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 MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020416SynopsisPharmacology/Drug DiscoveryEpidemiology/Public HealthCardiovascular MedicineEpidemiologyPharmacology and toxicologyFlying in the Face of the Evidence; Low Aspirin Use in the US Outpatient Setting Synopsis12 2005 15 11 2005 2 12 e416Copyright: © 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.
Underutilization of Aspirin Persists in US Ambulatory Care for the Secondary and Primary Prevention of Cardiovascular Disease
==== Body
There are many instances in medical practice of an intervention becoming adopted despite evidence of it not being of the highest quality. The converse is also true; despite good evidence for the effectiveness of a treatment, it may not be implemented for a variety of different reasons. Reasons may be social or financial or due to a lack of knowledge of the benefits of the intervention.
In this issue of PLoS Medicine, Randall Stafford and colleagues investigate the use of one such underused intervention, aspirin, in the United States. Many studies have shown that aspirin is beneficial as both primary and secondary prevention of cardiovascular disease in a wide range of patients who do not otherwise have contraindications such as increased bleeding risk. For example, the American College of Chest Physicians has issued a grade 1A recommendation (such recommendations are “strong and indicate that the benefits … outweigh risks, burden, and costs”) that aspirin is used for secondary prevention of acute coronary events. The American Diabetes Association recommends aspirin to all people with diabetes over 40, or younger if they have additional cardiovascular risk factors.
Stafford and colleagues used the 1993–2003 US National Ambulatory Medical Care Survey (NAMCS) and National Hospital Ambulatory Medical Care Survey (NHAMCS) to estimate aspirin use by cardiovascular risk. These validated surveys are comprehensive collections of data on prescribing; NAMCS captures health-care services provided by private office-based physicians, whereas NHAMCS captures services offered at hospital outpatient departments. Their data are representative of care provided in a huge number of outpatient visits—around 750 million in 2003.
What the authors found was that although aspirin use increased steadily over the time of the survey, even at its highest—in 2003—it was prescribed at only 32.8 % of high-risk visits; for low-risk visits, the rate was just 1%–3%. They also compared prescribing of aspirin and statins, and found that the rates of prescribing of statins overtook that of aspirin in 1997–1998, and rose steadily thereafter. There were also substantial differences in the age of patients prescribed aspirin, with lower usage found in those below 45 than those 45 or above, and lower usage in women. Also, there was lower prescribing by noncardiologists than cardiologists, in private practices versus hospital outpatient departments, and at return visits versus first-time visits. In people with diabetes who were at intermediate risk, aspirin use was only 11.7% by 2003.
So despite all the recommendations and evidence for the use of aspirin, rates of prescribing are low, even for conditions where clear guidelines exist. As the authors say, “gaps observed with secondary prevention are particularly concerning, given the existence of conclusive clinical evidence and unequivocal practice guidelines.” One particular cause of this lack of adherence to guidelines may be specific to the US—the less restrictive regulations on newer drug advertising of drugs such as statins, particularly widespread consumer advertising. Hence, despite good evidence and the cost effectiveness of aspirin, statins are increasingly preferred over aspirin
As the authors conclude, “marked changes in clinical practice are unlikely to occur unless more aggressive, innovative means are implemented to enhance health-care provider and patient adherence to consensus guidelines.”
|
0
|
PMC1283367
|
CC BY
|
2021-01-05 10:39:27
|
no
|
PLoS Med. 2005 Dec 15; 2(12):e416
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020416
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020418SynopsisNeuroscienceOtherNeurology/NeurosurgeryPathologyDementiaNeurologyPathologyWill Stopping Aβ Production Reverse the Damage in Alzheimer Disease? Synopsis12 2005 15 11 2005 2 12 e418This is an open-access article distributed under the terms of the Creative Commons Public Domain Declaration, which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose.2005This is an open-access article distributed under the terms of the Creative Commons Public Domain declaration, which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose.
Persistent Amyloidosis following Suppression of Aβ Production in a Transgenic Model of Alzheimer Disease
==== Body
Dementia is a common condition in the elderly; around 6% of people over 65 and up to 50% over 90 have some form of dementia, about half of which are due to Alzheimer disease (AD). The dementia caused by AD has an insidious onset and a progressive course with slow deterioration in cerebral function, initially affecting short-term memory and cognitive skills, and later speech, motor functions, and personality. Death usually occurs within four to eight years after diagnosis.
The aim of treatment is to reverse cognitive decline and improve behavioral and psychological functions. Key questions in Alzheimer research are how best to halt the progression of disease to maintain and if possible restore cognitive skills, and when to initiate such interventions in order to be effective.
AD is identified at autopsy by the presence of hallmark lesions in key regions of the brain. These lesions, known as amyloid plaques, are formed by the aggregation of small peptides, called amyloid β peptide (Aβ), that are produced when amyloid precursor protein (APP) is cleaved by the action of two enzymes, β-APP cleaving enzyme and γ-secretase. One approach to the treatment of Alzheimer is, therefore, limiting the production of Aβ from its precursor by inhibiting one or both of these enzymes. However, it is not yet clear whether this approach will prevent the brain lesions and cognitive symptoms from getting worse, and if it will then promote the removal of preexisting plaques and reverse cognitive decline.
To answer such questions, Joanna Jankowsky and colleagues have developed mice that produce Aβ at levels sufficient to induce severe amyloid burden by six months of age. The animals carry an additional transgene that acts as a switch to control when Aβ is produced. Commonly known as the tet-off system, the switch is turned off when the mice are fed tetracycline or its analog, doxycycline. Once given the drug, Aβ production in the brains of these mice diminishes by more than 95% of pretreatment levels within two weeks. This system, thus, mimics the effect of shutting down Aβ production with enzyme inhibitors that are being developed for use in human patients.
Amyloid plaques, shown here as false-color images, are highly stable structures in vivo
In the study, the researchers used doxycycline to switch off production of Aβ, and examined what happened to the amyloid pathology. Not surprisingly, the increase in number and size of amyloid lesions that normally occurs as the mice get older was completely prevented by suppressing Aβ production. However, the researchers also found no substantial clearance of preexisitng plaques, even after six months of treatment (one-quarter of the normal mouse lifespan).
What do these findings mean for human Alzheimer research? First, the study provides evidence that the lesions found in AD may be more difficult for the brain to repair than protein aggregates found in other diseases such as Huntington or prion disease. Second, the findings suggest that the removal of plaques, once formed, may require more than simply halting the production of the Aβ peptide. However, as with all animal models, there are differences in comparison to the human disease, leading to both over- and underestimation of the relative importance of an effect in humans. The researchers do not yet know whether the plaques formed in mice may be more resistant to clearance than those seen in human disease. Conversely, the human brain, unlike the murine one, may have a more efficient way of clearing amyloid plaques. What this study makes clear is that treatments directed at reducing Aβ peptide production in AD will likely be most effective when started as early as possible.
|
0
|
PMC1283368
|
CC0
|
2021-01-05 10:39:26
|
no
|
PLoS Med. 2005 Dec 15; 2(12):e418
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020418
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627755610.1371/journal.pbio.0030372PrimerCell BiologyDevelopmentMammalsMus (Mouse)An Ideal Society? Neighbors of Diverse Origins Interact to Create and Maintain Complex Mini-Organs in the Skin PrimerMillar Sarah E 11 2005 15 11 2005 15 11 2005 3 11 e372Copyright: © 2005 Sarah E. Millar.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.
Molecular Dissection of Mesenchymal-Epithelial Interactions in the Hair Follicle
Molecular Signatures of the Developing Hair Follicle
Hair follicles are the focus of this primer on mammalian skin development.
==== Body
The largest organ in the body, the skin provides protection against environmental insults and dehydration, and is an important gateway for sensory input. In addition to forming an environmental barrier, the skin has evolved to produce an amazing variety of appendages, including scales, feathers, hair follicles, sweat glands, and mammary glands (Figure 1). These organs arise from embryonic skin progenitor cells and endow wide-ranging properties, including regulation of body temperature, and the ability to fly, nurse young, and attract mates. As the body's primary frontier, the skin is both vulnerable to disease, such as melanoma, and a source of therapeutic promise, harboring accessible stem cells (SCs) that can regenerate skin and potentially other organs.
Figure 1 Histological and Schematic Depictions of Mature Skin
(A) Hematoxylin/eosin-stained section of mouse skin at postnatal day 28, showing hair follicles in the anagen growth phase; major layers of the skin are indicated.
(B) Schematic depiction of postnatal skin showing a hair follicle in the growth phase. Biological processes occurring in the skin are listed and SC locations are indicated. Dark blue arrows indicate the movements of stem and matrix cell progeny; pale blue arrows indicate SC self-renewal. Pink, epidermis and hair follicle outer root sheath; yellow, inner root sheath; green, matrix; red, hair follicle DP; light brown, hair shaft precursors; darker brown, hair shaft; violet circles, SCs; black ovals, dermal fibroblasts.
Mammalian skin contains three major cell types: epithelial cells that form a stratified epidermis containing specialized intermediate filament proteins called keratins; mesenchymal cells that form the underlying dermis and, together with epithelial cells, contribute to hair follicles and other appendages; and melanocytes that provide the skin and hair follicles with pigmentation. In addition, the skin is highly innervated, and contains populations of specialized cells including dentritic Langerhans cells (a type of antigen presenting cell), Merkel mechanoreceptor cells (which form complexes with sensory axons), and mast cells (which produce histamine). These different cell types have diverse origins, and undergo extensive interactions, migration, proliferation, and differentiation during embryonic development.
Morphogenesis is controlled by a relatively small number of key intercellular signaling pathways that occur sequentially and in specific combinatorial fashions. Activation of these pathways is directed by secreted ligands, including members of the WNT, fibroblast growth factor, tumor necrosis factor, and Hedgehog families, and bone morphogenetic protein (BMP) and other transforming growth factor superfamily members, which act over relatively short distances and bind to specific receptor proteins on neighboring cells to regulate cell shape, size, adhesion, polarity, movements, proliferation, and transcriptional activity. The molecular events underlying skin development in mammals have been studied using a variety of systems and techniques. These include analysis of human genetic syndromes and spontaneously occurring and induced mouse mutants, production of transgenic and gene-targeted mouse models, in vitro culture of rodent and human skin cells, and grafting of human or rodent skin to immune-deficient mice.
Epidermal Origins
The epidermis originates from the outer layer of the embryo, the surface ectoderm. BMPs activate the epidermal differentiation program and induce the expression of keratin proteins via several known transcription factors [1,2]. The surface ectoderm proliferates and migrates from the dorsal midline to cover the embryo [3], and persists as a simple epithelium until approximately embryonic day (E) 9.5 of mouse embryogenesis. At this stage basal cells begin to express keratins 5 and 14, presaging epidermal stratification [2], which requires the activity of a key epidermal transcription factor that also regulates epidermal fate, proliferation, and adhesion [4–9]. By birth, the epidermis consists of a proliferative basal layer that differentiates to form suprabasal layers, and an outer, “cornified,” enucleated shell that constitutes the epidermal barrier and is continuously shed and replenished. Epidermal renewal continues throughout life, suggesting that this tissue harbors an SC population; maintenance of epidermal SCs in vivo has been recently shown to require the Rho guanosine triphosphatase Rac1 [10].
Dermal Origins
The dermis of the skin consists of loosely packed fibroblasts and has a remarkable variety of embryonic origins. Fate mapping of mammalian dermis, which traces the developmental path of dermal cells, has been achieved by introducing artificial genes into mice. Specifically, the bacterial Cre recombinase enzyme is introduced under the control of a promoter that is active in a particular embryonic region, in this case either lateral plate mesoderm or neural crest. This transgene is combined with a Cre-activatable lacZ reporter gene, resulting in lacZ expression in Cre-expressing embryonic cells and all of their descendants. This allows tracking of the fates of these descendants by X-gal staining to detect activity of the β-galactosidase enzyme encoded by lacZ. These experiments showed that the lateral plate mesoderm gives rise to ventral trunk dermis, while head dermis arises, at least in part, from neural crest, a transient population of migratory cells derived from the neural plate [11–14]. While fate mapping of dorsal trunk dermis in mammalian embryos has not yet been described, analysis of chick–quail chimeras revealed that in avian embryos the dorsal dermis arises from the dorsal regions of somites [15,16]. Embryonic dermal fibroblasts from different body regions have different inductive properties [17–19] and show position-specific differences in gene expression, even in the adult [20]. The molecular controls of mammalian dermal development remain relatively unexplored.
Pigmentation of the epidermis and hair follicles is supplied by melanocytes that manufacture melanin (black pigment) or phaeomelanin (yellow/orange pigment) and deposit it into melanosomes that are transferred to hair shaft or epidermal cells. Melanocytes differentiate from melanoblasts that arise from the neural crest. After undergoing an initial period of proliferation, melanoblasts migrate between the dermatome and overlying ectoderm, and from E10.5 of mouse embryogenesis migrate through the developing dermis. Starting at approximately E12.5, melanoblasts move from the dermis to the epidermis, and by E15.5 they begin to migrate into the developing hair follicles [21–24] (Figure 2). By two weeks of age melanoblasts are absent from haired regions of mouse epidermis [24]. In contrast, in human skin melanoblasts and melanocytes populate the epidermis as well as the hair follicles. Genetic analysis in mice has revealed requirements for several factors in melanoblast migration, proliferation, and/or survival [25,26]. Mutations in a variety of these genes are known to cause pigmentation and other neural-crest-related defects in humans as well as in mice [25].
Figure 2 Schematic Depiction of Directions of Melanoblast Migration in Embryonic Mouse Skin from E9.5 to E17.5
Pink, epithelium; black dots, dermal fibroblasts; blue ovals, melanoblasts; red dots, dermal condensate/DP.
Interactions between Neighboring Cells Control Hair Follicle Development
While epidermal cells, dermal fibroblasts, and melanocytes have differing embryonic origins and migration pathways [27–29], these cells interact extensively during development of the skin and its appendages [17,19,30,31]. Experiments in which epithelium and mesenchyme from different body regions, different developmental stages, or different species were combined and allowed to develop at ectopic sites revealed that signals from the dermis initiate hair follicle development by inducing formation of a regular array of local thickenings, or placodes, in the overlying surface epithelium [17]. Signals from each epithelial placode induce the clustering of a ball of underlying mesenchymal cells, the dermal condensate. Signaling from the dermal condensate to the epithelium results in downward growth of the epithelial cells (Figure 3), which surround the dermal condensate, thereafter known as the dermal papilla (DP) [17]. The DP directs the rapid proliferation of adjacent epithelial cells called matrix cells, which gradually exit from the proliferative compartment, and terminally differentiate to form the hair shaft and the inner root sheath that molds the shaft [32]. An outer root sheath that is contiguous with the epidermis surrounds the inner root sheath, and the follicle is bound by a dermal sheath.
Figure 3 Successive Stages of Hair Follicle Development in Embryonic Mouse Skin at E14.5–E15.5
Sections were stained with Hoechst dye to reveal the nuclei. The epithelial–mesenchymal boundary is marked by a dashed white line in each panel, and sites of hair follicle development are bracketed. The directions of inductive signals are indicated for each stage (green arrows). Left: placode stage; note larger size and columnar appearance of placode nuclei compared with those in adjacent epithelium. Middle: induction of the dermal condensate (DC). Right: formation of a germ stage hair follicle with a well-developed dermal condensate.
Surgical removal of the DP and the contiguous lower dermal sheath prevents hair growth, indicating the importance of the DP as a key signaling center for the follicle [33]. DPs or dermal sheaths implanted at ectopic sites demonstrate a remarkable ability to induce the formation of new follicles, even from terminally differentiated epithelium such as the central cornea of the eye [34–36]. This inductive ability is lost if the DP cells are cultured in the absence of epithelial cells or without the addition of WNT proteins [37,38], suggesting that signals from neighboring epithelium are necessary for maintenance of DP function (see Figure 1B).
Not surprisingly, these extensive epithelial–mesenchymal interactions require the activities of secreted ligands that allow communication between different cell types. WNT signals play key roles at the initiation of embryonic hair follicle development [39–42]. Fibroblast growth factor proteins and the tumor necrosis factor family member ectodysplasin also promote epithelial placode formation, while Sonic hedgehog controls proliferation of both embryonic and postnatal hair follicle epithelium [30]. The BMP inhibitor Noggin is essential for hair follicle development [43], and acts in conjunction with WNT signaling and transforming growth factor β2 to control invasion of the dermis by hair follicle epithelial cells [44,45]. Differentiation of matrix cells toward either inner root sheath or hair shaft requires BMP signaling and numerous transcription factors [30,46–49]. The precise mechanisms by which these factors facilitate communication between cells of different types to produce an organ as complex as the hair follicle remain only partially understood.
Hair Follicle Growth: The Role of SCs
The hair follicle undergoes cycles of growth and regression throughout life. At the onset of a new cycle of hair growth (anagen), signals from the DP are thought to stimulate the transient proliferation of epithelial SCs in the permanent bulge region of the follicle [50]. Bulge SC descendants give rise to all the epithelial layers of the regenerating hair follicle [51–55] (see Figure 1B). The extent to which these cells also contribute to epidermis in normal and pathological situations is currently not clear. Epithelial SCs coexist in the bulge with a population of melanocyte SCs that proliferates transiently at anagen onset and gives rise to differentiated melanocytes that populate the hair follicle bulb [56]. Hair graying in genetic mouse models and in humans is associated with loss of the melanocyte SCs [57]. Forced expression of a stabilized form of the WNT pathway effector â-catenin in hair follicle epithelial SCs causes the onset of a new cycle of growth of pigmented hair, suggesting that WNT signaling is critical for epithelial SC proliferation, and that secreted factors downstream of β-catenin can activate melanocyte SCs [58–60].
Several recent reports have identified cell populations in the skin that appear to be capable of differentiating into cell types other than hair follicle epithelium and epidermis, raising the exciting possibility that skin SCs could be used to regenerate organs beside the skin and hair follicles. For example, neural crest derivatives [14] and cells expressing a Nestin–green fluorescent protein (GFP) transgene [61] have been isolated from the bulge regions of vibrissa (whisker) follicles and can differentiate in culture into neurons, smooth muscle cells, and melanocytes. Multipotent cells termed skin-derived precursor cells have also been derived from the dermis of adult skin, and several lines of evidence suggest that the endogenous niche for these cells is the hair follicle DP [13]. In a separate study, clonal cell lines established from dissected rat vibrissa DPs and dermal sheaths had extended proliferative capacities and could differentiate into fat- and bone-related lineages [62]. Although it is possible that the skin-derived precursor cells and dissected vibrissa preparations could have been contaminated with non-DP cells, these data suggest that DP cells not only possess remarkable inductive capacities but also reside in a niche provided by the surrounding follicular cells that enables them to maintain a large proliferative capacity and the ability to differentiate into multiple cell types.
Relatively few cells reside in specialized niches such as the hair follicle DP and bulge, making molecular analyses of pure populations of these cells difficult. Recently, however, hair follicle bulge epithelial SCs have been isolated by utilizing a specific promoter to target GFP to adult bulge epithelium [52], or by taking advantage of the rarely cycling properties of epithelial bulge cells to target GFP expression to this population [55]. These approaches permit the use of fluorescence-activated cell sorting (FACS) to isolate and transcriptionally profile pure bulge epithelial SC populations.
In the current issue of PLoS Biology, Rendl et al. [63] take similar approaches to systematically analyze the transcriptional profile of the DP and four surrounding cell types: dermal fibroblasts, melanocytes, epithelial matrix cells, and outer root sheath cells, from mouse hair follicles four days after birth. At this time point, embryonic hair follicles have reached a late stage in their differentiation and are beginning to synthesize hair shafts. Rendl et al. marked each skin cell population with a unique combination of fluorescent tags, using a clever mix of cell-type-specific transgenic expression of red and green fluorescent proteins, together with immunolabeling of specific antigens. Each of the different cell populations was found to possess a distinct molecular signature. These experiments identified genes not previously associated with hair follicles, and revealed cell-type-specific expression for several genes affected in hair disorders.
Importantly, the results of this study support and extend previous observations that the DP expresses genes originally associated with neural SCs [13]. However, Rendl et al. find that the DP transcriptional profile is clearly distinct from that of neural SCs, neural crest, or melanocytes, and is most similar to that of dermal fibroblasts. These data support the concept that DP cells and dermal fibroblasts arise from similar precursors. Consistent with this view, use of a Hoxb6-Cre transgene to label derivatives of lateral plate mesoderm resulted in apparent marking of ventral trunk DPs as well as ventral trunk dermal fibroblasts, suggesting that at least some DP cells have common origins with neighboring dermal fibroblasts ([11]; S. I. Candille and G. S. Barsh, personal communication). Thus, differences in gene expression in DP cells compared with neighboring dermis may reflect the influences of follicular matrix and outer root sheath cells, rather than being the result of a distinct embryonic origin of the DP. Interestingly, WNT signaling and inhibition of BMP signaling are key elements in induction and maintenance of both hair follicles [39–43] and neural crest [1], and are featured in the DP signature. The re-utilization of these pathways in the complex signaling environment of the hair follicle may account in part for the unusual gene expression pattern in the DP, for the remarkable inductive properties of DP cells, and perhaps for their apparent ability to maintain multipotency.
The approach employed by Rendl et al. has produced a remarkably comprehensive picture of gene expression in different components of the skin and hair follicles in the early postnatal period. These data will likely form the basis for a multitude of further investigations into the molecular nature of mesenchymal–epithelial interactions in the hair follicle, and the functions of hair-follicle-expressed genes in both normal development and disease. Similar approaches could be employed to begin to address outstanding questions in skin and appendage biology. Profiling of different follicular compartments during the hair growth cycle may identify putative inductive signals produced by the DP, and could provide clues as to how cycling activity of epithelial and melanocyte SCs is coordinated. Profiling of different cell compartments in various developing appendages might begin to reveal the molecular mechanisms by which apparently similar sets of signaling pathways direct the formation of diverse organs. As the basic mechanisms of hair follicle development and hair growth are conserved between mice and humans [30], these studies are directly relevant to human hair follicle biology. However, some important aspects of human hair biology, such as the follicular response to androgens at puberty and in male pattern baldness, are absent in mice [64]. The development of antibody-based multicolor labeling systems to isolate pure populations of human hair follicle cells would therefore be a valuable additional tool for investigation of human hair disease.
The author would like to thank Gregory S. Barsh and Sophie I. Candille for sharing unpublished data and for helpful suggestions, and George Cotsarelis for his insightful comments. Research in Sarah Millar's lab is supported by NIH grants R01AR047709 and R01DE015342.
Citation: Millar SE (2005) An ideal society? Neighbors of diverse origins interact to create and maintain complex mini-organs in the skin. PLoS Biol 3(11): e372.
Sarah E. Millar is in the Departments of Dermatology and Cell and Developmental Biology, University of Pennsylvania, Philadelphia, Pennsylvania, United States of America. E-mail: [email protected]
Abbreviations
BMPbone morphogenetic protein
DPdermal papilla
E[number]embryonic day [number]
GFPgreen fluorescent protein
SCstem cell
==== Refs
References
Meulemans D Bronner-Fraser M Gene-regulatory interactions in neural crest evolution and development Dev Cell 2004 7 291 299 15363405
Byrne C Tainsky M Fuchs E Programming gene expression in developing epidermis Development 1994 120 2369 2383 7525178
Mintz B Padykula HA Gene expression in allophenic mice Control mechanisms in the expression of cellular phenotypes 1970 New York Academic Press 15 42
Koster MI Kim S Mills AA DeMayo FJ Roop DR p63 is the molecular switch for initiation of an epithelial stratification program Genes Dev 2004 18 126 131 14729569
Lechler T Fuchs E Asymmetric cell divisions promote stratification and differentiation of mammalian skin Nature 2005 437 275 280 16094321
Bakkers J Hild M Kramer C Furutani-Seiki M Hammerschmidt M Zebrafish DeltaNp63 is a direct target of Bmp signaling and encodes a transcriptional repressor blocking neural specification in the ventral ectoderm Dev Cell 2002 2 617 627 12015969
Yang A Schweitzer R Sun D Kaghad M Walker N p63 is essential for regenerative proliferation in limb, craniofacial and epithelial development Nature 1999 398 714 718 10227294
Lee H Kimelman D A dominant-negative form of p63 is required for epidermal proliferation in zebrafish Dev Cell 2002 2 607 616 12015968
Ihrie RA Marques MR Nguyen BT Horner JS Papazoglu C Perp is a p63-regulated gene essential for epithelial integrity Cell 2005 120 843 856 15797384
Benitah SA Frye M Glogauer M Watt FM Stem cell depletion through epidermal deletion of Rac1 Science 2005 309 933 935 16081735
Candille SI Raamsdonk CDV Chen C Kuijper S Chen-Tsai Y Dorsoventral patterning of the mouse coat by Tbx15
PLoS Biol 2004 2 e3 10.1371/journal.pbio.0020003 14737183
Chai Y Jiang X Ito Y Bringas P Han J Fate of the mammalian cranial neural crest during tooth and mandibular morphogenesis Development 2000 127 1671 1679 10725243
Fernandes KJ McKenzie IA Mill P Smith KM Akhavan M A dermal niche for multipotent adult skin-derived precursor cells Nat Cell Biol 2004 6 1082 1093 15517002
Sieber-Blum M Grim M Hu YF Szeder V Pluripotent neural crest stem cells in the adult hair follicle Dev Dyn 2004 231 258 269 15366003
Mauger A The role of somitic mesoderm in the development of dorsal plumage in chick embryos. I. Origin, regulative capacity and determination of the plumage-forming mesoderm J Embryol Exp Morphol 1972 28 313 341 4642991
Sengel P Morphogenesis of skin 1976 Cambridge Cambridge University Press 277
Hardy MH The secret life of the hair follicle Trends Genet 1992 8 55 60 1566372
Foley J Dann P Hong J Cosgrove J Dreyer B Parathyroid hormone-related protein maintains mammary epithelial fate and triggers nipple skin differentiation during embryonic breast development Development 2001 128 513 525 11171335
Miletich I Sharpe PT Neural crest contribution to mammalian tooth formation Birth Defects Res C Embryo Today 2004 72 200 212 15269893
Chang HY Chi JT Dudoit S Bondre C van de Rijn M Diversity, topographic differentiation, and positional memory in human fibroblasts Proc Natl Acad Sci U S A 2002 99 12877 12882 12297622
Mayer TC Interactions between normal and pigment cell populations mutant at the dominant-spotting (W) and steel (Sl) loci in the mouse J Exp Zool 1979 210 81 88 536714
Jordan SA Jackson IJ MGF (KIT ligand) is a chemokinetic factor for melanoblast migration into hair follicles Dev Biol 2000 225 424 436 10985860
Peters EM Tobin DJ Botchkareva N Maurer M Paus R Migration of melanoblasts into the developing murine hair follicle is accompanied by transient c-Kit expression J Histochem Cytochem 2002 50 751 766 12019292
Fitch KR McGowan KA van Raamsdonk CD Fuchs H Lee D Genetics of dark skin in mice Genes Dev 2003 17 214 228 12533510
Baxter LL Hou L Loftus SK Pavan WJ Spotlight on spotted mice: A review of white spotting mouse mutants and associated human pigmentation disorders Pigment Cell Res 2004 17 215 224 15140066
Van Raamsdonk CD Fitch KR Fuchs H de Angelis MH Barsh GS Effects of G-protein mutations on skin color Nat Genet 2004 36 961 968 15322542
Mintz B Silvers WK Histocompatibility antigens on melanoblasts and hair follicle cells Transplantation 1970 9 497 505 4910946
Mintz B Gene control of the mouse pigmentary system Genetics 1969 61 s41
Nesbitt MN Variegation patterns in mice heterozygous for X-linked markers Genetics 1970 64 s46
Millar SE Molecular mechanisms regulating hair follicle development J Invest Dermatol 2002 118 216 225 11841536
Jernvall J Thesleff I Reiterative signaling and patterning during mammalian tooth morphogenesis Mech Dev 2000 92 19 29 10704885
Sperling LC Hair anatomy for the clinician J Am Acad Dermatol 1991 25 1 17 1880235
Oliver RF Whisker growth after removal of the dermal papilla and lengths of follicle in the hooded rat J Embryol Exp Morphol 1966 15 331 347 5964281
Reynolds AJ Jahoda CA Cultured dermal papilla cells induce follicle formation and hair growth by transdifferentiation of an adult epidermis Development 1992 115 587 593 1425341
Reynolds AJ Lawrence C Cserhalmi-Friedman PB Christiano AM Jahoda CA Trans-gender induction of hair follicles Nature 1999 402 33 34 10573414
Pearton DJ Yang Y Dhouailly D Transdifferentiation of corneal epithelium into epidermis occurs by means of a multistep process triggered by dermal developmental signals Proc Natl Acad Sci U S A 2005 102 3714 3719 15738417
Inamatsu M Matsuzaki T Iwanari H Yoshizato K Establishment of rat dermal papilla cell lines that sustain the potency to induce hair follicles from afollicular skin J Invest Dermatol 1998 111 767 775 9804336
Kishimoto J Burgeson RE Morgan BA Wnt signaling maintains the hair-inducing activity of the dermal papilla Genes Dev 2000 14 1181 1185 10817753
Andl T Reddy ST Gaddapara T Millar SE WNT signals are required for the initiation of hair follicle development Dev Cell 2002 2 643 653 12015971
Huelsken J Birchmeier W New aspects of Wnt signaling pathways in higher vertebrates Curr Opin Genet Dev 2001 11 547 553 11532397
van Genderen C Okamura RM Farinas I Quo RG Parslow TG Development of several organs that require inductive epithelial-mesenchymal interactions is impaired in LEF-1-deficient mice Genes Dev 1994 8 2691 2703 7958926
Gat U DasGupta R Degenstein L Fuchs E De novo hair follicle morphogenesis and hair tumors in mice expressing a truncated beta-catenin in skin Cell 1998 95 605 614 9845363
Botchkarev VA Botchkareva NV Roth W Nakamura M Chen LH Noggin is a mesenchymally derived stimulator of hair-follicle induction Nat Cell Biol 1999 1 158 164 10559902
Jamora C DasGupta R Kocieniewski P Fuchs E Links between signal transduction, transcription and adhesion in epithelial bud development Nature 2003 422 317 322 12646922
Jamora C Lee P Kocieniewski P Azhar M Hosokawa R A signaling pathway involving TGF-β2 and Snail in hair follicle morphogenesis PLoS Biol 2005 3 e11 10.1371/journal.pbio.0030011 15630473
Kulessa H Turk G Hogan BL Inhibition of Bmp signaling affects growth and differentiation in the anagen hair follicle EMBO J 2000 19 6664 6674 11118201
Kobielak K Pasolli HA Alonso L Polak L Fuchs E Defining BMP functions in the hair follicle by conditional ablation of BMP receptor IA J Cell Biol 2003 163 609 623 14610062
Andl T Ahn K Kairo A Chu EY Wine-Lee L Epithelial Bmpr1a regulates differentiation and proliferation in postnatal hair follicles and is essential for tooth development Development 2004 131 2257 2268 15102710
Kaufman CK Zhou P Pasolli HA Rendl M Bolotin D
GATA-3: An unexpected regulator of cell lineage determination in skin
Genes Dev 2003 17 2108 2122 12923059
Cotsarelis G Sun TT Lavker RM Label-retaining cells reside in the bulge area of pilosebaceous unit: Implications for follicular stem cells, hair cycle, and skin carcinogenesis Cell 1990 61 1329 1337 2364430
Oshima H Rochat A Kedzia C Kobayashi K Barrandon Y Morphogenesis and renewal of hair follicles from adult multipotent stem cells Cell 2001 104 233 245 11207364
Morris RJ Liu Y Marles L Yang Z Trempus C Capturing and profiling adult hair follicle stem cells Nat Biotechnol 2004 22 411 417 15024388
Taylor G Lehrer MS Jensen PJ Sun TT Lavker RM Involvement of follicular stem cells in forming not only the follicle but also the epidermis Cell 2000 102 451 461 10966107
Blanpain C Lowry WE Geoghegan A Polak L Fuchs E Self-renewal, multipotency, and the existence of two cell populations within an epithelial stem cell niche Cell 2004 118 635 648 15339667
Tumbar T Guasch G Greco V Blanpain C Lowry WE Defining the epithelial stem cell niche in skin Science 2004 303 359 363 14671312
Nishimura EK Jordan SA Oshima H Yoshida H Osawa M Dominant role of the niche in melanocyte stem-cell fate determination Nature 2002 416 854 860 11976685
Nishimura EK Granter SR Fisher DE Mechanisms of hair graying: Incomplete melanocyte stem cell maintenance in the niche Science 2005 307 720 724 15618488
Van Mater D Kolligs FT Dlugosz AA Fearon ER Transient activation of beta-catenin signaling in cutaneous keratinocytes is sufficient to trigger the active growth phase of the hair cycle in mice Genes Dev 2003 17 1219 1224 12756226
Lo Celso C Prowse DM Watt FM Transient activation of beta-catenin signalling in adult mouse epidermis is sufficient to induce new hair follicles but continuous activation is required to maintain hair follicle tumours Development 2004 131 1787 1799 15084463
Lowry WE Blanpain C Nowak JA Guasch G Lewis L Defining the impact of beta-catenin/Tcf transactivation on epithelial stem cells Genes Dev 2005 19 1596 1611 15961525
Amoh Y Li L Katsuoka K Penman S Hoffman RM Multipotent nestin-positive, keratin-negative hair-follicle bulge stem cells can form neurons Proc Natl Acad Sci U S A 2005 102 5530 5534 15802470
Jahoda CA Whitehouse J Reynolds AJ Hole N Hair follicle dermal cells differentiate into adipogenic and osteogenic lineages Exp Dermatol 2003 12 849 859 14714566
Rendl M Lewis L Fuchs E Molecular dissection of mesenchymal–epithelial interactions in the hair follicle PLoS Biol 2005 3 e331 10.1371/journal.pbio.0030331 16162033
Cotsarelis G Millar SE Towards a molecular understanding of hair loss and its treatment Trends Mol Med 2001 7 293 301 11425637
|
16277556
|
PMC1283383
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Nov 15; 3(11):e372
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030372
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0030382Book Reviews/Science in the MediaEvolutionNoneDarwin's Other Books: “Red” and “Transmutation” Notebooks, “Sketch,” “Essay,” and Natural Selection
Book ReviewEldredge Niles 11 2005 15 11 2005 15 11 2005 3 11 e382A handwritten manuscript note by Charles Darwin from Natural Selection portfolio (Cambridge University Library) Copyright: © 2005 Niles Eldredge.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.Study of Darwin's unpublished works, freely available on-line through the American Natural History Museum, reveals the origins of his thoughts on evolution.
==== Body
Depending on how you count them up, Charles Darwin published just over twenty books in his lifetime. His first—the Journal of Researches [1], also known as The Voyage of the Beagle was his most famous—until Darwin, pressured by the arrival in 1858 of A. R. Wallace's manuscript on evolution through natural selection, stopped working on his “Big Species Book,” [2] and wrote instead his epochal On the Origin of Species by Means of Natural Selection. Or the Preservation of Favoured Races in the Struggle for Life [3]. In between, and thereafter, Darwin published monographs and specialized narratives on topics as disparate as barnacle taxonomy, coral reef development, and insectivorous plants. Yet it is, of course, the Origin of Species that changed the world, establishing Darwin as one of the great thinkers in Western cultural history.
So much is well-known. Far less appreciated is the fact that Darwin wrote several other books, all on evolution, none of which were published in his lifetime. Together, they form a series that preserves the “evolutionary” history of Darwin's ideas from their very inception to their most mature form—while also revealing the more prosaic development of Darwin's written rhetoric of the Origin of Species. All have been subsequently published, and are now freely available online, constituting the initial components of the Darwin manuscript project page of the American Museum of Natural History Digital Library of Evolution. The Darwin manuscript project complements the museum's exhibition, Darwin, opening 19 November 2005; further analysis of these works can be found in my companion volume to the exhibition [4].
The first of these books is actually a series of notebooks. I include them because they are the foundational writings for all Darwin's later, more discursive, discussions of evolution. Many of the themes—and, indeed, some of the original language—of the familiar passages of the Origin of Species are found in the pages of these notebooks—maddeningly interspersed in near-chaotic fashion with all manner of geological and biological observations and notations gleaned from the literature, and Darwin's already burgeoning correspondence and conversations. Indeed, only the “principle of divergence” came along later to add to Darwin's themes and arguments.
“Red” and “Transmutation” Notebooks
The “Red Notebook” [5] is the first of the series of notebooks in which Darwin established the essential elements of his evolutionary theory. Although apparently started while still on the Beagle in 1836—recording various latitude, longitude, and depth soundings—the last third of the notebook seems to have been filled out after Darwin returned to England in late 1836, early 1837. Historians still disagree whether or not—or the degree to which—Darwin had tumbled to the idea of evolution while still on the Beagle. I fully agree with Kohn et al. [6] that the famous passage in his Ornithological Notes, discussing the differentiation of “varieties” of mockingbirds and tortoises on various islands in the Galapagos and concluding that “if there is the slightest foundation for these remarks to zoology of Archipelagoes—will be well worth examining; for such facts would undermine the stability of Species” [7] in fact does establish that Darwin was thinking about evolution in the final months before the Beagle arrived home. But nothing else unambiguously written while still aboard ship has as yet turned up to support this view.
Darwin was a fully committed evolutionist by the time the evolutionary passages of the “Red Notebook” were written. As he would subsequently write in the topic sentence of the Origin of Species, Darwin had been greatly struck by “certain facts of the distribution of the inhabitants of South America” that “throw some light on the origin of species.” [3]. Elsewhere, Darwin makes clear that there were three distinct patterns of replacement of “allied” forms. First, the replacement of extinct species by modern ones—belonging to groups unique to that part of the world. For example, armadillos now live, while the obviously closely similar giant glyptodonts (which Darwin collected in Argentina) are now extinct. Both are edentate mammals—found only in the Americas. Second, in the living world, closely similar species tend to replace each other over broad expanses of mainland South America. The original example of this is the replacement of the common rhea (ostrich-like bird) by the lesser (Darwin's) rhea in more southerly stretches of South America. And third, the replacement by similar varieties or species of animals and plants on different islands (in the Galapagos especially—but he also mentions the two different forms of fox found on each of the two Falkland Islands). The mockingbirds and tortoises are early examples; as is well-known, Darwin did not himself see similar replacement patterns in finches, which later became known as “Darwin's finches,” and the equally riveting plant examples had to await expert analysis, forthcoming only after Darwin had been home for some time.
Darwin, famous for his views of gradual evolution through natural selection in the Origin of Species, is unexpectedly a saltationist in the “Red Notebook.” He thinks, given the lack of intergradations between fossil forms, or his rheas, that new species must arise suddenly from ancestral species. He maintains this view to some degree in Notebook B, first of the four famous “Transmutation Notebooks” [8], begun in the summer of 1837 and finished in early 1838. But with Notebook B, his attention turns to defining the first of three additional patterns, seeing these as expected observations if evolution is true. Darwin's initial three replacement patterns were inductive generations that took some while to dawn on his conscious mind. Now, with Notebook B he turns the tables and establishes the idea of evolution in a hypothetico-deductive framework.
There is grandeur in this view of life.
First of these new expected patterns is the nested set of taxa already recognized and embodied in Linnaeus's Systema Naturae [9]. We now know why, in other words, there seems to be a natural classification of species—an explanation that differs from creationism precisely because it does make predictions about what we should expect to observe if evolution is true. In what is Darwin's closest equivalent to Einstein's handwritten E = MC2, he writes (Notebook B, page 36) “I think,” [8] and sketches an abstract evolutionary tree. He goes on to add embryological resemblance and “the unity of type” (homology), all close correlates of the “natural system”—all seen as predicted observations under the theory of transmutation (descent with modification eventually equals evolution).
But Darwin wanted more: he was constantly searching for a mechanism. Finally, in Notebook D, after having read Thomas Malthus and learned for the first time that more organisms are born to each species each generation than can possibly survive and reproduce (otherwise, “the world would be standing room only in elephants after but a few thousand years,” [3] he wrote later in the Origin of Species), he formulated “natural selection.” As David Kohn [10] first pointed out, Darwin parses natural selection pithily on page 58 of Notebook E (1839): “Three principles will account for all: (1) Grandchildren like grandfathers (2) Tendency to small change «especially with physical change≫ (3) Great fertility in proportion to support of parents” [8]. In other words, (1) heredity, (2) variation (Darwin thought variation was induced in large measure spontaneously in the reproductive process and by the environment—views he held throughout his writings), and (3) the Malthusian principle of overproduction.
Therefore, natural selection—though not called such until the next “book” in our series (the 1842 “Sketch”). Darwin had been using the expression “my theory” to mean “evolution.” But now, the expression “my theory” more specifically means “evolution by natural selection.” It is in 1839, toward the end of the series of “Transmutation Notebooks,” that Darwin takes his next logical, if not fateful, step: in page 118 of Notebook E, he exhorts himself to rederive his original patterns in terms of his ideas on how natural selection works to produce evolutionary change. He is by now far beyond his initial attraction to saltational evolution: rather natural selection must produce finely gradational change. This puts him at odds with his very first evolutionary pattern, as Darwin is aware that paleontologists see little evidence of such change in their collections of fossilized plants and animals. He writes (Notebook E, page 6): “My very theory requires each form to have lasted for its time: but we ought in same bed if very thick to find some change in upper & lower layers.—good objection to my theory: a modern bed at present might be very thick & yet have same fossils” [8] Darwin, an intellectually very honest man, was troubled by this “good objection” throughout his evolutionary writings—devoting a chapter to the problem and essentially inventing the science of taphonomy (study of the formation of the fossil record) in Origin of Species.
After discovering natural selection in Notebooks D and E, Darwin turns renewed attention to both variation and the process of artificial selection in embryonic form— the analogy to what he would soon call “natural selection” by this time clear. The theme that varieties are incipient species—perhaps the most pervasive of Darwinian argumentative themes—is found in these notebooks [6], as are some other, more rhetorical, devices that show up in the later works, including the Origin of Species.
Particularly striking is Darwin's invocation of the travails of astronomers who labored so hard (occasionally relinquishing their lives!) to establish the laws of gravitation governing the behavior of celestial bodies. Darwin was not only fearful of attack on religious grounds, he also knew all too well that the only competing theory to explain the origin and diversity of life was in fact Judeo–Christian creationism. In Notebook B (page 101) [8], Darwin writes: “Astronomers might formerly have said that God ordered each planet to move in its particular destiny—in same manner God orders each animal created with certain form in certain country, but how much more simple and sublime power let attraction act according to certain laws such are as inevitable consequence let animal be created, then by the fixed laws of generation, such will be their successors—let the powers of transportal be such & so will be the form of one country to another—let geological changes go at such a rate, so will be the numbers & distribution of the species!!” Later in the notebooks, he mentions persecution of the astronomers—and also writes (Notebook D, page 36): “What a magnificent view one can take of the world Astronomical <& unknown> causes, modified by unknown ones. cause changes in geography & changes of climate superadded to change of climate from physical causes.—these superinduce changes of form in the organic world, as adaptation. & these changing affect each other, & their bodies, by certain laws of harmony keep perfect in these themselves.—instincts alter, reason is formed, & the world peopled with Myriads of distinct forms from a period short of eternity to the present time, to the future—How far grander than idea from cramped imagination that God created (warring against those very laws he established in all organic nature) the Rhinoceros of Java & Sumatra, that since the time of the Silurian, he has made a long succession of vile Molluscous animals—How beneath the dignity of him, who is supposed to have said let there be light and there was light…” [8].
This passage is “ancestral” to Darwin's most famous passage—concluding the Origin of Species some 21 years later. Here we have not only the analogy with scientific law replacing creationist belief in astronomy, but also the origin of the famous phrase “there is grandeur in this view of life” [3,11,12]. We even see here reference (albeit only in passing) to Javan and Sumatran rhinos—expanded and integral to the conclusions of each of Darwin's successive books on evolution.
The 1842 “Sketch” and 1844 “Essay”
In 1909, Francis Darwin (Charles and Emma's seventh child), on the 100th anniversary of his father's birth, published Foundations of the Origin of Species [13]. The book contained Francis's transcription of two of his father's unpublished, handwritten manuscripts, the “Sketch” of 1842 [11] and the much longer, discursive, and on the whole better-written “Essay” of 1844 [12].
The “Sketch” is Darwin's earliest known (and almost undoubtedly his very first) attempt to write out his evolutionary theory in essay form. The fact that it was never intended for publication, but rather served as a first-shot “dry-run” in setting out his views, is amply demonstrated by the sometimes elliptical, almost notebook-like passages with incomplete sentences and occasional reminders to himself on how to develop his arguments further. (Indeed, the last two paragraphs of Francis's 53-page edition are just these sorts of notes to himself).
The 1842 “Sketch” is an exciting read. Darwin is effectively organizing his thoughts and putting them in more coherent form for the first time. He adopts a two-part structure (retained in his 1844 “Essay,” with part 1 (three chapters) a succinct statement of his theory of the mechanisms of evolution, and a longer part 2 (seven chapters) the application of his ideas of evolution through natural selection to the, by now, familiar patterns of the biological world (geographic replacement, classification, embryology, unity of type (homology)—and the persistent problems with the fossil record).
It is in the second chapter of part 1 that we see the fateful two words “natural selection” as a subhead of a section that lays out by far his most coherent description of the process to date: “DeCandolle's war of nature—seeing contented face of nature—may be well at first doubted; we see it on borders of perpetual cold. But considering the enormous geometrical increase in every organism and as every country, in ordinary cases, must be stocked to full extent, reflection will show that this is the case. Malthus on man—in animals no moral restraint—they breed in time of year when provision most abundant, or season most favourable….The unavoidable effect of this is that many of every species are destroyed either in egg or young or mature….In the course of a thousand generations infinitesimally small differences must inevitably tell….Nature's variation far less, but such selection far more rigid and scrutinizing” [11].
The bulk of Darwin's 1842 text integrates all he has read in books, monographs, and correspondence about variation, artificial selection, patterns of geographic distribution of animals and plants, and gradation between varieties and distinct species—the main topics of his notebooks. He continues the hypothetico-deductive theme begun in Notebook B—showing that such patterns should be expected as the natural outcome of the evolutionary process.
His “Recapitulation and Conclusion” in the 1842 “Sketch” is a brilliant, impassioned summary of his ideas. It ends with the passage, already adumbrated in Notebook D, that remains virtually identical, not only in 1844 but in the Origin of Species itself: “There is a simple grandeur in the view of life with its powers of growth, assimilation, and reproduction, being originally breathed into matter under one or a few forms, and that whilst this our planet has gone circling on according to fixed laws, and land and water, in a cycle of change, have gone on replacing each other, that from so simple an origin, through the process of gradual selection of infinitesimal changes, endless forms most beautiful and most wonderful have been evolved” [11].
For the most part, Darwin's fervent intellectual search is over with the conclusion of the 1842 “Sketch.” The 1844 “Essay,” at 198 pages, is a much longer manuscript; with exactly the same structure and sequence of topics, it is essentially a smoothed out version of its predecessor—written completely in essay form now, all sentences complete, with no personal notes and queries to interrupt the flow of ideas. The bulk of it, for the most part, consists of vastly more examples bolstering Darwin's points throughout.
That said, the 1844 version of his ideas is far less exciting to read than the 1842 manuscript. It is very much as if the excitement is muted by the sheer bulk of the material reviewed—and probably as much by the fact that the ideas are no longer so novel to Darwin himself. Freshness is lost to familiarity and the sheer weightiness of his verbiage.
Natural Selection and On the Origin of Species
Much the same can be said of Darwin's so-called abstract of his views—The Origin of Species. This, his most famous book, was fresh and new to its readers in 1859, so successful had Darwin been in keeping his views private. But to anyone who has had the privilege of reading the 1830s notebooks, and the early manuscripts (especially 1842), the Origin of Species reads like a mature work in both the best and worst sense of the term. He has honed his arguments beautifully, but the ideas, no longer fresh in his own mind, are just not as enthrallingly expressed as when he was younger and much closer to their inception.
Darwin's Principle of Divergence—poorly understood by modern scholars—melds nascent ecological theory with various models of the origination of new species. Darwin was developing those ideas around the time he started writing his “Big Species Book,” eventually published (second part only) as Natural Selection [2]. And though one shudders at the sheer voluminous nature of this gigantic, yet partial, book, and is tempted to be glad that Darwin pared it down to the more manageably sized Origin of Species, it is true that, at least as far as the discussion of the Principle of Divergence is concerned, the discussion in this last of Darwin's unpublished-in-his-lifetime books is more cogent and complete than the Origin of Species itself. The Origin of Species, one is tempted to conclude, written as it was in such haste, relied heavily on the earlier manuscripts. (Darwin dispensed with the two-part structure, but maintained the same basic sequence of chapters and topics.) In many ways, the unpublished versions hold more rewards than what the Origin of Species—the book that shook the world—offers the modern reader.
There is much more that can be said. Darwin's thoughts about the relative importance of isolation, for example, changed over the years. It is possible with this treasure trove of Darwin's other books to pick themes and trace their development over time. Seldom has the history of ideas, so important as Darwin's evolution by natural selection, been so faithfully preserved as it has in this virtual “fossil record” preserved in this magnificent series of Darwin's other books.
The "Red" and "Transmutation" notebooks (1836-1839), the "Sketch" (1842), the "Essay" (1844), and Natural Selection (1856-1858) are freely available online at http://darwinlibrary.amnh.org
Citation: Eldredge N (2005) Darwin's other books: “Red” and “Transmutation” notebooks, “Sketch,” “Essay,” and Natural Selection. PLoS Biol 3(11): e382.
Niles Eldredge is at the American Museum of Natural History, New York, New York, United States of America. E-mail: [email protected]
==== Refs
References
Darwin C Journal of researches into the natural history and geology of the countries visited during the voyage round the world of H.M.S. Beagle 1839 London Henry Colburn
Stauffer RC Charles Darwin's natural selection 1975 Cambridge Cambridge University Press 692 Being the second part of his big species book written from 1856 to 1858
Darwin C On the origin of species by means of natural selection. Or the preservation of favoured races in the struggle for life 1859 London John Murray 502
Eldredge N Darwin. Discovering the tree of life 2005 New York W. W. Norton 236
Darwin C Herbert S Barrett PH Gautrey PJ Herbert S Kohn D Smith S Red notebook Charles Darwin's notebooks 1987 Ithaca Cornell University Press 17 81 1836 1844
Kohn D Murrell G Parker J Whitehorn M What Henslow taught Darwin Nature 2005 436 643 645 16079834
Barlow N Darwin's ornithological notes Bull Br Mus 1963 2 33 278
Darwin C Kohn D B–E. Barrett PH Gautrey PJ Herbert S Kohn D Smith S Transmutation notebooks Charles Darwin's notebooks 1987 Ithaca Cornell University Press 167 455 1836 1844
Linnaeus C Systema naturae 1758 Stockholm
Kohn D Tauber AI The aesthetic construction of Darwin's theory The elusive synthesis 1996 Amsterdam Kluwer Academic Publishers 13 47
Darwin C Darwin F Sketch The foundations of the origin of species 1842 Cambridge Cambridge University Press (1909) Two essays in 1842 and 1844 by Charles Darwin
Darwin C Darwin F Essay The foundations of the origin of species 1844 Cambridge Cambridge University Press (1909) Two essays in 1842 and 1844 by Charles Darwin
Darwin F The foundations of the origin of species 1909 Cambridge Cambridge University Press 262 Two essays in 1842 and 1844 by Charles Darwin
|
0
|
PMC1283389
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Nov 15; 3(11):e382
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030382
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627755710.1371/journal.pbio.0030392PrimerBioengineeringBioinformatics/Computational BiologyBiophysicsEvolutionSystems BiologyBiochemistryEubacteriaSaccharomycesMotifs, Control, and Stability PrimerDoyle John [email protected] Marie 11 2005 15 11 2005 15 11 2005 3 11 e392Copyright: © 2005 Doyle and Csete.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.
Dynamic Properties of Network Motifs Contribute to Biological Network Organization
Charting the Interplay between Structure and Dynamics in Complex Networks
The interactions of networks of transcription factors and signaling molecules can be understood, in part, through concepts from control theory and engineering.
==== Body
Many of the detailed mechanisms by which bacteria express genes in response to various environmental signals are well-known. The molecular players underlying these responses are part of a bacterial transcriptional regulatory network (BTN). To explore the properties and evolution of such networks and to extract general principles, biologists have looked for common themes or motifs, and their interconnections, such as reciprocal links or feedback loops. A BTN motif can be thought of as a directed graph with regulatory interactions connecting transcription factors to their operon targets (the set of related bacterial genes that are transcribed together). For example, Figure 1A shows a BTN motif that describes a part of the transcriptional response to heat (and other) stressors.
Figure 1 Cartoons of the Escherichia coli HS Control System
(A) The transcriptional motif showing its basic function, the manufacture of chaperones to refold and proteases to degrade denatured proteins. For simplicity, only one operon is shown in detail.
(B) The same network including control elements. See text for explanation.
But biological networks are not just static physical constructs, and it is, in fact, their dynamical properties that determine their function. In this issue of PLoS Biology, Prill et al. [1] show that the relative abundance of small motifs in biological networks, including the BTN, may be explained by the stability of their dynamics across a wide range of cellular conditions. In a dynamical system, control engineers define “stability” as preservation of a specific behavior over time under some set of perturbations. The definitions of stability vary somewhat depending on the types of system, behavior, and perturbation specified [2]. For the BTN example, Prill et al. [1] study stability of gene expression levels, as modeled by a set of linear differential equations. Given interactions from a BTN motif, “structural stability” is robustness of stability to arbitrary signs and magnitudes of interactions. This is such a stringent notion of stability that it would be satisfied by few systems, yet Prill et al. [1] show that all BTN motifs are stable for all signs and magnitudes of interactions. For several other biological networks, they show a level of correlation between abundance and structural stability that is highly unlikely to occur at random. The significance of these results as well as those in recent related papers (see references in [1], particularly those of Alon and colleagues) can be better appreciated within the larger context of well-known concepts from biology and engineering, particularly control theory [3]. For additional mathematical details underlying the qualitative arguments presented here, see the online supplement (Text S1 and S2).
Motifs, Networks, and Dynamics
Prill et al. [1] point out that their motifs are small parts of networks in at least two distinct senses, and the heat shock (HS) response will be used to illustrate these points. A motif such as that shown in Figure 1A is just one of many motifs that make up the BTN, but is also part of an even larger network involving protein–protein interactions (PPI) illustrated in the more detailed Figure 1B [4]. The HS response ultimately works on proteins—repairing or degrading misfolded proteins before they damage the cell. The small motif cartoon of Figure 1A consists of the rpoH gene, which encodes a transcription factor called the alternative sigma factor σ32, which recognizes the HS gene promoters to induce HS-specific gene expression. For simplicity, only one operon is shown.
HS genes encode molecular “chaperones” (such as DnaK)—proteins that help refold denatured proteins—and “proteases” (such as Lon)—enzymes that degrade unfolded, dysfunctional proteins. Regulatory aspects of this motif are shown schematically in Figure 1B. Briefly, in addition to binding unfolded proteins, chaperones can also bind to σ32 (denoted in Figure 1 as σ), sequestering and preventing it from binding with RNA polymerase (RNAP), thus providing a negative feedback loop to modulate σ32 activity. The protease FtsH degrades bound σ32, a negative feedback that further fine-tunes the HS response. A feedforward response is implemented in the heat sensitivity of σ32 translation, which is enhanced at high temperatures. These additional layers of control beyond the BTN motif alone yield a system that by engineering standards is efficient, robust, and evolvable [4]. (Note: biologists use the term “regulatory” to describe networks such as that displayed in Figure 1A, but engineers typically reserve “regulatory” for actual controlling elements as in Figure 1B.)
The motif in Figure 1A is a simple tree and is perfectly structurally stable, with its stability completely independent of the specific concentrations and kinetics of the individual molecules that compose the network [1]. In contrast, by almost any reasonable definition (including appropriately generalizing the methods of Prill et al.), Figure 1B has no structural stability, as only a small subset of possible parameter values could confer a stable network. The small motif in Figure 1A is, thus, inherently stable, but Figure 1B requires a high level of fine-tuning for stability—which, in fact, has evolved for this network. An essentially parallel story holds for other motifs, as all motifs in the BTN are structurally stable. Indeed, the entire BTN from which the motifs were extracted (without the PPI elements) is perfectly structurally stable, since it has no nontrivial feedback loops (i.e., other than self-loops, where a protein regulates its own synthesis). And to the extent that analogous PPI dynamics are known for other motifs, they too require exquisite fine-tuning for stability.
Figure 2 Pendulum in Up and Down Positions
(A) Pendulum held in up (unstable) position
(B) Pendulum held in down (stable) position
The fact that the bacterial “transcriptional networks” have such strong structural stability and that this stability is completely lost with the inclusion of protein–protein and other regulatory interactions has many possible and quite different interpretations. Structural stability is clearly not an intrinsic feature of the biology itself, but depends in a rather extreme way on the level of detail in the models chosen. Thus, based on the biology alone and the many caveats that define the way these motifs were extracted, the results based on structural stability in Prill et al. might appear to be at best speculative, and at worst misleading. We will argue, however, that their results reveal highly significant organizational principles.
Plants, Controllers, and Disturbances
Control theory uses an abstraction that is useful in interpreting biological models like those in Figure 1 ([2]; Text S1 and S2). A system like Figure 1B is often decomposed into a “plant” (as in manufacturing plant), from which the basic function of the system can be inferred, and a “controller,” which implements feedback and feedforward manipulations to improve stability and robustness of this function. In this sense, robustness means that a specific plant function (such as low levels of unfolded proteins) is maintained in the face of certain disturbances (such as temperature or chemical insults). Robustness is usually used in a broader sense than stability, with the latter usually restricted to infinite time horizons and the former including additionally transient behavior and wider ranges of perturbations. Thus, “robust stability” is typically used to describe stability that is robust to some large set of perturbations. For example, in Figure 1A, heat can be viewed as the external disturbance on a plant consisting of folded and unfolded proteins plus chaperones and proteases. The controller in Figure 1B adds feedback and feedforward mechanisms to enhance robustness and efficiency in the control of unfolded protein levels, particularly in transient response to temperature change [4].
Many organizational features of this system have been experimentally studied. Removing σ32 (by creating bacteria lacking the rpoH gene) causes death of bacterial cells at high temperatures [5]. But this lethal knockout can be rescued by constitutive overexpression of the HS operons, essentially implementing the motif in Figure 1A as the complete system [6]. Thus, in principle, a network with the topology in Figure 1A is viable, provided it is implemented appropriately. The controller in Figure 1B enhances robustness and efficiency but is not required for basic function, and other less complex control schemes could be used, but with degraded robustness and performance compared to wild type (nonmutants) [4]. Again, to the extent that details are known, this is a common story for BTN motifs in general [7], in which a transcriptional motif provides a core plant that performs some basic function. The actual biological network, however, often has a controller involving PPI or other mechanisms, typically of much greater complexity than the motif itself, and which provides additional robustness, efficiency, and flexibility. In engineering design, controller or plant decompositions can be nested, with one plant plus controller collectively functioning as the controller for another plant, and so on. Far more complex layered control strategies are commonly used in designing technological systems, and are presumably ubiquitous in biology. For example, the hierarchical organization of bacterial regulation includes such elements as stimulons, modulons, regulons, operons, PPI control elements, etc [7].
Such decompositions are not unique, and plant and controller can, in principle, be arbitrarily chosen components, but the choices are typically used to highlight particular organizational features of the complete system. For example, Figure 1B can, instead, be decomposed into a plant consisting of just the folded or unfolded protein levels of the cell, with heat disturbance, and a controller consisting of the entire bottom part of Figure 1B. This decomposition is more natural from an engineering perspective, but highlights the BTN motif less than when Figure 1A is viewed as the plant plus disturbance. Note that because the plant motifs in the BTN involve only transcription, they typically must have much slower dynamics than the controller. For example, the slowest dynamic in Figure 1B is the synthesis of the HS operon from the plant in Figure 1A. Indeed, the rest of the controller is implemented entirely in the relatively faster PPI and σ32 translation, while transcription of rpoH is not regulated and so does not contribute to the dynamics of the controlled system [4].
The Implications of Structural Stability
The lack of structural stability in full systems of plant and controller may lead to the speculation that they are not robust. In fact, all complex, controlled systems lack structural stability when viewed at the full system level with controller dynamics included. A complete answer to this apparent paradox is a large subject in its own right, but some aspects are easily explained. One basic point is that the signs—whether interactions are activating (positive) or inhibiting (negative)—in most biological and technological networks must be fine-tuned, but once the signs of constants are appropriately fixed, their absolute values can often vary substantially with little effect. The number of different sign combinations in n constants grows exponentially as 2n; thus, one (fine-tuned) choice of signs becomes a vanishingly small fraction of the total number of possibilities in any sufficiently large network. In other words, if signs are important, and they are in control systems, the resulting network cannot be structurally stable. It is also true that in both technology and biology it is much easier to manufacture components with robustly fixed signs than with precise absolute values.
Thus, while structural stability as defined in Prill et al. may be too strong of a notion to be helpful in distinguishing between different control systems, this very strength further underscores the significance of the authors' results. All the plant motifs in the BTN, indeed, the entire BTN plant itself, are not merely stable, but have extremely strong structural stability. But the necessarily fine-tuned controller can stabilize these plants, and, furthermore, unstable plants are common in biology and technology, so the absence of unstable plants here still needs further explanation.
The Costs of Plant Instability
Perhaps the most relevant concept from control theory is that unstable plants are intrinsically more difficult to control than stable ones, and are generally avoided unless the instability confers some great functional advantage, which it often does [2,3]. A classic illustration of instability and control, the simple inverted pendulum experiment, can be easily tried at home, and illustrates the essential point without the mathematical details. Here the pendulum is the plant, and the human is the controller. The experiment can be done with sticks of different lengths or with an extendable pointer, holding the proximal tip between thumb and forefinger so that it is free to rotate but not otherwise slip. With the controlling hand fixed, this system has two equilibria, down and up, which are stable and unstable, respectively. By watching the distal tip and controlling hand motion, the up case can be stabilized if the stick is long enough. For an external disturbance, imagine that there is a virtual object making small motions in the vicinity of the distal tip, and your goal is to move the hand in such a way as to track this motion.
You will soon find that it is much easier to control the distal tip down than up, even though the components in both cases are the same. Because the up configuration is unstable, certain hand motions are not allowed because they produce large, unstable tip movements. This presents an obstacle in the space of dynamic hand movements that must be avoided, making control more difficult. If you make the stick shorter, it gets more unstable in the up case, evident in the short time it takes the uncontrolled stick to fall over. Shorter pendulums get harder and ultimately impossible to control in the up case, while length has little such effect on the down case. Also, the up stick cannot be stabilized for any length if only the proximal tip is watched, so the specific sensor location is crucial as well. This exercise is a classical demonstration of the principle that the more unstable a system the harder it is to control robustly, and control theory has formally quantified this effect in several ways (Text S1).
With this general context, a plausible conjecture is that the stability of the slow dynamics in the BTN plant is there, in part, to make control easier. It is the controller in Figure 1B that must be robust to the plant and disturbance in Figure 1A, not the other way around. Yet as the pendulum example illustrates, the stability of the plant can have a large impact on the achievable robustness of the controller. The BTN plant stability could additionally be a consequence of evolutionary constraints, in that the slow dynamics may have existed first (such as Figure 1A), and control was later layered (as in Figure 1B). If the slow dynamics are simply vestiges of an original, uncontrolled, and structurally stable network, their preservation, even by accident, still facilitates the full system level control. Perhaps such preservation could be the result of selective pressure on this system for robustness.
A Place for the Unstable Plant
While unstable plants are difficult to control, they are used when function requires it. Modern rockets are unstable in a manner similar to the up pendulum, and must be stabilized by active control systems. Toy rockets or fireworks without active control create stable plants by using large fins or tails that passively move the centers of pressure and gravity to make the dynamics more like the down pendulum, but at the expense of greater drag. Technology has abundant examples where similar efficiency and performance trade-offs lead to unstable plants with actively stabilizing controllers. Even bacteria have systems with unstable components that are nevertheless combined into feedback systems that are stable and robust.
Chemotaxis, cell movement toward a chemical attractant, is an example whereby an uncontrolled plant consisting of only cell and flagella would move essentially randomly, and, thus, would not be stable in any conventional sense. Yet with the full signal transduction system in place, the controlled runs and tumbles are biased to create effective chemotaxis, apparently using strategies common in engineering [8]. Glycolysis is often drawn as a “molecular motif” as in Figure 3A, without loops and therefore structurally stable, and the relationship between Figure 3A and the larger controlled Figure 3B is even more subtle than between Figure 1A and 1B. Figure 3B shows both positive autocatalytic feedback of ATP needed to fuel glycolysis and negative regulatory feedback, a combination that when sufficiently perturbed can lead to well-known instabilities, even with the control system intact [9]. These complex features of HS, chemotaxis, and glycolysis may not be accidental, but may be necessary consequences of unavoidable trade-offs, and as briefly sketched here, control theory supports this notion. Perhaps more persuasive is that these are three of the most thoroughly studied small networks in biology, and apparently no one has found alternatives, even theoretically, that convincingly improve on the efficiency and robustness of the wild type networks.
Figure 3 Cartoons of a Generic Glycolysis Network in Bacteria
(A) A simple graph showing the main metabolites of glycolysis and their relationships.
(B) The same metabolites but also including reactions (solid blue lines), autocatalytic feedbacks (solid purple lines), and regulatory feedbacks (dotted black lines).
BTN motifs exhibit an extremely strong version of structural stability. Yet because of the organization involving both plant and controller, this apparently severe restriction on the BTN plants does not necessarily create a correspondingly severe constraint on function. For example, basic function of the plant motif in Figure 1A of manufacturing HS proteins is simple enough that only a minimal network is needed. Most of the network complexity is in the controller in Figure 1B, and to maximize the speed of the HS response, it is important to minimize the effects of any additional transcriptional events, which implies the plant must be kept simple. The result is that the BTN network as a whole is very flat with few long paths, which we conjecture by analogy to HS, allows the controlled system to have rapid response [10]. Perhaps a constructive next step would be to systematically contrast the strongly stable BTN plants with the less stable plants in chemotaxis and glycolysis. From an engineering perspective, all of these well-studied bacterial networks appear highly efficient and robust, tolerating trade-offs to achieve this well-engineered state [11]. And now, Prill et al. put at least one feature of bacterial transcriptional network motifs, their structural stability, into a much broader context.
Supporting Information
Text S1 Supplementary Notes: Elementary Feedback Concepts
(154 KB PDF).
Click here for additional data file.
Text S2 Feedback Control Theory
(4.2 MB PDF).
Click here for additional data file.
Citation: Doyle J, Csete M (2005) Motifs, control, and stability. PLoS Biol 3(11): e392.
John Doyle is in the Department of Control and Dynamical Systems, California Institute of Technology, Pasadena, California, United States of America. Marie Csete is in the Departments of Anesthesiology and Cell Biology, Emory University, Atlanta, Georgia, United States of America.
Abbreviations
BTNbacterial transcriptional regulatory network
HSheat shock
PPIprotein–protein interactions
==== Refs
References
Prill R Iglesias PA Levchenko A Dynamic properties of network motifs contribute to biological network organization PLoS Biol 2005 3 e343 10.1371/journal.pbio.0030343 16187794
[Anonymous] Control theory 2005 St. Petersburg (Florida) Wikipedia Available: http://en.wikipedia.org/wiki/Control_theory#Stability . Accessed 21 September 2005
Doyle JC Francis BA Tannenbaum A Feedback control theory 1992 New York Macmillan Publishing Available: http://www.hot.caltech.edu/dft.pdf . Accessed 21 September 2005
El-Samad H Kurata H Doyle JC Gross CA Khammash M Surviving heat shock: Control strategies for robustness and performance Proc Natl Acad Sci U S A 2005 102 2736 2741 15668395
Yamamori T Yura T Genetic-control of heat-shock protein synthesis and its bearing on growth and thermal-resistance in Escherichia coli K-12 Proc Natl Acad Sci U S A 1982 79 860 864 7038687
Baker TA Grossman AD Gross CA A gene regulating the heat shock response in Escherichia coli also affects proteolysis Genetics 1984 81 6779 6783
Lengeler JW Drews G Schlegel HG Biology of the prokaryotes 1999 Malden (Massachusetts) Blackwell Science 955
Yi TM Huang Y Simon MI Doyle J Robust perfect adaptation in bacterial chemotaxis through integral feedback control Proc Natl Acad Sci U S A 2000 97 4649 4653 10781070
Heinrich R Montero F Klipp E Waddell TG Melendez-Hevia E Theoretical approaches to the evolutionary optimization of glycolysis Eur J Biochem 1997 243 191 201 9030739
Yeger-Lotem E Sattath S Kashtan N Itzkovitz S Milo R Network motifs in integrated cellular networks of transcription–regulation and protein–protein interaction Proc Natl Acad Sci U S A 2004 101 5934 5939 15079056
Csete ME Doyle JC Bow ties, metabolism, and disease Trends Biotechnol 2004 22 446 450 15331224
|
16277557
|
PMC1283396
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Nov 15; 3(11):e392
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030392
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627755810.1371/journal.pbio.0030394Community PageEcologyAnimalsPlantsLinking Biodiversity Conservation and Livelihoods in India Community PageShanker Kartik [email protected] Ankila Bawa Kamal 11 2005 15 11 2005 15 11 2005 3 11 e394Copyright: © 2005 Shanker, 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.The Ashoka Trust for Research in Ecology and the Environment was established in 1996 to curtail the rapid loss of India's biological resources and natural ecosystems.
==== Body
In a country like India, millions of people rely on products from natural ecosystems to sustain their livelihood. Natural ecosystems provide clean water from watersheds, retention of soil and soil fertility, sequestration of carbon, as well as pollinators and natural predators of pests. For these reasons, rapid, often irreversible, loss of species and ecosystems is of more than just academic concern. Indeed, our understanding of biodiversity in natural ecosystems remains so woefully inadequate that we are unable to fully comprehend the consequences of its loss. With impending climate change and increasing spread of invasive species, the biodiversity crisis is likely to get worse, with far-reaching effects on human societies.
To meet these challenges, we need institutions and scholars that can generate new knowledge, and apply it to resolve our most pressing environmental issues. The Ashoka Trust for Research in Ecology and the Environment (ATREE) (http://www.atree.org) was established in 1996 to curtail the rapid loss of India's biological resources and natural ecosystems, and to address the environmental, social, and economic dimensions of this decline. We highlight below two examples of ATREE's work from very different ecosystems and regions.
Tribals and Non-Timber Forest Products in the Western Ghats
The Biligiri Rangaswamy Temple Wildlife Sanctuary (BRT), a 540-square-kilometer protected area, forms a part of India's Western Ghats, one of the global hotspots of biodiversity. The area has traditionally been inhabited by an indigenous community, the Soligas, and is also a habitat for a number of endangered plants and animals. Soligas have harvested forest products for centuries for their own use and more recently for markets. The interrelated issues of livelihood enhancement of the Soligas and biodiversity conservation have been at the heart of ATREE's work in BRT for close to a decade—along with a partner non-governmental organization, the Vivekananda Girijana Kalyana Kendra; a local community organization, the Soliga Abhivrudhi Sangha; and the Karnataka Forest Department [1]. A detailed understanding of the drivers that cause forest loss and degradation is the first step toward its preservation.
The forestland of India's Western Ghats is a target for conservation, which must involve local communities
(Photo: Smitha Krishnan)
We have used demographic models to analyze changes in population structure of nelli (Phyllanthus emblica and Phyllanthus indofischeri), one of the most important NTFPs in BRT. Nelli is an edible fruit high in vitamin C, extracts from which are key ingredients in traditional Indian medicine, and in cosmetics. Our results indicate that population growth rates, on average, are close to rates that would allow full replacement of individuals. Moreover, it is not harvest, per se, but rather the method of harvest involved (e.g., whether or not it involves the lopping off of branches or cutting of small trees) and the spread of Loranthus, a plant parasite that infests mature trees, which affects population growth [2]. Management efforts, therefore, need to focus on control of the parasite, and on the use of nondestructive harvest techniques.
Discussion of harvest techniques and parasite removal form part of the annual participatory monitoring meetings with Soliga harvesters. These participatory monitoring meetings also focus on the temporal and spatial patterns in availability of various NTFPs, which are then used to guide harvest decisions. Information collected as part of this participatory monitoring program corroborate the results of scientific monitoring, namely, that populations of nelli, and also those of other important NTFP species, such as the Asian honeybee (Apis dorsata), a source of honey, have remained relatively stable over the last ten years.
It is essential that the benefits accruing to harvesters be maximized for their continued participation in management and conservation. A large part of our effort to strengthen existing institutions has, therefore, been directed toward reform of the Large-Scale Adivasi (tribal) Multi-Purpose Society, a government-established cooperative society, and a key element in the success of efforts at forest conservation. Non-timber products harvested from the forest can only be sold to the Large-Scale Adivasi Multi-Purpose Society. ATREE is also working with Soliga farmers to increase agricultural productivity, enhance on-farm diversity, and improve soil and water conservation. It is hoped that through simple agricultural interventions we can achieve a greater on-farm contribution toward subsistence and cash needs, thereby reducing the extent of dependence on NTFP.
BRT is the only forest area in India where production and extraction of NTFPs are being monitored, and where the local community is involved in such monitoring. In a recent meeting with the Forest Department, a committee comprising members of the Soliga community, Vivekananda Girijana Kalyana Kendra, and ATREE was proposed to provide suggestions to the Forest Department on management of the protected area. If formalized, this would make BRT the first protected area to have such a three-way collaboration between managers, the local community, and researchers, and would be a model for other protected areas in the country.
Protecting Turtles with Fisherfolk
On the other side of the country, on the coast of Orissa, are the nesting grounds of olive ridley sea turtles (Lepidochelys olivacea). This is one of three rookeries worldwide where arribadas, the synchronous mass nesting of thousands of ridley turtles, occurs. Genetic studies demonstrate that this is a unique population, which may be ancestral to olive ridleys in other ocean basins [3]. In the last decade, however, 10,000 turtles have been counted dead on the Orissa coast each year due to fishery-related activities. And many more are likely to have died, since not all turtles killed in nets are washed ashore [4].
Research indicates that olive ridleys remain in small offshore congregations during the breeding season, and are not diffusely distributed along the entire coast. Discrete reproductive patches off the mass nesting beaches are usually no more than 50 square kilometers; however, the location of these patches may vary over time [5]. Thus, the creation of sanctuaries or protected areas with fixed boundaries may not be effective. Instead, conservationists have focused on the enforcement of laws, including existing fishery regulations such as the 1983 Orissa Marine Fisheries Regulation Act, which stipulates that mechanized fishing is prohibited within five or ten kilometers of the coast, depending on boat size.
Despite the investment of large amounts of effort and funds by the government and civil society groups to patrol nearshore waters, trawlers continue to fish illegally in nearshore waters, causing continuing mortality of olive ridleys. Rather, the anti-trawler programs, coupled with the media coverage, have severely polarized fishing communities and conservation groups. Even traditional fish-worker associations in Orissa joined in protests with the trawler owners, since they perceived turtle conservation as being anti-people, even though most of the Orissa Marine Fisheries Regulation Act regulations were designed to protect traditional fishing rights rather than turtles.
In fact, if fishery laws had been enforced for the reasons that they were originally instituted, namely, to protect traditional fisherfolk and their livelihoods, it is likely that their implementation would have received far wider support. And as a result, sea turtles would then have been protected from mechanized fishing. To achieve this goal, the Coastal and Marine Programme at ATREE created and facilitated a common platform for sea turtle conservation in Orissa. In December 2004, ATREE organized a meeting in Bhubaneshwar that was attended by key fish-worker organizations (Orissa Traditional Fish Workers' Union and United Artists Association), local community organizations, and non-governmental organizations such as Project Swarajya, Wildlife Society of Orissa, Worldwide Fund for Nature, Greenpeace, and others. The group named itself the Orissa Marine Resources Conservation Consortium, and has been working together to achieve common marine conservation goals.
The Orissa Marine Resources Conservation Consortium has held numerous follow-up meetings in 2005. The activities identified to meet these common goals included meetings for fisherfolk representatives and other stakeholders on fisheries management and turtle conservation legislation in April 2005. ATREE has produced illustrative local language booklets on fisheries conservation and turtle protection measures in Orissa to help local communities understand their rights and regulations. Booklets, along with other visual aids such as posters illustrating fishing regulations, have been distributed to various stakeholders at the mass nesting beaches. The Orissa Marine Resources Conservation Consortium plans to promote community-based marine conservation practices and appropriate environmentally sustainable coastal development, addressing the issues of marine biodiversity and resource use.
The Big Picture
With its headquarters in Bangalore, ATREE also has offices in Delhi, and northeast India, where its programs are directed toward conservation of the eastern Himalayan region. ATREE's current activities are grouped under research and action, education, and outreach. Specifically, ATREE uses interdisciplinary approaches to (1) generate knowledge that fosters conservation and judicious management of biodiversity, (2) provide the best scientific information to policymakers, (3) design management systems that emphasize decentralization, fairness, and equity in the use of resources by civil society, (4) organize and disseminate information for conservation and sustainable use of biodiversity, and (5) train a new generation of leaders to meet current challenges in biodiversity conservation and environmental protection.
The various activities are organized under three centers: the Center for Conservation Science; the Center for Conservation, Governance, and Policy; and the Center for Ecoinformatics (http://www.ecoinfoindia.org). Most of the existing staff work in diverse areas of natural and social sciences including biodiversity characterization, forest ecology, taxonomy, conservation genetics, landscape ecology, hydrology, environmental sociology, and ecological economics, in the Center for Conservation Science. All three centers are, or would be, engaged in education and outreach activities designed to build the capacity of academic, governmental, and non-governmental organizations to meet the growing list of contemporary environmental challenges.
Citation: Shanker K, Hiremath A, Bawa K (2005) Linking biodiversity conservation and livelihoods in India. PLoS Biol 3(11): e394.
Kartik Shanker, Ankila Hiremath, and Kamal Bawa are at Ashoka Trust for Research in Ecology and the Environment, Bangalore, India. Kamal Bawa is also at University of Massachusetts, Boston, United States of America.
Abbreviations
ATREEAshoka Trust for Research in Ecology and the Environment
BRTBiligiri Rangaswamy Temple Wildlife Sanctuary
NTFPnon-timber forest product
==== Refs
References
Bawa KS Prescriptions for conservation Biodiversity Conservation Network 1999 Washington (D.C.) Biodiversity Support Program 48 55 Final stories from the field
Sinha A Bawa KS Harvesting techniques, hemiparasites and fruit production in two non-timber forest tree species in south India For Ecol Manage 2002 168 289 300
Shanker K Pandav B Choudhury BC An assessment of the olive ridley turtle (Lepidochelys olivacea ) nesting population in Orissa, India Biol Conserv 2004 115 149 160
Shanker K Ramadevi J Choudhury BC Singh L Aggarwal RK Phylogeography of olive ridley turtles (Lepidochelys olivacea ) on the east coast of India: Implications for conservation theory Mol Ecol 2004 13 1899 1909 15189212
Pandav B Conservation and management of olive ridley sea turtles on the Orissa coast [thesis] 2000 Bhubaneshwar (India) Utkal University 106 Available from Wildlife Institute of India, Dehradun, India
|
16277558
|
PMC1283397
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Nov 15; 3(11):e394
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030394
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627755910.1371/journal.pbio.0030395EssayCell BiologyAnimalsThe Cell Nucleus and Aging: Tantalizing Clues and Hopeful Promises EssayScaffidi Paola Gordon Leslie Misteli Tom [email protected] 2005 15 11 2005 15 11 2005 3 11 e395Copyright: © 2005 Public Library of Science.2005This is an open-access article distributed under the terms of the Creative Commons Public Domain declaration, which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose.Recent evidence links structural proteins in the cell nucleus with aging.
==== Body
There are a handful of biological questions that affect all of us directly in everyday life. How are emotions formed, what is the basis for consciousness, and why do we look the way we do? One that strikes particularly close to home is the question of how we age. The sheer complexity of this problem has had many scientists throw up their hands in frustration and most of the postulated theories have been vague and generally have involved ill-defined wear-and-tear mechanisms. But the pursuit of the biological basis of aging has been revitalized within the last decade by studies in yeast, worms, flies, and mice that have firmly established that there indeed exist specific molecular mechanisms that contribute to the aging process [1]. These efforts point to several distinct, likely interrelated, mechanisms, ranging from improper protein metabolism, to alterations of specific signaling pathways, progressive damage due to generation of oxidative free radicals, and increased genome instability.
Although much has been learned about the aging process from simple model organisms, one intuitively suspects that things might be somewhat different when it comes to human aging. So how does one best study the molecular basis of human aging? The answer might be premature aging diseases, or progeroid syndromes. The advantage of these often rare diseases is that they are mostly monogenic and thus experimentally tractable. On the other hand, one should keep in mind that such disorders usually only mimic some of the features of normal aging and it can be difficult to distinguish true aging symptoms from unrelated developmental defects. Regardless, it appears that progeroid syndromes may be legitimately used as model systems to investigate the physiological processes contributing to aging. In fact, the study of some of these diseases has recently brought tantalizing clues as to how we age. One of the most intriguing ones is a possible involvement of structural components of the cell nucleus [2].
Aging and the Cell Nucleus
The cell nucleus in higher organisms is now recognized as a complex, highly organized repository of an individual's genetic information. The typical nucleus contains distinct functional neighborhoods made up by nonrandomly positioned chromosomes and proteinaceous subcompartments in which specific processes, including gene expression, occur [3]. Which molecular mechanisms establish and maintain the structural integrity of the nucleus is largely unknown and represents one of the most exciting fields in modern cell biology. Having said that, one of the major structural elements of the nucleus, the nuclear lamina, has been known and studied for decades [2]. The lamina is made up of A-, and B-type lamins, which are intermediate filament proteins that form an interwoven network situated at the very periphery of the nucleus underlying the nuclear membrane. This structure has long been thought to act as a shield to protect the genome from mechanical stress. More recently, this architectural feature of the nucleus has also been recognized as potentially playing a regulatory role in gene expression because lamins interact with chromatin and might serve to anchor and organize genome regions in space [2].
The first hint to a surprising connection between nuclear architecture and aging came from yeast, when Leonard Guarente and colleagues found that a protein, Sir4, whose mutation results in extension of life span, localizes to the nucleolus, one of the most prominent subcompartments of the cell nucleus [4]. The link between aging and nuclear organization was further strengthened by the observation that the localization of the protein and the morphology of the nucleolus itself changed as yeast cells aged. However puzzling and provocative these observations were, it was unclear whether these structural reorganizations were a cause or consequence of aging, and it seemed a stretch to imagine that the same mechanisms might apply to human cells.
Hutchinson-Gilford Progeria Syndrome
The definitive proof for a causal connection between nuclear architecture and human aging came with a stunning discovery in the summer of 2003, when the groups of Francis Collins and Nicolas Levy identified mutations in the lamin A gene (LMNA) as the genetic cause of the segmental premature aging disease Hutchinson-Gilford progeria syndrome (HGPS) [5,6] (Box 1). Children with HGPS usually experience normal fetal and early postnatal development but die of severe atherosclerosis at an average age of 13 years [7]. The initial physical signs of HGPS include severe failure to thrive, heralding severe lipoatrophy, bony abnormalities, a small, beaked nose and receding mandible, complete hair loss, and speckled hypopigmentation with some areas of tight, hard skin. As the disease progresses, vascular plaques become pervasive, leading to strokes and heart attacks. In short, these children give the distinct physical impression of being many decades older that they really are. However, HGPS children are neurologically unaffected, so that their emotional and developmental stages are age-appropriate. A six-year-old with HGPS may look physically like an old person, but is ready to enter first grade along with the rest of his or her peers (Figure 1). In a sense, for families and friends, having a child whose mind and personality progress normally is a great fortune. In another sense, watching a child whose mind is so full of promise and joy experience angina, strokes, and heart attacks that are usually reserved for the elderly is devastating.
Box 1. The Cell Biology of HGPS
In more than 80% of cases the gene defect responsible for HGPS is a single spontaneous mutation in codon 608 of the LMNA gene, which encodes both lamin A and lamin C [6]. This base change produces a silent amino acid mutation and would therefore have no consequences for the lamin A protein. Unfortunately, the mutation activates a cryptic splice site in LMNA, resulting in aberrant removal of a part of the LMNA RNA during the RNA splicing reaction and generation of a protein lacking 50 amino acids towards its C-terminal end. This mutant protein is referred to as progerin. The progerin protein appears to act in a dominant negative fashion, and although its modus operandi is unclear at the moment, its action seems to be related to its extensive post-translational modifications. Normal lamin A is produced as a pre-lamin A protein that undergoes a complex set of modifications starting with carboxymethylation, followed by cleavage of the terminal three amino acids, farnesylation at the C-terminus, and subsequent proteolysis of its terminal 15 amino acids, leading to the removal of the farnesyl group. Although it can be farnesylated, pre-progerin lacks the endoprotease recognition site necessary for executing the final cleavage step and thus accumulates in a farnesylated form [19]. Since the farnesyl group is important for the protein's localization to the nuclear periphery, progerin accumulates in the lamina and this is presumably where it exerts its negative functional effects. Consistent with that notion, loss of the protease responsible for the cleavage event leads to premature aging in mice and humans [20,21]. Given the critical importance of the lamins in nuclear architecture, it is not surprising that the cells of HGPS patient are characterized by dramatic aberrations in nuclear architecture, particularly changes in the formation of their shape, loss of internal heterochromatin, and changes in the distribution of numerous nuclear proteins [14,22].
The Molecular Basis of Nuclear Defects in HGPS
The involvement of lamin A in this disease was initially puzzling. It is not apparent how a protein whose function is to maintain proper nuclear architecture may cause premature aging. Two hypotheses rapidly emerged based on what cell biologists had learned about this intensely studied protein in the last two decades. It could be that the mutant lamin A protein weakens the nuclear lamina and in that way reduces the resistance of the cells of HGPS patients to the types of mechanical stress encountered in the body, heralding cellular dysfunction and death. This explanation makes much sense since many of the primarily affected tissues, such as skin and vasculature, are under intense mechanical stress. On the other hand, recent work showing that lamins directly and indirectly interact with chromatin point to the possibility that changes in the lamina might lead to misregulation of sets of genes in HGPS patients. Although plausible, it is not obvious how the gene misregulation model can account for the aging effect seen in patients, particularly since it has so far proven difficult to identify sets of genes that are commonly misregulated in these patients. Even the mechanical stress model is not completely satisfying since it is based on one of those ill-defined wear-and-tear scenarios so often invoked in the aging process. The distinct possibility that aspects of both models might apply is suggested by the fact that several genes that respond to mechanical signal transduction via the cytoskeleton are affected in cells lacking lamin A [8].
A possible breakthrough in understanding how a structural protein of the cell nucleus may be a key player in aging came just a few months ago. Zhang and colleagues discovered that cells in HGPS patients have increased genomic instability and DNA damage and that the DNA repair machinery does not function as effectively as it does in healthy individuals [9]. In particular, it appears that in the cells of individuals with HGPS, the repair machinery is not as efficiently recruited to sites of damage, and, consequently, DNA breaks are less efficiently repaired. Although unproven at this point, one can imagine that defects in the lamina structure could prevent the efficient capture of repair factors [9,10]. Admittedly, increased genome instability must also be considered a wear-and-tear mechanism, but it deserves more credence than some of the other mechanisms because genome instability is a feature shared amongst virtually all premature aging disorders.
Genomic Instability and Aging
The clearest link between genomic instability and accelerated aging comes from a different progeroid disease, Werner syndrome (WS). This is an autosomal recessive disorder resulting from loss of function of a DNA helicase, the WRN RecQ helicase [11]. Compared to HGPS, this condition has a later onset, with patients developing normally until puberty and showing aging symptoms in early adulthood. Death generally occurs in the fifth decade of life, mainly because of cancer and cardiovascular disease. Extensive research over the past decade has shed light on the molecular and cellular defects underlying the disease phenotype, making WS a perfect example of how important cell biology can be in unraveling mechanisms of premature aging. The WRN protein is involved in DNA replication and recombination and its main function is to reinitiate stalled replication forks. In the cells of WS patients, the absence of the WRN protein results in defective replication, inefficient DNA repair, and chromosome rearrangements. Thus, altered DNA metabolism in WS cells directly causes defective maintenance of genome integrity, and this, in turn, results in cancer predisposition in patients [11].
The complex nature of the interplay between nuclear architecture, DNA metabolism, and aging becomes clear when considering a subset of patients who were diagnosed with WS based on their clinical symptoms, but, puzzlingly, do not carry a mutation in the WRN gene—instead, they have mutations in LMNA [12]. Clearly, the fact that WS symptoms can be recapitulated by mutations in a gene that gives rise to another premature aging disease is hardly coincidental. This remarkable confluence of observations reinforces the idea that HGPS and WS might share a similar underlying molecular mechanism, possibly genomic instability.
Accelerated aging, however, probably does not involve only genomic instability. WS has also been critical in revealing a potential role of cellular senescence in the aging process [13]. Cells entering senescence, i.e., permanent arrest of cell division, undergo phenotypic changes and display several functional abnormalities compared to their proliferating counterparts. A second hallmark of cells in WS patients is short replicative life span in culture, due to telomere dysfunction. Although the molecular details of how the WRN protein affects telomere metabolism are not fully understood, the picture emerging is that accumulation of senescent, dysfunctional cells in WS patients might compromise the effectiveness of tissues and organs, leading to accelerated aging. Extending this model, cellular senescence has been postulated to also play a key role in physiological aging.
These observations on WS and the occurrence of genomic instability in premature aging diseases crystallize a tantalizing antagonism between aging and cancer [13]. That these two are linked is clear from the fact that age is the single largest risk factor for the development of cancer. In addition, genomic instability as seen during aging is a hallmark of cancer cells. Paradoxically, what we have learned from WS suggests that increased senescence appears to lead to premature aging, but on the other hand, cellular senescence is now also recognized as a defense mechanism to effectively stop potentially harmful, cancer-forming cells from proliferating. Clearly, organisms must find a fine balance between these opposing effects. How this equilibrium is achieved during the physiological aging process is one of the fascinating mysteries of aging.
Finding the Fountain of Youth
One of the lures of aging research is of course to find a fountain of youth—an elixir that prolongs life. Although several commercial entities are unleashing their resources at this tempting goal, given the inconclusive nature of most molecular aging pathways currently under consideration, these efforts seem unlikely to come to fruition in the near future. However, therapeutic intervention may be realistic for premature aging disorders as their molecular basis is becoming well defined. The feasibility of anti-aging intervention is made clear in the case of HGPS. Since the discovery of its genetic cause less than two years ago, at least two fascinating, yet realistic, strategies of molecular treatment have been opened. The first is based on correcting the primary defect at the level of pre-mRNA. This strategy involves correcting the aberrant splicing event caused by the HGPS mutation (Box 1). Proof of principle that this is possible has recently been provided by introduction of small oligonucleotides specifically targeted to the HGPS splice site, preventing its use and restoring normal splicing and cellular behavior [14]. If these results from cell culture experiments can be extended to organisms, they might provide a promising avenue of intervention. A second potential therapeutic approach for HGPS is founded in the complex post-translational processing of the lamin A protein by farnesylation (Box 1). Inhibitors of the farnesylation reaction, so called farnesyl transferase inhibitors, have recently been developed as potential cancer therapeutics and might also be effective in HGPS. The hope would be that these inhibitors will decrease the amount of harmful farnesylated pre-progerin in the cell and thus alleviate cellular HGPS symptoms. Fortunately, ongoing clinical trials on farnesyl transferase inhibitors show little toxicity, and early studies in cell culture have already shown a reversal of some of the cellular defects associated with HGPS [15–18].
The story of HGPS is an impressive example of the interplay of basic and clinical science. It is also a showcase for how modern biology deals with disease (Figure 2). Within less than two years, we have gone from only knowing the symptoms of the disease to identifying the disease-causing gene and learning much about how the mutant protein behaves in patient cells. The challenge ahead is to close the circle and to apply what we have learned about the cell biology of this disease to the development of therapeutic approaches. The insights from HGPS have also opened an entirely new and fascinating vista on the aging process. Who would have guessed two years ago that architectural elements of the cell nucleus might contribute to aging? These advances have put us in the rare and desirable situation of possibly being able to making significant steps towards understanding one of the most fascinating problems in biology—and at the same time do some good for patients and their families.
Figure 1 Hutchinson-Gilford Progeria Syndrome
HGPS is a childhood disorder caused by mutations in one of the major architectural proteins of the cell nucleus. In HGPS patients the cell nucleus has dramatically aberrant morphology (bottom, right) rather than the uniform shape typically found in healthy individuals (top, right).
Figure 2 The Disease–Discovery–Therapy Cycle
Diseases are clinically defined by their symptoms. The genetic basis of disease is revealed by gene discovery. The cellular and molecular basis of a disease is characterized by cell biological approaches. These in turn provide the foundation to develop targeted therapies to alleviate symptoms and to fully understand the pathophysiology of the organismal disease symptoms.
Citation: Scaffidi P, Gordon L, Misteli T (2005) The cell nucleus and aging: Tantalizing clues and hopeful promises. PLoS Biol 3(11): e395.
Paola Scaffidi and Tom Misteli are at the National Cancer Institute, National Institutes of Health, Bethesda, Maryland, United States of America. Leslie Gordon is at Brown Medical School, Providence, Rhode Island, United States of America.
Abbreviations
HGPSHutchinson-Gilford progeria syndrome
LMNAlamin A gene
==== Refs
References
Guarente L Kenyon C Genetic pathways that regulate ageing in model organisms Nature 2000 408 255 262 11089983
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
Misteli T Concepts in nuclear architecture Bioessays 2005 27 477 487 15832379
Kennedy BK Gotta M Sinclair DA Mills K McNabb DS Redistribution of silencing proteins from telomeres to the nucleolus is associated with extension of life span in S. cerevisiae
Cell 1997 89 381 391 9150138
De Sandre-Giovannoli A Bernard R Cau P Navarro C Amiel J Lamin a truncation in Hutchinson-Gilford progeria Science 2003 300 2055 12702809
Eriksson M Brown WT Gordon LB Glynn MW Singer J Recurrent de novo point mutations in lamin A cause Hutchinson-Gilford progeria syndrome Nature 2003 423 293 298 12714972
DeBusk FL The Hutchinson-Gilford progeria syndrome. Report of 4 cases and review of the literature J Pediatr 1972 80 697 724 4552697
Lammerding J Hsiao J Schulze PC Kozlov S Stewart CL Abnormal nuclear shape and impaired mechanotransduction in emerin-deficient cells J Cell Biol 2005 170 781 791 16115958
Liu B Wang J Chan KM Tjia WM Deng W Genomic instability in laminopathy-based premature aging Nat Med 2005 11 780 785 15980864
Misteli T Scaffidi P Genome instability in progeria: When repair gets old Nat Med 2005 11 718 719 16015360
Martin GM Genetic modulation of senescent phenotypes in Homo sapiens
Cell 2005 120 523 532 15734684
Chen L Lee L Kudlow BA Dos Santos HG Sletvold O LMNA mutations in atypical Werner's syndrome Lancet 2003 362 440 445 12927431
Campisi J Cancer and ageing: Rival demons? Nat Rev Cancer 2003 3 339 349 12724732
Scaffidi P Misteli T Reversal of the cellular phenotype in the premature aging disease Hutchinson-Gilford progeria syndrome Nat Med 2005 11 440 445 15750600
Yang SH Bergo MO Toth JI Qiao X Hu Y Blocking protein farnesyltransferase improves nuclear blebbing in mouse fibroblasts with a targeted Hutchinson-Gilford progeria syndrome mutation Proc Natl Acad Sci U S A 2005 102 10291 10296 16014412
Glynn MW Glover TW Incomplete processing of mutant lamin A in Hutchinson-Gilford progeria leads to nuclear abnormalities, which are reversed by farnesyltransferase inhibition Hum Mol Genet. 2005 E-pub ahead of print
Capell BC Erdos MR Madigan JP Fiordalisi JJ Varga R Inhibiting farnesylation of progerin prevents the characteristic nuclear blebbing of Hutchinson-Gilford progeria syndrome Proc Natl Acad Sci U S A 2005 102 12879 12884 16129833
Toth JI Yang SH Qiao X Beigneux AP Gelb MH Blocking protein farnesyltransferase improves nuclear shape in fibroblasts from humans with progeroid syndromes Proc Natl Acad Sci U S A 2005 102 12873 12878 16129834
Hennekes H Nigg EA The role of isoprenylation in membrane attachment of nuclear lamins. A single point mutation prevents proteolytic cleavage of the lamin A precursor and confers membrane binding properties J Cell Sci 1994 107 1019 1029 8056827
Navarro CL De Sandre-Giovannoli A Bernard R Boccaccio I Boyer A Lamin A and ZMPSTE24 (FACE-1) defects cause nuclear disorganization and identify restrictive dermopathy as a lethal neonatal laminopathy Hum Mol Genet 2004 13 2493 2503 15317753
Varela I Cadinanos J Pendas AM Gutierrez-Fernandez A Folgueras AR Accelerated ageing in mice deficient in Zmpste24 protease is linked to p53 signalling activation Nature 2005 E-pub ahead of print
Goldman RD Shumaker DK Erdos MR Eriksson M Goldman AE Accumulation of mutant lamin A causes progressive changes in nuclear architecture in Hutchinson-Gilford progeria syndrome Proc Natl Acad Sci U S A 2004 101 8963 8968 15184648
|
16277559
|
PMC1283398
|
CC0
|
2021-01-05 08:21:45
|
no
|
PLoS Biol. 2005 Nov 15; 3(11):e395
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030395
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627756010.1371/journal.pbio.0030396FeatureCell BiologyEvolutionNoneJump-Starting a Cellular World: Investigating the Origin of Life, from Soup to Networks FeatureRobinson Richard 11 2005 15 11 2005 15 11 2005 3 11 e396Copyright: © 2005 Richard Robinson.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.If we agree that complex life evolved ultimately from single-celled organisms, how do we explain the origins of the cell itself?
==== Body
A physicist, a chemist, and a mathematician are stranded on a desert isle, when a can of food washes up on the beach. The three starving scientists suggest, in turn, how to open the can and ease their hunger. The physicist suggests they hurl it upon the rocks to split it open, but this fails. The chemist proposes they soak it in the sea and let the salt water eat away at the metal; again, no luck. They turn in desperation to the mathematician, who begins, “Assume we have a can opener….”
When discussing the evolution of life, biologists can often sound a bit like that mathematician. Beginning with a single cell, Darwinian evolution provides a simple, robust, and powerful algorithm for deriving all the astonishing richness of life, from bacteria to brains. Natural selection and other evolutionary forces, acting on surplus populations of replicating cells and multicellular organisms, lead inevitably to evolution and adaptation. Give biologists a cell, and they'll give you the world. But beyond assuming the first cell must have somehow come into existence, how do biologists explain its emergence from the prebiotic world four billion years ago?
The short answer is that they can't, yet. But this question may be a little closer to being answered as new money enters the field, and two new discoveries provide support for two competing models of prebiotic evolution.
While the past half century has seen an explosion of knowledge about the evolution of life after it began, there has been relatively little progress in the past half century on how it began—the so-called origin question. In part, the problem is financial: research money has flooded many other areas in biology, but remains in short supply in this one. “The funding is a big part of it,” says Jack Szostak, a Howard Hughes investigator and Professor of Genetics at Harvard Medical School. As a result, there is a shortage of researchers willing to commit their professional careers to finding out how life began. “This is a risky field to be involved with. The problems are hard. You can train students, but there may not be jobs waiting for them afterwards.” To that end, Harvard University recently announced a plan to fund origin-of-life research to the tune of one million dollars per year, which Szostak says is a good start.
Beyond assuming the first cell must have somehow come into existence, how do biologists explain its emergence from the prebiotic world four billion years ago?
But finding the answer to the origin question will require not only money but also progress in understanding how the most basic of biological molecules were put together before life began, how they became organized and self-sustaining, and how they developed into the membrane-bound cells that are our ancestors. Scientists have come a long way from the early days of supposing that all this would inevitably arise in the “prebiotic soup” of the ancient oceans; indeed, evidence eventually argued against such a soup, and the concept was largely discarded as the field progressed. But significant problems persist with each of the two competing models that have arisen—usually called “genes first” and “metabolism first”—and neither has emerged as a robust and obvious favorite. Now, two papers published in mid-2005 offer each camp some encouragement.
The Origin of Origin-of-Life Experiments
Some 50 years ago, Stanley Miller, then a graduate student at the University of Chicago and now in the Department of Chemistry at University of California at San Diego, got the field of origins research started with a bang—literally. He passed high-voltage electric sparks—a stand-in for lightning—through a gaseous mixture of water, methane, hydrogen, and ammonia, thought to be the major constituents of the ancient atmosphere. The liquid in the reaction flask eventually became a bouillon-like mix of amino acids and other small organic molecules. Miller's results predicted that, over time, the early oceans would have become a rich prebiotic soup, replete with amino acids, nucleic acids, and sugars. His results implied it was only a matter of time before these building blocks combined to form complex polymers and ultimately a replicating cell.
In the beginning, according to the so-called genes-first camp, was a single RNA molecule, both code and catalyst. Such a “replicase” would have catalyzed its own replication, and also provided the template on which the copy was made.
“The initial Miller experiment was earth-shaking,” says Harold Morowitz, Professor of Biology at George Mason University, and a long-time theorist and researcher in this area. The suggestion that random chemistry could produce the molecules of life “held the field for a long time.” But later calculations appeared to show that the early atmosphere contained much more carbon dioxide and much less hydrogen than Miller's model required, and correcting these concentrations cast doubt on the likelihood that complex molecules would form in abundance Where, then, might organic precursors have come from? There is some, albeit scant, evidence for their arrival on comets colliding with the earth, but there is little enthusiasm for this as a solution. Finally, there is no geologic evidence, in either sediments or metamorphic rocks, that such a soup ever existed.
An RNA World Needs Nucleotides
In the early 1980s, just as Miller-type chemistry was falling out of favor, RNA emerged as the rising star of origin-of-life research, based on a startling discovery. Up to this point, evolution appeared to have a severe chicken-and-egg problem: information-bearing DNA codes for protein, but catalytic proteins are essential to make DNA. That the two could have arisen independently but still work in concert seemed highly unlikely. But RNA, which was well-known in its role as temporary information carrier, also turned out to be catalytic. Indeed, a host of functions in modern cells that were once thought to be the province of proteins are instead supervised by catalytic RNA. It is only a small intellectual leap from here back to an “RNA world,” in which RNA, not DNA, is the molecule of heredity, and RNA, not protein, is the catalytic engine of the cell. In the beginning, according to the so-called genes-first camp, was a single RNA molecule, both code and catalyst. Such a “replicase” would have catalyzed its own replication, and also provided the template on which the copy was made.
Work by Jack Szostak of Harvard University has lent support to this elegant model. He has shown that certain catalytic RNAs can, indeed, join smaller RNA sequences together, hinting at the potential for self-replication. Given the right starting conditions, such a self-replicating RNA might increase its number at the expense of the “lifeless” ones surrounding it. Successive rounds of copying, with minor mutations, could lead the original replicator to acquire new abilities. Life, Szostak speculates, “starts simple, beginning with one gene, probably a replicase, and accretes additional functionality over time.”
It is a highly appealing concept, and has driven a great deal of good research. But how would the original replicase arise? James Ferris, Professor of Chemistry at Rensselaer Polytechnic Institute, has discovered that on the surface of montmorillonite, a common clay, activated RNA nucleotides—the monomeric building blocks of the RNA polymer—will spontaneously link together to form longer chains. While the sequences of these products are entirely random, Szostak has shown that within such a random pool of RNAs, some are likely to be catalytic. Szostak has also recently shown that replicating RNAs inside a lipid membrane vesicle cause the vesicle to grow, mimicking behavior of actual cells.
Box 1. How Did Life Become Handed?
To date, none of the models have proposed a solution to one of the more vexing origin problems: chirality. Three-dimensional molecules such as sugars and amino acids can exist in two mirror-image forms, like left and right hands (chiros is Greek for hand). Any nonbiological synthesis of such molecules, as would have occurred before life arose, produces equal amounts of each type. Nonetheless, modern cells use exclusively left-handed amino acids and right-handed ribose sugars, and interference from the wrong kind shuts down biological reactions. How could chiral life arise in the presence of so much interference?
“It's a serious problem,” Orgel admits, “but not an overwhelmingly serious one.” Orgel suggests that one of several possible solutions may be chance, a “frozen accident” that brought together, and kept together, molecules of the right chirality. Such an accident is perhaps not so unlikely, says Martin, who calculates that a mixture of every possible left- and right-handed combination of a 25–amino acid peptide (amino acid chain) would weigh 25 kilograms. “Any smaller sample is imperfect,” he says.
Martin also points out the problem may be a bit easier than it seems, since the chirality of a molecule such as a sugar is usually maintained as that molecule wends its way through a metabolic pathway. An enzyme at the head of that pathway could act as a “filter,” allowing only those molecules of the correct chirality to enter, thus fixing chirality for that pathway and others that branch off of it. Exactly this feature is seen in the central metabolic pathway for sugar formation found in all cells.
But working back even further, where do the nucleotides come from to form these chains? Here we come up against the “can opener” problem on the molecular level. “The biggest concern about the RNA world is that there has been no convincing prebiotic creation of the activated monomers” in any plausible prebiotic world, says Ferris. Despite years of experiments with dozens of different strategies, no one has figured out how to make this most essential of starting ingredients for an RNA world. “There is a growing realization that we may need to look beyond RNA,” Szostak says, to molecules whose chemistry is a bit more tractable, such as a peptide nucleic acid (PNA), a synthetic amino acid–nucleotide hybrid. These original replicators might then have given way to RNA, says Leslie Orgel, senior fellow and research professor at the Salk Institute of Biological Studies.
The case for PNA is weak, though. While modern cells still bear traces of a catalytic RNA world within them, “there is absolutely nothing that I know of to suggest there is evidence for PNA or other such molecules in present cells,” says Orgel. If they ever contributed to the development of life, all traces of their existence appear to have been wiped clean.
Whether the original replicator was RNA or PNA or some other molecule, any genes-first model relies on an abundance of building blocks in the environment, a requirement that seems to depend on the discounted idea of Miller's prebiotic soup. But in June of 2005, the prebiotic soup got a new lease on life. New calculations appear to show that there was considerably more hydrogen in the early atmosphere than once thought. “This could resurrect Miller's chemistry,” says Orgel. Nonetheless, “there is still an enormous way to go” to get the full set of RNA precursor molecules.
Metabolism More Ancient than Replication?
In 1988, even while the RNA world was enjoying its intellectual honeymoon, a German biochemist and patent attorney, Günther Wächtershäuser, proposed a radical alternative theory of the origin of life based on, of all things, fool's gold. Iron disulfide—pyrite or fool's gold—can catalyze a variety of crucial biochemical reactions. There are iron sulfide or iron–nickel sulfide clusters at the heart of several ancient and vital enzymes in use in all cells today. Wächtershäuser proposed that the earliest living system was not a nucleotide-based replicator but a mineral-based metabolizer, converting simple and abundant inorganic compounds—carbon dioxide, hydrogen sulfide—into more complex organic ones on the surface of a pyrite crystal, probably at deep-sea hydrothermal vents.
This metabolism-first model has a strong champion in Harold Morowitz, who paints the two major models for life's origin as “heaven and hell.” Miller-type scenarios, including the genes-first model, rely on a wealth of precursors raining down from above. The standard model of the RNA world, he says, “requires an environment that is impossibly improbable.” The alternative is a much smaller set of molecules, at much higher concentrations, bubbling up from below. “I really like theories that go from simplicity to complexity,” Morowitz says.
The possibility that metabolism first began at hydrothermal vents has been advanced most recently by Michael Russell, Research Professor of Geology at the Scottish Universities Environmental Research Centre in Glasgow, and William Martin, Professor at the University of Düsseldorf. Russell and Martin propose that life's metabolism developed not on a two-dimensional pyrite surface but within tiny cavities lined with iron monosulfide, through which percolated an energy-rich mix of hydrogen and carbon dioxide dissolved in seawater.
The standard model of the RNA world...“requires an environment that is impossibly improbable.” The alternative is a much smaller set of molecules, at much higher concentrations, bubbling up from below.
In the early 1980s, Russell discovered fossil fields of iron sulfide “chimneys,” formed on the ocean floor 350 million years ago (Figures 1 and 2). Unlike the more famous and much larger “black smokers”—deep-sea hydrothermal vents found in mid-ocean ridges that spew hot, mineral-rich water out of their sulfide chimneys—each of Russell's chimneys is no more than ten centimeters high and little more than a centimeter across. The entire field covers tens of square meters, and is composed, in part, of many thousands of millimeter-sized cavities, formed by outgassing of hydrogen and carbon dioxide, which bubbled up through cracks in the crust. While the particular structures Russell discovered formed well after life's origins, similar ones almost certainly existed in the prebiotic ocean, Russell says. Each would have remained stable over the course of thousands of years, a little experimental reaction vessel at the bottom of the sea.
Figure 1 Tynagh Chimneys
A view from above a chimney field, showing the chimneys (round black circles) and bubbles, which contain chambers. The object placed for scale is two centimeters across. These fossil chimneys were formed well after life's origin, but may be similar to those in which, according to one hypothesis, metabolism first began
Figure 2 Botyroidal Cross-Section
A cross-section through an iron sulfide deposit shows the small chambers within. One hypothesis of life's origin suggests that in such chambers metabolism first began, as hydrogen and carbon dioxide bubbled through and reacted to form simple organic compounds
And unlike the 400 °C water spewing out of a black smoker, the water flowing up through these vents was much cooler, not much more than 100 °C. Outside the chamber, the ocean would have been much cooler still, and more acidic and more oxidized than the solution within, creating a set of strong temperature and electrochemical gradients across the microscopically porous surface of the chamber. “Life loves to live at the gradients,” says Russell.
The slow trickle of hydrogen and carbon dioxide through such chambers and across the iron sulfide catalyst promotes formation of acetate, according to Russell and Martin. Acetate is a key intermediate in virtually all biosynthetic pathways, and in modern cells, enters these reactions tethered to sulfur. In modern bacteria, the two enzymes that make acetate depend on a catalytic core of iron, nickel, and sulfur, arranged almost exactly as they are in the free mineral itself. “In other words,” Russell and Martin have written, these enzymatic metal clusters “are not inventions of the biological world, rather they are mimics of minerals that are indisputably older, and which themselves have catalytic activity in the absence of protein” [1].
These chambers also suggest a radical solution to a heretofore stubborn problem, one with no other obvious resolution. Modern cells without nuclei are grouped into two domains, the Eubacteria and the Archaebacteria. These ancient lineages share the same energy metabolism and the same genetic code, presumably reflecting a single common ancestor. But they differ profoundly in how they synthesize the lipids in their membranes. One explanation, which Martin dismisses, is that the common ancestor had a membrane, probably similar to the eubacterial structure, and the archaebacterial ancestor “had to completely reinvent its cell wall chemistry.” Such proposals are “completely decoupled from microbial physiology,” he says. The alternative favored by Russell and Martin is that these lipid differences reflect a divergence in the two lines after the last common ancestor already had its carbon biochemistry and genetic code intact, but before the development of lipid membranes. The chambers served as the original cell compartment, and were only replaced by lipids after the eubacterial and archaebacterial lines split.
This metabolism-first model is not an alternative to life based on RNA...But it does propose that geology at hydrothermal vents provided the structure in which life emerged
Russell and Martin's model also provides a solution to another thorny issue in jump-starting life, that of concentration. An essential feature of all cells is their ability to maintain high concentrations of materials that are in short supply in the world around them. In the absence of a cell membrane, how did proto-life forms collect raw materials, and prevent products from dissipating into the vastness of the environment around them? Russell's chambers solve this problem in essentially the same way modern cells do, with an external boundary that is permeable to small reactant molecules, but much less so to larger product ones. In Russell's and Martin's scenario, then, the stable metabolism that developed within these chambers eventually gave rise to a genetic system, probably dependent on RNA, which encoded simple proteins, probably through direct accretion of RNA and amino acids on the surface of a mineral catalyst. Finally, these proto-organisms developed membranes, completing their evolution into recognizable cells.
This metabolism-first model is not an alternative to life based on RNA. “We can't work without an RNA world either,” says Martin. But it does propose that geology at hydrothermal vents provided the structure in which life emerged, and suggests that understanding prebiotic organic chemistry at these vents may provide the key to understanding the emergence of life from nonlife.
Self-Organizing Metabolic Networks
While not necessarily convinced of the details of this proposal, Morowitz applauds the focus on bacterial physiology as a guide to understanding early life. “Metabolism recapitulates biogenesis,” he proposes.
But for Morowitz, the most exciting development in the metabolism-first camp, “the really new idea,” is that small organic molecules, such as amino acids, can catalyze the formation of other small organic molecules, such as nucleic acids. “This has emerged only in the last two years,” he says. This view has found strong support from a new finding published in the journal Chemistry in August 2005, which indicates that single amino acids can catalyze the creation of sugars from simple starting materials with enzyme-like specificity.
“What has emerged is a very strong self-organizing principle,” says Morowitz. In this view, while iron sulfide may have been the original catalyst, it did not remain the only one for long. As products of the original reactions catalyzed new reactions, metabolic networks quickly arose. Feedback loops developed when two molecules regulated one another's synthesis. “The system can piggyback its way upward,” he says.
While the study of such networks is still in its infancy, Morowitz suggests they hold the key to a host of knotty problems, including that of RNA synthesis. “It's not a problem in this network point of view. Very early on you get the precursor compounds,” while formation of the complete nucleotide arises later. “Even today this is the core network of biochemistry.”
It is still unclear how, or whether, these competing models will fit together, and whether they will lead to a robust scenario for life's origin. Indeed, all may eventually prove wrong, and the real solution may lie hidden in some discovery yet to be made. Whatever the difficulties, says Morowitz, the allure of the field lies in its potential to answer the biggest question of them all. “You're not going to make drugs or better agriculture. You're going to make a philosophical impact.” Szostak agrees: “These are the big questions. Anybody who thinks has to be grabbed by these.”
Citation: Robinson R (2005) Jump-starting a cellular world: Investigating the origin of life, from soup to networks. PLoS Biol 3(11): e396.
Richard Robinson is a freelance science writer from Sherborn, Massachusetts, United States of America. E-mail: [email protected]
Abbreviation
PNApeptide nucleic acid
==== Refs
Reference
Russell MJ Martin W The rocky roots of the acetyl-CoA pathway Trends Biochem Sci 2004 29 358 363 15236743
|
16277560
|
PMC1283399
|
CC BY
|
2021-01-05 08:21:45
|
no
|
PLoS Biol. 2005 Nov 15; 3(11):e396
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030396
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627756110.1371/journal.pbio.0030409EditorialEvolutionNoneScientific Longevity EditorialParthasarathy Hemai 11 2005 15 11 2005 15 11 2005 3 11 e409Copyright: © 2005 Hemai Parthasarathy.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.This issue of PLoS Biology includes the first in a series of "historical" book reviews.
==== Body
When I was a graduate student, among the most common laments from senior scientists was the lack of scientific scholarship in younger colleagues. If a paper predated or was otherwise excluded from the listing of abstracts in Medline, so they said, it was likely to be at best uncited and at worst recapitulated as an entirely novel discovery. The transition from bound and dusty volumes of printed journals to increasingly backdated electronic archives is, however, progressing (ISI's Web of Science now goes back to 1945), and with it comes the hope of renewed interest in past scientific insights.
Indeed, one promise of open-access publishing is that observations that were out of scientific context and seemingly uninterpretable—thereby relegated to publication in obscure niche journals—will have the opportunity to come together in digital space and inspire synthesis. Another promise is that science educators will have more opportunity to highlight the research process by pointing directly to the source of discovery, rather than solely to the digested contents of textbooks.
We are therefore pleased to present the first in a series of reviews of scientific books that have moved into the public domain, digitized and freely accessible, which promote open-access publishing by virtue of longevity. The author whom we have chosen to highlight first is certainly not obscure—and, indeed, his 19th century theories are the source of renewed controversy in America today. However, Charles Darwin wrote more than just The Origin of Species, and as described by the evolution scholar Niles Eldredge in this issue of PLoS Biology (DOI: 10.1371/journal.pbio.0030382), it is his earlier unpublished works—the “Red” and “Transmutation” notebooks (1836–1839), the “Sketch” (1842), the “Essay” (1844), and Natural Selection (1856–1858)—that help us to trace the development of Darwin's great intellectual achievements.
Darwin's writings have been digitized and made freely available (http://darwinlibrary.amnh.org) by the American Natural History Museum in New York, which will feature an in-depth exhibition on this highly original theoretician, botanist, geologist, and naturalist. The exhibition opens 19 November 2005, and will run until 29 May 2006. It promises to feature live Galápagos tortoises, along with actual fossil specimens collected by Darwin. It will include an elaborate reconstruction of his study at Down House, where he first proposed evolution by natural selection. In short, the organizers hope to bring alive for the public the science and thinking behind a theory that scientists embrace as the most powerful unifying force in modern biological thought. We hope that by highlighting his original words, PLoS Biology will foster a deeper interest in and understanding of the origins of evolutionary biology.
More generally, we hope these historical reviews will encourage readers to explore science at its origins. Unlike the contemporary scientific literature—much of which is filled with jargon and acronyms, articulated in the “least publishable unit,” and lacking in elegant prose—the science of the past was often first disseminated in book form, intended to reach contemporary scientific colleagues and Renaissance men alike. Although foci shift and terminology changes, the fundamental approach to scientific problems changes much less quickly. We hope that the words of our greatest thinkers of the past inspire you to address the challenges of the future.
The long-lived Galápagos tortoise
(Image: Catriona MacCallum)
Citation: Parthasarathy H (2005) Scientific longevity. PLoS Biol 3(11): e409.
Hemai Parthasarathy is Managing Editor of PLoS Biology. E-mail: [email protected]
|
16277561
|
PMC1283408
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Nov 15; 3(11):e409
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030409
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627756110.1371/journal.pbio.0030409EditorialEvolutionNoneScientific Longevity EditorialParthasarathy Hemai 11 2005 15 11 2005 15 11 2005 3 11 e409Copyright: © 2005 Hemai Parthasarathy.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.This issue of PLoS Biology includes the first in a series of "historical" book reviews.
==== Body
When I was a graduate student, among the most common laments from senior scientists was the lack of scientific scholarship in younger colleagues. If a paper predated or was otherwise excluded from the listing of abstracts in Medline, so they said, it was likely to be at best uncited and at worst recapitulated as an entirely novel discovery. The transition from bound and dusty volumes of printed journals to increasingly backdated electronic archives is, however, progressing (ISI's Web of Science now goes back to 1945), and with it comes the hope of renewed interest in past scientific insights.
Indeed, one promise of open-access publishing is that observations that were out of scientific context and seemingly uninterpretable—thereby relegated to publication in obscure niche journals—will have the opportunity to come together in digital space and inspire synthesis. Another promise is that science educators will have more opportunity to highlight the research process by pointing directly to the source of discovery, rather than solely to the digested contents of textbooks.
We are therefore pleased to present the first in a series of reviews of scientific books that have moved into the public domain, digitized and freely accessible, which promote open-access publishing by virtue of longevity. The author whom we have chosen to highlight first is certainly not obscure—and, indeed, his 19th century theories are the source of renewed controversy in America today. However, Charles Darwin wrote more than just The Origin of Species, and as described by the evolution scholar Niles Eldredge in this issue of PLoS Biology (DOI: 10.1371/journal.pbio.0030382), it is his earlier unpublished works—the “Red” and “Transmutation” notebooks (1836–1839), the “Sketch” (1842), the “Essay” (1844), and Natural Selection (1856–1858)—that help us to trace the development of Darwin's great intellectual achievements.
Darwin's writings have been digitized and made freely available (http://darwinlibrary.amnh.org) by the American Natural History Museum in New York, which will feature an in-depth exhibition on this highly original theoretician, botanist, geologist, and naturalist. The exhibition opens 19 November 2005, and will run until 29 May 2006. It promises to feature live Galápagos tortoises, along with actual fossil specimens collected by Darwin. It will include an elaborate reconstruction of his study at Down House, where he first proposed evolution by natural selection. In short, the organizers hope to bring alive for the public the science and thinking behind a theory that scientists embrace as the most powerful unifying force in modern biological thought. We hope that by highlighting his original words, PLoS Biology will foster a deeper interest in and understanding of the origins of evolutionary biology.
More generally, we hope these historical reviews will encourage readers to explore science at its origins. Unlike the contemporary scientific literature—much of which is filled with jargon and acronyms, articulated in the “least publishable unit,” and lacking in elegant prose—the science of the past was often first disseminated in book form, intended to reach contemporary scientific colleagues and Renaissance men alike. Although foci shift and terminology changes, the fundamental approach to scientific problems changes much less quickly. We hope that the words of our greatest thinkers of the past inspire you to address the challenges of the future.
The long-lived Galápagos tortoise
(Image: Catriona MacCallum)
Citation: Parthasarathy H (2005) Scientific longevity. PLoS Biol 3(11): e409.
Hemai Parthasarathy is Managing Editor of PLoS Biology. E-mail: [email protected]
|
0
|
PMC1283410
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Nov 15; 3(11):e428
|
latin-1
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030428
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627426310.1371/journal.pbio.0030387Research ArticleBioinformatics/Computational BiologyEvolutionGenetics/Genomics/Gene TherapyNeuroscienceStatisticsHomo (Human)PrimatesAncient and Recent Positive Selection Transformed Opioid cis-Regulation in Humans PDYN EvolutionRockman Matthew V [email protected]
1
¤Hahn Matthew W
1
2
¤Soranzo Nicole
3
Zimprich Fritz
4
Goldstein David B
1
3
5
Wray Gregory A
1
5
1Department of Biology, Duke University, Durham, North Carolina, United States of America,2Center for Population Biology, University of California, Davis, California, United States of America,3Department of Biology, University College, London, United Kingdom,4Department of Clinical Neurology, Medical University of Vienna, Vienna, Austria5Institute for Genome Sciences and Policy, Duke University, Durham, North Carolina, United States of AmericaDermitzakis Emmanouil T. Academic EditorThe Wellcome Trust Sanger InstituteUnited Kingdom12 2005 15 11 2005 15 11 2005 3 12 e38719 7 2005 13 9 2005 Copyright: © 2005 Rockman 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.
Selection on a Neural Gene Regulator Sheds Light on Human Evolution
Changes in the cis-regulation of neural genes likely contributed to the evolution of our species' unique attributes, but evidence of a role for natural selection has been lacking. We found that positive natural selection altered the cis-regulation of human prodynorphin, the precursor molecule for a suite of endogenous opioids and neuropeptides with critical roles in regulating perception, behavior, and memory. Independent lines of phylogenetic and population genetic evidence support a history of selective sweeps driving the evolution of the human prodynorphin promoter. In experimental assays of chimpanzee–human hybrid promoters, the selected sequence increases transcriptional inducibility. The evidence for a change in the response of the brain's natural opioids to inductive stimuli points to potential human-specific characteristics favored during evolution. In addition, the pattern of linked nucleotide and microsatellite variation among and within modern human populations suggests that recent selection, subsequent to the fixation of the human-specific mutations and the peopling of the globe, has favored different prodynorphin cis-regulatory alleles in different parts of the world.
Strong positive selection has resulted in changes to the regulation of the human prodynorphin gene, with evidence for increased expression as compared with chimp. Additionally, recent selection has led to different alleles in different parts of the world.
==== Body
Introduction
Discovering the genetic changes that accompanied the origins of modern humans and pinpointing the subset of changes driven by natural selection remain central problems in evolutionary anthropology. These changes are likely to have included changes in the complement of genes, changes in the amino acid sequences of proteins, and changes in cis-regulation. While divergence in gene complement [1–4] and amino acid sequence [5–9] are discernable from genome sequences, functional divergence in cis-regulatory regions is largely invisible in sequence data. Consequently, while we now know of many uniquely human aspects of gene complement and protein sequence, we possess only a few documented examples of human-specific cis-regulation [10,11]. Moreover, while statistical tests for discerning the signature of positive selection in protein-coding sequences are well developed, and genomic surveys have identified many human genes showing evidence of positive selection [12,13] or diminished negative selection [14], in only a single instance has positive selection been implicated in cis-regulatory divergence between humans and other apes [15]. Our ignorance of cis-regulatory divergence is all the more remarkable considering the importance assigned to such changes in models of evolutionary novelty [16–20]. Thirty years have passed since King and Wilson [21] argued that human evolution owes more to changes in gene regulation than to changes in gene structure, and although their theoretical justifications remain strong, empirical study of human regulatory evolution has not kept pace [22].
An understanding of the genetic basis for human traits necessarily focuses on the evolution of the brain. Inquiries into human brain-specific gene regulation have relied on phenotypic analyses, particularly microarray measurements of gene expression in post-mortem human and ape brain tissue. These analyses document extensive differences in gene expression between humans and other great apes [23–26], but the genetic basis of such differences—if any—remains unknown. Moreover, a specific class of change in gene regulation, change in transcriptional inducibility, is invisible to studies of post-mortem tissues. Our species is distinguished by the ability to respond to and manipulate environmental cues; the responsiveness of genes to such cues, and not merely their constitutive activity, may play a role in human evolution.
Given the difficulties associated with identifying DNA sequence changes responsible for changes in inducibility, we focused on a candidate region whose role in inducibility in humans has already been demonstrated. The a priori designation of a functional regulatory element allows us to apply the tools of molecular evolution developed for protein sequences—specifically, rate comparisons among classes of nucleotide sites [20]. Of necessity, we also investigated the functional consequences of the species differences experimentally; we used transient transfection of promoter-reporter constructs into cultured human cells, a method with a record of success in identifying functional variation within our species [27–30].
We studied the cis-regulatory evolution of the opioid neuropeptide precursor prodynorphin (PDYN) (OMIM 131340). The opioid neuropeptides (endorphins) are the endogenous ligands for the opiate receptors. They mediate the anticipation and experience of pain [31,32], they influence behaviors including social attachment and bonding [32,33], and they affect learning and memory [32,34]. One of PDYN's products is dynorphin, a peptide whose pharmacological analogs specifically affect perception [35]. A 68 base-pair (bp) tandem repeat polymorphism in the human PDYN promoter, 1,250 bp upstream from the start of transcription, influences the inducibility of the gene [36], and association studies have tentatively implicated the polymorphism in schizophrenia [37], cocaine addiction [38], and epilepsy [39]. These genetic associations are supported by physiological associations between PDYN expression and each of the phenotypes [40–42].
Results
Phylogenetic Evidence: Accelerated Evolution of a Functional Element in the Human PDYN Promoter
To understand the evolutionary basis for the functional variation, we sequenced 3 kilobases (kb) of PDYN regulatory DNA from 74 human chromosomes and 32 chromosomes from seven species of non-human primates, experimentally determining haplotypic phase by cloning each allele. The non-human primates bear a single copy of the 68-bp regulatory element, and the pattern of substitutions implies that the duplication of the element is specific to the human lineage. All human copies of the element carry five substitutions that differentiate them from the sequence inferred for the last common ancestor of humans and chimpanzees. A sixth difference is variable among repeats in some human haplotypes (Figure 1).
Figure 1 Divergence of the 68-bp Element in Humans
Arrows indicate five differences fixed on the human lineage. The asterisk indicates a site that varies among human repeats. In the sample of 74 human haplotypes, all one-repeat and most two-repeat alleles bear G at this site. Complete haplotype data are given in Table S1. Below, schematic of the study region showing the position of the element and the non-coding first exon with respect to the start of transcription.
The five substitutions fixed on the human lineage are dramatically more than expected for 68 bp of neutrally evolving sequence. Under a model of spatially random mutation, the expected number of substitutions is fewer than 0.5, and the observed number is extremely improbable (Poisson p < 0.0001), whether we calculate the expectation from the density of substitutions across the PDYN promoter, the average divergence between human and chimpanzee on a chromosomal scale [43], or the estimated great ape substitution rate [44].
The elevated number of substitutions may be due to locally elevated mutation rate or to positive selection increasing the probability of fixation of new mutations. If the local mutation rate is intrinsically elevated, other species should also exhibit rapid evolution in the 68-bp region. We therefore tested a molecular clock, using phylogenetic likelihood ratio tests [45], to ask whether the 68-bp region is evolving rapidly due to an elevated mutation rate. The evolution of the 68-bp element is significantly accelerated exclusively along the branch leading to humans from our last common ancestor with chimpanzees (p = 0.005). The other branches of the evolutionary tree show no departure from rate constancy (p = 0.657), and the remainder of the promoter region and the coding sequence of PDYN also show no acceleration (Table 1). To control for possible lineage specific rate variation, we applied the relative ratio test [46], which allows for lineage-specific rates and for DNA region-specific rates, and tests for lineage-by-region interactions. Here again, the human 68-bp repeat exhibits a significant departure from the neutral expectation (p = 0.001 for proportionality to the rest of the promoter, p = 0.015 for proportionality to the coding sequence; Table 2); the remaining lineages and regions exhibit no such departures. The phylogenetic data imply that the rapid evolution of the human 68-bp element is due to positive selection.
Table 1 Molecular Clock Tests
Table 2 Relative Ratio Tests
The molecular evolution of the PDYN protein sequence, unlike the regulatory DNA, is consistent with a history dominated by negative selection. In a sample of the complete coding sequences from multiple chromosomes of eight primate species, 25 of the 254 amino acids in the PDYN protein vary, but none of the variants affect the 56 amino acids that comprise the neuroactive peptides. The exclusion of variation from the mature opioid peptides (Poisson p = 0.004) implies negative selection to maintain function. Phylogenetic likelihood ratio tests found no support for positive selection shaping the amino acid sequence of the remainder of the preprotein (model 1 versus model 2 of Yang et al. [47], p = 0.62). Nielsen et al. [13], in a genome scan of human-chimpanzee orthologs, also found no evidence for selection on the PDYN protein. No amino acid polymorphisms are known among humans, and we found none by directly sequencing the coding regions from chromosomes bearing each of the four repeat-number alleles of the promoter.
Population Genetic Evidence: An Excess of High-Frequency-Derived Mutations Flanking the Selected Element
Positive selection alters the frequency spectrum of linked neutral mutations. As the selected mutations are driven rapidly to fixation, linked alleles are dragged along to high frequency [48]. The linked alleles may be dragged to fixation, but they may also be driven to high frequency and then decoupled from the selected mutation by recombination or allelic gene conversion. As a result, an excess of high-frequency-derived mutations flanking a fixed difference provides evidence for positive selection [49]. Our sample of 74 experimentally phased haplotypes from an Austrian population exhibits such a pattern (Table S1). Fay and Wu's H statistic is −8.13, strongly supporting a departure from neutrality and consistent with positive selection (p = 0.004). The three polymorphisms nearest to the 68-bp element have derived allele frequencies greater than 0.95 in all repeat-number allelic classes, consistent with a selective sweep that fixed mutations in the 68-bp region, and that thus predated the origin of different repeat alleles by tandem duplication. As the 68-bp element is tandemly repeated in all sampled human populations (Table 3), the signature of selection in all Austrian repeat-number allelic classes also implies that the selective events predate the global human diaspora. A sample of 20 chimpanzee haplotypes, though exhibiting many more polymorphic sites than the human haplotypes, and hence more power to detect a departure from neutrality, shows no such departure (H = 1.62).
Table 3
PDYN Repeat Allele Frequencies
The human-specific accelerated evolution of the 68-bp element is best explained as the result of positive selection favoring the fixation of mutations. Although the rate is elevated by a factor of more than ten over the neutral expectation, the selection intensity required to explain this excess is quite modest. The rate of substitution (k) is equal to the rate at which new mutations arise in the population (2Neμ) times the probability that a new mutation will become fixed [50], which is 1/2Ne for neutral mutations and approximately 2s for advantageous mutations in a population of constant size [51], where Ne is the effective population size, μ is the mutation rate, and s is the selective advantage of the mutant allele. If we let fa be the fraction of non-deleterious mutations in the 68-bp element that are advantageous, then the rate acceleration (kobserved/kneutra
l
) is the ratio of the substitution rate in the human 68-bp element ([1 − fa]μ + 4Nesμfa) to that expected in the absence of positive selection (μ). We can place bounds on fa by recognizing that there are only 204 base-substituting mutations possible in a 68-bp sequence. For the usual estimate [52] of long-term human effective population size, Ne = 10,500, s falls in the range 0.0002 to 0.045; for fa greater than 2.2% (e.g., if all five fixed mutations were advantageous) s is less than 0.01 (Figure S1), well below the estimated selection coefficients of lactase persistence in Northern Europe [53] and G6PD deficiency [54] in regions of endemic malaria.
Functional Evidence: The Selected Element Increases Inducible PDYN Expression
To determine the effect of the selected nucleotide substitutions on PDYN transcription, we transiently transfected the human neural cell line SH-SY5Y with constructs bearing 3 kb of human or chimpanzee PDYN cis-regulatory DNA linked to a luciferase reporter. Downstream of the 68-bp repeat, in the non-coding first exon of PDYN, is a downstream regulatory element (DRE), a binding site for the repressor protein DRE-antagonist modulator (DREAM) [55]. A single nucleotide substitution in humans alters the DRE from the sequence found in the other primate species. To isolate the effect of the substitutions in the 68-bp element from the effect of the substitution in the DRE, we generated chimeric constructs containing either the human DRE or the human 68-bp element in the context of the chimpanzee promoter (Figure 2).
Figure 2 The Human 68-bp Element Increases Induced PDYN Expression
We tested four 3-kb constructs, encompassing the region shown in Figure 1.
(A) The human and chimpanzee constructs differ at the sites indicated by vertical bars. The two chimeric constructs incorporated the human 68-bp element or the human DRE (DREAM binding site) into the chimpanzee construct.
(B and C) Panels show luciferase activity for each construct (± SEM, five to seven transfections), standardized to that observed for promoterless luciferase vectors (white bars), in SH-SY5Y and JAR cells, with and without added caffeine, which causes the release of intracellular Ca2+ and the release of DREAM from the DRE.
We found that the human DRE sequence conferred slightly elevated expression of the reporter under basal conditions, though the effect was not significant (Figure 2B; analysis of variance p = 0.11). The sequence of the 68-bp element has no effect under these conditions (p = 0.66). When the effect of DREAM is removed by stimulating the cells to release intracellular Ca2+, which binds DREAM and causes it to release from DNA, the effect of the substitutions in the 68-bp element is conspicuous. Under these conditions, the human 68-bp element drives significantly higher expression than the chimpanzee sequence, regardless of the source of flanking sequence (Figure 2B; p = 0.002). Each relevant pairwise contrast is significant by t-test (human versus chimpanzee, 120%, p = 0.006; chimpanzee with human element versus chimpanzee, 115%, p = 0.037; human versus chimpanzee with human DRE, 120%, p = 0.007). In a three-factor analysis of variance, incorporating Ca2+ stimulation and the sequences of the DRE and the 68-bp element, the main effects of Ca2+ (p < 0.001) and the 68-bp element (p = 0.011) are significant and the interaction between Ca2+ and the 68-bp element is nearly significant (p = 0.054).
In contrast to the SH-SY5Y results, we observed no difference between chimpanzee and human constructs in the non-neural JAR cell line (Figure 2C), which serves as a control for the biological relevance of the cis-regulatory differences. Because PDYN is expressed in a broad range of neural and endocrine cell types and is induced by a diverse array of stimuli, our limited survey of potential functional consequences of human-specific regulatory substitutions is unlikely to have identified all such changes. Although transient transfection entails the removal of the regulatory DNA from its chromosomal context and the possible loss of biologically important interactions, the experimental results imply that the substitutions in the 68-bp element are visible to the cell.
Continuing Selection: PDYN Exhibits Elevated Differentiation among Populations and Reduced Variation within Them
The evidence for positive selection on the functional 68-bp element, and hence for increased PDYN expression in humans, raises the possibility that selection has also acted more recently on the alleles that differ in the number of tandem repeats of the element following the origin of modern humans. Intraspecific PDYN variation is a plausible target for selection because variation in the number of repeats has been shown to affect inducibility by the phorbol ester TPA [36] and has been associated with protection against cocaine dependency [38] and with neurological disease [37,39]. Moreover, evidence for selection among human populations would corroborate the functional importance of the 68-bp element, and hence support the inference of selection in human origins.
Population genetics predicts that recent selection in human populations will leave two types of signatures in patterns of genetic variation: departures from neutral expectations in the pattern of differentiation among populations, and departures from neutral expectations in the pattern of variation within populations. These predictions have given rise to a battery of statistical tests: FST -based tests to examine differentiation among populations, and θ-based tests to examine diversity within populations [56].
We initially genotyped the repeat polymorphism in six Old World populations and compared differentiation among populations (measured by FST) at the repeat locus to the differentiation expected at loci evolving neutrally. Elevated FST is a signature of geographically heterogeneous positive selection, driving allele frequencies to differ among populations more rapidly than they would if genetic drift and migration only were acting [57]. We estimated the neutral distribution of FST values from a set of 18 mutually unlinked candidate neutral single nucleotide polymorphisms (SNPs) typed in the same individuals [58]. Each of the candidate neutral SNPs was selected for this preliminary screening on the basis of its high heterozygosity in Europe and its distance (more than 200 kb) from known genes. FST values are constrained by the overall level of variation at a locus, so high heterozygosity is a useful filter for a pool of informative marker SNPs. Similarly, because genes and their regulatory elements are more likely to be under selection than arbitrary non-coding DNA, SNPs distant from genes are good candidates for neutral mutations.
Alleles with one or four copies of the PDYN repeat element are rare in every population we examined, but the frequencies of the two- and three-repeat alleles differ dramatically among populations (Table 3). The three-repeat allele ranges in frequency from less than 10% in China and New Guinea to more than 60% in Italy and Ethiopia. The differentiation at the repeat locus is higher than all 18 neutral markers for four of fifteen pairwise comparisons (Table 4), and the degree of elevation is substantial (Figure 3A-D). Although the small number of loci in our neutral proxy dataset makes it difficult to estimate precise significance values, we may approximate a denser probability distribution by bootstrapping over loci [58]. In this test, the difference between the PDYN FST and the 18-locus estimate of FST is significantly higher than the bootstrapped differences (p < 0.001) in the four comparisons. Moreover, PDYN has the second or third highest FST in four more comparisons; the sum of FST ranks across all 15 comparisons is significantly low (p = 0.01), although this p-value cannot be taken at face value due to the non-independence of the pairwise comparisons.
Figure 3 Elevated Differentiation at PDYN
(A–D) In four pairwise comparisons, FST at the PDYN 68-bp element (red) is markedly elevated above the FST estimated from 18 candidate neutral markers (blue) typed in the same individuals.
(E) Genetic differentiation between European- and Chinese-Americans, measured as a 15-SNP running FST average, for the entire p-arm of Chromosome 20. PDYN falls under a large FST peak (shaded), high above the arm average (red line). The RefSeq and chromosome band annotation is from the University of California, Santa Cruz Human Genome Browser (hg17), http://genome.ucsc.edu [79]. Perlegen SNP positions were matched to the hg17 assembly by the UCSC LiftOver utility.
(F) A finer-scale sliding window analysis shows that the region of elevated FST includes only two genes, PDYN and STK35, shown according to their RefSeq annotations.
(G) FST as a function of expected global heterozygosity. Red triangles represent the 52 SNPs in the Perlegen dataset in the 170-kb interval bounded by the 3′ ends of PDYN and STK35. The contours define the genome-wide density of FST conditioned on heterozygosity; for each heterozygosity, the lines represent the FST of SNPs in the specified FST percentile.
Table 4 Pairwise FST at PDYN and at Neutral Markers
If the elevated FST at PDYN is due to positive selection favoring different alleles in different populations, the signature of selection should also be visible in nearby variants, whose evolutionary fates are tied to the selected variant by linkage. We therefore asked whether the PDYN locus falls within an extended region of elevated FST. We investigated only Chinese-European FST, the population contrast for which our data suggested elevated FST (Figure 3A) and for which a genomic dataset was available. We used a dataset of 1,236,401 autosomal SNPs genotyped in African-, European-, and Chinese-Americans [59]. Because SNP ascertainment can influence the distribution of polymorphism statistics, we limited ourselves to SNPs ascertained by a single scheme: specifically, array-based resequencing of chromosomes from the National Institutes of Health Polymorphism Discovery Resource, a global sample. Because the 1.2 million SNPs share a common ascertainment bias, variation in FST along the chromosomes will reflect only variation in the demographic and selective history of genomic regions.
As an initial screen, we generated a 15-SNP sliding window plot of FST, considering only SNPs whose expected global heterozygosity exceeds 0.30. This filter is necessary to remove the dependence of FST on heterozygosity; otherwise, the plot would primarily reflect variation in the allele frequencies of the genotyped SNPs rather than differentiation among populations. As Figure 3E shows, PDYN falls within a tall and broad peak in FST. A finer scale sliding window plot (Figure 3F) indicates that the region of elevated FST encompasses two genes, PDYN and a serine/threonine kinase (STK35) implicated in cytoskeletal regulation [60]. These genes are divergently transcribed, and their intergenic region therefore likely contains the majority of cis-regulatory DNA for both genes. The 3′ flanking regions of each gene also exhibit elevated FST.
The genome-wide empirical distribution of FST is shaped by both demography and selection, and therefore the tail probabilities of SNPs estimated from the empirical distribution represent a very conservative test for selection. Nevertheless, the SNPs within the PDYN-STK35 FST peak exhibit significantly elevated FSTs. In Figure 3G, we plot FST versus expected global heterozygosity for all 52 genotyped SNPs in the 170-kb interval defined by PDYN and STK35 (i.e., excluding the 3′ flanking SNPs). We also plot the contours of the genome-wide FST distribution conditioned on heterozygosity; note that the median FST is below 0.06 for all heterozygosities. Six of the 52 SNPs in this region (12%) have FSTs in the top 0.5% of the genome-wide distribution, and 20 of the 52 (38%) are in the top 5%.
The number and location of selected variants driving elevation of FST remain unclear. However, neither PDYN nor STK35 is known to contain any non-synonymous variants, and neither protein sequence exhibits evidence of positive selection during human evolution [13]. The target or targets of selection are therefore likely to be cis-regulatory and to include the alleles of the 68-bp element.
Positive selection driving differentiation between populations should also decrease variation within populations; as a selected allele increases in frequency, its haplotype replaces other haplotypes before accumulating new variation. Microsatellites are particularly sensitive monitors of linked selection because of their high levels of polymorphism and high mutation rate. We asked whether the microsatellite nearest the PDYN promoter 68-bp element, a (CA)13–27 dinucleotide microsatellite 1.3 kb further upstream, exhibits the predicted signatures of selection. We genotyped the microsatellite in our panel of six populations (Figure 4A), and we used repeat-number variance and expected heterozygosity as summary statistics (Table 5).
Figure 4 Altered Variation at the PDYN Microsatellite
(A) The allele frequency distribution of the PDYN microsatellite for six populations. The most common allele has 18 CA repeats in each population except Papua New Guinea, where the 22-repeat allele is most common; the overall range is 13 to 27 repeats. The distributions show a reduction in allelic variation outside of the Cameroon population.
(B) The empirical probability density of lnRV for a panel of genomically distributed microsatellites is plotted for each population, using panel A as the color key. The distributions are based on 193 microsatellite loci for Ethiopia and 377 loci for the other populations. For clarity, a single negative outlier from the New Guinea population has been omitted from the figure. The arrows indicate lnRV of the PDYN microsatellite for each population, in the left tails of the distributions, indicating a locus-specific reduction in repeat-number variance.
(C) The empirical probability density for lnRH. Again, the PDYN microsatellite exhibits significantly negative lnRH values, indicating a locus-specific reduction in heterozygosity at PDYN in the non-West African populations.
Table 5
PDYN Microsatellite Summary Statistics
Repeat-number variance and heterozygosity are functions of θ = 4Neμ [61]. Because microsatellites vary in their mutation rates (μ) and recombinational contexts (which influences Ne), we used test statistics that control for these effects. For a given microsatellite, mutation rate and recombinational context are expected to be shared among populations, so they cancel out in a ratio. The ratio, Rθ, therefore estimates the relative effective sizes of two populations controlling for locus-specific phenomena; remaining variation among neutral microsatellites is attributable to stochastic variation in the outcomes of a neutral coalescent process [62,63]. Positive selection in one population will reduce heterozygosity and repeat-number variance at a linked microsatellite, causing it to appear in the tails of the estimated distributions of lnRV and lnRH (where repeat-number variance and heterozygosity are used in place of θ).
We estimated lnRθ distributions empirically from a genome-wide dataset of 337 autosomal loci [64, 65]. Because our FST data do not indicate recent selection in the sample from Cameroon, we used Cameroon as the denominator in all ratios, and we tested for positive selection in the other populations. Those in which positive selection has acted are predicted to exhibit significantly negative lnRθ at the PDYN microsatellite, unless the Cameroon sample has experienced equal or more extreme positive selection at a PDYN-linked locus.
We found a significant reduction in repeat-number variance at the PDYN microsatellite (Figure 4B) in three populations (Italy, p = 0.031; India, p = 0.034; China, p = 0.021), but not in Ethiopia (p = 0.103) or Papua New Guinea (p = 0.209). The sum of lnRV ranks across populations places PDYN in the 2.5% tail of lowest sums among all the microsatellites. The reduction in heterozygosity at PDYN (Figure 4C) is even more extreme (p < 0.003 for Italy and India, p < 0.006 for Ethiopia, p = 0.016 for China, and p = 0.072 for Papua New Guinea). The PDYN microsatellite is the locus with the lowest lnRH rank summed over populations.
The relationship between the events reducing variation at the PDYN microsatellite and the events elevating FST at the 68-bp repeat is most obvious when the haplotypic phase between the two elements is considered. We calculated expected heterozygosity and repeat-number variance in subsets of our experimentally determined haplotypes from an Austrian population. As shown in Table 5, the overall reduction in microsatellite heterozygosity and repeat-number variance is driven by the rapid elevation in frequency of the three-repeat allele at the 68-bp element.
The combination of elevated FSTs and reduced lnRθs implies that the selection occurred in multiple populations, favoring the two-repeat allele in China and India, and the three-repeat allele in Italy and Ethiopia. However, it remains possible that the 68-bp element in the PDYN promoter is not itself the target of selection, as the entire PDYN-STK35 region bears the signature of recent positive selection.
Discussion
The phylogenetic and population genetic data described above are difficult to reconcile with a simple selective scenario. Instead, they point to a complex selective history at the human PDYN locus, with ancient positive selection acting across the species and more recent positive selection favoring different alleles in local contexts.
The data allow us to construct a coherent and plausible model that accounts for each observation. During the course of human descent from our last common ancestor with chimpanzees, multiple non-coding mutations arising upstream of the start of PDYN transcription altered the gene's cis-regulation and swept to fixation due to positive selection, as indicated by analysis of the primate sequence data and the elevated frequency of derived alleles flanking the fixations. Concurrently, mutations altering the neuroactive peptide products of PDYN were eliminated by negative selection. The proximate effect of the fixed cis-regulatory mutations is the upregulation of PDYN transcription, particularly when induced by intracellular calcium release. Subsequent to the selective sweeps, but prior to the peopling of the globe, the 68-bp region encompassing the fixed mutations duplicated. The timing of these events is supported by the presence of the fixed differences on all copies of the repeat, the high frequency of flanking derived mutations in all repeat-number allele classes, and the presence of the duplication in all sampled human populations. The duplication segregates today as a tandem repeat polymorphism, with one to four repeats. After the global human diaspora, human populations in different parts of the world experienced different regimes of selection on PDYN cis-regulation, as indicated by the elevated FST values. Selection drove an increase in the frequency of the three-repeat allele in Europe and East Africa and independently increased the frequency of the two-repeat allele in India and China, according to the significantly reduced lnRθ values in each of the populations implicated by the FST data.
The convergence of independent lines of evidence—phylogenetic, population genetic, and functional evidence for ancient selection, and evidence of recent selection in patterns of variation within and among modern human populations—underscores the importance of PDYN to human biology. Though PDYN has received little attention from human geneticists, our evolutionary genetic data suggest that the locus would repay further investigation. While we may point to possible environmental and cultural agents of recent selection, including differences in use of plant opiates and environmental inducers of endogenous opioids, such as acupuncture [66], the phenotype targeted by the ancient selection is unknown.
For thousands of years, people have used opiates to alter consciousness and ameliorate pain. Our data indicate that the evolution of our species involved changes in the inducibility of an endogenous opioid precursor, and that these changes were driven by positive natural selection. Changes in neuropeptide expression are known to have accompanied behavioral evolution in other species [67,68], but the difficulties in studying such changes in gene expression in living human brains have prevented their discovery until now. PDYN, a natural candidate for human-specific traits by virtue of its documented role in perception, emotion, nociception, and learning is the first documented instance of a neural gene whose cis-regulation has been shaped by positive selection during human origins. Although the transcriptional effects of the selected changes in the PDYN promoter appear to be subtle, slight changes in gene expression are capable of substantial effects on organismal phenotypes [69]. In keeping with the predictions of King and Wilson [21], our data imply that minor changes in gene regulation played a significant role in the evolution of the traits that make us human.
Materials and Methods
Cloning and sequencing
Our human haplotypes are from an anonymized collection of genomic DNA samples from an Austrian population [39]. To guarantee recovery of rare one- and four-repeat alleles, we selected DNAs of known repeat-number genotype (note that representative samples were chosen for population genetic analyses, described below). We PCR amplified 3-kb fragments of PDYN promoter from genomic DNA, using high-fidelity Phusion polymerase (Finnzymes, Espoo, Finland). For repeat-number heterozygotes, PCR products were cloned into pGL3-basic vector, using an invariant Acc65I site at the 5′ end of the promoter and a NheI site incorporated into the PCR primer at the 3′ end. For each haplotype, we completely sequenced multiple clones, and all singletons were verified by bidirectional direct sequencing. Repeat-number homozygotes were sequenced directly from PCR products; in cases of multiple-site heterozygosity, these PCR products were also cloned and multiple clones sequenced to determine phase. Non-human primate DNA was acquired from Coriell Repositories (Camden, New Jersey, United States) (Pan troglodytes, Pan paniscus, Gorilla gorilla, Pongo pygmaeus, Macaca nemestrina, and Macaca mulatta), and as gifts from A. Stone (Pan troglodytes and Pan paniscus), and D. Loisel (Papio papio). For each sample, Phusion PCR products were cloned and sequenced as above. We sequenced the two coding exons directly from PCR products. In addition to non-human primates, we included four Austrian samples with known promoter repeat genotype to ensure recovery of coding sequence linked to each repeat-number allele. Sequences were scored using Sequencher (GeneCodes Corporation, Ann Arbor, Michigan, United States).
Phylogenetic and population genetic analyses
For all tests of substitution densities and rates, we conservatively assume that the site segregating among human repeat alleles represents a new mutation within humans and not a sixth fixed difference between humans and chimpanzees. To calculate the Poisson probability of five substitutions in 68 bp on the human branch, we found the expectation by considering the local average substitution rate, the genomic average divergence from chimpanzees, or the estimated genomic average mutation rate. The local average substitution rate, estimated from the human branch length for the entire 3-kb promoter sequence, including the 68-bp element, yields an expectation of 0.46 substitutions per 68 bp. At the broader scale of whole chromosomes, human and chimpanzee differ by nucleotide substitutions at an average of 1.44% of sites [43]; if one half of the divergence occurred on the human branch, the expected number of substitutions per 68 bp is 0.49. If instead of divergence, we consider the germline mutation rate, estimated at 0.99 × 10−9 per site, and assuming 5 to 7 million years of evolution since the last common ancestor of humans and chimpanzees [44], we expect 0.34 to 0.47 substitutions per 68 bp. Note that none of the substitutions occurs in a CpG context, although CpG may have been an intermediate in the adjacent substitutions that changed CpA in non-human primates to GpG in humans. The five substitutions are C→G, A→G, A→G, T→C, and C→A, three transitions and two transversions.
Molecular clock and relative ratio tests were implemented in HYPHY (http://www.hyphy.org) using an eight-sequence dataset, with a single allele representing each species. As our human exemplar we used the most common one-repeat haplotype from our sample of ten Austrian one-repeat alleles; we chose a one-repeat haplotype to facilitate comparison with the one-repeat sequences of non-human primates. The most common chimpanzee haplotype represented that species, while for other species, because each haplotype is unique in our small sample, a haplotype was selected randomly. Molecular clock tests used best-fit time reversible substitution models selected using ModelTest [70]. For the coding sequence and the promoter excluding the repeat, the favored model is HKY with Γ-distributed among-site rate variation. We used the maximum likelihood-estimated transition/transversion ratio and rate variation shape parameter and empirical base frequencies. For the repeat, the favored model is K2P. The relative ratio tests were performed using the HKY + Γ model, but results with K2P are very similar.
Negative selection on neuropeptides was tested by calculating the Poisson probability that zero of 25 variable positions would fall in the 56 of 254 amino acids comprising the neuropeptides. To test for positive selection in the remainder of the protein, we compared models 1 and 2 (2δ = 0.24364, p = 0.62) and models 7 and 8 (2δ = 0.0001, p = 0.99) from Yang et al. [47], in HYPHY, using the Goldman-Yang parameterization with base frequencies independent of codon position. The dataset included one sequence from each species.
For haplotype-based tests, we generated a representative population sample [71] by drawing 74 haplotypes from the sequenced Austrian haplotypes according to the population frequency of the different repeat-number alleles [36]. Summary statistics and their p-values were found using DNAsp [72].
Intensity of selection
Estimation of the rate acceleration factor (A) involves three data partitions whose evolution is consistent with neutrality. First, we found the maximum likelihood estimate of the ratio of substitution rates between the 68-bp element and the remainder of the promoter, excluding the human lineage (ratio = 1.804). Next, we estimated the substitution rate on the human lineage for the portion of the promoter excluding the 68-bp element (rate = 0.00506 substitutions per site), holding the substitution model parameters constant. The product of these gives the expectation for the human 68-bp repeat under neutrality, 0.00913 substitutions per site. (Note that the Poisson probability of five substitutions, given that expectation, is 0.00005, and three or four mutations also fall in the 0.025 tail of the Poisson probability.) The maximum likelihood estimate of the substitution rate in the 68-bp element along the human lineage, 0.0926 substitutions per site, represents an acceleration factor A = 10.1. All estimates employed the HKY + Γ substitution model.
The genic selection coefficient s is estimated from the relations
and f
a + f
d + f
0 = 1, where fa, fd, and f0 are the fraction of mutations that are advantageous, deleterious, and neutral, respectively. Figure S1 shows s as a function of the nuisance parameters fa and fd. We make the approximating assumption that fa, fd, and f0 are constant over the course of the selective history of the locus.
To convert the acceleration factor to s, we consider the case of sequential fixations and ignore the effect of interference among independent advantageous mutations. The effect of interference is likely to be modest, as the expectation of the conditional fixation time of advantageous alleles, ~(2/s)(ln2N) [73], is less than 10,000 generations for s > 0.002, while the time available for fixations is roughly 300,000 generations (6 million years, 20 years per generation). Our estimate of s is based on the long-term effective population size since the divergence of humans and chimpanzees, which may be much larger than the estimate for modern humans. Larger Ne translates into even lower estimates of s. The fixation probability (~2s for constant Ne) is sensitive to fluctuations in effective population size [74], increasing during population expansions and decreasing during bottlenecks. Our simple approach assumes constant effective size.
The magnitude of the estimated rate acceleration excludes non-reciprocal exchange processes subsequent to duplication (e.g., gene conversion and unequal crossing-over) as possible explanations for the human-specific acceleration. Unbiased non-reciprocal exchanges do not alter substitution rates; although the number of sites available to mutate is increased by the number of repeat elements (n), the probability that a new mutation will spread among the repeats is 1/n. Biased processes can accelerate substitution [75], but only when the bias consistently favors new mutations over ancestral alleles. Even in the most extreme case, where every inter-repeat conversion event replaces an ancestral allele with a new mutation, the maximum rate acceleration is n. In human PDYN, n averages less than three and never exceeds four, and the long-term effective number of repeats (the harmonic mean of repeat number over the duration of the human lineage) is likely quite close to one. Strong bias is at any rate ruled out by the presence of a segregating variant among the repeats. So non-reciprocal exchange cannot produce the observed acceleration factor under neutrality. Under positive selection, however, tandem repeats, with or without biased conversion, can increase the power of deterministic forces relative to drift by increasing the effective population size to Nen [76].
Vectors, cell culture, and transfection
We used the pGL3basic luciferase reporter (Promega, Madison, Wisconsin, United States). Chimpanzee and human constructs were generated as described above (“Cloning and sequencing”). To generate chimeric constructs with DRE site swaps, we cut the inserts with BstAPI and exchanged the DRE-containing fragments. To generate a 68-bp element chimera, we excised a human repeat using BspHI and DrdI and inserted it into a chimpanzee vector in the equivalent position. Vectors were verified by sequencing. We cultured JAR choriocarcinoma cells in RPMI 1640 with 2 mM L-glutamine and 10 mM HEPES, supplemented with 10% FBS. SH-SY5Y neuroblastoma cells were cultured in a 1:1 mixture of Ham's F-12 and EMEM with 1 mM sodium pyruvate and 0.1 mM non-essential amino acids, supplemented with 10% FBS. Both cell lines were acquired from
ATCC and maintained at 37 °C with 5% CO2. We performed transfections in 24-well plates with JAR cells at 90% confluence and SH-SY5Y cells at 50% confluence. The transfection mix, in OPTI-MEM, included 2 μl of Lipofectamine2000, 0.72 μg of pGL3, and 0.08 μg of Renilla-TK (Promega) as a co-reporter to control for variation in transfection efficiency. At 26 h, medium was supplemented with growth medium with or without caffeine (final concentration 10 mM). Cells were harvested 16 h later. Luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega) and a Turner Designs 20/20 luminometer. Results are reported as ratios of firefly:Renilla luciferase, standardized by setting the pGL3-basic ratio to one. Lysates from mock transfected cells were used to blank for machine background. All transfections were performed five to seven times, and effects assessed by analysis of variance and pairwise t-tests.
FST and lnRθ analysis
For the six-population analysis of FST, neutral markers, DNA samples, FST calculations, and bootstrap resampling are as previously described [58]. We genotyped the 68-bp repeat and the PDYN microsatellite by scoring the length of labeled PCR products run on an ABI 3700 capillary gel machine. We verified genotypes for 10% of samples by direct sequencing of PCR products.
Analysis of Perlegen data was limited to autosomal SNPs ascertained according to scheme A of Hinds et al. [59], array-based resequencing of National Institutes of Health Polymorphism Discovery Resource chromosomes. SNP data were downloaded from http://genome.perlegen.com/browser/download.html, and we used Perlegen's precalculated FST values. To calculate expected global heterozygosity, we averaged the allele frequencies for the three genotyped populations and used 2p-2p
2
, where p is the average frequency of the global minor allele. To generate the percentile plot for FST conditioned on expected heterozygosity, we sorted the SNPs into ten bins, each covering five percentage points of expected global heterozygosity range, we ranked SNPs within bins by FST, and then we recovered the FST for the SNP whose rank coincides with the relevant percentile within that bin. In Figure 3G, the contours are connecting the data points for the ten bins for each percentile; e.g., the point where the contours hit expected heterozygosity 0.5 represents the FST percentile for SNPs with heterozygosities 0.45 to 0.5.
Microsatellite data were downloaded from N. Rosenberg's Web site, http://www.cmb.usc.edu/people/noahr/diversity.html, in structure format. The Rosenberg data lacked a population to match to our Ethiopian population, but data for 193 of the 377 loci were available from the dataset of Kayser et al. [65]. To represent our populations, we selected the populations in the Rosenberg data that are geographically coincident or proximate. As the distributions are quite similar for all populations (Figure 4B and 4C), the precision of population matching appears unimportant (also found by [65]). We selected as follows from the Rosenberg et al. data: Cameroon: Yoruba; China: Han Chinese; India: pooled samples from the populations included in Rosenberg et al.'s South Asia cluster, specifically Brahui, Balochi, Hazara, Makrani, Sindhi, Pathan, and Burusho; Italy: pooled samples from Sardinia, Tuscany, and Bergamo; Papua New Guinea: Papuan. For each microsatellite locus, expected heterozygosity was calculated as
, where n is the number of chromosomes sampled, K is the number of alleles, and p
k is the frequency of the kth allele. Repeat-number variance was calculated as
where x is the number of repeats in the nth chromosome.
The statistical properties of lnRV and lnRH allow us to estimate significance values in a parametric context, reducing the influence of any non-neutral outliers in the tails of the empirical distribution. Each of the five lnRV distributions is consistent with normality, according to a Kolmogorov-Smirnov test. Standardizing our observed test statistics according to the empirical mean and standard deviation, and using the tail probabilities of the standard normal distribution, we recover p-values nearly identical to those drawn from the empirical distribution (Italy: 0.034; India: 0.035; China: 0.016; Ethiopia: 0.119; Papua New Guinea: 0.236). Of the lnRH distributions, only the Ethiopian sample conforms to normality according to the Kolmogorov-Smirnov test, conferring a parametric p-value of 0.0002 on the Ethiopian PDYN microsatellite.
The causes of the departures of lnRH from normality are unclear, but ascertainment bias is an obvious possibility. The 377 microsatellites were ascertained in a European population and may be biased against microsatellites with low heterozygosity in Europe. In general, ascertainment bias is expected to be very modest for microsatellites because low-heterozygosity microsatellites are quite rare [62,64,77]. As Rosenberg et al. [64] note, their microsatellite data are very similar to data from microsatellites ascertained in independent, geographically diverse panels. Nevertheless, we must consider the possibility that ascertainment bias contributes to the shapes of the empirical lnRV and lnRH distributions. The expected effect of ascertainment bias is the truncation of the left tails of the distributions, due to the exclusion of loci with low heterozygosity in Europeans, and consequently the extension of the right tails. Two predictions are thus a departure from normality and a significantly positive skew, measured by the third moment about the mean. As noted, all lnRV distributions are consistent with normality and have well-behaved tail probabilities. Only the Ethiopian lnRV distribution has a significantly positive skew. For lnRH, four of five distributions fail the Kolmogorov-Smirnov test for normality, but the departures appear to be due largely to elevated kurtosis, not to positive skew. Only the Italian distribution has a significantly positive skew. We can construct a conservative test by drawing p-values from the right tails of the lnRθ distributions, which should be enlarged relative to the unbiased case; all lnRθ values are as extreme relative to the right tails as to the left, except for Ethiopian lnRθ values, whose p-values rise to 0.016 (lnRH) and 0.119 (lnRV). Because lnRH has a smaller coalescent variance than lnRV, lnRH is exquisitely sensitive to selection [63], and the observed departures from normality may therefore simply reflect the occurrence of selection at sites linked to some subset of the 377 loci [65,78]. The complete microsatellite dataset is presented in Table S2.
Supporting Information
Figure S1 Intensity of Positive Selection
The average selection coefficient (s) of advantageous mutations can be estimated from the rate acceleration of the human 68-bp element, conditioned on the fractions of all mutations that are advantageous, neutral, and deleterious. When the advantageous fraction is more than 2.5%, the average selection coefficient is less than 0.01. Over most of the parameter space, s is less than 0.001. The red line illustrates the case in which all and only the five fixed mutations are advantageous.
(1.7 MB EPS).
Click here for additional data file.
Table S1 Experimentally Determined PDYN Haplotypes from 74 Austrian Chromosomes
Each unique haplotype is shown, from a sample of chromosomes selected to overrepresent the rare one- and four-repeat alleles. Derived alleles are highlighted in red. Haplotypes were determined by complete sequencing of multiple clones of each allele. Msat refers to the CAn microsatellite 1.3 kb upstream of the 68-bp repeat. Position 2370 segregates TC7 and TC9 alleles. The ancestral states at Msat and 2370 are uncertain, as both sites vary among and within the other primate species.
(74 KB PDF).
Click here for additional data file.
Table S2 Microsatellite Summary Statistics
For the PDYN microsatellite and those used to generate the genome-wide empirical distributions, we report the sample sizes, expected heterozygosities, and repeat-number variances, as well as the test statistics lnRH and lnRV.
(223 KB XLS).
Click here for additional data file.
Accession Numbers
DNA sequences have been submitted to GenBank (http://www.ncbi.nlm.nih.gov.Genbank) with accession numbers AY902542–AY902679.
We thank Lisa Bukovnik, Manny Lopez, and Anjali Patel for assistance in the lab and Cliff Cunningham, Greg Gibson, Fred Nijhout, and Mark Rausher for helpful comments. Thanks to Mark Stoneking for sharing data. This work was supported by a Royal Society/Wolfson Research Merit Award to DBG, and by grants and fellowships from the Leverhulme Trust to NS and DBG, the National Science Foundation to MVR, MWH, and GAW, and NASA to GAW.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. MVR, MWH, NS, FZ, DBG, and GAW conceived and designed the experiments, performed the experiments, analyzed the data, contributed reagents/materials/analysis tools, and wrote the paper.
¤a Current address: Lewis-Sigler Institute for Integrative Genomics, Princeton University, Princeton, New Jersey, United States of America
¤b Current address: Department of Biology and School of Informatics, Indiana University, Bloomington, Indiana, United States of America
Citation: Rockman MV, Hahn MW, Soranzo N, Zimprich F, Goldstein DB, et al. (2005) Ancient and recent positive selection transformed opioid cis-regulation in humans. PLoS Biol 3(12): e387.
Abbreviations
bpbase-pair
DREdownstream regulatory element
DREAMDRE-antagonist modulator
kbkilobase
PDYNprodynorphin
SNPsingle nucleotide polymorphism
==== Refs
References
Chou HH Hayakawa T Diaz S Krings M Indriati E Inactivation of CMP-N-acetylneuraminic acid hydroxylase occurred prior to brain expansion during human evolution Proc Natl Acad Sci U S A 2002 99 11736 11741 12192086
Stedman HH Kozyak BW Nelson A Thesier DM Su LT Myosin gene mutation correlates with anatomical changes in the human lineage Nature 2004 428 415 418 15042088
Winter H Langbein L Krawczak M Cooper DN Jave-Suarez LF Human type I hair keratin pseudogene phihHaA has functional orthologs in the chimpanzee and gorilla: Evidence for recent inactivation of the human gene after the Pan-Homo divergence Hum Genet 2001 108 37 42 11214905
Fortna A Kim Y MacLaren E Marshall K Hahn G Lineage-specific gene duplication and loss in human and great ape evolution PLoS Biol 2004 2 e207 10.1371/journal.pbio.0020207 15252450
Enard W Przeworski M Fisher SE Lai CS Wiebe V Molecular evolution of FOXP2, a gene involved in speech and language Nature 2002 418 869 872 12192408
Andres AM Soldevila M Navarro A Kidd KK Oliva B Positive selection in MAOA gene is human exclusive: Determination of the putative amino acid change selected in the human lineage Hum Genet 2004 5 377 386
Zhang J Evolution of the human ASPM gene, a major determinant of brain size Genetics 2003 165 2063 2070 14704186
Ferland RJ Eyaid W Collura RV Tully LD Hill RS Abnormal cerebellar development and axonal decussation due to mutations in AHI1 in Joubert syndrome Nat Genet 2004 36 1008 1013 15322546
Evans PD Anderson JR Vallender EJ Choi SS Lahn BT Reconstructing the evolutionary history of microcephalin, a gene controlling human brain size Hum Mol Genet 2004 13 1139 1145 15056607
Huby T Dachet C Lawn RM Wickings J Chapman MJ Functional analysis of the chimpanzee and human apo(a) promoter sequences: Identification of sequence variations responsible for elevated transcriptional activity in chimpanzee J Biol Chem 2001 276 22209 22214 11301336
Heissig F Krause J Bryk J Khaitovich P Enard W Functional analysis of human and chimpanzee promoters Genome Biol 2005 6 R57 15998446
Clark AG Glanowski S Nielsen R Thomas PD Kejariwal A Inferring nonneutral evolution from human-chimp-mouse orthologous gene trios Science 2003 302 1960 1963 14671302
Nielsen R Bustamante C Clark AG Glanowski S Sackton TB A scan for positively selected genes in the genomes of humans and chimpanzees PLoS Biol 2005 3 e170 10.1371/journal.pbio.0030170 15869325
Dorus S Vallender EJ Evans PD Anderson JR Gilbert SL Accelerated evolution of nervous system genes in the origin of Homo sapiens
Cell 2004 119 1027 1040 15620360
Hahn MW Rockman MV Soranzo N Goldstein DB Wray GA Population genetic and phylogenetic evidence for positive selection on regulatory mutations at the factor VII locus in humans Genetics 2004 167 867 877 15238535
Lowe CJ Wray GA Radical alterations in the roles of homeobox genes during echinoderm evolution Nature 1997 389 718 721 9338781
Stern DL Evolutionary developmental biology and the problem of variation Evolution Int J Org Evolution 2000 54 1079 1091
Carroll SB Grenier JK Weatherbee SD From DNA to diversity: Molecular genetics and the evolution of animal design 2001 London Blackwell Science 214
Davidson EH Genomic regulatory systems: Development and evolution 2001 San Diego Academic Press 261
Wray GA Hahn MW Abouheif E Balhoff JP Pizer M The evolution of transcriptional regulation in eukaryotes Mol Biol Evol 2003 20 1377 1419 12777501
King MC Wilson AC Evolution at two levels in humans and chimpanzees Science 1975 188 107 116 1090005
Carroll SB Evolution at two levels: On genes and form PLoS Biol 2005 3 e245 10.1371/journal.pbio.0030245 16000021
Enard W Khaitovich P Klose J Zollner S Heissig F Intra- and interspecific variation in primate gene expression patterns Science 2002 296 340 343 11951044
Hsieh WP Chu TM Wolfinger RD Gibson G Mixed-model reanalysis of primate data suggests tissue and species biases in oligonucleotide-based gene expression profiles Genetics 2003 165 747 757 14573485
Karaman MW Houck ML Chemnick LG Nagpal S Chawannakul D Comparative analysis of gene-expression patterns in human and African great ape cultured fibroblasts Genome Res 2003 13 1619 1630 12840040
Caceres M Lachuer J Zapala MA Redmond JC Kudo L Elevated gene expression levels distinguish human from non-human primate brains Proc Natl Acad Sci U S A 2003 100 13030 13035 14557539
Rockman MV Wray GA Abundant raw material for cis-regulatory evolution in humans Mol Biol Evol 2002 19 1991 2004 12411608
Hoogendoorn B Coleman SL Guy CA Smith K Bowen T Functional analysis of human promoter polymorphisms Hum Mol Genet 2003 12 2249 2254 12915441
Buckland PR Hoogendoorn B Guy CA Coleman SL Smith SK A high proportion of polymorphisms in the promoters of brain expressed genes influences transcriptional activity Biochim Biophys Acta 2004 1690 238 249 15511631
Horan M Millar DS Hedderich J Lewis G Newsway V Human growth hormone 1 (GH1) gene expression: Complex haplotype-dependent influence of polymorphic variation in the proximal promoter and locus control region Hum Mutat 2003 21 408 423 12655556
Cheng HY Pitcher GM Laviolette SR Whishaw IQ Tong KI DREAM is a critical transcriptional repressor for pain modulation Cell 2002 108 31 43 11792319
Rodgers RJ Cooper SJ Endorphins, opiates and behavioural processes 1988 New York Wiley 361
Moles A Kieffer BL D'Amato FR Deficit in attachment behavior in mice lacking the mu-opioid receptor gene Science 2004 304 1983 1986 15218152
Wagner JJ Terman GW Chavkin C Endogenous dynorphins inhibit excitatory neurotransmission and block LTP induction in the hippocampus Nature 1993 363 451 454 8099201
Roth BL Baner K Westkaemper R Siebert D Rice KC Salvinorin A: A potent naturally occurring nonnitrogenous kappa opioid selective agonist Proc Natl Acad Sci U S A 2002 99 11934 11939 12192085
Zimprich A Kraus J Woltje M Mayer P Rauch E An allelic variation in the human prodynorphin gene promoter alters stimulus-induced expression J Neurochem 2000 74 472 477 10646497
Ventriglia M Bocchio Chiavetto L Bonvicini C Tura GB Bignotti S Allelic variation in the human prodynorphin gene promoter and schizophrenia Neuropsychobiology 2002 46 17 21 12207142
Chen AC LaForge KS Ho A McHugh PF Kellogg S Potentially functional polymorphism in the promoter region of prodynorphin gene may be associated with protection against cocaine dependence or abuse Am J Med Genet 2002 114 429 435 11992566
Stogmann E Zimprich A Baumgartner C Aull-Watschinger S Hollt V A functional polymorphism in the prodynorphin gene promotor is associated with temporal lobe epilepsy Ann Neurol 2002 51 260 263 11835385
Hurd YL Subjects with major depression or bipolar disorder show reduction of prodynorphin mRNA expression in discrete nuclei of the amygdaloid complex Mol Psychiatry 2002 7 75 81 11803449
Hurd YL Herkenham M Molecular alterations in the neostriatum of human cocaine addicts Synapse 1993 13 357 369 7683144
Solbrig MV Koob GF Epilepsy, CNS viral injury, and dynorphin Trends Pharmacol Sci 2004 25 98 104 15102496
Watanabe H Fujiyama A Hattori M Taylor TD Toyoda A DNA sequence and comparative analysis of chimpanzee chromosome 22 Nature 2004 429 382 388 15164055
Yi S Ellsworth DL Li WH Slow molecular clocks in Old World monkeys, apes, and humans Mol Biol Evol 2002 19 2191 2198 12446810
Felsenstein J Evolutionary trees from DNA sequences: A maximum likelihood approach J Mol Evol 1981 17 368 376 7288891
Muse SV Gaut BS Comparing patterns of nucleotide substitution rates among chloroplast loci using the relative ratio test Genetics 1997 146 393 399 9136027
Yang Z Nielsen R Goldman N Pedersen AM Codon-substitution models for heterogeneous selection pressure at amino acid sites Genetics 2000 155 431 449 10790415
Maynard Smith J Haigh J The hitch-hiking effect of a favourable gene Genet Res 1974 23 23 35 4407212
Fay JC Wu CI Hitchhiking under positive Darwinian selection Genetics 2000 155 1405 1413 10880498
Kimura M The neutral theory of molecular evolution 1983 Cambridge Cambridge University Press 367
Haldane JBS The mathematical theory of natural and artificial selection Proc Camb Philos Soc 1927 23 838 844
Yu N Jensen-Seaman MI Chemnick L Kidd JR Deinard AS Low nucleotide diversity in chimpanzees and bonobos Genetics 2003 164 1511 1518 12930756
Bersaglieri T Sabeti PC Patterson N Vanderploeg T Schaffner SF Genetic signatures of strong recent positive selection at the lactase gene Am J Hum Genet 2004 74 1111 1120 15114531
Tishkoff SA Varkonyi R Cahinhinan N Abbes S Argyropoulos G Haplotype diversity and linkage disequilibrium at human G6PD: Recent origin of alleles that confer malarial resistance Science 2001 293 455 462 11423617
Carrion AM Link WA Ledo F Mellstrom B Naranjo JR DREAM is a Ca2+-regulated transcriptional repressor Nature 1999 398 80 84 10078534
Storz JF Using genome scans of DNA polymorphism to infer adaptive population divergence Mol Ecol 2005 14 671 688 15723660
Lewontin RC Krakauer J Distribution of gene frequency as a test of the theory of the selective neutrality of polymorphisms Genetics 1973 74 175 195 4711903
Rockman MV Hahn MW Soranzo N Goldstein DB Wray GA Positive selection on a human-specific transcription factor binding site regulating IL4 expression Curr Biol 2003 13 2118 2123 14654003
Hinds DA Stuve LL Nilsen GB Halperin E Eskin E Whole-genome patterns of common DNA variation in three human populations Science 2005 307 1072 1079 15718463
Vallenius T Makela TP Clik1: A novel kinase targeted to actin stress fibers by the CLP-36 PDZ-LIM protein J Cell Sci 2002 115 2067 2073 11973348
Ohta T Kimura M A model of mutation appropriate to estimate the number of electrophoretically detectable alleles in a finite population Genet Res 1973 22 201 204 4777279
Schlotterer C A microsatellite-based multilocus screen for the identification of local selective sweeps Genetics 2002 160 753 763 11861576
Kauer MO Dieringer D Schlotterer C A microsatellite variability screen for positive selection associated with the “out of Africa” habitat expansion of Drosophila melanogaster
Genetics 2003 165 1137 1148 14668371
Rosenberg NA Pritchard JK Weber JL Cann HM Kidd KK Genetic structure of human populations Science 2002 298 2381 2385 12493913
Kayser M Brauer S Stoneking M A genome scan to detect candidate regions influenced by local natural selection in human populations Mol Biol Evol 2003 20 893 900 12717000
Han JS Acupuncture and endorphins Neurosci Lett 2004 361 258 261 15135942
Saetre P Lindberg J Leonard JA Olsson K Pettersson U From wild wolf to domestic dog: Gene expression changes in the brain Brain Res Mol Brain Res 2004 126 198 206 15249144
Lim MM Wang Z Olazabal DE Ren X Terwilliger EF Enhanced partner preference in a promiscuous species by manipulating the expression of a single gene Nature 2004 429 754 757 15201909
Oliver F Christians JK Liu X Rhind S Verma V Regulatory variation at glypican-3 underlies a major growth QTL in mice PLoS Biol 2005 3 e135 10.1371/journal.pbio.0030135 15799711
Posada D Crandall KA MODELTEST: Testing the model of DNA substitution Bioinformatics 1998 14 817 818 9918953
Hudson RR Kaplan NL On the divergence of alleles in nested subsamples from finite populations Genetics 1986 113 1057 1076 3744026
Rozas J Rozas R DnaSP version 3: An integrated program for molecular population genetics and molecular evolution analysis Bioinformatics 1999 15 174 175 10089204
Nei M Molecular evolutionary genetics 1987 New York Columbia University Press 512
Otto SP Whitlock MC The probability of fixation in populations of changing size Genetics 1997 146 723 733 9178020
Nagylaki T Petes TD Intrachromosomal gene conversion and the maintenance of sequence homogeneity among repeated genes Genetics 1982 100 315 337 7106560
Walsh JB Selection and biased gene conversion in a multigene family: Consequences of interallelic bias and threshold selection Genetics 1986 112 699 716 3957008
Weissenbach J Gyapay G Dib C Vignal A Morissette J A second-generation linkage map of the human genome Nature 1992 359 794 801 1436057
Storz JF Payseur BA Nachman MW Genome scans of DNA variability in humans reveal evidence for selective sweeps outside of Africa Mol Biol Evol 2004 21 1800 1811 15201398
Kent WJ Sugnet CW Furey TS Roskin KM Pringle TH The human genome browser at UCSC Genome Res 2002 12 996 1006 12045153
|
16274263
|
PMC1283535
|
CC BY
|
2021-01-05 08:21:45
|
no
|
PLoS Biol. 2005 Dec 15; 3(12):e387
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030387
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627983810.1371/journal.pbio.0030402Research ArticleAnimal BehaviorNeurosciencePhysiologyRattus (Rat)Theta Rhythms Coordinate Hippocampal–Prefrontal Interactions in a Spatial Memory Task Spatial Memory-Coordinated Theta RhythmsJones Matthew W
1
¤Wilson Matthew A [email protected]
1
1The Picower Institute for Learning and Memory, Department of Brain and Cognitive Sciences, RIKEN-MIT Neuroscience Research Center, Massachusetts Institute of Technology, Cambridge, Massachusetts, United States of AmericaMorris Richard Academic EditorUniversity of EdinburghUnited Kingdom12 2005 15 11 2005 15 11 2005 3 12 e40229 7 2005 22 9 2005 Copyright: © 2005 Jones and Wilson.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.
Synchronized Brain Interactions Associated with Memory and Decision-Making
Decision-making requires the coordinated activity of diverse brain structures. For example, in maze-based tasks, the prefrontal cortex must integrate spatial information encoded in the hippocampus with mnemonic information concerning route and task rules in order to direct behavior appropriately. Using simultaneous tetrode recordings from CA1 of the rat hippocampus and medial prefrontal cortex, we show that correlated firing in the two structures is selectively enhanced during behavior that recruits spatial working memory, allowing the integration of hippocampal spatial information into a broader, decision-making network. The increased correlations are paralleled by enhanced coupling of the two structures in the 4- to 12-Hz theta-frequency range. Thus the coordination of theta rhythms may constitute a general mechanism through which the relative timing of disparate neural activities can be controlled, allowing specialized brain structures to both encode information independently and to interact selectively according to current behavioral demands.
Simultaneous electrophysiological recordings from CA1 of the rat hippocampus and medial prefrontal cortex reveal enhanced correlated activity between the two structures selectively during behavior that recruits spatial working memory.
==== Body
Introduction
The coordinated, rhythmic activity of neuronal populations gives rise to oscillations in local field potentials (LFP) and electroencephalograms at a broad range of frequencies [1]. Throughout the brain, these oscillations potentially constitute clocking mechanisms against which to reference and coordinate the timing of neural firing. Synchronization of these rhythmic activities is likely to reflect or underlie functional interactions between neurons within a defined brain structure, or between disparate populations in distinct structures [2,3]. Equally, abnormal synchronization may impair functional interactions and contribute to complex cognitive disorders such as schizophrenia [4,5] and attention deficit/hyperactivity disorder [6–8].
At the level of single neurons, synchrony is evident in the consistent temporal relationships between the firing patterns of interconnected cells. These are most commonly quantified using cross-correlation techniques [9–11]. For example, the correlated firing of cortical neurons is implicated in visual processing [12], attention [13], and motor learning [14]. However, the majority of these studies are based on recordings from single brain regions, and are unable to address the nature of correlated activities in networks spanning multiple structures. Do correlations also underlie coordination between anatomically and functionally related brain regions? If so, are they also mediated or reflected by oscillatory population activities at the LFP level?
Theta rhythms are 4- to 12-Hz oscillations consistently associated with complex behaviors presumed to require mnemonic processing and/or decision-making, for example spatial exploration in rodents [15], working memory in primates [16], and navigation and working memory in humans [17,18]. Dynamic, behavioral modulation of theta rhythms may therefore indicate or mediate cross-neuronal and/or cross-structural interactions during these behaviors. Theta rhythms are found in many mammalian brain structures, but are most prominent in the rodent hippocampus [19]. Here, the firing of individual “place cells”—hippocampal principal excitatory neurons with spatial receptive fields [20]—is coordinated (“phase-locked”) with respect to the local theta rhythm. Thus the action potentials of a given neuron tend to occur during a preferred phase of the theta cycle. This phase-locking of hippocampal spike-timing to ongoing LFP oscillations is an important example of temporal coding in the brain [21] and—in concert with the related phenomenon of phase precession—has been proposed to allow higher-order coding of spatial information than that imparted by the firing-rate pattern alone [22]. It has also been proposed that hippocampal theta rhythms may coordinate neural activity during sensorimotor integration [23] or information encoding [24,25].
Neuronal firing phase-locked to the hippocampal theta rhythm has also been described in cingulate cortex [26], amygdala [27], entorhinal cortex [28], striatum [29], and, most recently, the rat prefrontal cortex [30,31]. As in the hippocampus, phase-locking in the prefrontal cortex is also accompanied by phase precession [32]. This raises the possibility that phase-locking may play a broader role in defining the temporal relationships between cross-structural activities. However, it remains to be established how these phase relationships influence the firing of connected neurons, and how—or whether—they relate to behavior or hippocampal function. For example, while Siapas et al. [30] included radial arm maze and T-maze tasks among the conditions during which they collected data, analyses made no attempt to relate medial prefrontal cortex (mPFC) phase-locking to ongoing behaviors. If mPFC phase-locking remains constant while behavioral demands vary, it is unlikely to reflect functional interactions between coactive structures. Similarly, Hyman et al. [31] recorded during running on a linear track and random foraging, and were therefore unable to explicitly relate variations in phase-locking with variations in behavioral demands.
Siapas et al. [30] suggested that mPFC phase-locking may play some role in the formation of long-term memories that require that transfer of information from the hippocampus to the neocortex. Consistent with this role, monosynaptic projections from hippocampus to the deep layers of mPFC [33,34] do exhibit activity-induced plasticity [35], and LFP oscillations in the two structures are correlated during slow-wave sleep [36]. However, lesions or disruptions of either mPFC or the hippocampus impair spatial working memory [37–39], providing functional evidence that the two structures also interact on-line during behaviors that require short-term mnemonic processing [40]. To investigate the nature of functional interactions that relate to decision-making, we made simultaneous tetrode recordings of extracellular action potentials and LFP from CA1 and mPFC to examine coordination of activity in these two structures during a spatial working-memory task designed to dissociate neural activity related to performance (e.g., running and orienting behaviors) from activity related to mnemonic or decision-making processes. We find that spike timing and theta-rhythmic activities in CA1 and mPFC become more coordinated during epochs of the task associated with peak mnemonic and decision-making load.
Results
Behavior
Data are presented from eight recording sessions from six rats. Rats ran 15–25 trials per 20- to 30-min recording session. Each trial of the task comprised a “forced-turn” (sample) and “choice” (test) epoch (Figure 1A), and was subdivided into a number of stages corresponding to different sections of the maze (Figure 1B). Rats were trained to asymptotic performance prior to electrode implantation (Figure 1C), and performed the task at 83% ± 5.0% (mean ± standard error of the mean) correct during the eight sessions presented here. Rats began at a reward point (F1 or F2), then ran towards the central arm of the maze (stage 1 in Figure 1B). A moveable barrier directed them down the central arm towards the choice point (stage 2), when they were required to choose a left turn if the trial started from F1, or a right turn if the trial started from F2. If the correct turn was made (stage 3), chocolate reward was delivered remotely to C1 or C2; incorrect turns were not reinforced. Rats then returned to the choice point (stage 4) and were directed by another moveable barrier back down the central arm towards the forced-turn end of the maze (stage 5). Here, the barrier directed them to reward at F1 or F2 (stage 6), with left or right turns selected at random from trial to trial. The location of the forced-turn end of the maze was varied between animals.
Figure 1 Experimental Design and Performance during the Spatial Working-Memory Task
(A) Schematic of the maze and a pair of runs comprising a single trial (forced-turn direction C2 to F1, solid arrow; choice direction F1 to C1, dotted arrow). Grey rectangle marks the moveable barrier.
(B) The task was broken down into distinct behavioral stages for analysis: 1, running away from the “cue” reward point towards the central arm; 2, crossing the central arm in the choice direction (only activity on the central three-quarters section of the arm was considered for analysis to avoid divergent routes near the turning points and inconsistent running behavior); 3, post-choice running to reward point (rats were rewarded for choosing C1 if the trial started at F1 and for choosing C2 if the trial started at F2); 4, returning to the central arm (where a second barrier blocked the route to the opposite reward point); 5, crossing the central arm in the “forced-turn” direction; and 6, returning to one of the two reward points (F1 or F2 chosen at random for each trial). Stages 1 and 2, marked by the red arrows, were presumed to involve working memory and/or decision-making.
(C) The six rats were trained to asymptotic working-memory performance for at least 12 d before tetrode implantation.
Stages 1 and 2 (choice epoch) presumably invoked spatial working-memory processes: rats were required to “hold in mind” the location of the starting reward—or the direction of the turn between stages 1 and 2—in order to choose between C1 and C2 at the opposite end of the maze. By stage 3, their decision had already been made, and any working-memory requirement was negated. Stages 4–6 (forced-turn epoch) never required active working memory, since routes back to F1 or F2 were always predetermined by the barriers.
A primary concern during subsequent analyses was to dissociate overt behavioral differences (such as running speed) from differences in neural activity. For example, stages 1 and 6 both occurred on the same sections of the maze and yet made contrasting demands upon spatial working memory. However, rats ran more than twice as fast during stage 6 (44 ± 1.0 cm/s) than they did during stage 1 (18 ± 5.8 cm/s), a behavioral difference that may confound interpretation of differences between neural activities during the two stages. Similarly, rats ran at 17 ± 2.1 cm/s during stage 3 and at 31 ± 1.8 cm/s during stage 4. Most analyses therefore focused on the central three-quarters section of the central arm, where mean running speeds were similar during both choice and forced-turn epochs (38 ± 1.9 and 40 ± 2.0 cm/s, respectively). Restricting analyses to the central arm also excluded sections of divergent running trajectories at the very ends of the arm (Figure S1).
As rats crossed the central arm during forced-turn epochs, their route was predetermined by the moveable barrier. In contrast, during choice-direction epochs, rats were required to choose between left or right reward arms, employing spatial working memory to guide their decision. Since overt behavior was similar during both epochs, differences between neural activity during runs across the central arm in the forced-turn and choice directions therefore reflect mnemonic and/or decision-making processes, rather than the simple behavioral demands of the task.
Behavioral Correlates of CA1 and mPFC Firing
One hundred and sixty-five mPFC and 149 CA1 neurons (ten neurons from ventral CA1) active on the maze were classified as putative pyramidal neurons or fast-spiking interneurons (4% of recorded mPFC neurons, 5% of CA1 neurons) on the basis of spike width, firing rate, and burst-firing characteristics. The classification was based on similar schemes derived from intracellular recordings [41,42] (Table 1), although only pyramidal neurons were used for subsequent analyses. The data of Siapas et al. [30] and Hyman et al. [31] set a precedent for examining the coordination of dorsal CA1 and mPFC activities. However, the most prominent hippocampal projections to mPFC arise from ventral CA1/subiculum [43]; initial experiments were therefore designed to compare the properties and interactions of dorsal CA1 neurons with ventral CA1/subiculum neurons. The low yield of well-isolated units from ventral regions precluded a systematic comparison between dorsal and ventral subregions in this study, although the basic properties of ventral CA1 pyramidal cells were comparable with those of dorsal neurons (see Table 1).
Table 1 Spike Waveform, Burst Firing, Firing Rate, and Spatial Firing Parameters for the mPFC and CA1 Populations
The behavioral correlates of mPFC neuronal activity tended to be more spatially distributed than for CA1 neurons (Figure 2), with no population bias towards one maze region (mean mPFC firing rate 4.7 ± 0.4 Hz on reward arms [including reward points] and 5.2 ± 0.5 Hz on the central arm). Neither was mPFC population activity biased towards one task epoch. For example, overall central-arm firing rates were comparable in forced-turn and choice directions (5.3 ± 0.5 Hz and 5.2 ± 0.5 Hz, respectively).
Figure 2 Recording Details and Typical Hippocampal and Prefrontal Firing Properties on the Maze
(A) mPFC tetrodes targeted deep layers of prelimbic and infralimbic cortices. Photograph shows a typical lesion site (triangle) marking the tip of a tetrode. The partial brain section is superimposed on a schematic of a coronal section taken 3.7 mm rostral of bregma, showing the boundary of the prelimbic cortex (denoted by PrL).
(B) Spike amplitude clusters for a typical mPFC tetrode. The cluster in red was for the neuron shown in D. The points in the six panels plot extracellular-action-potential amplitude on wire 1 of the tetrode versus wire 2, wire 1 versus wire 3, etc.
(C) and (D) Activities of a typical CA1 place cell and two mPFC pyramidal cells, respectively (see also Figure S2). The upper mPFC neuron in D was recorded simultaneously with the CA1 neuron in C. Spikes were binned into positional pixels, and mean pixel firing rate was color-coded to generate the firing-rate maps on the left. Graphs show corresponding inter-spike interval distributions (10 Hz marked by the blue line; note logarithmic time scale). Waveforms show averaged extracellular action potentials recorded on a single wire of each tetrode (horizontal and vertical scale bars, 1 ms and 400 μV, respectively).
(E) Overlap in the distributions of spatial information carried by spikes from CA1 (black) and mPFC (blue) populations.
Despite these similar population firing rates during different task epochs, the central-arm firing rates of individual mPFC neurons did tend to distinguish between runs in the two directions and between different routes in the choice direction (Figure 3). A “directional index” for each neuron was defined as the magnitude of the difference between mean firing rates during forced-turn and choice directions, divided by the overall mean firing rate on the central arm. “Preference index” was defined as the magnitude of the difference in mean firing rates during F1 → C1 and F2 → C2 trials divided by the mean choice-direction firing rate. Thus both indices ranged from zero (firing rate identical in both epochs/routes) to one (fired only in one epoch). The mean directional index of mPFC neurons was 0.27 ± 0.02, and the mean preference index was 0.34 ± 0.04. This biased activity was also evident in CA1 (mean directional index 0.60 ± 0.05, preference index 0.62 ± 0.16) and was reminiscent of the CA1 activity previously described on similar linear tracks [44,45]. Slight differences in running trajectory or head direction may contribute to these firing-rate effects. However, these behavioral parameters did not vary consistently with trial type (Figure S1) and are unlikely to explain the firing-rate biases in their entirety. mPFC firing tended to be more sustained than that of CA1, with the average mPFC neuron firing at more than 10% of its maximum rate across 52% ± 3% of the central-arm area, whereas average CA1 place-cell firing covered only 22% ± 2% of the central arm. Similar “delay firing” activity in rat mPFC has been reported by Baeg et al. [46]. However, these firing-rate data do not address the nature of interactions between CA1 and mPFC.
Figure 3 Directional Bias of mPFC and CA1 Firing Rates
(A) Firing rate of a single mPFC neuron split into four trial types (shown by arrows). This neuron tended to fire at higher rates during runs in the choice direction, with the central arm firing highest during F2 → C2 trials. The magnified boxes show the central three-quarters section of the central arm. White lines mark the boundary of the positional pixels traversed by the rat on F1 → C1 trials. These are superimposed on the firing-rate map for F2 → C2 trials, showing the overlap between positions visited on both trial types.
(B) Central-arm firing rates of both mPFC (blue) and CA1 (black) neurons tended to distinguish between runs in the forced-turn and choice directions (directional index > 0) and choice-direction runs in either F1 → C1 or F2 → C2 trials (preference index > 0; see Results). However, there was no overall tendency for CA1 or mPFC populations to fire at higher rates during any one epoch or trial type.
Behavioral Modulation of Cross-Correlations between CA1 and mPFC Spike Times
We first investigated coordination of hippocampal and prefrontal activities by quantifying the temporal alignment of neuronal firing using cross-correlation of spike times from CA1–mPFC unit pairs coactive during different behavioral epochs. Peak cross-correlation coefficients (bin size 100 ms, maximum lag ± 200 ms) were normalized by spike counts, and bias-corrected by subtracting values obtained when trials were shuffled with respect to one another. Spike trains from 50 CA1–mPFC pairs coactive during stages 1 and 6 (see Figure 1B) showed significantly higher cross-correlation coefficients during stage 1 than during stage 6 (0.029 ± 0.002 versus 0.017 ± 0.002; p < 0.01 by Wilcoxon rank sum test on animal means). These differences cannot be explained by changes in firing rate, as the mean firing rates of these neurons were not significantly different during the two epochs (6.3 ± 2.1 Hz and 5.9 ± 2.0 Hz for epochs 1 and 6, respectively, in CA1; 7.7 ± 1.1 Hz and 8.8 ± 1.7 Hz in mPFC). However, running speeds were lower during stage 1 than during stage 6 (see Behavior); could this explain the differences in correlated activity? To address this, we also compared the degree of correlation between 49 coactive CA1–mPFC unit pairs during stages 3 and 4, neither of which required spatial working memory. Again, running speeds were higher during stage 3 than during stage 4. However, mean cross-correlations were not significantly different under these conditions (0.019 ± 0.002 during both stages). Independent of spatial location (epochs 1 and 6 correspond to the same sections of the maze), correlations between CA1 and mPFC activities were therefore significantly enhanced during the epoch presumed to recruit spatial working memory. These data are summarized in Table 2.
Table 2 Summary of Mean Firing Rates and Mean Peak Cross-Correlation Coefficients between CA1–mPFC Neuron Pairs during the Different Task Epochs Shown in Figure 1B
In order to confirm that differences in correlated activity related to varied working-memory or decision-making processes rather than overt behavioral state, we compared CA1–mPFC cross-correlations on the central arm during forced-turn and choice-direction epochs (stages 5 and 2), when running behavior was at its most uniform (Figure 4). Furthermore, we subdivided choice-direction runs into correct and incorrect trials. For 72 unit pairs coactive on the central arm, the mean peak cross-correlation during forced-turn runs was 0.009 ± 0.002 (Figure 4C). Correlated activity was significantly higher during choice-direction runs on correct trials (0.024 ± 0.003; p < 0.01). Importantly, 49 CA1–mPFC neuron pairs coactive during error trials showed significantly reduced correlations relative to correct-choice trials (0.015 ± 0.002; p < 0.05).
Figure 4 Enhanced Cross-Correlations between Spike Trains of CA1–mPFC Neuron Pairs during Behavioral Epochs Requiring Working Memory and Decision-Making
(A) Mean running speeds, CA1 firing rates, and mPFC firing rates were comparable during forced-turn (grey, epoch 5), choice-correct (red, epoch 2), and choice-error runs (hatched red) across the central arm.
(B) Example cross-correlogram (bin size 10 ms, maximum time lag ± 1,000 ms) for a single CA1–mPFC neuron pair (referenced to the CA1 firing at time 0), showing that correlated activity was higher during choice runs (red) than forced-turn runs (grey). The width of the central peak at 50% of its maximum value is 120 ms. This compares with a mean peak width of 156 ± 41 ms for 29 neuron pairs with peak cross-correlation coefficients (bin size 10 ms) of at least 0.0005. For comparison across task epochs, peak cross-correlation coefficients were quantified at the ± 200-ms time range with a bin size of 100 ms (inset to the right).
(C) Mean correlations for all neuron pairs that fired at least 50 spikes each during the three run types. CA1–mPFC correlation coefficients were significantly higher during choice runs (72 pairs) than during forced-turn (72) or error runs (49 pairs; ** p < 0.01, * p < 0.05 Wilcoxon rank sum test for grouped animal means).
These data demonstrate that cross-structural synchronization of neuronal firing at the neuron-pair level is modulated during this working-memory task. Specifically, mPFC activity is more highly correlated with CA1 place-cell activity during epochs of the task associated with spatial working-memory or decision-making processes. This synchronization may reflect the transfer of hippocampal spatial information to a mPFC working-memory system. Consistent with this, mPFC firing during runs in the choice direction carried significantly more spatial information [47] than during forced-turn runs (0.30 ± 0.06 versus 0.24 ± 0.05 bits per spike, two-tailed p < 0.05, Student's t-test). Rats may make errors in this task for a number of reasons, including failures of attention, working memory, or decision-making. Nevertheless, the attenuated cross-correlation of CA1–mPFC activity on the central arm during error trials does indicate that coordinated hippocampal–prefrontal activity is selectively associated with accurate behavioral performance.
Phase-Locking of CA1 and mPFC Spike Times to Local and Remote Theta Rhythms
How might the level of cross-structural coordination evident in the cross-correlation analysis be orchestrated and modulated? Siapas et al. [30] recently described phase-locking of mPFC spike-timing to the CA1 theta rhythm. During periods of phase-locking, spikes tend to occur during consistent time windows imposed by ongoing theta rhythms. It follows that the relative timing of spikes from multiple phase-locked neurons may also become more consistent under these conditions. Indeed, Siapas et al. [30] showed that mPFC neurons significantly phase-locked to the CA1 theta rhythm showed greater covariance with CA1 spike-timing than non-phase-locked neurons. However, this previous study made no attempt to link phase-locking to function by examining its relationship to ongoing behavior. We therefore went on to examine phase-locking of mPFC neurons during different epochs of this spatial working-memory task.
As detailed by Siapas et al. [30], the degree of phase-locking can be quantified by the circular-concentration coefficient, κ, of each neuron's phase distribution. κ was estimated by the maximum-likelihood method [48] and is a measure of concentration around the mean preferred phase; κ is inversely related to the Rayleigh p-value, with κ = 0 for uniform distributions. Phase distributions of CA1 pyramidal-cell spikes with respect to the local CA1 theta rhythm were significantly nonuniform for 71% of the population (Rayleigh test of uniformity p < 0.01). The mean circular-concentration coefficient across the entire population (including neurons with statistically uniform phase distributions) was κ = 0.21 ± 0.01. Adding random jitter of up to 100 ms to the spike times of these cells reduced mean κ to 0.05 ± 0.02, close to the zero value expected for uniform distributions. As expected, a cross-structural phase relationship was also evident: mPFC firing was significantly phase-locked to CA1 theta rhythm in 49% of the population (mean κ = 0.08 ± 0.01 for the entire population averaged over all firing on the maze; see examples in Figure 5). Again, adding random jitter to mPFC spike times reduced circular-concentration coefficients to 0.04 ± 0.02.
Figure 5 Spike-Timing in CA1 and mPFC Populations Was Phase-Locked to CA1 Theta Rhythm
Firing-rate maps for representative mPFC (A) and CA1 (B) pyramidal cells. Graphs show inter-spike interval distributions (blue line marks 10 Hz). Thick black lines show CA1 LFP band pass filtered at 4–12 Hz during single central-arm crossings (scale bar 0.5 mV, length 1.3 s and 1.4 s in A and B, respectively) with spike times of the two neurons above marked by ticks. Raw LFP is shown by the thin black line in A. Rose histograms show phase distributions for these single mPFC (blue) and CA1 (black) neurons with respect to CA1 theta rhythm. Thick black lines mark mean preferred phase. The numbers on the outer circular axis give spike counts. Circular-concentration coefficients are given by κ. Both distributions are significantly nonuniform (p < 0.01, Rayleigh test).
Siapas et al. [30] suggested that CA1 activity tends to lead to mPFC activity, as mPFC firing locked more reliably to the preceding CA1 theta cycle than to the simultaneous oscillation. In agreement with this, we found that shifting the relative timing of CA1 LFP forward by an average of 30 ± 10 ms maximized the values of κ for mPFC phase distributions. In contrast, shifting the timing of CA1 theta relative to CA1 spike times (by times of up to ± 100 ms) did not significantly improve their phase relationship.
The rhythmic activity underlying hippocampal theta is particularly apparent in LFP recordings because of the laminar organization of current sinks and sources in CA1. The nature of neocortical LFP is more ambiguous, yet we were able to distinguish clear instances of theta-frequency oscillations in mPFC LFP during running on the maze. These were distinct from the 7- to 12-Hz high-voltage spindles observed during immobile states [49,50] (data not shown). We therefore also calculated phase distributions for the firing of each neuron in relation to mPFC theta peak times. Significantly nonuniform distributions (κ = 0.05 ± 0.01 and κ = 0.09 ± 0.01) were shown by 29% of CA1 and 43% of mPFC populations, respectively. Importantly, the circular-concentration coefficient of any given neuron versus CA1 theta was positively correlated with its circular-concentration coefficient versus mPFC theta (r = 0.65, p < 0.001), indicating that theta-frequency activities in the two structures were related.
Behavioral Modulation of Phase-Locking
If the consistent timing relationship between neural spiking and ongoing theta rhythms provides a mechanism through which to coordinate mPFC and CA1 activity, and if the coordination between these structures depends upon task demands, the degree of phase-locking might be expected to vary with task epoch. In particular, phase-locking should be enhanced during the task epochs associated with working memory or decision-making and enhanced CA1–mPFC spike cross-correlations.
In every animal tested, the spikes of mPFC units active on the central arm showed a greater concentration around their preferred phase of CA1 theta during choice-direction epochs on correct trials than during forced-turn epochs (Figure 6A). Thus, although approximately 40% of the active mPFC population showed significantly uniform phase distributions in both epochs (44% during choice and 39% during forced-turn), the circular-concentration coefficients of these distributions were significantly higher during correct, choice-direction trials (κ = 0.19 ± 0.02 versus κ = 0.10 ± 0.02, n = 39; p < 0.01 Wilcoxon rank sum test based on animal means). Furthermore, as for the spike train cross-correlations, phase-locking of mPFC neurons on the central arm was significantly attenuated during error trials (κ = 0.10 ± 0.03, n = 27; p < 0.05 versus correct trials). In contrast, the degree of phase-locking shown by CA1 neurons was similar during choice-correct, forced-turn, and choice-error epochs (Figure 6B; κ = 0.43 ± 0.13, n = 16; κ = 0.45 ± 0.11, n = 16; κ = 0.53 ± 0.16, n = 10, respectively).
Figure 6 Theta Phase-Locking of mPFC Spike Timing to the CA1 Theta Rhythm Was Enhanced during Choice Epochs Relative to Forced-Turn and Choice-Error Runs
Phase distributions for single mPFC (A) and CA1 (B) neurons during forced-turn (grey), choice-direction (red), and choice-error (hatched red) epochs. Bar graphs show mean-population circular-concentration coefficients (κ) during the three epochs, and the significant (** p < 0.01, * p < 0.05) increase in κ for the mPFC population during choice epochs (39 neurons) relative to forced-turn and choice-error epochs (39 and 27 neurons, respectively). In contrast, the CA1 population showed a similar degree of phase-locking during all epoch types (for 26, 26, and 15 neurons, respectively).
(C) The results in A cannot be explained by changes in mean LFP theta power, which was comparable during forced-turn, choice, and choice-error epochs in both CA1 and mPFC.
The increases in phase-locking of mPFC neurons were not accompanied by overt changes in running behavior, increased population firing rates (see Figure 4A), or increased theta power (see Figure 6C) during choice epochs. They must therefore reflect a more consistent timing relationship between mPFC spikes and the CA1 theta rhythm during the behavioral epochs associated with peak working-memory load and decision-making. Again, error trials were associated with impaired coordination between CA1 and mPFC activities. Thus enhanced phase-locking of mPFC activity to the CA1 theta rhythm paralleled the increases in correlated activity at the neuron-pair level, consistent with the suggestion that phase-locking to the theta rhythm constitutes a mechanism through which to temporally coordinate populations of neurons in these two structures.
Coherence between CA1 and mPFC LFP
The enhanced phase-locking of mPFC single-unit activity to the theta rhythm suggests a broader coordination of CA1 and mPFC population activities in the theta-frequency band. A direct measure of such covariation at the population level is the coherence between LFPs in the two structures. LFP measures of “averaged” population activity constitute a useful adjunct to the spike-timing analyses, since they are unlikely to be sensitive to trial-by-trial fluctuations in the firing rates of individual neurons. We quantified multi-taper estimates of coherence in the same 4- to 12-Hz theta-frequency range as unit-LFP phase-locking (Figure 7). Considering only LFP sections corresponding to central-arm crossings, theta-frequency coherence assessed on a trial-by-trial basis was significant during 55% ± 2% of crossings in the choice direction, and only 32% ± 6% of crossings in the forced-turn direction. Hence the absolute value of mean 4- to 12-Hz coherence on the central arm was significantly higher during choice epochs on correct trials than during forced-turn epochs (0.32 ± 0.03 versus 0.19 ± 0.04, respectively; p < 0.05). Like the cross-correlation and phase-locking measures, this measure of CA1–mPFC coordination was reduced during error trials (to 0.20 ± 0.06, p < 0.05, versus correct trials).
Figure 7 CA1–mPFC LFP Coherence Showed a Significant Peak in the Theta-Frequency Range and Was Enhanced during Choice Epochs
(A) Raw LFP from dorsal CA1 (black) and mPFC (blue) during consecutive single central-arm crossings in the choice (left) and forced-turn (right) directions. Thick lines show theta-filtered LFP. Horizontal scale bar 0.5 s; vertical scale bar 0.8 mV. Red lines highlight the timing relationship between CA1 and mPFC theta peaks (red circles). Numbers above raw LFP traces give coherence in the 4- to 12-Hz range during these two example trials.
(B) Trial-averaged, central-arm coherence during a single run-session (17 trials). Central-arm coherence is subdivided into forced-turn (grey) and choice (red) directions. Dashed line marks 95% confidence level, with shaded band thickness corresponding to jackknife error bars (estimated over trials and nine tapers). Significant coherence was seen only in the theta-frequency range, and only during choice epochs on the central arm.
(C) Mean coherence at delta (1–4 Hz) and theta (4–12 Hz) frequencies, pooled across animals during forced-turn (grey), choice-correct (red), and choice-error (hatched red) epochs (* p < 0.05). Like theta CA1–mPFC spike cross-correlations and theta phase-locking of mPFC units, theta-frequency CA1–mPFC LFP coherence peaked during choice-direction runs across the central arm.
Given the restricted sample lengths (2–2.5 s for central-arm crossings) and 4-Hz bandwidth, coherence estimates at frequencies below 4 Hz are not statistically robust. Nevertheless, mean 1–4 Hz (delta) coherence did not differ markedly between choice-correct, forced-turn, and choice-error epochs (0.16 ± 0.02, 0.15 ± 0.02, and 0.12 ± 0.03, respectively). Neither did we find any consistent coherence at frequencies above 12 Hz. LFP coherence therefore paralleled phase-locking of mPFC units to the CA1 LFP, both in terms of its theta-frequency range and its enhancement during behavioral epochs that required spatial working memory or decision-making.
Discussion
A critical role of working memory is the dynamic and selective incorporation of task-relevant information into decision-making processes [51]. Working memory therefore exemplifies conditions during which multiple disparate brain structures must interact transiently yet coherently. The hub of these interactions is presumed to lie in the prefrontal cortex, whose working-memory functions subserve its broader, integrative roles in establishing context and guiding behavior appropriately [52]. Within the cortex, there is mounting theoretical [53] and electrophysiological evidence from humans and primates [16–18] suggesting a role for rhythmic activity in working memory. Do cortical rhythms relate to rhythms elsewhere in the brain? Are they indicative of a broader functional network, allowing subcortical structures to participate in decision-making processes? How are oscillations recorded at the LFP level reflected by the activities of single neurons? Which frequency bands are key, and do different frequencies play different roles?
Our simultaneous recordings from prefrontal cortex and hippocampus during spatial working memory provide evidence for the rapid configuration of functional connectivity through the theta-frequency entrainment of oscillatory networks across these two brain regions. This entrainment was specific to a 4- to 12-Hz frequency range and was evident at every level examined, from individual pairs of coactive neurons, to the theta phase-locking of neurons to LFP, to hippocampal–prefrontal LFP coherence.
The behavioral correlates of mPFC firing during this task were varied; mPFC activity presumably reflects and drives many aspects of behavior. However, the firing of a significant proportion of the mPFC population did carry spatial information in the hippocampal range. In particular, increased spatial information content of mPFC firing coincided with epochs of increased phase-locking to hippocampal theta. Therefore, rather than mPFC neurons “becoming place cells” during this spatial task, the selective refinement of phase-locking between mPFC pyramidal-cell firing and ongoing theta rhythms acted alongside enhanced theta-frequency coherence to integrate hippocampal and prefrontal activities when required by the task.
Behaviorally modulated phase-locking potentially imparts great flexibility to the mPFC, allowing any given mPFC neuron to join different functional networks according to prevailing behavioral demands and in line with its current relationship to ongoing hippocampal activity. The periodicity of 4- to 12-Hz theta rhythms may make them particularly suitable reference signals. For example, groups of mPFC and CA1 neurons locked to the downward (>180°) phase of theta will have mutually increased firing probabilities during repeated ∼50-ms windows. The net effect will be increased correlated activity amongst these groups of neurons. This proved to be the case during choice epochs, when improved cross-correlations between CA1–mPFC unit pairs paralleled enhanced phase-locking and coherence. Similarly, the value of each neuron's mean preferred phase during a given behavioral epoch will consistently dictate the order in which that neuron fires relative to neurons with different mean preferred phases. Theta rhythms can therefore mediate the consistent timing relationships needed to establish functional connectivity between two structures, in this case serving to dynamically incorporate currently relevant spatial information into decision-making processes.
How is CA1–mPFC synchrony enhanced during specific behavioral epochs? One possibility is simply through simultaneously increased theta-modulated activity in these two structures. However, the enhanced correlations and phase-locking were independent of changes in theta power, since overall theta power was similar during forced-turn and choice epochs. In addition, the coherence measures are normalized by the power spectra of CA1 and mPFC LFPs and are consequently insensitive to power changes. Together, these data therefore imply selective alignment of CA1 and mPFC theta-rhythmic activity during choice-direction runs across the central arm of the maze. Since the instantaneous frequency of theta rhythms can vary rapidly over time, this alignment requires active synchronization of CA1 and mPFC rhythms in order to maintain their coordination with respect to each other. Similarly, maintenance of enhanced phase-locking of mPFC neurons to hippocampal theta cannot arise simply as a consequence of firing rates and LFP being modulated similar frequencies, but rather necessitates precise temporal control of spiking relative to theta rhythms [30].
This level of control may be exerted by monosynaptic projections from the hippocampus, which have direct influence on mPFC interneurons [54]. Since the synchronization of oscillatory networks is often attributed to the activity of local inhibitory interneuronal networks [55], these projections may provide the anatomical and physiological foundations for coherence between these two structures. Interestingly, the most dorsal part of CA1 does not contribute direct projections to mPFC [43], raising the possibility that dorsal CA1 interacts with mPFC via ventral CA1/subiculum. Whilst we did not observe qualitative differences between ventral and dorsal CA1 neurons in terms of their correlations with coactive mPFC neurons or their phase-locking properties, further experiments are required to establish the functional consequences of the known anatomical connections in this system.
There is no evidence for direct reciprocal projections from mPFC back to the hippocampal formation [56], making it tempting to speculate that the hippocampus drives mPFC firing rather than vice versa. Although our data do not unequivocally address the directionality of hippocampal–prefrontal interactions, this is supported by Siapas et al. [30], who suggest that mPFC neurons phase-lock to CA1 theta that occurs ∼50 ms in advance of their spikes; this also proved to be the case during the working-memory task employed here. However, whether behavioral-dependent enhancement of these coordinated activities is achieved entirely through hippocampal–prefrontal connectivity or via some third party that influences both CA1 and mPFC remains to be established.
Whilst enhanced theta-frequency coordination coincided with peak working-memory load, it is possible that theta-frequency interactions between CA1 and mPFC do not pertain solely to working memory or decision-making. For example, attention or reward expectancy may also vary between choice and forced-turn behavioral epochs. It should also be noted that some degree of phase-locking and coherence remained evident during runs across the central arm in the forced-turn direction. These residual interactions may reflect some working-memory-related aspect of spatial behavior common to both task epochs, such as updating route or task-rule information. CA1–mPFC synchrony—neuron-pair correlations, phase-locking, and LFP coherence—also fell to these control levels during runs towards the choice point on error trials. We cannot determine the stage of the task at which the errors originated, or whether they were due to failures in mnemonic, attentional, or decision-making components. However, the fact that the degree of CA1–mPFC synchrony can be used to predict behavioral outcome strongly suggests that these electrophysiological phenomena are indeed signatures of cross-structural interactions.
In summary, our data reveal correlations between behavioral demands and cross-structural neural synchrony: theta-frequency coordination between CA1 and mPFC peaks during behavioral epochs presumed to require effective communication between these two structures. It follows that disruption of such complex cross-structural communication is likely to generate behavioral impairments. For example, schizophrenia is associated with altered GABAergic function in hippocampal and prefrontal interneurons [57], and is widely presumed to involve disrupted functional connectivity of the prefrontal cortex [4,58,59]. Interestingly, schizophrenic patients do show spatial working-memory impairments [60]. The theta-rhythm-mediated coordination of hippocampal–prefrontal activity that we describe here may reflect the nature of cross-structural coordination at network and neuronal levels, and may contribute to both the clinical diagnosis of the impaired interactions likely to underlie cognitive disorders and to characterizing animal models of these diseases.
Materials and Methods
All procedures were performed in accordance with the Massachusetts Institute of Technology Committee on Animal Care and the National Institutes of Health guidelines. Six male Long-Evans rats (2–6 mo) were mildly food-deprived (to 85% of free-feeding body weight) and trained to run a continuous spatial-alternation task (see Figure 1A). Each trial comprised distinct sample and test epochs. The contingency was set such that, for example, a rat forced to turn to his right during the forced-turn epoch had to choose a left-hand turn to win reward during the subsequent choice epoch. Forced-turn direction was varied randomly, with no more than three consecutive trials in one direction. The relative location of the forced-turn end of the maze was varied between animals. Every effort was made to constrain running to overlapping linear trajectories by using a narrow track (6 cm). Furthermore, analysis epochs excluded reward points and turning points.
Each rat was trained to asymptotic performance (two consecutive days of at least 80% choice-correct) over a period of 12–14 d before surgery, then implanted with arrays of adjustable tetrode recording electrodes targeted to the mPFC (+3.2 mm, +0.6 mm from bregma) and ipsilateral dorsal CA1 (−3.6 mm, +2.2 mm). In two rats, tetrodes were also targeted to ventral CA1 (−6.3 mm, +6.2 mm). Differential recordings of extracellular action potentials (sampled at 31.25 kHz per channel, filtered between 600 Hz and 6 kHz) and continuous LFP (sampled at 3.125 kHz per channel, filtered between 1 and 475 Hz) were made using Keithley Instruments acquisition boards (DAS-1802HC [http://www.keithley.com]). Local reference electrodes were placed in overlying (for dorsal CA1 recordings) or adjacent (for ventral CA1 recordings) white matter, or in a proximal cortical region without spiking activity (2.4–2.7 mm below the pial surface for mPFC recordings). Positioning in white matter was achieved on the basis of characteristically flat LFP recordings (with no hippocampal sharp waves or ripples) and an absence of action-potential activity in this region. Only hippocampal LFP data taken from dorsal CA1 are presented here. Electrolytic lesions established tetrode tip positions at the end of each experiment (Figure 1A).
Data presented here are taken from eight recording sessions from the six rats. Action potentials were assigned to individual neurons by off-line, manual clustering using Xclust software (M. A. Wilson). Subsequent analyses employed a combination of in-house software (M. A. Wilson) and custom Matlab code (MathWorks, Natick, Massachusetts, United States). Firing with inter-spike intervals of between 2 and 15 ms was defined as bursting (minimum inter-burst interval 150 ms). The Complex Spike Index (Table 1) combined a measure of bursting with a measure of the likelihood that spikes later in bursts were smaller in amplitude than spikes earlier in bursts [61]. Spatial information was calculated according to Skaggs et al. [47]. Most analyses compared firing on the central three-quarters section of the central arm in the two running directions, and were restricted to neurons that fired at least 50 spikes in both choice and forced-turn epochs (in order to allow reliable circular statistics).
LFPs were down-sampled (to 600 Hz) and band pass filtered between 4 and 12 Hz, then maxima and minima detected and thresholds established to extract theta peak and trough times. Only peaks or troughs greater than one standard deviation from the mean amplitude of the filtered LFP were included (63% ± 6.6% of all maxima during choice epochs, 61% ± 9.5% during forced-turn epochs). Each spike was assigned a theta phase between 0 and 360° by linear interpolation of the spike time relative to the enveloping pair of peak (phase 180°) and trough (phase 0 or 360°) times. The Rayleigh test of uniformity was used to assess the resulting phase distributions for deviations from the circular uniform distribution. Circular statistics were calculated according to Fisher [48].
Multi-taper spectral analysis [62] was used to calculate power spectra and coherence for LFP data. This technique takes advantage of short-time-window Fourier analysis to reduce artifacts caused by non-stationary elements in the data (since data can be assumed to be stationary within the short sliding time-windows). The significance of trial-by-trial magnitude of the coherence during central-arm crossings was calculated according to Jarvis and Mitra (with coherence values greater than 2/√[(number of trials) × (number of tapers)] considered significant at p = 0.05) [63]. Data in the text and figures are given as mean ± standard error of the mean. Statistical comparisons between forced-turn, choice-correct, and choice-error conditions were performed on the two groups of animal means using Wilcoxon rank sum tests.
Supporting Information
Figure S1 Details of Movement Trajectories and Running Speed
(A) Raw data showing positional samples (taken from LEDs mounted on the rat's head, 30-Hz sampling rate) during a single run-session. Blue box marks the central three-quarters section of the central arm used throughout analyses comparing choice and forced-turn directions. Scale bar 6 cm.
(B) Mean central-arm trajectories averaged across all rats and trials for the different behavioral epochs (± 1 standard deviation marked by the width of the shaded area). Upper panel compares choice-correct runs (solid red line and dark red shading) with choice-error runs (dashed red line and lighter shading). Lower panel compares choice-correct with forced-turn runs (grey line and shading). Trajectories show considerable overlap on this section of the maze; systematic variations in trajectory are therefore unlikely to explain the enhanced coordination seen during choice-direction runs.
(C) Mean running speeds across the central arm (same color scheme as in [B]).
(2.8 MB PDF).
Click here for additional data file.
Figure S2 Raster Plots Illustrating Trial-By-Trial Firing on the Central Arm of the Maze for the Three Neurons Shown in Figure 2
A CA1 pyramidal cell is shown in (A), while (B) and (C) show mPFC pyramidal cells; the two neurons in A and B were recorded simultaneously. Spikes (shown by the tick marks on the rasters) were parsed into choice-direction (middle column) and forced-turn-direction (right column) runs. Corresponding linearized mean firing rates (Gaussian-smoothed, with a kernel width of 8 cm) are shown below each raster.
(2.3 MB PDF).
Click here for additional data file.
This work was supported by the Wellcome Trust (MWJ) and National Institute of Mental Health grant NIMH-R01 MH 61976 (MAW). Thanks also to students and staff of Neuroinformatics 2002 (Marine Biological Laboratory, Woodshole, Massachusetts, United States) and, in particular, to Bijan Pesaran for discussion and provision of Matlab code relating to coherence analysis, and to the Miller Laboratory (Massachusetts Institute of Technology) for helpful discussion of the manuscript.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. MWJ and MAW conceived and designed the experiments and wrote the paper. MWJ performed the experiments and analyzed the data.
¤ Current address: Department of Physiology, University of Bristol, Bristol, United Kingdom
Citation: Jones MW, Wilson MA (2005) Theta rhythms coordinate hippocampal–prefrontal interactions in a spatial memory task. PLoS Biol 3(12): e402.
Abbreviations
LFPlocal field potentials
mPFCmedial prefrontal cortex
Abbreviations
LFPlocal field potentials
mPFCmedial prefrontal cortex
==== Refs
References
Steriade M Gloor P Llinas RR Lopes de Silva FH Mesulam MM Report of IFCN Committee on Basic Mechanisms. Basic mechanisms of cerebral rhythmic activities Electroencephalogr Clin Neurophysiol 1990 76 481 508 1701118
Singer W Neuronal synchrony: A versatile code for the definition of relations? Neuron 1999 24 49 65 10677026
Varela F Lachaux JP Rodriguez E Martinerie J The brainweb: Phase synchronization and large-scale integration Nat Rev Neurosci 2001 2 229 239 11283746
Friston KJ Frith CD Schizophrenia: A disconnection syndrome? Clin Neurosci 1995 3 89 97 7583624
Spencer KM Nestor PG Niznikiewicz MA Salisbury DF Shenton ME Abnormal neural synchrony in schizophrenia J Neurosci 2003 23 7407 7411 12917376
Mann CA Lubar JF Zimmerman AW Miller CA Muenchen RA Quantitative analysis of EEG in boys with attention-deficit-hyperactivity disorder: Controlled study with clinical implications Pediatr Neurol 1992 8 30 36 1558573
Barry RJ Clarke AR Johnstone SJ A review of electrophysiology in attention-deficit/hyperactivity disorder: I. Qualitative and quantitative electroencephalography Clin Neurophysiol 2003 114 171 183 12559224
Willis WG Weiler MD Neural substrates of childhood attention-deficit/hyperactivity disorder: Electroencephalographic and magnetic resonance imaging evidence Dev Neuropsychol 2005 27 135 182 15737945
Salinas E Sejnowski TJ Correlated neuronal activity and the flow of neural information Nat Rev Neurosci 2001 2 539 550 11483997
Averbeck BB Lee D Coding and transmission of information by neural ensembles Trends Neurosci 2004 27 225 230 15046882
Brown EN Kass RE Mitra PP Multiple neural spike train data analysis: State-of-the-art and future challenges Nat Neurosci 2004 7 456 461 15114358
Singer W Gray CM Visual feature integration and the temporal correlation hypothesis Annu Rev Neurosci 1995 18 555 586 7605074
Steinmetz PN Roy A Fitzgerald PJ Hsiao SS Johnson KO Attention modulates synchronized neuronal firing in primate somatosensory cortex Nature 2000 404 187 190 10724171
Laubach M Wessberg J Nicolelis MA Cortical ensemble activity increasingly predicts behaviour outcomes during learning of a motor task Nature 2000 405 567 571 10850715
Vanderwolf CH Hippocampal electrical activity and voluntary movement in the rat Electroencephalogr Clin Neurophysiol 1969 26 407 418 4183562
Lee H Simpson GV Logothetis NK Rainer G Phase locking of single neuron activity to theta oscillations during working memory in monkey extrastriate visual cortex Neuron 2005 45 147 156 15629709
Kahana MJ Sekuler R Caplan JB Kirschen M Madsen JR Human theta oscillations exhibit task dependence during virtual maze navigation Nature 1999 399 781 784 10391243
Raghavachari S Kahana MJ Rizzuto DS Caplan JB Kirschen MP Gating of human theta oscillations by a working memory task J Neurosci 2001 21 3175 3183 11312302
Buzsaki G Theta oscillations in the hippocampus Neuron 2002 33 325 40 11832222
O'Keefe J Nadel L The hippocampus as a cognitive map 1978 London Oxford University Press 570
Mehta MR Lee AK Wilson MA Role of experience and oscillations in transforming a rate code into a temporal code Nature 2002 417 741 746 12066185
O'Keefe J Recce ML Phase relationship between hippocampal place units and the EEG theta rhythm Hippocampus 1993 3 317 330 8353611
Bland BH Oddie SD Theta band oscillation and synchrony in the hippocampal formation and associated structures: The case for its role in sensorimotor integration Behav Brain Res 2001 127 119 136 11718888
Hasselmo ME Bodelon C Wyble BP A proposed function for hippocampal theta rhythm: Separate phases of encoding and retrieval enhance reversal of prior learning Neural Comput 2002 14 793 817 11936962
Jensen O Lisman JE Hippocampal sequence-encoding driven by a cortical multi-item working memory buffer Trends Neurosci 2005 28 67 72 15667928
Colom LV Christie BR Bland BH Cingulate cell discharge patterns related to hippocampal EEG and their modulation by muscarinic and nicotinic agents Brain Res 1988 460 329 338 3224265
Pare D Gaudreau H Projection cells and interneurons of the lateral and basolateral amygdala: Distinct firing patterns and differential relation to theta and delta rhythms in conscious cats J Neurosci 1996 16 3334 3350 8627370
Frank LM Brown EN Wilson MA A comparison of the firing properties of putative excitatory and inhibitory neurons from CA1 and the entorhinal cortex J Neurophysiol 2001 86 2029 2040 11600659
Berke JD Okatan M Skurski J Eichenbaum HB Oscillatory entrainment of striatal neurons in freely moving rats Neuron 2004 43 883 896 15363398
Siapas AG Lubenov EV Wilson MA Prefrontal phase locking to hippocampal theta oscillations Neuron 2005 46 141 151 15820700
Hyman JM Zilli EA Paley AM Hasselmo ME Medial prefrontal cortex cells show dynamic modulation with the hippocampal theta rhythm dependent on behavior Hippocampus 2005 15 739 749 16015622
Jones MW Wilson MA Phase precession of medial prefrontal cortical activity relative to the hippocampal theta rhythm Hippocampus 2005 15 867 873 16149084
Swanson LW A direct projection from Ammon's horn to prefrontal cortex in the rat Brain Res 1981 217 150 154 7260612
Jay TM Glowinski J Thierry AM Selectivity of the hippocampal projection to the prelimbic area of the prefrontal cortex in the rat Brain Res 1989 505 337 340 2598054
Jay TM Burette F Laroche S NMDA receptor-dependent long-term potentiation in the hippocampal afferent fibre system to the prefrontal cortex in the rat Eur J Neurosci 1995 7 247 250 7757261
Siapas AG Wilson MA Coordinated interactions between hippocampal ripples and cortical spindles during slow-wave sleep Neuron 1998 21 1123 1128 9856467
Brito GN Thomas GJ Davis BJ Gingold SI Prelimbic cortex, mediodorsal thalamus, septum, and delayed alternation in rats Exp Brain Res 1982 46 52 58 7067791
Kesner RP Hunt ME Williams JM Long JM Prefrontal cortex and working memory for spatial response, spatial location, and visual object information in the rat Cereb Cortex 1996 6 311 318 8670659
Floresco SB Seamans JK Phillips AG Selective roles for hippocampal, prefrontal cortical, and ventral striatal circuits in radial-arm maze tasks with or without a delay J Neurosci 1997 17 1880 1890 9030646
Jones MW A comparative review of rodent prefrontal cortex and working memory Curr Mol Med 2002 2 639 647 12420803
Yang CR Seamans JK Gorelova N Electrophysiological and morphological properties of layers V-VI principal pyramidal cells in rat prefrontal cortex in vitro J Neurosci 1996 16 1904 1921 8774458
Degenetais E Thierry AM Glowinski J Gioanni Y Electrophysiological properties of pyramidal neurons in the rat prefrontal cortex: An in vivo intracellular recording study Cereb Cortex 2002 12 1 16 11734528
Jay TM Witter MP Distribution of hippocampal CA1 and subicular efferents in the prefrontal cortex of the rat studied by means of anterograde transport of Phaseolus vulgaris-leucoagglutinin J Comp Neurol 1991 313 574 586 1783682
Frank LM Brown EN Wilson M Trajectory encoding in the hippocampus and entorhinal cortex Neuron 2000 27 169 178 10939340
Wood ER Dudchenko PA Robitsek RJ Eichenbaum H Hippocampal neurons encode information about different types of memory episodes occurring in the same location Neuron 2000 27 623 633 11055443
Baeg EH Kim YB Huh K Mook-Jung I Kim HT Dynamics of population code for working memory in the prefrontal cortex Neuron 2003 40 177 188 14527442
Skaggs WE McNaughton BL Gothard KM Markus EJ Hanson SJ Cowan JD Giles CL An information-theoretic approach to deciphering the hippocampal code Advances in neural processing systems 1993 Morgan Kauffman 1030 1037
Fisher NI Statistical analysis of circular data 1993 Cambridge Cambridge University Press
Buzsaki G Bickford RG Armstrong DM Ponomareff G Chen KS Electric activity in the neocortex of freely moving young and aged rats Neuroscience 1988 26 735 744 3200427
Sakata S Yamamori T Sakurai Y 7–12 Hz cortical oscillations: Behavioral context and dynamics of prefrontal neuronal ensembles Neuroscience 2005 134 1099 1111 16019153
Baddeley A Working memory Science 1992 255 556 559 1736359
Miller EK Cohen JD An integrative theory of prefrontal cortex function Annu Rev Neurosci 2001 24 167 202 11283309
Lisman JE Idiart MA Storage of 7 +/- 2 short-term memories in oscillatory subcycles Science 1995 267 1512 1515 7878473
Tierney PL Degenetais E Thierry AM Glowinski J Gioanni Y Influence of the hippocampus on interneurons of the rat prefrontal cortex Eur J Neurosci 2004 20 514 524 15233760
Klausberger T Marton LF Baude A Roberts JD Magill PJ Spike timing of dendrite-targeting bistratified cells during hippocampal network oscillations in vivo Nat Neurosci 2004 7 41 47 14634650
Sesack SR Deutch AY Roth RH Bunney BS Topographical organization of the efferent projections of the medial prefrontal cortex in the rat: An anterograde tract-tracing study with Phaseolus vulgaris leucoagglutinin J Comp Neurol 1989 290 213 242 2592611
Coyle JT The GABA-glutamate connection in schizophrenia: Which is the proximate cause? Biochem Pharmacol 2004 68 1507 1514 15451393
Weinberger DR A connectionist approach to the prefrontal cortex J Neuropsychiatry Clin Neurosci 1993 5 241 253 8369632
Winterer G Coppola R Goldberg TE Egan MF Jones DW Prefrontal broadband noise, working memory, and genetic risk for schizophrenia Am J Psychiatry 2004 161 490 500 14992975
Park S Holzman PS Schizophrenics show spatial working memory deficits Arch Gen Psychiatry 1992 49 975 982 1449384
McHugh TJ Blum KI Tsien JZ Tonegawa S Wilson MA Impaired hippocampal representation of space in CA1-specific NMDAR1 knockout mice Cell 1996 87 1339 1349 8980239
Thompson DJ Spectrum estimation and harmonic analysis Proc IEEE 1982 70 1055 1096
Jarvis MR Mitra PP Sampling properties of the spectrum and coherency of sequences of action potentials Neural Comput 2001 13 717 749 11255566
|
16279838
|
PMC1283536
|
CC BY
|
2021-01-05 08:21:45
|
no
|
PLoS Biol. 2005 Dec 15; 3(12):e402
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030402
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627983910.1371/journal.pbio.0030405Research ArticleBioinformatics/Computational BiologyBiotechnologyMolecular Biology/Structural BiologyBiochemistryYeastDrosophilaCaenorhabditisHomo (Human)Systematic Discovery of New Recognition Peptides Mediating Protein Interaction Networks Systematic Discovery of New Recognition PeptidesNeduva Victor
1
Linding Rune
1
Su-Angrand Isabelle
1
Stark Alexander
1
de Masi Federico
1
Gibson Toby J
1
Lewis Joe
1
Serrano Luis
1
Russell Robert B [email protected]
1
2
1European Molecular Biology Laboratory, Heidelberg, Germany2European Molecular Biology Laboratory–European Bioinformatics Institute, Hinxton, United KingdomMatthews Rowena Academic EditorUniversity of MichiganUnited States of America12 2005 15 11 2005 15 11 2005 3 12 e4059 8 2005 27 9 2005 Copyright: © 2005 Neduva 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.
A Systematic Way to Find Linear Motifs Mediating Protein-Protein Interactions
Many aspects of cell signalling, trafficking, and targeting are governed by interactions between globular protein domains and short peptide segments. These domains often bind multiple peptides that share a common sequence pattern, or “linear motif” (e.g., SH3 binding to PxxP). Many domains are known, though comparatively few linear motifs have been discovered. Their short length (three to eight residues), and the fact that they often reside in disordered regions in proteins makes them difficult to detect through sequence comparison or experiment. Nevertheless, each new motif provides critical molecular details of how interaction networks are constructed, and can explain how one protein is able to bind to very different partners. Here we show that binding motifs can be detected using data from genome-scale interaction studies, and thus avoid the normally slow discovery process. Our approach based on motif over-representation in non-homologous sequences, rediscovers known motifs and predicts dozens of others. Direct binding experiments reveal that two predicted motifs are indeed protein-binding modules: a DxxDxxxD protein phosphatase 1 binding motif with a K
D of 22 μM and a VxxxRxYS motif that binds Translin with a K
D of 43 μM. We estimate that there are dozens or even hundreds of linear motifs yet to be discovered that will give molecular insight into protein networks and greatly illuminate cellular processes.
Many protein interactions are mediated by short amino acid motifs. The authors describe a new approach to identify these interaction motifs and experimentally validate some of their binding predictions.
==== Body
Introduction
Protein interactions are central to all cellular processes. At the molecular level they can occur in a variety of ways. Probably the best known involve specific contacts between globular domains (∼100–200 residues) present in the interacting proteins. These are seen in many different contexts ranging from different subunits in large molecular machines (e.g., RNA polymerase II [1]), to more transient interactions (e.g., cyclins binding to CDK2 [2]).
However not all interactions are mediated by pairs of globular domains. Many involve the binding of a domain in one protein to short regions (approximately three to eight residues) in another [3,4]. These regions often show a particular sequence pattern, or “linear motif,” which captures the key residues involved in function or binding [5]. Linear motifs are critical to many processes including signal transduction (e.g., SH3 domains bind PxxP [6]), gene expression (e.g., Groucho→WRPW [7]) and DNA replication (e.g., PCNA→QxxxxxFF [8]).
In contrast to domains, which are readily detectable by sequence comparison, linear motifs are difficult to discover due to their short length, a tendency to reside in disordered regions in proteins, and limited conservation outside of closely related species. To date they have typically been found by time-consuming experiments, meaning that only a few hundred motifs are known compared to thousands of domains that might bind them. Although it is at present difficult to estimate just how many such interaction motifs exist, it is likely that many interactions are mediated by those not yet discovered. Here we perform the first systematic attempt to discover new motif candidates and their corresponding binding partners using results of genome-scale interaction datasets.
Results
Methodology
Our central hypothesis is that proteins with a common interaction partner will share a feature that mediates binding, either a domain or a linear motif. In the absence of a shared domain, a linear motif could well be the only common sequence feature and might thus be detectable simply by virtue of over-representation, which is the basis of our approach (Figure1).
Figure 1 Schematic of the Linear Motif Discovery Strategy
Interaction maps are probed for interaction sets (A): Partners of proteins with multiple interactions are clustered together when there are no known sequence features present (B). Domains and homologous regions are then identified (B) and removed prior to running exhaustive pattern discovery (C) to produce a list of motifs ranked by their probabilities P (D). Hypothetical motifs are shown as coloured squares in (C) and (D). “Proteins” in (D) gives the set of proteins containing at least one copy of the motif.
Given a set of proteins sharing an interaction partner we first remove sequence regions unlikely to contain linear motifs: globular domains, trans-membrane segments, coiled-coils, collagen regions, and signal peptides. This is justified because only 15% of known linear motifs [5] occur within these regions, and including them can give rise to misleading motif signals, particularly if common domains are found in more than one protein in a set. Most importantly, this avoids the detection of repetitive, purely structural patterns, such as β-turns, coiled-coil heptads, or collagen repeats, because these are unlikely to occur in the unstructured parts of proteins that remain after this filtering. We also compare all sequences in a set to each other and leave only one representative of any homologous segments. We do this in order to measure over-representation that is not the result of homology; our assumption is that each of the remaining instances of a particular motif has arisen convergently and is thus an independent observation. We specifically avoid removing regions of low complexity because linear motifs frequently occur within them.
We then find all three to eight residue motifs in the remaining sequence [9], and score their over-representation as the binomial probability (P) of seeing them randomly in a similar set of sequences (see Materials and Methods). This allows multiple observations of an otherwise insignificant motif to become statistically significant by over-representation, and readily accounts for sets of different sizes and composition. For example, the SH3-binding pattern RxPxxP readily occurs in about one out of 20 randomly selected proteins, but its occurrence in seven sequences in a set of nine becomes highly significant. We also compute P for all closely related species based on whether or not the same motifs are seen in the corresponding orthologues, and multiply these to give a final score (Scons; see Materials and Methods).
We applied our approach to interacting sets of proteins from Saccharomyces cerevisiae, Drosophila melanogaster, Caenorhabditis elegans, and Homo sapiens [10–14]. For the first three species, these datasets are from yeast two-hybrid screens; human data comes from the human Proteome Resource Database (HPRD) [14] and consists of hand-curated interactions extracted from the literature (see Protocol S1). For each dataset, we constructed a control by selecting random sets of proteins of a similar length and number, and performed the same calculations. We then defined a confidence threshold (p-value < 0.001) for Scons for each dataset (see Materials and Methods). Note that this threshold does not necessarily reflect the accuracy in terms of identifying binding motifs, only that the particular sequence pattern reported is very unlikely to arise by chance. It is possible that patterns can arise for other reasons, including localization signals or other sequence features common to protein performing similar function.
Known motifs come in different flavours, for instance canonical SH3-binding motifs (PxxP) are embellished with different amino acids, which determine the specific SH3-containing protein they bind (e.g., RxPxxP and PxxPxK). The sets of proteins above (i.e., those sharing an interaction partner) are appropriate for finding such motif flavours because each protein containing a particular instance of a domain (e.g., SH3) is considered separately. However, it is also beneficial to detect more general motifs specific to a domain family. To do this we simply merge sets if the common binding partners shared a particular domain. We refer to these as “domain” sets in the sections that follow (see Protocol S1 and Figure S1).
Benchmark
The Eukaryotic Linear Motif resource (ELM) [5] contains a curated set of experimentally validated instances of binding motifs (i.e., their location in a particular protein). This provides several pertinent sets of proteins to test the approach, namely each set of proteins containing a known instance of a particular motif (e.g., all PxxP motif–containing sequences known to interact with SH3 domains). Of 58 different sets, 22 contained at least four non-homologous instances of the motif, and could be used to test our approach. We ran the procedure on each set and monitored where the known motif (or a variant) was found in the list of all motifs ranked according to Scons. Despite many thousands of possibilities, the approach detected the correct motif as the very best ranked for 14 out of 22 and among the top ten for an additional three (Table 1). Applying the confidence threshold left eleven correct motifs at first rank, and no false predictions (see legend to Table 1). Inspection showed that those motifs that were either missed or scored poorly were generally highly degenerate in nature (e.g., the sumoylation site (VILAFP)Kx(EDNGP)).
Table 1 Detection of Known Linear Motifs in Experimentally Verified Sets from the Eukaryotic Linear Motif
Motifs in Genome-Scale Interaction Sets
Considering the genome-scale interactions, each dataset produced a number of protein sets sharing a common interaction partner: yeast, 191; fly, 632; nematode, 367; and human, 1,986. Only a small fraction of these produced one or more confident motifs (as assessed by the binomial probability): yeast, 11; fly, 26; nematode, 27; and human, 112. In all cases, known motifs were among those produced, though to varying degrees: yeast, 1 (domain set); fly, 9; nematode, 4 (domain set); and human, 48 (all significant motifs from the protein set are given in Protocol S1 and Table S1; all motifs, including those with poorer significance, are available at http://lmd.embl.de). Figure 2 shows a summary of the 26 motifs found in the fly set, highlighting the nine rediscoveries of known motifs (including one likely nuclear localization signal). The better results in human data (i.e., 48/112) are undoubtedly because the hand-curated interactions (HPRD, [14]) contain fewer errors than those from the comparatively noisy high-throughput yeast two-hybrid screens for the other organisms. Here we found motifs spanning virtually the full range of those known (SH3→PxxP, 14–3–3 proteins→RxxSxP, Clathrin→LDxL, etc.), in addition to several that appear to be novel.
Figure 2 Overview of Motifs Found in the Fly
Significant predictions from the yeast two-hybrid set for the fly. Blue dots in the center of each cluster represent proteins with four or more interaction partners (red and white dots) containing at least one confidently predicted motif (p-value < 0.001; Scons ≤ 8 × 10−15). Partner proteins containing the motif are represented by red dots, whereas proteins lacking the motif are indicated by white dots. Clusters are labelled as gene name→detected motif. Yellow circles enclose known motifs: SH3→PxxP [38], PP1→RVxF [22], C-terminal binding protein (CtBP)→PxDLS [52], SR splicing factors RS-rich segments [53], and CG6843→SxKSKxxK, a likely nuclear localization signal. The Translin→VxxxRxYS motif was experimentally tested (Figure 3). The grey circles enclose clusters with low-complexity patterns. Two additional known motifs were also found in the fly using more relaxed criteria than those used for the other motifs in the figure: Groucho→WRPW [7] and Dynein light chain→TQT [26] as the variant A(TI)QT(DE). The latter was also identified as significant in the domain sets. Proteins are denoted either by their FlyBase accession codes or protein names when available.
Inspection showed that known motifs were typically missed because the sets contained too few sequences with the correct motif to reach significance. For example, in yeast interaction data, just four out of 23 proteins interacting with the protein phosphatase 1 (PP1) domain contained the established (RK)VxF motif. A similar situation occurred with WW domains, where no more than three instances of known motifs were found among their interaction partners. It could be that certain motifs are just too rare in the interacting set to be detected. However, it is also well established that the yeast two-hybrid system, particularly when applied in genome-screens, can miss known interactions [15] and, moreover, make false predictions that cloud the signal from true motifs. The prediction accuracy and coverage will certainly increase when more comprehensive and reliable interaction data become available. The error prone nature of the underlying yeast two-hybrid data for yeast, fly, and nematode might be expected to yield inconsistencies (i.e., different motifs for the same protein) when comparing predictions from different species. Encouragingly, however, we found very few of these, and indeed in one case (see PP1, in Experimental Testing of New Motifs), we think the apparent inconsistency corresponds to two distinct motifs that bind to the same protein, each detected in a different species.
Many of the motifs detected in the protein sets were also found when interaction partners were pooled owing to the presence of a common domain (domain sets). Frequently a more general motif was found in the domain than in the protein sets. For example, in the fly we identified the canonical TQT motif as the binding site for Dynein light chain domains, but found the more specific pattern A(TI)QT(DE) for the specific Dynein homologue Cdlc2. Other motifs were seen only in one of the protein or domain sets. For example, in yeast a correct SH3 motif was only found in the domain sets, because no single SH3-domain protein had a sufficient number of interaction partners for the motifs to be found with significance. The reverse was true in the fly in which the correct SH3 motif was found only in the protein set, because the domain set had too many proteins lacking the canonical motif, meaning that the signal was lost.
There is also a problem of ambiguity in both protein and domain sets. For multi-domain proteins predicted to bind a motif, it is not possible to discern which domain is mediating the interaction. This can be partly resolved by considering domain sets, but even here there are still some examples where genuine motifs were predicted for the wrong domains. Inspection showed that this was either because the process selected the wrong domain of a frequently co-occurring pair (e.g., SH2 domains predicting to bind SH3 ligands) or selected an activator/inhibitor of the correct binding domain (e.g., protein phosphatase inhibitor 2 (IPP2) binding to PP1-like RVxF-like motifs [16]). The latter highlights the possibility of the yeast two-hybrid system identifying indirect interactions [17]. These domain ambiguities can likely only be resolved by experiment.
Experimental Testing of New Motifs
For a selection of new protein→motif associations, we tested direct binding via fluorescence anisotropy, using labelled peptides corresponding to the regions containing the predicted motifs (Protocol S1). Because the motifs we have predicted might be the lowest common denominator of what could be a slightly longer binding region, we included two additional residues to the N- and C-terminus of each peptide (extracted from the original sequence). We first selected candidate motifs in yeast, fly, and nematode based on the feasibility of expressing and purifying the common interaction partner (i.e., the protein predicted to bind the motif). Of the 55 significant novel motif predictions, only 13 contained a single globular domain, were not excessively long (≤ 650 amino acids), and lacked long regions of predicted disorder or low complexity. From these we selected five that had available clones and established purification protocols. These spanned a range of novelty, ranging from variations on a known motif, to those for which there was some supporting, but not direct, evidence in the literature, to those lacking any additional support. Of the five selected, we could obtain clones for four and could purify three.
We tested a highly significant Translin→VxxxRxYS motif found in fly data (Figure 3A). Translin is a protein thought to be involved in chromosomal rearrangements, and binds double-stranded RNA and DNA [18,19]. The fluorescence polarisation assay shows that it binds the peptide motif specifically compared to a mutated counterpart, or randomly selected peptides (Figure 3B). The affinity of binding (K
D = 43 ± 15 μM) is within the range typical for known linear motifs when considered in isolation (5–150 μM; see Table 1). Mutated controls or arbitrarily chosen peptides do not show specific binding (Figure 3B; note that the apparent linear increase in both is due to the high protein concentrations reached). We can only speculate what role this motif plays in modulating Translin function. However, there are several precedents for interaction motifs playing critical regulatory roles by binding to other DNA- or RNA-binding proteins, such as PCNA [20] or CtBP [21].
Figure 3 A Novel Fly VxxxRxYS Motif That Binds Translin
(A) Translin (left) shown surrounded by interaction partners containing the predicted motif VxxxRxYS. Proteins are shown as lines with domains (labelled shapes), predicted coiled coils (light blue/green segments), and the location of motifs (blue vertical bars). Sequences for the motif-containing region are shown aligned to the best homologues in closely related species. Amino acids are coloured according to residue type: blue, positive; red, negative; light blue, small; yellow, hydrophobic; green, aromatic; magenta, polar; and orange, proline. Those constituting the predicted motif are denoted by circles. Aga, Anopheles gambiae; Dme, D. melanogaster; Dps, D. pseudoobscura.
(B) Saturation curves, showing bound fraction (fluorescently labelled peptides at saturation) as a function of Translin concentration. Polarization values (mP) at zero concentration and Bmax were normalised to give the bound fraction. K
D was computed by non-linear regression on values from three independent experiments. The lower panel shows the alignment of the native and mutated peptides together with the arbitrary peptide (selected randomly). Black triangles show positions specifying the motif (VxxxRxYS). The alignment is coloured as described in (A).
We also tested a DxxDxxxD motif found in 10 of 12 interaction partners of yeast protein phosphatase 1 (PP1, Figure 4A). Eight are well-known PP1 interactors, and five contain the canonical RVxF PP1-binding motif. Fluorescence polarisation shows that a peptide corresponding to the region in Scd5 binds specifically to PP1 (K
D = 22 ± 5 μM), compared to arbitrary peptides (Figure 4B). Inspection of other PP1-binding proteins [22] reveals that 12 of 33 also contain the new motif, with an additional 15 containing a more relaxed pattern (permitting Glu). Deletions of the canonical RVxF motif do not always disrupt PP1-binding, and have led others to suggest additional binding sites [23]. Interestingly, deletions of some segments containing this new motif can affect PP1 binding in other proteins [22]. Other support comes from pull-down studies, which identified a similar region (RVRLDDDDE) critical for the Cdk5–PP1 interaction [24] and the recent crystal structure of human PP1 bound to a myosin-targeting subunit MYPT1, which led the authors to propose a positively charged surface on which a similar acidic stretch could interact [25]. Interestingly, the mutated control appears to retain some affinity, probably owing to the presence of additional negative charges that have not been mutated to alanine, and indeed a near match to the motif (DxxxExxD) is still present in the mutant. Arbitrary peptides did not show any specific binding (Figure 4B).
Figure 4 An Acidic Yeast PP1 Binding Motif
(A) PP1 (Glc7) with the set of interaction partners containing the DxxDxxxD motifs. Details are as for Figure 3A. Here the location of RVxF motifs (defined as matches to (RK)x0–1(VI)x(FW)) are shown as yellow bars, and low-complexity regions are magenta. The figure also shows the structure of PP1 bound to RVxF (red spheres) [54] with a hypothetical helix containing the motif. Blue spheres show the location of Arg or Lys residues, and the active site is circled with critical Arginines shown in ball-and-stick. Red arrows show hypothetical interactions of the motif either with sites on PP1 or elsewhere. Ani, Aspergillus nidulans; Cal, Candida albicans; Ego, Eremothecium gossypii; Gze, Gibberella zeae; Mgr, Magnaporthe grisea; Sce, S. cerevisiae; Spo, Schizosaccharomyces pombe; Str, Salinospora tropicalis; Uma, Ustilago maydis; Xla, Xenopus laevis.
(B) Saturation curves, showing bound fraction as a function of PP1 concentration. The polarization values (mP) were normalized to an extrapolated Bmax because Bmax could not be reached experimentally. Other details are as given in Figure 3B. Red triangles in the lower panel show the location of the near match to the motif in the mutated sequence.
Lastly, we tested a variant of the well-known Dynein light chain binding motif. The canonical motif has a consensus sequence (KR)xTQT and mediates interactions important for cell trafficking [26]. We found the canonical motif in the fly, but noticed a variant, IQTE, among three partners of Cdlc2 (Dynein light chain 2), which is similar to one present in the protein swallow from D. pseudoobscura [27]. We could detect no binding of the Cdlc2 to fluorescently labelled peptides over a range of protein concentrations (5–400 μM). Surprisingly, a true instance of the motif, known to bind Cdlc2 in vivo and in vitro [27], also did not give a signal using this procedure, suggesting that the experimental assay might not be suitable for Dynein light chain interactions (see Protocol S1).
Other Promising Predicted Motifs
For other predictions, we scrutinized the literature for previous experiments hinting that the motifs could be genuine. For example, among several interesting predictions in the fly was an Elongin C→LxxLCxR motif, which has been described previously only as part of a longer sequence called the SOCS box [28]. Only three of four interacting proteins with the motif contain the full SOCS box. The protein lacking it (CG18171) is not well understood; the interaction has not been reported apart from the genome screen. Deletion and mutagenesis experiments have shown that this region is important for the interaction with Elongin C [29]. Our finding agrees with this and further suggests that the motif could be sufficient on its own for mediating the interaction.
We found the motif SxPxxxS in 11 of 17 interaction partners of the nematode MAP-kinase lit-1 involved in wnt signalling and morphogenesis (Figure 5). These include three well-known regulators/interactors of lit-1: two nuclear proteins (wrm-1 and mom-4) and another morphogenesis protein (pop-1). Deletions have already demonstrated that regions containing the motif are critical for lit-1 binding (yellow boxes in Figure 5): a 148 N-terminal segment in wrm-1 [30], a 21-residue stretch in mom-4 [31], and a 45-residue region just six residues N-terminal to the motif in pop-1 [32] all disrupt lit-1 binding.
Figure 5 A Lit-1 MAP Kinase SxPxxxS Motif
The MAP kinase lit-1 surrounded by its interaction partners containing the SxPxxxS motif. Details are as for Figure 3. Yellow boxes show the location of deletion mutants known to affect the interaction. Cbr, C. briggsae; Cel, C. elegans.
Among several compelling new motifs in human was a T(PL)QP motif predicted to bind to the transcription factor PC4 (Positive Cofactor 4). PC4 binds double-stranded DNA and promotes the assembly of the preinitiation complex via a mechanism that is not fully understood [33,34]. The five proteins containing the putative motif all participate in transcription, but share no common globular domain that could mediate binding to PC4. Such a proline-rich motif could be a good candidate to bind one of the several aromatic patches on the surface of the PC4 protein [35].
Low-Complexity Linear Motifs
In both fly and nematode, several very significant motifs arose from regions of low sequence complexity (i.e., dominated by a few amino acids). These included examples already known to mediate interactions, and others not described previously, including His-, Ser-, Lys-, and Glu/His-rich motifs. We could find no motifs like these in random sets, which suggested that they are not the result of the general prevalence of low-complexity regions within proteins, but just what they mean is an open question. They might well be true, biologically meaningful interactions, and indeed for some sets the proteins show similarities in function. This idea is supported by the fact that many known motifs, including the protein/RNA binding RS/SR motifs [36], the Tudor domain→(RG)n [37], and SH3→poly-proline ligands [38], are themselves low complexity. Alternatively, they could be the result of some artefact of the yeast two-hybrid system. The last possibility is supported by the fact that we found fewer such motifs in the less error-prone human data.
How Many Protein–Motif Interaction Pairs Are Still to Be Found?
Both our experiments and those done previously suggest that many of our findings are genuine motifs that have not yet been reported. This raises the question as to how many new interaction motifs there are yet to be discovered. An estimate can come by considering what fraction of the previously known motifs we found and extrapolating this to the new discoveries. For example, in fly we predict 26 motifs of which nine are known, from a total number of roughly 60 that are known in this organism [5]. If we assume that all the remaining motifs are correct, and assume an equal distribution of motifs in fly proteins not seen in the yeast two-hybrid data (4,683/13,833), we estimate 334 additional motifs (the equivalent number for human is 405). Even the more modest assumption of between 10%–20% of the predicted motifs being correct (roughly the fraction for which we could see direct binding experimentally, which is clearly a lower estimate) gives estimates of 33–67 new motifs in fly (40–80 in human). There are very likely dozens to hundreds of new motifs to be discovered, which will correspond potentially to thousands of individual binding sites. To date we have just scratched the surface of what is likely a sophisticated network of peptide-mediated interactions in the cell.
Discussion
Many studies continue to highlight the importance of networks mediated by linear motifs [39,40], and each new discovery opens new lines of research into critical aspects of cell function [41]. We have shown here that these very simple features can be detected successfully, even in error-prone data, provided they occur with a sufficient frequency in otherwise unrelated proteins. The approach need not be restricted to protein-protein interactions. It can also be applied in other contexts: Any set of proteins or nucleic acids can be probed for short sequences responsible for a common biological feature (cellular location, modifications, etc.).
Both globular domains and linear motifs are modular in the sense that they are reused in different functional contexts, but they probably differ in how they arise. Domain shuffling involves duplication of part of a gene and its insertion into another. In contrast, the short length of linear motifs makes them likely to arise convergently in proteins by evolutionary drift [42]. This suggests that there are probably many near matches to the motifs just waiting for an appropriate point mutation to induce a function. They are, in effect, powerful switches for nature to explore during the evolution of complex functions. In this regard they are highly similar to transcription factor–binding sites [43,44] or microRNA target sequences [45]. In all three cases, molecular recognition is mediated by very short and fast-evolving sequences that are relatively unspecific in isolation, with more than one often being required for function. Identifying the correct sequence is a true needle-in-a-haystack problem, for nature and computational techniques alike.
New motifs are a treasure trove for investigations to deduce the molecular details of protein–protein interactions, particularly to understand those not mediated by domains alone. Given the essential regulatory functions of the motifs already known, we expect our new discoveries to have a profound impact on understanding the complex network of macromolecular interactions that exists in all living cells.
Materials and Methods
For proteins in all sets, we identified domains using SMART [46], including domains from Pfam-A [47]. We also removed regions showing similarity between members in a set of sequences using BLAST (E ≤ 0.001) [48], which removes the redundant measurements. We used TEIRESIAS [9] to detect all non-overlapping motifs of three to eight residues, requiring at least two identical positions. The method essentially detects all motifs of a variable length (i.e., three to eight) in which positions can either be specified as a particular amino acid, or represented by a wildcard (i.e., “x”). We did not allow for conservative substitutions (e.g., D/E), and ignored any motif that occurred in fewer than three sequences in the set.
We assessed the significance of a particular motif occurring a certain number of times within a set of sequences (interaction set) using the binomial distribution:
where p is the probability of seeing the motif in a background database, n is how often the motif was seen in the set of proteins, and M the size of the set.
The probability (p) was computed as a frequency of the motif in the background database of 15,000 randomly selected proteins. These proteins were taken from the SWISSPROT [49] and were subjected to the same filtering procedure as the test protein sets.
Values agree well with intuition: Motifs that are complex and thus rare need only be observed a few times to be significant, for example, the motif PxVPLR occurring in four out of 21 proteins gives a probability of 10−11. More common motifs must be seen more often to reach the same significance; for example, the VxxR (a subset of the first motif) must be seen in 19 out of 21 to reach a similar probability.
True instances of linear motifs are typically conserved across closely related species [42]. It is thus an advantage to use the information from the same (i.e., orthologous) protein in multiple genomes. Information from orthologues can be readily combined into a single value (Scons), which is the product of all binomial probabilities from the genomes considered:
This procedure will decrease the final value (and thus increase the significance) for all conserved motifs, but will have no effect if the motifs (or indeed the orthologues) are missing. The combined value is no longer a true probability, because the motifs from related species are not independent, but rather are a measure of likelihood of a conserved motif to occur at random in the set. To estimate significance we thus compare the values to those generated from random sets of proteins. These combined values greatly improve the sensitivity and specificity of the procedure: More known motifs are recovered and fewer clearly false predictions are made.
To get confidence thresholds for Scons, we created 50 random sets of sequences of the same number and length as seen in the interaction sets for each organism using the complete proteomes. We then ran the complete procedure for each random set and computed the distribution of Scons, which gave thresholds (p-value < 0.001) for each dataset: 3.0 × 10−17 for yeast, 7.5 × 10−14 for nematode, 8.0 × 10−15 for fly, and 7.0 × 10−38 for human. The differences between the thresholds are due largely to differences in the number and similarity of closely related species with complete genomes available: Four substantially similar genomes were available for human but only one for the fly and nematode.
We extracted orthologues from the STRING database [50] and aligned those using MUSCLE [51] with default parameters. We considered only closely related species because known instances of linear motifs are rarely conserved outside of them. We considered orthologues in the four other completely sequenced yeast genomes (Kluyveromyces lactis, Ashbya gossypii, Debaryomyces hansenii, and Candida glabrata) for yeast (S. cerevisiae) motifs, D. pseudoobscura for fly (D. melanogaster), C. briggsae for nematode (C. elegans), and Mus musculus, Rattus norvegicus, Gallus gallus, and Fugu rubripes for motifs found in human (H. sapiens) proteins.
The Linear Motif Discovery (LMD) program and all data related to this paper are available online (http://lmd.embl.de).
Supporting Information
Protocol S1 Supplementary Information
(289 KB PDF).
Click here for additional data file.
Figure S1 Schematic of Discovery Process of Linear Motifs Recognized by Protein Domains
(77 KB PDF).
Click here for additional data file.
Table S1 All Significant Motifs from the Protein Sets and the Rediscovered Motif:Domain Sssociations for Yeast, Fly, Nematode, and Human Interaction Datasets
(173 KB PDF).
Click here for additional data file.
We are grateful to the other members of the Eukaryotic Linear Motif (ELM) consortium, particularly Rein Aasland, Pål Puntervoll, and Morten Mattingsdal (University of Bergen, Norway) for advice, to Lars Juhl Jensen (European Molecular Biology Laboratory [EMBL]) and Ewan Birney (European Bioinformatics Institute [EBI], Hinxton, United Kingdom) for detailed help with the statistics, to Hugo Ceulemans and Mathieu Bollen (Katholieke Universiteit Leuven, Belgium) for advice and PP1 clones, Sean Hooper (EMBL) for NetView used in Figure 2, and Christian von Mering (EMBL) for help defining orthologues. We thank Patrick Aloy and Peer Bork (EMBL) for a critical reading of the manuscript. Part of this work was supported by a grant from the European Commission.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. VN and RBR conceived and designed the experiments. VN and ISA performed the experiments. VN and RBR analyzed the data. RL, AS, FdM, TJG, JL, and LS contributed reagents/materials/analysis tools. VN, RL, AS, and RBR wrote the paper.
Citation: Neduva V, Linding R, Su-Angrand I, Stark A, de Masi F, et al. (2005) Systematic discovery of new recognition peptides mediating protein interaction networks. PLoS Biol 3(12): e405.
Abbreviations
PP1protein phosphatase 1
Sconsthe product of all binomial probabilities for sets of orthologous proteins from a given set of genomes
==== Refs
References
Poglitsch CL Meredith GD Gnatt AL Jensen GJ Chang WH Electron crystal structure of an RNA polymerase II transcription elongation complex Cell 1999 98 791 798 10499796
Jeffrey PD Russo AA Polyak K Gibbs E Hurwitz J Mechanism of CDK activation revelated by the structure of a cyclinA-CDK2 complex Nature 1995 376 313 320 7630397
Pawson T Scott JD Signaling through scaffold, anchoring, and adaptor proteins Science 1997 278 2075 2080 9405336
Sudol M From Src Homology domains to other signaling modules: Proposal of the ‘protein recognition code' Oncogene 1998 17 1469 1474 9779993
Puntervoll P Linding R Gemund C Chabanis-Davidson S Mattingsdal M ELM server: A new resource for investigating short functional sites in modular eukaryotic proteins Nucleic Acids Res 2003 31 3625 3630 12824381
Pawson T Nash P Assembly of cell regulatory systems through protein interaction domains Science 2003 300 445 452 12702867
Paroush Z Finley RL Kidd T Wainwright SM Ingham PW Groucho is required for Drosophila neurogenesis, segmentation, and sex determination and interacts directly with hairy-related bHLH proteins Cell 1994 79 805 815 8001118
Prelich G Stillman B Coordinated leading and lagging strand synthesis during SV40 DNA replication in vitro requires PCNA Cell 1988 53 117 126 2894900
Rigoutsos I Floratos A Combinatorial pattern discovery in biological sequences: The TEIRESIAS algorithm Bioinformatics 1998 14 55 67 9520502
Uetz P Giot L Cagney G Mansfield TA Judson RS A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae Nature 2000 403 623 627 10688190
Ito T Chiba T Ozawa R Yoshida M Hattori M A comprehensive two-hybrid analysis to explore the yeast protein interactome Proc Natl Acad Sci U S A 2001 98 4569 4574 11283351
Giot L Bader JS Brouwer C Chaudhuri A Kuang B A protein interaction map of Drosophila melanogaster
Science 2003 302 1727 1736 14605208
Li S Armstrong CM Bertin N Ge H Milstein S A map of the interactome network of the metazoan C. elegans
Science 2004 303 540 543 14704431
Peri S Navarro JD Amanchy R Kristiansen TZ Jonnalagadda CK Development of human protein reference database as an initial platform for approaching systems biology in humans Genome Res 2003 13 2363 2371 14525934
von Mering C Krause R Snel B Cornell M Oliver SG Comparative assessment of large-scale data sets of protein-protein interactions Nature 2002 417 399 403 12000970
Terry-Lorenzo RT Elliot E Weiser DC Prickett TD Brautigan DL Neurabins recruit protein phosphatase-1 and inhibitor-2 to the actin cytoskeleton J Biol Chem 2002 277 46535 46543 12270929
Aloy P Russell RB The third dimension for protein interactions and complexes Trends Biochem Sci 2002 27 633 638 12468233
Han J Gu W Hecht N Testis-brain RNA-binding protein, a testicular translational regulatory RNA-binding protein, is present in the brain and binds to the 3′ untranslated regions of transported brain mRNAs Biol Reprod 1995 53 707 717 7578697
Aoki K Suzuki K Sugano T Tasaka T Nakahara K A novel gene, Translin, encodes a recombination hotspot binding protein associated with chromosomal translocations Nat Genet 1995 10 167 174 7663511
Warbrick E The puzzle of PCNA's many partners Bioessays 2000 22 997 1006 11056476
Chinnadurai G CtBP, an unconventional transcriptional corepressor in development and oncogenesis Mol Cell 2002 9 213 224 11864595
Bollen M Combinatorial control of protein phosphatase-1 Trends Biochem Sci 2001 26 426 431 11440854
Chang JS Henry K Wolf BL Geli M Lemmon SK Protein phosphatase-1 binding to scd5p is important for regulation of actin organization and endocytosis in yeast J Biol Chem 2002 277 48002 48008 12356757
Agarwal-Mawal A Paudel HK Neuronal Cdc2-like protein kinase (Cdk5/p25) is associated with protein phosphatase 1 and phosphorylates inhibitor-2 J Biol Chem 2001 276 23712 23718 11320080
Terrak M Kerff F Langsetmo K Tao T Dominguez R Structural basis of protein phosphatase 1 regulation Nature 2004 429 780 784 15164081
Lo KW Naisbitt S Fan JS Sheng M Zhang M The 8-kDa dynein light chain binds to its targets via a conserved (K/R)XTQT motif J Biol Chem 2001 276 14059 14066 11148209
Schnorrer F Bohmann K Nusslein-Volhard C The molecular motor dynein is involved in targeting swallow and bicoid RNA to the anterior pole of Drosophila oocytes Nat Cell Biol 2000 2 185 190 10783235
Kamura T Sato S Haque D Liu L Kaelin WG The Elongin BC complex interacts with the conserved SOCS-box motif present in members of the SOCS, ras, WD-40 repeat, and ankyrin repeat families Genes Dev 1998 12 3872 3881 9869640
Zhang JG Farley A Nicholson SE Willson TA Zugaro LM The conserved SOCS box motif in suppressors of cytokine signaling binds to elongins B and C and may couple bound proteins to proteasomal degradation Proc Natl Acad Sci U S A 1999 96 2071 2076 10051596
Rocheleau CE Yasuda J Shin TH Lin R Sawa H WRM-1 activates the LIT-1 protein kinase to transduce anterior/posterior polarity signals in C. elegans
Cell 1999 97 717 726 10380924
Shin TH Yasuda J Rocheleau CE Lin R Soto M MOM-4, a MAP kinase kinase kinase-related protein, activates WRM-1/LIT-1 kinase to transduce anterior/posterior polarity signals in C. elegans
Mol Cell 1999 4 275 280 10488343
Lo MC Gay F Odom R Shi Y Lin R Phosphorylation by the beta-catenin/MAPK complex promotes 14–3–3-mediated nuclear export of TCF/POP-1 in signal-responsive cells in C. elegans
Cell 2004 117 95 106 15066285
Kretzschmar M Kaiser K Lottspeich F Meisterernst M A novel mediator of class II gene transcription with homology to viral immediate-early transcriptional regulators Cell 1994 78 525 534 8062392
Ge H Roeder RG Purification, cloning, and characterization of a human coactivator, PC4, that mediates transcriptional activation of class II genes Cell 1994 78 513 523 8062391
Brandsen J Werten S van der Vliet PC Meisterernst M Kroon J C-terminal domain of transcription cofactor PC4 reveals dimeric ssDNA binding site Nat Struct Biol 1997 4 900 903 9360603
Shen H Kan JLC Green MR Arginine-serine-rich domains bound at splicing enhancers contact the branchpoint to promote prespliceosome assembly Molecular Cell 2004 13 367 376 14967144
Sprangers R Groves MR Sinning I Sattler M High-resolution X-ray and NMR structures of the SMN Tudor domain: Conformational variation in the binding site for symmetrically dimethylated arginine residues J Mol Biol 2003 327 507 520 12628254
Musacchio A How SH3 domains recognize proline Adv Protein Chem 2002 61 211 268 12461825
Yaffe MB Leparc GG Lai J Obata T Volinia S A motif-based profile scanning approach for genome-wide prediction of signaling pathways Nat Biotechnol 2001 19 348 353 11283593
Landgraf C Panni S Montecchi-Palazzi L Castagnoli L Schneider-Mergener J Protein interaction networks by proteome peptide scanning PLoS Biol 2004 2 E14 14737190
Benes CH Wu N Elia AE Dharia T Cantley LC The C2 domain of PKCdelta is a phosphotyrosine binding domain Cell 2005 121 271 280 15851033
Neduva V Russell RB Linear motifs: Evolutionary interaction switches FEBS Lett 2005 579 3342 3345 15943979
Xie X Lu J Kulbokas EJ Golub TR Mootha V Systematic discovery of regulatory motifs in human promoters and 3′ UTRs by comparison of several mammals Nature 2005 434 338 345 15735639
Ettwiller LM Rung J Birney E Discovering novel cis-regulatory motifs using functional networks Genome Res 2003 13 883 895 12727907
Stark A Brennecke J Russell RB Cohen SM Identification of Drosophila microRNA targets PLoS Biol 2003 1 E60 14691535
Letunic I Goodstadt L Dickens NJ Doerks T Schultz J Recent improvements to the SMART domain-based sequence annotation resource Nucleic Acids Res 2002 30 242 244 11752305
Bateman A Birney E Cerruti L Durbin R Etwiller L The Pfam protein families database Nucleic Acids Res 2002 30 276 280 11752314
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
Boeckmann B Bairoch A Apweiler R Blatter MC Estreicher A The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003 Nucleic Acids Res 2003 31 365 370 12520024
von Mering C Huynen M Jaeggi D Schmidt S Bork P STRING: A database of predicted functional associations between proteins Nucleic Acids Res 2003 31 258 261 12519996
Edgar RC MUSCLE: Multiple sequence alignment with high accuracy and high throughput Nucleic Acids Res 2004 32 1792 1797 15034147
Schaeper U Boyd JM Verma S Uhlmann E Subramanian T Molecular cloning and characterization of a cellular phosphoprotein that interacts with a conserved C-terminal domain of adenovirus E1A involved in negative modulation of oncogenic transformation Proc Natl Acad Sci U S A 1995 92 10467 10471 7479821
Cazalla D Zhu J Manche L Huber E Krainer AR Nuclear export and retention signals in the RS domain of SR proteins Mol Cell Biol 2002 22 6871 6882 12215544
Egloff MP Johnson DF Moorhead G Cohen PT Cohen P Structural basis for the recognition of regulatory subunits by the catalytic subunit of protein phosphatase 1 EMBO J 1997 16 1876 1887 9155014
Lee C Chang JH Lee HS Cho Y Structural basis for the recognition of the E2F transactivation domain by the retinoblastoma tumor suppressor Genes Dev 2002 16 3199 3212 12502741
Stomski FC Dottore M Winnall W Guthridge MA Woodcock J Identification of a 14–3–3 binding sequence in the common beta chain of the granulocyte-macrophage colony-stimulating factor (GM-CSF), interleukin-3 (IL-3), and IL-5 receptors that is serine-phosphorylated by GM-CSF Blood 1999 94 1933 1942 10477722
Molloy DP Milner AE Yakub IK Chinnadurai G Gallimore PH Structural determinants present in the C-terminal binding protein binding site of adenovirus early region 1A proteins J Biol Chem 1998 273 20867 20876 9694833
Bruning JB Shamoo Y Structural and thermodynamic analysis of human PCNA with peptides derived from DNA polymerase-delta p66 subunit and flap endonuclease-1 Structure (Camb) 2004 12 2209 2219 15576034
|
16279839
|
PMC1283537
|
CC BY
|
2021-01-05 08:21:45
|
no
|
PLoS Biol. 2005 Dec 15; 3(12):e405
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030405
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1627983910.1371/journal.pbio.0030405Research ArticleBioinformatics/Computational BiologyBiotechnologyMolecular Biology/Structural BiologyBiochemistryYeastDrosophilaCaenorhabditisHomo (Human)Systematic Discovery of New Recognition Peptides Mediating Protein Interaction Networks Systematic Discovery of New Recognition PeptidesNeduva Victor
1
Linding Rune
1
Su-Angrand Isabelle
1
Stark Alexander
1
de Masi Federico
1
Gibson Toby J
1
Lewis Joe
1
Serrano Luis
1
Russell Robert B [email protected]
1
2
1European Molecular Biology Laboratory, Heidelberg, Germany2European Molecular Biology Laboratory–European Bioinformatics Institute, Hinxton, United KingdomMatthews Rowena Academic EditorUniversity of MichiganUnited States of America12 2005 15 11 2005 15 11 2005 3 12 e4059 8 2005 27 9 2005 Copyright: © 2005 Neduva 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.
A Systematic Way to Find Linear Motifs Mediating Protein-Protein Interactions
Many aspects of cell signalling, trafficking, and targeting are governed by interactions between globular protein domains and short peptide segments. These domains often bind multiple peptides that share a common sequence pattern, or “linear motif” (e.g., SH3 binding to PxxP). Many domains are known, though comparatively few linear motifs have been discovered. Their short length (three to eight residues), and the fact that they often reside in disordered regions in proteins makes them difficult to detect through sequence comparison or experiment. Nevertheless, each new motif provides critical molecular details of how interaction networks are constructed, and can explain how one protein is able to bind to very different partners. Here we show that binding motifs can be detected using data from genome-scale interaction studies, and thus avoid the normally slow discovery process. Our approach based on motif over-representation in non-homologous sequences, rediscovers known motifs and predicts dozens of others. Direct binding experiments reveal that two predicted motifs are indeed protein-binding modules: a DxxDxxxD protein phosphatase 1 binding motif with a K
D of 22 μM and a VxxxRxYS motif that binds Translin with a K
D of 43 μM. We estimate that there are dozens or even hundreds of linear motifs yet to be discovered that will give molecular insight into protein networks and greatly illuminate cellular processes.
Many protein interactions are mediated by short amino acid motifs. The authors describe a new approach to identify these interaction motifs and experimentally validate some of their binding predictions.
==== Body
Introduction
Protein interactions are central to all cellular processes. At the molecular level they can occur in a variety of ways. Probably the best known involve specific contacts between globular domains (∼100–200 residues) present in the interacting proteins. These are seen in many different contexts ranging from different subunits in large molecular machines (e.g., RNA polymerase II [1]), to more transient interactions (e.g., cyclins binding to CDK2 [2]).
However not all interactions are mediated by pairs of globular domains. Many involve the binding of a domain in one protein to short regions (approximately three to eight residues) in another [3,4]. These regions often show a particular sequence pattern, or “linear motif,” which captures the key residues involved in function or binding [5]. Linear motifs are critical to many processes including signal transduction (e.g., SH3 domains bind PxxP [6]), gene expression (e.g., Groucho→WRPW [7]) and DNA replication (e.g., PCNA→QxxxxxFF [8]).
In contrast to domains, which are readily detectable by sequence comparison, linear motifs are difficult to discover due to their short length, a tendency to reside in disordered regions in proteins, and limited conservation outside of closely related species. To date they have typically been found by time-consuming experiments, meaning that only a few hundred motifs are known compared to thousands of domains that might bind them. Although it is at present difficult to estimate just how many such interaction motifs exist, it is likely that many interactions are mediated by those not yet discovered. Here we perform the first systematic attempt to discover new motif candidates and their corresponding binding partners using results of genome-scale interaction datasets.
Results
Methodology
Our central hypothesis is that proteins with a common interaction partner will share a feature that mediates binding, either a domain or a linear motif. In the absence of a shared domain, a linear motif could well be the only common sequence feature and might thus be detectable simply by virtue of over-representation, which is the basis of our approach (Figure1).
Figure 1 Schematic of the Linear Motif Discovery Strategy
Interaction maps are probed for interaction sets (A): Partners of proteins with multiple interactions are clustered together when there are no known sequence features present (B). Domains and homologous regions are then identified (B) and removed prior to running exhaustive pattern discovery (C) to produce a list of motifs ranked by their probabilities P (D). Hypothetical motifs are shown as coloured squares in (C) and (D). “Proteins” in (D) gives the set of proteins containing at least one copy of the motif.
Given a set of proteins sharing an interaction partner we first remove sequence regions unlikely to contain linear motifs: globular domains, trans-membrane segments, coiled-coils, collagen regions, and signal peptides. This is justified because only 15% of known linear motifs [5] occur within these regions, and including them can give rise to misleading motif signals, particularly if common domains are found in more than one protein in a set. Most importantly, this avoids the detection of repetitive, purely structural patterns, such as β-turns, coiled-coil heptads, or collagen repeats, because these are unlikely to occur in the unstructured parts of proteins that remain after this filtering. We also compare all sequences in a set to each other and leave only one representative of any homologous segments. We do this in order to measure over-representation that is not the result of homology; our assumption is that each of the remaining instances of a particular motif has arisen convergently and is thus an independent observation. We specifically avoid removing regions of low complexity because linear motifs frequently occur within them.
We then find all three to eight residue motifs in the remaining sequence [9], and score their over-representation as the binomial probability (P) of seeing them randomly in a similar set of sequences (see Materials and Methods). This allows multiple observations of an otherwise insignificant motif to become statistically significant by over-representation, and readily accounts for sets of different sizes and composition. For example, the SH3-binding pattern RxPxxP readily occurs in about one out of 20 randomly selected proteins, but its occurrence in seven sequences in a set of nine becomes highly significant. We also compute P for all closely related species based on whether or not the same motifs are seen in the corresponding orthologues, and multiply these to give a final score (Scons; see Materials and Methods).
We applied our approach to interacting sets of proteins from Saccharomyces cerevisiae, Drosophila melanogaster, Caenorhabditis elegans, and Homo sapiens [10–14]. For the first three species, these datasets are from yeast two-hybrid screens; human data comes from the human Proteome Resource Database (HPRD) [14] and consists of hand-curated interactions extracted from the literature (see Protocol S1). For each dataset, we constructed a control by selecting random sets of proteins of a similar length and number, and performed the same calculations. We then defined a confidence threshold (p-value < 0.001) for Scons for each dataset (see Materials and Methods). Note that this threshold does not necessarily reflect the accuracy in terms of identifying binding motifs, only that the particular sequence pattern reported is very unlikely to arise by chance. It is possible that patterns can arise for other reasons, including localization signals or other sequence features common to protein performing similar function.
Known motifs come in different flavours, for instance canonical SH3-binding motifs (PxxP) are embellished with different amino acids, which determine the specific SH3-containing protein they bind (e.g., RxPxxP and PxxPxK). The sets of proteins above (i.e., those sharing an interaction partner) are appropriate for finding such motif flavours because each protein containing a particular instance of a domain (e.g., SH3) is considered separately. However, it is also beneficial to detect more general motifs specific to a domain family. To do this we simply merge sets if the common binding partners shared a particular domain. We refer to these as “domain” sets in the sections that follow (see Protocol S1 and Figure S1).
Benchmark
The Eukaryotic Linear Motif resource (ELM) [5] contains a curated set of experimentally validated instances of binding motifs (i.e., their location in a particular protein). This provides several pertinent sets of proteins to test the approach, namely each set of proteins containing a known instance of a particular motif (e.g., all PxxP motif–containing sequences known to interact with SH3 domains). Of 58 different sets, 22 contained at least four non-homologous instances of the motif, and could be used to test our approach. We ran the procedure on each set and monitored where the known motif (or a variant) was found in the list of all motifs ranked according to Scons. Despite many thousands of possibilities, the approach detected the correct motif as the very best ranked for 14 out of 22 and among the top ten for an additional three (Table 1). Applying the confidence threshold left eleven correct motifs at first rank, and no false predictions (see legend to Table 1). Inspection showed that those motifs that were either missed or scored poorly were generally highly degenerate in nature (e.g., the sumoylation site (VILAFP)Kx(EDNGP)).
Table 1 Detection of Known Linear Motifs in Experimentally Verified Sets from the Eukaryotic Linear Motif
Motifs in Genome-Scale Interaction Sets
Considering the genome-scale interactions, each dataset produced a number of protein sets sharing a common interaction partner: yeast, 191; fly, 632; nematode, 367; and human, 1,986. Only a small fraction of these produced one or more confident motifs (as assessed by the binomial probability): yeast, 11; fly, 26; nematode, 27; and human, 112. In all cases, known motifs were among those produced, though to varying degrees: yeast, 1 (domain set); fly, 9; nematode, 4 (domain set); and human, 48 (all significant motifs from the protein set are given in Protocol S1 and Table S1; all motifs, including those with poorer significance, are available at http://lmd.embl.de). Figure 2 shows a summary of the 26 motifs found in the fly set, highlighting the nine rediscoveries of known motifs (including one likely nuclear localization signal). The better results in human data (i.e., 48/112) are undoubtedly because the hand-curated interactions (HPRD, [14]) contain fewer errors than those from the comparatively noisy high-throughput yeast two-hybrid screens for the other organisms. Here we found motifs spanning virtually the full range of those known (SH3→PxxP, 14–3–3 proteins→RxxSxP, Clathrin→LDxL, etc.), in addition to several that appear to be novel.
Figure 2 Overview of Motifs Found in the Fly
Significant predictions from the yeast two-hybrid set for the fly. Blue dots in the center of each cluster represent proteins with four or more interaction partners (red and white dots) containing at least one confidently predicted motif (p-value < 0.001; Scons ≤ 8 × 10−15). Partner proteins containing the motif are represented by red dots, whereas proteins lacking the motif are indicated by white dots. Clusters are labelled as gene name→detected motif. Yellow circles enclose known motifs: SH3→PxxP [38], PP1→RVxF [22], C-terminal binding protein (CtBP)→PxDLS [52], SR splicing factors RS-rich segments [53], and CG6843→SxKSKxxK, a likely nuclear localization signal. The Translin→VxxxRxYS motif was experimentally tested (Figure 3). The grey circles enclose clusters with low-complexity patterns. Two additional known motifs were also found in the fly using more relaxed criteria than those used for the other motifs in the figure: Groucho→WRPW [7] and Dynein light chain→TQT [26] as the variant A(TI)QT(DE). The latter was also identified as significant in the domain sets. Proteins are denoted either by their FlyBase accession codes or protein names when available.
Inspection showed that known motifs were typically missed because the sets contained too few sequences with the correct motif to reach significance. For example, in yeast interaction data, just four out of 23 proteins interacting with the protein phosphatase 1 (PP1) domain contained the established (RK)VxF motif. A similar situation occurred with WW domains, where no more than three instances of known motifs were found among their interaction partners. It could be that certain motifs are just too rare in the interacting set to be detected. However, it is also well established that the yeast two-hybrid system, particularly when applied in genome-screens, can miss known interactions [15] and, moreover, make false predictions that cloud the signal from true motifs. The prediction accuracy and coverage will certainly increase when more comprehensive and reliable interaction data become available. The error prone nature of the underlying yeast two-hybrid data for yeast, fly, and nematode might be expected to yield inconsistencies (i.e., different motifs for the same protein) when comparing predictions from different species. Encouragingly, however, we found very few of these, and indeed in one case (see PP1, in Experimental Testing of New Motifs), we think the apparent inconsistency corresponds to two distinct motifs that bind to the same protein, each detected in a different species.
Many of the motifs detected in the protein sets were also found when interaction partners were pooled owing to the presence of a common domain (domain sets). Frequently a more general motif was found in the domain than in the protein sets. For example, in the fly we identified the canonical TQT motif as the binding site for Dynein light chain domains, but found the more specific pattern A(TI)QT(DE) for the specific Dynein homologue Cdlc2. Other motifs were seen only in one of the protein or domain sets. For example, in yeast a correct SH3 motif was only found in the domain sets, because no single SH3-domain protein had a sufficient number of interaction partners for the motifs to be found with significance. The reverse was true in the fly in which the correct SH3 motif was found only in the protein set, because the domain set had too many proteins lacking the canonical motif, meaning that the signal was lost.
There is also a problem of ambiguity in both protein and domain sets. For multi-domain proteins predicted to bind a motif, it is not possible to discern which domain is mediating the interaction. This can be partly resolved by considering domain sets, but even here there are still some examples where genuine motifs were predicted for the wrong domains. Inspection showed that this was either because the process selected the wrong domain of a frequently co-occurring pair (e.g., SH2 domains predicting to bind SH3 ligands) or selected an activator/inhibitor of the correct binding domain (e.g., protein phosphatase inhibitor 2 (IPP2) binding to PP1-like RVxF-like motifs [16]). The latter highlights the possibility of the yeast two-hybrid system identifying indirect interactions [17]. These domain ambiguities can likely only be resolved by experiment.
Experimental Testing of New Motifs
For a selection of new protein→motif associations, we tested direct binding via fluorescence anisotropy, using labelled peptides corresponding to the regions containing the predicted motifs (Protocol S1). Because the motifs we have predicted might be the lowest common denominator of what could be a slightly longer binding region, we included two additional residues to the N- and C-terminus of each peptide (extracted from the original sequence). We first selected candidate motifs in yeast, fly, and nematode based on the feasibility of expressing and purifying the common interaction partner (i.e., the protein predicted to bind the motif). Of the 55 significant novel motif predictions, only 13 contained a single globular domain, were not excessively long (≤ 650 amino acids), and lacked long regions of predicted disorder or low complexity. From these we selected five that had available clones and established purification protocols. These spanned a range of novelty, ranging from variations on a known motif, to those for which there was some supporting, but not direct, evidence in the literature, to those lacking any additional support. Of the five selected, we could obtain clones for four and could purify three.
We tested a highly significant Translin→VxxxRxYS motif found in fly data (Figure 3A). Translin is a protein thought to be involved in chromosomal rearrangements, and binds double-stranded RNA and DNA [18,19]. The fluorescence polarisation assay shows that it binds the peptide motif specifically compared to a mutated counterpart, or randomly selected peptides (Figure 3B). The affinity of binding (K
D = 43 ± 15 μM) is within the range typical for known linear motifs when considered in isolation (5–150 μM; see Table 1). Mutated controls or arbitrarily chosen peptides do not show specific binding (Figure 3B; note that the apparent linear increase in both is due to the high protein concentrations reached). We can only speculate what role this motif plays in modulating Translin function. However, there are several precedents for interaction motifs playing critical regulatory roles by binding to other DNA- or RNA-binding proteins, such as PCNA [20] or CtBP [21].
Figure 3 A Novel Fly VxxxRxYS Motif That Binds Translin
(A) Translin (left) shown surrounded by interaction partners containing the predicted motif VxxxRxYS. Proteins are shown as lines with domains (labelled shapes), predicted coiled coils (light blue/green segments), and the location of motifs (blue vertical bars). Sequences for the motif-containing region are shown aligned to the best homologues in closely related species. Amino acids are coloured according to residue type: blue, positive; red, negative; light blue, small; yellow, hydrophobic; green, aromatic; magenta, polar; and orange, proline. Those constituting the predicted motif are denoted by circles. Aga, Anopheles gambiae; Dme, D. melanogaster; Dps, D. pseudoobscura.
(B) Saturation curves, showing bound fraction (fluorescently labelled peptides at saturation) as a function of Translin concentration. Polarization values (mP) at zero concentration and Bmax were normalised to give the bound fraction. K
D was computed by non-linear regression on values from three independent experiments. The lower panel shows the alignment of the native and mutated peptides together with the arbitrary peptide (selected randomly). Black triangles show positions specifying the motif (VxxxRxYS). The alignment is coloured as described in (A).
We also tested a DxxDxxxD motif found in 10 of 12 interaction partners of yeast protein phosphatase 1 (PP1, Figure 4A). Eight are well-known PP1 interactors, and five contain the canonical RVxF PP1-binding motif. Fluorescence polarisation shows that a peptide corresponding to the region in Scd5 binds specifically to PP1 (K
D = 22 ± 5 μM), compared to arbitrary peptides (Figure 4B). Inspection of other PP1-binding proteins [22] reveals that 12 of 33 also contain the new motif, with an additional 15 containing a more relaxed pattern (permitting Glu). Deletions of the canonical RVxF motif do not always disrupt PP1-binding, and have led others to suggest additional binding sites [23]. Interestingly, deletions of some segments containing this new motif can affect PP1 binding in other proteins [22]. Other support comes from pull-down studies, which identified a similar region (RVRLDDDDE) critical for the Cdk5–PP1 interaction [24] and the recent crystal structure of human PP1 bound to a myosin-targeting subunit MYPT1, which led the authors to propose a positively charged surface on which a similar acidic stretch could interact [25]. Interestingly, the mutated control appears to retain some affinity, probably owing to the presence of additional negative charges that have not been mutated to alanine, and indeed a near match to the motif (DxxxExxD) is still present in the mutant. Arbitrary peptides did not show any specific binding (Figure 4B).
Figure 4 An Acidic Yeast PP1 Binding Motif
(A) PP1 (Glc7) with the set of interaction partners containing the DxxDxxxD motifs. Details are as for Figure 3A. Here the location of RVxF motifs (defined as matches to (RK)x0–1(VI)x(FW)) are shown as yellow bars, and low-complexity regions are magenta. The figure also shows the structure of PP1 bound to RVxF (red spheres) [54] with a hypothetical helix containing the motif. Blue spheres show the location of Arg or Lys residues, and the active site is circled with critical Arginines shown in ball-and-stick. Red arrows show hypothetical interactions of the motif either with sites on PP1 or elsewhere. Ani, Aspergillus nidulans; Cal, Candida albicans; Ego, Eremothecium gossypii; Gze, Gibberella zeae; Mgr, Magnaporthe grisea; Sce, S. cerevisiae; Spo, Schizosaccharomyces pombe; Str, Salinospora tropicalis; Uma, Ustilago maydis; Xla, Xenopus laevis.
(B) Saturation curves, showing bound fraction as a function of PP1 concentration. The polarization values (mP) were normalized to an extrapolated Bmax because Bmax could not be reached experimentally. Other details are as given in Figure 3B. Red triangles in the lower panel show the location of the near match to the motif in the mutated sequence.
Lastly, we tested a variant of the well-known Dynein light chain binding motif. The canonical motif has a consensus sequence (KR)xTQT and mediates interactions important for cell trafficking [26]. We found the canonical motif in the fly, but noticed a variant, IQTE, among three partners of Cdlc2 (Dynein light chain 2), which is similar to one present in the protein swallow from D. pseudoobscura [27]. We could detect no binding of the Cdlc2 to fluorescently labelled peptides over a range of protein concentrations (5–400 μM). Surprisingly, a true instance of the motif, known to bind Cdlc2 in vivo and in vitro [27], also did not give a signal using this procedure, suggesting that the experimental assay might not be suitable for Dynein light chain interactions (see Protocol S1).
Other Promising Predicted Motifs
For other predictions, we scrutinized the literature for previous experiments hinting that the motifs could be genuine. For example, among several interesting predictions in the fly was an Elongin C→LxxLCxR motif, which has been described previously only as part of a longer sequence called the SOCS box [28]. Only three of four interacting proteins with the motif contain the full SOCS box. The protein lacking it (CG18171) is not well understood; the interaction has not been reported apart from the genome screen. Deletion and mutagenesis experiments have shown that this region is important for the interaction with Elongin C [29]. Our finding agrees with this and further suggests that the motif could be sufficient on its own for mediating the interaction.
We found the motif SxPxxxS in 11 of 17 interaction partners of the nematode MAP-kinase lit-1 involved in wnt signalling and morphogenesis (Figure 5). These include three well-known regulators/interactors of lit-1: two nuclear proteins (wrm-1 and mom-4) and another morphogenesis protein (pop-1). Deletions have already demonstrated that regions containing the motif are critical for lit-1 binding (yellow boxes in Figure 5): a 148 N-terminal segment in wrm-1 [30], a 21-residue stretch in mom-4 [31], and a 45-residue region just six residues N-terminal to the motif in pop-1 [32] all disrupt lit-1 binding.
Figure 5 A Lit-1 MAP Kinase SxPxxxS Motif
The MAP kinase lit-1 surrounded by its interaction partners containing the SxPxxxS motif. Details are as for Figure 3. Yellow boxes show the location of deletion mutants known to affect the interaction. Cbr, C. briggsae; Cel, C. elegans.
Among several compelling new motifs in human was a T(PL)QP motif predicted to bind to the transcription factor PC4 (Positive Cofactor 4). PC4 binds double-stranded DNA and promotes the assembly of the preinitiation complex via a mechanism that is not fully understood [33,34]. The five proteins containing the putative motif all participate in transcription, but share no common globular domain that could mediate binding to PC4. Such a proline-rich motif could be a good candidate to bind one of the several aromatic patches on the surface of the PC4 protein [35].
Low-Complexity Linear Motifs
In both fly and nematode, several very significant motifs arose from regions of low sequence complexity (i.e., dominated by a few amino acids). These included examples already known to mediate interactions, and others not described previously, including His-, Ser-, Lys-, and Glu/His-rich motifs. We could find no motifs like these in random sets, which suggested that they are not the result of the general prevalence of low-complexity regions within proteins, but just what they mean is an open question. They might well be true, biologically meaningful interactions, and indeed for some sets the proteins show similarities in function. This idea is supported by the fact that many known motifs, including the protein/RNA binding RS/SR motifs [36], the Tudor domain→(RG)n [37], and SH3→poly-proline ligands [38], are themselves low complexity. Alternatively, they could be the result of some artefact of the yeast two-hybrid system. The last possibility is supported by the fact that we found fewer such motifs in the less error-prone human data.
How Many Protein–Motif Interaction Pairs Are Still to Be Found?
Both our experiments and those done previously suggest that many of our findings are genuine motifs that have not yet been reported. This raises the question as to how many new interaction motifs there are yet to be discovered. An estimate can come by considering what fraction of the previously known motifs we found and extrapolating this to the new discoveries. For example, in fly we predict 26 motifs of which nine are known, from a total number of roughly 60 that are known in this organism [5]. If we assume that all the remaining motifs are correct, and assume an equal distribution of motifs in fly proteins not seen in the yeast two-hybrid data (4,683/13,833), we estimate 334 additional motifs (the equivalent number for human is 405). Even the more modest assumption of between 10%–20% of the predicted motifs being correct (roughly the fraction for which we could see direct binding experimentally, which is clearly a lower estimate) gives estimates of 33–67 new motifs in fly (40–80 in human). There are very likely dozens to hundreds of new motifs to be discovered, which will correspond potentially to thousands of individual binding sites. To date we have just scratched the surface of what is likely a sophisticated network of peptide-mediated interactions in the cell.
Discussion
Many studies continue to highlight the importance of networks mediated by linear motifs [39,40], and each new discovery opens new lines of research into critical aspects of cell function [41]. We have shown here that these very simple features can be detected successfully, even in error-prone data, provided they occur with a sufficient frequency in otherwise unrelated proteins. The approach need not be restricted to protein-protein interactions. It can also be applied in other contexts: Any set of proteins or nucleic acids can be probed for short sequences responsible for a common biological feature (cellular location, modifications, etc.).
Both globular domains and linear motifs are modular in the sense that they are reused in different functional contexts, but they probably differ in how they arise. Domain shuffling involves duplication of part of a gene and its insertion into another. In contrast, the short length of linear motifs makes them likely to arise convergently in proteins by evolutionary drift [42]. This suggests that there are probably many near matches to the motifs just waiting for an appropriate point mutation to induce a function. They are, in effect, powerful switches for nature to explore during the evolution of complex functions. In this regard they are highly similar to transcription factor–binding sites [43,44] or microRNA target sequences [45]. In all three cases, molecular recognition is mediated by very short and fast-evolving sequences that are relatively unspecific in isolation, with more than one often being required for function. Identifying the correct sequence is a true needle-in-a-haystack problem, for nature and computational techniques alike.
New motifs are a treasure trove for investigations to deduce the molecular details of protein–protein interactions, particularly to understand those not mediated by domains alone. Given the essential regulatory functions of the motifs already known, we expect our new discoveries to have a profound impact on understanding the complex network of macromolecular interactions that exists in all living cells.
Materials and Methods
For proteins in all sets, we identified domains using SMART [46], including domains from Pfam-A [47]. We also removed regions showing similarity between members in a set of sequences using BLAST (E ≤ 0.001) [48], which removes the redundant measurements. We used TEIRESIAS [9] to detect all non-overlapping motifs of three to eight residues, requiring at least two identical positions. The method essentially detects all motifs of a variable length (i.e., three to eight) in which positions can either be specified as a particular amino acid, or represented by a wildcard (i.e., “x”). We did not allow for conservative substitutions (e.g., D/E), and ignored any motif that occurred in fewer than three sequences in the set.
We assessed the significance of a particular motif occurring a certain number of times within a set of sequences (interaction set) using the binomial distribution:
where p is the probability of seeing the motif in a background database, n is how often the motif was seen in the set of proteins, and M the size of the set.
The probability (p) was computed as a frequency of the motif in the background database of 15,000 randomly selected proteins. These proteins were taken from the SWISSPROT [49] and were subjected to the same filtering procedure as the test protein sets.
Values agree well with intuition: Motifs that are complex and thus rare need only be observed a few times to be significant, for example, the motif PxVPLR occurring in four out of 21 proteins gives a probability of 10−11. More common motifs must be seen more often to reach the same significance; for example, the VxxR (a subset of the first motif) must be seen in 19 out of 21 to reach a similar probability.
True instances of linear motifs are typically conserved across closely related species [42]. It is thus an advantage to use the information from the same (i.e., orthologous) protein in multiple genomes. Information from orthologues can be readily combined into a single value (Scons), which is the product of all binomial probabilities from the genomes considered:
This procedure will decrease the final value (and thus increase the significance) for all conserved motifs, but will have no effect if the motifs (or indeed the orthologues) are missing. The combined value is no longer a true probability, because the motifs from related species are not independent, but rather are a measure of likelihood of a conserved motif to occur at random in the set. To estimate significance we thus compare the values to those generated from random sets of proteins. These combined values greatly improve the sensitivity and specificity of the procedure: More known motifs are recovered and fewer clearly false predictions are made.
To get confidence thresholds for Scons, we created 50 random sets of sequences of the same number and length as seen in the interaction sets for each organism using the complete proteomes. We then ran the complete procedure for each random set and computed the distribution of Scons, which gave thresholds (p-value < 0.001) for each dataset: 3.0 × 10−17 for yeast, 7.5 × 10−14 for nematode, 8.0 × 10−15 for fly, and 7.0 × 10−38 for human. The differences between the thresholds are due largely to differences in the number and similarity of closely related species with complete genomes available: Four substantially similar genomes were available for human but only one for the fly and nematode.
We extracted orthologues from the STRING database [50] and aligned those using MUSCLE [51] with default parameters. We considered only closely related species because known instances of linear motifs are rarely conserved outside of them. We considered orthologues in the four other completely sequenced yeast genomes (Kluyveromyces lactis, Ashbya gossypii, Debaryomyces hansenii, and Candida glabrata) for yeast (S. cerevisiae) motifs, D. pseudoobscura for fly (D. melanogaster), C. briggsae for nematode (C. elegans), and Mus musculus, Rattus norvegicus, Gallus gallus, and Fugu rubripes for motifs found in human (H. sapiens) proteins.
The Linear Motif Discovery (LMD) program and all data related to this paper are available online (http://lmd.embl.de).
Supporting Information
Protocol S1 Supplementary Information
(289 KB PDF).
Click here for additional data file.
Figure S1 Schematic of Discovery Process of Linear Motifs Recognized by Protein Domains
(77 KB PDF).
Click here for additional data file.
Table S1 All Significant Motifs from the Protein Sets and the Rediscovered Motif:Domain Sssociations for Yeast, Fly, Nematode, and Human Interaction Datasets
(173 KB PDF).
Click here for additional data file.
We are grateful to the other members of the Eukaryotic Linear Motif (ELM) consortium, particularly Rein Aasland, Pål Puntervoll, and Morten Mattingsdal (University of Bergen, Norway) for advice, to Lars Juhl Jensen (European Molecular Biology Laboratory [EMBL]) and Ewan Birney (European Bioinformatics Institute [EBI], Hinxton, United Kingdom) for detailed help with the statistics, to Hugo Ceulemans and Mathieu Bollen (Katholieke Universiteit Leuven, Belgium) for advice and PP1 clones, Sean Hooper (EMBL) for NetView used in Figure 2, and Christian von Mering (EMBL) for help defining orthologues. We thank Patrick Aloy and Peer Bork (EMBL) for a critical reading of the manuscript. Part of this work was supported by a grant from the European Commission.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. VN and RBR conceived and designed the experiments. VN and ISA performed the experiments. VN and RBR analyzed the data. RL, AS, FdM, TJG, JL, and LS contributed reagents/materials/analysis tools. VN, RL, AS, and RBR wrote the paper.
Citation: Neduva V, Linding R, Su-Angrand I, Stark A, de Masi F, et al. (2005) Systematic discovery of new recognition peptides mediating protein interaction networks. PLoS Biol 3(12): e405.
Abbreviations
PP1protein phosphatase 1
Sconsthe product of all binomial probabilities for sets of orthologous proteins from a given set of genomes
==== Refs
References
Poglitsch CL Meredith GD Gnatt AL Jensen GJ Chang WH Electron crystal structure of an RNA polymerase II transcription elongation complex Cell 1999 98 791 798 10499796
Jeffrey PD Russo AA Polyak K Gibbs E Hurwitz J Mechanism of CDK activation revelated by the structure of a cyclinA-CDK2 complex Nature 1995 376 313 320 7630397
Pawson T Scott JD Signaling through scaffold, anchoring, and adaptor proteins Science 1997 278 2075 2080 9405336
Sudol M From Src Homology domains to other signaling modules: Proposal of the ‘protein recognition code' Oncogene 1998 17 1469 1474 9779993
Puntervoll P Linding R Gemund C Chabanis-Davidson S Mattingsdal M ELM server: A new resource for investigating short functional sites in modular eukaryotic proteins Nucleic Acids Res 2003 31 3625 3630 12824381
Pawson T Nash P Assembly of cell regulatory systems through protein interaction domains Science 2003 300 445 452 12702867
Paroush Z Finley RL Kidd T Wainwright SM Ingham PW Groucho is required for Drosophila neurogenesis, segmentation, and sex determination and interacts directly with hairy-related bHLH proteins Cell 1994 79 805 815 8001118
Prelich G Stillman B Coordinated leading and lagging strand synthesis during SV40 DNA replication in vitro requires PCNA Cell 1988 53 117 126 2894900
Rigoutsos I Floratos A Combinatorial pattern discovery in biological sequences: The TEIRESIAS algorithm Bioinformatics 1998 14 55 67 9520502
Uetz P Giot L Cagney G Mansfield TA Judson RS A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae Nature 2000 403 623 627 10688190
Ito T Chiba T Ozawa R Yoshida M Hattori M A comprehensive two-hybrid analysis to explore the yeast protein interactome Proc Natl Acad Sci U S A 2001 98 4569 4574 11283351
Giot L Bader JS Brouwer C Chaudhuri A Kuang B A protein interaction map of Drosophila melanogaster
Science 2003 302 1727 1736 14605208
Li S Armstrong CM Bertin N Ge H Milstein S A map of the interactome network of the metazoan C. elegans
Science 2004 303 540 543 14704431
Peri S Navarro JD Amanchy R Kristiansen TZ Jonnalagadda CK Development of human protein reference database as an initial platform for approaching systems biology in humans Genome Res 2003 13 2363 2371 14525934
von Mering C Krause R Snel B Cornell M Oliver SG Comparative assessment of large-scale data sets of protein-protein interactions Nature 2002 417 399 403 12000970
Terry-Lorenzo RT Elliot E Weiser DC Prickett TD Brautigan DL Neurabins recruit protein phosphatase-1 and inhibitor-2 to the actin cytoskeleton J Biol Chem 2002 277 46535 46543 12270929
Aloy P Russell RB The third dimension for protein interactions and complexes Trends Biochem Sci 2002 27 633 638 12468233
Han J Gu W Hecht N Testis-brain RNA-binding protein, a testicular translational regulatory RNA-binding protein, is present in the brain and binds to the 3′ untranslated regions of transported brain mRNAs Biol Reprod 1995 53 707 717 7578697
Aoki K Suzuki K Sugano T Tasaka T Nakahara K A novel gene, Translin, encodes a recombination hotspot binding protein associated with chromosomal translocations Nat Genet 1995 10 167 174 7663511
Warbrick E The puzzle of PCNA's many partners Bioessays 2000 22 997 1006 11056476
Chinnadurai G CtBP, an unconventional transcriptional corepressor in development and oncogenesis Mol Cell 2002 9 213 224 11864595
Bollen M Combinatorial control of protein phosphatase-1 Trends Biochem Sci 2001 26 426 431 11440854
Chang JS Henry K Wolf BL Geli M Lemmon SK Protein phosphatase-1 binding to scd5p is important for regulation of actin organization and endocytosis in yeast J Biol Chem 2002 277 48002 48008 12356757
Agarwal-Mawal A Paudel HK Neuronal Cdc2-like protein kinase (Cdk5/p25) is associated with protein phosphatase 1 and phosphorylates inhibitor-2 J Biol Chem 2001 276 23712 23718 11320080
Terrak M Kerff F Langsetmo K Tao T Dominguez R Structural basis of protein phosphatase 1 regulation Nature 2004 429 780 784 15164081
Lo KW Naisbitt S Fan JS Sheng M Zhang M The 8-kDa dynein light chain binds to its targets via a conserved (K/R)XTQT motif J Biol Chem 2001 276 14059 14066 11148209
Schnorrer F Bohmann K Nusslein-Volhard C The molecular motor dynein is involved in targeting swallow and bicoid RNA to the anterior pole of Drosophila oocytes Nat Cell Biol 2000 2 185 190 10783235
Kamura T Sato S Haque D Liu L Kaelin WG The Elongin BC complex interacts with the conserved SOCS-box motif present in members of the SOCS, ras, WD-40 repeat, and ankyrin repeat families Genes Dev 1998 12 3872 3881 9869640
Zhang JG Farley A Nicholson SE Willson TA Zugaro LM The conserved SOCS box motif in suppressors of cytokine signaling binds to elongins B and C and may couple bound proteins to proteasomal degradation Proc Natl Acad Sci U S A 1999 96 2071 2076 10051596
Rocheleau CE Yasuda J Shin TH Lin R Sawa H WRM-1 activates the LIT-1 protein kinase to transduce anterior/posterior polarity signals in C. elegans
Cell 1999 97 717 726 10380924
Shin TH Yasuda J Rocheleau CE Lin R Soto M MOM-4, a MAP kinase kinase kinase-related protein, activates WRM-1/LIT-1 kinase to transduce anterior/posterior polarity signals in C. elegans
Mol Cell 1999 4 275 280 10488343
Lo MC Gay F Odom R Shi Y Lin R Phosphorylation by the beta-catenin/MAPK complex promotes 14–3–3-mediated nuclear export of TCF/POP-1 in signal-responsive cells in C. elegans
Cell 2004 117 95 106 15066285
Kretzschmar M Kaiser K Lottspeich F Meisterernst M A novel mediator of class II gene transcription with homology to viral immediate-early transcriptional regulators Cell 1994 78 525 534 8062392
Ge H Roeder RG Purification, cloning, and characterization of a human coactivator, PC4, that mediates transcriptional activation of class II genes Cell 1994 78 513 523 8062391
Brandsen J Werten S van der Vliet PC Meisterernst M Kroon J C-terminal domain of transcription cofactor PC4 reveals dimeric ssDNA binding site Nat Struct Biol 1997 4 900 903 9360603
Shen H Kan JLC Green MR Arginine-serine-rich domains bound at splicing enhancers contact the branchpoint to promote prespliceosome assembly Molecular Cell 2004 13 367 376 14967144
Sprangers R Groves MR Sinning I Sattler M High-resolution X-ray and NMR structures of the SMN Tudor domain: Conformational variation in the binding site for symmetrically dimethylated arginine residues J Mol Biol 2003 327 507 520 12628254
Musacchio A How SH3 domains recognize proline Adv Protein Chem 2002 61 211 268 12461825
Yaffe MB Leparc GG Lai J Obata T Volinia S A motif-based profile scanning approach for genome-wide prediction of signaling pathways Nat Biotechnol 2001 19 348 353 11283593
Landgraf C Panni S Montecchi-Palazzi L Castagnoli L Schneider-Mergener J Protein interaction networks by proteome peptide scanning PLoS Biol 2004 2 E14 14737190
Benes CH Wu N Elia AE Dharia T Cantley LC The C2 domain of PKCdelta is a phosphotyrosine binding domain Cell 2005 121 271 280 15851033
Neduva V Russell RB Linear motifs: Evolutionary interaction switches FEBS Lett 2005 579 3342 3345 15943979
Xie X Lu J Kulbokas EJ Golub TR Mootha V Systematic discovery of regulatory motifs in human promoters and 3′ UTRs by comparison of several mammals Nature 2005 434 338 345 15735639
Ettwiller LM Rung J Birney E Discovering novel cis-regulatory motifs using functional networks Genome Res 2003 13 883 895 12727907
Stark A Brennecke J Russell RB Cohen SM Identification of Drosophila microRNA targets PLoS Biol 2003 1 E60 14691535
Letunic I Goodstadt L Dickens NJ Doerks T Schultz J Recent improvements to the SMART domain-based sequence annotation resource Nucleic Acids Res 2002 30 242 244 11752305
Bateman A Birney E Cerruti L Durbin R Etwiller L The Pfam protein families database Nucleic Acids Res 2002 30 276 280 11752314
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
Boeckmann B Bairoch A Apweiler R Blatter MC Estreicher A The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003 Nucleic Acids Res 2003 31 365 370 12520024
von Mering C Huynen M Jaeggi D Schmidt S Bork P STRING: A database of predicted functional associations between proteins Nucleic Acids Res 2003 31 258 261 12519996
Edgar RC MUSCLE: Multiple sequence alignment with high accuracy and high throughput Nucleic Acids Res 2004 32 1792 1797 15034147
Schaeper U Boyd JM Verma S Uhlmann E Subramanian T Molecular cloning and characterization of a cellular phosphoprotein that interacts with a conserved C-terminal domain of adenovirus E1A involved in negative modulation of oncogenic transformation Proc Natl Acad Sci U S A 1995 92 10467 10471 7479821
Cazalla D Zhu J Manche L Huber E Krainer AR Nuclear export and retention signals in the RS domain of SR proteins Mol Cell Biol 2002 22 6871 6882 12215544
Egloff MP Johnson DF Moorhead G Cohen PT Cohen P Structural basis for the recognition of regulatory subunits by the catalytic subunit of protein phosphatase 1 EMBO J 1997 16 1876 1887 9155014
Lee C Chang JH Lee HS Cho Y Structural basis for the recognition of the E2F transactivation domain by the retinoblastoma tumor suppressor Genes Dev 2002 16 3199 3212 12502741
Stomski FC Dottore M Winnall W Guthridge MA Woodcock J Identification of a 14–3–3 binding sequence in the common beta chain of the granulocyte-macrophage colony-stimulating factor (GM-CSF), interleukin-3 (IL-3), and IL-5 receptors that is serine-phosphorylated by GM-CSF Blood 1999 94 1933 1942 10477722
Molloy DP Milner AE Yakub IK Chinnadurai G Gallimore PH Structural determinants present in the C-terminal binding protein binding site of adenovirus early region 1A proteins J Biol Chem 1998 273 20867 20876 9694833
Bruning JB Shamoo Y Structural and thermodynamic analysis of human PCNA with peptides derived from DNA polymerase-delta p66 subunit and flap endonuclease-1 Structure (Camb) 2004 12 2209 2219 15576034
|
16262889
|
PMC1283743
|
CC BY
|
2021-01-04 16:38:20
|
no
|
Ann Clin Microbiol Antimicrob. 2005 Oct 31; 4:17
|
latin-1
|
Ann Clin Microbiol Antimicrob
| 2,005 |
10.1186/1476-0711-4-17
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1371624204610.1186/1471-2407-5-137Research ArticleShort- and long-term cause-specific survival of patients with inflammatory breast cancer Tai Patricia [email protected] Edward [email protected] Ross [email protected] Juan [email protected] Kurian [email protected] Evgeny [email protected] Shazia [email protected] University of Saskatchewan, Faculty of Medicine, Saskatoon; Department of Radiation Oncology, Regina, Canada2 Division of Radiation Oncology, Department of Oncology, University of Western Ontario, London, Ontario, Canada2005 22 10 2005 5 137 137 3 5 2005 22 10 2005 Copyright © 2005 Tai 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
Inflammatory breast cancer (IBC) had been perceived to have a poor prognosis. Oncologists were not enthusiastic in the past to give aggressive treatment. Single institution studies tend to have small patient numbers and limited years of follow-up. Most studies do not report 10-, 15- or 20-year results.
Methods
Data was obtained from the population-based database of the Surveillance, Epidemiology, and End Results program of the National Cancer Institute from 1975–1995 using SEER*Stat5.0 software. This period of 21 years was divided into 7 periods of 3 years each. The years were chosen so that there was adequate follow-up information to 2000. ICD-O-2 histology 8530/3 was used to define IBC. The lognormal model was used for statistical analysis.
Results
A total of 1684 patients were analyzed, of which 84% were white, 11% were African Americans, and 5% belonged to other races. Age distribution was < 30 years in 1%, 30–40 in 11%, 40–50 in 22%, 50–60 in 24%, 60–70 in 21%, and > 70 in 21%. The lognormal model was validated for 1975–77 and for 1978–80, since the 10-, 15- and 20-year cause-specific survival (CSS) rates, could be calculated using the Kaplan-Meier method with data available in 2000. The data were then used to estimate the 10-, 15- and 20-year CSS rates for the more recent years, and to study the trend of improvement in survival. There were increasing incidences of IBC: 134 patients in the 1975–77 period to 416 patients in the 1993–95 period. The corresponding 20-year CSS increased from 9% to 20% respectively with standard errors of less than 4%.
Conclusion
The improvement of survival during the study period may be due to introduction of more aggressive treatments. However, there seem to be no further increase of long-term CSS, which should encourage oncologists to find even more effective treatments. Because of small numbers of patients, randomized studies will be difficult to conduct. The SEER population-based database will yield the best possible estimate of the trend in improvement of survival for patients with IBC.
==== Body
Background
Inflammatory breast cancer (IBC) occurs rarely [1]. Signs and symptoms of this condition include the presence of erythema, edema or peau d'orange appearance of the skin, and other clinical signs of disease. Diagnosis is made by skin biopsy. The definition of IBC varies in the literature and leads to some disparities. In this study, the pathological definition is used.
It is known that IBC have a poor prognosis. Oncologists were not enthusiastic to administer aggressive treatment in the past. Nowadays, treatment for this aggressive form of breast cancer is multi-modal, and includes chemotherapy, surgery, radiation therapy, and hormonal therapy [2]. The optimal sequence of the different modalities is still a subject of research [3]. Development of novel therapeutic agents continues and is based on an expanding understanding of the biology of tumor development and progression. Advances in treatment continue to improve the prognosis for this disease [4]. With a few notable exceptions, many publications on IBC do not have long periods of follow-up [5,6]. These single-institution studies are from academic centers. To our knowledge, long-term results of cases treated in the community are not available. This study examines the changes in the prognosis of IBC over the years with the Surveillance, Epidemiology, and End Results (SEER) database [7].
There is a parametric lognormal model, proposed by Boag [8-10] that has been validated retrospectively in the literature, and can be used prospectively for predicting long-term survival rates several years earlier than would otherwise be possible using the Kaplan-Meier method of calculation [11].
Boag's lognormal model for long-term cancer survival rates has been available for use for some 50 years. When the lognormal model was first proposed in the 1940s, it was difficult to implement because of a lack of computing power, and lack of good quality long-term follow-up data from cancer registries. Since 1970s the model has been used by authors on breast cancer, cervix uteri cancer, head and neck cancer, intraocular melanoma, choroidal-ciliary body melanoma, and small cell lung cancer [12-17]. Currently, although available computing power is adequate, good quality follow-up data on a sufficient number of patients are seldom available, and so can limit the application of Boag's model. Studies from single institutions tend to have small number of patients and limited years of follow-up for IBC. Use of a large data registry such as the SEER database with good long-term follow-up data can overcome these potential limitations.
Methods
From the population database of the Surveillance, Epidemiology, and End Results program of the National Cancer Institute from 1975–1995, data were extracted using SEER*Stat5.0 software from the 9 registries: San Francisco-Oakland, Connecticut, Metropolitan Detroit, Hawaii, Iowa, New Mexico, Seattle (Puget Sound), Utah, and Metropolitan Atlanta. This period of 21 years was divided into 7 periods of 3 years each. The years of diagnosis were 1975–77, 1978–80, 1981–83, 1984–86, 1987–89, 1990–92, and 1993–95. These years were chosen so as to provide adequate follow-up information to 2000. ICD-O-2 histology 8530/3 was used to define IBC. The data used in the study were survival time, vital status, and cause of death.
The cause-specific survival (CSS) was defined as the interval from the date of diagnosis to the date of death from breast cancer or to the last follow-up date for censoring purposes, if the patient was alive and was still being followed at the time of data cut-off.
The lognormal model was used for statistical analysis. Using short-term follow-up data, the lognormal model can predict long-term survival rates comparable in accuracy with those calculated by the Kaplan-Meier method using long-term follow-up [18]. The assumption of the lognormal model is that the survival times of the patients died of a specific cancer follows a lognormal distribution. What lognormal distribution means is that it becomes a normal distribution when the variables are converted by taking logarithmic transformation. The lognormal model has three parameters: the standard deviation S, the mean M and the proportion cured C. The proportion cured is defined as the portion of all the patients treated remaining alive and symptom free for a long period, some of those who died of intercurrent diseases are presumably cured of the cancer. This lognormal model used a maximum likelihood method to estimate long-term CSS (e.g., 10-year, 15-year and 20-year survival rates) from only short-term follow-up data. The CSS rates at time τ is calculated as [C+ (1-C)·Q]·100%, where C is the proportion cured of patients and Q is the integral of the lognormal distribution between the limits of time τ and infinity.
The long-term survival rates were predicted by Boag's method using a computer program run by Microsoft Excel. In this parametric lognormal model, the standard deviation S was fixed; only the two remaining parameters, mean M and proportion cured C, were kept floating when using the maximum likelihood method. A range (0.35–0.55) of S with step 0.01 was tested. The value of S was chosen for the best fit to the first five years known survival curve obtained by the Kaplan-Meier method, and also multiple iterations converged to a stable solution for M and C. The parameters obtained are shown in Table 1.
A 3-year period of diagnosis was selected and patients were followed as a cohort for an additional 3 years. For example, for cases diagnosed during the 3-year period, 1975–1977, prediction of the long-term survival rates was made using follow-up data to December 31, 1980 (i.e., 3 years after 1977). The predicted long-term survival rates for patients diagnosed during 1975–1977, and 1978–1980 were compared to the Kaplan-Meier estimates.
Confidence intervals are calculated by +/- 1.96 (standard error), assuming that the errors are normally distributed.
Results
A total of 1684 patients were extracted from the SEER database: 84% were white, 11% were African-Americans, and 5% belonged to other races. Age distribution was < 30 years in 1%, 30–40 in 11%, 40–50 in 22%, 50–60 in 24%, 60–70 in 21%, and > 70 in 21%. Table 2 shows the patient characteristics of the 7 periods in the study.
The proportions cured as shown in Table 1 for the different periods are almost linearly increasing (correlation coefficient of determination, R2 = 0.93) across the years. The number of breast cancer deaths and the number of total deaths for the 7 periods in the study are shown in Table 3 at different time of follow-up.
The 5-, 10-, 15-, and 20- year CSS by period of diagnosis, estimates by the lognormal model and the non-parametric Kaplan-Meier method if applicable are shown in Table 4. The standard errors were less than 4%. For patients diagnosed in 1975–77, the 5-year CSS was 18% and 20-year CSS was 9%. In the modern era, 1993–95, the 5-year CSS increased to 29% and 20-year CSS was estimated to be 20%. Table 5 shows the short-term CSS comparison obtained by the Kaplan-Meier method and the lognormal model. Figures 1, 2, 3, 4, 5, 6, 7 show both lognormal model estimations and Kaplan-Meier curves for the different periods.
Discussion
IBC and its outcome after treatment
IBC is a distinct entity different from the usual locally advanced breast cancer. Chang et al. [19] studied the incidence of IBC in the SEER database. IBC patients were significantly younger at diagnosis than non-IBC patients. Among both IBC and non-IBC patients, African Americans were younger than whites. Overall survivals (OS) were significantly worse for IBC patients than for non-IBC patients and for African Americans than for whites. Among whites, the 3-year survival improved more for IBC patients than for non-IBC patients between 1975–1979 and 1988–1992, increasing from 32% to 42% for IBC patients (P = 0.0001) and from 80% to 85% for non-IBC patients (P = 0.0001).
Low et al. [20] compared IBC versus non-IBC among patients in the National Cancer Institute. The 46 IBC patients had a median overall survival of 3.8 years and event free survival of 2.3 years, compared with 12.2 and 9.0 years, respectively, in stage IIIA breast cancer patients. Fifteen-year OS survival was 20% for IBC versus 50% for stage IIIA patients and 23% for stage IIIB non-IBC patients.
Table 6[3,6,21-28] summarizes the key results in research for IBC. Chang et al. [22] evaluated the effects of obesity and menopausal status on survival in a cohort of 177 female IBC patients diagnosed between 1974 and 1993. They found that factors associated with larger body size at diagnosis may contribute to shorter survival among postmenopausal IBC women but not among pre-menopausal IBC women. The latter were found to have poor survival regardless of body size.
In the table of patient characteristics (Table 2) of the different periods, the age distributions are similar for the different periods. However, the earlier 3 periods have similar stage distribution, and the later 4 periods have another similar stage distribution, with more distant stage patients compared with the earlier 3 periods. Despite an increasing proportion of distant stage, survival is increasing. This likely reflects the vigilance of staging investigations in recent time periods.
The present study shows a gradual increase in CSS rate over time, for both the Kaplan-Meier method and the lognormal model. The estimations of the long-term CSS rates by the lognormal model for 1975–77 and 1978–80 were validated within one standard error of those rates calculated by using Kaplan-Meier method. The above results show what is achievable in different institutions over a wide area of United States. Published single institution studies from major or tertiary referral centers do not reflect the true picture of care in the community. The long-term CSS calculated by the lognormal model for the cohort diagnosed in the years 1993–1995 is stable at 20%. There is little further drop of CSS beyond 10 years. The improvement in survival has been achieved by years of research, as noted below.
Perez et al. [23] analyzed 179 patients with histologically confirmed inflammatory carcinoma of the breast. Minimum follow-up was 2 years (maximum, 12 years; median, 4 years in the surviving patients). Clearly better loco-regional tumor control, i.e. in the breast and regional nodal drainage area, was observed in patients who underwent a surgical procedure: 79% with three modalities, 76% with irradiation and surgery, and only 30% with irradiation alone or in combination with chemotherapy. The addition of mastectomy to irradiation significantly improved loco-regional tumor control, disease free survival (DFS), and CSS. The combination of chemotherapy, surgery, and irradiation had a significant impact on loco-regional tumor control and incidence of distant metastases compared with surgery plus irradiation, and a lesser impact on DFS and CSS.
The literature indicates that chemotherapy does not negate the importance of radiation in optimizing loco-regional control in patients with high-risk breast cancer. The results of recent randomized trials studying postmastectomy radiation show that improved loco-regional control improves OS. Thus many authors believe that all breast cancer patients who have high-risk primary breast cancer and who are treated with chemotherapy should receive radiation as a component of their treatment [26]. Liao et al. [27] studied 115 patients with nonmetastatic IBC and tested the use of twice a day (b.i.d.) radiotherapy treatment with total dose of 60 Gy and 66 Gy at different time periods. Accelerated, b.i.d. fractionation at 1.5 Gy per fraction to a dose of 45 Gy plus 15 Gy chest wall boost was used for most patients from 1982 to 1985. From 1986–1993, 51 Gy in 34 fractions was delivered over 17 treatment days, followed by a 15 Gy chest wall boost in 10 fractions over a period of 5 treatment days. Chemotherapy regimens used did not change significantly during the period of that study. Long-term complications of radiation, such as arm edema of more than 3 cm (in 7 patients), rib fracture (in 10 patients), severe chest wall fibrosis (in 4 patients), and symptomatic pneumonitis (in 5 patients), were comparable in the two groups of 60 Gy versus 66 Gy, indicating that the dose escalation did not result in increased morbidity. Significant differences in the rates of loco-regional control (P = 0.03) and OS (P = 0.03), and a trend towards better DFS (P = 0.06) were observed among those recently treated patients who received higher doses of irradiation. For the entire patient group who received radiotherapy either once or twice daily, the 5- and 10-year local control rates were 73.2% and 67.1%, respectively. The 5- and 10-year DFS were 32.0% and 28.8%, respectively, and the overall survival rates for the entire group were 40.5% and 31.3%, respectively.
In France, a study on the impact of intensity of chemotherapy was performed [28] on 74 women with nonmetastatic IBC consecutively treated between 1976 and 2000. Patients received primary anthracycline-based chemotherapy either at conventional doses (n = 20) or at high does with hematopoietic stem cell support (HSCS) (n = 54). In multivariate analysis, the strongest independent prognostic factor was the delivery of high-dose chemotherapy (HDC). The 5-year DFS and OS of patients were respectively 28% and 50% with HDC and 15% and 18% with conventional chemotherapy. These results suggest that HDC with HSCS may have a role in the treatment of IBC.
Recent study on the use of trastuzumab and paclitaxel may well lead to further research on the use of different combinations of chemotherapy and biological response modifiers [29] for the treatment of IBC. After completion of chemotherapy, for patients whose tumors showed receptor-positive tumors; additional tamoxifen therapy (if post-menopausal) or gonadotropin-releasing hormone (GnRH)-analogues (if pre-menopausal) were given. However, it is still uncertain, whether better prognosis can be achieved by treatment with GnRH-analogues [30].
Ueno et al. [6] found from the long-term follow-up data on patients treated with a combined-modality (chemotherapy, then mastectomy, then chemotherapy and radiotherapy) approach, a significant fraction of patients (estimated to be 28%) remained free of disease beyond 15 years. There were virtually no recurrences after 10 years. Estimated DFS at 5 years was 32%, at 10 years was 28%, and at 15 years was 28%. Estimated OS at 5 years was 40%, at 10 years was 33%, and at 15 years was 29%. By contrast, single-modality treatment (radiotherapy or surgery alone) gave a DFS of less than 5% beyond 15 years. Thus, combined-modality treatment is recommended as the standard of care for IBC.
Possible explanations to the improved survival
IBC is a distinct clinicopathologic entity separate from noninflammatory locally advanced breast carcinoma [31,32]. Improvements in population-based survival represent the extent to which therapies with demonstrated efficacy are translated to the real population. Thus, they represent the effect of dissemination of new therapies and effectiveness. In the early 1970s, the commonly used regimen was FAC (5-fluorouracil, adriamycin, cyclophosphamide) before and after radiotherapy. In the late 1970s, FAC was given before and after mastectomy [6]. Taxanes become increasingly used in America since 2001 [33].
There are several other possible explanations to the improved survival other than treatment changes: change of the definition and classification of IBC, proportion of cause specific deaths not based on autopsy, change of patient population (age distribution, stages [IIIB and IV]). Obesity is a poor prognostic factor and so the improvement of IBC survival is not related to increasing obesity noted in the American population. Better treatment and the above factors all account for the improved survival.
Conclusion
The improvement of survival during the study period may be due to introduction of more aggressive treatments. However, there seem to be no further increase of long-term CSS, which should encourage oncologists to find even more effective treatments. Because of small numbers of patients, randomized studies will be difficult to conduct. The SEER population-based database will yield the best possible estimate of the trend in improvement of survival for patients with IBC.
List of abbreviations
C: Proportion cured
CSS: Cause-specific survival
DFS: Disease-free survival
IBC: Inflammatory breast cancer
M: Mean
OS: Overall survival
S: Standard deviation
SEER: Surveillance, Epidemiology, and End Results
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PT: Data analysis and writing of the manuscript.
EY, RS, JP, KJ, ES, SM: Critical appraisal of 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
Saskatchewan Cancer Agency Research Grant Award 2792.
Figures and Tables
Figure 1 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1975–1977.
Figure 2 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1978–1980.
Figure 3 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1981–1983.
Figure 4 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1984–1986.
Figure 5 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1987–1989.
Figure 6 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1990–1992.
Figure 7 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1993–1995.
Table 1 Parameters in the lognormal model estimations with standard errors in brackets.
Years of diagnosis Selected S Estimated M Estimated C
1975–77 0.50 1.25(0.06) 0.07(0.06)
1978–80 0.48 1.32(0.06) 0.10(0.06)
1981–83 0.48 1.37(0.06) 0.13(0.06)
1984–86 0.43 1.30(0.04) 0.15(0.05)
1987–89 0.40 1.33(0.04) 0.16(0.04)
1990–92 0.47 1.36(0.04) 0.16(0.04)
1993–95 0.39 1.33(0.03) 0.19(0.03)
S = standard deviation,
M = mean,
C = proportion cured.
Table 2 Patient characteristics.
Years of diagnosis 1975–77 1978–80 1981–83 1984–86 1987–89 1990–92 1993–95
No. of patients 134 145 181 210 258 340 416
Race:
White 117 124 162 184 207 276 352
Black 16 15 9 19 37 49 40
Others 1 6 8 7 13 15 24
Unknown 0 0 2 0 1 0 0
Age:
Median 58 56 60 56 55 59 56
Range 26–97 29–91 22–89 25–91 25–101 34–94 26–97
Stage:
Localized 6 3 6 1 1 5 5
Regional 84 100 101 41 27 20 21
Distant 35 37 72 161 218 304 385
Unstaged 9 5 2 7 12 11 5
Table 3 Number of deaths due to breast cancer and number of total deaths at six-month intervals for the 7 periods of diagnosis.
1975–77 1978–80 1981–83 1984–86 1987–89 1990–92 1993–95
Interval BCD TD BCD TD BCD TD BCD TD BCD TD BCD TD BCD TD
0–6 25 26 15 19 19 22 21 26 23 28 29 36 35 47
6–12 15 20 15 18 15 22 27 32 27 29 47 51 49 52
12–18 12 16 26 28 24 26 31 31 37 40 41 46 44 49
18–24 13 15 10 12 14 14 15 17 27 30 31 36 38 38
24–30 13 13 11 11 11 13 15 16 14 15 23 26 33 34
30–36 6 6 7 8 16 16 15 16 18 20 11 15 24 30
36–42 4 6 8 9 7 9 9 11 15 15 11 12 11 13
42–48 1 1 4 5 7 9 5 6 9 13 8 9 15 17
48–54 7 7 4 4 3 3 3 4 5 5 7 9 15 16
54–60 4 6 2 3 5 5 3 4 2 2 6 7 11 12
60–66 1 1 3 3 2 3 3 3 4 4 5 6 4 5
66–72 1 1 1 1 1 2 4 4 4 4 11 12 4 6
72–78 1 1 1 3 3 3 2 2 3 4 6 7 1 5
78–84 0 0 0 2 1 2 3 3 0 1 0 1 2 2
84–90 1 1 0 1 1 3 2 2 1 1 0 0 2 2
90–96 1 1 1 1 1 2 1 1 3 5 1 5 0 0
96–102 0 0 0 0 0 0 1 1 2 3 2 5
102–108 1 1 1 1 1 1 0 1 0 0 1 1
108–114 1 2 1 1 1 1 0 0 1 2 1 1
114–120 0 0 0 0 1 1 1 2 0 0 0 1
120–126 0 0 0 0 1 1 0 0 0 0 0 0
126–132 0 0 0 0 0 0 0 0 0 1 0 0
132–138 0 1 1 1 0 0 1 2 1 2
138–144 0 0 0 1 0 0 1 2 0 1
144–150 0 0 0 0 0 1 0 0 0 2
150–156 0 0 0 1 0 1 0 1 0 1
156–162 1 1 1 1 1 3 0 0 0 0
162–168 0 0 1 1 0 2 1 1
168–174 0 0 0 0 1 2 0 1
174–180 0 0 1 1 1 1 0 2
180–186 1 1 0 0 0 0 0 1
186–192 0 0 0 0 0 0 0 0
192–198 0 0 0 0 0 0 0 0
198–204 0 0 0 1 0 1 0 0
204–210 0 0 0 0 0 0
210–216 0 0 0 0 0 0
216–222 0 0 0 0 0 0
222–228 0 0 0 0 0 0
228–234 0 0 0 0 0 0
234–240 0 2 0 0
240–246 0 0 0 2
246–252 0 0 0 1
252–258 0 0 0 0
258–264 0 0 0 1
264–270 0 0 0 0
270–276 0 0 0 0
276–282 0 1
282–288 0 0
288–294 0 0
294–300 0 0
300–306 0 0
306–312 0 0
BCD = breast cancer deaths
TD = total deaths
Table 4 5-, 10-, 15- and 20-year cause-specific survival (CSS) rates, in percentages, estimated by Kaplan-Meier method compared with those estimated by lognormal model [in square brackets] with 95% confidence intervals (in round brackets).
Years of diagnosis 5 yr CSS 10 yr CSS 15 yr CSS 20 yr CSS
1975–77 18(16.5–19.4), [20](18.4–21.5) 11(10.3–11.6), [11](10.3–11.6) 9(8.4–9.5), [9](8.4–9.5) 9(8.4–9.5), [8](7.5–8.4)
1978–80 23(21.1–24.8), [24](22.1–25.8) 16(15.0–16.9), [14](13.1–14.8) 11(10.3–11.6), [12](11.2–12.7) 11(10.3–11.6), [11](10.3–11.6)
1981–83 27(24.8–29.1), [30](27.6–32.2) 18(16.9–19.0), [19](17.8–20.1) 14(13.1–14.8), [15](14.1–15.8) NA, [14](13.1–14.8)
1984–86 27(25.4–28.5), [26](24.4–27.5) 17(16.0–17.9), [18](16.9–19.0) 15(14.1–15.8), [16](15.0–16.9) NA, [15](14.1–15.8)
1987–89 27(25.4–28.5), [27](25.4–28.5) 19(17.8–20.1), [18](16.9–19.0) NA, [16](15.0–16.9) NA, [16](15.0–16.9)
1990–92 32(30.1–33.8), [32](30.1–33.8) 21(19.7–22.2), [22](20.7–23.2) NA, [19](17.8–20.1) NA, [18](16.9–19.0)
1993–95 29(27.8–30.1), [29](27.8–30.1) NA, [21](19.7–22.2) NA, [20](18.8–21.1) NA, [20](18.8–21.1)
NA = not available by Kaplan-Meier method
Table 5 0.5-, 1-, 2-, 3- and 4-year cause-specific survival (CSS) rates, in percentages, estimated by Kaplan-Meier method compared with those estimated by lognormal model [in square brackets] with 95% confidence intervals (in round brackets).
Years of diagnosis 0.5 yr CSS 1 yr CSS 2 yr CSS 3 yr CSS 4 yr CSS
1975–77 81(76.2–85.7), [84](79.0–88.9) 70(64.5–75.4), [66](60.8–71.1) 49(45.1–52.8), [44](40.5–47.4) 33(30.4–35.5), [32](29.4–34.5) 28(25.8–30.1), [25](23.0–26.9)
1978–80 89(83.7–94.2), [89](83.7–94.2) 79(74.3–83.6), [73](68.7–77.2) 52(47.9–56.0), [51](47.0–54.9) 38(35.0–40.9), [38](35.0–40.9) 28(25.8–30.1), [30](27.6–32.3)
1981–83 89(85.5–92.4), [90](86.4–93.5) 81(76.2–85.7), [76](71.5–80.4) 58(53.4–62.5), [55](50.6–59.3) 41(37.7–44.2), [43](39.6–46.3) 32(29.4–34.5), [35](32.2–37.7)
1984–86 90(86.4–93.5), [90](86.4–93.5) 77(72.4–81.5), [74](69.6–78.3) 53(48.8–57.1), [51](47.0–54.9) 38(35.7–40.2), [39](36.7–41.2) 30(28.2–31.7), [31](29.1–32.8)
1987–89 91(87.4–94.5), [91](87.4–94.5) 80(75.2–84.7), [77](72.4–81.5) 54(50.8–57.1), [53](49.8–56.1) 41(38.5–43.4), [40](37.6–42.3) 31(29.1–32.8), [32](30.1–33.8)
1990–92 91(87.4–94.5), [90](86.4–93.5) 77(73.9–80.0), [77](73.9–80.0) 54(50.8–57.1), [57](53.6–60.3) 43(40.4–45.5), [45](42.3–47.6) 37(34.8–39.1), [37](34.8–39.1)
1993–95 91(89.2–92.7), [92](90.1–93.8) 79(75.9–82.0), [79](75.9–82.0) 58(55.7–60.2), [56](53.8–58.1) 44(41.4–46.5), [42](39.5–44.4) 36(34.5–37.4), [34](32.6–35.3)
Table 6 Summary of key findings in research for inflammatory breast cancer.
Author Ref. Institute Year Patient No. Main results
Palangie [21] Institut Curie, France 1977–1987 223 Better DFS – rapid and complete remission with induction treatment
Chang [22] MD Anderson, US 1974–1993 177 Shorter OS – for postmenopausal IBC women with larger body size
Liauw [3] U. Florida, US 1982–2001 61 Improved CSS – tumor size <4 cm, upfront surgery, local disease control
Perez [23] Mallinckrodt, US 1958–1989 179 Improved DFS, CSS, LR – addition of mastectomy
Panades [24] British Columbia, Canada 1980–2000 485 Improved LC – mastectomy in addition to chemotherapy and radiotherapy
Improved CSS – modern chemotherapy regimens
Fleming [25] MD Anderson, US 1974–1993 178 Improved LC – mastectomy in addition to radiotherapy 16.3% vs 35.7%, P = 0.015
Buchholz [26] MD Anderson, US 1989–1997 12 Chemotherapy with radiation improved OS
Liao [27] MD Anderson, US 1977–1993 115 Improved LC, DFS, OS – twice-daily postmastectomy radiation to 66 Gy
Bertucci [28] Institut Paoli-Calmettes, France 1976–2000 74 High-dose chemotherapy: 5y DFS 28%, OS 50% Conventional dose chemotherapy: 5y DFS 15%, OS 18%
Ueno [6] MD Anderson, US 1973–1993 178 Improved DFS – combined modality: 15y DFS 28%, single-modality: 15y DFS <5%
DFS = disease free survival,
OS = overall survival,
CSS = cause-specific survival,
LC = local recurrence.
==== Refs
Wingo PA Jamison PM Young JL Gargiullo P Population-based statistics for women diagnosed with inflammatory breast cancer (United States) Cancer Causes Control 2004 15 321 328 15090727 10.1023/B:CACO.0000024222.61114.18
Gradishar WJ Inflammatory breast cancer: the evolution of multimodality treatment strategies Semin Surg Oncol 1996 12 352 363 8873324 10.1002/(SICI)1098-2388(199609/10)12:5<352::AID-SSU11>3.0.CO;2-R
Liauw SL Benda RK Morris CG Mendenhall NP Inflammatory breast carcinoma: outcomes with trimodality therapy for nonmetastatic disease Cancer 2004 100 920 928 14983486 10.1002/cncr.20083
Esteva FJ Hortobagyi GN Adjuvant systemic therapy for primary breast cancer Surg Clin North Am 1999 79 1075 1090 10572552 10.1016/S0039-6109(05)70062-1
Harris EE Schultz D Bertsch H Fox K Glick J Solin LJ Ten-year outcome after combined modality therapy for inflammatory breast cancer Int J Radiat Oncol Biol Phys 2003 55 1200 1208 12654428 10.1016/S0360-3016(02)04201-3
Ueno NT Buzdar AU Singletary SE Ames FC McNeese MD Holmes FA Theriault RL Strom EA Wasaff BJ Asmar L Frye D Hortobagyi GN Combined-modality treatment of inflammatory breast carcinoma: twenty years of experience at M. D. Anderson Cancer Center Cancer Chemother Pharmacol 1997 40 321 329 9225950 10.1007/s002800050664
Surveillance, Epidemiology, and End Results Program Public Use Data (1973–2000) National Cancer Institute, released April 2003 Bethesda: National Cancer Institute
Boag JW The presentation and analysis of the results of radiotherapy. Part I. Introduction Br J Radiol 1948 21 128 138
Boag JW The presentation and analysis of the results of radiotherapy. Part II. Mathematical theory Br J Radiol 1948 21 189 203
Boag JW Maximum likelihood estimates of the proportion of patients cured by cancer therapy J R Stat Soc (Ser B) 1949 11 15 53
Kaplan EL Meier P Nonparametric estimation from incomplete observations J Am Stat Assoc 1958 53 457 481
Rutqvist LE On the utility of the lognormal model for analysis of breast cancer survival in Sweden 1961–1973 Br J Cancer 1985 52 875 883 4074639
Mould RF Boag JW A test of several parametric statistical models for estimating success rate in the treatment of carcinoma cervix uteri Br J Cancer 1975 32 529 550 1212415
Mould RF Hearnden T Palmer M White GC Distribution of survival times of 12,000 head and neck cancer patients who died with their disease Br J Cancer 1976 34 180 190 962995
Gamel JW Greenberg RA McLean IW A stable linear algorithm for fitting the lognormal model to survival data Comput Biomed Res 1988 21 38 47 3345651 10.1016/0010-4809(88)90040-7
Gamel JW George SL Edwards MJ Seigler HF The long-term clinical course of patients with cutaneous melanoma Cancer 2002 95 1286 1293 12216097 10.1002/cncr.10813
Tai P Tonita J Yu E Skarsgard D A 20-year follow-up study of the long-term survival of limited stage small cell lung cancer and an overview of prognostic and treatment factors Int J Radiat Oncol Biol Phys 2003 56 626 633 12788167 10.1016/S0360-3016(03)00070-1
Mould RF Lederman M Tai P Wong JK Methodology to predict long-term cancer survival from short-term data using Tobacco Cancer Risk and Absolute Cancer Cure models Phys Med Biol 2002 47 3893 3924 12476973 10.1088/0031-9155/47/22/301
Chang S Parker SL Pham T Buzdar AU Hursting SD Inflammatory breast carcinoma incidence and survival: the surveillance, epidemiology, and end results program of the National Cancer Institute, 1975–1992 Cancer 1998 82 2366 2372 9635529 10.1002/(SICI)1097-0142(19980615)82:12<2366::AID-CNCR10>3.0.CO;2-N
Low JA Berman AW Steinberg SM Danforth DN Lippman ME Swain SM Long-term follow-up for locally advanced and inflammatory breast cancer patients treated with multimodality therapy J Clin Oncol 2004 22 4067 4074 15483018 10.1200/JCO.2004.04.068
Palangie T Mosseri V Mihura J Campana F Beuzeboc P Dorval T Garcia-Giralt E Jouve M Scholl S Asselain B Pouillart P Prognostic factors in inflammatory breast cancer and therapeutic implications Eur J Cancer 1994 30 921 927 10.1016/0959-8049(94)90115-5
Chang S Alderfer JR Asmar L Buzdar AU Inflammatory breast cancer survival: the role of obesity and menopausal status at diagnosis Breast Cancer Res Treat 2000 64 157 163 11194451 10.1023/A:1006489100283
Perez CA Fields JN Fracasso PM Philpott G Soares RL JrTaylor ME Lockett MA Rush C Management of locally advanced carcinoma of the breast. II. Inflammatory carcinoma Cancer 1994 74 466 476 8004622
Panades M Olivotto IA Speers CH Shenkier T Olivotto TA Weir L Allan SJ Truong PT Evolving treatment strategies for inflammatory breast cancer: a population-based survival analysis J Clin Oncol 2005 23 1941 1950 15774787 10.1200/JCO.2005.06.233
Fleming RYD Asmar L Buzdar AU McNeese MD Ames FC Ross MI Singletary SE Effectiveness of mastectomy by response to induction chemotherapy for control in inflammatory breast carcinoma Ann Surg Oncol 1997 4 452 461 9309333
Buchholz TA Tucker SL Moore RA McNeese MD Strom EA Jhingrin A Hortobagyi GN Singletary SE Champlin RE Importance of radiation therapy for breast cancer patients treated with high-dose chemotherapy and stem cell transplant Int J Radiat Oncol Biol Phys 2000 46 337 343 10661340 10.1016/S0360-3016(99)00429-0
Liao Z Strom EA Buzdar AU Singletary SE Hunt K Allen PK McNeese MD Locoregional irradiation for inflammatory breast cancer: effectiveness of dose escalation in decreasing recurrence Int J Radiat Oncol Biol Phys 2000 47 1191 1200 10889372 10.1016/S0360-3016(00)00561-7
Bertucci F Tarpin C Charafe-Jauffret E Bardou VJ Braud AC Tallet A Gravis G Viret F Goncalves A Houvenaeghel G Blaise D Jacquemier J Maraninchi D Viens P Multivariate analysis of survival in inflammatory breast cancer: impact of intensity of chemotherapy in multimodality treatment Bone Marrow Transplant 2004 33 913 920 15004544 10.1038/sj.bmt.1704458
Okawa Y Sugiyama K Aiba K Hirano A Uno S Hagino T Kawase K Shioya H Yoshida K Usui N Kobayashi M Kobayashi T Successful combination therapy with trastuzumab and Paclitaxel for adriamycin-and docetaxel-resistant inflammatory breast cancer Breast Cancer 2004 11 309 312 15550852
Volz J Volz E Bohm W Schneider V Inflammatory cancer of the breast. Therapy and follow-up Geburtshilfe Frauenheilkd 1992 52 206 209 1607113
Hance KW Anderson WF Devesa SS Young HA Levine PH Trends in inflammatory breast carcinoma incidence and survival: the surveillance, epidemiology, and end results program at the National Cancer Institute J Natl Cancer Inst 2005 97 966 975 15998949
Anderson WF Chu KC Chang S Inflammatory breast carcinoma and noninflammatory locally advanced breast carcinoma: distinct clinicopathologic entities? J Clin Oncol 2003 21 2254 2259 12805323 10.1200/JCO.2003.07.082
Somlo G Doroshow JH Synold T High-dose paclitaxel in combination with doxorubicin, cyclophosphamide and peripheral blood progenitor cell rescue in patients with high-risk primary and responding metastatic breast carcinoma: Toxicity profile, relationship to paclitaxel pharmacokinetics and short-term outcome Br J Cancer 2001 84 1591 1598 11401310 10.1054/bjoc.2001.1835
|
16242046
|
PMC1283744
|
CC BY
|
2021-01-04 16:03:08
|
no
|
BMC Cancer. 2005 Oct 22; 5:137
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-137
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1371624204610.1186/1471-2407-5-137Research ArticleShort- and long-term cause-specific survival of patients with inflammatory breast cancer Tai Patricia [email protected] Edward [email protected] Ross [email protected] Juan [email protected] Kurian [email protected] Evgeny [email protected] Shazia [email protected] University of Saskatchewan, Faculty of Medicine, Saskatoon; Department of Radiation Oncology, Regina, Canada2 Division of Radiation Oncology, Department of Oncology, University of Western Ontario, London, Ontario, Canada2005 22 10 2005 5 137 137 3 5 2005 22 10 2005 Copyright © 2005 Tai 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
Inflammatory breast cancer (IBC) had been perceived to have a poor prognosis. Oncologists were not enthusiastic in the past to give aggressive treatment. Single institution studies tend to have small patient numbers and limited years of follow-up. Most studies do not report 10-, 15- or 20-year results.
Methods
Data was obtained from the population-based database of the Surveillance, Epidemiology, and End Results program of the National Cancer Institute from 1975–1995 using SEER*Stat5.0 software. This period of 21 years was divided into 7 periods of 3 years each. The years were chosen so that there was adequate follow-up information to 2000. ICD-O-2 histology 8530/3 was used to define IBC. The lognormal model was used for statistical analysis.
Results
A total of 1684 patients were analyzed, of which 84% were white, 11% were African Americans, and 5% belonged to other races. Age distribution was < 30 years in 1%, 30–40 in 11%, 40–50 in 22%, 50–60 in 24%, 60–70 in 21%, and > 70 in 21%. The lognormal model was validated for 1975–77 and for 1978–80, since the 10-, 15- and 20-year cause-specific survival (CSS) rates, could be calculated using the Kaplan-Meier method with data available in 2000. The data were then used to estimate the 10-, 15- and 20-year CSS rates for the more recent years, and to study the trend of improvement in survival. There were increasing incidences of IBC: 134 patients in the 1975–77 period to 416 patients in the 1993–95 period. The corresponding 20-year CSS increased from 9% to 20% respectively with standard errors of less than 4%.
Conclusion
The improvement of survival during the study period may be due to introduction of more aggressive treatments. However, there seem to be no further increase of long-term CSS, which should encourage oncologists to find even more effective treatments. Because of small numbers of patients, randomized studies will be difficult to conduct. The SEER population-based database will yield the best possible estimate of the trend in improvement of survival for patients with IBC.
==== Body
Background
Inflammatory breast cancer (IBC) occurs rarely [1]. Signs and symptoms of this condition include the presence of erythema, edema or peau d'orange appearance of the skin, and other clinical signs of disease. Diagnosis is made by skin biopsy. The definition of IBC varies in the literature and leads to some disparities. In this study, the pathological definition is used.
It is known that IBC have a poor prognosis. Oncologists were not enthusiastic to administer aggressive treatment in the past. Nowadays, treatment for this aggressive form of breast cancer is multi-modal, and includes chemotherapy, surgery, radiation therapy, and hormonal therapy [2]. The optimal sequence of the different modalities is still a subject of research [3]. Development of novel therapeutic agents continues and is based on an expanding understanding of the biology of tumor development and progression. Advances in treatment continue to improve the prognosis for this disease [4]. With a few notable exceptions, many publications on IBC do not have long periods of follow-up [5,6]. These single-institution studies are from academic centers. To our knowledge, long-term results of cases treated in the community are not available. This study examines the changes in the prognosis of IBC over the years with the Surveillance, Epidemiology, and End Results (SEER) database [7].
There is a parametric lognormal model, proposed by Boag [8-10] that has been validated retrospectively in the literature, and can be used prospectively for predicting long-term survival rates several years earlier than would otherwise be possible using the Kaplan-Meier method of calculation [11].
Boag's lognormal model for long-term cancer survival rates has been available for use for some 50 years. When the lognormal model was first proposed in the 1940s, it was difficult to implement because of a lack of computing power, and lack of good quality long-term follow-up data from cancer registries. Since 1970s the model has been used by authors on breast cancer, cervix uteri cancer, head and neck cancer, intraocular melanoma, choroidal-ciliary body melanoma, and small cell lung cancer [12-17]. Currently, although available computing power is adequate, good quality follow-up data on a sufficient number of patients are seldom available, and so can limit the application of Boag's model. Studies from single institutions tend to have small number of patients and limited years of follow-up for IBC. Use of a large data registry such as the SEER database with good long-term follow-up data can overcome these potential limitations.
Methods
From the population database of the Surveillance, Epidemiology, and End Results program of the National Cancer Institute from 1975–1995, data were extracted using SEER*Stat5.0 software from the 9 registries: San Francisco-Oakland, Connecticut, Metropolitan Detroit, Hawaii, Iowa, New Mexico, Seattle (Puget Sound), Utah, and Metropolitan Atlanta. This period of 21 years was divided into 7 periods of 3 years each. The years of diagnosis were 1975–77, 1978–80, 1981–83, 1984–86, 1987–89, 1990–92, and 1993–95. These years were chosen so as to provide adequate follow-up information to 2000. ICD-O-2 histology 8530/3 was used to define IBC. The data used in the study were survival time, vital status, and cause of death.
The cause-specific survival (CSS) was defined as the interval from the date of diagnosis to the date of death from breast cancer or to the last follow-up date for censoring purposes, if the patient was alive and was still being followed at the time of data cut-off.
The lognormal model was used for statistical analysis. Using short-term follow-up data, the lognormal model can predict long-term survival rates comparable in accuracy with those calculated by the Kaplan-Meier method using long-term follow-up [18]. The assumption of the lognormal model is that the survival times of the patients died of a specific cancer follows a lognormal distribution. What lognormal distribution means is that it becomes a normal distribution when the variables are converted by taking logarithmic transformation. The lognormal model has three parameters: the standard deviation S, the mean M and the proportion cured C. The proportion cured is defined as the portion of all the patients treated remaining alive and symptom free for a long period, some of those who died of intercurrent diseases are presumably cured of the cancer. This lognormal model used a maximum likelihood method to estimate long-term CSS (e.g., 10-year, 15-year and 20-year survival rates) from only short-term follow-up data. The CSS rates at time τ is calculated as [C+ (1-C)·Q]·100%, where C is the proportion cured of patients and Q is the integral of the lognormal distribution between the limits of time τ and infinity.
The long-term survival rates were predicted by Boag's method using a computer program run by Microsoft Excel. In this parametric lognormal model, the standard deviation S was fixed; only the two remaining parameters, mean M and proportion cured C, were kept floating when using the maximum likelihood method. A range (0.35–0.55) of S with step 0.01 was tested. The value of S was chosen for the best fit to the first five years known survival curve obtained by the Kaplan-Meier method, and also multiple iterations converged to a stable solution for M and C. The parameters obtained are shown in Table 1.
A 3-year period of diagnosis was selected and patients were followed as a cohort for an additional 3 years. For example, for cases diagnosed during the 3-year period, 1975–1977, prediction of the long-term survival rates was made using follow-up data to December 31, 1980 (i.e., 3 years after 1977). The predicted long-term survival rates for patients diagnosed during 1975–1977, and 1978–1980 were compared to the Kaplan-Meier estimates.
Confidence intervals are calculated by +/- 1.96 (standard error), assuming that the errors are normally distributed.
Results
A total of 1684 patients were extracted from the SEER database: 84% were white, 11% were African-Americans, and 5% belonged to other races. Age distribution was < 30 years in 1%, 30–40 in 11%, 40–50 in 22%, 50–60 in 24%, 60–70 in 21%, and > 70 in 21%. Table 2 shows the patient characteristics of the 7 periods in the study.
The proportions cured as shown in Table 1 for the different periods are almost linearly increasing (correlation coefficient of determination, R2 = 0.93) across the years. The number of breast cancer deaths and the number of total deaths for the 7 periods in the study are shown in Table 3 at different time of follow-up.
The 5-, 10-, 15-, and 20- year CSS by period of diagnosis, estimates by the lognormal model and the non-parametric Kaplan-Meier method if applicable are shown in Table 4. The standard errors were less than 4%. For patients diagnosed in 1975–77, the 5-year CSS was 18% and 20-year CSS was 9%. In the modern era, 1993–95, the 5-year CSS increased to 29% and 20-year CSS was estimated to be 20%. Table 5 shows the short-term CSS comparison obtained by the Kaplan-Meier method and the lognormal model. Figures 1, 2, 3, 4, 5, 6, 7 show both lognormal model estimations and Kaplan-Meier curves for the different periods.
Discussion
IBC and its outcome after treatment
IBC is a distinct entity different from the usual locally advanced breast cancer. Chang et al. [19] studied the incidence of IBC in the SEER database. IBC patients were significantly younger at diagnosis than non-IBC patients. Among both IBC and non-IBC patients, African Americans were younger than whites. Overall survivals (OS) were significantly worse for IBC patients than for non-IBC patients and for African Americans than for whites. Among whites, the 3-year survival improved more for IBC patients than for non-IBC patients between 1975–1979 and 1988–1992, increasing from 32% to 42% for IBC patients (P = 0.0001) and from 80% to 85% for non-IBC patients (P = 0.0001).
Low et al. [20] compared IBC versus non-IBC among patients in the National Cancer Institute. The 46 IBC patients had a median overall survival of 3.8 years and event free survival of 2.3 years, compared with 12.2 and 9.0 years, respectively, in stage IIIA breast cancer patients. Fifteen-year OS survival was 20% for IBC versus 50% for stage IIIA patients and 23% for stage IIIB non-IBC patients.
Table 6[3,6,21-28] summarizes the key results in research for IBC. Chang et al. [22] evaluated the effects of obesity and menopausal status on survival in a cohort of 177 female IBC patients diagnosed between 1974 and 1993. They found that factors associated with larger body size at diagnosis may contribute to shorter survival among postmenopausal IBC women but not among pre-menopausal IBC women. The latter were found to have poor survival regardless of body size.
In the table of patient characteristics (Table 2) of the different periods, the age distributions are similar for the different periods. However, the earlier 3 periods have similar stage distribution, and the later 4 periods have another similar stage distribution, with more distant stage patients compared with the earlier 3 periods. Despite an increasing proportion of distant stage, survival is increasing. This likely reflects the vigilance of staging investigations in recent time periods.
The present study shows a gradual increase in CSS rate over time, for both the Kaplan-Meier method and the lognormal model. The estimations of the long-term CSS rates by the lognormal model for 1975–77 and 1978–80 were validated within one standard error of those rates calculated by using Kaplan-Meier method. The above results show what is achievable in different institutions over a wide area of United States. Published single institution studies from major or tertiary referral centers do not reflect the true picture of care in the community. The long-term CSS calculated by the lognormal model for the cohort diagnosed in the years 1993–1995 is stable at 20%. There is little further drop of CSS beyond 10 years. The improvement in survival has been achieved by years of research, as noted below.
Perez et al. [23] analyzed 179 patients with histologically confirmed inflammatory carcinoma of the breast. Minimum follow-up was 2 years (maximum, 12 years; median, 4 years in the surviving patients). Clearly better loco-regional tumor control, i.e. in the breast and regional nodal drainage area, was observed in patients who underwent a surgical procedure: 79% with three modalities, 76% with irradiation and surgery, and only 30% with irradiation alone or in combination with chemotherapy. The addition of mastectomy to irradiation significantly improved loco-regional tumor control, disease free survival (DFS), and CSS. The combination of chemotherapy, surgery, and irradiation had a significant impact on loco-regional tumor control and incidence of distant metastases compared with surgery plus irradiation, and a lesser impact on DFS and CSS.
The literature indicates that chemotherapy does not negate the importance of radiation in optimizing loco-regional control in patients with high-risk breast cancer. The results of recent randomized trials studying postmastectomy radiation show that improved loco-regional control improves OS. Thus many authors believe that all breast cancer patients who have high-risk primary breast cancer and who are treated with chemotherapy should receive radiation as a component of their treatment [26]. Liao et al. [27] studied 115 patients with nonmetastatic IBC and tested the use of twice a day (b.i.d.) radiotherapy treatment with total dose of 60 Gy and 66 Gy at different time periods. Accelerated, b.i.d. fractionation at 1.5 Gy per fraction to a dose of 45 Gy plus 15 Gy chest wall boost was used for most patients from 1982 to 1985. From 1986–1993, 51 Gy in 34 fractions was delivered over 17 treatment days, followed by a 15 Gy chest wall boost in 10 fractions over a period of 5 treatment days. Chemotherapy regimens used did not change significantly during the period of that study. Long-term complications of radiation, such as arm edema of more than 3 cm (in 7 patients), rib fracture (in 10 patients), severe chest wall fibrosis (in 4 patients), and symptomatic pneumonitis (in 5 patients), were comparable in the two groups of 60 Gy versus 66 Gy, indicating that the dose escalation did not result in increased morbidity. Significant differences in the rates of loco-regional control (P = 0.03) and OS (P = 0.03), and a trend towards better DFS (P = 0.06) were observed among those recently treated patients who received higher doses of irradiation. For the entire patient group who received radiotherapy either once or twice daily, the 5- and 10-year local control rates were 73.2% and 67.1%, respectively. The 5- and 10-year DFS were 32.0% and 28.8%, respectively, and the overall survival rates for the entire group were 40.5% and 31.3%, respectively.
In France, a study on the impact of intensity of chemotherapy was performed [28] on 74 women with nonmetastatic IBC consecutively treated between 1976 and 2000. Patients received primary anthracycline-based chemotherapy either at conventional doses (n = 20) or at high does with hematopoietic stem cell support (HSCS) (n = 54). In multivariate analysis, the strongest independent prognostic factor was the delivery of high-dose chemotherapy (HDC). The 5-year DFS and OS of patients were respectively 28% and 50% with HDC and 15% and 18% with conventional chemotherapy. These results suggest that HDC with HSCS may have a role in the treatment of IBC.
Recent study on the use of trastuzumab and paclitaxel may well lead to further research on the use of different combinations of chemotherapy and biological response modifiers [29] for the treatment of IBC. After completion of chemotherapy, for patients whose tumors showed receptor-positive tumors; additional tamoxifen therapy (if post-menopausal) or gonadotropin-releasing hormone (GnRH)-analogues (if pre-menopausal) were given. However, it is still uncertain, whether better prognosis can be achieved by treatment with GnRH-analogues [30].
Ueno et al. [6] found from the long-term follow-up data on patients treated with a combined-modality (chemotherapy, then mastectomy, then chemotherapy and radiotherapy) approach, a significant fraction of patients (estimated to be 28%) remained free of disease beyond 15 years. There were virtually no recurrences after 10 years. Estimated DFS at 5 years was 32%, at 10 years was 28%, and at 15 years was 28%. Estimated OS at 5 years was 40%, at 10 years was 33%, and at 15 years was 29%. By contrast, single-modality treatment (radiotherapy or surgery alone) gave a DFS of less than 5% beyond 15 years. Thus, combined-modality treatment is recommended as the standard of care for IBC.
Possible explanations to the improved survival
IBC is a distinct clinicopathologic entity separate from noninflammatory locally advanced breast carcinoma [31,32]. Improvements in population-based survival represent the extent to which therapies with demonstrated efficacy are translated to the real population. Thus, they represent the effect of dissemination of new therapies and effectiveness. In the early 1970s, the commonly used regimen was FAC (5-fluorouracil, adriamycin, cyclophosphamide) before and after radiotherapy. In the late 1970s, FAC was given before and after mastectomy [6]. Taxanes become increasingly used in America since 2001 [33].
There are several other possible explanations to the improved survival other than treatment changes: change of the definition and classification of IBC, proportion of cause specific deaths not based on autopsy, change of patient population (age distribution, stages [IIIB and IV]). Obesity is a poor prognostic factor and so the improvement of IBC survival is not related to increasing obesity noted in the American population. Better treatment and the above factors all account for the improved survival.
Conclusion
The improvement of survival during the study period may be due to introduction of more aggressive treatments. However, there seem to be no further increase of long-term CSS, which should encourage oncologists to find even more effective treatments. Because of small numbers of patients, randomized studies will be difficult to conduct. The SEER population-based database will yield the best possible estimate of the trend in improvement of survival for patients with IBC.
List of abbreviations
C: Proportion cured
CSS: Cause-specific survival
DFS: Disease-free survival
IBC: Inflammatory breast cancer
M: Mean
OS: Overall survival
S: Standard deviation
SEER: Surveillance, Epidemiology, and End Results
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PT: Data analysis and writing of the manuscript.
EY, RS, JP, KJ, ES, SM: Critical appraisal of 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
Saskatchewan Cancer Agency Research Grant Award 2792.
Figures and Tables
Figure 1 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1975–1977.
Figure 2 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1978–1980.
Figure 3 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1981–1983.
Figure 4 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1984–1986.
Figure 5 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1987–1989.
Figure 6 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1990–1992.
Figure 7 Fitting of the lognormal model estimation to the Kaplan-Meier curve for 1993–1995.
Table 1 Parameters in the lognormal model estimations with standard errors in brackets.
Years of diagnosis Selected S Estimated M Estimated C
1975–77 0.50 1.25(0.06) 0.07(0.06)
1978–80 0.48 1.32(0.06) 0.10(0.06)
1981–83 0.48 1.37(0.06) 0.13(0.06)
1984–86 0.43 1.30(0.04) 0.15(0.05)
1987–89 0.40 1.33(0.04) 0.16(0.04)
1990–92 0.47 1.36(0.04) 0.16(0.04)
1993–95 0.39 1.33(0.03) 0.19(0.03)
S = standard deviation,
M = mean,
C = proportion cured.
Table 2 Patient characteristics.
Years of diagnosis 1975–77 1978–80 1981–83 1984–86 1987–89 1990–92 1993–95
No. of patients 134 145 181 210 258 340 416
Race:
White 117 124 162 184 207 276 352
Black 16 15 9 19 37 49 40
Others 1 6 8 7 13 15 24
Unknown 0 0 2 0 1 0 0
Age:
Median 58 56 60 56 55 59 56
Range 26–97 29–91 22–89 25–91 25–101 34–94 26–97
Stage:
Localized 6 3 6 1 1 5 5
Regional 84 100 101 41 27 20 21
Distant 35 37 72 161 218 304 385
Unstaged 9 5 2 7 12 11 5
Table 3 Number of deaths due to breast cancer and number of total deaths at six-month intervals for the 7 periods of diagnosis.
1975–77 1978–80 1981–83 1984–86 1987–89 1990–92 1993–95
Interval BCD TD BCD TD BCD TD BCD TD BCD TD BCD TD BCD TD
0–6 25 26 15 19 19 22 21 26 23 28 29 36 35 47
6–12 15 20 15 18 15 22 27 32 27 29 47 51 49 52
12–18 12 16 26 28 24 26 31 31 37 40 41 46 44 49
18–24 13 15 10 12 14 14 15 17 27 30 31 36 38 38
24–30 13 13 11 11 11 13 15 16 14 15 23 26 33 34
30–36 6 6 7 8 16 16 15 16 18 20 11 15 24 30
36–42 4 6 8 9 7 9 9 11 15 15 11 12 11 13
42–48 1 1 4 5 7 9 5 6 9 13 8 9 15 17
48–54 7 7 4 4 3 3 3 4 5 5 7 9 15 16
54–60 4 6 2 3 5 5 3 4 2 2 6 7 11 12
60–66 1 1 3 3 2 3 3 3 4 4 5 6 4 5
66–72 1 1 1 1 1 2 4 4 4 4 11 12 4 6
72–78 1 1 1 3 3 3 2 2 3 4 6 7 1 5
78–84 0 0 0 2 1 2 3 3 0 1 0 1 2 2
84–90 1 1 0 1 1 3 2 2 1 1 0 0 2 2
90–96 1 1 1 1 1 2 1 1 3 5 1 5 0 0
96–102 0 0 0 0 0 0 1 1 2 3 2 5
102–108 1 1 1 1 1 1 0 1 0 0 1 1
108–114 1 2 1 1 1 1 0 0 1 2 1 1
114–120 0 0 0 0 1 1 1 2 0 0 0 1
120–126 0 0 0 0 1 1 0 0 0 0 0 0
126–132 0 0 0 0 0 0 0 0 0 1 0 0
132–138 0 1 1 1 0 0 1 2 1 2
138–144 0 0 0 1 0 0 1 2 0 1
144–150 0 0 0 0 0 1 0 0 0 2
150–156 0 0 0 1 0 1 0 1 0 1
156–162 1 1 1 1 1 3 0 0 0 0
162–168 0 0 1 1 0 2 1 1
168–174 0 0 0 0 1 2 0 1
174–180 0 0 1 1 1 1 0 2
180–186 1 1 0 0 0 0 0 1
186–192 0 0 0 0 0 0 0 0
192–198 0 0 0 0 0 0 0 0
198–204 0 0 0 1 0 1 0 0
204–210 0 0 0 0 0 0
210–216 0 0 0 0 0 0
216–222 0 0 0 0 0 0
222–228 0 0 0 0 0 0
228–234 0 0 0 0 0 0
234–240 0 2 0 0
240–246 0 0 0 2
246–252 0 0 0 1
252–258 0 0 0 0
258–264 0 0 0 1
264–270 0 0 0 0
270–276 0 0 0 0
276–282 0 1
282–288 0 0
288–294 0 0
294–300 0 0
300–306 0 0
306–312 0 0
BCD = breast cancer deaths
TD = total deaths
Table 4 5-, 10-, 15- and 20-year cause-specific survival (CSS) rates, in percentages, estimated by Kaplan-Meier method compared with those estimated by lognormal model [in square brackets] with 95% confidence intervals (in round brackets).
Years of diagnosis 5 yr CSS 10 yr CSS 15 yr CSS 20 yr CSS
1975–77 18(16.5–19.4), [20](18.4–21.5) 11(10.3–11.6), [11](10.3–11.6) 9(8.4–9.5), [9](8.4–9.5) 9(8.4–9.5), [8](7.5–8.4)
1978–80 23(21.1–24.8), [24](22.1–25.8) 16(15.0–16.9), [14](13.1–14.8) 11(10.3–11.6), [12](11.2–12.7) 11(10.3–11.6), [11](10.3–11.6)
1981–83 27(24.8–29.1), [30](27.6–32.2) 18(16.9–19.0), [19](17.8–20.1) 14(13.1–14.8), [15](14.1–15.8) NA, [14](13.1–14.8)
1984–86 27(25.4–28.5), [26](24.4–27.5) 17(16.0–17.9), [18](16.9–19.0) 15(14.1–15.8), [16](15.0–16.9) NA, [15](14.1–15.8)
1987–89 27(25.4–28.5), [27](25.4–28.5) 19(17.8–20.1), [18](16.9–19.0) NA, [16](15.0–16.9) NA, [16](15.0–16.9)
1990–92 32(30.1–33.8), [32](30.1–33.8) 21(19.7–22.2), [22](20.7–23.2) NA, [19](17.8–20.1) NA, [18](16.9–19.0)
1993–95 29(27.8–30.1), [29](27.8–30.1) NA, [21](19.7–22.2) NA, [20](18.8–21.1) NA, [20](18.8–21.1)
NA = not available by Kaplan-Meier method
Table 5 0.5-, 1-, 2-, 3- and 4-year cause-specific survival (CSS) rates, in percentages, estimated by Kaplan-Meier method compared with those estimated by lognormal model [in square brackets] with 95% confidence intervals (in round brackets).
Years of diagnosis 0.5 yr CSS 1 yr CSS 2 yr CSS 3 yr CSS 4 yr CSS
1975–77 81(76.2–85.7), [84](79.0–88.9) 70(64.5–75.4), [66](60.8–71.1) 49(45.1–52.8), [44](40.5–47.4) 33(30.4–35.5), [32](29.4–34.5) 28(25.8–30.1), [25](23.0–26.9)
1978–80 89(83.7–94.2), [89](83.7–94.2) 79(74.3–83.6), [73](68.7–77.2) 52(47.9–56.0), [51](47.0–54.9) 38(35.0–40.9), [38](35.0–40.9) 28(25.8–30.1), [30](27.6–32.3)
1981–83 89(85.5–92.4), [90](86.4–93.5) 81(76.2–85.7), [76](71.5–80.4) 58(53.4–62.5), [55](50.6–59.3) 41(37.7–44.2), [43](39.6–46.3) 32(29.4–34.5), [35](32.2–37.7)
1984–86 90(86.4–93.5), [90](86.4–93.5) 77(72.4–81.5), [74](69.6–78.3) 53(48.8–57.1), [51](47.0–54.9) 38(35.7–40.2), [39](36.7–41.2) 30(28.2–31.7), [31](29.1–32.8)
1987–89 91(87.4–94.5), [91](87.4–94.5) 80(75.2–84.7), [77](72.4–81.5) 54(50.8–57.1), [53](49.8–56.1) 41(38.5–43.4), [40](37.6–42.3) 31(29.1–32.8), [32](30.1–33.8)
1990–92 91(87.4–94.5), [90](86.4–93.5) 77(73.9–80.0), [77](73.9–80.0) 54(50.8–57.1), [57](53.6–60.3) 43(40.4–45.5), [45](42.3–47.6) 37(34.8–39.1), [37](34.8–39.1)
1993–95 91(89.2–92.7), [92](90.1–93.8) 79(75.9–82.0), [79](75.9–82.0) 58(55.7–60.2), [56](53.8–58.1) 44(41.4–46.5), [42](39.5–44.4) 36(34.5–37.4), [34](32.6–35.3)
Table 6 Summary of key findings in research for inflammatory breast cancer.
Author Ref. Institute Year Patient No. Main results
Palangie [21] Institut Curie, France 1977–1987 223 Better DFS – rapid and complete remission with induction treatment
Chang [22] MD Anderson, US 1974–1993 177 Shorter OS – for postmenopausal IBC women with larger body size
Liauw [3] U. Florida, US 1982–2001 61 Improved CSS – tumor size <4 cm, upfront surgery, local disease control
Perez [23] Mallinckrodt, US 1958–1989 179 Improved DFS, CSS, LR – addition of mastectomy
Panades [24] British Columbia, Canada 1980–2000 485 Improved LC – mastectomy in addition to chemotherapy and radiotherapy
Improved CSS – modern chemotherapy regimens
Fleming [25] MD Anderson, US 1974–1993 178 Improved LC – mastectomy in addition to radiotherapy 16.3% vs 35.7%, P = 0.015
Buchholz [26] MD Anderson, US 1989–1997 12 Chemotherapy with radiation improved OS
Liao [27] MD Anderson, US 1977–1993 115 Improved LC, DFS, OS – twice-daily postmastectomy radiation to 66 Gy
Bertucci [28] Institut Paoli-Calmettes, France 1976–2000 74 High-dose chemotherapy: 5y DFS 28%, OS 50% Conventional dose chemotherapy: 5y DFS 15%, OS 18%
Ueno [6] MD Anderson, US 1973–1993 178 Improved DFS – combined modality: 15y DFS 28%, single-modality: 15y DFS <5%
DFS = disease free survival,
OS = overall survival,
CSS = cause-specific survival,
LC = local recurrence.
==== Refs
Wingo PA Jamison PM Young JL Gargiullo P Population-based statistics for women diagnosed with inflammatory breast cancer (United States) Cancer Causes Control 2004 15 321 328 15090727 10.1023/B:CACO.0000024222.61114.18
Gradishar WJ Inflammatory breast cancer: the evolution of multimodality treatment strategies Semin Surg Oncol 1996 12 352 363 8873324 10.1002/(SICI)1098-2388(199609/10)12:5<352::AID-SSU11>3.0.CO;2-R
Liauw SL Benda RK Morris CG Mendenhall NP Inflammatory breast carcinoma: outcomes with trimodality therapy for nonmetastatic disease Cancer 2004 100 920 928 14983486 10.1002/cncr.20083
Esteva FJ Hortobagyi GN Adjuvant systemic therapy for primary breast cancer Surg Clin North Am 1999 79 1075 1090 10572552 10.1016/S0039-6109(05)70062-1
Harris EE Schultz D Bertsch H Fox K Glick J Solin LJ Ten-year outcome after combined modality therapy for inflammatory breast cancer Int J Radiat Oncol Biol Phys 2003 55 1200 1208 12654428 10.1016/S0360-3016(02)04201-3
Ueno NT Buzdar AU Singletary SE Ames FC McNeese MD Holmes FA Theriault RL Strom EA Wasaff BJ Asmar L Frye D Hortobagyi GN Combined-modality treatment of inflammatory breast carcinoma: twenty years of experience at M. D. Anderson Cancer Center Cancer Chemother Pharmacol 1997 40 321 329 9225950 10.1007/s002800050664
Surveillance, Epidemiology, and End Results Program Public Use Data (1973–2000) National Cancer Institute, released April 2003 Bethesda: National Cancer Institute
Boag JW The presentation and analysis of the results of radiotherapy. Part I. Introduction Br J Radiol 1948 21 128 138
Boag JW The presentation and analysis of the results of radiotherapy. Part II. Mathematical theory Br J Radiol 1948 21 189 203
Boag JW Maximum likelihood estimates of the proportion of patients cured by cancer therapy J R Stat Soc (Ser B) 1949 11 15 53
Kaplan EL Meier P Nonparametric estimation from incomplete observations J Am Stat Assoc 1958 53 457 481
Rutqvist LE On the utility of the lognormal model for analysis of breast cancer survival in Sweden 1961–1973 Br J Cancer 1985 52 875 883 4074639
Mould RF Boag JW A test of several parametric statistical models for estimating success rate in the treatment of carcinoma cervix uteri Br J Cancer 1975 32 529 550 1212415
Mould RF Hearnden T Palmer M White GC Distribution of survival times of 12,000 head and neck cancer patients who died with their disease Br J Cancer 1976 34 180 190 962995
Gamel JW Greenberg RA McLean IW A stable linear algorithm for fitting the lognormal model to survival data Comput Biomed Res 1988 21 38 47 3345651 10.1016/0010-4809(88)90040-7
Gamel JW George SL Edwards MJ Seigler HF The long-term clinical course of patients with cutaneous melanoma Cancer 2002 95 1286 1293 12216097 10.1002/cncr.10813
Tai P Tonita J Yu E Skarsgard D A 20-year follow-up study of the long-term survival of limited stage small cell lung cancer and an overview of prognostic and treatment factors Int J Radiat Oncol Biol Phys 2003 56 626 633 12788167 10.1016/S0360-3016(03)00070-1
Mould RF Lederman M Tai P Wong JK Methodology to predict long-term cancer survival from short-term data using Tobacco Cancer Risk and Absolute Cancer Cure models Phys Med Biol 2002 47 3893 3924 12476973 10.1088/0031-9155/47/22/301
Chang S Parker SL Pham T Buzdar AU Hursting SD Inflammatory breast carcinoma incidence and survival: the surveillance, epidemiology, and end results program of the National Cancer Institute, 1975–1992 Cancer 1998 82 2366 2372 9635529 10.1002/(SICI)1097-0142(19980615)82:12<2366::AID-CNCR10>3.0.CO;2-N
Low JA Berman AW Steinberg SM Danforth DN Lippman ME Swain SM Long-term follow-up for locally advanced and inflammatory breast cancer patients treated with multimodality therapy J Clin Oncol 2004 22 4067 4074 15483018 10.1200/JCO.2004.04.068
Palangie T Mosseri V Mihura J Campana F Beuzeboc P Dorval T Garcia-Giralt E Jouve M Scholl S Asselain B Pouillart P Prognostic factors in inflammatory breast cancer and therapeutic implications Eur J Cancer 1994 30 921 927 10.1016/0959-8049(94)90115-5
Chang S Alderfer JR Asmar L Buzdar AU Inflammatory breast cancer survival: the role of obesity and menopausal status at diagnosis Breast Cancer Res Treat 2000 64 157 163 11194451 10.1023/A:1006489100283
Perez CA Fields JN Fracasso PM Philpott G Soares RL JrTaylor ME Lockett MA Rush C Management of locally advanced carcinoma of the breast. II. Inflammatory carcinoma Cancer 1994 74 466 476 8004622
Panades M Olivotto IA Speers CH Shenkier T Olivotto TA Weir L Allan SJ Truong PT Evolving treatment strategies for inflammatory breast cancer: a population-based survival analysis J Clin Oncol 2005 23 1941 1950 15774787 10.1200/JCO.2005.06.233
Fleming RYD Asmar L Buzdar AU McNeese MD Ames FC Ross MI Singletary SE Effectiveness of mastectomy by response to induction chemotherapy for control in inflammatory breast carcinoma Ann Surg Oncol 1997 4 452 461 9309333
Buchholz TA Tucker SL Moore RA McNeese MD Strom EA Jhingrin A Hortobagyi GN Singletary SE Champlin RE Importance of radiation therapy for breast cancer patients treated with high-dose chemotherapy and stem cell transplant Int J Radiat Oncol Biol Phys 2000 46 337 343 10661340 10.1016/S0360-3016(99)00429-0
Liao Z Strom EA Buzdar AU Singletary SE Hunt K Allen PK McNeese MD Locoregional irradiation for inflammatory breast cancer: effectiveness of dose escalation in decreasing recurrence Int J Radiat Oncol Biol Phys 2000 47 1191 1200 10889372 10.1016/S0360-3016(00)00561-7
Bertucci F Tarpin C Charafe-Jauffret E Bardou VJ Braud AC Tallet A Gravis G Viret F Goncalves A Houvenaeghel G Blaise D Jacquemier J Maraninchi D Viens P Multivariate analysis of survival in inflammatory breast cancer: impact of intensity of chemotherapy in multimodality treatment Bone Marrow Transplant 2004 33 913 920 15004544 10.1038/sj.bmt.1704458
Okawa Y Sugiyama K Aiba K Hirano A Uno S Hagino T Kawase K Shioya H Yoshida K Usui N Kobayashi M Kobayashi T Successful combination therapy with trastuzumab and Paclitaxel for adriamycin-and docetaxel-resistant inflammatory breast cancer Breast Cancer 2004 11 309 312 15550852
Volz J Volz E Bohm W Schneider V Inflammatory cancer of the breast. Therapy and follow-up Geburtshilfe Frauenheilkd 1992 52 206 209 1607113
Hance KW Anderson WF Devesa SS Young HA Levine PH Trends in inflammatory breast carcinoma incidence and survival: the surveillance, epidemiology, and end results program at the National Cancer Institute J Natl Cancer Inst 2005 97 966 975 15998949
Anderson WF Chu KC Chang S Inflammatory breast carcinoma and noninflammatory locally advanced breast carcinoma: distinct clinicopathologic entities? J Clin Oncol 2003 21 2254 2259 12805323 10.1200/JCO.2003.07.082
Somlo G Doroshow JH Synold T High-dose paclitaxel in combination with doxorubicin, cyclophosphamide and peripheral blood progenitor cell rescue in patients with high-risk primary and responding metastatic breast carcinoma: Toxicity profile, relationship to paclitaxel pharmacokinetics and short-term outcome Br J Cancer 2001 84 1591 1598 11401310 10.1054/bjoc.2001.1835
|
16242025
|
PMC1283745
|
CC BY
|
2021-01-04 16:38:31
|
no
|
Cardiovasc Ultrasound. 2005 Oct 20; 3:33
|
latin-1
|
Cardiovasc Ultrasound
| 2,005 |
10.1186/1476-7120-3-33
|
oa_comm
|
==== Front
Health Qual Life OutcomesHealth and Quality of Life Outcomes1477-7525BioMed Central London 1477-7525-3-651624890010.1186/1477-7525-3-65ResearchPsychometric properties of the 25-item National Eye Institute Visual Function Questionnaire (NEI VFQ-25), Japanese version Suzukamo Yoshimi [email protected] Tetsuro [email protected] Mitsuko [email protected] Yoshihiro [email protected] Atsuo [email protected] Kotaro [email protected] Carol M [email protected] Joseph [email protected] Shunichi [email protected] Department of Epidemiology and Healthcare Research, Kyoto University, Yoshida -Konoe-cho, Sakyo-ku, Kyoto, Japan2 Institute for Health Outcomes and Process Evaluation Research, Kudan Building 2nd floor, Iidabashi 1-4-7, Chiyoda-ku, Tokyo, Japan3 Department of Ophthalmology, Institute of Clinical Medicine, University of Tsukuba, 1-1-1 Tennoudai, Tsukuba, Ibaraki, Japan4 Department of Ophthalmology, Nihon University Hospital, 1-8-13, Kanda-Surugadai, Chiyoda-ku, Tokyo, Japan5 Inouye Eye Hospital, 4-3, Kanda-Surugadai, Chiyoda-ku, Tokyo, Japan6 Department of Ophthalmology, Graduate School of Medicine, the University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo, Japan7 Oki Eye Surgery Center, Sojukai Medical foundation, 2-17-1, Ikebukuro, Toshima-ku, Tokyo, Japan8 Division of General Internal Medicine and Health Services Research, Department of Medicine, David Geffen School of Medicine at UCLA, 911 Broxton Plaza, Room 313, Los Angeles, CA, USA9 Graduate School of Medicine, University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo, Japan2005 26 10 2005 3 65 65 1 7 2005 26 10 2005 Copyright © 2005 Suzukamo et al; licensee BioMed Central Ltd.2005Suzukamo 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 importance of evaluating the outcomes of health care from the standpoint of the patient is now widely recognized. The purpose of this study is to develop and test a Japanese version of the National Eye Institute Visual Function Questionnaire (NEI VFQ-25).
Methods
A Japanese version was developed with a previously standardized method. The questionnaire and optional items were completed by 245 patients with cataracts, glaucoma, or age-related macular degeneration, by 110 others before and after cataract surgery, and by a reference group (n = 31). We computed rates of missing data, measured reproducibility and internal consistency reliability, and tested for convergent and discriminant validity, concurrent validity, known-groups validity, factor structure, and responsiveness to change.
Results
Based on information from the participants, some items were changed to 2-step items (asking if an activity was done, and if it was done, then asking how difficult it was). The near-vision and distance-vision subscales each had 1 item that was endorsed by very few participants, so these items were replaced with items that were optional in the English version. For example, more than 60% of participants did not drive, so the driving question was excluded. Reliability and validity were adequate for all subscales except driving, ocular pain, color vision, and peripheral vision. With cataract surgery, most scores improved by at least 20 points.
Conclusion
With minor modifications from the English version, the Japanese NEI VFQ-25 can give reliable, valid, responsive data on vision-related quality of life, for group-level comparisons or for tracking therapeutic outcomes.
==== Body
Background
The importance of evaluating the outcomes of health care from the standpoint of the patient is now widely recognized. Measures of health-related quality of life (HRQOL) have been used to track outcomes for many eye diseases [1-6]. HRQOL refers to health status in the physical, mental, and social domains, and to the effect of a disease, its symptoms, and treatments on patients' lives. Conventional clinical measures such as visual acuity and visual field assessments do not fully capture the influence of visual disability on daily visual functioning and on abilities to perform activities of daily living that are valued by patients.
In response to a need for a vision-targeted measure of quality of life, the National Eye Institute (NEI) funded the development of such an instrument in the mid-1990s. The resulting 51-item questionnaire is known as the National Eye Institute Visual Function Questionnaire (NEI VFQ) [7,8]. To lessen the burden on respondents and to improve data quality, a shorter version was developed: the NEI VFQ-25 [9]. The NEI VFQ-25 has 25 items that measure vision-targeted HRQOL and are grouped into 12 subscales: general health (GH, 1 item); general vision (GV, 1 item); ocular pain (OP, 2 items); difficulty with near-vision activities (NV, 3 items); difficulty with distance-vision activities (DV 3 items); limitation of social functioning due to vision (SF, 2 items); mental health problems due to vision (MH, 4 items), role limitations due to vision (RL, 2 items); dependency on others due to vision (DP, 3 items); driving difficulties (DR, 2 items); difficulty with color vision (CV, 1 item); and difficulty with peripheral vision (PV, 1 item). Each subscale score is converted to a score between 0 and 100, and higher scores indicate better vision-specific HRQOL. The composite VFQ-25 score is the mean score of all items except for the general health item. The VFQ-25 has adequate reliability and validity, and subscale scores from the shorter form correlate highly with scores on the original long version. This questionnaire has been translated into Italian, French, Spanish, and German, and validated [10-13], and it has been widely used to describe the HRQOL of patients with ocular disease and to assess the treatment of ocular disease [14-20].
We developed a Japanese version of the NEI VFQ-25 (Appendix [see additional file 1]). and evaluated its psychometric characteristics. We investigated three points in particular. First, we looked at each question item in the Japanese version quantitatively and qualitatively, taking into consideration Japanese lifestyles, and made the necessary adaptations. Second, although composite NEI VFQ-25 scores can be computed, there is no published evidence of this scale's uni-dimensionality. Therefore, on the basis of the Japanese version's factor structure and other psychometric characteristics, we propose a particular combination of subscales that can be used to compute an appropriate composite score. Third, research on the responsiveness of the NEI VFQ-25 is limited [4,21], so we quantified its responsiveness, using data obtained before and after cataract surgery.
Methods
Development of the Japanese version
One of us (CMM) was a developer of the original NEI VFQ-25. The Japanese version was developed in conformance with standard methods that have been adopted internationally [22], including forward translation, back-translation, examination of the translation quality and adjudication by bilingual speakers, and a pilot test on 15 persons. One item was changed to make then more appropriate to Japanese life style and culture (details below). The content of the translated questionnaire was reviewed by one of the original developers of the English version, and the Japanese version was considered appropriate for administration in a psychometric field test.
Study design and population
Two groups of patients were studied. The first group was a convenience sample of 276 outpatients who visited the departments of ophthalmology at 5 hospitals. To participate, patients had to be 21 years of age or older, had to have clinical evidence of age-related cataracts, glaucoma, or age-related macular degeneration (ARMD), and had to have been seen at least twice in the past 3 months at the participating hospital. For patients with cataracts, the inclusion criteria were having cataracts in both eyes and 20/30 or worse visual acuity in the better eye. Inclusion criteria for patients with glaucoma were binocular primary open-angle glaucoma, binocular abnormalities as measured with a Humphrey field analyzer, defects in the optic nerve, at least one documented instance (in each eye) of intraocular pressure greater than 21 mmHg, and no incisional surgery for treatment of glaucoma during the previous 3 months. For patients with ARMD, there were three inclusion criteria: having at least one of the following 5 conditions: abnormal retinal pigmented epithelium, sub-retinal neovascular membrane, disciform scar, previous laser treatment to the macula, or geographic atrophy involving the fovea; having small drusen in other areas; and binocular involvement. Also included in Sample 1 was a reference group of patients with refractive error only and hospital employees.
The second sample consisted of 110 patients who had been recruited from 6 different departments of ophthalmology and were scheduled for bilateral cataract surgery (phacoemulsification and implantation of foldable intraocular lenses). Inclusion criteria for these patients were bilateral cataracts and preoperative corrected visual acuity of 20/30 or better in both eyes.
Attending physicians explained the research and ethical considerations to the participants, who then indicated their understanding by signing an informed-consent form. This study was done in accord with the Declaration of Helsinki.
Data collection
All surveys were administered by a trained interviewer. The interviewers had no direct involvement in the medical care of the patients. The interviews included the Japanese version of the NEI VFQ-25 and 14 optional items about aspects of vision-specific HRQOL (which were not presented to patients who underwent cataract surgery), and SF-36 to measure general HRQOL [23,24].
The attending physician recorded, on a structured form, the type of eye disease, duration of disease, uncorrected vision, maximally refracted vision, vision with habitual correction, and ocular pressure. In addition, severity of age-related cataracts was graded with the Lens Opacities Classification System (LOCS) III (slit lamp, standard testing conditions [25]), and in participants with glaucoma visual field was assessed with a Humphrey field analyzer 30-2. In patients with ARMD, the type of ARMD and the size and location of absolute scotoma were recorded. The data were managed by ID number, and were analyzed in a way that maintained the participants' privacy.
Statistical analysis
All statistical analyses were done with SPSS version 12 for Windows (SPSS Inc, Chicago, IL).
Descriptive analysis and item analysis
The item analysis was done using the data from the multi-condition group (Sample 1). The percentage of missing values was examined for each item. We also examined whether each item's distribution of responses was strongly skewed (large ceiling effect or floor effect).
Reliability
Cross-sectional data from the multi-condition group (Sample 1) were used to quantify reliability. Cronbach's alpha coefficient [26] was used as the index of internal consistency for each subscale. To quantify test-retest reliability, intraclass correlation coefficients [27] were used. The test-retest data were obtained from clinically stable patients with age-related cataracts, in surveys done 2 weeks apart.
Validity
The use of multi-trait analysis to evaluate convergent and discriminant validity has been described previously in detail [28]. What follows is a brief summary of the method: Each item is hypothesized to belong to only one multi-item subscale. For each item, correlations between the score on that item and the scores on all the subscales are computed. Then, for each item, if the correlation between the score on that item and the score on the subscale to which that item belongs is 0.4 or higher, that item is said to have "passed" the test of convergent validity. Also for each item, if the correlation between the score on that item and the score on the subscale to which that item belongs is greater than the correlations between the score on that item and the scores on all the subscales to which it that item does not belong, then that item is said to have "passed" the test of discriminant validity [29].
To assess concurrent validity, we computed correlations between scores on the NEI VFQ-25 and on the SF-36 subscales. We hypothesized that the NEI VFQ-25 "mental health", "social functioning", "role difficulties" and "dependency" scores would be associated more strongly with the SF-36 subscale scores that measured similar domains.
The subscale scores of participants with poor visual acuity were compared to those of participants with better visual acuity. Also, by analysis of variance, the subscale scores were compared among those with age-related cataracts, ARMD, and the reference group. In addition, scores on the peripheral-vision subscale in the patients with glaucoma were compared to those in the reference group. We also computed the correlations between subscale scores and visual acuity with habitual correction in the better and worse eye and deficits in visual fields as measured by the Humphrey Field Analyzer 30-2 in the better and worse eye.
Finally, we used factor analysis to assess the uni-dimensionality of the scale, in preparation for computing a composite score. Factor analysis was done using 10 subscales ('General Health' and 'Driving' were not included), with the maximum-likelihood solution and promax rotation. The 'Driving' subscale was not included because more than 60% of the responses on this subscale were missing.
Responsiveness
Responsiveness was studied using data from the reference group and from the patients who completed the survey before and 2 months after cataract surgery. Differences related to cataract surgery were analyzed with Student's t-test for paired data, and with the responsiveness statistic of Guyatt [30]. The responsiveness statistic is the ratio of the clinically important difference (sometimes denoted by the Greek letter delta in sample-size calculations) to the variability in stable subjects (the square root of twice the mean square error).
Results
Translation and pilot test
On the basis of the translations and discussions among the developers, one item was changed to conform better to Japanese norms. In The item "Because of your eyesight, how much difficulty do you have visiting with people in their homes, at parties, or in restaurants?", "visit at parties" was changed to "going to gatherings". Also, when a pilot test was done in 5 subjects without eye disease and 10 subjects with eye conditions, we found no expression equivalent to 'not applicable'. Therefore each item was rewritten so that it had a stem, in which the participants were asked whether they did the activity. If they indicated that they did the activity, then they were asked about the degree of difficulty in doing it. If they indicated that they did not do the activity, then they were asked whether this was due to vision problems. All such changes were discussed with, and approved by, one of the original NEI VFQ developers (CMM).
Subjects
Sample 1 had 276 participants and Sample 2 had 110. All those in Sample 1 were included in the analytic sample. In Sample 2, 4 patients did not answer the questionnaire and 11 did not respond after cataract surgery, thus 95 patients were in the analytic sample for this group. The characteristics of the participants are shown in Table 1.
Table 1 Sociodemographic and clinical characteristics of the two samples
Sample 1, n = 276 Sample 2, n = 95
Mean of age (range) 66.8 (21 to 95) 71.9 (52 to 86)
Female (%) 141 (51.0) 71 (74.7)
Visual acuity (Snellen fraction)
Better eye, mean (range) 20/120 (20/13 to 20/2000) 20/110 (20/16 to 20/2000)
Worse eye, mean (range) 20/200 (20/13 to 20/2000) 20/200 (20/20 to 20/2000)
Chronic eye disease, number (%)
Age-related cataract 96 (34.8) 95 (100)
Glaucoma 69 (25.0) Not applicable
Age-related macular degeneration 80 (29.0) Not applicable
Normal reference 31 (11.2) Not applicable
Medical comorbidities*
0 104 (37.7) Not applicable
1 97 (35.1) Not applicable
2 or more 75 (27.2) Not applicable
Of the patients with ARMD, 7 had only dry change in both eye, 8 had exudative changes in one eye, 56 patients had exudative changes in both eyes, and the status 9 patients was unknown. In the patients with glaucoma, their mean dB threshold values were -12.8 for the right eye and -12.9 for the left eye with Humphrey 30-2 threshold perimetry test. In the cataract patients of sample 1, the mean values measured by LOCS III were 2.04 for nuclear color, 2.07 for nuclear opalescence, 2.42 for cortical opacity, and 1.84 for posterior subcapsular opacity in better eye. The mean values in sample-2 patients were 2.76, 2.78, 3.31 and 2.18 respectively.
Item analysis
Percentages of missing values for each item and proportions of responses at the floor (the lowest possible score) and ceiling (the highest possible score) are shown in Table 2. 'Finding objects on crowded shelf' which was included in the 'Near Vision' subscale was not endorsed by 28% of the respondents, while 'going out to movies/plays' which was included in the 'Distance Vision' subscale was not endorsed by 32% of the sample. Three items each from the 'Near Vision' and 'Distance Vision' subscales in the optional item pool were included in the questionnaire (NV: reading small print, reading mail/bills, shaving/styling hair, DV: recognizing faces in room, participating in sports, seeing television). Subsequently, items with low rates of missing data were substituted for those with high rates, as long as the percentage of responses at the ceiling or floor did not exceed 50%. The result was that 'reading small print' was selected for the 'Near Vision' subscale and 'seeing television program' was selected for the 'Distance Vision' subscale.
Table 2 Results of item analysis. Number and percentage of missing data and of responses at the floor and ceiling (n = 276)
Subscale and Item Missing Number (%) Floor Number (%) Ceiling Number (%)
General health: GH
5-level health rating 1 (<1) 11 (4) 7 (3)
General vision: GV
5-level general vision 2 (<1) 6 (2) 1 (<1)
Near vision: NV
Reading normal newsprint 16 (6) 49 (18) 38 (14)
See well up close 36 (13) 43 (16) 43 (16)
Finding objects on crowded shelf 78 (28) 31 (11) 47 (17)
Distance vision: DV
Going out to movies/plays 88 (32) 72 (26) 37 (13)
Going down stairs at night 24 (9) 38 (14) 36 (13)
Reading street signs 11 (4) 27 (10) 53 (19)
Driving: DR
Daylight familiar places 169 (61) 43 (16) 39 (14)
Driving at night 221 (80) 15 (5) 7 (3)
Peripheral vision: PV
Seeing objects off to side 6 (2) 8 (3) 47 (17)
Color vision: CV
Difficulty matching clothes 27 (10) 5 (2) 158 (57)
Ocular pain: OP
Amount pain 1 (<1) 2 (<1) 122 (44)
Amount time: pain 0 7 (3) 189 (69)
Role limitations: RL
Accomplish less 1 (<1) 31 (11) 85 (31)
Limited in endurance 3 (1) 29 (11) 105 (38)
Dependency: DP
Need much help from others 0 32 (12) 143 (52)
Stay home most of time 1 (<1) 38 (14) 127 (46)
Rely too much on other's word 0 29 (11) 154 (56)
Social function: SF
Seeing how people react 40 (14) 22 (8) 64 (23)
Visiting others 25 (9) 22 (8) 98 (36)
Mental health: MH
Amount true: frustrated 1 (<1) 30 (11) 117 (42)
Amount true: embarrassment 1 (<1) 31 (11) 134 (49)
Amount true: no control 2 (<1) 57 (21) 86 (31)
Amount true: worry 1 (<1) 40 (15) 32 (12)
Optional items
Near vision: NV
Reading small print 6 (2) 53 (19) 36 (13)
Reading mail/bills accurately 34 (12) 41 (15) 48 (17)
Shaving/styling hair/makeup 3 (1) 3 (1) 150 (54)
Distance vision: DV
Recognizing faces in room 19 (7) 25 (9) 78 (28)
Participating in sports/outdoors 105 (38) 41 (15) 62 (23)
Seeing television program 5 (2) 17 (6) 91 (33)
More than 60% of the answers were missing for the 'Driving' subscale, which was much higher than the 16% and 31% obtained from surveys done in the United States.
Reliability
Cronbach's alpha (the index of internal consistency reliability) was 0.7 or higher for almost all of the subscales. It was lower for the 'Ocular Pain' and 'Driving' subscales. With regard to test-retest reliability, the intraclass correlation coefficient was 0.7 or higher for all of the subscales except 'General Health', 'General Vision', and 'Peripheral Vision' (Table 3). These values are considered to indicate adequate reliability for group-level comparisons [31]. Substitution of items in the 'Near Vision' and 'Distance Vision' subscales (described above) did not affect the reliability of those subscales.
Table 3 Internal consistency and test-retest reliability of NEI VFQ-25 subscales
Number of items Cronbach's alpha Intraclass correlation, for test-retest reliability Range of item-scale correlations Convergent validity* Discriminant validity**
General health 1 NA*** 0.51 NA NA NA
General vision 1 NA 0.48 NA NA NA
Near vision**** 3 0.87 (0.85) 0.76 (0.69) 0.65 – 0.75 (0.69 – 0.73) 100 (100) 93.9 (100)
Distance vision**** 3 0.84 (0.79) 0.85 (0.69) 0.67 – 0.75 (0.57 – 0.69) 100 (100) 90.9 (100)
Driving 2 0.58 0.99 0.49 – 0.49 100 50.0
Peripheral vision 1 NA 0.62 NA NA NA
Color vision 1 NA 0.74 NA NA NA
Ocular pain 2 0.44 0.75 0.28 – 0.28 0 81.8
Vision-specific
Role limitation 2 0.82 0.88 0.70 – 0.70 100 90.9
Dependency 3 0.87 0.90 0.71 – 0.82 100 97.0
Social function 2 0.74 0.88 0.58 – 0.58 100 59.1
Mental health 4 0.84 0.94 0.62 – 0.75 100 90.9
25-item composite 25 0.96 0.94 NA NA NA
* The percentage of items that passed the test of convergent validity (as described in the text).
** The percentage of items that passed the test of discriminant validity (as described in the text).
*** NA: Not applicable to single-item scales
**** Scores were recomputed after the substitution of items described in the text. Results of the recomputations are in parentheses.
Validity
All items passed the test of convergent validity, and 80% passed the test of discriminant validity. The success rates for the 'Near Vision' and 'Distance Vision' subscales were higher after item substitution than before (Table 3).
For concurrent validity, there were high correlations between scores on the NEI VFQ-25 subscales and similar domains of the SF-36 (Table 4). For example, The highest correlations were with the "Vitality" and "Mental Health" subscales, followed by the "Role Physical" and "Role Emotional" subscales. Correlations with the "Bodily Pain" and "Physical Functioning" subscales were low.
Table 4 Correlation of NEI-VFQ 25 subscales and the SF-36
SF-36
Physical Functioning Role Physical Bodily Pain General Health Vitality Social Functioning Role Emotional Mental Health
General health .308 .220 .305 .658 .473 .234 .233 .310
General vision .225 .266 .125 .164 .176 .116 .200 .231
Near vision* .327 .404 .089 .123 .158 .294 .307 .299
Distance vision* .392 .423 .145 .160 .183 .306 .344 .325
Driving .448 .398 .186 .135 .019 .278 .285 .115
Peripheral vision .195 .363 .279 .184 .253 .236 .238 .265
Color vision .412 .313 .145 .207 .141 .297 .296 .204
Ocular pain .276 .279 .306 .249 .269 .321 .266 .377
Vision-specific
Role limitation .340 .418 .173 .200 .200 .292 .376 .332
Dependency .430 .447 .145 .179 .220 .383 .403 .390
Social function .362 .405 .123 .118 .145 .313 .359 .294
Mental health .346 .453 .192 .268 .235 .392 .382 .416
25-item composite .448 .519 .222 .240 .264 .410 .441 .432
*Scores were recomputed after the substitution of items described in the text. Results of the recomputations are in parentheses.
The mean scores and the standard errors after adjustment for sex, age, and number of comorbid conditions are shown in Table 5. All scores were lower for those patients with age-related cataracts than for those in the reference group, with the exception of the 'Peripheral Vision', 'Color Vision', 'Ocular Pain', and 'Dependency' subscales. In addition, the subscales scores were significantly lower for those with ARMD than for those in the reference group, with the exception of the 'Peripheral Vision', 'Color Vision', and 'Ocular Pain' subscales. The item substitution described above resulted in slightly lower scores on the 'Near Vision' subscale and slightly higher scores on the 'Distance Vision' subscale. We tried the comparison of the explanation of variance caused by the influence of medical condition and visual acuity (Table 5). Two models associated with the NEI-VFQ score similarly.
Table 5 NEI VFQ-25 subscale scores and composite score, by condition* and the comparison of R2 between medical condition model and visual acuity model
R2 ****
Subscales Cataract n = 96 Glaucoma n = 69 Age-related Macular Degeneration n = 78 Reference group n = 31 medical condition model visual acuity model
General health 46.9 ± 2.0** 43.6 ± 2.3** 45.7 ± 2.4** 59.6 ± 4.4 .198 .209
General vision 56.0 ± 2.0** 62.7 ± 2.3** 41.2 ± 2.4** 74.0 ± 4.5 .273 .269
Near vision*** 63.0 ± 2.6** (59.5 ± 5.4**) 69.5 ± 3.0 (67.9 ± 2.8) 38.6 ± 3.1** (31.9 ± 2.9**) 77.4 ± 5.8 (74.5 ± 5.4) .341 .346
Distance vision*** 59.1 ± 2.7** (65.1 ± 2.3**) 63.3 ± 3.1 (71.0 ± 2.7**) 40.0 ± 3.2** (47.4 ± 2.8**) 75.9 ± 6.0 (83.7 ± 5.1) .268 .325
Driving 52.2 ± 5.5** 55.3 ± 6.0** 12.8 ± 5.2** 85.0 ± 10.9 .495 .424
Peripheral vision 57.3 ± 2.7 56.9 ± 3.1 64.1 ± 3.3 69.0 ± 6.1 .029 .126
Color vision 85.2 ± 2.1 89.6 ± 2.4 90.0 ± 2.6 88.1 ± 4.5 .081 .119
Ocular pain 80.5 ± 2.0 81.5 ± 2.4 83.2 ± 2.5 83.3 ± 4.6 .036 .100
Vision-specific
Role limitation 71.3 ± 2.6** 73.5 ± 3.1 38.4 ± 3.2** 85.5 ± 5.9 .340 .252
Dependency 75.6 ± 2.8 83.9 ± 3.3 51.3 ± 3.5** 85.8 ± 6.3 .300 .306
Social function 73.5 ± 2.6** 80.0 ± 2.9 56.3 ± 3.1** 88.1 ± 5.6 .227 .252
Mental health 65.5 ± 2.5** 68.8 ± 2.9** 37.1 ± 3.1** 89.8 ± 5.7 .330 .300
25-item composite 66.0 ± 1.6** 69.8 ± 1.9** 51.0 ± 2.0** 80.1 ± 3.7 .331 .331
* Data are presented as mean ± SE adjusted by sex, age, and number of comorbid conditions.
** Significantly different from the reference group.
*** Scores were recomputed after the substitution of items described in the text. Results of the recomputations are in parentheses.
**** R2 showed the comparison of the explanation of variance caused by the influence of medical condition and visual acuity. The medical condition model have 'medical condition', 'sex', 'age', and 'number of comorbid condisions' as the explanatory variables and the visual acuity model have 'visual acuity', 'sex', 'age', and 'number of comorbid condisions'.
Visual acuity in the better eye (logMAR, the logarithm of the minimum angle of resolution) was strongly correlated with subscales that are influenced by the ability to use central vision: 'General vision', 'Near Vision', and 'Distance Vision' (Table 6). As would be expected, the logMAR was only weakly correlated with the subscales that are less dependent on the quality of central vision: 'General Health', 'Peripheral vision', 'Ocular Pain', and 'Color Vision'. In patients with glaucoma, visual field deficits were strongly correlated with scores on three subscales: 'Distance Vision', 'Driving', and 'Peripheral Vision' (Table 6). These correlations are similar to those observed between clinical measures and NEI VFQ scores in the NEI psychometric field test [9].
Table 6 Pearson correlations of NEI VFQ-25 subscale scores with visual acuity and visual field
Subscale Visual acuity* Visual field**
Better eye Worse eye Better eye Worse eye
General health 0.06 0.06 0.03 -0.01
General vision 0.55 0.50 0.33 0.39
Near vision*** 0.60 (0.64) 0.56 (0.59) 0.33 (0.34) 0.27 (0.30)
Distance vision*** 0.60 (0.59) 0.51 (0.52) 0.60 (0.54) 0.52 (0.41)
Driving 0.58 0.56 0.61 0.44
Peripheral vision 0.06 0.06 0.45 0.41
Color vision 0.28 0.23 0.01 -0.06
Ocular pain -0.02 -0.02 -0.15 -0.20
Vision-specific
Role limitation 0.51 0.46 0.36 0.19
Dependency 0.59 0.51 0.49 0.40
Social function 0.56 0.47 0.47 0.28
Mental health 0.55 0.49 0.49 0.41
25-item composite 0.61 0.54 0.49 0.39
Bold characters indicate correlation coefficients of 0.4 or greater.
* Visual acuity (logMAR) while wearing usual correction, in all subjects in Sample 1
** Visual field testing with the Humphrey Field Analyzer 30-2 (only glaucoma, n = 69)
*** Scores were recomputed after the substitution of items described in the text. Results of the recomputations are in parentheses.
The results of factor analysis done with 10 subscales ('General Health' and 'Driving' were excluded) are shown in Table 7. Two factors were extracted. The 'Peripheral Vision', 'Ocular Pain', and 'Color Vision' subscales were included in the second factor. The correlation between the two factors was 0.47. The results of factor analysis done with 22 items (1 item on 'General Health' and 2 items on 'Driving' were excluded) had the similar to the structure of scale-level analysis.
Table 7 Results of factor analysis on 10 subscales of VFQ-25 ('General Health' and 'Driving' were excluded): factor loadings after promax rotation
Subscale Factor 1 Factor 2
Near vision 0.909 -0.118
Mental health 0.866 0.052
Role limitation 0.850 -0.043
Dependency 0.838 0.053
Distance vision 0.828 0.067
General vision 0.757 -0.095
Social functioning 0.724 0.072
Peripheral vision -0.033 0.701
Ocular pain -0.094 0.575
Color vision 0.243 0.421
Responsiveness
The mean of visual acuity of patients with cataract were 20/200 before surgery, and after surgery it had improved to 20/50 in the better eye. In the reference group, scores were stable over two months. In the cataract surgery group, surgery was associated with significant increases in the composite score and in 8 subscale scores: 'General Vision', 'Near Vision', 'Distance Vision', 'Ocular Pain', 'Social Functioning', 'Mental Health', 'Role Limitation', and 'Dependency' (Figure 1). Guyatt's index of responsiveness for those subscale scores ranged from 1.91 to 7.35. Even the lower limit of that range would be considered to be extremely high [32]. The only exception was the 'General Health' subscale, which would not have been expected to be strongly influenced by cataract surgery.
Figure 1 Adjusted change in NEI VFQ-25 scores in the cataract-surgery group and the reference group. Change score adjusted for sex and age.
On the basis of the result of factor analysis, we computed 3 different composite scores: composite 11 (all VFQ-25 subscales except 'General Health'), composite 10 (all VFQ-25 subscales except 'General Health' and 'Driving'), and composite 7 (only those 7 subscales of the VFQ-25 that loaded heavily on Factor 1, as indicated in Table 7). The responsiveness indexes of these three composite scores were 7.18, 8.03, and 8.86, respectively, all of which are acceptable from a psychometric perspective.
Discussion
We developed a Japanese version of the NEI VFQ-25, and documented its psychometric characteristics in patients with various chronic eye conditions. Overall, we found that the Japanese version can provide data that are reliable, valid, and responsive to change in visual function.
In developing the Japanese version, a few changes to the content of the questionnaire were needed. For some of the items in the 'Near Vision' and 'Distance Vision' subscales, we found that the rates of missing data in the Japanese version were much higher than in the original English version. To minimize the rates of missing data and thereby to increase the measurement precision, we propose substituting items that are appropriate for patients in Japan. Specifically, instead of 'finding objects on crowded shelf', 'reading small print' can be used in the 'Near vision' subscale; and instead of 'going out to movies/plays', 'seeing television program' can be used in the 'Distance vision' subscale (both are from the pool of optional NEI VFQ items). Rates of missing data were much lower after those substitutions than before. The 'Near Vision' score was slightly higher and the 'Distance Vision' score was slightly lower, but their reliability and validity were virtually unchanged.
The 'Driving' subscale also had a high rate of missing data. We suggest that in Japan the 'Driving' subscale should be optional.
Composite scores can be useful summaries of visual function, particularly when the content of such a score is based on the results of factor analysis. In this study, factor analysis indicted that most of the subscales that are influenced by central vision correlated strongly with the first factor, while the 'Ocular Pain', 'Peripheral Vision', and 'Color Vision' subscales correlated strongly with the second factor. Therefore, if only one composite score is to be computed, that score should not include the 'Ocular Pain', 'Peripheral Vision', or 'Color Vision' subscales. Nonetheless, for studies of interventions involving small numbers of subjects we suggest using the 7-subscale composite score, given the caveat that it would not reflect problems with color vision, peripheral vision, or ocular pain. Furthermore, we suggest using the 10-subscale composite score when evaluating patients who have ocular pain or a disorder involving color vision.
Few reports [4,25] were available up until now on the responsiveness of the NEI VFQ-25. Almost no changes were observed in the VFQ-25 scores over 2 months in the reference group. In contrast, in the patients who underwent cataract surgery, many subscale scores increased by about 20 points. These increases occurred not only in the scores on subscales related directly to vision ('General Vision', 'Near Vision', and 'Distance Vision'), but also in the scores on subscales that are less vision-specific ('Mental Health', 'Dependency', 'Social Functioning', and 'Role Limitation'). Scores on the 'General Health', 'Color Vision', and 'Peripheral Vision' subscales did not change with cataract surgery. These results show that with the Japanese version of the NEI VFQ-25, one can easily detect clinically important changes such as those resulting from cataract surgery.
Interpretation of these results is limited in at least four ways. First, this study did not include patients with diabetic retinopathy, low vision, and a large number of other eye conditions. Thus, whether these findings are applicable to patients with diseases other than cataracts, glaucoma, and ARMD remains to be studied. Second, we used a convenience sample of persons with these conditions, and they may not represent the full clinical spectrum of each disease. Third, it is unclear whether the mode of administration (self-administered or interviewer administered) would have important effects on the results. However, we obtained the present data with trained interviewers, and we note that the findings are similar to those obtained in a field survey with the original English version, even though the questionnaire in that survey was self-administered. Fourth, the responsiveness results were obtained without the aforementioned substitutions in the 'Near Vision' and 'Distance Vision' subscales. Thus, the responsiveness of those two subscales should be examined again, after the recommended substitutions.
Conclusion
In conclusion, psychometric testing indicates that data obtained with the Japanese version of the NEI VFQ-25 are sufficiently reliable, valid, and responsive for group-level comparisons. For reasons described in detail above, we suggest that a few items be substituted and that a few be removed from the composite score. Using this scale in vision-related clinical research in Japan should facilitate evaluations of clinical care and outcomes from the standpoint of the patient.
Authors' contributions
Suzukamo Y assumed the coordination and design of this study, training of interviewers, data analysis and interpretation, and drafting this article. Oshika T, Yuzawa M, Tokuda Y, Tomidokoro A and Oki K contributed in the design of this study and acquisition of data. Mangione CM, Green J and Fukuhara S contributed in the concept and design of this study, interpretation of the data, and revising the article critically for important intellectual content.
Supplementary Material
Additional File 1
Appendix: The list of items on the Japanese NEI-VFQ 25.
Click here for file
Acknowledgements
We are grateful to Gentaro Sugita, Ken Hayashi, Shuichiro Eguchi, Kazunori Miyata, and Tadahiko Kozawa for assistance in collecting the data, and to Natsuko Takahashi for assistance in analyzing the data.
==== Refs
Submacular Surgery Trials Research Group Health- and vision-related quality of life among patients with ocular histoplasmosis or idiopathic choroidal neovascularization at enrollment in a randomized trial of submacular surgery: Submacular Surgery Trials Report No. 5 Arch Ophthalmol 2005 123 78 88 15642816 10.1001/archopht.123.1.78
Cahill MT Banks AD Stinnett SS Toth CA Vision-related quality of life in patients with bilateral severe age-related macular degeneration Ophthalmology 2005 112 152 158 15629836 10.1016/j.ophtha.2004.06.036
Tranos PG Ghazi-Nouri SM Rubin GS Adams ZC Charteris DG Visual function and subjective perception of visual ability after macular hole surgery Am J Ophthalmol 2004 138 995 1002 15629291 10.1016/j.ajo.2004.07.049
Hawkins BS Miskala PH Bass EB Bressler NM Childs AL Mangione CM Marsh MJ Submacular Surgery Trials Research Group Surgical removal vs observation for subfoveal choroidal neovascularization, either associated with the ocular histoplasmosis syndrome or idiopathic: II. Quality-of-life findings from a randomized clinical trial: SST Group H Trial: SST Report No. 10 Arch Ophthalmol 2004 122 1616 1628 15534122 10.1001/archopht.122.11.1616
Childs AL Bressler NM Bass EB Hawkins BS Mangione CM Marsh MJ Miskala PH Submacular Surgery Trials (SST) Research Group Surgery for hemorrhagic choroidal neovascular lesions of age-related macular degeneration: quality-of-life findings: SST report no. 14 Ophthalmology 2004 111 2007 2014 15522365 10.1016/j.ophtha.2004.07.024
Scilley K DeCarlo DK Wells J Owsley C Vision-specific health-related quality of life in age-related maculopathy patients presenting for low vision services Ophthalmic Epidemiol 2004 11 131 146 15255028 10.1076/opep.11.2.131.28159
Mangione CM Berry S Identifying the content area for the 51-item National Eye Institute Visual Functioning Questionnaire: results from focus groups with visually impaired persons Arch Ophthalmol 1998 116 227 233 9488276
Mangione CM Lee PP Pitts J Gutierrez P Berry S Hays RD Psychometric properties of the National Eye Institute Visual Function Questionnaire (NEI-VFQ). NEI-VFQ Field Test Investigators Arch Ophthalmol 1998 116 1496 1504 9823352
Mangione CM Lee PP Gutierrez PR Spritzer K Berry S Hays RD National Eye Institute Visual Function Questionnaire Field Test Investigators Development of the 25-item National Eye Institute Visual Function Questionnaire Arch Ophthalmol 2001 119 1050 1058 11448327
Rossi GC Milano G Tinelli C The Italian version of the 25-item National Eye Institute Visual Function Questionnaire: translation, validity, and reliability J Glaucoma 2003 12 213 220 12782838 10.1097/00061198-200306000-00006
Nordmann JP Viala M Sullivan K Arnould B Berdeaux G Psychometric Validation of the National Eye Institute Visual Function Questionnaire – 25 (NEI VFQ-25) French version: in a population of patients treated for ocular hypertension and glaucoma Pharmacoeconomics 2004 22 197 206 14871166
Broman AT Munoz B West SK Rodriguez J Sanchez R Snyder R Klein R Psychometric properties of the 25-item NEI-VFQ in a Hispanic population: Proyecto VER Invest Ophthalmol Vis Sci 2001 42 606 613 11222517
Franke GH Esser J Reimer J Maehner N Vision targeted quality of life under different degrees of visual impairment Revista Portuguesa de Psicossomática 2002 4 39 49
COMS Quality of Life Study Group Quality of life assessment in the collaborative ocular melanoma study: design and methods. COMS-QOLS Report No. 1 Ophthalmic Epidemiol 1999 6 5 17 10384680 10.1076/opep.6.1.5.1565
Cole SR Beck RW Moke PS Gal RL Long DT The National Eye Institute Visual Function Questionnaire: experience of the ONTT. Optic Neuritis Treatment Trial Invest Ophthalmol Vis Sci 2000 41 1017 1021 10752936
Miskala PH Bressler NM Meinert CL Relative contributions of reduced vision and general health to NEI-VFQ scores in patients with neovascular age-related macular degeneration Arch Ophthalmol 2004 122 758 766 15136325 10.1001/archopht.122.5.758
Smith JA Vitale S Reed GF Grieshaber SA Goodman LA Vanderhoof VH Calis KA Nelson LM Dry eye signs and symptoms in women with premature ovarian failure Arch Ophthalmol 2004 122 151 156 14769589 10.1001/archopht.122.2.151
Deramo VA Cox TA Syed AB Lee PP Fekrat S Vision-related quality of life in people with central retinal vein occlusion using the 25-item National Eye Institute Visual Function Questionnaire Arch Ophthalmol 2003 121 1297 1302 12963613 10.1001/archopht.121.9.1297
Clemons TE Chew EY Bressler SB McBee W Age-Related Eye Disease Study Research Group. National Eye Institute Visual Function Questionnaire in the Age-Related Eye Disease Study (AREDS): AREDS Report No. 10 Arch Ophthalmol 2003 121 211 217 12583787
Schiffman RM Jacobsen G Whitcup SM Visual functioning and general health status in patients with uveitis Arch Ophthalmol 2001 119 841 849 11405835
Cahill MT Stinnett SS Banks AD Freedman SF Toth CA Quality of life after macular translocation with 360 degrees peripheral retinectomy for age-related macular degeneration Ophthalmology 2005 112 144 151 15629835 10.1016/j.ophtha.2004.06.035
Acqudro C Jambon B Ellis D Marquis P Spilker B Language and translations issues Quality of life and pharmacoeconomics in clinical trials 1995 2 Philadelphia(PA): Lippincott Raven 575 585
Fukuhara S Bito S Green J Hsiao A Kurokawa K Translation, adaptation, and validation of the SF-36 Health Survey for use in Japan J Clin Epidemiol 1998 51 1037 1044 9817121 10.1016/S0895-4356(98)00095-X
Fukuhara S Ware JE JrKosinski M Wada S Gandek B Psychometric and clinical tests of validity of the Japanese SF-36 Health Survey J Clin Epidemiol 1998 51 1045 1053 9817122 10.1016/S0895-4356(98)00096-1
Chylack LT JrWolfe JK Singer DM Leske MC Bullimore MA Bailey IL Friend J McCarthy D Wu SY The Lens Opacities Classification System III. The Longitudinal Study of Cataract Study Group Arch Ophthalmol 1993 111 831 836 8512486
Cronbach LJ Coefficient alpha and the internal structure of tests Psychometrika 1951 16 297 334
Kramer MS Feinstein AR Clinical biostatistics LIV: the biostatistics of concordance Clin Pharmacol Ther 1981 29 111 123 7460469
Campbell DT Fiske DW Convergent and discriminant validation by the multitrait- multimethod matrix Psychol Bull 1959 56 81 105 13634291
Hays R Hayashi T Beyond internal consistency reliability: Rationale and user's guide for Multirait Analysis Program on the microcomputer Behav Res Methods Instrum Comput 1990 22 167 175
Guyatt G Walter S Norman G Measuring change over time: assessing the usefulness of evaluative instruments J Chron Dis 1987 40 171 178 3818871 10.1016/0021-9681(87)90069-5
Nunnally JC Psychometric Theory 1978 New York, NY: McGraw-Hill Inc 190 255
Husted JA Cook RJ Farewell VT Gladman DD Methods for assessing responsiveness: a critical review and recommendations J Clin Epidemiol 2000 53 459 468 10812317 10.1016/S0895-4356(99)00206-1
|
16248900
|
PMC1283746
|
CC BY
|
2021-01-04 16:38:13
|
no
|
Health Qual Life Outcomes. 2005 Oct 26; 3:65
|
utf-8
|
Health Qual Life Outcomes
| 2,005 |
10.1186/1477-7525-3-65
|
oa_comm
|
==== Front
Hum Resour HealthHuman Resources for Health1478-4491BioMed Central London 1478-4491-3-111625313710.1186/1478-4491-3-11ResearchThe effect of performance-related pay of hospital doctors on hospital behaviour: a case study from Shandong, China Liu Xingzhu [email protected] Anne [email protected] Abt Associates Inc., Bethesda, Maryland, USA2 Health Economics and Financing Programme, London School of Hygiene and Tropical Medicine, London, UK2005 27 10 2005 3 11 11 17 5 2005 27 10 2005 Copyright © 2005 Liu and Mills; licensee BioMed Central Ltd.2005Liu and Mills; 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
With the recognition that public hospitals are often productively inefficient, reforms have taken place worldwide to increase their administrative autonomy and financial responsibility. Reforms in China have been some of the most radical: the government budget for public hospitals was fixed, and hospitals had to rely on charges to fill their financing gap. Accompanying these changes was the widespread introduction of performance-related pay for hospital doctors – termed the "bonus" system. While the policy objective was to improve productivity and cost recovery, it is likely that the incentive to increase the quantity of care provided would operate regardless of whether the care was medically necessary.
Methods
The primary concerns of this study were to assess the effects of the bonus system on hospital revenue, cost recovery and productivity, and to explore whether various forms of bonus pay were associated with the provision of unnecessary care. The study drew on longitudinal data on revenue and productivity from six panel hospitals, and a detailed record review of 2303 tracer disease patients (1161 appendicitis patients and 1142 pneumonia patients) was used to identify unnecessary care.
Results
The study found that bonus system change over time contributed significantly to the increase in hospital service revenue and hospital cost recovery. There was an increase in unnecessary care and in the probability of admission when the bonus system switched from one with a weaker incentive to increase services to one with a stronger incentive, suggesting that improvement in the financial health of public hospitals was achieved at least in part through the provision of more unnecessary care and drugs and through admitting more patients.
Conclusion
There was little evidence that the performance-related pay system as designed by the sample of Chinese public hospitals was socially desirable. Hospitals should be monitored more closely by the government, and regulations applied to limit opportunistic behaviour. Otherwise, the containment of government financing for public facilities may result in an increase in the provision of unnecessary care, an increase in health costs to society, and a waste in social resources.
==== Body
Background
Health policy researchers and policy makers have increasingly recognized that health care providers have a powerful influence on health care provision and the use of health care resources. With the recognition that public hospitals are often productively inefficient, reforms have taken place worldwide to increase the administrative autonomy and financial responsibility of public hospitals [1,2]. Reforms in China are perhaps some of the most radical. Starting from the early 1980s, the government budget for public hospitals was fixed, and hospitals had to rely on charges to fill the gap between hospital expenditure and income from the government. Medical prices regulated by government were increased and hospitals allowed to earn profits from certain services and from drugs. Accompanying these changes was the widespread introduction of performance-related pay for hospital doctors – termed the "bonus" system. The policy objective was to improve productivity and cost recovery [3].
The bonus system is widespread and used by almost all hospitals in China. The types of hospital bonus system can be summarized into three different forms: flat bonus, quantity-related bonus and revenue-related bonus [4-12]. A flat bonus is distributed among hospital staff equally or almost equally, with the amount depending on the overall financial status of the hospital. A quantity-related bonus is paid according to the quantity of services provided (visits, admissions, inpatient days, medical procedures, and tests and examinations), usually with a quantity target above which the bonus is paid. A revenue-related bonus depends on the revenue generated by doctors through provision of services and drugs over a revenue target.
A survey of bonus systems in all county hospitals in Shandong province in 1997 [13] found that 78% had a revenue-related system and the remaining 22% a quantity-related system. The average bonus per month (83 yuan) was around 10% of the monthly salary but depended on the financial status of the hospital. In 1997, 11% of hospitals paid no bonus, and 37% paid an average bonus of over 100 yuan per month. Average bonus amounts were not significantly different by type of bonus system.
The objectives of the study reported here were to assess the effects of the bonus system on hospital revenue, cost recovery and productivity, and to explore whether bonus pay was associated with the provision of unnecessary care. The primary hypothesis tested in the study was that the impact of a bonus system on revenue, cost recovery, productivity and unnecessary care would depend on the strength and direction of the economic incentive of different bonus systems, and doctors' responses to these incentives. It was assumed that an increase in provision of necessary care would provide utility gains for doctors in the form of income and a utility loss in the form of greater effort, and that the provision of unnecessary care would also provide income utility gains, but at a utility loss stemming from ethical concerns as well as greater effort. Thus the behaviour of doctors would be determined by the trade-off of utility gains and losses, and the motivation for doctors to provide unnecessary care would be constrained by their desire for leisure and for ethical behaviour.
It was hypothesized that a flat bonus might motivate doctors to provide necessary care but would be less likely to motivate them to provide unnecessary care. If there is enough demand, the effort of an individual doctor to provide more necessary care would benefit patients, the hospital and the doctor as well. Productivity, cost recovery and quality of care could all improve. If there is insufficient demand, the doctor could induce demand but might be less motivated to do so because the bonus income would be distributed throughout the hospital. A quantity-related bonus would provide a stronger economic incentive for doctors to provide a greater quantity of services, and might help to improve productivity, cost recovery and quality of care if there is sufficient demand. But when demand is insufficient, doctors may be motivated to induce demand by providing more care, regardless of need. What types of service would be overprovided would depend on how quantity is defined and measured. Finally, a revenue-related bonus provides the strongest incentive for doctors to induce patient demand for both services and drugs.
Methods
Definition and measurement of key indicators
Productivity is generally defined as the ratio of a hospital's output to its input. This study employed both unidimensional ratio analysis (outpatient visits per doctor, admissions per doctor, bed occupancy rate, length of stay); and Data Envelopment Analysis, a linear programming method that measures the technical efficiency of production [14] and has been used to measure hospital productivity by numerous authors (e.g. [15]). The relative level of hospital productivity was indicated by the DEA efficiency score, which ranges from 0 to 100, with, for a given set of inputs, 0 meaning no output and 100 meaning the maximization of output. The objective function of hospital o compared with the n hospitals in the data set is:
where o represents the hospital being evaluated in the set of j = 1,...,n hospitals; E is the efficiency score; ur is the weight for the rth output; yro is the rth output for the oth hospital; s is the number of outputs; vi is the weight of the ith input; xio is the ith input for the oth hospital; m is the number of inputs. The DEA efficiency score was calculated using software developed by Warwick Business School [16], assuming constant returns to scale. Hospital inputs included the number of doctors, the number of nurses, the monetary value of hospital fixed assets, the number of hospital beds and the monetary value of supplies. The hospital outputs included the number of admissions, outpatient visits and surgical operations.
Cost recovery was defined as service revenue expressed as a percentage of recurrent and total costs. Unnecessary care was defined as services and drugs provided that were judged by a panel of doctors to lead to no improvement in patient outcomes. Unnecessary care can be assessed only in relation to the nature of the cases being treated, so appendicitis and child pneumonia were selected as tracer diseases on the basis that they were common so there were enough cases for each hospital and year; they had clear-cut diagnoses so the sample would be homogeneous; and both had a standard plan of treatment, so variation in treatment would not be due to treatment uncertainty. Six surgeons (for appendicitis cases) and six paediatricians (for pneumonia cases) worked together with the investigators to develop guidelines for appropriate management of the two tracer diseases and for identification of unnecessary care. These guidelines were positive lists of types and quantities of services and drugs necessary for the improvement of the health outcomes of patients. The unnecessary care indicator was unnecessary care expenditure as a percent of total expenditure. A detailed description of the methodology is in Liu and Mills [17].
Samples and data
Based on a census of all 127 county general hospitals in Shandong province [18], 25 hospitals were selected that had experienced change of bonus system and that had complete inpatient files for the previous 10 years. These hospitals were categorized into three groups based on county income level, and two hospitals randomly selected from each group. Data were collected from these six panel hospitals for the period 1978–1997 on the type of bonus system in different years; total revenue by source; recurrent and capital cost; staff numbers; and activity data (outpatients, inpatients, operations, CT scans, average length of stay, bed occupancy rate).
Inpatient records were collected from the six panel hospitals for each change of bonus system and encompassing the year of the switch and the two years before and after. Inpatient records for each year, disease and hospital were drawn from the beginning of the year until the sample size reached 30 or until there were no more records available. Altogether the study included 1161 appendicitis patients and 1142 pneumonia patients.
Patient files were randomly distributed to the relevant doctors who reviewed the files and recorded the types and quantities of services and drugs actually used, and the types and quantities of unnecessary services and drugs. If the removal of unnecessary items was considered to result in inadequate treatment, substitute necessary services and drugs were added in. Finally, actual expenditures and unnecessary expenditures were computed by accountants according to the 1997 provincial fee schedule.
Because of the burden of work, each record was reviewed by only one doctor, but to check for consistency, 61 patient records for appendicitis and 57 for pneumonia were selected and randomly distributed to the panel doctors for re-reviews, without their knowledge, and the results compared. Means were very similar: none of the ps of t-tests were less than 0.05 and most were very close to 1.
Data analysis
In the data analysis, we first described the historical changes in bonus system, hospital revenue, productivity and cost recovery, and then analysed the level of unnecessary care.
Trend analysis was performed to examine the changes in hospital revenue, cost recovery, productivity, and the rate of unnecessary expenditure following the bonus switches. In the trend analysis, the indicators were assessed for a continuous five years for each of the hospital bonus switches, including the year of bonus switch and the two years before and the two years after the switch.
In correlation and regression analysis, we examined the relationships of the bonus system with the four key variables (hospital revenue, cost recovery, hospital productivity and unnecessary care). First, the interrelationships among these four variables were examined through correlation analysis of each pair of variables. The observation units were hospital-years. The type of bonus was measured by a dummy variable with values reflecting the expected strength of economic incentives to overprovide (non-bonus = 1; flat bonus = 2; revenue-related bonus = 3). Cost recovery was measured by the rate of recovery of recurrent cost, and hospital productivity by the DEA efficiency score.
The subsequent set of analyses used stepwise regression analysis to examine whether and how hospital revenue, cost recovery, unnecessary care and productivity were related to each other, and to what extent they were explained by the bonus system. Each of the four variables was taken in turn as a dependent variable, and the factors that might explain the variations in the dependent variable examined (the results of the four stepwise regression analyses yielded results similar to a general regression analysis in terms of R-squares and statistical significance levels of the independent variables). Besides indicators of revenue, cost recovery, unnecessary care and productivity, the year, names of hospitals and bonus type were put into the regression models as independent variables. The six hospital names were arranged into five dummy variables. The year measured all the factors that changed with time (e.g. medical price inflation, technology improvement and the increase in demand for care). The names of hospitals measured all the factors that were related to each hospital (e.g. the level of demand faced by an individual hospital, management capacity, degree of observance of ethical codes, etc.).
Results
Historical changes in bonus system, hospital revenue, productivity and cost recovery
All six panel hospitals had experienced bonus switches from no-bonus to flat-bonus and then to revenue-related bonus (Table 1). None had experience of quantity-related bonus. Before the early 1980s, no hospital had a bonus system; by 1988 all hospitals had a flat bonus system; and by the middle of the 1990s, all hospitals had a revenue-related bonus system.
Table 1 Bonus systems of the six panel hospitals (1978 – 1997)
Year Zhaoyuan Liangshan Qixia Weishan Changyi Yanzhou
1978 1 1 1 1 1 1
1979 1 1 1 1* 1 1
1980 1 1 1* 1* 1 1
1981 1 1 1* 2* 1 1
1982 1 1 2* 2* 1 1*
1983 1 1 2* 2* 1 1*
1984 1 1 2* 2 2 2*
1985 1 1 2 3 2 2*
1986 1 1 2 3 2 2*
1987 2 1 2 3 2 2
1988 2 2 2 3 2 2
1989 2 2 2 3 2 2
1990 2 2 2 3 2 2
1991 2 2 2 3 2 2
1992 2 2 2 3 2 2
1993 3 2 3 1 2 2
1994 3 3 3 1 3 3
1995 3 3 3 3 3 3
1996 3 3 3 3 3 3
1997 3 3 3 3 3 3
Notes:
1) 1: No bonus; 2: Flat bonus; 3: Revenue-related.
2) Numbers in italics indicate years for which inpatient records were collected.
3) * indicates data on tracer diseases were not available around the years of the bonus switch.
Coinciding with the change in bonus systems, over the period 1978–97, there was a remarkable increase in hospital revenue, an increase in hospital cost recovery, a doubling of admissions, a decrease in visits and a tripling of operations (Table 2). The average revenue increase in real terms was 16.3% per year, and 3.7% per year for admissions. In 1978, only 3% of outpatients were admitted into hospital and only 17% of inpatients were operated on, while by 1997 these had increased to 7% and 25%. Staff numbers and hospital beds increased over time and because the increase in inputs exceeded the increase in outputs, most productivity indicators decreased (Table 3) with the exception of operations per doctor. Quality changes might have occurred but could not be assessed with the available data.
Table 2 Changes in average activity levels of the panel hospitals, 1978 – 1997
Year Change in revenue (1975 = 100) Recovery of recurrent cost (%) Change in the number of: (1975 = 100)
Admissions Visits Operations No. of visits per admission
1978 100 75.4 100 100 100 31.7
1979 124 79.3 111 102 103 29.3
1980 132 79.3 123 105 116 27.1
1981 140 82.0 124 106 114 27.2
1982 155 82.1 104 108 118 32.9
1983 168 80.3 110 105 122 30.3
1984 206 77.9 106 112 131 33.4
1985 206 73.3 108 111 120 32.4
1986 270 82.7 119 115 141 30.6
1987 341 89.1 138 121 161 27.9
1988 436 97.8 149 120 181 25.6
1989 525 104.9 161 112 196 22.0
1990 634 101.4 160 113 202 22.5
1991 728 93.8 171 120 220 22.3
1992 839 98.3 149 115 218 24.6
1993 1,046 100.2 157 97 245 19.6
1994 1,063 90.8 158 85 254 17.0
1995 1,280 99.4 172 83 265 15.2
1996 1,552 104.1 185 90 268 15.4
1997 1,755 102.0 201 91 283 14.3
Table 3 Changes in productivity in the panel hospitals, 1978 – 1997
Year Bed occupancy rate (%) Length of stay (days) Visits per doctor Admissions per doctor Operations per doctor DEA efficiency score
1978 86 9.1 2041 64.3 11.3 97.1
1979 91 9.0 1955 68.7 10.9 97.1
1980 88 9.0 1893 72.0 11.5 96.9
1981 90 9.2 1712 64.0 10.2 95.6
1982 89 10.0 1660 51.2 9.8 87.9
1983 90 10.2 1561 52.0 9.8 86.5
1984 87 10.5 1480 43.5 9.6 84.8
1985 91 10.7 1374 41.9 8.2 74.7
1986 94 11.3 1378 44.9 9.3 79.5
1987 93 11.2 1289 44.6 9.0 77.0
1988 95 11.1 1297 48.4 10.5 75.2
1989 92 11.3 1063 45.6 9.9 77.1
1990 94 11.8 1066 45.1 10.3 74.9
1991 92 11.1 1165 50.3 11.6 78.9
1992 88 12.2 1019 39.6 10.4 76.2
1993 86 11.7 839 40.9 11.4 75.1
1994 79 11.3 720 40.2 11.3 74.8
1995 82 10.8 675 42.2 11.4 74.7
1996 82 10.2 689 44.2 11.1 73.5
1997 79 10.6 684 48.2 11.8 72.6
Unnecessary expenditure
There was a large amount of unnecessary expenditure for both tracer diseases. The average expenditure (1997 prices) for an appendicitis patient was 774 yuan (USD 95), of which 18% was considered unnecessary. For a pneumonia patient these figures were 559 yuan (USD 68) and 19%. Further analyses showed that more than one third of the expenditure for drugs was deemed unnecessary for both appendicitis (38%) and pneumonia (34%), and this made up 49% (appendicitis) and 73% (pneumonia) of total unnecessary expenditure. Unnecessary expenditure for doctors' and nurses' services, associated with excessive lengths of stay, made up the second largest share of unnecessary expenditure, accounting for 43% (appendicitis) and 21% (pneumonia). In terms of expenditure for doctors' and nurses' services, 13% (appendicitis) and 9% (pneumonia) of this was unnecessary. Although more than 50% of the expenditure for examinations for appendicitis patients was unnecessary, it made up only about 1% of the total unnecessary expenditure since little was spent on examinations. Unnecessary expenditure for laboratory tests was small in terms of both its contribution to total unnecessary expenditure (1%), and its percentage of expenditure for tests (7%).
The relationship of bonus switch to hospital revenue, cost recovery, productivity and unnecessary care: trend analysis
The above data show that there had been considerable changes over time in hospital revenue and productivity, and also that there was a considerable amount of unnecessary care. But were these features related? This section examines this question through analysis of trends; the subsequent section examines the question through correlation and regression analysis.
Table 4 summarizes the changes in key indicators by hospital and type of bonus switch. When three hospitals changed their bonus system from non-bonus to flat bonus:
Table 4 Changes in revenue, cost recovery, productivity and unnecessary care with changes in bonus system
Name of hospital Rate of change of revenue Rate of cost recovery Visits per admission Admissions per operation DEA efficiency score Rate of unnecessary care
Switch from non bonus to flat bonus
Zhaoyuan Increase Increase Decrease Increase Decrease Increase
Liangshan Increase* Increase* Decrease Increase Decrease Decrease
Changyi Increase Increase* Increase Decrease Decrease Increase
Switch from flat to revenue-related bonus
Zhaoyuan Increase Increase Decrease Decrease Increase Increase
Liangshan Increase* Increase* Decrease Decrease Decrease Decrease
Qixia Increase Decrease Decrease Decrease Increase No change
Changyi Increase Decrease Decrease Decrease Increase Increase
Yanzhou Increase Increase Decrease Decrease Increase Decrease
Weishan Increase Increase Decrease Decrease Increase Increase
Switch from revenue-related to flat bonus
Weishan Decrease Increase Decrease Increase Decrease Decrease
• the rate of growth of revenue increased in all three;
• cost recovery increased in all three;
• two hospitals out of three showed a decrease in the visits/admission ratio, meaning that more patients were admitted out of those attending for outpatient care;
• one hospital out of three showed a decrease in the admissions/operation ratio, meaning that a higher share of inpatients were operated on;
• productivity decreased in all three;
• unnecessary care increased in two out of three.
When hospitals conducted a further switch from flat bonus to revenue-related bonus:
• all six hospitals increased their rate of growth of revenue
• the majority (4/6) showed a decrease in cost recovery
• all six hospitals showed a decrease in the visits/admission ratio
• all six hospitals showed a decrease in the admissions/operation ratio
• five of the six hospitals experienced increases in productivity
• half showed an increase in unnecessary care.
Despite some inconsistent results, likely to reflect location-specific factors, an overall pattern emerges. First, the implementation of a flat bonus where previously there had been no bonus system seemed to be associated with an increase in hospital revenue and cost recovery, a decrease in hospital productivity and an increase in unnecessary care. Second, the implementation of a revenue-related bonus following a flat bonus system appeared to have increased hospital revenue and cost recovery, encouraged higher admissions and operations rates, and increased hospital productivity, with an unclear effect on unnecessary care.
The interrelationships among bonus system, hospital revenue, cost recovery, productivity and unnecessary care
Results of correlation analysis (the correlation coefficients and their p values for the null hypothesis that the correlation coefficients are zero) are shown in Table 5. Bonus type was negatively correlated with DEA efficiency score (p < 0.01) but positively correlated with unnecessary care (p < 0.05), with service revenue (p < 0.01) and the rate of cost recovery (p < 0.01). These results mean that with a change in the bonus payment from non-bonus to flat bonus and to revenue-related bonus, hospital revenue increased and cost recovery improved. However, the improvement in financial status was associated with reductions in hospital productivity and increase in unnecessary care.
Table 5 Correlations among bonus system, productivity, cost recovery and unnecessary care
Type of bonus Efficiency score Service revenue Cost recovery
Efficiency score -0.4075
(0.001)
Service revenue 0.7353 -0.2952
(0.001) (0.001)
Cost recovery 0.3002 -0.2559 0.3328
(0.009) (0.005) (0.0002)
Unnecessary care 0.3502 -0.2989 0.2374 0.0534
(0.012) (0.035) (0.0969) (0.7128)
Sample sizes: n = 50 for the correlation with unnecessary care; for all the others n = 120
The negative correlations between hospital productivity and cost recovery (p < 0.01) and hospital productivity and hospital revenue (p < 0.01) are consistent with the previous trend analysis, which showed an increase in cost recovery (and revenue) but a decrease in hospital productivity over time. Reduction in hospital productivity may have resulted from increased competition between providers, which encouraged county hospitals to provide more drugs and services per patient and to charge them more in order to increase revenue, and/or the increasing cost may have deterred people from seeking care.
Hospital productivity and unnecessary care were negatively correlated (p < 0.05). This relationship implies that if hospitals provide less unnecessary care and the expenditure of patients is therefore less, the hospital will be able to attract more patients. This is theoretically as expected because reduction in price will lead to increase in patient demand.
The relationships were not statistically significant between unnecessary care and hospital revenue, and unnecessary care and cost recovery (p > 0.05 for both). Individual hospital analysis showed that when the bonus switched from non-bonus to flat bonus or from flat bonus to revenue-related bonus, both unnecessary care and cost recovery tended to increase. In theory, there should indeed be a positive relationship between unnecessary care and hospital cost recovery. The reason why the relationship was not statistically significant may be explained by the sample size (only 50 hospital years for the correlation analysis with unnecessary care) and large variations in both unnecessary care and rates of cost recovery.
Table 6 shows the stepwise regression results for hospital revenue. The fitted model was statistically significant at the p < 0.001 level and could explain 86% of the total variation in hospital revenue (R2 = 0.8636). Bonus type was selected into the model and was statistically significant (p < 0.01), explaining about 6% of the total variation in hospital service revenue. On average, as the parameter estimate shows, a switch to a bonus system with an expected stronger economic incentive increased hospital revenue by about two million yuan. The DEA efficiency score was not statistically significant and the unnecessary care indicator was not even selected into the model at the level of p = 0.15, implying that the relationship between revenue and productivity and the relationship between revenue and unnecessary care (as shown in the correlation analysis) are in fact the effect of the bonus type. In other words, bonus type may have affected hospital revenue through increases in hospital productivity and unnecessary care.
Table 6 Factors explaining the variation in hospital revenue (n = 50)
Variable Parameter estimate F P Partial R**2 Model R**2
INTERCEP -99 008 144 63.24 0.0001
Year 1 080 386 56.76 0.0001 0.7084 0.7084
Weishan -4 029 314 13.42 0.0007 0.0708 0.7792
Bonus type 1 887 830 17.38 0.0001 0.0577 0.8369
Qixia 3 540 575 7.68 0.0082 0.0168 0.8357
DEA score 58 905 3.21 0.0800 0.0100 0.8636
Unnecessary care (not selected at P = 0.15 level)
Model: F = 55.73 P = 0.0001 R2 = 0.8636
As the partial R2 shows, the year, which was selected with p = 0.0001, explained 70% of the total variation in hospital revenue. This is not surprising, given the change in revenue over time described earlier. The two hospital dummy variables jointly explained 9% of the revenue variation. These results imply that a dominant proportion of the revenue variation among years and hospitals can be explained by factors related to time and individual hospital characteristics. Although bonus type mattered, it could explain only a small portion of the variation.
Table 7 shows the stepwise regression results for rate of recovery of recurrent cost. The model was statistically significant (p < 0.001) with an R2 of 0.6084. As with the regression on hospital revenue, the DEA efficiency score and unnecessary care were not selected into the model at the p = 0.15 level, and the bonus type was selected with a p value of 0.0129. The bonus type explained more than 3% of the variation in the rate of cost recovery, and on average a bonus switch to one with an expected stronger economic incentive contributed a two percentage-point increase to the rate of cost recovery. In contrast to the regression on hospital revenue, the year variable was no longer statistically significant and a dominant proportion (55%) of the variation in the rate of cost recovery was explained by hospital characteristics.
Table 7 Factors explaining the variation in hospital cost recovery (n = 50)
Variable Parameter estimate F P Partial R**2 Model R**2
INTERCEP 50.4365 1.58 0.2147
Changyi -24.1669 28.67 0.0001 0.1880 0.1880
Zhaoyuan -19.2626 18.93 0.0001 0.1370 0.3250
Qixia -23.7887 21.40 0.0001 0.1486 0.4736
Liangshan -9.9650 5.87 0.0196 0.0818 0.5554
Bonus type 2.1861 6.38 0.0129 0.0306 0.5860
Year 0.6534 2.34 0.1336 0.0224 0.6084
DEA score (not selected at p = 0.15 level)
Unnecessary care (not selected at p = 0.15 level)
Model: F = 12.04 p = 0.0001 R2 = 0.6084
Table 8 shows the stepwise regression results for unnecessary care. The model was statistically significant (p < 0.01) and the R2 of the model was 0.3371. Bonus type was selected as a significant factor (p = 0.05) with a partial R2 of 0.058, meaning that the difference in bonus system explained about 6% of the variation in unnecessary care. As the parameter shows, with a change in bonus type from non-bonus to flat or from flat bonus to revenue-related bonus, the unnecessary care indicator increased by about one percentage point – not as much as had been expected. After controlling for bonus type and hospital characteristics, the year and DEA efficiency score were not selected as significant factors explaining the difference in unnecessary care. The difference in characteristics of individual hospitals explained 27% of the total difference in unnecessary care and more than 60% of the variance was unexplained by this model.
Table 8 Factors explaining the variation in unnecessary care (n = 50)
_Variable Parameter estimate F P Partial R**2 Model R**2
INTERCEP 79.7257 2298.21 0.0001
Liangshan -5.3433 13.05 0.0007 0.1629 0.1629
Zhaoyuan -3.6315 5.52 0.0232 0.1161 0.2790
Bonus type 1.0492 4.04 0.0582 0.0582 0.3372
Year (not selected at p = 0.15 level)
DEA score (not selected at p = 0.15 level)
Model: F = 7.80 P = 0.0003 R2 = 0.3371
Table 9 shows the stepwise regression results for hospital productivity. The fitted model was statistically significant (p < 0.001) with an R2of 0.35. Unnecessary care was selected as a significant factor affecting hospital productivity, and the difference in unnecessary care explained 8.6% of the total variation in hospital productivity. A decrease in unnecessary care was associated with an increase in hospital productivity. This finding, and the fact that the DEA efficiency score was not selected as a significant factor explaining the difference in unnecessary care, imply that the relationship between unnecessary care and hospital productivity is such that the increase in unnecessary care reduced hospital productivity, rather than that the reduction in hospital productivity forced hospitals to provide more unnecessary care. Bonus type in this model was not found to be a significant factor, suggesting that the historical reduction in hospital productivity was not caused by switches in bonus system and that bonus switches failed to improve hospital productivity. Although the descriptive data showed that hospital productivity fell over time, year was not statistically significant in this model.
Table 9 Factors explaining the variation in hospital productivity (n = 50)
Variable Parameter estimate F P Partial R**2 Model R**2
INTERCEP -55.8719 1.08 0.3038
Year -0.7421 2.89 0.0959 0.1075 0.1075
Unnecessary care 0.7010 4.10 0.0489 0.0859 0.1934
Zhaoyuan 14.6042 11.64 0.0014 0.0930 0.2864
Liangshan 6.2018 2.18 0.1468 0.0337 0.3201
Changyi 15.4109 12.83 0.0008 0.0321 0.3522
Bonus type (not selected at p = 0.15 level)
Model: F = 12.05, P = 0.0001, R2 = 0.3522
Discussion
Given the absence of computerized information systems, this study was constrained by what data could feasibly be collected by hand. In particular this limited the study of unnecessary care, which was extremely time-consuming. If inpatient records had covered more years and more hospital years had been employed in the regression analyses, the factors that were not significant might have been statistically significant, and the goodness of fit of the models might have improved. Moreover, only two tracer conditions were studied, and only inpatient care was assessed. These conditions allow for only a limited degree of overprovision, especially in terms of lab tests and drugs, and so are likely to have underestimated the amount of unnecessary care.
Productivity assessment, for both unidimensional ratio analysis and DEA, had two major shortcomings: changes in the quality of care (other than the component of unnecessary care) were ignored, and changes in case mix were not adjusted for. Over time it is likely that the quality of care improved, and that case mix changed; thus this study is likely to have exaggerated the degree to which productivity declined over time. However, these points do not necessarily affect the finding that there was a lack of a statistically significant relationship between bonus type and hospital productivity. It is possible that hospital expansion and increased numbers of staff per hospital may have obscured any increase in quantity stimulated by the bonus payment.
The effect of bonus payment is likely to depend on both the type and the amount of bonus. However, data on the amount of bonus was a sensitive question, and panel hospitals either refused to provide data or provided data that were not considered reliable. In a related hospital census survey [18], it was found that there was no statistically significant difference in the average amount of bonus per doctor across the types of bonus. However, the variation of bonus payment among hospital doctors increased with the progression of bonus types. These findings suggest that exclusion of bonus amount in the analysis may not have introduced much error, because the average amount did not vary much, and it was the way in which the bonus was distributed that mattered.
In the analyses of individual hospitals and of trends, it was found that for some hospitals, the changes in indicators happened in the year of switch and for others the following year. There are two possible reasons. First, when hospitals responded to the bonus switch would depend on whether the bonus switch happened early or late in the year. Second, the speed of effects would also depend on how hospitals responded. Some may have responded quickly by admitting more patients, providing more existing services and prescribing more and costlier drugs. Others may have had to wait for the purchase of equipment (e.g. CT scanner) or the training of personnel (e.g. for open-heart surgery).
It was clear from the analysis that the increase in hospital cost recovery was not a result of improvement in hospital productivity, since between 1978 and 1997, cost recovery increased from 71% to 96% and the DEA efficiency score decreased from 97% to 73%. Apart from the increase in the proportion of unnecessary expenditure and the bonus system changes, there are at least four additional factors that may have affected the increase in hospital cost recovery.
First, over the previous 20 years the Chinese government had been raising medical care prices. Although these on average were set at about 50% of total cost [19], prices were higher relative to their cost than 10 years previously.
Second, due to the liberalization of the pharmaceutical market, the prices of drugs had increased considerably, doubling from 1980 to 1990 [8]. This benefited hospitals, because they were allowed to sell drugs at a 20% mark-up.
Third, the development of new technologies encouraged the introduction of new treatments and drugs that usually had much higher regulated prices than traditional treatments and drugs. For example, before 1980, imported drugs accounted for less than 1% of the Chinese drug market, while by 1993 the sale of imported drugs made up 30%–55% of the market in major cities such as Beijing and Shanghai [20]. Before 1980 there was no CT or MRI in China, while by 1995, CT scanners had became very popular in county hospitals and MRI could be found in any city at and above the prefecture level. Because the prices of high technology services could be set above cost, and the prices for the new imported drugs were 5 to 10 times the prices of the traditional and domestic drugs, hospitals obtained much more profit from using these services and drugs.
Finally, government budget reform increased the financial accountability of public hospitals. Reducing waste and saving costs became a major management concern, helping to improve hospital cost recovery. However, this study suggests that hospital characteristics explained around 55% of the variation in hospital cost recovery, implying that hospital management capacity and ability to control cost varied a great deal.
Between 1978 and 1997, the visits/admission ratio and admissions/operation ratio went down. The analysis showed that a bonus switch from one with a weaker economic incentive to one with a stronger incentive was one of the factors explaining the decreases in the visits/admission ratio and in the admissions/operation ratio, suggesting there must be other factors affecting these ratios. The number of visits to county hospitals was increasing until around the middle of the 1980s, and then started falling. Beginning at the end of the 1970s, rural economic reform brought about a rapid increase in peasants' income, which could have encouraged an increased demand for health care [21]. This may have been felt particularly at higher-level health institutions, such as county hospitals, because of the collapse of the rural Cooperative Medical System and the decrease in the number of rural doctors after the rural economic reform [22]. However, the increase in the number of rural doctors after the mid-1980s, when rural private practice was permitted, may have pulled patients back from county hospitals. In addition, the increase at that time in the number of county level health institutions, such as stations of maternal and child health and hospitals of Chinese traditional medicine, and the increase in medical prices, would have decreased the demand for county hospital care.
It is difficult to explain fully the steady increase in the number of admissions and the number of operations. Since inpatients can be admitted only through the outpatient department and only inpatients can be operated on, it is obvious that the hospitals admitted more and more from among the outpatients and performed more and more operations for the inpatients. There are several possible reasons for this. First, the case mix may have changed so that more patients needed to be admitted and operated on. Second, the development of technology and changes in medical criteria for admissions and operations may have led to more patients' being admitted and operated on. Third, the increase in the numbers of beds and doctors may have permitted needed admissions and operations that had not been possible before due to lack of inputs. Finally, related to the major hypothesis of this research, the changes in doctor payment system may have motivated staff to provide more unnecessary admissions and surgical operations. The study has shown that a bonus switch appeared to bring about an increase in the visits/admission ratio, but case notes did not permit a judgement as to whether an admission was necessary or unnecessary.
Although the DEA efficiency score decreased when panel hospitals switched from non-bonus to flat bonus, this does not mean that the bonus switch helped to decrease hospital productivity. This is, first, because the DEA efficiency score was generally decreasing over time; and second, because in two of the three hospitals (Liangshan and Changyi), the bonus switch appears to have helped to slow down the rate of decrease in the DEA efficiency score.
Conclusions and policy implications
Based on these analyses, we can draw several conclusions. First, there was a steady increase in hospital revenue, and bonus type was a significant factor explaining its variation across hospitals and years.
Second, a considerable proportion of unnecessary expenditure out of total expenditure was identified, and there was a relationship between the bonus system and unnecessary care. Analyses showed that the bonus system was positively correlated with the unnecessary care indicator, implying that the higher the expected incentive of the bonus system, the higher the proportion of unnecessary expenditure.
Third, although hospital productivity decreased over time, a bonus switch from flat bonus to revenue-related bonus appeared to increase hospital productivity in the year of the switch. A bonus switch from non-bonus to flat bonus was not similarly able to reverse the trend of hospital productivity, but it seemed that the rate of decrease in hospital productivity was slowed down by a bonus switch.
Fourth, hospital cost recovery increased over time. The study suggests that the bonus switch brought about an increase in cost recovery and that the bonus system was positively correlated with hospital cost recovery.
Last and in general, the research suggests that the bonus change over time contributed significantly to the increase in hospital service revenue and hospital cost recovery. The increase in unnecessary care and increase in the number of admissions out of the existing number of outpatients, with the bonus system switching from one with a weaker incentive to one with a stronger incentive, suggests that the improvement in hospital cost recovery was achieved at least in part through the provision of more unnecessary care and drugs and through admitting more patients.
There are two policy implications from this study. First, there is little evidence that the performance-related pay system as designed by Chinese public hospitals is socially desirable. It could improve hospital financial sustainability, but did not necessarily lead to improvements in efficiency from a social perspective. The key barrier to achieving the social objectives of performance-related pay was the inappropriate link (whether direct or indirect) between bonus payment and hospital revenue. Hospital bonus distribution should be based on doctor performance measured by indicators that are in line with the desired overall performance of the health care system.
Second, reforms in various countries are characterized by increased exposure of public hospitals to financial risk, in order to increase financial accountability, efficiency and productivity. However there is a risk of encouraging revenue maximization and rent-seeking. Chinese experiences show that when increasing public hospital autonomy, hospitals should be monitored closely by the government, and regulations applied to limit opportunistic behaviour. Otherwise, the containment of government financing to public facilities may result in an increase in the provision of unnecessary care, an increase in health costs to society and a waste in social resources.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
XL designed the study, carried out the fieldwork, undertook the data analysis and wrote the initial draft. AM guided and supervised the design and data analysis and participated in writing and finalizing the paper. Both authors read and approved the final manuscript.
Acknowledgements
This study was funded by support from the UNDP/World Bank/WHO Special Programme for Research and Training in Tropical Diseases, and from the UK Department for International Development (DFID). DFID supports policies, programmes and projects to promote international development and provided funds for this study as part of that objective, but the views and opinions expressed are those of the authors alone. We are grateful for advice on data analysis from Dr Lilani Kumaranayake and for her comments on an earlier draft.
==== Refs
Sharma S Hotchkiss DR Developing financial autonomy in public hospitals in India: Rajasthan's model. Health Policy 2001 55 1 18 11137185 10.1016/S0168-8510(00)00106-8
Preker AS Harding A Innovations in service delivery 2003 Washington DC, World Bank
Health M The resolution of strengthening health work. 1997 Beijing, People's Health Press 9815492
Chang SC Does the hospital responsibility system increase the social benefit of hospital services? Chinese Health Economics 1986 5 52 54
Jiang GH The reasons for the reduction in the effect of bonus system and related countermeasures. Chinese Health Economics 1990 3 21 23
Yi QY The mistakes of hospitals' motivation mechanism. Chinese Health Economics 1990 4 26 27
Wang GZ Lin SY The health economic problems of bonus system. Chinese Health Economics 1992 6 10 11
Li Q The prospect of the Chinese pharmaceutical market in 1996. Chinese Pharmaceutical Information 1997 2 30 37
Bian ZP Reforming hospital management and the income distribution of medical staff. Chinese Health Economics 1994 8 44 45
Zhang MN Lin YY The fairness and efficiency of bonus distribution. Chinese Health Economics 1994 8 46 47
Han T Qiao TY Cui P An effective way of designing the bonus system. Chinese Health Economics 1994 8 48 48
Jiang DJ Reforming the distribution mechanism and increase the activeness of the hospitals. Chinese Health Economics 1995 4 21 22
Liu X Impact of payment reforms to hospital-based doctors on the efficiency of hospital services: a study of county general hospitals in China. PhD thesis. 1999 London, University of London
Sherman HD Hospital efficiency measurement and evaluation. Medical Care 1984 22 922 935 6436590
Valdmanis V Ownership and technical efficiency of hospitals. Medical Care 1990 28 552 562 2355759
Thanassoulis E Emrouznejad A Warwick Windows DEA User's Guide 1998 Warwick, Warwick Business School
Liu X Mills A Evaluating payment mechanisms: how can we measure unnecessary care. Health Policy and Planning 1999 14 409 413 10787657 10.1093/heapol/14.4.409
Liu X Mills A The influence of bonus payments to doctors on hospital revenue: results of a quasi-experimental study Applied Health Economics and Health Policy 2003 2 91 98 14619280
Health HDP The current status of medical price management and its reform. Chinese Journal of Health Economics 1993 2 33 34
Zhang WK The further improvement of drugs administration and inspection. Chinese Pharmacy 1992 6 4 6
Hsiao WC Transformation of health care in China. New England Journal of Medicine 1987 14 932 936
Liu X Cao H The Chinese cooperative medicine: its historical transformation and the future trend of development Journal of Public Health Policy 1992 4 517 526
|
16253137
|
PMC1283747
|
CC BY
|
2021-01-04 16:37:43
|
no
|
Hum Resour Health. 2005 Oct 27; 3:11
|
utf-8
|
Hum Resour Health
| 2,005 |
10.1186/1478-4491-3-11
|
oa_comm
|
==== Front
Nutr Metab (Lond)Nutrition & Metabolism1743-7075BioMed Central London 1743-7075-2-271622974110.1186/1743-7075-2-27EditorialWhen is a high fat diet not a high fat diet? Feinman Richard D [email protected] Department of Biochemistry, State University of New York Downstate Medical Center, Brooklyn, NY 11203 USA2005 17 10 2005 2 27 27 4 10 2005 17 10 2005 Copyright © 2005 Feinman; licensee BioMed Central Ltd.2005Feinman; 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.
==== Body
The observation that a high fat/low carbohydrate (CHO) diet has a beneficial effect on a mouse model of Alzheimer's disease (AD) published today is notable given previous results showing that high fat diets have a deleterious effect on AD. Van de Auwera, et al. [1] reported that mice fed a ketogenic diet (<1% carbohydrate, 80% fat) were found to have a 25% decrease in the protein Aβ42 compared to mice fed a standard high-carbohydrate, low-fat chow diet. Aβ42 is a particularly amyloidogenic mutant form of the amyloid precursor protein whose proteolytic product β-amyloid peptide is contained in the plaques and neurofibrillar tangles that are characteristic of AD.
Any suggestion of an environmental or dietary attack on AD is welcome given the devastating effects of the disease. Beyond the potential application, however, Van de Auwera, et al.'s results have general implications for nutritional approaches in biochemistry and cell biology, and ultimately on disease processes. The physiologic effect of dietary fat can be significantly modulated by the presence of carbohydrate and the associated hormonal changes. The description "high fat diet" is thus an inadequate way to characterize a diet. One must also specify the level of carbohydrate.
The principle that dietary fat might play a relatively passive role in metabolism and that the disposition of fat is regulated by the hormonal state stimulated by carbohydrate is taught in elementary courses in biochemistry but remains an under-appreciated factor in many studies, possibly due to the emphasis on low-fat recommendations of nutritional agencies. Because of the requirement of brain cells for glucose (or ketones) for energy metabolism and, in particular, because of the known involvement of insulin in regulating secretase (proteolytic enzyme in β-amyloid production) (e.g. [2]), it is pertinent to inquire about the role of macronutrient composition in the diet in neuronal disorders. The role of energy metabolism in brain function has been discussed in a recent review and hypothesis by Mukherjee and Seyfried [3].
In explaining the importance of macronutrient composition to students we emphasize energy metabolism and gain or loss of body weight and we stress the need to get away from the principle that "you are what you eat," and replace it with the idea that "you are what you do with what you eat [4]." A common analogy, that fat is the bomb and carbohydrate is the fuse, or in its original description, powder keg and tinder box, may be too broad for appreciation of fine control of metabolism but is probably good enough to illustrate the principle here. Although there are many effects of dietary change, to a first order approximation, carbohydrate is the major stimulus for insulin secretion and as an anabolic hormone, leads to repression of lipolysis and glycogenolysis. Continued hyperinsulinemia, therefore, may predispose to a state where dietary fat is stored rather than oxidized. In addition, current thinking on insulin resistance emphasizes the role of free fatty acids and other fat metabolites (e.g. [5,6]). One theory of the etiology of insulin resistance is that insulin resistance in the adipocyte represents down regulation of response due to continued hyperinsulinemia. This causes increased lipolysis and excessive liberation of fatty acids which may have several effects in peripheral tissues. Thus, the regulation of the TAG-fatty acid axis may be more important than the dietary levels of fat itself.
Similarity of starvation and carbohydrate restriction
One of the areas bearing on the idea that disposition of fat is controlled by insulin is the observation made by several groups that the metabolic response to starvation resembles the response to carbohydrate restriction [7-11]. An important but, in our view, under-appreciated study is that of Klein & Wolfe [11] who compared responses of subjects on an 84 hour fast to the same subjects on a similar fast in which lipids were infused at a level equal to resting energy requirements. Table 1 shows that the levels of fatty acid, rates of oxidation and levels of glucose and ketone bodies were similar in the two groups (Table 1). These rather dramatic results were summarized by the authors as demonstrating that "carbohydrate restriction, not the presence of a negative energy balance, is responsible for initiating the metabolic response to fasting." It might be said that this is the key experiment in understanding the interaction between fat and carbohydrate. Similarly, Bisschop, et al. [8] demonstrated a similarity between high carbohydrate, low fat diets (CHO:Lipid:Protein = 85:0:15) and control (44:41:15) diets in FFA rate of appearance and oxidation but significant differences with a low carbohydrate, high fat diet (CHO:Lipid:Protein = 2:83:15).
Table 1 Similarity of starvation and carbohydrate restriction
Reference [11] FFA (μmol/l) fat oxidation (μmol/kg/min) Glucose (mg/dl) β-hydroxybutyrate (mM)
Fast 84 h 0.92 1.94 68 2.56
Fast + Lipid 1.02 1.67 66 2.54
Data from reference [11]. Procedure as described in the text.
High fat diets in obesity
The interaction of fat and carbohydrate bear on the mechanism of weight loss strategies based on carbohydrate restriction. In considering the problem, it is important to recognize that percentages are misleading. There are really three degrees of freedom in design or analysis of a weight loss experiment: two of the three macronutrients and the total caloric intake. It is unlikely that the percentage rather than the absolute amount of macronutrients is the controlling variable and at least three published studies show that carbohydrate reduction is not necessarily accompanied by replacement with either fat or protein but rather caloric reduction due to the carbohydrate removed [12-14]. Such diets are effectively high fat by percentages but lead to substantial weight loss.
To understand the relative impact of carbohydrate and fat, one has to consider experiments in which one macronutrient is replaced by another. Several studies in the literature have demonstrated that in isocaloric comparisons, the replacement of dietary carbohydrate with fat leads to greater weight loss [15-19], that is, the absence of the fuse prevents an explosive effect of dietary fat on body mass. The limitations of the fat bomb per se are reviewed by Willett and Leibel [20].
Cardiovascular disease
Many of the objections over the years of controlling the effect of dietary fat by replacing dietary carbohydrate has centered on the "obvious" danger that high fat diets present to plasma lipids and coronary risk. The subject is more complicated than obesity in that the mechanistic role of insulin is less well understood and the type of fat becomes much more important. Moreover, the relative importance of different risk factors LDL vs. triglycerides and HDL may be a matter of clinical judgment. However, what we know from epidemiology is that, whereas replacing unsaturated fat (UF) with saturated fat is deleterious to CVD outcome, replacing UF with carbohydrate is worse [21]. Similar effects are seen on the effect of dietary replacement on total cholesterol/HDL [22]. The generally beneficial effects of carbohydrate reduction on lipid profile has been reviewed by Volek [23].
Diabetes
Control of lipid metabolism by insulin is of greatest importance in diabetes where insulin insufficiency or resistance is the key variable. The subject is a matter of current controversy since official organizations counsel against high-fat diets: the American Diabetes Association position statement, for example, presents an expert consensus for 60–70% of energy intake as carbohydrates for diabetics. Such high intake of the major insulin secretagogue as a therapeutic strategy remains counter-intuitive and may not even be known to or implemented by many practicing physicians and the idea has recently been strongly criticized [24,25].
A widely quoted example of experiments bearing on this question is the study of Garg, et al. [26,27]. In a four-center randomized crossover trial, carbohydrate was replaced with monounsaturated fats (CHO: 55% → 40; total fat: 30% → 45) or, conversely, MUF was replaced with CHO. It was found that replacement of MUF with carbohydrate "caused persistent deterioration of glycemic control and accentuation of hyperinsulinemia, as well as increased plasma triglyceride and very-low-density lipoprotein cholesterol levels, which may not be desirable [26]," that is, the high fat arm was beneficial, again, suggesting that what happened to the fat was more important than its concentration in the diet.
To determine the relative importance of fat and carbohydrate and protein, Gannon and coworkers have measured the effect of replacement of carbohydrate with protein and fat in patients with type 2 diabetes [28]. Figure 1 shows the effects on glycemic control of the replacement of carbohydrate with protein and fat. Again, the 50% fat diet provides better glycemic control although, in this case, the relative contributions of carbohydrate reduction and increase in protein is unknown.
Figure 1 Effect of diet on glucose. Mean plasma glucose concentration before (triangles) and after 5 weeks on control diet (yellow circles: (CHO:fat:protein = 55:30:15)) or 5 weeks on the higher fat diet (blue circles: (20:50:30)). Meals are Breakfast (B), lunch (L) and dinner(D) plus 2 snacks (S1, S2). Data from reference [28].
Conclusion
The long range implication of Van de Auwera's study for AD remains to be seen but the general lesson is that in dietary recommendations or in testing animal models a diet should not be characterized as high fat without also specifying the level of carbohydrate.
==== Refs
Van der Auwera I Wera S Van Leuven F Henderson ST A ketogenic diet reduces amyloid beta 40 and 42 in a mouse model of Alzheimer's disease Nutr Metab (Lond) 2005 2 28 16229744 10.1186/1743-7075-2-28
Solano DC Sironi M Bonfini C Solerte SB Govoni S Racchi M Insulin regulates soluble amyloid precursor protein release via phosphatidyl inositol 3 kinase-dependent pathway Faseb J 2000 14 1015 1022 10783157
Mukherjee P Seyfried TN Metabolic Targeting of Brain Cancer Nutr Metab (Lond) 2005 2 30 16242042 10.1186/1743-7075-2-30
Feinman RD Makowske M Metabolic Syndrome and Low-Carbohydrate Ketogenic Diets in the Medical School Biochemistry Curriculum. Metabolic Syndrome and Related Disorders 2003 1 189 198 10.1089/154041903322716660
Boden G Laakso M Lipids and glucose in type 2 diabetes: what is the cause and effect? Diabetes Care 2004 27 2253 2259 15333497
Kim JK Fillmore JJ Sunshine MJ Albrecht B Higashimori T Kim DW Liu ZX Soos TJ Cline GW O'Brien WR Littman DR Shulman GI PKC-theta knockout mice are protected from fat-induced insulin resistance J Clin Invest 2004 114 823 827 15372106 10.1172/JCI200422230
Azar GJ Bloom WL Similarities Of Carbohydrate Deficiency And Fasting. II. Ketones, Nonesterified Fatty Acids And Nitrogen Excretion Arch Intern Med 1963 112 338 343 14045277
Bisschop PH Ackermans MT Endert E Ruiter AF Meijer AJ Kuipers F Sauerwein HP Romijn JA The effect of carbohydrate and fat variation in euenergetic diets on postabsorptive free fatty acid release Br J Nutr 2002 87 555 559 12067425 10.1079/BJNBJN2002578
Bloom WL Azar GJ Similarities Of Carbohydrate Deficiency And Fasting. I. Weight Loss, Electrolyte Excretion, And Fatigue Arch Intern Med 1963 112 333 337 14045276
Fery F Bourdoux P Christophe J Balasse EO Hormonal and metabolic changes induced by an isocaloric isoproteinic ketogenic diet in healthy subjects Diabetes Metab 1982 8 299 305 6761185
Klein S Wolfe RR Carbohydrate restriction regulates the adaptive response to fasting Am J Physiol 1992 262 E631 6 1590373
Boden G Sargrad K Homko C Mozzoli M Stein TP Effect of a low-carbohydrate diet on appetite, blood glucose levels, and insulin resistance in obese patients with type 2 diabetes Ann Intern Med 2005 142 403 411 15767618
Larosa JC Fry AG Muesing R Rosing DR Effects of high-protein, low-carbohydrate dieting on plasma lipoproteins and body weight J Am Diet Assoc 1980 77 264 270 7410754
Miller BV Bertino J Reed TG Burrington C Davidson LK Green A Gartung A Nafziger A An Evaluation of the Atkins' Diet Metabolic Syndrome and Related Disorders 2003 1 299 309 10.1089/1540419031361426
Baba NH Sawaya S Torbay N Habbal Z Azar S Hashim SA High protein vs high carbohydrate hypoenergetic diet for the treatment of obese hyperinsulinemic subjects Int J Obes Relat Metab Disord 1999 23 1202 1206 10578211 10.1038/sj.ijo.0801064
Golay A Eigenheer C Morel Y Kujawski P Lehmann T de Tonnac N Weight-loss with low or high carbohydrate diet? Int J Obes Relat Metab Disord 1996 20 1067 1072 8968851
Rabast U Kasper H Schonborn J Comparative studies in obese subjects fed carbohydrate-restricted and high carbohydrate 1,000-calorie formula diets Nutr Metab 1978 22 269 277 662209
Sharman MJ Kraemer WJ Love DM Avery NG Gomez AL Scheett TP Volek JS A ketogenic diet favorably affects serum biomarkers for cardiovascular disease in normal-weight men J Nutr 2002 132 1879 1885 12097663
Volek JS Sharman MJ Gomez AL Judelson DA Rubin MR Watson G Sokmen B Silvestre R French DN Kraemer WJ Comparison of energy-restricted very low-carbohydrate and low-fat diets on weight loss and body composition in overweight men and women Nutr Metab (Lond) 2004 1 13 15533250 10.1186/1743-7075-1-13
Willett WC Leibel RL Dietary fat is not a major determinant of body fat Am J Med 2002 113 Suppl 9B 47S 59S 12566139 10.1016/S0002-9343(01)00992-5
Hu FB Manson JE Willett WC Types of dietary fat and risk of coronary heart disease: a critical review J Am Coll Nutr 2001 20 5 19 11293467
Mensink RP Zock PL Kester AD Katan MB Effects of dietary fatty acids and carbohydrates on the ratio of serum total to HDL cholesterol and on serum lipids and apolipoproteins: a meta-analysis of 60 controlled trials Am J Clin Nutr 2003 77 1146 1155 12716665
Volek JS Sharman MJ Forsythe CE Modification of lipoproteins by very low-carbohydrate diets J Nutr 2005 135 1339 1342 15930434
Arora SK McFarlane SI The case for low carbohydrate diets in diabetes management Nutr Metab (Lond) 2005 2 16 16018812 10.1186/1743-7075-2-16
McFarlane SI Dietary recommendations for people with diabetes: Time to reduce the carbohydrate loads Endocrine Today 2005 3 4 5
Garg A Bantle JP Henry RR Coulston AM Griver KA Raatz SK Brinkley L Chen YD Grundy SM Huet BA Effects of varying carbohydrate content of diet in patients with non-insulin-dependent diabetes mellitus Jama 1994 271 1421 1428 7848401 10.1001/jama.271.18.1421
Garg A Grundy SM Koffler M Effect of high carbohydrate intake on hyperglycemia, islet function, and plasma lipoproteins in NIDDM Diabetes Care 1992 15 1572 1580 1468287
Gannon MC Nuttall FQ Effect of a high-protein, low-carbohydrate diet on blood glucose control in people with type 2 diabetes Diabetes 2004 53 2375 2382 15331548
|
16229741
|
PMC1283748
|
CC BY
|
2021-01-04 16:37:46
|
no
|
Nutr Metab (Lond). 2005 Oct 17; 2:27
|
utf-8
|
Nutr Metab (Lond)
| 2,005 |
10.1186/1743-7075-2-27
|
oa_comm
|
==== Front
Aust New Zealand Health PolicyAustralia and New Zealand Health Policy1743-8462BioMed Central London 1743-8462-2-251625313210.1186/1743-8462-2-25Research"There's no place like home" A pilot study of perspectives of international health and social care professionals working in the UK Moran Anna [email protected] Susan [email protected] Allister [email protected] Trent RDSU, Institute of General Practice and Primary Care, School of Health and Related Research, The University of Sheffield, ICOSS Building, 219 Portobello, S1 4DP, Sheffield, UK2 Faculty of health, Canterbury Christ Church University College, Kent, UK2005 27 10 2005 2 25 25 3 8 2005 27 10 2005 Copyright © 2005 Moran et al; licensee BioMed Central Ltd.2005Moran 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 countries are reporting health workforce shortages across a range of professions at a time of relatively high workforce mobility. Utilising the global market to supply shortage health skills is now a common recruitment strategy in many developed countries. At the same time a number of countries report a 'brain drain' resulting from professional people leaving home to work overseas. Many health and social care professionals make their way to the UK from other countries. This pilot study utilises a novel 'e-survey' approach to explore the motives, experiences and perspectives of non-UK health and social care professionals who were working or had worked in the UK. The study aims to understand the contributions of international health and social care workers to the UK and their 'home' countries. The purpose of the pilot study is also in part to test the appropriateness of this methodology for undertaking a wider study.
Results
A 24-item questionnaire with open-ended and multiple choice questions was circulated via email to 10 contacts who were from a country outside the UK, had trained outside the UK and had email access. These contacts were requested to forward the email to other contacts who met these criteria (and so on). The email was circulated over a one month pilot period to 34 contacts. Responses were from physiotherapists (n = 11), speech therapists (n = 4), social workers (n = 10), an occupational therapist (n = 1), podiatrists (n = 5), and others (n = 3). Participants were from Australia (n = 20), South Africa (n = 10), New Zealand (n = 3) and the Republic of Ireland (n = 1). Motives for relocating to the UK included travel, money and career opportunities. Participants identified a number of advantages and disadvantages of working in the UK compared to working in their home country health system. Respondents generally reported that by working in the UK, they had accumulated skills and knowledge that would allow them to contribute more to their profession and health system on their return home.
Conclusion
This pilot study highlights a range of issues and future research questions for international learning and comparison for the health and social care professions as a result of international workforce mobility. The study also highlights the usefulness of an e-survey technique for capturing information from a geographically diverse and mobile group of professionals.
==== Body
Background
World over, countries use strategies to manage health workforce shortages such as improving retention, attracting 'non-traditional' entrants and attracting back 'returnees' [1]. International recruitment is another popular solution to overcoming shortfalls of health providers [2-6]. Recruitment strategies aside, health workforce mobility is increasing, particularly the flow of health and social care professionals to and from the UK.
Health and social care professionals in the UK
Most of the available data surrounding health worker migration relates to the nursing and medical professions. For instance, in the UK and Ireland there were more overseas additions to the in 2000/2001 UK nursing register than home country registrants [1] and two thirds of new registrants to the UK General Medical Council in 2003 were from overseas, mostly from outside the European Economic Area [7]. Less research has been undertaken into the mobility of the allied health and social care workforce.
During the 1990s, registration of non-UK trained physiotherapists ranged from 26 to 42 percent; sourced mainly from Australia, South Africa and New Zealand [8]. The Health Professions Council (HPC), the regulatory body responsible for registration of allied health professionals and clinical scientists in the UK, captures data on applications for registration by profession and country of origin. In the 2004 / 05 financial year there were 3,515 international registrations. Nearly one third of these (n = 1,339) were physiotherapists, followed by radiographers (n = 681), occupational therapists (n = 668) and biomedical scientists (n = 363). These reflect the relative proportions of the total registrations for the respective professions. The major donor 'continents' during the same period were Europe (excluding the UK) (n = 129), Africa (n = 110), Asia (n = 107), followed by Oceania (n = 105) (which includes Australia and New Zealand) [Source: HPC 2005].
Australian and New Zealand diaspora
The Australian and New Zealand diasporas have similar histories. It is estimated that one million (out of twenty million) Australians are overseas at any one time [9] and the primary destination of resident professionals leaving Australia and New Zealand on a permanent or long-term basis is the United Kingdom [9-11]. In addition, recent changes to international immigration laws for skilled professionals has made professional migration more seamless [9,11].
'Brain drain'
Not surprisingly, the concept of 'brain drain' resulting from professionals leaving their country of residence to work overseas has created great concern for 'source' or 'donor' countries, who also report shortages of health and social care professionals [9,12-15]. Brain drain is deemed particularly pernicious in developing nations, particularly African countries, and compounded by the fact that emigrating skilled workers are more likely to stay in their host country [6]. It is well documented that recruitment of health personnel from developing countries threatens the operation of crucial health programs in these countries[3].
'Brain circulation'
In contrast, recent research indicates international migration of skilled workers is often temporary and the mobility of this workforce generates global benefits by improving knowledge flows and satisfying the demand for skills, often termed 'brain circulation' [9,11,13]. However to date there is little research specifically analysing the benefits of skilled migration in health and social care.
Push and Pull factors
It has been suggested that push and pull factors motivate workers to leave one country and seek employment in another [8,16] where push factors are motives to leave home countries such as low pay, limited career opportunities, unemployment or civil unrest and pull factors are motivations or conditions that attract migrants to other countries such as demand for workers or a higher standard of living. This idea has been further developed to categorise workers as permanent or temporary movers based on their motives for leaving their home country [16]. For example permanent movers may be 'economic migrants' who are attracted to better standards of living and who may send money to their home country or 'career migrants' who are attracted to enhanced career opportunities. Temporary movers include those on a 'working holiday' where expertise is used to finance travel, or 'the study tour' where new knowledge and techniques are acquired for use when they return home.
Westcott and Whitcombe [17] suggest that the benefits offered by the globalisation of occupational therapy have not been fully realised, particularly in reference to education. One study has been conducted to inform writers of UK speech and language therapy curricula by utilising the perspectives of current and past international students [18].
The motivations, experiences and perspectives of international health and social care professionals however have not been thoroughly explored, particularly in reference to workforce dynamics, workforce flexibility [19-23] and the understanding and management of the global flow of health and social care workers. This pilot study aims to capture preliminary data on the perspectives and motives of health and social care professionals of non-UK origin and training who have at some stage moved to and worked in the UK.
Methods
The pilot study used a novel approach to data collection to capture a range of views from health and social care workers on their experiences of working in the UK. The data collection relied on the assumption that many foreign trained workers living in the UK have access to email and that many international workers have wide international networks.
As a result, we piloted the use of an 'e-survey' which was distributed by the researchers via email to 10 contacts inviting them to participate on the basis that they had worked in the UK at some stage. The participants represented a range of disciplines, having worked in different settings (hospital, community etc) and different geographic locations across the UK. The first round participants were accessed through personal and professional networks. Each of those contacts was asked to forward the e-survey to their international contacts and so on creating a snowball sampling effect. We allowed a four week time frame for replies. This method was chosen over other more traditional survey methods as it was a quick and inexpensive way to gain preliminary insight into an internationally diverse and geographically varied group of health and social care workers. The purpose of the pilot study was, in part, to test the appropriateness of this methodology for undertaking a wider study.
Inclusion criteria were health and social care professionals from any country other than the UK, who trained outside the UK and who have previously worked or are currently working in the UK. UK nationals and UK trained health and social care professionals were excluded. The intention of the research was to capture information from this pool of relatively mobile international health and social care staff.
The 24 item questionnaire included closed and open-ended questions and was designed to capture demographic information about the participants, their profession, perceptions of training, career development opportunities and learning experiences. It was piloted with four expatriate allied health professionals resulting in the deletion of two questions.
The survey was initially circulated as a Microsoft Word attachment. However feedback from some participants indicated that it was not always possible to download or open the attachment, so the survey was then embedded into the body of the email. The researchers established a web-site on which detailed information about the research was available, including the protocol and details of the researchers and from which further questionnaires could be downloaded. All but two surveys were returned electronically. These were faxed and posted to the researchers.
All data were entered into a Microsoft Excel spreadsheet. Numeric data have been presented in descriptive numerical form. Where large amounts of qualitative data were received, they have been summarised for this paper. For the purpose of this paper, the phrase 'country of origin' refers to the nationality of the professional.
Results
Response
Ten e-surveys were emailed in the first round and responses were received from 34 participants within the one month pilot, of which one was incomplete. The profile of the respondents is summarised in Table 1.
Table 1 Profile of Respondents
N %
Responses Complete 33
Incomplete 1
Profession Physiotherapy 11 32
Social Work 10 29
Podiatry 5 15
Speech and Language Therapy 4 12
Occupational Therapy 1 3
Nurse 1 3
Medical Doctor 1 3
Medical Transcript Editor 1 3
Age Mean 31 (25–55), Median 28
Country of Origin Australia 20 60
South Africa 10 30
New Zealand 3
Rep of Ireland 1
Number of years qualified Mean 7 (2–19), Median 4.5
Few respondents reported difficulties obtaining a UK visa, the majority using a working holiday visa (available to Commonwealth citizens aged 17–30) or work permit. Professionals had worked in a median number of 2 cities (range 1–15, mean 3) and usually as a 'locum' where locum work is defined as temporary or contractual employment. The median time spent in the UK was 3 years (range 3 months – 8 years, mean 2.5 years). Employment was typically gained through a recruitment agency in the UK or in their home country. There were 27 professionals still in the UK, most of whom were working in Social Services (n = 9), for the NHS full time (n = 8) or as a locum (n = 7) or in private practice (n = 6). Over half of the respondents (67%, n = 23) reported they would not stay in the UK permanently. Table 2 summarises these results.
Table 2 Work details of respondents
N %
Number of locations worked Mean 3 (1–15), Median 2
One location 14
Two locations 6
Three locations 5
> Three locations 9
Main location in UK London 9 26
Essex 6 17
Sheffield 3 8
Oxford 3 8
Other (Edinburgh, Worcester, Cambridge, Nottingham, Clacton-on-sea, Cardiff, Gosport, Loughton, Halifax, West Grinstead, Rotherham, Falkirk)
Visa difficulties Yes 6 18
No 28 82
Visa Type * Working Holiday 13
Work Permit / sponsorship 11
Ancestry 6
British Passport 4
Highly skilled migrant programme 1
Other 2
Means of securing job in UK+ Agency in home country 10
Locum Agency 6
Agency in the UK 15
Advertisement in the home country 4
Advertisement in the UK 6
Word of mouth 7
Other 1
Why work in the UK$ Travel 29
Money 23
Career 16
Partner 3
Other 4
Areas worked in UK$ Locum 21
NHS 15
Social Services 10
Private practice 7
Research 5
Self Employed 4
Teaching 2
Current area of work in UK$ Locum 7
NHS 8
Social Services 9
Private 6
Research 4
Self Employed 2
Teaching 1
N/A 5
Incomplete 1
* May have held more than one visa
+May have held more than one job
$ Multi-answer question
Motivation to work in the UK
Respondents were asked why they initially chose to work in the UK. Travel (n= 29), money (n = 23) and career opportunities (n = 16) were the primary motives expressed. One respondent answered ...to experience living in a country other than my home country.' (Physiotherapist, Australia)
Expectations prior to working in the UK
Respondents were asked what their expectations of working in the UK were prior to their arrival and how they compared to their experience. There were mixed responses, many reporting they believed the UK would be superior to their home country in terms of resources, professional expertise and funding (n = 5) ; others assumed it would be the same (n = 15).
Thought it would be similar to Australia – maybe not quite so advanced with their techniques. Experience was pretty much what I thought it would be – depends on the different hospitals, which is the same back home. (Speech and language therapist, Australia)
I thought that the workforce would be more superior and be able to provide good guidance. I expected there to be far more resources to enable service users to achieve an element of self actualization. Experience: poor management, poor team and case planning, lack of resources and especially money to provide for the needs of service users. (Podiatrist, Australia)
I just expected to work and earn money to travel but in reality it is a time when you can really work on your professional development, which is what I am doing. (Physiotherapist, Australia)
Skills and Training
Respondents were asked what skills or training could have better equipped them to work in the UK and to describe their perceptions of the quality of training in the UK. Many felt they had adequate skills and training to work in the UK (n = 15) but highlighted greater knowledge of the health and social care systems would have been beneficial particularly those working in social services (n = 6).
My training and skills were of English standards. The only adjustments I had to develop skills around was to know the culture of the community and adjust strategies of intervention. (Speech and language therapist, Australia)
Most perceived that the undergraduate training in the UK was of a lower standard than their country of origin (n = 23) but that opportunities for continuing professional education were superior in the UK (n = 10). Some felt that their undergraduate training equipped them more thoroughly to enter the workforce with more confidence in their role than their UK counterparts.
I believe the broad 4 year undergraduate training in Australia is of very high quality. This allows for multi-skilling and confidence from day one. Therapists trained in Australia tend to be a lot more confident in their skills and are more used to working within multidisciplinary teams. (Occupational therapist, Australia)
Others felt that each system had its advantages.
The UK training is more practice based and reflective. SA training is more theory based. (Social worker, South Africa)
UK pre-qualifying training inferior due to shorter course length. Good first year graduate program compensates for this. (Physiotherapist, New Zealand)
Attractiveness of working in the UK
Respondents were asked what was good about working in the UK. The most attractive features were greater access to Continuing Professional Development (CPD), wider variety of specialisation, more career opportunities and a well-defined career structure. Experiencing a different system and culture was also a theme as well as travel and greater earning power.
Accessibility to latest research and professional development (remote location of many Australian practices limits this). Number of jobs available with very acute caseload (far fewer in Aust). Opportunity to develop quickly as a therapist given our undergraduate skills and general confidence as therapists. Close proximity to Europe and travel opportunities. (Speech and language therapist, Australia)
Difficulties working in the UK
Respondents were asked what was not good about working in the UK. The most unappealing features included large waiting lists and correspondingly large caseloads, poor recognition or respect as a professional, the bureaucracy, the weather, or for some professional groups, racism.
Understaffing of all health professions; Too much paperwork and repetition of paperwork; Reduced hospital standards; Distance from home. (Podiatrist, Australia)
Some people find it difficult to accept that although not trained in the UK the level of skill you bring into the profession is of high value. (Social worker, South Africa)
Participants were also asked how the status of their profession in the UK compared to their country of origin. The majority (n = 24) felt the status of their profession was lower in the UK than in their home country compared to eight who felt it was the same.
I expected to retain the same high status that my profession has back home with other professionals and the communities who appreciate the services provided by the profession. This was completely opposite when I got here (UK) and shocking to me. This discrepancy creates difficulties in working with partners to bring about desired change. (Social worker, South Africa)
Benefits and suggestions for country of origin
When asked how their country of origin could benefit or learn from their experience of working in the UK, most responded they had gained a much broader skill base and knowledge of how a different system works.
I have such a vast array of experiences now to draw on, both good and bad which I can take home with me. I think I am much more worldly now. (Physiotherapist, New Zealand)
I have experienced many management styles, and government agendas, and would be able to take the advantages and disadvantages of these systems back to Aust. and formulate better solutions to problems. (Speech and language therapist, Australia)
Due to excellent continuing education [in the UK] I feel I will have a more up to date knowledge base which I will attempt to pass on when I begin working again at home. (Physiotherapist, Australia)
Discussion
Method
The e-survey technique was chosen over other more traditional survey techniques as it is a quick and cost-effective way to gain preliminary insight into a geographically and demographically diverse group of professionals. The mobility of the diasporas makes them a particularly difficult group to access in a systematic way and, as this study has demonstrated, health and social care workers may work in a number of different cities during their time away. Additionally the e-survey was deemed the most appropriate method given that this project was proposed as a pilot study with the intention to trial the methodology to identify pertinent themes for future research, rather than generate statistically generalisable findings.
The e-survey technique provided an opportunity to capture a global 'before and after' perspective for many professionals who were now living and working at 'home', in a new city or another new country. These professionals may have proved more difficult to sample using a professional register such as the Health Professions Council which captures 'inflow' registration information only[8]. In order that a larger, more systematic e-survey be repeated, electronic networks for diasporas which exist in some countries [24] could be accessed. Alternatively, professional newsletters and journals may facilitate a more targeted approach to specific disciplines or groups.
The limitations of the e-survey approach include the inability to follow-up non-respondents, as once the initial round of surveys had been circulated the researchers had no control over the distribution network. For the same reason, no response rate can be calculated as the denominator is unknown.
Additionally, it was difficult to avoid the potential sources of bias inherent in this type of study. For instance, by circulating the survey electronically, we could only access those workers who use email, and there was no way of knowing how the IT literacy of the respondents impacted on the response rate. The selection of the initial ten respondents may also have introduced bias, however the researchers reiterate this intention of this pilot study was to identify future research themes.
Push and Pull factors
The primary motivations for this group of health and social care professionals to work in the UK were travel, money and career opportunities. These motivations fit with research conducted by Buchan [16], whereby most respondents in this study were 'temporary movers', pursuing 'the working holiday'. This was particularly true for those from Australia and New Zealand. Equally, respondents were relatively young. These findings correspond to research conducted by Birrell et al [9] who reports 70% of Australian professionals who work overseas return and are usually aged between 20 and 30. Also demonstrated in this study was a trend for internationally trained health workers to fill temporary, locum positions in the NHS. A report for the UK Chartered Society of Physiotherapists [8] recognises a key 'pull' factor for overseas physiotherapists is the relative ease with which they can find comparatively well paid temporary work in the UK, giving them greater choice over the location and duration of employment. These findings are supported by Allan and Larsen [26] for international nurses and O'Hagan [26] for Australian medical radiation graduates. The results also reinforce the perception that South Africans are more likely to be 'permanent movers', a trend recognised by Cerventes [6].
Perceptions and Experiences
When examining the perceptions and experiences of the respondents, it is important to remember their original motivations to move to the UK and how this may affect their experiences and perceptions. For example the majority of respondents were motivated to move to the UK by the opportunity to travel. Perceptions may then be from the perspective of a holiday maker, working to fund travels rather than from the perspective of a full time employee working to pay a mortgage, for example. Motives can also change over time depending on different circumstances, for example the motives of and incentives for migrating nurses to the UK have been shown to change over time as personal and socio-economic conditions alter [25]. Respondents' experience of their 'home' health system and organisational culture would also significantly contribute to the forming perceptions and opinions about the UK health organisation and culture. Additionally respondents have experienced a mixture of 'British' culture and NHS organisational culture which together have influenced their perceptions and opinions.
Resources
The qualitative responses demonstrate themes of dissatisfaction and discontent with NHS bureaucracy and lack of resources. This has also been reported in the UK with claims that people leave the UK public sector primarily due to bureaucracy and paper work, lack of resources, lack of autonomy and feeling undervalued[27]. Similarly, a cohort of UK physiotherapy students and professionals perceived physiotherapy in the NHS to have high levels of stress and workload, staff shortages and poor equipment [28].
Undergraduate skills and training
There was a clear perception that undergraduate training is comparatively better outside the UK. This may in part be explained by a discrepancy in the length of training undertaken. South African trained social workers have traditionally had a longer programme of training at undergraduate level, and their education has been granted the status of full degree for longer than their UK equivalents [29]. This is also true for Australian allied health and social work undergraduate degrees, many of which are 4 year qualifications compared to the 3 year UK equivalent [30]. As these professionals have not directly experienced training in the UK, these findings need to be interpreted cautiously. Given these perceptions, it is not surprising respondents felt their skills and training gained at 'home' adequately equipped them to undertake work in the UK health and social care sectors. Further research comparing these perceptions to UK trained professionals working in Australia, South Africa or New Zealand would add value to these findings.
Continuing Professional Development
Many of the respondents reported on the 'value adding' attributes of practicing in the UK. These included the extensive opportunity for post qualifying training or Continuing Professional Development (CPD), a more defined and progressive career structure and greater availability of specialisation. One study confirms these perceptions, reporting that UK physiotherapists and prospective UK physiotherapy students perceive that the career structure in the NHS and variety in work are desirable qualities of NHS physiotherapy as a career [28].
In many cases local health authorities in the UK offer financial contributions towards tuition fees as well as protected time to pursue both academic coursework and CPD. The growth in and support for CPD in Britain is in part due to explicit Department of health NHS policies and frameworks which outlined the need for delivery of high quality care and clinical excellence in the NHS [31,32]. Although Australian professional bodies are equally as attentive to CPD [30], improved access to CPD in the UK may also be explained by the size of the British health and social care system proportionally providing greater numbers of courses for a larger health workforce. Further research is needed to explore this possibility. A smaller and more convenient geographic area for accessing CPD in the UK compared to remote, rural areas of Australia or South Africa may also be a contributing factor. South Africa however has not yet developed a coherent curriculum that focuses on CPD for social work graduates [29].
Professional Status
Undergraduate educational differences may also speak of the issue of professional identity and the reported contrast between professional status of health and social care work in the UK and other countries. Turner [33] compared the status of physiotherapy in Australia to the UK finding Australian general public and physiotherapy students perceive physiotherapy to have a higher occupational prestige than their UK equivalents. The health and social care professionals who responded to our study indicated that they entered the UK system from a very different professional perspective, accounting in part for the difference in perceived professional image. Another study [28] of UK physiotherapy students and professionals, showed that they perceive the general public and other health care professionals to have a lack of recognition for physiotherapy. Negative perceptions of professional status have also been noted within podiatry in the UK [34-37], the USA [39] and, to some extent, Australia [36]. An interesting discrepancy therefore emerges, whereby the perception of lower professional status in the UK is reported along side the perception of improved pay and career opportunities.
International information sharing
This pilot study has highlighted the extent to which international information sharing and collaboration can benefit both 'home' and the UK, particularly in reference to service development, education and career development. The extent to which this mutual learning is realised is dependent on how accessible international registration is [23]; how compatible training and education are; the degree to which health and social care communities embrace the knowledge and expertise foreign workers offer their country and the wisdom returning professionals bring home.
Implications for future research
Many themes have emerged from this study that give rise to further research questions. Of particular interest is the relationship between status, pay and career opportunities in the UK and other countries; the effects of different undergraduate training on accessibility and quality of a 'global' health workforce; and the effect of culture on health and social care systems. Further studies utilising a larger sample size may aid in exploring these themes.
Conclusion
This pilot study demonstrates that international health and social care professionals who have worked in the UK have accumulated vast amounts of experience and knowledge. The study captured the perspectives and experiences of a group of professionals who have gained experience working in different health systems and cultures. This combined with a high percentage of those professionals returning to their country of origin, makes for a new class of highly resourceful and skilled professionals, with global ideas and resources to share with colleagues. It would be valuable to further pursue in depth what this growing group of internationally skilled health and social care professionals offer both their country of origin and the UK. As one respondent commented:
It's a bit like doing rotations to different departments but to different countries.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AM and SN contributed jointly to the research design and data collection. Both authors were involved in the analysis, interpretation of data and preparation of the final manuscript. AB was consulted about the design of the research, provided literature, access to networks of clinicians and South African perspective.
Acknowledgements
The authors would like to thank all of the participants who took part in this study.
==== Refs
Buchan J Nurse migration and international recruitment Nursing Inquiry 2001 8 203 204 11844041 10.1046/j.1440-1800.2001.00112.x
Buchan J International rescue? The dynamics and policy implications of the international recruitment of nurses to the UK. Journal of Health Services Research and Policy 2004 9 Suppl 1 10 116 15006223 10.1258/135581904322724086
Scott ML Whelan A Dewdney J Zwi AB "Brain drain" or ethical recruitment? Solving health workforce shortages with professionals from developing countries Medical Journal of Australia 2004 180 174 176 14960139
Brush BL Berger AM Sending for nurses: foreign nurse migration Nursing and Health Policy Review 2002 1 103 115
Brown RP Connell J The migration of doctors and nurses from South Pacific Island Nations Social Science and Medicine 2004 58 2193 2210 15047077 10.1016/j.socscimed.2003.08.020
Cervantes M Guellec D The brain drain: Old myths, new realities OECD Observer 2002
Buchan J International recruitment of health professionals British Medical Journal 2005 330 210 15677637
Buchan J O'May F International recruitment of physiotherapists 2000 Edinburgh, Queen Margaret University College 38
Birrell B Dobson I Rapson V Smith TF Skilled Labour: Gains and losses 2001 Melbourne, Centre for Population and Urban Research, Monash University
OECD Towards high-performing health systems 2004 Paris, OECD 21
OECD International mobility of the Highly Skilled 2002 , OECD
Chu J How to plug Europe's Brain Drain. TIME Europe Magazine 2004 163
Glass H Choy WK Brain drain or brain exchange? Treasury Working Papers 01/22 2001 New Zealand,
Meyer JB Brown M Kaplan D Assessing the South African brain drain: a statistical comparison. Emigration and immigration: the South African figures. 2000 Cape Town, Development Policy Research Unit (DPRU), University of Cape Town, South Africa
Khadria B Migration of highly skilled Indians: Case studies of IT and health professionals STI working paper 2004
Buchan J Seccombe I Thomas S Overseas mobility of UK based nurses International journal of Nursing Studies 1997 34 54 62 9055121 10.1016/S0020-7489(96)00015-6
Westcott L Whitcombe S Globalistaion and Occupational Therapy: Poles Apart? British Journal of Occuptational Therapy 2003 66 328 330
Marshall J Golbart J Evans R International students of Speech and Language therapy in the UK: do we meet their needs? International Journal of Language and Communication Disorders 2004 39 269 284 15204456 10.1080/13682820310001615788
Nancarrow SA Promoting flexibility in the intermediate care workforce Journal of Integrated Care 2003 11 25 230
Nancarrow SA Borthwick AM Dynamic professional boundaries in the health care workforce Sociology of Health and Illness
Nancarrow SA Dynamic role boundaries in intermediate care Journal of Interprofessional Care 2004 18 141 151 15203673 10.1080/13561820410001686909
Duckett SJ Interventions to facilitate health workforce restructure Australia and New Zealand Health Policy 2005 2
Boyce R The future health care workforce: workforce reform in the allied health professions 2004 , School of Health and Rehabilitation Sciences
SANSA South African Network of Skills Abroad (SANSA) 2005
Allan H Larsen J “We need respect”: experiences of internationally recruited nurses in the UK: ; Surrey. 2003
O'Hagan J Radiation science victim of brain drain Sydney Morning Herald 2003 Sydney,
Pearson R Reilly P Robinson D Recruiting and developing an effective workforce in the British NHS. Health Services Research and Policy 2004 9 17 23 10.1258/135581904322724095
Park JR Coombs CR Wilkinson AJ Loan-Clarke A Arnold J Preston D Attractiveness of Physiotherapy in the National Health Service as a Career Choice Physiotherapy 2003 89 575 583 10.1016/S0031-9406(05)60056-9
Butler A Challenges in modernising social work education and training in the United Kingdom: discrete discipline status versus interprofessional and shared learning: ; Melbourne. 2005
Nancarrow SA Clark J Accountability and Accreditation in the Australian Allied Health Context Practice Development in Health Care 2003 2 33 348
Department of Health A first class service. Quality in the NHS: Lifelong Learning 1998 London, Department of Health Paragraph 3.32
Department of Health The new NHS Modern, Dependable 1997 London, Department of Health Paragraph 6.10
Turner P The occupational prestige of physiotherapy: Perceptions of student physiotherapists in Australia Australian Journal of Physiotherapy 2001 47 191 197 11552875
Borthwick A Perspectives on podiatric biomechanics: Foucault and the professional project British Journal of Podiatry 1999 2 21 228
Larkin GV Allsop J and Saks M Regulating the allied health professions Regulating the Health Professions 2002 London, Sage
Mandy A Tinley P Burnout and occupational stress in Australian podiatrists Australasian Journal of Podiatric Medicine 2002 36 101 108
Vernon W Borthwick A Farndon L Nancarrow S Walker J Issues of podiatry status in the U.K. British Journal of Podiatry 2005 8 6 10
Chumbler NR Grimm JW Podiatrists in a Weberian Spirit: Medical Care Specialists Seeking Full Professional Status and Recognition Sociological Focus 1992 25 345 347
Skipper JKJ Hughes JE Characteristics of podiatrists and podiatric practice J-Am-Podiatry-Assoc 1984 74 618 622 6520315
|
16253132
|
PMC1283969
|
CC BY
|
2021-01-04 16:38:28
|
no
|
Aust New Zealand Health Policy. 2005 Oct 27; 2:25
|
utf-8
|
Aust New Zealand Health Policy
| 2,005 |
10.1186/1743-8462-2-25
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2551622568810.1186/1471-2105-6-255Research ArticleThe modeled structure of the RNA dependent RNA polymerase of GBV-C Virus suggests a role for motif E in Flaviviridae RNA polymerases Ferron François [email protected] Cécile [email protected] Hélène [email protected] Bruno [email protected] Architecture et Fonction des Macromolécules Biologiques, UMR 6098 CNRS et Université Aix-Marseille I et II, ESIL, Campus de Luminy, 13288 Marseille Cedex 09, France2 Boston Biomedical Research Institute, 64, Grove St, Watertown 02472, MA, USA2005 14 10 2005 6 255 255 18 1 2005 14 10 2005 Copyright © 2005 Ferron et al; licensee BioMed Central Ltd.2005Ferron 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 Flaviviridae virus family includes major human and animal pathogens. The RNA dependent RNA polymerase (RdRp) plays a central role in the replication process, and thus is a validated target for antiviral drugs. Despite the increasing structural and enzymatic characterization of viral RdRps, detailed molecular replication mechanisms remain unclear. The hepatitis C virus (HCV) is a major human pathogen difficult to study in cultured cells. The bovine viral diarrhea virus (BVDV) is often used as a surrogate model to screen antiviral drugs against HCV. The structure of BVDV RdRp has been recently published. It presents several differences relative to HCV RdRp. These differences raise questions about the relevance of BVDV as a surrogate model, and cast novel interest on the "GB" virus C (GBV-C). Indeed, GBV-C is genetically closer to HCV than BVDV, and can lead to productive infection of cultured cells. There is no structural data for the GBV-C RdRp yet.
Results
We show in this study that the GBV-C RdRp is closest to the HCV RdRp. We report a 3D model of the GBV-C RdRp, developed using sequence-to-structure threading and comparative modeling based on the atomic coordinates of the HCV RdRp structure. Analysis of the predicted structural features in the phylogenetic context of the RNA polymerase family allows rationalizing most of the experimental data available. Both available structures and our model are explored to examine the catalytic cleft, allosteric and substrate binding sites.
Conclusion
Computational methods were used to infer evolutionary relationships and to predict the structure of a viral RNA polymerase. Docking a GTP molecule into the structure allows defining a GTP binding pocket in the GBV-C RdRp, such as that of BVDV. The resulting model suggests a new proposition for the mechanism of RNA synthesis, and may prove useful to design new experiments to implement our knowledge on the initiation mechanism of RNA polymerases.
==== Body
Background
The Flaviviridae virus family comprises three genera pestivirus, hepacivirus, and the large group of flavivirus. HCV causes acute and chronic hepatitis that may lead to cirrhosis and/or liver cancer. HCV is a major human pathogen, with 170 million people infected worldwide and 3 to 4 million of newly infected people each year [1]. Despite its large socio-economic impact, there is neither a vaccine nor an efficient, side-effect free therapy against this virus. Thus, the identification of potent drugs would be a major public health achievement. However, convenient small-animal models or productively infected cell systems to study HCV are still lacking. Consequently, compounds are often directly validated in HCV infected chimpanzees, or in cultured cells infected with related, surrogate viruses such as pestiviruses. The latter are animal pathogens showing similarity to hepaciviruses and flaviviruses [2] in genome structure, replication strategy, and individual gene products.
The RNA-dependent RNA polymerase (RdRp) is an enzyme playing a key role in the RNA replication process. Despite the increasing number of studies on the characterization of RdRp activity and structure, the precise molecular mechanism remains unclear. The postulated RNA replication process is a two-step mechanism. First, the initiation step of RNA synthesis begins at or near the 3' end of the (+) RNA template by means of a primer-independent (de novo) mechanism [3]. The de novo initiation consists in the addition of a nucleotide tri-phosphate (NTP) to the 3'-OH of the first initiating NTP. During the following so-called elongation phase, this nucleotidyl transfer reaction is repeated with subsequent NTPs to generate the complementary RNA product [3-6].
The structure of the RdRp of HCV (NS5B) has been determined [7,8]. It serves as reference in the uncovering of mechanism [9,10] and as link between structure and biochemical data for RNA polymerases [11]. The HCV polymerase shape resembles a semi-closed right hand and is made of three subdomains: fingers, palm and thumb (Figure 1). Computational and structural analysis of viral RdRp sequences has identified five universal motifs (F, A, B, C, E) located in (or close to) the palm (additional file 1). These motifs are both catalytic and structural. Fingers are made of a β-strand subdomain (four strands β1, β2, β4, β5 and an α-helix α1) and an α-helix rich subdomain (seven helices αA αB, αC, αD, αE, αF, αH). The palm is made of three stranded anti-parallel β-sheet (β3, β6, β7, and three helices (αG, αJ, αK). The thumb is mainly made of α-helices αN, αM, αL, αQ, αO, αP, αR) and a two-stranded antiparallel β-sheet (β10, β11) forming an extra structure called "the flap". The flap is proposed to play a role in the initiation mechanism, allowing only ssRNA to access the active site, and helping the correct positioning of the first two nucleotides [8].
Figure 1 Ribbon representation of the RdRp structure. HCV and BVDV RdRps are represented with their different subunits and domains. The thumb is colored in dark blue and yellow, fingers are colored in green and purple and the palm is colored in red. The image was generated using PYMOL.
Based on the structure of HCV RdRp solved in complex with NTPs [8], several GTP and NTP binding sites have been proposed. One is located behind the thumb, in a pocket on the surface of the structure, and has been called the allosteric (or surface) GTP binding site. The second one is in the catalytic cavity, where NTP can bind at various sites called P (priming), C (catalytic), and I (interrogating). Recently, the crystal structure of the RdRp of Bovine viral diarrhea virus (BVDV) has been published [12]. Another GTP binding site was found in the catalytic site, distinct from the P, C, and I sites of HCV NS5B. In the latter structure, this site corresponds to a cavity filled with water.
BVDV and HCV polymerases share a similar fold (Figure 1), but exhibit differences in the fingers and thumb subdomains due to differences in the number of secondary structure elements. As for the HCV polymerase, the shape of the BVDV polymerase is a semi-closed right hand made of fingers, palm, and thumb. Fingers are made of eleven β-strands, and twelve α-helices. The palm domain shows great conservation with the HCV palm domain. It consists of four strands forming a central β-sheet surrounded by three α-helices. The thumb contains height α-helices and five β-strands. The flap is lacking in BVDV RNA polymerase although Choi & et al [12] proposed that two β-strands with their connecting loops play the same role.
A number of structural differences in the flap and other subdomains raise the question of the relevance of BVDV as a surrogate model to discover HCV RNA polymerase inhibitors. Few years ago, "GB" viruses were identified and characterized as Flaviviridae agents leading to hepatitis [2] but not belonging to hepacivirus. Previous phylogenetic studies of GBV viruses were based on NS3 sequence comparisons [2]. Out of the three GB viruses identified so far, namely GBV-A, -B, and -C, two of them (GBV-A and GBV-B) are most likely monkey viruses while GBV-C can infect humans. HCV and GB virus genomes are organized in a similar way [13,14]. This similarity has been extended to the functional level with the characterization of the polymerase activity carried out by NS5B [15,16]. GBV-C virus allows a productive infection of cultured cells, that makes it a relevant alternate virus to be used as a model for HCV antiviral drug screening. In this study, we show using a NS5B-based phylogenetic analysis that GB viruses indeed carry the closest known RdRp to HCV in Flaviviridae. We have built a structural model for the GBV-C polymerase, which allows comparative analysis with HCV, and BVDV polymerase. Results presented in this paper suggest a novel model for the initiation of RNA synthesis in Flaviviridae. Due to its phylogenetic closeness to HCV, GBV-C might be an alternate and more relevant surrogate viral system than BVDV to HCV. Finally, the GBV-C polymerase model proposed in this study might help drug discovery and guide the characterization of the RNA polymerization mechanism.
Results and Discussion
Sequence analysis and phylogenic distribution
To compare Flaviviridae RdRps, we have used the set of sequences defined in VaZyMolO [17], that includes all sequences of completely sequenced viral genomes (Table 1).
The polymerase gene product alignment is based both on motif conservation and structural superimposition or conservation of secondary structures. We observe a great disparity depending on the genera of the compared sequences. Based on the alignment, a tree was derived (Figure 2 and additional file 1). Three major groups appear corresponding to the respective genus. Pestiviruses form a clear group distant from hepacivirus and flavivirus. This latter is the largest group of the family. It may be divided into several groups and isolated viruses reflecting adaptation. GB viruses cluster with hepacivirus in one group. This phylogenic distribution suggests that, in terms of a most relevant model polymerase useful in the screening of anti-viral drugs, GBV-C is closer to HCV than BVDV. The PSI-BLAST [18] search against non-redundant data bases (nrdb) using the GBV-C polymerase as an input sequence converges after one iteration and retrieves the HCV polymerase only, with an E-value of 9510-59.
Table 1 A listing of Flaviviridae. Viruses used in the study, together with their correspondent VaZyMolO and NCBI accession numbers.
Flaviviridae
Data Base Accession Virus NCBI Acc Protein Genus
VaZy 268 Dengue virus type 2 NP_056776.1
VaZy 270 Omsk hemorrhagic fever virus [Bogoluvovska] NP_878909.1
VaZy 345 West Nile virus NP_041724.2
VaZy 387 Kamiti River virus – isolate SR-82 NP_891560.1
VaZy 389 Yokose virus [Oita 36] NP_872627.1
VaZy 506 Hepatitis C virus type 1a – isolate H77 NP_671491.1
VaZy 508 Murray Valley encephalitis virus NP_051124.1
VaZy 509 Japanese encephalitis virus NP_059434.1
VaZy 510 Pestivirus type 1 [NADL] NP_040937.1
VaZy 511 Cell fusing agent virus NP_041725.1
VaZy 512 Yellow fever virus [Flavivirus (mosquito-borne)] NP_041726.1
VaZy 513 Hepatitis GB virus B NP_056931.1
VaZy 514 Bovine viral diarrhea virus genotype 2 [C413] NP_044731.1
VaZy 515 Pestivirus type 3 [X818; Clover Lane] NP_620062.1
VaZy 516 Tick-borne encephalitis virus NP_043135.1
VaZy 518 Powassan virus [LB] NP_620099.1
VaZy 519 Hepatitis GB virus C NP_043570.1
VaZy 520 Pestivirus type 2 [Eystrup] NP_075354.1
VaZy 521 Langat virus [TP21] NP_620108.1
VaZy 522 Louping ill virus [369/T2] NP_044677.1
VaZy 523 Deer tick virus [ctb30] – isolate CT95 NP_476520.1
VaZy 524 Tamana bat virus NP_658908.1
VaZy 525 Hepatitis GB virus A NP_045010.1
VaZy 526 Modoc virus [M544] NP_619758.1
VaZy 527 Montana myotis leukoencephalitis virus NP_689391.1
VaZy 528 Rio Bravo virus [RiMAR] NP_620044.1
VaZy 529 Alkhurma virus [1176] NP_722551.1
VaZy 530 Apoi virus [ApMAR] NP_620045.1
VaZy 531 Pestivirus – isolate reindeer-1 V60-Krefeld NP_620051.1
VaZy 532 Pestivirus – isolate giraffe-1 H138 NP_620053.1
GB group
Pestivirus
Hepacivirus
Flavivirus
sub group
Figure 2 The phylogenetic tree of Flaviviridae RdRps. Numbers at nodes indicate the statistical support of the branching order by bootstrap criteria. The bar at the bottom of the phylogram indicates the evolutionary distance, to which the branch lengths are scaled based on the estimated divergence. The dashed yellow line indicates flavivirus genus, the blue line indicates pestivirus genus and the red line indicates hepacivirus genus.
Homology modeling of the GBV-C Virus RNA Polymerase
A sequence alignment of GBV-C and HCV polymerases is presented in Figure 3. It is based on sequence and structure comparison taking into account the prediction of secondary structure for GBV-C. In order to validate our method to predict the secondary structure, we have first used the HCV polymerase NS5B as a test sequence. Using the software PREDICT PROTEIN [19], 50% of the β-sheets and 84% of the α-helix are correctly predicted in the HCV polymerase, and using PSI-PRED [20] we obtain 87.5% of correctly predicted structural elements. Such prediction results make us confident with respect to the reliability of the GBV-C prediction. The secondary structure elements of HCV polymerase and the structural prediction of the GBV-C polymerase are superimposed on the sequence alignment shown in Figure 3. The comparison between the secondary structure elements observed in the HCV crystal structure and the prediction made for GBV-C polymerase (Figure 3) shows that β˜
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacuaHYoGygaacaaaa@2E5C@ strands and α˜
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacuaHXoqygaacaaaa@2E5A@ helices are almost perfectly superimposed, albeit small gaps are located in few α-helices or loops. The alignment shows 32% identity and 72% similarity. Insertions and deletions localize in loops primarily. The amino acid conservation in the fingers and palm is close to 40% identity. Both motifs (F, A, B, C, E) and the residues involved in the I site (Arg 45, Lys 48, Lys 145, and Arg 151) match very well. As in the crystal structure of the HCV polymerase where the 55 C-terminal amino acids are deleted, we did not include the last 47 amino acids at the GBV-C polymerase C-terminus.
Figure 3 Alignment of the structural template (HCV) and the sequence of GBV-C. Sequence alignment of the HCV polymerase and the GBV-C polymerase. Identical amino acids are boxed in red. We superimposed secondary structure elements from the HCV polymerase in pink, the predicted structural elements of the GBV-C polymerase in blue and the secondary structure element of our final model in black. The HCV numbering according to [7] is given in pink. The numbering in dark red corresponds to the structure elements which have been observed with a better resolution. The dots in the alignment and structural elements (predicted or average) symbolise gaps. Green letters show universal motifs of RNA polymerases. Green arrows indicate amino acids involved in NTP binding. The green star indicates the amino acid supposed to stack the priming base. The numbering is that of the GBV-C polymerase. Numbers in dark green indicate cysteines involved in a putative disulfide bridge in our model. Residues forming the allosteric GTP binding site are underlined in black.
Both the sequence alignment and predicted secondary structure shown in Figure 3 were used in SWISS-MODEL (see methods) [21] to build and refine the GBV-C polymerase model (Figure 4). Alternative models were also generated using SCRWL [22], 3D-JIGSAW [23-25] and MODELLER [26] and evaluated using VERIFY3D [27]. The results are presented in the additional files 4 and 5. All models are evaluated as good by VERIFY3D [27], and are very similar, although some differences exist in flexible loops. Key residues of the active site are perfectly superimposed, unlike side chains because of their flexibility (additional file 4). These similar results make us very confident of the reliability of the GBV-C polymerase model, and for clarity, we will focus on the model generated by SWISS-MODEL [21]. The modeled structure was then evaluated using PROCHECK [28], "WHAT IF" [29], and VERIFY3D [27]. Results are shown in Table 2 A/B and additional file 5. The Ramachandran plot is correct, and according to theses programs, scores are within expected ranges for well-refined structures. Nevertheless, several residues located in flexible loops fall into disallowed regions of the Ramachandran plot (additional file 2A): Ser 100, Val 255, Thr 256 and Cys 215. The Ramachandran Plot statistics given by PROCHECK (additional file 2B) shows clearly that 99% of the residues are in allowed regions. The score corresponding to the chi-1/chi-2 angles of all residues is within expected ranges for well-refined structures (Table 2). The model has a normal distribution of residue types over the inside and the outside of the protein. Again, the backbone conformation analysis gives a score that is normal for correctly refined protein structures. The RMS Z-score given in Table 2 is expected to be around 1.0 for a normally restrained data set, and this is indeed observed as in the case of high-resolution X-ray structures. In the GBV-C polymerase model, bond angles and lengths can be considered to deviate normally from the mean standard bond angles.
Table 2 Quality of the model. A: Parameters reflecting the quality of the model checked by «WHAT IF» [29]. B: Quality of chain of the model. The model is verified at 2Å resolution. Parameter values in the table represent observed values for the GBV-C polymerase model compared with typical values obtained for well refined structures at the same resolution [28].
A
Structure Z-scores:
1st generation packing quality -1.577
2nd generation packing quality -2.94
Ramachandran plot appearance -0.74
chi-1/chi-2 rotamer normality -0.224
Backbone conformation -0.904
RMS Z-scores, should be close to 1.0:
Bond lengths 0.950
Bond angles 1.426
Omega angle restraints -0.923
Inside/Outside distribution 1.096
B
Stereochemical parameter N° of data points Parameter valueTypical valueBand width N° of bandwidth
Stereochemistry of main-chain
Percentage residues in A, B, L 438 89 83.8 10 0.5
Omega angle S.D 507 6.8 6 3 0.3
Bad contacts 100 residues 3 0.6 4.2 10 -0.4
Zeta angle S.D. 476 2.8 3.1 1.6 -0.2
Hydrogen bond energy S.D. 305 0.7 0.8 0.2 -0.4
Stereochemistry of side-chain
Chi-1 gauche minus S.D. 79 15.7 18.1 6.5 -0.4
Chi-1 trans S.D. 103 13.2 19 5.3 -1.1
Chi-1 gauche plus S.D. 205 11.4 17.5 4.9 -1.2
Chi-1 pooled S.D. 387 13 18.2 4.8 -1.1
Chi-2 trans S.D. 114 15.7 20.4 5 -0.9
Figure 4 Structural comparison of GBV-C and HCV RdRps. A: The model of the GBV-C polymerase is presented as a front view highlighting the Flap and the histidine residue pointing to the catalytic site. The color scheme is the same as in Figure 1. Images were generated using POV-RAY. B: a 180° rotation view of the GBV-C model show in A. Images were generated using POV-RAY. C: Superimposition of the X-ray structure of the HCV polymerase (in red) and the GBV-C polymerase model (in purple and yellow). A zoomed view of the Flap region is presented in the upper side box in order to highlight the perfect superimposition of the aromatic ring of the histidine found in the GBV-C polymerase and the tyrosine found in the HCV polymerase. Images were generated using POV-RAY.
As expected with such good scores, the model of the GBV-C polymerase is similar to that of HCV, and displays the essential features of the typical RNA dependent RNA polymerase fold (Figure 4A and 4B). However, we note two small differences between the HCV structure and the GBV-C model. First, Cys 283 and Cys 308 are spatially close enough to model a disulphide bridge (Figure 3 and additional file 3). This bond connects the fingers and the palm, and may stabilize the protein. Second, the superimposition of the GBV-C model and the HCV structure (Figure 4C) shows little but notable differences in the palm and thumb. The secondary structure elements are conserved in place and type, but they are shorter in the model than in the structure. These secondary structure elements should have similar functions, though. For example His 428 overlaps Tyr 448 of the HCV flap (Figure 4D) and replacement of the aromatic ring of the tyrosine by the histidine ring could play the same role during initiation (see discussion below).
Surface analysis
We note several differences between the surface shapes (Figure 5) of HCV RdRp and the GBV-C model. As the two backbones are superimposed these differences are only due to the variability of side chains. The sequence conservation reported for the GBV-C model (additional file 3) shows that amino acids oriented toward the inner side of the protein are conserved whereas the amino acid which are pointing to the surface show low identity. This surface variability may be explained by the fact that the GBV-C polymerase form a complex with other viral proteins, as it is the case for the HCV polymerase which interacts with NS3 or NS5A proteins, or as observed in the case of the poliovirus polymerase [30]. These other viral proteins may differ in their NS5B binding domain between HCV and GBV-C. Moreover, it has been shown that the HCV polymerase dimerizes and can form higher order structures after oligomerization. This multimerization is required for the HCV polymerase activity [31]. As the GBV-C polymerase is similar to HCV polymerase, the same oligomerization may also occur in the case of the GBV-C polymerase. Surface amino-acids have then to be specific to the virus to allow correct dimerization of the polymerase and/or interaction with the other components of the replicative complex. The electrostatic potential comparison is presented in Figure 5. It shows that the charges distribution on the surface of the model is globally equivalent to those located on the surface of HCV polymerase. We observe that the thumb in both cases is negatively charged (Figure 5B and 5E). The positive channel supposed to guide the RNA template to the catalytic site is very well conserved, and the flap is partially obstructing this cavity. The difference appears near the NTP tunnel (Figure 5C and 5F). In the HVC polymerase structure, the surface is clearly positively charged whereas in the GBV-C polymerase model the positive charge is less apparent.
Figure 5 Surface comparison of GBV-C and HCV RdRps. A, B, C correspond to different views of the GBV-C polymerase surfaces calculated using GRASP. The surface is colored according to the electrostatic potential. The red correspond to negative charges, the white is neutral, and the blue corresponds positive charges. D, E, F correspond to the surface of HCV polymerase in similar and respective orientations. The color ramp is the same as for the GBV-C polymerase surface.
NTP-binding sites
In the HCV polymerase, the allosteric site forms a pocket where GTP binds. Such a pocket does exist in GBV-C despite sequence variability (Figure 3), and is located behind the thumb subdomain. The surface analysis shows that the pocket has a hydrophobic nature, except for the side chains of Asp 30 and Lys 473 that may however participate in the binding of a GTP molecule (see below).
In the HCV structure, several NTP molecules can bind to the catalytic site at P, C, and I sites. Indeed, up to 9 phosphate moieties can be seen in the crystal structure. Only the nucleotide bound at the C site is well defined, although its nucleobase is probably incorrectly located in the absence of the RNA template [8]. Clearly, a better definition of nucleotides and template is needed to understand the RNA synthesis process. On the other hand, the BVDV polymerase structure in complex with GTP in the catalytic cavity suggests a role for this nucleotide in the initiation of RNA synthesis, as proposed below.
Docking of GTP in GBV-C
The analysis of the thumb in terms of structure and sequence comparison proved to be informative to propose an RNA synthesis initiation mechanism. Previously, in HCV polymerase the E motif has been proposed as a part of the site that accommodates the first NTP incorporated during initiation of RNA synthesis (P site). Motif E is defined by the CS-18X-R signature (Figure 3 and additional file 1) [8]. In the case of BVDV, the polymerase structure has also been solved in complex with GTP [12]. This GTP is found in a binding pocket that is mainly constituted by amino acids within motif E. Their side chains effectively stabilize the phosphate chains of GTP with an Arginine (Arg 529) further away in the sequence. The NS5B sequence comparison of Flaviviridae showed that motif E could be extended to CS-18X-[RKT]-x(8)-[RK] as a signature sequence (Figure 6). In the BVDV polymerase structure, the GTP molecule has been compared to a vestigial RNA molecule acting as a primer [12]. In the HCV polymerase structure, this GTP position corresponds to a cavity filled with water molecules. In the GBV-C model such a pocket exists, but its shape is different. Based on the GTP localization in the BVDV polymerase structure, we have docked a GTP molecule in GBV-C and HCV polymerase structures to see if these pockets could accommodate a GTP molecule in a similar manner. These three pockets are similar regarding position and nature of the conserved residues. This characteristic allows a perfect fitting of the molecule into the GBV-C and HCV pockets (Figure 7). In all cases, part of the cavity is positively charged contributing to the stabilization of the GTP-phosphate chain in the pocket. This stabilization involves Thr 367 and Arg 371 in GBV-C motif E and the corresponding Arg 386 and Arg 394 in HCV. The Ser 349 in GBV-C (Ser 367 for HCV) of the CS motif forms the bottom of the cavity. In the structure and both models, the cavity is obstructed by a proline (Pro 189 GBV-C; Pro 321 BVDV; Pro 197 HCV). However, amino acids stabilizing the guanine base are different. While the base is stabilized only by hydrogen bonds with Tyr 187 in GBV-C and Tyr 195 in HCV, Thr 320 and Tyr 581 stabilize it in the case of BVDV. In GBV-C and HCV an aromatic residue located at the extremity of the flap, His 448 in GBV-C and Tyr 448 in HCV forms the top of the cavity stabilizing the cycle of the base (Figure 7). Although the pocket is conserved in charged residues, the GTP position in the pocket is different. Indeed, because the GBV-C pocket is somehow smaller than in the case of BVDV, the GTP ribose is flipped and the phosphate chain bends to follow the surface of the pocket (Figure 7, compare A and B). In HCV polymerase, the cavity is larger than the GBV-C pocket and therefore the binding of the GTP molecule is closer to what is observed in the case of the BVDV polymerase structure (Figure 7 compare B and C).
Figure 6 General alignment of the E motif in Flaviviridae RdRps. The conserved motif is labeled according to the nomenclature described for the RNA polymerase family. Invariant residues are highlighted in red, while conserved residues are boxed yellow highlighted in bold. Consensus sequence with 70% similarity is shown down the alignment. The sequences are sorted by genera.
Figure 7 Docking a GTP in the GBV-C polymerase. A to C: Views of GTP-binding pockets. The surface is colored according to the electrostatic potential nomenclature. Hydrogen bonds are indicated in dotted lines and the numbering indicates the distance (in Å) between amino acids. A: The proposed GTP pocket in the GBV-C polymerase model with a docked GTP molecule. B: The BVDV polymerase structure in which a GTP molecule is co-crystallized. C: The proposed GTP pocket in the HCV polymerase structure with a docked GTP molecule. D to F: LIGPLOT presenting residues involved in the stabilization of GTP. Hydrogen bonds are indicated in dotted lines and the numbering indicates the distance (in Å) between amino acids. D: View of the GBV-C GTP pocket. E: same view of the BVDV GTP pocket. F: LIGPLOT of the HCV GTP pocket. Images were generated using PYMOL.
Based on our docking results, we propose that motif E is the signature sequence of a GTP binding site in which GTP is required to hold the initiation complex tight. In our structural model, the GTP itself is too remote to act as a platform for the nucleotide positioned at the P site. The modeled GTP binding site together with the observed position of the flap lead us to suggest a mechanism for de novo initiation (Figure 8). We propose that once the first reaction of initiation is achieved (Figure 8C and 8D), the initiated template enters the pocket where the motif E GTP is located, and stacks against the guanine base (Figure 8E). This stacking induces a rearrangement of the base, which now contacts the flap. This latter interaction induces the opening of the flap leading to GTP release and further major structural changes within the polymerase (Figure 8F and 8G). The movement of the flap is supposed to occur to open the cavity allowing the elongation of the neo-synthesized RNA. The opening of the cavity implies that the thumb moves. It has been already observed that the fingers and the palm rotate as rigid body around the axis against the thumb domain [9]. In our model, the flap is spatially conserved suggesting that the same movement may occur during the elongation step of the GBV-C polymerization. Additionally, the position of the amino acid closing the cavity of the polymerase (flap in the case of GBV-C and HCV, or the β-sheet in the case of BVDV) suggests that the opening movement is specific for each virus. This movement would be best described as an opening from the top for HCV and GBV-C and, lateral for BVDV. Recently, we have characterized the initiation steps of RNA synthesis kinetically [32]. It is interesting to note that our present model is in agreement with the kinetic data showing that the N2 to N3 polymerization reaction is strongly rate limiting, and corresponds to the first partial opening of the flap to release GTP as proposed in Figure 8 panel F, whereas the other rate-limiting step from N4 to N6 corresponds to the other complete flap opening allowing dsRNA to exit from the active site as proposed in panel G.
Figure 8 A model for de novo RNA synthesis at the hepacivirus NS5B active site. A: The polymerase is represented schematically to illustrate key points in the reaction mechanism. B: The RNA template is represented as clear blue squares. NTP are as red squares, the allosteric GTP is represented as a dark blue square, and the bound GTP as a green blue square. C: Binding of the first NTP in the active site. D: The initiation reaction is presented with a yellow lightening. Upon incorporation of the third NTP, the template and the neo synthesized RNA slide to the cavity pushing the GTP towards the flap. E: Intermediate position where the flap, GTP and RNA template are stacked. F: Opening of the flap and release of GTP. G: The polymerase shifts to the elongation mode; the thumb moves to fully open the cavity, and the elongation resumes.
Conclusion
The recently published high-resolution three-dimensional structure of BVDV and HCV polymerase has allowed the structural comparison of the two polymerases. Major differences in fingers and thumb suggest that molecular interactions during the initiation mechanism are different. BVDV has been used as a model in the study of hepaciviruses. However, phylogenic analysis shows that GBV-C is more closely related to HCV than BVDV. We propose here a reliable model of the GBV-C polymerase structure.
The model of the GBV-C polymerase is poorly defined in loopy regions where most of the gaps have been introduced. Despite this imprecision, the very good scores of the structural indicators make us very confident of the reliability of our model. Moreover, the model is consistent with the known three-dimensional structure of RNA dependent RNA polymerases, and show conservation of all structural elements involved in polymerization (catalytic site, RNA positive channel, NTP tunnel). As expected after the alignment and prediction study, the GBV-C model is very close to the HCV structure, even with a conserved allosteric GTP binding site. Based on the BVDV polymerase/GTP complex structure, we generated a model of a corresponding complex of GBV-C. We propose a role for the GTP molecule bound at a site involved in the initiation of RNA synthesis. Our study provides useful information of the location of residues involved in the polymerization process and hence presents a useful resource for future biochemical analysis and drug discovery.
Methods
Sequence Retrieval
The sequences related to the different kind of polymerase were retrieved with a PSI-BLAST [18] with standard parameters from the public available protein database Swiss-Prot [33], Protein Data Bank (PDB) [34] and VaZyMolO [17]. For this study we have used different structures of HCV (PDB code: [1GX5, 1GX6]), and BVDV (PDB code: [1S48, 1S49]).
Sequence alignment comparison
Alignment of representative sequences from several members of Flaviviridae were performed using CLUSTALW [35] with the following parameter. Slow Algorithm, Identity matrix for pairwise alignment and BLOSUM series matrix for multiple alignments. The alignment was then carefully analyzed and optimized with SEAVIEW [36], taking into account the secondary structure prediction and structural elements when existing. This alignment was cross checked using 3DJURY [37].
The secondary structure predictions were carried out using JPRED2 [38], PSI-PRED [20] and PREDICT-PROTEIN Server [19]. We used PREDICT-PROTEIN with a window of 150 amino acids in order to increase the sensitivity of the prediction. 20 amino acids overlap with each common superimposed window. The results presented are consensus. Sequence alignment with structural information (structure or predictions) and the comparison of the structure one dimension of the known viral polymerases was performed using ESPript 2.0 [39] and ENDscript 1.0 [40].
To visualize conserved region in amino acids composition on the reference structure, we used BOBSCRIPT [41]. The similarity scores were calculated from the CLUSTALW [35] alignment and they are shown on this structure with a white (low score) to red (identity) color ramp.
Phylogenetic analysis
The sampling variance of the distance values was estimated from 1000 bootstrap resamplings of the alignment columns. The evolutionary inference was performed according to the Neighbor-joining method. Multiple runs were conducted with randomized sequence input order to avoid the tree being caught in a local statistical minimum. The tree was generated using Phylodendron (©1997 Gilbert).
Model building, refinement and evaluation
The resulting multiple sequence alignment with the consensus secondary structure prediction was used as template to generate the threading alignment. The derived pairwise alignment serves as reference for preparing the file for the model. SWISS-PDB VIEWER [21] was used to generate a first threading model. The three dimensional model of the GBV-C RdRp was constructed using the crystal structure coordinates of the HCV polymerase [8,7] (PDB code: 1GX5, 1QUV). Main gaps appear in loops and smaller ones in helices. This alignment and the threading model serve as a template file for SWISS-MODEL [21]. The non-modeled loops were manually built after scanning the loop database. The model was then minimized with a cut off of 10 Å with 40 cycles of steepest descent until the gradient fell below 10 Kcal/mol and 20 cycles of conjugate gradient. The computations were done in vacuum with using GROMOS 96 [42,43] force field. To generate alternate models, we have used the 3D-JIGSAW [23-25] server, SCRWL [22] and MODELLER [26]. In this latter, positions of predicted catalytic residues and secondary structure elements were used as spatial restraints.
Surface comparison of the template and the model were performed with GRASP [44]. The generated models were checked using PROCHECK [285] "WHAT IF" [29] and/or VERIFY3D [27].
Docking GTP molecule in GBV-C
The 3D model of the GBV-C RNA polymerase was used as a target for the docking of GTP. We first superimposed the structure of BVDV RNA polymerase/GTP complex (PDB code 1S49) with our 3D model. This step was performed with the program Turbo-Frodo [44]. A docking study was performed to explore the presence or absence of a GTP binding pocket like, as it was described in the BVDV polymerase structure. For the docking procedure, the program AUTODOCK 3.0.5 [45] was used with a grid spacing of 0.375 Å and 40 × 40 × 40 number of points. The grid was centered on the mass center of the GTP molecule. The GA-LS method was adopted using the default settings. Amber united atoms were assigned to the protein using the program AUTODOCK TOOLS. 250 possible binding conformations were generated. The results of AUTODOCK run were clustered using a RMSD tolerance of 1.0 Å. We considered the structure of the first cluster. To validate the use of the AUTODOCK program, the docking study was performed on the BVDV polymerase with GTP as a reference. This program successfully reproduced the experimental binding conformation with acceptable root-mean-square deviation (RMSD) of atom coordinates. Finally, the interaction models of GTP with the binding pocket were produced using the LIGPLOT program [46].
Authors' contributions
FF carried out the sequence retrieval, alignments, modeling, phylogenic studies, and, the structure and docking analysis. CB and HD performed the docking and structure analysis. BC conceived of the study, and participated in its design, analysis, and coordination. FF, CB, HD and BC all contributed to writing the final manuscript and interpretation of data.
Note
Table 1 – A listing of Flaviviridae
Viruses used in the study, together with their correspondent VaZyMolO and NCBI accession numbers.
Table 2 – Quality of the model
A: Parameters reflecting the quality of the model checked by « WHAT IF » [26].
B: Quality of chain of the model. The model is verified at 2Å resolution. Parameter values in the table represent observed values for the GBV-C polymerase model compared with typical values obtained for well refined structures at the same resolution [25].
Supplementary Material
Additional File 1
Multiple alignment of Flaviviridae RNA polymerase palm subdomains. The conserved motifs are labeled according to the nomenclature described for the RNA polymerase family. Invariant residues are highlighted in red, while conserved residues are boxed yellow highlighted in bold. A consensus sequence with 70% similarity is shown below the alignment. The sequences are sorted by genera.
Click here for file
Additional File 2
Ramachandran plot of the GBV-C Model with PROCHECK statistics. A: Ramachandran plot of GBV-C polymerase model. Favoured and allowed regions are in red and yellow, respectively. All residues are represented by black boxes (■) except glycine (▲). Red boxes () highlight residues in forbidden regions.
Click here for file
Additional File 4
Superimposition of the models generated with different programs. A. Models generated using SWISS-MODEL (represented in light blue), and MODELLER: model 1 (represented in light green), model 3 (represented in magenta) or model 2 (represented in yellow) were superimposed. B. 90° rotation view of the same superimposed models. C. Zoom view of the superimposed amino acids of the GTP pocket.
Click here for file
Additional File 5
score of the different model generated according to VERIFY3D.
Click here for file
Additional File 3
Residues conservation plotted on the structure. Calculated homology based on the superimposition of the structure of HCV polymerase on the GBV-C polymerase model. The figure was done using BOBSCRIPT. The similarity is shown on this structure by a white (low score) to red (identity) colour ramp. The green doted line indicates the position of the disulfide bridge.
Click here for file
Acknowledgements
The authors thanks Dr. Barbara Selisko, Dr. Sonia Longhi, Dr Yana Khalina and Jean-Marie Bourhis for critical reading of the manuscript. This work was support by Association Nationale de Recherche sur le Sida (ANRS), by the European Community (Flavitherapeutics European Contract N° QLK3-CT-2001-00506) and by the Centre National de Recherche Scientifique (CNRS).
==== Refs
World Health Organization Global surveillance and control of hepatitis C J Viral Hepatitis 1999 6 35 47 10.1046/j.1365-2893.1999.6120139.x
Lindenbach BD Rice CM Knipe DM, Howley PM Flaviridae : The viruses and their replication Fields virology 2001 1 Fourth 991 1042 virology.
Kao CC Singh P Ecker DJ De novo initiation of viral RNA-dependent RNA synthesis Virology 2001 287 251 260 11531403 10.1006/viro.2001.1039
Kao CC Del Vecchio AM Zhong W De novo initiation of RNA synthesis by a recombinant flaviviridae RNA-dependent RNA polymerase Virology 1999 253 1 7 9887312 10.1006/viro.1998.9517
Kim MJ Zhong W Hong Z Kao CC Template nucleotide moieties required for de novo initiation of RNA synthesis by a recombinant viral RNA-dependent RNA polymerase J Virol 2000 74 10312 10322 11044075 10.1128/JVI.74.22.10312-10322.2000
Luo G Hamatake RK Mathis DM Racela J Rigat KL Lemm J Colonno RJ De novo initiation of RNA synthesis by the RNA-dependent RNA polymerase (NS5B) of hepatitis C virus J Virol 2000 74 851 863 10623748 10.1128/JVI.74.2.851-863.2000
Ago H Adachi T Yoshida A Yamamoto M Habuka N Yatsunami K Miyano M Crystal structure of the RNA-dependent RNA polymerase of hepatitis C virus Structure Fold Des 1999 7 1417 1426 10574802 10.1016/S0969-2126(00)80031-3
Bressanelli S Tomei L Rey FA De Francesco R Structural analysis of the hepatitis C virus RNA polymerase in complex with ribonucleotides J Virol 2002 76 3482 3492 11884572 10.1128/JVI.76.7.3482-3492.2002
Adachi T Ago H Habuka N Okuda K Komatsu M Ikeda S Yatsunami K The essential role of C-terminal residues in regulating the activity of hepatitis C virus RNA-dependent RNA polymerase Biochim Biophys Acta 2002 1601 38 48 12429501
Ranjith-Kumar CT Sarisky RT Gutshall L Thomson M Kao CC De novo initiation pocket mutations have multiple effects on hepatitis C virus RNA-dependent RNA polymerase activities J Virol 2004 78 12207 12217 15507607 10.1128/JVI.78.22.12207-12217.2004
Zhong W Ingravallo P Wright-Minogue J Uss AS Skelton A Ferrari E Lau JY Hong Z RNA-dependent RNA polymerase activity encoded by GB virus-B non-structural protein 5B J Viral Hepat 2000 7 335 342 10971821 10.1046/j.1365-2893.2000.00226.x
Choi KH Groarke JM Young DC Kuhn RJ Smith JL Pevear DC Rossmann MG The structure of the RNA-dependent RNA polymerase from bovine viral diarrhea virus establishes the role of GTP in de novo initiation Proc Natl Acad Sci USA 2004 101 4425 4430 15070734 10.1073/pnas.0400660101
Muerhoff AS Leary TP Simons JN Pilot-Matias TJ Dawson GJ Erker JC Chalmers ML Schlauder GG Desai SM Mushahwar IK Genomic organization of GB viruses A and B: two new members of the Flaviviridae associated with GB agent hepatitis J Virol 1995 69 5621 5630 7637008
Simons JN Leary TP Dawson GJ Pilot-Matias TJ Muerhoff AS Schlauder GG Desai SM Mushahwar IK Isolation of novel virus-like sequences associated with human hepatitis Nat Med 1995 1 564 569 7585124 10.1038/nm0695-564
Ranjith-Kumar CT Gutshall L Kim MJ Sarisky RT Kao CC Requirements for de novo initiation of RNA synthesis by recombinant flaviviral RNA-dependent RNA polymerases J Virol 2002 76 12526 12536 12438578 10.1128/JVI.76.24.12526-12536.2002
Ranjith-Kumar CT Santos JL Gutshall LL Johnston VK Lin-Goerke J Kim MJ Porter DJ Maley D Greenwood C Earnshaw DL Enzymatic activities of the GB virus-B RNA-dependent RNA polymerase Virology 2003 312 270 280 12919733 10.1016/S0042-6822(03)00247-2
Ferron F Rancurel C Longhi S Cambillau C Henrissat B Canard B VaZyMolO: a tool to define and classify modularity in viral proteins J Gen Virol 2005 86 743 749 VaZyMolO 15722535 10.1099/vir.0.80590-0
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
Rost B PHD: predicting one-dimensional protein structure by profile-based neural networks Methods Enzymol 1996 266 525 539 8743704
McGuffin LJ Bryson K Jones DT The PSIPRED protein structure prediction server Bioinformatics 2000 16 404 405 10869041 10.1093/bioinformatics/16.4.404
Guex N Peitsch MC SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling Electrophoresis 1997 18 2714 2723 9504803 10.1002/elps.1150181505
Canutescu A Shelenkov A Dunbrack RJ A graph-theory algorithm for rapid protein side-chain prediction Protein Science 2003 12 2001 2014 12930999 10.1110/ps.03154503
Bates PA Sternberg MJE Model Building by Comparison at CASP3: Using Expert Knowledge and Computer Automation Proteins: Structure, Function and Genetics 1999 47 54 10.1002/(SICI)1097-0134(1999)37:3+<47::AID-PROT7>3.0.CO;2-F
Bates PA Enhancement of Protein Modelling by Human Intervention in Applying the Automatic Programs 3D-JIGSAW and 3D-PSSM Proteins: Structure, Function and Genetics 2001 39 46 10.1002/prot.1168
Contreras-Moreira B Bates PA Domain fishing: a first step in protein comparative modelling Bioinformatics 2002 18 1141 1142 12176841 10.1093/bioinformatics/18.8.1141
Fiser A Sali A Modeller: generation and refinement of homology-based protein structure models Methods Enzymol 2003 374 461 491 14696385
Eisenberg D Luthy R Bowie JU VERIFY3D: assessment of protein models with three-dimensional profiles Methods Enzymol 1997 277 396 404 9379925
Laskowski RA Macarthur MW Moss DS Thornton JM Procheck – a Program to Check the Stereochemical Quality of Protein Structures J Appl Crystallogr 1993 26 283 291 10.1107/S0021889892009944
Vriend G WHAT IF: a molecular modeling and drug design program J Mol Graph 1990 8 52 56 29 2268628 10.1016/0263-7855(90)80070-V
Lyle JM Bullitt E Bienz K Kirkegaard K Visualization and functional analysis of RNA-dependent RNA polymerase lattices Science 2002 296 2218 2222 12077417 10.1126/science.1070585
Qin W Luo H Nomura T Hayashi N Yamashita T Murakami S Oligomeric interaction of hepatitis C virus NS5B is critical for catalytic activity of RNA-dependent RNA polymerase J Biol Chem 2002 277 2132 2137 11673460 10.1074/jbc.M106880200
Dutartre H Boretto J Guillemot JC Canard B A relaxed discrimination of 2'-O-methyl-GTP relative to GTP between de novo and Elongative RNA synthesis by the hepatitis C RNA-dependent RNA polymerase NS5B J Biol Chem 2005 280 6359 6368 15561709 10.1074/jbc.M410191200
Bairoch A Apweiler R The SWISS-PROT protein sequence database and its supplement TrEMBL in 2000 Nucleic Acids Res 2000 28 45 48 10592178 10.1093/nar/28.1.45
Berman HM Westbrook J Feng Z Gilliland G Bhat TN Weissig H Shindyalov IN Bourne PE The Protein Data Bank Nucleic Acids Res 2000 28 235 242 10592235 10.1093/nar/28.1.235
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
Galtier N Gouy M Gautier C SEAVIEW and PHYLO_WIN: two graphic tools for sequence alignment and molecular phylogeny Comput Appl Biosci 1996 12 543 548 9021275
Ginalski K Elofsson A Fischer D Rychlewski L 3D-Jury: a simple approach to improve protein structure predictions Bioinformatics 2003 19 1015 1018 12761065 10.1093/bioinformatics/btg124
Cuff JA Barton GJ Application of multiple sequence alignment profiles to improve protein secondary structure prediction Proteins 2000 40 502 511 10861942 10.1002/1097-0134(20000815)40:3<502::AID-PROT170>3.0.CO;2-Q
Gouet P Courcelle E Stuart DI Metoz F ESPript: analysis of multiple sequence alignments in PostScript Bioinformatics 1999 15 305 308 10320398 10.1093/bioinformatics/15.4.305
Gouet P Courcelle E ENDscript: a workflow to display sequence and structure information Bioinformatics 2002 18 767 768 12050076 10.1093/bioinformatics/18.5.767
Esnouf RM An extensively modified version of MolScript that includes greatly enhanced coloring capabilities J Mol Graph Model 1997 15 132 134 112-133. 9385560 10.1016/S1093-3263(97)00021-1
Gunsteren WFv HJCB Engl E Computer Simulation of Molecular Dynamics: Methodology, Applications and Perspectives in Chemistry Angew Chem Int 1990 29 992 1023 10.1002/anie.199009921
Gunsteren WFv Hünenberger PH Mark AE Smith PE Tironi IG Computer simulation of protein motion Computer Phys Communications 1995 91 305 319 10.1016/0010-4655(95)00055-K
Sharp K Fine R Honig B Computer simulations of the diffusion of a substrate to an active site of an enzyme Science 1987 236 1460 1463 3589666
Roussel A Cambillau C Silicon Graphics MV TURBO-FRODO 1991 CA: In Silicon Graphics Geometry Partners Directory 86
Morris GM Goodsell DS Halliday RS Huey R Hart WE Belew RK Olson AJ Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function Journal of Computational Chemistry 1998 19 1639 1662 10.1002/(SICI)1096-987X(19981115)19:14<1639::AID-JCC10>3.0.CO;2-B
Wallace AC Laskowski RA Thornton JM LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions Protein Eng 1995 8 127 134 7630882
|
16225688
|
PMC1283970
|
CC BY
|
2021-01-04 16:27:46
|
no
|
BMC Bioinformatics. 2005 Oct 14; 6:255
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-255
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2581623231210.1186/1471-2105-6-258Methodology ArticleDynamic modeling of cis-regulatory circuits and gene expression prediction via cross-gene identification Lin Li-Hsieh [email protected] Hsiao-Ching [email protected] Wen-Hsiung [email protected] Bor-Sen [email protected] Lab. of System Biology, National Tsing Hua University, 101, Sec 2, Kuang Fu Road, Hsinchu, 300, Taiwan.2 Department of Life Science & Institute of Bioinformatics and Structural Biology, National Tsing Hua University, Hsinchu, 300, Taiwan.3 Department of Ecology and Evolution, University of Chicago, USA.4 Genomics Research Center, Academia Sinica, Taipei, Taiwan.2005 18 10 2005 6 258 258 19 3 2005 18 10 2005 Copyright © 2005 Lin et al; licensee BioMed Central Ltd.2005Lin 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
Gene expression programs depend on recognition of cis elements in promoter region of target genes by transcription factors (TFs), but how TFs regulate gene expression via recognition of cis elements is still not clear. To study this issue, we define the cis-regulatory circuit of a gene as a system that consists of its cis elements and the interactions among their recognizing TFs and develop a dynamic model to study the functional architecture and dynamics of the circuit. This is in contrast to traditional approaches where a cis-regulatory circuit is constructed by a mutagenesis or motif-deletion scheme. We estimate the regulatory functions of cis-regulatory circuits using microarray data.
Results
A novel cross-gene identification scheme is proposed to infer how multiple TFs coordinate to regulate gene transcription in the yeast cell cycle and to uncover hidden regulatory functions of a cis-regulatory circuit. Some advantages of this approach over most current methods are that it is based on data obtained from intact cis-regulatory circuits and that a dynamic model can quantitatively characterize the regulatory function of each TF and the interactions among the TFs. Our method may also be applicable to other genes if their expression profiles have been examined for a sufficiently long time.
Conclusion
In this study, we have developed a dynamic model to reconstruct cis-regulatory circuits and a cross-gene identification scheme to estimate the regulatory functions of the TFs that control the regulation of the genes under study. We have applied this method to cell cycle genes because the available expression profiles for these genes are long enough. Our method not only can quantify the regulatory strengths and synergy of the TFs but also can predict the expression profile of any gene having a subset of the cis elements studied.
==== Body
Background
Cis-regulatory circuits have been applied to model cell cycle control and other developmental processes [1,2]. Recently, the genetic regulatory and transcriptional networks of yeast have been studied [3-8]. However, in order to understand the regulation of any particular gene in the network, one should study carefully how the genes in the network collectively operate with remarkable precision in response to environmental cues and the structure and function of the cis-regulatory circuit of the gene [7]. The cis-regulatory circuit of a gene consists of its cis elements, i.e., binding motifs of transcription factor (TF), and the interactions among their recognizing TFs. The cis elements of a gene can be considered as the information processing units in the regulatory circuit; they receive multiple inputs from the TFs that bind the cis elements of the gene. The output is the instruction for the transcription apparatus to determine whether the gene is to be expressed at a specific rate or to be repressed [9,10].
A cis-regulatory circuit may be regarded as a control device that is called into play by the TFs that have target sites inside the promoter [14-20]. A well-known example is the promoter region of the developmentally regulated endo 16 gene of the sea urchin [12-14]. It is about ~2300 bp in length and consists of several clusters of target sites for distinct functions. Yuh et al. [12-14] have explored the function of each subregion of the endo 16 system and every target site within each subregion, using a cis-regulatory logic model.
However, a drawback of most current methods for inferring cis-regulatory circuits is that they rely on changing or deleting some binding site sequences (e.g., [12-14]), which may not provide intact functional information for reconstructing the cis-regulatory circuit. The deletion or mutation experiments may change or destroy the original cis-regulatory circuit structure. Using such data, one may lose significant interactions among transcription factors (TFs). Obviously, it is appealing to develop a method that can infer intact cis-regulatory circuits. Recently, there are some statistical and system level approaches to study the genome-wide transcription regulation and address cooperativity among TFs ([7-11,15,16]). Important advances have been made toward understanding transcriptional regulatory networks. One strategy infers global networks directly from whole genome microarray data, and another strategy focuses on the identification of shared cis elements in the promoters of co-regulated genes, signified by similar expression profiles [8]. In this paper we develop a new method to combine microarray data and TF-binding location data by chromatin immunoprecipitation [5,18] to study the regulatory and interaction functions of various cis elements with regard to the target gene. The data of Lee et al. 2002 [5] can reveal the in vivo physical interactions of TFs with their cis elements on the promoter and therefore can provide a more reliable view of functional interactions between TFs and cis elements. Combining these types of data and microarray data, we propose a novel cross-gene identification scheme to infer how multiple TFs coordinate to regulate gene transcription. Our approach is rather different from most existing statistical and system level methods for analyzing gene expression data. Our results show that this novel method is suitable for deciphering the complex TFs interactions and for predicting the gene expression. In addition, we also identify the dynamic regulatory functions of TFs interaction in the yeast cell cycle, which cannot be achieved by current methods.
Results
Characterizing the cis-regulatory circuit of a gene
In this study, there are two steps for characterizing the cis-regulatory circuit of a gene. The first is to find a cluster of genes that includes the gene of interest and a number of other genes each of which shares a subset of cis elements with the gene of interest. Assuming a certain regulatory function for each of the TFs that recognize the cis elements of genes in the cluster and certain interaction functions among the TFs of a circuit, we set up dynamic equations for the cis-regulatory circuits of the genes in the cluster to describe their expression profiles. Since each gene in the cluster shares some cis elements with the gene of interest, a matrix of cis elements for the cluster of genes can be constructed. In this model, the regulatory functions of individual TFs and the interactions among TFs can be estimated from microarray data. In the second step, a cross-gene identification scheme is developed with an array of expression profiles of genes in the cluster (e.g., Spellman et al., 1998 [18]) to identify regulatory functions of the TFs and their possible interactions for each gene in the cluster; the parameters are estimated by the least square estimation algorithm.
Finally, plugging these estimated regulatory functions and interactions into the dynamic equations, one can explicitly describe the cis-regulatory circuit of the gene of interest.
Choice of a cluster of genes
As an illustration, suppose some genes of the yeast cell cycle are of interest. We find a cluster of genes for each gene of interest according to their cis elements found in Simon et al. 2001 [4]. To simplify the analysis, we consider only the nine TFs that are currently known to be important cell cycle TFs of the yeast (i.e., Mbp1, Swi4, Swi6, Mcm1, Fkh1, Fkh2, Ndd1, Swi5, and Ace2). The cluster of genes for the gene of interest is called the reference gene cluster (RGC). In an RGC, we assume that each gene shares some cis elements of the gene of interest. Furthermore, the regulatory functions and the interactions of the TFs recognizing the same cis elements are assumed to be the same for all genes in the RGC. For example, in Figure 1a gene MFA2 is the gene of interest; it causes cell cycle arrest and is essential for mating in yeast [19]. This gene has three main cis elements, Ndd1, Mcm1 and Swi5 [4], from which we want to reconstruct the cis-regulatory circuit of MFA2 from yeast microarray data. The cis elements of MFA2 are denoted as follows:
MFA2:{Ndd1, Mcm1, Swi5}. (1)
Figure 1 Dynamic model of the cis-regulatory circuit of gene MFA2 (a) and of gene CLB2 (b). The genome-wide TF-binding location data obtained using chromatin immunoprecipitation [4] is used to identify the transcription factor binding motifs (cis elements). A binding transcription factor p has a regulatory function gp(t) and interacts with other recognizing TFs to produce the regulatory functions gp,q(t) and gp,q,r(t). These regulatory functions are the inputs of the cis-regulatory circuit and generate the dynamic output (i.e., the expression profile) of the target gene. Different phases of the cell cycle are indicated by the colored bar at the right lower corner.
Some genes chosen for the cluster and their cis elements are:
YAL022C:{Swi5}, YAR018C:{Ndd1, Mcm1}, YKL163W:{Mcm1, Swi5} ....
The cluster of genes can be represented by a connectivity matrix of cis elements as shown in Table 1 in which "1" denotes the connection of a cis element with a gene, while "0" means no connection. Similarly, the RGC for gene CLB2 can be represented by the connectivity matrix in Table 2.
Table 1 The reference gene clusters (RGCs) of MFA2. Target Gene MFA2 and the connectivities to cis elements.
Fkh1 Fkh2 Ndd1 Mcm1 Ace2 Swi5 Mbp1 Swi4 Swi6
MFA2 (YNL145W) 0 0 1 1 0 1 0 0 0
Reference genes cluster (RGC) and their connectivities to cis elements
ORF Fkh1 Fkh2 Ndd1 Mcm1 Ace2 Swi5 Mbp1 Swi4 Swi6 ORF Fkh1 Fkh2 Ndd1 Mcm1 Ace2 Swi5 Mbp1 Swi4 Swi6
YAL022C 0 0 0 0 0 1 0 0 0 YKL209C 0 0 0 1 0 0 0 0 0
YAR018C 0 0 1 1 0 0 0 0 0 YLR274W 0 0 0 1 0 0 0 0 0
YDR150W 0 0 1 0 0 0 0 0 0 YML050W 0 0 1 1 0 0 0 0 0
YFL026W 0 0 0 1 0 0 0 0 0 YML125C 0 0 0 0 0 1 0 0 0
YGL032C 0 0 0 1 0 0 0 0 0 YMR001C 0 0 1 0 0 0 0 0 0
YIL050W 0 0 0 0 0 1 0 0 0 YMR002W 0 0 1 0 0 0 0 0 0
YIL129C 0 0 0 0 0 1 0 0 0 YMR253C 0 0 0 1 0 0 0 0 0
YJL079C 0 0 1 1 0 0 0 0 0 YNL056W 0 0 1 1 0 0 0 0 0
YJL157C 0 0 0 1 0 0 0 0 0 YNL058C 0 0 1 1 0 0 0 0 0
YKL163W 0 0 0 1 0 1 0 0 0 YNL145W 0 0 1 1 0 1 0 0 0
YKL164C 0 0 0 1 0 1 0 0 0 YOR066W 0 0 0 1 0 0 0 0 0
Table 2 The reference gene clusters (RGCs) of CLB2. Target Gene CLB2 and the connectivities to cis elements.
Fkh1 Fkh2 Ndd1 Mcm1 Ace2 Swi5 Mbp1 Swi4 Swi6
CLB2 (YPR119W) 1 1 1 1 0 0 0 1 1
Reference genes cluster (RGC) and their connectivities to cis elements
ORF Fkh1 Fkh2 Ndd1 Mcm1 Ace2 Swi5 Mbp1 Swi4 Swi6 ORF Fkh1 Fkh2 Ndd1 Mcm1 Ace2 Swi5 Mbp1 Swi4 Swi6
YAR018C 0 0 1 1 0 0 0 0 0 YKR013W 0 0 0 0 0 0 0 1 1
YBR133C 0 1 0 0 0 0 0 0 0 YLR056W 0 0 0 0 0 0 0 1 1
YBR138C 1 0 1 0 0 0 0 0 0 YLR084C 0 1 1 1 0 0 0 1 0
YBR139W 1 0 1 0 0 0 0 0 0 YLR131C 0 1 1 1 0 0 0 0 0
YCL063W 1 1 0 0 0 0 0 0 0 YLR190W 0 1 1 1 0 0 0 0 0
YDL227C 0 0 0 0 0 0 0 1 1 YLR209C 1 0 0 0 0 0 0 0 0
YDR033W 0 1 1 0 0 0 0 0 0 YLR210W 1 0 0 0 0 0 0 0 0
YDR146C 0 1 1 1 0 0 0 0 0 YLR274W 0 0 0 1 0 0 0 0 0
YDR150W 0 0 1 0 0 0 0 0 0 YLR342W 0 0 0 0 0 0 0 1 1
YDR224C 0 0 0 0 0 0 0 1 1 YML050W 0 0 1 1 0 0 0 0 0
YDR225W 0 0 0 0 0 0 0 1 1 YML064C 1 1 0 0 0 0 0 0 0
YDR507C 0 0 0 1 0 0 0 1 1 YMR001C 0 0 1 0 0 0 0 0 0
YEL017W 1 0 0 0 0 0 0 0 0 YMR002W 0 0 1 0 0 0 0 0 0
YEL040W 0 1 0 1 0 0 0 1 1 YMR015C 1 0 0 0 0 0 0 1 0
YER001W 0 0 0 0 0 0 0 1 1 YMR183C 1 0 0 0 0 0 0 0 0
YFL026W 0 0 0 1 0 0 0 0 0 YMR253C 0 0 0 1 0 0 0 0 0
YGL032C 0 0 0 1 0 0 0 0 0 YMR305C 0 0 0 0 0 0 0 1 1
YGL038C 0 0 0 0 0 0 0 1 1 YMR307W 0 0 0 0 0 0 0 1 1
YGL116W 0 1 1 1 0 0 0 0 0 YNL056W 0 0 1 1 0 0 0 0 0
YGR014W 0 0 0 0 0 0 0 1 1 YNL058C 0 0 1 1 0 0 0 0 0
YGR099W 1 0 0 0 0 0 0 0 0 YNL231C 0 1 0 0 0 0 0 1 1
YGR151C 0 0 0 0 0 0 0 1 0 YNL300W 0 0 0 0 0 0 0 1 1
YGR152C 0 0 0 0 0 0 0 1 0 YOL011W 0 0 0 0 0 0 0 1 0
YGR153W 0 0 0 0 0 0 0 1 0 YOL030W 1 0 0 0 0 0 0 0 0
YGR221C 0 0 0 0 0 0 0 1 1 YOL114C 0 0 0 0 0 0 0 1 1
YGR279C 0 0 0 0 0 0 0 1 1 YOR066W 0 0 0 1 0 0 0 0 0
YHR061C 0 1 0 0 0 0 0 1 1 YOR073W 0 1 0 0 0 0 0 0 0
YIL056W 0 1 1 0 0 0 0 1 1 YOR372C 0 0 0 0 0 0 0 1 1
YIL121W 0 0 0 0 0 0 0 1 0 YPL032C 1 0 0 0 0 0 0 0 0
YIL123W 0 1 0 1 0 0 0 1 1 YPL116W 1 0 0 0 0 0 0 0 0
YIL158W 0 1 1 1 0 0 0 0 0 YPL127C 0 0 0 0 0 0 0 1 1
YJL051W 0 1 1 1 0 0 0 0 0 YPL141C 1 0 0 0 0 0 0 0 0
YJL079C 0 0 1 1 0 0 0 0 0 YPL155C 0 1 0 0 0 0 0 0 0
YJL157C 0 0 0 1 0 0 0 0 0 YPL163C 0 0 0 0 0 0 0 1 1
YJL158C 0 1 1 0 0 0 0 1 1 YPL255W 0 0 0 0 0 0 0 0 1
YJR054W 0 0 0 0 0 0 0 1 1 YPL256C 0 0 0 0 0 0 0 0 1
YJR092W 1 1 1 1 0 0 0 0 0 YPR013C 1 0 0 0 0 0 0 1 0
YJR110W 0 1 0 0 0 0 0 0 0 YPR119W 1 1 1 1 0 0 0 1 1
YKL096W 0 1 0 0 0 0 0 1 1 YPR149W 0 1 1 0 0 0 0 1 0
YKL103C 0 0 0 0 0 0 0 1 0 YPR159W 0 0 0 0 0 0 0 1 1
YKL209C 0 0 0 1 0 0 0 0 0
Dynamic modeling of cis-regulatory circuits
Figure 1 illustrates the leaky integrator-based dynamic models of the cis-regulatory circuits of two yeast genes (MFA2 and CLB2) [1,2]. The dynamics of gene expression can be modeled by a simple first-order nonlinear differential equation that is well established and analyzed in [17]; each model includes the possible regulatory functions of the individual TFs and possible interactions among the TFs. For the target gene MFA2 (Figure 1a), the cis-regulatory circuit is modeled by the following dynamic equation
Y˙MFA2(t)=gNdd1(t)+gMcm1(t)+gSwi5(t)+gNdd1,Mcm1(t)+ gNdd1,Swi5(t)+gMcm1,Swi5(t)+gNdd1,Mcm1,Swi5(t)−λMFA2YMFA2(t)+εMFA2(t), (2)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@C2AB@
where εMFA2(t) denotes the noise (data uncertainty), gNdd1(t), gMcm1(t) and gSwi5(t) are the regulatory functions of transcription factors Ndd1, Mcm1, and Swi5 or the incident transcriptional regulations at the Ndd1, Mcm1, and Swi5 cis elements, respectively, λMFA2 represents the mRNA decay rate of the target gene and we used the degradation rate measured by Wang et al. 2002 [20]. The gNdd1,Mcm1(t), gNdd1,Swi5(t), gMcm1,Swi5(t) and gNdd1,Mcm1,Swi5(t) denote the following nonlinear interactions among the three TFs:
gNdd1,Mcm1(t) =: f(gNdd1(t), gMcm1(t)),
gNdd1,Swi5(t) =: f(gNdd1(t), gSwi5(t)), (3)
gMcm1,Swi5(t) =: f(gMcm1(t), gSwi5(t)),
and
gNdd1,Mcm1,Swi5(t) =: f(gNdd1(t), gMcm1(t), gSwi5(t)). (4)
The biological meaning of Equation (2) is that the change in the mRNA expression level of gene MFA2 is due to the productions of regulatory functions of individual TFs and interactions among the TFs, i.e., gNdd1(t) + gMcm1(t) + gSwi5(t) + gNdd1,Mcm1(t) + gNdd1,Swi5(t) + gMcm1,Swi5(t) + gNdd1,Mcm1,Swi5(t), and -λMFA2YMFA2(t), which is the degradation of mRNA. Similarly, the cis-regulatory circuit of the target gene CLB2 in Figure 1b is modeled by
Y˙CLB2(t)=gFkh1(t)+gFkh2(t)+⋯+gFkh1,Fkh2(t)+⋯+gFhk1,Fkh2,Ndd1(t)+⋯ (5) −λCLB2YCLB2(t)+εCLB2(t).
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@B83F@
For simplicity, the indices of target genes and cis elements are denoted by numerical notation, so that the cis-regulatory circuit of gene i can be written as
Y˙i(t)=∑pvgp(t)+∑pqgp,q(t)+∑pqrgp,q,r(t)+⋯−λiYi(t)+εi(t), (6)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@780B@
where v is the total number of cis elements in gene i and its corresponding degradation rate λi measured by Wang et al. 2002 [20]. If λi is unavailable, it should be estimated together with the parameters gp(t), gp,q(t), gp,q,r(t) (see Methods). For the cis-regulatory circuits in Equations (2), (5) and (6), one obviously cannot estimate the multiple unknowns gp(t), gp,q(t), gp,q,r(t), … by only the expression profile (i.e., Yi(t)) of the ith target gene. However, since the functions gp(t), gp,q(t), gp,q,r(t), … are assumed to be the same for all genes in the RGC and since there are overlaps of cis elements among genes in this RGC, one can estimate these functions from an array of expression profiles Y1(t), Y2(t), …, YN(t) of the genes in the RGC simultaneously, taking advantage of cross information enhancement. The RGCs for MFA2 and CLB2 are shown in Table 1 and Table 2, respectively. In this situation, a cross-gene identification method is proposed as follows. By integrating the dynamic equations of cis-regulatory circuits for N genes in the RGC of the gene of interest, we obtain the following array of dynamic equations
(Y˙1(t)Y˙2(t)⋮⋮Y˙i(t)⋮⋮Y˙N(t))=(10⋯01⋯0⋯111⋯11⋯1⋯0⋮⋮⋮⋮⋮⋮⋮⋮01⋯00⋯0⋯1⋮⋮⋮⋮⋮⋮⋮⋮11⋯10⋯0⋯1)•(g1(t)g2(t)⋮g1,2(t)g1,3(t)⋮g1,2,3(t)⋮gp,q,r(t))−(λ1Y1(t)λ2Y2(t)⋮⋮λiYi(t)⋮⋮λNYN(t))+(ε1(t)ε2(t)⋮⋮εi(t)⋮⋮εN(t)). (7)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@277F@
In the cross-gene dynamic equations in (7), the process to identify regulatory functions gp(t), gp,q(t) and gp,q,r(t) from microarray data Yi(t), i = 1, 2,..., N is called the cross-gene identification approach, in which the regulatory functions gp(t), gp,q(t) and gp,q,r(t) are shared by genes in the RGC. Therefore, the estimation of the regulatory functions of one gene can also use the information from the profiles Y1(t), Y2(t), …, YN(t) of other genes in RGC to improve the identification ability of the regulatory functions to reconstruct the cis-regulatory circuit of the gene of interest, which is called cross information enhancement.
Remark 1 : Suppose that the gene of interest in Equation (6) has cis elements p = 1,..., v. Then all genes whose cis elements are subsets of these v cis elements are included in the same RGC of the gene of interest.
Cross-gene identification scheme
Since the number of functions gp(t), gp,q(t), gp,q,r(t), … is finite, we can estimate these functions if the number N of dynamic equations in Equation (7) is large enough. Equation (7) can be rewritten in an algebraic form
X(t) = A·G(t) + E(t), (8)
where
X(t)=(Y˙1(t)+λ1Y1(t)Y˙2(t)+λ2Y2(t)⋮⋮Y˙i(t)+λiYi(t)⋮⋮Y˙N(t)+λNYN(t)), G(t)=(g1(t)g2(t)⋮g1,2(t)g1,3(t)⋮g1,2,3(t)⋮gp,q,r(t)), E(t)=(ε1(t)ε2(t)⋮⋮εi(t)⋮⋮εN(t))
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@FECF@
Remark 2 : In order to calculate the derivatives in X(t) from undersampled data, a cubic spline interpolation method [21,22] is employed for curve fitting to obtain more accurate differential values and to learn more reliable models.
In Equation (8), X(t) can be calculated from microarray data for the RGC of the gene of interest, which can then be used to estimate the vector G(t) by the least squares method, leading to the following solution:
G^(t)=(ATA)-1ATX(t) (9)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuqGhbWrgaqcaiabcIcaOiabdsha0jabcMcaPiabg2da9iabcIcaOiabbgeabnaaCaaaleqabaGaemivaqfaaOGaeeyqaeKaeeykaKYaaWbaaSqabeaacqqGTaqlcqqGXaqmaaGccqqGbbqqdaahaaWcbeqaaiabdsfaubaakiabbIfayjabcIcaOiabdsha0jabcMcaPiaaxMaacaWLjaWaaeWaaeaacqaI5aqoaiaawIcacaGLPaaaaaa@43C0@
for all t. After G^(t)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuqGhbWrgaqcaiabcIcaOiabdsha0jabcMcaPaaa@30F6@ is estimated from Equation (9), the regulatory functions g1(t), …, g1,2(t), … and gp,q,r(t) in Equation (8) can be reconstructed for the genes in the RGC at every time point. If a cis-regulatory circuit is free of any function gp(t), gp,q(t), or gp,q,r(t), the value of the estimated function should be very small or zero. After the functions are estimated, they can be plugged into Equation (6) and the reconstruction of the cis-regulatory circuit of the gene of interest is completed. The flowchart for modeling, identifying and predicting a cis-regulatory circuit is shown in Figure 2.
Figure 2 The overall flowchart of the modeling, identification and prediction of a cis-regulatory circuit.
In order to obtain more accurate cis-regulatory circuits, the model should include the triple interactions among the recognizing TFs; i.e., the vector G(t) in Equation (7) should include gp,q,r(t).
Remark 3: If the degradation parameters λi in (6) are unavailable, the estimation procedure of G(t) and λi from equation (7) to equation (9) should be modified as in Methods.
Two application examples
I. The cis-regulatory circuit of MFA2
Suppose that the cis-regulatory circuit of the MFA2 gene is of interest. We construct a dynamic model of the cis-regulatory circuits of the genes in the RGC of MFA2 in Table 1 and then estimate the regulatory functions and interactions by the cross-gene identification scheme. The estimated transcriptional regulatory functions gp(t) and interactions gp,q(t) and gp,q,r(t) are shown as the insets in Figure 1a. These insets indicate that cis elements Ndd1, Mcm1 and Swi5 in MFA2 all have cell cycle regulatory abilities in the late G1 phase. In addition, every individual cis element has a positive regulatory function on the MFA2 gene; for example, the function g4(t) for Mcm1 has an obvious peak value in the transition phase late G1 of the cell cycle. Michaelis and Herskowitz [19] found that the MFA2 gene causes the cell cycle arrest at the G1 phase and is required for mating in yeast. Note that the interaction g3,6(t) between TFs Ndd1 and Swi5 is very weak or absent in the cell cycle. In contrast, the interaction g4,6(t) between TFs Mcm1 and Swi5 is dynamic; it has a high positive peak value in the late G1 phase, which coincides with MFA2's activity phase. This interaction seems to play an important role of positive regulation in this cis-regulatory circuit. On the other hand, the regulatory function g3,4,6(t) of the interaction among TFs Ndd1, Mcm1 and Swi5 is negative on gene MFA2. This regulation may repress the expression of gene MFA2 to make the expression decay to the steady state. If there is no repression function such as g3,4,6(t), the expression of MFA2 will increase and may disrupt in the cell cycle.
II. The cis-regulatory circuit of CLB2
Clb proteins are crucial cyclins for completing the G2/M transition of the mitotic cell cycle and the most typical one is the B-type mitotic cyclin Clb2, which is required for entry into mitosis [23]. Suppose that the cis-regulatory circuit of gene CLB2 is of interest. Fkh1, Fkh2, Ndd1, Mcm1, Swi4 and Swi6 have been identified as the TFs that bind to the promoter sequence of CLB2[3,4,24,25]. As shown in Figure 1b, the possible cis-regulatory circuit of CLB2 is very complex. Using the cross-gene identification scheme, we reconstructed the cis-regulatory functions shown in Figure 3. Although there are 41 possible regulatory functions, including gp(t), gp,q(t) and gp,q,r(t), only 20 regulatory functions are found to have nonzero values (Figure 3).
Figure 3 All estimated cis-regulatory functions, including the regulatory function gp(t) of each individual TF and the interactions gp,q(t) and gp,q,r(t) among the TFs that recognize the cis elements of the CLB2 gene. The numerical notation of regulatory functions is given in the box at the bottom of the figure. Different phases in the cell cycle are indicated by the colored bar near the right lower corner.
The two cis elements Fkh1 and Fkh2 are found to have very similar regulatory functions g1(t) and g2(t), in agreement with the experimental evidence that the forkhead family members Fkh1 and Fkh2 of transcription factors have overlapping roles in the control of the G2/M transition [25,26]. The regulatory function g2(t) of Fkh2 has a distinct cell cycle regulatory ability (Figure 3) and especially, the interaction function g2,3(t) between Fkh2 and Ndd1 has a strong regulatory contribution to the gene expression profile in the M/G1 phase. The regulatory function g4(t) of Mcm1 has two peaks. There is experimental evidence that Mcm1 is a member of an evolutionarily conserved class of transcription factors that have related to DNA binding sequences and dimerization domains. In addition, Mcm1 binds the early cell cycle box (ECB) that contains a Mcm1 cis element in the SWI4, CLN3, CDC6, and CDC47 promoters and activates M/G1-specific transcription [27].
The cell cycle genes that are activated during the late G1 or S phase have SBF or MBF sequence-specific transcription factors that bind the cis elements in their promoter region. SBF (the Swi4-Swi6 cell cycle box binding factor) is a heterodimer of Swi4 and Swi6 [3,28,29]. The regulatory functions g8(t) and g9(t) of Swi4 and Swi6 and their interaction function g8,9(t) are estimated using the dynamic expression model (Figure 3). It is well-known that neither Swi4 nor Swi6 alone has obvious cell cycle regulation ability, and indeed we found that g8(t) has only one peak and so shows no cycle and that g9(t) shows no capability of cell cycle regulation (Figure 3). In contrast, the combination of Swi6 and Swi4 to make the complex SBF enables the cis elements Swi6 and Swi4 to provide cell cycle regulation capacity; that is, the interaction function g8,9(t) of Swi6 and Swi4 has a peak during the G1/S phase of the cell cycle. Ndd1 and Fkh2 are bound to identical promoters throughout the cell cycle and their interaction g2,3(t) is an important transcriptional process targeted by the Cdk activity [24,30]. In addition, there is another obvious positive interaction g3,4,9(t) among Ndd1, Mcm1, and Swi6 (Figure 3). It has a large regulatory ability in the G1/S phase, which almost dominates the expression profile of CLB2. In contrast, g3,4(t) has a negative regulation contribution. We therefore propose that the key factor Swi6 in the interaction g3,4,9(t) is similar to its role in SBF and MBF. At any rate, our model suggests that Swi6 plays a key role in the interaction among Ndd1, Mcm1, and Swi6. This is a new finding in the cis-regulatory circuit of CLB2.
We also confirm the well-known interaction among Ndd1, Fkh2, and Mcm1 in the cis-regulatory circuits of the CLB2 and SWI5 genes because the interaction function g2,3,4(t) has a distinct regulatory ability in the G2/M phase of the cell cycle (Figure 3). Interestingly, the interaction function g3,4,9(t) among Ndd1, Mcm1 and Swi6 is about two times higher than any of the other functions in Figure 3. In addition, the interaction g1,3,4(t) among Fkh1, Ndd1 and Mcm1 is highly positive, the interaction g3,4,8(t) among Ndd1, Mcm1 and Swi4 is mildly positive, while the interaction g3,8,9(t) among Ndd1, Swi4 and Swi6 is negative. These observations are in agreement with the fact that the regulatory ability of an interaction among TFs is usually much stronger than that of an individual TF; in other words, there is synergy among TFs.
In summary, there are many experimental observations that support the cis-regulatory functions identified by the dynamic model and our model provides novel insights into the quantitative regulation of the cis-regulatory circuit of a gene of interest.
Support from expression phases of TFs in the cell cycle
In this paper, the question of why the strengths of regulatory functions in the cis-regulatory circuits are different in different phases of the cell cycle is investigated. Based on the mRNA expression profiles of transcription factor genes from experiments, the distribution of the expressions of the nine TF genes in the different phases of the cell cycle is shown in Figure 4. In support of our results, the large positive interaction functions (peaks) among a set of TFs always occur during the expression phases of the genes of the interacting TFs. For example, for the cis-regulatory circuit of MFA2 (Figure 1a), there is an obvious peak for the function g4,6(t) of the interaction between TFs Mcm1 and Swi5 during the M phase, and in Figure 4, this peak indeed occurs during the expression phases of the two TF genes [18,27]. As another example, for the cis-regulatory circuit of CLB2, there is a strong interaction (g3,4,9 (t)) among Ndd1, Mcm1 and Swi6 starting from the G2 phase (Figure 3). We can therefore infer that the expression of a cell cycle gene in a specific phase of the cell cycle needs a specific inducing signal, which is mainly from the interactions of certain specific TFs that bind the cis elements of the gene.
Figure 4 The gene expression phases match the main regulatory functions of TFs. It is seen that the main interaction functions of TFs have a peak value and always occur during or soon after the mRNA expression phases of the corresponding genes. For example, the gene MFA2 has the main interaction regulatory function gMcm1,Swi5(t) which has a peak during the expression phases between the TFs Mcm1 and Swi5 by identifying the cis-regulatory circuit. As another example, the gene UTR2 has the main interaction regulatory function gSwi4,Swi6(t), which has a peak during the expression phases between the TFs Swi4 and Swi6 by identifying the cis-regulatory circuit. These results indicate that the main regulatory functions have a peak value phase to match the gene expression phase. Therefore, we can estimate the gene expression phase by identifying the main regulatory function. The width of a colored band in the inner circle is approximately proportional to the expression level of the TF gene of interest in the cell cycle. A pink line points to the main expression phase of a target gene in the pink box. Different phases in the cell cycle are indicated by the colored bar at the right lower corner.
Accuracy of reconstructed cis-regulatory circuits
The accuracy of the reconstructed cis-regulatory circuit of a gene can be evaluated by reconstructing the expression profile of the gene using the reconstructed cis-regulatory circuit
Y^˙i(t)=∑pvg^p(t)+∑pqg^p,q(t)+∑pqrg^p,q,r(t)+⋯−λiY^i(t), (10)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@7216@
where g^p(t)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuWGNbWzgaqcamaaBaaaleaacqWGWbaCaeqaaOGaeiikaGIaemiDaqNaeiykaKcaaa@32D7@, g^p,q(t)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuWGNbWzgaqcamaaBaaaleaacqWGWbaCcqGGSaalcqWGXbqCaeqaaOGaeiikaGIaemiDaqNaeiykaKcaaa@3522@ and g^p,q,r(t)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuWGNbWzgaqcamaaBaaaleaacqWGWbaCcqGGSaalcqWGXbqCcqGGSaalcqWGYbGCaeqaaOGaeiikaGIaemiDaqNaeiykaKcaaa@376F@ have been estimated by the cross-gene identification scheme. The reconstructed profile and the observed profile are compared in Figure 5. We find that if the number of the cis elements of a gene is large enough, the reconstructed expression profile is very accurate; otherwise, it may be inaccurate. Fortunately, although the reconstructed expression profile is not accurate in some genes, the trend of the expression profile for a gene is always correct.
Figure 5 Comparison between the actual gene expression profiles (with cubic spline) and the reconstructed gene expression profiles. The examples shown were randomly chosen. The reconstructed gene expression profiles were obtained by integrating the estimated cis-regulatory functions and the chromatin immunoprecipitation data. When there is only one blue line in a figure it means that the reconstructed function is very close to the actual gene expression profile. Different phases in the cell cycle are indicated by the colored bar.
Prediction of gene expression profile
In the above, each regulatory circuit was identified using 95% of genes in its RGC, and the remaining 5% of genes in the RGC will now be used for predicting expression profiles, i.e., for cross validation. Our cross-gene identification scheme assumes that the regulatory functions of TFs are universal in the cluster of genes with similar functions. Under this assumption, our method should be able to predict the expression profiles of other genes in the same cluster that have not been employed in Equation (7) to reconstruct the cis-regulatory circuits. This is one way to validate our model.
In Figure 6, we randomly chose 100 RGCs to test the prediction accuracy; that is, for each RGC we show the prediction result for one of the unused genes. Figure 6a shows the predicted target gene and the mean square error (MSE) of prediction results which has the maximum of 2.055 and the minimum of 0.025. In addition, three examples of the detailed comparison between the actual and the predicted gene expression profiles are shown in Figure 6b. In general, the predicted profiles are satisfactory approximations of the observed expression profiles. We found that the smaller the number of the cis elements, the less accurate the prediction results. However, if some cis elements of a gene have strong regulatory functions, the expression profiles of this gene can be predicted accurately even when the number of cis elements is small. If some genes in the RGC have the same cis elements but have different observed gene expression profiles, these expression profiles will lead to poor estimation of parameters. This is the main cause of prediction error. Why does this situation arise? It may be that some cis elements of these genes have not yet been identified or there are some errors in the inference of the cis elements. For example, MMR1 may have another cis element Gcr2 [5] and this may be why the predicted profile is quite different from the observed profile (Figure 6b).
Figure 6 The regulatory functions integrated with ChIP-chip data to predict gene expression profiles. To test the prediction performance of our model, (a) 100 yeast cell cycle genes that have not been employed in the reconstruction of the cis-regulatory circuits are randomly chosen from their corresponding reference gene clusters (RGCs). The maximum mean square error (MSE) of prediction results is 2.055 and the minimum is 0.025. (b) Three examples of the comparison between the actual (blue) and the predicted (red) gene expression profiles. Different phases in the cell cycle are indicated by the colored bar.
Discussion
In contrast to current methods, our method uses all possible expression profile information from the cluster of genes to reconstruct the cis-regulatory circuit of a target gene. In particular, our method is capable of extracting dynamic interactions among TFs. For this reason, the analysis and interpretation of output expression profiles become straightforward. Therefore, our method has a high potential for applications such as studying variations of the cis-regulatory circuit of the same gene in different yeast strains to investigate the regulatory evolution of the gene.
The contributions of this study include: (1) a nonlinear dynamical model is developed for cis-regulatory circuits in terms of regulatory functions and interactions among TFs, (2) a cross-gene identification scheme is proposed to estimate many parameters involved in the dynamical model of cis-regulatory circuits from the expression profiles of genes in the reference gene cluster, (3) a detailed identification of the dynamic cis-regulatory abilities of TFs, which vary with time, and (4) a gene expression prediction method is developed by the proposed dynamic cis-regulatory circuit, assuming that the cis-regulatory functions of the same TFs in different circuits are the same. Three advantages of our method over current methods are that the cis-regulatory circuit is constructed with the circuit structure intact, that it uses the expression profiles of many genes simultaneously to obtain extra information, and that it is dynamic and quantitative.
Significantly, our model not only can confirm known regulations but also can provide conjectures for experimental verification. Consider the key positive cis-regulatory function g4,6(t) in Figure 1a. We propose that during the expression of gene MFA2, the transcription factor Ndd1 (the G2/M phase) communicates with the transcription factor Mcm1 (the M phase) to transmit a specific signal to induce the expression of the MFA2 gene. Such conjectures from the reconstructed cis-regulatory circuits may be useful for studying the regulatory evolution of genes by comparing the cis-regulatory functions of different strains, or for predicting the gene expression behavior before conducting an experiment.
However, we found poor results in some cases. For example, in Figure 3, we were unable to find the basal regulatory function g9(t) of individual Swi6 or the interaction g2,4(t) between Fkh2 and Mcm1 for gene CLB2 [31]. These regulatory functions have not been identified by our scheme because they have no obvious specific phase regulatory ability. Besides, from the cis-regulatory functions in Figures 1 and 2, several cis-regulatory function profiles did not show a clear periodicity. A possible reason may be that the original microarray data were noisy and the use of cubic spline interpolation and linear transformation of microarray data in our scheme had introduced new noise and distortions. Most probably, the cis element information used to construct the cis-regulatory circuits under yeast cell cycle is not complete; only nine significant cis elements were considered in this study to reduce the complexity of the mathematical model. Another possible source of error is that we have not considered the order of the cis elements on the promoter region, which may affect the strength of the interaction between TFs [36]. Such differences, however, can be incorporated by putting, say both gp,q(t) and gp,q,r(t), into the model.
In view of the facts that there are uncertainties about the cis elements of some of the genes studied and that microarray data are noisy, it is remarkable that our method gave accurate results for the expression profiles of the majority of the cell cycle genes studied and also gave fairly accurate predictions of the expression profiles of other cell cycle genes. In the future, if better cis elements data and more accurate and longer gene expression profile data become available, we should be able to improve the reconstruction of cis-regulatory circuits. Also, our approach may be extended to reconstruct cis-regulatory circuits in more diverse conditions and more complex eukaryotes. After cis-regulatory circuits are accurately described by explicit dynamical equations, some applications will be straightforward.
Conclusion
In this study, we assume that the regulatory functions of the same cis elements and the interaction functions among their TFs are similar across genes within the cluster of genes with overlapping cis elements; i.e., the regulatory functions and interaction functions are universal in this cluster of genes. Under this assumption, the cross-gene identification scheme takes advantage of cross-information enhancement to improve the accuracy of parameter estimation. The number of genes used should be large enough, so that we have a large number of outputs (i.e., their microarray data) for parameter estimation.
After the parameters of the cis-regulatory circuits of interest have been estimated, the circuits can be explicitly described by plugging these parameters into their corresponding dynamic equations. Moreover, these estimated functions and interactions can be used to predict the expression profiles of other genes that share the same cis-regulatory elements but have not been used to identify the cis-regulatory circuits. In this manner, we can evaluate the performance of the proposed dynamic model of cis-regulatory circuits. From a number of examples, we have found that the predicted results are in most cases satisfactory, confirming the validity of the proposed dynamic model of cis-regulatory circuits. Our modeling represents a new approach to studying cis-regulatory circuits from cross-gene expression data. It is a systems biology approach because we consider the regulatory circuits of many genes and many TFs at the same time and we use system identification techniques to estimate the parameters of the circuits. The results of expression prediction from experimental data suggest that our novel approach is suitable for deciphering the regulatory functions and the cooperativity of the TFs that regulate the expression of a gene.
Methods
Experimental data
To identify the cis-regulatory circuit of a gene of interest in the yeast cell cycle, we apply our approach to the data of Spellman et al. 1998 [18], which contains expression profiles of 6178 open reading frames (ORFs) in the yeast Saccharomyces cerevisiae during the cell cycle [33]. Our analysis was applied to the α-factor arrest data set. The raw data were transformed into a linear scale from the original log ratio carried out by Spellman et al. 1998 [18]. To reduce the effect of noise and to overcome undersampled microarray data in the estimation of cis-regulatory circuits, the cubic spline was used for data interpolation and smoothing to obtain a less sensitive first derivative of the expression pattern and to learn a more reliable model. Furthermore, the noise is modeled in the noise term εi(t) in Equation (7).
From the RGC of the gene of interest, the cross-gene identification method from Equations (8) to (9) is employed to reconstruct the cis-regulatory circuit. The connectivity information between TFs and their target genes was obtained from the yeast cell cycle analysis [4]. We focused on the nine transcription factors that have been identified to play important roles in the transcription regulation of a set of yeast genes whose expressions are cell-cycle dependent; these nine transcription factors are Mbp1, Swi4, Swi6, Mcm1, Fkh1, Fkh2, Ndd1, Swi5, and Ace2 [24,26,27,34] (See additional file 1: Table for the original data used to perform this analysis).
Estimation of degradation rate
If the mRNA degradation rate λi in Equation (6) has not been estimated experimentally, it should be estimated together with the parameters gp(t), gp,q(t), gp,q,r(t),. The algorithm to estimate the λi is described as follows. First, Equation (6) is changed to
Y˙i(t)=∑pvgp(t)+∑pqgp,q(t)+∑pqrgp,q,r(t)+⋯−λi(t)Yi(t)+εi(t), (11)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@7BEB@
where v is the total number of cis elements in gene i, and the degradation rate is substituted as λi(t). Similarly, since the functions gp(t), gp,q(t), gp,q,r(t), … are assumed to be the same for all genes in the RGC and since there are overlaps of cis elements among genes in this RGC, one can estimate these functions from an array of expression profiles Y1(t), Y2(t), …, YN(t) of the genes in the RGC simultaneously, taking advantage of cross information enhancement. The RGCs for MFA2 and CLB2 are shown in Table 1 and Table 2, respectively.
Second, by integrating the dynamic equations of cis-regulatory circuits for N genes in the RGC of the gene of interest, we obtain the following array of dynamic equations
(Y˙1(t)Y˙2(t)⋮⋮Y˙1(t)⋮⋮Y˙N(t))=(10⋯01⋯0⋯111⋯11⋯1⋯0⋮⋮⋮⋮⋮⋮01⋯00⋯0⋯1⋮⋮⋮⋮⋮⋮11⋯10⋯01|−Y1(t)0↔00−Y2(t)⋱↕−Y1(t)↕⋱00↔0−YN(t))•(g1(t)g2(t)⋮g1,2(t)g1,3(t)⋮g1,2,3(t)⋮gp,q,r(t)λ1(t)λ2(t)⋮⋮λi(t)⋮⋮λN(t))+(ε1(t)ε2(t)⋮⋮εi(t)⋮⋮εN(t)). (12)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@17D5@
In the dynamic equations in (12), the regulatory functions gp(t), gp,q(t) and gp,q,r(t) are shared by genes in the RGC. Therefore, the estimation of these functions from expression profiles Y1(t), Y2(t), …, YN(t) can use also information from other genes to enhance our ability to reconstruct the cis-regulatory circuit of the gene of interest.
Finally, since the number of functions gp(t), gp,q(t), gp,q,r(t), … is finite, we can estimate these functions and the decay rates λ1(t), λ2(t), …, λN(t) if the number N of dynamic equations in Equation (12) is large enough. Equation (12) can be rewritten in an algebraic form
X˜(t)=A˜(t)⋅G˜(t)+E(t), (13)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaaieaacuWFybawgaacaiabcIcaOiabdsha0jabcMcaPiabg2da9iqb=feabzaaiaGaeiikaGIaemiDaqNaeiykaKIaeyyXICTaf83raCKbaGaacqGGOaakcqWG0baDcqGGPaqkcqGHRaWkcqWFfbqrcqGGOaakcqWG0baDcqGGPaqkcqGGSaalcaWLjaGaaCzcaiabcIcaOiabigdaXiabiodaZiabcMcaPaaa@47BA@
where
X˜(t)=(Y˙1(t)Y˙2(t)⋮⋮Y˙i(t)⋮⋮Y˙N(t)), G˜(t)=(g1(t)g2(t)⋮g1,2(t)⋮gp,q,r(t)λ1(t)λ2(t)⋮⋮λN(t)), E(t)=(ε1(t)ε2(t)⋮⋮εi(t)⋮⋮εN(t)).
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@C6D3@
Then, we have the following dynamic equations for all time profiles
(
X˜(t1)
X˜(t2)⋮
X˜(tm))=(
A˜(t1)0↔00
A˜(t2)↕↕⋱00↔0
A˜(tm))·(
G˜(t1)
G˜(t2)⋮
G˜(tm))+(E(t1)E(t2)⋮E(tm)). (14)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@94B1@
Let us denote the above equations in the following simple algebraic form
X˜=Φ˜⋅Θ˜+M. (15)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuqGybawgaacaiabg2da9iqbfA6agzaaiaGaeyyXICTafuiMdeLbaGaacqGHRaWkcqqGnbqtcqqGUaGlcaWLjaGaaCzcaiabbIcaOiabbgdaXiabbwda1iabbMcaPaaa@3C07@
In Equation (15),
X˜
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuqGybawgaacaaaa@2DF4@ and Φ˜
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuqHMoGrgaacaaaa@2E37@ can be calculated from microarray data for the RGC of the gene of interest, which can then be used to estimate the vector Θ˜^
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuqHyoqugaacgaqcaaaa@2E43@ by the least squares method, leading to the following solution:
Θ˜^=(Φ˜TΦ˜)−1Φ˜
X˜. (16)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuqHyoqugaacgaqcaiabg2da9iabcIcaOiqbfA6agzaaiaWaaWbaaSqabeaacqWGubavaaGccuqHMoGrgaacaiabcMcaPmaaCaaaleqabaGaeyOeI0IaeGymaedaaOGafuOPdyKbaGaacuqGybawgaacaiabc6caUiaaxMaacaWLjaGaeiikaGIaeGymaeJaeGOnayJaeiykaKcaaa@401C@
After Θ˜^
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuqHyoqugaacgaqcaaaa@2E43@ is estimated from Equation (16), the regulatory functions
G˜(t)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuqGhbWrgaacaiabcIcaOiabdsha0jabcMcaPaaa@30F5@ in Equation (13) can be reconstructed for the genes in the RGC at every time point. However, in Equation (12), the degradation rate λ(t) is a time-varying function and is affected by both the error terms and experimental data. Therefore, in order to reduce the influence on degradation rate, we simplify Equation (12) and average the negative gradients of λ(t) to obtain the constant value λ^
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuaH7oaBgaqcaaaa@2E72@. Then the estimated λ^
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuaH7oaBgaqcaaaa@2E72@ is substituted into Equation (7) to re-identify the cis-regulatory functions to derive the final regulatory functions. Using this procedure, we can avoid the effects of the time-varying function λ(t) on the identification process and reduce the influence by different experimental data. Hence, the degradation rate can be estimated. After the functions are estimated, they can be plugged into Equation (11) and then the reconstruction of the cis-regulatory circuit of the gene of interest is completed.
Authors' contributions
L.H. Lin carried out the model design and computation of this study, and drafted the manuscript. H.C. Lee participated in the design of the study and drafted the manuscript. W.H. Li amended and improved the design and the presentation of the study. B.S. Chen gave the topic and suggestions and was responsible for the entire study. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
The cell cycle genes and their connectivities to cis elements. 769 cell cycle genes defined by Spellman et al. [18] were selected from a total of 6178 genes in the data set. "1" denotes the connection of a cis element with a gene, while "0" means no connection. The main cis element data were compiled from the data set of Simon et al. 2001 [4] by choosing a P value (significance level) ≤ 0.0015. Under this threshold, many interactions among cis elements for genes confirmed by the conventional data are included [35,36]. Additionally, we modified some cis element data, using well-known experimental information to correct false negatives [35,36].
Click here for file
Acknowledgements
We thank Ming-Che Shih, Geoff Morris, Jake Byrnes, Josh Rest and Ya-Wen Chang for helpful comments. This study was supported by an NSC grant NSC 91-2321-B-007-002, by Academia Sinica, Taiwan and by NIH grants.
==== Refs
Shea M Ackers G The OR control system of bacteriophage lambda. A physical-chemical model for gene-regulation J Molec Biol 1985 181 211 230 3157005 10.1016/0022-2836(85)90086-5
Mjolsness E Sharp DH Reinitz J A connectionist model of development J Theor Biol 1991 152 429 453 1758194
Iyer VR Horak CE Scafe CS Botstein D Snyder M Brown PO Genomic binding sites of the yeast cell-cycle transcription factors SBF and MBF Nature 2001 409 533 538 11206552 10.1038/35054095
Simon I Barnett J Hannett N Harbison CT Rinaldi NJ Volkert TL Wyrick JJ Zeitlinger J Gifford DK Jaakkola TS Young RA Serial regulation of transcriptional regulators in the yeast cell cycle Cell 2001 106 697 708 11572776 10.1016/S0092-8674(01)00494-9
Lee TI Rinaldi NJ Robert F Odom DT Bar-Joseph Z Gerber GK Hannett NM Harbison CT Thompson CM Simon I Zeitlinger J Jennings EG Murray HL Gordon DB Ren B Wyrick JJ Tagne JB Volkert TL Fraenkel E Gifford DK Young RA Transcriptional regulatory networks in Saccharomyces cerevisiae Science 2002 298 799 804 12399584 10.1126/science.1075090
Gao F Foat BC Bussemaker HJ Defining transcriptional networks through integrative modeling of mRNA expression and transcription factor binding data BMC Bioinformatics 2004 5 31 15113405 10.1186/1471-2105-5-31
Wei H Kaznessis Y Inferring gene regulatory relationships by combining target-target pattern recognition and regulator-specific motif examination Biotechnol Bioeng 2005 89 53 77 15540196 10.1002/bit.20305
Haverty PM Hansen U Weng Z Computational inference of transcriptional regulatory networks from expression profiling and transcription factor binding site identification Nucleic Acids Res 2004 32 179 188 14704355 10.1093/nar/gkh183
Kato M Hata N Banerjee N Futcher B Zhang MQ Identifying combinatorial regulation of transcription factors and binding motifs Genome Biol 2004 5 R56 15287978 10.1186/gb-2004-5-8-r56
Nachman I Regev A Friedman N Inferring quantitative models of regulatory networks from expression data Bioinformatics 2004 20 I248 I256 15262806 10.1093/bioinformatics/bth941
Wang W Cherry JM Nochomovitz Y Jolly E Botstein D Li H Inference of combinatorial regulation in yeast transcriptional networks: a case study of sporulation Proc Natl Acad Sci USA 2005 102 1998 2003 15684073 10.1073/pnas.0405537102
Yuh CH Moore JG Davidson EH Quantitative functional interrelations within the cis-regulatory system of the S. purpuratus Endo16 gene Development 1996 122 4045 4056 9012524
Yuh CH 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
Yuh CH Bolouri H Davidson EH Cis-regulatory logic in the endo16 gene: switching from a specification to a differentiation mode of control Development 2001 128 617 629 11171388
Banerjee N Zhang MQ Identifying cooperativity among transcription factors controlling the cell cycle in yeast Nucleic Acids Res 2003 31 7024 7031 14627835 10.1093/nar/gkg894
Das D Banerjee N Zhang MQ Interacting models of cooperative gene regulation Proc Natl Acad Sci USA 2004 101 16234 16239 15534222 10.1073/pnas.0407365101
Chen HC Lee HC Lin TY Li WH Chen BS Quantitative characterization of the transcriptional regulatory network in the yeast cell cycle Bioinformatics 2004 20 1914 1927 15044243 10.1093/bioinformatics/bth178
Spellman PT Sherlock G Zhang MQ Iyer VR Anders K Eisen MB Brown PQ Bostein D Futcher B Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cererisiae by microarray hybridization Mol Biol Cell 1998 9 3273 3297 9843569
Michaelis S Herskowitz I The a-factor pheromone of Saccharomyces cerevisiae is essential for mating Mol Cell Biol 1988 8 1309 1318 3285180
Wang Y Liu CL Storey JD Tibshirani RJ Herschlag D Brown PO Precision and functional specificity in mRNA decay Proc Natl Acad Sci USA 2002 99 5860 6865 11972065 10.1073/pnas.092538799
de Boor C A Practical Guide to Splines 1978 New York: Springer-Verlag Press
Bar-Joseph Z Gerber G Gifford DK Jaakkola TS Simon I A new approach to analyzing gene expression time series data The Sixth Annual International Conference on Research in Computational Molecular Biology (RECOMB) 2002 Washington DC 39 48
Maher M Cong F Kindelberger D Nasmyth K Dalton S Cell cycle-regulated transcription of the CLB2 gene is dependent on Mcm1 and a ternary complex factor Mol Cell Biol 1995 15 3129 3137 7760809
Koranda M Schleiffer A Endler L Ammerer G Fork-head-like transcription factors recruit Ndd1 to the chromatin of G2/M-specific promoters Nature 2000 406 94 98 10894549 10.1038/35017589
Zhu G Spellman PT Volpe T Brown PO Botstein D Davis TN Futcher B Two yeast forkhead genes regulate the cell cycle and pseudohyphal growth Nature 2000 406 90 94 10894548 10.1038/35021046
Kumar R Reynolds DM Shevchenko A Goldstone SD Dalton S Forkhead transcription factors, Fkh1p and Fkh2p, collaborate with Mcm1p to control transcription required for M-phase Curr Biol 2000 10 896 906 10959837 10.1016/S0960-9822(00)00618-7
McInerny CJ Partridge JF Mikesell GE Creemer DP Breeden LL A novel Mcm1-dependent element in the SWI4, CLN3, CDC6, and CDC47 promoters activates M/G1-specific transcription Genes Dev 1997 11 1277 1288 9171372
Dirick L Moll T Auer H Nasmyth K A central role for SWI6 in modulating cell cycle start-specific transcription in yeast Nature 1992 357 508 513 1608451 10.1038/357508a0
Ho Y Costanzo M Moore L Kobayashi R Andrews BJ Regulation of transcription at the Saccharomyces cerevisiae start transition by Stb1, a Swi6-binding protein Mol Cell Biol 1999 19 5267 5278 10409718
Loy CJ Lydall D Surana U Ndd1, a high-dosage suppressor of cdc28-1N, is essential for expression of a subset of late-S-phase-specific genes in Saccharomyces cerevisiae Mol Cell Biol 1999 19 3312 3327 10207056
Lim FL Hayes A West AG Pic-Taylor A Darieva Z Morgan BA Oliver SG Sharrocks AD Mcm1p-induced DNA bending regulates the formation of ternary transcription factor complexes Mol Cell Biol 2003 23 450 461 12509445 10.1128/MCB.23.2.450-461.2003
Beer MA Tavazoie S Predicting gene expression from sequence Cell 2004 117 185 198 15084257 10.1016/S0092-8674(04)00304-6
Yeast Cell Cycle Analysis Project
McBride HJ Yu Y Stillman DJ Distinct regions of the Swi5 and Ace2 transcription factors are required for specific gene activation J Biol Chem 1999 274 21029 21036 10409653 10.1074/jbc.274.30.21029
Cross FR Hoek M McKinney JD Tinkelenberg AH Role of Swi4 in cell cycle regulation of CLN2 expression Mol Cell Biol 1994 14 4779 4787 8007977
Costanzo M Schub O Andrews B G1 Transcription factors are differentially regulated in Saccharomyces cerevisiae by the Swi6-binding protein Stb1 Mol Cell Biol 2003 23 5064 5077 12832490 10.1128/MCB.23.14.5064-5077.2003
|
16232312
|
PMC1283971
|
CC BY
|
2021-01-04 16:27:46
|
no
|
BMC Bioinformatics. 2005 Oct 18; 6:258
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-258
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1421622567310.1186/1471-2164-6-142Research ArticleMacrodissection versus microdissection of rectal carcinoma: minor influence of stroma cells to tumor cell gene expression profiles de Bruin Elza C [email protected] de Pas Simone [email protected] Esther H [email protected] Eijk Ronald [email protected] der Zee Minke MC [email protected] Marcel [email protected] Wezel Tom [email protected] Corrie AM [email protected] Krieken J Han JM [email protected] Jan Paul [email protected] de Velde Cornelis JH [email protected] Paul HC [email protected] Lucy TC [email protected] Department of Clinical Oncology, Leiden University Medical Center, Albinusdreef 2, 2333 ZA Leiden, The Netherlands2 Department of Pathology, Leiden University Medical Center, Albinusdreef 2, 2333 ZA Leiden, The Netherlands3 Department of Pathology, University Medical Center St. Radboud, Geert Grooteplein-Zuid 10, 6525 GA Nijmegen, The Netherlands4 Department of Surgery, Leiden University Medical Center, Albinusdreef 2, 2333 ZA Leiden, The Netherlands5 Department of Medical Statistics, Leiden University Medical Center, Wassenaarseweg 62, 2333 AL, Leiden, The Netherlands2005 14 10 2005 6 142 142 17 12 2004 14 10 2005 Copyright © 2005 de Bruin et al; licensee BioMed Central Ltd.2005de Bruin 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 molecular determinants of carcinogenesis, tumor progression and patient prognosis can be deduced from simultaneous comparison of thousands of genes by microarray analysis. However, the presence of stroma cells in surgically excised carcinoma tissues might obscure the tumor cell-specific gene expression profiles of these samples. To circumvent this complication, laser microdissection can be performed to separate tumor epithelium from the surrounding stroma and healthy tissue. In this report, we compared RNAs isolated from macrodissected, of which only surrounding healthy tissue had been removed, and microdissected rectal carcinoma samples by microarray analysis in order to determine the most reliable approach to detect the expression of tumor cell-derived genes by microarray analysis.
Results
As microdissection yielded low tissue and RNA quantities, extra rounds of mRNA amplification were necessary to obtain sufficient RNA for microarray experiments. These second rounds of amplification influenced the gene expression profiles. Moreover, the presence of stroma cells in macrodissected samples had a minor contribution to the tumor cell gene expression profiles, which can be explained by the observation that more RNA is extracted from tumor epithelial cells than from stroma.
Conclusion
These data demonstrate that the more convenient procedure of macrodissection can be adequately used and yields reliable data regarding the identification of tumor cell-specific gene expression profiles.
==== Body
Background
Microarray technology permits simultaneous analyses of the expression profiles of thousands of genes. These analyses allow identification of profiles correlating with prognosis and permit tumor classifications [1-3], but can also be used to identify genes that are involved in several molecular processes, like carcinogenesis, metastasis and responses to treatment (reviewed by ref [4]).
To ensure that the expressions of tumor cell-derived genes are identified by microarray analysis of surgically excised carcinomas, the samples can be enriched for tumor cells by removing the surrounding healthy tissue. However, besides tumor epithelium with infiltrating cells, these macrodissected samples contain stroma cells as well. Evidently, after RNA isolation of such macrodissected samples, tumor epithelium-derived RNA cannot be separated from RNA specific for stroma. Although informative, the presence of stroma might obscure the tumor cell gene expressions, thereby preventing accurate data on tumor cell expression profiles. Because in rectal carcinoma the percentages of stroma versus tumor epithelium vary widely among patients, this high variation might complicate comparisons of different tumor samples even more.
To circumvent this problem, microdissection, such as Laser Microdissection and Pressure Catapulting (LMPC), can be used to select tumor epithelial cells exclusively. Although contamination of infiltrating cells will in this case also be present and important micro environmental information of the tumor cells will be missed, RNA extracted from such microdissected samples is expected to be more specific for tumor epithelial gene expression than RNA isolated from macrodissected samples. Comparisons of gene expression profiles of a small number of carcinoma samples obtained using macrodissection or microdissection, indeed led to the conclusion that stroma cells disturb the tumor gene expression profiles [5]. However, it has also been demonstrated that some degradation of RNA occurs during the lengthy procedure of laser capture microdissection, resulting in a decreased correlation between macro- and randomly microdissected samples [6]. Another disadvantage of microdissection can be the limited amount of extracted RNA, requiring an extra amplification round to get sufficient RNA for microarray experiments. There are several publications addressing the effect of amplification on gene expression profiles. T7 polymerase-based mRNA amplification is demonstrated to be reproducible and to maintain the relative abundances of mRNA transcripts, although lower correlation coefficients are always observed when amplified samples are compared to non-amplified samples [7-9]. This amplification effect becomes more serious with less starting material, which is the case for microdissected samples. In addition, a second round of amplification does have a further effect on reproducibility [10,11].
In this study, we determined the most reliable way to detect the expression of tumor cell-derived genes by microarray analysis: macrodissection or microdissection. Comparing gene expression profiles of macrodissected and microdissected rectal carcinoma samples in the same experimental setting allowed evaluation of the effect of a second round of RNA amplification as well as evaluation of the presence of varying amounts of stroma. Quantification of both effects demonstrated that the second amplification round had a high impact on gene expression profiles. In addition, epithelial tumor cells as compared to stroma cells had a much higher contribution to gene expression profiles than is expected from the quantified surface percentage. We conclude that the obscuring effect of stroma on the tumor epithelium gene expression profiles appears to be minimal and that therefore in clinical settings the convenient procedure of macrodissection is the preferable method to examine rectal carcinomas by microarray analysis.
Results
Gene expression profiles of macrodissected and microdissected rectal carcinomas
In the panel of excised rectal carcinoma samples used for this study, a high variation in surface percentages of tumor epithelium versus stroma was observed; percentages of epithelium ranged from 11 to 82% (Table 1). In order to compare macro- and microdissection of these carcinoma samples in microarray experiments, RNA was extracted from carcinoma tissue where surrounding healthy tissue had been removed (macrodissection), as well as from tumor epithelium isolated by LMPC (microdissection) of the same carcinoma samples. The microdissection procedure of tumor epithelium resulted on average in 30 ng of total RNA. Because 1 μg of mRNA is normally required for microarray experiments, two rounds of mRNA amplification were necessary, yielding on average 15 μg aRNA. For macrodissected samples one round of mRNA amplification was sufficient to get an adequate amount of aRNA. To be able to examine the effect of the second round of mRNA amplification on the gene expression profiles, several macrodissected samples were amplified a second round as well (Table 1).
Table 1 Percentages epithelial tumor surface and used amplification scheme
sample % tumor amplification rounds
epithelium macro micro
tumor stroma
1 11 1 2 nd
2 15 1, 2 2 2
3 17 1 2 2*
4 21 1 2 nd
5 35 1, 2 2 nd
6 37 1 2 nd
7 43 1 2 nd
8 50 1, 2 2 2
9 52 1 2 nd
10 59 1, 2 2 nd
11 59 1, 2 2 2
12 61 1 2 nd
13 71 1 2 nd
14 82 1, 2 2 2
* sample not used for microarray experiment because of insufficient aRNA after amplification.
All samples were Cy5-labeled and mixed with an equal amount of Cy3-labeled reference probe, consisting of equal amounts of RNA of all macrodissected samples. After hybridization of cDNA arrays, data were normalized and filtered, resulting in a set of 2358 genes that gave sufficient signal on all arrays. Based on the expression of these 2358 genes, hierarchical clustering was performed to group samples according to similarity in gene expressions, without information of sample identity (Figure 1). This unsupervised clustering distinguished two main clusters according to the number of amplification cycles. Macrodissected samples that were amplified a second round were more similar to twice-amplified microdissected samples than to their original once-amplified samples. To determine the statistical significance of the effect of the second amplification cycle, Pearson correlation coefficients between once-amplified macrodissected samples and their corresponding twice-amplified samples were calculated. The resulting coefficients were low, in contrast to the coefficients of independently amplified samples or of duplicate labeling experiments (Table 2), excluding that such a change in expression profiles was induced by experimental variation.
Figure 1 Unsupervised hierarchical clustering of macro- and microdissected rectal carcinoma samples. Macrodissected samples (squares), microdissected tumor epithelium samples (triangles) and microdissected stroma samples (circles) were clustered based on average correlation. Open symbols indicate RNA analyzed after one round of amplification and closed symbols indicate two rounds of amplification. Numbers correspond to the carcinoma samples in Table 1.
Table 2 Effect of the second round of amplification. Pearson correlation coefficients evaluating the effect of the second round of amplification on the gene expression profiles. Twice-amplified macrodissected samples were compared to the corresponding once-amplified macrodissected samples. Correlation of duplicate amplification and labeling experiments are presented as well. In case of repeated experiments, Pearson correlation coefficients were calculated for each experiment and averaged.
Sample Correlation Coefficient p-value
2 0.24 <0.001
5 0.24 <0.001
8 0.00 0.961
10 0.19 <0.001
11 0.17 <0.001
14 0.15 <0.001
labelling 0.95 <0.001
amplification 0.81 <0.001
Taken together, these findings indicate that, in this experimental setting, expression profiles were hardly preserved during the extra round of amplification performed with random primers, and therefore exclude reliable cross-comparison of once- and twice-amplified samples.
Evaluation of the bias introduced by second round amplification
Low correlation coefficients indicate that the overall gene expression profile was changed. However, such coefficients do not specify whether the expression of all genes was slightly changed or whether the expression of a proportion of genes was altered dramatically. To evaluate the amplification-induced change in more detail, the number of genes that were significantly preserved by the second round of amplification, a "conservative set of genes", was defined by t-tests according to Nygaard et al. [9]. These calculations indicated that 42% of the genes were on average not significantly influenced. Calculating a "rejected set of genes" indicated that for 20% of the genes the expressions were significantly changed, and most of these rejected genes (70%) were changed at least three-fold. This suggests that a substantial proportion of the expression profile was significantly affected by the second round of amplification with random primers. However, closer examination of this "rejected gene-set" revealed that for the majority of the rejected genes, the amplification-induced change was in the same direction over all tumor samples, indicating that the bias for these genes could be constant. Although changed significantly, a constant bias might not influence the outcome as long as all tested samples are amplified for the same number of cycles.
To analyze whether the amplification-induced bias was constant for all carcinoma samples, we determined the actual variation for the whole set of 2358 genes. This variation between once- and twice-amplified macrodissected samples is more indicative for the reproducibility of the amplification effect on the gene expression profiles. Therefore, the amplification-induced fold-changes of each gene were calculated for all tumor samples. These values were then averaged, which allow calculation of a standard deviation (SD) of the fold-change for each gene (Figure 2). A high SD indicates that the amplification-induced change of that gene was less reproducible over the different samples. For instance, the expression of a given gene may be induced ten-fold in one sample, while reduced ten-fold in the next sample. On average, the amplification-induced change is zero, suggesting no effect. However, the variation of this particular gene among samples is 100-fold, which is too high to be regarded as reproducible. To determine which variation-range is acceptable, we used the 95% normal confidence interval, which is defined by the mean ± 1.96*SD. An interval with a ten-fold variation-range has an SD of 0.25 on a log10-based scale, and an interval with a four-fold variation-range has an SD value of 0.15. Figure 2 demonstrates that several genes (8%) have a higher SD than 0.25, indicating that for these genes the amplification resulted in a highly dispersed (>ten-fold) expression pattern. When the cut-off point of the SD was set at 0.15, a range we propose to be acceptable, it turned out that 39% of all genes had a higher standard deviation. For this substantial proportion of genes, we concluded that the amplification effect was not constant over the different samples.
Figure 2 Variation in amplification-induced fold-change for the conserved and rejected gene-sets. Per gene the fold-change induced by the second round of amplification over all samples was averaged (x-axis) and plotted against the standard deviation (SD) of the fold-change (y-axis). Statistically conserved genes (black) and rejected genes (gray) displayed high variation in standard deviations. A cut-off value at which the 95% normal CI lies between 0.5 and 2 times the expression value (± 0.3 on log10-scale) corresponds to an SD of 0.15 (the 95% normal CI lies within 1.96 standard deviations of the mean; in this case SD = 0.3/1.96 = 0.15).
Larger contribution of tumor epithelium than stroma to gene expression profiles
In the unsupervised clustering (Figure 3) all twice-amplified samples clustered together. In this subgroup the macrodissected samples clustered closer to microdissected tumor samples than to stroma samples. This observation suggested that in general the effect of stroma on gene expression profiles of macrodissected samples was smaller than the contribution of tumor epithelium. To determine the contributions of epithelial tumor cells and of stroma cells in the gene expression profiles of macrodissected samples, linear regression analysis was performed on the twice-amplified macrodissected samples with their corresponding microdissected tumor and stroma samples (Table 3). In this analysis, the relationship between the macrodissected sample and the corresponding tumor and stroma samples were quantified according to the formula: gene expressions of macrodissected sample = α*tumor expressions + β*stroma expressions. The relative contribution of tumor epithelium can then be calculated by α/(α+β). In case of carcinoma sample 14, with 18% stroma and 82% tumor epithelial surface in the macrodissected section, the relative contributions of stroma and tumor mRNA were 7% and 93%, respectively. For carcinoma sample 2, which had only 15% tumor epithelium surface in the macrodissected section, stroma and tumor mRNAs contributed equally to the gene expression profile of the macrodissected sample. These linear regression analyses demonstrate that the macrodissected gene expression profile depends much more on the tumor epithelium than would be expected from the percentage of epithelial tumor surface.
Table 3 Involvement of tumor epithelium and stroma. Linear regression was used to quantify the relative contributions of tumor epithelium and stroma to the gene expression profile of the macrodissected sample. If, for one gene, s is the amount of RNA measured in microdissected stroma, t is the amount measured in microdissected tumor, and r is the amount RNA measured in the macrodissected sample, we assume r = αt+βs, where α and β are unknown coefficients. The last column is the relative contribution of tumor epithelium, α/(α+β), assuming that the contributions of stroma and tumor together are 100%. Because we are considering a sum of contributions on the linear RNA scale, the regression has to be performed on the non-logged data. The values are averaged in case of duplicate labeling experiments and standard errors of coefficients α and β are given.
Sample surface % epithelium tumor α (std error) stroma β (std error) relative tumor contribution: α/(α+β)
2 15 0.44 (0.01) 0.46 (0.01) 49%
5 35 0.80 (0.01) nd nd
8 50 0.93 (0.01) 0.06 (0.02) 94%
10 59 0.91 (0.01) nd nd
11 59 0.66 (0.01) 0.14 (0.02) 83%
14 82 0.98 (0.02) 0.07 (0.03) 93%
The relatively high contribution of epithelial tumor cells suggested that more RNA could be extracted from tumor epithelium than from stroma. Therefore, the yields of total RNA isolated per volume microdissected tissue were compared (Figure 3). Although similar volumes of tumor epithelium and stroma were microdissected, yields of total RNA of epithelial tumor samples were on average 3.5-fold higher than yields of stroma RNA (p = 0.001). This difference between tumor epithelium and stroma increased when the aRNA yields after two rounds of amplification were compared. On average, the amount of aRNA generated from microdissected tumor samples was eight times higher than the aRNA of equal volumes of stroma samples (p < 0.001). This difference in mRNA quantities explained the minor contribution of stroma to the gene expression profiles of macrodissected samples.
Figure 3 Total RNA and amplified RNA yields of equal volumes of microdissected tumor epithelium and stroma. The average yield of total RNA (left axis; closed symbols) isolated from tumor epithelium (triangles; mean 50 ng/mm2 of 10 μm thick sections), was higher than total RNA isolated from stroma (circles; mean 14 ng/mm2; p = 0.001). After two rounds of amplification, a higher difference was observed between yields of RNA (right axis; open symbols) of microdissected tumor epithelium (mean 33 μg/mm2) and stroma (mean 4 μg/mm2; p < 0.0001).
Discussion
Surgically resected rectal carcinomas contain epithelial tumor cells as well as stroma cells. In microarray experiments of such specimens, both components will contribute to the gene expression profiles. The influence of stroma cells might therefore prevent accurate analysis of gene expressions specific for epithelial tumor cells, especially when high percentages of stroma are present in the carcinoma samples. For rectal carcinomas, the observed high variation in percentages of epithelial tumor surface might complicate interpretations of microarray data even more. Therefore, the question arose whether these samples had to be microdissected to obtain reliable tumor epithelial gene expression data.
In this study, we compared gene expression profiles of several macrodissected rectal carcinoma samples, where only surrounding healthy tissue was removed, with the same samples microdissected by LMPC. Both the effect of a second amplification round as well as the effect of stroma on the gene expression profiles was analyzed in order to determine the best dissection method to detect the expression of epithelial tumor-derived genes by microarray analysis. Unsupervised clustering of the gene expression profiles resulted in two main clusters according to the number of amplification rounds. This observation indicates that the second round of amplification, needed for microdissected samples to get sufficient RNA for microarray experiments, affected the overall gene expression profiles.
The T7 RNA polymerase-based linear amplification protocol [12] is one of the most widely used among the available amplification techniques. In this procedure, the amplification reaction consists of transcription via an oligo(dT)-primer harboring a T7 promoter sequence. When second amplification rounds were required, as is the case for microdissected samples, an additional cDNA synthesis step was performed with second round primers followed by the T7-based amplification reaction. Because these second round primers are random primers, transcript sizes will decrease. Quality analysis of the amplified RNA samples demonstrated that indeed the second round of amplification slightly reduced transcript fragments (data not shown). This effect was more pronounced for microdissected samples, probably because of lower amounts of input RNA for the amplification procedure and some degradation occurring during the time-consuming process of microdissection [6].
Most studies determined the amplification effect by comparing expression ratios of two non-amplified RNA samples versus the ratios of the same RNAs amplified. These studies show that the majority of expression differences were maintained by the amplification procedure although a slight decrease in correlation coefficients was observed [13,14], and the intensity levels were not preserved [7,9,15]. In order to evaluate macro- versus microdissection, we determined the effect of the for microdissection required second amplification reaction on the gene expression profiles by comparing once- and twice-amplified samples. The low Pearson correlation coefficients and the calculated significantly "conserved" and "rejected" gene-sets according to Nygaard et al. [9] demonstrate that the overall gene expression profile was changed by the second round of amplification. In this cross-comparison analysis, the extreme low correlation coefficients might be the consequence of the above-suggested loss of intensity levels.
Such a cross-comparison analysis of once- and twice-amplified samples indicates that the gene expression profile is changed by the amplification reaction, but not whether this change is reproducible for all samples. Other studies established that amplification-induced changes are particularly sequence dependent and not abundance dependent [10,15], suggesting a fairly constant bias. Therefore, the variation of the amplification-induced change over the different samples was determined, as this variation will be indicative for the consistency of the bias. When we take a standard deviation of 0.15 as an approximate quantitative criterion (95% normal confidence interval, allowing a four-fold variation in expressions induced by amplification), for 39% of the genes the variation in the gene expression introduced by amplification was outside this confidence interval. This analysis indicates that for a substantial proportion of the genes, the amplification-induced change was not constant.
Importantly, such a high variation was observed with similar frequencies in the "conserved" and "rejected" gene-sets. Therefore, although twice-amplified genes might be called "conserved" based on a t-test, indicating that the change on average is around zero, the variation over the different samples will be changed by the amplification, resulting in more false-negative and false-positive genes. Since genes extracted from the microarray analysis require verification by other biochemical experiments, false-positive genes will be recognized and can be reclassified. Putative interesting genes that are false negative will be missed from the analysis.
Since the second round of amplification affected the gene expression profiles, the use of once-amplified samples is highly preferred. The fact that for macrodissected samples one round of amplification suffices to get enough labeled mRNA, which results in a far more convenient and cost-effective procedure [16], supports the use of macrodissected samples. Although the first amplification round might induce some changes in gene expression as well, the amplification-induced bias is reported to be larger when the amounts of input material is low [10,17]. The yields of RNA isolated from microdissected samples were small, while for macrodissected samples the recommended quantity of 1 μg total RNA could be used in the amplification reaction. Therefore, the amplification-induced bias is probably slightly higher for microdissected samples than for macrodissected samples. Of note, one round of amplification is demonstrated to be more sensitive to low abundance transcripts than using total RNA [18-20].
A possible disadvantage of macrodissected samples is the presence of stroma cells that might disturb the epithelial tumor-specific gene expression profiles. We therefore evaluated the contribution of stroma in the macrodissected gene expression profiles. In the unsupervised clustering of twice-amplified samples, macrodissected samples clustered closer to microdissected tumor samples than to microdissected stroma samples, suggesting that epithelial tumor cells had a higher contribution to gene expression profiles than stroma cells. This observation was confirmed by linear regression analysis, indicating that the involvement of stroma in macrodissected gene expression profiles was minor. In the unsupervised clustering, samples 2 and 14 (low en high percentage of tumor epithelium, respectively) clustered relatively together with their corresponding microdissected stroma and tumor sample. Although this observation suggested an association between the surface percentage of tumor epithelium in the macrodissected sample and the degree of clustering of this sample with the microdissected tumor sample, such a clear-cut correlation could not be established. The contribution of stroma to the gene expression profiles was not strictly related to the surface percentage and was smaller than expected from the surface percentage of the stroma. However, for the sample with 15% tumor epithelium the contributions of stroma and tumor were equal, indicating that this sample contained probably too much stroma for adequate analysis of tumor-derived genes. For such samples a further enrichment for tumor epithelium is necessary and can probably be attained by macrodissection.
An explanation for our finding that the contribution of stroma is relatively small is provided by the observation that the yields of total RNA as well as of amplified mRNA from stroma samples were much lower than from equal volumes of tumor tissue. These findings are presumably due to a higher density of tumor cells and/or more transcription activity in tumor epithelium compared to stroma. Although these data are obtained by analysis of rectal carcinoma samples, our conclusions are probably applicable to other tumor types with a stroma component as well. The fact that far less mRNA is isolated from stroma than from epithelium suggests that the contribution of stroma to the overall gene expression profile will always be minor with the consequence that macrodissection might be the preferred method for other carcinoma types as well. Furthermore, in case it is absolutely necessary to discard the stroma gene expression, it might be an option to perform in silico microdissection [21-23]. These computational approaches have the advantage that macrodissected samples can be used, thereby leaving out the biases caused by the required second round of amplification in case of manual microdissection. However, it is important to realize that most of the in silico approaches are based on the assumption that tumor epithelium and stroma will equally contribute to the overall gene expression profile. In this study, we demonstrated that the stroma contribution is much smaller than expected from the surface area of the rectal carcinoma sample, which should be included in the in silico analysis.
Although the influence of stroma-derived RNA on the expression profiles of genes which are expressed by stroma as well as by tumor epithelium is small, expression of genes which are specific for stroma cells, might still be detectable when using macrodissection [24]. This is an additional advantage of macrodissection, because increasing evidence supports an important role for the microenvironment in carcinoma formation and progression, and therefore these stroma cells might be of great interest. For instance, expression of some stroma-specific genes appeared to be correlated with patient prognosis [2,25]. Fromique et al. [26] showed that signaling between epithelial tumor cells and fibroblasts influenced the gene expression pattern of the tumor cells. For rectal carcinoma it has been demonstrated that apart from the pathological characteristics of tumor cells the amount and type of infiltrate is also relevant for the control of cancer [27]. When tumor epithelium is selected by LMPC, this stroma-specific information is missed.
Conclusion
Because rectal carcinoma samples contained varying amounts of stroma versus tumor epithelium, the question arose whether macrodissection could be used or whether the samples should be microdissection for gene expression profiling. Purification of tumor epithelium by laser microdissection was supposed to give the most reliable tumor-specific gene expression profiles. However, we showed that these overall gene expression profiles are affected by the required second round of mRNA amplification with random primers. The contribution of stroma to gene expression profiles of macrodissected samples was much smaller than expected on the basis of the quantified surfaces. And of even more importance, the interference of stroma cells with the overall gene expression profiles appeared to be minor. Therefore, we recommend RNA isolation of clinically resected carcinomas samples that are only enriched for tumor epithelium by macrodissection for microarray experiments.
Methods
Macrodissection and microdissection of tissue samples
The experimental outline of this study is depicted in Figure 4. Fresh frozen rectal carcinoma samples were obtained from 14 different patients who underwent surgery in either the Leiden University Medical Center or the Leyenburg Hospital. All samples were macrodissected in a cryostat at -20°C by removing surrounding healthy tissue. Of these, two sections of 30 μm were collected for total RNA extraction. For the microdissection procedure [28], sections of 10 μm were cut and adhered to polyethylene-naphtalate (PEN) membrane slides (P.A.L.M. Microlaser Technologies AG, Bernried, Germany), followed by hydration by rinsing the slides in 100%, 75% and 50% ethanol. The samples were stained with Mayer's haematoxylin, briefly rinsed in diethylpyrocarbonate (DEPC)-treated water and dehydrated in graded ethanols. All slides were finally air-dried and stored dry at -80°C until microdissection was performed using the PALM® Micro Beam microscope (P.A.L.M. Microlaser Technologies AG) for non-contact laser microdissection and pressure catapulting (LMPC). Microdissection of 0.5–1 mm2 tissue took 30 to 120 minutes per sample.
Figure 4 Schematic overview of the strategy used to compare macrodissection and microdissection. All 14 rectal carcinoma samples are macrodissected or microdissected for tumor epithelium, 6 macrodissected samples were in addition amplified (amp) a second round and of 4 samples the stroma was microdissected as well.
Sections of 5 μm of each macrodissected sample were stained with classical haematoxylin and eosin staining and examined by light microscopy to quantify the surface percentages of tumor epithelium versus stroma. Table 1 gives an overview of the samples used and the quantified percentages of tumor epithelium.
Total RNA extraction, mRNA amplification and labeling
For macrodissected samples, the sections were homogenized by vortexing with glass beads in RNA-Bee reagent (Tel-Test Inc., Friendswood, TX). Total RNA was extracted according to the manufacturer's protocol of RNA-Bee and purified using the Qiagen RNeasy mini kit with on-column DNase digestion according to manufacturer's instructions (Qiagen Sciences, Germantown, MD). For microdissected tissues, total RNA was isolated using the Qiagen RNeasy mini kit with on-column DNAse treatment (Qiagen). Quality of total RNAs was assessed with lab-on-a-chips on the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, California). All samples were shown to be free of DNA contamination and for each sample the ratio 28S/18S was >1.5.
Amplifications were performed using Ambion's MessageAmp™ kit and protocol (Ambion Inc., Austin, TX). For macrodissected samples, of which on average 30 μg total RNA was isolated, the first amplification round was started with 1.0 μg total RNA. This first amplification round of macrodissected samples yielded on average 24 μg amplified mRNA (aRNA). The second round of amplification, using random second round primers, was started with 1.0 μg aRNA in case of macrodissected samples. For microdissected specimens the whole quantity of isolated total RNA was used (on average 30 ng total RNA) for the first round of amplification, and all aRNA was used for the second round of amplification. Yield of aRNA of these twice-amplified microdissected samples was on average 15 μg aRNA. Quality of each aRNA was checked on lab-on-a-chip (Agilent Technologies). Quantification of aRNA was performed by spectrometry at 260 nm wavelength.
Per microarray experiment, 1.0-μg aliquots of aRNA were labeled with Cy5-dUTPs (Amersham Biosciences, Buckinghamshire, UK) by direct incorporation during a reverse transcriptase reaction using the CyScribe kit, according to manufacturer's instructions (Amersham Biosciences). The labeled cDNAs were mixed with equal amounts of Cy3-dUTP-labeled cDNA from a once-amplified reference probe, consisting of equal amounts of RNA from all macrodissected samples.
cDNA microarray
The mixture of labeled reference and sample was purified on YM30 Microcon columns (Millipore Corporation, Bedford, MA) together with 20 μg human COT-1 DNA (Invitrogen, Carlsbad, CA). After purification, 8 μg yeast tRNA (Invitrogen) and 20 μg polyadenylic acid (Sigma-Aldrich, St. Louis, MO) were added. Preheated hybridization buffer (25% formamide, 5× SSC, 0.1% SDS) was added just before hybridization at 42°C o/n in to human 18K cDNA microarrays slides, manufactured at the Central Microarray Facility (CMF) of the Netherlands Cancer Institute. Protocols, GeneID list and information about arrays are available at the website of the CMF [29].
Data preparation
Of each slide, two images were scanned using the GeneTAC LSIV laser scanner (Genomic Solutions, Ann Arbor, MI) at different gain settings, one at which hardly any of the spots were saturated and one with a higher gain to obtain data from lowly expressed genes. Spots were quantified by using GenePix Pro 4.1 software (Axon Instruments Inc., Union City, CA). For spot selection an MS-Excel macro was used [30]. Briefly: spots were corrected for local background noise. Per dye, the intensity of each spot was normalized to the median of all spots on the array and for each spot, the ratio of the sample to the reference was calculated. Because arrays were scanned at two different settings, ratios from high gain-saturated spots were used from the low gain scans, while lowly expressed genes were used from the high gain scans. Genes saturated in both gains were rejected from analysis. For other spots, the mean of the ratios of the two scans was calculated. Finally, ratios were log10-transformed. Because the goal of this study was to analyze the effects of macro- and microdissection on overall gene expression profiles and not to select specific genes, only qualified genes that were present on all 45 arrays (2358 genes) were selected for further statistical analyses. The data discussed in this publication have been deposited in NCBIs Gene Expression Omnibus (GEO) [31] and are accessible through GEO Series accession number GSE2738.
Array data analysis
Unsupervised clustering of the genes and samples was performed with Spotfire 7.2 software (Spotfire AB, Göteborg, Sweden) based on hierarchical clustering of average linkage correlation of the log10-transformed data. Pearson correlation coefficients and linear regression were calculated with SPSS 11.0 software for Windows (SPSS Inc., Chicago, IL). The "conservative" and "rejected" gene sets were calculated according to Nygaard et al. [9]. In summary, for the "conservative set of genes", each gene with a p-value less than 0.1 in a t-test assuming unequal variances or in a t-test assuming equal variances was removed. The "rejected set of genes" are genes which are significantly changed when two groups of samples are compared in a t-test according to the Benjamini-Hochberg procedure [32] with a false discovery rate of 1%.
Authors' contributions
EB and SP carried out all experiments and data analysis with assistance of EL, RE and MZ. ML and TW designed the Excel macro to normalize the microarray data. CM, JK, JM, CV and LP initiated the study and supervised the data generation and analyses. PE was involved in the statistical analyses. All authors read and approved that final manuscript.
Acknowledgements
We thank Joana Cardoso and Dr. Riccardo Fodde from the Department of Pathology, Josephine Nefkens Institute, Erasmus Medical Center, Rotterdam and Dr. Judith Boer from the Center for Human and Clinical Genetics, Leiden University Medical Center for advice and providing protocols on microdissection and amplification techniques. We are indebted to Dr. Paul Blok from the Department of Pathology, Leyenburg hospital, Den Haag and Dr. Hans Morreau from the Department of Pathology, Leiden University Medical Center for providing several rectal carcinoma samples and for assistance with the pathological evaluation as well. This study was financially supported by the Dutch Cancer Society: RUL 2002–2733.
==== Refs
Ramaswamy S Tamayo P Rifkin R Mukherjee S Yeang CH Angelo M Ladd C Reich M Latulippe E Mesirov JP Poggio T Gerald W Loda M Lander ES Golub TR Multiclass cancer diagnosis using tumor gene expression signatures Proc Natl Acad Sci U S A 2001 98 15149 15154 11742071 10.1073/pnas.211566398
van't Veer LJ Dai HY van de Vijver MJ He YDD Hart AAM Mao M Peterse HL van der Kooy K Marton MJ Witteveen AT Schreiber GJ Kerkhoven RM Roberts C Linsley PS Bernards R Friend SH Gene expression profiling predicts clinical outcome of breast cancer Nature 2002 415 530 536 11823860 10.1038/415530a
Cromer A Carles A Millon R Ganguli G Chalmel F Lemaire F Young J Dembele D Thibault C Muller D Poch O Abecassis J Wasylyk B Identification of genes associated with tumorigenesis and metastatic potential of hypopharyngeal cancer by microarray analysis Oncogene 2004 23 2484 2498 14676830 10.1038/sj.onc.1207345
Russo G Zegar C Giordano A Advantages and limitations of microarray technology in human cancer Oncogene 2003 22 6497 6507 14528274 10.1038/sj.onc.1206865
Sugiyama Y Sugiyama K Hirai Y Akiyama F Hasumi K Microdissection is essential for gene expression profiling of clinically resected cancer tissues Am J Clin Pathol 2002 117 109 116 11789716 10.1309/G1C8-39MF-99UF-GT2K
Michel C Desdouets C Sacre-Salem B Gautier JC Roberts R Boitier E Liver gene expression profiles of rats treated with clofibric acid - Comparison of whole liver and laser capture microdissected liver Am J Pathol 2003 163 2191 2199 14633594
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
Zhao H Hastie T Whitfield ML Borresen-Dale AL Jeffrey SS Optimization and evaluation of T7 based RNA linear amplification protocols for cDNA microarray analysis BMC Genomics 2002 3 31 12445333 10.1186/1471-2164-3-31
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
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
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
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
Luzzi V Mahadevappa M Raja R Warrington JA Watson MA Accurate and reproducible gene expression profiles from laser capture microdissection, transcript amplification, and high density oligonucleotide microarray analysis J Mol Diagn 2003 5 9 14 12552074
Wang J Hu L Hamilton SR Coombes KR Zhang W RNA amplification strategies for cDNA microarray experiments Biotechniques 2003 34 394 400 12613262
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
Croner RS Guenther K Foertsch T Siebenhaar R Brueckl WM Stremmel C Hlubek F Hohenberger W Reingruber B Tissue preparation for gene expression profiling of colorectal carcinoma: three alternatives to laser microdissection with preamplification J Lab Clin Med 2004 143 344 351 15192650 10.1016/j.lab.2004.03.003
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
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
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
Feldman AL Costouros NG Wang E Qian M Marincola FM Alexander HR Libutti SK Advantages of mRNA amplification for microarray analysis Biotechniques 2002 33 906 12, 914 12398200
Venet D Pecasse F Maenhaut C Bersini H Separation of samples into their constituents using gene expression data Bioinformatics 2001 17 Suppl 1 S279 S287 11473019
Stuart RO Wachsman W Berry CC Wang-Rodriguez J Wasserman L Klacansky I Masys D Arden K Goodison S McClelland M Wang Y Sawyers A Kalcheva I Tarin D Mercola D In silico dissection of cell-type-associated patterns of gene expression in prostate cancer Proc Natl Acad Sci U S A 2004 101 615 620 14722351 10.1073/pnas.2536479100
Lahdesmaki H Shmulevich L Dunmire V Yli-Harja O Zhang W In silico microdissection of microarray data from heterogeneous cell populations BMC Bioinformatics 2005 6 54 15766384 10.1186/1471-2105-6-54
Perou CM Jeffrey SS van de RM Rees CA Eisen MB Ross DT Pergamenschikov A Williams CF Zhu SX Lee JC Lashkari D Shalon D Brown PO Botstein D Distinctive gene expression patterns in human mammary epithelial cells and breast cancers Proc Natl Acad Sci U S A 1999 96 9212 9217 10430922 10.1073/pnas.96.16.9212
Ramaswamy S Ross KN Lander ES Golub TR A molecular signature of metastasis in primary solid tumors Nat Genet 2003 33 49 54 12469122 10.1038/ng1060
Fromigue O Louis K Dayem M Milanini J Pages G Tartare-Deckert S Ponzio G Hofman P Barbry P Auberger P Mari B Gene expression profiling of normal human pulmonary fibroblasts following coculture with non-small-cell lung cancer cells reveals alterations related to matrix degradation, angiogenesis, cell growth and survival Oncogene 2003 22 8487 8497 14627989 10.1038/sj.onc.1206918
Nagtegaal ID Marijnen CA Kranenbarg EK Mulder-Stapel A Hermans J van de Velde CJ van Krieken JH Local and distant recurrences in rectal cancer patients are predicted by the nonspecific immune response; specific immune response has only a systemic effect--a histopathological and immunohistochemical study BMC Cancer 2001 1 7 11481031 10.1186/1471-2407-1-7
Cardoso J Molenaar L de Menezes RX Rosenberg C Morreau H Moslein G Fodde R Boer JM Genomic profiling by DNA amplification of laser capture microdissected tissues and array CGH Nucleic Acids Res 2004 32 e146 15514107 10.1093/nar/gnh142
Central Microarray Facility of the Netherlands Cancer Institute 2005
Rozeman LB Hameetman L van Wezel T Taminiau AHM Cleton-Jansen AM Hogendoorn PCW Bovee JVMG cDNA expression profiling of chondrosarcomas: Ollier disease resembles solitary tumours and alteration in genes coding for components of energy metabolism occurs with increasing grade Journal of Pathology 2005 207 61 71 16007578 10.1002/path.1813
NCBI Gene Expression Omnibus 2005
Benjamini Y Hochberg Y Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing J R Stat Soc (Series B) Stat Methodol 1995 57 289 300
|
16225673
|
PMC1283972
|
CC BY
|
2021-01-04 16:32:46
|
no
|
BMC Genomics. 2005 Oct 14; 6:142
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-142
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-951626289710.1186/1471-2334-5-95Research ArticleIncreased oxidative stress associated with the severity of the liver disease in various forms of hepatitis B virus infection Bolukbas Cengiz [email protected] Fusun Filiz [email protected] Mehmet [email protected] Mehmet [email protected] Hakim [email protected] Ozcan [email protected] Department of Internal Medicine, Gastroenterology Division, Harran University, Medical Faculty, Sanliurfa, Turkey2 Department of Internal Medicine, Harran University, Medical Faculty, Sanliurfa, Turkey3 Department of Biochemistry, Harran University, Medical Faculty, Sanliurfa, Turkey2005 31 10 2005 5 95 95 28 5 2005 31 10 2005 Copyright © 2005 Bolukbas et al; licensee BioMed Central Ltd.2005Bolukbas 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
Oxidative stress can be defined as an increase in oxidants and/or a decrease in antioxidant capacity. There is limited information about the oxidative status in subjects with hepatitis B virus infection. We aimed to evaluate the oxidative status in patients with various clinical forms of chronic hepatitis B infection.
Methods
Seventy-six patients with hepatitis B virus infection, in whom 33 with chronic hepatitis, 31 inactive carriers and 12 with cirrhosis, and 16 healthy subjects were enrolled. Total antioxidant response and total peroxide level measurement, and calculation of oxidative stress index were performed in all participants.
Results
Total antioxidant response was significantly lower in cirrhotics than inactive HbsAg carriers and controls (p = 0.008 and p = 0.008, respectively). Total peroxide level and oxidative stress index was significantly higher in cirrhotic (p < 0.001, both) and chronic hepatitis B subjects (p < 0.001, both) than inactive HbsAg carriers and controls. Total antioxidant response was comparable in chronic hepatitis B subjects, inactive HbsAg carriers and controls (both, p > 0.05/6). Total peroxide level and oxidative stress index were also comparable in inactive HBsAg carriers and controls (both, p > 0.05/6). Serum alanine amino transferase level was positively correlated with total peroxide level and oxidative stress index only in chronic hepatitis B subjects (p = 0.002, r = 0.519 and p = 0.008, r = 0.453, respectively).
Conclusion
Oxidative stress occurs secondarily to increased total lipid peroxidation and inadequate total antioxidant response and is related to severity of the disease and replication status of virus in hepatitis B infection.
==== Body
Background
Reactive oxygen species (ROS) are oxygen-containing molecules that produced during normal metabolism [1]. The organism has enzymatic and non-enzymatic antioxidant systems neutralizing the harmful effects of the endogenous ROS products [2]. Under certain conditions, the oxidative or anti-oxidative balance shifts towards the oxidative status as a result of increase in ROS and/or impairment in antioxidant mechanism [3,4].
It has been suggested that reactive oxygen species and lipid peroxidation products likely contribute to both onset and progression of hepatic fibrosis [5]. In addition, oxidative stress is one of the reasons of DNA damage, which might be associated with the development of hepatocellular carcinoma (HCC) in chronic viral hepatitis [6].
In several studies [7-10], increased oxidative stress has been suggested to be responsible from the hepatocellular damage caused by chronic hepatitis B infection (CHB). However, most of them have evaluated oxidative status using individual antioxidants measurement and the information about the total antioxidant response (TAR) of subjects with CHB and cirrhosis due to hepatitis B virus (HBV) infection is limited [11]. To our knowledge, there is no information in the literature about the oxidants in subjects with cirrhosis due to HBV infection, and neither oxidants nor antioxidants in inactive hepatitis B carriers.
In the present study, we aimed to measure the TAR in CHB, cirrhosis due to HBV infection and inactive HbsAg carrier subjects to evaluate their antioxidant status using a novel automated method [12]. As a reciprocal measure, the total peroxide levels of the same plasma samples were also measured. The percent ratio of the total plasma peroxide level to the plasma TAR value was regarded as oxidative stress index [13].
Methods
Enrollment of patients
CHB subjects (n = 33, 20 male/13 female; mean age; 39 ± 12.8 years), inactive HBsAg carrier subjects (n = 31, 19 male/12 female; mean age; 38 ± 8.3 years), subjects with cirrhosis due to HBV infection (n = 12, 7 male/5 female; mean age 40.9 ± 11.6 years) and healthy controls (n = 16, 9 male/7 female; mean age 31.6 ± 6 years) were enrolled in the present study. Patient selection has been started at March 2003 and finished at December 2004, when the numbers of study subjects have reached to provide a power for statistical analysis. Sample size was calculated with an expected parameter estimate based on a pilot study performed in our department. An assuming a mean of 1.71 TAR in control group and a mean of 1.40 TAR in Cirrhosis-HBV Group with a 0.25 standard deviation, the minimum sample size thus required to be approximately 13 in each study groups within a 95% confidence and 80% power.
All cirrhotic subjects had compensated and Child A cirrhosis according to the Child-Pugh classification. All participants were age and sex matched. The study protocol was carried out in accordance with the Helsinki Declaration as revised in 1989. All subjects were informed about the study protocol and the written consent was obtained from each one.
Exclusion criteria
Exclusion criteria included the use of supplemental vitamins, serum total bilirubin level higher than 2 mg/dL, history of diabetes mellitus, coronary artery disease, rheumatoid arthritis, cancer, systemic or local infection, the existence of alcohol intake, poor nutritional status, pregnancy, decompensated and Child B or C cirrhosis, concomitant chronic hepatitis C (CHC) or hepatitis D or other well known liver diseases such as metabolic or autoimmune disorders and various infectious states of the liver, non-alcoholic steatohepatitis and HBV-DNA negativity in patients with CHB and cirrhosis.
HCC was detected using ultrasonographic examination of the liver and measurement of serum alpha fetoprotein level. The ultrasound and AFP determinations were performed before the enrolment and especially before drawing the blood samples in all subjects.
Initial evaluation
Diagnosis of chronic hepatitis B was based on 6 months history of HBsAg and HBV-DNA positivity with at least 2 times higher alanine amino transferase (ALT) than upper limit of normal level. Findings of chronic hepatitis B were supported by the histopathological evaluation based on the modified Knodell score [14]. Inactive HBsAg carrier state for HBV infection was diagnosed on the basis of at least 1 year of HBsAg positivity with normal ALT levels and negative HBV-DNA.
Cirrhosis due to HBV was diagnosed on the basis of the clinical, laboratory, virological, radiological, and/or histopathological findings.
Control group consisted of healthy individuals with normal medical history, physical examination, blood biochemistry and negative hepatitis C virus (HCV) antibody and HBV serum markers.
A wide panel of biochemical and haematological parameters were evaluated by standard automated techniques.
Virological studies
Anti-HCV, HBsAg, anti-HBs, HBeAg, anti-Hbe were assayed by micro particle enzyme immunoassay (MEIA) (Abbott axsym system, IL USA). The presence of hepatitis D infection was detected by enzyme immunoassay for detection of antibodies against hepatitis delta [(Abbott Murex, Dartford UK) (organon Teknika)]. HBV-DNA was investigated using real time polymerase chain reaction (PCR) method [HBV QNP 2.0 HBV-DNA quantitative kits, Iontek, Istanbul, Turkey) (BioRad ÝCycler)]. Upper and lower limit of HBV-DNA level with real time PCR were 2 × 102 and 2 × 107 copy/ml, respectively.
Blood collection
Blood samples were obtained following an overnight fasting state. Smoker subjects were not permit to smoke during those fasting period. Samples were withdrawn from a cubital vein into heparinised tubes and immediately stored on ice at 4°C. The plasma was then separated from the cells by centrifugation at 3000 rpm for 10 min. Because we aimed to analyze all plasma samples simultaneously for the measurement of TAR and total peroxide level, and the collection of the samples were thought to be longer than 1 month of period, plasma samples were stored at -80°C until analysis as described elsewhere [15,16].
Measurement of the total antioxidant status of plasma
The total antioxidant status of the plasma was measured using a novel automated colorimetric measurement method for TAR developed by Erel [12]. In this method the hydroxyl radical, the most potent biological radical, is produced by the Fenton reaction, and reacts with the colourless substrate O-dianisidine to produce the dianisyl radical, which is bright yellowish-brown in colour. Upon the addition of a plasma sample, the oxidative reactions initiated by the hydroxyl radicals present in the reaction mix are suppressed by the antioxidant components of the plasma, preventing the colour change and thereby providing an effective measure of the total antioxidant capacity of the plasma. The assay results are expressed as mmol Trolox eq./L, and the precision of this assay is excellent, being lower than 3% [17].
Measurement of total plasma peroxide concentration
The total plasma peroxide concentrations were determined using the FOX2 method [18] with minor modifications [13]. The FOX2 test systemis based on the oxidation of ferrous iron to ferric iron by the various types of peroxides contained in the plasma samples, in the presence of xylenol orange which produces a coloured ferric-xylenol orange complex whose absorbance can bemeasured. TheFOX2 reagent was prepared by dissolving ammonium ferrous sulphate (9.8 mg) in 250 mM H2SO4 (10 ml) to give a final concentration of 250 mM ferrous iron in acid. This solution was then added to 90 ml HPLC-grade methanol containing 79.2 mg butylated hydroxytoluene (BHT). Finally, 7.6 mg xylenol orange was added, with stirring, to make the working reagent (250 mM ammonium ferrous sulphate, 100 mMxylenol orange, 25 mM H2SO4, and 4 nM BHT, in 90% (v/v) methanol in a final volume of 100 ml). The blank reagent contained all the components of the solution except ferrous sulphate.
Aliquots (200 mL) of plasma were mixed with 1.8 ml FOX2 reagent. After incubation at room temparature for 30 min, the vials were centrifuged at 12,000 g for 10 min. The absorbance of the supernatant was then determined at 560 nm. The total peroxide content of the plasma samples was determined as a function of the difference in absorbance between the test and blank samples using a solution of H2O2 as standard. The coefficient of variation for individual plasma samples was less than 5%.
Oxidative stress index
The percent ratio of the total peroxide to the total anti-oxidant potential gave the oxidative stress index, an indicator of the degree of oxidative stress [13].
Statistical analysis
Continuous variables were compared by Kruskal-Wallis one-way analysis of variance for non-parametric data with a post hoc analysis using a Mann-Whitney U test. Parametric variables were compared using One-way analysis of variance with post hoc analysis using Tukey test. Fisher's exact test was used to test the sex differences between groups. Spearman's correlation analysis was used to find out the relationship of alanine aminotransferase with TAR, total peroxide level or OSI. Data were presented as median and range for nonparametric variables and mean ± SD for parametric variables. Differences were regarded as significant at 0.05/6 in Kruskal-Wallis one-way analysis and p < 0.05 in other analysis.
Results
Mean age, gender distribution were equal in each group. Serum ALT levels were higher in both CHB and cirrhotic subjects than controls (p = 0.001, p = 0.032, respectively). While CHB subjects had higher serum ALT levels than inactive HBsAg carriers (p = 0.001), both cirrhotics and control groups did not show any significant difference in term of serum ALT levels with inactive HBsAg carriers. There was no statistically significant difference in respect to percent of smokers, numbers of cigarettes smoking in a day and smoking duration of the groups (all p > 0.05). All subjects who smoked were smoking filtered cigarettes.
Smoking habit details and other clinical and demographic data are shown in Table 1.
Table 1 The clinical and demographic data of the study groups
CHB Group Inactive HBsAg Carrier Group Cirrhosis-HBV Group Control Group
n 33 31 12 16
Age (years) 39 ± 12.8 38 ± 8.3 40.9 ± 11.6 31.6 ± 6
Gender (M/F) 20/13 19/12 7/5 9/7
ALT IU/L 100 ± 32* 24.8 ± 9.3 40.6 ± 18.5** 17.5 ± 8.4
Smoking Habit
Percent 45.45 41.93 41.66 43.75
Cigarettes/day 8.8 ± 4.1 7.8 ± 3.1 9.2 ± 3.8 9.8 ± 3.1
Duration(years) 15.5 ± 9.4 16.6 ± 7.0 14.4 ± 8.9 12 ± 5.9
Data were presented as mean ± SD.
* p < 0.001 CHB group vs. inactive HBsAg carriers and control.
** p < 0.05 HBV related cirrhosis group vs. control.
Qualitative variables were assessed by Fisher's exact tests. Differences in continuous variables were evaluated by One-way analysis of variance with post hoc analysis using Tukey test. A p value of < 0.05 was considered statistically significant.
CHB, chronic hepatitis B; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; ALT, alanine aminotransferase.
TAR was significantly lower in cirrhotic subjects than the inactive HBsAg carrier and controls (p = 0.008 and p = 0.008, respectively). The difference between subjects with CHB and cirrhotic subjects in respect to TAR was not statistically significant (p > 0.05/6). There was no significant difference in TAR between chronic hepatitis B and inactive HBsAg carrier or controls (all p > 0.05/6). (Table 2)
Table 2 Oxidative and antioxidative parameters in each group.
CHB Group Inactive HBsAg Carrier Group Cirrhosis-HBV Group Control Group
TAR (mmol Trolox eq./L) 1.58 (1.16–2.23) 1.62 (1.3–1.98) 1.36 (1.2–1.9)‡ 1.7 (1.46–1.96)
Total peroxide (μmol H2O2/L) 32.2(18.9–66)* 25.9(6.6–32) 38.5(22.5–59)** 24.8(18.5–31)
OSI (AU) 1.9 (1–3.9)* 1.6 (0.4–2.4) 2.66(1.1–4.9)** 1.4 (1–1.8)
Data were presented as median and range.
‡ p < 0.008 HBV related cirrhosis vs. inactive HBsAg carriers and control.
* p < 0.001 CHB vs. inactive HBsAg carriers and control.
** p < 0.001 HBV related cirrhosis vs. inactive HBsAg carriers and control.
Differences were regarded as significant at 0.05/4 in Kruskal-Wallis one-way analysis.
CHB, chronic hepatitis B; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; TAR, total antioxidant response; OSI, oxidative stress index.
Total plasma peroxide level of CHB or cirrhotic subjects was significantly higher than inactive HBsAg carrier and controls (p < 0.001, p = 0.001 and p < 0.001, p = 0.001, respectively). The total plasma peroxide level and OSI, an indicator of the degree of oxidative stress, were not significantly higher in cirrhotic than chronic hepatitis B subjects (all p > 0.05/6). (Table 2)
OSI was significantly higher in CHB and cirrhotic subjects than inactive HBsAg carriers and controls (p < 0.001, p < 0.001 and p < 0.001, p < 0.001 respectively) (Table 2).
Inactive HbsAg carriers and controls had comparable results in term of TAR, total peroxide level and OSI (all p > 0.05/6).
TAR of chronic hepatitis B subjects was not significantly correlated with serum ALT level (p > 0.05). Serum ALT level were positively correlated with total peroxide level and OSI in subjects with chronic hepatitis B (p = 0.002, r = 0.519 and p = 0.008, r = 0.453, respectively) (Fig. 1 and 2).
Figure 1 Serum ALT values were positively correlated with total peroxide levels in subjects with chronic hepatitis B (p = 0.002, r = 0.519). ALT, alanine amino transferase.
Figure 2 Serum ALT values were positively correlated with oxidative stress index in subjects with chronic hepatitis B (p = 0.008, r = 0.453). OSI, oxidative stress index; AU, arbitrary unit; ALT, alanine amino transferase.
There was no statistically significant correlation with ALT level, and TAR, total peroxide level and OSI in cirrhotic subjects or inactive HbsAg carriers or controls (all p > 0.05).
Discussion
Normal cell functions and integrity of cell structures may be broken via considerable reactivity of ROS. The organism has enzymatic (e.g. superoxide dismutase, catalase, glutathione peroxidase) and non-enzymatic (e.g. vitamin C, vitamin E) antioxidant mechanisms that work as scavenger for this harmful ROS. Radical-scavenging antioxidants are consumed by the increased free radical activity associated with several conditions, and the total antioxidant response has been used to indirectly assess of free radical activity. The effects of various antioxidants in plasma are additive and the cooperation of antioxidants in human serum provides protection of the organism against attacks by free radicals [3,19]. Therefore, the measurement of TAR may reflect accurately the antioxidant status of the organism [3,12,15,20].
Oxidative stress can be defined as an increase in oxidants and/or a decrease in antioxidant capacity. Although determination of either oxidants or antioxidant components alone may give information about the oxidative stress, determination of oxidants along with antioxidants is more useful in this context. Therefore, oxidants and anti-oxidant capacity should be measured simultaneously to assess oxidative stress more exactly. In addition, the ratio of the total plasma peroxide level o TAR, regarded as OSI and an indicator of oxidative stress, reflects the redox balance between oxidation and anti-oxidation. Recently, it has been reported that OSI may reflect the oxidative status more accurately than TAR or total peroxide level alone [13,21].
Various methods have been developed for the measurement of total antioxidant status. However, there is not yet an accepted "gold standard" reference method [18,22-24], and decisions concerning standardization, and the terms and units used for the measurement of TAR have not yet been made [15]. This implies that this topic needs to be studied further [12]. The most widely used methods for TAR measurement are colorimetric, or involve either fluorescence or chemiluminescence [22,23,25]. However, the fluorescence and chemiluminescence methods need sophisticated techniques, are not appropriate for routine usage and not present in most routine clinical biochemistry laboratories.
In the present study, antioxidant capacity of subjects was determined using TAR, and oxidants and antioxidant capacity were determined simultaneously to determine oxidative stress. The novel method that used in the present study provides several major advantages in comparison with other currently available methods. It is simple and cheap, and can easily be fully automated. It is also reliable and sensitive, and does not interact with commonly occurring serum components such as bilirubin, serum lipids, and anticoagulants. Accurate measurements of TAR can be obtained as little as 10 minutes, making this assay eminently suitable for the clinical biochemistry laboratory [12].
The evidence of oxidative damage in human chronic viral hepatitis is accompanied by a significant rise of the plasma level of the fibrogenic cytokines TNFa and TGFb. In particular, the latter cytokine was shown increased already in plasma of patients with mild tissue inflammation in direct relation with the degree of tissue damage and fibrosis [5]. In addition, excess amounts of reactive species generated in inflamed tissues can cause injury to host cells and also induce DNA damage and mutations [26] and oxidative DNA damage has been suggested to play an important role in the development of HCC [6].
In several studies [7-9,11], increase in oxidative components or decrease in antioxidants or both have been reported in subjects with either acute or chronic HBV infection. Total antioxidant capacity in either acute or chronic HBV infection was measured in only in study of Irhsad et al [11]. The remaining was used individual antioxidants measurement to assess antioxidant response of the organism. At the same way, simultaneously measurement of the oxidants and antioxidant components of the plasma in CHB infection was performed in only at study of Demirdag et al [9].
The information in the literature about the antioxidant components in subjects with cirrhosis due to HBV infection is limited. Irhad et al [11] found that total antioxidant capacity of subjects either with cirrhosis due to HBV infection or other liver disease due to viral etiology is either comparable to or higher than control. To our knowledge, there is no information in the literature about the oxidants in subjects with cirrhosis due to HBV infection, and neither oxidants nor antioxidants in inactive HbsAg carrier subjects.
In order to reflect the true state of oxidative stress in the liver, measurement of lipid peroxidation markers and antioxidant components in hepatic tissue is more ideal than plasma. Nevertheless, ethical and practical considerations make this very difficult for research purposes. Liver biopsy caries a significant morbidity and even mortality risk and it is impossible to perform multiple tests with current techniques on very limited amounts of biopsy specimen that obtained in needle biopsy. Thus, in the present study, we have chosen to perform the measurement of oxidative stress markers in plasma samples. Indeed, in various disorders of the liver, increase in oxidants and/or decrease in antioxidants have been shown in both plasma and liver tissue samples [27,28].
It is well known that serum bilirubin has an antioxidant property [15]. Additionally, poor nutritional status caused modifications to the enzymatic antioxidant systems, with a lower ability to reduce oxidative compounds and a state of lipid peroxidation [29]. These two factors are frequently found in subjects with advanced stages of cirrhosis. Thus, in the present study, we included only the subjects with compensated Child A cirrhosis to evaluate the effects of cirrhosis due to HBV on oxidative status more accurately, and to exclude the effects of other additional factors.
In the present study, we found that TAR of CHB subjects was equivalent to inactive HBsAg carrier and controls. However, total peroxide level, a parameter of oxidative stress, and OSI was significantly higher in CHB subjects than inactive HBsAg carrier and controls. At the same way, there was a strict positive correlation between ALT level, and total peroxide level and OSI in CHB subjects, while no correlation between ALT level and TAR. Additionally, there was an inverse correlation between total peroxide level and OSI with TAR in cirrhotic subjects vs. inactive HBV carriers and controls.
Inactive HBsAg carrier and control subjects had comparable results in term of TAR, total lipid peroxide level and OSI. The lack of increase in TAR in the presence of increased oxidative components, and the strict correlation of ALT levels with total peroxide level and OSI are suggestive for the role of oxidative stress in the pathogenesis of CHB infection.
In the lightening of these findings, we concluded that oxidative stress may have a critical role in hepatic injury and is associated with the severity of disease and the replication status of virus in hepatitis B infection. The novel automated calorimetric assay is a useful, reliable, simple and easily applicable method in the assesment of the total plasma antioxidant response in various forms of hepatitis B virus infection.
Abbreviations
ROS, reactive oxygen species; CHB, chronic hepatitis B infection; HCC, hepatocellular carcinoma; TAR, total antioxidant response; HBV, hepatitis B virus; HbsAg, hepatitis B surfage antigen; OSI, oxidative stress index; CHC, chronic hepatitis C; ALT, alanine amino transferase; HCV, hepatitis C virus; MEÝA, micro particle enzyme immunoassay; PCR, polymerase chain reaction.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
CB, FFB, OE, MA: Conception and design; CB, MH, HC: Analysis and interpretation of the data; CB, FFB, MH: Drafting of the article; CB, FFB, OE, MH, MA, HC: Critical revision of the article for important intellectual content; CB, FFB, OE, MH, MA, HC: final approval of the article; provision of study materials or patients; CB, FFB, MA; CB, MH: Statistical expertise; OE, MA, HC: Collection and assembly of data.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Chance B Sies H Boveris A Hydroperoxide metabolism in mammalian organs Physiol Rev 1979 59 527 605 37532
Halliwell B Free radicals, antioxidants, and human disease: curiosity, cause, or consequence? Lancet 1994 344 721 724 7915779 10.1016/S0140-6736(94)92211-X
Halliwell B Gutteridge JMC Free Radicals in Biology and Medicine 1999 3 Oxford; Oxford Science Publications
Aruoma OI Characterization of drugs as antioxidant prophylactics Free Radic Biol Med 1996 20 675 705 8721614 10.1016/0891-5849(95)02110-8
Poli G Pathogenesis of liver fibrosis: role of oxidative stress Mol Aspects Med 2000 21 49 98 10978499 10.1016/S0098-2997(00)00004-2
Hagen TM Huang S Curnutte J Fowler P Martinez V Wehr CM Ames BN Chisari FV Extensive oxidative DNA damage in hepatocytes of transgenic mice with chronic active hepatitis destined to develop hepatocellular carcinoma Proc Natl Acad Sci U S A 1994 91 12808 12812 7809125
Swietek K Juszczyk J Reduced glutathione concentration in erythrocytes of patients with acute and chronic viral hepatitis J Viral Hepat 1997 4 139 141 9097271
Chrobot AM Szaflarska-Szczepanik A Drewa G Antioxidant defense in children with chronic viral hepatitis B and C Med Sci Monit 2000 6 713 718 11208397
Demirdag K Yilmaz S Ozdarendeli A Ozden M Kalkan A Kilic SS Levels of plasma malondialdehyde and erythrocyte antioxidant enzyme activities in patients with chronic hepatitis B Hepatogastroenterology 2003 50 766 770 12828081
Tanyalcin T Taskiran D Topalak O Batur Y Kutay F The effects of chronic hepatitis C and B virus infections on liver reduced and oxidized glutathione concentrations Hepatol Res 2000 18 104 109 10936561 10.1016/S1386-6346(99)00088-1
Irshad M Chaudhuri PS Joshi YK Superoxide dismutase and total anti-oxidant levels in various forms of liver diseases Hepatol Res 2002 23 178 184 12076713 10.1016/S1386-6346(01)00181-4
Erel O A novel automated method to measure total antioxidant response against potent free radical reactions Clin Biochem 2004 37 112 119 14725941 10.1016/j.clinbiochem.2003.10.014
Harma M Harma M Erel O Increased oxidative stress in patients with hydatidiform mole Swiss Med Wkly 2003 133 563 566 14691728
Ishak K Baptista A Bianchi L Callea F De Groote J Gudat F Denk H Desmet V Korb G MacSween RN Phillips MJ Portmann BG Poulsen H Scheuer PJ Schmid M Thaler H Histological grading and staging of chronic hepatitis J Hepatol 1995 22 696 699 7560864 10.1016/0168-8278(95)80226-6
Erel O A novel automated direct measurement method for total antioxidant capacity using a new generation, more stable ABTS radical cation Clin Biochem 2004 37 277 285 15003729 10.1016/j.clinbiochem.2003.11.015
Erel O A new automated colorimetric method for measuring total oxidant status Clin Biochem 2005 xx xxx xxx (Article in press-DOI information: 10.1016/j.clinbiochem.2005.08.008)
Cao G Prior RL Comparison of different analytical methods for assessing total antioxidant capacity of human serum Clin Chem 1998 44 1309 1315 9625058
Miyazawa T Determination of phospholipid hydroperoxides in human blood plasma by a chemiluminescence-HPLC assay Free Radic Biol Med 1989 7 209 217 2806945 10.1016/0891-5849(89)90017-8
Wayner DD Burton GW Ingold KU Barclay LR Locke SJ The relative contributions of vitamin E, urate, ascorbate and proteins to the total peroxyl radical-trapping antioxidant activity of human blood plasma Biochim Biophys Acta 1987 924 408 419 3593759
Miller NJ Rice-Evans C Davies MJ Gopinathan V Milner A A novel method for measuring antioxidant capacity and its application to monitoring the antioxidant status in premature neonates Clin Sci (Lond) 1993 84 407 412 8482045
Harma M Harma M Erel O Oxidative stress in women with preeclampsia Am J Obstet Gynecol 2005 192 656 657 15696019 10.1016/j.ajog.2004.07.094
Janaszewska A Bartosz G Assay of total antioxidant capacity: comparison of four methods as applied to human blood plasma Scand J Clin Lab Invest 2002 62 231 236 12088342 10.1080/003655102317475498
Prior Rl Cao G In vivo total antioxidant capacity: comparison of different analytical methods Free Radic Biol Med 1999 27 1173 1181 10641708 10.1016/S0891-5849(99)00203-8
Koracevic D Koracevic G Djordjevic V Andrejevic S Cosic V Method for the measurement of antioxidant activity in human fluids J Clin Pathol 2001 54 356 361 11328833 10.1136/jcp.54.5.356
Schlesier K Harwat M Bohm V Bitsch R Assessment of antioxidant activity by using different in vitro methods Free Radic Res 2002 36 177 187 11999386 10.1080/10715760290006411
Ohshima H Tatemichi M Sawa T Chemical basis of inflammation-induced carcinogenesis Arch Biochem Biophys 2003 417 3 11 12921773 10.1016/S0003-9861(03)00283-2
Videla LA Rodrigo R Orellana M Fernandez V Tapia G Quinones L Varela N Contreras J Lazarte R Csendes A Rojas J Maluenda F Burdilas P Diaz JC Smok G Thielemann L Poniachik J Oxidative stress-related parameters in the liver of non-alcoholic fatty liver disease patients Cli Sci 2004 106 261 268 10.1042/CS20030285
Farinati F Cardin R Degan P De Maria N Floyd RA Van Thiel DH Naccarato R Oxidative DNA damage in circulating leukocytes occurs as an early event in chronic HCV infection Free Radic Biol Med 1999 27 1284 1291 10641722 10.1016/S0891-5849(99)00161-6
Grattagliano I Vendemiale G Caraceni P Domenicali M Nardo B Cavallari A Trevisani F Bernardi M Altomare E Starvation impairs antioxidant defense in fatty livers of rats fed a choline-deficient diet J Nutr 2000 130 2131 2136 10958803
|
16262897
|
PMC1283973
|
CC BY
|
2021-01-04 16:28:16
|
no
|
BMC Infect Dis. 2005 Oct 31; 5:95
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-95
|
oa_comm
|
==== Front
BMC Med ImagingBMC Medical Imaging1471-2342BioMed Central London 1471-2342-5-71620217610.1186/1471-2342-5-7Research ArticleMindboggle: Automated brain labeling with multiple atlases Klein Arno [email protected] Brett [email protected] Satrajit [email protected] Jason [email protected] Joy [email protected] fMRI Research Center, Columbia University, New York, USA2 Parsons Institute for Information Mapping, The New School, New York, USA3 New York State Psychiatric Institute, Columbia University, New York, USA4 Speech Communication Group, Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, USA5 Department of Cognitive and Neural Systems, Boston University, Boston, USA2005 5 10 2005 5 7 7 19 2 2005 5 10 2005 Copyright © 2005 Klein 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 make inferences about brain structures or activity across multiple individuals, one first needs to determine the structural correspondences across their image data. We have recently developed Mindboggle as a fully automated, feature-matching approach to assign anatomical labels to cortical structures and activity in human brain MRI data. Label assignment is based on structural correspondences between labeled atlases and unlabeled image data, where an atlas consists of a set of labels manually assigned to a single brain image. In the present work, we study the influence of using variable numbers of individual atlases to nonlinearly label human brain image data.
Methods
Each brain image voxel of each of 20 human subjects is assigned a label by each of the remaining 19 atlases using Mindboggle. The most common label is selected and is given a confidence rating based on the number of atlases that assigned that label. The automatically assigned labels for each subject brain are compared with the manual labels for that subject (its atlas). Unlike recent approaches that transform subject data to a labeled, probabilistic atlas space (constructed from a database of atlases), Mindboggle labels a subject by each atlas in a database independently.
Results
When Mindboggle labels a human subject's brain image with at least four atlases, the resulting label agreement with coregistered manual labels is significantly higher than when only a single atlas is used. Different numbers of atlases provide significantly higher label agreements for individual brain regions.
Conclusion
Increasing the number of reference brains used to automatically label a human subject brain improves labeling accuracy with respect to manually assigned labels. Mindboggle software can provide confidence measures for labels based on probabilistic assignment of labels and could be applied to large databases of brain images.
==== Body
Background
When comparing structures or functions across brains, it is common to label the gross anatomy of brain image data and to compare the structures or functions that lie within anatomically labeled regions. Since brains differ in their anatomy [1-10], it would seem reasonable to refer to the anatomy of many brains when labeling an individual subject's brain image. Atlases are manually labeled brains used as references. Using every atlas from a group of atlases independent of each other was found to give labeling results superior to those obtained by selecting the closest matching single atlas from the group, the average atlas, or an individual atlas, for the case of confocal microscopy images of bee brains [11]. However, labeling a subject's brain image with many different brains presents unreasonable demands on human labelers, who may not be consistent in their label assignments [12-15]. Fully automated labeling would facilitate large-scale labeling efforts while adding efficiency and consistency.
Image registration software (reviewed in [16-18]) may be used to coregister subject and atlas brain images, thereby labeling the subject images with superimposed atlas labels. There exist many different nonlinear image registration and feature-matching approaches to this problem [19-62]. Mindboggle software (see below) offers certain advantages over most of these approaches: it does not make the same assumptions about preserving topography from brain to brain, is relatively fast, and it performed well in comparison tests with standard image registration software packages (AIR, SPM2, ANIMAL, and linear registration with FLIRT) and in artificial lesion tests [63].
Having an automated registration or feature-matching program and a database of atlases introduces the problem of how to reconcile the multiple atlas label sets when labeling a single subject's brain. Labels could be assigned based on the selection or construction of similar or representative anatomy from these atlases. It is becoming more common to label subject brain image data with a single, composite atlas representing some average of multiple brain atlases (an average brain atlas) or retaining information about the differences between the atlases or between the atlases and the subject brain image (a probabilistic brain atlas).
Average brain atlases attempt to assign to each voxel (volume element) a representative value associated with image intensity or anatomical label. An intensity-based average brain atlas is the voxelwise mean intensity across individual brain images after linear [64-66] or nonlinear [67] coregistration. Additionally or alternatively, an average brain atlas may represent average sulcus shapes and positions computed in the original brain image space [8,68,69] or in an alternative space such as on a sphere [70]. An example of a label-based average brain atlas was constructed by Hammers, et. al. [71], where the majority label was computed for each voxel across 20 manually labeled brains after nonlinear registration to the MNI152 [64] template using SPM99 [27]. The use of an average atlas presupposes that there is such a thing as a representative brain and does not usually account for variability across brains.
Probabilistic brain atlases, on the other hand, do provide additional statistical information across the population used to construct the atlas [62,72-82]. This information may be related to the variance of landmark positions [73], probability of anatomical labels [44,79,83,84], probability of tissue classes [80], or multiple anatomical dimensions, for example characteristics of surface geometry and Bayesian priors associated with neighborhood relations between labels [62], and the multi-dimensional atlases under development by Mazziotta and Zilles and their colleagues [72,77,78,81]. An abstract representation of a database of manually labeled brains can also serve as a probabilistic atlas; for example, expert neural networks trained on a learning database of such brains [48] or graphs relating parametric surfaces [36]. However, there are only two examples known by the authors in which a complete cortical atlas is constructed from multiple label sets where each label set was assigned manually [62,71], rather than by automating the labeling of many brains without independent validation of the labeling technique. As with average brain atlases, probabilistic atlases have primarily been used as templates to which a subject brain is transformed and compared. This comparison presupposes that the single transform will account for differences between the subject brain image and each of the multiple brain images that were used to construct the atlas.
In this paper, we have chosen to extend the use of an individual atlas to multiple atlases in a recently introduced, fully automated, feature-based nonlinear labeling method called Mindboggle (freely downloadable, open source Matlab code) [63,85]. Rather than use a single (average or probabilistic) atlas, Mindboggle employs each atlas in a database independently to label the cortical voxels of a subject brain image, and for each voxel chooses the majority label assigned by the different atlases. We explore the effects of using two different labeling schemes and variable numbers of atlases on labeling accuracy and on the numbers of labels assigned per voxel.
Methods
Image acquisition
We used two sets of T1-weighted MRI data from a total of 20 young, healthy adult subjects. The first group of 10 subjects was scanned at the MGH/MIT/HMS Athinoula A. Martinos Center for Biomedical Imaging using a 3T Siemens scanner and standard head coil (TE: 2.9 ms, TR: 6.6 ms, flip angle: 8°). The in-plane resolution was approximately 1 × 1 mm, the slice thickness was 1.33 mm, and the dimensions and field of view were 256 × 256 voxels. These subjects consist of four men and six women between the ages of 22 and 29 years old (μ = 25.3). All are right-handed. The data were bias-corrected, affine-registered to the MNI152 template [64], and segmented using SPM2 software [27].
The second group of 10 subjects was scanned at Columbia University on a 1.5T GE scanner (TE: 5 ms, TR: 34 ms, flip angle: 45°). Slice thickness was 1.5-mm axial, in-plane resolution was 0.86 mm. Images were resliced coronally to a slice thickness of 3 mm, rotated into cardinal orientation, then segmented and parcellated using Cardviews software from MGH. These subjects consist of five men and five women between the ages of 26 and 41 years old (μ = 32.7).
Image processing before applying Mindboggle algorithm
Mindboggle calls on third-party software to perform three preliminary steps on a subject brain image: (1) cropping non-brain matter, (2) linear coregistration with the MNI152 template [64], and (3) segmentation into gray matter, white matter, and cerebrospinal fluid. For this study, these steps were performed by (1) BET [90], (2) FLIRT [91] set to correlation ratios, 12-parameter affine transforms and trilinear interpolation, and (3) SPM2 [27] for the first group of 10 brains and FAST [92] for the second group of 10 brains.
Mindboggle algorithm
Mindboggle is a freely downloadable, open source software package written in Matlab (version 6, release 13, with the Image Processing Toolbox, The Mathworks Inc., USA) and has been tested on different models of desktop and laptop computers running different distributions of Linux, as well as MacOSX and Windows. The general system requirements are the basic requirements of the Matlab environment. The system used to conduct the following tests consists of a 2.2 GHz Pentium IV processor running Redhat Linux 9.0 on a PC with 1 GB memory. Mindboggle was selected as the nonlinear method because it was created by one of the authors (AK) and performed favorably in comparisons with the popular nonlinear methods AIR, ANIMAL, and SPM2 [63].
Mindboggle's general strategy is to fill a subject's cortical gray matter mask with atlas labels, based on correspondences found between structures in a subject image and in one or more atlases (see Figure 1). Details of the original algorithm may be read in [63], and consist of the following five steps performed on a subject's brain image data:
If we divide the voxels into groups, by the number of different labels per voxel, as in Figure 10, we may see that there is an inverse relationship between the number of different labels and the label agreement with manual labels. Therefore, the number of labels per voxel provides a rough confidence measure for the majority label assigned to each voxel.
(1) extract cerebral cortical sulci,
(2) prepare hundreds of pieces from image-processed versions of these sulci,
(3) match each piece from an atlas with a combination of pieces from the subject,
(4) translate local atlas label boundaries according to the difference in position between each match, and
(5) warp the atlas label volume to the transformed boundaries and propagate these labels to fill a subject mask. Mindboggle optionally resets planar boundaries for frontal and temporal poles as well as the occipital lobes, if the atlas itself is labeled using these planar boundaries.
Mindboggle extracts cerebral cortical sulci in the following manner (see Note 1 in Appendix). First, Mindboggle crops exposed brain surface by eroding the segmented cortex three voxels deep. Mindboggle also crops subcortex and cerebellum with a mask constructed from a union of two of the Montreal Neurological Institute's atlases: the single-subject atlas [93] and the MNI152 template [64]. All registration and labeling by Mindboggle is performed in MNI152 space (resolution of 1 × 1 × 1 mm and dimensions of 181 × 217 × 181 voxels).
Sulcus pieces are constructed as follows (see Figure 2). The segmented gray matter with cerebrospinal fluid is thinned to a pixel-wide skeleton for each slice (Matlab's bwmorph.m function). All of the skeletonized slices are stacked to create a 3-D skeleton. This skeleton is split by an interhemispheric plane formed by warping a vertical plane to the medial slab of the skeleton using a modified Self-Organizing Map algorithm (see Note 2 in Appendix). The skeleton is then broken up into pieces as follows. Starting from the top slice of the skeleton, each set of connected pixels is considered a separate piece. Each pixel in the slice below is assigned membership to the nearest piece in the above slice. The latter operation is repeated from top to bottom, as well as from bottom to top, resulting in two independent sets of candidate pieces, with each pixel having two assignments, one for each set. A single set of pieces is obtained by identifying the unique set of pairs of assignments. The 3-D pieces are then fragmented using a k-means algorithm and regrouped together if they share extensive borders. This last regrouping step is conducted so that compact structures with a low surface-to-volume ratio such as a ball do not get broken up in arbitrary ways by the k-means algorithm. "Extensive borders" is defined as a ratio of border to surface voxels equal to at least one-tenth, where a border voxel has at least one other piece in its immediate neighborhood of six voxels, and a surface voxel has fewer than six occupied voxels in its neighborhood.
Finding similar pieces in an atlas helps to determine how to transform atlas label boundaries, and therefore how to distribute atlas labels in the subject brain. Matching each piece from an atlas with a combination of (up to three) similar pieces from the subject is performed by minimizing a cost function. The cost function consists of a sum of normalized quantities derived from differences in: mean position, number of voxels, number of subvolumes, and non-overlap. Differences in mean position and number of voxels are measures of the differences in location and size between the atlas and subject pieces. The number of subvolumes for a given piece is the number of 5 × 5 × 5-voxel boxes dividing the image volume that contain the piece. This measure is useful for distinguishing between pieces that have different spatial distributions, such as between a tight ball and an extensive sheet. Non-overlap of two pieces, P1 and P2, is equal to the fraction of subvolumes of P1 that do not overlap P2 added to the fraction of subvolumes of P2 that do not overlap P1. This measure is useful for distinguishing between differently shaped pieces that may otherwise be similar according to the other three measures.
Atlas label boundaries are locally translated according to the difference in position between nearby atlas and matching subject pieces. The translation is the difference of the mean of the local boundary from the mean of the subject piece(s), plus the difference between the mean of the atlas piece from the mean of the local boundary (after scaling by the ratio of the atlas and subject piece bounding boxes).
The atlas label volume is then warped to the transformed atlas label boundaries as follows (see Note 3 in Appendix). The atlas label that was closest to each original boundary point moves to the transformed boundary point, carrying along its neighboring labels as a function of their distance from the point (according to a Gaussian distribution function). After warping, each unlabeled voxel within the segmented gray matter mask is assigned the majority label in its 5 × 5 × 5-voxel neighborhood; this last step is repeated several times.
Evaluation
We evaluated labels assigned by Mindboggle to a brain image (in MNI152 space) by comparing them with the manual labels for that brain (linearly registered to MNI152 space). The manual labels used for evaluation were also used to construct Mindboggle atlases. They were assigned by a single human labeler to each of the 20 subject brains (before linear registration to the MNI152 space), according to one of two different parcellation schemes. The first group of 10 subjects was labeled by Jason Tourville according to a scheme that is a modified version of Cardviews (see below) and implemented in a software tool developed by Satrajit Ghosh at the Department of Cognitive and Neural Systems, Boston University [94]. The second group of 10 subjects was labeled by Olga Kambalov according to the Cardviews parcellation scheme, created at the Center for Morphometric Analysis, Massachusetts General Hospital, and implemented in Cardviews software [12]. The labeler for each group of subjects is an expert in Cardviews.
For both parcellation schemes, 74 cortical labels were selected from the original 96 labels and merged to give 36 labels (18 per hemisphere): superior, middle, and inferior frontal and temporal gyrii, frontal and temporal poles, pre- and postcentral gyrii, superior and inferior parietal lobules, occipital lobe, fusiform, lingual/parahippocampal, and orbital (frontal) gyrii, insula, and cingulate gyrus. The anatomical divisions are coarser than those of Cardviews primarily because regions divided by planes in the Cardviews approach are combined.
Figure 3 presents an isosurface representation of a single manually labeled subject brain. To determine whether increasing the number of atlases would improve the accuracy of Mindboggle labeling, we compared the manual label for each voxel of a subject image with the majority of all Mindboggle labels for that voxel, for an increasing number of atlases used to assign labels. When determining the majority label, ties were broken by random selection. Each subject is automatically labeled by a random selection of atlases for each number of atlases. For comparisons up to nine atlases, the atlases were randomly selected from within the same subject pool; for comparisons up to 19 atlases, atlases were randomly selected from either subject pool.
The primary evaluation measure we employ is percent label agreement between atlas labels and manual labels assigned to a subject's segmented gray matter mask, with each gray matter voxel having one manual and one automated (Mindboggle) label. The agreement between atlas label set Ai and manual label set Mi is defined as the volume of intersection divided by the volume of the manually labeled region, computed in voxels and summed over a set of multiple labeled regions each with index i, where |.| indicates number of voxels:
Our type I error, a measure of how many incorrect labels are found in a given manually labeled region, is simply equal to one minus the label agreement for that region. We define a type II error for a given manually labeled region as the number of automatically labeled voxels outside the region that have been assigned that region's label, divided by the total number of automatically labeled voxels with that label. This is equal to one minus the fraction of voxels automatically assigned a given label that lies within the corresponding region:
These error measures assume that the manual labels are correct, and they can range from zero to one; a value of zero is achieved when automated and manual labels perfectly overlap for each label.
Another evaluation measure we employ is percent label accord [12], the intersection between two similarly labeled regions divided by the mean volume of the two regions:
The above voxel-based measures ignore misregistration within a labeled region. Any conclusions based on them must therefore be restricted to the labeled volumes and may not be applicable to finer resolutions.
Results and discussion
We found there to be greater disagreement between atlases as the number of atlases increases, as one would expect. This is clearly demonstrated in Figures 4 and 5. Figure 4 displays the anatomical distribution of the number of different labels assigned by the atlases to each voxel. Figure 5 plots the total number of voxels with a given number of different labels per voxel. Both figures present their data as a function of the number of atlases. If we compare Figure 4 with Figure 3 (it is the same subject), the disagreements are clustered about anatomical boundaries, with the highest numbers of labels per voxel at the boundaries between multiple anatomical regions, as one would expect.
Figure 6A demonstrates the variability in labeling errors when different single atlases are used to label one subject. Figure 6B demonstrates the effect of the use of multiple atlases on labeling errors for the same subject. Figure 6B indicates that increasing the number of atlases reduces labeling errors in Mindboggle. Figure 7 demonstrates that the two subject populations, manually labeled with slightly different parcellation schemes, give clearly separable labeling results, and that the label agreement between manual labels and (voxelwise majority) Mindboggle labels remains distinct between the two subject groups even as the number of atlases increases. Each member from the first group of 10 subjects was labeled with one atlas from the same group, then two, three, up to nine atlases, with each atlas selected at random from the remaining unselected atlases. The same procedure was repeated for each member from the second group of 10 subjects. For each member of the combined subject population, one to 19 atlases are selected at random from either subject group.
Therefore, Mindboggle is sensitive to variance in the subject population and to the parcellation scheme used to manually label the atlases, in particular to the vertical planes that are used to define boundaries to large regions (occipital lobe, frontal and temporal poles). These planes are not positioned by the sulcus piece matching stage but by an automated identification and matching of specific anatomical landmarks. The definitions of these landmarks may be different between parcellation schemes and may not be as consistently or as accurately determined manually or automatically in one scheme versus another. Some of the differences between the results obtained by the two subject groups (see Methods: Image Acquisition) may be attributed to the broader sampling in the second group of subjects (three races versus one, unknown vs. right-handed, and much wider age range). We can expect even greater deviations from brains that are very young, very old, or inflicted with a pathological condition, something we are presently investigating.
Even with this dependence of absolute results on parcellation scheme, we may determine whether there is a relative improvement of results across all subjects as a function of the number of atlases used to obtain the voxelwise majority labels. From Table 1 and the accompanying graph in Figure 8, we may see that increasing the number of atlases asymptotically increases mean label agreement with manual labels. A one-way ANOVA was performed to test if the means are the same for the label agreements obtained by the different numbers of atlases. A multiple comparison test was then performed using Tukey's honestly significantly difference criterion to determine which pairs of means are significantly different. We see from Figure 9 that simply increasing the number of atlases from one to at least four results in a statistically significant increase in label agreement (p ≪ 10-6 for all comparisons), and further increasing the number of atlases to at least nine (or at least seven for the first set of subjects) results in a statistically significant increase in label agreement compared with using three atlases. However, the increase in label agreement from four to five or more atlases is not statistically significant for the mixed subject group.
One should not conclude based on these data that atlas databases need only contain four or five atlases to be representative. The standard deviations for our subject pool were high enough to warrant further investigation into sources of error. These sources include morphological dissimilarities between subject subpopulations, different parcellation schemes, and limitations of the Mindboggle algorithm. Interestingly, Kittler et al. [86] found that the classification performance of the voting rule applied to face and voice biometric data also peaked at four to five experts (atlases).
These results corroborate the conclusion of a study on atlas selection strategies applied to confocal microscopy images of bee brains, that labeling a brain image using every one of a group of atlases gives results superior to selecting an individual atlas [11]. However, when they tested the individual atlas condition, they chose only a single favorable atlas from a group of 20, whereas in the present study we ran tests using each and every single individual atlas from a group of 20.
The majority voting rule is probably not the optimal way to decide on a voxel's label [86-89], especially if the selected atlases deviate considerably from the subject brain to be labeled. A missing or unusual structure in a subject brain represented in only a minority of the atlases would most likely result in an inappropriate label. Rather than simply weighting the contribution of each of the atlases equally, each atlas vote for each subject voxel could be weighted by a function of the matching cost for the structure containing that voxel, since Mindboggle's matching cost function is intended to determine degree of correspondence between structures across brains.
We further separated the results by labeled region, to compare label agreement and type II errors between manual and Mindboggle labels for each label. As may be seen in Table 2, different numbers of atlases provide significantly higher label agreements for specific brain regions. Caviness et al. [12] found the percent label accord between two expert human labelers, the manual inter-rater reliability, to be 80.23% (σ = 8.08%) averaged across all 96 labels in four brains. Since we found, as did Caviness et al., a weak correlation between percent accord and region size, we should expect that a manual inter-rater reliability for our parcellation's fewer and larger regions to be somewhat higher than 80%. The problem with making a direct comparison between the same number and sizes of parcellation units is that Mindboggle relies solely on structural features to define anatomical boundaries whereas the Caviness approach also uses planes that extend far from the structural features used to construct the planes. We are presently evaluating Mindboggle on the entire set of 96 labels. The percent label accords obtained by Mindboggle in this study range in value across the different labeled regions, and average to 79.86% (a = 4.18%) for subject group 1 and 76.23% (σ = 5.17%) for subject group 2 (9 atlases for each subject), with the highest accords (> 90%) for the largest regions, the frontal and temporal poles and occipital lobes, and the lowest accords (< 70%) for the postcentral gyrii. The fact that the Mindboggle vs. manual accuracy is comparable to the reported inter-rater reliability is very encouraging.
For a single atlas to label a single subject, Mindboggle presently takes less than 17 minutes after linear registration and gray matter segmentation on a 2.2 GHz Pentium IV processor running Redhat Linux 9.0 on a PC with 1 GB memory: 1.3 minutes to construct a sulcus skeleton, 2.5 minutes to divide the skeleton with an interhemispheric plane, 3.4 minutes to construct and tally data on sulcus pieces, 2.5 minutes to find matching pieces in the atlas and to transform them from the atlas to the subject brain, and the remaining 7 minutes to warp and propagate labels through the gray matter mask. For each additional atlas, matching, warping, and labeling takes under 10 minutes if performed sequentially. For example, labeling a subject using five atlases would take 17 minutes if conducted in parallel, or an hour if conducted sequentially. The run time would reduce significantly not only by running Mindboggle for each atlas in parallel, but also by implementing faster preprocessing algorithms and optimized code rewritten in a lower-level language such as C as opposed to Matlab.
We conclude that by using multiple atlases, the overall label agreement between manual labels and the majority labels assigned by these atlases significantly improves when using a nonlinear procedure such as Mindboggle.
We are now in the process of applying this multiple atlas extension of Mindboggle to anatomically label functional activity data. Combining a confidence measure for anatomical boundaries derived from multiple atlases with statistical maps of functional activity data across subjects should help to establish our level of confidence in reported functional findings.
Appendix
Note 1
Since this study was conducted, Mindboggle no longer crops any part of the brain's gray matter; sulci are instead extracted by creating a mask by morphologically closing white matter (using Matlab's imclose.m function). This was visually determined to result in better sulcus extraction.
Note 2
Mindboggle now splits the skeleton much more quickly and accurately with a surface constructed by selecting a medial slab of the skeleton, flattening the slab into a mean surface along the x-axis, and applying a median filter to the surface x values (Matlab's medfilt2.m).
Note 3
Mindboggle no longer warps the atlas label volume or fills unlabeled regions by the majority label in its neighborhood. Instead, it simply fills the transformed boundaries with nearby labels according to a distance function (Matlab's bwdist.m).
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AK invented Mindboggle, conceived of and executed the study, and drafted the manuscript. BM, SG, and JT contributed manually labeled brain images. JH sponsored and supported 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
Arno Klein would like to thank Jack Grinband for helpful discussions. Satrajit Ghosh and Jason Tourville were supported by NIH grant R01 DC02852 (Frank Guenther, PI).
Figures and Tables
Figure 1 Mindboggle flowchart.
Figure 2 Piece construction in Mindboggle. Shown on the left, from top to bottom, are the five steps Mindboggle takes to construct pieces from a subject brain image. Darker pixels (non-white matter) of a (1) segmented horizontal slice are (2) thinned to a skeleton which is (3) split into left and right hemispheres. (4) Contiguous pixels of the skeleton slice are grouped into 2-D pieces, and (5) these 2-D pieces are used to construct 3-D pieces, shown in cross-section (bottom) and in 3-D (right figure, showing the left side of the brain with the frontal pole facing left).
Figure 3 Manual labels. Manual labels for a single subject (left side, frontal pole facing left), drawn from the first subject pool (modified Cardviews labels). After these labels are registered to the common (MNI152) space, they are processed to construct one of the atlases for labeling with Mindboggle. This figure is an isosurface representation constructed with a Gaussian filter of radius three voxels. Missing data in vertical strips are due to incompletely labeled coronal sections.
Figure 4 Spatial distribution of the number of labels per voxel. The isosurface representations of this subject are colored to indicate the number of different labels assigned to each voxel by the different atlases (for example, gray indicates one label, when all atlases agree). From left to right, each brain has been labeled by Mindboggle using an increasing number of atlases (2, all 9 from the same subject group, and all 19 from both subject groups). As one would expect, increasing the number of atlases inceases the average number of different labels assigned to each voxel. Missing data in vertical strips are a result of incompletely labeled coronal sections, as in Figure 1. The data for all subjects and for every number of atlases are graphed in Figure 3.
Figure 5 Quantity of voxels with a given number of labels per voxel. This is a graph of all the subject data labeled with an increasing number of atlases, from which the single subject in Figure 2 was drawn. The total quantity of labeled subject voxels, representing the total volume of labeled gray matter, was around 700,000 voxels on average. Remaining data for voxels with five or more atlas labels were not included for clarity.
Figure 6 Labeling errors. Fig. 4A demonstrates the variability in the spatial distribution of labeling errors for a single atlas labeling a subject, across all atlases. Blue indicates voxels where at least one atlas disagrees with the subject's manual labels (union). Green indicates voxels where every atlas disagrees with the subject's manual labels (intersection). Fig. 4B demonstrates the effect of the use of multiple atlases on labeling errors. Red voxels are those whose manually assigned label disagrees with the majority of the labels assigned by Mindboggle using multiple atlases. If we look from left to right, we see that increasing the number of atlases reduces labeling errors. Atlas selection and isosurface representation match the conditions of Figure 2.
Figure 7 Percent label agreement by subject pool. The two subject groups are manually labeled with slightly different parcellation schemes. Each member from the first group of 10 subjects (green) was labeled with one atlas from the same group, then two, three, up to nine atlases, with each atlas selected at random from the remaining unselected atlases. The same procedure was repeated for each member from the second group of 10 subjects (magenta). As may be seen here, the percent label agreements obtained by Mindboggle are clearly separable between the two groups. Therefore, Mindboggle is sensitive to variance in the subject population, and to the parcellation scheme used to manually label the atlases.
Figure 8 Change in label agreement as a function of the number of atlases. Increasing the number of atlases results in an asymptotic increase in the mean label agreement between labels assigned manually and by Mindboggle. The error bars extend one standard deviation about the mean. Data from the first group of subjects alone are in green and from the second group alone are in magenta. Data from both groups, where one to 19 atlases are selected at random from either subject group, are in black. Table 1 contains the data used in this figure.
Figure 9 Comparison between label agreements obtained with different numbers of atlases. A one-way ANOVA was performed to test if the means are the same for the label agreements obtained by the different numbers of atlases. A multiple comparison test was then performed using Tukey's honestly significantly difference criterion to determine which pairs of means are significantly different. The graph displays the mean for each number of atlases with a 95% confidence interval around the mean, based on the Studentized range distribution. If intervals are disjoint, their means are considered significantly different. The label agreement obtained with a single atlas is in blue and any significantly different result is in red or green. Green results are significantly higher than gray results (using three atlases). Using at least four atlases resulted in significantly higher label agreements and lower type II errors than when using one atlas (p ≪ 10-6), suggesting that Mindboggle should be used with at least four atlases to benefit from the multiple atlas approach.
Figure 10 Label agreement for different numbers of atlases and labels per voxel. The data of Figure 6 are broken up here into subsets of voxels according to the number of different atlas labels assigned to each voxel. Voxel populations with fewer label assignments (greater agreement between the atlases) have higher label agreements with manual labels. Therefore, the number of labels per voxel provides a rough confidence measure for each voxel's label. The error bars extend one standard deviation about the mean. Remaining data for voxels with five or more labels are not included for clarity.
Table 1 Percent label agreement as a function of the number of atlases
A All subjects σ II Group 1 σ II Group 2 σ II
1 74.39 (3.90) 0.20 79.05 (1.06) 0.17 73.55 (2.43) 0.21
2 73.75 (2.81) 0.21 78.38 (1.32) 0.18 72.68 (2.07) 0.21
3 76.54 (2.93) 0.19 80.19 (1.42) 0.17 75.30 (2.00) 0.20
4 77.73 (2.78) 0.18 80.95 (1.43) 0.16 76.31 (1.40) 0.19
5 77.88 (2.67) 0.18 81.98 (1.39) 0.15 77.24 (1.83) 0.19
6 78.30 (2.52) 0.18 81.58 (1.42) 0.16 77.45 (1.66) 0.18
7 78.56 (2.75) 0.18 82.09 (1.37) 0.15 77.76 (1.78) 0.18
8 78.67 (2.86) 0.18 82.21 (1.28) 0.15 77.87 (1.93) 0.18
9 79.18 (2.59) 0.17 82.50 (1.20) 0.15 78.15 (1.94) 0.18
10 78.91 (2.60) 0.17
11 79.25 (2.75) 0.17
12 79.47 (2.78) 0.17
13 79.47 (2.86) 0.17
14 79.41 (2.84) 0.17
15 79.41 (2.83) 0.17
16 79.48 (2.57) 0.17
17 79.59 (2.74) 0.17
18 79.61 (2.76) 0.17
19 79.62 (2.74) 0.17
Percent label agreements and type II errors (II) are given for each number of atlases (A) used to label subject group 1, group 2, and all subjects. The accompanying graph is in Figure 6. Standard deviations are in parentheses (for type II errors, σ equals 0.02 for all subjects and 0.01 for groups 1 and 2).
Table 2 Regions whose label agreement improves with multiple atlases
Labels L R
frontal pole
sup. frontal
mid. frontal 11
inf. frontal
orbital 13 8
precentral 11
postcentral
sup. parietal 8 9
inf. parietal
temporal pole 4
sup. temporal 6 11
mid. temporal 17 12
inf. temporal 17 11
fusiform 7 11
lingual
occipital lobe
cingulate 9
insula 9 3
The effect of multiple atlases on label agreement varies by manually labeled region. Here we compare label agreements obtained by Mindboggle for each region by different numbers of atlases randomly selected from either of the two subject groups (up to 19 atlases for each of the 20 subjects). Numerical entries denote the minimum number of atlases that result in significantly higher label agreements than for single atlas data, by region (p ≪ 0.0001 for all regions except inferior temporal gyms: p < 0.01 and right temporal pole: p < 0.0002). For example, an entry of "9" means that significantly higher results were obtained using 9, 10, 11,... to 19 atlases versus using one atlas). The significance test is the same as that applied to the whole brain label agreement data.
==== Refs
Jouandet ML Tramo MJ Herron DM Hermann A Loftus WC Bazell J Gazzaniga MS Brainprints: computer-generated two-dimensional maps of the human cerebral cortex in vivo Journal of Cognitive Neuroscience 1989 1 88 117
Ono M Kubik S Abernathey CD Atlas of the cerebral sulci 1990 Stuttgart: Georg Thieme Verlag
Steinmetz H Seitz RJ Functional anatomy of language processing: neuroimaging and the problem of individual variability Neuropsychologia 1991 29 1149 1161 1791929 10.1016/0028-3932(91)90030-C
Rademacher J Caviness VS JrSteinmetz H Galaburda AM Topographical variation of the human primary cortices: implications for neuroimaging, brain mapping, and neurobiology Cerebral Cortex 1993 3 313 329 8400809
Steinmetz H Herzog A Huang Y Hacklander T Discordant brain-surface anatomy in monozygotic twins [correspondence] New England Journal of Medicine 1994 331 952 953 10.1056/NEJM199410063311419
Zilles K Shleicher A Langemann C Amunts K Morosan P Palomero-Gallagher N Schormann T Mohlberg H Bürgel U Steinmetz H Schlaug G Roland PE Quantitative analysis of sulci in the human cerebral cortex: development, regional heterogeneity, gender difference, asymmetry, intersubject variability and cortical architecture Human Brain Mapping 1997 5 218 221 10.1002/(SICI)1097-0193(1997)5:4<218::AID-HBM2>3.0.CO;2-6
Le Goualher G Argenti AM Duyme M Baaré WFC Hulshoff Pol HE Boomsma DI Zouaoui A Barillot C Evans A Statistical sulcal shape comparisons: application to the detection of genetic encoding of the central sulcus shape Neurolmage 2000 11 564 574 10.1006/nimg.2000.0559
Thompson PM Woods RP Mega MS Toga AW Mathematical/computational challenges in creating deformable and probabilistic atlases of the human brain Human Brain Mapping 2000 9 81 92 10680765 10.1002/(SICI)1097-0193(200002)9:2<81::AID-HBM3>3.0.CO;2-8
Zilles K Kawashima R Dabringhaus A Fukuda H Schormann T Hemispheric shape of European and Japanese brains: 3-D MRI analysis of intersubject variability, ethnical, and gender differences Neurolmage 2001 13 262 271 10.1006/nimg.2000.0688
Wright IC Sham P Murray RM Weinberger DR Bullmore ET Genetic contributions to regional variability in human brain structure: methods and preliminary results Neurolmage 2002 17 256 271 10.1006/nimg.2002.1163
Rohlfing T Brandt R Menzel R Maurer CR Jr Evaluation of atlas selection strategies for atlas-based image segmentation with application to confocal microscopy images of bee brains Neurolmage 2004 21 1428 1442 10.1016/j.neuroimage.2003.11.010
Caviness VS JrMeyer J Makris N Kennedy DN MRI-based topographic parcellation of human neocortex: an anatomically specified method with estimate of reliability Journal of Cognitive Neuroscience 1996 8 566 587
Fiez JA Damasio H Grabowski TJ Lesion segmentation and manual warping to a reference brain: intra- and interobserver reliability Human Brain Mapping 2000 9 192 211 10770229 10.1002/(SICI)1097-0193(200004)9:4<192::AID-HBM2>3.0.CO;2-Y
Lancaster JL Woldorff MG Parsons LM Liotti M Freitas CS Rainey L Kochunov PV Nickerson D Mikiten SA Fox PT Automated Talairach Atlas labels for functional brain mapping Human Brain Mapping 2000 10 120 131 10912591 10.1002/1097-0193(200007)10:3<120::AID-HBM30>3.0.CO;2-8
Towle VL Khorasani L Uftring S Pelizzari C Erickson RK Spire JP Hoffmann K Chu D Scherg M Noninvasive identification of human central sulcus: a comparison of gyral morphology, functional MRI, dipole localization, and direct cortical mapping Neurolmage 2003 19 684 697 10.1016/S1053-8119(03)00147-2
Maintz JBA An overview of medical image registration methods Symposium of the Belgian hospital physicists association (SBPH/BVZF) 1996 12 1 22
Maintz JBA Viergever MA A survey of medical image registration Medical Image Analysis 1998 2 1 36 10638851 10.1016/S1361-8415(01)80026-8
Toga EAW Brain Warping 1999 San Diego: Academic Press
Broit C Optimal registration of deformed images PhD thesis 1981 Department of Computer and Information Science, University of Pennsylvania, Philadelphia
Bajcsy R Kovacic S Multiresolution elastic matching Computer Vision Graphics Image Processing 1989 46 1 21
Bookstein FL Principal warps: thin plate splines and the decomposition of deformations IEEE Transactions on Pattern Analysis and Machine Intelligence 1989 11 567 585 10.1109/34.24792
Gee JC Reivich M Bajcsy R Elastically deforming 3D atlas to match anatomical brain images Journal of Computer Assisted Tomography 1993 17 225 236 8454749
Miller MI Christensen GE Amit Y Grenander U Mathematical textbook of deformable neuroanatomies Proceedings of the National Academy of Sciences 1993 90 11944 11948
Christensen GE Rabbitt RD Miller MI 3D brain mapping using a deformable neuroanatomy Phys Med Biol 1994 39 609 618 15551602 10.1088/0031-9155/39/3/022
Collins DL Holmes CJ Peters TM Evans AC Automatic 3-D model-based neuroanatomical segmentation Human Brain Mapping 1995 3 190 208 10.1002/hbm.460030304
Declerck J Subsol G Thirion JP Ayache N Ayache N Automatic retrieval of anatomical structures in 3D medical images Proc 1st Computer Vision, Virtual Reality and Robotics in Medicine, Nice, France 1995 Springer-Verlag, Berlin 153 162
Friston KJ Ashburner J Poline JB Frith CD Heather JD Frackowiak RSJ Spatial Registration and Normalization of Images Human Brain Mapping 1995 2 165 189 10.1002/hbm.460030303
Gee JC LeBriquer L Barillot C Haynor DR Bajcsy R Bayesian approach to the brain image matching problem Medical Imaging 1995: Image Processing, Proc SPIE, Bellingham, WA 1995 2434 145 156
Davatzikos C Spatial normalization of 3D brain images using deformable models Journal of Computer Assisted Tomography 1996 20 656 665 8708076 10.1097/00004728-199607000-00031
Thompson PM Toga AW A surface-based technique for warping three-dimensional images of the brain IEEE Transactions on Medical Imaging 1996 15 402 417 10.1109/42.511745
Christensen GE Rabbitt RD Miller MI Deformable templates using large deformation kinematics IEEE Transactions on Image Processing 1996 5 1435 1447 10.1109/83.536892
Ashburner J Neelin P Collins DL Evans A Friston K Incorporating prior knowledge into image registration Neurolmage 1997 6 344 352 10.1006/nimg.1997.0299
Sandor S Leahy R Surface-based labeling of cortical anatomy using a deformable atlas IEEE Transactions on Medical Imaging 1997 16 41 54 9050407 10.1109/42.552054
Davatzikos C Mapping image data to stereotaxic spaces: applications to brain mapping Human Brain Mapping 1998 6 334 338 9788069 10.1002/(SICI)1097-0193(1998)6:5/6<334::AID-HBM2>3.0.CO;2-7
Gaens T Maes F Vandermeulen D Suetens P Wells WM, Colchester A, Delp S Non-rigid multimodal image registration using mutual information Proc 1st Medical Image Computing and Computer-Assisted Intervention, Cambridge, MA, USA, Lecture Notes in Computer Science 1998 1496 Springer-Verlag, Heidelberg 1099 1106
Le Goualher G Collins L Barillot C Evans A Wells WM, Colchester A, Delp S Automatic identification of cortical sulci using a 3D probabilistic atlas Proc 1st Medical Image Computing and Computer-Assisted Intervention, Cambridge, MA, USA, Lecture Notes in Computer Science 1998 1496 Springer-Verlag, Heidelberg 509 518
Lohmann G Yves von Cramon D Burkhardt H, Neumann B Automatic detection and labelling of the human cortical folds in magnetic resonance data sets Proc 5th European Conference on Computer Vision, Freiburg, Germany 1998 Springer-Verlag, Berlin 369 381
Schormann T Zilles K Three-dimensional linear and nonlinear transformations: an integration of light microscopical and MRI data Human Brain Mapping 1998 6 339 347 9788070 10.1002/(SICI)1097-0193(1998)6:5/6<339::AID-HBM3>3.0.CO;2-Q
Woods RP Grafton ST Watson JDG Sicotte NL Mazziotta JC Automated image registration: II. Intersubject validation of linear and nonlinear models Journal of Computer Assisted Tomography 1998 22 153 165 9448780 10.1097/00004728-199801000-00028
Ashburner J Andersson JLR Friston KJ High–dimensional image registration using symmetric priors Neurolmage 1999 9 619 628 10.1006/nimg.1999.0437
Ashburner J Friston KJ Nonlinear spatial normalization using basis functions Human Brain Mapping 1999 7 254 266 10408769 10.1002/(SICI)1097-0193(1999)7:4<254::AID-HBM4>3.0.CO;2-G
Caunce A Taylor CJ Kuba A, Sámal M, Todd-Pokropek A Using local geometry to build 3D sulcal models Proc 16th Information Processing in Medical Imaging, Visegrád, Hungary, Lecture Notes in Computer Science 1999 1613 Springer-Verlag, Heidelberg 196 209
Christensen GE Kuba A, Sámal M, Todd-Pokropek A Consistent linear-elastic transformations for image matching Proc 16th Information Processing in Medical Imaging, Visegrád, Hungary, Lecture Notes in Computer Science 1999 1613 Springer-Verlag, Heidelberg 224 237
Collins DL Evans AC Toga AW Animal: automatic non-linear image matching and anatomical labeling Brain Warping 1999 San Diego: Academic Press 123 142
Collins DL Zijdenbos AP Baaré FC Evans AC Kuba A, Sámal M, Todd-Pokropek A ANIMAL+INSECT: Improved cortical structure segmentation Proc 16th Information Processing in Medical Imaging, Visegrád, Hungary, Lecture Notes in Computer Science 1999 1613 Springer-Verlag, Heidelberg 210 223
Gee JC On matching brain volumes Pattern Recognition 1999 32 99 111 10.1016/S0031-3203(98)00093-4
Kochunov P Lancaster J Thompson P Boyer A Hardies J Fox P Evaluation of octree regional spatial normalization method for regional anatomical matching Human Brain Mapping 2000 11 193 206 11098797 10.1002/1097-0193(200011)11:3<193::AID-HBM50>3.0.CO;2-Z
Rivière D Mangin JF Papadopoulos D Martinez JM Frouin V Régis J Delp SL, DiGioia AM, Jaramaz B Automatic recognition of cortical sulci using a congregation of neural networks Proc 3rd Medical Image Computing and Computer-Assisted Intervention, Pittsburgh, PA, USA, Lecture Notes in Computer Science 2000 1935 Springer-Verlag, Heidelberg 40 49
Cachier P Mangin JF Pennec X Rivière D Papadopoulos-Orfanos D Régis J Ayache N Niessen WJ, Viergever MA Multisubject non-rigid registration of brain MRI using intensity and geometric features Proc 4th Medical Image Computing and Computer-Assisted Intervention, Utrecht, Netherlands, Lecture Notes in Computer Science 2001 2208 Springer-Verlag, Heidelberg 734 742
Chui H Win L Schultz R Duncan J Rangarajan A Insana MF, Leahy RM A unified feature registration method for brain mapping Proc 17th Information Processing in Medical Imaging, Davis, CA, USA, Lecture Notes in Computer Science 2001 2082 Springer-Verlag, Heidelberg 300 314
Guimond A Roche A Ayache N Meunier J Three-dimensional multimodal brain warping using the demons algorithm and adaptive intensity corrections IEEE Transactions on Medical Imaging 2001 20 58 69 11293692 10.1109/42.906425
D'Agostino E Maes F Vandermeulen D Suetens P Dohi T, Kikinis R A viscous fluid model for multimodal non-rigid image registration using mutual information Proc 5th Medical Image Computing and Computer-Assisted Intervention, Tokyo, Japan, Lecture Notes in Computer Science 2002 2489 Springer-Verlag, Heidelberg 541 548
Fischl B Salat DH Busa E Albert M Dieterich M Haselgrove C van der Kouwe A Killiany R Kennedy D Klaveness S Montillo A Makris N Rosen B Dale AM Whole brain segmentation: Automated labeling of neuroanatomical structures in the human brain Neuron 2002 33 341 355 11832223 10.1016/S0896-6273(02)00569-X
Jaume S Macq B Warfleld SK Dohi T, Kikinis R Labeling the brain surface using a deformable multiresolution mesh Proc 5th Medical Image Computing and Computer-Assisted Intervention, Tokyo, Japan, Lecture Notes in Computer Science 2002 2488 Springer-Verlag, Heidelberg 451 458
Meier D Fisher E Parameter space warping: shape-based correspondence between morphologically different objects IEEE Transactions on Medical Imaging 2002 21 31 47 11838662 10.1109/42.981232
Rettman ME Han X Xu C Prince JL Automated sulcal segmentation using watersheds on the cortical surface Neurolmage 2002 15 329 344 10.1006/nimg.2001.0975
Shen D Davatzikos C HAMMER: Hierarchical Attribute Matching Mechanism for Elastic Registration Transactions on Medical Imaging 2002 21 1421 1439 10.1109/TMI.2002.803111
Cachia A Mangin JF Rivière D Papadopoulos-Orfanos D Kherif F Bloch I Régis J A generic framework for parcellation of the cortical surface into gyri using geodesic Voronoï diagrams Medical Image Analysis 2003 7 403 416 14561546 10.1016/S1361-8415(03)00031-8
D'Agostino E Maes F Vandermeulen D Suetens P An information theoretic approach for non-rigid image registration using voxel class probabilities Biomedical Image Registration (Second International Workshop), Philadelphia, PA, USA, Lecture Notes in Computer Science 2003 2717 Springer-Verlag, Heidelberg 122 131
D'Agostino E Modersitzki J Maes F Vandermeulen D Fischer B Suetens P Free-form registration using mutual information and curvature regularization Biomedical Image Registration (Second International Workshop), Philadelphia, PA, USA, Lecture Notes in Computer Science 2003 2717 Springer-Verlag, Heidelberg 11 20
Liu T Shen D Davatzikos C Goos G, Hartmanis J, van Leeuwen J Deformable registration of cortical structures via hybrid volumetric and surface warping Medical Image Computing and Computer-Assisted Intervention, Lecture Notes in Computer Science 2003 2879 Springer-Verlag, Heidelberg 780 787
Fischl B van der Kouwe A Destrieux C Halgren E Ségonne F Salat DH Busa E Seidman LJ Goldstein J Kennedy D Caviness V Makris N Rosen B Dale AM Automatically parcellating the human cerebral cortex Cerebral Cortex 2004 14 11 22 14654453 10.1093/cercor/bhg087
Klein A Hirsch J Mindboggle: a scatterbrained approach to automate brain labeling Neurolmage 2005 24 261 280 10.1016/j.neuroimage.2004.09.016
Evans AC Collins DL Milner B An MRI-based stereotactic brain atlas from 300 young normal subjects Proc of the 22nd Symposium of the Society for Neuroscience, Anaheim 1992 408
Andreasen NC Arndt S Swayze V 2ndCizadlo T Flaum M O'Leary DS Ehrhardt JC Yuh WTC Thalamic abnormalities in schizophrenia visualized through magnetic resonance image averaging Science 1994 266 294 298 7939669
Collins DL Neelin P Peters TM Evans AC Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space Journal of Computer Assisted Tomography 1994 18 192 205 8126267
Bookstein FL Colchester ACF, Hawkes DJ Thin-Plate Splines and the Atlas Problem for Biomedical Images Proc 12th Information Processing in Medical Imaging, Wye, UK, Lecture Notes in Computer Science 1991 511 Springer-Verlag, Heidelberg 326 342
Roland PE Graufelds CJ Wåhlin J Ingelman L Andersson M Ledberg A Pedersen J ÅAAkerman S Dabringhaus A Zilles K Human Brain Atlas: for high-resolution functional and anatomical mapping Human Brain Mapping 1994 1 173 184 10.1002/hbm.460010303
Guimond A Meunier J Thirion JP Wells WM, Colchester A, Delp S Automatic computation of average brain models Proc 1st Medical Image Computing and Computer-Assisted Intervention, Cambridge, MA, USA, Lecture Notes in Computer Science 1998 1496 Springer-Verlag, Heidelberg 631 640
Fischl B Sereno MI Tootell RBH Dale AM High-resolution intersubject averaging and a coordinate system for the cortical surface Human Brain Mapping 1999 8 272 284 10619420 10.1002/(SICI)1097-0193(1999)8:4<272::AID-HBM10>3.0.CO;2-4
Hammers A Allom R Koepp MJ Free SL Myers R Lemieux L Mitchell TN Brooks DJ Duncan JS Three-dimensional maximum probability atlas of the human brain, with particular reference to the temporal lobe Human Brain Mapping 2003 19 224 247 12874777 10.1002/hbm.10123
Mazziotta JC Toga AW Evans A Fox P Lancaster J A probabilistic atlas of the human brain: theory and rationale for its development. The International Consortium for Brain Mapping (ICBM) Neurolmage 1995 2 89 101 10.1006/nimg.1995.1012
Bookstein FL Höhne KH, Kikinis R Visualizing group differences in outline shape: methods from biometrics of landmark points Proc 4th Visualization in Biomedical Computing, Hamburg, Germany, Lecture Notes in Computer Science 1996 1131 Springer-Verlag, Heidelberg 405 410
Thompson PM Schwartz C Toga AW High-resolution random mesh algorithms for creating a probabilistic 3D surface atlas of the human brain Neurolmage 1996 3 19 34 10.1006/nimg.1996.0003
Thompson PM Toga AW Detection, visualization and animation of abnormal anatomic structure with a deformable probabilistic brain atlas based on random vector field tranformations Neurolmage 1996 1 271 294
Le Briquer L Gee JC Duncan JS, Gindi G Design of a statistical model of brain shape Proc 15th Information Processing in Medical Imaging, Vermont, USA, Lecture Notes in Computer Science 1997 1230 Springer-Verlag, Heidelberg 477 482
Mazziotta J Toga A Evans A Fox P Lancaster J Zilles K Woods R Paus T Simpson G Pike B Holmes C Collins L Thompson P MacDonald D Iacoboni M Schormann T Amunts K Palomero-Gallagher N Geyer S Parsons L Narr K Kabani N Le Goualher G Boomsma D Cannon T Kawashima R Mazoyer B A probabilistic atlas and reference system for the human brain: International Consortium for Brain Mapping (ICBM) Philosophical Transactions of the Royal Society of London, Series B: Biological Sciences 2001 356 1293 1322 10.1098/rstb.2001.0915
Mazziotta J Toga A Evans A Fox P Lancaster J Zilles K Woods R Paus T Simpson G Pike B Holmes C Collins L Thompson P MacDonald D Iacoboni M Schormann T Amunts K Palomero-Gallagher N Geyer S Parsons L Narr K Kabani N Le Goualher G Feidler J Smith K Boomsma D Hulshoff Pol H Cannon T Kawashima R Mazoyer B A four-dimensional probabilistic atlas of the human brain Journal of the American Medical Informatics Association 2001 8 401 430 11522763
Van Essen DC Drury HA Dickson J Harwell J Hanlon D Anderson CH An integrated software suite for surface-based analyses of cerebral cortex Journal of the American Medical Informatics Association 2001 8 443 459 11522765
Hill DLG Hajnal JV Rueckert D Smith SM Hartkens T McLeish K Dohi T, Kikinis R A dynamic brain atlas Proc 5th Medical Image Computing and Computer-Assisted Intervention, Tokyo, Japan, Lecture Notes in Computer Science 2002 2488 Springer-Verlag, Heidelberg 532 539
Abbott A A new atlas of the brain Nature (News Feature) 2003 424 249 250 10.1038/424249a
Koo B Lee JM Kim JS Lee JS Kim IY Kim JJ Lee DS Kwon JS Kim SI A probabilistic and statistical approach for Korean standard brain atlas [abstract] 9th Annual Meeting of the Organization for Human Brain Mapping, New York City 2003
Evans AC Collins DL Neelin P MacDonald D Kamber M Marrett TS Thatcher RW, Hallett M, Zeffiro T, John ER, Huerta M Three-dimensional correlative imaging: applications in human brain mapping Functional Neuroimaging: Technical Foundations 1994 San Diego: Academic Press 145 162
Dinov ID Mega MS Thompson PM Lee L Woods RP Holmes CJ Sumners DW Toga AW Analyzing functional brain images in a probabilistic atlas: a validation of subvolume thresholding Journal of Computer Assisted Tomography 2000 24 128 138 10667672 10.1097/00004728-200001000-00024
Klein A Mindboggle
Kittler J Alkoot FM Sum versus vote fusion in multiple classifier systems IEEE Transactions on Pattern Analysis and Machine Intelligence 2003 25 110 115 10.1109/TPAMI.2003.1159950
Kittler J Hatef M Duin RPW Matas J On combining classifiers IEEE Transactions on Pattern Analysis and Machine Intelligence 1998 20 226 239 10.1109/34.667881
Rohlfing T Russakoff DB Maurer CR Taylor CJ, Noble JA Expectation maximization strategies for multi-atlas multi-label segmentation Proc 18th Information Processing in Medical Imaging, Ambleside, UK, Lecture Notes in Computer Science 2003 2732 Springer-Verlag, Heidelberg 210 221
Warfield SK Zou KH Wells WM Simultaneous Truth and Performance Level Estimation (STAPLE): an algorithm for the validation of image segmentation IEEE Transactions on Medical Imaging 2004 23 903 921 15250643 10.1109/TMI.2004.828354
Smith S Fast robust automated brain extraction Human Brain Mapping 2002 17 143 155 12391568 10.1002/hbm.10062
Jenkinson M Smith S A global optimisation method for robust affine registration of brain images Medical Image Analysis 2001 5 143 156 11516708 10.1016/S1361-8415(01)00036-6
Zhang Y Brady M Smith S Segmentation of brain MR images through a hidden Markov random field model and the expectation maximization algorithm IEEE Transactions on Medical Imaging 2001 20 45 57 11293691 10.1109/42.906424
Tzourio-Mazoyer N Landeau B Papathanassiou D Crivello F Etard O Delcroix N Mazoyer B Joliot M Automated anatomical labelling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single subject brain [abstract] 8th Annual Meeting of the Organization for Human Brain Mapping, Sendai, Japan 2002
Tourville JA Guenther FH A cortical parcellation scheme for speech studies Boston University Technical Report CAS/CNS-03-022 2003 Boston, MA: Boston University
|
16202176
|
PMC1283974
|
CC BY
|
2021-01-04 16:03:35
|
no
|
BMC Med Imaging. 2005 Oct 5; 5:7
|
utf-8
|
BMC Med Imaging
| 2,005 |
10.1186/1471-2342-5-7
|
oa_comm
|
==== Front
BMC NephrolBMC Nephrology1471-2369BioMed Central London 1471-2369-6-111625091910.1186/1471-2369-6-11Research ArticleBarriers to successful care for chronic kidney disease Lenz Oliver [email protected] Durga P [email protected] Daniel V [email protected] Alessia [email protected] David [email protected] David [email protected] Division of Nephrology and Hypertension, University of Miami Miller School of Medicine, Miami, FL, USA2005 27 10 2005 6 11 11 31 5 2005 27 10 2005 Copyright © 2005 Lenz et al; licensee BioMed Central Ltd.2005Lenz 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 National Kidney Foundation has formulated clinical practice guidelines for patients with chronic kidney disease (K/DOQI). However, little is know about how many patients actually achieve these goals in a dedicated clinic for chronic kidney disease.
Methods
We performed a cross-sectional analysis of 198 patients with an estimated glomerular filtration rate of less than 30 ml/min/1.73 m2 and determined whether K/DOQI goals were met for calcium, phosphate, calcium-phosphate product, parathyroid hormone, albumin, bicarbonate, hemoglobin, lipids, and blood pressure.
Results
We found that only a small number of patients achieved K/DOQI targets. Recent referral to the nephrologist, failure to attend scheduled clinic appointments, African American ethnicity, diabetes, and advanced renal failure were significant predictors of low achievement of K/DOQI goals.
Conclusion
We conclude that raising awareness of chronic kidney disease and K/DOQI goals among primary care providers, early referral to a nephrologist, the exploration of socioeconomic barriers and cultural differences, and both patient and physician education are critical to improve CKD care in patients with Stage 4 and 5 CKD.
==== Body
Background
The National Kidney Foundation has recently launched a major effort to define Chronic Kidney Disease (CKD) and formulate clinical practice guidelines [1,2]. It has been clearly shown that complications of CKD, such as anemia, metabolic acidosis, nutritional deficits, secondary hyperparathyroidism, and hypertension, significantly contribute to morbidity and mortality [3-10]. It has been proposed that care for patients with CKD be best delivered in dedicated CKD clinics that provide a multidisciplinary approach to patients with CKD [11,12]. Typically, these clinics are staffed with nephrologists, dieticians, social workers, and educators, and the team works closely with vascular surgeons for access placement. However, little is known about the effectiveness of these clinics at academic centers. The purpose of this cross-sectional analysis is determine to what extent K/DOQI goals are achieved in a dedicated CKD clinic serving a urban, socio-economically disadvantaged minority population.
Methods
Patients
IRB approval for this study was obtained from the Human Subjects Research Office of the University of Miami, Miami, FL (protocol number 2004–3071). All study procedures were carried out in accordance with the Declaration of Helsinki regarding research involving human subjects. Our chronic kidney disease (CKD) clinic focuses on the care of patients with an estimated GFR, using the modified MDRD formula [1], of less than 30 ml/min/1.73 m2 (Stage 4 and 5). The clinic setting is described in more detail below. We screened 268 patients with an appointment scheduled between January and August of 2004 in the CKD clinic at Jackson Memorial Hospital, Miami, FL. Of those, 35 were excluded because they were not seen: 25 had already started renal replacement therapy, 1 patient had died, and 9 patients were lost to follow-up. Of the remaining 233 patients, an additional 35 were excluded: 7 patients did not have any laboratory data and thus the degree of their renal impairment could not be determined, and 28 patients had CKD Stages 1, 2 or 3 after having recovered from an episode of acute renal failure (Figure 1). No dialysis patients were included in this study. For patients who initiated renal replacement therapy during the study period, their last pre-dialysis parameters were used for analysis.
Figure 1 Recruitment. 268 patients had an appointment in the CKD clinic. Of those, 233 patients were seen at least once during the study period, and 198 met inclusion criteria.
Dedicated CKD clinic
All patients with Stages 4 or 5 CKD are followed in our dedicated CKD clinic. The clinic is located in an urban county hospital that predominantly serves socio-economically disadvantaged minority populations. The clinic takes place once a week. The clinic is staffed by two academic nephrologists and four senior nephrology fellows, a nurse practitioner, a social worker, a nurse educator, a case manager, a dietician, and a pharmacist. Patients with stage 4 CKD are seen by a physician at least every 3 months, more frequently if they have laboratory abnormalities or uncontrolled blood pressure. Patients with stage 5 CKD are seen monthly. Patients receiving recombinant erythropoietin for anemia or active vitamin D for secondary hyperparathyroidism are seen monthly by the nurse practitioner. The nurse educator gives a short lecture about dialysis options, vascular access, and complications of CKD at the beginning of each clinic session. In addition, one-on-one educational sessions are offered to patients who decide on a dialysis modality. The dietician accepts walk-in appointments during regular CKD clinic hours in addition to elective appointments for nutritional counseling. The case manager assists with referrals and follow-up appointments. The social worker assists mainly with financial issues, since most of our patients belong to socio-economically disadvantaged minority populations, as well as referrals to the transplant center. The pharmacist is available to answer patients' questions about their medications, drug and food interactions, adverse events, as well as programs offering financial assistance.
Nephrology fellows act as the patients' primary nephrologist. Each case is discussed with a staff nephrologist. Following a checklist, current data for core indicators for CKD care including anemia, calcium, phosphate and PTH metabolism, nutrition, metabolic acidosis, hyperlipidemia, blood pressure, and dialysis options including vascular access, are compared to target values. Algorithms are in place to start and adjust phosphate binders, calcium supplements, active vitamin D preparations, recombinant erythropoietin, iron supplementations, and to make decision regarding vascular access placement and transplant referral. These algorithms are based on published treatment guidelines [2,13-15]. K/DOQI guidelines do not contain treatment recommendations for low HDL, which we commonly treat with extended-release niacin.
Laboratory data are obtained quarterly for patients with stage 4 CKD and monthly for patients with stage 5 CKD. In patients not at goal, and in those receiving active vitamin D preparations or recombinant human erythropoietin supplementation, data are obtained monthly, and medication doses are adjusted accordingly.
The nephrology clinics receive referrals mostly from primary care providers (PCP), internists, cardiologists, and endocrinologists practicing within the hospital system and its affiliated community clinics. All referrals are triaged by their estimated GFR: patients with stage 4 or 5 CKD are directly seen in the CKD clinic, while all others are seen in the general nephrology clinic. PCP are strongly encouraged to refer all patients with Stage 3 CKD, or whenever there is a doubt in regards to diagnosis or therapy. During the study period, of the patients seen for the first time in the CKD clinic about half came from the general nephrology clinic with progressive renal failure, while the other half were new referrals. In patients seen in the nephrology clinics who reach Stage 4 CKD the nephrologist typically manages all problems related to chronic kidney disease.
Data collection
Laboratory parameters for serum calcium, phosphate, intact PTH, albumin, bicarbonate, and hemoglobin were obtained from chart review. For each patient, the most recent value prior to their last clinic appointment within the study period was used. Laboratory data had to be obtained no more than 3 months prior to the patients scheduled visit, otherwise they were entered as "not at goal". This is based on the K/DOQI recommendation to obtain laboratory data for the parameters investigated in this study at least quarterly. The serum calcium concentration was corrected for serum albumin using the formula [Cacorrected] = [Cameasured] + 0.8 × (4 - [Albumin]), where [Ca] is expressed in mg/dl and [Albumin] is expressed in g/dl [16]. The calcium-phosphate product was obtained by multiplying the corrected serum calcium with the serum phosphate concentration and expressed as mg2/dl2. The following treatment targets were used for patients with CKD Stages 4 and 5, respectively [2,13-15]: calcium 8.4–10.3/8.4–9.5 mg/dl, phosphate 2.7–4.6/3.5–5.5 mg/dl, and intact PTH 70–110/150–300 pg/ml. Identical treatment targets were set for patients with CKD Stage 4 and 5, respectively, for calcium-phosphate product (<55 mg2/dl2), albumin (≥3.5 g/dl), bicarbonate (>22 mmol/l), hemoglobin (≥ 11 g/dl), and blood pressure (less than 130/80 mmHg). The goals for lipid control were set at LDL < 100 mg/dl, triglycerides < 500 mg/dl, and non-HDL < 150 mg/dl. If any of the three lipid goals was not met, the subject was classified as "not at goal". Thus, a total of nine parameters were evaluated: 8 laboratory parameters plus blood pressure. Patient age, gender, ethnicity, and the following co-morbidities were abstracted from chart review: diabetes, hypertension, and hyperlipidemia. Failure to attend scheduled clinic visits half the time or more (no show rate ≥ 50%) was used as a surrogate measure of non-adherence. The length of nephrology care in months was tabulated; time spent in the CKD clinic, the general nephrology clinic, or under the care of a nephrologist in private practice was added. Patients who had been seen for 6 months or less by a nephrologist were categorized as having received short nephrology care (SNC), while all others were categorized having received long nephrology care (LNC).
Statistical analysis
All statistical analyses ware carried out using the SPSS statistical software package (SPSS Inc., Chicago, IL). We chose poor achievement of K/DOQI goals as the outcome variable, which was defined as having less than half the parameters, i.e., 0 to 4 out of 9, at goal, because we felt that it is clinically relevant if a patient has more than half of the investigated parameters not at goal. Categorical variables were compared by chi square analysis. Continuous variables were compared with Student's t-test if two groups were present or ANOVA followed by post hoc analysis if more than two groups were present. Welch's correction for unequal variances and Bonferroni's correction for multiple testing were employed where indicated. Associations were tested by logistic regression. The following variables were thought to be clinically relevant and entered into the model: age (converted into decades), gender, African American ethnicity, short nephrology care (SNC), failure to attend scheduled clinic visits (no show rate above 50%), stage 5 CKD, diagnosis of diabetes, hypertension, and hyperlipidemia.
Results
Patient characteristics
Patient characteristics are shown in Table 1. Median follow-up time for SNC patients was 2 months, while the median follow-up for LNC patients with nephrology care was 33 months. 44% of LNC patients had been followed for 3 years or more. Important differences between LNC patients and SNC were noted. There were fewer African Americans and more young patients among SNC patients. SNC patients were less likely to carry the diagnosis of hypertension and hyperlipidemia but more likely to have diabetes. About 43% of the patients were poor and uninsured, 18% had Medicaid only, and 26% had Medicare as their primary insurance. There was no difference in insurance status between SNC or LNC patients (data not shown).
Table 1 Patient characteristics
All (N = 198) LNC† (N = 146) SNC† (N = 52) P-value
Ethnicities
Hispanic 46% 44% 54% 0.214
African American 43% 48% 31% 0.032
Haitian 7% 7% 8% 0.839
Caucasian 3% 1% 5% 0.083
Asian 1% 0% 2% 0.093
Age [Years]
Mean ± SD 56 ± 13.4 51 ± 14 58 ± 12.7 0.001
<31 4% 1% 10% 0.006
31–40 10% 8% 14% 0.270
41–50 20% 19% 21% 0.758
51–60 25% 23% 31% 0.241
61–70 28% 32% 17% 0.041
71–80 12% 13% 8% 0.304
>80 3% 3% 2% 0.176
Gender
Female 53% 55% 48% 0.405
CKD Stage
Stage 4 47% 48% 44% 0.645
Length of Nephrology Care [Months]
0–1 9% 0% 35% -
2–6 17% 0% 65% -
7–12 14% 20% 0% -
13–24 15% 22% 0% -
24–36 15% 21% 0% -
>36 30% 44% 0% -
Co-Morbidities
Hypertension 92% 95% 83% 0.004
Hyperlipidemia 66% 71% 52% 0.015
Diabetes 50% 45% 65% 0.008
Estimated GFR at Referral to Nephrology [ml/min/1.73 m2]
Mean ± SD 26 ± 14.9 29 ± 15.7 17 ± 8.3 <0.001
<15 21% 13% 42% <0.001
15–29 52% 51% 52% 0.945
30–59 24% 31% 6% <0.001
>59 4% 5% 0% 0.108
† LNC: seen by a nephrologist for more than 6 months; SNC: seen by a nephrologist for 6 months or less. Data for LNC and SNC were compared using Chi Square Analysis for categorical variables or by unpaired t-test for continuous variables. Welch's correction was used if unequal variances were present.
Achievement of K/DOQI goals
The median values along with the 25th and 75th percentiles for estimated GFR, serum calcium, phosphate, calcium-phosphate product, intact PTH, bicarbonate, albumin, hemoglobin, lipids, and blood pressure are shown in Table 2. The proportion of patients achieving K/DOQI goals are shown in Table 3.
Table 2 Mean laboratory parameters and blood pressure values
Stage 4 CKD Stage 5 CKD
SNC† (N = 23) LNC† (N = 70) SNC† (N = 29) LNC† (N = 76) p-values
Estimated GFR [ml/min/1.73 m2] 22 ± 4* 21 ± 4# 9.1 ± 3* 9.1 ± 3# *#<0.001
Calcium [mg/dl] 9.5 ± 0.6 9.3 ± 0.6* 8.9 ± 0.9 8.9 ± 0.8* *0.015
Phosphate [mg/dl] 4.6 ± 1.1 4.1 ± 0.9* 5.2 ± 1.2 5.3 ± 1.5* *<0.001
Calcium-Phosphate Product [mg2/dl2] 45 ± 11 38 ± 9*# 46 ± 10# 47 ± 13* *<0.001; #0.035
iPTH [pg/ml] 155 ± 153* 203 ± 152# 532 ± 476* 486 ± 385# *0.008; #<0.001
Bicarbonate [mmol/l] 23 ± 4 25 ± 4*# 21 ± 5# 21 ± 4* *#<0.001
Albumin [g/dl] 3.3 ± 0.8 3.8 ± 0.5 3.5 ± 0.7 3.6 ± 0.6
Hemoglobin [g/dl] 11.3 ± 2.0 11.8 ± 1.2*# 10.3 ± 1.9# 10.5 ± 1.8* *<0.001; #0.003
Total Cholesterol [mg/dl] 190 ± 37 189 ± 40 180 ± 56 172 ± 41
Triglycerides [mg/dl] 153 ± 72 176 ± 120 183 ± 115 149 ± 85
LDL [mg/dl] 111 ± 30 103 ± 34 100 ± 49 94 ± 33
HDL [mg/dl] 48 ± 16 53 ± 15* 43 ± 15* 47 ± 16 *0.033
Non-HDL [mg/dl] 142 ± 38 137 ± 39 137 ± 54 125 ± 37
Systolic Blood Pressure [mmHg] 148 ± 27 135 ± 26* 155 ± 27* 147 ± 28
Diastolic Blood Pressure [mmHg] 82 ± 21 73 ± 14* 84 ± 17* 77 ± 16 *0.049
Shown are the most recent data obtained within the 3-months period preceding the last clinic visit. All data are shown as mean ± standard deviation. Stage 4 = GFR 15–29, Stage 5 = GFR <15 ml/min/1.73 m2. The glomerular filtration rate was estimated using the modified MDRD formula [1]. † SNC: nephrology care for 6 months or less; LNC: nephrology care for more than 6 months. Data in each row were analyzed by one-way ANOVA followed by pair wise post hoc comparisons. Welch's correction for unequal variances was used where indicated. Bonferroni's correction for multiple comparisons was applied within each row. Pairs compared in each row bear the same symbol. Only p < 0.05 are shown.
Table 3 Proportion of patients achieving K/DOQI targets
Stage 4 CKD Stage 5 CKD
SNC† (N = 23) LNC† (N = 70) p-value SNC† (N = 29) LNC† (N = 76) p-value
Calcium 87% 91% 0.529 79% 82% 0.791
Phosphate 48% 80% 0.003 52% 66% 0.185
Calcium-Phosphate Product 65% 94% <0.001 83% 78% 0.789
iPTH 9% 20% 0.213 24% 21% 0.733
Bicarbonate 70% 81% 0.230 38% 49% 0.323
Albumin 43% 76% 0.004 48% 62% 0.208
Hemoglobin 57% 77% 0.056 38% 42% 0.697
Lipids 35% 49% 0.249 34% 49% 0.191
Triglycerides 91% 94% 0.614 90% 96% 0.207
LDL 39% 53% 0.253 48% 55% 0.521
Non-HDL 52% 63% 0.253 66% 76% 0.249
Blood Pressure 17% 34% 0.132 10% 17% 0.382
Shown are the percentages of patients in each group with laboratory values within K/DOQI target ranges. Laboratory data were the most recent values within the 3 months period preceding the last clinic visit. See methods section for individual ranges. Stage 4 = GFR 15–29, Stage 5 = GFR <15 ml/min/1.73 m2. The glomerular filtration rate was estimated using the modified MDRD formula [1]. † LNC: seen by a nephrologist for more than 6 months; SNC: seen by a nephrologist for 6 months or less. Data for LNC and SNC were compared using Chi Square Analysis. Only p-values < 0.05 are shown.
Except for SNC patients with Stage 5 CKD more than 80% of all patients achieved calcium goals. Those outside the target range were more likely to have low calcium levels suggesting secondary hyperparathyroidism; calcium levels above target range were rare.
80% of LNC patients with Stage 4 CKD had phosphate levels at goal, and only 1 patient had a serum phosphate above 6 mg/dl. Only 66% of LNC patients with Stage 5 CKD had phosphate levels within the target range, 22% had phosphate levels above 6 mg/dl, and 13% had phosphate levels above 7 mg/dl. The highest phosphate level in this group was 10.5 mg/dl. Among SNC patients, only about half of all patients achieved phosphate goals, and 13% with Stage 4 CKD and 24% with Stage 5 CKD had phosphate levels above 6 mg/dl. None of these patients had serum phosphate levels above 7 mg/dl, suggesting a shorter disease course or poorer nutrition.
Given the relative high percentage of patients with low calcium concentrations, calcium-phosphate products were within target range for more than three quarters of all patients. The lower percentage of SNC patients with Stage 4 CKD achieving goal is due to the high number of missing data that were coded as not at goal. Missing data resulted from missing phosphate determinations.
Secondary hyperparathyroidism presented a serious problem in our cohort. Only between 9% and 24% of all patients achieved PTH goals. Among those with Stage 4 CKD who had a PTH determination on file, PTH levels were below target in 6%, 41% were at goal, 110–200 pg/ml in 24%, 200–400 pg/ml in 24%, and above 400 pg/ml in 5% of the patients. None of the patients with Stage 4 CKD had PTH levels above 1000 pg/ml. For those with Stage 5 CKD who had a PTH determination on file, PTH levels were below target in 15%, at goal in 29%, 300–600 pg/dl in 28%, 600–900 pg/ml in 17% and above 900 pg/ml in 11% of the patients. Particularly in SNC patients, missing PTH determinations that were coded as not at goal resulted in the overall very low number of patients achieving PTH targets.
Between 70% and 80% of patients with Stage 4 CKD had good control of metabolic acidosis, while only 38% to 49% of patients with Stage 5 CKD had adequate bicarbonate levels. There was no significant difference between SNC and LNC patients.
Albumin was chosen as a surrogate marker for malnutrition. A higher proportion of LNC patients had normal serum albumin concentrations than SNC patients; however, this difference was no longer significant in patients with Stage 5 CKD. 9% of LNC patients with Stage 4 CKD and 13% of LNC patients with Stage 5 CKD had albumin concentrations below 3 g/dl. For SNC patients, the corresponding proportions were 35% and 24%, respectively.
More than three quarters of LNC patients with Stage 4 CKD had adequate anemia management, while only 57% of SNC patients with Stage 4 CKD had desirable hemoglobin concentrations. Hemoglobin concentrations for LNC patients with Stage 4 CKD were between 9 and 10 g/dl in 6%, and between 10 and 11 g/dl in 17% of the cases, while for SNC patients they were below 9 g/dl in 9%, between 9 and 10 g/dl in 17%, and between 10 and 11 g/dl in 17% of the cases. Anemia control was worse in patients with stage 5 CKD. For SNC patients and LNC patients hemoglobin concentrations were below 9 g/dl in 21% and 16%, between 9 and 10 g/dl in 28% and 17%, and between 10 and 11 g/dl in 14% and 24% of the cases, respectively.
Hypertriglyceridemia was uncommon in all groups. While only about half of all patients achieved an LDL of less than 100 mg/dl, 64% had an LDL less than 110 mg/dl and 79% had an LDL of less than 130 mg/dl. Total cholesterol levels tended to be lower in patients with Stage 5 CKD, leading to a greater proportion of patients with non-HDL cholesterol at goal.
Blood pressure control was poor in all groups. Stage 1 hypertension [17] was present in 27% and 37% of LNC patients with Stages 4 and 5 CKD, respectively, and in 30% and 52% of SNC patients with Stages 4 and 5 CKD, respectively. Stage 2 hypertension [17] was present in 14% and 28% of LNC patients with Stages 4 and 5 CKD, respectively, and in 35% and 34% of SNC patients with Stages 4 and 5 CKD, respectively. Overall, the prevalence of stage 1 or stage 2 hypertension combined was 82% in SNC and 44% in LNC patients (p < 0.003)
Table 4 shows the percentage of patients with 0–1, 2–3, 4–5, 6–7, and 8–9 of the nine investigated parameters at goal. As became already evident in table 3, our results indicate that LNC patients with Stage 4 CKD who had been seen by a nephrologist for more than 6 months were more likely to achieve goals than SNC patients seen for 6 months or less. In Stage 5 CKD patients, length of nephrology care did not appear to make a big difference. Overall, only a very small fraction of all patients had eight or nine parameters at goal.
Table 4 Proportion of patients achieving a set number of K/DOQI targets
Stage 4 CKD# Stage 5 CKD#
Parameters at goal* SNC† (N = 23) LNC† (N = 70) p-value SNC† (N = 29) LNC† (N = 76) p-value
0–1 0% 0% 0% 5%
2–3 39% 3% 34% 28%
4–5 26% 30% 52% 26%
6–7 35% 54% 14% 36%
8–9 0% 13% 0% 5%
Mean ± SD 4.3 ± 1.8 6.0 ± 1.3 0.001 4.1 ± 1.4 4.6 ± 2.0 0.477
* Number of parameters shown in table 3 that are within the range recommended by K/DOQI. See methods section for individual ranges. Each row shows the proportion of patients within each category achieving the indicated number of parameters at goal. The last row shows the mean number of parameters at goal within each category. P-values were determined by ANOVA with Welch's correction for unequal variances followed by Tamhane's post hoc pairwise comparison of SNC and LNC within each Stage of CKD. # Stage 4 = GFR 15–29, Stage 5 = GFR <15 ml/min/1.73 m2. The glomerular filtration rate was estimated using the modified MDRD formula [1]. † SNC: seen for 6 months or less, LNC: seen for more than 6 months
Predictors of low achievement of K/DOQI goals
We used logistic regression to determine predictors for the failure to achieve K/DOQI goals (Figure 2). The following variables were entered into the model: failure to attend scheduled clinic appointments (no show rate ≥50% versus <50%), having received short nephrology care (≤6 months versus >6 months), CKD Stage 5 (versus CKD Stage 4), African American ethnicity (versus all others), age below 60 years (versus ≥60), female gender (versus male), and having (versus not having) diabetes, hypertension or hyperlipidemia. In addition, all two-way interaction terms were considered; however, no interactions were detected. Short nephrology care (OR = 3.3), failure to attend scheduled clinic appointments (OR = 3.2), African American background (OR = 2.2), diabetes (OR = 2.2), and Stage 5 CKD (OR = 2.2) were significant predictors of low achievement of K/DOQI goals. Age, gender, hyperlipidemia, and hypertension did not reach the level of significance set as p < 0.05.
Figure 2 Logistic regression analysis to identify predictors of poor achievement of K/DOQI Goals. The outcome was defined as having less than five of the nine parameters shown in Table 3 at goal. Shown are the odds ratio (OR), 95% confidence interval (CI), and p-value. All 2-way interaction terms were tested and no interaction was detected. Thus, interaction terms were excluded from the full model. All variables entered are shown. Short nephrology care (SNC) was defined as having been seen by a nephrologist for 6 months or less. African Americans were compared to all other ethnicities. No show ≥ 50% indicates failure to attend scheduled clinic appointments at least half the time. Age was entered after transformation into decades.
Discussion
Dedicated CKD clinics have been established based on the conviction that such clinics will help implement K/DOQI goals and thus improve outcomes [11,18]. Recent data show that multidisciplinary pre-dialysis care is associated with significantly lower mortality after starting dialysis [19]. However, in this study, Curtis et al also report that anemia management only achieved a mean hemoglobin of 102 ± 18 g/l in the multidisciplinary group at the time dialysis was initiated. This is comparable to our results achieved in patients with stage 5 CKD (median: 10.5 mg/dl) and falls short of K/DOQI goals. Similar to our findings, preliminary data from the United Kingdom showed difficulties in achieving K/DOQI goals [20]: in patients with stage 4 and stage 5 CKD, blood pressure goals were achieved in 22.2% and 16.9%, and bicarbonate goals in 41.2% and 29.7%, respectively. Achievement of PTH and hemoglobin goals was better in that study, however, cut-off values were different from the ones recommended by K/DOQI, making a direct comparison difficult. Thus, it appears that although there are proven benefits of having CKD clinics, outcomes might be improved even further if a better achievement of K/DOQI goals could be realized. We found that several factors can be identified that contribute to our failure to reach K/DOQI targets in a larger number of patients.
For most of the parameters investigated, SNC patients with Stage 4 CKD were less likely to achieve goals than LNC patients. Since our referral sources are PCP working either out of the main campus or affiliated satellite clinics, our findings raise the hypothesis that PCP may not be familiar with CKD and its associated co-morbidities. This is most evident in the finding that more than half the patients referred from PCP did not have their PTH level checked, and a quarter had no phosphate determinations on record. The observation that about 25% of the cohort in this study represented SNC patients, i.e., patients referred to the nephrologist for the first time when they already had Stage 4 or 5 CKD, corroborates this impression. The prevalence of CKD is high and with the rising incidence of diabetes it is not expected to decrease [21]. Thus, it will be of vital importance to include the PCP in the care of CKD patients and promote early screening, early initiation of treatment, and timely nephrology referral [22,23]. Educating PCP about early CKD care will be an important step in including them in the care of CKD patients.
We found that the most difficult to manage parameter was PTH; no more than 25% of the patients achieved K/DOQI targets. Similar results were reported by others in dialysis patients despite extensive use of active Vitamin D preparations [24]. A review of our patients' charts revealed that virtually every patient with an elevated PTH was prescribed active Vitamin D, with the exception of those with elevated phosphate levels (above 5.5 mg/dl) or calcium-phosphate products (above 55 mg2/dl2). However, Vitamin D doses used were quite low compared to what is customary in hemodialysis patients (data not shown). Thus, the problems may not to be lack of use, but failure to utilize the correct dose of active Vitamin D preparations, even though an algorithm was in place. The preferred active vitamin D preparation at this institution was oral doxercalciferol. Based on the prevalence and severity of observed hyperparathyroidism in our clinic and the dosage algorithm used, we expected an average dose of 1.49 micrograms daily in patients with stage 4 CKD and 1.75 micrograms daily in patients with stage 5 CKD. The observed dose in patients that did receive active vitamin D was about 25% lower than the expected dose. "Clinical inertia" on the part of clinicians, which has been reported in the care of patients with other chronic diseases such as diabetes, may represent a significant problem [25], as well as concerns about adynamic bone disease, even though this condition is uncommon in the pre-dialysis population [26-29]. In addition, patients referred late by the PCP may already have hyperplasia of the hyperparathyroid gland that is more difficult to treat than the hypertrophy seen earlier in the course of the disease. Finally, we noticed that a number of patients did not understand the difference between over-the-counter vitamin preparations and active Vitamin D. Taken together these data suggest that both physician and patient education are critical components of CKD care. In light of a recent report showing that the combination of high PTH and normal serum calcium and phosphate concentrations was associated with the lowest mortality in prevalent hemodialysis patients [6], prospective studies correlating PTH levels with outcomes in pre-dialysis patients are needed.
Patients with Stage 5 CKD were less likely to achieve K/DOQI targets when compared to patients with Stage 4 CKD. In fact, the benefit of LNC was virtually lost among patients with stage 5 CKD. Among those with Stage 5 CKD, patients requiring renal replacement therapy had the worst results (data not shown). One might argue that the renal team initiates renal replacement therapy too late, i.e., dialysis is only started at a point when medical management becomes impossible. We noted that a number of patients in our clinic are very reluctant to initiate renal replacement therapy in the absence of severe uremic symptoms despite intense counseling. In fact, only half the patients with a GFR of less than 10 ml/min/1.73 m2 started renal replacement therapy. There are no strong data to suggest that an earlier start of dialysis is beneficial [30-33], however, there is data to suggest that treatment goals for complications of CKD such as anemia are more frequently achieved after initiation of hemodialysis [21]. It is possible that this is secondary better adherence since many medications, such as erythropoiesis-stimulating agents, iron, or active vitamin D preparations, are administered in intravenous form during the hemodialysis session. In addition, regular, mandatory clinical quality assurance measures in dialysis units may lead to a stricter adherence to treatment protocols. Thus, in the CKD population, additional vigilance may be needed in patients approaching the need for renal replacement therapy to achieve better outcomes.
Our finding that failure to attend scheduled clinic appointments, younger age, and being African American were all associated with the failure to achieve K/DOQI goals may suggest that cultural or socioeconomic barriers exist. Prior studies investigating access to nephrology care have reported similar findings [34,35]. It is interesting to note that a large proportion of Hispanic patients have a poor command of English, yet being African American conferred a higher risk of not achieving K/DOQI goals, suggesting that language per se may not be a barrier for successful CKD care.
Hyperlipidemia was associated with a better achievement of K/DOQI goals, although statistical significance was not reached. This was surprising since hyperlipidemia has been associated with higher cardiovascular morbidity and mortality, and has been postulated to promote faster progression of renal failure. For this reason, we expected the opposite association. However, since virtually all patients in our cohort carrying the diagnosis of hyperlipidemia are treated with a HMG-CoA reductase inhibitor, we hypothesize that it may be the treatment for, rather than the diagnosis of hyperlipidemia that promotes achievement of K/DOQI goals. Additional research is needed to explore this possibility.
A significant number of variables associated with failure to achieve K/DOQI goals were patient characteristics, such as failure to attend clinic appointments, younger age, and African American ethnicity. This is an important observation, because physician profiling is gaining momentum as a way to improve quality of patient care and adherence to practice guidelines [36]. Our data show that it may be very difficult to judge physicians' performances based on report cards derived from a set of laboratory data that are compared to published practice guidelines, without taking into account the characteristics of the patient population.
Limitations of the study
Our study has several important limitations. Given that this is a single center study, our population has a unique ethnic mix, which may make it difficult to compare our results with what is seen at other centers. However, minorities are among the fastest growing segments in the United States population, and minorities have the highest incidence and prevalence of chronic kidney disease and end-stage renal disease [37]. Thus, it is of vital importance to focus on the health needs of these populations. The population size is small in comparison to some of the published data from managed care providers [38]. This may decrease our power to detect parameters associated with a poorer outcome. However, our study identified important predictors of poor outcomes that can be targeted in future prospective studies. We chose the most recent laboratory value rather than the mean of a number of tests. Thus, patients who had a recent worsening of their status, for example due to hospitalization, may inadvertently have been classified as 'not at goal'. However, averaging several values may have introduced a bias as well, since SNC patients are expected to have a trend towards improvement. Being a retrospective analysis, our data only show associations and do not reveal causalities. Thus, the hypotheses generated by this study will need to be tested prospectively. We have initiated a prospective clinical study to address several of the key points raised by the current work, such as raising the awareness of CKD among referring physicians, frequent quality checks and direct feed back to providers to avoid "clinical inertia", a modified teaching program for patients to improve the patient-provider interaction and enhance empowerment, and intensified case management to address socio-economic barriers to CKD care. It is not known, and this study was not designed to test, whether achieving K/DOQI goals will improve or potentially worsen long-term-outcomes in patients with Stages 4 and 5 CKD, such as cardiovascular morbidity and mortality. Additional research is needed to address this important question. Finally, it is likely that the parameters collected from a given patient are not independent. In addition, multiple interactions exist; for example, the treatment of hyperphosphatemia may influence metabolic acidosis, the treatment of anemia may affect hypertension, and well-nourished patients may have a higher chance to develop hyperphosphatemia. Given the small sample size in our study, we were not able to identify predictors of poor outcomes for each category, such as anemia, while adjusting for all other terms. While it would be clinically important to test these scenarios, this likely will require a multi-center effort.
Conclusion
In summary, we found that K/DOQI goals are achieved in only a small proportion of patients cared for in a dedicated CKD clinic. While numerous publications show that dedicated CKD clinics lead to better outcomes, it appears that there is room for improvement. Raising awareness of CKD and K/DOQI goals among PCP, early referral to a nephrologist, timely initiation of renal replacement therapy, the exploration of socioeconomic barriers and cultural differences, and both continuous patient and physician education are critical to improve CKD care. Prospective clinical trials are needed to explore the impact of these measures on cardiovascular morbidity and mortality in the pre-dialysis arena.
Competing interests
Oliver Lenz has acted as a consultant for Abbott Laboratories, Inc.
Authors' contributions
Oliver Lenz designed the study, obtained IRB approval, collected most data, performed all statistical analyses, and wrote the manuscript. Durga P. Mekala, Daniel V. Patel, Alessia Fornoni, David Metz, and David Roth were major collaborators in study design, data collection, and manuscript preparation. All authors approved the final version of this manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank Dr. Orlando Gomez for assistance with the statistical analyses and careful review of the manuscript. Local IRB approval was obtained for this study. All study procedures were carried out in accordance with the Declaration of Helsinki regarding research involving human subjects.
==== Refs
Levey AS Bosch JP Lewis JB Greene T Rogers N Roth D A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group Ann Intern Med 1999 130 461 470 10075613
Foundation NK K/DOQI Clinical Practice Guidelines for Chronic Kidney Disease: Evaluation, Classification and Stratification Am J Kidney Dis 2002 39 S1 S000 11904577 10.1053/ajkd.2002.32799
Collins AJ Anaemia management prior to dialysis: cardiovascular and cost-benefit observations. Nephrol Dial Transplant Suppl 2003 18 ii2 ii6
Xue JL St Peter WL Ebben JP Everson SE Collins AJ Anemia treatment in the pre-ESRD period and associated mortality in elderly patients Am J Kidney Dis 2002 40 1153 1161 12460033 10.1053/ajkd.2002.36861
Eknoyan G Levin A Levin NW Foundation NK Bone metabolism and disease in chronic kidney disease Am J Kidney Dis Suppl 2003 42 1 201
Stevens LA Djurdjev O Cardew S Cameron EC Levin A Calcium, phosphate, and parathyroid hormone levels in combination and as a function of dialysis duration predict mortality: evidence for the complexity of the association between mineral metabolism and outcomes J Am Soc Nephrol 2004 15 770 779 14978180 10.1097/01.ASN.0000113243.24155.2F
Eustace JA Astor B Muntner PM Ikizler TA Coresh J Prevalence of acidosis and inflammation and their association with low serum albumin in chronic kidney disease Kidney Int 2004 65 1031 1040 14871424 10.1111/j.1523-1755.2004.00481.x
Locatelli F Pozzoni P Tentori F del Vecchio L Epidemiology of cardiovascular risk in patients with chronic kidney disease. Nephrol Dial Transplant Suppl 2003 18 vii2 vii9
Initiative KDOQ K/DOQI clinical practice guidelines on hypertension and antihypertensive agents in chronic kidney disease Am J Kidney Dis Suppl 2004 43 S1 S290
John R Webb M Young A Stevens PE Unreferred chronic kidney disease: a longitudinal study Am J Kidney Dis 2004 43 825 835 15112173 10.1053/j.ajkd.2003.12.046
Ghossein C Serrano A Rammohan M Batlle D The role of comprehensive renal clinic in chronic kidney disease stabilization and management: The Northwestern experience. Semin Nephrol 2002 22 526 532 12430097
Yeoh HH Rasgon S Rutkowski M Crooks P Idroos M Diesto J Management of patients with chronic kidney disease at Kaiser Permanente Nephrol News Issues 2003 17 25 7, 30 12971222
Foundation NK NKF-K/DOQI Clinical Practice Guidelines for Anemia of Chronic Kidney Disease: update 2000 Am J Kidney Dis 2001 37 S182 S238 11229970
Foundation NK K/DOQI Clinical Practice Guidelines for Bone Metabolism and Disease in Chronic Kidney Disease Am J Kidney Dis 2003 42 S1 S202 14520607
Foundation NK K/DOQI Clinical Practice Guidelines on Hypertension and Antihypertensive Agents in Chronic Kidney Disease Am J Kidney Dis 2004 43 S1 S290 15114537
Payne RB Little AJ Williams RB Milner JR Interpretation of serum calcium in patients with abnormal serum proteins Br Med J 1973 4 643 646 4758544
Chobanian AV Bakris GL Black HR Cushman WC Green LA Izzo JLJ Jones DW Materson BJ Oparil S Wright JTJ Roccella EJ Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure Hypertension 2003 42 1206 1252 14656957 10.1161/01.HYP.0000107251.49515.c2
St Peter WL Schoolwerth AC McGowan T McClellan WM Chronic kidney disease: issues and establishing programs and clinics for improved patient outcomes Am J Kidney Dis 2003 41 903 924 12722025 10.1016/S0272-6386(03)00188-4
Curtis BM Ravani P Malberti F Kennett F Taylor PA Djurdjev O Levin A The short- and long-term impact of multi-disciplinary clinics in addition to standard nephrology care on patient outcomes Nephrol Dial Transplant 2005 20 147 154 15585514 10.1093/ndt/gfh585
R.J. M Foley RN New JP O'Donoghue DJ Standard Achievement in Chronic Kidney Disease in CRISIS; Results from Chronic Renal Insufficiency Standards Implementation Study. J Am Soc Nephrol 2004 15 F-PO636
USRDS 2004 Annual Data Report: Atlas of End-Stage Renal Disease in the United States 2005 Bethesda, MD, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases
Israni A Korzelius C Townsend R Mesler D Management of chronic kidney disease in an academic primary care clinic Am J Nephrol 2003 23 47 54 12373080 10.1159/000066296
Ifudu O Friedman EA Managing chronic renal insufficiency--intertwined roles of non-nephrologist and nephrologist J Natl Med Assoc 2003 95 43 54 12656449
Al Aly Z Gonzalez EA Martin KJ Gellens ME Achieving K/DOQI laboratory target values for bone and mineral metabolism: an uphill battle Am J Nephrol 2004 24 422 426 15308874 10.1159/000080087
Phillips LS Branch WT Cook CB Doyle JP El Kebbi IM Gallina DL Miller CD Ziemer DC Barnes CS Clinical inertia Ann Intern Med 2001 135 825 834 11694107
Ho LT Sprague SM Renal osteodystrophy in chronic renal failure Semin Nephrol 2002 22 488 493 12430093 10.1053/snep.2002.31714
Coen G Ballanti P Bonucci E Calabria S Costantini S Ferrannini M Giustini M Giordano R Nicolai G Manni M Sardella D Taggi F Renal osteodystrophy in predialysis and hemodialysis patients: comparison of histologic patterns and diagnostic predictivity of intact PTH Nephron 2002 91 103 111 12021526 10.1159/000057611
Wang M Hercz G Sherrard DJ Maloney NA Segre GV Pei Y Relationship between intact 1-84 parathyroid hormone and bone histomorphometric parameters in dialysis patients without aluminum toxicity Am J Kidney Dis 1995 26 836 844 7485142
Hamdy NA Kanis JA Beneton MN Brown CB Juttmann JR Jordans JG Josse S Meyrier A Lins RL Fairey IT Effect of alfacalcidol on natural course of renal bone disease in mild to moderate renal failure BMJ 1995 310 358 363 7677827
Beddhu S Samore MH Roberts MS Stoddard GJ Ramkumar N Pappas LM Cheung AK Impact of timing of initiation of dialysis on mortality J Am Soc Nephrol 2003 14 2305 2312 12937307 10.1097/01.ASN.0000080184.67406.11
Beddhu S Samore MH Roberts MS Stoddard GJ Pappas LM Cheung AK Creatinine production, nutrition, and glomerular filtration rate estimation J Am Soc Nephrol 2003 14 1000 1005 12660334 10.1097/01.ASN.0000057856.88335.DD
Korevaar JC Jansen MA Dekker FW Jager KJ Boeschoten EW Krediet RT Bossuyt PM Netherlands Cooperative Study on the Adequacy of Dialysis Study Group When to initiate dialysis: effect of proposed US guidelines on survival Lancet 2001 358 1046 1050 11589934 10.1016/S0140-6736(01)06180-3
Lameire N Biesen WV Vanholder R Initiation of dialysis--is the problem solved by NECOSAD? Nephrol Dial Transplant 2002 17 1550 1552 12198203 10.1093/ndt/17.9.1550
Arora P Obrador GT Ruthazer R Kausz AT Meyer KB Jenuleson CS Pereira BJ Prevalence, predictors, and consequences of late nephrology referral at a tertiary care center J Am Soc Nephrol 1999 10 1281 1286 10361866
Cass A Cunningham J Snelling P Wang Z Hoy W Urban disadvantage and delayed nephrology referral in Australia Health Place 2003 9 175 182 12810325 10.1016/S1353-8292(02)00037-0
Mannion R Davies HT Reporting health care performance: learning from the past, prospects for the future J Eval Clin Pract 2002 8 215 228 12180369 10.1046/j.1365-2753.2002.00331.x
Li S McAlpine DD Liu J Li S Collins AJ Differences between blacks and whites in the incidence of end-stage renal disease and associated risk factors Adv Ren Replace Ther 2004 11 5 12 14730534 10.1053/j.arrt.2003.10.005
Keith DS Nichols GA Gullion CM Brown JB Smith DH Longitudinal follow-up and outcomes among a population with chronic kidney disease in a large managed care organization Ann Intern Med 2004 164 659 663 10.1001/archinte.164.6.659
|
16250919
|
PMC1283975
|
CC BY
|
2021-01-04 16:39:31
|
no
|
BMC Nephrol. 2005 Oct 27; 6:11
|
utf-8
|
BMC Nephrol
| 2,005 |
10.1186/1471-2369-6-11
|
oa_comm
|
==== Front
BMC Pulm MedBMC Pulmonary Medicine1471-2466BioMed Central London 1471-2466-5-131621612810.1186/1471-2466-5-13Research ArticleHepatocyte and keratinocyte growth factors and their receptors in human lung emphysema Bonay Marcel [email protected] Anne [email protected]çon-Malas Véronique [email protected] Joëlle [email protected] Paul [email protected] Michel [email protected] Guy [email protected] Monique [email protected] Bruno [email protected] INSERM U 700, Faculté Xavier Bichat, Paris, France2 Service de Physiologie-Explorations Fonctionnelles, Hôpital Bichat-Claude Bernard, Assistance Publique-Hôpitaux de Paris, Université Paris 7, Paris, France3 Service de Biochimie A, Hôpital Bichat-Claude Bernard, Assistance Publique-Hôpitaux de Paris, Université Paris 7, Paris, France4 Service de Pneumologie Hôpital Bichat-Claude Bernard, Assistance Publique-Hôpitaux de Paris, Université Paris 7, Paris, France5 Service de Pneumologie, Hôpital Beaujon, Assistance Publique-Hôpitaux de Paris, Université Paris 7, Paris, France6 Service de Chirurgie Thoracique, Hôpital Beaujon, Assistance Publique-Hôpitaux de Paris, Université Paris 7, Paris, France2005 10 10 2005 5 13 13 6 7 2005 10 10 2005 Copyright © 2005 Bonay et al; licensee BioMed Central Ltd.2005Bonay 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
Hepatocyte and keratinocyte growth factors are key growth factors in the process of alveolar repair. We hypothesized that excessive alveolar destruction observed in lung emphysema involves impaired expression of hepatocyte and keratinocyte growth factors or their respective receptors, c-met and keratinocyte growth factor receptor. The aim of our study was to compare the expression of hepatocyte and keratinocyte growth factors and their receptors in lung samples from 3 groups of patients: emphysema; smokers without emphysema and non-smokers without emphysema.
Methods
Hepatocyte and keratinocyte growth factor proteins were analysed by immunoassay and western blot; mRNA expression was measured by real time quantitative polymerase chain reaction.
Results
Hepatocyte and keratinocyte growth factors, c-met and keratinocyte growth factor receptor mRNA levels were similar in emphysema and non-emphysema patients. Hepatocyte growth factor mRNA correlated negatively with FEV1 and the FEV1/FVC ratio both in emphysema patients and in smokers with or without emphysema. Hepatocyte and keratinocyte growth factor protein concentrations were similar in all patients' groups.
Conclusion
The expression of hepatocyte and keratinocyte growth factors and their receptors is preserved in patients with lung emphysema as compared to patients without emphysema. Hepatocyte growth factor mRNA correlates with the severity of airflow obstruction in smokers.
==== Body
Background
Emphysema is one of the most important cause of respiratory insufficiency with increasing mortality and morbidity [1]. Lung emphysema is defined pathologically by the destruction of alveolar walls with abnormal permanent enlargement of the air spaces distal to the terminal bronchiole. Precise mechanims contributing to the destruction of alveolar walls remain unclear. Cigarette smoking may induce emphysema by stimulating neutrophils and alveolar macrophages to produce proteases leading to the degradation of the alveolar extracellular matrix and oxidant injury contributing to alveolar destruction [2]. T lymphocytes contribute to the recruitment and activation of these inflammatory cells and may be involved in the apoptosis and destruction of alveolar epithelium [3-5]. Recent evidence of increased apoptosis of alveolar epithelial and endothelial cells in emphysematous lung suggests that primary alterations of the alveolar epithelium and endothelium might participate in the pathogenesis of the disease [6-8].
The alveolar epithelium is essential for maintenance of the integrity of the alveolar spaces. Functional restoration of the alveolar epithelium after an injury requires the proliferation and migration of type 2 pneumocytes and their differentiation into type 1 pneumocytes, a tightly regulated phenomenon. Growth factors have been shown to control both phases of the process [9-12]. Among these factors, the hepatocyte growth factor (HGF), a heterodimeric protein obtained through the cleavage of an inactive precursor, called 90-kD proHGF, and the keratinocyte growth factor (KGF, also named fibroblast growth factor-7, FGF-7) have been identified as key factors in the process of alveolar repair, both in acute or chronic conditions [13-15]. HGF and KGF respectively act through specific receptors c-met, a membrane bound tyrosine kinase [16] and FGFR2IIIb, also named KGF-R [17]. HGF and KGF productions by lung fibroblasts from emphysema have been shown to be reduced when compared with controls [18]. Recently, Shigemura et al reported that decreased HGF expression due to a failure in sustained endogenous production after injury was associated with emphysema-related histopathologic and physiological changes in a rat model of elastase-induced emphysema [19]. In this animal model, HGF could have a therapeutic effect [20,19].
We hypothesized that a defective expression of growth factors involved in human alveolar epithelium repair such as HGF and KGF or their specific receptors might participate in the pathophysiology of lung emphysema. We therefore evaluated the expression of HGF, KGF, and their receptors c-met and KGF-R in lung biopsies from patients with emphysema and from non-emphysema patients, according to their smoking status.
Methods
This study was approved by the ethics committee of Saint Germain-en-Laye hospital.
Patients
Lung samples were obtained during surgery in adult patients (>18 years) from Beaujon university hospital (Clichy, France) and from Bichat university hospital (Paris, France).
Patients with lung emphysema
Seventeen patients with radiographically defined emphysema (E group) were included. All patients were active or ex-smokers (Table 1). CT-scan, pulmonary function tests and α1-AT deficiency were systematically documented. Patients with α1-AT deficiency were excluded. They underwent bullectomy (n = 3), lobectomy (n = 4), lung transplantation (n = 3) or lung volume reduction (n = 7). Pulmonary function tests demonstrated mild to severe airflow obstruction and lung distension (Table 2). Tissue samples were taken from the resected parenchyma in a macroscopically emphysematous region. Nine patients with emphysema were receiving corticosteroids, either oral (n = 2) and/or inhaled (n = 8). In all patients, lung emphysema was suspected on CT-scan and confirmed by the pathological examination of lung resection samples. The severity of emphysema was approached through pulmonary function abnormalities.
Table 1 Clinical characteristics of patients with and without emphysema
Emphysema (E) Non-emphysema Smokers (S) Non-emphysema Non-Smokers (NS) Between groups differences
n 17 8 10
Age (yr) 60 ± 7.4 55.8 ± 11.2 52.7 ± 19.4 ns
Sex ratio F/M 1/16 0/8 2/8 ns
Smoking (Pack.yr) 51 ± 21.4* 33.3 ± 14.8 0 p < 0.001
Time since smoking cessation (yearrs) 7.0 ± 7.1 10 (n = 1) Non relevant Non relevant
Active smokers/ex-smokers 7/10 7/1 0/0 p < 0.001
*: vs S, p < 0.05; as assessed by Mann-Whitney U-test. ns: non significant.
Table 2 Pulmonary function tests of patients with and without emphysema.
Emphysema (E) Non-emphysema Smokers (S) Non-emphysema Non-Smokers (NS) Between groups differences
n 17 8 10
FEV1 % predicted value (% pred) 41.7 ± 26.6 *,† 71 ± 21.3 82 ± 11.1 p < 0.001
FEV1/FVC 48.7 ± 13.9 *,† 67.6 ± 14.8† 86.9 ± 11.6 p < 0.001
RV (%pred) 211.6 ± 69.4 *,† 123 ± 19.7 88 ± 23.9 p < 0.001
TLC (%pred) 121.5 ± 17.8 *,† 100 ± 10.4 † 83 ± 13.5 p < 0.001
PaO2 (mmHg) 69.7 ± 10.1 *,† 81 ± 15.8 85.2 ± 7.6 p < 0.005
PaCO2 (mmHg) 41.3 ± 5 38.7 ± 7.7 41.8 ± 2.6 ns
*: vs S, p < 0.01; †: vs NS, p < 0.01; as assessed by Mann-Whitney U-test. ns: non significant
PaO2: arterial oxygen pressure, PaCO2: arterial carbon dioxide pressure, FEV1: forced expiratory volume in one second, RV: residual volume, FVC: forced vital capacity, TLC: total lung capacity.
Non-emphysema patients
Normal tissue was obtained from 18 non-emphysema patients. Eight patients were smokers (non-emphysema smoker, S) and 10 were non-smokers (non-emphysema non-smoker, NS). S patients were undergoing surgery for the resection of a localised primary lung carcinoma (n = 7) or a benign lesion (n = 1). NS patients were undergoing surgery for the resection of a localised primary lung carcinoma (n = 3), lung metastases (n = 2) or a benign lesion (n = 5). Tissue samples were taken at a site distant from the pathological process and without macroscopical and microscopical evidence of emphysema. S patients had mild to moderate alterations of pulmonary functions tests (table 2). Increased cumulative tobacco exposure was observed in the E group as compared with the S group (Table 1). No difference was observed between groups for age and sex ratio (Table 1). Two patients without emphysema (S group) received inhaled corticosteroids.
Processing of lung samples
Lung tissue fragments (about 0.2 cm3) were immediately frozen in liquid N2 and stored at -80°C until RNA and protein analysis. The histopathology of biopsies was evaluated on paraffin-embedded sections to verify the features of emphysema or normal lungs. The concentration of proteins in biopsies was evaluated from 100 mg of lung samples homogenised with 0.5 ml PBS containing 200 μM phenylmethylsulfonyl fluoride, 1 mg/ml leupeptin and 1 mg/ml aprotinin. The homogenates were centrifuged at 10,000 g for 10 min at 4°C to remove tissue fragments and the supernatants were collected and stored at -80°C until measurement.
Quantitative analysis of mRNA expression
Total RNA was extracted from frozen lung tissue and reverse transcribed. Each sample was analysed by reverse transcriptase-real-time polymerase chain reaction (RT-PCR) with specific primers (table 3) to quantify the expression of mRNA of HGF, KGF, c-met and KGF-R as described previously [21].
Table 3 Primers and PCR cycling conditions.
Primers Sequences Denaturation annealing Cycles PCR products
HGF: Forward
Reverse 5'-CAGAGGGACAAAGGAAAAGAA-3'
5'-GCAAGTGAATGGAAGTCCTTTA-3' 94°C, 15s 58°C, 60s 40 167 bp
KGF: Forward
Reverse 5'-GAACAAGGAAGGAAAACTCTATGCAA-3'
5'-AAGTGGGCTGTTTTTTGTTCTTTCT-3' 94°C, 15s 60°C, 60s 40 201 bp
HGF-R: Forward
Reverse 5'-GTTTACTTGTTGCAAGGGAGAAGACT-3'
5'-TAGGGTGCCAGCATTTTAGCA-3' 94°C, 15s 58°C, 60s 40 88 bp
KGF-R: Forward
Reverse 5'-TTAAGCAGGAGCATCGCATTG-3'
5'-AACATCCAGGTGGTACGTGTGAT-3' 94°C, 15s 60°C, 60s 40 151 bp
Ubiquitin-c: Forward
Reverse 5'-CACTTGGTCCTGCGCTTGA-3'
5'-TTTTTTGGGAATGCAACAACTTT-3' 94°C, 15s 60°C, 60s 40 105 bp
HGF and KGF concentration in lung homogenates
The proteins HGF and KGF (Quantikine®, R&D Systems; respective detection limits were 40 and 15 pg/ml for HGF, KGF) were measured in lung homogenates from 26 out of 35 patients when enough lung sample was available (13 E, 5 S, 8 NS).
HGF western blotting
Lung homogenates from 4 patients (2 patients with emphysema and 2 patients without emphysema) were examined by Western blotting as previously described [22].
Statistical analysis
Data were analysed by Statview software (Abacus Concepts, Inc.) and displayed as mean ± SD. Between-group differences were first assessed by non-parametric analysis of variance (Kruskal-Wallis test). In the case of global significant difference, between two groups comparisons were assessed by the non-parametric Mann-Whitney U-test. Correlations were assessed by the Spearman's rank order test. Categorical data were analysed using the Chi-squared test. A p value < 0.05 was regarded as significant.
Results
Lung expression of KGF and KGF receptor
KGF mRNA (figure 1A) and KGF protein (figure 1C) were detected in the lungs of all patients. No difference was observed according to the presence of emphysema, and the smoking status.
Figure 1 Expression of KGF (A) and KGF-R (B) mRNA in lung tissue from emphysema and non-emphysema patients. Results are expressed as a ratio to Ubiquitin in arbitrary units. Individual and mean values (bar) are presented. No difference between groups was found. E: emphysema patients; S and NS: non-emphysematous smoker and non-smoker patients. KGF concentrations in lung homogenates were measured by immunoassay (C). The detection limit was 15 pg/ml. The results are displayed per 1 mg of lung tissue.
KGF-R mRNA (figure 1B) was detected in the lungs of all patients. A considerable variability was observed in KGF-R transcript levels in patients with or without emphysema. No difference was observed according to the presence of emphysema and the smoking status.
KGF mRNA, KGF protein and KGF-R mRNA did not correlate with age, cumulative tobacco exposure, period since smoking cessation, use of inhaled corticosteroids, pulmonary function tests and arterial blood gases.
Lung expression of HGF and HGF receptor
Although a considerable variability of HGF mRNA expression was observed, no difference was found between emphysema and non-emphysema groups (figure 2A). In emphysema patients, HGF mRNA correlated positively with total lung capacity (TLC) (Rho = 0.51, p = 0.04, n = 17) and negatively with forced expiratory volume in one second (FEV1) (Rho = -0.53, p = 0.03, n = 17) and FEV1/FVC (forced vital capacity) (Rho = -0.54, p = 0.03, n = 17). When all smokers were studied together (smokers with or without emphysema), again, significant correlations between HGF mRNA and FEV1 (Rho = -0.53, p = 0.009, n = 25), and FEV1/FVC (Rho = -0.49, p = 0.017, n = 25) were found. However, the correlation between HGF mRNA and TLC was no more significant (Rho = 0.28, p = 0.18, n = 24) (figure 3). There was no correlation between HGF mRNA and cumulative tobacco exposure.
Figure 2 Expression of HGF (A) and c-met (B) mRNA in lung tissue from emphysema and non-emphysema patients. Results are expressed as a ratio to Ubiquitin in arbitrary units. Individual and mean values (bar) are presented. No difference between groups was found. E: emphysema patients; S and NS: non-emphysematous smoker and non-smoker patients. HGF concentrations in lung homogenates were measured by immunoassay (C). The detection limit was 40 pg/ml. The results are displayed per 1 mg of lung tissue.
Figure 3 Correlation between HGF mRNA and pulmonary function in smoker patients. The ratio of lung HGF to Ubiquitin mRNA was correlated with: (A) forced expiratory volume in one second (FEV1 % predicted), (B) FEV1/FVC (forced vital capacity), but not with (C) total lung capacity (TLC % predicted). Full circle (•): emphysema patients; open circle (o): smoker patients without emphysema.
HGF protein was detected in lung homogenates from all patients as assessed by immunoassay. HGF concentration was not different between groups (figure 2C). HGF protein correlated positively with residual volume (RV) (Rho = 0.43, p = 0.04, n = 24), TLC (Rho = 0.45, p = 0.03, n = 24) and negatively with FEV1/FVC (Rho = -0.45, p = 0.03, n = 24). Although, no difference of HGF protein concentrations was found between patients' groups, a significant correlation between HGF protein concentrations and cumulative tobacco exposure was observed (Rho = 0.49, p = 0.01, n = 26). As HGF immunoassay measured both proHGF and mature HGF, a western blot analysis was performed to characterize which form of HGF was present in lung homogenates. Western blot (figure 4) demonstrated that HGF was present mainly in the cleaved mature form (presence of the 69-kD alpha chain) both in the non-emphysematous and the emphysematous lungs.
Figure 4 Western blot analysis of HGF in lung tissue. Lung homogenates from 4 patients (2 patients with emphysema [E] and 2 non-emphysema patients [NonE]). Recombinant human HGF (R&D Systems: according to the manufacturer, rhHGF is a mixture of proHGF and cleaved mature HGF) was loaded and blotted in parallel. HGF was in the cleaved mature form as evidenced by the detection of the 69-kD α chain and the absence of the 90-kD proHGF form.
HGF-R mRNA was detected in all patients. We found no difference between groups (figure 2B). Strong correlations were observed between HGF-R mRNA and KGF-R mRNA (Rho = 0.82, p < 0.0001, n = 35) and between HGF mRNA and KGF mRNA (Rho = 0.61, p = 0.004, n = 35) when all patients were taken together.
Discussion
The involvement of KGF and HGF in lung repair has been widely documented. Numerous studies in vitro and in vivo have demonstrated that KGF and HGF have protective effects in experimental lung injury [15,23]. To our knowledge, this is the first study of KGF and HGF lung expression in human emphysema. Proteases, oxidant injury [2], chronic inflammation [3,5] and apoptosis [6-8] all contribute to the excessive alveolar wall destruction observed in lung emphysema. We hypothezised that a defect of the lung repair process might be associated with the pathophysiology of lung emphysema. Our results show that lung KGF mRNA and KGF protein are not altered in emphysema. In bleomycin-induced lung injury in rats, KGF and HGF increase in the lung [24]. A few clinical studies have assessed KGF concentrations in acute lung injury. Verghese et al reported that KGF was not increased in lung edema fluid whereas HGF was increased and associated with higher mortality [25]. Stern et al reported that KGF and HGF were increased in bronchoalveolar lavage fluid in acute respiratory distress syndrome and associated with a poor prognosis [13]. Recently, Danan et al observed the highest KGF concentrations in tracheal aspirates from premature infants who survived without bronchopulmonary dysplasia, leading to the conclusion that KGF may prevent injury to lung epithelium and enhanceits repair [26].
This study has some methodological limitations. A limited number of patients was studied in each group, especially in non emphysema groups which were mostly composed of lung biopsies obtained at a site distant from localised carcinoma. Furthermore, the patients could only be evaluated at one time point in the course of their disease. Inclusion of smokers without emphysema allowed the differentiation of emphysema-related and tobacco-related events. Because only one tissue sample from surgically resected material was available for examination, the expression of HGF, KGF and their receptors reflects regional disease activity and may be unrepresentative of the entire lung. Indeed, it is well known that emphysema affects different lung regions to a varying extent. Moreover, we evaluated HGF and KGF in lung homogenates only. Future studies should address the expression of HGF and KGF in a more cell-specific fashion.
In our study, although HGF mRNA lung expression was similar in emphysema and non emphysema patients, a correlation was found between HGF mRNA and the deterioration of pulmonary function tests in emphysema patients. The correlation between airflow obstruction and HGF mRNA level was similarly observed when all smokers with or without emphysema were studied, suggesting that emphysema was not a main determinant of HGF mRNA level in the lung. This strong correlation between airflow obstruction and HGF mRNA in smokers suggests that the increase of HGF mRNA was not related to the presence of emphysema but rather to the degree of airflow obstruction. This observation is supported by the correlation between HGF protein in lung homogenates and the FEV1/FVC ratio in our population. These results are in agreement with the observations of Sauleda et al, who reported that HGF protein concentrations were increased in broncho-alveolar lavage of patients with chronic obstructive pulmonary disease as compared to smokers and non-smoker controls [27]. Interestingly, the increased lung expression of other growth factors (fibroblast growth factors 1 and 2 and their receptors) has already been reported in chronic obstructive pulmonary disease [28].
The mechanisms underlying the correlation between airflow obstruction and HGF mRNA in smokers are unclear. Although speculative, we can propose that the mechanical constraints applied to alveolar tissue secondary to airflow obstruction may stimulate HGF production by alveolar epithelial cells, since Yamamoto et al showed that mechanical stretch induced HGF in alveolar type II cells in vitro [29]. Furthermore, airway inflammation could contribute to increase local HGF expression by neutrophils [22] and macrophages [30]. Interestingly, Aharinejad et al have shown that serum HGF concentrations increased at the time of lung graft rejection, a situation associated with airflow obstruction [31].
As HGF and KGF are key factors in the process of alveolar repair [15], we suggest that their production might be not adapted to the degree of alveolar injury. Indeed, in view of the chronic lung injury observed in emphysema, one could expect an increase of HGF and KGF expression as observed in acute lung injury in rats [24] and in humans [25,13]. However the direct assessment of HGF content in lung homogenates is technically difficult. Indeed, HGF is a heparin binding growth factor. High concentrations of inactive precursor of HGF (proHGF) are probably bound to proteoglycans of the extracellular matrix and may not be assayed in the lung homogenates by immunoassay. In this study, western blot analysis showed that HGF was only in the mature active form, both in lung biopsies from emphysema and non-emphysema patients.
Recently, HGF has been shown to stimulate pulmonary regeneration and to improve pulmonary function in animal models of elastase-induced lung emphysema [20,19]. Preserved expression of HGF-R could allow therapeutic use of growth factors in lung emphysema. Further studies are needed to assess the therapeutic potential of HGF and KGF in lung emphysema.
Conclusion
The main results of our study are that: i) KGF and HGF lung expression is preserved in emphysema patients, ii) HGF-R and KGF-R mRNA are consistently expressed in the lung of emphysema patients and are not modified by the smoking status, iii) HGF mRNA correlates with the severity of airflow obstruction in smokers.
List of abbreviations
α1-AT: α1-antitrypsin
E: emphysema
FEV1: forced expiratory volume in one second
FGF: fibroblast growth factor
FVC: forced vital capacity
HGF: hepatocyte growth factor
KGF: keratinocyte growth factor
mRNA: messenger ribonucleic acid
NS: non-smoker without emphysema
RT-PCR: reverse transcriptase-real-time polymerase chain reaction
RV: residual volume
S: smoker without emphysema
SD: standard deviation
TLC: total lung capacity
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MB and AB equally participated in the design of the study, conducted the majority of the research experiments and drafted the manuscript.
VL participated in the majority of the research experiments.
MF participated in the design of the study.
JM and PS conducted some experiments.
GL participated in the design of the study.
MD and BC conceived of the study, participated in its design, and in drafting the manuscript.
All authors read and approved the final version of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Part of this work was supported by the legs Poix, Chancellerie des Universités de Paris.
The authors thank Anne Barnier for helpful technical assistance, and Joanna Shore for helpful criticism of the manuscript.
==== Refs
Barnes PJ Chronic obstructive pulmonary disease N Engl J Med 2000 343 269 280 10911010 10.1056/NEJM200007273430407
Shapiro SD The macrophage in chronic obstructive pulmonary disease Am J Respir Crit Care Med 1999 160 S29 32 10556166
Finkelstein R Fraser RS Ghezzo H Cosio MG Alveolar inflammation and its relation to emphysema in smokers Am J Respir Crit Care Med 1995 152 1666 1672 7582312
Retamales I Elliott WM Meshi B Coxson HO Pare PD Sciurba FC Rogers RM Hayashi S Hogg JC Amplification of inflammation in emphysema and its association with latent adenoviral infection Am J Respir Crit Care Med 2001 164 469 473 11500352
Turato G Zuin R Miniati M Baraldo S Rea F Beghe B Monti S Formichi B Boschetto P Harari S Papi A Maestrelli P Fabbri LM Saetta M Airway inflammation in severe chronic obstructive pulmonary disease: relationship with lung function and radiologic emphysema Am J Respir Crit Care Med 2002 166 105 110 12091179 10.1164/rccm.2111084
Kasahara Y Tuder RM Cool CD Lynch DA Flores SC Voelkel NF Endothelial cell death and decreased expression of vascular endothelial growth factor and vascular endothelial growth factor receptor 2 in emphysema Am J Respir Crit Care Med 2001 163 737 744 11254533
Kasahara Y Tuder RM Taraseviciene-Stewart L Le Cras TD Abman S Hirth PK Waltenberger J Voelkel NF Inhibition of VEGF receptors causes lung cell apoptosis and emphysema J Clin Invest 2000 106 1311 1319 11104784
Aoshiba K Yokohori N Nagai A Alveolar wall apoptosis causes lung destruction and emphysematous changes Am J Respir Cell Mol Biol 2003 28 555 562 12707011 10.1165/rcmb.2002-0090OC
Kim HJ Sammak PJ Ingbar DH Hepatocyte growth factor stimulates migration of type II alveolar epithelial cells on the provisional matrix proteins fibronectin and fibrinogen Chest 1999 116 94S 95S 10424614 10.1378/chest.116.suppl_1.94S
Mason RJ Leslie CC McCormick-Shannon K Deterding RR Nakamura T Rubin JS Shannon JM Hepatocyte growth factor is a growth factor for rat alveolar type II cells Am J Respir Cell Mol Biol 1994 11 561 567 7524567
Ohmichi H Matsumoto K Nakamura T In vivo mitogenic action of HGF on lung epithelial cells: pulmotrophic role in lung regeneration Am J Physiol 1996 270 L1031 9 8764230
Panos RJ Patel R Bak PM Intratracheal administration of hepatocyte growth factor/scatter factor stimulates rat alveolar type II cell proliferation in vivo Am J Respir Cell Mol Biol 1996 15 574 581 8918364
Ware LB Matthay MA Keratinocyte and hepatocyte growth factors in the lung: roles in lung development, inflammation, and repair Am J Physiol Lung Cell Mol Physiol 2002 282 L924 40 11943656
Stern JB Fierobe L Paugam C Rolland C Dehoux M Petiet A Dombret MC Mantz J Aubier M Crestani B Keratinocyte growth factor and hepatocyte growth factor in bronchoalveolar lavage fluid in acute respiratory distress syndrome patients Crit Care Med 2000 28 2326 2333 10921560 10.1097/00003246-200007000-00024
Crestani B Dehoux M Hayem G Lecon V Hochedez F Marchal J Jaffre S Stern JB Durand G Valeyre D Fournier M Aubier M Differential role of neutrophils and alveolar macrophages in hepatocyte growth factor production in pulmonary fibrosis Lab Invest 2002 82 1015 1022 12177240
Weidner KM Sachs M Birchmeier W The Met receptor tyrosine kinase transduces motility, proliferation, and morphogenic signals of scatter factor/hepatocyte growth factor in epithelial cells J Cell Biol 1993 121 145 154 8384622 10.1083/jcb.121.1.145
Dionne CA Crumley G Bellot F Kaplow JM Searfoss G Ruta M Burgess WH Jaye M Schlessinger J Cloning and expression of two distinct high-affinity receptors cross-reacting with acidic and basic fibroblast growth factors Embo J 1990 9 2685 2692 1697263
Plantier L Marchand-Adam S Marchal-Somme J Leseche G Fournier M Dehoux M Aubier M Crestani B Defect of hepatocyte growth factor production by fibroblasts in human pulmonary emphysema Am J Physiol Lung Cell Mol Physiol 2005 288 L641 7 15579628 10.1152/ajplung.00249.2004
Shigemura N Sawa Y Mizuno S Ono M Ohta M Nakamura T Kaneda Y Matsuda H Amelioration of pulmonary emphysema by in vivo gene transfection with hepatocyte growth factor in rats Circulation 2005 111 1407 1414 15781752 10.1161/01.CIR.0000158433.89103.85
Ishizawa K Kubo H Yamada M Kobayashi S Suzuki T Mizuno S Nakamura T Sasaki H Hepatocyte growth factor induces angiogenesis in injured lungs through mobilizing endothelial progenitor cells Biochem Biophys Res Commun 2004 324 276 280 15465014 10.1016/j.bbrc.2004.09.049
Marchand-Adam S Marchal J Cohen M Soler P Gerard B Castier Y Leseche G Valeyre D Mal H Aubier M Dehoux M Crestani B Defect of hepatocyte growth factor secretion by fibroblasts in idiopathic pulmonary fibrosis Am J Respir Crit Care Med 2003 168 1156 1161 12947024 10.1164/rccm.200212-1514OC
Grenier A Chollet-Martin S Crestani B Delarche C El Benna J Boutten A Andrieu V Durand G Gougerot-Pocidalo MA Aubier M Dehoux M Presence of a mobilizable intracellular pool of hepatocyte growth factor in human polymorphonuclear neutrophils Blood 2002 99 2997 3004 11929792 10.1182/blood.V99.8.2997
Ray P Devaux Y Stolz DB Yarlagadda M Watkins SC Lu Y Chen L Yang XF Ray A Inducible expression of keratinocyte growth factor (KGF) in mice inhibits lung epithelial cell death induced by hyperoxia Proc Natl Acad Sci U S A 2003 100 6098 6103 12732722 10.1073/pnas.1031851100
Adamson IY Bakowska J Relationship of keratinocyte growth factor and hepatocyte growth factor levels in rat lung lavage fluid to epithelial cell regeneration after bleomycin Am J Pathol 1999 155 949 954 10487852
Verghese GM McCormick-Shannon K Mason RJ Matthay MA Hepatocyte growth factor and keratinocyte growth factor in the pulmonary edema fluid of patients with acute lung injury. Biologic and clinical significance Am J Respir Crit Care Med 1998 158 386 394 9700111
Danan C Franco ML Jarreau PH Dassieu G Chailley-Heu B Bourbon J Delacourt C High concentrations of keratinocyte growth factor in airways of premature infants predicted absence of bronchopulmonary dysplasia Am J Respir Crit Care Med 2002 165 1384 1387 12016100 10.1164/rccm.200112-134BC
Sauleda J Noguera A Blanquer D Fuster A Pons AR Pons J Agusti A Growth factors in chronic obstructive pulmonary disease [abstract] Eur Respir J 2004 24 (suppl) 320s
Kranenburg AR De Boer WI Van Krieken JH Mooi WJ Walters JE Saxena PR Sterk PJ Sharma HS Enhanced expression of fibroblast growth factors and receptor FGFR-1 during vascular remodeling in chronic obstructive pulmonary disease Am J Respir Cell Mol Biol 2002 27 517 525 12397010
Yamamoto H Teramoto H Uetani K Igawa K Shimizu E Stretch induces a growth factor in alveolar cells via protein kinase Respir Physiol 2001 127 105 111 11504583 10.1016/S0034-5687(01)00244-4
Morimoto K Amano H Sonoda F Baba M Senba M Yoshimine H Yamamoto H Ii T Oishi K Nagatake T Alveolar macrophages that phagocytose apoptotic neutrophils produce hepatocyte growth factor during bacterial pneumonia in mice Am J Respir Cell Mol Biol 2001 24 608 615 11350831
Aharinejad S Taghavi S Klepetko W Abraham D Prediction of lung-transplant rejection by hepatocyte growth factor Lancet 2004 363 1503 1508 15135596 10.1016/S0140-6736(04)16148-5
|
16216128
|
PMC1283976
|
CC BY
|
2021-01-04 16:30:12
|
no
|
BMC Pulm Med. 2005 Oct 10; 5:13
|
utf-8
|
BMC Pulm Med
| 2,005 |
10.1186/1471-2466-5-13
|
oa_comm
|
==== Front
BMC Struct BiolBMC Structural Biology1472-6807BioMed Central London 1472-6807-5-201624203110.1186/1472-6807-5-20Research ArticleAsymmetry in the burial of hydrophobic residues along the histone chains of Eukarya, Archaea and a transcription factor Silverman B David [email protected] IBM Thomas J. Watson Research Center P. O. Box 218, Yorktown Heights, NY 10598 USA2005 21 10 2005 5 20 20 18 7 2005 21 10 2005 Copyright © 2005 Silverman; 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 histone fold is a common structural motif of proteins involved in the chromatin packaging of DNA and in transcription regulation. This single chain fold is stabilized by either homo- or hetero-dimer formation in archaea and eukarya. X-ray structures at atomic resolution have shown the eukaryotic nucleosome core particle to consist of a central tetramer of two bound H3-H4 dimers flanked by two H2A-H2B dimers. The c-terminal region of the H3 histone fold involved in coupling the two eukaryotic dimers of the tetramer, through a four-fold helical bundle, had previously been shown to be a region of reduced burial of hydrophobic residues within the dimers, and thereby provide a rationale for the observed reduced stability of the H3-H4 dimer compared with that of the H2A-H2B dimer. Furthermore, comparison between eukaryal and archaeal histones had suggested that this asymmetry in the distribution of hydrophobic residues along the H3 histone chains could be due to selective evolution that enhanced the coupling between the eukaryotic dimers of the tetramer.
Results and discussion
The present work describes calculations utilizing the X-ray structures at atomic resolution of a hyperthermophile from Methanopyrus kandleri (HMk) and a eukaryotic transcription factor from Drosophila melanogaster (DRm), that are structurally homologous to the eukaryotic (H3-H4)2 tetramer. The results for several other related structures are also described. Reduced burial of hydrophobic residues, at the homologous H3 c-terminal regions of these structures, is found to parallel the burial at the c-terminal regions of the H3 histones and is, thereby, expected to affect dimer stability and the processes involving histone structural rearrangement. Significantly different sequence homology between the two histones of the HMk doublet with other archaeal sequences is observed, and how this might have occurred during selection to enhance tetramer stability is described.
==== Body
Background
The histone fold [1] is a commonly conserved structural motif found in proteins that interact with DNA [2-6]. This monomeric fold consists of a helix-strand-helix motif stabilized by dimer formation. X-ray structures at atomic resolution, in either the absence [1] or presence of DNA [7], have shown the octamer of the eukaryotic nucleosome to be composed of histone dimers in a "handshake conformation" [1], with the H3-H4 dimers coupled forming a central tetramer flanked by two H2A-H2B dimers.
The dimers of the tetramer are bound by a tenuous four-fold helical bundle that involves residues over the c-terminal regions of the H3 histone folds. Consequently, side chain orientations in these regions while contributing to tetramer stability do not contribute to the stability of the dimers in the tetramer conformation by optimally burying residues within the dimer interiors. Such orientations, if comparably maintained in the isolated dimers would, therefore, reduce dimer stability relative to that of dimers that are not required to perform such dual role of hydrophobic burial, namely, burial with respect to the interface between dimers as well as with respect to the dimer interiors. This has been proposed [8] as a rationale for the observed [9] reduced stability of the H3-H4 dimer compared with that of the H2A-H2B dimer.
X-ray structures at atomic resolution of an ancestral nuclear protein histone of the hyperthermophile, HMk, from Methanopyrus kandleri [10] and of the amino-terminal portions of the TFIID transcription factor TATA box-binding associated factors (TAFIIs), dTAFII42 and dTAFII62, from Drosophila, DRm [11], have been found to be structurally homologous to the eukaryotic (H3-H4)2 tetramer. Since the unusual fold of the hyperthermophile contains two histone fold motifs arranged in tandem within a single chain [10,12], its structure, homologous to the eukaryotic tetramer, is a dimer. The coupling of the HMk monomers and TFIID dimers that yield the higher-order dimeric and tetrameric structures, respectively, involves a four-fold helical bundle as does the (H3-H4)2 tetramer. Since the c-terminal region of the H3 histone fold involved in coupling the two eukaryotic nucleosomal dimers of the tetramer through the four-fold helical bundle had previously been shown [8] to be a region of reduced burial of hydrophobic residues within the H3-H4 dimers, it is of interest to investigate the burial of hydrophobic residues over the homologous c-terminal regions of the HMk monomer and DRm dimer. Reduced stability of these lower-order structures, due to the burial of hydrophobic residues within the four-fold helical bundles, holds implications for dynamical processes involving DNA as well for nucleosome assembly.
Calculations have, therefore, been performed for structures of the HMk monomer and DRm dimer, as performed previously for archaeal methanogens and eukaryotic nucleosomal dimers [8]. An interesting pattern emerges. All chains involved in the coupling of the lower-order structures, e.g. the HMk monomers and DRm dimers that yield dimers and tetramers, respectively, exhibit a difference in the burial of hydrophobic residues along their length. Reduced burial in the lower-order structures arises from the region of the coupling, namely, the region of the four-fold helical bundle. While these results are consistent with the difference in burial of hydrophobic residues from the n- to c-terminal regions of the H3 histone chains of the nucleosome, the magnitudes of the burial over the homologous c-terminal regions of the HMk and DRm chains are found to be less than that of the H3 magnitudes. On the other hand, all eukaryotic chains that are not involved in the helical coupling yielding the tetramers, do not exhibit such difference in hydrophobic burial of residues along their length.
From a structural perspective, one expects the H3 histone to play a key role in organizing the nucleosome. From an evolutionary perspective, the n-terminal histone of HMk is clearly cast in that role [13]. While the archaeal n-terminal histone of HMk is found to exhibit little or no sequence homology with other archaeal histones, BLAST2 as well as PSI-BLAST searches, show the c-terminal histone, the histone structurally homologous with the H4 eukaryotic histone, to exhibit significant homology with other archaeal histones. How selection, involving the n-terminal histone of HMk, to enhance tetramer stability could have evolved within the context of such dichotomy is described.
Early recognition of the homology between the sequences of the Drosophila transcription factor and the eukaryotic H3 and H4 histones had suggested [14] the presence of a histone octamer-like TAF complex within TFIID that would interact with DNA in a manner similar to that of the histone octamer. Subsequent investigations [15,16] have provided chromatographic evidence suggesting the presence of a TAF octamer within the TFIID initiation complex. Comparisons of the structural and sequence homology between the tetramers of HMk, DRm and the eukaryotic H3 and H4 histones show, as expected, that the structures and sequences of species belonging to the same domain align more closely than those belonging to different domains. While the HMk dimer appears to be structurally homologous to the eukaryotic tetramer, a detailed analysis highlights differences.
Methods
Calculations have been performed utilizing the PDB structures of the histone monomer and dimer of the ancestral hyperthermophile from Methanopyrus kandleri, HMk [10] (PDB id 1F1E) and the eukaryotic transcription TATA box-binding associated factors (TAFIIs) dTAFII42 and dTAFII62 from Drosophila melanogaster, DRm [11] (PDB id 1TAF), respectively. Only the structural segments with histone fold motifs, i.e., the helix-strand-helix-strand-helix substructure segments as defined by the PDB files are extracted and used in the calculations. The HMk single chain is thus clipped in the middle and the HMk structure then appears as a histone dimer. The histone nearest the n-terminal end of the HMk monomer will be simply called the "first histone". The histone nearest the c-terminal end of the monomer will be called the "second histone". Consequently the n-terminal region of the HMk monomer may be referred to as the n-terminal region of the first histone and the c-terminal region of the HMk monomer referred to as the c-terminal region of the second histone. Dependent upon the context of discussion, the HMk monomer may be referred to as an HMk dimer and the dimer referred to as a tetramer. This terminology supplements that used previously [10] enabling a more unequivocal and parallel description of the histone homologies between HMk and other species. The "c-terminal region" of the first HMk histone is then the region involved in the four-fold helical coupling that binds the HMk dimers of the tetramer. This region is structurally homologous with the c-terminal region of the H3 eukaryotic histone. One might have been tempted to label the two segments of HMk, domain1 and domain2. Since this protein nomenclature is usually reserved for units that appear to fold independently and the histone fold is stabilized by dimerization, we have not used this terminology. The dimer of the eukaryotic Drosophila melanogaster transcription factors, dTAFII42 and dTAFII62 will be referred to as the DRm dimer. Consistent with previous usage [10], the hyperthermophile Methanopyrus kandleri will be referred to as HMk. The HMk dimer (tetramer) and the DRm tetramer may be referred to as the higher-order structures, collectively.
Calculations have also been performed on the archaeal histone dimer, HPha from Pyrococcus horikoshii [17], and on two other histone-like transcription associated factor (TAF) dimeric structures. One is from the hetero-trimeric transcription factor, NF-Y [18] (PDB id 1N1J), and the other is from the ternary complex of the negative cofactor 2, NC2 [19] (PDB id 1JFI)). The results of calculations performed on these structures for which structural homology with the eukaryotic tetramer is not expected, are useful for comparison with the results obtained for the HMk and DRm structures.
The present calculations, which provide a measure of residue distance from the interior of the dimers, and consequently, correlation coefficients between these distances and residue hydrophobicities, are based upon the residue side-chain locations. The center-of-geometry of the ith residue, or residue centroid, , is calculated with inclusion of only the backbone α – carbon atom and exclusion of the hydrogen atoms. This distribution of points in three-dimensional space enables calculation of the geometric center of the distribution, , namely, the centroid of all residue side-chain centroids of the protein. This will be called the "center-of-the-protein".
n is the total number of residues.
An ellipsoidal characterization of protein shape is obtained as follows.
A second rank geometric tensor
with , the unit dyadic, is diagonalized to provide the moments-of-geometry, g1, g2, and g3. These moments are the moments-of-inertia of a discrete distribution of points of unit mass. The moments provide an ellipsoidal characterization of protein shape.
xp, yp, zp, are coordinates in the frame of the principal axes with the centroid of the structure as origin. If the magnitudes are ordered as,
g1 <g2 <g3 [4]
the major principal axis is of length, (d2/g1)1/2.
Each ith residue at location, xip, yip, zip, in the frame of the principal axis, resides on an ellipsoid with the length of its major principal axis equal to, , namely,
For a compact globular protein, the residue associated with the largest di specifies the ellipsoid that defines a presumed "protein surface". Residues with the same di, namely, residues residing on the same ellipsoid are at the same radial fractional distance from the center-of-the-protein to the protein ellipsoidal surface.
Rewriting equation 5 as:
with
enables to be used as a measure of the radial fractional distance of the ith residue from the center-of-the-protein to the protein surface.
This distance, which will be called the ellipsoidal distance, is used in the calculations. It is just the value of the principal major axis of the ellipsoid upon which the residue centroid is found. It has been shown to correlate more closely with residue solvent accessibility than the radial distance from the center-of-the-protein to the residue centroid [20].
The scale of residue hydrophobicity chosen is that of Neumaier [21]. This scale provided greater correlation between residue hydrophobicity and ellipsoidal distance than a number of other scales initially considered. Calculations are only performed on the histone folds when in the conformation of the dimer. Consequently the residue distances obtained from the calculations are distances from the center of the dimer. This enables an examination of how the burial of residue hydrophobicity within the four-fold helical bundle trades-off against such burial within the interior of the dimer. The degree of such trade-off is quantitatively mirrored by the correlation coefficient between residue distances from the dimer interior and residue hydrophobicities.
A number of web based programs, i.e., PSI-BLAST [22], BLAST2 [23], CE [24] and CONSURF [25] have been used in examining the homology between the histones of HMk and DRm with other archaeal and eukaryal sequences.
Results and discussion
Tables 1 and 2 list the correlation coefficients previously obtained [8] between residue distances from the dimer interior and residue hydrophobicity of the amino acids of the H3, H4, H2A, and H2B histone chains. Histones from the Xenopus laevis [26] (PDB id 1KX5), Gallus gallus [27] (PDB id 1EQZ) and Saccharomyces cerevisae [28] (PDB id 1ID3) species had been used in the calculations. The correlation coefficients are provided for the entire histone chain as well as for the n- and c-terminal halves of the chain. Figure 1 highlights the c-terminal regions of the A and E chains of the H3 histones of the 1KX5 structure involved in the four-fold helical coupling that binds the dimers of the tetramer. Residues in these regions comprise a majority of the residues of the c-terminal halves of the chains. Table 1 shows significant reduction in correlation coefficient between amino acid distance from the center of the dimer and hydrophobicity over the c-terminal halves of the H3 chains of the three species compared with their n-terminal halves. The c-terminal regions of the H3 histone folds are regions of reduced burial of hydrophobic residues within the dimers and thereby provide a rationale for the observed reduced stability of the H3-H4 dimer compared with that of the H2A-H2B dimer [9]. Tables 1 and 2 also show that histones not involved in the four-fold helical coupling yielding the higher order tetrameric structures, namely, H4, H2A, and H2B, do not exhibit the extent of differential burial of hydrophobic residues over their chain-lengths as found for the H3 histones.
Table 3 shows the correlation coefficients between the histone amino acid residue distances from the interior of the dimers and their values of residue hydrophobicity for the histones of the Gallus gallus [1] (PDB id 2HIO) nucleosomal structure in the absence of DNA. One notes that the differential burial along the H3 chain is maintained as well as the distinction in values between species. It would be of interest to see if such differences are comparably maintained for the isolated dimers. If so, this would support the contention [8] that the difference in hydrophobic residue burial along the H3 histone chain is responsible for the difference in stability observed between the H3-H4 and H2A-H2B dimers in solution [9]. X-ray structures at atomic resolution have previously been obtained [29] for the individual homodimers of the A and B histones of the Methanothermus fervidus hyperthermophilic archaeon. In the presence of DNA, these dimers form tetramers [30] which are structurally homologous to the H3 and H4 eukaryotic nucleosomal tetramer.
Table 4 lists the correlation coefficients for the archaea. This includes results obtained for the x-ray structures [29] of the archaeal histones, HMfA (PDB id: 1B67) and HMfB (PDB id: 1A7W) of Methanothermus Fervidus While the difference in the correlation coefficients between residue distance and hydrophobicity from the n- to c-terminal regions of HMfA and HMfB is comparable with the difference from the n- to c-terminal regions of the eukaryotic H3 histones, the magnitude of burial over the homologous c-terminal regions is significantly less for the archaea. This difference in the burial of hydrophobic residues from the n- to c-terminal regions of the HMfA and HMfB archaeal chains can also be visually discerned from the asymmetry in values of the longer wavelength variations in hydrophobicity along the HMfA and HMfB sequences as seen in figure 8 of reference 8. Table 4 also lists the correlation coefficients for the histone chains of the HMk and HPhA hyperthermophiles. The HMk chain as well as the chain of HPhA show comparable differential burial of residues from the n- to c-terminal regions over their length as found for the methanogens, HMfA and HMfB.
HMk is particulary interesting for a number of reasons. First, the HMk monomer differs from the HMf and eukaryotic histones by containing two histone fold motifs in a single chain. Such anomaly is one of a number of strange properties exhibited by this archaeon [12,31,32]. Second, this monomeric doublet forms a dimer in the crystal [10] that appears to be structurally homologous to the (H3-H4)2 tetramer of the eukaryotic nucleosome. Gel filtration chromatography and chemical fixation have also indicated that in the absence of DNA the dominant form of HMk in solution is a stable dimer [33]. Furthermore, even though the two histones of HMk exhibit 28% sequence identity, they are exceedingly different in one respect. The first histone is distantly related to other known Archaeal sequences in the Swiss-Prot database, whereas the second histone has many close Archaeal neighbors in this database. Figure 2 shows the results of a BLAST2 search [23] of these two different histones. Interestingly, the closest hits of the first histone are four mammalian TATA box-binding proteins. The remaining Archaeal correspondences obtained are six-orders of magnitude greater in E-value than those obtained between the second histone and other Archaeal proteins in the data base. A PSI-BLAST search [22] reduces this difference to two-orders of magnitude; however, the four mammalian TATA box-binding proteins remain as nearest neighbors of the first histone. Figure 3 illustrates the segregated alignments of the two histones. The BLAST2 search [23] has been performed for the entire sequence of the doublet histone monomer. The color coded matches are shown along the full sequence. The four most distant correspondences of the ten shown are between the TATA box-binding proteins and the first histone.
Perhaps the most dramatic way to exhibit this difference between the two HMk histones is achieved by displaying the conservation of their amino acid residues. This is simply provided by a CONSURF analysis [25]. CONSURF accepts a PDB file (in the present case, the 1F1E PDB file), performs a PSI-BLAST sequence analysis and uses the results to assign an integer amino acid conservation score for each residue. The integer score is assigned a color which can be used to paint the atoms of the PDB structure and provide a visual display of amino acid conservation. The greatest integers, 9 (maroon) and 8 (magenta), are reserved for the most highly conserved residues. Figure 4 shows the two separate chains, the n- and c-terminal histones (first and second histones) of the 1F1E structure, painted with values obtained by an analysis of the full chain. Interestingly, whereas the c-terminal histone displays maroon and magenta, the n-terminal histone doesn't display any. The amino acid residues of the n-terminal histone have no 9's or 8's assigned, whereas the c-histone has been assigned 32, 9's and 8's.
Early analyses of the 16S rRNA sequence of HMk placed it close to the root of the Euryarchaeotic tree [31] while more recent studies based on more extensive sets of sequences [12,32] have grouped HMk with other archaeal methanogens. From this latter observation, it thus comes as a surprise that its genome appears to contain large numbers of genes not present in the genomes of any of the other sequenced archaeal methanogens and it contains the largest fraction of genes for which function cannot automatically be assigned based on sequence similarity [34]. Such equivocacy justifies speculation regarding the phylogeny of the HMk monomer within the context of two very different scenarios; one in which the HMk monomer evolved from a grouping with other methanogenic histones and one from a lineage in which it bore little resemblance to other archaeal histones.
In the former scenario, the unique structural feature of HMk, namely, the tandem repeat of two histones in a single chain may have been the result of gene fusion. This, as well as other HMk unique protein fusions and splittings have been observed [12,31,32,34,35]. Reduced sequence homology between the first histone and other archaeal and eukaryotic H3 histones, from that observed with the HMf methanogens, may have resulted from gene duplication and subfunctionalization involving the n-terminal HMk histone. Relaxed constraints [13] combined with the selective pressures exerted by the extreme environment of the HMk organism could have enhanced tetramer stability at the cost of sequence homology with other methanogens. This is not inconsistent with the general observation that structurally related proteins need not be related in sequence. Significant modifications in sequence are, also, consistent with the high evolutionary rate inferred for this archaeon by a phylogenetic analysis of HMk's transcription apparatus [32]. Sequence homology between the c-terminal or second histone of HMk with the other archaeal histones would be relatively maintained. Figure 5 shows the close ClustalW alignment [36] between the sequence of the second histone and that of the methanogen, HMfA. The second HMk histone also aligns most closely with the H4 histones, which is consistent with H4's role of remaining constant throughout eukaryotic evolution.
For the second scenario, the ancestral HMk histones would have had little resemblance to other archaeal histones and differences between the first HMk histone and other archaeal histones would thereby be accounted for. Selection, however, would still have yielded the helices and structures required to wrap DNA and to bury the appropriate hydrophobic residues required to couple the monomers yielding structurally homologous dimers to the eukaryotic tetramers. Homology between the sequence of the second histone and other archaeal sequences as well as with the H4 eukaryotic histone would now require explanation. This homology could have been the result of lateral gene transfer (LGT). Such transfers have been found to occur; for example, the RNA polymerase subunit H of HMk has been apparently replaced by a protein from a distantly related protein of archaeal lineage [32]. This particular transfer is strongly supported by the observation of a well conserved insert of five or six amino acid residues shared only by the RNA polymerase subunits H from M. kandleri and Thermoplasmatales. Interestingly, the nearest BLAST2 [23] neighbor of the second histone of HMk is an archaeal histone of Methanobacterium formicicum. This histone and the second HMk histone share five identical amino acids that span the fold coupling the final two c-terminal histone helices. So, to summarize, either phylogenetic scenario would be consistent with the differences observed between the two histone sequences of HMk that had occurred during selection to enhance tetramer stability.
The HPhA and HMf archaea align with approximately 60% identity and their alignments with other archaeal histones are comparable. The differential asymmetry in the burial of hydrophobic residues from the n- to c-terminal ends of the HMk and HPhA chains is also comparable. For the HPhA dimers as for the HMf dimers, one might expect this bias to have been a consequence of selection that assisted in the coupling of the dimers to form tetramers while in the presence of DNA.
Table 5 lists the correlation coefficients for three histone-like dimers from the transcription factors of, TFIID [11], NF_Y [18] (PDB id; 1N1J), and NC2 [19] (PDB id 1JFI). The asymmetry of burial of hydrophobic residues from the n- to c-terminal region of the A chain of the 1TAF structure is comparable in magnitude with that found for the Archaea. This is consistent with the role played by this chain in binding the dimers of the tetramer. Such asymmetric burial is not observed for the B chain of the 1TAF structure. These observations are consistent with the homologies, originally identified [37,38] between the A and B chain sequences of the 1TAF PDB structure with the eukaryotic H3 and H4 histone sequences, respectively. As had been previously noted [18,19], the A and B chains of the histone-like structures of 1N1J and 1JFI more closely resemble the H2A and H2B histones than the H3, and H4 histones. Consequently differential burial along the chain lengths of these histone-like structures is not expected and the values of their correlation coefficients listed in Table 5 support this contention.
Major interest in the1TAF and 1F1E structures is a result of their apparent structural homology with the eukaryotic (H3-H4)2 tetramer. Both of these histones, however, differ in at least two respects. First, Drosophila melanogaster, DRm, is a member of the eukaryotic domain, whereas Methanopyrus kandleri, HMk, belongs to the archaeal domain and consequently at a greater phylogenetic distance from the eukaryotic structures. Secondly, the results of sequence alignments and BLAST2 [23] searches are very different for these two histone structures. The sequence of the A chain of 1TAF does resemble the nucleosomal H3 sequence and a BLAST2 [23] search, in contrast with the search of the first HMk sequence, turns up numerous related sequences. A detailed comparison of the apparently homologous HMk and DRm tetrameric structures is, therefore, of interest.
Figure 6 shows ribbon diagrams of the three homologous tetrameric structures. They appear as inverted V-shaped shaped structures. From the figure it is seen that the arms of the V of HMk (figure 6c) are drawn together more closely than the arms of the eukaryotic structures (figures 6a and 6b). Closing the arms of the HMk structure places the c-terminal ends of the chains, homologous to the H4 eukaryotic chains, in close proximity. This proximity of the amino acids at the c-terminal ends of the HMk chains apparently contributes to the stability of the tetramer [10]. Figure 7 shows a wire diagram of the 1F1E and 1TAF α-carbon atom locations, CE [24] aligned with those of the eukaryotic 1KX5 structure. Only chains involved in the four-fold helical binding of the tetramer are shown. CE alignment is achieved by coupling the pairs of chains homologous to the H3 chains, as well as the H3 chains. Note that the alignment between the two eukaryotic structures is within the alignment threshold for the CE chain extension cut-off; consequently the chains are completely aligned. The RMSD between the alpha-carbon atom coordinates over the entire chain is 2.9 Angstroms with a sequence identity of 21 %. The CE alignment of the n-terminal histones of HMk with the H3 histones of 1KX5 is not possible over the entire length of the two arms of the structures, while still remaining below the CE threshold for chain extension. The longer length of HMk chain that is below this threshold is demarcated by the dashed line in figure 7. Over this aligned region, the RMSD is 2.7 Angstroms with a sequence identity of 8 %. Therefore, while the structure of the archaeal HMk dimer appears to be globally homologous with the tetrameric structure of the eukaryotic nucleosome and transcription factor, it exhibits well defined differences in structure as well as in sequence.
Conclusion
The present investigation was predominantly motivated by the availability of structures at atomic resolution of the histone dimer of the archaeal hyperthermophile, HMk, and the histone-like tetramer of the eukaryotic transcription factor of Drosophila melanogaster, DRm. The asymmetry in burial of hydrophobic residues along the lengths of the histone chains of these structures was investigated as well as sequence and structural homology with other archaeal and eukaryotic, nucleosomal and transcription factor histones. Whereas previous studies have emphasized structural similarities between the HMk histone dimer and eukaryotic tetramers, the present study has emphasized differences in structure as well as in sequence.
The calculations show that the histone chains of the Drosophila transcription factor involved in the helical coupling that yield the tetramer, exhibit the asymmetry in the burial of hydrophobic residues previously observed for the homologous histone chains of the nucleosome. The magnitude of the burial over the c-terminal region of the A chain of the 1TAF structure is, however, less compared with that calculated for the chains of the nucleosomal proteins. On the other hand, all eukaryotic chains investigated that are not involved in such four-fold helical coupling, whether from the histone-like structures of transcription factors or from the histones of the nucleosome, do not exhibit this asymmetry. Consequently, while contributing to an instability or reduction in the binding of the lower order structures, the asymmetry may also provide a marker for the presence of higher order multimers currently unobserved.
The archaeal HMk histones have also been shown to exhibit the moderate asymmetry in residue burial comparable with that found for other archaeal histones and for the histones of the 1TAF structure. On the other hand, the histones of the HMk dimer that are involved in the four-fold helical coupling of the monomers, namely, the n-terminal histones of the monomer, have been shown to have questionable sequence and structural homology with archaeal and eukaryotic histones. The HMk histone not involved in such coupling, namely the c-terminal histone, is found to be homologous to numerous other archaeal histones in the Swiss-Prot database. How selection may have enhanced the stability of the tetramer by modifications that would be consistent with this difference between the two HMk histones has been described within the context of two different phylogenetic scenarios. The limited amount of data, however, makes these speculations tentative. It will be interesting to see how the story evolves as further structures are determined.
Figures and Tables
Figure 1 Highlighted region of the four-fold helical coupling of the (H3-H4)2 tetramer.
Figure 2 BLAST2 results for the individual first and second histones of HMk.
Figure 3 BLAST2 results for the entire sequence of the HMk doublet histone monomer.
Figure 4 CONSURF coloring of the amino acid conservation of the histones of the HMk monomer.
Figure 5 ClustalW alignment of the second HMk histone sequence with the archaeal HMfA sequence.
Figure 6 Ribbon diagrams of the homologous tetrameric histone structures from (a) Drosophila melanogaster (PDB id 1TAF) (b) Xenopus laevis (PDB id 1KX5) (c) Methanopyrus kandleri (PDB id 1F1E).
Figure 7 Combinatorial Extension (CE) alignment of the tetrameric histone structures of (a) Methanopyrus kandleri (PDB id 1F1E) and (b) Drosophila melanogaster (PDB id 1TAF) with Xenopus laevis (PDB id 1KX5).
Table 1 H3/H4 Eukaryotic Histones
chain amino half carboxyl half % change
1KX5 H3 A chain correlation coefficient -0.363 -0.510 -0.242 -53
H4 B chain correlation coefficient -0.549 -0.596 -0.511 -14
1EQZ H3 C chain correlation coefficient -0.361 -0.470 -0.280 -40
H4 D chain correlation coefficient -0.550 -0.590 -0.517 12
1ID3 H3 A chain correlation coefficient -0.437 -0.602 -0.293 -51
H4 B chain correlation coefficient -0.562 -0.639 -0.494 -22
Table 2 H2A/H2B Eukaryotic Histones
chain amino half carboxyl half % change
1KX5 H2A C chain correlation coefficient -0.622 -0.662 -0.584 -12
H2B D chain correlation coefficient -0.571 -0.597 -0.531 -11
1EQZ H2A A chain correlation coefficient -0.617 -0.675 -0.546 -19
H2B B chain correlation coefficient -0.575 -0.600 -0.536 -11
1ID3 H2A C chain correlation coefficient -0.619 -0.645 -0.600 -7
H2B D chain correlation coefficient -0.611 -0.662 -0.557 -16
Table 3 Eukaryotic Histones of 2HIO
chain amino half carboxyl half
H3 C chain correlation coefficient -0.389 -0.458 -0.276
H4 D chain correlation coefficient -0.569 -0.618 -0.559
H2A A chain correlation coefficient -0.558 -0.587 -0.546
H2B B chain correlation coefficient -0.542 -0.580 -0.545
Table 4 Archaeal Histones
chain amino half carboxyl half % change
1B67 correlation coefficient -0.516 -0.641 -0.390 -39
1A7W correlation coefficient -0.537 -0.633 -0.443 -30
1F1E First histone (A chain) correlation coefficient -0.600 -0.772 -0.437 -43
Second Histone (A chain) correlation coefficient -0.561 -0.637 -0.464 -27
1KU5 correlation coefficient -0.539 -0.695 -0.446 -36
Table 5 Transcription Factor Histones
chain amino half carboxyl half % change
1TAF dTAF42 A chain correlation coefficient -0.506 -0.610 -0.398 -34
dTAF62 B chain correlation coefficient -0.460 -0.458 -0.491 7
1N1J A chain correlation coefficient -0.595 -0.585 -0.620 5
B chain correlation coefficient -0.523 -0.473 -0.571 21
1JFI A chain correlation coefficient -0.657 -0.642 -0.686 7
B chain correlation coefficient -0.663 -0.658 -0.670 2
==== Refs
Arents G Burlingame RW Wang B Love WE Moudrianakis EN The nucleosomal core histone octamer at 3.1 A resolution: A tripartite assembly and a left-handed superhelix Proc Natl Acad Sci USA 1991 88 10148 10152 1946434
Arents G Moudrianakis EN The histone fold: A ubiquitous architectural motif utilized in DNA compaction and protein dimerization Proc Natl Acad Sci USA 1995 92 11170 11174 7479959
Baxevanis AD Arents G Moudrianakis EN Landsman D A variety of DNA-binding and multimeric proteins contain the histone fold motif Nucleic Acids Res 1995 14 2685 2691
Burley SK Xie X Clark KL Shu F Histone-like transcription factors in eukaryotes Curr Opin Struct Biol 1997 7 94 102 9032065 10.1016/S0959-440X(97)80012-7
Luger K Richmond TJ DNA binding within the nucleosome core Curr Opin Struct Biol 1998 8 33 40 9519294 10.1016/S0959-440X(98)80007-9
Sullivan S Sink DW Trout KL Makalowska I Taylor PM Baxevanis AD Landsman D The Histone Database Nucleic Acids Res 2002 30 341 342 11752331 10.1093/nar/30.1.341
Luger K Mader AW Richmond RK Sargent DF Richmond TJ Crystal Structure of the nucleosome core particle at 2.8 A resolution Nature 1997 389 251 260 9305837 10.1038/38444
Silverman BD The hydrophobicity of the H3 Histone Fold differs from the Hydrophobicity of the other three Folds J Mol Evol 2005 60 354 364 15871046 10.1007/s00239-004-0193-6
Banks DD Gloss LM Equilibrium Folding of the Core Histones: the H3-H4 Tetramer is Less Stable than the H2A-H2B Dimer Biochemistry 2003 42 6827 6839 12779337 10.1021/bi026957r
Fahrner RL Cascio D Lake JA Slesarev A An ancestral nuclear protein assembly: Crystal structure of the Methanopyrus kandleri Protein Science 2001 10 2002 2007 11567091 10.1110/ps.10901
Xie X Kokubo T Cohen SL Mirza UA Hoffman A Chait BT Roeder RG Nakatani Y Burley SK Structural similarity between TAFs and the heterotetrameric core of the histone octamer Nature 1996 380 316 322 8598927 10.1038/380316a0
Slesarev S Mezhevaya KV Makarova KS Polushin NN Shcherbinina OV Shakhova VV Belova GI Aravind L Natale DA Rogozin IB Tatusov RL Wolf YI Stetter KO Malykh AG Koonin EV Kozyavkin SA The complete genome of hyperthermophile Methanopyrus kandleri AV19 and monophyly of archaeal methanogens Proc Natl Acad Sci USA 2002 99 4644 4649 11930014 10.1073/pnas.032671499
Malik HS Henikoff S Phylogenetics of the nucleosome Nature Struct Biol 2003 10 882 891 14583738 10.1038/nsb996
Hoffman A Chiang C Oeigeschlager T Xie X Burley SK Nakatani Y Roeder RG A histone octamer-like structure within TFIID Nature 1996 380 356 359 8598932 10.1038/380356a0
Selleck W Howley R Fang Q Podolny V Fried MG Buratowski Tan S A histone fold TAF octamer within the yeast TFIID transcriptional coactivator Nature Struct Biol 2001 8 695 700 11473260 10.1038/90408
Shao H Revach M Moshonov S Tzuman Y Gazit K Albeck S Unger T Dikstein R Core Promoter Binding by Histone-Like TAF Complexes Molecular and Cellular Biology 2005 25 206 219 15601843 10.1128/MCB.25.1.206-219.2005
Li T Sun F Ji X Feng Y Zihe R Structure Based Hyperthermostability of Archaeal Histone from Pyrococcus horikoshii J Mol Biol 2003 325 1031 1037 12527306 10.1016/S0022-2836(02)01285-8
Romier C Cocchiarella F Mantovani R Moras D The NF-YB-YC Structure Gives Insight into DNA Binding and Transcription Regulation by CCAAT Factor NF-Y J Biol Chem 2003 278 1336 1345 12401788 10.1074/jbc.M209635200
Kamada K Shu F Chen H Malik S Stelzer G Roeder RG Meisterernet M Burley SK Crystal Structure of Negative Cofactor 2 Recognizing the TBP-DNA Transcription Complex Cell 2001 106 71 81 11461703 10.1016/S0092-8674(01)00417-2
Silverman BD Hydrophobic Moments of Tertiary Protein Structures Proteins: Struct Funct, & Genet 2003 53 880 888 14635130 10.1002/prot.10494
Neumaier A Huyer W Bornberg-Bauer E WWW-document The signs of the values of amino acid hydrophobicity have been revesed in the present calculations 1998
Shindyalov IN Bourne PE Protein structure alignment by incremental combinatorial extension (CE) of the optimal path Protein Engineering 1998 11 739 747 9796821 10.1093/protein/11.9.739
Glaser F Pupko T Paz I Bell RE Bechor-Shental D Martz E Ben-Tal N ConSurf: identification of functional regions in proteins by surface-mapping of phylogenetic information Bioinformatics 2003 19 163 164 12499312 10.1093/bioinformatics/19.1.163
Davey CA Sargent DF Luger K Maeder AW Richmond TJ Solvent Mediated Interactions in the Structure of the Nucleosome Core Particle at 1.9 A Resolution J Mol Biol 2002 391 1097 1113 10.1016/S0022-2836(02)00386-8
Harp JM Hanson BL Timm DE Bunick GJ Asymmetries in the nucleosome core particle at 2.5 A resolution Acta Crystallogr D Biol Crystallogr 2000 12 1513 1534 10.1107/S0907444900011847
White CL Suto RK Luger K Structure of the yeast nucleosome core particle reveals fundamental changes in internucleosome interactions EMBO 2001 20 5207 5218 10.1093/emboj/20.18.5207
Decanniere K Sandman K Reeve JN Heinemann U Crystal structures of recombinant histones HMfA and HMfB from the hyperthermophilic archaeon Methanothermus fervidus J Mol Biol 2000 303 35 47 11021968 10.1006/jmbi.2000.4104
Grayling RA Sandman K Reeve JN DNA stability and DNA binding proteins Adv protein chem 1996 48 436 467
Burggraf S Stetter KO Rouviere P Woese CR Methanopyrus kandleri An Archaeal Methanogen Unrelated to all Other Known Methanogens Sysytem Appl Microbiol 1991 14 346 351
Brochier C Forterre P Gribaldo S Archaeal phylogeny based on proteins of the transcription and translation machineries: tackling the Methanopyrus kandleri paradox Genome Biology 2004 5 R17 15003120 10.1186/gb-2004-5-3-r17
Pavlov NA Cherny DI Jovin TM Slesarev AI Nucleosome-like Complex of the Histone from the Hyperthermophile Methanopyrus kandleri (MkaH) with Linear DNA Journal of Biomolecular Structure & Dynamics 2002 20 207 214 12354072
Jensen LJ Skovgaaard M Sicheritz-Ponten Jorgensen MK Lundegaard C Pedersen CC Petersen N Ussery D Analysis of two large functionally uncharacterized regions in the Methanopyrus kandleri AV19 genome BMC Genomics 2003 4 12 12697059 10.1186/1471-2164-4-12
Krah R Kozyavkin SA Slesarev AI Gellert M A two-subunit type I DNA topoisomerase (reverse gyrase) from an extreme hyperthermophile Proc Natl Acad Sci USA 1996 93 106 110 8552584 10.1073/pnas.93.1.106
Kokubo t Gong D Wooten JC Horikoshi M Roeder RG Nakatani Y Molecular cloning of Drosophila TFIID subunits Nature 1994 367 484 487 7545910 10.1038/367484a0
Nakatani Y Bagby S Ikura M The Histone Folds in Transcription Factor TFIID The Journal of Biological Chemistry 1996 22 6575 6578
|
16242031
|
PMC1283977
|
CC BY
|
2021-01-04 16:03:53
|
no
|
BMC Struct Biol. 2005 Oct 21; 5:20
|
utf-8
|
BMC Struct Biol
| 2,005 |
10.1186/1472-6807-5-20
|
oa_comm
|
==== Front
Clin Pract Epidemiol Ment HealthClin Pract Epidemiol Ment HealthClinical Practice and Epidemiology in Mental Health : CP & EMH1745-0179Bentham Science Publishers 1745-0179-1-211621612510.1186/1745-0179-1-21ReviewComorbidity issues in the pharmacological treatment of pathological gambling: a critical review Dell'Osso Bernardo [email protected] Andrea [email protected] Eric [email protected] Compulsive, Impulsive and Anxiety Disorders Program, Department of Psychiatry, Mount Sinai School of Medicine, One Gustave L. Levy Place, New York, NY 10029, USA2 Compulsive, Impulsive and Anxiety Disorders Program, Department of Psychiatry, Mount Sinai School of Medicine, One Gustave L. Levy Place, New York, NY 10029, USA3 Compulsive, Impulsive and Anxiety Disorders Program, Department of Psychiatry, Mount Sinai School of Medicine, One Gustave L. Levy Place, New York, NY 10029, USA2005 10 10 2005 1 21 21 24 8 2005 10 10 2005 Copyright ©2005 Dell'Osso et al; licensee BioMed Central Ltd.2005Dell'Osso et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Background
Pathological Gambling (PG) is an impulse control disorder often comorbid with other psychopathology, particularly bipolar spectrum disorders, attention deficit/hyperactivity disorder, obsessive-compulsive disorder (OCD) and substance abuse. This paper reviews the published literature on the pharmacological management of PG, highlighting how clinical and subclinical comorbid psychopathology influences the choice of pharmacological treatment.
Methods
Using Medline, the authors reviewed relevant articles published on this topic from1995 to 2005, focusing on the best-designed studies for inclusion.
Results
Much of the literature on PG-treatment presupposes different theories regarding this disorder. Data suggest the utility of differentiating the pharmacotherapy of pathological gamblers in light of their comorbid profile, specifically assessing for comorbid bipolar, ADHD, OCD, and substance abuse disorders.
Conclusion
Decisions about pharmacological treatment of PG should take into account current and previous comorbid disorders which influence treatment selection.
Pathological GamblingImpulse Control DisordersComorbidityCompulsive-Impulsive SpectrumSubthreshold symptoms
==== Body
Background
Pathological gambling (PG) is an impulse control disorder not otherwise specified (ICD-NOS) [1] that is characterized by recurrent and maladaptive patterns of gambling behavior and significantly disrupts the patient's functioning in the personal, familial, or vocational spheres. It is assumed to be a chronic disorder, with a clinical course that is continuous, unremitting, or episodic [2]. Its prevalence ranges from 1% to 3% of the US adult population [3,4], and there has been a dramatic increase in PG over the last decade, due to the legalization and availability of new forms of gambling in most Western countries. Despite a prevalence even higher than that of schizophrenia or bipolar disorder, little is known regarding effective treatments, particularly pharmacotherapies, for PG. In addition, currently, no medications have been approved by the U.S. Food and Drug Administration for the treatment of this impairing and common disorder.
A crucial issue to consider in approaching PG is represented by the high rates of comorbidity among pathological gamblers. The majority of these patients, at least those seeking treatment, have been found to score significantly higher than control populations on measures of depression [5], and have shown high incidences of various psychiatric disorders, including bipolar, anxiety and substance use disorders [6]. This frequent comorbidity is not surprising if we focus on the psychopathological core features of PG: impulsivity, compulsive drive to gamble, addictive features such as withdrawal symptoms during gambling abstinence, and bipolar features such as urges, pleasure seeking and decreased judgment due to unrealistic appraisal of the individuals' own abilities. Several authors have associated some of these core features to neurobiological data and clinical aspects of treatment-response, and have conceptualized PG as belonging to different spectrum classification models, in which the main psychiatric disorders of reference are obsessive-compulsive disorder (OCD), addictive disorders, and affective disorders. These models of categorization provide the basis and rationale for the use of specific pharmacological treatments in pathological gamblers. In addition, they may also suggest, according to consistent findings reported in some trials, the presence of specific subgroups of patients with similar core features, comorbid profiles and treatment-responses within the population of pathological gamblers.
Main classification models for Pathological Gambling
The traditional nosographic model includes PG within ICDs-NOS; evidence supporting this categorization is the elevated rates of comorbidity between these disorders, and the similarity in phenomenology between PG and other ICDs. These similarities include the temptation to perform some behavior notwithstanding its detrimental consequences for the person, a growing emotional tension before performing the act, a gratifying feeling while performing the behavior, and sometimes, a feeling of guilt following the behavior. However, in addition to this classification model, at least three other conceptualizations have been historically proposed for the classification of PG [7].
PG has also been conceptualized as an obsessive-compulsive (OC) spectrum disorder, within the impulsive cluster [8]. Patients with OC spectrum disorders, in fact, experience unpleasant feelings and physiological activation that result in an intense desire to perform a specific behavior in order to relieve the unpleasant feelings [9-11]; this is the case in PG. In addition, a reduced capacity to resist gambling thoughts would lead to excessive gambling, in particular in the advanced phases of the disorder [12]. However, these patients differ from patients with OCD in important ways. Gambling behavior and thoughts are often experienced by these patients as ego-syntonic, while OCD obsessions and compulsions are generally ego-dystonic. In addition, the excessive doubt, frequently experienced by OCD patients [10,13,14] as well as their compulsions, characterized by harm avoidance, risk aversion and anticipatory anxiety [14], are not characteristic of pathological gamblers. OC spectrum disorders do differ along the dimension of risk aversion vs risk taking; the compulsive disorders are characterized by an overestimation of harm and by risk aversion while the impulsive disorders are characterized by an underestimation of risk and by risk seeking.
PG has occasionally been characterized as an affective spectrum disorder. Notwithstanding the high rates of comorbidity between depression and PG [15-18] and the frequent presence of suicidality and suicidal ideation among these patients [19-22], the link between these two disorders has been questioned by several authors. With regard to suicidality, for example, is possible to guess that high levels of impulsivity, often present in pathological gamblers, may lead to suicidality independently from depression. On the other hand, data from several studies suggest that PG is a bipolar spectrum disorder and, considering that the comorbidity between these two disorders is estimated to be approximately 24% [23], have led some authors to the conclusion that impulsivity and bipolarity are related [24]. Other authors have stressed that certain PG core features resemble some characteristic aspects of bipolar spectrum disorders. The experience of the urge in PG, for example, may represent a potential overlap between pathological gamblers and patients with bipolar spectrum disorders. The pleasure seeking and impaired judgment resulting from an unrealistic evaluation of one's own abilities that are seen in pathological gamblers resemble characteristics seen in individuals with bipolar disorder [25]. Both disorders involve potentially harmful but pleasurable behavior and acting without forethought. Finally, these disorders generally have their onset in early adulthood and have an episodic course.
Another interesting model proposed for PG is as a non-substance addiction. PG and substance use disorders, in fact, share several features. Important features common to these mental disorders include the intense desire to satisfy a need; the lack of control over the substance use or behavior; certain aspects of abstinence and tolerance; the obsessive thoughts about the substance use or the activity; and the continuous engagement in the behavior despite experiencing social and occupational consequences [26]. For example, pathological gamblers often increase the frequency of their bets or the amount of money gambled in order to achieve the desired level of excitement and this behavior is suggestive of drug tolerance. Finally, high comorbidity rates between PG and substance use disorders [23,27-33] particularly alcohol and nicotine abuse and dependence, give further support to this model.
Being able to explain PG using different models does not necessarily imply that these models are incompatible; rather it may suggest the possible presence of subtypes of pathological gamblers. Being able to identify different subtypes that share similar phenomenological and clinical characteristics might not only lead to a better understanding of this disorder, but better use of pharmacological options. In fact, if it is possible to distinguish different clinical subtypes, it would seem logical that pharmacological treatments could be optimized to fit specific subgroups. Several core symptoms of PG could conceivably be targeted for treatment: impulsive symptoms, compulsive symptoms, bipolar symptoms and addictive symptoms. A rational pharmacological choice would take into account which of these symptom domains seem to dominate, as well as a patient's comorbid disorders, because treatment should ultimately target all clinically significant symptoms in the individual patient. Actually, the ability of specific pharmacological treatments to improve gambling behavior in some patients and the lack of efficacy or, even the worsening, in some other patients, suggests the presence of specific subtypes of pathological gamblers and supports the possibility of targeting pharmacotherapy based on the patient's unique symptoms and comorbidities.
Methods
In this review we used a Medline search to locate the published articles from 1995 to 2005. The most weight was given to double blind, placebo-controlled studies due to the high placebo response rates reported in most trials; open-label studies and case-reports have also been included. In addition, we focused on studies that were relevant to understanding the connection between the characteristics of the samples, the comorbidity profiles, the drug administrated and the clinical response achieved.
The variety of drugs employed in the treatment of PG includes serotonin reuptake inhibitors (SRIs) and other antidepressants, mood stabilizers, and opioid antagonists. We did not include the atypical antipsychotics since there has been little systematic investigation of them in PG, and there is little evidence of their effectiveness in this disorder.
Results
Serotonin Reuptake Inhibitors (see Tables 1, 2 - additional file: 1)
Serotonin reuptake inhibitors, considered first-line treatment for OCD [34], have been shown to be effective in treating impulsivity in other ICDs and OC spectrum disorders [35-39]. Furthermore, neurobiological data indicating serotonergic dysfunction in PG [40-42], a phenomenological link to compulsivity [8], and a possible response in PG to the SRI clomipramine [43], have given further support for the use of the SRIs in this disorder. To date, the following SRI compounds have been tested for the treatment of PG: fluvoxamine, paroxetine, citalopram and clomipramine.
Fluvoxamine
The possible efficacy of fluvoxamine in the treatment of PG has been tested by three randomized, placebo-controlled studies. In an initial pilot study conducted by our group [44], 16 pathological gamblers were enrolled in a single-blind, 16-week, crossover study at the end of which 7 of the 10 completers receiving fluvoxamine were found to be treatment responders with notable improvements on the main outcome measures: the Pathological Gambling Yale-Brown Obsessive Compulsive Scale (PG-YBOCS, reduction ≥ 25%)[45] and the Clinical Global Impressions Scale[46] (CGI score 2, very much improved or 1, much improved). The mean fluvoxamine dose at the end of the study was 207 mg/day and the 7 treatment responders reported a total abstinence from gambling behavior. It is noteworthy that among the 3 fluvoxamine non-responders, 2 patients had a history of cyclothymia, the mildest form of bipolar disorder, which raised the possibility of a symptomatologic exacerbation with this SRI as well as a gambling relapse.
In a subsequent double-blind crossover study [47] comparing fluvoxamine to placebo, a group of 10 male patients completed the 16-week trial. The percent improvement on the CGI-PG was significantly greater for fluvoxamine (40.6%) than for placebo (16.6%), and the percent improvement on PG-YBOCS, although not statistically significant, was greater for the group treated with fluvoxamine (33.4%) than with placebo (28%). The average fluvoxamine dosage at the end point was 195 mg/day. In this study, however, the patient group was limited and relatively homogenous, excluding gamblers with current comorbid drug or alcohol abuse as well as bipolar I and II patients.
In another double-blind, parallel, placebo-controlled trial [48], 13 pathological gamblers of the 32 enrolled completed a 6 month-study, receiving a mean fluvoxamine dosage of 200 mg/ day at endpoint. Blanco and colleagues, however, reported a statistically significant improvement with fluvoxamine compared with placebo only for males and for young pathological gamblers. Furthermore, a high drop-out rate in the fluvoxamine group, a high rate of placebo response (59%) and the concomitant psychotherapy received by some patients complicate the interpretation of the results. We cannot rule out that, notwithstanding the absence of another current Axis I disorder in the sample, the presence of subthreshold comorbid psychopathology might have decreased the homogeneity of the sample, suggesting different subtypes of pathological gamblers.
Taken as a whole, fluvoxamine trials support the efficacy of this compound in PG, although in some cases the efficacy was not statistically significant, as compared with placebo. Drug dosages ranged between 100 and 250 mg/day. It's noteworthy that at least in the first study different patterns of treatment-response were related to different comorbidity profiles.
Paroxetine
In a study conducted by Kim and colleagues, 45 patients were randomized to paroxetine up to 60 mg/day or to placebo in a double-blind, 8-week trial [49]. At endpoint, there were greater, and statistically significant, improvements (G-SAS [50], p = .042 and CGI, p = .025) in the paroxetine group than in the placebo group. As the authors reported, all subjects had low baseline scores at the Hamilton Depression Rating Scale (HDRS) [51] and Hamilton Anxiety Rating Scale (HARS) [52] suggesting low comorbidity rates in the sample for depression and anxiety. In addition, patients with other Axis I disorders were excluded from the study. However, it is possible to assume that the low presence of comorbidity might in turn be indicative of a specific subgroup of gamblers particularly sensitive to the SRI treatment.
In a later multicenter, double-blind, placebo-controlled trial [53], Grant and co-workers randomly assigned 76 pathological gamblers to paroxetine, up to 60 mg/day, or to placebo for 16 weeks. A total of 45 patients completed the study. The paroxetine group showed a greater percentage of responders (59%) at each study visit, compared with placebo (49%), but failed to demonstrate statistical significance over placebo on the outcome measures (PG-CGI scores of 1 or 2). Current Axis I disorders as well as a past history of bipolar, psychotic, alcohol or substance use disorders were exclusion criteria. The results of this study did not replicate the previous findings by Kim and colleagues, and found a notable placebo response for pathological gamblers without relevant comorbid conditions. Grant and colleagues suggested that a possible interpretation of these results might be the tendency to reduce unwanted behaviors when greater attention is focused on them, which might also increase motivation. However, the presence of subthreshold psychopathology was not investigated and whether or not this lack of response was related to particular subthreshold comorbid conditions.
Citalopram
In the only open-label study performed with citalopram [54], Zimmerman and colleagues reported a clinically significant improvement (PG-CGI scores of 1 or 2) for 13 of the 15 patients participating in the trial. The final mean dosage of citalopram was approximately 35 mg/day. Given the design of the study, which did not control for placebo response, it is not possible to calculate the actual effect of citalopram. However, it is noteworthy that while psychotic disorders, mania and hypomania, as well as drug or alcohol dependence were excluded, depression, anxiety, eating or impulse-control disorders were not. This difference could be important in the understanding the impressive clinical response in this study, suggesting that these particular comorbid disorders might not represent a specific contraindication to the use of the SSRIs
Other antidepressants (see Table 3 - additional file: 1)
Nefazodone
In 2002, Pallanti and our group enrolled 14 patients with PG in an 8-week open-label oral nefazodone trial [55]. The sample included other Axis I comorbid disorders such as bipolar II, cyclothymia, depression, panic disorder and social phobia. At the endpoint, 9 of the 12 completers were classified as responders (reduction ≥ 25% PG-Y-BOCS and score of 1 or 2 on PG-CGI). The mean endpoint dosage of nefazodone was 350 mg/ day. This phenylpiperazine antidepressant with 5HT2 receptor antagonist properties and mixed noradrenergic/serotonergic reuptake inhibitor effects showed a good response profile in this group of pathological gamblers with other comorbid psychopathology. However, the study design as well as the limited size of the sample does not allow definitive conclusions.
Dopamine Reuptake Inhibitors
More recently, Black reported positive findings in an open-label trial [56] involving 10 pathological gamblers treated with bupropion up to 300 mg/day for 8 weeks. All subjects reported improvement (reduction from 20.3 to 8.8 on the PG-Y-BOCS and score of 1 or 2 on PG-CGI). Given patients' initial mild to moderate levels of ADHD traits, the author suggested that bupropion might reduce impulsiveness and improve attention span. In addition, he hypothesized that pathological gamblers with comorbid ADHD might represent a distinct subgroup of patients requiring a specific treatment.
Opioid Antagonists (see Table 4 - additional file: 1)
Naltrexone, a long-acting opioid antagonist, is the only compound in this category that has been found to be effective in the treatment of PG. This compound blocks the effect of endogenous endorphins on central opiate receptors and also inhibits dopamine release in the nucleus accumbens, acting on neuronal pathways involved in reward, pleasure and urge. The inhibition of dopamine release in the nucleus accumbens, through the disinhibition of γ-aminobutyric acid (GABA) input to the dopamine neurons in the ventral tegmental area [57-66], is one of the most consistent reasons given for the use of naltrexone in ICDs. Another important pharmacological action of naltrexone in the central nervous system is the antagonism of the μ-opioid receptor, which is the site of action of beta-endorphins, morphine and heroin. Shinohara and colleagues [67] have suggested a possible role of this system in the physiological responses to gambling, given its involvement in the processing of reward, pleasure and pain. In addition, some neurobiological and neuroimaging reports [68,69] would confirm the relationship between some specific brain areas (nucleus accumbens, orbitofrontal cortex and motor limbic system), in which naltrexone is supposed to act, and their physiological role in the processes of reward and urge, typically abnormal in pathological gamblers.
So far, the clinical use of naltrexone has shown mixed results in treating urge related disorders such as alcohol dependence, OCD, bulimia nervosa, kleptomania, and self-injurious behaviors [70-74]. In a case report [70] of an open label treatment, Crockford and el-Guebaly reported that naltrexone at 50 mg/ day was effective in a patient suffering from both PG and alcohol dependence. In another case-report [75], a 55-year-old man with PG and compulsive shopping markedly improved while taking 100 mg/day of naltrexone. In a 6-week open-label study [76], 17 pathological gamblers without severe psychiatric comorbidity showed a statistically significant reduction in both gambling behavior and urges on a mean dosage of 157 mg/ day of naltrexone. In an 11-week double-blind, placebo-controlled study [50], 83 patients with PG were randomly assigned to naltrexone (up to 250 mg/day, mean final dosage of 188 mg/day), or to placebo. Of the 45 completers receiving naltrexone, 75% showed improvement on the outcome measures (patient- and clinician-rated CGI), compared to 24% of patients receiving placebo. The authors reported that depressive symptoms were mild or absent in the completers receiving naltrexone. Furthermore, patients with moderate or higher levels of gambling urges at baseline had a better response to naltrexone than other patients. It is possible to argue that urges, which represent one of the core symptoms of PG, might also be the main target of naltrexone. As suggested by Kim's group, baseline urge level might be used as a stratification variable that could enhance group differences in outcome and could also be used to predict response to naltrexone. However, the side-effect profile of this compound may be problematic. Since higher naltrexone doses are needed in PG treatment than are usually employed for alcohol and drug dependence, the increase in transaminase levels must be carefully monitored due to the risk of hepatotoxicity. In addition, other common side-effects reduce patient compliance. Currently, research is being conducted on another opioid receptor antagonist, nalmefene, to assess its efficacy in treating pathological gambling with preliminary encouraging findings.
Mood Stabilizers (see Table 5 - additional file: 1)
Evidence of the effectiveness of mood stabilizers in ICDs has been reported [77-85], suggesting these compounds can modify, and treat successfully, some core features of these disorders. In 1980, Moskowitz [86] reported the effectiveness of lithium in treating 3 pathological gamblers with bipolar features. In 1994, Haller and Hinterhuber reported, in a placebo controlled studyl [87], a single case of chronic PG successfully treated with carbamazepine at a dosage of 600 mg/day. Subsequently, a single-blind placebo-controlled study [88] was conducted by our group in order to evaluate the efficacy and safety of lithium and valproate in non-bipolar pathological gamblers. At the end of the 14th week, 14 of the 23 patients receiving lithium (mean dose 795 mg/day) and 11 of the 19 taking valproate (mean dose 870 mg/ day) were considered responders (PG-CGI and PG-YBOCS). Although patients with bipolar disorder were excluded from thus study as well as patients with current alcohol/drug addiction or schizophrenic spectrum disorders, the majority of patients had a past history of other psychiatric conditions, including depressive episodes (7/45), alcohol or drug abuse (23/45), panic disorder (15/45), OCD (9/45), antisocial personality disorder (8/45) and other ICDs (16/45). Therefore, even if a specific "anti-impulsive" action of mood stabilizers may be hypothesized, we have to consider the impact of lifetime comorbidities, which may currently be subthreshold, when considering treatment-response data. In addition, although the Structured Clinical Interview for DSM-IV (SCID) [89] was used in the study to exclude the comorbid bipolar disorders, the accuracy of this instrument in detecting bipolar disorder, particularly type II, is still matter of debate. Consequently, the possible inclusion of bipolar II patients in the study should be kept in mind.
Recently, our group conducted the first placebo-controlled treatment study [90] of sustained-release lithium carbonate in pathological gamblers with bipolar spectrum disorders. Patients with bipolar I disorder were excluded because it was a placebo-controlled trial, but patients with diagnoses of bipolar II, bipolar disorder NOS and cyclothymia were included in the trial. Among the 29 completers at the end of the 10th week, 12 patients received lithium (mean dose 1150 mg/day) and 17 received placebo. At endpoint, 10/12 patients in the lithium group were considered responders (based on CGI of 1 or 2 and a score reduction of 35% on the PG-YBOCS), and improvement in impulsive gambling significantly correlated with increases in affective stability. Of note, the percentage of placebo responders found in this study (29%) appears to be significantly lower than that reported in other studies. This study, deliberately designed to evaluate the effectiveness of a mood stabilizer in pathological gamblers with comorbid bipolar spectrum features confirms previous reports of the effectiveness of these compounds in PG, and suggests that there may be significant advantages in subtyping pathological gamblers. The identification of bipolar spectrum PG patients is relevant to the choice of pharmacological treatment; in this subtype, in fact, a mood stabilizer might have a higher probability of efficacy than other treatments such as SSRIs, which may exacerbate affective instability and cause a gambling relapse.
Recently, Dannon and his group randomized 31 male, pathological gamblers to topiramate or fluvoxamine (both titrated up to 200 mg/day) in a 12 week blind-rater comparison trial [91]. At endpoint, both groups reported improvement on the PG-CGI, although the improvement for fluvoxamine did not quite reach statistical significance (p < 0.08 for fluvoxamine; p < 0.01 for topiramate); there were no statistically significant differences between the 2 groups. In addition, a larger number of drop-outs were reported in the fluvoxamine group. Notwithstanding the lack of a placebo-controlled group and the partial blind design of this trial, it is noteworthy that a mood stabilizer was at least as effective as an SSRI in a group of pathological gamblers without comorbid conditions, providing, therefore, more than one choice for this specific subgroup of gamblers.
Among the 18 studies reported, 6 were double-blind, placebo-controlled studies (4 with SRIs, 1 with naltrexone and 1 with a mood stabilizer), 3 were single-blind (1 with an SRI and 1 comparing mood stabilizers, and one comparing an SRI with a mood stabilizer), 4 were open-label (1 with an SRI, 1 with nefazadone, 1 with naltrexone, and one with bupropion) and 4 were case-reports (2 with naltrexone and 2 with mood stabilizers, one of which had a double blind design).
With regard to reported SRI trials, a total of 127/200 completed single- and double-blind trials (41 with fluvoxamine and 86 with paroxetine), 15 patients completed an open-label trial with citalopram (see Tables 1 and 2 - additional file: 1). In addition, 12 patients received open-label treatment with nefazodone and 10 patients with bupropion (Table 3 - additional file: 1).
Thirty-six out of forty-five patients completed a double-blind trial with naltrexone, 14/17 patients completed an open-label trial (table 4 - additional file: 1), and 2 patients were reported in 2 different case reports to have received open label treatment.
Seventy-two out of ninty-seven patients completed single- and double-blind trials with mood stabilizers (27 with lithium, 16 with valproate, 12 with topiramate and 17 treated with placebo Table 5), and 4 cases of patients treated with mood stabilizers were reported, one of which treated in a double-blind design.
The majority of pathological gamblers have been treated with SRIs, and these drugs seem to be generally well tolerated. Effective does in PG resemble those used in the treatment of OCD; approximately 200 mg/day for fluvoxamine, and up to 60 mg/day for paroxetine were effective in PG trials. These studies demonstrated a marked placebo effect, and therefore they need an adequate period of time (between 8 and 12 weeks) in order to monitor and assess the improvement achieved. Only one SRI trial [44] looked at the relationship between worsening PG symptoms in non-responders and comorbid cyclothymia. In all the studies, the efficacy of the SRIs, when statistically significant, was independent of underlying depressive or anxious comorbidity. This supports the hypothesis that a serotonin sensitive PG subtype exists in which the marked impulsivity is driven by serotonergic dysregulation. In this subtype, the use of an SRI might normalize the 5-HT dysfunction and improve the clinical condition. An important exception, however, is the subgroup of pathological gamblers with bipolar features, for whom treatment with an SRI could precipitate worsening of the overall clinical picture. Therefore, although SRIs have been shown to be effective in patients with mild obsessive-compulsive, depressive and anxious comorbidity, additional research should be conducted on PG cormorbid with bipolar spectrum disorders. In this research, it would be important to consider not only current comorbid diagnoses but also the past history of the patient and careful attention to other current symptoms to assess the presence of subthreshold bipolar psychopathology [88]. Future double-blind studies are necessary in order to acquire further and more detailed information about the efficacy of the SRIs already employed in the treatment of PG, as well as of those which have never been tested.
Although only 1 double-blind study has been conducted in PG with naltrexone, its results may have important implications. The target of this opioid antagonist seems to be the urge to gamble. Even if this core symptom is also a feature of bipolar disorder, it is sensitive to the action of naltrexone, and its predominance in specific cases might indicate this opioid antagonist should be considered. In addition, the presence of comorbid substance and alcohol abuse/dependence might also suggest that naltrexone should be tried, as shown by the case reports. However, although naltrexone has demonstrated specific anti-impulsive properties, it could be difficult to consider this drug a first-choice treatment for PG, given the high dosage required to be effective, its potential hepatotoxicity risk and the side-effects profile reported. However, pathological gamblers with specific comorbid profiles and without hepatic diseases may be successfully treated with this opioid antagonist. Finally, promising preliminary findings [92] have been shown with another opioid antagonist, nalmefene, which may have better tolerability than naltrexone.
The studies conducted with mood stabilizers confirm what has been postulated about the SRIs. A specific anti-impulsive property of mood stabilizers can be hypothesized. In addition, the impulsivity may be related to affective instability and, therefore, a decrease in affective instability with mood stabilizers might reduce impulsivity, improving the overall clinical condition. The findings support the hypothesis of a PG subtype in which the administration of SRIs would be useless or even worsen the associated bipolar spectrum symptoms. Specific contraindications for the use of mood-stabilizers in PG have not yet been reported. Although additional confirming research is necessary, the possible efficacy of these compounds in a wide spectrum of pathological gamblers was recently supported by an open-label study in which fluvoxamine and topiramate showed similar efficacy in treating PG.
The most recent double-blind PG study reported by our group [92] included only PG patients with comorbid bipolar II, cyclothymia and bipolar NOS diagnoses. The recognition of this as a possibly important subgroup of PG led to this study of the efficacy of lithum in gamblers with PG and bipolar spectrum disorders. These patients demonstrated statistically significant improvement compared to placebo on all key outcome measures, with 83% of the patients on lithium considered responders compared to 29% on placebo, a notably smaller placebo effect than normally reported in SRIs trials. The inclusion of adequate clinical instruments and interviews in order to obtain precise comorbid diagnoses may enable complex findings to be interpreted and lead to more precisely targeted treatments. Only a minority of studies have done this. Generally authors prefer to select samples of patients without other current psychopathology, in order to obtain samples that are as homogeneous as possible and to have a sample of "pure" PG patients. However, this procedure, though excluding complicating comorbid pathologies, does not exclude the possible influence of the subthreshold psychopathology, which may influence the treatment-response. In addition, such a procedure may lead to treatments which are effective for only a small and atypical group of pathological gamblers.
Future studies are needed in order to confirm the clinical validity of dividing pathological gamblers into specific subgroups. For example, neuroimaging studies might assess if different subgroups of patients show common patterns of neuronal activation, metabolism and perfusion of specific brain areas.
Conclusion
Treatment data supports the hypothesis that there are specific subgroups of pathological gamblers. These subgroups were suggested by the different models of PG that were discussed. The assessment of clinical and subclinical comorbid psychopathology seems to represent a rational and valid approach to selecting pharmacological treatment for the different subgroups. However, few studies reported to date have specifically explored the comorbid profile of pathological gamblers. Nevertheless, positive findings shown in some studies and interesting observations reported in other studies, suggest this approach support is promising and might lead to more effective treatments for specific subgroups of gamblers.
Competing interests
Disclosure: Dr. Dell'Osso and Dr. Allen do not have an affiliation or financial interest in any organization that might pose a conflict of interest. Prof. Hollander is on the advisory board of Abbott, Ortho-McNeil and Solvay.
Supplementary Material
Additional file 1
Click here for file
Acknowledgements
Finding/Support: Prof. Hollander has received grant support from The National Institute of Drug Addiction, The National Institute of Neurological Diseases and Stroke, The National Institute of Mental Health, Abbott, Ortho-McNeil, Pfizer, Solvay, UCB Pharma and Wyeth.
==== Refs
American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision 2000 4 Washington, DC, American Psychiatric Association
Hollander E Buchalter AJ DeCaria C Pathological gambling Psychiatr Clin North Am 2000 23 629 642 10.1016/S0193-953X(05)70185-4 10986732
Shaffer HJ Hall MN Vander Bilt J Estimating the prevalence of disordered gambling behavior in the United States and Canada: a research synthesis Am J Public Health 1999 89 1369 1376 10474555
Shaffer HJ Hall MN Updating and refining prevalence estimates of disorderd gambling behaviour in the United States and Canada Can J Public Health 2001 92 168 172 11496623
Roy A Adinoff B Roehrich L Depressed pathological gamblers Acta Psychiatr Scand 1988 77 163 165 3364202
National Research Council Pathological Gambling: a critical review 1999 Washington DC, National Academy Press
Moreyra P Ibanez A Saiz-Ruiz J Grant JE, Potenza MN Categorization in: Pathological Gambling a clinical guide to treatment American Psychiatric Publishing Inc, Washington DC 55 68
Hollander E Obsessive-Compulsive Related Disorders 1993 Washington, DC, American Psychiatric Press
Hantouche E Merckaert P Nosological classifications of obsessive-compulsive disorder Ann Med Psychol 1991 149 393 408
Hollander E Skodol A Oldham J Impulsivity and compulsivity 1996 Washington, DC, American Psychiatric Press
Cartwright C DeCaria C Hollander E Pathological gambling: a clinical review Pratical Psychiatry and Behavioral Health 1998 4 277 286
Lesieur HR The compulsive gambler's spiral of options and involvement Psychiatry 1979 42 79 87 760136
Leisieur HR Custer RL Pathological gambling: roots, phases and treatment Ann Acad Poll Soc Sci 1984 474 146 156
Rasmussen SA Eisen JL The epidemiology and differential diagnosis of obsessive-compulsive disorder J Clin Psychiatry 1992 53 4 10 1564054
Cusack JR Malaney KR Depry DL Insights about pathological gamblers: "chasing losses" in spite of the cosequences Postgrad Med 1993 9 169 176 8460075
Blaszczynsky AP McConaghy N SCL-90 assessed psychopathology in pathological gamblers Psychol Rep 1988 62 547 552 3406270
McCormick RA Taber JI Kruedelback N Personality profiles of hospitalised pathological gamblers: the California Personality inventory J Clin Psychol 1987 43 521 527 3667946
Saiz-Ruiz J Moreno I Lopez-Ibor JJ Pathological gambling: a clinical and therapeutic-evolutive study of a group of pathologic gamblers Actas Luso Esp Neurol Psychiatr Cienc Afines 1992 20 189 197
Ibanez A Blanco C Donahue E Psychiatric comorbidity in pathological gamblers seeking treatment Am J Psychiatry 2001 158 1733 1735 10.1176/appi.ajp.158.10.1733 11579014
Thorson JA Powell FC Hilt M Epidemiology of gambling and depression in an adult sample Psychol Rep 1994 74 987 994 8058890
Sullivan S Abbott M McAvoy B Pathological gamblers: will they use a new telephone hotline? N Z Med J 1994 107 313 315 8052468
Petry NM Kiluk BD Suicidal ideation and suicide attempts in treatment-seeking pathological gamblers J Nerv Ment Dis 2002 190 462 469 10.1097/00005053-200207000-00007 12142848
Linden RD Pope HG Jonas JM Pathological gambling and major affective disorders: preliminary findings J Clin Psychiatry 1986 47 201 203 3957881
McElroy S Pope HG jrKech PE jrAre impulse control disorders related to bipolar disorder? Compr Psychiatry 1996 37 229 240 10.1016/S0010-440X(96)90001-2 8826686
Sood ED Pallanti S Hollander E Diagnosis and treatment of pathological gambling Curr Psychiatry Rep 2003 5 9 15 12685996
World Health Organization International Classification of Diseases and Related Health Problems, 10th Revision 1992 Geneva, World Health Organization
Daghestani AN Elenz E Crayton JV Pathological gambling in hospitalized substance abusing veterans J Clin Psychiatry 1996 57 360 363 8752019
Haberman PW Drinking and other self-indulgences: complements or counterattractions? Int J Addict 1969 4 157 167
Leisieur HR Blume SP Zoppa RM Alcoholism, drug abuse and gambling Alcohol Clin Exp Res 1986 10 33 38 3515989
Lesieur HR Heineman M Pathological gambling among youthful multiple substance abusers in a therapeutic community Br J Addict 1988 83 765 771 3207936
McCormick RA Russo AM Ramirez LF Affective disorders among pathological gamblers seeking treatment Am J Psychiatry 1984 141 215 218 6691482
Spunt B Lesieur H Hunt D Gambling among metadone patients Int J Addict 1995 30 929 962 7558485
Voldberg RA Gambling and Problem Gambling in New York, a 10 year replication survey, 1986 to 1996 1996 Albany, NY, New York Council on Problem Gambling
Hollander E Pallanti S Davis KL, Charney D, Coyle JT Current and experimental therapeutics of obsessive-compulsive disorder, in Neuropsychopharmacology: The Fifth Generation of Progress: An Official Publication of the American College of Neuropsychopharmacology 2002 Philadelphia, PA Lippincott Williams & Wilkins 1647 1664
McElroy S Kech PE jrPope HG jrPharmacological treatment of kleptomania and bulimia nervosa J Clin Psychopharmacol 1989 9 358 360 2677062
Black D Compulsive Buying: a review J Clin Psychiatry 1996 57 50 55 8698681
Simeon D Stein J Gross S A double-blind trial of fluoxetine in pathologic skin picking J Clin Psychiatry 1997 58 341 347 9515971
Hollander E Allen A Kwon J Clomipramine vs desipramine crossover trial in body dysmorphic disorder: selective efficacy of a serotonin reuptake inhibitor in imagined ugliness Arch Gen Psychiatry 1999 56 1033 1039 10.1001/archpsyc.56.11.1033 10565503
Philips KA Albertini RS Rasmussen SA A randomized placebo-controlled trial of fluoxetine in body dysmorphic disorder Arch Gen Psychiatry 2002 59 381 388 10.1001/archpsyc.59.4.381 11926939
Blanco C Orensanz-Munoz L Blanco-Jerez C Pathological gambling and platelet MAO activity: a psychobiological study Am J Psychiatry 1996 153 119 121 8540570
DeCaria CM Hollander E Grossman R Diagnosis, neurobiology and treatment of Pathological gambling J Clin Psychiatry 1996 57 80 83 8698687
Moreno I Saiz-Ruiz J Lopez-Ibor JJ Serotonin and gambling dependence Hum Psychopharmacol 1996 Suppl 6 9 12
Hollander E Frenkel M DeCaria C Treatment of pathological gambling with clomipramine Am J Psychiatry 1992 149 710 711 1575267
Hollander E DeCaria CM Mari E Short-term single blind fluvoxamine treatment of pathological gambling Am J psychiatry 1998 155 1781 1783 9842795
DeCaria CM Hollander E Begaz T Reliability and validity of a pathological gambling modification of the Yale-Brown Obsessive Compulsive Scale (PG-YBOCS): preliminary findings, in Proceedings of the 12th national conference on Problem Gambling 1998
Guy W ed ECDEU Assessment Manual for Psychopharmacology: Publication ADM 76–338 1976 Washington DC, US Department of Health, Education and Welfare 218 222
Hollander E DeCaria CM Finkell NJ A randomized double blind fluvoxamine/placebo crossover trial in pathological gambling Biol Psychiatry 2000 47 813 817 10.1016/S0006-3223(00)00241-9 10812040
Blanco C Petkova E Ibanez A A Pilot Placebo-Controlled Study of Fluvoxamine for Pathological Gambling Ann Clin Psychiatry 2002 14 9 15 10.1023/A:1015215809770 12046642
Kim SW Grant JE Adson DE A double-blind placebo controlled study of the efficacy and safety of paroxetine in the treatment of pathological gambling J Clin Psychiatry 2002 14 9 15 10.1023/A:1015215809770
Kim SW Grant JE Adson DE Double-blind naltrexone and placebo comparison study in the treatment of pathological gambling Biol Psychiatry 2001 49 914 921 10.1016/S0006-3223(01)01079-4 11377409
Hamilton M A rating scale for depression J Neurol Neurosurg Psychiatry 1960 23 56 62 14399272
Hamilton M The assessment of anxiety states by rating Br J Med Psychol 1959 32 50 55 13638508
Grant JE Kim SW Potenza MN Paroxetine treatment of pathological gambling: a multi centre randomized controlled trial Int Clin Psychopharmacol 2003 18 243 249 10.1097/00004850-200307000-00007 12817159
Zimmerman M Breen RB Posternak MA An open-label study of citalopram in the treatment of pathological gambling J Clin Psychiatry 2002 63 44 48 11838625
Pallanti S Rossi NB Sood E Nefazodone treatment of pathological gambling: a prospective open-label controlled trial J Clin Psychiatry 2002 63 1034 9 12444818
Black DW An open-label trial of Bupropion in the treatment of pathological gambling J Clin Psychopharmacol 2004 24 108 110 10.1097/01.jcp.0000114844.58996.4a 14709965
Phillips AG Lepiane FG Reinforcing effects of morphine microinjection onto the ventral tegmental area Pharmacol Biochem Behav 1980 12 965 968 10.1016/0091-3057(80)90460-8 7403209
van Wolfswinkel L van Ree JM Effects of morphine and naloxone on thresholds of ventral tegmental electrical self-stimulation Naunyn Schmiedebergs Arch Pharmacol 1985 330 84 92 10.1007/BF00499899 4047179
Stewart J Reinstatement of heroin and cocaine self-administration behavior in the rat by intracerebral application of morphine in the ventral tegmental area Pharmacol Biochem Behav 1984 20 917 923 10.1016/0091-3057(84)90017-0 6463075
Pei Q Zatterstrom T Leslie RA 5HT3 receptor antagonists inhibit morphine-induced stimulation of mesolimbic dopamine release and function in the rat Eur J Pharmacol 1993 230 63 68 10.1016/0014-2999(93)90410-J 8381355
Montgomery AM Rose IC Herberg LJ The effect of a 5HT3 receptor, ondansetron, on brain stimulation reward, and its interaction with direct and indirects stimulants of central dopaminergic transmission J Neural Trasm Gen Sect 1993 91 1 11 10.1007/BF01244914
Carboni E Acquas E Leone P 5HT3 receptor antagonists block morphine- and nicotine- but not amphetamine-induced reward Psychopharmacology 1989 97 175 178 10.1007/BF00442245 2498925
Higgins GA Joharchi N Nguyen P Effect of the 5HT3 receptor antagonists, MDL72222 and ondansetron on morphine place conditioning Psychopharmacology 1992 106 315 320 1533288
Acquas E Carboni E Garau L Blockade of acquisition of drug-conditioned place aversion by 5-HT3 antagonists Psychopharmacology 1990 100 459 463 2320706
Matthews RT German DC Electrophysiological evidence for excitation of rat ventral tegmental area dopamine neurons by morphine Neuroscience 1984 11 617 625 10.1016/0306-4522(84)90048-4 6717805
Broekkamp CL Phillips AG Facilitation of self-stimulation behavior following intracerebral microinjections of opioids into the ventral tegmental area Pharmacol Biochem Behav 1979 11 289 295 10.1016/0091-3057(79)90137-0 504308
Shinohara K Yanagisawa A Kagota I Physiological changes in Pachinko players; beta-endorphin, catecholamines, immune system substances and heart rate Applied Human Science 1999 18 37 42 10.2114/jpa.18.37 10388157
Volkow ND Fowler JS Wolf AP Changes in brain glucose metabolism in cocaine dependence and withdrawal Am J Psychiatry 1991 148 621 626 2018164
McGuire PK Bench CJ Marks IM Functional anatomy of obsessive compulsive phenomena Br J Psychiatry 1994 164 459 468 8038933
Crockford DN el-Guebaly N Naltrexone in the treatment of pathological gambling and alcohol dependence Can J Psychiatry 1998 43 86 9494755
Grant JE Kim SW An open label study of naltrexone in the treatment of kleptomania J Clin Psychiatry 2002 63 349 356 12000210
Keuler DJ Altemus M Michelson D Behavioral effects of naloxone infusion in obsessive-compulsive disorder Biol Psychiatry 1996 40 154 156 10.1016/0006-3223(95)00632-X 8793049
Roth AS Ostroff RB Hoffman RE Naltrexone as a treatment for repetitive self-injurious behavior: an open-label trial J Clin Psychiatry 1996 57 233 237 8666558
Volpicelli JR Alterman AI Hayashida M Naltrexone in the treatment of alcohol dependence Arch Gen Psychiatry 1992 49 876 880 1345133
Kim SW Opioid antagonists in the treatment of impulse-control disorders J Clin Psychiatry 1998 59 159 164 9590665
Kim SW Grant JE An open naltrexone treatment study in pathological gambling disorder Int Clin Psychopharmacol 2001 16 285 289 10.1097/00004850-200109000-00006 11552772
Christenson GA Popkin MK Mackenzie TB Lithium treatment of chronic hair pulling J Clin Psychiatry 1991 52 116 120 1900831
Donovan SJ Susser ES Nunes EV Divalproex treatment of disruptive adolescents: a report of 10 cases J Clin Psychiatry 1997 58 12 15 9055831
Donovan SJ Divalproex treatment for youth with esplosive temper, and mood lability: a double-blind, placebo-controlled crossover design Am J Psychiatry 2000 157 1038 10.1176/appi.ajp.157.5.818 10831513
Kavoussi RJ Coccaro EF Divalproex sodium for impulsive aggressive behavior in patients with personalità disorder J Clin Psychiatry 1998 59 676 680 9921702
Kmetz GF McElroy SL Collins DJ Response of kleptomania and mixed mania to valproate Am J Psychiatry 1997 154 580 581 9090355
Stein DJ Simeon D Frenkel M An open trial of valproate in borderline personality disorder J Clin Psychiatry 1995 56 506 510 7592502
Hollander E Allen A Lopez RP A preliminary double-blind, placebo-controlled trial of divalproex sodium in borderline personality disorder J Clin Psychiatry 2001 62 199 203 11305707
Hollander E Tracy KA Swann AC Divalproex in the treatment of impulsive aggression: efficacy in cluster B personality disorders Neuropsychopharmacol 2003 28 1186 1197
McElroy SL Arnold LM Shapira NA Topiramate in the treatemnt of binge eating disorder associated with obesity: a randomised placebo controlled trial Am J Psychiatry 2003 160 612 10.1176/appi.ajp.160.2.255
Haller R Hinterhuber H Treatment of Pathological gambling with carbamazepine Pharmacopsychiatry 1994 27 129 8078954
Moskowitz JA Lithium and lady luck; use of lithium carbonate in compulsive gambling N Y State J Med 1980 80 785 8 6930557
Pallanti S Quercioli L Sood E Lithium and Valproate treatment of pathological gambling: a randomized single-blind study J Clin Psychiatry 2002 63 559 564 12143910
First MB Spitzer RL Gibbon M Structured Clinical Interview for DSM-IV Axis I (SCID-I), Clinician Version 1997 Washington DC: American Psychiatric Press
Hollander E Pallanti S Allen A Does sustained-release lithium reduce impulsive gambling and affective instability versus placebo in pathological gamblers with bipolar spectrum disorders? Am J Psychiatry 2005 162 137 145 10.1176/appi.ajp.162.1.137 15625212
Dannon P Lowengrub K Gonopolsky Y Topiramate versus fluvoxamine in the treatment of pathological gambling: a randomized, blind-rater, comparison study Clin Neuropharmacol 2005 28 6 10 10.1097/01.wnf.0000152623.46474.07 15711432
Grant John A multicenter Investigation of the Opioid Antagonist Nalmefene in the Treatment of Pathological Gambling. [ABSTRACT] Presented at the 45th NCDEU Meeting, Boca Raton, Florida, June 6–9, 2005
|
16216125
|
PMC1283978
|
CC BY
|
2021-01-04 22:25:40
|
no
|
Clin Pract Epidemiol Ment Health. 2005 Oct 10; 1:21
|
utf-8
|
Clin Pract Epidemiol Ment Health
| 2,005 |
10.1186/1745-0179-1-21
|
oa_comm
|
==== Front
Health Qual Life OutcomesHealth and Quality of Life Outcomes1477-7525BioMed Central London 1477-7525-3-661625962710.1186/1477-7525-3-66ResearchThe Oxford hip score: the patient's perspective Wylde Vikki [email protected] Ian D [email protected] Victoria J [email protected] Academic Orthopaedic Unit, University of Bristol, Avon Orthopaedic Centre, Southmead Hospital, Westbury-on-Trym, Bristol BS10 5NB UK2005 31 10 2005 3 66 66 24 8 2005 31 10 2005 Copyright © 2005 Wylde et al; licensee BioMed Central Ltd.2005Wylde 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 the last 25 years, assessment of orthopaedic intervention has become patient focused, with the development of self-completion patient-centred outcome measures. The Oxford hip score (OHS) is a joint specific outcome measure tool designed to assess disability in patients undergoing total hip replacement (THR). Although the psychometric properties of the OHS have been rigorously examined, there is little research on the patient's perspective of the OHS. Therefore, the aim of this study is to assess whether the OHS is an adequate disability measure from the patient's perspective using qualitative analysis of annotations written on the OHS by patients.
Methods
In total, 276 orthopaedic patients completed an OHS between April 2004 and May 2005. One hundred and fifty six pre-operative patients listed for a THR completed the OHS during a pre-admission assessment clinic, and 120 post-operative patients completed the OHS postally in the home setting. Patient's unprompted annotations in response to the questions on the OHS were recorded and grouped into thematic categories.
Results
In total, 46 (17%) patients made 52 annotations when completing the OHS. These annotations identified five main areas of difficulty that patients experienced: lack of question clarity (particularly concerning the use of aids), difficulty in reporting measurements of pain, restrictive and irrelevant questions, the influence of co-morbidities on responses, and double-barrelled questions.
Conclusion
Although the OHS is a useful short tool for the assessment of disability in patients undergoing THR, this study identified several problem areas that are applicable to patient-centred outcome tools in general. To overcome these current limitations, further work is underway to develop a more individualised patient-centred outcome measure of disability for use in patients with osteoarthritis.
==== Body
Background
During the last decade, the assessment of outcomes in orthopaedic surgery has shifted from the success or failure of an implant towards patient satisfaction and quality of life [1]. Initially, surgeon assessment of total hip replacement (THR) outcome was accepted, with the development of tools such as the Harris Hip Score [2] and the Charnley score [3]. However, these measures presuppose a concordance between the views of patients and clinicians, which has been proved to be an erroneous assumption [4,5], particularly in subjective domains such as pain [6]. Consequently, the last 25 years has witnessed the development of generic and disease-specific self-completion patient-centred outcome measures. Generic measures such as the SF-12 [7] and Nottingham Health Profile [8] endeavour to assess all important dimensions of health-related quality of life [9]. Disease-specific tools such as the Arthritis Impact Measurement Scale (AIMS) [10] and the Western Ontario and McMaster University Osteoarthritis Index (WOMAC) [11] focus on specific aspects of disability relating to a particular condition. These are supplemented by joint specific measures such as the Oxford hip score (OHS) [12] and the Hip Disability and Osteoarthritis Outcome Score (HOOS) [13].
The OHS is a patient-centred questionnaire that is designed to assess functional ability and pain from the patient's perspective. It is a short, twelve-item questionnaire developed for completion by patients undergoing THR [12] and is extensively referenced in the orthopaedic literature [14-21]. The OHS has been demonstrated to be highly sensitive to change in patients undergoing primary THR [12,16,21,19,22] and revision THR [15,16]. It correlates well with patient satisfaction [15,19] and other patient-centred instruments, such as the Euroqol 5D [15]. Responsiveness of the OHS to change has been found to be greater than generic measures such as the SF-36 [16,18] and disease specific measures such as the WOMAC [21]. The OHS has been utilized in a broad range of contexts, including studies comparing different prostheses [14], surgeon and patient expectations [20], and the outcomes of NHS and private patients [17].
Although the OHS has been shown to have internal consistency and produce data of high reliability and validity [12], there is a shortage of published data on the patient's perception of the OHS. During the validation of the questionnaire, there was no reference to difficulties that patients experienced when completing the OHS, beyond a brief statement that "the patients had little difficulty in completing it" [12]. Previous research has explored patient's perception of the OHS, and found that patients encountered several limitations of the OHS relating to question specificity, response category clarity, exclusion of comorbidities, and experience of pain [23]. However, this study was limited to a small sample size and during the past half decade no further work has been published investigating the patient's experience of the OHS. Therefore, the aim of this study was to determine, from the patient's perspective, if the OHS is an adequate questionnaire for measuring disability. This was achieved by analysing unprompted, spontaneous annotations generated by patients completing a paper copy of the OHS.
Methods
Between April 2004 and May 2005 patients attending orthopaedic preadmission assessment clinic at the Avon Orthopaedic Centre, under the care of one consultant orthopaedic surgeon (IDL), and awaiting THR, were administered an OHS as part of a routine questionnaire pack used in the clinic. These patients were sampled as they were expected to be unfamiliar with the OHS, as the introduction of this questionnaire into routine clinical care in this clinic was initiated in April 2004. Between January 2005 and May 2005, consecutive patients with 12-months follow-up, who received a THR under the care of a consultant orthopaedic surgeon (IDL), completed a postally administered OHS as part of their on-going clinical care. In addition, all patients that had an IPS Stem (DePuy) between 1997 and 2004, under the care of one consultant orthopaedic surgeon (IDL) complete a postal OHS as part of another study. During administration the patients were not instructed to annotate or comment on the questions on the OHS.
The OHS consists of 12 questions about pain and disability experienced over the past four weeks. Each item has five response categories, given a score of between 1–5 (low disability to high disability). Scoring involves summating the total for each item to produce a final score between 12–60, with a higher score indicating greater disability. In this study the OHS was presented on a double-sided sheet of A4 paper, with six questions on each side. The response categories to each question were formatted as a Likert scale (Figure 1) with the coding frame integrated into the questionnaire. Both the questionnaires administered in the pre-admission clinic and the postal questionnaires were formatted in an identical manner.
Figure 1 The format of the Oxford hip score.
Each questionnaire was reviewed for spontaneously generated annotations and these annotations were then grouped into thematic categories.
Results
In total, 276 patients completed the OHS questionnaire. In pre-admission assessment clinic, 156 consecutive patients, listed for a THR, completed an OHS. For patients that attended the clinic twice between April 2004 and May 2005, only the first questionnaire was included in the analysis. Post-operatively, 120 patients completed a postally administered OHS. The pre-operative mean OHS was 44.1 (SD 8.3, range 21–59) and the mean post-operative OHS was 24.1 (SD 11.3, range 12–55). Patients who completed the OHS after surgery had a mean follow-up period of 24 months (SD 19, range 12–77 months). The sample consisted of 169 women (61%) and 107 men (39%) with a mean age of 58 years (SD 15.9, range 14–82 years). Patient's diagnoses are listed in Table 1.
Table 1 Diagnosis of patients who completed the Oxford hip score
Diagnosis Number of patients % of patients
Osteoarthritis 211 76%
Development hip dysplasia 33 12%
Avascular necrosis 15 5%
Juvenile chronic arthritis 5 2%
Rheumatoid arthritis 4 1%
Ankylosing spondylosis 3 1%
Hip fracture 3 1%
Psoriatic arthritis 2 1%
Forty-six (17%) patients annotated a total of 52 questions (Table 2). Five patients drew 16 arrows linking boxes, signalling that they felt they were unable to place themselves in a single category provided by the OHS. Question six, which asks "During the past 4 weeks, for how long have you been able to walk before pain from your hip becomes severe (with or without a stick)?" most frequently elicited annotation, whereas question eleven, which asks "how much has pain from your hip interfered with your usual work (including housework)?" was the only question that was not annotated by any of the patients. The annotations were broadly grouped into five main categories, each highlighting difficulties the patients experienced when completing the OHS (Table 3).
Table 2 Annotations on the Oxford hip score
Question number Annotations (n) Total number of annotations
1 Pain not constant in intensity (5), depends of medication (2) 7
2 Depends on part of body (2), due to other co-morbidities 3
3 More difficulty using public transport (4) depends on which side of the car, uses adapted taxi 6
4 Difficulty with and without an aid (4), depends whether it is socks, tights or stockings (2), Due to other co-morbidities 7
5 Due to other co-morbidities 1
6 Description of pain (5), difficulty with and without a stick/crutches (5), due to other co-morbidities (4), pain not constant in intensity, depends on medication (2) 17
7 Description of how stairs are climbed (4), uses stair lift 5
8 Pain not constant in intensity 1
9 In a wheelchair, reason for limp 2
10 Pain not constant in intensity, causes of pain 2
11 -------------------------------------------------------------------------------------- 0
12 Causes of pain 1
Table 3 Categories of annotations made by patients on the Oxford hip score
Category of annotation Purpose of annotation n (%) annotations
Question clarity To expand and explain answers 15 (29)
Measurement of pain To explain nature of pain 12 (23)
Restrictive and irrelevant questions To describe pain and alterations to activities, and comment on non applicable questions 12 (23)
Co-morbidities To explain influence of co-morbidities on answer 7 (13)
Double-barrelled questions To give two or more answers to a single question 6 (12)
Discussion
The mean pre-operative OHS of 44.1 and post-operative score of 24.1 are similar to previous results [12,16,19], indicating that the sample in this study was representative of other lower limb orthopaedic patients. The pre-operative and the post-operative groups were purposively sampled as separate cohorts to avoid familiarly with the OHS, which could comprise the validity of the results. As the completion of the OHS was only introduced into this clinic in April 2004, the postal OHS completed by the post-operative patients was likely to be their initial contact with the questionnaire. Similarly, the patients attending the pre-admission assessment clinic should not have previously encountered the questionnaire. However, a limitation of the study was that patients may have previously completed the OHS for their GP or under the care of a different consultant, and this prior exposure to the OHS may have influenced the patient's responses.
This study has highlighted several pitfalls and limitations of the OHS, and of available disability measures in general. However, although the current study identified substantial areas of difficulty, analysing unprompted annotations has limitations. The results are confined to the difficulties encountered by individuals who were self-motivated to comment upon these problems. As a result of this methodology, conclusions are drawn from the responses of only 17% of the patients sampled. For the remaining 83% of patients, the OHS could have been adequate from their perspective or alternatively, they could have encountered problems, but not have documented them on paper as they were not instructed to do so. Therefore, further research needs to be undertaken, in which patient are explicitly encourage to comment upon any difficulties when completing the OHS, in order to assess the extent of it's applicability. Alternatively, qualitative interviews could be employed to explore the patient's perspective on the OHS in greater depth, although findings from qualitative work have raised similar areas of difficulty to those in the current study [23].
The five general themes of difficulties that emerged from the analysis of annotations is discussed in more detail below.
Seventeen percent of patients annotated answers they provided on the OHS, suggesting that the patients felt that the questions were inadequate to suitably express themselves. Five general themes emerged from the analysis of annotations and each thematic category is discussed in more detail below.
Question clarity
The aspect of the OHS that appeared to cause the greatest difficulty for the patients, with 29% of annotations, was the lack of question clarity. Within this theme, the predominant area of uncertainty was whether the questions were enquiring about actual level of disability or the level of disability after accounting for the use of aids or specialised devices, such as long handled shoehorns or helping hands. When responding to question four, which asks respondents "have you been able to put on a pair of socks, stocking or tights?", a number of individuals answered accounting for the use of an aid, and other people gave two answers; one referring to the level of disability in performing the activity when using an aid and one when not using an aid. The same lack of clarity has resulted from this question previously [23]. Question six, which asks the respondent "long have you been able to walk before pain from your hip becomes severe (with or without a stick)?", acknowledges that many individuals need to use a walking stick. However, it is not specified in the question whether the patients should provide a response for actual or relative disability. Consequently, inconsistent results were obtained, with patients providing two answers i.e. distance walked with and without a walking stick. Therefore, the score becomes dependant on whether the respondent chooses to take account of the walking stick. These findings suggest that many respondents perceive the question as ambiguous. Further evidence for the lack of question clarity is based upon a large study of pre-operative patients, who most frequently omitted question six when completing the OHS [19].
In summary, it appears that the predominant area of ambiguity due to lack of question clarity on the OHS is whether patients should take into consideration the use of aids or specialised devices when responding to questions. Not taking consideration of the use of aids and devices, and indeed any assistance in activities, is a common oversight of many patient-centred measures of disability, such as the WOMAC [11]. Individuals who take into consideration the use of an aid when answering a question will appear less disabled than they are in reality. This lack of clarity could confound results, resulting in patients with the highest level of disability, who utilize specialised equipment in many activities, appearing to be the least disabled on paper. To enhance question clarity and gain consistent results it would appear advisable to specify to patients whether they should account for the use of aids or devices when responding to the question. However, modification of validated outcome measures can be fraught with problems [24], and therefore it may be more advisable to use an outcome tool that considers the modifying effect of aids and assistance on disability.
Measurement of pain
Nearly a quarter of all the annotations provided an explanation of the nature of pain. Frequently patients commented that the intensity of pain can fluctuate greatly over four weeks and that the level of pain is heavily dependant on factors such as medication and activity. As a consequence, several patients felt they could not give an 'average' level of pain for the last four weeks. Therefore, a limitation of the OHS is that it attempts to categorise patients into a single category of pain when in fact pain, predominantly arthritic pain, is not static, but rather a dynamic entity. In a previous study, when interviewed about difficulties encountered when completing the OHS, individuals explained that they learned to ignore the pain, and that it could be masked by medication, and as a consequence struggled to complete the questions referring to pain [23]. Thus questions relating to 'average' pain appear inadequate to capture the experience of individuals with arthritic pain.
Restrictive and irrelevant questions
Twenty three percent of annotations were descriptive or explanatory comments, supplementing the information recorded by the question. These annotations included descriptions of pain or how activities had to be modified as a consequence of disability, such as climbing stairs backwards, and the causes of pain. Furthermore, inadequate response categories resulted in 16 arrows being drawn between boxes, indicating patients were unable to place themselves into a single category. The original article on the OHS does not indicate how these responses should be scored [12]. Although it has recently been suggested that the highest score should be used, it may be argued that this is not a true reflection of the patient's answer and the clinician is introducing bias by selecting which answer to accept [24].
Expansion of answers was necessary for several patients to explain that, although they had answered the question, it was not applicable to them. Comments written in response to question seven, which asks "have you been able to climb a flight of stairs?", suggest that climbing stairs is not applicable to everyone as some individuals have stair lifts installed or they live in a bungalow. In reply to question nine, "have you been limping when walking because of your hip?", a respondent answered that they don't limp but explained this was a result of them being confined to a wheelchair. Although the questionnaire accounts for people that cannot drive by asking about difficulty travelling by public transport in question three, this question was not applicable to a patient who used an adapted taxi. The OHS appears to restrict individual's answers and fails to allow them to express themselves adequately, as well as including questions that are not relevant to all individuals.
Co-morbidities
The OHS was designed as a site-specific outcome measure for orthopaedic evaluation, and as such, has been favoured over more generic outcome measures [16,21]. However, an underlying theme in the annotations was the difficulties that patients encountered when attempting to separate the disability and pain resulting from the affected hip from that arising from other co-morbidities. Contrary to Dawson and colleagues finding that the OHS is not influenced by co-morbidities [22], the effect size of the OHS has been found to be substantially smaller in patients with other mobility limiting conditions, compared with patients with unilateral hip osteoarthritis (OA), suggesting that other co-morbidities do influence the OHS [21]. The Oxford knee score, which has a comparable format to the OHS, produced similar results for patients with and without knee pain, in the presence of other co-morbidities, providing evidence that the questionnaire is not joint specific [25]. Patients with consistently high scores on the OHS have been found to suffer from multiple co-morbidities [18] and patients have verbalised that they find it difficult to separate pain from their hip from pain arising from other sites [23]. Therefore, co-morbidities appear to compromise the specificity of the OHS in evaluating disability resulting from hip symptoms, although joint specific questionnaires are designed to exclude the effects of co-morbidities.
In addition to the influence of diffuse co-morbidities, patients found it difficult to distinguish between pain originating from bilateral hips, highlighting a limitation of the OHS in considering only a single joint, which does not reflect the pattern of OA. In a sample of 500 OA patients, 53% of patients had more than one symptomatic joint [26]. Recently, this issue has been addressed by the modification of the OHS to ask about bilateral hip joints, although the success of this new design is questionable as 41% of the patients completed the OHS for the operated side only and 12% of patients did not discriminate between the two joints [24].
Double-barrelled questions
During the validation process it is advisable to eliminate any double-barrelled questions [27], yet question three asks two questions in one: "have you had any trouble getting in and out of a car or using public transport because of your hip?". Several patients answered the two parts of the question separately as it is common to use both modes of transport. Similarly, question four asks three questions in one: "have you been able to put on a pair of socks, stocking or tights?". Again some patients answered this as three questions, with women often finding tights harder to put on than socks.
Conclusion
The OHS is a useful short tool that is frequently utilised to assess the patient's perception of hip function, mobility and pain. It is quick both for the patient to complete and the clinician to score. Although the OHS is a widely used and validated patient-centred outcome tool, it appears that the OHS is not without problems, in concordance with previous findings [23]. It is unclear to patients whether the questions are asking about level of disability before or after accounting for the use of aids and devices. Individuals found it difficult to respond to questions about the severity of their symptoms due to the dynamic nature of pain and the use of medications to mask the pain. They also had difficulty separating other co-morbidities from the symptoms of the affected hip. Also double-barrelled questions caused confusion and not all questions on the OHS were relevant, or important, to the patient.
It could be argued that the difficulties patients experience with the OHS are due to the brevity of the scale, and could be reduced by the inclusion of additional questions. However, although there is little research on the problems experienced by patients while completing longer scales, such as the WOMAC [11] or HOOS [13], it appears that the limitations highlighted in the OHS could be applied to these longer questionnaires. The WOMAC does not account for the use of aids or devices, includes questions asking patients about their average pain level over the past 4 weeks, and has double barrelled questions such as "what degree of difficulty do you have with getting in/out of bath/shower?". In addition, the WOMAC items have been found to be influenced by other co-morbidities, such as low back pain [28]. Previous research has found that the items on the WOMAC are unimportant, or irrelevant, to some individuals with OA [13]. This latter limitation is applicable to many validated patient-centred outcome measures. No single activity is important to all individuals, nor is the importance of being able to perform that activity necessarily stable over time [29]. Hence, an ideal would be to weight items of disability with the importance of performing that activity. This would allow non-applicable items to be rated as of no importance and thus not contribute to the score, producing a more individualised patient-centred outcome measure. Further work is underway to develop a personal impact of disability in osteoarthritis.
Authors' contributions
VW was involved in the acquisition, analysis and interpretation of the data, and drafted the manuscript
IDL was involved in the conception of the study, revision of the manuscript and gave final approval of the version to be published
VJC was involved in the conception and design of the study and revision of the manuscript
Acknowledgements
We would like to give thanks to the staff and patients at the Avon Orthopaedic Centre for their cooperation in this study.
No external funding was received for this study.
==== Refs
Katz JN Phillips CB Poss R Harrast JJ Fossel AH Liang MH Sledge CB The validity and reliability of a Total Hip Arthroplasty Outcome evaluation questionnaire J Bone Joint Surg Am 1995 77 1528 1534 7593061
Harris WH Traumatic arthritis of the hip after dislocation and acetabular fractures: treatment by mold arthroplasty. An end-result study using a new method of result evaluation J Bone Joint Surg Am 1969 51 737 755 5783851
Charnley J The long-term results of low-friction arthroplasty of the hip performed as a primary intervention J Bone Joint Surg Br 1972 54 61 76 5011747
Hewlett SA Patients and clinicians have different perspectives on outcomes in arthritis J Rheumatol 2003 30 877 879 12672220
Rothwell PM McDowell Z Wong CK Dorman PJ Doctors and patients don't agree: cross sectional study of patients' and doctors' perceptions and assessments of disability in multiple sclerosis BMJ 1997 314 1580 1583 9169401
Janse AJ Gemke RJ Uiterwaal CS van der Tweel I Kimpen JL Sinnema G Quality of life: patients and doctors don't always agree: a meta-analysis J Clin Epidemiol 2004 57 653 661 15358393 10.1016/j.jclinepi.2003.11.013
Ware J JrKosinski M Keller SD A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity Med Care 1996 34 220 233 8628042 10.1097/00005650-199603000-00003
Hunt SM McKenna SP McEwen J Backett EM Williams J Papp E A quantitative approach to perceived health status: a validation study J Epidemiol Community Health 1980 34 281 286 7241028
Guyatt G A taxonomy of health status instruments J Rheumatol 1995 22 1188 1190 7674253
Meenan RF Gertman PM Mason JH Measuring health status in arthritis. The arthritis impact measurement scales Arthritis Rheum 1980 23 146 152 7362665
Bellamy N Buchanan WW Goldsmith CH Campbell J Stitt LW Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee J Rheumatol 1988 15 1833 1840 3068365
Dawson J Fitzpatrick R Carr A Murray D Questionnaire on the perceptions of patients about total hip replacement J Bone Joint Surg Br 1996 78 185 190 8666621
Nilsdotter AK Lohmander LS Klassbo M Roos EM Hip disability and osteoarthritis outcome score (HOOS) – validity and responsiveness in total hip replacement BMC Musculoskelet Disord 2003 4 10 12777182 10.1186/1471-2474-4-10
Dawson J Jameson-Shortall E Emerton M Flynn J Smith P Gundle R Murray D Issues relating to long-term follow-up in hip arthroplasty surgery: a review of 598 cases at 7 years comparing 2 prostheses using revision rates, survival analysis and patient-based measures J Arthroplast 2000 15 710 717 10.1054/arth.2000.7109
Dawson J Fitzpatrick R Frost S Gundle R McLardy-Smith P Murray D Evidence for the validity of a patient-based instrument for assessment of outcome after revision hip replacement J Bone Joint Surg Br 2001 83 1125 1129 11764424 10.1302/0301-620X.83B8.11643
Dawson J Fitzpatrick R Murray D Carr A Comparison of measures to assess outcomes in total hip replacement surgery Qual Health Care 1996 5 81 88 10158596
Field RE Cronin MD Singh PJ The Oxford hip scores for primary and revision hip replacement J Bone Joint Surg Br 2005 87 618 622 15855360 10.1302/0301-620X.87B5.15390
Fitzpatrick R Dawson J Health-related quality of life and the assessment of outcomes of total hip replacement surgery Health Psychol 1997 12 793 803
Fitzpatrick R Morris R Hajat S Reeves B Murray DW Hannen D Rigge M Williams O Gregg P The value of short and simple measures to assess outcomes for patients of total hip replacement surgery Qual Health Care 2000 9 146 150 10980074 10.1136/qhc.9.3.146
Moran M Khan A Sochart DH Andrew G Expect the best, prepare for the worst: surgeon and patient expectation of the outcome of primary total hip and knee replacement Ann R Coll Surg Engl 2003 85 204 206 12831497 10.1308/003588403321661415
Ostendorf M van Stel HF Buskens E Schrijvers AJ Marting LN Verbour AJ Dhert WJ Patient-reported outcome in total hip replacement. A comparison of five instruments of health status J Bone Joint Surg B 2004 86 801 808 10.1302/0301-620X.86B6.14950
Dawson J Fitzpatrick R Murray D Carr A The problem of 'noise' in monitoring patient-based outcomes: generic, disease-specific and site-specific instruments for total hip replacement J Health Serv Res Polic 1996 1 224 231
McMurray R Heaton J Sloper P Nettleton S Measurement of patient perceptions of pain and disability in relation to total hip replacement: the place of the Oxford hip score in mixed methods Qual Health Care 1999 8 228 233 10847884
Pynsent PB Adams DJ Disney SP The Oxford hip and knee outcome questionnaires for arthroplasty J Bone Joint Surg Br 2005 87 241 248 15736751 10.1302/0301-620X.87B2.15095
Harcourt WG White SH Jones P Specificity of the Oxford Knee status questionnaire. The effect of disease of the hip or lumbar spine on patients' perception of knee disability J Bone Joint Surg Br 2001 83 345 347 11341417 10.1302/0301-620X.83B3.11298
Cushnaghan J Dieppe P Study of 500 patients with limb joint osteoarthritis. I. Analysis by age, sex, and distribution of symptomatic joint sites Ann Rheum Dis 1991 50 8 13 1994877
Streiner DL Norman GR Selecting the items Health Measurement Scales: A Practical Guide to Their Development and Use 1995 Oxford: Oxford University Press 54 66
Wolfe F Determinants of WOMAC function, pain and stiffness scores: evidence for the role of low back pain, symptom counts, fatigue and depression in osteoarthritis, rheumatoid arthritis and fibromyalgia Rheumatology (Oxford) 1999 38 355 61 10378714 10.1093/rheumatology/38.4.355
O'Boyle CA McGee H Hickey A O'Malley K Joyce CR Individual quality of life in patients undergoing hip replacement Lancet 1992 339 1088 1091 1349111 10.1016/0140-6736(92)90673-Q
|
16259627
|
PMC1283979
|
CC BY
|
2021-01-04 16:38:13
|
no
|
Health Qual Life Outcomes. 2005 Oct 31; 3:66
|
utf-8
|
Health Qual Life Outcomes
| 2,005 |
10.1186/1477-7525-3-66
|
oa_comm
|
==== Front
Mol PainMolecular Pain1744-8069BioMed Central London 1744-8069-1-311624204710.1186/1744-8069-1-31ResearchContrasting phenotypes of putative proprioceptive and nociceptive trigeminal neurons innervating jaw muscle in rat Connor Mark [email protected] Ligia A [email protected] Edwin W [email protected] Vollum Institute, Oregon Health & Sciences University, Portland, Oregon, USA2 Pain Management Research Institute, Kolling Institute, University of Sydney at Royal North Shore Hospital E25, St Leonards, NSW 2065, Australia3 Department of Physiology and Biophysic, Federal University of Minas Gerais, Belo Horizonte, Brazil2005 24 10 2005 1 31 31 6 9 2005 24 10 2005 Copyright © 2005 Connor et al; licensee BioMed Central Ltd.2005Connor 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
Despite the clinical significance of muscle pain, and the extensive investigation of the properties of muscle afferent fibers, there has been little study of the ion channels on sensory neurons that innervate muscle. In this study, we have fluorescently tagged sensory neurons that innervate the masseter muscle, which is unique because cell bodies for its muscle spindles are in a brainstem nucleus (mesencephalic nucleus of the 5th cranial nerve, MeV) while all its other sensory afferents are in the trigeminal ganglion (TG). We examine the hypothesis that certain molecules proposed to be used selectively by nociceptors fail to express on muscle spindles afferents but appear on other afferents from the same muscle.
Results
MeV muscle afferents perfectly fit expectations of cells with a non-nociceptive sensory modality: Opiates failed to inhibit calcium channel currents (ICa) in 90% of MeV neurons, although ICa were inhibited by GABAB receptor activation. All MeV afferents had brief (1 msec) action potentials driven solely by tetrodotoxin (TTX)-sensitive Na channels and no MeV afferent expressed either of three ion channels (TRPV1, P2X3, and ASIC3) thought to be transducers for nociceptive stimuli, although they did express other ATP and acid-sensing channels. Trigeminal masseter afferents were much more diverse. Virtually all of them expressed at least one, and often several, of the three putative nociceptive transducer channels, but the mix varied from cell to cell. Calcium currents in 80% of the neurons were measurably inhibited by μ-opioids, but the extent of inhibition varied greatly. Almost all TG masseter afferents expressed some TTX-insensitive sodium currents, but the amount compared to TTX sensitive sodium current varied, as did the duration of action potentials.
Conclusion
Most masseter muscle afferents that are not muscle spindle afferents express molecules that are considered characteristic of nociceptors, but these putative muscle nociceptors are molecularly diverse. This heterogeneity may reflect the mixture of metabosensitive afferents which can also signal noxious stimuli and purely nociceptive afferents characteristic of muscle.
==== Body
Background
The masseter muscle is involved in many painful conditions which are grouped under the general heading of temporomandibular disorders [1]. Although a role of primary afferents innervating the masseter muscle in the development and maintenance of some of these pain states has been generally accepted, the cellular properties of masseter afferents have not been extensively investigated [2]. The trigeminal sensory system is unusual because the cell bodies of the trigeminal primary afferent neurons are located both in the trigeminal ganglion (TG) and in the mesencephalic nucleus of the 5th cranial nerve (MeV) in the brainstem. The proprioceptive afferents located in the MeV include those arising from the masseter muscle [3], and this unique anatomical segregation enables direct comparison of the expression of proteins involved sensory transduction and its modulation between proprioceptive and other muscle afferents.
We have previously compared the types of P2X receptor and acid gated ion channel (ASIC)-mediated response between nociceptive and non-nociceptive sensory afferents [4,5], utilizing the MeV proprioceptors as the non-nociceptive neuronal population and either tooth pulp or cardiac afferents as the exemplar nociceptors. However, we have not previously compared the properties of different types of sensory afferent innervating the same tissue. We hypothesize that afferents from the masseter muscle that have different sensory functions will have different complements of signal detection and transduction molecules from each other, and that these different molecular profiles in part determine the response properties of the neurons to sensory stimuli. to In this study we have investigated the expression of sensory transduction-related ion channels and modulatory receptors in masseter afferent neurons located in the TG, which presumably represent a population of sensory neurons with mixed sensory modalities, and the proprioceptive masseter afferents located in the MeV. We determined the responses of these cell populations to the sensory mediators ATP, capsaicin and acid and potential analgesic agents such as μ-opioid agonists and baclofen [6,7]. There were significant differences between TG masseter afferents and MeV afferents in their voltage-gated sodium current (INa) expression, responses to sensory mediators and μ opioid agonists.
Results
We made recordings from 143 TG neurons and 31 brainstem MeV neurons labeled by injection of DiI into the masseter muscle. The diameter of the labeled ganglion cells ranged between 13 μm and 50 μm, the diameter of the cells isolated from the MeV nucleus ranged between 29 μm and 60 μm. We also recorded currents from 70 unlabeled TG neurons, with diameters between 12 μm and 50 μm.
Action Potentials and Sodium Currents
Action potentials in masseter afferent neurons were examined in whole cell current clamp recordings (Figure 1). Neurons were held at -70 mV and sufficient current injected (0.5–3 ms duration) to elicit an action potential. Action potential duration was measured at 0 mV. Action potentials elicited from masseter afferents isolated from the TG varied considerably in width, with durations between 0.6 ms and 5.2 ms at 0 mV (average duration 2.2 ± 0.2 ms, n = 27, Figure 1B). By contrast, MeV masseter afferent action potentials were narrower and displayed much less variability with a mean AP duration of 0.8 ± 0.05 ms (range 0.6 ms to 1.1 ms, n = 13, Figure 1B).
Figure 1 Contrasting action potential characteristics of masseter muscle sensory afferents. Action potentials were elicited by injecting a brief depolarizing current step to neurons held at -70 mV and the width measured at 0 mV. A(i) illustrates 2 example action potentials from trigeminal ganglion masseter afferents while A(ii) illustrates an action potential from a typical MeV nucleus masseter afferent. Action potentials widths are plotted against cell body diameter in B, note the wide range of action potentials widths in masseter afferents isolated from the trigeminal ganglion contrasting with the very tightly grouped action potentials widths from MeV masseter afferents.
The types of voltage-dependent sodium channel currents (INa) contributing to the action potentials in masseter afferents were examined using conventional voltage clamp recordings made in low extracellular Na (20 mM or 40 mM) in order to minimize the size of evoked INa. The peak INa was determined by stepping the neurons from a holding potential of -90 mV to potentials between -100 mV and +45 mV. Even in conditions of reduced external Na, the peak amplitude of evoked currents was substantial; peak INa averaged 11 ± 3 nA in ganglion afferents (n = 27) and 22 ± 4 nA in masseter afferents isolated from the MeV nucleus (n = 15). Application of tetrodotoxin (TTX, 300 nM), a potent inhibitor of many types of INa, inhibited the peak INa in ganglion masseter afferents in a highly variable manner (Figure 2). The average inhibition by TTX was 38 ± 6% (n = 27) but the inhibition ranged in individual cells from nothing to 100%. In contrast, TTX completely inhibited the peak INa in masseter afferents isolated from the MeV nucleus (99 ± 1%, n = 15, Figure 2).
Figure 2 Contrasting sodium currents in masseter afferents. MeV masseter afferents express only TTX-sensitive INa, but trigeminal ganglion masseter afferents display great variation in the proportion of TTX-sensitive to TTX-resistant INa among cells. INa were elicited by voltage steps from a holding potential of -90 mV, as diagrammed next to each set of traces. TTX-sensitive INa were obtained by digitally subtracting the INa remaining in TTX (300 nM) from the total INa. A trigeminal ganglion masseter afferent is illustrated in A; (i) illustrates the TTX-sensitive and TTX-resistant INa at -20 mV, a potential where TTX-sensitive current predominates while (ii) shows the total INa and TTX-insensitive INa at the potential where TTX-sensitive INa is maximal. The peak amplitudes of the total INa, TTX-sensitive and -resistant INa are plotted in (iii). A typical MeV masseter afferent is illustrated in B, (i) illustrates the INa and TTX-insensitive inward current at the test potential where INa is maximal; (ii) shows the TTX-sensitive INa over a range of test potentials. The peak amplitudes of the total INa, TTX-sensitive and residual current (ICa, see Table 1) are plotted in (iii). C) Illustrates the proportion of the peak inward current in TG and MeV masseter afferents that was sensitive to TTX (300 nM). The proportion of TTX-sensitive INa for the cell illustrated in A) would be calculated at a test potential of +5 mV.
Modulation of ICa
The calcium channel current (ICa) density of neurons was determined by repetitively stepping the membrane potential from a holding potential of -90 mV to test potentials between -60 and +60 mV. The ICa density of masseter afferents was similar to that of unlabeled cells (115 ± 6 pA/pF, n = 104 vs 109 ± 9 pA/pF, n = 53). The calcium channel density of masseter afferents isolated from the MeV nucleus was 36 ± 8 pA/pF (n = 12).
The sensitivity of the masseter afferents to opioids was determined by examining the modulation of ICa by agonists selective for μ-, δ- and κ-opioid receptors and the nociceptin receptor ORL1 (Figure 3). Cells were considered responsive if there was a reversible inhibition of the ICa evoked by step from -90 mV to 0 mV of at least 10%. Maximally effective concentrations of the μ-opioid agonist DAMGO (3 μM–10 μM, [8]) inhibited ICa in 74 of 92 labeled TG neurons (80%), and in 29 of 49 unlabeled neurons assessed at the same time (59%; P < 0.01, χ 2). In responsive masseter afferents DAMGO (3–10 μM) inhibited ICa by 46 ± 3%, in responsive unlabeled cells the inhibition of ICa was 39 ± 4%. We have previously arbitrarily divided trigeminal neurons into small (diameter <30 μm), intermediate (diameter between 30 μm–40 μm) and large (diameter >40 μm) cells [8]. DAMGO inhibited ICa in 85% of small masseter afferents, in 73% of intermediate diameter cells and in 82% of large cells (Figure 3C). The inhibition of ICa by DAMGO was 48 ± 4% in small masseter afferents, 50 ± 4% in intermediate cells and 36 ± 4% in large masseter afferents. Neither the κ-opioid agonist U69-593 (3 μM, n = 10) or the δ-opioid agonist deltorphin II (n = 13) inhibited ICa in masseter afferents. U69-593 inhibited ICa in 1 of 11 unlabeled cells tested (by 57%), deltorphin II did not inhibit ICa in any unlabeled cells (n = 11). The ORL1 agonist nociceptin (300 nM-1 μM) inhibited ICa in 6/15 (40%) of masseter afferents (inhibition of ICa was 41 ± 9%) and in 5/12 (42%) of unlabeled cells (inhibition of ICa was 38 ± 12%). The GABAB receptor agonist baclofen inhibited ICa in most ganglion masseter afferents (31/35; 89%, inhibition of ICa was 26 ± 4%) and unlabeled neurons (15/20; 75%, inhibition of ICa was 24 ± 4%).
Figure 3 Contrasting opioid modulation of masseter afferent ICa. ICa were elicited by stepping from a holding potential of -90 mV to test potential of 0 mV. A) (i) Example ICa traces from a trigeminal ganglion masseter afferent recorded in control conditions and in the presence of the μ-opioid agonist DAMGO and the GABA-B receptor agonist baclofen. (ii) A timeplot of the ICa amplitude at 0 mV for the cell illustrated in (i). Drugs were applied for the duration of the bars. B) (i) Example ICa traces from a MeV masseter afferent recorded in control conditions and in the presence of the μ-opioid agonist DAMGO the GABA-B receptor agonist baclofen and the ICa blocker Cd2+. (ii) A timeplot of the ICa amplitude at 0 mV for the cell illustrated in (i). Drugs were applied for the duration of the bars. C) The response of masseter afferents to a high concentration of DAMGO (3 or 10 μM) is plotted. A reversible inhibition of ICa of 10% or greater was considered a significant response. Note that cells of all sizes responded to DAMGO, although only 1 MeV masseter afferent was DAMGO sensitive. All MeV masseter afferents were sensitive to baclofen.
We have previously reported that somatostatin receptors preferentially inhibit ICa in large tooth pulp nociceptors when compared with small nociceptors [8]. Somatostatin inhibited ICa in 11/39 labelled masseter afferents and 3/15 unlabelled afferents. Somatostatin inhibited ICa in 5/19 small masseter afferents and 3/7 large masseter afferents (P = 0.4, χ2). The vast majority of masseter afferents that did (8/11) or did not (22/28) respond to SRIF also responded to DAMGO.
DAMGO (3 μM) inhibited ICa in only 1/11 MeV masseter afferents examined (by 18%). Baclofen (30 μM–100 μM) inhibited ICa in all MeV masseter afferents examined, by an average of 33 ± 2% (n = 10, Figure 3B).
Nociceptive Channels
TRPV1
The sensitivity of trigeminal neurons to the TRPV1 agonist capsaicin (3 μM) was determined by superfusing capsaicin onto neurons voltage clamped at -70 mV. Neurons with reversible capsaicin-induced currents of greater than 500 pA were considered to be sensitive. Many masseter afferents (65/137; 47%) and most unlabeled neurons (37/62; 60%) responded to capsaicin, with inward currents of up to 30 nA. Capsaicin produced inward currents in cells of all diameters, from 12 μm to 50 μm. Masseter afferents isolated from the MeV nucleus did not respond to capsaicin (n = 8).
ATP
The sensitivity of trigeminal neurons to ATP (50 μM) was determined by a 500 ms application of neurotransmitter to neurons voltage clamped at -70 mV. Neurons with reversible ATP-induced currents of greater than 500 pA were considered to be sensitive. Most masseter afferents (40/60; 66%) and unlabeled neurons (12/17; 70%) isolated from the TG were sensitive to ATP, with currents of up to 9500 pA. Masseter afferent responses to by ATP application were kinetically diverse, with many cells displaying a rapidly activating current that was more than 90% inactivated after 500 ms (19/40; 48%), or a mixture of rapidly activating and sustained currents (15/40; 38%). Some cells displayed slowly activating currents that did not significantly inactivate at the end of the 500 ms pulse (6/40; 15%). Masseter afferents isolated from the MeV displayed small sustained currents in response to ATP application (average amplitude of 150 ± 36 pA, n = 8).
Acid
The sensitivity of masseter afferents to extracellular acidification was assessed by varying the pH of the external solution from pH 7.4 to pHs between 7.0 and 5.0. Neurons were voltage clamped at -70 mV and acid solutions perfused for 5 s. Cells with an inward current of at least 500 pA for a given pH were considered sensitive. Most masseter afferents (45/70; 64%) displayed robust inward currents when the extracellular pH was changed from 7.4 to 6.8 (average amplitude 4.9 ± 0.5 nA). Similarly, changing extracellular pH from 7.4 to 6.8 produced large inward currents in most unlabeled ganglion afferents (11/19 cells (58%), average amplitude 6.9 ± 1.4 nA). The inward currents elicited by changing pH in ganglion masseter afferents could be distinguished by the degree of inactivation of the current during a pH application to pH 6.0 (Figure 4). The peak inward current in response to a pH step from 7.4 to 6.0 was 11.8 ± 1.5 nA, which declined to 1.2 ± 0.3 nA by 1.5 s (n = 45). In 33/45 (73%) of cells the inward current at 1.5 s was less than 10% of the peak (average sustained current 3 ± 0.5% of peak), in 12/45 (27%) of cells the sustained current was greater than 10% of the peak (average sustained current 24 ± 4% of peak).
Figure 4 Contrasting masseter afferent responses to nociceptive mediators. In these experiments cells were voltage clamped at -70 mV and mediators applied for the duration of the bars. The percentage of all tested cells that responded with currents similar to those illustrated is shown below each example trace. A) Perfusion of the TRPV1 agonist capsaicin produced an inward current in many (i) trigeminal ganglion masseter afferents but no (ii) MeV masseter afferents. B) Perfusion of ATP produced inward currents in most (i) trigeminal ganglion masseter afferents and all (ii) MeV masseter afferents. The ATP currents in ganglion afferents exhibited a variety of kinetic profiles while those elicited from MeV neurons were uniformly small and non-desensitizing. C) Changing the pH of the perfusion solution from 7.4 to various more acidic values produced large inward currents in most (i) trigeminal ganglion masseter afferents and all (ii) MeV masseter afferents. Note that MeV neurons required greater changes in pH (7.4 to 6.5) to produce detectable inward currents than TG neurons (7.4 to 7.0). Acid-induced currents in TG neurons most commonly inactivated completely within about 500 ms, but in about 30% of cells a substantial inward current remained at the end of the 2 sec acid perfusion to pH 6.0.
In masseter afferents isolated from the MeV nucleus changing extracellular pH from 7.4 to 6.8 produce an inward current of only 120 ± 25 pA (n = 11). However, changing the pH from 7.4 to 6.0 produced inward currents of 3.5 ± 1 nA in MeV cells, and in 3 cells further decreasing the pH to 5.0 produced even larger inward currents (7.5 ± 0.3 nA).
Co-expression of Nociceptive Channels
We were able to examine the co-expression of all 3 of the putative nociception-related ion channels in 55 TG masseter afferents (Figure 5). Only 4/55 (7%) of neurons failed to express one of either TRPV1, ASIC or P2X3-like current (Figure 5a). Conversely, only 3/55 (5%) of cells expressed all 3 types of current. About half the neurons (29/55, 53%) expressed at least 2 of the channels while the remainder (22/55, 40%) expressed only one of the nociception-related channels. The data for co-expression of channels is illustrated in Figure 5b. ASIC3-like currents were the most commonly co-expressed channel in neurons expressing TRPV1 or P2X3-like currents, while no neurons expressed only TRPV1 and P2X3-like currents.
Figure 5 Prevalence and co-expression of putative markers of nociceptive sensory neurons in trigeminal ganglion masseter afferents. A) The proportion of masseter afferents from TG that expressed substantial (> 500 pA) currents mediated by TRPV1-, ASIC3- and P2X3-containing channels is plotted. Virtually all ganglion masseter afferents expressed at least one of these channels. B) Co-expression of nociceptive channels in cells expressing (i) TRPV1-like currents, (ii) ASIC-3 like currents and (iii) P2X3-like currents. TRPV1 expressing cells often expressed ASIC3-like currents, and P2X3-like currents were only found in the cells also expressing ASIC3-like currents. Cells with ASIC3-like currents often expressed either TRPV1 or P2X3-like currents and occasionally both together. Cells with P2X3-like currents usually expressed ASIC3-like currents but TRPV1-like currents were only found in the population of these cells also expressing ASIC3-like currents. The numbers of cells in each population can be found in the Results.
Discussion
Recordings from muscle afferent fibers have provided a wealth of information about the response properties of these cells in physiological and pathophysiological situations, and how these responses are modified by sensory mediators [9]. However, ion channel activation and the signal transduction cascades modulating primary afferent excitability are most directly studied by making electrophysiological or optical recordings from sensory neuron cell bodies, and the present study provides some of the first descriptions of the electrophysiological properties of isolated sensory neurons innervating muscle. The results highlight the differences in the molecular signatures of proprioceptive muscle afferents and other muscle afferents, as well as the differences between muscle afferents and those which innervate other structures in the head such as teeth [4,8].
Sensory neuron modality can only be determined in in vivo or intact ex vivo preparations, thus we cannot assign a definitive physiological function to the cells in the present study. However, sensory neurons that detect potentially noxious stimuli (nociceptors) are thought to preferentially express a number of ion channels not normally found in other primary afferents. For example, expression of the TTX-resistant sodium channels NaV1.8 and NaV1.9 has been strongly correlated with a nociceptive sensory modality in in vivo recordings made from sensory neuron cell bodies [10,11]. Channels such as the vanilloid receptor TRPV1 are thought to be expressed exclusively by nociceptors because they are normally activated by demonstrably noxious stimuli and there is a strong correlation between selective pharmacological activation of the channels and human sensations [7,12]. Other channels are thought to be associated with nociceptors because their biophysical properties are sufficient to explain a response to a noxious stimulus by a subset of sensory neurons. For example, ASIC3 channels are activated by the modest changes in extracellular calcium and pH that accompany cardiac ischaemia and are highly expressed in a subset of cardiac sensory afferents that are presumed to transmit the pain of angina [5]. The assignment of ion channels to nociceptive neurons has also been made based on correlating channel expression with other putative markers of nociceptors including small soma diameter, expression of substance P, calcitonin gene related peptide or TRPV1 and expression in sensory neurons projecting to tissues from which the only conscious sensation is pain [4].
Almost all the masseter afferents isolated from the TG expressed significant amounts of at least one of the putative "nociceptive" ion channels we examined in this study; capsaicin activated TRPV1 channels, acid activated ASIC-3-like channels or ATP activated P2X3-like channels. With the exception of cells expressing TRPV1, a nociceptive phenotype cannot be reasonably inferred from the expression of any one channel, however significant numbers of masseter afferents expressed two or more of the channels we examined. Almost all cells expressing P2X3-like channels also expressed ASIC3-like currents and about 25% also expressed TRPV1; most cells expressing ASIC3-like currents also expressed either P2X3-like currents or TRPV1, and more than 50% of the TRPV1 expressing cells also expressed either P2X3-like currents or ASIC3-like currents. These data indicate that most TG masseter afferents can detect a noxious stimulus, but whether this represents their primary or only function remains unknown. By contrast, masseter afferents isolated from the MeV nucleus did not express TRPV1, P2X3-like or ASIC3-like channels, consistent with their function as purely proprioceptive muscle spindle afferents [3].
There is considerable evidence that a proportion of muscle afferents can reliably signal stimuli in both the innocuous and noxious range, and the properties of some of these afferents are consistent with the expression patterns of the channels in muscle afferents found in the present study [9]. In particular, afferents that are activated by the changing metabolic state of muscle (metaboreceptors) [13] appear to express channels classically associated with nociceptors, such as TRPV1, but clearly signal non-noxious information as well. Thus, lactic acid stimulation of muscle afferents in rat produces a classic cardiovascular pressor response that is sensitive to the ASIC channel antagonist amiloride but not to the TRPV1 antagonist capsazepine [14]. However, the lactic acid-induced response is attenuated after pretretament with the potent TRPV1 agonist resiniferatoxin, which desensitizes or destroys TRPV1-expressing nerves [14]. Further, while capsaicin produces a pressor response, blocking TRPV1 does not inhibit a contraction-induced pressor response [15]. Recordings from muscle afferents also show that capsaicin activates a population of Group III and Group IV afferents, some of which also proton-sensitive [16]. Thus it seems that a significant proportion of muscle afferents involved in producing activity-induced cardiovascular reflexes, perhaps mediated by activation of ASIC channels, also express TRPV1. Injection of capsaicin into human masseter muscle is painful [17], so there is no question that there are TRPV1 expressing afferents in muscle that transduce noxious stimuli. Our findings that about 50% of TRPV1 containing masseter afferents also expressed ASIC channels, and 30% of ASIC expressing afferents expressed TRPV1 are consistent with these results. There is no other information about the co-expression of TRPV1, ASIC and P2X receptors in afferents from the masseter muscle, although a relatively limited co-expression of TRPV1 and P2X3 receptors has been reported in gastrocnemius-soleus muscle afferents [16].
The ASIC channels in trigeminal masseter afferents seemed to be largely comprised of homomeric ASIC3 channels or ASIC channel heteromers containing ASIC3. ASIC3 channels are highly sensitive to changes in extracellular pH and lactate and if activated by a substantial change in pH they desensitize significantly more rapidly than other ASIC channels [5,18,19]. The pH 6-induced currents in most trigeminal ganglion masseter afferents desensitized more than 90% during the 1.5 s proton application, while in the remaining neurons the significant residual current (> 10% of the peak) suggested the presence of other ASIC subunits in these cells, probably ASIC1 [18]. Thus, the majority of ASIC currents observed in TG ganglion afferents had similar properties to those found in rat cardiac afferents (18), which are thought to be ASIC3-mediated. However, in the absence of selective blockers of ASIC subunits, we cannot definitively assign the currents we observed to specific ASIC subunits or combinations of subunits. MeV masseter afferents exhibited robust acid-induced currents but these were less sensitive to changes in extracellular pH and desensitized much more slowly than ganglion neuron ASIC currents.
In the only previous study of ASIC channel function in muscle afferents, 50% of sensory neurons labeled from the gastrocnemius muscle responded to pH 5.0 solution with robust inward currents [20]. The currents elicited in 30% of the cells were tentatively assigned to ASIC3/ASIC2b heteromers. The crucial role of ASIC3 channels in muscle-associated sensory function is underlined by the main finding of that study, which is that ASIC3 channel expression is required for the long lasting hyperalgesia produced by repeated acid injection in muscle [20]. The currents we observed in the majority of rat masseter afferents differ from those reported in mouse dorsal root ganglion neurons [21], primarily due to the lack of a significant sustained current component at pH 6.0-in our experiments this component was only 3% of the peak current. However, our conclusion that ASIC3 forms an essential part of masseter afferent ASIC channel complexes, is similar to that reached by others based on experiments in mouse DRG neurons from ASIC-null mice [21].
Action Potentials and Sodium Channels in Masseter Afferents
The action potentials of the MeV masseter afferents were narrow and lacked an inflection on the downward component of the current, consistent with previous recordings from acutely isolated MeV neurons [22] and MeV neurons in brain slices [23,24]. The INa recorded from MeV masseter afferents were completely blocked by TTX. These data are consistent with reports that muscle spindle afferents do not express detectable NaV1.8 or NaV1.9 immunoreactivity [10,11]. The narrow action potentials of proprioceptive afferents are consistent with the very rapid firing rates that these neurons achieve-exceeding 200 Hz (e.g. [25]). By contrast, the masseter afferents isolated from the TG had a wide range of action potential widths and shapes, and most cells had a significant component of TTX-resistant INa. TTX-resistant INa are subject to acute regulation by a variety sensory mediators acting via G protein-coupled or tyrosine kinase-linked receptors, particularly prostaglandins, bradykinin and nerve growth factor [26]. The changes in INa availability produced by these mediators mean that afferents expressing TTX-resistant INa are likely to be subject to rapid changes in excitability reflecting the state of the tissue they innervate. Wider action potentials and greater amounts of TTX-resistant INa are strongly correlated with a nociceptive modality, but these properties vary between afferents of different conduction velocity classes as well as between afferents of different modality within a class [27], and one cannot define a neuron as nociceptive simply on the basis of a action potential duration or its sensitivity to TTX. Nevertheless, within the TG masseter afferents, which presumably contain cells with a nociceptive function, smaller neurons tended to have wider action potentials. There was no such relationship apparent within the proprioceptive MeV masseter afferents.
P2X Receptors
We found a wide variety of ATP-induced currents in TG masseter afferents, similar to results from other studies in sensory neurons [28-30]. Messenger RNA and receptor-like immunoreactivity for 6 of the 7 cloned subtypes of P2X receptor are found in the trigeminal ganglion [31,32] and the currents we recorded are likely to be comprised of a mixture of homo- and heteromeric P2X receptor channels. Although attempting to define the P2X subunits responsible for the variety of ATP currents was beyond the scope of this study (but see [28]) we attributed the rapidly desensitizing ATP current observed in some masseter afferents to P2X3 receptor activation. Rapidly desensitizing ATP currents in sensory neurons have been reported to depend on the presence of the P2X3 gene or have been identified pharmacologically as P2X3 receptors [33-35] and although the kinetically similar P2X1 receptor has been shown to be present in sensory neurons by immunohistochemical methods, there is little electrophysiological evidence for currents mediated by P2X1 receptors in rat or mouse sensory neurons [4,29].
P2X3 subunits make a major contribution to the ATP currents in trigeminal neurons projecting to the tooth pulp, both as P2X3 receptor homomers and putative P2X2/P2X3 heteromers [4]. Relatively fewer masseter afferents express P2X3-like immunoreactivity [36]. In the present study we observed rapidly desensitizing ATP currents either alone or in combination with other P2X currents in about 55% of TG masseter afferents, which is similar to the proportion of tooth pulp nociceptors which displayed fast ATP currents (44%, [4]). P2X3-like immunoreactivity has been reported in about 25% of masseter afferents [36], which is a considerably smaller proportion than suggested by the present report. These differing results may reflect differing sensitivities of immunohistochemistry and electrophysiology, they may be due to the presence of some P2X1-containing currents in masseter afferents or perhaps arise from the short time the isolated neurons spend in culture. 30 – 40% of tooth pulp afferents can be labeled with P2X3 antiserum [4,37] but it is interesting to note that more than 50% of tooth pulp afferents challenged with ATP also displayed persistent currents. This indicates that while P2X3-containing ATP receptors are found in many putative nociceptors, they are not the only P2X subunits that may detect noxious stimuli signalled by ATP.
Opioid Modulation of Calcium Channels
The relatively high μ-opioid receptor sensitivity of jaw muscle afferents is similar to that reported in afferents projecting to hindlimb muscles, where about 75% of cells were sensitive to DAMGO [38]. The apparently high sensitivity of muscle afferent ICa to μ-opioid agonists contrasts with the reported low frequency of μ-opioid agonist modulation of ICa in skin and colonic afferents (approximately 10%, [39]). The relative insensitivity of MeV ICa to DAMGO (1 of 11 cells responding) is consistent with the extremely low mRNA abundance in these cells (2 of 72, [40]). The modulation of ICa in MeV neurons by the GABAB receptor agonist baclofen is in contrast to the lack of affect of baclofen on the membrane properties of MeV neurons in slices [41].
Interestingly, the μ-opioid receptor sensitivity of masseter muscle afferents differs markedly from that reported for the "purely nociceptive" afferents from tooth pulp [8,40]. DAMGO inhibited ICa in more masseter afferents than tooth pulp afferents, regardless of cell size (80% versus 42% respectively [40]), and most strikingly, DAMGO was equally effective in small and large masseter afferents (85% and 82% of cells inhibited respectively). By contrast, DAMGO inhibited ICa in only 30% of large tooth pulp afferents [8]. As 30% of large masseter afferents had substantial capsaicin currents (> 500 pA), indicating that these cells are likely to be nociceptors, our data suggest that μ-opioid receptors may be differentially expressed in distinct populations of nociceptors projecting to different tissues in the head, i.e. preferentially expressed in muscle nociceptors versus tooth pulp nociceptors. These data suggest that opioid analgesics may be better at relieving some types of head pain than others, and further that the endogenous opioid analgesic systems of the periphery may display differential effectiveness against nociceptive stimuli arising from distinct structures. The apparently high expression of opioid receptors in TG masseter afferents also suggests that these receptors may have functions in muscle physiology other than simple inhibition of nociception.
Conclusion
This study demonstrates that most masseter muscle afferents isolated from the trigeminal ganglion express one or more ion channels associated with detecting noxious stimuli, while masseter muscle proprioceptive afferents isolated from the MeV a different array of ion channels consistent with their non-nociceptive phenotype. It remains to be seen whether this profile is typical of sensory innervation of skeletal muscle, and whether the phenotypes described in this study undergo significant changes in chronic pathologies of the masseter muscle or associated nerves.
Methods
Cell labelling
All experiments were carried out using protocols approved by the OHSU Institutional Animal Care and Use Committee. Masseter afferents were labeled as outlined in detail in [42]. Briefly, male Sprague-Dawley rats between 5–8 weeks old were anesthetized with a s.c. injection of "rat cocktail" consisting of ketamine (55 mg kg -1), xylazine (5.5 mg kg -1 and acepromazine (1.1 mg kg -1). A small incision was made in skin overlying each masseter muscle and 5 × 1 μl injections of DiI (5% in DMSO) were made into each muscle with a Hamilton 10 μl syringe. The wound was closed with cyanoacrylate glue and the animals returned to their cages. Sensory neurons were isolated 2 weeks after surgery.
Cell Isolation
Cells were isolated from trigeminal ganglia essentially as described in [42]. Briefly, rats were anaesthetized with halothane (4%), and killed by decapitation. The trigeminal ganglia were removed and placed in cold Ca2+/Mg2+-Free Hanks Solution (CMF Hanks). Ganglia were cut up with iridectomy scissors and incubated at 35°C for 20 minutes in CMF Hanks plus papain (20 U ml-1), followed by 20 minutes in CMF Hanks plus dispase (4 mg ml-1) and collagenase (3 mg ml-1). The enzyme incubation was stopped with F-12 media supplemented with 10% fetal bovine serum (FBS), 50 U/ml penicillin/streptomycin, and the cells were released by gentle trituration through decreasing bore Pasteur pipettes with fire-polished tips. Cells were plated on plastic culture dishes precoated with poly-D-lysine and laminin. After the cells had settled they were cultured in a humidified chamber at room temperature in Leibovitz's L-15 medium supplemented with 10% FBS, 50 ng ml-1 NGF, 5 mM glucose, 5 mM NaHEPES and 50 U ml-1 penicillin/streptomycin.
Cells were isolated from the MeV nucleus by a modification of the methods outlined in [42]. Briefly, rats were anaesthetized with halothane and killed by a blow to the chest. The brain was rapidly removed and a block containing the brainstem immersed in ice cold artificial cerebrospinal fluid of composition (mM): NaCl 126, KCl 2.5, NaH2PO4 1.4, MgCl2 1.2, CaCl2 2.4, glucose 11, NaHCO3 25. Slices of the brainstem (400 μM) containing the MeV region were made with a Leica vibratome and the region containing the MeV cells subdissected with fine needles. The tissue chunks were placed in modified HBS (Solution 1 containing 10 mM MgCl2, 2 mM CaCl2) containing papain, 20 U ml-1 and incubated for 3–5 minutes at 37°C. The enzyme incubation was stopped with F-12 media supplemented with 10% FBS, 25 ng ml -1 NT-3 and 50 U/ml penicillin/streptomycin, and the cells were released by gentle trituration through decreasing bore Pasteur pipettes with fire-polished tips. The cells were plated onto culture plates with confluent, quiescent glia and cultured overnight at 37°C.
Electrophysiological Recording
Ionic currents from trigeminal neurons were recorded in the whole-cell configuration of the patch-clamp method [43] at room temperature (22–24°C), as described in [44]. The solutions used to record different types of current are listed in Table 1. Dishes were continually perfused with HEPES buffered saline (HBS, Solution 1). Calcium channel currents (ICa), were recorded in Solution 2, recordings of sodium channel currents (INa) were made in Solution 3. Recordings of ICa and INa were made with fire polished borosilicate pipettes (A-M Systems #603500, Carlsborg WA) filled with (in mM): CsCl 120, MgATP 5, NaCl 5, Na2GTP 0.3, EGTA 10, CaCl2 2 and HEPES 20, pH 7.3, resistance approximately 2 MΩ. In the cells where ICa and INa were recorded, capsaicin currents were also recorded with the above internal solution. Action potentials were recorded in Solution 4, with an internal solution that consisted of (mM): K methanesulphonate 115, KCl 5, NaCl, 8, MgCl2, 1, MOPS 10, MgATP 2, Na2GTP 0.3, BAPTA-K4 10, pH 7.0. In these recordings thin walled 7052 type glass (Garner Glass Company, Claremont CA) was used.
Table 1 External Solutions
Solution 1
HBS Solution 2
ICa buffer Solution 3
INa buffer Solution 4
AP buffer*
NaCl (mM) 140 20 or 40 145
KCl (mM) 2.5 5
CaCl2 (mM) 2.5 2.5 1 2
MgCl2 (mM) 1 1 3 1
HEPES (mM) 10 10 10 10
MES (mM) 10
TEA (mM) 140 120 or 100
CsCl (mM) 5
All external solutions contained glucose (*5 mM or 10 mM) and the pH was adjusted to 7.4 with either NaOH or CsOH as appropriate. Osmolarity was 300–330 mOsmol.
Recordings were made using either an Axopatch 1D amplifier (Axon Instruments, Union City, CA, USA) or a HEKA EPC 9 amplifier using Pulse acquisition and analysis software (HEKA). Currents were filtered at 3–5 kHz, sampled at 20–100 KHz, and recorded on hard disk for later analysis. Series resistance ranged from 2–7 MOhm and was compensated by 80% in all experiments. An approximate value of whole cell capacitance was determined by nulling the amplifier capacitance compensation circuit (Axoptach 1D) or automatically by EPC 9. Leak current was subtracted on line using a P/8 protocol. Cells were exposed to drugs via a series of flow pipes positioned about 200 μM from the cells. Fast application of ATP, capsaicin and acid were made with valve controlled sewer pipes.
Data analysis
Significant differences between means were tested using unpaired, two tailed Students t-test as noted. All data are expressed as mean ± S.E.mean unless otherwise indicated.
Drugs and Chemicals
DAMGO ([Tyr-D-Ala-Gly-MePhe-Gly-ol]enkephalin), U-69593 ((+)-(5-, 7-, 8-)-N-methyl-N-[7-(1-pyrrolidinyl)-1-oxaspiro[4.5]dec-8-yl]benzeneacetamide), deltorphin II, capsaicin, laminin, polylysine were from Sigma. Buffer salts were from Sigma. F-12, L-15 and fetal bovine serum were from GIBCO. Tetrodotoxin and human NT-3 were from Alomone Laboratories. NGF (mouse 2.5 S) was from either Upstate or Sigma. Papain and collagenase were from Worthington Biochemical Corporation (Freehold, NJ, USA), dispase was from Roche Applied Sciences.
Competing interests
MC, LAN & EWM declare that they have no conflicts of interest relating to this work.
Authors' contributions
MC performed experiments and wrote the draft of the paper, LAN performed experiments and EWM conceived the study. All the authors were involved in the experimental design, data analysis and writing the final paper. All the authors have all seen and approved the final paper.
Acknowledgements
Supported by NH&MRC of Australia 302002 and ARC Grant DP0449575 to MC, a scholarship from CAPES, Brazil to LAN, and NIH grants NS37010 and HL64840 to EWM.
We thank Robert Mouton and the Westbrook lab for their help with glial cultures, and JT Williams' lab for a lend of their vibratome.
==== Refs
Svensson P Graven-Nielsen T Craniofacial muscle pain: Review of mechanisms and clinical manifestations J Orofac Pain 2001 15 117 145 11443825
Ambalavanar R Moritani M Haines A Hilton T Dessem D Chemical phenotypes of muscle and cutaneous afferent neurons in the rat trigeminal ganglion J Comp Neurol 2003 460 167 179 12687682 10.1002/cne.10655
Cody FW Lee RW Taylor A A functional analysis of the components of the mesencephalic nucleus of the fifth nerve in the cat J Physiol 1972 226 249 261 4263681
Cook SP Vulchanova L Hargreaves KM Elde R McCleskey EW Distinct ATP receptors on pain-sensing and stretch sensing neurons Nature 1997 387 505 508 9168113 10.1038/387505a0
Sutherland SP Benson CJ Adelman JP McCleskey EW Acid-sensing ion channel 3 matches the acid-gated current in cardiac ischemia-sensing neurons Proc Natl Acad Sci USA 2001 98 711 716 11120882 10.1073/pnas.011404498
McCleskey EW Gold MS The ion channels of nociception Annu Rev Physiol 1999 61 835 856 10099712 10.1146/annurev.physiol.61.1.835
Julius D Basbaum AI Molecular mechanisms of nociception Nature 2001 403 203 210 11557989 10.1038/35093019
Taddese A Nah SY McCleskey EW Selective opioid inhibition of small nociceptive neurons Science 1995 270 1366 1369 7481826
Mense S Nociception from skeletal muscle in relation to clinical muscle pain Pain 1993 54 241 289 8233542 10.1016/0304-3959(93)90027-M
Fang X Djouhri L Black JA Dib-Hajj SD Waxman SG Lawson SN The presence and role of the tetrodotoxin-resistant sodium channel Nav1.9 (NaN) in nociceptive primary afferent neurons J Neurosci 2002 22 7425 7433 12196564
Djouhri L Fang X Okuse K Wood JN Berry CM Lawson SN The TTX-resistant sodium channel Nav1.8 (SNS/PN3) expression and correlation with membrane properties in nociceptive primary afferent neurons J Physiol 2003 550 739 752 12794175 10.1113/jphysiol.2003.042127
Handwerker HO Kobal G Psychophysiology of experimentally induced pain Physiol Rev 1993 73 693 671
McCloskey DI Mitchell JH Reflex cardiovascular and respiratory responses originating in exercising muscle J Physiol 1972 224 173 186 5039977
Li J Maile MD Sinoway AN Sinoway LI Muscle pressor reflex: potential role of vanilloid type 1 receptor and acid-sensing ion channel J Appl Physiol 2004 97 1709 1714 15220301 10.1152/japplphysiol.00389.2004
Kindig AE Heller TB Kaufman MP VR1 receptor blockade attenuates the pressor response to capsaicin, but has no effect on the pressor response to contraction in cats Am J Physiol Heart Circ Physiol 2005 288 H1867 H1873 15563536 10.1152/ajpheart.00735.2004
Hoheisel U Reinohl J Unger T Mense S Acidic pH and capsaicin activate mechanosensitive group IV muscle receptors in rat Pain 2004 110 149 157 15275762 10.1016/j.pain.2004.03.043
Sohn MK Graven-Nielsen T Arendt-Nielsen L Svensson P Inhibition of motor unit firing during muscle pain in humans Muscle Nerve 2000 23 1219 1226 10918259 10.1002/1097-4598(200008)23:8<1219::AID-MUS10>3.0.CO;2-A
Waldmann R Lazdunski M H+-gated cation channels: neuronal acid sensors in the NaC/DEG family of ion channels Curr Opin Neurobiol 1998 8 418 424 9687356 10.1016/S0959-4388(98)80070-6
Immke DI McCleskey EW Lactate enhances the acid-sensing Na+ channel on ischemia-sensing neurons Nat Neurosci 2001 4 869 870 11528414 10.1038/nn0901-869
Sluka KA Price MP Breese NM Stucky CL Wemmie JA Welsh MJ Chronic hyperalgesia induced by repeated acid injections in muscle is abolished by the loss of ASIC3 but not ASIC1 Pain 2003 106 229 239 14659506 10.1016/S0304-3959(03)00269-0
Benson CJ Xie J Wemmie JA Price MP Henss JM Welsh MJ Snyder PM Heteromultimers of DEG/EnaC subunits form H+-gated channels in mouse sensory neurons Proc Natl Acad Sci USA 2002 99 2338 2343 11854527 10.1073/pnas.032678399
Yoshida S Oka H Membrane properties of dissociated trigeminal mesencephalic neurons of the adult rat Neurosci Res 1998 30 227 234 9593333 10.1016/S0168-0102(98)00003-0
Henderson G Pepper CM Shefner SA Electrophysiological properties of neurons contained in the locus coeruleus and mesencephalic nucleus of the trigeminal nerve in vitro Exp Brain Res 1982 45 29 37 6120088 10.1007/BF00235760
Verdier D Lund JP Kolta A GABAergic control of action potential propagation along axonal branches of mammalian sensory neurons J Neurosci 2003 23 2002 2007 12657657
Masri R Ro YJ Capra N The effect of experimental muscle pain on the amplitude and velocity sensitivity of jaw closing muscle spindle afferents Brain Res 2005 1050 138 147 15982645 10.1016/j.brainres.2005.05.039
Lai J Porreca F Hunter JC Gold MS Voltage-gated sodium channels and hyperalgesia Annu Rev Pharmacol Toxicol 2004 44 371 397 14744251 10.1146/annurev.pharmtox.44.101802.121627
Lawson SN Phenotype and function of somatic primary afferent nociceptive neurons with C-, Aδ-or Aα/β-fibres Exp Physiol 2002 87 239 244 11856969 10.1113/eph8702350
Burgard EC Niforatos W Van Bisen T Lynch KJ Touma E Metzger RE Kowaluk EA Jarvis MF P2X receptor-mediated ionic currents in dorsal root ganglion neurons J Neurophysiol 1999 82 1590 1598 10482772
Petruska JC Cooper BY Gu JG Rau KK Johnson RD Distribution of P2X1, P2X2 and P2X3 subunits in primary afferents: relation to population makers and specific cell types J Chem Neuroanat 2000 20 141 162 11118807 10.1016/S0891-0618(00)00080-6
Spehr J Spehr M Hatt H Wetzel CH Subunit-specific P2X-receptor expression defines chemosensory properties of trigeminal neurons Eur J Neurosci 2004 19 2497 2510 15128403 10.1111/j.0953-816X.2004.03329.x
Collo G North RA Kawashima E Merlo-Pich E Neidhart S Surprenant A Buell G Cloning of P2X5 and P2X6 receptors and the distribution and properties of an extended family of ATP-gated ion channels J Neurosci 1996 16 2495 2507 8786426
Xiang Z Bo X Burnstock G Localization of ATP-gated P2X receptor immunoreactivity in rat sensory and sympathetic ganglia Neurosci Lett 1998 256 105 108 9853714 10.1016/S0304-3940(98)00774-5
Rae MG Rowan EG Kennedy C Pharmacological properties of P2X3-receptors present in neurons of the rat dorsal root ganglia Brit J Pharmacol 1998 124 176 180 9630357 10.1038/sj.bjp.0701803
Cockayne DA Hamilton SG Zhu Q-M Dunn PM Zhong Y Novakovic S Malmberg AB Cain G Berson A Kassotakis L Hedley L Lachnit WG Burnstock G McMahon SB Ford APDW Urinary bladder hyporeflexia and reduced pain-related behavior in P2X3-deficient mice Nature 2000 407 1011 1015 11069181 10.1038/35039519
Souslova V Cesare P Ding Y Akopian AN Stanfa L Suzuki R Carpenter K Dickenson A Boyce S Hill R Nebenius-Oosthuizen D Smith AJH Kidd E Wood JN Warm-coding deficits and aberrant inflammatory pain in mice lacking P2X3 receptors Nature 2000 407 1015 1017 11069182 10.1038/35039526
Ambalavanar R Moritani M Dessem D Trigeminal P2X3 receptor expression differs from dorsal root ganglion and is modulated by deep tissue inflammation Pain 2005 117 280 291 16153775 10.1016/j.pain.2005.06.029
Ichikawa H Sugimoto T The co-expression of P2X3 receptor with VR1 and VRL-1 in the rat trigeminal ganglion Brain Res 2004 998 130 135 14725976 10.1016/j.brainres.2003.11.019
Schroeder JE McCleskey EW Inhibition of Ca2+ currents by a μ-opioid in a defined subset of rat sensory neurons J Neurosci 1993 13 867 873 7678862
Su X Wachtel RE Gebhart GF Inhibition of calcium currents in rat colon sensory neurons by κ- but not μ- or δ-opioids J Neurophysiol 1998 80 3112 3119 9862909
Silbert SC Beacham DW McCleskey EW Quantitative single-cell differences in mu-opioid receptor mRNA distinguish myelinated and unmyelinated nociceptors J Neurosci 2003 23 34 42 12514199
Hayar A Poulter MO Pelkey K Feltz P Marshall KC Mesencephalic trigeminal neuron responses to γ-aminobutyric acid Brain Res 1997 753 120 127 9125438 10.1016/S0006-8993(97)00002-4
Eckert SP Taddese A McCleskey EW Isolation and culture of rat sensory neurons having distinct sensory modalities J Neurosci Meth 1997 77 183 190 10.1016/S0165-0270(97)00125-8
Hamill OP Marty A Neher E Sackmann B Sigworth FJ Improved patch-clamp technique for high resolution current recordings from cell and cell-free membrane patches Pflugers Archiv 1981 391 85 100 6270629 10.1007/BF00656997
Borgland SL Connor M Christie MJ Nociceptin inhibits calcium channel currents in a subpopulation of small nociceptive trigeminal ganglion neurons in mouse J Physiol (London) 2001 536 35 47 11579155 10.1111/j.1469-7793.2001.t01-1-00035.x
|
16242047
|
PMC1283980
|
CC BY
|
2021-01-04 16:40:08
|
no
|
Mol Pain. 2005 Oct 24; 1:31
|
utf-8
|
Mol Pain
| 2,005 |
10.1186/1744-8069-1-31
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1141621612610.1186/1465-9921-6-114ResearchRhinovirus infection induces cytotoxicity and delays wound healing in bronchial epithelial cells Bossios Apostolos [email protected] Stelios [email protected] Dimitrios [email protected] Chrysanthi L [email protected] Andreas G [email protected] Photini [email protected] Nikolaos G [email protected] Allergy Department, 2nd Pediatric Clinic, University of Athens, Athens, Greece2005 10 10 2005 6 1 114 114 7 5 2005 10 10 2005 Copyright © 2005 Bossios et al; licensee BioMed Central Ltd.2005Bossios 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
Human rhinoviruses (RV), the most common triggers of acute asthma exacerbations, are considered not cytotoxic to the bronchial epithelium. Recent observations, however, have questioned this knowledge. The aim of this study was to evaluate the ability of RV to induce epithelial cytotoxicity and affect epithelial repair in-vitro.
Methods
Monolayers of BEAS-2B bronchial epithelial cells, seeded at different densities were exposed to RV serotypes 1b, 5, 7, 9, 14, 16. Cytotoxicity was assessed chromatometrically. Epithelial monolayers were mechanically wounded, exposed or not to RV and the repopulation of the damaged area was assessed by image analysis. Finally epithelial cell proliferation was assessed by quantitation of proliferating cell nuclear antigen (PCNA) by flow cytometry.
Results
RV1b, RV5, RV7, RV14 and RV16 were able to induce considerable epithelial cytotoxicity, more pronounced in less dense cultures, in a cell-density and dose-dependent manner. RV9 was not cytotoxic. Furthermore, RV infection diminished the self-repair capacity of bronchial epithelial cells and reduced cell proliferation.
Conclusion
RV-induced epithelial cytotoxicity may become considerable in already compromised epithelium, such as in the case of asthma. The RV-induced impairment on epithelial proliferation and self-repair capacity may contribute to the development of airway remodeling.
==== Body
Background
The bronchial epithelium plays a unique role as a protective physical and functional barrier between external environment and underlying tissues. As a result of this role it is frequently injured and epithelial integrity is damaged. A repair process starts quickly which includes migration of the remaining basal airway epithelial cells to repopulate damaged areas, and subsequent proliferation and differentiation until epithelial integrity has been restored [1,2].
Epithelial damage is a key feature of asthma. As a result of inflammation, a large portion of columnar epithelial cells shed and form Creola bodies, detected in sputum and during bronchoscopy in asthmatic patients [3]. This cycle of damage and repair has been proposed as a key mechanism leading to thickening of the airway wall, and other pathologic alterations collectively characterized as airway remodeling [4], which in turn has been associated with incompletely reversible airway narrowing, bronchial hyper-responsiveness and asthma symptoms [5].
Many factors can be cytotoxic to the bronchial epithelium, including eosinophil products [6], allergens [7] and respiratory viruses. Virus-induced cytotoxicity has been well documented for the majority of these agents, including influenza, parainfluenza, adenovirus and respiratory syncytial virus (RSV) [8]. In contrast, human rhinoviruses (RVs), although the most preponderant viruses associated with acute asthma exacerbations [9], have been shown to induce minimal, if any, cytotoxicity [10-12]. We have recently shown that RVs are able to replicate in human primary bronchial epithelial cells [13]. An unexpected finding in that study was that exposure of sparely seeded cell monolayers resulted in a considerable RV-specific cytopathic effect (CPE). RV-induced CPE was also reported in another study, in which case it was attributed to specific RV serotypes [14].
Based on the above, we hypothesized that RV infection may be conditionally able to affect epithelial cell viability and life-death cycle. Therefore, in this study we used BEAS-2B cells, a well-established in-vitro lower respiratory epithelium model of RV infection [15,16], used in parallel studies with primary bronchial epithelial cells [17,18]as well in cell death studies [19]to systematically investigate the ability of RV to induce cytotoxicity in bronchial epithelial cells. Furthermore, the effect of RV on an in-vitro model of epithelial wound repair was assessed.
Methods
Cell cultures
BEAS-2B cells, a human continuous bronchial epithelial cell line and Ohio-HeLa cells (obtained from ATCC and the MRC Cold Unit, UK, respectively) were cultured in Eagle's minimal essential medium (E-MEM) buffered with NaHCO3 and supplemented with 10% (v/v) fetal bovine serum (FBS) and 40 μg/ml of gentamycin, in a humified 5% CO2 incubator. Cells were spilt twice weekly.
Primary human bronchial epithelial cells (HBECs), initially deriving from an adult non-asthmatic volunteer in the course of another ongoing study, were available frozen in liquid nitrogen. They were isolated as described earlier [13]. Cells were rapidly thawed and cultured on plates pre-coated with collagen type-I (Nutacon, Holland), submerged in Clonetics BEGM (Cambrex, ML, USA). Medium was replaced daily. Cells were used at passage 2, at a confluence of 50%. It should be pointed out that cells in these submerged cultures are undifferentiated and do not form tight junctions [20].
All culture reagents were purchased from Gibco-Invitrogen Corp. (Carlsbad, CA, USA) and Falcon (Becton Dickinson, Labware, NJ, USA) and biochemicals were from Sigma (St. Louis, MO, USA), unless otherwise specified.
Virus cultures and titration
Rhinovirus types 5, 7, 9, 14 and 16 (major subtypes) and 1b (minor subtype) were propagated in Ohio-HeLa cells in large quantities at 33°C, in a humified, 5% CO2 incubator, as previously described[16]. Briefly, when full cytopathic effect (CPE) developed, cells and supernatants were harvested, pooled, frozen and thawed twice, clarified, sterile-filtered, aliquoted and stored at -70°C. Lysates of parallel Ohio-HeLa cell cultures, not infected with virus, were used as controls.
In order to determine RV titers, Ohio-HeLa cells were seeded in 96-well plates reaching 60–70% confluence at the time of infection. Logarithmic dilutions of RVs were made in multiple wells and after five days of culture the plates were fixed and stained with Crystal Violet Buffer (5% formaldehyde, 5% ethanol and 0.1% crystal Violet in PBS). The end-point titer was defined as the highest dilution at which a CPE was detected in at least half of the wells and expressed as the inverse logarithm of this dilution (MOI-multiplicity of infection-infectious units/cell). For each experiment, a new vial was rapidly thawed and used immediately [21].
In order to assess the specificity of RV-mediated responses, RV preparations were exposed to 58°C for 1 h. The successful inactivation was confirmed by lack of RV replication in Ohio-HeLa cells.
Cytotoxicity assay
BEAS-2B cells were plated in 48-well plates in serial dilutions and allowed to grow for 48 hours, reaching confluence of 100%, 50%, 25% and 12.5%. Cell numbers and respective confluence were assessed by standard Neubauer cytometer in initial experiments. Cells were then exposed to rhinovirus as previously described [16]. Briefly wells were washed with HBSS and virus was added at the desirable MOI in parallel to non-infected Ohio HeLa cell lysate negative controls. The amount of the virus added was proportional to the number of the cells. After 1 hour of gentle shaking at room temperature, fresh medium was added, to a final volume of 0.5 ml. Eagle's MEM supplemented with 4% FCS, 1% MgCl and 4% tryptose phosphate broth and 40 μg/ml of gentamycin, was used for the experiments. After 48 hours of incubation, cells were washed twice in PBS and a volume of crystal violet staining buffer equal to 1/5th of the original culture medium was added to the wells as indicator of cell viability [22,23]. Cells were incubated for 30 min at room temperature followed by extensive washing with distilled water. After air drying 0.2 ml of a destain buffer (16.6% v/v glacial acetic acid, 50% v/v methanol in ultra pure water) was added to the wells for 5 min. Cells were fully destained and the produced color was transferred to a clear 96 well ELISA plate and optical density was measured with a photometer at 595 nm (Ceres 900C, Bio-Tec Instruments, Inc, Winooski VT, USA) [13]. Cytotoxicity was estimated as % of the negative control (1- O.D RV infected/O.D HeLa * 100).
Epithelial repair assay
Confluent monolayers of BEAS-2B cells were grown in 48-well plates. Cells were then damaged mechanically by crossing three times with a 10–200 μL volume universal pipette tip (Corning, NY, USA) [1]. After washing twice with HBSS cells were infected with RV1b at MOI 0.5, in parallel to non-damaged monolayers as well as HeLa lysate controls as described above and incubated in a humified 5% CO2 incubator. Immediately after infection (t = 0) and at 24, 48 and 72 hours a plate was stained with crystal violet. Wells were photographed and the area of unpopulated cells was calculated with image analysis, (Scion Image software, b4.0.2, NIH). Furthermore, cytotoxicity was estimated, as described above.
In some experiments, cells were fixed and stained with 4, 6 diamino-2-phenylindole (DAPI), a DNA-binding dye. They were then viewed using a UV-visible Zeiss Axioplan 2 fluorescent microscope and fluorescence images were captured using a CCD camera.
Proliferation Assay
BEAS-2B cells were cultured in 25 cm2 flasks until confluent. After infection with RV 1b at 0.5 MOI or control, cells were incubated for an additional 24 hours at 33°C. They were then washed twice with HBSS, detached using a non-enzymatic cell dissociation buffer (Gibco, UK), split 1:2 in new flasks and re-incubated. At that time, as well as at 24, 48 and 72 hours later proliferation was estimated by staining with Proliferating Cell Nuclear Antigen (PCNA), a proliferation marker correlates with other markers of the S phase of cell cycle like tritiated thymidine and Bromodeoxyuridine labeling [24]. PCNA assessed with flow cytometry [25].
Flow Cytometry
BEAS-2B cells were harvested non-enzymatically and resuspended at a density of 1 × 105 cells/100 μl in washing buffer (PBS with 1% FBS). For ICAM-1 analysis cells were incubated with 20 μL anti-ICAM, phycoerythin-conjugated monoclonal antibody (Pharmingen, Becton Dickinson, Jan Hose, CA, USA) for 30 min at 4°C. After washing twice, cells were fixed with 0.5 ml of 1% paraformaldehyde in PBS and counted with a FACSort (Becton Dickinson, Jan Hose, CA, USA) flow cytometer. Fluorescence data were collected on 104cells and histogram analysis was performed with the use of Cell Quest software™.
For PCNA analysis, cells were permeabilized in a buffer comprising of 0.2 mg/ml Na2HPO4-2H2O, 1 mg/ml KH2PO4, 45% v/v acetone and 9.25% v/v formaldehyde [25], followed immediately by staining with 10 μL of an anti-PCNA, fluorescein-conjugated monoclonal antibody (Pharmingen, Becton Dickinson, Jan Hose, Ca, USA). Fluorescence data from 104 cells were collected and histogram analysis was performed with Cell Quest software.
Cell viability was assessed by staining with 7-aminoactinomycin D (7-AAD) (Becton-Dickinson, San Jose, Calif., USA).
Statistical Analysis
Data are expressed as mean ± standard error of mean. Statistical analysis was conducted with the SPSS 11.0 for Windows software. Linear regression analysis was used to evaluate the effect of cell density, and ANOVA for time and dose comparisons. Means were compared by non-parametric tests. P values less than 5% were considered significant.
Results
Rhinoviruses induce cytotoxicity in bronchial epithelial cells in a serotype and cell density- depended manner
BEAS-2B cultures were infected with RV1b, RV5, RV7, RV9, RV14 and RV16 at an MOI of 1 and confluences of 12.5%, 25%, 50% and 100%. The extent of RV-induced cytotoxicity differed between RV serotypes: RV9 was not cytotoxic at all at this MOI. RV1b and RV7 were the most cytotoxic, able to induce cytotoxicity even on confluent monolayers, while killing over 65%–70% of less dense cultures. RV14 and RV5 were moderately cytotoxic while RV16 could kill only sparsely seeded cells. Differences in RV-induced cell death between RV serotypes were statistically significant at all cell densities (p = 0.00 in all cases, ANOVA). Furthermore, a statistically significant inverse correlation between cell density and RV-induced cytotoxicity was observed for RV1b, RV7, RV14 and RV16, (p = 0.000, 0.000, 0.014 and 0.03 respectively, linear regression); RV5 was moderately cytotoxic at all cell densities. Figure 1 shows the % cytotoxicity of each RV serotype at different cell densities.
Figure 1 Cytotoxicity of different RV serotypes; 1b, 7, 14, 5, 16 and 9, at MOI = 1 on BEAS-2B cells, cultured until reaching different densities (100, 50, 25 and 12.5%). An inverse correlation between cell density and RV-induced cytotoxicity is observed for RV1b, RV7, RV14 and RV16, (*p < 0.05, n = 3–26, linear regression).
HBEC infected with 1 MOI of RV 1b, RV7 and RV16 at 50% confluence (n = 4), showed cytotoxicity levels of 50 ± 1%, 52 ± 3% and 0% respectively, almost identical to those observed in BEAS-2B cells under the same conditions, confirming that the described phenomenon is reproducible in primary cells.
Rhinovirus-induced cytotoxicity is dose depended
Subsequently, dose-dependence of cytotoxicity was assessed using RV1b, RV5, RV9 and RV16. At MOI-5, RV9 remained non-cytotoxic (data not shown). RV16 (Figure 2A), but even more RV 5 (Figure 2B), became cytotoxic on a cell-density dependent manner (p = 0.007 and p = 0.000, respectively, linear regression). RV1b at an MOI of 5 was able to kill almost 50% of a confluent monolayer, reaching a plateau of 85%–89% of cytoxicity in less dense cultures (Figure 2C). A dose-response was observed when comparing cytotoxicity of RV1b both in 100% confluent (16.5% ± 4.45, 34.66% ± 1.85, 49.65% ± 5.24 at 0.5, 1 and 5 MOI respectively, p = 0.000, ANOVA) and 50% confluent monolayers (23.76% ± 9.66, 58.75% ± 3.62, 86.46% ± 0.62 at 0.5, 1 and 5 MOI respectively, p = 0.000, ANOVA) (Figure 2D).
Figure 2 Cytotoxicity of RV16 (A), RV5 (B) and RV1b (C) at MOI = 5 on BEAS-2B cells, cultured until reaching different densities (100, 50, 25 and 12.5%, n = 3–8). RVs became more cytotoxic at this MOI, and density dependence appeared for RV5. Dose dependence is shown for RV1b (D) at 0.5, 1, and 5 MOI for both 100% and 50% cell densities (p = 0.000 in both cases, n = 6, ANOVA).
RV- induced cytotoxicity is specific
To determine whether the observed cytotoxicity is specific to RV and not associated with factors in the virus preparation, we exposed a 50% confluent monolayer to 1 MOI of heat-inactivated RV 1b. Inactivated RV1b lost its capacity to induce cell death (6.43% ± 3.68 vs. 55.87% ± 2.68 of live virus, p = 0.021, Mann-Whitney) (Figure 3).
Figure 3 Cytotoxicity of active and heat inactivated RV1b (MOI = 1) on 50% confluent BEAS-2B cells. Inactivated virus is no longer cytotoxic (*p = 0.021, n = 4, Mann Whitney).
ICAM-1 expression is not affected by cell density
To test whether cell-density dependent, differential susceptibility of BEAS-2B cells to RV cytotoxicity may relate to variations of ICAM-1 expression, the major RV receptor, cells were cultured as described above and expression of ICAM-1 was measured by flow cytometry. In all densities cells expressed ICAM-1 over 98.5%, without differences on fluorescence intensity (1030.6 ± 88.76, 980 ± 48.47, 964.29 ± 25.52 and 987.38 ± 37.5, at cell densities of 100%, 50%, 25% and 12.5% respectively).
RV infection delays epithelial wound repair
To test whether infection with RV may affect the self-repair capacity of bronchial epithelial cell monolayers, digital photos were taken immediately after mechanical damage (t = 0) as well as 24, 48 and 72 hours later in infected and non-infected monolayers. The damaged area not populated with cells was calculated by image analysis. Control cells demonstrated a fast response in repopulating the damaged area (from 133.07 ± 14.67 mm2 at t = 0 to 72.92 ± 3.59, 28.09 ± 3.11 and 13.49 ± 1.9 at 24, 48 and 72 hours respectively, p = 0.000, ANOVA). The damaged area in infected monolayers was repopulated considerably more slowly, while it seemed to plateau at 48 hours (t = 0, 124.41 ± 9.26 mm2, t = 24 h, 89.13 ± 5.55, t = 48 h, 52.18 ± 10.5, t = 72 h, 69.5 ± 6.3, p = 0.00, ANOVA). When infected and non-infected cells were compared, differences were significant at all time points (p = 0.024, 0.031 and 0.001 at 24, 48 and 72 hours respectively, Mann Whitney), (Figure 4).
Figure 4 Damaged epithelium (t = 0) is suboptimally repopulated after RV-infection in comparison to control BEAS-2B cells. DAPI stained cells (A). Repopulation of damaged epithelium, expressed as unpopulated area in mm2, in RV-infected and non-infected BEAS-2B cells, immediately after damage (t = 0) and at 24, 48 and 72 hours later (B). Repopulation in infected cells is significantly reduced (*p < 0.05, **p = 0.001, n = 8, Mann Whitney).
Furthermore, intact and wounded monolayers did not differ in susceptibility to RV-mediated cytotoxicity (17.13% ± 4.11 versus 17.64% ± 2.6 at 48 hours after infection for intact and wounded respectively), suggesting that epithelial wounding leaves unaffected the remaining cells of the monolayer in respect to RV-induced cytotoxicity.
RV infection decreases epithelial cell proliferation
The expression of PCNA, reflecting proliferative activity of epithelial cells, increased 24 hours after seeding, followed by a trend towards return to baseline at 48 and 72 h. However, PCNA expression (Mean Fluorescence Intensity, MFI) was significantly lower at all time points in RV-infected cells (Figure 5). Cell viability, assessed by 7ADD staining, was over 90% in these experiments.
Figure 5 Proliferation of BEAS-2B cells, as mean fluorescence intensity (MFI) of PCNA in RV-infected and control cells at various time points after reculture (A). The proliferation rate is significantly reduced for RV-infected cells at all time points (*p = 0.012, ** p < 0.05, n = 3–7, Mann Whitney). Representative histograms at t = 0 and 24 hours are shown (B). Closed line: isotype control, thick line: RV-infected, dashed line: non-infected control cells.
Discussion
In contrast to previous knowledge, but in line with recent observations, this study demonstrates that human rhinoviruses, the agents most frequently associated with acute asthma exacerbations [26], are able to become cytotoxic in an in-vitro model of human bronchial epithelium. A continuous cell line model was used for the majority of experiments; however, the finding was also confirmed in primary bronchial cells. Furthermore, it is shown for the first time that RV infection may delay epithelial wound healing by affecting epithelial cell proliferation.
It has been generally accepted that RVs do not induce cytotoxicity in-vitro or in-vivo [27-29], even in heavy colds [10-12,30]. However, two recent studies designed to assess the ability of RV to infect primary human bronchial epithelial cells have unexpectedly observed RV-associated cytotoxicity: in the study of Schroth et al [14], RV16 and RV49 were used and cytotoxicity was observed only with the latter serotype; the authors hypothesized that a higher viral binding and/or larger yield by RV49 may explain their observation, noting however the need for additional studies. This was also the case in the study of Papadopoulos et al. [13] in which RV cytotoxicity was observed when sparsely seeded cultures were exposed to the virus. The current study, which systematically addressed these possibilities, demonstrates that they are both reproducible, and in fact different RV serotypes differ in their cytotoxic capacity, which in most cases, is nevertheless cell density dependent. The latter finding can also explain why RV cytotoxicity was not observed in previous in-vitro studies, which were conducted with confluent cultures [28,29].
A recent comprehensive study from Deszcz et al [31] is in support of our findings, as it shows that RV14 can induce high levels of cytotoxicity in a bronchial epithelial cell line 16HBE14o-. Furthermore, they demonstrate that a possible mechanism is the induction of apoptosis via the mitochondrial pathway, a phenomenon also shown in primary cells from asthmatic subjects [32].
It has been shown that differentiated bronchial epithelial cells grown in air-liquid interface and developing tight junctions, are considerably resistant to RV infection [20]. In this respect, the results of this study, using submerged cultures that lead to non-differentiated cells, may overestimate the in-vivo situation. However, a characteristic of asthma is the significant loss of columnar epithelial cells leading to loss of its integrity and density [4], epithelial damage also correlates to the severity of the disease [33] In this respect, sparsely seeded bronchial epithelial cell cultures, can be considered as an extreme, but relevant model of asthmatic epithelium. Under such conditions, as show herein, RV-associated cytotoxicity increases considerably, with almost linear density-dependence, suggesting that virus-induced exacerbations may have increased sequels in more severe patients [34].
The fact that different RV serotypes are not equally capable of killing epithelial cells, ranging from no to extensive cytotoxicity, supports the possibility that this phenomenon may contribute to asthma exacerbation severity variations observed in clinical practice [35].
RV infects a small proportion of exposed cells [15]; biopsy data show that in human RV infection epithelial inflammation, potentially resulting from infection, is patchy [27,36]: there has been, however, no direct comparison between normal and asthmatic individuals in regard to RV-induced cytotoxicity in-vivo, a study complicated by the fact that the epithelial integrity and viability is considerably affected in asthmatics at baseline. In a recent study, Wark et al showed increased RV proliferation in primary epithelial cells obtained from asthmatic patients in comparison to normal controls [32]. We have also observed that exposure of BEAS-2B cells to culture supernatants modeling an 'atopic' environment, was also able to increase RV proliferation, and at the same time increase RV-induced cytotoxicity [37]. These observations further suggest that RV-induced cytotoxicity may be relevant in asthma exacerbation pathogenesis.
There are several possibilities in respect to the mechanism(s) underlying this phenomenon, which have not, however, been addressed in this study. One possibility might be that rapidly dividing cells, as is the case of sparse cultures, may be more permissive to RV infection. Moreover, differential expression of soluble factors, such as interferons, may regulate either susceptibility to infection or the proliferative potential of RV. These hypotheses, which may well not be mutually exclusive and could all contribute to RV cytotoxicity, are currently under investigation.
Independent of the causative mechanism(s), in an already affected epithelium, RV infection may lead to more profound damage. This would eventually lead to activation of repair mechanisms: deposition of extracellular matrix, proliferation and migration of epithelial cells in order to repopulate the damaged area, followed by cell differentiation [1,2]. Hence, we used a previously validated wound model [38,39] to investigate the role of RV infection on the repair process, describing for the first time an RV-mediated delay in epithelial wound healing, associated with reduced proliferation of RV infected cells. This finding may be of significance as altered restitution of airway structure is one of hallmarks of asthmatic inflammation leading to airway remodeling [4]. Damaged asthmatic epithelium has been previously reported to have proliferation defects during the repair process [40]. Dysregulated proliferation in bronchial epithelial cells from asthmatic patients has been associated with increased expression of the cyclin-dependent kinase inhibitor p21waf [41,42]. In severe, corticosteroid-dependent asthma, markers of epithelial cell proliferation are coexpressed with markers of activation, suggesting that, in at least that case, the repair process is associated with a persistent activation state of the epithelial cells [41]. The above findings have led investigators to propose that a repair/activation imbalance may be the central mechanism of airway remodeling in asthma [5]. In this respect, RV-induced cytotoxicity, an event frequently occurring and able to activate epithelial cells into an inflammatory response [16], may be implicated in the development of remodeling. The possibility that a viral infection may reprogram epithelial responses towards a 'remodeled' phenotype has also been proposed, based on a mouse model of paramyxoviral infection [43].
Conclusion
In conclusion, several human RV serotypes are able to become cytotoxic to human bronchial epithelial cells, especially when these are sparsely cultured; RVs are also able to delay epithelial wound healing. Previously unrecognised, RV-induced cytotoxicity may become important in the context of asthma in which the epithelium is already affected and consequently contribute to the induction and/or perpetuation of airway remodelling.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AB carried out the major part of experiments, participated in the sequence alignment and drafted the manuscript.
SP participated in epithelial repair assay.
DG participated in the study design and in the sequence alignment.
CLS performed primary epithelial cell experiments
PSP participated in the study desigh and helped to draft the manuscript.
AGC participated in the sequence alignment.
NGP, conceived of the study, participated in its design and coordination and participated in the writing of the manuscript.
Acknowledgements
We are grateful to Dr A. Kolialexi from Department of Medical Genetics, Athens University for her help in taking the DAPI pictures
==== Refs
Adam EC Holgate ST Fildew CJ Lackie PM Role of carbohydrates in repair of human respiratory epithelium using an in vitro model Clin Exp Allergy 2003 33 1398 1404 14519146 10.1046/j.1365-2222.2003.01771.x
Sacco O Silvestri M Sabatini F Sale R Defilippi AC Rossi GA Epithelial cells and fibroblasts: structural repair and remodelling in the airways Paediatr Respir Rev 2004 5 Suppl A S35 S40 14980241 10.1016/S1526-0542(04)90008-5
Woltmann G Ward RJ Symon FA Rew DA Pavord ID Wardlaw AJ Objective quantitative analysis of eosinophils and bronchial epithelial cells in induced sputum by laser scanning cytometry Thorax 1999 54 124 130 10325916
Bousquet J Jeffery PK Busse WW Johnson M Vignola AM Asthma. From bronchoconstriction to airways inflammation and remodeling Am J Respir Crit Care Med 2000 161 5 1720 1745 10806180
Davies DE Wicks J Powell RM Puddicombe SM Holgate ST Airway remodeling in asthma: new insights J Allergy Clin Immunol 2003 111 2 215 225 12589337 10.1067/mai.2003.128
Erjefalt JS Greiff L Andersson M Matsson E Petersen H Linden M Ansari T Jeffery PK Persson CG Allergen-induced eosinophil cytolysis is a primary mechanism for granule protein release in human upper airways Am J Respir Crit Care Med 1999 160 1 304 312 10390416
Hewitt CRA Horton H Jones RM Pritchard DI Heterogeneous proteolytic specificity and activity of the house dust mite proteinase allergen Der p I Clin Exp Allergy 1997 27 2 201 207 9061221 10.1046/j.1365-2222.1997.d01-492.x
Folkerts G Busse WW Nijkamp FP Sorkness R Gern JE Virus-induced airway hyperresponsiveness and asthma Am J Respir Crit Care Med 1998 157 6 1708 1720 9620896
Johnston SL Pattemore PK Sanderson G Smith S Lampe F Josephs L Symington P O'Toole S Myint SH Tyrrell DA Holgate ST Community study of role of viral infections in exacerbations of asthma in 9–11 year old children BMJ 1995 310 6989 1225 1229 7767192
Fraenkel DJ Bardin PG Sanderson G Lampe F Johnston SL Holgate ST Lower airways inflammation during rhinovirus colds in normal and in asthmatic subjects Am J Respir Crit Care Med 1995 151 879 886 7881686
Gern JE French DA Grindle KA Brockman-Schneider RA Konno S Busse WW Double-stranded RNA induces the synthesis of specific chemokines by bronchial epithelial cells Am J Respir Cell Mol Biol 2003 28 6 731 737 12600836 10.1165/rcmb.2002-0055OC
Yamaya M Sasaki H Rhinovirus and asthma Viral Immunol 2003 16 2 99 109 12828863 10.1089/088282403322017857
Papadopoulos NG Bates PJ Bardin PG Papi A Leir SH Fraenkel DJ Meyer J Lackie PM Sanderson G Holgate ST Johnston SL Rhinoviruses infect the lower airways J Infect Dis 2000 181 6 1875 1884 10837165 10.1086/315513
Schroth MK Grimm E Frindt P Galagan DM Konno SI Love R Gern JE Rhinovirus replication causes RANTES production in primary bronchial epithelial cells Am J Respir Cell Mol Biol 1999 20 6 1220 1228 10340941
Subauste MC Jacoby DB Richards SM Proud D Infection of a human respiratory epithelial cell line with rhinovirus. Induction of cytokine release and modulation of susceptibility to infection by cytokine exposure J Clin Invest 1995 96 1 549 557 7615827
Papadopoulos NG Papi A Meyer J Stanciu LA Salvi S Holgate ST Johnston SL Rhinovirus infection up-regulates eotaxin and eotaxin-2 expression in bronchial epithelial cells Clin Exp Allergy 2001 31 7 1060 1066 11467997 10.1046/j.1365-2222.2001.01112.x
Donninger H Glashoff R Haitchi HM Syce JA Ghildyal R van Rensburg E Bardin PG Rhinovirus induction of the CXC chemokine epithelial-neutrophil activating peptide-78 in bronchial epithelium J Infect Dis 2003 187 11 1809 1817 12751040 10.1086/375246
Spurrell JC Wiehler S Zaheer RS Sanders SP Proud D Human airway epithelial cells produce IP-10 (CXCL10) in vitro and in vivo upon rhinovirus infection Am J Physiol Lung Cell Mol Physiol 2005 289 1 L85 95 15764644 10.1152/ajplung.00397.2004
Wadsworth SJ Freyer AM Corteling RL Hall IP Biosynthesized matrix provides a key role for survival signaling in bronchial epithelial cells Am J Physiol Lung Cell Mol Physiol 2004 286 3 L596 603 14617518 10.1152/ajplung.00217.2003
Lopez-Souza N Dolganov G Dubin R Sachs LA Sassina L Sporer H Yagi S Schnurr D Boushey HA Widdicombe JH Resistance of differentiated human airway epithelium to infection by rhinovirus Am J Physiol Lung Cell Mol Physiol 2004 286 2 L373 L381 14711802 10.1152/ajplung.00300.2003
Papadopoulos NG Sanderson G Hunter J Johnston SL Rhinoviruses replicate effectively at lower airway temperatures J Med Virol 1999 58 1 100 104 10223554 10.1002/(SICI)1096-9071(199905)58:1<100::AID-JMV16>3.0.CO;2-D
Clement MV Stamenkovic I Fas and tumor necrosis factor receptor-mediated cell death: similarities and distinctions J Exp Med 1994 180 2 557 567 7519240 10.1084/jem.180.2.557
Conaldi PG Biancone L Bottelli A Wade-Evans A Racusen LC Boccellino M Orlandi V Serra C Camussi G Toniolo A HIV-1 kills renal tubular epithelial cells in vitro by triggering an apoptotic pathway involving caspase activation and Fas upregulation J Clin Invest 1998 102 12 2041 9854039
Zeymer U Fishbein MC Forrester JS Cercek B Proliferating cell nuclear antigen immunohistochemistry in rat aorta after balloon denudation. Comparison with thymidine and bromodeoxyuridine labeling Am J Pathol 1992 141 3 685 690 1381560
Papadopoulos NG Syrigou EI Bossios A Manou O Gourgiotis D Saxoni-Papageorgiou P Correlation of lymphocyte proliferating cell nuclear antigen expression with dietary cow's milk antigen load in infants with allergy to cow's milk Int Arch Allergy Immunol 1999 119 1 64 68 10341323 10.1159/000024177
Papadopoulos NG Psarras S Manoussakis E Saxoni-Papageorgiou P The role of respiratory viruses in the origin and exacerbations of asthma Curr Opin Allergy Clin Immunol 2003 3 1 39 44 12582313 10.1097/00130832-200302000-00007
Winther B Gwaltney JM JrMygind N Turner RB Hendley JO Sites of rhinovirus recovery after point inoculation of the upper airway JAMA 1986 256 3 1763 1767 3018306 10.1001/jama.256.13.1763
Winther B Gwaltney JM Hendley J Respiratory virus infections of monolayer cultures of nasal epithelial cells Am Rev Respir Dis 1990 141 4 pt 1 839 845 2158258
Yamaya M Sekizawa K Suzuki T Yamada N Furukawa M Ishizuka S Nakayama K Terajima M Numazaki Y Sasaki H Infection of human respiratory submucosal glands with rhinovirus: effects on cytokine and ICAM-1 production Am J Physiol 1999 277 2 Pt 1 L362 371 10444531
Fraenkel DJ Bardin PG Sanderson G Lampe F Johnston SL Holgate ST Immunohistochemical analysis of nasal biopsies during rhinovirus experimental colds Am J Respir Crit Care Med 1994 150 4 1130 1136 7921447
Deszcz L Gaudernak E Kuechler E Seipelt J Apoptotic events induced by human rhinovirus infection J Gen Virol 2005 86 Pt 5 1379 89 15831950 10.1099/vir.0.80754-0
Wark PA Johnston SL Bucchieri F Powell R Puddicombe S Laza-Stanca V Holgate ST Davies DE Asthmatic bronchial epithelial cells have a deficient innate immune response to infection with rhinovirus J Exp Med 2005 201 6 937 947 15781584 10.1084/jem.20041901
Gibson PG Saltos N Borgas T Airway mast cells and eosinophils correlate with clinical severity and airway hyperresponsiveness in corticosteroid-treated asthma J Allergy Clin Immunol 2000 105 4 752 759 10756226 10.1067/mai.2000.105319
Corne JM Marshall C Smith S Schreiber J Sanderson G Holgate ST Johnston SL Frequency, severity, and duration of rhinovirus infections in asthmatic and non-asthmatic individuals: a longitudinal cohort study The Lancet 2002 359 9309 831 834 11897281 10.1016/S0140-6736(02)07953-9
Salmeron S Liard R Elkharrat D Muir J Neukirch F Ellrodt A Asthma severity and adequacy of management in accident and emergency departments in France: a prospective study The Lancet 2001 358 9282 629 635 11530150 10.1016/S0140-6736(01)05779-8
Mosser AG Vrtis R Burchell L Lee WM Dick CR Weisshaar E Bock D Swenson CA Cornwell RD Meyer KC Jarjour NN Busse WW Gern JE Quantitative and Qualitative Analysis of Rhinovirus Infection in Bronchial Tissues Am J Respir Crit Care Med 2005 171 6 645 651 15591468 10.1164/rccm.200407-970OC
Xatzipsalti M Effect of immune response to rhinovirus (RV) on RV-induced bronchial epithelial inflammation, cytotoxicity, and viral load J Allergy Clin Immunol 2005 115 2, Sup 1 175 10.1016/j.jaci.2004.12.710
Oertel M Graness A Thim L Buhling F Kalbacher H Hoffmann W Trefoil factor family-peptides promote migration of human bronchial epithelial cells: synergistic effect with epidermal growth factor Am J Respir Cell Mol Biol 2001 25 4 418 424 11694446
Takami K Takuwa N Okazaki H Kobayashi M Ohtoshi T Kawasaki S Dohi M Yamamoto K Nakamura T Tanaka M Nakahara K takuwa Y Takizawa H Interferon-gamma inhibits hepatocyte growth factor-stimulated cell proliferation of human bronchial epithelial cells: upregulation of p27(kip1) cyclin-dependent kinase inhibitor Am J Respir Cell Mol Biol 2002 26 2 231 238 11804875
Puddicombe SM Polosa R Richter A Krishna MT Howarth PH Holgate ST Davies DE Involvement of the epidermal growth factor receptor in epithelial repair in asthma FASEB J 2000 14 10 1362 1374 10877829 10.1096/fj.14.10.1362
Vignola AM Chiappara G Siena L Bruno A Gagliardo R Merendino AM Polla BS Arrigo AP Bonsignore G Bousquet J Chanez P Proliferation and activation of bronchial epithelial cells in corticosteroid-dependent asthma J Allergy Clin Immunol 2001 108 5 738 746 11692098 10.1067/mai.2001.119160
Puddicombe SM Torres-Lozano C Richter A Bucchieri F Lordan JL Howarth PH Vrugt B Albers R Djukanovic R Holgate ST Wilson SJ Davies DE Increased Expression of p21(waf) cyclin-dependent kinase inhibitor in asthmatic bronchial epithelium Am J Respir Cell Mol Biol 2003 28 1 61 68 12495933 10.1165/rcmb.4715
Holtzman MJ Morton JD Shornick LP Tyner JW O'Sullivan MP Antao A Lo M Castro M Walter MJ Immunity, inflammation, and remodeling in the airway epithelial barrier: epithelial-viral-allergic paradigm Physiol Rev 2002 82 1 19 46 11773608
|
16216126
|
PMC1283981
|
CC BY
|
2021-01-04 16:36:25
|
no
|
Respir Res. 2005 Oct 10; 6(1):114
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-114
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1171622344610.1186/1465-9921-6-117ResearchGender differences in respiratory symptoms in 19-year-old adults born preterm Vrijlandt Elianne JLE [email protected] Jorrit [email protected] H Marike [email protected] Eric J [email protected] Dutch POPS-19 Collaborative Study Group* 1 Department of Pediatric Pulmonology, Beatrix Children's Hospital Groningen, UMCG University of Groningen, Hanzeplein 1 9713 GZ Groningen The Netherlands2 Department of Epidemiology and bioinformatics, University Medical Center Groningen, University of Groningen, Hanzeplein 1 9713 GZ Groningen The Netherlands2005 13 10 2005 6 1 117 117 19 2 2005 13 10 2005 Copyright © 2005 Vrijlandt et al; licensee BioMed Central Ltd.2005Vrijlandt 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.
Objective
To study the prevalence of respiratory and atopic symptoms in (young) adults born prematurely, differences between those who did and did not develop Bronchopulmonary Disease (BPD) at neonatal age and differences in respiratory health between males and females.
Methods
Design: Prospective cohort study.
Setting: Nation wide follow-up study, the Netherlands.
Participants: 690 adults (19 year old) born with a gestational age below 32 completed weeks and/or with a birth weight less than 1500 g. Controls were Dutch participants of the European Community Respiratory Health Survey (ECRHS).
Main outcome measures: Presence of wheeze, shortness of breath, asthma, hay fever and eczema using the ECRHS-questionnaire
Results
The prevalence of doctor-diagnosed asthma was significantly higher in the ex-preterms than in the general population, whereas eczema and hay fever were significant lower. Women reported more symptoms than men. Preterm women vs controls: asthma 13% vs 5% (p < 0.001); hay fever 8% vs 20% (p < 0.001); eczema 10% vs 42% (p < 0.001). Preterm men vs controls: asthma 9% vs 4% (p = 0.007); hay fever 8% vs 17% (p = 0.005); eczema 9% vs 31% (p < 0.001) Preterm women reported more wheeze and shortness of breath during exercise (sob) than controls: wheeze 30% vs 22% (p = 0.009); sob 27% vs 16% (p < 0.001); 19-year-old women with BPD reported a higher prevalence of doctor diagnosed asthma compared to controls (24% vs 5% p < 0.001) and shortness of breath during exercise (43% vs 16% p = 0.008). The prevalence of reported symptoms by men with BPD were comparable with the controls.
Conclusion
Our large follow-up study shows a higher prevalence of asthma, wheeze and shortness of breath in the prematurely born young adults. 19-year-old women reported more respiratory symptoms than men. Compared to the general population atopic diseases as hay fever and eczema were reported less often.
==== Body
Background
Neonatal respiratory distress syndrome (RDS), previously called hyaline membrane disease, is mainly seen in preterm infants. The main causative factors leading to respiratory distress in preterm infants are structural immaturity of the lungs, surfactant deficiency and surfactant dysfunction. Most infants recover from RDS. However in infants with a birth weight between 500 and 1500 gram, 3 to 43%, develop chronic lung disease (CLD), also called bronchopulmonary dysplasia (BPD)[1,2].
High rates of respiratory illnesses and other morbidities have been reported in survivors born prematurely in the 1970s and 1980s [3-5]. However reports on long term outcome of respiratory health in adolescents and young adults born prematurely are limited. This can be explained by the continuously changing approach to the treatment of the preterm infant with neonatal RDS, and BPD being a relatively young disorder firstly described about 35 years ago[1]. As the prevalence of both preterm birth and BPD is on the rise and treatment is assessable also in younger infants (from 25 weeks of gestational age), the number of survivors of prematurity will increase[6].
In most reported studies the rates of re-hospitalisation of preterm and/or (extremely) low birth weight infants during the first two years of life, approach or exceed 50%[7]. Overall, respiratory infections are the most common indication for re-hospitalisation[8]. The hygiene- hypothesis states that environmental changes in the industrialised world have lead to reduced microbial contact at an early age and thus resulted in the growing epidemic of atopic diseases as eczema and rhinoconjunctivitis. We were wondering whether the respiratory infections during early life in preterm children resulted in a low prevalence of atopic diseases later.
The aim of the study was to examine the presence or development of respiratory or atopic symptoms in the whole group of (young) adults born prematurely and specifically if premature born did encounter irreversible injuries. In addition, we studied differences in respiratory health at adulthood between those who did and did not develop BPD at neonatal age. Male gender is a risk factor for neonatal RDS, BPD and even death [9-12]. Boys with neonatal RDS seem to have more health problems than girls during the neonatal period [13]. This lead to the question whether we could find differences in respiratory health between young adult males and females.
Methods
Respiratory health was studied in adults born prematurely in a prospective cohort study. In the early eighties a nation-wide survey was started by the Division of Perinatology of the Dutch Paediatric Association. Information was collected on the incidence of very preterm and very low birth weight infants and subsequently on their outcome on mortality, morbidity and handicap [14,15]. Pre-, peri-, and neonatal data of Dutch infants born alive with a gestational age (GA) below 32 completed weeks and/or with a birth weight less than 1500 g, were collected prospectively. The study ultimately consisted of 1338 infants, constituting 94% of the eligible infants born in 1983 in the Netherlands. All 998 infants surviving the initial hospital stay were enlisted for long term follow-up. Between their birth and the follow up visit in 2002 379 children died, leaving 959 living participants at age 19. BPD was defined as clinical signs of respiratory distress, with an abnormal chest X-ray and an oxygen requirement after 28 days of age.
The European Community Respiratory Health Survey (ECRHS) questionnaire, was mailed to the 959 living participants[16]. This standardised questionnaire was used to assess the prevalence of respiratory symptoms and asthma, in relation to well-known (environmental) risk factors. The Dutch part of the ECRHS data in the youngest age group (20–45 years) was used as control group (644 male and 666 female randomly selected subjects from the general population). The study has been approved by the Ethical Committee of TNO Leiden, the Netherlands.
Statistical analysis
Data were analysed using the Chi2test to compare the prevalence of respiratory symptoms among both ex-preterms and those of the general population. If numbers were too small to use Chi2test, we made use of Fisher's exact test. We studied the independent effects of birth weight, gestational age, gender, duration of mechanical ventilation, smoking habits of the parents during the youth of the child and family history of atopic diseases and asthma on the presence of wheeze using multiple logistic regression analysis. Likewise we studied the independent effect of these potential risk factors on the presence of asthma, shortness of breath (with or without exercise), hay fever and eczema respectively. All statistical procedures were performed using SPSS 10.0. P-values less than 0.05 were considered to be significant (2 sided tests).
Results
The overall response rate was 72% (n = 690). Patient characteristics are shown in Table 1. The non-responders were more likely to be male, have foreign nationalities, lower social-economic status, disabilities and lower school performances at an earlier age than the responders. No differences in birth weight, gestational age or duration of mechanical ventilation were assessed.
Table 1 Characteristics participants "Project On Preterm and Small for gestational age infants" (POPS) followed up at the age of 19 year
total followed-up Gestational age ≤ 32 weeks Gestational age >32 weeks
n = 690 n = 508 n = 182
Gestational age (weeks), mean (range), SD 31 (26–41) ± 2.5 30 (26–32) ± 1.5 34 (32–41) ± 1.5
Birthweight (grams) mean (range), SD 1309 (560–2580) ± 293 1320 (560–2580) ± 325 1277 (600–1495) ± 175
male/female (%) 320/370 (46/54) 244/264 (48/52) 76/106 (42/58)
BPD (yes/no/unknown) (n) 58/505/127 55/362/91 3/143/36
duration mechanical ventilation (days) mean (range), SD 4.5 (0–55) ± 8.1 5.7 (0–55) ± 8.9 1.1 (0–41) ± 3.8
Premature born
The results of the analyses of the ex-preterms (GA ≤ 32 weeks) are shown in Table 2. The prevalence of doctor-diagnosed asthma and shortness of breath during exercise was significantly higher in the preterm than in the general population, whereas eczema and hay fever were significantly lower. The premature born women reported more symptoms like wheeze than the controls. Women with a birth weight less than 1500 gram (GA >32 weeks) reported more often wheeze and shortness of breath but less allergy and eczema than the female controls. We found no such differences in males.
Table 2 Prevalences of symptoms. The children born with a gestational age (GA) ≤ 32 weeks were compared with those born with a GA >32 weeks (birth weight (bw) <1500 g) and with controls according to gender
Symptom GA ≤ 32 w GA >32 w controls p-value GA ≤ 32 w
vs controls p-value GA >32 w
vs controls p-value GA ≤ 32 w
vs GA >32w
yes (%) yes (%) yes (%)
Have you had wheezing in
your chest at any time in
the last twelve months? females 79(29.9) 32(30.5) 145(21.8) 0.009 0.05 0.9
males 42(17.4) 15(19.7) 122(18.9) 0.6 0.9 0.6
Have you had this wheezing
when you did not have cold? females 35(13.3) 13(12.4) 87(13.1) 0.9 0.8 0.8
males 25(10.4) 8(10.5) 73(11.3) 0.7 0.8 0.96
Are you troubled by shortness
of breath when Hurrying on level ground
or walking up a slight hill? females 60(26.8) 28(34.1) 104(16.3) <0.001 <0.001 0.2
males 19(10.4) 8(12.1) 61(9.9) 0.8 0.5 0.7
Do you get short of breath
walking with other people of your
own age on level ground? females 17(7.5) 6(7.5) 11(1.7) <0.001 0.001 1.0
males 5(2.8) 2(3) 9(1.5) 0.23 0.3 1.0
Do you have to stop
for breath when walking at your
own pace on level ground? females 12(5.3) 4(5.1) 1(0.2) <0.001 <0.001 0.9
males 6(3.3) 0(0) 4(0.6) 0.004* 0.5 0.3
Have you ever had asthma? females 33(12.7) 7(6.7) 31(4.7) <0.001 0.3 0.1
males 21(9) 12(16) 28(4.3) 0.007 <0.001 0.08
Have you had an attack of asthma
in the last twelve months? females 12(4.6) 4(3.8) 15(2.3) 0.05 0.33 0.9
males 5(2.4) 2(2.7) 2(0.3) 0.004 0.009 0.7
Do you have hay fever? females 20(7.6) 8(7.5) 135(20.4) <0.001 0.002 0.9
males 23(8.4) 8(10.5) 109(17) 0.005 0.15 0.8
Do you have eczema? females 26(9.8) 9(8.5) 276(41.7) <0.001 <0.001 0.7
males 23(9.4) 8(10.5) 196(30.5) <0.001 <0.001 0.8
BPD
111 Children developed BPD (8.2%); 28 of them (25%) died. Boys (n = 72) were more prone to develop BPD than girls (n = 39). The response rate among BPD-patients was 69%. Since the number of BPD patients with a GA >32 weeks was only 3, we decided to analyse the results of the patients with a GA = 32 weeks (table 3). Compared to female controls, 19-year-old females with BPD reported a higher prevalence of doctor-diagnosed asthma, wheeze and shortness of breath during exercise. The BPD males reported significant less hay fever and eczema than the male controls.
Table 3 Prevalence of symptoms in participants with a gestational age ≤ 32 weeks with & without BPD and controls according to gender
Symptom BPD No BPD controls p BPD p no BPD vs PBPD vs
yes (%) yes (%) yes (%) vs controls vs controls no BPD
Have you had wheezing
in your chest at any time in
the last twelve months? females 7 (41.1) 58 (29.5) 145 (21.8) 0.06 0.025 0.3
males 9 (23.7) 25 (15.3) 122 (18.9) 0.18 0.28 0.2
Have you had this wheezing
when you did not have cold? females 6 (35.3) 23 (11.7) 87 (13.1) 0.008 0.6 0.006
males 4 (10.8) 15 (9.2) 73 (11.3) 0.9 0.4 0.7
Are you troubled by shortness
of breath when Hurrying on level
ground or walking up a slight hill? females 6 (42.9) 41 (20.8) 104 (16.3) 0.008 0.01 0.13
males 3 (12.0) 13 (10.2) 61 (9.9) 0.7 0.9 0.7
Do you get short of breath
walking with other people of
your own age on level ground? females 0 (0) 12 (7.1) 11 (1.7) 0.6 <0.001 0.6
males 0 (0) 5 (3.9) 9 (1.5) 0.5 0.06 0.6
Do you have to stop
for breath when walking at your
own pace on level ground? females 0 (0) 6 (3.6) 1 (0.2) 0.8 <0.001 1.0
males 0 (0) 6 (4.7) 4 (0.6) 0.7 <0.001 0.6
Have you ever had asthma? females 4 (23.5) 23 (12) 31 (4.7) <0.001 <0.001 0.2
males 3 (8.1) 13 (8.2) 28 (4.3) 0.28 0.05 1.0
Have you had an attack of
asthma in the last twelve months? females 1 (6.25) 10 (17.2) 15 (2.3) 0.32 0.007 1.0
males 0 (0) 4 (2.5) 2 (0.3) 0.73 0.004 0.5
Do you have hay fever? females 1 (5.9) 16 (8.1) 135 (20.4) 0.13 <0.001 0.7
males 1 (2.6) 20 (12) 109 (17) 0.02 0.1 0.1
Do you have eczema? females 5 (29.4) 18 (9.1) 276 (41.7) 0.3 <0.001 0.01
males 3 (7.9) 13 (7.9) 196 (30.5) 0.002 <0.001 1.0
Respiratory symptoms and atopy
In regression analyses dyspnea, asthma, wheeze, dyspnea on exertion, hay fever and eczema were assessed as outcome parameters. Dyspnea was significantly related to long term mechanical ventilation and BPD, maternal asthma and current smoking. An inverse relation was found with gestational age. Asthma was significantly related to maternal asthma. Wheeze was significantly related to female gender and current smoking habits and tended to be related to maternal smoking during the youth of the participant. Shortness of breath during exercise was related to female gender and smoking in the past. We found no significant associations of birth weight, gestational age, duration of mechanical ventilation, gender, smoking habits or BPD to hay fever and eczema (see table 4). Young adults with recurrent respiratory infections in infancy reported more asthmatic symptoms than those without respiratory infections (p < 0.001). No significant differences were found between recurrent respiratory infections and hayfever or eczema. Young adults with sepsis during the neonatal period reported less hayfever than those without sepsis (p = 0.03), but no significant differences were found between sepsis and asthma or eczema.
Table 4 Odds ratios (95% confidence intervals) for respiratory symptoms, hay fever and eczema, determined by multiple regression analysis. Significant relations are printed in bold. Birth weight, gestational age, duration of mechanical ventilation and smoking habits are entered as categorical covariates.
dyspnea asthma wheeze SOBDE* hayfever eczema
birth weight (gram) 500–1000 0.4 (0.2–1.1) 0.6 (0.2–2.4) 1.6 (0.7–4.0) 0.6 (0.2–1.9) 1.2 (0.3–4.4) 1.2 (0.3–4.5)
1000–1500 0.5 (0.3–1.2) 0.8 (0.3–2.0) 1.7 (0.8–3.6) 1.0 (0.4–2.3) 1.4 (0.5–3.6) 1.5 (0.6–4.3)
Gestational age (weeks) till 28 0.4 (0.2–0.9) † 0.8 (0.2–2.6) 0.9 (0.4–1.9) 1.1 (0.5–2.7) 1.3 (0.4–4.2) 0.9 (0.3–2.7)
28–31 0.5 (0.2–0.9) † 1.1 (0.4–2.6) 1.0 (0.6–1.9) 0.7 (0.3–1.4) 1.2 (0.5–2.9) 0.9 (0.4–2.3)
Mechanical ventilation (days) 1–7 days 1.1 (0.6–216) 1.3 (0.6–3.1) 1.0 (0.5–1.9) 1.1 (0.5–2.2) 1.4 (0.6–3.2) 1.21 (0.5–2.8)
8–28 days 0.9 (0.4–2.2) 0.3 (0.1–1.5) 1.2 (0.6–2.6) 0.7 (0.3–1.9) 0.7 (0.2–2.3) 1.1 (0.4–3.4)
>28 5.2 (1.2–23.3) † 0.3 (0.0–4.0) 1.6 (0.4–6.9) 0.2 (0.2–2.7) 0.7 (0.1–7.7) 0.4 (0.0–3.7)
female gender 1.6 (1.0–2.7) 1.2 (0.6–2.3) 2.0 (1.2–3.2) † 3.8 (2.0–7.3) ‡ 0.7 (0.1–7.7) 1.3 (0.7–2.5)
Maternal smoking 1.1 (0.6–1.9) 1.6 (0.7–3.5) 1.6 (1.0–2.8) 1.3 (0.7–2.5) 0.6 (0.3–1.3) 1.0 (0.5–2.1)
Maternal asthma 2.5 (1.2–5.4) † 4.2 (1.8–10.5) ‡ 1.3 (0.6–2.9) 1.6 (0.6–4.1) 1.8 (0.6–5.2) 2.0 (0.7–5.3)
BPD 3.1 (1.2–8.2) † 3.1 (0.7–14.6) 1.5 (0.6–3.6) 2.0 (0.6–6.9) 0.4 (0.1–2.3) 2.4 (0.8–7.4)
Smoking participant past 1.2 (0.6–2.5) 0.6 (0.2–1.7) 0.9 (0.4–1.9) 2.2 (1.0–4.8) † 1.5 (0.6–3.6) 1.4 (0.6–3.4)
"party" 0.5 (0.1–1.5) 0.9 (0.3–2.9) 1.9 (0.8–4.0) 0.9 (0.3–2.5) 0.3 (0.0–2.3) 1.3 (0.5–3.8)
daily 2.7 (1.5–5.1) ‡ 0.4 (0.1–1.1) 2.6 (1.5–4.6) ‡ 1.5 (0.7–2.9) 1.6 (0.7–3.7) 1.0 (0.4–2.5)
*SOBDE = shortness of breath during exercise, † p < 0.05, ‡ p = 0.001
Discussion
In this long-term follow-up of ex-preterms into adulthood we found a higher prevalence of asthma, wheezing and shortness of breath during exercise in the ex-preterms (especially the women) compared to the general population. Atopy (i.e. hay fever, rhino-conjunctivitis and atopic dermatitis) was significantly lower in the ex-preterms compared with the controls. In this study, we did not perform lung function, skin prick or RAST tests to confirm the diagnoses. However, the relation between subject reported symptoms on the basis of the used ECHRS questionnaire and lung function is studied earlier. Subject reported symptoms were related to impaired lung function and to increased variability of peak flow[17].
Long-term reports on respiratory health in infants born prematurely are limited and contradictory. Respiratory health of preterm children of birth weight ≤ 1500 g at 14 years of age has been reported to be comparable to that of term controls [18]. Others found that infants born prematurely with and without a history of neonatal RDS, but who did not develop BPD, have an increased prevalence of airway hyperreactivity compared to full term controls which can persist into early adult life [3,4]. At school age bronchial obstruction and increased bronchial responsiveness have been demonstrated in prematurely born children [19].
Preterm birth and asthma
The pathophysiology of neonatal RDS is not completely understood, but it has been demonstrated that factors such as mechanical ventilation and oxygen lead to an inflammatory process, which could result in an early Th1- response. Moreover, in most reported studies the rates of re-hospitalisation of preterm and/or (extremely) low birth weight infants during the first two years of life, approach or exceed 50%[7]. Respiratory illnesses and especially respiratory infections are the most common indication for re-hospitalisation in this patient group. Also in our cohort the re-hospitalisation-rate in early childhood was high (34%)[14]. In contrast re-admission rates for normal birth weight infants are much lower (about 20%)[7]. Asthma is often characterised by symptoms like shortness of breath and wheeze; reversible airway obstruction; airway hyper-responsiveness and airway inflammation. In children and young adults, asthma is associated with atopy through IgE-dependent mechanisms, and airway-inflammation is partly related to helper T type 2 (Th2) lymphocytes and eosinophil mediation [20]. Preterm born adults report asthma-like symptoms, but less allergy compared to controls. Decreased risk of atopy is also found in a Finnish prospective birth cohort study comparing term and preterm adults: high gestational age increased the risk of atopy at the age of 31[21]. The early Th1- response, in combination with serious infections in the first two years of life, could be an explanation for the lower prevalence of atopy, which is in line with the hygiene-hypothesis [22]. Others found that children who were septic in the neonatal period were less likely to have asthma[23]. We could not confirm this. However, young adults who were septic during the neonatal period did report less hayfever. In our study, adults with recurrent respiratory infections in infancy did not report less but more asthma. This is remarkable considering that early exposure to endotoxins or other allergens enhance Th1-type cytokine responses tip the balance away from Th2-type responses that favour the development of allergic diseases including asthma[24]. The high rate of respiratory symptoms might be due to sustained increased vulnerability of the immature airways in a way that mimics asthma, but is not exactly the same.
Gender
Male gender is a risk factor for neonatal RDS and BPD [9-11]. Boys with neonatal RDS seem to have more health problems than girls during the neonatal period and school age[13,23]. However, long-term outcome shows gender differences in e.g. school-performances, but not in respiratory health. We found that particularly women reported symptoms as wheeze and shortness of breath. In the 'general' population both incidence and prevalence of wheeze and asthma is higher in males than in females until the age of 16 year [25,26]. In adulthood, asthma occurs more frequently among women[25,26]. The observed variation between males and females in the general population has partly been explained by dys-synnaptic lung growth: the independent growth of the airways in comparison with the lung parenchyma and air spaces. In girls, growth of the airways is proportional to growth of lung parenchyma, whereas in boys growth of the airways lags behind that of lung parenchyma, causing a discrepancy between airway and lung size [27]. Different pubertal patterns of thoracic growth between the sexes results in an approximately 25% higher lung function in males than in females of identical height at the end of puberty. We speculate that a similar process takes place in the preterm born population, although the underlying mechanism is not understood. Another explanation might be that large individual differences exist in physical symptom reports. Women may require a greater amount of cognitive analysis (and thus more attention) to make judgements about physical symptoms compared to men[28].
There have been few reports of respiratory health during exercise. Our finding of a high percentage of participants that reported shortness of breath during exercise, is in agreement with a study showing low oxygen consumption in low birth weight children compared to children with a normal birth weight [29]. The authors suggested that extremely low birth weight children have a lower level of fitness than controls.
BPD
Airway obstruction and airway hyper-reactivity persisted in children and adolescents with BPD [3,4,30]. Long-term studies in children who had BPD as infants showed persisting lung function abnormalities consisting of airway obstruction, airway hyper-reactivity, and hyperinflation[5,31]. Both BPD and asthma are characterised by increased smooth muscle contraction and symptoms of both diseases are therefore perhaps difficult to distinguish. As stated above, airway inflammation is an important feature in children and adults with asthma. Studies showed that inflammation plays an important role in the pathogenesis of BPD. Contrary to asthma, however, the BAL-fluid reflects a Th1-cell subtype[32,33]. Even more than preterm infants without BPD, infants with BPD are likely to be re-hospitalised early in childhood with a respiratory illness[8,34]. The same mechanism as described above could be an explanation for the low prevalence of hay fever and eczema, despite the asthma-like symptoms.
Analysis of risk factors for respiratory symptoms
The regression analysis confirmed the association between dyspnea and respectively long-term mechanical ventilation, BPD and smoking of the participant. We expected to find high risks for respiratory symptoms in the young adults with a (very) low birth weight or born (very) prematurely due to the immaturity of the airways at birth. However, the degree of prematurity or dysmaturity did not increase the risk at all. As a matter of fact, the risk for dyspnea was even lower in the children born very prematurely. In seeking to understand this we speculate that these young adults have a bias toward symptom detection and the feeling of distress because they are used to physical limitations. Future research should investigate the extend to which physical symptoms correlate with lung function abnormalities.
A limitation of our study might be that the age range of the preterms and the general population sample is not exactly the same. However, the prevalence of respiratory symptoms is probably increasing with age. Therefore, the differences might even be more obvious when the results of young adults born prematurely could be compared with peers from the general population. As the complete cohort was inhomogeneous in the sense that it consisted of either preterm or small for gestational age infants, we choose to analyse the data of the preterm children (GA ≤ 32 weeks). The possibility that symptoms will disappear and that ex-preterms will "grow out" of their disease after adolescence is likely to be very small because the lungs stop growing and developing after that age. It might even be possible that symptoms come back or become more severe during adulthood, as has been observed in long-term follow-up of asthma[35].
Conclusion
Our study clearly demonstrated that more than a third of young adults born preterm suffer from respiratory symptoms (higher prevalence of asthma, wheeze and shortness of breath) and need more medical care than peers. Not only paediatricians, but also family doctors and chest physicians should be aware of this 'new' group of patients in which respiratory symptoms will never disappear. Especially women seem to be more vulnerable on their way to adulthood and report more respiratory symptoms than controls. Future research should investigate to what extend physical symptoms correlate with lung function abnormalities. On the other hand, our findings are encouraging because a lot of young adults born preterm, survive with no or only minor respiratory problems and compared to the general population atopic diseases as hay fever and eczema were reported less often.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
EV participated in design and co-ordination of the study, analysis & interpretation of the data and drafting of the article.
JG and ED have made substantial contributions to the design of the study, the interpretation of data and drafting the article
HB has made substantial contributions to (statistical) analysis, interpretation of the data and drafting the article
All authors read and approved the final manuscript
Funding
major funding was provided by the "Stichting Astmabestrijding"
Note
* Participants of the Dutch POPS-19 Collaborative Study Group:
TNO Prevention and Health, Leiden (ETM Hille, CH de Groot, H Kloosterboer-Boerrigter, AL den Ouden, A Rijpstra, SP Verloove-Vanhorick, JA Vogelaar); Emma Children's Hospital AMC, Amsterdam (JH Kok, A Ilsen, M van der Lans, WJC Boelen-van der Loo, T Lundqvist, HSA Heymans); University Hospital Groningen, Beatrix Children's Hospital, Groningen (EJ Duiverman, WB Geven, ML Duiverman, LI Geven, EJLE Vrijlandt); University Hospital Maastricht, Maastricht (ALM Mulder, A Gerver); University Medical Center St Radboud, Nijmegen (LAA Kollée, L Reijmers, R Sonnemans); Leiden University Medical Center, Leiden (JM Wit, FW Dekker, MJJ Finken); Erasmus MC – Sophia Children's Hospital, University Medical Center Rotterdam (N Weisglas-Kuperus, MG Keijzer-Veen, AJ van der Heijden, JB van Goudoever); VU University Medical Center, Amsterdam (MM van Weissenbruch, A Cranendonk, HA Delemarre-van de Waal, L de Groot, JF Samsom); Wilhelmina Children's Hospital, UMC, Utrecht (LS de Vries, KJ Rademaker, E Moerman, M Voogsgeerd); Máxima Medical Center, Veldhoven (MJK de Kleine, P Andriessen, CCM Dielissen-van Helvoirt, I Mohamed); Isala Clinics, Zwolle (HLM van Straaten, W Baerts, GW Veneklaas Slots-Kloosterboer, EMJ Tuller-Pikkemaat); Royal Effatha Guyot Group, Zoetermeer (MH Ens-Dokkum); Association for Parents of Premature Babies (GJ van Steenbrugge).
Acknowledgements
The POPS study at 19 years of age was supported by grants from the Netherlands Organisation for Health Research and Development (ZonMw), Edgar Doncker Foundation, Foundation for Public Health Fundraising Campaigns, Phelps Foundation, Swart-van Essen Foundation, Foundation for Children's Welfare Stamps, TNO Prevention and Health, Netherlands Organisation for Scientific Research (NWO), Dutch Kidney Foundation, Sophia Foundation for Medical Research, Stichting Astmabestrijding, Royal Effatha Guyot group.
==== Refs
Northway WHJ Rosan RC Porter DY Pulmonary disease following respirator therapy of hyaline-membrane disease. Bronchopulmonary dysplasia N Engl J Med 1967 276 357 368 5334613
Bancalari E Claure N Sosenko IR Bronchopulmonary dysplasia: changes in pathogenesis, epidemiology and definition Semin Neonatol 2003 8 63 71 12667831 10.1016/S1084-2756(02)00192-6
Bertrand JM Riley SP Popkin J Coates AL The long-term pulmonary sequelae of prematurity: the role of familial airway hyperreactivity and the respiratory distress syndrome N Engl J Med 1985 312 742 745 3974651
MacLusky IB Stringer D Zarfen J Smallhorn J Levison H Cardiorespiratory status in long-term survivors of prematurity, with and without hyaline membrane disease Pediatr Pulmonol 1986 2 94 102 3714346
Northway WHJ Moss RB Carlisle KB Parker BR Popp RL Pitlick PT Eichler I Lamm RL Brown BWJ Late pulmonary sequelae of bronchopulmonary dysplasia N Engl J Med 1990 323 1793 1799 2247118
Manktelow BN Draper ES Annamalai S Field D Factors affecting the incidence of chronic lung disease of prematurity in 1987, 1992, and 1997 Arch Dis Child Fetal Neonatal Ed 2001 85 F33 F35 11420319 10.1136/fn.85.1.F33
Doyle LW Ford G Davis N Health and hospitalistions after discharge in extremely low birth weight infants Semin Neonatol 2003 8 137 145 15001150 10.1016/S1084-2756(02)00221-X
Chye JK Gray PH Rehospitalization and growth of infants with bronchopulmonary dysplasia: a matched control study J Paediatr Child Health 1995 31 105 111 7794609
Bancalari E Gonzalez A Bland RD and Coalson JJ Clinical course and lung function abnormalities during development of neonatal chronic lung disease Chronic Lung Disease in Early Infancy Lung Biology in health and disease Lenfant C 2000 3 1 New York; Basel, Marcel Dekker, Inc 41 64
Farrell PM Wood RE Epidemiology of hyaline membrane disease in the United States: analysis of national mortality statistics Pediatrics 1976 58 167 176 951131
Khoury MJ Marks JS McCarthy BJ Zaro SM Factors affecting the sex differential in neonatal mortality: the role of respiratory distress syndrome Am J Obstet Gynecol 1985 151 777 782 3976790
Stevenson DK Wright LL Lemons JA Oh W Korones SB Papile LA Bauer CR Stoll BJ Tyson JE Shankaran S Fanaroff AA Donovan EF Ehrenkranz RA Verter J Very low birth weight outcomes of the National Institute of Child Health and Human Development Neonatal Research Network, January 1993 through December 1994 Am J Obstet Gynecol 1998 179 1632 1639 9855609
Elsmen E Hansen Pupp I Hellstrom-Westas L Preterm male infants need more initial respiratory and circulatory support than female infants Acta Paediatr 2004 93 529 533 15188982 10.1080/08035250410024998
DM ZA Verloove-Vanhorick SP Brand R Ruys JH The use of health services in the first 2 years of life in a nationwide cohort of very preterm and/or very low birthweight infants in The Netherlands: rehospitalisation and out-patient care Paediatr Perinat Epidemiol 1991 5 11 26 2000331
Walther FJ den Ouden AL Verloove-Vanhorick SP Looking back in time: outcome of a national cohort of very preterm infants born in The Netherlands in 1983 Early Hum Dev 2000 59 175 191 10996273 10.1016/S0378-3782(00)00094-3
Rijcken B Kerkhof M de Graaf A Boezen HM Droste JHJ Kremer AM Europees Luchtweg Onderzoek Nederland 1996 Rijksuniversiteit Groningen, Epidemiologie 29 33
Boezen HM Schouten JP Postma DS Rijcken B Relation between respiratory symptoms, pulmonary function and peak flow variability in adults Thorax 1995 50 121 126 7701448
Doyle LW Cheung MMH Ford GW Olinsky A Davis NM Callanan C Birth weight < 1501 g and respiratory health at age 14 Archives of Disease in Childhood 2001 84 40 44 11124782 10.1136/adc.84.1.40
Pelkonen AS Hakulinen AL Turpeinen M Bronchial lability and responsiveness in school children born very preterm Am J Respir Crit Care Med 1997 156 1178 1184 9351619
Vignola AM La Grutta S Chiappara G Benkeder A Bellia V Bonsignore G Cellular network in airways inflammation and remodelling Paediatr Respir Rev 2002 3 41 46 12065181 10.1053/prrv.2001.0166
Pekkanen J Xu B Jarvelin MR Gestational age and occurrence of atopy at age 31--a prospective birth cohort study in Finland Clin Exp Allergy 2001 31 95 102 11167956 10.1046/j.1365-2222.2001.00930.x
Strachan DP Hay fever, hygiene, and household size BMJ 1989 299 1259 1260 2513902
Grischkan J Storfer-Isser A Rosen CL Larkin EK Kirchner HL South A Wilson-Costello DC Martin RJ Redline S Variation in childhood asthma among former preterm infants J Pediatr 2004 144 321 326 15001935 10.1016/j.jpeds.2003.11.029
Martinez FD Maturation of a hypothesis Mediators Inflamm 2001 10 306 307 11817670 10.1080/09629350152700984
Sennhauser FH Kuhni CE Prevalence of respiratory symptoms in Swiss children: is bronchial asthma really more prevalent in boys? Pediatr Pulmonol 1995 19 161 166 7792118
Yunginger JW Reed CE O'Connell EJ Melton LJIII O'Fallon WM Silverstein MD A community-based study of the epidemiology of asthma. Incidence rates, 1964-1983 Am Rev Respir Dis 1992 146 888 894 1416415
Boezen HM Jansen DF Postma DS Sex and gender differences in lung development and their clinical significance Clin Chest Med 2004 25 237 245 15099885 10.1016/j.ccm.2004.01.012
Williams PG Wiebe DJ Individual differences in self-assessed health: gender, neuroticism and physical symptom reports Personality and individual differences 1999 28 823 835 10.1016/S0191-8869(99)00140-3
Kilbride HW Gelat MC Sabath RJ Pulmonary function and exercise capacity for ELBW survivors in preadolescence: Effect of neonatal chronic lung disease Journal of Pediatrics 2003 143 488 493 14571227 10.1067/S0022-3476(03)00413-X
Koumbourlis AC Motoyama EK Mutich RL Mallory GB Walczak SA Fertal K Longitudinal follow-up of lung function from childhood to adolescence in prematurely born patients with neonatal chronic lung disease Pediatr Pulmonol 1996 21 28 34 8776263 10.1002/(SICI)1099-0496(199601)21:1<28::AID-PPUL5>3.0.CO;2-M
Allen J Zwerdling R Ehrenkranz R Gaultier C Geggel R Greenough A Kleinman R Klijanowicz A Martinez F Ozdemir A Panitch HB Nickerson B Stein MT Tomezsko J Van Der Anker J Statement on the care of the child with chronic lung disease of infancy and childhood Am J Respir Crit Care Med 2003 168 356 396 12888611 10.1164/rccm.168.3.356
Papoff P Christensen RD Calhoun DA Juul SE Granulocyte colony-stimulating factor, granulocyte macrophage colony-stimulating factor and neutrophils in the bronchoalveolar lavage fluid of premature infants with respiratory distress syndrome Biol Neonate 2001 80 133 141 11509813 10.1159/000047132
Pierce MR Bancalari E The role of inflammation in the pathogenesis of bronchopulmonary dysplasia Pediatr Pulmonol 1995 19 371 378 7567218
Sauve RS Singhal N Long-term morbidity of infants with bronchopulmonary dysplasia Pediatrics 1985 76 725 733 4058982
Vonk JM Postma DS Boezen HM Grol MH Schouten JP Koeter GH Gerritsen J Childhood factors associated with asthma remission after 30 year follow up Thorax 2004 59 925 929 15516465 10.1136/thx.2003.016246
|
16223446
|
PMC1283982
|
CC BY
|
2021-01-04 16:36:27
|
no
|
Respir Res. 2005 Oct 13; 6(1):117
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-117
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1211624203610.1186/1465-9921-6-121ResearchSub-chronic inhalation of high concentrations of manganese sulfate induces lower airway pathology in rhesus monkeys Dorman David C [email protected] Melanie F [email protected] Elizabeth A [email protected] Brian A [email protected] Paul C [email protected] CIIT Centers for Health Research, 6 Davis Drive, P.O. Box 12137, Research Triangle Park, NC 27709-2137, USA2 Experimental Pathology Laboratories, Inc., P.O. Box 12766, Research Triangle Park, NC 27709, USA2005 21 10 2005 6 1 121 121 30 3 2005 21 10 2005 Copyright © 2005 Dorman et al; licensee BioMed Central Ltd.2005Dorman 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
Neurotoxicity and pulmonary dysfunction are well-recognized problems associated with prolonged human exposure to high concentrations of airborne manganese. Surprisingly, histological characterization of pulmonary responses induced by manganese remains incomplete. The primary objective of this study was to characterize histologic changes in the monkey respiratory tract following manganese inhalation.
Methods
Subchronic (6 hr/day, 5 days/week) inhalation exposure of young male rhesus monkeys to manganese sulfate was performed. One cohort of monkeys (n = 4–6 animals/exposure concentration) was exposed to air or manganese sulfate at 0.06, 0.3, or 1.5 mg Mn/m3 for 65 exposure days. Another eight monkeys were exposed to manganese sulfate at 1.5 mg Mn/m3 for 65 exposure days and held for 45 or 90 days before evaluation. A second cohort (n = 4 monkeys per time point) was exposed to manganese sulfate at 1.5 mg Mn/m3 and evaluated after 15 or 33 exposure days. Evaluations included measurement of lung manganese concentrations and evaluation of respiratory histologic changes. Tissue manganese concentrations were compared for the exposure and control groups by tests for homogeneity of variance, analysis of variance, followed by Dunnett's multiple comparison. Histopathological findings were evaluated using a Pearson's Chi-Square test.
Results
Animals exposed to manganese sulfate at ≥0.3 mg Mn/m3 for 65 days had increased lung manganese concentrations. Exposure to manganese sulfate at 1.5 mg Mn/m3 for ≥15 exposure days resulted in increased lung manganese concentrations, mild subacute bronchiolitis, alveolar duct inflammation, and proliferation of bronchus-associated lymphoid tissue. Bronchiolitis and alveolar duct inflammatory changes were absent 45 days post-exposure, suggesting that these lesions are reversible upon cessation of subchronic high-dose manganese exposure.
Conclusion
High-dose subchronic manganese sulfate inhalation is associated with increased lung manganese concentrations and small airway inflammatory changes in the absence of observable clinical signs. Subchronic exposure to manganese sulfate at exposure concentrations (≤0.3 mg Mn/m3) similar to the current 8-hr occupational threshold limit value established for inhaled manganese was not associated with pulmonary pathology.
==== Body
Background
There is growing evidence to suggest that a wide variety of respirable particles can induce lung injury under certain exposure conditions. Clinical recognition of this hazard originally stemmed from occupational studies examining workplace exposure to metals, asbestos, silica, coal, and other biologically active particles [1,2]. However, particulate-induced lung injury is not confined to the workplace. There is strong epidemiologic evidence for significant associations between respiratory morbidity, including exacerbations of asthma and mortality, with exposure to relatively low ambient particulate matter concentrations [3,4]. This association has contributed to the adoption of more stringent ambient air quality standards for respirable particulate matter by the United States Environmental Protection Agency and other health organizations.
Particulate matter is not a single entity but rather a mixture of many subclasses of pollutants including metals, sulfate, acids, ammonium, nitrate, organic compounds, and minerals. Depending on emission source, metals may represent a significant proportion of a particulate matter sample [5]. Soluble metals have been implicated in particulate matter-associated cardiopulmonary disease in healthy and compromised individuals [6-8]. One metal found in ambient air is manganese. Airborne manganese sources include wind erosion of dusts and soils, anthropogenic fugitive dusts, and emissions from automobiles, power plants, coke ovens, municipal waste incinerators, and metal smelting operations [5]. Ambient air manganese concentrations are typically quite low, ranging between 5 and 33 ng Mn/m3 [9]. Significant occupational manganese exposure (≥0.2 mg Mn/m3) can occur in some workers involved in ferroalloy production, iron and steel foundries, and welding [9].
Workers exposed to high atmospheric manganese concentrations frequently demonstrate an increased incidence of cough and other signs associated with bronchitis [10,11]. Acute inhalation of air with extremely high manganese concentrations (≥1 mg Mn/m3) can result in pneumonitis [12]. Although manganese-induced pneumonitis has been recognized since the mid-1940's, histological characterization of the pulmonary response remains incomplete. Most experimental exposures in laboratory animals have shown only minor lung pathology despite the administration of high doses of manganese oxides by intratracheal instillation or inhalation [13-18]. Far less is known about the potential respiratory effects induced by exposure to other inorganic forms of manganese. Manganese chloride instillation into rabbits failed to induce significant pulmonary pathology [19]. Dorman et al. [20] showed that subchronic exposure of rats to manganese sulfate (MnSO4) was associated with rhinitis in the anterior part of the nose. To our knowledge, histologic assessment of pulmonary changes occurring in association with subchronic inhalation of MnSO4 has not been performed and is the subject of our study. Herein, we report that subchronic high-dose inhalation exposure of nonhuman primates to MnSO4 resulted in mild subacute bronchiolitis, alveolar duct inflammation, and proliferation of bronchus-associated lymphoid tissue (BALT), in the absence of rhinitis or other forms of nasal pathology.
Methods
Chemicals
Manganese (II) sulfate monohydrate (MnSO4·H2O) was obtained from Aldrich Chemical Company, Inc. (Milwaukee, WI).
Animals
This study was conducted under federal guidelines for the care and use of laboratory animals [21] and was approved by the CIIT Centers for Health Research (CIIT) Institutional Animal Care and Use Committee. Additional endpoints evaluated in this study, but not presented in the present manuscript, included determination of manganese concentrations in additional tissues and magnetic resonance imaging (MRI) of the brain. We chose to use rhesus monkeys since they are extensively used in toxicology studies, manganese-exposed monkeys develop distribution patterns for this metal within the brain that mimic those seen in heavily exposed people [22], and there are anatomically-based simulation models for air flow in the macaque upper and lower respiratory tracts [23,24]. Thirty-six male rhesus monkeys purchased from Covance Research Products, Inc. (Alice, TX) were used in this study. Monkeys were 17 to 22 months old at the time of their arrival at CIIT. Animals were screened for the nasal parasite Anatrichosoma spp. herpes B, simian immunodeficiency virus, simian respiratory virus, pulmonary acariasis, and tuberculosis, and were subjected to a thorough clinical examination including an evaluation of pre-exposure blood samples for routine hematology and clinical chemistry. The results of these evaluations were within normal limits.
Assignment of animals to treatment cohorts considered the age of the animal. Assignment occurred so that the animals were between 20 and 24 months of age at the start of the inhalation exposure. Randomization of animals to treatment groups occurred prior to the start of the inhalation exposure and was based upon a weight randomization procedure. Animals were acclimated to the facility for at least 43 days prior to the start of the first inhalation exposure. Additional endpoints evaluated in this study, but not presented in the present manuscript, included magnetic resonance imaging (MRI) of the brain, post-exposure clinical chemistry and hematological evaluations, and determination of tissue manganese concentrations in the central nervous system and other organs.
Animal Husbandry
Animal rooms were maintained at daily temperatures of 22 ± 4°C, relative humidity of 30–70%, and an air flow rate sufficient to provide 10–15 air changes per hour. Lighting was controlled by automatic controls (lights on approximately 0600–1800). All exposures were conducted during the animal's light cycle (approximately from 0800 to 1400). All animals were housed in animal rooms or exposure chambers within CIIT's animal facility. This facility is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care, International. A certified primate chow (# 5048) diet from Purina Mills (St. Louis, MO) was fed twice-a-day (total daily amount fed was approximately 4% of the animal's body weight). Dietary supplements were also used as part of CIIT's nonhuman primate enrichment program. These supplements included fruits (e.g., oranges, raisins, apples), vegetables (e.g., carrots), and treats (e.g., honey, candies, cereal, fruit juices) purchased from a local grocery store. Reverse osmosis purified water was available ad libitum. During non-exposure periods, domiciliary stainless steel cages (0.4 m2 × 0.8 m tall) suitable for housing macaque monkeys (Lab Products, Inc.; Seaford, DE) were used to individually house monkeys. On each exposure day, animals were transferred to 0.2 m2 × 0.6 m tall stainless steel cages (Lab Products, Inc.; Seaford, DE) that were designed to fit within the 8-m3 inhalation chambers. Animals were moved back to their domiciliary cages after the end of each 6-hr exposure.
Manganese Exposures
MnSO4 aerosol concentrations of 0.18, 0.92, and 4.62 mg MnSO4/m3, corresponding to 0.06, 0.3, and 1.5 mg Mn/m3, were generated for this study. Control animals were exposed to filtered air. Animal exposures were conducted as described in Figure 1. Four 8-m3 stainless steel and glass inhalation exposure chambers with glass doors and windows for animal observation were used. Animal position within the inhalation chambers was rotated weekly to minimize the impact of any undetected differences in the environment or in the MnSO4 exposure concentrations. Air flow through the 8-m3 inhalation chamber was typically maintained at a rate sufficient to provide at least 12 air changes per hour. Airflow through each chamber was monitored continuously during each exposure and recorded every 30 minutes. Temperature and relative humidity inside each inhalation chamber were recorded every 30 minutes during each 6-hr exposure. The average chamber temperature and relative humidity during the 6-hr exposure period were maintained at 18–26°C and 50 ± 20%, respectively. Methods describing the generation of the MnSO4 atmosphere with a dry powder generator (Wright Dust Feeder, Model WDF-II, BGI, Inc., Waltham, MA) and the characterization of the subsequent aerosol have been previously described [25]. MnSO4 was packed in separate dry powder generator cups at pressures between 2000 and 3000 psi (Model 3912, Carver Inc., Wabash, IN) using a hydraulic press (Model C, Carver Inc., Menomonee Falls, WI).
Figure 1 Experimental design overview. Group size equals 4 monkeys per exposure group, with the exception of the 0 and 0.06 mg Mn/m3 exposure groups (n = 6 monkeys/exposure concentration). Lung manganese concentrations and respiratory histologic changes were evaluated after 15, 33, or 65 exposure days or 45 or 90 days after the 65th exposure day. #Denotes animals assigned to cohort 2.
Necropsy Procedures
Necropsies were performed the day following the last inhalation exposure (i.e., 12–18 hr after termination of the final inhalation exposure). Food was withheld overnight prior to necropsy. Monkeys were anesthetized with ketamine (20 mg/kg, IM, Fort Dodge Animal Health, Fort Dodge, IA) and euthanized with pentobarbital (80–150 mg/kg, IV, Henry Schein Inc., Port Washington, NY) followed by exsanguination. Following euthanasia, the lungs and other thoracic organs were removed, weighed, and inspected for gross lesions. The left primary bronchus was ligated and the left lung separated for determination of tissue manganese concentration. The right lung and trachea were then inflated with 10% neutral-buffered formalin using 30-cm of hydrostatic pressure [26]. The olfactory epithelium was excised for chemical analysis and the remaining nasal tissues, including the nasopharynx and larynx, were stored in 10% neutral-buffered formalin. The larynx samples were decalcified in 10% formic acid (Fisher Scientific International, Inc., Hampton, NH) for two days. Cranial tissues were decalcified in RDO® (Apex Engineering Products Corporation, Plainfield, IL) for up to 6 days. Following decalcification, tissues were washed in running tap water for at least 6 hr. Following fixation (and decalcification when appropriate), representative samples of the lung, trachea, larynx, oropharynx, tracheobronchial lymph nodes, and nose were collected from each animal. Tissue samples were trimmed, embedded in paraffin, and five-μm thick sections were cut and stained with hematoxylin and eosin for light microscopic evaluation. Histologic specimens were examined by an experienced veterinary pathologist (Howroyd).
Tissue manganese concentrations
Tissue samples collected for chemical analyses were stored in individual plastic containers, frozen in liquid nitrogen, and stored at approximately -80°C until chemical analyses were performed. Lung and olfactory epithelium manganese concentrations were determined by graphite furnace atomic absorption spectrometry using previously published methods [27].
Statistics
Tissue manganese concentrations were compared for the exposure and control groups by tests for homogeneity of variance (Levene's test), analysis of variance (ANOVA), and Dunnett's multiple comparison procedure for significant ANOVA. Histopathological findings were evaluated using a Chi-Square test. Statistical analyses were performed using SAS Statistical Software. A probability value of <0.01 was used for Levene's test, while <0.05 was used as the critical level of significance for all other statistical tests. Unless otherwise noted, data presented are mean values ± standard error of the mean (SEM).
Results
Test atmospheres
No significant differences in the test aerosol characteristics were observed between the two exposure cohorts (Table 1). Particles of unknown composition (arising from animal dander and other background sources) were present in the control chamber at an overall average concentration ± standard deviation (SD) of 0.004 ± 0.002 mg/m3. The calculated mass median aerodynamic diameter (MMAD) for the particles in the control chamber was 3.9 μm.
Table 1 Characteristics of manganese aerosols generated for whole-body exposures in this study (means ± SD)
Nominal MnSO4 exposure concentration (mg/m3)
0.18 0.92 4.62a 4.62b
Actual exposure concentration (mg MnSO4/m3)c 0.19 ± 0.01 0.97 ± 0.06 4.55 ± 0.33 4.45 ± 0.35
Geometric mean diameter (μm)d 1.04 1.07 1.12 1.04
Geometric standard deviation (σg)d 1.51 1.54 1.58 1.50
Mass median aerodynamic diameter (μm)e 1.73 1.89 2.12 1.72
a Cohort 1
b Cohort 2
c Based on continuous chamber monitoring with a calibrated optical particle sensor (Real-Time Aerosol Sensors, Model RAM-S, MIE, Inc., Billerica, MA).
d Based on biweekly aerodynamic particle size spectrometry (Aerodynamic Particle Sizer, Model 3320, TSI, Inc., St. Paul, MN) measurements.
e Calculated value [56]
Lung and olfactory epithelium manganese concentrations following MnSO4 exposure
Lung and olfactory epithelium manganese concentrations are presented in Table 2. Animals exposed to MnSO4 at ≥0.3 mg Mn/m3 for 65 exposure days developed increased lung manganese concentrations. Animals exposed to MnSO4 at ≥0.06 mg Mn/m3 for 65 exposure days developed increased olfactory epithelium manganese concentrations. Increased lung and olfactory epithelium manganese concentrations developed within three weeks of exposure to MnSO4 at 1.5 mg Mn/m3. Within 45 days after completion of the 65-day inhalation exposure to MnSO4 at 1.5 mg Mn/m3 regimen, lung and olfactory epithelium manganese concentrations were not different from those seen in air-exposed controls.
Table 2 Olfactory epithelial and lung manganese concentrations in young monkeys following exposure to air or MnSO4. Manganese concentrations were determined by graphite furnace atomic absorption spectrometry and are expressed as mean ± SEM μg Mn/g tissue wet weight. Young male rhesus monkeys (n = 4 except where noted) were exposed to either air or MnSO4 6 hours/day, 5 days/week.
Tissue MnSO4 exposure concentration (mg Mn/m3) 15 33 Exposure Day 65 65 [+45]a 65 [+90]a
Olfactory epithelium 0 0.42 ± 0.01b
0.06 1.22 ± 0.15*b
0.3 2.96 ± 0.46*
1.5 6.10 ± 0.39* 7.34 ± 0.70* 7.10 ± 2.01* 0.65 ± 0.04 0.69 ± 0.11
Lung 0 0.15 ± 0.03b
0.06 0.18 ± 0.01b
0.3 0.25 ± 0.02*
1.5 0.39 ± 0.06* 0.35 ± 0.02* 0.33 ± 0.04* 0.09 ± 0.01 0.06 ± 0.01
a Number in brackets indicates number of days post exposure assessment.
b n = 6.
* p < 0.05
Respiratory tract pathology following MnSO4 exposure
Manganese exposure did not affect absolute or relative lung weights and did not result in coughing, dyspnea, or other respiratory signs (data not shown). High-dose exposure to MnSO4 was associated with an increased incidence of minimal to mild subacute bronchiolitis (Table 3). These lesions consisted of infiltrates of lymphocytes, along with neutrophils and occasional eosinophils, primarily surrounding the terminal and respiratory bronchioles and/or alveolar ducts, but sometimes extending into the lamina propria (Figure 2). Macrophages with moderate amounts of pale-staining cytoplasm were occasionally observed in the adjacent airway lumen. Although the overlying epithelium generally appeared intact, because it is normally very thin at the level of the distal airways, it is difficult to confirm that epithelial integrity was unaffected. These changes appeared to be reversible upon cessation of MnSO4 exposure. An increased incidence of enhanced proliferation of BALT in association with smaller (≤500 μm) airways also occurred in monkeys exposed to the highest MnSO4 concentration (1.5 mg Mn/m3) (Figure 3). Some BALT foci included germinal center formation. The incidence of increased BALT was highest in monkeys exposed to 1.5 mg Mn/m3for 33 exposure days (Table 3) suggesting that BALT proliferation may subside even in the face of ongoing manganese exposure. Proliferation of BALT occurred in one animal exposed to MnSO4 at 0.3 mg Mn/m3; however, this increase was only minimal and was not statistically significant.
Table 3 Incidence of MnSO4-induced microscopic lesions observed in young male rhesus monkeys exposed to MnSO4. Incidence is expressed as number affected/number examined.
Lesion MnSO4 exposure concentration (mg Mn/m3) 15 33 Exposure Day 65 65 [+45]a 65 [+90]a
Subacute bronchiolitis/alveolar duct inflammation 0 0/6
0.06 0/6
0.3 1/4b
1.5 3/4* 4/4* 3/4* 0/4 1/4
Increased bronchus associated lymphoid tissue 0 0/6
0.06 0/6
0.3 1/4
1.5 2/4† 3/4* 1/4 2/4† 1/4
a Number in brackets indicates number of days post exposure assessment
b Subacute bronchiolitis was observed in one monkey; however, in this animal only, this lesion was observed in conjunction with aspirated food particles.
* p < 0.05 (Pearson's chi-square test)
† p = 0.053 (Pearson's chi-square test)
Figure 2 Bronchiolitis. Moderate subacute bronchiolitis in a monkey exposed to the highest (1.5 mg Mn/m3) MnSO4 exposure concentration for 15 exposure days (right). Normal appearing bronchioles present in an air-exposed control monkey (left). (10×)
Figure 3 BALT proliferation. Peribronchial BALT proliferation in a monkey exposed to the highest (1.5 mg Mn/m3) MnSO4 exposure concentration for 65 exposure days (right). Normal appearing BALT present in an air-exposed control monkey (left). (4×)
Several of the MnSO4-exposed animals had minimal acute alveolitis. However, this finding was considered unlikely to be treatment-related since no statistically significant dose-response relationship was observed. Minimal changes consistent with chronic bronchiolitis were observed in several control as well as MnSO4-exposed animals. As such, these changes were not considered to be treatment-related. The majority of lung sections examined, including those taken from controls, had scattered deposits of brown-green pigment in the interstitial tissue. Such chronic bronchiolitic changes characterized by lung pigment deposits are commonly seen in monkeys with lung mites [28,29]. The most common lung mite is Pneumonyssus simicola and this agent occurs with nearly 100% incidence in rhesus monkeys [28,29]. Animals used on this study were treated with ivermectin prior to use; thus it is unlikely that superimposed active mite infection occurred. This conclusion is further supported by the absence of microscopic evidence of mites.
One animal from each of the 0.06 and 1.5 mg Mn/m3 exposure groups had minimal or mild, basophilic foci in the nerves of the nose. The spherical basophilic foci were approximately 40 μm in diameter and consisted of concentric lamellae (Figure 4). The foci were Periodic acid-Schiff stain positive but failed to stain with alizarin red, von Kossa's, or Perl's stains (for calcium or iron) and were not birefringent in polarized light (data not shown). Similar foci were present in the nasal epithelium of three other animals, including one control animal. Thus, it is unlikely that these foci were induced by MnSO4 exposure. X-ray microanalysis of foci taken from the decalcified nasal epithelium of a single monkey from the 1.5 mg Mn/m3 exposure group showed the presence of sulfur but no other ions (data not shown). Thus, the basophilic foci noted in the nasal nerves were most likely glycoproteinaceous inclusion bodies (i.e., corpora amylacea) or psammoma bodies. Similar deposits have been observed in the olfactory nerves of untreated rhesus monkeys and rats (Howroyd, unpublished observations), in the olfactory tracts of humans [30], and in the brains of mice [31] and monkeys [32,33].
Figure 4 Basophilic foci in nerves. Basophilic foci (arrow) in nerves of olfactory mucosa from a monkey exposed to MnSO4 at 0.06 mg Mn/m3, for 65 days (40×).
Discussion
Pulmonary inflammation is a common response to the inhalation of various types of particles including manganese [34]. In the present study, monkeys exposed to the highest MnSO4 concentration (1.5 mg Mn/m3) developed subacute bronchiolitis and proliferation of BALT. Bronchiolitis included an infiltrate of acute inflammatory cells (neutrophils) in the peribronchiolar connective tissues in the centriacinar region. These lesions developed relatively rapidly as they were observed in monkeys exposed to MnSO4 for only 15 exposure days. However, with ongoing MnSO4 exposure, the bronchiolitic lesion progressed, affecting a larger percentage of respiratory bronchioles examined. Bronchiolitis resolved rapidly after cessation of MnSO4 exposure and was absent 45 days after the end of the 13 week exposure to MnSO4 at 1.5 mg Mn/m3.
Monkeys and human beings share similar lung anatomy [35], and their respiratory bronchioles and other small airways have similar sensitivity to inhaled toxicants such as cigarette smoke, coal dust, and ozone [36-38].
Following bronchiolitis, people are often noted to have increased wheeze, cough, and asthma, increased airway responsiveness, and reduced lung function due primarily to airflow obstruction [39]. In the present study, signs referable to the respiratory system were not recognized in the monkeys developing manganese-induced bronchiolitis. Although pulmonary function was not assessed in our study, reduced forced expiratory volume in one second (FEV1) and forced vital capacity (FVC) have been reported in workers that have been chronically exposed to high levels of manganese dust [10,40].
Particles deposited in the lung can be retained in the lung interstitium [41] or cleared via the mucociliary apparatus or through the lymphatic system [42]. Alternatively, particles may be transported from the lung via alveolar macrophages or neutrophils with subsequent accumulation in the BALT and tracheobronchial lymph node [42,43]. Thus, as a lymphoepithelial organ, the BALT is critical to the immune defense of the lung and to alveolar clearance of particles and lung pathogens. Proliferation of BALT has been observed in rodents following inhalation exposure to carbon black and silica [44,45]. There is evidence that BALT proliferation occurs in humans with panbronchiolitis, chronic hypersensitivity pneumonitis, and other chronic inflammatory airway disease [46,47]. Proliferative BALT lesions observed in these disease conditions are comparable to those observed in the MnSO4-exposed monkeys, suggesting that BALT proliferation in the manganese-exposed monkeys may be related to the local airway immune response secondary to inflammation induced by MnSO4.
Some agents that target the lung may also affect the upper airways as well. Our laboratory has recently reported that rats exposed subchronically to MnSO4 (at 0.5 mg Mn/m3) develop a mild reversible inflammatory change consisting of pleocellular inflammatory infiltrates and fibrinonecrotic debris within the nasal respiratory epithelium [20]. These lesions occurred primarily in high airflow regions and were consistent with mild irritation. In the present study, however, despite a 15-fold increase in nasal epithelial manganese concentrations, we did not observe any chemical-related nasal pathology in monkeys exposed to MnSO4. Although nasal pathology did not occur, the fate of the manganese that is initially deposited in the nose and subsequently absorbed by the olfactory epithelium remains toxicologically important. Experiments from our laboratory have shown that manganese deposited on the olfactory epithelium can undergo transport along the olfactory nerve with subsequent delivery to the olfactory bulb [48].
Our interest in MnSO4 stems from the use of manganese in the gasoline fuel additive methylcyclopentadienyl manganese tricarbonyl (MMT). Automobiles that use MMT in the fuel and are equipped with catalytic converters emit manganese primarily in the phosphate and sulfate forms with smaller amounts of manganese oxides also being discharged [49]. The manganese exposure concentrations used in this study bracket several human exposure scenarios. Prolonged human exposure to the highest MnSO4 concentration used in this study (1.5 mg Mn/m3) can occur among manganese miners and prolonged exposure is associated with frank neurotoxicity and respiratory disease [9]. Our mid-dose exposure concentration is analogous to the current 8-hr Threshold Limit Value (TLV) for inhaled manganese of 0.2 mg Mn/m3 that has been established by the American Conference of Governmental Industrial Hygienists (ACGIH). Our lowest exposure concentration (0.06 mg Mn/m3) is > 2,000-fold higher than typical air manganese concentrations observed in the ambient air including Canadian cities where MMT is extensively used in gasoline [9,50]. In the present study, exposure concentrations associated with increased lung manganese concentrations and lung pathology were respectively 1.5-and 7.5-fold higher than the current TLV.
The MnSO4 particle size used in this study had an MMAD of 1.72 to 2.12 μm. Aerodynamic size is an important factor that influences particle deposition [51]. Several models based on airway geometry have been developed to describe particle deposition patterns in humans, monkeys, and rats [52,53]. The model of Asgharian and coworkers (1995) predicts a pulmonary deposition efficiency of an aerosol with a particle size of 1.5 μm of approximately 35% for humans and rhesus monkeys while the rat had much lower deposition efficiency (6%) due to higher nasal uptake [52]. The rate of particle clearance from the alveolar region also differs among species. Rodents clear particles from the lung more quickly than either monkeys or humans [54]. Anatomical differences between rodents and primates also affect particle deposition, retention, and clearance. Rodents lack respiratory bronchioles and have simple acini. Macaque monkeys and humans have larger alveoli and alveolar ducts than rats [55] and have similar numbers of respiratory bronchiole generations between the terminal bronchiole and the alveolar duct [35,56]. Our results may be especially important for human risk assessment owing to the fact that monkeys and human beings share similar lung anatomy (i.e., both species have extensive respiratory bronchioles comprised of bronchiolar epithelium and gas exchange epithelium) [35].
Conclusion
High-dose subchronic manganese sulfate inhalation is associated with increased lung manganese concentrations, mild subacute bronchiolitis, alveolar duct inflammation, and proliferation of bronchus-associated lymphoid tissue. Bronchiolitis and alveolar duct inflammatory changes were absent 45 days post-exposure, suggesting that these lesions are reversible upon cessation of subchronic high-dose manganese exposure. These small airway changes occurred in the absence of observable clinical signs. Subchronic exposure to manganese sulfate at exposure concentrations (≤0.3 mg Mn/m3) similar to the current 8-hr occupational threshold limit value established for inhaled manganese was not associated with pulmonary or nasal pathology.
Competing interests
This publication is based on a study sponsored and funded by Afton Chemical Corporation in satisfaction of registration requirements arising under Section 211(a) and (b) of the Clean Air Act and corresponding regulations at 40 C. F. R. Subsections 79.50 et seq.
Authors' contributions
DCD conceived of the study, was the principal investigator and participated in all phases of the study, and drafted the manuscript. MFS participated in the design, coordination, and conduct of the study. EAG participated in and supervised the necropsy and preparation of histology specimens. BAW designed the exposure generation system and characterization of the aerosol. PCH was the study pathologist and conducted the histopathological evaluation of tissues. All authors contributed to, read, and approved the final manuscript.
Acknowledgements
The authors would like to thank Marianne Marshall, Carl Parkinson, Paul Ross and the staff of the CIIT animal care facility for their contributions. We also thank Drs. Jamie Bonner, Jan Dye, Jeff Everitt, Owen Moss, and Elizabeth Roberts for their critical review of this manuscript.
==== Refs
De Vuyst P Camus P The past and present of pneumoconioses Curr Opin Pulm Med 2000 6 151 156 10741776 10.1097/00063198-200003000-00012
Kelleher P Pacheco K Newman LS Inorganic dust pneumonias: the metal-related parenchymal disorders Environ Health Perspect 2000 108 Suppl 4 685 696 10931787
Pope CA 3rd What do epidemiologic findings tell us about health effects of environmental aerosols? J Aerosol Med 2000 13 335 354 11262440
Schwartz J Air pollution and daily mortality: a review and meta analysis Environ Res 1994 64 36 52 8287841 10.1006/enrs.1994.1005
United States Environmental Protection Agency(U.S. EPA) Air Quality Criteria for Particulate Matter EPA/600/P-95/001 1996 2 Washington DC: Office of Research and Development
Chapman RS Watkinson WP Dreher KL Costa DL Ambient particulate matter and respiratory and cardiovascular illness in adults: particle-borne transition metals and the heart-lung axis Environ Toxicol Pharmacol 1997 4 331 338 10.1016/S1382-6689(97)10031-X
Costa DL Dreher KL Bioavailable transition metals in particulate matter mediate cardiopulmonary injury in healthy and compromised animal models Environ Health Perspect 1997 105 Suppl 5 1053 1060 9400700
Gavett SH Madison SL Dreher KL Winsett DW McGee JK Costa DL Metal and sulfate composition of residual oil fly ash determines airway hyperreactivity and lung injury in rats Environ Res 1997 72 162 172 9177658 10.1006/enrs.1997.3732
Agency for Toxic Substances and Disease Registry (ATSDR) Toxicological Profile for Manganese 2000 Atlanta, GA: US Department of Health and Human Services
Roels H Lauwerys R Buchet JP Genet P Sarhan MJ Hanotiau I deFays M Bernard A Stanescu D Epidemiological survey among workers exposed to manganese: effects on lung, central nervous system, and some biological indices Am J Ind Med 1987 11 307 327 3578289
Saric M Occupational and environmental exposures and nonspecific lung disease – a review of selected studies Isr J Med Sci 1992 28 509 512 1428803
Lloyd Davies TA Harding HE Manganese pneumonitis. Further clinical and experimental observations Br J Ind Med 1949 6 82 90
Adkins B JrLuginbuhl GH Gardner DE Acute exposure of laboratory mice to manganese oxide Am Ind Hyg Assoc J 1980 41 494 500 7415970
Bergstrom R Acute pulmonary toxicity of manganese dioxide Scand J Work Environ Health 1977 3 Suppl 1 1 41 905805
Shanker R Dogra RK Sahu AP Zaidi SH Experimental manganese lymphadenopathy in guinea pigs Arch Toxicol 1976 36 151 157 1036890
Singh J Kaw JL Zaidi SH Early biochemical response of pulmonary tissue to manganese dioxide Toxicology 1977 8 177 184 929625 10.1016/0300-483X(77)90006-3
Ulrich CE Rinehart W Brandt M Evaluation of the chronic inhalation toxicity of a manganese oxide aerosol. III. Pulmonary function, electromyograms, limb tremor, and tissue manganese data Am Ind Hyg Assoc J 1979 40 349 353 111536
Zaidi SH Dogra RK Shanker R Chandra SV Experimental infective manganese pneumoconiosis in guinea pigs Environ Res 1973 6 287 297 4583443 10.1016/0013-9351(73)90041-8
Camner P Curstedt T Jarstrand C Johannsson A Robertson B Wiernik A Rabbit lung after inhalation of manganese chloride: a comparison with the effects of chlorides of nickel, cadmium, cobalt, and copper Environ Res 1985 38 301 309 4065078 10.1016/0013-9351(85)90094-5
Dorman DC McManus BE Parkinson CU Manuel CA McElveen AM Everitt JI Nasal toxicity of manganese sulfate and manganese phosphate in young male rats following subchronic (13-week) inhalation exposure Inhal Toxicol 2004 16 481 488 15204764 10.1080/08958370490439687
National Research Council Guide for the Care and Use of Laboratory Animals 1996 Washington DC: National Academy Press
Newland MC Animal models of manganese's neurotoxicity Neurotoxicology 1999 20 415 432 10385901
Kepler GM Richardson RB Morgan KT Kimbell JS Computer simulation of inspiratory nasal airflow and inhaled gas uptake in a rhesus monkey Toxicol Appl Pharmacol 1998 150 1 11 9630447 10.1006/taap.1997.8350
Martonen TB Katz IM Musante CJ A nonhuman primate aerosol deposition model for toxicological and pharmaceutical studies Inhal Toxicol 2001 13 307 324 11295864 10.1080/089583701750127412
Dorman DC Struve MF James RA Marshall MW Parkinson CU Wong BA Influence of particle solubility on the delivery of inhaled manganese to the rat brain: manganese sulfate and manganese tetroxide pharmacokinetics following repeated (14-day) exposure Toxicol Appl Pharmacol 2001 170 79 87 11162771 10.1006/taap.2000.9088
Mellick PW Dungworth DL Schwartz LW Tyler WS Short term morphologic effects of high ambient levels of ozone on lungs of rhesus monkeys Lab Invest 1977 36 82 90 401528
Dorman DC McManus BE Marshall MW James RA Struve MF Old age and gender influence the pharmacokinetics of inhaled manganese sulfate and manganese phosphate in rats Toxicol Appl Pharmacol 2004 197 113 124 15163547 10.1016/j.taap.2004.02.010
Brack M Histochemistry of the lung mite pigment in infections of Pneumonyssus sp. in nonhuman primates Parasitology 1972 64 47 52 4110820
Hull WB Respiratory mite parasites in nonhuman primates Lab Anim Care 1970 20 402 406 4245586
Fuller GN Burger PC Sternberg SS Central nervous system Histology for Pathologists 1997 2 Philadelphia: Lippincott Williams & Wilkins 243 282
Morgan KT Johnson BP Frith CH Townsend J An ultrastructural study of spontaneous mineralization in the brains of aging mice Acta Neuropathol 1982 58 120 124 7180386 10.1007/BF00691652
Kast A Peil H Weisse I Calcified foci at the junction between adrenal cortex and medulla of rhesus monkeys Lab Anim 1994 28 80 89 8158974
Wadsworth PF Jones HB Cavanagh JB The topography, structure and incidence of mineralized bodies in the basal ganglia of the brain of cynomolgus monkeys (Macaca fascicularis) Lab Anim 1995 29 276 281 7564211
Donaldson K Tran CL Inflammation caused by particles and fibers Inhal Toxicol 2002 14 5 27 12122558 10.1080/089583701753338613
Tyler NK Plopper CG Morphology of the distal conducting airways in rhesus monkey lungs Anat Rec 1985 211 295 303 3993981 10.1002/ar.1092110310
Heppleston AG The pathological anatomy of simple pneumoconiosis in coal workers J Pathol Bact 1953 66 235 246 13109636 10.1002/path.1700660127
Hyde DM Hubbard WC Wong V Wu R Pinkerton K Ploppper CG Ozone-induced acute tracheobronchial epithelial injury: relationship to granulocyte emigration in the lung Am J Respir Cell Mol Biol 1992 6 481 497 1581072
Niewoehner DE Kleinerman J Rice DB Pathologic changes in the peripheral airways of young cigarette smokers N Engl J Med 1974 291 755 758 4414996
Ryu JH Myers JL Swensen SJ Bronchiolar disorders Am J Respir Crit Care Med 2003 168 1277 1292 14644923 10.1164/rccm.200301-053SO
Boojar MM Goodarzi F A longitudinal follow-up of pulmonary function and respiratory symptoms in workers exposed to manganese J Occup Environ Med 2002 44 282 290 11911030
Nikula KJ Vallyathan V Green FH Hahn FF Influence of exposure concentration or dose on the distribution of particulate material in rat and human lungs Environ Health Perspect 2001 109 311 318 11335177
Harmsen AG Mason MJ Muggenburg BA Gillett NA Jarpe MA Bice DE Migration of neutrophils from lung to tracheobronchial lymph node J Leukoc Biol 1987 41 95 103 3468196
Lee KP Kelly DP Translocation of particle-laden alveolar macrophages and intra-alveolar granuloma formation in rats exposed to Ludox colloidal amorphous silica by inhalation Toxicology 1993 77 205 222 8384734 10.1016/0300-483X(93)90161-K
Henderson RF Driscoll KE Harkema JR Lindenschmidt RC Chang IY Maples KR Barr EB A comparison of the inflammatory response of the lung to inhaled versus instilled particles in F344 rats Fundam Appl Toxicol 1995 24 183 197 7737430 10.1006/faat.1995.1022
Hiramatsu K Azuma A Kudoh S Desaki M Takizawa H Sugawara I Inhalation of diesel exhaust for three months affects major cytokine expression and induces bronchus-associated lymphoid tissue formation in murine lungs Exp Lung Res 2003 29 607 622 14594659
Sato A Chida K Iwata M Hayakawa H Study of bronchus-associated lymphoid tissue in patients with diffuse panbronchiolitis Am Rev Respir Dis 1992 146 473 478 1489143
Suda T Chida K Hayakawa H Imokawa S Iwata M Nakamura S Sato A Development of bronchus-associated lymphoid tissue in chronic hypersensitivity pneumonitis Chest 1999 115 357 363 10027432 10.1378/chest.115.2.357
Brenneman KA Wong BA Bucellato MA Costa ER Gross EA Dorman DC Direct olfactory transport of inhaled manganese (54MnCl2) to the rat brain: Toxicokinetic investigations in a unilateral nasal occlusion model Toxicol Appl Pharmacol 2000 169 238 248 11133346 10.1006/taap.2000.9073
Molders N Schilling PJ Wong J Roos JW Smith IL X-ray fluorescence mapping and micro-XANES spectroscopic characterization of exhaust particulates emitted from auto engines burning MMT-added gasoline Environ Sci Technol 2001 35 3122 3129 11505987 10.1021/es0010958
Pellizzari ED Clayton CA Rodes C Mason RE Piper LL Fort B Pfeifer G Lynam D Particulate matter and manganese exposures in Toronto, Canada Atmos Environ 1999 33 721 734 10.1016/S1352-2310(98)00229-5
Schlesinger RB McClellan RO, Henderson RF Deposition and clearance of inhaled particles Concepts in Inhalation Toxicology 1995 Taylor and Francis, Washington, DC 191 224
Asgharian B Wood R Schlesinger RB Empirical modeling of particle deposition in the alveolar region in the lungs: A basis for interspecies extrapolation Fundam Appl Toxicol 1995 27 232 238 8529818 10.1006/faat.1995.1128
Snipes MB Long-term retention and clearance of particles inhaled by mammalian species Crit Rev Toxicol 1989 20 175 211 2692607
Mercer RR Crapo JD Structure of the gas exchange region of the lungs determined by three dimensional reconstruction Toxicology of the Lung 1988 New York:Raven Press 43 70
Tyler WS Small airways and terminal units: comparative subgross anatomy of lungs Am Rev Respir Dis 1983 128 S32 S36 6881704
Hinds WC Aerosol Technology 1999 2 New York: John Wiley and Sons
|
16242036
|
PMC1283983
|
CC BY
|
2021-01-04 16:36:27
|
no
|
Respir Res. 2005 Oct 21; 6(1):121
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-121
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1221624203810.1186/1465-9921-6-122ResearchA role for airway remodeling during respiratory syncytial virus infection Becnel David [email protected] Dahui [email protected] Joshua [email protected] Dawn M [email protected] Stephania A [email protected] Department of Biological Sciences, 202 Life Sciences Bldg., Baton Rouge, LA 70803, USA2 Deparment of Biochemistry & Molecular Biology, 13400 East Shea Boulevard, Scottsdale, AZ 85259, USA2005 21 10 2005 6 1 122 122 9 3 2005 21 10 2005 Copyright © 2005 Becnel et al; licensee BioMed Central Ltd.2005Becnel 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
Severe respiratory syncytial virus infection (RSV) during infancy has been shown to be a major risk factor for the development of subsequent wheeze. However, the reasons for this link remain unclear. The objective of this research was to determine the consequences of early exposure to RSV and allergen in the development of subsequent airway hyperreactivity (AHR) using a developmental time point in the mouse that parallels that of the human neonate.
Methods
Weanling mice were sensitized and challenged with ovalbumin (Ova) and/or infected with RSV. Eight days after the last allergen challenge, various pathophysiological endpoints were examined.
Results
AHR in response to methacholine was enhanced only in weanling mice exposed to Ova and subsequently infected with RSV. The increase in AHR appeared to be unrelated to pulmonary RSV titer. Total bronchoalveolar lavage cellularity in these mice increased approximately two-fold relative to Ova alone and was attributable to increases in eosinophil and lymphocyte numbers. Enhanced pulmonary pathologies including persistent mucus production and subepithelial fibrosis were observed. Interestingly, these data correlated with transient increases in TNF-α, IFN-γ, IL-5, and IL-2.
Conclusion
The observed changes in pulmonary structure may provide an explanation for epidemiological data suggesting that early exposure to allergens and RSV have long-term physiological consequences. Furthermore, the data presented here highlight the importance of preventative strategies against RSV infection of atopic individuals during neonatal development.
respiratory syncytial viruspulmonaryinflammationage factorsasthmamice
==== Body
Introduction
Several experimental studies have shown a synergistic interaction between respiratory viral infections and allergic inflammation that exacerbate asthma [1-3]. RSV is the most common respiratory pathogen during infancy, and the majority of children worldwide have been infected with it by 2 years of age[4]. Several retrospective studies have suggested a link between RSV lower respiratory tract infections in infancy and the later development of asthma [5-10]. In a more recent study conducted by Sigurs and colleagues using a cohort of 140 children (47 of whom were hospitalized for RSV bronchiolitis during the first year of life and a control population of 93 infants with no history of RSV infection), it was found that by 7 years of age 30% of the children in the RSV group had experienced recurrent physician-diagnosed "wheeze" (ie, asthma) as compared to 3% of the control group[11]. Multivariant analyses further demonstrated that the greatest risk factor for asthma was RSV bronchiolitis independent of a family history of atopy. Moreover, the Tucson Children's Respiratory Study demonstrated that even children with mild RSV infections were 4 times more likely to have recurrent wheeze by 6 years of age[12]. Cumulatively, the data suggest that RSV bronchiolitis in infancy is associated with an increased risk of wheeze, which may persist for several years and is not adequately explained by a family history of atopy. Whether RSV directly contributes to the development of asthma remains unclear.
A recent study using an animal model of RSV infection demonstrated that "infections in early life play an important role" in shaping the secondary immune response to antigen leading to long-term consequences for the host[13]. In this study, Culley and colleagues showed that the age of initial infection played a significant role in the secondary response of these same animals to rechallenge with RSV at 12 weeks of age. Interestingly, if the mice were initially infected at 1 – 7 days of age, then their immune response to rechallenge at 12 weeks of age was characterized by enhanced bronchoalveolar lavage (BAL) cellularity including increased eosinophil and neutrophil numbers, increased CD8+ T cell numbers, and increased CD4+ T cell production of intracellular IL-4. The data seemed to indicate that a CD4+ response was important in creating the immune memory response of the neonatal mice to RSV, while a CD8+ response seemed important in the older animals (age 4 – 8 wks). Furthermore, these data suggested that the Th2 skewing of the immune response was likely to be important in eliciting disease pathogenesis. Although this work was seminal in demonstrating the importance of timing (7 d vs. 4 wks) of the initial infection on the subsequent T cell responses, it did not link these events with enhanced pulmonary dysfunction or demonstrate the prolonged contribution of these T cell responses to pathophysiological events involved in airway remodeling associated with asthma. Recently, Dakhama and colleagues[14] demonstrated that indeed timing of initial infection was a critical factor in determining the airway response to subsequent RSV infection. However, other risk factors for the development of pulmonary inflammation and wheeze (i.e., asthma) in humans exist and the enhancement of pulmonary inflammation by RSV may be dependent on the individual's atopic background and current exposure to allergens or other environmental factors.
Our hypothesis was that early exposure to RSV and allergen act synergistically to illicit inflammatory responses and long-term functional changes in the developing lung. We further hypothesized that these changes were the result of changes in airway structure (i.e., remodeling) and therefore, would compromise adult lung function. In the present study, weanling mice were exposed to RSV and/or ovalbumin (Ova) to examine the effect of early exposures on pulmonary pathophysiology. We report that weanling mice infected with RSV then exposed, Ova fail to develop airway hyperreactivity (AHR) or long-term pathophysiologic changes, while weanling mice exposed first to Ova then infected with RSV developed increased AHR and long-term pulmonary pathologies. This increase in AHR was accompanied by pulmonary inflammation due to increased eosinophil and lymphocyte cell numbers, mucus cell hypertrophy, and enhanced mucus production. Intriguingly, the mice also exhibited signs of airway remodeling including subepithelial fibrosis. The observed remodeling events were correlated with increased levels of various cytokines including TNF-α, IFN-γ, IL-5, and IL-2. Collectively, these data demonstrate that RSV infections, when combined with an allergic predisposition, can have long-term consequences for the lung and may contribute to the development of inflammatory disease states, such as asthma.
Methods
Mice
BALB/cJ mice, 6 – 10 weeks of age, were purchased from Jackson Labs and were maintained in ventilated micro-isolator cages housed in a specific pathogen-free animal facility. Sentinel mice within this animal colony were negative for antibodies to viral and other known mouse pathogens. All animal protocols were prepared in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, 1996) and approved by the Institutional Animal Care and Use Committees at Mayo Clinic Arizona and Louisiana State University.
Viral Preparation and Infection of Mice with RSV
RSV strain A2 (a kind gift of Dr. Barney Graham; NIH) was originally provided by Dr. R Chanock (NIH) and has since been maintained in culture by passage in HEp-2 cells. Master stocks and working stocks of virus were prepared as described elsewhere[15]. Prior to infection all mice were anesthetized with 3% isoflurane. Mice were subsequently infected intratracheally (i.t.) with RSV (104 TCID50/g body weight) or mock infected (i.e., culture media alone) at 21 or 47 days of age (Figure 1). Four days post-infection viral titer was determined using the TCID50 method of Spearman-Kärber[16,17].
Figure 1 Schematic of study protocol and exposure groups. Weanling mice (21 days of age) were injected i.p. with ovalbumin complexed to Imject Alum (Ova, ROO, and OOR groups) or with isotonic saline (Sal, RSS, and SSR groups) on protocol days 0 and 14. After 1 h, the ROO and RSS groups were infected with RSV (104 TCID50/g body weight). These mice were then exposed to aerosolized ovalbumin or saline for 20 minutes on protocol days 24, 25, and 26. On protocol day 26, a subset of Sal and Ova mice were infected with RSV at 104 TCID50/g body weight (SSR and OOR groups, respectively).
Ovalbumin Sensitization and Challenge Protocol
Mice were sensitized and challenged with chicken ovalbumin (Ova; crude grade IV; Sigma, St. Louis, MO) as previously described [18]. Mice were sensitized with an intraperitoneal (i.p.) injection of 0.1 ml (20 μg) Ova complexed with 2 mg Imject Alum (Al [OH]3/Mg [OH]2; Pierce, Rockford, IL) on protocol days 0 and 14 (Figure 1). Mice were subsequently challenged with an aerosol generated from an Ova solution (1% Ova w/v in saline) for 20 minutes on protocol days 24, 25, and 26 using an ultrasonic nebulizer (DeVilbiss, Somerset, PA). Control animals were injected i.p. with saline on days 0 and 14 and challenged with aerosolized saline on protocol days 24, 25, and 26 as described above. The mice were assessed for pulmonary cellular infiltrates, histopathologies, and lung function on protocol day 34.
Study Protocol
The study protocol is outlined in Figure 1. The mice were divided into six groups. The SAL group was mock-allergen sensitized and challenged. The OVA group was Ova sensitized and challenged. The RSS group was mock-allergen sensitized and challenged and RSV-infected at 21 d of age. The ROO group was Ova sensitized and challenged and subsequently RSV-infected at 21 d of age. The SSR group was mock-allergen sensitized and challenged and subsequently RSV-infected at 47 d of age. The OOR group was Ova sensitized and challenged and subsequently RSV-infected at 47 d of age. Three mice in each group were sacrificed four days post-infection to assess pulmonary viral titers as described above. Assessment of airway reactivity and collection of tissue and BAL samples, as detailed below, were performed on protocol day 34 (55 days of age and 8 days following RSV infection).
Assessment of Airway Reactivity in Response to Methacholine
Whole body plethysmography
Airway responsiveness to methacholine (MeCh), a muscarinic agonist, was assessed by whole body plethysmography (Buxco Electronics, Troy, NY and EMKA Technologies, Falls Church, VA) as described previously[19]. Mice were exposed for 3 minutes to aerosolized saline and subsequently to increasing concentrations of aerosolized MeCh (0, 3.125, 6.25, 12.5, 25, and 50 mg/ml in isotonic saline; Sigma). Following each nebulization, data including minute volume, tidal volume, breathing frequency, and enhanced paused (Penh) were recorded for 3 minutes. The Penh values measured during each 3-minute sequence were averaged and expressed for each dose of MeCh. Baseline Penh values did not differ significantly between any of the groups.
Invasive measurements of respiratory mechanics
Pulmonary resistance was measured using the forced oscillation technique as previously described[20]. Anesthetized animals were mechanically ventilated with a tidal volume of 10 ml/kg and a frequency of 2.5 Hz using a computer-controlled piston ventilator (Flexivent, SCIREQ; Montreal, Canada). Responses were determined in response to increasing concentrations of aerosolized MeCh (0, 1.875, 3.75, 7.5, 15, and 30 mg/ml in isotonic saline). All data that did not result in a coefficient of determination that was greater than 0.9 were excluded. The average value for each dose was calculated; and the percent difference from baseline per dose was then plotted.
Sample Collection
On protocol day 34, all mice were euthanized and the following samples were obtained:
Lavage fluid
Bronchoalveolar lavage (BAL) of the lungs was performed using 1 ml of PBS containing 2% FBS. Total BAL cellularity was determined with the use of a hemocytometer. Cytospin slides were fixed and stained using the Diff-Quick kit (IMEB, Chicago, IL) and differential cell counts by unbiased observers were based on counts of 200–300 cells using standard morphological criteria to classify individual leukocyte populations. Four mice from each group were used for these analyses.
Pulmonary histology
Lungs were inflated with 1 ml of 10% neutral-buffered formalin via a tracheostomy tube. After instillation of fixative, the trachea was ligated, and the lung was excised and fixed in formalin for 24 hours at 4°C. These tissues were then embedded in paraffin, cut in 4 μm frontal sections, mounted onto slides, and stained with either hematoxylin and eosin (H&E), periodic acid-Schiff (PAS) to quantitate mucus, Masson's trichrome (MT) to quantitate airway collagen deposition, toludine blue to quantitate mast cells, or anti-MBP (major basic protein) antibodies to specifically evaluate tissue-infiltrating eosinophils as previously described[21]
Morphometric analyses
The MT data were analyzed morphometrically by digital image analysis using Image-Pro Plus software (version 5.0.1, Media Cybermetics, Inc., Silver Spring, MD). The following calculation was used to determine airway mucus: % airway mucus = (the area of airway epithelium staining positive for mucus/the total area of airway epithelium) × 100. To calculate the thickness of collagen deposition within the basement membranes, a random starting point was chosen and a single measurement was made between two points on either side of the collagen deposition at right angles to a tangent marking the perimeter of the basement membrane. For each airway, measurements at approximately 50 μm intervals from a randomly chosen starting point were made around the entire airway. The measured values were averaged for the airways of each animal and the mean values for each group were determined.
Cytokine Assays
Cytokine levels in whole lung homogenates were determined using the Mouse Th1/Th2 Cytokine Cytometric Bead Array Kit (BD Biosciences) as per the manufacturer's instructions. The sensitivity of the assay was as follows: TNF-α – 6.3 pg/ml, INF-γ – 2.5 pg/ml, and IL-5, IL-4 and IL-2 – 5 pg/ml. The data was resolved in the FL3 channel and acquired with a BD FACScan™ flow cytometer. Data analyses were performed using the BD Cytometric Bead Array Software to generate standard curves for each cytokine and to determine sample cytokine levels.
Statistical Analysis
Data are presented as mean ± SEM obtained from experiments with n = 8 for whole body plethysmography analysis of AHR, n = 4 for invasive measurements of pulmonary mechanics, pulmonary viral titers, and histology, n = 3 for cytokine assays. For AHR and BAL cellularity, differences between groups were evaluated by means of two-way ANOVA. Bonferroni post-tests were performed to compare between pairs of groups. To determine statistical significance of the morphometric data, we employed a Kruskal-Wallis test with a Dunn's post-test. A one-way ANOVA was used to compare the mean cytokine levels among the various groups followed by the Tukey-Kramer multiple comparisons tests for significance between the groups. This was repeated for each individual cytokine. Differences between means were considered significant when p < 0.05.
Results
AHR is Enhanced in Weanling Mice Exposed First to Ova then Infected with RSV
In order to investigate the role of RSV infection in relation to other environmental factors such as allergen, we sensitized and challenged weanling mice with Ova as shown in Figure 1. One hour post-sensitization, a subset of mice was infected with RSV (104 TCID50/g body weight). On the last day of allergen challenge, another subset of mice was infected with RSV. Eight days following the last Ova challenge, mice were exposed to increasing doses of MeCh and AHR was assayed. As shown in Figure 2A, RSV infection before (RSS) or after (SSR) saline administration failed to alter AHR. However, weanling mice exposed to Ova and then RSV (OOR) had significantly greater airway hyperresponsiveness to MeCh compared to all other groups (p < 0.001). AHR of weanling mice exposed first to RSV and then Ova (ROO) was similar to that of mice exposed only to Ova (OVA).
Figure 2 Enhanced airway hyperreactivity and pulmonary resistance in weanling mice exposed to RSV and OVA. A) Airway hyperreactivity (Penh) of each group is plotted as a function of increasing doses of inhaled MeCh. Data points are mean ± SEM from 8–10 mice per group. Groups as outlined in Figure 1. B) Lung resistance values were obtained by a forced oscillation technique and are plotted as a function of increasing doses of inhaled MeCh. Values presented are means ± SEM (n = 4 mice/group). *p < 0.001 for OOR vs. all other groups.
The consequences of altered airway responsiveness were further examined in vivo by invasive measurements of airway resistance. Respiratory system mechanics were assessed using the single-compartment model. The lung mechanics of weanling mice exposed to Ova and RSV (OOR) were significantly elevated as compared to control mice in response to MeCh administration (Figure 2B; p < 0.001).
Allergen Sensitization and Challenge Decreases Pulmonary RSV Titer
Four days post-infection, we assessed pulmonary viral titers for all groups of mice. As expected, mice from the OVA and SAL groups displayed no evidence of viral replication. The mean viral titer in the lungs of weanling mice exposed to RSV alone (RSS) was 3.2 ± 0.08 log10 TCID50/g of lung, while viral titers in adult mice exposed to RSV (SSR) were 5.3 ± 0.1 log10 TCID50/g. Intriguingly, the reverse was true for mice infected with RSV in the presence of allergen sensitization. If mice were infected with RSV prior to Ova sensitization and challenge (ROO), then viral titers were high (5.2 ± 0.08 log10 TCID50/g); however if mice were sensitized and challenged with Ova prior to RSV infection (OOR) then the resulting viral titer (3.2 ± 0.2 log10 TCID50/g) was significantly lower (p < 0.001).
BAL Cellularity Increases in Mice Exposed to Ova and RSV
To evaluate the pulmonary immune response to RSV and Ova exposure, BAL fluid cells were recovered, and the total cellularity and composition of leukocytes among the groups of animals were compared. Total cell counts from the BAL fluid of the SAL, OVA, RSS, and SSR groups were not significantly different as shown in Figure 3. However, total cell counts from the BAL fluid of mice exposed to both RSV and Ova (ROO and OOR) were significantly greater than any other group (7.73 × 105 ± 0.19 and 4.65 × 105 ± 0.25; respectively). A comparison of individual cell numbers per ml of BAL fluid reveals that these increases were mainly due to increases in eosinophil and lymphocyte cell populations. To compare cell proportions among the different exposure groups, BAL cellularities were expressed as the product of the total number of cells recovered and the percentages of each cell type derived from differentials. RSV infection, in the presence of allergen challenge, led to a significant increase in total BAL cellularity (ROO and OOR). Interestingly, this increase was highest in the group of mice that were infected prior to allergen challenge (ROO). A significant reduction in the number of BAL macrophages was observed in the OVA, SSR, and OOR groups. Eosinophilia was observed only in the groups that received Ova (OVA, ROO, and OOR groups; p < 0.001). No significant difference in neutrophil numbers was observed in any of the groups.
Figure 3 BAL fluid cellularity is altered in mice exposed to RSV and/or OVA. Differential cell counts were performed on Diff-quick stained cytospin preparations by two unbiased observers counting > 300 cells per sample. BAL cellularities are expressed as the product of the total number of cells recovered and the percentages of each cell type derived from differentials. Data are expressed as means ± SEM.
Enhanced Pulmonary Pathology is Observed in Weanling Mice Exposed to Ova and Subsequently Infected with RSV
The pulmonary histology induced by allergic sensitization and RSV infection is illustrated in Figure 4. Weanling mice exposed to Ova followed by infection with RSV (OOR) exhibited a significantly greater degree of pulmonary inflammation in both the peribronchial and perivascular regions (Figure 4, panel A). Additional histological analyses of the lungs of OOR mice demonstrated that other changes were present. Lung sections stained with PAS to detect mucus showed a significant increase in mucus production and mucus cell hypertrophy (Figure 4, panel B). Morphometric analyses of the PAS stained lung sections revealed a 2.3 fold increase in the percentage of airway mucus in the OOR group compared to the Ova group (35 ± 0.4% vs. 15 ± 2.2%, respectively; p < 0.05). Immunofluorescent staining using an anti-MBP antibody, which is specific for eosinophils, demonstrated an increase in the number of eosinophils that were associated with areas of expanded bronchus-associated lymphoid tissue (BALT) (Figure 4, panel C). There was no evidence of pulmonary inflammation, eosinophilia, or mucus production in the SAL, RSS, or SSR groups. In addition, no significant recruitment (i.e., average of 2 per entire lung section) of mast cells was observed in any of the exposure groups.
Figure 4 Enhanced pulmonary histopathology is observed in mice exposed to Ova and RSV. Lung sections from formalin-fixed, paraffin-embedded tissue were stained for A) cellularity, B) mucus (purple), and C) eosinophils (green) as described in the materials and methods section. The photographs are representative of the staining that occurs in the bronchioles of SAL, OVA, SSR, and OOR exposed mice. Although not shown, the RSS resembled the SSR group and the ROO group was not significantly different from the OVA mice. Scale bar = 50 μm.
Visual analysis of the OOR lung sections indicated thickening of the subepithelial reticular layer indicative of airway remodeling. To investigate the role of airway remodeling in the enhanced pulmonary pathophysiology observed in these groups lung sections were stained with Masson's trichrome. All groups exposed to Ova (OVA, ROO, and OOR) displayed an increase in subepithelial fibrosis that was accompanied by an increase in airway collagen deposition compared to the SAL control group (Figure 5). In fact, a two-fold increase in the amount of basement membrane associated-collagen was observed in the OOR group relative to Ova alone (8.3 ± 0.9 μm vs. 4.2 ± 0.4 μm; p < 0.01). Neither group exposed to RSV alone (RSS and SSR) developed observable airway pathologies.
Figure 5 Airway remodelling is evident in weanling mice exposed to Ova and RSV. Fibrosis and deposition of collagen was observed in the subepithelial, reticular layer of the airways in the OVA, ROO, and OOR mice. Morphometric analyses using data collected at 50 μm intervals over the entire basement membrane revealed that these differences were significant. (8–10 measurements at ~50 μm intervals were collected for at least 5 airways; n = 4 mice per group). ROO vs. OOR, *p < 0.05 and OVA vs. OOR, *p < 0.01.
Allergen Sensitization of Weanling Mice and Subsequent Infection with RSV Increases the Levels of both Th1 and Th2 Cytokines
To determine how exposure of weanling mice to Ova and RSV enhanced AHR and led to pulmonary fibrosis, the concentration of various cytokines in whole lung homogenates was measured. When lung homogenates were isolated on the final day of the protocol (i.e., day 34), no significant difference in cytokine levels for TNF-α, IFN-γ, IL-5, IL-4, or IL-2 were observed. In standard Ova models demonstrating allergic inflammation and AHR, expression of Th2 cytokines such as IL-4 and IL-13 typically peak within 48 hours of final allergen challenge and in most cases return to baseline levels within 96 hours[20,22-24]. Therefore, these studies were repeated using whole lung homogenates that were isolated 4 days post infection (i.e., protocol day 4 for the ROO group and day 30 for the OOR group). Cytokine levels from the ROO group were not significantly different from the OVA or SAL groups (data not shown). Interestingly, significantly elevated levels of TNF-α, IFN-γ, IL-5, and IL-2 were observed in the OOR group (151 ± 20 pg/ml, 3663 ± 121 pg/ml, 58 ± 7.5 pg/ml, 34 ± 6.2 pg/ml; respectively) as compared to the OVA group (15 ± 3.4 pg/ml, 7.4 ± 0.12 pg/ml, 10 ± 0.97 pg/ml, 13 ± 1.9 pg/ml; respectively) (Figure 6). Levels of IL-4 in the OOR group were similar to that of the OVA group (24 ± 1.9 pg/ml vs. 16 ± 2.7 pg/ml, respectively).
Figure 6 Th1 and Th2 cytokine levels are elevated in the lungs of OOR mice. CBA analysis was used to determine the concentration of TNF-α, IFN-γ, IL-5, IL-4, and IL-2 in whole lung homogenates (n = 3 mice per group; *p < 0.01) isolated on protocol day 30. Data are expressed as means ± SEM.
Discussion
In the present study, we have shown that exposure of weanling mice to RSV followed by Ova does not lead to (AHR) or pulmonary pathologies greater than that induced by Ova sensitization and challenge alone. In contrast, exposure of weanling mice to Ova followed by RSV infection leads to long-term pulmonary consequences at the pathophysiological level. RSV infection of weanling mice (unless accompanied by Ova sensitization and challenge) was unable to induce significant histopathologies or AHR. However, early sensitization and challenge with Ova followed by infection with RSV induced a 2.6 fold increase in AHR over SAL controls and a 2 fold increase over Ova alone (Figure 2A). The enhanced AHR coincided with increased: 1) total BAL cellularity (increases were specifically observed in eosinophil and lymphocyte cell numbers); 2) pulmonary inflammation in both the perivascular and peribronchial regions of the lungs; 3) mucus cell hypertrophy and airway mucus production; 4) elevated levels of TNF-α, IFN-γ, IL-5, and IL-2 and airway collagen deposition; and 5) subepithelial fibrosis. These data suggest that pulmonary remodeling events are occurring in weanling mice exposed to allergen followed by RSV infections and, furthermore, that these exposures synergistically enhance pulmonary pathology and physiology. In fact, preliminary data suggest that AHR is prolonged only in the OOR group (> 4 weeks post infection), whereas AHR in the OVA and ROO groups was no longer significantly different from SAL controls forty-eight hours after the last allergen challenge.
Rarely, in humans or mice, is eosinophilia associated with primary RSV infection. However, RSV infection in combination with Ova sensitization and challenge resulted in significant pulmonary eosinophilia in both the pulmonary tissue and BAL fluid (OVA: 1.3 × 105 ± 0.1; ROO: 2.3 × 105 ± 0.3; and OOR: 1.8 × 105 ± 0.2). These results seem inconsistent with those of Peebles et al. [25], who reported decreases in allergen-induced pulmonary eosinophilia when RSV infection preceded allergen challenge. Although the exact cause for this discrepancy is unknown, there are several methodological differences between our studies, which may be pertinent. Peebles and colleagues: 1) used adult mice in their studies, 2) exposed their mice to daily Ova aerosol challenges for eight days, and 3) infected their mice 14 days post Ova challenge. Whereas we: 1) used weanling mice, 2) exposed mice to aerosolized Ova for 20 min for three consecutive days, and 3) infected our mice prior to Ova sensitization (ROO) or immediately following the last Ova challenge (OOR). Although the enhanced recruitment of any one cell did not correlate with pulmonary pathologies, the ratio of total macrophages to eosinophils in the BAL fluid correlated fairly well with pathophysiology (OVA: 0.3 vs. OOR: 0.4).
Culley and colleagues in their neonatal model of RSV infection [13] demonstrated that RSV infection of young mice (0–14 d of age) produced more severe disease initially and upon subsequent rechallenge than did infection of adult mice. In fact, the younger the mouse upon initial RSV infection, the stronger the Th2 polarized immune response to RSV upon secondary infection. Furthermore, human data suggested a link between RSV lower respiratory tract infections in infancy and the later development of asthma [5-10]. Thus, we were expecting to see pulmonary changes in weanling mice (21 d of age) exposed to RSV alone (RSS). However, we failed to see any pathophysiological response to RSV alone (RSS and SSR) or to weanling exposure to RSV followed by allergen (ROO). Although this data was unexpected, we feel that it further supports Culley's original work[13] and the more recent work of Dakhama and colleagues[14] demonstrating the importance of timing of the initial exposure to RSV. Although not presented here, we have preliminary data suggesting that earlier infection with RSV during neonatal development (7 d of age) is sufficient to induce long-term functional consequences in the mouse even in the absence of allergic inflammation. In summary, the age of primary RSV infection has a crucial role in determining disease outcome and suggests that immunity in weanling mice may have matured beyond the polarized Th2 responses of the neonate.
Histological analyses of weanling mice first exposed to Ova and then infected with RSV (OOR group) revealed changes consistent with airway remodeling including subepithelial fibrosis, collagen deposition, smooth muscle hypertrophy, and mucus cell hyperplasia. Histological changes observed in the ROO group were similar to mice receiving Ova alone (OVA group). The amount of airway remodeling observed was positively correlated with increased airway hyperresponsiveness to aerosolized MeCh and with macrophage/eosinophil ratios in the OVA and OOR groups. Our data (presented here and elsewhere[20]) along with the data of others[26] suggest that eosinophils play a prominent role in orchestrating the local pulmonary immune responses and pathologies associated with allergic inflammation (i.e., asthma). It will be interesting to see if the specific loss of eosinophils leads to improvement of pulmonary function parameters and pathologies associated with neonatal RSV infection.
Analysis of cytokine data demonstrated elevations in both Th2 and Th1 cytokines. Previous studies have clearly demonstrated that TNF-α contributes to early clearance of RSV [27]; however, continued production of TNF-α exacerbates RSV-induced illness [28]. Furthermore, neutralization of TNF-α has been shown to reduce clinical illness without an impairment on viral clearance[29]. Interestingly, the OOR group exhibited greater than 10 times the amount of TNF-α observed in any other group. One might also argue that IFN-γ expression in the OOR group, which is upregulated almost 500 fold over the OVA group, is important in the physiologic and histologic injury observed in the OOR group. However, TNF-α levels were still detectable at 17.2 ± 2.4 pg/ml long after (i.e., 8 days) IFN-γ expression was no longer detectable. Though the mechanisms are unclear, the data presented here suggest that the extended upregulation of TNF-α may be important in the lung injury and remodeling events observed in the OOR group.
Interestingly, pulmonary viral titers appeared to be dependent on age at initial RSV infection and allergic phenotype of the mice. Recall that the RSS and OOR groups had significantly lower viral titers than the ROO and SSR groups. We believe that low RSV titers in the RSS group are the result of the Th2-bias that is known to occur in the early neonatal immune system of both mice and humans [30-32], while the low RSV titers in the OOR group are the result of Ova-induced Th2 immune responses. Furthermore, our previous studies demonstrated that eosinophil associated ribonuclease 11 (Ear11) transcripts increase in response to Th2 inflammatory events, such as Ova sensitization and challenge, and that this increase is paralleled by a concomitant increase in RNase activity in the BAL fluid[18]. Ear11 belongs to a family of eosinophil associated RNases, including eosinophil-derived neurotoxin (EDN) and eosinophil cationic protein (ECP). Both EDN and ECP have been shown to possess antiviral activities against a range of single-stranded RNA viruses including RSV [33-35]. Thus, these data cumulatively suggest that the expression of Ear11 in response to a Th2 environment provides the protective (i.e., antiviral) effects observed in the RSS and OOR groups. Ongoing studies in this laboratory with Ear11 depletion strategies are expected to resolve this question.
Severe RSV infections (i.e., requiring hospitalization) during infancy are associated with the development of subsequent wheeze and pulmonary dysfunction including a diagnosis of asthma in later life [6,11,12,36-41]. One hypothesis is that RSV bronchiolitis is simply more severe in atopic individuals. In fact, our studies demonstrated that interactions between RSV infection and allergen-induced immune responses exacerbate pulmonary inflammation and airway physiology. Furthermore, the synergism between RSV and allergen leads to long-term pulmonary consequences such as airway remodeling and may explain progressive lung disease in humans. An alternative hypothesis is that severe RSV infections during infancy are a predisposing factor for the development of airway inflammatory disease states, such as asthma. In this scenario, RSV infection during infancy, when the immune system is still developing and is in a Th2-skewed state, initiates a Th2 polarized primary immune response and subsequently a Th2 polarized memory response to RSV [41]. Such a response may even be heightened in atopic individuals. Although compelling evidence in support of the later hypothesis comes from recently published data [13], we did not find evidence of this in our studies using weanling mice. It is important to point out that this may be due entirely to age at primary infection, since data from other labs[13,14] and unpublished data from our laboratory demonstrate that infections at earlier time points (i.e., 7 days post-partum) leads to long-term immune and pulmonary consequences for the host.
Respiratory tract viral infections account for approximately 85% of asthma exacerbations in children, and 80% of those children have allergic asthma[11,38,39,42,43]; therefore, it is imperative that we understand the mechanisms by which viral infections lead to asthma exacerbations. Although it remains unclear what is responsible for RSV-enhanced allergic inflammation in the lung, what is obvious is that AHR in RSV and Ova exposed mice does not correlate with RSV titer. In contrast, AHR does appear to correlate with increased pulmonary eosinophilia, lymphocyte cell numbers, a decrease in macrophage numbers, and most importantly pulmonary remodeling. Our data suggest that exposure to allergens and RSV leads to increased structural changes in the conducting airways. And it is these structural changes in the developing lung that may be ultimately responsible for the progressive development of the chronic inflammatory disease state known as adult asthma.
Conclusion
Weanling mice exposed to allergen and RSV exhibit enhanced immune cell responses, which are accompanied by long-term changes in airway structure and function. Although this increased pathologic response was associated with RSV infection, the enhanced pathologies were not dependent upon pulmonary viral titer. Airway remodeling was evident in the lungs of adult mice exposed to RSV and Ova and may provide an explanation for observations suggesting that viral exacerbations of asthma have long-term physiological consequences. Airway remodeling was correlated with elevated levels of TNF-α in the lungs. Extrapolation of these studies to exposures occurring in human neonates highlights the importance of preventative strategies against RSV infection of atopic individuals during neonatal development.
Abbreviations
RSV, respiratory syncytial virus; AHR, airway hyperreactivity; Ova, ovalbumin; Sal, saline; BAL, bronchoalveolar lavage; BALT, bronchus-associated lymphoid tissue; MeCh, methacholine; Penh, enhanced pause
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DB performed mast cell counts in the lung, counted the BAL fluid differentials, performed the invasive pulmonary function studies, and assisted in the preparation of the manuscript. DY determined pulmonary viral titers, performed cytokine assays, and assisted in the preparation of the manuscript. JE collected the airway pathology data, counted the differentials, and assisted in the preparation of the micrographs. DD preformed all of the cell counting and cytospin preparations. SC conceived of the study, performed/assisted in all experiments, and prepared the manuscript. All authors read and approved the final manuscript.
Acknowledgements
Special thanks to my former advisor Dr. James "Jamie" Lee for his encouragement. Insightful comments and critical review of the manuscript was provided by Dr. Karin Peterson and Mr. Tim Jensen. The authors wish to thank those at the Mayo Clinic Arizona Core facilities (Histology: Lisa Barbarisi; Clinical Engineering: Joseph Caplette) and at the Louisiana State University School of Veterinary Medicine Core facilities (Histology: Cheryl Crowder). This publication was made possible by NIH grant number P20 RR020159 from the LSU/Tulane COBRE-CEIDR Program of the National Center for Research Resources. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NIH.
==== Refs
Fraenkel DJ Bardin PG Sanderson G Lampe F Johnston SL Holgate ST Lower airways inflammation during rhinovirus colds in normal and in asthmatic subjects Am J Respir Crit Care Med 1995 151 3 Pt 1 879 886 7881686
Grunberg K Smits HH Timmers MC de Klerk EP Dolhain RJ Dick EC Hiemstra PS Sterk PJ Experimental rhinovirus 16 infection. Effects on cell differentials and soluble markers in sputum in asthmatic subjects Am J Respir Crit Care Med 1997 156 2 Pt 1 609 616 9279247
Lemanske RFJ Dick EC Swenson CA Vrtis RF Busse WW Rhinovirus upper respiratory infection increases airway hyperreactivity and late asthmatic reactions J Clin Invest 1989 83 1 1 10 2536042
Glezen WP Taber LH Frank AL Kasel JA Risk of primary infection and reinfection with respiratory syncytial virus Am J Dis Child 1986 140 6 543 546 3706232
McConnochie KM Roghmann KJ Bronchiolitis as a possible cause of wheezing in childhood: new evidence Pediatrics 1984 74 1 1 10 6739202
Pullan CR Hey EN Wheezing, asthma, and pulmonary dysfunction 10 years after infection with respiratory syncytial virus in infancy Br Med J (Clin Res Ed) 1982 284 6330 1665 1669 6805648
Sims DG Downham MA Gardner PS Webb JK Weightman D Study of 8-year-old children with a history of respiratory syncytial virus bronchiolitis in infancy Br Med J 1978 1 6104 11 14 620129
Mok JY Simpson H Outcome for acute bronchitis, bronchiolitis, and pneumonia in infancy Arch Dis Child 1984 59 4 306 309 6721555
Murray M Webb MS O'Callaghan C Swarbrick AS Milner AD Respiratory status and allergy after bronchiolitis Arch Dis Child 1992 67 4 482 487 1580676
Noble V Murray M Webb MS Alexander J Swarbrick AS Milner AD Respiratory status and allergy nine to 10 years after acute bronchiolitis Arch Dis Child 1997 76 4 315 319 9166022
Sigurs N Bjarnason R Sigurbergsson F Kjellman B Respiratory syncytial virus bronchiolitis in infancy is an important risk factor for asthma and allergy at age 7 Am J Respir Crit Care Med 2000 161 5 1501 1507 10806145
Stein RT Sherrill D Morgan WJ Holberg CJ Halonen M Taussig LM Wright AL Martinez FD Respiratory syncytial virus in early life and risk of wheeze and allergy by age 13 years Lancet 1999 354 9178 541 545 10470697 10.1016/S0140-6736(98)10321-5
Culley FJ Pollott J Openshaw PJ Age at first viral infection determines the pattern of T cell-mediated disease during reinfection in adulthood Journal of Experimental Medicine 2002 196 10 1381 1386 12438429 10.1084/jem.20020943
Dakhama A Park JW Taube C Joetham A Balhorn A Miyahara N Takeda K Gelfand EW The Enhancement or Prevention of Airway Hyperresponsiveness during Reinfection with Respiratory Syncytial Virus Is Critically Dependent on the Age at First Infection and IL-13 Production J Immunol 2005 175 3 1876 1883 16034131
Graham BS Perkins MD Wright PF Karzon DT Primary respiratory syncytial virus infection in mice J Med Virol 1988 26 2 153 162 3183639
Karber G Beitrag zur kollektiven behandlung pharmakologischer reihenversuche Arch exp Path Pharmak 1931 162 480 483 10.1007/BF01863914
Spearman C The Method of Right and Wrong Cases (constatn stimuli) Without Gauss's Formulae. Brit J Psychol 1908 2 227 242
Cormier SA Yuan S Crosby JR Protheroe CA Dimina DM Hines EM Lee NA Lee JJ T(H)2-mediated pulmonary inflammation leads to the differential expression of ribonuclease genes by alveolar macrophages Am J Respir Cell Mol Biol 2002 27 6 678 687 12444027
Hamelmann E Schwarze J Takeda K Oshiba A Larsen GL Irvin CG Gelfand EW Noninvasive measurement of airway responsiveness in allergic mice using barometric plethysmography Am J Respir Crit Care Med 1997 156 3 Pt 1 766 775 9309991
Lee JJ Dimina D Macias MMP Ochkur SI McGarry MP O'Neill KR Protheroe C Pero R Nguyen T Cormier SA Lenkiewicz E Colbert D Rinaldi L Ackerman SJ Irvin CG Lee NA Defining a Link with Asthma in Mice Congenitally Deficient in Eosinophils Science 2004 305 5691 1773 1776 15375267 10.1126/science.1099472
Nguyen TD Dobos KA Cormier SA Hulberd EMH Lenkiewicz E Protheroe CA Dimina DM Ochkur SI McGarry MP Lee NA Lee JJ The Development and Characterization of Novel Eosinophil-Specific Polyclonal Antibody with Pan-Species Specificity That Includes Humans International Conference of the American Thoracic Society 2003 Orlando, FL, USA
Denzler KL Borchers MT Crosby JR Cieslewicz G Hines EM Justice JP Cormier SA Lindenberger KA Song W Wu W Hazen SL Gleich GJ Lee JJ Lee NA Extensive eosinophil degranulation and peroxidase-mediated oxidation of airway proteins do not occur in a mouse ovalbumin-challenge model of pulmonary inflammation Journal of Immunology 2001 167 3 1672 1682
Denzler KL Farmer SC Crosby JR Borchers MT Cieslewicz G Larson KA Cormier-Regard S Lee NA Lee JJ Eosinophil major basic protein-1 does not contribute to allergen-induced airway pathologies in mouse models of asthma Journal of Immunology 2000 165 10 5509 5517
Shen HHH Ochkur SI McGarry MP Crosby JR Hines EM Borchers MT Wang H Biechele TL O'Neill KR Ansay TL Colbert DC Cormier SA Justice JP Lee NA Lee JJ A causative relationship exists between eosinophils and the development of allergic pulmonary pathologies in the mouse Journal of Immunology 2003 170 3296 3305
Peebles RSJ Hashimoto K Collins RD Jarzecka K Furlong J Mitchell DB Sheller JR Graham BS Immune interaction between respiratory syncytial virus infection and allergen sensitization critically depends on timing of challenges J Infect Dis 2001 184 11 1374 1379 11709778 10.1086/324429
Humbles AA Lloyd CM McMillan SJ Friend DS Xanthou G McKenna EE Ghiran S Gerard NP Yu C Orkin SH Gerard C A Critical Role for Eosinophils in Allergic Airways Remodeling Science 2004 305 5691 1776 1779 15375268 10.1126/science.1100283
Rutigliano JA Graham BS Prolonged Production of TNF-{alpha} Exacerbates Illness during Respiratory Syncytial Virus Infection J Immunol 2004 173 5 3408 3417 15322205
Aung S Rutigliano JA Graham BS Alternative mechanisms of respiratory syncytial virus clearance in perforin knockout mice lead to enhanced disease J Virol 2001 75 20 9918 9924 11559824 10.1128/JVI.75.20.9918-9924.2001
Hussell T Pennycook A Openshaw PJ Inhibition of tumor necrosis factor reduces the severity of virus-specific lung immunopathology Eur J Immunol 2001 31 9 2566 2573 11536154 10.1002/1521-4141(200109)31:9<2566::AID-IMMU2566>3.0.CO;2-L
Prescott SL Macaubas C Holt BJ Smallacombe TB Loh R Sly PD Holt PG Transplacental priming of the human immune system to environmental allergens: universal skewing of initial T cell responses toward the Th2 cytokine profile J Immunol 1998 160 10 4730 4737 9590218
Holt PG Yabuhara A Prescott S Venaille T Macaubas C Holt BJ Bjorksten B Sly PD Allergen recognition in the origin of asthma Ciba Found Symp 1997 206 35 49; discussion 49-55, 106-10 9257004
Adkins B Development of neonatal Th1/Th2 function Int Rev Immunol 2000 19 157 10763707
Domachowske JB Rosenberg HF Eosinophils inhibit retroviral transduction of human target cells by a ribonuclease-dependent mechanism Journal of Leukocyte Biology 1997 62 3 363 368 9307075
Domachowske JB Dyer KD Bonville CA Rosenberg HF Recombinant human eosinophil-derived neurotoxin/RNase 2 functions as an effective antiviral agent against respiratory syncytial virus Journal of Infectious Diseases 1998 177 6 1458 1464 9607820
Domachowske JB Dyer KD Adams AG Leto TL Rosenberg HF Eosinophil cationic protein/RNase 3 is another RNase A-family ribonuclease with direct antiviral activity Nucleic Acids Research 1998 26 14 3358 3363 9649619 10.1093/nar/26.14.3358
Taussig LM Wright AL Holberg CJ Halonen M Morgan WJ Martinez FD Tucson children's respiratory study: 1980 to present J Allergy Clin Immunol 2003 111 4 661 675 12704342 10.1067/mai.2003.162
Schauer U Hoffjan S Bittscheidt J Kochling A Hemmis S Bongartz S Stephan V RSV bronchiolitis and risk of wheeze and allergic sensitisation in the first year of life Eur Respir J 2002 20 5 1277 1283 12449185 10.1183/09031936.02.00019902
Kneyber MCJ Steyerberg EW de Groot R Moll HA Long-term effects of respiratory syncytial virus (RSV) bronchiolitis in infants and young children: a quantitative review Acta Paediatr 2000 89 6 654 660 10914957 10.1080/080352500750043945
Sigurs N Clinical perspectives on the association between respiratory syncytial virus and reactive airway disease Respir Res 2002 3 Suppl 1 S8 S14 12119052 10.1186/rr186
Sigurs N Epidemiologic and clinical evidence of a respiratory syncytial virus-reactive airway disease link Am J Respir Crit Care Med 2001 163 3 Pt 2 S2 6 11254543
Holt PG Sly PD Interactions between RSV infection, asthma, and atopy: unraveling the complexities J Exp Med 2002 196 10 1271 1275 12438419 10.1084/jem.20021572
Glezen WP Greenberg SB Atmar RL Piedra PA Couch RB Impact of respiratory virus infections on persons with chronic underlying conditions Jama 2000 283 4 499 505 10659876 10.1001/jama.283.4.499
Tuffaha A Gern JE Lemanske RFJ The role of respiratory viruses in acute and chronic asthma Clin Chest Med 2000 21 2 289 300 10907589 10.1016/S0272-5231(05)70267-7
|
16242038
|
PMC1283984
|
CC BY
|
2021-01-04 16:36:27
|
no
|
Respir Res. 2005 Oct 21; 6(1):122
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-122
|
oa_comm
|
==== Front
Saline Syst
Saline Syst
Saline Systems
1746-1448
BioMed Central
1746-1448-1-8
16242015
10.1186/1746-1448-1-8
Research
Endospores of halophilic bacteria of the family Bacillaceae isolated from non-saline Japanese soil may be transported by Kosa event (Asian dust storm)
Echigo Akinobu [email protected]
Hino Miki [email protected]
Fukushima Tadamasa [email protected]
Mizuki Toru [email protected]
Kamekura Masahiro [email protected]
Usami Ron [email protected]
1 Department of Applied Chemistry, Faculty of Engineering, Toyo University, 2100 Kujirai, Kawagoe, Saitama 350-8585, Japan
2 Bio-Nano Electronics Research Centre, Toyo University, 2100 Kujirai, Kawagoe, Saitama 350-8585, Japan
3 Noda Institute for Scientific Research, 399 Noda, Noda, Chiba 278-0037, Japan
2005
20 10 2005
1 88
11 7 2005
20 10 2005
Copyright © 2005 Echigo et al; licensee BioMed Central Ltd.
2005
Echigo et al; licensee BioMed Central Ltd.
https://creativecommons.org/licenses/by/2.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Generally, extremophiles have been deemed to survive in the extreme environments to which they had adapted to grow. Recently many extremophiles have been isolated from places where they are not expected to grow. Alkaliphilic microorganisms have been isolated from acidic soil samples with pH 4.0, and thermophiles have been isolated from samples of low temperature. Numerous moderately halophilic microorganisms, defined as those that grow optimally in media containing 0.5–2.5 Molar (3–15%) NaCl, and halotolerant microorganisms that are able to grow in media without added NaCl and in the presence of high NaCl have been isolated from saline environments such as salterns, salt lakes and sea sands. It has tacitly been believed that habitats of halophiles able to grow in media containing more than 20% (3.4 M) are restricted to saline environments, and no reports have been published on the isolation of halophiles from ordinary garden soil samples.
Results
We demonstrated that many halophilic bacteria that are able to grow in the presence of 20% NaCl are inhabiting in non-saline environments such as ordinary garden soils, yards, fields and roadways in an area surrounding Tokyo, Japan. Analyses of partial 16S rRNA gene sequences of 176 isolates suggested that they were halophiles belonging to genera of the family Bacillaceae, Bacillus (11 isolates), Filobacillus (19 isolates), Gracilibacillus (6 isolates), Halobacillus (102 isolates), Lentibacillus (1 isolate), Paraliobacillus (5 isolates) and Virgibacillus (17 isolates). Sequences of 15 isolates showed similarities less than 92%, suggesting that they may represent novel taxa within the family Bacillaceae.
Conclusion
The numbers of total bacteria of inland soil samples were in a range from 1.4 × 107/g to 1.1 × 106/g. One tenth of the total bacteria was occupied by endospore-forming bacteria. Only very few of the endospore-forming bacteria, roughly 1 out of 20,000, are halophilic bacteria. Most of the halophilic bacteria were surviving as endospores in the soil samples, in a range of less than 1 to about 500/g soil. Samples collected from seashore in a city confronting Tokyo Bay gave the total numbers of bacteria and endospores roughly 1000 time smaller than those of inland soil samples. Numbers of halophilic bacteria per gram, however, were almost the same as those of inland soil samples. A possible source of the halophilic endospore originating from Asian dust storms is discussed.
==== Body
pmcBackground
Extremophiles are microorganisms adapted to grow in conditions such as extreme pH, temperature, salinity and absence of oxygen [1]. The representatives are acidophiles (Thiobacillus ferroxidans), alkaliphiles (Bacillus alcalophilus), hyperthermophiles (Thermoproteus tenax), extreme halophiles (Halobacterium salinarum) and methanogens (Methanobacterium formicicum). In general, it has been believed that they survive in the extreme environments to which they had adapted to grow. Many extremophiles, however, have been isolated from places where they are not expected to grow. Alkaliphilic microorganisms were isolated from acidic soil samples with pH 4.0 as well as from neutral and alkaline soil [2]. Thermophiles have been isolated from environments of high temperature and also from samples of lower temperature such as soil, food, animal's excrement and seawater [3]. For example, Bacillus stearothermophilus (now Geobacillus stearothermophilus) and Clostridium thermoautotrophicus (now Moorella thermoautotrophica) were isolated from ordinary soil. Strictly anaerobic bacteria such as methanogens, sulfate-reducers, and homoacetogens were isolated from rice paddies during dry fallow period, arable soils, and even from desert soils [4,5]. Thus, the notion that isolation of an organism from a given environment does not mean that the organism is growing in that environments, but just surviving is now generally accepted.
Halophilic microorganisms are adapted to conditions of high salinity and require a certain concentration of NaCl for their optimum growth [6,7]. They have been isolated from various saline environments such as salt lakes (eg. the Dead Sea, the Great Salt Lake), salterns, solar salts and subsurface salt formation. Extremely halophilic microorganisms require high concentration of NaCl for their growth, with optimum concentrations of 2.5–5.2 M (15–30%). Haloarcula vallismortis and Haloterrigena turkmenica for example, have been isolated from salt pool of Death Valley, California, and saline soil of Turkmenia, respectively [8,9]. Moderate halophiles are defined as those that grow optimally in media containing 0.5–2.5 M (3–15%) NaCl, such as Halomonas maura isolated from a saltern in Morocco, and Marinococcus halophilus isolated from sea sands [10,11]. Halotolerant microorganisms possess the ability to grow in media without added NaCl and also in the presence of high concentrations of NaCl. For example, Halobacillus salinus isolated from a salt lake in Korea is able to grow without added salt and in media containing up to 23% NaCl [12].
Are halophiles inhabiting non-saline environments such as garden soil, yards and field? Bacillus clarkii, B. agaradhaerens and B. pseudofirmus are examples of halotolerant bacteria isolated from soil samples that were shown to be tolerant up to 16% or 17% NaCl [13]. It has, however, been tacitly believed that habitats of halophiles able to grow in media containing higher concentrations, let's say 20% (3.4 M), are restricted to saline environments [14,15], and no reports have been published on the isolation of microorganisms able to grow at 20% or higher NaCl concentrations from ordinary, non-saline soil samples. In 1980 Onishi et al. [14] surveyed extensively the occurrence of halophilic bacteria in more or less saline samples collected in Japan. They adopted enrichment culture in a medium containing 4 M (23.4%) NaCl, a customary concentration for the cultivation of Halobacterium spp. They isolated 168 strains finally, but no enrichment was obtained from one third of 287 samples of sea sands and seaweeds collected on seashore. They did not include ordinary garden soil samples. It should be pointed out that a non-pigmented haloarchaeon strain 172P1 (designated later as Natrialba asiatica [16]) was isolated during their survey from dry beach sands with granular salts attached.
In this study, we defined "halophilic bacteria", for convenience, as microorganisms that form colonies on agar plates of a complex medium with 20% added NaCl, and demonstrated that halophilic bacteria are inhabiting in non-saline environments such as ordinary garden soils, yards, fields and roadways in an area surrounding Tokyo, Japan. Phylogenetic analyses of the isolates suggested that they were halophiles belonging to genera of the family Bacillaceae.
Results
Isolation of halophilic bacteria from soil samples
Soil samples (0.5 g each) taken from 360 places were spread on agar plates containing 20% NaCl, with pH adjusted to 5.0, 7.0 and 9.0 respectively. The pH of the soil samples ranged between 5.0 and 6.0. After incubation of plates for 3 weeks at 37°C, colony formations were observed in 132 soil samples (red circles in Fig. 1), at least on one of the three agar plates. Numbers of colonies per plate ranged from 1 on 51 plates to 40 on 1 plate. The sum of colonies amounted to 49 from the medium of pH 9.0, 534 from the pH 7.0 medium, and 61 from the medium of pH 5.0. By inspecting each plate, representative colonies were picked up and transferred to fresh plates and purified by plating out of serial dilutions. A few isolates gradually failed to form colonies on fresh plates. When colonies failed to grow on fresh 20% NaCl plates, concentration of NaCl was decreased to 15 or 10%. It was observed that 26 strains failed to grow in the presence of 20% NaCl. According to our definition of the present paper, these strains are not 'halophilic bacteria', but these strains were included in the further characterization (see discussion).
Figure 1 Collection sites of the 360 soil samples. Red circles; colonies were detected from at least one of the three plates of different pH, black circles; colonies were not detected. A white double circle indicates Tokyo Station, and a white circle indicates Narita Airport.
Finally, 176 strains were obtained (Table 1): 27 strains from 23 samples on alkaline medium (pH 9.0), 139 strains from 120 samples on neutral medium (pH 7.0), and 10 strains from 9 samples on acidic medium (pH 5.0). Endospores were observed by microscope after spore staining [17]. These strains have been kept at 5°C on agar plates of 10% NaCl.
On the other hand, from 228 soil samples (black circles in Fig. 1), about two thirds of the 360 samples collected, no colonies appeared on any plates of the three different pH values. There was no distinct bias in the distribution of black and red circles. In order to estimate if those soil samples contain indeed no microorganisms able to grow at 20% NaCl, two soil samples were randomly picked up, and 0.5 g each was spread on to 10 agar plates of pH 7.0. The colony numbers per plate ranged from 0 on 3 plates to 5 in 1 plate, amounting to 14 in sample 1. From another sample the numbers were from 0 on 3 plates to 4 on 1 plate, amounting to 14 colonies. These data may suggest that halophilic bacteria able to grow at 20% NaCl inhabit any soil samples, at least at a frequency of 1 c.f.u. (colony forming unit)/g soil, in the area we investigated.
Table 1 Strains isolated from ordinary soil samples on agar plates containing 20% NaCl.
Strain No. Sampling site Pig. NaCl (M) pH Similarity Tentatively assigned to
Range Optimum Range Optimum (%)
3 Omiya, S - 0.9–2.6 0.9–1.7 6.5–10.0 8.5–9.5 100 B. haloalkaliphilus (AJ238041)
12 Showa, S - 0.9–4.3 1.7–2.6 6.5–10.0 8.5–9.5 100 B. haloalkaliphilus
27 Takasaki, G - 1.7–3.4 1.7–2.6 6.5–10.0 8.5–9.5 100 B. haloalkaliphilus
29 Kamagaya, C - 1.7–4.3 1.7–2.6 6.5–10.0 8.5–9.5 100 B. haloalkaliphilus
1 Wako, S - 0.9–4.3 1.7–2.6 6.5–10.0 8.5–9.5 99.8 B. haloalkaliphilus
28 Sakado, S - 1.7–2.6 1.7–2.6 6.5–10.0 8.5–9.5 99.8 B. haloalkaliphilus
7 Higashichichibu, S - 0.9–1.7 0.9–1.7 6.5–10.0 8.5–9.5 99.6 B. haloalkaliphilus
8 Higashichichibu, S - 0.9–2.6 0.9–1.7 6.5–10.0 8.5–9.5 99.6 B. haloalkaliphilus
2 Yachiyo, C - 0–3.4 0.9–1.7 6.5–10.0 8.5–9.5 98.4 B. haloalkaliphilus
18 Kawagoe, S - 0.9–4.3 1.7–2.6 6.5–10.0 8.5–9.5 98.0 B. haloalkaliphilus
31 Okabe, S - 0–3.4 1.7–2.6 6.5–10.0 6.5–7.5 97.2 F. milosensis (AJ238042)
19 Wako, S - 0.9–4.3 1.7–2.6 6.5–10.0 8.5–9.5 94.0 F. milosensis
9 Shiki, S - 0–3.4 0.9–1.7 6.5–10.0 6.5–7.5 96.1 G. halotolerans (AB101591)
17 Urawa, S - 0–4.3 0.9–1.7 6.5–10.0 8.5–9.5 91.4 B. agaradhaerens (X76445)
22 Kawagoe, S - 1.7–4.3 1.7–2.6 6.5–10.0 8.5–9.5 88.8 H. trueperi (AJ310149)
14 Kawagoe, S Y 0.9–3.4 1.7–2.6 6.5–10.0 8.5–9.5 88.7 H. trueperi
13 Kawagoe, S Y 0.9–3.4 1.7–2.6 6.5–10.0 8.5–9.5 88.2 H. trueperi
25 Okegawa, S - 1.7–3.4 1.7–2.6 6.5–10.0 8.5–9.5 88.0 H. trueperi
10 Tsurugashima, S B 1.7–2.6 1.7–2.6 6.5–10.0 8.5–9.5 88.0 'B. nitritophilus' (AJ309562)
11 Showa, S Y 0–3.4 1.7–2.6 6.5–10.0 8.5–9.5 87.8 'B. nitritophilus'
16 Iruma, S B 1.7–3.4 1.7–2.6 6.5–10.0 8.5–9.5 87.7 'B. nitritophilus'
4 Matsubushi, S - 1.7–4.3 1.7–2.6 6.5–10.0 8.5–9.5 87.5 'B. nitritophilus'
21 Omiya, S B 1.7–4.3 1.7–2.6 6.5–10.0 8.5–9.5 87.5 'B. nitritophilus'
15 Omiya, S B 1.7–3.4 1.7–2.6 6.5–10.0 8.5–9.5 87.3 'B. nitritophilus'
5 Kawasaki, K B 1.7–4.3 1.7–2.6 6.5–10.0 8.5–9.5 86.9 'B. nitritophilus'
24 Soka, S - 0–4.3 0.9–1.7 6.5–10.0 8.5–9.5 86.9 'B. nitritophilus'
30 Okabe, S - 1.7–4.3 1.7–2.6 6.5–10.0 8.5–9.5 87.3 'Pc. psychrotoleratus' (AF324659)
61 Omiya, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 97.3 F. milosensis (AJ238042)
66 Koga, I - 0–2.6 0–0.9 6.5–8.0 6.5–7.5 97.1 F. milosensis
173 Tsurugashima, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 97.1 F. milosensis
106 Showa, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 97.0 F. milosensis
112 Yachiyo, C - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 97.0 F. milosensis
113 Yachiyo, C Y 0–3.4 0–0.9 6.5–8.0 6.5–7.5 97.0 F. milosensis
117 Niiza, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 97.0 F. milosensis
168 Tsurugashima, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 97.0 F. milosensis
176 Kawagoe, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 97.0 F. milosensis
70 Okegawa, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.9 F. milosensis
105 Kashiwa, C - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.9 F. milosensis
116 Niiza, S Y 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.9 F. milosensis
172 Ageo, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.9 F. milosensis
69 Kitamoto, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.7 F. milosensis
170 Asaka, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.7 F. milosensis
60 Kasukabe, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.5 F. milosensis
185 Urawa, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.4 F. milosensis
74 Hanyu, S - 0–3.4 0–0.9 5.5–8.0 6.5–7.5 96.3 G. halotolerans (AB101591)
75 Hanyu, S - 0–2.6 0–0.9 6.5–8.0 6.5–7.5 96.3 G. halotolerans
76 Katsushika, T - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.3 G. halotolerans
102 Omiya, S - 0–2.6 0–0.9 5.5–8.0 6.5–7.5 96.3 G. halotolerans
72 Kawagoe, S - 0–2.6 0–0.9 6.5–8.0 6.5–7.5 96.1 G. halotolerans
77 Katsushika, T - 0–3.4 1.7–2.6 6.5–8.0 6.5–7.5 94.8 H. karajensis (AJ486874)
51 Okegawa, S - 0–3.4 1.7–2.6 6.5–8.0 6.5–7.5 100 H. litoralis (X94558)
54 Fujimi, S - 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 100 H. litoralis
92 Fujimi, S Y 0.9–2.6 0.9–1.7 6.5–8.0 6.5–7.5 100 H. litoralis
188 Ageo, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 100 H. litoralis
195 Fujimi, S - 0–4.3 0–0.9 6.5–8.0 6.5–7.5 99.5 H. litoralis
200 Kawagoe, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 99.4 H. litoralis
91 Itabashi, T - 0.9–3.4 0.9–2.6 6.5–8.0 6.5–7.5 98.8 H. litoralis
130 Sakado, S - 0.9–2.6 0.9–1.7 6.5–8.0 6.5–7.5 98.8 H. litoralis
196 Urawa, S Y 0–3.4 0.9–1.7 5.5–8.0 6.5–7.5 98.6 H. litoralis
152 Kawagoe, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 97.2 H. litoralis
136 Urawa, S - 0–3.4 1.7–2.6 6.5–8.0 6.5–7.5 97.0 H. litoralis
97 Higashichichibu, S - 0–3.4 0–0.9 5.5–8.0 6.5–7.5 96.7 H. litoralis
154 Ina, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.6 H. litoralis
78 Shiki, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.4 H. litoralis
132 Urawa, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.4 H. litoralis
180 Toda, S - 0.9–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.4 H. litoralis
107 Showa, S Y 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 96.3 H. litoralis
143 Nerima, T - 0–3.4 0.9–1.7 5.5–8.0 6.5–7.5 96.3 H. litoralis
162 Kawagoe, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.3 H. litoralis
86 Wako, S - 0–2.6 0.9–2.6 6.5–8.0 6.5–7.5 96.2 H. litoralis
88 Fujimi, S - 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 96.2 H. litoralis
119 Toride, I - 0.9–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.2 H. litoralis
131 Iruma, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.2 H. litoralis
133 Urawa, S - 0.9–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.2 H. litoralis
164 Ota, T - 0.9–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.2 H. litoralis
192 Tama, T - 0–4.3 0.9–1.7 6.5–8.0 6.5–7.5 96.2 H. litoralis
55 Wako, S - 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 96.1 H. litoralis
68 Okegawa, S - 0–3.4 1.7–2.6 6.5–8.0 6.5–7.5 96.1 H. litoralis
84 Higashimurayama, T Y 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 96.1 H. litoralis
85 Toshima, T - 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 96.1 H. litoralis
95 Soka, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.1 H. litoralis
100 Higashichichibu, S - 0–3.4 0.9–2.6 5.5–8.0 6.5–7.5 96.1 H. litoralis
104 Omiya, S Y 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.1 H. litoralis
124 Koto, T - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.1 H. litoralis
144 Sakado, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.1 H. litoralis
145 Sakado, S - 0–4.3 0.9–1.7 6.5–8.0 6.5–7.5 96.1 H. litoralis
151 Tsurugashima, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.1 H. litoralis
184 Yoshimi, S Y 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.1 H. litoralis
203 Omiya, S - 0.9–3.4 0.9–1.7 5.5–8.0 6.5–7.5 96.1 H. litoralis
56 Fujimi, S - 0.9–2.6 0.9–1.7 6.5–8.0 6.5–7.5 96.0 H. litoralis
62 Higashimatsuyama, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.0 H. litoralis
65 Yachiyo, C - 0.9–2.6 0.9–1.7 6.5–8.0 6.5–7.5 96.0 H. litoralis
83 Nerima, T Y 0.9–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.0 H. litoralis
89 Fujimi, S - 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 96.0 H. litoralis
109 Kyowa, I Y 0–4.3 0.9–2.6 6.5–8.0 6.5–7.5 96.0 H. litoralis
120 Toride, I - 0.9–3.4 0.9–2.6 6.5–8.0 6.5–7.5 96.0 H. litoralis
128 Tsurugashima, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 96.0 H. litoralis
146 Kawagoe, S - 0–2.6 0–0.9 5.5–8.0 6.5–7.5 96.0 H. litoralis
175 Tsurugashima, S - 0–4.3 0–0.9 6.5–8.0 6.5–7.5 96.0 H. litoralis
182 Niiza, S - 0.9–4.3 0.9–1.7 6.5–8.0 6.5–7.5 96.0 H. litoralis
189 Ageo, S - 0–2.6 0.9–1.7 5.5–8.0 6.5–7.5 96.0 H. litoralis
57 Itabashi, T - 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 95.9 H. litoralis
63 Nerima, T - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 95.9 H. litoralis
110 Kyowa, I Y 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 95.9 H. litoralis
127 Tsurugashima, S - 0.9–3.4 0.9–1.7 6.5–8.0 6.5–7.5 95.9 H. litoralis
187 Koga, I - 0.9–3.4 0.9–1.7 6.5–8.0 6.5–7.5 95.9 H. litoralis
80 Koshigaya, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 95.8 H. litoralis
155 Omiya, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 95.8 H. litoralis
166 Tsurugashima, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 95.8 H. litoralis
202 Omiya, S - 0.9–2.6 0.9–1.7 6.5–8.0 6.5–7.5 95.8 H. litoralis
59 Higashichichibu, S Y 0–3.4 0–0.9 6.5–8.0 6.5–7.5 95.7 H. litoralis
79 Shiki, S Y 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 95.7 H. litoralis
159 Omiya, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 95.7 H. litoralis
156 Omiya, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 95.6 H. litoralis
178 Urawa, S - 0–4.3 0.9–1.7 6.5–8.0 6.5–7.5 95.6 H. litoralis
165 Warabi, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 95.2 H. litoralis
52 Omiya, S Y 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 100 H. trueperi (AJ310149)
64 Shiki, S - 0.9–3.4 1.7–2.6 5.5–8.0 6.5–7.5 100 H. trueperi
67 Okegawa, S - 0.9–3.4 1.7–2.6 6.5–8.0 6.5–7.5 100 H. trueperi
82 Omiya, S - 1.7–2.6 1.7–2.6 6.5–8.0 6.5–7.5 100 H. trueperi
96 Soka, S - 0.9–3.4 1.7–2.6 6.5–8.0 6.5–7.5 100 H. trueperi
98 Higashimatsuyama, S Y 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 100 H. trueperi
99 Higashimatsuyama, S Y 0.9–3.4 0.9–1.7 6.5–8.0 6.5–7.5 100 H. trueperi
111 Kyowa, I - 0–3.4 0.9–1.7 5.5–8.0 6.5–7.5 100 H. trueperi
114 Shiki, S - 0.9–3.4 0.9–1.7 5.5–8.0 6.5–7.5 100 H. trueperi
118 Toride, I Y 0.9–2.6 0.9–1.7 6.5–8.0 6.5–7.5 100 H. trueperi
122 Urawa, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 100 H. trueperi
150 Tsurugashima, S Y 0–3.4 0–0.9 6.5–8.0 6.5–7.5 99.8 H. trueperi
181 Higashimatsuyama, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 99.7 H. trueperi
157 Nagareyama, C - 0–4.3 1.7–2.6 6.5–8.0 6.5–7.5 99.6 H. trueperi
58 Kamagaya, C Y 0–3.4 0–0.9 5.5–8.0 6.5–7.5 99.5 H. trueperi
53 Kamagaya, C Y 0–3.4 0–0.9 5.5–8.0 6.5–7.5 99.4 H. trueperi
87 Kawasaki, K Y 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 99.3 H. trueperi
94 Soka, S - 0–3.4 0.9–2.6 6.5–8.0 6.5–7.5 99.3 H. trueperi
153 Warabi, S Y 0–3.4 0–0.9 6.5–8.0 6.5–7.5 99.2 H. trueperi
163 Kawagoe, S - 0–3.4 0–0.9 5.5–8.0 6.5–7.5 99.2 H. trueperi
142 Kawagoe, S Y 0–3.4 0.9–1.7 5.5–8.0 6.5–7.5 99.0 H. trueperi
140 Urawa, S Y 0–4.3 0.9–2.6 5.5–8.0 6.5–7.5 98.9 H. trueperi
139 Urawa, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 98.6 H. trueperi
81 Kasukabe, S - 0–3.4 0.9–1.7 5.5–8.0 6.5–7.5 98.3 H. trueperi
129 Tsurugashima, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 98.3 H. trueperi
193 Wako, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 98.1 H. trueperi
90 Hatoyama, S - 1.7–2.6 1.7–2.6 6.5–8.0 6.5–7.5 97.1 H. trueperi
158 Nagareyama, C - 0.9–1.7 0.9–1.7 6.5–8.0 6.5–7.5 96.4 H. trueperi
108 Iwatsuki, S Y 1.7–3.4 1.7–2.6 6.5–8.0 6.5–7.5 100 L. salicampi (AY057394)
101 Omiya, S - 0–3.4 0–0.9 5.5–8.0 6.5–7.5 96.0 P. ryukyuensis (AB087828)
174 Higashimatsuyama, S Y 0–4.3 0–0.9 5.5–8.0 6.5–7.5 96.0 P. ryukyuensis
177 Kamifukuoka, S - 0–3.4 0–0.9 5.5–8.0 6.5–7.5 95.8 P. ryukyuensis
198 Kiyose, T - 0–3.4 0–0.9 5.5–8.0 6.5–7.5 95.6 P. ryukyuensis
169 Ageo, S - 0–4.3 0–0.9 5.5–8.0 6.5–7.5 95.5 P. ryukyuensis
121 Toride, I P 0.9–3.4 0.9–1.7 6.5–8.0 6.5–7.5 93.5 V. carmonensis (AJ316302)
141 Ranzan S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 99.9 V. halodenitrificans (AB021186)
148 Sakado, S - 0–3.4 0–0.9 5.5–8.0 6.5–7.5 99.7 V. halodenitrificans
147 Matsubushi, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 99.5 V. halodenitrificans
160 Fujioka, Tg - 0–3.4 0–0.9 5.5–8.0 6.5–7.5 94.4 V. halodenitrificans
137 Urawa, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 94.1 V. halodenitrificans
201 Omiya, S - 0–3.4 0–0.9 5.5–8.0 6.5–7.5 100 V. marismortui (AJ009793)
126 Tsurugashima, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 98.3 V. marismortui
123 Urawa, S - 0–3.4 0–0.9 6.5–8.0 6.5–7.5 97.3 V. marismortui
161 Tsurugashima, S P 0–3.4 0–0.9 6.5–8.0 6.5–7.5 96.9 V. picturae (AJ315060)
138 Urawa, S P 0.9–1.7 0.9–1.7 6.5–8.0 6.5–7.5 96.7 V. picturae
179 Urawa, S - 1.7–4.3 1.7–2.6 6.5–8.0 6.5–7.5 95.9 V. picturae
149 Ryugasaki, I - 0.9–2.6 0.9–1.7 6.5–8.0 6.5–7.5 95.8 V. picturae
134 Kamifukuoka, S P 0.9–3.4 0.9–1.7 6.5–8.0 6.5–7.5 95.4 V. picturae
135 Urawa, S P 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 95.1 V. picturae
73 Sakado, S - 1.7–4.3 1.7–2.6 6.5–8.0 6.5–7.5 88.4 B. agaradhaerens (X76445)
219 Kasukabe, S - 0–3.4 0.9–1.7 5.5–8.0 6.5–7.5 100 B. megaterium (AY553118)
216 Sekijo, I - 0–1.7 0–0.9 5.5–8.0 6.5–7.5 99.3 H. litoralis (X94558)
214 Sayama, S - 1.7–2.6 1.7–2.6 5.5–8.0 6.5–7.5 96.3 H. litoralis
211 Higashichichibu, S - 0–3.4 0.9–1.7 5.5–8.0 6.5–7.5 96.1 H. litoralis
213 Sekijo, I - 0–1.7 0–0.9 5.5–8.0 6.5–7.5 95.9 H. litoralis
215 Kawagoe, S - 0–4.3 0.9–1.7 5.5–8.0 6.5–7.5 95.9 H. litoralis
212 Sayama, S - 0–3.4 0.9–1.7 6.5–8.0 6.5–7.5 94.9 H. litoralis
218 Toride, I - 0–4.3 0–0.9 6.5–8.0 6.5–7.5 97.5 H. trueperi (AJ310149)
220 Hachioji, T - 0–3.4 0.9–1.7 5.5–8.0 6.5–7.5 100 V. halodenitrificans (AB021186)
217 Itabashi, T - 0–3.4 0–0.9 5.5–8.0 6.5–7.5 94.5 V. necropolis (AJ315056)
Strains No. 1–31 isolated on alkaline medium (pH 9.0), strains No. 51–203 isolated on neutral medium (pH 7.0) and strains No. 211–220 isolated on acidic medium (pH 5.0). Abbreviations of prefectures: C, Chiba; G, Gunma; I, Ibaraki; K, Kanagawa; S, Saitama; T, Tokyo; Tg, Tochigi. Abbreviations of pigmentation:–, None; B, Brown; P, Pink; Y, Yellow. Abbreviations of generic names: B., Bacillus; F., Filobacillus; G., Gracilibacillus; H., Halobacillus; L., Lentibacillus; P., Paraliobacillus; Pc., Planococcus; V., Virgibacillus.
Anaerobic halophiles
Five soil samples which gave considerable numbers of colonies on the aerobic cultures were spread on agar plates (pH 7.0) and incubated in anaerobic jar for 3 weeks. No colonies were observed at all, while 30 to 40 colonies appeared from the same soil samples incubated aerobically.
Growth range of NaCl concentration and pH in liquid media
Of the 27 strains isolated on the alkaline medium, 21 strains did not grow in media without NaCl, and all except one (strain No. 31) showed optimal growth at alkaline pH, 8.5–9.5 in the presence of 10% NaCl. On the other hand, about 78% (116/149 strains) of the strains isolated on neutral and acidic media were shown to be able to grow without added NaCl, and all strains showed optimal growth at pH 6.5–7.5 (Table 1).
Altogether, 176 strains were divided into 3 groups. Strains of group I and group II may be classified as moderately halophilic bacteria, according to the classification proposed by Kushner et al. [6].
Group I (54 isolates) showed optimal growth between 1.7 and 2.6 M NaCl, and no growth in the absence of added NaCl.
Group II (62 isolates) showed optimal growth between 0.9 and 1.7 M NaCl, and growth in the absence of added NaCl.
Group III (60 isolates) showed optimal growth between 0 and 0.9 M NaCl, and growth in the absence of added NaCl.
Tentative identification of the isolates by partial 16S rRNA gene sequences
Sequences of PCR-amplified partial 16S rRNA genes were determined (about 500 nucleotides), and the 176 strains were tentatively identified by comparing to sequences deposited in databases (Table 1). Summaries of tentative identifications are given in Table 2.
Isolates from the alkaline medium
Ten out of 27 strains showed more than 98% sequence similarities to Bacillus haloalkaliphilus. It was noteworthy that these isolates differed considerably in the range of NaCl for growth; from 0.9–1.7 M to 1.7–4.3 M. Two isolates possessed 97.2 and 94.0% similarities to Filobacillus milosensis, and one isolate was most closely related to Gracilibacillus halotolerans (96.1% similarities). Fourteen other isolates had less than 92% sequence similarities to any deposited sequences. Eight isolates showed 86.9–88.0% similarities to 'Bacillus nitritophilus', and one isolate 87.3% similarity to 'Planococcus psychrotoleratus' but these species have not been validly published. Similarities of complete sequences of the 14 isolates (data not shown) were less than 92%, thus they may represent novel taxa. Out of these 14 isolates, 5 strains were pigmented brown and 2 isolates were yellow.
Table 2 Tentative identification of the isolates by partial 16S rRNA gene sequences.
Tentatively assigned to pH of isolation media
5.0 7.0 9.0
B. haloalkaliphilus (AJ238041) 10
B. megaterium (AY553118) 1
F. milosensis (AJ238042) 17 2
G. halotolerans (AB101591) 5 1
H. karajensis (AJ486874) 1
H. litoralis (X94558) 6 66
H. trueperi (AJ310149) 1 28
L. salicampi (AY057394) 1
P. ryukyuensis (AB087828) 5
V. carmonensis (AJ316302) 1
V. halodenitrificans (AB021186) 1 5
V. marismortui (AJ009793) 3
V. necropolis (AJ315056) 1
V. picturae (AJ315060) 6
No closely related species 1 14
Total 10 139 27
Tentatively assigned to Group I Group II Group III
B. haloalkaliphilus 9 1
B. megaterium 1
F. milosensis 1 2 16
G. halotolerans 1 5
H. karajensis 1
H. litoralis 17 38 17
H. trueperi 9 10 10
L. salicampi 1
P. ryukyuensis 5
V. carmonensis 1
V. halodenitrificans 3 3
V. marismortui 1 2
V. necropolis 1
V. picturae 4 1 1
No closely related species 12 3
Total 54 62 60
Abbreviations of generic names: B., Bacillus; F., Filobacillus; G., Gracilibacillus; H., Halobacillus; L., Lentibacillus; P., Paraliobacillus; Pc., Planococcus; V., Virgibacillus.
Isolates from the neutral medium
Sixty six out of 139 strains showed more than 95% sequence similarities to Halobacillus litoralis: 28 isolates possessed more than 98% similarities to Halobacillus trueperi: 17 isolates more than 96% similarities to Filobacillus milosensis: 6 isolates more than 95% similarities to Virgibacillus picturae: 5 isolates more than 94% similarities to Virgibacillus halodenitrificans: 3 isolates 97.3, 98.3 and 100% similarities to Virgibacillus marismortui: 5 isolates more than 96% similarities to Gracilibacillus halotolerans: 5 isolates more than 95% similarities to Paraliobacillus ryukyuensis. Three other isolates were most closely related to Halobacillus karajensis (94.8%), Virgibacillus carmonensis (93.5%) and Lentibacillus salicampi (100%), respectively. One isolate had less than 89% similarities to any deposited sequences. In the strains from the neutral medium, 29 strains pigmented yellow, and 5 strains were pigmented pink.
Isolates from the acidic medium
Sequences of 6 out of 10 strains were most similar to that of Halobacillus litoralis, with more than 94% sequence similarities. Four other strains were most similar to Virgibacillus necropolis (94.5%), Virgibacillus halodenitrificans (100%), Halobacillus trueperi (97.5%) and Bacillus megaterium (100%).
Ratios of halophilic bacteria to total bacteria
The numbers of total bacteria (c.f.u. on plates with no NaCl) and halophilic bacteria were determined in six inland soil samples; three samples had 30–40 colonies on the 20% NaCl agar plates (pH 7.0), and three samples had just 1 colony on the same medium. As shown in Table 1, the number of c.f.u. on the plates ranged from 340 × 1,000 to 28 × 1,000, thus the total bacteria of inland soil samples were in a range from 1.4 × 107/g (340,000 × 20 × 2) to 1.1 × 106/g (28,000 × 20 × 2). Roughly speaking, one tenth of the total bacteria were occupied by endospore-forming bacteria, and only very few of the endospore-forming bacteria, roughly 1 out of 20,000 or more cells, are halophilic bacteria. Table 3(A) also suggests that most, if not all, of halophilic bacteria are surviving as endospores in the soil samples, in a range of less than 1 to about 500/g soil.
Table 3 Numbers of colony-forming units of samples of inland soil and seashore sands.
Numbers of c.f.u. Numbers of endospore
Sampling site NaCl 0% NaCl 20% NaCl 0% NaCl 20%
(A) Inland soils
Warabi, S 340,000 14 89,000 3
Koto, T 186,000 6 11,000 5
Toshima, T 209,000 8 27,000 1
Ina, S 180,000 0 20,000 0
Shiki, S 113,000 0 13,000 0
Higashimurayama, T 28,000 0 2,000 0
(B) Seashore sands
Tateyama, C (20 m from sea) 282 6 64 6
Tateyama, C (20 m from sea) 99 0 17 0
Tateyama, C (5 m from sea) 501 13 24 4
Tateyama, C (5 m from sea) 381 7 93 1
Tateyama, C (0 m from sea) 244 12 8 1
Tateyama, C (0 m from sea) 289 5 34 4
For the determination of numbers of c.f.u., six inland and six seashore samples (0.5 g) were suspended in distilled water or 10% NaCl solution (2.0 ml), and serially dilutions were spread on agar media (0.1 ml/plate) with no or 20% NaCl, and incubated at 37°C. Numbers of endospore forming bacteria, were determined after heat treatment of the soil suspension at 80°C for 60 min. Numbers are averages of three experiments.
In a separate experiment, 0.5 g of a soil sample was suspended in 2 ml of sterile 10% NaCl, and heated at 80°C. After incubation for 0, 5, 10, 30, and 60 min, three 0.1 ml aliquots were taken, spread on 20% NaCl agar medium (pH 7.0), and incubated for 2 weeks. Data in Table 4 clearly indicated that numbers of c.f.u. (halophilic bacteria) showed little decrease upon heating for 60 min, indicating again that the most of halophilic bacteria were surviving as endospores in soil.
Table 4 Numbers of colonies after heat treatment at 80°C for varying time.
Number of c.f.u
Heat treatment Plate 1 Plate 2 Plate 3 Average
0 min 10 18 14 14.0
5 min 17 13 13 14.3
10 min 14 11 16 13.7
30 min 19 12 13 14.7
60 min 16 13 14 14.3
The soil sample used was 84-10 (Toride, I).
Halophilic bacteria in outdoor accumulations?
Outdoor accumulations (dust, fine sands etc.) were collected from places like roofs of buildings, veranda, cars, and barks of trees, where heavy rainfall would wash away the previous ones. Numbers of bacterial cells and endospores present ranged from 0.8 × 106 to 7.6 × 106/g and from 89 × 103 to 812 × 103/g, respectively. No colonies, however, were observed on any 20% NaCl plates (pH 5.0, 7.0, and 9.0) from 0.5 g samples, even after 8 weeks incubation. Repeated incubations of outdoor accumulations collected 2 weeks after rainfall gave no colonies at all.
Halophilic bacteria in seashore soil (sands)
Six samples were collected from three spots of seashore in Tateyama of Chiba prefecture, a city confronting Tokyo Bay (Fig. 1). Samples were suspended, heated at 80°C, and subjected to colony counting. Table 3(B) showed clearly that the total numbers of bacteria and endospores were roughly 1000 time smaller than those of inland soil samples. Numbers of halophilic bacteria, however, were almost the same as those of inland soil samples.
NaCl contents of samples
Analyses of Cl content of soil samples suggested that NaCl contents of soil samples taken from near seashore were as high as 15–20 mg NaCl/g, whereas the 360 inland soil samples contained less than 1 mg NaCl/g.
Haloarchaea in soil samples?
The agar plate with 20% NaCl used in this study was based on the medium No. 168 recommended by JCM (Japan Collection of Microorganisms) for the cultivation of haloarchaea. All of the pink to brown colonies that might be haloarchaea were picked up, but they all belonged of the family Bacillaceae. To ascertain that haloarchaea are not present in the soil samples, at least to the limit of detection, soil samples were spread on agar plates of 20% NaCl, pH 7.0 and 9.0, supplemented with 30 μg/ml ampicillin. Generally, numbers of colonies decreased to less than one tenth of those obtained on plates without ampicillin. From five soil samples out of 107 samples tested, four colonies were obtained on plates of pH 7.0, and 12 colonies on those of pH 9.0. No colony appeared from seven samples from seashore at Tateyama. The colonies were transferred to liquid media containing ampicillin, and 10 strains that grew were subjected DNA extraction and PCR amplification using both bacterial and archaeal 16S rRNA gene primer sets. All of them yielded amplification bands only with bacterial primers, suggesting they were not haloarchaea but halophilic bacteria harboring plasmids with β-lactamase genes.
Discussion
Definition of "moderate halophiles", "extreme halophiles" and "halotolerant" has long been given by Larsen [18] and Kushner & Kamekura [6]. In this paper, we defined "halophilic bacteria", for convenience, as microorganisms able to form colonies on agar plates containing 20% (3.4 M) added NaCl. Strictly speaking, there existed moderate halophiles that were not able to grow in the presence of 20% NaCl.
Is the distribution of halophilic Bacteria and Archaea restricted?
Since halophilic bacteria have been recognized to live in the Dead Sea [18], numerous halophilic and halotolerant microorganisms, both aerobic and anaerobic, both Bacteria and Archaea, have been isolated from saline environments. Thanks to the enthusiastic devotion of A. Oren on halophilic microorganisms [7], we know that halophilic Bacteria are not restricted to the class Bacilli (Bacillus sensu lato) but distributed through classes of Cyanobacteria, α-, β-, γ-, and δ-proteobacteria, Clostridia, Actinobacteria, Flavobacteria, etc. Although Oren defined the "halophilic" as tolerance to 10% (100 g/L) NaCl in his book [7], some of the halophilic microorganisms are able to grow in the presence of 20% NaCl. We also know that all microorganisms that were intentionally isolated as halophiles are inhabitants of saline environments. On the other hand, some bacteria isolated from soil are able to tolerate high NaCl concentrations. For example, Bacillus clarkii, B. agaradhaerens, and B. pseudofirmus are tolerant up to 16% or 17% NaCl [13]. To the best knowledge of the authors of this paper, no reports have been published on the isolation of microorganisms able to grow at 20% or higher NaCl concentrations from ordinary non-saline soil samples. It has tacitly been believed that habitats of halophiles able to grow in media containing more than 20% are restricted to saline environments [14,15].
Halophilic bacteria are isolated from soil samples
In the present study, we have demonstrated that halophilic bacteria that are able to grow in the presence of 20% NaCl are inhabiting almost everywhere in non-saline environments such as ordinary garden soils, yards, fields and roadways in an area surrounding Tokyo. We isolated 176 strains, and analysis of partial sequences of their 16S rRNA genes showed that some of them possessed similarities higher than 94.8% with those of Bacillus haloalkaliphilus [20], Filobacillus milosensis [21], Gracilibacillus halotolerans [22], Halobacillus karajensis [23], Halobacillus litoralis [24], Halobacillus trueperi [24], Lentibacillus salicampi [25], Paraliobacillus ryukyuensis [26], Virgibacillus halodenitrificans [27], Virgibacillus marismortui [28] and Virgibacillus picturae [29]. Most of the strains of these species have been isolated from saline environments, and were reported to be halophilic, capable of growth at 20% NaCl. All strains of species of the genera of the family Bacillacea were endospore formers, except Bacillus saliphilus [30]. Sequences of 15 isolates (14 isolates from alkaline medium, and one isolated from neutral medium) showed similarities less than 92% to any deposited sequences, thus they may represent novel taxa within the family Bacillaceae. For unknown reason(s), cells of some colonies on the initial isolation plates failed to grow when transferred to fresh plates, but grew on plates with lower NaCl concentrations. Some growth factors present in soil might be responsible for this phenomenon [31].
A large number of halophilic bacteria of group I (no grow without added NaCl) were shown to be alkaliphilic, and most of the group II and III, which grew without added NaCl, were neutrophilic. The haloalkaliphiles, halophilic and alkaliphilic bacteria [2], have been found mainly in extremely alkaline and saline environments, which were distributed in the Rift Valley lakes of East Africa, soda lakes of the United States and Inner Mongolia of China, etc.
Quite interesting is the fact that none of the isolates of halophilic bacteria showed similarities with any halophilic microorganisms of the classes mentioned above other than the endospore-forming Bacillus sensu lato. There remains a possibility, however, that colonies of halophilic bacteria other than Bacilli on agar plates unfortunately escaped from being picked up for purification. Another possibility is that those halophilic bacteria lost the ability to form colonies on the particular agar plates we used during repeated transfers in the purification procedures, or that they simply did not form colonies because of unsuitableness of the composition of agar plates to them.
Are the halophilic Bacilli indigenous to soil?
Spore-formers of the family Bacillaceae were easily isolated from a number of environments by suspending a sample in water and heating at 80°C for 10 to 30 min, even from environments unrelated to their growth requirements. For example, a thermophile Geobacillus stearothermophilus was isolated from ordinary soil, and many of the alkaliphilic Bacillus species have been isolated from soils that were not particularly alkaline [32]. Now we know that roughly one tenth of the culturable total bacteria present in ordinary soil were endospore-forming bacteria, only very few of them were halophilic bacteria, and most of the halophilic bacteria were surviving as endospores. Also, the exact distribution of so called "soil microorganisms" of each taxon may fluctuate depending on the soil and also on season, number of cells of alkaliphiles and methanogens have been shown to be an order of 104 to 105/g of neutral soil [2,4]. The numbers of endospores of the halophilic bacteria ranged from less than 1 to 500/g in our experiments. A question is "Are the endospores of halophilic bacteria indigenous to soil, and if not, where did they come from?" Here we remember that (i) NaCl contents of soil samples taken from near seashore contained as high as 15–20 mg NaCl/g, whereas the 360 inland soil samples contained less than 1 mg NaCl/g, and that (ii) numbers of endospores of halophilic bacteria were almost the same in the inland soil samples and the seashore sand samples. These facts strongly suggest that the endospores of halophilic bacteria are neither from the sea nor from minute highly saline niche produced by evaporation of seawater on seashore. Then, where are they from?
Bacteria in Asian Dust?
Although we have no confirming data at present, we may speculate that the halophilic bacteria have been transported by westerlies either as vegetative cells or as endospores from the indigenous highly saline environments, such as salt lakes in Inner Mongolia or salterns confronting Yellow Sea or East Sea (Japan Sea) in Korea. In fact, several novel genera and species have been isolated from these areas recently. It has been realized that Aeolian dust (mineral dust) and sand storms are plaguing North-East Asia [33]. The storms originate in the arid inland parts of China and Mongolia and blow across the Korean peninsula and Japan. It is believed that cold air masses from Siberia whip deserts and soils eastward after the dry continental winter. The dusts kicked up into the jet stream are carried by the prevailing westerlies across mainland Asia, over the Sea of Japan and Pacific Ocean and reach into the main land United States in just five days. It is demonstrated that the dusts reach even to Hawaii, which is over 6,000 km away [34,35]. China's sand storms are referred to as Huangsha in China, "Asian dust" or Whangsa in Korea, and "Yellow sand" or Kosa event in Japan. The authors of this study believe that bacteria, at least their endospores, thriving in salt lakes, soda lakes, and the surrounding saline soils in the arid region have been kept carried to Japan for thousands of years together with the mineral dusts. A long-distance transport of fungal spores by winds across the English Channel was demonstrated by Hirst et al. [36]
Cells of non-endospore-forming halophilic bacteria and halophilic archaea (see below) thriving in the saline environments would be flown to Japan by westerlies together with the endospores, but they would die sooner or later after they arrive at soil, at least because of the hypotonic conditions caused by rainfall.
How long have they thrived in soil?
If the above speculation is correct, the next question is "At what density are there the endospores of halophilic bacteria in the yellow sand and how long do they survive in soil?" The fact that dusts accumulated for several days after the last rainfall gave no colonies on 20% NaCl agar plates suggests that frequencies of the presence of endospores of halophilic bacteria transported by the westerlies are very low, and 40 colonies that appeared on the agar plate are the result of precipitation of the endospores for a very long time span. Endospores are known to be able to survive for quite a long time, at least several decades, or even thousands of years. There is a famous case that a endospore of bacteria closely related to extant Bacillus sphaericus was revived and cultured from the abdominal contents of extinct bees preserved for 25 to 40 million years in buried Dominican amber [37]. An even more spectacular claim was made that a bacteria closely related to Virgibacillus marismortui was isolated from fluid inclusions in rock salt crystals of Permian age, over 250 million years old [38]. Although there is no well-documented data on the longevity of endospores in the soil, we may expect they can survive at least for decades.
Halophilic strains of the class Bacilli from non-saline environments
Recently, several moderately halophilic bacteria, representing novel species of the genus Halobacillus (97.1 to 98.4% similarities to Halobacillus litoralis in the 16S rRNA gene sequences) were isolated by enrichment culture in a medium containing 20% NaCl from damaged medieval wall paintings and building materials in Austria [39]. To the best knowledge of authors of this paper, this is the only exception of the isolation of halophilic bacteria from non-saline environments. Virgibacillus carmonensis, V. necropolis and V. picturae isolated from samples of biofilm formation on the mural paintings in Spain [28] are moderately halophilic with optimal growth at NaCl concentration of 5 to 10%, but it is not clear if they are able to grow at 20% NaCl.
Conclusion
An answer to a question "Are the endospores of halophilic bacteria indigenous to soil, and if not, where did they come from?" will be that the endospores of halophilic bacteria are NOT indigenous to soil, and they are neither from sea nor from minute highly saline niche produced by evaporation of seawater on seashore. We may speculate that the endospores of halophilic bacteria detected in this study were the results of precipitation in long-term, for years at least, from atmosphere that have been transported by westerlies from the indigenous highly saline environments, such as salt lakes in Inner Mongolia or salterns in Korea.
Methods
Samples of soils, accumulations on roofs etc., and seashore sands
A total of 360 soil samples were collected in an area (103 km by 126 km) surrounding Tokyo, Japan (Tokyo, Saitama-, Chiba-, Kanagawa- and Ibaraki- prefecture) (Fig. 1). Each soil sample was taken from surfaces such as gardens, fields, yards and roadways, which was separated each other by at least 1 km. There exist no highly saline environments in this region such as salterns and salt lakes. On the other hands, 6 accumulation samples were taken from roof of a car, roofs of three buildings, and veranda of two rooms of building in the campus of Toyo University or nearby cities. Several samples were also taken from seashore of Tateyama, a city southern part of Chiba-prefecture confronting Tokyo-Bay (Fig. 1). Each soil sample was collected into 50 ml sterile FALCON tubes (Becton Dickinson) with sterile spatula, and kept at room temperature until use.
Agar media for the isolation of halophilic bacterial and archaeal strains
A complex growth medium contained the following ingredients (per liter). 5.0 g casamino acids (Difco), 5.0 g yeast extract (Difco), 1.0 g sodium glutamate·H2O, 3.0 g trisodium citrate·2H2O, 2.0 g KCl, 0.2 g MgSO4·7H2O, 36 mg FeCl2·4H2O, 200 g (3.4 M) NaCl and 20 g Bacto-agar (Difco), pH 7.2. After autoclaving, pH was adjusted to 5.0, 7.0 or 9.0 by adding pre-calculated amounts of diluted sterile H2SO4 or Na2CO3 solutions. Soil samples (0.5 g each) were placed directly on the three agar plates of different pH, spread with spatula, and incubated at 37°C for 3 weeks in plastic bags to prevent desiccation. Colonies were picked up, transferred to fresh agar plates of the same pH, and pure cultures were obtained by plating serial dilutions and repeated transfers on agar plates of the same medium. In some experiments, the agar plates of pH 7.0 and 9.0 were supplemented with 30 μg/ml ampicillin for the selective isolation of haloarchaeal strains.
Determination of growth range
Growth was determined by inoculating pre-cultures of purified strains into 30 ml test tubes each containing 3 ml liquid media with varying NaCl concentrations (0, 0.9, 1.7, 2.6, 3.4, 4.3 and 5.2 M, pH 7 or 9) and pH (5.0, 5.5, 6.0, 6.5, 7.0, 7.5, 8.0, 8.5, 9.0, 9.5 and 10.0, in the presence of 1.7 M NaCl) and shaken at 37°C with 120 rpm. Growth was monitored by taking 0.1 ml culture broth periodically and measuring absorbance at 660 nm.
Measurements of numbers of total bacterial cells, halophilic bacteria, and endospore-forming bacteria
For the estimation of the numbers of colony forming unit and halophilic bacteria present in the soil samples, 0.5 g samples were suspended in 2 ml of distilled water or sterile 10% NaCl solution, diluted serially with the same solution, and 0.1 ml each was spread on agar plates without added NaCl or of 20% NaCl, respectively, pH 7.0, followed by incubation at 37°C for 2 weeks. Separate experiments on 9 soil samples have shown that increasing concentrations of NaCl of agar plates for colony counting (0, 5, 10, 15 and 20%) just decreased the number of c.f.u. The numbers of total endospore-forming bacteria were determined after heat treatment of the soil suspension at 80°C for 60 min, followed by dilution and spreading on the plates without added NaCl. Heat resistance of endospores is known to differ in different strains of Bacillaceae and depends on the NaCl concentration during heating [40,41].
Spore-staining
Endospore formations of the isolates were examined with a phase-contrast microscope (×1,000) or after spore-staining according to the method of Wirtz-Conklin [17].
Anaerobic growth
For the detection of anaerobic halophilic bacteria, 5 soil samples were spread on the agar plates as described above, and incubated at 37°C in an anaerobic jar using a deoxygenation reagent, Anaero Pauch Anaero, and Anaero Pack Rectangular Jar (Mitsubishi Gas Chemical Co., Inc., Tokyo).
Phylogenetic analysis
Total DNAs were extracted by the method of Cline et al. [42]. The 16S rRNA genes were amplified by PCR and directly sequenced by using the ABI PRISM® BigDye Terminator v3.1 Cycle Sequencing Kits (Applied Biosystems) with the following forward and reverse primers for Bacteria: 5'-AGAGTTTGATCCTGGCTCAG-3' (positions 8–27 according to Escherichia coli numbering) and 5'-GACTACCAGGGTATCTAATC-3' (positions 805–786) on the ABI PRISM® 310 Genetic Analyzer (Applied Biosystems). In some experiments, the complete 16S rRNA genes were amplified by PCR with the forward primer and a reverse primer: 5'-GGCTACCTTGTTACGACTT-3' (positions 1510–1492). Archaeal primer sets were 5'-ATTCCGGTTGATCCTGCCGG-3' (positions 6–25 according to E. coli numbering) and 5'-AGGAGGTGATCCAGCCGCAG-3' (positions 1540–1521). The closest known relatives of the sequenced organisms were determined by sequence database searches. These sequences and those of known related strains retrieved from the DNA Data Bank of Japan [43-45] were aligned using the CLUSTAL W Multiple Sequence Alignment Program [46]. The phylogenetic tree was reconstructed by the neighbour-joining method [47] and was evaluated by bootstrap sampling [48].
Estimation of NaCl contents, pH, and water content of soil samples
Ten grams of each soil sample was suspended in 100 ml of distilled water, shaken for 1 h, and supernatants were obtained by centrifugation. NaCl content in soil samples was calculated by determining Cl content of the soil extract by the method of Mohr [49]. The pH of the soil extract was determined with a pH meter. Water content was determined by measuring the decrease of weight after heating 1 g soil sample at 120°C for 1, 2 and 3 hours.
Authors' contributions
AE carried out isolation of strains and characterization, sequencing of 16S rRNA genes, analyses, study design, and drafted the manuscript.
MH carried out isolation of strains and characterization.
TF participated in the design of study.
TM participated in the design of study.
MK participated in the study design and drafted the manuscript.
RU directed the research and drafted the manuscript.
All authors have read and approved the final manuscript.
Acknowledgements
Part of this study has been supported by a grant for the 21st Century's Center of Excellence Programs organized by the Ministry of Education, Culture, Sports, Science and Technology, Japan, since 2003.
==== Refs
Horikoshi K Grant WD eds Extremophiles – microbial life in extreme environments 1998 New York, Wiley-Liss
Horikoshi K Alkaliphiles: some applications of their products for biotechnology Microbiol Mol Biol Rev 1999 63 735 750 10585964
Sneath PHA Endospore-forming Gram-positive rods and cocci Bergey's Manual of Systematic Bacteriology 1986 2 Baltimore, Williams & Wilkins 1104 1207
Mayer HP Conrad R Factors influencing the populations of methanogenic bacteria and the initiation of methane production upon flooding of paddy soil FEMS Microbiol Ecol 1990 73 103 111 10.1016/0378-1097(90)90656-B
Peters V Conrad R Methanogenic and other strictly anaerobic bacteria in desert soil and other oxic soils Appl Environ Microbiol 1995 61 1673 1676 16535011
Kushner DJ Kamekura M Physiology of halophilic eubacteria Halophilic bacteria 1988 I Boca Raton, CRC Press 109 140
Oren A Halophilic Microorganisms and their environments 2002 Dordrecht, Kluwer Academic
Gonzalez C Gutierrez C Ramirez C Halobacterium vallismortis sp. nov. an amylolytic and carbohydrate-metabolizing, extremely halophilic bacterium Can J Microbiol 1978 24 710 715 667737
Zvyagintseva IS Tarasov AL Extreme halophilic bacteria from saline soils Mikrobiologiya 1987 56 839 844
Bouchotroch S Quesada E del Moral A Llamas I Bejar V Halomonas maura sp. nov., a novel moderately halophilic, exopolysaccharide-producing bacterium Int J Syst Evol Microbiol 2001 51 1625 32 11594589
Hao MV Kocur M Komagata K Marinococcus gen nov., a new genus for motile cocci with meso-diaminopimelic acid in the cell wall; and Marinococcus albus sp. nov. and Marinococcus halophilus (Novitsky and Kushner) comb. nov J Gen Appl Microbiol 1984 30 449 459
Yoon JH Kang KH Park YH Halobacillus salinus sp. nov., isolated from a salt lake on the coast of the East Sea in Korea Int J Syst Evol Microbiol 2003 53 687 93 10.1099/ijs.0.02421-0 12807188
Nielsen P Fritze D Priest FG Phenetic diversity of alkaliphilic Bacillus strains: proposal for nine new species Microbiology 1995 141 1745 1761
Onishi H Fuchi H Konomi K Hidaka O Kamekura M Isolation and distribution of a variety of halophilic bacteria and their classification by salt-response Agric Biol Chem 1980 44 1253 1258
Saiz-Jimenez C Laiz L Occurrence of halotolerant/halophilic bacterial communities in deteriorated monuments Int Biodeterior Biodegrad 2000 46 319 326 10.1016/S0964-8305(00)00104-9
Kamekura M Dyall-Smith ML Taxonomy of the family Halobacteriaceae and the description of two new genera Halorubrobacterium and Natrialba J Gen Appl Microbiol 1995 41 333 350
Murray PR Baron EJ Pfaller MA TeNover FC Yolken RH Manual of Clinical Microbiology 1999 7 Washington, D.C., ASM Press
Larsen H Halophilism The Bacteria 1962 4 297 342
Wilkansky B Life in the Dead Sea Nature 1936 138 467
Fritze D Bacillus haloalkaliphilus sp. nov Int J Syst Bacteriol 1996 46 98 101
Schlesner H Lawson PA Collins MD Weiss N Wehmeyer U Volker H Thomm M Filobacillus milensis gen. nov., sp. nov., a new halophilic spore-forming bacterium with Orn-D-Glu-type peptidoglycan Int J Syst Evol Microbiol 2001 51 425 431 11324591
Waino M Tindall BJ Schumann P Ingvorsen K Gracilibacillus gen. nov., with description of Gracilibacillus halotolerans gen. nov., sp. nov.; transfer of Bacillus dipsosauri to Gracilibacillus dipsosauri comb. nov., and Bacillus salexigens to the genus Salibacillus gen. nov., as Salibacillus salexigens comb. nov Int J Syst Bacteriol 1999 49 821 831 10319508
Amoozegar MA Malekzadeh F Malik KA Schumann P Sproer C Halobacillus karajensis sp. nov., a novel moderate halophile Int J Syst Evol Microbiol 2003 53 1059 1063 10.1099/ijs.0.02448-0 12892126
Spring S Ludwig W Marquez MC Ventosa A Schleifer KH Halobacillus gen. nov., with descriptions of Halobacillus litoralis sp. nov., and Halobacillus trueperi sp. nov., and transfer of Sporosarcina halophila to Halobacillus halophilus comb. nov Int J Syst Bacteriol 1996 46 492 496
Yoon JH Kang KH Park YH Lentibacillus salicampi gen. nov., sp. nov., a moderately halophilic bacterium isolated from a salt field in Korea Int J Syst Evol Microbiol 2002 52 2043 2048 10.1099/ijs.0.02335-0 12508866
Ishikawa M Ishizaki S Yamamoto Y Yamasato K Paraliobacillus ryukyuensis gen. nov., sp. nov., a new Gram-positive, slightly halophilic, extremely halotolerant, facultative anaerobe isolated from a decomposing marine alga J Gen Appl Microbiol 2002 48 269 279 12501437
Yoon JH Oh TK Park YH Transfer of Bacillus halodenitrificans Denariaz et al. 1989 to the genus Virgibacillus as Virgibacillus denitrificans comb. nov Int J Syst Evol Microbiol 2004 54 2163 2167 10.1099/ijs.0.63196-0 15545452
Arahal DR Marquez MC Volcani BE Schleifer KH Ventosa A Bacillus marismortui sp. nov., a new moderately halophilic species from the Dead Sea Int J Syst Bacteriol 1999 49 521 530 10319473
Heyrman J Logan NA Busse HJ Balcaen A Lebbe L Rodriguez-Diaz M Swings J De Vos P Virgibacillus carmonensis sp. nov., Virgibacillus necropolis sp. nov. and Virgibacillus picturae sp. nov., three novel species isolated from deteriorated mural paintings, transfer of the species of the genus Salibacillus to Virgibacillus, as Virgibacillus marismortui comb. nov. and Virgibacillus salexigens comb. nov., and emended description of the genus Virgibacillus Int J Syst Evol Microbiol 2003 53 501 511 10.1099/ijs.0.02371-0 12710619
Romano I Lama L Nicolaus B Gambacorta A Giordanao A Bacillus saliphilus sp. nov., isolated from a mineral pool in Campania, Italy Int J Syst Evol Microbiol 2005 55 159 163 10.1099/ijs.0.63298-0 15653870
Aagot N Nybroe O Nielsen P Johnsen K An altered Pseudomonas diversity is recovered from soil by using nutrient-poor Pseudomonas-selective soil extract media Appl Environ Microbiol 2001 67 5233 5239 10.1128/AEM.67.11.5233-5239.2001 11679350
Krulwich TA Guffani AA Alkalophilic bacteria Annu Rev Microbiol 1989 43 435 463 10.1146/annurev.mi.43.100189.002251 2679360
North-East Asian dust and sand storms growing in scale and intensity; UNEP warns of 'The Globalization of Environmental Problems' http://www.un.org/News/Press/docs/2004/unep216.doc.htm
Duce RA Unni CK Ray BJ Prospero JM Merrill JT Long-range atmospheric transport of soil dust from Asia to the tropical north Pacific: temporal variability Science 1980 209 1522 1524 17745962
Chadwick OA Derry LA Vitousek PM Huebert BJ Hedin LO Changing sources of nutrients during four million years of ecosystem development Nature 1999 397 491 497 10.1038/17276
Hirst JM Stedman OJ Hurst GW Long-distance spore transport: Vertical sections of spore clouds over the sea J Gen Microbiol 1957 48 357 377
Cano RJ Borucki MK Revival and identification of bacterial spores in 25- to 40-million-year-old Dominican amber Science 1995 268 1060 1064 7538699
Vreeland RH Rosenzweig WD Powers DW Isolation of a 250 million-year-old halotolerant bacterium from a primary salt crystal Nature 2000 407 8 10.1038/35038060 10993047
Piñar G Ramos C Rölleke S Schabereiter-Gurtner C Vybiral D Lubiz W Denner EBM Detection of indigenous Halobacillus populations in damaged ancient wall paintings and building materials: molecular monitoring and cultivation Appl Environ Microbiol 2001 67 4891 4895 10.1128/AEM.67.10.4891-4895.2001 11571198
Roberts TA Hitchins AD Gould GW, Hurst A Resistance of spores The Bacterial Spore 1969 London, Academic Press 611 670
Murrell WG Warth AD Campbell LL, Halvorson HO Composition and heat resistance of bacterial spores Spores III 1965 Ann Arbor, American Society for Microbiology 1 24
Cline SW Schalkwyk LC Doolittle WF Transformation of the archaebacterium Halobacterium volcanii with genomic DNA J Bacteriol 1989 171 4987 4991 2768194
Miyazaki S Sugawara H Gojobori T Tateno Y DNA Data Bank of Japan (DDBJ) in XML Nucl Acids Res 2003 30 13 16 10.1093/nar/gkg088
Pearson WR Lipman DJ Improved tools for biological sequence comparison Proc Natl Acad Sci USA 1988 85 2444 2448 3162770
Lipman DJ Pearson WR Rapid and sensitive protein similarity searches Science 1985 227 1435 1441 2983426
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 Nucl Acids Res 1994 22 4673 4680 7984417
Saitou N Nei M The neighbour-joining methods: a new method for reconstructing phylogenetic trees Mol Biol Evol 1987 4 406 425 3447015
Felsenstein J Confidence limits on phylogenies: an approach using the bootstrap Evolution 1985 39 783 791 28561359
American Public Health Association Standard methods for the examination of water and wastewater 1998 20 American Public Health Association, Washington, D.C
|
16242015
|
PMC1283985
|
CC BY
|
2022-09-08 23:19:43
|
no
|
Saline Syst. 2005 Oct 20; 1:8
|
utf-8
|
Saline Syst
| 2,005 |
10.1186/1746-1448-1-8
|
oa_comm
|
==== Front
PLoS Comput BiolPLoS Comput. BiolpcbiplcbploscompPLoS Computational Biology1553-734X1553-7358Public Library of Science San Francisco, USA 1629958910.1371/journal.pcbi.001006205-PLCB-RA-0178R3plcb-01-06-04Research ArticleBioinformatics - Computational BiologyMolecular Biology - Structural BiologyNeuroscienceMus (Mouse)The Association of Tetrameric Acetylcholinesterase with ColQ Tail: A Block Normal Mode Analysis Association of AChE with ColQ TailZhang Deqiang 123*McCammon J. Andrew 12341 Howard Hughes Medical Institute, University of California, San Diego, California, United States of America
2 Department of Chemistry and Biochemistry, University of California, San Diego, California, United States of America
3 Center for Theoretical Biological Physics, University of California, San Diego, California, United States of America
4 Department of Pharmacology, University of California, San Diego, California, United States of America
Murray Diana EditorWeill Medical College of Cornell University, United States of America*To whom correspondence should be addressed. E-mail: [email protected] 2005 18 11 2005 19 10 2005 1 6 e6226 7 2005 19 10 2005 Copyright: © 2005 Zhang and McCammon.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.Acetylcholinesterase (AChE) rapidly hydrolyzes acetylcholine in the neuromuscular junctions and other cholinergic synapses to terminate the neuronal signal. In physiological conditions, AChE exists as tetramers associated with the proline-rich attachment domain (PRAD) of either collagen-like Q subunit (ColQ) or proline-rich membrane-anchoring protein. Crystallographic studies have revealed that different tetramer forms may be present, and it is not clear whether one or both are relevant under physiological conditions. Recently, the crystal structure of the tryptophan amphiphilic tetramerization (WAT) domain of AChE associated with PRAD ([WAT]4PRAD), which mimics the interface between ColQ and AChE tetramer, became available. In this study we built a complete tetrameric mouse [AChET]4–ColQ atomic structure model, based on the crystal structure of the [WAT]4PRAD complex. The structure was optimized using energy minimization. Block normal mode analysis was done to investigate the low-frequency motions of the complex and to correlate the structure model with the two known crystal structures of AChE tetramer. Significant low-frequency motions among the catalytic domains of the four AChE subunits were observed, while the [WAT]4PRAD part held the complex together. Normal mode involvement analysis revealed that the two lowest frequency modes were primarily involved in the conformational changes leading to the two crystal structures. The first 30 normal modes can account for more than 75% of the conformational changes in both cases. The evidence further supports the idea of a flexible tetramer model for AChE. This model can be used to study the implications of the association of AChE with ColQ.
Synopsis
Acetylcholinesterase (AChE) breaks down acetylcholine in the neuromuscular junction and other cholinergic synapses to terminate neuronal signals. AChE exists as tetramers anchored by structural subunits to the cell membranes in the brain or the basal lamina in the neuromuscular junction. Based on a crystal structure of the tetramerization domain of AChE with a proline-rich attachment domain of the anchoring proteins, a symmetric model of the complex of AChE tetramer with the anchoring protein tail was constructed. Block normal mode analysis revealed the presence of several low-frequency, low-barrier normal modes corresponding to inter-subunit motions. Previous crystal structures of AChE tetramer could be rationalized using these normal modes. These low-frequency modes are due to the presence of a flexible hinge in the structure of AChE. This study paints a picture of a flexible AChE tetramer with different conformational states interconverting easily under physiological conditions, which has important implications on the function of AChE. In particular, AChE is not trapped in the compact tetramer structure, for which access of substrate to two of the active sites is somewhat limited. Rather, the tetramer fluctuates to expose all four of its active sites to ensure rapid removal of acetylcholine.
Citation:Zhang D, McCammon JA (2005) The association of tetrameric acetylcholinesterase with ColQ tail: A block normal mode analysis. PLoS Comput Biol 1(6): e62.
==== Body
Introduction
Acetylcholinesterase (AChE; E.C. 3.1.1.7) rapidly hydrolyzes acetylcholine to terminate neurotransmissions at cholinergic synapses [1,2]. The reaction is very fast, approaching the diffusion limit. AChE has three different molecular forms due to an alternate splicing scheme at the C-terminus [3]. The T-subtype (AChET) with a 40-residue C-terminal “t-peptide” is the only form expressed in the brain and adult muscles of normal adult mammals [4]. In vertebrate cholinergic synapses, tetramers of AChET are associated with either collagen-like Q subunit (ColQ) or transmembrane proline-rich membrane-anchoring protein (PRiMA) [5,6]. ColQ is a structural protein that anchors AChET to the synaptic basal lamina [5,7], and PRiMA is a membrane protein that anchors AChET to the membrane of neuronal synapses in the brain [6]. They both contain a proline-rich attachment domain (PRAD) near the N-terminus, which is the site for interacting with the t-peptide of AChE. The PRAD has three and five consecutive proline residues, and it has been shown that synthetic polyproline could replace PRAD in its association with AChET tetramers [8]. In AChET the t-peptide is absolutely required in its association with PRAD [9]. The sequence of the t-peptide is highly conserved throughout vertebrates, with a cysteine at −4 position from the C-terminus and a series of seven aromatic residues, including four equally spaced tryptophans. Because the t-peptide constitutes an autonomous interacting domain, it has been named “the tryptophan amphiphilic tetramerization” (WAT) domain. In this notation, AChET is equivalent to AChE + WAT [10].
Recently the crystal structure of PRAD/WAT complex was solved at 2.35 Å resolution [11]. The complex has the expected [WAT]4PRAD stoichiometry. Four parallel α-helical WAT chains wrap around a single antiparallel PRAD helix, which itself has a left-handed polyproline II conformation. Each WAT helix assumes a coiled-coil conformation, and all four of them form a left-handed supercoil around the PRAD (Figure 1A). The WWW motif in the WAT makes repetitive hydrophobic stacking and hydrogen bond interactions with the PRAD. The four WAT chains are related by a 4-fold screw axis around the PRAD. The strength of PRAD–WAT interaction is very tight, with no monomer of WAT detected in the range of 10−10 to 10−12 M [8].
Figure 1 The Structures Related to AChE Tetramerization
(A) The [WAT]4PRAD complex structure. (B) The compact tetramer structure. (C) The loose tetramer structure. (D) The [AChET]4–ColQ complex model constructed according to the [WAT]4PRAD complex structure. Each chain is colored differently (A, blue; B, red; C, gray; D, orange; ColQ, yellow). The catalytic S203 was shown as a green ball model for each AChE subunit, and the cyan surfaces are residues near the peripheral site.
It remains unknown how the four AChE subunits are arranged in the tetramer associated with the PRAD. Low-resolution crystallographic studies revealed two distinct three-dimensional structures of AChE tetramer [12]. Both crystal structures show a dimer of dimers, i.e., there is no 4-fold symmetry to relate all four subunits. In one structure two AChE dimers are close with all four C-terminal sequences aligning to the same direction (referred to as the compact tetramer; Figure 1B), and in the other structure the space between the two dimers is large and the four C-terminal sequences are aligned antiparallel to the middle (referred to as the loose tetramer; Figure 1C). The crystals were grown from trypsin-digested, collagen-tailed AChE and should both have WAT and PRAD preserved; although electronic densities were seen, it was not possible to resolve them. It was suggested that the flexibility of AChE tetramers might be related to the regulation of catalysis [2,12]. To test this, reaction-rate calculations were conducted using these two tetramer structures and a morphed intermediate structure. The results showed that the rate per active site was reduced due to active site occlusion and sink–sink competition compared to the monomeric form, but could be partly compensated by electrostatic enhancements in the tetramers [13]. The rate reduction due to active site occlusion was particularly notable for the compact tetramer [13].
Efforts to combine the PRAD/WAT structure and the two AChE tetramer structures were not very successful. The problem was that in both tetramer structures the four AChE subunits lacked 4-fold symmetry as seen in the [WAT]4PRAD complex crystal structure [11]. Since the four WAT chains are staggered in the structure, it is impossible, without substantial distortion of the [WAT]4PRAD complex and/or the AChE tetramer structures, to dock the [WAT]4PRAD complex along the 2-fold axis between the pairs of AChE dimers. In fact, the superhelical axis of the [WAT]4PRAD complex structure is at an angle of about 30° to this 2-fold axis [11].
Considering the tight association between PRAD and WAT, the weak affinity of the AChE dimer, and the limited contact between the two dimers in the tetramer, it is reasonable to assume that the PRAD–WAT interaction dominates over the inter-subunit interaction in the association of AChE tetramer with ColQ. In fact, AChE only exists as a soluble monomer if the WAT sequence is deleted [9], indicating that the dimerization forces are weak. In a previous study of subunit association in cholinesterases, this interaction was identified as the weak hydrophobic interaction, compared with the strong interaction seen between WAT and PRAD [14]. It has also been shown that if two proteins, for example, AChE and alkaline phosphatase—both with WAT sequences at their C-termini—are mixed with PRAD at various ratios, tetramers containing three subunits of one protein and one of the other are underrepresented [10]. This points to a symmetric [AChET]4–ColQ complex form in physiological conditions. Since the two crystal structures of AChE tetramer were generated experimentally, they have to be accessible from the physiological model through conformational changes.
In this study, we built an [AChET]4–ColQ complex model strictly following the PRAD–WAT interaction. Although molecular dynamics (MD) is a useful tool to observe dynamics in conformational changes, the size of the system (more than 2,300 residues) poses too much a challenge in this case [15]. Block normal mode analysis (BNMA) uses coarse-grained representation for the protein to reduce the computational cost and has been shown to be a useful tool in predicting low-frequency motions seen in large protein assemblies, for example, the swelling of a virus capsid [16–18], the ratchet motion of the ribosome [19], the collective motions in HIV-1 transcriptase [20], myosin [21,22], ATPase [22,23], and chaperonin GroEL [24], etc. These low-frequency motions are often related to the conformational changes required by the biological functions of the protein assemblies.
Here we applied BNMA to the [AChET]4–ColQ complex model and calculated the 100 lowest normal modes. By projecting the conformational changes onto these normal modes, it was found that the two AChE tetramer crystal structures could be rationalized by using these low-frequency normal modes.
Results/Discussion
Our [AChET]4–ColQ complex model has a quasi-4-fold axis as shown in Figure 1D. The bending angle between the WAT helix and the C-terminal helix of AChE is approximately 45°. Structural relaxation changed the model by 0.95 Å in root mean square deviation (RMSD) of the backbone atoms calculated by the ptraj module of Amber8. If viewed from the presynaptic side (i.e., the WAT sequence on the top), each gorge to the active site is located at the clockwise side of AChE and orients almost exactly parallel to the tetramer plane. In this model each AChE subunit has its own quarter of space for attracting substrate, and no active site is blocked by adjacent subunits. Presumably the sink–sink competition is less in such an arrangement, because the average sink–sink distance is larger than that in the two crystal structures of the tetramer. These two factors can both increase the diffusion-controlled reaction rate for the [AChET]4–ColQ complex. Reaction-rate calculations are currently underway using this model.
A simplified BNMA was done to calculate the low-frequency modes, using the [AChET]4–ColQ structure model. The first 100 lowest frequency normal modes were obtained. These normal modes are sufficient to capture all the collective motions, and motions with higher frequency are usually localized to a small domain [21,25]. Figure 2 shows the root mean square fluctuations (RMSFs) derived from Equation 1. It appears that the largest structural fluctuations are located at the WAT and PRAD sequences. This is consistent with the fact that these sequences in the two crystal structures (1C2O and 1C2B) of the AChE tetramer were disordered. As can be seen in the correlation map of motions discussed below, the motion experienced by this region is rigid-body in nature. Therefore it does not contradict our hypothesis that WAT and PRAD have very high affinity. Other regions showing high fluctuation are residues 380 to 390 and residues 265 to 276. They are both loops connecting α helices. Residues 380 and 390 are located at the interface between the catalytic domain of AChE and ColQ; therefore the large fluctuation indicates that there is no strong association between them at this region.
Figure 2 The RMSF of the [AChET]4–ColQ Complex as Calculated from the 100 Lowest Frequency Modes
All residues are numbered continuously (chain A: 1–583, B: 584-1166, C: 1167–1749, D: 1750–2332, ColQ: 2333–2379).
The correlation map can identify collective motions, which are often important large-scale motions related to the protein's biological function. Due to the large amount of data, only one AChET subunit and the ColQ are presented in the motion correlation map in Figure 3. The WAT domain, which corresponds to residues 544 to 583 in Figure 3, has little or no correlation with the catalytic domain of AChE. Instead, it moves together with ColQ (residues 584 to 630), as implied by the high correlation between the WAT and ColQ. This further demonstrates that the interaction between AChE and ColQ is weak, and the interaction between WAT and ColQ is strong. An interesting region in the correlation map can be found for residues 355 to 410. These residues form an α helix bundle themselves and are isolated in the correlation map from the rest of the catalytic domain of AChE. Therefore it appears that these residues form a subdomain. Structurally, Pro410 breaks this subdomain from a long helix, and Ser355 is connected to a flexible loop. They can be considered as hinges connecting different domains. We note that Pro410 is almost universally conserved in all cholinesterases. In our [AChET]4–ColQ complex model, this subdomain makes contact with the C-terminal extension of ColQ, as is evident from the correlation map. In addition, this subdomain participates in the inter-subunit interface with the clockwise neighboring subunit.
Figure 3 The Motion Correlation Map of the [AChET]4–ColQ Complex as Predicted by BNMA
Only one AChE subunit and ColQ were plotted here (AChE: 1–583, ColQ: 584–630).
Conformational changes in proteins can be considered as linear combinations of displacements along low-frequency normal modes. Since the two crystal structures of AChE tetramer were real snapshots of the [AChET]4–ColQ complex, they represent two conformational states with large conformational changes to the complex structure. By projecting the conformational changes onto the eigenvector of each normal mode, it is possible to identify the degree to which each mode is involved in the conformational change [22].
Figure 4 shows the results for the involvement analysis of these conformational changes. In the case of the compact tetramer, the lowest frequency mode (except the trivial modes corresponding to translation and rotation) has the highest coefficient at 0.53 as seen in Figure 4A. Such a large involvement coefficient indicates that this mode is highly relevant to the conformational change from the symmetric model to the compact tetramer structure. Figure 5A illustrates the collective motions in this mode by plotting the displacement onto each residue. The frequency of this mode is 1.39 cm−1. It seems that in this mode the subunits move closer to each other, as the inter-subunit distance is shorter in the compact tetramer structure compared to the symmetric model of [AChET]4–ColQ complex. The cumulative involvement plot in Figure 4C shows that the first 30 lowest frequency modes can account for 80% of the motions in the conformational change to the compact tetramer structure.
Figure 4 The Involvement Analysis of the Low-Frequency Modes of the [AChET]4–ColQ Complex
The involvement of the modes was shown for the conformational change to (A) the compact and (B) loose tetramer structures. The cumulative involvement of the modes was shown for (C) the compact and (D) loose tetramer structures.
Figure 5 Illustration of Motions in the Two Lowest Frequency Modes from BNMA
The arrow represents the amplitude and direction of the displacement experienced by each residue in the conformational change. Frequencies of the two modes are (A) 1.49 cm−1 and (B) 1.66 cm−1.
In the case of the loose tetramer structure, the most involved modes are the second and fourth lowest frequency modes with involvement coefficients of 0.35 and 0.44, respectively. The vibrational frequencies of these two modes are 1.66 cm−1 and 1.89 cm−1, respectively. Figure 5B illustrates the second lowest frequency mode by plotting displacements onto each residue. In this mode, one AChE subunit moves away from others, thus increasing the inter-subunit distance. In the fourth lowest frequency mode, one subunit rotates relative to the others (not shown in Figure 5). This motion can change the relative orientation of the gorges, as the gorge openings are in an angle in the crystal structures, while they are all in a plane in our model. In fact, all first seven lowest frequency modes are collective inter-subunit motions involving the four AChE catalytic domains, with little or no involvement from the WAT domain or ColQ. The first 30 modes can account for 75% of the motion in the conformational change in the loose tetramer as seen in Figure 4D.
All the seven inter-subunit motions involve a flexible hinge in residues 540 to 545 (D544 is the first residue in the WAT sequence), with T543 at the center of the flexible hinge. These residues have no correlation in motion with either AChE or the WAT/PRAD complex as seen in Figure 3. Visualization of each individual normal mode confirmed these hinge residues. This is in excellent agreement with secondary structure prediction that these residues form a loop and with the experimental finding that cysteins introduced in this region can efficiently form homomeric disufide bonds in dimers [26]. The conformational flexibility of these hinge residues allows WAT and AChE to each maintain their secondary structures. If these residues were deleted or made more rigid, it would affect the tetrameric association of AChE with PRAD as the secondary structure of either AChE or WAT has to be broken at the joint.
Using a simplified linear combination of these 100 normal modes, we generated two series of transient models of the [AChET]4–ColQ complex that minimize the RMSD to the two crystal structures. The RMSD between the models of the [AChET]4–ColQ complex and the compact tetramer structure decreased from 23.9 Å to 15.7 Å, while for the loose tetramer structure it went from 28.9 Å to 17.6 Å. The large RMSD values are due to the large translational and rotational movement required for each AChE subunit to realign between the [AChET]4–ColQ complex model and the crystal structures. As a comparison, the RMSD for the two crystal structures is 34.9 Å, i.e., even larger than the initial RMSD between the [AChET]4–ColQ complex model and the two crystal structures. When the two final models of the [AChET]4–ColQ complex from projection were examined, it was found that they were significantly distorted in respect to the monomeric AChE. However, both models show a similar dimer of dimers as seen in the two crystal structures. The orientations of the active site gorges were originally all parallel to the tetramer plane. After applying projected motions from BNMA, two gorges were pointing downward and the other two slightly upward. It should be noted that this is a rather approximate demonstration using a simplified linear combination of normal mode motions. The application of normal modes for proteins is strictly only valid for very small conformational motions.
Normal mode analysis (NMA) can tell us what intrinsic low-frequency or large-scale motions the molecule or molecular assembly has. Conformational changes can be achieved by activating these modes to cross low-energy barriers by ligand binding or environmental changes. As indicated by the low-frequency inter-subunit normal modes, each AChE subunit in the [AChET]4–ColQ complex may possess significant flexibility due to the presence of a flexible hinge at the interface of AChE and WAT. Under physiological conditions, there is a distribution of many different conformational states, including the two crystal structures and the symmetric tetramer model we constructed based on the [WAT]4PRAD complex structure. Due to the low barrier uncovered by our analyses, these conformational states can easily interconvert. Experimental findings support such a conformational transition. For example, under conditions that the compact AChE tetramer is grown, all four active site gorges can be occupied by fasciculin [12], which is too large to access the two partially blocked active sites in the compact tetramer structure. From fluorescence polarization spectroscopy measurement it was found that discrete segments of AChE tetramer might move independently [27]. The binding of bulky ligands such as fasciculin or antibodies can fix the tetramer in certain conformational states.
The presence of a spectrum of conformations for AChE tetramers further complicates the function of AChE. It has been known that the rapid opening/closing motion of aromatic residues in the peripheral anionic site of AChE functions as a gate for substrate entering the active site [28,29]. Future modeling of AChE function will have to consider these low-frequency large-scale motions as well.
Materials and Methods
[AChET]4–ColQ complex model building.
The crystal structure of the [WAT]4PRAD complex was downloaded from the Protein Data Bank [11]. Chains A–D and I were selected for modeling. In building a complete [AChET]4–ColQ complex model, we added the missing residues 35–40 in WAT and extended PRAD to 16 more residues in the N-terminus and 19 more residues in the C-terminus using SYBL (Tripos, St. Louis, Missouri, United States). The conformations of these residues were copied from the existing chains, i.e., the area with residues 35–40 was the same coiled α helix as the rest of the WAT, and all residues in ColQ were in polyproline II conformation. Mouse sequence of ColQ tail was downloaded from the Swiss-Prot database. While the sequence in the [WAT]4PRAD complex crystal structure was from human, the only differences were D35E in the WAT and T3M in the PRAD from human to mouse. SeMet21 was mutated back to Met.
The next step was to connect WAT to AChE to form AChET. The mouse AChE structure was taken from a preoptimized structure based on the PDB entry (see Accession Numbers section) [29]. Since both the C-terminus of AChE catalytic domain and the N-terminus of the WAT domain have α-helical conformations, we initially connected the two using continuous α-helical conformation. A perfect α-helix sequence was used as a bridge in matching the two ends. However, the AChE had a severe clash with the C-terminal extension of PRAD in ColQ. The phi and psi angles of residues A542, T543, and D544 were manually adjusted to bend the helix to move AChE away from ColQ using known helix-bending structures in PDB. Most of the helix bending structures involve a proline residue in the kink region [30], including P410 and P537 in the mouse AChE structure itself. Here we borrowed the same helical bend from the M2 helix of the nicotinic acetylcholine receptor structure at position L11′(see Accession Numbers section) [31], which does not involve a proline residue. SYBYL version 7.0 (Tripos) was used to copy the main chain conformation of L11′ and the two adjacent residues. By geometrical matching using the 4-fold screw symmetry, we added three other chains to build a complete mouse [AChET]4–ColQ complex model. VMD version 1.82 [32] was used to match structures in a tcl script. The complex structure was further optimized with Amber94 force field [33], using the sander module in Amber8 [34]. The initial structure had some severely bad contacts, which caused the sander minimization module in Amber8 to crash. We found that the old-style velocity quenching in NAMD v2.5 [35] worked better in removing bad clashes. The partially relaxed structure of the [AChET]4–ColQ complex model was further optimized by 5,000 steps of conjugate-gradient energy minimization, using the sander module of Amber8 (first 100 steps were steepest descent search). The solvation effect was mimicked with a distance-dependent dielectric coefficient of 4r. A 20-Å cutoff was used in evaluating nonbond interactions. The final RMS force in energy minimization was 0.01 kcal/mol/Å.
BNMA.
Although MD is useful in observing conformational changes in protein simulations, the size of the system prevents us from running MD for the extended time required for observing large conformational changes [15]. NMA is another useful tool to predict conformational changes from low-frequency modes [25]. Recently NMA has been extended to large biomolecules using the Elastic Network Model or BNMA [16,22,24,36]. Here we use the BNMA method to analyze the low-frequency motions of the [AChET]4–ColQ complex model. The advantage of BNMA is that the real atomic potential can be used and projected to a smaller dimension using blocks of atoms [22]. Each residue was chosen as a block in the current study. Therefore the degrees of freedom of the system can be reduced from 3N to 6M, where N and M are the total number of atoms and residues, respectively. The factor of six in block space comes from the fact that each block has three degrees of freedom from translation and three from rotation. The rotational degrees of freedom within a block have been shown to be less important and can be omitted to further reduce the computational cost [21].
In BNMA, each component of the Hessian matrix, i.e., the second derivative of energy, was calculated and projected to the block space “on the fly” [22]. The rotational degrees of freedom for each block were omitted in this study. In this system, there are 36,836 atoms in 2,379 residues. Using BNMA reduced the total degrees of freedom from 36,836 × 3 = 11,0508 to 2,379 × 3 = 7,137. The block Hessian matrix was diagonalized by using the nmode module of Amber8. Due to the use of cutoff for nonbond interactions, the matrix is sparse and easy to solve. The whole procedure took about 2 h in a Pentium 2.8 GHz Linux machine.
From the BNMA results, RMSF of each residue can be calculated by summing over contributions from all the normal modes [37]:
where kB is the Boltzmann constant, T is the temperature, mi is the mass for residue i, aij is the i-th component in the eigenvector corresponding to mode j, and ωj is the angular frequency for mode j. The summation is over all the nontrivial normal modes.
The correlation map defines the correlation of motions of each residue pair and can be computed as follows [37]:
and
A conformational change can be considered as the linear combination of motions from all the normal modes. Given a conformational change, the involvement coefficient of mode k can be calculated as follows [22],
where vk is the eigenvector corresponding to mode k, and ΔX is the conformational change vector, which can be computed as the difference in coordinates of two different conformations. A large involvement coefficient indicates that the motion along this normal mode is highly relevant to the conformational change being examined. The cumulative involvement for the first n normal modes can be obtained by [22]
Obviously, if the summation is over all the modes (3N), the cumulative involvement is expected to be one.
Supporting Information
Accession Numbers
The accession numbers for the proteins described in this paper can be found in the SwissProt database (http://www.ebi.ac.uk/swissprot): AChE (P21836), ColQ (O35348), and PRiMA (Q810F0); and in the Protein Data Bank (http://www.rcsb.org/pdb/): PRAD/WAT complex (1VZJ), compact AChE tetramer structure (1C2O), loose AChE tetramer structure (1C2B), [WAT]4PRAD complex (1VZJ), mouse AChE (1MAH), and acetylcholine receptor structure (1OED).
The authors thank Dr. Michal Harel for helpful discussions and Dr. Xiaolin Cheng for technical assistance. This work was in part supported by the National Institutes of Health, the National Science Foundation (NSF), the Howard Hughes Medical Institute, the NSF Center for Theoretical and Biological Physics, the National Biomedical Computing Resource, the W. M. Keck Foundation, and Accelrys.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. DZ and JAM conceived and designed the experiments. DZ performed the experiments. DZ analyzed the data and contributed reagents/materials/analysis tools. DZ and JAM wrote the paper.
A previous version of this article appeared as an Early Online Release on October 19, 2005 (DOI: 10.1371/journal.pcbi.0010062.eor).
Abbreviations
AChEacetylcholinesterase
AChETacetylcholinesterase T-subtype
BNMAblock normal mode analysis
ColQcollagen-like Q subunit
MDmolecular dynamics
NMAnormal mode analysis
PRADproline-rich attachment domain
PRiMAproline-rich membrane-anchoring protein
RMSDroot mean square deviation
RMSFroot mean square fluctuation
WATtryptophan amphiphilic tetramerization
==== Refs
References
Massoulie J Pezzementi L Bon S Krejci E Vallette FM 1993 Molecular and cellular biology of cholinesterases Prog Neurobiol 41 31 91 8321908
Taylor P Radic Z 1994 The cholinesterases: From genes to proteins Annu Rev Pharmacol Toxicol 34 281 320 8042853
Massoulie J 2002 The origin of the molecular diversity and functional anchoring of cholinesterases Neurosignals 11 130 143 12138250
Legay C Huchet M Massoulie J Changeux JP 1995 Developmental regulation of acetylcholinesterase transcripts in the mouse diaphragm—Alternative splicing and focalization Eur J Neurosci 7 1803 1809 7582132
Krejci E Coussen F Duval N Chatel JM Legay C 1991 Primary structure of a collagenic tail peptide of torpedo acetylcholinesterase—Coexpression with catalytic subunit induces the production of collagen-tailed forms in transfected cells EMBO J 10 1285 1293 1840520
Perrier AL Massoulie J Krejci E 2002 PRiMA: The membrane anchor of acetylcholinesterase in the brain Neuron 33 275 285 11804574
Krejci E Thomine S Boschetti N Legay C Sketelj J 1997 The mammalian gene of acetylcholinesterase-associated collagen J Biol Chem 272 22840 22847 9278446
Bon S Coussen F Massoulie J 1997 Quaternary associations of acetylcholinesterase .2: The polyproline attachment domain of the collagen tail J Biol Chem 272 3016 3021 9006950
Duval N Massoulie J Bon S 1992 H and T subunits of acetylcholinesterase from Torpedo, expressed in COS cells, generate all types of globular forms J Cell Biol 118 641 653 1639848
Simon S Krejci E Massoulie J 1998 A four-to-one association between peptide motifs: Four C-terminal domains from cholinesterase assemble with one proline-rich attachment domain (PRAD) in the secretory pathway EMBO J 17 6178 6187 9799227
Dvir H Harel M Bon S Liu W-Q Vidal M 2004 The synaptic acetylcholinesterase tetramer assembles around a polyproline II helix EMBO J 23 4394 4405 15526038
Bourne Y Grassi J Bougis PE Marchot P 1999 Conformational flexibility of the acetylcholinesterase tetramer suggested by X-ray crystallography J Biol Chem 274 30370 30376 10521413
Zhang DQ Suen J Zhang YJ Song YH Radic Z 2005 Tetrameric mouse acetylcholinesterase: Continuum diffusion rate calculations by solving the steady-state Smoluchowski equation using finite element methods Biophys J 88 1659 1665 15626705
Giles K 1997 Interactions underlying subunit association in cholinesterases Protein Eng 10 677 685 9278281
Elcock AH 2002 Modeling supramolecular assemblages Curr Opin Struct Biol 12 154 160 11959491
Tama F Brooks CL 2002 The mechanism and pathway of pH induced swelling in cowpea chlorotic mottle virus J Mol Biol 318 733 747 12054819
Tama F Brooks CL 2005 Diversity and identity of the mechanical properties of icosahedral viral capsids studied with elastic network normal mode analysis Biophys J 88 231a 232a
Tama F Brooks CL 2005 Diversity and identity of mechanical properties of icosahedral viral capsids studied with elastic network normal mode analysis J Mol Biol 345 299 314 15571723
Tama F Valle M Frank J Brooks CL 2003 Dynamic reorganization of the functionally active ribosome explored by normal mode analysis and cryo-electron microscopy Proc Natl Acad Sci U S A 100 9319 9323 12878726
Bahar I Erman B Jernigan RL Atilgan AR Covell DG 1999 Collective motions in HIV-1 reverse transcriptase: Examination of flexibility and enzyme function J Mol Biol 285 1023 1037 9887265
Li GH Cui Q 2002 A coarse-grained normal mode approach for macromolecules: An efficient implementation and application to Ca2+-ATPase Biophys J 83 2457 2474 12414680
Li GH Cui Q 2004 Analysis of functional motions in Brownian molecular machines with an efficient block normal mode approach: Myosin-II and Ca2+-ATPase Biophys J 86 743 763 14747312
Cui Q Li GH Ma JP Karplus M 2004 A normal mode analysis of structural plasticity in the biomolecular motor F-1-ATPase J Mol Biol 340 345 372 15201057
Ma JP Karplus M 1998 The allosteric mechanism of the chaperonin GroEL: A dynamic analysis Proc Natl Acad Sci U S A 95 8502 8507 9671707
Case DA 1994 Normal-mode analysis of protein dynamics Curr Opin Struct Biol 4 285 290
Bon S Dufourcq J Leroy J Cornut I Massoulie J 2004 The C-terminal t peptide of acetylcholinesterase forms an alpha helix that supports homomeric and heteromeric interactions Eur J Biochem 271 33 47 14686917
Berman HA Yguerabide J Taylor P 1985 Flexibility of the molecular forms of acetylcholinesterase measured with steady-state and time-correlated fluorescence polarization spectroscopy Biochemistry 24 7140 7147 4084569
Zhou HX Wlodek ST McCammon JA 1998 Conformation gating as a mechanism for enzyme specificity Proc Natl Acad Sci U S A 95 9280 9283 9689071
Tara S Elcock AH Kirchhoff PD Briggs JM Radic Z 1998 Rapid binding of a cationic active site inhibitor to wild type and mutant mouse acetylcholinesterase: Brownian dynamics simulation including diffusion in the active site gorge Biopolymers 46 465 474 9838872
Sansom MSP Weinstein H 2000 Hinges, swivels and switches: The role of prolines in signaling via transmembrane alpha-helices TiPS 21 445 451 11121576
Miyazawa A Fujiyoshi Y Unwin N 2003 Structure and gating mechanism of the acetylcholine receptor pore Nature 423 949 955 12827192
Humphrey W Dalke A Schulten K 1996 VMD: Visual molecular dynamics J Mol Graph 14 33 38 8744570
Cornell WD Cieplak P Bayly CI Gould IR Merz KM 1995 A 2nd generation force-field for the simulation of proteins, nucleic-acids, and organic-molecules J Am Chem Soc 117 5179 5197
Case DA Darden TA Cheatham TEI Simmerling CL Wang JM 2004 Amber8 [computer program] San Francisco Department of Pharmaceutical Chemistry, University of California
Kale L Skeel R Bhandarkar M Brunner R Gursoy A 1999 NAMD2: Greater scalability for parallel molecular dynamics J Comput Phys 151 283 312
Marques O Sanejouand Y 1995 Hinge-bending motion in citrate synthase arising from normal mode calculations Proteins 23 557 560 8749851
Tama F Sanejouand YH 2001 Conformational change of proteins arising from normal mode calculations Protein Eng 14 1 6 11287673
|
16299589
|
PMC1285061
|
CC BY
|
2021-01-05 09:20:35
|
no
|
PLoS Comput Biol. 2005 Nov 18; 1(6):e62
|
utf-8
|
PLoS Comput Biol
| 2,005 |
10.1371/journal.pcbi.0010062
|
oa_comm
|
==== Front
PLoS Comput BiolPLoS Comput. BiolpcbiplcbploscompPLoS Computational Biology1553-734X1553-7358Public Library of Science San Francisco, USA 1629959010.1371/journal.pcbi.001006305-PLCB-RA-0160R3plcb-01-06-03Research ArticleBioinformatics - Computational BiologyEvolutionImmunologyMicrobiologyEukaryotesEubacteriaVertebratesArthropodsCnidaria (JellyfishHydraCoralsEtc.)Stealth Proteins: In Silico Identification of a Novel Protein Family Rendering Bacterial Pathogens Invisible to Host Immune Defense In Silico Identification of Novel Protein FamilySperisen Peter 1Schmid Christoph D 1Bucher Philipp 12*Zilian Olav 2¤1 Swiss Institute of Bioinformatics, Epalinges, Switzerland
2 Swiss Institute for Experimental Cancer Research, Epalinges, Switzerland
Bork Peer EditorEMBL Heidelberg, Germany* To whom correspondence should be addressed. E-mail: [email protected]¤ Current address: Helvea, Geneva, Switzerland
11 2005 18 11 2005 1 6 e6311 7 2005 20 10 2005 Copyright: © 2005 Sperisen 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.There are a variety of bacterial defense strategies to survive in a hostile environment. Generation of extracellular polysaccharides has proved to be a simple but effective strategy against the host's innate immune system. A comparative genomics approach led us to identify a new protein family termed Stealth, most likely involved in the synthesis of extracellular polysaccharides. This protein family is characterized by a series of domains conserved across phylogeny from bacteria to eukaryotes. In bacteria, Stealth (previously characterized as SacB, XcbA, or WefC) is encoded by subsets of strains mainly colonizing multicellular organisms, with evidence for a protective effect against the host innate immune defense. More specifically, integrating all the available information about Stealth proteins in bacteria, we propose that Stealth is a D-hexose-1-phosphoryl transferase involved in the synthesis of polysaccharides. In the animal kingdom, Stealth is strongly conserved across evolution from social amoebas to simple and complex multicellular organisms, such as Dictyostelium discoideum, hydra, and human. Based on the occurrence of Stealth in most Eukaryotes and a subset of Prokaryotes together with its potential role in extracellular polysaccharide synthesis, we propose that metazoan Stealth functions to regulate the innate immune system. Moreover, there is good reason to speculate that the acquisition and spread of Stealth could be responsible for future epidemic outbreaks of infectious diseases caused by a large variety of eubacterial pathogens. Our in silico identification of a homologous protein in the human host will help to elucidate the causes of Stealth-dependent virulence. At a more basic level, the characterization of the molecular and cellular function of Stealth proteins may shed light on fundamental mechanisms of innate immune defense against microbial invasion.
Synopsis
The immune system is a complex and highly developed system of specialized cells and organs that protects an organism against bacterial, parasitic, fungal, and viral infections. Broadly speaking, the different types of immune responses subdivide the immune system into two categories: innate (or nonadaptive) and adaptive immune system. The innate immune system serves as a first line of defense but lacks the ability to recognize certain pathogens and to provide the specific protective immunity that prevents reinfection. Just as metazoans have developed many different defenses against pathogens, so have pathogens evolved elaborate strategies to evade these defenses. Based on a comparative genomics approach and data mining, the authors have discovered a new family of proteins with a striking phylogenetic distribution, occurring in most eukaryotes and in subsets of mostly pathogenic or commensal prokaryotes. While the precise functions of these proteins remain unknown, prokaryotic versions have been implicated in the synthesis of extracellular polysaccharides known to be potent regulators of the innate immune system. This previously unrecognized link hints towards a potentially novel regulatory mechanism of the innate immune system. It remains to be shown if drugs selectively inhibiting Stealth in pathogens will help fight Stealth-mediated infections.
Citation:Sperisen P, Schmid CD, Bucher P, Zilian O (2005) Stealth proteins: In silico identification of a novel protein family rendering bacterial pathogens invisible to host immune defense. PLoS Comput Biol 1(6): e63.
==== Body
Introduction
Colonization of hosts by microorganisms is a complex process that determines if the microorganism will coexist with the host as commensal, become an invasive pathogen, or be efficiently eliminated by the host's immune defense [1,2]. Consequently, microorganisms have developed a variety of measures to cope with the increasingly sophisticated defense strategies of the host's immune system [3–7]. Amongst them, the generation of an extracellular coat made of polysaccharides has proved to be a simple but effective strategy. Bacterial surface polysaccharides can be either amorphous exopolysaccharides, anchored in the lipid layer (lipopolysaccharides, another known regulator of the immune system), or organized as a capsule (capsule polysaccharides [CPSs]). The latter have been shown to mediate adherence to cells and, more importantly, protection against the host's innate immune system [8–11].
Different strategies to escape host immune surveillance have evolved through vertical evolution but also through horizontal gene transfer [12–15]. Though a subject of long-standing controversy, there is increasing evidence suggesting that horizontal gene transfer also occurs from eukaryotes to prokaryotes [16]. Even though the recombined bacteria seemed to have preferentially retained individual domains of proteins [16], a first example was recently reported in which certain bacterial strains kept an entire open reading frame [17].
Here we describe a novel protein family named “Stealth.” Based on a comparative genomics approach, we propose a biological function and an evolutionary scenario for this new protein family.
Results/Discussion
Identification of Stealth
In a screen of the human genome for Notch-related proteins, a novel protein containing two copies of Lin-12/Notch repeats was identified. The protein also showed strong sequence similarity to a number of animal and bacterial proteins, including several virulence factors of human pathogens published under different names. This previously unknown protein family was named “Stealth” because experimentally characterized members of this family appear to render bacterial and protozoan invaders invisible to the host's immune surveillance system.
Stealth proteins are characterized by four conserved regions (CRs) referred to as CR1 to CR4 (Figure 1). The N-terminal CR1 consists of a short but strongly conserved sequence motif, IDVVYTF or very similar. The second region, CR2, is approximately 100 residues long and constitutes the most conserved part of this protein family. A standard BLAST search [18] with any CR2 domain identifies all other members of the Stealth family in the current database with highly significant E-values. CR3 is about 50 residues long but less well conserved. Finally, the C-terminal CR4 includes an almost universally conserved tetrapetide, CLND or CIND. Adjacent and between these domains are divergent sequence regions of variable length that may contain additional domains (Figures 1 and 2A).
Figure 1 Multiple Alignments of CRs
Multiple alignments of the four CRs for a representative set of protein sequences (>15% dissimilarity over all four CRs) are shown. Sequences are identified by a species code (see Table 1), protein name (from literature as proposed in this paper), and database accession number, where available. The lengths of the sequences omitted between or within CRs are indicated in square brackets. The last row shows the secondary structure prediction obtained by jnetpred [65] for the human Stealth protein, where H stands for helices and E for beta-sheets. The color scheme used is the ClustalX default scheme, with the colors for conserved amino acids being more intense than those for nonconserved ones.
Figure 2 Domain Architecture and Genome Structure
(A) CR1 to CR4, found through multiple alignments, are represented by rectangles ranging from light blue (CR1) to dark blue (CR4). Other motifs are represented as follows: predicted signal peptides as magenta rectangles, transmembrane regions as orange rectangles, Lin-12/Notch repeats as red pentagons, and EF-hands as green circles.
(B) The genome structure of the human and fly Stealth homologs is represented, with the exons depicted as green rectangles separated by introns of indicated size.
(C) Two splice variants lead to different N-terminal sequences, as supported by mouse EST sequences. Splicing reconstructs a codon for tyrosine (Y). Both proteins contain a predicted signal peptide.
Table 1 Summary of All Species Containing Stealth Proteins
Table 1 Continued
Taxonomic Distribution
Stealth proteins are found encoded in the genomes of chordates, echinodermates, hydras, fungi, and flies but appear to be absent from nematodes and plants. Interestingly, a few organisms contain multiple Stealth genes (Table 1). Stealth proteins also occur in the protist genomes of Dictyostelium, Giardia, Leishmania, Entamoeba, and Phytophthora, and among the hitherto sequenced bacteria, they are found in the following phyla: alpha-, beta-, and gamma-proteobacteria (mostly pathogens), firmicutes (mostly the commensals), and actinobacteria (some animal pathogens) (Table 1; Figure S1). It is noteworthy that the large majority of completely sequenced bacterial genomes do not harbor Stealth. The species that do contain a member of this family are not necessarily closely related, and include Gram-positive as well as Gram-negative bacteria.
Stealth in Bacteria
Several of the documented bacterial Stealth genes belong to capsule group II biosynthesis operons generating carbohydrate-phosphodiester-containing CPSs [19–24]. In the case of Stealth-expressing bacteria, these CPSs turned out to inhibit complement-mediated lysis, as shown for serogroup A and X of Neisseria meningitidis [23,24] and to correlate with serum and phagocyte survival abilities as shown for Aeromonas hydrophila [25].
The majority of Stealth-expressing bacteria that have been analyzed so far for the composition of their exopolysaccharides turned out to build phosphoglycans consisting of phosphodiester-linked hexose mono- or disaccharide building blocks [26–29]. On the other hand, certain bacteria living in a biofilm community contain CPSs consisting of phosphodiester-linked hexa- or heptasaccharide repeating units [30,31]. These carbohydrates, also called receptor polysaccharides, are synthesized by a series of different glycosyltransferases, with Stealth amongst them [22]. Strains encoding Stealth carry a hexose phosphodiester linker [31] in their receptor polysaccharides, whereas strains lacking Stealth build receptor polysaccharides with a pentose phosphodiester linker.
Definite proof for an essential function of Stealth in CPS biosynthesis was shown in N. meningitidis serogroup A by selective deletion of the gene sacB (i.e., Stealth), giving rise to virtually unencapsulated mutants [23], and by deletion of part of the gene xcbA (i.e., Stealth), together with flanking open reading frames in a serogroup X strain, which resulted in complement-sensitive mutants [24]. Moreover, when the gene cps1A (i.e., Stealth) was deleted in Actinobacillus pleuropneumoniae, the resulting strains lost their pathogenicity in pigs [20].
Taken together, all of the above data suggest that Stealth is a D-hexose-1-phosphoryl transferase that generates interglycosidic phosphate diester linkages.
Characteristics of Metazoan Stealth
Unlike the bacterial Stealth proteins, the vertebrate members of this family are not properly represented in current protein databases. We have manually reconstructed the gene and protein sequences for a number of species with the aid of EST sequences and cross-genome comparisons (Table 1). The human gene consists of 21 exons (Figure 2B), and the translated protein sequence is identical to the RefSeq entry NP_077288. The intron–exon structures of genes found in other vertebrates are essentially the same. In the mouse, however, there is a facultative intron near the start codon spliced out predominantly in transcripts from dendritic cells. This alternative splicing leads to two protein variants with different N-termini (Figure 2C). The hypothetical Drosophila melanogaster and D. yakuba Stealth genes, however, have a completely different intron–exon structure (Figure 2B). Finally, pieces of Stealth-encoding sequences were also found in the preliminary genomes or ESTs of other mammals (Table 1).
Metazoan Stealth proteins are characterized by additional domains. There is a predicted signal peptide and, near the C-terminus, a transmembrane helix. One or two Notch/Lin-12 repeats [32] are inserted between CR2 and CR3, and an EF-hand domain [33] appears between CR3 and CR4. So far, all reconstructed Stealth proteins contain these domains, and in some of the cases where only pieces of sequences are available one can identify these motifs. The strong conservation of the Stealth domain architecture suggests that this protein plays an essential role.
No experimental knowledge is available about the function of metazoan Stealth proteins today (note, however, that Stealth-deficient mice have been generated by O. Z. and coworkers and will be made available upon request). In view of the high degree of sequence similarity to their bacterial homologs, it is reasonable to speculate that they have a similar molecular function and thus are also implicated in exopolysaccharide synthesis. Public expression profiles derived from SAGE experiments indicate a rather broad tissue distribution. The Stealth-dependent polysaccharides could be host-specific structural surface elements exploited by the immune system for self-recognition. In this case, the Stealth-dependent resistance of human pathogens to complement-mediated lysis and other host defense mechanisms would be a straightforward case of molecular mimicry. Alternatively, host-encoded Stealth proteins may play an active role in down-regulating the immune response. The presence of Stealth in both insects and urochordates further suggests that this protein interferes with processes related to innate rather than adaptive immunity [34,35].
Stealth and Protists
Although higher eukaryotes haven't yet been investigated for the presence of phosphoglycan structures similar to the CPSs, such structures have been identified in D. discoideum and in Leishmania species. In D. discoideum such polysaccharides were found on lysosomal cysteine proteinases and spore coat proteins [36,37]. The lysosomal enzymes of D. discoideum have two types of carbohydrate modifications [38,39] found in two separate sets of lysosomal vesicles [40,41]. The major component of Leishmania lipophosphoglycan is a heteropolymer of 10–40 phosphodiester-linked disaccharide units, depending on species and developmental stage [42]. Lipophosphoglycan is predominantly expressed by promastigotes, is essential for intracellular survival in macrophages and for the virulence of Leishmania major and L. donovani, and disappears when the pathogen intracellularly differentiates into amastigotes within host phagolysosomes [43–47]. The genes encoding these hexose-phosphoryl transferases have been identified neither in D. discoideum nor in Leishmania. Given, however, Stealth's presumed enzymatic activity and its comparative biochemical characterization from three different Leishmania species using synthetic acceptor substrate analogs [48], the two Stealth proteins found in Leishmania and those found in D. discoideum are good candidates for this function.
Evolution of Stealth
The peculiar taxonomic distribution of Stealth (Figure 3) could be the outcome of two different evolutionary scenarios: (i) differential loss of an ancient protein already present in an ancestral form of life, or (ii) horizontal gene transfer between eukaryotes and eubacteria. The second hypothesis appears to be the more plausible, but the direction of the transfer is more difficult to assess. Overall, the protein tree largely follows species phylogeny, at least with regard to the higher level taxonomic groups. This indicates that transfer between eukaryotes and prokaryotes must have been an ancient event. However, several observations suggest that Stealth proteins continue to be horizontally transferred within and between certain bacterial groups. In Gram-negative bacteria, Stealth is inserted into group II capsule operons, which exhibit strong sequence similarity across many species, thus facilitating horizontal gene transfer via homologous recombination [49,50]. Moreover, certain Stealth genes have significantly lower G+C content than the remaining part of the genome [19,21,24,51], which is indicative of a recent acquisition from another species, and some of these genes are flanked by recombination-promoting IS insertion elements or residual fragments thereof [21,24].
Figure 3 Phylogenetic Tree
Trees were calculated from amino acid sequence alignments of the four CRs. As in Figure 1, sequences are identified by a species code (see Table 1), protein name (from literature as proposed in this paper), and database accession number, and are color-coded. Dissimilarities are represented by the length of the branches (all with posterior probabilities above 0.95).
Materials and Methods
Sequence analysis.
Multiple amino acid sequence alignments of the four CRs were generated using T-Coffee [52]. The signal peptides were predicted with SignalP v2.0 using the combined NN/HMM-based method [53,54], the transmembrane predictions were made using TMHMM v2.0 [55,56], and the Lin-12/Notch repeats were identified using the profile PS50258 in PROSITE [57]. The EF-hand domains were detected using the Pfam HMM PF00036 [58].
The human and the fly gene structures were constructed with the aid of the trome database [59–61].
Sequence database searches.
Other members of the Stealth protein family were identified by searching with either the human or the Streptomyces coelicolor CR2 using BLAST [18] on either nucleic acid or protein databases.
Calculation of sequence trees.
For each CR a separate multiple amino acid sequence alignment was generated. These multiple alignments were concatenated, resulting in a multiple alignment that represents the four CRs. CRs that are absent in certain species are represented as gaps in the multiple alignment. Processed alignments were used to derive tree topologies using Bayesian inference of phylogeny as implemented by MrBayes v3.0 [62,63]. MrBayes was used with four heated chains over 200,000 generations, sampling every 20 trees. The likelihoods of these trees were examined to estimate the length of the burn-in phase, and all trees sampled 20,000 generations later than this point were used to create a consensus tree using the 50% majority rule. MrBayes was used with the mixed model of amino acid substitution, assuming the presence of invariant sites and using a gamma distribution approximated by four different rate categories to model rate variation between sites, estimating amino acid frequencies from the alignment. The consensus tree was displayed using DRAWGRAM of the PHYLIP package [64].
Supporting Information
Figure S1 Taxonomic Distribution of Stealth in Bacteria
(57 KB DOC)
Click here for additional data file.
Part of this work has been supported by grant SKL 1125–02–2001 from the Swiss Cancer League (to OZ). We thank Denis-Luc Ardiet for stimulating discussions and prompting us to kreisler.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. PS, CDS, PB, and OZ conceived and designed the experiments. PS and CDS performed the experiments. PS, CDS, PB, and OZ analyzed the data and wrote the paper.
Abbreviations
CPScapsule polysaccharide
CRconserved region
==== Refs
References
Merrell DS Falkow S 2004 Frontal and stealth attack strategies in microbial pathogenesis Nature 430 250 256 15241423
Waterfield NR Wren BW Ffrench-Constant RH 2004 Invertebrates as a source of emerging human pathogens Nat Rev Microbiol 2 833 841 15378047
Irving P Troxler L Hetru C 2004 Is innate enough? The innate immune response in Drosophila
C R Biol 327 557 570 15330255
Brown DR 2002 Mycoplasmosis and immunity of fish and reptiles Front Biosci 7 D1338 D1346 11991833
Kalia A Bessen DE 2004 Natural selection and evolution of streptococcal virulence genes involved in tissue-specific adaptations J Bacteriol 186 110 121 14679231
Claverys JP Prudhomme M Mortier-Barriere I Martin B 2000 Adaptation to the environment: Streptococcus pneumoniae, a paradigm for recombination-mediated genetic plasticity? Mol Microbiol 35 251 259 10652087
Fleckenstein JM Kopecko DJ 2001 Breaching the mucosal barrier by stealth: An emerging pathogenic mechanism for enteroadherent bacterial pathogens J Clin Invest 107 27 30 11134175
Costerton JW Cheng KJ Geesey GG Ladd TI Nickel JC 1987 Bacterial biofilms in nature and disease Annu Rev Microbiol 41 435 464 3318676
Roberts IS Saunders FK Boulnois GJ 1989 Bacterial capsules and interactions with complement and phagocytes Biochem Soc Trans 17 462 464 2666197
Cross AS 1990 The biologic significance of bacterial encapsulation Curr Top Microbiol Immunol 150 87 95 2404691
Moxon ER Kroll JS 1990 The role of bacterial polysaccharide capsules as virulence factors Curr Top Microbiol Immunol 150 65 85 2404690
Lorenz MG Wackernagel W 1994 Bacterial gene transfer by natural genetic transformation in the environment Microbiol Rev 58 563 602 7968924
Hartl DL Dykhuizen DE Berg DE 1984 Accessory DNAs in the bacterial gene pool: Playground for coevolution Ciba Found Symp 102 233 245 6319094
Moreira D 2000 Multiple independent horizontal transfers of informational genes from bacteria to plasmids and phages: Implications for the origin of bacterial replication machinery Mol Microbiol 35 1 5 10632872
Sundstrom L 1998 The potential of integrons and connected programmed rearrangements for mediating horizontal gene transfer APMIS Suppl 84 37 42 9850680
Koonin EV Makarova KS Aravind L 2001 Horizontal gene transfer in prokaryotes: Quantification and classification Annu Rev Microbiol 55 709 742 11544372
Budd A Blandin S Levashina EA Gibson TJ 2004 Bacterial alpha2-macroglobulins: Colonization factors acquired by horizontal gene transfer from the metazoan genome? Genome Biol 5 R38 15186489
Altschul SF Gish W Miller W Myers EW Lipman DJ 1990 Basic local alignment search tool J Mol Biol 215 403 410 2231712
Satola SW Schirmer PL Farley MM 2003 Genetic analysis of the capsule locus of Haemophilus influenzae serotype f Infect Immun 71 7202 7207 14638817
Bandara AB Lawrence ML Veit HP Inzana TJ 2003 Association of Actinobacillus pleuropneumoniae capsular polysaccharide with virulence in pigs Infect Immun 71 3320 3328 12761114
Almiron-Roig E Mulholland F Gasson MJ Griffin AM 2000 The complete cps gene cluster from Streptococcus thermophilus NCFB 2393 involved in the biosynthesis of a new exopolysaccharide Microbiology 146 2793 2802 11065358
Xu DQ Thompson J Cisar JO 2003 Genetic loci for coaggregation receptor polysaccharide biosynthesis in Streptococcus gordonii 38 J Bacteriol 185 5419 5430 12949094
Swartley JS Liu LJ Miller YK Martin LE Edupuganti S 1998 Characterization of the gene cassette required for biosynthesis of the (alpha1→6)-linked N-acetyl-D-mannosamine-1-phosphate capsule of serogroup A Neisseria meningitidis
J Bacteriol 180 1533 1539 9515923
Tzeng YL Noble C Stephens DS 2003 Genetic basis for biosynthesis of the (alpha 1→4)-linked N-acetyl-D-glucosamine 1-phosphate capsule of Neisseria meningitidis serogroup X Infect Immun 71 6712 6720 14638755
Zhang YL Lau YL Arakawa E Leung KY 2003 Detection and genetic analysis of group II capsules in Aeromonas hydrophila
Microbiology 149 1051 1060 12686647
Liu TY Gotschlich EC Dunne FT Jonssen EK 1971 Studies on the meningococcal polysaccharides. II. Composition and chemical properties of the group B and group C polysaccharide J Biol Chem 246 4703 4712 4327325
Liu TY Gotschlich EC Jonssen EK Wysocki JR 1971 Studies on the meningococcal polysaccharides. I. Composition and chemical properties of the group A polysaccharide J Biol Chem 246 2849 2858 4995120
Bundle DR Jennings HJ Kenny CP 1974 Studies on the group-specific polysaccharide of Neisseria meningitidis serogroup X and an improved procedure for its isolation J Biol Chem 249 4797 4801 4211095
Branefors-Helander P Kenne L Lindqvist B 1980 Structural studies of the capsular antigen from Haemophilus influenzae type f Carbohydr Res 79 308 312 6965611
Reddy GP Abeygunawardana C Bush CA Cisar JO 1994 The cell wall polysaccharide of Streptococcus gordonii 38: Structure and immunochemical comparison with the receptor polysaccharides of Streptococcus oralis 34 and Streptococcus mitis J22 Glycobiology 4 183 192 8054717
Cisar JO Sandberg AL Reddy GP Abeygunawardana C Bush CA 1997 Structural and antigenic types of cell wall polysaccharides from viridans group streptococci with receptors for oral actinomyces and streptococcal lectins Infect Immun 65 5035 5041 9393793
Vardar D North CL Sanchez-Irizarry C Aster JC Blacklow SC 2003 Nuclear magnetic resonance structure of a prototype Lin12-Notch repeat module from human Notch1 Biochemistry 42 7061 7067 12795601
Ikura M 1996 Calcium binding and conformational response in EF-hand proteins Trends Biochem Sci 21 14 17 8848832
Cooper EL 2003 Comparative immunology Curr Pharm Des 9 119 131 12570663
Khalturin K Panzer Z Cooper MD Bosch TC 2004 Recognition strategies in the innate immune system of ancestral chordates Mol Immunol 41 1077 1087 15476919
Srikrishna G Wang L Freeze HH 1998 Fucosebeta-1-P-Ser is a new type of glycosylation: Using antibodies to identify a novel structure in Dictyostelium discoideum and study multiple types of fucosylation during growth and development Glycobiology 8 799 811 9639541
Mreyen M Champion A Srinivasan S Karuso P Williams KL 2000 Multiple O-glycoforms on the spore coat protein SP96 in Dictyostelium discoideum . Fuc(alpha1–3)GlcNAc-alpha-1-P-Ser is the major modification J Biol Chem 275 12164 12174 10766852
Freeze HH 1997
Dictyostelium discoideum glycoproteins: Using a model system for organismic glycobiology Montreuil J Vliegenhart JFG Schachter H Glycoproteins II New York Elsevier 91 105
Gustafson GL Gander JE 1984 O beta-(N-acetyl-alpha-glucosamine-1-phosphoryl)serine in proteinase I from Dictyostelium discoideum
Methods Enzymol 107 172 183 6438439
Mehta DP Ichikawa M Salimath PV Etchison JR Haak R 1996 A lysosomal cysteine proteinase from Dictyostelium discoideum contains N-acetylglucosamine-1-phosphate bound to serine but not mannose-6-phosphate on N-linked oligosaccharides J Biol Chem 271 10897 10903 8631906
Souza GM Mehta DP Lammertz M Rodriguez-Paris J Wu R 1997
Dictyostelium lysosomal proteins with different sugar modifications sort to functionally distinct compartments J Cell Sci 110 2239 2248 9378773
Descoteaux A Turco SJ 1999 Glycoconjugates in Leishmania infectivity Biochim Biophys Acta 1455 341 352 10571023
McNeely TB Turco SJ 1990 Requirement of lipophosphoglycan for intracellular survival of Leishmania donovani within human monocytes J Immunol 144 2745 2750 2319134
Dermine JF Scianimanico S Prive C Descoteaux A Desjardins M 2000
Leishmania promastigotes require lipophosphoglycan to actively modulate the fusion properties of phagosomes at an early step of phagocytosis Cell Microbiol 2 115 126 11207568
Scianimanico S Desrosiers M Dermine JF Meresse S Descoteaux A 1999 Impaired recruitment of the small GTPase rab7 correlates with the inhibition of phagosome maturation by Leishmania donovani promastigotes Cell Microbiol 1 19 32 11207538
Spath GF Lye LF Segawa H Sacks DL Turco SJ 2003 Persistence without pathology in phosphoglycan-deficient Leishmania major
Science 301 1241 1243 12947201
Spath GF Garraway LA Turco SJ Beverley SM 2003 The role(s) of lipophosphoglycan (LPG) in the establishment of Leishmania major infections in mammalian hosts Proc Natl Acad Sci U S A 100 9536 9541 12869694
Routier FH Higson AP Ivanova IA Ross AJ Tsvetkov YE 2000 Characterization of the elongating alpha-D-mannosyl phosphate transferase from three species of Leishmania using synthetic acceptor substrate analogues Biochemistry 39 8017 8025 10891083
Coffey TJ Dowson CG Daniels M Zhou J Martin C 1991 Horizontal transfer of multiple penicillin-binding protein genes, and capsular biosynthetic genes, in natural populations of Streptococcus pneumoniae
Mol Microbiol 5 2255 2260 1766389
Kroll JS Moxon ER 1990 Capsulation in distantly related strains of Haemophilus influenzae type b: Genetic drift and gene transfer at the capsulation locus J Bacteriol 172 1374 1379 2137816
Tettelin H Saunders NJ Heidelberg J Jeffries AC Nelson KE 2000 Complete genome sequence of Neisseria meningitidis serogroup B strain MC58 Science 287 1809 1815 10710307
Notredame C Higgins DG Heringa J 2000 T-Coffee: A novel method for fast and accurate multiple sequence alignment J Mol Biol 302 205 217 10964570
Nielsen H Engelbrecht J Brunak S von Heijne G 1997 Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage sites Protein Eng 10 1 6
Nielsen H Krogh A 1998 Prediction of signal peptides and signal anchors by a hidden Markov model Proc Int Conf Intell Syst Mol Biol 6 122 130 9783217
Krogh A Larsson B von Heijne G Sonnhammer EL 2001 Predicting transmembrane protein topology with a hidden Markov model: Application to complete genomes J Mol Biol 305 567 580 11152613
Sonnhammer EL von Heijne G Krogh A 1998 A hidden Markov model for predicting transmembrane helices in protein sequences Proc Int Conf Intell Syst Mol Biol 6 175 182 9783223
Hofmann K Bucher P Falquet L Bairoch A 1999 The PROSITE database, its status in 1999 Nucleic Acids Res 27 215 219 9847184
Bateman A Birney E Durbin R Eddy SR Finn RD 1999 Pfam 3.1: 1313 multiple alignments and profile HMMs match the majority of proteins Nucleic Acids Res 27 260 262 9847196
Iseli C Stevenson BJ de Souza SJ Samaia HB Camargo AA 2002 Long-range heterogeneity at the 3′ ends of human mRNAs Genome Res 12 1068 1074 12097343
Stevenson BJ Iseli C Beutler B Jongeneel CV 2003 Use of transcriptome data to unravel the fine structure of genes involved in sepsis J Infect Dis 187 S308 S314 12792844
Sperisen P Iseli C Pagni M Stevenson BJ Bucher P 2004 trome, trEST and trGEN: Databases of predicted protein sequences Nucleic Acids Res 32 D509 D511 14681469
Huelsenbeck JP Ronquist F 2001 MRBAYES: Bayesian inference of phylogenetic trees Bioinformatics 17 754 755 11524383
Ronquist F Huelsenbeck JP 2003 MrBayes 3: Bayesian phylogenetic inference under mixed models Bioinformatics 19 1572 1574 12912839
Felsenstein J 1989 PHYLIP—Phylogeny Inference Package (version 3.2) Cladistics 5 164
Cuff JA Barton GJ 1999 Evaluation and improvement of multiple sequence methods for protein secondary structure prediction Proteins 34 508 519 10081963
|
16299590
|
PMC1285062
|
CC BY
|
2021-01-05 09:18:23
|
no
|
PLoS Comput Biol. 2005 Nov 18; 1(6):e63
|
utf-8
|
PLoS Comput Biol
| 2,005 |
10.1371/journal.pcbi.0010063
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1629958610.1371/journal.pgen.001006205-PLGE-RA-0181R2plge-01-05-05Research ArticleBioinformatics - Computational BiologyEvolutionMicrobiologyGenetics/GenomicsGenetics/Functional GenomicsEubacteriaArchaeaVirusesEvidence of a Large Novel Gene Pool Associated with Prokaryotic Genomic Islands Novel Genes in Genomic IslandsHsiao William W. L 1Ung Korine 1Aeschliman Dana 2Bryan Jenny 2Finlay B. Brett 3Brinkman Fiona S. L 1*1 Department of Molecular Biology and Biochemistry, Simon Fraser University, Burnaby, British Columbia, Canada
2 Department of Statistics, University of British Columbia, Vancouver, British Columbia, Canada
3 Michael Smith Laboratory, University of British Columbia, Vancouver, British Columbia, Canada
Fraser Claire EditorThe Institute for Genomic Research, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 18 11 2005 13 10 2005 1 5 e621 8 2005 13 10 2005 Copyright: © 2005 Hsiao 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.Microbial genes that are “novel” (no detectable homologs in other species) have become of increasing interest as environmental sampling suggests that there are many more such novel genes in yet-to-be-cultured microorganisms. By analyzing known microbial genomic islands and prophages, we developed criteria for systematic identification of putative genomic islands (clusters of genes of probable horizontal origin in a prokaryotic genome) in 63 prokaryotic genomes, and then characterized the distribution of novel genes and other features. All but a few of the genomes examined contained significantly higher proportions of novel genes in their predicted genomic islands compared with the rest of their genome (Paired t test = 4.43E-14 to 1.27E-18, depending on method). Moreover, the reverse observation (i.e., higher proportions of novel genes outside of islands) never reached statistical significance in any organism examined. We show that this higher proportion of novel genes in predicted genomic islands is not due to less accurate gene prediction in genomic island regions, but likely reflects a genuine increase in novel genes in these regions for both bacteria and archaea. This represents the first comprehensive analysis of novel genes in prokaryotic genomic islands and provides clues regarding the origin of novel genes. Our collective results imply that there are different gene pools associated with recently horizontally transmitted genomic regions versus regions that are primarily vertically inherited. Moreover, there are more novel genes within the gene pool associated with genomic islands. Since genomic islands are frequently associated with a particular microbial adaptation, such as antibiotic resistance, pathogen virulence, or metal resistance, this suggests that microbes may have access to a larger “arsenal” of novel genes for adaptation than previously thought.
Synopsis
More than 250 microbial genomes have been sequenced to date. A significant proportion of the genes in these genomes have no apparent similarity to known genes and their functions are unknown (i.e., they appear to be novel). As the number of sequenced genomes increases, the number of these novel genes continues to increase. In this paper, the authors now show, through an analysis of a diverse range of prokaryotic genomes, that novel genes are more prevalent in regions called genomic islands. Genomic islands are clusters of genes in genomes that show evidence of horizontal origins. This study is notable since genomic islands disproportionately contain many genes of medical, agricultural, and environmental importance (e.g., animal and plant pathogen virulence factors, antibiotic resistance genes, phenolic degradation genes, etc.). The observation that high proportions of novel genes are also localized to genomic islands suggests that microbes may have access to a larger “arsenal” of novel genes for important adaptations than previously thought. These results also imply that there are different gene pools associated with recently horizontally transmitted genomic regions versus regions that are primarily vertically inherited. The authors suggest that further studies involving large-scale environmental genomic sampling are required to help characterize this understudied gene pool.
Citation:Hsiao WWL, Ung K, Aeschliman D, Bryan J, Finlay BB, et al. (2005) Evidence of a large novel gene pool associated with prokaryotic genomic islands. PLoS Genet 1(5): e62.
==== Body
Introduction
Since the publication of the first bacterial genome [1], a consistent observation made by biologists is that a significant portion of a prokaryotic genome encodes putative proteins with no known functions. Even in well-studied free-living microbes such as Escherichia coli and Bacillus subtilis, more than 35% of their predicted proteomes do not have functional assignment [2,3]. Peer Bork and others have observed that the functions of less than 70% of proteins in unicellular genomes can be predicted with reasonable confidence, a phenomenon which he termed “the 70% hurdle” [4]. Despite the ever increasing number of genomes becoming available, this observation still holds true in the majority of the genomes sequenced. Moreover, the total number of hypothetical genes is steadily increasing as more genomes are sequenced [5]. This suggests that there may be a genetic pool that is being neglected in functional studies of genes to date. With the exploration of sequence data from environmental samples [6,7], scientists have begun to further appreciate the vast number of novel genes in the environment (in particular those with no detectable homologs versus “conserved hypothetical” genes) that appear to be harbored by yet unculturable and unstudied organisms.
In our studies of selected genomic islands (GIs), defined as horizontally acquired genomic regions that may have mutated to obfuscate or destroy their modes of transmission and integration, we anecdotally observed that the distribution of genes annotated as hypothetical in prokaryotic genomes is non-random. The name genomic island is derived from the term pathogenicity island (PAI), originally coined to describe a cluster of virulence genes identified in uropathogenic E. coli [8] but not found in closely related strains or species. PAIs have been noted for their important roles in bacterial pathogenesis. For example, the pathogenicity island SPI-2 of Salmonella typhimurium encodes a type III secretion system required for intracellular proliferation and systemic infection in a mouse model [9,10]. Mutants of the SPI-2-encoded genes result in attenuation of virulence suggesting that these genes are intricately involved in the infection process [11,12]. Subsequently, genetic elements, which share the same structural features of PAIs, were found in non-pathogenic microorganisms to serve other adaptive functions; these PAI-like elements are collectively referred to as GIs [13]. In the few short years since their discovery, GIs have already been associated with many important adaptive functions that contribute to different microbes' unique life styles. For instance, nitrogen fixation in Rhizobiaceae species is encoded by “symbiosis islands”[14], genes for phenolic compound degradation in Pseudomonas putida are found on “metabolic islands”[15], and the iron-uptake ability of many pathogens are conveyed by “adaptive islands” [16]. Since GIs have been noted to contribute to a microorganism's fitness, metabolic versatility, and adaptability, we decided to develop a method to computationally identify microbial GIs in a large dataset of completely sequenced microbial genomes and to investigate further what features are noted in these agents of microbial innovation. GIs have been previously detected by the genetic features reported to be associated with them [17] and by comparative genomic and phylogenetic approaches [18,19]. Features reported to be associated with GIs include the presence of flanking repeats, mobility genes (e.g., integrases and transposases), proximal transfer RNAs (tRNAs), and atypical guanine and cytosine content [20]. More recently, we and others have used additional species-specific DNA signatures such as oligonucleotide biases and codon adaptation index to identify GIs [21–25]. However, there is a need to better quantify exactly which features and methods best identify GIs. Then we can use more objective criteria to investigate additional features and properties of these important genomic regions.
In this study, we performed a comprehensive analysis of a dataset of 95 known GIs and related prophages to determine which features best identify these genomic regions. We then used sets of objective criteria based on these features to predict putative GIs on a genome-wide scale for 63 prokaryotic organisms. Through analysis of additional features associated with islands, we found that novel hypothetical genes (genes with no detectable homologs using two independent sequence similarity search methodologies, as discussed below) are significantly more prevalent in GIs versus the rest of the genome, irrespective of what method of novel gene identification is used. From this and additional analyses, we propose that there is a large, separate gene pool associated with such horizontally transferred genomic regions; and this gene pool is a more notable source of innovation in a wide range of taxa involving both bacteria and archaea.
Results/Discussion
Prevalence of Features Associated with Reported GIs: Dinucleotide Bias and Associated Mobility Genes Are the Best Predictors
As part of our analysis of features associated with GIs, we created a curated dataset of 95 previously reported horizontally acquired genetic elements (containing 4,553 genes) that we collectively refer to as “known islands” (see Materials and Methods, Table S1). The number of genes in each of these islands ranges from four to 579, reflecting the diversity of these elements. We then inspected each of these known islands for the presence of four GI-associated features (Table 1). This analysis, plus additional analyses of the degree to which each feature overlaps an island (Table S2), indicates that dinucleotide bias is much more sensitive versus conventional %G+C analysis in identifying putative GIs. Using our dataset, the dinucleotide bias approach detected almost three times more of the known islands than the %G+C approach (59 of the islands contain dinucleotide bias versus 23 with %G+C bias). In fact, less than a quarter of the islands examined have abnormal %G+C according to a previously developed cutoff suggesting that by using %G+C alone, many potential GIs may be missed. Only two of the 95 islands examined have abnormally high %G+C, while low %G+C islands are ten times more common. This is in agreement with the observation by Daubin et al. [26,27] that A+T rich genes are preferentially acquired.
Table 1 List of GI-Associated Features and the Number and Percentage of Islands Meeting Each Criterion
Highly expressed genes such as ribosomal proteins have been found to exhibit anomalous sequence compositions [28] and can be a source of false positives in dinucleotide bias analysis. We therefore examined the possibility of incorporating other features into our methodology for island detection. Mobility genes and structural RNA (tRNA and tmRNA) genes also appear to be good indicators of horizontal gene transfer (HGT). Of the 41 GIs inspected, 19 have tRNAs suggesting that phage or phage-like elements may be the precursor for these GIs since some phages are noted for using tRNAs as preferred sites for integration [29]. Three-quarters of the islands inspected contain one or more mobility genes making it the most prevalent feature in our dataset. By combining the two best predictors (mobility genes and dinucleotide bias), approximately 50% of the islands satisfy both criteria. Moreover, among approximately 300 predicted islands in our ORF_ALL dataset (see Materials and Methods) we detected only four potential false positives (i.e., ribosomal protein operons with associated mobility genes that may not be HGT).
In subsequent systematic analyses, we predicted islands using criteria that utilized both a dinucleotide bias-based approach alone (the DINUC dataset with higher sensitivity, also referred to as recall in computer science) and a combined dinucleotide bias method and mobility gene identification (the DIMOB dataset with higher specificity, also referred to as precision). By examining our data using both methods, we were able to assess whether trends we examined held true regardless of whether the method favors sensitivity or specificity.
Overview of Analysis of Predicted GIs: Prevalence of Islands in a Given Microorganism Reflects the Life Styles of the Organism
Our island prediction results are listed in Table S3 (DINUC dataset) and Table S4 (DIMOB dataset). The number of islands and the number of genes in the islands for each of the organisms examined are summarized in Table S5. According to the more precise DIMOB criteria (see Table 2 for summary), 12 organisms did not contain GIs. Most of these organisms are strict intracellular organisms that have restricted access to external gene pools and little or no proposed HGT [30–34], or these organisms appear to be undergoing genome reduction [35]. This supports past statements that HGT occurs more commonly in bacteria that have access to a horizontal gene pool [36].
Table 2 Summary of Organisms without Genomic Islands Based on the DIMOB Criterion
There are some notable limitations to this analysis. Our examination of known islands demonstrates that the DIMOB criterion, though more accurate, does under-identify islands. Furthermore, if a horizontally acquired region shares similar sequence composition features with the host sequence, no composition-based approaches will detect such GIs. This may explain why there were few islands detected in Neisseria meningitidis despite their natural competency for DNA uptake and exchange and their lack of clonality. Neisseriaceae are noted for horizontal DNA exchange [37]; however, this exchange occurs primarily between Neisseria species, which have similar genome sequence compositions. Our analysis represents an examination of HGT between more distantly related organisms (or gene sources) that have different sequence compositions.
Comparative Analysis of GIs versus Non-GIs: Distributions of Gene Function Categories Differ
It has been proposed that certain types of genes are more likely to be horizontally transferred. For example, based on a phylogenetic analysis, Jain et al. observed that informational genes (translation and transcription) are far less likely to be horizontally transferred than operational (housekeeping) genes [38]. A recent paper by Nakamura et al. [24] also found cell surface, DNA binding, and pathogenicity-related genes to be more prevalent in horizontally acquired regions. We analyzed all organisms in our dataset to see if the distributions of protein function categories differ for proteins encoded by genes in islands versus those outside of islands. Genes were classified into 22 clusters of orthologous groups of proteins (COG) functional categories plus a “none” category for proteins without COG assignments (i.e., the “none” category implies that the protein does not have three or more orthologs in other species and so is a relatively “novel” gene).
The most striking observation was that the proportion of genes in the “none” category, which, for readability, we refer to as “proportion of novel genes,” was higher in islands than outside for almost all organisms. On average, 42% of the genes in islands are novel compared to 26% of the genes outside of islands for an organism using the DINUC criteria. The result for the DIMOB dataset is also consistent (53% for islands genes and 28% for outside genes). The actual proportions do vary widely between organisms (though the general trend is consistent) so caution is required when interpreting the means. Figure 1 shows the pair-wise comparison for each organism in the DIMOB dataset and Table S6 tabulates the results for all criteria examined. This observation of a higher proportion of novel genes in islands was statistically significant regardless of whether the DINUC (Paired t test, p-value = 1.27E-18) or the DIMOB (p-value = 1.20E-18; Figure 1) criterion was used to define putative GIs. Since this observation has not been rigorously validated in the past, we decided to characterize and validate this observation further. The other category of genes that is over-represented within islands in both sets of predicted islands (DINUC and DIMOB) is the genes involved in DNA replication, recombination, and repair. Conversely, genes involved in macromolecule biosynthesis (transport and metabolism genes for lipid, amino acid, nucleotide, carbohydrate, and co-enzymes) are present in significantly lower proportions in the predicted islands versus outside of islands. We also found that there is no difference in the proportion of transcriptional genes in islands versus outside of islands. This result may appear to contradict observations made by Jain et al. [38]; however, the differences are likely due more to differences in the type of HGT being detected. Their dataset from six organisms consisted of a rather small set of homologous genes which were more likely to be subject to orthologous displacement than to de novo acquisition. Since sequence compositional approaches are less able to detect orthologous displacement from organisms with similar compositions, our results suggest that if these genes have indeed undergone HGT, the mode of transfer is likely to be homologous recombination between closely related species or ancient HGT that has been subject to amelioration. See Protocol S1 for details regarding this analysis and Table S7 for the tabulated results.
Figure 1 Proportion of Novel Genes in Genomic Islands (Red Bars) versus the Rest of the Genome (Blue Bars) according to a COG-Based Analysis
Proportions of novel genes are calculated as a percentage of all genes within islands or outside of islands, respectively, for each genome (listed on the x axis). A paired t test indicates that significantly more genes in islands versus non-islands do not have a COG classification (p = 1.20E-18). This phenomenon is uniform across prokaryotic lineages and domains. Similar results are also observed if different datasets are analyzed, or different methods for identifying novel genes are used (Table 3).
Table 3 Summary of p-Values Using Different Datasets and Methods
Notably, unlike the “none” category, there is no difference in the proportions of genes in the “general function prediction” and “unknown function” COG categories. These two categories primarily consist of genes encoding conserved hypothetical proteins (i.e., proteins of unknown function that are not “novel” to a given species). This implies that there is not necessarily a bias in terms of what conserved genes have been functionally studied in GIs versus non-GIs to date. The increase in hypothetical genes in GIs is primarily due to the increased occurrence of novel, relatively unconserved genes in these genomic regions.
Higher Proportions of Novel Genes in Predicted GIs Are Independent of the Method Used to Identify Novel Genes
COG analysis is suitable for detecting orthologous genes but may fail to identify more “distant” homologs that have complex evolution histories (e.g., multiple duplications and deletions among lineages). To complement the COG-based analysis of novel genes in GIs, we adapted another independent method to detect novel hypothetical proteins called SUPERFAMILY analysis [39]. We chose this method because of its reported accuracy and its ability to detect more remote homologs based on structurally conserved similarities [40]. Genes that cannot be assigned to a SUPERFAMILY do not have detectable structural domain homologs in the SCOP database, and therefore are more likely to be novel genes. In both the DINUC and DIMOB datasets, we observed that the proportion of such novel genes is significantly higher in islands compared with outside of islands (see Table 3 for a list of p-values from paired t tests). Pair-wise comparison for the DIMOB dataset is illustrated in Figure 2 and tabulated in Table S6 together with the other datasets. This SUPERFAMILY analysis, like the COG-based analysis, indicates that the vast majority of microbes examined have more novel genes in GIs. The SUPERFAMILY analysis, however, may be considered to be more rigorous and subject to less sampling bias in the Tree of Life than the COG-based analysis, because it can detect more distantly related homologs. Notably, the proportion of novel genes outside of islands according to the SUPERFAMILY analysis is remarkably consistent regardless of the lineages of the organisms. Any variability in novel gene content between organisms does appear to primarily occur in GI regions.
Figure 2 Proportion of Novel Genes in Genomic Islands (Red Bars) versus the Rest of the Genome (Blue Bars) according to a SUPERFAMILY-Based Analysis
Proportions of novel genes are calculated as a percentage of all genes within islands or outside of islands, respectively, for each genome (listed on the x axis). A paired t test indicates that significantly higher proportions of genes in islands (red bars) versus outside islands (non-islands; purple bars) do not have a SUPERFAMILY prediction (potential novel genes; p = 4.43E-14).
Higher Proportions of Novel Genes in Predicted GIs Are Not Due to Less Accurate Gene Prediction in GI Regions
One possible explanation for the higher proportion of novel genes in GIs is that genes in GI regions are more frequently mispredicted. Most commonly used gene prediction algorithms today incorporate genomic composition measures such as codon usage to aid in the identification of genes. They also require training with a subset of known genes in an organism in order to become familiar with that organism's genomic composition [41,42]. Gene prediction could presumably be failing more frequently in GI regions because of their differing genomic compositions that lead to more false predictions of genes and consequently more predicted “novel” genes in these regions. Also, since our method of calculating dinucleotide bias uses gene clusters rather than sliding windows of a fixed size (in base pairs), shorter genes may reduce the sampling size to the point of increasing the chance of biased sampling. We addressed this issue by re-examining our data after removal of open reading frames (ORFs) less than 300 bps from our gene sets for each organism. Since longer ORFs are more likely to encode truly functional genes (see Materials and Methods), this reduces the probability that a novel gene is falsely predicted. We used the same two criteria for GI identification, namely dinucleotide bias alone and dinucleotide bias plus mobility gene identification, to generate two lists of islands, which we called the “DINUC_300” and “DIMOB_300” datasets respectively. Because the island detection process was carried out after genes less than 300 bps were removed, this resulted in slightly different lists of the number of islands (see Table S5), as well as which genes were present in each island. Despite these adjustments, the proportion of novel genes in islands was still statistically significantly higher (compared to outside of islands) for both the COG- and SUPERFAMILY-based analyses using either the DINUC or DIMOB dataset (results tabulated in Table S8). Pair-wise t tests for these analyses ranged from a p-value of 2.04E-10 to 1.05E-17 (for complete list, see Table 3). These p-values, while still highly significant, are slightly higher than the ones derived from full gene set. This variation suggests that the ORFs less than 300 bps in length do influence the analysis, but that the contribution is very minor since the p-value is still very significant. We therefore conclude that the over-representation of novel genes in GIs is not predominantly due to more falsely predicted genes in such regions. There appears to be a genuine increase in the number of novel genes in GIs.
Higher Proportions of Novel Genes in Islands Are Not Due to Domain Coverage Bias of COG and SUPERFAMILY
COGs are constructed from fully sequenced microbial genomes which contain prophages and phage-like elements, but exclude plasmids, phages, and other extrachromosomal elements. Therefore COG may have better coverage of the domain comprising prokaryotic chromosomal proteins versus phage and plasmid-associated proteins. SUPERFAMILY, which is based on the SCOP structural classification database, and therefore includes proteins from all domains of life, albeit at different ratios, may be less subject to this bias. Since phage and plasmid mobile elements are potential sources of HGT, higher proportions of novel genes in our DINUC and DIMOB islands may be due to the coverage bias of the methodologies. To investigate this, we searched all of the translated products of the novel genes in and outside of islands against proteins encoded by prokaryotic plasmids and phage genomes using BLAST. With the criteria and database we used, we would expect some novel genes to encode homologs of plasmid and phage-associated proteins, but we wished to discover whether they would be disproportionately associated with genomic islands or not. The results showed that while some of the novel genes did indeed have detectable homologs in our plasmid and phage dataset (Table 4), the majority do not. Moreover, the proportions of novel coding genes with similarity to plasmid and phage proteins are almost identical in the DINUC islands and outside of islands (~30%, see Table 4). For the DIMOB islands, since the dataset is enriched with elements that are more likely to have phage and plasmid origins (by incorporating transposases and integrases as part of the definition of islands and by reducing the number of potential false positives, such as highly expressed genes not associated with these mobile elements), we would expect to see an enrichment of genes with phage and plasmid homologs. Indeed, this is what we observed (Table 4). However, even after taking this potential bias into account by down-adjusting the novel gene counts in DIMOB islands by 11.5% (40.47% minus 28.98%), the proportion of novel genes in islands is still significantly higher than outside (the paired t test p value is 4.76E-16). Therefore, we can conclude that while COG and SUPERFAMILY searches missed some phage or plasmid encoded genes, this omission does not significantly contribute to the observation of higher proportion of novel genes in islands. Notably, the observation suggests that the current sampling of plasmids and phage genomes, which are mostly from culturable prokaryotic hosts, does not account for most of the horizontal gene pool contributing to islands.
Table 4 Proportions of Novel Genes with BLAST Hits in the Phage and Plasmid Database at Expect Value Cutoff of 1E-5
Higher Proportion of Novel Genes in Islands Is Statistically Significant in Many Organisms while the Reverse Is Never Observed
We further assessed the proportion of novel genes at the level of an individual organism (for organisms with more than one chromosome, though each chromosome was analyzed independently). We used a chi-square test of independence or Fisher's exact test (when the number of novel genes in islands is small) to see whether the proportion of novel genes is the same for the within-GI gene pool and for the outside-GI gene pool. Our results (summarized in Table 5) showed that regardless of the GI prediction criteria (DINUC or DIMOB) or the novel gene prediction method (COG or SUPERFAMILY) used the majority of organisms show significantly higher proportion of novel genes in islands. Even after taking multiple testing into account by drastically adjusting the p-values upward using the Bonferroni correction, the observation still holds true. While this biased occurrence of novel genes in GIs is relatively independent of the prokaryotic lineage examined, certain organisms do not exhibit statistically higher proportions of novel genes in islands. This may reflect a genuine reduced access to our described novel gene pool (e.g., hyperthermophilic microorganisms at the base of the Tree of Life may have reduced access to the relevant phage, which as we discuss below, is a probable source of HGT), or it may simply reflect a bias in analysis of organisms with few close relatives. In the latter case, however, one would expect that the proportion of novel genes in non-island regions to be higher, reflecting a lack of similarity to genes in other organisms. However, visual examination of both Figures 1 and 2 indicates that the proportion of novel genes in non-island regions is not notably higher for those organisms with insignificant novel gene bias. It is intriguing to note that organisms with an observed lower proportion of novel genes in islands never achieved statistical significance. This further confirms that the source of genetic material for the GIs analyzed is different and less well characterized than the source of the more stable “core” genome.
Table 5 Number of Organisms Distributed by Proportion of Novel Genes and Statistical Test Significance
Further Analysis of the Proteins Encoded in Islands Indicates that Their Subcellular Localization Distribution Is Different, and Similar to Phage
We used PSORTb version 2.0 [43] for de novo prediction of bacterial protein subcellular localization in order to gain insight into whether genes in islands encode proteins with preferential localizations. PSORTb currently generates the most precise predictions available (more than 95% precision) with similar recall as other methods. Biases toward particular predicted subcellular localizations could provide further clues regarding the origin and function of these novel predicted proteins. We found (Table 6) that proteins encoded in islands are less often predicted to be localized to the cytoplasmic membrane (CM) of both Gram-negative and Gram-positive bacteria (Paired t tests ~5.0E-7 and 8.7E-9, respectively). This is notable, because CM proteins can be predicted with very high accuracy (more than 95% precision and recall). There have been indications that some GIs are of phage origin [26], therefore we performed an additional analysis of CM proteins associated with phage by examining the subcellular localization of deduced proteins from annotated phage or phage-associated genes in E. coli and B. subtilus genomes. We found that the proportion of integrated phage proteins predicted to be in the CM is lower versus other proteins for such organisms. Only 4–5% of phage-associated proteins are predicted to target the CM compared to 15–20% of bacterial genes. These observations further support proposals that phages may be the source of HGT in prokaryotic organisms.
Table 6 Distribution of Predicted Subcellular Localization of Proteins in Islands Compared to Outside of Islands
Implication of Higher Proportions of Novel Genes in Islands: The Big Picture
HGT has been found to be a formidable force in prokaryotic innovation [44]. In this study, we first evaluated several GI-associated features and determined that dinucleotide bias and mobility genes are more sensitive indictors of HGT than the more commonly used %G+C anomaly. Using these indicators, we then constructed a couple of objective criteria to define putative GIs. We have now shown through the largest, most comprehensive analysis of its kind, that novel genes are more likely to be present in GIs and related horizontally acquired regions, than in the rest of a prokaryotic genome. This biased distribution is observed on the majority of taxa examined to date and does not appear to be an artifact of false gene prediction. Moreover, we have shown that the reverse (i.e., higher proportions of novel genes outside of islands) does not reach statistical significance. Also, conserved hypothetical genes do not have a similar distribution bias, suggesting that this observation is not more generally associated with genes of unknown function, but rather is specific to genes that are relatively novel, or specific, to a lineage. Clearly our analysis only detects and examines a subset of horizontally acquired genes as horizontal acquisition from organisms with similar genome sequence compositions and ancient HGT would not be detected by our methods. However, significant implications can still be drawn from our observations.
First, the gene content associated with GIs and related regions is significantly different from gene content in other regions. This implies that the gene pool associated with such GI elements have different composition and characteristics. Furthermore, the increased prevalence of novel genes in such regions suggests that the associated gene pool may be larger, or otherwise subject to more innovation, than in the vertical gene pool. In a recent study by Lerat et al. [45], they took a phylogenetic approach to look at genomic repertoires of gamma-proteobacteria and noticed that single unique genes within the phyla are predominately most parsimoniously explained by HGT from distant sources rather than gene duplication and loss. Based on this and other observations, Lerat et al. also suggested a large pool of available genes for gamma-proteobacteria. Our study indicates that this observation is universal to a wide range of prokaryotes—both from the Bacteria and Archaea domains of life. In this context it is therefore perhaps ominous that GIs are noted for their association with particular adaptations of a microbe, such as antibiotic resistance or pathogen virulence. Significantly higher proportion of novel genes in GIs therefore cautions us that microbes may have a larger “arsenal” of novel genes for adaptation to environments, including resistance to anti-microbial approaches, than previously thought.
We also noted that other biased features of genes in GIs, such as the subcellular localization of their deduced proteins, were consistent with phage. There is increasing evidence that phage and GIs are related [26,46]. While transformation, transduction, and conjugation all have been implicated as mechanisms for HGT, recent analyses have indicated that phage transduction is the predominant force in cross-taxa transfer [29]. With phage diversity approximately ten times that of the prokaryotic diversity, several researchers have proposed that phage can contribute to the genetic individuality of bacterial strains at a much higher level than previously believed [47]. Our results support this, and further support that there is a gene pool with considerable diversity, potentially related to phage, that affects a wide diversity of bacteria of medical and economic importance, as well as archaea. Furthermore, our results pointed out that sampling phages, commonly associated with culturable prokaryotes, are insufficient to elucidate this diverse gene pool and that additional sources still need to be characterized. Metagenomic approaches of environmental samples may provide further clues regarding the nature of these sources.
This work is consistent with the hypothesis that genomes are composed of a more stable set of “core” genes and adaptive “life style” genes [48]. Since genomic sequences only provide snapshots of an organism's evolutionary history, the fate of these “life style” genes is largely unknown. Some evidence suggests that at least some of these life style genes are maintained and have become useful to the hosts [27,45]. Having the ability to draw novel genes from the environment to satisfy short-term needs provides an economical and effective strategy for survival. By characterizing these genes, we can gain new insights into what makes an organism unique and able to adapt to its current environment. While headway has been made in the characterization of conserved hypothetical proteins [49], our most valuable in silico tools for protein characterization are still predominantly based on sequence similarity. Little success has been achieved in de novo characterization of hypothetical proteins. As a result, studying novel hypothetical genes in silico or at the bench provides a significant challenge to researchers. Our inability to effectively functionally characterize these novel hypothetical genes may hamper our ability to devise tailored strategies to combat different microbial pathogens and resistance mechanisms.
Recent efforts focused on environmental genomic sampling and metagenomic projects [6,50,51] will help us obtain sequences that may elucidate the sources of these novel hypothetical genes from organisms whose genomes have yet to be sequenced. Our results provide a strong rationale for the continuation of these efforts. Regardless of the source of this innovation, it appears that novel genes are being acquired by prokaryotes disproportionately through GIs. A wide range of microbial research areas are impacted by this adaptation strategy due to the association between GIs and microbial adaptations of importance.
Materials and Methods
Genome sequence data and organisms examined.
Sequence and annotation of each ORF of completely sequenced prokaryotic genomes were downloaded from the National Center for Biotechnology Information (NCBI) FTP site in April, 2004. We limited our final dataset to the 63 organisms used in the most recent publication that analyzed COG [52] because we wished to adopt the most consistent and accurate COG dataset in our analysis. The selection of 63 organisms, nevertheless, represents a wide range of taxa [52]. We examined only chromosomal sequences, not plasmid data. To reduce the number of falsely predicted ORFs and to avoid sequence compositional bias due to short ORFs, we also constructed a separate dataset by excluding any ORFs smaller than 300 bps. We labeled the first set of ORFs containing all the predicted ORFs, “ORF_ALL,” and the second set “ORF_300.”
GI dataset development and validation of GI features.
We constructed a verified dataset of 41 known GIs and 54 prophages from 14 well-studied organisms (ten species) through a manual literature research (Tables S1 and S2). We then examined the prevalence of four sequence and annotation features commonly reported with GIs. These four features (%G+C bias, dinucleotide bias [53], presence of tRNA genes, and presence of mobility genes) were identified using our previously developed IslandPath software [21]. For each available prokaryotic genome, IslandPath generates a graphic representation of the genome and superimposes these features on the image. The genetic elements in our dataset were inspected manually, using the IslandPath analysis, for the presence or absence of the four GI-associated features. For mobility and tRNA genes, an island is scored positively if it contains at least one gene annotated as such. An island is considered to exhibit %G+C or dinucleotide bias if more than half of the ORFs in that island have these biases as determined by IslandPath [21]. ORFs with %G+C more than 4.62% above or below the genome average are marked as “High %G+C” or “Low %G+C”, respectively. All the other ORFs are noted as “Normal %G+C.” We derived this cutoff from a previous study of genome %G+C variation in obligate intracellular bacteria that are thought to be subject to little horizontal gene transfer [36]. This cutoff is thought to reflect the inherent %G+C variation of an organism due to other factors such as gene expression level [36].
GI prediction.
Based on our GI feature validation results, we defined a putative GI as eight or more consecutive ORFs with dinucleotide bias (DINUC dataset), or eight or more consecutive ORFs with dinucleotide bias plus at least one mobility gene present in the region (DIMOB dataset). Mobility genes were identified using the NCBI annotation and PFAM hidden Markov models (HMMs) searches. The PFAM HMM search was conducted as follows: in order to identify putative mobility genes in a large number of genomes in a reasonable amount of time, we used the Paracel GeneMatcher system, a hardware-based solution for carrying out similarity search in parallel. To further speed up the search, instead of searching all of the predicted ORFs against all of the PFAM HMMs, we identified and searched against 46 PFAM HMMs representing mobility genes (e.g., integrases and transposases). Results with expect values (similar to BLAST E-value) smaller than 0.01 were retained. Manual inspection of results from a randomly selected set of five species did not reveal any obvious false positives using this cutoff. Genomic regions that satisfied the above criteria were extracted and the genes in these regions were labeled as “islands.” The rest of the genes were labeled as “outside of islands.”
We also performed a prediction of GIs after removal of all genes that are less than 300 bps in length (ORF_300) to reduce the possible impact of incorrectly predicted small genes on our analysis. We chose a 300-bps cutoff (corresponding to 100 amino acids) because we and others have previously found, through comparisons of the genome-wide gene predictions of closely related organisms not subject to much HGT, that annotation of genes shorter than this cutoff by separate groups becomes more inconsistent [36]. In addition, a 300-bps cutoff has been commonly used for some genome annotation processes [33].
Functional characterization of genes in GIs.
To avoid inconsistencies in genome annotation from different sequencing projects, we used two independent bioinformatic tools to assign ORFs to different functional categories. We chose COG and SUPERFAMILY [54] because of their complementarities. COG is suitable for predicting “closely related” homologs which are likely to be orthologous because a COG is defined as three or more proteins that all share the highest sequence similarity with each other. Detailed description of how a COG was constructed and subsequently updated can be found in [55] and [52] SUPERFAMILY, on the other hand, provides functional assignments to protein sequences at the superfamily level of the SCOP protein structural classification system [56]. Proteins in a SCOP superfamily are likely to share a common evolutionary origin based on their structural similarities. As a result, the SUPERFAMILY predictors are useful in detecting more remote homologs that have similar structural features. Both programs have been shown to make reliable assignments and have been widely used [40,57,58]. COG assignment results were obtained from the NCBI FTP site. We used the dataset published with the updated COG paper rather than the subsequent assignments associated with the NCBI genome “ptt” files since there appears to be some inconsistency and omissions of COG assignments in these files. Pre-computed SUPERFAMILY genome assignment results (Version ass_09 May, 2004) were obtained, with permission, from Julian Gould, the original lead author of the SUPERFAMILY database. With both COG and SUPERFAMILY assignments, we used the cutoffs set by the respective authors for filtering out non-significant hits. For SUPERFAMILY, the expect value cutoff used was 0.02 (provided by the authors of the database). Since we are not trying to identify specific functions using SUPERFAMILY and can tolerate some false assignments, this more relaxed cutoff seems adequate. For COG, the cutoff(s) used is not reported and as argued by the authors of the COG database, the absolute cutoff is not crucial since all COGs have to satisfy the “best BLAST hit to multiple other organisms” constraint.
The phage genome and plasmid records were obtained from the NCBI Entrez Genome site (http://www.ncbi.nih.gov/entrez/query.fcgi?db=Genome) in September, 2005. There are 284 phage genomes and 716 plasmids records. Protein FASTA records associated with these genomic sequences were downloaded by following the NCBI Protein Linkouts of these records. These protein records were converted into a local BLAST database and searched against using the NCBI BLASTP program. Queries (translated products of either COG-based or SUPERFAMILY-based novel genes) that have database matches with an expect value less than 1E-5 were considered to have homologs in this phage and plasmid database.
PSORTb version 2.0 [43] was used to predict protein subcellular localization for deduced proteins from all complete genomes analyzed. Custom Perl scripts were used to combine records obtained from various sources and to link annotations.
Statistical analyses.
Each COG functional category was assessed for over- and under-representation in predicted GIs across all species. This was done by first expressing the number of genes in a category as a percentage of all the genes in islands for a given organism. The percentage of genes outside of islands for the same category was likewise calculated. We calculated the two percentages for each organism and for each category including a “none” category into which we assigned genes without a COG category. For each category, we could then determine if the genes in that category are over- or under-represented in islands through a paired t test analysis (in island versus outside island) across all organisms. We carried out the same t test analysis to determine whether genes lacking a SUPERFAMILY prediction are over-represented in island across all organisms. For each organism, we also determined if the proportions of “novel hypothetical” genes (genes without COG or SUPERFAMILY assignments) in islands are significantly different from those outside of islands using chi-square test of independence. In a few cases where the numbers of these novel hypothetical genes are small in islands, we used Fisher Exact test instead. We considered p-values smaller than 0.05 to be significant. Statistical analyses were done using R statistics package.
Supporting Information
Protocol S1 Additional Analysis and Discussion Regarding COG Categories
(22 KB DOC)
Click here for additional data file.
Table S1 List of Reported and Known Genomic Islands and Phages
(28 KB XLS)
Click here for additional data file.
Table S2 List of Reported and Known Genomic Islands and Phages (Graded)
(24 KB XLS)
Click here for additional data file.
Table S3 List of Genes in Predicted Genomic Islands (Criterion: DINUC_ALL)
(3187 KB XLS)
Click here for additional data file.
Table S4 List of Genes in Predicted Genomic Islands (Criterion: DIMOB_ALL)
(903 KB XLS)
Click here for additional data file.
Table S5 Number of Islands and Number of Genes in Islands by Organisms
(25 KB PDF)
Click here for additional data file.
Table S6 Summary of Gene Counts in Islands and outside of Islands (All ORFs Included)
(27 KB PDF)
Click here for additional data file.
Table S7 Gene Count in Each of the COG Categories (Excluding the None Category)
(73 KB PDF)
Click here for additional data file.
Table S8 Summary of Gene Counts in Islands and outside of Islands (ORFS < 300 bps Excluded)
(29 KB PDF)
Click here for additional data file.
We thank the National Center for Biotechnology, United States for providing genomic sequence and annotation files, the Canadian Bioinformatic Resource (National Research Council, Canada) for hosting GeneMatcher, and Julian Gough for providing the SUPERFAMILY pre-computed genomic data. WWLH is supported by a Michael Smith Foundation for Health Research (MSFHR) Trainee Award and a Canadian Institutes of Health Research (CIHR) Doctoral Scholarship. FSLB and JB are MSFHR Scholars and FSLB is a CIHR New Investigator. Funding was provided in part by the Functional Pathogenomics of Mucosal Immunity Project, which is supported by Genome Prairie/Genome BC and Inimex Pharmaceuticals.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. WWLH, BF, and FSLB conceived and designed the experiments. WWLH and KU performed the experiments. WWLH, DA, and JB analyzed the data. WWLH and FSLB wrote the paper.
A previous version of this article appeared as an Early Online Release on October 13, 2005 (DOI: 10.1371/journal.pgen.0010062.eor).
Abbreviations
bpbase pair
CMcytoplasmic membrane
COGclusters of orthologous groups of proteins
GIgenomic island
HGThorizontal gene transfer
HMMhidden Markov models
ORFopen reading frame
PAIpathogenicity island
==== Refs
References
Fleischmann RD Adams MD White O Clayton RA Kirkness EF 1995 Whole-genome random sequencing and assembly of Haemophilus influenzae Rd Science 269 496 512 7542800
Kunst F Ogasawara N Moszer I Albertini AM Alloni G 1997 The complete genome sequence of the gram-positive bacterium Bacillus subtilis
Nature 390 249 256 9384377
Blattner FR Plunkett G 3rd Bloch CA Perna NT Burland V 1997 The complete genome sequence of Escherichia coli K-12 Science 277 1453 1474 9278503
Bork P 2000 Powers and pitfalls in sequence analysis: The 70% hurdle Genome Res 10 398 400 10779480
Siew N Fischer D 2003 Analysis of singleton ORFans in fully sequenced microbial genomes Proteins 53 241 251 14517975
Venter JC Remington K Heidelberg JF Halpern AL Rusch D 2004 Environmental genome shotgun sequencing of the Sargasso Sea Science 304 66 74 15001713
Breitbart M Wegley L Leeds S Schoenfeld T Rohwer F 2004 Phage community dynamics in hot springs Appl Environ Microbiol 70 1633 1640 15006788
Hacker J Bender L Ott M Wingender J Lund B 1990 Deletions of chromosomal regions coding for fimbriae and hemolysins occur in vitro and in vivo in various extraintestinal Escherichia coli isolates Microb Pathog 8 213 225 1974320
Ochman H Soncini FC Solomon F Groisman EA 1996 Identification of a pathogenicity island required for Salmonella survival in host cells Proc Natl Acad Sci U S A 93 7800 7804 8755556
Hensel M 2000
Salmonella pathogenicity island 2 Mol Microbiol 36 1015 1023 10844687
Coombes BK Brown NF Kujat-Choy S Vallance BA Finlay BB 2003 SseA is required for translocation of Salmonella pathogenicity island-2 effectors into host cells Microbes Infect 5 561 570 12787732
Cirillo DM Valdivia RH Monack DM Falkow S 1998 Macrophage-dependent induction of the Salmonella pathogenicity island 2 type III secretion system and its role in intracellular survival Mol Microbiol 30 175 188 9786194
Hacker J Kaper JB 2000 Pathogenicity islands and the evolution of microbes Annu Rev Microbiol 54 641 679 11018140
Sullivan JT Ronson CW 1998 Evolution of rhizobia by acquisition of a 500-kb symbiosis island that integrates into a phe-tRNA gene Proc Natl Acad Sci U S A 95 5145 5149 9560243
Ravatn R Studer S Springael D Zehnder AJ van der Meer JR 1998 Chromosomal integration, tandem amplification, and deamplification in Pseudomona s putida F1 of a 105-kilobase genetic element containing the chlorocatechol degradative genes from Pseudomonas sp. Strain B13 J Bacteriol 180 4360 4369 9721270
Dobrindt U Hochhut B Hentschel U Hacker J 2004 Genomic islands in pathogenic and environmental microorganisms Nat Rev Microbiol 2 414 424 15100694
Moss JE Cardozo TJ Zychlinsky A Groisman EA 1999 The selC-associated SHI-2 pathogenicity island of Shigella flexneri
Mol Microbiol 33 74 83 10411725
Karaolis DK Johnson JA Bailey CC Boedeker EC Kaper JB 1998 A Vibrio cholerae pathogenicity island associated with epidemic and pandemic strains Proc Natl Acad Sci U S A 95 3134 3139 9501228
McClelland M Sanderson KE Spieth J Clifton SW Latreille P 2001 Complete genome sequence of Salmonella enterica serovar Typhimurium LT2 Nature 413 852 856 11677609
Hacker J Blum-Oehler G Muhldorfer I Tschape H 1997 Pathogenicity islands of virulent bacteria: Structure, function, and impact on microbial evolution Mol Microbiol 23 1089 1097 9106201
Hsiao W Wan I Jones SJ Brinkman FS 2003 IslandPath: Aiding detection of genomic islands in prokaryotes Bioinformatics 19 418 420 12584130
Lio P Vannucci M 2000 Finding pathogenicity islands and gene transfer events in genome data Bioinformatics 16 932 940 11120683
Tu Q Ding D 2003 Detecting pathogenicity islands and anomalous gene clusters by iterative discriminant analysis FEMS Microbiol Lett 221 269 275 12725938
Nakamura Y Itoh T Matsuda H Gojobori T 2004 Biased biological functions of horizontally transferred genes in prokaryotic genomes Nat Genet 36 760 766 15208628
Merkl R 2004 SIGI: Score-based identification of genomic islands BMC Bioinformatics 5 22 15113412
Daubin V Lerat E Perriere G 2003 The source of laterally transferred genes in bacterial genomes Genome Biol 4 R57 12952536
Daubin V Ochman H 2004 Bacterial genomes as new gene homes: The genealogy of ORFans in E. coli
Genome Res 14 1036 1042 15173110
Karlin S Mrazek J Campbell AM 1998 Codon usages in different gene classes of the Escherichia coli genome Mol Microbiol 29 1341 1355 9781873
Canchaya C Fournous G Chibani-Chennoufi S Dillmann ML Brussow H 2003 Phage as agents of lateral gene transfer Curr Opin Microbiol 6 417 424 12941415
Dykhuizen DE Baranton G 2001 The implications of a low rate of horizontal transfer in Borrelia
Trends Microbiol 9 344 350 11435109
Tamas I Klasson L Canback B Naslund AK Eriksson AS 2002 50 million years of genomic stasis in endosymbiotic bacteria Science 296 2376 2379 12089438
Andersson JO Andersson SG 1999 Insights into the evolutionary process of genome degradation Curr Opin Genet Dev 9 664 671 10607609
Kawarabayasi Y Hino Y Horikawa H Yamazaki S Haikawa Y 1999 Complete genome sequence of an aerobic hyper-thermophilic crenarchaeon, Aeropyrum pernix K1 DNA Res 6 83 101 145 152 10382966
Slesarev AI Mezhevaya KV Makarova KS Polushin NN Shcherbinina OV 2002 The complete genome of hyperthermophile Methanopyrus kandleri AV19 and monophyly of archaeal methanogens Proc Natl Acad Sci U S A 99 4644 4649 11930014
Cole ST Eiglmeier K Parkhill J James KD Thomson NR 2001 Massive gene decay in the leprosy bacillus Nature 409 1007 1011 11234002
Brinkman FS Blanchard JL Cherkasov A Av-Gay Y Brunham RC 2002 Evidence that plant-like genes in Chlamydia species reflect an ancestral relationship between Chlamydiaceae, cyanobacteria, and the chloroplast Genome Res 12 1159 1167 12176923
Elkins C Thomas CE Seifert HS Sparling PF 1991 Species-specific uptake of DNA by gonococci is mediated by a 10-base-pair sequence J Bacteriol 173 3911 3913 1904861
Jain R Rivera MC Lake JA 1999 Horizontal gene transfer among genomes: The complexity hypothesis Proc Natl Acad Sci U S A 96 3801 3806 10097118
Gough J Chothia C 2002 SUPERFAMILY: HMMs representing all proteins of known structure. SCOP sequence searches, alignments, and genome assignments Nucleic Acids Res 30 268 272 11752312
Okazaki Y Furuno M Kasukawa T Adachi J Bono H 2002 Analysis of the mouse transcriptome based on functional annotation of 60,770 full-length cDNAs Nature 420 563 573 12466851
Delcher AL Harmon D Kasif S White O Salzberg SL 1999 Improved microbial gene identification with GLIMMER Nucleic Acids Res 27 4636 4641 10556321
Lukashin AV Borodovsky M 1998 GeneMark.hmm: New solutions for gene finding Nucleic Acids Res 26 1107 1115 9461475
Gardy JL Laird MR Chen F Rey S Walsh CJ 2005 PSORTb v.2.0: Expanded prediction of bacterial protein subcellular localization and insights gained from comparative proteome analysis Bioinformatics 21 617 623 15501914
Ochman H Lawrence JG Groisman EA 2000 Lateral gene transfer and the nature of bacterial innovation Nature 405 299 304 10830951
Lerat E Daubin V Ochman H Moran NA 2005 Evolutionary origins of genomic repertoires in bacteria PLoS Biol 3 e130. DOI: 10.1371/journal.phio.0030130 15799709
Boyd EF Davis BM Hochhut B 2001 Bacteriophage-bacteriophage interactions in the evolution of pathogenic bacteria Trends Microbiol 9 137 144 11303502
Canchaya C Fournous G Brussow H 2004 The impact of prophages on bacterial chromosomes Mol Microbiol 53 9 18 15225299
Makarova KS Aravind L Galperin MY Grishin NV Tatusov RL 1999 Comparative genomics of the Archaea (Euryarchaeota): evolution of conserved protein families, the stable core, and the variable shell Genome Res 9 608 628 10413400
Kolker E Makarova KS Shabalina S Picone AF Purvine S 2004 Identification and functional analysis of “hypothetical” genes expressed in Haemophilus influenzae
Nucleic Acids Res 32 2353 2361 15121896
Tyson GW Chapman J Hugenholtz P Allen EE Ram RJ 2004 Community structure and metabolism through reconstruction of microbial genomes from the environment Nature 428 37 43 14961025
Breitbart M Felts B Kelley S Mahaffy JM Nulton J 2004 Diversity and population structure of a near-shore marine-sediment viral community Proc R Soc Lond B Biol Sci 271 565 574
Tatusov RL Fedorova ND Jackson JD Jacobs AR Kiryutin B 2003 The COG database: An updated version includes eukaryotes BMC Bioinformatics 4 41 12969510
Karlin S 2001 Detecting anomalous gene clusters and pathogenicity islands in diverse bacterial genomes Trends Microbiol 9 335 343 11435108
Madera M Vogel C Kummerfeld SK Chothia C Gough J 2004 The SUPERFAMILY database in 2004: Additions and improvements Nucleic Acids Res 32 Database issue D235–239
Tatusov RL Koonin EV Lipman DJ 1997 A genomic perspective on protein families Science 278 631 637 9381173
Andreeva A Howorth D Brenner SE Hubbard TJ Chothia C 2004 SCOP database in 2004: Refinements integrate structure and sequence family data Nucleic Acids Res 32 Database issue D226–229
Raymond J Zhaxybayeva O Gogarten JP Gerdes SY Blankenship RE 2002 Whole-genome analysis of photosynthetic prokaryotes Science 298 1616 1620 12446909
Jordan IK Rogozin IB Wolf YI Koonin EV 2002 Essential genes are more evolutionarily conserved than are nonessential genes in bacteria Genome Res 12 962 968 12045149
Faguy DM Doolittle WF 1999 Lessons from the Aeropyrum pernix genome Curr Biol 9 R883 886 10607556
Brochier C Forterre P Gribaldo S 2004 Archaeal phylogeny based on proteins of the transcription and translation machineries: Tackling the Methanopyrus kandleri paradox Genome Biol 5 R17 15003120
Shigenobu S Watanabe H Hattori M Sakaki Y Ishikawa H 2000 Genome sequence of the endocellular bacterial symbiont of aphids Buchnera sp. APS Nature 407 81 86 10993077
Parkhill J Wren BW Mungall K Ketley JM Churcher C 2000 The genome sequence of the food-borne pathogen Campylobacter jejuni reveals hypervariable sequences Nature 403 665 668 10688204
Eppinger M Baar C Raddatz G Huson DH Schuster SC 2004 Comparative analysis of four Campylobacterales
Nat Rev Microbiol 2 872 885 15494744
Smith DR Doucette-Stamm LA Deloughery C Lee H Dubois J 1997 Complete genome sequence of Methanobacterium thermoautotrophicum deltaH: functional analysis and comparative genomics J Bacteriol 179 7135 7155 9371463
Fraser CM Gocayne JD White O Adams MD Clayton RA 1995 The minimal gene complement of Mycoplasma genitalium
Science 270 397 403 7569993
Ogata H Audic S Renesto-Audiffren P Fournier PE Barbe V 2001 Mechanisms of evolution in Rickettsia conorii and R. prowazekii
Science 293 2093 2098 11557893
Nelson KE Eisen JA Fraser CM 2001 Genome of Thermotoga maritima MSB8 Methods Enzymol 330 169 180 11210497
|
16299586
|
PMC1285063
|
CC BY
|
2021-01-05 07:59:36
|
no
|
PLoS Genet. 2005 Nov 18; 1(5):e62
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010062
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1629958710.1371/journal.pgen.001006305-PLGE-RA-0152R1plge-01-05-04Research ArticleDevelopmentNeuroscienceGenetics/Gene DiscoveryGenetics/Gene FunctionGenetics/Complex TraitsDrosophila
Drosophila tan Encodes a Novel Hydrolase Required in Pigmentation and Vision Drosophila tan in Pigmentation and Vision
True John R 1*Yeh Shu-Dan 1Hovemann Bernhard T 2Kemme Tobias 2Meinertzhagen Ian A 3Edwards Tara N 4Liou Shian-Ren 1Han Qian 5Li Jianyong 51 Department of Ecology and Evolution, State University of New York, Stony Brook, New York, United States of America
2 Fakultät für Chemie, Ruhr-Universität Bochum, Bochum, Germany
3 Life Sciences Centre, Dalhousie University, Halifax, Nova Scotia, Canada
4 Department of Biology, Dalhousie University, Halifax, Nova Scotia, Canada
5 Department of Veterinary Pathology, College of Veterinary Medicine, University of Illinois, Urbana, Illinois, United States of America
Barsh Gregory EditorStanford University School of Medicine, United States of America*To whom correspondence should be sent. E-mail: [email protected] 2005 18 11 2005 15 10 2005 1 5 e635 7 2005 14 10 2005 Copyright: © 2005 True 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.Many proteins are used repeatedly in development, but usually the function of the protein is similar in the different contexts. Here we report that the classical Drosophila melanogaster locus tan encodes a novel enzyme required for two very different cellular functions: hydrolysis of N-β-alanyl dopamine (NBAD) to dopamine during cuticular melanization, and hydrolysis of carcinine to histamine in the metabolism of photoreceptor neurotransmitter. We characterized two tan-like P-element insertions that failed to complement classical tan mutations. Both are inserted in the 5′ untranslated region of the previously uncharacterized gene CG12120, a putative homolog of fungal isopenicillin-N N-acyltransferase (EC 2.3.1.164). Both P insertions showed abnormally low transcription of the CG12120 mRNA. Ectopic CG12120 expression rescued tan mutant pigmentation phenotypes and caused the production of striking black melanin patterns. Electroretinogram and head histamine assays indicated that CG12120 is required for hydrolysis of carcinine to histamine, which is required for histaminergic neurotransmission. Recombinant CG12120 protein efficiently hydrolyzed both NBAD to dopamine and carcinine to histamine. We conclude that D. melanogaster CG12120 corresponds to tan. This is, to our knowledge, the first molecular genetic characterization of NBAD hydrolase and carcinine hydrolase activity in any organism and is central to the understanding of pigmentation and photoreceptor function.
Synopsis
True et al. describe the identification and characterization of the Drosophila melanogaster enzyme Tan. The gene encoding Tan was originally discovered in the early 20th century as a mutant strain lacking the dark pigmentation of wild-type flies, hence the name tan. Flies lacking Tan function also exhibited mysterious abnormalities in vision, for example, in responses to light. The new findings by True et al. help to explain the vastly different functions of Tan in pigmentation and vision. In the developing epidermal cells that secrete the adult cuticle, the enzyme encoded by tan is required for the production of dopamine, which is needed for dark melanin pigmentation. In the eye, the Tan enzyme converts carcinine, a modified form of the neurotransmitter histamine, back to histamine, which is necessary for the rapid and constant neurotransmission events involved in vision. These two enzyme activities have not been previously characterized in any organism. Surprisingly, Tan appears to be closely related to an enzyme in fungi that is used for production of the antibiotic penicillin.
Citation:True JR, Yeh SD, Hovemann BT, Kemme T, Meinertzhagen IA, et al. (2005) Drosophila tan encodes a novel hydrolase required in pigmentation and vision. PLoS Genet 1(5): e63.
==== Body
Introduction
One of the most important generalizations to emerge from contemporary developmental genetics is that gene products are typically used more than once during development. Usually multiple functions involve the protein performing the same task, e.g., activating transcription or binding to the cytoskeleton. A more surprising form of multifunctionality occurs in enzymes adapted to perform a catalytic function on different substrates in distinct developmental or metabolic pathways. The products encoded by the Drosophila melanogaster ebony and tan genes, mutations in which cause reciprocal pigmentation defects but related neurological phenotypes, have been proposed as such a system [1–6]. ebony encodes N-β-alanyl dopamine (NBAD) synthase, which in epidermal cells converts the melanin precursor dopamine to N-β-alanyl dopamine, a precursor to tan-colored pigment, during cuticle development in late pupae [2,6].
The molecular nature of the tan gene, however, has remained a mystery. The tan gene product has been proposed to catalyze the conversion of NBAD to dopamine via hydrolysis, which is the reverse of the activity catalyzed by the Ebony protein [2]. The reciprocal pigmentation effects of tan and ebony mutants can thus be explained: excess NBAD and deficient dopamine result in the abnormally light phenotype of tan, whereas excess dopamine and deficient NBAD result in the melanic phenotype of ebony, in which excess dopamine is converted to melanin.
The Ebony protein is related to a family of fungal and bacterial non-ribosomal peptide synthetases sharing ATP-dependent carboxy acid activation via acyladenylate and a thioester module that binds a 4′ phosphopantetheinyl cofactor during conversion from the apo to holo form [2,4]. Richardt et al. [4] demonstrated that Ebony protein is capable of capturing a number of biogenic amines, including histamine, and of binding these amines to β-alanine, as predicted from the biochemical genetic data. The extended lineage of ebony suggests that the putative ebony–tan circuit may be quite ancient. However, the apparent absence of NBAD in vertebrates suggests that the complementary functions of ebony and tan in dopamine metabolism and pigmentation may have been derived in protostomes or, more restrictively, in arthropods. More definitive evidence on the evolutionary history of this circuit could be provided by characterizing the tan locus. A further motivation to characterize the role of tan in pigment formation is provided by its potential involvement in pigment pattern evolution in insects [7,8].
Recent studies have begun to shed light on the neural functions of tan and ebony. The detailed characterization of Ebony function and expression [4,5] strongly suggests that the Ebony–Tan circuit plays a role in the metabolism of histamine, a photoreceptor neurotransmitter. The optomotor and visual defects of ebony mutants are consistent with reduced transmission from the terminals of the photoreceptors R1–R6. This possibility was initially suggested by the loss in ebony mutants of the “on” and “off” transients of electroretinograms (ERGs), first reported by Hotta and Benzer [9], but without explanation at that time. This defect is now attributable to the loss of transmission to lamina interneurons [10]. More recently, Richardt et al. [5] demonstrated that Ebony protein is expressed in non-neuronal cells of the first and second optic neuropiles, the lamina and medulla, respectively, in particular the epithelial glia of the lamina and the neuropile glia of the distal medulla. These glia are located at the sites of histamine release from the photoreceptors [5]. Collectively, the localization data and the biochemical function of Ebony as a general β-alanyl biogenic amine synthase [4] suggested that Ebony's role is to deactivate histamine by conjugating it with β-alanine, to form N-β-alanyl histamine, also known as carcinine, after histamine release from the photoreceptor terminal. Rapid histamine deactivation and reuptake are necessary for photoreceptor function because histamine release rates are high [11,12], whereas histamine synthesis and degradation are relatively slow [13].
Evidence for a role of the Ebony–Tan circuit in regulating histamine conjugation and regeneration in vivo prior to its presumed reuse by photoreceptors was provided by Borycz et al. [3], who demonstrated that fly heads rapidly convert microinjected histamine to carcinine and that tan mutants have abnormally high quantities of carcinine, hydrolysis of which is blocked. The same study also demonstrated that both ebony and tan mutants have substantially reduced head histamine content, as well as fewer synaptic vesicles at photoreceptor terminals. Together with the evidence on Ebony localization and function [4,5], these observations have led to a model of Ebony–Tan function in the fly's visual system. Histamine released at the photoreceptor terminals is captured by conversion to carcinine by Ebony in the surrounding glial cells. This carcinine is subsequently hydrolyzed by the product of the tan gene, which liberates the histamine for reuse by the presynaptic photoreceptor neuron. Neither the function nor the localization of Tan has yet been established, however.
Validation of this model requires molecular identification and characterization of the tan locus and its product, a putative NBAD/carcinine hydrolase that has yet to be identified in any organism. Here we present genetic and biochemical evidence that the D. melanogaster tan locus corresponds to predicted gene CG12120 and confirm that it encodes a bifunctional NBAD/carcinine hydrolase. Tan/CG12120 is related to fungal isopenicillin-N N-acyltransferases (IATs), suggesting that like Ebony, insect NBAD/carcinine hydrolase also represents an ancient enzyme. We discuss the cellular and evolutionary implications of this discovery for models of both photoreceptor function and pigmentation development.
Results
Two P-element Insertions in CG12120 Exhibit tan-Like Phenotypes and Fail to Complement tan Mutants
The tan locus was originally mapped to the cytological interval 8C3-9E3 by the inclusion of tan in Df(1)t282–1 [14]. We used meiotic mapping of local P-element insertions to determine that tan maps 0.51 cM proximal to P-element P{EPgy2}EY04394 (near CG12118) at position 8D1 and 0.51 cM distal to P-element P{EPgy2}EY05996 (near CG17754) at position 8D3 (Figure 1). This map interval contains 16 known or predicted genes. We subsequently obtained two new P insertions in this interval, P{d07784} [15] and P{g1557} [16], both of which were associated with a tan-like pigmentation phenotype (Figure 1E and 1F; compare with wild-type, Figure 1C, and tan mutant tan5, Figure 1D). One of these P inserts, P{d07784} was crossed with four classical tan mutants (tan1, tan2, tan4, and tan5) and failed to complement them for the pigmentation phenotype. These two P insertion lines were sequenced to determine the positions of the P insertions. In both lines, the P-element is inserted in the 5′ end of the first exon of predicted gene CG12120, at positions +9 and +21 bp, respectively (Figure 1B).
Figure 1
P-Element Insertions in CG12120 Cause tan-Like Pigmentation Phenotypes and Ectopic Expression of CG12120 Rescues the tan Phenotype and Causes Ectopic Melanin Formation
(A) The tan region of the D. melanogaster X chromosome. Known and predicted genes with directions of transcription are depicted below the line. Cytological divisions and flanking P-element insertions, as well as meiotic mapping data (see text), are indicated above the line.
(B) Transcription unit of predicted gene CG12120. Boxes indicate exons. Black indicates coding region. Transcription start site, inferred from EST data [15], is at the position designated +1. Start codon is at position +250. Below are indicated positions of P{d07784} and P{g1557} insertions, just 3′ to the transcription initiation site (+1) of CG12120. Transcription unit is indicated in bold.
(C–F) Adult (3–5 d) female body coloration of (C) CantonS (wild-type), (D) tan5, (E) w P{d07784}, and (F) w P{g1557}.
(G and H) w tan1/w tan1; CyO/+; P{w+ C765-GAL4}/+ control female (G) and w tan1/w tan1; P{w+ UAS-CG12120}/+; P{ w+ C765-GAL4}/+ female (H) showing rescue of abdomen phenotype (arrows).
(I and J) w/w; CyO/+; P{w+ pnr-GAL4}/+ control female (I) and w/w; P{w+ UAS-CG12120}/+; P{w+ pnr-GAL4}/+ female (J) showing ectopic melanin pattern along dorsal midline (arrowheads).
All flies depicted were 3–5 d of age.
CG12120 Expression Rescues the tan Pigmentation Phenotype
The complete 2,009-bp CG12120 cDNA was cloned into the pUAST vector [17] and used to transform a yellow white (yw) strain of D. melanogaster. Transformant lines were crossed into y+w tan+ and y+w tan backgrounds. We tested whether CG12120 expression under the control of the ubiquitous driver C765-GAL4 [18] in a tan1 background could rescue the pigmentation phenotype. C765-GAL4-mediated CG12120 expression is sufficient to rescue the tan1 pigmentation phenotype in the adult epidermis (Figure 1G and 1H). Similar rescue was seen with C765-GAL4 driving UAS-CG12120 in a tan5 background, as well as with patched-GAL4; UAS-CG12120 in both tan1 and tan5 backgrounds (data not shown).
Ectopic CG12120 Expression Produces Ectopic Melanin Pigmentation
We predicted that ectopic tan expression should result in ectopic melanin patterns by reversing the melanin-inhibiting role of ebony and thus providing dopamine for melanin synthesis [6–8]. To test this prediction, we drove UAS-CG12120 expression with a panel of GAL4 lines (complete dataset available upon request). A typical result is shown in Figure 1I and 1J. pannier-GAL4; UAS-CG12120 caused a dark stripe of ectopic melanin to appear in the pannier pattern along the dorsal midline, especially evident in the notum (arrowheads in Figure 1I an 1J). This phenotype is similar to driving ectopic yellow expression in an ebony mutant background [6].
CG12120 Is Expressed in the Optic Lobes of Adult Flies
In order to verify that tan/CG12120 is expressed in the Drosophila visual system we performed in situ hybridization experiments using a digoxigenin-labeled CG12120 cDNA antisense RNA probe (Figure 2). Head cryosections clearly revealed labeling in the retina. In horizontal sections the label appeared in thread-like structures resembling photoreceptor cell expression (Figure 2A). In sections that cut the retina in a sagittal plane, label showed the typical ommatidial structure of the retina (Figure 2B). Light areas resembling the ommatidial cavity were surrounded by dark label constituting a ring-like structure composed of photoreceptor cell somata. This pattern clearly indicates that the photoreceptor cell expression of tan differs from, and is complementary to, the lamina epithelial glial expression of ebony (see [5]).
Figure 2 Tan Expression and Function in the Eye and Optic Lobes
(A) RNA in situ hybridization with an antisense tan cDNA probe to wild-type adult head sections. Horizontal section cut through the eye and the optic lobes.
(B) Sagittal section through the ommatidial array of the eye.
(C) Control sense probe does not show significant staining (horizontal section).
(D–G) ERG phenotypes of wild-type flies (D) and tan mutants (E–G). (D) CantonS (wild-type) ERG showing normal “on” and “off” transients. Upper trace in each recording indicates duration of light input stimulus (see Materials and Methods). (E) tan2 mutant showing complete absence of “on” and “off” transients. (F) P{d07784} exhibits a normal “on” and “off” transient even though it has a tan-like pigmentation phenotype. (G) P{g1557} lacks “on” and “off” transients.
(H and I) Normalized “on” (H) and “off” (I) transients for wild-type, tan2 (t[2]), tan-like P excision lines, and “revertant” P excision lines (see text) expressed as a percentage of the respective sustained negative potential. Insets: Correlation between head histamine content and the normalized magnitude of ERG ‘on’ and ‘off’ transients (see text). Error bars: ± 1 standard error in both dimensions.
(J) Head histamine contents for tan-like and revertant P excision lines (see text), relative to control w1118 and tan1w1118 flies. Head histamine increased in all flies that drank 0.5% carcinine in 4% glucose (black bars), relative to controls that drank only 4% glucose (open bars). Revertant and tan-like excision lines are each arranged in rank order. Note that excision line flies that drank only 4% glucose have histamine contents too small to show above the abscissa. Error bars represent ± 1 standard error.
La, lamina; Me, medulla; Re, retina. Scale bars = 50 μm.
tan Mutants Exhibit Reduced ERG Transients
Like ebony mutants, tan flies lack normal “on” and “off” transients in their ERG ([9]; Figure 2E and 2G). In order to more thoroughly examine the association between tan pigmentation and ERG phenotypes, ERGs were recorded and the lamina “on” and “off” transients measured for the two tan P insertion lines, as well as for four tan-like and four “revertant” excisions of the P{d07784} insertion (see Materials and Methods). P{d07784} (Figure 2F) exhibited normal “on” and “off” ERG transients (compare with CantonS; Figure 2D). However P{g1557} (Figure 2G) lacked both “on” and “off” transients, somewhat similar to tan2 (Figure 2E). For three of four tan-like P excision lines, normalized “on” transients (Figure 2H) reached values of only 0%–5.6%, compared with 23.3% for wild-type eyes, from which they differed significantly (p ≤ 0.001, t-test). ERG “off” transients for these same three lines were between 0.2% and 3.9%, compared with a wild-type value of 63.3%, also significantly different (p < 0.001). Excision line 27A exhibited normal “on” transients, and markedly reduced, but not absent, “off” transients (Figure 2I). The amplitudes of these transients reached 27.7% (p >> 0.05) and 39.3% (p < 0.05), respectively, thus revealing partial rescue of the wild-type phenotype.
ERG Transients Are Restored in Revertant P Excision Lines
Flies from three of four revertant P excision lines exhibited normalized transients comparable in amplitude with wild-type (Figure 2H and 2I), line 017B reaching 21.8% (p >> 0.05), line 038B 22.3% (p >> 0.05), and line 051C 19.7% (p = 0.034) for “on,” compared with a wild-type CantonS value of 23.3% (Figure 2H). The normalized amplitudes of the “off” transients reached 48.3% in 017B (p = 0.032), while in 038B and 051C the values were 56.6% and 54.5%, respectively (both p >> 0.05), compared with a wild-type transient value of 63.3% (Figure 2I). Line 019A showed a less pronounced rescue effect: the “on” transient was restored, with a value of 16.2% (p > 0.05) (Figure 2H), while the “off” transient value of 15.3% remained significantly less than the wild-type level (p < 0.05) (Figure 2I).
tan-Like P Excision Lines Have Reduced Head Histamine and Impaired Conversion of Carcinine to Histamine
Flies from the four tan-like P excision lines (20A, 37C, 42A, and 27A) were given either 4% glucose to drink or 4% glucose laced with 0.5% carcinine. After drinking 4% glucose, all lines except 27A had reduced head histamine contents compared with corresponding w1118 control flies (Figure 2J). The reductions were to between 0.7% (37C) and 7.3% (20A) of w1118 histamine levels. Line 27A had, by contrast, 13% more head histamine than w1118. After the flies drank glucose plus carcinine, the head histamine contents were much larger than in flies that drank only glucose. These differences were significant for all excision lines (p < 0.05, t-test). The increases were not in proportion to the original head histamine content. The accumulated histamine levels after carcinine feeding in all P excision lines were far less than for the corresponding control w1118 flies, in which the differences were 2.15 times greater than in the tan-like excision line 27A. The differences between all excision alleles and w1118 were significant (p < 0.001, t-test) in carcinine-fed flies. However, there was no significant difference in head histamine contents between the excision lines 20A, 37C, and 42A and flies from a control w1118
tan1 double-mutant line that also drank a 0.5 % carcinine solution (p > 0.005). After ingesting carcinine, w1118 flies had a histamine head content 47 times greater than the w1118
tan1 controls, confirming the defective ability of flies mutant for tan to liberate histamine from exogenous carcinine.
Revertant P Excision Lines Exhibit Partially Restored Head Histamine Levels and Conversion of Carcinine to Histamine
Flies from the four revertant P excision lines (019A, 051C, 038B, and 017B) were also fed carcinine and control solutions. Control head histamine contents in flies that drank a 4.0% glucose solution were either similar to (019A and 051C: p > 0.05, t-test) or even slightly greater than (017B and 038B: p < 0.005) those in w1118 control flies (Figure 2J). After drinking carcinine, the head histamine increased in all revertant lines. The increases were significant compared with controls that did not drink carcinine (t-test, p < 0.005). Compared with w1118 control flies that also drank carcinine, the increases in head histamine were less, by between 16% (019A) and 63% (017B) of w1118 levels. The rank order in head histamine increases was roughly the same as the rank-ordered original head histamine contents in control flies before drinking carcinine. Thus, fly lines in which tan function was rescued most completely with respect to control head histamine content also had the largest histamine increases after drinking carcinine.
Correlation of Total Histamine Content and Normalized ERG Transients
Values of normalized transients were plotted against the total amount of histamine per head for the corresponding tan-like and revertant excision lines (insets, Figure 2H, I). The relationship between the size of the transients and the histamine content was approximately linear for both line types, supported by high regression coefficients (0.87 and 0.93 for tan-like and revertant excisions, respectively). Thus, in a general way, the more head histamine made available by tan function, the larger the ERG transients generated by the release of histamine during transmission in the lamina. The correlation coefficients for both relationships were greater than 0.87. The “off” transients were more sensitive to head histamine than were the “on” transients. Values fit a linear relationships having the equations y = 21.31x
ON − 2.96 (R
2 = 0.87) for the “on” transients and y = 51.13x
OFF − 13.92 (R
2 = 0.93) for the “off” transients, where y represents histamine content and x
ON and x
OFF represent the normalized “on” and “off” transients, respectively. This difference supports the separable origins of the transients [19]. The relationship takes no account of the histamine located outside the visual system, of which there are several sources [20], which may explain the residual histamine content (0.2–0.3 ng/head) when the transients were zero.
The CG12120 Protein Is Conserved among Insects and Is Related to Fungal IAT
CG12120 encodes a 387-amino-acid polypeptide with a predicted molecular weight of approximately 44 kD. CG12120 homologs are present in all sequenced Drosophila genomes, as well as in the genomes of Anopheles gambiae and Bombyx mori. Sequence identity among insect CG12120 proteins extends over the entire length of the protein for the two Diptera (79.8% identical between D. melanogaster and A. gambiae;
Figure 3A and 3B) and the first 230–240 amino acids between dipteran and Bombyx mori sequences (45%–48% identical between dipterans and B. mori;
Figure 3C). The B. mori sequence may be an incomplete fragment, to be clarified pending the release of an annotated version of the genome sequence. Interestingly, several fungal IATs (Pencillium chrysogenum shown in Figure 3A and 3B) are approximately 50% similar (based on Gonnet series in CLUSTALW [21]) and 20% identical to insect CG12120 proteins, suggesting the conservation of a very ancient gene present in the common ancestor of fungi and metazoans. Fungal IATs are one of three enzymes in the penicillin biosynthetic pathway and catalyze the substitution of the L-α-aminoadipyl side chain of isopenicillin-N with aromatic acyl side chains [22]. A BLAST search of P. chrysogenum IAT to the D. melanogaster genome turned up two other proteins, but these have less substantial similarity to IAT: CG12140, a predicted electron-transferring flavoprotein dehydrogenase (25% identical, 38% similar over a 148-amino-acid region from residues 90–228 of IAT), and CG8864, a predicted monooxygenase/oxidoreductase electron transporter (29% identical, 38% similar over a 78-amino-acid region from residues 224–289 of IAT). Therefore, it appears that CG12120 is the only protein in the Drosophila genome with strong homology to fungal IATs.
Figure 3
Drosophila CG12120 Is Conserved across Insects and Is Related to Fungal IAT
(A) Alignment of D. melanogaster, D. pseudoobscura, and A. gambiae CG12120 orthologs and P. chrysogenum IAT protein. Residues in blue are identical (*) across all four species, and residues in green are functionally similar, showing strong (:) or weak (.) similarity. Residue highlighted “1” indicates conserved arginine residue at position 217 that is mutated to proline in tan1 and tan4 mutants. Residue highlighted “2” indicates methionine residue at position 256 that is mutated to isoleucine in tan5 mutant. Cyan rectangle indicates auto-processing site of P. chrysogenum at which pro-IAT is cleaved between glycine 102 and cysteine 103.
(B) Percent sequence identity (blue in upper right) and sequence similarity (red in lower left) among Drosophila spp., A. gambiae, and P. chrysogenum proteins.
(C) Percent sequence identity (blue in upper right) and sequence similarity (red in lower left) of N-terminal Drosophila spp., A. gambiae, and P. chrysogenum proteins with presumptive B. mori CG12120 ortholog (see text).
The presence of completely conserved sites between metazoans and fungi through virtually the entire length of the CG12120/IAT proteins strongly suggests that the molecular mechanism underlying enzyme activity has been conserved, even though these proteins function in very different pathways in fungi and metazoans and in two different functional pathways even within insects. One site of particular interest is the conserved glycine–cysteine domain at position 102–103, which is the site of autocatalytic processing of fungal IATs [23]. Conservation of this domain among insects suggests that autocatalytic cleavage may also be present in CG12120.
The CG12120 amino acid sequence was highly conserved between wild-type flies and the five classical tan mutants that were sequenced. Only two amino acid substitutions were found. In tan1 and tan4, a highly conserved arginine is changed to proline at position 217 (Figure 3A). Since these two mutants did not exhibit extremely different levels of CG12120 transcript from wild-type (data not shown), it is likely that this substitution of an evolutionarily conserved amino acid is functionally important and responsible for the tan phenotype. One other substitution, a methionine to isoleucine substitution at position 256 in tan5, a region of little sequence conservation except among Drosophila species, is not predicted to cause an extreme functional change in the protein. The tan5 allele may represent a more profound disruption of the CG12120 locus, given that PCR amplification of the 3′ end of this allele, including the last two exons, was not successful in several attempts.
CG12120 Possesses Both NBAD Hydrolase and N-β-Alanyl Histamine Hydrolase Activity
The tan gene product is predicted to encode a multifunctional hydrolase that catalyzes the hydrolysis of NBAD into β-alanine and dopamine, and the hydrolysis of carcinine (N-β-alanyl histamine) into β-alanine and histamine, respectively. To test whether CG12120 possesses these predicted activities, we produced recombinant CG12120 protein using a baculovirus expression system in insect cell culture (Figure 4). After soluble proteins from either uninfected Spodoptera frugiperda (Sf9) insect cells or Sf9 cells infected with an alanine glyoxylate transaminase (AGT) recombinant baculovirus were mixed into a NBAD solution. Production of dopamine in the reaction mixture was not observed (Figure 4A), suggesting that Sf9 cells do not have a protein capable of mediating NBAD hydrolysis, and infection of baculovirus itself also did not stimulate the production of a protein with NBAD-hydrolyzing activity (data not shown). In contrast, when soluble proteins from CG12120 recombinant baculovirus-infected cells were mixed into a NBAD solution, accumulation of dopamine in the reaction mixture was observed and the amounts of dopamine produced in the reaction mixture were approximately proportional to the applied incubation periods (Figure 4B and 4C). During hydrolysis, an equal amount of β-alanine was produced in the reaction mixture (data not shown), but could not be detected electrochemically, because β-alanine is not electrochemically active.
Figure 4 The CG12120 Gene Product Possesses NBAD Hydrolase and N-β-Alanyl Histamine (Carcinine) Hydrolase Activities
HPLC-ED chromatograms from activity assays.
(A) Lysate from control (recombinant AGT) transfected Sf9 cells exhibited no activity. Oxidation product (OP) peak denotes uncharacterized presumptive oxidation product of NBAD present in NBAD substrate in all experiments.
(B) NBAD incubated with CG12120 protein purified from Sf9 cells expressing CG12120 baculovirus construct after 20 min incubation with NBAD. Dopamine (DA) peak is evident.
(C) Same experiment as in (B) but after 50 min incubation, showing increased accumulation of dopamine and depletion of NBAD.
(D) Control HPLC chromatogram containing dopamine and NBAD standards.
(E) Extract from control (recombinant AGT) transfected Sf9 cells exhibits no production of β-alanine and histamine when incubated with carcinine (CA). Unknown contaminant (UC) peak denotes an unknown contaminant in carcinine. β-mercaptoethanol (BME) peak denotes β-mercaptoethanol present in the reaction solution.
(F) Extract from Sf9 cells transfected with CG12120 baculovirus incubated with carcinine demonstrates production of histamine (HA) and β-alanine (BA). Under the applied conditions, essentially all carcinine was hydrolyzed to histamine and β-alanine.
(G) Control HPLC chromatogram containing β-mercaptoethanol, carcinine, histamine, and β-alanine standards. β-alanine and histamine detection were enabled by OPT conjugation (see Materials and Methods).
A similar assay was used to examine carcinine hydrolysis. In this case, detection of β-alanine and histamine products was enabled by o-phthaldialdehyde thiol (OPT) conjugation (see Materials and Methods). After proteins from uninfected (not shown) or AGT recombinant baculovirus-infected insect cells were mixed into a solution of carcinine, hydrolysis of the carcinine was not observed (Figure 4E). However, after soluble proteins from CG12120 recombinant baculovirus-infected cells were mixed into a carcinine solution, rapid accumulation of β-alanine and histamine was indeed observed in the reaction mixture (Figure 4F). Hydrolysis of both NBAD and carcinine by soluble proteins from CG12120 recombinant baculovirus-infected insect cells provides direct and convincing evidence for the NBAD and carcinine hydrolase identity of the CG12120 protein.
Discussion
The molecular identity of tan has been a longstanding question critical to the biology of melanin pigmentation and synaptic transmission at photoreceptors in insects. We have provided genetic, developmental, and biochemical evidence that the predicted D. melanogaster gene CG12120 encodes tan. P insertions in the first exon of CG12120 are associated with tan-like phenotypes and fail to complement classical tan mutants. Reduction of CG12120 transcript levels is correlated with tan pigmentation, ERG, carcinine hydrolysis, and histamine phenotypes. Ectopic expression of CG12120 in transgenic D. melanogaster rescues the tan pigmentation phenotype in tan mutant backgrounds and causes ectopic melanization, analogous to loss of ebony function, in wild-type backgrounds. The CG12120 protein exhibits the two predicted enzyme activities, NBAD hydrolase and carcinine hydrolase, in vitro. Taken together, our data demonstrate that CG12120 is tan.
The Tan–Ebony Circuit Occupies a Pivotal Position in Melanin Biosynthesis
The molecular identification of tan helps clarify a crucial step in dopamine metabolism and melanin biosynthesis in epidermal cells. All developing adult epidermal cells in insects are capable of secreting catecholamine precursors of melanin and sclerotin, and current models [1,6,24] propose that the patterns of adult melanin reflect the differential spatial regulation of four parallel branches from the core dopamine pathway catalyzed by tyrosine hydroxylase and dopa decarboxylase (Figure 5A). One of the four branches produces dopa melanin, which is under the control of yellow [6,25], the exact function of which is unknown, and at least two Yellow-related proteins, Yellow-f and Yellow-f2, which convert dopachrome to 5,6-dihydroxyindole [26]. Dopamine is also secreted and converted into dopamine melanin through an as yet uncharacterized pathway. Areas of the cuticle that are not melanized secrete NBAD, produced by the action of the Ebony protein [2,6], resulting in yellow or light tan cuticle, or N-acetyl dopamine, produced by the action of the arylalkylamine N-acyltransferases [27], which results in transparent cuticle (J. R. T., unpublished data). All of these precursors are extracellularly polymerized and crosslinked to cuticle proteins, probably through the action of a common set of enzymes, including phenol oxidases [28], the functions of which in the developing cuticle are not well characterized. Tyrosine and catecholamines are also provided to some degree from the hemolymph [1,29], and a hemolymph supply of melanin precursors is required for wing pigmentation [24].
Figure 5 Tan Functions in Diverse Developmental and Metabolic Pathways
(A) In developing adult epidermal cells, Tan catalyzes the production of dopamine from NBAD during pigment development. This is one of four parallel pathways by which dopa or dopamine derivatives are secreted into the developing cuticle as precursors for distinct pigments. aaNAT, arylalkylamine-N-acetyl transferase; DDC, dopa decarboxylase; NADA, N-acetyl dopamine; PO, phenol oxidase; TH, tyrosine hydroxylase.
(B) In the photoreceptor, Tan catalyzes the hydrolysis of N-β-alanyl histamine (carcinine) to histamine for re-uptake by the presynaptic photoreceptor cell (R). CA, carcinine; EG, epithelial glial cell; HA, histamine.
Normal melanization depends in part on Tan function to provide dopamine by hydrolyzing sequestered NBAD. It is currently unclear why this dopamine is produced from NBAD rather than directly from dopa by dopa decarboxylase. One possible explanation for an Ebony–Tan “shunt” would be if epidermal cells require rapid or precise temporal regulation of dopamine secretion during cuticle development. For example, long-term sequestration of dopamine awaiting this developmental time window could be injurious to the cell. Alternatively, conversion of dopamine to NBAD by Ebony may be a constitutive ancestral state in insects, and conversion of some of this NBAD back to dopamine for melanin production may be a derived condition in some insects. NBAD synthase activity has been demonstrated in lepidopterans [30], in which NBAD is a precursor to yellow papiliochrome pigment. Isolation and functional characterization of tan and ebony gene homologs from more basal insects will be needed to test these alternative hypotheses.
The production of dopamine melanin depends on Tan function, which in turn depends on Ebony to produce its substrate. As predicted by this relationship, ebony is epistatic to tan (J. R. T., unpublished data). Production of melanin from both dopa and dopamine is an apparent degeneracy that occurs in insects but not vertebrates, which produce melanin primarily from L-dopa [31]. The final dark black color of many insects reflects contributions of both types of melanin, which continuously darken during cuticle maturation and hardening. There is evidence in D. melanogaster that the two melanin pathways are not independent. The presence of Ebony appears to determine whether melanin will be produced, even in the presence of ectopic Yellow, which gains access to the core dopamine pathway upstream at the dopa stage. Only in the absence of Ebony function is ectopic Yellow able to promote ectopic melanin production [6]. This suggests that normally most dopa is converted to dopamine and then to NBAD (or N-acetyl dopamine), but when the dopamine-to-NBAD step is blocked in an ebony mutant more dopa may be available for Yellow-mediated conversion to dopa melanin, possibly because of product inhibition of dopa decarboxylase [32]. Note that back-conversion of dopamine to dopa has not been observed in insects [1]. ebony mutants accumulate excess levels of dopamine, which is shunted to dopamine melanin. This mechanism has long been a candidate for naturally occurring melanism, which is an extremely common type of polymorphism in insects [7,33]. Thus, ebony itself is a candidate gene for such polymorphisms. However, D. melanogaster ebony mutants do not show the complete dominance typical of naturally occurring melanic alleles in other insects. Another important candidate is tan, which we have demonstrated here is mutable, via gain of function, to dominant production of ectopic melanin.
Role of tan in Insect Vision
Compared with melanin biosynthesis, less can be said of tan's involvement in histamine metabolism, but our findings do help to explain the action of tan, and thus clarify the Ebony–Tan pathway, in the visual system. The essential feature is that Tan localizes to the photoreceptor cells, and thus presumably their synaptic terminals, in a complementary position to that of Ebony, which localizes to the surrounding glial cells. This result helps explain early mosaic studies indicating that tan acts autonomously either within or very close to the eye [34]. Thus, histamine released from the photoreceptor terminals must apparently enter the epithelial glia and become converted to carcinine by Ebony, and the carcinine must return to the photoreceptor terminal, where hydrolysis can liberate histamine (Figure 5B). Uptake mechanisms and pathways are unknown, not only for histamine and carcinine, but also for the β-alanine co-liberated from carcinine by photoreceptor Tan, and identification of these now constitutes a next line of inquiry.
Histamine liberated in the photoreceptor terminal is presumed to finally become available for pumping into newly endocytosed synaptic vesicles. The site of the latter function is now clear: in the lamina, endocytotic recovery of new synaptic vesicles is localized to the stalk region of capitate projections [35], invaginations of the photoreceptor terminal, from epithelial glia [36]. Mutant tan1 flies have significantly more penetrating capitate projections than wild-type flies [37], a phenotype that has been used to suggest that the capitate projection is an integrated recycling organelle site for the endocytotic retrieval of membrane and the recycling of histamine [35]. Mutant tan flies have been suggested to lack ERG transients because they have an insufficient pool of photoreceptor histamine to release [3], but the differential action we report in tan alleles for the “on” and “off” transients suggests that this may at best be a partial explanation.
Tan Is the First NBAD/Carcinine Hydrolase Characterized in Any Organism
Like Ebony [2], Tan is related to a microbial protein, in this case an enzyme involved in penicillin synthesis. Given their close reciprocal functions in dopamine and histamine metabolism, it is tempting to speculate that ebony and tan are evolutionarily ancient molecular partners in insects, and possibly in arthropods or even ecdysozoans. If this is the case, then the two genes may have been co-opted together or consecutively from a microbial genome, perhaps at the base of the metazoan lineage. Alternatively, tan and ebony could have descended from genes present in the common ancestor of fungi and metazoans, in which case they have apparently been lost in the chordate or deuterostome lineage. There are no proteins related in sequence to Tan and Ebony in any vertebrate genome sequenced to date, and NBAD has not been found in vertebrates. Carcinine was originally characterized in the crab Carcinus maenas [38], but its biological function until recently has been a mystery. Carcinine possesses several pharmacological properties in mammals, including antioxidant effects [39] and cardiac vasodilation [40], but an endogenous role for this compound has not been demonstrated in any deuterostome.
Characterization of the structure and function of the Tan/CG12120 protein, including the conserved putative autoprocessing site at positions 102–103 and the conserved arginine residue that is mutated to a proline in tan1 and tan4 mutants, will help reveal any possible conservation of function of this protein in insects and fungi. As further arthropod and protostome genomes are sequenced, the presence or absence and sequence evolution of tan/CG12120 homologs will help indicate whether other invertebrate species utilize this protein for melanin production and/or neurotransmitter metabolism. An intriguing possibility in insects is that coevolution of pigment patterns and behaviors [7] may have involved tan and other genes that have pleiotropic actions in both pigmentation and the nervous system.
Conclusions
We provide genetic, developmental, neurophysiological, and biochemical evidence that D. melanogaster tan corresponds to the predicted gene CG12120. tan encodes a multifunctional enzyme that hydrolyzes both NBAD to dopamine and carcinine (N-β-alanyl histamine) to histamine. In this study, this enzyme is characterized at the molecular level for the first time, to our knowledge, in any organism. Confirmation of these two enzyme activities of the Tan protein provides important clarification of the pleiotropic function of this gene in pigment development and in histamine metabolism at the photoreceptor organ. tan is also an important candidate gene for melanin pattern polymorphisms and species differences. Further study of Tan function in photoreceptor neurons will help clarify how transmitter released by photoreceptors is recovered for reuse during insect vision.
Materials and Methods
Drosophila strains and culture.
All D. melanogaster crosses were performed at 25 °C (23 °C for histamine assays) with a 12 h light:12 h dark cycle. Flies were cultured on standard corn meal/molasses/agar medium.
C765-GAL4 [18] was provided by S. Carroll (University of Wisconsin, Madison, Wisconsin, United States). pannier-GAL4 [41] was provided by G. Morata (Universidad Autonoma de Madrid, Madrid, Spain). CantonS, OregonR, w1118; +; P{Δ2–3}, TM3, Sb/TM6B, and Pro,os/FM6,B,w were provided by W. Eanes (State University of New York, Stony Brook, New York, United States). w1118 P{Mae-UAS.6.11} CG12120[g1557](P{g1557}) [16] was provided by U. Schäfer (Max-Planck-Institut für Biophysikalische Chemie, Göttingen, Germany). The following lines were provided by the Bloomington Drosophila StockCenter (http://flystocks.bio.indiana.edu/): w1118, P{w[+mC]=XP} CG12120[d07784] (P{d07784}), w1118, t 1, t 2 v 1 f 1 (t 2), t 3, br 1 w e ec 1 rb 1 t 4/FM1 lz 4 (t 4), t 5 v 1 r 1/FM7c (t 5), P{EPgy2}EY04394, and P{EPgy2}EY05996.
All flies were examined at 3–5 d of age, after the full adult pigmentation appeared. UAS-CG12120 rescue and ectopic expression genotypes showed very little variation. Rescue and ectopic expression phenotypes occurred in 100% of the flies that inherited both the GAL4 and UAS-CG12120 elements.
P excision lines.
w1118, P{XP}CG12120 d07784 females were crossed to w1118; +; P{Δ2–3}, TM3, Sb/TM6B males, and dysgenic male F1s were crossed to Pro,os/FM6,B,w females. Female F2 progeny lacking eye color (w1118, P*/FM6 B w; +; +) were crossed to FM6,B,w/ Y males, and their non-FM6 male progeny were inspected for pigmentation phenotype. Four tan-like lines and four “revertant” lines with wild-type pigmentation were chosen for further analysis.
The four excisions classified as tan-like all contained imprecise excisions, two of which, 20A and 37C, contained large deletions (953 bp and 1,641 bp, respectively) that included the presumptive promoter region. The other two tan-like excisions, 27A and 42A, left small insertions (38 bp and 77 bp, respectively) at the P-element site (data available upon request). All four revertant P excisions were also imprecise, leaving P-element fragments ranging from 12 bp to 1.2 kb in size (data available upon request), but none of these contained deletions of the endogenous CG12120 transcription unit or promoter region.
The tan-like excision lines showed significantly lower CG12120 mRNA levels on average than wild-type lines by quantitative reverse transcriptase PCR assay (data not shown). Revertant excision lines showed no differences on average from wild-type lines in mean CG12120 mRNA expression levels. These results were found at two different developmental stages, 60–75 h after puparium formation and 0–8 h after eclosion. The complete CG12120 mRNA expression dataset is available upon request.
Plasmid constructs.
For UAS-CG12120, the complete CG12120 cDNA was obtained from the Drosophila Genomics Resource Center (http://dgrc.cgb.indiana.edu/) as clone RH41996 (barcode 17763), consisting of the 2,009-bp CG12120 cDNA in plasmid vector pFLC-1. A 2,133-bp NotI-Acc65I fragment containing the CG12120 cDNA was cloned into the pUAST vector [17] to produce the UAS-CG12120 construct. This construct was used to transform a D. melanogaster yw host strain as described [42], and transformant lines were homozygosed, mapped, and crossed into a y+w background for tan rescue and ectopic CG12120 expression experiments.
For the CG12120 baculovirus expression construct, a 1,568-bp XbaI-BstBI fragment, containing the complete 1,164-bp predicted CG12120 ORF, was cloned from RH41996 into the pBlueBac 4.5 baculovirus expression vector (Invitrogen, Carlsbad, California, United States). Then, in order to place the start codon as close as possible to the pBlueBac polyhedrin promoter, the CG12120 cDNA was amplified from the pBlueBac 4.5–CG12120 clone using a forward primer (5′-GCT
AGC
ATG TCC TCC TTA AAG ATC CTG-3′) containing a NheI restriction site (underlined), and a reverse primer (5′-AAG
CTT CTA CTT GTA GAG CAG CGG CAG-3′) containing a HindIII restriction site (underlined). The start codon of the CG12120 ORF is indicated in bold in the forward primer. The amplified DNA fragment was inserted into a PCR2.1-TOPO TA cloning vector and then cloned into pBlueBac 4.5 between the NheI and HindIII restriction sites. The recombinant transfer vectors were sequenced and verified to ensure that the inserted genes were in-frame and controlled under the downstream polyhedrin promoter. Recombinant NP 572543 (CG12120) pBlueBac 4.5 transfer vector was cotransfected with linearized Bac-N-Blue (AcMNPV, Autographa californica multiple nuclear polyhedrosis virus) viral DNA to Sf 9 insect cells (Invitrogen) to generate recombinant CG12120 baculovirus. The recombinant baculovirus was purified by the plaque assay procedure.
Quantitative RT-PCR.
RNA was isolated from 20–30 individuals sorted by sex from two stages: pooled P8–P11 stage pupae (roughly 60–75 h after puparium formation, when eye color is present but macrochaetes and body cuticle are not yet pigmented [43]), and adults 0–8 h after eclosion. RNA isolation used a Stratagene (La Jolla, California, United States) Absolutely RNA Microprep kit and protocol. RNA yields were quantified by OD260 reading on a spectrophotometer. For each RT-PCR reaction, 300 ng was loaded. One-step quantitative RT-PCR analysis used the SYBR-green-based reagents and protocols in the Stratagene One-Step Brilliant QRT-PCR kit. The following RT-PCR products were quantified: for CG12120, a 117-bp fragment from position 327 to 443 of the CG12120 cDNA amplified using forward primer 5′-CTT CGA TAT GGG TCG CAC ATT TGG-3′ and reverse primer 5′-TTG TAG ATC TGC CTG CCT TTT GG-3′; for α-GPDH (CG9042), a 187-bp fragment from position 1210 to 1396 of the GPDH-RC cDNA [15] using forward primer 5′-ATC TGA TCA CGA CGT GTT ACG GTG-3′ and reverse primer 5′-AAC AGG GGG AAT TTG TCC TCC AGA C-3′; and for RPII-215 (CG1554), a 211-bp fragment from position 2674 to 2884 of the RPII-215 cDNA using forward primer 5′-TAT CCC AGG TTA TTG CTT GTG TGG G-3′ and reverse primer 5′-GCA GTA TCG ATA AGA CCT TCA CGA CC-3′. Quantitative RT-PCR was performed and product yields were quantified on a Stratagene MX3000P real-time thermal cycler connected to a Gateway PC laptop computer running version 2.0 of the MX3000P software. The following amplification procedure was used: 50 °C for 20 min (RT step), 95 °C for 15 min, 40 amplification cycles (94 °C for 30 s, 53 °C for 30 s, and 72 °C for 30 s), and 79 °C for 11 s (fluorescence reading), followed by melting curve analysis to confirm expected product T
m. Controls without reverse transcriptase were run to estimate background signal, if any, due to amplification from genomic DNA. These background amounts were generally three to four orders of magnitude below experimental reactions and were subtracted from yields of experimental reactions prior to the calculation of CG12120/RPII-215 and CG12120/GPDH ratios.
ERG recordings.
Flies were immobilized in cut-off pipette tips with the head protruding from the opening. The indifferent and recording electrodes, filled with Drosophila Ringer's solution, were inserted into the posterior part of the head capsule and placed on the surface of the retina, respectively. After a stable baseline was obtained, the light impulse was triggered by removing a shutter from a Schott KL 150 B light source with a Xenophot 150-W halogen photo optic lamp (Osram, Augsburg, Germany), thus directing the light beam onto the fly's eye. The stimulus lasted for 1 s, and this cycle was repeated ten times for each individual fly after allowing the signal to return to baseline. Potentials were recorded over a 5-s time frame by a Hameg Instruments (Mainhausen, Germany) digital oscilloscope and stored with Hameg SP107 software.
ERG plots for individual flies were obtained by calculating the mean of the ten light/dark cycles. To normalize the transients, the sustained negative potential (which reports the photoreceptor response that drives transmission in the lamina, represented by the transients [10,44]), was determined from the averaged potential 20 ms prior to the offset of the light stimulus, and the relative size of the transients was then calculated as a percentage of this sustained negative potential.
In situ hybridization to RNA.
For in situ hybridization, fly heads were mounted in Tissue-Tek O.C.T. compound (Microm, Walldorf, Germany) and were shock-frozen in liquid nitrogen. Sections of 10 μm were cut and fixed with 4% paraformaldehyde. After acetylation and prehybridization, subsequent hybridization with a digoxigenin-labeled RNA CG12120-cDNA probe was performed overnight at 55 °C. Specimens were blocked with normal goat serum in Tris-buffer saline/0.1% Triton X-100 and were then treated with an alkaline phosphatase-coupled anti-digoxigenin antiserum (1:1,000 dilution in Tris-buffer saline/0.1% Triton X-100). NBT/BCIP color development was controlled under the microscope for 20–60 min.
Histamine assays.
Flies for histamine determinations were aged for at least 3 d prior to preparation for high performance liquid chromatography (HPLC) to ensure the completion of the critical period for lamina development [45], when histamine content has stabilized (J. Borycz and I. A. M., unpublished data). To determine Tan function by carcinine conversion to histamine, flies were dehydrated for 2 h prior to feeding with aqueous solutions of 4% glucose or 0.5% carcinine in glucose, and then allowed to drink from these solutions overnight for 16 h. Flies were collected 2 h after lights on, then rapidly frozen and stored at −80 °C until assayed by HPLC. To compare head histamine with the reduction in lamina ERG transients, head histamine was determined from flies that were fed on medium only.
In all cases, histamine determinations were performed on groups of 20–50 isolated heads using HPLC with electrochemical detection as reported for D. melanogaster [3,13,46], and values are reported for the means of 3–12 such samples.
Recombinant CG12120 expression.
Sf 9 cells were cultured in 25-cm culture flasks in the presence of 10 ml of TNM-FH medium containing 10% fetal bovine serum (Invitrogen). High titer recombinant CG12120 baculovirus was inoculated into the cell culture at 70% confluence. Sf 9 cells were harvested 3 d post-inoculation by centrifugation (800 g for 15 min at 4 °C). The pellet cells were washed twice with PBS. Cells were lysed by sonication in 50 mM phosphate buffer. The lysate was centrifuged at 20,000 g for 20 min to obtain supernatant that was subsequently used for NBAD and carcinine hydrolyase activity assays.
Enzyme activity assays.
Cell lysate supernatant from recombinant tan baculovirus-infected cells was assayed for NBAD and carcinine hydrolase activities. Cell lysate supernatant from AGT baculovirus-infected cells served as a control. A typical reaction mixture consisted of 50 μl of cell lysate supernatant, 100 μl of phosphate buffer (50 mM [pH 7.0]), and 50 μl of 8 mM NBAD (provided by the National Institute of Mental Health Chemical Synthesis and Drug Supply Program; http://nimh-repository.rti.org/) or N-β-alanyl histamine (a kind gift of M. Feigel, Ruhr-Universität Bochum, Bochum, Germany), prepared in 50 mM phosphate buffer. At different incubation periods, 50 μl of reaction mixture was withdrawn and mixed with an equal volume of 0.8 M formic acid (NBAD reactions) or an equal volume of absolute ethanol (N-β-alanyl histamine reactions) to stop the enzymatic reaction. The formic-acid-treated reaction mixture was centrifuged at 20,000 g for 15 min at 4 °C, and supernatant was analyzed by HPLC with electrochemical detection. Hydrolase activity to NBAD was determined based on the detection of dopamine in the reaction mixture, and hydrolase activity to N-β-alanyl histamine was determined based on the detection of OPT β-alanine and OPT histamine conjugates [47].
Supporting Information
Accession Numbers
The GenBank (http://www.ncbi.nlm.nih.gov/Genbank/) accession numbers for the gene products discussed in this paper are A. gambiae CG12120 (AAAB01008846), B. mori CG12120 (BAAB070172112), D. melanogaster CG12120 (NM 132315), and D. pseudoobscura CG12120 (CH379064). The Swiss-Prot (http://www.ebi.ac.uk/swissprot/) accession number for P. chrysogenum IAT is P15802.
We thank N. Gompel, B. Prud'homme, and S. Carroll for technical help in producing D. melanogaster transgenic lines and S. Wagner for expert technical assistance on the in situ hybridization experiments. B. McOmber assisted in P-element screens for tan. P. Wittkopp, A. Kopp, B. Prud'homme, N. Gompel, and S. Carroll provided many stimulating discussions during the course of this work. P. Gergen, M. Kernan, D. Stoebel, B. Prud'homme, P. Wittkopp, T. Merritt, W. Eanes, T. F. C. Mackay, R. Geeta, G. Barsh, and three anonymous reviewers provided valuable comments during the preparation and revision of this manuscript. B. Allen and B. Leger provided invaluable statistical assistance. We are grateful to R. Yukilevich, W. Eanes, S. Carroll, G. Morata, U. Schäfer, A. Richardt, and the Bloomington Drosophila Stock Center for providing D. melanogaster strains. M. Feigel generously provided carcinine, and the National Institute of Mental Health Chemical Synthesis and Drug Supply Program provided NBAD for enzyme activity assays. JRT was funded during this project by State University of New York (SUNY) at Stony Brook. BTH was supported by Deutsche Forschungsgemeinschaft grant 714/9–4. IAM was supported by Canadian Institutes of Health Research Regional Partnership Program grant ROP-6740. JL was supported by National Institutes of Health grant AI 37789. Quantitative RT-PCR and DNA sequencing were performed in the National Science Foundation–funded MEAD Laboratory in the SUNY Stony Brook Department of Ecology and Evolution. This paper is number 1141 contributed by the Graduate Program of the Department of Ecology and Evolution at SUNY Stony Brook.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. JRT, BTH, IAM, and JL conceived and designed the experiments. JRT, SY, BTH, TK, TNE, SL, QH, and JL performed the experiments. JRT, SY, BTH, TK, IAM, TNE, and JL analyzed the data. JRT, BTH, IAM, and JL wrote the paper.
A previous version of this article appeared as an Early Online Release on October 15, 2005 (DOI: 10.1371/journal.pgen.0010063.eor).
Abbreviations
AGTalanine glyoxylate transaminase
ERGelectroretinogram
HPLChigh performance liquid chromatography
IATisopenicillin-N N-acyltransferase
NBADN-β-alanyl dopamine
OPTo-phthaldialdehyde thiol
Sf9
Spodoptera frugiperda
yw
yellow white
==== Refs
References
Wright TRF 1987 The genetics of biogenic amine metabolism, sclerotization, and melanization in Drosophila melanogaster
Adv Genet 24 127 222 3124532
Hovemann BT Ryseck RP Walldorf U Sortkuhl KF Dietzel ID 1998 The Drosophila ebony gene is closely related to microbial peptide synthetases and shows specific cuticle and nervous system expression Gene 221 1 9 9852943
Borycz J Borycz JA Loubani M Meinertzhagen IA 2002
tan and ebony genes regulate a novel pathway for transmitter metabolism at fly photoreceptor terminals J Neurosci 22 10549 10557 12486147
Richardt A Kemme T Wagner S Schwarzer D Marahiel MA 2003 Ebony, a novel nonribosomal peptide synthetase for beta-alanine conjugation with biogenic amines in Drosophila
J Biol Chem 278 41160 41166 12900414
Richardt A Rybak J Störtkuhl KF Meinertzhagen IA Hovemann BT 2002 Ebony protein in the Drosophila nervous system: Optic neuropile expression in glial cells J Comp Neurol 452 93 102 12205712
Wittkopp PJ True JR Carroll SB 2002 Reciprocal functions of the Drosophila Yellow and Ebony proteins in the development and evolution of pigment patterns Development 129 1849 1858 11934851
True JR 2003 Insect melanism: The molecules matter Trends Ecol Evol 18 640 647
Wittkopp PJ Carroll SB Kopp A 2003 Evolution in black and white: Genetic control of pigment patterns in Drosophila
Trends Genet 19 495 504 12957543
Hotta Y Benzer S 1969 Abnormal electroretinograms in visual mutants of Drosophila
Nature 222 354 356 5782111
Coombe PE 1986 The large monopolar cells L1 and L2 are responsible for ERG transients in Drosophila
J Comp Physiol 159 655 665
Laughlin SB de Ruyter van Steveninck RR 1996 Measurements of signal transfer and noise suggest a new model for graded transmission at an adapting retinal synapse Nature 379 642 645
Borycz JA Borycz J Kubów A Kostyleva R Meinertzhagen IA 2005 Histamine compartments of the Drosophila brain with an estimate of the quantum content at the photoreceptor synapse J Neurophysiol 93 1611 1619 15738275
Melzig J Burg M Gruhn M Pak WL Buchner E 1998 Selective histamine uptake rescues photo- and mechanoreceptor function of histidine decarboxylase-deficient Drosophilia mutant J Neurosci 18 7160 7166 9736639
Demerec M Kaufmann BP Fano U Sutton E Sansome ER 1942 The gene Carnegie Inst Yearb 41 190 199
Drysdale RA, Crosby MA, FlyBase Consortium 2005 FlyBase: Genes and gene models Nucleic Acids Res 33 D390 D395 15608223
Beinert N Werner M Dowe G Chung HR Jäckle H 2004 Systematic gene targeting on the X chromosome of Drosophila melanogaster
Chromosoma 113 271 275 15480728
Brand AH Perrimon N 1993 Targeted gene expression as a means of altering cell fates and generating dominant phenotypes Development 118 401 415 8223268
Gomez-Skarmeta JL Modolell J 1996
araucan and caupolican provide a link between compartment subdivisions and patterning of sensory organs and veins in the Drosophila wing Genes Dev 10 2935 2945 8918894
Rajaram S Nash HA 2004 A specific alteration in the electroretinogram of Drosophila melanogaster is induced by halothane and other volatile general anesthetics Anesth Analg 98 1705 1711 15155332
Pollack I Hofbauer A 1991 Histamine-like immunoreactivity in the visual system and brain of Drosophila melanogaster
Cell Tissue Res 266 391 398 1684918
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
Queener SW Neuss N 1982 The biosynthesis of beta-lactam antibiotics Morin EB Morgan M Beta-lactam antibiotics, Volume 3 London Academic Press 1 81
Barredo JL van Solingen P Diez B Alvarez E Cantoral JM 1989 Cloning and characterization of the acyl-coenzyme A: 6-aminopenicillanic-acid-acyltransferase gene of Penicillium chrysogenum
Gene 83 291 300 2555269
True JR Edwards KA Yamamoto D Carroll SB 1999
Drosophila wing melanin patterns form by vein-dependent elaboration of enzymatic prepatterns Curr Biol 9 1382 1391 10607562
Walter MF Black BC Afshar G Kermabon AY Wright TRF 1991 Temporal and spatial expression of the yellow gene in correlation with cuticle formation and DOPA decarboxylase activity in Drosophila development Dev Biol 147 32 45 1879614
Han Q Fang J Ding H Johnson JK Christensen BM 2002 Identification of Drosophila melanogaster yellow-f and yellow-f2 proteins as dopachrome-conversion enzymes Biochem J 368 333 340 12164780
Brodbeck D Amherd R Callaerts P Hintermann E Meyer UA 1998 Molecular and biochemical characterization of the aaNAT1 (Dat) locus in Drosophila melanogaster : Differential expression of two gene products DNA Cell Biol 17 621 633 9703021
Ashida M Yamazaki HI 1990 Biochemistry of the phenoloxidase system in insects: With special reference to its activation Ohnishi E Ishizaki H Molting and metamorphosis Tokyo Japan Scientific Societies Press 239 265
Hazel MH Ianowski JP Christensen RJ Maddrell SH O'Donnell MJ 2003 Amino acids modulate ion transport and fluid secretion by insect Malpighian tubules J Exp Biol 206 79 91 12456699
Koch PB Behnecke B Ffrench-Constant RH 2000 The molecular basis of melanism and mimicry in a swallowtail butterfly Curr Biol 10 591 594 10837223
Prota G 1992 Melanins and melanogenesis London Academic Press 290 p.
Black BC Smarrelli J Jr 1986 A kinetic analysis of Drosophila dopa decarboxylase Biochim Biophys Acta 870 31 40 3081033
Majerus MEN 1998 Melanism: Evolution in action Oxford Oxford University Press 338 p.
Hotta Y Benzer S 1970 Genetic dissection of the Drosophila nervous system by means of mosaics Proc Natl Acad Sci U S A 67 1156 1163 5274445
Fabian-Fine R Verstreken P Hiesinger PR Horne JA Kostyleva R 2003 Endophilin promotes a late step in endocytosis at glial invaginations in Drosophila photoreceptor terminals J Neurosci 23 10732 10744 14627659
Stark WS Carlson SD 1986 Ultrastructure of capitate projections in the optic neuropil of Diptera Cell Tissue Res 246 481 486 3098431
Meinertzhagen IA Wang Y 1997
Drosophila mutants tan and ebony have altered numbers of capitate projections, glial invaginations into photoreceptor terminals [abstract] Elsner N Wässle H Neurobiology: From membrane to mind (25th Göttingen Neurobiology Conference; Göttingen, Germany), Volume II Stuttgart Georg Thieme Verlag p. 457
Arnould JM 1985 Biosynthesis of carcinine (beta-alanyl-histamine) in vivo Arch Int Physiol Biochim 93 351 361 2421673
Babizhayev MA Seguin MC Gueyne J Evstigneeva RP Ageyeva EA 1994 L-carnosine (beta-alanyl-L-histidine) and carcinine (beta-alanylhistamine) act as natural antioxidants with hydroxyl-radical-scavenging and lipid-peroxidase activities Biochem J 304 509 516 7998987
Steinberg C Notterman DA 1996 Hemodynamic effects of carcinine in the anesthetized, instrumented, open-chest rat Crit Care Med 24 2042 2045 8968274
Calleja M Moreno E Pelaz S Morata G 1996 Visualization of gene expression in living adult Drosophila
Science 274 252 255 8824191
Spradling AC 1986 P-element mediated transformation Roberts DB Drosophila: A practical approach Oxford IRC Press Limited 175 197
Bainbridge SP Bownes M 1981 Staging the metamophosis of Drosophila melanogaster
J Embryol Exp Morphol 66 57 80 6802923
Heisenberg M 1971 Separation of receptor and lamina potentials in the electroretinogram of normal and mutant Drosophila
J Exp Biol 55 85 100 5001616
Barth M Hirsch HV Meinertzhagen IA Heisenberg M 1997 Experience-dependent developmental plasticity in the optic lobe of Drosophila melanogaster
J Neurosci 17 1493 1504 9006990
Borycz J Vohra M Tokarczyk G Meinertzhagen IA 2000 The determination of histamine in the Drosophila head J Neurosci Methods 101 141 148 10996374
Han Q Fang J Li J 2001 Kynurenine aminotransferase and glutamine transaminase K of Escherichia coli: Identity with aspartate aminotransferase Biochem J 360 617 623 11736651
|
16299587
|
PMC1285064
|
CC BY
|
2021-01-05 08:00:24
|
no
|
PLoS Genet. 2005 Nov 18; 1(5):e63
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010063
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1629958810.1371/journal.pgen.001006705-PLGE-RA-0248R2plge-01-05-08Research ArticleGenetics/Chromosome BiologyDrosophilaAll Paired Up with No Place to Go: Pairing, Synapsis, and DSB Formation in a Balancer Heterozygote Meiotic pairing in
DrosophilaGong Wei J 1McKim Kim S 2Hawley R. Scott 13*1 Stowers Institute for Medical Research, Kansas City, Missouri, United States of America
2 Waksman Institute and Department of Genetics, Rutgers, the State University of New Jersey, Piscataway, New Jersey, United States of America
3 Department of Molecular Biosciences, University of Kansas, Lawrence, Kansas, United States of America
Frankel Wayne EditorThe Jackson Laboratory, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 18 11 2005 20 10 2005 1 5 e6723 8 2005 20 10 2005 Copyright: © 2005 Gong 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.The multiply inverted X chromosome balancer FM7 strongly suppresses, or eliminates, the occurrence of crossing over when heterozygous with a normal sequence homolog. We have utilized the LacI-GFP: lacO system to visualize the effects of FM7 on meiotic pairing, synapsis, and double-strand break formation in Drosophila oocytes. Surprisingly, the analysis of meiotic pairing and synapsis for three lacO reporter couplets in FM7/X heterozygotes revealed they are paired and synapsed during zygotene/pachytene in 70%–80% of oocytes. Moreover, the regions defined by these lacO couplets undergo double-strand break formation at normal frequency. Thus, even complex aberration heterozygotes usually allow high frequencies of meiotic pairing, synapsis, and double-strand break formation in Drosophila oocytes. However, the frequencies of failed pairing and synapsis were still 1.5- to 2-fold higher than were observed for corresponding regions in oocytes with two normal sequence X chromosomes, and this effect was greatest near a breakpoint. We propose that heterozygosity for breakpoints creates a local alteration in synaptonemal complex structure that is propagated across long regions of the bivalent in a fashion analogous to chiasma interference, which also acts to suppress crossing over.
Synopsis
One of the more intriguing mysteries in chromosome biology lies in the ability of homologous chromosomes to pair during meiosis, the process that creates haploid gametes. This pairing is the crucial first step in seeing to it that each gamete receives one, and only one, copy of each chromosome. The later steps in this process include recombination and the actual segregation of paired homologs into different daughter cells. During the last century of study, people who worked on meiosis believed that changes in chromosome structure that disrupted the meiotic processes did so by impeding the pairing process. Here the authors show that pairing occurs quite normally even in cells carrying a highly rearranged chromosome. Surprisingly, even recombination is normally initiated, but not completed. These data are allowing them to reconsider several old and cherished views of the process called meiosis.
Citation:Gong WJ, McKim KS, Hawley RS (2005) All paired up with no place to go: Pairing, synapsis, and DSB formation in a balancer heterozygote. PLoS Genet 1(5): e67.
==== Body
Introduction
Despite recent advances in our understanding of the meiotic process, the mechanisms that underlie meiotic pairing and the establishment of synapsis remain poorly understood. This is particularly true for Drosophila female meiosis, both because the earlier stages of female meiosis are rapid and therefore difficult to analyze by standard cytogenetic techniques and because of the paucity of mutants that affect the pairing process. Recently, Sherizen et al. [1] in Drosophila females and Vazquez et al. [2] in Drosophila males have presented evidence that meiotic pairings in Drosophila could be an extension of existing pre-meiotic pairings. In other words, the pairing events that take place in cycles 14–15 of Drosophila embryos [3] could be maintained throughout germline differentiation and development, without necessitating a period of re-pairing in meiotic prophase. This observation supports the assertion made by Roeder and Weiner et al. [4,5] that the ability of Drosophila females to form a synaptonemal complex (SC) between homologous chromosomes in the absence of double-strand breaks (DSBs) [6] reflects the fact that these chromosomes enter meiosis as paired.
However, the suggestion that meiotic pairing is a continuation of pre-existing somatic pairings assumes that somatic pairings are maintained through the different phases of the cell cycle. In fact, there are notable examples where somatic pairing is lost in Drosophila somatic cells. For example, although Vazquez et al. [2] suggested that somatic pairing was maintained through pre-meiotic S-phase in the male germline, homolog pairing is reduced or lost during S-phase in larval neuroblasts [7] and during anaphase in embryos [8]. These observations suggest that both meiotic and mitotic pairings might need to be re-established, perhaps more than once, during each cell cycle. Thus, it is possible that despite previous somatic pairings, pairing might still need to be re-established in female meiotic prophase immediately prior to SC formation. In other words, rather than proposing that meiotic pairings are extensions of somatic pairings, it is possible that homolog pairing during prophase I in Drosophila oocytes could occur by an efficient and rapid mechanism that functions in somatic cells as well.
For obvious reasons then, the study of meiotic pairing in Drosophila must be re-phrased in terms of three distinct sets of questions. First, how do the somatic pairings that occur in Drosophila embryonic cells and in other tissues take place? Second, are those pairings maintained through division and development? Third, regardless of pre-existing somatic pairings, how does the meiotic cell facilitate meiotic synapsis and recombination? In this paper we address the third of these questions by investigating how heterozygosity for structurally altered homologs affects the maintenance of meiotic pairing, the assembly of the SC, and the initiation of genetic recombination.
In Drosophila, and in most other higher organisms, heterozygous chromosome aberrations act as dominant and region-specific suppressors of meiotic crossing over [9,10]. It has been reasonably assumed these exchange suppressions result from defects in either meiotic pairing or synapsis. However, studies of translocation heterozygotes in tomato by Herickhoff et al. [11] and more recently in Drosophila by Sherizen et al. [1] have shown that pairing and synapsis occur normally in translocation heterozygotes, even in regions close to the breakpoint, suggesting that the processes that mediate pairing and synapsis in higher eukaryotes may be insensitive to breakpoint heterozygosity. In order to examine the effects of more severe structural rearrangement on pairing, synapsis, and exchange, we focused on the effects of a multiply inverted balancer chromosome that, when heterozygous, suppresses exchange along the entire chromosome.
Specifically, we set out to examine the effects of an X chromosome balancer known as FM7 [12]. As shown in Figure 1A, the FM7 chromosome differs from a normal sequence homolog by three separate but overlapping paracentric inversions: a large inversion, (In(1)sc8), that spans the length of the X chromosome, and two smaller inversions, (In(1)dl-49 and In(1)15DE-20AE), that both lie within In(1)sc8. Thus, FM7 differs in sequence from a normal sequence X chromosome by six breakpoints distributed along the length of the X chromosome, four of which (1B, 4D, 11F, and 15DE) disrupt the euchromatin. The remaining two breaks (20A and 20E) disrupt the centric heterochromatin. The complex juxtaposition of homologs that would be required to fully pair both X chromosomes in FM7/X heterozygotes is displayed in Figure 1B.
Figure 1 The Multiply Inverted Balancer Chromosome FM7
(A) Schematic diagram of the generation of chromosomal inversions from X to FM7. Color represents the regions that are involved in inversions, and arrows are used to indicate the orientation of each region with respect to the centromere. Circles represent the centromeres. The numbers below each structure represent positions of the standard polytene map.
(B) A hypothetical structure displaying the pairing relations of the two homologs in an FM7/X heterozygote.
(C) A picture of DAPI-stained chromosome at meiotic metaphase I in FM7/X oocytes (courtesy of W. Gilliland). The two X and 4th chromosomes are positioned between the two autosomal bivalents and the poles, with the dot-like 4th chromosomes located closer to the poles. The FM7 chromosome (denoted by the brighter DAPI staining at its tip) is located between the autosomes and the upper pole, while the normal sequence X chromosome lies in the lower half of the spindle. Note that the X and FM7 chromosomes are not connected by chiasmata.
Heterozygosity for FM7 results in a complete, or near complete, suppression of exchange along the length of the X chromosome. Three lines of evidence demonstrate that heterozygosity for FM7 actually suppresses the formation of crossover products and does not simply prevent or preclude their recovery, as is characteristic of many large paracentric inversions [13,14]. First, as exemplified in Figure 1C, a cytological analysis of metaphase I figures in FM7/X heterozygotes by Theurkauf and Hawley [15] revealed that the X chromosomes were usually, if not always, achiasmate. Indeed, in 28/28 metaphase figures observed by Theurkauf and Hawley [15], both the X and FM7 chromosomes were always observed as well separated elements, such that one X chromosome was positioned between one pole and the chiasmate chromosomes while the other X chromosome was observed at a symmetrical position on the other half of the spindle. These observations parallel the behavior of the obligately achiasmate 4th chromosome.
Second, the segregation of the X chromosomes in FM7/X heterozygotes is entirely dependent on the functioning of the achiasmate-specific distributive system [16–19]. In Drosophila females that are homozygous or hemizygous for null mutants that specifically ablate the achiasmate segregation system, such as ald, alpha-Tub67C, nod, and mtrm, the frequency of X nondisjunction in FM7/X females approaches 50%, the value expected for random segregation if all, or nearly all, oocytes are achiasmate with respect to the X chromosomes [16–19].
Finally, the use of sophisticated genetic regimes by Novitski and Braver [10] to recover crossovers that do occur within females heterozygous only for the In(1)dl-49 chromosome demonstrated that exchange within this single inversion is reduced to less than 25% of normal and exchange is also strongly suppressed immediately distal to this inversion. The imposition of the remaining two inversions that make up the FM7 chromosome is expected to further reduce or eliminate even these residual levels of exchange. Indeed, Hutter [20] has attempted to recover rare exchanges within the proximal In(1)15DE;20AE inversion of FM7/X heterozygotes and estimates that if such exchanges occur at all, they do so at frequencies less than 10%–20% of what might be expected based on the genetic length of this interval. These studies all argue that, when heterozygous, the FM7 chromosome functions to suppress crossing over, rather than simply eliminating crossovers that do occur [13,14]. This cannot be due to any direct effect of the FM7 chromosome, the genes it carries, or the process of crossing over itself, because crossing over in FM7 homozygotes is normal [16].
Despite extensive study of its effects on recombination, very little is known about the effects of heterozygosity for FM7 (or of any balancer chromosome) on meiotic pairing and synapsis in Drosophila oocytes. Dernburg et al. [21] showed that the blocks of homologous heterochromatin remained paired throughout meiotic prophase in FM7/X heterozygotes. This observation forms the linchpin of the model that shows that achiasmate segregations are mediated by the maintenance of heterochromatic pairing, but no data on the pairing or synapsis of euchromatic regions are available. To assess pairing in oocytes, we took advantage of the LacI-GFP: lacO system developed by Robinett et al. [22] and applied it to the Drosophila male meiotic system by Vazquez et al. [2]. In this system, specific sites on a given pair of homologs are marked by the insertion of an array of lacO binding sites. Expression of the LacI-GFP fusion protein under the control of a germline-specific promoter (in this case, the nanos promoter) results in the binding of LacI-GFP to the array of lacO sites and thus allows the pairing of a given site to be assayed as unpaired or paired.
Surprisingly, our analysis revealed a high frequency of pairing and synapsis in FM7/X females. Moreover, at least at the level of light microscopy, the pairing and synapsis we observe is similar to that observed in females carrying two structurally normal X chromosomes. Thus, the strong reduction of recombination observed in these oocytes cannot be accounted for by a correspondingly strong defect in pairing and synapsis. We also observed a normal frequency of DSB formation on the X chromosomes in FM7/X heterozygotes, suggesting that the events that initiate meiotic recombination occur normally and are not impeded by structural heterozygosity. Indeed, the frequency of DSBs per oocyte is unchanged even in oocytes that are heterozygotes for balancers that suppress exchange on all three chromosomes.
Results
Use of the LacI-GFP: lacO System to Assess Chromosome Pairing and Synapsis in Drosophila Oocytes with Two Normal Sequence X Chromosomes
We set out to assess pairing and synapsis in Drosophila females that either carry two normal sequence X chromosomes or are heterozygous for a normal sequence X chromosome and for the multiply inverted FM7 chromosome [12]. To obtain lacO sites in corresponding positions on both the X chromosome and the FM7 balancer chromosome, we mobilized a lacO array located on Chromosome 2 to multiple sites on both the normal sequence X and FM7 and mapped the positions of these insertions on the X chromosome genomic sequence by inverse PCR.
Our initial analysis of chromosome pairing and synapsis in Drosophila oocytes focused on the study of four allelic pairs of lacO arrays located at 1C, 9B, 11A, and 18C on a pair of normal sequence X chromosomes (Table 1 and Figures 2 and 3). In SC-positive oocytes the two lacO sites are considered as paired and synapsed if any of the following three criteria are met: 1) there is only one visible green fluorescent protein (GFP) focus associated with a stretch of SC; 2) there are two clearly overlapping GFP foci associated with a stretch of SC (see the penultimate row in Figure 2); or 3) there are two distinct GFP foci that lie on opposite sides of a stretch of SC (see Figure 2). Using this method, the observed frequencies of failed synapsis for the four allelic pairs of lacO insertions studied in X/X oocytes ranged from 1.7% for the lacO insertion at 9B, to values ranging from 4.2%–4.6% for the lacO insertions at 1C, 11A, and 18C.
Table 1 Chromosome Synapsis As Assayed by LacI-GFP Tagging
Figure 2 Synapsed lacO Foci Flanking Stretches of SC in FM7/X or X/X Oocytes
In both X/X and FM7/X oocytes, paired lacO foci flank regions of SC. The positions of paired/synapsed lacO sites studied in X/X and FM7/X oocytes in zygotene/pachytene are shown at the left-most of each row. One to three optical sections show two partially overlapping or adjacent GFP foci (green) associated with a segment of SC (red). Distances between those GFP foci are shown at the right-most in each row. Due to the difficulty of accurately measuring the distance between overlapping foci, “<0.25” is used. Bars = 1 μm.
Figure 3 Distributions of Distances between GFP Foci in Both X/X and FM7/X Oocytes
These data report, in histogram form, the distribution of distances between GFP foci in those oocytes with two distinct GFP foci (including those oocytes with overlapping foci). For those oocytes with overlapping foci, the distance is measured as < 0.25 μm.
(A) The distribution of distances in oocytes containing two lacO insertions at allelic sites in X/X females (upper panel) and nearby lacO couplets in FM7/X oocytes (lower panel).
(B) The distribution of distances for non-allelic lacO insertion sites in X/X and FM7/X females. The distribution for an allelic pair of lacO insertions at 11A is provided as a control.
(C) The distribution of distances in oocytes containing two lacO insertions at an allelic site (11A) in X/X; c(3)G females (left) and a nearby lacO couplet (18A/18C) in FM7/X ; c(3)G oocytes (right panel).
However, as noted in the Materials and Methods, on average any given lacO array was detectable in only 70% of the oocytes, and thus two well-separated lacO arrays would be detectable in only approximately 50% of the oocytes in which they occurred. This required us to use two additional metrics to estimate the frequency of failed pairing and synapsis. First, we provide a more accurate measurement of pairing/synapsis failure by multiplying the observed fraction of oocytes with unpaired lacO foci by a factor of two. Second, we obviate the detection problem by considering only that subset of oocytes that exhibit two discernable foci. In Table 1, the number of synapsed GFP foci that were discernable as two distinct or overlapping dots flanking the SC is indicated in parentheses. Those oocytes in which the two foci were either touching or separated only by the width of an SC are considered synapsed, while those in which the GFP foci were well separated are scored as unsynapsed. Comparisons of these three methods of estimation are presented in Table 2.
Table 2 Summary of Synapsis Assays in X/X and FM7/X Oocytes
As an alternative to the simple qualitative characterization of two GFP foci as “synapsed or unsynapsed,” we also measured the distances between lacO sites in all oocytes in which we could clearly distinguish two GFP foci (even if they were overlapping). The histograms describing those distributions for lacO sites on normal sequence X chromosomes are presented in Figure 3A. Our analysis of X/X oocytes with two allelic lacO arrays revealed that those paired and synapsed foci that flanked a stretch of SC were never separated by more than 0.7 μm and were usually separated by only the width of the C(3)G signal that defines the SC ( ~0.4 μm). Thus, for purposes of the comparison of these data with the qualitative data on synapsis and non-synapsis presented in Table 1, those foci less than 0.7 μm apart may be considered as paired and synapsed and those foci separated by a distance of greater than 0.7 μm may be considered as unpaired or unsynapsed. Although distant or unpaired foci were rare in all four X/X genotypes studied, oocytes were occasionally observed in which either the 1C or 18C arrays were separated by distances substantially greater than 0.7 μm. While these observations demonstrate that most homologous sites are properly paired during meiosis, cases of failed pairing and synapsis do occur even in oocytes with iso-sequential X chromosomes.
Examining Oocytes with lacO Arrays Located at Different Sites on Two Normal Sequence X Chromosomes
The analysis presented above assumes that if two lacO arrays were frequently separated, even by small distances, we would still be able to visualize them as two separate dots. To estimate the effect of displacing two lacO arrays on the separation of GFP foci, we also assessed pairing and synapsis in X/X females carrying a lacO insertion at position 10A on one homolog and a lacO insertion at position 11A on the other. These two sites are separated by a physical distance of 0.9 Mb. Figure 4 presents examples of two nuclei in which the two foci were associated with a stretch of SC. In the upper case the two GFP foci were opposite from each other across the SC and scored as paired and synapsed, while in the lower case the two foci were well separated on the same stretch of SC and were considered to be unpaired. As shown in Table 1, among the 57 oocytes with two distinct GFP foci we saw 26 examples in which the two foci appeared as synapsed foci separated only by a stretch of SC. Among those 31 cases in which the two foci were not paired, 25 were nonetheless still associated with the same stretch of SC, and there were also six oocytes in which distant foci were not connected by SC. There were also 43 oocytes with just one GFP focus.
Figure 4 Synapsed and Unsynapsed lacO Sites in X/X (10A/11A) Oocytes
Note that in the lower (unsynapsed) case, the two GFP foci are displaced along the length of the same stretch of SC. In these images, which consist of one to two optical sections, two GFP foci (green) are associated with a segment of SC (red). Distances between those GFP foci are shown at the right-most in each row. Bars = 1 μm.
Figure 3B presents the distance distributions for the GFP foci in 10A/11A heterozygotes as well as distributions for two more distant pairs of lacO sites. These lacO insertions, which were also located on two normal sequence X chromosomes, are separated by physical distances of 1.6 Mb (11A/9B) and 11.7 Mb (12D/2F) (For comparison, the physical length of the X chromosome euchromatin is 22 Mb.). The distance distribution for the pair of allelic sites at 11A is presented for comparison. It is clear that as the distance between the lacO sites increases, so does the average distance between GFP foci.
Assessing Chromosome Pairing in X/X Oocytes Homozygous for the c(3)G Mutation
We were concerned that the associations of homologous chromosomes into regional domains within the nucleus might constrain both X chromosomes into a small enough nuclear region that we might fail to see unpaired lacO couplets even if they did occur. To confirm that high frequencies of failed pairing could be observed, if they indeed occurred, lacO pairing was analyzed in oocytes homozygous for the c(3)G mutation, which disrupts the pairing of euchromatic regions during zygotene/pachytene [1] (Figure 5). These experiments differ from those presented above only in that we used Orb staining to identify meiotic nuclei [23] rather than C(3)G itself. Examining oocytes homozygous for both a lacO insertion at 11A and for c(3)G revealed 15/33 nuclei with two unpaired foci. There were only three nuclei in which two distinct GFP foci were paired (for examples of unpaired foci in this genotype, see Figure 5). The distribution of distances between foci in oocytes with two GFP foci is shown in Figure 3C. By multiplying the frequency of unpaired foci by two or by computing the fraction of two foci nuclei that were unpaired (15/18), we can estimate that the lacO arrays were unpaired in 83%–91% of the oocytes examined. Thus, we can easily observe a failure in homolog pairing in X/X females that are homozygous for c(3)G.
Figure 5 Unpaired lacO Sites in c(3)G Mutant Oocytes for Both X/X and FM7/X Oocytes
Cytoplasmic protein marker ORB was used to identify the 16-cell cysts. It is present from region 2a, where it is evenly distributed in the 16 cells. At region 2b and region 3, ORB concentrates in the pro-oocytes and the oocyte. The GFP foci (green) in merge images in conjunction with DNA (blue, DAPI staining) and ORB (red). Two optical sections are shown in wildtype, while six to seven optical sections are shown in c(3)G mutants. Bars = 1 μm.
Use of the LacI-GFP: lacO System to Assess Chromosome Pairing and Synapsis in Drosophila Oocytes Heterozygous for the FM7 Balancer Chromosome and a Normal Sequence X Chromosome
To obtain lacO sites in corresponding positions on FM7, we mobilized lacO arrays to multiple sites on the FM7 chromosome and then chose those insertions that are located close to the positions of lacO sites on normal sequence X chromosomes. These include insertions at positions 1E, 8F, and 18A. By combining FM7 chromosomes carrying these insertions with normal sequence X chromosomes carrying a lacO insertion at a nearby site, we created the following lacO couplets: a couplet at 1E (FM7)/1C(X), which defines a region just proximal to the distal break of In(1)sc8 at 1B, a couplet at 8F(FM7)/9B(X) which defines a region within In(1)dl-49, and a couplet at 18A(FM7)/18C(X), which lies within In(1)15DE-20AE. The positions of these couplets in FM7/X heterozygotes are diagrammed in Figure 2. We compared the pairing and synapsis behavior of these three couplets of lacO arrays with the behavior of four allelic pairs of lacO arrays located at 1C, 9B, 11A, and 18C on a pair of normal sequence X chromosomes (see Table 1 and Figures 2 and 3).
The lower half of Table 1 displays the frequencies of failed pairing and synapsis for the three lacO couplets studied in FM7/X oocytes, as assayed by measuring the frequency of oocytes with two well-separated spots. Quite surprisingly, in most oocytes examined, these three couplets were paired and synapsed. Nonetheless, the observed frequencies of failed synapsis were higher than observed for lacO allelic pairs at similar positions in X/X females. Indeed, the observed frequencies of failed synapsis for the 8F/9B and 18A/18C lacO couplets were 10.2% and 9.6%, respectively. The frequency of unsynapsed foci for the 1E/1C couplet (16%) that lies close to the breakpoint of In(1)sc8 at 1B was substantially higher than observed for the two couplets with breakpoints in the middle of the two smaller inversions. Using the correction of a factor of two required to compensate for the fact that each lacO site is detectable in only 70% of the oocytes suggests that actual frequencies of failed pairing lie between 19.2% and 32.0% (Table 2).
The corrected frequencies of failed synapsis presented in Table 2 correlate well with the frequencies of failed synapsis calculated by only using oocytes with two discernable GFP foci (see also Table 2). For example, in the case of the 1E/1C couplet we observed 60 nuclei in which the two foci were paired and straddled an intact region of SC (for examples, see Figure 2) and 30 cases in which the two foci were well separated, suggesting that the frequency of failed synapsis is 33.3%. For the 8F/9B couplet, 18 out of 79 nuclei with two GFP foci were unsynapsed (22.8%); and for 18A/18C, 16 out of 65 nuclei with two GFP foci were unsynapsed (24.6%). It is important to note that in FM7/X oocytes in those cases in which two GFP foci were scored as unsynapsed, the GFP foci were not observed to be connected by a stretch of SC. Thus, this situation is unlike the case described above for X/X oocytes that were doubly heterozygous for lacO insertions at 10A and 11A, in which even well-separated (unsynapsed) GFP foci were still found on opposite sides of a contiguous SC. Rather, these instances of unsynapsed foci in FM7/X oocytes appear to represent real cases of failed pairing and synapsis in the presence of the balancer chromosome, rather than simply the result of the distance between the two sites of lacO insertion on the X and FM7 chromosomes. The histograms describing distributions of the distances between the foci in each of these genotypes are presented in Figure 3A, and no significant difference from those three allelic lacO pairs on X/X chromosomes was observed (for all three comparisons, p > 0.09).
We also note that the frequencies of failed synapsis are similar for oocytes in regions 2a, 2b, and 3 of Drosophila oocytes that correspond to the zygotene and pachytene stages of meiotic prophase (Table 1). The fact that the frequencies of failed synapsis are stable throughout meiotic prophase argues strongly that synaptic adjustment does not occur in Drosophila oocytes.
High Frequencies of Failed Pairing in FM7/X Oocytes Can Be Induced by Homozygosity for c(3)G
The data presented above suggest that the three regions being assayed in FM7/X females are properly paired in approximately 70% of oocytes. However, given the dramatic exchange suppression observed in this genotype, we wondered whether or not we might have missed failed pairings in FM7/X heterozygotes for structural reasons. Perhaps the conformational twisting resulting from the need to maintain heterochromatic associations in the presence of heterozygosity for an inversion with heterochromatic breakpoints might restrict the X and FM7 chromosomes to a small enough nuclear territory or domain that lacO foci might appear paired even in the absence of proper pairing. This possibility can be directly tested by examining lacO pairing in oocytes homozygous for the c(3)G mutation that causes a failure of pairing in euchromatic regions during zygotene/pachytene [1] without disrupting heterochromatic associations [21]. Examining FM7/X oocytes carrying the lacO couplet at 18A/18C and homozygous for c(3)G revealed 13/23 nuclei with two unpaired foci, one nucleus with two paired foci, and nine nuclei with a single GFP focus (for examples of paired and unpaired foci in this genotype, see Figure 5). The distribution of distances in oocytes with two GFP foci is shown in Figure 3C. By multiplying the frequency of unpaired foci by two or by computing the fraction of two foci nuclei that were unpaired (13/14), we can estimate that the lacO arrays were unpaired in 93%–100% of the oocytes examined. Thus, we can easily observe a failure in homolog pairing in FM7/X females that are homozygous for c(3)G.
In addition to the oocytes considered above, we also observed two oocytes that had four separated foci. A small number of oocytes with 3–4 FISH signals were also observed by Sherizen et al. [1] in c(3)G homozygotes. As suggested by those authors, the existence of these oocytes presumably reflects a role of the C(3)G protein in the maintenance of euchromatic sister chromatid cohesion as well as in the maintenance of homolog–homolog association. It is worth noting that both of these roles appear to be restricted to euchromatin; c(3)G oocytes show normal pairing in the heterochromatin [21].
Comparing the Frequencies of Failed Pairing and Synapsis in X/X and FM7/X Oocytes
Table 2 compares the frequency of failed synapsis for allelic lacO sites in X/X females and lacO couplets in FM7/X females by the three metrics considered above (observed frequency of two separated foci, corrected frequency of separated foci, and fraction of oocytes with two foci in which the foci are unsynapsed). Although all regions tested appear to be paired and synapsed in the majority of oocytes of both genotypes, it is clear that the frequencies of failed synapses are higher in FM7/X heterozygotes. The highest frequency of failed synapsis in FM7/X oocytes (~33%) was observed for the 1E/1C lacO couplet. This region is of specific interest because in FM7/X oocytes these two lacO arrays lie immediately proximal to the distal breakpoint of In(1)sc8, a distance less than 3% of the length of the X euchromatin. The presence of proper pairing and synapsis in the remaining two-thirds of FM7/X oocytes argues strongly that breakpoints do not usually disrupt pairing and synapsis even in the immediate vicinity of the breakpoint. For the remaining two lacO couplets (8F/9B and 18A/18C), which define regions lying in the middle of In(1)dl-49 and In(1)15D-20AE, the frequencies of apparent proper pairing and synapsis are estimated to be 75%–80%.
These observations argue strongly that the ability of the FM7 balancer chromosome to suppress exchange by more than 100-fold when heterozygous cannot be explained by a corresponding strong suppression of pairing and/or synapsis. However, the types of observations presented here cannot exclude the possibility that there are subtle differences in homolog–homolog associations or synapsis in FM7/X heterozygotes, or defects in SC structure, that cannot be resolved by the techniques employed here. Nonetheless, as shown in the next two sections, if such differences do exist, they do not preclude the normal initiation of recombination as evidenced by DSB formation.
X Chromosomes in FM7/X Heterozygotes Experience a Normal Number of DSBs
As noted in the Introduction, the FM7 chromosome functions to suppress crossing over, rather than simply eliminating crossovers that do occur. Still, our failure to detect a defect in either pairing or synapsis makes the absence of crossing over in FM7/X heterozygotes difficult to understand. To further investigate the mechanism underlying the crossover suppression in FM7/X heterozygotes, we sought to determine whether the suppression could be the result of the prevention of DSB formation between FM7 and the normal sequence X chromosome.
To determine whether DSBs were formed along the length of the synapsed FM7/X pair, we visualized sites of DSB formation using γ-HIS2AV antibody [24,25] as well as paired lacO sites in C(3)G-positive nuclei. Such an analysis requires finding only those cytologically favorable nuclei in which a well-separated length of SC is marked by paired GFP foci, and then assaying that stretch of SC for the presence of a γ-HIS2AV focus (or foci) indicative of DSB formation.
In this study, allelic pairs at 9B and 18C in X/X females were compared with lacO couplets at 8F/9B and 18A/18C in FM7/X females. As shown in Figure 6 and Table 3, such cases of SC stretches doubly marked with GFP foci and a γ-HIS2AV focus were observed in ~5% of the SC-positive nuclei for all four genotypes studied. Indeed, combining the data for the two nearby lacO couplets studied in FM7/X oocytes, we observed 12 cases in which the same stretch of SC was marked by both a GFP focus (or paired foci) and by a γ-HIS2AV focus out of 237 oocytes examined in FM7/X oocytes (5.1%). Similarly (again combining the data for the two allelic pairs of lacO insertions studied in X/X females), nine cases in which the same stretch of SC was marked by both a GFP focus (or paired foci) and by a γ-HIS2AV focus were observed in 190 oocytes from X/X females (4.7%). Furthermore, the average distances between a GFP focus and a γ-HIS2AV focus were not significantly different between FM7/X and X/X oocytes (p = 0.86; the average distance in X/X is 0.79 μm while the value is 0.81 μm in FM7/X). While it is not possible to compare either of these frequencies to some absolute expectation of the number of such foci per given length of SC, it is clear that there is no obvious reduction in the frequency of DSBs in balancer heterozygotes.
Figure 6 DSB Formation on the Same Stretch of SC As Paired lacO Couplets in FM7/X and X/X Oocytes
The representatives from FM7/X (18A/18C) and X/X (18C/18C) oocytes are shown in this figure. Single C(3)G-positive (red) meiotic cells are shown in each row. DSBs are indicated by γ-HIS2AV staining (blue, arrows). GFP foci (green, arrows) represent the pairing FM7 and X or X and X chromosomes. Two to three optical sections of images were projected. Bars = 1 μm.
Table 3 DSB Formation and Synapsed lacO Sites on the Same Stretch of SC
Even a Global Suppression of Crossing Over Does Not Alter the Frequency of DSB Formation
Because of the difficulties inherent in finding cytologically favorable nuclei in which a well-separated length of C(3)G staining is marked by paired GFP foci, and then assaying that stretch of SC for the presence of a γ-HIS2AV focus, we chose to simply measure the number of DSBs occurring in oocytes in which exchange is suppressed on all five chromosome arms as a result of heterozygosity for three balancer chromosomes FM7, SM1, and TM3. The SM1 and TM3 balancer chromosomes each involve six euchromatic breakpoints. They also strongly suppress exchange when heterozygous with normal sequence homologs, as evidenced by their sensitivity to nondisjunction induced by mutants that impair the achiasmate segregation systems [26,27].
Data for both γ-HIS2AV staining and C(3)G staining in oocytes doubly or triply heterozygous for these balancers are presented in Figure 7. We saw no obvious difference in either the number of γ-HIS2AV foci per nucleus during the length of meiotic prophase or in the general structure or organization of the SC when comparing wildtype oocytes. This observation suggests that even when confronted with two or three balancer chromosomes, both extensive synapsis and DSB formation still occur in Drosophila oocytes. Moreover, this experiment also suggests that the well-documented ability of heterozygous inversions to increase the frequency of recombination elsewhere in the genome, referred to as the “interchromosomal effect” [28], is not mediated by either a substantial increase in the total number of DSBs or by an obvious change in the timing of their appearance or disappearance.
Figure 7 DSBs in Oocytes Heterozygous for Three Balancer Chromosomes
(A) Representatives from the oocytes with normal or heterozygous balancer chromosomes. Maximum intensity projection of image stacks of nuclei showing C(3)G (red) in conjunction with DNA (blue, DAPI staining) and DSB (green, γ-HIS2AV staining). Bars = 1 μm.
(B) Distribution graph representing the number of γ-HIS2AV foci per C(3)G staining nucleus at different developmental regions in germaria. The sample size of oocytes in wildtype is 64, 34, and 9, representing region 2a, 2b, and 3; in FM7/+;SM1/+;TM3/+, 48, 43, and 11; in +/+;SM1/+;TM3/+, 66, 38, and 10, respectively. Data are presented as means ± SD.
Effects of the FM7 Balancer Chromosome on Pre-Meiotic Pairing Are Similar to Its Effects on Meiotic Pairing
Table 4 presents data for pre-meiotic pairing of allelic lacO insertion sites in X/X germlines and of nearby lacO couplets in FM7/X germlines. These interphase nuclei were obtained from regions 1 and 2a of the germarium and include mitotically dividing cystoblast and cystocyte cells (region 1) as well as nuclei from 16 cell cysts that have not yet assembled SC (region 2a). For X/X nuclei, the frequencies of failed pairing (as indicated by two well-separated GFP foci) ranged from 1.4 % (1C) to 6.4 % (18C). While the frequencies of failed pairings were higher for the FM7/X lacO couplets (5.9% to 18.3%), they are still substantially less than frequency of unpaired GFP foci (39.2%, n = 51) observed in pre-meiotic nuclei in X/X females carrying a lacO insertion at position 10A on one homolog and a lacO insertion at position 11A on the other (a physical distance of 0.9 Mb). The observation that the effects of FM7 on pre-meiotic pairing are similar to its effects on meiotic pairing support an emerging view that homolog pairing relationships are established early in Drosophila development and maintained until the completion of synapsis during meiosis [1,2]. These data also confirm and extend the preliminary cytological observations of Becker [29] that suggested that at least for the large In(1)sc8 inversion, the sequences within the inversion can pair with a normal sequence homolog in somatic cells.
Table 4 Chromosome Pairing As Assayed by LacI-GFP Tagging
Discussion
The data presented above argue that while the oocytes heterozygous for the FM7 balancer chromosome and for a normal sequence X chromosome do exhibit a higher frequency of failed pairing and synapsis than do oocytes carrying two normal sequence X chromosomes, the effect is small in comparison to the global defect in exchange observed in FM7/X females. Moreover, the fact that the frequencies of failed pairing and synapsis do not increase throughout meiotic prophase argues strongly that synaptic adjustment does not occur in Drosophila oocytes. In that sense, our data confirm and extend the studies of translocation heterozygotes performed by Sherizen et al. [1] and demonstrate that despite their ability to suppress exchange over large distances, heterozygous breakpoints do not create corresponding strong defects in pairing or synapsis. However, we do see at least a weak defect in pairing that appears to be strongest in the interval closest to a breakpoint. The possible significance of these defects is discussed below. Finally, our data also allow us to conclude that the exchange suppression generated by breakpoint heterozygosity is not the result of a strong decrease in the frequency of DSBs.
While we observed that X chromosomal bivalents are paired and synapsed in the large majority of FM7/X oocytes, we nonetheless do observe an increased frequency of failed pairing and synapsis in FM7/X heterozygotes. The effect on pairing and synapsis that we see may parallel an effect on synapsis in the vicinity of the breakpoints in translocation heterozygotes observed by Sherizen et al. [1]. These authors, who used FISH to analyze pairing and synapsis near the breakpoints, observed that the SC staining in translocation heterozygotes was sometimes less intense than was observed in wildtype controls and was missing entirely in 10%–20% of nuclei. Although we do not see a decrease in C(3)G intensity in FM7/X heterozygotes, the frequency of synapsis failures observed by Sherizen et al. [1] (10%–20%) is roughly similar to the frequencies of failed pairing and synapsis observed in our study. Thus, in both inversion and translocation heterozygotes, it appears that pairing and synapsis still occurs between the rearrangement and the normal sequence homolog in the majority of oocytes.
How Might Heterozygosity for a Breakpoint Suppress Exchange, without Suppressing Pairing and Synapsis in Drosophila?
Several aspects of its meiotic process make the Drosophila oocyte different from meiotic cells in many other species. First, in both male and female meiosis, Drosophila homologs either enter meiosis in a fashion that preserves existing pre-meiotic pairings [1,2, and this study], or they are able to rapidly establish pairing following cell division. Second, although the maturation of such pairings to synapsis via SC assembly does not require the formation of DSBs [6], DSB formation does occur in the absence of synapsis [25]. Third, the maturation of the DSBs to either reciprocal crossover events or gene conversion events absolutely requires the presence of the C(3)G Zip1-like protein [30,31]. Breaks that occur in the absence of C(3)G function are evidently repaired by conversion-like events involving either the sister chromatid or the homolog, because only slight, if any, increases in the frequency of sister chromatid exchange are observed in females homozygous for null mutants in c(3)G [6,32]. However, Carlson has presented evidence that inter-homolog conversion events are quite rare, if they occur at all [30].
Within the context of this meiotic system, we can propose two general classes of models to explain the ability of breakpoints to suppress exchange over relatively large distances, without obviously affecting pairing. According to the first model, proposed by both Hawley [33] and more fully by Sherizen et al. [1], the conversion of DSBs into crossovers would require a long region of uninterrupted continuity of the SC. Either the absence of that continuity or a structural change in the SC at the site of a breakpoint would serve to suppress crossover formation over long distances from the breakpoint.
It is possible that breakpoints might suppress exchange by a mechanism that is functionally similar to crossover interference. Indeed, the process of synapsis across a breakpoint might require a distortion or twisting of chromosome axes as they switch from pairing with one chromosome to the other (in the case of a translocation heterozygote), or from one chromosomal region to another (in the case of an inversion heterozygote), that mimics the effect of an actual crossover event on axis structure, and, in doing so, propagates a signal along the length of the SC that diminishes the likelihood of an actual exchange. This model has the intriguing feature that it explains the observation that the ability of breakpoints to suppress exchange in an organism with strong interference, such as Drosophila, is quite robust compared with their ability to suppress exchange in Saccharomyces cerevisiae, in which interference is weak [34]. It also explains the well-documented observation that genetic or environmental factors that reduce the level of interference (such as heat, age, the inter-chromosomal effect, or heterozygosity for c(3)G) significantly elevate the levels of recombination in regions suppressed by breakpoint heterozygosity. Finally, this model makes the prediction that heterozygous breakpoints should be much poorer suppressors of exchange in the closely related species D. mauritiana, in which interference is weak or absent [35], than they are in D. melanogaster.
According to the second model, one could imagine that breakpoints do create subtle or short-lived defects in pairing and synapsis that propagate over long distances and are not observable by current methods. While such “invisible defects” are, of course, hard to prove or disprove, their existence might explain the observations of Sherizen et al. [1] that both reciprocal recombination and gene conversion are suppressed by breakpoint heterozygosity. Given that exchange, and apparently gene conversion as well, requires proper synapsis in Drosophila (or at least the presence of the C(3)G protein) [30,31], a subtle defect in synapsis propagated over long distances might explain both defects. One mechanistic view of this model might take the form of proposing that there is a critical period early during zygotene/pachytene in which either C(3)G must be present and/or SC must properly form in order to allow DSBs to be properly matured to either reciprocal exchanges or gene conversion events. If breakpoint heterozygosity delayed C(3)G action or incorporation within or beyond that brief temporal window, one might see dramatic reductions in exchange. The subsequent proper assembly of the SC might well mask such a brief and early defect. This model has the benefit that with development of the technology to visualize meiotic prophase in living oocytes, it might eventually be testable. An alternative version of this model might suggest that the pairing and synaptic failures that occur close to a breakpoint result in a propagated disruption in SC structure that, while too subtle to be observed by the techniques proposed here, is nonetheless sufficient to prevent the maturation of DSBs into crossovers.
Both of these models allow us to propose roles for the crossover-suppression boundary sites mapped by both Hawley [33] and Sherizen et al. [1]. While it now seems unlikely that such sites play a role in mediating meiotic pairing, it seems likely that they define regions in which proper synapsis or SC structure can be restored in such a way that the effects of a heterozygous breakpoint on synapsis or SC structure are damped out across the lengths of these regions. Such a function may well be consistent with the finding that at least the sites mapped by Hawley [33] reside in regions of intercalary heterochromatin. Such a role for these sites as “fasteners” of synapsis is consistent with the proposal of Sherizen et al. [1] that these sites are involved in defining large chromosomal domains that control crossover formation, perhaps by playing roles in either initiating SC formation or correcting deformations in SC structure.
Two other possibilities, which we deem less likely, also need to be at least mentioned. The first is that chromosomes that dominantly suppress the recovery of crossovers may well acquire a significant amount of sequence divergence as a consequence of reduced recombination. This may be especially true for balancer chromosomes such as FM7 that suppress exchange along their entire length. One could imagine that the accumulation of such sequence divergence might suppress the formation of those recombinational intermediates that facilitate both reciprocal exchange and gene conversion. While this model is attractive in terms of simplicity, it fails to explain how a translocation breakpoint might suppress exchange. Moreover, our limited amount of sequence analysis on the FM7 balancer chromosome, performed in the vicinity of the Axs locus, suggests a relatively low level of sequence polymorphism (~2%) over a 2-kb region (Gustafson and Hawley, unpublished data). This level of polymorphism is unlikely to greatly reduce the rate of recombination [36]. Finally, Coyne and his collaborators have characterized one pericentric inversion that is coupled to a cis-acting mutant that dominantly suppresses exchange within the inverted region [37]. While such mutants clearly exist, the fact that most aberrations (including FM7) allow normal, or near normal, levels of exchange when homozygous renders this possibility unlikely.
What Is the Fate of DSBs in Balancer Heterozygotes?
Given that the DSBs that occur along the length of the FM7/X bivalent are not matured into crossovers, it becomes important to understand just how they might be repaired. Studies of exchange in females heterozygous for both FM7 and a ring-X chromosome (FM7/R(1) females) presented in McKim et al. [6] failed to show a substantial level of ring chromosome loss, as might be expected if DSBs were frequently processed to sister chromatid exchanges [38]. One possibility is that these events are repaired by either inter-homolog gene conversion events or by gene conversion events involving the sister chromatid. The effect of breakpoint heterozygosity on conversion is unclear. Sherizen et al. [1] found a greater than 6-fold reduction of inter-homolog conversion events near a breakpoint in translocation heterozygotes. Perhaps then, in the vicinity of the breakpoint, sister chromatid conversion events predominate, while at greater distances from the breakpoint repair by inter-homolog conversion events becomes more frequent.
However, Chovnick [39] found little or no effect on gene conversion when comparing conversion at the rosy locus in females with two normal sequence third chromosomes and in females heterozygous for a paracentric inversion that includes rosy. One possible explanation for this discrepancy might lie in the fact that the two breakpoints studied by Sherizen et al. [1] were within one numbered polytene division of the rosy locus, while both breakpoints of the inversion studied by Chovnick [39] were greater than four polytene units away from rosy. Indeed, the inversion studied by Chovnick [39] included two “boundary or pairing site” elements (see [1]) that might function to restore pairing and synapsis.
Heterozygosity for Aberration Breakpoints and Exchange Suppression in Other Species
Numerous meiotic systems have been characterized in which breakpoint heterozygosity leads to absent or aberrant synapsis. As reviewed by Koehler et al. [40], studies of pairing and synapsis of simple inversion heterozygotes in various organisms have revealed three major patterns of pairing and synapsis. In the first pattern, homologous synapsis, the sequences within the inversion pair and synapse properly with their homolog. In the second process, synaptic adjustment, the inversion loop that is initially formed by homologous pairing is gradually re-adjusted by progressive heterologous synapsis until the loop has been replaced by a linear stretch of SC running from end to end of the bivalent. In the third process, heterologous synapsis, no loop is formed and heterologous synapsis occurs concomitantly with the establishment of homologous synapsis elsewhere in the genome. Although both synaptic adjustment and heterologous synapsis are well documented, organisms, and even individual aberrations, appear to differ in terms of which process predominates (for review, see Koehler et al. [40]). Our data suggest that the frequency of pairing at the three sites we monitored does not change during meiotic prophase. Indeed, the frequencies of failed pairing in SC positive cells are quite similar to those observed in pre-meiotic nuclei. Thus, it seems unlikely that a process analogous to synaptic adjustment occurs in Drosophila oocytes.
Materials and Methods
Drosophila strains.
The transgenic construct expressing LacI-GFP and the autosome lacO transgenic lines were gifts from A. S. Belmont and J. W. Sedat, and described in Vazquez et al. [2]. We then moved the lacO P-element insertion from sites on the autosomes to new locations on normal X and FM7 chromosomes by providing the transposase source Δ2–3. The positions of the new lacO insertions were subsequently determined by inverse PCR. Genomic DNA was digested by Sau3AI or MspI; after ligation, PCR was performed using primer pairs CGGATATATGTCGGCTACTCCTTGC and CACCCAAGGCTCTGCTCCCACAAT, or CTAGGTACGGCATCTGCGTTGAGTC and ATTGAGACGAAATGAACCACTCGGA. c(3)G68 mutant lines were described in [41].
Antibodies and immunofluorescence.
All the immunolocalization experiments were carried out as described in [41]. The mouse anti-C(3)G antibody [41] was used at 1:500; both mouse monoclonal Orb antibodies 4H8 and 6H4 [23] were used together at 1:30. Secondary antibodies Cy3-conjugated (Jackson ImmunoResearch, West Grove, Pennsylvania, United States) and Alexa 647-conjugated (Molecular Probes, Eugene, Oregon, United States) anti-mouse IgG were used at a dilution of 1:500.
For detecting DSBs, the anti-phospho-H2AV(Ser139) rabbit polyclonal antibody (Upstate Biotechnology, Lake Placid, New York, United States) was used at 1:100, and the secondary antibody Cy3-conjugated anti-rabbit (Jackson ImmunoResearch) was used at 1:300. All distinct γ-HIS2AV foci were counted, regardless of the size of the individual. The size distribution of γ-HIS2AV foci appeared to be comparable in all the different genotypes of flies.
Microscopy and image analyses.
Images were collected using a DeltaVision microscopy system (Applied Precision, Issaquah, Washington, United States), equipped with an Olympus IX70 inverted microscope and high-resolution CCD camera. Image data were deconvolved using the softWoRx v. 2.5 software package (Applied Precision). The distance between the centers of two GFP foci was measured using the same software.
Identifying oocytes within the germarium.
In the Drosophila ovariole, germline stem cells are located at the anterior-most tip of the germarium. Following stem cell division, primary oogonial cells (cystoblasts) undergo four mitotic divisions with incomplete cytokinesis to create 16 cell cysts in which the cells are inter-connected by ring canals. We identified meiotic cells by staining for the C(3)G protein, which comprises the transverse filaments of the SC. In region 2a, two to four cells of each ball-shaped 16-cell cyst enter meiosis (as identified by SC formation). After this point, cysts move toward the posterior of the germarium, where they first flatten out to a pancake-like shape (region 2b) and then are enveloped by a monolayer of follicle cells at the posterior end of the germarium (region 3).
Assessing the frequency of pairing in meiotic and mitotic cells.
Our initial analysis of chromosome pairing and synapsis in Drosophila oocytes focused on the study of four allelic pairs of lacO arrays located at 1C, 9B, 11A, and 18C on a pair of normal sequence X chromosomes (see Table 1 and Figures 2 and 3). In its simplest form, such an analysis would look at each oocyte and determine whether or not the two lacO sites were paired, as evidenced by either the presence of a single GFP focus or two nearby foci that flanked a SC, or unpaired, as evidenced by two well-separated lacO foci. Unfortunately, such a simple type of analysis presumes that each lacO array is always detected in 100% of the oocytes. However, an analysis of oocytes that were heterozygous for a single lacO array showed that any given lacO array is visible in only approximately 70% of the oocytes. For three lacO arrays located on normal sequence X chromosomes at positions 1C, 9B, and 18C, the frequencies of oocytes carrying a single copy of this array that exhibited the expected single GFP focus were 65.3 % (n = 118), 72.8% (n = 114), and 68.0% (n = 103), respectively. Similarly, with three lacO arrays located on the FM7 balancer chromosome at positions 1E, 8F, and 18A, the frequencies of oocytes carrying a single copy of this array that exhibited the expected single GFP focus were 74.3 % (n = 101), 65.4 % (n = 81), and 71.0% (n = 107), respectively.
Our ability to detect a single lacO array in only ~70% of the cases suggests that two well-separated lacO foci would be detectable in only 49% of the instances in which they occurred. Indeed, when we examined females that were doubly heterozygous for lacO foci located at distant sites along the X chromosome, the frequency of oocytes that exhibited the expected two foci was indeed approximately 50%. As expected, in the case of the 10A/11A double heterozygotes in X/X females, we observed only two discrete foci in 57% (n = 100) of the oocytes examined. Similar frequencies of nuclei with two GFP foci were also observed in females doubly heterozygous for lacO insertions at 9B and 11A (52.8%, n = 89) and for insertions at 2F and 12D (66.2%, n = 77). These observations suggest that we may miss one of the two lacO arrays in approximately 50% of oocytes. For this reason, our estimates of the frequency of failed pairing and synapsis may be under-estimated by as much as 2-fold, suggesting that the frequency of synapsis failure may actually range from 3%–10% for the four sites examined (Table 2). Similar observations were made in FM7/X oocytes that were doubly heterozygous for lacO insertions at 18A and 9B and at 8F and 18C. In the case of the 9B/18A double heterozygote, we were able to visualize two foci in 47.4% (n = 78) of the oocytes, while in 8F/18C double heterozygotes we were able to visualize two foci in 57.1% (n = 91) of the oocytes. These observations suggest that, as was the case for the lacO insertions studied in X/X females, we may be underestimating the frequency of failed pairing and/or synapsis by as much as a factor of two.
For these reasons, we have evaluated the frequency of failed pairing by three separate parameters. First, as exemplified in Table 1, we simply report the fraction of oocyte with two clearly separate GFP foci. While this metric is clearly an underestimate, it nonetheless can be used to compare the severity of pairing failures between the genotypes examined. Second, we provide a more accurate measurement of pairing failure by multiplying the observed fraction of oocytes with unpaired lacO foci by a factor of two. Finally, we obviate the detection problem by considering only that subset of oocytes that exhibit two discernable foci. Those oocytes in which the two foci were either touching or separated only by the width of an SC are considered paired, while those in which the GFP foci were well separated are viewed as unpaired. This approach also allows us to provide more quantitative estimates of the frequency of failed pairing by measuring the distances between the two foci in each oocyte. Comparisons of these three methods of estimation are presented in Table 2.
Finally, we were concerned that our frequency of failed pairing might be over-estimated by cases in which two foci were created by sister chromatid separation. We can discount this possibility for three reasons. First, because each lacO array has a characteristic intensity we could tell the difference between two unpaired arrays and two separated sisters. Second, as indicated by their absence in Table 1, we did not see oocytes with the three or four foci that might be expected if sister separation was common. We did however, see such oocytes (those with three to four dots) in c(3)G oocytes, confirming the observations of Sherizen et al. [1]. Third, we examined lacO:LacI-GFP interactions in oocytes with but one copy of the lacO array. In such oocytes, sister separation could be easily detected but the frequency of such events was extremely low. A discernable separation of nearby or overlapping foci was observed in 1%–3% of oocytes. For all of these reasons, it seems very unlikely that our estimates of the frequency of failed pairing might be greatly over-estimated due to sister chromatid separation.
Statistics.
Statistical analyses were performed using Graphpad Instat software for Macintosh. The distance distributions were analyzed using the Mann-Whitney test. For statistical analyses of the distances of GFP foci (Figure 3A), the distance of overlapping foci was assigned as 0.125 μm.
We thank Abby Dernburg, Scott Page, Joseph Kramer, Cathy Lake, William Gilliland, Wei Cui, and Youbin Xiang for comments on the manuscript. We also thank Kathy Teeter for technical support. We are grateful to John Sedat for both the original autosomal lacO insertion lines and for advice and encouragement throughout the course of this study. This research was supported by a grant from the National Institutes of Health to RSH and the American Cancer Society to KSM.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. WJG, KSM, and RSH conceived and designed the experiments. WJG performed the experiments. WJG, KSM, and RSH analyzed the data. WJG and RSH contributed reagents/materials/analysis tools, and wrote the paper.
A previous version of this article appeared as an Early Online Release on October 20, 2005 (DOI: 10.1371/journal.pgen.0010067.eor).
Abbreviations
DSBdouble-strand break
GFPgreen fluorescent protein
SCsynaptonemal complex
==== Refs
References
Sherizen D Jang JK Bhagat R Kato N McKim KS 2005 Meiotic recombination in Drosophila females depends on chromosome continuity between genetically defined boundaries Genetics 169 767 781 15545646
Vazquez J Belmont AS Sedat JW 2002 The dynamics of homologous chromosome pairing during male Drosophila meiosis Curr Biol 12 1473 1483 12225662
Hiraoka Y Dernburg AF Parmelee SJ Rykowski MC Agard DA 1993 The onset of homologous chromosome pairing during Drosophila melanogaster embryogenesis J Cell Biol 120 591 600 8425892
Roeder GS 1997 Meiotic chromosomes: It takes two to tango Genes Dev 11 2600 2621 9334324
Weiner BM Kleckner N 1997 Chromosome pairing via multiple interstitial interactions before and during meiosis in yeast Cell 77 977 991
McKim KS Green-Marroquin BL Sekelsky JJ Chin G Steinberg C 1998 Meiotic synapsis in the absence of recombination Science 279 876 878 9452390
Csink AK Henikoff S 1998 Large-scale chromosomal movements during interphase progression in Drosophila
J Cell Biol 143 13 22 9763417
Fung JC Marshall WF Dernburg A Agard DA Sedat JW 1998 Homologous chromosome pairing in Drosophila melanogaster proceeds through multiple independent initiations J Cell Biol 141 5 20 9531544
Dobzhansky T 1934 Studies on chromosome conjugation Z Indukt Abstamm Vererbungsl 68 134 162
Novitski E Braver G 1954 An analysis of crossing over within a heterozygous inversion in Drosophila melanogaster
Genetics 39 197 209 17247476
Herickhoff L Stack S Sherman J 1993 The relationship between synapsis recombination nodules and chiasmata in tomato translocation heterozygotes Heredity 71 373 385
Merriam JR 1969
FM7 : A ‘new' first chromosome balancer D I S 44 101
Sturtevant AH Beadle GW 1936 The relations of inversions in the X chromosome of Drosophila melanogaster to crossing over and disjunction Genetics 21 554 604 17246812
Roberts PA 1970 Screening for X-ray induced crossover suppressors in Drosophila melanogaster : Prevalence and effectiveness of translocations Genetics 65 429 448 5519644
Theurkauf WE Hawley RS 1992 Meiotic spindle assembly in Drosophila females: Behavior of nonexchange chromosomes and the effects of mutations in the nod kinesin-like protein J Cell Biol 116 1167 1180 1740471
Zhang P Hawley RS 1990 The genetic analysis of distributive segregation in Drosophila melanogaster . II. Further genetic analysis of the nod locus Genetics 125 115 127 2111262
Matthies HJ Messina LG Namba R Greer KJ Walker MY 1999 Mutations in the alpha-tubulin 67C gene specifically impair achiasmate segregation in Drosophila melanogaster
J Cell Biol 147 1137 1144 10601329
Harris D Orme C Kramer J Namba L Champion M 2003 A deficiency screen of the major autosomes identifies a gene (matrimony ) that is haplo-insufficient for achiasmate segregation in Drosophila oocytes Genetics 165 637 652 14573476
Gilliland WD Wayson SM Hawley RS 2005 The meiotic defects of mutants in the Drosophila mps1 gene reveal a critical role of Mps1 in the segregation of achiasmate homologs Curr Biol 15 672 677 15823541
Hutter P 1990 ‘Exceptional sons' from Drosophila melanogaster mothers carrying a balancer X chromosome Genet Res 55 159 164 2118468
Dernburg AF Sedat JW Hawley RS 1996 Direct evidence of a role for heterochromatin in meiotic chromosome segregation Cell 86 135 146 8689681
Robinett CC Straight A Li G Willhelm C Sudlow G 1996 In vivo localization of DNA sequences and visualization of large-scale chromatin organization using lac operator/repressor recognition J Cell Biol 135 1685 1700 8991083
Lantz V Chang JS Horabin JI Bopp D Schedl P 1994 The Drosophila orb RNA-binding protein is required for the formation of the egg chamber and establishment of polarity Genes Dev 8 598 613 7523244
Liu H Jang JK Kato N McKim KS 2002
mei-P22 encodes a chromosome-associated protein required for the initiation of meiotic recombination in Drosophila melanogaster
Genetics 162 245 258 12242237
Jang JK Sherizen DE Bhagat R Manheim EA McKim KS 2003 Relationship of DNA double-strand breaks to synapsis in Drosophila
J Cell Sci 116 3069 3077 12799415
Carpenter ATC 1973 A meiotic mutant defective in distributive disjunction in Drosophila melanogaster
Genetics 73 393 428 4633612
Zitron AE Hawley RS 1989 The genetic analysis of distributive segregation in Drosophila melanogaster
Genetics 122 801 821 2503421
Lucchesi JC 1976 Interchromosomal effects Ashburner M Novitski E The genetics and biology of Drosophila London Academic Press 315 329
Becker HJ 1969 The influence of heterochromatin, inversion-heterozygosity and somatic pairing on X-ray induced mitotic recombination in Drosophila melanogaster
Mol Gen Genet 105 203 218 5369216
Carlson PS 1972 The effects of inversions and the C(3)G mutation on intragenic recombination in Drosophila
Genet Res 19 129 132 4624988
Page SL Hawley RS 2004 The genetics and molecular biology of the synaptonemal complex Annu Rev Cell Dev Biol 20 525 558 15473851
Hall JC 1977 Recombination influenced by two alleles of the meiotic mutant c(3)G
D I S 52 143 144
Hawley RS 1980 Chromosomal sites necessary for normal levels of meiotic recombination in Drosophila melanogaster . I. Evidence for and mapping of the sites Genetics 94 625 646 6772522
Sherman F Helms C 1978 A chromosomal translocation causing overproduction of iso-2-cytochrome c in yeast Genetics 88 689 707 206486
True JR Mercer JM Laurie CC 1996 Differences in crossover frequency and distribution among three sibling species of Drosophila
Genetics 142 507 523 8852849
Hoskins RA Phan AC Naeemuddin M Mapa FA Ruddy DA 2001 Single nucleotide polymorphism markers for genetic mapping in Drosophila melanogaster
Genome Res 11 1100 1113 11381036
Coyne JA Meyers W Crittenden AP Sniegowski P 1993 The fertility effects of pericentric inversions in Drosophila melanogaster
Genetics 134 487 496 8325485
Webber HA Howard L Bickel SE 2004 The cohesion protein ORD is required for homologue bias during meiotic recombination J Cell Biol 164 819 829 15007062
Chovnick A 1973 Gene conversion and transfer of genetic information within the inverted region of inversion heterozygotes Genetics 75 123 131 4202769
Koehler KE Millie EA Cherry JP Schrump SE Hassold TJ 2004 Meiotic exchange and segregation in female mice heterozygous for paracentric inversions Genetics 166 1199 1214 15082541
Page S Hawley RS 2001
c(3)G encodes a Drosophila synaptonemal complex protein Genes Dev 15 3130 3143 11731477
|
16299588
|
PMC1285065
|
CC BY
|
2021-01-05 08:00:25
|
no
|
PLoS Genet. 2005 Nov 18; 1(5):e67
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010067
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1628583910.1371/journal.pmed.0020334Research ArticleGenetics/Genomics/Gene TherapyEpidemiology/Public HealthStatisticsGeneticsResearch MethodsLocal Literature Bias in Genetic Epidemiology: An Empirical Evaluation of the Chinese Literature Gene Epidemiology: Local Literature BiasPan Zhenglun
1
Trikalinos Thomas A
2
3
Kavvoura Fotini K
2
Lau Joseph
3
Ioannidis John P.A
2
3
4
*1Department of Rheumatology, Shandong Provincial Hospital, Jinan 250021, Shandong, China2Clinical and Molecular Epidemiology Unit, Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina, Greece,3Institute for Clinical Research and Health Policy Studies, Department of Medicine, Tufts-New England Medical Center, Tufts University School of Medicine, Boston, Massachusetts, United States of America,4Biomedical Research Institute, Foundation for Research and Technology–Hellas, Ioannina, GreeceGwinn Marta Academic EditorU.S. Centers for Disease Control and PreventionUnited States of America*To whom correspondence should be addressed. E-mail: [email protected]
Competing Interests: The authors have declared that no competing interests exist.
Author Contributions: The original idea for this study arose in an electronic conversation between ZP and JPAI. JPAI drafted the protocol that was critically evaluated by all other authors. ZP performed all the database searches. ZP performed the data extraction that was duplicated by FKK and JL. TAT performed the statistical analyses. JPAI wrote the final draft. All authors interpreted the results, commented critically on the manuscript, and approved its final version.
12 2005 22 11 2005 2 12 e3349 5 2005 15 8 2005 Copyright: © 2005 Pan 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.
Selection Bias in Meta-Analyses of Gene-Disease Associations
Bias in Reporting of Genetic Association Studies
Background
Postulated epidemiological associations are subject to several biases. We evaluated whether the Chinese literature on human genome epidemiology may offer insights on the operation of selective reporting and language biases.
Methods and Findings
We targeted 13 gene-disease associations, each already assessed by meta-analyses, including at least 15 non-Chinese studies. We searched the Chinese Journal Full-Text Database for additional Chinese studies on the same topics. We identified 161 Chinese studies on 12 of these gene-disease associations; only 20 were PubMed-indexed (seven English full-text). Many studies (14–35 per topic) were available for six topics, covering diseases common in China. With one exception, the first Chinese study appeared with a time lag (2–21 y) after the first non-Chinese study on the topic. Chinese studies showed significantly more prominent genetic effects than non-Chinese studies, and 48% were statistically significant per se, despite their smaller sample size (median sample size 146 versus 268, p < 0.001). The largest genetic effects were often seen in PubMed-indexed Chinese studies (65% statistically significant per se). Non-Chinese studies of Asian-descent populations (27% significant per se) also tended to show somewhat more prominent genetic effects than studies of non-Asian descent (17% significant per se).
Conclusion
Our data provide evidence for the interplay of selective reporting and language biases in human genome epidemiology. These biases may not be limited to the Chinese literature and point to the need for a global, transparent, comprehensive outlook in molecular population genetics and epidemiologic studies in general.
Using the chinese literature as an example, John Ioannidis and colleagues show that selective reporting and language biases occur frequently in human genome epidemiology
==== Body
Introduction
Research conducted in non-English-speaking countries may be published either in English-language journals that are usually indexed in major international bibliographic databases or in domestic journals, many of which are not indexed in international databases. There is some empirical evidence that the decision to publish in international versus domestic journals may be influenced by the nature of the results: Significant results may be published in international journals, while nonsignificant results appear in the local literature, resulting in language bias (the “tower of Babel” bias) [1,2]. The opposite phenomenon, a reverse tower of Babel bias, nevertheless has also been described [3] in which most of the locally produced and published literature is spuriously statistically significant. Moreover, other investigators have questioned whether the inclusion or not of non-English studies makes any meaningful difference in the overall picture of the evidence [4].
The available evidence on these biases stems from the literature of randomized controlled trials. However, there are other fields in which language biases may be particularly important to appreciate. Genetics poses some special challenges. There are millions of polymorphisms in the human genome, and an exponentially increasing number of studies are trying to associate genetic polymorphisms with the risk of common diseases or treatment outcomes [5]. The risk conferred by each one of these genetic markers is usually small [5], with odds ratios between 1.1 and 1.4. Therefore, selective publication of studies with different results may potentially invalidate the overall picture about genetic risk factors. Moreover, there is major debate on whether there are differences in the strength of the genetic effects across people of different “racial” descent [6–8]. Language-related biases would tend to affect predominantly literature that refers to populations of specific “racial” descent, thus affecting the larger debate on “racial” descent differences.
The Chinese literature is a prominent example of possible bias, because a plethora of domestic scientific journals are not cataloged in international databases. China accounts for one-fifth of the world population, and this research is of major importance not only for China, but also internationally. It has been estimated that overall, for each internationally indexed publication from China, there are 18 publications in local nonindexed journals [9]. The consequences of potential selective publication and language biases for human genome epidemiology research and for biomedical research in general are unknown. Here we aimed to evaluate the extent to which genetic association studies are published in local Chinese journals not indexed in PubMed. We tried to understand whether the results of the Chinese literature differs from the results of the non-Chinese literature and what the implications would be for the total evidence on postulated epidemiological associations and inherent biases.
Methods
Definitions
The primary comparison addressed the results of Chinese versus non-Chinese studies. “Chinese studies” refers to studies performed in the People's Republic of China, regardless of the language of publication. All of them have been performed in people of Chinese descent. Chinese studies are further classified according to whether they are indexed in PubMed or not. “Non-Chinese studies” refers to studies performed outside of China, regardless of the language of publication and regardless of the “racial” descent of the studied populations. Non-Chinese studies are further classified according to whether they evaluated people of Asian or non-Asian descent.
Database of Meta-Analyses of Gene-Disease Associations
We used published meta-analyses of gene-disease associations with binary outcomes and unrelated subjects. Whenever a publication provided data on more than one “racial” descent group, these were split and counted as separate studies. We started from a dataset of 55 meta-analyses previously used in an evaluation of differences between small and larger genetic association studies with binary outcomes [10]. The exact search strategy and eligibility criteria for these meta-analyses have been described previously [10,11]. For each one of them, we updated searches until December 2004, in order to identify more recent meta-analyses on exactly the same topic and containing more studies. More comprehensive meta-analyses replaced the older ones. Then we focused only on meta-analyses in which at least 15 non-Chinese studies were already available. We took this approach because there is evidence that the early literature on gene-disease associations often provides unreliable, inflated results [11,12]. Moreover, Chinese studies may not appear for at least a few years after the appearance of the first non-Chinese studies, thus meta-analyses with few non-Chinese studies may not have had any Chinese studies published yet. Meta-analyses were selected regardless of whether or not they already included any studies from China or individuals of Chinese ethnic descent. None of these meta-analyses had access to Chinese journals not indexed in PubMed, and all included studies were PubMed-indexed.
Search for Additional Studies from the Chinese Literature
For each of the eligible meta-analyses, we searched the national Chinese database of biomedical literature (last search December 2004) for potentially additional gene-disease association studies published in local Chinese journals that would fulfill the eligibility criteria of the original meta-analysis. The Chinese Journal Full-Text Database covers 8,000 journals since 1994, and it is accessible with username and password via the Web site of Tsinghua University. We excluded family-based studies, since they are based on linkage analyses, and these have also been excluded from the original meta-analyses as well.
The search strategy for each topic used the name of each genetic marker (using the abbreviated name of the gene, the expanded name of the gene, and the polymorphism) in combination with terms pertaining to the disease and/or outcome of interest. Retrieved abstracts and articles were further screened for eligibility by the same native Chinese investigator (ZP) who performed the literature search. When in doubt, two other investigators (JPAI and JL), one of them Chinese-speaking (JL), decided on the study's eligibility.
Data Extraction
From each eligible Chinese study, we recorded the name of the first author, journal of publication, year of publication, ethnic descent, and data on the 2 × 2 table for the association (data necessary to derive the crude odds ratio and standard error thereof for the probed association). For consistency, the same genetic contrast was used as in the original meta-analysis. We also recorded whether the study was also indexed in PubMed.
We also examined, in each Chinese article, whether the disease was defined with specific criteria, whether any effort was described to ensure that the controls were indeed disease-free or otherwise appropriate, whether it was specified that genotyping was performed blinded to the clinical status, whether there was any mention that the disease-free controls were tested for conformity to Hardy-Weinberg equilibrium, whether any authors were involved from countries other than China, and whether the article was published in an international or national versus a local journal.
Data extraction was performed by a native Chinese investigator (ZP). Key data were independently verified by a second investigator (FKK) whenever tables in English were available or from another Chinese-speaking investigator (JL) otherwise. The few discrepancies were discussed and consensus was reached with a third arbitrator.
Data Analysis
Descriptive statistics summarized the number of studies, total sample size, number and percentage of studies with statistically significant results on their own, and year of publication. Sample sizes were compared between groups of studies with the Mann-Whitney U test and with median regression adjusted for topic (bootstrap p-values). The proportion of studies with statistically significant results was compared with the χ2 test.
For each meta-analysis and for each group of studies, we estimated the summary odds ratio with inverse variance random effects models, which allow for between-study heterogeneity and incorporate it in the calculations [13]. We tested for between-study heterogeneity with the χ2-distributed Q statistic (considered significant at p < 0.10) [13], and estimated its extent with the I
2 statistic. I
2 ranges between 0% and 100% and represents the proportion of between-study variability that can be attributed to heterogeneity rather than chance (considered large for values of 75% and higher) [14]. Given the prominent differences in effect sizes between different groups of studies, it was considered inappropriate to obtain an overall summary effect including all of them. Instead, we estimated whether the results of different groups of studies differed between themselves beyond chance. A standardized z-score statistic was employed, as previously described [15].
For Chinese studies, for each study we estimated the probability that it would have a formally statistically significant result at the α = 0.05 level, conditional on the sample size of its case and control groups, the genetic marker frequency in the controls, and the summary genetic effect seen across Chinese studies. This calculation was performed as a regular power calculation for a case-control study. The sum of these probabilities across Chinese studies (the expected number of studies with statistically significant results) was then compared to the observed number of statistically significant findings using a χ2 test.
We also compared PubMed-indexed versus not PubMed-indexed Chinese studies as to all the other study and quality characteristics mentioned in the Data Extraction section above.
Analyses were conducted in Intercooled Stata 8.2 (Stata Corp., College Park, Texas, United States) using the metan module. p-Values are two-tailed.
Results
Data on Chinese and Non-Chinese Studies
Thirteen published meta-analyses were found with at least 15 non-Chinese studies [16–26]. Data on any Chinese studies could be retrieved for 12 of those, and these 12 topics are considered from now on (for the association of DRD2 TaqIA polymorphism with alcoholism [26], no Chinese study was identified; Table 1). Overall, there were 161 eligible Chinese studies, only 20 of which were indexed in PubMed. Of the 20 Chinese studies indexed in PubMed (two on ID1, two on ID2, one on ID3, two on ID4, five on ID10, one on ID11, and seven on ID12; Table 1), only six had already been included in the published meta-analyses (one on ID11 and five on ID12), while the others were more recent; only seven of the 20 were published in full-text English journals. Of the 309 non-Chinese studies already included in the published meta-analyses, 44 pertained to populations of Asian descent (Japan, n = 25; Korea, n = 7; Chinese people outside of China, n = 5; Taiwan, n = 4; Malaysia, n = 2; and Singapore n = 1), and 265 to people of non-Asian descent (Figure 1).
Figure 1 Categorization of the Examined Genetic Association Studies
IQR, interquartile range; N, sample size (as median and interquartile range); StatSig, statistically significant at the 0.05 level.
Table 1 Eligible Meta-Analyses
For six topics we retrieved an extensive Chinese literature from the Chinese database (14–35 studies for each), while for the other topics the Chinese studies were sparse (four or fewer studies per topic) (Table 1). Chinese data were typically sparse if the disease was relatively uncommon in China compared with other countries, e.g., bladder cancer (bladder cancer is almost 10-fold less common in China than in Europe or the United States) [27] and alcoholism (at least until the early 1990s) [28]; or if the disease was not very common globally (e.g., systemic lupus erythematosus and schizophrenia). Chinese studies were plentiful if the disease was common (e.g., cancer in general, lung cancer, coronary heart disease, and diabetic nephropathy), with the exception of the postulated association of the ITGB3 gene with coronary artery disease, for which only one Chinese study was available.
With one exception, where the first Chinese study was published in the same year as the first non-Chinese study, the first Chinese study always appeared with a considerable time lag compared with the remaining world literature (2–21 y; Table 1).
Study Sample Sizes
The sample size for Chinese studies was significantly smaller than for non-Chinese studies (p < 0.001 both by U test and topic-adjusted median regression; Figure 1). Although non-Chinese studies of non-Asian descent populations overall seemed to be larger than studies on non-Chinese studies of Asian descent populations (p < 0.001 by U test), the difference was lost after adjusting for topic (p = 0.72). Chinese studies indexed or not indexed in PubMed did not differ in sample size (p = 0.79 by U test, p = 0.55 by median regression; Figure 1).
Statistically Significant Results
Overall, 78 (48%) of the 161 Chinese studies had formally statistically significant results. There was some heterogeneity in this proportion across topics (exact p = 0.041). Conversely, only 57 (18%) of 309 non-Chinese studies had significant results, despite the larger sample size, and the percentage differed greatly across the 12 topics (exact p < 0.001). As shown in Figure 1, the proportion of formally statistically significant studies differed between PubMed-indexed Chinese studies, non-PubMed-indexed Chinese studies, non-Chinese studies of Asian-descent populations, and non-Asian studies (65%, 46%, 27%, and 17%, respectively; p < 0.001 by χ2). None of the five studies on Chinese-descent people living outside of China had statistically significant results.
Changes in Study Sample Sizes and Significant Results over Time
The sample size of Chinese studies increased over time (Spearman correlation coefficient for publication year and sample size, 0.32, p < 0.001), while this was not seen for non-Chinese studies (correlation coefficient 0.00, p = 0.95). As for non-Chinese studies, the proportion of Chinese studies with formally significant results did not increase over time; if anything, there was a trend towards decrease (47/89 [53%] in 1993–2000 versus 41/72 [43%] in 2001–2004; p = 0.27).
Comparison of Genetic Effects
Table 2 summarizes the genetic effect sizes. As shown, whenever there was a sizeable literature of Chinese studies, the gene-disease association was always formally significant in both non-Chinese and Chinese studies, but Chinese studies always showed a larger genetic effect than the non-Chinese studies (Figure 1). In five of the six topics the observed difference was even beyond chance (p < 0.05 on the z-score). Even with limited data, Chinese studies suggested larger estimates than non-Chinese studies also in the other three topics where there was some overall evidence for the presence of a gene-disease association; the genetic effect difference was beyond chance in one of the three topics (Table 2).
Table 2 Genetic Effects in Chinese and Non-Chinese Studies
PubMed-indexed Chinese studies were too few for formal comparisons, but the available data suggested that they often tended to provide extreme estimates of genetic effects (Figure 2). In three of the five topics where at least two such studies were available, their summary estimate was the most extreme observed compared with any other group of studies (non-PubMed Chinese, non-Chinese Asian, and non-Chinese non-Asian).
Figure 2 Meta-Analyses of Gene-Disease Associations in a Large Number of Both Non-Chinese and Chinese Studies
Each study is shown by its odds ratio and 95% confidence intervals (CIs). The box of the point estimate is proportional to the study weight. Also shown are summary estimates by random effects calculations (diamonds). Summary estimates are obtained separately for Chinese studies indexed in PubMed (red), Chinese studies not indexed in PubMed (pink), non-Chinese studies of Asian descent populations (green), and studies of persons of non-Asian descent (blue). An odds ratio of 1 means no genetic effect, odds ratios larger than 1 mean genetic predisposition, and odds ratios less than 1 mean genetic protection.
Non-Chinese studies of Asian descent populations were available for eight topics. In seven of the eight cases, the estimated genetic effect size was stronger in these Asian-descent studies than in the non-Chinese non-Asian descent studies (Table 3). The difference was beyond chance in two topics (the associations of MTHFR C677T polymorphism with coronary heart disease [ID10], and of GSTM1 gene deletion with lung cancer [ID12]). In topics for which several studies of different groups were available, the non-Chinese studies of Asian-descent populations seemed to have effect sizes somewhere between the effect sizes of Chinese studies and non-Asian studies (see Figure 1).
Table 3 Comparison of Genetic Effects in Non-Chinese Studies of Asian-Descent Populations and Non-Asian-Descent Studies
Expected versus Observed Significant Findings in Chinese Studies
Power calculations based on asymptotic statistical testing suggested that even if the large summary genetic effects claimed by the Chinese studies were genuine, one would expect 56.6 formally statistically significant studies, substantially fewer than the 78 observed in the database (p < 0.001). Based on exact statistical testing, one would expect 61.1 significant studies instead of the 81 observed (p = 0.001).
Qualitative Comparison of Chinese Studies Indexed versus Not Indexed in PubMed
PubMed-indexed Chinese studies did worse than Chinese studies not indexed in PubMed in defining disease with specific criteria (17/20 [85%] versus 137/141 [97%], respectively; exact p = 0.042), and in ascertaining the eligibility of controls (13/20 [65%] versus 129/141 [92%], respectively; exact p = 0.003). However, the only three Chinese studies mentioning blinding of the genotyping personnel to disease status were PubMed-indexed (exact p = 0.002). There was no difference in testing for violations of the Hardy-Weinberg law in the controls (5/20 [20%] versus 38/141 [27%], respectively; exact p = 1.00). Only PubMed-indexed Chinese studies had any representation of authors from countries other than China (4/20 [20%] versus 0/141 [0%], respectively; exact p < 0.001). As expected, almost all (19/20 [95%]) PubMed-indexed Chinese studies were published in international or national Chinese journals, while only 22 (16%) of the 141 studies not indexed in PubMed were published in national journals.
Discussion
This empirical evaluation reveals a large Chinese literature on human genome epidemiology that deserves more attention from the international community. The vast majority of this literature does not reach PubMed. Chinese studies usually appear with a time lag of several years after an epidemiological association is first postulated in the world literature, but many such studies are published, especially when the disease is perceived to be common in China. Chinese studies typically suggest much stronger genetic effects than non-Chinese studies, and this may be even more prominent for the few studies that reach PubMed. Although Chinese studies are smaller than non-Chinese studies and thus even more underpowered [5], surprisingly half of them reach formal statistical significance for the evaluated gene-disease association. This exaggeration is seen across very diverse topics.
The larger genetic effects in Chinese studies are unlikely to reflect genuine heterogeneity in the effects of genetic risk factors across various “racial” descent populations [8]. Heterogeneity due to ancestry should not have led always to larger effect sizes in all probed gene-disease associations. Therefore, the most likely explanation is publication bias against “negative” results [29–32] or other selection biases in the chase for statistically significant findings [33]. This explanation is further supported by our analysis of the expected number of statistically significant findings. Even if the average genetic effects in the Chinese studies were indeed as large as those observed, one would expect far fewer Chinese studies to have reached formal statistical significance on their own, given their small sample sizes. The alternative explanation that Chinese investigators may be targeting high-risk populations with particularly strong genetic effects is unlikely given these data.
Language may be a marker for other confounding characteristics of these studies, or even of the whole research and publication milieu. Moreover, even within the English-language studies, strong biases may occasionally operate in the confirmation process. Cultural issues may also be involved with unstated pressures to find positive results for various reasons in different settings around the globe. Various compromises of research quality may ensue.
We focused on gene-disease associations for which a considerable number of studies have been published in the English language. It is possible that there could be a reluctance to submit and publish “negative” or inconclusive results when a large body of English-language literature has shown the presence of genetic effects. Also attempts to confirm multiply supported findings may be more likely to be made, especially with limited resources. However, such pressure for unilateral confirmation destroys the independence and thus also the importance of confirmation.
Our observation is reminiscent also of the randomized trial literature on acupuncture, where studies from China, Russia, Japan, Hong Kong, and Taiwan almost always yielded statistically significant results, in contrast to studies performed in other countries [3]. A predilection for the dissemination of statistically significant results in some non-English speaking countries has also been suspected in other fields, such as lung cancer chemotherapy trials [34]. To our knowledge, there has been no prior documentation of this phenomenon in molecular medicine. Given the rapid pace of production of information in molecular genetics and other modern disciplines, this bias may become a serious problem in the appraisal of cutting-edge science and may jeopardize the credibility of molecular discovery research.
We also found some evidence that superimposed language bias [2] is also operating in this literature. Typically, the few PubMed-indexed Chinese studies showed the most extreme genetic effects, and two-thirds of them reached formally statistically significant results on their own, even though their sample size was very small. Therefore, analyses limited to PubMed-indexed studies may sometimes yield spurious results, if the summary estimates are driven by these extreme findings. PubMed-indexed Chinese studies had worse quality ratings in case and control definitions and ascertainment than Chinese studies not indexed in PubMed. Language bias may not be limited to China, but may also be pertinent to other Asian countries with considerable scientific production, and beyond. We found that non-Chinese Asian studies also tend consistently to show relatively larger genetic effects than non-Asian studies, although data were too sparse to be definitive. The relative extent of selective reporting, publication bias, and language bias is difficult to disentangle here and may vary across topics and across local literatures. It would also be useful to analyze the local literatures for Japanese and Korean studies, where a considerable number of local journals also exist.
The Chinese literature is essential for the evaluation of evidence on genetic risk factors. China is making rapid scientific progress in this field, as in many others. It is already participating in the Human Genome project, and the Southern China National Genome Research Center established in Shanghai in 2001 creates new frontiers for gene-disease association studies. Evidence on population genetics, as well as for any other field pertinent to population health, is extremely important to obtain for China from a global perspective. Moreover, it is unlikely that biases are limited to China, as we discussed above. Also, European and American studies are not necessarily unbiased. There is strong evidence that early-published European and American studies that appeared in the most prestigious journals tended to have inflated results [11,35,36].
Here we did not update further the existing non-Chinese data from the published meta-analyses. Our investigation focused on the Chinese literature, and we tried to focus on meta-analyses with a large number of included studies that should hopefully have reached a stable effect estimate. Nevertheless, for at least two of the postulated associations examined here (the associations of ACE with cardiac outcomes), a very large study [37] conducted after the meta-analysis found absolutely no effect, while the previous studies had found modest, but statistically significant, genetic effects. Thus not only was the discrepancy against the Chinese studies even larger than what was found in our analyses, but the evidence from the earlier European-descent studies, in particular the smaller ones [16], had also been biased.
Since most effect sizes in genetic epidemiology, and most other molecular medicine fields, are small or modest, one wonders whether many of the postulated associations are generated from the interplay of various reporting and local literature biases that leave no country immune. In some of these postulated associations, the observed effect sizes may simply be estimates of the prevailing bias [38].
One might argue that the inclusion of poor-quality research may contaminate the better literature rather than provide a more accurate, comprehensive picture. Large-scale aggregate evidence may arrive at erroneous conclusions if studies are automatically included without some critical appraisal. However, it is unfair to judge the quality of research on the basis of its regional origin. Chinese studies may often be as good as or even better than many or most studies from countries publishing routinely in the English language [39–41]. Efforts to improve the quality of research around the globe should run in parallel with enhanced access to global research results.
Our findings have two broad implications. First, language bias may be important to consider in meta-analyses of observational studies in general, and its impact may be as large as or larger than its impact on randomized evidence. Second, human genome epidemiology in particular is a global enterprise, and a critical and comprehensive global view is important to decipher artifacts from true genetic effects. Large studies are useful to validate postulated gene-disease associations [12]. However, such studies are difficult to conduct, they are not completely immune from biases, and their targets must be carefully selected given the plethora of test hypotheses in molecular genetics [5,42]. Besides large studies, registration of investigators and data collections is useful to consider [42,43]. In contrast to randomized trials [44], study registration is impractical in molecular medicine, since investigators would be reluctant to share their hypotheses in public. However, if all investigators working on the genetics of a specific disease were registered in a common network, then it would be easier to trace additional unpublished or non-indexed data. Common networks would also, hopefully, help to improve the quality of research. Such networks should aim for a global, inclusive outlook. The Chinese research output, as well as the output of other non-English-speaking countries, should be appropriately captured. Failure to maintain a global outlook may result in a scientific literature that is driven by the opportunistic dissemination of selected results.
Patient Summary
Background
There are many different places that medical research can be published. However, some research is never published, which leads to so-called publication bias. One of the biggest divides is between English and non-English research. Research done in non-English-speaking countries can be published in English journals that are usually indexed in major international databases, but more often is published in domestic journals, many of which are not indexed in international databases. This selective publication is called language bias. Scientists have questioned what difference the inclusion or not of non-English studies makes to the total evidence. Publication bias is of concern, especially in genetics, which is a very fast-moving area of research.
Why Was This Study Done?
China is a prominent example of a nation with many domestic journals that are not indexed in the international databases. This study looked at Chinese genetic association studies. Understanding the quality and findings of genetics research in China, home to one-fifth of the world's population, is essential for the evaluation of evidence on genetic risk factors. The authors hoped this study would help understand more about selective reporting and language biases.
What Did the Researchers Do and Find?
They found that there were many studies (14–35 per topic) for each of the topics they chose to assess, which covered diseases common in China. Generally, Chinese studies appeared a considerable time (two to 21 years) after the first non-Chinese study on the topic. Chinese studies showed stronger genetic effects than non-Chinese studies, despite in some cases having smaller sample sizes. The largest genetic effects were often seen in Chinese studies indexed in Western databases.
What Do These Findings Mean?
It seems that there is a combination of selective reporting of studies with interesting findings, and language biases in human gene studies. The main reason most studies didn't appear in the international literature was probably a combination of publication bias and superimposed language bias. It is important to note that such biases are not limited to Chinese literature. Researchers should consider language bias when doing analyses of groups of studies, and efforts to improve the quality of research around the globe should run in parallel with enhanced access to global research results.
Where Can I Get More Information Online?
The Cochrane Collaboration has a small section explaining the different types of publication bias:
http://www.cochrane-net.org/openlearning/HTML/mod15-2.htm
BMJ has a presentation on publication bias by a BMJ editor:
http://bmj.bmjjournals.com/talks/publication_bias/index.htm
TAT and FKK were funded by PENED grants from the General Secretariat for Research and Technology, Greece, and the European Commission. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Citation: Pan Z, Trikalinos TA, Kavvoura FK, Lau J, Ioannidis JPA (2005) Local literature bias in genetic epidemiology: An empirical evaluation of the Chinese literature. PLoS Med 2(12): e334.
==== Refs
References
Gregoire G Derderian F Le Lorier J Selecting the language of the publications included in a meta-analysis: is there a Tower of Babel bias? J Clin Epidemiol 1995 48 159 163 7853041
Egger M Zellweger-Zahner T Schneider M Junker C Lengeler C Language bias in randomised controlled trials published in English and German Lancet 1997 350 326 329 9251637
Vickers A Goyal N Harland R Rees R Do certain countries produce only positive results? A systematic review of controlled trials Control Clin Trials 1998 19 159 166 9551280
Juni P Holenstein F Sterne J Bartlett C Egger M Direction and impact of language bias in meta-analyses of controlled trials: Empirical study Int J Epidemiol 2002 31 115 123 11914306
Ioannidis JP Genetic associations: False or true? Trends Mol Med 2003 9 135 138 12727138
Thomas DC Witte JS Population stratification: A problem for case-control studies of candidate-gene associations? Cancer Epidemiol Biomarkers Prev 2002 11 505 512 12050090
Wacholder S Rothman N Caporaso N Counterpoint: Bias from population stratification is not a major threat to the validity of conclusions from epidemiological studies of common polymorphisms and cancer Cancer Epidemiol Biomarkers Prev 2002 11 513 520 12050091
Ioannidis JP Ntzani EE Trikalinos TA “Racial” differences in genetic effects for complex diseases Nat Genet 2004 36 1312 1318 15543147
International Working Party to Promote and Revitalise Academic Medicine Academic medicine: The evidence base BMJ 329 789 792
Ioannidis JP Trikalinos TA Ntzani EE Contopoulos-Ioannidis DG Genetic associations in large versus small studies: An empirical assessment Lancet 2003 361 567 571 12598142
Ioannidis JP Ntzani EE Trikalinos TA Contopoulos-Ioannidis DG Replication validity of genetic association studies Nat Genet 2001 29 306 309 11600885
Trikalinos TA Ntzani EE Contopoulos-Ioannidis DG Ioannidis JP Establishment of genetic associations for complex diseases is independent of early study findings Eur J Hum Genet 2004 12 762 769 15213707
Lau J Ioannidis JP Schmid CH Quantitative synthesis in systematic reviews Ann Intern Med 1997 127 820 826 9382404
Higgins JP Thompson SG Quantifying heterogeneity in a meta-analysis Stat Med 2002 21 1539 1558 12111919
Cappelleri JC Ioannidis JP Schmid CH de Ferranti SD Aubert M Large trials vs meta-analysis of smaller trials: how do their results compare? JAMA 1996 276 1332 1338 8861993
Agerholm-Larsen B Nordestgaard BG Tybjaerg-Hansen A
ACE gene polymorphism in cardiovascular disease. Meta-analyses of small and large studies in whites Arterioscler Thromb Vasc Biol 2000 20 484 492 10669647
Di Castelnuovo A de Gaetano G Donati MB Iacoviello L Platelet glycoprotein receptor IIIa polymorphism PLA1/PLA2 and coronary risk: A meta-analysis Thromb Haemost 2001 85 626 633 11341496
Karassa FB Trikalinos TA Ioannidis JPA the FcgRIIa-SLE Meta-Analysis Investigators Role of the Fcg receptor IIa polymorphism in susceptibility to systemic lupus erythematosus and lupus nephritis Arthritis Rheum 2002 46 1563 1571 12115187
Jonsson EG Flyckt L Burgert E Crocq MA Forslund K Dopamine D3 receptor gene Ser9Gly variant and schizophrenia: association study and meta-analysis Psychiatr Genet 2003 13 1 12 12605094
Benhamou S Lee WJ Alexandrie AK Boffetta P Bouchardy C Meta- and pooled analyses of the effects of glutathione S-transferase M1 polymorphisms and smoking on lung cancer risk Carcinogenesis 2002 23 1343 1350 12151353
Klerk M Verhoef P Clarke R Blom HJ Kok FJ MTHFR 677C–>T polymorphism and risk of coronary heart disease: a meta-analysis JAMA 2002 288 2023 2031 12387655
Engel LS Taioli E Pfeiffer R Garcia-Closas M Marcus PM Pooled analysis and meta-analysis of glutathione S-transferase M1 and bladder cancer: A HuGE review Am J Epidemiol 2002 156 95 109 12117698
Tarnow L Gluud C Parving HH Diabetic nephropathy and the insertion/deletion polymorphism of the angiotensin-converting enzyme gene Nephrol Dial Transplant 1998 13 1125 1130 9623540
Marcus PM Vineis P Rothman N NAT2 slow acetylation and bladder cancer risk: A meta-analysis of 22 case-control studies conducted in the general population Pharmacogenetics 2000 10 115 122 10761999
Krontiris TG Devlin B Karp DD Robert NJ Risch N An association between the risk of cancer and mutations in the HRAS1 minisatellite locus N Engl J Med 1993 329 517 523 8336750
Noble EP The D2 dopamine receptor gene: A review of association studies in alcoholism and phenotypes Alcohol 1998 16 33 45 9650634
Parkin DM Bray F Ferlay J Pisani P Global cancer statistics, 2002 CA Cancer J Clin 2005 55 74 108 15761078
Li HZ Rosenblood L Exploring factors influencing alcohol consumption patterns among Chinese and Caucasians J Stud Alcohol 1994 55 427 433 7934050
Easterbrook PJ Berlin JA Gopalan R Matthews DR Publication bias in clinical research Lancet 1991 337 867 872 1672966
Dickersin K Min YI Publication bias: The problem that won't go away Ann N Y Acad Sci 1993 703 135 148 8192291
Munafo MR Clark TG Flint J Assessing publication bias in genetic association studies: Evidence from a recent meta-analysis Psychiatry Res 2004 129 39 44 15572183
Agema WR Jukema JW Zwinderman AH van der Wall EE A meta-analysis of the angiotensin-converting enzyme gene polymorphism and restenosis after percutaneous transluminal coronary revascularization: Evidence for publication bias Am Heart J 2002 144 760 768 12422143
Cohen J The earth is round (P-less-than .05) Amer Psychol 1994 49 997 1003
Ioannidis JP Polycarpou A Ntais C Pavlidis N Randomised trials comparing chemotherapy regimens for advanced non-small cell lung cancer: Biases and evolution over time Eur J Cancer 2003 39 2278 2287 14556918
Gelernter J Goldman D Risch N The A1 allele at the D2 dopamine receptor gene and alcoholism. A reappraisal JAMA 1993 269 1673 1677 8095994
Ioannidis JP Contradicted and initially stronger effects in highly cited clinical research JAMA 2005 294 218 228 16014596
Keavney B McKenzie C Parish S Palmer A Clark S Large-scale test of hypothesised associations between the angiotensin-converting-enzyme insertion/deletion polymorphism and myocardial infarction in about 5000 cases and 6000 controls. International Studies of Infarct Survival (ISIS) Collaborators Lancet 2000 355 434 442 10841123
Ioannidis JP Why most published research findings are false PLoS Medicine 2005 2 e124 16060722
Salanti G Amountza G Ntzani EE Ioannidis JP Hardy-Weinberg equilibrium in genetic association studies: An empirical evaluation of reporting, deviations, and power Eur J Hum Genet 2005 13 840 848 15827565
Bogardus ST Concato J Feinstein AR Clinical epidemiological quality in molecular genetic research: The need for methodological standards JAMA 1999 281 1919 1926 10349896
Attia J Thakkinstian A D'Este C Meta-analyses of molecular association studies: Methodologic lessons for genetic epidemiology J Clin Epidemiol 2003 56 297 303 12767405
Ioannidis JP Rosenberg PS Goedert JJ O'Brien TR International Meta-analysis of HIV Host Genetics. Commentary: Meta-analysis of individual participants' data in genetic epidemiology Am J Epidemiol 2002 156 204 210 12142254
Ioannidis JP Bernstein J Boffetta P Danesh J Dolan S A network of investigator networks in human genome epidemiology Am J Epidemiol 2005 162 302 304 16014777
De Angelis C Drazen JM Frizelle FA Haug C Hoey J Clinical trial registration: A statement from the International Committee of Medical Journal Editors N Engl J Med 2004 351 1250 1251 15356289
|
16285839
|
PMC1285066
|
CC BY
|
2021-01-05 10:39:20
|
no
|
PLoS Med. 2005 Dec 22; 2(12):e334
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020334
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1636391110.1371/journal.pmed.0020409PerspectivesGenetics/Genomics/Gene TherapyGeneticsResearch MethodsSelection Bias in Meta-Analyses of Gene-Disease Associations PerspectivesTang Jin Ling Jin Ling Tang is the director of the Chinese Cochrane Centre Hong Kong Branch, based in the Faculty of Medicine, the Chinese University of Hong Kong, Shatin, New Territories, Hong Kong, The People's Republic of China. E-mail: [email protected]
Competing Interests: The author declares that there are no competing interests.
12 2005 22 11 2005 2 12 e409Copyright: © 2005 Jin Ling Tang.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.
Local Literature Bias in Genetic Epidemiology: An Empirical Evaluation of the Chinese Literature
Jin-Ling Tang discusses the issues arising from Ioannidis and colleagues' paper on biases in the Chinese genetic epidemiology literature, and possible solutions
==== Body
Many studies have too small a sample size for their findings to be conclusive, but large studies are expensive and time-consuming. Meta-analysis is an alternative to conducting large studies in tackling the problem of small sample size, by combining available small studies to increase the total sample size. Since the 1980s, meta-analysis has been widely used in summarizing results from clinical trials of medical interventions and has also recently gained increasing attention in studying gene-disease associations. However, selection bias may occur in meta-analyses due to the inability to identify and include all conducted and relevant studies. Such selection bias can cause exaggerated or even false-positive gene-disease associations [1].
Failure to include all relevant studies is largely caused by selective publication of studies with certain results (publication bias), and the inability to identify studies published in languages other than English (language bias). Selection bias has been well recognized in meta-analyses of clinical trials [2–4]. Less is known about selection bias in meta-analyses of studies of gene-disease associations; such studies generally address weak associations and thus are particularly vulnerable to biases.
A Study of the Chinese Literature
In their study published in this issue of PLoS Medicine, Pan and colleagues compared genetic studies conducted in mainland China with those from other places [5]. The researchers identified 12 gene-disease associations and compared a total of 161 Chinese studies and 309 non-Chinese studies. The Chinese studies were on average smaller in sample size than non-Chinese studies and appeared in the literature a few years after the first non-Chinese studies. Chinese studies in general reported a stronger gene-disease association and more frequently a statistically significant result. These two characteristics were more likely to occur in Chinese studies identified through PubMed than in those accessible only locally.
These findings suggest a variation or heterogeneity in the strength of the gene-disease association (often expressed in an odds ratio) observed between Chinese and non-Chinese studies. These studies are primarily case-control studies. Many factors may contribute to the variation in the estimate of odds ratio across such studies, such as the genetic make-up of the population studied, the type of patients included, the selection of controls, the quality of the study design, and the quality of the laboratory work. These factors could lead to either over- or under-estimation of the true odds ratio. However, it is difficult to conceive that any single factor, or combination of these factors, could consistently cause the exaggerated odds ratio in Chinese studies in all the topics (gene-disease associations) examined by Pan and colleagues. Selective publication is therefore a very likely and worrying explanation for their findings.
Implications for Clinical Practice and Research
Selective publication can cause publication bias, which in turn could lead to false gene-disease associations in meta-analyses. It would be a disaster if a genetic screening program (in which healthy people are tested for a gene and offered a treatment if they test positive) were based on such a false association. Even if such a false gene-disease association were only subjected to further related investigations, this would be a waste of valuable resources for medical research.
Selective publication of positive studies in China and a few other Asian countries has been observed in clinical trials of acupuncture [6,7]. However, selective publication by no means exists in only the Chinese literature. It is probably a common phenomenon in the entire field of biomedical research. Given the fact that positive studies are more likely to be published than negative ones, and given the pressure on researchers worldwide to publish in indexed journals (especially in international journals with high impact factors), selective publication is likely to continue in the foreseeable future. As compared with English-speaking countries, selective publication is perhaps more likely to occur in non-English-speaking countries where there are a small number of indexed journals to publish local studies.
Addressing the Problem
Journals accessible through PubMed or other major biomedical databases are unlikely to have the same mechanism of selection for publication as local journals that are less accessible to researchers outside the country, such as the Chinese journals. Thus, meta-analyses that include only internationally accessible studies (which is, currently, often the case for meta-analyses) are likely to have language or location bias. Meta-analyses that included only local studies could be even worse, as implied by Pan and colleagues' study. Inclusion of every study published worldwide would probably still not totally solve the problem, as many studies are never published or their publication is delayed. Odds ratios thus estimated would normally be an over-estimate.
Registration of studies is ideal and has been widely advocated for clinical trials [8]. Before such registration becomes universal practice, it would be important for journals, in selecting papers for publication, to emphasize the quality of the study rather than the size and direction of the odds ratio and the p-value of the statistical test. However, such an emphasis on quality (rather than the size and direction of the odds ratio) would not be much help to researchers who are currently doing meta-analyses.
Current researchers must strive to not only identify relevant studies but also examine the possibility of publication bias in the results. Although better tools have yet to be developed [9,10], current methods for detection and adjustment for publication bias in meta-analyses of clinical trials would be useful for meta-analyses of gene-disease associations [1,11,12].
Citation: Tang JL (2005) Selection bias in meta-analyses of gene-disease associations. PLoS Med 2(12): e409.
==== Refs
References
Thornton A Lee P Publication bias in meta-analysis: Its causes and consequences J Clin Epidemiol 2000 53 207 216 10729693
Easterbrook PJ Berlin JA Gopalan R Matthews DR Publication bias in clinical research Lancet 1991 337 867 872 1672966
Dickersin K Min YI Meinert CL Factors influencing publication of research results: Follow-up of applications submitted to two institutional review boards JAMA 1992 267 374 378 1727960
Egger M Davey-Smith G Bias in location and selection of studies BMJ 1998 316 61 66 9451274
Pan ZL Trikalinos TA Kavvoura FK Lau J Ioannidis JPA Local literature bias in genetic epidemiology: An empirical evaluation of the Chinese literature PLoS Med 2005 2 e334 10.1371/journal.pmed.0020334 16285839
Vickers A Goyal N Harland R Rees R Do certain countries produce only positive results? A systematic review of controlled trials Control Clin Trials 1998 19 159 166 9551280
Tang JL Zhan SY Ernst E Review of randomised controlled trials of traditional Chinese medicine Br Med J 1999 319 160 161 10406751
De Angelis C Drazen JM Frizelle FA Haug C Hoey J Clinical trial registration: A statement from the International Committee of Medical Journal Editors N Engl J Med 2004 351 1250 1251 15356289
Tang JL Liu JLY Misleading funnel plot for detection of bias in meta-analysis J Clin Epidemiol 2000 53 477 484 10812319
Macaskill P Walter SD Trwig L A comparison of methods to detect publication bias in meta-analysis Stat Med 2001 20 641 654 11223905
Hedges LV Modeling publication selection effects in meta-analysis Stat Sci 1992 7 246 255
Duval S Tweedie R A non-parametric “Trim and Fill” method of accounting for publication bias in meta-analysis J Am Stat Assoc 2000 95 89 98
|
16363911
|
PMC1285067
|
CC BY
|
2021-01-05 10:39:26
|
no
|
PLoS Med. 2005 Dec 22; 2(12):e409
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020409
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020419SynopsisGenetics/Genomics/Gene TherapyEpidemiology/Public HealthStatisticsGeneticsResearch MethodsBias in Reporting of Genetic Association Studies Synopsis12 2005 22 11 2005 2 12 e419This is an open-access article distributed under the terms of the Creative Commons Public Domain Declaration, which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose.2005This is an open-access article distributed under the terms of the Creative Commons Public Domain declaration, which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose.
Local Literature Bias in Genetic Epidemiology: An Empirical Evaluation of the Chinese Literature
==== Body
One of the tools in the scientist's armory for resolving a medical issue or consolidating a body of clinical trials is the systematic review of the published medical literature. This technique involves doing a literature search and critical appraisal of individual studies, and in addition, may also use statistical techniques to combine the results of these studies. One of the aims of such reviews is to assess and then, ideally, include all appropriate studies that address the question of the review. But finding all studies is not always possible, and researchers have no way of knowing what they have missed. But does it matter if some studies are left out?
It would definitely matter if the missing studies differed significantly from the included ones. And the worst-case scenario is that the accumulation of evidence might point to the wrong answer if the studies included are unrepresentative of all those that have been done.
Studies of publication bias have noted that papers with significant positive results are easier to find than those with nonsignificant or negative results. As a result, overrepresentation of positive studies in systematic reviews might mean that such reviews are biased toward a positive result. Publication bias is just one in a group of related biases, all of which potentially lead to overrepresentation of significant or positive studies in systematic reviews. Other types of bias include time lag bias (positive studies are more likely to be published rapidly); multiple publication bias (positive studies are more likely to be published more than once); citation bias (positive studies are more likely to be cited by others); and language bias (positive studies are more likely to be published in English).
In PLoS Medicine, John Ioannidis and colleagues have taken a closer look at bias in Chinese genetics studies. Research done in non-English-speaking countries has two outlets. A study might be published in English-language journals, which are usually indexed in major international bibliographic databases such as PubMed, or in domestic journals, many of which are not indexed in international databases. The Chinese literature is a prominent example of where domestic scientific journals are not catalogued in international databases. There is some evidence that the decision to publish in international versus domestic journals might be influenced by the results. For example, significant results are often published in international journals, whereas nonsignificant results appear in the local literature, resulting in a language bias—although, the reverse situation has also been described.
Different results in local literature studies
Genetics studies pose particular problems for impartial reporting. There are millions of polymorphisms in the human genome, and an exponentially increasing number of studies are trying to associate genetic polymorphisms with risk of disease or treatment outcomes. Selective publication might invalidate the overall picture of genetic risk factors.
The authors examined 13 gene–disease associations. Studies were more likely to be published when the disease was considered common in China. They found 161 Chinese studies on 12 of these gene–disease associations, only 20 of which were indexed in PubMed. Chinese studies had significantly more prominent genetic effects than non-Chinese studies, and 48% were statistically significant per se, despite their smaller sample size. Moreover, the largest, most exaggerated genetic effects were often seen in PubMed-indexed Chinese studies. Chinese studies usually appeared several years after their equivalent was first postulated in the world literature.
The larger genetic effects in Chinese studies are unlikely to reflect genuine heterogeneity and are more likely to do with publication bias operating within the Chinese literature, say the authors. It is possible that there was reluctance to submit and publish negative or inconclusive results when a large body of English-language literature has shown the presence of genetic effects. However, such “forced” confirmation negates the importance of independent confirmation of research results. This problem is probably not limited to the Chinese literature. These phenomena haven't been noted in molecular medicine before, but could become a serious problem in such a fast-moving field. Moreover, the inclusion of poor-quality research and additional selectively reported data may contaminate the better literature rather than provide a more accurate, comprehensive picture.
The findings have two broad implications. First, language bias might be important to consider in meta-analyses of observational studies, where its effect might be larger than its effect on randomized evidence. Second, because human genome epidemiology is a global enterprise, a comprehensive global view is important to help decipher artifacts from true genetic effects. The Chinese literature in particular will be essential for the evaluation of evidence on genetic risk factors. China is making rapid scientific progress in this field and joining in international collaborative projects, such as the Human Genome Project. To develop a global perspective, one way forward might be for all investigators working on the genetics of a specific disease to register with a common network, making it easier to trace additional unpublished or nonindexed data.
|
0
|
PMC1285068
|
CC0
|
2021-01-05 10:39:28
|
no
|
PLoS Med. 2005 Dec 22; 2(12):e419
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020419
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1628734210.1371/journal.pmed.0030010Research ArticleInfectious DiseasesMicrobiologyVirologyEpidemiology/Public HealthHIV/AIDSObstetrics/GynecologyPediatricsInfectious DiseasesHIV Infection/AIDSPediatricsObstetricsPregnancyMaternal–Fetal Microtransfusions and HIV-1 Mother-to-Child Transmission in Malawi Placental Microtransfusions and HIV MTCTKwiek Jesse J
1
*Mwapasa Victor
2
Milner Danny A Jr
3
Alker Alisa P
1
Miller William C
1
Tadesse Eyob
4
Molyneux Malcolm E
5
Rogerson Stephen J
6
Meshnick Steven R
1
1Department of Epidemiology, University of North Carolina, Chapel Hill, North Carolina, United States of America2Department of Community Health, College of Medicine, University of Malawi, Blantyre, Malawi3Department of Pathology, Brigham and Women's Hospital, Boston, Massachusetts, United States of America4Department of Obstetrics and Gynaecology, College of Medicine, University of Malawi, Blantyre, Malawi5Malawi-Liverpool-Wellcome Trust Clinical Research Programme, College of Medicine, University of Malawi, Blantyre, Malawi6Department of Medicine, University of Melbourne, Parkville, Victoria, AustraliaGuay Laura Academic EditorJohns Hopkins Medical SchoolUnited States of America*To whom correspondence should be addressed. E-mail: [email protected]
Competing Interests: The authors have declared that no competing interests exist.
Author Contributions: JJK participated in hypothesis generation, data collection, and data analysis, and was the primary author of this study. VM participated in the study design, data collection, and the writing of the manuscript. DAM diagnosed chorioamnionitis. APA contributed to the data analysis. WCM suggested the case-cohort design and participated in data analysis and manuscript preparation. ET contributed to data collection and was the Chief Obstetrician. MEM participated in participant recruitment and study management/supervision. SJR diagnosed placental malaria, developed the hypothesis, participated in data analysis, and provided study oversight. SRM developed the hypothesis; participated in the study design, data analysis, and study coordination; and was the Principal Investigator. All of the authors edited and revised the final manuscript.
1 2006 22 11 2005 3 1 e102 8 2005 7 10 2005 Copyright: © 2006 Kwiek 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.
Placental Microtransfusions Associated with Increased HIV Transmission from Mother to Child
Background
Between 25% and 35% of infants born to HIV-infected mothers become HIV-1 infected. One potential route of mother-to-child transmission (MTCT) could be through a breakdown in the placental barrier (i.e., maternal–fetal microtransfusions).
Methods and Findings
Placental alkaline phosphatase (PLAP) is a 130-kD maternal enzyme that cannot cross the intact placental barrier. We measured PLAP activity in umbilical vein serum as an indicator of maternal–fetal microtransfusion, and related this to the risk of HIV-1 MTCT. A case-cohort study was conducted of 149 women randomly selected from a cohort of HIV-1-infected pregnant Malawians; these women served as a reference group for 36 cases of in utero MTCT and 43 cases of intrapartum (IP) MTCT. Cord PLAP activity was measured with an immunocatalytic assay. Infant HIV status was determined by real-time PCR. The association between cord PLAP activity and HIV-1 MTCT was measured with logistic regression using generalized estimating equations. Among vaginal deliveries, PLAP was associated with IP MTCT (risk ratio, 2.25 per log10 ng/ml PLAP; 95% confidence interval, 0.95–5.32) but not in utero MTCT. In a multivariable model adjusted for HIV-1 RNA load, chorioamnionitis, and self-reported fever, the risk of IP MTCT almost tripled for every log10 increase in cord PLAP activity (risk ratio, 2.87; 95% confidence interval, 1.05–7.83).
Conclusion
These results suggest that during vaginal deliveries, placental microtransfusions are a risk factor for IP HIV-1 MTCT. Future studies are needed to identify factors that increase the risk for microtransfusions in order to prevent IP HIV-1 MTCT.
Placental microtransfusions as measured by placental alkaline phosphatase levels in cord blood are a risk factor for mother-to-child transmission during vaginal deliveries.
==== Body
Introduction
By the end of 2004, sub-Saharan Africa was home to 13.3 million HIV-1-infected women, and in many sub-Saharan countries, the HIV-1 prevalence in antenatal women exceeded 20% [1]. In the absence of interventions such as antiretroviral drug prophylaxis, elective cesarean sections, and replacement feeding, between 25% and 35% of the children born to HIV-1-positive women themselves become infected [2]. Of the infant infections, approximately one-half occur during labor and delivery through an unknown mechanism [3].
One potential mechanism of HIV-1 mother-to-child transmission (MTCT) is through direct contact of infant mucosa with HIV-1-infected maternal blood, amniotic fluid, or cervical/vaginal secretions (the “all mucosal” mechanism) [4]. Evidence for the importance of this mechanism during intrapartum (IP) MTCT includes the observations that elective caesarean sections reduce HIV-1 MTCT [5], and that higher quantities of HIV-1 secreted into the birth canal are associated with increased HIV-1 MTCT [6,7]. On the other hand, both birth canal disinfection and emergency cesarean sections reduce birth canal exposure to HIV-1, but neither intervention significantly reduces HIV-1 MTCT [5,8–11].
In addition to the “all mucosal” hypothesis, HIV-1 MTCT could also occur via a breakdown in the maternal–fetal barrier followed by placental microtransfusions. Placental microtransfusions have previously been suggested as a route of HIV-1 transmission [12–15], and they are considered a plausible route of hepatitis B, C, and G vertical transmission [16–18]. Recently, an assay has been developed to quantify placental microtransfusions based on the detection of placental alkaline phosphatase (PLAP) activity in umbilical cord serum [13,15]. PLAP is a 130-kD glycoprotein synthesized from the eighth week of gestation until parturition, and its large molecular size is believed to preclude its passive diffusion across the placental barrier. Based on these characteristics and the observation that infants produce small amounts of PLAP compared to pregnant women, cord PLAP can be used as a surrogate marker of placental microtransfusions [13].
Although the cause of placental microtransfusions is unknown, they are hypothesized to arise during the first stage of labor, when uterine contractions intensify and membranes rupture [3,12,13]. In support of this hypothesis, cesarean sections performed before the onset of labor result in significantly less cord PLAP activity than both emergency cesarean deliveries and spontaneous vertex deliveries [13,15]. To build on these observations and to expand our knowledge of the biological mechanism of HIV-1 MTCT, we measured cord PLAP activity in Malawian mother–offspring pairs and evaluated placental microtransfusions as a risk factor for HIV-1 MTCT.
Methods
Participant Recruitment
This case-cohort study was derived from the Malaria and HIV in Pregnancy (MHP) prospective cohort study that was approved by the College of Medicine Research Committee at the University of Malawi and the Institutional Review Boards of the University of Michigan and the University of North Carolina at Chapel Hill. From December 2000 until March 2004, women in the Antenatal Ward at Queen Elizabeth Central Hospital in Blantyre, Malawi, were screened for eligibility to participate in a prospective cohort study designed to determine the association between malaria and HIV-1 MTCT. Women were ineligible for the study if they were in the active phase of labor, were participating in other research studies, lived outside the Blantyre district, were less than 15 y of age, were hypertensive, or had altered consciousness. Consenting women received HIV pre- and post-test counseling, and all HIV-1-infected women and their offspring received nevirapine according to the HIVNET 012 protocol [19]. The association between malaria and HIV-1 viral load in the MHP cohort has been described previously [20].
Case-Cohort Design
We constructed a case cohort study from the parent MHP prospective cohort. With this study design, a rare disease assumption is unnecessary and the odds ratio provides a direct estimate of the risk ratio (RR) [21]. Between December 2000 and February 2003, 2,557 pregnant women were recruited in the MHP study, of whom 744 (29.1%) were HIV infected. After identifying all cases of in utero (IU) and IP MTCT in the parent cohort, we randomly selected 160 women, regardless of their HIV-1 transmission status, from the 744 HIV-1-positive women to serve as a reference cohort. Of these mother–offspring pairs, three were excluded because of multiple gestations, six were excluded because they delivered at home (therefore cord serum was unavailable), and two were excluded because the mother died during delivery; the remaining 149 HIV-infected women formed the reference cohort of the case-cohort study (Figure 1). All cases of IU MTCT were selected from women enrolled in the MHP study through February 2003, and all cases of IP MTCT were selected from women enrolled in the MHP study through June 2003. IP transmission cases selected from February to June 2003 did not differ from the other IP transmission cases in terms of age, peripheral HIV-1 RNA load, PLAP activity, hemoglobin concentration, placental malaria infection, or mode of delivery (data not shown). HIV-1 MTCT cases were considered IU transmissions if the infant was positive for HIV-1 DNA within 48 h of birth, and were considered IP transmissions (i.e., HIV-1 infection occurring at or around the time of delivery) if the infant was both negative for HIV-1 DNA within 48 h of birth and positive for HIV-1 DNA 6 wk after delivery [22].
Figure 1 Case-Cohort Profile
Laboratory/Pathology Testing
Chorioamnionitis was assessed by a board-certified pathologist (DM) according to the methods described in [23]. Peripheral malaria infection was assessed on thick blood films stained with field stain. Placental malaria was diagnosed from formalin-fixed placental biopsies as described in [24]. Maternal hemoglobin concentration was determined by HemoCue hemoglobinometer (HemoCue, Ängelholm, Sweden), and CD4-positive T cells were quantified by FACScan (Becton Dickinson, San Jose, California, United States).
HIV-1 Testing
Within the MHP cohort, maternal HIV-1 status was determined concurrently with both the Determine HIV-1/2 Rapid Test (Abbott Laboratories, Abbott Park, Illinois, United States) and the SeroCard HIV-1/2 Rapid Test (Trinity Biotech, Bray, Ireland). HIV-1 RNA was quantified using Amplicor HIV-1 Monitor v1.5 (Roche Diagnostics, Branchburg, New Jersey, United States), with plasma HIV-1 RNA concentrations less than 400 copies/ml assigned a value of 400 copies/ml. Infants were considered HIV-1 infected based on the detection of HIV-1 DNA with a real-time PCR assay against the HIV-1 long terminal repeat [25]. Maternal HIV-1 proviral load was quantified according to the methods presented in [6].
PLAP Assay
Immediately after delivery the umbilical cord was clamped and cut. A section of the umbilical cord 3 cm from its point of insertion into the placenta was washed with saline, the umbilical vein was located, and cord blood was aspirated with a large-bore needle. Serum was prepared from the blood and stored at −80 °C. PLAP was isolated from 100 μl of cord serum using an isoform-specific anti-PLAP antibody (clone B431, Biomeda, Foster City, California, United States), and its activity was measured with an immunocatalytic assay according to the methods of Hirano et al. [26] with the following modification: a fluorescent alkaline phosphatase substrate (5 μM fluorescein diphosphate, Molecular Probes, Eugene, Oregon, United States) was used, and its fluorescence was measured on a PerkinElmer (Wellesley, Massachusetts, United States) fluorimeter (excitation λ = 490 nm, 10-nm slit width; emission λ = 514 nm, 2.5-nm slit width, 515-nm cutoff filter). Cord PLAP concentration was determined by interpolation from a standard curve of purified human PLAP (Sigma, St. Louis, Missouri, United States). The standard curve was linear from 12 ng/ml to 3,250 ng/ml (R
2 = 0.992), with PLAP activity less than 12 ng/ml assigned a value of 12 ng/ml.
Statistical Methods
IU and IP MTCT cases were analyzed as independent outcomes. We used generalized estimating equations with a logit link, binomial distribution, and an independent correlation structure to conduct bivariable and multivariable assessments of the relationship between maternal features and MTCT, while accounting for the lack of independence of the 23 MTCT cases included in the sub-cohort. To reflect the monotonically increasing risk of IP transmission as PLAP increased, log10 cord PLAP was coded as a continuous variable. In the multivariable models, prior to the outcome analysis, we assessed heterogeneity of the odds ratio by testing for interaction terms at α = 0.1. Subsequently, confounding was assessed by the backward elimination method [21] based on the change in the point estimate; variables that changed the estimate (RR) more than 10% were retained in the final model, with the exception of maternal HIV-1 viral load, which was included a priori. Variables assessed for interaction and potential confounding of HIV-1 MTCT prior to the outcome analysis include the following: HIV-1 RNA load, HIV-1 DNA load, CD4 T cell count < 200 cells/μl, chorioamnionitis (present or absent), episiotomy, self-reported fever in the week prior to enrollment, gestational age, placental malaria (any Plasmodium falciparum–infected erythrocytes on placental histology), and rupture of membranes more than 4 h prior to delivery. Owing to the correlation between HIV-1 RNA load, HIV-1 DNA load, and CD4 T cell count, the multivariable model of MTCT contained only one of these measures of HIV burden. Before the analysis of the association between maternal features and log10 cord PLAP, we eliminated duplicate values by removing the 23 cases included in the sub-cohort. The relationships between PLAP and dichotomous maternal factors were assessed with a two-tailed unpaired t-test (α = 0.05), and with continuous maternal factors the relationship was assessed with Pearson's correlation coefficient (α = 0.05). Statistical analysis was performed with S
TATA v8.2 (StataCorp, College Station, Texas, United States).
Results
We designed a case-cohort study to independently compare 36 IU and 43 IP MTCT cases to a sub-cohort of 149 HIV-positive mothers. The IU and IP MTCT cases were similar to the sub-cohort in terms of maternal age, infant birth weight, maternal hemoglobin concentration, gestational age, and gravidity (Table 1). The groups did not differ in the proportion of chorioamnionitis infections, nor in the proportion of peripheral or placental malaria infections. Of 205 deliveries, 149 (73%) were spontaneous vertex deliveries, 40 (20%) were emergency cesarean sections, seven (3%) were instrumental vaginal deliveries, five (2%) were elective cesarean sections, and four (2%) were breech deliveries; the mode of delivery did not differ significantly between study groups.
Table 1 Enrollment Characteristics
As expected, HIV-1 RNA concentration was associated with MTCT. The median HIV-1 RNA load of the mother in IU MTCT cases was twice the median HIV-1 RNA load of the sub-cohort (67,646 copies/ml versus 35,241 copies/ml), and in a univariable regression, the risk of IU transmission increased 65% for every log10 increase in HIV RNA load (RR, 1.65; 95% confidence interval [CI], 1.06–2.58). Similarly, the median HIV-1 RNA load of the mother in IP MTCT cases was approximately twice the median RNA load of the sub-cohort (76,544 copies/ml versus 35,241 copies/ml), and the risk of IP MTCT increased 76% per log10 increase in HIV-1 RNA (RR, 1.76; 95% CI, 1.10–2.81). No statistically significant association between MTCT and either HIV DNA level or CD4 count was observed.
To validate PLAP as a marker for placental microtransfusions, we determined whether cord PLAP was associated with mode of delivery and gestational age. Cord serum was available for 177/205 (87%) of the enrolled mother–offspring pairs. Mean cord PLAP activity was lower in elective cesarean deliveries than in spontaneous vertex deliveries (1.43 log10 ng/ml versus 1.82 log10 ng/ml, t = −2.09, p = 0.039), and it was also lower in elective cesarean deliveries than in emergency cesarean sections (1.43 log10 ng/ml versus 1.81 log10 ng/ml, t = −1.78, p = 0.083). In addition, cord PLAP activity was directly correlated with gestational age (n = 174, correlation coefficient = 0.18, p = 0.017), and therefore preterm deliveries (<37 wk) had lower cord PLAP than term deliveries (37–40 wk) (1.68 log10 ng/ml versus 1.84 log10 ng/ml, t = 1.91, p = 0.058). In contrast, log10 cord PLAP was not associated with the following maternal variables: rupture of membranes more than 4 h prior to delivery (n = 163, t = 0.74, p = 0.46), placental malaria (n = 167, t = −0.12, p = 0.90), chorioamnionitis (n = 162, t = 0.85, p = 0.40), duration of labor (n = 177, correlation coefficient = −0.0001, p = 0.99), log10 HIV-1 RNA load (n = 170, correlation coefficient = 0.012, p = 0.88), or CD4 T cell count (n = 171, correlation coefficient = 0.12, p = 0.105). Therefore, the only maternal factors associated with cord PLAP were mode of delivery and gestational age.
The association between IP HIV MTCT and cord PLAP concentration was also measured. This association varied by the mode of delivery (p = 0.058), so prior to the analysis, the data were stratified by mode of delivery. In a univariable model of 119 spontaneous vaginal deliveries, elevated cord PLAP activity increased the risk of IP MTCT (RR, 2.25 per log10 ng/ml PLAP; 95% CI, 0.95–5.32); a bivariable model that included log10 HIV RNA load yielded a similar association (RR, 2.01; 95% CI, 0.84–4.79; n = 115). After adjusting the model for self-reported fever and chorioamnionitis, cord PLAP activity was significantly associated with IP MTCT (RR, 2.82 per log10 increase in cord PLAP; 95% CI, 1.04–7.67; n = 103; Table 2). The risk of IP MTCT associated with cord PLAP increased slightly if maternal CD4 T cell count or HIV-1 DNA load was substituted for HIV-1 RNA load, although the precision of the estimates varied (Table 2). For emergency cesarean section deliveries (n = 32), although the trend of the data suggests an inverse relationship, log10 PLAP was not associated with IP MTCT (RR, 0.32; 95% CI, 0.04–2.68). Small sample sizes precluded the analysis of an association between cord PLAP and HIV MTCT during instrumental vaginal, elective cesarean section, and breech deliveries. Thus, among vaginal deliveries, placental microtransfusions appear to be a risk factor for IP HIV-1 transmission.
Table 2 Risk of IP HIV-1 MTCT during Vaginal Delivery per log10 Increase in Cord PLAP
Finally, we analyzed the risk of IU MTCT from placental microtransfusions. In a univariable analysis of 159 mother–offspring pairs, log10 cord PLAP activity was not significantly associated with IU MTCT (RR, 0.54; 95% CI, 0.26–1.13). Inclusion of HIV-1 RNA load, preterm delivery, syphilis infection, chorioamnionitis, and placental malaria infection as covariates in the model did not change the magnitude of the relationship between cord PLAP and IU MTCT, which remained nonsignificant (data not shown). Thus, these data provide no evidence of an association between placental microtransfusions and IU HIV transmission.
Discussion
In this case-cohort study of Malawian mother–offspring pairs, we tested the hypothesis that placental microtransfusions, as measured by cord PLAP activity, are an additional mechanism of HIV-1 MTCT. In support of this hypothesis, our data show that for every log10 increase in cord PLAP, the risk of IP transmission during vaginal deliveries almost tripled. Importantly, this increased risk of transmission remained after adjustment for HIV-1 RNA load, which is the most consistently reported risk factor for HIV-1 vertical transmission (reviewed in [27]).
In contrast to IP transmission during vaginal deliveries, placental microtransfusions were not significantly associated with IU HIV-1 transmission. The lack of association between IU MTCT and placental microtransfusions may result from the short half-life of PLAP in infants (~5 d) [13]; therefore, because PLAP activity is measured at the time of delivery, PLAP that passed into fetal circulation in the weeks prior to parturition might not persist until the time of delivery, and an association with IU MTCT, if it existed, would be missed.
Although a possible inverse relationship between PLAP and IP MTCT during emergency cesarean sections was observed, the small number of IP transmission cases in this stratum (n = 6) and the wide confidence intervals of this association (95% CI, 0.04–2.68) preclude a reliable conclusion.
The cord PLAP measurements in this study are consistent with previous studies in two ways. First, gestational age was directly correlated with cord PLAP activity, which is a result of the increase in maternal PLAP production over the course of a pregnancy [28]. Second, compared to both vaginal and emergency cesarean section deliveries, elective cesarean sections produced the smallest amount of microtransfusion [13,15]. Because only elective cesarean sections eliminate labor, the relationship between mode of delivery and cord PLAP activity has been attributed to a disruption of the placental barrier by labor and contractions. In support of this theory, Kaneda and colleagues reported a direct correlation between cord PLAP activity and prolonged labor (≥5 h) [13]. We did not observe a similar correlation among women who entered labor, and this could be explained in two ways: recording the duration of labor was not the primary concern of the study, so it is possible that our measurements were imprecise, or other features of labor such as the frequency and/or intensity of contractions may more strongly influence maternal–fetal transfusions.
Besides labor, placental infection or inflammation could also compromise the maternal–fetal barrier. A common source of placental pathology in sub-Saharan Africa is malaria, and although 17% of the women in this study had active placental malaria at the time of delivery, we detected no association between cord PLAP and placental malaria. A second potential source of placental inflammation is chorioamnionitis, which has been associated with HIV-1 MTCT [29]. Although 30% of the women in this study had chorioamnionitis, it also was not associated with cord PLAP. Based on these observations, there is no evidence from this study that placental malaria or chorioamnionitis increases placental microtransfusions.
The most probable confounding factor in this study is gestational age, which has been associated with both HIV transmission and maternal PLAP concentration. However, although the gestational age differed between the case-cohort groups, inclusion of gestational age in our IP transmission model did not change the risk estimate.
Two potential sources of uncertainty in our RR estimate are the use of PLAP as a marker of placental microtransfusions and the misclassification of HIV-1 transmission. Any measurement error in the PLAP assay (exposure) is likely to be nondifferential among the cases and the sub-cohort. Owing to the timing of infant blood collection (48 h and 6 wk after delivery), it is also possible that some of the HIV-1 transmission classified as IP was actually acquired late IU or during early breastfeeding [3]. This misclassification should be independent of PLAP activity, and therefore our RR estimate is most likely biased towards the null [21].
In this study investigating the role of maternal–fetal microtransfusions in HIV-1 MTCT, our data suggest that, independent of maternal HIV-1 viral load, placental microtransfusions during spontaneous vaginal deliveries increase the risk of IP HIV-1 MTCT. Future studies on the etiology of placental microtransfusions should provide greater insight into the mechanism of HIV-1 MTCT and suggest new strategies to prevent IP HIV-1 transmission in the developing world.
Patient Summary
Background
Without intervention, between 25% and 35% of the children born to HIV-positive mothers will themselves be infected. In about 50% of the cases, transmission from mother to child occurs during labor and delivery. Scientists don't yet understand how exactly this transmission happens. Even so, they have found that some treatments can prevent most of the mother-to-child transmission of HIV. The problem is that for many of the HIV-positive pregnant women in developing countries, these treatments are not available or not acceptable.
Why Was This Study Done?
The hope is that better understanding of mother-to-child transmission will lead to more effective, more available, and more acceptable treatments. One possible way children are infected is through placental microtransfusions, which are exchanges of small amounts of blood between the mother and the baby. Some degree of placental microtransfusion occurs in most pregnancies once labor starts, because the contractions cause small areas of rupture in the placenta, but the overall amount of blood exchange is likely to differ from delivery to delivery. Until recently, it was not possible to measure the extent of these microtransfusions for a particular delivery, but now scientists have developed an assay in umbilical cord blood that can do this. In this study, the researchers made use of this new assay to ask whether there is a link between the extent of placental microtransfusion and the likelihood of HIV transmission.
What Did the Researchers Do and Find?
They studied a group of mothers and children in Malawi. All of the mothers were HIV-positive, and some of them transmitted the virus to their children. This transmission occurred either during the pregnancy or around delivery, and the researchers knew the timing for each case. They also knew how the children were born: approximately three-quarters by vaginal delivery and one-fifth by emergency caesarean section. They determined the level of placental microtransfusion from the umbilical cord blood and then looked for correlations between HIV transmission and level of microtransfusion. They found no correlation for the cases where HIV was transmitted during pregnancy. For cases of transmission around delivery, however, higher levels of microtransfusion were associated with a higher risk of HIV transmission for vaginal deliveries.
What Does This Mean?
This suggests that better understanding of what causes microtransfusions might help researchers devise new strategies to prevent transmission. However, this will take some time. Because effective ways to prevent transmission already exist, the immediate goal must be to make them available to women in developing countries where transmission still occurs at high frequencies.
Where Can I Find More Information Online?
The following Web sites provide information on mother-to-child transmission of HIV.
World Health Organization pages:
http://www.who.int/reproductive-health/rtis/MTCT/
Joint United Nations Programme on HIV/AIDS pages:
http://www.unaids.org/Unaids/EN/In+focus/Topic+areas/ Mother-to-child+transmission.asp
Unversity of California at San Francisco HIV inSite pages:
http://hivinsite.ucsf.edu/InSite?page=kbr-07–02–03
Centers for Disease Control and Prevention page:
http://www.cdc.gov/nchstp/od/gap/pmtct/
Los Alamos HIV database:
http://hiv-web.lanl.gov/content/index
We thank the Malawian mothers who participated in this study. We thank Chin-Yih Ou for his technical assistance; Debbie Kamwendo, Paul Wilson, Patrick Mkundika (deceased), Benson Thindwa, and Visopo Harawa for their logistical and technical support; Ebbie Chalaluka and the study nurses for their invaluable contributions to participant recruitment, data collection, and patient care; and Marc Bulterys for his critical reading of the manuscript. This research was presented in part at the 53rd annual meeting of the American Society of Tropical Medicine and Hygiene and at the 11th Annual Conference on Retroviruses and Opportunistic Infections.
Support for this research was provided by the University of North Carolina at Chapel Hill Center for AIDS Research (P30-AI50410) and the National Institutes of Heath National Institute of Allergy and Infectious Disease (AI07151–26, AI065369–01, and AI49084). MEM is supported by a Research Leave Fellowship and SJR is supported by a Senior Fellowship, both from The Wellcome Trust. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Citation: Kwiek JJ, Mwapasa V, Milner DA Jr, Alke AP, Miller WC, et al. (2006) Maternal–fetal microtransfusions and HIV-1 mother-to-child transmission in Malawi. PLoS Med 3(1): e10.
Abbreviations
CIconfidence interval
IPintrapartum
IUin utero
MHPMalaria and HIV in Pregnancy
MTCTmother-to-child transmission
PLAPplacental alkaline phosphatase
RRrisk ratio
==== Refs
References
Joint United Nations Programme on HIV/AIDS, World Health Organization Joint United Nations Programme on HIV/AIDS, World Health Organization (2004) AIDS epidemic update: December 2004 2004 Available: http://www.unaids.org/wad2004/EPI_1204_pdf_en/EpiUpdate04_en.pdf . Accessed 12 October 2005
De Cock KM Fowler MG Mercier E de Vincenzi I Saba J Prevention of mother-to-child HIV transmission in resource-poor countries: Translating research into policy and practice JAMA 2000 283 1175 1182 10703780
Kourtis AP Bulterys M Nesheim SR Lee FK Understanding the timing of HIV transmission from mother to infant JAMA 2001 285 709 712 11176886
Van de Perre P Mother-to-child transmission of HIV-1: the ‘all mucosal' hypothesis as a predominant mechanism of transmission AIDS 1999 13 1133 1138 10397545
European Mode of Delivery Collaboration Elective caesarean-section versus vaginal delivery in prevention of vertical HIV-1 transmission: A randomised clinical trial Lancet 1999 353 1035 1039 10199349
Montano M Russell M Gilbert P Thior I Lockman S Comparative prediction of perinatal human immunodeficiency virus type 1 transmission, using multiple virus load markers J Infect Dis 2003 188 406 413 12870122
John GC Nduati RW Mbori-Ngacha DA Richardson BA Panteleeff D Correlates of mother-to-child human immunodeficiency virus type 1 (HIV-1) transmission: Association with maternal plasma HIV-1 RNA load, genital HIV-1 DNA shedding, and breast infections J Infect Dis 2001 183 206 212 11120927
Gaillard P Mwanyumba F Verhofstede C Claeys P Chohan V Vaginal lavage with chlorhexidine during labour to reduce mother-to-child HIV transmission: Clinical trial in Mombasa, Kenya AIDS 2001 15 389 396 11273219
Biggar RJ Miotti PG Taha TE Mtimavalye L Broadhead R Perinatal intervention trial in Africa: Effect of a birth canal cleansing intervention to prevent HIV transmission Lancet 1996 347 1647 1650 8642957
Kind C Rudin C Siegrist CA Wyler CA Biedermann K Prevention of vertical HIV transmission: Additive protective effect of elective Cesarean section and zidovudine prophylaxis. Swiss Neonatal HIV Study Group AIDS 1998 12 205 210 9468370
Mandelbrot L Landreau-Mascaro A Rekacewicz C Berrebi A Benifla JL Lamivudine-zidovudine combination for prevention of maternal-infant transmission of HIV-1 JAMA 2001 285 2083 2093 11311097
Schafer A Materno-fetal transmission of human immune deficiency virus Infect Dis Obstet Gynecol 1997 5 115 120 18476163
Kaneda T Shiraki K Hirano K Nagata I Detection of maternofetal transfusion by placental alkaline phosphatase levels J Pediatr 1997 130 730 735 9152281
Biggar RJ Mtimavalye L Justesen A Broadhead R Miley W Does umbilical cord blood polymerase chain reaction positivity indicate in utero (pre-labor) HIV infection? AIDS 1997 11 1375 1382 9302448
Lin HH Kao JH Hsu HY Mizokami M Hirano K Least microtransfusion from mother to fetus in elective cesarean delivery Obstet Gynecol 1996 87 244 248 8559532
Lin HH Lee TY Chen DS Sung JL Ohto H Transplacental leakage of HBeAg-positive maternal blood as the most likely route in causing intrauterine infection with hepatitis B virus J Pediatr 1987 111 877 881 3681555
Lin HH Kao JH Chen PJ Chen DS Mechanism of vertical transmission of hepatitis G Lancet 1996 347 1116
Lin HH Kao JH Chen DS Mother-to-child HCV transmission Lancet 2001 357 142 143
Guay LA Musoke P Fleming T Bagenda D Allen M Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial Lancet 1999 354 795 802 10485720
Mwapasa V Rogerson SJ Molyneux ME Abrams ET Kamwendo DD The effect of Plasmodium falciparum malaria on peripheral and placental HIV-1 RNA concentrations in pregnant Malawian women AIDS 2004 18 1051 1059 15096809
Rothman KJ Greenland S Modern epidemiology, 2nd ed 1998 Philadelphia Lippincott-Raven 737
Bryson YJ Luzuriaga K Sullivan JL Wara DW Proposed definitions for in utero versus intrapartum transmission of HIV-1 N Engl J Med 1992 327 1246 1247 1406816
Abrams ET Milner DA Kwiek J Mwapasa V Kamwendo DD Risk factors and mechanisms of preterm delivery in Malawi Am J Reprod Immunol 2004 52 174 183 15274659
Rogerson SJ Pollina E Getachew A Tadesse E Lema VM Placental monocyte infiltrates in response to Plasmodium falciparum malaria infection and their association with adverse pregnancy outcomes Am J Trop Med Hyg 2003 68 115 119
Luo W Yang H Rathbun K Pau CP Ou CY Detection of HIV-1 DNA in dried blood spots using a duplex real-time PCR Assay J Clin Microbiol 2005 43 1851 1857 15815008
Hirano K Matsumoto H Tanaka T Hayashi Y Iino S Specific assays for human alkaline phosphatase isozymes Clin Chim Acta 1987 166 265 273 3621603
Bulterys M Nolan ML Jamieson DJ Dominguez K Fowler MG Advances in the prevention of mother-to-child HIV-1 transmission: Current issues, future challenges. AIDScience 2 2002 Available: http://www.aidscience.org/Articles/aidscience017.asp . Accessed 17 October 2005
Contractor SF Holmes-Ievers E Morgan A Oakey M Staines NA A monoclonal antibody based solid-phase enzyme-binding assay to measure levels of placental alkaline phosphatase in serum of women during pregnancy J Immunol Methods 1985 79 99 108 3889164
Goldenberg RL Vermund SH Goepfert AR Andrews WW Choriodecidual inflammation: A potentially preventable cause of perinatal HIV-1 transmission? Lancet 1998 352 1927 1930 9863804
|
16287342
|
PMC1285069
|
CC BY
|
2021-01-05 10:40:32
|
no
|
PLoS Med. 2006 Jan 22; 3(1):e10
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0030010
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1628734210.1371/journal.pmed.0030010Research ArticleInfectious DiseasesMicrobiologyVirologyEpidemiology/Public HealthHIV/AIDSObstetrics/GynecologyPediatricsInfectious DiseasesHIV Infection/AIDSPediatricsObstetricsPregnancyMaternal–Fetal Microtransfusions and HIV-1 Mother-to-Child Transmission in Malawi Placental Microtransfusions and HIV MTCTKwiek Jesse J
1
*Mwapasa Victor
2
Milner Danny A Jr
3
Alker Alisa P
1
Miller William C
1
Tadesse Eyob
4
Molyneux Malcolm E
5
Rogerson Stephen J
6
Meshnick Steven R
1
1Department of Epidemiology, University of North Carolina, Chapel Hill, North Carolina, United States of America2Department of Community Health, College of Medicine, University of Malawi, Blantyre, Malawi3Department of Pathology, Brigham and Women's Hospital, Boston, Massachusetts, United States of America4Department of Obstetrics and Gynaecology, College of Medicine, University of Malawi, Blantyre, Malawi5Malawi-Liverpool-Wellcome Trust Clinical Research Programme, College of Medicine, University of Malawi, Blantyre, Malawi6Department of Medicine, University of Melbourne, Parkville, Victoria, AustraliaGuay Laura Academic EditorJohns Hopkins Medical SchoolUnited States of America*To whom correspondence should be addressed. E-mail: [email protected]
Competing Interests: The authors have declared that no competing interests exist.
Author Contributions: JJK participated in hypothesis generation, data collection, and data analysis, and was the primary author of this study. VM participated in the study design, data collection, and the writing of the manuscript. DAM diagnosed chorioamnionitis. APA contributed to the data analysis. WCM suggested the case-cohort design and participated in data analysis and manuscript preparation. ET contributed to data collection and was the Chief Obstetrician. MEM participated in participant recruitment and study management/supervision. SJR diagnosed placental malaria, developed the hypothesis, participated in data analysis, and provided study oversight. SRM developed the hypothesis; participated in the study design, data analysis, and study coordination; and was the Principal Investigator. All of the authors edited and revised the final manuscript.
1 2006 22 11 2005 3 1 e102 8 2005 7 10 2005 Copyright: © 2006 Kwiek 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.
Placental Microtransfusions Associated with Increased HIV Transmission from Mother to Child
Background
Between 25% and 35% of infants born to HIV-infected mothers become HIV-1 infected. One potential route of mother-to-child transmission (MTCT) could be through a breakdown in the placental barrier (i.e., maternal–fetal microtransfusions).
Methods and Findings
Placental alkaline phosphatase (PLAP) is a 130-kD maternal enzyme that cannot cross the intact placental barrier. We measured PLAP activity in umbilical vein serum as an indicator of maternal–fetal microtransfusion, and related this to the risk of HIV-1 MTCT. A case-cohort study was conducted of 149 women randomly selected from a cohort of HIV-1-infected pregnant Malawians; these women served as a reference group for 36 cases of in utero MTCT and 43 cases of intrapartum (IP) MTCT. Cord PLAP activity was measured with an immunocatalytic assay. Infant HIV status was determined by real-time PCR. The association between cord PLAP activity and HIV-1 MTCT was measured with logistic regression using generalized estimating equations. Among vaginal deliveries, PLAP was associated with IP MTCT (risk ratio, 2.25 per log10 ng/ml PLAP; 95% confidence interval, 0.95–5.32) but not in utero MTCT. In a multivariable model adjusted for HIV-1 RNA load, chorioamnionitis, and self-reported fever, the risk of IP MTCT almost tripled for every log10 increase in cord PLAP activity (risk ratio, 2.87; 95% confidence interval, 1.05–7.83).
Conclusion
These results suggest that during vaginal deliveries, placental microtransfusions are a risk factor for IP HIV-1 MTCT. Future studies are needed to identify factors that increase the risk for microtransfusions in order to prevent IP HIV-1 MTCT.
Placental microtransfusions as measured by placental alkaline phosphatase levels in cord blood are a risk factor for mother-to-child transmission during vaginal deliveries.
==== Body
Introduction
By the end of 2004, sub-Saharan Africa was home to 13.3 million HIV-1-infected women, and in many sub-Saharan countries, the HIV-1 prevalence in antenatal women exceeded 20% [1]. In the absence of interventions such as antiretroviral drug prophylaxis, elective cesarean sections, and replacement feeding, between 25% and 35% of the children born to HIV-1-positive women themselves become infected [2]. Of the infant infections, approximately one-half occur during labor and delivery through an unknown mechanism [3].
One potential mechanism of HIV-1 mother-to-child transmission (MTCT) is through direct contact of infant mucosa with HIV-1-infected maternal blood, amniotic fluid, or cervical/vaginal secretions (the “all mucosal” mechanism) [4]. Evidence for the importance of this mechanism during intrapartum (IP) MTCT includes the observations that elective caesarean sections reduce HIV-1 MTCT [5], and that higher quantities of HIV-1 secreted into the birth canal are associated with increased HIV-1 MTCT [6,7]. On the other hand, both birth canal disinfection and emergency cesarean sections reduce birth canal exposure to HIV-1, but neither intervention significantly reduces HIV-1 MTCT [5,8–11].
In addition to the “all mucosal” hypothesis, HIV-1 MTCT could also occur via a breakdown in the maternal–fetal barrier followed by placental microtransfusions. Placental microtransfusions have previously been suggested as a route of HIV-1 transmission [12–15], and they are considered a plausible route of hepatitis B, C, and G vertical transmission [16–18]. Recently, an assay has been developed to quantify placental microtransfusions based on the detection of placental alkaline phosphatase (PLAP) activity in umbilical cord serum [13,15]. PLAP is a 130-kD glycoprotein synthesized from the eighth week of gestation until parturition, and its large molecular size is believed to preclude its passive diffusion across the placental barrier. Based on these characteristics and the observation that infants produce small amounts of PLAP compared to pregnant women, cord PLAP can be used as a surrogate marker of placental microtransfusions [13].
Although the cause of placental microtransfusions is unknown, they are hypothesized to arise during the first stage of labor, when uterine contractions intensify and membranes rupture [3,12,13]. In support of this hypothesis, cesarean sections performed before the onset of labor result in significantly less cord PLAP activity than both emergency cesarean deliveries and spontaneous vertex deliveries [13,15]. To build on these observations and to expand our knowledge of the biological mechanism of HIV-1 MTCT, we measured cord PLAP activity in Malawian mother–offspring pairs and evaluated placental microtransfusions as a risk factor for HIV-1 MTCT.
Methods
Participant Recruitment
This case-cohort study was derived from the Malaria and HIV in Pregnancy (MHP) prospective cohort study that was approved by the College of Medicine Research Committee at the University of Malawi and the Institutional Review Boards of the University of Michigan and the University of North Carolina at Chapel Hill. From December 2000 until March 2004, women in the Antenatal Ward at Queen Elizabeth Central Hospital in Blantyre, Malawi, were screened for eligibility to participate in a prospective cohort study designed to determine the association between malaria and HIV-1 MTCT. Women were ineligible for the study if they were in the active phase of labor, were participating in other research studies, lived outside the Blantyre district, were less than 15 y of age, were hypertensive, or had altered consciousness. Consenting women received HIV pre- and post-test counseling, and all HIV-1-infected women and their offspring received nevirapine according to the HIVNET 012 protocol [19]. The association between malaria and HIV-1 viral load in the MHP cohort has been described previously [20].
Case-Cohort Design
We constructed a case cohort study from the parent MHP prospective cohort. With this study design, a rare disease assumption is unnecessary and the odds ratio provides a direct estimate of the risk ratio (RR) [21]. Between December 2000 and February 2003, 2,557 pregnant women were recruited in the MHP study, of whom 744 (29.1%) were HIV infected. After identifying all cases of in utero (IU) and IP MTCT in the parent cohort, we randomly selected 160 women, regardless of their HIV-1 transmission status, from the 744 HIV-1-positive women to serve as a reference cohort. Of these mother–offspring pairs, three were excluded because of multiple gestations, six were excluded because they delivered at home (therefore cord serum was unavailable), and two were excluded because the mother died during delivery; the remaining 149 HIV-infected women formed the reference cohort of the case-cohort study (Figure 1). All cases of IU MTCT were selected from women enrolled in the MHP study through February 2003, and all cases of IP MTCT were selected from women enrolled in the MHP study through June 2003. IP transmission cases selected from February to June 2003 did not differ from the other IP transmission cases in terms of age, peripheral HIV-1 RNA load, PLAP activity, hemoglobin concentration, placental malaria infection, or mode of delivery (data not shown). HIV-1 MTCT cases were considered IU transmissions if the infant was positive for HIV-1 DNA within 48 h of birth, and were considered IP transmissions (i.e., HIV-1 infection occurring at or around the time of delivery) if the infant was both negative for HIV-1 DNA within 48 h of birth and positive for HIV-1 DNA 6 wk after delivery [22].
Figure 1 Case-Cohort Profile
Laboratory/Pathology Testing
Chorioamnionitis was assessed by a board-certified pathologist (DM) according to the methods described in [23]. Peripheral malaria infection was assessed on thick blood films stained with field stain. Placental malaria was diagnosed from formalin-fixed placental biopsies as described in [24]. Maternal hemoglobin concentration was determined by HemoCue hemoglobinometer (HemoCue, Ängelholm, Sweden), and CD4-positive T cells were quantified by FACScan (Becton Dickinson, San Jose, California, United States).
HIV-1 Testing
Within the MHP cohort, maternal HIV-1 status was determined concurrently with both the Determine HIV-1/2 Rapid Test (Abbott Laboratories, Abbott Park, Illinois, United States) and the SeroCard HIV-1/2 Rapid Test (Trinity Biotech, Bray, Ireland). HIV-1 RNA was quantified using Amplicor HIV-1 Monitor v1.5 (Roche Diagnostics, Branchburg, New Jersey, United States), with plasma HIV-1 RNA concentrations less than 400 copies/ml assigned a value of 400 copies/ml. Infants were considered HIV-1 infected based on the detection of HIV-1 DNA with a real-time PCR assay against the HIV-1 long terminal repeat [25]. Maternal HIV-1 proviral load was quantified according to the methods presented in [6].
PLAP Assay
Immediately after delivery the umbilical cord was clamped and cut. A section of the umbilical cord 3 cm from its point of insertion into the placenta was washed with saline, the umbilical vein was located, and cord blood was aspirated with a large-bore needle. Serum was prepared from the blood and stored at −80 °C. PLAP was isolated from 100 μl of cord serum using an isoform-specific anti-PLAP antibody (clone B431, Biomeda, Foster City, California, United States), and its activity was measured with an immunocatalytic assay according to the methods of Hirano et al. [26] with the following modification: a fluorescent alkaline phosphatase substrate (5 μM fluorescein diphosphate, Molecular Probes, Eugene, Oregon, United States) was used, and its fluorescence was measured on a PerkinElmer (Wellesley, Massachusetts, United States) fluorimeter (excitation λ = 490 nm, 10-nm slit width; emission λ = 514 nm, 2.5-nm slit width, 515-nm cutoff filter). Cord PLAP concentration was determined by interpolation from a standard curve of purified human PLAP (Sigma, St. Louis, Missouri, United States). The standard curve was linear from 12 ng/ml to 3,250 ng/ml (R
2 = 0.992), with PLAP activity less than 12 ng/ml assigned a value of 12 ng/ml.
Statistical Methods
IU and IP MTCT cases were analyzed as independent outcomes. We used generalized estimating equations with a logit link, binomial distribution, and an independent correlation structure to conduct bivariable and multivariable assessments of the relationship between maternal features and MTCT, while accounting for the lack of independence of the 23 MTCT cases included in the sub-cohort. To reflect the monotonically increasing risk of IP transmission as PLAP increased, log10 cord PLAP was coded as a continuous variable. In the multivariable models, prior to the outcome analysis, we assessed heterogeneity of the odds ratio by testing for interaction terms at α = 0.1. Subsequently, confounding was assessed by the backward elimination method [21] based on the change in the point estimate; variables that changed the estimate (RR) more than 10% were retained in the final model, with the exception of maternal HIV-1 viral load, which was included a priori. Variables assessed for interaction and potential confounding of HIV-1 MTCT prior to the outcome analysis include the following: HIV-1 RNA load, HIV-1 DNA load, CD4 T cell count < 200 cells/μl, chorioamnionitis (present or absent), episiotomy, self-reported fever in the week prior to enrollment, gestational age, placental malaria (any Plasmodium falciparum–infected erythrocytes on placental histology), and rupture of membranes more than 4 h prior to delivery. Owing to the correlation between HIV-1 RNA load, HIV-1 DNA load, and CD4 T cell count, the multivariable model of MTCT contained only one of these measures of HIV burden. Before the analysis of the association between maternal features and log10 cord PLAP, we eliminated duplicate values by removing the 23 cases included in the sub-cohort. The relationships between PLAP and dichotomous maternal factors were assessed with a two-tailed unpaired t-test (α = 0.05), and with continuous maternal factors the relationship was assessed with Pearson's correlation coefficient (α = 0.05). Statistical analysis was performed with S
TATA v8.2 (StataCorp, College Station, Texas, United States).
Results
We designed a case-cohort study to independently compare 36 IU and 43 IP MTCT cases to a sub-cohort of 149 HIV-positive mothers. The IU and IP MTCT cases were similar to the sub-cohort in terms of maternal age, infant birth weight, maternal hemoglobin concentration, gestational age, and gravidity (Table 1). The groups did not differ in the proportion of chorioamnionitis infections, nor in the proportion of peripheral or placental malaria infections. Of 205 deliveries, 149 (73%) were spontaneous vertex deliveries, 40 (20%) were emergency cesarean sections, seven (3%) were instrumental vaginal deliveries, five (2%) were elective cesarean sections, and four (2%) were breech deliveries; the mode of delivery did not differ significantly between study groups.
Table 1 Enrollment Characteristics
As expected, HIV-1 RNA concentration was associated with MTCT. The median HIV-1 RNA load of the mother in IU MTCT cases was twice the median HIV-1 RNA load of the sub-cohort (67,646 copies/ml versus 35,241 copies/ml), and in a univariable regression, the risk of IU transmission increased 65% for every log10 increase in HIV RNA load (RR, 1.65; 95% confidence interval [CI], 1.06–2.58). Similarly, the median HIV-1 RNA load of the mother in IP MTCT cases was approximately twice the median RNA load of the sub-cohort (76,544 copies/ml versus 35,241 copies/ml), and the risk of IP MTCT increased 76% per log10 increase in HIV-1 RNA (RR, 1.76; 95% CI, 1.10–2.81). No statistically significant association between MTCT and either HIV DNA level or CD4 count was observed.
To validate PLAP as a marker for placental microtransfusions, we determined whether cord PLAP was associated with mode of delivery and gestational age. Cord serum was available for 177/205 (87%) of the enrolled mother–offspring pairs. Mean cord PLAP activity was lower in elective cesarean deliveries than in spontaneous vertex deliveries (1.43 log10 ng/ml versus 1.82 log10 ng/ml, t = −2.09, p = 0.039), and it was also lower in elective cesarean deliveries than in emergency cesarean sections (1.43 log10 ng/ml versus 1.81 log10 ng/ml, t = −1.78, p = 0.083). In addition, cord PLAP activity was directly correlated with gestational age (n = 174, correlation coefficient = 0.18, p = 0.017), and therefore preterm deliveries (<37 wk) had lower cord PLAP than term deliveries (37–40 wk) (1.68 log10 ng/ml versus 1.84 log10 ng/ml, t = 1.91, p = 0.058). In contrast, log10 cord PLAP was not associated with the following maternal variables: rupture of membranes more than 4 h prior to delivery (n = 163, t = 0.74, p = 0.46), placental malaria (n = 167, t = −0.12, p = 0.90), chorioamnionitis (n = 162, t = 0.85, p = 0.40), duration of labor (n = 177, correlation coefficient = −0.0001, p = 0.99), log10 HIV-1 RNA load (n = 170, correlation coefficient = 0.012, p = 0.88), or CD4 T cell count (n = 171, correlation coefficient = 0.12, p = 0.105). Therefore, the only maternal factors associated with cord PLAP were mode of delivery and gestational age.
The association between IP HIV MTCT and cord PLAP concentration was also measured. This association varied by the mode of delivery (p = 0.058), so prior to the analysis, the data were stratified by mode of delivery. In a univariable model of 119 spontaneous vaginal deliveries, elevated cord PLAP activity increased the risk of IP MTCT (RR, 2.25 per log10 ng/ml PLAP; 95% CI, 0.95–5.32); a bivariable model that included log10 HIV RNA load yielded a similar association (RR, 2.01; 95% CI, 0.84–4.79; n = 115). After adjusting the model for self-reported fever and chorioamnionitis, cord PLAP activity was significantly associated with IP MTCT (RR, 2.82 per log10 increase in cord PLAP; 95% CI, 1.04–7.67; n = 103; Table 2). The risk of IP MTCT associated with cord PLAP increased slightly if maternal CD4 T cell count or HIV-1 DNA load was substituted for HIV-1 RNA load, although the precision of the estimates varied (Table 2). For emergency cesarean section deliveries (n = 32), although the trend of the data suggests an inverse relationship, log10 PLAP was not associated with IP MTCT (RR, 0.32; 95% CI, 0.04–2.68). Small sample sizes precluded the analysis of an association between cord PLAP and HIV MTCT during instrumental vaginal, elective cesarean section, and breech deliveries. Thus, among vaginal deliveries, placental microtransfusions appear to be a risk factor for IP HIV-1 transmission.
Table 2 Risk of IP HIV-1 MTCT during Vaginal Delivery per log10 Increase in Cord PLAP
Finally, we analyzed the risk of IU MTCT from placental microtransfusions. In a univariable analysis of 159 mother–offspring pairs, log10 cord PLAP activity was not significantly associated with IU MTCT (RR, 0.54; 95% CI, 0.26–1.13). Inclusion of HIV-1 RNA load, preterm delivery, syphilis infection, chorioamnionitis, and placental malaria infection as covariates in the model did not change the magnitude of the relationship between cord PLAP and IU MTCT, which remained nonsignificant (data not shown). Thus, these data provide no evidence of an association between placental microtransfusions and IU HIV transmission.
Discussion
In this case-cohort study of Malawian mother–offspring pairs, we tested the hypothesis that placental microtransfusions, as measured by cord PLAP activity, are an additional mechanism of HIV-1 MTCT. In support of this hypothesis, our data show that for every log10 increase in cord PLAP, the risk of IP transmission during vaginal deliveries almost tripled. Importantly, this increased risk of transmission remained after adjustment for HIV-1 RNA load, which is the most consistently reported risk factor for HIV-1 vertical transmission (reviewed in [27]).
In contrast to IP transmission during vaginal deliveries, placental microtransfusions were not significantly associated with IU HIV-1 transmission. The lack of association between IU MTCT and placental microtransfusions may result from the short half-life of PLAP in infants (~5 d) [13]; therefore, because PLAP activity is measured at the time of delivery, PLAP that passed into fetal circulation in the weeks prior to parturition might not persist until the time of delivery, and an association with IU MTCT, if it existed, would be missed.
Although a possible inverse relationship between PLAP and IP MTCT during emergency cesarean sections was observed, the small number of IP transmission cases in this stratum (n = 6) and the wide confidence intervals of this association (95% CI, 0.04–2.68) preclude a reliable conclusion.
The cord PLAP measurements in this study are consistent with previous studies in two ways. First, gestational age was directly correlated with cord PLAP activity, which is a result of the increase in maternal PLAP production over the course of a pregnancy [28]. Second, compared to both vaginal and emergency cesarean section deliveries, elective cesarean sections produced the smallest amount of microtransfusion [13,15]. Because only elective cesarean sections eliminate labor, the relationship between mode of delivery and cord PLAP activity has been attributed to a disruption of the placental barrier by labor and contractions. In support of this theory, Kaneda and colleagues reported a direct correlation between cord PLAP activity and prolonged labor (≥5 h) [13]. We did not observe a similar correlation among women who entered labor, and this could be explained in two ways: recording the duration of labor was not the primary concern of the study, so it is possible that our measurements were imprecise, or other features of labor such as the frequency and/or intensity of contractions may more strongly influence maternal–fetal transfusions.
Besides labor, placental infection or inflammation could also compromise the maternal–fetal barrier. A common source of placental pathology in sub-Saharan Africa is malaria, and although 17% of the women in this study had active placental malaria at the time of delivery, we detected no association between cord PLAP and placental malaria. A second potential source of placental inflammation is chorioamnionitis, which has been associated with HIV-1 MTCT [29]. Although 30% of the women in this study had chorioamnionitis, it also was not associated with cord PLAP. Based on these observations, there is no evidence from this study that placental malaria or chorioamnionitis increases placental microtransfusions.
The most probable confounding factor in this study is gestational age, which has been associated with both HIV transmission and maternal PLAP concentration. However, although the gestational age differed between the case-cohort groups, inclusion of gestational age in our IP transmission model did not change the risk estimate.
Two potential sources of uncertainty in our RR estimate are the use of PLAP as a marker of placental microtransfusions and the misclassification of HIV-1 transmission. Any measurement error in the PLAP assay (exposure) is likely to be nondifferential among the cases and the sub-cohort. Owing to the timing of infant blood collection (48 h and 6 wk after delivery), it is also possible that some of the HIV-1 transmission classified as IP was actually acquired late IU or during early breastfeeding [3]. This misclassification should be independent of PLAP activity, and therefore our RR estimate is most likely biased towards the null [21].
In this study investigating the role of maternal–fetal microtransfusions in HIV-1 MTCT, our data suggest that, independent of maternal HIV-1 viral load, placental microtransfusions during spontaneous vaginal deliveries increase the risk of IP HIV-1 MTCT. Future studies on the etiology of placental microtransfusions should provide greater insight into the mechanism of HIV-1 MTCT and suggest new strategies to prevent IP HIV-1 transmission in the developing world.
Patient Summary
Background
Without intervention, between 25% and 35% of the children born to HIV-positive mothers will themselves be infected. In about 50% of the cases, transmission from mother to child occurs during labor and delivery. Scientists don't yet understand how exactly this transmission happens. Even so, they have found that some treatments can prevent most of the mother-to-child transmission of HIV. The problem is that for many of the HIV-positive pregnant women in developing countries, these treatments are not available or not acceptable.
Why Was This Study Done?
The hope is that better understanding of mother-to-child transmission will lead to more effective, more available, and more acceptable treatments. One possible way children are infected is through placental microtransfusions, which are exchanges of small amounts of blood between the mother and the baby. Some degree of placental microtransfusion occurs in most pregnancies once labor starts, because the contractions cause small areas of rupture in the placenta, but the overall amount of blood exchange is likely to differ from delivery to delivery. Until recently, it was not possible to measure the extent of these microtransfusions for a particular delivery, but now scientists have developed an assay in umbilical cord blood that can do this. In this study, the researchers made use of this new assay to ask whether there is a link between the extent of placental microtransfusion and the likelihood of HIV transmission.
What Did the Researchers Do and Find?
They studied a group of mothers and children in Malawi. All of the mothers were HIV-positive, and some of them transmitted the virus to their children. This transmission occurred either during the pregnancy or around delivery, and the researchers knew the timing for each case. They also knew how the children were born: approximately three-quarters by vaginal delivery and one-fifth by emergency caesarean section. They determined the level of placental microtransfusion from the umbilical cord blood and then looked for correlations between HIV transmission and level of microtransfusion. They found no correlation for the cases where HIV was transmitted during pregnancy. For cases of transmission around delivery, however, higher levels of microtransfusion were associated with a higher risk of HIV transmission for vaginal deliveries.
What Does This Mean?
This suggests that better understanding of what causes microtransfusions might help researchers devise new strategies to prevent transmission. However, this will take some time. Because effective ways to prevent transmission already exist, the immediate goal must be to make them available to women in developing countries where transmission still occurs at high frequencies.
Where Can I Find More Information Online?
The following Web sites provide information on mother-to-child transmission of HIV.
World Health Organization pages:
http://www.who.int/reproductive-health/rtis/MTCT/
Joint United Nations Programme on HIV/AIDS pages:
http://www.unaids.org/Unaids/EN/In+focus/Topic+areas/ Mother-to-child+transmission.asp
Unversity of California at San Francisco HIV inSite pages:
http://hivinsite.ucsf.edu/InSite?page=kbr-07–02–03
Centers for Disease Control and Prevention page:
http://www.cdc.gov/nchstp/od/gap/pmtct/
Los Alamos HIV database:
http://hiv-web.lanl.gov/content/index
We thank the Malawian mothers who participated in this study. We thank Chin-Yih Ou for his technical assistance; Debbie Kamwendo, Paul Wilson, Patrick Mkundika (deceased), Benson Thindwa, and Visopo Harawa for their logistical and technical support; Ebbie Chalaluka and the study nurses for their invaluable contributions to participant recruitment, data collection, and patient care; and Marc Bulterys for his critical reading of the manuscript. This research was presented in part at the 53rd annual meeting of the American Society of Tropical Medicine and Hygiene and at the 11th Annual Conference on Retroviruses and Opportunistic Infections.
Support for this research was provided by the University of North Carolina at Chapel Hill Center for AIDS Research (P30-AI50410) and the National Institutes of Heath National Institute of Allergy and Infectious Disease (AI07151–26, AI065369–01, and AI49084). MEM is supported by a Research Leave Fellowship and SJR is supported by a Senior Fellowship, both from The Wellcome Trust. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Citation: Kwiek JJ, Mwapasa V, Milner DA Jr, Alke AP, Miller WC, et al. (2006) Maternal–fetal microtransfusions and HIV-1 mother-to-child transmission in Malawi. PLoS Med 3(1): e10.
Abbreviations
CIconfidence interval
IPintrapartum
IUin utero
MHPMalaria and HIV in Pregnancy
MTCTmother-to-child transmission
PLAPplacental alkaline phosphatase
RRrisk ratio
==== Refs
References
Joint United Nations Programme on HIV/AIDS, World Health Organization Joint United Nations Programme on HIV/AIDS, World Health Organization (2004) AIDS epidemic update: December 2004 2004 Available: http://www.unaids.org/wad2004/EPI_1204_pdf_en/EpiUpdate04_en.pdf . Accessed 12 October 2005
De Cock KM Fowler MG Mercier E de Vincenzi I Saba J Prevention of mother-to-child HIV transmission in resource-poor countries: Translating research into policy and practice JAMA 2000 283 1175 1182 10703780
Kourtis AP Bulterys M Nesheim SR Lee FK Understanding the timing of HIV transmission from mother to infant JAMA 2001 285 709 712 11176886
Van de Perre P Mother-to-child transmission of HIV-1: the ‘all mucosal' hypothesis as a predominant mechanism of transmission AIDS 1999 13 1133 1138 10397545
European Mode of Delivery Collaboration Elective caesarean-section versus vaginal delivery in prevention of vertical HIV-1 transmission: A randomised clinical trial Lancet 1999 353 1035 1039 10199349
Montano M Russell M Gilbert P Thior I Lockman S Comparative prediction of perinatal human immunodeficiency virus type 1 transmission, using multiple virus load markers J Infect Dis 2003 188 406 413 12870122
John GC Nduati RW Mbori-Ngacha DA Richardson BA Panteleeff D Correlates of mother-to-child human immunodeficiency virus type 1 (HIV-1) transmission: Association with maternal plasma HIV-1 RNA load, genital HIV-1 DNA shedding, and breast infections J Infect Dis 2001 183 206 212 11120927
Gaillard P Mwanyumba F Verhofstede C Claeys P Chohan V Vaginal lavage with chlorhexidine during labour to reduce mother-to-child HIV transmission: Clinical trial in Mombasa, Kenya AIDS 2001 15 389 396 11273219
Biggar RJ Miotti PG Taha TE Mtimavalye L Broadhead R Perinatal intervention trial in Africa: Effect of a birth canal cleansing intervention to prevent HIV transmission Lancet 1996 347 1647 1650 8642957
Kind C Rudin C Siegrist CA Wyler CA Biedermann K Prevention of vertical HIV transmission: Additive protective effect of elective Cesarean section and zidovudine prophylaxis. Swiss Neonatal HIV Study Group AIDS 1998 12 205 210 9468370
Mandelbrot L Landreau-Mascaro A Rekacewicz C Berrebi A Benifla JL Lamivudine-zidovudine combination for prevention of maternal-infant transmission of HIV-1 JAMA 2001 285 2083 2093 11311097
Schafer A Materno-fetal transmission of human immune deficiency virus Infect Dis Obstet Gynecol 1997 5 115 120 18476163
Kaneda T Shiraki K Hirano K Nagata I Detection of maternofetal transfusion by placental alkaline phosphatase levels J Pediatr 1997 130 730 735 9152281
Biggar RJ Mtimavalye L Justesen A Broadhead R Miley W Does umbilical cord blood polymerase chain reaction positivity indicate in utero (pre-labor) HIV infection? AIDS 1997 11 1375 1382 9302448
Lin HH Kao JH Hsu HY Mizokami M Hirano K Least microtransfusion from mother to fetus in elective cesarean delivery Obstet Gynecol 1996 87 244 248 8559532
Lin HH Lee TY Chen DS Sung JL Ohto H Transplacental leakage of HBeAg-positive maternal blood as the most likely route in causing intrauterine infection with hepatitis B virus J Pediatr 1987 111 877 881 3681555
Lin HH Kao JH Chen PJ Chen DS Mechanism of vertical transmission of hepatitis G Lancet 1996 347 1116
Lin HH Kao JH Chen DS Mother-to-child HCV transmission Lancet 2001 357 142 143
Guay LA Musoke P Fleming T Bagenda D Allen M Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial Lancet 1999 354 795 802 10485720
Mwapasa V Rogerson SJ Molyneux ME Abrams ET Kamwendo DD The effect of Plasmodium falciparum malaria on peripheral and placental HIV-1 RNA concentrations in pregnant Malawian women AIDS 2004 18 1051 1059 15096809
Rothman KJ Greenland S Modern epidemiology, 2nd ed 1998 Philadelphia Lippincott-Raven 737
Bryson YJ Luzuriaga K Sullivan JL Wara DW Proposed definitions for in utero versus intrapartum transmission of HIV-1 N Engl J Med 1992 327 1246 1247 1406816
Abrams ET Milner DA Kwiek J Mwapasa V Kamwendo DD Risk factors and mechanisms of preterm delivery in Malawi Am J Reprod Immunol 2004 52 174 183 15274659
Rogerson SJ Pollina E Getachew A Tadesse E Lema VM Placental monocyte infiltrates in response to Plasmodium falciparum malaria infection and their association with adverse pregnancy outcomes Am J Trop Med Hyg 2003 68 115 119
Luo W Yang H Rathbun K Pau CP Ou CY Detection of HIV-1 DNA in dried blood spots using a duplex real-time PCR Assay J Clin Microbiol 2005 43 1851 1857 15815008
Hirano K Matsumoto H Tanaka T Hayashi Y Iino S Specific assays for human alkaline phosphatase isozymes Clin Chim Acta 1987 166 265 273 3621603
Bulterys M Nolan ML Jamieson DJ Dominguez K Fowler MG Advances in the prevention of mother-to-child HIV-1 transmission: Current issues, future challenges. AIDScience 2 2002 Available: http://www.aidscience.org/Articles/aidscience017.asp . Accessed 17 October 2005
Contractor SF Holmes-Ievers E Morgan A Oakey M Staines NA A monoclonal antibody based solid-phase enzyme-binding assay to measure levels of placental alkaline phosphatase in serum of women during pregnancy J Immunol Methods 1985 79 99 108 3889164
Goldenberg RL Vermund SH Goepfert AR Andrews WW Choriodecidual inflammation: A potentially preventable cause of perinatal HIV-1 transmission? Lancet 1998 352 1927 1930 9863804
|
0
|
PMC1285070
|
CC BY
|
2021-01-05 11:13:40
|
no
|
PLoS Med. 2006 Jan 22; 3(1):e23
|
latin-1
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0030023
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0030026SynopsisInfectious DiseasesVirologyHIV/AIDSObstetrics/GynecologyInfectious DiseasesHIV Infection/AIDSObstetricsPregnancyPlacental Microtransfusions Associated with Increased HIV Transmission from Mother to Child Synopsis1 2006 22 11 2005 3 1 e26Copyright: © 2006 PLoS Medicine.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.
Maternal-Fetal Microtransfusions and HIV-1 Mother-to-Child Transmission in Malawi
==== Body
Mother-to-child transmission (MTCT) is the predominant way that children become infected with HIV. MTCT can occur prenatally during pregnancy, perinatally during labor and delivery, and postnatally through breastfeeding. In the absence of specific interventions, approximately one-third of children born to mothers who are HIV-positive become infected themselves. Specific interventions, including prenatal HIV counseling and testing, antiretroviral prophylaxis, elective cesarean delivery, and avoidance of breastfeeding, have reduced MTCT to less than 2% in developed countries. In many less-developed countries, however, these interventions are not readily available, and even effective, shorter, and cheaper antiretroviral prophylaxis strategies have not yet been widely implemented or accepted. As a result, MTCT rates in lower-income countries remain high, accounting for an estimated 2,000 new pediatric HIV infections per day.
An estimated one-half of the infant infections occur during labor and delivery, but the exact mechanisms remain poorly understood. Two possible ways of transmission have been proposed: direct contact of infant mucosa with HIV-infected maternal body fluids, and placental microtransfusions after breakdown of the maternal–fetal barrier through contractions at the beginning of labor. However, the evidence implicating either transmission route has been inconclusive.
To further examine a possible connection between placental microtransfusions and HIV MTCT, Jesse Kwiek and colleagues have taken advantage of a recently developed assay that provides a surrogate measure for placental microtransfusions, based on the amount of placental alkaline phosphatase (PLAP) in umbilical cord blood. PLAP is a large maternal enzyme that cannot cross the intact placental barrier. Infants produce very low levels of PLAP, and amounts found in umbilical cord blood are thought to result from leakage of maternal protein into the fetal circulation, caused by placental microtransfusions.
The researchers measured PLAP activity in umbilical cord blood as an indicator of maternal–fetal microtransfusions, and related this to risk of MTCT in a case-cohort study of mothers who were HIV-positive in Malawi. In the study, 149 women randomly selected from a larger cohort of pregnant women infected with HIV served as a reference group for 36 cases of prenatal MTCT and 43 cases of perinatal MTCT. The researchers saw no correlation between PLAP levels and prenatal MTCT. However, among the cases of perinatal transmission in women who had vaginal deliveries, elevated PLAP levels were associated with higher transmission risks. The connection was also seen after adjusting for some potential confounding factors such as HIV viral RNA load and chorioamnionitis.
While these are preliminary results and many questions remain, the reviewers felt that the design of the study enabled an efficient first test of the maternal–fetal microtransfusion hypothesis of MCTC. As such, it should encourage other researchers to look at this issue in their datasets. If a connection between microtransfusions and transmission is confirmed, it might help to improve the timing of short-term prophylaxis regimens and possibly lead to the development of new strategies for preventing MTCT of HIV.
|
0
|
PMC1285071
|
CC BY
|
2021-01-05 10:40:32
|
no
|
PLoS Med. 2006 Jan 22; 3(1):e26
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0030026
|
oa_comm
|
==== Front
BMC Cell BiolBMC Cell Biology1471-2121BioMed Central London 1471-2121-6-371625091710.1186/1471-2121-6-37Research ArticleHDACs and the senescent phenotype of WI-38 cells Place Robert F [email protected] Emily J [email protected] Charles [email protected] Department of Molecular and Cell Biology, University of Connecticut, Storrs, Connecticut 06269, USA2005 26 10 2005 6 37 37 8 6 2005 26 10 2005 Copyright © 2005 Place et al; licensee BioMed Central Ltd.2005Place 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
Normal cells possess a limited proliferative life span after which they enter a state of irreversible growth arrest. This process, known as replicative senescence, is accompanied by changes in gene expression that give rise to a variety of senescence-associated phenotypes. It has been suggested that these gene expression changes result in part from alterations in the histone acetylation machinery. Here we examine the influence of HDAC inhibitors on the expression of senescent markers in pre- and post-senescent WI-38 cells.
Results
Pre- and post-senescent WI-38 cells were treated with the HDAC inhibitors butyrate or trichostatin A (TSA). Following HDAC inhibitor treatment, pre-senescent cells increased p21WAF1 and β-galactosidase expression, assumed a flattened senescence-associated morphology, and maintained a lower level of proteasome activity. These alterations also occurred during normal replicative senescence of WI-38 cells, but were not accentuated further by HDAC inhibitors. We also found that HDAC1 levels decline during normal replicative senescence.
Conclusion
Our findings indicate that HDACs impact numerous phenotypic changes associated with cellular senescence. Reduced HDAC1 expression levels in senescent cells may be an important event in mediating the transition to a senescent phenotype.
==== Body
Background
Normal somatic cells possess a limited proliferative life span after which they enter a state of irreversible growth arrest. This process, known as replicative senescence, can be signaled by shortened telomeres that result from repeated rounds of DNA replication in the absence of telomerase expression. Once the telomeres erode to an average size of 4–6 kilobases, senescence is triggered and cells stop dividing [1,2]. Replicative senescence plays an important role in maintaining the structural integrity of tissues by limiting the excessive clonal expansion of cells [3,4]. However, the accumulation of senescent cells is also believed to contribute to the age-related decline in tissue function [5]. Replicative senescence can therefore be viewed as both a mechanism of tumor suppression and a contributor in pathologies associated with age. The role of replicative senescence in tumorigenesis is highlighted by the fact that the most common mutations in human cancers occur in genes encoding p53 and members of the pRB pathway, which are the critical effectors of replicative senescence [4,6,7].
A number of fundamental metabolic and biochemical changes occur as a cell enters senescence and begins to age. Numerous studies have reported dramatic changes in protein turnover. The proteasome, the primary non-lysosomal protease responsible for degrading intracellular proteins including misfolded, oxidized and ubiquitinated proteins, has been reported to decline in function with age [8-13]. Several reports have indicated that the expression of certain proteasome subunits drops after cells enter replicative senescence [14-17]. In addition, proteasome inhibition, or "clogging", has been observed as aging cells accumulate damaged proteins [12,13,18]. The resulting drop in protein turnover may contribute to the accumulation of protein deposits, such as lipofuscin, which can further compromise cell function [19]. In addition, the drop in proteasome activity is likely to alter the activity of numerous cellular signal transduction pathways that involve the proteasome.
Replicative senescence is accompanied by many changes in gene expression that contribute to the senescence-associated phenotypes. Of particular importance are the cell cycle inhibitors p16INK4a and p21WAF1, which are induced upon replicative senescence to halt cell proliferation [20,21]. Interestingly, many genes involved in the regulation of cellular growth arrest and differentiation are regulated by histone acetylation. For example, in proliferating fibroblasts, the stable association of HDAC1 with the Sp1/Sp3 transcription factors bound to the p21WAF1 promoter suppresses p21WAF1 expression. Upon senescence, HDAC1 is displaced from to the p21WAF1 promoter, due in part to the actions of p53 [22].
HDAC inhibitors have long been known to induce differentiation, growth arrest, and apoptosis in cancer cells [23-25]. The aberrant utilization of HDACs is believed to be a contributing factor in carcinogenesis. However, only recently have HDAC inhibitors been shown to induce premature senescence in normal human fibroblasts [26,27]. HDACs may therefore play a critical role in modulating cell physiology during the aging process, as well as contribute to the cellular changes associated with transformation. Here we examine the interplay between cellular HDAC activity and a number of phenotypic changes that accompany cell senescence. We find that replicative senescence is accompanied by a drop in cellular HDAC1 expression, the activation of the cell cycle inhibitory protein p21WAF1, and a reduction in cellular proteasome activity and subunit expression. The critical role of HDACs in regulating these events is supported by the finding that HDAC inhibitors selectively trigger these changes in pre-senescent, but not post-senescent cells. Our findings indicate that a drop in HDAC expression may be a critical event in mediating the transition from a proliferating to a senescent phenotype.
Results
HDAC inhibitors induce a senescence-like phenotype in proliferating WI-38 cells
HDAC inhibitors can induce growth arrest in many cell types, and have recently been reported to induce a senescence-like state in normal human fibroblasts [26,27]. Therefore, we sought to determine if the HDAC inhibitors butyrate and TSA could induce premature senescence in proliferating WI-38 cells. One molecular marker of senescence in normal human fibroblasts is p21WAF1 expression [28]. As shown in Figure 1A, treatment with butyrate or TSA for 24 hours induced the expression of p21WAF1 in proliferating WI-38 cells. Distinct morphological changes also occurred when WI-38 cells enter replicative senescence. Senescent cells became larger and assumed irregular shapes, while proliferating WI-38 cells formed long and striated parallel arrays (Figure 1B). As shown in Figure 1B, treatment of young WI-38 cells with HDAC inhibitors butyrate or TSA caused cells to rapidly acquire a senescent-like morphology.
Figure 1 HDAC inhibition induces markers of senescence. (A) Butyrate and TSA induce the expression of the senescence-associated cell cycle inhibitor protein p21WAF1. Proliferating WI-38 cells were treated with butyrate (BA) or TSA for 24 hours. Cytosolic extracts were prepared and equivalent protein levels were analyzed by immunoblot using an anti-p21WAF1 antibody. Untreated samples (0 hours) consisted of cytosolic extracts prepared from naïve cells. Results are shown in triplicate. (B) Butyrate and TSA rapidly induce senescent-like morphologies in young WI-38 cells. Proliferating WI-38 cells were treated with butyrate or TSA for 0, 24, 48, and 72 hours, as indicated. Phase contrast images of cell morphology were taken at 100 × magnification. An image of WI-38 cells propagated to replicative senescence is also shown.
Another biomarker for replicative senescence is senescence-associated-β-galactosidase (SA-β-gal) activity [29]. Young WI-38 cells were cultured for 14 days in 0.5 mM butyrate or 9 days in 0.5 μM TSA. These concentrations allowed for the prolonged exposure of WI-38 cells to the HDAC inhibitors with minimal cytotoxicity. As shown in Figure 2, young WI-38 cells cultured in the presence of either HDAC inhibitor acquired the perinuclear staining for SA-β-gal activity normally associated with senescent cells. Untreated proliferating WI-38 cells propagated in parallel had no SA-β-gal activity (Figure 2). This data further supports the findings that HDAC inhibition induces a senescent-like phenotype in proliferating fibroblasts [26,27].
Figure 2 HDAC inhibition induces the senescence-associated-β-glacatosidase (SA-β-gal) activity in proliferating WI-38 cells. Proliferating WI-38 cells were propagated for 14 days in 0.5 mM butyrate (Proliferating WI-38; Butyrate) or 9 days in 0.5 μM TSA (Proliferating WI-38; TSA) and stained for SA-β-gal activity. The level of SA-β-gal staining was also determined in untreated young WI-38 cells propagated in parallel (Proliferating WI-38; Control), as well as WI-38 cells propagated to replicative senescence (Senescent WI-38; Control). Phase contrast images are shown in duplicate at 100 × magnification.
Proteasome activity is reduced in senescent WI-38 cells
Declines in proteasome function during senescence and aging have been observed in cultured cells and in tissues from a variety of organisms [8-10,14,15,17]. Our aim was to verify and characterize the changes in proteasome activity following senescence in the human fibroblast WI-38 cell line. Cytosolic extracts prepared from proliferating and senescent WI-38 cells were tested for proteasome activity using the synthetic substrate Suc-LLVY-AMC [30]. As shown in Figure 3A, proteasome activity was significantly lower (reduced by ~30%) in the older WI-38 cells. Since a decrease in proteasome activity may also cause a general increase in the presence of polyubiqitinated proteins, cytsolic extracts from proliferating and senescent WI-38 cells were analyzed by immunoblotting for polyubiquitinated proteins. As indicated in Figure 3B, the accumulation of high molecular weight ubiquitin-conjugated proteins was accentuated in the senescent WI-38 cells, which is consistent with a drop in proteasome activity.
Figure 3 Proteasome activity is reduced in senescent WI-38 cells. (A) Cell lysates were prepared from proliferating or senescent WI-38 cells. Protein concentrations were normalized and proteasome activity was determined using a synthetic fluorogenic substrate (± standard error; n = 4). The reduction in proteasome activity was found to be significant in the older cells (**, p < 0.01). (B) Polyubiquitinated proteins found in the cytosolic extracts of proliferating and senescent WI-38 cells. Sample concentrations were normalized and the level of polyubiquitin-protein conjugates was determined by immunoblot using an antibody specific for ubiquitin.
Previous reports have documented that certain proteasome subunits are down-regulated in senescent WI-38 cells [14,15]. However, these analyses were limited to only a select subset of proteasome subunits. We therefore analyzed the expression of each constitutive β-type subunit to further characterize differences in proteasome subunit expression between proliferating and senescent cells (Figure 4). As shown in Figure 4, senescent WI-38 cells expressed lower levels of the three catalytic proteasome subunits: β5 (X), β1 (Y), and β2 (Z). However, the expression levels of the other β-type subunits did not change in the older cells. Figure 4 also shows the increased expression of p21WAF1 in senescent WI-38 cells [20]. The protein expression levels of the β5 subunit were additionally quantified by optical densitometry from immunoblots (Figure 5). The β5 protein levels were reduced by ~30% in senescent WI-38 cells, which corresponds to the ~30% decline in proteasome activity (as shown in Figure 3A).
Figure 4 Expression levels of the β-type proteasome subunits in proliferating and senescent WI-38 cells. Cytosolic extracts were prepared from proliferating and senescent WI-38 cells. Samples were analyzed by immunoblot using antibodies specific for each of the β-type proteasome subunits. Actin served as a loading control. Levels of p21WAF1 (p21) were also determined to serve as an inducible marker for replicative senescence. Results from duplicate cultures are shown.
Figure 5 Quantified levels of the β5 proteasome subunit and actin protein in proliferating and senescent WI-38 cells. Cytosolic extracts were prepared and equivalent protein levels were analyzed for β5 subunit expression by immunoblotting. Actin levels served as a loading control. The graphical display indicates the relative intensities of β5 and actin levels determined by optical densitometry from corresponding immunoblots (± standard error; n = 3). The decline in β5 levels was found to be significant in senescent cells (**, p < 0.01).
Senescent WI-38 cells are resistant to HDAC inhibitors
The expression of p21WAF1 is regulated by aceytlation and readily activated by HDAC inhibitors [31,32]. We therefore determined the effect of HDAC inhibitors on p21WAF1 expression in proliferating and senescent WI-38 cells. As shown in Figure 6A, treatment with HDAC inhibitors butyrate or TSA induced p21WAF1 expression in proliferating WI-38 cells. However, in senescent cells the endogenous levels of p21WAF1 were high and not further enhanced by either HDAC inhibitor (Figure 6A). These data suggest that p21WAF1activation in senescent cells may result from a reduction in cellular HDAC activity.
Figure 6 Senescent WI-38 cells are resistant to the effects of HDAC inhibition. (A) Butyrate and TSA did not further enhance the expression of cell cycle inhibitor protein p21WAF1 in senescent WI-38 cells. Proliferating and senescent WI-38 cells were treated with butyrate (BA) or TSA for 24 hours, as indicated. Cytosolic extracts were prepared and equivalent protein levels were analyzed by immunoblot using an antibody specific for p21WAF1. Actin served as a loading control and was determined by immunoblotting for the actin protein. Control samples (0 hours) consisted of cytosolic extracts prepared from untreated cells. (B) The effect of HDAC inhibitors on proteasome activity in proliferating and senescent WI-38 cells. Cytoslic extracts were prepared from proliferating (Young;◆) and senescent (Old;) WI-38 cells treated with butyrate (+ Butyrate) or TSA (+ TSA) for 0, 24, 48 and 72 hours, as indicated. Equivalent protein concentrations were determined and analyzed for proteasome activity using a synthetic fluorogenic substrate (± standard error; n = 4; values of non-treated samples set to 1). Senescent cells were found to be significantly less responsive to proteasome inhibition by butyrate and TSA.
HDAC inhibitors have also been reported to suppress proteasome activity and subunit expression in several transformed cell lines [33-35]. We hypothesized that HDAC inhibitors may suppress proteasome activity in proliferating WI-38 cells, as well. Cytosolic extracts were prepared from young WI-38 cells treated with butyrate or TSA for 0, 24, 48, and 72 hours. The synthetic substrate Suc-LLVY-AMC was then utilized to measure proteasome activity in each sample. As shown in Figure 6B, proteasome activity decreased in young WI-38 cells treated with either butyrate or TSA. To determine if senescent WI-38 cells were also sensitive to HDAC inhibitor-induced proteasome suppression, the proteasome activity of senescent WI-38 cells was analyzed following butyrate or TSA treatment. Although proteasome activity was lower in senescent WI-38 cells (as shown in Figure 3A), it was significantly less sensitive to the inhibitory effects of the HDAC inhibitors (Figure 6B). This data suggests that replicative senescence and HDAC inhibitor-induced senescence impacts proteasome activity through a common pathway.
Reduced expression of the β5 proteasome subunit in proliferating WI-38 cells by HDAC inhibitors
We determined whether the HDAC inhibitors butyrate and TSA could suppress the expression of the catalytic β5 subunit of the proteasome in proliferating WI-38 cells. The immunoblots in Figure 7A indicate that β5 expression levels decreased in these cells following butyrate or TSA treatments. The expression levels of the β5 subunit were additionally quantified by optical densitometry from immunoblots (Figure 7B). This data indicates that reduced proteasome activity following HDAC inhibition (Figure 6B) may be due in part to reduced proteasome subunit expression.
Figure 7 HDAC inhibitors down-regulate β5 proteasome subunit expression in proliferating WI-38 cells. (A) Proliferating WI-38 cells were treated with butyrate (BA) or TSA for 0, 24, 48, or 72 hours, as indicated. Cytosolic extracts were prepared and equivalent protein levels (determined by a Bradford assay) were analyzed by immunoblotting using an antibody specific for the β5 proteasome subunit. Actin served as a loading control. (B) Immunoblots from Figure 7A were quantified by optical densitometry. The graphical display shows the relative levels of the β5 subunit (normalized to actin) from the corresponding immunoblots (± standard error; n = 3). The decline in β5 levels was found to be significant at the indicated time points (**, p < 0.01).
HDAC1 is down-regulated in senescent WI-38 cells
The class I histone deacetylase protein HDAC1 is a component of the corepressor complex involved in suppressing the transcription of p21WAF1 and other cell cycle inhibitory genes [22]. We therefore determined if HDAC1 expression was altered upon replicative senescence. As shown in Figure 8, HDAC1 levels decreased in senescent WI-38 cells. (It should be noted that HDAC1 is predominantly a nuclear protein, but diffuses into the cyoplasmic fraction during protein extraction.) The levels of another class I histone deacetylase, HDAC3, was found to be equivalent in pre- and post-senescent cells (Figure 8). The drop in HDAC1 expression may contribute to the induction of p21WAF1 in senescent WI-38 cells (as shown in Figure 4; panel p21). Likewise, the decline in HDAC1 may contribute to the appearance of other senescent phenotypes, such as a drop in proteasome activity.
Figure 8 HDAC1 and HDAC3 levels in proliferating and senescent WI-38 cells. Cytosolic and nuclear extracts were prepared from proliferating and senescent WI-38 cells. HDAC1 and HDAC3 levels were determined in both fractions by immunoblotting. Actin served as a loading control. The results from duplicate cultures are shown.
Discussion
Replicative senescence marks the end of the proliferative life span of normal cells. This is accompanied by distinct alterations in the pattern of gene expression. It has been suggested that changes in gene expression during senescence and aging may result in part from alterations in protein acetylation [36-38]. Figure 9 illustrates a potential mechanism by which HDACs (e.g. HDAC1) contribute to the senescence phenotype. As WI-38 cells senesce, HDAC activity decreases to facilitate changes in gene expression. Reductions in HDAC levels, in association with increased transcriptional activity of p53 in senescent cells, contributes to the induction of p21WAF1 expression and subsequent growth arrest [39,40]. This model also envisions HDACs contributing to the age-related decline in proteasome activity, since HDAC inhibitors can reduce proteasome expression and activity [33-35,41,42].
Figure 9 Model: the age-related decline in HDAC levels contributes to replicative senescence. As WI-38 cells enter senescence, cellular HDAC activity is envisioned to decrease to alter gene expression. The reduction in HDAC activity induces the expression of p21WAF1 to assist in senescence-associated growth arrest. It is also envisioned that the decline in HDAC activity contributes to the senescence-associated drop in proteasome activity.
The identification of HDACs as a component in replicative senescence, and hence growth arrest, is interesting because data has shown that HDACs can promote tumor growth and stem cell proliferation. For example, it has been reported that HDAC1 overexpression occurs in 68% of primary human gastric cancer, and contributes to colony formation and proliferation of prostate and breast cancer cells [43-45]. Some transformed cell types may exaggerate the expression of HDACs to circumvent replicative senescence. In this regard, cancer cells are similar to stem cells, where HDAC1 is required for full cellular growth potential [46]. This further supports the idea that replicative senescence, and the associated decline in HDAC1 expression, has a tumor suppressing role [4,47].
It is not entirely clear how HDACs are regulating proteasome subunit expression. In yeast, a common mode of transcriptional regulation of the proteasomal subunits has already been identified [48-50]. Nearly all the yeast subunit homologs have been found to possess proteasome-associated control elements within their promoters. The transcription factor RPN4 has been identified as the component within yeast involved in binding these elements to modulate gene transcription [48]. Remarkably, no homolog of RPN4 has been identified in humans. However, it is still possible that another common transcriptional mechanism is shared amongst the catalytic subunits in human cells. The activity of these putative regulatory proteins may be regulated by acetylation, such that an increased level of acetylation reduces proteasome subunit expression.
Our analysis of HDAC1 and HDAC3 indicates that replicative senescence is not accompanied by a global decline in HDAC expression. Rather, it appears to occur through the down-regulation of HDAC1, and potentially other HDACs. Other groups have also reported a senescence-specific form of the HDAC2 protein [38]. In addition, the NAD+-dependent Sir2 histone deacetylase has been identified to contribute to the replicative life-span in yeast, thus suggesting that the mammalian Sir2-related class III HDACs may contribute to senescence in normal human cell types, as well [51,52]. It should be noted that it is not clear if the decline in HDAC1 is a cause or a consequence of replicative senescence. However, it seems reasonable to hypothesize that age-related modulations in HDAC levels could be a contributing factor in senescence. Further analysis of individual HDAC proteins may identify their individual functions within the senescence machinery. Anti-aging and anti-cancer strategies may be aimed at increasing or decreasing the activity of specific HDAC proteins.
Conclusion
Our findings indicate that cellular HDAC activity regulates numerous phenotypic changes associated with cellular senescence. Reduced cellular HDAC expression and activity, in association with other events, may be important for mediating the transition to a senescent phenotype.
Methods
Cell culture and treatments
The WI-38 human lung fibroblast cell line was purchased from American Type Culture Collection (Manassas, VA). Cells were propagated in minimal essential media containing 2 mM L-glutamine and Earle's salts (E-MEM) supplemented with 10% fetal bovine serum, 0.1 mM non-essential amino acids, 1 mM Sodium Pyruvate, streptomycin (50 mg/ml), and penicillin (50 U/ml). All medium components were purchased from Invitrogen Life Technologies (Carlsbad, CA). WI-38 cells entered senescence at about 50 CPD (Cumulative Population Doublings). Early-passage WI-38 cells (CPD < 30) are referred to as young or pre-senescent cells and displayed high proliferative potential. Late-passage WI-38 cells (CPD > 50) are classified as old or post-senescent cells and exhibited very low proliferative potential. Sodium butyrate (Sigma-Aldrich, St. Louis, MO) was used at the final concentration of 4 mM (unless stated otherwise). TSA (Calbiochem, San Diego, CA) was used at a 2 μM concentration (unless stated otherwise). Cells treated with TSA were given fresh media supplemented with new TSA every 24 hours.
Immunoblotting
Cytosolic extracts were prepared as described in Inan et al. [53]. For immunoblotting studies, 25 μg of cytoplasmic protein (quantified by the Bio-Rad protein assay) was denatured under reducing conditions, separated on 10% sodium dodecyl sulfate (SDS) polyacrylamide gels, and transferred to nitrocellulose by voltage gradient transfer. The resulting blots were blocked with 5% nonfat dry milk. Specific proteins were detected with appropriate antibodies using enhanced chemiluminescence detection (Santa Cruz Biotechnology, Santa Cruz, CA). Immunoblotting antibodies used were: subunit β1 PW8140, subunit β2 PW8145, subunit β3 PW8130, subunit β4 PW8890, subunit β5 PW8895, subunit β6 PW9000, and subunit β7 PW8135, (Affiniti Research Products Ltd., Mamhead, Exeter, UK); Ubiquitin P1A6, (Santa Cruz Biotechnology, Santa Cruz, CA); p21 C-19, (Santa Cruz Biotechnology, Santa Cruz, CA); and Actin I-19, (Santa Cruz Biotechnology, Santa Cruz, CA). The antibodies specific for ubiquitin and p21WAF1 were diluted 1:500 for immunoblotting. All other antibodies were employed at a 1:1000 dilution. For optical densitometry, immunoblot images were scanned on a UMAX Astra 1220P scanner and analyzed with NIH Image version 1.62. Statistical significance was determined by a paired Student's t-test.
Proteasome activity assay
Proteasome activity was quantified by using a fluorogenic proteasome-specific substrate. The assay is based on the detection of the fluorophore AMC (7-amino-4-methylcoumarin) after cleavage from the synthetic proteasome substrate Suc-LLVY-AMC (Calbiochem, San Diego, CA). Cytosolic extract (5 μg of total protein in 5 μl) was incubated in a 100 μl reaction containing 20 mM Tris-HCL (pH 7.8), 0.5 mM EDTA, 0.035% SDS, and 70 μM Suc-LLVY-AMC for 10 minutes at room temperature. The change in fluorescence (substrate consumption) was measured over an interval of 40 minutes using a microtiter plate fluorometer (excitation, 360 nm; emission, 460 nm). Proteasome-independent activity was determined by performing the assay in the presence of proteasome inhibitor MG-132 (final concentration 60 μM) (Calbiochem, San Diego, CA). Proteasome activity values were derived by subtracting the fluorescence obtained in the presence of this inhibitor from the values obtained in its absence. The values shown represent the ratio in proteasome activity from each sample compared to the activity in young WI-38 cell extracts. Assays were performed in quadruplicate, and statistical significance was determined by a paired Student's t-test.
Senescence-associated β-galactosidase staining
Staining for β-galactosidase activity in WI-38 cells was performed as previously described [29]. WI-38 cells were washed with PBS, fixed in 0.2% glutaraldehyde/2% formaldehyde for 10 minutes at room temperature, and washed again with PBS. Cells were then stained at 37°C (in the absence of CO2) with fresh senescence-associated β-gal (SA-β-gal) staining solution (150 mM NaCl, 2 mM MgCl2, 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, and 40 mM citric acid/sodium phosphate, pH 6.0) containing 1 mg/ml 5-bromo-4-chloro-3-indolyl-β-D-galactoside (X-gal). Once staining was maximal (12–16 hrs), cells were washed with PBS and overlaid in 70% glycerol. Images were taken at 100 × magnification as viewed by phase contrast.
Abbreviations
HDAC, histone deacetylase; BA, butyrate; TSA, trichostatin A; SA-β-gal, senescence-associated-β-galactosidase; E-MEM, minimal essential media with Earle's salts; Suc-LLVY-AMC, N-succinyl-Leu-Leu-Val-Tyr-7-amino-4-methylcoumarin.
Authors' contributions
RFP performed all the experiments, designed the study, and wrote the manuscript. EJN helped capture all microscopic images and discussed these results. CG was the principal investigator who gave advice in designing the study and edited the manuscript.
Acknowledgements
This work was supported in part by an award from the National Cancer Institute to C.G. (R29CA79656)
==== Refs
Harley CB Futcher AB Greider CW Telomeres shorten during ageing of human fibroblasts Nature 1990 345 458 460 2342578 10.1038/345458a0
Allsopp RC Chang E Kashefi-Aazam M Rogaev EI Piatyszek MA Shay JW Harley CB Telomere shortening is associated with cell division in vitro and in vivo Exp Cell Res 1995 220 194 200 7664836 10.1006/excr.1995.1306
Kipling D Telomeres, replicative senescence and human ageing Maturitas 2001 38 25 37; discussion 37-8 11311583 10.1016/S0378-5122(00)00189-4
Campisi J Cellular senescence as a tumor-suppressor mechanism Trends Cell Biol 2001 11 S27 31 11684439
Faragher RG Kipling D How might replicative senescence contribute to human ageing? Bioessays 1998 20 985 991 10048298 10.1002/(SICI)1521-1878(199812)20:12<985::AID-BIES4>3.0.CO;2-A
Chen QM Bartholomew JC Campisi J Acosta M Reagan JD Ames BN Molecular analysis of H2O2-induced senescent-like growth arrest in normal human fibroblasts: p53 and Rb control G1 arrest but not cell replication Biochem J 1998 332 ( Pt 1) 43 50 9576849
Shay JW Pereira-Smith OM Wright WE A role for both RB and p53 in the regulation of human cellular senescence Exp Cell Res 1991 196 33 39 1652450 10.1016/0014-4827(91)90453-2
Friguet B Bulteau AL Chondrogianni N Conconi M Petropoulos I Protein degradation by the proteasome and its implications in aging Ann N Y Acad Sci 2000 908 143 154 10911955
Bulteau AL Szweda LI Friguet B Age-dependent declines in proteasome activity in the heart Arch Biochem Biophys 2002 397 298 304 11795886 10.1006/abbi.2001.2663
Conconi M Szweda LI Levine RL Stadtman ER Friguet B Age-related decline of rat liver multicatalytic proteinase activity and protection from oxidative inactivation by heat-shock protein 90 Arch Biochem Biophys 1996 331 232 240 8660703 10.1006/abbi.1996.0303
Hayashi T Goto S Age-related changes in the 20S and 26S proteasome activities in the liver of male F344 rats Mech Ageing Dev 1998 102 55 66 9663792 10.1016/S0047-6374(98)00011-6
Sitte N Merker K Von Zglinicki T Davies KJ Grune T Protein oxidation and degradation during cellular senescence of human BJ fibroblasts: part II--aging of nondividing cells Faseb J 2000 14 2503 2510 11099468 10.1096/fj.00-0210com
Sitte N Merker K Von Zglinicki T Grune T Davies KJ Protein oxidation and degradation during cellular senescence of human BJ fibroblasts: part I--effects of proliferative senescence Faseb J 2000 14 2495 2502 11099467 10.1096/fj.00-0209com
Chondrogianni N Stratford FL Trougakos IP Friguet B Rivett AJ Gonos ES Central role of the proteasome in senescence and survival of human fibroblasts: induction of a senescence-like phenotype upon its inhibition and resistance to stress upon its activation J Biol Chem 2003 278 28026 28037 12736271 10.1074/jbc.M301048200
Chondrogianni N Gonos ES Proteasome inhibition induces a senescence-like phenotype in primary human fibroblasts cultures Biogerontology 2004 5 55 61 15138382 10.1023/B:BGEN.0000017687.55667.42
Petropoulos I Conconi M Wang X Hoenel B Bregegere F Milner Y Friguet B Increase of oxidatively modified protein is associated with a decrease of proteasome activity and content in aging epidermal cells J Gerontol A Biol Sci Med Sci 2000 55 B220 7 10819308
Keller JN Huang FF Markesbery WR Decreased levels of proteasome activity and proteasome expression in aging spinal cord Neuroscience 2000 98 149 156 10858621 10.1016/S0306-4522(00)00067-1
Davies KJ Degradation of oxidized proteins by the 20S proteasome Biochimie 2001 83 301 310 11295490 10.1016/S0300-9084(01)01250-0
Sitte N Huber M Grune T Ladhoff A Doecke WD Von Zglinicki T Davies KJ Proteasome inhibition by lipofuscin/ceroid during postmitotic aging of fibroblasts Faseb J 2000 14 1490 1498 10928983 10.1096/fj.14.11.1490
Fang L Igarashi M Leung J Sugrue MM Lee SW Aaronson SA p21Waf1/Cip1/Sdi1 induces permanent growth arrest with markers of replicative senescence in human tumor cells lacking functional p53 Oncogene 1999 18 2789 2797 10362249 10.1038/sj.onc.1202615
Dai CY Enders GH p16 INK4a can initiate an autonomous senescence program Oncogene 2000 19 1613 1622 10763818 10.1038/sj.onc.1203438
Lagger G Doetzlhofer A Schuettengruber B Haidweger E Simboeck E Tischler J Chiocca S Suske G Rotheneder H Wintersberger E Seiser C The tumor suppressor p53 and histone deacetylase 1 are antagonistic regulators of the cyclin-dependent kinase inhibitor p21/WAF1/CIP1 gene Mol Cell Biol 2003 23 2669 2679 12665570 10.1128/MCB.23.8.2669-2679.2003
Saito A Yamashita T Mariko Y Nosaka Y Tsuchiya K Ando T Suzuki T Tsuruo T Nakanishi O A synthetic inhibitor of histone deacetylase, MS-27-275, with marked in vivo antitumor activity against human tumors Proc Natl Acad Sci U S A 1999 96 4592 4597 10200307 10.1073/pnas.96.8.4592
Glick RD Swendeman SL Coffey DC Rifkind RA Marks PA Richon VM La Quaglia MP Hybrid polar histone deacetylase inhibitor induces apoptosis and CD95/CD95 ligand expression in human neuroblastoma Cancer Res 1999 59 4392 4399 10485488
Butler LM Agus DB Scher HI Higgins B Rose A Cordon-Cardo C Thaler HT Rifkind RA Marks PA Richon VM Suberoylanilide hydroxamic acid, an inhibitor of histone deacetylase, suppresses the growth of prostate cancer cells in vitro and in vivo Cancer Res 2000 60 5165 5170 11016644
Munro J Barr NI Ireland H Morrison V Parkinson EK Histone deacetylase inhibitors induce a senescence-like state in human cells by a p16-dependent mechanism that is independent of a mitotic clock Exp Cell Res 2004 295 525 538 15093749 10.1016/j.yexcr.2004.01.017
Ogryzko VV Hirai TH Russanova VR Barbie DA Howard BH Human fibroblast commitment to a senescence-like state in response to histone deacetylase inhibitors is cell cycle dependent Mol Cell Biol 1996 16 5210 5218 8756678
Noda A Ning Y Venable SF Pereira-Smith OM Smith JR Cloning of senescent cell-derived inhibitors of DNA synthesis using an expression screen Exp Cell Res 1994 211 90 98 8125163 10.1006/excr.1994.1063
Dimri GP Lee X Basile G Acosta M Scott G Roskelley C Medrano EE Linskens M Rubelj I Pereira-Smith O A biomarker that identifies senescent human cells in culture and in aging skin in vivo Proc Natl Acad Sci U S A 1995 92 9363 9367 7568133
Rock KL Gramm C Rothstein L Clark K Stein R Dick L Hwang D Goldberg AL Inhibitors of the proteasome block the degradation of most cell proteins and the generation of peptides presented on MHC class I molecules Cell 1994 78 761 771 8087844 10.1016/S0092-8674(94)90462-6
Richon VM Sandhoff TW Rifkind RA Marks PA Histone deacetylase inhibitor selectively induces p21WAF1 expression and gene-associated histone acetylation Proc Natl Acad Sci U S A 2000 97 10014 10019 10954755 10.1073/pnas.180316197
Chai F Evdokiou A Young GP Zalewski PD Involvement of p21(Waf1/Cip1) and its cleavage by DEVD-caspase during apoptosis of colorectal cancer cells induced by butyrate Carcinogenesis 2000 21 7 14 10607727 10.1093/carcin/21.1.7
Yin L Laevsky G Giardina C Butyrate suppression of colonocyte NF-kappa B activation and cellular proteasome activity J Biol Chem 2001 276 44641 44646 11572859 10.1074/jbc.M105170200
Catley L Weisberg E Tai YT Atadja P Remiszewski S Hideshima T Mitsiades N Shringarpure R LeBlanc R Chauhan D Munshi NC Schlossman R Richardson P Griffin J Anderson KC NVP-LAQ824 is a potent novel histone deacetylase inhibitor with significant activity against multiple myeloma Blood 2003 102 2615 2622 12816865 10.1182/blood-2003-01-0233
Place RF Noonan EJ Giardina C HDAC inhibition prevents NF-kappa B activation by suppressing proteasome activity: down-regulation of proteasome subunit expression stabilizes I kappa B alpha Biochem Pharmacol 2005 70 394 406 15950952 10.1016/j.bcp.2005.04.030
Burzynski SR Gene silencing--a new theory of aging Med Hypotheses 2003 60 578 583 12615527 10.1016/S0306-9877(03)00050-1
Chang KT Min KT Regulation of lifespan by histone deacetylase Ageing Res Rev 2002 1 313 326 12067588 10.1016/S1568-1637(02)00003-X
Wagner M Brosch G Zwerschke W Seto E Loidl P Jansen-Durr P Histone deacetylases in replicative senescence: evidence for a senescence-specific form of HDAC-2 FEBS Lett 2001 499 101 106 11418121 10.1016/S0014-5793(01)02524-8
Atadja P Wong H Garkavtsev I Veillette C Riabowol K Increased activity of p53 in senescing fibroblasts Proc Natl Acad Sci U S A 1995 92 8348 8352 7667293
Kulju KS Lehman JM Increased p53 protein associated with aging in human diploid fibroblasts Exp Cell Res 1995 217 336 345 7698234 10.1006/excr.1995.1095
Mitsiades CS Mitsiades NS McMullan CJ Poulaki V Shringarpure R Hideshima T Akiyama M Chauhan D Munshi N Gu X Bailey C Joseph M Libermann TA Richon VM Marks PA Anderson KC Transcriptional signature of histone deacetylase inhibition in multiple myeloma: biological and clinical implications Proc Natl Acad Sci U S A 2004 101 540 545 14695887 10.1073/pnas.2536759100
Mariadason JM Corner GA Augenlicht LH Genetic reprogramming in pathways of colonic cell maturation induced by short chain fatty acids: comparison with trichostatin A, sulindac, and curcumin and implications for chemoprevention of colon cancer Cancer Res 2000 60 4561 4572 10969808
Kawai H Li H Avraham S Jiang S Avraham HK Overexpression of histone deacetylase HDAC1 modulates breast cancer progression by negative regulation of estrogen receptor alpha Int J Cancer 2003 107 353 358 14506733 10.1002/ijc.11403
Choi JH Kwon HJ Yoon BI Kim JH Han SU Joo HJ Kim DY Expression profile of histone deacetylase 1 in gastric cancer tissues Jpn J Cancer Res 2001 92 1300 1304 11749695
Halkidou K Gaughan L Cook S Leung HY Neal DE Robson CN Upregulation and nuclear recruitment of HDAC1 in hormone refractory prostate cancer Prostate 2004 59 177 189 15042618 10.1002/pros.20022
Lagger G O'Carroll D Rembold M Khier H Tischler J Weitzer G Schuettengruber B Hauser C Brunmeir R Jenuwein T Seiser C Essential function of histone deacetylase 1 in proliferation control and CDK inhibitor repression Embo J 2002 21 2672 2681 12032080 10.1093/emboj/21.11.2672
Pereira-Smith OM Smith JR Evidence for the recessive nature of cellular immortality Science 1983 221 964 966 6879195
Mannhaupt G Schnall R Karpov V Vetter I Feldmann H Rpn4p acts as a transcription factor by binding to PACE, a nonamer box found upstream of 26S proteasomal and other genes in yeast FEBS Lett 1999 450 27 34 10350051 10.1016/S0014-5793(99)00467-6
Xie Y Varshavsky A RPN4 is a ligand, substrate, and transcriptional regulator of the 26S proteasome: a negative feedback circuit Proc Natl Acad Sci U S A 2001 98 3056 3061 11248031 10.1073/pnas.071022298
Owsianik G Balzi l L Ghislain M Control of 26S proteasome expression by transcription factors regulating multidrug resistance in Saccharomyces cerevisiae Mol Microbiol 2002 43 1295 1308 11918814 10.1046/j.1365-2958.2002.02823.x
Kennedy BK Gotta M Sinclair DA Mills K McNabb DS Murthy M Pak SM Laroche T Gasser SM Guarente L Redistribution of silencing proteins from telomeres to the nucleolus is associated with extension of life span in S. cerevisiae Cell 1997 89 381 391 9150138 10.1016/S0092-8674(00)80219-6
Kaeberlein M McVey M Guarente L The SIR2/3/4 complex and SIR2 alone promote longevity in Saccharomyces cerevisiae by two different mechanisms Genes Dev 1999 13 2570 2580 10521401 10.1101/gad.13.19.2570
Inan MS Rasoulpour RJ Yin L Hubbard AK Rosenberg DW Giardina C The luminal short-chain fatty acid butyrate modulates NF-kappaB activity in a human colonic epithelial cell line Gastroenterology 2000 118 724 734 10734024 10.1016/S0016-5085(00)70142-9
|
16250917
|
PMC1285358
|
CC BY
|
2021-01-04 16:31:29
|
no
|
BMC Cell Biol. 2005 Oct 26; 6:37
|
utf-8
|
BMC Cell Biol
| 2,005 |
10.1186/1471-2121-6-37
|
oa_comm
|
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-561623231310.1186/1471-2148-5-56Research ArticleLineage-specific variations of congruent evolution among DNA sequences from three genomes, and relaxed selective constraints on rbcL in Cryptomonas (Cryptophyceae) Hoef-Emden Kerstin [email protected] Hoang-Dung [email protected] Michael [email protected] Universität zu Köln, Botanisches Institut, Lehrstuhl I; Gyrhofstr. 15, 50931 Köln, Germany2005 18 10 2005 5 56 56 24 5 2005 18 10 2005 Copyright © 2005 Hoef-Emden et al; licensee BioMed Central Ltd.2005Hoef-Emden 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
Plastid-bearing cryptophytes like Cryptomonas contain four genomes in a cell, the nucleus, the nucleomorph, the plastid genome and the mitochondrial genome. Comparative phylogenetic analyses encompassing DNA sequences from three different genomes were performed on nineteen photosynthetic and four colorless Cryptomonas strains. Twenty-three rbcL genes and fourteen nuclear SSU rDNA sequences were newly sequenced to examine the impact of photosynthesis loss on codon usage in the rbcL genes, and to compare the rbcL gene phylogeny in terms of tree topology and evolutionary rates with phylogenies inferred from nuclear ribosomal DNA (concatenated SSU rDNA, ITS2 and partial LSU rDNA), and nucleomorph SSU rDNA.
Results
Largely congruent branching patterns and accelerated evolutionary rates were found in nucleomorph SSU rDNA and rbcL genes in a clade that consisted of photosynthetic and colorless species suggesting a coevolution of the two genomes. The extremely accelerated rates in the rbcL phylogeny correlated with a shift from selection to mutation drift in codon usage of two-fold degenerate NNY codons comprising the amino acids asparagine, aspartate, histidine, phenylalanine, and tyrosine. Cysteine was the sole exception. The shift in codon usage seemed to follow a gradient from early diverging photosynthetic to late diverging photosynthetic or heterotrophic taxa along the branches. In the early branching taxa, codon preferences were changed in one to two amino acids, whereas in the late diverging taxa, including the colorless strains, between four and five amino acids showed changes in codon usage.
Conclusion
Nucleomorph and plastid gene phylogenies indicate that loss of photosynthesis in the colorless Cryptomonas strains examined in this study possibly was the result of accelerated evolutionary rates that started already in photosynthetic ancestors. Shifts in codon usage are usually considered to be caused by changes in functional constraints and in gene expression levels. Thus, the increasing influence of mutation drift on codon usage along the clade may indicate gradually relaxed constraints and reduced expression levels on the rbcL gene, finally correlating with a loss of photosynthesis in the colorless Cryptomonas paramaecium strains.
==== Body
Background
Ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) plays a key role in the photosynthetic Calvin cycle as the carbon dioxide fixating enzyme [1,2]. The most common type of RuBisCO, form I RuBisCO, is found in Viridiplantae, cyanobacteria (green-like RuBisCO), in most non-green algae, and some proteobacteria (red-like RuBisCO) [3]. Eight large subunits and an equal number of small subunits make up a functional holoenzyme of form I RuBisCO [1]. Their genes, rbcL and rbcS, are plastid-encoded and co-transcribed in non-green algae, whereas in the Viridiplantae, rbcS was transferred to the nucleus and evolved to a multi-gene family [3,4]. To compensate for its inefficient and slow catalytic mechanism, RuBisCO is usually expressed at high rates in plastids, making it "the most abundant protein in the world" [5]. Surprisingly, in some colorless algae and holoparasitic land plants, functional RuBisCO was found [6,7]. One possible explanation for a function of RuBisCO outside of the Calvin cycle was reported only recently. In developing Brassica napus seeds, RuBisCO recycles CO2 that was released in the pyruvate dehydrogenase step prior to fatty-acid biosynthesis [8].
Cryptophyte algae are flagellates with complex plastids originating from a secondary endosymbiosis between a phagotrophic host cell and a red alga [9]. The cryptophyte plastid consists of two nested compartments, each with its own genome (nucleomorph in the periplastidial space between inner and outer pairs of plastid membranes and plastid genome) [9,10]. Cryptophytes of the genus Cryptomonas thrive exclusively in freshwater habitats [11]. The leukoplast-bearing freshwater cryptophytes were formerly considered a separate genus Chilomonas, but have been shown in phylogenetic analyses to be colorless Cryptomonas cells; the diagnosis of the genus Cryptomonas was emended accordingly [11]. In previous molecular phylogenetic analyses, accelerated evolutionary rates in either nuclear ribosomal DNA (internal transcribed spacer 2 [ITS2] and partial LSU ribosomal DNA [LSU rDNA]) or nucleomorph SSU ribosomal DNA sequences were found in three independently evolved colorless Cryptomonas lineages and in closely related photosynthetic strains [12].
Accelerated evolutionary rates are usually considered indicative for relaxed selective constraints and were also found in rbcL genes of land plants [13,14]. Previous studies have shown that relaxed selective constraints and different levels of gene expression in protein-coding genes correlate with biases in codon usage [15,16]. Possibly, by preferring the so-called "major codons" in highly expressed genes, the efficiency of the translation process is increased [17]. The differences in codon usage between highly and lowly expressed genes are most obvious for two-fold degenerate NNY codons, i.e. pairs of triplets that code for the same amino acid with either C or T(U) at third positions. In highly expressed genes, NNC is preferred over NNT in most two-fold degenerate NNY codons, whereas the codon preference reverses, if functional constraints are relaxed and expression levels decrease [16,18]. Under neutral mutation, DNA displays a strong bias towards increased A+T contents. Such a genome compositional bias has been found also in endosymbiotic or organellar genomes [9,19,20]. Thus, in two-fold degenerate NNY codons, codon bias due to selection operates in opposite direction to genome compositional bias [16].
In this study, we compare the phylogeny of the rbcL gene as a representative for the plastid genome to phylogenies of nuclear rDNA (concatenated SSU rDNA, ITS2 and partial LSU rDNA) and nucleomorph SSU rDNA. To obtain a congruent taxon sampling across the three genomes of all strains, twenty-three rbcL genes and fourteen nuclear SSU rDNA genes were newly sequenced. As putative indicators for differences of functional constraints and expression levels in the rbcL genes, codon usages of two-fold degenerate NNY codons were compared among photosynthetic and colorless Cryptomonas species.
Results
Results of the phylogenetic analyses
Both ribosomal data sets passed the chi-square test for homogeneity of base frequencies across taxa, whereas the rbcL data set failed the test (Additional file 1). After exclusion of third codon positions, the rbcL data set passed the test, indicating that the heterogeneity of base frequencies was restricted to third codon positions. This became also obvious by separate computation of mean values and standard deviations of base frequencies for first, second and third codon positions across all twenty-three taxa. The second codon position was most homogeneous concerning standard deviations (A: 27.8 ± 0.3%; C: 21.4 ± 0.4%; G: 19.9 ± 0.4%; T: 30.9 ± 0.3%), the first codon position had an intermediate position (A: 24.6 ± 1.0%; C: 17.8 ± 1.4, G: 38.9 ± 0.7, T: 18.7 ± 1.1), whereas the third codon position was most heterogeneous (A: 29.0 ± 2.6%; C: 16.1 ± 4.6%; G: 10.3 ± 3.8%; T: 44.6 ± 3.2%). Phylogenetic analyses with bootstrap resampling under all optimality criteria and a Bayesian analysis, however, showed that despite of an obvious bias, these positions contributed the most to the support of the clades in the rbcL data set. This was confirmed by separate phylogenetic analyses of first, second and third codon positions (Additional file 1). The highly conserved rbcL protein sequences consistently failed to recover three clades that were otherwise highly supported in all DNA sequence data sets (Cryptomonas marssonii, C. ovata and C. pyrenoidifera), because phylogenetic information was predominantly based on synonymous substitutions. Therefore, we used the rbcL gene tree with complete codon positions for a comparison with the nuclear and nucleomorph ribosomal DNA phylogenies (Figure 1A to 1C).
Figure 1 Unrooted maximum likelihood trees of DNA sequences representing three different genomes of the cryptophyte genus Cryptomonas. Figure 1A – Tree inferred from concatenated nuclear SSU rDNA, ITS2 and partial LSU rDNA sequences. Evolutionary model, GTR+I+Γ [51]; -ln L = 9254.5. Figure 1B – Nucleomorph SSU rDNA phylogeny. Evolutionary model, TVM+I+Γ [51]; -ln L = 4899.1. Figure 1C – Tree inferred from plastid-encoded rbcL genes (for a rooted tree including rbcL genes of other cryptophyte genera, see Additional file 3). Evolutionary model, GTR+I+Γ [51]; -ln L = 7857.4. Figure 1D (inlet) – Nuclear (top), nucleomorph (middle) and plastid (bottom) phylogeny scaled to the same substitution rate. Gray shaded areas in Figures 1A to C, presumed position of the root. In a rooted phylogeny inferred from a concatenated data set of nuclear (ITS2 excluded), nucleomorph and plastid DNA sequences with Guillardia theta as an outgroup, the root inserted between clade NoPyr and all other taxa (see Additional file 4). Evolutionary models were chosen according to the results of the Akaike information criterion in Modeltest (see Additional file 1 and Methods). Support values from left to right, maximum likelihood bootstrap/maximum parsimony bootstrap/distance (neighbor-joining) bootstrap/posterior probabilities (Figures 1A and B) or maximum likelihood bootstrap/maximum parsimony bootstrap/distance (neighbor-joining) bootstrap/logdet transformation bootstrap/posterior probabilities (Figure 1C). Cbo, Cryptomonas borealis; Ccu, C. curvata; Cgy, C. gyropyrenoidosa; Clu, C. lundii; Cma, C. marssonii; Cov, C. ovata; Cpa, C. paramaecium (colorless); Cpy, C. pyrenoidifera; Cte, C. tetrapyrenoidosa; blue, taxa of clade LB; red branches and strain designations, loss of photosynthesis; scale bars, substitutions per site.
Almost all Cryptomonas clades were unequivocally recovered with significant or at least moderate support in nuclear, nucleomorph and plastid gene trees (Figure 1A to 1C). This refers to C. curvata (Ccu), C. marssonii (Cma), C. ovata (Cov), C. paramaecium (colorless strains; Cpa), C. pyrenoidifera (Cpy), and C. tetrapyrenoidosa (Cte; clades named according to Hoef-Emden and Melkonian 2003). C. borealis (Cbo) was significantly supported in nuclear and nucleomorph phylogenies but not in all phylogenetic analyses in the rbcL phylogeny (Figure 1C). Significant support for this clade, however, was found in the rbcL protein phylogeny apparently due to nonsynonymous substitutions (tree not shown; for support values, see Additional file 1). Clade NoPyr (for no pyrenoids [11]), otherwise highly supported in nuclear and nucleomorph phylogenies, could not be resolved in the rbcL phylogeny (Figure 1C).
In all phylogenies, C. borealis, C. gyropyrenoidosa and C. lundii formed a "super-clade" together with the colorless C. paramaecium (termed clade LB for long-branch in [11]; Figure 1A to 1C). Only in the nucleomorph SSU rDNA tree, however, convincing support for clade LB was found (Figure 1B). In the nucleomorph SSU rDNA and rbcL phylogenies, representing the two genomes of the complex plastid, evolutionary rates and topologies of the strains in clade LB resembled each other. In both phylogenies, evolutionary rates were extremely accelerated in clade LB, and the branching pattern was similar, except for the position of C. gyropyrenoidosa, which was a sister to C. lundii in the nucleomorph SSU rDNA (but without bootstrap support), but not in the rbcL phylogeny (where it was the first divergence). In the nuclear-encoded ribosomal DNA phylogeny, predominantly the strains of clade NoPyr displayed increased evolutionary rates, whereas evolutionary rates were less pronounced in C. borealis and C. paramaecium of clade LB (Figure 1A). In clade NoPyr, an acceleration of evolutionary rates was also present to a lesser extent in the nucleomorph SSU rDNA; in the rbcL phylogeny, however, branch lengths of this clade were inconspicuous (Figure 1B and 1C).
In Figure 1D, the phylogenetic trees of Figure 1A to 1C were scaled to the same substitution rate and in Figure 2, the maximum likelihood distances of C. pyenoidifera strain M1077 to the other taxa were plotted in a chart diagram for a direct comparison of genetic divergences. Among the three data sets, the rbcL data displayed generally the highest substitution rates and genetic distances. In the nucleomorph SSU rDNA, the evolutionary rates of C. gyropyrenoidosa, C. borealis and C. paramaecium were in an intermediate position. Apparently, evolutionary rates in clade LB increased successively from host to nucleomorph to plastid genome.
Figure 2 Chart diagram displaying genetic divergences among the taxa and across the three data sets. A strain from a clade with inconspicuous branch lengths in all three phylogenies, Cryptomonas pyrenoidifera strain M1077, was chosen as a reference. The distance values represent the genetic divergences of strain M1077 to the other taxa. The distance values were extracted from the maximum likelihood distance matrices used otherwise by Paup to infer the neighbor-joining trees during phylogenetic analyses, and fed into a spread-sheet program. Strains CCMP 152, CCAC 0031 and M2180 were genetically identical to strains M1077, CCAP 979/46 and CCAC 0056, respectively, thus, were omitted from the chart diagram. Nucleus, concatenated nuclear SSU rDNA, ITS2 and partial LSU rDNA; nucleomorph, nucleomorph SSU rDNA; plastid, rbcL gene. Taxon designations (abscissa): py, C. pyrenoidifera CCAP 979/61; ma1, C. marssonii CCAC 0086; ma2, C. marssonii CCAC 0103; cu1, C. curvata CCAC 0006; cu2, C. curvata CCAC 0080; te1, C. tetrapyrenoidosa M1092; te2, C. tetrapyrenoidosa NIES 279; ov1, C. ovata CCAC 0064; ov2, C. ovata M1171; NP1, NoPyr strain CCAP 979/46; NP2, NoPyr strain CCAC 0109; NP3, NoPyr strain M0741; gy, C. gyropyrenoidosa CCAC 0108; lu, C. lundii CCAC 0107; bo1, C. borealis CCAC 0113; bo2, C. borealis SCCAP K-0063; pa1, C. paramaecium M2452; pa2, C. paramaecium CCAP 977/1; pa3, C. paramaecium CCAC 0056.
Codon usage analysis
In Table 1, the codon usages of the six amino acids with two-fold degenerate NNY codons (asparagine, histidine, aspartate, tyrosine, cysteine and phenylalanine) are listed in absolute counts computed from the 396 codons that were included in the phylogenetic analyses. Cysteine was exceptional in codon usage in that it mostly showed a preference for UGU over UGC, thus it will not be further discussed (Table 1). For the remaining five amino acids, NNC codons were always preferred over NNU codons in C. marssonii, C. pyrenoidifera, C. tetrapyrenoidosa and in clade NoPyr (Table 1). In all strains of clade LB, on the other hand, indications for a change of codon usage were found, although to different extent. In almost all strains of C. paramaecium (except for histidine in strain CCAP 977/1) and C. borealis (except for asparagine in strain SCCAP K-0063) codon preferences were inversed from NNC to NNU for these amino acids (Table 1). C. lundii and C. gyropyrenoidosa were in an intermediate position concerning codon usages. In C. lundii only in two amino acids, aspartate and tyrosine, codon usage was inversed to prefer GAU over GAC (aspartate) and UAU over UAC (tyrosine), whereas in C. gyropyrenoidosa only one amino acid, aspartate, was affected (Table 1). Inversed codon usages were also found in two clades that were not part of LB, C. curvata (histidine and aspartate in strain CCAC 0080, aspartate in strain CCAC 0006) and C. ovata (aspartate and tyrosine; Table 1). In the three phylogenies, C. ovata displayed slightly increased evolutionary rates in nucleomorph SSU rDNA and rbcL phylogenies, whereas C. curvata had slightly longer branches only in the nuclear ribosomal DNA phylogeny (Figure 1A to 1C).
Table 1 Codon usage of two-fold degenerate NNY codons in Cryptomonas sp. and Guillardia theta rbcL
Clade Asn His Asp Tyr Cys Phe
Strain AAC/AAU CAC/CAU GAC/GAU UAC/UAU UGC/UGU UUC/UUU
Guillardia theta 15/0 8/2 12/8 12/3 0/7 12/5
C. curvata
CCAC 0006 13/2 7/3 7/13 10/5 4/5 14/3
CCAC 0080 13/2 4/6 10/10 11/4 3/6 11/6
C. marssonii
CCAC 0086 16/0 6/4 14/6 12/3 1/7 11/6
CCAC 0103 15/1 6/4 10/9 13/2 2/5 10/3
C. ovata
CCAC 0064 8/8 8/2 7/13 5/10 0/8 10/7
M1171 10/6 5/5 7/13 5/10 1/7 10/7
C. pyrenoidifera
CCAP 979/61 15/0 8/2 12/8 13/3 0/7 12/5
CCMP 152 and M1077 15/0 9/1 13/7 12/3 1/6 13/4
C. tetrapyrenoidosa
M1092 13/3 8/2 13/6 10/5 3/5 12/5
NIES 279 12/4 8/2 10/9 10/5 1/7 13/4
NoPyr
CCAC 0031 and CCAP 979/46 15/0 10/0 14/6 13/2 4/3 16/1
CCAC 0109 15/0 9/1 15/5 10/5 1/5 15/2
M0741 15/0 6/4 13/7 9/6 3/3 15/2
clade LB (basally diverging taxa first)
C. gyropyrenoidosa
CCAC 0108 12/4 5/5 9/10 11/4 1/5 12/6
C. lundii
CCAC 0107 12/4 6/4 4/15 4/11 0/7 9/8
C. borealis
CCAC 0113 6/9 2/8 2/18 3/12 0/8 5/12
SCCAP K-0063 8/7 2/8 5/14 4/11 1/7 5/12
C. paramaecium
CCAC 0056 and M2180 6/9 3/7 5/15 6/9 0/7 4/13
CCAP 977/1 6/9 6/4 4/16 4/11 1/6 5/12
M2452 5/10 3/7 7/12 7/8 1/6 6/11
The entries for the rbcL genes in Table 1 refer to absolute counts per 396 amino acids (= 1188 nucleotide positions). Also for the Guillardia theta rbcL gene only the corresponding 396 of the 488 codons were used. The plastid of G. theta contains genes for 30 tRNAs [10]. Only the NNC codons (underlined) are served by an exactly matching tRNA. As a code table, the eubacterial/plastid code was used (code table 11). Bold face, changed codon usage from NNC to NNU = shift from selection bias to genome composition bias.
Discussion
Lineage-specific parallel evolution across three genomes in Cryptomonas
Most of the Cryptomonas clades were recovered with high support values in phylogenies of the concatenated nuclear ribosomal DNA sequences (SSU rDNA, ITS2 and partial LSU rDNA), of the nucleomorph SSU rDNA and of the plastid-encoded rbcL gene, but obvious differences in evolutionary rates among the different clades and genomes were displayed.
In the "super-clade" LB, consisting of three photosynthetic (C. borealis, C. gyropyrenoidosa and C. lundii) and one heterotrophic Cryptomonas species (C. paramaecium), largely congruent branching patterns and extreme evolutionary rates in the nucleomorph and plastid gene phylogenies suggested coevolution under similar selective constraints, as if the two genomes of the complex plastid were a genetic unit in this clade. In the nuclear ribosomal DNA phylogeny, an increase in evolutionary rates was in part also present but less pronounced. Support for this clade was low in the nuclear ribosomal DNA phylogeny, although several parts of the nuclear ribosomal operon were concatenated to improve resolution (the nuclear SSU rDNA alone failed to recover clade LB, but increased evolutionary rates in C. paramaecium and C. borealis were more pronounced than in the concatenated data set; not shown).
In a different clade, that also consists of photosynthetic and colorless Cryptomonas taxa, clade NoPyr, the situation was reversed; coevolution with increased evolutionary rates seemed to have taken place in the nuclear and nucleomorph genes [12], whereas no acceleration of evolutionary rates could be observed in the rbcL gene phylogeny (this study). We did not obtain an rbcL PCR product, however, from the colorless strains of clade NoPyr [[12], this study].
Cho et al. [21] demonstrated that extremely accelerated evolutionary rates were present in three mitochondrial genes (two protein-coding genes, cox1, atp1, and one RNA-coding gene, rrn16, the gene for the SSU rDNA in mitochondria) in the flowering plant genus Plantago, but not in plastid or nuclear genes of the same taxa. We chose two RNA-coding genes and a protein-coding gene as representatives for three of the four genomes in Cryptomonas. Despite their differing functions, the phylogenetic trees suggested that at least two (clade NoPyr), or even all three genomes (clade C. borealis and C. paramaecium) may have evolved in parallel under similar selective constraints or by interacting with each other.
Evidence for relaxed functional constraints on RNA- or protein-coding genes
Possible explanations for accelerated evolutionary rates of DNA sequences include relaxation or loss of functional constraints due to either changes in mode of nutrition, adaptations to new environmental conditions, genetic bottlenecks or obligate asexuality [22-25]. Endosymbiotic, parasitic and organellar genomes are notorious for high A+T contents in their genomes likely caused by biased substitution rates under neutral mutational pressure [19,20,26,27]. Minimum amounts of G and C are required to maintain the codon information for a functional protein or to preserve the secondary structure of an RNA. Depending on the strengths of the functional constraints, the resulting selection bias may differ from the genome composition bias to varying degrees (reviewed for protein-coding genes in [16]). Thus, lineage-specific relaxed selective constraints may be identified by increases in A+T content.
This notion is supported by the observation that the nucleomorph SSU rRNA genes in clade LB accumulate mononucleotide repeats of A and T in highly variable regions [11,12]. In functional protein-coding genes, the triplet structure constrains mutation rates by selection. Synonymous substitutions do not replace amino acids, thus are more likely to occur than nonsynonymous substitutions. In previous studies, however, also synonymous substitutions were reported to be skewed towards specific codons in correlation with expression levels of the respective protein [15,28,29]. The codon biases were explained as a result of a competition between selection and genome compositional bias [17]. In highly expressed genes, codons with abundant or perfectly matching tRNAs (major codons) are apparently preferred over codons are translated by rare or "wobbling" tRNAs (minor codons) [29,30]. In plastid genomes, usually only 30 to 31 tRNAs are available to translate all 61 codons (in the Guillardia theta plastome, 30 tRNAs were found) [10,16]. In two-fold degenerate NNY codons, the preferred major codon in highly expressed genes is usually NNC, thus codon bias due to selection can be comparably easily distinguished from codon bias due to mutation drift [16,18]. Among the six amino acids with two-fold degenerate codons (asparagine, aspartate, histidine, tyrosine, phenylalanine and cysteine), cysteine seems to be the sole exception [[16], this study].
In the rbcL genes of most Cryptomonas clades, the major NNC codons for asparagine, aspartate, histidine, tyrosine, or phenylalanine were preferred over their NNU alternatives, however, codon preferences were reversed in several or all of these amino acids in clade LB [this study]. There was even a gradient of decreasing selective constraints and presumably also expression levels along the LB clade: In the early diverging C. gyropyrenoidosa and C. lundii, reversed codon usages were found in only one or two amino acids, whereas in the late diverging C. borealis and C. paramaecium in four or five NNY-coded amino acids, NNU codons were preferred. Morton and Levin used the codon adaptation index (CAI) to compare codon usage of two-fold degenerate NNY codons in psbA genes among dicot and monocot plants, and discussed putatively decreasing selective constraints from basally to terminally diverging lineages [18]. However, no hemi- or holoparasitic angiosperm plants were included in their study.
In previous studies, convergent codon usage resulted in artificial tree topologies [31]. Despite an obvious bias in codon usage, the rbcL phylogeny was largely confirmed by the nuclear and nucleomorph ribosomal DNA phylogenies. It is likely that the rbcL genes examined in this study had not yet diverged enough to cause artifacts. It may have been different, though, if rbcL had been used to infer phylogenetic trees across cryptophyte genera. For higher level phylogenies, it may, thus, be a better option to use protein sequences instead.
Potential causes for lineage-specific accelerated rates and relaxed functional constraints
One of the explanations for accelerated evolutionary rates and relaxed functional constraints in plastid genomes is loss of photosynthesis, since this usually results in large-scale degradation and compaction of plastomes leading to loss of almost all photosynthetic genes, except perhaps for rbcL [22,32,33].
In the cryptophyte Guillardia theta, photosynthesis genes are spread across plastid (46 genes), nucleomorph (30 genes) and nucleus (α-subunits of phycoerythrin, and an unknown number of additional photosynthetic genes) [9,10,34,35]. Thus, loss of photosynthesis is not an unlikely explanation for a parallel acceleration of evolutionary rates across three genomes in C. paramaecium. However, observations of elevated evolutionary rates in closely related photosynthetic taxa contradict this notion. Instead of being the cause for increased substitution rates, loss of photosynthesis may rather be a result of an accelerated evolution that had started already in the photosynthetic ancestors of the colorless lineages [[12], this study]. Similar observations have been made in plastid gene phylogenies of hemi- and holoparasitic land plants [36].
Previous studies have shown that mutations in genes of DNA repair or DNA replication may result in overall increases of substitution rates in bacterial and eukaryotic genomes, e.g. [37]. Many genes of the cryptophyte nucleomorph have been transferred to the host nucleus including DNA polymerases [9,10,35]. Thus, some potential mutator genes in cryptophytes can be expected to be nuclear-encoded. It is tempting to speculate that a spontaneous mutation in a nuclear-encoded plastid-targeted protein, for example in the proofreading subunit of a DNA polymerase III or in a DNA repair enzyme, could have accelerated successively mutation rates in nuclear, nucleomorph and plastid DNA in the photosynthetic ancestors of clade LB. Accelerated mutation rates may have resulted in loss of photosynthesis in C. paramaecium, which in turn perhaps resulted in less functional constraints on the rbcL protein and, thus, in further increase of evolutionary rates.
Another possible cause for accelerated evolutionary rates in clade LB was discussed previously [12]. The genus Cryptomonas is dimorphic, a feature that usually correlates with sexual reproduction. Final proof for sexual reproduction is still missing, but, however, in clade LB only strains with campylomorph cells were found[11,12]. It may, thus, as well be possible that loss of sexual reproduction caused an increase in mutation rates by loss of recombination affecting also nuclear-encoded plastid-targeted proteins. However, the observed increase in evolutionary rates from host to nucleomorph to plastid genome suggests that the evolutionary processes may have started in the plastid genome.
Conclusion
An rbcL phylogeny of twenty-three Cryptomonas strains was compared with phylogenetic trees inferred from nucleomorph and nuclear ribosomal DNA sequences. In a super-clade comprising photosynthetic and colorless Cryptomonas species, a congruent increase in evolutionary rates and a similar branching pattern were found in data sets representing the two genomes of the complex plastid, the nucleomorph SSU rDNA and the rbcL data set. In both data sets, the colorless strains displayed the highest substitution rates. A direct comparison of the genetic distances across nuclear, nucleomorph and plastid data sets showed that the evolutionary rates in the long-branch super-clade were highest in the rbcL genes and lowest in the nuclear ribosomal DNA. Perhaps evolutionary rates first accelerated in the plastid genome, then in the nucleomorph genome. The increased evolutionary rates of nucleomorph SSU rDNA and rbcL gene evolved in parallel with a gradual shift in codon usage of the rbcL gene towards a relax in functional constraints and decreasing expression levels. Strongest evidence for relaxed functional constraints and decreased expression levels in rbcL were found in the terminally diverging photosynthetic species Cryptomonas borealis and in the colorless species C. paramaecium. Either loss of photosynthesis was a gradual at first hidden process starting already in pigmented ancestors of the colorless C. paramaecium strains or the accelerated evolutionary rates caused defects in the photosynthetic genes resulting in loss of photosynthesis.
Methods
Algal cultures
Photosynthetic and heterotrophic Cryptomonas strains were obtained from different algal culture collections (Table 2). Photosynthetic strains were maintained in modified WARIS-H freshwater culture medium [38,39], and heterotrophic strains in biphasic soil/water medium with one-eighth of a pea for supply with organic substances. Strains were grown at 15°C under a 14/10 h light/dark regime (15–35 μmol photons m-2 s-1; photosynthetic strains) or in the dark (colorless strains).
Table 2 List of Cryptomonas strains examined in this study with accession numbers to EMBL/GenBank/DDBJ entries
Species/Clade Strain Nucleus Nucleom. Plastid
ITS2+LSUp SSU rDNA SSU rDNA rbcL
C. borealis CCAC 0113 [EMBL:AJ566160] [EMBL:AM051188] [EMBL:AJ566185] [EMBL:AM051202]
SCCAP K-0063 [EMBL:AJ566159] [EMBL:AJ420696] [EMBL:AJ420688] [EMBL:AM051203]
C. curvata CCAC 0006 [EMBL:AJ566147] [EMBL:AJ007280] [EMBL:AJ420682] [EMBL:AM051204]
CCAC 0080 [EMBL:AJ566148] [EMBL:AM051189] [EMBL:AJ715462] [EMBL:AM051205]
C. gyropyrenoidosa CCAC 0108 [EMBL:AJ566154] [EMBL:AJ421149] [EMBL:AJ420686] [EMBL:AM051206]
C. lundii CCAC 0107 [EMBL:AJ566161] [EMBL:AM051190] [EMBL:AJ566184] [EMBL:AM051207]
C. marssonii CCAC 0086 [EMBL:AJ566155] [EMBL:AM051191] [EMBL:AJ566173] [EMBL:AM051208]
CCAC 0103 [EMBL:AJ715444] [EMBL:AM051192] [EMBL:AJ566174] [EMBL:AM051209]
C. ovata CCAC 0064 [EMBL:AJ566153] [EMBL:AM051193] [EMBL:AJ566178] [EMBL:AM051210]
M1171 [EMBL:AJ566152] [EMBL:AJ420695] [EMBL:AJ420687] [EMBL:AM051211]
C. paramaecium CCAC 0056 [EMBL:AJ566158] [EMBL:AJ007276] [EMBL:AJ420676] [EMBL:AM051212]
CCAP 977/1 [EMBL:AJ715445] [EMBL:AM051194] [EMBL:AJ715465] [EMBL:AM051213]
M2180 [EMBL:AJ715451] [EMBL:AM051195] [EMBL:AJ715471] [EMBL:AM051214]
M2452 [EMBL:AJ715452] [EMBL:AM051196] [EMBL:AJ715472] [EMBL:AM051215]
C. pyrenoidifera CCAP 979/61 [EMBL:AJ566142] [EMBL:AJ421147] [EMBL:AJ420684] [EMBL:AM051216]
CCMP 152 [EMBL:AJ566140] [EMBL:AJ421150] [EMBL:AJ420675] [EMBL:AM051217]
M1077 [EMBL:AJ566144] [EMBL:AM051197] [EMBL:AJ566180] [EMBL:AM051218]
C. tetrapyrenoidosa M1092 [EMBL:AJ566146] [EMBL:AM051198] [EMBL:AJ566182] [EMBL:AM051219]
NIES 279 [EMBL:AJ715455] [EMBL:AM051199] [EMBL:AJ566183] [EMBL:AM051220]
NoPyr CCAC 0031 [EMBL:AJ566166] [EMBL:AJ007281] [EMBL:AJ420685] [EMBL:AM051221]
CCAC 0109 [EMBL:AJ566165] [EMBL:AJ420697] [EMBL:AJ420683] [EMBL:AM051222]
CCAP 979/46 [EMBL:AJ566167] [EMBL:AM051200] [EMBL:AJ566171] [EMBL:AM051223]
M0741 [EMBL:AJ566163] [EMBL:AM051201] [EMBL:AJ566172] [EMBL:AM051224]
Abbreviations: CCAC, Culture Collection of Algae at the University of Cologne (Germany); CCAP, Culture Collection of Algae and Protozoa (UK); CCMP, Provasoli-Guillard Center for the Culture of Marine Phytoplankton (USA); ITS2, nuclear internal transcribed spacer 2; LSUp, partial nuclear large subunit ribosomal DNA (28S rDNA or LSU rDNA, approx. 800 nt of 5' terminus); M, Algal Culture Collection Melkonian at the University of Cologne (Germany); NIES, Culture Collection of The National Institute for Environmental Studies (Japan); rbcL, large subunit gene of ribulose-1,5-bisphosphate carboxylase/oxygenase; SCCAP, Scandinavian Culture Centre for Algae and Protozoa (Denmark); SSU rDNA, small subunit ribosomal DNA (18S rDNA, nuclear or nucleomorph).
New sequences: acc. nos. AM051188 to AM051224.
Isolation of DNA, PCR amplification and sequencing
Total genomic DNA was isolated from the cells with the DNeasy Plant Mini Kit according to the manufacturer's protocol (Qiagen, Hilden, Germany). PCR amplification of nuclear SSU rDNA, ITS2 and partial LSU rDNA, and of nucleomorph SSU rDNA with nucleus- or nucleomorph-specific primers followed previously described protocols [11,40]. For PCR amplification of cryptophyte rbcL genes, new primers were designed using an alignment of bangiophyte or florideophycean red algal and cryptophyte rbcL sequences (cryptophyte sequences: Chroomonas sp., acc. no. AY119781; Cryptomonas paramaecium, acc. no. AY119780; Guillardia theta, acc. no. AF041468; Pyrenomonas helgolandii, acc. no. AY199782). Similarly, new sequencing primers were constructed using the same alignment (sequences of PCR primers and sequencing primers for rbcL are listed in Additional file 2). For PCR amplification of rbcL DNA sequences, the same cycling protocol as for the ribosomal DNA sequences was used except for a decrease of the annealing temperature (predenaturation for 3 min. at 95°C; 30 cycles: 1 min. at 95°C, 2 min. at 45 or 50°C, 3 min. at 68°C). PCR products were purified with the Dynabead M-280 system according to the manufacturer's protocol (Dynal, Oslo, Norway). For bidirectional sequencing, two sets of primer pairs were used for each PCR product; the forward primers were labeled with IRDye-800 and the reverse primers with IRDye-700 (see Additional file 2). Double-stranded sequences were determined with a Li-Cor 4200L bidirectional sequencer (Li-Cor Biosciences, Bad Homburg, Germany).
Phylogenetic analyses
The rbcL nucleotide and protein sequences were prealigned with clustalw and refined by eye using the multiple alignment sequence editor SeaView [41]. The ribosomal DNA sequences were manually aligned according to secondary structure; non-alignable regions were excluded prior to the phylogenetic analyses.
Since the taxon sampling was congruent for plastid-, nucleomorph- and nucleus-encoded sequences, all unrooted data sets comprised 23 taxa (accession nos. are listed in Table 2). The unrooted rbcL nucleotide data set consisted of 1188 positions and was translated to perform phylogenetic analyses of protein sequences or modified for phylogenetic analyses of single codon positions (396 positions each data set; see Additional file 1 for additional information about the data sets). A rooted data set of rbcL nucleotide sequences consisted of 46 taxa and 990 positions, including 14 rhodophyte rbcL sequences as outgroup taxa (Additional file 3). The nuclear ribosomal DNA sequences were concatenated for phylogenetic analyses resulting in a data set with a total length of 2623 nucleotides (complete nuclear ITS2 and partial nuclear LSU rDNA comprising approx. 800 nt of the 5' terminus: 1083 positions; nuclear SSU rDNA: 1540 positions). The nucleomorph SSU rDNA data set comprised 1496 positions.
All nucleotide data sets were subjected to distance, maximum likelihood, maximum parsimony and Bayesian analyses. To determine the evolutionary model fitting best the data according to the Akaike Information Criterion (AIC), Modeltest 3.6 was used [42]). Distance, maximum likelihood and maximum parsimony analyses were performed with the program PAUP* 4.0b10 [43]. Distance analyses were run under minimum evolution and set to the maximum likelihood parameters proposed by Modeltest. Data sets with heterogeneous base frequencies were also analyzed using the LogDet transformation. For both types of analyses, trees were inferred with the neighbor-joining algorithm. Maximum likelihood analyses were done using the proposed evolutionary model settings of Modeltest with three random addition replicates and heuristic tree search algorithm with tree bisection and reconnection (TBR). Unweighted maximum parsimony analyses were performed using 10 random addition replicates also in combination with the heuristic tree search algorithm. For all analyses under the distance or maximum parsimony criterion, 1000 bootstrap replicates were calculated; for maximum likelihood, 500 bootstrap replicates were computed. Bayesian analyses were performed using MrBayes 3.0B4 [44]. For the nucleotide data sets, likelihood settings were set to GTR, gamma-distributed among-site rate variation and covarion (includes proportion of invariable sites). Samples were drawn every 100th generation for at least 3.5 million generations with one cold and three heated chains. Burn-in was determined for the individual data set according to the sump plot.
The protein data set was also subjected to distance, maximum likelihood, maximum parsimony and Bayesian analyses. The evolutionary model fitting best the data was determined with ProtTest 1.2.6 according to the AIC [45,46] and used for maximum likelihood analysis with Phyml 2.4.4 [47]. Distance analysis was performed using protdist from the Phylip 3.62 package (set to JTT+Γ with global rearrangements; progam suite by Joe Felsenstein [48]). The shape parameter α for the gamma distribution in protdist was calculated using Tree-Puzzle 5.2 [49]. Maximum parsimony analysis was done using PAUP* 4.0b10 (10 random addition sequence replicates). For Bayesian analyses with MrBayes 3.0b4, prior expectations were set to AAmodel=mixed (priors for all amino acid substitution matrices considered equal) and likelihood settings to gamma-distributed among-site rate variation and proportion of invariable sites. Samples were drawn every 100th generation for 5 million generations using one cold and three heated chains. Burn-in was determined according to the sump plot.
Codon usage analysis
The countcodon program from the web site of the Codon Usage Database [50] was used to determine absolute counts of all codons of the twenty-three rbcL sequences (396 codons). The DNA sequences were translated using the eubacterial/plastid codon table (code table 11).
Authors' contributions
KHE sequenced fourteen nuclear SSU rDNAs, five rbcL sequences, aligned the nuclear and nucleomorph data sets, performed the phylogenetic analyses and the codon usage analysis, and wrote the manuscript; HDT sequenced eighteen rbcL sequences and did the rbcL alignment; MM contributed to planning of the study and critically revised the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Evolutionary models and support values for the Cryptomonas clades in the unrooted rbc L data set.
Click here for file
Additional File 2
PCR and sequencing primers for cryptophyte rbcL
Click here for file
Additional File 3
Rooted maximum likelihood tree of cryptophyte rbc L sequences including all codon positions.
Click here for file
Additional File 4
Rooted maximum likelihood tree of concatenated nuclear SSU rDNA, partial nuclear LSU rDNA, nucleomorph SSU rDNA and rbc L sequences
Click here for file
Acknowledgements
This study was supported by a grant from the International Graduate School in Genetics and Functional Genomics at the University of Cologne to HDT.
==== Refs
Kellogg EA Juliano ND The structure and function of RuBisCO and their implications for systematic studies Am J Bot 1997 84 413 428
Spreitzer RJ Salvucci ME Rubisco: structure, regulatory interactions, and possibilities for a better enzyme Annu Rev Plant Biol 2002 53 449 475 12221984 10.1146/annurev.arplant.53.100301.135233
Tabita FR Microbial ribulose 1,5-bisphosphate carboxylase/oxygenase: A different perspective Photosynthesis Res 1999 60 1 28 10.1023/A:1006211417981
Rodermel S Subunit control of Rubisco biosynthesis – a relic of an endosymbiotic past? Photosynthesis Res 1999 59 105 123 10.1023/A:1006122619851
Ellis RJ The most abundant protein in the world Trends Biochem Sci 1979 4 241 244 10.1016/0968-0004(79)90212-3
Bricaud CH Thalouarn P Rinaudin S Ribulose-1,5-bisphosphate carboxylase activity in the holoparasite Lathraea clandestina L J Plant Physiol 1986 125 367 370
Siemeister G Hachtel W Structure and expression of a gene encoding the large subunit of ribulose-1,5-bisphosphate carboxylase (rbcL) in the colorless euglenoid flagellate Astasia longa Plant Mol Biol 1990 14 825 833 2102860 10.1007/BF00016515
Schwender J Goffman F Ohlrogge JB Shachar-Hill Y Rubisco without the Calvin cycle improves the carbon efficiency of developing green seeds Nature (Lond) 2004 432 779 782 15592419 10.1038/nature03145
Douglas SE Zauner S Fraunholz M Beaton M Penny S Deng L-T Wu X Reith M Cavalier-Smith T Maier U-G The highly reduced genome of an enslaved algal nucleus Nature (Lond) 2001 410 1091 1096 11323671 10.1038/35074092
Douglas SE Penny SL The plastid genome of the cryptophyte alga, Guillardia theta: Complete sequence and conserved synteny groups confirm its common ancestry with red algae J Mol Evol 1999 48 236 244 9929392
Hoef-Emden K Melkonian M Revision of the genus Cryptomonas (Cryptophyceae): a combination of molecular phylogeny and morphology provides insights into a long-hidden dimorphism Protist 2003 154 371 409 14658496 10.1078/143446103322454130
Hoef-Emden K Multiple independent losses of photosynthesis and differing evolutionary rates in the genus Cryptomonas (Cryptophyceae) – combined phylogenetic analyses of DNA sequences of the nuclear and nucleomorph ribosomal operons J Mol Evol 2005 60 183 195 15785847 10.1007/s00239-004-0089-5
Bousquet J Strauss SH Doerksen AH Price RA Extensive variation in evolutionary rate of rbcL sequences among seed plants Proc Natl Acad Sci USA 1992 89 7844 7848 1502205
Wolfe AD de Pamphilis CW The effect of relaxed functional constraints on the photosynthetic gene rbcL in photosynthetic and nonphotosynthetic parasitic plants Mol Biol Evol 1998 15 1243 1258 9787431
Sharp PM Li W-H The rate of synonymous substitution in enterobacterial genes is inversely related to codon usage bias Mol Biol Evol 1987 4 222 230 3328816
Morton BR Codon bias and the context dependency of nucleotide substitutions in the evolution of plastid DNA Evol Biol 2000 31 55 103
Bulmer M The selection-mutation-drift theory of synonymous codon usage Genetics 1991 129 897 907 1752426
Morton BR Levin JA The atypical codon usage of the plant psbA gene may be the remnant of an ancestral bias Proc Natl Acad Sci USA 1997 94 11434 11438 9326627 10.1073/pnas.94.21.11434
Lynch M Mutation accumulation in nuclear, organelle, and prokaryotic transfer RNA genes Mol Biol Evol 1997 14 914 925 9287424
Moran NA Accelerated evolution and Muller's ratchet in endosymbiotic bacteria Proc Natl Acad Sci USA 1996 93 2873 2878 8610134 10.1073/pnas.93.7.2873
Cho Y Mower JP Qiu Y-L Palmer JD Mitochondrial substitution rates are extraordinarily elevated and variable in a genus of flowering plants Proc Natl Acad Sci USA 2004 101 17741 17746 15598738 10.1073/pnas.0408302101
dePamphilis CW Palmer JD Loss of photosynthetic and chlororespiratory genes from the plastid genome of a parasitic flowering plant Nature (Lond) 1990 348 337 339 2250706 10.1038/348337a0
Charlesworth D Morgan MT Charlesworth B Mutation accumulation in finite outbreeding and inbreeding populations Genet Res 1993 61 39 56
Lynch M and Blanchard JL Deleterious mutation accumulation in organelle genomes Genetica 1998 103 29 39 9720269 10.1023/A:1017022522486
Dufresne A Garczarek L Partensky F Accelerated evolution associated with genome reduction in a free-living prokaryote Genome Biol 2005 6 1 10 10.1186/gb-2005-6-2-r14
Maier UG Douglas SE Cavalier-Smith T The nucleomorph genomes of the cryptophytes and the chlorarachniophytes Protist 2000 151 103 109 10965950 10.1078/1434-4610-00011
Wernegreen JJ Funk DJ Mutation exposed: A neutral explanation for extreme base composition of an endosymbiont genome J Mol Evol 2004 59 849 858 15599516 10.1007/s00239-003-0192-z
Gouy M Gautier C Codon usage in bacteria: Correlation with gene expressivity Nucleic Acids Res 1982 10 7055 7074 6760125
Ikemura T Codon usage and tRNA content in unicellular and multicellular organisms Mol Biol Evol 1985 2 13 34 3916708
Anderson SGE Kurland CG Codon preferences in free-living microorganisms Microbiol Rev 1990 54 198 210 2194095
Inagaki Y Simpson AGB Dacks JB Roger AJ Phylogenetic artifacts can be caused by leucine, serine, and arginine codon usage heterogeneity: Dinoflagellate plastid origins as a case study Syst Biol 2004 53 582 593 15371248 10.1080/10635150490468756
Delavault PM Russo NM Lusson NA Thalouarn PA Organization of the reduced plastid genome of Lathraea clandestina, an achlorophyllous parasitic plant Physiol Plant 1996 96 674 682 10.1034/j.1399-3054.1996.960418.x
Gockel G Hachtel W Complete gene map of the plastid genome of the nonphotosynthetic euglenoid flagellate Astasia longa Protist 2000 151 347 351 11212895 10.1078/S1434-4610(04)70033-4
Jenkins J Hiller RG Speirs J Godovac-Zimmermann J A genomic clone encoding a cryptophyte phycoerythrin α-subunit. Evidence for three α-subunits and an N-terminal membrane transit sequence FEBS Letters 1990 273 191 194 2226853 10.1016/0014-5793(90)81082-Y
Deane JA Fraunholz M Su V Maier U-G Martin W Durnford DG McFadden GI Evidence for nucleomorph to host nucleus gene transfer: Light-harvesting complex proteins from cryptomonads and chlorarachniophytes Protist 2000 151 239 252 11079769 10.1078/1434-4610-00022
dePamphilis CW Young ND and Wolfe AD Evolution of the plastid gene rps2 in a lineage of hemiparasitic and holoparasitic plants: Many losses of photosynthesis and complex patterns of rate variation Proc Natl Acad Sci USA 1997 94 7367 7372 9207097 10.1073/pnas.94.14.7367
Dzierzbicki P Koprowski P Fikus MU Malc E Ciesla Z Repair of oxidative damage in mitochondrial DNA of Saccharomyces cerevisiae: involvment of the MSH1-dependent pathway DNA Repair 2004 3 403 411 15010316 10.1016/j.dnarep.2003.12.005
Kies L Über die Zweiteilung und Zygotenbildung bei Roya obtusa (Bre.) West & West Mitt Staatsinst Allg Bot Hamb 1967 12 35 42
McFadden GI Melkonian M Use of HEPES buffer for microalgal culture media and fixation for electron microscopy Phycologia 1986 25 551 557
Hoef-Emden K Marin B Melkonian M Nuclear and nucleomorph SSU rDNA phylogeny in the Cryptophyta and the evolution of cryptophyte diversity J Mol Evol 2002 55 161 179 12107593 10.1007/s00239-002-2313-5
Galtier N Gouy M Gautier C SeaView and Phylo_win, two graphic tools for sequence alignment and molecular phylogeny Comput Applic Biosci 1996 12 543 548
Posada D Crandall KA Modeltest: testing the model of DNA substitution Bioinformatics 1998 14 817 818 9918953 10.1093/bioinformatics/14.9.817
Swofford D PAUP*: Phylogenetic Analyses Using Parsimony (* and Other Methods) 40 Beta for UNIX or OpenVMS 2002 Sunderland, Massachusetts: Sinauer Associates
Ronquist F Huelsenbeck JP MrBayes 3: Bayesian phylogenetic inference under mixed models Bioinformatics 2003 19 1572 1574 12912839 10.1093/bioinformatics/btg180
Drummond A Strimmer K PAL: an object-oriented programming library for molecular evolution and phylogenetics Bioinformatics 2001 17 662 663 11448888 10.1093/bioinformatics/17.7.662
Abascal F Zardoya R Posada D Prottest: selection of best-fit models of protein evolution Bioinformatics 2005 21 2104 2105 15647292 10.1093/bioinformatics/bti263
Guindon S Gascuel O A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood Syst Biol 2003 52 696 704 14530136 10.1080/10635150390235520
The Phylip web site
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
The Codon Usage Database
Rodríguez J Oliver JF Marín A Medina JR The general stochastic model of nucleotide substitution J Theor Biol 1990 142 485 501 2338834
Adachi J Waddell PJ Martin W Hasegawa M Plastid genome phylogeny and a model of amino acid substitution for proteins encoded by chloroplast DNA J Mol Evol 2000 50 348 358 10795826
|
16232313
|
PMC1285359
|
CC BY
|
2021-01-04 16:29:16
|
no
|
BMC Evol Biol. 2005 Oct 18; 5:56
|
utf-8
|
BMC Evol Biol
| 2,005 |
10.1186/1471-2148-5-56
|
oa_comm
|
==== Front
BMC GenetBMC Genetics1471-2156BioMed Central London 1471-2156-6-501622568110.1186/1471-2156-6-50Research ArticleCharacterization of a likelihood based method and effects of markers informativeness in evaluation of admixture and population group assignment Yang Bao-Zhu [email protected] Hongyu [email protected] Henry R [email protected] Joel [email protected] Yale University School of Medicine, Department of Psychiatry, New Haven, CT, USA2 VA CT Healthcare Center, West Haven, CT, USA3 Yale University School of Medicine, Departments of Epidemiology and Public Health, and Genetics, New Haven, CT, USA4 University of Connecticut Health Center, Farmington, CT, USA2005 14 10 2005 6 50 50 6 6 2005 14 10 2005 Copyright © 2005 Yang et al; licensee BioMed Central Ltd.2005Yang 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
Detection and evaluation of population stratification are crucial issues in the conduct of genetic association studies. Statistical approaches useful for understanding these issues have been proposed; these methods rely on information gained from genotyping sets of markers that reflect population ancestry. Before using these methods, a set of markers informative for differentiating population genetic substructure (PGS) is necessary. We have previously evaluated the performance of a Bayesian clustering method implemented in the software STRUCTURE in detecting PGS with a particular informative marker set. In this study, we implemented a likelihood based method (LBM) in evaluating the informativeness of the same selected marker panel, with respect to assessing potential for stratification in samples of European Americans (EAs) and African Americans (AAs), that are known to be admixed. LBM calculates the probability of a set of genotypes based on observations in a reference population with known specific allele frequencies for each marker, assuming Hardy Weinberg equilibrium (HWE) for each marker and linkage equilibrium among markers.
Results
In EAs, the assignment accuracy by LBM exceeded 99% using the most efficient marker FY, and reached perfect assignment accuracy using the 10 most efficient markers excluding FY. In AAs, the assignment accuracy reached 96.4% using FY, and >95% when using at least the 9 most efficient markers. The comparison of the observed and reference allele frequencies (which were derived from previous publications and public databases) shows that allele frequencies observed in EAs matched the reference group more accurately than allele frequencies observed in AAs. As a result, the LBM performed better in EAs than AAs, as might be expected given the dependence of LBMs on prior knowledge of allele frequencies. Performance was not dependent on sample size.
Conclusion
The performance of the LBM depends on the efficiency and number of markers, and depends greatly on how representative the available reference allele frequencies are for those of the population being assigned. This method is of value when the parental population is known and relevant allele frequencies are available.
==== Body
Background
Population stratification is a crucial issue in conducting genetic association studies, in particular, for case-control study designs, such that if it is not accounted for study results could be invalid – either false positive or false negative [1]. Methods to address the issue have been proposed [2-17]. Before using these methods, an informative set of markers is necessary; this is known as a set of ancestry informative markers (AIMs). In this study, we implemented a likelihood based method (LBM), as an alternative to popular Bayesian methods such as that implemented in STRUCTURE [3,13], and used it to evaluate the informativeness of a selected marker panel and to assess potential for stratification in a sample of European Americans (EAs) and African Americans (AAs) that are known to be admixed.
Likelihood-based methods (LBMs) provide a framework for assignment of individuals to specific populations based on observed allele frequencies in AIMs. LBMs for the classification of individuals into subgroups can be implemented by calculating the probability of a marker genotype profile (i.e., a set of genotypes) based on observations in a reference population with known specific allele frequencies for each marker ("training frequencies"), assuming Hardy Weinberg equilibrium (HWE) for each marker and linkage equilibrium among markers [18]. The LBM method is also called an "assignment test" and is widely applied in molecular ecology and animal forensics for identifying population genetic substructures for animals or plants [18-24]. Research on the assignment test or LBM has not yet focused on the performance of the test or of specific markers in differentiating the PGS in human subjects. In theory, LBM may be better for probabilistic classification of individuals to subpopulations, if certain conditions are met. The most important of these conditions is availability of an accurate set of training frequencies. Obviously, this method may be applicable only if the populations from which the sample to be classified are already known or can be determined. This condition can be met in most situations; for example, the AA population is well known to have principally African and European American ancestry.
In the present study, we compared the performance of LBMs to that of the popular Bayesian approach used by the software program STRUCTURE. We predicted that, if the conditions for successful LBM application are met, LBMs would be more efficient that Bayesian methods for population group assignment, because they make use of more information (i.e., known ancestral population allele frequencies, which are provided a priori rather than inferred from the data presented to the program).
Results
We calculated the measure of marker efficiency by the metric δ for each marker. (Note that δ as defined here is different from that defined in Rosenberg et al. 2003 [25]). We designated δstudy-AA-EA as the measure of marker efficiency between EA and AA in our study populations, and δreference-study-EA or δreference-study-AA as the quantitative difference in efficiency between marker characteristics as they were reported previously, and as we observed them in the study populations. We observed that the maximum δstudy-AA-EA was 0.82, for the marker FY, and the minimum δstudy-AA-EA was 0.15. The mean was 0.32 and median was 0.28. Larger observed δstudy-AA-EA corresponded to greater marker efficiency for differentiating the EA and AA study populations. Furthermore, smaller values of the δreference-study (including δreference-study-EA or δreference-study-AA) indicate that the marker as observed is more similar to the marker as described in the reference (and therefore the reported allele frequencies were relatively accurate for LBM training). For markers with higher values of this measure, since they did not match the training frequencies as well, their utility in practice was reduced. An efficient classification marker would be one with bigger δstudy-AA-EA and smaller δreference-study when the reference allele frequencies are used for training for the LBM. Figure 1 shows the relationship of these three δ measures; the straight line in the Figure 1(1) indicates the equality of δstudy-AA-EA and δreference-study. Thus, Figure 1(1) illustrates that the majority of the markers have δstudy-AA-EA > δreference-study, and Figure 1(2) shows the ratio of δreference-study-AA to δreference-study-EA with a horizontal line specifying δreference-study-AA = δreference-study-EA. (Twenty-two of 36 markers studied (61%) are above the horizontal line, which indicates that they are less representative (of prior reports) for AAs than for EAs. This reduced correspondence of the observed AA allele frequency compared to the prior reports relative to our observations in EA populations, also causes decreased assignment accuracy in AAs compared to EA – in fact, the assignment accuracy in AAs never reaches 100%. Even with imperfect training frequencies, the LBM using the selected makers to classify individuals into subpopulations still performed very well, with average assignment accuracy of 96.8% and 99.9% for AA and EA respectively.) These results illustrate, further, that the selected marker panel is a relatively informative marker set in differentiating between EAs and AAs.
Figure 1 Marker efficiency in terms of the metric δ. (1) Comparison of delta between for AAs and EAs as observed in our sample, and as reported in the prior literature: δstudy-AA-EA versus δreference-study-EA (red triangle) or δreference-study-AA (blue dot). (2) Ratio for deltas for each marker (AA/EA) as observed in our sample compared to the prior literature: δstudy-AA-EA versus the ratio of δreference-study-AA to δreference-study-EA.
Assignment accuracy
In order to ascertain the smallest sufficient marker set and identify how many makers are needed to reach reasonable assignment accuracy, we took the approach of selecting markers by marker efficiency, as we did previously in evaluating the Bayesian method [1]. The relative assignment accuracy was evaluated by adding markers one-by-one up to 36 markers, with the order of δ either descending or ascending; the results are shown in Figure 2 (This result by LBM can be compared with results from STRUCTURE in Yang et al. 2005 [1]; cf. Figure 3, p. 308). FY was the most informative marker, and due to its unique value in distinguishing the EA and AA populations under study, we performed analyses separately either including or excluding this marker.
Figure 2 Assignment accuracy by LBM. Assignment accuracy by LBM. The markers are adding one by one either by δ descending or ascending. Assignment accuracy without FY, the most efficient marker in the panel studied, was also evaluated.
In EAs (Figure 2, (1)), the assignment accuracy by LBM exceeded 99% using the most efficient marker FY, and reached 100% using the 10 most efficient markers excluding FY (when FY was excluded, the assignment accuracy using the next most efficient marker D11S936 dropped by 9%). In contrast, it would take 29 markers to reach >99% assignment accuracy when the least efficient markers are selected or the seven most efficient markers are omitted. In AAs (Figure 2, (2)), the assignment accuracy reached 96.4% using FY, and then the assignment accuracy changed inconsistently as more markers were added up to 21 markers, at which point assignment accuracy stabilized at 97.6%, achieving the maximum of 98.8% when all 36 markers were used. Overall, using LBM, it can exceed 95% when using at least the 9 most efficient markers. When FY was excluded, the assignment accuracy dropped by 38%.
This 38% drop, which reflects the difference in accuracy between the most efficient marker, FY, and the second most efficient one, D11S936, was further investigated by a corresponding analysis in which the study sample was randomly split into two groups and one group was treated as a reference sample. The drop declined to 6%, which was more comparable to the 9% in EAs. Thus, this reduced accuracy was in large part attributable to mismatch between reported training allele frequencies and frequencies that are more representative of our Northeastern US AA population. LBM never reaches perfect assignment accuracy for AAs in this sample even when all the 36 markers were used, but accuracy did reach 98.8%.
Comparison of observed and reference allele frequencies
The high assignment accuracy by LBMs was observed notwithstanding the deviation between our observed allele frequencies and the reference frequencies described above. We further compared our observed allele frequencies with published reference allele frequencies using the χ2 test. In EAs, after adjusting for sample size, there were 19 markers that differed at p < 0.05, while in AAs, the corresponding number of markers was 29. In other words, allele frequencies observed in EAs matched the reference group more closely than did allele frequencies observed in AAs. As a result, the LBM performed better in EAs than AAs, as might be expected given the dependence of LBMs on prior knowledge of allele frequencies.
Evaluation of the influence of mismatched reference allele frequencies on assignment accuracy by means of split samples
As noted above, in many cases our observed allele frequencies showed nominally significant differences from population reference frequencies. This could reflect, for example, sampling error, or differences in allele frequency for population groups with similar self-identified ethnicity that are assessed at different geographic locations. To further assess the impact of the reference group on the assignment accuracy for LBM, we randomly split our EA and AA study datasets each into two equal-sized samples, treating one as the study group and the other as the reference group. Thus, we were able to model geographically appropriate allele frequencies for each group, at the expense of reducing the analysis sample size by a factor of two. The distributions of the allele frequencies for the two split samples are the same in EAs and AAs for all the markers based on the χ2 test (p-value ranges from > 0.57 to 1). The results (Figure 3) for AAs using internal split samples improved dramatically compared to the results using the external reference group in AAs (Figure 2). These results (Figure 3) illustrate that the performance of the LBM depends greatly on how representative the reference allele frequencies are to those of the population being assigned when the parental population is known.
Figure 3 Assignment accuracy by LBM for Split samples. Assignment accuracy by LBM for Split samples. Split samples were used to evaluate the impact of reference group of allele frequencies on the assignment accuracy by LBM.
Logarithm likelihood ratio
We also calculated the logarithm of the likelihood ratio, expressing the comparison of the probability of being in the EA group compared to the AA group, based on formula (2) (Methods section), and generated a visual display of correct or misplaced group assignment for each individual, adding the markers one by one using a descending value of δ. Figure 4 shows the 12 most efficient markers. The horizontal line represents a log likelihood ratio of zero; those above zero are allocated to EA, and below zero to AA (refer to equation (2)). The vertical line separates the groups. Therefore, those in the upper right and lower left quadrants are misclassified based on self-identified race. The first graph represents the allocation of each individual using only the most efficient marker, FY. As markers are added to the analyses, the log likelihood ratios increase and the separation between clusters become more and more marked. (Note that the Y-axis scale is not constant.)
Figure 4 Logarithm of likelihood ratio for each individual. Logarithm of likelihood ratio for each individual grouping by their self-identified ethnicity. Markers were added one by one with δ descending. The first marker is FY.
One individual in the AA series appeared to be misclassified; see Figure 4 with 9 to 12 markers. Based on this observation, we examined the phenotypic information for this subject, and determined that, although self-identified as AA, the subject had one AA and one EA parent.
Comparison of LBM results with Bayesian results obtained using STRUCTURE
We compared the performance of LBM with results obtained using STRUCTURE and the same panel of markers by Yang et al. 2005 [1] (Figure 5); the samples used for Figure 5 are exactly the same as those for Figure 3 in Yang et al. 2005 [1] (cf. Figure 3, p. 308). In EAs (Figure 5 – (1)), the LBM provided more accurate group assignment than STRUCTURE, with the FY locus included or excluded. In AAs (Figure 5 – (2)), the relative performance of STRUCTURE and LBM was mixed.
Figure 5 Comparison of STRUCTURE and LBM on assignment accuracy. The markers are adding one by one either by delta descending or ascending. Assignment accuracy without FY, the most efficient marker in the panel studied, was also evaluated.
Discussion
The LBM is appealing for population group assignment because it is straightforward and easily implemented, provided that sufficiently accurate reference allele frequencies are available. We provide a set of allele frequencies for all markers herein that will prove useful for classifying populations similar to those discussed in the present article [see Additional file 1]. Under these circumstances, the LBM should classify individuals at least as accurately as STRUCTURE, and probably more accurately. However, a representative reference population may be difficult to establish in some cases. With a good reference group, as shown in the analysis of split samples (p-values of the χ2 test range from >0.57 to 1 for distributions of allele frequencies for the two split samples), LBM performed very well. In EAs, the clustering by LBM is as good as by STRUCTURE (using an ancestry model of admixture and an allele frequency dependence model) for δ descending, but LBM performs better for ascending values of δ. For AAs, LBM and STRUCTURE cluster the groups equally well. STRUCTURE retains certain advantages, such as the ability to classify individuals by proportional ancestry for subsequent application of the structured association method, as discussed elsewhere [1]. It should be noted that the superior performance of LBMs over STRUCTURE, when observed, depends on LBM having more data available than STRUCTURE in the form of reference allele frequencies.
The observed allele frequencies in this study matched reference allele frequencies better for EA than AA populations. Subjects from some populations from different geographic areas might have quite different admixture proportions and ancestral origins. This is demonstrably the case for African-Americans, since in different parts of the US the percent admixture from EAs is known to range at least from 12% to 23% [26]. Another issue with LBM involves justification for the multiplication of allele frequencies across loci under the assumption of linkage equilibrium. If the allele frequencies of different STRs vary among subpopulations, then the loci are not in complete linkage equilibrium or are not statistically independent even if they are genetically unlinked. However, we did assume linkage equilibrium within the subpopulations. This is also an underlying assumption for STRUCTURE [4]. This assumption might prove to be problematic under some circumstances, but the practical impact seemed minimal for the present study, as evidenced by the fact that LBM performed well.
The result from the single most informative marker, FY, could exceed 99% and 96% assignment accuracy in EAs and AAs, respectively. This result is, of course, sample-specific to some extent; AA subjects who are homozygous for the allele more characteristic of European ancestry (i.e., FY (+/+)), should have a population frequency of about 4%, given a 20% admixture rate from EA, and would be misclassified into the EA group if based only on this marker; this misclassification rate is equal to what we observed, about 4% in AAs. Likewise, EAs heterozygous for the FY(-) allele characteristic of AAs are observed as well, and they are liable to be misclassified as AAs. Our Northeastern US AA population had approximated the expected European admixture rate, based on the information from FY.
The sample size of the populations being assigned is not an issue for LBM, while it is for STRUCTURE. The Bayesian cluster approach taken by STRUCTURE requires building a likelihood function from the observed samples to infer allele frequencies, such that if the sample size is insufficient, the estimated allele frequencies might not be accurate. As a result, sample size in each subgroup affects the assignment accuracy, and our simulation result [1] shows that approximately fifty subjects are required to have stable assignment accuracy by STRUCTURE. LBM, in contrast, uses allele frequencies from the reference populations; there is no need to estimate allele frequencies by LBM. Thus, even a single individual can be assigned accurately using the LBM.
We conclude that assignment accuracy by LBM depends on the efficiency of the markers selected (FY alone can separate EAs and AAs with accuracy that can approach 99% for excluding AAs from a presumed EA sample), the number of markers (other things being equal, more markers produce higher assignment accuracy), and greatly on how representative the parental population reference allele frequencies are for the populations being queried.
Methods
Subjects
Three hundred sixty-six individuals recruited in the Northeastern US (classified as 282 EAs, 84 AAs) were studied. These individuals were selected from a larger sample for evaluation of this likelihood based method because they had complete marker data for all markers described below. All subjects provided informed consent as approved by the appropriate institutional review boards.
Markers genotyped
Detailed marker and genotyping information was described previously [1]. Briefly, two different sets of STR markers were used. First, we used the AmpFLSTR Identifiler PCR Amplification Kit (Applied Biosystems (ABI), Foster City, CA), which provides data from a set of 16 loci useful for forensic purposes (D8S1179, D21S11, D7S820, CSF1PO, D3S1358, TH01, D13S317, D16S539, D2S1338, D19S443, vWA, TPOX, D18S51, D5S818, FGA, and amelogenin). Amelogenin is used for sex identification rather than for polymorphism content, so information from that locus was not included in any analyses. Second, we selected 21 markers known to have high δ between EAs and AAs, and in some cases Hispanic and Asian populations, based on the report of Smith et al. 2001 [27]. This marker panel includes markers D1S196, D1S2628, D2S162, D2S319, D5S407, D5S410, D6S1610, D7S640, D7S657, D8S272, D8S1827, D9S175, D10S197, D10S1786, D11S935, D12S352, D14S68, D15S1002, D16S3017, D17S799, and D22S274. We also genotyped marker FY, added to the 36 STRs because of its known value in identifying individuals of primarily African ancestry.
Measures of marker efficiency
δ was used to measure the marker efficiency. The definition and properties of δ are described in Yang et al. 2005 [1]. Briefly, δ is half the sum of the absolute difference in population frequency over all alleles for each marker between two populations.
Analysis with the likelihood-based method (LBM)
We assumed HWE among alleles for each marker within populations and linkage equilibrium between markers. The likelihood, or the probability of observed genotype profile, for each individual to be in a specific population is calculated as
where X is a vector of genotypes of marker loci, Z is the proposed population of origin, PZ(p11, p12,..., ) is the set of reference allele frequencies p11, p12,..., for the nm alleles of m markers of population Z, and h is a dummy variable for homozygosity (i.e., when the locus is homozygous, h is 1, otherwise h is 0) for each marker locus. When an allele is absent for a given population in the reference frequencies, the corresponding allele frequency in the study group is estimated and used in the calculation of likelihood. An individual is assigned to a population if the maximum likelihood results from assignment to that population among all possible population-specific likelihoods calculated. For assigning individuals into one of two populations A or B, an individual is assigned to population A if the logarithm of likelihood ratio is greater than zero, or otherwise to B, as shown in equation (2).
An individual was considered to be assigned accurately to a group when the greatest likelihood among all the calculated likelihoods assigned an individual the same ethnicity as the self-identified population group of that individual. Assignment accuracy in each population group was defined as the proportion of correctly assigned ethnicities. (The above decision rule is optimal if we have equal priors of proportion for the two ethnic groups. However, when there are more people from one group, a priori, then the prior information needs to be incorporated to improve the overall performance in terms of misclassification rate.) The method was realized in R/S-Plus; the function codes are available upon request from the authors.
The initial set of reference population-specific allele frequencies (training frequencies) for the 36 markers were derived from ABI reference materials [27] or Smith et al. 2001 [28], depending on the source of each marker. Since ABI uses different nomenclature (in some cases) and we redesigned some primers referenced by Smith to facilitate efficient genotyping, each observed allele had to be matched to the corresponding allele for each marker. Alleles at one marker locus (D6S1610) described by Smith et al. 2001 [28] could not be matched accurately to our data from the same marker (however, the value of EA/AAδ that we derived for that marker, 0.336, was similar to the value reported by Smith et al., which was 0.337). The χ2 test was used to compare the allele distributions of the study group and the reference group.
Evaluation of the impact of the training frequencies on population group assignment accuracy
To evaluate the impact of the training frequencies on population group assignment accuracy, we compared the literature-derived training allele frequencies (described above) with training allele frequencies computed from our specific populations. To do so, we randomly split the 282 EAs and 84 AAs into two equal-sized groups. One was treated as the study group, and the other was treated as the reference group, from which the training allele frequencies were estimated.
Authors' contributions
BZY participated in study design, wrote the computer code, carried out the statistical analyses and drafted the manuscript. HZ participated in study design, and conceptual and technical assistance. HRK provided sample recruitment and phenotyping, and commented on the manuscript. JG designed the study, coordinated genotyping efforts and helped draft the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Population allele frequencies in the EA and AA samples for the 35 STRs markers and the FY marker studied. For each marker, the marker name and alleles are listed with the allele frequencies.
Click here for file
Acknowledgements
Greg Dalton-Kay and Ann Marie Lacobelle provided excellent technical assistance. This work was supported in part by funds from NIH: MH14276 (Biological Sciences Training Program support to BZY), the U.S. Department of Veterans Affairs (the VA Medical Research Program [Merit Review to JG], and the VA CT REAP (Research Enhancement Award Program)), NIMH grant K02-MH01387, NIDA grants DA12690, DA12849, and DA12468, NIAAA grants AA11330, AA12870, AA13736, AA03510, and RR06192 (University of Connecticut General Clinical Research Center), and NIGMS grant GM59507.
==== Refs
Yang BZ Zhao H Kranzler H Gelernter J Practical population group assignment with selected informative markers: characteristics and properties of Bayesian clustering via STRUCTURE Genetic Epidemiology 2005 28 302 312 15782414 10.1002/gepi.20070
Devlin B Roeder K Genomic control for association studies Biometrics 1999 55 997 1004 11315092 10.1111/j.0006-341X.1999.00997.x
Pritchard JK Stephens M Donnelly P Inference of population structure using multilocus genotype data Genetics 2000 155 945 959 10835412
Pritchard JK Stephens M Rosenberg NA Donnelly P Association mapping in structured populations Am J Hum Genet 2000 67 170 181 10827107 10.1086/302959
Reich DE Goldstein DB Detecting association in a case-control study while correcting for population stratification Genet Epidemiology 2001 20 4 16 11119293 10.1002/1098-2272(200101)20:1<4::AID-GEPI2>3.0.CO;2-T
Ripatti S Pitkaniemi J Sillanpaa MJ Joint modeling of genetic association and population stratification using latent class models Genet Epidemiology 2001 S409 14 11793709
Satten GA Flanders WD Yang Q Accounting for unmeasured population substructure in case-control studies of genetic association using a novel latent-class model Am J Hum Genet 2001 68 466 477 11170894 10.1086/318195
Sillanpaa MJ Kilpikari R Ripatti S Onkamo P Uimari P Bayesian association mapping for quantitative traits in a mixture of two populations Genetic Epidemiology 2001 S692 9 11793763
Zhang S Zhao H Quantitative similarity-based association test using population samples Am J Hum Genet 2001 69 601 614 11479834 10.1086/323037
Pfaff C Kittles R Shriver MD Adjusting for population structure in admixed populations Genetic Epidemiology 2002 22 196 198 11788964 10.1002/gepi.0126
Zhang S Zhu X Zhao H On a semi-parametric test to detect associations between quantitative traits and candidate genes using unrelated individuals Genetic Epidemiology 2003 24 44 56 12508255 10.1002/gepi.10196
Chen HS Zhu X Zhao H Zhang S Qualitative semi-parametric test to detect genetic association in case-control design under structured population Annals of Human Genetics 2003 67 250 264 12914577 10.1046/j.1469-1809.2003.00036.x
Falush D Stephens M Pritchard JK Inference of population structure using multilocus genotype data: Linked loci and correlated allele frequencies Genetics 2003 164 1567 1587 12930761
Hoggart CJ Parra EJ Shriver MD Bonilla C Kittles RA Clayton DG McKeigue PM Control of confounding of genetic associations in stratified populations Am J Hum Genet 2003 72 1492 1504 12817591 10.1086/375613
Patterson N Hattangadi N Lane B Lohmueller KE Hafler DA Oksenberg JR Hauser SL Smith MW O'Brien SJ Altshuler D Daly MJ Reich D Methods for high-density admixture mapping of disease genes Am J Hum Genet 2004 74 979 1000 15088269 10.1086/420871
Smith MW Patterson N Lautenberger JA Truelove AL McDonald GJ Waliszewska A Kessing BD Malasky MJ Scafe C Le E De Jager PL Mignault AA Yi Z De The G Essex M Sankale JL Moore JH Poku K Phair JP Goedert JJ Vlahov D Williams SM Tishkoff SA Winkler CA De La Vega FM Woodage T Sninsky JJ Hafler DA Altshuler D Gilbert DA O'Brien SJ Reich D A high-density admixture map for disease gene discovery in African Americans Am J Hum Genet 2004 74 1001 13 15088270 10.1086/420856
Tang H Peng J Wang P Risch NJ Estimation of individual admixture: Analytical and study design considerations Genetic Epidemiology 2005 28 289 301 15712363 10.1002/gepi.20064
Paetkau D Calvert W Sterling I Strobeck C Microsatellite analysis of population structure in Canadian polar bears Molecular Ecology 1995 4 347 354 7663752
Paetkau D Waits LP Clarkson PL Craighead L Strobeck C An empirical evaluation of genetic distance statistics using microsatellite data from bear (Ursidae) populations Genetics 1997 147 1943 1957 9409849
Rannala B Mountain JL Detecting immigration by using multilocus genotypes Proc Natl Acad Sci USA 1997 94 9197 9201 9256459 10.1073/pnas.94.17.9197
Cornuet JM Piry S Luikart G Estoup A Solignac M New methods employing multilocus genotypes to select or exclude populations as origins of individuals Genetics 1999 153 1989 2000 10581301
Guinand B Topchy A Page KS Burnham-Curtis MK Punch WF Scribner KT Comparison of likelihood and machine learning methods of individual classification The Journal of Heredity 2002 93 260 269 12407212 10.1093/jhered/93.4.260
Manel S Berthier P Luikart G Detecting wildlife poaching: Identifying the origin of individuals with Bayesian assignment tests and multilocus genotypes Conservation Biology 2002 16 650 659 10.1046/j.1523-1739.2002.00576.x
Piry S Alapetite A Cornuet JM Paetkau D Baudouin L Estoup A GENECLASS2: a software for genetic assignment and first-generation migrant detection J Hered 2004 95 536 9 15475402 10.1093/jhered/esh074
Rosenberg NA Li LM Ward R Pritchard JK Informativeness of genetic markers for inference of ancestry Am J Hum Genet 2003 73 1402 22 14631557 10.1086/380416
Parra EJ Marcini A Akey J Martinson J Batzer MA Cooper R Forrester T Allison DB Deka R Ferrell RE Shriver MD Estimating African American admixture proportions by use of population-specific alleles Am J Hum Genet 1998 63 1839 1851 9837836 10.1086/302148
Website title
Smith MW Lautenberger JA Shin HD Chretien JP Shrestha S Gilbert DA O'Brien SJ Markers for mapping by admixture linkage disequilibrium in African American and Hispanic populations Am J Hum Genet 2001 69 1080 94 11590548 10.1086/323922
|
16225681
|
PMC1285360
|
CC BY
|
2021-01-04 16:30:39
|
no
|
BMC Genet. 2005 Oct 14; 6:50
|
utf-8
|
BMC Genet
| 2,005 |
10.1186/1471-2156-6-50
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1431622569810.1186/1471-2164-6-143Research ArticleEST analysis in Ginkgo biloba: an assessment of conserved developmental regulators and gymnosperm specific genes Brenner Eric D [email protected] Manpreet S [email protected] Dennis W [email protected] Stephen A [email protected] Andrew W [email protected] Walter N [email protected] Richard W [email protected] Suzan J [email protected] Giulia M [email protected] WR [email protected] Gloria M [email protected] The New York Botanical Garden, 200th St. and Kazimiroff, Bronx, NY 10458-5126, USA2 New York University, Department of Biology 1009 Main Building, New York, NY 10003, USA3 Centre for Biotechnology, Tykistökatu 6, FIN-20521 Turku, Finland4 Genome Research Center, Cold Spring Harbor Laboratory, 500 Sunnyside Blvd, Woodbury, NY 11797, USA5 Biology Department, Duke University, Box 91000, Durham, North Carolina, 277086 Department of Plant Biology, Cornell University, Ithaca NY 14850, USA2005 15 10 2005 6 143 143 4 7 2005 15 10 2005 Copyright © 2005 Brenner et al; licensee BioMed Central Ltd.2005Brenner 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
Ginkgo biloba L. is the only surviving member of one of the oldest living seed plant groups with medicinal, spiritual and horticultural importance worldwide. As an evolutionary relic, it displays many characters found in the early, extinct seed plants and extant cycads. To establish a molecular base to understand the evolution of seeds and pollen, we created a cDNA library and EST dataset from the reproductive structures of male (microsporangiate), female (megasporangiate), and vegetative organs (leaves) of Ginkgo biloba.
Results
RNA from newly emerged male and female reproductive organs and immature leaves was used to create three distinct cDNA libraries from which 6,434 ESTs were generated. These 6,434 ESTs from Ginkgo biloba were clustered into 3,830 unigenes. A comparison of our Ginkgo unigene set against the fully annotated genomes of rice and Arabidopsis, and all available ESTs in Genbank revealed that 256 Ginkgo unigenes match only genes among the gymnosperms and non-seed plants – many with multiple matches to genes in non-angiosperm plants. Conversely, another group of unigenes in Gingko had highly significant homology to transcription factors in angiosperms involved in development, including MADS box genes as well as post-transcriptional regulators. Several of the conserved developmental genes found in Ginkgo had top BLAST homology to cycad genes. We also note here the presence of ESTs in G. biloba similar to genes that to date have only been found in gymnosperms and an additional 22 Ginkgo genes common only to genes from cycads.
Conclusion
Our analysis of an EST dataset from G. biloba revealed genes potentially unique to gymnosperms. Many of these genes showed homology to fully sequenced clones from our cycad EST dataset found in common only with gymnosperms. Other Ginkgo ESTs are similar to developmental regulators in higher plants. This work sets the stage for future studies on Ginkgo to better understand seed and pollen evolution, and to resolve the ambiguous phylogenetic relationship of G. biloba among the gymnosperms.
==== Body
Background
Ginkgo biloba is a widely popular tree that is native to China and has been cultivated for well over a millennium. In Asia, G. biloba is used medicinally and its seeds are also a popular cuisine item. In the West, Ginkgo leaf extracts are commonly used for a variety of folk remedies (for review see: [1]) including as a treatment for improving cognitive function [2,3]. Today's Ginkgo biloba is the sole surviving species of an ancient group (Ginkgophytes) of seed plants that may even date from the Permian (approximately 150–200 million years ago) [4]. The genus Ginkgo itself goes back to the Jurassic period – approximately 170 million years ago [5]. Although it is widely believed that the survival of G. biloba depended upon Buddhist monks, who venerated the tree cultivated in their temple grounds, molecular evidence suggests that some stands in China (Wuchuan, Guizhou) are of natural origin representing vestige populations [6]. As a living fossil, Ginkgo biloba has changed little in morphology from its extinct relatives [5]. Along with the Cycadales, Coniferales and Gnetales, the Ginkgoales is one of four orders of non-flowering seed plants (gymnosperms) that form a sister group to the angiosperms (Figure 1).
Figure 1 Gene tree of extant gymnosperms. Ginkgo displays characters suggesting it forms a basal subgroup among the gymnosperms with cycads. Alternately Ginkgo is a sister group of the conifers. Here the placement of Ginkgo is shown as ambivalent between these two scenarios.
Morphological [7,8] and molecular analysis have not yet succeeded in defining the precise phylogenetic hierarchy of the four gymnosperm clades [9]. Ginkgo potentially forms a sister group with the Coniferales (partly due to similar characteristics such as axillary branching and simple leaves). Another model, based on molecular sequence data, places Ginkgo with the Cycadales [10-12]. Interestingly cycads and Ginkgo both share certain plesiomorphic (ancestral) characters found in early fossil seed plants such as haustorial pollen [13,14], which release motile male gametes [15] as well as a large four celled opening in the neck of the archegonia [13,16]. Despite the presence of these and other early seed-plant characteristics, surprisingly little work has been performed on Ginkgo and cycads. Some recent molecular [17] and genomic [18] research on cycads have been conducted and molecular studies of Ginkgo genes have been initiated as well [19-21]. However, no genomic work on Ginkgo biloba has been completed to date.
To begin our genomic treatment of Ginkgo biloba, we focused our initial efforts on developing reproductive and vegetative tissues (Figure 2). Separating Ginkgo male and female structures at an early stage is straightforward because Ginkgo is strictly a dioecious plant (male and female organs on separate individuals). Organ emergence can generally be pinpointed to a specific time of the year in that both reproductive and vegetative tissues are regularly produced in the beginning of May at our collection site in New York. The reproductive structures, megasporangia bearing ovules (Figure 2A–C) (from female trees) and microsporangia bearing pollen (Figure 2D–F) (from male trees), emerge at the apex of short, determinate (spur) shoots. A discreet flush of leaves are also produced in male and female short-shoots (Figure 2A and 2D) [13,16,22]. Long shoots (not shown) exhibit indeterminate growth and yield only vegetative organs. Long-shoots are identifiable by their obvious longer internodes, whereas short shoots (Figure 2A and 2D) have telescoped internodes. Each season, short-shoots might exhibit extensive internode growth and be transformed into long-shoots and vice versa. Consequently, reproductive shoots can become vegetative or vegetative shoots can become reproductive.
Figure 2 Ginkgo male and female short shoots. (A) The fertile female structure (megasporangia) has just emerged from the bud. Two ovules set on a green stalk are visible. Young, unfurled leaves, which have also emerged have extended above the megasporangia. The bracts, which originally enclosed the bud, are now completely opened below the leaves and megasporangia (B) Scanning EM of an ovule, which is completely enclosed by an integument. (C) A longitudinal cross section of the megasporangia reveals the integument enclosing the nucellus. (D) The male reproductive structure is a cluster of microsporangia. In the center of the bud are partly emerged leaves (E) Scanning EM shows two microsporangial lobes containing ripening pollen sacks attached to a stalk. (F) Longitudinal cross section shows a large mucilage containing cavity juxtaposed from the microsporangia filled with immature pollen. C integument; N, nucellus; I, microsporngia; L, mucilage cavity. O, ovule
Until now, little is known regarding the genetic regulation of development in the oldest living seed plants. In order to uncover the genetic controls directing growth and development in Ginkgo biloba, we generated expressed sequence tags (ESTs) from cDNA libraries of very young, recently emerged organs of fertile short-shoots where a large number of regulatory genes are expected to be present. Below is an analysis of these ESTs from Ginkgo biloba. In all three tissues examined, vegetative, microsporangia (male), and megasporangia (female), we found a large number of ESTs with similarity to angiosperm developmental genes. Conversely, a certain number of Ginkgo biloba ESTs were uncovered with homology to genes only found in gymnosperms and non-seed plants, including a set of Ginkgo ESTs that were only common to our cycad EST dataset, further strengthening their classification as gymnosperm specific.
Results and discussion
Construction of a cDNA library from Ginkgo biloba fertile and vegetative tissue
Young organs (Figure 2A and 2D) were collected during the spring from the opening buds of short shoots immediately after their emergence. At this stage, the megasporangium consists of an axis typically bearing two ovules (Figure 2A–C). The ovule is composed of a single integument surrounding a developing nucellus (Figure 2C). The male structure consists of a main axis bearing two or more microsporangia (Figure 2D–F). RNA was extracted from the following organs: megagasporangia, microsporangia and two sets of leaves collected from either male or female trees. mRNA isolated from all four tissues was used to construct four separate cDNA libraries. (Both male and female leaf sequences were pooled during subsequent bioinformatic analysis). Size fractionation was used to enrich for full-length cDNAs during library construction. From this cDNA library, 6,434 sequence reads (Expressed Sequence Tags, ESTs) were generated. All Ginkgo biloba EST reads have been deposited in GenBank. It was determined that 3,739 (58%) of the cDNA clones were over 500 bp long. 3618 of the reads were generated from the 5' end of the cDNA, and 2816 were sequenced from the 3' end. Cluster analysis on the EST sequence produced a unigene set of 3,830 contigs consisting of 2,851 singletons and 979 assemblies. Of the clusteredESTs, the longest contig was 2,172 bp. The entire unigene set or complete Ginkgo BLAST files can be downloaded at the following website [23]. Each G. biloba contig is given a numeric identifier. The constituent ESTs for each contig can be obtained at this website. Additional bioinformatic analysis of the Ginkgo biloba dataset can be accessed at the open Sputnik Comparative Genomics Platform at [24]. This site features sequence annotations, peptide sequence predictions, protein domain architectures and putative molecular markers (ISSRs) for the ginkgo EST derived unigenes. The sequence can be downloaded either as a fasta file, a clustered fasta file or as the derived peptide fasta file. In addition, BLAST analysis can be performed with the clustered ESTs from a given ginkgo organ against all genes in Arabidopsis thaliana or distinct plant clustered EST datasets using the ViCoGenTa program available at the New York Plant Genomics Consortium website [23].
Ginkgo contig matches to genes in angiosperms, gymnosperms and non-seed plants
TBLASTX (expect < 1eX10-5) was used to compare the G. biloba unigenes against all available plant ESTs from TIGR (The Institute for Genomic Research) [25], and the Plant Genome DataBase (Plant GDB) [26]. ESTs from these databases were downloaded and clustered into unigenes, which were used in the comparison. Next the Ginkgo unigene set was compared against the Arabidopsis and rice genome annotated protein sequences downloaded from TIGR. All genes used in this comparison against Ginkgo were divided into one of three taxonomically relevant categories: 1. angiosperms, 2. gymnosperms, and 3. non-seed plants. The angiosperm category encompasses all annotated rice and Arabidopsis genes identified from their respective genomic sequences, as well as all higher plant ESTs. The majority of the gymnosperm ESTs came from the conifer groups pine and spruce but also include ESTs generated from the Plant Genomics Consortium containing ESTs from the two other gymnosperm clades, Cycadales and Gnetales. The non-seed plant category consisted of genes from all remaining plant ESTs including ferns, fern allies, bryophytes and algae available with the majority of the sequences originating from Physcomitrella patens and Chlamydamonas reinhardtii.
A Venn diagram shown in Figure 3A displays the number of Ginkgo contigs, which are shared between one or more of the plant EST datasets at low BLAST stringency value (expect < 1eX10-5). From the Venn diagram, it can be seen that a majority of Ginkgo unigenes (2749/3830) match genes in other plants, and 1081 have no match to other plant genes. Of those 2749 Ginkgo biloba unigenes with matches to other plant genes, a subgroup of 256 unigenes had no corresponding match to genes in the angiosperm dataset. Of these 256 Ginkgo genes that do not match angiosperms, 4 also match genes in non-seed plants.
Figure 3 A Venn diagram illustrating the number of Ginkgo contigs with shared homology to genes found in non-seed plants, gymnosperms and/or angiosperms (A). A BLASTX E value >e 10-5 was used as a cut-off. (B) The Ginkgo contigs with similarity to gymnosperms but (no match to angiosperm genes) were further subgrouped according to their BLAST score homology (E value >e 10-5) within gymnosperm taxa.
The 252 Ginkgo biloba unigenes that only match gymnosperm genes were next partitioned into matches between the three other gymnosperm orders: Cycadales, Coniferales, and Gnetales (Figure 3B). Since there are significantly more conifer unigenes (>60,000) than cycad unigenes (5459) that were used in this comparison, one would expect that the number of matches between Ginkgo and conifers would be significantly greater than matches between Ginkgo and cycads. However the actual number of matches between Ginkgo and conifers (215) is only slightly more then for Ginkgo and cycads (163). In other words despite the fact that there is over 10 times the number of conifer genes then cycad genes used in this comparison, Ginkgo matches to conifers are only 1.3 times greater then matches between Ginkgo and cycads. Of the matches between Ginkgo and gymnosperms, 31 match only cycads, (22 match Cycas rumphii, 6 match only unigenes from the cycad species, Zamia furfuracea, (753 contigs deposited in Genbank from Brenner et al, unpublished data); and 3 Ginkgo contigs match unigenes from both Cycas rumphii and Zamia furfuracea).
As one might expect, unigenes with matches to other plants are somewhat longer then those that have no match to other plants. Of all Ginkgo unigenes with matches to other plants, 89% are greater than 300 bp, whereas 72% are greater than 300 bp then those Ginkgo unigenes with no matches to other plants.
Common genes between cycads and Ginkgo
Our comparative analysis of the Ginkgo EST dataset builds upon our results from a previous genomic study on the cycad, Cycas rumphii [18]. In our current analysis, three (CB090673:GinkgoA3816, CycadCB089620:GingkoA3730, and CycadCB089926:GinkgoA1532) of the fourteen unigenes from cycads previously found only among gymnosperms (after the full-length clone was sequenced), were also homologues to Ginkgo genes found only in gymnosperms. Considering the relatively small number of unigenes from Ginkgo (3,830) and cycads (4706) available for our comparative studies, the detection of the same gene match in Ginkgo and cycads with homology to only gymnosperms strengthens the argument that these genes are gymnosperms specific.
Ginkgo matches to non-seed plants but not angisosperms
An additional four Ginkgo unigenes that are not found in angiosperms were detected in non-seed plants. Three of these Ginkgo unigenes (GinkgoA2411, GinkgoA3214 and GinkgoA325) match non-seed plants and other gymnosperm genes whereas the forth Ginkgo gene (GinkgoA2273) only matches non-seed plants with similarity to a gene in Chlamydomonas.
Classification of G. biloba ESTs by functional categories
Each contig from the Ginkgo dataset was automatically assigned to a functional category (FunCat) based on its top match against the MIPS FunCat list of functionally annotated gene sequences from S. cerevisiae and A. thaliana databases using BLASTP. A non-stringent expect value (E-value) of < e-10 was chosen as the threshold. Table 1 below illustrates the relative fraction that each functional category comprises within the entire unigene set compared to our previous study in Cycas rumphii [18]. The four largest categories of Ginkgo ESTs according to this functional categorization are: "cellular organization" (19%), "metabolism" (11%), "unclassified proteins" (14%), and "protein synthesis" (8%). In general, these same categories are also the highest in the cycad EST library from our previous work, except for "protein synthesis", which appears increased in Ginkgo, whereas interestingly, the category of cell growth, cell division and DNA synthesis is reduced in Ginkgo compared to cycads.
Table 1 Placement of Ginkgo unigenes into functional categories (funcats). Ginkgo genes with a BLASTP expect value (E-value) of > e-10 were assigned into funcats based on their similarity score. The analysis was performed at the Centre for Biotechnology, Turku, Finland. Previous funcat analysis data from Cycas rumphii is shown.
Functional Category Ginkgo% Cycad%
Metabolism 11.0 10.3
Energy 5.0 5.1
Cell Growth, Cell Division and DNA Synthesis 5.4 9.2
Transcription 3.6 5.0
Protein Synthesis 8.4 5.8
Protein Destination 5.8 8.3
Transport Facilitation 2.0 1.3
Intracellular Transport 3.4 4.3
Cellular Biogenesis 3.5 3.3
Cellular Communication/Signal Transduction 3.5 4.2
Cell Rescue, Defense, Cell Death and Ageing 7.2 6.8
Ionic Homeostasis 0.5 0.1
Cellular Organization 18.8 21.6
Classification Not Yet Clear-Cut 7.3 4.6
Unclassified Proteins 14.1 10.3
Ginkgo genes involved in development
Analysis of the Ginkgo biloba dataset revealed a number of ESTs with highest BLAST similarity to genes with known roles in higher plant developmental processes. A sampling of some of these genes is shown below (Table 2). These genes included the Polycomb gene CURLY LEAF [27,28] as well as LATERAL ORGAN BOUNDARIES (LOB) [29], EARLY FLOWERING 5 (ELF5) [30], FLOWERING LOCUS T (FT) [31], and CONSTANS (for review see [32] as well as five ESTs that match MADS box genes, some of which appear identical to previously cloned fragments of MADS genes from Ginkgo biloba including the G. biloba ortholog of AGAMOUS [19]. Other genes in our EST library have homologies to proteins that regulate development through protein turnover including SPA-1 [33]COP1 [34] and COP9 [35,36]. This sampling reveals that the EST dataset of Ginkgo biloba is a rich source of genes encoding proteins with known roles in development at the transcriptional as well as post-transcriptional stage.
Table 2 Similarity match of Ginkgo unigenes to genes involved in development. The G. biloba unigene set was compared to Genbank using a BLASTP cut-off score < e-5. The top match is listed under subject description. The organ(s) from which the listed ESTs were detected are: G = megagametophyte, I = microgametophyte, and L = leaf.
Genes in Ginkgo biloba with similarity to developmental genes in other plants
Contig id. organ BLAST homology match E-value
A2725 G gi|1903019|emb|CAA71599.1| curly leaf [A. thaliana] 4.00E-37
A3095 I gi|15231388|ref|NP_188001.1| LOB domain family protein [A. thaliana] 2.00E-21
A2815 I gi|42541156|gb|AAS19471.1| EARLY FLOWERING 5 [A. thaliana] 1.00E-33
A3351 I gi|4903139|dbj|BAA77836.1| extensive homology to FT (FLOWERING LOCUS T, AB027504) 4.00E-10
A241 I, L gi|41323976|gb|AAS00054.1| CONSTANS-like protein CO1 [Populus deltoides] 1.00E-30
A1591 L gb|AAG43405.1|AF172931_1 homeobox 1 [Picea abies] 5.00E-23
A1591 G gi|14715183|emb|CAC44080.1| putative MADS-domain transcription factor DEFH7 [A. majus] 2.00E-39
A2737 G gi|30230270|gb|AAM76208.1| AGAMOUS-like MADS-box transcription factor [Ginkgo biloba] 5.00E-28
A2850 I gi|25307918|pir||S51935 probable MADS-box protein dal1 – Norway spruce dal1 [Picea abies] 5.00E-72
A629 G, L gi|7446554|pir||T10751 MADS-box protein MADS9 – Monterey pine 6.00E-17
A352 G, I gi|7446559|pir||T09571 MADS box protein MADS2 – Monterey pine 4.00E-75
A914 G, I gi|18401293|ref|NP_565632.1| COP9 / CSN signalosome complex subunit [A. thaliana] 1.00E-45
A2730 G gi|15225760|ref|NP_180854.1| COP1 regulatory protein [A. thaliana] 2.00E-60
A944 G, I gi|30694320|ref|NP_849784.1|argonaute protein(AG01) [A. thaliana] 1.00E-56
Conclusion
The importance of Ginkgo for the study of plant evolution
As the sole remaining species of an ancient genus of plants which has survived nearly 170 million years from the Jurassic [5], Ginkgo biloba is a taxonomic and geographic relict that may be even older because fossils displaying a "ginkgophyte" vegetative morphology have been found as early as the Permian [4]. Ginkgo has a number of plesiomorphic (unspecialized) as well as apomorphic (derived) traits that make it a valuable tool to study the evolution of seed plants. Here we used a genomic approach to investigate the genes involved in regulating development in Ginkgo by creating an EST library from both reproductive and vegetative tissues.
Similar to our previous analysis in Cycas rumphii, our Ginkgo EST study has found significant BLAST homology between Ginkgo ESTs with plant genes in gymnosperms and non-seed plants but not in angiosperms. Since ESTs, even when clustered in contiguous genes, may not represent the complete gene [37], often one will find homology to angiosperm genes when the remaining Ginkgo gene sequence is revealed. For example in the gymnosperm, Pinus taeda EST collection, contigs of increasing length have a higher likelihood then shorter contigs matching a known gene in the Arabidopsis genome [38]. However, in this same study a significant subcategory of very long contigs (>1900 bp) have no homology to Arabidopsis [38]. It is likely that at least some of these long contigs with no match to angiosperm genes represent full length genes that are specific to the gymnosperm and/or seed-less plant clades. Our strategy to address this question involves screening for these same genes in additional taxa of gymnosperms, in the case of this study, Ginkgo biloba. In our analysis three Ginkgo genes that were only found in gymnosperms also matched the fourteen ESTs from our previous study of gymnosperm common cycad genes.
Along these same lines, our results suggest the presence of genes common to non-seed plants and gymnosperms that are not present in angiosperms. This non-seed plant/gymnosperm grouping is not surprising considering the fact that gymnosperms have morphologically common characters that are not found in the angiosperms – particularly in their reproductive structures. For example, the megagametophyte is highly reduced both in cell number and in structural organization in angiosperms when compared to gymnosperms. Although these results cannot say for certain that these genes are specific to non-seed plants and gymnosperms, or more specifically that these genes are found in gymnosperm structures that are not found in seed plant, it nonetheless represents an important starting point to correlate the presence or absence of gymnosperm genes in angiosperms and/or lower plants.
Are cycads and Ginkgo sister taxa?
One result from our study found that the number of Ginkgo contig matches to conifers are only 1.3 times greater then matches between Ginkgo and cycads despite the fact that there is over 10 times the number of conifer genes then cycad genes used in this comparison,. Taken together these results might indicate a closer evolutionary association between Ginkgo and cycads then between Ginkgo and conifers. This bias towards cycad/Ginkgo similarity correlates with the fact that the majority of molecular phylogenetic studies place as the cycads sister group to Ginkgo. Hopefully, this preliminary data will encourage further phylogenomic studies to fully resolve the hierarchy among extant gymnosperm orders. Until the full genome sequence becomes available for key gymnosperm taxa, EST sampling provides an important initial step for large scale identification of molecular markers to generate robust phylogenetic trees.
Developmental regulators in Ginkgo
In Ginkgo biloba we note here a variety of genes with similarity to developmental regulators in angiosperms. We also note below that homologues to some of these developmental regulators are also present in our Cycas rumphii library as either orthologs to those found in higher plants or at least, belonging to the same gene family. An EST from Ginkgo biloba that was detected in the megagametophyte library has high similarity to the Arabidopsis CURLY LEAF (CLC) gene, which belongs to the Polycomb-group proteins (PcGs). PcGs epigenetically regulate downstream target genes [28]. PcGs modify chromatin-protein complexes that repress homeotic gene transcription and influence cell proliferation. In Arabidopsis PcG genes have been shown to regulate MADS box genes [39]. The CLC protein product regulates the expression of AGAMOUS [27], a gene controlling floral organ identity [40]. Interestingly, an ortholog for angiosperm AGAMOUS was also detected in the Ginkgo megagametophyte library (Table 2). Ginkgo AGAMOUS, (previously named GBM5) was identified in a study where the MADS domains were examined in Ginkgo [19]. In this work Ginkgo AGAMOUS was shown via RT-PCR to be expressed in not only female but also in male and vegetative tissue. In our analysis, five total MADS box homologues were also detected in the Ginkgo EST dataset. Three of the Ginkgo ESTs from our library, GinkgoA2340, GinkgoA2730 and GinkgoA2850, are perfectly identical to the MADS domain gene fragments previously cloned by [19] as degenerate PCR products. The other two unigenes from our dataset have homologies to MADS genes (GinkgoA629 and GinkgoA352), but do not specifically match any of the PCR fragments isolated in their study. These two MADS box unigenes either do not include the small region amplified in their degenerate PCR screen or could alternative be unique MADS genes not isolated in their study. Unlike the degenerate primer approach used to isolate MADS genes, our EST approach offers the additional advantage of cloning entire genes or at least substantially large gene fragments. Among the few developmental genes examined in gymnosperms, most attention has focused on the expression of MADS homologs [41,42].
Other developmental genes found in the Ginkgo EST library include those with homology to regulators of flowering such as EARLY FLOWERING 5 (ELF5), which controls the levels of the gene FLC, which itself is a central regulator of flowering [30]. Another Ginkgo EST includes FLOWERING LOCUS T (FT), which belongs to a small family of genes (FT/TFL1) that act to promote flowering as a downstream component from CONSTANS [43]. CONSTANS is a transcription factor that has a critical role integrating circadian rhythms and light signals (for review see [32]). As one would expect an EST homolog for the CONSTANS gene family was found in Ginkgo. CONSTANS belongs to a large gene family, which may have redundant roles in plants [44]. Not surprisingly, we also found homologs to CONSTANS in our previous study on cycad leaf ESTs [18]. In that flowering plants are believed to have evolved from gymnosperms, a survey of CONSTANS, ELF, and FT in gymnosperms, particularly in very young reproductive tissue might help define the origins of reproductive induction in non-flowering plants. Among the other genes related to developmental regulators includes a homologue to LATEROL ORGAN BOUNDARIES (LOB) domain gene family which in Arabidopsis has over 40 members [29]. The molecular mechanism of LOB domain containing genes is unknown, but one gene in Arabidopsis, ASYMMETRIC LEAVES2, is required for normal leaf development, by potentially acting as a regulatory repressor of KNOX genes [45]. A KNOX homolog is also present in our EST library and was found in male reproductive tissues and HOX genes were also detected in our previous analysis in C. rumphii
Another important component regulating development occurs at the level of protein degradation. A gene recognized in our EST library includes COP1. COP1, serves as an E3 ubiquitin targeting photomorphogenic factors such as HY5 for degradation [33]. Another Ginkgo EST from the library has highest similarity to COP9. In our previous EST analysis in Cycas rumphii an EST was also isolated with similarity to COP9 [18]. COP9 is a subunit of the COP9 signalosome complex that controls multiple signaling pathways that regulate development in all eukaryotes [35,36]. In Arabidopsis, the cop9 and cop1 mutants are constitutively photomorphogenic in dark grown seedlings [46]. Unlike angiosperms, seedlings from conifers are constitutively photomorphogenic when grown in the dark [47,48]. In Ginkgo, chlorophyll and chloroplast development is completely dependent on light, however this process proceeds at a markedly slower pace then in flowering plants. That is, photomorphogenic development in Ginkgo seedlings is strongly delayed after transfer from dark grown conditions to light grown conditions when compared to seed plants [20,21]. The dark grown phenotype of cycads is unreported. Considering this variability in photomorphic development among and between the gymnosperms and the angiosperms, the discovery of genes encoding photomorphogenic regulators in gymnosperms will help understand the evolution of photomorphogenesis in seed plants.
Taken together, our genomics analysis of Ginkgo biloba is an important additional step to analyze the role of molecular development of early seed plants. Thus the stage is set to further determine the role of these genes during the development of ancillary structures found between Ginkgo, cycads and other gymnosperms with higher plants as well as the role of those in structures that are unique to gymnosperms and/or the non-seed plants as a step to understand the evolution of the seed plant habit.
Methods
Tissue collection and library construction and DNA purification
Newly emerged microsporangia from accession 76163B, megasporangia from accession 76163D, and immature leaves from both accessions were collected from newly opened buds of Ginkgo biloba growing in the New York Botanical Garden outdoor collection on April 12, 2002. Organs were snap frozen in liquid nitrogen. RNA was collected from each organ and a cDNA library was constructed from fractionated cDNA according to [49].
Microscopy
Ginkgo apices were collected on April 19. Bract tissues were removed from the apex leaving the leaves and reproductive tissue, which was fixed in FAA (50% ethanol, 5% glacial acetic Acid, 3.7% formaldehyde) under vacuum (20 In. Hg) at room temperature. Fresh FAA was vacuum infiltrated two additional times. Tissue was stored in 70% ethanol at 4°C.
For histology, tissue was prepared by sequential (overnight 4 C incubation at each alcohol grade) dehydration in 80, 90, 95 and finally 100% ethanol plus Eosin Y (National Medicinal Products) followed by two treatments in 100% ethanol for 2 hours at room temperature. The tissue was next placed in a 1:1 solution of ethanol and toluene, then twice in toluene alone, each time for 2 hours at room temperature. The tissue was then placed in toluene with a quarter-volume of paraffin (PARAPLAST X-TRA® (Fisher)) chips at 60°C overnight. The tissue was then embedded in melted paraffin with six wax changes over the course of three days at 55°C. Apices were sectioned on a MICROM HM 355 microtome. 8 μm thick sections were taken using a blade angle of 9°. The tissue was stained with Astra Blue and Safranin. After mounting on slides, sections were imaged using a Nikon DXM1200F digital microscope camera.
For scanning electron microscopy, fixed materials were dissected, dehydrated in ethanol and critical point dried in a Denton critical point dryer. Dried materials were affixed to aluminium E. M. stubs and coated with between 80–240 A of palladium in a Hummer II Sputter Coater. Coated materials were then observed using a Jeol scanning electron microscope at 15 or 20 kV. Images were digitally recorded and evaluated using Adobe Photoshop 9.0.
EST sequencing and gene analysis
Plasmid DNA was collected as described in the manual (Stratagene), catalogue number 200450 in the in vivo mass excision section. Sequence analysis was performed at CSHL using an ABI 3700 Capillary sequencer for separation and nucleotide detection. Reactions were performed using a 1/16 Big Dye Terminator. Sequencing was performed with either the -21 M13 forward and/or reverse primer. ESTs were assembled using Phrap [50,51] and clustered into contigs using the CAP3 program [52]
Peptide extraction
Peptide sequences were derived for all unigenes using the ESTScan application [53] run with the default parameters. Prior to the ESTScan predictions, a Ginkgo species-specific ESTScan model was created. ESTScan was trained with Ginkgo ORFs identified from the best match of BLASTX analyses performed on the unigene sequence against the Swissprot protein database. All BLASTX matches were filtered using the arbitrary expectation value of 1e-10.
Sequence annotation
Sequence annotation on each of the Ginkgo cluster consensus sequences and derived peptides were performed within the openSputnik application [54]. Results were assessed for possible contamination by searching for homology to the E. coli and human genomes and were scored for homology to a wide range of non-coding RNAs and plant chloroplast and mitochondrial genomes. Homology searches were performed using the BLAST application [55] and results were filtered using the expectation value < 1e-10. Functional assignment was performed on both cluster consensus sequence and the peptide sequence. Assignments were made using BLASTX and BLASTP respectively against the MIPS catalogue of functionally assigned proteins (funcat) [50,51], tentative functional assignments were filtered using the expectation value < 1e-10.
Categorization of Ginkgo contigs
All Ginkgo contigs sequences were aligned against a PlantEST database using TBLASTX [55] and BLASTX against the NR (aa) database. The PlantEST database was created by downloading all plant ESTs in GenBank and assembling them using Phrap [50,51]. Todd Wood from Clemson University provided the PERL script that creates the PlantEST databases as described above. The NR (aa) database is a non-redundant database of protein sequences from GenBank.
Determination of gymnosperm specific genes
All available plant ESTs were downloaded from GenBank and separated into three datasets consisting of angiosperms (monocots and dicots), gymnosperms, or non-seed plants (ferns, mosses and algae). Downloaded ESTs were assembled using Phrap [50,51]. All matches with an expect value < 1e 10-5 are considered significant.
Authors' contributions
EB conceived of this project. He participated in its design, experiments and drafted the manuscript. DS and GC also conceived this project and participated in its design and coordination. MK played the major role in the bioinformatics analysis. SAR performed the funcat analysis and built the Sputnik website, AD performed the scanning electron microscopy work, WM and GS performed the histological sectioning, RT and SJR performed the cDNA library construction, RM facilitated the EST sequencing. All authors read and approved the final manuscript.
Acknowledgements
We would like to thank Vivekanand Balija for sequence generation. Funding for this work comes from the Plant Genomics Consortium. The plant Genomics Consortium is made possible by the generosity of the Altria Group, Inc., The Mary Flagler Cary Charitable Trust, The Eppley Foundation for Research, Inc., The Ambrose Monell Foundation, The Wallace Genetic Foundation, Inc., and the National Science Foundation Plant Genome Group grant number DBI-0421604.
==== Refs
Hori T Ridge RW Tulecke W Del Tredici P Tremouillaux-Guiller J Tobe H Ginkgo biloba-A Global Treasure 1997 Tokyo, Springer
Curtis-Prior P Vere D Fray P Therapeutic value of Ginkgo biloba in reducing symptoms of decline in mental function J Pharm Pharmacol 1999 51 535 541 10411212 10.1211/0022357991772817
Gold PE Cahill L Wenk GL The lowdown on Ginkgo biloba Sci Am 2003 288 86 91 12661320
Rothwell GW Holt B Hori T, Ridge RW, Tulecke W, Del Tredici P, Trémouillaux-Guiller J and Tobe H Fossils and phenology in the evolution of Ginkgo biloba Ginkgo-biloba-A Global Treasure 1997 Tokyo, Springer-Verlag 223 230
Zhou Z Zheng S The missing link in Ginkgo evolution Nature 2003 423 821 822 12815417 10.1038/423821a
Fagard M Boutet S Morel JB Bellini C Vaucheret H AGO1, QDE-2, and RDE-1 are related proteins required for post-transcriptional gene silencing in plants, quelling in fungi, and RNA interference in animals Proc Natl Acad Sci USA 2000 97 11650 11654 11016954 10.1073/pnas.200217597
Doyle JA Donoghue MJ Seed plant phylogeny and the origin of angiosperms: an experimental cladistic approach. Botanical Reviews 1986 52 321 431
Donoghue MJ Doyle JA Seed plant phylogeny: Demise of the anthophyte hypothesis? Curr Biol 2000 10 R106 9 10679315 10.1016/S0960-9822(00)00304-3
Magollón S Sanderson MJ Relationships among seed plants inferred from highly conserved genes: sorting confilicting phylogenetic signals among ancient lineages Am J Bot 2002 89 1991 2006
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 USA 2000 97 4086 4091 10760277 10.1073/pnas.97.8.4086
Bowe LM Coat G dePamphilis CW Phylogeny of seed plants based on all three genomic compartments: extant gymnosperms are monophyletic and Gnetales' closest relatives are conifers Proc Natl Acad Sci USA 2000 97 4092 4097 10760278 10.1073/pnas.97.8.4092
Hasebe M Hori T, Ridge RW, Tulecke W, Del Tredici P, Trémouillaux-Guiller J and Tobe H Molecular phylogeny of Ginkgo biloba: close relationship between Ginkgo biloba and cycads Ginkgo biloba-a Global Treasure 1997 Tokyo, Springer-Verlag 173 181
Chamberlain CJ Gymnosperms. Structure and Evolution 1935 Chicago, University of Chicaco Press
Friedman WE Morphogenesis and experimental aspects of growth and development of the male gametophyte of Ginkgo biloba in vitro American Journal of Botany 1987 74 1816 1830
Ikeno S Hirase S Spermatozoids in gymnosperms Ann of Bot 1897 11 344 345
Foster AS Gifford EM Kennedy D and Park RB Comparative Morphology of Vascular Plants 1974 second edition San Francisco, W. H. Freeman and Company
Zhang P Tan HT Pwee KH Kumar PP Conservation of class C function of floral organ development during 300 million years of evolution from gymnosperms to angiosperms Plant J 2004 37 566 577 14756763 10.1046/j.1365-313X.2003.01983.x
Brenner ED Stevenson DW McCombie RW Katari MS Rudd SA Mayer KF Palenchar PM Runko SJ Twigg RW Dai G Martienssen RA Benfey PN Coruzzi GM Expressed sequence tag analysis in Cycas, the most primitive living seed plant Genome Biol 2003 4 R78 14659015 10.1186/gb-2003-4-12-r78
Jager M Hassanin A Manuel M Guyader HL Deutsch J MADS-box genes in Ginkgo biloba and the evolution of the AGAMOUS Family Mol Biol Evol 2003 20 842 854 12679535 10.1093/molbev/msg089
Chinn E Silverthorne J Light-dependent chloroplast development and expression of a light-harvesting chlorophyll a/b-binding protein gene in the gymnosperm Ginkgo biloba Plant Physiol 1993 103 727 732 8022934 10.1104/pp.103.3.727
Chinn E Silverthorne J Hohtola A Light-regulated and organ-specific expression of types 1, 2, and 3 light-harvesting complex b mRNAs in Ginkgo biloba Plant Physiol 1995 107 593 602 7724674 10.1104/pp.107.2.593
Bierhorst DW Morphology of Vascular Plants 1971 , Macmillan
The New York Plant Genomics Consortium
OpenSputnik Comparative Genomics Platform
The Institute for Genomic Research
Plant Genomic Database
Goodrich J Puangsomlee P Martin M Long D Meyerowitz EM Coupland G A Polycomb-group gene regulates homeotic gene expression in Arabidopsis Nature 1997 386 44 51 9052779 10.1038/386044a0
Kohler C Grossniklaus U Epigenetic inheritance of expression states in plant development: the role of Polycomb group proteins Curr Opin Cell Biol 2002 14 773 779 12473353 10.1016/S0955-0674(02)00394-0
Shuai B Reynaga-Pena CG Springer PS The lateral organ boundaries gene defines a novel, plant-specific gene family Plant Physiol 2002 129 747 761 12068116 10.1104/pp.010926
Noh YS Bizzell CM Noh B Schomburg FM Amasino RM EARLY FLOWERING 5 acts as a floral repressor in Arabidopsis Plant J 2004 38 664 672 15125772 10.1111/j.1365-313X.2004.02072.x
Weigel D Ahn JH Blazquez MA Borevitz JO Christensen SK Fankhauser C Ferrandiz C Kardailsky I Malancharuvil EJ Neff MM Nguyen JT Sato S Wang ZY Xia Y Dixon RA Harrison MJ Lamb CJ Yanofsky MF Chory J Activation tagging in Arabidopsis Plant Physiol 2000 122 1003 1013 10759496 10.1104/pp.122.4.1003
Hayama R Coupland G The molecular basis of diversity in the photoperiodic flowering responses of Arabidopsis and rice Plant Physiol 2004 135 677 684 15208414 10.1104/pp.104.042614
Saijo Y Sullivan JA Wang H Yang J Shen Y Rubio V Ma L Hoecker U Deng XW The COP1-SPA1 interaction defines a critical step in phytochrome A-mediated regulation of HY5 activity Genes Dev 2003 17 2642 2647 14597662 10.1101/gad.1122903
Seo HS Yang JY Ishikawa M Bolle C Ballesteros ML Chua NH LAF1 ubiquitination by COP1 controls photomorphogenesis and is stimulated by SPA1 Nature 2003 423 995 999 12827204 10.1038/nature01696
Chamovitz DA Glickman M The COP9 signalosome Curr Biol 2002 12 R232 11937035 10.1016/S0960-9822(02)00775-3
Hellmann H Estelle M Plant development: regulation by protein degradation Science 2002 297 793 797 12161644 10.1126/science.1072831
Rudd S Expressed sequence tags: alternative or complement to whole genome sequences? Trends Plant Sci 2003 8 321 329 12878016 10.1016/S1360-1385(03)00131-6
Kirst M Johnson AF Baucom C Ulrich E Hubbard K Staggs R Paule C Retzel E Whetten R Sederoff R Apparent homology of expressed genes from wood-forming tissues of loblolly pine (Pinus taeda L.) with Arabidopsis thaliana Proc Natl Acad Sci U S A 2003 100 7383 7388 12771380 10.1073/pnas.1132171100
Katz A Oliva M Mosquna A Hakim O Ohad N FIE and CURLY LEAF polycomb proteins interact in the regulation of homeobox gene expression during sporophyte development Plant J 2004 37 707 719 14871310 10.1111/j.1365-313X.2003.01996.x
Bowman JL Smyth DR Meyerowitz EM Genetic interactions among floral homeotic genes of Arabidopsis Development 1991 112 1 20 1685111
Brenner ED Stevenson DW Williams CG Genomic Approaches to Understand the Origins of Seeds Landscapes, Genomics and Transgenic Conifers 2005 Amsterdam, Publisher: Kluwer-Springer
Theissen G Becker A Di Rosa A Kanno A Kim JT Munster T Winter KU Saedler H A short history of MADS-box genes in plants Plant Mol Biol 2000 42 115 149 10688133 10.1023/A:1006332105728
Kardailsky I Shukla VK Ahn JH Dagenais N Christensen SK Nguyen JT Chory J Harrison MJ Weigel D Activation tagging of the floral inducer FT Science 1999 286 1962 1965 10583961 10.1126/science.286.5446.1962
Lagercrantz U Axelsson T Rapid evolution of the family of CONSTANS LIKE genes in plants Mol Biol Evol 2000 17 1499 1507 11018156
Lin WC Shuai B Springer PS The Arabidopsis LATERAL ORGAN BOUNDARIES-domain gene ASYMMETRIC LEAVES2 functions in the repression of KNOX gene expression and in adaxial-abaxial patterning Plant Cell 2003 15 2241 2252 14508003 10.1105/tpc.014969
Suzuki G Yanagawa Y Kwok SF Matsui M Deng XW Arabidopsis COP10 is a ubiquitin-conjugating enzyme variant that acts together with COP1 and the COP9 signalosome in repressing photomorphogenesis Genes Dev 2002 16 554 559 11877375 10.1101/gad.964602
Bogdanovic M Chlorophyll formation in the dark Physiol Plant 1973 29 17 18
Peer W Silverthorne J Peters JL Developmental and light-regulated expression of individual members of the light-harvesting complex b gene family in Pinus palustris Plant Physiol 1996 111 627 634 8787030 10.1104/pp.111.2.627
Brenner ED Stevenson DW Twigg RW Cycads: evolutionary innovations and the role of plant-derived neurotoxins. Trends Plant Sci 2003 8 446 452 13678912 10.1016/S1360-1385(03)00190-0
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
Huang X Madan A CAP3: A DNA sequence assembly program Genome Res 1999 9 868 877 10508846 10.1101/gr.9.9.868
Iseli C Jongeneel CV Bucher P ESTScan: a program for detecting, evaluating, and reconstructing potential coding regions in EST sequences Proc Int Conf Intell Syst Mol Biol 1999 138 148 10786296
Rudd S openSputnik--a database to ESTablish comparative plant genomics using unsaturated sequence collections Nucleic Acids Res 2005 33 (Database Issue) D622 7 15608275
Altschul SF Gish W Miller W Myers EW Lipman DJ Basic local alignment search tool J Mol Biol 1990 215 403 410 2231712 10.1006/jmbi.1990.9999
|
16225698
|
PMC1285361
|
CC BY
|
2021-01-04 16:32:47
|
no
|
BMC Genomics. 2005 Oct 15; 6:143
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-143
|
oa_comm
|
==== Front
BMC Med EducBMC Medical Education1472-6920BioMed Central London 1472-6920-5-361623231010.1186/1472-6920-5-36Research ArticleMistreatment of university students most common during medical studies Rautio Arja [email protected] Vappu [email protected] Matti [email protected] Marja [email protected] Department of Pharmacology and Toxicology, University of Oulu, FIN-90014 University of Oulu, Finland2 Department of Educational Sciences and Teacher Education, University of Oulu, FIN-90014 University of Oulu, Finland3 Department of Paediatrics, University of Oulu, FIN-90014 University of Oulu, Finland2005 18 10 2005 5 36 36 22 6 2005 18 10 2005 Copyright © 2005 Rautio et al; licensee BioMed Central Ltd.2005Rautio 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
This study concerns the occurrence of various forms of mistreatment by staff and fellow students experienced by students in the Faculty of Medicine and the other four faculties of the University of Oulu, Finland.
Methods
A questionnaire with 51 questions on various forms of physical and psychological mistreatment was distributed to 665 students (451 females) after lectures or examinations and filled in and returned. The results were analysed by gender and faculty. The differences between the males and females were assessed statistically using a test for the equality of two proportions. An exact two-sided P value was calculated using a mid-P approach to Fisher's exact test (the null hypothesis being that there is no difference between the two proportions).
Results
About half of the students answering the questionnaire had experienced some form of mistreatment by staff during their university studies, most commonly humiliation and contempt (40%), negative or disparaging remarks (34%), yelling and shouting (23%), sexual harassment and other forms of gender-based mistreatment (17%) and tasks assigned as punishment (13%). The students in the Faculty of Medicine reported every form of mistreatment more commonly than those in the Faculties of Humanities, Education, Science and Technology. Experiences of mistreatment varied, but clear messages regarding its patterns were to be found in each faculty. Female students reported more instances of mistreatment than males and were more disturbed by them. Professors, lecturers and other staff in particular mistreated female students more than they mistreated males. About half of the respondents reported some form of mistreatment by their fellow students.
Conclusion
Students in the Faculty of Medicine reported the greatest amount of mistreatment. If a faculty mistreats its students, its success in the main tasks of universities, research, teaching and learning, will be threatened. The results challenge university teachers, especially in faculties of medicine, to evaluate their ability to create a safe environment conducive to learning.
==== Body
Background
The main tasks of universities are research, teaching and learning. The teaching atmosphere during undergraduate studies is important not only for learning but also for building positive attitudes towards one's professional identity and towards life-long learning. Attitudes, positive or negative, adopted during university studies will have an impact on the values and behaviour of students in their future working lives.
Various forms of mistreatment have been reported to occur in a variety of workplaces, including schools [1], universities [2-5] and the police force [6]. Mistreatment is a problem on a personal level and on the organisational and societal levels as well. In some cases mistreatment can even lead to alcohol and drug abuse [2,7]. Exposure to mistreatment has a significant inverse correlation with both job satisfaction and psychological health and well being [8].
Mistreatment is perceived by undergraduate students in the United States as a major source of stress [9], and such perceptions and their consequences are more prevalent among medical students than either students or faculty staff believe [3,4,10,11]. More than a third of the students at medical school have considered dropping out, and one fourth report that they would have chosen a different profession had they known in advance about the extent of the mistreatment they would experience in American medical schools [12]. Corresponding results have been obtained in Finland [13]. Also, more generally, high proportions of students who experience mistreatment suffer measurable psychopathological consequences [3,7]. Perceived mistreatment has been found to be a major source of stress during medical internship [9], and especially when this is consistent and systematic, it may significantly impair mental health and well-being among both university students and employees and affect their overall satisfaction with their work [2,4,7].
Becher [14] found in the UK and US that mistreatment in the cultures of different university disciplines can vary. Disciplines and departments differ both at the level of epistemic issues and in the quality of social relations and atmospheres and their ways of controlling and punishing students. Becher thinks that it is the moral order that defines the basic beliefs, values, norms and aspirations prevailing in each disciplinary culture. This forms the background ethos for each discipline, determining what is regarded as normal and ordinary and what is impossible, imaginary or extraordinary.
Objectives
A report from two medical schools in Finland in the early 1990s showed that three out of every four students had experienced some kind of mistreatment by classmates, teachers, hospital staff or patients during their education [11]. The present study was undertaken to evaluate the prevalence of physical and psychological mistreatment among students in all faculties of Oulu University, including the Faculty of Medicine. One special purpose was to see whether there were any differences in the treatment of students between the Faculty of Medicine and the other faculties and whether it would be meaningful to discuss the characteristics in terms of a moral order.
Methods
Study design
Permission to perform this survey was obtained from the rector of the university and from the chief administrator and chief academic officer of each faculty. The protocol was accepted by the university ethics committee. The work was carried out mainly within a 3-week period.
After briefly explaining the survey and its purpose to the students, we distributed the questionnaire forms (see Additional file 1) to them after a lecture or an examination and continued to be present while they filled them in. The students were not allowed to discuss the questionnaire, but were told that they were to give their own personal, honest answer to each question anonymously.
Survey questionnaire
The questionnaire was modified from that used in 1991 to evaluate two medical schools in Finland [11], which had in turn been based on that of Sheehan et al. [12] and Baldwin et al. [15]. In order to keep the questionnaire valid and to be able to compare the present results with those obtained earlier using a similar questionnaire, we kept modification of the questionnaire to an absolute minimum. Since it had originally been used only among students in medical faculties, we modified it to be applicable to students in all faculties by changing few phrasings, e.g. we did not specifically ask about mistreatment by nurses.
The first 13 questions (out of the total of 51) covered the student's background, i.e. faculty, age, gender, native language, marital status, religion, years of study and curriculum, socio-economic status and level of education of the person's father and mother. These were followed by 36 structured and 2 open-ended questions (see Additional file 1) covering different types of physical and psychological mistreatment such as sexual harassment and discrimination, verbal and psychological mistreatment and physical threats (Table 1). Each staff group was listed separately under each question: "How often, if ever, have any of the following persons mistreated you (each type of mistreatment was asked separately)?" and the following options were given: "never", "rarely (1–2 times)", "sometimes (3–4 times)" and "often (5 times or more)". Each item also had space for a written answer and an opportunity to give an example of the mistreatment. If the respondent reported mistreatment, he/she also answered the question: "How much did this mistreatment bother you?" In addition to personal perceptions of mistreatment, we also attempted to evaluate its general occurrence in the university by asking: "How often does each type of mistreatment occur at your university?" The same options were given: never, rarely, sometimes and often.
Table 1 Topics addressed in the questionnaire (see Additional file 1).
Number of questions Topics addressed
1–13 Student background
14–31 Mistreatment and harassment
32–35 Sexual harassment or mistreatment
36–39 Racial, ethnic, religious or age discrimination
40–42 Threats to fail or give a low grade
43–46 Negative or disparaging remarks on study performance
47–49 Sleep deprivation
50 Immoral, unethical or other unacceptable treatment during studies (open-ended question)
51 Other forms of mistreatment (open-ended question)
Students
The main target groups were first and second-year students and those who had been studying for four years or more, to investigate the occurrence of mistreatment in relation to the number of years of study. Altogether 665 students participated in the survey, representing 7% of the total at the university. The proportion varied according to faculty, being 11.5% in the Humanities, 6.1% in Education, 6.6% in Science, 18.9% in Medicine and 3.5% in Technology. The sample size for each faculty was designed to be sufficiently large that no single student or teacher could be identified in the analyses. The students who had been studying for more than three years in the Faculties of Education and Technology were doing their practical training period outside the university and could not be reached. Only a few students refused to fill in the questionnaire, and those who returned it had answered all the questions. The exact figures according to faculty, gender and study year, and the proportion (%) of female students in each faculty and among the respondents are given in Tables 2 and 3. The median number of years of study was two for the males and almost three for the females, and the median age was roughly the same for both sexes, between 22 and 23 years. All the participants were Finnish, and 90% of them were members of the Evangelical-Lutheran church. The female students were more often married (43%) than the male ones (34%).
Table 2 Percentages of female students in the faculties and among the students participating in the survey (n = 662).
Faculties Percentage of women
In the faculty % In the survey %
Humanities 75.6 77.8
Education 77.2 88.0
Medicine 67.4 80.2
Science 50.5 75.5
Technology 14.8 7.3
Table 3 Numbers of students participating in the survey, by faculty and study year (n = 634; 18 females and 10 males did not report the study year).
Students by years of study Number Total Number (%)
Year 1 2 3 4 >4
Sex* F M F M F M F M F M F M
Faculties:
Humanities 23 3 16 8 17 7 15 9 55 9 126 (19.9) 36 (10.4)
Education 12 0 22 2 10 5 14 0 8 2 66 (10.4) 9 (1.4)
Medicine 1 0 35 9 16 5 23 5 39 9 114 (18.0) 28 (6.2)
Science 19 1 29 17 34 9 13 9 25 3 120 (18.9) 39 (6.2)
Technology 0 11 4 50 2 11 0 10 1 7 7 (1.1) 89(14.0)
Total 55 15 106 86 79 37 65 33 128 30 433 (68.3) 201 (31.7)
*F = female, M = male
Statistical procedures
The data were analysed using SPSS (Statistical Package for Social Sciences, version 7.0) and the differences between the males and females were assessed using a test for the equality of two proportions [16] in the Arcus Quickstat Biomedical software (Research Solutions). An exact two-sided P value was calculated using a mid-P approach to Fisher's exact test (the null hypothesis being that there is no difference between the two proportions) [16].
Results
Mistreatment by staff
Our results showed that mistreatment is common in the university, since 40% of the men and 55% of the women had experienced some mistreatment by staff or faculty members during their university studies (Table 4). Females more commonly reported mistreatment than males (p < 0.0005), and were more disturbed by it. Twenty-one percent of students reported at least one instance of mistreatment, and 12.6% reported having experienced four or more different types of mistreatment (Table 4). The most common form was belittlement and humiliation (40%) (Figure 1), followed by negative or disparaging remarks about the respondent's academic performance (34%), yelling and shouting (23%), sexual harassment and other forms of gender-based mistreatment (17%) and tasks assigned as punishment (13%). Research fellows and senior research fellows and lecturers were most often reported as the sources of this mistreatment. (Table 5). Professors, lecturers and other (non-academic) staff mistreated female students significantly more frequently than males (Table 5).
Table 4 Total numbers of episodes of mistreatment by staff (range 0–10) and fellow students (range 0–8) reported by students during their university studies.
Frequency Staff Fellow students
Male
(N = 193) % Female
(N = 378) % All
(N = 571) % Male
(N = 205) % Female
(N = 426) % All
(N = 631) %
Never 60.1 44.7 49.9 54.1 48.8 50.6
Once 21.2 20.9 21.0 24.9 26.3 25.8
2–3 times 9.3 20.1 16.5 14.2 19.7 17.9
4–5 times 5.1 7.7 6.9 3.9 4.5 4.3
6 or more times 4.3 6.6 5.7 2.9 0.7 1.4
The data were obtained by counting how many types of mistreatment a single student reported having experienced. We recognised ten types of mistreatment by staff (questions 14,17,20,23, 26,29,32,36,40,43) and eight types of mistreatment by fellow students (questions 14,17,23,26, 29,32,36,43).
Figure 1 Percentages (%) of female and male students who reported given types of mistreatment by staff (n = 647 – 652).
Table 5 Frequency (%) of all types of mistreatment by given categories of staff experienced by students.
Staff categories Male % Female % All %
Professors 12.4 18.6* 16.6
Associate professors 13.4 13.7 13.6
Research/senior research fellows 26.2 28.6 27.7
Lecturers 16.9 33.3** 27.9
Other staff 8.7 15.5* 13.2
Differences between female and male students, *p < 0.05; **p < 0.0001.
Belittlement and humiliation were the most common forms of student mistreatment for every staff group. The second most common among the professors was sexual harassment or gender-based mistreatment, together with negative or disparaging remarks. Among the lecturers it was sexual harassment or gender-based mistreatment together with assignments given as a punishment (Figure 2). Research fellows and senior research fellows were reported to shout and yell at students and to assign tasks as punishment. Shouting and yelling were the second most common form of mistreatment by other groups of staff.
Figure 2 Occurrence and frequency of different types of mistreatment from professors and lecturers as reported by students. Rarely (1–2 times), sometimes (3–4) and often (5 or more times).
Mistreatment by fellow students
51.2% of the female students and 45.9% of the males reported having experienced mistreatment from fellow students at least once (Table 4). 24.5% of the females and 19.2% of the males reported contempt and humiliation, and derogatory remarks concerning the career chosen by the informant were common, as also was students taking credit for someone else's work (Table 6). The results showed that the students did not appreciate the fields of study pursued in other faculties, this being especially evident in the answers given by the students of the humanities and technology concerning each other's fields of study.
Table 6 Occurrence (%) of mistreatment by staff and fellow students during the 2nd year and during or after the 4th year, as reported by male and female students.
2nd year % ≥4th year % All %
Staff Fellow student Staff Fellow student Staff Fellow student
Number of answer
(M = male, F = female) M 84 – 86 F 102 – 104 M 85 – 86 F 102 – 104 M 61 – 62 F 184 – 189 M 62 F 187 – 191 M 208 – 211 F 433 – 446 M 208 – 211 F 437 – 451
Shouting or yelling 7.0 11.5 11.6 9.6 14.5 15.3 8.1 18.8 8.2 12.6 13.0 12.8
Belittlement or humiliation 10.5 23.1 15.1 24.0 26.2 28.9 29.0 31.7 15.5 26.3 19.2 24.5
Assignment for punishment 7.0 9.7 - - 12.9 12.2 - - 10.1 11.7 - -
Someone else took credit for student's work 2.0 3.8 11.8 17.3 4.8 6.3 21.0 20.1 3.9 4.1 15.9 18.5
Threats to harm 3.5 2.9 3.5 0 3.2 6.3 9.7 2.7 3.4 5.7 6.7 2.0
Slapping, pushing etc. 0 0 8.1 1.0 1.6 0 14.5 2.6 0.5 0.2 8.7 2.5
Gender-based harassment or mistreatment 3.5 10.6 7.0 7.7 8.1 23.2 4.8 9.0 4.8 15.1 7.7 6.1
Racial, ethnic harassment or discrimination etc. 2.3 4.9 3.5 10.7 4.8 3.7 9.7 6.4 2.4 3.0 5.8 8.2
Threat to fail or give a low mark 2.3 3.9 - - 9.8 9.1 - - 5.3 6.2 - -
Negative remarks 5.8 8.8 21.2 25.5 4.9 10.3 17.7 25.7 5.3 8.5 20.3 23.3
Mistreatment in relation to years of study
The reported occurrence of mistreatment both by staff and fellow students increased with the number of years of study (Table 6). This was especially true of sexual harassment or gender-based mistreatment and threats to fail a student or give a low mark, which were reported 2–4 times as often during or after the 4th year as in the 2nd year both by males and females (Table 6). The same tendency was also observed in mistreatment by fellow students. Female students who had been studying for 4 years or more reported shouting and yelling to be twice as common as those who had been studying for 1 or 2 years (Table 6). Men reported the largest increase in belittlement and humiliation and in some one else taking credit for their work. There was no change in the reported occurrence of sexual harassment by fellow students over the years.
Sexual harassment and other forms of gender-based mistreatment
The female students reported gender-based mistreatment significantly more commonly than the males (p < 0.0001) and the frequency of this increased with the duration of studying (Figure 3). 21% of the female students and 10% of the males had either personally experienced or observed some form of sexual harassment or gender-based mistreatment during their studies. The occurrences of different forms of this on the part of teachers or other staff, as reported by female and male students, are given in Table 7. The most common types were derogatory remarks (sexist slurs), affecting 11.5% of the female students and 3.4% of the males, while 9.0% of the female students and 2.9% of the males had experienced gender-based discrimination (favouritism). Equal percentages of men and women (3.4%) reported having experienced sexual approaches (advances). The faculty staff mistreated female students more often than male ones (p = 0.0002), but mistreatment by fellow students was equally common among both. Sexual harassment or gender-based mistreatment was reported most often by the female students in the Faculty of Medicine (28.4%) and the Faculty of Humanities (24.2%), and the lowest figures reported by women were in the Faculty of Sciences (10.5%) (Table 8). 24.1% of the male respondents in the Faculty of Medicine reported sexual harassment or other forms of gender-based mistreatment. Of the categories of staff, lecturers were most often reported as sources of sexual harassment or discrimination (Figure 2).
Figure 3 Sexual harassment or mistreatment by staff (%) as reported by male and female students, by years of study (N = number of students).
Table 7 Reported occurrence (%) of different types of gender-based harassment or mistreatment by staff, by gender of the respondent.
Type of sexual harassment or mistreatment Males (N = 211) % Females (N = 451) % All (N = 662) %
Denied opportunities 1.9 3.7 3.2
Exchange of rewards for sexual favours 1.0 0 0.3
Advances 3.4 3.4 3.4
Sexist slurs 3.4 11.5 8.9
Sexist teaching material 1.0 2.0 1.7
Malicious gossip 2.9 1.6 2.0
Favouritism 2.9 9.0 7.0
Poor evaluations 2.4 3.8 3.4
Table 8 Occurrence of different types of mistreatment by staff, by respondent's gender and faculty.
Faculties Medicine Humanities Education Technology Science
Male
N = 28–29 Female
N = 106–111 Male
N = 36–38 Female
N = 130–134 Male
N = 11 Female
N = 67–68 Male
N = 90–91 Female
N = 8 Male
N = 40 Female
N = 123–124
% (N) % (N) % (N) % (N) (N)a) % (N) % (N) (N)a) %(N) % (N)
Shouting or yelling 14.3 (4) 21.6 (24) 2.6 (1)* 21.6 (29)* (3) 11.8 (8) 12.1 (11) (2) 10.0 (4) 11.3 (14)
Belittlement or humiliation 32.1 (9) 43.2 (48) 21.1 (8) 32.6 (43) (1) 35.3 (24) 13.2 (12) (3) 30.0 (12) 29.3 (36)
Assignment for punishment 27.6 (8) 22.5 (25) 5.3 (2) 7.5 (10) (1) 19.1 (13) 4.4 (4) (2) 15.0 (6) 10.5 (13)
Threats of harm 6.9 (2) 7.3 (8) 5.3 (2) 5.3 (7) 0 7.4 (5) 4.4 (4)** (1)** 7.5 (3) 5.6 (7)
Threat to fail or
give a low mark 17.2 (5) 6.5 (7) 10.5 (4) 7.6 (10) (1) 5.9 (4) 4.4 (4) 0 5.0 (2) 5.7 (7)
Negative or
disparaging remarks 13.8 (4) 18.9 (20) 13.9 (5) 10.0 (13) (1) 9.0 (6) 3.3 (3) 0 7.5 (3) 10.6 (13)
Gender-based harassment
or mistreatment 24.1 (7) 28.4 (31) 13.2 (5) 24.2 (32) (3) 22.1 (15) 3.3 (3)*** (3)*** 7.5 (3) 10.5 (13)
a) No percentage is given because of the small number of answers.
Statistical significances of differences between females and males within the same faculty: * p < 0.01, ** p < 0.05, *** p < 0.0001
Immoral, unethical and other unacceptable treatment during studies
There were two open questions concerning immoral and unethical treatment or other unacceptable treatment. 47 students reported that they had to do something immoral or unethical during their studies (Table 9), the female students reporting this more frequently than the males (9.6% vs. 3.0%; p = 0.0031). The largest student group reporting immoral and unethical treatment was female students in the Faculty of Medicine (17.1%). The medical students wrote in their open answers that they had to treat patients who were "too sick" or dying, and that this caused them anxiety. The students of science quoted environmental problems with chemicals. The only form of religious discrimination, reported by one student, was favouritism in the attitudes of teachers who were members of a small local religious group towards students belonging to the same group.
Table 9 Occurrence (%) of immoral, unethical and other unacceptable treatment during studies, as reported by male and female students.
Immoral, unethical treatment Other unacceptable treatment
Faculty Male (N = 203) % Female (N = 429) % Male (N = 203) % Female (N = 425) %
Medicine 3.6 17.1 10.3 15.7
Humanities 0 3.1 16.2 17.3
Education 0 5.9 0 14.7
Technology 2.3 14.3 10.2 14.3
Science 7.7 11.6 7.9 17.4
Total 3.0* 9.6* 10.3** 16.5**
*Differences between male and female students, *p = 0.0031; **p = 0.041
The last question concerned all other types of mistreatment during university studies. This was answered by 91 students, with the female students reporting such things more than the males (p = 0.041). Other unacceptable treatment was reported most often by females in the faculties of Medicine (15.7%), Humanities (17.3%) and Science (17.4%). Most of these comments were concerned with teaching skills, poor teacher-student relations, the atmosphere in a department, nasty behaviour by office secretaries and practical training in teaching. The answers to these open-ended questions reported disagreements within and between faculties, e.g. between dentists and medical students or doctors and nurses, or between students in the Faculty of Technology and either the Faculty of Humanities or the Faculty of Education. Students in the Faculty of Education also reported that teachers gave assignments as a means of punishment, that they threatened students, and that they were lacking in punctuality and were in the habit of cancelling their lectures at the last moment.
Overall estimation of mistreatment in the university
The frequencies of personally experienced mistreatment (Figure 1) were lower than the overall perceptions of mistreatment during university studies (Figure 4). Female students reported more mistreatment in the university overall than did the male students (Figure 4).
Figure 4 Overall perceptions of different types of mistreatment as reported by students (n = 629 – 650) when asked in the form: "How often does each type of mistreatment occur at your university?".
Differences between the faculties
The results suggest that the faculties have their own "typical" modes of mistreatment (Table 8). Belittlement or humiliation were especially often reported by males in the Faculties of Medicine (32.1%) and Science (30.0%), whereas the respective figure in Technology was 13.2%. Among females the highest figures were in Medicine (43.2%), followed by Education (35.3%). Negative or disparaging remarks were reported most often by females (18.9%) and males (13.8%) in Medicine, in contrast to only 3.3% of males (3 out of 90) in Technology. Shouting and yelling was rarely reported by males in the Humanities (2.6%), whereas the females in the same faculty reported this 9 times more often (21.6%). Threats to fail a student or to give a low grade were reported most often by male students of Medicine (17.2%). Assignments given as punishment were evident in the Faculties of Medicine (males 27.6%; females 22.5%) and Education (females 19.1%). The gender that was in the minority in a faculty regularly reported mistreatment more often than the majority gender, e.g. female students in the Faculty of Technology.
Discussion
Mistreatment appeared to be very common, since about half of the students had experienced some form of mistreatment by staff or faculty members and every fifth student reported at least one instance of mistreatment. This can sometimes be explained by racial, ethnic or social discrimination [17,18], but that is not the case here, since the social, ethnic and racial backgrounds of our students were very homogeneous. Some students will evidently be taken as victims of extensive mistreatment, as can be seen from the fact that 25 female students (6.6%) and eight male students (4.3%) reported having experienced as many as six or more different types of mistreatment. These high figures demand particular attention. The sense of victimization is a complex issue related to negative identification, the sociological and psychological environment and personality [19]. The reporting of any form of abuse or mistreatment is to some extent subjective and depends on the personality and psycho-social structure of the respondent, and thus vulnerable to over reporting. It is difficult to generalize on results of this kind, but in the light of figures reported also by other authors [3,5,7,15,20] mistreatment of medical students seems to be very common.
Also, about half of the students reported some form of mistreatment by their fellow students. Belittlement, humiliation and negative remarks were astonishingly common. "Inappropriate comments" by fellow-students were also the major category of harassment identified by White [5]. In our material students seemed not to appreciate each other's fields of study, and the present atmosphere in certain disciplines evidently does not support collaboration. It is important that the teachers should not express attitudes of their own that belittle other faculties or disciplines.
The occurrence of mistreatment generally increased with the number of years of study, as also reported earlier [5,10,21,22]. A significant difference has been reported, for example, between the percentages of second-year and third-year medical students quoting any experience of mistreatment (37% vs. 76%) [21]. Furthermore, in accordance with some other studies [3,5,23,24], the women were treated worse than men and were more seriously disturbed by the treatment. It was surprising that medical students reported mistreatment most often in the present survey, even though Uhari et al. [11] had also reported high rates of negative experiences, up to 70%, among students in two Finnish faculties of medicine in the early 1990s and a frequency of 37% for sex-based mistreatment. Very similar rates was reported by White [5] at an Australian medical school, where 37.9% of the students reported some form of sexual harassment (male 24.6%; female 47.9%). The female respondents accounted for 80% of all the events reported, and often reported the same form of harassment more than once.
Twenty-one percent of the female students and 10% of the male students had either personally experienced or observed some form of gender-based mistreatment or discrimination during their studies, the frequency being 2.5-fold among fourth- year students or over by comparison with first-year students. These figures are in accordance with those of White [5], who reported that sexual harassment at medical school was highly significantly (P < 0.001) more common during the clinical years 4–6 than during the pre-clinical years 1–3. Sexual harassment and other sex-based forms of mistreatment experienced especially by women can be ways of trying to marginalize women in university settings [25]. They can have this effect in practice whether or not they are intended as such, because it is not exhilarating to seek a career in a field where you may be harassed sexually and marginalized [26]. High figures for sexual and gender-related harassment have been reported by Larsson [3] in Sweden, for example.
The high figures for mistreatment by lecturers and teachers without tenure may point to signs of frustration. Teaching is no longer held in high regard in universities by comparison with research work, which is appreciated and respected far more. Recent changes in the working environment have also resulted in insecurity, which may be reflected in the treatment of students by such categories of staff. These issues nevertheless cannot provide a full explanation for the mistreatment that students encounter while studying at university.
It is possible that the use of aversive methods to make students learn and behave better has been passed down from teacher to learner, i.e. there is a "transgenerational legacy" that leads to future mistreatment of others by those who themselves have been mistreated. Also, to add to the picture, it is possible that the attitudes, forms of behaviour and values that characterise study cultures and atmospheres in faculties may be connected with discipline-based or professional socialisation, or different moral orders, to use the terminology of Becher [14] discussed at the beginning of this paper. The students' experiences of mistreatment in different faculties gave some messages regarding attitudes and hidden assumptions that can be interpreted as being connected with the kind of discipline-based moral order that Becher discusses [14]. This type of discipline-based mistreatment was most evident in medicine and education, fields in which academic socialisation can be greatly influenced by the norms of socialisation that are traditionally emphasised in the professions for which the students are being educated. These norms can exercise a highly covert influence and form tacit but fundamental assumptions and prejudices [14].
The forms of mistreatment discussed by the students of medicine were very often connected with the position of the student. In a faculty of medicine, as in a hospital, one's position (doctor, nurse, patient, student) seems to be central, and is based on a clear hierarchy within the system. Sexual harassment can be conceptualized as an institutionally sanctioned display of the power that the harasser believes he/she possesses in relation to the victim [5]. This power is often thought to be obtained and derived either from gender or from formal status in the workplace, but it may also play a role in situations where there is no apparent power over the victim, such as a patient harassing a doctor. In such cases the victim may hold the power but the harasser exercises a contra-power and may cause harassment in spite of the apparent formal power imbalance. This may also be true in the case of harassment by a fellow student.
The position of a student in the hierarchy is not so clear as that of a professional, especially when practising in a hospital. The position of an individual in a hospital is often connected with the work done with patients: e.g. who is in a position to decide about injections, operations etc, whose task is it to perform each action, who is allowed to have a voice in each situation and who is not. In that kind of setting it is easy to understand that the atmosphere can be characterised by a strong measure of position-based control and the punishment and mistreatment connected with it, i.e. the use of power. These issues thus characterise the moral order of the study culture. The treatment of students of education – especially by their teachers – points to different areas of control and punishment from those that apply to medicine, but which also relate to the use of power, since teachers control their pupils' use of time, the nature of the tasks they have to do, etc. Correspondingly, the special topics emphasised by the trainee teachers were the control maintained over their time and attendance. The only comment on an obligation to be present in lectures or exercises in spite of being ill was made by a student of teacher education.
The topic of sexual harassment can also be approached through the concept of moral order. Another study conducted by us [27] gives more detailed information on the situation in the Faculty of Technology, noting that although the respondents indicated that they had not experienced sexual harassment, some other students claimed to have left that faculty because of the sexually harassing atmosphere they had experienced. It is thus possible to think that the moral order of the Faculty of Technology, or of some of its departments, may include the idea that as a female student you must abide a sexually harassing atmosphere if you wish to study there.
In both the Faculties of Medicine and in that of Education the female students reported mistreatment more often than the males, as if the order were stricter for them. The question of treatment is especially important in these faculties, for several reasons. One is that most of the students in these faculties are female and mistreatment seems to disturb them more than males. On the other hand, the nature of this treatment is an especially important question in the professional areas concerned, as it is a question of how teachers treat the pupils for whom they are responsible and how doctors treat their patients. These aspects need further study.
Conclusion
For most people the vision of universities is that they are peaceful sanctuaries protected from the "real world", where students are taught high-level academic skills, read, write and do research. It is also thought that equity, dignity, respect and justice are emphasized in these environments. But, as earlier [17,18,28,29], the present survey shows that this is not true. Students are not treated equally in our universities.
The main tasks of a university – research, teaching and learning – are seriously threatened if members of staff mistreat their students, and this is especially true where women are concerned. The teaching atmosphere during one's studies is important not only for learning but also for building up a positive professional identity. The attitudes adopted during university studies, positive or negative, will have an impact on graduates' values and behaviour in their future working life. And of course, a university hires its researchers and teachers of tomorrow from today's students. Therefore, it is very important to prevent mistreatment from being transmitted to the next generation, in order to increase professionalism in medicine [30], as in other disciplines. This may be achieved by various methods, such as role playing [30] and educational strategies [31].
Our findings emphasize the need to develop and maintain a good, impartial and supportive atmosphere within medical studies and training in order to develop the personality traits needed to practise as a doctor. Serious discussion is obviously needed on the behaviour and habits of teaching staff. Indeed, some degree of educational intervention is needed in all faculties of a university. Issues of the hidden moral order should also be included in these discussions. Several interesting perspectives for educational interventions have been provided by White [31] and Heru [30], for example, and by B Sandler and R Shoop [32].
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AR participated in the planning, execution, analysing and writing the manuscript. VS participated in analysing and writing the manuscript. MN participated in the planning, execution, analysing and writing the manuscript. ML participated in the planning, execution, analysing 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:
Supplementary Material
Additional File 1
Questionnaire.
Click here for file
==== Refs
Borg MG The extent and nature of bullying among primary and secondary schoolchildren Educ Res 1999 41 137 153
Richman JA Rospenda KM Nawyn SJ Flaherty JA Fedrich M Drum ML Johnson TP Sexual harassment and generalized workplace abuse among university employees: Prevalence and mental health correlates Am J Public Health 1999 89 358 363 10076485
Larsson C Hensing G Allbeck P Sexual and gender-related harassment in medical education and research training: results from a Swedish survey Med Educ 2003 37 39 50 12535114 10.1046/j.1365-2923.2003.01404.x
Maida AM Vasquez A Herskovic V Calderon JL Jacard M Pereira A Widdel L A report on student abuse during medical training Med Teach 2003 25 497 501 14522671 10.1080/01421590310001605615
White GE Sexual harassment during medical training: the perceptions of medical students at a university medical school in Australia Med Educ 2000 34 980 986 11123560 10.1046/j.1365-2923.2000.00684.x
Brown J Cambell EA Fife-Schawn C Adverse impacts experienced by police officers following exposure to sex discrimination and sexual harassment Stress Med 1995 11 221 228
Richman JA Flaherty JA Rospenda KM Christensen ML Mental health consequences and correlates of reported medical student abuse JAMA 1992 267 692 694 1731137 10.1001/jama.267.5.692
Einarsen S Raknes BI Harassment in the workplace and the victimization of men Violence Vict 1997 12 247 263 9477540
Daugherty SR Baldwin DC JrRowley BD Learning, satisfaction and mistreatment during medical internship: A national survey of working conditions JAMA 1998 279 1194 1199 9555759 10.1001/jama.279.15.1194
Lebenthal A Kaiserman I Lernau O Student abuse in medical school: a comparison of students' and faculty's perceptions Isr J Med Sci 1996 32 229 238 8606140
Uhari M Kokkonen J Nuutinen M Vainionpää L Rantala H Lautala P Väyrynen M Medical student abuse: An international phenomenon JAMA 1994 271 1049 1051 8139064 10.1001/jama.271.13.1049
Sheehan KH Sheehan DV White K Leibowitz A Baldwin DC Jr A pilot study of medical student "abuse" JAMA 1990 263 533 537 2294325 10.1001/jama.263.4.533
Ryynänen K Constructing physician's professional identity – explorations of students' critical experiences in medical education Academic dissertation 650 Acta Univ Oul D 2001
Becher T Academic tribes and territories: Intellectual enquiry and the cultures of disciplines 1989 Stony Stratford: Society for Research into Higher Education & Open University Press
Baldwin DC JrDaugherty SR Eckenfels EJ Student perceptions of mistreatment and harassment during medical school: a survey of ten United States schools West J Med 1991 155 140 145 1926843
Armitage P Berry G Statistical methods in medical research 1994 London, Blackwell
Brooks A Academic women Society for research into higher education 1997 Suffolk, St Edmundsbury Press Ltd
Margolis E Romero M The department is very male, very white, very old and very conservative Harv Educ Rev 1998 68 1 23
Heru AM The linkages between gender and victimhood Int J Soc Psych 2001 47 10 20
Wolf TM Randall HM von Almen K Tynes LL Perceived mistreatment and attitude changes by graduating medical student: a retrospective study Med Educ 1991 25 182 190 1857273
Richardson DA Becker M Frank RR Sokol RJ Assessing medical students' perceptions of mistreatment in their second and third years Acad Med 1997 72 728 730 9282152
Lee D Sexual harassment in PhD supervision Gend Educ 1998 10 299 312 10.1080/09540259820916
Lacy FJ Sheehan BA Job satisfaction among academic staff: an international perspective Higher Educ 1997 34 305 322 10.1023/A:1003019822147
Toren N Moore D The academic "hurdle race": A case study Higher Educ 1998 35 267 283 10.1023/A:1003122603226
Eyre L The discursive framing of sexual harassment in a university community Gend Educ 2000 12 293 307
Benson D Thomas G Sexual harassment on a university campus: the confluence of authority relations, sexual interest and gender stratification Soc Probl 1982 29 236 251
Sunnari V Heikkinen M Rautio A Väyrynen M Nuutinen M Teräs L, Sunnari V, Kailo K. Raahe, Rannikon Laatupaino Sexual Harassment and Molestation as a Part of Sustaining Segregation in the University – a Sudy Based on Materials Collected during the Period of a Decade Education, Gender Socialisation and Technology – Perspectives to Segregation 2005 83 99 English summary
Poole M Bornholt L Summers F An international study of the gendered nature of academic work: some cross-cultural explorations Higher Educ 1997 34 373 396 10.1023/A:1003075907126
Erwin L Maurutto P Beyond Access: Considering gender deficits in science education Gender Educ 1998 10 51 69 10.1080/09540259821096
Heru AM Using role playing to increase residents' awareness of medical student mistreatment Acad Med 2003 78 35 38 12525407
White GE Setting and maintaining professional role boundaries: an educational strategy Med Educ 2004 38 903 910 15271052 10.1111/j.1365-2929.2004.01894.x
Sandler B Shoop R Sexual harassment on campus A guide for administrators, faculty, and students 1997 Boston, Allyn and Bacon
|
16232310
|
PMC1285362
|
CC BY
|
2021-01-04 16:30:56
|
no
|
BMC Med Educ. 2005 Oct 18; 5:36
|
utf-8
|
BMC Med Educ
| 2,005 |
10.1186/1472-6920-5-36
|
oa_comm
|
==== Front
BMC MicrobiolBMC Microbiology1471-2180BioMed Central London 1471-2180-5-601622344410.1186/1471-2180-5-60Research ArticleGenometrics as an essential tool for the assembly of whole genome sequences: the example of the chromosome of Bifidobacterium longum NCC2705 Guy Lionel [email protected] Dimitri [email protected] Philippe [email protected] Claude-Alain H [email protected] Département de Microbiologie Fondamentale, Faculté de Biologie et Médecine, Université de Lausanne, CH-1015 Lausanne, Switzerland2005 13 10 2005 5 60 60 29 4 2005 13 10 2005 Copyright © 2005 Guy 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
Analysis of the first reported complete genome sequence of Bifidobacterium longum NCC2705, an actinobacterium colonizing the gastrointestinal tract, uncovered its proteomic relatedness to Streptomyces coelicolor and Mycobacterium tuberculosis. However, a rapid scrutiny by genometric methods revealed a genome organization totally different from all so far sequenced high-GC Gram-positive chromosomes.
Results
Generally, the cumulative GC- and ORF orientation skew curves of prokaryotic genomes consist of two linear segments of opposite slope: the minimum and the maximum of the curves correspond to the origin and the terminus of chromosome replication, respectively. However, analyses of the B. longum NCC2705 chromosome yielded six, instead of two, linear segments, while its dnaA locus, usually associated with the origin of replication, was not located at the minimum of the curves. Furthermore, the coorientation of gene transcription with replication was very low.
Comparison with closely related actinobacteria strongly suggested that the chromosome of B. longum was misassembled, and the identification of two pairs of relatively long homologous DNA sequences offers the possibility for an alternative genome assembly proposed here below. By genometric criteria, this configuration displays all of the characters common to bacteria, in particular to related high-GC Gram-positives. In addition, it is compatible with the partially sequenced genome of DJO10A B. longum strain. Recently, a corrected sequence of B. longum NCC2705, with a configuration similar to the one proposed here below, has been deposited in GenBank, confirming our predictions.
Conclusion
Genometric analyses, in conjunction with standard bioinformatic tools and knowledge of bacterial chromosome architecture, represent fast and straightforward methods for the evaluation of chromosome assembly.
==== Body
Background
Bifidobacterium longum is an obligate anaerobe, belonging to the Actinomycetales, a branch of the high-GC Gram-positive bacteria which includes, among others, corynebacteria, mycobacteria and streptomycetes. B. longum is a natural colonizer of the gastrointestinal tract (GIT) and the vagina [1]. It is one of the very first bacteria which colonize the sterile GIT of newborns, predominating in breast-fed infants until weaning [2]. Thereafter, its numerical importance decreases, while Bacteroides and other taxa replace it [3]. B. longum, a harmless bacterium, considered to play an important role in maintaining a healthy GIT by preventing diarrhea, improving lactose intolerance, and participating to immunomodulation [1], is now widely used in health-promoting foods.
Recently, the whole genome of B. longum strain NCC2705 has been sequenced [4]. Comparison with other high-GC Gram-positives revealed high levels of protein homology with Streptomyces coelicolor A3(2) (34% of best hits), Mycobacterium tuberculosis (9.3% of best hits) and, to a lesser extent, with other actinobacteria as well as with some unrelated genera such as Clostridium and Streptococcus. Surprisingly, it contains a very high number of genetic entities related to mobile elements such as transposons and plasmids. There are 14 integrases/recombinases, 16 intact insertion sequences (ISs), many remnants of ISs, one integrated plasmid, many possible remnants of integrated plasmids and prophages. The origin and the terminus of chromosome replication of B. longum NCC2705 could not be accurately localized along the initial whole genome sequence. Today, DJO10A, another strain of B. longum, is almost fully sequenced, but not assembled.
The presence of many ISs and IS remnants in B. longum NCC2705 leaves open the possibility of major chromosomal rearrangements [4]. These internal recombination events were already advanced to explain the poor conservation of gene order during the evolution of prokaryotic genomes [5].
It appears that major chromosomal rearrangements almost always consist in inversions of a segment of the chromosome centered on the origin of replication [6,7]. Other inversions are probably counter-selected [8], since, unlike inversions around the origin of replication, they change the orientation of transcription relative to DNA replication or they change the length of chromosome arms. Such events have adverse effects on both replication speed and transcription [9,10]. Alternatively, it has been proposed that rearrangements preferentially centered on the origin of replication are favored by the bidirectional DNA replication: starting simultaneously from the origin of chromosome replication, the two replication forks are at the same distance from it and are likely to be in close contact [6]. Thus, DNA breaks produced by topoisomerases would generate structures suitable for recombinations between the two chromosomal arms, leading to origin-centered rearrangements [6].
The coorientation between gene transcription and DNA replication is apparently a fundamental feature of bacterial chromosome architecture. More specifically, ORFs and tRNA genes follow a similar tendency and all so far identified ribosomal RNA operons are cooriented with DNA replication [11,12]. The asymmetric bias in the nucleotide composition at the genome level is another relevant feature (for a review see [13]). The leading strand, defined by chromosome replication, is generally enriched in guanines (Gs) and depleted in cytosines (Cs). To explain this observation several proposals have been advanced: (i) a preferential usage of certain codons to avoid frameshifting during translation [14], (ii) the enrichment of coding sequences in purines so as to avoid mRNAs secondary structures [15,16], (iii) mutational biases targeting single-stranded DNA present during transcription [17] or (iv) during DNA replication [18]. Mechanisms that would lead to the observed biases in models (i) to (iii) rest on the widespread coorientation of gene transcription and chromosome replication (see above). These asymmetric biases have allowed to unambiguously determine the origin of replication in almost all bacteria [12,18-20] as well as the terminus of replication in a large majority of the species [12].
Genomic rearrangements are often highlighted by comparison of whole chromosomal sequences belonging to the same species or genus. For example, dot-plot analyses revealed two recombination events in Streptococcus pyogenes SSI-1, with respect to other S. pyogenes strains, leading to an inversion around the origin of replication [21]. Since they do not change the orientation of transcription relative to DNA replication, these symmetric rearrangement were not revealed by nucleotide bias (skew) analysis.
However, several examples of asymmetric rearrangements are known, pointed out by nucleotide skew analyses. Several isolates of the original clone of Pseudomonas aeruginosa PAO1, a high-GC gamma-proteobacterium and an opportunistic pathogen, have an inversion of a third of their chromosome. The inversion occurs by homologous recombination between two rRNA loci: rrnA and rrnB. As a consequence, the circular chromosome is divided into two inequal arms of one and two thirds, instead of the usual two halves [22]. This asymmetry is obvious on a cumulative GC- or TA skew curve (see the P. aeruginosa PAO1 page on Comparative Genometrics website [23]). The citrus pathogen Xylella fastidiosa 9a5c is another example of asymmetric rearrangements, that is highlighted when compared with another X. fastidiosa strain, Temecula1 [24,25]. In this case, the rearrangements occur between three pairs of prophage-related elements [25], also dividing the chromosome of strain 9a5c in two arms of inequal lenghts (one third and two thirds), as in P. aeruginosa PAO1 (see pages for X. fastidiosa strains on Comparative Genometrics [23]). In Yersinia pestis strains, the high number of insertion sequence (IS) copies leads to frequent recombination events, inverting segments of the chromosome and changing their orientation of transcription with respect to replication. These inversions are easily spotted on a GC skew plot (see pages for Y. pestis strains on Comparative Genometrics [23]). In all three above cases, the rearrangements occur naturally, and do not constitute an incorrect genome assembly.
In this contribution we assess the assembly of the initially deposited genome sequence of B. longum NCC2705 by genometric methods, rapid and efficient tools suitable for testing the assembly of prokaryotic chromosomes [23]. Our analysis, strongly suggesting that the chromosome of B. longum NCC2705 was initially misassembled, was confirmed by Schell et al. [26] during the review of this contribution.
Results and discussion
Analysis of the initial Bifidobacterium longum NCC2705 genome sequence
Investigation with genometric tools of the initially released nucleotide sequence of B. longum NCC2705 (Configuration I [GenBank:NC_004307.1]) revealed several atypical features.
First, cumulative GC skew on the first codon position, as well as the cumulative ORF orientation skew, yielded a curve with six significant changes of the slope sign. Furthermore, the dnaA gene was not located at the lowest minimum of the curve (Figure 1A and 1B). This is clearly different from all known high-GC Gram-positives (see [23,27]), since they essentially exhibit one maximum and one minimum, the latter being generally located in the vicinity of the dnaA gene [12].
The presence of dnaA at a place other than the minimum of the cumulative ORF orientation curve has never been reported in high-GC Gram-positive bacteria. For a large majority of bacterial species, it was shown that dnaA, a gene whose product binds to the origin of replication and participates in the initiation of replication, is located close to the origin [12,28]. More generally, in archaea, gene orc1/cdc6, which encodes the archaeal counterpart of DnaA, is very often also located close to the origin of chromosome replication. Finally, in sequenced genomes, half of archaea and most of bacterial genes encoding origin binding proteins are close to the minimum of the cumulative GC skew and ORF orientation curves [12].
Second, in the first published B. longum NCC2705 sequence, coorientation indexes (CI), i.e. the proportion of genes or of a given subset of genes that are transcribed in the direction of chromosome replication, revealed significant anomalies. Indeed, the CI of protein encoding genes and tRNAs was 0.48 and 0.47, respectively, while only one out of the four rRNA operons was cooriented (Table 1). These low CIs are most uncommon in bacteria where it has been shown that the majority of protein encoding genes [11] as well as of tRNA genes are cooriented with chromosome replication (i.e. CIs higher than 0.5), while, so far, a strict coorientation of rRNA operons constitutes a universal rule in prokaryotes [12]. More specifically, CIs of protein encoding genes of M. tuberculosis CDC1551 and S. coelicolor A3(2), two high G+C Gram-positives related to B. longum NCC2705, are 0.59 and 0.55, respectively. CIs of tRNAs of the same species are similar, i.e. 0.62 and 0.57, respectively, whereas all rRNA operons are cooriented (Table 1).
Third, between-species whole-genome alignments of B. longum NCC2705 and M. tuberculosis CDC1551 or S. coelicolor A3(2) revealed a very poor conservation of gene order (Figure 2) [6,7]. Indeed, correlation coefficients measuring the conservation of gene order between species are close to zero, whereas in the ideal case the coefficient of correlation should be 1 and -1 for the direct and respectively indirect homologous DNA segment subsets.
In summary, genometric analyses revealed major anomalies in the organization of the B. longum NCC2705 genome: (i) several changes in the sign of the slope of the cumulative nucleotide skew curves, and location of the dnaA gene far from the minimum of the curve, (ii) low gene coorientation indexes and (iii) absence of correlation between B. longum NCC2705 and related species in between-species whole-genome alignments.
Relationship of B. longum strains NCC2705 and DJO10A
Availability of numerous contigs of the genome of DJO10A, another B. longum strain, strongly suggested that the initially reported sequence of NCC2705 chromosome could have been incorrectly assembled, or had undergone major chromosomal rearrangements. Indeed, BLAST results reveal three DJO10A long scaffolds -number 1, 8 and 9 – each with a large number of hits in two different regions of the B. longum NCC2705 chromosome (Figure 1A and 1B). These homology discontinuities occur in four regions encompassing extrema of the cumulative skew curves (Figure 1A). Within each of these four regions, an insertion sequence was identified: ISBlo2a and ISBlo2b, belonging to the IS21 family, and ISBlo5c and ISBlo5d, belonging to the IS256 family (Figure 1A, 1B and Table 2)[4].
The presence of these insertion sequences (Table 2) at putative recombination sites offers a straightforward way to account for chromosomal rearrangements which can mediate the shift between initial configuration of strain NCC2705 and the putative configuration of strain DJO10A, here below designated configurations I and II, respectively (Figure 3). Indeed, translocation and inversion of the two large segments a and d, achieved by two homologous recombination events between ISBlo2a and 2b, on one side, and ISBlo5c and 5d, on the other side, would allow the interconversion between configurations I and II. This assumption is supported by the cross-exchange of direct repeats in both IS pairs, as noted in the IS Finder database [29].
Analysis of the configuration II of the B. longum NCC2705 chromosome
Genometric analyses of the B. longum NCC2705 chromosome in configuration II reveal a genome architecture typical of high-GC Gram-positive organisms. Indeed, the cumulative GC-skew curve performed on the first codon positions and the cumulative ORF orientation skew curves are very similar to those characteristic of high-GC Gram-positive chromosomes (Figure 1C). First, both skew curves exhibit essentially one minimum and one maximum corresponding, respectively, to the origin and the terminus of chromosome replication; dnaA being at the minimum of the curve while the integrated plasmid is close to the probable terminus of replication, at the maximum of the skew curves. The two minor peaks correspond to the extremities of the integrated plasmid. This genetic element is antioriented, so that the majority of its genes are transcribed in the opposite direction with respect to chromosome replication, leading to a short reversal both in the cumulative ORF orientation skew and the cumulative first codon GC skew curves. This fact was already reported for other integrated elements (for example in Parachlamydiaceae UWE25 [30]) and is possibly a consequence of the instability of the integration. Integration of foreign DNA stretches close to the terminus is common, for example in prophages [25]. A higher frequency of recombination at the terminus of replication was proposed as the source of this instability [31,32]. It appears that of the six changes in the sign of the skew slopes of strain NCC2705 in configuration I, one does actually correspond to the origin of chromosome replication and a recombination site, three to other recombination sites, one to the likely terminus of replication, and the last one to the distal extremity of an integrated plasmid. Second, in configuration II, the CIs of protein encoding- and tRNA genes are 0.66 and 0.61, respectively, while all rRNA operons are cooriented, a situation characteristic of prokaryotes (Table 1). Third, between-species whole-genome alignments of B. longum in configuration II and S. coelicolor A3(2) or M. tuberculosis CDC1551 display a fair conservation of gene order (Figure 2C, 2D). Correlation coefficients of type II regressions for direct and indirect homologous DNA segment subsets are much closer to 1 or -1, respectively, than those obtained with the sequence of B. longum NCC2705 in configuration I (Figure 2C, 2D). Finally, each of the scaffolds 1, 8 and 9 of B. longum DJO10A has hits in one single region of the B. longum NCC2705 chromosome in configuration II (Figure 1C).
A genome sequence corresponding approximatively to configuration II of B. longum NCC2705 [GenBank:NC_004307.2] has been recently deposited in the GenBank database by Schell et al. [26]. Whereas we hypothesized that the two pairs of IS, ISBlo2a, ISBlo2b, ISBlo5c and ISBlo5d were the only four chromosomal rearrangement loci, these authors found experimental evidences that, moreover, the initial sequence of NCC2705 had been misassembled at the level of the three ribosomal RNA operons. The sequence in configuration II proposed in this contribution has three DNA segments (totalizing 226 kb, i.e. 10% of the genome) which are differently assembled than the corrected sequence. These assembly discrepancies have only very limited consequences on the results of our analyses.
Thus, configuration II, similar to the recently deposited sequence of the B. longum NCC2705 genome [Genbank:NC_004307.2], is endowed with all chromosomal features common to high-GC Gram-positive bacteria: (i) cumulative GC-and ORF orientation skew curves are typical and the dnaA gene is located at the minimum of the curves, (ii) between-species whole-genome alignments provide the expected X-shape and the coefficients of correlation are relatively close to 1 or -1 and (iii) relevant contigs of B. longum DJO10A are each homologous to a single continuous region of the proposed NCC2705 chromosome.
Conclusion
Genometric analyses – nucleotide skews, coorientation indexes, BLAST comparisons and between-species whole-genome alignments – revealed a most peculiar chromosomal architecture of the initially reported sequence of the B. longum NCC2705 genome. This observation may have two explanations.
First, it is highly probable that in the final stages of the sequencing process, the genome of B. longum NCC2705 was misassembled, a possibility presently favored by Schell and coworkers, in particular since independently performed long-range PCR experiments confirm the presence of configuration II, but could not detect configuration I (F. Arigoni, personal communication). This is in full agreement with genometric analyses of more than 150 genome sequences ([23] and supplementary material of Tillier and Collins [20,33]) which reveal a near universal architecture of prokaryotic chromosomes, also found in configuration II of B. longum NCC2705.
Second, the genome of NCC2705 may possibly undergo major chromosomal rearrangements, yielding either of the two alternative configurations I and II. The interconversion between them, achieved by two crossovers between two pairs of homologous insertion sequences (IS), would have drastic consequences. In particular, in configuration II, the cooriented transcription of the majority of the genes, including all rRNA operons, with chromosome replication would allow higher growth rates in suitable conditions. Absence of significant coorientation in configuration I due to the inversion of large segments b and d, representing about 50% of the chromosome, would probably considerably increase the generation time because of collisions between the DNA- and the RNA polymerase [9]. However, as discussed here above, the adverse effects of inversions, other than those around the origin of replication, would render the existence of configuration I highly unlikely. Actually, the latter has apparently not been detected in long range PCR experiments (F. Arigoni, personal communication). However, inversions of relatively long segments – leading to the antiorientation of the majority of the genes in the segment – have been reported, for example, in Yersinia pestis [34-37], and thus may not be completely excluded in B. longum NCC2705.
For the first time, our analyses illustrate the potential of fast and straightforward genometric methods to test genome assembly. They almost immediately revealed gross anomalies of the B. longum NCC2705 initially published sequence, pointing to an incorrect assembling. In conclusion, although their results have to be supported by experimental verification, these simple and powerful tools are essential for the assembly of a chromosome sequence, and for its final validation.
Methods
Sequences
Full genome sequences and annotation files of B. longum NCC2705,S. coelicolor A3(2), M. tuberculosis CDC1551 and contigs of an unfinished sequence of B. longum DJO10A were retrieved from NCBI database [38,39]. For B. longum NCC2705, the initial [GenBank:NC_004307.1] and the second [GenBank:NC_004307.2] versions of the chromosome (released on August, 27th, 2002 and on January, 21st, 2005 respectively) were downloaded. Sequence of configuration II as proposed in this contribution, and both initial and corrected versions of the B. longum NCC2705 genome are available in fasta format [40]. As proposed by Cebrat et al. [41], we term the genome sequences available on databases and those of the complementary strands the Watson- and Crick strands, respectively.
Genome analyses
Genomes were investigated by cumulative genomic GC skew, first codon position GC skew, and ORF orientation skew [19,20,42]. We used the algorithms described in [27,30] and implemented in the Genometrician's Scooter [43].
Nucleotide skews
As defined by Lobry [18], a GC skew is the difference between the number of Gs and Cs normalized to the G+C content. In our contribution we used the non-normalized nucleotide skew, calculated in 1-kb windows along the genome. In the genomic GC skew, the whole genome sequence is used. For GC skew on the first codon position, only nucleotides at the first position of codons are considered for the skew calculation.
Cumulative nucleotide skews
Slightly different from the definition of Grigoriev [19], the cumulative nucleotide skew of any given window is the nucleotide skew of the latter (see above) added to the sum of skews of all preceding windows.
Cumulative ORF orientation skews
As in [20], in the ORF orientation skew analysis, the value attributed to each ORF corresponds to its length, considered as positive if the ORF is located on the Watson strand, and negative if encoded on the Crick strand. The cumulative ORF orientation analysis is calculated as a cumulative nucleotide skew by replacing windows and GC skews by genes and ORF orientation skews: the value corresponding to a given ORF is added to the sum of the values of all upstream-located ORFs. A cumulative ORF orientation skew is represented as a function of the position of the center of each gene. We used the number of nucleotide per gene, and not the number of ORFs to normalize the signal to the length of the gene, otherwise, in the cumulative ORF orientation skew plot, small genes would have a greater importance than long ones.
Coorientation Indexes (CI)
For all genomes, coorientation indexes (CI), i.e. the percentage of all or of certain categories of genes – protein encoding genes, rRNAs, tRNAs – transcribed in the direction of DNA replication, were calculated according to [12]. For that purpose, the origin and the terminus of chromosome replication are determined by cumulative GC skew. For B. longum NCC2705, where the cumulative GC skew did not reveal the origin and/or the terminus of replication, the first codon position cumulative GC skew and ORF orientation skews were used. In most so far sequenced bacterial genomes, the origin of replication is located at the minimum of the cumulative skew curves. S. coelicolor A3(2), that has an extremely high G+C content, is an exception since its origin of replication is located at the maximum of the genomic GC skew curve. Generally, the origin of replication was shown to be close to the dnaA gene. The terminus of replication is assumed to be at the maximum of the skew curves, except in S. coelicolor A3(2), where it is assumed to be at the minimum, corresponding to both ends of the linear chromosome. However, for the first reported sequence of B. longum NCC2705, where the skew analyses did not provide the origin or the terminus of replication, we assumed that they are respectively located close to dnaA and at the putative terminus of replication in the integrated plasmid, about 180° from the dnaA gene on the circular chromosome.
BLAST
Basic Local Alignment Search Tool (BLAST) 2.2.4 [44] analysis was performed with the software kindly provided by the NCBI [38] using as a cutoff an expected E-value of 10-2 for alignments of the full genome sequence of B. longum NCC2705 vs. the available contigs of B. longum DJO10A. An E-value of 10-2 indicates that a hit with the same or a better alignment score occurs with a probability of 10-2 when searching the same database with a random sequence. BLAST results with an alignment length below 1000 nucleotides were discarded. BLAST analysis was performed with an expected E-value of 10 for alignments of S. coelicolor A3(2) and M. tuberculosis CDC1551 vs. B. longum NCC2705 in its actual as well as putative alternative chromosomal configuration. For the latter analyses only, hits with an alignment score below 100 were discarded. A hit is defined as direct or indirect if the DNA segments are in the same, or respectively opposite, orientation in both genomes.
Between-species alignments of whole genomes
Also called dot-plot analyses [6,21], genome-to-genome comparisons were achieved according to [45]. The relative positions of homologous segments in pairwise comparisons of bacterial genomes were determined by BLAST (see above).
Correlation coefficients of type II regression (major axis regression) were determined for both direct and indirect hit subsets. If a genome had undergone only exactly symmetric rearrangements around the origin of replication, the correlation coefficients of the direct- and indirect BLAST hit sets would be 1 and -1, respectively. Correlation coefficients close to zero show no correlation between relative chromosomal positions of homologous segments.
Accession numbers
Bifidobacterium longum NCC2705, [GenBank:NC_004307.1] and [GenBank:NC_004307.2]; S. coelicolor A3(2), [GenBank:NC_003888]; M. tuberculosis CDC1551, [GenBank:NC_002755]; B. longum DJO10A, [GenBank:NZ_AABM00000000].
List of abbreviations
kb, kilobase; A, adenine; C, cytosine; G, guanine; T, thymine; rRNA, ribosomal RNA; tRNA, transfer RNA.
Authors' contributions
LG carried out the analyses during his MSc thesis, supervised by CAR who designed and managed the project. LG and CAR proposed the genome configuration II. LG, DK, PM and CAR participated to the interpretation of the results. LG drafted the manuscript in collaboration with CAR and DK. All authors read and approved the final manuscript.
Acknowledgements
We would like to thank Alexandre Panchaud who drew our attention at the particular genome configuration presented by the first reported sequence of B. longum NCC2705. We warmly thank Fabrizio Arigoni and Bernard Berger for sharing unpublished results in discussions initiated by our poster presentation at Genomes 2004: International Conference on the Analysis of Microbial and Other Genomes (Guy L., Karamata D., Moreillon P. and Roten CAH, The genome of Bifidobacterium longum NCC2705: an example of major chromosomal rearrangements revealed by genometric analyses. April 14–17, 2004, The Wellcome Trust Conference Centre, Cambridge, UK). We are particularly grateful to them for informing us in late 2004, after submission of the first version of our paper (July 2004), that they and their colleagues consider that their initial published sequence of B. longum NCC2705 was misassembled.
Figures and Tables
Figure 1 Genometric analyses of the two chromosomal configurations of B. longum NCC2705. Cumulative GC- and ORF orientation skew analyses of chromosomal configurations I (A, B) and II (C) of B. longum NCC2705. Configuration I [GenBank:NC_004307.1] is plotted without modification with respect to the version available on databases (Watson strand), (A) as well as the Crick strand of same sequence starting with the gene dnaA (B). Cumulative GC skew of the first codon position (black curve) and cumulative ORF orientation skew (grey curve) are plotted as a function of their position on the genome. Distance between two graduation marks represents an excess of 104 Gs over Cs in the first codon position cumulative GC skew and of 2·103 nucleotides in the ORF orientation skew. Thin arrows indicate relevant changes of the slope sign. The thick arrow corresponds to the position and the orientation of dnaA. The location of the integrated plasmid described in [4] is indicated. Vertical lines specify the position of copies of ISBlo2 (dashed line) and ISBlo5 (full line) insertion sequences. Segments delimited by these insertion sequences are designated a, b, c and d. Horizontal lines below the curves are the hits of scaffolds 1 (black), 8 (light grey) and 9 (dark grey) of B. longum DJO10A when BLASTed against the B. longum NCC2705 genome in configurations I (A, B) or II (C).
Figure 2 Between-species comparisons of the two chromosomal configurations of Bifidobacterium longum NCC2705 and of related species. Relative positions of homologous segments on sequences of configurations I (A, C) and II (B, D) of B. longum NCC2705 against S. coelicolor A3(2) (A, B) and M. tuberculosis CDC1551 (C, D). Full circles and open circles indicate pairs of homologous DNA segments that have the same, respectively the opposite orientation on the chromosome. Since the S. coelicolor A3(2) genome is linear, its origin of replication cannot be placed at position 1. Therefore, in A and B, both configurations of the B. longum genome are linearized by placing the terminus of replication at the origin of the graph. For B. longum NCC2705 in configuration I (A), the probable terminus of replication of the integrated plasmid has been chosen as the terminus of replication and placed at the origin of the graph. For the B. longum in configuration II, the maximum of the first codon position cumulative GC skew has been chosen as the terminus. On (A) and (B), position and orientation of the dnaA gene on each genome is indicated by a thick arrow. Lines represent type II regressions of direct (full line) and of indirect hits (dashed line). For each regression, the corresponding correlation coefficient is indicated.
Figure 3 Chromosome rearrangements generating the transition between the two Bifidobacterium longum configurations. Maps of B. longum NCC2705 in configurations I (A), II (B) and the proposed mechanism mediating the transition between the two configurations (C). Segments a, b, c and d are indicated. Gene dnaA and the integrated plasmid are represented by an arrow and a thin black line, respectively. ISBlo2a and b are indicated by open triangles, ISBLo5c and d by solid ones. For each segment, thick black arrows indicate the orientation of transcription of the majority of the genes.
Table 1 Coorientation indexes of different genome subsets in some high G+C Gram-positive bacteria
ORFs rRNA genes tRNA genes
B. longum NCC2705 initial sequence 0.48 0.25 0.47
B. longum NCC2705 configuration II 0.66 1 0.61
M. tuberculosis CDC1551 0.59 1 0.62
S. coelicolor A3(2) 0.55 1 0.57
Table 2 Genetic elements on the initial sequence of Bifidobacterium longum NCC2705
5' end location 3' end location Length (nt)
ISBlo2a 295604 298058 2454
ISBlo2b 1522936 1525390 2454
dnaA 794998 796500 1502
ISBlo5c 853915 855280 1365
ISBlo5 1886182 1887547 1365
Integrated plasmid 1813659 1870484 56825
==== Refs
Biavati B Mattarelli P Dworkin M, Falkow S, Rosenberg E, Schleifer KH and Stackebrandt E The family Bifidobacteriaceae The Prokaryotes: An Evolving Electronic Resource for the Microbiological Community 2001 3rd edition, relase 3.7 New York, Springer-Verlag 1 70
Favier CF Vaughan EE De Vos WM Akkermans ADL Molecular monitoring of succession of bacterial communities in human neonates Appl Environ Microbiol 2002 68 219 226 11772630 10.1128/AEM.68.1.219-226.2002
Harmsen HJ Wildeboer-Veloo AC Raangs GC Wagendorp AA Klijn N Bindels JG Welling GW Analysis of intestinal flora development in breast-fed and formula-fed infants by using molecular identification and detection methods J Pediatr Gastroenterol Nutr 2000 30 61 67 10630441 10.1097/00005176-200001000-00019
Schell MA Karmirantzou M Snel B Vilanova D Berger B Pessi G Zwahlen MC Desiere F Bork P Delley M Pridmore RD Arigoni F The genome sequence of Bifidobacterium longum reflects its adaptation to the human gastrointestinal tract Proc Natl Acad Sci U S A 2002 99 14422 14427 12381787 10.1073/pnas.212527599
Casjens S The diverse and dynamic structure of bacterial genomes Annu Rev Genet 1998 32 339 377 9928484 10.1146/annurev.genet.32.1.339
Tillier ERM Collins RA Genome rearrangement by replication-directed translocation Nature Genet 2000 26 195 197 11017076 10.1038/79918
Eisen J Heidelberg J White O Salzberg S Evidence for symmetric chromosomal inversions around the replication origin in bacteria Genome Biol 2000 1 research0011.1 11.9 11178265 10.1186/gb-2000-1-6-research0011
Brewer BJ When polymerases collide: replication and the transcriptional organization of the E. coli chromosome Cell 1988 53 679 686 3286014 10.1016/0092-8674(88)90086-4
French S Consequences of replication fork movement through transcription units in vivo Science 1992 258 1362 1365 1455232
Mirkin EV Mirkin SM Mechanisms of transcription-replication collisions in bacteria Mol Cell Biol 2005 25 888 895 15657418 10.1128/MCB.25.3.888-895.2005
McLean MJ Wolfe KH Devine KM Base composition skews, replication orientation, and gene orientation in 12 prokaryote genomes J Mol Evol 1998 47 691 696 9847411
Guy L Roten CAH Genometric analyses of the organization of circular chromosomes: a universal pressure determines the direction of ribosomal RNA genes transcription relative to chromosome replication Gene 2004 340 45 52 15556293 10.1016/j.gene.2004.06.056
Frank AC Lobry JR Asymmetric substitution patterns: a review of possible underlying mutational or selective mechanisms Gene 1999 238 65 77 10570985 10.1016/S0378-1119(99)00297-8
Trifonov EN Translation framing code and frame-monitoring mechanism as suggested by the analysis of messenger-RNA and 16S ribosomal RNA nucleotide sequences J Mol Biol 1987 194 643 652 2443708 10.1016/0022-2836(87)90241-5
Szybalski W Kubinski H Sheldrick O Pyrimidine clusters on the transcribing strand of DNA and their possible role in the initiation of RNA synthesis Cold Spring Harb Symp Quant Biol 1966 31 123 127 4966069
Lao PJ Forsdyke DR Thermophilic bacteria strictly obey Szybalski's transcription direction rule and politely purine-load RNAs with both adenine and guanine Genome Res 2000 10 228 236 10673280 10.1101/gr.10.2.228
Francino MP Ochman H Strand asymmetries in DNA evolution Trends Genet 1997 13 240 245 9196330 10.1016/S0168-9525(97)01118-9
Lobry JR Asymmetric substitution patterns in the two DNA strands of bacteria Mol Biol Evol 1996 13 660 665 8676740
Grigoriev A Analyzing genomes with cumulative skew diagrams Nucleic Acids Res 1998 26 2286 2290 9580676 10.1093/nar/26.10.2286
Tillier ERM Collins RA The contributions of replication orientation, gene direction, and signal sequences to base-composition asymmetries in bacterial genomes J Mol Evol 2000 50 249 257 10754068
Nakagawa I Kurokawa K Yamashita A Nakata M Tomiyasu Y Okahashi N Kawabata S Yamazaki K Shiba T Yasunaga T Hayashi H Hattori M Hamada S Genome sequence of an M3 strain of Streptococcus pyogenes reveals a large-scale genomic rearrangement in invasive strains and new insights into phage evolution Genome Res 2003 13 1042 1055 12799345 10.1101/gr.1096703
Stover CK Pham XQ Erwin AL Mizoguchi SD Warrener P Hickey MJ Brinkman FS Hufnagle WO Kowalik DJ Lagrou M Garber RL Goltry L Tolentino E Westbrock-Wadman S Yuan Y Brody LL Coulter SN Folger KR Kas A Larbig K Lim R Smith K Spencer D Wong GK Wu Z Paulsen IT Reizer J Saier MH Hancock RE Lory S Olson MV Complete genome sequence of Pseudomonas aeruginosa PA01, an opportunistic pathogen Nature 2000 406 959 964 10984043 10.1038/35023079
Comparative Genometrics Database
Van Sluys MA de Oliveira MC Monteiro-Vitorello CB Miyaki CY Furlan LR Camargo LEA da Silva ACR Moon DH Takita MA Lemos EGM Machado MA Ferro MIT da Silva FR Goldman MHS Goldman GH Lemos MVF El-Dorry H Tsai SM Carrer H Carraro DM de Oliveira RC Nunes LR Siqueira WJ Coutinho LL Kimura ET Ferro ES Harakava R Kuramae EE Marino CL Giglioti E Abreu IL Alves LMC do Amaral AM Baia GS Blanco SR Brito MS Cannavan FS Celestino AV da Cunha AF Fenille RC Ferro JA Formighieri EF Kishi LT Leoni SG Oliveira AR Rosa VEJ Sassaki FT Sena JAD de Souza AA Truffi D Tsukumo F Yanai GM Zaros LG Civerolo EL Simpson AJG Almeida NFJ Setubal JC Kitajima JP Comparative analyses of the complete genome sequences of Pierce's disease and citrus variegated chlorosis strains of Xylella fastidiosa J Bacteriol 2003 185 1018 1026 12533478 10.1128/JB.185.3.1018-1026.2003
Canchaya C Fournous G Brussow H The impact of prophages on bacterial chromosomes Molecular Microbiology 2004 53 9 18 15225299 10.1111/j.1365-2958.2004.04113.x
Schell MA Karmirantzou M Snel B Vilanova D Berger B Pessi G Zwahlen MC Desiere F Bork P Delley M Pridmore RD Arigoni F Correction for Schell et al., The genome sequence of Bifidobacterium longum reflects its adaptation to the human gastrointestinal tract, PNAS 2002 99:14422-14427 Proc Natl Acad Sci U S A 2005 102 9430
Roten CA Gamba P Barblan JL Karamata D Comparative Genometrics (CG): a database dedicated to biometric comparisons of whole genomes Nucleic Acids Res 2002 30 142 144 11752276 10.1093/nar/30.1.142
Messer W The bacterial replication initiator DnaA. DnaA and oriC, the bacterial mode to initiate DNA replication Fems Microbiol Rev 2002 26 355 374 12413665 10.1016/S0168-6445(02)00127-4
IS Finder
Greub G Collyn F Guy L Roten CA A genomic island present along the bacterial chromosome of the Parachlamydiaceae UWE25, an obligate amoebal endosymbiont, encodes a potentially functional F-like conjugative DNA transfer system BMC Microbiol 2004 4 48 15615594 10.1186/1471-2180-4-48
Louarn JM Louarn J Francois V Patte J Analysis and possible role of hyperrecombination in the termination region of the Escherichia coli chromosome Journal of Bacteriology 1991 173 5097 5104 1650344
Bierne H Michel B When replication forks stop Mol Microbiol 1994 13 17 23 7984091
Elisabeth R. M. Tillier Website
Parkhill J Wren BW Thomson NR Titball RW Holden MTG Prentice MB Sebaihia M James KD Churcher C Mungall KL Baker S Basham D Bentley SD Brooks K Cerdeno-Tarraga AM Chillingworth T Cronin A Davies RM Davis P Dougan G Feltwell T Hamlin N Holroyd S Jagels K Karlyshev AV Leather S Moule S Oyston PCF Quail M Rutherford K Simmonds M Skelton J Stevens K Whitehead S Barrell BG Genome sequence of Yersinia pestis, the causative agent of plague Nature 2001 413 523 527 11586360 10.1038/35097083
Deng W Burland V Plunkett III G Boutin A Mayhew GF Liss P Perna NT Rose DJ Mau B Zhou S Schwartz DC Fetherston JD Lindler LE Brubaker RR Plano GV Straley SC McDonough KA Nilles ML Matson JS Blattner FR Perry RD Genome sequence of Yersinia pestis KIM J Bacteriol 2002 184 4601 4611 12142430 10.1128/JB.184.16.4601-4611.2002
Wren BW The yersiniae - a model genus to study the rapid evolution of bacterial pathogens Nat Rev Micro 2003 1 55 64 10.1038/nrmicro730
Chain PSG Carniel E Larimer FW Lamerdin J Stoutland PO Regala WM Georgescu AM Vergez LM Land ML Motin VL Brubaker RR Fowler J Hinnebusch J Marceau M Medigue C Simonet M Chenal-Francisque V Souza B Dacheux D Elliott JM Derbise A Hauser LJ Garcia E Insights into the evolution of Yersinia pestis through whole-genome comparison with Yersinia pseudotuberculosis Proc Natl Acad Sci U S A 2004 101 13826 13831 15358858 10.1073/pnas.0404012101
National Center for Biotechnology Information Website
Wheeler DL Church DM Lash AE Leipe DD Madden TL Pontius JU Schuler GD Schriml LM Tatusova TA Wagner L Rapp BA Database resources of the National Center for Biotechnology Information Nucleic Acids Res 2001 29 11 16 11125038 10.1093/nar/29.1.11
Comparative Genometrics - Bifidobacterium longum NCC2705 Webpage
Cebrat S Dudek MR Gierlik A Kowalczuk M Mackiewicz P Effect of replication on the third base of codons Physica A 1999 265 78 84 10.1016/S0378-4371(98)00657-8
Kano-Sueoka T Lobry JR Sueoka N Intra-strand biases in bacteriophage T4 genome Gene 1999 238 59 64 10570984 10.1016/S0378-1119(99)00296-6
Genometrician's company
Altschul SF Madden TL Schaffer AA Zhang JH 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
Myllykallio H Lopez P Lopez-Garcia P Heilig R Saurin W Zivanovic Y Philippe H Forterre P Bacterial mode of replication with eukaryotic-like machinery in a hyperthermophilic archaeon Science 2000 288 2212 2215 10864870 10.1126/science.288.5474.2212
|
16223444
|
PMC1285363
|
CC BY
|
2021-01-04 16:03:40
|
no
|
BMC Microbiol. 2005 Oct 13; 5:60
|
utf-8
|
BMC Microbiol
| 2,005 |
10.1186/1471-2180-5-60
|
oa_comm
|
==== Front
BMC PhysiolBMC Physiology1472-6793BioMed Central London 1472-6793-5-161626643510.1186/1472-6793-5-16Research ArticleAbsence of force suppression in rabbit bladder correlates with low expression of heat shock protein 20 Batts Timothy W [email protected] John S [email protected] Richard A [email protected] Christopher M [email protected] Cardiovascular Division, Department of Internal Medicine, University of Virginia, Charlottesville, Virginia 22908 USA2 Department of Molecular Physiology and Cellular Biophysics, University of Virginia, Charlottesville, Virginia 22908 USA2005 2 11 2005 5 16 16 4 5 2005 2 11 2005 Copyright © 2005 Batts 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
Nitroglycerin can induce relaxation of swine carotid artery without sustained reductions in [Ca2+]i or myosin regulatory light chain (MRLC) phosphorylation. This has been termed force suppression and been found to correlate with ser16-phosphorylation of heat shock protein 20 (HSP20). We tested for the existence of this mechanism in a smooth muscle that is not responsive to nitric oxide.
Methods
Isometrically mounted mucosa free rabbit bladder strips were contracted with carbachol and relaxed with 8-Br-cGMP, forskolin, or isoprenaline.
Results
Contraction was associated with a highly cooperative relation between MRLC phosphorylation and force such that very small increases in MRLC phosphorylation induced large increases in force. Relaxation induced by 8-Br-cGMP, forskolin, or isoprenaline did not shift the MRLC phosphorylation-force relation from that observed with carbachol alone, i.e. there was no force suppression. HSP20 content was negligible (approximately two hundred-fold less than swine carotid).
Conclusion
The lack of force suppression in the absence of HSP20 is consistent with the hypothesized role for HSP20 in the force suppression observed in tonic smooth muscles.
==== Body
Background
Ca2+-dependent phosphorylation of the myosin regulatory light chain (MRLC) is accepted as the primary mechanism regulating contraction of smooth muscle in response to excitatory stimuli [1,2]. However, this myosin-linked phosphorylation mechanism does not fully explain all observations. Of particular physiological interest is relaxation induced in activated tissues by NO or NO-donors that increase [cGMP]. In both vascular smooth muscle [3] and corpus cavernosum [4,5], there are two mechanism for NO donors to reduce tone. 1) NO donors can reduce in myoplasmic [Ca2+], which decreases MRLC phosphorylation, a process termed "deactivation." Deactivation is demonstrated when a decline in force is associated with a dependence of force on MRLC phosphorylation similar to that observed with contractile agonists alone. 2) NO donors can also reduce force without reductions in myoplasmic [Ca2+] or MRLC phosphorylation, a mechanism termed "force suppression" [6]. Force suppression is demonstrated when a decline in force is associated with a rightward shift in the dependence of force on MRLC phosphorylation similar to that observed with contractile agonists alone [6]. We hypothesized that a thin filament mechanism, specifically ser16 phosphorylation of heat shock protein was the mediator of force suppression [7].
The human umbilical artery was found to not express HSP20 and not exhibit NO dependent relaxation [8], suggesting a possible linkage between HSP20 and NO dependent relaxation. The rabbit bladder is a phasic urogenital smooth muscle that does not relax in response to NO. We therefore hypothesized that the rabbit bladder could be another test of the role of HSP20 in force suppression. We therefore tested 1) whether HSP20 is expressed in rabbit bladder and 2) whether force suppression occurs with relaxing agents in the rabbit bladder.
Methods
Tissues
Male New Zealand white rabbits (Burleson Enterprises, Inc., 2.3 – 2.7 kg) were euthanized by halothane inhalation according to an IACUC approved protocol. The urinary bladder was isolated at 4°C in a bicarbonate-buffered Krebs solution containing (in mM) 118.0 NaCl, 4.75 KCl, 24.80 NaHCO3, 1.18 KH2PO4, 1.27 CaCl2, 1.18 MgSO4·7H2O, and 10.0 D-glucose saturated with 95% O2 and 5% CO2. An incision was made from the bladder neck up to the dome following either the dorsal or ventral vasculature. The bladder was pinned out with the mucosa facing down. This protocol caused ridges to form from which strips were dissected from the abluminal surface. Histological examination showed that the cells were aligned in the longitudinal axis of the preparation in which length and force were measured (not illustrated). The bladder strips were tied to the two posts on the apparatus using silk sutures; one post to a micrometer to change length, and the other to a FT0.3 Grass force transducer. The length was incrementally increased until a constant force of 1 g was maintained, approximating Lo. The preparations responded with sustained contractions when exposed to 3 μM carbachol. K+-depolarization elicited transient contractions diagnostic of a phasic smooth muscle [9]. Tissues exhibiting spontaneous oscillatory activity were excluded from the analysis. A total of 45 tissues were included in the analysis.
MRLC and HSP20 phosphorylation
Bladder strips treated pharmacologically and then frozen in 20 ml of acetone cooled with 20 ml crushed dry ice. They were then slowly (2.5 hours) thawed to room temperature to dehydrate the tissues, air dried and weighed. The dry samples were homogenized in ground glass tissue homogenizers on ice in 1% (w/v) sodium dodecyl sulfate (SDS), 10% (v/v) glycerol, 0.1% pefabloc (a protease inhibitor), 0.1 % microcystin, and 30 mM dithiothreitol (0.22 ml/mg tissue dry weight), and then centrifuged at 14,000 × g for 10 min. Trichorloacetic acid was not included since it did not alter MRLC phosphorylation estimates. Serial dilutions (1/2, 1/4, 1/8, 1/16 and 1/32) of homogenates in homogenization buffer were loaded onto 12% acrylamide/glycerol-urea slab gels for isoelectric focusing at 250 volts overnight on a pH 4.0–6.5 gradient for MRLC [10] and a pH 4.5–8.0 gradient for HSP20 [6]. Gels were focused at 250 V constant voltage for 12 h at 8°C. Separated proteins were transferred to a nitrocellulose membrane by electroblotting in Towbin's transfer buffer (25 mM Tris, 192 mM glycine, 20% methanol, 0.1% SDS) at 200 mA constant current for 2 h at 8°C. The membranes were first washed in a 0.1% Tris-buffered saline-Tween solution (TBST: 10 mM Tris, 0.05% NaCl, 0.1% Tween-20). The membranes were then blocked overnight in TBST containing 1% bovine serum albumin and 0.01% sodium azide. After rinsing in TBST, the membranes were incubated in either 1:2000 anti-MRLC antibody (20 kD MRLC from Sigma) or 1:1000 rabbit anti-HSP20 (made by the authors) antibody for 1 h. After rinsing in TBST, the membranes were incubated with a horseradish peroxidase conjugate secondary (1:15000) for 1 h. After rinsing twice with TBST and once with TBS (TBST without Tween-20), antibodies were detected with enhanced chemiluminescence was scanned using a Molecular Dynamics laser densitometer and analyzed by NIH Image software. The relative protein content was estimated assuming that antibody binding was the same for phospho- and dephospho-MRLC and corrections made for offset and saturation errors as described [10].
Results
Carbachol (0.3 μM) alone induced a sustained contraction that measured 56 ± 9 % of a maximal (3 μM) carbachol contraction. Therefore, 0.3 μM carbachol was used to test relaxing agents. Preliminary experiments showed that 0.3 μM carbachol stimulated rabbit bladder did not relax with 100 μM nitroprusside (data not shown). Carbachol (0.3 μM) stimulated rabbit bladder relaxed when treated with 8-bromo-cGMP (a cell permeant cGMP analog – force was 30 ± 4 % of a 3 μM carbachol contraction), forskolin (a non receptor activator of adenyl cyclase – force was 12 ± 5 %), or isoprenaline (a non receptor activator of adenyl cyclase – force was 18 ± 8 %).
The open circles in Fig. 1 shows the steep dependence of force on MRLC phosphorylation when rabbit urinary bladder was activated with 0.01–100 μM carbachol (these data were previously published [11]). The filled symbols in Fig. 1 show the dependence of force on MRLC phosphorylation when 0.3 μM carbachol stimulated rabbit bladder were relaxed with 8-bromo-cGMP (filled circle), forskolin (square), or isoprenaline (triangle). These three treatments induced a dependence of force on MRLC phosphorylation that did not differ from that observed with carbachol alone (open circles), suggesting the relaxation occurred by deactivation. If there had been force suppression, there would have been a rightward shift in the dependence of force on MRLC phosphorylation.
HSP20 immunostaining was low in rabbit bladder (Fig. 2, lanes 2–7) compared to swine carotid (Fig. 2, left lane 1). Overall, there was 70.5 fold less HSP20 immunostaining in the bladder homogenates compared to swine carotid homogenates. Since this comparison was normalized on tissue dry weight, we also compared MRLC immunostaining from the same samples. There was 2.9 fold more MRLC immunostaining in the bladder homogenates compared to swine carotid homogenates (data not shown). When HSP20 immunostaining was normalized by MRLC immunostaining, there was 204 fold less HSP20 immunostaining in the bladder compared to the swine carotid immunostaining.
Discussion
These data suggest that rabbit bladder does not express significant levels of HSP20 (Fig. 2) and does not exhibit force suppression (Fig. 1). This result is consistent with the hypothesis that HSP20 is the mediator of force suppression [7]. Our results are reminiscent of data showing that umbilical vein does not express significant levels of HSP20 and does not relax to NO donors [8].
We think it most likely that the low level of HSP20 immunostaining is caused by HSP20 present in the vascular smooth muscle of the bladder vasculature. However, we cannot rule out a low level of HSP20 expression in bladder smooth muscle. We assumed that carotid and bladder have similar cellular MRLC concentration (13) so that MRLC immunostaining could be used to normalize HSP20 immunostaining. The accuracy of this assumption is not that crucial given the relatively small amount of HSP20 immunostaining in the bladder (approximately two hundred-fold less).
We confirmed that the rabbit bladder does not relax to NO donors. However, there was a relaxation to 8-bromo-cGMP, a cell permeant cGMP analog, suggesting that the lack of response to NO donors resides in the generation of cGMP, rather than a defect in the response to cGMP. Non-receptor activators of adenyl cyclase (forskolin and isoprenaline) also induced relaxation, suggesting that rabbit bladder can relax to increases in cAMP. Importantly, these relaxations cGMP and cAMP mediated relaxations were not associated with force suppression (which would have produced a rightward shift in the dependence of force on MRLC phosphorylation). This is consistent with the hypothesis that HSP20 is involved in force suppression. The relaxations in response to cGMP analogs and agents that increase [cAMP] suggest that the relaxation is caused by deactivation mechanisms such as mechanisms that reduce [Ca2+]i or possibly increase myosin phosphatase activity [12,13].
Conclusion
These results suggest that some phasic smooth muscles, such as rabbit bladder, do not exhibit force suppression. Force suppression was found in corpus cavernosum [4,5], a NO responsive phasic smooth muscle. Unlike the human umbilical vein study [8], we report that the lack of HSP20 is not only associated with a lack in relaxation, but also a lack of force suppression.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
TB and JW performed the studies. RM and CR conceived the study. TB and CR drafted the study.
Acknowledgements
We thank Gwaltney of Smithfield (Smithfield, VA) for contribution of swine carotid arteries. Robin Woodson provided technical assistance. Grants from the NIH (and DK56034, HL71191, and HL07284) and the Mid-Atlantic AHA (0151586U) supported this research.
Figures and Tables
Figure 1 The steady-state relationship between active force and percent MRLC phosphorylation in the rabbit urinary bladder detrusor strips. Force is expressed as a percent of the initial response to 3 μM carbachol. The open circles show results from tissues were stimulated by varying doses of carbachol (0.01, 0.03, 0.1, 0.3, 1.0, 3.0, 10, 30, and 100 μM) until steady state was achieved (the data with carbachol alone previously published [11]). The symbol at 56% of force represents tissues stimulated with 0.3 μM carbachol. The filled symbols show results from tissues were stimulated 0.3 μM carbachol until steady state was achieved followed by addition of 300 μM 8-bromo-cGMP (filled circle), 10 μM forskolin (filled square), or 0.3 μM isoprenaline (filled triangle) until steady state was achieved.
Figure 2 Representative HSP20 immunoblot of homogenates from a swine carotid tissue (lane 1, 15 μl loaded) and six different rabbit bladder tissues (lanes 2–7, 30 μl loaded). The bands have been identified in swine carotid as unphosphorylated (labeled U), monophosphorylated at serine 157 (labeled S157), monophosphorylated at serine 16 (labeled S16), diphosphorylated at serine 16 and 157 (labeled D), There was significantly less HSP20 immunostaining in the bladder compared to swine carotid, suggesting that the HSP20 in bladder could come from contaminating vasculature.
==== Refs
Horowitz A Menice CB Laporte R Morgan KG Mechanisms of smooth muscle contraction Physiol Rev 1996 76 967 1003 8874491
Hai CM Murphy RA Ca2+, crossbridge phosphorylation, and contraction Annu Rev Physiol 1989 51 285 298 2653183
McDaniel NL Chen XL Singer HA Murphy RA Rembold CM Nitrovasodilators relax arterial smooth muscle by decreasing [Ca2+]i, [Ca2+]i sensitivity, and uncoupling stress from myosin phosphorylation Am J Physiol Cell Physiol 1992 263 C461 C467
Chuang AT Strauss JD Steers WD Murphy RA cGMP mediates corpus cavernosum smooth muscle relaxation with altered cross-bridge function Life Sci 1998 63 185 194 9698048 10.1016/S0024-3205(98)00259-8
Chuang AT Strauss JD Murphy RA Steers WD Sildenafil, a type-5 CGMP phosphodiesterase inhibitor, specifically amplifies endogenous cGMP-dependent relaxation in rabbit corpus cavernosum smooth muscle in vitro J Urol 1998 160 257 261 9628660 10.1097/00005392-199807000-00094
Rembold CM O'Connor MJ Clarkson M Wardle RL Murphy RA HSP20 phosphorylation in nitroglycerin- and forskolin-induced sustained reductions in swine carotid media tone J Appl Physiol 2001 91 1460 1466 11509549
Rembold CM Foster B Strauss JD Wingard CJ Van Eyk JE cGMP mediated phosphorylation of heat shock protein 20 may cause smooth muscle relaxation without myosin light chain dephosphorylation J Physiol (Lond ) 2000 524 865 878 10790164 10.1111/j.1469-7793.2000.00865.x
Brophy CM Beall A Lamb S Dickinson M Ware DJ Small heat shock proteins and vasospasm in human umbilical artery smooth muscle Biol Reprod 1997 1354 1359 9408240 10.1095/biolreprod57.6.1354
Somlyo AP Somlyo AV Vascular smooth muscle. I. Normal structure, pathology, biochemistry, and biophysics Pharmacol Rev 1968 20 197 272 4888390
Walker JS Walker LA Etter EF Murphy RA A dilution immunoassay to measure myosin regulatory light chain phosphorylation Analytical Biochem 2000 284 173 182 10.1006/abio.2000.4704
Rembold CM Wardle RL Wingard CJ Batts TW Etter EF Murphy RA Cooperative attachment of cross bridges predicts regulation of smooth muscle force by myosin phosphorylation American Journal of Physiology-Cell Physiology 2004 287 C594 C602 15151901 10.1152/ajpcell.00082.2004
Rembold CM Barany M Electromechanical and Pharmacomechanical Coupling Biochemistry of Smooth Muscle Contraction 1996 Chicago, Academic Press 227 239
Etter EF Eto M Wardle RL Brautigan DL Murphy RA Activation of myosin light chain phosphatase in intact arterial smooth Muscle during nitric oxide-induced relaxation J Biol Chem 2001 276 34681 34685 11461918 10.1074/jbc.M104737200
|
16266435
|
PMC1285364
|
CC BY
|
2021-01-04 16:03:51
|
no
|
BMC Physiol. 2005 Nov 2; 5:16
|
utf-8
|
BMC Physiol
| 2,005 |
10.1186/1472-6793-5-16
|
oa_comm
|
==== Front
BMC Plant BiolBMC Plant Biology1471-2229BioMed Central London 1471-2229-5-201619427310.1186/1471-2229-5-20Research ArticleGenetic chimerism of Vitis vinifera cv. Chardonnay 96 is maintained through organogenesis but not somatic embryogenesis Bertsch Christophe [email protected] Flore [email protected] Pascale [email protected] Sibylle [email protected] Gisèle [email protected] Didier [email protected] Bernard [email protected] Université de Haute-Alsace, Laboratoire Vigne Biotechnologies et Environnement, 33, rue de Herrlisheim 68000 Colmar, France2 INRA, UMR 1131, Vigne et Vin d'Alsace, 28, rue de Herrlisheim 68000 Colmar, France2005 29 9 2005 5 20 20 25 4 2005 29 9 2005 Copyright © 2005 Bertsch 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
Grapevine can be a periclinal chimera plant which is composed at least of two distinct cell layers (L1, L2). When the cell layers of this plant are separated by passage through somatic embryogenesis, regenerated plants could show distinct DNA profiles and a novel phenotype which proved different from that of the parent plant.
Results
Genetically Chardonnay clone 96 is a periclinal chimera plant in which is L1 and L2 cell layers are distinct. Plants obtained via organogenesis through meristematic bulks are shown to be composed of both cell layers. However, plants regenerated through somatic embryogenesis starting from anthers or nodal explants are composed only of L1 cells. These somaclones do not show phenotypic differences to the parental clone up to three years after regeneration. Interestingly, the only somaclone showing an atypical phenotype (asymmetric leave) shows a genotypic modification.
Conclusion
These results suggest that the phenotype of Chardonnay 96 does not result from an interaction between the two distinct cell layers L1 and L2. If phenotype conformity is further confirmed, somatic embryogenesis will result in true-to-type somaclones of Chardonnay 96 and would be well suitable for gene transfer.
==== Body
Background
Improvement of grapevine rootstocks and scion varieties can be achieved through either inter-specific hybridization or clonal selection. Grapevine clonal selection consists in choosing in a variety one plant presenting desired characteristics. This selected plant is further propagated by vegetative multiplication known to maintain trueness-to-type. But clonal selection is restricted to the natural variability of a given cultivar, within the limits of the characteristics on which the trueness-to-type is based. Somatic embryogenesis could be an additional possibility for varietal improvement. A broader perspective of improvement of grapevine cultivars for characteristics such as resistance to pests and diseases has been opened by transgenic technologies which in most cases are based on somatic embryogenesis of grapevine [1] and shoot organogenesis [2]. One question to be addressed is the phenotypic and genotypic variability of grapevine clones raised through somatic embryogenesis or shoot organogenesis. Recently, Desperriers et al. [3] presented results of a ten years observation of Vitis vinifera Gamay somaclones which showed variations in fertility as well as sugar content, size and level of maturity of the grapes. These observations illustrate that somaclones can differ from the original parent without changing the fundamental typicity of the wine. One possible origin of variability may be the separation of cell layers of the mother plant from which clones are grown.
The grapevine meristem is considered to be composed at least of two distinct cell layers L1 and L2 [4], which can produce a chimeric tissue structure. For example, Vitis vinifera cv. Pinot Meunier phenotype is due to the interaction of genetically distinct cell layers. When the cell layers of Pinot Meunier periclinal chimera were separated by passage through somatic embryogenesis, regenerated plants showed distinct DNA profiles which proved to be different from that of the parent plant. Regenerated somaclones also showed a novel phenotype [5].
About five hundred microsatellite markers of the grapevine genome are now available and widely as well as very efficiently used for identification of cultivars [6-8]. Riaz et al. [9] used microsatellite markers for the detection of reproducible intra-cultivar polymorphism in Vitis vinifera Chardonnay and Pinot noir. Furthermore, some of these microsatellite markers made it possible to differentiate the two cell layers L1 and L2 in some clones of both cultivars [9,10].
A previous paper [11], showed that in the leaf tissue of Chardonnay 96 the microsatellite marker VMC 5g7 revealed two standard alleles (198:220 bp) and a variant allele (222 bp) previously defined by Riaz et al. [9]. Wood tissues and roots only presented the two standard alleles with VMC 5g7. With a second microsatellite marker (VMC 6g10), two standard alleles (114:140 bp) and a mutant allele (142 bp) were detected in leaves. This mutant allele replaced one of the standard alleles in woods and roots (114:142 bp). In addition, we showed, that somaclones regenerated from anthers of a single inflorescence all derived from L1 cells exclusively.
In the present paper we report results of a wider genotypic analysis of somaclones regenerated either from anthers of different inflorescences or from nodal explants of Chardonnay 96. The same microsatellite markers were also used to compare the genotype of different clones obtained through shoot organogenesis. The genotypic identities (L1; L2) of the regenerated clones are in accordance with their respective phenotypic characteristics.
Results
Genotypic analysis of various tissues of Chardonnay 96 mother clone
Three alleles were detected in leaves of Chardonnay 96 mother clone with the microsatellite marker VMC 6c10: the 114 bp and 140 bp standard alleles and an additional 142 bp variant allele (table 1). The standard alleles were defined by Riaz et al. [9] as the most frequently detected alleles in different clones of the same cultivar. VMC 5g7 revealed in leaves the two 198 bp and 220 bp standard alleles and a 222 bp variant allele (table 1). The same genotype was observed with leaves from in vitro or greenhouse-grown plants. In the DNA extracted from rootlets and wood tissues, only the two pairs of alleles 114:142 and 198:220 were detected with VMC 6c10 and VMC 5g7 respectively (table 1). In berry skin, the triallelic profiles were detected with both markers. It is known that wood and roots are composed of L2 cells only and that leaves comprise L1 and L2 cell layers. Our results lead to the conclusion that the VMC 6c10 114 bp and 142 bp alleles as well as the VMC 5g7 198 bp and 220 bp alleles are present in L2 cells of Chardonnay 96. We can also deduce that L1 cells have the VMC 6c10 114 bp and 140 bp alleles and the VMC 5g7 the 198 bp and 222 bp alleles.
Regeneration of plants through somatic embryogenesis or shoot organogenesis
Four different inflorescences of Chardonnay 96 were used to collect anthers. Primary somatic embryos were obtained from anther-derived embryogenic calli after a 2 month period (figure 1A). Secondary embryos were obtained 1 month after initiation of embryogenic calli from primary embryos. Embryogenic calli and primary embryos from nodal explants were obtained after 2–4 months (figure 1B). Meristematic bulks developed after successive transfers of shoots on IM medium with benzyladenine concentrations increasing up to 13.2 μM. High numbers of adventitious shoots regenerated from slices cut from the meristematic bulk tissues, and rooted plantlets further developed (figure 1C).
Efficient plant growth was further obtained with a number of somaclones. The plants did not show any atypical phenotype in visual comparison with vegetatively propagated Chardonnay 96 grown in the growth chamber and the greenhouse. Except one somaclone (n° 21) originated from an anther-derived callus.
Genotypic analysis of Chardonnay 96 somaclones obtained from anthers and from nodal explants
The same genotypic profiles were obtained with a total of 29 primary and 7 secondary somaclones regenerated from anthers collected from 4 different inflorescences and 16 clones from nodal explants (table 2). In leaves from in vitro and greenhouse-grown plants, the standard genotype (114:140) was visualised with VMC 6c10. For VMC 5g7 in addition to the standard 198 bp allele, the 222 bp variant allele was detected in leaves from all somaclones (table 2). In the rootlets, the standard diallelic genotype was visualised with VMC 6c10. With VMC 5g7 the 198 bp standard allele and the 222 bp variant allele were detected (table 2). These results show that all somaclones regenerated from L1 cells of the Chardonnay 96 explants.
Phenotypic and genotypic analysis of somaclone n°21
Compared to the Chardonnay 96 mother clone, which presents mature circular leaves undivided or with five lobes, all leaves from somaclone n° 21 show one circular undivided half and the other half with separated lobes (figure 2). This phenotype was observed after the acclimatization and is still observed after 3 years on greenhouse-grown plants only for this somaclone.
In leaves from in vitro and greenhouse-grown somaclone 21, the triallelic profile was detected with VMC 6c10. For VMC 5g7 in addition to the standard 198 bp allele, the 222 bp variant allele was detected. The same results were obtained with berry skin. In the rootlets, the standard allele (114) and the mutant allele (142) were visualised with VMC 6c10 and with VMC 5g7 the standard allele (198) and the mutant allele (222) were detected (table 3). These results indicate that the genotype of somaclone 21 is the same as that of the mother clone for VMC 6c10, but for VMC 5g7 only 2 alleles present in the L1 cell layer of the mother plant are present. A mutation in the microsatellite sequence VMC 6c10 from the L1 cell type of the mother clone probably occurred during embryogenesis of clone 21.
Genotypic analysis of clones obtained through shoot organogenesis
Leaves from all 7 clones obtained through shoot organogenesis show the triallelic profile both for VMC 6c10 and VMC 5g7 (table 2). This suggests that plants obtained via organogenesis through meristematic bulks are derived from both L1 and L2 cell layers of the Chardonnay 96 tissues from which they grew.
Discussion
Vitis vinifera Chardonnay clone 96 is a periclinal chimera
In grapevine, apical meristems are composed of two or more cell layers forming the tunica in addition to a corpus [4,12]. In the leaf tissue from Chardonnay 96, which is derived from both the outer tunica layer L1 and the inner cell layer L2, the microsatellite marker VMC 5g7 revealed the two standard alleles and a variant allele previously defined by Riaz et al. [9]. Wood tissues and roots, which originate exclusively from the L2 layer, presented only the two standard alleles. The presence of a third allele in leaf suggests that Chardonnay is a periclinal chimera in which a mutant allele is present only in the L1 layer, as described by Riaz et al. [9]. For the microsatellite marker VMC 6g10, a mutant allele was detected in leaves, wood tissue and roots. This mutant allele replaced one of the standard alleles in woods and roots, whereas in leaves the mutant allele was present simultaneously with the two standard alleles [11]. These results suggest that the 2 bp-mutation resulting in the replacement of the VMC 6c10 140 bp allele by the 142 bp allele only occurred in L2. Riaz et al. [9] propose that the mutation most likely occurred in an L1 or L2 cell and then came to populate both layers of the meristem rather than two identical mutations appearing independently in the L1 and L2. Results reported by Riaz et al. [9] were based on visual interpretation of electrophoretic profiles of amplification products. We suggest that such observations could be biased by the DNA polymerase slippage. Our analyses with ABI PRISM allowed to differentiate between an allele and the different stutter bands, without ambiguity.
Somaclones develope only from L1 cells of Chardonnay 96
In the present study, we show that somaclones obtained not only from anthers from different inflorescences but also from nodal explants all derived from L1 cells. Embryogenic calli were composed exclusively of cells showing the genetic profile of L1 cells of the mother clone Chardonnay 96 (data not shown), suggesting that L2 cells could not multiply into callus at least in our culture conditions. On another hand, clones raised through shoot organogenesis are composed of cells showing the genetic profiles of both L1 and L2 cells of the mother clone. These observations suggest that the genetic chimerism of Chardonnay 96 is maintained through shoot organogenesis but not through somatic embryogenesis.
Anthers are the most widely used organ for initiation of grapevine somatic embryos. It has been shown that embryogenic cells derive from the anther filament. Both L1 and L2 cell layers seem to be competent to form embryogenic calli in some conditions, as reported for Pinot Meunier by Franks et al. [9]. Filaments from anthers of Chardonnay 96 are composed of L1 and L2 cells (data not shown). But, in our conditions only L1 cells of Chardonnay 96 developed into embryos. A similar result was reported recently with Pinot gris from which only the L1 cell layer is competent to form embryogenic callus [10].
Phenotype variation of Chardonnay 96 somaclones
Embryogenesis might generate new grapevine phenotypes when the mother plant is a chimera of genetically distinct cell layers. For example, the separation of chimeric cell layers of Pinot Meunier through somatic embryogenesis generated plants that had distinct DNA profiles and had novel phenotypes which were different from those of the parent plants [5]. The phenotype of all the somaclones we obtained from anthers has been observed for 2–3 years in a glasshouse: no visible modification was noticed (except for the somaclone n°21) in comparison to the mother clone Chardonnay 96. This phenotypic conformity suggests that the L2 genotype would not significantly participate in the phenotypic expression in the L1–L2 chimeric Chardonnay 96. More subtle variations in fruit setting and wine quality may be only detected in the future, when the somaclones will be grown in the vineyard. Variation of colour intensity, sugar content, size and level of maturity of the grapes has been reported for adult somaclones of Gamay after ten years observation [3]. Though no genotypic analyses were done for the different Gamay somaclones, it can be hypothesised that this somaclonal variation is of epigenetic origin [13]. The Chardonnay 96 mother clone shows the typical ampelographic traits of the cultivar Chardonnay with mature circular leaves undivided or with five lobes, petiolar sinus slightly open, often limited through veins at petiole end. Some adult leaves of somaclone 21 have an asymmetric shape, one half undivided or slightly lobed, the other half deeply lobed. Somaclone 21 probably arose from L1 cell(s) of Chardonnay 96 as all the other somaclones. We suggest that during an early cell division a mutation occurred in locus 6c10 which replaced the 140 bp allele by a 142 bp allele. The L2 cell layer of somaclone 21 originated from the diallelic mutated cell (114:142), whereas cells with the non mutated diallele (114:140) further multiplied giving the L1 cell layer of somaclone 21. This hypothesis is in accordance with the fact that the most frequently observed allelic size variation is the addition of one motif [14].
Conclusion
Regeneration via embryogenesis for Chardonnay 96 could result generally in a non chimeric plant and in an unchanged phenotype and would be well suited for gene transfer.
Methods
Plant material
Vitis vinifera cv. Chardonnay clone 96, was obtained from ENTAV (Etablissement National Technique pour l'Amélioration de la Viticulture, Le Grau du Roi, France), the national repository for registered grape clones in France. Forced adult plants were maintained in a growth chamber at 25 ± 0.5°C, 70 ± 10 % RH with a 16 h – photoperiod.
Initiation of embryogenic calli from anthers
Anthers were dissected and grown as described by Mauro et al. [15]. For long-term culture of embryogenic callus, subcultures were performed every three weeks on the MPM medium described by Perrin et al. [16]. All the cultures were maintained at 25 ± 0.5°C, 70 ± 10 % RH with a 16 h – photoperiod, except the very first step -the culture of detached anthers- which was performed in the dark.
For production of secondary embryos, primary embryos were cut off and transferred onto half strength Murashige and Skoog medium (MS medium) [17] supplemented with 20 g.l-1 sucrose, 0.7% Bacto-agar, 2.5 μM 2,4-dichlorophenoxyacetic acid (2,4-D) and 0.5 μM 6-benzylaminopurine (BA; N6-benzyladenine). Incubation was done at 25 ± 0.5°C in the dark, during 3 weeks. Then calli were subcultured every three weeks on MPM medium and maintained under the same conditions as previously described.
Initiation of embryogenic calli from nodal explants
Nodal explants were excised from plantlets grown in vitro and plated on half strength MS medium supplemented with 25 g.l-1 sucrose, 0.7 % Bacto-agar, 9 μM 2,4-D and 4.5 μM BA. Subculture was performed on half strength MS medium supplemented with 60 g.l-1 sucrose, 0.7 % Bacto-agar, 20 μM indole-3-acetic acid (IAA), 10 μM 2-naphtoxyacetic acid (NOA) and 1 μM BA. Embryogenic calli were obtained after several transfers onto this last medium. For long-term culture of embryogenic callus, subcultures were performed every three weeks on the same medium. All the cultures were maintained at 25 ± 0.5°C, 70 ± 10 % RH with a 16 h – photoperiod, except the callus initiation which was performed in the dark.
Regeneration and acclimatization of somaclones
Embryos were carefully excised and transferred onto half-strength MS medium containing 20 g.l-1 sucrose, 0.7 % agar and 0.4 μM BA. After two weeks at 25 ± 0.5°C and 16 h light, the growing embryos were individually transferred into tubes containing half-strength MS medium without plant growth factor. Plantlets with rootlets were transferred to soil and allowed to acclimatize in a growth chamber for about three weeks before transfer to a greenhouse.
Ampelographic observation
Conformity of the different 3 years old somaclones were based on visual observation of leaf morphology and phyllotaxy compared to Chardonnay clone 96.
Shoot organogenesis
In vitro propagation of V. vinifera Chardonnay 96 was initiated on MS basal medium supplemented with 4.4 μM BA, 30 g.l-1 sucrose and 0.7 % Bacto agar. Meristematic bulks were initiated on IM medium of Mezzetti et al. [2] supplemented with 0.05 μM NAA (α-naphtalene-acetic acid) and increasing concentrations of BA : 4.4 μM for the first 30 days subculture, 8.8 μM for the second subculture and finally 13.2 μM. At each transplantation, the apical dome was eliminated.
In order to induce shoot organogenesis, thin slices were cut from the inner part of the meristematic bulks and transferred onto IM medium containing 13.2 μM BA. Growing shoots were transferred to rooting medium of Quoirin and Lepoivre [18] containing 4.9 μM indole-3-butyric acid (IBA) and 5.7 μM IAA [2]. Growing plantlets were further propagated on hormone free MS medium with half strength macroelements.
DNA extraction
Leaves were harvested from plants regenerated through somatic embryogenesis or shoot organogenesis: these plants were in vitro and greenhouse-grown somaclones and in vitro-grown clones from organogenesis. Leaves from greenhouse-grown plants of Chardonnay 96 mother clone were used as a control. Rootlets were taken from in vitro-grown plantlets. Wood tissue was obtained from the dormant canes after the bark and cambium were first scraped away. Berry skin, removed using a scalpel, was also analysed. About 80 mg tissue was ground in liquid nitrogen using a grinder (Retsch MM200) and total DNA was extracted with the Qiagen Dneasy Plant mini-kit (Qiagen, Hilden, Germany) as described by the supplier.
Amplification of microsatellites and polymorphism detection
DNA was analysed using two pairs of primers flanking two different microsatellite regions: VMC6c10 and VMC 5g7 (Vitis Microsatellites Consortium, Dr. Rosa Arroyo Garcia and Dr. Kirsten Wolff), respectively marked with the fluorophores HEX and FAM. Amplification reactions were performed in a total volume of 50 μl with 10 ng of template DNA, 0.35 μM of forward primer labelled either with 6-FAM or HEX fluorophore, 0.35 μM of non-labelled reverse primer, 200 μM dNTP (Invitrogen), 1.5 mM MgCl2, 1X PCR Buffer and 0.2 unit Platinium® Taq DNA Polymerase (Invitrogen). The PCR was carried out using a GeneAmp® PCR System 2700 thermocycler (Applied Biosystems). The cycling program consisted of the following steps: 2 min at 94°C followed by 35 cycles of 40 s at 92°C, 1 min at 57°C and 1 min at 72°C and a final extension step of 7 min at 72°C. The amplification products were separated by capillary electrophoresis and detected with an ABI PRISM 310 Genetic Analyser (Applied Biosystem), using HD400-ROX as an internal size standard. The PCR fragments were detected with the GeneScan™ analysis software version 3.1 and the alleles were scored using the Genotyper™ DNA fragment analysis software version 2.5.2 (Applied Biosystems).
Authors' contributions
CB carried out the molecular studies, participated in the analyses of tissues and extract, and drafted the manuscript. FK and SF participated in the production of embryogenic calli from anthers. PM participated in the production of embryogenic calli from nodal explants. GM and DM partipated in the analyses of tissues. BW initiated shoot organogenesis and envisioned and supervised all the studies, as well as drafting the manuscript.
Acknowledgements
The authors are grateful to Dr. Rosa Arroyo Garcia and Dr. Kirsten Wolff for the VMC 6c10 and VMC 5g7 primer sequences.
Figures and Tables
Figure 1 Somatic embryogenesis and shoot organogenesis of Vitis vinifera Chardonnay 96. A) Embryogenic callus from anthers and plant regeneration, B) Embryogenic callus from nodal explants and plant regeneration, C) Shoot organogenesis from meristematic bulk.
Figure 2 Asymmetric leaf of a 3 year-old atypical somaclone n°21 plant.
Table 1 Genotypes of Chardonnay clone 96 at the VMC 5g7 and VMC 6c10 loci.
Locus alleles (bp)
DNA origin VMC 6c10 VMC 5g7 cell layer
Leaves 114:140:142 198:220:222 L1 + L2
Wood 114:142 198:220 L2
Rootlets 114:142 198:220 L2
Berry skin 114:140:142 198:220:222 L1 + L2
Table 2 Genotypes at the VMC 5g7 and VMC 6c10 loci of clones regenerated through embryogenesisor shoot organogenesis
Locus alleles (bp)
Plant origin DNA origin VMC 6c10 VMC 5g7 cell layer of origin
anthers
Inflorescence A
Primary (20) or secondary (6) somaclones Leaves iv + gh 114:140 198:222 L1
Rootlets iv 114:140 198:222 L1
Inflorescence B
Primary somaclones (3) Leaves iv 114:140 198:222 L1
Inflorescence C
Primary (5) or secondary (1) somaclones Leaves iv + gh 114:140 198:222 L1
Inflorescence D
Primary somaclone (1) Leaves iv + gh 114:140 198:222 L1
nodal explants
Primary somaclones (16)
Leaves iv 114:140 198:222 L1
Rootlets iv 114:140 198:222 L1
meristematic bulk
clones (7)
Leaves iv 114:140:142 198:220:222 L1 + L2
from:
– anthers from 4 different inflorescences (A, B, C, D)
– nodal explants
or through shoot organogenesis from meristematic bulks
(N°) : number of clones, iv : in vitro, gh : greenhouse
Table 3 Genotype of the somaclone 21 at the VMC 5g7 and VMC 6c10 loci
Locus alleles (bp)
DNA origin VMC 6c10 VMC 5g7
Leaves 114:140:142 198:222
Rootlets 114:142 198:222
Berry skin 114:140:142 198:222
==== Refs
Martinelli L Mandolino G Transgenic Grapes (Vitis species) Biotechnology in Agriculture and Forestry 2003 47 325 338
Mezzetti B Pandolfini T Navacchi O Landi L Genetic transformation of Vitis vinifera via organogenesis BMC Biotechnology 2002 2 1 10 11818033 10.1186/1472-6750-2-18
Desperrier JM Berger JL Bessis R Fournioux JC Labroche C Création clonale dirigée par embryogenèse somatique Bulletin de l'OIV 2003 76 871 872
Thompson NM Olmo HP Cytohistological studies of cytochimeric and tetraploid grapes Amer J Bot 1963 50 901 906
Franks T Botta R Thomas MR Chimerism in grapevines: implications for cultivar identity, ancestry and genetic improvement Theor Appl Genet 2002 104 192 199 12582686 10.1007/s001220100683
Sefc KM Lefort F Grando MS Scott K Steinkellner H Thomas MR Roubelakis-Angelakis KA Microsatellite markers for grapevine: a state of the art Molecular Biology and Biotechnology of Grapevine 2001 Kluwer Publishers, Amsterdam 433 463 ISBN 0-7923-6949-1
This P Jung A Boccacci P Borrego J Botta R Costantini L Crespan M Dangl GS Eisenheld C Ferreira-Monteiro F Grando S Ibanez J Lacombe T Laucou Magalhaes R Meredith CP Milani N Peterlunger E Regner F Zulini L Maul E Development of a standard set of microsatellite reference alleles for identification of grape cultivars Theor Appl Genet 2004 109 1448 1458 15565426 10.1007/s00122-004-1760-3
Merdinoglu D Butterlin G Bevilacqua L Chiquet V Adam-Blondon AF Decroocq S Development and characterization of a large set of microsatellite markers in grapevine (Vitis vinifera) suitable for multiplex PCR Molecular Breeding 2005
Riaz S Garrison KE Dangl GS Boursiquot JM Meredith CP Genetic divergence and chimerism within ancient asexually propagated wine grape cultivars J Amer Soc Hort Sci 2002 127 508 514
Hocquigny S Pelsy F Dumas V Kindt S Heloir MC Merdinoglu D Diversification within grapevine cultivars goes through chimeric states Genome 2004 47 579 589 15190375 10.1139/g04-006
Bertsch C Kieffer F Triouleyre C Butterlin G Merdinoglu D Walter B Molecular profiling of Vitis vinifera Chardonnay obtained by somatic embryogenesis J Int Sci Vigne Vin 2003 34 223 227
Morrison J Bud development in Vitis vinifera L Bot Gaz 1991 152 304 315 10.1086/337894
Henry RJ Jain SM, Brar DS, Ahloowalia BS Molecular and biochemical characterization of somaclonal variation Somaclonal variation and induced mutations in crop improvement 1998 Kluwer Academic Publishers, Dordrecht, Boston, London 485 499
Hocquigny S La diversité génétique intra-variétale chez la vigne: caractérisation et origines PhD thesis 2003 Université Louis Pasteur, Strasbourg 138
Mauro M Nef C Fallot J Stimulation of somatic embryogenesis and plant regeneration from anther culture of Vitis vinifera cv. Cabernet-Sauvignon Plant Cell Reports 1986 5 377 380 10.1007/BF00268606
Perrin M Martin D Joly D Demangeat G This P Masson JE Medium-dependent response of grapevine somatic embryogenic cells Plant Science 2001 161 107 116 10.1016/S0168-9452(01)00385-5
Murashige T Skoog F A revised medium for rapid growth and bioassays with tobacco tissue cultures Physiol Plant 1962 15 473 497
Quoirin M Lepoivre P Improved media from in vitro culture of Prunus spp Acta Horticulturae 1997 78 437 442
|
16194273
|
PMC1285365
|
CC BY
|
2021-01-04 16:03:52
|
no
|
BMC Plant Biol. 2005 Sep 29; 5:20
|
utf-8
|
BMC Plant Biol
| 2,005 |
10.1186/1471-2229-5-20
|
oa_comm
|
==== Front
Cerebrospinal Fluid ResCerebrospinal Fluid Research1743-8454BioMed Central London 1743-8454-2-91622344810.1186/1743-8454-2-9ResearchCharacterization of cytoskeletal and junctional proteins expressed by cells cultured from human arachnoid granulation tissue Holman David W [email protected] Deborah M [email protected] Bhavya C [email protected] Steven E [email protected] Martin [email protected] Biomedical Engineering Center, The Ohio State University, 260 Bevis Hall, 1080 Carmack Rd, Columbus, OH 43210, USA2 Neuroophthalmic Research Group, Department of Ophthalmology, The Ohio State University, Cramblett Hall 5A, 456 W. 10th Ave., Columbus, Ohio 43210, USA3 Department of Chemical and Biomolecular Engineering, The Ohio State University, 125A Koffolt Laboratories, 140 W. 19th Ave., Columbus, OH 43210, USA2005 13 10 2005 2 9 9 10 6 2005 13 10 2005 Copyright © 2005 Holman et al; licensee BioMed Central Ltd.2005Holman 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 arachnoid granulations (AGs) are projections of the arachnoid membrane into the dural venous sinuses. They function, along with the extracranial lymphatics, to circulate the cerebrospinal fluid (CSF) to the systemic venous circulation. Disruption of normal CSF dynamics may result in increased intracranial pressures causing many problems including headaches and visual loss, as in idiopathic intracranial hypertension and hydrocephalus. To study the role of AGs in CSF egress, we have grown cells from human AG tissue in vitro and have characterized their expression of those cytoskeletal and junctional proteins that may function in the regulation of CSF outflow.
Methods
Human AG tissue was obtained at autopsy, and explanted to cell culture dishes coated with fibronectin. Typically, cells migrated from the explanted tissue after 7–10 days in vitro. Second or third passage cells were seeded onto fibronectin-coated coverslips at confluent densities and grown to confluency for 7–10 days. Arachnoidal cells were tested using immunocytochemical methods for the expression of several common cytoskeletal and junctional proteins. Second and third passage cultures were also labeled with the common endothelial markers CD-31 or VE-cadherin (CD144) and their expression was quantified using flow cytometry analysis.
Results
Confluent cultures of arachnoidal cells expressed the intermediate filament protein vimentin. Cytokeratin intermediate filaments were expressed variably in a subpopulation of cells. The cultures also expressed the junctional proteins connexin43, desmoplakin 1 and 2, E-cadherin, and zonula occludens-1. Flow cytometry analysis indicated that second and third passage cultures failed to express the endothelial cell markers CD31 or VE-cadherin in significant quantities, thereby showing that these cultures did not consist of endothelial cells from the venous sinus wall.
Conclusion
To our knowledge, this is the first report of the in vitro culture of arachnoidal cells grown from human AG tissue. We demonstrated that these cells in vitro continue to express some of the cytoskeletal and junctional proteins characterized previously in human AG tissue, such as proteins involved in the formation of gap junctions, desmosomes, epithelial specific adherens junctions, as well as tight junctions. These junctional proteins in particular may be important in allowing these arachnoidal cells to regulate CSF outflow.
==== Body
Background
Our understanding of cerebrospinal fluid (CSF) egress remains limited regarding fluid movement from the subarachnoid space across the arachnoid granulations (AGs) and into the venous sinuses. The classical view of CSF egress is that arachnoid granulations are herniations of the arachnoid membrane which project into the dural venous sinuses and function to return CSF to the systemic venous circulation [1,2]. In addition, it has long been recognized that there may be a lymphatic component to CSF drainage, recent tracer studies in sheep have suggested that extra-cranial lymphatics might account for as much as 40–48% of CSF outflow [3,4]. Similar results have not yet been demonstrated conclusively in humans, and the relative importance of the two routes at physiologic and non-physiologic intracranial pressures is uncertain. Hence, a study of arachnoidal cell cultures and their proteins will help in the understanding of CSF dynamics.
Impaired CSF circulation can result in increased intracranial pressure, causing hydrocephalus, severe headaches, tinnitus, diplopia, and transient visual obscurations. If left untreated, chronic intracranial pressure can cause intractable headache and compressive optic nerve damage, causing irreversible blindness. To study the role of the arachnoidal cells in CSF outflow and its pathologies, we have developed an in vitro model of the CSF outflow pathway across the arachnoid granulations. This model can be applied to physiological as well as pathological conditions of increased intracranial pressure, such as idiopathic intracranial hypertension where it has been suggested that CSF egress is impaired [5] by an increased resistance to outflow at the AGs [6-8]. This in vitro model utilizes arachnoidal cells cultured from human AG tissue seeded onto filter membranes as a model for the CSF outflow pathway.
The first step in utilizing this model effectively is to confirm that human arachnoidal cells in vitro express some of the cytoskeletal proteins [9,10] and junctional complexes [11] that have been described previously in fixed AG tissue using immunohistochemistry and electron microscopy. In particular, the junctional complexes including gap junctions [11,12], desmosomes [10,11,13-18], epithelial specific cell adhesion molecules (E-cadherin) [19-21], and tight junctions [11,15-18] are important in mediating cell-cell adhesion and communication. These proteins allow the arachnoid cells to form a barrier to CSF egress, regulating the return of CSF to the venous circulation.
The mechanism by which the AG cells facilitate fluid transport is still a topic of debate, though the similarity to the drainage of aqueous humour across the endothelia of Schlemm's canal has been noted by several authors [22,23]. These studies have suggested that AG cells regulate fluid flow by a process of large transcellular vacuoles that originate at the basal membrane of the cell and function as one-way valves [22,24,25]. Ultrastructural studies of human AGs by Yamashima [16,17] have identified large vacuoles and have also suggested that extracellular cisterns between AG cells contribute to the passive transport of CSF, presumably between transiently altered tight junctions.
This paper describes the culture and characterization of the cytoskeletal and junctional proteins expressed by arachnoidal cells grown from human arachnoid granulation explants. Using immunofluorescent microscopy and flow cytometry, we have demonstrated the growth of arachnoidal cells from human AG explants. We have shown that these cells in culture express many of the same cytoskeletal and junctional proteins that have previously been identified in human AG tissue using immunohistochemistry and electron microscopy.
Methods
Collection of brain tissue
Human brain tissue was obtained within 24 hours post-mortem from the Ohio State University Regional Autopsy Center. Tissue donors ranged in age from 22 to 88 years old. Brain tissue was collected in accordance with the guidelines and regulations set forth by the Office of Responsible Research Practices Institutional Review Board for human subjects at The Ohio State University (IRB#2002H3018). At autopsy, AGs were collected by resecting the superior sagittal sinus and lateral lacunae. Prior to explantation the tissue was washed 3X in sterile Dulbecco's phosphate buffered saline (D-PBS) (Cellgro Mediatech, Herndon, VA), and then placed into a Petri dish containing fresh culture media. Primary cell culture medium was Dulbecco's Modified Eagle Medium/Ham's F-12 Nutrient medium (50:50 v/v), with L-glutamine, penicillin/streptomycin, amphotericin B (all Cellgro Mediatech), and 10% newborn calf serum (Invitrogen Gibco, Carlsbad, CA).
Explant procedure and cell culture
The explant procedure was performed under a dissecting microscope. An individual granulation was secured with micro-surgical forceps adjacent to the apical cap cell portion of the granulation. Micro-surgical spring scissors were used to cut just below the forceps as close as possible to the cap of the granulation. Dissecting the apical portion of the granulation ensured that the cap cell cluster was explanted while minimizing the possibility that the fibrous capsule, central core, or underlying arachnoid membrane was explanted as well. The number of explants from each tissue donor depended on the frequency of the AGs on the surface of the brain and in the sinuses. Typical tissue donors generated between 24–48 explants.
The granulation was washed again in fresh culture medium, and explanted into a 24-well culture plate coated with a fibronectin solution of 30 μg human fibronectin (Sigma, St. Louis, MO) per ml M199 culture medium (Cellgro Mediatech). Culture medium was added to each well, and the explants were allowed to remain undisturbed in the incubator for a period of three to four days. The medium was subsequently changed every three to four days as needed. Upon confluency, the cells were washed with D-PBS, and dissociated with 0.05% trypsin EDTA in Hanks buffered saline (Cellgro Mediatech). Trypsinization would typically dissociate the explanted AG tissue as well. At this point sterile forceps were used to remove and discard the explanted tissue. Culture media containing 10% serum was added to the dissociated cells to inhibit the trypsin reaction and the cells were spun down at 1300 RPM. The cell pellet was resuspended and plated to a T-25 cm2 culture flask. In subsequent passages, cells were grown in T-75 cm2 flasks.
Antibodies
The following monoclonal antibodies were used for immunofluorescence microscopy at the dilutions indicated: mouse anti-human cytokeratin antibody clones AE1/AE3 (1:50, DakoCytomation, Carpinteria, CA), mouse anti-connexin43 antibody (1:100, Zymed, San Francisco, CA), Cy3 conjugated mouse anti-vimentin antibody (1:100, Sigma), mouse anti-human desmoplakin 1&2 antibody (1:40, Chemicon International, Temecula, CA), FITC conjugated mouse anti E-cadherin antibody (1:50, Becton Dickinson, Franklin Lakes, NJ), and FITC conjugated mouse anti ZO-1 antibody (1:50, Zymed). The secondary antibodies used were a fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG1 secondary antibody (Sigma) or an Alexa Fluor 555 conjugated donkey anti-mouse IgG1 antibody (Molecular Probes, Eugene, OR) at a 1:50 dilution for 45 minutes at 37°C. All primary and secondary antibodies were diluted in 10% serum from the same species in which the secondary antibody was produced (i.e. goat or donkey serum).
Immunocytochemistry
Second or third passage cells were seeded onto fibronectin-coated coverslips (Becton Dickinson) and grown to confluency. Cell cultures were tested at 1–1.5 weeks post-confluency for the presence of cytokeratins, vimentin, connexin43, desmoplakins 1&2, E-cadherin, and ZO-1. The cells were washed 3X with sterile D-PBS and fixed with 3.7% paraformaldehyde for 10 minutes, then permeabilized with 0.2% Triton X-100 (Sigma) in PBS at 37°C for 5 minutes. To block non-specific binding of the primary antibody, the cells were incubated for 30 minutes in 10% serum diluted in D-PBS from the same species in which the secondary antibody was produced. Next, the cells were incubated with the primary antibodies at the dilutions indicated above for 60 minutes at 37°C. When appropriate, the cells were washed in D-PBS and incubated with the secondary antibody for 45 minutes. After incubation with the secondary antibody, cells were washed in D-PBS, counterstained with 4',6-diamidino-2-phenylindole (DAPI), and mounted with an antifade reagent (Prolong Gold with DAPI, Molecular Probes) onto glass slides for visualization. The cell cultures were visualized using a Zeiss Axiocam inverted microscope equipped with DAPI, FITC, and Cy3 filter sets. As a negative control, cells grown on a cover slip were stained following the same procedure except the primary antibody was omitted. A lack of staining was interpreted as a high specificity of the primary antibody.
Labeling with endothelial markers for flow cytometry
Anti-VE-cadherin (CD144) purified IgG1 isotype with an anti-IgG1 secondary antibody conjugated to phycoerythrin (PE) (Becton Dickinson), and a FITC conjugated anti-CD31 antibody (Becton Dickinson) were used for flow cytometry analysis of approximately 3–5 × 106 cells. Cell cultures at second or third passage were harvested and spun down, in the same way as for passaging. The cells were resuspended at a concentration of 106 cells/100 μL. Unlabeled cells were used for control analysis and compensation (106 cells). The remaining cells were labeled with antibodies VE-cadherin (5 μL /106 cells) or CD31-FITC (5 μL /106 cells). The cells were incubated with the primary antibody at 4°C for 25–30 minutes, and then washed with labeling buffer (PBS with 2 mM EDTA, 0.5% bovine serum albumin). For the VE-cadherin experiments cells were then incubated with an anti IgG1 PE secondary antibody (10 μL/106 cells) for 25 minutes. The cells were washed with labeling buffer and finally resuspended in 2% paraformaldehyde.
Flow cytometry analysis
Analysis was performed on a Becton-Dickinson FACS Calibur equipped with 4 photo multiplier tubes, allowing for 4-color analysis using a 488 nm air-cooled argon and a 633 nm helium-neon laser as excitation wavelengths. VE-cadherin-PE or CD31-FITC expression was assessed within labeled cell populations by comparing their fluorescent expression to unstained control cells. Cells were gated based on their forward and side scattered light (FSC and SSC, respectively) to exclude cellular debris (low SSC and FSC) as well as clumps of cells (large SSC and FSC) that may give erroneous fluorescent readings. Only cells that fell within the defined gate in the light scatter plot were subsequently analyzed for fluorescent expression.
In the subsequent fluorescent analysis, unlabeled cells were examined to determine their intrinsic PE or FITC fluorescence. To determine their intrinsic PE fluorescence, unstained cells were excited with a 488 nm argon laser and their fluorescent emission was detected in the FL2 channel which detects wavelengths from 564–604 nm. Based on their fluorescent emission in the FL2 channel, a gate was established in the PE fluorescence histogram that was considered the intrinsic PE fluorescence by unlabeled arachnoidal cells. In analyzing cells labeled with VE-cadherin-PE, cells were gated based on their light scatter (SSC and FSC) using the same gate as for the unlabeled cells. In the following fluorescent analysis, cells falling within the predefined gate for intrinsic PE fluorescence were considered negative for VE-cadherin-PE expression.
A similar procedure was followed in analyzing cells for CD31-FITC expression except that intrinsic FITC fluorescence was determined in unlabeled arachnoidal cells by exciting cells with a 488 mn argon laser and detecting their fluorescent emission in the FL1 channel which detects fluorescent wavelengths from 515–545 nm.
Results
Arachnoidal cell culture
Cell migration from human AG explants was generally seen within 7–10 days (Figure 1A). At confluency, arachnoidal cells exhibited contact inhibition and grew in monolayers with a densely packed polygonal morphology and a cobblestone-like appearance characteristic of epithelial cell types (Figure 1B). Cells could remain in confluent culture for as long as a month and passaged as many as 5–6 times before exhibiting signs of senescence, manifested by enlargement of the cells and the dramatic slowing of cell growth rate.
Figure 1 Human AG explants and arachnoidal cells in culture. Human AG tissue was obtained at autopsy, usually within 24 hours post-mortem and explanted into cell culture plates coated with fibronectin. A: Cells were seen migrating from the explant within 7–10 days. Bar = 200 μm. B: Arachnoidal cells grown from AG explants in confluent cultures exhibited polygonal cell morphology. Confluent cultures of arachnoidal cells packed in quite densely and displayed a cobblestone-like appearance common to epithelial cells in culture. Bar = 200 μm.
In limited cases, some cultures became overgrown by fibroblasts, identified as elongated spindle-shaped cells with many processes. Their growth was marked by rapid proliferation that could overgrow the arachnoidal cells in cultures. In instances of fibroblast overgrowth, these cultures were discarded and were not used for immunofluorescent labeling or flow cytometry analysis.
Immunocytochemical labeling of cytoskeletal and junction proteins
Cells cultured from human AGs expressed the intermediate filament proteins vimentin and cytokeratin. Second passage cell cultures were immunoreactive to the anti-vimentin antibody (Figure 2A), and cells expressed this protein uniformly throughout the cell cytoplasm. Additionally, a subpopulation of cells in second passage cultures was immunoreactive to the anti-human cytokeratin antibody (AE1/AE3), which recognizes a wide range of human cytokeratins (Moll's designation 1–8, 10, 13–16, 19) [26]. The expression of cytokeratin in the arachnoidal cell cultures varied between experiments. Often, only a subset of stained cells expressed cytokeratins. Immunoreactive cells expressed cytokeratin intermediate filaments in a perinuclear pattern, with long filaments surrounding the nucleus in a basket-like structure (Figure 2B).
Figure 2 Immunocytochemical staining of confluent cell cultures from human AG tissue on fibronectin coverslips. A: Arachnoidal cells in culture were incubated with a Cy3 conjugated anti-vimentin antibody. Cells expressed this intermediate filament protein uniformly throughout the cytoplasm. B: Arachnoidal cells were labeled with a broad spectrum anti-cytokeratin antibody and visualized with a FITC conjugated secondary antibody. Cells expressed the epithelial specific intermediate filament protein cytokeratin in a perinuclear pattern, though the expression of this protein was not uniform. C: Cells cultured from AG tissue were incubated with a connexin43 antibody and then visualized with a FITC conjugated secondary antibody. Immunolocalization of this protein in a punctuate pattern at cell-cell borders (red arrows) indicates that arachnoidal cells are able to form gap junctions in confluent culture. D: Tight junctions in confluent cultures were identified by immunoreactivity to a FITC conjugated ZO-1 antibody. The antibody deposition pattern can be seen at cell-cell borders (white arrow) with overlapping filapodia consisting of short linear structures in parallel. E: Arachnoidal cells in culture were labeled with an anti-desmoplakin antibody and revealed with an Alexa Fluor 555 conjugated secondary antibody. Confluent cultures were able to form desmosomes as evidenced by the punctuate staining along the membranes of adjacent cell borders (white arrows). F: E-cadherin immunoreactivity was demonstrated by incubating cells with a FITC conjugated antibody to E-cadherin. This epithelial specific cell adhesion molecule was expressed at the periphery of the cells, at cell-cell contacts (white arrows) in a pattern similar to that of connexin43 or ZO-1. All immunofluorescent images were taken at the same magnification. Bar= 50 μm.
Cells cultured from AGs also stained positively for connexin43, with immunofluorescent microscopy showing punctate staining at cell-cell borders (Figure 2C, red arrows). Expression of the ZO-1 protein at the cell membrane between adjacent cells is seen in Figure 2D (white arrows). The antibody deposition pattern can be seen at cell-cell borders with overlapping cellular processes seen as short linear structures in parallel. The expression of desmoplakin 1&2 was similar to that of connexin43, with punctate staining seen along the plasma membrane of two adjoining cells (Figure 2E, white arrows). Finally, Figure 2F shows cells labeled with the FITC-conjugated E-cadherin antibody. This antibody labeled cells at their periphery, at cell-cell contacts (Figure 2D, white arrows), in a pattern similar to the connexin43 or ZO-1 labeled cells.
Cells were also incubated with a FITC conjugated goat anti-mouse IgG1 secondary antibody only. The lack of staining in the absence of the primary antibody indicates a high specificity between the primary and secondary antibodies (results not shown). Similar results were obtained for cells incubated with only the Alexa Fluor 555 conjugated donkey anti-mouse IgG1 secondary antibody (results not shown).
Immunocytochemical labeling experiments with antibodies to cytoskeletal and junctional proteins were repeated at least once for cultures grown from the same tissue donor. Immunofluorescent labeling of cytoskeletal and junctional proteins was repeated in at least three additional tissue donors with positive staining seen in all cultures. Images obtained from these experiments were similar to those presented in Figure 2.
Flow cytometry analysis of endothelial markers
Cells cultured from human AGs at second and third passage were labeled with the antibodies to VE-cadherin-PE or CD31-FITC and their expression of these endothelial markers was quantified using flow cytometry. Figure 3A shows the light scatter plot for the unlabeled control cells in the VE-cadherin-PE experiments. The gate R1 (in red) was selected to exclude any cellular debris (low SSC and FSC) and clumps of cells (high SSC and FSC). Only events (cells) falling within this gate were further analyzed for PE fluorescence. Figure 3B shows PE fluorescence histogram used to determine the intrinsic PE fluorescence in unlabeled arachnoidal cell cultures. Gate M1 was created to define the intrinsic PE fluorescence. Of the 4,497 events that fell within R1 and were subsequently analyzed for fluorescence, 4,487 (99.8%) also fell within the gate M1. Figures 3C and 3D show the light scatter plot and PE fluorescence histogram respectively for the cells labeled with the VE-cadherin-PE antibody. The position of the gates R1 and M1 remained unchanged from the control analysis. Of the 2,601 events that fell within R1 and were subsequently analyzed for VE-cadherin-PE expression, 2,308 (88.7%) also fell within gate M1 and were considered negative for VE-cadherin expression.
Figure 3 Flow cytometry analysis of cells cultured from AG tissue for expression of the endothelial marker VE-cadherin-PE. A: A scatter plot of unlabeled cells, showing a gate, R1, used to define the events subsequently analyzed for fluorescence. SSC and FSC are side scattered light and forward scattered light respectively. B: A histogram was created to determine the autofluorescence of unlabeled cells. The gate M1 is defined as the intrinsic PE fluorescence of unlabeled cells. C: A scatter plots of cells labeled with the endothelial marker VE-cadherin-PE. The position of the gate R1 was unchanged from the control analysis. D: A histogram of the cells labeled with VE-cadherin-PE that fell within the gate R1 and were analyzed for PE fluorescence. Of the 2,601 events that were analyzed, 2,308 (88.7%) fell within the gate, M1, and were considered negative for VE-cadherin expression. The position of the gate, M1, remained unchanged from the control analysis.
Similar results were obtained for the CD31-FITC labeled cells (Figure 4). The analysis of the unstained control cells is shown in figure 4A and 4B. The light scatter plot in figure 4A also shows the gate, R2 (in red), that was used for subsequent FITC fluorescence analysis. Figure 4B displays the fluorescence histogram that was used to determine the intrinsic FITC fluorescence in unlabeled arachnoidal cell cultures. Of the 3,704 events that fell within the gate R2 and were subsequently analyzed for FITC fluorescence, 3,697 (99.8%) also fell within the gate M2 that was considered the intrinsic FITC fluorescence. Figures 4C and 4D show the light scatter plot and FITC fluorescence histogram respectively for the cells labeled with the CD31-FITC antibody. The position of the gates R2 and M2 remained unchanged from the control analysis. Of the 3,767 events that fell within R2 and were subsequently analyzed for CD31-FITC expression, 3,761 (99.8%) also fell within the gate M2 and were considered negative for CD31 expression. The failure of these cultures to express the endothelial specific proteins CD31 and VE cadherin in significant quantities indicated that cells cultured from human AG explants were not significantly contaminated by endothelia from the venous sinus lumen.
Figure 4 Flow cytometry analysis of cells cultured from AG tissue for expression of the endothelial marker CD31-FITC. A: A Scatter plot of unlabeled cells showing a gate, R2, which was created to define events that were subsequently analyzed for fluorescence. SSC and FSC are side scattered and forward scattered light respectively. B: Histogram used to determine the autofluorescence of unlabeled cells, with the gate M2 defined as the intrinsic FITC fluorescence. C: A scatter plot of cells labeled with the CD31-FITC antibody. The position of the gate R2 in the scatter plot was unchanged from the control analysis. D: Histogram of the 3,767 events that fell within R2 and were subsequently analyzed for CD31-FITC expression. Almost all events (99.8%) also fell within the gate M2, indicating that cells cultured from AG tissue did not express the endothelial marker CD31-FITC. The position of gate, M2, did not change from control analysis.
Flow cytometry experiments quantifying the expression of the endothelial markers were repeated with cells cultured from two additional tissue donors. Similar results were obtained to those presented in Figures 3 and 4. Thereafter, it was concluded that endothelial cells were likely not a contaminating cell type and further experiments were not performed
Discussion
The role of the arachnoid granulations in CSF outflow has been the focus of attention since the works of Weed nearly 90 years ago [1,2]. While the ultrastructure of the arachnoid membrane and granulations has been studied extensively with light and electron microscopy [11,15-18,22,24,25,27-34], functional studies have been few. Because they are located intracranially, it is difficult to observe and measure flow across human AGs in vivo. Hence functional studies have been restricted primarily to animal models. Welch and Pollay have described in vitro preparations in both monkey and canine [34,35], where excised pieces of dura and arachnoid villi were perfused with colloidal gold, yeast, erythrocytes, and polystyrene microspheres. However, a significant proportion of the preparations developed leaks (7 of 22 in monkeys). In addition, the structure of the arachnoid villi and granulations in primate and canines may not accurately represent the structure of human arachnoid granulations [15,23]. To more closely study the role of arachnoid cells in CSF outflow, we have undertaken efforts to isolate these cells from human AG tissue.
To our best knowledge, this is the first demonstration of primary arachnoidal cell cultures from human AG tissue. Previously leptomeningeal cells have been cultured from the human pia and arachnoid membranes [36-42], but cultures from AG tissue have not been described. This study showed that cells cultured from AG tissue have specific immunological characteristics and can be prepared without contamination of adjacent endothelial cells.
In some limited cases, cultures of arachnoidal cells became overgrown by fibroblasts. Fibroblast contamination was marked by a rapid growth rate that could over grow the arachnoidal cells in culture. However, with the current explant technique, it was possible to consistently grow cultures that were predominantly free of fibroblasts contamination. This was assessed by uniform cobblestone morphology and a homogenous expression of junctional proteins using immunofluorescence.
While care was taken when dissecting the AG tissue, it is possible in some instances that portions of the underlying arachnoid membrane were explanted as well. In such cases, it would be difficult to distinguish cells of the arachnoid membrane from arachnoidal cells from AGs. Differences in protein expression between cells lining the arachnoid granulation and the arachnoid membrane have not been reported and would not be expected, as the AGs are continuous with the arachnoid membrane proper, and are often described as projections of the arachnoid membrane into the dural sinuses. There remain functional differences inferred from differences in ultrastructure between the cells of the arachnoid membrane and those of the arachnoid granulation. The cells lining the arachnoid granulations allow the passage of CSF to the venous sinuses by forming transcellular vacuoles and extracellular cisterns [16-18,22,24,25,43].
The possibility existed that several of the junctional proteins could have been expressed by contaminating endothelial cells from the venous sinus lumen. The flow cytometry analysis, however, demonstrated that arachnoidal cell cultures did not express endothelial markers in significant quantities. Compared to unlabeled control cells, approximately 10% of VE-cadherin-PE labeled cells expressed this marker positively. It was possible that this constituted a small endothelial population within these cultures, though further inspection of this histogram (Figure 3D) did not reveal a second population of PE positive cells. Taken together with the nearly complete lack of CD31 expression (>99%), this strongly suggests that endothelial cells were not present in significant quantities in cultures from AG explants.
To characterize the profile of immunological markers expressed by arachnoidal cells grown from human AG tissue, cultures were labeled with fluorescently conjugated antibodies to several cytoskeletal and junctional proteins. Vimentin intermediate filament expression has been characterized extensively using immunohistochemistry in the cells of the arachnoid membrane, granulations, and meningiomas [9,10,14,44,45], as well as in cultured human leptomeningeal cells [37,39-41]. Cells cultured from AG tissue uniformly expressed vimentin intermediate filaments.
The connexins are found in cells containing gap junctions, providing a pathway for direct intercellular communication for ions, amino acids, and nucleotides. Gap junctions have been identified in human arachnoid membranes and AGs using electron microscopy, freeze fracture, and immunohistochemistry [11,12]. Grafstein et al. [39] demonstrated connexin43 expression in cultures of human leptomeningeal cells and showed that these cells could propagate calcium waves suggesting a pathway for intercellular communication. The presence of gap junctions in cell cultured from AGs may provide a pathway for intercellular communication, allowing the regulation of CSF passage.
Desmosomal junctions are typically found in cells of epithelial origin. They function to anchor bundles of intermediate filaments at desmoplakin proteins [46]. While most desmoplakins associate with cytokeratin filaments, exceptions have been found, including the arachnoid cells that anchor vimentin intermediate filaments at these plaques. The presence of desmosomes has been identified with electron microscopy, freeze fracture, and immunohistochemistry [10,11]. Desmosomal junctions are also expressed by in vitro cultures of leptomeningeal cells and have been used as a specific marker for these cells [40]. The presence of desmosomes in the in vitro cultures from human AGs indicates that these cultures are of arachnoidal origin.
Cadherins are cell adhesion molecules that are largely responsible for calcium dependent cell-cell adhesion. Several types have been described, including N-cadherin in neural and muscle tissue, P-cadherin in placental tissue, E-cadherin found in epithelial cells, and VE-cadherin expressed exclusively in endothelial cells [47,48]. The expression of E-cadherin has been demonstrated in the flash frozen arachnoid membrane, granulations and meningioma tissue using standard immunohistochemistry [19-21]. E-cadherin functions in AG cells to bind the arachnoid cells lining the arachnoid villi and granulations together flexibly and may allow the formation of extracellular cisterns during the bulk outflow of CSF [20]. The in vitro expression of E-cadherin by cells cultured from AG tissue demonstrated that these cells were epithelial and could form cell-cell junctions that may be necessary to regulate the passage of fluid.
The zonula occludens (ZO proteins) are peripheral membrane proteins that associate with other proteins in tight junctions where they serve to anchor the actin cytoskeleton. In epithelial cells, these junctions maintain polarity and regulate the paracellular passage of ions, macromolecules, and water. While tight junctions have been recognized in the arachnoid cells lining the AGs [15-17], these junctions have not been described in cells cultured from the leptomeninges. Demonstrating the presence of tight junctions in cells cultured from AG tissue cultures, suggest that these cells have the potential to form a barrier to fluid flow.
The expression of cytokeratin filaments in the arachnoid membrane and granulations has been the subject of debate. The association of intermediate filaments with desmosomal junctions in meningiomas, arachnoid membrane, and arachnoid granulations was first reported by Kartenbeck et al. and Schwecheimer et al. [10,14], who demonstrated that desmosomal plaques anchored vimentin filaments, while cytokeratin immunoreactivity within these tissues was not found. Subsequent immunohistochemical studies have also failed to find cytokeratin expression in the meninges and AGs [44,45,49]. However, cytokeratin positive cells within the arachnoid have been recognized in limited cases [50,51]. Several published reports of cultured leptomeningeal cells show that these cells express cytokeratin uniformly [38,40,41]. Of note are studies by Frank et al. [38] and Murphy et al. [40] focusing on cultured leptomeningeal cells grown from human arachnoid membranes.
The variability in cytokeratin expression may be related to the embryological origin of the meninges. The development of the meninges has been described in detail by O'Rahilly and Müller [52], who suggest that there are several possible sources for the human cranial meninges, including the parachordal mesoderm, mesectoderm, and other neuroectodermal elements. Embryological contribution from mesodermal as well as neuroectodermal elements may explain the later mesenchymal and epithelial properties of the adult meninges. Arachnoid cells appear to express both epithelial and mesenchymal properties [36,44,45,49]; mesenchymal by their ability to express vimentin intermediate filaments and synthesize basement membrane proteins (collagen type I and IV, fibronectin, laminin) [37], and epithelial through the formation of cell-cell junctions [11].
Since cytokeratin expression has not been found consistently in human arachnoid membrane and granulation tissue, it seems anomalous that cells cultured from these tissues should express cytokeratins uniformly. In identifying arachnoidal cells within these cultures, the formation of cell-cell junctions may be more relevant to recapitulate these cells' in vivo function, regulating the outflow of CSF.
Future studies will focus on characterizing these cells' ability to form an occluding barrier to fluid flow using a perfusion chamber and hydrostatic pressure control. Cells are perfused and fixed under pressure, and the cellular ultrastructure is examined using electron microscopy. These studies will also try to determine whether arachnoidal cells in vitro can mimic the one-directional flow of CSF in vivo.
Conclusion
These results show what we believe to be the first demonstration and characterization of arachnoidal cells cultured from human AG tissue. We have demonstrated the in vitro expression of several important cytoskeletal and junctional proteins previously identified in AG tissue by immunohistochemistry and electron microscopy. The expression of these junctional proteins is an important first step in demonstrating that arachnoidal cells grown from human AG tissue in vitro can exhibit some of the barrier properties previously recognized in intact AG tissue and may allow these cells to regulate CSF outflow.
List of Abbreviations
AG: Arachnoid granulation, CSF: cerebrospinal fluid, DAPI: 4',6-diamidino-2-phenylindole, D-PBS: Dulbecco's phosphate buffered saline, FITC: fluorescein isothiocyanate, FSC: forward scatter, PE: phycoerythrin, SSC: side scatter
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DH: collected and dissected brain tissue, performed immunocytochemical staining, with BM prepared and analyzed cells with flow cytometry, and drafted and revised the manuscript.
DG: conceived of the study, developed the initial explant procedures, participated in the study design and coordination, and helped to draft and revise the manuscript.
BM: with DH prepared and analyzed cells with flow cytometry.
SEK: participated in the design and coordination of the study, helped to revise the manuscript.
ML: participated in the design and coordination of the study, helped to revise the manuscript.
All authors read and approved the final manuscript.
Acknowledgements
This research was funded by the Davis Medical Research Grant and the Ohio Lions Eye Research Foundation.
==== Refs
Weed LH Studies on the Cerebrospinal Fluid: No. III The Pathways of Escape from Subarachnoid Spaces with Particular Reference to the Arachnoid Villi J Med Res 1914 31 51 91
Weed LH The Absorption of the Cerebrospinal Fluid into the Venous System Am J Anat 1923 31 191 221 10.1002/aja.1000310302
Johnston M The importance of lymphatics in cerebrospinal fluid transport Lymphat Res Biol 2003 1 41 4; discussion 45 15624320 10.1089/15396850360495682
Johnston M Relationship between cerebrospinal fluid and extracranial lymph Lymphology 2000 33 1 3 10769809
Johnston I Hawke S Halmagyi M Teo C The pseudotumor syndrome. Disorders of cerebrospinal fluid circulation causing intracranial hypertension without ventriculomegaly Arch Neurol 1991 48 740 747 1859302
Martins AN Resistance to drainage of cerebrospinal fluid: clinical measurement and significance J Neurol Neurosurg Psychiatry 1973 36 313 318 4541080
Johnston I The definition of reduced CSF absorption syndrome: a reapraisal of benign intracranial hypertension and related conditions Med Hypotheses 1975 1 10 14 1105091 10.1016/0306-9877(75)90035-3
Levine DN Ventricular size in pseudotumor cerebri and the theory of impaired CSF absorption J Neurol Sci 2000 177 85 94 10980304 10.1016/S0022-510X(00)00348-8
Yamashima T Tachibana O Nitta H Yamaguchi N Yamashita J Ultrastructural immunogold labelling of vimentin filaments on postembedding ultrathin sections of arachnoid villi and meningiomas Histol Histopathol 1989 4 47 53 2520446
Kartenbeck J Schwechheimer K Moll R Franke WW Attachment of vimentin filaments to desmosomal plaques in human meningiomal cells and arachnoidal tissue J Cell Biol 1984 98 1072 1081 6365927 10.1083/jcb.98.3.1072
Hasegawa M Yamashima T Kida S Yamashita J Membranous ultrastructure of human arachnoid cells J Neuropathol Exp Neurol 1997 56 1217 1227 9370232
Arishima H Sato K Kubota T Immunohistochemical and ultrastructural study of gap junction proteins connexin26 and 43 in human arachnoid villi and meningeal tumors J Neuropathol Exp Neurol 2002 61 1048 1055 12484567
Akat K Mennel HD Kremer P Gassler N Bleck CK Kartenbeck J Molecular characterization of desmosomes in meningiomas and arachnoidal tissue Acta Neuropathol (Berl) 2003 106 337 347 12845453 10.1007/s00401-003-0736-8
Schwechheimer K Kartenbeck J Moll R Franke WW Vimentin filament-desmosome cytoskeleton of diverse types of human meningiomas. A distinctive diagnostic feature Lab Invest 1984 51 584 591 6387275
Kida S Yamashima T Kubota T Ito H Yamamoto S A light and electron microscopic and immunohistochemical study of human arachnoid villi J Neurosurg 1988 69 429 435 3404241
Yamashima T Functional ultrastructure of cerebrospinal fluid drainage channels in human arachnoid villi Neurosurgery 1988 22 633 641 3374774
Yamashima T Ultrastructural study of the final cerebrospinal fluid pathway in human arachnoid villi Brain Res 1986 384 68 76 3791000 10.1016/0006-8993(86)91220-5
Yamashima T Kida S Yamamoto S Ultrastructural comparison of arachnoid villi and meningiomas in man Mod Pathol 1988 1 224 234 3237703
Tohma Y Yamashima T Yamashita J Immunohistochemical localization of cell adhesion molecule epithelial cadherin in human arachnoid villi and meningiomas Cancer Res 1992 52 1981 1987 1551127
Yamashima T Tohma Y Yamashita J Expression of cell adhesion molecule E-cadherin in human arachnoid villi J Neurosurg 1992 77 749 756 1403118
Figarella-Branger D Pellissier JF Bouillot P Bianco N Mayan M Grisoli F Rougon G Expression of neural cell-adhesion molecule isoforms and epithelial cadherin adhesion molecules in 47 human meningiomas: correlation with clinical and morphological data Mod Pathol 1994 7 752 761 7824509
Tripathi RC The functional morphology of the outflow systems of ocular and cerebrospinal fluids Exp Eye Res 1977 25 Suppl 65 116 590403
Ohta K Inokuchi T Hayashida Y Mizukami T Yoshida T Kawahara T Regional diminution of von Willebrand factor expression on the endothelial covering arachnoid granulations of human, monkey and dog brain Kurume Med J 2002 49 177 183 12652968
Tripathi R Tracing the bulk outflow route of cerebrospinal fluid by transmission and scanning electron microscopy Brain Res 1974 80 503 506 4214009 10.1016/0006-8993(74)91033-6
Tripathi RC Ultrastructure of the arachnoid mater in relation to outflow of cerebrospinal fluid. A new concept Lancet 1973 2 8 11 4123339 10.1016/S0140-6736(73)91945-4
Moll R Franke WW Schiller DL Geiger B Krepler R The catalog of human cytokeratins: patterns of expression in normal epithelia, tumors and cultured cells Cell 1982 31 11 24 6186379 10.1016/0092-8674(82)90400-7
Alksne JF Lovings ET Functional ultrastructure of the arachnoid villus Arch Neurol 1972 27 371 377 5078893
Alksne JF Lovings ET The role of the arachnoid villus in the removal of red blood cells from the subarachnoid space. An electron microscope study in the dog J Neurosurg 1972 36 192 200 5008733
Gomez DG DiBenedetto AT Pavese AM Firpo A Hershan DB Potts DG Development of arachnoid villi and granulations in man Acta Anat (Basel) 1982 111 247 258 7064645
Gomez DG Potts DG Effects of pressure on the arachnoid villus Exp Eye Res 1977 25 Suppl 117 125 412689
Gomez DG Potts DG The surface characteristics of arachnoid granulations Arch Neurol 1974 31 88 93 4834969
Gomez DG Potts DG Deonarine V Arachnoid granulations of the sheep. Structural and ultrastructural changes with varying pressure differences Arch Neurol 1974 30 169 175 4810895
Gomez DG Potts G Deonarine V Reilly KF Effects of pressure gradient changes on the morphology of arachnoid villi and granulations of the monkey Lab Invest 1973 28 648 657 4197445
Pollay M Welch K The function and structure of canine arachnoid villi J Surg Res 1962 2 307 311 14487638
Welch K Pollay M Perfusion of particles through arachnoid villi of the monkey Am J Physiol 1961 201 651 654 14005969
Rutka JT Giblin J Dougherty DV McCulloch JR DeArmond SJ Rosenblum ML An ultrastructural and immunocytochemical analysis of leptomeningeal and meningioma cultures J Neuropathol Exp Neurol 1986 45 285 303 3083053
Rutka JT Kleppe-Hoifodt H Emma DA Giblin JR Dougherty DV McCulloch JR De Armond SJ Rosenblum ML Characterization of normal human brain cultures. Evidence for the outgrowth of leptomeningeal cells Lab Invest 1986 55 71 85 3724065
Frank EH Burge BW Liwnicz BH Lotspeich LJ White JC Wechsler SL Mayfield FH Keller JT Cytokeratin provides a specific marker for human arachnoid cells grown in vitro Exp Cell Res 1983 146 371 376 6192005 10.1016/0014-4827(83)90138-6
Grafstein B Liu S Cotrina ML Goldman SA Nedergaard M Meningeal cells can communicate with astrocytes by calcium signaling Ann Neurol 2000 47 18 25 10632097 10.1002/1531-8249(200001)47:1<18::AID-ANA6>3.0.CO;2-N
Murphy M Chen JN George DL Establishment and characterization of a human leptomeningeal cell line J Neurosci Res 1991 30 475 483 1724784 10.1002/jnr.490300304
Motohashi O Suzuki M Yanai N Umezawa K Shida N Shirane R Yoshimoto T Primary culture of human leptomeningeal cells in serum-free medium Neurosci Lett 1994 165 122 124 8015711 10.1016/0304-3940(94)90724-2
Motohashi O Suzuki M Yanai N Umezawa K Shida N Yoshimoto T Thrombin and TGF-beta promote human leptomeningeal cell proliferation in vitro Neurosci Lett 1995 190 105 108 7644116 10.1016/0304-3940(95)11513-V
Yamashima T On Arachnoid Villi and Meningiomas: Functional Implication of Ultrastructure, Cell Adhesion Mechanisms, and Extracellular Matrix Composition Pathol Oncol Res 1996 2 144 149 11173597
Schnitt SJ Vogel H Meningiomas. Diagnostic value of immunoperoxidase staining for epithelial membrane antigen Am J Surg Pathol 1986 10 640 649 2428264
Winek RR Scheithauer BW Wick MR Meningioma, meningeal hemangiopericytoma (angioblastic meningioma), peripheral hemangiopericytoma, and acoustic schwannoma. A comparative immunohistochemical study Am J Surg Pathol 1989 13 251 261 2648875
Franke WW Moll R Mueller H Schmid E Kuhn C Krepler R Artlieb U Denk H Immunocytochemical identification of epithelium-derived human tumors with antibodies to desmosomal plaque proteins Proc Natl Acad Sci U S A 1983 80 543 547 6340102
Breviario F Caveda L Corada M Martin-Padura I Navarro P Golay J Introna M Gulino D Lampugnani MG Dejana E Functional properties of human vascular endothelial cadherin (7B4/cadherin-5), an endothelium-specific cadherin Arterioscler Thromb Vasc Biol 1995 15 1229 1239 7627717
Lampugnani MG Resnati M Raiteri M Pigott R Pisacane A Houen G Ruco LP Dejana E A novel endothelial-specific membrane protein is a marker of cell-cell contacts J Cell Biol 1992 118 1511 1522 1522121 10.1083/jcb.118.6.1511
Ng HK Tse CC Lo ST Meningiomas and arachnoid cells: an immunohistochemical study of epithelial markers Pathology 1987 19 253 257 2448733
Kasper M Perry G Stosiek P Cytokeratin expression in human spinal meninges and ependymal cells J Hirnforsch 1991 32 19 25 1725782
Achtstatter T Fouquet B Rungger-Brandle E Franke WW Cytokeratin filaments and desmosomes in the epithelioid cells of the perineurial and arachnoidal sheaths of some vertebrate species Differentiation 1989 40 129 149 2474473
O'Rahilly R Muller F The meninges in human development J Neuropathol Exp Neurol 1986 45 588 608 3746345
|
16223448
|
PMC1285366
|
CC BY
|
2021-01-04 16:37:38
|
no
|
Cerebrospinal Fluid Res. 2005 Oct 13; 2:9
|
utf-8
|
Cerebrospinal Fluid Res
| 2,005 |
10.1186/1743-8454-2-9
|
oa_comm
|
==== Front
Lipids Health DisLipids in Health and Disease1476-511XBioMed Central London 1476-511X-4-251624201810.1186/1476-511X-4-25ResearchHDL enhances oxidation of LDL in vitro in both men and women Solakivi T [email protected] O [email protected]äki A [email protected] N [email protected] S [email protected]äki T [email protected] H [email protected] ST [email protected] Department of Medical Biochemistry, University of Tampere, Medical School, Tampere, Finland2 Institute of Medical Technology, University of Tampere, Tampere, Finland3 Laboratory of Atherosclerosis Genetics, Department of Clinical Chemistry, Tampere University Hospital, Tampere, Finland4 Department of Internal Medicine, Tampere University Hospital, Tampere, Finland2005 20 10 2005 4 25 25 29 9 2005 20 10 2005 Copyright © 2005 Solakivi et al; licensee BioMed Central Ltd.2005Solakivi 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
Oxidative modification of low-density lipoprotein (LDL) is a key event in the oxidation hypothesis of atherogenesis. Some in vitro experiments have previously suggested that high-density lipoprotein (HDL) co-incubated with LDL prevents Cu2+-induced oxidation of LDL, while some other studies have observed an opposite effect. To comprehensively clarify the role of HDL in this context, we isolated LDL, HDL2 and HDL3 from sera of 61 free-living individuals (33 women and 28 men).
Results
When the isolated LDL was subjected to Cu2+-induced oxidation, both HDL2 and HDL3 particles increased the rate of appearance and the final concentration of conjugated dienes similarly in both genders. Oxidation rate was positively associated with polyunsaturated fatty acid content of the lipoproteins in that it was positively related to the content of linoleate and negatively related to oleate. More saturated fats thus protected the lipoproteins from damage.
Conclusion
We conclude that in vitro HDL does not protect LDL from oxidation, but is in fact oxidized fastest of all lipoproteins due to its fatty acid composition, which is oxidation promoting.
==== Body
Background
Epidemiological studies show an inverse correlation between high-density lipoprotein (HDL) concentration and the risk of developing coronary artery disease [1]. According to a widely accepted hypothesis, HDL or its subtractions play an important role in recruiting and transporting cholesterol from peripheral tissues to the liver for excretion, a series of events known as reverse cholesterol transport [2]. Other properties of HDL link its antiatherogenic functions to its antioxidative effects. Some studies have shown that co incubation of LDL with HDL in the presence of divalent copper prevents the oxidative modification of LDL [3]. In some reports this finding could not be confirmed, and in fact it has been demonstrated that in vitro HDL is oxidized faster than other lipoproteins [4]. When HDL is oxidatively modified, it alters to a form that causes macrophages to accumulate cholesterol. [5]. It has been suggested that systemic inflammation gives rise to prooxidant and proinflammatory HDL particles [6]. Oxidatively modified HDL is found in atheromatous plaques from human aorta [7]. Oxidatively modified HDL is no longer capable of removing cholesterol from cells, and it enhances LDL oxidation [8].
The contradictory findings on the role of HDL on LDL oxidation in vitro may be due to rather small study populations, and the reported heterogeneity of oxidation kinetics between lipoprotein preparations in vitro [9] which might be due to individual intrinsic properties of the lipoproteins. In the present paper we report the results of a study of the effect of HDL subtractions and gender on ex vivo oxidation of LDL from a population of 61 healthy free-living human subjects.
Results
Background characteristics of the men and women participating in the study are shown in Table 1. Compared with women, men had higher body mass indices, serum total cholesterol as well as LDL and apoB concentrations. Men had smaller LDL size and smaller concentrations of serum HDL and apoA-I than women.
Table 1 Characteristics of the study subjects.
Women Men All
N 32 27 59
Age (years) 39.3 ± 10.5 39.3 ± 11.0 39.3 ± 10.6
Body mass index (kg/m2) 23.2 ± 3.1 25.8 ± 3.6** 24.4 ± 3.6
Total cholesterol (mmol/l) 5.23 ± 0.87 5.76 ± 1.02* 5.47 ± 0.97
Triacylglycerol (mmol/l) 0.96 ± 0.41 1.81 ± 1.27*** 1.35 ± 0.71
HDL cholesterol (mmol/l) 1.88 ± 0.32 1.38 ± 0.23*** 1.65 ± 0.38
LDL cholesterol (mmol/l) 2.91 ± 0.84 3.63 ± 0.92** 3.21 ± 0.96
ApoA-I (g/l) 1.70 ± 0.20 1.46 ± 0.13*** 1.59 ± 0.21
ApoB (g/l) 0.81 ± 0.20 1.00 ± 0.22** 0.90 ± 0.23
Lp(a) (U/l) 266 ± 226 165 ± 201 220 ± 219
fB-Glucose (mmol/l)a 4.4 ± 0.4 4.6 ± 0.6 4.5 ± 0.5
LDL diameter (nm) 26.94 ± 0.45 26.40 ± 0.55*** 26.70 ± 0.51
Values are mean ± SD. a fB, fasting blood, * p < 0.05, ** p < 0.01, *** p < 0.001 compared to women by Mann-Whitney U-test.
The fatty acid compositions of ultracentrifugally isolated LDL, HDL2 and HDL3 fractions were analyzed by gas liquid chromatography (Table 2). There were no gender differences in the total amounts of saturated, monounsaturated and polyunsaturated fatty acids of LDL, HDL2 and HDL3. The calculated peroxidizability indices were also similar in men and women. In both genders this index increased significantly from LDL to HDL2 to HDL3 (p < 0.001, Wilcoxon's matched pairs test).
Table 2 Percentage composition of fatty acids of LDL, HDL2 and HDL3 in 59 healthy subjects.
Women Men
Fatty acid LDL HDL2 HDL3 LDL HDL2 HDL3
14:0 0.80 ± 0.22 0.67 ± 0.19 0.61 ± 0.17 0.97 ± 0.29* 0.94 ± 1.41*** 0.76 ± 0.26*
16:0 19.50 ± 1.27 22.68 ± 1.41 22.45 ± 1.51 19.39 ± 1.23 22.79 ± 1.30 22.25 ± 1.10
16:1(n-7) 2.28 ± 0.64 1.72 ± 0.51 1.69 ± 0.51 2.22 ± 0.79 1.82 ± 0.71 1.87 ± 1.21
18:0 5.44 ± 0.58 8.18 ± 0.91 8.24 ± 0.90 5.71 ± 0.35 8.41 ± 0.67 8.71 ± 0.60
18:1T 0.35 ± 0.11 0.38 ± 0.12 0.36 ± 0.12 0.38 ± 0.15 0.45 ± 0.19 0.42 ± 0.16
18:1(n-9) 21.44 ± 1.63 17.79 ± 1.21 16.91 ± 1.02 21.12 ± 2.37 19.62 ± 3.09 17.77 ± 2.29
18:2(n-6) 34.92 ± 2.88 30.20 ± 2.93 30.71 ± 3.00 34.92 ± 4.16 28.38 ± 3.90 29.41 ± 3.72
18:3(n-3) 0.87 ± 0.22 0.67 ± 0.18 0.65 ± 0.18 0.87 ± 0.20 0.77 ± 0.23 0.69 ± 0.18
18:3(n-6) 0.43 ± 0.14 0.28 ± 0.09 0.28 ± 0.01 0.53 ± 0.26 0.34 ± 0.20 0.35 ± 0.20
20:0 0.31 ± 0.04 0.26 ± 0.03 0.22 ± 0.03 0.26 ± 0.04*** 0.23 ± 0.03*** 0.19 ± 0.02**
20:3(n-6) 1.22 ± 0.29 1.85 ± 0.44 1.97 ± 0.49 1.34 ± 0.26 1.86 ± 0.39 2.09 ± 0.40
20:4(n-6) 5.40 ± 0.91 7.08 ± 1.04 7.70 ± 1.20 5.53 ± 1.14 6.67 ± 1.42 7.64 ± 1.49
20:5(n-3) 1.12 ± 0.59 1.28 ± 0.67 1.38 ± 0.75 1.26 ± 0.59 1.33 ± 0.55 1.50 ± 0.59
22:0 0.91 ± 0.10 0.75 ± 0.13 0.63 ± 0.14 0.86 ± 0.11 0.60 ± 0.12 0.55 ± 0.13
22:5(n-3) 0.40 ± 0.11 0.65 ± 0.16 0.68 ± 0.16 0.47 ± 0.07** 0.74 ± 0.12* 0.82 ± 0.11**
22:6(n-3) 2.18 ± 0.50 3.46 ± 0.73 3.70 ± 0.79 1.96 ± 0.53 3.25 ± 0.86 3.37 ± 0.88
24:0 0.84 ± 0.07 0.65 ± 0.09 0.56 ± 0.08 0.83 ± 0.14 0.63 ± 0.13 0.56 ± 0.12
24:1(n-9) 1.59 ± 0.21 1.43 ± 0.23 1.23 ± 0.21 1.39 ± 0.28** 1.17 ± 0.24** 1.04 ± 0.19*
ΣSAFA 27.82 ± 1.21 33.21 ± 1.16 32.77 ± 1.23 27.88 ± 1.18 33.54 ± 1.50 32.94 ± 1.12
ΣMUFA 25.33 ± 2.03 20.98 ± 1.54 19.86 ± 1.34 24.48 ± 2.67 22.28 ± 3.16 20.51 ± 2.63
ΣPUFA 46.51 ± 2.55 45.44 ± 2.00 47.02 ± 1.82 47.26 ± 3.66 43.75 ± 4.13 46.14 ± 3.48
PI 83.2 ± 7.4 96.1 ± 8.3 101.2 ± 9.3 83.7 ± 9.2 92.2 ± 12.0 99.2 ± 10.9
Values are mean ± SD. * p < 0.05. ** p < 0.01, *** p < 0.001 compared to women. ΣSAFA, sum of percentages of saturated fatty acids, ΣMUFA, sum of percentages of monounsaturated fatty acids, ΣPUFA, sum of percentages of polyunsaturated fatty acids, PI, peroxidizability index (see methods).
The oxidation of LDL of all subjects gave rise to typical conjugated diene vs. time -curves, where the different phases of hydro peroxide formation were clearly discernible. Co incubations of LDL with either HDL2 or HDL3 produced biphasic profiles with faster oxidation in the beginning, followed by a slower rate and finally a faster propagation phase. The profiles looked similar in all participants.
Co incubation of LDL with HDL2 or HDL3 decreased the mean lag time of diene formation in both women and men (Table 3). Likewise, the mean propagation rate and the maximum diene concentration increased significantly in the presence of HDL. These oxidation parameters did not differ between women and men.
Table 3 Kinetic parameters of LDL, LDL + HDL2 and LDL + HDL3 oxidation in 59 healthy subjects.
Women Men All
LDL
Lag time (min) 60.9 ± 7.9 59.3 ± 6.9 60.2 ± 7.4
Rate (μmol/l/min)a 0.509 ± 0.067 0.502 ± 0.067 0.506 ± 0.066
Max (nmol/mg)b 541 ± 41 537 ± 51 539 ± 45
LDL + HDL2
Lag time (min) 56.0 ± 5.7*** 56.0 ± 6.7*** 56.0 ± 6.1***
Rate (μmol/l/min)a 0.687 ± 0.062*** 0.654 ± 0.084*** 0.671 ± 0.073***
Max (nmol/mg)b 774 ± 40*** 745 ± 71*** 762 ± 56***
LDL + HDL3
Lag time (min) 55.8 ± 5.4*** 54.8 ± 5.4*** 55.3 ± 5.4***
Rate (μmol/l/min)a 0.616 ± 0.064*** 0.607 ± 0.070*** 0.612 ± 0.066***
Max (nmol/mg)b 687 ± 40*** 674 ± 60*** 681 ± 50***
Values are mean ± SD. a Rate means maximal formation rate of conjugated dienes during oxidation. Calculation of the diene concentration is based on ε = 29500 of the conjugated dienes b Max is the maximal amount of dienes produced per mg of LDL protein. *** p < 0.001 in comparison with LDL alone in Wilcoxon's matched pairs test.
Multiple forward stepwise regression analysis was performed to estimate the effect of lipids and factors related to lipoprotein metabolism on oxidation parameters. Predictors for the multivariate analysis were selected on the basis of initial correlation analyses using Spearman's correlation coefficients. The resulting models formed consistent patterns of predictors. The results of the models for the mixtures of LDL + HDL2 are shown in Table 4. The results were similar for mixtures of LDL and HDL3, and for LDL alone (not shown). In these incubations, an increase in lag time was related to fasting blood glucose concentration, and a decrease in lag time was related to the peroxidizability index. Oxidation rate was positively associated with PUFA content of the lipoproteins. Maximum concentration of dienes was positively related to the content of linoleate and to the ratio of LDL to apoB, and negatively related to oleate.
Table 4 Multivariate regression models of factors predicting oxidation parameters in mixtures of LDL and HDL2.
Dependent variable Independent variable Standardized regression coefficient β p-Value Total model
LDL + HDL2
Lag time (min) fB-Glucose 0.410 0.00037 R2 = 0.296
PI of HDL2 -0.351 0.00276 p < 0.00005
LDL + HDL2
Oxidation rate (μmol/ml/min) ΣPUFA of LDL 0.424 0.0325 R2 = 0.57
ΣPUFA of HDL2 0.352 0.0129 p < 0.00000
LDL + HDL2
Maximum diene HDL2 18:2n-6 0.429 0.00015 R2 = 0.55
concentration (nmol/mg) LDL 18:1n-9 -0.292 0.0071 p < 0.00000
LDL/apoB 0.287 0.0033
Abbreviations: PI, peroxidizability index; ΣPUFA, sum of percentages of polyunsaturated fatty acids; LDL/apoB, the ratio of LDL cholesterol to apoB concentration. Stepwise forward regression analysis
Although the mean lag time was shorter in the presence of HDL2 or HDL3, there were nine subjects who had longer lag time when HDL2 was co incubated with LDL. We analyzed, whether there were any differences between these nine subjects and the rest of the study group that could explain the increased lag time. These nine subjects had a significantly smaller peroxidizability index of HDL2 (88.0 ± 12.3 vs. 95.4 ± 9.5, p < 0.05) than the rest of the group. Their LDL lag time was slightly shorter (55.7 ± 4.7 vs. 61.0 ± 7.7 μmol/l/min, p < 0.05) and their fasting blood glucose concentration was lower (4.1 ± 0.5 vs. 4.5 ± 0.5 mmol/l, p < 0.05).
Discussion
We found that co incubation of HDL2 or HDL3 with LDL in the presence of Cu2+ resulted in shortening of the mean lag time and acceleration of the oxidation rate in comparison with that of incubation of LDL alone. If lag time or propagation rate are thought of as indices of oxidation resistance, this outcome contradicts the role of HDL as an antioxidant. Our findings are in line with the studies by Bowry et al. [4], Suzukawa et al. [10], Schnitzer et al. [11], Ohmura et al. [12] and Raveh et al. [9], who have come to the conclusion that HDL is more easily oxidized than LDL. In other studies, HDL has appeared to be less prone to oxidation and even to protect LDL against copper-induced oxidation [3,13-15]. In the study of Kontush et al. [16] all subtractions of HDL exhibited limited capacities to protect LDL at early stages of oxidation. At later phases, small dense HDL particles were the most potent inhibitors of LDL oxidation under mildly oxidative conditions. If strongly oxidative conditions were used [5 μmol/l Cu2+), none of the HDL subtractions offered any protection to LDL. The results were fairly similar whether the subspecies were isolated from serum or EDTA-plasma despite their widely differing paraoxonase activities suggesting that paraoxonase may have had a smaller role in the inhibition. In our study the HDL2 subtraction of the majority of the subjects had properties that enhanced the onset of propagation phase. However, in 9 participants this phase was delayed in the presence of a moderate concentration of Cu2+ emphasizing that the individual variation of intrinsic characteristics of lipoproteins can not be overlooked.
In all, kinetic analyses by different investigators of the effects of HDL on copper-induced peroxidation of LDL are difficult to compare. (a) Firstly, the concentrations of LDL and HDL have been inconsistent and their expression has been variably based on protein, total lipid, total mass, phospholipid or cholesterol concentration, molar concentrations or particle numbers. The investigations into the kinetics of lipoprotein oxidation of Raveh et al. [9] and Ziouzenkova et al. [17] showed that the lag time and the propagation rate are dependent on LDL concentration. (b) Secondly, copper concentrations have also differed between the experiments. This has profound implications, since it has been shown in kinetic experiments that the lag time and the oxidation rate are correlated with the copper concentration until a saturating concentration is reached [9,16,18]. However, until more data are available, there is reason to think that the number of copper binding sites of lipoproteins is not constant but varies [18]. (c) Thirdly, the ratio of copper to lipoprotein has varied between studies. It has been found that the kinetic profile of LDL oxidation changes in response to copper concentration and that the familiar monophasic, auto accelerating profile is only obtained when the copper concentration is relatively high [17]. All of the compound kinetic curves in our study had a biphasic shape. The first phase of rapid oxidation in such a profile, whether observed with HDL or LDL, has been interpreted to occur via a tocopherol-mediated mechanism [9] where vitamin E acts as a prooxidant. Because the rate of oxidation is regulated by the ratio of bound copper/lipoprotein as outlined above, the addition of HDL to LDL should, theoretically, have lengthened the lag time and oxidation rate instead of shortening it, since HDL bound part of the copper. Obviously, many factors are involved in determining the outcome of this kind of experiment. Furthermore, it is apparent that ex vivo oxidization experiments with lipoproteins require standardization.
Earlier studies have shown that there are several intrinsic properties of lipoproteins that can affect their susceptibility to oxidation. Lipoprotein antioxidant content [19,20], fatty acid composition [21,22], presence of various enzymes [13] and LDL and HDL size [16,23] are among the factors that have been shown to have an impact on oxidation parameters, the former especially in supplementation studies. Also, long-term habitual diets with different fatty acid contents have been shown to influence LDL oxidation susceptibility [24]. We analyzed the fatty acid compositions of LDL, HDL2 and HDL3 particles. Fatty acids are highly intercorrelated, and therefore their use in multivariate analysis as predictors is problematic. We tried to overcome this difficulty by uniting the information in the fatty acid profiles into a single term – the peroxidizability index – which describes the combined reactivity of fatty acids towards reactive oxygen species [25]. The results (Table 4) show that this index was significantly larger in both HDL2 and HDL3 than in LDL in men as well as in women, suggesting that HDL particles might be more susceptible to oxidation than LDL. This opinion was further strengthened by the findings that in multiple regression analysis the peroxidizability index of LDL, HDL2 or HDL3 in combination with fasting blood glucose concentration were the best predictors of lag time when LDL was oxidized alone or in mixtures with HDL2 or HDL3, respectively. Furthermore, our finding of a smaller peroxidizability index in those subjects whose lag time lengthened in the presence of HDL2 is in line with the suggestion that the rate of oxidation is governed by the ratio of bound copper to oxidizable lipids [26]. The results of our experiments also confirm the findings of earlier studies and suggest that the proportions of polyunsaturated fatty acids as well as those of linoleic acid and oleic acid [27-29] are related to the oxidation rate and the amount of dienes formed during in vitro oxidation. We have no ready explanation as to why the glucose concentration had a positive correlation with lag time of oxidation especially since all our subjects were normoglycemic. It has been shown that LDL isolated from patients with poorly controlled type I diabetes is more susceptible to copper-induced oxidation than LDL from control subjects [30]. Consequently, it has been suggested that glycated LDL might be particularly prone to oxidation. Nonetheless, our result is more in line with the results obtained in well-controlled type I diabetics, where glycated LDL gave a longer lag time than that of nonglycated LDL [31].
Conclusion
In conclusion, we report that the lag time, the maximum propagation rate for the formation of dienes and the amount of dienes formed by Cu2+ -induced oxidation of LDL alone or in the presence of HDL2 or HDL3 do not differ between healthy men and women despite significant differences in lipid concentrations. Our findings do not support the concept that co incubation of LDL with HDL in the presence of divalent copper prevents its oxidative modification. Rather, our findings support previous results that in vitro HDL is oxidized fastest of all lipoproteins [4] partly because of its fatty acid composition, which is oxidation promoting.
Methods
Subjects
61 healthy subjects from the personnel and medical students of the Department of Medical Sciences of Tampere University and Tampere University Hospital volunteered. The age range of the subjects was 20 to 58 years. 33 were women and 28 were men. All participants filled in a questionnaire, where emphasis was given to their health status (diseases and use of medication) in addition to health related behavior (smoking, use of alcohol and vitamins). Ten subjects were current smokers and two abstained from alcohol. Fasting blood glucose concentration was ≤5.7 mmol/l in all subjects. Nine women and two men reported the use of vitamins and 12 women used hormone preparations. The results of two of the participants were later removed from analysis because of reported diseases Thus, 59 subjects remained. All participants gave their written consent to the study. The study protocol was approved by the ethics committee of the Tampere University Hospital.
Blood Samples
Fasting (12 h) blood samples were taken into suitable tubes (Vacuette, Greiner) from the antecubital vein in a sitting position after a 15-min rest using minimal stasis. Samples for the isolation of lipoproteins and for LDL size determination were taken into pre-chilled EDTA tubes, which were immediately placed in ice. Plasma was separated after centrifugation (Heraeus, 2000 xg, +4°C). EDTA plasmas were supplemented with sucrose (0.6 % w/v final concentration). This procedure has been shown to preserve LDL from oxidation for at least two months and the oxidation curve does not differ from that of a fresh sample [32]. All samples were kept frozen at -70°C until analyzed. Fasting blood glucose concentration was determined from capillary blood using Hemocue Glucose Analyzer (Hemocue, Ängelholm Sweden).
Analysis of Lipids and Lipoproteins
Cholesterol, HDL cholesterol, triacylglycerol, apoA-I and apoB concentrations were measured with Cobas Integra 700 automatic analyzer (Roche Diagnostics, Basel, Switzerland) using reagents and calibrators as recommended by the manufacturer. LDL cholesterol was calculated according to Friedewald. Lp(a) concentrations were analyzed by radioimmunoassay (Pharmacia, Uppsala Sweden) according to the manufacturer's instructions.
Isolation of Lipoproteins
Lipoproteins were fractionated by isopycnic density gradient ultracentrifugation using a Beckman SW40 Ti rotor in a Beckman L60 centrifuge (36000 rpm, 40 hours, 10°C). 2.0 ml of plasma was gently mixed with 4.0 ml of d 1.35 g/l NaCl-KBr solution in a polyallomer 14 × 95 mm tube. The mixture was then successively over layered with 4.5 ml of a d 1.006 salt solution and 1.0 ml of distilled water. The gradients were fractionated as described [33] and 0.4-ml fractions were collected. The fractions belonging to LDL, HDL2 and HDL3 were pooled on the basis of the absorbance curve. A part of the pooled fractions were immediately frozen to -70°C and a part was used for the oxidation experiments.
Oxidation of Lipoproteins
The susceptibility of LDL and mixtures of LDL and HDL subtractions to in vitro copper-catalyzed oxidation was assessed by the technique described in [34] as modified from Esterbauer et al. [35]. LDL (50 μg protein/ml ≈ 0.1 μM) was incubated either alone or mixed with autologous HDL2 (50 μg protein/ml ≈ 0.35 μM) or HDL3 (50 μg protein/ml ≈ 0.53 μM). The protein concentrations were determined using the method of Markwell et al. [36]. Oxidation was started by adding 10 μl of CuSO4 to a final concentration of 1.65 μM Cu2+. The spectrophotometer was computer-operated (UVWinlab 2.1). This program also collected the absorbance data at 2-min intervals during the oxidation. Several characteristic indices were obtained from the resulting absorbance versus time curves [32]. To control the in vitro oxidation procedure we prepared an LDL pool as described [37] and stored it at -70°C in 0.15-M NaCl/1 mM EDTA solution containing 0.6 % sucrose. One control LDL was analyzed in every oxidation run. The inter-assay coefficient of variation for lag time was 3.1 %. This LDL preparation was also used as a standard in gradient gel electrophoresis.
Electrophoretic Analysis of Lipoprotein Size
For the estimation of lipoprotein particle size in EDTA-plasma samples we used the nondenaturing gradient gel electrophoretic method of Krauss and Burke [38]. However, the 2–16 % polyacrylamide gels were cast in-house according to the instructions given by Pharmacia (Uppsala, Sweden) as described [39]. A control plasma sample (peak particle diameter 27.00 nm) stored at -70°C was included in every gel. The inter-assay coefficient of variation during this study was 1.1 %.
Fatty Acid Composition of Lipoproteins
The fatty acid compositions of the ultracentrifugally isolated LDL, HDL2 and HDL3 fractions were analyzed by capillary gas-liquid chromatography. Lipids were extracted with chloroform/methanol, partitioned, and the chloroform phase was dried under N2 [40]. The lipids were then transesterified with H2SO4 in dry methanol at 85°C for 2 h under N2. Following the addition of water, methyl esters of the fatty acids were extracted with petroleum ether and analyzed in a Shimadzu GC-14A gas chromatograph (Shimadzu Corporation, Kyoto, Japan) with a flame ionization detector using a Supelco SP 2560 capillary column (100 m, 0.25 mm I.D., 0.20 μm film thickness). The carrier gas was helium. The column temperature was held at 180°C for 15 min and thereafter programmed to increase at 3°C/min until 230°C and held at this temperature for 40 min. The individual fatty acids were identified with the aid of a standard mixture of methyl esters (Lipid standards 189-15 and 189-17, Sigma). The areas were measured with a Shimadzu C-R4A Chromatopac Integrator and the results expressed as percentages of the sum of all fatty acids from 14:0 to 24:1. As a control sample we used a pool of isolated HDL that was kept frozen at -70°C. The inter-assay coefficients of variation for the percentages of different fatty acids ranged from 0.3 to 4.4 %. From fatty acid compositions, the following indices were calculated: saturated fatty acids (SAFA) = Σ(%) of saturated fatty acids; monounsaturated fatty acids (MUFA) = Σ(%) of monoenoic fatty acids; polyunsaturated fatty acids (PUFA) = Σ(%) of polyunsaturated fatty acids; peroxidizability index (PI) = [(Σ mol% monoenoic FAs × 0.025) + (Σ mol% dienoic FAs × 1) + (Σ mol% trienoic FAs × 2) + (Σ mol% tetraenoic FAs × 4) + (Σ mol% pentaenoic FAs × 6) + (Σ mol% hexaenoic FAs × 8)] [25].
Statistical Analysis
Results are expressed as means ± standard deviation. Plasma triacylglycerol and Lp(a) concentrations were used as their logarithms but reported as their original results. Comparisons were conducted by analysis of variance or covariance and Mann-Whitney U-test. For pair wise comparisons we used Wilcoxon's matched pairs test. Univariate associations between variables were analyzed using Spearman's correlation coefficients. Predictors for the multivariate analysis were selected on the basis of the correlation analyses. Multivariate analysis was done using the stepwise forward linear regression technique. The Statistica for Windows (version 5.1) software package (Statsoft Inc., Oklahoma, USA) was used for the statistical analyses.
List of abbreviations
HDL, high density lipoprotein; LDL, low density lipoprotein; SAFA, saturated fatty acids; MUFA, monounsaturated fatty acids;, PUFA, polyunsaturated fatty acids; PI, peroxidizability index
Authors' contributions
TS and OJ conceived of the study, participated in its design, performed the statistical analysis and drafted the manuscript, NP and AS carried out the laboratory analyses, MS planned and analyzed the health questionnaire, TL, HJ and STN participated in the coordination of the study and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
The authors thank Marita Koli, Marja Jousimies and Marjo Virkki for expert laboratory assistance. This study was supported by the Medical Research Fund of Tampere University Hospital (TS, OJ, TL, HJ, STN), The Finnish Foundation of Cardiovascular Research (TL) and the Finnish Association of Clinical Biochemists (TS).
==== Refs
Nikkilä EA Letter: Serum high-density-lipoprotein and coronary heart-disease Lancet 1976 2 320 59895 10.1016/S0140-6736(76)90779-0
Sviridov D Nestel P Dynamics of reverse cholesterol transport: protection against atherosclerosis Atherosclerosis 2002 161 245 254 11888506 10.1016/S0021-9150(01)00677-3
Parthasarathy S Barnett J Fong LG High-density lipoprotein inhibits the oxidative modification of low-density lipoprotein Biochim Biophys Acta 1990 1044 275 283 2344447
Bowry VW Stanley KK Stocker R High density lipoprotein is the major carrier of lipid hydroperoxides in human blood plasma from fasting donors Proc Natl Acad Sci USA 1992 89 10316 10320 1332045
La Ville AE Sola R Balanya J Turner PR Masana L In vitro oxidised HDL is recognised by the scavenger receptor of macrophages: implications for its protective role in vivo Atherosclerosis 1994 105 179 189 8003094 10.1016/0021-9150(94)90048-5
Navab M Hama SY Reddy ST Ng CJ Van Lenten BJ Laks H Fogelman AM Oxidized lipids as mediators of coronary heart disease Curr Opin Lipidol 2002 13 363 372 12151851 10.1097/00041433-200208000-00003
Nakajima T Origuchi N Matsunaga T Kawai S Hokari S Nakamura H Inoue I Katayama S Nagata A Komoda T Localization of oxidized HDL in atheromatous plaques and oxidized HDL binding sites on human aortic endothelial cells Ann Clin Biochem 2000 37 179 186 10735361 10.1258/0004563001899186
Nagano Y Arai H Kita T High density lipoprotein loses its effect to stimulate efflux of cholesterol from foam cells after oxidative modification Proc Natl Acad Sci USA 1991 88 6457 6461 1862074
Raveh O Pinchuk I Schnitzer E Fainaru M Schaffer Z Lichtenberg D Kinetic analysis of copper-induced peroxidation of HDL, autoaccelerated and tocopherol-mediated peroxidation Free Rad Biol Med 2000 29 131 146 10980402 10.1016/S0891-5849(00)00332-4
Suzukawa M Ishikawa T Yoshida H Nakamura HJ Effect of in-vivo supplementation with low-dose vitamin E on susceptibility of low-density lipoprotein and high-density lipoprotein to oxidative modification J Am Coll Nutr 1995 14 46 52 7706609
Schnitzer E Pinchuk I Fainaru M Schafer Z Lichtenberg D Copper-induced lipid oxidation in unfractionated plasma: the lag preceding oxidation as a measure of oxidation-resistance Biochem Biophys Res Commun 1995 216 854 861 7488204 10.1006/bbrc.1995.2700
Ohmura H Watanabe Y Hastumi C Sato H Daida H Mokuno H Yamaguchi H Possible role of high susceptibility of high-density lipoprotein to lipid peroxidative modification and oxidized high-density lipoprotein in genesis of coronary artery spasm Atherosclerosis 1999 142 179 184 9920519 10.1016/S0021-9150(98)00235-4
Mackness MI Durrington PN HDL, its enzymes and its potential to influence lipid peroxidation Atherosclerosis 1995 115 243 253 7661883 10.1016/0021-9150(94)05524-M
Tribble DL Chu BM Gong EL Van-Venrooij F Nichols AV HDL antioxidant effects as assessed using a nonexchangeable probe to monitor particle-specific peroxidative stress in LDL-HDL mixtures J Lipid Res 1995 36 2580 2589 8847484
Kunitake ST Jarvis MR Hamilton RL Kane JP Binding of transition metals by apolipoprotein A-I-containing plasma lipoproteins: inhibition of oxidation of low density lipoproteins Proc Natl Acad Sci USA 1992 89 6993 6997 1495991
Kontush A Chantepie S Chapman MJ Small, dense HDL particles exert potent protection of atherogenic LDL against oxidative stress Arterioscler Thromb Vasc Biol 2003 23 1881 1888 12920049 10.1161/01.ATV.0000091338.93223.E8
Ziouzenkova O Gieseg SP Ramos P Esterbauer H Factors affecting resistance of low density lipoproteins to oxidation Lipids 1996 S71 76 8729097
Gieseg SP Esterbauer H Low density lipoprotein is saturable by pro-oxidant copper FEBS Lett 1994 343 188 194 8174699 10.1016/0014-5793(94)80553-9
Abbey M Nestel PJ Baghurst PA Antioxidant vitamins and low-density lipoprotein oxidation Am J Clin Nutr 1993 58 525 532 8379508
Goulinet S Chapman MJ Plasma LDL and HDL subspecies are heterogenous in particle content of tocopherols and oxygenated and hydrocarbon carotenoids. Relevance to oxidative resistance and atherogenesis Arterioscler Thromb Vasc Biol 1997 17 786 796 9108795
Reaven PD Parthasarathy S Grasse BJ Miller E Steinberg D Witztum JL Effects of oleate-rich and linoleate-rich diets on the susceptibility of low-density lipoproteins to modification in mildly hypercholesterolemic volunteers J Clin Invest 1993 91 668 676 8432867
Schnell JW Anderson RA Stegner JE Schindler SP Weinberg RB Effects of high polyunsaturated fat diet and vitamin E on high-density lipoprotein oxidation in humans Atherosclerosis 2001 159 459 466 11730827 10.1016/S0021-9150(01)00525-1
Chait A Brazg RL Tribble DL Krauss RM Susceptibility of small, dense low-density lipoproteins to oxidative modification in volunteers with the atherogenic lipoprotein phenotype, Pattern B Am J Med 1993 94 350 356 8475928 10.1016/0002-9343(93)90144-E
Korpela R Seppo L Laakso J Lilja J Karjala K Lähteenmäki T Solatunturi E Vapaatalo H Tikkanen MJ Dietary habits affect the susceptibility of low-density lipoprotein to oxidation Eur J Clin Nutr 1999 52 802 807 10556987 10.1038/sj.ejcn.1600860
Lambert AJ Portero-Otin M Pamplona R Merry BJ Effect of ageing and caloric restriction on specific markers of protein oxidative damage and membrane peroxidizability in rat liver mitochondria Mech Ageing Dev 2004 125 529 538 15336910 10.1016/j.mad.2004.06.002
Pinchuk I Schnitzer E Lichtenberg D Kinetic analysis of copper-induced peroxidation of LDL Biochim Biophys Acta 1998 1389 155 172 9461257
Bonanome A Pagnan A Biffanti S Opportuno A Sorgato F Dorella M Maiorino M Ursini F Effect of monounsaturated and polyunsaturated fatty acids on the susceptibility of plasma low density lipoproteins to oxidative modification Arterioscler Thromb 1992 12 529 533 1558840
Kleinveld HA Naber AH Stalenhoef AF Demacker PN Oxidation resistance, oxidation rate and extent of oxidation of human low-density lipoprotein depend on the ratio of oleic acid content to linoleic acid content: Studies in vitamin E deficient subjects Free Rad Biol Med 1993 15 273 280 8406127 10.1016/0891-5849(93)90074-5
Scheffer PG Bakker SJ Popp-Snijders C Heine RJ Schulgens RB Teerlink T Composition of LDL as determinant of its susceptibility to in vitro oxidation in patients with well-controlled type 2 diabetes Diabetes Metab Res Rev 2001 17 459 466 11757082 10.1002/dmrr.231
Tsai EC Hirsch IB Brunzell JD Chait A Reduced plasma peroxyl-trapping capacity and increased susceptibility of LDL to oxidation in poorly controlled IDDM Diabetes 1994 43 1010 1014 8039594
Jenkins AJ Thorpe SR Alderson NL Hermayer KL Lyons TJ King LP Chassereau CN Klein RL In vivo glycated low-density lipoprotein is not more susceptible to oxidation than nonglycated low-density lipoprotein in type I diabetes Metabolism 2004 53 969 976 15281003 10.1016/j.metabol.2004.01.002
Ramos P Gieseg SP Schuster B Esterbauer H Effect of temperature and phase transition on oxidation resistance of low density lipoprotein J Lipid Res 1995 36 2113 2128 8576638
Jaakkola O Solakivi T Tertov VV Orekhov AN Miettinen TA Nikkari T Characteristics of low-density lipoprotein subfractions from patients with coronary artery disease Coronary Artery Dis 1993 4 379 385 8261211
Leinonen JS Rantalaiho V Solakivi T Koivula T Wirta O Pasternack A Alho H Lehtimäki T Susceptibility of LDL to oxidation is not associated with the presence of coronary heart disease or renal dysfunction in NIDDM patients Clin Chim Acta 1998 275 163 174 9721074 10.1016/S0009-8981(98)00083-7
Esterbauer H Striegl G Puhl H Rotheneder M Continuous monitoring of in vitro oxidation of human low density lipoprotein Free Rad Res Commun 1989 6 67 75 2722022
Markwell MA Haas SM Tolbert NE Bieber LL Protein determination in membrane and lipoprotein samples: manual and automated procedures Methods Enzymol 1981 72 296 303 6796803
Jaakkola O Solakivi T Ylä-Herttuala S Nikkari T Receptor-mediated binding and degradation of subfractions of human plasma low-density lipoprotein in cultured fibroblasts Biochim Biophys Acta 1989 1005 118 122 2775766
Krauss RM Burke DJ Identification of multiple subfractions of plasma low density lipoproteins in normal humans J Lipid Res 1982 23 97 104 7057116
Siirtola A Antikainen M Ala-Houhala M Solakivi T Jokela H Lehtimäki T Holmberg C Salo MK Studies of LDL particle size and susceptibility to oxidation and association with glucose metabolism in children after heart transplantation J Heart Lung Transplant 2004 23 418 426 15063401 10.1016/S1053-2498(03)00201-8
Folch J Lees M Stanley GHS A simple method for the isolation and purification of total lipids from animal tissues J Biol Chem 1957 66 375 400 13428781
|
16242018
|
PMC1285367
|
CC BY
|
2021-01-04 16:39:19
|
no
|
Lipids Health Dis. 2005 Oct 20; 4:25
|
utf-8
|
Lipids Health Dis
| 2,005 |
10.1186/1476-511X-4-25
|
oa_comm
|
==== Front
Lipids Health DisLipids in Health and Disease1476-511XBioMed Central London 1476-511X-4-251624201810.1186/1476-511X-4-25ResearchHDL enhances oxidation of LDL in vitro in both men and women Solakivi T [email protected] O [email protected]äki A [email protected] N [email protected] S [email protected]äki T [email protected] H [email protected] ST [email protected] Department of Medical Biochemistry, University of Tampere, Medical School, Tampere, Finland2 Institute of Medical Technology, University of Tampere, Tampere, Finland3 Laboratory of Atherosclerosis Genetics, Department of Clinical Chemistry, Tampere University Hospital, Tampere, Finland4 Department of Internal Medicine, Tampere University Hospital, Tampere, Finland2005 20 10 2005 4 25 25 29 9 2005 20 10 2005 Copyright © 2005 Solakivi et al; licensee BioMed Central Ltd.2005Solakivi 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
Oxidative modification of low-density lipoprotein (LDL) is a key event in the oxidation hypothesis of atherogenesis. Some in vitro experiments have previously suggested that high-density lipoprotein (HDL) co-incubated with LDL prevents Cu2+-induced oxidation of LDL, while some other studies have observed an opposite effect. To comprehensively clarify the role of HDL in this context, we isolated LDL, HDL2 and HDL3 from sera of 61 free-living individuals (33 women and 28 men).
Results
When the isolated LDL was subjected to Cu2+-induced oxidation, both HDL2 and HDL3 particles increased the rate of appearance and the final concentration of conjugated dienes similarly in both genders. Oxidation rate was positively associated with polyunsaturated fatty acid content of the lipoproteins in that it was positively related to the content of linoleate and negatively related to oleate. More saturated fats thus protected the lipoproteins from damage.
Conclusion
We conclude that in vitro HDL does not protect LDL from oxidation, but is in fact oxidized fastest of all lipoproteins due to its fatty acid composition, which is oxidation promoting.
==== Body
Background
Epidemiological studies show an inverse correlation between high-density lipoprotein (HDL) concentration and the risk of developing coronary artery disease [1]. According to a widely accepted hypothesis, HDL or its subtractions play an important role in recruiting and transporting cholesterol from peripheral tissues to the liver for excretion, a series of events known as reverse cholesterol transport [2]. Other properties of HDL link its antiatherogenic functions to its antioxidative effects. Some studies have shown that co incubation of LDL with HDL in the presence of divalent copper prevents the oxidative modification of LDL [3]. In some reports this finding could not be confirmed, and in fact it has been demonstrated that in vitro HDL is oxidized faster than other lipoproteins [4]. When HDL is oxidatively modified, it alters to a form that causes macrophages to accumulate cholesterol. [5]. It has been suggested that systemic inflammation gives rise to prooxidant and proinflammatory HDL particles [6]. Oxidatively modified HDL is found in atheromatous plaques from human aorta [7]. Oxidatively modified HDL is no longer capable of removing cholesterol from cells, and it enhances LDL oxidation [8].
The contradictory findings on the role of HDL on LDL oxidation in vitro may be due to rather small study populations, and the reported heterogeneity of oxidation kinetics between lipoprotein preparations in vitro [9] which might be due to individual intrinsic properties of the lipoproteins. In the present paper we report the results of a study of the effect of HDL subtractions and gender on ex vivo oxidation of LDL from a population of 61 healthy free-living human subjects.
Results
Background characteristics of the men and women participating in the study are shown in Table 1. Compared with women, men had higher body mass indices, serum total cholesterol as well as LDL and apoB concentrations. Men had smaller LDL size and smaller concentrations of serum HDL and apoA-I than women.
Table 1 Characteristics of the study subjects.
Women Men All
N 32 27 59
Age (years) 39.3 ± 10.5 39.3 ± 11.0 39.3 ± 10.6
Body mass index (kg/m2) 23.2 ± 3.1 25.8 ± 3.6** 24.4 ± 3.6
Total cholesterol (mmol/l) 5.23 ± 0.87 5.76 ± 1.02* 5.47 ± 0.97
Triacylglycerol (mmol/l) 0.96 ± 0.41 1.81 ± 1.27*** 1.35 ± 0.71
HDL cholesterol (mmol/l) 1.88 ± 0.32 1.38 ± 0.23*** 1.65 ± 0.38
LDL cholesterol (mmol/l) 2.91 ± 0.84 3.63 ± 0.92** 3.21 ± 0.96
ApoA-I (g/l) 1.70 ± 0.20 1.46 ± 0.13*** 1.59 ± 0.21
ApoB (g/l) 0.81 ± 0.20 1.00 ± 0.22** 0.90 ± 0.23
Lp(a) (U/l) 266 ± 226 165 ± 201 220 ± 219
fB-Glucose (mmol/l)a 4.4 ± 0.4 4.6 ± 0.6 4.5 ± 0.5
LDL diameter (nm) 26.94 ± 0.45 26.40 ± 0.55*** 26.70 ± 0.51
Values are mean ± SD. a fB, fasting blood, * p < 0.05, ** p < 0.01, *** p < 0.001 compared to women by Mann-Whitney U-test.
The fatty acid compositions of ultracentrifugally isolated LDL, HDL2 and HDL3 fractions were analyzed by gas liquid chromatography (Table 2). There were no gender differences in the total amounts of saturated, monounsaturated and polyunsaturated fatty acids of LDL, HDL2 and HDL3. The calculated peroxidizability indices were also similar in men and women. In both genders this index increased significantly from LDL to HDL2 to HDL3 (p < 0.001, Wilcoxon's matched pairs test).
Table 2 Percentage composition of fatty acids of LDL, HDL2 and HDL3 in 59 healthy subjects.
Women Men
Fatty acid LDL HDL2 HDL3 LDL HDL2 HDL3
14:0 0.80 ± 0.22 0.67 ± 0.19 0.61 ± 0.17 0.97 ± 0.29* 0.94 ± 1.41*** 0.76 ± 0.26*
16:0 19.50 ± 1.27 22.68 ± 1.41 22.45 ± 1.51 19.39 ± 1.23 22.79 ± 1.30 22.25 ± 1.10
16:1(n-7) 2.28 ± 0.64 1.72 ± 0.51 1.69 ± 0.51 2.22 ± 0.79 1.82 ± 0.71 1.87 ± 1.21
18:0 5.44 ± 0.58 8.18 ± 0.91 8.24 ± 0.90 5.71 ± 0.35 8.41 ± 0.67 8.71 ± 0.60
18:1T 0.35 ± 0.11 0.38 ± 0.12 0.36 ± 0.12 0.38 ± 0.15 0.45 ± 0.19 0.42 ± 0.16
18:1(n-9) 21.44 ± 1.63 17.79 ± 1.21 16.91 ± 1.02 21.12 ± 2.37 19.62 ± 3.09 17.77 ± 2.29
18:2(n-6) 34.92 ± 2.88 30.20 ± 2.93 30.71 ± 3.00 34.92 ± 4.16 28.38 ± 3.90 29.41 ± 3.72
18:3(n-3) 0.87 ± 0.22 0.67 ± 0.18 0.65 ± 0.18 0.87 ± 0.20 0.77 ± 0.23 0.69 ± 0.18
18:3(n-6) 0.43 ± 0.14 0.28 ± 0.09 0.28 ± 0.01 0.53 ± 0.26 0.34 ± 0.20 0.35 ± 0.20
20:0 0.31 ± 0.04 0.26 ± 0.03 0.22 ± 0.03 0.26 ± 0.04*** 0.23 ± 0.03*** 0.19 ± 0.02**
20:3(n-6) 1.22 ± 0.29 1.85 ± 0.44 1.97 ± 0.49 1.34 ± 0.26 1.86 ± 0.39 2.09 ± 0.40
20:4(n-6) 5.40 ± 0.91 7.08 ± 1.04 7.70 ± 1.20 5.53 ± 1.14 6.67 ± 1.42 7.64 ± 1.49
20:5(n-3) 1.12 ± 0.59 1.28 ± 0.67 1.38 ± 0.75 1.26 ± 0.59 1.33 ± 0.55 1.50 ± 0.59
22:0 0.91 ± 0.10 0.75 ± 0.13 0.63 ± 0.14 0.86 ± 0.11 0.60 ± 0.12 0.55 ± 0.13
22:5(n-3) 0.40 ± 0.11 0.65 ± 0.16 0.68 ± 0.16 0.47 ± 0.07** 0.74 ± 0.12* 0.82 ± 0.11**
22:6(n-3) 2.18 ± 0.50 3.46 ± 0.73 3.70 ± 0.79 1.96 ± 0.53 3.25 ± 0.86 3.37 ± 0.88
24:0 0.84 ± 0.07 0.65 ± 0.09 0.56 ± 0.08 0.83 ± 0.14 0.63 ± 0.13 0.56 ± 0.12
24:1(n-9) 1.59 ± 0.21 1.43 ± 0.23 1.23 ± 0.21 1.39 ± 0.28** 1.17 ± 0.24** 1.04 ± 0.19*
ΣSAFA 27.82 ± 1.21 33.21 ± 1.16 32.77 ± 1.23 27.88 ± 1.18 33.54 ± 1.50 32.94 ± 1.12
ΣMUFA 25.33 ± 2.03 20.98 ± 1.54 19.86 ± 1.34 24.48 ± 2.67 22.28 ± 3.16 20.51 ± 2.63
ΣPUFA 46.51 ± 2.55 45.44 ± 2.00 47.02 ± 1.82 47.26 ± 3.66 43.75 ± 4.13 46.14 ± 3.48
PI 83.2 ± 7.4 96.1 ± 8.3 101.2 ± 9.3 83.7 ± 9.2 92.2 ± 12.0 99.2 ± 10.9
Values are mean ± SD. * p < 0.05. ** p < 0.01, *** p < 0.001 compared to women. ΣSAFA, sum of percentages of saturated fatty acids, ΣMUFA, sum of percentages of monounsaturated fatty acids, ΣPUFA, sum of percentages of polyunsaturated fatty acids, PI, peroxidizability index (see methods).
The oxidation of LDL of all subjects gave rise to typical conjugated diene vs. time -curves, where the different phases of hydro peroxide formation were clearly discernible. Co incubations of LDL with either HDL2 or HDL3 produced biphasic profiles with faster oxidation in the beginning, followed by a slower rate and finally a faster propagation phase. The profiles looked similar in all participants.
Co incubation of LDL with HDL2 or HDL3 decreased the mean lag time of diene formation in both women and men (Table 3). Likewise, the mean propagation rate and the maximum diene concentration increased significantly in the presence of HDL. These oxidation parameters did not differ between women and men.
Table 3 Kinetic parameters of LDL, LDL + HDL2 and LDL + HDL3 oxidation in 59 healthy subjects.
Women Men All
LDL
Lag time (min) 60.9 ± 7.9 59.3 ± 6.9 60.2 ± 7.4
Rate (μmol/l/min)a 0.509 ± 0.067 0.502 ± 0.067 0.506 ± 0.066
Max (nmol/mg)b 541 ± 41 537 ± 51 539 ± 45
LDL + HDL2
Lag time (min) 56.0 ± 5.7*** 56.0 ± 6.7*** 56.0 ± 6.1***
Rate (μmol/l/min)a 0.687 ± 0.062*** 0.654 ± 0.084*** 0.671 ± 0.073***
Max (nmol/mg)b 774 ± 40*** 745 ± 71*** 762 ± 56***
LDL + HDL3
Lag time (min) 55.8 ± 5.4*** 54.8 ± 5.4*** 55.3 ± 5.4***
Rate (μmol/l/min)a 0.616 ± 0.064*** 0.607 ± 0.070*** 0.612 ± 0.066***
Max (nmol/mg)b 687 ± 40*** 674 ± 60*** 681 ± 50***
Values are mean ± SD. a Rate means maximal formation rate of conjugated dienes during oxidation. Calculation of the diene concentration is based on ε = 29500 of the conjugated dienes b Max is the maximal amount of dienes produced per mg of LDL protein. *** p < 0.001 in comparison with LDL alone in Wilcoxon's matched pairs test.
Multiple forward stepwise regression analysis was performed to estimate the effect of lipids and factors related to lipoprotein metabolism on oxidation parameters. Predictors for the multivariate analysis were selected on the basis of initial correlation analyses using Spearman's correlation coefficients. The resulting models formed consistent patterns of predictors. The results of the models for the mixtures of LDL + HDL2 are shown in Table 4. The results were similar for mixtures of LDL and HDL3, and for LDL alone (not shown). In these incubations, an increase in lag time was related to fasting blood glucose concentration, and a decrease in lag time was related to the peroxidizability index. Oxidation rate was positively associated with PUFA content of the lipoproteins. Maximum concentration of dienes was positively related to the content of linoleate and to the ratio of LDL to apoB, and negatively related to oleate.
Table 4 Multivariate regression models of factors predicting oxidation parameters in mixtures of LDL and HDL2.
Dependent variable Independent variable Standardized regression coefficient β p-Value Total model
LDL + HDL2
Lag time (min) fB-Glucose 0.410 0.00037 R2 = 0.296
PI of HDL2 -0.351 0.00276 p < 0.00005
LDL + HDL2
Oxidation rate (μmol/ml/min) ΣPUFA of LDL 0.424 0.0325 R2 = 0.57
ΣPUFA of HDL2 0.352 0.0129 p < 0.00000
LDL + HDL2
Maximum diene HDL2 18:2n-6 0.429 0.00015 R2 = 0.55
concentration (nmol/mg) LDL 18:1n-9 -0.292 0.0071 p < 0.00000
LDL/apoB 0.287 0.0033
Abbreviations: PI, peroxidizability index; ΣPUFA, sum of percentages of polyunsaturated fatty acids; LDL/apoB, the ratio of LDL cholesterol to apoB concentration. Stepwise forward regression analysis
Although the mean lag time was shorter in the presence of HDL2 or HDL3, there were nine subjects who had longer lag time when HDL2 was co incubated with LDL. We analyzed, whether there were any differences between these nine subjects and the rest of the study group that could explain the increased lag time. These nine subjects had a significantly smaller peroxidizability index of HDL2 (88.0 ± 12.3 vs. 95.4 ± 9.5, p < 0.05) than the rest of the group. Their LDL lag time was slightly shorter (55.7 ± 4.7 vs. 61.0 ± 7.7 μmol/l/min, p < 0.05) and their fasting blood glucose concentration was lower (4.1 ± 0.5 vs. 4.5 ± 0.5 mmol/l, p < 0.05).
Discussion
We found that co incubation of HDL2 or HDL3 with LDL in the presence of Cu2+ resulted in shortening of the mean lag time and acceleration of the oxidation rate in comparison with that of incubation of LDL alone. If lag time or propagation rate are thought of as indices of oxidation resistance, this outcome contradicts the role of HDL as an antioxidant. Our findings are in line with the studies by Bowry et al. [4], Suzukawa et al. [10], Schnitzer et al. [11], Ohmura et al. [12] and Raveh et al. [9], who have come to the conclusion that HDL is more easily oxidized than LDL. In other studies, HDL has appeared to be less prone to oxidation and even to protect LDL against copper-induced oxidation [3,13-15]. In the study of Kontush et al. [16] all subtractions of HDL exhibited limited capacities to protect LDL at early stages of oxidation. At later phases, small dense HDL particles were the most potent inhibitors of LDL oxidation under mildly oxidative conditions. If strongly oxidative conditions were used [5 μmol/l Cu2+), none of the HDL subtractions offered any protection to LDL. The results were fairly similar whether the subspecies were isolated from serum or EDTA-plasma despite their widely differing paraoxonase activities suggesting that paraoxonase may have had a smaller role in the inhibition. In our study the HDL2 subtraction of the majority of the subjects had properties that enhanced the onset of propagation phase. However, in 9 participants this phase was delayed in the presence of a moderate concentration of Cu2+ emphasizing that the individual variation of intrinsic characteristics of lipoproteins can not be overlooked.
In all, kinetic analyses by different investigators of the effects of HDL on copper-induced peroxidation of LDL are difficult to compare. (a) Firstly, the concentrations of LDL and HDL have been inconsistent and their expression has been variably based on protein, total lipid, total mass, phospholipid or cholesterol concentration, molar concentrations or particle numbers. The investigations into the kinetics of lipoprotein oxidation of Raveh et al. [9] and Ziouzenkova et al. [17] showed that the lag time and the propagation rate are dependent on LDL concentration. (b) Secondly, copper concentrations have also differed between the experiments. This has profound implications, since it has been shown in kinetic experiments that the lag time and the oxidation rate are correlated with the copper concentration until a saturating concentration is reached [9,16,18]. However, until more data are available, there is reason to think that the number of copper binding sites of lipoproteins is not constant but varies [18]. (c) Thirdly, the ratio of copper to lipoprotein has varied between studies. It has been found that the kinetic profile of LDL oxidation changes in response to copper concentration and that the familiar monophasic, auto accelerating profile is only obtained when the copper concentration is relatively high [17]. All of the compound kinetic curves in our study had a biphasic shape. The first phase of rapid oxidation in such a profile, whether observed with HDL or LDL, has been interpreted to occur via a tocopherol-mediated mechanism [9] where vitamin E acts as a prooxidant. Because the rate of oxidation is regulated by the ratio of bound copper/lipoprotein as outlined above, the addition of HDL to LDL should, theoretically, have lengthened the lag time and oxidation rate instead of shortening it, since HDL bound part of the copper. Obviously, many factors are involved in determining the outcome of this kind of experiment. Furthermore, it is apparent that ex vivo oxidization experiments with lipoproteins require standardization.
Earlier studies have shown that there are several intrinsic properties of lipoproteins that can affect their susceptibility to oxidation. Lipoprotein antioxidant content [19,20], fatty acid composition [21,22], presence of various enzymes [13] and LDL and HDL size [16,23] are among the factors that have been shown to have an impact on oxidation parameters, the former especially in supplementation studies. Also, long-term habitual diets with different fatty acid contents have been shown to influence LDL oxidation susceptibility [24]. We analyzed the fatty acid compositions of LDL, HDL2 and HDL3 particles. Fatty acids are highly intercorrelated, and therefore their use in multivariate analysis as predictors is problematic. We tried to overcome this difficulty by uniting the information in the fatty acid profiles into a single term – the peroxidizability index – which describes the combined reactivity of fatty acids towards reactive oxygen species [25]. The results (Table 4) show that this index was significantly larger in both HDL2 and HDL3 than in LDL in men as well as in women, suggesting that HDL particles might be more susceptible to oxidation than LDL. This opinion was further strengthened by the findings that in multiple regression analysis the peroxidizability index of LDL, HDL2 or HDL3 in combination with fasting blood glucose concentration were the best predictors of lag time when LDL was oxidized alone or in mixtures with HDL2 or HDL3, respectively. Furthermore, our finding of a smaller peroxidizability index in those subjects whose lag time lengthened in the presence of HDL2 is in line with the suggestion that the rate of oxidation is governed by the ratio of bound copper to oxidizable lipids [26]. The results of our experiments also confirm the findings of earlier studies and suggest that the proportions of polyunsaturated fatty acids as well as those of linoleic acid and oleic acid [27-29] are related to the oxidation rate and the amount of dienes formed during in vitro oxidation. We have no ready explanation as to why the glucose concentration had a positive correlation with lag time of oxidation especially since all our subjects were normoglycemic. It has been shown that LDL isolated from patients with poorly controlled type I diabetes is more susceptible to copper-induced oxidation than LDL from control subjects [30]. Consequently, it has been suggested that glycated LDL might be particularly prone to oxidation. Nonetheless, our result is more in line with the results obtained in well-controlled type I diabetics, where glycated LDL gave a longer lag time than that of nonglycated LDL [31].
Conclusion
In conclusion, we report that the lag time, the maximum propagation rate for the formation of dienes and the amount of dienes formed by Cu2+ -induced oxidation of LDL alone or in the presence of HDL2 or HDL3 do not differ between healthy men and women despite significant differences in lipid concentrations. Our findings do not support the concept that co incubation of LDL with HDL in the presence of divalent copper prevents its oxidative modification. Rather, our findings support previous results that in vitro HDL is oxidized fastest of all lipoproteins [4] partly because of its fatty acid composition, which is oxidation promoting.
Methods
Subjects
61 healthy subjects from the personnel and medical students of the Department of Medical Sciences of Tampere University and Tampere University Hospital volunteered. The age range of the subjects was 20 to 58 years. 33 were women and 28 were men. All participants filled in a questionnaire, where emphasis was given to their health status (diseases and use of medication) in addition to health related behavior (smoking, use of alcohol and vitamins). Ten subjects were current smokers and two abstained from alcohol. Fasting blood glucose concentration was ≤5.7 mmol/l in all subjects. Nine women and two men reported the use of vitamins and 12 women used hormone preparations. The results of two of the participants were later removed from analysis because of reported diseases Thus, 59 subjects remained. All participants gave their written consent to the study. The study protocol was approved by the ethics committee of the Tampere University Hospital.
Blood Samples
Fasting (12 h) blood samples were taken into suitable tubes (Vacuette, Greiner) from the antecubital vein in a sitting position after a 15-min rest using minimal stasis. Samples for the isolation of lipoproteins and for LDL size determination were taken into pre-chilled EDTA tubes, which were immediately placed in ice. Plasma was separated after centrifugation (Heraeus, 2000 xg, +4°C). EDTA plasmas were supplemented with sucrose (0.6 % w/v final concentration). This procedure has been shown to preserve LDL from oxidation for at least two months and the oxidation curve does not differ from that of a fresh sample [32]. All samples were kept frozen at -70°C until analyzed. Fasting blood glucose concentration was determined from capillary blood using Hemocue Glucose Analyzer (Hemocue, Ängelholm Sweden).
Analysis of Lipids and Lipoproteins
Cholesterol, HDL cholesterol, triacylglycerol, apoA-I and apoB concentrations were measured with Cobas Integra 700 automatic analyzer (Roche Diagnostics, Basel, Switzerland) using reagents and calibrators as recommended by the manufacturer. LDL cholesterol was calculated according to Friedewald. Lp(a) concentrations were analyzed by radioimmunoassay (Pharmacia, Uppsala Sweden) according to the manufacturer's instructions.
Isolation of Lipoproteins
Lipoproteins were fractionated by isopycnic density gradient ultracentrifugation using a Beckman SW40 Ti rotor in a Beckman L60 centrifuge (36000 rpm, 40 hours, 10°C). 2.0 ml of plasma was gently mixed with 4.0 ml of d 1.35 g/l NaCl-KBr solution in a polyallomer 14 × 95 mm tube. The mixture was then successively over layered with 4.5 ml of a d 1.006 salt solution and 1.0 ml of distilled water. The gradients were fractionated as described [33] and 0.4-ml fractions were collected. The fractions belonging to LDL, HDL2 and HDL3 were pooled on the basis of the absorbance curve. A part of the pooled fractions were immediately frozen to -70°C and a part was used for the oxidation experiments.
Oxidation of Lipoproteins
The susceptibility of LDL and mixtures of LDL and HDL subtractions to in vitro copper-catalyzed oxidation was assessed by the technique described in [34] as modified from Esterbauer et al. [35]. LDL (50 μg protein/ml ≈ 0.1 μM) was incubated either alone or mixed with autologous HDL2 (50 μg protein/ml ≈ 0.35 μM) or HDL3 (50 μg protein/ml ≈ 0.53 μM). The protein concentrations were determined using the method of Markwell et al. [36]. Oxidation was started by adding 10 μl of CuSO4 to a final concentration of 1.65 μM Cu2+. The spectrophotometer was computer-operated (UVWinlab 2.1). This program also collected the absorbance data at 2-min intervals during the oxidation. Several characteristic indices were obtained from the resulting absorbance versus time curves [32]. To control the in vitro oxidation procedure we prepared an LDL pool as described [37] and stored it at -70°C in 0.15-M NaCl/1 mM EDTA solution containing 0.6 % sucrose. One control LDL was analyzed in every oxidation run. The inter-assay coefficient of variation for lag time was 3.1 %. This LDL preparation was also used as a standard in gradient gel electrophoresis.
Electrophoretic Analysis of Lipoprotein Size
For the estimation of lipoprotein particle size in EDTA-plasma samples we used the nondenaturing gradient gel electrophoretic method of Krauss and Burke [38]. However, the 2–16 % polyacrylamide gels were cast in-house according to the instructions given by Pharmacia (Uppsala, Sweden) as described [39]. A control plasma sample (peak particle diameter 27.00 nm) stored at -70°C was included in every gel. The inter-assay coefficient of variation during this study was 1.1 %.
Fatty Acid Composition of Lipoproteins
The fatty acid compositions of the ultracentrifugally isolated LDL, HDL2 and HDL3 fractions were analyzed by capillary gas-liquid chromatography. Lipids were extracted with chloroform/methanol, partitioned, and the chloroform phase was dried under N2 [40]. The lipids were then transesterified with H2SO4 in dry methanol at 85°C for 2 h under N2. Following the addition of water, methyl esters of the fatty acids were extracted with petroleum ether and analyzed in a Shimadzu GC-14A gas chromatograph (Shimadzu Corporation, Kyoto, Japan) with a flame ionization detector using a Supelco SP 2560 capillary column (100 m, 0.25 mm I.D., 0.20 μm film thickness). The carrier gas was helium. The column temperature was held at 180°C for 15 min and thereafter programmed to increase at 3°C/min until 230°C and held at this temperature for 40 min. The individual fatty acids were identified with the aid of a standard mixture of methyl esters (Lipid standards 189-15 and 189-17, Sigma). The areas were measured with a Shimadzu C-R4A Chromatopac Integrator and the results expressed as percentages of the sum of all fatty acids from 14:0 to 24:1. As a control sample we used a pool of isolated HDL that was kept frozen at -70°C. The inter-assay coefficients of variation for the percentages of different fatty acids ranged from 0.3 to 4.4 %. From fatty acid compositions, the following indices were calculated: saturated fatty acids (SAFA) = Σ(%) of saturated fatty acids; monounsaturated fatty acids (MUFA) = Σ(%) of monoenoic fatty acids; polyunsaturated fatty acids (PUFA) = Σ(%) of polyunsaturated fatty acids; peroxidizability index (PI) = [(Σ mol% monoenoic FAs × 0.025) + (Σ mol% dienoic FAs × 1) + (Σ mol% trienoic FAs × 2) + (Σ mol% tetraenoic FAs × 4) + (Σ mol% pentaenoic FAs × 6) + (Σ mol% hexaenoic FAs × 8)] [25].
Statistical Analysis
Results are expressed as means ± standard deviation. Plasma triacylglycerol and Lp(a) concentrations were used as their logarithms but reported as their original results. Comparisons were conducted by analysis of variance or covariance and Mann-Whitney U-test. For pair wise comparisons we used Wilcoxon's matched pairs test. Univariate associations between variables were analyzed using Spearman's correlation coefficients. Predictors for the multivariate analysis were selected on the basis of the correlation analyses. Multivariate analysis was done using the stepwise forward linear regression technique. The Statistica for Windows (version 5.1) software package (Statsoft Inc., Oklahoma, USA) was used for the statistical analyses.
List of abbreviations
HDL, high density lipoprotein; LDL, low density lipoprotein; SAFA, saturated fatty acids; MUFA, monounsaturated fatty acids;, PUFA, polyunsaturated fatty acids; PI, peroxidizability index
Authors' contributions
TS and OJ conceived of the study, participated in its design, performed the statistical analysis and drafted the manuscript, NP and AS carried out the laboratory analyses, MS planned and analyzed the health questionnaire, TL, HJ and STN participated in the coordination of the study and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
The authors thank Marita Koli, Marja Jousimies and Marjo Virkki for expert laboratory assistance. This study was supported by the Medical Research Fund of Tampere University Hospital (TS, OJ, TL, HJ, STN), The Finnish Foundation of Cardiovascular Research (TL) and the Finnish Association of Clinical Biochemists (TS).
==== Refs
Nikkilä EA Letter: Serum high-density-lipoprotein and coronary heart-disease Lancet 1976 2 320 59895 10.1016/S0140-6736(76)90779-0
Sviridov D Nestel P Dynamics of reverse cholesterol transport: protection against atherosclerosis Atherosclerosis 2002 161 245 254 11888506 10.1016/S0021-9150(01)00677-3
Parthasarathy S Barnett J Fong LG High-density lipoprotein inhibits the oxidative modification of low-density lipoprotein Biochim Biophys Acta 1990 1044 275 283 2344447
Bowry VW Stanley KK Stocker R High density lipoprotein is the major carrier of lipid hydroperoxides in human blood plasma from fasting donors Proc Natl Acad Sci USA 1992 89 10316 10320 1332045
La Ville AE Sola R Balanya J Turner PR Masana L In vitro oxidised HDL is recognised by the scavenger receptor of macrophages: implications for its protective role in vivo Atherosclerosis 1994 105 179 189 8003094 10.1016/0021-9150(94)90048-5
Navab M Hama SY Reddy ST Ng CJ Van Lenten BJ Laks H Fogelman AM Oxidized lipids as mediators of coronary heart disease Curr Opin Lipidol 2002 13 363 372 12151851 10.1097/00041433-200208000-00003
Nakajima T Origuchi N Matsunaga T Kawai S Hokari S Nakamura H Inoue I Katayama S Nagata A Komoda T Localization of oxidized HDL in atheromatous plaques and oxidized HDL binding sites on human aortic endothelial cells Ann Clin Biochem 2000 37 179 186 10735361 10.1258/0004563001899186
Nagano Y Arai H Kita T High density lipoprotein loses its effect to stimulate efflux of cholesterol from foam cells after oxidative modification Proc Natl Acad Sci USA 1991 88 6457 6461 1862074
Raveh O Pinchuk I Schnitzer E Fainaru M Schaffer Z Lichtenberg D Kinetic analysis of copper-induced peroxidation of HDL, autoaccelerated and tocopherol-mediated peroxidation Free Rad Biol Med 2000 29 131 146 10980402 10.1016/S0891-5849(00)00332-4
Suzukawa M Ishikawa T Yoshida H Nakamura HJ Effect of in-vivo supplementation with low-dose vitamin E on susceptibility of low-density lipoprotein and high-density lipoprotein to oxidative modification J Am Coll Nutr 1995 14 46 52 7706609
Schnitzer E Pinchuk I Fainaru M Schafer Z Lichtenberg D Copper-induced lipid oxidation in unfractionated plasma: the lag preceding oxidation as a measure of oxidation-resistance Biochem Biophys Res Commun 1995 216 854 861 7488204 10.1006/bbrc.1995.2700
Ohmura H Watanabe Y Hastumi C Sato H Daida H Mokuno H Yamaguchi H Possible role of high susceptibility of high-density lipoprotein to lipid peroxidative modification and oxidized high-density lipoprotein in genesis of coronary artery spasm Atherosclerosis 1999 142 179 184 9920519 10.1016/S0021-9150(98)00235-4
Mackness MI Durrington PN HDL, its enzymes and its potential to influence lipid peroxidation Atherosclerosis 1995 115 243 253 7661883 10.1016/0021-9150(94)05524-M
Tribble DL Chu BM Gong EL Van-Venrooij F Nichols AV HDL antioxidant effects as assessed using a nonexchangeable probe to monitor particle-specific peroxidative stress in LDL-HDL mixtures J Lipid Res 1995 36 2580 2589 8847484
Kunitake ST Jarvis MR Hamilton RL Kane JP Binding of transition metals by apolipoprotein A-I-containing plasma lipoproteins: inhibition of oxidation of low density lipoproteins Proc Natl Acad Sci USA 1992 89 6993 6997 1495991
Kontush A Chantepie S Chapman MJ Small, dense HDL particles exert potent protection of atherogenic LDL against oxidative stress Arterioscler Thromb Vasc Biol 2003 23 1881 1888 12920049 10.1161/01.ATV.0000091338.93223.E8
Ziouzenkova O Gieseg SP Ramos P Esterbauer H Factors affecting resistance of low density lipoproteins to oxidation Lipids 1996 S71 76 8729097
Gieseg SP Esterbauer H Low density lipoprotein is saturable by pro-oxidant copper FEBS Lett 1994 343 188 194 8174699 10.1016/0014-5793(94)80553-9
Abbey M Nestel PJ Baghurst PA Antioxidant vitamins and low-density lipoprotein oxidation Am J Clin Nutr 1993 58 525 532 8379508
Goulinet S Chapman MJ Plasma LDL and HDL subspecies are heterogenous in particle content of tocopherols and oxygenated and hydrocarbon carotenoids. Relevance to oxidative resistance and atherogenesis Arterioscler Thromb Vasc Biol 1997 17 786 796 9108795
Reaven PD Parthasarathy S Grasse BJ Miller E Steinberg D Witztum JL Effects of oleate-rich and linoleate-rich diets on the susceptibility of low-density lipoproteins to modification in mildly hypercholesterolemic volunteers J Clin Invest 1993 91 668 676 8432867
Schnell JW Anderson RA Stegner JE Schindler SP Weinberg RB Effects of high polyunsaturated fat diet and vitamin E on high-density lipoprotein oxidation in humans Atherosclerosis 2001 159 459 466 11730827 10.1016/S0021-9150(01)00525-1
Chait A Brazg RL Tribble DL Krauss RM Susceptibility of small, dense low-density lipoproteins to oxidative modification in volunteers with the atherogenic lipoprotein phenotype, Pattern B Am J Med 1993 94 350 356 8475928 10.1016/0002-9343(93)90144-E
Korpela R Seppo L Laakso J Lilja J Karjala K Lähteenmäki T Solatunturi E Vapaatalo H Tikkanen MJ Dietary habits affect the susceptibility of low-density lipoprotein to oxidation Eur J Clin Nutr 1999 52 802 807 10556987 10.1038/sj.ejcn.1600860
Lambert AJ Portero-Otin M Pamplona R Merry BJ Effect of ageing and caloric restriction on specific markers of protein oxidative damage and membrane peroxidizability in rat liver mitochondria Mech Ageing Dev 2004 125 529 538 15336910 10.1016/j.mad.2004.06.002
Pinchuk I Schnitzer E Lichtenberg D Kinetic analysis of copper-induced peroxidation of LDL Biochim Biophys Acta 1998 1389 155 172 9461257
Bonanome A Pagnan A Biffanti S Opportuno A Sorgato F Dorella M Maiorino M Ursini F Effect of monounsaturated and polyunsaturated fatty acids on the susceptibility of plasma low density lipoproteins to oxidative modification Arterioscler Thromb 1992 12 529 533 1558840
Kleinveld HA Naber AH Stalenhoef AF Demacker PN Oxidation resistance, oxidation rate and extent of oxidation of human low-density lipoprotein depend on the ratio of oleic acid content to linoleic acid content: Studies in vitamin E deficient subjects Free Rad Biol Med 1993 15 273 280 8406127 10.1016/0891-5849(93)90074-5
Scheffer PG Bakker SJ Popp-Snijders C Heine RJ Schulgens RB Teerlink T Composition of LDL as determinant of its susceptibility to in vitro oxidation in patients with well-controlled type 2 diabetes Diabetes Metab Res Rev 2001 17 459 466 11757082 10.1002/dmrr.231
Tsai EC Hirsch IB Brunzell JD Chait A Reduced plasma peroxyl-trapping capacity and increased susceptibility of LDL to oxidation in poorly controlled IDDM Diabetes 1994 43 1010 1014 8039594
Jenkins AJ Thorpe SR Alderson NL Hermayer KL Lyons TJ King LP Chassereau CN Klein RL In vivo glycated low-density lipoprotein is not more susceptible to oxidation than nonglycated low-density lipoprotein in type I diabetes Metabolism 2004 53 969 976 15281003 10.1016/j.metabol.2004.01.002
Ramos P Gieseg SP Schuster B Esterbauer H Effect of temperature and phase transition on oxidation resistance of low density lipoprotein J Lipid Res 1995 36 2113 2128 8576638
Jaakkola O Solakivi T Tertov VV Orekhov AN Miettinen TA Nikkari T Characteristics of low-density lipoprotein subfractions from patients with coronary artery disease Coronary Artery Dis 1993 4 379 385 8261211
Leinonen JS Rantalaiho V Solakivi T Koivula T Wirta O Pasternack A Alho H Lehtimäki T Susceptibility of LDL to oxidation is not associated with the presence of coronary heart disease or renal dysfunction in NIDDM patients Clin Chim Acta 1998 275 163 174 9721074 10.1016/S0009-8981(98)00083-7
Esterbauer H Striegl G Puhl H Rotheneder M Continuous monitoring of in vitro oxidation of human low density lipoprotein Free Rad Res Commun 1989 6 67 75 2722022
Markwell MA Haas SM Tolbert NE Bieber LL Protein determination in membrane and lipoprotein samples: manual and automated procedures Methods Enzymol 1981 72 296 303 6796803
Jaakkola O Solakivi T Ylä-Herttuala S Nikkari T Receptor-mediated binding and degradation of subfractions of human plasma low-density lipoprotein in cultured fibroblasts Biochim Biophys Acta 1989 1005 118 122 2775766
Krauss RM Burke DJ Identification of multiple subfractions of plasma low density lipoproteins in normal humans J Lipid Res 1982 23 97 104 7057116
Siirtola A Antikainen M Ala-Houhala M Solakivi T Jokela H Lehtimäki T Holmberg C Salo MK Studies of LDL particle size and susceptibility to oxidation and association with glucose metabolism in children after heart transplantation J Heart Lung Transplant 2004 23 418 426 15063401 10.1016/S1053-2498(03)00201-8
Folch J Lees M Stanley GHS A simple method for the isolation and purification of total lipids from animal tissues J Biol Chem 1957 66 375 400 13428781
|
16305754
|
PMC1287063
|
CC BY
|
2021-01-04 16:37:20
|
no
|
Immunome Res. 2005 Sep 20; 1:1
|
latin-1
|
Immunome Res
| 2,005 |
10.1186/1745-7580-1-1
|
oa_comm
|
==== Front
Immunome ResImmunome Research1745-7580BioMed Central London 1745-7580-1-21630575510.1186/1745-7580-1-2DatabaseAn ontology for immune epitopes: application to the design of a broad scope database of immune reactivities Sathiamurthy Muthuraman [email protected] Bjoern [email protected] Huynh-Hoa [email protected] John [email protected] John [email protected] Stephen S [email protected] Ward [email protected] Deborah L [email protected] Philip E [email protected] Alessandro [email protected] La Jolla Institute of Allergy and Immunology, 3030 Bunker Hill Street, Suite 326, San Diego, California, 92109, USA2 Knowledge Systems, Artificial Intelligence Laboratory, Stanford University and McGuinness Associates, Stanford, CA 94305, USA3 San Diego Supercomputer Center, P.O. Box 85608, San Diego, California 92186-5608, USA2005 20 9 2005 1 2 2 26 6 2005 20 9 2005 Copyright © 2005 Sathiamurthy et al; licensee BioMed Central Ltd.2005Sathiamurthy 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
Epitopes can be defined as the molecular structures bound by specific receptors, which are recognized during immune responses. The Immune Epitope Database and Analysis Resource (IEDB) project will catalog and organize information regarding antibody and T cell epitopes from infectious pathogens, experimental antigens and self-antigens, with a priority on NIAID Category A-C pathogens () and emerging/re-emerging infectious diseases. Both intrinsic structural and phylogenetic features, as well as information relating to the interactions of the epitopes with the host's immune system will be catalogued.
Description
To effectively represent and communicate the information related to immune epitopes, a formal ontology was developed. The semantics of the epitope domain and related concepts were captured as a hierarchy of classes, which represent the general and specialized relationships between the various concepts. A complete listing of classes and their properties can be found at .
Conclusion
The IEDB's ontology is the first ontology specifically designed to capture both intrinsic chemical and biochemical information relating to immune epitopes with information relating to the interaction of these structures with molecules derived from the host immune system. We anticipate that the development of this type of ontology and associated databases will facilitate rigorous description of data related to immune epitopes, and might ultimately lead to completely new methods for describing and modeling immune responses.
==== Body
Background
An epitope can be defined as the molecular structure recognized by the products of immune responses. According to this definition, epitopes are the specific molecular entities engaged in binding to antibody molecules or specific T cell receptors. An extended definition also includes the specific molecules binding in the peptide binding sites of MHC receptors. We have previously described [1] the general design of the Immune Epitope Database and Analysis Resource (IEDB), a broad program recently initiated by National Institute of Allergy and Infectious Diseases (NIAID). The overall goal of the IEDB is to catalog and organize a large body of information regarding antibody and T cell epitopes from infectious pathogens and other sources [2]. Priority will be placed on NIAID Category A-C pathogens () and emerging/re-emerging infectious diseases. Epitopes of human and non-human primates, rodents, and other species for which detailed information is available will be included. It is envisioned that this new effort will catalyze the development of new methods to predict and model immune responses, will aid in the discovery and development of new vaccines and diagnostics, and will assist in basic immunological investigations.
The IEDB will catalog structural and phylogenetic information about epitopes, information about their capacity to bind to specific receptors (i.e. MHC, TCR, BCR, Antibodies), as well as the type of immune response observed following engagement of the receptors (RFP-NIH-NIAID-DAIT-03/31: ).
In broad terms, the database will contain two general categories of data and information associated with immune epitopes – intrinsic and extrinsic (context-dependent data). Intrinsic features of an epitope are those characteristics that can be unequivocally defined and are specified within the epitope sequence/structure itself. Examples of intrinsic features are the epitope's sequence, structural features, and binding interactions with other immune system molecules. To describe an immune response associated with a specific epitope, context information also needs to be taken into account. Contextual information includes, for example, the species of the host, the route and dose of immunization, the health status and genetic makeup of the host, and the presence of adjuvants. In this respect, the IEDB project transcends the strict boundaries of database development and reaches into a systems biology application, attempting for the first time to integrate structural information about epitopes with comprehensive details describing their complex interaction with the immune system of the host, be it an infected organism or a vaccine recipient [1-3].
For these reasons, it was apparent at the outset of the project that it was crucial to develop a rigorous conceptual framework to represent the knowledge related to the epitopes. Such a framework was key to sharing information and ideas among developers, scientists, and potential users, and to allowing the design of an effective logical structure of the database itself. Accordingly, we decided to develop a formal ontology. Over the years, the term "ontology" has been defined and utilized in many ways by the knowledge engineering community [4]. We will adopt the definition of "ontology" as "the explicit formal specifications of the terms in a domain and the relationships among them" [5]. According to Noy and McGuinness [6], "ontology defines a common vocabulary for researchers who need to share information in a domain and helps separate domain knowledge from operational knowledge". Thus, availability of a formal ontology is relevant in designing a database, in cataloging the information, in communicating the database structure to researchers, developers and users, and in integrating multiple database schema designs and applications.
Several existing databases catalog epitope related data. We gratefully acknowledge that we have been able use these previous experiences in the design and implementation of the IEDB. MHCPEP [7], SYFPEITHI [8], FIMM [9], HLA Ligand Database [10], HIV Immunology Database [11], JenPep [12], AntiJen [13], and MHCBN [14] are all publicly available epitope related databases. In general, these databases provide information relating to epitopes, but do not catalog in-depth information relating to their interactions with the host's immune system. It should also be noted that none of these databases has published a formal ontology, but all of them rely on informal or implicit ontologies. We have taken into account as much as possible these ontologies, inferring their structure by informal communications with database developers or perusal of the databases websites.
The ontology developed for IEDB and described herein complements two explicit ontologies that are presently available: the IMGT-Ontology and the Gene Ontology (GO). The IMGT-Ontology [15] was created for the international ImMunoGeneTics Database (IMGT), which is an integrated database specializing in antigen receptors (immunoglobulin and T Cell receptors) and MHC molecules of all vertebrate species. This is, to the best of our knowledge, the first ontology in the domain of immunogenetics and immunoinformatics. The GO project [16] provides structured, controlled vocabularies that cover several domains of molecular and cellular biology. GO provides an excellent framework for genes, gene products, and their sequences, but it does not address the specific epitope substructure of the gene products. The IMGT provides an excellent ontological framework for the immune receptors but lacks information relating to the epitopes themselves. Therefore it was necessary to expand the available ontologies and to create an ontology specifically designed to represent the information of immune receptor interaction with immune epitopes. Wherever possible, the IEDB ontology conforms to standard vocabularies for capturing values for certain fields. For capturing disease names, IEDB uses the International Classification for Diseases (ICD-10) [17]. The NCBI Taxonomy database nomenclature [18,19] is used to capture species and strain names, and HLA Allele names are consistent with the HLA nomenclature reports [20].
Construction and Content
The IEDB is being developed as a web-accessible database using Oracle 10g and Enterprise Java (J2EE). Industry standard software design has been followed and it is expected that IEDB will be available for public users by the end of 2005.
Protégé was used to design and document the IEDB ontology. Protégé is a free, open source ontology editor and knowledge-base framework, written in Java. It provides an environment for creating ontologies and the terms used in those ontologies. Protégé supports class, slot, and instance creation, allowing users to specify relationships between appropriate entities. Two features that IEDB ontology effort used extensively were Protégé's support for creating ontology terms and for viewing the term hierarchies and the definitions. The support for a central repository on ontologies, along with browsing support, is key in reviewing and reusing ontologies.
While there are several open source tools available [21] for developing ontologies, we selected Protégé because of its extensibility to a variety of plug-ins that are readily available for integration. It also has the ability to export to different formats including the Ontology Web Language (OWL) (), which allows interoperability with other ontologies.
We have previously described some of the general concepts relating to the IEDB design [1,2]. More information relating to various aspects of the project can be accessed at . Herein, we report a detailed description of the novel aspects of the IEDB ontology. In designing our application architecture, we have followed the common system engineering practice of first determining the scope and nature of the data involved. A first essential step is to understand the semantics of the domain and to capture that knowledge in an agreed-upon format. Arranging the domain concepts in a taxonomy is one of the initial organizing steps in the ontology design process. The class hierarchy represents the generalization and specialization relationships between the various classes of objects in a domain [6]. Briefly, classes describe concepts in the domain. Subclasses represent concepts (classes) that are more specific than the superclass and these subclasses can have their own unique properties. Slots represent properties of the classes. For example, in Figure 1, we see that there is a class named Reference and three more specific subclasses of Reference: Journal Article, Patent Application, and Direct Submission. Figure 1 also shows that the class Epitope has a number of properties (slots) associated with it such as "has Epitope Structure" and "has Epitope Source".
Figure 1 Overview of IEDB Class Hierarchy.
The IEDB Ontology: Reference, Epitope Structure, Epitope Source, and Assay Information classes
Our approach for creating the class hierarchy was a top-down development process where we defined each class in a domain and then identified its properties before building the hierarchy. The main classes identified for IEDB are Reference, Epitope Structure, Epitope Source, MHC Binding, Naturally Processed Ligand, T Cell Response, and B Cell Response (Figure 1). The Epitope class is the main class that encompasses all the individual concepts that were identified. The individual concepts are related to other classes. The primary relationships use the sub-class relationship or use a property (shown in the figures by the arcs labeled "has") that has a restriction on the type of the value that may fill that slot.
"Reference" is the class encompassing information related to the data source from which an epitope and its related information are extracted into the IEDB. We have identified three broad subclasses of References that describe where epitope information will be obtained. They are Journal Article, Patent Application, and Direct Submission. The complete listing of slots (fields) encompassed by the Journal Article, Patent Application, and Submission classes are provided in Figure 2. The Journal Article class refers to manuscripts published in peer-reviewed journals. The Patent Application class captures all the reference fields for a patent application that contain epitope information. The Submission class captures information about sources that contribute data to the IEDB directly. Data deposited by the Large Scale Antibody and T Cell Epitope Discovery contracts [3] and those transferred from other websites fall into this class.
Figure 2 Detailed classification of Reference class showing its subclasses and slots.
The Epitope Structure and Epitope Source classes capture intrinsic features of an epitope. The Epitope Structure class captures the physical and chemical features of an epitope. Virtually any molecular structure may provoke an immune response, such as proteins, carbohydrates, DNA, and lipids. In the Epitope Structure class, structural information relating to linear sequences and 2-D structures of the epitope, if available, are catalogued. The Epitope Source class captures the phylogenetic source of an epitope, including species of origin, gene name, protein name, and links to other databases for more detailed information about proteins and genes. Figures 3A and 3B show the listings of properties (slots/fields) encompassing the Epitope Structure and Epitope Source classes.
Figure 3 Detailed listing of properties of Epitope Structure (A), Epitope Source (B), and Assay Information (C) class.
The experimental data and information about specific experiments and the methodology utilized are captured in the Assay Information class. The name of the assay used, the type of response measured in the assay, and the readout of the assay are examples of information captured in the Assay Information class. This important class is used as a superclass of several other classes (and thus its properties are inherited by those classes). A complete listing the properties (slots/fields) in the Assay Information class is shown in Figure 3C.
Immunization, Antigen, and Antigen Presenting Cell classes
As with Assay Information, the classes Immunization, Antigen, and Antigen Presenting Cell are used in multiple other class descriptions. Features relating to the induction of the immune response are captured in the Immunization class (Figure 4A). It has relationships to other classes like Immunized Species, Immunogen, In vivo Immunization, and In vitro Immunization. Immunized Species contains information relating to the host that is being immunized. The Immunogen class describes the molecules that induce the immune response and an associated carrier molecule, if present. Features relating to how the immunogen was introduced to the immunized species are captured under the In vivo and In vitro Immunization classes.
Figure 4 High-level classification of Immunization (A), Antigen (B), and Antigen Presenting Cells (C) class.
Similarly, antigens are defined as the whole molecules that react with the products of an immune response (as opposed to the epitopes which are the specific structures, contained within the antigen that engages the immune receptor). Information relating to the antigen and any associated carrier molecule is captured in the Antigen class (Figure 4B). During immune responses, antigen-presenting cells process antigens and present peptide epitopes complexed with MHC molecules. This information is captured in the Antigen Presenting Cells class, which has a relationship to the MHC Molecules and the Source Species classes (Figure 4C). The Source Species class describes the species information from which the antigen presenting cells are derived.
The MHC Binding, Naturally Processed Ligand, T Cell Response, and B Cell Response classes
The MHC Binding class captures the details relating to the interaction of the epitope with specific MHC molecules and information relating to the MHC molecule along with any available Epitope-MHC complex structure details. This class also has a slot that is restricted to be an instance of the Assay Information class (Figure 5A).
Figure 5 High-level classification of MHC Binding (A) and Naturally Processed Ligand (B) class.
Extrinsic features of an epitope are captured by Naturally Processed Ligand, T Cell Response, and B Cell Response classes. Extrinsic features are context-dependent attributes, being dependent upon specific experimental conditions. The Naturally Processed Ligand class captures data related to epitopes that are naturally processed and presented on the cell surface. This class has properties that are instances of classes including Antigen Presenting Cell, Antigen, and Assay Information (Figure 5B).
The Naturally Processed Ligand class differs from the MHC Binding class in that information related to the antigen that was processed and the cell types in which the processing occurred is represented. MHC Binding class captures data relating to in vitro MHC binding assays, which assess the epitope's binding capacity to the MHC molecule. Hence the MHC Binding class does requires neither the Antigen class not the Antigen Presenting Cells class. In general, naturally processed ligands are assessed in the absence of a T cell response, for example, identified by direct elution from MHC molecules extracted from infected cells or antigen presenting cells. Thus, the Immunization class is not used as a value restriction by the Naturally Processed Ligand class.
The T Cell Response class captures all of the T cell mediated immunity-related information (Figure 6A). It has properties that are of type: Immunization, Effector Cells, Antigen Presenting Cell, Antigen, Assay Information, and Epitope-MHC-TCR Complex. The Effector Cell class describes the cells that are elicited upon immunization and that acquire measurable functions as a result. The B Cell Response class describes antibody responses that are related to the epitope (Figure 6B). This class has properties that are of type: Immunization, Antibody Molecule, Antigen, Assay Information, and Antigen-Antibody Complex. Because B cell responses do not require MHC binding and antigen presenting cells, the respective classes related to MHC Molecule and Antigen Presenting Cells are not used as restrictions on properties of the B Cell Response class.
Figure 6 High-level classification of T Cell Response (A) and B Cell Response (B) class.
Classes capturing 3D structures
There are three classes that capture information about the 3D structure of complexes: Epitope-MHC Complex, Epitope-MHC-TCR Complex, and Antigen-Antibody Complex. The Epitope-MHC Complex, Epitope-MHC-TCR Complex, and Antigen-Antibody Complex classes are used as restrictions on properties of the MHC Binding, T Cell Response, and B Cell Response classes respectively (Figures 5A, 6A, and 6B). These Complex classes capture the Protein Data Bank (PDB) Identifier, which provides detailed information about 3D structures. The Protein Data Bank [22,23] contains approximately 1600 3D structures that are of immunological interest. Other information that is not available in PDB, such as the atom pairs that are involved in the interactions between molecules, the specific residues, the contact area of the molecules, and allosteric effect, is also captured here.
IEDB Class Hierarchy and Data Dictionary
Each class has numerous slots that capture detailed information associated with epitopes. As mentioned above, a complete list of all the classes, their properties, and relationship, can be found at . One of the files provided as supplementary material contains two examples of how two literature references [24,25] containing epitope information are extracted into the IEDB ontology (additional file 1). Along with the class hierarchy, the IEDB's data dictionary (additional file 2) provides more detailed information about the fields that are defined for the IEDB. The data dictionary contains a textual overview description and a listing of fields that are required to be completed for IEDB entries. The data dictionary also allows database users to provide comments and suggestions to IEDB team to enhance the formal ontology.
Utility and Discussion
The IEDB will be a comprehensive resource pertaining to epitopes of the immune system. By extensively curating both intrinsic and extrinsic features associated with epitopes, the IEDB is expected to provide substantially greater detail about specific epitopes than any other databases presently available. The IEDB will be populated with data derived from three main sources, namely the peer-reviewed literature, patent applications, and direct submission. Epitope data published in the literature and patent applications are curated manually by the IEDB's curation team. Data from already existing epitope databases, whose authors have agreed to share their data, will also be imported into the IEDB. Apart from these, a main data source will be the direct submission of data from the Large-scale Antibody and T Cell Epitope Discovery programs [3] that are funded by NIAID. Presently, fourteen contracts have been awarded under this program, and all of them will submit their data to the IEDB. Direct antibody and T cell epitope submissions will also be sought from the broader research community, with an emphasis on antibody epitopes to NIAID Category A-C pathogens. Because of the large scale of the IEDB project, a formal ontology is critical to ensure consistency in the representation of data.
Communication between database developers, researchers, analysis tool developers, and team members is crucial, and can be performed in harmony only when a common vocabulary is established. An ontology, which is an explicit formal specification of the terms in the domain and relationships among them, is an effective way to share the knowledge contained in that domain. Accordingly, since the IEDB's domain is epitope-related data, we have created an ontology that captures detailed conceptual structure related to these data.
The development of this ontology has relevance for the expansion and modification of the epitope knowledge base. Our ontology design defines individual concepts as separate classes and then defined relationships between these classes and other objects in the domain. These classes serve to restrict the values that will describe properties of objects in the database. For example, the species is a separate concept defined in its own class. Depending on the context, this can refer to an immunized species or the species from which antigen presenting cells are derived. Similarly MHC Molecules is defined as a separate class, and it is used as a value restriction by concepts like MHC Binding and Antigen Presenting Cells. Defining concepts as separate classes and using them to restrict the values of properties in other classes facilitates the expansion and modification of our ontology. Adding properties (slots) to concepts is a task easily accomplished when there are well-defined class descriptions that may serve as value restrictions on the properties, and providing that these class descriptions are general enough to apply in all instances. We have ensured in our design that each concept is atomic and that it can be re-used by various classes.
The development of a formal ontology is valuable to database users and in particular to scientists contributing data to, and downloading data from, the IEDB. We anticipate that the availability of a formal ontology will ensure that a common language and shared understanding of concepts will inspire this type of communication, thus ensuring maximum efficiency and accuracy. The formal ontology developed will most likely require refinement and fine tuning when users provide suggestions and new technologies for performing experiments are discovered. The IEDB website will provide mechanisms for the users to provide suggestions and participate in the enhancement of the ontology. The IEDB Data Dictionary has a separate column for the users to provide comments on specific data fields. The IEDB website will also host web forms that will guide users to conform to the ontology definitions when submitting data. Apart from the web forms, an XML schema definition (XSD) will be available on the website for users to inspect and use in their data submission. Users will also be able to download epitope records from the website.
In the process of developing new ontologies, it is good practice to leverage existing community standards. In our initial analysis, we confirmed that there were no explicit ontologies that efficiently captured epitope details as per the scope of the IEDB program. As mentioned above, we have utilized, as much as possible, inferred ontologies from existing epitope databases. Among the ontologies that we analyzed, IMGT-Ontology and Gene Ontology were the only two formal ontologies that were related to the epitope domain. The IMGT-Ontology was designed for the ImMunoGeneTics database. IMGT is an integrated database specializing in antigen receptors (immunoglobulins and T-cell receptors) and the major histocompatibility complex of all vertebrate species. The ontology developed for this database has specific immunological content, describing the classification and specification of terms needed for immunogenetics. The IEDB does conform to IMGT's standards about receptors and MHC molecule chains in the sense that all the chain names follow IMGT's controlled vocabulary.
GO provides structured controlled vocabularies for genes, gene products, and sequences annotated for many organisms. The IEDB complements GO in terms of epitopes of immunological interest since GO is incomplete in this area. Antigens, which are primary sources of epitopes, are annotated in GO. Thus, in essence, the IEDB could be utilized to provide an extension of GO for antigens that contain epitope-related information.
Conclusion
Perhaps the most important element in the development of the IEDB ontology is that, to the best of our knowledge, this represents the first immunological ontology specifically designed to capture both intrinsic biochemical and extrinsic context dependent information. In this respect, it is similar in spirit, but different in approach, from other knowledge resources relating to systems biology. We anticipate that the development of this type of ontology and associated databases might lead to completely new methods for describing and modeling immune responses. Accordingly, this new program might represent a novel tool to assist in the design, testing, and development of new ways to combat infectious diseases and other immune related pathologies such as cancer and autoimmune diseases.
Availability
A complete listing of IEDB's class hierarchy and its properties is available at
List of Abbreviations Used
IEDB – Immune Epitope Database and Analysis Resource
MHC – Major Histocompatibility Complex
TCR – T Cell Receptor
BCR – B Cell Receptor
IMGT – Immunogenetics Database
GO – Gene Ontology
J2EE – Java 2 Enterprise Edition
OWL – Ontology Web Language
PDB – Protein Data Bank
NIAID – National Institute of Allergy and Infectious Diseases
XML – Extensible Markup Language
XSD – XML Schema Definition
Authors' contributions
MS generated the formal ontology using Protégé. MS, BP, HB, JS, and AS designed the initial ontology. SW, JM, WF, DM, PB provided critical insight in enhancing the initial design and creating the formal ontology. All authors participated in the preparation of the manuscript.
Supplementary Material
Additional File 1
Sample Curation. This file contains a table that shows how epitope and its related information were extracted from a literature reference and mapped into the IEDB ontology.
Click here for file
Additional File 2
IEDB Data Dictionary v12-5. The Data Dictionary contains the textual overview description and a listing of fields that are captured for IEDB.
Click here for file
Acknowledgements
This work was supported by the National Institutes of Health Contract HHSN26620040006C. The authors like to thank Bette Korber, Marie-Paule LeFranc, William Hildebrand, Vladimir Brusic, Anne De Groot, Darren Flower, Pam Surko, Scott Stewart, and Scott Way for the helpful discussions. The authors also thank Alison Deckhut for critical review of the manuscript.
==== Refs
Peters B Sidney J Bourne P Bui HH Buus S Doh G Fleri W Kronenberg M Kubo R Lund O Nemazee D Ponomarenko JV Sathiamurthy M Schoenberger SP Stewart S Surko P Way S Wilson S Sette A The design and implementation of the immune epitope database and analysis resource Immunogenetics 2005
Peters B Sidney J Bourne P Bui HH Buus S Doh G Fleri W Kronenberg M Kubo R Lund O Nemazee D Ponomarenko JV Sathiamurthy M Schoenberger S Stewart S Surko P Way S Wilson S Sette A The immune epitope database and analysis resource: from vision to blueprint PLoS Biol 2005 3 e91 15760272 10.1371/journal.pbio.0030091
Sette A Fleri W Peters B Sathiamurthy M Bui HH Wilson S A roadmap for the immunomics of category A-C pathogens Immunity 2005 22 155 161 15773067 10.1016/j.immuni.2005.01.009
Guarino N Giaretta P Mars NJI Ontologies and Knowledge Bases: Towards a Terminological Clarification Towards Very Large Knowledge Bases: Knowledge Building and Knowledge Sharing 1995 Amsterdam, IOS Press 25 32
Gruber T A Translational Approach to Portable Ontologies Knowledge Acquisition 1993 5 199 220 10.1006/knac.1993.1008
Noy NF McGuinness DL Ontology Development 101: A Guide to Creating Your First Ontology Stanford Knowledge Systems Laboratory technical report KSL-01-05 and Stanford Medical Informatics technical report SMI-2001-0880 2001
Brusic V Rudy G Harrison LC MHCPEP, a database of MHC-binding peptides: update 1997 Nucleic Acids Res 1998 26 368 371 9399876 10.1093/nar/26.1.368
Rammensee H Bachmann J Emmerich NP Bachor OA Stevanovic S SYFPEITHI: database for MHC ligands and peptide motifs Immunogenetics 1999 50 213 219 10602881 10.1007/s002510050595
Schonbach C Koh JL Flower DR Wong L Brusic V FIMM, a database of functional molecular immunology: update 2002 Nucleic Acids Res 2002 30 226 229 11752300 10.1093/nar/30.1.226
Sathiamurthy M Hickman HD Cavett JW Zahoor A Prilliman K Metcalf S Fernandez Vina M Hildebrand WH Population of the HLA ligand database Tissue Antigens 2003 61 12 19 12622773 10.1034/j.1399-0039.2003.610102.x
Korber BTM Brander C Haynes BF Koup R Moore JP Walker BD Watkins DI HIV Immunology and HIV/SIV Vaccine Databases 2003 , Los Alamos National Laboratory, Theoretical Biology and Biophysics.Los Alamos, New Mexico. LA-UR 04-8162
Blythe MJ Doytchinova IA Flower DR JenPep: a database of quantitative functional peptide data for immunology Bioinformatics 2002 18 434 439 11934742 10.1093/bioinformatics/18.3.434
AntiJen Database
Bhasin M Singh H Raghava GP MHCBN: a comprehensive database of MHC binding and non-binding peptides Bioinformatics 2003 19 665 666 12651731 10.1093/bioinformatics/btg055
Giudicelli V Lefranc MP Ontology for immunogenetics: the IMGT-ONTOLOGY Bioinformatics 1999 15 1047 1054 10745995 10.1093/bioinformatics/15.12.1047
Harris MA Clark J Ireland A Lomax J Ashburner M Foulger R Eilbeck K Lewis S Marshall B Mungall C Richter J Rubin GM Blake JA Bult C Dolan M Drabkin H Eppig JT Hill DP Ni L Ringwald M Balakrishnan R Cherry JM Christie KR Costanzo MC Dwight SS Engel S Fisk DG Hirschman JE Hong EL Nash RS Sethuraman A Theesfeld CL Botstein D Dolinski K Feierbach B Berardini T Mundodi S Rhee SY Apweiler R Barrell D Camon E Dimmer E Lee V Chisholm R Gaudet P Kibbe W Kishore R Schwarz EM Sternberg P Gwinn M Hannick L Wortman J Berriman M Wood V de la Cruz N Tonellato P Jaiswal P Seigfried T White R The Gene Ontology (GO) database and informatics resource Nucleic Acids Res 2004 32 D258 61 14681407 10.1093/nar/gkh066
International Classification of Diseases (ICD)
Benson DA Karsch-Mizrachi I Lipman DJ Ostell J Rapp BA Wheeler DL GenBank Nucleic Acids Res 2000 28 15 18 10592170 10.1093/nar/28.1.15
Wheeler DL Chappey C Lash AE Leipe DD Madden TL Schuler GD Tatusova TA Rapp BA Database resources of the National Center for Biotechnology Information Nucleic Acids Res 2000 28 10 14 10592169 10.1093/nar/28.1.10
Marsh SG Albert ED Bodmer WF Bontrop RE Dupont B Erlich HA Geraghty DE Hansen JA Hurley CK Mach B Mayr WR Parham P Petersdorf EW Sasazuki T Schreuder GM Strominger JL Svejgaard A Terasaki PI Trowsdale J Nomenclature for Factors of the HLA System, 2004 Hum Immunol 2005 66 571 636 15935895 10.1016/j.humimm.2005.02.002
Duineveld AJ Stoter R Weiden MR Kenepa B Benjamins VR Wondertools? A comparative study of ontological engineering tools International Journal of Human-Computer Studies 2000 52 1111 -1133 10.1006/ijhc.1999.0366
Berman HM Westbrook J Feng Z Gilliland G Bhat TN Weissig H Shindyalov IN Bourne PE The Protein Data Bank Nucleic Acids Res 2000 28 235 242 10592235 10.1093/nar/28.1.235
Berman HM Bhat TN Bourne PE Feng Z Gilliland G Weissig H Westbrook J The Protein Data Bank and the challenge of structural genomics Nat Struct Biol 2000 7 Suppl 957 959 11103999 10.1038/80734
Terajima M Cruz J Raines G Kilpatrick ED Kennedy JS Rothman AL Ennis FA Quantitation of CD8+ T cell responses to newly identified HLA-A*0201-restricted T cell epitopes conserved among vaccinia and variola (smallpox) viruses J Exp Med 2003 197 927 932 12668642 10.1084/jem.20022222
Lin Y Shen X Yang RF Li YX Ji YY He YY Shi MD Lu W Shi TL Wang J Wang HX Jiang HL Shen JH Xie YH Wang Y Pei G Shen BF Wu JR Sun B Identification of an epitope of SARS-coronavirus nucleocapsid protein Cell Res 2003 13 141 145 12862314
|
16305755
|
PMC1287064
|
CC BY
|
2021-01-04 16:37:20
|
no
|
Immunome Res. 2005 Sep 20; 1:2
|
utf-8
|
Immunome Res
| 2,005 |
10.1186/1745-7580-1-2
|
oa_comm
|
==== Front
Immunome ResImmunome Research1745-7580BioMed Central London 1745-7580-1-51630575610.1186/1745-7580-1-5ResearchModelling vaccination schedules for a cancer immunoprevention vaccine Motta Santo [email protected] Filippo [email protected] Pierluigi [email protected] Francesco [email protected] Department of Mathematics and Computer Science, University of Catania, Catania, Italy2 Sezione di Cancerologia, Dipartimento di Patologia Sperimentale, University of Bologna, and Centro Interdipartimentale di Ricerche sul Cancro "Giorgio Prodi", Italy3 Istituto per le Applicazioni del Calcolo, Consiglio Nazionale delle Ricerche, Roma, Italy4 Faculty of Pharmacy, University of Catania, Italy2005 7 10 2005 1 5 5 9 6 2005 7 10 2005 Copyright © 2005 Motta et al; licensee BioMed Central Ltd.2005Motta 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.
We present a systematic approach to search for an effective vaccination schedule using mathematical computerized models. Our study is based on our previous model that simulates the cancer vs immune system competition activated by tumor vaccine. This model accurately reproduces in-vivo experiments results on HER-2/neu mice treated with the immuno-prevention cancer vaccine (Triplex) for mammary carcinoma. In vivo experiments have shown the effectiveness of Triplex vaccine in protection of mice from mammary carcinoma. The full protection was conferred using chronic (prophylactic) vaccination protocol while therapeutic vaccination was less effcient.
In the present paper we use the computer simulations to systematically search for a vaccination schedule which prevents solid tumor formation. The strategy we used for defining a successful vaccination schedule is to control the number of cancer cells with vaccination cycles. We found that, applying the vaccination scheme used in in-vivo experiments, the number of vaccine injections can be reduced roughly by 30%.
==== Body
1 Introduction
Tumor immunologists have long sought ways to turn experimental results into effective therapies for human cancer patients. The best results so far are provided by using various monoclonal antibodies directed against tumor cells, which won approval from regulatory agencies and entered clinical practice. Other approaches, such as therapeutic vaccines that aim at stimulating the immune response of the host against tumor cells, were much less successful [1]. Experimental evidence clearly shows that vaccines elicit effective responses against early, microscopic tumors, but are ineffective against established, large tumor masses. A similar situation is found in infectious immunity: prophylactic vaccines protect millions of individuals worldwide from pathogens, whereas therapeutic vaccines are mostly ineffective. Such results led some tumor immunologists to the idea that the effort should be directed towards the development of prophylactic, rather than therapeutic, cancer vaccines [2]. Prophylactic vaccines against viruses that increase the risk of cancer, such as hepatitis B virus (HBV) or papilloma virus (HPV), have already shown a significant effcacy in reducing cancer incidence [3,4]. The challenge is now to devise immunological strategies for the design of prophylactic vaccines for prevention of those human cancers that are not related to viral infections.
Standard vaccines against viruses induce the primary response of the adaptive components of the immune system against the non-self antigen in order to activate the immune system memory which will then elicit the much stronger secondary response when the antigen will enter the body. There is no need to break self tolerance as the antigen is non self.
At variance tumor vaccines need to break tolerance of the tumor associated antigens (TAA) that otherwise would be treated as self and not attacked by the immune system.
The effectiveness of tumor immuno-prevention was demonstrated in the past ten years in several mouse models of tumor development. Tumors in these models are induced either by chemical carcinogens or by transgenic expression of oncogenes. Among the most thoroughly investigated is the HER-2/neu transgenic mouse model. HER-2/neu is an oncogene involved in human breast and ovary carcinomas. The protein product of HER-2/neu, p185neu, is a membrane tyrosine kinase that transduces proliferative signals. A deregulation of HER-2/neu, for example due to gene amplification, leads to uncontrolled cell proliferation. Over-expression of a rat HER-2/neu transgene in mice was obtained using tissue-specific regulatory sequences derived from mammary tumor virus (MMTV) long terminal repeats (LTR). Various transgenic mouse lines were obtained using either a mutant, constitutively activated HER-2/neu gene or a normal, non mutated gene. Here we will refer to mice carrying the mutant oncogene, which are prone to a very aggressive mammary carcinogenesis, invariably leading to the development of invasive mammary carcinomas by the age of 6 months. The natural history of mammary carcinoma in HER-2/neu transgenic mice was thoroughly investigated and found to be remarkably similar to that of human breast cancer [5]. Immuno-prevention of mammary carcinoma in HER-2/neu transgenic mice was attempted using various immunological strategies, including cytokines, non-specific stimulators of the immune response, and HER-2/neu specific vaccines made of DNA, proteins, peptides, or whole cells. Most approaches achieved a delay of mammary carcinogenesis, but a complete prevention of tumor onset was not attained, particularly in the most aggressive tumor models [6]. We achieved the first complete success at preventing mammary carcinoma in HER-2/neu transgenic mice using a vaccine that combined three different stimuli for the immune system. The first was p185neu, protein product of HER-2/neu, which in this system is at same time the oncogene driving carcinogenesis and the target antigen. p185neu was combined with the two non specific adjuvants, allogeneic class I major histocompatibility complex (MHCI) glycoproteins and interleukin 12 (IL-12). MHCI glycoproteins are responsible for some of the most intense immune responses observed during the rejection of allogeneic organ transplants. Unlike conventional antigens, allogeneic MHCI molecules stimulate a relatively large fraction of all T cell clones, up to 10% of the available repertoire. IL-12 is a cytokine normally produced by antigen presenting cells (APC) such as dendritic cells (DC) to stimulate T helper cells and other cells of the immune system, such as natural killer cells (NK) [7]. IL-12 was initially administered systemically, but a more recent formulation of the Triplex vaccine used genetically modified vaccine cells transduced with IL-12 genes, thus allowing cytokine production at locally high levels that more closely mimicked the natural release of IL-12 [8]. The vaccine cell (VC) that we will model here is similar to the latter one and consists of a HER-2/neu transgenic mammary carcinoma cell allogeneic with respect to the host and transduced with IL-12 genes.
A complete prevention of mammary carcinogenesis with the Triplex vaccine was obtained when vaccination cycles started at 6 weeks of age and continued for the entire duration of the experiment, at least one year (chronic vaccination). One vaccination cycle consisted of four intraperitoneal administrations of non-replicating (mitomycin-treated) VC over two weeks followed by two weeks of rest [8]. We made various attempts at reducing the number of vaccination cycles, in particular we studied the effects of just three cycles starting at 6 weeks of age (early vaccination), or at 10 weeks of age (late vaccination), or at 16 weeks of age (very late vaccination). Early vaccination produced a significant delay in the onset of tumors, but all mice eventually succumbed to mammary carcinoma. Late vaccination was less effective than early vaccination, whereas the very late protocol was completely devoid of effect in comparison to untreated control mice (Figure 1).
Figure 1 Tumor-free survival curves of HER-2/neu transgenic mice receiving the Triplex vaccine according to different protocols from [6]. Each arrow at the bottom of the graph represents one cycle of vaccination. The sequence of neoplastic progression in untreated mice is outlined under the x axis; CIS, carcinoma in situ.
The question whether the Chronic protocol is the minimal vaccination protocol yielding complete protection from tumor onset, or whether a lower number of vaccination cycles would provide a similar degree of protection is still an open question. Finding an answer to this question via a biological solution would be too expensive in time and money as it would require an enormous number of experiments, each lasting at least one year. For this reason we developed an accurate model of immune system responses to vaccination and we use this model as a virtual laboratory to search for effective vaccination protocols. The paper is organized as follows: section §2 will introduce the general requirements needed to model the problem; section §3 describes in details our virtual laboratory, i.e. the model, its structure and biological details, and its computer implementation (the simulator). In section §4 we present in silico experiments for different vaccination schedules and we show how the virtual laboratory can be used for designing a new and better protocol. Finally in section §5 we discuss our results, draw some conclusions, and plans for future investigations.
2 The general framework for modelling
In dealing with modelling of cancer – immune system interaction and competition one should be aware that this competition can play a crucial role besides therapeutical actions. This competition may possibly end up either with the elimination of the cancer cells, or the progressive invasion by cancer cells of other tissues or organs.
The goal of medical treatments is to enhance the immune response by activating the immune defense and/or specializing the ability of immune cells to identify the presence of the tumor.
The immune competition is a phenomenon which involves aggressive cells or particles (either external non-self pathogens or self modified or corrupted cells) and the various populations of the immune system. The immune system appears to be a distributed system which lacks central control, but which, nevertheless, performs its complex task in an extremely effective and effcient way. Complexity, in this framework, is driven by the fact that interactions are developed at different scales (i.e. the cellular dynamics is ruled by sub-cellular interactions) and different mechanisms operate on the same subject (mechanical for the dynamics and biological for the immune competition). The state of the art of the immune mechanisms from the view point of molecular biology are described in specialized literature [9-11]. Owing to the rapid progress of biological knowledge this is rapidly changing.
A model, which is a mapping from a real-world domain to a mathematical domain, highlights some of the essential properties while ignoring not relevant (or believed not relevant) ones. A good model must be relevant, capturing the essential properties of the phenomenon, computable, driving computational knowledge into mathematical representation; understandable, offering a conceptual framework for thinking about the scientific domain; and extensible, allowing the discovery of additional real properties in the same mathematical framework.
In the framework of immune system competition, relevant means that a model should be able to capture the essential properties of the system, the system entities organization and their dynamic behavior. A computable model should then be able to simulate both the dynamic behavior and the evolution and interactions of the system entities which play the game. An understandable model must reproduce concept and ideas of tumor immunology while opening new computational possibilities for understanding the immune competition. Finally extensible models should allow the inclusion of new knowledge with a limited effort. The immune system is characterized by a great complexity so that it is very diffcult, or even impossible, to develop a detailed mathematical description of all phenomena related to the immune competition which satisfies all the above properties. A significant effort has recently been devoted in searching for an appropriate mathematical approach to describe the Immune System – tumor competition (see [12] for a recent review). However, if one focuses the attention on specific types of interactions, one may attempt to develop ad hoc models for a specific phenomenon at the chosen observation and representation scale. Extensive description of subcellular vs larger scales of modelling can be found in [13-15]. Methods based on generalized kinetic model look very promising for system description: however they are not yet so detailed to model a specific therapeutic vaccine.
3 SimTriplex Model
To model the action of a specific tumor vaccine, we used a computational approach which reproduces the ab initio kinetic model that describes the interactions and diffusion of each relevant biological entity. This approach is biologically very flexible; the behavior of entities is modelled using present biological knowledge and can be easily modified to reflect observations from new biological experimental results. Compared to the complexity of the real biological system our model is still very naive and it can be extended in many aspects. However the model is complete enough to describe the major aspect of the phenomenon and, after tuning the model parameters, it can predict the response to a vaccination schedule that prevented the formation of solid tumors in mice.
Our model, hereafter referred as SimTriplex model, describes the immune competition using an agent based method. These methods are nowadays very popular as they find application in various fields. However the idea of discrete agents whose global dynamics is able to reproduce macroscopic behavior has been introduced more than fifty years ago in the framework of fluid-dynamics.
We used a Lattice Gas Automata (LGA) approach which allows to describe, in a defined space, the immune system entities with their different biological states and the interactions between different entities. The immune system evolution in a 2D physical space and in time is generated from the interactions and diffusion of the different entities. Extension to a 3D physical space is possible, but it would obviously have an higher computational cost. The major advantage of this technique is that the entities and the relationships can be described in terms that are much similar to the biological world. The intrinsic non linearity of the system is treated with no additional efforts. Models based on this class of techniques reproduce the biological knowledge of the system; so they are relevant, understandable and extensible. They are naturally computable but the computational effort increases drastically with increasing biological details (e.g. the immune repertoire and biological details). This class of model can be seen as the computational counterpart of the generalized kinetic model.
To describe the cancer – immune system competition one needs to include all the entities (cells, molecules, adjuvants, etc.) which biologists recognize as relevant in the competition. These entities have internal states, birth, age and death (i.e. ages structure). Interactions between different entities will stochastically change the internal state of one or both the interacting entities. Space changes in the system are achieved with diffusion instead of collision (see 3.2.4). It is worth to remind that in the microscopic cellular framework one is interested only in the initial and final state. The model of the mechanisms which produce the change of state is deferred to sub-cellular models.
Our model is driven by the experimental data on Triplex, an engineered vaccine for mammary carcinoma tested on HER-2/neu transgenic mice.
The model, which has been fully described in [16] include the entities described in table 1. Here we want to recall entities representation in order to highlight the computational approach.
Table 1 List of symbols and half life. Half life is expressed in times teps; one times tep represents 8 hours.
entity symbol half life
B Cells B 3.33 days
Antibody Secreting Plasma Cells PLB 10
T-helper lymphocytes TH 10
T-cytotoxic lymphocytes TC 10
Macrophages MP 3.33 days
Dendritic Cells DC 10
Interleukin-2 IL-2 5
Immunoglobulins IgG 5
Danger Signal D 5
Major Histocompatibility Complex Class I MHCI -
Major Histocompatibility Complex Class II MHCII -
Tumor Associated Antigens Ag 3
Immunocomplexes IC 100
Cancer Cells CC 1095
Vaccine Cells VC 5
Natural Killer Cells NK 10
Interleukin-12 IL-12 5
3.1 Entities Representation
The primary function of the immune system is a recognition function. Working at cellular scale we need to represent only those characteristics of entities which are relevant for recognition and interaction: state, recognition site and lifetime.
From the computational point of view, the major difference between cellular and molecular entities is that cells have a state attribute and may be classified on the basis of these attributes. Entities with no biological state, like antibodies are not followed individually, instead we consider only their total number for each lattice point. Their age structure is obtained increasing or decreasing the population number. Antigens (Ag or TAA) are however followed as individuals for their specific role.
Entities with biological states, i.e. cellular entities, are tracked individually in order to keep track of state change owing to interactions. The state of a cell is an artificial label introduced by the logical representation of the cells behavior. They have a standard lifetime which can eventually increase with interactions.
Most cellular and molecular entities have recognition sites (receptors). The set of lymphocyte's receptors is represented by bit-strings of length h [17] which then forms the so called "shape space" [18]. A clonal set of cells is characterized by the same clonotypic receptor, i.e. by the same bit-string of length l. The potential repertoire of receptors scales as 2l.
Simple entities, like antibodies, are represented using their binding site (receptors). Tumor associated antigens are also described using their binding site but we keep also track of their lifetime. Representation of cellular entities include receptors, binding site and lifetime.
The receptor-coreceptor binding among the entities are described in terms of matching between binary strings with fixed directional reading frame. Bit-strings represent the generic "binding site" between cells (through their receptors) and target molecules (through peptides and epitopes). Every cellular entity is represented by a number of molecules, including the receptor. The repertoire is then defined as the cardinality of the set of possible instances of entities that differ in, at least, one bit of the whole set of binary strings used to represent its attributes.
Indeed, the cells equipped with binding sites and the antibodies, have a potential repertoire of , where Ne indicates the number of binary strings used to represent receptors, MHC-peptide complexes, and epitopes of the entity e. Other entities do not need to be specified by binary strings so their repertoire is represented by a single entity (i.e. Ne = 0). Examples include cytokine molecules such as interferon-γ and the danger signal [19].
Figures 2 and 3 show a graphical representation of lymphocytes and antigens. The bit-string is labelled with its decimal representation. The figures are taken from ImmSim v.3 userguide 1.
Figure 2 Grafical representation of lymphocytes and antigen presenting cells.
Figure 3 Graphical representation of antigens.
As receptors' entities are represented by bit strings, the only information available is a "similarity" between bit strings. A standard measure of similarity between two bit strings is the so called Hamming distance which is just the number of mismatching bits 2. The interactions between two entities equipped with receptors are defined by a probability measure, called affnity potential, which is a function of the Hamming distance between the binary strings representing the two entities' binding site. For two strings s and s' such probability is max when all corresponding bits are complementary (0 ↔ 1), that is, when the Hamming distance between s and s' is equal to the bit string length. If l is the bit string length and m is the Hamming distance between the two strings, the affnity potential is defined in the range 0, . . ., l as
where vc ∊ (0,1) is a free parameter which determines the slope of the function where as mc (l/2 <mc ≤ l) is a cut-off (or threshold) value below which no binding is allowed.
The cells are free to diffuse across the lattice sites. At each time-step, representing 8 hours of real time, the cells and molecules residing on the same lattice site they can interact each other. The tissue is represented as two-dimensional triangular lattice (six neighbor sites) L × L, with periodic boundary conditions in both directions (up-down, left-right). The lattice is taken to represent here a portion of mammary tissue of the mouse.
All various classes of immune functional activity, phagocytosis, immune activation, opsonization, infection and cytotoxicity are described by computational rules.
The model also includes the mechanism of haematopoiesis as described in [16]. The rules which implement these actions are executed in a randomized order.
3.2 Coding SimTriplex
The high-level architecture of the code can be explained using the following pseudo-code.
SimTriplex Simulator
Input: Accept pre-determined inputs (e.g., user-specified vaccine injections, random generation of new B- and T-cells, etc.).
Time cycle start
Activate Thymus and Bone Marrow functions.
Haematopoiesis function is activated and operates for all cycles.
Interaction-driven dynamics. Enables the pairing of entities and modification of their states, creates new entities in response to these interactions, etc. in response to these interactions.
Internal dynamics. Makes allowance for internal, non interaction-driven dynamics (e.g. aging and natural death, differentiation, mutation, etc.).
Diffusion. Allows cellular entities to diffuse on the lattice checking the physical space constraints. Change the density of molecular entities in order to mimic diffusion.
Output. Stores a trace of the state of the system at time t.
Time cycle ends
code end
A set-up routine will create the lattice, register the values of the various input parameters and stochastically fill the lattice with all population of cells. Time step interactions will proceeds up to a preselected final time.
Cellular and molecular entities are treated differently in the model. Cellular entities are treated individually while molecular entities are modelled as populations. We describe them separately.
3.2.1 Cellular entities
The common features of cellular entities are : Position, Specificity, States and Age.
Position is the crucial parameter which defines which entities can interact as interactions between entities can occur only between those entities which are co-located in the same lattice cell;
In the shape space [17] entities interact according to their specificity represented by binary string. This simulates recognition of the entities by their paratopes (complementary shapes of peptides and receptors). Each entity, except for plasma cells, has at least one receptor (or paratope/epitope, depending on whether they are cellular or molecular entities) which is its primary identifier, i.e., its specificity.
The key constituents of specificity are MHC class I (MHC-I) and MHC class II (MHC-II), T cell receptors (TCR) and B-cell receptors (BCR). The different class of cellular entities we consider are: B-cells, T helper cells (TH), cytotoxic T cells (TC), antigen presenting cells (APC) and plasma cells (PLB). B cells are endowed with MHC-II and BCR; TC cells are equipped only with TCR (CD8) and TH include TCR (CD4).
With the name APC cells one refers to Macrophages (MP), Dendritic cells (DC) and B-cells. MP are aspecific APC and do not contain any specific receptor. They phagocytose antigens (in our case TAA) and expose their bit string as MHC-II. DC are also aspecific APC. They internalize TAA and expose as MHC-I and MHC-II. Finally B-cells act as specific APC. If they recognize TAA they internalize it and expose as MHC-II. Plasma cells have no specific receptor (but contains the information of the original BCR). They produce antibodies with the same receptor as the B cell from which they originate.
Internal states of each cell type are summarized in Table 2.
Table 2 Entities' states. Internal states (columns) of each cell type (rows) are labelled with a •. The rows show the states of each entity.
Cell Active Re-sting Intern PresI PresII Duplica Bound-ToAb
B • • • • •
TH • • •
TC • • •
DC • • • • •
MP • • • •
CC • • •
VC • •
NK •
Each cells can be in different internal states and all cells are tracked individually throughout the course of an experimental run. In order to describe state changes we will look (see § 3.2.3) at the evolutions of each cellular entity since its initially entry in the lattice. Here we briefly describe the biological meaning of entities states.
APC, TH and B cells are initialized as Active; TC cells are initialized as Resting; VC are initialized as PresI while CC are stochastically set as PresI with an high probability. All entities are initialized with their default life time (see Age).
All cellular entities will then change their status following a successful interaction with another entity (hereafter referred as positive interaction) or by internal processing. Conditions under which a positive interaction (or the opposite negative interaction) occurs will be described later (see § 3.2.3).
• APC (MP, DC and B cells) will change their status to Intern when they interact positively with TAA; Internal processes will then change the state to PresI (MHC class I) or PresII (MHC class II). DC may present both in MHC-I and MHC-II. Positive interaction of a presenting MHC-II APC with an active TH cell will change both B cells APC and TH state to Duplica and clonal proliferation phase begins for B cells. If a PresII APC negatively interacts with a TH cell, its status can return to Active. This simulates the biological event that a presenting APC that is not stimulated by TH-cell may loose the presentation status.
• B cells change their status only when they positively interact with a TAA or with a TH cell as described above.
• Cytotoxic T cells become Active when positively interact with a presenting MHC-I DC, or with a VC in presence of IL-12 adjuvant or with a CC in presence of IL-2 previously released by an activated TH cell. Active TC cells can recognize and kill a CC by lysis. In such a case, the status of the TC changes to Duplica and clonal division starts.
• Helper T cells change their status only when they positively encounter an APC as described above.
• Cancer and vaccine cells positively interact with Antibodies and change their status to BoundToAb.
Antibodies can then kill them by complement mediated cytotoxicity or can act as signals for NK cells. Cellular entities have age structure. They are born, they interact and duplicate (and eventually get anergic) and, after a finite lifetime, they die by apoptosis or by lysis. We need to keep track of the age of cellular entities; we do this by keeping a count of the number of time-steps since cell birth (from stem cells or by clonal division). To simulate memory cells, we increase the halflife of TH, TC and B cells after successful interaction with target antigens. The death probability reaches 1 when the age gets to twice the half life. The simulator performs the process of the thymic selection of T cells (TH + TC) as described in [16]. Selection has two phases: negative and positive (both stochastic).
The simulated immune system achieves repertoire completeness by a mechanism of mutation of B-cell receptor (BCR). Mutation occurs when a new B cell is formed, i.e. in the duplication phase. This effect is included in the simulator as a stochastic event; this increases the immune system recognition ability. The setup routine initially set entities concentration in the lattice. The Leukocytes form three general classes (Granulocytes; Lymphocytes and Monocytes) which are present in different ratios in blood, tissues and various organs. In our virtual mouse we initially consider 4500 leukocytes. Of these there are 4200 lymphocytes, comprising 1512 B cells, 1512 T cells (subdivided in 1008 T helper cells and 504 cytotoxic T cells) and 1176 natural killer cells. There are 300 monocytes divided in 150 macrophages and 150 dendritic cells. The model does not have any granulocyte.
3.2.2 Molecular entities
Molecular entities included in the simulator are Antigens, Antibodies, Cytokines and Damage (Danger Signal, see [19]). Molecular entities do not have internal states and thus do not need to be modelled individually (only TAA are treated individually). Rather, we define populations which represent different specificity of molecular entities, on the lattice. Diffusion on the lattice is performed by appropriate change of the concentrations of entities on the lattice. The age structure is stochastically increased/decreased as function of production and half life time respectively.
As antigen we consider only the tumor associated antigen (p185) released by injected vaccine cells and cancer cells when they die. Antigens are represented by a number of binary segments consisting of a fixed number of bits. These segments represent the antigenic sites (epitopes and peptides) as shown in Figure 3. Epitopes are defined as the external portion of an antigen that is recognized by a B-cell receptor. Peptides are defined as portions of an antigen that can be bound by an MHC molecule and be recognized by an appropriate T cell. The epitopes and peptides are specified separately in the model. The minimum effective antigen we consider is therefore two segments long, one for B-cell epitope and one for peptide.
Antibodies (Ab) are also represented by bit strings, and have a paratope which is identical to the paratope/receptor of the B cell (plasma cell) that secretes them.
Cytokines (IL-2 and IL-12) and a danger signal are encoded as aspecific entities. Their presence in the lattice site will induce or prevent interactions by increasing or decreasing probabilities.
3.2.3 Interactions
In this section we will describe interactions following the evolution of the simulator for the first few steps. First of all we must remember that interactions occurs only if two entities stay in the same site. Taking into account that a time step is 8 hours we can say that entities in a site are those entities that a single entity encounters during 8 hours. Time t = 0 corresponds to the atypical hyperplasia, i.e. first appearing of tumor cells. For "early schedule" time t = 0 corresponds also to the first vaccine injection. An interaction between two entities is a complex action which eventually end with a state change of one or both entities. Interactions can be specific or aspecific. Specific interactions need a recognition phase between the two entities (e.g. B ↓ TAA); recognition is based on Hamming distance and affnity function and is eventually enhanced by adjuvants. We refers to positive interaction when this first phase occurs successfully. Aspecific interaction do not have a recognition phase (e.g. DC ↓ TAA). When two entities, which may interact, lie in the same lattice site then they interact with a probabilistic law. All entities which may interact and are in the same site have a positive interaction.
First positive interaction is between vaccine cells and cytotoxic T cells (TC – VC Interaction). vaccine cells are engineered in such a way that are presenting MHC class I (status Presl) and TC are in the state in which are released from thymus (status Active). Then, if TC (CD8) cell receptor matches with a non-zero affnity with the allogeneic MHCI, VC dies by lysis and release TAA. Allogeneic-MHC present in the vaccine guarantees a non-zero affnity. Positive interaction produces TC duplication (state change into Duplica) and increase TC lifetime of one time cycle (8 hours). Once TAA are released they can interact with Antigen Presenting Cells (APC), (i.e. Macrophages (MP), Dendritic cells (DC) and B cells) or antibodies. Positive APC ↓ TAA interaction will have the following effect: i) TAA is ingested by APC; ii) APC will change state becoming presenting. A presenting APC is able to stimulate other cells (TC, TH). Stimulated TH produce Interleukin-2 (IL-2). A positive interaction TH B will change the state of the B cell into Plasma Cell (PLB) and the humoral response begin by antibodies production. This briefly describes the major interactions included in SimTriplex. They can be divided into standard interaction of the immune system (B ↓ Ag; Ab ↓ Ag; TH ↓ B; TH ↓ MP; MP ↓ IC; MP ↓ Ag; DC ↓ Ag; TH ↓ DC;) and interactions which occurs in presence of Tumor and vaccine (TC ↓ CC; Ab ↓ CC; NK ↓ CC; Ab ↓ VC; NK ↓ VC). Allowed interactions are shown in table 3.
Table 3 Simulator's interactions. An interaction between two entities occurs if the row/column interaction is marked with a •. The table is obviously symmetric.
Entity 1 Entity 2 B Ag Ab TH TC MP DC IC VC CC NK
B • •
Ag • • • •
Ab • • • •
TH • • •
TC • •
MP • • •
DC • •
IC •
VC • • •
CC • • •
NK • • • •
3.2.4 Diffusion
All entities are allowed to move with uniform probability between neighboring lattices in the grid with equal diffusion coeffcient. In the present release of the simulator chemotaxis is not implemented.
4 Results
The systematic search for a new schedule is driven by the known in vivo results that we have reproduced with our model. In the following we first describe how we set up our virtual lab in-silico experiments which reproduced the in vivo results (section § 4.1); we then analyze results of the different schedules with to analyze the immune system reaction to vaccine injections (section § 4.2). Finally, by taking into account what we learned from computer experiments, we systematically search for a new schedule alternative to the chronic one, which could prevent the solid tumor formation (section 4.3).
4.1 Setting in silico experiments
In vivo experiments on sets of HER/2-neu mice have been carried, for all the schedules, up to the time in which a solid tumor is formed in the mouse. In fact, one observes that the effectiveness of vaccine action is depleted after a solid tumor is formed. In order to mimic these experiments in silico one needs to define a solid tumor on a lattice. Taking into account the number of simulated Immune System entities in the lattice we assume that a solid tumor is formed when the number of cancer cells in the lattice becomes greater than 105.
The vaccination protocols that have been used to test the effectiveness of Triplex are described in section 1. In in silico experiments we try to reproduce in vivo experiments. For this we considered all the different protocols mentioned above and the case of untreated mice. The last one is analyzed in order to tune with experimental tumor growth and ensure that simulation shows no significant immune response.
We performed in silico experiments using the standard good practice statistical procedure: i) We considered a large population of individual mice. Each individual mouse is characterized by a sequence of uniform numbers which will determine the probabilistic events. ii) We randomly extract from this population two statistical samples of 100 individual mice (hereafter referred as S1 and S2) to perform numerical experiments.
The computational time begins when the mouse is six weeks old (the observed time of atypical hyperplasia) and proceeds up to the formation of a solid tumor or up to 2 years. For each protocol we treat all mice in the sample and we measure the time in which the solid tumor is formed. The percentage of tumor free mice as function of age is shown in Figure 4 for sample S2 (the same result for sample S1 has been shown in [16]). Comparison with Figure 1 shows excellent agreement with in vivo experiments.
Figure 4 Tumor-free survival curves of virtual mice receiving the Triplex vaccine according to different protocols. Each arrow at the bottom of the graph represents one cycle of vaccination. The sequence of neoplastic progression in untreated mice is outlined under the x axis; CIS, carcinoma in situ.
4.2 Analysis of schedules' results
The general behavior of the most relevant quantities versus time has been already described in [16]. Here we take the matter again to analyze in detail the system reaction to vaccine injections in order to envisage a new vaccination protocol which prevents the solid tumor formation. To better appreciate this reaction with respect to vaccine injection timing we consider the plots of the relevant quantities separately for each schedule and we mark the vaccination times. As in [16] we consider, for the S1 statistical samples, the mean values, total number of immune cells and cancer cells in the lattice as function of time. These quantities are plotted in Figures 6, 7, 8, 9, respectively for Very Late, Early, Late and Chronic, for each schedule. Figure 5 show the same quantities for the untreated mice in S1.
Figure 5 Immune response. The immune response activation is shown for untreated virtual mice, versus time in days.
Figure 6 Immune response. The immune response activation due to the vaccine effect is shown for VERY LATE vaccination schedules, versus time in days. Red ticks above x axis represent the timing of vaccine administration.
Figure 7 Immune response. The immune response activation due to the vaccine effect is shown for EARLY vaccination schedules, versus time in days. Red ticks above x axis represent the timing of vaccine administration.
Figure 8 Immune response. The immune response activation due to the vaccine effect is shown for LATE vaccination schedules, versus time in days. Red ticks above x axis represent the timing of vaccine administration.
Figure 9 Immune response. The immune response activation due to the vaccine effect is shown for CHRONIC vaccination schedules, versus time in days. Red ticks above x axis represent the timing of vaccine administration.
The error analysis shows that in the regions where the sample has strong statistical significance (i.e. there still is a suffcient large number of mice which has not formed a solid tumor), the standard deviation always reaches a maximum of 5–8% for all entities.
First of all we notice that very late and untreated protocols show similar curves. Thus very late schedule is not relevant as it does not affect tumor growth. We will not discuss further this schedule.
Early and late protocol show a similar general behavior for all quantities. The chronic protocol shows a behavior which is different from all the others protocols.
This last three schedules show an initial transient phase characterized by a strong cytotoxic response which corresponds to a drastic reduction of cancer cells. This initial, burst like, transient phase appears also, for all the three schedules, in T helpers, cytotoxic T cells, B cells and antibodies. The antibodies grow immediately after the first vaccine injection. This effect is due to the antibodies released by vaccine cells. The general behavior is similar for all the schedules but there are significant differences:
i) the vaccine's effect on the growth of cancer cells appears after the second cycle of vaccination;
ii) cancer cells drop off roughly two months after the end of the early schedule, while in the late protocol the drop begins in between second and third cycle of vaccination;
iii) the same delay can be observed in the other entities behavior (except for natural killer and dendritic cells as explained in [16]);
iv) Cytotoxic T cells response is higher in late protocol than in early one. This is probably due to the fact that in late vaccination there are many more cancer cells than in the early;
v) Antibody response is more powerful in early than late. This is crucial for controlling tumor progression [6]: early treated mice survive about 20 weeks more than late treated mice. Early protocol is then successful in delaying the formation of the solid tumor via humoral response.
As already mentioned chronic schedule plots show a transient phase as in other schedules. This is followed by a quasi steady phase. All the plots, during the quasi steady phase, are mostly flat and characterized by small humps with maximum at the end of each vaccination cycle. The number of antibodies, after the initial increase, keeps constant. This show, as found in the in vivo data, that cancer growth is controlled by antibodies. The number of cancer cells is always in the range 103 ÷ 104.
4.3 Computer aided search for a new schedule
The search for a new schedule can be envisaged from the results, previously shown, of the three protocols giving, at least, an initial positive response, i.e. early, late and chronic.
This search is driven by the following observation: cancer cells drop off roughly two weeks later the last vaccine injection of the third cycle of the early vaccination. Following this observation we randomly choose one mouse and provide a complete early vaccination, i.e. three cycles roughly two weeks before the observed minimum of cancer cells. The time of the second vaccination was chosen from the plots of early schedule of the mice, Figure 10(a). Figure 10(b) shows that the repeated early vaccination schedule is not able to stop cancer cells from growing in number. Another complete early vaccination needed to decrease the number of cancer cells as shown in Figure 10(c). To control the tumor growth up to the end of simulation we were again forced to apply twice a complete early vaccinations as shown in Figure 10(d,e,f). The time setting for injections was done heuristically and many trials were necessary. After a number of attempts we envisage a possible alternative therapy. We then tested it in-silico for all mice of sample S1 and then for all mice of sample S2.
Figure 10 Searching a new schedule: results after early cycles for a single mice. Red ticks above x axis represent the timing of vaccine administration.
We get the tumor control for 85% of mice of the sample with the schedule shown in Figure 11, which shows the mean values for all mice of sample S1 in order to compare with previous figures.
Figure 11 Immune response. The immune response activation due to the vaccine effect is shown for a possible vaccination schedules, versus time in days. Red ticks above x axis represent the timing of vaccine administration.
This schedule is able to control the tumor formation with a number of vaccine injections 30% less than the chronic one.
5 Conclusion
We presented an in silico search for effective cancer vaccination protocols using a model describing the action of a tumor vaccine in stimulating immune response and the ensuing competition between the immune system and tumor cells. This model applies to the very early stage of tumor genesis, i.e. before a solid tumor is formed. In silico experiments show excellent agreement with in vivo experiments both for the time of formation of solid tumor in mice and the role of antibody response in controlling tumor growth [6,8]. The model and its computer implementation are very flexible and new biological entities, behavior, and interactions can be easily added. This helps achieving a realistic description of the immune responses that target solid tumor formation.
We found that a possible protocol which prevents solid tumor formation for the mice lifetime and uses a number of vaccine injections 30% less than Chronic schedule. This result shows that, at least in principle, a computerized mathematical model can be used to search for better protocols. The vaccination schedule we found is better than the Chronic one. However the question if this schedule is optimal, i.e. is the effective schedule with the minimum number of injections, or nearly-optimal is still open. The answer to this question is not trivial. First one should properly define the biological meaning of an optimal schedule; then an optimal/nearly optimal solution can be found with different techniques available in the mathematical market. We plan to further investigate this point using alternative strategies, or optimal search algorithms like simulated annealing or genetic algorithms, taking into account biological constraints.
Moreover the model we used is still naive and can be improved greatly. We are working to a more detailed model to dissert the contribution of individual vaccine's components to the protective immune response. We plan also to consider a multi-organ model in order to consider metastasis formation, and the vaccine's effect on metastasis. Work in these directions is in progress and results will be published in due course.
Note
1
2Paul E. Black, "Hamming distance", from Dictionary of Algorithms and Data Structures, Paul E. Black, ed., NIST.
Acknowledgements
F.P. and S.M. acknowledge partial support from University of Catania research grant and MIUR (PRIN 2004: Problemi matematici delle teorie cinetiche). Part of this work has been done while F.P. is research fellow of the Faculty of Pharmacy of University of Catania.
P.-L.L. acknowledges financial support from the University of Bologna, the Department of Experimental Pathology ("Pallotti" fund) and MIUR.
==== Refs
Rosemberg SA Yang JC Restifo NP Cancer immunotherapy: moving beyond current vaccines Nat Med 2004 10 909 915 15340416 10.1038/nm1100
Forni G Lollini PL Musiani P Colombo MP Immunoprevention of cancer: is the time ripe? Cancer Res 2000 60 2571 2575 10825122
Chang MH Chen CJ Lai MS Hsu HM Wu TC Kong MS Liang DC Shau WY Chen DS Universal hepatitis B vaccination in Taiwan and the incidence of hepatocellular carcinoma in children N Engl J Med 1997 336 1855 1859 9197213 10.1056/NEJM199706263362602
Koutsky LA Ault KA Wheeler CM R BD Barr E Alvarez FB Chiacchierini LM Jansen KU Proof of Principle Study Investigators. A controlled trial of a human papillomavirus type 16 vaccine N Engl J Med 2002 347 1645 1651 12444178 10.1056/NEJMoa020586
Di Carlo E Diodoro MG Boggio K Modesti A Modesti M Nanni P Forni G Musiani P Analysis of mammary carcinoma onset and progression in HER-2/neu oncogene transgenic mice reveals a lobular origin Lab Invest 1999 79 1261 1269 10532589
Lollini PL De Giovanni C Pannellini T Cavallo F Forni G Nanni P Cancer immunoprevention Future Oncology 2005 1 57 66 16555976 10.1517/14796694.1.1.57
Colombo MP Trinchieri G Interleukin-12 in anti-tumor immunity and immunotherapy Cytokine Growth Factor Rev 2002 13 155 168 11900991 10.1016/S1359-6101(01)00032-6
De Giovanni C Nicoletti G Landuzzi L Astolfi A Croci S Comes A Ferrini S Meazza R Iezzi M Di Carlo E Musiani P Cavallo F Nanni P Lollini P Immunoprevention of HER-2/neu transgenic mammary carcinoma through an interleukin 12-engineered allogeneic cell vaccin Cancer Res 2004 64 4001 4009 15173014 10.1158/0008-5472.CAN-03-2984
Delves PJ Roitt YM The immune system Advances in Immunology 2000 343 37 49
Forni G Foa R Santoni A Frati L Eds Cytokine Induced Tumor Immunogeneticity 1994 New York: Academic Press
Greller L Tobin F Poste G Tumor hetereogenity and progression: conceptual foundation for modeling Invasion and Metastasis 1996 16 177 208
Bellomo N De Angelis E Preziosi L Multiscale Modeling and Mathematical Problems Related to Tumor Evolution and Medical Therapy Journal of Theoretical Medicine 2003 5 111 136 10.1080/1027336042000288633
Bellomo N Preziosi L Modelling Mathematical Methods and Scientific Computation 1995 Boca, Raton: CRC Press
Bellomo N Pulvirenti M Modeling in Applied Sciences: A Kinetic Theory Approach 2000 Birkhauser, Boston
Motta S Brusic V Modelling in Molecular Biology Springer 2004 chap. Mathematical modelling of the immune system 193 218
Pappalardo F Lollini PL Castiglione F Motta S Modelling and Simulation of Cancer Immunoprevention vaccine Bioinformatics 2005 21 2891 2897 15817697 10.1093/bioinformatics/bti426
Farmer JD Packard N Perelson AS The immune system, adaption and machine learning Phisica D 1986 22 187 204 10.1016/0167-2789(86)90240-X
Segel LA Perelson AS Some reflections on memory in shape spaces 1989 Springer Coll of Theories of Immune Networks 63 70
Matzinger P An innate sense of danger Immunology 1998 10 399 415
|
16305756
|
PMC1287065
|
CC BY
|
2021-01-04 16:37:20
|
no
|
Immunome Res. 2005 Oct 7; 1:5
|
utf-8
|
Immunome Res
| 2,005 |
10.1186/1745-7580-1-5
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1629298110.1371/journal.pbio.0030410Research ArticleGenetics/Genomics/Gene TherapyOtherStatisticsHomo (Human)Tracing the Origin and Spread of Agriculture in Europe Origin of Agriculture in the Near East and EuropePinhasi Ron [email protected]
1
Fort Joaquim
2
Ammerman Albert J
3
1School of Human and Life Sciences, Whitelands College, Roehampton University, London, United Kingdom,2Departament de Fisica, E.P.S. P-II, Universitat de Girona, Campus de Montilivi, Catalonia, Spain,3Department of Classics, Colgate University, Hamilton, New York, United States of AmericaTyler-Smith Chris Academic EditorThe Wellcome Trust Sanger InstituteUnited Kingdom12 2005 29 11 2005 29 11 2005 3 12 e4105 4 2005 29 9 2005 Copyright: © 2005 Pinhasi 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.
Go West, Early Man: Modeling the Origin and Spread of Early Agriculture
The origins of early farming and its spread to Europe have been the subject of major interest for some time. The main controversy today is over the nature of the Neolithic transition in Europe: the extent to which the spread was, for the most part, indigenous and animated by imitation (cultural diffusion) or else was driven by an influx of dispersing populations (demic diffusion). We analyze the spatiotemporal dynamics of the transition using radiocarbon dates from 735 early Neolithic sites in Europe, the Near East, and Anatolia. We compute great-circle and shortest-path distances from each site to 35 possible agricultural centers of origin—ten are based on early sites in the Middle East and 25 are hypothetical locations set at 5° latitude/longitude intervals. We perform a linear fit of distance versus age (and vice versa) for each center. For certain centers, high correlation coefficients (R > 0.8) are obtained. This implies that a steady rate or speed is a good overall approximation for this historical development. The average rate of the Neolithic spread over Europe is 0.6–1.3 km/y (95% confidence interval). This is consistent with the prediction of demic diffusion (0.6–1.1 km/y). An interpolative map of correlation coefficients, obtained by using shortest-path distances, shows that the origins of agriculture were most likely to have occurred in the northern Levantine/Mesopotamian area.
An analysis of radiocarbon dates from early Neolithic sites reveals that agriculture in Europe most likely originated in the northern Levant and Mesopotamia and spread by population growth and migration, rather than by cultural diffusion.
==== Body
Introduction
The study of the origins of farming in the Near East and its dispersal to Europe has been a subject of major interest to archaeologists, anthropologists, linguists, and geneticists. The interest in agricultural origins can be traced to Gordon Childe [1], who proposed in 1942 that the Neolithic populations of the Near East were under substantial economic and demographic pressures triggered by marked population growth following the successful development of the Neolithic lifestyle. In his later book The Dawn of European Civilisation [2], Childe applied his demographic/Malthusian concept of population pressure and territorial expansion to the study of Neolithic Europe, asserting that the first Neolithic crops and domesticated animals did not reach Europe by means of trade or exchange but by means of migration or the colonization of farmers and shepherds from the Near East.
Clark [3,4] was the first to study the Neolithic dispersal by looking at the spatiotemporal pattern of radiocarbon dates in Europe and the Near East. Clark allocated the few carbon-14 dates available at the time to three temporal classes: Group 1, dates equal to or earlier than 5,200 BC; Group 2, dates between 5,200 and 4,000 BC; and Group 3, dates between 4,000 and 2,800 BC. His map shows a basic trend from east to west for the early Neolithic in Europe that is consistent with Childe's ideas [2].
In 1971, the first quantitative analysis of the spread of early farming in Europe was undertaken by Ammerman and Cavalli-Sforza [5], who then went on to develop a new perspective on the processes at work behind the Neolithic dispersal [6]. To measure the average rate of spread, Ammerman and Cavalli-Sforza [5] collected the radiocarbon dates from 53 early Neolithic sites, which were representative of the arrival of early farming in different parts of Europe, and performed a regression analysis. Four archaeological sites in the Near East (Jericho, Jarmo, Çayönü, and Ali Kosh) were taken as probable centers of agriculture—a fifth center used in their study was the center of gravity of their four sites—and the great-circle distance from each European site to a given center was then calculated. Diffusion rates for the respective centers were obtained from linear fits of the radiocarbon ages and the geographic distances. The results from all of the centers gave an average rate of about 1 km/y. The correlation coefficients were relatively high (R > 0.8) for each of their five centers, indicating that a regular rate offers a good overall description of the spread.
Ammerman and Cavalli-Sforza [6,7] stressed that, in principle, the observed rate could be explained as the consequence of cultural diffusion (the spread of crops and farming technology without the movement of people) or demic diffusion (the spread of farmers themselves)—or even some combination of the two. At the same time, they were the first to emphasize the role of demic diffusion in the Neolithic transition and to draw attention to the agreement between the observed average rate of spread and the one predicted by a demic wave-of-advance model [7]. Their wave-of-advance model, borrowed from the field of population biology, proposes that active population growth at the periphery of the farmers' range, in combination with local migratory activity (isotropic in character), would produce a population range expansion that moves outwards in all directions and advances at a relatively steady rate. They also predicted that the mixing of Neolithic and Mesolithic populations would lead to genetic gradients with extreme gene frequencies in the areas with the oldest Neolithic sites [6]. This prediction was confirmed in 1978 by the analysis of classical genetic markers [8]. The first principal component of the classical polymorphisms shows a geographic cline, from the south-east to the north-west of Europe [9,10], as expected under the hypothesis of demic diffusion and the interaction of Neolithic and Mesolithic populations. However, it must be added that clines can arise through several processes [11,12]. Moreover, even if a genetic cline is associated with a demic-diffusion process, it does not in itself indicate the time in which it was established [13]. On the other hand, a strong correlation is observed between genetic and archaeological distances, and this correlation supports the hypothesis of demic diffusion [14].
Molecular studies using mitochondrial DNA, Y-chromosome DNA, and nuclear DNA differ in their assessment of the contribution of Near Eastern farmers to the European gene pool. Some mitochondrial-DNA studies suggest that the contribution of Near Eastern farmers to the European gene pool is about 20% [15,16]. A similar percentage (22%) is suggested by a Y-chromosome study carried out by Semino et al. [17]. However, the data in [17] were reexamined by Chikhi et al. [18], who found (through a different methodology) an average contribution of between 50% and 65% by Near Eastern farmers to the European gene pool. Estimations depend not only on the markers employed but also on the model used (and its inherent assumptions). A recent study that makes use of mitochondrial-DNA, Y-chromosome DNA, and other autosomal markers [19] finds that the Neolithic contribution is much higher than 20%, and decreases from east to west, as expected under the Near Eastern demic-diffusion model. Finally, nuclear-DNA studies support a substantial contribution of Near Eastern populations to the European gene pool [20]. Thus, many genetic studies tend to support the idea of demic diffusion at some level, but there is still a lack of consensus with regard to the percentage of the contribution of early Near Eastern farmers to the European gene pool (see also [21,22]).
Recent archaeobotanical, archaeological, and craniometric studies suggest that, in all probability, the spread of farming to Europe was a complex process, and these studies point to the occurrence of an “aceramic” or “pre-pottery” dispersal to Cyprus, Crete, and the Argolid from various locations in the Near Eastern zone [23–26]. These studies and others highlight the complexity of the historical process of the spread of farming, suggesting significant regional variations in the dispersal process, with varying degrees of demic diffusion and cultural diffusion.
Some points that have been relatively neglected are: (1) the identification of the area in the Near East from which the spread began; (2) the computation of a statistically-significant error range for the observed speed; (3) the computation of the speed range predicted by demic diffusion; (4) the comparison of these observed and predicted ranges; (5) the effect of the Mediterranean Sea as a barrier (by computing shortest-path in addition to great-circle distances); and (6) the calibration of dates. Below we address these issues.
Our work involves a reassessment of the wave-of-advance model using a sample of 735 dates from early Neolithic sites in Anatolia, the Near East, and Europe (Table S1 and Text S1). We try to take into account all available sites that have standard errors of the mean of less than 200 radiocarbon years, including those from the Alpine region and from various regions in the Near East. We assess the correlation coefficient, R, and the rate-of-advance parameters for 25 hypothetical centers of origin of agriculture (HOA) and ten probable centers of origin of agriculture (POA). The 25 HOA are defined solely on the basis of their geographic location. The ten POAs consist of nine archaeological sites that have yielded some of the earliest evidence for cereal domestication in the Near East, plus the center of gravity from the original analysis mentioned above [5]. We then calculate four sets of R-values (one for each distance and dating method) and apply statistical methods to these sets in order to determine: (1) the most likely average speed of the spread over Europe; (2) its error range; and (3) the most likely area of the origins of agriculture. We also compute the speed range predicted by a demic diffusion model and compare it to the range inferred from the observed data.
Results
Determination of the Observed Rate of Spread of Agriculture
The values of the correlation coefficient, R, derived from the linear regressions for the ten POAs (Figure 1) are presented in Table 1. Let us first consider great-circle distances. The POA with the highest correlation coefficient is Center 3 (Abu Madi). However, eight out of the nine other POAs have values of R that overlap with the range for Abu Madi (R = 0.827 ± 0.026, using uncalibrated dates). Therefore, their R-values are similar, and they can be regarded as likely places of origin for the dispersal. Center 1 (Çatal Höyük) is the only POA with an R-range that does not overlap with that of the center with the highest value of R; it has a substantially lower value (using either uncalibrated or calibrated ages). Interestingly, Table 1 shows that this conclusion changes when we consider shortest-path distances (see Text S2 for details on the computation of shortest-path distances). When shortest-path distances are used, the center with the highest value of R is no longer the most southern one (Abu Madi) of the POAs shown in Figure 1. At the same time, all of the ten POAs now have overlapping ranges of R-values (using either uncalibrated or calibrated dates). In short, the analyses based on shortest-paths yield an area for the origins of agriculture located to the north of the one identified by the use of great-circle distances (this topic is analyzed in detail below).
Figure 1 Location of the 735 Archaeological Sites Used in the Analysis as well as the Ten POAs Listed in Table 1
Table 1 Several Sites That Have Been Considered as POAs to Europe, Their Locations, and Values of the Correlation Coefficient, R, Using the 735 Sites in Figure 1 (with 95% Confidence-Interval Errors, Obtained by Bootstrap Resampling)
In order to estimate the speed of the agricultural wave of advance, we use distances relative to the POA with the highest R-values in Table 1: Abu Madi for great circles (Figure 2A) and Cayönü for shortest paths (Figure 2B). This yields a speed range of 0.7–1.1 km/y using great circles and 0.8–1.3 km/y using shortest paths (95% confidence interval, see the caption of Figure 2). The shortest-path rate is obviously higher because the corresponding distances are equal or longer than great-circle distances, but what is very interesting is that the speed range is almost identical whether we use great circles or shortest paths. This also holds if we use calibrated dates (which yield 0.6–1.0 km/y for great circles and 0.7–1.1 km/y for shortest paths; see the caption of Figure 2). All of the other POAs (Table 1) yield essentially the same speed range (0.6–1.3 km/y). The time at which the spread began can be estimated, under the same hypothesis of linearity (straight fits in Figure 2), to fall within the interval of 9,000–10,500 years before present (BP; uncalibrated years) or 10,000–11,500 BP (calibrated years).
Figure 2 Linear Regression Fits to the Data (n = 735 Sites) for Uncalibrated Dates in Years BP and Distances of Sites Computed from the POA with the Highest R-Value in Table 1
(A) Based on great-circle distances. The speed implied by the distance-versus-time regression is the slope of the dashed line, namely 0.71 ± 0.04 km/y (in agreement with statistical theory, the error range of 0.04 km/y has been computed as twice the standard error of the slope and corresponds to a 95% confidence interval). The speed implied by the time-versus-distance regression (full line) is the inverse of the corresponding regression slope, namely 1.04 ± 0.05 km/y (95% confidence interval). Therefore, we estimate the overall speed range as 0.7–1.1 km/y. If calibrated dates are used in the analysis (top axis), the result is 0.6–1.0 km/y (see the first figure in Protocol S1).
(B) Based on shortest-path distances. The distance-versus-time regression yields 0.85 ± 0.04 km/y, whereas the time-versus-distance regression yields 1.22 ± 0.06 km/y. The overall estimated speed range is thus 0.8–1.3 km/y. If calibrated dates are used (top axis), the result is 0.7–1.1 km/y (see the second figure in Protocol S1).
Speed Predicted by a Demic-Diffusion Model
The results from Figure 2 strongly suggest that the average rate of the Neolithic transition was in the range of 0.6–1.3 km/y, and that the advance of the front took a form that was approximately linear (R > 0.8). The spread of early agro-pastoralism swept over Europe, taken as a whole, essentially at a regular speed—a rate that shows no overall trend either toward acceleration or toward deceleration over time [7]. As explained below, this range of values is compatible with that predicted by the time-delayed theory of the Neolithic transition [27]. The time-delayed theory is nothing but a refinement of the wave-of-advance model developed by Ammerman and Cavalli-Sforza [6,7]. The refinement takes into consideration the diffusive delay which is due to the generation time during which children remain with their parents and do not relocate their place of residence [27]. The time-delayed theory agrees with other human and non-human range expansions [28–30], as well as with the spread of viral infections [31].
As far as we know, no cultural-diffusion model to date has been able to derive a speed compatible with the observed range (0.6–1.3 km/y). This is an important point that has been neglected in the literature up to now. In contrast, the time-delayed demic model [27] predicts that the speed is
where a is the initial growth rate of the population number, m is the mobility, and T is the mean generation time [27]. The values of a and m have been carefully derived in previous work from plots of the population number versus time (a) and records of individual movements (m). Data from anthropological studies gathered hitherto yield estimates of 0.029–0.035/y for a, 900–2,200 km2/generation for m, and 29–35 y for T (Text S3). Using these ranges, the above formula yields a speed range of 0.6–1.1 km/y. Thus, the speed range predicted by demic diffusion, namely 0.6–1.1 km/y, is compatible with that observed, namely 0.6–1.3 km/y (obtained above from Figure 2). Our conclusion at this point is that demic diffusion predicts a speed compatible with the archaeological observations, whereas no cultural-diffusion model has been developed so far that can explain the observed speed.
Interpolative Determination of the Most Likely Region of the Origin of Agriculture
Finally, we consider a larger sample by adding 30 sites in Arabia (see Materials and Methods). The results of the HOA regressions are given in Table 2. The spatial distribution of these R-values was examined using ArcMap 8.3. We interpolated R-values using ordinary kriging (see Materials and Methods). We also checked that other methods of spatial interpolation (such as the Inverse Distance Method [32]) yield almost the same results. The two maps obtained by spatial interpolation of the R-values in Table 2 are presented in Figure 3A and 3B. They show varying grades or clines that differ in their steepness and geographic extent. The lighter a grade, the less likely it is that agriculture originated in that region. The darkest area is that with the highest interpolated value of R (R > 0.811; the lower limit, R = 0.811, was chosen in such a way that different zones can be clearly distinguished in Figure 3A and 3B). Here, progressively lighter grades surround the darkest area in an approximately concentric fashion. Using great-circle distances (Figure3A), the area of highest R-values focuses upon the Levant, and yet it also includes the north-west part of the Arabian Peninsula and the northern part of the Nile Valley. In terms of current archaeological knowledge, the latter are less likely to be involved in the origins of agriculture. Interestingly, this subregion disappears when shortest-path distances are used in the analysis (Figure 3B). When the two maps are compared, the most likely area is found to be located more to the north in the shortest-path analysis (Figure 3B). This is, in all likelihood, the better of the two maps for tracing the origins of agriculture. Figure 3B thus provides quantitative support for the view that agriculture is most likely to have originated in the area that today includes north-east Syria, northern Mesopotamia, and part of south-east Turkey near the site of Çayönü. The use of calibrated dates yields similar results (Protocol S1).
Figure 3 Interpolation Map of R-Values of HOAs (n = 765 Sites)
Using great-circle distances (A) and shortest-path distances (B), these maps are based on uncalibrated dates and a slightly larger number of sites than those used in Figures 1 and 2 and Table 1. As a consistency test, the dataset now includes 30 Arabian sites. However, the results for the speed range are very similar to those obtained for the set of 735 sites. In addition, the use of calibrated dates does not lead to substantial changes in the maps (see the third and fourth figures in Protocol S1).
Table 2 Correlation Coefficients, R, for HOAs and Uncalibrated Radiocarbon Dates (n = 765 Sites), Using Great-Circle Distances (Upper Entry in Each Cell, Used in Figure 3A) and Shortest-Path Distances (Lower Entry in Each Cell, Used in Figure 3B, See Text S1)
Discussion
We estimated the overall speed of the spread to be in the range of 0.6–1.3 km/y. The R-values in Table 1 agree well with those reported by Ammerman and Cavalli-Sforza [5,7]. The correlation coefficients that they obtained (R = 0.89 for Jericho, R = 0.83 for Jarmo, R = 0.83 for Çayönü, R = 0.84 for Ali Kosh, and R = 0.86 for their center of gravity [our tenth POA]) are slightly higher than ours. This is not surprising since Ammerman and Cavalli-Sforza [5,7] chose to leave out sites in the Alps as well as those at high latitudes in northern Europe (to avoid the time delays in the arrival of early farming owing to the ecological adaptations called for in such places). They are included here, and this leads to an increase in the data dispersion (i.e., to lower R-values). This also explains why our speed range (0.7–1.1 km/y, using great circles and uncalibrated dates as did Ammerman and Cavalli-Sforza [5,7]), is slightly lower than theirs (the 53 sites and dates used by Ammerman and Cavalli-Sforza [5,7] yield 0.8–1.3 km/y, taking Jericho as the center [which yielded their highest R-value], with a 95% confidence interval). In agreement with Ammerman and Cavalli-Sforza [5,7], we find that a number of the correlation coefficients for our POAs are greater than 0.8 (see Tables 1 and 2). The implication here is that the phenomenon, as examined at the macro level (Europe as a whole), unfolded basically in a linear fashion (see the linear fits in Figure 2A and 2B). These results are particularly noteworthy, because the average rate for the spread (about 1 km/y) is now confirmed by a dataset that is some 15 times larger than the one used more than 30 y ago in the original analysis.
In addition, our rate of advance (0.6–1.3 km/y) is similar to the one determined by Gkiasta et al. [33], who obtained a speed of 1.3 km/y in their regression analysis. They did not estimate an error range for the rate—something that is essential if one intends to develop a comparative analysis of the observed speed and the speed predicted on the basis of a model (see Results). Their value for the correlation coefficient was R = 0.73—thus lower than ours. The differences between their rate (and their R-value) and ours may be due to the following: (1) all of their sites are more recent than 8,200 BP, whereas we included sites dating back to 11,000 BP; (2) they make the working assumption that the center of origin is Jericho, whereas we performed a more comprehensive analysis of the ten POAs shown in Figure 1 and then turned our attention to those with the highest R-values; and (3) we used 735 sites (a dataset about 50% larger than the one they used).
The observed rate (0.6–1.3 km/y, from Figure 2A and 2B) is consistent with that predicted by a demic-diffusion model (0.6–1.1 km/y, from equation 1). As mentioned earlier, we are not aware of any cultural-diffusion model that predicts a range consistent with the observed speed.
It is worth noting how slow the rate is on the ground (that is, in terms of a human generation). Although there is a tendency to imagine the spread racing across the map of Europe, it actually took more than 3,000 y (or 100 human generations) for the Neolithic transition to reach north-west Europe. What is involved—again on the macro level for Europe as a whole—is a slow, gradual process. At the same time, in the light of the early maritime spread of farming to Cyprus from the mainland, one can ask the following question: why did it then take almost 1,000 y to get to Crete, the next offshore island in the Mediterranean? At several sites on Cyprus, there is now good evidence for the arrival of the Neolithic package of domesticated crops and animals from the mainland by around 8,200 BC (calibrated). The fact that people were already using boats on a regular basis is also shown by the occurrence of obsidian (a volcanic glass used for making chipped-stone tools), which has its sources in Anatolia, at the same sites in Cyprus.
We reach much the same conclusion about the use of boats in the case of southern Italy, where obsidian from nearby islands is found at the oldest Neolithic sites in the region. Given the common use of boats in both parts of the Mediterranean, one might expect a faster rate for the spread between Cyprus and Italy than the one we observe. Why, in a maritime context, was the average speed in the central Mediterranean so slow? This is a puzzle that calls for further investigation. In fact, the slowness of the overall spread and its essentially linear character, as shown by the present analysis, may offer one of the best lines of argument for demic diffusion. Cultural diffusion can, and probably should, go faster. An excellent example is pottery, which appeared after the aceramic Neolithic and spread more rapidly than early farming [34].
The results of the great-circle analysis indicate that the area with the highest R-values (R > 0.811) encompasses the southern Levant and southern Mesopotamia; it contains five of the POAs (see Figure 3A). The shortest-path treatment, which takes into account the role of the Mediterranean as a barrier (see Text S2), gives an area with high R-values (R > 0.811), and this area shows a better match with the evidence regarding plant domestication since it does not include northern Egypt and the Red Sea. The area that it identifies as the most likely source for the spread of early agriculture is located in the northern Levant and the northern part of Mesopotamia; it contains six out of ten of the POAs (Figure 3B).
College et al. [25] examined the archaeobotanical remains recovered from 40 aceramic Neolithic sites in the Near East and south-east Europe. They note the similarity of most of the southern Levantine, Cypriot, and Aegean sites. They conclude that the contrast between these sites and those in Anatolia and the Euphrates Valley/central Steppe region of Syria points to the possibility of two dispersal routes. One route is a maritime-based colonization of Cyprus, central Anatolia, Crete, and Greece starting from a Levantine core region, as previously suggested by Van Andel and Runnels [35]. The second route is a land route from central/western Anatolia, reaching Thrace and south-east Europe. It would be of interest, in future work, to try to test statistically the two-route model put forward by College et al. [25].
Pinhasi and Pluciennik [26], in their analysis of craniometric affinities between populations, point to the homogeneity between Çatal Höyük and early Neolithic Greek and south-eastern European groups. This homogeneity contrasts with the pronounced heterogeneity found among other Pre-Pottery Neolithic groups in the Near East. On the basis of these results, they hypothesize that a founder population from central Anatolia (represented by specimens from Çatal Höyük) spread into south-east and central Europe. The results of the shortest-path analysis of the POAs could be consistent with their position, since they suggest that Çatal Höyük falls in the region adjacent to the one with the maximum R-values (Figure 3B).
We concur with Özdog˘an's assertion that “an unbiased reassessment of the evidence strongly implies that there were multiple paths in the westward movement of the Neolithic way of life” ([36], pp. 51–52). Aceramic Neolithic levels at sites on Cyprus (late ninth millennium BC [calibrated]), Crete and the Argolid (eight and early seventh millennia BC [calibrated]) are strongly suggestive of an initial population dispersal wave from one or more centers in the Near East [37]. At the present time, it is unclear whether farming reached south-east Europe by means of a secondary demic expansion from Anatolia or as a continuation of the initial dispersal involving Cyprus, Crete, and mainland south-east Greece. In any event, Figure 3B does provide, at this stage of research, spatial information regarding differing grades of likelihood for tracing the origins of agriculture.
In closing, we would like to stress again that our aim here is not to deny the existence of regional variability in Europe, nor to deny that local populations of late hunters and gatherers may have made a significant contribution to the Neolithic transition in certain regions [38,39]. On the other hand, for many areas of southern Europe, it remains an open question as to whether or not local populations of foragers were actually living there in the time just before the Neolithic transition. In countries such as Greece and Italy, where many archaeological surveys have been carried out over the past 30 y, very few late Mesolithic sites have come to light so far [40]. Our purpose here is to return to the big picture. Indeed, the pattern of dispersion shown in Figure 2B implies that the processes involved may have been extremely complex and at least to some extent geographically non-homogeneous. This is precisely why it is important to consider more fully what makes it possible for the very simple formula shown in equation 1 to account for the average rate of spread over Europe. While our analysis takes a mathematical approach to the overall Neolithic spread, and by doing so, we are not in a position to tackle the question of a mosaic of regional processes, we nevertheless think that the high R-values obtained in our new analysis show that, at the macro level of human population biology, the wave-of-advance model is not just a mathematical artifact. Rather, it points to an overall spatiotemporal pattern in the spread of the Neolithic lifestyle, which best agrees with an initial dispersal from the Levantine/Mesopotanian core region to Europe, and which does not exclude subsequent range expansions, colonizations, and jump dispersals.
Materials and Methods
Samples
Our radiocarbon dataset includes uncalibrated dates of the earliest Neolithic occupation from the earliest-dated levels of 735 sites in the Near East, Europe, and Asia (see Figure 1), as well as from 30 Arabian sites (all dates are available, see Table S1). The dates were obtained from the following online databases: UK Archaeology Data Service (http://www.ads.ahds.ac.uk, Canew database (http://canew.org, Near Eastern Radiocarbon Context database (http://www.context-database.de, and the Radon Workgroup database (http://www.jungsteinsite.de/radon/radon.htm). We selected the earliest date of Neolithic occupation for each site. We used only dates that have standard errors of the mean of less than 200 radiocarbon years. We omitted all dates with higher error intervals, as well as outlier dates (i.e., early occupation dates that are rejected by most archaeologists as being erroneously too early or too late, see Table S2). We preferred, whenever possible, to take dates coming from charcoal or bone collagen rather than shells. The sites and corresponding dates provide a secure sample for the earliest appearance of each of the early Neolithic archaeological cultures of Europe (such as the Linienbandkermik, Starçevo, Körös, Çris, Cardial, and so forth). Dates from the Alpine and Scandinavian regions were also included, despite the late appearance of Neolithic occupation in these regions owing to the “delayed” adoption of agriculture in these zones for environmental reasons [5].
Our basic approach to the analysis is to include all of the robust data that are available, without trying to be more selective or rigorous about dates that give a good estimate of the first appearance of the Neolithic in a given area. On the one hand, there is a virtue in this approach: one avoids bias (making choices that are arbitrary). On the negative side, however, one includes some data that are weak. The measured correlation coefficients will be lower, in all likelihood, than those obtained by using only high-quality data (for example, accelerator mass spectrometry [AMS] dates). AMS dates have two main advantages: (1) they can be obtained directly from seeds and bones, and (2) the smaller size means that one is often in a position to date samples of higher quality than previously. There is a consensus that AMS dating represents a major advance for the study of the Neolithic transition in Europe [40]. Out of our total of 765 dates, comparatively few have been obtained by AMS. Clearly, a relevant point is the trade-off between the quality and the quantity of the dates used. In measuring the overall rate, this probably makes little difference. For the estimation of regional rates in Europe (not the aim of the present paper), it may make more of a difference. The advantage of our global approach is the sheer quantity of data that are available today.
In Table 1, Center 1 has been investigated by Pinhasi [41]. Center 4 (Jericho) has been considered, among others, by Gkiasta et al. [33]. We chose the other centers (Centers 1–9) on the basis of early Neolithic sites that are POAs [42]. For comparative purposes, we also include Center 10, which was proposed by Ammerman and Cavalli-Sforza [5,7]: Center 10 is the geographic center of gravity of the sites of Jericho, Jarmo, Cayönü, and Ali Kosh.
In Table 2 and Figure 3, we added 30 sites in Arabia to the 735 sites used in Table 1 and Figure 2, primarily as a consistency test of the results from Figure 2A and 2B. We chose not to include sites in North Africa because they are much more controversial and are few in number [43,44]; some of them also have problems associated with their chronometric provenance. In addition, there are no reliable dates from Egypt and the Sahara, as the Neolithic occupation there involved nomadic tribes with domesticated cattle (and in some cases pottery, but without permanent dwellings and other Neolithic criteria such as early cereals). In any case, even if reliable dates become available in the future, using North-African sites will presumably not significantly change the interpolative R-value maps (see Figure 3A and 3B) as they are located far to the west of the Near Eastern region. In contrast, the interpolated maps are sensitive to the inclusion of the 30 Arabian sites because they are geographically close to the Near East (moreover, there are no other early sites further to the south). In any case, when the 30 Arabian sites were included, we obtained the same speed range (0.6–1.3 km/y) as that obtained above with the 735-site sample only. Therefore, we can say that the 30 Arabian sites serve as a self-consistency check of the speed range computed.
Calibration
Uncalibrated radiocarbon dates are based on the premise that the atmospheric ratio of carbon-14 to carbon-12 has been constant over time. However, this premise is only approximately valid. Briefly, carbon-14 dates can be calibrated by using tree-ring, glacial, ice-core, and other known climatic sequences. We applied the CalPal calibration software package (www.calpal.de) to all dates and their standard errors of the mean, using specifically the CalPal January 2004 calibration curve, which is based on six datasets comprising tree-ring, lacustrine, and glacial data. For more details, together with figures corresponding to Figure 2A and 2B and Figure 3A and 3B, but obtained using calibrated instead of uncalibrated dates, see Protocol S1. It can be seen in Protocol S1 that the figures do not change appreciably after calibration is taken into account, so that the conclusions remain the same.
Distances computations
A great-circle distance between two geographic points is the shortest distance along the circle on the Earth's surface (considered as a sphere) that contains both points. Shortest-path distances take into account the fact that some great-circle distances are not realistic, owing primarily to the presence of the Mediterranean Sea in our case (Text S2).
Statistical analysis
We calculated two linear regressions for each of the ten POAs in Figure 1, namely the distance-against-date and the date-against-distance regressions. Both were computed using the radiocarbon dates of the 735 sites and their distances (in km) from a given POA. The distance-versus-date regression corresponds to fitting a linear model to predict the position (distance) of the front of the population spread after a given time has elapsed, but it also makes sense to fit a linear model the other way (date-versus-distance). It corresponds to predicting the time that it takes for the population wave front to travel a given distance. Combining both fits is useful in order to estimate a wider, more reasonable, error range for the observed speed (see Figure 2A, caption). Comparing the speed predicted by a theoretical model (in this case, a demic reaction-diffusion model, by means of equation 1) to the observed speed is much more meaningful if, as is done here, the error ranges of both the observed and predicted speeds are determined.
Spatiotemporal analysis
The interpolative method of ordinary kriging [32] (used in Figure 3A and 3B) takes into account the R-values at surrounding locations in order to obtain the R-value at another location. This method fits the R-values to a sum of two functions. The first function is a polynomial of latitude and longitude. The second function has zero average, and its difference between two spatial points does not depend on their locations but only on the distance between them. This second function is used in an attempt to control for autocorrelation between the values of a geographic variable at nearby points, a basic principle in geography [32]. We also found that other methods of spatial interpolation yield almost the same result as that shown in Figure 3A and 3B (e.g., the Inverse Distance Method, where the second function does not make use of autocorrelation but instead makes use of a simple algorithm based on distance [32]).
Supporting Information
Protocol S1 Computation and Effect of Calibrated Dates
Protocol S1 includes four figures, which correspond with and are very similar to, Figure 2A and 2B and Figure 3A and 3B, but use calibrated (instead of uncalibrated) dates. Therefore, using calibrated dates does not lead to any substantial change in the results and conclusions made here.
(474 KB PDF).
Click here for additional data file.
Table S1 Information about 765 Neolithic Sites
Latitude/longitude, radiocarbon date, and additional archaeological information.
(1.1 MB XLS).
Click here for additional data file.
Table S2 List of Discarded Sites
See Materials and Methods.
(15 KB XLS).
Click here for additional data file.
Text S1 Neolithic Data
An explanation of the entries included in Table S1 is presented.
(102 KB PDF).
Click here for additional data file.
Text S2 Computation of Shortest-Path Distances
An explanation is presented of the approach that we used to compute the shortest-path and great-circle distances included in Table S1. Comparisons are provided of a great-circle distance, the corresponding shortest-path-on-Earth distance, and the intermediate approach we used. The latter computes shortest-path distances taking into account the possibility of some sea travel (as implied by the presence of early Neolithic sites on islands such as Cyprus, Crete, Lipari, and Sardinia) but not long-distance voyaging (e.g., from one end of the Mediterranean to the other).
(165 KB PDF).
Click here for additional data file.
Text S3 Demographic Data
Comparison and comment are presented with regard to the demographic data available for the determination of the parameters a, m, and T, which are used in equation 1 to predict the speed of the wave of advance.
(99 KB PDF).
Click here for additional data file.
JF was supported in part by the Generalitat de Catalunya under grant SGR-2005–00087, and by the Ministry of Education and Culture grant REN-2003–00185. We would like to thank Stephen Shennan and Stuart Semple for their comments on and corrections to the manuscript. We also thank William Kilbride for providing access to data from the Archaeology Data Service, and Nicolas Ray and Núria Roura for their help with the GIS software.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. RP gathered the data (Tables S1 and S2), calibrated the dates, computed the great-circle and shortest-path distances, and prepared Figures 1 and 3. RP and JF jointly prepared the tables, wrote the paper, and prepared Text S1. JF conceived the computation of shortest-path distances, summarized the computation in Text S2, prepared Text S3 and Figure 2, and computed the observed and predicted speed ranges. AJA made suggestions on the interpretation of the results, on the shortest-path distance computations, on the calibration of dates, and on the write-up of the paper.
Citation: Pinhasi R, Fort J, Ammerman AJ (2005) Tracing the origin and spread of agriculture in Europe. PLoS Biol 3(12): e410.
Abbreviations
AMSaccelerator mass spectrometry
BPbefore present
HOAhypothetical center of origin of agriculture
POAprobable center of origin of agriculture
==== Refs
References
Childe VG What happened in history 1942 Harmondsworth (UK) Penguin Books 303
Childe VG The dawn of European civilisation 1968 London Routledge and Kegan Paul 368
Clark JGD Radiocarbon dating and the spread of farming economy Antiquity 1965 39 45 48
Clark JGD Radiocarbon dating and the expansion of farming culture from the Near East over Europe Proc Prehist Soc 1965 31 58 73
Ammerman AJ Cavalli-Sforza LL Measuring the rate of spread of early farming in Europe Man NS 1971 6 674 688
Ammerman AJ Cavalli-Sforza LL Renfrew C A population model for the diffusion of early farming in Europe The explanation of culture change 1973 London Duckworth 343 357
Ammerman AJ Cavalli-Sforza LL The Neolithic transition and the genetics of populations in Europe 1984 Princeton Princeton University Press 176
Menozzi P Piazza A Cavalli-Sforza LL Synthetic maps of human gene frequencies in Europeans Science 1978 201 786 792 356262
Cavalli-Sforza LL Menozzi P Piazza PA Demic expansions and human evolution Science 1993 259 639 646 8430313
Cavalli-Sforza LL Menozzi P Piazza PA The history and geography of human genes 1994 Princeton Princeton University Press 428
Barbujani G Bartorelle G Genetics and the population history of Europe Proc Natl Acad Sci U S A 2001 98 22 25 11136246
Fix AG Gene frequency clines in Europe: Demic diffusion or natural selection? J R Anthropol Inst 1996 2 625 643
Currat M Excoffier L The effect of the Neolithic expansion on European molecular diversity Proc Biol Sci 2005 272 679 688 15870030
Sokal RR Oden NL Wilson C Genetic evidence for the spread of agriculture in Europe by demic diffusion Nature 1991 351 143 144 2030731
Richards MB Macaulay V Bandelt HJ Sykes BC Phylogeography of mitochondrial DNA in Western Europe Ann Hum Genet 1998 62 241 260 9803269
Richards M Macaulay V Hickey E Vega E Sykes B Tracing European founder lineages in the Near Eastern mtDNA pool Am J Hum Genet 2000 62 241 260
Semino O Passarino G Oefner PG Lin AA Arbuzova S The genetic legacy of Paleolithic Homo sapiens : A Y chromosome perspective Science 2000 290 1155 1159 11073453
Chikhi L Nichols RA Barbujani G Beaumont MA Y genetic data support the Neolithic demic diffusion model Proc Natl Acad Sci U S A 2002 99 11008 11013 12167671
Dupanloup I Bertorelle G Chikhi L Barbujani G Estimating the impact of prehistoric admixture on the genome of Europeans Mol Biol Evol 2004 21 1361 1372 15044595
Chikhi L Destro-Bisol G Bertorelle G Pascali V Barbujani G Clines of nuclear DNA markers suggest a largely Neolithic ancestry of the European gene pool Proc Natl Acad Sci U S A 1998 95 9053 9058 9671803
Cavalli-Sforza LL Ammerman AJ Biagi P Returning to the Neolithic transition in Europe The widening harvest 2003 Boston Archaeological Institute of America 297 313
Sykes B Ammerman AJ Biagi P European ancestry: The mitochondrial landscape The widening harvest 2003 Boston Archaeological Institute of America 315 326
Gronenborn D A variation on the basic theme: The transition to farming in southern central Europe J World Prehist 1999 2 23 210
Özdog˘an M The beginning of Neolithic economies in southeastern Europe: An Anatolian perspective J Eur Arch 1997 5 1 33
College S Conolly J Shennan S Archaeobotanical evidence for the spread of farming in the Eastern Mediterranean Curr Anthropol 2004 45 S35 S58
Pinhasi R Pluciennik M A regional biological approach to the spread of farming in Europe: Anatolia, the Levant, South-Eastern Europe, and the Mediterranean Curr Anthropol 2004 45 S59 S82
Fort J Méndez V Time-delayed theory of the Neolithic transition in Europe Phys Rev Lett 1999 82 867 871
Fort J Population expansion in the western Pacific (Austronesia): A wave of advance model Antiquity 2004 77 520 530
Fort J Pujol T Cavalli-Sforza LL Palaeolithic populations waves of advance Camb Archaeol J 2004 14 53 61
Ortega-Cejas V Fort J Méndez V The role of the delay time in the modelling of biological range expansions Ecology 2004 85 258 264
Fort J Méndez V Time-delayed spread of viruses in a growing plaque Phys Rev Lett 89 2002 1 4 (178101)
Johnston K ver Hoef JM Krivoruchko K Lucas N Using ArcGIS Geostatistical Analyst 2000 New York ESRI Press 300
Gkiasta M Russel T Shennan S Steele J The Neolithic transition in Europe: The radiocarbon record revisited Antiquity 2003 77 45 62
Cavalli-Sforza LL Bellwood P Renfrew C Demic diffusion as the basic process of human expansions Examining the farming/language dispersal hypothesis 2002 Cambridge McDonald Institute for Archaeological Research 79 88
Van Andel TH Runnels CN The earliest farmers in Europe Antiquity 1995 69 481 500
Özdog˘an M Comments on College et al. article Curr Anthropol 2004 45 S51 S52
Perlès C The Early Neolithic in Greece: The first farming communities in Europe 2001 Cambridge Cambridge University Press 356
Price TD Europe's first farmers 2000 Cambridge Cambridge University Press 395
Rowley-Conwy P How the west was lost Curr Anthropol 2004 45 S83 S113
Ammerman AJ Biagi P The widening harvest: The Neolithic transition in Europe—Looking back, looking forward 2003 Boston Archaeological Institute of America 230
Pinhasi R A new model for the spread of the first farmers in Europe Documenta Praehistorica 2004 30 1 76
Van Zeist W Bakker-Heeres JAH Archaeobotanical studies in the Levant I. Neolithic sites in the Damascus Basin: Aswad, Ghoraifé, Ramad Palaeohistoria 1982 24 165 256
Ehret C Keita SOY Newman P The origins of Afroasiatic Science 2004 306 1680
Bellwood P Response to “The origins of Afroasiatic” Science 2004 306 1680
|
16292981
|
PMC1287502
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e410
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030410
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1633604910.1371/journal.pbio.0030412Research ArticleNeuroscienceHomo (Human)Neural Substrate of Body Size: Illusory Feeling of Shrinking of the Waist Construction of Body Image in Parietal CortexEhrsson H. Henrik [email protected]
1
Kito Tomonori
2
Sadato Norihiro
3
Passingham Richard E
1
4
Naito Eiichi
2
5
1Wellcome Department of Imaging Neuroscience, Institute of Neurology, London, United Kingdom,2Graduate School of Human and Environmental studies, Kyoto University, Sakyo-Ku, Kyoto, Japan,3Department of Cerebral Research, National Institute for Physiological Sciences, Okazaki Aichi, Japan,4Department of Experimental Psychology, University of Oxford, Oxford, United Kingdom,5ATR Computational Neuroscience Labs, Hikaridai, Seika-Cho, Soraku-Gun, Kyoto, JapanLackner James Academic EditorBrandeis UniversityUnited States of America12 2005 29 11 2005 29 11 2005 3 12 e41214 3 2005 3 10 2005 Copyright: © 2005 Ehrsson 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.
Neural Basis of Body Image: How to Lose Inches at the (Perceived) Flick of the Wrist
The perception of the size and shape of one's body (body image) is a fundamental aspect of how we experience ourselves. We studied the neural correlates underlying perceived changes in the relative size of body parts by using a perceptual illusion in which participants felt that their waist was shrinking. We scanned the brains of the participants using functional magnetic resonance imaging. We found that activity in the cortices lining the left postcentral sulcus and the anterior part of the intraparietal sulcus reflected the illusion of waist shrinking, and that this activity was correlated with the reported degree of shrinking. These results suggest that the perceived changes in the size and shape of body parts are mediated by hierarchically higher-order somatosensory areas in the parietal cortex. Based on this finding we suggest that relative size of body parts is computed by the integration of more elementary somatic signals from different body segments.
Functional magnetic resonance imaging (fMRI) during a perceptual illusion suggests that activity in the parietal cortex is important for the integration of body image information.
==== Body
Introduction
The term “body image” commonly refers to the perception of the spatial dimensions of the body, its size, shape, and relative configuration of its parts [1,2]. The perception of the size and shape of body parts is archetypical and an important aspect of the body image. Everyday examples in which information about body size is used include when feeling ourselves as thin or large, or when walking through a narrow doorway. Unlike more elementary bodily senses such as limb movement, touch, and pain, there are no specialized receptors in the body that provide information to the brain about the size and shape of body segments. Furthermore, the somatotopically organized maps of the body surface in the somatosensory cortex that receive the sensory inputs from the peripheral receptors do not contain any explicit information about the relative size of body parts. Thus, an important question in sensory neuroscience is how the central nervous system computes the relative size and shape of the body and its parts.
Psychophysical studies suggest that the perceived relative size of body parts depends on the integration and comparison of somatic signals from different body segments [3–5] and of visual information from the body [6]. The size of body parts is probably represented in a relative sense, that is, relative to the size of other body parts and objects in the external environment. These notions are supported by the fact that people can experience illusions that the size and shape of a body part is changing when the central nervous system receives conflicting sensory signals from different body parts [3–5,7]. Likewise, in the absence of afferent sensory inputs, for example during anaesthesia of a limb or after amputation, subjects sometimes perceive changes in the size and shape of body parts [8,9]. Under certain pathological conditions affecting the parietal cortex, such as stroke [10], epilepsy with parietal focus [11–13], or somesthetic auras during migraine [14–16], people can experience changes in the size and shape of one or several limbs and body parts. Although these studies suggest that the brain processes information about body size and shape, and indicate that the parietal lobes are involved, the underlying neuronal substrate and mechanisms remain uncertain.
Here we used functional magnetic resonance imaging (fMRI) to investigate the neural correlates of perceptual changes of the size and shape of a body part. To experimentally manipulate the body image, we took advantage of a perceptual illusion—the “Pinocchio illusion” [4]—during which subjects feel that a body part changes its size and shape. This illusion has been demonstrated to work for both the length of the nose and for the width, height, and shape of various other body parts [4]. These illusions make use of the fact that vibration of the skin over the tendon of a joint extensor muscle elicits a vivid kinaesthetic illusion that the joint is passively flexing [17–21]. It is now well established that the illusory movements are caused by the excitation of muscle spindles in the vibrated muscle [18,22,23]. The afferent signals from this stimulation reach the primary somatosensory cortex [24–26] and primary motor cortex [19–21,26,27]. What Lackner [4] demonstrated was that if the hand is in direct contact with another body part, e.g., the nose or the waist, the subjects will not only feel that the vibrated wrist is bending but also experience the other body part being stretched or shrinking. In these situations the distortion of the body image is determined by the pattern of sensory stimulation according to a strict perceptual logic, so that the changes in shape and size of a body part appear to be caused by the illusory movement of the hand [4] (Protocol S1; Figures S1 and S2). For example, if the hand is grasping the nose and the biceps tendon is vibrated, one experiences the illusion that the hand is moving away from the face and the nose is becoming elongated. In contrast, when the triceps tendon is vibrated one feels that the hand is moving towards the face and that the nose is becoming shorter. We made use of the “waist-shrinking illusion.” The subjects put their hands so that the palms are in direct contact with the lateral sides of waist and the hips (see Figure 1). Then, when the tendons of the wrist extensor muscles are vibrated, the participants not only feel that the hands are bending inwards, but they also have the experience that the waist and the hips are shrinking.
Figure 1 The Position of the Two Hands Relative to the Body and the Factorial Design of the Experiment
When the palms of the hands were in contact with the body, the vibration of the two wrists elicited the illusion that the wrists were passively flexing and the waist and hips were shrinking (A, lower right). When the hands were not in contact with the body, the vibration of the wrists only elicited the illusion that the hands were flexing (A, top right). In two additional conditions, we vibrated the skin over the styloid bone beside the tendon, which does not elicit any illusions (A, top left and lower left). The neural effect of the shrinking-body illusion can be modelled as the interaction term between hand position and site of vibration in a 2 × 2 factorial design (see [B], [tendon contact– skin contact] – [tendon free – skin free]).
We hypothesised that activity in higher-order somatosensory areas in the parietal cortex would reflect the perceived changes in waist size. These areas receive somatic information from different body parts [7,28–32] and thus, theoretically, have the capacity to integrate this information to compute the relative size of different body parts.
To reveal the activity associated with the feeling of waist shrinking, we used a 2 × 2 factorial design where the illusion was modelled as the interaction between hand position (attached to body [contact]/not attached [free]) and site of vibration (wrist tendon [tendon]/skin beside the tendon [skin]); see Figure 1 and Materials and Methods]. Before the brain scan commenced, we tested all subjects to make sure that they experienced a strong and reliable shrinking-waist illusion (see Results and Materials and Methods). Also, during these test sessions we quantified the illusion and we later used this measure to relate the strength of the illusion to the brain activity.
Results
Psychophysics
Pilot experiments on 24 participants had shown that the shrinking-waist illusion starts quickly, persists throughout a 30-s period of stimulation, and can be repeatedly elicited for as many times as required for the fMRI experiment. These initial observations were confirmed by the psychophysical experiments before the brain scans.
All 17 subjects that participated in the brain scan reported that they felt as if their hands were flexing passively and their waist shrinking during the tendon contact condition when asked to describe their experiences without leading questions. When asked to select a picture out of six pictures showing different conceivable body-image distortions, all subjects selected the picture showing that the waist was shrinking and the hands were bending towards the body (see Materials and Methods). It is important to note that participants did not report the sensation that the hands were moving into the body.
The participants rated the vividness of the shrinking body illusion as 6.8 ± 1.7 (mean ± standard deviation [SD]: rating from 0–9) and continuance as 8.1 ± 1.3 (mean ± SD) in tendon contact. They also reported that the illusion started after 3.3 ± 2.0 s (mean ± SD) of vibration. Thus, the illusion of waist shrinkage was vivid and reliable, and it started quickly. Furthermore, the quantification of the illusion showed that the subjects experienced that their wrists flexed by 13.6° ± 7.5° (mean ± SD for right and left wrist). This corresponded to a reduction in waist width by 9.4 ± 3.5 cm (mean ± SD), which is a 28% reduction in width.
We also quantified the illusion of wrist movement in the condition in which the hands did not touch the body (tendon free). Consistent with previous studies [21], they felt that their wrists were flexing by 22.9° ± 15.4° (mean ± SD) in this condition. The fact that the illusory wrist movements were greater in this control condition than in the shrinking-waist condition was expected because hand contact with body parts tends to reduce the wrist illusions [21].
Finally, the electromyograms (EMGs) showed no muscular activity in 12 of the 17 subjects. Five subjects showed some weak (<150 μV; in the order of 1%–3% of maximal voluntary contraction) and brief (<5 s) EMG responses. These activities were recorded both in the extensor carpi radialis (ECR) and flexor carpi radialis (FCR), and did not seem to have any relationship with the illusions. Importantly, the muscle activity was not significantly different in tendon contact and tendon free when we performed a quantitative integrated EMG analysis (p > 0.05 paired t-test; see Figure S3). Thus, the weak muscular activity that occurred in some subjects was matched in the comparisons between the tendon conditions and could not therefore have influenced our imaging results.
Brain Imaging
First, we analysed the activity that reflected the shrinking-body illusion that could not be attributed to the effects of vibrating the wrist tendon or the position of the arms. This activity is given by the interaction term in the factorial design ([tendon contact – skin contact] – [tendon free – skin free]) (see Methods and Methods; Figure 1). We found one cluster of active voxels in the whole brain that was located in the left parietal lobe (size: 200 mm3; p < 0.05 corrected; see Figure 2). The cluster was located in the anterior part of the intraparietal sulcus and extended rostrally to the postcentral sulcus. This cluster contained two distinct peaks of activation (Figure 2). The anterior peak was located in the postcentral sulcus (x = −54, y = −30, z = 57 [x, y, and z coordinates in the standard space of the Montreal Neurological Institute]; t = 4.76; p < 0.001 uncorrected) near the junction of this sulcus and the intraparietal sulcus. The posterior peak was located at the border between the superior parietal convexity and the anterior part of intraparietal sulcus (x = −45, y = −39, z = 60; t = 3.86; p < 0.001 uncorrected). In the right hemisphere, there was a statistical trend for activation in the corresponding parietal sites (intraparietal sulcus: x = 42, −30, 60; t = 2.91; p = 0.005 uncorrected; superior parietal gyrus: x = 48, y = −33, z = 72; t = 3.15; p = 0.003; not shown in the Figures). The activity in the two left parietal foci was not significantly greater than the activity in the right corresponding areas (p > 0.19 uncorrected). Hence, the parietal cortex appears to be bilaterally engaged, albeit with only a statistical trend for activation on the right side.
Figure 2 Activity that Reflects the Illusion that the Waist Was Shrinking (Interaction Effect: p < 0.001 Uncorrected)
Top row: Activation of the cortices lining postcentral sulcus near its junction with the intraparietal sulcus. Lower row: Activation of cortex lining the intraparietal sulcus. The activations (colour) are superimposed on a normalized high-resolution T1-weighted image of a representative participant (black and white). The coordinates for the displayed slices are shown, and the crossing of the blue lines indicates the location of the activation peaks. R and L denote the right and left hemispheres, respectively. The plots to the right show the contrast estimates with the standard bars corresponding to the standard error (SE).
When we analysed the interaction term that reflected the shrinking-waist illusion, no activation, not even at a very low significance level (p < 0.05 uncorrected), was detected in the primary somatosensory cortex (areas 3a, 3b, and 1), the parietal operculum (SII), or the primary motor cortex (areas 4a and 4p). This observation supports our notion that the experimental design successfully matched the effects related to skin vibration, tendon vibration, and arm position. When these effects are not carefully matched, we know from our earlier studies that these areas are activated [21].
Next, we investigated whether there was a relationship between the activity in the parietal cortex and the strength of the body-image illusion. Because we had quantified the strength of the illusion for each subject in the test sessions prior to the scans, we could examine how the blood oxygenation level-dependent (BOLD) signal in the parietal cortex related to these illusion ratings. First, we used a linear regression model to search for voxels in the left intraparietal cortex in which the activity was related to the degree of body shrinkage across subjects (Figure 3). We found a peak of activation in the most anterior part of the intraparietal sulcus (Figure 3B; x = −48, y = −33, z = 54; t = 2.69; p < 0.009 uncorrected, R
2 = 0.32, Pearson's R = 0.57). This peak was located within the cluster of active voxels detected in the factorial design above, and therefore it probably corresponds to the same area. Also just adjacent, we found a peak in the border zone between the anterior part of the intraparietal sulcus and the left superior parietal gyrus (Figure 3A; x = −39, y = −42, z = 72, t = 3.81; p < 0.001 uncorrected, R
2 = 0.4916, Pearson's R = 0.70). Second, we examined exactly those peak voxels that were detected in the interaction analysis. As shown in Figure 3C and 3D, there was a linear relationship between the degree of illusory waist shrinking and the BOLD activity at these sites (p < 0.05 uncorrected). Taken together, these findings demonstrate that the subjects who reported the strongest shrinking-waist illusion also showed the strongest BOLD signal in the left postcentral sulcus and in the anterior part of left the intraparietal cortex.
Figure 3 Linear Relationship between Parietal Activity and the Strength of the Shrinking-Waist Illusion
Each dot represents the values for one individual subject. The data is fitted with a least-squares regression line. In (A) and (B) we plot the activity from the peaks in the intraparietal region that showed the most significant relation between illusion strength and neuronal activity ([A]: x = −39, y = −42, z = 72, t = 3.81; p < 0.001 uncorrected, R
2 = 0.4916, Pearson's R = 0.70; [B]: x = −48, y = −33, z = 54; t = 2.69; p < 0.009 uncorrected, R
2 = 0.32, Pearson's R = 0.57). These peaks were identified by using SPM2 to search for parietal voxels using a second-level linear regression model. (C) and (D) show the relationship between illusion and activity at exactly those peak voxels detected in the interaction analysis (see Figure 2). (C) shows the cortex at the junction between the postcentral sulcus and the intraparietal sulcus (p < 0.027, y = 0.0385x + 0.1424, R
2 = 0.23, Pearson's R = 0.48), and (D) illustrates the anterior part of the intraparietal cortex (p < 0.016, y = 0.0552x + 0.0488, R
2 = 0.27, Pearson's R = 0.52). In all plots the y-axis indicates the BOLD response (contrast estimates for interaction effect) in the parietal cortex, and the x-axis indicates the illusory displacement of the wrists when in contact with the body (which corresponds to the degree of waist shrinking). These regressions are not driven by outliers because all four remained significant (p < 0.05) when we used a least-square fitting procedure that minimizes the effects of outliers (Robustfit in MATLAB; see Results for details).
To ensure that these results were not dominated by outliers, we used a linear regression analysis with a modified least-squares algorithm that is much less sensitive to outliers (Robustfit in MATLAB 6.5). This algorithm gives lower weight to points that do not fit well. Using this approach we found significant relationships between the BOLD signal and the illusion in all four regions presented in Figure 3: Figure 3A (x = −39, y = −42, z = 72): p < 0.001 (one-tailed); t = 3.54, df = 15; Figure 3B (x = −48, y = −33, z = 54): p < 0.025 (one-tailed); t = 2.45, df = 15; Figure 3C (x = −54, y = −30, z = 57): p < 0.05 (one-tailed); t = 1.864, df = 15; Figure 3D (x = −45, y = −39, z = 60): p < 0.05 (one-tailed); t = 1.94, df = 15. This means that the regression was not driven by outliers.
Finally, no activity was observed in the intraparietal or postcentral sulci when the subjects felt illusory wrist flexion when their hands were not touching the body (i.e., no waist-shrinking illusion; tendon free − skin free;
p < 0.001 uncorrected) As in the previous experiments of Naito et al. [19,21,33], the illusory hand movements in the present study activated the bilateral primary motor cortices, dorsal premotor cortices, supplementary motor areas, right inferior parietal cortex, right inferior frontal cortex, and bilateral cerebellum (p < 0.001 uncorrected; some of these regions are shown in Figure S4).
Discussion
Taken together, our results show that neural activity in the parietal cortex reflected the illusory sensation that the size and shape of the waist were changing. This illusion is elicited when the hands are in contact with the waist and the tendons of both hands are vibrated. Thus in this situation the brain receives conflicting sensory information from the vibrated wrists and the contact surfaces between the hands and waist. The input from the vibrated wrist muscles signals to the brain that the hands are flexing, whereas the tactile signals from the palms remained stable, signalling that the hands were in contact with the waist and hips. This conflict is resolved by recalibrating the size and shape of the waist and hips, so that it feels as if the waist/hip region is shrinking as the hands are bending inwards. We found activity in the cortices lining the left postcentral and left intraparietal sulci reflecting the shrinking-waist illusion. Furthermore, there was a linear relationship between the level of activity in these parietal areas and strength of the illusion across subjects. In other words, the subjects that reported the strongest illusion in the psychophysical test session also displayed the strongest parietal activity. The activity in the cortices lining the postcentral and anterior intraparietal sulci probably reflects the neuronal computations associated with the recalibration of the size and shape of the waist. Thus, these parietal areas are likely to be important for the construction of the body image.
The parietal activity can not be explained in terms of the sensory stimulation or the different postures of the arms and hands. The effects of vibrating the skin and the wrist muscles and changing the arm postures were matched in the factorial design ([tendon contact – skin contact] – [tendon free – skin free]), as were the effects related to the kinaesthetic illusions of wrist movement. Furthermore, illusory wrist movements do not activate the parietal areas in question (p > 0.05 uncorrected), but rather activate other areas such as the primary motor cortex [19–21] (see also Figure S4).
The activity associated with the waist-size illusion lies in the postcentral sulcus at its junction with the intraparietal sulcus and in the most anterior part of the intraparietal sulcus. This activity is located in the border region between somatosensory area 2 and the intraparietal sulcus (http://www.bic.mni.mcgill.ca/cytoarchitectonics/), but probably anterior to AIP (the anterior intraparietal area) [34]. The sulcal cortex anterior to AIP responds to somatic stimulation [34]. In the monkey brain, somatosensory area 2, and area 5 which lies posterior to area 2, are considered to be higher-order somatosensory areas [32,35,36]. Cells in these areas are active when different limbs and other parts of the body are touched or moved, i.e., they have complex receptive fields that include several body parts [28,32,35,36]. For example, some cells discharge when the hand, arm, or torso are touched [28], and many cells have bilateral receptive fields [37]. Such cells are not found in the primary somatosensory areas 3a, 3b, or 1. Thus, the neuronal populations in the cortices lining the postcentral sulcus and anterior part of the intraparietal cortex have the capacity to integrate tactile and proprioceptive information from different body parts. Because the shrinking-body illusion depends on the integration and interpretation of somatosensory inputs from different body parts, the postcentral and intraparietal activity could reflect this integration and recalibration process. This result supports the general hypothesis of hierarchical processing in the somatosensory system [32] and extends this principle to the representation of the body image. Afferent inputs from skin, joints, and muscles primarily reach the primary somatosensory cortex [24–26,38,39] and the primary motor cortex [19–21,27,40–42]. From these somatotopically organized primary representations (areas 4, 3a, 3b, and 1), somatic signals from different body parts converge onto higher-order somatosensory regions where the neuronal computations critical for the recalibration of body-part size may be performed
In our previous studies [21,43], we found activity in the right inferior parietal cortex (supramarginal cortex) when the subjects experienced illusory movements of the right hand, left hand, or both hands. In the present study we also observed activity in the right supramarginal cortex both during the illusion of both hands bending (tendon free– rest; p < 0.001 uncorrected) and during the illusion that the hands were bending and the waist was shrinking (tendon contact– rest; p < 0.001 uncorrected). Thus, the inferior parietal activation was eliminated when we examined the interaction term to look for activity specifically associated with the shrinking-waist illusion. This means that the right inferior posterior parietal cortex does not seem to differentiate between the different types of kinaesthetic illusions, and its exact role in body perception remains unclear [43]. This is consistent with a variety of body-image disturbances that can be seen after lesions involving the inferior parietal cortex [44–47]. However, though the lesions probably included the inferior parietal cortex, they are typically very large, thus also including the superior parietal cortex.
Our findings also imply a functional–anatomical dissociation between the central representations of limb movement and perceived changes in the size and shape of body parts. Perception of passive limb movement engage frontal motor areas and parietal areas that are different [19,21,26,43,48] from those associated with changes in waist size (see also Figure S4). The reason for this dissociation probably relates to how the information is derived. Although limb movement can be represented by the analysis of afferent somatic input from a single limb in primary sensorimotor representations, the derivation of information about the relative size of body parts probably requires the integration of information from different body segments in higher-order areas.
The parietal activity can be associated with the illusory sensation that the size of the waist is changing, rather than with a complex kinaesthetic illusion involving two body parts more generally. Two points support this. First, we observed a significant relationship between the reported degree of illusory waist shrinking and the level of parietal activity (see the linear regression analysis and Figure 3). Second, in a previous imaging study [21] we studied effects of tendon vibration of the one wrist while both hands were mutually in contact palm to palm. This elicited an illusion that both hands were bending but without changes in body size [21,49]. Although this illusion also critically depends on the integration of somatosensory signals from different body parts (two limbs) and changes in position and orientation of the hand, we did not observe any activations in the parietal areas reported in the present study (p > 0.001 uncorrected using a sensitive fixed-effect analysis). Thus, the parietal activity is probably related to the illusory feeling that the size of the waist is changing.
We do not say that this activity is specific to waist shrinking as opposed to expansion. We predict the same activity for waist expansion because the BOLD signal cannot distinguish between the directions of movement. Likewise it is an open question as to what extent the location of the parietal activation we detected would depend on the body part that underwent the size-changes. Because the somatotopical organization of the posterior parietal cortex is coarse with extensive overlaps of the somatic receptive fields of different body parts, we would only predict small changes in the location of the activation peaks within the same parietal area when people feel that other body parts are shrinking.
It is important to clarify that our results cannot be explained by passive transduction of vibration from the vibration site to the waist and the abdomen. We know from the psychophysical experiments that passive spread from hands to abdomen does not cause any body-image illusions (e.g., during the skin contact condition). Moreover the spread of vibration was too weak to activate the posterior parietal cortex, as evident from the lack of activity in this area in the contrast skin contact – rest. Further, when we examined the interaction term, we were protected from effects related to passive spread because it occurred in both tendon contact and skin contact. The amount of passive spread in these conditions should be similar given that the vibrator is only moved 3–4 cm on the hands. Finally, passive transduction of vibration can never explain the correlation we observed between subjective ratings of the waist-shrinking illusion and the parietal activity.
In summary, we have shown that higher-order somatosensory areas in the junction between the postcentral and intraparietal sulci (probably areas 2/5) are involved in perceived changes in the size and shape of the waist. We suggest that the underlying mechanism is that these areas compute the relative size and shape of body parts by integrating multiple somatic signals from different body parts. Our finding is important because it provides direct neurophysiological evidence that the parietal cortex is involved in the construction of the body image.
Materials and Methods
Prescanning Psychophysical Test
We tested 24 blindfolded potential healthy subjects on the “shrinking-waist illusion” [4] in a separate experiment before the brain scans. All subjects were right handed and had given their informed consent. The local ethical committee had approved the study. We tested the same four stimulation conditions we later used in the brain scan (tendon contact, skin contact, tendon free, and skin free) as described in the section below on scanning. Each condition lasted 30 s and was repeated three times in a pseudo-randomized order. Seven subjects reported that they did not reliably feel the shrinking illusion (during tendon contact), and they were not scanned because the aim of the present study was to identify the neural correlates of perceived body-size changes.
In the condition in which the hands were in contact, the subjects were requested to indicate the onset of the illusion that the waist was shrinking by making a verbal response (saying “now”). An experimenter timed this response using a stopwatch. After the 30-s period of vibration of both wrists, the subjects were first asked “What did you feel?” and we noted the response. Then we asked them to select one picture out of six different body configurations that best corresponded to their experiences during the stimulation. The relevant picture showed the illusion of waist shrinking; the control pictures showed (1) waist enlargement, (2) hands moving into the waist but no waist shrinking, (3) longer arms, (4) shrinking torso and head, and (5) no changes in body image or hand movement. By these questions we could confirm that the participants experienced the shrinking-waist illusion. The subjects that felt the illusion were also asked to rate the vividness and continuance of the illusion on an analogue scale from 0 to 9. The vividness was defined as how realistic the illusion was when it was experienced (9 being “absolutely realistic”). The continuance score reflected the persistence of the illusion (equivalent to the percentage of time that the illusion was experienced). We also quantified the degree of perceived change in waist size. Directly after the 30-s vibration period, the subjects were asked to display the maximum perceived displacement of the wrists by holding the hands just above the body and flexing the wrists. We measured the angle of illusory wrist displacement and the distance between hands. These measurements reflect the degree of waist shrinkage because the subjects felt as if the waist was shrinking as much as the hands were bending inwards.
During these tests, we simultaneously recorded EMGs during the tendon contactand tendon free conditions (the ECR and the FCR of both forearms). We used a pair of 8-mm diameter Ag/AgCl electrodes (NT-211U; Nihon kohden, Tokyo, Japan) and an amplifier (AB-610J; Nihon kohden) for the digital registration and analysis of the muscle signals (PowerLab/16SP; ADInstruments, Sydney, Australia). We then calculated the integrated EMG (iEMG) to quantify the muscle activity during the different conditions.
Brain Scanning: Experimental Design
On the basis of the results from the initial psychophysical testing described above, 17 subjects (four female, age 20 to 35 y; mean = 24 ± 3.2 y) were selected to take part in the fMRI experiment. Whilst the scanning was being performed, the blindfolded subjects rested comfortably in a supine position on the bed in the MRI scanner. We used two non-magnetic vibrators that were driven by constant air pressure provided by two air-compressors (Umihira Ltd, ILLUSOR, Kyoto, Japan). The frequency was approximately 110 Hz (amplitude: ± 3.5 mm) and the skin surface vibrated was approximately 1 cm2. Two experimenters in the scanner room manually operated the vibrators by applying them to the skin with a light pressure. To provide the two experimenters with synchronized instructions about the conditions and the onset and offset of the vibration, computer-generated visual cues were projected onto the white surface of the scanner (the blindfolded participants could not see this visual information). The participants were instructed to relax during the scans and not to make any movements.
There were four experimental conditions and two resting baselines. In tendon contact, both of the subject's hands were attached to the lateral sides of the waist and legs, palm to body. We vibrated the skin surface over the left and right tendons of ECR muscles (the muscle that extends the wrist). This stimulation causes a kinaesthetic illusion that both wrists are passively flexing and that the waist and upper parts of the legs are shrinking (see Figure 1A). In skin contact,the subjects had their hands in contact with the body (exactly as in tendon contact) but we vibrated the skin over the left and right processes styloideus ulnae, i.e., the skin beside the tendon. In this condition the subjects felt no illusion [21]. In the tendon freeand skin freeconditions, the hands did not touch the body but were positioned in a semi-pronated position so that the palms were towards the lateral sides of the body but not touching it (10-cm distance). In tendon free, we vibrated the tendon of the right and left ECR muscles, which caused the subjects to experience an illusory flexion of both wrists (but no change in body size; see Figure 1B). In skin free, we vibrated the skin beside the tendon over the processes styloideus ulnae, and the participants felt no illusion. Finally, in rest contactand rest free,we did not apply any vibratory stimuli and the subjects had their hands either in contact with the body (rest contact) or not in contact with the body (rest free). To reveal the activity that reflected the shrinking-body illusion, we examined the interaction between site of vibration and hand position using a 2 × 2 factorial design ([tendon contact – skin contact] – [tendon free – skin free]; see also Figure 1). The rationale of this design is that the interaction term reveals activity that reflects the shrinking body illusion and that cannot be attributed to the effects of vibrating the muscle tendon and hand position, i.e., to the sum of the main effects.
A complete experiment consisted of six experimental runs, each lasting 5 min and 36 s. In three runs, the hands lay freely beside the body without touching it and the arms were supported. In these runs we tested the three conditions tendon free, skin free, and rest free. In the three other experimental runs, the palms of the hands were in direct contact with the lateral sides of the body. A strap was used to attach the hands to the body, allowing the subjects to completely relax their arms. In these runs we collected data for the conditions tendon contact, skin contact, and rest contact. To eliminate time effects, the two types (free or contact) of experimental runs were performed in an alternating order that was counterbalanced across subjects.
Each condition lasted for 30 s. The vibration conditions were repeated three times in each run, and the rest condition was performed five times. During the runs we always had rest conditions before and after each vibration condition, and we alternated between tendon and skin vibration.
Acquisition and Analysis of Functional Imaging Data
The functional imaging was conducted by using a Siemens Allegra 3.0T scanner (Erlangen, Germany) to acquire gradient echo T2*-weighted echo-planar images with BOLD contrast as an index of local increases in synaptic activity [50]. The image parameters used were: matrix size = 64 by 64, voxel size = 3 mm by 3 mm, echo time (TE) = 40 ms, and repetition time (TR) = 3,000 ms. A functional image volume comprised 48 slices of 3-mm thickness which ensured that the whole brain was within the field of view. For each of the six experimental runs (see above), we collected 112 image volumes, with one volume being collected every 3 s. A high-resolution T1-weighted structural image was also collected. The fMRI data was analysed using the Statistical Parametric Mapping Software [51] (SPM2, http://www.fil.ion.ucl.ac.uk/spm; Wellcome Department of Imaging Neuroscience, London, United Kingdom). The images were realigned to correct for head movements, co-registered with each subject's anatomical MRI, and transformed to the standard anatomical format. Thus, all coordinates refer to the standard space of the Montreal Neurological Institute (MNI) The functional images were spatially smoothed with a 10-mm full width at half maximum (FWHM) isotropic Gaussian kernel, and smoothed in time by a 4-s FWHM Gaussian kernel.
For each individual subject, we fitted a linear regression model (general linear model) to the data. Each condition was modelled with a boxcar function delayed by 4 s and convoluted with the standard SPM2 hemodynamic response function. Because we knew from the psychophysical test before the scans that the illusion of body shrinking started after 3.3 ± 2.0 s (see Results), we omitted the first 4 s of all conditions by defining these periods as conditions of no interest in the model. We defined linear contrasts in the general linear model to test our hypothesis. The result from this analysis was the estimated BOLD signals for this contrast from each of the 17 subjects (contrast images). To accommodate inter-subject variability, the contrast images from all subjects were entered into a random-effect group analysis (second-level analysis). One-sample t-tests were used (16 df). We used the threshold of p < 0.001 uncorrected in the whole brain. Because we had a priori anatomical hypothesis that the somatosensory section of the parietal cortex would be active, we also used a small volume correction in this region. We defined regions of interest using spheres of 20-mm radius around the most significant peaks of activity observed in the bilateral parietal cortex in the main-effect contrast of all four vibration conditions versus rest (x = −51, y = −42, z = 51, and x = 51, y = −42, z = 48). This main-effect contrast identifies somatosensory areas and can be used to define regions of interest because it is orthogonal to the interaction effect, i.e., statistically independent. The interaction effect in the left parietal cortex corresponded to p < 0.05 corrected. We do not report areas that did not show an increase in activity relative to rest (p < 0.01 uncorrected). In these cases the interaction effect was caused by a deactivation in the control stimulation conditions rather than an increase related to the shrinking-waist illusion.
Finally, to investigate the relationship between the strength of the illusion and the neural activity, we used linear regression analyses. Because we wanted to corroborate the results from the interaction analysis that had revealed activation in the parietal cortex, we restricted this analysis to the left and right parietal cortices. For each subject, we related the activity obtained during the shrinking-body illusion (interaction term) to the mean illusory displacement of the wrist in the tendon contact condition as measured in the tests conducted outside the scanner. This approach is valid because we knew from pilot experiments and our previous experiments that kinaesthetic illusions are consistent across test sessions within the same subject and that there are substantial differences in illusion strength between subjects [19]. First, we searched for active areas within the parietal lobes using the SPM2 regression model. This allowed us to identify the parietal region that showed the most significant relationship between illusion strength and fMRI activity. Second, we examined the relationship in exactly those two coordinates that corresponded to the peaks of the activations in the interaction analysis (see above).
The anatomical localization of the activations was related to the major sulci and gyri [52], distinguishable on a mean MRI generated from the standardized anatomical MRIs from the 17 subjects.
Supporting Information
Figure S1 Quantification of the Illusions of Waist Shrinking and Waist Expansion
The perceived changes in wrist angle when the hands were in contact palm to waist and the wrist flexor or extensor muscles were vibrated are shown. The illusory changes in the wrist angle (top row), in the distance between the hands (middle row), and the vividness of the illusion (lower row) are shown in the graphs. Error bars denote standard errors of means.
(2.7 MB TIF).
Click here for additional data file.
Figure S2 Quantification the Illusions of Head Shrinking and Head Expansion
The perceived changes in wrist angle when the subjects put their right or left hand on top of their head and we vibrated the tendon of either the wrist flexor or extensor muscles are shown. The illusory changes in the wrist angle (top row), in the hand position (middle row), and vividness of the illusions (lower row) are shown. Error bars denote standard errors of means.
(851 KB TIF).
Click here for additional data file.
Figure S3 Analysis of EMG Data
We compared the iEMG from the right and left ECR and the FCR muscles during the two experimental conditions in which we vibrated the muscle tendons (tendon contact; tendon free). Error bars indicate SD. The iEMGs were calculated for a vibration period of 30 s for each subject (n = 17 subjects who participated in fMRI study). There was no difference in muscle activity between the two conditions (paired t-tests; p > 0.05).
(2.8 MB TIF).
Click here for additional data file.
Figure S4 Activation Maps Associated with Feeling Illusory Hand Flexion and Waist Shrinking or Just Illusory Flexion of the Hands without Waist Shrinking
Activation maps associated with feeling illusory hand flexion and waist shrinking are shown in the right column (tendon attached – skin attached); activation maps associated with the illusory flexion of the hands without waist shrinking are shown in the left column (tendon free – skin free). Note that the left intraparietal cortex (yellow circle) was activated only when subjects felt waist shrinking (top left) and not when they only felt that their wrists were bending (top right; compare with Figure 2). Further, activity associated with illusory wrist movements is seen bilaterally in the premotor cortex (top row), primary motor cortex (top row), and inferior parietal cortex (lower row).
(7.1 MB TIF).
Click here for additional data file.
Protocol S1 Psychophysical Experiment
(31 KB DOC).
Click here for additional data file.
We thank Mr. Nobuhiro Hagura, Mr. Toshihiro Hashimoto, Ms. Sara Bengtsson, and Ms. Kumiko Yoshimura. HHE was supported by a post-doctoral stipend from the Human Frontier Science Program. REP was supported by the Wellcome Trust. This study was supported by the 21st Century COE Program (D-2 to Kyoto University), Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. HHE and EN conceived and designed the experiments. HHE, TK, NS, and EN performed the experiments. HHE and TK analyzed the data. HHE, REP, and EN wrote the paper.
Citation: Ehrsson HH, Kito T, Sadato N, Passingham RE, Naito E (2005) Neural substrate of body size: Illusory feeling of shrinking of the waist. PLoS Biol 3(12): e412.
Abbreviations
BOLDblood oxygenation level-dependent
ECRextensor carpi radialis
EMGelectromyogram
FCRflexor carpi radialis
fMRIfunctional magnetic resonance imaging
iEMGintegrated electromyogram
SDstandard deviation
==== Refs
References
Head H Holmes G Sensory disturbances from cerebral lesions Brain 1911 34 102 254
Gandevia SC Rowell LB Shepherd JT Kinesthesia: Roles for afferent signals and motor commands Handbook on integration of motor, circulatory, respiratory and metabolic control during exercise 1996 Bethesda (Maryland) American Physiological Society 128 172
Craske B Kenny FT Keith D Modifying an underlying component of perceived arm length: Adaptation of tactile location induced by spatial discordance J Exp Psychol Hum Percept Perform 1984 10 307 317 6232347
Lackner JR Some proprioceptive influences on the perceptual representation of body shape and orientation Brain 1988 111 281 297 3378137
Ramachandran VS Hirstein W Gallagher S Shear J Three laws of qualia: What neurology tells us about the biological functions of consciousness, qualia, and the self Models of the self 1999 Thorverton (United Kingdom) Imprint Academic 83 111
Taylor-Clarke M Jacobsen P Haggard P Keeping the world a constant size: Object constancy in human touch Nat Neurosci 2004 7 219 220 14966526
Graziano MS Botvinick MM Prinz W Hommel B How the brain represents the body: Insights from neurophysiology and psychology Common mechanisms in perception and action. Volume XIX, Attention and performance 2001 Oxford Oxford University Press 136 157
Gandevia SC Phegan CM Perceptual distortions of the human body image produced by local anaesthesia, pain and cutaneous stimulation J Physiol 1999 514 609 616 9852339
Paqueron X Leguen M Rosenthal D Coriat P Willer JC The phenomenology of body image distortions induced by regional anaesthesia Brain 2003 126 702 712 12566290
Critchley M The parietal lobes 1953 London Edward Arnold 480
Todd J The syndrome of Alice in Wonderland Can Med Assoc J 1955 73 701 704 13304769
Sveinbjornsdottir S Duncan JS Parietal and occipital lobe epilepsy: A review Epilepsia 1993 34 493 521 8504783
Salanova V Andermann F Rasmussen T Olivier A Quesney LF Parietal lobe epilepsy. Clinical manifestations and outcome in 82 patients treated surgically between 1929 and 1988 Brain 1995 118 607 627 7600082
Lippman CW Certain hallucinations peculiar to migraine J Nerv Ment Dis 1952 116 346 351 12991095
Kew J Wright A Halligan PW Somesthetic aura: The experience of “Alice in Wonderland.” Lancet 1998 351 1934 9654271
Robinson D Podoll K Macrosomatognosia and microsomatognosia in migraine art Acta Neurol Scand 2000 101 413 416 10877160
Goodwin GM McCloskey DI Matthews PB Proprioceptive illusions induced by muscle vibration: contribution by muscle spindles to perception? Science 1972 175 1382 1384 4258209
Goodwin GM McCloskey DI Matthews PB The contribution of muscle afferents to kinaesthesia shown by vibration induced illusions of movement and by the effects of paralysing joint afferents Brain 1972 95 705 748 4265060
Naito E Ehrsson HH Geyer S Zilles K Roland PE Illusory arm movements activate cortical motor areas: A positron emission tomography study J Neurosci 1999 19 6134 6144 10407049
Naito E Ehrsson HH Kinesthetic illusion of wrist movement activates motor-related areas Neuroreport 2001 12 3805 3809 11726799
Naito E Roland PE Ehrsson HH I feel my hand moving: A new role of the primary motor cortex in somatic perception of limb movement Neuron 2002 36 979 988 12467600
Roll JP Vedel JP Kinaesthetic role of muscle afferents in man, studied by tendon vibration and microneurography Exp Brain Res 1982 47 177 190 6214420
Roll JP Vedel JP Ribot E Alteration of proprioceptive messages induced by tendon vibration in man: A microneurographic study Exp Brain Res 1989 76 213 222 2753103
Phillips CG Powell TP Wiesendanger M Projection from low-threshold muscle afferents of hand and forearm to area 3a of baboon's cortex J Physiol 1971 217 419 446 5097607
Iwamura Y Tanaka M Sakamoto M Hikosaka O Rostrocaudal gradients in the neuronal receptive field complexity in the finger region of the alert monkey's postcentral gyrus Exp Brain Res 1993 92 360 368 8454001
Romaiguere P Anton JL Roth M Casini L Roll JP Motor and parietal cortical areas both underlie kinaesthesia Brain Res Cogn Brain Res 2003 16 74 82 12589891
Colebatch JG Sayer RJ Porter R White OB Responses of monkey precentral neurones to passive movements and phasic muscle stretch: Relevance to man Electroencephalogr Clin Neurophysiol 1990 75 44 55 1688773
Taoka M Toda T Iwamura Y Representation of the midline trunk, bilateral arms, and shoulders in the monkey postcentral somatosensory cortex Exp Brain Res 1998 123 315 322 9860270
Graziano MS Cooke DF Taylor CS Coding the location of the arm by sight Science 2000 290 1782 1786 11099420
Iriki A Tanaka M Iwamura Y Coding of modified body schema during tool use by macaque postcentral neurones Neuroreport 1996 7 2325 2330 8951846
Obayashi S Tanaka M Iriki A Subjective image of invisible hand coded by monkey intraparietal neurons Neuroreport 11 3499 3505
Iwamura Y Hierarchical somatosensory processing Curr Opin Neurobiol 1998 8 522 528 9751655
Naito E Kochiyama T Kitada R Nakamura S Matsumura M Internally simulated movement sensations during motor imagery activate cortical motor areas and the cerebellum J Neurosci 2002 22 3683 3691 11978844
Culham JC Visually guided grasping produces fMRI activation in dorsal but not ventral stream brain areas Exp Brain Res 2003 153 180 189 12961051
Duffy FH Burchfiel JL Somatosensory system: Organizational hierarchy from single units in monkey area 5 Science 1971 172 273 275 4994137
Sakata H Takaoka Y Kawarasaki A Shibutani H Somatosensory properties of neurons in the superior parietal cortex (area 5) of the rhesus monkey Brain Res 1973 64 85 102 4360893
Iwamura Y Iriki A Tanaka M Bilateral hand representation in the postcentral somatosensory cortex Nature 1994 369 554 556 8202155
Mountcastle VB Powell TP Central nervous mechanisms subserving position sense and kinesthesis Bull Johns Hopkins Hosp 1959 105 173 200 14424737
Iwamura Y Tanaka M Sakamoto M Hikosaka O Functional subdivisions representing different finger regions in area 3 of the first somatosensory cortex of the conscious monkey Exp Brain Res 1983 51 315 326
Fetz EE Finocchio DV Baker MA Soso MJ Sensory and motor responses of precentral cortex cells during comparable passive and active joint movements J Neurophysiol 1980 43 1070 1089 6766994
Lemon RN Hanby JA Porter R Relationship between the activity of precentral neurones during active and passive movements in conscious monkeys Proc R Soc Lond B Biol Sci 1976 194 341 373 11492
Lemon RN Functional properties of monkey motor cortex neurones receiving afferent input from the hand and fingers J Physiol 1981 311 497 519 7264981
Naito E Dominance of the right hemisphere and role of area 2 in human kinesthesia J Neurophysiol 2005 93 1020 1034 15385595
Goldenberg G Imitating gestures and manipulating a mannikin–the representation of the human body in ideomotor apraxia Neuropsychologia 1995 33 63 72 7731541
Felician O Ceccaldi M Didic M Thinus-Blanc C Poncet M Pointing to body parts: A double dissociation study Neuropsychologia 2003 41 1307 1316 12757904
Berlucchi G Aglioti S The body in the brain: Neural bases of corporeal awareness Trends Neurosci 1997 20 560 564 9416668
Devinsky O D'Esposito M Devinsky O D'Esposito M The right hemisphere, interhemispheric communication, and consciousness Neurology of cognitive and behavioral disorders 2004 Oxford Oxford University Press 68 102
Radovanovic S Korotkov A Ljubisavljevic M Lyskov E Thunberg J Comparison of brain activity during different types of proprioceptive inputs: A positron emission tomography study Exp Brain Res 2002 143 276 285 11889505
Lackner JR Taublieb AB Reciprocal interactions between the position sense representations of the two forearms J Neurosci 1983 3 2280 2285 6631480
Logothetis NK Pauls J Augath M Trinath T Oeltermann A Neurophysiological investigation of the basis of the fMRI signal Nature 2001 412 150 157 11449264
Frackowiak RSJ Friston KJ Frith C Dolan R Price CJ Human brain function, 2nd ed 2004 San Diego (California) Academic Press 1,144 p
Duvernoy HM The human brain: Surface, blood supply, and three-dimensional sectional anatomy 1991 New York Springer-Verlag 499
|
16336049
|
PMC1287503
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e412
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030412
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1629689310.1371/journal.pbio.0030415Research ArticleAnimal BehaviorGenetics/Genomics/Gene TherapyNeuroscienceDiabetes/Endocrinology/MetabolismGeriatricsMus (Mouse)Effects of Hypothalamic Neurodegeneration on Energy Balance Neurodegeneration and ObesityXu Allison Wanting
1
Kaelin Christopher B
1
Morton Gregory J
3
Ogimoto Kayoko
3
Stanhope Kimber
4
Graham James
4
Baskin Denis G
5
Havel Peter
4
Schwartz Michael W
3
Barsh Gregory S [email protected]
1
2
1Department of Genetics, Stanford University School of Medicine, Stanford, California, United States of America,2Department of Pediatrics, Stanford University School of Medicine, Stanford, California, United States of America,3Department of Medicine, Harborview Medical Center, University of Washington, Seattle, Washington, United States of America,4Department of Nutrition, University of California, Davis, California, United States of America,5VA Puget Sound Health Care System and University of Washington School of Medicine, Seattle, Washington, United States of AmericaSaper Cliff Academic EditorBeth Israel Deaconess Medical CenterUnited States of America12 2005 29 11 2005 29 11 2005 3 12 e41522 7 2005 6 10 2005 Copyright: © 2005 Xu 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.
What Makes Mice Fat? How the Brain Controls Energy Balance
Normal aging in humans and rodents is accompanied by a progressive increase in adiposity. To investigate the role of hypothalamic neuronal circuits in this process, we used a Cre-lox strategy to create mice with specific and progressive degeneration of hypothalamic neurons that express agouti-related protein (Agrp) or proopiomelanocortin (Pomc), neuropeptides that promote positive or negative energy balance, respectively, through their opposing effects on melanocortin receptor signaling. In previous studies, Pomc mutant mice became obese, but Agrp mutant mice were surprisingly normal, suggesting potential compensation by neuronal circuits or genetic redundancy. Here we find that Pomc-ablation mice develop obesity similar to that described for Pomc knockout mice, but also exhibit defects in compensatory hyperphagia similar to what occurs during normal aging. Agrp-ablation female mice exhibit reduced adiposity with normal compensatory hyperphagia, while animals ablated for both Pomc and Agrp neurons exhibit an additive interaction phenotype. These findings provide new insight into the roles of hypothalamic neurons in energy balance regulation, and provide a model for understanding defects in human energy balance associated with neurodegeneration and aging.
Mice are genetically engineered for the progressive degeneration of hypothalamic neurons containing the neuropeptides Pomc and Agrp. Their phenotypes suggest a model for energy balance changes associated with aging.
==== Body
Introduction
The mechanisms by which warm-blooded animals balance energy intake with energy expenditure have come under increasing scrutiny in recent years. Important components of this process are the changes associated with aging. In humans, rats, and mice, young adults exhibit a remarkable ability to regulate fuel stores so that the net change in adiposity over long periods of time is a small fraction of energy consumed. As part of the normal aging process, however, a slow but progressive increase in adiposity occurs throughout most of adulthood (reviewed in [1,2]). In both humans and rodents, these changes occur independently of environmental variation, and are likely to be caused by the progressive impairment of mechanisms that normally control energy homeostasis [3–5].
Studies in rodents suggest the underlying causes of age-related changes in energy homeostasis are likely to involve the central melanocortin system [5–7], in which two subgroups of neurons in the arcuate nucleus of the hypothalamus, those marked by the expression of proopiomelanocortin (Pomc) or agouti-related protein (Agrp), sense total body adiposity by sampling circulating indicators such as leptin and insulin. Pomc and Agrp neurons integrate and relay this information to downstream central nervous system effectors that act to balance energy stores via changes in both energy intake and expenditure, with activation of Pomc neurons promoting negative energy balance, and activation of Agrp neurons promoting positive energy balance [8,9].
To some extent, experimental mouse genetic manipulations support this view: deletion of the Pomc gene [10] or transgenic overexpression of Agrp [11] causes hyperphagia and obesity. The results of knockout experiments, however, have been more difficult to interpret. Agrp-deficient mice have been reported to have no detectable defects in energy balance [12], although treatment of adult mice with RNAi against Agrp causes a transient decrease in energy expenditure [13].
An important consideration in interpreting the results of these experiments is the relationship between the action of Pomc or Agrp and that of the Pomc- or Agrp-expressing neurons. The view of energy homeostasis as a system based on neuronal subtypes rather than neuropeptides per se derives in part from the observations that Pomc and Agrp are each coexpressed with other neuropeptides that have similar effects, Cart and neuropeptide Y (Npy), respectively [8], and in part from the observations that the different neuronal subtypes exhibit reciprocal patterns of electrical and/or signaling activity [14,15]. From this perspective, perturbing Pomc or Agrp gene expression may not fully reveal the function of the Pomc or Agrp neuron, respectively.
To better understand the role of Pomc and Agrp neurons in energy balance, and to model how altered activity of these neurons might account for age-related changes in adiposity, we constructed animals in which Pomc- or Agrp-expressing cells were progressively ablated by deleting the mitochondrial transcription factor A (Tfam) gene using a Cre-lox strategy. Tfam activity is required for mitochondrial genome transcription and replication, and Tfam-deficient cells exhibit progressive loss of cellular respiration and eventual cell death [16,17]. Here we report that animals in which the Tfam gene is removed from Pomc- or Agrp-expressing cells exhibit adult-onset defects in energy balance that provide insight into the normal role of specific hypothalamic neuronal subtypes. Our findings demonstrate that central nervous system regulation of body weight by Agrp neurons involves compensation by redundant genes, and provide a model for understanding defects in energy balance associated with aging and neurodegeneration.
Results
Construction and Characterization of Pomc- and Agrp-Neuronal Ablation Animals
In previous experiments using a Cre recombinase transgene controlled by Pomc or Agrp regulatory elements, we demonstrated that expression of the transgene was specific for the different neuronal subtypes in adult animals [9]. To further characterize the timing and pattern of transgene expression, we examined different tissues from mice carrying the PomcCre or AgrpCre transgene together with the lacZ reporter allele Gt(Rosa)26Sortm1Sor (R26R) [18].
In the arcuate nucleus of the hypothalamus, activation of R26R by the PomcCre transgene became apparent by embryonic day 17.5; activation of R26R by the AgrpCre transgene, however, was not detectable until 2–3 wk of age (Figure 1). These results correspond to what has been described previously for normal development of the melanocortinergic system, with the hypothalamic Pomc system becoming established during fetal brain development, but the Agrp system developing after birth [19,20]. Activation of R26R by the PomcCre transgene also occurs in the nucleus tract solitarius of the hindbrain, and the anterior and intermediate lobe of the pituitary in neonatal animals (unpublished data), which are areas in which Pomc is normally expressed. The early onset of Pomc expression in the arcuate nucleus suggests that the Pomc neurons may play a role in the development of the hypothalamus. Expression of Pomc has also been reported in the skin, but we did not detect activation of the R26R transgene outside the brain or pituitary gland.
Figure 1 Pattern of R26R Activation in Mice Carrying the PomcCre or AgrpCre Transgene
Whole brains from Tg.PomcCre/+; R26R/+ or Tg.AgrCre/+; R26R/+ mice of the indicated age were fixed and stained with Xgal as described in [9]. Photographs show the ventral brain surface, and coronal sections (not shown) indicate that Xgal-stained neurons lie in the arcuate nucleus of the hypothalamus. As assessed by the extent of Xgal staining, Cre-induced recombination in Pomc neurons is complete by embryonic day 17.5; however, Cre-induced recombination in Agrp neurons is just beginning at 2 wk of age (arrows) and is not complete until 3–4 wk of age.
The loxP-flanked Tfam allele, Tfamflox, was originally constructed by Larsson and colleagues [16] and has been used for a number of cell type–specific ablation studies [17,21,22]. Animals with global Tfam deficiency die in early embryogenesis, but animals with Tfam deficiency in the neocortex are normal until 5–6 mo of age, when animals die suddenly with massive neuronal cell death. In these latter studies, Cre recombinase was driven by regulatory elements from the calcium/calmodulin-dependent protein kinase II alpha gene (Camk2a), which is maximally expressed at 1 mo of age, and the 4- to 5-mo delay between Tfam deficiency and neuronal cell death is thought to be due to the gradual depletion of mitochondrial gene products from postmitotic cells [17]. Thus, we anticipated that ablation of Pomc or Agrp neurons due to loss of Tfam would be complete by 6 mo of age.
We intercrossed Tfam flox
/flox and Tg.PomcCre/+; R26R/+ or Tg.AgrpCre/+; R26R/+ animals, and identified groups of F2 animals as mutant ablation (Tfam flox
/flox; Tg.Cre/?), Tfam control (Tfam flox/+; Tg.Cre/?), or Cre recombinase control (Tfam flox/flox; +/+). Consistent with our earlier characterization of the PomcCre and AgrpCre transgenic lines, we observed no phenotypic differences in the different types of control animals; in the studies described below, results for Tfam control and Cre recombinase control animals are pooled. We also generated animals deficient for both Pomc and Agrp cells (Tfam flox
/flox; Tg.PomcCre/+; Tg.AgrpCre/+), referred to below as “double-ablation” mice.
We examined the pattern and extent of immmunohistochemical staining for α–melanocyte-stimulating hormone (α-MSH) and Agrp in Tfam mutant and control mice at multiple time points. At 2 and 4 mo of age, α-MSH and Agrp neurons in Tfam mutant mice appeared no different from control mice (Figure 2). However, at 7 mo of age, we found that in Pomc-specific Tfam mutant mice, immunohistochemical staining for α-MSH was dramatically reduced compared to control animals, but immunohistochemical staining for Agrp was unaffected (Figure 3). Similarly, in Agrp-specific Tfam mutant mice, immunohistochemical staining for Agrp was dramatically reduced compared to control animals, but immunohistochemical staining for α-MSH was unaffected (Figure 3). In both cases, there was no detectable gliosis or effect on brain architecture, indicating that the ablation was highly specific.
Figure 2 Immunostaining of α-MSH or Agrp in Control and Tfam Mutant Animals of Different Ages
Hypothalami from control (Tfam flox/+; Tg.PomcCre/+ or Tfam flox/flox; +/+), Pomc-specific Tfam mutant (Tfam flox/flox; Tg.PomcCre/+), and Agrp-specific Tfam mutant (Tfam flox/flox; Tg.AgrpCre/+) animals were harvested at the indicated ages and immunostained for α-MSH or Agrp as indicated.
IR, immmunoreactivity.
Figure 3 Effect of Pomc-Specific Tfam Deficiency on Pomc and Agrp Immunostaining
Each column of panels shows coronal sections from a 7-mo-old control (Tfam flox/+; Tg.PomcCre/+ or Tfam flox/flox; +/+), Pomc-specific Tfam mutant (Tfam flox/flox; Tg.PomcCre/+), and Agrp-specific Tfam mutant (Tfam flox/flox; Tg.AgrpCre/+) on the left, middle, and right, respectively. Each section was stained for α-MSH or Agrp as indicated. The three lower panels show coronal sections stained to reveal cell nuclei (DAPI) and immunostained for GFAP, and indicate that ablation of Pomc or Agrp neurons does not grossly disturb tissue architecture, cell number, or stimulate astrocytosis. The sections shown are representative of three different animals that were examined.
IR, immmunoreactivity.
As an additional measure of ablation, we examined Xgal staining of hypothalamic sections from experimental and control mice that carried the R26R reporter gene, since, for example, loss of Agrp neurons from Tg.AgrpCre/+; Tfam flox
/flox ; R26R/+ mice should be reflected in a reduced number of Xgal-positive neurons. For both Pomc-specific and Agrp-specific Tfam mutant mice, very few Xgal-positive cells were detectable; furthermore, mRNA in situ hybridization revealed a dramatic reduction in the number of Agrp-expressing cells (Figure 4). Thus, loss of Tfam induced by a PomcCre or AgrpCre transgene gives rise to postnatal loss of Pomc or Agrp neurons that is specific and extensive.
Figure 4 Expression of the R26R Cre Reporter Gene in Pomc- or Agrp-Specific Tfam Mutant Animals
Animals carrying the PomcCre or AgrpCre transgene together with the lacZ reporter allele for Cre recombination Gt(Rosa)26Sortm1Sor (R26R) were intercrossed with Tfam flox/flox animals. F2 progeny of the indicated genotype were sacrificed at 7 mo of age, and hypothalamic coronal sections stained for Xgal. In this situation, Xgal staining serves as an autonomous histologic marker, and loss of Xgal staining in Tfam mutant animals is therefore secondary to cell death. Sections are representative of three to four per genotype group that were examined. The panel on the lower right shows the number of Agrp-expressing neurons in the control and Agrp-Tfam mutants as determined by fluorescence in situ hybridization to Agrp mRNA.
**, p ≤ 0.01. Error bars = standard error of the mean.
Effects of Pomc and/or Agrp Ablation on Energy Balance and Neuroendocrine Axis
Pomc-ablation mice exhibited a progressive adult-onset obesity in which increased weight gain compared to controls was first detectable at 4 mo of age, and the magnitude of the effect was greater in females (∼30% increase in body weight) than in males (∼15% increase in body weight) (Figure 5A and 5B). Agrp-ablation mice exhibited little difference in body weight compared to control animals, although at every time point, the mean body weight for mutant animals was slightly less than control animals for both males (Figure 5C) and females (Figure 5D). A more significant role for Agrp neurons in total body weight was uncovered by examining double-ablation mice; animals deficient for both neuronal subtypes weigh more than control animals but less than Pomc single-ablation mice in both males (Figure 5E) and females (Figure 5F).
Figure 5 Effect of Pomc- and/or Agrp-Specific Tfam Mutations on Body Weight
(A–D) show longitudinal measurements of body weight in animals of the indicated genotype and sex. (E) and (F) show comparisons of single (Pomc-specific) and double (Pomc- and Agrp-specific) Tfam deficiency for 6- and 9-mo-old animals.
*, p ≤ 0.05; **, p ≤ 0.01. Error bars = standard error of the mean.
Using dual energy X-ray absorptiometry (DEXA) or MRI to measure adipose versus nonadipose tissue, we observed that in 7- to 10-mo-old animals, loss of Pomc-expressing cells caused an increase in both compartments, with a relative effect that was slightly greater on fat than on lean body mass (Figure 6A and 6B). Agrp ablation caused a small but significant decrease of fat mass in females but not in males (Figure 6D). We also measured rates of food intake and energy expenditure in Pomc-ablation animals, and observed an effect on both components of the energy balance equation. Pomc-ablation mice consumed approximately 10% more than control animals, and exhibited rates of O2 consumption approximately 10% less than control animals (Figure 7).
Figure 6 Lean Body Composition of Animals with Pomc- or Agrp-Specific Tfam Deficiency
Lean body mass of animals of the indicated genotype was determined as described in Materials and Methods. For Pomc-specific Tfam mutant animals (A), numbers of animals were: control n = 10, mutant n = 4 (male); and control n = 6, mutant n = 4 (female). For Agrp-specific Tfam mutant animals (C), numbers of animals were control n = 4, mutant n = 4 (male); and control n = 6, mutant n = 8 (female). (B) and (D) show comparisons of control and mutant animals of the indicated genotype and sex for fat masses of 7- to 10-mo-old animals. For Pomc-specific Tfam mutant animals (B), numbers of animals were control n = 10, mutant n = 4 (male); and control n = 6, mutant n = 4 (female). For Agrp-specific Tfam mutant animals (D), numbers of animals were control n = 4, mutant n = 4 (male); and control n = 6, mutant n = 8 (female).
*, p ≤ 0.05; **, p ≤ 0.01. Error bars = standard error of the mean.
Figure 7 Food Intake and Energy Expenditure in Animals with Pomc- or Agrp-Specific Tfam Deficiency
(A and B) Daily food intake was measured for 7 d as described in Materials and Methods. For (A), numbers of animals used (control and Pomc-Tfam mutant) were 6-mo-old male (n = 13, n = 7); 6-mo-old female (n = 10, n = 6); > 8-mo-old male (n = 10, n = 6); and > 8-mo-old female (n= 13, n = 6).
(C, D) O2 consumption was measured over a 24-h period as described in Materials and Methods; panels illustrate results from a single control and a single Pomc-Tfam mutant animal before and after (10–12 d) corticosterone replacement; these results are representative of four control and four mutant animals that were examined.
(E) Daily food intake was significantly increased after corticosterone replacement in Pomc-Tfam mutant animals (n = 6) but not in control (n = 9) animals.
(F) Measurement of total serum T4 levels in control (n = 7) and Pomc-Tfam mutant (n = 5) animals.
*, p ≤ 0.05; **, p ≤ 0.01. Error bars = standard error of the mean.
The Pomc-ablation mice lost Pomc-expressing pituitary cells in addition to Pomc neurons, and therefore had chronic ACTH insufficiency with reduced basal corticosterone levels (unpublished data). Although both Pomc knockout and the Pomc-ablation mice showed corticosterone deficiency due to lack of ACTH, the Pomc-ablation mice developed adrenal glands (unpublished data), whereas adrenal glands in the Pomc knockout mice were largely absent [10]. We used long-term corticosterone replacement to investigate whether or not hypoadrenalism contributed to abnormal energy balance in Pomc-ablation mice. Corticosterone pellets (10 mg; 21 d) were implanted into both Pomc-ablation and control mice, and measurements of food intake and energy expenditure were made after 10–12 d. In control animals, corticosterone replacement had no effect on energy expenditure (Figure 7C and 7D) or food intake (Figure 7E). In Pomc-ablation animals, corticosterone replacement had no effect on energy expenditure (Figure 7C and 7D) but stimulated food intake by approximately 20% over a level that was already elevated relative to control (Figure 7E). These observations suggest that ablation of Pomc neurons is sufficient to reduce energy expenditure and increase food intake; in fact, concomitant adrenocortical deficiency in Pomc-ablation mice would, if anything, mitigate the obesity phenotype.
α-MSH is also thought to stimulate the thyroid axis via thyroid-releasing hormone, but measurements of total T4 in Pomc-ablation mice were not significantly different from those of control mice (Figure 7F).
Impaired Compensatory Refeeding in Pomc-Ablation Mice
Because Pomc neurons serve as primary hypothalamic sensors for adiposity signals, we suspected that hyperphagia and reduced energy expenditure in Pomc-ablation mice were caused by an underlying abnormality in the hypothalamic circuitry that normally maintains peripheral energy stores. To test this idea directly, we examined the ability of 6-mo-old mutant and control animals to increase food intake acutely following a period of food deprivation. After a 48-h fast, animals normally compensate with a transient hyperphagia sufficient to restore adipose depots to prefasting levels, a phenomenon typically associated with a 2-fold increase in food intake over 24 h. We found that Pomc-ablation mice exhibit an impaired compensatory refeeding response, consuming approximately 25% less than expected (Figure 8A); conversely, Agrp-ablation mice exhibited a normal refeeding response (Figure 8B). These effects also were apparent in weight gain after refeeding; control and Agrp-ablation mice were able to recover their body weight precisely within 24 h after fasting, while Pomc-ablation mice were not (Figure 8C and 8D).
Figure 8 Compensatory Refeeding and Neuropeptide mRNA levels in Pomc-Specific and Agrp-Specific Tfam Deficiency
(A–D) 24-h compensatory refeeding after a 48-h fast was measured as described in Materials and Methods; data are shown either as (A and B) the ratio of food consumed over 24 h (refeeding) to normal daily food intake (averaged over 7 h prior to food deprivation), or as (C and D) percentage of weight recovery after 24 h of refeeding. For the Pomc-Tfam experiment (A and C), number of animals used was n = 23 (control) and n = 13 (Pomc-Tfam mutant); for the Agrp-Tfam experiment (B and D), number of animals used was n = 21 (control) and n = 15 (Agrp-Tfam).
(E) The same refeeding defects were observed in Pomc-Tfam mutants after corticosterone replacement (control, n = 9; Pomc-Tfam mutant, n = 6).
(F) Expression of Pomc, Agrp, and Npy in Pomc-Tfam mutants. Mice were fasted for 48 h 10 d after implanting corticosterone pellets, and expression of Pomc, Agrp, and Npy in the hypothalamus was measured by semi-quantitative RT-PCR. Values are shown as relative levels compared to free-fed controls. Numbers of animals used were n = 5 (control fed); n = 5 (control fasted); n = 3 (mutant fed); and n = 3 (mutant fasted).
*, p ≤ 0.05; **, p ≤ 0.01. Error bars = standard error of the mean.
To investigate the extent to which adrenocortical impairment might contribute to the refeeding abnormality in Pomc-ablation mice, we carried out the experiment before and after implanting corticosterone replacement pellets. Although this approach cannot completely mimic the normal circadian and metabolic fluctuations of glucocorticoids, we found that during the period of corticosterone replacement, blood corticosterone levels were increased after fasting in both controls and mutants (unpublished data), indicating that the adrenal glands of the Pomc-ablation mice were able to mount a stress response. Even though corticosterone replacement stimulated food intake (Figure 7E) and weight gain (unpublished data), mutant mice were (Figure 8E) unable to compensate for body weight lost by fasting.
Agrp and Npy mRNA levels are normally upregulated by fasting, and have been suggested to contribute to compensatory refeeding. Surprisingly, we found that Pomc-ablation mice exhibited a normal response to fasting in terms of their ability to upregulate Agrp and Npy mRNA (Figure 8F). These findings suggest that impaired compensatory refeeding in Pomc-ablation mice is caused not by inability to upregulate orexigenic neuropeptides, but instead by the absence of Pomc neurons. Thus, Pomc-ablation mice represent an interesting paradox in which mutant animals are hyperphagic at baseline, yet are unable to increase their food intake to compensate for food deprivation.
Discussion
Although physical or chemical lesioning studies carried out 60 years ago first drew attention to the ventromedial area and the lateral hypothalamic area of the hypothalamus as sites important for regulating energy balance [23,24], functional heterogeneity within the arcuate nucleus of the hypothalamus was not recognized until recently. Our results address functional heterogeneity in a specific way by ablating subpopulations of neurons with a Cre-lox strategy. Distinguishing between the action of a neuropeptide and the action of a neuropeptide-specific cell is also apparent by considering previous studies of Agrp and Pomc knockout mice. In contrast to Agrp knockout or Agrp; Npy double-knockout mice, which are normal [12], Agrp-ablation mice show reduced adiposity. This reduction in adiposity is most apparent in the Agrp; Pomc double-ablation mice, an observation that is reminiscent of what was previously described for the interaction of Npy and Lepob [25]. This observation points to an important role for other neuropeptides and neurotransmitters besides Agrp and Npy. One such candidate is gamma-aminobutyric acid, which has been shown to be released from Npy neurons and inhibit the activity of Pomc neurons [14]. The relatively mild weight phenotype of the Agrp-ablation mice could be due to a small fraction of the Agrp neurons that escape ablation, differences in genetic background, potential compensation developed during the course of progressive cell loss, or a combination of these mechanisms. One such potential compensatory mechanism is neuronal plasticity and rewiring of the feeding circuits as described by Horvath and colleagues [26].
Nonetheless, our findings and similar results recently published by Bewick et al. [27], Gropp et al. [28], and Luquet et al. [29] demonstrate that ablation of Agrp neurons yields a phenotype that is more dramatic than a knockout of Agrp (or combined ablation of Agrp and Npy [12]). Furthermore, acute ablation of the Agrp neurons over a time frame of days [28,29] yields a more severe phenotype than progressive ablation of the Agrp neurons over a time frame of months due to neurodegeneration conferred by Tfam deficiency. These observations highlight the potential for neuronal plasticity and developmental compensation [26], and have implications for both experimental strategies and therapeutic approaches to energy homeostasis. By contrast, Pomc knockout mice [10,30,31] exhibit a phenotype that is more dramatic than ablation of Pomc neurons, with obesity starting at 4–8 wk of age and an eventual ∼50% increase in body weight, whereas Pomc-ablation mice develop obesity starting at 4 mo of age with an eventual 15%–30% increase in body weight (Figure 5). The later time of onset can be attributed simply to the difference between germline inheritance of a knockout allele and somatic loss of mitochondrial function in specific cells, but several factors could account for the difference in magnitude, including an inbred versus a mixed genetic background, a small fraction of Pomc neurons that escape ablation, or additional products released by Pomc neurons that counter the effects of Pomc.
The surprising result that the Pomc-ablation mice failed to defend their body weight normally after food deprivation is somewhat paradoxical given their chronic hyperphagia and obesity. We propose that normal compensatory refeeding requires either downregulation of Pomc neuronal activity, decreased Pomc release, or both, and that normal downregulation is abolished in Pomc-ablation mice. This suggestion is consistent with the observation that widespread transgenic overexpression of Pomc also gives rise to impaired compensatory refeeding thought to be due to disruptions of Pomc regulation; however, these animals exhibit a chronic state of negative rather than positive energy balance [32].
Electrophysiologic studies have suggested a melanocortinergic neuronal circuit in which Agrp neurons control the activity of Pomc neurons [14,33], but the reciprocal question—do Pomc neurons control the activity of Agrp neurons—has not been addressed. However, Agrp and Npy mRNA levels responded normally to food deprivation in our Pomc-ablation mice, which indicates that a potential pathway from Pomc neurons to Agrp neurons, should one exist, does not act by controlling production of Agrp or Npy.
Our strategy for genetic ablation of Pomc and Agrp neurons is based on a prior study in which Cre-mediated loss of Tfam driven by the Camk2a promoter caused widespread neuronal cell death due to loss of mitochondrial ATP production [17]. In some ways, this approach is similar to that used by Sakurai and colleagues [34] for the ablation of hypocretin/orexin (Hcrt)-expressing neurons, in which transgenic expression of a polyglutamine-containing protein caused Hcrt-expressing neurons to die in the immediate postnatal period. In fact, while this manuscript was under review, Bewick et al. [27] reported that transgenic animals carrying a polyglutamine-containing protein driven by the Agrp promoter exhibit reduced adiposity and body weight, similar to our observations. The later onset of neuronal cell death induced by Tfam deficiency is likely explained by a gradual depletion of mitochondrial gene products from postmitotic cells, and therefore provides a model for pathophysiologic changes related to the mitochondrial theory of aging, in which progressive impairment of mitochondrial function associated with oxidative damage is thought to lead to a vicious circle that culminates in neuronal apoptosis [35–37]. In a recent test of this model, Trifunovic et al. [38] found that animals genetically engineered to accumulate mitochondrial mutations developed several features of premature aging. While not specifically tested by Trifunovic et al. [38], our results predict that these animals should exhibit a defective energy balance with an impairment in compensatory hyperphagia.
From this perspective, the defective energy balance that developed in our Pomc-ablation or double-ablation mice at 4 mo of age provides a useful model for age-related obesity. In both rodents and humans, adiposity progressively increases throughout most of adult life. Even in elderly individuals, where total body weight (and subcutaneous fat) declines toward the end of life, visceral adiposity continues to increase, pointing to a fundamental age-related defect in energy balance [1,3,5]. Elderly humans exhibit defects in both the normal reduction in energy expenditure caused by overfeeding [39], and the normal increase in perceived hunger caused by underfeeding [40]. Defects in compensatory hyperphagia similar to what we observed in our Pomc-ablation or double-ablation mice have also been described in studies of young (3-mo-old), middle-aged (12-mo-old), and elderly (24-mo-old) rats from Matsumoto and colleagues [6,41]. Notably, impaired energy homeostasis in the rat model is associated with altered expression of Pomc [6] and Npy [41]. Thus, the neuronal-ablation mice we describe here represent a model for premature dysfunction of hypothalamic circuits implicated in the normal aging process.
Materials and Methods
Animal studies
Animals carrying the TfamloxP allele [16,17] were kindly provided by N. Larsson (Karolinska Institute, Stockholm, Sweden) on a mixed C57BL/6J and 129 background. Animals carrying the R26R LacZ reporter allele were obtained from P. Soriano (University of Washington, Seattle, Washington, United States), and are maintained in our laboratory as homozygotes on a mixed background that includes contributions from FVB/N, C57BL/6J, and 129 strains. The Tg.PomcCre and Tg.AgrpCre lines are maintained as heterozygotes on the same mixed background; their construction and characterization has been described previously [9]. We generated a total of 128 animals for the Pomc-ablation mice (44 mutants, 62 Tfam controls, and 22 Cre recombinase controls), 133 animals for the Agrp-ablation mice (44 mutants, 54 Tfam controls, and 35 Cre recombinase controls), and 42 double-ablation mice. To help control for differences in litter size and modifier genes, results were analyzed by two-way ANOVA in which sibship was included as a variable. Although the genetic background was heterogeneous, there was less variation (both environmental and genetic) within litters than between litters.
For feeding studies, animals were singly housed for 2 wk, then food intake was measured for a consecutive 6–7 d; results are shown as an average over 24 h. To measure compensatory refeeding, food was removed 3 h before the onset of the dark cycle, mice were fasted for 48 h, then food was added back and food intake measured for the next 24 h.
Metabolic monitoring was carried out with a system (AccuScan Instruments, Columbus, Ohio, United States) located in the UC Davis Department of Nutrition animal facility. The system consists of an O2 analyzer, CO2 analyzer, and PhysioScan analyzer, which monitors vertical and horizontal movement via light beam interruption. A flow controller/channelyzer allows for flow rate adjustments and sequential channeling of airflow from a reference line and the four animal chambers through the CO2 and O2 analyzers. The flow rate for these experiments was 0.5 l/min. The dimensions of the plexiglas chambers were 12 in × 8 in × 8 in. The Integra ME software includes O2 and CO2 analyzer calibration, data collection, and analysis programs. Reported and calculated values include O2 consumption, CO2 production, RQ, heat production (energy expenditure), total ambulatory movement, and total rest time. Animals were acclimated to the chambers for 20 h immediately prior to a 24-h data collection period. Each animal received its normal diet and water while in the chamber.
Body fat mass and lean mass were determined either by dual X-ray absorptiometry (Pomc-ablation mice) or by MRI (Agrp-ablation mice). In the former case, mice were anesthetized by intraperitoneal injection of 2.5% avertin, and body composition was measured using a PIXImus2 instrument. In the latter case, conscious animals were analyzed with an MRI whole-body composition analyzer (Echo Medical Systems, Houston, Texas, United States) at the Rodent Energy Metabolism core at the University of Washington.
For studies involving corticosterone replacement, a 10-mg 21-d release corticosterone pellet (Innovative Research of America, Sarasota, Florida, United States) was implanted subcutaneously on the lateral side of the neck of both ablation and control animals. This dose of corticosterone is sufficient for physiologic replacement and has been shown to prevent adrenalectomy-induced anorexia [42].
Molecular biology
Measurements of mRNA levels were carried out by quantitative RT-PCR on RNA extracted from dissected hypothalamic tissue. Total RNA for each hypothalamus was quantified by spectrophotometry after purification using TRIzol reagent (Invitrogen, Carlsbad, California, United States) and an RNeasy mini kit (Qiagen, Valencia, California, United States). 250 ng each total RNA sample was reverse-transcribed, then PCR-amplified using a Lightcycler (Roche Applied Science, Basel, Switzerland) and SYBR green to measure relative cDNA levels. Agrp primers were TGCTACTGCCGCTTCTTCAA and CTTTGCCCAAACAACATCCA; Npy primers were TAACAAGCGAATGGGGCTGT and ATCTGGCCATGTCCTCTGCT; Pomc primers were AGGCCTGACACGTGGAAGAT and AGGCACCAGCTCCACACAT; and Actb primers were CTGCGTTTTACACCCTTTCTTTG and gccatgccaatgttgtctcttat. Efficiency for each primer set was estimated from standard curves made with serial cDNA dilutions.
Xgal staining and immunohistochemistry for α-MSH or AGRP was carried out as described previously [9]. For glial fibrillary acidic protein (GFAP) immunofluorescence, a polyclonal anti-GFAP antibody (Dako, Glostrup, Denmark) was used at 1:1500 dilution according to the protocol described in [9]. Goat–anti-rabbit Alexa488 (Molecular Probes, Eugene, Oregon, United States; 1:200) was used for secondary antibody detection. Sections were mounted using Vectashield with DAPI (Vector Laboratories, Burlingame, California, United States). Fluorescence images were captured using a black-and-white digital camera (AxioCam MRm) attached to a Zeiss Axioplan2 imaging system (Zeiss, Oberkochen, Germany), and later pseudocolored to red (GFAP) and blue (DAPI) and superimposed.
For fluorescence in situ hybridization to Agrp mRNA, brains were frozen on dry ice, and 14-μm cryostat sections mounted on slides using RNAse-free conditions. Labeling and hybridization were carried out as described previously [43].
We thank N. G. Larsson for providing the TfamloxP
/loxP mice. This work was supported by grants from the National Institutes of Health to GSB (DK48506 and DK68384) and MWS (DK68384).
Competing interests. The authors have declared that no competing interests exist.
Author contributions. AWX, MWS, and GSB conceived and designed the experiments. AWX, CBK, GJM, KO, KS, JG, and DGB performed the experiments. AWX, CBK, GJM, and DGB analyzed the data. CBK, PH, and MWS contributed reagents/materials/analysis tools. AWX, MWS, and GSB wrote the paper.
Citation: Xu AW, Kaelin CB, Morton GJ, Ogimoto K, Stanhope K, et al. (2005) Effects of hypothalamic neurodegeneration on energy balance. PLoS Biol 3(12): e415.
Abbreviations
α-MSHα–melanocyte-stimulating hormone
Agrpagouti-related protein
GFAPglial fibrillary acidic protein
Npyneuropeptide Y
Pomcproopiomelanocortin
==== Refs
References
Elia M Obesity in the elderly Obes Res 2001 9 244S 248S 11707549
Roberts SB Dallal GE Effects of age on energy balance Am J Clin Nutr 1998 68 975S 979S 9771882
Roberts SB Energy regulation and aging: Recent findings and their implications Nutr Rev 2000 58 91 97 10835898
Roberts SB A review of age-related changes in energy regulation and suggested mechanisms Mech Ageing Dev 2000 116 157 167 10996016
Mizuno T Shu IW Makimura H Mobbs C Obesity over the life course Sci Aging Knowledge Environ 2004 re4
Gruenewald DA Matsumoto AM Age-related decrease in proopiomelanocortin gene expression in the arcuate nucleus of the male rat brain Neurobiol Aging 1991 12 113 121 1711159
Miller MM Joshi D Billiar RB Nelson JF Loss during aging of beta-endorphinergic neurons in the hypothalamus of female C57BL/6J mice Neurobiol Aging 1991 12 239 244 1876229
Schwartz MW Woods SC Porte D Seeley RJ Baskin DG Central nervous system control of food intake Nature 2000 404 661 671 10766253
Xu AW Kaelin CB Takeda K Akira S Schwartz MW PI3K integrates the action of insulin and leptin on hypothalamic neurons J Clin Invest 2005 115 951 958 15761497
Yaswen L Diehl N Brennan MB Hochgeschwender U Obesity in the mouse model of pro-opiomelanocortin deficiency responds to peripheral melanocortin Nat Med 1999 5 1066 1070 10470087
Ollmann MM Wilson BD Yang YK Kerns JA Chen Y Antagonism of central melanocortin receptors in vitro and in vivo by agouti-related protein Science 1997 278 135 138 9311920
Qian S Chen H Weingarth D Trumbauer ME Novi DE Neither agouti-related protein nor neuropeptide Y is critically required for the regulation of energy homeostasis in mice Mol Cell Biol 2002 22 5027 5035 12077332
Makimura H Mizuno TM Mastaitis JW Agami R Mobbs CV Reducing hypothalamic AGRP by RNA interference increases metabolic rate and decreases body weight without influencing food intake BMC Neurosci 2002 3 18 12423556
Cowley MA Smart JL Rubinstein M Cerdan MG Diano S Leptin activates anorexigenic POMC neurons through a neural network in the arcuate nucleus Nature 2001 411 480 484 11373681
Van Den Top M Lee K Whyment AD Blanks AM Spanswick D Orexigen-sensitive NPY/AgRP pacemaker neurons in the hypothalamic arcuate nucleus Nat Neurosci 2004 7 493 494 15097991
Larsson NG Wang J Wilhelmsson H Oldfors A Rustin P Mitochondrial transcription factor A is necessary for mtDNA maintenance and embryogenesis in mice Nat Genet 1998 18 231 236 9500544
Sorensen L Ekstrand M Silva JP Lindqvist E Xu B Late-onset corticohippocampal neurodepletion attributable to catastrophic failure of oxidative phosphorylation in MILON mice J Neurosci 2001 21 8082 8090 11588181
Soriano P Generalized lacZ expression with the ROSA26 Cre reporter strain Nat Genet 1999 21 70 71 9916792
Pelletier G Regulation of proopiomelanocortin gene expression in rat brain and pituitary as studied by in situ hybridization Ann N Y Acad Sci 1993 680 246 259 8512219
Nilsson I Johansen JE Schalling M Hokfelt T Fetissov SO Maturation of the hypothalamic arcuate agouti-related protein system during postnatal development in the mouse Brain Res Dev Brain Res 2005 155 147 154 15804403
Silva JP Kohler M Graff C Oldfors A Magnuson MA Impaired insulin secretion and beta-cell loss in tissue-specific knockout mice with mitochondrial diabetes Nat Genet 2000 26 336 340 11062475
Wang J Wilhelmsson H Graff C Li H Oldfors A Dilated cardiomyopathy and atrioventricular conduction blocks induced by heart-specific inactivation of mitochondrial DNA gene expression Nat Genet 1999 21 133 137 9916807
Hetherington AW Ranson SW Hypothalamic lesions and adiposity in the rat Anat Rec 1940 78 149 172
Elmquist JK Elias CF Saper CB From lesions to leptin: Hypothalamic control of food intake and body weight Neuron 1999 22 221 232 10069329
Erickson JC Hollopeter G Palmiter RD Attenuation of the obesity syndrome of ob/ob mice by the loss of neuropeptide Y Science 1996 274 1704 1707 8939859
Pinto S Roseberry AG Liu H Diano S Shanabrough M Rapid rewiring of arcuate nucleus feeding circuits by leptin Science 2004 304 110 115 15064421
Bewick GA Gardiner JV Dhillo WS Kent AS White NE Postembryonic ablation of AgRP neurons in mice leads to a lean, hypophagic phenotype FASEB J 2005 19 1680 1682 16099943
Gropp E Shanabrough M Borok E Xu AW Janoschek R Agouti-related peptide-expressing neurons are mandatory for feeding Nat Neurosci 2005 8 1289 1291 16158063
Luquet S Perez FA Hnasko TS Palmiter RD NPY/AgRP neurons are essential for feeding in adult mice but can be ablated in neonates Science 2005 In press
Smart JL Low MJ Lack of proopiomelanocortin peptides results in obesity and defective adrenal function but normal melanocyte pigmentation in the murine C57BL/6 genetic background Ann N Y Acad Sci 2003 994 202 210 12851317
Challis BG Coll AP Yeo GS Pinnock SB Dickson SL Mice lacking pro-opiomelanocortin are sensitive to high-fat feeding but respond normally to the acute anorectic effects of peptide-YY(3–36) Proc Natl Acad Sci U S A 2004 101 4695 4700 15070780
Mizuno TM Kelley KA Pasinetti GM Roberts JL Mobbs CV Transgenic neuronal expression of proopiomelanocortin attenuates hyperphagic response to fasting and reverses metabolic impairments in leptin-deficient obese mice Diabetes 2003 52 2675 2683 14578285
Roseberry AG Liu H Jackson AC Cai X Friedman JM Neuropeptide Y-mediated inhibition of proopiomelanocortin neurons in the arcuate nucleus shows enhanced desensitization in ob/ob mice Neuron 2004 41 711 722 15003171
Hara J Beuckmann CT Nambu T Willie JT Chemelli RM Genetic ablation of orexin neurons in mice results in narcolepsy, hypophagia, and obesity Neuron 2001 30 345 354 11394998
Wallace DC A mitochondrial paradigm for degenerative diseases and ageing Novartis Found Symp 2001 235 247 263 263 266 discussion 11280029
Dufour E Larsson NG Understanding aging: Revealing order out of chaos Biochim Biophys Acta 2004 1658 122 132 15282183
Balaban RS Nemoto S Finkel T Mitochondria, oxidants, and aging Cell 2005 120 483 495 15734681
Trifunovic A Wredenberg A Falkenberg M Spelbrink JN Rovio AT Premature ageing in mice expressing defective mitochondrial DNA polymerase Nature 2004 429 417 423 15164064
Roberts SB Fuss P Dallal GE Atkinson A Evans WJ Effects of age on energy expenditure and substrate oxidation during experimental overfeeding in healthy men J Gerontol A Biol Sci Med Sci 1996 51 B148 B157 8612099
Moriguti JC Das SK Saltzman E Corrales A McCrory MA Effects of a 6-week hypocaloric diet on changes in body composition, hunger, and subsequent weight regain in healthy young and older adults J Gerontol A Biol Sci Med Sci 2000 55 B580 B587 11129387
Gruenewald DA Marck BT Matsumoto AM Fasting-induced increases in food intake and neuropeptide Y gene expression are attenuated in aging male brown Norway rats Endocrinology 1996 137 4460 4467 8828508
Jacobson L Glucocorticoid replacement, but not corticotropin-releasing hormone deficiency, prevents adrenalectomy-induced anorexia in mice Endocrinology 1999 140 310 317 9886840
Hahn TM Breininger JF Baskin DG Schwartz MW Coexpression of Agrp and NPY in fasting-activated hypothalamic neurons Nat Neurosci 1998 1 271 272 10195157
|
16296893
|
PMC1287504
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e415
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030415
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1629298210.1371/journal.pbio.0030416Research ArticleCancer BiologyCell BiologyGenetics/Genomics/Gene TherapyMolecular Biology/Structural BiologyOncologyHomo (Human)Securin Is Not Required for Chromosomal Stability in Human Cells Securin and Chromosomal StabilityPfleghaar Katrin
1
2
Heubes Simone
3
Cox Jürgen
4
Stemmann Olaf
3
Speicher Michael R [email protected]
1
2
1Institute of Human Genetics, Technical University Munich, Munich, Germany,2Institute of Human Genetics, GSF National Research Center for Environment and Health, Neuherberg, Germany,3Department of Molecular Cell Biology, Max Planck Institute of Biochemistry, Martinsried, Germany,4Genedata GmbH, Martinsried, GermanyHawley R. Scott Academic EditorStowers Institute for Medical ResearchUnited States of America12 2005 29 11 2005 29 11 2005 3 12 e41611 8 2005 11 10 2005 Copyright: © 2005 Pfleghaar 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.
An Insecure Role for Securin in Chromosome Segregation
Abnormalities of chromosome number are frequently observed in cancers. The mechanisms regulating chromosome segregation in human cells are therefore of great interest. Recently it has been reported that human cells without an hSecurin gene lose chromosomes at a high frequency. Here we show that, after hSecurin knockout through homologous recombination, chromosome losses are only a short, transient effect. After a few passages hSecurin−/− cells became chromosomally stable and executed mitoses normally. This was unexpected, as the securin loss resulted in a persisting reduction of the sister-separating protease separase and inefficient cleavage of the cohesin subunit Scc1. Our data demonstrate that securin is dispensable for chromosomal stability in human cells. We propose that human cells possess efficient mechanisms to compensate for the loss of genes involved in chromosome segregation.
Speicher et al. show that previously reported chromosome instability and loss due to Securin gene knockout is a transient effect; human cells may have compensatory mechanisms to overcome Securin gene loss.
==== Body
Introduction
A number of factors are involved in ensuring that in dividing cells chromosomes are copied exactly once and then distributed correctly to daughter cells. Chromosome cohesion is established during chromosome replication in S-phase and is mediated by the multisubunit cohesin complex, which forms a giant ring structure possibly encircling sister chromatids [1]. Sister chromatid separation in anaphase depends on the removal of cohesin complexes from chromosomes [2]. In vertebrates removal of cohesin from chromosomes occurs in at least two steps. The “prophase pathway” removes the bulk of cohesin from chromosome arms during prophase and prometaphase [3,4]. By metaphase only minor amounts of cohesin remain on chromosomes, preferentially at centromeres [4]. Centromere-specific factors, such as shugoshin, protect the cohesion between sister centromeres from the prophase pathway [5,6]. At the metaphase-to-anaphase transition, residual cohesion is dissolved by the large cysteine endopeptidase separase, which cleaves the so-called kleisin subunit of cohesin (Scc1/Rad21 in mitosis; Rec8 in meiosis). This cleavage allows sister chromatids to move apart [7,8] and is, in fact, essential for anaphase to occur [9].
For most of the cell cycle, separase activity is inhibited by binding of an inhibitory chaperone called securin [10–12] or by phosphorylation-dependent complex formation with Cdk1 [13,14]. Separase is eventually activated by proteolysis of securin or the cyclin B subunit of Cdk1, which in both cases is mediated by a ubiquitin protein ligase named anaphase promoting complex or cyclosome (APC/C) and its cofactor Cdc20 [15,16].
Thus, securin is a key substrate of the APC/CCDC20 pathway. Though conserved in function, securins from different phyla are highly divergent in sequence [17]. Earlier studies had already implicated securin in functional mechanisms related to cell-cycle control and tumorigenesis [18,19]. To further address securin's function, both copies of the gene encoding hSecurin were inactivated via homologous recombination in the karyotypically stable human colorectal cancer cell line HCT116 [20]. The results indicated that hSecurin is indeed needed for chromosomal stability in human cells, as hSecurin-deficient cells exhibited high rates of chromosome missegregation, similar to those observed in many cancers. Furthermore, the data suggested that hSecurin, through its chaperone activity, plays a crucial role for the proper function of separase, especially for separase-dependent cleavage of the cohesin subunit Scc1 [20]. (Our group contributed to that paper some of the cytogenetic data using fixed cell suspensions provided by C. Lengauer's laboratory.)
However, the important role of hSecurin elucidated in the Jallepalli et al. study [20] contrasts with the results of another investigation, which found mice lacking securin to be viable and apparently normal [21]. Furthermore, only 20% of mouse cells without securin exhibit gains or losses of chromosomes [22].
To resolve this discrepancy, we conducted further studies with the hSecurin−/− cell line. Here, we show that the initial missegregation phenotype was superseded by a regaining of chromosomal stability in only a few passages. The karyotype of chromosomally stable hSecurin−/− cells was indistinguishable from that of the parent cell line with intact securin. Surprisingly, the initially described biochemical defects caused by the lack of securin, i.e., significantly reduced levels of separase and inefficient cleavage of the cohesin subunit Scc1 [20], were still present.
This indicates that securin is not required for faithful chromosome segregation and that alternative mechanisms may compensate for the absence of securin and/or reduced separase levels.
Results
Analysis of Chromosomal Instability in the hSecurin−/− Cell Line at Different Passages
In an initial step, metaphase spreads of the hSecurin−/− cell line were karyotyped by multiplex fluorescence in situ hybridization (M-FISH) at different passages (Figure 1). For passages 2 and 3, we confirmed the loss of numerous chromosomes in the majority of analyzed cells (Figure 1A and 1B). About 60% (12/20) of metaphase spreads showed losses of at least one chromosome. Surprisingly however, the high rate of chromosome losses in the hSecurin−/− cell line had almost vanished by passage 8 (Figure 1C), when chromosome losses were noted in only 10% (2/20) of cells. By passage 12, we observed no chromosome losses (Figure 1D). In the latter two analyses, merely one metaphase spread each had a gain of a single chromosome (Figure 1C and 1D).
Figure 1 hSecurin−/− Cells Regain Chromosomal Stability Quickly after hSecurin Knockout by Homologous Recombination: Summary of M-FISH Analysis of hSecurin−/− Cells at Different Passages
(A–D) Graphic summary of M-FISH data from hSecurin−/− cells at passages 2 (A), 3 (B), 8 (C), and 12 (D). At each passage point 20 or 30 metaphase spreads were painted by M-FISH and analyzed for alterations of chromosome structure and number. Loss of a single copy of a given chromosome is marked in red, loss of both copies is marked in crimson, and gain of a single chromosome is marked in green. Rows indicate the analyzed metaphase spreads (m1 to m30 or m20); columns indicate the chromosome number (1–22 and X).
(E) Graphic representation of the percentages of metaphase spreads with chromosomal copy number aberrations at different passages for the series of experiments shown in (A–D) (blue line) and for a repeat experiment (purple line).
(F) M-FISH karyotype of a passage 12 hSecurin−/− cell, showing that the karyotype is identical to that of the parent cell line HCT116 (for details, see text).
The entire experiment was repeated, and it again showed the same phenomenon, i.e., decreasing chromosome losses in the hSecurin−/− cell line with increasing passage numbers (Figure 1E).
As expected, the parent cell line HCT116 was chromosomally stable and remained stable throughout all analyses (unpublished data).
When we karyotyped the hSecurin−/− cells at passage 12, we found that the karyotype of the cells was identical to the karyotype of the parent cell line HCT116 (Figure 1F; a detailed karyotype description is given in Materials and Methods). Thus, the hSecurin−/− cells showed all known structural rearrangements from the parental cell line HCT116, which we had described before [20,23]. There were no additional, new changes, and none of the known aberrations was lost.
Interphase FISH confirmed that hSecurin−/− cells at passage 12 were indeed chromosomally stable, similarly to the parental cell line HCT116 (Figure 2).
Figure 2 Assessment of Chromosomal Stability in Interphase Nuclei of Parental HCT116 Cells and Chromosomally Stable hSecurin−/− Cells Using Centromere-Specific Probes for Chromosomes 7, 8, 11, and 17
(A and B) Representative interphase FISH images of parental HCT116 (hSecurin+/+) (A) and hSecurin−/− (B) cell nuclei after hybridization of a four-color probe set consisting of centromere probes for chromosomes 7 (Cy5.5; purple), 8 (FITC; green), 11 (Cy5; blue), and 17 (Cy3; yellow). In each nucleus, two signals are visible for each probe.
(C and D) Graphic summary of chromosome gains and losses in parental HCT116 (C) and hSecurin−/− (D) cells. The percentage of signals per nucleus for chromosomes 7, 8, 11, and 17 was determined from 300 cells of each genotype (100 cells each in three separate experiments).
Confirmation that Chromosomally Stable Cell Lines Are Indeed hSecurin−/− Cells
In the next step we confirmed that chromosomally stable hSecurin
−/− cells were indeed hSecurin deficient. We used PCR primer pairs spanning the second and third exons of the securin gene, as described previously [20] (Figure 3A and 3B). In addition, we designed a primer set flanking exon 3 to specifically demonstrate that the hSecurin
−/− cells lack exon 3 of the securin gene (Figure 3A and 3C). Genomic PCR analyses with these primer sets using DNA extracted from chromosomally stable hSecurin
−/− cells demonstrated that the cells had indeed a homozygous deletion of the exon 3 region of the hSecurin gene. Primers for exons 8 and 9 of p53 were used as an additional control (Figure 3B).
Figure 3 Verification that the Chromosomally Stable Cells Indeed Lack Part of hSecurin by Analyses of Genomic DNA from Parental HCT116 Cells (+/+) and Chromosomally Stable hSecurin−/− Cells (−/−)
(A) Transcript structure of the hSecurin gene with its six exons. The lengths of introns and exons are drawn to scale based on the NCBI 35 assembly of the human genome (http://www.ensembl.org). Exons 2 and 3, with the locations of the respective primer pairs, are depicted enlarged.
(B) As a control, PCR analysis was done with primers located in exons 8 and 9 of the p53 gene and resulted in the expected amplification product for both cell lines (lanes 2 and 3). In contrast, PCR with primers PTTG-R6 and PTTG-R1, located in the second and third exon of the hSecurin gene (arrows above exons 2 and 3 in [A]), yielded an amplification product only for the parental HCT116 cells (+/+; lane 5) and not for the chromosomally stable hSecurin−/− cells (−/−; lane 6). Lane 1 shows the 100-bp ladder as a size marker, and lanes 4 and 7 are negative controls for the respective primer pairs.
(C) PCR analyses with primers SecP1l, located in exon 2, and SecP2r, located in intron 3–4 (arrows below exons 2 and 3 in [A]), resulted in amplification products with different sizes (lanes 2 and 3), reflecting the deletion of exon 3. Lane 1 shows the 100-bp ladder; lane 4 is the negative control.
Normal Execution of Anaphase in hSecurin−/− Cells
Previously, it was reported that cells lacking hSecurin grew somewhat more slowly than wild-type cells [20]. In contrast, the growth pattern of chromosomally stable hSecurin−/− cells was indistinguishable from that of the HCT116 parent cell line (unpublished data). Therefore, we performed immunofluorescence experiments to examine the distribution of centromeres during mitosis with cells from passages 12 or higher. hSecurin
+/+ cells and chromosomally stable hSecurin
−/− cells in various stages of mitosis were stained with the CREST (calcinosis-Raynaud's phenomenon-esophageal dismobility-sclerodactyly-telangiectasia syndrome of scleroderma) antibody, which recognizes kinetochore proteins (Figure 4). In addition, we used cyclin B1 as a marker for mitotic stage. Anaphase cells were first identified by virtue of chromosome condensation and lack of cyclin B staining and then scored for unsegregated chromatids remaining at the metaphase plate. In previous experiments, about 30% of hSecurin
−/− cells in anaphase still had paired sister chromatids left behind at the metaphase plate, when most of the other chromosomes had segregated to the poles [20]. When we analyzed a total of 75 cells in three separate experiments in each hSecurin
+/+ (Figure 4A–4G) and hSecurin
−/− (Figure 4H–4N) cell line, we found no differences. In fact, the vast majority of analyzed anaphase cells in each cell line displayed a complete separation of sister chromatids and migration of centromeres to opposite poles (Figure 4O).
Figure 4 Analysis of Defective Sister Chromatid Separation in hSecurin
−/− Cells
hSecurin
+/+ cells and hSecurin
−/− cells were stained with DAPI as a counterstain, a cyclin B1 antibody (green/FITC), which stains cells in the early mitosis but not in anaphase, and a CREST antibody (yellow/Cy3) to visualize kinetochores. In each image are telophase cells showing a complete separation of sister chromatids.
(A–D) Analysis of several cells of the HCT116 parent cell line (hSecurin
+/+). The sequence of images illustrates the DAPI (A), FITC (B), and Cy3 (C) channels, while (D) shows the merged FITC and Cy3 images. One cell is stained with the cyclin B1 antibody; the majority of cells are in prophase (characteristic “double-dot” pattern of paired centromeres). There are two telophase cells.
(E–G) Two telophase hSecurin
+/+ cells with complete separation of sister chromatids. (E) DAPI; (F) Cy3; (G) merged DAPI and Cy3 image.
(H–K) Cells from the hSecurin
−/− cell line. The cell in the upper-right corner is stained with the cyclin B1 antibody; at the bottom is a normal telophase cell with complete separation of sister chromatids. (H) DAPI; (I) FITC; (J) Cy3; (K) merged FITC and Cy3 image.
(L–N) Prophase and telophase hSecurin
−/− cells. The telophase cell demonstrates a complete separation of sister chromatids. (L) DAPI; (M) Cy3; (N) merged DAPI and Cy3 image.
(O) Quantitation of the chromatid separation defect in hSecurin
+/+ and hSecurin
−/− cells. The percentage of anaphase cells with unsegregated sister chromatids at the metaphase plate was determined from 75 cells of each genotype (25 cells each in three separate experiments; the error bars indicate standard deviation).
Furthermore, as in the first paper [20], we did not observe premature sister chromatid separation when cells were exposed to microtubule poisons such as colchecine (unpublished data).
These results indicate that chromosomally stable hSecurin
−/− cells execute anaphase normally, with complete sister chromatid separation.
Cleavage of Separase and Scc1 in hSecurin−/− Cells
During mitosis, separase undergoes proteolytic auto-cleavage resulting in carboxy (C)- and amino (N)-terminal fragments. Previously, it was reported that in hSecurin
−/− cells separase levels and activity were both reduced [20].
We analyzed separase levels in chromosomally stable hSecurin
−/− cells synchronized by nocodazole. Lysates from hSecurin
+/+ and hSecurin
−/− cells were probed with antibodies to separase. For each cell line we detected both the full-length and the cleaved forms of separase (Figure 5A). However, both the full-length and the cleaved forms of separase were consistently 3- to 4-fold weaker in the chromosomally stable hSecurin
−/− cells.
Figure 5 Chromosomally Stable hSecurin
−/− Cells of Passage 12 and Higher Show Reduction in Both the Level and the Activity of Separase
(A) Quantitation of full-length separase and the N-terminal cleavage product in both hSecurin
+/+ and hSecurin
−/− cells. Lysates from nocodazole-arrested cells were analyzed by immunoblotting with an antibody against the N-terminus of separase. The chromosomally stable hSecurin
−/− cells show reduced levels of both the full-length and the cleaved N-terminal form of separase. β-tubulin was used as a loading control.
(B) Separase was immunoprecipitated from nocodazole-arrested hSecurin
+/+ and hSecurin
−/− cells, activated by incubation in Xenopus anaphase extracts, and incubated with 35S-hScc1 for 0, 20, or 90 min before analysis by SDS-PAGE and autoradiography. For these experiments we used four times as many hSecurin
−/− cells as hSecurin
+/+ cells. Note the absence of detectable Scc1 cleavage fragments in the hSecurin
−/− samples.
(C) Separase used for the activity assay in (B) was analyzed by Western blotting before (−) and after (+) exposure to Xenopus anaphase extracts. The hSecurin
+/+ cells clearly demonstrate an increase in self-cleavage of separase upon activation in the extract. Occurrence of auto-cleavage in hSecurin
−/− cells even before incubation in Xenopus extract suggests deregulation of separase, at least under the given conditions of this in vitro experiment.
We reconstituted the cleavage reaction, which dissociates Scc1 from the centromeric regions, by using immunoprecipitated separase complexes that were first incubated with Xenopus egg extracts as a source of mitotic APC/CCDC20. In the case of HCT116 parent cells, incubation of activated separase with in vitro-translated 35S-hScc1 resulted in typical cleavage fragments that were readily detectable after a 20-min incubation (Figure 5B). Four times as much starting cell material was used to purify separase from hSecurin
−/− cells as was used to purify the same amount of separase from wild-type cells (Figure 5C). However, the separase from hSecurin
−/− cells did not display any cleavage activity towards Scc1, even after a 90-min incubation (Figure 5B).
Despite the absence of activity in vitro, separase auto-cleavage products (Figure 5A) demonstrate the presence of at least some separase activity in hSecurin
−/− cells, which, apparently, is sufficient to execute anaphase normally (see above).
Interestingly, immunoprecipitated separase from nocodazole-arrested cells showed a higher degree of self-cleavage in hSecurin
−/− cells compared to that in wild-type cells. This suggests that separase might be partly deregulated in the hSecurin
−/− cells (Figure 5C).
Discussion
We report here that hSecurin
−/− cells are capable of compensating securin loss and rapidly regain chromosomal stability within a few passages. Our findings were unexpected, as chromosomally stable hSecurin
−/− cells continue to have the initially described biochemical defect, i.e., reduction of both the amount and the activity of separase [20; this study].
Our data may explain why mice lacking securin are viable and normal [21]. However, mouse cells without securin have little change in the level of separase [22]. This is in contrast to our observations in human cells, where absence of securin resulted in the aforementioned significant reduction of both separase and its cleavage product. Furthermore, about 20% of mouse embryonic stem cells without securin were aneuploid [22], while the percentage of aneuploid human hSecurin
−/− cells was reduced to background levels similar to those observed in the chromosomally stable parent cell line HCT116. These data indicate that significant differences in separase regulation between human and mouse cells must exist.
The reconstitution of chromosomal stability and complete separation of sister chromatids in hSecurin
−/− cells suggest that significantly lower than normal amounts of separase are sufficient for normal execution of anaphase. This is in agreement with separase at wild-type levels being able to efficiently remove from chromosomes even vastly increased amounts of cohesin [7].
In contrast to budding yeast, human cells lacking hSecurin still manage to arrest sister chromatid separation in the presence of spindle poisons [20; unpublished data]. Therefore, additional mechanisms that regulate the removal of cohesin in human cells must exist. One additional, securin-independent mechanism of separase inhibition involves phosphorylation by Cdk1 and subsequent binding of the kinase [13,22]. As securin and Cdk1 bind separase in a mutually exclusive manner [14], Cdk1 may be capable of compensating for the loss of securin. Indeed, mouse embryonal stem cells lacking both forms of separase regulation suffer from precocious sister chromatid separation under mitotic checkpoint arrest [22]. Another level of control is probably exerted by shugoshin, which prevents removal of centromere-specific cohesin before the onset of anaphase [5].
Our findings demonstrate that deletion of hSecurin has little or no effect on long-term chromosome segregation fidelity in human cells. In fact, our results even raise the possibility that the chromosomal instability (CIN) phenotype observed in early passages of the hSecurin
−/− cells might have been caused by insults during the deletion process rather than by loss of securin per se.
Alternatively, cells might upregulate other control pathways in response to loss of securin, thereby regaining chromosomal stability. Comparative gene expression profiling before and after loss of hSecurin might reveal compensatory changes in the expression of genes involved in anaphase regulation. We therefore compared the transcriptomes of the HCT116 parent and the hSecurin
−/− cell line using the Affymetrix U133A chip. Indeed, significantly different expression levels were found for PLK2 (Polo-like kinase 2), RCC1 (regulator of chromosome condensation 1, also known as chromosome condensation 1 [CHC1]), and SMC6L1 (SMC6 structural maintenance of chromosomes 6-like 1) (unpublished data). However, future experiments will be required to determine the physiological significance of these findings and whether the above proteins might play a currently unknown role in the other two known regulations of sister chromatid separation.
In summary, we have shown that securin is not required for chromosomal stability in human cells. Our results affect current mathematical models of colorectal cancer investigating the role of genetic instability in tumorigenesis [24]. The crucial effect attributed to CIN is acceleration of the mutation rate. However, our data indicated that a CIN-causing mutation may not reach fixation in a given cell compartment, which should therefore change existing assumptions on the evolution of CIN lesions and their growth rates.
Finally, implications of this study extend beyond mechanisms leading to chromosomal instability and affect possible strategies for cancer therapy. It has been suggested that, as stability pathways of tumor cells are defective, cancer cells may be more sensitive to stress-inducing agents and they should be especially susceptible to attack by instability-inducing drugs [25]. However, our results suggest that targeting a particular pathway may not destroy a cell but rather activate alternative pathways. A search and detailed characterization of these alternative pathways will provide further clues to the nature of CIN in human cancers.
Materials and Methods
Cell lines
The colorectal cell line HCT116 was used as parent cell line. This is a chromosomally stable cell line; the karyotype has been described before as 45,X,-Y,der(10)dup(10)(q24q26)t(10;16)(q26;q24), der(16)t(8;16)(q13;p13),der(18)t(17;18)(q21;p11.3) [20,23]. The hSecurin−/− cell line was generated by homologous recombination as described previously [20]. Early passage stocks of both cell lines were generously provided by C. Lengauer and B. Vogelstein (both Johns Hopkins Oncology Center, Baltimore, Maryland, United States).
Culturing of parent and hSecurin knockout cells
HCT116 cells and HCT116 hSecurin
−/− cells were cultured in McCoy's 5A medium (Gibco Invitrogen, Karlsruhe, Germany) supplemented with 10% fetal bovine serum (FBS) (Biochrom, Berlin, Germany), 100 units/ml penicillin, and 0.1 mg/ml streptomycin. Monolayer cultures were grown at 37 °C in a 5% CO2 atmosphere and were split 1:3 twice a week.
M-FISH and interphase FISH
M-FISH [26] was done with 7-fluorochromes as described previously [27]. For interphase FISH we used centromere probes, which were generously provided by M. Rocchi (for detailed information, see http://www.biologia.uniba.it/rmc/index.html). We assembled a four-centromere probe set consisting of centromere-specific probes for chromosomes 7 (PZ7.6B; indirectly labeled with Cy5.5), 8 (PZ8.4; directly labeled with FITC), 11 (PRB11; directly labeled with Cy5), and 17 (PZ17–14; directly labeled with Cy3).
PCR verification of hSecurin knockout
To verify that part of the hSecurin locus was deleted by homologous inactivation, we used the same four PCR primers as previously described [20]: two primers located in exon 2 (PTTG-R6 [AAAATGGAGAACCAGGCACC] and PTTG-gen01 [ACCCGTGTGGTTGCTAAGGA]) and two primers within exon 3 (PTTG-R1 [GGTCCCTTGGTCTTTACAGA] and PTTG-R4 [GTGGGCATCGAACGTTTTG]). These primers define two STS markers of the hSecurin locus that were homozygously deleted in hSecurin−/− cells [20]. In addition, we used the following set of primers to specifically demonstrate that hSecurin−/− cells lack exon 3 of the hSecurin gene: SecP1l (GATGGGCTGAAGCTGGG), which is located in exon 2, and SecP2r (TGCTTGCTAACCTCTATTTCCC), which is within intron 3–4.
As an additional control we used primers for exons 8 and 9 within TP53. The 3′ primer was CATGATTCAGAACCCTGGAG; the 5′ primer was AGGACCTGATTTCCTTACTGC.
Analysis of sister chromatid separation
Cells were grown on coverslips for 24 h in McCoy's 5A medium (Gibco Invitrogen) plus 100 units/ml penicillin and 0.1 mg/ml streptomycin without FBS. Aphidicolin was added to a final concentration of 0.15 μg/ml in McCoy's 5A medium with 10% FBS. After 14 h, the medium was removed, and cells were washed four times with PBS and cultured for 8–10 h in McCoy's 5A medium plus 10% FBS to obtain cells in anaphase.
Cells were fixed in a 4% paraformaldehyde solution, washed three times with PBS/0.2% Tween, and permeabilized in 0.5% TritonX in PBS/Tween for 15 min.
After being blocked with 4% BSA in PBS/Tween, cells were incubated with CREST serum (Euroimmun Corp., Gross Groenau, Germany) in a 1:100 dilution.
After being washed as above, cells were incubated with rabbit anti-human antibody conjugated with Cy3 (Dianova GmbH, Hamburg, Germany) and anti-cyclin B1 monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, California, United States). After being washed as above, cells were counterstained with DAPI and mounted in antifade for fluorescence microscopy.
Separase quantification by immunoblotting
HCT116 wild-type and hSecurin−/− cells were grown in McCoy's medium (10% FBS, 100 units/ml penicillin, 0.1 mg/ml streptomycin) and synchronized by addition of nocodazole (0.2 μg/ml final concentration) for 14 h. Cells from six 75-ml dishes (70% confluent) of each cell line were lysed in 1 ml of 20 mM Tris/HCl (pH 7.7), 100 mM NaCl, 10 mM NaF, 20 mM β-glycerophoshate, 5 mM MgCl2, 0.1% Triton X100, 5% glycerol, 1 μM microcystin-LR, and Complete protease inhibitors (Roche, Basel, Switzerland). After ultracentrifugation at 100,000 g, supernatants were analyzed by Western blotting using an antibody directed against the N-terminus of separase [13]. Signals were quantified by normalizing to β-tubulin (monoclonal antibody obtained from the Developmental Studies Hybridoma Bank, Iowa City, Iowa, United States).
Separase activity assay
HCT116 wild-type (four 75-ml dishes) and hSecurin−/− cells (12 75-ml dishes) were lysed as above, and separase was immunoprecipitated with a rabbit polyclonal antibody raised against the sequence GSDGEDSASGGKTPA of human separase. For each immunoprecipitation, 8 μg of antibodies was prebound to 30 μl of Protein G Sepharose 4 Fast Flow (Amersham Biosciences, Little Chalfont, United Kingdom). Separase activation in Xenopus extract and Scc1 cleavage assays were done as described elsewhere [13], except that the Scc1 cleavage reaction was performed in the presence of 1.3 μg/μl antigenic peptide. Amounts and self-cleavage of separase were analyzed by immunoblotting aliquots before and after incubation in the extract.
This work was supported by grants from the Deutsche Krebshilfe and the Bundesministerium für Bildung und Forschung (BMBF, NGFN-2; PTJ-BIO/0313377A) to MRS. OS was supported by grants from the Deutsche Forschungsgemeinschaft (Emmy Noether Program) and from the Human Frontier Science Program. SH was supported by a fellowship of the Boehringer Ingelheim Foundation. We are grateful to Dr. Christoph Lengauer for helpful discussions, to Dr. Sabine Langer for evaluation of experiments, and to Cora Beier for critically reading the manuscript.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. KP, OS, and MRS conceived and designed the experiments. KP and SH performed the experiments. KP, SH, JC, OS, and MRS analyzed the data. OS and MRS wrote the paper.
Citation: Pfleghaar K, Heubes S, Cox J, Stemmann O, Speicher MR (2005) Securin is not required for chromosomal stability in human cells. PLoS Biol 3(12): e416.
Abbreviations
M-FISHmultiplex fluorescence in situ hybridization
==== Refs
References
Gruber S Haering CH Nasmyth K Chromosomal cohesin forms a ring Cell 2003 112 765 777 12654244
Haering CH Nasmyth K Building and breaking bridges between sister chromatids Bioessays 2003 25 1178 1191 14635253
Losada A Hirano M Hirano T Identification of Xenopus SMC protein complexes required for sister chromatid cohesion Genes Dev 1998 12 1986 1997 9649503
Waizenegger IC Hauf S Meinke A Peters JM Two distinct pathways remove mammalian cohesin from chromosome arms in prophase and from centromeres in anaphase Cell 2000 103 399 410 11081627
McGuinness BE Hirota T Kudo NR Peters JM Nasmyth K Shugoshin prevents dissociation of cohesin from centromeres during mitosis in vertebrate cells PLoS Biol 2005 3 e86 10.1371/journal.pbio.0030086 15737064
Stemmann O Boos D Gorr IH Rephrasing anaphase: separase FEARs shugoshin Chromosoma 2005 113 409 417 15703941
Giménez-Abián JF Sumara I Hirota T Hauf S Gerlich D Regulation of sister chromatid cohesion between chromosome arms Curr Biol 2004 14 1187 1193 15242616
Hauf S Roitinger E Koch B Dittrich CM Mechtler K Dissociation of cohesin from chromosome arms and loss of arm cohesion during early mitosis depends on phosphorylation of SA2 PLoS Biol 2005 3 e69 10.1371/journal.pbio.0030069 15737063
Hauf S Waizenegger IC Peters JM Cohesin cleavage by separase required for anaphase and cytokinesis in human cells Science 2001 293 1320 1323 11509732
Ciosk R Zachariae W Michaelis C Shevchenko A Mann M An ESP1/PDS1 complex regulates loss of sister chromatid cohesion at the metaphase to anaphase transition in yeast Cell 1998 93 1067 1076 9635435
Uhlmann F Lottspeich F Nasmyth K Sister-chromatid separation at anaphase onset is promoted by cleavage of the cohesin subunit Scc1 Nature 1999 400 37 42 10403247
Zou H McGarry TJ Bernal T Kirschner MW Identification of a vertebrate sister-chromatid separation inhibitor involved in transformation and tumorigenesis Science 1999 285 418 422 10411507
Stemmann O Zou H Gerber SA Gygi SP Kirschner MW Dual inhibition of sister chromatid separation at metaphase Cell 2001 107 715 726 11747808
Gorr IH Boos D Stemmann O Mutual inhibition of separase and Cdk1 by two-step complex formation Mol Cell 2005 19 135 141 15989971
Cohen-Fix O Peters JM Kirschner MW Koshland D Anaphase initiation in Saccharomyces cerevisiae is controlled by the APC-dependent degradation of the anaphase inhibitor Pds1p Genes Dev 1996 10 3081 3093 8985178
Funabiki H Yamano H Kumada K Nagao K Hunt T Cut2 proteolysis required for sister-chromatid separation in fission yeast Nature 1996 381 438 441 8632802
Nasmyth K Peters JM Uhlmann F Splitting the chromosome: Cutting the ties that bind sister chromatids Science 2000 288 1379 1385 10827941
Saez C Japon MA Ramos-Morales F Romero F Segura DI hpttg is over-expressed in pituitary adenomas and other primary epithelial neoplasias Oncogene 1999 18 5473 5476 10498901
Ramos-Morales F Dominguez A Romero F Luna R Multon MC Cell cycle regulated expression and phosphorylation of hpttg proto-oncogene product Oncogene 2000 19 403 409 10656688
Jallepalli PV Waizenegger IC Bunz F Langer S Speicher MR Securin is required for chromosomal stability in human cells Cell 2001 105 445 457 11371342
Mei J Huang X Zhang P Securin is not required for cellular viability, but is required for normal growth of mouse embryonic fibroblasts Curr Biol 2001 11 1197 1201 11516952
Huang X Hatcher R York JP Zhang P Securin and separase phosphorylation act redundantly to maintain sister chromatid cohesion in mammalian cells Mol Biol Cell 2005 16 4725 4732 16030258
Bunz F Fauth C Speicher MR Dutriaux A Sedivy JM Targeted inactivation of p53 in human cells does not result in aneuploidy Cancer Res 2002 62 1129 1133 11861393
Michor F Iwasa Y Lengauer C Nowak MA Dynamics of colorectal cancer Semin Cancer Biol 2005 15 484 493 16055342
Cahill DP Kinzler KW Vogelstein B Lengauer C Genetic instability and Darwinian selection in tumours Trends Cell Biol 1999 9 M57 M60 10611684
Speicher MR Ballard SG Ward DC Karyotyping human chromosomes by combinatorial multi-fluor FISH Nat Genet 1996 12 368 375 8630489
Azofeifa J Fauth C Kraus J Maierhofer C Langer S An optimized probe set for the detection of small interchromosomal aberrations by use of 24-color FISH Am J Hum Genet 2000 66 1684 1688 10762552
|
16292982
|
PMC1287505
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e416
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030416
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1633605110.1371/journal.pbio.0030422Research ArticleEvolutionZoologyMolluscsDNA Barcoding: Error Rates Based on Comprehensive Sampling Barcoding Error RatesMeyer Christopher P [email protected]
1
Paulay Gustav
1
1Florida Museum of Natural History, University of Florida, Gainesville, Florida, United States of AmericaGodfray Charles Academic EditorImperial College at Silwood ParkUnited Kingdom12 2005 29 11 2005 29 11 2005 3 12 e4224 7 2005 11 10 2005 Copyright: © 2005 Meyer and Paulay.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.
DNA Barcodes Perform Best with Well-Characterized Taxa
DNA barcoding has attracted attention with promises to aid in species identification and discovery; however, few well-sampled datasets are available to test its performance. We provide the first examination of barcoding performance in a comprehensively sampled, diverse group (cypraeid marine gastropods, or cowries). We utilize previous methods for testing performance and employ a novel phylogenetic approach to calculate intraspecific variation and interspecific divergence. Error rates are estimated for (1) identifying samples against a well-characterized phylogeny, and (2) assisting in species discovery for partially known groups. We find that the lowest overall error for species identification is 4%. In contrast, barcoding performs poorly in incompletely sampled groups. Here, species delineation relies on the use of thresholds, set to differentiate between intraspecific variation and interspecific divergence. Whereas proponents envision a “barcoding gap” between the two, we find substantial overlap, leading to minimal error rates of ~17% in cowries. Moreover, error rates double if only traditionally recognized species are analyzed. Thus, DNA barcoding holds promise for identification in taxonomically well-understood and thoroughly sampled clades. However, the use of thresholds does not bode well for delineating closely related species in taxonomically understudied groups. The promise of barcoding will be realized only if based on solid taxonomic foundations.
An examination of the efficacy of barcoding using a comprehensive sample of marine gastropods reveals that the method performs poorly for identifying closely related species in taxonomically understudied groups.
==== Body
Introduction
The Controversy
DNA barcoding, the recently proposed DNA-based project for species identification, has attracted much attention and controversy [1–6]. Proponents envision that a short fragment of DNA can be used to diagnose taxa, increasing the speed, objectivity, and efficiency of species identification. Initial tests of genetic barcoding using mitochondrial markers on animals reported near-100% accuracy, indicating that the method can be highly accurate under certain conditions [1,7,8]. Accurate species identification—assignment of an unknown to a known—requires a comprehensive comparative molecular database against which unknowns can be compared. However, it is clear that most of the biological diversity in the world is undocumented [9,10]. Therefore, a stated second goal of DNA barcoding is to facilitate the species-discovery process [11–13]. Such a proposal has raised the concern of the systematics community, which claims that adopting barcoding would be a step backwards [14–16], returning taxonomy to typology [17]. Opponents also note that mitochondrial DNA (mtDNA) sequences alone may be insufficient to diagnose species, because genetic differentiation does not necessarily track species boundaries [18,19]. Thus, Funk and Omland [20] found that ca. 23% of surveyed metazoan species are genetically polyphyletic or paraphyletic, implying that they would not be differentiable by barcoding techniques.
What Does Accuracy Depend on? The Barcoding “Gap”
The critical issue in barcoding is accuracy. How well does a single gene sequence perform in delineating and identifying species? Accuracy depends especially on the extent of, and separation between, intraspecific variation and interspecific divergence in the selected marker. The more overlap there is between genetic variation within species and divergence separating sister species, the less effective barcoding becomes. Initial efforts to test barcoding suggested a significant barcoding “gap” between intra- and interspecific variation, but these efforts have greatly undersampled both intraspecific variation (mostly 1–2 individuals per species sampled) and interspecific divergence (because of incomplete or geographically restricted sampling) [1,7,8].
Overlap between Intra- versus Interspecific Variation
When the coalescent has yet to sort between incipient species (ancestral polymorphism), intraspecific variation overlaps with interspecific divergence and gives rise to genetically polyphyletic or paraphyletic species (Figure 1) [18,21,22]. When such overlap is real (i.e., not the result of poor taxonomy), then that marker cannot reliably distinguish among those species. Overlap between intraspecific and interspecific variation can also occur broadly within a tree, even when each species is reciprocally monophyletic to all others. This occurs when intraspecific variation in parts of the tree exceeds interspecific divergence in other parts of the tree—i.e., when the range of intra- and interspecific variation overlaps. Such overlap will not affect identification of unknowns in a thoroughly sampled tree, where they should fall within the coalescent of already characterized species. However, such overlap can have a substantial impact during the discovery phase (i.e., in an incompletely sampled group), as the status of unknowns that fall outside the coalescent of previously sampled species is problematic to evaluate.
Figure 1 Phylogenetic Relationships and Terminology
(A) Reciprocal monophyly. Members of each species share a unique common ancestor. For each species, the white star represents the coalescent, the point at which all extant haplotypes share a common ancestry.
(B) Paraphyly. One species (Y), is monophyletic, but nests within another recognized species (X). Thus, the coalescent of species Y (small star) is contained within the coalescent of species X (large star).
(C) Polyphyly. Neither species X or Y are monophyletic, and both coalesce to the white star.
Gap versus Overlap: The Efficacy of Thresholds
The proposed mechanism for the evaluation of unknowns within a partially sampled phylogeny is through the implementation of thresholds, chosen to separate intraspecific variation from interspecific differences. An unknown differing from an existing sample by less than a threshold value is assumed to represent that species, but one differing from existing sequences by more than the threshold value is assumed to represent a new taxon. This method is vulnerable to both false positives and false negatives. False positives are the identification of spurious novel taxa (splitting) within a species whose intraspecific variation extends deeper than the threshold value; false negatives are inaccurate identification (lumping) within a cluster of taxa whose interspecific divergences are shallower than the proposed value. The accuracy of a threshold-based approach critically depends upon the level of overlap between intra- and interspecific variation across a phylogeny. While Hebert et al. [7] suggest that a wide gap between intra- and interspecific variation makes a threshold approach promising (Figure 2A), Moritz and Cicero [23] argue that the overlap is considerably greater when a larger proportion of closely related taxa are included, making the method problematic (Figure 2B). To evaluate the performance of this method, we need to assess the extent of and overlap between intra- and interspecific genetic variation comprehensively, within a thoroughly sampled clade [23,24].
Figure 2 Schematic of the Inferred Barcoding Gap
The distribution of intraspecific variation is shown in red, and interspecific divergence in yellow. (A) Ideal world for barcoding, with discrete distributions and no overlap. (B) An alternative version of the world with significant overlap and no gap.
Here we present the first dataset sufficiently comprehensive to robustly evaluate the efficacy of DNA barcoding: the cowrie genetic database [25,26]. This dataset includes sequences from >2,000 individuals in 263 taxa, representing >93% of recognized cowrie (marine gastropods of the family Cypraeidae) species worldwide, with multiple individuals from >80%, and at least five individuals from >50% of the taxa. These data provide near comprehensive sister-species coverage, and a broad survey of intraspecific variation. We use this dataset to address several questions. How accurate is molecular identification of unknowns in a thoroughly sampled tree? What are the reasons for failures in such identifications? How much do intraspecific variation and interspecific divergence overlap across this well-sampled phylogeny? How much error is associated with threshold-based identifications, and what threshold value minimizes this error? Finally we use data from two smaller but similarly exhaustively sampled clades (of limpets [27] and turbinid gastropods [28]), to evaluate the generality of these patterns. Cowries encompass a diversity of species attributes: recent versus ancient, planktonic versus direct development, common versus rare, and large Indo-west Pacific-wide ranges versus single island endemics. All cowries have internal fertilization, mostly with feeding larvae, whereas limpets and turbinids are external fertilizers, with non-feeding larvae. While all three examples are gastropods, their range of species attributes implies that these findings are likely applicable to a wide range of taxa.
The effectiveness of barcoding is critically dependent upon species delineation: splitting decreases while lumping increases both intraspecific variation and interspecific divergence. Taxonomically, cowries are one of the most extensively studied marine gastropod families, both morphologically [29–33] and genetically [25,26], thus their species are well circumscribed. We analyze and compare barcoding performance for two types of species-level taxa based on different levels of taxonomic analysis: (1) traditional, morphological species, as defined by the most recent morphology-based revision [33], and (2) evolutionary significant units (ESUs), as defined through an integrative taxonomic analysis of combined and extensively sampled genetic and morphological data (slightly modified from [25,26]). We thus compare the efficacy of barcoding across a 2 × 2 matrix: performance with traditional species versus ESUs in an identification versus discovery setting. Traditional species provide a test of barcoding when substantial morphological information is available, but remain untested with genetic tools. This level of knowledge is comparable to biotic checklists, which are often used to guide sampling in barcoding efforts. In contrast, ESUs provide the best of integrative taxonomy, a system where population-level and geographically extensive genetic sampling has tested species-level boundaries described by extensive morphological studies. Because ESUs are defined as reciprocally monophyletic units, they exclude the possibility of, and errors associated with, paraphyletic or polyphyletic species, and thus provide the optimal units for barcoding. Given that at present their reciprocal monophyletic status is based on the same genetic marker used for barcoding, they should lead to 100% accuracy in species identification tests. Presently, cowrie ESUs exclude potentially valid, young species that are not reciprocally monophyletic in cytochrome c oxidase I (COI) sequences; however, additional work may demonstrate some of these to be valid species. ESUs fulfill the phylogenetic species concept; however, we choose to recognize them only as ESUs, to emphasize that although they are genetically divergent and distinctive, they all are not, or destined to become, biological species.
The correspondence between ESU definitions and traditional morphological taxonomy is high. Remarkably, 255 ESUs (97%) have been recognized previously at either the specific or subspecific level and are therefore supported by independent morphological criteria in addition to molecular data. Only eight ESUs are genetically distinct but have not been previously recognized by traditional taxonomy; all of these are allopatric, genetically divergent lineages. So defined, the 263 ESUs sampled include >93% of the 233 recognized cowrie species and 56 recognized subspecies. From here on, we use “ESU” to denote taxa recognized through an integrated approach with the aid of molecular criteria, and “species” to refer to taxa recognized at that level in traditional cowrie taxonomy. The same definition led to the recognition of 12 ESUs in the Patelloida profunda group of limpets [27] and 30 ESUs in the Astralium rhodostomum complex of turbinid gastropods [28]. In both groups traditional taxonomic study lags substantially behind cowries, and many ESUs represent undescribed, but morphologically recognizable, species.
Results/Discussion
Accuracy in Thoroughly Sampled Phylogenies: Identification
Identification of unknowns against a thoroughly sampled phylogeny was prone to error when traditional species were utilized, but accurate when ESUs formed the basis of the phylogeny. Assignment of unknowns to a phylogeny comprised of exemplars of every traditional species was correct 80% of the time using a neighbor-joining approach (see Materials and Methods). Eight percent of the assignments were incorrect, while 12% were ambiguous, with the unknown falling as sister to a clade comprised of its species plus its sister species. Parsimony analyses were unambiguously correct 79%, incorrect 7%, and ambiguous 10% of the time, while the correct placement was one of multiple, equally parsimonious placements in 4% of the cases. Ambiguous assignments also represent failures of the barcoding method, as although the unknowns “belong” to sampled species, they fall outside of that species as characterized by an exemplar approach, and could represent a novel taxon. This approximately 20% failure rate at the species level is consistent with Funk and Omland's [20] assessment that 23% of metazoan species are not monophyletic.
In contrast, identification of unknowns was 98% accurate with a neighbor-joining approach against an ESU phylogeny. Similar analyses of turbinid and limpet datasets had success rates of 100% and 99%, respectively. These results are not unexpected, however, as the reciprocal monophyly criteria for circumscribing units predisposed the system for success. More surprising is the 2% failure rate (1% each from incorrect assignment and ambiguity). In these incorrect identifications, improper assignment involved a recently derived sister ESU. These failures occur because only a single exemplar was used to define ESUs in the phylogenies. The rooting of the three-taxon arrangement between the sample, correct ESU, and sister ESU is tenuous, and vulnerable to artifacts of incomplete sampling. If all sequenced haplotypes were included in the analyses, the unknown would have been correctly assigned. Nevertheless, these high success rates are encouraging, particularly since only a single exemplar was used for comparison [34], and many of the divergences between sister taxa are shallow.
What are the sources for the 20% failure rate in species-level analyses? Non-monophyly at the species level leads to barcoding failure both in thoroughly sampled and threshold approaches, and represents the greatest challenge for the method. Funk and Omland [20] recognize five reasons for species-level non-monophyly; two of these account for most non-monophyly in cowries: imperfect taxonomy and incomplete lineage sorting. Imperfect taxonomy can cause non-monophyly either through lack of recognition of multiple taxa within a traditional species (overlumping) or when morphotypes are inappropriately recognized as species (oversplitting). Overlumping is common in cowries and readily identified via thorough genetic sampling: 16 recognized cowrie species (7%) are nested ESUs within other, paraphyletic species comprised of multiple ESUs (e.g., Palmadusta artuffeli within P. clandestina; Figure 3). Oversplitting is more difficult to resolve because young species that remain within their sister species' coalescent lead to the same polyphyletic, genetic signature. Of 218 traditional cowrie species tested [25,26], 18 (8%) are polyphyletic with respect to another recognized species. These are either young species (incomplete lineage sorting), or artificially split forms (imperfect taxonomy); additional research is needed to resolve their status. Note that such young species are also neglected by the ESU approach and represent the ultimate limit for barcoding: non-monophyly that cannot be eliminated at the marker (COI) used.
Figure 3 Intraspecific and Interspecific Estimations
A subclade of five cowrie ESUs shows how both coalescent and divergence depths are generated. The two most disparate individuals are culled from within each ESU (left—red) and used in a constrained phylogeny with a molecular clock enforced (right) to recover both the maximum coalescent depth (red) and the divergence depths between sisters (yellow). Two young ESUs (stars) would be missed (false negatives) if a 3% threshold cutoff (shown) was employed. Note that Palmadusta artuffeli, a Japanese endemic species, is nested among monophyletic subspecies of the paraphyletic species P. clandestina. The black circle indicates the coalescent for the species P. clandestina, and the black star indicates the interspecific divergence for species-level analyses.
Using the ESU concept in hindsight, we can ascribe the failures in our species-level test to artifacts of paraphyly or polyphyly (Figure 1). Ten percent of the failures can be attributed to overlumped, paraphyletic species, while nine percent are the results of either oversplit or young (incompletely sorted) polyphyletic species. The remaining 1% is real error based on single exemplars of the type mentioned previously.
The other three causes of species non-monophyly (inadequate phylogenetic information, unrecognized paralogy, and introgression) identified by Funk and Omland [20] are of minor importance in these studies. Since all three gastropod datasets are well circumscribed using morphological, anatomical, geographic, and molecular attributes, we have minimized the problems of inadequate phylogenetic information. We can estimate error rates associated with paralogy and the presence of nuclear copies of mtDNA (NUMTs; [35]). In generating sequence data for 2,026 cowrie individuals, seven sequences (0.3%) have been generated that are thought to be NUMTs, all within three species. Low levels of NUMTs (<1%) were also reported by Hebert et al. [12] in their study of Astraptes butterflies. NUMTs can be problematic in some taxa (e.g., [36]), but their presence is usually ascertained by translation shifts in amino acid patterns or signal deterioration in electropherograms derived from non-cloned products. The final source of non-monophyly is introgression. Hybrid individuals have been reported within cowries, and indeed, mtDNA data reveal that individuals assigned conchologically to certain species or subspecies possess haplotypes of closely related lineages, indicating some past introgressive or hybridization event. Using only mtDNA sequences, these individuals would be identified incorrectly. How frequently does this occur? Less than 2% of cowrie individuals, 1% of turbinids, and 0% of limpets possess COI sequences inconsistent with their morphology, indicating that the impact of introgression has been minor. Nevertheless, this low frequency should be included in error estimation. Therefore, our overall empirical error in the best of situations (ESUs) for species identification is 4%–12%: 2% because of the use of single exemplars, 2% from introgression, and 0%–8% from polyphyletic species.
Accuracy in Undersampled Phylogenies Using Thresholds: Discovery
To evaluate the efficacy of thresholds for species delineation in a partially sampled clade, we examined the overlap between intra- and intertaxon divergences at both ESU and species levels using a phylogenetic approach. Three different metrics were used to characterize intraspecific variation: (1) average pairwise intraspecific difference (K2P distance) between all individuals sampled within species/ESU, as employed by previous researchers [7,8]; (2) average theta (θ), where theta is the mean pairwise distance within each taxon, thereby eliminating bias associated with uneven sampling among taxa; and (3) average coalescent depth, the depth of the node linking all sampled extant members of a taxon, bookending intraspecific variability (see Materials and Methods). Genetic distance between terminal taxa and their closest sister was used to characterize interspecific divergence.
A wide range of intraspecific variation was encountered among ESUs in all three datasets, with generally less variation in turbinids and limpets than cowries. Sampling effort was designed to capture the greatest intraspecific variation by targeting the most disparate populations in a taxon, whenever possible. Thus, while coalescent depths generally increase with sample size (Figure 4A), they are variable, and ESUs with n ≥ 2, n ≥ 5, and n ≥ 10 samples overlap broadly (Figure 4B). The distribution of all intraspecific, pairwise genetic distances approximates a Poisson distribution (Figure 5A). Calculated values of theta for cowrie ESUs with ≥ten samples are normally distributed, and are highly correlated with estimated coalescent depth (Figure 5B). All three measures of intraspecific variability (average pairwise distances, theta, and coalescent depth) are substantially higher in cowries than in turbinids or limpets (Table 1). This may be a result of smaller effective population sizes in the latter two groups [37], reflecting their poor dispersal abilities because of non-feeding larvae, and resultant narrow ESU ranges. A similar pattern is evident within cowries: taxa that lack planktonic larvae and consequently have restricted dispersal and narrow ranges, have a smaller mean theta (0.0029) than cowries that possess planktonic larvae (0.0070).
Figure 4 Sample Size Effect on Intraspecific Variation
A. Coalescent depth vs. sample size. B. Histograms for coalescent depths of various sample size classes. Mean coalescent depth increases with increased sample size, from 0.0049 for n ≥ 2, to 0.0057 for n ≥ 5, to 0.0068 to 0.0070 for n ≥ 10.
Figure 5 Alternative Metrics for Intraspecific Variation
(A) Distribution of all intraspecific pairwise K2P distances for cowrie ESUs with n ≥ 10, turbinids and limpets. Left y-axis for cowries; right y-axis for others.
(B) Comparison between estimated theta versus estimated maximum coalescent for each cowrie ESU with n ≥ 10; r2 = 0.837.
(C) Distribution of theta values for cowrie (n ≥ 10), turbinid and limpet ESUs.
Table 1 Intraspecific Variation
As with intraspecific variation, a wide range of interspecific differences is found in all three gastropod groups, indicating that divergences are spread out over time (Figure 6). It is interesting to note that intraspecific variation (as measured by coalescent depth) is not correlated with interspecific divergence for ESUs ≥ five individuals (p = 0.12), indicating that older species (those without close extant relatives) do not have more intraspecific variation than younger species (those with close relatives).
Figure 6 Interspecific Variation
Distribution of divergence depths between terminal ESUs and their sister ESU(s) in cowries, turbinids and limpets.
Gap or overlap? Efficacy of thresholds with ESUs
We found broad overlap between levels of intraspecific variation and interspecific divergence at the ESU level in cowries. Intraspecific variation is well constrained: only five ESUs (2%–3% of ESUs with n ≥ 10, n ≥ 5, and n ≥ 2 samples/ESU) have coalescent depths >1.5% (=3% threshold), and none have >2% (=4% threshold) (Table 2). Coalescent depths are recorded as nodal depths, and thus are half the value of pairwise distances commonly reported for threshold values. Therefore, if an unknown was >3% divergent from all other samples, we could say with ~98% confidence that it represents an independent evolutionary lineage. Such false-positive errors become rapidly more common at lower thresholds, as 20%, 15%, 11% of ESUs (with n ≥ 10, n ≥ 5, and n ≥ 2 samples/ESU) have coalescent depths >1% (=2% threshold). In turbinids and limpets, all coalescent depths are <1%, thus none yield a false positive at even a 2% threshold. Because these error rates are determined by maximum coalescent depth, this assessment of performance is conservative. Two randomly chosen individuals within an ESU will likely be less divergent than the two most disparate individuals. For direct comparison with Hebert et al. [7] and Barrett and Hebert [8], examination of all intraspecific pairwise distances (Figure 5A) yields 99% and 95% confidence values at thresholds set at 2.85% and 1.99% in cowries, 1.12% and 0.81% for turbinids, and 1.38% and 0.52% for limpets, respectively.
Table 2 False-Positive Error Rates
In contrast, interspecific ESU divergences are much less constrained, extending at their lower end well into the range of intraspecific variation (Figure 7A). Thus, high divergence thresholds miss many young ESUs. Of the 263 cowrie ESUs sampled, 16% would be artificially lumped with another ESU at a 3% threshold, and 8% would be lumped even at a 2% cutoff (Table 3). Most (79% at the 3% threshold) of the lumped ESUs are allopatrically distributed sister taxa, yet more than half (22 of 42) are traditionally recognized species. A similar percentage of taxa would be overlooked at a 3% cutoff in turbinids (20%) and limpets (17%) (Table 3). This high incidence of false negatives reflects both the comprehensive phylogenetic sampling and increased taxonomic scrutiny these taxa have received.
Figure 7 Barcoding Overlap: Cowrie ESUs
(A) Relative distributions of intraspecific variability (coalescent depth—red) and interspecific divergence between ESUs (yellow), demonstrating significant overlap and the lack of a barcoding gap. Note that the x-axis scale shifts to progressively greater increments above 0.02.
(B) Cumulative error based on false positives plus false negatives for each threshold value. The optimum threshold value is 0.013 (2.6%), where error is minimized at 17%.
Table 3 False-Negative Error Rates
How high should thresholds be set to minimize error? False-positive and false-negative error rates can be totaled for any threshold value across the phylogeny, and combined error minimized (Figure 7B). In cowries the lowest overall error (17%) was at a threshold values of 2.6%, and error varied little (17%–19%) between 2.4%–3.4% thresholds. Errors at these levels are largely the result of missing young taxa, not of false recognition of additional species. In turbinids, as in cowries, the distribution of intra- and interspecific divergences overlap, and combined error is lowest (7%) at thresholds values of 1.2%–1.6% (Figure S1). In contrast, thresholds are effective and error can be entirely eliminated in limpets: there is no overlap at a threshold of 1.7% (Figure S2). The better performance of turbinids and limpets is likely in part the result of their shallower coalescents and lower diversity.
A 3% threshold has been cited as sufficient genetic disparity to characterize different species [1]. The actual threshold value that researchers would be willing to accept as indicative of a new taxon, if any, varies depending upon philosophy, marker choice, and group of organisms. For our three datasets, a 3% threshold would work well at minimizing false positives, but it would create many false negatives. Alternatively, Hebert et al. [7], in order to screen for novel taxa, proposed to set a standard sequence threshold value that minimizes false positives at ten times the mean intraspecific variation. This would set the threshold at around 8% in cowries (based on 60 ESUs with n ≥ 10; Table 1), well above the 4% level where all false positives are eliminated, considerably above the optimum (2.6%), and leading to a 34% error rate (all false negatives).
Substantial variation in the relationship between intraspecific variation and optimal threshold values to either minimize combined error or to eliminate all false positives makes setting the latter on the former problematic. The optimum threshold values to minimize total error correspond to 3.2–4.1 times the level of intraspecific variation in cowries, depending on which measure is used (Table 4). The factors range from 4.9–6.3× if one were to use a conservative threshold that eliminates false positives. The corresponding ranges for turbinids are 4.8–7.8×, while for limpets it is at 5.7–6.8× (Table 4). The range of values among these gastropods and Hebert et al.'s [7] bird samples indicate that no simple formula based on intraspecific variation will yield a robust threshold to minimize error across groups.
Table 4 Threshold Correction Factors
Thresholds can be used to either minimize total error or to cleanly screen for novel taxa. Our results imply that they serve poorly for the former, as high error rates remain at even optimal threshold values. However thresholds can certainly be set in a way to guarantee that sequences beyond them represent novel taxa. In cowries, a 4% screening threshold eliminates all known intra-ESU variants, and guarantees that such divergent taxa are novel. However the same threshold will also miss 21% of novel taxa, as they will register less divergent. Thus, thresholds can assist in the species discovery process by guaranteeing the distinctiveness of genetically deep variants, at some cost.
Efficacy of thresholds with traditional taxonomy
Error rates almost double if we replace cowrie ESUs with the currently recognized species. This increase in error is the result of a simultaneous increase in the range of intraspecific variation and interspecific divergence, creating a wider overlap between the two in species than in ESUs. Intraspecific variability is substantially higher in traditional species than ESUs for all three metrics. The distribution of all intraspecific pairwise comparisons is multi-modal, reflecting the lumping of discrete ESUs (Figure 8A). The means of all intraspecific pairwise distances and theta are both three times as high within species (2.97%, θ = 1.86%) than within ESUs (0.81%, θ = 0.63%) (Figure 8). The range of interspecific divergence is also increased because numerous traditionally recognized cowrie species are not monophyletic in their COI, either because their coalescents have not sorted, or because they represent forms recognized by splitters that are not based on biological species. As a result, overlap between intra- and interspecific differences and error rates associated with thresholds both are greater when traditional species are used (Figure 9). The optimal threshold for recognized species is 5%, with an error rate of 33%. The 2.6% threshold, optimal for ESUs, yields a 37% error rate. Thus, thresholds fare poorly even in a thoroughly sampled phylogeny, if the basis for sampling is traditionally recognized species. This result is a strong warning against limited sampling to exemplars for taxa based on species checklists, even for relatively well-known groups. Had we not sampled the various subspecies of cowries and geographic locations as in the turbinids and limpets, we would have had a very different perspective on intra- and interspecific divergences.
Figure 8 Intraspecific Variation Based on Recognized Species
(A) The distribution of all intraspecific pairwise distances for traditionally recognized cowrie species with n ≥ 10. The white bars represent intraspecific distances where the two specimens compared fall into separate ESUs.
(B) Theta values for traditionally recognized cowrie species. Black bars are species that correspond to an ESU; white bars are species that include multiple ESUs.
Figure 9 Barcoding Overlap: Cowrie Species
Data are presented as in Figure 7; however, estimates of intraspecific variation and interspecific divergence are based on traditionally recognized cowrie species.
(A) Relative distributions of intraspecific variability (coalescent depth—red) and interspecific divergence between species (yellow), demonstrating a more pronounced overlap than when utilizing ESUs. Note that the x-axis scale shifts to progressively greater increments above 0.02.
(B) Cumulative error based on false positives plus false negatives for each threshold value. The optimum threshold value is 0.025 (5%), where error is minimized at 33%.
Global versus regional sampling
This broad overlap contrasts with Hebert et al.'s [7] and Barrett and Hebert's [8] findings of a wide separation between intraspecific variation and interspecific divergence in a sample of North American birds and spiders. What causes this difference? This difference likely reflects differential intensity (number of samples per species/ESU) and scale (regional versus global) of sampling, rather than differences among birds, spiders, and snails. First, Hebert et al.'s [7] appraisal of intraspecific variation was limited, and thus they underestimated intraspecific variation [23,24]. Second, they substantially undersampled true sister species pairs [23,24,38], and thus overestimated interspecific divergence. Regional studies [1,7,8] undersample the most closely related species, which are frequently allopatric, and thus underestimate global error rates. The purported barcoding “gap” reported in these studies is the best-case scenario, and can only get worse (decrease or disappear) with increased intraspecific and interspecific sampling. Error rates are lower in regionally scaled analyses if the geographic scale of the study excludes allopatric sister taxa, thus artificially increasing observed interspecific divergence levels. For instance, a barcoding gap does exist if only cowries from the island of Moorea were investigated. The geographic scale where such reduction in error occurs is dependent on the geographic mode and scale of speciation of the group. While marine gastropods sampled at a single island would generally not include any allopatric sister taxa, terrestrial gastropods sampled at that same island may include many shallow sisters, if the landsnail group has undergone in situ radiation [39]. Consequently, error rates can be high even in geographically restricted analyses if diversification is local—through fine-scale allopatric speciation, sympatric speciation, polyploidy, or rapid attainment of sympatry following allopatry—or when invasive species homogenize the biota.
Conclusions
Two principal elements are proposed in DNA barcoding: (1) the ability to assign an unknown sample to a known species, and (2) the ability to detect previously unsampled species as distinct. The prospect of assigning an unknown to a known is promising especially for well-known, comprehensively sampled groups that have been extensively studied by genetic and morphological taxonomy. In such globally comprehensive and well-circumscribed datasets, the majority of individuals (>96% in these snails) may be successfully identified by a short fragment of mtDNA. However, even in such extensively studied taxa, a certain percentage of young species (0%–8% in cowries) will not be discernable because of ancestral polymorphism. DNA barcoding is much less effective for identification in taxa where taxonomic scrutiny has not been thorough, and species recognition is limited to a few traditional character sets, untested by additional studies and tools. In such modestly known groups, which represent the bulk of life on Earth, many species will appear to be genetically non-monophyletic because of imperfect taxonomy [20], contributing to a high error rate for barcode-based identification. Thus, to create an effective environment for identification through barcoding, comprehensive, taxonomically thoroughly studied, comparative databases are necessary. The barcoding movement will play a leading role in generating the standards and protocols for establishing these databases, and facilitating their development.
The promise of barcoding for species discovery based on methodologies currently proposed should be tempered. The use of thresholds for species delineation is not promising and is strongly discouraged, as levels of overlap between intra- and interspecific differences are likely to be significant in most major clades, particularly within diverse yet poorly documented groups. Thresholds can be effective in screening for substantially divergent novel taxa, but our data indicate such use will overlook at least one-fifth of life's forms that are distinct but less divergent. More elegant methodologies will be required that incorporate principles of population genetics, knowledge of intraspecific variability, and sister group attributes. Identifications or discoveries may be placed within a statistical framework [40], allowing statements such as “based on the data at hand, sample X is 83% likely to be a member of taxon A.” The Data Analysis Working Group (DAWG) associated with the Consortium for the Barcoding of Life (CBOL) is pursuing these analytical challenges.
While the barcode is certainly a link out and can provide access to life's encyclopedia, this book needs to be written in collaboration with taxonomists, systematists, and ecologists, in an integrative taxonomic framework [17,41,42]. Barcoding on a global scale can only achieve high accuracy once the majority of evolutionary units have been sampled and taxonomically assessed. This critical first step was achieved for the studied gastropod taxa by centuries of careful, traditional taxonomic consideration (cowries) and large sample sizes (for all three). Without this initial phase, a threshold approach is likely to fail for ~20% of the taxa and individuals at the species discovery phase.
Materials and Methods
We sequenced 2,026 cowries for 614 bp of COI mtDNA, the traditional Folmer primer region proposed for barcoding most metazoans. Two or more individuals were sequenced from 82% (216) of ESUs, ≥5 from 54% (143), and ≥10 from 23% (60). To maximize recovery of the greatest intraspecific variation and test for geographical structuring, sequences were generated from the most geographically distant populations available. Molecular methods followed standard procedures and are reviewed in Meyer [25,26], Kirkendale and Meyer [27], and Meyer et al. [28].
We used standard, tree-based methods to address accuracy of identification in a thoroughly sampled phylogeny using both a species-level and ESU approach. One exemplar from each recognized species (the nominal subspecies if the species included multiple subspecies) or each identified ESU was used as the reference “barcode” exemplar in topological comparisons. We randomly selected 1,000 sequences from the cowrie COI dataset, excluding barcode exemplars, and limiting representation of each species or ESU to 15 or ten sequences, respectively, to minimize bias toward well-sampled taxa. Hybrid individuals (see above) were excluded. These 1,000 sequences were tested one at a time, and their placement relative to the barcoding exemplars evaluated in both neighbor-joining (K2P) and parsimony phylogenies. Identification was considered correct if the sister taxon of the test sequence was the exemplar sequence of its corresponding species or ESU. Identification was considered incorrect if the sister taxon was wrong. If the random sequence fell below a node linking two recognized sister taxa including the corresponding species, the identification was considered ambiguous, as assignment to one or the other is equivocal, as the unknown could also represent a novel taxon. Similar analyses were performed with the turbinid (n = 200 from 278) and limpet (n = 100 from 125) datasets.
Pairwise K2P distances, theta, and coalescent depth were used to characterize intraspecific variation. Genetic distance between terminal taxa and their closest sister was used to characterize interspecific divergence. While the phylogenies used are based upon sequence data from two mtDNA markers (16S and COI: [26–28]), only COI was used for these analyses. The two most genetically distant individuals within each ESU (based on pairwise comparisons) were chosen to bookend genetic diversity and recover coalescent depth (maximum intra-ESU variability). These two individuals replaced the exemplar taxon used to construct the overall phylogeny (Figure 3). A likelihood ratio test (GTR + G with and without a clock enforced) was used to test for clock-like behavior (using only COI) in the resulting tree. A clock could not be falsified for turbinids and limpets (p > 0.05); but was falsified (p = 0.007) for cowries. Coalescent depths and interspecific divergence estimates throughout are based on topologies with a molecular clock enforced, although the overall cowrie data marginally rejected rate constancy. We estimated theta by calculating the average intraspecific difference using K2P distances. All analyses were conducted using PAUP* version 4.0b10 [43]. A listing of ESUs, number of individuals examined, interspecific divergence, and intraspecific metrics can be found in the supporting information for cowries (Table S1), turbinids (Table S2), and limpets (Table S3).
Supporting Information
Figure S1 Barcoding Overlap in Turbinids
(A) Relative distributions of intraspecific variability (coalescent depth—red) and interspecific divergence between ESUs (yellow). Note that the x-axis scale shifts to progressively greater increments above 0.01.
(B) Cumulative totals of false positives plus false negatives for each threshold value. The optimum threshold value is between 0.005 or 0.007 (1.0%–1.4%), where error is minimized at 7%.
(169 KB EPS).
Click here for additional data file.
Figure S2 Barcoding Gap in Limpets
(A) Relative distributions of intraspecific variability (coalescent depth—red) and interspecific divergence between ESUs (yellow). Note that the x-axis scale shifts to progressively greater increments above 0.01.
(B) Cumulative totals of false positives plus false negatives for each threshold value. A gap exists at a threshold of 0.0085 (1.7%), where error is eliminated.
(166 KB EPS).
Click here for additional data file.
Table S1 ESU Listing for Cowries
The table contains the taxon name, number of individuals sequenced, interspecific divergence (lineage) depth (GTR + G), coalescent depth (GTR + G), and estimated theta value (K2P). Asterisk (*) denotes direct developers, lacking planktonic larvae.
(343 KB DOC).
Click here for additional data file.
Table S2 ESU Listing for Turbinids
The table contains the taxon name, number of individuals sequenced, interspecific divergence (lineage) depth (GTR + G), coalescent depth (GTR + G), and estimated theta value (K2P).
(55 KB DOC).
Click here for additional data file.
Table S3 ESU Listing for Limpets
The table contains the taxon name, number of individuals sequenced, interspecific divergence (lineage) depth (GTR + G), coalescent depth (GTR + G), and estimated theta value (K2P).
(36 KB DOC).
Click here for additional data file.
Accession Numbers
The GenBank (http://www.ncbi.nlm.nih.gov/Genbank) accession numbers for sequences discussed in this paper are: cowrie (AY161637–AY161846, AY534433–AY534503, and DQ206992–DQ207351), limpet (AY628240–AY628327), and turbinid (AY787233–AY787400). The complete datasets are available also by request from CPM or at the Cowrie Genetic Database Web site (http://www.flmnh.ufl.edu/cowries).
Funding for these studies was provided by grants from the National Science Foundation (DEB-9807316/0196049 and OCE-0221382) to GP; and grants from Conchologists of America, the University of California Museum of Paleontology, and the National Institutes of Health (training grant in genetics) to CPM. Over 100 persons and many institutions contributed tissue samples for this study. They are acknowledged at the CGDP Web site. We would like to thank Brent Mishler, Craig Moritz, Mark Stoeckle, Kevin Omland, Felix Sperling, and two anonymous reviews for helpful comments or discussion on earlier drafts of this manuscript.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. CPM conceived and designed the experiments, performed the experiments, and analyzed the data. CPM and GP contributed reagents/materials/analysis tools and wrote the paper.
Citation: Meyer CP, Paulay G (2005) DNA barcoding: Error rates based on comprehensive sampling. PLoS Biol 3(12): e422.
Abbreviations
COIcytochrome c oxidase I
ESUevolutionary significant unit
mtDNAmitochondrial DNA
NUMTnuclear copy of mtDNA
==== Refs
References
Hebert PDN Cywinska A Ball SL DeWaard JR Biological identifications through DNA barcodes Proc R Soc Lond B Biol Sci 2003 270 313 321
Will KW Rubinoff D Myth of the molecule: DNA barcodes for species cannot replace morphology for identification and classification Cladistics 2004 20 47 55
Marshall E Taxonomy—Will DNA bar codes breathe life into classification? Science 2005 307 1037 15718446
Ebach MC Holdrege C DNA barcoding is no substitute for taxonomy Nature 2005 434 697
Schindel DE Miller SE DNA barcoding a useful tool for taxonomists Nature 2005 435 17
Gregory TR DNA barcoding does not compete with taxonomy Nature 2005 434 1067
Hebert PDN Stoeckle MY Zemlak TS Francis CM Identification of birds through DNA barcodes PLoS Biol 2004 2 e312 10.1371/journal.pbio.0020312 15455034
Barrett RDH Hebert PDN Identifying spiders through DNA barcodes Can J Zool 2005 83 481 491
Wilson EO The encyclopedia of life Trends Ecol Evol 2003 18 77 80
May RM Raven PH The dimensions of life of earth Nature and human society: The quest for a sustainable world 1997 Washington, D.C National Academy Press 30 45
Stoeckle M Taxonomy, DNA, and the bar code of life Bioscience 2003 53 796 797
Hebert PDN Penton EH Burns JM Janzen DH Hallwachs W Ten species in one: DNA barcoding reveals cryptic species in the neotropical skipper butterfly Astraptes fulgerator
Proc Natl Acad Sci U S A 2004 101 14812 14817 15465915
Janzen DH Now is the time Philos Trans R Soc Lond B Biol Sci 2004 359 731 732 15253359
Mallet J Willmott K Taxonomy: Renaissance or Tower of Babel? Trends Ecol Evol 2003 18 57 59
Seberg O Humphries CJ Knapp S Stevenson DW Petersen G Shortcuts in systematics? A commentary on DNA-based taxonomy Trends Ecol Evol 2003 18 63 65
Lipscomb D Platnick N Wheeler Q The intellectual content of taxonomy: A comment on DNA taxonomy Trends Ecol Evol 2003 18 65 66
Will KW Mishler BD Wheeler QD The perils of DNA bar-coding and the need for integrative taxonomy Syst Biol 2005 54 844 851 16243769
Avise JC Phylogeography: The history and formation of species 2000 Cambridge (Massachusetts) Harvard University Press 447
Avise JC Molecular markers, natural history and evolution 2004 Sunderland (Massachusetts) Sinauer Associates 541
Funk DJ Omland KE Species-level paraphyly and polyphyly: Frequency, causes, and consequences, with insights from animal mitochondrial DNA Annu Rev Ecol Evol Syst 2003 34 397 423
Rosenberg NA The shapes of neutral gene genealogies in two species: Probabilities of monophyly, paraphyly, and polyphyly in a coalescent model Evolution 2003 57 1465 1477 12940352
Neigel JE Avise JC Nevo E Karlin S Phylogenetic relationships on mitochondrial DNA under various demographic models of speciation Evolutionary processes and theory 1986 New York Academic Press 515 534
Moritz C Cicero C DNA barcoding: Promise and pitfalls PLoS Biol 2004 2 e354 10.1371/journal.pbio.0020354 15486587
Prendini L Comment on “Identifying spiders through DNA barcodes” Can J Zool 2005 83 498 504
Meyer CP Molecular systematics of cowries (Gastropoda: Cypraeidae) and diversification patterns in the tropics Biol J Linn Soc Lond 2003 79 401 459
Meyer CP Toward comprehensiveness: Increased molecular sampling within Cypraeidae and its phylogenetic implications Malacologia 2004 46 127 156
Kirkendale LA Meyer CP Phylogeography of the Patelloida profunda group (Gastropoda: Lottidae): Diversification in a dispersal-driven marine system Mol Ecol 2004 13 2749 2762 15315686
Meyer CP Geller JB Paulay G Fine scale endemism on coral reefs: Archipelagic differentiation in turbinid gastropods Evolution 2005 59 113 125 15792232
Schilder FA Schilder M Prodrome of a monograph on living Cypraeidae Proc Malacol Soc Lond 1938 22–23 119 231
Schilder M Schilder FA A catalogue of living and fossil cowries. Taxonomy and bibliography of Triviacea and Cypraeacea, Gastropoda Prosobranchia Memoires Institut Royal des Sciences Naturelles de Belgique, Deuxième Série 1971 85 1 246
Burgess CM Cowries of the world 1985 Cape Town (South Africa) G. Verhoef Seacomber Publications 288
Lorenz F Hubert A A guide to worldwide cowries 1993 Wiesbaden (Germany) C. Hemmen 571
Lorenz F New worldwide cowries 2002 Hackenheim (Germany) ConchBooks 292
Zwickl DJ Hillis DM Increased taxon sampling greatly reduces phylogenetic error Syst Biol 2002 51 588 598 12228001
Bensasson D Zhang DX Hartl DL Hewitt GM Mitochondrial pseudogenes: Evolution's misplaced witnesses Trends Ecol Evol 2001 16 314 321 11369110
Williams ST Knowlton N Mitochondrial pseudogenes are pervasive and often insidious in the snapping shrimp genus Alpheus
Mol Biol Evol 2001 18 1484 1493 11470839
Tajima F Evolutionary relationship of Dna-sequences in finite populations Genetics 1983 105 437 460 6628982
Johnson NK Cicero C New mitochondrial DNA data affirm the importance of Pleistocene speciation in North American birds Evolution 2004 58 1122 1130 15212392
Paulay G Meyer CP Diversification in the tropical Pacific: Comparisons between marine and terrestrial systems Am Zool 2001 41 1550 1551
Matz MV Nielsen R A likelihood ratio test for species membership based on DNA sequence data Philos Trans R Soc Lond B Biol Sci 2005 360 1969 1974 16214754
Dayrat B Towards integrative taxonomy Biol J Linn Soc Lond 2005 85 407 415
de Queiroz K Howard DJ Berlocher SH The general lineage concept of species, species criteria, and the process of speciation: A conceptual unification and terminological recommendations Endless forms: Species and speciation 1998 Oxford University Press pp 57
Swofford DL PAUP*: Phylogenetic Analysis Using Parsimony (*and Other Methods), version 4.0b10 [computer program] 2002 Sunderland (Massachusetts) Sinauer
|
16336051
|
PMC1287506
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e422
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030422
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1629298310.1371/journal.pbio.0030423Research ArticlePhysiologyMus (Mouse)MammalsVertebratesAnimalsHomo (Human)Ankyrin-B Coordinates the Na/K ATPase, Na/Ca Exchanger, and InsP3 Receptor in a Cardiac T-Tubule/SR Microdomain Ankyrin-B-Based Complex of Channels/TransportersMohler Peter J [email protected]
1
Davis Jonathan Q
2
Bennett Vann [email protected]
2
1Department of Pathology, Vanderbilt University, Nashville, Tennessee, United States of America,2Howard Hughes Medical Institute and Departments of Cell Biology, Biochemistry, and Neurosciences, Duke University Medical Center, Durham, North Carolina, United States of AmericaBenjamin Ivor Academic EditorUniversity of Utah Health Sciences CenterUnited States of America12 2005 29 11 2005 29 11 2005 3 12 e4235 7 2005 12 10 2005 Copyright: © 2005 Mohler 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.
Ankyrin-B Binds Disparate Proteins to Keep Calcium Flowing in the Heart
We report identification of an ankyrin-B-based macromolecular complex of Na/K ATPase (alpha 1 and alpha 2 isoforms), Na/Ca exchanger 1, and InsP3 receptor that is localized in cardiomyocyte T-tubules in discrete microdomains distinct from classic dihydropyridine receptor/ryanodine receptor “dyads.” E1425G mutation of ankyrin-B, which causes human cardiac arrhythmia, also blocks binding of ankyrin-B to all three components of the complex. The ankyrin-B complex is markedly reduced in adult ankyrin-B+/− cardiomyocytes, which may explain elevated [Ca2+]i transients in these cells. Thus, loss of the ankyrin-B complex provides a molecular basis for cardiac arrhythmia in humans and mice. T-tubule-associated ankyrin-B, Na/Ca exchanger, and Na/K ATPase are not present in skeletal muscle, where ankyrin-B is expressed at 10-fold lower levels than in heart. Ankyrin-B also is not abundantly expressed in smooth muscle. We propose that the ankyrin-B-based complex is a specialized adaptation of cardiomyocytes with a role for cytosolic Ca2+ modulation.
The authors describe an ankyrin-B-based macromolecular complex localized in cardiomyocyte T-tubules, which the authors propose is a specialized adaptation of these cells with a role for cytosolic Ca2+ modulation.
==== Body
Introduction
Defects in Ca2+ homeostasis underlie major diseases of the heart including congestive heart failure, cardiac hypertrophy, and fatal cardiac arrhythmias [1,2]. Ca2+ ions enter cardiomyocytes through voltage-sensitive Ca2+ channels (dihydropyridine receptor [DHPR]) located in invaginations of the plasma membrane known as transverse tubules (T-tubules). DHPR is localized in a microdomain of the T-tubule that is synapsed with sites in the sarcoplasmic reticulum (SR) that are enriched in Ca2+-release channels (ryanodine receptor [RyR]; [3,4]). Ca2+ that enters through DHPR must be balanced in each contraction cycle (~100 ms in mouse) by Ca2+ export. Ca2+ export is accomplished primarily by the Na/Ca exchanger 1 (NCX1), which is driven by the transmembrane Na+ gradient provided by the Na/K ATPase (NKA) [5]. The requirement for rapid export of Ca2+ is a specialized feature of heart that is not present in skeletal muscle, where DHPR directly activates RyR without Ca2+ import.
Ca2+ export has historically been an important therapeutic target in the management of heart failure. Cardiac glycosides increase [Ca2+]i by inhibiting NKA, thus elevating [Na+]i and indirectly inhibiting Ca2+ efflux through NCX1 [6]. Several considerations suggest that NCX1 and NKA operate together in diffusion-limited physiological spaces. Cardiac glycosides do not elevate averaged cytoplasmic [Na+] to levels sufficient to inhibit NCX1, suggesting NKA and NCX are coupled [6,7]. Arrhythmia caused by cardiac glycoside toxicity is believed to result from a transient inward current carried by NCX1 [8]. However, NCX1 would not be expected to generate inward flux of Na+ if averaged [Na+] was elevated sufficiently to inhibit Ca+2 efflux. Numerous studies support a tight functional coupling between NCX1, NKA, and intracellular Ca2+ stores in heart [9–13]. However, while co-localization between NCX1, NKA, and intracellular Ca2+ stores is described in smooth muscle [14], the relative localizations of these proteins in heart are undefined. Additionally, no biochemical evidence exists for a direct link between NKA, NCX1, and other SR proteins in cardiomyocytes.
Ankyrin-B is a multivalent adapter present in cardiomyocytes that binds individually to NCX1, NKA, and inositol 1,4,5-trisphosphate receptors (InsP3Rs), and potentially could play a role in functional coupling of these proteins [15–17]. Loss-of-function mutations in ankyrin-B cause a dominantly inherited human cardiac arrhythmia syndrome associated with sudden cardiac death [15,18]. Mice heterozygous for a null mutation in ankyrin-B (ankyrin-B+/− mice) are haploinsufficient and have a similar cardiac phenotype as humans heterozygous for loss-of-function mutations of ankyrin-B [15,18]. Adult ankyrin-B+/− cardiomyocytes exhibit elevated [Ca2+]i transients and, in the presence of beta-catecholamines, exhibit delayed and early afterdepolarization events and extrasystoles [15]. In contrast, action potential duration, inward Ca2+ current, and diastolic [Ca2+] are normal in ankyrin-B+/− cardiomyocytes [15].
We report here that NKA (alpha 1 and alpha 2 isoforms), NCX1, and InsP3R are complexed with ankyrin-B within a microdomain of cardiomyocyte T-tubules, and that the complex is deficient in ankyrin-B+/− cardiomyocytes. Additionally, we present evidence that loss of the ankyrin-B-based complex is the molecular defect in cardiac arrhythmia due to ankyrin-B mutation in humans and mice. The ankyrin-B complex is not present in skeletal muscle, smooth muscle, or brain, and may have evolved in the context of specialized requirements for cytosolic Ca2+ regulation in cardiomyocytes.
Results
Ankyrin-B Is Required for T-Tubule Localization of NKA, NCX1, and InsP3R in Cardiomyocytes
Ankyrin-B as well as NKA, NCX1, and InsP3R are selectively lost from Z-line/T-tubule sites in haploinsufficient ankyrin-B+/− cardiomyocytes [15]. The localization of these proteins was further resolved by three-dimensional rendering of consecutive confocal Z-sections of cardiomyocytes labeled by immunofluorescence (Figure 1). Wild-type ankyrin-B is organized in an intracellular tubular lattice in parallel with both the M-line and Z-line/T-tubules, but also including perpendicular axial branches that connect M-line and Z-line staining (Figure 1C). Z-line ankyrin-B staining is co-linear with the dyad marker DHPR in two-dimensional images, but is distinct from the DHPR in three dimensions (<2% of DHPR-positive voxels [three-dimensional (3D) pixels] overlap with ankyrin-B-positive voxels; not shown). Z-line ankyrin-B staining significantly overlaps in three dimensions with InsP3R (~45% of InsP3R-positive voxels co-localize with ankyrin-B-positive voxels; Figure 1D) as well as T-tubule-associated NCX1 (~53%; Figure 1E) and NKA (~51%; Figure 1F).
Figure 1 NCX1, NKA, and InsP3R Localization Require Ankyrin-B
Boxes in differential interference contrast images (A and B) represent sites that were imaged in (C–F). Immunolocalization of ankyrin-B (red) and (C) DHPR (green), (D) InsP3R (green), (E) NCX1 (green), and (F) NKA (green) in wild-type (left) and ankyrin-B+/− cells (right). Ankyrin-B+/− myocytes were labeled and imaged using identical protocols. M-lines (middle of A-band) are denoted by arrowheads; T-tubules over I-bands are denoted by arrows. Scale bar = 2 μm. Distance between arrowheads is ~1.8 μm. Images represent at least twenty Z-sections at 0.18-μm intervals. In (C), perpendicular axial branches are denoted by white arrowheads.
In contrast to wild-type cardiomyocytes, ankyrin-B+/− cells lack Z-line staining as well as the axial lattice of ankyrin-B that connects Z-line- and M-line-associated populations (Figure 1C–1F, right panels). DHPR staining and organization is unaffected in ankyrin-B+/− cardiomyocytes. InsP3R immunofluorescence (Figure 1D), as well as NCX1 and NKA isoform staining (Figure 1E and 1F) are markedly reduced at T-tubule sites. Residual InsP3R, NKA, and NCX1 rarely co-localize with ankyrin-B (levels of InsP3R, NKA, and NCX1 are reduced to levels too low for accurate determination of overlap).
Ankyrin-B-Coupled NKA, NCX1, and InsP3R Are Co-Localized
The Na/Ca exchanger has been localized at sites on cardiac T-tubules distinct from DHPR and RyR [19]. Moreover, NKA isoforms, NCX1, and InsP3R have been individually localized at T-tubule sites in cardiomyocytes [15,19–21]. However, localizations of all three proteins relative to each other have not been addressed. The relative localization of NCX1 with NKA alpha 1 and alpha 2 isoforms, InsP3R, DHPR, and RyR was evaluated using multiple combinations of double-labeled 3D images of mouse cardiomyocytes generated using confocal Z-stacks (Figure 2; Table 1). As expected, DHPR and RyR are co-localized, with ~55% voxel overlap of DHPR with RyR and ~42% overlap of RyR with DHPR (Figure 2B; Table 1), values comparable to voxel overlaps observed previously [19]. T-tubule NCX1 immunofluorescence (also localized at the sarcolemma [21]) is distinct from DHPR staining, with a voxel overlap of less than 5%, as reported previously [19] (Figure 2C; Table 1).
Figure 2 The Ankyrin-B-Based Complex of NKA, NCX1, and InsP3R Is Localized in a Specialized T-Tubule Microdomain of Cardiomyocytes
Adult cardiomyocytes were labeled with indicated antibodies. 3D reconstructions of each cell (A, differential interference contrast) are shown for each combination. Staining pairs include (B) DHPR (red) and RyR (green), (C) DHPR (red) and NCX (green), (D) NCX (red) and NKA alpha1 (green), (E) NCX (red) and NKA alpha 2 (green), (F) NKA alpha 1 (red) and NKA alpha 2 (green), (G) InsP3R (red) and RyR (green), and (H) InsP3R (red) and NCX (green). Voxel co-localization was performed for non-sarcolemmal voxels. T-tubule/Z-lines are indicated by white arrows. Scale bar = 5 μm.
Table 1 Analyses of the Extent of Voxel Co-Localization for Select Cardiac Protein Pairs
A new finding is that T-tubule NCX1 immunofluorescence co-localizes in submicron-sized domains with the T-tubule population of both NKA alpha 1 (Figure 2D; Table 1) and alpha 2 isoforms (Figure 2E; Table 1), with 55% (alpha 1) and 50% (alpha 2) voxel overlap. T-tubule NCX1 also co-localizes with InsP3R (Figure 2H; Table 1), with 56% of InsP3R voxels overlapping NCX1 and 50% of NCX1 voxels overlapping InsP3R. In contrast, InsP3R and RyR are not co-localized at T-tubule sites, with less than 5% voxel overlap (Figure 2G; Table 1). This is the first direct evidence to our knowledge that RyR and InsP3R are localized in spatially defined compartments of the endoplasmic reticulum and SR of ventricular cardiomyocytes.
NKA alpha 1 and alpha 2 isoforms were similarly distributed over the T-tubule (Figure 2F; Table 1) and sarcolemma (not shown). The T-tubule co-localization occurred in three dimensions as voxels with alpha 1 signal containing alpha 2 signal ~52% of the time, while voxels containing alpha 2 signal contained alpha 1 signal ~46% of the time. NKA isoforms have been proposed to have unique functions based on differences in localization and/or affinity for cardiac glycosides [22]. These differences may depend on the species and cell type, and have not been reported in mouse cardiomyocytes [23]. Our results in mouse ventricular cardiomyocytes suggest no major differences in NKA alpha 1 and alpha 2 localization at T-tubules (Figure 2) or sarcolemma (not shown). These results demonstrate co-clustering of NCX1, NKA, and InsP3R within microdomains along the T-tubule/SR that are distinct from classic T-tubule/SR junctions populated by DHPR and RyR. Moreover, the clusters of these proteins as well as ankyrin-B are reduced or absent in ankyrin-B+/− cardiomyocytes.
T-Tubule-Associated Ankyrin-B Is a Specialized Adaptation of Cardiomyocytes
The expression of 220-kDa ankyrin-B in skeletal muscle is nearly 10-fold lower than in heart (Figure 3A). Moreover, in contrast to ankyrin-B localization in cardiomyocytes (Figure 3B), ankyrin-B is not present over T-tubules of skeletal muscle, but instead is concentrated at punctate sites on the sarcolemma, over the A-band, and at costameres (Figure 3B). Additionally, in contrast to cardiac muscle, NCX1 and NKA isoforms are nearly undetectable over T-tubules of skeletal muscle, but are instead concentrated at the sarcolemma (Figure 3C). Finally, ankyrin-B expression is nearly absent from smooth muscle (Figure S1), and ankyrin-B-based complexes of NCX1, NKA isoforms, and InsP3R are not detectable in brain (see below). Therefore, the ankyrin-B-based complex of NKA, NCX1, and InsP3R is a specialized feature of cardiac myocytes.
Figure 3 The Ankyrin-B-Based Complex Is a Specialized Feature of Heart
(A) Relative expression of 220-kDa ankyrin-B in adult mouse heart and skeletal muscle (n = 3, p < 0.05). For the immunoblot in the left panel, 50 μg of total lysate was analyzed.
(B) Ankyrin-B (AnkB) immunostaining (red) in frozen sections of heart (H) and skeletal muscle (Sk Mus, SkM). The Z-lines (arrowheads, also asterisks in right skeletal muscle panel) overlap with alpha-actinin (green). Arrows indicate A-bands. Ankyrin-B is localized at costameres in skeletal muscle, visualized as staining continuous with alpha-actinin near the sarcolemma, but does not extend into the interior. Ankyrin-B localizes over the A-band in both heart and skeletal muscle. Scale bars = 5 μm.
(C) NCX1 (left panel), NKA alpha 1 (middle panel), and NKA alpha 2 immunostaining (right panel) in skeletal muscle. Scale bars = 18 μm.
Ankyrin-B Coordinates NKA, NCX1, and InsP3R in a Macromolecular Protein Complex in Cardiomyocytes
The finding that ankyrin-B is co-localized with NKA, NCX1, and InsP3R in cardiac T-tubule microdomains and that all of these proteins are coordinately reduced in ankyrin-B+/− cardiomyocytes raises the question of their molecular organization. Given previous evidence that ankyrin-R can form heterocomplexes between two ankyrin-binding proteins [24], we wondered whether ankyrin-B could form a multi-protein complex involving NKA, NCX1, and InsP3R in cardiomyocytes. We therefore performed a series of immunoprecipitations from detergent extracts of mouse heart with antibodies against NCX1, alpha 1 and alpha 2 isoforms of NKA, and InsP3R, followed by immunoblots to detect associated proteins (Figure 4). As reported previously, ankyrin-B antibody co-immunoprecipitated NKA isoforms, NCX1, and InsP3R, but not DHPR, SR Ca2+ ATPase (SERCA2), or calsequestrin (Figure 4A) [15]. NCX1 antibody co-immunoprecipitated 220-kDa ankyrin-B as well as NKA alpha 1 and alpha 2 isoforms and InsP3R; DHPR, SERCA2, and calsequestrin were not co-immunoprecipitated (Figure 4). Moreover, NKA alpha 1– and alpha 2–specific antibodies co-immunoprecipitated 220-kDa ankyrin-B as well as NCX1 and InsP3R, but, again, not DHPR, SERCA2, or calsequestrin (Figure 4A). Finally, antibody specific for InsP3R also co-immunoprecipitated 220-kDa ankyrin-B along with NKA alpha 1 and alpha 2 and NCX1, but not DHPR, SERCA2, or calsequestrin. DHPR-, SERCA2-, and calsequestrin-specific antibodies did not co-immunoprecipitate 220-kDa ankyrin-B, NCX1, InsP3R, or NKA isoforms. These mutual co-immunoprecipitations provide evidence for a macromolecular protein complex in heart containing ankyrin-B coupled to alpha 1 and alpha 2 isoforms of NKA, NCX1, and InsP3R. While other proteins may be in this protein complex, components of the classic T-tubule/SR junction (DHPR [also RyR; not shown]) as well as components of the SR (SERCA2 and calsequestrin) are not included.
Figure 4 Ankyrin-B Forms a Macromolecular Complex with NKA, NCX1, and InsP3R That Is Missing in Ankyrin-B+/− Heart
(A) Immunoprecipitations and co-immunoprecipitations from detergent-soluble extracts from adult mouse heart.
(B and C) Detergent-soluble lysates from wild-type or ankyrin-B+/− mouse hearts were used for immunoprecipitations with indicated antibodies (IB, immunoblot; IP, immunoprecipitation). Immunoprecipitations of ankyrin-B+/− extracts employed doubled amounts of input lysate to compensate for 50% reduction of ankyrin-B.
(D) InsP3R co-immunoprecipitates 220-kDa ankyrin-B, NCX1, and NKA from detergent-soluble heart lysates (Input = 10%). In contrast, InsP3R co-immunoprecipitates 220-kDa ankyrin-B, but not NCX1 or NKA, from detergent-soluble lysates of mouse brain (Input = 10%).
ANKB, ankyrin-B; C. IG, control Ig.
We next asked whether ankyrin-B was required for mutual co-immunoprecipitation of NKA, NCX1, and InsP3R by comparing wild-type hearts and ankyrin-B+/− hearts, which are deficient in ankyrin-B (Figure 4B and 4C). Ankyrin-B+/− hearts express reduced levels of 220-kDa ankyrin-B (decreased ~50%), NKA alpha 1 and alpha 2 (both reduced ~15%), NCX1 (reduced ~16%), and InsP3R (reduced ~33%) [15]. Strikingly, ankyrin-B+/− heart lysates exhibited over 60% loss of the ability of ankyrin-B antibody to co-immunoprecipitate NKA, InsP3R, or NCX1, even when the quantity of lysate was doubled to equalize the starting amount of ankyrin-B (Figure 4B). Moreover, a similar reduction in NCX1 co-immunoprecipitation of Na/K pump isoforms and InsP3R occurred using doubled ankyrin-B+/− lysates (Figure 4C). NKA alpha 1 and alpha 2 antibody also failed to co-immunoprecipitate a significant fraction of NCX1 or InsP3R from ankyrin-B+/− doubled lysates. Finally, InsP3R antibody immunoprecipitated minimal levels of NCX1 or NKA isoforms from ankyrin-B+/− heart (Figure 4C). These results demonstrate that a specialized population of ankyrin-B, which is reduced in ankyrin-B+/− heart, is critical for ankyrin-B interactions with NKA alpha 1 and alpha 2, NCX1, and InsP3R.
Ankyrin-B, NKA, InsP3R, and NCX1 are all expressed in brain at levels comparable to those in heart tissue. However, while immunoprecipitation of 100,000g detergent extracts of brain tissue with antibody against the InsP3R co-immunoprecipitated ankyrin-B, NKA and NCX1 were not present (Figure 4D). Moreover, NKA and NCX1 were also missing when the immunoprecipitation was performed with antibody against ankyrin-B (not shown). These results are in contrast to the report that NCX1, NKA, InsP3R, and ankyrin-B co-immunoprecipitate along with several other proteins from 27,000g supernatants of detergent extracts from brain [25]. The difference could result from use of a 27,000g supernatant in the other study and a 100,000g supernatant in our experimental protocol. A major complication with a lower speed supernatant is the likely presence of large complexes such as those connected by short actin filaments that would be removed with more centrifugation. Our results demonstrate that co-expression of ankyrin-B with NCX1 and NKA in the same tissue is not sufficient for formation of a complex from a 100,000g supernatant.
Reconstitution of an Ankyrin-B-Based Complex of NKA, NCX1, and InsP3R
Co-immunoprecipitation experiments as presented in Figure 4 provide evidence for interactions between ankyrin-B and its partners in vivo. We next evaluated whether ankyrin-B could form a complex with the NKA, NCX1, and InsP3R in vitro using purified proteins (See Materials and Methods). We first confirmed, using 125I-labeled proteins and immobilized ankyrin-B membrane-binding domain, that purified ankyrin-B membrane-binding domain directly interacts in vitro with purified NCX1 expressed in Sf9 cells (K
d = 5 nM), purified NKA from kidney (K
d = 50 nM), and purified InsP3R from cerebellum (K
d = 3 nM) (Figure 5).
Figure 5 Purified NCX1, NKA, and InsP3R Directly Interact with Purified Ankyrin-B Membrane-Binding Domain
Saturation binding of purified (A) 125I-labeled NCX1, (B) 125I-labeled NKA (NKA), and (C) 125I-labeled InsP3R with glutathione Sepharose–immobilized GST-ankyrin-B membrane-binding domain or GST control Sepharose. Inset for each panel represents Scatchard analysis of ankyrin-B interactions following subtraction of nonspecific binding to GST-Sepharose. Proteins were isolated as described in Materials and Methods.
We next asked whether ankyrin-B could form a multivalent complex with these proteins. Association of InsP3R with NKA and NCX1 in the presence or absence of soluble ankyrin-B membrane-binding domain (purified as a GST-fusion protein and then cleaved from the GST-tag; Figure 6A) was assessed using biotinylated InsP3R bound to neutravidin-Dynabeads (Figure 6B) and 125I-labeled NCX1 and NKA. 125I-labeled NCX1 and 125I-labeled NKA associated with InsP3R-coated beads only in the presence of ankyrin-B membrane-binding domain (Figure 6C). In fact, while the intensity of the NCX1 band is ~50% of the band intensity of NKA (not shown), the picomoles of each protein bound to the InsP3R-coated beads was approximately equal (Figure 6C; specific activity of 125I-NKA ~504,000 cpm; 125I-NCX1 ~270,000 cpm). In contrast, 125I-labeled NCX1 and 125I-labeled NKA failed to bind to InsP3R-coated beads in the absence of ankyrin-B (Figure 6C). Additionally, in the presence of ankyrin-B, InsP3R-coated beads simultaneously associated with both 125I-labeled NCX1 and 125I-labeled NKA with no decrease in binding capacity compared to reactions where only one labeled protein was used (Figure 6C). These results demonstrate that interaction of InsP3R with either NCX1 or NKA is ankyrin-B-dependent and that these proteins can assemble in vitro in the absence of additional co-factors or regulatory proteins.
Figure 6 Reconstitution of the NCX1, NKA, and InsP3R Complex Requires Ankyrin-B
(A) Purified NKA (α/β subunits), InsP3R, NCX1, and ankyrin-B membrane-binding domain (ANKB MBD) were examined by SDS-PAGE and Coomassie blue.
(B) Stained gel of control beads and beads plus purified InsP3R.
(C) NKA and NCX1 were 125I-labeled and incubated with control Dynabeads or Dynabeads coated with InsP3R with or without ankyrin-B membrane-binding domain. Bound protein was analyzed by a gamma-counter (C) and by SDS-PAGE and phosphorimaging (not shown; see Materials and Methods for protein measurement, n = 3, p < 0.05).
Human E1425G Mutation Abolishes Ankyrin-B Association with NCX1, NKA, and InsP3R
One test of the physiological importance of the ankyrin-B-based complex is whether mutations in ankyrin-B resulting in loss of the complex also cause arrhythmia. E1425G mutation of ankyrin-B causes human cardiac arrhythmia and loss of activity of ankyrin-B in restoring normal Ca2+ waves to ankyrin-B+/− neonatal cardiomyocytes [15]. The mechanism for loss of function due to the E1425G mutation, which is located close to the C-terminal regulatory domain and distant from the membrane-binding domain (Figure 7A), is not known. However, the regulatory domains of ankyrins (Figure 7A) can modulate activities of N-terminal membrane- and spectrin-binding domains [26,27].
Figure 7 Human Ankyrin-B E1425G Mutation Abolishes Binding to NKA, NCX1, and InsP3R
(A) 220-kDa ankyrin-B domain organization. The human LQT4 E1425G mutation is marked.
(B) Immunoblot of expressed wild-type and mutant GFP-ankyrin-B proteins.
(C–E) Relative binding of rat heart lysate (C) InsP3R, (D) NCX1, and (E) NKA to wild-type and mutant GFP-ankyrin-B proteins. Bound NCX1, NKA, and InsP3R were evaluated following quantitative immunoblot (pan-InsP3R, pan-NKA, and NCX1 Ig) and phosphorimaging (n = 3, p < 0.05).
The effect of the E1425G mutation on the ability of ankyrin-B to bind to NKA, NCX1, and InsP3R was evaluated using detergent extracts of heart tissue (not shown) and using purified proteins isolated as in Figure 6A. Evaluation of the binding properties of the E1425G mutant protein requires full-length 220-kDa ankyrin-B. We have not yet successfully generated full-length 220-kDa ankyrin-B in bacteria. Therefore, we used mammalian HEK293 cells to generate full-length wild-type and mutant ankyrin-B polypeptides for our binding studies. Recombinant green fluorescent protein (GFP)–220-kDa ankyrin-B that was either wild-type, with the E1425G mutation, or with a E1425D mutation was expressed and immuno-isolated from HEK293 cells using an affinity-purified antibody against GFP immobilized on Protein A agarose.
The levels of immobilized GFP-ankyrins were all equivalent in these assays (Figure 7B). E1425G ankyrin-B exhibited a 60%–70% loss of association with NCX1, NKA alpha 1 and alpha 2, and InsP3R from cardiac lysates (not shown), and as pure proteins (Figure 7C–7E). The conservative E1425D mutation at this site had no effect on binding of NCX1, NKA, or InsP3R (Figure 7C–7E). The finding that the E1425G mutation abolishes the ability of ankyrin-B to bind to NCX1, NKA, and InsP3R (Figure 7) suggests that interaction of ankyrin-B with either all three or some combination of these proteins is required for its function. It is possible that the E1425G mutation affects other protein interactions of ankyrin-B, although these remain to be identified.
Discussion
This study presents the discovery of an ankyrin-B-based macromolecular complex of NKA (alpha 1 and alpha 2 isoforms), NCX1, and InsP3R in cardiomyocytes. The complex is localized in a microdomain along cardiomyocyte T-tubules resolved by 3D confocal microscopy as distinct from the classic dyad formed by DHPR and RyR. This microdomain was first described by Moore and colleagues, who also distinguished the T-tubule NCX1 from dyad proteins RyR and DHPR and from voltage-gated sodium channels by light microscopy using image deconvolution and wide-field epifluorescence microscopy [19]. T-tubule-associated ankyrin-B is a specialized adaptation of cardiomyocytes and is not evident in smooth muscle, which does not express significant ankyrin-B levels, or in skeletal muscle, where ankyrin-B is expressed at 10-fold lower levels than in heart. The T-tubule domain containing ankyrin-B-coupled NKA, NCX1, and InsP3R thus is a specialized adaptation of cardiac cells that is not present in other types of muscle cells.
We propose a scale model for the ankyrin-B-based complex (Figure 8) based on previous structural reports and on evidence from this study that ankyrin-B can promote association between purified NKA, NCX1 and InsP3R (see Figure 6). In this scheme, the extended ankyrin-B membrane-binding domain adapts the NKA and NCX1 to the InsP3R in a configuration that would allow for regulation of cytosolic Ca2+ in a spatially privileged domain (Figure 8). It is likely that all participants in such an assembly have mutually interacting surfaces. In this case, the role of ankyrin-B could be to stabilize the assembly and/or possibly direct its cellular localization. The resulting macromolecular complex capable of coupled transport would accomplish the intended purpose of “restricted space” previously invoked to explain the action of cardiac glycosides [28]. However, the dimensions of a complex would be on the order of 10–20 nm, while an anatomical space or “synapse” between the endoplasmic reticulum and plasma membrane is 500–1,000 nm in size and would not provide a effective barrier to diffusion of small ions with radii less than 1 nm. A test of the idea of coupled transport by ankyrin-B-complexed proteins would be to selectively interfere with participation of individual members of the complex by knocking in mutants lacking ankyrin-B-binding activity.
Figure 8 Model of the Cardiac Ankyrin-B Complex
(Left) Model of ankyrin-B-dependent complex of NKA, NCX1, and InsP3R at T-tubule/SR sites distinct from the classic “dyad.”
(Right) Scale model of ankyrin-B complex based on approximate dimensions of represented proteins. The protein ratios are not representative for in vivo couplings as there are likely 50–100 RyR2 for each InsP3R in a ventricular cardiomyocyte [32].
While binding and localization data are consistent with simultaneous interaction of a single ankyrin-B molecule with NCX1, NKA, and InsP3 receptor, it also is possible that only one or two ankyrin-B-associated proteins are bound at a given time. It will be important in future experiments to isolate ankyrin-B-based macromolecular assemblies and directly determine stoichiometries of component proteins. A current challenge is that ankyrins also associate with spectrin and spectrin/actin complexes (reviewed in [17]), as well as proteins such as obscurin [29,30].
A role for InsP3R in heart is unknown. InsP3R-dependent Ca2+ signaling has been proposed to regulate excitation–contraction coupling in atrial myocytes by modulation of the activity (priming) of juxtaposed RyR [31]. However, based on the low ratio of InsP3R to RyR [32], the high Ca2+ buffering capacity of the cytosol [33], and now the distinct localizations of these Ca2+-release channels, it is unlikely that InsP3R Ca2+ release could affect the activity of RyR-mediated Ca2+-induced Ca2+ release in ventricular cardiomyocytes. Also, a role of InsP3R in Ca2+ signaling is difficult to reconcile with an environment where [Ca2+]i transients occur continuously [2]. Our model suggests a counterintuitive role for InsP3R as a “Ca2+ pressure valve” for export of excess SR Ca2+ from the cell (Figure 8). Consistent with this idea is experimental evidence for functional coupling of SR Ca2+ stores with Ca2+ efflux [9,11].
Loss of the ankyrin-B-based complex may provide an explanation for the cardiac arrhythmia syndrome due to ankyrin-B mutations in humans and mice. The E1425G mutation of ankyrin-B, which causes human cardiac arrhythmia, also blocks binding of ankyrin-B to all three components (NCX1, NKA, and InsP3R) of the complex (see Figure 7). Moreover, ankyrin-B+/− mice have a related cardiac arrhythmia, and ankyrin-B+/− cardiomyocytes are also deficient in the complex, while the expression and subcellular localization of other cardiac ion channels and transporters (e.g., Nav channels, which associate with a second ankyrin gene product, ankyrin-G) remain normal [15,18,34]. The electrical basis for ankyrin-B-dependent cardiac arrhythmia has been proposed, based on observations with ankyrin-B+/− cardiomyocytes, to be due to elevated Ca2+ transients that provoke afterdepolarizations and extrasystoles following catecholamine-induced stress [15]. These predictions of a calcium-based phenotype are also supported by absence of abnormalities in the localization or expression of Nav channels (also normal cardiac action potentials) and K channels in ankyrin-B+/− and ankyrin-B−/− cardiomyocytes [15,18]. Absence of the ankyrin-B complex would be predicted to result in less efficient export of calcium from the SR and could result in elevated calcium transients.
Ankyrin-B+/− cardiomyocytes display preferential loss of ankyrin-B immunoreactivity at Z-line/T-tubule domains compared with M-line staining (see Figure 1). Potential explanations for this preferential loss may include reduced T-tubule ankyrin-B protein stability (half-life), increased T-tubule/SR membrane turnover, or differences in the association of each ankyrin-B population with the underlying cytoskeleton. Alternatively, reduced expression of the T-tubule/SR population of ankyrin-B in ankyrin-B+/− cardiomyocytes may result from differences in the molecular identities of ankyrin-B polypeptides at each domain. For example, ankyrin-B immunoreactivity at the M-line may represent an ankyrin-B splice form that lacks Ank2 exon 23 (exon targeted in the ankyrin-B knock-out mouse) but still reacts with ankyrin-B Ig.
Interaction between InsP3R and ouabain-associated Na/K pump has been reported to be responsible for slow Ca2+ oscillations in cultured renal proximal tubule and Cos7 cells [35]. Our results with pure proteins suggest that InsP3R and the NKA do not interact directly, at least not with high affinity (see Figure 7). Thus, it will be of interest to evaluate possible participation of ankyrin-B or possibly other adaptor proteins in this system. More generally, determinants of cellular localization and partnerships with physiologically related proteins likely are an essential aspect of function for all ion channels and transporters. Ankyrins are ubiquitously expressed and display diversity in protein interactions. Based on the findings of this study, and previous findings that ankyrin-G is required for coordinating voltage-gated Na channels and L1CAM cell adhesion molecules at axon initial segments [36,37], we predict that ankyrins are likely to contribute to higher order organization of multiple channels and transporters in a variety of tissues.
Materials and Methods
Animals
Mice used in these studies were adult WT C57BL/6 mice and ankyrin-B+/− littermates (C57BL/6), 3–6 mo of age and weighing 30–40 g. Animals were handled according to approved protocols and animal welfare regulations of the Institutional Review Board of Duke University Medical Center. Mouse ventricular cardiomyocytes were isolated as described in [15].
Immunofluorescence
Antibodies not described in [15] include NCX1 (Affinity Bioreagents, Golden, Colorado, United States; Swant, Bellinzona, Switzerland), alpha 1 and alpha 2 ATPase (Transduction), DHPR (ABR, Alomone), affinity-purified GFP polyclonal Ig, and affinity-purified pan-InsP3R polyclonal Ig generated against the C-terminus of mouse InsP3R (residues 2592 −2750). When unavoidable, mouse cells were immunostained with monoclonal antibodies that had been first affinity-purified. For these monoclonal antibodies, we confirmed that our staining was specific by control experiments in rat cells. Additionally, Alexa anti-mouse secondary antibodies were examined for background immunoreactivity. Adult cardiomyocytes were stained as described [15]. Isolated ventricular mouse cardiomyocytes were double-labeled and imaged in three dimensions by rendering of confocal Z-scans obtained at 0.18-μm increments near the center of isolated cells using a 100 power/1.45 NA objective (LSM 510, Zeiss, Oberkochen, Germany). LSM Z-stacks were transferred to Volocity software (Improvision, Lexington, Massachusetts, United States), and identical protocols were used for 3D rendering of WT and ankyrin-B+/− cells. Volocity Classification software or LSM 510 software was used to measure voxel or pixel co-localization. Data represent at least three separate experiments with at least five areas measured for each experiment. Areas measured do not include sarcolemmal membrane voxels. Using monoclonal and polyclonal antibody directed against the same protein, cardiac double-labeling, and voxel co-localization revealed that the maximal co-localization for the same protein was ~65% consistent with previous studies [19].
Ankyrin-B mutagenesis.
GFP–220-kDa ankyrin-B mutants E1425G and E1425D were created using site-directed mutagenesis. The mutated region was subcloned into a native GFP–220-kDa ankyrin-B plasmid, and the plasmid was completely sequenced to verify that no additional mutations were introduced.
Statistics
When appropriate, data were analyzed using a two-tailed Student's t-test, and values less than p < 0.05 were considered significant. Values are expressed as the mean ± standard deviation.
Protein modeling
3D protein structures for the model in Figure 8 were approximated based on published structures [38–43].
Immunoprecipitation and solubilization of heart proteins
Adult heart immunoprecipitations and quantitative immunoblotting were performed as described [15]. Briefly, adult mouse heart and brain were dissected and rinsed in PBS plus 0.32 M sucrose and 2 mM Na EDTA, flash frozen in liquid nitrogen, and ground into a fine powder. The powder was resuspended in 4 volumes of 50 mM Tris HCl (pH 7.35), 10 mM NaCl, 0.32 M sucrose, 5 mM Na EDTA, 2.5 mM Na EGTA, 1 mM PMSF, 1 mM 4-(2-aminoethyl) benzenesulfonylfluoride hydrochloride (AEBSF), 10 μg/ml leupeptin, and 10 μg/ml pepstatin using a Dounce homogenizer (Kimble/Kontes, Vineland, New Jersey, United States). The homogenate was centrifuged at 1,000g to remove nuclei. Triton X-100 and deoxycholate were added to the post-nuclear supernatant for final concentrations of 1.5% Triton X-100 and 0.75% deoxycholate. The lysate was pelleted at 100,000g for 1 h at 4 °C, and the supernatant was re-cleared at 100,000g for 1 h to remove residual large membranes or vesicles. The resulting supernatant was used for immunoprecipitation (see Figure 4) as described [16], or for binding experiments.
Binding studies
GFP–220-kDa ankyrin-B and mutants (E1425G and E1425D) were expressed in HEK293 cells and purified using affinity-purified GFP Ig coupled to Protein A agarose beads. Briefly, cells were lysed in above homogenization buffer plus 1.0% Triton X-100 and 0.5% deoxycholate. The extract was centrifuged at 100,000g, and the supernatant was incubated with GFP affinity-purified Ig coupled to Protein A sepharose. The beads were washed with homogenization buffer plus 1.0% Triton X-100. Purified proteins were incubated with 10 μg of affinity-purified GFP Ig or control Ig coupled to Protein A sepharose beads for 4 h at 4 °C. The beads were washed four times with homogenization buffer plus 1.0% Triton X-100. Protein bound to each mutant GFP–220-kDa ankyrin-B was eluted, analyzed by quantitative 125I-labeled Protein A immunoblot, normalized for relative GFP–ankyrin-B expression, and then compared to WT GFP–220-kDa ankyrin-B binding.
For in vitro complex reconstitution experiments, 0.5 ml of Dynabeads M-270 Epoxy (1 × 109) beads (Dynal Biotech, Brown Deer, Wisconsin, United States) were washed in PBS and incubated with 5 mg of neutravidin in PBS in a final volume of 1 ml for 48 h at 25 °C. Beads were then washed in BSA binding buffer (20 mM Hepes [pH 7.3], 50 mM NaCl, 1 mM Na EDTA, 0.1% Triton X-100, 1 mM sodium azide, and 5 mg/ml BSA). Purified InsP3R (0.5 ml; 200 μg/ml) was incubated with a 20-fold molar excess of NHS-LC-biotin (Pierce Biotechnology, Rockford, Illinois, United States) overnight at 4 °C. For control biotin, the same biotin was incubated overnight in PBS. Biotin-InsP3R was then dialyzed against binding buffer without BSA to remove unbound biotin. Then 50% of the neutravidin Dynabeads were pre-incubated with the control biotin for 2 h at 4 °C, washed, and resuspended in BSA binding buffer (control neutravidin Dynabeads). Neutravidin Dynabeads were incubated with the dialyzed biotin-InsP3R while control neutravidin Dynabeads were incubated with the same concentration of unlabeled InsP3R for 2 h at 4 °C. Dynabeads were then washed 3× in BSA binding buffer and used for binding experiments. Coated Dynabeads (InsP3R at 20 nM) were incubated with 125I-labeled NCX1 (20 nM final concentration; specific activity 274,000 cpm/pmol) and/or 125I-labeled NKA (20 nM final concentration; specific activity ~504,000 cpm/pmol) in the presence or absence of a pre-incubation with ankyrin-B membrane-binding domain (20 nM final concentration) in a final volume of 50 μl of BSA binding buffer (20 mM Hepes [pH 7.3]; 50 mM NaCl, 1 mM Na EDTA, 0.2% Triton X-100, 1 mM NaN3, and 5 mg/ml BSA). Following 4 h, the beads were washed in binding buffer minus BSA, and both pellet and supernatant samples were assayed for 125I in a gamma counter (n = 3). The samples were then examined by SDS-PAGE and phosphorimaging (n = 3). Values for picomoles bound of 125I-labeled NCX1 or 125I-labeled NKA in experiments where only one labeled ligand was used were calculated from counts of 125I-labeled protein bound and specific activity. Values for picomoles bound of 125I-labeled NCX1 and 125I-labeled NKA when two labeled proteins were incubated together (125I-labeled NCX1 + 125I-labeled NKA) were calculated by first determining a ratio of counts of 125I-labeled protein bound/band intensity for unique bands of 125I-labeled NCX or 125I-labeled NKA in single protein binding experiments (i.e., 125I-labeled NCX + ankyrin-B + InsP3R beads). The intensity of these same bands was measured in the gel lanes where the two proteins (i.e., 125I-labeled NCX + 125I-labeled NKA + ankyrin-B + InsP3R beads) were combined to determine the number of picomoles bound of each protein. We observed approximately equal picomoles of 125I-labeled NCX1 and 125I-labeled NKA bound to InsP3R beads when ankyrin-B was included in the binding reaction. However, because of the lower specific activity of 125I-labeled NCX1, the band intensity on the gel was approximately 50% of that of 125I-labeled NKA. Saturation binding was performed essentially as described in [16] but using glutathione beads. Briefly, ankyrin-B membrane-binding domain was purified as described in [16]. Increasing concentrations of 125I-labeled InsP3R, 125I-labeled NCX1, or 125I-labeled NKA were incubated for 2 h at 25 °C with glutathione Sepharose-immobilized GST-ankyrin-B membrane-binding domain or GST. The beads were washed and counted in a gamma counter. The data were corrected for nonspecific binding at each concentration by subtracting values obtained with GST-coated beads.
Protein purification
Full-length human NCX1 was cloned from a human heart library (Clontech, Palo Alto, California, United States) into pBacPak9 (Clontech) using standard molecular techniques. For purification of NCX1, a His-tag was engineered to the C-terminus. NCX1 was expressed in SF21 insect cells using a generated recombinant baculovirus. Cells were infected in monolayer cultures with a MOI of ten for 72 h at 27 °C. Cells were harvested and washed in PBS, and cell pellets were snap frozen and stored at −80 °C. All subsequent procedures were performed at 4 °C in the presence of protease inhibitors (100 μg/ml AEBSF, 100 μg/ml benzamidine, 30 μg/ml leupeptin, and 10 μg/ml pepstatin). Cells were syringed and sonicated in cell homogenization buffer (PBS, 1 mM Na EDTA, 1 mM DTT, and 1 mM sodium azide) to break the cell membranes, then centrifuged at 100,000g for 30 min to collect membranes. Cell membranes were pre-extracted with 20 mM CHAPS (pH 12) for 30 min followed by 20 mM PB (pH 7.3), 0.5 M NaCl, 0.5 M urea, 0.5% Triton X-100, and 0.5 mM beta mercaptoethanol. The cell residue was resuspended in extraction buffer (50 mM PB (pH 8.0), 0.3 M NaCl, 10 mM imidazole, 0.2% Triton X-100, 1 mM beta mercaptoethanol, 1 mM sodium azide, and 2% Sarkosyl) for 20 min. The extract was centrifuged at 100,000g for 1 h, and the supernatant collected and diluted 10-fold in buffer lacking Sarkosyl. The diluted extract was applied to a column of Ni-NTA Sepharose, washed with 10–20 column volumes of dilution buffer, and eluted with buffer plus 0.3 M imidazole. Peak fractions were pooled, adjusted to 10% glycerol, snap frozen, and stored at −80 °C. Sheep kidney NKA was isolated in membrane-bound form from outer medulla as previously described [44]. The NKA was extracted and purified as previously described [45]. The InsP3R was purified from frozen bovine brain cerebellum by a modification of published procedures [16,46]. All procedures were carried out at 4 °C in the presence of protease inhibitors (100 μg/ml AEBSF, 100 μg/ml benzamidine, 30 μg/ml leupeptin, and 10 μg/ml pepstatin). Cerebellum was homogenized using a polytron in five volumes (weight/volume) of homogenization buffer (10 mM Hepes [pH 7.3], 0.32 M sucrose, 2 mM EGTA, 1 mm DTT, and 1 mM sodium azide), and centrifuged at 2,000 rpm for 10 min. Membranes were then collected at 30,000g for 1 h. Membranes were prewashed in wash buffer (50 mM Tris HCl [pH 8.0], 1 mM Na EGTA, 1 mM DTT, and 1 mM Na azide), then resuspended to the homogenization volume with the wash buffer. InsP3R was extracted from the membranes by the addition of 2% final Triton X-100 for 30 min, and supernatants collected after centrifugation at 30,000g for 1 h. The extract was adjusted to 0.25 M NaCl and applied to a 50-ml heparin Sepharose column equilibrated in 0.25 M NaCl and 0.2% Triton X-100 extraction buffer. The heparin Sepharose was washed with ten column volumes of equilibration buffer, and then eluted with 0.5 M NaCl buffer. Peak fractions were pooled and dialyzed against ten volumes of column buffer lacking NaCl and 20 mM Tris HCl (pH 8.0). A precipitate formed after dialysis and was collected by centrifugation at 100,000g for 20 min. The pellet was resuspended in column buffer with the addition of 1.0 M NaCl and was re-centrifuged as above. The InsP3R released into the supernatant was then adjusted to 0.2 mM CaCl2 and 0.2 mM MnCl2, and applied to a 4-ml ConA Sepharose column. The column was washed in 20 column volumes of buffer, the elution started with the addition of 1 M mannose, the elution stopped, and the column allowed to sit in elution buffer overnight. The elution was continued the following day and fractions collected, aliquoted, snap frozen, and stored at −80 °C.
Supporting Information
Figure S1 Ankyrin-B-Based Complex Is a Specialized Feature of Cardiac Myocytes
Ankyrin-B is expressed in ventricular cardiomyocytes but not in smooth muscle lining large arteries. Image represents adult mouse heart immunostained with ankyrin-B-specific Ig.
(2.7 MB TIF).
Click here for additional data file.
Accession Number
The NCBI (http://www.ncbi.nlm.nih.gov/) accession number for ankyrin-B is NM_020977.
We gratefully acknowledge the following research support and grants: the Howard Hughes Medical Institute and a focused giving grant from Johnson and Johnson to VB. PJM is supported by a National Scientist Development Award from the American Heart Association.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. PJM, JQD, and VB conceived and designed the experiments. PJM and JQD performed the experiments. PJM, JQD, and VB analyzed the data and contributed reagents/materials/analysis tools. PJM and VB wrote the paper.
Citation: Mohler PJ, Davis JQ, Bennett V (2005) Ankyrin-B coordinates the Na/K ATPase, Na/Ca exchanger, and InsP3 receptor in a cardiac T-tubule/SR microdomain. PLoS Biol 3(12): e423.
Abbreviations
3Dthree-dimensional
AEBSF4-(2-aminoethyl) benzenesulfonylfluoride hydrochloride
DHPRdihydropyridine receptor
GFPgreen fluorescent protein
InsP3Rinositol 1,4,5-trisphosphate receptor
NCX1Na/Ca exchanger 1
NKANa/K ATPase
RyRryanodine receptor
SERCA2sarcoplasmic reticulum Ca2+ ATPase
SRsarcoplasmic reticulum
T-tubuletransverse tubule
==== Refs
References
Keating MT Sanguinetti MC Molecular and cellular mechanisms of cardiac arrhythmias Cell 2001 104 569 580 11239413
Marks AR Calcium and the heart: A question of life and death J Clin Invest 2003 111 597 600 12618512
Carl SL Felix K Caswell AH Brandt NR Ball WJ Immunolocalization of sarcolemmal dihydropyridine receptor and sarcoplasmic reticular triadin and ryanodine receptor in rabbit ventricle and atrium J Cell Biol 1995 129 672 682
Sun XH Protasi F Takahashi M Takeshima H Ferguson DG Molecular architecture of membranes involved in excitation-contraction coupling of cardiac muscle J Cell Biol 1995 129 659 671 7730402
Philipson KD Nicoll DA Sodium-calcium exchange: A molecular perspective Annu Rev Physiol 2000 62 111 133 10845086
Akera T Brody TM The role of Na+,K+-ATPase in the inotropic action of digitalis Pharmacol Rev 1977 29 187 220 150607
Arnon A Hamlyn JM Blaustein MP Ouabain augments Ca(2+) transients in arterial smooth muscle without raising cytosolic Na(+) Am J Physiol Heart Circ Physiol 2000 279 H679 H691 10924067
Bers DM Excitation-contraction coupling and cardiac contractile force 2001 Dordrecht Kluwer Academic Publishers 427
Gilbert JC Shirayama T Pappano AJ Inositol trisphosphate promotes Na-Ca exchange current by releasing calcium from sarcoplasmic reticulum in cardiac myocytes Circ Res 1991 69 1632 1639 1954683
Fujioka Y Matsuoka S Ban T Noma A Interaction of the Na+-K+ pump and Na+-Ca2+ exchange via [Na+]i in a restricted space of guinea-pig ventricular cells J Physiol 1998 509 457 470 9575295
Janiak R Lewartowski B Langer GA Functional coupling between sarcoplasmic reticulum and Na/Ca exchange in single myocytes of guinea-pig and rat heart J Mol Cell Cardiol 1996 28 253 264 8729058
Reuter H Henderson SA Han T Ross RS Goldhaber JI The Na+-Ca2+ exchanger is essential for the action of cardiac glycosides Circ Res 2002 90 305 308 11861419
Su Z Zou A Nonaka A Zubair I Sanguinetti MC Influence of prior Na+ pump activity on pump and Na+/Ca2+ exchange currents in mouse ventricular myocytes Am J Physiol 1998 275 H1808 H1817 9815089
Moore ED Etter EF Philipson KD Carrington WA Fogarty KE Coupling of the Na+/Ca2+ exchanger, Na+/K+ pump and sarcoplasmic reticulum in smooth muscle Nature 1993 365 657 660 8413629
Mohler PJ Schott JJ Gramolini AO Dilly KW Guatimosim S Ankyrin-B mutation causes type 4 long-QT cardiac arrhythmia and sudden cardiac death Nature 2003 421 634 639 12571597
Mohler PJ Davis JQ Davis LH Hoffman JA Michaely P Inositol 1,4,5-trisphosphate receptor localization and stability in neonatal cardiomyocytes requires interaction with ankyrin-B J Biol Chem 2004 279 12980 12987 14722080
Bennett V Baines AJ Spectrin and ankyrin-based pathways: Metazoan inventions for integrating cells into tissues Physiol Rev 2001 81 1353 1392 11427698
Mohler PJ Splawski I Napolitano C Bottelli G Sharpe L A cardiac arrhythmia syndrome caused by loss of ankyrin-B function Proc Natl Acad Sci U S A 2004 101 9137 9142 15178757
Scriven DR Dan P Moore ED Distribution of proteins implicated in excitation-contraction coupling in rat ventricular myocytes Biophys J 2000 79 2682 2691 11053140
Kieval RS Bloch RJ Lindenmayer GE Ambesi A Lederer WJ Immunofluorescence localization of the Na-Ca exchanger in heart cells Am J Physiol 1992 263 C545 C550 1514597
Frank JS Mottino G Reid D Molday RS Philipson KD Distribution of the Na(+)-Ca2+ exchange protein in mammalian cardiac myocytes: An immunofluorescence and immunocolloidal gold-labeling study J Cell Biol 1992 117 337 345 1373142
Blaustein MP Golovina VA Structural complexity and functional diversity of endoplasmic reticulum Ca(2+) stores Trends Neurosci 2001 24 602 608 11576675
Muller-Ehmsen J Juvvadi P Thompson CB Tumyan L Croyle M Ouabain and substrate affinities of human Na(+)-K(+)-ATPase alpha(1)beta(1), alpha(2)beta(1), and alpha(3)beta(1) when expressed separately in yeast cells Am J Physiol Cell Physiol 2001 281 C1355 C1364 11546674
Michaely P Bennett V Mechanism for binding site diversity on ankyrin. Comparison of binding sites on ankyrin for neurofascin and the Cl-/HCO3- anion exchanger J Biol Chem 1995 270 31298 31302 8537399
Lencesova L O'Neill A Resneck WG Bloch RJ Blaustein MP Plasma membrane-cytoskeleton-endoplasmic reticulum complexes in neurons and astrocytes J Biol Chem 2004 279 2885 2893 14593108
Hall TG Bennett V Regulatory domains of erythrocyte ankyrin J Biol Chem 1987 262 10537 10545 3038887
Davis LH Davis JQ Bennett V Ankyrin regulation: An alternatively spliced segment of the regulatory domain functions as an intramolecular modulator J Biol Chem 1992 267 18966 18972 1388161
Juhaszova M Blaustein MP Na+ pump low and high ouabain affinity alpha subunit isoforms are differently distributed in cells Proc Natl Acad Sci U S A 1997 94 1800 1805 9050859
Bagnato P Barone V Giacomello E Rossi D Sorrentino V Binding of an ankyrin-1 isoform to obscurin suggests a molecular link between the sarcoplasmic reticulum and myofibrils in striated muscles J Cell Biol 2003 160 245 253 12527750
Kontrogianni-Konstantopoulos A Jones EM Van Rossum DB Bloch RJ Obscurin is a ligand for small ankyrin 1 in skeletal muscle Mol Biol Cell 2003 14 1138 1148 12631729
Mackenzie L Bootman MD Laine M Berridge MJ Thuring J The role of inositol 1,4,5-trisphosphate receptors in Ca(2+) signalling and the generation of arrhythmias in rat atrial myocytes J Physiol 2002 541 395 409 12042347
Moschella MC Marks AR Inositol 1,4,5-trisphosphate receptor expression in cardiac myocytes J Cell Biol 1993 120 1137 1146 8382205
Berlin JR Bassani JW Bers DM Intrinsic cytosolic calcium buffering properties of single rat cardiac myocytes Biophys J 1994 67 1775 1787 7819510
Mohler PJ Rivolta I Napolitano C Lemaillet G Lambert S Nav1.5 E1053K mutation causing Brugada syndrome blocks binding to ankyrin-G and expression of Nav1.5 on the surface of cardiomyocytes Proc Natl Acad Sci U S A 2004 101 17533 17538 15579534
Miyakawa-Naito A Uhlen P Lal M Aizman O Mikoshiba K Cell signaling microdomain with Na,K-ATPase and inositol 1,4,5-trisphosphate receptor generates calcium oscillations J Biol Chem 2003 278 50355 50361 12947118
Jenkins SM Bennett V Developing nodes of Ranvier are defined by ankyrin-G clustering and are independent of paranodal axoglial adhesion Proc Natl Acad Sci U S A 2002 99 2303 2308 11842202
Zhou D Lambert S Malen PL Carpenter S Boland LM AnkyrinG is required for clustering of voltage-gated Na channels at axon initial segments and for normal action potential firing J Cell Biol 1998 143 1295 1304 9832557
Hamada K Terauchi A Mikoshiba K Three-dimensional rearrangements within inositol 1,4,5-trisphosphate receptor by calcium J Biol Chem 2003 278 52881 52889 14593123
Hebert H Purhonen P Vorum H Thomsen K Maunsbach AB Three-dimensional structure of renal Na,K-ATPase from cryo-electron microscopy of two-dimensional crystals J Mol Biol 2001 314 479 494 11846561
Jiang QX Thrower EC Chester DW Ehrlich BE Sigworth FJ Three-dimensional structure of the type 1 inositol 1,4,5-trisphosphate receptor at 24 A resolution EMBO J 2002 21 3575 3581 12110570
Rice WJ Young HS Martin DW Sachs JR Stokes DL Structure of Na+,K+-ATPase at 11-A resolution: Comparison with Ca2+-ATPase in E1 and E2 states Biophys J 2001 80 2187 2197 11325721
Serysheva II Bare DJ Ludtke SJ Kettlun CS Chiu W Structure of the type 1 inositol 1,4,5-trisphosphate receptor revealed by electron cryomicroscopy J Biol Chem 2003 278 21319 21322 12714606
Michaely P Tomchick DR Machius M Anderson RG Crystal structure of a 12 ANK repeat stack from human ankyrinR EMBO J 2002 21 6387 6396 12456646
Jorgensen PL Purification of Na+,K+-ATPase: Enzyme sources, preparative problems, and preparation from mammalian kidney Methods Enzymol 1988 156 29 43 2835612
Davis JQ Bennett V The anion exchanger and Na+K(+)-ATPase interact with distinct sites on ankyrin in in vitro assays J Biol Chem 1990 265 17252 17256 2170369
Maeda N Niinobe M Mikoshiba K A cerebellar Purkinje cell marker P400 protein is an inositol 1,4,5-trisphosphate (InsP3) receptor protein. Purification and characterization of InsP3 receptor complex EMBO J 1990 9 61 67 2153079
|
16292983
|
PMC1287507
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e423
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030423
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0030434SynopsisPhysiologyAnimalsHomo (Human)MammalsMus (Mouse)VertebratesAnkyrin-B Binds Disparate Proteins to Keep Calcium Flowing in the Heart Synopsis12 2005 29 11 2005 29 11 2005 3 12 e434Copyright: © 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.
Ankyrin-B Coordinates the Na/K ATPase, Na/Ca Exchanger, and InsP3 Receptor in a Cardiac T-Tubule/SR Microdomain
==== Body
Muscle cells of the heart regulate their calcium levels tightly, and no wonder—an influx of calcium triggers contraction of the cell, and the beating of the heart. After it enters, calcium must be quickly pumped back across the cell membrane, a job that falls to the sodium/calcium exchanger (NCX1), which trades incoming Na+ ions for outgoing Ca2+ ions. The sodium, in turn, is pumped out by the workhorse of membrane gradients, Na/K ATPase.
Defects in calcium equilibrium, or homeostasis, underlie major diseases of the heart, including arrhythmia, an inability to regulate the heartbeat. One cause of arrhythmia is a mutation leading to loss of the protein ankyrin-B. This mutation increases calcium within heart muscle cells. In this issue, Peter Mohler, Vann Bennett, and colleagues show that ankyrin-B binds to both NCX1 and Na/K ATPase, as well as to a third protein; that mutations in ankyrin-B disrupt this complex; and that the loss of this complex is the likely reason for arrhythmia from ankyrin-B mutation.
Ankyrin-B was known to bind individually to both proteins, as well as a third one, the inositol 1,4,5-trisphosphate receptor (InsP3R). To determine if the entire group formed a single complex, the authors stained the various proteins, and using three-dimensional microscopy, showed that the staining pattern for each largely overlapped. They next used antibodies to precipitate each of the four proteins in turn. They found that, in each case, precipitation of one protein brought the others along with it, strongly suggesting the four formed a single multiprotein complex. This conclusion was further strengthened when they found that the purified proteins created in vitro could also link together. Microscopy revealed that this complex was embedded in an invagination of the plasma membrane called the transverse tubule, or T-tubule. The T-tubule also holds the proteins that allow calcium into the cell, but the staining pattern showed that these were located apart from the ankyrin-B complexes.
Finally, the authors examined how well mutant ankyrin-B binds to the other proteins in the complex. They found that the mutant lost 60% of its ability to bind the other three proteins. Since the physiological effect of the mutation is loss of calcium regulation in heart cells, these results strongly suggest that binding to ankyrin-B is critical for efficiently coordinating the function of the sodium/calcium exchanger with that of the Na/K ATPase, to remove calcium from the cell. The authors note that their results do not explain the function of the InsP3R protein, whose role in the heart is currently unknown. Earlier evidence suggested it may cooperate with other proteins to regulate calcium influx, but that seems less likely now, given its localization on this complex.
Along with explaining the mechanism of a known defect in cardiac calcium regulation, these results also highlight the important role played by “adapter” proteins such as ankyrin-B in creating “molecular machines.” Such multiprotein complexes are common in the cell, and their working depends on the close proximity of member subunits. By bringing together subunits of disparate structures but related functions, such adapter proteins increase the efficiency of the unit as a whole. —Richard Robinson
The localization pattern of ankyrin-B (red) and dihydropyridine receptor (green) in cardiomyocytes
|
0
|
PMC1287508
|
CC BY
|
2021-01-05 08:21:47
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e434
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030434
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0030435SynopsisEvolutionZoologyMolluscsDNA Barcodes Perform Best with Well-Characterized Taxa Synopsis12 2005 29 11 2005 29 11 2005 3 12 e435Copyright: © 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.
DNA Barcoding: Error Rates Based on Comprehensive Sampling
.
==== Body
With species around the world disappearing faster than biologists can identify them, the need for rapid, accurate methods of classifying life has never been more pressing. Toward this end, many scientists pinned their hopes on DNA barcoding, a recently proposed strategy that treats a short fragment of DNA as a sort of universal product code to identify species by running unknown sequences through a database that links DNA barcodes to organisms. But this approach generated controversy from the start, with advocates touting the benefits—rapid identification of unknown individuals and discovery of novel species—and skeptics bristling at the notion that a single gene fragment could perform such a tall task.
For most animals, the DNA barcode consists of just over 600 base pairs of a mitochondrial gene called cytochrome oxidase subunit I (COI). In September 2004, PLoS Biology published a paper that tested COI barcode performance using a proportion of North American birds. The study found that all these well-studied species had a different barcode, and that the variation between species was much higher than variation within species. Based on this gap, the study proposed a screening threshold of sequence difference (ten times the average within-species difference) that could speed the discovery of new animal species. In a new study, Christopher Meyer and Gustav Paulay revisit the issue with a diverse, extensively studied snail group, the ubiquitous, tropical marine cowries whose shell can command over $30,000. Meyer and Paulay found that while the barcode worked well for identifying specimens in highly characterized groups, thresholds would miss many novel species.
After ten years of collecting and sequencing cowries from around the world, Meyer and Paulay assembled a database of over 2,000 cowrie COI sequences from 218 species. To capture the full range of within-species variation and geographic differences in population structure, they included sequences from multiple individuals and geographic extremes. Meyer and Paulay tested barcode performance in species identification and discovery against traditional morphology-based species lists and against an integrated taxonomic approach that determines “evolutionary significant units” (ESUs) based on morphology and sequence data. ESUs are what's called reciprocally monophyletic—two ESUs each have a unique ancestor, and, therefore, a unique genetic signature. But genetic variation doesn't always track with species distinctions. The common ancestor of some species nests within another species' variation (called paraphyly), and sometimes different members of what is thought to be one species can be related to another species and not share a most recent common ancestor (called polyphyly).
The charismatic cowrie
Meyer and Paulay found that barcodes could accurately identify unknown samples against a well-characterized database using ESUs, but were “prone to error”—with a 20% failure rate—when traditional species checklists were used, likely reflecting taxonomic problems mentioned above (lumping similar forms that turn out to be distinct species, for example, or erroneously classifying a specimen with an odd morphology as separate species).
When Meyer and Paulay looked at thresholds to delineate species in cowries, they found considerable abundance of young taxa between intra- and intertaxon variation at both ESU and species levels. Within-species variation among cowries was “substantially higher” than that found in two other marine snails, limpets and turbinids, demonstrating the value of comparative analyses in generalizing limits for intra-species variation. The three groups also showed a wide range of interspecies variation. Still, using a barcode threshold to constrain intraspecies variation worked well for ESUs (98% of the taxa had less than 3% variation).
But error rates were substantial when applying thresholds to species discovery because of the abundance of young taxa. For instance, of the 263 ESUs, 16% artificially lumped with another ESU at the 3% threshold; similar patterns were seen in the turbinids and limpets. Because many traditionally recognized cowrie species are not reciprocally monophyletic based on their COI barcode, when Meyer and Paulay replaced cowrie ESUs with recognized species, both intra- and interspecies variation increased, bumping the error rate above 30%.
This comprehensive analysis demonstrates that relying solely on DNA barcodes masks fine-tuned species boundaries not readily captured in DNA sequences without extensive sampling. The barcode performs best in identifying individuals against a well-annotated sequence database—as demonstrated here with ESUs—and the authors argue that the barcoding movement is well-equipped to help in this effort. But barcoding methods for discovering new species need refinement, they argue, and should be developed in collaboration with taxonomists, systematists, and ecologists into a comprehensive taxonomic framework. Once databases are fully annotated with taxonomically evaluated sequences, error rates should go down. With just 1.7 million species described and some 10 million to go, there's a lot of work to be done. —Liza Gross
|
0
|
PMC1287509
|
CC BY
|
2021-01-05 08:21:47
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e435
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030435
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0030436SynopsisGenetics/Genomics/Gene TherapyOtherStatisticsHomo (Human)Go West, Early Man: Modeling the Origin and Spread of Early Agriculture Synopsis12 2005 29 11 2005 29 11 2005 3 12 e436Copyright: © 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.
Tracing the Origin and Spread of Agriculture in Europe
==== Body
After the last major ice age some 10,000 years ago, things began to look up for early humans. Forbidding climes yielded to more hospitable weather patterns, and people began to settle down and domesticate plants and animals. Archeologist Gordon Childe, who in 1942 called the transition from hunting and gathering to agriculture the Neolithic Revolution, proposed that unchecked population growth triggered economic and social problems among Near Eastern populations and forced farmers and shepherds to search for new lands. In this demic diffusion model, dispersing populations introduced Europeans to the Neolithic lifestyle. Alternately, Europeans may have learned to farm by imitating Neolithic practitioners they encountered through trade or other interactions (the cultural diffusion model).
Childe's ideas of westward migration found support in a 1965 study that mapped the spatiotemporal pattern of a small sample of radiocarbon dates (determined from animal bones and other carbon remains) from Neolithic sites. A landmark study by Albert Ammerman and Luigi Cavalli-Sforza in 1971 used more data—radiocarbon dates from 53 early Neolithic sites—and used a population biology model to investigate Neolithic spread. Their “wave of advance” model proposed that population growth at the agricultural fringes coupled with local migrations would produce steady population expansions in all directions. They calculated an average rate of spread of about one kilometer per year.
But the controversy between the cultural and demic diffusion models still remains today. Now, over 30 years later, Ron Pinhasi and Joaquim Fort revisited the question along with Ammerman, using a substantially larger dataset with new locations—radiocarbon-dated bones and charcoal from 735 Neolithic sites in Europe, the Near East, and Asia—and reaffirm the wave-of-advance model. The authors combined mathematical and geospatial techniques to estimate the timing and likely center of agricultural origins, as well as the rate of spread. Their results support a model of demic diffusion and, for the first time, pinpoint the geographic origin of agriculture within the Fertile Crescent.
Pinhasi et al. calculated the correlation between the straight distance versus age of the 735 radiocarbon dates and the likely spread from 25 hypothetical centers of origin (based on location only) and ten probable centers (sites that included the oldest remains, as well as a center proposed in the 1971 study). The most southern point, Abu Madi in Egypt, had the highest correlation, though eight of the other probable centers had similar scores. However, charting the shortest paths (which take into account the barrier effect of the Mediterranean Sea), pointed to an origin in the north. Focusing on the centers that seemed most likely, Pinhasi et al. used both approaches (one based on straight paths, one based on shortest paths) to estimate the speed of agricultural spread, and came up with nearly the same figure: 0.7–1.1 kilometers per year versus 0.8–1.3 kilometers per year. An error range for this speed was estimated (which had not been done before), so the authors could also compare this observed rate with that predicted by a model.
While no cultural diffusion model is known so far that can explain the observed rate (calculated from the archeological evidence), a kilometer or so a year is consistent with a time-delayed demic diffusion model. (This model, which was proposed by Fort and co-workers in 1999, also agrees with data from other human and nonhuman population expansions, as well as with the observed speeds of virus infections.) While many genetic studies also support demic diffusion, they do not agree on the extent to which Near Eastern farmers contributed to the European gene pool. Assuming a linear advance, agricultural expansion began some 9,000–11,500 years ago, falling in line with a gradual wave of advance. Rather than “racing across the map of Europe,” the authors argue, the Neolithic transition took over 3,000 years, or 100 generations, reflecting the time children stay with their parents before moving on to greener pastures. This is precisely the time-delay effect that classical diffusion models are unable to capture, but that is accounted for in the model by Fort and co-workers. Finally, the authors incorporated radiocarbon data from 30 sites in Arabia to find the most likely birthplace of agriculture. Their shortest-path analysis points to northern Levant and northern Mesopotamia (whereas the straight-path, or classical, approach pointed to a southern origin).
The authors' approach did not address whether migrants traveled by land or by sea or whether farmers displaced foragers. But the pattern and processes of dispersal were likely complex, Pinhasi et al. conclude, with multiple paths and mechanisms fueling the western expansion of the Neolithic lifestyle. And with a newly bolstered wave-of-advance model and the approach outlined here, geneticists, anthropologists, and other researchers investigating the origin and spread of human populations have a more detailed roadmap to follow.
Note for international readers: During the time of Westward expansion in the 19th century, American essayist Horace Greeley famously advocated Manifest Destiny by exhorting, “Go West, young man!” —Liza Gross
|
0
|
PMC1287510
|
CC BY
|
2021-01-05 08:28:16
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e436
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030436
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0030437SynopsisCancer BiologyCell BiologyGenetics/Genomics/Gene TherapyOncologyHomo (Human)An Insecure Role for Securin in Chromosome Segregation Synopsis12 2005 29 11 2005 29 11 2005 3 12 e437Copyright: © 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.
Securin Is Not Required for Chromosomal Stability in Human Cells
==== Body
Just as good parents try to prevent inequities among their progeny, a dividing cell must ensure that its daughters inherit all the chromosomes they are entitled to. But with cells, such evenhandedness goes beyond matters of equity to those of life and death. The progeny of cancer cells, for example, typically suffer rampant chromosome losses. Though these losses should eventually cause the cells' demise, they can also reveal or induce mutations that encourage proliferation, which explains why cancer cells accommodate widespread chromosomal instability. Understanding how normal cells unerringly transmit a full chromosome set to their daughters has long been an important part of the fight against cancer.
In the ballet of cell division, or mitosis, chromosomes, intracellular fibers, and the cell outer membrane execute carefully choreographed steps and partner shifts. First, cells replicate their DNA, creating twin sets of each chromosome, known as sister chromatids. Mitosis, per se, starts when the chromatids condense their DNA into compact bodies that are captured in a spindle of tubulin fibers (called microtubules). The microtubules first line up the chromatid pairs along the spindle's middle plane; then they pull them apart, hauling the members of each pair to opposite ends of the spindle. The mother cell then pinches its membrane along the spindle's middle plane, splitting into two daughter cells with full chromosome sets. Keeping the chromatids of a pair together at the beginning of mitosis, and allowing their timely separation at the end, are both crucial steps for proper chromosome segregation into the daughter cells.
From the time DNA replication begins, sister chromatids are held together by a protein complex named cohesin. After the chromatid pairs are all neatly positioned in the center of the spindle, they can be safely segregated, a job performed by separase, an enzyme that dissolves cohesin's grip by breaking down one of its protein components. The dissolving power of separase is tightly controlled to avoid precocious or delayed chromatid separation. One of separase's regulators is securin, a protein known as a chaperone that appears to hold separase captive until just the right time. A recent study indicated that human cells devoid of securin underwent abnormal mitoses that led to widespread chromosome losses, making securin a key player in chromosome segregation and a promising entry point into cancer therapy. But in a new study in PLoS Biology, Katrin Pfleghaar, Michael Speicher, and colleagues have repeated and expanded on these experiments, and come to somewhat different conclusions.
Both teams carried out the same experiments in the same system: they counted chromosomes and examined the mitotic process in cultures of human cells lacking the securin gene. And both teams found that in the first weeks of cell culture, cells were losing chromosomes at a very high rate and most mitoses showed abnormal chromatid distribution. But in this study, Pfleghaar observed the cells for longer periods and found that as time went by, the culture recovered: cells with abnormal mitoses and chromosome counts became rarer until, after a few weeks, the cells appeared indistinguishable from their relatives with an intact securin gene. Interestingly, both studies found that the amounts and activity levels of separase were low in securin-deficient cells, which confirms that securin regulates separase. Pfleghaar and her colleagues speculate that securin normally plays an important role in mitoses, but that in its absence, cells tap into compensatory mechanisms to restore proper chromosome segregation.
The implications for cancer treatment are potentially great, as mathematical models of cancer growth do not usually include the possibility that cell populations might recover from chromosomal instability. In addition, such recoveries might interfere with therapies that aim to kill cancer cells by exacerbating their chromosome losses. —Francoise Chanut
|
0
|
PMC1287511
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e437
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030437
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0030438SynopsisAnimal BehaviorGenetics/Genomics/Gene TherapyNeuroscienceDiabetes/Endocrinology/MetabolismGeriatricsMus (Mouse)What Makes Mice Fat? How the Brain Controls Energy Balance Synopsis12 2005 29 11 2005 29 11 2005 3 12 e438Copyright: © 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.
Effects of Hypothalamic Neurodegeneration on Energy Balance
==== Body
In most animals, food intake and energy expenditure vary greatly from day to day. Yet, in healthy young animals, cumulative energy intake over several days matches energy use very closely. This balancing act or “energy homeostasis” is controlled by complex neuronal circuitry and numerous signaling molecules. When these control mechanisms go wrong, the result is weight loss or obesity. Middle-aged spread, for example, is probably caused by a progressive impairment of energy homeostasis.
Two types of neurons in the hypothalamus—a region deep in the brain that controls many aspects of physiology—help to regulate fat buildup, or adiposity. First, there are proopiomelanocortin (Pomc) neurons, so called because they make proopiomelanocortin. This is a precursor for melanocortins, peptides that bind to melanocortin receptors elsewhere in the brain to limit food intake and increase energy expenditure. Then there are agouti-related protein (Agrp) neurons. These make agouti-related protein, named for its similarity to a protein mutated in an obese mouse with a characteristic yellow coat. By blocking melanocortin receptors, Agrp increases food uptake—the scientific term for this is hyperphagia—and decreases energy use. Both types of neurons detect circulating indicators of body adiposity such as leptin, and then act to keep energy stores constant.
Normal mouse hypothalamic tissue (left) and mutant tissue (right), with Pomc neurons deleted
Support for this model for energy homeostasis comes from rodent studies in which the hypothalamus was damaged or stimulated, or in which leptin and other peptides were injected directly into the brain. Genetic experiments in mice provide further support but also some conflicting evidence. While deletion of the Pomc gene or overexpression of Agrp increase appetite and obesity as predicted by the model, unexpectedly, deletion of the Agrp gene does not disturb energy balance. One explanation for this is that Pomc and Agrp neurons might play a role in energy homeostasis even when they don't express their defining peptides; it is known, for instance, that Pomc and Agrp neurons express additional neuropeptides with effects similar to Pomc and Agrp, and that Agrp neurons control the activity of Pomc neurons.
To investigate more fully the roles that Pomc and Agrp neurons play in energy homeostasis, Allison Wanting Xu et al. have constructed mouse strains in which the Pomc or Agrp neurons are lost progressively after birth. They took advantage of a technique that selectively deleted the gene for the mitochondrial transcription factor A (Tfam) in Pomc- or Agrp-expressing neuronal cells. Tfam is required for transcription of the mitochondrial genome, which encodes proteins required for cellular respiration and thus cell survival. By six months old, the researchers report, the engineered mice had lost many of their Pomc or Agrp neurons but no other neurons.
Like aging humans, mice in which Pomc neurons had died became progressively fatter because of an increased food intake and reduced energy expenditure. Mice that had lost Agrp neurons weighed slightly less than control animals, and mice engineered so that both types of neurons died weighed more than control mice but less than those lacking just Pomc neurons. These results indicate that the regulation of adiposity by Pomc and Agrp neurons is not simply a matter of releasing these two neuropeptides.
Xu et al. made an additional, unexpected observation. After food deprivation, mice normally increase their food intake acutely until their fat stores return to prefasting levels—a process called compensatory hyperphagia. Paradoxically, mice without Pomc neurons showed reduced compensatory hyperphagia despite overeating under normal conditions. Since aging humans also fail to increase their food intake after fasting, these mouse strains that gradually lose specific hypothalamic neurons provide a potentially informative model of human age-related obesity. In addition, by studying such mice, scientists may gain important insights into the full complexity of how hypothalamic neurons regulate energy balance that could help to reverse the current human obesity epidemic. —Jane Bradbury
|
0
|
PMC1287512
|
CC BY
|
2021-01-05 08:21:47
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e438
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030438
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0030439SynopsisNeuroscienceHomo (Human)Neural Basis of Body Image: How to Lose Inches at the (Perceived) Flick of the Wrist Synopsis12 2005 29 11 2005 29 11 2005 3 12 e439Copyright: © 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.
Neural Substrate of Body Size: Illusory Feeling of Shrinking of the Waist
==== Body
Wouldn't it be nice if you could change your body image by placing a vibrating gadget on your wrist? As it happens, you can—though under controlled circumstances. Vibrating skin over the tendon of a joint extensor muscle triggers the vivid sensation that the joint is passively flexing, even though it's not. When the hand is touching the waist, nose, or some other body part, a person can feel the wrist bending and the body part stretching or shrinking—in what's aptly called the Pinocchio illusion.
Vibrations on the skin over a muscle tendon cause the perceptual illusion by exciting sensory nerve endings in the tendon that send signals to brain areas that process touch and motor control, the primary somatosensory cortex and the primary motor cortex. The somatosensory cortex creates neural maps of the body surface, and receives sensory inputs from receptors in the peripheral nervous system. But these peripheral receptors carry no information about the relative size of body parts, and the brain has no specialized neurons to receive such information. The neural map of body size and shape are likely represented in a relative way by integrating signals from the relevant body parts and visual cues. The parietal lobes may play a role, based on reports that patients with parietal cortex injuries imagine changes in the size and shape of various body parts. Still, it's not clear how the brain integrates the relevant information to compute body image.
Higher-order somatosensory areas in the parietal cortex mediate perceived changes in body shape and size
To investigate the neural correlates of body image, H. Henrik Ehrsson, Eiichi Naito, and their colleagues recruited 24 participants to model the “waist shrinking illusion,” and then scanned their brains with functional magnetic resonance imaging (fMRI). The authors hypothesized that higher-order somatosensory areas in the parietal cortex would reflect the perceived changes in waist size, and designed the study to isolate illusion-linked brain activity by varying participants' hand position (body contact/no contact, or free) and the vibration site (tendon/skin, or beside the tendon).
After participants experienced each possible combination of hand position and vibration site, they answered “now” when they felt the illusion, and then chose a picture from six different body configurations that best represented their experience. They rated the vividness of the experience, on a scale of zero to nine (absolutely realistic), and then moved their wrists to show what they felt so the authors could measure the angle. At the same time, electromyograms (EMGs) recorded muscle stimulation.
Seven participants did not reliably experience the illusion and so were not scanned. The other 17 participants underwent six experimental trials (two baselines, with hands resting, were added) while their brains were scanned (while lying in the fMRI machine). In three trials, participants' hands lay freely, but supported, by their side without touching the body (tendon free/skin free/rest free). In the other three trials, the palms of the hand were in direct contact with their sides (tendon contact/skin contact/skin free), while a strap allowed them to relax their arms.
During the tendon contact condition, all 17 participants sensed their hands flexing and their waist shrinking. The degree of flexion corresponded to a 28% waist shrinkage. This sensation was vivid, reliable, and quick to start. The EMGs showed no muscle activity in over 70% of the participants, and muscle activity wasn't significantly different in tendon contact and tendon free, confirming that muscle stimulation did not account for the illusion. The brain regions showing most activity during the illusory perception were in the left parietal lobe, within the anterior intraparietal sulcus (a sulcus is an inward fold of the brain) and extending toward the postcentral sulcus.
Participants who reported the strongest shrinking waist illusion also showed the strongest activity in the postcentral sulcus and the anterior left intraparietal cortex—activity that was not observed in participants who felt illusory wrist movement when their hands were not touching their body—confirming a link between these brain regions and the shrinking waist illusion. When the brain receives conflicting sensory information from the vibrated wrists and the sensory inputs of the hands on the waist, the brain recalibrates the relative size of the wrist and shape of the waist, creating the illusion that the waist is shrinking as the hands are bending inward. Altogether, these results suggest that the brain computes body image by integrating signals from the skin, joints, and muscles through hierarchical processing in the somatosensory system. The researchers could elicit this illusion as many times as needed for the fMRI experiment, but there's no indication that a portable device will hit the consumer market anytime soon. —Liza Gross
|
0
|
PMC1287513
|
CC BY
|
2021-01-05 08:28:16
|
no
|
PLoS Biol. 2005 Dec 29; 3(12):e439
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0030439
|
oa_comm
|
==== Front
PLoS PathogPLoS PathogppatplpaplospathPLoS Pathogens1553-73661553-7374Public Library of Science San Francisco, USA 1630460710.1371/journal.ppat.001002405-PLPA-RA-0078R3plpa-01-03-04Research ArticleCell BiologyImmunologyParasitologyMus (mouse)In VitroDisruption of Toxoplasma gondii Parasitophorous Vacuoles by the Mouse p47-Resistance GTPases Vacuolar Disruption by p47 GTPasesMartens Sascha 1¤Parvanova Iana 1Zerrahn Jens 2Griffiths Gareth 3Schell Gudrun 4Reichmann Gaby 4Howard Jonathan C 1*
1 Institute for Genetics, University of Cologne, Cologne, Germany
2 Max Planck Institute for Infection Biology, Berlin, Germany
3 European Molecular Biology Laboratory, Heidelberg, Germany
4 Institute for Medical Microbiology, Heinrich Heine University Duesseldorf, Duesseldorf, Germany
Boothroyd John EditorStanford University, United States of America* To whom correspondence should be addressed. E-mail: [email protected]¤ Current address: Medical Research Council Laboratory of Molecular Biology, Cambridge, United Kingdom
11 2005 18 11 2005 1 3 e2421 6 2005 4 10 2005 Copyright: © 2005 Martens 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.The p47 GTPases are essential for interferon-γ-induced cell-autonomous immunity against the protozoan parasite, Toxoplasma gondii, in mice, but the mechanism of resistance is poorly understood. We show that the p47 GTPases, including IIGP1, accumulate at vacuoles containing T. gondii. The accumulation is GTP-dependent and requires live parasites. Vacuolar IIGP1 accumulations undergo a maturation-like process accompanied by vesiculation of the parasitophorous vacuole membrane. This culminates in disruption of the parasitophorous vacuole and finally of the parasite itself. Over-expression of IIGP1 leads to accelerated vacuolar disruption whereas a dominant negative form of IIGP1 interferes with interferon-γ-mediated killing of intracellular parasites. Targeted deletion of the IIGP1 gene results in partial loss of the IFN-γ-mediated T. gondii growth restriction in mouse astrocytes.
Synopsis
Toxoplasma gondii is a small unicellular parasite infecting virtually every warm-blooded animal including humans. After infection, T. gondii does not stay in extracellular fluids such as the blood, but actively invades body cells. The parasite has developed elaborate mechanisms enabling it to form a so-called parasitophorous vacuole (PV) within the cell it invades. Within this vacuole the parasite multiplies until the host cell ruptures and the progeny are released into the extracellular space to infect further cells. Host cells have developed several special mechanisms to combat the parasite. In mice, these mechanisms include a protein family, the p47 GTPases, which are induced by immune-alert factors called interferons. This study begins to address how the mouse p47 GTPases function. The study shows that the p47 GTPases assemble on the PV very shortly after infection, apparently to form a “membrane attack complex.” Within an hour the PV membrane shows signs of damage, bulging into small out-foldings that separate from the membrane in small vesicles. Shortly afterward the PV membrane ruptures and the parasite deteriorates. The p47 GTPase have several properties in common with the dynamin GTPases, which deform cellular membranes, suggesting that the p47 GTPases function in a mechanistically similar manner.
Citation:Martens S, Parvanova I, Zerrahn J, Griffiths G, Schell G, et al. (2005) Disruption of Toxoplasma gondii parasitophorous vacuoles by the mouse p47-resistance GTPases. PLoS Pathog 1(3): e24.
==== Body
Introduction
Many pathogens taken into cells by phagocytosis are killed by phagosomal maturation [1]. It is, however, unclear how cells eliminate parasites such as the apicomplexan protoza Toxoplasma gondii and Plasmodium which enter host cells by active invasion [2–4]. Within host cells, Plasmodium and T. gondii replicate in parasitophorous vacuoles (PVs) formed during invasion by invagination of the plasma membrane [5]. Many host plasma membrane proteins are excluded from the parasitophorous vacuole membrane (PVM) [6,7] and the PV does not fuse with the host cell endocytic compartment [8,9]. In contrast, the vacuole is massively modified by the parasite, including the recruitment of host cell endoplasmic reticulum (ER) and mitochondria [10], the establishment of an intravacuolar network [11,12], and the insertion of pores into the PVM allowing the free diffusion of molecules below 1,300 Da [13].
Early production of IL-12, after infection of mice, stimulates IFN-γ secretion (mainly by CD4+ and NK cells) that is required for an efficient immune response to T. gondii and other protozoan pathogens such as Plasmodium [14–17]. Experiments in mice using bone marrow chimeras have shown that efficient control of T. gondii is dependent on expression of the IFN-γ receptor in both the hemopoietic and non-hemopoietic compartment [18]. Perforin-deficient mice are still able to control acute T. gondii infections suggesting that control of intracellular T. gondii occurs in a non-cytolytic manner [19]. Together these results suggest the IFN-γ-mediated induction of cell-autonomous resistance mechanisms in hemopoietic and non-hemopoietic cells that control replication and dissemination of T. gondii.
Consistent with the cell-autonomous non-cytolytic control of T. gondii, astrocytes isolated from neonatal mice are able to kill intracellular parasites after activation with IFN-γ [20,21]. This killing was shown to be independent of inducible nitric oxide synthase and indoleamine 2,3-dioxygenase but required the presence of the IFN-γ-inducible p47 GTPase IGTP [22]. Recently it has been shown that the p47 GTPases IGTP and LRG-47 are both required for IFN-γ-mediated growth inhibition of T. gondii in bone marrow-derived macrophages [23]. Indeed, the interferon-inducible p47 GTPases are potent resistance factors in mice against T. gondii [24,25] and other intracellular pathogens [26,27]. The mouse genome encodes 23 p47 GTPases [28] among which are IRG-47 [29], GTPI [30], IGTP [31], LRG-47 [32], TGTP [33], and IIGP1 [30]. Mice deficient for IGTP, LRG-47, or IRG-47 showed complete (IGTP, LRG-47) or partial (IRG-47) loss of host resistance to T. gondii infection despite an otherwise intact immune system and normal IFN-γ production [24,25]. Like the IFN-γ receptor, the p47 GTPase IGTP must be expressed in hemopoietic and non-hemopoietic cells for efficient control of T. gondii infection [34].
After induction by IFN-γ, the p47 GTPases are predominantly associated with intracellular membranes including the Golgi apparatus and the ER [35–37]. LRG-47 associates with the cis-Golgi apparatus by means of an amphipathic helix while the membrane association of IIGP1 is aided by an N-terminal myristoylation moiety [35]. IRG-47 is an exception in being almost completely cytosolic [35]. IIGP1 is equally distributed between the cytosol and the ER [35], and is the best understood member of the p47 family, both biochemically and structurally, with micromolar affinity for guanine nucleotides, cooperative GTP hydrolysis, and GTP-dependent oligomerization in vitro [38,39]. These properties, which are shared by the IFN-inducible Mx proteins and GBPs, relate the p47 and other interferon-inducible GTPases to the dynamin family of GTPases mediating membrane tubulation and vesicle scission [40].
Despite clear evidence placing the p47 GTPases at the center of initial defense against intracellular protozoan and bacterial pathogens in mice, their mode of action is unknown [24–27]. We show here that the IFN-γ-mediated killing of intracellular T. gondii is accompanied by the accumulation of multiple p47 GTPases at the T. gondii PV in primary mouse astrocytes. The vacuolar accumulation of the p47 GTPase IIGP1 requires live parasites and GTP binding. The accumulations undergo a maturation-like process resulting in vesiculation and disruption of the T. gondii PVM and subsequent killing of parasites. We further show that IIGP1 contributes to vacuolar disruption. Our study shows that the protozoan parasite T. gondii is eliminated from cells in a p47 GTPase-dependent process that is unrelated to phagosomal maturation.
Results
IFN-γ-Mediated Killing of T. gondii in Primary Mouse Astrocytes
We studied the p47 GTPase-mediated mechanism of resistance against the ME49 strain of T. gondii in primary astrocytes isolated from neonatal C57BL/6 mice. T. gondii replication was strongly inhibited by IFN-γ in a dose-dependent manner (Figure 1A) [20]. Growth inhibition is not dependent on indoleamine 2,3-dioxygenase or inducible nitric oxide synthase [20] (unpublished data) but requires at least one p47 GTPase, IGTP [22]. The induction of IIGP1 (Figure 1A), and other p47 GTPases (Figure S3), correlated with IFN-γ-induced growth inhibition of T. gondii. In untreated cells 24 h after infection, the frequency of PVs relative to total cells dropped to 58% of the 2-h value, probably reflecting astrocyte replication, while in the IFN-γ-treated cells the frequency of PVs dropped sharply over 24 h to only 17% of the 2-h value (Figure 1B). Thus, IFN-γ promotes killing of intracellular parasites and loss of PVs from infected mouse astrocytes [21]. In both IFN-γ-treated and untreated cells, PVs contained one parasite at 2 h and 8 h post-infection, while in both cases at 24 h, vacuoles contained up to eight organisms. Thus interferon-dependent killing of parasites occurs shortly after infection, and parasites replicate normally if they survive this phase (unpublished data).
Figure 1 IFN-γ-Mediated Growth Inhibition and Intracellular Killing of T. gondii Are Accompanied by Accumulation of p47 GTPases at the PV
(A) Astrocytes were induced with the indicated concentrations of IFN-γ and infected with T. gondii 24 h later for 68 h. The growth of intracellular parasites was monitored by uracil incorporation assay. (Inset) Lysates of astrocytes induced with the indicated concentrations of IFN-γ for 24 h were probed for IIGP1 protein by Western blotting.
(B) Untreated or IFN-γ induced astrocytes were infected with T. gondii. After 2 h, extracellular parasites were washed away and cells were either fixed or incubated further for a total of 8 h or 24 h. Shown are the mean values of three independent counts representing a total number of 650–997 cells per time point.
Accumulation of p47 GTPases at PVs
We monitored p47 GTPases in IFN-γ-stimulated astrocytes during infection. At 2 h post-infection, IGTP, GTPI, TGTP, and IRG-47 (Figure 2A–D) and IIGP1 (Figure 2E) accumulated markedly at PVs. The accumulation was most intense for IIGP1, TGTP, and IRG-47 (Figure 2 and Figure S1), somewhat less so for GTPI, and least intense for IGTP. We detected no accumulation of LRG-47 at the vacuole, but the A19 antiserum had a marked background stain on the T. gondii organism that set a lower limit of detectability (Figure S1). We attempted to detect vacuolar localization of LRG-47 in other ways. Unfortunately, neither N- nor C-terminally tagged LRG-47 is correctly localized [35], but a C-terminal tetracysteine-modified LRG-47 was localized correctly in interferon-treated cells and could also not be detected significantly at the vacuole (unpublished data). We tentatively conclude, with Butcher et al. [23], that LRG-47 does not localize significantly to the T. gondii vacuole. In uninfected, IFN-γ-stimulated cells the p47 GTPases are associated with ER (IIGP1, IGTP, and TGTP) and Golgi (LRG-47, GTPI) membranes or are cytosolic (IRG-47) [35–37] (Figure S2, unpublished data). Accumulation of p47 GTPases at the T. gondii vacuole could not be explained by the recruitment of ER cisternae to the vacuole [10]. The ER-localized chaperones, ERP60, protein disulphide isomerase (PDI), and calnexin, were hardly detectable at the PVs (Figure 3), while IIGP1 localization was extremely intense (Figure 3); indeed a useful image of IIGP1 localization at the vacuole could be obtained only under photographic conditions under which the ER localization of this abundant protein was invisible (Figure 3A–3C). As noted above, IIGP1 was also markedly more concentrated at the PV than the other completely ER-localized p47 GTPase, IGTP [36] (Figure 2A). By immunogold electron microscopy for IIGP1 (Figures 4A and S3) we saw intense label clearly localized to the PVM around many vacuoles in IFN-γ-induced cells (Figure 4A), while only weak label was detectable on ER cisternae. These results show that IIGP1, ER-localised in the interferon-induced cell, is repositioned and intensely concentrated at the PVM shortly after T. gondii infection. By implication, the same is true for TGTP. IRG-47 is almost exclusively cytoplasmic in the interferon-induced cell [35], while GTPI is almost exclusively localized to the Golgi (Figure S2). The repositioning of these two p47 GTPases to the PV therefore also cannot be secondary to the accumulation of ER cisternae at the PV. We conclude that the p47 GTPases, with the possible exception of LRG-47, accumulate at the PV shortly after T. gondii infection of interferon-stimulated cells; this accumulation is independent of the accumulation of ER cisternae.
Figure 2 The Accumulation of p47 GTPases at the PV Is Dependent on Active Invasion by T. gondii
(A–D) IFN-γ-induced astrocytes were infected with T. gondii for 2 h, fixed, and stained for IGTP (A), GTPI (B), TGTP1 (C), or IRG-47 (D).
(E) IFN-γ-induced astrocytes were infected with T. gondii for 2 h, fixed, and stained for IIGP1 (red) and T. gondii (green).
(F) IFN-γ-induced cells were infected with T. gondii, fixed 24 h later, and stained for IIGP1. White arrowheads point to PVs containing replicating parasites.
(G and H) IFN-γ-induced cells were infected with heat-killed (G) or live (H) parasites, fixed 2 h later, and stained for IIGP1 (green) and LAMP1 (red). White arrowheads in (G) point to parasites residing in a LAMP1-positive but IIGP1-negative compartment. (H) Shows single sections of a 3D deconvoluted Z-series. Nuclei of host cells and parasites were stained with DAPI.
Figure 3 The Vacuolar Accumulations of IIGP1 Do Not Reflect Host Cell ER Recruitment by the Parasite
Astrocytes were induced with IFN-γ or left untreated and infected with T. gondii 24 h later for 2 h. Cells were fixed and stained for the indicated proteins. (A) Shows a cell that was stained for IIGP1 (red) and calnexin (green). The vacuolar calnexin signal is markedly less concentrated at the PV than IIGP1. (A') shows the same cell as in (A) but with an electronically enhanced IIGP1 signal to reveal its non-vacuolar ER localization. Note the dramatic difference in the ratio of the ER versus PV signal between IIGP1 and calnexin.
(B and C) Shows astrocytes stained for IIGP1- and the ER-localized PDI. No PDI accumulation at the PV was detected.
(D) Astrocytes were treated as above but stained for IIGP1 (red) and ERP60 (green). Nuclei were stained with DAPI.
Figure 4 IIGP1 Associates Directly with the PVM; the Morphology of the Vacuolar IIGP1 Accumulation Changes in a Time-Dependent Manner
(A) IFN-γ-induced astrocytes were infected with T. gondii for 6 h, fixed, and subjected to ultra-thin cryosectioning. Sections were labeled for IIGP1 using the 165 antiserum and 10 nm gold particles coupled to protein A. The right side is an enlarged view of the boxed region showing that the IIGP1 label was found in close proximity to the PVM (black arrowhead: PVM; white arrowhead: T. gondii plasma membrane; open arrowhead: T. gondii inner membrane complex; bars 200 nm and 100 nm [inset]).
(B) IFN-γ-induced astrocytes were fixed at the indicated times post-infection (MOI of 10) and 110–160 IIGP1-positive vacuoles were counted per time point. Shown is the percentage of smooth (white), rough (hatched), and disrupted vacuoles (black).
(C) IFN-γ-induced astrocytes were infected with T. gondii, fixed 2 h later, and stained for IIGP1 with the 10D7 monoclonal antibody (left) or the 165 antiserum (right). The images show maximum projections of 3D deconvoluted Z-series.
We next analyzed the behavior of IIGP1 in more detail. The recruitment of IIGP1 to the PV is stimulated by active invasion of host cells by the parasites. Heat-killed T. gondii were efficiently internalized by phagocytosis, but no IIGP1 was detected on vacuoles surrounding the dead parasites (Figure 2G). Rather, this compartment was LAMP-1 positive and presumably corresponds to phagolysosomes. LAMP-1 was absent from IIGP1-positive vacuoles containing live T. gondii (Figure 2H) at any time point after infection. Surprisingly, IIGP1 does not apparently accumulate on every PV. Approximately 30% of the PVs had already accumulated IIGP1 by 15 min after infection. By 1 to 2 h post-infection, approximately 75% of vacuoles were IIGP1-positive (about 50% strongly positive), falling to about 20% after 8 h. At 24 h after infection, PVs containing apparently dividing parasites were IIGP1 negative and in many of the infected cells at this time the total cellular IIGP1 signal was markedly lower than in uninfected cells (Figure 2F).
Maturation of the Vacuolar IIGP1 Accumulations
The IIGP1 accumulations seen at the PV by immunofluorescence showed different vacuolar morphologies that we termed smooth, rough, and disrupted (Figure 4B). Rough vacuoles showed a less compact morphology and the IIGP1-positive zone around the PV was broader. Also, the IIGP1 signal appeared foamy and less homogenous compared with smooth vacuoles. At disrupted vacuoles, IIGP1 localized to very bright aggregate-like structures that did not completely surround the parasite. Long IIGP1-positive filaments frequently emanated from rough and disrupted vacuoles (Figure 4C). The percentage of smooth PVs decreased with time after infection and the percentage of rough and disrupted vacuoles increased (Figure 4B). At later time points (6 to 8 h post-infection), disrupted PVs are probably underestimated because they disintegrate and become uncountable. Initially smooth IIGP1-positive PVs therefore probably mature via a rough to a disrupted IIGP1 morphology. Similar morphologies were observed for TGTP-positive PVs, most of which were also positive for IIGP1 (Figure S5). The number of TGTP-positive PVs tended to be somewhat higher than the number of IIGP1-positive PVs.
Loss of GRA7 from Maturing PVs
GRA7 is a T. gondii-encoded protein that is released by intracellular parasites shortly after invasion and localizes to the intravacuolar network and the PVM [41,42]. At smooth PVs, IIGP1 and the PVM-associated T. gondii protein, GRA7, co-localized accurately on apparently intact PVs (Figure 5A). However, rough and disrupted PVs had a much altered GRA7 distribution, generally following the localization of IIGP1 and including the IIGP1-positive filaments (Figure 5B, open arrowheads), and the GRA7 signal was drastically reduced. A similar result was observed for the T. gondii PVM-localized rhoptry protein ROP2 [43] (Figure S4). In general, the intensity of the GRA7 signal correlated well with the maturation status of the IIGP1- and TGTP-positive vacuoles. Early rough vacuoles were still intensely labeled for GRA7 whereas late disrupted vacuoles displayed no GRA7 signal that would identify the former vacuole. As the GRA7 signal weakened at the vacuole, so the GRA7 became increasingly distributed throughout the cytoplasm of infected cells. In particular, at later time points we observed IFN-γ-stimulated cells displaying a strong cytoplasmic GRA7 signal but no apparent vacuole, suggesting that disseminated GRA7 is a relic of vacuolar destruction. Sometimes as early as 2 h after infection, a strong GRA7 signal was seen throughout the cytoplasm of interferon-stimulated, infected cells containing no visible vacuole defined by GRA7. Cytoplasmic GRA7 was never seen in unstimulated, infected cells (Figure 5C).
Figure 5 The Morphological Changes of the IIGP1 Accumulations at the PV Are Accompanied by Loss of T. gondii GRA7 from the PV and its Dissemination throughout the Cytoplasm
(A and B) IFN-γ-induced astrocytes were infected with T. gondii for 2 h (A) or 6 h (B) and stained for IIGP1 (green) and GRA7 (red) (filled arrowheads: IIGP1-negative PVs, open arrowheads: IIGP1-positive PVs). Nuclei were stained with DAPI. (C) Uninduced (left) and IFN-γ-induced (right) astrocytes were infected with T. gondii and stained for GRA7 at 4 h post-infection. Exposure conditions for the two images were the same.
Disruption of T. gondii PVs by Vesiculation
These results pointed to disintegration of the PVM in maturing PVs carrying a high density of p47 GTPases, beginning earlier than 2 h after infection. By electron microscopy of interferon-induced infected cells, the PVM of IIGP1-positive PVs often appeared disrupted at several sites (Figure 6A). At these sites IIGP1 localized to small vesicular forms with an electron-dense coat apparently derived from the adjacent PVM. Frequently the plasma membrane of the parasite itself appeared damaged and IIGP1 label was associated with internal parasite membranes (Figure 6C) suggesting destruction of the parasites themselves. We also observed obviously defunct parasites associated with disrupted IIGP1-positive PVs by immunofluorescence (Figure S4). Frequently, large IIGP1-positive aggregates were seen in infected cells (Figure 6B) often at some distance from the parasite. These aggregates were strongly labeled for IIGP1 and contained apparently vesicular membranes with an electron-dense coat similar to the structures seen at the dissolving PVM. The same structures were also labeled for GRA7 (Figure 6D and 6E). None of these appearances were seen at the PVs of infected cells not induced with interferon. The PVM remained intact and no accumulations of coated vesicles were detected adjacent to the vacuoles. T. gondii inside PVs in uninduced cells were apparently normal and viable.
Figure 6 IIGP1-Positive Vesicular Structures Are Located at Sites Where the PVM Is Disrupted
Astrocytes were induced with IFN-γ, infected with T. gondii and fixed 6 h later. (A–C) Ultra- thin cryosections were labeled for IIGP1 using the 165 antiserum and 10 nm gold coupled to protein A. The insets in (A) show enlarged views of the boxed regions. The black arrows in the bottom left inset point to IIGP1-labeled vesicular profiles with an apparent electron-dense coat. (D and E) Astrocytes were labeled with the anti-GRA7 mAb and 10 nm gold coupled to protein A. (open white arrowhead: T. gondii inner membrane complex [IMC]; filled white arrowhead: T. gondii plasma membrane; black arrowhead: PVM). Bars: 250 nm.
GTP-Dependent Vacuolar Accumulation of IIGP1 Is Required for Efficient Killing of T. gondii
These results associated IIGP1 with disruption of the PVM without establishing a causal link. We therefore asked whether over-expression of IIGP1 would lead to accelerated maturation and disruption of PVs in IFN-γ-induced astrocytes. To detect transfected IIGP1 in an IFN-γ-induced background we generated an epitope-tagged version of the molecule. The N-terminus of IIGP1 is not available for tagging due to its N-terminal myristoylation site [35], and some C-terminal tags interfere with its enzymatic properties [38]. We therefore generated a specific antiserum against a synthetic C-terminal extension of IIGP1 (ctag1) that does not interfere with the structure and known biochemical properties of the GTPase (called IIGP-M in [38]). IIGP1ctag1, under control of a strong CMV promoter, was transfected into IFN-γ-induced astrocytes that were infected with T. gondii 24 h later. In uninfected IFN-γ-induced cells IIGP1ctag1 localized in a reticular ER-like pattern (Figure S2). In many transfected cells, however, IIGP1ctag1 localized to the PVs in greatly increased amounts (Figure 7A), accompanied by accelerated maturation of these vacuoles (Figure 7C). The percentage of rough and disrupted PVs was increased in the transfected cells by 1 h after infection and persisted up to 4 h post-infection. The increase of IIGP1 accumulation at the PV and the accelerated vacuolar maturation accompanied faster loss of GRA7 from the PV (Figure S5).
Figure 7 IIGP1 Contributes to Vacuolar Maturation and Parasite Killing
(A) Astrocytes were transfected with IIGP1ctag1 and simultaneously induced with IFN-γ. After 24 h, cells were infected with T. gondii, fixed 2 h later, and stained for ctag1 (green) and IIGP1 (red).
(B) Cells were treated as in (A) but transfected with IIGP1K82Actag1. White arrowheads point to T. gondii-containing vacuoles.
(C) Astrocytes were treated as in (A), fixed at the indicated time points, and the number of vacuoles displaying a smooth, rough, or disrupted morphology was counted. Shown are the mean values of two independent experiments. (White bars: smooth vacuoles; hatched bars: rough vacuoles, black bars: disrupted vacuoles)
(D) Astrocytes were stimulated with IFN-γ and transfected with IIGP1ctag1 (black bars) or IIGP1K82Actag1 (white bars). Cells were infected with T. gondii 24 h later and fixed at the shown time points. Three independent experiments (A–C) are shown. 217–570 cells per time point and condition were counted.
(E) Astrocytes isolated from neonatal IIGP1−/− mice or wild-type littermates were induced with the indicated concentrations of IFN-γ and infected with T. gondii 24 h later for a total of 48 h. The growth of intracellular parasites was monitored by uracil incorporation assay (p values: ** 0.0015, *** 0.0001 (unpaired Student t-test)).
In contrast to wild-type IIGP1, IIGP1K82Actag1, a GTP-binding deficient mutant (unpublished data), did not localize to the PV (Figures 7B and S5) in IFN-γ-induced cells. Furthermore, IIGP1K82Actag1 behaved as a dominant negative, inhibiting endogenous IIGP1 accumulation at PVs: endogenous IIGP1 and IIGP1K82Actag1 were co-localized and distributed generally across intracellular membranes. The distribution was clearly distinct from the ER localization typical of interferon-induced wild-type IIGP1 in the absence of the dominant negative (Figure 7B). The close co-localization of the dominant negative and wild-type IIGP1 may suggest that they interact directly. Possibly the dominant negative effect is implemented by interference with GTP-dependent oligomerisation or accelerated GTP hydrolysis [38]. In many cells, IIGP1K82A and endogenous IIGP1 co-localized, also in puncta (Figure S5). TGTP1K69A-Flag carrying the same mutation at the homologous position also behaved as a dominant negative (unpublished data).
The killing of T. gondii by IFN-γ-stimulated astrocytes was also partially inhibited by expression of dominant negative IIGP1K82Actag1. At 8 h and 24 h post-infection, markedly more PVs and intracellular parasites were found in IFN-γ-induced cells transfected with IIGP1K82Actag1 than in those transfected with wild-type IIGP1ctag1 (Figure 7D; see Figure 1B). However, dominant negative IIGP1K82A did not completely block parasite killing since some disrupted PVs and cytoplasmic GRA7 signals were also seen in IIGP1K82A-transfected cells (unpublished data). We therefore asked whether other p47 GTPases, such as TGTP, would still accumulate at T. gondii PVs in these cells. TGTP did indeed accumulate strongly at PVs in some IIGP1K82A-expressing cells (Figure S5) and the maturation of TGTP-positive PVs was not inhibited (Figure S5).
Next we analyzed the inhibition of T. gondii growth by astrocytes from mice with a targeted deletion of the IIGP1 gene. Unlike mice deficient for IGTP, LRG-47, or IRG-47 [24] [25], these mice do not show marked acute or delayed susceptibility to infection with the ME49 strain of T. gondii (unpublished data). However, compared with astrocytes isolated from wild-type littermates, the IIGP1−/− astrocytes showed a limited but highly significant defect in IFN-γ-mediated restriction of intracellular T. gondii growth (Figure 7E). IIGP1 has very recently also been implicated by RNAi in resistance to Chlamydia trachomatis infection in mouse vaginal epithelium culture, another organism with a remarkably modified vacuole [44]. The incomplete loss of IFN-γ-mediated resistance suggests that additional IFN-γ-induced factors, including, of course, other p47 GTPases [22], act with IIGP1 in this system in a partly redundant manner. Our recent preliminary findings suggest TGTP as a strong candidate for this role. Dominant negative TGTPK69A blocked IFN-γ-induced T. gondii resistance to the same extent as IIGP1K82A, as measured by parasite yield per transfected cell. Furthermore, the inhibition of IFN-γ-induced T. gondii resistance was even more pronounced when assayed in IIGP1−/− astrocytes (unpublished data).
Accumulation of the Autophagy-Associated LC3 Protein at Disrupted PVs
Several recent studies have implicated autophagic processes in the destructive stage of elimination of phagosomal bacterial pathogens [45]. We therefore looked for vacuolar accumulation of the autophagy-associated LC3 protein during the elimination of T. gondii from interferon-stimulated astrocytes. An LC3 construct labeled N-terminally with GFP was transfected into interferon-stimulated cells that were subsequently infected with T. gondii. At 2 hour after infection smooth vacuoles with strong IIGP1 accumulation showed no accumulation of LC3 (Figure 8A). LC3 fluorescence was uniformly distributed in the cell. However the initially homogenous GFP-LC3 signal condensed into vesicular structures in close proximity to rough IIGP1- and TGTP-positive vacuoles. At 6 h, cells containing disrupted vacuoles with intense IIGP1 and TGTP accumulations showed concentrations of LC3 in the immediate vicinity of these vacuoles (Figure 8B, 8C, and 8D). Virtually every rough or disrupted IIGP1- or TGTP-positive vacuole showed closely apposed GFP-LC3 concentrations. Notably, the IIGP1 and TGTP signals around the vacuole did not co-localize with the GFP-LC3 positive vesicular structures (Figure 8B, 8C, and 8D, unpublished data) It therefore appears likely that the disrupted vacuoles stimulate autophagic activity, though arguably the vesicular accumulations are the primary target, not the vacuole itself. At the electron microscopical level no investing membrane structures surrounding disrupted vacuoles were seen that could be interpreted as autophagic membranes. Furthermore, the spreading of GRA7 throughout the cytoplasm in cells containing disrupted vacuoles suggests that the vacuoles at the disruption stage are not surrounded by an autophagic membrane that would confine the released GRA7.
Figure 8 Induction of Autophagosomes in Vicinity of Disrupted Vacuoles
Astrocytes were transfected with pEGFP-C3-LC3 and induced with IFN-γ. Cells were infected with T. gondi 24 h later and fixed after 2 h (A) or 6 h (B, C, and D). In cells containing only smooth IIGP1 vacuoles GFP-LC3 remained diffusely distributed throughout the cytoplasm (A). In cells containing disrupted IIGP1 PVs GFP-LC3 localizes to vesicular and filamentous structures that are in close proximity to, but do not engulf the IIGP1-positive PVs (B and C). The arrowheads point to IIGP1-positive PVs. The images shown in (A) were processed by 2D deconvolution. (D) Shows maximum projections of 3D deconvoluted Z-series.
Discussion
This study aimed to elucidate the mechanism by which the p47 GTPases confer resistance to intracellular pathogens, using T. gondii as a model. We made the surprising observation that many p47 GTPases, including IIGP1, assemble together at the T. gondii vacuole followed by the disintegration of the vacuolar compartment and killing of T. gondii. This distinguishes the mechanism of T. gondii killing clearly from phagosomal maturation. Indeed we did not detect IIGP1 on phagosomes containing dead T. gondii (Figure 2G). The accumulation of IIGP1 is therefore dependent on the active invasion of host cells by the parasite and is unlikely to be initiated by receptors such as Toll-like receptors recognizing pathogen-associated molecular patterns on the T. gondii surface [46,47]. The PV formed by T. gondii or Plasmodium represents a distinctive compartment [4]. Most host cell proteins are excluded from the PVM while a number of parasite-encoded proteins are inserted into the PVM [6,7,48]. It will be of great interest to see how cells in general and the p47 GTPases in particular recognize these alien compartments in the cytoplasm of infected cells.
A recently published study failed to show an association of IGTP and LRG-47 with vacuoles containing live or dead T. gondii [23]. We also failed to detect LRG-47 at the vacuole, but we report the clear localization of IGTP at vacuoles formed by live parasites. This discrepancy might be due to the different cell types used in the two studies (mouse astrocytes versus mouse bone marrow-derived macrophages). However in our system the association of IGTP with PVs was indeed less marked than that of the other p47 GTPases and it is thus possible that this association was missed in the experimental system of Butcher et al [23].
In IFN-γ-induced uninfected cells the p47 GTPases localize to overlapping but distinct compartments [35–37]. In T. gondii infected cells however, IIGP1, TGTP, IRG-47, GTPI, and IGTP are closely co-localized at the PVM. In the specific cases of IGTP and IIGP (and probably also TGTP), which are primarily located at the ER in interferon-treated cells in the absence of infection [35,36], it might be argued that the vacuolar association observed in our studies reflects, not repositioning of the GTPases, but rather the well-known accumulation of ER cisternae at the T. gondii vacuole [10]. In the case of IIGP1, however, the accumulation of the GTPase is far more intense than that of any of three other ER proteins examined as markers for ER accumulation at the vacuole (Figure 3). The case that IGTP is repositioned from the ER to the PV is less clear than that for IIGP1 since the intensity of this p47 at the PV is less striking than that of the other GTPases, though still relatively more intense than the ER markers. Evidently accumulation of ER cisternae at the PV also fails to account for the PV accumulation of the Golgi-localized GTPI (Figure S2) or of the cytosolic IRG-47 [35]. Although the trigger for the striking relocation of the p47 GTPases is currently unknown, it requires GTP binding at least for TGTP1 and IIGP1 since IIGP1K82A and TGTPK69A fail to relocate (Figure 7B, unpublished data). We previously reported the relocation of LRG-47 from Golgi membranes to plasma membrane ruffles triggered by phagocytosis [35], and another study reported a brefeldin A-sensitive association of LRG-47 with Mycobacterium tuberculosis-containing phagosomes [26]. It therefore appears that the initial location of the p47 GTPases in cells is a resting or storage location from which the p47 GTPases are recruited to plasma membrane-derived compartments upon pathogen uptake or invasion.
It is remarkable that a proportion of T. gondii-containing PVs with apparently normal released GRA7 do not appear to accumulate p47 GTPases, while other vacuoles in the same cell are intensely labeled. In astrocytes we report approximately 70% of vacuoles carrying IIGP1 and a slightly higher percentage carrying TGTP. This stochastic behavior is unlikely to have a purely kinetic basis since no difference was observed when T. gondii infections were synchronized by centrifugation of the parasites onto the cell layers (unpublished data). We have been able to show in reconstruction experiments in fibroblasts that the p47 GTPases are the only interferon-inducible components necessary for the initiation of p47 accumulation at the PV (unpublished data). Nevertheless other constitutive cellular components required for PV accumulation may be limiting. It is also not excluded that T. gondii itself may resist the accumulation of p47 GTPases at the PV through one or more of the many components known to be secreted into the host cell during the infection process [49].
IGTP and LRG-47 have already been shown to be required for resistance against T. gondii in astrocytes and/or bone marrow-derived macrophages [22,23] and we now add IIGP1 to the list of p47 GTPases mediating cell-autonomous resistance against this organism. Furthermore IGTP, LRG-47, and IRG-47 are required for resistance against T. gondii in infected mice [24,25]. The participation of so many highly diversified members of the p47 GTPase family in resistance to T. gondii and their co-localization on the PV suggests that they may act in a cooperative manner. Furthermore the individual p47 GTPases are non-equivalent against different pathogens [24–27,44] suggesting a complex relationship between individual members of the p47 family. We have been able to show that interactions with other p47 GTPases are required for the accumulation of IIGP1 at the PV during T. gondii infection (unpublished data). Thus the normal function of the p47 GTPases is indeed interactive, with some members of the family possibly fulfilling a regulatory function and others an effector function. Such a situation could explain the apparent anomaly that the two p47 GTPases, whose elimination leads to the strongest phenotypes in T. gondii infection, are IGTP and LRG-47, one of which, IGTP, is only weakly associated with the PV, possibly via ER accumulation, and the other (LRG-47) is probably not associated with the PV at all.
IIGP1 directly associates with the PVM and localizes to apparently PVM-derived vesicles with an electron-dense coat (Figures 4A and 6). In view of the intense concentration of p47 GTPases on the PVM it is plausible that the electron-dense coat indeed consists of p47 GTPases. Biochemically, IIGP1 shows micromolar affinities for nucleotides, nucleotide-dependent oligomerization, and cooperative GTP hydrolysis [38,39]. These properties relate IIGP1 to the dynamin family of GTPases with a well established role in membrane fission and deformation processes [40]. It is therefore conceivable, though not yet formally shown, that IIGP1 and also probably other p47 GTPases act directly on the PVM causing its deformation and vesiculation and thereby disruption. Vesiculation on the scale observed may lead to the net abstraction of enough material from the PVM to cause loss of membrane integrity. The sequestration by T. gondii of ER cisternae to the PVM [10] may allow the supply of new lipids to the PVM at a high rate as a defense mechanism. Transport of membrane material carrying IIGP1 and T. gondii proteins from the PV (Figures 5 and 6) could entail active interaction with microtubules, and association has been shown between IIGP1 and the microtubule motor binding protein, Hook-3 [50]. Active transport of PVM-derived material could give rise to the observed long filamentous structures emanating from the vacuole (Figure 4). However, filamentous projections containing T. gondii-derived proteins, which were still connected to the PVM, have also been detected in cells not induced with IFN-γ permissive for T. gondii replication [42,51].
The parasite itself deteriorates after the disruption of its vacuole. It remains to be established whether the p47 GTPases also contribute to perforation of the T. gondii plasma membrane or whether the p47 GTPase-dependent removal of the protective PVM renders the parasite accessible to cytosolic factors mediating its disintegration. We have been unable to find convincing evidence that autophagy participates at this stage in pathogen elimination. LC3 accumulates in the vicinity of disrupted vacuoles (Figure 8) but does not appear to surround the pathogen and no profiles resembling autophagic vacuoles were seen engulfing disrupted vacuoles by cryo-electron microscopy. While it is not clear why the cytosol may be an inimical environment for T. gondii, there are precedents showing that bacteria that normally replicate in modified phagosomes are inviable when introduced into the cytosol [52,53]. Interestingly, the anti-microbial activity of the host cell cytosol was cell-type dependent and, in RAW 264.7 macrophages, induced by the pre-exposure of the host cell to the pathogen [53]. There is no independent experience of the viability of T. gondii released into the cytosol. However, the intensive modification of the PV by the parasite surely indicates a high level of co-adaptation between the organism and its constructed microenvironment.
The disruption of T. gondii PVs accompanied by cytoplasmic dissemination of parasite-encoded proteins such as GRA7 (Figure 5) may also accelerate presentation of antigens to the adaptive immune system via the class I pathway. Release of GRA7 into the cytosol and association with host cytoplasmic structures has been reported before in the absence of interferon treatment. In this case, however, GRA7 release occurred much later, possibly coincident with the tachyzoite-bradyzoite transition. The possible implication of this late GRA7 release for class I antigen presentation was also noted [42]. A recent study showed the MHC class I restricted, TAP-dependent presentation of a T. gondii-encoded protein that was secreted into the vacuolar space by splenocytes of infected mice [60]. It is conceivable that the p47 GTPase-dependent disruption of the PVs contributes to the accessibility of this protein to the antigen presentation machinery.
We have identified a previously unknown mechanism of cell-autonomous resistance in mice against the intracellular protozoan pathogen T. gondii. This involves a direct attack on the PVM leading to vesiculation and ultimately disruption of the membrane, and requires the participation of multiple interferon-inducible p47 GTPases. In view of the wide sequence divergence and distinctive resistance properties associated with the different p47 GTPases [28] it will be of interest to find out whether a role in membrane vesiculation is their exclusive mode of action on pathogen-containing vacuolar compartments, including phagosomal derivatives, or whether their contribution to pathogen resistance is more diverse than this. Thus MacMicking et al. [26] have concluded from their own studies on resistance in mice to Mycobacterium tuberculosis that LRG-47 acts in the interferon-stimulated cell to accelerate phagosomal acidification, an activity which is difficult to reconcile with the phenomena observed here, where the T. gondii PVs never enter the lysosomal system as defined by the acquisition of LAMP1.
As we have shown [28], there are 23 p47 GTPase genes in mice, of which at least 14 are inducible by interferon. Only four of these (IGTP, LRG-47, IRG-47, and IIGP1) have yet been analyzed experimentally to any significant extent and all these have been shown to be active, each in its distinctive way, in resistance to vacuolar pathogens. They have so far been implicated in resistance to a remarkable range of major pathogens (Gram negative bacteria: Salmonella typhimurium; Gram positive bacteria: Listeria monocytogenes; Mycobacteria: M. tuberculosis and Mycobacterium avium; Protozoa: T. gondii, Leishmania donovani, Trypanosoma cruzi) [24–27,54]. There is no reason to doubt that the complexity and functional diversity of the p47 resistance system will be further extended as new family members are analyzed in detail. It is evident that the entire structure of immunity in mice against vacuolar pathogens is critically dependent on the effector mechanisms delivered by the p47 GTPases. Against this background, therefore, it is quite an extraordinary fact that the p47 resistance system is completely absent in man [28]. The implications of this remarkable difference in the deployment of immune mechanisms between man and his principal experimental model organism, the mouse, will require extensive experimental analysis.
Materials and Methods
In vitro passage of T. gondii.
Tachyzoites from T. gondii strain ME49 were maintained by serial passage in confluent monolayers of human foreskin fibroblasts (HS27, ATCC). Following infection of fibroblasts, 3 days later parasites were harvested from the culture supernatant and purified from host cell debris by differential centrifugation (5 min at 50 × g, 15 min at 500 × g). Parasites were resuspended in medium and immediately used for inoculation of host cells.
Preparation and culture of murine primary astrocytes.
Primary cultures of murine astrocytes were obtained from the brains of newborn C57BL/6 mice as previously described [55]. Briefly, following removal from the meninges, cortices were minced and a single cell suspension was prepared by triturating the tissue with a Pasteur pipette. The cell suspension was filtered through a 70 μm cell strainer and centrifuged at 200 × g. Cells were resuspended in DMEM/10% FCS/2 mM L-glutamine/50 μM 2-mercaptoethanol and seeded in 6-well plates at 1 × 106 cells/well. After reaching confluence at day 7 to 10 of culture, cells were harvested by trypsinization and replated at a concentration of 0.3 × 106 cells/ml in tissue-culture plates or onto glass cover slips. Cells were used for experiments 2–7 days later. Approximately 90% of the cells were glial fibrillary acidic protein-positive astrocytes as controlled by immunofluorescence staining. All animal experiments were conducted according to institutional guidelines.
Generation of C57BL/6 astrocytes bearing a targeted deletion of IIGP1.
A targeted deletion was introduced into Bruce4 C57BL/6 ES cells [56] by homologous recombination. The targeting construct flanked the long coding exon of IIGP1 with LoxP sites in the same orientation; the whole open reading frame of IIGP1, including the neomycin resistance gene used for in vitro selection, was excised in vivo by crossing to a C57BL/6 Cre-deleter strain [57]. Astrocytes were prepared from individually typed newborn mice derived from heterozygous matings segregating for the deleted chromosome. A detailed description of the targeting strategy and the phenotype of the targeted mice is in preparation.
Infection of astrocytes with T. gondii.
Murine astrocytes were stimulated with IFN-γ (R&D Systems, Minneapolis, Minnesota, United States) at 2–200 U/ml as indicated for 24 h prior to infection while control cultures were left untreated. In some experiments astrocytes were transfected with the indicated expression construct using Fugene6 reagent (Roche, Basel, Switzerland) and simultaneously stimulated with IFN-γ. For determination of T. gondii growth via uptake of [3H]-uracil, astrocytes were inoculated with T. gondii at a multiplicity of infection (MOI) of 1 (Figure 1A) or MOI of 0.1 (Figure 7E). After 48 h (Figure 1A) or 24 h (Figure 7E) of incubation, cultures were labeled with 1μCi/well [3H]-uracil (Hartmann Analytical) for an additional 20 h (Figure 1A) or 24 h (Figure 7E). The amount of incorporated uracil, directly proportional to the parasite growth [58], was determined by liquid scintillation counting.
For immunostaining, astrocytes were inoculated with T. gondii at a MOI of 5 or 10 for a maximum of 2 h. At that time point, extracellular parasites were removed by extensive washing with PBS. Cells were then either fixed or incubated further (up to 24 h post-infection) in fresh medium in the absence or presence of IFN-γ before fixation.
Vacuoles containing intracellular parasites were identified by immunostaining for the T. gondii dense granule protein GRA7, a 29 kDa-predicted transmembrane protein that is released into the vacuole by intracellular parasites shortly after invasion and associates with the intravacuolar network and the PV membrane [41,42]. Intracellular parasites were counted by DAPI staining.
Generation of expression constructs.
The coding regions of IIGP1, TGTP, and LC3 were amplified by PCR from full-length cDNAs described in [30] from IFN-γ-stimulated mouse embryonic fibroblasts according to standard procedures using Pfu-polymerase (Promega, Madison, Wisconsin, United States) and primers from Invitrogen Life Technologies (Carlsbad, California, United States). Restriction enzymes were from New England Biolabs (Beverly, Massachusetts, United States). The PCR fragments were cloned into the SalI site of pGW1H (British Biotech, Oxford, United Kingdom). Mutations were introduced into the coding regions according to the “QuikChange” site-directed mutagenesis kit (Stratagene, La Jolla, California, United States) protocol. All constructs were verified by sequencing. The ctag1 C-terminal modification of IIGP1 replaces the last two residues (RN) with the sequence KLGRLERPHRD.
Serological reagents.
Serological reagents used were: Anti-IIGP1 165 rabbit antiserum [35], anti-IIGP1 10E7, and 10D7 mouse monoclonal antibodies (mAb), anti-IGTP I68120 mAb (BD Transduction Laboratories, Lexington, Kentucky, United States), anti-TGTP1 A20 goat antiserum (Santa Cruz Biotechnology, Santa Cruz, California, United States), anti-LRG-47 A19 goat antiserum (Santa Cruz), anti-GTPI H53 rabbit antiserum raised against the N-terminal peptide MEEAVESPEVKEFEY, anti-IRG-47 2078 rabbit antiserum raised against the peptides CKTPYQHPKYPKVIF, and CDAKHLLRKIETVNVA, anti-T. gondii rabbit antiserum (BioGenex, San Ramon, California, United States), anti-LAMP1 1D4B rat mAb (University of Iowa, Iowa City, Iowa, United States), anti-GRA7 5–241–178 mouse mAb (gift from R. Ziemann, Abbott Laboratories, Abbot Park, Illinois, United States) [41], anti-ROP2/3/4 T24A7 mouse mAb (gift of J. Dubremetz, Montpellier, France) [59], anti-ctag1 2600 rabbit antiserum raised against the peptide CLKLGRLERPHRD, anti-ERP60 rabbit antiserum (gift from T. Wileman, BBSRC, Pirbright, United Kingdom), SPA-265 anti-calnexin rabbit antiserum (Stressgene), anti-PDI mAb (BD Transduction Laboratory), anti-Gm130 mAB (BD Transduction Laboratory), goat anti-mouse Alexa 546/488, goat anti-rabbit Alexa 546/488, donkey anti-goat Alexa 546/488, donkey anti-mouse Alexa 488, donkey anti-rabbit Alexa 488, donkey anti-rat Alexa 488, goat anti-rabbit Alexa 680 (Molecular Probes, Eugene, Oregon, United States).
Immunofluorescence analysis.
Cells were washed with PBS and fixed in 3% paraformaldehyde for 20 min at room temperature. Cells were permeabilized with 0.1% saponin and blocked with 3% BSA (Roth). The cells were analyzed using an Axioplan II fluorescence microscope (Zeiss, Oberkochen, Germany) equipped with a cooled CCD camera (Quantix, Photometrix, Tucson, Arizona, United States). Image processing and 2D deconvolution was done with the Metamorph software (Version 4.5r3, Universal Imaging, Downington, Pennsylvania, United States). For 3D deconvolution the Auto Deblur software (version 6.001, AutoQuant Imaging, Watervliet, New York, United States) was used.
Electron microscopy.
For cryo-electron microscopy astrocytes were treated as indicated and fixed at 6 h post infection with 4% paraformaldehyde in 0.2M HEPES (pH 7.4). Cells were released from the plastic with a Teflon edge, pelleted in 1% gelatin in PBS and washed several times with 100 mM glycine in PBS. Pellets were incubated overnight in 2.3 M sucrose in PBS and shock frozen in liquid nitrogen. Ultra-thin sections were cut at −120 C and placed on formvar coated grids. Immunogold labeling was performed as follows: Sections were blocked in PBG (0.8% BSA, 1% fish skin gelatin in PBS) followed by incubation with the primary antibody diluted in PBG. After seven washes with PBS, grids were incubated with protein A coupled to 10 nm gold particles diluted in PBG. Grids were then washed ten times with PBS and ten times with water, and finally contrasted with methylcellulose/uranacetate (8.5 parts/1.5 parts).
Western blotting.
Astrocytes were treated as indicated and lysed in 1% T×100/PBS containing “complete mini” protease inhibitor cocktail (Boehringer, Ingelheim, Germany). Lysates were spun at 23,000 × g for 15 min at 4C and supernatants were subjected to SDS-PAGE. Proteins were transferred to a nitrocellulose membrane and probed for IIGP1 by incubation with the anti-IIGP1 165 antiserum. Bound primary antiserum was detected using goat anti-rabbit-Alexa-680 antiserum and blots were scanned using the Odyssey system (LI-COR Biotechnology, Lincoln, Nebraska, United States) (IIGP1) or by conventional chemiluminescence.
Supporting Information
Figure S1 Specificity of Anti-p47 GTPase Antisera
Astrocytes were induced with IFN-γ, or left untreated and infected with T. gondii 24 h later for 2 h. Cells were fixed and stained for the indicated p47 GTPase. With the exception of the A19 anti-LRG-47 antiserum (E) no or a very low signal was detected in uninduced cells. The arrowheads point to representative vacuoles in the uninduced cells. The images for the −FN-γ controls were taken with the same exposure time as the +IFN-γ images.
(1.0 MB JPG)
Click here for additional data file.
Figure S2 Click here for additional data file.
GTPI Localizes to the Golgi Apparatus
(A) Astrocytes were induced with IFN-γ, fixed 24 h later, and stained for GTPI and Gm130. The GTPI signal accurately, though not perfectly, overlaps with Gm130 localizing to the cis-Golgi. An additional vesicular signal throughout the cytoplasm and a weak signal at plasma membrane ruffles are also apparent. The nucleus was stained with DAPI.
(B) Astrocytes were induced with IFN-γ and simultaneously transfected with an IIGP1ctag1-expression construct. Cells were fixed 24 h later and stained for ctag1 using the anti-ctag1 2600 antiserum.
(200 KB JPG)
Figure S3
T. gondii in Infected IFN-γ Induced and Uninduced Primary Astrocytes
(A) Lysates of astrocytes induced with the indicated concentrations of IFN-γ for 24 h were probed for the indicated proteins by Western blotting.
(B) Transmission electron micrograph of ultra-thin cryosectioned uninduced astrocytes infected with T. gondii 6 h post-infection. Cells were labeled with the 165 anti-IIGP1 antiserum and protein A coupled to 10 nm gold particles (open white arrowhead: T. gondii inner membrane complex [IMC]; filled white arrowhead: T. gondii plasma membrane; black arrowhead: PVM) Bars: 0.5μm and 0.25μm (inset).
(2.5 MB JPG)
Click here for additional data file.
Figure S4 Disintegration of T. gondii in IFN-γ-Induced Cells
(A) IFN-γ- induced astrocytes 2 h post-infection with T. gondii stained for IIGP1 (green) and ROP2/3/4 (red). The arrowheads point to IIGP1-positive, but ROP2/3/4-low vacuoles.
(B) IFN-γ- induced astrocytes 6 h post-infection with T. gondii. Cells were stained for IIGP1 with 10E7 (green) and an anti-T. gondii antiserum (red). The white arrowhead points to a disrupted parasite. Nuclei were stained with DAPI.
(628 KB JPG)
Click here for additional data file.
Figure S5 Co-localization of IFN-γ-Induced and Transfected p47 GTPases at the PV
IFN-γ induced astrocytes were infected with T. gondii and fixed 2 h (A) or 6 h (B) post-infection. Cells were stained with 10D7 for IIGP1 (green) and A20 for TGTP1 (red). (C) Astrocytes were transfected with IIGP1ctag1 and stimulated with IFN-γ. Cells were infected 24 h later with T. gondii, fixed 2 h post-infection, and stained for ctag1 (green) and GRA7 (red). (D) Astrocytes were transfected with IIGP1K82Actag1 and stimulated with IFN-γ. Cells were infected 24 h later with T. gondii, fixed 2 h post-infection, and stained for ctag1 (green) and TGTP1 (red). (E) Astrocytes were transfected with the indicated expression constructs and stimulated with IFN-γ. Cells were infected 24 h later with T. gondii at a MOI of 5, fixed at the indicated time points, stained for TGTP1 and ctag1. The number of TGTP1-positive vacuoles in transfected cells displaying a smooth, rough, or disrupted morphology was counted. (White bars: smooth vacuoles; grey bars: rough vacuoles, black bars: disrupted vacuoles.) Nuclei were stained with DAPI. (B) Shows a maximum projection of a 3D deconvoluted Z-series.
(2.8 MB JPG)
Click here for additional data file.
Table S1 List of p47 GTPases Discussed in this Article
(28 KB DOC)
Click here for additional data file.
Accession Numbers
The Mouse Genome Initiative Database (http://www.ncbi.nlm.nih.gov/Genbank/) accession numbers for genes and gene products discussed in this paper are GTPI (AJ007972, CAA 07799), IGTP (U53219, AAC53007), IIGP1 (AJ007971, CAA 07798), IRG-47 (NM_008330, NP_032356), LRG-47 (NM_008326, NP_032352), and TGTP (L38444, AAA64914).
We thank A. Mausberg for help in the preparation of astrocytes. We are especially grateful to R. Ziemann (Abbott Laboratories) and J.F. Dubremetz for their generous gifts of antibodies. We thank Natasa Papic for developing the ctag1 tagged constructs and specific antiserum. SM, IP, and JCH were supported by the Deutsche Forschungsgemeinschaft through programs SP1110 “Innate Immunity” and SFB635 “Post-Translational Control of Protein Function,” and by the Land Nordrhein-Westfalen, through the University of Cologne. IP was additionally supported by a stipend from the DFG Graduiertenkolleg “Genetics of Cellular Systems.”
Competing interests. The authors have declared that no competing interests exist.
Author contributions. SM, GR, and JCH conceived and designed the experiments. SM, IP, GS, and GR performed the experiments. SM, IP, GG, GR, and JCH analyzed the data. SM, IP, JZ, GG, GR, and JCH contributed reagents/materials/analysis tools. SM and JCH wrote the paper.
Abbreviations
ERendoplasmic reticulum
MOImultiplicity of infection
PDIprotein disulphide isomerase
PVparasitophorous vacuole
PVMparasitophorous vacuole membrane
==== Refs
References
Underhill DM Ozinsky A 2002 Phagocytosis of microbes: Complexity in action Annu Rev Immunol 20 825 852 11861619
Morisaki J Heuse E Sibley LD 1995 Invasion of Toxoplasma gondii occurs by active penetration of the host cell J Cell Sci 108 2457 2464 7673360
Dobrowolski J Sibley LD 1996
Toxoplasma invasion of mammalian cells is powered by the actin cytoskeleton Cell 84 933 939 8601316
Sibley LD 2004 Intracellular parasite invasion strategies Science 304 248 253 15073368
Suss-Toby E Zimmerberg J Ward G 1996
Toxoplasma invasion: The parasitophorous vacuole is formed from host cell plasma membrane and pinches off via a fission pore Proc Natl Acad Sci U S A 93 8413 8418 8710885
Charron AJ Sibley LD 2004 Molecular partitioning during host cell penetration by Toxoplasma gondii
Traffic 5 855 867 15479451
Mordue DG Desai N Dustin M Sibley LD 1999 Invasion by Toxoplasma gondii establishes a moving junction that selectively excludes host cell plasma membrane proteins on the basis of their membrane anchoring J Exp Med 190 1783 1792 10601353
Joiner K, Fuhrman S, Miettinen H, Kasper L, I M 1990
Toxoplasma gondii: Fusion competence of parasitophorous vacuoles in Fc receptor-transfected fibroblasts Science 249 641 646 2200126
Mordue D HÂkansson S Niesman I Sibley L 1999
Toxoplasma gondii resides in a vacuole that avoids fusion with host cell endocytic and exocytic vesicular trafficking pathways Exp Parasitol 92 87 99 10366534
Sinai AP Webster P Joiner KA 1997 Association of host cell endoplasmic reticulum and mitochondria with Toxoplasma gondii parasitophorous vacuole membrane: A high affinity interaction J Cell Sci 110 2117 2128 9378762
Sibley L Niesman I Parmley S Cesbron-Delauw M 1995 Regulated secretion of multi-lamellar vesicles leads to formation of a tubulo-vesicular network in host-cell vacuoles occupied by Toxoplasma gondii
J Cell Sci 108 1669 1677 7615684
Mercier C Cesbron-Delauw M Sibley L 1998 The amphipathic alpha helices of the Toxoplasma protein GRA2 mediate post-secretory membrane association J Cell Sci 111 2171 2180 9664038
Schwab J Beckers C Joiner K 1994 The parasitophorus vacuole membrane surrounding intracellular Toxoplasma gondii functions as a molecular sieve Proc Natl Acad Sci U S A 91 509 513 8290555
Scharton-Kersten T Wynn T Denkers E Bala S Grunvald E 1996 In the absence of endogenous IFN-gamma, mice develop unimpaired IL-12 responses to Toxoplasma gondii while failing to control acute infection J Immunol 157 4045 4054 8892638
Tsuji M Miyahira Y Nussenzweig R Aguet M Reichel M 1995 Development of antimalaria immunity in mice lacking IFN-gamma receptor J Immunol 154 5338 5344 7537305
Scharton-Kersten T Nakajima H Yap G Sher A Leonard WJ 1998 Infection of mice lacking the common cytokine receptor gamma-Chain (gamma-c) reveals an unexpected role for CD4+ T lymphocytes in early IFN-gamma-dependent resistance to Toxoplasma gondii
J Immunol 160 2565 2569 9510152
Gazzinelli R Wysocka M Hayashi S Denkers E Hieny S 1994 Parasite-induced IL-12 stimulates early IFN-gamma synthesis and resistance during acute infection with Toxoplasma gondii
J Immunol 153 2533 2543 7915739
Yap GS Sher A 1999 Effector cells of both nonhemopoietic and hemopoietic origin are required for interferon (IFN)-gamma- and tumor necrosis factor (TNF)-alpha-dependent host resistance to the intracellular pathogen, Toxoplasma gondii
J Exp Med 189 1083 1092 10190899
Denkers E Yap G Scharton-Kersten T Charest H Butcher B 1997 Perforin-mediated cytolysis plays a limited role in host resistance to Toxoplasma gondii
J Immunol 159 1903 1908 9257855
Halonen SK Chiu FC Weiss LM 1998 Effect of cytokines on growth of Toxoplasma gondii in murine astrocytes Infect Immun 66 4989 4993 9746608
Halonen SK Weiss LM 2000 Investigation into the mechanism of gamma interferon-mediated inhibition of Toxoplasma gondii in murine astrocytes Infect Immun 68 3426 3430 10816494
Halonen SK Taylor GA Weiss LM 2001 Gamma interferon-induced inhibition of Toxoplasma gondii in astrocytes is mediated by IGTP Infect Immun 69 5573 5576 11500431
Butcher BA Greene RI Henry SC Annecharico KL Weinberg JB 2005 p47 GTPases regulate Toxoplasma gondii survival in activated macrophages Infect Immun 73 3278 3286 15908352
Taylor GA Collazo CM Yap GS Nguyen K Gregorio TA 2000 Pathogen-specific loss of host resistance in mice lacking the IFN-gamma- inducible gene IGTP Proc Natl Acad Sci U S A 97 751 755 10639151
Collazo CM Yap GS Sempowski GD Lusby KC Tessarollo L 2001 Inactivation of LRG-47 and IRG-47 reveals a family of interferon gamma-inducible genes with essential, pathogen-specific roles in resistance to infection J Exp Med 194 181 188 11457893
MacMicking JD Taylor GA McKinney JD 2003 Immune control of tuberculosis by IFN-gamma-inducible LRG-47 Science 302 654 659 14576437
Taylor GA Feng CG Sher A 2004 p47 GTPases: Regulators of immunity to intracellular pathogens Nat Rev Immunol 4 100 109 15040583
Bekpen C Hunn CP Rohde C Parvanova I Guethlein L (2005) The interferon-inducible p47 (IRG) GTPases in vertebrates: Loss of the cell-autonomous resistance mechanism in the human lineage Genome Biol In press.
Gilly M Wall R 1992 The IRG-47 gene is IFN-g induced in B cells and encodes a protein with GTP-binding motifs J Immunol 148 3275 3281 1578148
Boehm U Guethlein L Klamp T Ozbek K Schaub A 1998 Two families of GTPases dominate the complex cellular response to IFN-g J Immunol 161 6715 6723 9862701
Taylor G Jeffers M Largaespada D Jenkins N Copeland N 1996 Identification of a novel GTPase, the inducibly expressed GTPase, that accumulates in response to interferon gamma J Biol Chem 271 20399 20405 8702776
Sorace J Johnson R Howard D Drysdale B 1995 Identification of an endotoxin and IFN-inducible cDNA: Possible identification of a novel protein family J Leukoc Biol 58 477 484 7561525
Carlow D Marth J Clark-Lewis I Teh H 1995 Isolation of a gene encoding a developmentally regulated T cell-specific protein with a guanine nucleotide triphosphate-binding motif J Immunol 154 1724 1734 7836757
Collazo C Yap GS Hieny S Caspar P Feng CG 2002 The function of gamma interferon-inducible GTP-binding protein IGTP in host resistance to Toxoplasma gondii is Stat1 dependent and requires expression in both hematopoietic and nonhematopoietic cellular compartments Infect Immun 70 6933 6939 12438372
Martens S Sabel K Lange R Uthaiah R Wolf E 2004 Mechanisms regulating the positioning of mouse p47 resistance GTPases LRG-47 and IIGP1 on cellular membranes: Retargeting to plasma membrane induced by phagocytosis J Immunol 173 2594 2606 15294976
Taylor G Stauber R Rulong S Hudson E Pei V 1997 The inducibly expressed GTPase localizes to the endoplasmic reticulum, independently of GTP binding J Biol Chem 272 10639 10645 9099712
Zerrahn J Schaible UE Brinkmann V Guhlich U Kaufmann SHE 2002 The IFN-inducible Golgi- and endoplasmic reticulum—associated 47-kDa GTPase IIGP is transiently expressed during listeriosis J Immunol 168 3428 3436 11907101
Uthaiah RC Praefcke GJK Howard JC Herrmann C 2003 IIGP1, an interferon-g inducible 47 kDa GTPase of the mouse, showing cooperative enzymatic activity and GTP-dependent multimerisation J Biol Chem 278 29336 29343 12732635
Ghosh A Uthaiah R Howard J Herrmann C Wolf E 2004 Crystal structure of IIGP1: A paradigm for interferon-inducible p47 resistance GTPases Mol Cell 15 727 739 15350217
Praefcke GJK McMahon HT 2004 The dynamin superfamily: Universal membrane tubulation and fission molecules? Nat Rev Mol Cell Biol 5 133 147 15040446
Bonhomme A Maine GT Beorchia A Burlet H Aubert D 1998 Quantitative immunolocalization of a P29 protein (GRA7), a new antigen of Toxoplasma gondii
J Histochem Cytochem 46 1411 1422 9815283
Fischer HG Stachelhaus S Sahm M Meyer HE Reichmann G 1998 GRA7, an excretory 29 kDa Toxoplasma gondii dense granule antigen released by infected host cells Mol Biochem Parasitol 91 251 262 9566518
Sinai AP Joiner KA 2001 The Toxoplasma gondii protein ROP2 mediates host organelle association with the parasitophorous vacuole membrane J Cell Biol 154 95 108 11448993
Nelson DE Virok DP Wood H Roshick C Johnson RM 2005 Chlamydial IFN-gamma immune evasion is linked to host infection tropism Proc Natl Acad Sci U S A 102 10658 10663 16020528
Levine B 2005 Eating oneself and uninvited guests: Autophagy-related pathways in cellular defense Cell 120 159 162 15680321
Del Rio L Butcher BA Bennouna S Hieny S Sher A 2004
Toxoplasma gondii triggers myeloid differentiation factor 88-dependent IL-12 and chemokine ligand 2 (monocyte chemoattractant protein 1) responses using distinct parasite molecules and host receptors J Immunol 172 6954 6960 15153515
Bowman CC Rasley A Tranguch SL Marriott I 2003 Cultured astrocytes express toll-like receptors for bacterial products Glia 43 281 291 12898707
Lingelbach K Joiner K 1998 The parasitophorous vacuole membrane surrounding Plasmodium and Toxoplasma: An unusual compartment in infected cells J Cell Sci 111 1467 1475 9580555
Bradley PJ Ward C Cheng SJ Alexander DL Coller S 2005 Proteomic analysis of rhoptry organelles reveals many novel constituents for host-parasite interactions in Toxoplasma gondii
J Biol Chem 280 34245 34258 16002398
Kaiser F Kaufmann SHE Zerrahn J 2004 IIGP, a member of the IFN inducible and microbial defense mediating 47 kDa GTPase family, interacts with the microtubule binding protein hook3 J Cell Sci 117 1747 1756 15075236
Dubremetz JF Achbarou A Bermudes D Joiner KA 1993 Kinetics and pattern of organelle exocytosis during Toxoplasma gondii/ host-cell interaction Parasitol Res 79 402 408 8415546
Goetz M Bubert A Wang G Chico-Calero I Vazquez-Boland JA 2001 Microinjection and growth of bacteria in the cytosol of mammalian host cells Proc Natl Acad Sci U S A 98 12221 12226 11572936
Beuzon CR Salcedo SP Holden DW 2002 Growth and killing of a Salmonella enterica serovar Typhimurium sifA mutant strain in the cytosol of different host cell lines Microbiology 148 2705 2715 12213917
Feng CG Collazo-Custodio CM Eckhaus M Hieny S Belkaid Y 2004 Mice deficient in LRG-47 display increased susceptibility to mycobacterial infection associated with the induction of lymphopenia J Immunol 172 1163 1168 14707092
Fischer HG Nitzgen G Reichmann G Hadding U 1997 Cytokine responses induced by Toxoplasma gondii in astrocytes and microglial cells Eur J Immunol 27 1539 1548 9209508
Kontgen F Suss G Stewart C Steinmetz M Bluethmann H 1993 Targeted disruption of the MHC class II Aa gene in C57BL/6 mice Int Immunol 5 957 964 8398989
Schwenk F Baron U Rajewsky K 1995 A cre-transgenic mouse strain for the ubiquitous deletion of loxP-flanked gene segments including deletion in germ cells Nucleic Acids Res 23 5080 5081 8559668
Pfefferkorn ER Pfefferkorn LC 1977 Specific labeling of intracellular Toxoplasma gondii with uracil J Protozool 24 449 453 21288
Sadak A Taghy Z Fortier B Dubremetz JF 1988 Characterization of a family of rhoptry proteins of Toxoplasma gondii
Mol Biochem Parasitol 29 203 211 3045541
Gubbels MJ Striepen B Shastri N Turkoz M Robey EA 2005 Class I major histocompatibility complex presentation of antigens that escape from the parasitophorous vacuole of Toxoplasma gondii
Infect Immun 73 703 711 15664908
|
16304607
|
PMC1287907
|
CC BY
|
2021-01-05 12:11:29
|
no
|
PLoS Pathog. 2005 Nov 18; 1(3):e24
|
utf-8
|
PLoS Pathog
| 2,005 |
10.1371/journal.ppat.0010024
|
oa_comm
|
==== Front
PLoS PathogPLoS PathogppatplpaplospathPLoS Pathogens1553-73661553-7374Public Library of Science San Francisco, USA 1630460810.1371/journal.ppat.001002605-PLPA-RA-0052R1plpa-01-03-03Research ArticleImmunologyGenetics/Gene ExpressionParasitologyPlasmodium
Plasmodium falciparum Variant Surface Antigen Expression Patterns during Malaria P. falciparum var Expression during Malaria
Bull Peter C 12*Berriman Matthew 3Kyes Sue 1Quail Michael A 3Hall Neil 4Kortok Moses M 2Marsh Kevin 12Newbold Chris I 1
1 Nuffield Department of Clinical Medicine, John Radcliffe Hospital, University of Oxford, Oxford, United Kingdom
2 Wellcome Trust/Kenya Medical Research Institute Collaborative Programme, Kilifi, Kenya
3 Wellcome Trust Sanger Institute, Hinxton, United Kingdom
4 The Institute for Genomic Research, Rockville, Maryland, United States of America
Burleigh Barbara EditorHarvard School of Public Health, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 18 11 2005 1 3 e2631 5 2005 11 10 2005 Copyright: © 2005 Bull 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.The variant surface antigens expressed on Plasmodium falciparum–infected erythrocytes are potentially important targets of immunity to malaria and are encoded, at least in part, by a family of var genes, about 60 of which are present within every parasite genome. Here we use semi-conserved regions within short var gene sequence “tags” to make direct comparisons of var gene expression in 12 clinical parasite isolates from Kenyan children. A total of 1,746 var clones were sequenced from genomic and cDNA and assigned to one of six sequence groups using specific sequence features. The results show the following. (1) The relative numbers of genomic clones falling in each of the sequence groups was similar between parasite isolates and corresponded well with the numbers of genes found in the genome of a single, fully sequenced parasite isolate. In contrast, the relative numbers of cDNA clones falling in each group varied considerably between isolates. (2) Expression of sequences belonging to a relatively conserved group was negatively associated with the repertoire of variant surface antigen antibodies carried by the infected child at the time of disease, whereas expression of sequences belonging to another group was associated with the parasite “rosetting” phenotype, a well established virulence determinant. Our results suggest that information on the state of the host–parasite relationship in vivo can be provided by measurements of the differential expression of different var groups, and need only be defined by short stretches of sequence data.
Synopsis
Hope that it will be possible to develop a malaria vaccine is supported by the fact that individuals who have grown up in malaria endemic regions learn to carry malarial infections without suffering disease. Surprisingly little is still known about how this immunity develops. Much current research focuses on how the host develops immune responses to parasite antigens that are exposed to the host immune system. A major family of such antigens are inserted into the surface of parasite-infected erythrocytes, where they undergo antigenic switching to evade a developing antibody response. These proteins are encoded by a family of approximately 60 var genes, variants of which are present in every parasite genome.
The extreme diversity of the var genes has prevented meaningful comparison of their expression in clinical isolates. However, the authors of this paper show that var genes can be placed in groups that have a similar representation in the genomes of all parasites that the authors collected from Kenyan children. Having demonstrated an underlying similarity at the genomic level, the authors show that the var expression patterns vary markedly between different patients. The expression levels of specific groups of var genes was associated with poorly developed antibody responses in the children and a well-established parasite virulence phenotype. The study provides tools for exploring how host and parasite adapt to one another as immunity develops.
Citation:Bull PC, Berriman M, Kyes S, Quail MA, Hall N, et al. (2005) Plasmodium falciparum variant surface antigen expression patterns during malaria. PLoS Pathog 1(3): e26.
==== Body
Introduction
In sub-Saharan Africa, Plasmodium falciparum malaria infection is a major cause of childhood mortality. Adults, though still susceptible to infection, are protected against severe forms of malaria. Despite considerable attention over the last decade, this naturally acquired immunity is poorly understood at the molecular level. Even less understood is why, despite similar exposure levels, some children get severe malaria and die whereas others never succumb to life-threatening disease. Molecular tools to type infecting parasites and to give meaningful information about the host–parasite interaction in vivo are needed urgently.
P. falciparum erythrocyte membrane protein 1 (PfEMP1) plays a central role in the host–parasite interaction. Members of this family of molecules are inserted into the surface of infected erythrocytes by parasites during the asexual stage of growth. PfEMP1 molecules are encoded by a family of about 60 highly diverse var genes [1] that undergo rapid switching in vitro and are thought to be largely responsible for the well characterized phenomenon of clonal antigenic variation [2–5]. In addition, they appear to be central to changes in cytoadherence properties that lead to the sequestration of infected erythrocytes in capillary beds, potentially a key step in the pathology of severe disease [6,7]. The molecules are made up of combinations of different domains, each mediating a specific range of interactions with molecules on host endothelial cells [8–10], platelets [11], uninfected erythrocytes [12,13], and dendritic cells [14].
PfEMP1 proteins are presently the best candidates for the variant surface antigens (VSAs) proposed as targets for naturally acquired immunity to malaria [15]. Following acute disease, children develop specific immune responses to the repertoire of VSAs that caused the infection. The anti-VSA antibodies carried by the host at the time of disease impose a selection pressure on the repertoire of VSAs expressed during an infection [16–18]. Thus, naturally acquired immunity may develop through the piecemeal acquisition of a large repertoire of anti-VSA antibodies [16]. This is supported by the demonstration that PfEMP1-based vaccines provide protection against experimental infection with a specific parasite genotype [19].
PfEMP1 proteins have generally been considered to be too diverse to be of use in a malaria vaccine. This diversity appears to be generated, at least in part, by intragenic recombination between var genes [20,21], raising fears that it may be impossible to classify these genes in any meaningful sense. However, other observations suggest this diversity may be finite. First, VSAs expressed in children with severe malaria show evidence of having restricted diversity. Parasite isolates from children are recognised at different frequencies by plasma collected from the childhood population in the same geographical location. This frequency of recognition [17,22] is dependent on the immune status of the host, being negatively associated with host age and positively associated with disease severity [17,18,22,23]. Commonly recognised VSAs also appear to have a broad geographical distribution [24]. Second, complete genome sequencing of laboratory P. falciparum line 3D7 [1] revealed genetic structuring of the var repertoire within the genome. Different subsets of var genes exist that are associated with different upstream control elements [25,26] and functional properties [27,28]. Importantly, differences in the functional properties of the proteins appear to be reflected in the sequence of the Duffy-binding-like (DBL) α domain, the only var domain that is PCR amplifiable from nearly all var genes (See Figure 1A for more details).
Figure 1 Organizational Features of var Genes
(A) var gene organization. The var genes are complex, multi-domain structures composed of variable numbers of DBL domains (rectangles) of different sequence classes (DBLα, -β, -γ, -δ, and -ɛ plus several heterogeneous DBLX) and cysteine-rich interdomain regions (CIDR, ovals), also of different classes (CIDRα, -β, and -γ). Despite this complexity, the majority of var genes in the 3D7 genome can be described according to (1) whether they belong to a small set of long genes that encode PfEMP1 molecules with 6–9 domains or to a much larger set of short genes that encode short PfEMP1 molecules with only four domains; (2) their telomeric or internal position within the chromosome; (3) their direction of transcription, either towards the telomere or towards the centromere; (4) the sequence of their second most N-terminal DBL domain (DBL2), either DBL2β, DBL2δ, or DBL2γ; and (5) their association with one of five upstream (ups) regulatory sequences, upsA, upsB, upsC, upsD, or upsE [1,25,26]. The genes do not fall randomly within these categories. For example, of the long var genes, most have a DBL2β and are associated with upsA, whereas most short genes have a DBL2δ. Of the telomeric genes, most of those that are transcribed towards the telomere are associated with upsA, whereas all those transcribed towards the centromere are associated with upsB [28,33]. This apparent genetic structuring is associated with functional specialization. A subset of CIDR regions (CIDRα) situated immediately 3′ of DBLα bind to CD36 when expressed as recombinant peptides (teal shaded oval). These regions are generally replaced by non-CD36-binding CIDR regions in long var genes [27]. In 3D7, long var genes tend to be associated with a distinct subset of DBLα domains called “DBLα1” [27]. This observation is potentially very useful since DBLα and DBLα1 sequences can be PCR amplified from nearly all var genes using a universal set of degenerate primers [29]. Examples of long and short var genes found in the 3D7 genome are shown. A white square is used to represent the conserved exon 2 at the C-terminal end. Short genes are relatively conserved in domain structure. Long genes have variable organization. Two forms of PCR product were cloned and sequenced. DBLα-specific primers (DBLαAF′ and DBLαBR) were used to amplify DBLα sequences from cDNA and genomic DNA from each parasite isolate. DBLαAF′ and BetaR primers were used to amplify genomic DNA from isolates 4111 and 4161. Four further primers were used in different combinations to amplify between DBLα and ups.
(B) Distinct DBLα sequences categorized according to the number of cysteine residues in the sequence. The number of cysteines present in each distinct DBLα sequence is plotted against the length of the sequence.
(C) The location of sequence features. PoLVs are in gray. The conserved amino acids to which they are anchored are in blue. Cysteine residues are highlighted in green. The derived “sequence signature” for each clone is indicated.
A key question is whether the genetic structuring observed in 3D7 is universal enough to allow the development of a biologically meaningful var gene typing system. Here we have addressed this question using large-scale sequencing of short var sequence tags from the DBLα domain of genomic DNA and expressed transcripts from clinical parasite isolates from Kenya. Though diverse, the DBLα domain present in most genes in 3D7 can be readily amplified using a set of universal primers [29]. We demonstrate a high degree of underlying similarity between the distributions of var sequences in Kenya and var sequences present within the genome of a fully sequenced parasite isolate 3D7. Second, we show how specific sequence features can be used to classify the sequences into groups that allow the var expression patterns of different clinical parasite isolates to be compared directly.
Results
A total of 1,746 var DBLα clones were successfully sequenced from 12 field isolates using the DBLαAF′ and DBLαBR primers (Figure 1A;. see Table S1 for patient information). Of these, 722 clones were sequenced from cDNA and 1,024 from genomic DNA (Table S1; see Dataset S1 for a complete list of the sequences). Overall, a total of 878 non-identical sequences were identified (see Text S1). The sequences were too diverse for direct comparisons between isolates, and there was virtually no overlap between sequences of different isolates. We therefore focused on features from these sequences that would provide more general information about the var genes to which they belong.
DBLα Sequences Contain Semi-Conserved Features
Smith et al. [30] have previously noted from the analysis of var sequences from laboratory isolates that the different domains of var genes contain islands of homology that can be used to distinguish different classes of DBLα domains, called DBLα and DBLα1. Within the region we amplified, the positions that were most discriminatory between DBLα and DBLα1 domains corresponded with two cysteine residues [27]. We therefore analysed the amino acid composition of the amplified sequences from Kilifi, Kenya. For all the amino acids apart from cysteine, the frequencies were distributed around single modal values (data not shown). In contrast, the majority of DBLα sequences contained either two or four cysteine residues, with only a minority containing one, three, five, or six cysteines (Figure 1B; such sequences are hereafter referred to as cys2, cys4, and cysX types, respectively). This is entirely consistent with the 3D7 var genes. DBLα1 sequences are cys2 while most DBLα sequences are cys4. The functional importance of cysteine residues within the DBLα domain is supported by the observation that parasites from severe malaria cases from Brazil tend to express var genes containing DBLα domains with reduced numbers of cysteine residues [31].
Next, amino acid motifs occurring at four fixed positions within the sequenced regions were chosen. These will be referred to hereafter as positions of limited variability 1 to 4 (PoLV1–4; Figure 1C). Each PoLV was four amino acids in length and situated directly adjacent to conserved amino acid residues at the fringes of the previously defined islands of homology. Thus, each PoLV was located at identical relative positions within each DBLα sequence. The PoLV motifs and cysteine count were used as features to classify the DBLα sequences further.
Sequences from Kilifi and from a Laboratory Isolate Contain Similar Distributions of Semi-Conserved Features
The distribution of DBLα features between different var genes was examined in the full genome sequence of 3D7 [1]. In the entire 3D7 repertoire of 59 var genes there are 17 variants of PoLV1, six of PoLV2, 13 of PoLV3, and eight of PoLV4. We compared the DBLα features in 3D7 with those found among different Kilifi sequences. The majority of Kilifi sequences contained PoLV motifs that were found in the 3D7 genome (Figure 2A; Text S2). Furthermore, there was a close similarity between the distribution of PoLV motifs among var genes from the 3D7 genome and among the Kilifi sequences (Figure 2A). In both sets of sequences a similar hierarchy was evident in the frequency of variants of each sequence feature, with the same features being common and rare in each. The similarity between Kilifi and 3D7 sequences extended to the associations between different DBLα features. For example, Figure 2B shows the tendency of the different PoLV motifs to be associated with cys2 sequences. The same PoLV motifs tended to be associated with cys2 sequences among both 3D7 and Kilifi sequences, and the overall pattern of positive and negative associations was strikingly similar (Mantel-Haenszel test, p = 0.000002).
Figure 2 Conservation of Sequence Features between 3D7 and Kilifi Field Isolates
(A) Distribution of sequence features within the 3D7 genome (black bars) and in Kilifi sequences (green bars). Sequence features not shared between Kenyan isolates and 3D7 are marked “other”.
(B) Relationships between sequence features in sequences from Kilifi and 3D7: distribution of sequence features among cys2 sequences relative to those with non-cys2 sequences. The Cramer's V statistic (y-axis) indicates whether each of the listed sequence features was positively (V > 0) or negatively (V < 0) associated with cys2 sequences. Sequences from Kilifi are indicated with green bars; those from 3D7 are indicated by black bars.
(C) The distribution of DBLα sequence features within var genes containing a DBL2β domain. PCR was performed on genomic DNA from two field isolates, 4161 and 4111 (see Figure 1A for amplification details). The percentage of distinct sequences containing each of the features listed is shown for Kilifi sequences (green bars) and is compared to the distribution of similar sequences from the 3D7 genome sequence (black bars).
To test whether similarity between Kilifi and 3D7 sequences extended as far as the next downstream DBL region (see Figure 1A), genomic DNA from two field isolates, 4111 and 4161, was amplified with primers DBLαAF′ and BetaR located within the DBLα and DBL2β regions, respectively. Cloned PCR products were sequenced at both ends to determine the DBLα sequence at the 3′ end and confirm correct priming within DBL2β at the 5′ end. Figure 2C shows the distribution of DBLα features among distinct sequences. Among these clones there was a clear bias towards sequence features that are associated with DBL2β in the 3D7 genome, namely PoLV1LFLG, PoLV2LREA, PoLV4PTNL, and cys2. A similar conservation was evident upstream (see Text S3). Taken together, these observations suggest that despite being extremely diverse, DBLα sequences are built around a finite collection of building blocks whose relationships with one another follow underlying ground rules.
Assignment of DBLα Sequences to Groups
To simplify comparisons between the different isolates and to summarize the profile of expression, we sought an algorithm to assign the sequences to groups. Though in 3D7 cys2-type DBLα sequences correspond very well with those that were previously classified as DBLα1 [27], we did not expect to identify additional discrete subgroups of sequence because of the high frequency of recombination between var genes [20,21,32]. However, inspection of the sequences suggested an approach to identifying subgroups. As shown in Figure 1B, cys2 DBLα sequences were significantly shorter than cys4 DBLα sequences ( Mann-Whitney U test, p < 0.0001). This is consistent with these forming distinct sequence groups. We considered the possibility that additional sequence features may exist that are independently associated with DBLα sequence length. Using logistic regression analysis, two such groups of sequence features were identified (see Materials and Methods and Text S4). These were PoLV1MFK* and PoLV2*REY (with the asterisk marking degenerate positions). PoLV2*REY was associated with short sequences in both cys2 and cys4 sequences. PoLV1MFK* was found exclusively in cys2 sequences and was independently associated with short sequences (Figure 3A). Among the cys2 sequences there was a complete absence of sequences that contained both PoLV1MFK* and PoLV2*REY. This is a significant departure from a random distribution (Fisher's exact test, p < 0.001), suggesting that these features define subgroups of cys2 sequence.
Figure 3 Sequence Groups
(A) DBLα sequences were divided into six sequence groups: sequence groups 1–3 are those that contain two cysteine residues (cys2), and sequence groups 4 and 5 are sequences that contain four cysteine residues (cys4). Sequence group 6 includes sequences with one, three, five, or six cysteines (cysX). Sequence groups 2 and 5 contain PoLV2*REY. Sequence group 1 contains PoLV1MFK*. The length of each distinct DBLα sequence within each sequence group is indicated.
(B and C) The distribution of 3D7 var genes in each DBLα sequence group among groups previously defined on the basis of coding and upstream non-coding regions of full-length var sequences [1,33] (B) and the overall length of the genes (C). Genes are classified as short if they have 4–5 domains and long if they have 6–9 domains (see Text S6).
We used this information to assign each sequence to one of six groups (Figure 3A). Since discrimination by number of cysteine residues corresponded well with the previous classification of DBLα regions from 3D7 [27], sequences were first divided into cys2, cys4, and cysX sequences. Cys2 sequences were then divided into those containing PoLV1MFK* (group 1), those containing PoLV2*REY (group 2), and those containing neither (group 3). Cys4 sequences were divided into those without PoLV2*REY (group 4) and those with PoLV2*REY (group 5). CysX sequences were placed in group 6. Thus, groups 1, 2, and 5 were strictly defined using two features, groups 3 and 4 were defined with one feature, and group 6 contained the remaining unusual sequences.
We tested this system of classification on full-length var gene sequences from the 3D7. The full-length var genes have previously been classified into five major groups (A to E) using both coding and upstream non-coding regions [1,28,33]. Figure 3B shows how DBLα sequences from 3D7, classified using our algorithm, are distributed between the five var gene groups. Figure 3C shows how the DBLα sequences are distributed between short and long var genes. There were striking differences in the distribution of DBLα sequences particularly comparing group A, B, and C var genes and long and short var genes (see Text S3).
To determine the relationships between the six var groups, 30 randomly chosen sequences from each group were globally aligned using ClustalW analysis. A pairwise identity matrix was then constructed with the sequences sorted into their six groups (Figure 4A). It is clear from this comparison that, despite being defined using only a small amount of sequence information, groups 1, 2, and 5 form discrete sequence groups, since more sequence identity is shared between members of the same group than between groups. The distinction between groups 1 and 5 is particularly striking. Though groups 3 and 4 do not appear to form such discrete groups, the distinction between groups 2 and 4 was marked. Group 6 does not define a discrete sequence subset when analysed globally in this way, and may contain sequences derived from a variety of different recombination events.
Figure 4 Global Comparisons of Sequences Falling in Six Sequence Groups
Using ClustalW, pair-wise sequence identity comparisons were made between 30 randomly selected, distinct sequences from each sequence group. Pair-wise comparisons are expressed in an identity matrix, in which the percent identity between pairs of sequence is represented in different shades of gray.
(A) Full-length sequence comparisons.
(B) Sequence comparisons of the region from the 5′ end of PoLV3 to the 3′ end of PoLV4.
(C) Sequence comparisons of the region from the 5′ end of PoLV1 to the 3′ end of PoLV2.
Overall, this identity matrix suggested a complex web of relationships between the different sequence groups. Visual inspection of the sequences suggested that the similarity between groups 2 and 5 extended 5′ from PoLV2*REY. Therefore, to explore the relationships between these groups, we generated two further identity matrices, the first comparing the region from the 5′ end of PoLV3 to the 3′ end of PoLV4 (Figure 4B) and the second comparing the region from the 5′ end of PoLV1 to the 3′ end of PoLV2 (Figure 4C). These two comparisons gave strikingly different pictures of the interrelatedness of these groups. From Figure 3B it is clear that cys2 sequences (groups 1–3) are distinct from cys4 sequences (groups 4 and 5). However, Figure 3C shows that this distinction breaks down in the regions between PoLV1 and PoLV2. Thus, groups 2 and 5 and groups 3 and 4 share some identity within this region. From this analysis it is unclear whether these two related pairs of groups originated from ancestral hybrid sequences or whether recombination between cys2 and cys4 still occurs. What is clear is that group1 is distinct from cys4 sequences over the entire length of the sampled region.
Sequence Groups Are Consistently Represented in Genomic DNA from Clinical Isolates but Are Differentially Expressed
Using the above system of classification each of 12 clinical parasite isolates were compared (Figure 5). Figure 5A and 5C divide all 1,746 DBLα sequences by (1) whether they were cloned from genomic DNA (Figure 5A) or cDNA (Figure 5C), (2) the parasite isolate from which they were isolated, and (3) the group to which they were assigned. The cloning frequencies of genomic sequences from each group were fairly constant between parasite isolates and close to those expected from the distribution of var genes in the 3D7 genome (Figure 5A). This suggests that the number of sequences from each group was relatively constant between different parasite genomes. In contrast, there was considerable variation in the cloning frequency of cDNA-derived sequences (Figures 5C and S1). To highlight this, parasite isolates were sorted left to right according to increasing cloning frequency of cys2 cDNA sequences. Between the 12 isolates there was a significant correlation between the expression of group 1 and group 2 (r
s = 0.67, p = 0.02), suggesting that they may be under similar expression control.
Figure 5
var Gene Expression Profiling
(A–D) Each DBLα sequence was assigned to one of six sequence groups (Figure 3). The proportion of clones that fell into each of the six groups was calculated separately for genomic clones (A) and cDNA clones (C). (A) includes the distribution of sequences from the 3D7 genome (right). (B) and (D) show for each isolate the percent of genomic DNA (B) or cDNA clones (D) corresponding to the three most dominantly cloned genomic or cDNA sequences from that isolate. Isolates are ordered left to right according to the overall proportion of cys2 clones isolated from cDNA. Underlined ID numbers correspond to children with severe malaria.
(E) Northern blots of total RNA from each of the parasite isolates. Blots were hybridized to a generic var exon 2 probe, varc, corresponding to a conserved region within all var genes. The position of var genes expressed by the laboratory parasite line Palo Alto is indicated by lines to the left of each lane. These are approximately 9 kb and 11 kb in length.
(F) The VSA antibody repertoire carried at the time of acute disease by each patient. The y-axis shows the number of a panel of six parasite isolates that were recognised by the acute plasma from each child.
Based on their distribution within the 3D7 genome (see Figure 3C), we expected cys2 sequences to be associated with long var genes [1]. To confirm this, Northern blots of total RNA from each parasite isolate were hybridized to a generic var-specific probe from the relatively conserved 5′ exon 2 region. There was a good correspondence between the size of the bands (Figure 5E) and the dominant cDNA sequences from each isolate (Figure 5D). In five of seven samples that had a dominant band less than or equal to 9 kb in length, the dominant sequence was cys4 type. In five of five samples that had a dominant band greater than 9 kb, the dominant sequence was cys2 type (Fisher exact test, two-tailed, p = 0.03).
The Parasite Rosetting Phenotype Is Associated with Expression of Group 2 Sequence
As a test of the validity of our DBLα sampling and grouping strategy, we tested whether the parasite rosetting phenotype is associated with expression of specific DBLα sequence groups. Since this phenotype is mediated by the DBLα domain of PfEMP1 [12,13], we expected that specific sequence features associated with rosetting may be associated with the sequence features used to define our groups. In support of this, a striking positive association was observed between group 2 expression and the percentage of infected erythrocytes that formed rosettes (r
s = 0.92, p < 0.001, corrected for six comparisons (Bonferroni); Figures 6, S2C, and S2D). Furthermore, the two parasite isolates with the highest rosetting rates expressed dominant group 2 sequences with the same combination of sequence features. (i.e., the sequence “signature”; see Materials and Methods, Text S5, and Figure S3 for more details; this particular sequence signature was called “sig2” in Figure S3). These highly similar sequences are shown in Figure 6D.
Figure 6 Relationships between the Expression of Each DBLα Sequence Group and Markers of the Host–Parasite Relationship
In each graph the Spearman's rank correlation coefficient (r
s) is shown for each sequence group. Significance without Bonferroni correction is indicated as follows: *, p < 0.05; **, p < 0.001.
(A) Correlation between expression of each sequence group and parasite rosetting (percent of infected erythrocytes forming rosettes).
(B) Correlation between expression of each sequence group and host VSA antibody repertoire (the number of a panel of six isolates recognised by the patient plasma).
(C) Correlation between expression of each sequence group and severe malaria.
(D) Alignments of sequences associated with parasite rosetting. Similar sequences were found to be dominant in two isolates (4140 and 4187) with the highest rosetting frequency. In isolate 4187, the two most dominant sequences (4187_dom1 and 4187_dom2) were highly similar. All three sequences shown have the same sequence signature, “sig2”: LYLD-VERY-KAIT-2-PTNL.
var Gene Expression in the Infecting Parasite Population Reflects the Host VSA Antibody Response at the Time of Disease
Previous studies predicted that as children build up a repertoire of anti-VSA antibodies, the proportion of VSAs that can be expressed by the infecting parasite population is diminished [16]. More recently, mathematical modelling has suggested that sequential expression of single VSAs can be sustained by the anti-VSA antibodies [34]. Between parasite isolates in this study, there was considerable variation in the extent to which the cDNA sequences were dominated by a small number of sequences. Figure 5D shows the extent to which the most dominant sequences from each parasite isolate accounted for the entire collection of clones from that isolate among cDNA sequences (i.e., the homogeneity of the collection of sequences; see Materials and Methods). From Figure 5D it is clear that among the cDNA clones, dominant sequences were identified from each of groups 1–5 with no striking association between any particular group and the disease severity of the infected child (see also Figures 6C, S1, S2A, and S2B).
If the expressed var genes correspond with VSAs expressed on the infected erythrocyte, then, following from previous studies, we would expect to observe a positive association between the homogeneity of the var message and the repertoire of VSA antibodies carried by each child at the time of disease (see Figure 5F). In support of this, there was evidence for such an association among the cDNA sequences (r
s = 0.81, p = 0.002; Figure 5D) but not among the gDNA sequences (r
s = −0.37, p = 0.23; Figure 5B). Previous serological studies have further led to the suggestion that a subset of relatively conserved VSAs is under particularly high immune selection [17,18]. To test whether any of the DBLα groups show evidence of being under high immune selection relative to the other groups, we tested for a negative association between the relative expression of each of the groups and the VSA antibody repertoire of the infected child. Evidence for such an association was found for sequence group 1 (r
s = −0.68, p = 0.015, p = 0.088 after Bonferroni correction for six comparisons; see Figures 6B, S2E, and S2F). Though this association clearly needs to be confirmed in larger studies, the fact that group1 sequences were relatively well conserved between isolates agrees with predictions from the previous serological data.
Discussion
Despite years of research very little is known about how the host–parasite relationship changes as naturally acquired anti-malarial immunity develops. More specifically, we lack molecular tools for measuring changes in the parasite as it adapts to the development of clinical immunity in vivo. Such tools could provide a powerful means of dissecting the protective components of host response, a first step in the identification of new vaccine candidates. A main requirement for such tools is that they can be used in field-based studies. Here we have assessed a simple approach using large-scale sequencing of short stretches of sequence from DBLα, a region that, though highly diverse, is present in the majority of var genes.
Expectations that such an approach could generate data that would reflect the host–parasite relationship at the time of disease have until recently been low. This has been due to uncertainty about whether the high recombination rate between var genes and their extreme diversity would allow meaningful comparisons between isolates [20,21,32]. The 3D7 genome sequence [1] has provided more encouraging information. The location of var genes in both internal and telomeric locations and their mixed direction of transcription set up the conditions for genetic structuring. This is supported by the existence within the 3D7 genome of two groups of var genes encoding PfEMP1 with different functional properties [27]. The two groups of genes carry different DBLα sequences defined as DBLα and DBLα1 [27]. These observations opened up the possibility of obtaining functionally relevant data from field isolates using only limited sequence data. However, it was uncertain how useful these definitions would be to field studies.
Here we analysed a large number of different sequences from the DBLα region of var genes from Kilifi, Kenya, focusing on a limited number of semi-conserved sequence features. The data strongly support the existence of an underlying order that extends from the single genome to the parasite population as a whole. This enabled us to use the 3D7 genome as a basic reference for interpretation of the field data. Overall, the similarity between Kilifi and 3D7 sequences was extensive, in terms of (1) the range of sequence features observed and their similar frequency distribution, (2) their relationships with each other within the sequence, and (3) their relationship to features outside the sequence region that was sampled. Most notably, within the region we have sequenced the main defining feature of DBLα1 sequences in 3D7 is the existence of two cysteine residues (cys2) rather than the normal four cysteines (cys4). In the field isolates, apart from a small minority, the sequences could be classified as either cys2 or cys4. These observations together help clarify an interesting earlier observation from Brazil, where DBLα sequences containing reduced numbers of cysteines (corresponding to our cys2 sequences) tended to be expressed in children with severe malaria [31].
Comparison of the lengths of the DBLα sequences revealed that cys2 types were significantly shorter than cys4 types. Other sequence features independently associated with sequence length were subsequently identified and used to place the sequences in groups, providing a simple means of classifying the sequences. The practical usefulness of these groupings is supported by the striking association between sequence group 2 expression and parasite rosetting. The rosetting phenotype is a well established virulence phenotype mediated by binding of a subset of DBLα domains to complement receptor 1 (CR1) on erythrocytes and has been found in several previous studies to be associated with severe malaria [35,36]. Surprisingly, these sequences were not related to previously identified rosetting var genes such as R29 [12] or FCR3S1.2var1 [13], which fall in groups 1 and 4, respectively, suggesting that they may represent a novel class of rosetting var genes.
In most of the isolates there were clear dominant cDNA sequences, and the dominance of particular sequences in different infections was consistent with previous studies of var gene expression from field isolates [37,38]. This challenges previous ideas based on studies of var gene expression in laboratory isolates. Previous studies suggested that all var genes may be switched on in the immature ring stages, but only one is expressed in mature stages. These data suggested a post-transcriptional level of control that would prevent meaningful data being obtained from uncultured parasite isolates [39–41]. Though Kaestli et al. [38] took the precaution to pre-select for full-length transcripts to remove the possibility of amplifying incomplete transcripts, neither Peters et al. [37] nor ourselves performed this step, suggesting that background transcription may not be a cause for concern in the interpretation of field studies.
The primary aim of sampling var gene sequences from clinical isolates was to use the information to track changes in var gene expression associated with the development of naturally acquired immunity to malaria. It is important for such studies to be carried out over a long period of time and in different geographical locations. However, as a first step, it was encouraging to find that the repertoire of VSA antibodies carried by a child at the time of disease correlated with both the tendency of the cDNA sequences to be dominated by a small number of sequences, and their bias away from a small group of relatively conserved cys2 sequences (group 1). Both these observations fit in well with previous serological and theoretical studies that suggest that theVSA antibody response both supports sequential expression of single VSAs [34] and selects against those that are most conserved. Previous serological studies have led to the suggestion that a restricted subset of commonly recognised PfEMP1 molecules are associated with both low host immunity and severe malaria [17,18,23]. In an attempt to select for the expression of such molecules in vitro, Jensen et al. [42] selected the 3D7 parasite line on IgG from malaria-exposed children. Several var genes appeared to be specifically selected by these naturally acquired antibodies. The DBLα tag regions of the majority of these were found to be cys2 sequences, though not specifically from group 1. A key question for future research is why certain var genes would be maintained in the genome if they are particularly sensitive to immune selection. If such genes have specific functional properties, it would be important to examine these in detail to assess their potential usefulness as vaccine candidates.
However, in the present study there was no clear evidence for any particular sequence group being associated with severe malaria. As noted above, Kirchgatter et al. [31] previously observed that children with severe malaria tend to express DBLα sequences with cys2 sequences, and Bian et al. [43] have observed that parasites causing severe malaria tend to express long PfEMP1 molecules [43]. In the 3D7 genome both these are characteristics shared by var genes that lie downstream of upsA control elements (see Figure 1A). These observations together with those of Jensen et al. [42] may suggest a specific role for upsA var genes in severe malaria. In the present study, the clear bias in parasites from two severe cases away from expression of cys2 DBLα sequences suggests that some caution is needed in regard to this interpretation. However, the strong association of a subgroup of cys2 sequences (group 2) with rosetting and the observation that parasites from two of the six severe cases expressed very similar group 2 dominant sequences are consistent with the idea that some children with severe malaria express a restricted subset of cys2 var genes. More samples are clearly needed to confirm this observation.
In future studies with larger numbers of parasite isolates it will be interesting to explore DBLα expression patterns in relation to other aspects of the host–parasite interaction, such as the number of parasite genotypes present, host endothelial cell binding phenotype, and various components of the host immune response. Though initially it would be important to carry out these studies using DNA sequence data, the fact that sequence groups can be defined using short sequence motifs suggests that approaches based on microarray and real-time PCR analysis could be developed to distinguish between the expression of different groups of var sequences. In addition, the close relationship between PCR product length and the sequence group of the products raises the possibility that inexpensive approaches to var expression typing might be developed using PCR product length data.
In conclusion, we have shown that var genes from both field and laboratory isolates can be classified into biologically meaningful subsets based on small blocks of semi-conserved sequence. Further sequencing of var genes from a much larger number of parasites derived from patients that have been rigorously categorised with respect to clinical presentation and parasite phenotype is clearly necessary. By focusing attention on subgroups of var genes that are associated with parasite virulence and host immune status, such studies may provide further information with implications for malaria intervention.
Materials and Methods
Study site.
The study was carried out at Kilifi District Hospital, situated 50 km north of Mombasa on the coast of Kenya. The hospital has a high-dependency ward to treat children with severe life-threatening malaria, a paediatric ward to treat children with moderate malaria, and an outpatient department to treat children with mild malaria.
Sample collection.
Children were recruited if they had a primary diagnosis of malaria and parasitaemia ≥ one trophozoite per 100 uninfected erythrocytes [17]. Isolates were collected and white blood cells removed as described previously [22]. For each isolate a sample of acute plasma was stored at −20 °C. Parasites were collected from children attending hospital between July 1998 and February 1999 and have been described previously [17,44]. Twelve patients were selected for the study: six with mild disease and six with severe disease.
Serotyping of plasma.
Plasma from each of the 12 patients was tested by agglutination assay against six parasite isolates from blood group O individuals (ID numbers 4513, 4518, 1759, 4542, 4508, and 4528) who came to hospital with malaria either between January and August 2000, or, in the case of 1759, in December 1995 [44]. The VSA antibody repertoire carried by these plasma samples was defined as the number of the six target isolates that were agglutinated.
Agglutination assays.
Parasites were cultured until they were middle to late pigmented trophozoites, as described previously [22]. Assays were performed in microtitre plates (Falcon, Becton-Dickinson, Palo Alto, California, United States) at 4% haematocrit in RPMI at a parasitaemia of 1–2 trophozoites per 100 uninfected erythrocytes in a 12.5-μl total assay volume in the presence of 2.5 μl of plasma. Cells were rotated for 1 h as described previously [22]. Assays were scored using the dry agglutinate method as described previously [17].
Rosetting assay.
Cells (0.5 μl) were resuspended in 9.5 μl of RPMI containing 5 μg/ml acridine orange. Following the addition of 2.5 μl of non-immune European serum, cells were rotated for 30 min on a vertical rotator and the entire reaction volume pipetted onto a glass slide, covered with a coverslip, and observed under a fluorescence microscope (Nikon, Tokyo, Japan). Rosetting was scored as the percentage of 100 mature trophozoites that adhered to at least two uninfected erythrocytes.
Characterization of DBLα sequences.
Pellets (100 μm) of packed infected erythrocytes were collected and, following lymphocyte and phagocyte depletion, were stored in Trizol (Invitrogen, Paisley, United Kingdom) at −30 °C. RNA was prepared as described previously [45]. To amplify var from RNA, the RNA was first treated with DNAse I (DNAse Free, Ambion, Cambridge, United Kingdom) according to the manufacturer's instructions. The DNAse was removed using Ambion DNase inactivation reagent. RNA (2 μl) was reverse transcribed using reverse transcriptase (Invitrogen SuperscriptII). For each isolate a negative control reaction was performed in the absence of reverse transcriptase to ensure that all contaminating DNA had been removed by DNAseI pre-treatment. Sufficient DNA was present in the untreated RNA sample for PCR amplification of genomic DNA. Suspended sample (1 μl) was diluted in 10 μl of water, and 1 μl was amplified directly by PCR. DBLα sequences were amplified with the following primers: DBLαAF′, GCACG(A/C)AGTTT(C*/T)GC, and DBLαBR, GCCCATTC(G/C)TCGAACCA, modified from [29] (Figure 1A). The nucleotide marked with an asterisk indicates a modification from the originally described primer DBLαAF. This change was introduced to broaden the range of sequences that can be amplified. PCR amplifications between DBL1α and DBL2β were performed using the following primers: DBLαAF′, see above, and BetaR, GA/CCCAC/TTCIGC/TCATCCA. The following conditions were used. For isolation of DBLα sequences, 35 cycles of PCR were performed in 25 μl using an annealing temperature of 42 °C and a 30-s extension time at 65 °C in the presence of 0.2 U of Amplitaq polymerase (Applied Biosystems, Foster City, California, United States) and 3 mM MgCl2 to give a product of 400 bp. Amplification between DBLα and DBL2β was performed in the presence of BioXact polymerase (Bioline, London, United Kingdom) in the presence of 3 mM MgCl2 with an annealing temperature of 50 °C and extension time of 2 min at 65 °C to give a product length of approximately 2.3 kb. Following PCR, DBLα sequences were purified using Sephacryl (Amersham Biosciences, Amersham, United Kingdom). Products obtained by amplification between DBLαAF′ and BetaR were size selected on an ethidium-bromide-stained agarose gel and purified using a Qiagen (Valencia, California, United States) gel extraction kit. DNA was ligated into either TA vector or TOPO vector (Invitrogen) and used to transform TOP10 cells. From each clinical isolate we aimed to sequence approximately 100 genomic DNA and 50 cDNA DBLα clones. Sequencing was carried out using M13 reverse and T7 primers (3 pmol) with BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems). Samples were run on Applied Biosystems 3700 or 3730 sequencing machines.
Amplification of DNA upstream of DBLα.
The following primers were used to test the relationship between (1) DBLα sequence features PoLV1MFKR (amino acid sequence motif MFKR at PoLV1; see Figure 1C) or PoLV4PTYF and (2) upstream sequences upsA or upsB (see Figure 1A). Reverse primers MK3, TCATTACGTTTAAACATATC (specific to PoLV1MFKR), and PTYF3′, ACGTAGTCAAAATATGTGG (specific to PoLV3PTYF); forward primers upsA750, AACATKGTTCTATTTTCTC, and upsB, TTGCCTCTDTTGTTATCTC, specific to upsA and upsB, respectively. All reactions were performed in the presence of 3 mM MgCl2 using an annealing temperature of 47 °C, 35 cycles, and an extension time of 1 min at 65 °C with Amplitaq polymerase in the presence of Taqstart reagent (Clontech, Becton-Dickinson). See also Text S3.
Selection of sequences for analysis.
DNA subclones were selected for analysis if at least one of the pair of sequence reads contained a single open reading frame and began and ended within previously identified “homology blocks” of DBLα [30]. From the pair of sequence reads from each clone the best quality single read was chosen for analysis. Sequences selected for analysis were all open reading frames beginning at the position of the 5′ consensus motif DIGDI within homology block D and ending at the position of the 3′ consensus motif PQYLR within homology block H. Five different sequences (eight clones in total) were excluded from the analysis because they were from non-alpha DBL domains.
Extraction of sequence features.
Translated sequences were aligned in batches using ClustalW analysis (http://www.ebi.ac.uk/clustalw/) using the default settings (Gonnet250 matrix, gap opening penalty = 10.0, gap extension penalty = 0.2, gap closing penalty = −1, gap separation penalty = 4). Sequence features listed in Figure 1C were extracted from the sequence using GeneDoc software (http://www.psc.edu/biomed/genedoc/) and exported into Microsoft (Seattle, Washington, United States) Excel and Stata version 6.0 (StataCorp, College Station, Texas, United States) for further analysis.
Sequence analysis.
To test the association between sequence features within DBLα sequences, the Cramer's V statistic was used in Stata. This is a representation of χ2, but is bounded between −1 and +1. To identify sequence features that were independently associated with DBLα sequence length, the following strategy was used. (1) The association between each PoLV motif and sequence length was determined using the Mann Whitney U test. (2) PoLV motifs with a highly significant negative association with sequence length (p < 0.0001) were identified. (3) These sequence features were grouped allowing one degenerate position. (4) Logistic regression was used to screen each DBLα sequence feature or group of features simultaneously for those that were independently associated with sequence length. See Text S4 for more details.
Sequence signatures.
Because of the high overall sequence diversity, very few of the sequences had absolute matches between isolates. To make more general comparisons between different sequences and to identify common and rare sequence types within each group, DBLα sequence features were used to reduce each sequence to a “signature” of standard length. The signature consisted of the string of amino acids at each of the PoLVs together with the cysteine count (see Figure 1C for examples). Sequence signatures are discussed further in Text S5 and Figure S3.
Definition of “distinct” sequences.
Several of the analyses described here, in particular the identification of sequence groups, were performed on collections of “distinct” sequences. A robust definition of “distinctness” was required to help minimise repeated sampling of very similar sequences arising from PCR and sequencing errors. For this definition, two sequences were considered distinct if they had either (1) non-identical signatures or (2) different amino acid length.
Homogeneity of cDNA expression.
Homogeneity of cDNA expression was defined for each isolate as the total number of cDNA clones containing the dominant two sequences from that isolate, expressed as a percentage of all cDNA clones sequenced from that isolate.
Northern blot analysis.
For comparison of full-length ring-stage var RNA transcripts, Northern blots were prepared and hybridized with a generic var exon 2 (see Figure 1A) probe as previously described [45]. A sample of laboratory isolate Palo Alto RNA was included, to allow size comparison between samples run on different gels. The largest commercially available markers go up to 9.5 kb, whereas the Palo Alto sample has major var transcripts at approximately 9 kb and 11 kb. Exposures to autoradiography film ranged from 1 to 4 d.
Generation of sequence identity matrices.
Distinct sequences from each required category were picked at random using the RAND function in Microsoft Excel and subjected to ClustalW analysis as described above. Identity matrices were generated in the form of statistics reports using GeneDoc software and the report file saved with an *.xls extension. Files were opened in Microsoft Excel and conditional formatting was used to shade the matrix as follows: 80% gray (55%–100% identity), 50% gray (45%–54% identity), 25% gray (35%–44% identity), and white (< 35% identity).
Supporting Information
Dataset S1 The 1,746 Translated DNA Sequences Used in the Study
(248 KB TXT)
Click here for additional data file.
Figure S1
var Expression Profiling
Pie charts are used to show the number of distinct sequences of each group cloned from each parasite isolate. The size of each slice is proportional to the number of clones of that sequence identified.
(19 KB PDF)
Click here for additional data file.
Figure S2 Global Analysis of Expressed DBLα Sequences between Subsets of Parasites
To obtain a global picture of the expression of different groups of parasites and to test the usefulness of our sequence groupings we randomly selected 26 cDNA clones from each parasite isolate and constructed identity matrices of pair-wise comparisons of the sequences. Multiple sequence comparisons of all 312 sequences were first performed using ClustalW. The twelve isolates were then split into various groups of six, and pairs of identity matrices were constructed from the selected sequences as described in Materials and Methods: mild and severe cases (A and B), low and high rosetting (C and D), and VSA antibody positive and negative (E and F). The expression patterns further illustrate the associations described in the text and reveal subtle characteristics of expression patterns that need to be explored in future studies with larger samples of parasites. The most notable is the apparent emergence of large clusters of similar sequences in both rosetting parasites and those from antibody-negative children. The fact that this is not apparent in children with severe malaria may reflect heterogeneity in the var genes compatible with causing severe malaria. However, this can be tested only by comparisons using matrices generated from much larger pools of sequences.
(22 KB PDF)
Click here for additional data file.
Figure S3 Significant Sequence Signatures from Kilifi and Elsewhere
Forty-seven Kilifi sequence signatures are shown of the total 393 isolated, in addition to two signatures from previously identified var genes associated with rosetting that were not found among Kilifi sequences. Kilifi sequences were considered significant based on four criteria: (1) being among the dominant cDNA sequences represented in the cDNA from each isolate (dark green boxes); (2) being sequence signatures that were isolated from more than five isolates (including 3D7); (3) being sequence signatures of full-length sequences that were identical in two or more isolates (highlighted with a white X); or (4) being sequence signatures shared with previously identified var genes of note. Only sequences cloned from cDNA and representing greater than 20% of all the cDNA clones from that isolate are shown as dark green squares. Expressed signatures that were not the most dominant sequence or were present in less than 20% of the sequences from each isolate are represented as light green boxes. For dominant and second most dominant sequence signatures from each isolate, the percentage of cDNA sequences containing that signature is indicated. Signatures only identified in genomic DNA from a given isolate are indicated as light gray boxes. The sequence signatures are divided into sequence groups 1–6 and sorted, with the sequence signatures that were most frequently shared between isolates at the top of each group. Within the sequence signatures listed on the left, individual sequence features that were not found in the 3D7 genome are highlighted with brackets. Sequence features that were the most frequently represented within all the clones are highlighted in bold. Those that were most frequently represented within cys2 sequences are written in blue. The PoLV1MFK* features are highlighted in dark red, PoLV2*REY features are highlighted in light red. Previously described var genes that contain the sequence features listed here are indicated on the right: AFBR41 in the 3D7 genome was found to be dominantly expressed in a vaccinated volunteer [37]. 3D7chr5var and FCR3varCSA are collectively known as var1. var1-like genes isolated so far tend to have either 3D7chr5var-like or FCR3varCSA-like sequence signatures [46–49]. The tendency of the dd2var1 gene to be conserved between isolates has been noted previously (S. Kyes, unpublished data). R29, FCR3S1.2-var1, and A4-AFBR19 are associated with parasite rosetting [12,13,50]. For more on sig1 and sig2, see Text S5.
(14 KB PDF)
Click here for additional data file.
Table S1 Parasite Isolates and Patients Used in This Study
(57 KB DOC)
Click here for additional data file.
Table S2 Mann Whitney U Test Analysis of Associations between Sequence Features and DBLα Sequence Length
(66 KB DOC)
Click here for additional data file.
Table S3 Logistic Regression Analysis of Associations between Sequence Features and DBLα Tag Sequence Length
(29 KB DOC)
Click here for additional data file.
Text S1 PCR and Sequencing Errors
(19 KB DOC)
Click here for additional data file.
Text S2 Comparison of PoLV between Kilifi Sequences and Isolate 3D7
(26 KB DOC)
Click here for additional data file.
Text S3 The Relationship between DBLα and ups
(20 KB DOC)
Click here for additional data file.
Text S4 Screening for Sequence Motifs Associated with DBLα Sequence Length
(21 KB DOC)
Click here for additional data file.
Text S5 Sub-Classification of DBLα Sequences by Their Sequence Signatures
(21 KB DOC)
Click here for additional data file.
Text S6
var1 Genes
(23 KB DOC)
Click here for additional data file.
Accession Numbers
The EMBL Nucleotide Sequence Database (http://www.ebi.ac.uk/embl) accession numbers discussed in this paper are cDNA (AM114937-AM115658) and gDNA (AM115696-AM116719).
We thank the parents and children who were involved in this study; Carol Churcher, Rebecca Atkin, Tracey Chillingworth, Nancy Hamlin, Zahra Hance, and Sally Whitehead for producing the sequence data; Norbert Peshu, the director of the Centre for Geographic Medicine Research, Coast (CGMRC), at Kilifi; Brett Lowe and the staff at CGMRC; Britta Urban, Alex Rowe, Paul Horrocks, Claire Mackintosh, Joe Smith, and Man-Suen Chan for critical comments on the manuscript; and Arnab Pain, Greg Fegan, and Rosalind Harding for useful discussion. This paper is published with the permission of the director of Kenya Medical Research Institute. The work was supported by a Wellcome Trust Advanced Training Fellowship in Tropical Medicine (060678) to PB. KM was supported by a Wellcome Trust Senior Fellowship (631342).
Competing interests. The authors have declared that no competing interests exist.
Author contributions. PCB, SK, KM, and CIN conceived and designed the study. PCB, SK, MB, MMK and MAQ performed the experiments. MB provided overall management of DNA sequencing. MAQ managed the DBL libraries and ensured clones from each library were made available for sequencing. NH managed sample processing and DNA sequencing. PCB analyzed the data and wrote the paper. MB, SK, KM, and CIN revised drafts of the paper.
Abbreviations
CIDRcysteine-rich interdomain region
DBLDuffy-binding-like
PfEMP1
Plasmodium falciparum erythrocyte membrane protein 1
PoLVposition of limited variation
VSAvariant surface antigen
==== Refs
References
Gardner MJ Hall N Fung E White O Berriman M 2002 Genome sequence of the human malaria parasite Plasmodium falciparum
Nature 419 498 511 12368864
Roberts DJ Craig AG Berendt AR Pinches R Nash G 1992 Rapid switching to multiple antigenic and adhesive phenotypes in malaria Nature 357 689 692 1614515
Smith JD Chitnis CE Craig AG Roberts DJ Hudson-Taylor DE 1995 Switches in expression of Plasmodium falciparum var genes correlate with changes in antigenic and cytoadherent phenotypes of infected erythrocytes Cell 82 101 110 7606775
Baruch DI Pasloske BL Singh HB Bi X Ma XC 1995 Cloning the Plasmodium falciparum gene encoding PfEMP1, a malarial variant antigen and adherence receptor on the surface of parasitized human erythrocytes Cell 82 77 87 7541722
Su X Heatwole VM Wertheimer SP Guinet F Herrfeldt JA 1995 The large diverse gene family var encodes proteins involved in cytoadherence and antigenic variation of Plasmodium falciparum -infected erythrocytes Cell 82 89 100 7606788
MacPherson GG Warrell MJ White NJ Looareesuwan W Warrell DA 1995 Human cerebral malaria: Quantitative ultrastructural analysis of parasitized erythrocyte sequestration Am J Pathol 119 385 401
Pongponratn E Turner GD Day NP Phu NH Simpson JA 2003 An ultrastructural study of the brain in fatal Plasmodium falciparum malaria Am J Trop Med Hyg 69 345 359 14640492
Baruch DI Gormley JA Ma C Howard RJ Pasloske BL 1996
Plasmodium falciparum erythrocyte membrane protein 1 is a parasitized erythrocyte receptor for adherence to CD36, thrombospondin, and intercellular adhesion molecule 1 Proc Natl Acad Sci U S A 93 3497 3502 8622965
Smith JD Kyes S Craig AG Fagan T Hudson-Taylor D 1998 Analysis of adhesive domains from the A4VAR Plasmodium falciparum erythrocyte membrane protein-1 identifies a CD36 binding domain Mol Biochem Parasitol 97 133 148 9879893
Smith JD Craig AG Kriek N Hudson-Taylor D Kyes S 2000 Identification of a Plasmodium falciparum intercellular adhesion molecule-1 binding domain: A parasite adhesion trait implicated in cerebral malaria Proc Natl Acad Sci U S A 97 1766 1771 10677532
Pain A Ferguson DJ Kai O Urban BC Lowe B 2001 Platelet-mediated clumping of Plasmodium falciparum -infected erythrocytes is a common adhesive phenotype and is associated with severe malaria Proc Natl Acad Sci U S A 98 1805 1810 11172032
Rowe JA Moulds JM Newbold CI Miller LH 1997
Plasmodium falciparum rosetting is mediated by a parasite-variant erythrocyte membrane protein and complement-receptor 1 Nature 388 292 295 9230440
Chen Q Barragan A Fernandez V Sundstrom A Schlichtherle M 1998 Identification of Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) as the rosetting ligand of the malaria parasite P. falciparum
J Exp Med 187 15 23 9419207
Urban BC Ferguson DJ Pain A Willcox N Plebanski M 1999
Plasmodium falciparum -infected erythrocytes modulate the maturation of dendritic cells Nature 400 73 77 10403251
Bull PC Marsh K 2002 The role of antibodies to Plasmodium falciparum infected erythrocyte surface antigens in naturally acquired immunity to malaria Trends Microbiol 10 55 58 11827798
Bull PC Lowe BS Kortok M Molyneux CS Newbold CI 1998 Parasite antigens on the infected red cell are targets for naturally acquired immunity to malaria Nature Medicine 4 358 360
Bull PC Kortok M Kai O Ndungu F Ross A 2000
Plasmodium falciparum -infected erythrocytes: Agglutination by diverse Kenyan plasma is associated with severe disease and young host age J Infect Dis 182 252 259 10882604
Nielsen MA Staalsoe T Kurtzhals JA Goka BQ Dodoo D 2002
Plasmodium falciparum variant surface antigen expression varies between isolates causing severe and nonsevere malaria and is modified by acquired immunity J Immunol 168 3444 3450 11907103
Baruch DI Gamain B Barnwell JW Sullivan JS Stowers A 2002 Immunization of Aotus monkeys with a functional domain of the Plasmodium falciparum variant antigen induces protection against a lethal parasite line Proc Natl Acad Sci U S A 99 3860 3865 11904437
Freitas-Junior LH Bottius E Pirrit LA Deitsch KW Scheidig C 2000 Frequent ectopic recombination of virulence factor genes in telomeric chromosome clusters of P. falciparum
Nature 407 1018 1022 11069183
Taylor HM Kyes SA Newbold CI 2000
Var gene diversity in Plasmodium falciparum is generated by frequent recombination events Mol Biochem Parasitol 110 391 397 11071291
Bull PC Lowe BS Kortok M Marsh K 1999 Antibody recognition of Plasmodium falciparum erythrocyte surface antigens in kenya: Evidence for rare and prevalent variants Infect Immun 67 733 739 9916084
Lindenthal C Kremsner PG Klinkert MQ 2003 Commonly recognised Plasmodium falciparum parasites cause cerebral malaria Parasitol Res 91 363 368 13680376
Nielsen MA Vestergaard LS Lusingu J Kurtzhals JA Giha HA 2004 Geographical and temporal conservation of antibody recognition of Plasmodium falciparum variant surface antigens Infect Immun 72 3531 3535 15155661
Voss TS Kaestli M Vogel D Bopp S Beck HP 2003 Identification of nuclear proteins that interact differentially with Plasmodium falciparum var gene promoters Mol Microbiol 48 1593 1607 12791141
Voss TS Thompson JK Waterkeyn J Felger I Weiss N 2000 Genomic distribution and functional characterisation of two distinct and conserved Plasmodium falciparum var gene 5′ flanking sequences Mol Biochem Parasitol 107 103 115 10717306
Robinson BA Welch TL Smith JD 2003 Widespread functional specialization of Plasmodium falciparum erythrocyte membrane protein 1 family members to bind CD36 analysed across a parasite genome Mol Microbiol 47 1265 1278 12603733
Kraemer SM Smith JD 2003 Evidence for the importance of genetic structuring to the structural and functional specialization of the Plasmodium falciparum var gene family Mol Microbiol 50 1527 1538 14651636
Taylor HM Kyes SA Harris D Kriek N Newbold CI 2000 A study of var gene transcription in vitro using universal var gene primers Mol Biochem Parasitol 105 13 23 10613695
Smith JD Subramanian G Gamain B Baruch DI Miller LH 2000 Classification of adhesive domains in the Plasmodium falciparum erythrocyte membrane protein 1 family Mol Biochem Parasitol 110 293 310 11071284
Kirchgatter K del Portillo HA 2002 Association of severe noncerebral Plasmodium falciparum malaria in Brazil with expressed PfEMP1 DBL1α sequences lacking cysteine residues Mol Med 8 16 23 11984002
Ward CP Clottey GT Dorris M Ji DD Arnot DE 1999 Analysis of Plasmodium falciparum PfEMP-1/var genes suggests that recombination rearranges constrained sequences Mol Biochem Parasitol 102 167 177 10477185
Lavstsen T Salanti A Jensen AT Arnot DE Theander TG 2003 Sub-grouping of Plasmodium falciparum 3D7 var genes based on sequence analysis of coding and non-coding regions Malar J 2 27 14565852
Recker M Nee S Bull PC Kinyanjui SM Newbold CI 2004 Transient cross-reactive immune responses can orchestrate antigenic variation in malaria Nature 429 555 558 15175751
Carlson J Helmby H Hill AVS Brewster D Greenwood BM 1990 Human cerebral malaria: Association with erythrocyte rosetting and lack of anti-rosetting antibodies Lancet 336 1457 1460 1979090
Rowe A Obeiro J Newbold CI Marsh K 1995
Plasmodium falciparum rosetting is associated with malaria severity in Kenya Infect Immun 63 2323 2326 7768616
Peters J Fowler E Gatton M Chen N Saul A 2002 High diversity and rapid changeover of expressed var genes during the acute phase of Plasmodium falciparum infections in human volunteers Proc Natl Acad Sci U S A 99 10689 10694 12142467
Kaestli M Cortes A Lagog M Ott M Beck HP 2004 Longitudinal assessment of Plasmodium falciparum var gene transcription in naturally infected asymptomatic children in Papua New Guinea J Infect Dis 189 1942 1951 15122533
Scherf A Hernandezrivas R Buffet P Bottius E Benatar C 1998 Antigenic variation in malaria: In situ switching, relaxed and mutually exclusive transcription of var genes during intra-erythrocytic development in Plasmodium falciparum
EMBO J 17 5418 5426 9736619
Chen Q Fernandez V Sundstrom A Schlichtherle M Datta S 1998 Developmental selection of var gene expression in Plasmodium falciparum
Nature 394 392 395 9690477
Fernandez V Chen Q Sundstrom A Scherf A Hagblom P 2002 Mosaic-like transcription of var genes in single Plasmodium falciparum parasites Mol Biochem Parasitol 121 195 203 12034453
Jensen AT Magistrado P Sharp S Joergensen L Lavstsen T 2004
Plasmodium falciparum associated with severe childhood malaria preferentially expresses PfEMP1 encoded by group A var genes J Exp Med 199 1179 1190 15123742
Bian Z Wang G 2000 Antigenic variation and cytoadherence of PfEMP1 of Plasmodium falciparum -infected erythrocyte from malaria patients Chin Med J (Engl) 113 981 984 11776131
Bull PC Pain A Ndungu FM Kinyanjui SM Roberts DJ 2005
Plasmodium falciparum antigenic variation: Relationships between in-vivo selection, the acquired antibody response and disease severity J Infect Dis 192 1119 1126 16107968
Kyes S Pinches R Newbold C 2000 A simple RNA analysis method shows var and rif multigene family expression patterns in Plasmodium falciparum
Mol Biochem Parasitol 105 311 315 10693754
Kyes SA Christodoulou Z Raza A Horrocks P Pinches R 2003 A well-conserved Plasmodium falciparum var gene shows an unusual stage-specific transcript pattern Mol Microbiol 48 1339 1348 12787360
Salanti A Jensen AT Zornig HD Staalsoe T Joergensen L 2002 A sub-family of common and highly conserved Plasmodium falciparum var genes Mol Biochem Parasitol 122 111 115 12076777
Rowe JA Kyes SA Rogerson SJ Babiker HA Raza A 2002 Identification of a conserved Plasmodium falciparum var gene implicated in malaria in pregnancy J Infect Dis 185 1207 1211 11930336
Winter G Chen Q Flick K Kremsner P Fernandez V 2003 The 3D7var5.2 (var COMMON) type var gene family is commonly expressed in non-placental Plasmodium falciparum malaria Mol Biochem Parasitol 127 179 191 12672527
Horrocks P Pinches R Christodoulou Z Kyes SA Newbold CI 2004 Variable var transition rates underlie antigenic variation in malaria Proc Natl Acad Sci U S A 101 11129 11134 15256597
|
16304608
|
PMC1287908
|
CC BY
|
2021-01-05 12:11:25
|
no
|
PLoS Pathog. 2005 Nov 18; 1(3):e26
|
utf-8
|
PLoS Pathog
| 2,005 |
10.1371/journal.ppat.0010026
|
oa_comm
|
==== Front
PLoS PathogPLoS PathogppatplpaplospathPLoS Pathogens1553-73661553-7374Public Library of Science San Francisco, USA 1630460910.1371/journal.ppat.001002805-PLPA-RA-0128R1plpa-01-03-05Research ArticleBiochemistryInfectious DiseasesMicrobiologyMolecular Biology - Structural BiologyIn VitroMechanisms of Assembly and Cellular Interactions for the Bacterial Genotoxin CDT CDT Assembly and Cellular InteractionsNesic Dragana Stebbins C. Erec *Laboratory of Structural Microbiology, The Rockefeller University, New York, New York, United States of AmericaGalan Jorge EditorYale University, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 18 11 2005 1 3 e2812 8 2005 12 10 2005 Copyright: © 2005 Nesic and Stebbins.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.Many bacterial pathogens that cause different illnesses employ the cytolethal distending toxin (CDT) to induce host cell DNA damage, leading to cell cycle arrest or apoptosis. CDT is a tripartite holotoxin that consists of a DNase I family nuclease (CdtB) bound to two ricin-like lectin domains (CdtA and CdtC). Through the use of structure-based mutagenesis, biochemical and cellular toxicity assays, we have examined several key structural elements of the CdtA and CdtC subunits for their importance to toxin assembly, cell surface binding, and activity. CdtA and CdtC possess N- and C-terminal nonglobular polypeptides that extensively interact with each other and CdtB, and we have determined the contribution of each to toxin stability and activity. We have also functionally characterized two key binding elements of the holotoxin revealed from its crystal structure. One is an aromatic cluster in CdtA, and the other is a long and deep groove that is formed at the interface of CdtA and CdtC. We demonstrate that mutations of the aromatic patch or groove residues impair toxin binding to HeLa cells and that cell surface binding is tightly correlated with intoxication of cultured cells. These results establish several structure-based hypotheses for the assembly and function of this toxin family.
Synopsis
The cytolethal distending toxin is used by many bacteria to damage the DNA of infected organisms. This DNA damage prevents cells from dividing and eventually leads to cell death, which raises the possibility that this genomic damage may be a contributing factor to carcinogenesis. The cytolethal distending toxin is composed of three proteins that form a tightly associated complex. After secretion by the bacterium, two proteins in this complex adhere to the cell surface and achieve the delivery of the third protein into the cell, where it causes DNA lesions. This report examines how this toxin is assembled and how it adheres to host cell surfaces. A set of molecular features on the toxin is shown to be critical for this cell adherence and for the ability of the cytolethal distending toxin to inhibit cell division. These results tie together for the first time aspects of the molecular structure of the cytolethal distending toxin and its ability to adhere to host cell surfaces, contributing to mechanistic understanding of the activity of this genotoxin.
Citation:Nesic D, Stebbins CE (2005) Mechanisms of assembly and cellular interactions for the bacterial genotoxin CDT. PLoS Pathog 1(3): e28.
==== Body
Introduction
The cytolethal distending toxin (CDT) is a tripartite bacterial toxin that targets many types of eukaryotic cells. Most of the cells intoxicated with CDT predominantly arrest in G2/M transition of the cell cycle; they slowly distend over a period of 2−5 d, and eventually die [1−4]. The effect on lymphocyte cultures is somewhat faster and different, since they do not distend but undergo apoptosis [5−7]. CDT activity has been detected among many pathogenic gram-negative bacteria that can cause a panoply of diseases such as chancroid, endocarditis, diarrhea, periodontal disease, and chronic hepatitis [1−4].
A major advancement in the understanding of the action of CDT was taken with the discovery that CdtB (cytolethal distending toxin B) shares weak active site sequence similarity with DNase I−like nucleases and that the mutation of predicted active site residues leads to a loss of CDT activity [8,9]. Subsequent studies demonstrated that CdtB has weak nuclease activity in vitro [10,11] and that CDT inflicts damage to eukaryotic DNA in vivo [4,5,9,10,12−15]. The crystal structure of the CDT holotoxin from Haemophilus ducreyi revealed that CdtB is indeed a DNase I−like nuclease in which all key residues involved in the catalysis and binding to DNA are conserved [16]. A growing amount of evidence supports a model of CDT intoxication by which CdtB delivery into the cells is mediated by CdtA (cytolethal distending toxin A) and CdtC (cytolethal distending toxin C) [17−19]. Once inside the cell, CdtB can translocate to the nucleus and induce DNA lesions, which will activate DNA damage response cascades and cause cell cycle arrest [18−20].
Very limited data are available about the function of CdtA and especially of CdtC. Only recently have several groups adapted different binding assays to study the interactions of CdtA and CdtC with eukaryotic cells. Most of these assays find that both CdtA and CdtC are capable of binding to the cellular membrane [21−24], and some suggested that they might share the same receptor [22,23]. Reports indicating that antibodies raised against CdtC were protective against CDT toxicity corroborated these findings [25]. On the contrary, Mao and DiRienzo [11] were not able to detect CdtC on the cell surfaces by using immunofluorescent techniques. The crystal structure of CDT holotoxin uncovered that both CdtA and CdtC are lectin-type structures, similar to each other and to the binding component of plant toxin ricin [16]. This realization, together with the positioning of lectin subunits, and the presence of two notable surface elements at their interface, an aromatic patch and an adjacent deep groove surface, suggest that the role of CdtA and CdtC is to interact with the cell surface and to enable translocation of the holotoxin across the plasma membrane [16].
In the present report, we have used structure-guided mutagenesis, biochemical and cellular assays, to examine the main structural features of CdtA and CdtC. We have discovered nonglobular interactions critical for the holotoxin assembly, stability, and cytotoxicity. Mutational analysis of the surface exposed residues of the aromatic patch and the groove shows that these two elements are critical to cell surface binding and toxicity.
Results
The Role of CdtA Nonglobular Interactions
We recently determined the crystal structure of Haemophilus ducreyi CDT [16]. The structure revealed that the toxin is composed of three different subunits (Figure 1A): CdtA and CdtC, which both exhibit a lectin-type fold, and CdtB, which is a DNase I-like nuclease. All three subunits interact intimately with each other, forming three extensive globular protein-protein interfaces. In addition, CdtA and CdtC each have extended, nonglobular polypeptides at their N- and C-termini, and in both proteins these regions interact with other elements of the holotoxin to cement the assembly of the ternary complex. These four nonglobular extensions, or “tails,” account for nearly a third of the surface area buried upon complex formation. These tails, therefore, likely contribute significantly to the stability of the ternary complex. By creating a series of deletion mutants, we have examined the role of the CdtA and CdtC “tails” in holotoxin assembly and activity.
Figure 1 Structure of CDT and Mutational Locations
Four alternative orientations of the crystal structure of the H. ducreyi CDT are shown as a ribbon cartoon tracing the three polypeptide chains. CdtA, CdtB, and CdtC are shown in blue, red, and green, respectively. Red dots indicate CdtB peptide not modeled due to disorder. N, NH2-terminus; C, COOH-terminus. The final image shows a partially transparent surface illustration focusing on the groove (red) and aromatic patch (yellow). Residues mutated for cell binding studies are indicated in white.
The N-terminus of CdtA is particularly interesting. Residues 18−56 are disordered and invisible in the crystal structure, and we sought to determine their contribution to the assembly and activity of the holotoxin. The amino acids 56−67 of CdtA make only minor contacts with other subunits within a single holotoxin complex, but instead interact with the groove of an adjacent holotoxin complex in the crystals [16]. The amino acids 67−75 of CdtA, however, do contact CdtB and CdtC in the appropriate complex in the crystal and therefore represent a more realistic structural interaction. Therefore, we created three N-terminal deletions of CdtA (Δ18−56, Δ18−67, Δ18−75) and examined them for complex assembly and toxin activity.
All mutants with N-terminal deletions of CdtA were successfully purified and refolded into CDT complex. The complex integrity in all cases was preserved during ion-exchange chromatography (Figure 2). The refolded complexes were loaded on a cation-exchange column (SP Sepharose Fast Flow; GE Healthcare, Piscataway, New Jersey, United States) at 40 mM NaCl concentration and eluted by gradual increase of salt concentration (from 0 mM to 500 mM NaCl). The CDT holotoxin, wild-type and all mutants, elutes between 100 mM and 150 mM NaCl. In addition, in some preparations even in the case of the wild-type complex, we observed a second peak eluting at concentrations higher than 200 mM NaCl. This peak was composed of the CdtB subunit only, the only CDT subunit that binds to SP Sepharose under these conditions (data not shown). That is not unexpected as the calculated pI of CdtB is fairly basic at 8.5, whereas CdtA and CdtC have acidic pIs of 5.58 and 6.21, respectively. It is possible that single peak of CdtB is due to excess CdtB protein in the refolding reactions that refolded apart from CdtA and CdtC, which then binds more strongly to the cation-exchange resin than the holotoxin with the acidic lectin domains. It is also possible that the cation-exchange column destabilizes the CDT holotoxin to some degree, leading to an increase in the “CdtB peak.” CdtA and CdtC were present in the flow through of the column as well, and this is what we observed with CDT holotoxin with Δ18−75 truncation of CdtA (Figure 2D). Moreover, when refolded in the absence of CdtB, CdtA and CdtC do not bind the resin under these conditions (data not shown).
Figure 2 Cation-Exchange Chromatography of the CDT Holotoxin Containing N-Terminal Deletions of CdtA
Wild-type (A) or mutant CDT holotoxins [(B) holotoxin with CdtA (Δ18–56), (C) holotoxin with CdtA (Δ18–67), (D) holotoxin with CdtA (Δ18–75)] were run on Fast Flow SP Sepharose columns using an ÄKTA FPLC. Refolded CDT complexes were loaded and examined in a salt gradient as described in Materials and Methods. Individual fractions (5 ml) or pooled material from the elution peak were collected and examined by SDS-PAGE. Proteins were stained with Coomassie blue stain. Images of the gels are presented inside corresponding chromatograms. L, loaded material; fr, fraction number; Fl, flowthrough; P, pooled material from elution peak. Absorbance was measured at 280 nm.
On a Superdex 200 gel filtration column (GE Healthcare), using a gel filtration buffer containing 200 mM NaCl, 20 mM HEPES (pH 7.5), 2.5 mM dithiothreitol (DTT) (Figure 3A), all CdtA deletion mutants that were refolded into a CDT holotoxin were shifted significantly to a lower molecular weight in gel filtration (Figure 3A). At the same time, all three subunits appear in a stoichiometric ratio in the peak fractions (Figure 3B). Moreover, all of these deletions of CdtA do not reduce the toxicity of the holotoxin, as measured by cell cycle arrest of holotoxin-treated cells (Figure 3C). One explanation for the discrepancy between the results for biochemical assembly and activity of the CdtA N-terminal deletion mutants would be to postulate an enhanced stabilization of the ternary complex upon interaction with a cell surface ligand. Indeed, gel filtration chromatography is a more demanding assay for complex stability than other chromatographic methods, and the three CdtA deletion mutants discussed above all co-purified by ion-exchange chromatography (Figure 2).
Figure 3 Effects of N-Terminal Deletions of CdtA on Holotoxin Integrity and Toxicity
(A) Size exclusion chromatography profile of CDT holotoxin containing wild-type or deletion mutants of CdtA, wild-type CdtB, and myc-tagged CdtC. Following cation-exchange chromatography, 2-ml samples were injected onto a Superdex 200 HiLoad 16/60 column and run at flow rate of 2 ml/min in 200 mM NaCl, 20 mM HEPES (pH 7.5), and 2.5 mM DTT on an ÄKTA FPLC. Fractions of 2 ml were collected. Calculated molecular weights of globular proteins eluting at the same elution volumes are indicated in parentheses.
(B) Peak fractions were analyzed by SDS-PAGE, and CDT subunits were visualized with Coomassie blue stain. A, CdtA; B, CdtB; C, CdtC; A*, deletion mutants of CdtA.
(C) Cell cycle analysis of HeLa cells exposed to 1 ng/ml concentration of the wild-type or mutant holotoxin for 3 hr at 37 °C, 5% CO2. Cells were processed 48 hr after holotoxin treatment, and DNA content was measured by flow cytometry. The calculated percentages of cells in G0/G1, S, and G2/M are shown.
As an alternative to demonstrate complex formation, we have performed co-immunoprecipitation experiments. A Myc-tagged CdtC subunit was refolded together with wild-type CdtB and wild-type or deletion mutants of CdtA and subjected to ion-exchange and size exclusion chromatography. Gel filtration fractions were then subjected to immunoprecipitation with anti-MYC antibody. They were all able to co-immunoprecipitate, demonstrating that the subunits were still able to form complexes (Figure 4A) despite their disassembly during gel filtration chromatography. Nonspecific binding to protein G Sepharose was not detected (Figure 4B). All of these experiments suggest that contacts that the residues 18−75 of N-terminus CdtA make with other subunits contribute to the holotoxin stability, but they are not critical for complex formation and toxin activity.
Figure 4 Immunoprecipitation of CDT Complexes Containing Deletion Mutants of CdtA
(A) The complexes (10 μg each) were immunoprecipitated with 10 μg anti-myc antibody (myc tag was fused to C-terminus of CdtC) in 150 mM NaCl, 20 mM HEPES (pH 7.5), 0.1% Nonidet P40 substitute, followed by exposure to protein G Sepharose beads, and subjected to nondenaturing SDS-PAGE.
(B) Nonspecific binding of CDT complexes to protein G Sepharose was not detected under the same experimental conditions. A, CdtA; B, CdtB; C, CdtC; A*, truncated CdtA; A1, holotoxin with CdtA Δ18−56; A2, holotoxin with CdtA Δ18−67; A3, holotoxin with CdtA Δ18−75; WT, wild-type; Ab, only anti-myc antibody, no CDT; IgG (H+L), nonreduced immunoglobulin G; IgG H, immunoglobulin G heavy chain; IgG L, immunoglobulin G light chain; SN, supernatant.
The C-terminus of CdtA extends to CdtC and binds by forming an intermolecular β-sheet, whereas the C-terminal extension of CdtC interacts with both CdtA and CdtB (Figure 1A). The CdtA C-terminus possesses a more extensive interaction than the N-terminal region of this subunit, and it might be expected that deletions of this region would have a more pronounced effect. This is indeed the case, as removal of this polypeptide (CdtAΔ215−223) leads to an inability of the complex to co-refold and to a precipitation of CdtA (data not shown).
The Role of CdtC Nonglobular Interactions
We also made a series of deletion mutants of the CdtC N- and C- terminal tails, which demonstrate that the nonglobular interactions of CdtC with both CdtA and CdtB strongly contribute to toxin assembly, stability, and activity.
The N-terminal 13 amino acids of CdtC interact with active site residues of CdtB, and we have previously shown that these interactions play an autoinhibitory role in holotoxin activity in vitro [16]. We have previously shown that CdtCΔ(21−35) assembles into a biochemically stable holotoxin, and it is as active as the wild-type complex in toxicity assays [16]. A slightly larger deletion, CdtCΔ(21−39), which removes several residues that make a small number of additional contacts to CdtB and CdtA, destabilizes the complex. The mutant complex still co-purifies by cation-exchange chromatography, but there is a significant decrease in the holotoxin stability (Figure 6C). This is also evident by gel filtration chromatography (Figure 5A and 5B). This leads to impaired toxicity of the mutant holotoxin as determined by analysis of cell cycle of holotoxin-treated cells (Figure 5C).
Figure 5 Effects of N- and C-Terminal Deletions of CdtC on Holotoxin Integrity and Toxicity
(A) Size exclusion chromatography of recombinant holotoxin containing different CdtC mutants. After cation-exchange chromatography, 2-ml CDT holotoxin samples were injected onto a Superdex 200 HR 10/30 column and run at flow rate of 0.5 ml/min in 200 mM NaCl, 20 mM HEPES (pH 7.5), 2.5 mM DTT on an ÄKTA FPLC. Fractions of 1 ml were collected.
(B) Following elution from the gel filtration column, several peak fractions were subjected to SDS-PAGE and CDT subunits visualized with Coomassie blue stain. A, CdtA; B, CdtB; C, cdtC; C*, deletion mutants of CdtC.
(C) Cell cycle analysis of HeLa cells exposed to 1 ng/ml concentration of either wild-type or mutant holotoxin for 3 hr at 37 °C, 5% CO2. Cells were processed 48 hr after holotoxin treatment, and DNA content was measured by flow cytometry. The calculated percentages of cells in G0/G1, S, and G2/M are shown.
Figure 6 Cation-Exchange Chromatography of CDT Holotoxin Containing N- and C-Terminal Deletions of CdtC
Wild-type (A) or mutant CDT holotoxins [(B) holotoxin with CdtC (Δ21−35), (C) holotoxin with CdtC (Δ21−39), (D) holotoxin with CdtC (Δ179−186)] were run on Fast Flow SP Sepharose columns (20 ml) using an ÄKTA FPLC. Refolded CDT complexes were loaded on an SP Sepharose column in buffer containing 40 mM NaCl, 20 mM HEPES (pH 7.5), 2.5 mM DTT and analyzed by a gradient in salt concentration as described in Materials and Methods. Individual fractions (5 ml) or pooled material (P) from elution peak were collected and examined by SDS-PAGE. Proteins were stained with Coomassie blue dye. Images of the gels are presented inside corresponding chromatograms. The peak that elutes between 100 and 150 mM NaCl concentration (P1) contains an intact CDT holotoxin, and the second peak (P2) that elutes at concentrations higher than 200 mM NaCl contains only the CdtB subunit. A, CdtA; B, CdtB; C, CdtC; C*, deletion mutants of CdtC; L, loaded material; fr, fraction number; Fl, flowthrough; P, pooled material from elution peak. Absorbance was measured at 280 nm. In A, P1 contains fractions 14−22, and P2 contains fractions 31−34; in D, P1 contains fractions 20−22, and P2 contains fractions 32−34.
As the C-terminal tail of CdtC interacts with both CdtA and CdtB, it is expected that its removal would undermine complex stability and decrease toxicity. CdtCΔ179−186, which deletes residues at the C-terminus that are not visible in the crystals, partially destabilizes the holotoxin assembly, as judged by gel filtration and ion-exchange chromatography (Figures 5A and 6D), and significantly reduces toxicity against cells (Figure 5C). It is not possible with these data to determine whether the decrease in toxicity is due to partial destabilization of the ternary complex or to a role of the CdtC C-terminus in receptor binding. A larger C-terminal deletion, CdtCΔ169−186, presents a much more forceful phenotype, as it does not productively refold, indicating that the loss of residues 169−178, which make contact with both CdtA and CdtB, more significantly destabilizes the assembly of the ternary complex (data not shown).
Structural Basis of Cell Surface Binding
CdtA and CdtC have many similarities to the ricin B-chain in fold and in placement with respect to the active subunit within the holotoxin. Therefore, in analogy to ricin, it may be expected that these two subunits of CDT will have a function in host cell binding and internalization of the toxin. Several reports indicate that these subunits can adhere to cell surfaces [11,21−23]. The crystal structure shows that CdtA and CdtC together contribute toward the formation of two notable surfaces, a cluster of highly conserved aromatic residues, and a long and deep groove that is formed at the interface of two subunits (Figure 1B). These may represent regions responsible for cell contact and attachment. To address this possibility, we have altered these molecular surfaces by mutagenesis and examined mutant holotoxins for binding to cells and effectiveness to induce cell cycle arrest.
In order to be able to determine if different mutant holotoxins interact with the cells, we have developed a fluorescence-based binding assay. The holotoxin was fluorescently labeled with Alexa Fluor 488 fluorescent dye (Molecular Probes, Eugene, Oregon, United States). The Alexa Fluor 488 reactive dye has a tetrafluorophenyl ester moiety that reacts efficiently with primary amines of proteins to form stable dye-protein conjugates. It was necessary to test the activity of CDT-Alexa Fluor conjugate in a cellular intoxication assay, since there are reports suggesting that biotin labeling of CdtA disrupted complex formation and led to a reduction in toxicity [22]. Dye-conjugated CDT was as potent as unconjugated toxin and caused a comparable level of G2/M arrest of HeLa cells (Figure 7A). This enabled us to test the binding of fluorescently labeled holotoxin to HeLa cells by flow cytometry and confocal microscopy. Flow cytometry gave a quantitative evaluation of holotoxin binding, whereas confocal microscopy provided qualitative visualization of the cytometry results, while also revealing the localization of bound holotoxin. The labeled holotoxin specifically binds to cells in a manner easily detectable over background.
Figure 7 The Activity of CDT Holotoxin Labeled with Alexa Fluor 488
(A) The toxicity of CDT holotoxin labeled with Alexa Fluor 488. HeLa cells were treated with 1 ng/ml (black) or 10 ng/ml (gray) concentration of unconjugated or Alexa Fluor 488−conjugated CDT for 3 hr at 37°C, 5% CO2. Cells were processed 48 hr after holotoxin treatment, and DNA content was measured by flow cytometry. The calculated percentages of cells in G0/G1, S, and G2/M are shown.
(B) Binding of CDT-Alexa Fluor 488 to cells. Harvested HeLa cells were exposed for 2 hr to 5 and 10 μg/ml concentration of wild-type or mutant CDT-Alexa Fluor 488. The histogram shows the binding of 5 or 10 μg/ml concentration of wild-type and mutant CDT-Alexa Fluor 488 conjugates to HeLa cells. Mock represents cells in buffer only (2% FCS in PBS), and control is goat anti-mouse IgG conjugated with Alexa Fluor 488, which does not bind to HeLa cells. The level of fluorescence was analyzed by flow cytometry. The relative levels of fluorescent labeling of wild-type and mutant CDT holotoxin was maintained to be nearly equivalent, with the mutant holotoxins (groove and aromatic patch) possessing a slightly higher level of labeling than the wild-type (Materials and Methods).
We have previously demonstrated that a quadruple aromatic patch mutant (W91G/W98G/W100G/Y102A) is incapable of intoxicating cells [16] (Figure 8B) and postulated that this prominent structural element could be involved in cell surface interactions. Using a fluorescence-based cell binding assay, we were able to show that this mutant is indeed incapable of binding to HeLa cells. By both flow cytometry and confocal microscopy, we were not able to detect any significant binding of the aromatic patch mutant above background (Figures 7B and 9). This shows that the aromatic patch residues are vital for cell surface attachment. A triple aromatic mutant (W98G/W100G/Y102A) behaves in the same way as the quadruple (Figure 8B).
Figure 8 Effect of the Aromatic Patch and Groove Mutants on the Holotoxin Assembly and Toxicity
(A) Elution profile of the wild-type and groove mutant of the CDT holotoxin during size exclusion chromatography. Samples were loaded on a Superdex 200 HiLoad 16/60 column at a flow rate of 2 ml/min in 200 mM NaCl, 20 mM HEPES (pH 7.5), 2.5 mM DTT on ÄKTA FPLC. Peak fractions (2 ml) were analyzed by SDS-PAGE (15% gel), and CDT subunits were visualized with Coomassie blue staining. SDS-PAGE of the CDT holotoxin groove mutant is shown inside gel filtration chromatogram. Elution volumes of wild-type and mutant holotoxin are shown.
(B) Cell cycle analysis of HeLa cells exposed to 1 ng/ml (gray) or 10 ng/ml (black) concentration of either wild-type or mutant holotoxin for 3 hr at 37 °C, 5% CO2. Cells were processed 48 hr after holotoxin treatment, and DNA content was measured by flow cytometry, as detailed in Materials and Methods. The calculate percentages of cells in G0/G1, S, and G2/M are shown. CdtABC 4m Aromatic mutant (CdtA: W91G/W98G/W100G/Y102A); CdtABC 3m Aromatic mutant (CdtA: W98G/W100G/Y102A); CdtABC groove mutant (CdtA: P103A/Y106A, CdtC: R43K/Q49A).
Figure 9 Binding of Mutant CDT- Alexa Fluor 488 Holotoxin to Cells
HeLa cells were treated for 2 hr with 5 μg/ml concentration of wild-type or mutant fluorescently labeled holotoxin in 3% BSA/PBS at 37 °C, 5% CO2. Samples were co-stained with Hoechst 33258 dye to visualized nuclei. Toxin was washed out, and cells were fixed in 3.7% formaldehyde and mounted with ProLong Gold Antifade Reagent (Molecular Probes). Confocal images were taken under an inverted LSM510 confocal microscope (Zeiss) using ×60 objective lenses. The aromatic patch mutant contains wild-type CdtB and CdtC, and mutant CdtA (W91G/W98G/W100G/Y102A); the groove mutant contains wild-type CdtB and mutant CdtA (P103A/Y106A), and CdtC (R43K/Q49A).
We also mutagenized the extended groove formed by CdtA and CdtC to assess its importance. The groove mutant contains two mutations in CdtA (P103A and Y106A) and two in CdtC (R43K and Q49A). The rationale for these mutations was to include residues that are surface exposed and that also interact with the N-terminal tail of the symmetry-related CdtA molecule. The quadruple groove mutant did not affect complex stability as judged by ion-exchange (data not shown) and gel filtration chromatography (Figure 8A). However, the biological activity of the mutant holotoxin was significantly impaired (Figure 8B). We have examined the ability of the groove mutant to induce cell cycle arrest on HeLa cells using two different concentrations and found that the percentage of cells that arrest in G2/M is diminished at a 10 ng/ml concentration of toxin (compare wild-type [68.98%] and groove mutant [48.52%]), and it was almost undetectable at 1 ng/ml (the concentration that is usually sufficient to cause significant cell cycle arrest with the wild-type holotoxin). To characterize this mutant in more detail, we examined it in binding assays. Alexa Fluor 488−conjugated groove mutant exhibited noticeably diminished binding to HeLa cells as determined by both flow cytometry (Figure 7B) and confocal microscopy (Figure 9). It should be noted that the degree of labeling of the mutant CDT holotoxins was slightly higher than that of the wild-type CDT (see Materials and Methods), which excludes the possibility that lack of detection of binding was due to weaker fluorescent labeling. Therefore, these findings identify the groove as a second critical cell-contacting region of the holotoxin.
Discussion
Lara-Tejero and Galan [18] initially proposed that CDT represents an “AB” type of toxin in which CdtB serves as the “A,” or active, subunit, and CdtA and CdtC represent a heterodimeric “B,” or binding, subunit. The crystal structure of the H. ducreyi CDT holotoxin strongly substantiated that hypothesis [16]. By analogy to other AB toxins [26], it has been proposed that CDT travels from the plasma membrane through an endocytic pathway and then enters the Golgi network and endoplasmic reticulum, after which CdtB translocates directly into the nucleus [27], or is delivered by retrograde transport into the cytosol, and then translocates into the nucleus [17−20]. In order to try to elucidate the role of CdtA and CdtC in the toxin assembly and cellular binding, we have addressed the contribution of several critical structural elements of these molecules by biochemical and cellular assays.
Using deletion mutagenesis, we examined the role of several nonglobular interactions in toxin stability and function. We found that contacts of C-terminal tails of CdtA and CdtC with other subunits are vital for toxin assembly. Their removal results in the poor refolding of the holotoxin, destabilization, and, as a consequence, inefficient cellular intoxication. The N-terminal tail of CdtC is deeply inserted into the active site of CdtB, which may contribute to overall stability of the complex. Interestingly, the truncation of 35 amino acids from the N-terminus of CdtC does not disturb complex stability, but removal of just four more residues leads to destabilization of the holotoxin and impaired toxicity. On the other hand, the N-terminal tail of CdtA makes only a few contacts with other subunits, which appear to be inconsequential for the CdtA function in the holotoxin. Although removal of up to 75 residues from N-terminal tail of CdtA destabilizes the complex under gel filtration conditions, truncated mutants are still as potent as the wild-type holotoxin against target cells. This finding is in agreement with previous observations that deletion 19−49 of closely related Actinobacillus actinomycetemcomitans CdtA [28] or, in another report, the first 59 residues [29] are dispensable for holotoxin activity. Moreover, several groups have demonstrated that some preparations of holotoxin actually contain truncated CdtA [25,30]. In other bacterial pathogens, the region N-terminal to ricin fold of CdtA is very diverse and much longer than in H. ducreyi. It is possible that these long N-terminal tails play a role in the assembly and translocation of the holotoxin in bacterial cells.
Recently, several studies examined the binding of CDT subunits to target cells and addressed the role of both CdtA and CdtC. The results were somewhat conflicting in the conclusion of what contribution CdtC made to this process. While several groups have shown that both CdtA and CdtC can bind to the surface of HeLa cells and that preincubation with CdtA-CdtC complex can inhibit subsequent intoxication with holotoxin [21−24], others have observed only binding of CdtA to Chinese hamster ovary cells [11]. Furthermore, it was suggested that CdtA and CdtC might bind to the same cellular receptor [22,23]. The crystal structure of H. ducreyi CDT holotoxin demonstrated that both CdtA and CdtC are lectin-type structures, and they both contribute to the formation of two prominent molecular surfaces that could be involved in cellular interactions. The first is a region composed of highly conserved, surface-exposed aromatic residues. We have previously shown that mutations of four CdtA aromatic residues (W91G, W98G, W100G, Y102A) abolish the holotoxin activity [16]. Here, using fluorescence-based binding experiments, we were able to demonstrate that these residues are indeed involved in the cell surface binding. While binding of fluorescently labeled wild-type holotoxin to HeLa cells was readily detectable by flow cytometry and confocal microscopy, binding of an Alexa Fluor 488−conjugated aromatic mutant was undetectable in both assays. It appears that the disruption of only three aromatic residues of CdtA (W98, W100, Y102) is sufficient to abolish cellular interactions and abrogate holotoxin activity. It was recently shown that Escherichia coli CdtA-II and CdtC-II can interact with glycoproteins, and it was postulated that the putative cellular receptor would most probably have an N-linked fucose-containing structure [23]. It has been suggested that the aromatic patch could be a potential carbohydrate-interacting region [23,31], although the lipophilic nature of this cluster of aromatic residues suggests that it may also serve to interact with components of the plasma membrane. Indeed, it has been recently shown that interaction of CDT with the cells is sensitive to cholesterol depletion, and it was therefore suggested that cholesterol-rich lipid rafts could be involved in CDT binding [27].
We have also identified an additional cell-binding region on the surface of the CDT holotoxin. It is represented by a groove that is formed by CdtA and CdtC along their molecular surfaces. Mutations of four surface-exposed groove residues (two in CdtA:P103A/Y106A and two in CdtC:R43K/Q49A) significantly decrease binding to HeLa cells, which results in the decrease in CDT toxicity. The effect of these mutations was less severe than that of mutations of aromatic residues, representing an intermediate phenotype. By increasing concentrations of the groove mutant, it was possible to induce more robust cell cycle arrest. In contrast, aromatic mutants were completely incapable of causing any effect on treated cells. This could suggest that these two regions are involved in two independent binding events with the cell surface.
Our findings are an important step toward a better understanding of the molecular mechanisms underlying CDT holotoxin assembly and interaction with the target cell surfaces. Understanding toxin assembly and activity also provides potential targets for pharmacological disruption of CDT activity. Finally, the identification of host cell surface binding mutants of CDT provides important tools for the identification of host cell receptors mediating toxin entry.
Materials and Methods
Production of wild-type and mutant H. ducreyi holotoxin.
H. ducreyi CDT subunits were cloned, purified, and holotoxin complex reconstituted as previously described [16]. CdtA(18−223), CdtB(23−283), and CdtC(21−186) were cloned by PCR from genomic DNA (ATCC [American Type Culture Collection] number 700724D, Manassas, Virginia, United States) as an N-terminal hexahistidine fusion proteins in E. coli such that the predicted N-terminal secretion signals were removed [32]. cdtA was cloned in pET-28a vector (Novagen, Madison, Wisconsin, United States) with engineered in-frame rhinovirus 3C protease recognition sequence at the 5′ end of cdtA. cdtB was cloned into an engineered version of the pac28 vector [33], which also contains a 3C protease recognition sequence between the hexahistidine tag and the cdtB. cdtC was cloned in the pET-21a vector (Novagen), which has an insertion with a hexahistidine tag and an engineered 3C protease recognition site. The three subunits were expressed separately in BL-21 E. coli cells by the addition of 0.1 mM isopropyl-β-D-thiogalactopyranoside (IPTG; Gold Biotechnology, Inc., St. Louis, Missouri, United States) for 4 hr at 37 °C, 200 rpm. The subunits were purified under denaturing conditions (8 M urea, 10 mM Tris [pH 8.0], 0.1 M Na-phosphate) using nickel chelating affinity resin (GE Healthcare). Each subunit was eluted from nickel chelating affinity resin with the elution buffer that contained 500 mM imidazole (pH 8.0), 8 M urea, 10 mM Tris (pH 8.0), 0.1 M Na-phosphate. The usual yields per liter of bacterial culture were 50 mg of CdtA, 60 mg of CdtB, and 200 mg of CdtC. The CDT holotoxin was reconstituted by co-refolding all three subunits together via dialysis at 4°C into a native buffer consisting of 20 mM HEPES (pH 7.5), 200 mM NaCl, 2.5 mM DTT, 5% glycerol, and 2 mM EDTA. The total protein concentration during refolding was maintained under 0.1 mg/ml (20 mg of protein in 200 ml of denaturing buffer per 4 L of native buffer), and the native buffer was changed four times within a 24-hr period. The proteins were separated from the affinity tag through site-specific proteolytic cleavage with rhinovirus 3C protease (2 μg/ml 3C protease fused to a GST tag) for 12 hr at 4 °C. The protease was subsequently removed by passing the material through a Fast Flow GST-Sepharose column (GE Healthcare). The CDT holotoxin was further purified by cation-exchange chromatography (SP Sepharose Fast Flow, 20 ml bead volume; GE Healthcare) using an ÄKTA FPLC (GE Healthcare). Refolded material was diluted five times to lower the salt concentration to 40 mM NaCl prior to loading the SP Sepharose column. After equilibrating the column with washing buffer (20 mM HEPES [pH 7.5], 2.5 mM DTT), the column was subjected to a gradient increase in salt concentration (from 0 to 500 mM NaCl in 20 mM HEPES [pH 7.5], 2.5 mM DTT during 150 min at flow rate of 3 ml/min). The remaining material was bumped from the column with 1 M NaCl, 20 mM HEPES (pH 7.5), 2.5 mM DTT. Individual fractions (5 ml) or pooled material from elution peak was collected and examined by SDS-PAGE. As a final step in the purification, and at the same time as a test of complex integrity, fractions eluted from the SP Sepharose column that contained intact CDT holotoxin were pooled together, concentrated using Centricon Plus-20 (Millipore, Billerica, Massachusetts, United States), and run on a gel filtration column (Superdex 200 HiLoad 16/60 [120 ml] or Superdex 200 HR 10/30 [25 ml]; GE Health), as indicated in the figure legends. Gel filtration buffers contained 200 mM NaCl, 20 mM HEPES (pH 7.5), 2.5 mM DTT; a 2-ml sample was injected, and 2-ml (from Superdex 200 HiLoad 16/60) or 1-ml (from Superdex 200 HR 10/30) fractions were collected. The peak fractions were examined by SDS-PAGE, and proteins were visualized by staining with Coomassie blue staining solution (50% methanol, 0.05% Coomassie brilliant blue R-50, 10% acetic acid in water) followed by destaining solution (5% methanol, 7% acetic acid in water).
The deletion mutants were generated by PCR and inserted into the same vector as the wild-type genes. The amino acid substitutions were introduced into cdt genes by PCR using Pfu DNA Polymerase (Stratagene, La Jolla, California, United States) and primers containing the appropriate base changes. The template plasmid was subsequently removed by digestion with DpnI prior to transformation. Myc tag sequence (EQKLISEEDL) was fused in-frame to the 3′ end of the wild-type cdtC gene in two subsequent PCRs using the pET-21a-CdtC plasmid as template, a T7 promoter primer as a forward primer in both reactions, and myc1R (5′-GAGTTTCTGCTCGCTACCCTGATTTCTTCG-3′) and myc2R (5′-TTACAGATCCTCTTCAGAGATGAGTTTCTGCTCGCTACC-3′) as reverse primers. The PCR product was initially cloned into the pCR2.1 TOPO vector (Invitrogen, Carlsbad, California, United States), and after sequence verification subcloned into the pET21a. Purification of the mutant proteins and assembly into the mutant CDT holotoxin were performed as previously described for the wild-type complex.
Cell cycle analysis/cellular intoxication assay.
As a measure of toxin potency and cellular intoxication, we have examined by flow cytometry the cell cycle progression of HeLa cells treated with the CDT holotoxin. HeLa cells (2 × 106 cells) were treated with indicated concentrations of wild-type or mutant CDT holotoxin for 3 hr at 37 °C, 5% CO2, and were subsequently washed of toxin and maintained in culture media. At 48 hr after treatment with CDT, cells were harvested, fixed by the addition of cold 70% ethanol during continuous vortexing, and left at 4 °C for 16 hr. The cells were then stained for 2 hr at 23 °C with propidium iodide solution (25 μg/ml propidium iodide, 100 U/ml DNase-free RNase A, 0.04% Triton X-100 in PBS). Relative DNA content was measured by flow cytometry with an FACSCalibur flow cytometer (Becton-Dickinson, Palo Alto, California, United States). The data were collected from 20,000 cells, and analysis was performed with Cell Quest software.
Labeling of the CDT holotoxin with Alexa Fluor 488 fluorescent dye and flow cytometry.
The CDT holotoxin was conjugated with Alexa Fluor 488 dye using the Alexa Fluor 488 Monoclonal Antibody Labeling Kit (Molecular Probes) according to the manufacturer's instructions. Briefly, 100 μg of CDT holotoxin in PBS/0.1 M sodium bicarbonate (1 mg/ml final protein concentration) was added to the vial of reactive dye and incubated overnight at 4 °C on a rotator shaker. Labeled protein complex is then separated from unconjugated dye using a purification resin supplied by manufacturer. The degree of labeling was determined by measuring the absorbance at 280 nm and 494 nm using a NanoDrop ND-1000 spectrophotometer (Wilmington, Delaware, United States) and calculating the protein concentration and the fluorescent dye concentration using Protein & Labels module of ND-1000 software. The degree of labeling was calculated as moles of dye per mole of protein. This produced calculations of labeling of 2.98 ± 0.11 for wild-type CdtABC, 3.99 ± 0.32 for groove mutant CdtABC, and 5.64 ± 0.33 for aromatic mutant. Goat anti-mouse IgG was purchased already conjugated with Alexa 488 (Molecular Probes), and the degree of labeling was 6.4 moles of dye per mole of protein.
Binding of fluorescently labeled CDT holotoxin to HeLa cells was determined by flow cytometry. HeLa cells (1 × 106) were harvested, washed twice with PBS, and incubated for 2 hr at room temperature with indicated amounts of wild-type or mutant CDT holotoxin in 2% FCS/PBS. Goat anti-mouse IgG antibody conjugated with Alexa Fluor 488 (Molecular Probes) was used as a negative control. Samples were washed three times with PBS and fixed with 1% formaldehyde (Polysciences, Inc., Warrington, Pennsylvania, United States) in PBS prior analysis. The presence of green fluorescence on HeLa cells was measured by flow cytometry using an FACSCalibur flow cytometer (Becton-Dickinson). Analysis was performed with Cell Quest software.
Immunoprecipitation of CDT complexes.
CDT complexes composed of either mutant or wild-type CdtA, wild-type CdtB, and CdtC fused in-frame with myc tag at the C-terminus were immunoprecipitated using an anti-myc antibody (9E10; Santa Cruz Biotechnology, Santa Cruz, California, United States). Then, 10 μg of protein complex was incubated with 10 μg of antibody in binding buffer (20 mM HEPES [pH 7.5], 150 mM NaCl, 0.1% Nonidet P40 Substitute [Sigma, St. Louis, Missouri, United States]) for 3 hr at 4 °C. Immune complexes were incubated with protein G Sepharose 4 Fast Flow (Molecular Probes) in the binding buffer for additional 3 hr at 4 °C. Immunoprecipitates were collected by centrifugation, washed five times with binding buffer, and resuspended in nonreducing sample buffer (2% SDS, 50 mM Tris-Cl [pH 6.8], 0.1% bromophenol blue, 10% glycerol). The samples were boiled for 3 min at 95 °C, briefly centrifuged, and subjected to SDS-PAGE. Protein gels were stained with Coomassie blue dye. Nonreducing conditions were used to minimize dissociation of heavy and light immunoglobulin chains, which masked CDT subunits on a Coomassie blue−stained gel. It was not possible to avoid complete separation of the immunoglobulin chains due to the fact that CDT complexes were already in the buffer with DTT.
Confocal microscopy.
HeLa cells, cultured directly on microscope cover glass, were treated with 5 μg/ml concentration of mutant or wild-type CDT holotoxin labeled with Alexa Fluor 488 in PBS buffer containing 3% BSA and Hoechst 33258 dye (1:500) (Sigma). Following treatment, cells were washed three times with PBS and fixed by incubation in 3.7% formaldehyde for 15 min at room temperature. Fixative was replaced with PBS, and samples were mounted with ProLong Gold Antifade Reagent (Molecular Probes). Samples were dried for at least 48 hr prior to examination under inverted LSM510 confocal microscope (inverted Zeiss Axiovert 200 microscope, Thornwood, New York, United States) using ×60 objective lenses. Images were collected and analyzed using LSM510 software (Zeiss).
Supporting Information
Accession Numbers
The GenBank (http://www.ncbi.nlm.nih.gov/Genbank) accession number for Haemophilus ducreyi CDT is U53215. The Protein Data Bank (http://www.rcsb.org/pdb) accession number for the crystal structure of H. ducreyi CDT is 1SR4 and is shown in Figure 1.
We thank Alison North for assistance and access to the BioImaging Facility and Svetlana Mazel for access to the Flow Cytometry Facility at The Rockefeller University. This work was funded by research funds to C.E.S. from The Rockefeller University. The authors declare that they have no competing financial interests.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. DN and CES conceived and designed the experiments. DN performed the experiments. DN and CES analyzed the data and wrote the paper.
Abbreviations
CDTcytolethal distending toxin
DTTdithiothreitol
==== Refs
References
Lara-Tejero M Galan JE 2002 Cytolethal distending toxin: Limited damage as a strategy to modulate cellular functions Trends Microbiol 10 147 152 11864825
De Rycke J Oswald E 2001 Cytolethal distending toxin (CDT): A bacterial weapon to control host cell proliferation? FEMS Microbiol Lett 203 141 148 11583839
Johnson WM Lior H 1988 A new heat-labile cytolethal distending toxin (CLDT) produced by Campylobacter spp Microb Pathog 4 115 126 2849028
Frisan T Cortes-Bratti X Thelestam M 2002 Cytolethal distending toxins and activation of DNA damage-dependent checkpoint responses Int J Med Microbiol 291 495 499 11890549
Cortes-Bratti X Karlsson C Lagergard T Thelestam M Frisan T 2001 The Haemophilus ducreyi cytolethal distending toxin induces cell cycle arrest and apoptosis via the DNA damage checkpoint pathways J Biol Chem 276 5296 5302 11076947
Shenker BJ Hoffmaster RH Zekavat A Yamaguchi N Lally ET 2001 Induction of apoptosis in human T cells by Actinobacillus actinomycetemcomitans cytolethal distending toxin is a consequence of G2 arrest of the cell cycle J Immunol 167 435 441 11418680
Gelfanova V Hansen EJ Spinola SM 1999 Cytolethal distending toxin of Haemophilus ducreyi induces apoptotic death of Jurkat T cells Infect Immun 67 6394 6402 10569755
Elwell CA Dreyfus LA 2000 DNase I homologous residues in CdtB are critical for cytolethal distending toxin-mediated cell cycle arrest Mol Microbiol 37 952 963 10972814
Lara-Tejero M Galan JE 2000 A bacterial toxin that controls cell cycle progression as a deoxyribonuclease I-like protein Science 290 354 357 11030657
Elwell C Chao K Patel K Dreyfus L 2001
Escherichia coli CdtB mediates cytolethal distending toxin cell cycle arrest Infect Immun 69 3418 3422 11292766
Mao X DiRienzo JM 2002 Functional studies of the recombinant subunits of a cytolethal distending holotoxin Cell Microbiol 4 245 255 11952641
Cortes-Bratti X Frisan T Thelestam M 2001 The cytolethal distending toxins induce DNA damage and cell cycle arrest Toxicon 39 1729 1736 11595635
Hassane DC Lee RB Mendenhall MD Pickett CL 2001 Cytolethal distending toxin demonstrates genotoxic activity in a yeast model Infect Immun 69 5752 5759 11500452
Alby F Mazars R de Rycke J Guillou E Baldin V 2001 Study of the cytolethal distending toxin (CDT)-activated cell cycle checkpoint. Involvement of the CHK2 kinase FEBS Lett 491 261 265 11240139
Frisan T Cortes-Bratti X Chaves-Olarte E Stenerlow B Thelestam M 2003 The Haemophilus ducreyi cytolethal distending toxin induces DNA double-strand breaks and promotes ATM-dependent activation of RhoA Cell Microbiol 5 695 707 12969375
Nesic D Hsu Y Stebbins CE 2004 Assembly and function of a bacterial genotoxin Nature 429 429 433 15164065
Cortes-Bratti X Chaves-Olarte E Lagergard T Thelestam M 2000 Cellular internalization of cytolethal distending toxin from Haemophilus ducreyi
Infect Immun 68 6903 6911 11083812
Lara-Tejero M Galan JE 2001 CdtA, CdtB, and CdtC form a tripartite complex that is required for cytolethal distending toxin activity Infect Immun 69 4358 4365 11401974
Deng K Latimer JL Lewis DA Hansen EJ 2001 Investigation of the interaction among the components of the cytolethal distending toxin of Haemophilus ducreyi
Biochem Biophys Res Commun 285 609 615 11453636
Lewis DA Stevens MK Latimer JL Ward CK Deng K 2001 Characterization of Haemophilus ducreyi cdtA, cdtB, and cdtC mutants in in vitro and in vivo systems Infect Immun 69 5626 5634 11500438
Deng K Hansen EJ 2003 A CdtA-CdtC complex can block killing of HeLa cells by Haemophilus ducreyi cytolethal distending toxin Infect Immun 71 6633 6640 14573688
Lee RB Hassane DC Cottle DL Pickett CL 2003 Interactions of Campylobacter jejuni cytolethal distending toxin subunits CdtA and CdtC with HeLa cells Infect Immun 71 4883 4890 12933829
McSweeney LA Dreyfus LA 2005 Carbohydrate-binding specificity of the Escherichia coli cytolethal distending toxin CdtA-II and CdtC-II subunits Infect Immun 73 2051 2060 15784546
Akifusa S Heywood W Nair SP Stenbeck G Henderson B 2005 Mechanism of internalization of the cytolethal distending toxin of Actinobacillus actinomycetemcomitans
Microbiology 151 1395 1402 15870449
Frisk A Lebens M Johansson C Ahmed H Svensson L 2001 The role of different protein components from the Haemophilus ducreyi cytolethal distending toxin in the generation of cell toxicity Microb Pathog 30 313 324 11399138
Sandvig K Grimmer S Iversen TG Rodal K Torgersen ML 2000 Ricin transport into cells: Studies of endocytosis and intracellular transport Int J Med Microbiol 290 415 420 11111920
Guerra L Teter K Lilley BN Stenerlow B Holmes RK 2005 Cellular internalization of cytolethal distending toxin: A new end to a known pathway Cell Microbiol 7 921 934 15953025
Saiki K Gomi T Konishi K 2004 Deletion and purification studies to elucidate the structure of the Actinobacillus actinomycetemcomitans cytolethal distending toxin J Biochem (Tokyo) 136 335 342 15598890
Shenker BJ Besack D McKay T Pankoski L Zekavat A 2005 Induction of cell cycle arrest in lymphocytes by Actinobacillus actinomycetemcomitans cytolethal distending toxin requires three subunits for maximum activity J Immunol 174 2228 2234 15699156
Shenker BJ Besack D McKay T Pankoski L Zekavat A 2004
Actinobacillus actinomycetemcomitans cytolethal distending toxin (Cdt): Evidence that the holotoxin is composed of three subunits: CdtA, CdtB, and CdtC J Immunol 172 410 417 14688349
Rini JM 1995 X-ray crystal structures of animal lectins Curr Opin Struct Biol 5 617 621 8574697
Cope LD Lumbley S Latimer JL Klesney-Tait J Stevens MK 1997 A diffusible cytotoxin of Haemophilus ducreyi
Proc Natl Acad Sci U S A 94 4056 4061 9108104
Kholod N Mustelin T 2001 Novel vectors for co-expression of two proteins in E. coli Biotechniques 31 322 323 326 328 11515368
|
16304609
|
PMC1287909
|
CC BY
|
2021-01-05 12:11:28
|
no
|
PLoS Pathog. 2005 Nov 18; 1(3):e28
|
utf-8
|
PLoS Pathog
| 2,005 |
10.1371/journal.ppat.0010028
|
oa_comm
|
==== Front
PLoS PathogPLoS PathogppatplpaplospathPLoS Pathogens1553-73661553-7374Public Library of Science San Francisco, USA 1630461010.1371/journal.ppat.001003005-PLPA-RA-0114R2plpa-01-03-06Research ArticleImmunologyInfectious DiseasesYeast and FungiMus (mouse)
Aspergillus fumigatus Triggers Inflammatory Responses by Stage-Specific β-Glucan Display Inflammatory Responses to
A. fumigatusHohl Tobias M 12Van Epps Heather L. 2Rivera Amariliz 2Morgan Laura A 2Chen Patrick L 3Feldmesser Marta 3Pamer Eric G 12*
1 Infectious Diseases Service, Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York, New York, United States of America
2 Immunology Program, Sloan-Kettering Institute, Memorial Sloan-Kettering Cancer Center, New York, New York, United States of America
3 Division of Infectious Diseases, Albert Einstein College of Medicine, Bronx, New York, United States of America
Cormack Brendan EditorJohns Hopkins University, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 18 11 2005 1 3 e304 8 2005 13 10 2005 Copyright: © 2005 Hohl 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.Inhalation of fungal spores (conidia) occurs commonly and, in specific circumstances, can result in invasive disease. We investigated the murine inflammatory response to conidia of Aspergillus fumigatus, the most common invasive mold in immunocompromised hosts. In contrast to dormant spores, germinating conidia induce neutrophil recruitment to the airways and TNF-α/MIP-2 secretion by alveolar macrophages. Fungal β-glucans act as a trigger for the induction of these inflammatory responses through their time-dependent exposure on the surface of germinating conidia. Dectin-1, an innate immune receptor that recognizes fungal β-glucans, is recruited in vivo to alveolar macrophage phagosomes that have internalized conidia with exposed β-glucans. Antibody-mediated blockade of Dectin-1 partially inhibits TNF-α/MIP-2 induction by metabolically active conidia. TLR-2- and MyD88-mediated signals provide an additive contribution to macrophage activation by germinating conidia. Selective responsiveness to germinating conidia provides the innate immune system with a mechanism to restrict inflammatory responses to metabolically active, potentially invasive fungal spores.
Synopsis
Aspergillus fumigatus is a mold that forms spores that are often inhaled by mammals. Humans with normal immune systems inhale several hundred A. fumigatus spores per day without developing detectable disease. Immunocompromised hosts, on the other hand, can develop invasive A. fumigatus infections. In these cases, inhaled spores germinate and form tissue-invasive hyphae that invade blood vessels and disseminate to remote tissues. The aim of this study was to investigate the normal inflammatory response to inhaled spores in mouse lungs. The researchers found that the earliest stage of spore germination, referred to as “swelling,” triggered the recruitment of inflammatory cells into the lung airways. Consistent with this finding, lung-derived cells stimulated with swollen spores secreted copious amounts of proteins that attract inflammatory cells. Analysis of spores revealed that swelling is accompanied by surface expression of β-glucan polymers. These carbohydrates, which are not present on the surface of mammalian cells, induce signaling by the mammalian Dectin-1 receptor and activate the expression of genes that promote the inflammatory response. The results suggest that mammalian lungs have evolved a mechanism to distinguish swollen and potentially threatening spores from innocuous, dormant spores.
Citation:Hohl TM, Van Epps HL, Rivera A, Morgan LA, Chen PL, et al. (2005) Aspergillus fumigatus triggers inflammatory responses by stage-specific β-glucan display. PLoS Pathog 1(3): e30.
==== Body
Introduction
The innate immune system confronts a wide spectrum of microbes, extending from the innocuous to the highly pathogenic [1–3]. Overly robust inflammatory responses can compromise host tissues and organ function, which, in the case of the lungs, can be severely debilitating or even lethal. In contrast, inadequate immune responses to pathogens can promote tissue invasion and systemic dissemination with equally dire consequences. Lung airways are perpetually exposed to inhaled particulate materials that include pollens, viruses, and bacterial and fungal spores. While many of these particles are innocuous, some spores have the potential to germinate and cause invasive diseases. Distinguishing these rare pathogenic microbes from the innocuous majority and calibrating inflammatory responses to the invasive potential of the microbe are fundamental challenges faced by the pulmonary innate immune system.
Aspergillus fumigatus is a mold that forms conidia with a diameter of 2–3 μm [4]. Owing to their small size, A. fumigatus conidia can bypass mucociliary clearance mechanisms and are inhaled into terminal airways and phagocytosed by alveolar macrophages (AMØs) [5,6]. Conidia are killed in a phagocyte oxidase-dependent manner [7]. Neutrophils recruited to the site of infection form a second line of defense against germinating conidia [5].
Conidial germination begins with swelling and progresses to germ tube formation and hyphal extension [8]. Genetic, acquired, or pharmacologically induced states that impair macrophage and neutrophil function enable in vivo conidial germination and the formation of fungal hyphae that can invade pulmonary tissues, enter the bloodstream, and disseminate to remote tissues [9].
Hyphae contain four major carbohydrate polymers: chitin, galactomannan, branched β–1,3/β–1,6 glucans, and linear β–1,3/β–1,4 glucans [10]. In contrast to hyphae, conidia have morphologically distinct features—an outermost proteinaceous rodlet layer [11] and an inner cell wall layer that is exposed during swelling [12], a process that can occur within macrophages [7]. The complex composition of the conidial cell wall is incompletely defined [13], and the identity of conidial surface molecules that induce proinflammatory cytokine/chemokine responses through cognate host cell receptors has not been determined.
Germline encoded pattern recognition receptors constitute a major surveillance and defense mechanism against microbial invaders [1–3,14]. The 11 mammalian toll-like receptors (TLRs) differ in their subcellular localization, tissue distribution, and ligand specificity, yet signal through a limited set of adaptor proteins that includes myeloid differentiation factor 88 (MyD88). TLR-mediated signals induce proinflammatory responses, activate antimicrobial effector functions, and stimulate adaptive immune responses. In the case of A. fumigatus, TLR-2, TLR-4, and MyD88 have been implicated in the recognition of conidia and hyphae [15–20]. TLR-2-, TLR-4-, and MyD88-deficient animals survive pulmonary A. fumigatus infection [21]; this finding implies the existence of MyD88-independent pathways in host defense against inhaled conidia.
Dectin-1 [22,23], a type II transmembrane protein that belongs to the NK-like C-type lectin-like receptor family, is a pattern recognition receptor for β–1,3/β–1,6-linked glucans [24]. Dectin-1 binds to the inert, Saccharomyces cerevisiae–derived, β-glucan-rich particle zymosan [24,25], Candida albicans [26], murine sp. Pneumocystis carinii [27], and Coccidioides posadasii [28]. Dectin-1 activation results in phagocytic, proinflammatory, and antimicrobial responses [25,29–31]. The contribution of Dectin-1-mediated signals to innate immune defense against A. fumigatus infection remains undefined.
To characterize the innate inflammatory response following spore inhalation, we investigated cell recruitment into the airways of immune competent mice exposed to live and killed A. fumigatus conidia. The early pulmonary innate immune response to these two challenges is distinct; live conidia induce rapid neutrophil recruitment, while killed conidia are far less stimulatory. Live conidia display β-glucans on their cell surface during swelling and hyphal formation; this process triggers Dectin-1 recruitment to phagosomal membranes of AMØs in vivo. Exposure of macrophages to germinating conidia reveals a direct correlation between surface β-glucan display and inflammatory cytokine/chemokine induction. This induction is partially inhibited by antibody-mediated blockade of Dectin-1 signaling. The innate immune system, therefore, restricts responsiveness to conidia that are germinating and ignores the less threatening dormant conidia. This mechanism may limit pulmonary damage following microbial spore inhalation.
Results
Bronchoalveolar Cellular Infiltrates Vary in Response to Live and Heat-Killed Conidia
To examine in vivo inflammatory responses to metabolically active and inert A. fumigatus conidia, 107 live or heat-killed conidia were administered intratracheally to immune competent mice. Cell recruitment into the bronchoalveolar lavage (BAL) fluid collected 24 h later was quantified and analyzed by flow cytometry. In a representative experiment, BAL fluid recovered from mice that were infected with live conidia contained 3.0 ± 1.2 × 106 cells (Figure 1A) with a predominant neutrophilic infiltrate consisting of Ly6Ghi, CD11b+, and CD11c− cells (Figure 1B; unpublished data). In comparison, heat-killed conidia induced modest cell recruitment into the BAL (0.36 ± 0.1 × 106 cells), with a predominance of CD11c+, Ly6Gint AMØs. BAL fluid collected from mice administered PBS-Tween (vehicle) contained only 0.97 ± 0.17 × 105 cells with >90% AMØs. The 8- to 9-fold increase in total BAL cells in mice administered live rather than heat-killed conidia was almost entirely attributable to the increased influx of neutrophils.
Macrophages and Dendritic Cells Secrete TNF-α and MIP-2 in Response to Live but Not Heat-Killed Conidia
Neutrophil recruitment to sites of infection and inflammation depends, in part, on chemokine signals. In immune competent mice, the chemokine receptor CXCR2 is critical for host survival following A. fumigatus pulmonary challenge [32]. CXCR2 binds the neutrophil chemoattractants macrophage inflammatory protein-2 (MIP-2) and KC. The proinflammatory cytokine tumor necrosis factor-α (TNF-α) is produced in response to A. fumigatus challenge in vivo and influences multiple facets of the early innate immune response to inhaled conidia [33,34].
To determine whether live conidia differ from heat-killed conidia in their ability to stimulate inflammatory cytokine and chemokine production, bone marrow-derived macrophages (BMMØs) were stimulated for 18 h with both types of conidia. Cell culture supernatants from BMMØs stimulated with live conidia contained >50-fold higher TNF-α and >10-fold higher MIP-2 concentrations than cells exposed to heat-killed conidia (Figure 1C). Paraformaldehyde inactivation of resting conidia yielded similar results (unpublished data).
Since dendritic cells (DCs) [35,36] and AMØs [5] are implicated in innate immune defense against A. fumigatus, their responses to live and heat-killed conidia were investigated. While bone marrow-derived dendritic cells (BMDCs) (Figure 1D) stimulated with live conidia produced robust levels of TNF-α/MIP-2, AMØs were the most responsive cell type (Figure 1E). As noted for BMMØs, the BMDC and AMØ response to heat-killed conidia was markedly attenuated.
These experiments were performed in the presence of 0.5 μg/ml voriconazole to prevent fungal overgrowth of cell cultures. This drug concentration was chosen since it did not prevent conidial swelling or compromise viability during an overnight incubation ([37]; unpublished data). Hyphae were not observed under these conditions but formed rapidly if voriconazole was removed from the medium. The results from the macrophage/DC stimulation experiments thus indicate that conidial swelling is sufficient to induce inflammatory cytokine/chemokine production.
Figure 1 Live but Not Heat-Killed A. fumigatus Conidia Induce Inflammatory Responses
(A and B) In vivo cellular recruitment into lung airways.
(A) Absolute number of BAL cells, total neutrophils (Ly6Ghi, CD11c−) in BAL, and total macrophages (CD11c+, Ly6Gint) in BAL for C57BL/6 mice 24 h after intratracheal instillation of PBS-Tw (vehicle), 107 heat-killed, or 107 live conidia. The bar graphs show the average cell numbers + standard deviation from four mice per group. One of three representative experiments is shown.
(B) Flow cytometric analysis of live BAL cells stained for Ly6G and CD11c. Gates representing neutrophils (Ly6Ghi, CD11c−) and AMØs (CD11c+, Ly6Gint) are shown and the frequencies of these cell populations are indicated in representative BAL samples.
(C–E) Ex vivo TNF-α/MIP-2 secretion by BMMØs (C), BMDCs (D), and AMØs (E) stimulated with live or heat-killed A. fumigatus conidia, PBS-Tw, or LPS (100 ng/ml) for 18 h in medium containing 0.5 μg/ml voriconazole. TNF-α/MIP-2 concentrations in the culture supernatants were determined by ELISA. In (C), the values for TNF-α/MIP-2 secretion induced by LPS were reduced by a factor of ten. The bar graphs represent the average cytokine production ± standard deviation by cells in 3–4 wells per condition.
Heat-Killed Conidia That Have Initiated the Germination Process Recruit Neutrophils into the Airways and Induce TNF-α/MIP-2 Secretion
Heat inactivation arrests conidial swelling and germination at a defined time point. To examine the effect of conidial swelling on in vivo inflammatory responses, conidia were swollen, heat-killed to prevent further germination, and administered intratracheally into mice. In contrast to conidia killed in the resting state (see Figure 1B), conidia that were swollen prior to heat inactivation induced an inflammatory cell influx comparable both in number and cell type to that induced by an inoculum of live conidia (Figure 2A and 2B). This result indicates that the pulmonary innate immune system selectively recognizes conidia that have initiated germination; once swollen, conidial viability is no longer critical for the initiation of inflammatory responses.
Conidia heat-killed prior to significant conidial swelling (3 h incubation in RPMI medium) did not induce TNF-α/MIP-2 secretion by BMMØs (Figure 2C). In contrast, fungal preparations containing heat-killed swollen conidia (5 h incubation in RPMI medium) induced approximately the same amount of TNF-α/MIP-2 as live, resting conidia (see Figure 1C). As germination proceeds for longer time periods prior to heat-killing (7 h incubation in RPMI medium), TNF-α/MIP-2 production increases dramatically. These results indicate that conidial germination alters the fungal cell surface and renders it far more inflammatory.
β-Glucans Become Surface-Exposed during Swelling and Germ Tube Formation
β-glucans have been implicated as fungus-derived targets of mammalian innate immune receptors [29]. We generated a β-glucan specific monoclonal antibody (mAb 744) by immunizing mice with A. fumigatus conidia. This antibody binds immobilized laminarin (Mw < 10,000), a β–1,3 glucan polymer with β–1,6 interstrand linkages (Figure 3A). The mAb 744 binding is disrupted by addition of laminarinase, defining β-glucan as the cognate antigen. Confocal microscopy of unfixed fungal cells labeled with mAb 744 indicated that β-glucan cell surface immunoreactivity was prominent in swollen conidia and in early germlings, particularly at sites of initial hyphal extension (Figure 3B).
Resting conidia failed to stain with mAb 744, demonstrating that β-glucans are exposed only on germinating spores. No fungal cell surface fluorescence was observed if mAb 744 was depleted by a prior incubation with the insoluble β-glucan-rich particle zymosan (Figure 3C) or replaced with an isotype control antibody (Figure 3D). These results demonstrate that β-glucans are exposed on the conidial surface in a stage-specific fashion, either through de novo synthesis or exposure of a concealed layer.
Figure 2 Germinating A. fumigatus Conidia Are Highly Inflammatory
(A) Absolute number of BAL cells, total neutrophils in BAL, and total macrophages in BAL for C57BL/6 mice 24 h after intratracheal instillation with either heat-killed swollen conidia or live conidia. The bar graphs show the average cell numbers + standard deviation from four mice per group.
(B) Flow cytometric analysis of live BAL cells stained for Ly6G and CD11c as in Figure 1B.
(C) A. fumigatus conidia were allowed to initiate germination for 3, 5, or 7 h, heat-killed, and added to BMMØ cultures for 18 h. TNF-α/MIP-2 concentrations in the supernatants were determined by ELISA. The bar graphs represent the average ± standard deviation of three wells per condition. One of four experiments is shown.
β-Glucan Exposure Alters Dectin-1 Intracellular Distribution in AMØs
To determine the effect of β-glucan exposure on the in vivo distribution of Dectin-1 in AMØs, mice were inoculated intratracheally with either heat-killed swollen conidia (Figure 4A–4D) or heat-killed resting conidia (Figure 4E–4H). AMØs were harvested 45 min later by BAL and processed immediately for immunofluorescence microscopy. Immunostaining for Dectin-1 revealed that phagocytosis of heat-killed swollen conidia (β-glucan surface-positive) (Figure 4B) triggered recruitment of Dectin-1 to the phagosomal membrane, resulting in a ring pattern of fluorescence surrounding the ingested conidia (Figure 4C). Merging the fluorescence images confirmed that the red Dectin-1 fluorescence signal surrounded (and partially overlapped) the green fluorescence signal from the labeled, surface-exposed conidial β-glucans (Figure 4D). Voriconazole treatment and heat-killing did not appreciably influence β-glucan immunostaining on the surface of swollen conidia (see Figures 3B and 4B; unpublished data).
Heat-killed resting conidia are β-glucan surface-negative (Figure 4F), in contrast to heat-killed C. albicans yeast cells [26]. Phagocytosis of heat-killed resting conidia did not result in Dectin-1 redistribution around the ingested conidia (Figure 4G). These results suggest that Dectin-1 associates with inhaled A. fumigatus conidia in a β-glucan-dependent manner in vivo.
Conidia Activate Two Parallel Macrophage Signaling Pathways That Lead to TNF-α and MIP-2 Production
To examine whether Dectin-1 mediates proinflammatory responses to A. fumigatus, TNF-α/MIP-2 secretion by BMMØs in response to live conidia was measured in the presence of the anti-Dectin-1 blocking antibody 2A11. Cytochalasin D was added to these assays to prevent antibody degradation or dissociation in the vacuolar compartment. In comparison to cells incubated with an IgG2b control antibody, 2A11 administration reduced TNF-α/MIP-2 secretion in BMMØs by approximately 40%–50% (Figure 5A). On the basis of this result, Dectin-1-mediated signaling accounts for nearly one-half of the TNF-α/MIP-2 released by BMMØs.
To assess the role of MyD88-dependent TLR-signaling in TNF-α/MIP-2 release, MyD88−/− BMMØs were stimulated with live conidia. In MyD88−/− BMMØs, TNF-α/MIP-2 secretion was consistently reduced by approximately 40%–50% as compared to wild-type (WT) (MyD88+/+) control cells (Figure 5B). Since TLR-2 has been implicated in conidial recognition [17–19] and may interact with Dectin-1 [25,29] TNF-α/MIP-2 release was also examined in TLR-2−/− BMMØs. TLR-2−/− BMMØs released approximately 20% less TNF-α/MIP-2 than WT (TLR-2+/+) BMMØs (Figure 5B). This result suggests that TLR-2 engagement does not fully account for the MyD88-dependent TNF-α/MIP-2 release triggered by A. fumigatus conidia.
Since Dectin-1- and MyD88-dependent signals each account for 40%–50% of TNF-α/MIP-2 release by BMMØs, MyD88−/− cells were treated with the anti-Dectin antibody and stimulated with conidia to determine whether Dectin-1-mediated TNF-α/MIP-2 release relies on the presence of MyD88. MyD88−/− BMMØs incubated with anti-Dectin-1 antibody secreted approximately 70% less TNF-α/MIP-2 than MyD88−/− BMMØs treated with an isotype control antibody (Figure 5C). The extent of TNF-α/MIP-2 inhibition by Dectin-1 blockade was similar in MyD88−/− and WT BMMØs. This result suggests that conidia activate independent pathways that signal through MyD88 and Dectin-1 and that these pathways are additive with respect to TNF-α/MIP-2 secretion. A similar independence and summed contribution of MyD88- and Dectin-1-mediated signals to TNF-α/MIP-2 release was observed if either voriconazole (to exclude a drug-mediated effect) or cytochalasin D (to exclude an effect of blocking phagocytosis) was omitted from the samples (unpublished data).
Figure 3 Germinating but Not Resting A. fumigatus Conidia Display β-Glucans on the Cell Surface
(A) The mAb 744 detects β-glucans. Decreasing laminarin concentrations were coated onto ELISA plates in the presence or absence of laminarinase. 10 μg/ml mAb 744 was added to the wells and detected using an alkaline phosphatase conjugated anti-mouse IgM.
(B–D) A. fumigatus conidia were incubated in RPMI for 7 h, stained with mAb 744 (B), mAb 744 depleted by zymosan (C), or IgM control antibody (D), followed by Alexa Fluor 594-anti-mouse IgM, and examined by confocal microscopy. Representative DIC/epifluorescence confocal images of resting conidia (open arrowheads), swollen conidia (black arrowheads), and early germlings (white arrowheads) are shown in overlay. Scale bar = 10 μm.
The additive effect of Dectin-1- and MyD88-dependent signaling accounted for approximately 80%–85% of TNF-α/MIP-2 secretion by BMMØs (Figure 5C). In AMØs, the combination of Dectin-1 blockade and MyD88 deficiency led to a similar reduction in TNF-α/MIP-2 secretion, approximately 70%–80% as compared to the control condition with WT cells incubated with an isotype control antibody (Figure 5D).
Figure 4 Dectin-1 Is Recruited to AMØ Phagosomes Containing Swollen Conidia with Surface-Exposed β-Glucans
(A–H) AMØs were harvested from C57BL/6 mice 45 min after intratracheal instillation of 107 heat-killed swollen conidia (A–D) or heat-killed resting conidia (E–H), processed at 4 °C, and immunostained for ingested conidial β-glucans with mAb 744, followed by a FITC-coupled anti-mouse IgM (B and F) and for Dectin-1 (C and G) with goat anti-Dectin polyclonal antibodies, followed by Alexa Fluor 594-coupled anti-goat IgG. AMØs were examined by DIC microscopy (A and E) and epifluorescence confocal microscopy (B–D and F–H).
(D and H) Merged images of β-glucan and Dectin-1 immunofluorescence.
Scale bar = 10 μm.
Discussion
In this study, we show that innate immune responses to A. fumigatus conidia depend on host recognition of morphologic changes that occur during the first step of germination, conidial swelling. Our data indicate that remodeling and expansion of the cell wall during conidial swelling results in the exposure of β-glucan polymers on the fungal cell surface. Swollen conidia display surface β-glucan and associate with Dectin-1 in AMØs isolated from in vivo challenged mice. Two parallel innate immune signaling pathways respond to this process. One pathway signals through the TLR adaptor protein MyD88, the other through the β-glucan receptor Dectin-1.
Figure 5 Dectin-1- and MyD88-Mediated Signals Are Induced by A. fumigatus Conidia
(A–D) WT (white bars, [A–C]), MyD88−/− (dark grey bars, [B–D]), and TLR-2−/− (light grey bars, [B]) BMMØs (A–C) or AMØs (D) were stimulated with conidia for 18 h in medium containing 0.5 μg/ml voriconazole, and TNF-α/MIP-2 secretion was measured by ELISA. (A,C,D) BMMØs (A and C) and AMØs (D) were incubated with 2A11 (anti-Dectin-1 mAb) or an isotype control antibody in the presence of 2 μM cytochalasin D. (A and B) ELISA results from test conditions were averaged among three to five experiments and expressed as a percentage ± standard deviation of the averaged value obtained for the control condition. (C and D) show representative experiments.
All experiments were performed with three to six replicates per condition.
Several features of this response are striking. First, neither innate immune response pathway is activated by dormant conidia; each responds only to conidia that have initiated the germination process. We propose that restricted recognition of germinating conidia provides a mechanism to focus pulmonary inflammatory responses on spores that are most likely to cause invasive disease. Second, the contributions of the Dectin-1- and MyD88-mediated signaling pathways are, in the case of A. fumigatus conidia, additive. The ability of AMØs to generate Dectin-1-dependent, MyD88-independent inflammatory responses to germinating conidia may, in part, provide an explanation for the survival of TLR-2- and MyD88-deficient mice in a pulmonary infection model [21].
The relationship between MyD88- and Dectin-1-dependent signaling has not been examined previously in the host response to intact fungi. Studies with the β-glucan-rich particle zymosan revealed that TNF-α secretion by macrophages is fully dependent on the presence of TLR-2 and MyD88 and is blocked by the addition of glucan phosphate, a soluble high-molecular-weight β-glucan [25,29]. Experiments with transfected cell lines found that co-expression of Dectin-1- and TLR-2- enhanced zymosan-dependent transcriptional responses [25]. Zymosan also binds the extracellular domain of TLR-2 in vitro [38]. Taken together, these results suggest that Dectin-1 and TLR-2/MyD88 collaborate to mount zymosan-dependent inflammatory responses. In our experimental system, Dectin-1- and MyD88-mediated TNF-α/MIP-2 secretion did not depend on the functional presence of the other signaling molecule.
Our results, to our knowledge, differ from previous work examining TLR- and MyD88-dependent signaling in response to conidia and hyphae. This may be, in part, due to differences in TLR/MyD88- and Dectin-1-dependent signaling by macrophages isolated from different anatomical sites. Mambula et al. found that in peritoneal macrophages MyD88-dependent signals account for approximately 90% of TNF-α secreted in response to live resting and heat-killed swollen conidia [17]. In contrast, our results suggest that MyD88-dependent signals account for roughly 50% of the TNF-α secreted by BMMØs and AMØs in response to live, resting conidia. Macrophages from different sources may also react uniquely to killed, resting conidia. For example, Marr et al. also observed that BMMØs do not produce TNF-α in response to heat-killed resting conidia [20], while two other studies demonstrated that peritoneal macrophages secrete TNF-α in response to heat- or ethanol-killed conidia in a largely TLR-4-dependent fashion [18,19]. We have investigated AMØs given their central role as first responders to inhaled conidia. Unlike peritoneal macrophages, AMØs and BMMØs kill conidia and prevent germination [7,20,39].
Conidial ligands that activate TLR/MyD88-dependent signals remain unknown and may involve molecules derived from carbohydrates, proteins, or lipids. The complex interaction between conidia and mammalian cells involves multiple sets of ligand-receptor interactions beyond the β-glucan/Dectin-1 and the unknown ligand(s)-TLR/MyD88 pairs. Experiments using soluble carbohydrates or antibodies as blocking agents have implicated a number of receptors in conidial binding or internalization [35,40,41]; these include a mannosyl-fucosyl receptor (in murine AMØs), a β-glucan inhibitable receptor (in human monocytes), the macrophage mannose receptor, the DEC205 lectin, and the CD11b/CD18 integrin (all in murine pulmonary DCs). Conidial galactomannan binds the soluble pattern recognition receptor pentraxin 3 [42], a C-type lectin of galactomannan-specificity on Langerhans cells [43], as well as DC-specific intercellular adhesion molecule 3-grabbing nonintegrin (DC-SIGN) on human DCs [36]. However, it remains unknown whether DC-SIGN or any of these other cell surface receptors implicated in conidial binding or internalization contribute to inflammatory cytokine/chemokine induction and neutrophil recruitment.
Human fungal pathogens propagate and grow using different mechanisms. For example, the yeast C. albicans divides by separation of mother and daughter cells and can switch between yeast and filamentous growth. It has been recently demonstrated that the process of C. albicans cell division creates bud and birth scars with exposed β-glucans [26]. Unlike filamentous growth, C. albicans yeast growth is thus susceptible to Dectin-1-dependent recognition and antifungal responses [26].
Paracoccidioides brasiliensis, a thermal dimorph, forms conidia that are inhaled and transform into pathogenic yeast forms within the host. The dimorphic transition is associated with a change in the polymer linkage from β-glucan to α-glucan in the cell wall [44]. In contrast, the mold A. fumigatus forms conidia that swell and germinate prior to hyphal extension; our study indicates that this process exposes β-glucan polymers on the surface of swollen conidia. The appearance of surface-exposed β-glucans at specific stages of fungal growth and division as well as their conserved presence among fungal organisms renders these carbohydrate polymers ideal recognition molecules for innate immune receptors that trigger antifungal responses.
C. albicans, P. jiroveci, C. posadasii, and A. fumigatus vary significantly in terms of human disease and host susceptibility, tissue tropism and damage, and metastatic potential [15]. However, innate immune defenses against these organisms are mobilized, at least in part, through the recognition of exposed β-glucans by Dectin-1-dependent pathways. Despite their common β-glucans, yeasts, molds, and zymosan differ in terms of other surface and cell-wall components. It is likely that these disparities account for distinct innate immune responses to different fungal organisms and components. The identification of both common and distinct fungal components that activate innate immune responses will be an intense focus of further research. In turn, the repertoire of signaling pathways activated in leukocytes and other host cells will undoubtedly yield experimental strategies to modulate inflammatory states and responses.
Materials and Methods
Chemical reagents and antibodies.
Chemical reagents were purchased from Sigma-Aldrich (St. Louis, Missouri, United States) unless noted otherwise. All cell culture reagents were purchased from Invitrogen (Carlsbad, California, United States).
The mAb 744 (IgMκ) was generated by the fusion of splenocytes from a BALB/c mouse immunized with heat-killed resting conidia of A. fumigatus strain 293 to the Ag8–653 myeloma cell line. Hybridomas were screened for binding to the conidial surface by ELISA and were cloned twice on soft agar.
The anti-Dectin-1 mAb 2A11 (rat IgG2b) was obtained from G. Brown [24] for use in pilot experiments. The 2A11 mAb is commercially available from Cell Sciences (Canton, Massachusetts, United States). Affinity-purified goat anti-Dectin and goat IgG control antibodies were from R&D Systems (Minneapolis, Minnesota, United States), rat IgG2b isotype controls and antibodies for flow cytometry (see below) were from BD Biosciences Pharmingen (San Diego, California, United States), FITC-coupled donkey anti-mouse IgM were from Jackson Immunoresearch (West Grove, Pennsylvania, United States), and Alexa Fluor 594-coupled goat anti-mouse IgM and Alexa-Fluor 594-coupled donkey anti-goat IgG were from Molecular Probes (Eugene, Oregon, United States).
Aspergillus growth and culture.
A. fumigatus strain 293 was grown on Sabouraud dextrose agar slants incubated at 37 °C for 5–8 d. Conidia were dislodged from slants by gentle tapping and then resuspended in PBS containing 0.025% (w/v) Tween-20 (PBS-Tw), filtered twice through a 40-μm nylon cell strainer (BD Falcon), and, as required, heat-killed at 100 °C for 30 min in a heating block or at 121 °C for 15 min in an autoclave.
Conidia were incubated at 37 °C at a concentration of 5 × 106 conidia/ml in RPMI for the indicated times. Preparations incubated for 5 h contained swollen conidia. Preparations incubated for 7 h included early germlings with <10 μm hyphal extensions. For in vivo experiments shown in Figures 2A and 4A–D, homogeneous preparations of swollen conidia were prepared by incubating conidia in RPMI containing 0.5 μg/ml voriconazole (Pfizer, New York, New York, United States) for 12 h at 37 °C. Fungal cells were washed twice, adjusted to 2 × 108/ml in PBS-Tw, and stored at 4 °C for use within 48 h. The conidial concentration and the efficiency of heat-killing was verified by plating serial dilutions on Sabouraud dextrose agar.
Animal care.
C57BL/6 mice were purchased from Jackson Laboratory (Bar Harbor, Maine, United States). MyD88−/− and TLR-2−/− mice were backcrossed at least ten generations on the C57BL/6 background and maintained under pathogen-free conditions in the animal care facilities at Memorial Sloan-Kettering Cancer Center (New York, New York, United Sates). In vivo studies were performed in accordance with institutional guidelines regarding animal care.
Intratracheal injections and flow cytometry.
A blunt-end, 20-gauge needle was used to administer 107 conidia (in 50 μl PBS-Tw) intratracheally to mice anesthetized with isoflurane and immobilized in an upright position. Mice were sacrificed either 45 min or 24 h later for recovery of BAL cells though eight stepwise 0.5-ml rinses using a sterile Angiocath plastic catheter (Becton-Dickinson [BD], Palo Alto, California, United States) inserted into the trachea. The BAL specimens were harvested with PBS, 5% FCS for flow cytometry, or RP10 (see below) for immunofluorescence experiments.
BAL cells were counted using a hemocytometer, incubated in 5 μg/ml anti-Fcγ III/II receptor (clone 2.4G2), stained with anti-Ly6C-FITC (AL-21), anti-Ly6G-PE (1A8), anti-CD11b-PERCP (M1/70), and anti-CD11c-APC (HL3) in FACS buffer (PBS, 1% FCS, and 0.05% NaN3), and analyzed on a BD LSR flow cytometer. AMØs stained with anti-Ly6G-PE gave rise to a Ly6Gint fluorescence signal that reflects autofluorescence of this cell type.
Cell culture.
BMDCs were eluted from the tibias and femurs of 8- to 12-wk-old mice in RP10 (RPMI, 10% FCS, and 5 mM HEPES, 1.1 mM L-glutamine, 0.5 U/ml penicillin, 0.5 μg/ml gentamicin, 50 μg/ml streptomycin, and 50 μM 2-ME), filtered through a 100-μm nylon mesh, and cultured in RP20 (20% FCS) supplemented with 30% (v/v) L929 cell supernatant (source of M-CSF) at a density of 0.25–0.5 × 106 cells/ml (in 100-mm cell culture dishes) for the generation of BMMØs or in RP10 and 2% (v/v) p815 cell supernatant (source of GM-CSF) at a density of 106 cells/ml (in 24-well dishes) for the generation of BMDCs.
For BMMØ cultures, the medium was exchanged on day 3. Cells were plated in 0.2 ml RP10 at a density of 4–6 × 105 cells/ml in 96-well plates on day 5–6 for conidial stimulation 6–24 h later. For BMDC cultures, non-adherent cells were gently removed on day 2 and fresh medium was added every 2 d. On day 5, the medium was exchanged to RP10 (1 ml) prior to conidial stimulation.
To obtain AMØs, the BAL fluid from 10–12 uninfected mice was pooled and centrifuged at 300 × g for 5 min. BAL cells were plated in 0.2 ml RP10 at a concentration of 3.5–5.0 × 105 cells/ml in 96-well plates. Non-adherent cells were removed after overnight incubation, and AMØs were stimulated with conidia on the day following the harvest. AMØs were >95% pure as judged by microscopic analysis of cells spun on slides and stained with Diff Quik (Fisher Scientific, Pittsburgh, Pennsylvania, United States).
Conidial stimulation of macrophages and DCs.
Conidia were added at a 10:1 cell ratio to BMMØs, BMDCs, and AMØs and incubated for 18 h in RP10 in the presence of 0.5 μg/ml voriconazole (Pfizer), as indicated, and supernatants were collected for ELISA. For antibody inhibition experiments, cells were incubated in 0.2 ml of RP10 containing 5–10 μg/ml 2A11 or isotype control antibody. Conidia were added 30 min later in a volume of 5 μl of PBS-Tw, and no further antibody was added to the samples. Pilot experiments showed maximal inhibition of conidial-dependent TNF-α/MIP-2 secretion if the 2A11 concentration was ≥ 2 μg/ml. Cytochalasin D was added to a final concentration of 2 μM 30 min prior to conidial stimulation, as indicated.
ELISA.
Commercially available ELISA kits for TNF-α (BD Biosciences Pharmingen, San Diego, California, United States) and MIP-2 (R&D Systems) were used according to the manufacturers' instructions. The limit of detection was 15 pg/ml for both TNF-α and MIP-2.
To determine laminarin binding by mAb 744, a 10 μg/ml solution of laminarin from Laminaria digitata was added to wells of a 96-well EIA plate (Corning, Corning, New York, United States) and serially diluted 1:2 in buffer (PBS with 10 mM NaN3) across the plate. For laminarinase treatment, a 10 μg/ml solution of laminarin with 2 U/ml of laminarinase from Penicillium spp. was added to the wells as described above. After incubation at 37 °C for 20 h, the wells were washed, blocked for 1 h at 37 °C in PBS, 2% BSA, and 10 mM NaN3, and washed again. The mAb 744 was added to the wells at a final concentration of 10 μg/ml, and the plates were incubated for 1 h at 37 °C. After washing, a 1:1,000 dilution of alkaline phosphatase-conjugated anti-mouse IgM was added to wells and the ELISA developed with p-nitrophenyl phosphate as a substrate. The A405 was measured using a VersaMax Microplate Reader (Molecular Devices, Sunnyvale, California, United States).
Immunofluorescence and confocal microscopy.
Fungal cells grown in 4-well glass chamber slides (Lab-Tek, Campbell, California, United States) were blocked in RP10 containing 0.025% Tween-20 and 0.05% NaN3 and stained with 10 μg/ml mAb 744 (or a mouse IgM isotype control) followed by 10 μg/ml Alexa Fluor 594-goat anti-mouse IgM. To deplete mAb 744 with zymosan, 1 mg of zymosan was added to 10 μg of mAb 744 in 1 ml of PBS and incubated at 4 °C for 30 min with gentle agitation. Zymosan was removed by centrifugation, and the resulting supernatant was used for immunofluorescence staining.
AMØs harvested by BAL were processed at 4 °C, centrifuged on 8-well glass chamber slides (105 cells/well), fixed and permeabilized with ice-cold ethanol, blocked in PBS, 10% donkey serum, and 5 μg/ml Fc block, and decorated with goat anti-Dectin-1 and mAb 744. The samples were washed and incubated with Alexa Fluor 594-donkey anti-goat IgG and FITC-donkey anti-mouse IgM. Duplicate samples decorated with isotype control primary antibodies did not yield fluorescence signals. All samples were imaged in an upright Leica (Wetzlar, Germany) TCS SP2 AOBS confocal microscopy system using 488 and 594 nm laser lines with a 63x Leica HPX PL APO water-immersion objective (NA = 1,2).
The anti-Dectin-1 antibody, the A. fumigatus strain 293, and TLR-2−/− and MyD88−/− breeding pairs were generous gifts from Gordon Brown (University of Cape Town, South Africa), Michael Anderson (University of Manchester, United Kingdom), and Shizuo Akira (Osaka University, Japan), respectively. We thank Alexander Ploss for assistance with flow cytometry, Guillaume Dorothee for help with macrophage cultures, and Ingrid Leiner and Ewa Menet for technical support. This work was supported by the Sandler Program for Asthma Research to EGP, funds from an NIH T32 training grant to TMH, and NIH R03AI53623 grant to MF. This paper is dedicated to the memory of Peter Hohl.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. TMH, HLV, MF, and EGP conceived and designed the experiments. TMH, HLV, AR, LAM, PLC, and MF performed the experiments. TMH, MF, and EGP analyzed the data. TMH, MF, and EGP contributed reagents/materials/analysis tools. TMH and EGP wrote the paper.
Abbreviations
AMØalveolar macrophage
BALbronchoalveolar lavage
BMDCbone marrow-derived dendritic cell
BMMØbone marrow-derived macrophage
DCdendritic cell
mAbmonoclonal antibody
MIP-2macrophage inflammatory protein-2
MyD88myeloid differential factor 88
TLRtoll-like receptor
TNF-αtumor necrosis factor-α
WTwild-type
==== Refs
References
Janeway CA Jr Medzhitov R 2002 Innate immune recognition Annu Rev Immunol 20 197 216 11861602
Beutler B 2004 Innate immunity: An overview Mol Immunol 40 845 859 14698223
Akira S Takeda K 2004 Toll-like receptor signalling Nat Rev Immunol 4 499 511 15229469
Latge JP 1999
Aspergillus fumigatus and aspergillosis Clin Microbiol Rev 12 310 350 10194462
Schaffner A Douglas H Braude A 1982 Selective protection against conidia by mononuclear and against mycelia by polymorphonuclear phagocytes in resistance to Aspergillus. Observations on these two lines of defense in vivo and in vitro with human and mouse phagocytes J Clin Invest 69 617 631 7037853
Ibrahim-Granet O Philippe B Boleti H Boisvieux-Ulrich E Grenet D 2003 Phagocytosis and intracellular fate of Aspergillus fumigatus conidia in alveolar macrophages Infect Immun 71 891 903 12540571
Philippe B Ibrahim-Granet O Prevost MC Gougerot-Pocidalo MA Sanchez Perez M 2003 Killing of Aspergillus fumigatus by alveolar macrophages is mediated by reactive oxidant intermediates Infect Immun 71 3034 3042 12761080
Osherov N May GS 2001 The molecular mechanisms of conidial germination FEMS Microbiol Lett 199 153 160 11377860
Marr KA Patterson T Denning D 2002 Aspergillosis. Pathogenesis, clinical manifestations, and therapy Infect Dis Clin North Am 16 875 894 12512185
Fontaine T Simenel C Dubreucq G Adam O Delepierre M 2000 Molecular organization of the alkali-insoluble fraction of Aspergillus fumigatus cell wall J Biol Chem 275 27594 27607 10869365
Paris S Debeaupuis JP Crameri R Carey M Charles F 2003 Conidial hydrophobins of Aspergillus fumigatus
Appl Environ Microbiol 69 1581 1588 12620846
Tronchin G Bouchara JP Ferron M Larcher G Chabasse D 1995 Cell surface properties of Aspergillus fumigatus conidia: Correlation between adherence, agglutination, and rearrangements of the cell wall Can J Microbiol 41 714 721 7553454
Bernard M Latge JP 2001
Aspergillus fumigatus cell wall: Composition and biosynthesis Med Mycol 39 9 17
Taylor PR Martinez-Pomares L Stacey M Lin HH Brown GD 2005 Macrophage receptors and immune recognition Annu Rev Immunol 23 901 944 15771589
Romani L 2004 Immunity to fungal infections Nat Rev Immunol 4 11 24
Wang JE Warris A Ellingsen EA Jorgensen PF Flo TH 2001 Involvement of CD14 and toll-like receptors in activation of human monocytes by Aspergillus fumigatus hyphae Infect Immun 69 2402 2406 11254600
Mambula SS Sau K Henneke P Golenbock DT Levitz SM 2002 Toll-like receptor (TLR) signaling in response to Aspergillus fumigatus
J Biol Chem 277 39320 39326 12171914
Meier A Kirschning CJ Nikolaus T Wagner H Heesemann J 2003 Toll-like receptor (TLR) 2 and TLR4 are essential for Aspergillus-induced activation of murine macrophages Cell Microbiol 5 561 570 12864815
Netea MG Warris A Van der Meer JW Fenton MJ Verver-Janssen TJ 2003
Aspergillus fumigatus evades immune recognition during germination through loss of toll-like receptor-4-mediated signal transduction J Infect Dis 188 320 326 12854089
Marr KA Balajee SA Hawn TR Ozinsky A Pham U 2003 Differential role of MyD88 in macrophage-mediated responses to opportunistic fungal pathogens Infect Immun 71 5280 5286 12933875
Bellocchio S Montagnoli C Bozza S Gaziano R Rossi G 2004 The contribution of the Toll-like/IL-1 receptor superfamily to innate and adaptive immunity to fungal pathogens in vivo J Immunol 172 3059 3069 14978111
Ariizumi K, Shen GL, Shikano S, Xu S, Ritter R III, et al. 2000 Identification of a novel, dendritic cell-associated molecule, dectin-1, by subtractive cDNA cloning J Biol Chem 275 20157 20167 10779524
Brown GD Gordon S 2001 Immune recognition. A new receptor for beta-glucans Nature 413 36 37
Brown GD Taylor PR Reid DM Willment JA Williams DL 2002 Dectin-1 is a major beta-glucan receptor on macrophages J Exp Med 196 407 412 12163569
Gantner BN Simmons RM Canavera SJ Akira S Underhill DM 2003 Collaborative induction of inflammatory responses by dectin-1 and Toll-like receptor 2 J Exp Med 197 1107 1117 12719479
Gantner BN Simmons RM Underhill DM 2005 Dectin-1 mediates macrophage recognition of Candida albicans yeast but not filaments Embo J 24 1277 1286 15729357
Steele C Marrero L Swain S Harmsen AG Zheng M 2003 Alveolar macrophage-mediated killing of Pneumocystis carinii f. sp. muris involves molecular recognition by the Dectin-1 beta-glucan receptor J Exp Med 198 1677 1688 14657220
Viriyakosol S Fierer J Brown GD Kirkland TN 2005 Innate immunity to the pathogenic fungus Coccidioides posadasii is dependent on Toll-like receptor 2 and Dectin-1 Infect Immun 73 1553 1560 15731053
Brown GD Herre J Williams DL Willment JA Marshall AS 2003 Dectin-1 mediates the biological effects of beta-glucans J Exp Med 197 1119 1124 12719478
Herre J Marshall AS Caron E Edwards AD Williams DL 2004 Dectin-1 uses novel mechanisms for yeast phagocytosis in macrophages Blood 104 4038 4045 15304394
Rogers NC Slack EC Edwards AD Nolte MA Schulz O 2005 Syk-dependent cytokine induction by Dectin-1 reveals a novel pattern recognition pathway for C type lectins Immunity 22 507 517 15845454
Mehrad B Strieter RM Moore TA Tsai WC Lira SA 1999 CXC chemokine receptor-2 ligands are necessary components of neutrophil-mediated host defense in invasive pulmonary aspergillosis J Immunol 163 6086 6094 10570298
Mehrad B Strieter RM Standiford TJ 1999 Role of TNF-alpha in pulmonary host defense in murine invasive aspergillosis J Immunol 162 1633 1640 9973423
Brieland JK Jackson C Menzel F Loebenberg D Cacciapuoti A 2001 Cytokine networking in lungs of immunocompetent mice in response to inhaled Aspergillus fumigatus
Infect Immun 69 1554 1560 11179326
Bozza S Gaziano R Spreca A Bacci A Montagnoli C 2002 Dendritic cells transport conidia and hyphae of Aspergillus fumigatus from the airways to the draining lymph nodes and initiate disparate Th responses to the fungus J Immunol 168 1362 1371 11801677
Serrano-Gomez D Dominguez-Soto A Ancochea J Jimenez-Heffernan JA Leal JA 2004 Dendritic cell-specific intercellular adhesion molecule 3-grabbing nonintegrin mediates binding and internalization of Aspergillus fumigatus conidia by dendritic cells and macrophages J Immunol 173 5635 5643 15494514
Van Epps HL Feldmesser M Pamer EG 2003 Voriconazole inhibits fungal growth without impairing antigen presentation or T-cell activation Antimicrob Agents Chemother 47 1818 1823 12760853
Sato M Sano H Iwaki D Kudo K Konishi M 2003 Direct binding of Toll-like receptor 2 to zymosan, and zymosan-induced NF-kappa B activation and TNF-alpha secretion are down-regulated by lung collectin surfactant protein A J Immunol 171 417 425 12817025
Schaffner A Douglas H Braude AI Davis CE 1983 Killing of Aspergillus spores depends on the anatomical source of the macrophage Infect Immun 42 1109 1115 6642661
Kan VL Bennett JE 1988 Lectin-like attachment sites on murine pulmonary alveolar macrophages bind Aspergillus fumigatus conidia J Infect Dis 158 407 414 3042877
Kan VL Bennett JE 1991 Beta 1,4-oligoglucosides inhibit the binding of Aspergillus fumigatus conidia to human monocytes J Infect Dis 163 1154 1156 2019764
Garlanda C Hirsch E Bozza S Salustri A De Acetis M 2002 Non-redundant role of the long pentraxin PTX3 in anti-fungal innate immune response Nature 420 182 186 12432394
Persat F Noirey N Diana J Gariazzo MJ Schmitt D 2003 Binding of live conidia of Aspergillus fumigatus activates in vitro-generated human Langerhans cells via a lectin of galactomannan specificity Clin Exp Immunol 133 370 377 12930363
Borges-Walmsley MI Chen D Shu X Walmsley AR 2002 The pathobiology of Paracoccidioides brasiliensis
Trends Microbiol 10 80 87 11827809
|
16304610
|
PMC1287910
|
CC BY
|
2021-01-05 12:11:28
|
no
|
PLoS Pathog. 2005 Nov 18; 1(3):e30
|
utf-8
|
PLoS Pathog
| 2,005 |
10.1371/journal.ppat.0010030
|
oa_comm
|
==== Front
PLoS PathogPLoS PathogppatplpaplospathPLoS Pathogens1553-73661553-7374Public Library of Science San Francisco, USA 1630461110.1371/journal.ppat.001003205-PLPA-RA-0071R2plpa-01-03-07Research ArticleGastroenterology - HepatologyInfectious DiseasesMicrobiologyGenetics/Gene ExpressionEubacteriaMus (Mouse)None
Salmonella Pathogenicity Island 2 Is Expressed Prior to Penetrating the Intestine SPI-2 Expression In VivoBrown Nat F 1Vallance Bruce A 2Coombes Brian K 1Valdez Yanet 1Coburn Bryan A 1Finlay B. Brett 1*
1 Michael Smith Laboratories, University of British Columbia, Vancouver, British Columbia, Canada
2 Division of Gastroenterology, BC Children's Hospital, Vancouver, British Columbia, Canada
Isberg Ralph EditorTufts University School of Medicine, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 18 11 2005 1 3 e3215 6 2005 17 10 2005 Copyright: © 2005 Brown 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.
Salmonella enterica serovar Typhimurium is a facultative intracellular pathogen that causes disease in mice that resembles human typhoid. Typhoid pathogenesis consists of distinct phases in the intestine and a subsequent systemic phase in which bacteria replicate in macrophages of the liver and spleen. The type III secretion system encoded by Salmonella pathogenicity island 2 (SPI-2) is a major virulence factor contributing to the systemic phase of typhoid pathogenesis. Understanding how pathogens regulate virulence mechanisms in response to the environment, including different host tissues, is key to our understanding of pathogenesis. A recombinase-based in vivo expression technology system was developed to assess SPI-2 expression during murine typhoid. SPI-2 expression was detectable at very early times in bacteria that were resident in the lumen of the ileum and was independent of active bacterial invasion of the epithelium. We also provide direct evidence for the regulation of SPI-2 by the Salmonella transcription factors ompR and ssrB in vivo. Together these results demonstrate that SPI-2 expression precedes penetration of the intestinal epithelium. This induction of expression precedes any documented SPI-2-dependent phases of typhoid and may be involved in preparing Salmonella to successfully resist the antimicrobial environment encountered within macrophages.
Synopsis
Typhoid fever is a disease caused by specific Salmonella strains and is a significant cause of mortality in many regions of the developing world. Following a person's ingestion of Salmonella, the bacteria initially colonize the intestine, which they subsequently breach to reside in immune cells of the liver and spleen. The ability to survive inside immune cells directly contributes to the ability of Salmonella to cause typhoid, and is conferred upon Salmonella by the so-called Salmonella pathogenicity island 2 (SPI-2) type III secretion system. Previous work has shown that while SPI-2 is specifically turned on inside host cells, it is not active when grown in typical laboratory medium. Owing to these facts, it has been hypothesized that Salmonella specifically turn on SPI-2 inside host cells after breaching the host intestine. The researchers developed a sensitive system in Salmonella to test this hypothesis using a mouse model of typhoid. Interestingly, SPI-2 was specifically turned on before Salmonella breached the intestine, suggesting that SPI-2, which is integral to virulence, is active in a preemptive fashion to allow Salmonella to survive within the immune system.
Citation:Brown NF, Vallance BA, Coombes BK, Valdez Y, Coburn BA, et al. (2005) Salmonella pathogenicity island 2 is expressed prior to penetrating the intestine. PLoS Pathog 1(3): e32.
==== Body
Introduction
Salmonella is a Gram-negative bacterial pathogen that causes substantial morbidity and mortality worldwide. Human-adapted serovars cause typhoid, a systemic and life-threatening infection, while non-human-adapted serovars commonly cause enteritis. Following ingestion of contaminated food or water, the pathogenesis of both typhoid and Salmonella enteritis begins with an intestinal phase, while only typhoid progresses to a systemic phase. The intestinal phase of typhoid involves colonization of the intestine and penetration of the intestinal epithelium through two separate mechanisms. The first involves active bacterial invasion [1], and the second involves passive uptake of Salmonella during dendritic cell (DC) sampling of luminal microflora [2,3]. Once Salmonella has penetrated the intestinal epithelium, the systemic phase of typhoid begins by dissemination from the intestine via the lymphatics followed by colonization of macrophages of the liver and spleen [4,5]. The niche occupied by Salmonella within these cells is a membrane-bound compartment termed the Salmonella-containing vacuole. Much of our understanding of typhoid pathogenesis has come from mice infected with S. enterica serovar Typhimurium, which models human typhoid in several respects.
The current understanding of typhoid pathogenesis suggests that distinct virulence systems operate during the intestinal and systemic phases of infection and that these virulence systems display little overlap in their spatiotemporal activation. These virulence factors are type III secretion systems (T3SSs) that translocate numerous Salmonella virulence proteins (termed effectors) directly into the host cell, where they alter various host cell processes [6]. The T3SS encoded by Salmonella pathogenicity island 1 (SPI-1) allows Salmonella to invade non-phagocytic cells and penetrate the intestinal epithelium, and is the major factor involved during the intestinal phase of typhoid [7–9]. SPI-1 mutant Salmonella Typhimurium is fully virulent when inoculated intraperitoneally into mice, indicating that the role played by SPI-1 is limited to the intestinal phase of Salmonella infection [9]. In contrast, Salmonella pathogenicity island 2 (SPI-2) mediates Salmonella replication within macrophages at systemic sites, and SPI-2 mutant Salmonella is avirulent when inoculated intraperitoneally [10–12]. Studies of Salmonella-induced enteritis have found that the role of SPI-2 in the intestine is subtle when compared to the role of SPI-1 [13–16]. However, it should be noted that Salmonella-induced enteritis and typhoid are distinct diseases involving different intestinal pathologies, and as such the pathogenesis of the diseases is not directly comparable. In particular, typhoid does not typically involve significant intestinal inflammation [16]. This has led to models in which Salmonella employs a two-tiered expression of virulence systems that correspond to the biphasic pathogenesis of typhoid, with SPI-1 mediating intestinal pathogenesis and SPI-2 mediating systemic pathogenesis.
While it is clear that SPI-2 is a major virulence factor leading to mortality in murine typhoid, our current understanding of its molecular function is limited. Its primary role appears to be subversion of host cell membrane traffic, allowing replication of intracellular Salmonella [17]. Consistent with its role during intracellular growth, in vitro studies have shown that expression of SPI-2 is induced inside host cells and in culture medium that mimics the environment of the Salmonella-containing vacuole [18–20]. In vitro, SPI-2 expression is regulated by the SPI-2-encoded regulatory system SsrA/B [18,19], and efficient transcription of ssrAB requires the regulator OmpR [21]. Recent work by Hensel and colleagues has shown that the SPI-2 T3SS is typically expressed as one apparatus per bacterium, during growth in host cells or during inducing in vitro culture [22].
Countless studies have shown that bacteria regulate their expression profile by sensing the extracellular environment and responding accordingly. As an environment in which to study bacterial gene expression, the mammalian host poses unique challenges because of its complexity and dynamic nature. Consequently, analysis of the spatiotemporal expression pattern of virulence genes in vivo has rarely been undertaken. Of particular interest to the field of Salmonella research is the expression pattern of SPI-2 during typhoid because it is the central virulence mechanism and is considered to play a role specific to a particular phase of pathogenesis. Based on this, we hypothesized that SPI-2 expression would be confined to the peripheral lymphoid tissues, spleen, and liver. To address this, we examined the spatiotemporal expression pattern of three SPI-2 promoters during experimental murine typhoid. In contrast to the current model of SPI-2-mediated pathogenesis that argues for exclusive intracellular expression, we found that expression of SPI-2 occurs during initial stages of pathogenesis in the lumen of the intestine.
Results
SPI-2 Gene Expression In Vitro Assessed Using RIVET
Recombinase-based in vivo expression technology (RIVET) is an exquisitely sensitive reporter of gene expression. This system involves the construction of a transcriptional fusion to a site-specific recombinase, which mediates the loss of a selectable genetic marker (a process called resolution) [23]. This approach has been applied to elucidate the spatiotemporal expression patterns of the toxin-coregulated pilus and cholera toxin during Vibrio cholerae infection of mice [24]. We designed and constructed a Salmonella Typhimurium strain possessing all the genetic requirements for RIVET and lacZ fusion analysis of PsseA (see Materials and Methods). The ability of this strain to function as a RIVET reporter of PsseA activity was initially assessed during in vitro growth in media that have previously been shown to induce (LPM medium) or not induce (Luria-Bertani [LB] medium) expression from SPI-2 promoters (Figure 1). PsseA activity was assessed using RIVET and β-galactosidase activity as well as by monitoring cytoplasmic levels of SseB, a protein expressed from PsseA. A low level of PsseA activity was detected for each output in the non-inducing LB medium whereas high levels were detected from bacteria cultured in the inducing LPM medium, indicating that RIVET correlates well with standard methods. Additionally, RIVET was considerably more sensitive at detecting PsseA activity than lacZ transcriptional fusion and immunoblotting for native SseB levels. This was most obvious when the number of bacteria in the culture was below the sensitivity of conventional reporter systems. Furthermore, the high degree of sensitivity of RIVET was deemed important for the study of the monocopy expression level of the SPI-2 T3SS. Using RIVET, we also confirmed the role of SsrB and OmpR in regulating SPI-2 gene expression (Figure 1D).
Figure 1 Expression from PsseA during Growth of Salmonella in Inducing and Non-Inducing Culture Media
(A–C) The wild-type Salmonella RIVET strain (NB25) was inoculated into LPM (inducing) and LB (non-inducing) media at 1/100 of the culture volume, and cultures were incubated with shaking at 37 °C. At the indicated time points, samples were taken for measurement of culture OD600 (unpublished data), colony-forming units per millilitre (unpublished data), β-galactosidase activity, percent resolution, and native SseB levels. The results shown for percent resolution (A) and β-galactosidase activity (B) are the mean ± standard error of the mean from three independent experiments. The levels of SseB associated with bacteria were detected by immunoblotting (C). Protein loading was normalized to culture OD600 and was confirmed by immunoblotting for the abundantly expressed cytoplasmic protein DnaK.
(D) Wild-type (NB25), ssrB
− (NB7), and ompR
− (NB15) were grown in LPM medium and percent resolution was determined from the cultures at 10 h.
Salmonella infection of cultured mammalian cells is a model that is frequently used to approximate conditions encountered during infection in vivo. We determined the expression from PsseA at 1 h following uptake into various mammalian cells using the RIVET system (Figure 2). In HeLa cells (epithelial), where bacteria actively induce their own uptake using the SPI-1 T3SS, substantial induction of expression from PsseA occurred after 1 h in an SsrB- and an OmpR-dependent fashion. This demonstrates that SPI-2 is activated rapidly following SPI-1-mediated Salmonella invasion of an epithelial cell. The role of SPI-2 in mediating replication within macrophages has been thoroughly documented [17], and DCs are also a potentially important cell type encountered by Salmonella in vivo [2,3]. We therefore determined the kinetics of PsseA induction in RAW264.7 murine macrophages and murine DCs. A significant induction of SsrB- and OmpR-dependent PsseA expression could be detected within 1 h of Salmonella being added to either cell type (Figure 2).
Figure 2 Expression from PsseA during Infection of Mammalian Cells In Vitro
The human epithelial cell line HeLa, murine macrophage cell line RAW264.7, and bone-marrow-derived DCs were infected with wild-type (NB25), ssrB − (NB7), and ompR − (NB15) RIVET strains as described in Materials and Methods. At 1 h post-infection intracellular bacteria were recovered and the percent resolution was determined.
SPI-2 Gene Expression In Vivo Assessed Using RIVET
Once it was established that the RIVET system was sensitive and specific, the RIVET system was used to determine the spatiotemporal expression pattern of PsseA during infection of mice. To facilitate the synchronized arrival of a bacterial load in the distal ileum, the inoculum was delivered into a single loop constructed in the ileum of each mouse tested (see Materials and Methods). At times ranging from 15 min to 4 h following inoculation, the liver, spleen, mesenteric lymph nodes, and ileal loop were dissected, homogenized, and plated to determine the degree of PsseA induction using RIVET (Figure 3A). These results clearly demonstrated that expression from PsseA was induced within 15 min following inoculation and that this induction occurred within the ileum. RIVET reports gene expression in an irreversible fashion, and data on expression at systemic sites and later time points was not considered reliable as the majority of bacteria had undergone resolution within 15 min. We also assessed ssrB− and ompR− strains for their ability to induce expression from PsseA within the ileum 15 min following inoculation (Figure 3B). These data showed that both regulators play a significant role in the observed induction in the ileum and directly showed the involvement of these regulators in the expression of SPI-2 in vivo.
Figure 3 Kinetics of SPI-2 Expression during Murine Typhoid
(A) Mice were infected with the wild-type PsseA RIVET strain (NB25) as described in Materials and Methods. At the indicated times following inoculation, the ileal loops, mesenteric lymph nodes, spleen, and liver were taken and assessed for expression from PsseA by determining the percent resolution. The results shown are the mean ± standard error of the mean from three independent experiments.
(B) Mice were infected with the wild-type (NB25), ssrB
− (NB7), and ompR
− (NB15) PsseA RIVET strains, and at 15 min following inoculation, the ileal loop was removed and homogenized, and the percent resolution of the infecting Salmonella was determined.
(C) Mice were infected with the wild-type PspiC (NB33) and PssaG (NB31) RIVET strains for 15 and 30 min, the ileum was removed and homogenized, and the percentage resolution determined. The results shown are the mean ± standard error of the mean for three independent experiments.
The experiments described above showed that PsseA, the promoter driving expression of the SPI-2 T3SS translocon and effectors, is active within 15 min of inoculation into ileal loops. To see if the promoters driving expression of the remaining components of the SPI-2 T3SS were active on a similar timescale, we constructed strains for RIVET analysis of transcription from PspiC and PssaG. These strains were confirmed to have the typical SPI-2 expression characteristics during in vitro culture, including being dependent on the SPI-2-encoded transcription factor SsrB (data not shown). When these strains were inoculated into mouse ileal loops, expression after 15 min from both PspiC and PssaG could be observed (Figure 3C), similar to that of previous experiments with PsseA. These results showed that all SPI-2 T3SS components are expressed in a large proportion of Salmonella within 15 min of arriving in the ileum.
The short time frame in which SPI-2 expression was induced in vivo prompted us to investigate the location of the bacteria that had induced SPI-2 expression. Specifically, we wanted to know whether Salmonella was located within host cells when SPI-2 expression was induced. As SPI-1 mediates the major route of epithelial penetration [2,8], we tested expression from PsseA in a SPI-1 mutant (invA
−) in vivo. At 15 min following inoculation, the induction of expression from PsseA was independent of SPI-1 (Figure 4A), strongly suggesting that the SPI-2 expression we were observing was occurring in the lumen of the intestine. To further test this, we directly determined the localization of wild-type Salmonella 15 min following inoculation into ileal loops using confocal microscopy. As expected, the vast majority of bacteria were not located within host cells but rather were associated with the apical surface of the epithelium (Figure 4B). These data are inconsistent with the observed induction of SPI-2 occurring exclusively in an intracellular compartment in vivo.
Figure 4
Salmonella Induces Expression of SPI-2 in the Lumen of the Ileum
(A) Mouse ileal loops were infected with either wild-type or invA
− PsseA RIVET strains for 15 min before the loops were removed and homogenized, and the percent resolution was determined.
(B) Mouse ileal loops were infected with wild-type Salmonella for 15 min before being removed, fixed, stained, and analyzed by confocal microscopy with an oil immersion 40× 1.3 N.A. objective. A representative image is displayed showing host cell nuclei in blue, actin in green, and Salmonella in red.
To test whether Salmonella association with the cytoplasmic membrane of a host cell initiates expression from PsseA, we infected HeLa cells in vitro with noninvasive invA
−
Salmonella. Extracellular, cell-associated invA
−
Salmonella at 1 h post-infection had not induced significant expression from PsseA (3.54% ± 3.01% resolution). In contrast, internalized wild-type Salmonella controls showed a high level of expression (75.10% ± 5.75% resolution). This is consistent with cell culture medium providing a non-inducing environment for SPI-2 expression (data not shown) and shows that the host cell plasmalemma does not induce SPI-2 expression. This suggested that a stimulus for expression from PsseA exists in the luminal contents of the small intestine. We therefore conducted experiments to address this hypothesis. Wild-type, ssrB
−, and ompR
− RIVET strains were incubated for up to 30 min at 37 °C with contents collected from mouse small intestine and then resolution was measured. No resolution was observed for any strain, suggesting that the stimulus inducing expression of SPI-2 in the intestine is not capable of being extracted with naive luminal contents of the small intestine.
Discussion
By using a sensitive methodology and testing earlier time points than have been presented in previous studies of SPI-2 function, we have established that SPI-2 genes are expressed very early in the intestine. The data presented above are inconsistent with the currently held view that the initial induction of SPI-2 expression occurs in response to an intracellular environment. Our attempts to confirm the RIVET expression data using β-galactosidase reporters to measure SPI-2 expression in vivo failed to detect sufficient signal to reliably report SPI-2 expression (data not shown). However, extensive controls were performed in vitro, where sufficient numbers of bacteria can be analyzed to detect β-galactosidase reporters as well as SPI-2 proteins by western blotting. These control experiments confirmed that RIVET was a sensitive and specific reporter of SPI-2 expression. The failure of β-galactosidase to accurately report on SPI-2 expression in vivo is most likely a reflection of low bacterial numbers in the examined tissues. This highlights a major advantage of RIVET in its capacity to report on gene expression from small populations of bacteria.
Previous work has focused on mimicking the intracellular environment encountered by Salmonella to define the cues for inducing expression of SPI-2. Our results show that SPI-2 is induced in the gut lumen, prior to encountering an intracellular environment. We have investigated signals present in the small intestine as potential cues for SPI-2 expression and have determined that intestinal SPI-2 induction is not a result of association with host cell plasmalemma or luminal contents from the small intestine of uninfected mice. Additionally, we have been able to rule out low oxygen tension as a cue for SPI-2 induction in the intestine (data not shown). Our data are compatible with a rapid host response to Salmonella acting as a cue for the induction of SPI-2 expression. One such response would be the secretion of antimicrobial products by paneth cells, specialized epithelial cells that reside at the base of the crypts and respond to antigenic stimuli including Salmonella [25]. However, experiments with mmp7
−/− mice, which are deficient in paneth cell products, have indicated that these products are not a stimulus for SPI-2 expression in vivo (data not shown). Other potential host responses acting as a stimulus for SPI-2 expression will be a topic of future investigation.
Previous studies on SPI-2 and its translocated effectors have almost exclusively focused on the role played during the systemic phase of typhoid, and consequently little is known about the potential actions of SPI-2 during the intestinal phase of typhoid. Others have shown that SPI-2 mutant Salmonella colonizes the caecum and Peyer's patches to a lesser extent than wild-type Salmonella during typhoid [19]. This, together with our data on expression, firmly establishes a role for SPI-2 prior to colonization of systemic sites during typhoid. It is important to consider the localization of intestinal Salmonella in order to speculate on the role SPI-2 could be playing at this location. Although the vast majority of bacteria are present in the luminal space of the intestine, it is well established that Salmonella transits through epithelial cells and can reside within sub-epithelial phagocytic cells. Given that type III secretion mediates the direct delivery of effector proteins into the host cell cytosol [6] and that the system encoded by SPI-2 allows Salmonella to replicate within macrophages [17], this suggests that the location where SPI-2 is actually functioning is within phagocytic cells of the intestine. It is likely that SPI-2 expression initiated in the intestinal lumen primes Salmonella for residency in intestinal phagocytic cells prior to the bacteria reaching this niche. We do not conclude that SPI-2 enables replication of luminal Salmonella.
The rapid induction of SPI-2 in the lumen of the ileum suggests that for Salmonella to initiate a successful infection of sub-epithelial and systemic sites, it must be prepared for the harsh environment of the macrophage endosomal system prior to the encounter. This finding may establish a paradigm for all pathogens where preemptive synthesis of important virulence factors occurs prior to the transition from colonization of a mucosal surface to colonization of systemic sites. We also speculate that functions mediated by SPI-2 during early stages of infection may not be limited to the establishment of a Salmonella-containing vacuole that supports bacterial replication. Such possibilities include enhancing dissemination to systemic sites. By identifying additional SPI-2-dependent functions at these early intestinal stages of typhoid pathogenesis, we hope to elucidate key mechanisms that facilitate the successful parasitic lifestyle of Salmonella.
Materials and Methods
Bacterial strains and plasmids.
Bacterial strains and plasmids are shown in Table 1. All Salmonella strains used in this study were derived from the wild-type strain SL1344. All bacteria were routinely cultured in LB medium. LPM medium (pH 5.8) has been described previously [26]. Kanamycin was used at 50 μg ml−1, chloramphenicol at 10 μg ml−1, and ampicillin at 100 μg ml−1. Salmonella RIVET strains were always grown in LB medium containing ampicillin and chloramphenicol prior to conducting experiments.
Table 1 Bacterial Strains and Plasmids Used in This Study
For RIVET studies, it is desirable to insert the resolvable cassette (res-marker-res) into a neutral site within the genome. ushA is a silent gene in Salmonella Typhimurium because of an inactivating S139Y substitution in the expressed protein product [27], and as such this was considered to be a good site to insert the resolvable cassette for RIVET. A region of approximately 1.8 kb spanning ushA was amplified by PCR using the oligonucleotides ushako-f (5′ GCAACTAGTGCGATGTTGGAGATAGTAGGATGTG 3′) and ushako-r (5′ GATGTCGACCCTCACCATTTGGCGTAAACG 3′) and was cloned into pBLUESCRIPT KS+ using SpeI and SalI to create pUshA. A resolvable cassette mediating resistance to chloramphenicol was constructed by subcloning the res-npt-sacB-res-containing SphI fragment into pUC19 to create pUC-RES. The npt-sacB portion was replaced with cat to give pUC-RESCm. The res-cat-res cassette was then inserted into the middle of ushA in pUshA, and then the ushA::res-cat-res allele was cloned into pCVD442 using SacI and KpnI to give pUshA::RES KO. Using this plasmid, the ushA::res-cat-res allele was introduced into SL1344 and SL1344 ΔompR using standard allelic exchange methodologies [28] to give strains NB24 and NB13, respectively. NB24 was shown to have no detectable defect in virulence when inoculated intraperitoneally into mice by either a time-to-death assay (N. F. B. and B. A. V., unpublished data) or competitive index (B. K. C. and M. Wickham, unpublished data). Salmonella strains NB6 and NB16 were created by P22 generalized transduction of ssrB::Km and invA::Km alleles into NB24, respectively.
Transcriptional fusions were generated as follows. An approximately 1.5-kb region upstream of sseA was amplified by PCR using oligonucleotides pssea-f (5′ ATACTCGAGCGTATTCTTCATTTTCATCGGTG 3′) and pssea-r (5′ ATACAATTGCCCTTTCAGCAAGCTGTTGAC 3′) and cloned into pIVET5nMut135 using XhoI and MfeI. Other reporter strains were generated using the same strategy using oligonucleotides pssagfu-f (5′ ATGCTCGAGAATTACCTCCGTTAGCCTGAC 3′) and pssagfu-r (5′ ATGCAATTGCTGCCTGGTGCGCCATGTG 3′) for PssaG and oligonucleotides pssab-f (5′ ATTCTCGAGTCGGCGTGCAATTTGAAGG 3′) and pssab-r (5′ ATACAATTGCCAGCATGAATCCCTCCTCAGAC 3′) for PspiC. The resulting reporter plasmids were conjugated into the desired Salmonella strains from Escherichia coli SM10 λpir. The resulting strains had the reporter plasmid integrated into the chromosome by homologous recombination, resulting in a merodiploid genotype for the promoter region being studied.
Gene expression reporter assays.
Resolution assays were performed by plating serial dilutions of sample material onto LB ampicillin plates and incubating these overnight at 37 °C. The following day, plates with between 50 and 200 colonies were replica-plated onto LB ampicillin plates and LB ampicillin chloramphenicol plates, which were incubated overnight at 37 °C. Colonies that grew on the ampicillin only plates, but not on the ampicillin chloramphenicol plates were considered to have undergone the resolution event. β-galactosidase assays were performed as previously described [29] using Galacto-Star chemiluminescent substrate (Applied Biosystems, Foster City, California, United States) and were read using a Spectrafluor Plus (Tecan, Mannedorf, Switzerland) in luminescence mode.
Antibodies and reagents.
Antibodies to SseB have been described previously [30] and were used at a concentration of 1:1,000, and a monoclonal antibody to DnaK (Stressgen Biotechnologies, Victoria, British Columbia, Canada) was used at a concentration of 1:2,000. Anti-Salmonella Typhimurium LPS antiserum (Difco, Becton-Dickinson, Franklin Lakes, New Jersey, United States) was used at a concentration of 1:200. HRP-labeled anti-mouse and anti-rabbit antibodies were used at a concentration of 1:5,000 and were purchased from Jackson Immunoresearch Laboratories (West Grove, Pennsylvania, United States). Alexafluor 568–labeled anti-rabbit antibodies (Molecular Probes, Eugene, Oregon, United States) were used at a concentration of 1:200. Alexafluor 488–labeled phalloidin and DAPI were purchased from Molecular Probes.
Cell culture and infection of cultured cells.
HeLa and RAW264.7 cells were cultured using DMEM containing 10% FCS. Bone-marrow-derived DCs were derived as previously described [31]. Briefly, 2.5 × 106 bone marrow cells were cultured in 10 ml of Iscove's Modified Dulbeco's Medium supplemented with 10% FCS, GM-CSF (20 ng/ml), and IL-4 (10 ng/ml). The cells were harvested after 6 d of culture, and DCs were purified using anti-CD11c conjugated MACS microbeads and magnetic separation columns (Miltenyi Biotec, Bergisch Gladbach, Germany). The purity of DCs was assessed by FACS analysis and ranged from 70% to 90% CD11c+. Cells were incubated in an atmosphere containing 5% CO2 at 37 °C.
HeLa cells were infected with invasive Salmonella prepared according to previous studies [32]. RAW264.7 and DCs were infected with stationary phase Salmonella opsonized in 20% normal human serum for 30 min. An incubation of 10 min at 37 °C in 5% CO2 was performed once bacteria were applied to cell cultures to allow for internalization. Following this, the infecting medium was aspirated, cells were washed four times with PBS, fresh cell culture medium containing gentamicin (100 μg ml−1) was added, and incubation at 37 °C in 5% CO2 was continued. At 1 h post-infection, cell culture medium was removed and the cells were again washed four times with PBS. The cells were then lysed in a solution of Triton X-100 (1% v/v) and sodium dodecyl sulfate (0.1% w/v) to release intracellular bacteria. Samples were diluted and assayed for resolution.
Mouse infections.
Female C57BL/6 mice (6–10 wk of age) were purchased from Jackson Laboratory (Bar Harbor, Maine, United States), and housed in the animal facility at the University of British Columbia in direct accordance with guidelines drafted by the University of British Columbia's Animal Care Committee and the Canadian Council on the Use of Laboratory Animals. For ileal loop experiments, bacterial inocula of approximately 107 colony-forming units were prepared in 100 μl, and the resolution status of the strain was confirmed directly before inoculation. Ileal loop experiments were modified from those previously described [1]. In brief, mice were anaesthetized by intraperitoneal injection of ketamine and xylazine. Following a midline abdominal incision, the small bowel was exposed and the bowel was ligated twice, close to the cecum, to create a loop approximately 4 cm in length into which the inoculum was injected. The bowel was then returned to the abdominal cavity and the incision closed with discontinuous sutures. At given time points, the mice were euthanized and tissues collected for bacterial enumeration and RIVET. The intestinal lumen was rinsed with PBS to remove non-adherent bacteria. Tissues were homogenized in PBS using a Polytron homogenizer (Kinematica, Lucerne, Switzerland).
Immunohistochemistry.
Tissues were fixed for 3 h in 3% paraformaldehyde, and washed three times with PBS prior to cryoembedding and sectioning. Sections were fixed in acetone at −20 °C for 10 min and then air dried at room temperature. Tissue sections were outlined with a wax pen and blocked 30 min in 10% goat serum in PBS-BSA at room temperature. Sections were then washed three times in PBS-BSA prior to incubation for 30 min at room temperature with anti-Salmonella LPS antiserum. Sections were then washed three times in PBS-BSA prior to incubation with appropriate Alexafluor 568–conjugated anti-rabbit antibodies. Sections were then washed three times in PBS-BSA, incubated with Alexafluor 488–labeled phalloidin, washed a further three times in PBS-BSA, and then incubated in DAPI. Imaging was performed on a Nikon (Tokyo, Japan) TE2000 inverted microscope equipped with a Bio-Rad (Hercules, California, United States) Radiance 2000 confocal scan head and a two-photon laser source using a plan apochromat 40× 1.3 N.A. objective.
We thank members of the Finlay and Vallance laboratories for advice with experiments and critical reading of this manuscript. We also thank Andrew Camilli for providing reagents and advice on the use of RIVET and Ifor Beacham for advice on silent genes in Salmonella. NFB is a Michael Smith Foundation for Health Research (MSFHR) fellow, and BKC is a MSFHR and Canadian Institutes of Health Research (CIHR) fellow. BAC is supported by CIHR and MSFHR studentships. BAV is the CHILD Foundation Research Scholar, a MSFHR Scholar, and the Canada Research Chair in Pediatric Gastroenterology. BBF is a Howard Hughes Medical Institute (HHMI) International Research Scholar and the University of British Columbia Peter Wall Distinguished Professor. This work was supported by grants from CIHR and HHMI to BBF and from CIHR to BAV.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. NFB, BAV, BKC, YV, and BAC conceived and designed the experiments, performed the experiments, and analyzed the data. NFB contributed reagents/materials/analysis tools. NFB, BAV, BKC, YV, BAC, and BBF wrote the paper.
Abbreviations
DCdendritic cell
LBLuria-Bertani
RIVETrecombinase-based in vivo expression technology
SPI-1
Salmonella pathogenicity island 1
SPI-2
Salmonella pathogenicity island 2
T3SStype III secretion system
==== Refs
References
Jones BD Ghori N Falkow S 1994
Salmonella typhimurium initiates murine infection by penetrating and destroying the specialized epithelial M cells of the Peyer's patches J Exp Med 180 15 23 8006579
Vazquez-Torres A Jones-Carson J Baumler AJ Falkow S Valdivia R 1999 Extraintestinal dissemination of Salmonella by CD18-expressing phagocytes Nature 401 804 808 10548107
Rescigno M Urbano M Valzasina B Francolini M Rotta G 2001 Dendritic cells express tight junction proteins and penetrate gut epithelial monolayers to sample bacteria Nat Immunol 2 361 367 11276208
Richter-Dahlfors A Buchan AM Finlay BB 1997 Murine salmonellosis studied by confocal microscopy: Salmonella typhimurium resides intracellularly inside macrophages and exerts a cytotoxic effect on phagocytes in vivo J Exp Med 186 569 580 9254655
Salcedo SP Noursadeghi M Cohen J Holden DW 2001 Intracellular replication of Salmonella typhimurium strains in specific subsets of splenic macrophages in vivo Cell Microbiol 3 587 597 11553011
Hueck CJ 1998 Type III protein secretion systems in bacterial pathogens of animals and plants Microbiol Mol Biol Rev 62 379 433 9618447
Wallis TS Galyov EE 2000 Molecular basis of Salmonella -induced enteritis Mol Microbiol 36 997 1005 10844685
Penheiter KL Mathur N Giles D Fahlen T Jones BD 1997 Non-invasive Salmonella typhimurium mutants are avirulent because of an inability to enter and destroy M cells of ileal Peyer's patches Mol Microbiol 24 697 709 9194698
Galan JE Curtiss R III 1989 Cloning and molecular characterization of genes whose products allow Salmonella typhimurium to penetrate tissue culture cells Proc Natl Acad Sci U S A 86 6383 6387 2548211
Hensel M Shea JE Gleeson C Jones MD Dalton E 1995 Simultaneous identification of bacterial virulence genes by negative selection Science 269 400 403 7618105
Shea JE Hensel M Gleeson C Holden DW 1996 Identification of a virulence locus encoding a second type III secretion system in Salmonella typhimurium
Proc Natl Acad Sci U S A 93 2593 2597 8637919
Ochman H Soncini FC Solomon F Groisman EA 1996 Identification of a pathogenicity island required for Salmonella survival in host cells Proc Natl Acad Sci U S A 93 7800 7804 8755556
Bispham J Tripathi BN Watson PR Wallis TS 2001
Salmonella pathogenicity island 2 influences both systemic salmonellosis and Salmonella -induced enteritis in calves Infect Immun 69 367 377 11119526
Tsolis RM Adams LG Ficht TA Baumler AJ 1999 Contribution of Salmonella typhimurium virulence factors to diarrheal disease in calves Infect Immun 67 4879 4885 10456944
Hapfelmeier S Stecher B Barthel M Kremer M Muller AJ 2005 The Salmonella pathogenicity island (SPI)-2 and SPI-1 type III secretion systems allow Salmonella serovar typhimurium to trigger colitis via MyD88-dependent and MyD88-independent mechanisms J Immunol 174 1675 1685 15661931
Barthel M Hapfelmeier S Quintanilla-Martinez L Kremer M Rohde M 2003 Pretreatment of mice with streptomycin provides a Salmonella enterica serovar Typhimurium colitis model that allows analysis of both pathogen and host Infect Immun 71 2839 2858 12704158
Waterman SR Holden DW 2003 Functions and effectors of the Salmonella pathogenicity island 2 type III secretion system Cell Microbiol 5 501 511 12864810
Valdivia RH Falkow S 1997 Fluorescence-based isolation of bacterial genes expressed within host cells Science 277 2007 2011 9302299
Cirillo DM Valdivia RH Monack DM Falkow S 1998 Macrophage-dependent induction of the Salmonella pathogenicity island 2 type III secretion system and its role in intracellular survival Mol Microbiol 30 175 188 9786194
Deiwick J Nikolaus T Erdogan S Hensel M 1999 Environmental regulation of Salmonella pathogenicity island 2 gene expression Mol Microbiol 31 1759 1773 10209748
Lee AK Detweiler CS Falkow S 2000 OmpR regulates the two-component system SsrA-SsrB in Salmonella pathogenicity island 2 J Bacteriol 182 771 781 10633113
Chakravortty D Rohde M Jager L Deiwick J Hensel M 2005 Formation of a novel surface structure encoded by Salmonella Pathogenicity Island 2 EMBO J 24 2043 2052 15889142
Camilli A Beattie DT Mekalanos JJ 1994 Use of genetic recombination as a reporter of gene expression Proc Natl Acad Sci U S A 91 2634 2638 8146167
Lee SH Hava DL Waldor MK Camilli A 1999 Regulation and temporal expression patterns of Vibrio cholerae virulence genes during infection Cell 99 625 634 10612398
Ayabe T Satchell DP Wilson CL Parks WC Selsted ME 2000 Secretion of microbicidal alpha-defensins by intestinal Paneth cells in response to bacteria Nat Immunol 1 113 118 11248802
Coombes BK Brown NF Valdez Y Brumell JH Finlay BB 2004 Expression and secretion of Salmonella pathogenicity island-2 virulence genes in response to acidification exhibit differential requirements of a functional type III secretion apparatus and SsaL J Biol Chem 279 49804 49815 15383528
Innes D Beacham IR Beven CA Douglas M Laird MW 2001 The cryptic ushA gene (ushA(c)) in natural isolates of Salmonella enterica (serotype Typhimurium) has been inactivated by a single missense mutation Microbiology 147 1887 1896 11429465
Donnenberg MS Kaper JB 1991 Construction of an eae deletion mutant of enteropathogenic Escherichia coli by using a positive-selection suicide vector Infect Immun 59 4310 4317 1937792
Kehres DG Zaharik ML Finlay BB Maguire ME 2000 The NRAMP proteins of Salmonella typhimurium and Escherichia coli are selective manganese transporters involved in the response to reactive oxygen Mol Microbiol 36 1085 1100 10844693
Knodler LA Celli J Hardt WD Vallance BA Yip C 2002
Salmonella effectors within a single pathogenicity island are differentially expressed and translocated by separate type III secretion systems Mol Microbiol 43 1089 1103 11918798
Lutz MB Kukutsch N Ogilvie AL Rossner S Koch F 1999 An advanced culture method for generating large quantities of highly pure dendritic cells from mouse bone marrow J Immunol Methods 223 77 92 10037236
Steele-Mortimer O Meresse S Gorvel JP Toh BH Finlay BB 1999 Biogenesis of Salmonella typhimurium– containing vacuoles in epithelial cells involves interactions with the early endocytic pathway Cell Microbiol 1 33 49 11207539
Wray C Sojka WJ 1978 Experimental Salmonella typhimurium infection in calves Res Vet Sci 25 139 143 364573
Osorio CG Crawford JA Michalski J Martinez-Wilson H Kaper JB 2005 Second-generation recombination-based in vivo expression technology for large-scale screening for Vibrio cholerae genes induced during infection of the mouse small intestine Infect Immun 73 972 980 15664940
Dennis JJ Zylstra GJ 1998 Improved antibiotic-resistance cassettes through restriction site elimination using Pfu DNA polymerase PCR Biotechniques 25 772 774 776 9821576
Yanisch-Perron C Vieira J Messing J 1985 Improved M13 phage cloning vectors and host strains: Nucleotide sequences of the M13mp18 and pUC19 vectors Gene 33 103 119 2985470
|
16304611
|
PMC1287911
|
CC BY
|
2021-01-05 12:11:28
|
no
|
PLoS Pathog. 2005 Nov 18; 1(3):e32
|
utf-8
|
PLoS Pathog
| 2,005 |
10.1371/journal.ppat.0010032
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1631162410.1371/journal.pgen.001006505-PLGE-RA-0188R2plge-01-05-10Research ArticleBioinformatics - Computational BiologyEvolutionMicrobiologySystems BiologyEubacteriaLife in Hot Carbon Monoxide: The Complete Genome Sequence of Carboxydothermus hydrogenoformans Z-2901 Hydrogenogen GenomicsWu Martin 1Ren Qinghu 1Durkin A. Scott 1Daugherty Sean C 1Brinkac Lauren M 1Dodson Robert J 1Madupu Ramana 1Sullivan Steven A 1Kolonay James F 1Nelson William C 1Tallon Luke J 1Jones Kristine M 1Ulrich Luke E 2Gonzalez Juan M 3Zhulin Igor B 2Robb Frank T 3Eisen Jonathan A 14*1 The Institute for Genomic Research, Rockville, Maryland, United States of America
2 Center for Bioinformatics and Computational Biology, School of Biology, Georgia Institute of Technology, Atlanta, Georgia, United States of America
3 Center of Marine Biotechnology, University of Maryland Biotechnology Institute, Baltimore, Maryland, United States of America
4 Johns Hopkins University, Baltimore, Maryland, United States of America
Richardson Paul M EditorJoint Genome Institute, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 25 11 2005 19 10 2005 1 5 e658 8 2005 19 10 2005 Copyright: © 2005 Wu 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 report here the sequencing and analysis of the genome of the thermophilic bacterium Carboxydothermus hydrogenoformans Z-2901. This species is a model for studies of hydrogenogens, which are diverse bacteria and archaea that grow anaerobically utilizing carbon monoxide (CO) as their sole carbon source and water as an electron acceptor, producing carbon dioxide and hydrogen as waste products. Organisms that make use of CO do so through carbon monoxide dehydrogenase complexes. Remarkably, analysis of the genome of C. hydrogenoformans reveals the presence of at least five highly differentiated anaerobic carbon monoxide dehydrogenase complexes, which may in part explain how this species is able to grow so much more rapidly on CO than many other species. Analysis of the genome also has provided many general insights into the metabolism of this organism which should make it easier to use it as a source of biologically produced hydrogen gas. One surprising finding is the presence of many genes previously found only in sporulating species in the Firmicutes Phylum. Although this species is also a Firmicutes, it was not known to sporulate previously. Here we show that it does sporulate and because it is missing many of the genes involved in sporulation in other species, this organism may serve as a “minimal” model for sporulation studies. In addition, using phylogenetic profile analysis, we have identified many uncharacterized gene families found in all known sporulating Firmicutes, but not in any non-sporulating bacteria, including a sigma factor not known to be involved in sporulation previously.
Synopsis
Carboxydothermus hydrogenoformans, a bacterium isolated from a Russian hotspring, is studied for three major reasons: it grows at very high temperature, it lives almost entirely on a diet of carbon monoxide (CO), and it converts water to hydrogen gas as part of its metabolism. Understanding this organism's unique biology gets a boost from the decoding of its genome, reported in this issue of PLoS Genetics. For example, genome analysis reveals that it encodes five different forms of the protein machine carbon monoxide dehydrogenase (CODH). Most species have no CODH and even species that utilize CO usually have only one or two. The five CODH in C. hydrogenoformans likely allow it to both use CO for diverse cellular processes and out-compete for it when it is limiting. The genome sequence also led the researchers to experimentally document new aspects of this species' biology including the ability to form spores. The researchers then used comparative genomic analysis to identify conserved genes found in all spore-forming species, including Bacillus anthracis, and not in any other species. Finally, the genome sequence and analysis reported here will aid in those trying to develop this and other species into systems to biologically produce hydrogen gas from water.
Citation:Wu M, Ren Q, Durkin AS, Daugherty SC, Brinkac LM, et al. (2005) Life in hot carbon monoxide: The complete genome sequence of Carboxydothermus hydrogenoformans Z-2901. PLoS Genet 1(5): e65.
==== Body
Introduction
Carbon monoxide (CO) is best known as a potent human poison, binding very strongly and almost irreversibly to the iron core of hemoglobin. Despite its deleterious effects on many species, it is also the basis for many food chains, especially in hydrothermal environments such as the deep sea, hot springs, and volcanoes. In these environments, CO is a common potential carbon source, as it is produced both by partial oxidation of organic matter as well as by multiple microbial strains (e.g., methanogens). It is most readily available in areas in which oxygen concentrations are low, since oxidation of CO will convert it to CO2. In hydrothermal environments, CO use as a primary carbon source is dominated by the hydrogenogens, which are anaerobic, thermophilic bacteria or archaea that carry out CO oxidation using water as an electron acceptor [1]. This leads to the production of CO2 and H2. The H2 is frequently lost to the environment and the CO2 is used in carbon fixation pathways for the production of biomass. Hydrogenogens have attracted significant biotechnological interest because of the possibility they could be used in the biological production of hydrogen gas.
Hydrogenogens are found in diverse volcanic environments [2–7]. The phylogenetic types differ somewhat depending on the environments and include representatives of bacteria and archaea. Carboxydothermus hydrogenoformans is a hydrogenogen that was isolated from a hot spring in Kunashir Island, Russia [2]. It is a member of the Firmicutes Phylum (also known as low GC Gram-positives) and grows optimally at 78 °C. This species has been considered an unusual hydrogenogen, in part because unlike most of the other hydrogenogens, it was believed to be strictly dependent on CO for growth. The other species were found to grow poorly unless CO was supplemented with organic substrates. Thus it was selected for genome sequencing as a potential model obligate CO autotroph.
Surprisingly, initial analysis of the unpublished genome sequence data led to the discovery that this species is not an obligate CO autotroph [8]. We report here a detailed analysis of the genome sequence of C. hydrogenoformans strain Z-2901, the type strain of the species, hereafter referred to simply as C. hydrogenoformans.
Results/Discussion
Genome Structure
The C. hydrogenoformans genome is a single circular chromosome of 2,401,892 base pairs (bp) with a G+C content of 42.0% (Figure 1, Table 1). Annotation of the genome reveals 2,646 putative protein coding genes (CDSs), of which 1,512 can be assigned a putative function. The chromosome displays two clear GC skew transitions that likely correspond to the DNA replication origin and terminus (Figure 1). Overall, 3.0 % of the genome is made up of repetitive DNA sequences. Included in this repetitive DNA are two large-clustered, regularly interspaced short palindromic repeats (CRISPR, 3.9 and 5.6 kilobases, respectively). Each cluster contains 59 and 84 partially palindromic repeats of 30 bps, respectively (GTTTCAATCCCAGA[A/T]TGGTTCGATTAAAAC). Most repeats within each cluster are identical but they differ for one nucleotide in the middle between clusters. Repeats at ends of the smaller cluster degenerate to some extent. These types of repeats are widespread in diverse groups of bacteria and archaea [9]. The first one-third of the repeat sequence is generally conserved. Although the precise functions of these repeats are unknown, some evidence suggests they are involved in chromosome partitioning [10,11]. In addition, experiments in the thermophilic archaea Sulfolobus solfataricus have identified a genus-specific protein binding specifically to the repeats present in that species' genome [11].
Figure 1 Genomic Organization of C. hydrogenoformans
From the outside inward the circles show: (1, 2) predicted protein-coding regions on the plus and minus strands (colors were assigned according to the color code of functional classes; (3) prophage (orange) and CRISPR (pink) regions; (4) χ2-square score of tri-nucleotide composition; (5) GC skew (blue indicates a positive value and red a negative value); (6) tRNAs (green); (7) rRNAs (blue) and structural RNAs (red).
Table 1 General Features of the C. hydrogenoformans Genome
One 35-kilobase lambda-like prophage containing 50 CDSs was identified in the genome. It is flanked on one side by a tRNA suggesting this may have served as a site of insertion. Phylogenetic analysis showed this phage is most closely related to phages found in other Firmicutes, particularly the SPP1 phage infecting Bacillus subtilis.
As with other members of the Phylum Firmicutes, the directions of leading strand DNA replication and transcription are highly correlated, with 87% of genes located on the leading strand. This gene distribution bias is also highly correlated with the presence of a Firmicutes-specific DNA polymerase PolC in the genome [12]. In B. subtilis, PolC synthesizes the leading strand, and another distinct DNA polymerase, DnaE, replicates the lagging strand [13]. In other non-Firmicutes bacteria, DnaE replicates both strands. The asymmetric replication forks of Firmicutes were proposed to contribute to the asymmetry of their gene distributions [12]. One copy of PolC and two copies of DnaE have been identified in C. hydrogenoformans genome. At least some of the gene distribution bias can be caused by selection to avoid collision of the RNA and DNA polymerases as well [14,15]. Despite this apparent selection, the lack of significantly conserved gene order across Firmicutes indicates that genome rearrangements still occur at a reasonably high rate.
Phylogeny and Taxonomy
Analysis of the complete genome of C. hydrogenoformans suggests that the taxonomy of this species, as well as some other organisms, needs to be revised. More specifically, phylogenetic analysis based on concatenation of a few dozen markers (Figure 2) reveals a variety of conflicts between the organismal phylogeny and the classification of some of the Firmicutes. For example, C. hydrogenoformans is currently considered to be a member of the Family Peptococcaceae in the Order Clostridiales [16]. Thus it should form a clade with the Clostridium spp. to the exclusion of other taxa for which genomes are available (e.g., Thermoanaerobacter tengcongensis, which is considered to be a member of Thermoanaerobacteriales). The tree, however, indicates that this is not the case and that T. tengcongensis and the Clostridia spp. are more closely related to each other than either is to C. hydrogenoformans. Thus we believe C. hydrogenoformans should be placed in a separate Order from Clostridiales.
Figure 2 Genome Tree of Representatives of Firmicutes
A maximum likelihood tree was built from concatenated protein sequences of 31 universal housekeeping genes and rooted by two outgroup Actinobacteria (high GC Gram-positives) species: Corynebacterium glutamicum and Streptomyces coelicolor. Bootstrap support values (out of 100 runs) for branches of interest are shown beside them. Each species' ability to sporulate and its number of putative orthologs of the 175 known B. subtilis sporulation genes are also shown.
Perhaps more surprisingly, the concatenated genome tree shows C. hydrogenoformans grouping with Symbiobacterium thermophilum. S. thermophilum is a strictly symbiotic thermophile isolated from compost and is currently classified in the Actinobacteria (also known as high GC Gram-positives) based on analysis of its 16s rRNA sequence [17]. The grouping with Firmicutes is supported by the overall level of similarity of its proteome to other species [18]. We therefore believe the rRNA-based classification is incorrect and that S. thermophilum should be transferred to the Firmicutes. Such inaccuracies of the rRNA trees are relatively uncommon and may in this case be due to the mixing of thermophilic and non-thermophilic species into one group. This can cause artifacts when using rRNA genes for phylogenetic reconstruction since the G+C content of rDNA is strongly correlated to optimal growth temperature.
CO Dehydrogenases and Life in CO
Anaerobic species that make use of CO do so using nickel-iron CO dehydrogenase (CODH) complexes [19,20]. These enzymes all appear to catalyze the anaerobic interconversion of CO and CO2. However, they vary greatly in the cellular role of this conversion and in the exact structure of the complex [19]. Analysis of the genome reveals the presence of five genes encoding homologs of CooS, the catalytic subunit of anaerobic CODHs. These five CooS encoding genes are scattered around the genome, and analysis of genome context, gene phylogeny, and experimental studies in this and other CO-utilizing species suggests they are subunits of five distinct CODH complexes, which we refer to as CODH I-V (Figure 3). The CooS homologs are named accordingly. Specific details about each complex and proposed physiological roles are given in the following paragraphs.
Figure 3 Genome Locations of Genes Predicted to Encode Five CODH Complexes
The genome locations of the genes encoding the five CooS homologs (labelled CooS I-V) are shown. Also shown are neighboring genes that are predicted to encode the five distinct CODH complexes (CODH I-V) with each CooS homolog. Possible cellular roles for four of the five CODH complexes are indicated.
Energy Conservation (CODH-I)
A catalytic subunit (CooS-I, CHY1824) and an electron transfer protein (CooF, CHY1825) of CODH are encoded immediately downstream of a hydrogenase gene cluster (cooMKLXUH, CHY1832–27) that is closely related to the one found in Rhodospirillum rubrum [21]. These eight proteins form a tight membrane-bound enzyme complex that converts CO to CO2 and H2 in vitro [1,22]. In R. rubrum, this CODH/hydrogenase complex was proposed to be the site of CO-driven proton respiration where energy is conserved in the form of a proton gradient generated across the cell membrane [21]. Based on the high similarities in protein sequences and their gene organization, this set of genes were suggested to play a similar role in energy conservation in C. hydrogenoformans [1]. Consistent with this, this cooS gene is in the same subfamily as that from R. rubrum (Figure 4).
Figure 4 Phylogenetic Tree of CooS Homologs
The figure shows a maximum-likelihood tree of CooS homologs. The tree indicates the five CooS homologs in C. hydrogenoformans are not the result of recent duplications but instead are from distinct subfamilies. The other CooS homologs included in the tree were obtained from the NCBI nr database and include some from incomplete genome sequences generated by United States Department of Energy Joint Genome Institute (http://www.jgi.doe.gov/).
Carbon Fixation (CODH-III)
Anaerobic bacteria and archaea, such as methanogens and acetogens, can fix CO or CO2 using the acetyl-CoA pathway (also termed the Wood-Ljungdahl pathway), where two molecules of CO2, through a few steps, are condensed into one acetyl-CoA, a key building block for cellular biosynthesis and an important source of ATP [23]. The key enzyme of the final step (a CODH/acetyl-CoA synthase complex) has been purified from C. hydrogenoformans (strain DSM 6008) cultured under limited CO supply and shown to be functional in vitro [24]. Genes encoding this complex and other proteins predicted to be in this pathway are clustered in the genome (CHY1221–7). This cluster is very similar to the acs operon from the acetogen Moorella thermoacetica which encodes the acetyl-CoA pathway machinery [25]. The phylogenetic tree also shows that CooS-III is in the same subfamily as the corresponding gene in the M. thermoacetica acs operon (Figure 4), suggesting they have the same biological functions. In addition, all the genes in the acetyl-CoA pathway have been identified in the C. hydrogenoformans genome and activities of some of those gene products have been detected (Figure 5), prompting us to propose that this organism carries out autotrophic fixation of CO through this pathway. This is consistent with the observation that key enzymes for the other known CO2 fixation pathways, such as the Calvin cycle, the reverse tricarboxylic acid cycle, and 3-hydroxypropionate cycle are apparently not encoded in the genome.
Figure 5 Predicted Complete Acetyl-CoA Pathway of Carbon Fixation in C. hydrogenoformans
Genes predicted to encode each step in the acetyl-CoA pathway of carbon fixation were identified in the genome. The locus numbers are indicated on the figure.
Oxidative Stress Response (CODH-IV)
C. hydrogenoformans, though an anaerobe, has to deal with oxidative challenges present in the environment from time to time. Unlike aerobes, many anaerobes are proposed to use an alternative oxidative stress protection mechanism that depends on proteins such as rubrerythrin [26,27]. With few exceptions, rubrerythrin-like proteins have been found in complete genomes of all anaerobic and microaerophilic microbes but are absent in aerobic microbes [28]. Rubrerythrin is thought to play a role in the detoxification of reactive oxygen species by reducing the intermediate hydrogen peroxide, although the exact details remain elusive [28,29]. C. hydrogenoformans encodes three rubrerythrin homologs. One of them forms an operon with genes encoding CooS-IV, a CooF homolog, and a NAD/FAD-dependent oxidoreductase (CHY0735–8, Figure 3), suggesting that their functions are related. Here we speculate that this operon encodes a multi-subunit complex where electrons stripped from CO by the CODH are passed to rubrerythrin to reduce hydrogen peroxide to water, with CooF and the NAD/FAD-dependent oxidoreductase acting as the intermediate electron carriers. Therefore, CODH-IV may play an important role in oxidative stress response by providing the ultimate source of reductants.
Others
Two other homologs of CooS are encoded in the genome. The gene encoding CooS-II (CHY0085) was originally cloned with the neighboring cooF (CHY0086) [30] and the complex was purified as functional homodimers [1]. This complex (CODH-II) is membrane-associated and an in vitro study showed it might have an anabolic function of generating NADPH [1]. Its structure has been solved [31]. The role of CooS-V (CHY0034) is more intriguing as it is the most deeply branched of the CooSs (Figure 4) and is not flanked by any genes with obvious roles in CO-related processes.
Aerobic bacteria metabolize CO using drastically different CODHs that are unrelated to the anaerobic ones. The CODHs from aerobes are dimers of heterotrimers composed of a molybdoprotein (CoxL), a flavoprotein (CoxM), and an iron-sulfur protein (CoxS) and belong to a large family of molybdenum hydroxylases including aldehyde oxidoreductases and xanthine dehydrogenases [32]. These enzymes characteristically demonstrate high affinity for CO, and the oxidation is typically coupled to CO2 fixation via the reductive pentose phosphate cycle.
C. hydrogenoformans has one gene cluster (CHY0690–2) homologous to the coxMSL cluster in Oligotropha carboxidovorans, the most well-studied aerobic CODHs. However, our phylogenetic analysis showed that the C. hydrogenoformans homolog of CoxL does not group within the CODH subfamily. Therefore, we conclude that it is unlikely that this gene cluster in C. hydrogenoformans encodes a CODH, although that needs to be tested. Of the available published and unpublished genomes, only R. rubrum appears to have both an anaerobic CODH and a close relative of the aerobic O. carboxidovorans CODH. Accordingly, R. rubrum, a photosynthetic bacterium, can grow in the dark both aerobically and anaerobically using CO as an energy source.
Structures of both the Mo- and Ni-containing enzymes have been published recently. The crystal structure of CooS-II from C. hydrogenoformans is a dimeric enzyme with dual Ni-containing reaction centers each connected to the enzyme surface by 70-Å hydrophobic channels through which CO transits [31]. This channeling, also confirmed experimentally [33,34], explains the mechanism of CO use as a central metabolic intermediate despite its low solubility and generally low concentration in geothermal environments.
Sporulation
The C. hydrogenoformans genome encodes a large number of homologs of genes involved in sporulation in other Firmicutes, spanning all stages of sporulation (Table 2). Among those are the master switch gene spo0A and all sporulation-specific sigma factors, σH, σE, σF, σG, and σK. However, sporulation has not been previously reported for this species. With this in mind, we set out to re-examine the morphology of C. hydrogenoformans cells and found endospore-like structures when cultures were stressed (Figure 6).
Table 2 Orthologs of Known Bacillus subtilis Sporulation Genes in C. hydrogenoformans
Figure 6 An Electron Micrograph of a C. hydrogenoformans Endospore
The finding of homologs of many genes involved in sporulation in other species led us to test whether C. hydrogenoformans also could form an endospore. Under stressful growth conditions, endospore-like structures form. We note that even though homologs could not be found in the genome for many genes that in other species are involved in protective outer-layer (cortex, coat, and exosporium) formation, those structures seem to be visible and intact.
We then used phylogenetic profile analysis to look for other possible sporulation genes in the genome. Phylogenetic profiling works by grouping genes according to their distribution patterns in different species [35]. Proteins that function in the same pathways or structural complexes frequently have correlated distribution patterns. Phylogenetic profile analysis identified an additional set of 37 potential sporulation-related genes (Figure 7). Those genes are generally Bacillales- and Clostridiales-specific, consistent with the fact that endospores have so far only been found in these and other closely related Firmicutes. Most of the novel genes are conserved hypothetical proteins, whereas a few are putative membrane proteins. In support, a few of those novel sporulation genes have been shown to be involved in Bacillus subtilis sporulation by experimental studies [36,37]. The rest of the genes are thus excellent candidates for encoding known sporulation functions that have not been assigned to genes or previously unknown sporulation activities. Strikingly, within this group of genes, in addition to other known sporulation-specific sigma factors (σE, σF, σG, and σK), we identified a sigma factor (CHY1519) that was not known to be associated with sporulation previously. σI, its putative ortholog in B. subtilis, has shown some association with heat shock [38]. It remains to be determined experimentally whether this sigma factor is involved in sporulation, and if so, the regulatory network it controls.
Figure 7 Phylogenetic Profile Analysis of Sporulation in C. hydrogenoformans
For each protein encoded by the C. hydrogenoformans genome, a profile was created of the presence or absence of orthologs of that protein in the predicted proteomes of all other complete genome sequences. Proteins were then clustered by the similarity of their profiles, thus allowing the grouping of proteins by their distribution patterns across species. Examination of the groupings showed one cluster consisting of mostly homologs of sporulation proteins. This cluster is shown with C. hydrogenoformans proteins in rows (and the prediced function and protein ID indicated on the right) and other species in columns with presence of a ortholog indicated in red and absence in black. The tree to the left represents the portion of the cluster diagram for these proteins. Note that most of these proteins are found only in a few species represented in red columns near the center of the diagram. The species corresponding to these columns are indicated. We also note that though most of the proteins in this cluster, for which functions can be predicted, are predicted to be involved in sporulation and some have no predictable functions (highlighted in blue). This indicates that functions of these proteins' homologs have not been characterized in other species. Since these proteins show similar distribution patterns to so many proteins with roles in sporulation, we predict that they represent novel sporulation functions.
A search of known sporulation-related genes in B. subtilis against C. hydrogenoformans revealed that many of them are missing in the genome. Of the 175 B. subtilis sporulation-related genes we compiled from the genome annotation and literature [39,40], half have no detectable homologs in C. hydrogenoformans using BLASTP with an E-value cutoff of 1e-5. Putative orthologs defined by mutual-best-hit methodology are present for only one third of those genes in C. hydrogenoformans. Among those missing genes are spo0B and spo0F, which encode the key components of the complex phosphorelay pathway in B. subtilis that channels various signals such as DNA damage, the ATP level, and cell density to the master switch protein Spo0A and therefore governs the cell's decision to enter sporulation. C. hydrogenoformans hence uses either a simplified version of this pathway or an alternative signal transduction pathway to sense the environmental or physiological stimuli. A large number of genes involved in the protective outer layer (cortex, coat, and exosporium) formation, spore germination, and small acid-soluble spore protein synthesis, among a few genes in various stages of spore development, are also missing. A similar, but slightly different, set of genes are missing in the other spore-forming Clostridia species as well [41]. Absence of those genes is more pronounced in non-spore-forming Firmicutes such as Listeria spp., Staphylococcus spp., and Streptococcus spp., as they lack all the sporulation-specific genes. When overlaid onto the phylogeny of Firmicutes (Figure 2), this observation can be explained by either multiple independent gene-loss events along branches leading to non-Bacillus species or by independent gene-gain events along branches leading to Bacillus and Clostridia, or by both. Whatever the history is of the sporulation evolution, the core set of sporulation genes shared by Bacillus and Clostridia might be close to a “minimal” sporulation set, as so far only these two groups have been found to be capable of producing endospores. Alternatively, some spore specific functions may be carried out by non orthologous genes in different species, which would prevent us from identifying them by this type of analysis.
Strictly Dependent on CO?
Until very recently, C. hydrogenoformans was thought to be an autotroph strictly depending on CO for growth. An overview of the genome reveals features related to its autotrophic lifestyle. For example, it has lost the entire sugar phosphotransferase system and encodes no complete pathway for sugar compound degradation. However, many aspects of the gene repertoire are suggestive of heterotrophic capabilities. For example, among the transporters encoded in the genome are ones predicted to import diverse carbon compounds including formate, glycerol, lactate, C4-dicarboxylate (malate, fumarate, or succinate; the binding receptor for this has three paralogs in the genome), 2-keto-3-deoxygluconate, 2-oxoglutarate, and amino acids. In addition there is a diverse array of signal transduction pathways including chemotaxis not commonly found in the genomes of autotrophs (see below). Consistent with these observations, Henstra et al. recently showed that formate, lactate, and glycerol could be utilized as carbon source provided 9,10-anthraquinone-2,6-disulfonate was used as the electron acceptor [8]. Similarly, sulfite, thiosulfate, sulfur, nitrate, and fumarate were reduced with lactate as electron donor, although heterotrophic growth was relatively slow compared with cultures growing on pure CO [8]. It is not known what electron acceptors are likely to be coupled to these pathways in the isolation locale of C. hydrogenoformans, however it is clear that there is a more versatile complement of energy sources than initially concluded by Svetlichny et al. [2].
In terms of autotrophic lifestyle, although C. hydrogenoformans and S. thermophilum are close phylogenetically, they have gone separate ways in their lifestyles. S. thermophilum is an uncultivable thermophilic bacterium growing as part of a microbial consortium [18], while C. hydrogenoformans is a hot-spring autotroph that can survive efficiently on CO as its sole carbon and energy source. Accordingly, their metabolic capabilities are very different and only half of their proteomes are homologous. It is not clear why S. thermophilum is dependent on other microbes. Unlike other symbiotic microorganisms, no large-scale genome reductions have occurred in S. thermophilum [18]. On the other hand, C. hydrogenoformans has evolved to live preferably on CO, possibly by acquiring and/or expanding its complement of CODHs. As a result, it has lost many genes associated with a heterotrophic lifestyle, such as the phosphotransferase transporter system, and may be on the verge of becoming an obligate autotroph. Even though C. hydrogenoformans is more closely related to S. thermophilum than to T. tengcongensis, an anaerobic thermophile isolated also from freshwater hot springs [42], C. hydrogenoformans actually shares slightly less genes with S. thermophilum than with T. tengcongensis.
Signal Transduction
C. hydrogenoformans is poised to respond to diverse environmental cues through a suite of signal transduction pathways and processes. The organism has 83 one-component regulators and 13 two-component systems (including two chemotaxis systems), which are average numbers for such a genome size [43] (Table S1). Many of the genes encoding these two-component systems are next to transporters, possibly being involved in regulation of solute uptake, while others are adjacent to oxidoreductases. C. hydrogenoformans also possesses an elaborate cascade of chemotaxis genes, including 11 chemoreceptors, and a complete set of flagellar genes, most located within a large cluster of about 70 genes (CHY0963–1033). Chemotaxis allows microbes to respond to environmental stimuli by swimming toward nutrients or away from toxic chemicals. Generally, a heavy commitment to chemotaxis is not a characteristic of autotrophic microorganisms [44], and it is possible that C. hydrogenoformans is responding to gradients of inorganic nutrients, or gases such as CO, O2, H2, or CO2.
Critical for sensing CO, two CooA homologs occur in the C. hydrogenoformans genome, both of which are encoded within operons containing cooS genes. CooA proteins are heme proteins that act as both sensors for CO as well as transcriptional regulators. They belong to the cyclic adenosine monophosphate receptor protein family and induce CO-related genes upon CO binding [45]. CHY1835, encoding CooA1, is at the beginning of the R. rubrum-like coo operon. CHY0083, encoding CooA2, is at the end of the operon possibly involved in NADPH generation from CO [1] (Figure 3).
C. hydrogenoformans lacks certain subfamilies of transcription factors that are present in its close Clostridia relatives, such as those utilizing the following helix-turn-helix domains: iron-dependent repressor DNA-binding domain, LacI, PadR, and DeoR (Pfam nomenclature). The genome does not encode any proteins of the LuxR family, which are usually abundant in both one-component (e.g., quorum-sensing regulators) and two-component systems.
The largest family of transcriptional regulators in C. hydrogenoformans is sigma-54- dependent activators. Eight such regulators comprise one-component systems (CHY0581, CHY0788, CHY1254, CHY1318, CHY1359, CHY1376, CHY1547, and CHY2091) and another one is a response regulator of the two-component system (CHY1855). Seven one-component sigma-54-dependent regulators have at least one PAS domain as a sensory module. PAS domains are known to often contain redox-responsive cofactors, such as FAD, FMN, and heme and serve as intracellular oxygen and redox sensors [46]. Overall, there are 18 PAS domains in C. hydrogenoformans. It is a very significant number compared to only two PAS domains in Moorella thermoacetica (similar genome size) and nine in Desulfitobacterium hafniense (a much larger genome). The most abundant sensory domain of bacterial signal transduction, the LysR substrate-binding domain, which binds small molecule ligands, is present only in six copies in C. hydrogenoformans (there are 36 copies in D. hafniense), re-enforcing the notion that redox sensing via PAS domains might be the most critical signal transduction event for this organism.
The most intriguing signal transduction protein in C. hydrogenoformans is the sigma-54-dependent transcriptional regulator that has an iron hydrogenase-like domain as a sensory module (CHY1547). This domain contains 4Fe-4S clusters and is predicted to use molecular hydrogen for the reduction of a variety of substrates. Its fusion with the sigma-54 activator and the DNA-binding HTH_8 domain in the CHY1547 protein strongly suggests that this is a unique regulator that activates gene expression in C. hydrogenoformans in response to hydrogen availability. Interestingly, it is located immediately upstream of a ten-gene cluster encoding a Ni/Fe hydrogenase (CHY1537–46). Iron hydrogenases similar to the one in CHY1547 can be identified in several bacterial genomes including S. thermophilum, Dehalococcoides ethenogenes, and some Clostridia; however, they are not associated with DNA-binding domains. The only organisms where we found a homologous sigma-54 activator are M. thermoacetica, Geobacter metallireducens, G. sufurreducens, and Desulfuromonas acetoxidans.
Selenocysteine-Containing Proteins
C. hydrogenoformans possesses all known components of the selenocysteine (Sec) insertion machinery (CHY1803:SelA, CHY1802:SelB, CHY2058:SelD) and the Sec tRNA. A total of 12 selenocysteine-containing proteins (selenoproteins) were identified in C. hydrogenoformans genome by the Sec/Cys homology method (Table 3). For each of them, an mRNA stem-loop structure, the signature of the so-called Sec Insertion Sequence (SECIS) required for the Sec insertion, is present immediately downstream of the UGA codon. Although most of the identified selenoproteins are redox proteins, as has been shown for other bacteria and archaea [47], three are novel. Two are transporters (CHY0860, CHY0565), while the third is a methylated-DNA-protein-cysteine methyltransferase (CHY0809), a suicidal DNA repair protein that repairs alkylated guanine by transferring the alkyl group to the cysteine residue at its active site. It is striking that although this protein has been found in virtually every studied organism, only the one in C. hydrogenoformans has selenocysteine in place of cysteine at its active site. Therefore, this selenoprotein most likely evolved very recently, probably from a cysteine-containing protein. Similar patterns exist for the two selenocysteine-containing transporters, suggesting invention of new selenoproteins is an ongoing process in C. hydrogenoformans.
Table 3 Selenoproteins Identified in C. hydrogenoformans Genome
Translational Frameshifts
Analysis of the genome identified many potential cases of frameshifted genes. They are identified by having significant sequence similarity in two reading frames to a single homolog in another species. Examination of sequence traces suggests they are not sequencing errors. Some of these appear to be programmed frameshifts. Programmed frameshifting is a ubiquitous mechanism cells use to regulate translation or generate alternative protein products [48]. The frameshift in the gene prfB (CHY0163), encoding the peptide chain release factor 2, is a well-studied example of programmed frameshift that actually regulates its own translation [48].
However, many of the detected frameshifts appear to be the result of mutations from an ancestral un-frameshifted state. This is best exemplified by examination of the frameshift in the cooS-III gene (CHY1221), which as described above is predicted to encode one of the key components of the acetyl-CoA carbon fixation pathway. In cultures of another strain of this species (DSM 6008), a functional full-length (i.e., unframeshifted) version of this protein has been purified [24] and sequence comparisons of the gene from that strain with ours revealed many polymorphisms, including a deletion in our strain that gave rise to this frameshift (unpublished data). Studies of DSM 6008 show that in cultures grown in excess CO, the acetyl-CoA synthase (ACS, CHY1222) existed predominantly as monomer and only trace amount of CODH-III/ACS complex could be detected. On the other hand, when the CO supply was limited, CODH-III/ACS complex became the dominant form. It is plausible that CODH-III is not absolutely required for carbon fixation when the CO supply is high. Thus the frameshift and other mutations in cooS-III in Z-2901 may reflect the fact that it has been serially cultured in excess CO in the lab for many years. The putative lab-acquired mutations in Z-2901 are yet another reason to sequence type strains of species that have been directly acquired from culture collections and not submitted to extended laboratory culturing [49].
Conclusion
Living solely on CO is not a simple feat and the fact that C. hydrogenoformans does it so well makes it a model organism for this unusual metabolism. Our analysis of the genome sequence, and phylogenomic comparisons with other species, provide insights into this species' specialized metabolism. Perhaps most striking is the presence of genes that apparently encode five distinct carbon monoxide dehydrogenase complexes. Analysis of the genome has also revealed many new perspectives on the biology and evolution of this species, for example, leading us to propose its reclassification, providing further evidence that it is not a strict autotroph and revealing a previously unknown ability to sporulate. The analysis reported here and the availibility of the complete genome sequence should catalyze future studies of this organism and the hydrogenogens as a whole.
Materials and Methods
Medium composition and cultivation.
C. hydrogenoformans Z-2901were cultivated under strictly anaerobic conditions in a basal carbonate-buffered medium composed as described [2]. However, 1.5 g l−1 NaHCO3, 0.2 g l−1 Na2S · 9 H2O, 0.1 g l−1 yeast extract, and 2 μmol l−1 NiCl2 were used instead of reported concentrations, and the Na2S concentration was lowered to 0.04 g l −1. Butyl rubber-stoppered bottles of 120 ml contained 50 ml medium. Bottles were autoclaved for 25' at 121 °C. Gas phases were pressurized to 170 kPa and were composed of 20% CO2 and either 80% of N2, H2, or CO. Sporulation was induced by the addition of 0.01 mM MnCl2 to the medium and by a transient heat shock treatment (100 °C for 5 min).
EM of C. hydrogenoformans endospore.
Samples were fixed with 5% glutaraldehyde for 2 h and 1% OsO4 for 4 h at 4 °C and then embedded in Epon-812. The thin sections were stained with uranyl acetate and lead citrate according to the method described by Miroshnichenko et al. [50]. The samples were observed and photographed using a JEOL JEM-1210 electron microscope.
Genome sequencing.
Genomic DNA was isolated from exponential-phase cultures of C. hydrogenoformans Z-2901. This strain was acquired by Frank Robb from Vitali Svetlitchnyi (Bayreuth University, Germany) in 1995 after being serially grown in culture since its original isolation in 1990. Cloning, sequencing, assembly, and closure were performed as described [51,52]. The complete sequence has been assigned GenBank accession number CP000141 and is available at http://www.tigr.org.
Annotation.
The gene prediction and annotation of the genome were done as previously described [51,52]. CDSs were identified by Glimmer [53]. Frameshifts or premature stop codons within CDSs were identified by comparison to other species and confirmed to be “authentic” by either their high quality sequencing reads or re-sequencing. Repetitive DNA sequences were identified using the REPUTER program [54].
Comparative genomics.
To identify putative orthologs between two species, both of their proteomes were BLASTP searched against a local protein database of all complete genomes with an E-value cutoff of 1e-5. Species-specific duplications were identified and treated as one single gene (super-ortholog) for later comparison. Pair-wise mutual best-hits were then identified as putative orthologs.
Genome tree construction.
Protein sequences of 31 housekeeping genes (dnaG, frr, infC, nusA, pgk, pyrG, rplA, rplB, rplC, rplD, rplE, rplF, rplK, rplL, rplM, rplN, rplP, rplS, rplT, rpmA, rpoB, rpsB, rpsC, rpsE, rpsI, rpsJ, rpsK, rpsM, rpsS, smpB, tsf) from genomes of interest were aligned to pre-defined HMM models and ambiguous regions were auto-trimmed according to an embedded mask. Concatenated alignments were then used to build a maximum likelihood tree using phyml [55].
Phylogenetic profile analysis.
For each protein in C. hydrogenoformans, its presence or absence in every complete genome available at the time of this study was determined by asking whether a putative ortholog was present in that species (see above). Proteins were then grouped by their distribution patterns across species (bits of 1 and 0, 1 for presence and 0 for absence) using the CLUSTER program and the clusters were visualized using the TREEVIEW program (http://rana.lbl.gov/EisenSoftware.htm). Species were weighted by their closeness to each other to partially remove the phylogenetic component of the correlation [56].
Identification of selenoproteins.
Each CDS of C. hydrogenoformans that ends with stop codon TGA was extended to the next stop codon TAA or TAG. It was then searched with BLASTP against the nraa database. A protein with a TGA codon pairing with a conserved Cys site was identified as a putative selenoprotein. The secondary structure of the mRNA immediately downstream of the TGA codon was also checked using MFOLD [57] to look for a possible stem-loop structure.
Supporting Information
Table S1 Regulatory Genes in Clostridia Species
(22 KB DOC)
Click here for additional data file.
We would like to thank The Institute for Genomic Research's (TIGR) Bioinformatics Department for supporting the infrastructure associated with genome annotation and analysis; the TIGR IT Department for general IT support; Dan Haft for discussions regarding selenoproteins; Anne Ciecko, Kristi Berry, and Chris Larkin for initial work on genome closure; Patrick Eichenberger, Richard Losick, and Jim Brannigan for discussions about sporulation; Vitali Svetlichnyi for general discussions about the organism; Terry Utterback and Tamara Feldblyum for coordinating the shotgun sequencing; William C. Nierman for making the genomic libraries; and Claire M. Fraser for supporting the selection of this genome as part of the DOE project referenced below. The sequencing, annotation, and analysis of the genome were supported by United States Department of Energy, Office of Biological Energy Research, Co-Operative Agreement DE-FC0295ER61962. Support for other work in association with this publication came from the National Science Foundation (NSF/MCB-02383387 for FTR and JMG) and the National Institutes of Health (GM072285 for LEU and IBZ).
Competing interests. The authors have declared that no competing interests exist.
Author contributions. MW, FTR, and JAE conceived and designed the experiments. JMG and FTR performed the physiology experiments. MW, QR, LEU, IBZ, FTR, and JAE analyzed the data. MW, IBZ, FTR, and JAE wrote the paper. LJT and KMJ closed the genome. ASD, SCD, LMB, RJD, RM, SAS, JFK, and WCN annotated the genome.
A previous version of this article appeared as an Early Online Release on October 19, 2005 (DOI: 10.1371/journal.pgen.0010065.eor).
Abbreviations
CDSprotein coding sequence
COcarbon monoxide
CODHcarbon monoxide dehydrogenase
Note Added in Proof
It has come to our attention that a complementary comparison of sporulation genes in various Firmicutes was published in 2004 [58]. This study identified homologs of known sporulation genes in Firmicutes by experimental methods and genome analysis. The authors then used these results to study the evolution of sporulation and known sporulation genes.
==== Refs
References
Svetlitchnyi V Peschel C Acker G Meyer O 2001 Two membrane-associated NiFeS-carbon monoxide dehydrogenases from the anaerobic carbon-monoxide-utilizing eubacterium Carboxydothermus hydrogenoformans
J Bacteriol 183 5134 5144 11489867
Svetlichny VA Sokolova TG Gerhardt M Ringpfeil M Kostrikina NA 1991
Carboxydothermus hydrogenoformans gen. nov., sp. nov., a CO-utilizing thermophilic anaerobic bacterium from hydrothermal environments of Kunashir Island Syst Appl Microbiol 14 254 260
Svetlichny V Sokolova T Kostrikina N Lysenko A 1994 A new thermophilic anaerobic carboxydotrophic bacterium Carboxydothermus restrictus sp. nov Mikrobiologiya 63 294 297
Bonch-Osmolovskaya E Miroshnichenko M Slobodkin A Sokolova T Karpov G 1999 Biodiversity of anaerobic lithotrophic prokaryotes in terrestrial hot springs of Kamchatka Microbiology 68 343 351
Sokolova TG Gonzalez JM Kostrikina NA Chernyh NA Tourova TP 2001
Carboxydobrachium pacificum gen. nov., sp. nov., a new anaerobic, thermophilic, CO-utilizing marine bacterium from Okinawa Trough Int J Syst Evol Microbiol 51 141 149 11211251
Sokolova TG Gonzalez JM Kostrikina NA Chernyh NA Slepova TV 2004
Thermosinus carboxydivorans gen. nov., sp. nov., a new anaerobic, thermophilic, carbon-monoxide-oxidizing, hydrogenogenic bacterium from a hot pool of Yellowstone National Park Int J Syst Evol Microbiol 54 2353 2359 15545483
Sokolova TG Kostrikina NA Chernyh NA Tourova TP Kolganova TV 2002
Carboxydocella thermautotrophica gen. nov., sp. nov., a novel anaerobic, CO-utilizing thermophile from a Kamchatkan hot spring Int J Syst Evol Microbiol 52 1961 1967 12508854
Henstra AM Stams AJ 2004 Novel physiological features of Carboxydothermus hydrogenoformans and Thermoterrabacterium ferrireducens
Appl Environ Microbiol 70 7236 7240 15574922
Mojica FJ Diez-Villasenor C Soria E Juez G 2000 Biological significance of a family of regularly spaced repeats in the genomes of Archaea, Bacteria, and mitochondria Mol Microbiol 36 244 246 10760181
Mojica FJ Ferrer C Juez G Rodriguez-Valera F 1995 Long stretches of short tandem repeats are present in the largest replicons of the Archaea Haloferax mediterranei and Haloferax volcanii and could be involved in replicon partitioning Mol Microbiol 17 85 93 7476211
Peng X Brugger K Shen B Chen L She Q 2003 Genus-specific protein binding to the large clusters of DNA repeats (short regularly spaced repeats) present in Sulfolobus genomes J Bacteriol 185 2410 2417 12670964
Rocha E 2002 Is there a role for replication fork asymmetry in the distribution of genes in bacterial genomes? Trends Microbiol 10 393 395 12217498
Dervyn E Suski C Daniel R Bruand C Chapuis J 2001 Two essential DNA polymerases at the bacterial replication fork Science 294 1716 1719 11721055
Brewer BJ 1988 When polymerases collide: Replication and the transcriptional organization of the E. coli chromosome Cell 53 679 686 3286014
Liu B Alberts BM 1995 Head-on collision between a DNA replication apparatus and RNA polymerase transcription complex Science 267 1131 1137 7855590
Garrity GM Bell JA Lilburn TG 2004 Taxonomic outline of the prokaryotes Bergey's manual of systematic bacteriology, Second Edition New York Springer-Verlag 2816 p.
Ohno M Shiratori H Park MJ Saitoh Y Kumon Y 2000
Symbiobacterium thermophilum gen. nov., sp. nov., a symbiotic thermophile that depends on co-culture with a Bacillus strain for growth Int J Syst Evol Microbiol 50 Pt 5 1829 1832 11034494
Ueda K Yamashita A Ishikawa J Shimada M Watsuji TO 2004 Genome sequence of Symbiobacterium thermophilum, an uncultivable bacterium that depends on microbial commensalism Nucleic Acids Res 32 4937 4944 15383646
Ferry JG 1995 CO dehydrogenase Annu Rev Microbiol 49 305 333 8561463
Ragsdale SW Kumar M 1996 Nickel-containing carbon monoxide dehydrogenase/acetyl-CoA synthase(,) Chem Rev 96 2515 2540 11848835
Fox JD He Y Shelver D Roberts GP Ludden PW 1996 Characterization of the region encoding the CO-induced hydrogenase of Rhodospirillum rubrum
J Bacteriol 178 6200 6208 8892819
Soboh B Linder D Hedderich R 2002 Purification and catalytic properties of a CO-oxidizing:H2-evolving enzyme complex from Carboxydothermus hydrogenoformans
Eur J Biochem 269 5712 5721 12423371
Ragsdale SW 2004 Life with carbon monoxide Crit Rev Biochem Mol Biol 39 165 195 15596550
Svetlitchnyi V Dobbek H Meyer-Klaucke W Meins T Thiele B 2004 A functional Ni-Ni-[4Fe-4S] cluster in the monomeric acetyl-CoA synthase from Carboxydothermus hydrogenoformans
Proc Natl Acad Sci U S A 101 446 451 14699043
Roberts DL James-Hagstrom JE Garvin DK Gorst CM Runquist JA 1989 Cloning and expression of the gene cluster encoding key proteins involved in acetyl-CoA synthesis in Clostridium thermoaceticum: CO dehydrogenase, the corrinoid/Fe-S protein, and methyltransferase Proc Natl Acad Sci U S A 86 32 36 2911576
Adams MW Jenney FE Jr. Clay MD Johnson MK 2002 Superoxide reductase: Fact or fiction? J Biol Inorg Chem 7 647 652 12072972
Lynch MC Kuramitsu HK 1999 Role of superoxide dismutase activity in the physiology of Porphyromonas gingivalis
Infect Immun 67 3367 3375 10377114
Weinberg MV Jenney FE Jr. Cui X Adams MW 2004 Rubrerythrin from the hyperthermophilic archaeon Pyrococcus furiosus is a rubredoxin-dependent, iron-containing peroxidase J Bacteriol 186 7888 7895 15547260
Sztukowska M Bugno M Potempa J Travis J Kurtz DM Jr. 2002 Role of rubrerythrin in the oxidative stress response of Porphyromonas gingivali s Mol Microbiol 44 479 488 11972784
Gonzalez JM Robb FT 2000 Genetic analysis of Carboxydothermus hydrogenoformans carbon monoxide dehydrogenase genes cooF and cooS FEMS Microbiol Lett 191 243 247 11024270
Dobbek H Svetlitchnyi V Gremer L Huber R Meyer O 2001 Crystal structure of a carbon monoxide dehydrogenase reveals a [Ni-4Fe-5S] cluster Science 293 1281 1285 11509720
Schubel U Kraut M Morsdorf G Meyer O 1995 Molecular characterization of the gene cluster coxMSL encoding the molybdenum-containing carbon monoxide dehydrogenase of Oligotropha carboxidovorans
J Bacteriol 177 2197 2203 7721710
Maynard EL Lindahl PA 1999 Evidence of a molecular tunnel connecting the active sites for CO2 reduction and acetyl-coA synthesis in acetyl-coA synthase from Clostridium thermoaceticum
J Am Chem Soc 121 9221 9222
Seravalli J Ragsdale SW 2000 Channeling of carbon monoxide during anaerobic carbon dioxide fixation Biochemistry 39 1274 1277 10684606
Pellegrini M Marcotte EM Thompson MJ Eisenberg D Yeates TO 1999 Assigning protein functions by comparative genome analysis: Protein phylogenetic profiles Proc Natl Acad Sci U S A 96 4285 4288 10200254
Eichenberger P Jensen ST Conlon EM van Ooij C Silvaggi J 2003 The sigmaE regulon and the identification of additional sporulation genes in Bacillus subtilis
J Mol Biol 327 945 972 12662922
Molle V Fujita M Jensen ST Eichenberger P Gonzalez-Pastor JE 2003 The Spo0A regulon of Bacillus subtilis
Mol Microbiol 50 1683 1701 14651647
Zuber U Drzewiecki K Hecker M 2001 Putative sigma factor SigI (YkoZ) of Bacillus subtilis is induced by heat shock J Bacteriol 183 1472 1475 11157964
Stragier P Losick R 1996 Molecular genetics of sporulation in Bacillus subtilis
Annu Rev Genet 30 297 241 8982457
Kunst F Ogasawara N Moszer I Albertini AM Alloni G 1997 The complete genome sequence of the gram-positive bacterium Bacillus subtilis
Nature 390 249 256 9384377
Nolling J Breton G Omelchenko MV Makarova KS Zeng Q 2001 Genome sequence and comparative analysis of the solvent-producing bacterium Clostridium acetobutylicum
J Bacteriol 183 4823 4838 11466286
Bao Q Tian Y Li W Xu Z Xuan Z 2002 A complete sequence of the T. tengcongensis genome Genome Res 12 689 700 11997336
Ulrich LE Koonin EV Zhulin IB 2005 One-component systems dominate signal transduction in prokaryotes Trends Microbiol 13 52 56 15680762
Faguy DM Jarrell KF 1999 A twisted tale: The origin and evolution of motility and chemotaxis in prokaryotes Microbiology 145 (Pt 2) 279 281 10075408
He Y Shelver D Kerby RL Roberts GP 1996 Characterization of a CO-responsive transcriptional activator from Rhodospirillum rubrum
J Biol Chem 271 120 123 8550545
Taylor BL Zhulin IB 1999 PAS domains: Internal sensors of oxygen, redox potential, and light Microbiol Mol Biol Rev 63 479 506 10357859
Kryukov GV Gladyshev VN 2004 The prokaryotic selenoproteome EMBO Rep 5 538 543 15105824
Farabaugh PJ 1996 Programmed translational frameshifting Annu Rev Genet 30 507 528 8982463
Ward N Eisen J Fraser C Stackebrandt E 2001 Sequenced strains must be saved from extinction Nature 414 148
Miroshnichenko ML Bonch-Osmolovskaya EA Neuer A Kostrikina NA Chernych NA 1989
Thermococcus stetteri sp. nov., a new extremely thermophilic marine sulfur metabolizing archaebacterium Syst Appl Microbiol 12 257 262
Eisen JA Nelson KE Paulsen IT Heidelberg JF Wu M 2002 The complete genome sequence of Chlorobium tepidum TLS, a photosynthetic, anaerobic, green-sulfur bacterium Proc Natl Acad Sci U S A 99 9509 9514 12093901
Wu M Sun LV Vamathevan J Riegler M Deboy R 2004 Phylogenomics of the reproductive parasite Wolbachia pipientis wMel: A streamlined genome overrun by mobile genetic elements PLoS Biol 2 e69. DOI: 10.1371/journal.pbio.0020069 15024419
Salzberg SL Delcher AL Kasif S White O 1998 Microbial gene identification using interpolated Markov models Nucleic Acids Res 26 544 548 9421513
Kurtz S Choudhuri JV Ohlebusch E Schleiermacher C Stoye J 2001 REPuter: The manifold applications of repeat analysis on a genomic scale Nucleic Acids Res 29 4633 4642 11713313
Guindon S Gascuel O 2003 A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood Syst Biol 52 696 704 14530136
Eisen JA Wu M 2002 Phylogenetic analysis and gene functional predictions: Phylogenomics in action Theor Popul Biol 61 481 487 12167367
Zuker M 2003 Mfold web server for nucleic acid folding and hybridization prediction Nucleic Acids Res 31 3406 3415 12824337
Onyenwoke RU Brill JA Farahi K Wiegel J 2004 Sporulation genes in members of the low G+C Gram-type-positive phylogenetic branch (Firmicutes) Arch Microbiol 182 182 192 15340788
|
16311624
|
PMC1287953
|
CC BY
|
2021-01-05 08:00:24
|
no
|
PLoS Genet. 2005 Nov 25; 1(5):e65
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010065
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1631162510.1371/journal.pgen.0010066plge-01-05-07Research ArticleForward Genetic Analysis of Visual Behavior in Zebrafish Behavioral Screen in ZebrafishMuto Akira Orger Michael B ¤aWehman Ann M Smear Matthew C Kay Jeremy N ¤aPage-McCaw Patrick S ¤bGahtan Ethan ¤cXiao Tong Nevin Linda M Gosse Nathan J Staub Wendy Finger-Baier Karin Baier Herwig *Department of Physiology, Programs in Neuroscience, Genetics, and Developmental Biology, Center for Human Genetics, University of California, San Francisco, California, United States of AmericaMullins Mary EditorUniversity of Pennsylvania School of Medicine, United States of America* To whom correspondence should be addressed. E-mail: [email protected]¤a Current address: Department of Molecular and Cellular Biology, Harvard University, Cambridge, Massachusetts, United States of America
¤b Current address: Department of Biology, Rensselaer Polytechnic Institute, Troy, New York, United States of America
¤c Current address: Department of Psychology, Humboldt State University, Arcata, California, United States of America
11 2005 25 11 2005 1 5 e661 7 2005 19 10 2005 Copyright: © 2005 Muto 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.The visual system converts the distribution and wavelengths of photons entering the eye into patterns of neuronal activity, which then drive motor and endocrine behavioral responses. The gene products important for visual processing by a living and behaving vertebrate animal have not been identified in an unbiased fashion. Likewise, the genes that affect development of the nervous system to shape visual function later in life are largely unknown. Here we have set out to close this gap in our understanding by using a forward genetic approach in zebrafish. Moving stimuli evoke two innate reflexes in zebrafish larvae, the optomotor and the optokinetic response, providing two rapid and quantitative tests to assess visual function in wild-type (WT) and mutant animals. These behavioral assays were used in a high-throughput screen, encompassing over half a million fish. In almost 2,000 F2 families mutagenized with ethylnitrosourea, we discovered 53 recessive mutations in 41 genes. These new mutations have generated a broad spectrum of phenotypes, which vary in specificity and severity, but can be placed into only a handful of classes. Developmental phenotypes include complete absence or abnormal morphogenesis of photoreceptors, and deficits in ganglion cell differentiation or axon targeting. Other mutations evidently leave neuronal circuits intact, but disrupt phototransduction, light adaptation, or behavior-specific responses. Almost all of the mutants are morphologically indistinguishable from WT, and many survive to adulthood. Genetic linkage mapping and initial molecular analyses show that our approach was effective in identifying genes with functions specific to the visual system. This collection of zebrafish behavioral mutants provides a novel resource for the study of normal vision and its genetic disorders.
Synopsis
While many genes have previously been implicated in the development and function of the vertebrate central nervous system, no systematic attempt has been made to build a comprehensive catalog of genes important for its behavioral output. Motion evokes two visual reflexes in zebrafish larvae, the optomotor and the optokinetic response. After mutagenesis with ethylnitrosourea and inbreeding over two generations, the authors of this study searched for point mutations disrupting either, or both, of these innate responses. In almost 2,000 F2 families, they discovered 53 recessive mutations in 41 genetic loci. Developmental phenotypes included abnormal differentiation or absence of photoreceptors, and deficits in retinal ganglion cell differentiation or axon targeting. Physiological phenotypes include disruptions of phototransduction, light adaptation, and behavior-specific responses. Most of the mutants are morphologically indistinguishable from wild type, and many survive to adulthood. Genetic linkage mapping and initial molecular analyses revealed that the authors' approach identified genes with functions specific to the visual system. This collection of zebrafish behavioral mutants provides a novel resource for studying the genetic architecture of the vertebrate central nervous system.
Citation:Muto A, Orger MB, Wehman AM, Smear MC, Kay JN, et al. (2005) Forward genetic analysis of visual behavior in zebrafish. PLoS Genet 1(5): e66.
==== Body
Introduction
An animal's behavioral repertoire is deeply rooted in its genome. Mutations of behaviorally important genes may alter or disrupt either the physiology of neuronal circuits or their development. The first task of a research program aimed at identifying the genetic underpinnings of perception and behavior is to build a comprehensive catalog of genes with specific, non-lethal phenotypes, initially with no regard of when and where in the organism they are acting. Forward genetic screens are the method of choice to identify those genes in an unbiased fashion. This approach was pioneered over 30 years ago by Benzer in Drosophila melanogaster [1] and was quickly extended to Caenorhabditis elegans [2]. In these invertebrate species, the forward genetic strategy was particularly productive for the analysis of sensory systems, such as vision, mechanosensation, and olfaction, where these screens helped to discover many genes important for the patterning of sensory epithelia and for sensory transduction [3–7].
Very few behavioral screens have been attempted in vertebrates to date. In mice, Takahashi and colleagues carried out a screen for dominant mutations disrupting circadian behavior [8]. Other groups have carried out behavioral “shelf screens” of previously discovered mutants in both zebrafish and mice [9–11] or collected mutants in motility and locomotor coordination [12,13]. Here we report on the results of the first large-scale behavioral screen focused on a vertebrate sensory system. Following chemical mutagenesis, we searched for recessive mutations that disrupt visually evoked behaviors in zebrafish. Brockerhoff et al. first showed the utility of optokinetic behavior as a powerful screening tool to find visual mutants [14]. Here we used both the optokinetic response (OKR) and the optomotor response (OMR) as screening assays [9,14–16]. These two behaviors employ different motor outputs (swimming and eye movements, respectively), but they are both elicited by large-field motion and are dependent on the retina as the light-sensing organ [15,17]. In a high-throughput screen of almost 2,000 mutagenized genomes, we discovered 41 loci whose mutations lead to a broad spectrum of specific visual (or visuomotor) impairments. Some of the more striking phenotypes include new mutants in retinal axon targeting and in the adaptive dynamics of light responses. This first survey reveals the extent to which single-gene mutations can perturb visual behavior without affecting gross development or vital organ functions. The identities of the corresponding genes are beginning to provide novel insights into how the visual system is assembled and how cellular and molecular interactions shape sensory processing in the vertebrate brain.
Results
Design of an Efficient, Large-Scale Mutagenesis Screen in Zebrafish
We carried out a large-scale screen for mutants with defects in visually elicited behavior. Forty-one founder males (F0) treated with ethylnitrosourea (ENU; see Materials and Methods) were mated with wild-type (WT) females to generate more than 5,000 F1 fish. Adult F1 fish were mated with other F1 fish, or with WT fish, to generate more than 2,000 F2 families. In total, 3,171 F1 fish were used to generate the 1,896 F2 families (2,550 F1 fish for F1 × F1 crosses, and 621 F1 fish for F1 × WT) that gave at least one healthy clutch of F3 embryos in the subsequent generation. F3 embryos and larvae were obtained by random crosses between siblings from F2 families (6,468 F3 clutches in total, or 3.4 clutches per each F2 family on average). From each F3 clutch, typically 12 larvae were tested for OKR and 25 larvae for the OMR (see below). Fish were routinely scored on the seventh day postfertilization (7 dpf). Including retests, over 500,000 individual fish were screened in the course of three years. Calculations based on binomial statistics [18], taking into account the number of F1 fish used to generate the F2 families, the number of F2 families, the number of crosses for each F2 family, and the number of F3 larval fish tested, show that our screen encompassed 1,688 ENU-mutagenized genomes.
The efficiency of mutagenesis in the founder male germlines was determined by a specific-locus test, using sandy (sdy), a zebrafish tyrosinase mutant [19]. In this test, ENU-treated founder males mated with sdy heterozygous females produced six new sdy mutations in about 2,000 genomes screened. In the actual screen of F2 families, however, two new sdy mutant alleles were identified. The allele distribution of all loci, which was determined after completion of the screen and following extensive complementation tests, shows that our screen was not saturated (see Discussion). We nevertheless successfully identified new alleles of previously reported visual mutants, such as bel and nof (Table 1). Although we did not attempt to characterize mutations falling outside our screening criteria, i.e., those causing embryonic or larval lethality, we noticed (and most of the time discarded) new alleles of chk [20], bru/eby [21,22], ome, and nok [21] (unpublished data).
Table 1 Continued
Table 1 Zebrafish Visual Behavior Mutantsa
Two Behavioral Screening Assays, Executed in Parallel, Discovered 53 Visual Mutants
We screened for mutations disrupting behavioral responses to visual motion. A coarse grating that drifts across the fishes' visual field elicits either of two distinct responses, an OKR or an OMR. In the OMR, WT animals vigorously swim in the direction of the perceived motion (Video S1). When restrained from swimming and presented with a rotating whole-field motion stimulus, the fish show an OKR to cancel retinal slip: WT animals move their eyes to track the motion. These pursuit phases are interrupted at regular intervals by reset movements, or saccades (Video S2) [15]. To achieve high throughput, we automated both visual stimulation and analysis, as described elsewhere [16]. We found that the two screening assays were complementary: The OKR assay is slower and more labor-intensive, but has single-fish resolution; the OMR assay, on the other hand, is fast, but measures only population responses. For each assay, a behavioral index ranging from 0 (no response) to 1.0 (WT) was calculated (see Materials and Methods). Typical OMR and OKR mutant phenotypes are shown in Figure 1A and 1B.
Figure 1 Behavioral Screening Assays
(A) OMR. WT larvae in the racetrack reflexively swim in the same direction as a moving stimulus (top). Mutant larvae (for example, dlns393) with an OMR index of 0 fail to respond (bottom). A contrast-enhanced image outlining the fish is shown in the lower image. In this experiment, WT fish larvae were driven all the way to the right end of the racetrack, which differs slightly from our screening assay [16].
(B) OKR. Eye positions (angles shown by white arrows, far left image) were plotted over time during optokinetic stimulation in one direction. The OKR has a sawtooth profile, consisting of alternating quick and slow phases. OKR mutants show slowed eye movements (for example, nebos342), absence of the OKR (lims382), or no eye movements (flans513). Corresponding OKR indices are given in parentheses.
(C) VBA. WT (VBA index = 1) shows fully contracted melanophores in bright illumination. Mutants (edpos371, ymjs392, and amjs391) show three gradations of darker pigmentation, due to enhanced melanin dispersal. Scale bar is 1 mm.
(D) SSA. Movies of six fish per rectangular well, taken at 0.5 frame per second for 20 min, were subtracted frame by frame and projected into a single image to show the locomotor behavior over time. Blind mutants, such as mtis113 (OKR and OMR indices = 0), may show normal spontaneous activity (SSA index = 1). The mti mutants are also darker (VBA = 0.3), resulting in a higher-contrast image than WT. The walks536 mutants (OKR = 0.8; OMR = 0) show less activity, with some circling (SSA = 0.7), which could explain part of their OMR defect. In beats348 mutants, locomotion is severely compromised (SSA = 0.1). SSA-defective mutants were not systematically kept.
Mutants detected by at least one of the two assays in the primary screen were kept. To select against phenotypes with general defects, we discarded mutants with overt developmental problems, as well as those that were poor swimmers, with a few exceptions. Putative F2 carriers were mated at least twice more for confirmation of the phenotype in their progeny before they were outcrossed. The OKR screen initially picked up 241 putative mutants, or “putants.” Following two retests, 46 lines (23%) were outcrossed. The OMR screen picked up 361 putants, 34 (9%) of which were confirmed and successfully propagated. In addition to high-contrast stimuli, we also routinely used a lower-contrast grating to detect subtle and/or contrast-specific visual defects. The high percentage of false positives is mostly attributable to the use of these weak test stimuli. The OKR and OMR assays were used independently within the primary screen. A considerable number of OKR mutants were later found to be OMR-deficient, and vice versa, as discussed below (Table 1).
The initial false positive rate of this behavioral screen greatly exceeded that of a morphological screen for small-eye mutants carried out in parallel [23]. However, almost all behavioral mutants were recovered in the following generation. Our strategy of extensive retesting as part of the primary screen therefore dramatically decreased the number of false positives and made this screen practical. Mutants or putative mutants with low penetrance were not kept or are not reported here. The mutants presented in this paper, therefore, were found in about 25% of the population in a clutch. To establish potential complementation groups, we systematically crossed heterozygous carriers of mutants with similar phenotypes. Noncomplementing mutations (in which the transheterozygous progeny showed a mutant phenotype) were considered to be allelic (Table 1).
Secondary Screening Assays Allowed Classification of Behavioral Phenotypes
In addition to OMR and OKR, we also assessed the larvae's visually mediated background adaptation (VBA) at 5 dpf, as a complementary strategy to enrich for visual mutants. The VBA is a neuroendocrine response that is controlled by ambient light levels and appears to depend on the function of retinal ganglion cells (RGCs) [17]. Melanophores in the skin contract their melanin granules in a bright environment, while a dark environment induces melanin dispersal [9]. We tested the VBA only in response to long (over 20 min) exposure to bright light, i.e., the mutants' ability to become pale. Figure 1C shows gradations of the VBA defect in three representative mutants. We found that, of the 89 VBA mutants discovered in the screen, 19 (21%) also had specific OMR or OKR defects. The remaining 70 “dark” mutants were either behaviorally normal or had externally visible, morphological phenotypes and were not always maintained.
To identify defects in motor functions, we systematically tested spontaneous swimming activity (SSA) (Figure 1D) in all our mutants. We also made sure that all mutants listed in Table 1, except s513, showed spontaneous, conjugate eye movements similar to WT when presented with a stationary stimulus. Finally, to identify mutants with developmental defects, we systematically examined their retinal and tectal histology and their retinotectal projections (Table 1).
Mutations May Affect Some Visual Behaviors More than Others
Because OKR and OMR are both evoked by motion of a large field grating, but differ in their motor output, our collection of mutants presented us with an opportunity to ask how well single-gene mutations can dissociate these two related behaviors. Are there mutations that impair OMR and OKR in a differential manner (weak dissociation) or even disrupt only one of the behaviors, while leaving the other unaffected (strong dissociation)? Table 1 shows that none of our mutants showed a complete absence of either OMR or OKR together with no defect at all in the other behavior. However, the two behaviors were often affected to different degrees. To reveal potential correlations, we plotted the behavioral profiles of our mutant set (Figure 2). Each data point in Figure 2 corresponds to one mutant, measured repeatedly (n > 3 clutches), and was also shaded to represent that mutant's light-exposed VBA score. Although many mutants lacked any visual responses, for those with partial OMR and OKR phenotypes, there was no clear relationship between the magnitudes of the deficit in the two behaviors (correlation coefficient r = 0.4, when mutants with OKR = 0 and OMR = 0 were excluded). Perhaps surprisingly, the severity of the VBA phenotype was not positively related to either OMR (r = −0.5) or OKR (r = −0.4) defects. The overall correlation of all OMR and OKR indices (r = 0.75) and the absence of exclusively OMR- or OKR-specific mutants suggest that these behaviors are weakly dissociable by single-gene mutation. This indicates that OMR and OKR share a major portion of the underlying neural circuitry. In contrast, the VBA appears to employ a dedicated neural pathway largely segregated, and therefore genetically separable, from motion vision.
Figure 2 Distribution of Behavioral Phenotypes among the Three Visual Responses
OMR index is plotted over OKR index for each mutant. Each circle represents a mutant. The shading of the circles represents the VBA index for that mutant. Only mutants with SSA index greater than 0.6 are shown. OMR is strongly correlated with OKR only for very low scores (around 0). Mildly impaired mutants are often differentially affected. OMR and OKR performance is not correlated to VBA index.
Genes Required for Photoreceptor Differentiation and Survival
We discovered seven genes essential for photoreceptor differentiation and/or maintenance (Figures 3 and 4; Table 1). No other phenotypes could be discovered in these mutants, and at least four of them are adult viable. In two mutants (five alleles of wud and yois121), cone photoreceptors are present, but their shapes are shorter and thicker than in WT (see Figure 3). This “stumpy” morphology is not restricted to one particular cone type, as shown by labeling with zpr1, a double-cone-specific marker (Figure 3C and 3D). In five mutants (five alleles of mti, as well as goshs341, pdays351, lims382, and ssds386), all photoreceptors are lost before 6 dpf, except for a small population in the margins of the eye (Figure 4A–4J), where proliferation and differentiation of neuronal precursors continue throughout the life of the fish [23]. This suggests that some of the newborn cells select the photoreceptor fate, but die shortly after beginning differentiation. In mti mutants, degeneration spreads to the outer part of the inner nuclear layers (Figure 4F and 4H). This mutant is also the only one in this class with defective VBA (Figure 4K), as examined further below.
Figure 3 Example of a Mutant with Abnormal Morphology of Cone Photoreceptors
Photoreceptors in a retinal section stained with DAPI (A and B) and a marker for double cones, zpr1 (C and D) at 7 dpf in WT larva (A, C, and E) and yois121 mutant retina (B, D, and F). Merged images of DAPI (in green) and zpr1 (in magenta) are also shown (E and F). Both zpr1-positive and zpr1-negative cone photoreceptors in the mutant are “stumpy” when compared to those in the control retina (arrows). B, bipolar cells; C, cone photoreceptor cells; H, horizontal cells; ONL, outer nuclear layer; OPL, outer plexiform layer. Scale bar is 10 μm.
Figure 4 Examples of Mutants with Photoreceptor Degeneration
(A–J) WT and mutant retinas (A–H, mtis113; I and J, ssds386) were sectioned and stained with DAPI (A, B, E, F, and I) and zpr1 monoclonal antibody (double-cone photoreceptor marker) (C, D, G, H, and J). At 7 dpf, photoreceptors in the central part of the retina have degenerated in both mti (A–D) and ssd (I–J). In the mti retina at 14 dpf, degeneration has spread to the inner nuclear layer (INL). Arrows show the ciliary marginal zone, from which new cells are continually added to the growing retina. Scale bar is 100 μm.
(K) Mutants with photoreceptor degeneration may (mtis113) or may not (ssds386) be dark in VBA assay. Scale bar is 1 mm.
Six of the seven photoreceptor-defective mutants appear normal in their VBA response to light (Figure 4K). This is a curious finding, as it may suggest that classical cone/rod-mediated photoreception is not strictly required for this neuroendocrine response. It is conceivable that the pineal gland, a light-sensing organ in the dorsal forebrain, may control the VBA instead of, or together with, the retina. We therefore asked if presence of the VBA correlated with an intact pineal in our photoreceptor-degeneration mutants. Both VBA-normal and VBA-defective mutants showed a normal pineal, based on expression of shared marker zpr1 (Figure S1). This suggests that none of the mutated genes found here are necessary for the maintenance of the pineal photoreceptors. Moreover, it implies that pineal photoreceptors are not sufficient to control the VBA. This is consistent with the observation that lakritz mutants, which completely lack all RGCs due to mutation in the atonal homolog atoh7 (ath5), but which apparently have a normal pineal gland, show an extreme VBA defect (VBA = 0) [17]. Based on these combined genetic data, we propose that classical cone/rod photoreception is dispensable for this behavior and that other photosensitive cells, situated in the inner retina, signal ambient light levels to the VBA circuitry via the optic nerve.
Genes Required for General Visual Function, Including Phototransduction and Adaptation to Sudden Increases in Light
We identified 11 mutant alleles of nine genes (blds394, dadas503, dlns518, dlns393, edpos371, lajs304, mzrs130, nofs377, snevs102, zats125, and zats376) without detectable anatomical defects (unpublished data), but with complete absence of OKR and OMR (both indices 0.1 or less) (Figure 5; Table 1). The nofs377 mutation is a new allele of the alpha subunit of cone transducin [24], and the zat gene was shown by positional cloning to encode cone-specific guanylyl cyclase, Gc3 (unpublished data) [25]. Based on these findings, it is likely that some of the other seven genes in this category also encode components of the phototransduction cascade.
Figure 5 Example of an OKR Mutant with Normal Morphology
(A) WT sibling and zats125 mutants are indistinguishable in their appearance (shown here at 6 dpf).
(B) The mutant showed no OKR, but saccadic eye movements, which were not correlated to the motion stimulus. The zat gene encodes cone-specific Gc3.
Other mutants were found to have variable visual impairments. We speculated that some of these mutants were unable to adjust the gain of their visual responses due to defective light adaptation. We therefore rescreened mutants with partial impairments and normal histology, using a behavioral paradigm previously developed by us to test this process in zebrafish larvae [19]. In brief, initially light-adapted fish were placed in a dark environment for a period of 45 min and then tested for OKR after return to light. The recovery of visual responsiveness following the sudden transition from dark to light served as a convenient surrogate measurement for light adaptation, although we do not know how closely this paradigm mimics adaptation. We identified five mutants (nkis136, utas301, ututs357, ymjs392, and mdrs527) in which the measured light adaptation was severely delayed (example in Figure 6). In addition, another mutant, nbks342, had a chronic impairment of both OKR and OMR, which varied with genetic background and occasionally improved with repeated stimulus presentation (unpublished data). The mutated genes may be components of light-adaptation pathways, either in photoreceptors or in the retinal network.
Figure 6 Example of a Mutant with a Potential Defect in Light Adaptation
OKR is plotted at several time points before and after dark treatment for 45 min. WT sibling larvae (n = 6) recover quickly from the dark pulse, while nkis136 mutants (n = 6) show reduced responsiveness for several minutes after return to the light. Average number of saccades to a constant motion stimulus is shown for each time point. Error bars indicate standard deviation.
Genes Required for Ganglion Cell Differentiation and Axon Pathfinding
In WT animals, RGCs project to the contralateral brain and terminate in ten different arborization fields (AFs), of which AF-10, the tectum, is the largest [26]. In our collection of behavioral mutants, we found eight new mutants with specific retinofugal projection deficits (Figure 7): bojs307, darls327, walks536, exas174, misss522, michs314, drgs510, and drgs530, as well as a new allele of bel. In bels385 mutants, RGCs develop normally, but project, in variable proportions, to the ipsilateral side of the brain. The new allele was discovered in the OKR screen, because mutants showed reversed eye movements in response to a drifting grating, as is expected from a predominantly ipsilateral projection [9,27]. The reversed response is seen only when the grating rotates around the mutant, as in the OKR assay, because in this situation the direction of motion is opposite between the two eyes (e.g., temporal-to-nasal for the right eye and nasal-to-temporal for the left eye). In the OMR assay, both eyes are exposed to motion flowing in the same direction. Consequently, the OMR of bel mutants is intact.
Figure 7 Examples of Retinofugal Projection Mutants
(A and B) Sections of WT and bojs307 retina stained with DAPI. The mutant retina has a thinner RGC layer (arrow).
(C and D) Dorsal views of RGC axons from the right eye of a WT and a bojs307 mutant labeled with DiO, showing mutant axons in the ipsilateral tectum (arrow). To show that there is no ipsilateral projection in WT, the image is overexposed.
(E–J) Lateral views of RGC axons labeled with DiO after removal of the eye. Anterior is to the left, dorsal to the top. In WT, the tectum and other retinorecipient areas are clearly visible (E). The arrow indicates AF-4. In darls327, the ventral branch of the optic tract is missing (arrow), and only dorsal tectum is innervated (F). In walks536, innervation of AF-4 (arrow) is disorderly (G). In exas174, the posterior tectum (arrow) appears to be incompletely innervated, while AF-4 is larger than in WT (H). In misss522, AF-4 (arrow) is reduced in size (I). In michs314, there is an ectopic arborization (arrow) at the root of the optic tract (J). Scale bars are 100 μm.
The RGC layer of bojs307 mutants is dramatically reduced to about a third of that in WT (Figure 7A and 7B). The optic nerve is thinner, and a variable fraction (up to 50%) of the remaining RGC axons project ipsilaterally (Figure 7C and 7D). Although the axons make this abnormal choice at the midline, they nevertheless show appropriate targeting on the ipsilateral side, innervating the optic tectum as well as the other major AFs. The boj mutation complements mutations in both lakritz (encoding Atoh7/Ath5) [17] and daredevil (encoding an unknown protein) [28], two previously described genes important for RGC genesis or differentiation. The boj mutants are visually impaired to variable degrees, but most severely in the OMR. Based on our finding that the OMR is normal in bel, the OMR deficit in boj is likely due to the reduced number of RGCs, rather than the ipsilateral projection. Another possible cause could be an as-yet unknown patterning defect in the brain, which is often found in ipsilateral RGC projection mutants [29].
In darls327 mutants, the ventral branch of the optic tract is completely missing, and with it AF-2, AF-3, and AF-6; the dorsal optic tract (with AF-4, AF-5, AF-7, AF-8, and AF-9) appears intact (Figures 7F and 8). The tectum has normal size and histology, but only its dorsal half is innervated at 7 dpf; the ventral half is devoid of retinal input. We asked if the dorsal RGCs, which project their axons to the ventral branch of the optic tact in WT fish (Figure 8A), are missing in darls327 mutants. We detected differentiated RGCs throughout the retina, including the dorsal part (Figure S2). Axon tracing, following injection of 3,3′-dioctadecyloxacarbocyanine (DiO) and 1,1′-dioctadecyl-3,3,3′,3′- tetramethylindodicarbocyanine (DiD) into the nasal-dorsal and temporal-ventral quadrants of the eye, respectively, revealed that the dorsally located RGCs project into the dorsal, instead of the ventral, branch of the optic tract, sharing the same route as the ventral RGCs (Figure 8B). The absence of both the ventral optic tract and the ventral innervation of the tectum (Figure 8B and 8D) suggests that the darl gene is required for specifying dorsal RGC fate. Positional information along the temporal-nasal axis of the retina seems unaltered in the mutant. Despite the severity of the anatomical defect, this mutant's OMR and OKR scores are not substantially reduced. The VBA, however, is severely disrupted, suggesting that this neuroendocrine behavior requires input from dorsally specified RGCs.
Figure 8 The darl Mutant Shows Retinotectal Mapping Deficits
(A and B) The nasal-dorsal quadrant of the retina was labeled with DiO (green), and the temporal-ventral quadrant was labeled with DiD (magenta). In darls327, the ventral branch of the optic tract is missing (arrow). Scale bar is 100 μm.
(C and D) Dorsal view of the tectum in the same larvae as in A and B. The ventral half of the darls327 tectum is not innervated by the dorsal-nasal RGC axons. Anterior is to the left and ventral is to the bottom. Tectal neuropil is demarcated by the dotted line, based on DAPI counterstaining (blue). Scale bar is 50 μm.
The mutants walks536, exas174, and misss522 show specific axon targeting defects, best seen in, but not restricted to, AF-4. AF-4 is associated with the dorsal branch of the optic tract and normally has a well-ordered, compact structure (see Figure 7E). In walks536 and exas174, AF-4 is overelaborated and located at a greater distance from the optic tract (see Figure 7G and 7H). The tectum in the exas174 mutant shows an abnormal shape, particularly in the ventral-posterior region (Figure S3), and AF-9 is often missing or reduced (unpublished data). In misss522 mutants, on the other hand, AF-4 and AF-9 are reduced in size or undetectable (see Figure 7I). This mutant is completely unresponsive to motion, while the walks536 and exas174 mutants show residual OKR and OMR (Table 1). In all three mutants, AFs associated with the ventral tract appear normal. This observation, together with the finding that OMR and OKR are intact in darls327 mutants, which lack the ventral tract, suggest that one or more AFs in the dorsal tract play a key role in OMR and OKR.
In michs314 mutants, a subset of RGC axons make an abnormal turn shortly after crossing the midline and stall to form an ectopic AF (see Figure 7J). The location of this new retinorecipient area is highly consistent among individual mutants. Another OMR mutant, shirs362, has a severely retarded retinofugal projection at 5 dpf, which recovers by 7 dpf, although the dorsal optic tract remains thinner (Figure S4). Finally, in blins573 mutants, axon arbors in the tectal neuropil are disorganized and, in drg (two alleles), a subset of the RGC axons project to the incorrect layer of the tectum [28]. The axon-targeting phenotypes described here are, for the most part, so subtle and localized that they would have escaped previous lipophilic carbocyanine dye-tracing screens [30].
Genes Apparently Required for the Function of Specific Behavioral Pathways
Two mutants, ofrts373 and amjs391, show severe VBA defects with only minor OKR and OMR impairments. Strikingly, the VBA of amjs391 is reversed: The mutant turns dark in the light and light in the dark, which is the opposite of what is seen in WT. At what stage the photoresponse is inverted in this mutant will have to be elucidated. In addition, we discovered several mutants with VBA defects, but normal OMR and OKR, which are not included in Table 1.
Two other VBA mutants, dpgs128 and jakos326, showed normal OKR, but were impaired in the OMR. This selective deficit could not be explained by a locomotor problem, as both mutants show normal SSA and are adult viable. Specific deficits such as these may be either due to differential sensitivity to the stimuli presented in the two assays or due to differential effects of the mutation on the underlying neural circuits. Thus, our screen has discovered a small number of mutations that dissociate visual pathways underlying OMR and OKR.
Genes Required for Posture, Swimming, or Eye Movements
While we did not systematically keep OMR mutants with swimming defects or OKR mutants that did not move their eyes, we saved a small number of mutants whose phenotypes appeared to be informative with regard to specific neural pathways. The morphologically normal beats348, pahs374, slaks564, and flans513 mutants showed reduced OMR and/or OKR in combination with motor abnormalities. The pah gene was positionally cloned and shown to encode phenylalanine hydroxylase, an enzyme required for tyrosine and catecholamine synthesis (unpublished data). These mutations appear to primarily affect motor or other nonsensory central nervous system functions, although additional defects in visual processing may also be present.
Discussion
In this study, we took a forward genetic approach to identify genes involved in zebrafish visually controlled behaviors. In order to capture a large number of mutants, we screened almost 2,000 F2 families and cast a wide, dense net by screening with three complementary behavioral assays. We report here on the initial characterization of 53 specific mutations in 41 genes, only two of which had previously been described.
OKR versus OMR versus VBA as Screening Assays
Choice of a suitable assay is paramount to the success of any genetic screen. We found that each of the three assays employed had its specific strengths and limitations. The OKR assay requires each fish to be mounted individually, dorsal side up, in methylcellulose and is therefore much more time-consuming than the OMR assay, for which each group of fish can just be poured into an elongated tank. The OKR assay therefore dictated the pace of the screen, and we were thus unable to test as many fish as with the OMR assay (and may therefore have missed some mutants). However, since the OKR assay records fish individually, whereas the OMR assay records a population, the OKR has the potential to find less-penetrant phenotypes than the OMR. In the primary screen, OMR and OKR assays each discovered a largely nonoverlapping set of visual mutants, which, upon retesting, showed defects in either assay. Thus, the high throughput of the OMR assay complemented the specificity of the OKR assay. This tradeoff also applies to genetic linkage mapping, which we have so far completed for 25 of the 41 loci. We found that the OMR is most useful for presorting of mutants, while the OKR is most suitable for the subsequent “weeding-out” of false positives. The VBA response, on the other hand, is extremely effective in sorting mutants for linkage mapping, but is less suited as a primary screening assay, because it is prone to missing important mutant classes. Screening with all three assays increased the likelihood of finding all mutants and often provided independent confirmation of a behavioral phenotype.
How Many and What Kinds of Genes Control Visual Behavior?
We found that at least one-quarter, and probably more than half, of the behavioral mutations discovered here affect photoreception. Their phenotypes include defects in photoreceptor formation or maintenance, phototransduction, and adaptation to sudden light changes (whose likely cellular and molecular substrate is located in the outer retina). Another sizable fraction (at least a quarter) of mutations affect RGCs and their projections to the brain. As far as we can conclude so far from our ongoing analysis, mutations affecting the development of higher visual centers (beyond the retinofugal projections) are largely absent from our collection. This could mean that the genes involved in the formation of circuits in higher brain regions are either essential for embryonic development (i.e., their loss of function would lead to early lethality), or they are redundant, which would prevent their discovery by classical mutagenesis screens.
The number of genomes screened should have been sufficient to uncover at least one mutation in each gene of interest, based on the mutation rate measured in the F0 founder males. However, the empirical allele frequency clearly contradicts this optimistic scenario. Of the 41 loci in our collection, 35 are represented by a single allele and four by two alleles. The other two genes for which we found five alleles each, mti and wud, appear to be outliers. Excluding these two loci, and assuming that the probability of finding a mutation follows a Poisson distribution, the number of genes with no hits is estimated at about 150. This back-of-the-envelope calculation shows that our screen was not saturating, and that many more genes may be discovered using our approach. Potential obstacles to future screens include the intrinsic difficulty of detecting mutants in behavior, as opposed to, say, pigmentation (which was used to measure the mutation rate), and the low mutability of some loci, as has been observed in other large-scale zebrafish screens [31,32].
Satisfyingly, we discovered new alleles of several previously identified genes. These include mutants falling within the limits of our screening criteria, such as bel and nof (Table 1), as well as others with more severe phenotypes, such as chk [20], bru [21,22], ome, and nok [21] (unpublished data). It is possible that some of our mutations have generated weak (or maternally rescued) alleles of housekeeping or other essential genes, although the molecular identification of the first set of genes shows that this is not generally the case. For a precise estimate of the number of genes whose mutations lead to specific, nonlethal visual system phenotypes, a much larger screen will have to be carried out.
Genes Involved in Photopic Vision and Photoresponse Dynamics
Zebrafish fill an important niche for the genetic study of photoreception. Human pattern vision, like that of zebrafish, is largely cone-driven. Because most genetic work has been done on the rod-dominated retinas of rodents, less is known about phototransduction in cones. Here we have already discovered two mutant alleles of zatoichi (zats125 and zats376), the gene for cone-specific guanylyl cyclase (Gc3), as well as a new allele of nof, which encodes the alpha subunit of cone transducin [24]. It is likely that there are additional mutants in phototransduction in our collection, and it will be interesting to study their genetic interactions. Zebrafish are appealing for this work, because all their cone opsin genes have been identified [33], and their photoreceptors are amenable for biochemical [34] and psychophysical studies [35].
The visual system operates over a wide range of luminance intensities by adjusting its sensitivity to ambient light levels. At least two adaptation mechanisms are operational in the vertebrate retina, one acting on the phototransduction cascade itself [36–38] and the other on synaptic strengths within the network of neurons [39]. We have discovered five mutants that exhibit delayed recovery of the OKR following a sudden transition from dark to light. These mutants are otherwise normal and adult viable. We speculate that these mutants have defects in light adaptation, although further analyses, such as electroretinogram recordings, will be needed to define and localize the underlying defect. The mutations identified here should provide novel entry points into a molecular dissection of light adaptation.
Zebrafish Mutants as Human Eye Disease Models
We identified five genes whose mutations result in loss of photoreceptors. Several processes can lead to retinitis pigmentosa or macular degeneration in mammals, including structural defects of outer segments, excessive light illumination, and genetic disruption of the phototransduction cascade, but the molecular mechanisms of cell death induction are largely unknown [40]. Photoreceptors are lost quickly in our zebrafish mutants (over days), in contrast to rodent models of retinal degeneration, in which the same process takes months [40]. This is advantageous for the screening of therapeutic drugs that block photoreceptor degeneration. Tests of pharmacological rescue could be carried out in conjunction with our high-throughput behavioral assays. Our collection of zebrafish mutants with rapidly degenerating cones provides us with novel tools to examine the molecular mechanisms of macular degeneration in a model system that is not only genetically tractable, but amenable to small-molecule screens [41].
Axon Targeting and Functional Neuroanatomy
Our screen successfully identified a small assortment of specific axon-guidance mutants. These mutants will serve as starting points for the discovery of proteins involved in axon targeting and synaptic specificity in the visual pathway. But their phenotypes are also significant for assigning function to certain pathways in the zebrafish visual system [42]. While most RGCs project to the midbrain tectum, nine smaller areas, or AFs, also receive direct retinal input [26]. Different AFs are innervated by molecularly and spatially distinct subpopulations of RGCs [28] and probably mediate different visual behaviors. Laser ablations have shown that the tectum is required for localization of prey [43], but is dispensable for OMR, OKR, and VBA [44]. An intact AF-7 is also not necessary for OMR or OKR [44]. Some of the new mutants now help us narrow down the optomotor pathway further by providing “lesions” that are impossible to obtain using surgical, pharmacological, or optical ablation techniques. For instance, in the OMR-deficient misss522 mutant, AF-4 and AF-9 are reduced. This suggests, but does not prove, that one of these underdeveloped AFs is necessary for the OMR. Conversely, darls327 mutants lack AF-2, AF-3, and AF-6, but have an intact OMR, indicating that these three AFs are dispensable for this behavior. Based on these phenotypes, we predict that either AF-4 or AF-9 (or both) are required for the OMR.
Conclusions
Systematic forward genetic approaches have been applied with great success to many areas of biology in a variety of model species. Mutants are not only starting points for gene discovery; their phenotypes often elucidate underlying biological mechanisms even before molecular identification of the mutated genes (e.g., [45]). Our behavioral screen focusing on the zebrafish visual system has achieved three major goals. First, the mutant phenotypes found here have revealed novel genes, or new functions for known genes, which can be identified by positional cloning. Second, these mutations provide novel tools to study central nervous system development and behavior, to localize functions in the brain and to explore the ways in which neuronal circuits reorganize in response to genetic perturbations. Third, our unbiased screen is yielding fundamental insight into the genetic architecture of brain functions and their pathologies. A mutational approach to circuit formation and function, while being an essential first step, should be complemented in the future by targeted manipulations of cells and synapses. Zebrafish are slated to become an excellent system for an integrated genetic approach to unravel cellular and molecular mechanisms of behavior.
Materials and Methods
Fish strains, mutagenesis, and screening.
We used fish from the TL strain for mutagenesis and crossed them to fish from the WIK strain for linkage mapping (see below). Embryos and larval fish were kept in E3 solution (egg water): 5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, and 0.33 mM MgSO4 supplemented with 1:107 w/v methylene blue. Mutations in the zebrafish genome were induced in the spermatogonia of 41 founder males (F0) by three to five treatments with ENU (3 mM for 1 h each, at weekly intervals) and bred to homozygosity over two generations, as previously described [31, 32]. Details of the screen statistics and the specific-locus test used to measure the mutation rate are given in Results.
Genetic linkage mapping.
We used microsatellite-based linkage mapping methods to locate the mutation in the zebrafish genome [46]. Heterozygous carriers of the mutation (in the TL background) were crossed to the highly polymorphic WIK strain. Carrier pairs were identified from this hybrid progeny and mated repeatedly. Clutches were sorted for mutants and nonmutant siblings using behavioral assays (often a combination of OMR to quickly enrich for mutants, followed by OKR of the enriched population for unambiguous identification of mutants). Bulk-segregant analysis was performed using pooled DNA from siblings and mutants. This method involves PCR with a set of 192 polymorphic simple-sequence repeat markers (oligonucleotide primers targeted to unique sequences flanking dinucleotide repeats of variable length [46]). The markers were selected to cover the entire zebrafish genome (25 linkage groups) at roughly even intervals (K. F.-B., unpublished data). Candidate markers showing co-segregation with the mutant pool were confirmed by PCR of single-fish DNA. Map position was further verified by demonstration of linkage to additional markers located in the presumed chromosomal region.
Complementation tests.
We completed classical complementation crosses among all mutants with similar phenotypes (Table 1) or with reported mutants with similar phenotypes or similar map position (if available). Heterozygous nof carriers were obtained from S. Brockerhoff (University of Washington). Heterozygous bel carriers were obtained from C. B. Chien (University of Utah). Complementation tests for nok were carried out by S. Horne (UCSF). Complementation tests for bru were carried out by J. Malicki (Harvard).
Assessment of VBA.
Fish were kept on the fluorescent illuminator (950 cd/m2) for at least 20 min to light-adapt. The pigmentation of the fish was visually scored in four grades to determine the VBA score, with 1 = normal (WT), 0.7 = slightly dark, 0.3 = intermediate dark, and 0 = strongly dark. In this scoring system, the previously discovered, RGC-deficient lakritz mutant scored 0 [17] and served as a reference to calibrate the index. The VBA score for variably dark mutants was estimated by averaging over at least ten individuals.
Recording of the OMR.
The OMR assay was conducted as described previously [16]. Visual stimuli were displayed on a flat-screen CRT monitor that faced upward. The stimuli, which consisted of moving sinusoidal gratings, were generated in MATLAB (MathWorks, Natick, Massachusetts, United States), using the Psychophysics Toolbox extensions (http://psychtoolbox.org). The gamma function of the CRT was measured using a Minolta LS-100 (Tokyo, Japan) light meter, and corrected using MATLAB. The images of the fish before and after each stimulus were captured by a digital still camera (Nikon CoolPix [Tokyo, Japan]), which was triggered by MATLAB using a set of serial commands. These images were downloaded from the camera offline and analyzed using custom macros in Object-Image (http://simon.bio.uva.nl/object-image.html). Ten to 40 larvae (routinely 25) were placed in custom-built acrylic tanks, or “racetracks,” which allowed the larvae to swim in only two directions. Twelve racetracks were placed side by side on the monitor. After subtracting two consecutive images to remove the background, the position of each fish was determined by using the Analyze Particles function of Object-Image. The average position of the fish in each tank before a stimulus was then subtracted from the average position after 30 s of exposure to a standard motion stimulus. The OMR index of a recessive mutant was calculated for stimuli of 100% and 75% contrast by measuring the average distance swum by the 25% weakest responders in a clutch, divided by the distance swum by the 75% best responders. Each stimulus contrast and stimulus direction were repeated four times and the average OMR score was calculated offline.
Recording of the OKR.
The OKR assay was conducted as described previously [16]. An animation of sine-wave gratings was projected on the internal wall of a drum (height, 6 cm; inner diameter, 5.6 cm), using an LCD projector (InFocus LP755 [Wilsonville, Oregon, United States]) [44]. To focus the image at close distance, a wide-angle conversion lens (Kenko VC-050Hi [Tokyo, Japan]), a close-up lens (King CU+1 [Tokyo, Japan]), and a neutral density filter (Hoya ND4 [Tokyo, Japan]) were placed in front of the projector. Twelve zebrafish larvae were immobilized in 2.5% methylcellulose in E3 egg water with their dorsal sides up in the inverted lid of a 3.5-cm diameter petri dish and placed into the center of the drum. The fish were imaged using a dissecting microscope (Nikon SMZ-800) and a CCD camera (Cohu MOD8215–1300 [Tokyo, Japan]) to observe horizontal eye movements. Sinewave gratings with a spatial frequency of 20° per cycle moving at 10°/s were used. Image-J (http://www.rsb.info.nih.gov/ij/) was used for both stimulus generation and image analysis. Images were captured via an LG-3 video capture board (Scion; http://www.scioncorp.com) at two frames per second with Scion Java Package 1.0 for Image-J Windows. A custom-programmed Image-J plug-in (A. M., unpublished data) was used to calculate the changes in eye angles. The OKR index of a mutant was defined here as the saccade number per minute divided by the saccade number per minute observed in WT.
Surrogate light adaptation assay.
The dynamics of OKR in response to sudden changes in illumination was measured as described previously [19]. Fish larvae were put in the dark for 45 min to let them dark-adapt, then subjected to the OKR recording at 2, 8, 15, and 30 min after return to a bright environment (2,400 cd/m2 underneath the larvae; 400–600 cd/m2 at the internal drum wall, where the visual stimulus was projected).
Recording of spontaneous swimming activity.
Spontaneous swimming activity was measured as described [16]. Larvae at 7 dpf were tested in groups of six fish in a rectangular compartment (3 cm × 7.5 cm) of a four-well, clear acrylic plate (12.8 cm × 7.7 cm [Nunc, Roskilde, Denmark]). Fish images were captured by a digital camcorder (Sony TRV-9 [San Diego, California, United States]) at a rate of 0.5 Hz for 20 min in Adobe Premiere. Recorded movies were analyzed using Image-J. Each frame was subtracted (pixel by pixel) from the previous frame to extract the fish that moved during the inter-frame interval. Spontaneous activity was quantified by counting the number of moving fish across all frames. The SSA index was calculated by dividing the number of movement episodes seen in mutants by that seen in WT siblings.
Histology and immunohistochemistry.
Zebrafish larvae were fixed in 4% paraformaldehyde in PBS at 4 °C for 2–16 h, transferred to 30% sucrose in PBS plus 0.02% NaN3 for 16 h or more, mounted in O. C. T. Compound (Sakura Finetek USA, Torrance, California, United States), frozen, and sectioned at 10–12 μm. In some cases, after fixation, the sample was dehydrated in an ethanol series followed by xylene, embedded in paraffin, and sectioned at 6 μm. For immunohistochemistry, the section was incubated with primary antibodies, fluorescent dye-conjugated secondary antibodies (Molecular Probes, Eugene, Oregon, United States), counterstained with 4′,6-diamidino-2-phenylindole (DAPI), and mounted with Fluoromount-G (Southern Biotechnology Associates, Birmingham, Alabama, United States).
Fluorescent axon tracing of the optic tract.
Zebrafish larvae were fixed in 4% paraformaldehyde in half-strength PBS at 4 °C overnight. The fish eye was injected with 1% 1,1′-dioctadecyl-3,3,3′,3′- tetramethylindocarbocyanine (DiI), DiD, or DiO dissolved in chloroform [30]. Fluorescent images were observed with a confocal laser-scanning microscope (BioRad MRC 1024 [Hercules, California, United States] or Zeiss LSM [Oberkochen, Germany]).
Supporting Information
Figure S1 Pineal Photoreceptors Are Present in Retinal Photoreceptor Degeneration Mutants
Coronal sections of the forebrain at 7 dpf were stained with DAPI (A, C, E, and G) and zpr1, a marker of both retinal and pineal photoreceptors (B, D, F, and H). Pineal photoreceptors (arrow and inset) were consistently present in mutants in which retinal photoreceptors were depleted (D, F, and H). Scale bar is 100 μm for A–J and 25 μm for the insets.
(1.2 MB PDF)
Click here for additional data file.
Figure S2 Dorsal RGCs Are Present and Properly Differentiated in darl Mutants
Sagittal sections of WT (A and C) and darls327 retina (B and D) were stained with DAPI (A and B) and zn5 (C and D), a marker for differentiated RGCs. RGCs are present in the dorsal part of the retina and sending out axons into the optic nerve head in the mutant. The mutant eyes are reduced in size compared to WT.
(513 KB PDF)
Click here for additional data file.
Figure S3 The Tectum of exa Mutants Has an Abnormal Shape
RGC axon tracing, following whole-eye DiI fills at 7 dpf, reveals a subtle extension of the tectal neuropil (delineated by DAPI counterstaining) at the ventral-posterior margin (arrow). Scale bar is 50 μm
(1.7 MB PDF)
Click here for additional data file.
Figure S4 Retinal Axon Outgrowth Is Delayed in shir Mutants
Lateral views of the retinal ganglion cell axons labeled with DiO. Anterior to the left, dorsal to the top.
(A and B) At 7 dpf, the retinofugal projection in shirs362 (B) appears similar to WT (A), although the anterior portion may be less dense (arrow).
(C and D) At 5 dpf, RGC axon outgrowth in shirs362 (D) evidently lags behind WT (C). Scale bar is 100 μm.
(1.6 MB PDF)
Click here for additional data file.
Video S1 Optomotor Response
The movie shows a close-up of part of a racetrack tank during OMR testing. A visible light filter has been used to remove the stimulus, and the fish are visualized using infrared light (Sony TRV-9 video camera, night vision mode). The stimulus is represented below. Initially, a converging grating brings the fish into the field of view. After 8 s, the stimulus changes to a rightward-moving grating, and all the fish swim to the right, out of the field of view. At 18 s, the converging movie reappears, and the fish return. Playback in Quicktime runs at twice the actual speed.
(2.3 MB WMV)
Click here for additional data file.
Video S2 Optokinetic Response
The WT larva is on the left, and a zats125 mutant is on the right. For the first 60 s no stimulus is shown, and both fish show spontaneous eye movements. After 60 s, a clockwise-rotating striped pattern is projected on the drum around the fish. The WT fish responds by tracking the pattern slowly to the right and making fast reset saccades to the left. The mutant continues to make undirected spontaneous eye movements.
(2.21 MB MOV)
Click here for additional data file.
Accession Numbers
The GenBank (http://www.ncbi.nlm.nih.gov/) accession numbers of the Danio rerio genes discussed in this paper are retinal guanylyl cyclase 3 (gc3) (AY050505) and phenylalanine hydroxylase (pah) (BC056537).
We thank D. Stainier and S. Baraban, and their labs for collaboration in the screen, in particular L. D'Amico, B. Jungblut, I. Scott, D. Beis, P. Castro, and S. Jin. In addition, S. Brockerhoff, C. B. Chien, S. Horne, and J. Malicki kindly provided mutant carriers for complementation tests. We are grateful to W. Harris, P. Goldsmith, T. Roeser, and M. Taylor for advice and support and to K. Deere, A. Mrejeru, E. Janss, K. Menuz, B. Bogert, H. Haeberle, B. Griffin, M. Dimapasoc, and K. Takahashi for their assistance at various stages of the project. Doctoral and postdoctoral fellowship support came from Naito Foundation (AM), Uehara Memorial Foundation (AM), Howard Hughes Medical Institute (MBO), National Science Foundation (MCS, JNK, LMN), a National Research Service Award from the National Institutes of Health (NIH) (EG), American Heart Association (MCS), University of California, San Francisco Chancellor's Fund (MCS), an Achievement Reward for College Scientist/ARCS (AMW), American Association of University Women Educational Foundation (AMW), National Alliance for Research on Schizophrenia and Depression (PSPM), and an NIH neuroscience postdoctoral training grant (TX). HB was supported by the NIH (EY12406, EY13855, NS42328), by the Sandler Family, by the Sloan Foundation, by the Klingenstein Foundation, and by the David and Lucile Packard Foundation.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. HB conceived the project. AM, MBO, and HB designed the experiments. AM, MBO, AMW, MCS, JNK, PSPM, EG, TX, LMN, NJG, WS, KFB, and HB performed the experiments. AM, MBO, JNK, and HB analyzed the data. AM, MBO, MCS, PSPM, KFB, and HB contributed reagents/materials/analysis tools. AM and HB wrote the paper with input from all authors.
Abbreviations
AFarborization field
[number] dpfday [number] postfertilization
DiD1,1′-dioctadecyl-3,3,3′,3′- tetramethylindodicarbocyanine
DiI1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine
DiO3,3′-dioctadecyloxacarbocyanine
ENUethylnitrosourea
OKRoptokinetic response
OMRoptomotor response
PhRphotoreceptor cell
RGCretinal ganglion cell
SSAspontaneous swimming activity
VBAvisually mediated background adaptation
WTwild-type
==== Refs
References
Benzer S 1973 Genetic dissection of behavior Sci Am 229 24 37 4202065
Brenner S 1974 The genetics of Caenorhabditis elegans
Genetics 77 71 94 4366476
Bargmann CI 1993 Genetic and cellular analysis of behavior in C. elegans
Annu Rev Neurosci 16 47 71 8460900
Zuker CS 1996 The biology of vision of Drosophila
Proc Natl Acad Sci U S A 93 571 576 8570597
Clandinin TR Lee CH Herman T Lee RC Yang AY 2001
Drosophila LAR regulates R1-R6 and R7 target specificity in the visual system Neuron 32 237 248 11683994
Hardie RC Raghu P 2001 Visual transduction in Drosophila
Nature 413 186 193 11557987
Chalfie M Au M 1989 Genetic control of differentiation of the Caenorhabditis elegans touch receptor neurons Science 243 1027 1033 2646709
Takahashi JS Pinto LH Vitaterna MH 1994 Forward and reverse genetic approaches to behavior in the mouse Science 264 1724 1733 8209253
Neuhauss SC Biehlmaier O Seeliger MW Das T Kohler K 1999 Genetic disorders of vision revealed by a behavioral screen of 400 essential loci in zebrafish J Neurosci 19 8603 8615 10493760
Balkema GW Mangini NJ Pinto LH Vanable JW Jr. 1984 Visually evoked eye movements in mouse mutants and inbred strains. A screening report Invest Ophthalmol Vis Sci 25 795 800 6735643
Nicolson T Rusch A Friedrich RW Granato M Ruppersberg JP 1998 Genetic analysis of vertebrate sensory hair cell mechanosensation: The zebrafish circler mutants Neuron 20 271 283 9491988
Granato M van Eeden FJ Schach U Trowe T Brand M 1996 Genes controlling and mediating locomotion behavior of the zebrafish embryo and larva Development 123 399 413 9007258
Nolan PM Peters J Vizor L Strivens M Washbourne R 2000 Implementation of a large-scale ENU mutagenesis program: Towards increasing the mouse mutant resource Mamm Genome 11 500 506 10886012
Brockerhoff SE Hurley JB Janssen-Bienhold U Neuhauss SC Driever W 1995 A behavioral screen for isolating zebrafish mutants with visual system defects Proc Natl Acad Sci U S A 92 10545 10549 7479837
Easter SS Jr. Nicola GN 1996 The development of vision in the zebrafish (Danio rerio)
Dev Biol 180 646 663 8954734
Orger MB Gahtan E Muto A Page-McCaw P Smear MC 2004 Behavioral screening assays in zebrafish Methods Cell Biol 77 53 68 15602905
Kay JN Finger-Baier KC Roeser T Staub W Baier H 2001 Retinal ganglion cell genesis requires lakritz, a zebrafish atonal homolog Neuron 30 725 736 11430806
Mullins MC Hammerschmidt M Haffter P Nusslein-Volhard C 1994 Large-scale mutagenesis in the zebrafish: In search of genes controlling development in a vertebrate Curr Biol 4 189 202 7922324
Page-McCaw PS, Chung SC, Muto A, Roeser T, Staub W, et al. 2004 Retinal network adaptation to bright light requires tyrosinase Nat Neurosci 7 1329 1336 15516923
Loosli F Staub W Finger-Baier KC Ober EA Verkade H 2003 Loss of eyes in zebrafish caused by mutation of chokh/rx3
EMBO Rep 4 894 899 12947416
Malicki J Neuhauss SC Schier AF Solnica-Krezel L Stemple DL 1996 Mutations affecting development of the zebrafish retina Development 123 263 273 9007246
Goldsmith P Baier H Harris WA 2003 Two zebrafish mutants, ebony and ivory, uncover benefits of neighborhood on photoreceptor survival J Neurobiol 57 235 245 14608660
Wehman AM Staub W Meyers JR Raymond PA Baier H 2005 Genetic dissection of the zebrafish retinal stem-cell compartment Dev Biol 281 53 65 15848388
Brockerhoff SE Rieke F Matthews HR Taylor MR Kennedy B 2003 Light stimulates a transducin-independent increase of cytoplasmic Ca2+ and suppression of current in cones from the zebrafish mutant nof
J Neurosci 23 470 480 12533607
Hisatomi O Honkawa H Imanishi Y Satoh T Tokunaga F 1999 Three kinds of guanylate cyclase expressed in medaka photoreceptor cells in both retina and pineal organ Biochem Biophys Res Commun 255 216 220 10049688
Burrill JD Easter SS Jr. 1994 Development of the retinofugal projections in the embryonic and larval zebrafish (Brachydanio rerio)
J Comp Neurol 346 583 600 7983245
Rick JM Horschke I Neuhauss SC 2000 Optokinetic behavior is reversed in achiasmatic mutant zebrafish larvae Curr Biol 10 595 598 10837226
Xiao T Roeser T Staub W Baier H 2005 A GFP-based genetic screen reveals mutations that disrupt the architecture of the zebrafish retinotectal projection Development 132 2955 2967 15930106
Culverwell J Karlstrom RO 2002 Making the connection: Retinal axon guidance in the zebrafish Semin Cell Dev Biol 13 497 506 12468253
Baier H Klostermann S Trowe T Karlstrom RO Nusslein-Volhard C 1996 Genetic dissection of the retinotectal projection Development 123 415 425 9007259
Driever W Solnica-Krezel L Schier AF Neuhauss SC Malicki J 1996 A genetic screen for mutations affecting embryogenesis in zebrafish Development 123 37 46 9007227
Haffter P Granato M Brand M Mullins MC Hammerschmidt M 1996 The identification of genes with unique and essential functions in the development of the zebrafish, Danio rerio
Development 123 1 36 9007226
Chinen A Hamaoka T Yamada Y Kawamura S 2003 Gene duplication and spectral diversification of cone visual pigments of zebrafish Genetics 163 663 675 12618404
Taylor MR Van Epps HA Kennedy MJ Saari JC Hurley JB 2000 Biochemical analysis of phototransduction and visual cycle in zebrafish larvae Methods Enzymol 316 536 557 10800700
Orger MB Baier H 2005 Channeling of red and green cone inputs to the zebrafish optomotor response Vis Neurosci 22 275 281 16079003
Korenbrot JI Rebrik TI 2002 Tuning outer segment Ca2+ homeostasis to phototransduction in rods and cones Adv Exp Med Biol 514 179 203 12596922
Fain GL Matthews HR Cornwall MC Koutalos Y 2001 Adaptation in vertebrate photoreceptors Physiol Rev 81 117 151 11152756
Matthews HR 1992 The role of cytoplasmic calcium in photoreceptor light adaptation J Physiol Paris 86 147 155 1343592
Witkovsky P 2004 Dopamine and retinal function Doc Ophthalmol 108 17 40 15104164
Rattner A Sun H Nathans J 1999 Molecular genetics of human retinal disease Annu Rev Genet 33 89 131 10690405
Goldsmith P 2004 Zebrafish as a pharmacological tool: The how, why and when Curr Opin Pharmacol 4 504 512 15351356
Gahtan E Baier H 2004 Of lasers, mutants, and see-through brains: Functional neuroanatomy in zebrafish J Neurobiol 59 147 161 15007833
Gahtan E Tanger P Baier H 2005 Visual prey capture is controlled by identified reticulospinal neurons downstream of the tectum J Neurosci 25 9294 9303 16207889
Roeser T Baier H 2003 Visuomotor behaviors in larval zebrafish after GFP-guided laser ablation of the optic tectum J Neurosci 23 3726 3734 12736343
Nusslein-Volhard C Wieschaus E 1980 Mutations affecting segment number and polarity in Drosophila
Nature 287 795 801 6776413
Shimoda N Knapik EW Ziniti J Sim C Yamada E 1999 Zebrafish genetic map with 2000 microsatellite markers Genomics 58 219 232 10373319
|
16311625
|
PMC1287954
|
CC BY
|
2021-01-05 07:59:37
|
no
|
PLoS Genet. 2005 Nov 25; 1(5):e66
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010066
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1631162610.1371/journal.pgen.0010068plge-01-05-11Research ArticleGenetic Variation in the HSD17B1 Gene and Risk of Prostate Cancer HSD17B1 and Prostate CancerKraft Peter 1Pharoah Paul 2Chanock Stephen J 3Albanes Demetrius 4Kolonel Laurence N 5Hayes Richard B 4*Altshuler David 6Andriole Gerald 7Berg Christine 8Boeing Heiner 9Burtt Noel P 10Bueno-de-Mesquita Bas 11Calle Eugenia E 12Cann Howard 13Canzian Federico 14Chen Yen-Ching 1Crawford David E 15Dunning Alison M 16Feigelson Heather S 12Freedman Matthew L 17Gaziano John M 18Giovannucci Ed 19Gonzalez Carlos Alberto 20Haiman Christopher A 21Hallmans Goran 22Henderson Brian E 21Hirschhorn Joel N 10Hunter David J 119Kaaks Rudolf 23Key Timothy 24Marchand Loic Le 5Ma Jing 25Overvad Kim 26Palli Domenico 27Pike Malcolm C 21Riboli Elio 28Rodriguez Carmen 29Setiawan Wendy V 30Stampfer Meir J 31Stram Daniel O 32Thomas Gilles 13Thun Michael J 12Travis Ruth 24Trichopoulou Antonia 33Virtamo Jarmo 34Wacholder Sholom 41 Program in Molecular and Genetic Epidemiology, Harvard School of Public Health, Boston, Massachusetts, United States of America
2 CRC Human Cancer Genetics Research Group, University of Cambridge, Cambridge, United Kingdom
3 Core Genotyping Facility, National Cancer Institute, Gaithersburg, Maryland, United States
4 Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland, United States of America
5 Cancer Research Center, University of Hawaii, Honolulu, Hawaii, United States of America
6 Broad Institute at Harvard and MIT, Cambridge, Massachusetts, United States of America
7 Washington University, St. Louis, Missouri, United States of America
8 Division of Cancer Prevention, National Cancer Institute, Bethesda, Maryland, United States of America
9 Department of Epidemiology, German Institute of Human Nutrition, Potsdam, Germany
10 Whitehead/MIT Center for Genome Research, Cambridge, Massachusetts, United States of America
11 Centre for Nutrition and Health, National Institute for Public Health and the Environment, Bilthoven, the Netherlands
12 Department of Epidemiology and Surveillance Research, American Cancer Society, National Home Office, Atlanta, Georgia, United States of America
13 Fondation Jean Dausset, Centre d'Etude du Polymorphisme Humain, Paris, France
14 Genetic Susceptibility Group, International Agency for Research on Cancer, Lyon, France
15 Anschutz Cancer Pavillon, Aurora, Colorado, United States of America
16 Department of Oncology, University of Cambridge, Cambridge, United Kingdom
17 Whitehead Institute for Biomedical Research, Cambridge, Massachusetts, United States of America
18 Department of Medicine, Division of Aging, Brigham and Women's Hospital, Boston, Massachusetts, United States of America
19 Department of Nutrition, Harvard School of Public Health, Boston, Massachusetts, United States of America
20 Catalan Institute of Oncology, L'Hospitalet de Llobregat, Barcelona, Spain
21 Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America
22 Department of Public Health and Clinical Medicine, Umea University, Umea, Sweden
23 Hormones and Cancer Group, International Agency for Research on Cancer, Lyon, France
24 Epidemiology Unit, Cancer Research UK, Oxford, United Kingdom
25 Channing Laboratory, Brigham and Women's Hospital, Department of Medicine, Harvard Medical School, Boston, Massachusetts, United States of America
26 Department of Clinical Epidemiology, Aalborg Hospital, Aarhus University Hospital, Aalborg, Denmark
27 Molecular and Nutritional Epidemiology Unit, CSPO-Scientific Institute of Tuscany, Florence, Italy
28 Unit of Nutrition and Cancer, International Agency for Research on Cancer, Lyon, France
29 American Cancer Society, Atlanta, Georgia, United States of America
30 Department of Preventive Medicine, University of Southern California/Norris Comprehensive Cancer Center, Los Angeles, California, United States of America
31 Department of Epidemiology, Harvard School of Public Health, Boston, Massachusetts, United States of America
32 Division of Biostatistics and Genetic Epidemiology, Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California, United States of America
33 Department of Hygiene and Epidemiology, University of Athens Medical School, Athens, Greece
34 Cancer Prevention Unit, National Public Health Institute, Helsinki, Finland
Abecasis Goncalo EditorUniversity of Michigan, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 25 11 2005 1 5 e688 8 2005 21 10 2005 Copyright: © 2005 Kraft et al.2005This is an open-access article distributed under the terms of the Creative Commons Public Domain declaration, which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose.Steroid hormones are believed to play an important role in prostate carcinogenesis, but epidemiological evidence linking prostate cancer and steroid hormone genes has been inconclusive, in part due to small sample sizes or incomplete characterization of genetic variation at the locus of interest. Here we report on the results of a comprehensive study of the association between HSD17B1 and prostate cancer by the Breast and Prostate Cancer Cohort Consortium, a large collaborative study. HSD17B1 encodes 17β-hydroxysteroid dehydrogenase 1, an enzyme that converts dihydroepiandrosterone to the testosterone precursor Δ5-androsterone-3β,17β-diol and converts estrone to estradiol. The Breast and Prostate Cancer Cohort Consortium researchers systematically characterized variation in HSD17B1 by targeted resequencing and dense genotyping; selected haplotype-tagging single nucleotide polymorphisms (htSNPs) that efficiently predict common variants in U.S. and European whites, Latinos, Japanese Americans, and Native Hawaiians; and genotyped these htSNPs in 8,290 prostate cancer cases and 9,367 study-, age-, and ethnicity-matched controls. We found no evidence that HSD17B1 htSNPs (including the nonsynonymous coding SNP S312G) or htSNP haplotypes were associated with risk of prostate cancer or tumor stage in the pooled multiethnic sample or in U.S. and European whites. Analyses stratified by age, body mass index, and family history of disease found no subgroup-specific associations between these HSD17B1 htSNPs and prostate cancer. We found significant evidence of heterogeneity in associations between HSD17B1 haplotypes and prostate cancer across ethnicity: one haplotype had a significant (p < 0.002) inverse association with risk of prostate cancer in Latinos and Japanese Americans but showed no evidence of association in African Americans, Native Hawaiians, or whites. However, the smaller numbers of Latinos and Japanese Americans in this study makes these subgroup analyses less reliable. These results suggest that the germline variants in HSD17B1 characterized by these htSNPs do not substantially influence the risk of prostate cancer in U.S. and European whites.
Synopsis
Steroid hormones such as estrogen and testosterone are hypothesized to play a role in the development of cancer. This is the first substantive paper from the Breast and Prostate Cancer Cohort Consortium, a large, international study designed to assess the effect of variation in genes that influence hormone production and activity on the risk of breast and prostate cancer. The investigators first constructed a detailed map of genetic variation spanning HSD17B1, a gene involved in the production of estrogen and testosterone. This enabled them to efficiently measure common variation across the whole gene, capturing information about both known variants with a plausible function and unknown variants with an unknown function. Because of the results with a large number of study participants, the investigators could rule out strong associations between common HSD17B1 variants and risk of prostate cancer among U.S. and European whites. While this sheds some light on the carcinogenic effects of one enzyme involved in the complex process of steroid hormone production, it remains to be determined whether variants in other genes play a more important role or if the combined effects of several genes within these pathways have a larger impact.
Citation:Kraft P, Pharoah P, Chanock SJ, Albanes D, Kolonel LN, et al. (2005) Genetic variation in the HSD17B1 gene and risk of prostate cancer. PLoS Genet 1(5): e68.
==== Body
Introduction
Prostate cancer is a leading cause of mortality and morbidity in both western Europe and the United States, where it is the most commonly diagnosed nondermatological cancer and is the second leading cause of cancer death in men. Although epidemiological investigations over several decades have studied exogenous risk factors for prostate cancer, including diet, occupation, and sexually transmitted agents, the only established risk factors for this disease are age, ethnicity, and family history of prostate cancer.
A large body of evidence suggests that inherited genetic susceptibility plays an important role in prostate cancer etiology [1,2]. The risk of prostate cancer in first-degree male relatives of affected individuals is about twice that for the general population, with greater risks for those with an increased number of affected family members [3,4]. Twin studies show that the majority of this excess familial risk is due to inherited factors [5]. Unlike other cancers, however, high-penetrance prostate cancer susceptibility genes have not been consistently identified. Numerous studies have found suggestive linkage signals, but these have been difficult to replicate. Similarly, studies of candidate genes suggested by early linkage studies (e.g., ELAC2, RNASEL, and MSR1) have provided inconsistent evidence for association [1,2]. Carriers of mutations in BRCA1 and BRCA2 reportedly have increased risk of prostate cancer [6,7], but these mutations are rare and account for only a small fraction of excess familial risk. All of this evidence suggests multiple common variants that moderately increase prostate cancer risk have yet to be identified. Genes encoding proteins involved in hormone biosynthesis are plausible candidates for such low-penetrance variants.
Steroid hormones are believed to play an important role in prostate carcinogenesis for several reasons. First, androgens are essential for prostate maturation and functional integrity. Second, androgen ablation is standard therapy for metastatic prostate cancer. Third, androgens are generally needed to induce prostate cancer in animal models.
The results of studies of serum concentration of androgens in relation to prostate cancer risk have been somewhat inconsistent. A meta-analysis of eight prospective serum-based studies showed modest increases in prostate cancer risk associated with androstanediol glucuronide levels but not with testosterone, non–steroid hormone–binding globulin-bound testosterone, dihydrotestosterone, or androstendione levels [8], although the largest prospective study found increased risk with increasing levels of testosterone, after adjustment for serum sex hormone–binding globulin [9].
One endogenous source of variation in serum or tissue concentrations of steroid hormones may be functional variants in genes related to their synthesis and catabolism. Pursuing this line of reasoning, several investigators have examined polymorphisms in some of these genes [10–12]. For example, the missense mutation A49T in the steroid 5α-reductase type 2 gene (SRD5A2) increases enzyme activity for converting testosterone to the more potent dihydrotestosterone [13]. Several studies found that men who carry the T allele are at increased risk for prostate cancer, although these results are not conclusive [14]. Another study found that the T allele is associated with tumor aggressiveness [15]. Shorter repeats of the (CAG)n trinucleotide in the androgen receptor gene (AR) are associated with increased androgen response gene transactivation and show modest increases in risk in some, but not all, studies [10,16,17]. The CYP3A4*1B allele has also been consistently associated with prostate cancer onset and severity, although the functional impact of this allele remains controversial [18]. A number of studies have found evidence that combinations of multiple polymorphisms in steroid-pathway genes increase risk of prostate cancer [11,18,19], but these studies have had low power to detect gene-gene interactions (increasing the likelihood that these results are due to chance [20]) and to the best of our knowledge have not been replicated.
Thus, the combined evidence provides several clues about the role of steroid hormones in prostate cancer yet does not permit definitive conclusions, partly because of limitations in previous epidemiological study designs. Serum hormone studies are limited because serum levels may not reflect prostate tissue levels, and the studies have generally been small (case sample sizes have varied from 16 to 222). Further, hormone interrelationships (e.g., estrogen-androgen balance) have not been considered in detail. Genetic studies have assessed only a few of the potentially important gene variants in similarly underpowered investigations; and gene-gene and gene-environment interrelationships have yet to be effectively examined.
The Breast and Prostate Cancer Cohort Consortium (BPC3), a large, multicenter collaborative study, aims to examine the role of steroidal hormones in prostate cancer by comprehensively measuring variation in more than 30 genes involved in the steroidal hormone pathway and their associated receptors in 8,301 prostate cancer cases and 9,373 controls (unpublished data). The BPC3 has adopted a multistage approach combining genomic, statistical, and epidemiological methods that involves targeted resequencing in a multiethnic sample of 190 advanced breast and prostate cancer cases, followed by genotyping a dense set of common single nucleotide polymorphisms (SNPs) across a region spanning the gene in a multiethnic sample of 349 cancer-free subjects. These genotyping data are used to assess patterns of linkage disequilibrium and select efficient haplotype-tagging SNPs (htSNPs), which are then genotyped on the cases and controls in the main study. The BPC3 provides excellent statistical power to detect modest associations between common genetic variants and risk of prostate cancer and to assess the joint effect of genetic variation and other established risk factors.
Here we report on the association between prostate cancer and the gene encoding 17β-hydroxysteroid dehydrogenase 1 (17β-HSD-1), HSD17B1, which is situated on chromosome 17q21 near BRCA1. 17β-HSD-1 plays a role in estrogen and testosterone biosynthesis. We hypothesize that germline variation in HSD17B1 may lead to variation in 17β-HSD-1 activity. Specifically, 17β-HSD-1 catalyzes the conversion of estrone to the more reactive estradiol and may play a role in the conversion of adrenal-derived dehydroepiandrosterone to Δ5-androsterone-3β,17β-diol [21]. Δ5-Androsterone-3β,17β-diol has estrogenic activity and peroxisomal proliferation activity, via peroxisome proliferative activated receptor α [22]. It also acts as a substrate for conversion to testosterone by 3β-hydroxysteroid dehydrogenase/Δ4-Δ5 isomerase. Testosterone in turn can be metabolized to the more functionally active dihydrotestosterone by steroid-5-α-reductase [21,23]. Dehydroepiandrosterone, estrogens (estrone, estradiol), and androgens (testosterone, dihydrotestosterone) are hormones that affect prostate physiology and possibly carcinogenesis [24,25]. Thus, increased activity of 17β-HSD-1 may increase the levels of these hormones and the risk of prostate cancer. Functionally active 17β-HSD-1 is expressed in the testis [26], the primary site of testosterone synthesis. Although initial studies found evidence of HSD17B1 expression in prostate tissue [21,27], more recent studies of prostate cancer cell lines have found small amounts of the longer of the two HSD17B1 transcripts, which does not appear to correlate with 17β-HSD-1 protein levels [28–32].
While previous studies have evaluated whether germline variation in HSD17B1 is associated with breast or endometrial cancer [33–36], this is the first large prospective study to assess HSD17B1 in relation to prostate cancer among men from several ethnicities.
Results
Table 1 shows demographic and other characteristics of cases and controls from the seven cohorts. Most study subjects were U.S. or European whites (75%), followed by African Americans (10%), Latinos (7%), Japanese Americans (5%), and Native Hawaiians (1%). Of the 8,301 prostate cancer cases and 9,373 controls sent for genotyping, at least one SNP was successfully genotyped for 8,290 (>99.8%) cases and 9,367 (>99.9%) controls, with 7,713 (93%) cases and 8,715 (93%) controls genotyped for all four markers. Among those subjects with data on both genotypes and family history, 832 (14%) cases and 555 (9%) controls reported a father or a brother with prostate cancer. Cases and controls were comparable with respect to age, body mass index (BMI [kg/m2]), and height. Stage information was available on 71% of genotyped prostate cancer cases, and among these, 1,312 (22%) had advanced disease (defined as stage C or D disease at diagnosis or death due to prostate cancer). Gleason score was recorded for 66% of genotyped cases, with 990 cases (18% of those with Gleason scores exhibiting scores of eight or greater).
Table 1 Characteristics of the Study Population by Study, BPC3
Resequencing exons in 190 advanced cancer cases identified two novel nonsynonymous coding SNPs, one of which was seen more than once. (For more detailed resequencing results, see Gene_Summary_Table.xls available under “Genes: Data and Haplotypes” at http://www.uscnorris.com/Core/DocManager/OpenFile.aspx?DocID=9394.) The latter SNP and 25 common SNPs spanning a 42-kb region including HSD17B1 were genotyped in a multiethnic reference panel of 349 cancer-free subjects. Nineteen of these 26 SNPs formed a block of high linkage disequilibrium (Figure 1) that spans HSD17B1, including (5′ to 3′) the pseudogene HSD17BP1, the promoter region, and the gene TCFL4. We found three common haplotypes (>5% frequency) within this block among whites in the reference panel, with a cumulative frequency above 83% (Table S1). We chose four SNPs that predict these common haplotypes in whites with a minimum Rh
2 of 82% (Figure 1 and Table 2); we required that the nonsynonymous coding SNP S312G (rs605059) be included in the set of htSNPs. These four htSNPs also predicted common haplotypes (>5% frequency) in African Americans, Native Hawaiians, Japanese Americans, and Latinos with a minimum Rh
2 above 80% (Table S1). However, among African Americans, the cumulative frequency of common haplotypes was only 62%. To achieve a cumulative frequency above 70% in African Americans (i.e., to predict an additional two haplotypes with an Rh
2 >80%), additional htSNPs were needed. Because we did not genotype these extra SNPs for this analysis, our current analyses of African Americans are principally informative for haplotypes with frequency above 5%. None of the SNP genotype frequencies showed evidence of deviation from Hardy-Weinberg equilibrium at the 0.001 level among controls in any of the cohorts (stratified by ethnicity).
Figure 1 A Scale Map of the 26 SNPs Genotyped in the MEC Screening Panel and a Plot of the Pattern of Linkage Disequilibrium among Them (in Whites)
The four tag SNPs are markers with arrows, and the block of high linkage disequilibrium and limited haplotype diversity spanning HSD17B1 is highlighted.
Table 2 Characteristics of the htSNPs for HSD17B1
Four htSNP haplotypes had frequencies above 5% in white controls, with a cumulative frequency above 99% (Table 3). Haplotype frequencies were similar for whites across cohorts (Table S2), while some differences in haplotype frequencies were seen among whites, African Americans, Japanese Americans, and Native Hawaiians. Consistent with the greater genetic diversity in African Americans, we found one haplotype (CAAC) that was common only in African Americans (Table 3).
Table 3 Haplotype Frequencies in Controls, by Ethnicity
Global tests of association between HSD17B1 haplotypes and prostate cancer were not significant (likelihood-ratio test [LRT] χ2 = 5.25, 5 d.f., p = 0.39 for analysis using all subjects; LRT χ2 = 6.45, 5 d.f., p = 0.22 for analysis restricted to whites; see Table 4). However, the test for heterogeneity in haplotype odds ratios across ethnicities was significant (LRT χ2 = 44.66, 15 d.f., p < 0.0001). While no haplotype showed evidence of association with prostate cancer risk in African Americans, whites, or Native Hawaiians, haplotype CAGC was significantly associated with decreased prostate cancer risk in Latinos and Japanese Americans (Figure 2; more detailed cohort- and ethnicity-specific results are given in Table S3). The test for heterogeneity across cohorts in haplotype odds ratios among whites was not significant (LRT χ2 = 37.82, 23 d.f., p = 0.03).
Table 4 HSD17B1 Haplotypes and Prostate Cancer Risk, BPC3
Figure 2 Plot of Log Odds Ratios for CAGC (Relative to All Other Haplotypes) under an Additive Model
The boxes are proportional to the inverse of the parameter estimate variance; larger boxes denote more precise estimates. The error bars mark 99% confidence intervals.
Genotype-specific odds ratios for the four SNPs tested are shown in Table 5 for analyses restricted to whites and for analyses pooling all subjects. There was no evidence of an association between the nonsynonymous S312G SNP and prostate cancer (p = 0.40 for analysis using all subjects and p = 0.09 for analysis restricted to whites). None of the other SNPs showed any evidence of association with prostate cancer at the 0.01 level, and none of the SNP odds ratios showed significant evidence of heterogeneity across ethnicity (Table S4).
Table 5 HSD17B1 htSNPs and Prostate Cancer Risk, BPC3
We calculated stratum-specific SNP and haplotype odds ratios for strata defined by family history (at least one first-degree relative diagnosed with prostate cancer versus none), age at time of diagnosis or time of diagnosis of the matched case for controls (≤65 versus >65 years old), and BMI (<25, ≥25 but <30, >30). None of these stratum-specific tests of association were significant at the 0.01 level, and tests for departures from multiplicative interaction model (tests for “statistical interaction”) were also nonsignificant (Tables S5 and S6).
We also found no association between HSD17B1 haplotypes and advanced prostate cancer (Table 6).
Table 6
HSD17B1 Haplotypes and Risk of Advanced Stage (≥C) and High-Grade (Gleason ≥8) Prostate Cancer among All Cases
Discussion
After comprehensively screening HSD17B1 for variation in U.S. and European whites, we found no evidence of association between prostate cancer and common variants in HSD17B1. We observed that haplotype odds ratios for association with prostate cancer differed across ethnicity, with the CAGC haplotype showing a significant (p < 0.01) inverse association with prostate cancer effects in Latinos and Japanese Americans. However, the smaller sample size in these subgroups limits our power to detect an effect of the observed magnitude, leading to an increase in the probability that these results are false positives. We found no evidence that the odds ratios associated with common haplotypes or SNPs differed by cohort (among whites), family history, age, or BMI. We also found no evidence that common variants in HSD17B1 were related to disease severity among cases.
A major advantage of the BPC3 haplotype-tagging approach is that it allows a cost-effective approach to the identification of common susceptibility alleles across the entire gene region. This includes putative functional variants such as nonsynonymous coding SNPs as well as variants of unknown function in intronic and 5′ and 3′ untranslated regions. In the case of HSD17B1, there is evidence that several upstream regions participate in the regulation of HSD17B1 expression [28]. All of these lie well within the region of high linkage disequilibrium spanning HSD17B1; hence, common variants in these regulatory regions are accurately predicted by the four htSNPs analyzed here.
Another major strength of the BPC3 is its unprecedented sample size. With 8,290 cases and 9,367 controls, there is greater than 90% power to detect a dominant or log-additive odds ratio of 1.3 for an allele with 5% frequency at the 0.001 level, even after accounting for loss of effective sample size due to the haplotype-tagging approach. The large sample size of the BPC3 allows adequately powered investigation of differences in genetic effect by established or hypothesized prostate cancer risk factors. For example, there is still greater than 90% power to detect a stratum-specific dominant odds ratio of 1.7 for a 5% frequency variant at the 0.001 level when the stratum consists of only 20% of the total sample.
A limitation of the study is the inability of the assayed htSNPs to adequately capture several haplotypes in African Americans that have a frequency just below 5%, so the cumulative frequency of the haplotypes that are effectively predicted in this group by the htSNPs is only 60%. Moreover, power to detect possible associations between prostate cancer and genetic variation in HSD17B1 unique to nonwhite ethnicities is limited in this study, given the smaller sample sizes available for these groups. For example, power to detect the observed log-additive effect sizes for haplotype CAGC in Japanese Americans at the 0.01 level is approximately 59% (68% for Latinos). Thus, assuming prior probabilities of causality for this HSD17B1 haplotype of 1%, the false-positive probabilities for these associations are 64% for Japanese Americans and 59% for Latinos; assuming a more conservative prior probability of 0.1%, the false-positive probabilities are 95% for Japanese Americans and 94% for Latinos. Further analysis to assess whether HSD17B1 is associated with risk for prostate cancer in nonwhite ethnic groups will require larger samples accumulated through longer follow-up or new cohorts.
Another potential limitation of this study is that results are reported for a single gene. Prostate cancer risk may be a complex function of genotypes across several genes involved in steroid hormone metabolism [37]. For example, a mutation in one gene may only increase risk in the presence of a mutation in a different gene. Although a gene involved in such gene-gene interactions can be discovered using a marginal test (ignoring the other genes) [38], incorporating information about other genes may improve power to detect association [39]. As the BPC3 will eventually measure variation in over 30 genes related to the steroid hormone pathway, it will have the opportunity to investigate the combined contribution of multiple genes to the risk of prostate cancer.
In the present study, the absence of an association between HSD17B1 haplotypes and prostate cancer suggests that we can rule out large or moderate associations between common HSD17B1 variants and risk of prostate cancer among U.S. and European whites. If any variants affect the risk of prostate cancer, they are likely to have small effects or low frequency and are unlikely to contribute significantly to the overall incidence of prostate cancer in these populations. While this sheds some light on the clinical effects of one enzyme involved in the complex process of steroid synthesis and catabolism, it remains to be determined whether variants in other genes in steroidal hormone pathways play a more important role or if the combined effects of several genes within these pathways have a larger impact. The BPC3 plans to investigate these questions by comprehensively measuring variation in more than 30 genes involved in the steroidal hormone pathway and their associated receptors. Last, this study underscores the importance of large, cooperative consortia in evaluating the contribution of germline genetic variation to a common cancer, such as prostate cancer.
Materials and Methods
Study population.
The BPC3 has been described in detail elsewhere (unpublished data). For prostate cancer, the study combines the resources of seven large cohort studies of men: the American Cancer Society Cancer Prevention Study II (ACS CPS-II) [40], the Alpha-Tocopherol, Beta-Carotene Cancer Prevention (ATBC) Study [41], the EPIC Cohort (itself comprising cohorts from Denmark, Great Britain, Germany, Greece, Italy, the Netherlands, Spain, and Sweden) [42], the Health Professionals Follow-up Study (HPFS) [43], the Hawaii/Los Angeles Multi-ethnic Cohort Study (MEC) [44], the Physicians Health Study (PHS) [45], and the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial [46]. With the exception of the MEC, these cohorts are composed predominantly of whites of European descent. We do not have information on ethnicity or ancestry beyond country of residence for EPIC and have classified all EPIC participants as white. We anticipate that the number of EPIC participants of non-European ancestry is small. We plan further population genetic studies will to verify this, using the large number of SNPs the BPC3 will have genotyped. The MEC contributes African American, Latino, Japanese Americans, and Native Hawaiian cases and controls recruited from Los Angeles and Hawaii. The PLCO also includes over 650 African American subjects. We distinguish Spanish EPIC participants from MEC Latino participants, because the latter are principally of Mexican and Central American origin, with origins in European, Native American, and African populations [47,48].
Cases of prostate cancer were identified through population-based cancer registries or self-report confirmed by medical records. The BPC3 data for prostate cancer consist of a series of matched nested case-control studies from each cohort; controls were matched to cases on a number of potentially confounding factors, including age, ethnicity, and region of recruitment. For the current investigation, prostate cancer cases were matched to available controls by age in 5-year intervals, ethnicity, and cohort.
SNP discovery and htSNP selection.
We used a multistage approach to characterize genetic variation in and around HSD17B1 in our large sample of cases and controls. First, HSD17B1 exons in 190 advanced breast and prostate cancer cases were resequenced at the Broad Institute to discover novel coding SNPs. Then we genotyped a set of SNPs across a 42-kb region spanning HSD17B1 in a multiethnic reference sample to determine patterns of linkage disequilibrium and select htSNPs. This set consisted of 25 common (allele frequency >5%) SNPs selected from public databases and one nonsynonymous SNP discovered during resequencing; these 26 SNPs covered the target region at a density of one SNP per 1.6 kb. The target region included the gene N-acetylglucosaminidase alpha (NAGLU) and the pseudogene for HSD17B1 (HSD17BP1), both 5′ of HSD17B1, and coenzyme A synthase (COASY) and transcription factor-like 4 (TCFL4), both 3′ of HSD17B1 (see Figure 1). The reference sample consisted of equal numbers (70 each) of whites, African Americans, Latinos, and Japanese Americans and 69 Native Hawaiians.
We identified a region of high linkage disequilibrium and low haplotype diversity spanning HSD17B1 using the algorithm of Gabriel et al. [49] as implemented in Haploview [50]. Haplotype-tagging SNPs were then chosen in these regions based on Rh
2, a measure of the correlation between observed haplotypes and those predicted on the basis of htSNP genotypes [51]. This approach is based on the observation that within blocks of high linkage disequilibrium and limited haplotype diversity, common SNPs are highly correlated with common haplotypes [49]. Finally, we genotyped these htSNPs in BPC3 cases and controls and tested for association between htSNP haplotypes and disease.
Genotyping.
The 26 SNPs were genotyped in the multiethnic reference panel at the Broad Institute using Sequenom and Illumina platforms. Genotyping of cases and controls was performed in 4 laboratories using a fluorescent 5′ endonuclease assay and ABI PRISM 7900 for sequence detection (TaqMan; Applied Biosystems, Foster City, California, United States). Based on sequence information, TaqMan assays were designed for each SNP and synthesized in four separate batches of 12,000 reactions for the roughly 48,000 needed to complete the study of HSD17B1 in BPC3 breast and prostate cancer samples. Initial quality control was performed at the manufacturer (Applied Biosystems); an additional 500 test reactions were run by the Cohort Consortium on the multiethnic reference panel; greater than 99.5% concordance was observed across genotyping platforms. (Assay characteristics for the four htSNPs for HSD17B1 are available on the public Website: http://www.uscnorris.com/mecgenetics/CohortGCKView.aspx.) Sequence validation for each SNP assay was performed and 100% concordance observed (http://snp500cancer.nci.nih.gov) [52]. To assess interlaboratory variation, each center ran assays on a designated set of 94 samples from the SNP 500 cancer panel, showing completion and concordance rates of greater than 99% [52]. The internal quality of genotype data at each center was assessed by 5% to 10% blinded samples in duplicate or triplicate (depending on study).
Statistical analysis.
For each SNP, we used conditional logistic regression to simultaneously estimate the odds ratio for disease associated with carrying one copy of the minor allele relative to carrying no copies and the odds ratio associated with carrying two copies relative to carrying no copies. We estimated haplotype-specific odds ratios using an expectation-substitution approach to account for haplotype uncertainty given unphased genotype data [53,54]. Haplotype frequencies and subject-specific expected haplotype indicators were calculated separately for each cohort (and country or ethnicity within cohort). To test the global null hypothesis of no association between variation in HSD17B1 haplotypes and risk of prostate cancer, we used an LRT comparing a model with additive effects on the log odds scale for each common haplotype (treating the most common haplotype as the referent) to the intercept-only model. We considered haplotypes with greater than 5% frequency in at least one cohort or ethnic group to be “common.” All other haplotypes were pooled into a separate “rare haplotypes” category.
Although the matched analysis accounts for heterogeneity in risk-factor prevalence across study, we also tested for heterogeneity in odds ratio estimates across studies that might result from slightly different matching criteria or case definitions using an LRT. We also tested for heterogeneity in odds ratio estimates across ethnicity using an LRT that compares the model with common additive effects for each haplotype (except the referent) to the model with distinct additive effects for each ethnicity where the expected numbers of the haplotype in cases and controls under the null were above five. Thus, the three common haplotypes among Native Hawaiians contributed two ethnicity-specific haplotype effects; the five common haplotypes and the pooled rare haplotypes among African Americans contribute five. To assess whether other risk factors for prostate cancer modify the association with haplotype, we calculated risk stratum–specific odds ratios and tested for departures from a multiplicative interaction model. We performed case-only analyses to test for association between HSD17B1 variants and advanced prostate cancer (as defined above).
We calculated 99% confidence intervals and test for significance associations at the 0.01 level to minimize the chance of both false-positive and false-negative results. An upper bound on the probability of a false positive was estimated roughly as α(1 – π)/[α(1 – π) + π(1 – β)], where π is the prior probability that a variant has a relative risk of R or greater, α is the test size, and 1 – β is the power [20]. Our study has greater than 99% power to detect a dominant or log-additive odds ratio of 1.3 for an allele with 5% frequency at the 0.001 level. Thus, when α = 0.01, the probability of a false positive is 8% for the very optimistic prior probability of a 10% chance that HSD17B1 is associated with prostate cancer. The false-negative report probability, defined as β π/[(1 – α)(1 – π) + π β], is only 0.1% in this situation. For a prior probability of 1 in 100, the false-positive and false-negative report probabilities are 50% and 0.01%, respectively. Thus, for a range of priors, the probability that we would fail to reject at the .01 level if HSD17B1 were truly associated with disease is small. Power for individual SNPs and haplotypes was calculated using Quanto (http://hydra.usc.edu/gxe/), assuming an effective sample size of N Rh
2 to adjust for the loss in power inherent in genotyping surrogate tagging SNPs. Here N is the nominal sample size and Rh
2 is the design threshold of 0.8; this is somewhat conservative, as the achieved Rh
2 can be well above the threshold.
Supporting Information
Table S1 Haplotype Frequencies and htSNP Performance in MEC Reference Sample
(61 KB DOC)
Click here for additional data file.
Table S2 htSNP Haplotype Frequencies by Cohort Among Whites and African Americans.
(106 KB DOC)
Click here for additional data file.
Table S3 Tests of Haplotype-Prostate Cancer Association and Haplotype Odds Ratios, by Study (Whites) and Ethnicity
(439 KB DOC)
Click here for additional data file.
Table S4 Tests of Association between Individual htSNPs and Prostate Cancer and Odds Ratio Estimates, by Study (Whites) and Ethnicity
(175 KB DOC)
Click here for additional data file.
Table S5 Tests of Haplotype-Prostate Cancer Association and Haplotype Odds Ratios, Stratified by Age, Family History of Prostate Cancer, and BMI
(346 KB DOC)
Click here for additional data file.
Table S6 Tests of Between-Individual htSNPs and Prostate Cancer and Odds Ratio Estimates, Stratified by Age, Family History of Prostate Cancer, and BMI
(285 KB DOC)
Click here for additional data file.
Accession Numbers
The OMIM (http://www.ncbi.nlm.nih.gov/OMIM) accession numbers for genes mentioned in this paper are AR (313700), BRCA1 (113705), BRCA2 (600185), CYP3A4*1B (124010), ELAC2 (605367), HSD17B1 (109684), RNASEL (180435), MSR1 (153622), NAGLU (252920), SRD5A2 (607306), and TCFL4 (602976). The HGNC (http://www.gene.ucl.ac.uk/cgi-bin/nomenclature/searchgenes.pl) accession number for COASY is 29932.
The authors gratefully acknowledge the participants in the component cohort studies and express sincere gratitude to the investigators involved in the recruitment and follow-up of the EPIC cohorts: Jakob Linsisen, Division of Clinical Epidemiology, German Cancer Research Centre, Heidelberg, Germany; Vittorio Krogh, Epidemiology Unit, National Cancer Institute, Milan, Italy; Rosario Tumino, Cancer Registry Azienda Ospedaliera “Civile M P Arezzo,” Ragusa, Italy; Paolo Vineis, Environmental Epidemiology, Imperial College London, London, United Kingdom; Carmen Martinez-Garcia, Andalusian School of Public Health, Granada, Spain; Carmen Navarro, Miguel Rodriguez-Barranco, Department of Epidemiology, Murcia Regional Health Council, Murcia, Spain; Miren Dorronsoro, Department of Public Health of Guipuzcoa, San Sebastian, Spain; Sheila Bingham, Medical Research Council Dunn Nutrition Unit, Cambridge, United Kingdom; Goran Berglund, Malmo Diet and Cancer Study, Department of Medicine, Lund University, Sweden; and Anne Tjonneland, Institute of Cancer Epidemiology, Danish Cancer Society, Copenhagen, Denmark. We also express gratitude to the investigators involved in the recruitment and follow-up of the PLCO cohort: Philip C. Prorok, Division of Cancer Prevention, National Cancer Institute, Bethesda, Maryland, United States of America; Mona Fouad, University of Alabama at Birmingham, Birmingham, Alabama, United States of America; Paul A. Kvale, Henry Ford Health System, Detroit, Michigan, United States of America; Lance Yokochi, Pacific Health Research Institute, Honolulu, Hawaii, United States of America; Douglas Reding, Marshfield Clinic Research Foundation, Marshfield, Wisconsin, United States of America; Timothy R. Church, University of Minnesota, Minneapolis, Minnesota, United States of America; Joel L. Weissfeld, University of Pittsburgh, Pittsburgh, Pennsylvania, United States of America; Saundra Buys, University of Utah, Salt Lake City, Utah, United States of America; Thomas M. Beck, Mountain States Tumor Institute, University of Utah, Boise, Idaho, United States of America; and Edward P. Gelmann, Georgetown University Medical Center, Washington, District of Columbia, United States of America.
The authors acknowledge the expert contributions of Hardeep Ranu, Craig Labadie, Lisa Cardinale, and Shamika Ketkar at Harvard University; William Modi, Merideth Yeager, Robert Welch, Cynthia Glaser, and Laurie Burdett at the National Cancer Institute; and Loreall Pooler at the University of Southern California. This study was supported by NCI cooperative agreements UO1-CA98233, UO1-CA98710, UO1-CA98216, and UO1-CA98758.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. M. J. Thun, E. Riboli, S. Chanock, D. Albanes, D. Hunter, R. B. Hayes, B. Henderson, and D. Stram formed the BPC3 Publications Committee. D. Albanes, L. Kolonel, R. B. Hayes, H. Boeing, B. Bueno-de-Mesquita, E. E. Calle, H. S. Feigelson, J. M. Gaziano, E. Giovannucci, C. A. Gonzalez, G. Hallmans, B. Henderson, T. Key, L. Le Marchand, J. Ma, K. Overvad, D. Palli, E. Riboli, C. Rodriguez, M. Stampfer, M. J. Thun, A. Trichopoulou, and J. Virtamo performed cohort recruitment and follow-up. S. Chanock, D. Stram, C. Haiman, D. Altshuler, M. Freedman, J. Hirschhorn, N. Burtt, G. Thomas, and H. Cann conducted gene sequencing and haplotype construction. D. J. Hunter, A. Dunning, S. Chanock, L. Le Marchand, C. Haiman, J. Hirschhorn, N. Burtt, and G. Thomas coordinated the genotyping. M. J. Thun, E. E. Calle, H. S. Feigelson, Y. C. Chen, P. Kraft, D. J. Hunter, J. Ma, R. Travis, S. Chanock, R. B. Hayes, B. Henderson, D. Stram, C. Haiman, W. Setiawan, D. Altshuler, M. Freedman, and J. Hirschhorn pooled, managed, and analyzed data. P. Kraft, R. Kaaks, P. Pharoah, S. Wacholder, D. Stram, M. Pike, and G. Thomas developed statistical methodology and oversaw analysis. P. Kraft, P. Pharoah, S. Chanock, D. Albanes, L. Kolonel, and R. B. Hayes wrote the paper.
A previous version of this article appeared as an Early Online Release on October 21, 2005 (DOI: 10.1371/journal.pgen.0010068.eor).
Abbreviations
BMIbody mass index
BPC3Breast and Prostate Cancer Cohort Consortium
17β-HSD-117β-hydroxysteroid dehydrogenase 1
htSNPhaplotype-tagging single nucleotide polymorphism
LRTlikelihood-ratio test
SNPsingle nucleotide polymorphism
Note Added in Proof
The Breast and Prostate Cancer Cohort Consortium (BPC3) study, cited in this paper as unpublished data, is now in press [55]
==== Refs
References
Schaid DJ 2004 The complex genetic epidemiology of prostate cancer Hum Mol Genet 13 (Spec No 1) R103 R121 14749351
Simard J Dumont M Labuda D Sinnett D Meloche C 2003 Prostate cancer susceptibility genes: Lessons learned and challenges posed Endocr Relat Cancer 10 225 259 12790786
Bruner DW Moore D Parlanti A Dorgan J Engstrom P 2003 Relative risk of prostate cancer for men with affected relatives: Systematic review and meta-analysis Int J Cancer 107 797 803 14566830
Houlston R Peto J 2004 Genetics and the common cancers Eeles R Ponder B Easton D Eng C Genetic predisposition to cancer London Chapman & Hall 235 247
Lichtenstein P Holm N Verkasalo P Iliadou A Kaprio J 2000 Environmental and heritable factors in the causation of cancer—Analyses of cohorts of twins from Sweden, Denmark and Finland N Engl J Med 343 78 85 10891514
Angele S Falconer A Edwards SM Dork T Bremer M 2004 ATM polymorphisms as risk factors for prostate cancer development Br J Cancer 91 783 787 15280931
Thompson D Easton D 2002 Cancer incidence in BRCA1 mutation carriers J Natl Cancer Inst 94 1358 1365 12237281
Eaton N Reeves G Appleby R Key T 1999 Endogenous sex hormones and prostate cancer: A quantitative review of prospective studies Br J Cancer 80 930 934 10362098
Gann PH Hennekens CH Ma J Longcope C Stampfer MJ 1996 Prospective study of sex hormone levels and risk of prostate cancer J Natl Cancer Inst 88 1118 1126 8757191
Makridakis NM Reichardt JK 2001 Molecular epidemiology of hormone-metabolic loci in prostate cancer Epidemiol Rev 23 24 29 11588850
Coughlin SS Hall IJ 2002 A review of genetic polymorphisms and prostate cancer risk Ann Epidemiol 12 182 196 11897176
Makridakis NM Reichardt JK 2004 Molecular epidemiology of androgen-metabolic loci in prostate cancer: Predisposition and progression J Urol 171 S25 S28 discussion S28 S29 14713749
Makridakis NM Ross RK Pike MC Crocitto LE Kolonel LN 1999 Association of mis-sense substitution in SRD5A2 gene with prostate cancer in African-American and Hispanic men in Los Angeles, USA Lancet 354 975 978 10501358
Ntais C Polycarpou A Ioannidis JP 2003 SRD5A2 gene polymorphisms and the risk of prostate cancer: A meta-analysis Cancer Epidemiol Biomarkers Prev 12 618 624 12869400
Jaffe JM Malkowicz SB Walker AH MacBride S Peschel R 2000 Association of SRD5A2 genotype and pathological characteristics of prostate tumors Cancer Res 60 1626 1630 10749132
Tut TG Ghadessy FJ Trifiro MA Pinsky L Yong EL 1997 Long polyglutamine tracts in the androgen receptor are associated with reduced trans-activation, impaired sperm production, and male infertility J Clin Endocrinol Metab 82 3777 3782 9360540
Freedman ML Pearce CL Penney KL Hirschhorn JN Kolonel LN 2005 Systematic evaluation of genetic variation at the androgen receptor locus and risk of prostate cancer in a multiethnic cohort study Am J Hum Genet 76 82 90 15570555
Zeigler-Johnson C Friebel T Walker AH Wang Y Spangler E 2004 CYP3A4, CYP3A5, and CYP3A43 genotypes and haplotypes in the etiology and severity of prostate cancer Cancer Res 64 8461 8467 15548719
Chang BL Zheng SL Hawkins GA Isaacs SD Wiley KE 2002 Joint effect of HSD3B1 and HSD3B2 genes is associated with hereditary and sporadic prostate cancer susceptibility Cancer Res 62 1784 1789 11912155
Wacholder S Chanock S Garcia-Closas M El Ghormli L Rothman N 2004 Assessing the probability that a positive report is false: An approach for molecular epidemiology studies J Natl Cancer Inst 96 434 442 15026468
Labrie F Luu-The V Lin SX Labrie C Simard J 1997 The key role of 17 beta-hydroxysteroid dehydrogenases in sex steroid biology Steroids 62 148 158 9029730
Waxman DJ 1996 Role of metabolism in the activation of dehydroepiandrosterone as a peroxisome proliferator J Endocrinol 150 (Suppl) S129 S147 8943797
Mindnich R Moller G Adamski J 2004 The role of 17 beta-hydroxysteroid dehydrogenases Mol Cell Endocrinol 218 7 20 15130507
Arnold JT Le H McFann KK Blackman MR 2005 Comparative effects of DHEA vs. testosterone, dihydrotestosterone, and estradiol on proliferation and gene expression in human LNCaP prostate cancer cells Am J Physiol Endocrinol Metab 288 E573 E584 15536203
Ross RK Makridakis NM Reichardt JK 2003 Prostate cancer: Epidemiology and molecular endocrinology Henderson BE Ponder B Ross RK Hormones, genes, and cancer Oxford Oxford University Press
Peltoketo H Isomaa VV Ghosh D Vihko P 2003 Estrogen metabolism genes: HSD17B1 and HSD17B2 Henderson BE Ponder B Ross RK Hormones, genes, and cancer Oxford Oxford University Press
Martel C Rheaume E Takahashi M Trudel C Couet J 1992 Distribution of 17 beta-hydroxysteroid dehydrogenase gene expression and activity in rat and human tissues J Steroid Biochem Mol Biol 41 597 603 1314080
Peltoketo H Nokelainen P Piao YS Vihko R Vihko P 1999 Two 17beta-hydroxysteroid dehydrogenases (17HSDs) of estradiol biosynthesis: 17HSD type 1 and type 7 J Steroid Biochem Mol Biol 69 431 439 10419022
Carruba G Adamski J Calabro M Miceli MD Cataliotti A 1997 Molecular expression of 17 beta hydroxysteroid dehydrogenase types in relation to their activity in intact human prostate cancer cells Mol Cell Endocrinol 131 51 57 9256363
Castagnetta LA Carruba G Traina A Granata OM Markus M 1997 Expression of different 17beta-hydroxysteroid dehydrogenase types and their activities in human prostate cancer cells Endocrinology 138 4876 4882 9348218
Elo JP Akinola LA Poutanen M Vihko P Kyllonen AP 1996 Characterization of 17beta-hydroxysteroid dehydrogenase isoenzyme expression in benign and malignant human prostate Int J Cancer 66 37 41 8608963
Miettinen MM Mustonen MV Poutanen MH Isomaa VV Vihko RK 1996 Human 17 beta-hydroxysteroid dehydrogenase type 1 and type 2 isoenzymes have opposite activities in cultured cells and characteristic cell- and tissue-specific expression Biochem J 314 (Pt 3) 839 845 8615778
Wu AH Seow A Arakawa K Van Den Berg D Lee HP 2003 HSD17B1 and CYP17 polymorphisms and breast cancer risk among Chinese women in Singapore Int J Cancer 104 450 457 12584742
Setiawan VW Hankinson SE Colditz GA Hunter DJ De Vivo I 2004 HSD17B1 gene polymorphisms and risk of endometrial and breast cancer Cancer Epidemiol Biomarkers Prev 13 213 219 14973105
Mannermaa A Peltoketo H Winqvist R Ponder BA Kiviniemi H 1994 Human familial and sporadic breast cancer: Analysis of the coding regions of the 17 beta-hydroxysteroid dehydrogenase 2 gene (EDH17B2) using a single-strand conformation polymorphism assay Hum Genet 93 319 324 8125484
Feigelson H Coetzee G Kolonel L Ross R Henderson B 1997 A polymorphism in the CYP17 gene increases the risk of breast cancer Cancer Res 57 1063 1065 9067272
Thomas DC 2005 The need for a systematic approach to complex pathways in molecular epidemiology Cancer Epidemiol Biomarkers Prev 14 557 559 15767327
Clayton D McKeigue PM 2001 Epidemiological methods for studying genes and environmental factors in complex diseases Lancet 358 1356 1360 11684236
Marchini J Donnelly P Cardon LR 2005 Genome-wide strategies for detecting multiple loci that influence complex diseases Nat Genet 37 413 417 15793588
Calle EE Rodriguez C Jacobs EJ Almon ML Chao A 2002 The American Cancer Society Cancer Prevention Study II Nutrition Cohort: Rationale, study design, and baseline characteristics Cancer 94 500 511 11900235
Buring JE Hebert P Hennekens CH 1994 The alpha-tocopherol, beta-carotene lung cancer prevention trial of vitamin E and beta-carotene: The beginning of the answers Ann Epidemiol 4 75 8205274
Riboli E Hunt KJ Slimani N Ferrari P Norat T 2002 European Prospective Investigation into Cancer and Nutrition (EPIC): Study populations and data collection Public Health Nutr 5 1113 1124 12639222
Giovannucci E Pollak M Liu Y Platz EA Majeed N 2003 Nutritional predictors of insulin-like growth factor I and their relationships to cancer in men Cancer Epidemiol Biomarkers Prev 12 84 89 12582016
Kolonel LN Altshuler D Henderson BE 2004 The multiethnic cohort study: Exploring genes, lifestyle and cancer risk Nat Rev Cancer 4 519 527 15229477
Chan JM Stampfer MJ Ma J Gann P Gaziano JM 2002 Insulin-like growth factor-I (IGF-I) and IGF binding protein-3 as predictors of advanced-stage prostate cancer J Natl Cancer Inst 94 1099 1106 12122101
Hayes RB Reding D Kopp W Subar AF Bhat N 2000 Etiologic and early marker studies in the prostate, lung, colorectal and ovarian (PLCO) cancer screening trial Control Clin Trials 21 349S 355S 11189687
Bonilla C Parra EJ Pfaff CL Dios S Marshall JA 2004 Admixture in the Hispanics of the San Luis Valley, Colorado, and its implications for complex trait gene mapping Ann Hum Genet 68 (Pt 2) 139 153 15008793
Bonilla C Gutierrez G Parra EJ Kline C Shriver MD 2005 Admixture analysis of a rural population of the state of Guerrero, Mexico Am J Phys Anthropol Aug 23 [Epub ahead of print].
Gabriel SB Schaffner SF Nguyen H Moore JM Roy J 2002 The structure of haplotype blocks in the human genome Science 296 2225 2229 12029063
Barrett JC Fry B Maller J Daly MJ 2005 Haploview: Analysis and visualization of LD and haplotype maps Bioinformatics 21 263 265 15297300
Stram D Haiman C Hirschhorn J Altshuler D Kolonel L 2003 Choosing haplotype-tagging SNPs based on unphased genotype data using as preliminary sample of unrelated subjects with an example from the multiethnic cohort study Hum Hered 55 27 36 12890923
Packer BR Yeager M Staats B Welch R Crenshaw A 2004 SNP500Cancer: A public resource for sequence validation and assay development for genetic variation in candidate genes Nucleic Acids Res 32 (Database issue) D528–D532
Kraft P Cox D Paynter R Hunter D De Vivo I 2005 Accounting for haplotype uncertainty in association studies: A comparison of simple and flexible techniques Genet Epidemiol 28 261 272 15637718
Zaykin D Westfall P Young S Karnoub M Wagner M 2002 Testing association of statistically inferred haplotypes with discrete and continuous traits in samples of unrelated individuals Hum Hered 53 79 91 12037407
The Breast and Prostate Cancer Cohort Consortium 2005 The NCI Cohort Consortium on breast and prostate cancer: Rationale and design Nat Rev Cancer: In press
|
16311626
|
PMC1287955
|
CC0
|
2021-01-05 07:59:37
|
no
|
PLoS Genet. 2005 Nov 25; 1(5):e68
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010068
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1631162710.1371/journal.pgen.0010069plge-01-05-12Research ArticleIncreased Life Span due to Calorie Restriction in Respiratory-Deficient Yeast Calorie Restriction in YeastKaeberlein Matt 1*Hu Di 2Kerr Emily O 2Tsuchiya Mitsuhiro 2Westman Eric A 2Dang Nick 2Fields Stanley 13Kennedy Brian K 2*1 Departments of Genome Sciences and Medicine, University of Washington, Seattle, Washington, United States of America
2 Department of Biochemistry, University of Washington, Seattle, Washington, United States of America
3 Howard Hughes Medical Institute, University of Washington, Seattle, Washington, United States of America
Kim Stuart EditorStanford University School of Medicine, United States of America*To whom correspondence should be addressed. E-mail: [email protected] (MK), [email protected] (BKK)11 2005 25 11 2005 25 10 2005 1 5 e697 7 2005 24 10 2005 Copyright: © 2005 Kaeberlein 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.A model for replicative life span extension by calorie restriction (CR) in yeast has been proposed whereby reduced glucose in the growth medium leads to activation of the NAD+–dependent histone deacetylase Sir2. One mechanism proposed for this putative activation of Sir2 is that CR enhances the rate of respiration, in turn leading to altered levels of NAD+ or NADH, and ultimately resulting in enhanced Sir2 activity. An alternative mechanism has been proposed in which CR decreases levels of the Sir2 inhibitor nicotinamide through increased expression of the gene coding for nicotinamidase, PNC1. We have previously reported that life span extension by CR is not dependent on Sir2 in the long-lived BY4742 strain background. Here we have determined the requirement for respiration and the effect of nicotinamide levels on life span extension by CR. We find that CR confers robust life span extension in respiratory-deficient cells independent of strain background, and moreover, suppresses the premature mortality associated with loss of mitochondrial DNA in the short-lived PSY316 strain. Addition of nicotinamide to the medium dramatically shortens the life span of wild type cells, due to inhibition of Sir2. However, even in cells lacking both Sir2 and the replication fork block protein Fob1, nicotinamide partially prevents life span extension by CR. These findings (1) demonstrate that respiration is not required for the longevity benefits of CR in yeast, (2) show that nicotinamide inhibits life span extension by CR through a Sir2-independent mechanism, and (3) suggest that CR acts through a conserved, Sir2-independent mechanism in both PSY316 and BY4742.
Synopsis
Calorie restriction slows aging and increases life span in nearly every organism studied. The mechanism by which this occurs is one of the most important unanswered questions in biogerontology. One popular theory, based on work from the budding yeast Saccharomyces cerevisiae, proposes that calorie restriction works by causing a metabolic shift toward increased mitochondrial respiration, resulting in activation of a family of proteins known as Sirtuins. This study demonstrates that life span extension by calorie restriction does not require respiration and occurs even in cells completely lacking mitochondrial DNA. Interestingly, calorie restriction protects yeast cells against a severe longevity defect associated with absence of mitochondrial DNA, suggesting the possibility that the consequences of age-associated mitochondrial dysfunction might be alleviated or prevented by calorie restriction.
Citation:Kaeberlein M, Hu D, Kerr EO, Tsuchiya M, Westman EA, et al. (2005) Increased life span due to calorie restriction in respiratory-deficient yeast. PLoS Genet 1(5): e69.
==== Body
Introduction
Calorie restriction (CR) has been shown to slow aging in evolutionarily divergent species, including yeast, worms, flies, and rodents [1–5]. In addition to increasing longevity, CR is reported to cause additional phenotypes, including increased resistance to oxidative stress [6–8], enhanced DNA damage repair [9,10], decreased levels of oxidatively damaged proteins [11–13], improved glucose homeostasis and insulin sensitivity [14–16], altered levels of apoptosis [17], and delayed onset of a number of age-related diseases [18–21]. Although it has been known for more than 70 y that calorie restriction can increase life span in mammals [22], a mechanistic understanding of this phenomenon has remained elusive. It seems clear that nutrient and growth factor responsive pathways, such as those mediated by insulin, IGF-1, TOR, and Akt, are likely to represent important conduits through which these signals affect the aging rate. CR mediates enhancement of stress response pathways in mammals [23,24], and signaling through the insulin-like pathway in worms coordinates expression of a variety of antioxidant, chaperone, and anti-bacterial stress response proteins [25–27]. Similarly, TOR-mediated regulation of translational machinery appears to play a role in the response to nutrient deprivation in yeast [28], worms [29,30], flies [31], and mammals [32]. Finally, models postulating a role for Sir2-like protein deacetylases in CR-mediated life span extension have gained popularity for yeast [33], flies [34], and mammals, as well [4,35].
In the budding yeast Saccharomyces cerevisiae, CR can be imposed by reducing the concentration of glucose in the growth medium, resulting in a 20%–40% increase in replicative life span in multiple strain backgrounds [33,36,37]. In addition, genetic models of CR include deletion of the gene coding for hexokinase, HXK2, and mutations that decrease signaling through the cAMP-dependent protein kinase, PKA, such as deletion of the genes coding for the glucose sensing proteins GPA2 or GPR1, and temperature-sensitive alleles of adenylate cyclase (cdc35–1) or the RAS-associated GTPase (cdc25–10) [33].
CR has been proposed to increase yeast replicative life span by a mechanism involving activation of Sir2 [33], an NAD+–dependent histone deacetylase [38–40] that inhibits the formation of extrachromosomal rDNA circles (ERCs) [41]. ERCs are self-replicating DNA molecules that accumulate in the mother-cell nucleus with age and are thought to cause senescence [42]. Overexpression of Sir2 increases life span in multiple strain backgrounds [36,41,43], and deletion of Sir2 shortens life span by about 50% [41,44]. CR fails to increase the life span of short-lived sir2Δ cells [33], consistent with the idea that CR could be acting by a mechanism involving Sir2.
The question of how CR might activate Sir2 has been a source of considerable controversy [45]. Saccharomyces cerevisiae is a facultative anaerobe that, under standard laboratory growth conditions (2% glucose), generates ATP largely by fermentation. Under conditions of reduced glucose, such as CR, S. cerevisiae shifts from fermentation to respiration, resulting in increased transcription of respiratory genes and a higher rate of oxygen consumption [46]. In models put forth by Lin et al., this metabolic shift results in activation of Sir2, either through increased cellular NAD+ [46] or decreased cellular NADH [47]. Alternatively, Anderson et al. have reported that CR does not alter NAD+ levels [48], but leads to enhanced expression of PNC1 and a reduction in cellular nicotinamide [49]. Since nicotinamide is an inhibitor of the Sir2 deacetylation reaction, its decreased concentration could result in enhanced Sir2 activity [50,51]. Overexpression of Pnc1 suppresses the effect of exogenously added nicotinamide on Sir2-dependent silencing at HM loci, telomeres, and rDNA [52]; there are conflicting reports, however, on whether Pnc1 overexpression alters Sir2 activity at endogenous levels of nicotinamide [49,52].
More recently, we have questioned the importance of Sir2 in life span extension by CR [28,53]. In a long-lived strain background, BY4742, CR increases life span to a greater extent in cells lacking both Sir2 and the replication fork barrier protein Fob1 than in wild-type cells [36]. Based on this observation, and the fact that deletion of SIR2 shortens life span by approximately 50%, we proposed a model whereby the inability of CR to increase life span in sir2Δ FOB1 cells is explained as an indirect effect, resulting from the hyperaccumulation of ERCs [36]. Deletion of FOB1 in a sir2Δ background suppresses the hyperaccumulation of ERCs, as well as the longevity defect [41], thus allowing CR to dramatically increase the life span of cells in the absence of Sir2 [36].
Much of the early work suggesting a link between CR and Sir2 was carried out in the PSY316 strain background [33,37,46,47,49,50,54], a strain in which, paradoxically, overexpression of Sir2 fails to increase life span [55]. Guarente and Picard [35] have proposed that CR might act via different mechanisms in the BY4742 and PSY316 strain backgrounds. In order to further clarify the molecular mechanism(s) underlying replicative life span extension by calorie restriction in yeast, we have sought to directly test key components of the models for Sir2-dependent CR in both of these strains. Here, we examine in detail the role of respiratory metabolism in life span extension by CR, finding that (1) respiration is not required for life span extension by CR; and (2) CR suppresses the enhanced early mortality, only apparent in PSY316, due to loss of mitochondrial DNA. In contrast, exogenous addition of nicotinamide partially, but not completely, blocks Sir2-independent life span extension by CR.
Results
Life Span Extension by CR Is Independent of Respiration in BY4742
A central facet of the Sir2-dependent models for life span extension by CR is that a metabolic shift from fermentation to respiration in response to CR results in activation of Sir2 [46,47]. Since Sir2 is not required for life span extension by CR in BY4742 [36], we wished to determine whether respiration was also dispensable. The effect of CR in the absence of respiratory capacity was examined using rho0 cells, which completely lack mitochondrial DNA. Rho0 yeast lack three cytochrome c oxidase subunits (COX1, COX2, and COX3), three ATP synthase subunits (ATP6, ATP8, and ATP9), and apocytochrome b (CYTB), and are therefore incapable of respiratory metabolism [56,57]. BY4742 rho0 cells were generated (see Materials and Methods), and the absence of mitochondrial DNA was verified by staining cells with DAPI (Figure 1A). Lack of respiratory capacity in rho0 cells was confirmed by inability to grow on the non-fermentable carbon source, glycerol (Figure 1B). As previously observed [58], replicative life span under standard conditions (2% glucose) was not altered by loss of mitochondrial DNA in this strain (Figure 1C). CR (0.05% glucose) significantly enhanced the life span of both wild-type and rho0 cells to a comparable degree. Thus, respiration is not required for life span extension by CR in the long-lived BY4742 strain background.
Figure 1 Respiration Is Not Required for Life Span Extension by CR in BY4742
(A) BY4742 rho0 strains lack mitochondrial DNA. DAPI staining of BY4742 (i) wild-type or (ii) rho0 cells grown under standard conditions (2% glucose) and calorie-restricted (iii) wild-type or (iv) rho0 cells (CR = 0.05% glucose).
(B) BY4742 rho0 strains are unable to grow on glycerol as the sole carbon source. (i) BY4742 wild-type or (ii) rho0 cells on YEP supplemented with 2% glucose or 3% glycerol.
(C) CR increase life span in BY4742 rho0 cells. Replicative life span analysis for BY4742 wild-type and rho0 cells on 2% glucose and 0.05% glucose (CR). Mean life span is shown in parentheses.
Life Span Extension by CR Is Independent of Respiration in PSY316
Previous work indicating that life span extension by CR is dependent on respiration was carried out in the PSY316 strain background, in which it was reported that deletion of the gene coding for cytochrome c1, CYT1, prevents life span extension by CR [46]. In order to address whether life span extension by CR in mutants incapable of respiratory metabolism is specific to BY4742, we generated respiratory-deficient cyt1Δ and rho0 variants in the PSY316 background (Figure 2A and 2B) and measured life span on medium containing either 2% or 0.05% glucose. As in BY4742 rho0 cells, CR significantly increased both mean and maximum life span in PSY316 rho0 (Figure 2C) and cyt1Δ cells (Figure 2D).
Figure 2 Respiration Is Not Required for Life Span Extension by CR in PSY316
(A) PSY316AUT rho0 strains lack mitochondrial DNA. DAPI staining of PSY316 (i) wild-type or (ii) rho0 cells grown under standard conditions (2% glucose) and calorie-restricted (iii) wild-type or (iv) rho0 cells (CR = 0.05% glucose).
(B) PSY316AUT rho0 strains are unable to grow on glycerol as the sole carbon source. (i) PSY316AUT wild-type, (ii) cyt1Δ rho0, (iii) cyt1Δ, or (iv) rho0 cells on YEP supplemented with 2% glucose or 3% glycerol.
(C) CR increases life span in PSY316AUT rho0 cells. Replicative life span analysis for PSY316AUT wild-type and rho0 cells on 2% glucose and 0.05% glucose (CR). Mean life span is shown in parentheses.
(D) CR increases the life span of cyt1Δ cells. Replicative life span analysis for PSY316AUT wild-type and cyt1Δ cells on 2% glucose and 0.05% glucose (CR). Mean life span is shown in parentheses.
(E) CR increases the life span of cyt1Δ rho0 cells. Replicative life span analysis for PSY316AUT wild-type and cyt1Δ rho0 cells on 2% glucose and cyt1Δ rho0 cells on 0.05% glucose (CR). Mean life span is shown in parentheses.
Unlike the case in BY4742 in which deletion of mitochondrial DNA has no effect on life span, PSY316 rho0 variants exhibited a profound increase in early mortality under standard growth conditions. This phenotype was not observed in cyt1Δ cells, indicating that loss of mitochondrial DNA, rather than general respiratory deficiency, is responsible for the life span defect. Deletion of CYT1 in cells lacking mitochondrial DNA, however, resulted in a short life span comparable to that of rho0 cells (Figure 2E), demonstrating that rho0 is epistatic to cyt1Δ for this phenotype. As observed for rho0 cells, CR more than doubled the short life span of cyt1Δ rho0 cells, which contain both nuclear and mitochondrial mutations that prevent respiration.
Our observation that CR increased life span in PSY316 cells lacking CYT1 is in contrast to a previous report [46]. A potential explanation for this difference is that 0.5% glucose was used for CR in the prior study, rather than the 0.05% glucose concentration used in this study. We therefore measured the life span of respiratory deficient rho0 and cyt1Δ cells grown on 0.5% glucose. As we observed for cells grown on 0.05% glucose, growth on 0.5% glucose increased the life span of rho0 and cyt1Δ cells, although the magnitude of life span extension is reduced relative to 0.05% glucose (Figure 3A). Thus, the use of a non-optimal level of CR may have precluded detection of life span extension by CR in cyt1Δ mutants in the prior study.
Figure 3 Effect of Glucose Concentration on Life Span and Sir2 Activity in Respiratory-Deficient Mutants
(A) Mean replicative life span is significantly increased in PSY316 AUT cyt1Δ and rho0 cells as the glucose concentration is decreased to either 0.5% or 0.05%, relative to life span on 2% glucose. *p < 0.05, **p < 0.01.
(B) Sir2 activity is not increased by CR but is responsive to increased expression of Sir2 or to addition of exogenous nicotinamide. Transcriptional silencing of the telomeric URA3 marker in PSY316AUT (WT) was monitored by the survival of cells plated onto medium containing 5-FOA.
(C) Sir2 activity is not altered in respiratory deficient cyt1Δ cells and is unaffected by CR. Transcriptional silencing of the telomeric URA3 marker in PSY316AUT (WT) was monitored by the survival of cells plated onto medium containing 5-FOA.
We also examined the effect of CR on Sir2 activity in respiratory-deficient PSY316 cells. The PSY316AUT variant has both URA3 and ADE2 marker genes integrated near telomeres, thus allowing for efficient determination of Sir2-dependent telomeric silencing in response to genetic or environmental perturbations [55]. As previously reported, increased Sir2 activity can be achieved by overexpression of SIR2 in the PSY316 strain background [55], significantly enhancing telomere silencing and survival on FOA, while inhibition of Sir2 by addition of 5 mM nicotinamide to the medium decreased telomere silencing (Figure 3B). CR, however, had no detectable effect on Sir2-dependent silencing at telomeres. Similarly, respiratory deficiency caused by deletion of CYT1 also fails to impact Sir2-dependent silencing at either 2% or 0.05% glucose (Figure 3C). A decrease in survival on FOA was observed in rho0 cells relative to wild-type or cyt1Δ cells at 2% glucose (Figure S1); however, CR failed to result in a detectable increase in Sir2 activity in respiratory deficient or respiratory competent cells. Therefore, we find no evidence that a metabolic shift from fermentation toward respiration is involved in life span extension by CR or that Sir2 is activated in response to CR.
Nicotinamide Blocks Life Span Extension by CR
CR has also been proposed to activate Sir2 by reducing cellular pools of nicotinamide, a Sir2 inhibitor [50]. Addition of 5 mM nicotinamide to the medium prevents life span extension by CR in wild-type mother cells [49]; however, interpretation of this experiment is complicated by the fact that, like deletion of SIR2, high levels of nicotinamide result in a dramatically shortened life span. We took advantage of the fact that deletion of FOB1 suppresses the short life span and ERC hyperaccumulation phenotypes associated with deletion or inhibition of Sir2 [41] to ask whether nicotinamide could inhibit the longevity effect of calorie restriction, even in the absence of Sir2. As expected, growth on 5 mM nicotinamide reduced the life span of wild-type cells to a level not significantly different from that of sir2Δ cells (Figure 4A). The very short life span of sir2Δ cells or nicotinamide-treated wild-type cells is most likely due to the hyperaccumulation of ERCs in cells lacking Sir2 activity [36,41]. Also as expected, nicotinamide had no effect on the life span of sir2Δ fob1Δ double mutants (Figure 4B), because Sir2 is already absent from these cells. CR dramatically increased the life span of sir2Δ fob1Δ double mutants, but, unexpectedly, addition of nicotinamide decreased the magnitude of life span extension conferred by CR (Figure 4B). Thus, Sir2-independent life span extension by CR is partially prevented by nicotinamide.
Figure 4 Effect of Nicotinamide on Life Span Extension by CR
(A) Nicotinamide shortens the life span of wild-type cells. Replicative life span analysis for BY4742 wild-type and sir2Δ cells on 2% glucose containing or lacking 5 mM nicotinamide (nic). Mean life span is shown in parentheses.
(B) Nicotinamide partially prevents Sir2-independent life span extension by CR. Replicative life span analysis for BY4742 wild type on 2% glucose, along with sir2Δ fob1Δ double mutant cells on 2% glucose or 0.05% glucose (CR) containing or lacking 5 mM nicotinamide (nic). Mean life span is shown in parentheses.
Discussion
Role of Respiration and Nicotinamide in Life Span Extension by CR
Sir2 is dispensable for life span extension by CR in yeast [36]. It remains possible, however, that under some conditions, CR might be mediated by Sir2. Central to this possibility is the premise that CR results in activation of Sir2. One mechanism by which CR has been hypothesized to activate Sir2 involves altered levels of the nicotinamide adenine dinucleotide cofactors NAD+ and NADH, resulting from a metabolic shift toward increased respiration in response to CR [46,47]. The other mechanism by which CR has been suggested to activate Sir2 is through a reduction in nicotinamide levels [49].
An important test of the respiration-dependent model for CR is whether CR can increase life span in cells that are incapable of respiration. Contradictory to the prediction from this model, we find that respiration is dispensable for enhanced longevity in response to CR. Growth on reduced glucose resulted in increased life span in two distinct models of respiratory deficiency, cyt1Δ and rho0 (see Figures 1C and 2C–2E). These data, combined with the observation that inhibition of Sir2 cannot account for the effect of nicotinamide on life span extension by CR (Figure 4B), call into question the proposed molecular explanations for activation of Sir2 in response to CR. Further, we find no evidence that Sir2 activity is altered either by CR or by respiratory deficiency, as measured by Sir2-dependent transcriptional silencing near telomeres (see Figure 3B and 3C). This result does not rule out the possibility that CR specifically enhances Sir2 activity at the rDNA; however, like the case at telomeres, Sir2-dependent silencing of a modified URA3 marker gene inserted into the rDNA is not enhanced by CR (J. Smith, personal communication). Thus, we propose that life span extension by CR occurs through a conserved Sir2-independent, respiration-independent mechanism (Figure 5).
Figure 5 Genetic Pathways Determining Replicative Life Span in Yeast
Sir2 and CR act in parallel pathways to slow aging. Both pathways are affected by nicotinamide levels.
It should be noted that our results do not contradict previous findings that increased respiration correlates with replicative life span in PSY316. Overexpression of the HAP4 transcription factor, which results in transcriptional up-regulation of respiratory genes and increased oxygen consumption, or overexpression of a mitochondrial NADH oxidoreductase, are reported to increase life span in PSY316 [46,47]. It remains possible that these interventions do indeed activate Sir2 by altering levels of NAD+ or NADH, as proposed. Alternatively, these interventions may behave as genetic mimics of CR, increasing life span through a Sir2-independent mechanism.
Our data suggest that high levels of nicotinamide can alter the response of yeast cells to CR. How might nicotinamide interfere with life span extension by CR? We can imagine at least three possible models. First, nicotinamide could partially block CR by inhibiting the activity of the other yeast Sirtuins (Hst1–4). This model seems unlikely because CR increases the life span of yeast cells lacking both Sir2 and either Hst1, Hst2, Hst3, or Hst4, and CR increases the life span of a sir2Δ fob1Δ hst1Δ hst2Δ quadruple mutant by greater than 50% (unpublished data). Second, nicotinamide could specifically interfere with the longevity benefits of CR, but through a mechanism unrelated to Sirtuin action. Nicotinamide, conventionally classified as a vitamin, participates in many biological processes distinct from Sirtuins [59], and could conceivably alter the activity of any NAD+–binding protein in the cell. Third, a reduction in nicotinamide levels conferred by CR might be important to offset detrimental effects, resulting from growth on reduced glucose medium, that are themselves unrelated to replicative aging, but may shorten life span to an extent that it masks life span extension by CR. Further study will be required to distinguish between these models.
Mitochondrial Defects, CR, and Longevity
Defects in mitochondrial function cause several human diseases, and mutation of mitochondrial DNA has been suggested to result in age-associated phenotypes in mammals [60–62]. Yeast provides a unique model in which to study the phenotypic consequences of mutation to the mitochondrial genome. With respect to replicative life span, complete deletion of the mitochondrial genome (rho0) results in different phenotypic outcomes depending on the genetic background of the strain [58,63]. Indeed, we report here that rho0 cells of BY4742 have a life span comparable to that of wild-type cells, whereas, rho0 cells of PSY316 are extremely short-lived (compare Figure 1C with Figure 2C). Presumably, this difference is the result of polymorphisms present in the nuclear genomes of these strains. Interestingly, in the PSY316 strain background, a nuclear mutation (cyt1Δ) that prevents respiration results in a life span comparable to that of wild-type cells (see Figure 2D). Thus, the short life span of PSY316 rho0 cells is apparently caused by loss of mitochondrial DNA rather than a general consequence of respiratory deficiency.
Although the PSY316 rho0 variant is extremely short-lived, CR by growth on low glucose is capable of increasing the life span of these cells by more than 100% (see Figure 3A). In fact, CR increases life span of the rho0 strain to a level that is comparable to calorie-restricted wild-type cells. To the best of our knowledge, this is the first indication, in any organism, that CR has a beneficial effect on defects caused by deletion of mitochondrial DNA. It will be of interest to understand the molecular basis for this effect and to determine whether this is a general feature of CR in multicellular eukaryotes.
Conclusion
Three competing models of life span extension by CR in yeast have been put forward: (1) Sir2 activation through a metabolic shift to respiration [46,47], (2) Sir2 activation by decreased nicotinamide levels [49], and (3) Sir2-independent life span extension [28,36]. Although CR can increase life span by a Sir2-independent mechanism [36], it remains to be determined whether either of the Sir2-dependent models account for a portion of the longevity benefits of CR under any conditions. We show here that in two different strain backgrounds, one of which is the PSY316 strain background used to generate the data supporting the Sir2-dependent models, life span extension by CR does not require respiration. We also show that the partial inhibition of CR by addition of exogenous nicotinamide does not act through Sir2. Thus, activation of Sir2 through a metabolic shift to respiration or through depletion of intracellular nicotinamide cannot explain CR-mediated increases in longevity.
Materials and Methods
Strains and media.
Unless otherwise stated, all yeast strains were derived from the parent strain for the haploid yeast open reading frame (ORF) deletion collection [64], BY4742 (obtained from Research Genetics, Invitrogen, Carlsbad, California, United States) or from PSY316AUT [55]. Strains used in this study are listed in Table 1. Gene disruptions were carried out by transforming yeast with PCR-amplified deletion constructs containing 45 nucleotides of homology to regions flanking the ORF to be deleted and either HIS3, LEU2, or URA3 amplified from pRS403, pRS405, or pRS406 [65], respectively. In each case, the entire ORF of the deleted gene was removed. All gene disruptions were verified by PCR. Medium used for life span studies was YEP (2% bacto peptone, 1% yeast extract) supplemented with filter-sterilized glucose at the designated concentration. For nicotinamide supplementation experiments, nicotinamide was added to YEP from a 500 mM nicotinamide (100×) filter sterilized stock solution to a final concentration of 5 mM just prior to pouring plates. Nicotinamide was obtained from Sigma (St. Louis, Missouri, United States).
Table 1 Yeast Strains Used in This Study
Generation of rho0 strains and verification by DAPI staining.
The rho0 strains used for life span analysis were generated by treatment with ethidium bromide. In each case, life span was determined for more than one rho0 isolate in order to verify the observed phenotype. In the case of PSY316AUT rho0, four different rho0 isolates were examined, and the severe shortening in life span was observed in all four cases. Life span was also determined for spontaneously arising PSY316AUT rho0 cells, which showed a life span defect similar to that of rho0 cells generated by ethidium bromide. Absence of mitochondrial DNA was verified by fluorescence microscopy of log phase cells stained with DAPI.
Replicative life span analysis.
Replicative life span analysis was carried out as described [58]. For all life span experiments, strains were coded such that the researcher performing the life span experiment had no knowledge of the strain genotypes. Unless otherwise stated, standard life span medium was YEP + 2% glucose (YPD) and CR medium was YEP + 0.05% glucose. Life span experiments in the presence of nicotinamide were carried out at a final concentration of 5 mM nicotinamide in the plates. Cells were grown on experimental medium for at least 8 h prior to microdissection. Wilcoxon p-values were calculated using the MATLAB “ranksum” function, and strains are stated to have a significant difference in life span for p < 0.05.
FOA telomere silencing assays.
For the silencing experiment shown in Figure 3B and Figure S1, three independent cultures were inoculated from single colonies into liquid YPD for each genotype and grown overnight. The next morning, each overnight culture was diluted 1:100 into YPD or CR medium and grown for 4 h in a shaking incubator. Cultures were then diluted to a cell density of approximately 2 × 103 cells/ml in water, and plated in 100-μl aliquots onto synthetic complete (SC) or FOA medium, containing either 2% or 0.05% glucose, such that cells cultured in 2% glucose were plated onto 2% glucose plates and cells cultured in CR medium were plated onto 0.05% glucose plates (CR plates). Percent survival was calculated as the number of colonies arising on FOA medium divided by the number of colonies arising on SC medium. Nicotinamide silencing experiments were carried out as above, except that after the overnight culture, cells were preincubated for 4 h in YPD + 5 mM nicotinamide and plated onto SC + 5 mM nicotinamide or FOA + 5 mM nicotinamide.
For the silencing experiment shown in Figure 3C, cultures of wild-type or cyt1Δ cells were inoculated from single colonies into liquid YPD or CR medium. The next morning, each overnight culture was diluted 1:1000 into fresh control or CR medium, such that cells grown overnight in control medium were diluted in control medium and cells grown overnight in CR medium were diluted into CR medium, and grown for 8 h in a shaking incubator. Cell cycle division time for BY4742 control cells was approximately 95 min and for BY4742 CR cells was approximately 105 min. After outgrowth, cultures were then diluted to a cell density of approximately 2 × 103 cells/ml in water, and plated in 100-μl aliquots onto SC or FOA medium, containing either 2% or 0.05% glucose, such that cells cultured in 2% glucose were plated onto 2% glucose plates and cells cultured in CR medium were plated onto CR plates. Percent survival was calculated as the number of colonies arising on FOA medium divided by the number of colonies arising on SC medium.
Supporting Information
Figure S1 CR Has No Effect on Sir2 Activity in Respiratory-Competent or Respiratory-Deficient Cells
Transcriptional silencing of the telomeric URA3 marker in PSY316AUT was monitored by the survival of cells plated onto medium containing 5-FOA.
(42 KB PDF)
Click here for additional data file.
We would like to thank J. Smith, D. Gottschling, and T. Powers for helpful discussion. This work has been funded by a grant from the Ellison Medical Foundation. Support for this work has also been provided by the American Federation for Aging Research and the University of Washington Nathan Shock Center of Excellence for the Basic Biology of Aging. MK is supported by National Institutes of Health training grant P30 AG013280. SF is an investigator of the Howard Hughes Medical Institute. BKK is a Searle Scholar.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. MK, SF, and BKK conceived and designed the experiments. MK, DH, EOK, MT, ND, and BKK performed the experiments. MK and BKK analyzed the data. MK, DH, EAW, and BKK contributed reagents/materials/analysis tools. MK and BKK wrote the paper.
A previous version of this article appeared as an Early Online Release on October 25, 2005 (DOI: 10.1371/journal.pgen.0010069.eor).
Abbreviations
CRcalorie restriction
ERCextrachromosomal rDNA circles
ORFopen reading frame
SCsynthetic complete
==== Refs
References
Walker G Houthoofd K Vanfleteren JR Gems D 2005 Dietary restriction in C. elegans: From rate-of-living effects to nutrient sensing pathways Mech Ageing Dev 126 929 937 15896824
Merry BJ 2005 Dietary restriction in rodents—Delayed or retarded ageing Mech Ageing Dev 126 951 959 15893804
Partridge L Piper MDW Mair W 2005 Dietary restriction in Drosophila
Mech Ageing Dev 126 938 950 15935441
Sinclair DA 2005 Toward a unified theory of caloric restriction and longevity regulation Mech Ageing Dev 126 987 1002 15893363
Guarente L 2005 Calorie restriction and SIR2 genes: Towards a mechanism Mech Ageing Dev 126 923 928 15941577
Merry BJ 2004 Oxidative stress and mitochondrial function with aging—The effects of calorie restriction Aging Cell 3 7 12 14965349
Armeni T Pieri C Marra M Saccucci F Principato G 1998 Studies on the life prolonging effect of food restriction: Glutathione levels and glyoxylase enzymes in rat liver Mech Ageing Dev 101 101 110 9593316
De Cabo R Cabello R Rios M Lopez-Lluch G Ingram DK 2004 Calorie restriction attenuates age-related alterations in the plasma membrane antioxidant system in rat liver Exp Gerontol 39 297 304 15036389
Guo Z Heydari A Richardson A 1998 Nucleotide excision repair of actively transcribed versus nontranscribed DNA in rat hepatocytes: Effect of age and dietary restriction Exp Cell Res 245 228 238 9828120
Rao KS 2003 Dietary calorie restriction, DNA repair and brain aging Mol Cell Biochem 253 313 318 14619982
Yu BP 1996 Aging and oxidative stress: Modulation by dietary restriction Free Radic Biol Med 21 651 668 8891668
Youngman LD Park JY Ames BN 1992 Protein oxidation associated with aging is reduced by dietary restriction of protein or calories Proc Natl Acad Sci U S A 89 9112 9116 1409611
Sohol RS Weindruch R 1996 Oxidative stress, caloric restriction, aging Science 273 59 63 8658196
Masoro EJ Katz MS McMahan CA 1989 Evidence for the glycation hypothesis of aging from the food-restricted rodent model J Gerontol 44 B20 B22 2910985
Masoro EJ McCarter RJM Katz MS McMahan CA 1992 Dietary restriction alters the characteristics of glucose fuel use J Gerontol 47 B202 B208 1430849
Kemnitz JW Roecker EB Weindruch R Elson DF Baum ST 1994 Dietary restriction increases insulin sensitivity and lowers blood glucose in rhesus monkeys Am J Physiol 266 E540 E547 8178974
Zhang Y Herman B 2002 Ageing and apoptosis Mech Ageing Dev 123 245 260 11744038
Weindruch R Walford RL 1988 The retardation of aging and disease by dietary restriction Springfield (Illinois) Charles C. Thomas 436 p.
Jolly CA 2005 Diet manipulation and prevention of aging, cancer, and autoimmune disease Curr Opin Clin Nutr Metab Care 8 382 387 15930962
Mattson MP 2000 Emerging neuroprotective strategies for Alzheimer's disease: Dietary restriction, telomerase activation, and stem cell therapy Exp Gerontol 35 489 502 10959037
Saffrey MJ 2004 Ageing of the enteric nervous system Mech Ageing Dev 125 899 906 15563936
McCay CM Crowell MF Maynard LA 1935 The effect of retarded growth upon the length of life and upon ultimate size J Nutr 10 63 79
Masoro EJ 1998 Influence of caloric intake on aging and on the response to stressors J Toxicol Environ Health B Crit Rev 1 243 257 9644329
Mattson MP Duan W Guo Z 2003 Meal size and frequency affect neuronal plasticity and vulnerability to disease: Cellular and molecular mechanisms J Neurochem 84 417 431 12558961
Murphy CT McCarroll SA Bargmann CI Fraser A Kamath RS 2003 Genes that act downstream of DAF-16 to influence the lifespan of Caenorhabditis elegans
Nature 424 277 283 12845331
Kenyon C 2005 The plasticity of aging: Insights from long-lived mutants Cell 120 449 460 15734678
Braeckman BP Houthoofd K Vanfleteren JR 2001 Insulin-like signaling, metabolism, stress resistance and aging in Caenorhabditis elegans
Mech Ageing Dev 122 673 693 11322992
Kaeberlein M Powers RW Steffen KK Westman EA Hu D 2005 Regulation of yeast replicative life span by Tor and Sch9 in response to nutrients Science In press.
Meissner B Boll M Daniel H Baumeister R 2004 Deletion of the intestinal peptide transporter affects insulin and TOR signaling in Caenorhabditis elegans
J Biol Chem 279 36739 36745 15155758
Vellai T Takacs-Vellai K Zhang Y Kovacs AL Orosz L 2003 Genetics: Influence of TOR kinase on lifespan in C. elegans
Nature 426 620
Kapahi P Zid BM Harper T Koslover D Sapin V 2004 Regulation of lifespan in Drosophila by modulation of genes in the TOR signaling pathway Curr Biol 14 885 890 15186745
Sharp ZD Bartke A 2005 Evidence for down-regulation of phosphoinositide 3-kinase/Akt/mammalian target of rapamycin (PI3K/Akt/mTOR)-dependent translation regulatory signaling pathways in Ames dwarf mice J Gerontol A Biol Sci Med Sci 60 293 300 15860463
Lin SJ Defossez PA Guarente L 2000 Requirement of NAD and SIR2 for life-span extension by calorie restriction in Saccharomyces cerevisiae
Science 289 2126 2128 11000115
Rogina B Helfand SL 2004 Sir2 mediates longevity in the fly through a pathway related to calorie restriction Proc Natl Acad Sci U S A 101 15998 16003 15520384
Guarente L Picard F 2005 Calorie restriction—The SIR2 connection Cell 120 473 482 15734680
Kaeberlein M Kirkland KT Fields S Kennedy BK 2004 Sir2-independent life span extension by calorie restriction in yeast PLoS Biol 2 E296 15328540
Kaeberlein M Andalis AA Fink GR Guarente L 2002 High osmolarity extends life span in Saccharomyces cerevisiae by a mechanism related to calorie restriction Mol Cell Biol 22 8056 8066 12391171
Smith JS Brachmann CB Celic I Kenna MA Muhammad S 2000 A phylogenetically conserved NAD+ -dependent protein deacetylase activity in the Sir2 protein family Proc Natl Acad Sci U S A 97 6658 6663 10841563
Landry J Sutton A Tafrov ST Heller RC Stebbins J 2000 The silencing protein SIR2 and its homologs are NAD-dependent protein deacetylases Proc Natl Acad Sci U S A 97 5807 5811 10811920
Imai S Armstrong CM Kaeberlein M Guarente L 2000 Transcriptional silencing and longevity protein Sir2 is an NAD-dependent histone deacetylase Nature 403 795 800 10693811
Kaeberlein M McVey M Guarente L 1999 The SIR2/3/4 complex and SIR2 alone promote longevity in Saccharomyces cerevisiae by two different mechanisms Genes Dev 13 2570 2580 10521401
Sinclair DA Guarente L 1997 Extrachromosomal rDNA circles—A cause of aging in yeast Cell 91 1033 1042 9428525
Michel AH Kornmann B Dubrana K Shore D 2005 Spontaneous rDNA copy number variation modulates Sir2 levels and epigenetic gene silencing Genes Dev 19 1199 1210 15905408
Kennedy BK Austriaco NR Zhang J Guarente L 1995 Mutation in the silencing gene SIR4 can delay aging in S. cerevisiae
Cell 80 485 496 7859289
Couzin J 2004 Scientific community. Aging research's family feud Science 303 1276 1279 14988530
Lin SJ Kaeberlein M Andalis AA Sturtz LA Defossez PD 2002 Calorie restriction extends Saccharomyces cerevisiae life span by increasing respiration Nature 418 344 348 12124627
Lin SJ Ford E Haigis M Liszt G Guarente L 2004 Calorie restriction extends yeast life span by lowering the level of NADH Genes Dev 18 12 16 14724176
Anderson RM Latorre-Esteves M Neves AR Lavu S Medvedik O 2003 Yeast life-span extension by calorie restriction is independent of NAD fluctuation Science 302 2124 2126 14605207
Anderson RM Bitterman KJ Wood JG Medvedik O Sinclair DA 2003 Nicotinamide and PNC1 govern lifespan extension by calorie restriction in Saccharomyces cerevisiae
Nature 423 181 185 12736687
Bitterman KJ Anderson RM Cohen HY Latorre-Esteves M Sinclair DA 2002 Inhibition of silencing and accelerated aging by nicotinamide, a putative negative regulator of yeast sir2 and human SIRT1 J Biol Chem 277 45099 45107 12297502
Landry J Slama JT Sternglanz R 2000 Role of NAD(+) in the deacetylase activity of the SIR2-like proteins Biochem Biophys Res Comm 278 685 690 11095969
Gallo CM Smith DL Jr. Smith JS 2004 Nicotinamide clearance by Pnc1 directly regulates Sir2-mediated silencing and longevity Mol Cell Biol 24 1301 1312 14729974
Kaeberlein M Kennedy BK 2005 Large-scale identification in yeast of conserved ageing genes Mech Ageing Dev 126 17 21 15610758
Kaeberlein M Andalis AA Liszt G Fink GR Guarente L 2004
Saccharomyces cerevisiae SSD1-V confers longevity by a Sir2p-independent mechanism Genetics 166 1661 1672 15126388
Kaeberlein M McDonagh T Heltweg B Hixon J Westman EA 2005 Substrate-specific activation of sirtuins by resveratrol J Biol Chem 280 17038 17045 15684413
Tzagoloff A Myers AM 1986 Genetics of mitochondrial biogenesis Annu Rev Biochem 55 249 285 2427014
de Zamaroczy M Bernardi G 1986 The primary structure of the mitochondrial genome of Saccharomyces cerevisiae
Gene 47 155 177 3549452
Kaeberlein M Kirkland KT Fields S Kennedy BK 2005 Genes determining yeast replicative life span in a long-lived genetic background Mech Ageing Dev 126 491 504 15722108
Williams AC Ramsden DB 2005 Nicotinamide homeostasis: A xenobiotic pathway that is key to development and degenerative diseases Med Hypotheses 65 353 362 15922112
Wallace DC 2001 A mitochondrial paradigm for degenerative diseases and ageing Novartis Found Symp 235 247 266 11280029
Wallace DC Shoffner JM Trounce I Brown MD Ballinger SW 1995 Mitochondrial DNA mutations in human degenerative diseases and aging Biochim Biophys Acta 1271 141 151 7599200
Taylor RW Turnbull DM 2005 Mitochondrial DNA mutations in human disease Nat Rev Genet 6 389 402 15861210
Kirchman PA Kim S Lai CY Jazwinski SM 1999 Interorganelle signaling is a determinant of longevity in Saccharomyces cerevisiae
Genetics 152 179 190 10224252
Winzeler EA Shoemaker DD Astromoff A Liang H Anderson K 1999 Functional characterization of the S. cerevisiae genome by gene deletion and parallel analysis Science 285 901 906 10436161
Sikorski RS Hieter P 1989 A system of shuttle vectors and yeast host strains designed for efficient manipulation of DNA in Saccharomyces cerevisiae
Genetics 122 19 27 2659436
|
16311627
|
PMC1287956
|
CC BY
|
2021-01-05 08:00:25
|
no
|
PLoS Genet. 2005 Nov 25; 1(5):e69
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010069
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1631162710.1371/journal.pgen.0010069plge-01-05-12Research ArticleIncreased Life Span due to Calorie Restriction in Respiratory-Deficient Yeast Calorie Restriction in YeastKaeberlein Matt 1*Hu Di 2Kerr Emily O 2Tsuchiya Mitsuhiro 2Westman Eric A 2Dang Nick 2Fields Stanley 13Kennedy Brian K 2*1 Departments of Genome Sciences and Medicine, University of Washington, Seattle, Washington, United States of America
2 Department of Biochemistry, University of Washington, Seattle, Washington, United States of America
3 Howard Hughes Medical Institute, University of Washington, Seattle, Washington, United States of America
Kim Stuart EditorStanford University School of Medicine, United States of America*To whom correspondence should be addressed. E-mail: [email protected] (MK), [email protected] (BKK)11 2005 25 11 2005 25 10 2005 1 5 e697 7 2005 24 10 2005 Copyright: © 2005 Kaeberlein 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.A model for replicative life span extension by calorie restriction (CR) in yeast has been proposed whereby reduced glucose in the growth medium leads to activation of the NAD+–dependent histone deacetylase Sir2. One mechanism proposed for this putative activation of Sir2 is that CR enhances the rate of respiration, in turn leading to altered levels of NAD+ or NADH, and ultimately resulting in enhanced Sir2 activity. An alternative mechanism has been proposed in which CR decreases levels of the Sir2 inhibitor nicotinamide through increased expression of the gene coding for nicotinamidase, PNC1. We have previously reported that life span extension by CR is not dependent on Sir2 in the long-lived BY4742 strain background. Here we have determined the requirement for respiration and the effect of nicotinamide levels on life span extension by CR. We find that CR confers robust life span extension in respiratory-deficient cells independent of strain background, and moreover, suppresses the premature mortality associated with loss of mitochondrial DNA in the short-lived PSY316 strain. Addition of nicotinamide to the medium dramatically shortens the life span of wild type cells, due to inhibition of Sir2. However, even in cells lacking both Sir2 and the replication fork block protein Fob1, nicotinamide partially prevents life span extension by CR. These findings (1) demonstrate that respiration is not required for the longevity benefits of CR in yeast, (2) show that nicotinamide inhibits life span extension by CR through a Sir2-independent mechanism, and (3) suggest that CR acts through a conserved, Sir2-independent mechanism in both PSY316 and BY4742.
Synopsis
Calorie restriction slows aging and increases life span in nearly every organism studied. The mechanism by which this occurs is one of the most important unanswered questions in biogerontology. One popular theory, based on work from the budding yeast Saccharomyces cerevisiae, proposes that calorie restriction works by causing a metabolic shift toward increased mitochondrial respiration, resulting in activation of a family of proteins known as Sirtuins. This study demonstrates that life span extension by calorie restriction does not require respiration and occurs even in cells completely lacking mitochondrial DNA. Interestingly, calorie restriction protects yeast cells against a severe longevity defect associated with absence of mitochondrial DNA, suggesting the possibility that the consequences of age-associated mitochondrial dysfunction might be alleviated or prevented by calorie restriction.
Citation:Kaeberlein M, Hu D, Kerr EO, Tsuchiya M, Westman EA, et al. (2005) Increased life span due to calorie restriction in respiratory-deficient yeast. PLoS Genet 1(5): e69.
==== Body
Introduction
Calorie restriction (CR) has been shown to slow aging in evolutionarily divergent species, including yeast, worms, flies, and rodents [1–5]. In addition to increasing longevity, CR is reported to cause additional phenotypes, including increased resistance to oxidative stress [6–8], enhanced DNA damage repair [9,10], decreased levels of oxidatively damaged proteins [11–13], improved glucose homeostasis and insulin sensitivity [14–16], altered levels of apoptosis [17], and delayed onset of a number of age-related diseases [18–21]. Although it has been known for more than 70 y that calorie restriction can increase life span in mammals [22], a mechanistic understanding of this phenomenon has remained elusive. It seems clear that nutrient and growth factor responsive pathways, such as those mediated by insulin, IGF-1, TOR, and Akt, are likely to represent important conduits through which these signals affect the aging rate. CR mediates enhancement of stress response pathways in mammals [23,24], and signaling through the insulin-like pathway in worms coordinates expression of a variety of antioxidant, chaperone, and anti-bacterial stress response proteins [25–27]. Similarly, TOR-mediated regulation of translational machinery appears to play a role in the response to nutrient deprivation in yeast [28], worms [29,30], flies [31], and mammals [32]. Finally, models postulating a role for Sir2-like protein deacetylases in CR-mediated life span extension have gained popularity for yeast [33], flies [34], and mammals, as well [4,35].
In the budding yeast Saccharomyces cerevisiae, CR can be imposed by reducing the concentration of glucose in the growth medium, resulting in a 20%–40% increase in replicative life span in multiple strain backgrounds [33,36,37]. In addition, genetic models of CR include deletion of the gene coding for hexokinase, HXK2, and mutations that decrease signaling through the cAMP-dependent protein kinase, PKA, such as deletion of the genes coding for the glucose sensing proteins GPA2 or GPR1, and temperature-sensitive alleles of adenylate cyclase (cdc35–1) or the RAS-associated GTPase (cdc25–10) [33].
CR has been proposed to increase yeast replicative life span by a mechanism involving activation of Sir2 [33], an NAD+–dependent histone deacetylase [38–40] that inhibits the formation of extrachromosomal rDNA circles (ERCs) [41]. ERCs are self-replicating DNA molecules that accumulate in the mother-cell nucleus with age and are thought to cause senescence [42]. Overexpression of Sir2 increases life span in multiple strain backgrounds [36,41,43], and deletion of Sir2 shortens life span by about 50% [41,44]. CR fails to increase the life span of short-lived sir2Δ cells [33], consistent with the idea that CR could be acting by a mechanism involving Sir2.
The question of how CR might activate Sir2 has been a source of considerable controversy [45]. Saccharomyces cerevisiae is a facultative anaerobe that, under standard laboratory growth conditions (2% glucose), generates ATP largely by fermentation. Under conditions of reduced glucose, such as CR, S. cerevisiae shifts from fermentation to respiration, resulting in increased transcription of respiratory genes and a higher rate of oxygen consumption [46]. In models put forth by Lin et al., this metabolic shift results in activation of Sir2, either through increased cellular NAD+ [46] or decreased cellular NADH [47]. Alternatively, Anderson et al. have reported that CR does not alter NAD+ levels [48], but leads to enhanced expression of PNC1 and a reduction in cellular nicotinamide [49]. Since nicotinamide is an inhibitor of the Sir2 deacetylation reaction, its decreased concentration could result in enhanced Sir2 activity [50,51]. Overexpression of Pnc1 suppresses the effect of exogenously added nicotinamide on Sir2-dependent silencing at HM loci, telomeres, and rDNA [52]; there are conflicting reports, however, on whether Pnc1 overexpression alters Sir2 activity at endogenous levels of nicotinamide [49,52].
More recently, we have questioned the importance of Sir2 in life span extension by CR [28,53]. In a long-lived strain background, BY4742, CR increases life span to a greater extent in cells lacking both Sir2 and the replication fork barrier protein Fob1 than in wild-type cells [36]. Based on this observation, and the fact that deletion of SIR2 shortens life span by approximately 50%, we proposed a model whereby the inability of CR to increase life span in sir2Δ FOB1 cells is explained as an indirect effect, resulting from the hyperaccumulation of ERCs [36]. Deletion of FOB1 in a sir2Δ background suppresses the hyperaccumulation of ERCs, as well as the longevity defect [41], thus allowing CR to dramatically increase the life span of cells in the absence of Sir2 [36].
Much of the early work suggesting a link between CR and Sir2 was carried out in the PSY316 strain background [33,37,46,47,49,50,54], a strain in which, paradoxically, overexpression of Sir2 fails to increase life span [55]. Guarente and Picard [35] have proposed that CR might act via different mechanisms in the BY4742 and PSY316 strain backgrounds. In order to further clarify the molecular mechanism(s) underlying replicative life span extension by calorie restriction in yeast, we have sought to directly test key components of the models for Sir2-dependent CR in both of these strains. Here, we examine in detail the role of respiratory metabolism in life span extension by CR, finding that (1) respiration is not required for life span extension by CR; and (2) CR suppresses the enhanced early mortality, only apparent in PSY316, due to loss of mitochondrial DNA. In contrast, exogenous addition of nicotinamide partially, but not completely, blocks Sir2-independent life span extension by CR.
Results
Life Span Extension by CR Is Independent of Respiration in BY4742
A central facet of the Sir2-dependent models for life span extension by CR is that a metabolic shift from fermentation to respiration in response to CR results in activation of Sir2 [46,47]. Since Sir2 is not required for life span extension by CR in BY4742 [36], we wished to determine whether respiration was also dispensable. The effect of CR in the absence of respiratory capacity was examined using rho0 cells, which completely lack mitochondrial DNA. Rho0 yeast lack three cytochrome c oxidase subunits (COX1, COX2, and COX3), three ATP synthase subunits (ATP6, ATP8, and ATP9), and apocytochrome b (CYTB), and are therefore incapable of respiratory metabolism [56,57]. BY4742 rho0 cells were generated (see Materials and Methods), and the absence of mitochondrial DNA was verified by staining cells with DAPI (Figure 1A). Lack of respiratory capacity in rho0 cells was confirmed by inability to grow on the non-fermentable carbon source, glycerol (Figure 1B). As previously observed [58], replicative life span under standard conditions (2% glucose) was not altered by loss of mitochondrial DNA in this strain (Figure 1C). CR (0.05% glucose) significantly enhanced the life span of both wild-type and rho0 cells to a comparable degree. Thus, respiration is not required for life span extension by CR in the long-lived BY4742 strain background.
Figure 1 Respiration Is Not Required for Life Span Extension by CR in BY4742
(A) BY4742 rho0 strains lack mitochondrial DNA. DAPI staining of BY4742 (i) wild-type or (ii) rho0 cells grown under standard conditions (2% glucose) and calorie-restricted (iii) wild-type or (iv) rho0 cells (CR = 0.05% glucose).
(B) BY4742 rho0 strains are unable to grow on glycerol as the sole carbon source. (i) BY4742 wild-type or (ii) rho0 cells on YEP supplemented with 2% glucose or 3% glycerol.
(C) CR increase life span in BY4742 rho0 cells. Replicative life span analysis for BY4742 wild-type and rho0 cells on 2% glucose and 0.05% glucose (CR). Mean life span is shown in parentheses.
Life Span Extension by CR Is Independent of Respiration in PSY316
Previous work indicating that life span extension by CR is dependent on respiration was carried out in the PSY316 strain background, in which it was reported that deletion of the gene coding for cytochrome c1, CYT1, prevents life span extension by CR [46]. In order to address whether life span extension by CR in mutants incapable of respiratory metabolism is specific to BY4742, we generated respiratory-deficient cyt1Δ and rho0 variants in the PSY316 background (Figure 2A and 2B) and measured life span on medium containing either 2% or 0.05% glucose. As in BY4742 rho0 cells, CR significantly increased both mean and maximum life span in PSY316 rho0 (Figure 2C) and cyt1Δ cells (Figure 2D).
Figure 2 Respiration Is Not Required for Life Span Extension by CR in PSY316
(A) PSY316AUT rho0 strains lack mitochondrial DNA. DAPI staining of PSY316 (i) wild-type or (ii) rho0 cells grown under standard conditions (2% glucose) and calorie-restricted (iii) wild-type or (iv) rho0 cells (CR = 0.05% glucose).
(B) PSY316AUT rho0 strains are unable to grow on glycerol as the sole carbon source. (i) PSY316AUT wild-type, (ii) cyt1Δ rho0, (iii) cyt1Δ, or (iv) rho0 cells on YEP supplemented with 2% glucose or 3% glycerol.
(C) CR increases life span in PSY316AUT rho0 cells. Replicative life span analysis for PSY316AUT wild-type and rho0 cells on 2% glucose and 0.05% glucose (CR). Mean life span is shown in parentheses.
(D) CR increases the life span of cyt1Δ cells. Replicative life span analysis for PSY316AUT wild-type and cyt1Δ cells on 2% glucose and 0.05% glucose (CR). Mean life span is shown in parentheses.
(E) CR increases the life span of cyt1Δ rho0 cells. Replicative life span analysis for PSY316AUT wild-type and cyt1Δ rho0 cells on 2% glucose and cyt1Δ rho0 cells on 0.05% glucose (CR). Mean life span is shown in parentheses.
Unlike the case in BY4742 in which deletion of mitochondrial DNA has no effect on life span, PSY316 rho0 variants exhibited a profound increase in early mortality under standard growth conditions. This phenotype was not observed in cyt1Δ cells, indicating that loss of mitochondrial DNA, rather than general respiratory deficiency, is responsible for the life span defect. Deletion of CYT1 in cells lacking mitochondrial DNA, however, resulted in a short life span comparable to that of rho0 cells (Figure 2E), demonstrating that rho0 is epistatic to cyt1Δ for this phenotype. As observed for rho0 cells, CR more than doubled the short life span of cyt1Δ rho0 cells, which contain both nuclear and mitochondrial mutations that prevent respiration.
Our observation that CR increased life span in PSY316 cells lacking CYT1 is in contrast to a previous report [46]. A potential explanation for this difference is that 0.5% glucose was used for CR in the prior study, rather than the 0.05% glucose concentration used in this study. We therefore measured the life span of respiratory deficient rho0 and cyt1Δ cells grown on 0.5% glucose. As we observed for cells grown on 0.05% glucose, growth on 0.5% glucose increased the life span of rho0 and cyt1Δ cells, although the magnitude of life span extension is reduced relative to 0.05% glucose (Figure 3A). Thus, the use of a non-optimal level of CR may have precluded detection of life span extension by CR in cyt1Δ mutants in the prior study.
Figure 3 Effect of Glucose Concentration on Life Span and Sir2 Activity in Respiratory-Deficient Mutants
(A) Mean replicative life span is significantly increased in PSY316 AUT cyt1Δ and rho0 cells as the glucose concentration is decreased to either 0.5% or 0.05%, relative to life span on 2% glucose. *p < 0.05, **p < 0.01.
(B) Sir2 activity is not increased by CR but is responsive to increased expression of Sir2 or to addition of exogenous nicotinamide. Transcriptional silencing of the telomeric URA3 marker in PSY316AUT (WT) was monitored by the survival of cells plated onto medium containing 5-FOA.
(C) Sir2 activity is not altered in respiratory deficient cyt1Δ cells and is unaffected by CR. Transcriptional silencing of the telomeric URA3 marker in PSY316AUT (WT) was monitored by the survival of cells plated onto medium containing 5-FOA.
We also examined the effect of CR on Sir2 activity in respiratory-deficient PSY316 cells. The PSY316AUT variant has both URA3 and ADE2 marker genes integrated near telomeres, thus allowing for efficient determination of Sir2-dependent telomeric silencing in response to genetic or environmental perturbations [55]. As previously reported, increased Sir2 activity can be achieved by overexpression of SIR2 in the PSY316 strain background [55], significantly enhancing telomere silencing and survival on FOA, while inhibition of Sir2 by addition of 5 mM nicotinamide to the medium decreased telomere silencing (Figure 3B). CR, however, had no detectable effect on Sir2-dependent silencing at telomeres. Similarly, respiratory deficiency caused by deletion of CYT1 also fails to impact Sir2-dependent silencing at either 2% or 0.05% glucose (Figure 3C). A decrease in survival on FOA was observed in rho0 cells relative to wild-type or cyt1Δ cells at 2% glucose (Figure S1); however, CR failed to result in a detectable increase in Sir2 activity in respiratory deficient or respiratory competent cells. Therefore, we find no evidence that a metabolic shift from fermentation toward respiration is involved in life span extension by CR or that Sir2 is activated in response to CR.
Nicotinamide Blocks Life Span Extension by CR
CR has also been proposed to activate Sir2 by reducing cellular pools of nicotinamide, a Sir2 inhibitor [50]. Addition of 5 mM nicotinamide to the medium prevents life span extension by CR in wild-type mother cells [49]; however, interpretation of this experiment is complicated by the fact that, like deletion of SIR2, high levels of nicotinamide result in a dramatically shortened life span. We took advantage of the fact that deletion of FOB1 suppresses the short life span and ERC hyperaccumulation phenotypes associated with deletion or inhibition of Sir2 [41] to ask whether nicotinamide could inhibit the longevity effect of calorie restriction, even in the absence of Sir2. As expected, growth on 5 mM nicotinamide reduced the life span of wild-type cells to a level not significantly different from that of sir2Δ cells (Figure 4A). The very short life span of sir2Δ cells or nicotinamide-treated wild-type cells is most likely due to the hyperaccumulation of ERCs in cells lacking Sir2 activity [36,41]. Also as expected, nicotinamide had no effect on the life span of sir2Δ fob1Δ double mutants (Figure 4B), because Sir2 is already absent from these cells. CR dramatically increased the life span of sir2Δ fob1Δ double mutants, but, unexpectedly, addition of nicotinamide decreased the magnitude of life span extension conferred by CR (Figure 4B). Thus, Sir2-independent life span extension by CR is partially prevented by nicotinamide.
Figure 4 Effect of Nicotinamide on Life Span Extension by CR
(A) Nicotinamide shortens the life span of wild-type cells. Replicative life span analysis for BY4742 wild-type and sir2Δ cells on 2% glucose containing or lacking 5 mM nicotinamide (nic). Mean life span is shown in parentheses.
(B) Nicotinamide partially prevents Sir2-independent life span extension by CR. Replicative life span analysis for BY4742 wild type on 2% glucose, along with sir2Δ fob1Δ double mutant cells on 2% glucose or 0.05% glucose (CR) containing or lacking 5 mM nicotinamide (nic). Mean life span is shown in parentheses.
Discussion
Role of Respiration and Nicotinamide in Life Span Extension by CR
Sir2 is dispensable for life span extension by CR in yeast [36]. It remains possible, however, that under some conditions, CR might be mediated by Sir2. Central to this possibility is the premise that CR results in activation of Sir2. One mechanism by which CR has been hypothesized to activate Sir2 involves altered levels of the nicotinamide adenine dinucleotide cofactors NAD+ and NADH, resulting from a metabolic shift toward increased respiration in response to CR [46,47]. The other mechanism by which CR has been suggested to activate Sir2 is through a reduction in nicotinamide levels [49].
An important test of the respiration-dependent model for CR is whether CR can increase life span in cells that are incapable of respiration. Contradictory to the prediction from this model, we find that respiration is dispensable for enhanced longevity in response to CR. Growth on reduced glucose resulted in increased life span in two distinct models of respiratory deficiency, cyt1Δ and rho0 (see Figures 1C and 2C–2E). These data, combined with the observation that inhibition of Sir2 cannot account for the effect of nicotinamide on life span extension by CR (Figure 4B), call into question the proposed molecular explanations for activation of Sir2 in response to CR. Further, we find no evidence that Sir2 activity is altered either by CR or by respiratory deficiency, as measured by Sir2-dependent transcriptional silencing near telomeres (see Figure 3B and 3C). This result does not rule out the possibility that CR specifically enhances Sir2 activity at the rDNA; however, like the case at telomeres, Sir2-dependent silencing of a modified URA3 marker gene inserted into the rDNA is not enhanced by CR (J. Smith, personal communication). Thus, we propose that life span extension by CR occurs through a conserved Sir2-independent, respiration-independent mechanism (Figure 5).
Figure 5 Genetic Pathways Determining Replicative Life Span in Yeast
Sir2 and CR act in parallel pathways to slow aging. Both pathways are affected by nicotinamide levels.
It should be noted that our results do not contradict previous findings that increased respiration correlates with replicative life span in PSY316. Overexpression of the HAP4 transcription factor, which results in transcriptional up-regulation of respiratory genes and increased oxygen consumption, or overexpression of a mitochondrial NADH oxidoreductase, are reported to increase life span in PSY316 [46,47]. It remains possible that these interventions do indeed activate Sir2 by altering levels of NAD+ or NADH, as proposed. Alternatively, these interventions may behave as genetic mimics of CR, increasing life span through a Sir2-independent mechanism.
Our data suggest that high levels of nicotinamide can alter the response of yeast cells to CR. How might nicotinamide interfere with life span extension by CR? We can imagine at least three possible models. First, nicotinamide could partially block CR by inhibiting the activity of the other yeast Sirtuins (Hst1–4). This model seems unlikely because CR increases the life span of yeast cells lacking both Sir2 and either Hst1, Hst2, Hst3, or Hst4, and CR increases the life span of a sir2Δ fob1Δ hst1Δ hst2Δ quadruple mutant by greater than 50% (unpublished data). Second, nicotinamide could specifically interfere with the longevity benefits of CR, but through a mechanism unrelated to Sirtuin action. Nicotinamide, conventionally classified as a vitamin, participates in many biological processes distinct from Sirtuins [59], and could conceivably alter the activity of any NAD+–binding protein in the cell. Third, a reduction in nicotinamide levels conferred by CR might be important to offset detrimental effects, resulting from growth on reduced glucose medium, that are themselves unrelated to replicative aging, but may shorten life span to an extent that it masks life span extension by CR. Further study will be required to distinguish between these models.
Mitochondrial Defects, CR, and Longevity
Defects in mitochondrial function cause several human diseases, and mutation of mitochondrial DNA has been suggested to result in age-associated phenotypes in mammals [60–62]. Yeast provides a unique model in which to study the phenotypic consequences of mutation to the mitochondrial genome. With respect to replicative life span, complete deletion of the mitochondrial genome (rho0) results in different phenotypic outcomes depending on the genetic background of the strain [58,63]. Indeed, we report here that rho0 cells of BY4742 have a life span comparable to that of wild-type cells, whereas, rho0 cells of PSY316 are extremely short-lived (compare Figure 1C with Figure 2C). Presumably, this difference is the result of polymorphisms present in the nuclear genomes of these strains. Interestingly, in the PSY316 strain background, a nuclear mutation (cyt1Δ) that prevents respiration results in a life span comparable to that of wild-type cells (see Figure 2D). Thus, the short life span of PSY316 rho0 cells is apparently caused by loss of mitochondrial DNA rather than a general consequence of respiratory deficiency.
Although the PSY316 rho0 variant is extremely short-lived, CR by growth on low glucose is capable of increasing the life span of these cells by more than 100% (see Figure 3A). In fact, CR increases life span of the rho0 strain to a level that is comparable to calorie-restricted wild-type cells. To the best of our knowledge, this is the first indication, in any organism, that CR has a beneficial effect on defects caused by deletion of mitochondrial DNA. It will be of interest to understand the molecular basis for this effect and to determine whether this is a general feature of CR in multicellular eukaryotes.
Conclusion
Three competing models of life span extension by CR in yeast have been put forward: (1) Sir2 activation through a metabolic shift to respiration [46,47], (2) Sir2 activation by decreased nicotinamide levels [49], and (3) Sir2-independent life span extension [28,36]. Although CR can increase life span by a Sir2-independent mechanism [36], it remains to be determined whether either of the Sir2-dependent models account for a portion of the longevity benefits of CR under any conditions. We show here that in two different strain backgrounds, one of which is the PSY316 strain background used to generate the data supporting the Sir2-dependent models, life span extension by CR does not require respiration. We also show that the partial inhibition of CR by addition of exogenous nicotinamide does not act through Sir2. Thus, activation of Sir2 through a metabolic shift to respiration or through depletion of intracellular nicotinamide cannot explain CR-mediated increases in longevity.
Materials and Methods
Strains and media.
Unless otherwise stated, all yeast strains were derived from the parent strain for the haploid yeast open reading frame (ORF) deletion collection [64], BY4742 (obtained from Research Genetics, Invitrogen, Carlsbad, California, United States) or from PSY316AUT [55]. Strains used in this study are listed in Table 1. Gene disruptions were carried out by transforming yeast with PCR-amplified deletion constructs containing 45 nucleotides of homology to regions flanking the ORF to be deleted and either HIS3, LEU2, or URA3 amplified from pRS403, pRS405, or pRS406 [65], respectively. In each case, the entire ORF of the deleted gene was removed. All gene disruptions were verified by PCR. Medium used for life span studies was YEP (2% bacto peptone, 1% yeast extract) supplemented with filter-sterilized glucose at the designated concentration. For nicotinamide supplementation experiments, nicotinamide was added to YEP from a 500 mM nicotinamide (100×) filter sterilized stock solution to a final concentration of 5 mM just prior to pouring plates. Nicotinamide was obtained from Sigma (St. Louis, Missouri, United States).
Table 1 Yeast Strains Used in This Study
Generation of rho0 strains and verification by DAPI staining.
The rho0 strains used for life span analysis were generated by treatment with ethidium bromide. In each case, life span was determined for more than one rho0 isolate in order to verify the observed phenotype. In the case of PSY316AUT rho0, four different rho0 isolates were examined, and the severe shortening in life span was observed in all four cases. Life span was also determined for spontaneously arising PSY316AUT rho0 cells, which showed a life span defect similar to that of rho0 cells generated by ethidium bromide. Absence of mitochondrial DNA was verified by fluorescence microscopy of log phase cells stained with DAPI.
Replicative life span analysis.
Replicative life span analysis was carried out as described [58]. For all life span experiments, strains were coded such that the researcher performing the life span experiment had no knowledge of the strain genotypes. Unless otherwise stated, standard life span medium was YEP + 2% glucose (YPD) and CR medium was YEP + 0.05% glucose. Life span experiments in the presence of nicotinamide were carried out at a final concentration of 5 mM nicotinamide in the plates. Cells were grown on experimental medium for at least 8 h prior to microdissection. Wilcoxon p-values were calculated using the MATLAB “ranksum” function, and strains are stated to have a significant difference in life span for p < 0.05.
FOA telomere silencing assays.
For the silencing experiment shown in Figure 3B and Figure S1, three independent cultures were inoculated from single colonies into liquid YPD for each genotype and grown overnight. The next morning, each overnight culture was diluted 1:100 into YPD or CR medium and grown for 4 h in a shaking incubator. Cultures were then diluted to a cell density of approximately 2 × 103 cells/ml in water, and plated in 100-μl aliquots onto synthetic complete (SC) or FOA medium, containing either 2% or 0.05% glucose, such that cells cultured in 2% glucose were plated onto 2% glucose plates and cells cultured in CR medium were plated onto 0.05% glucose plates (CR plates). Percent survival was calculated as the number of colonies arising on FOA medium divided by the number of colonies arising on SC medium. Nicotinamide silencing experiments were carried out as above, except that after the overnight culture, cells were preincubated for 4 h in YPD + 5 mM nicotinamide and plated onto SC + 5 mM nicotinamide or FOA + 5 mM nicotinamide.
For the silencing experiment shown in Figure 3C, cultures of wild-type or cyt1Δ cells were inoculated from single colonies into liquid YPD or CR medium. The next morning, each overnight culture was diluted 1:1000 into fresh control or CR medium, such that cells grown overnight in control medium were diluted in control medium and cells grown overnight in CR medium were diluted into CR medium, and grown for 8 h in a shaking incubator. Cell cycle division time for BY4742 control cells was approximately 95 min and for BY4742 CR cells was approximately 105 min. After outgrowth, cultures were then diluted to a cell density of approximately 2 × 103 cells/ml in water, and plated in 100-μl aliquots onto SC or FOA medium, containing either 2% or 0.05% glucose, such that cells cultured in 2% glucose were plated onto 2% glucose plates and cells cultured in CR medium were plated onto CR plates. Percent survival was calculated as the number of colonies arising on FOA medium divided by the number of colonies arising on SC medium.
Supporting Information
Figure S1 CR Has No Effect on Sir2 Activity in Respiratory-Competent or Respiratory-Deficient Cells
Transcriptional silencing of the telomeric URA3 marker in PSY316AUT was monitored by the survival of cells plated onto medium containing 5-FOA.
(42 KB PDF)
Click here for additional data file.
We would like to thank J. Smith, D. Gottschling, and T. Powers for helpful discussion. This work has been funded by a grant from the Ellison Medical Foundation. Support for this work has also been provided by the American Federation for Aging Research and the University of Washington Nathan Shock Center of Excellence for the Basic Biology of Aging. MK is supported by National Institutes of Health training grant P30 AG013280. SF is an investigator of the Howard Hughes Medical Institute. BKK is a Searle Scholar.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. MK, SF, and BKK conceived and designed the experiments. MK, DH, EOK, MT, ND, and BKK performed the experiments. MK and BKK analyzed the data. MK, DH, EAW, and BKK contributed reagents/materials/analysis tools. MK and BKK wrote the paper.
A previous version of this article appeared as an Early Online Release on October 25, 2005 (DOI: 10.1371/journal.pgen.0010069.eor).
Abbreviations
CRcalorie restriction
ERCextrachromosomal rDNA circles
ORFopen reading frame
SCsynthetic complete
==== Refs
References
Walker G Houthoofd K Vanfleteren JR Gems D 2005 Dietary restriction in C. elegans: From rate-of-living effects to nutrient sensing pathways Mech Ageing Dev 126 929 937 15896824
Merry BJ 2005 Dietary restriction in rodents—Delayed or retarded ageing Mech Ageing Dev 126 951 959 15893804
Partridge L Piper MDW Mair W 2005 Dietary restriction in Drosophila
Mech Ageing Dev 126 938 950 15935441
Sinclair DA 2005 Toward a unified theory of caloric restriction and longevity regulation Mech Ageing Dev 126 987 1002 15893363
Guarente L 2005 Calorie restriction and SIR2 genes: Towards a mechanism Mech Ageing Dev 126 923 928 15941577
Merry BJ 2004 Oxidative stress and mitochondrial function with aging—The effects of calorie restriction Aging Cell 3 7 12 14965349
Armeni T Pieri C Marra M Saccucci F Principato G 1998 Studies on the life prolonging effect of food restriction: Glutathione levels and glyoxylase enzymes in rat liver Mech Ageing Dev 101 101 110 9593316
De Cabo R Cabello R Rios M Lopez-Lluch G Ingram DK 2004 Calorie restriction attenuates age-related alterations in the plasma membrane antioxidant system in rat liver Exp Gerontol 39 297 304 15036389
Guo Z Heydari A Richardson A 1998 Nucleotide excision repair of actively transcribed versus nontranscribed DNA in rat hepatocytes: Effect of age and dietary restriction Exp Cell Res 245 228 238 9828120
Rao KS 2003 Dietary calorie restriction, DNA repair and brain aging Mol Cell Biochem 253 313 318 14619982
Yu BP 1996 Aging and oxidative stress: Modulation by dietary restriction Free Radic Biol Med 21 651 668 8891668
Youngman LD Park JY Ames BN 1992 Protein oxidation associated with aging is reduced by dietary restriction of protein or calories Proc Natl Acad Sci U S A 89 9112 9116 1409611
Sohol RS Weindruch R 1996 Oxidative stress, caloric restriction, aging Science 273 59 63 8658196
Masoro EJ Katz MS McMahan CA 1989 Evidence for the glycation hypothesis of aging from the food-restricted rodent model J Gerontol 44 B20 B22 2910985
Masoro EJ McCarter RJM Katz MS McMahan CA 1992 Dietary restriction alters the characteristics of glucose fuel use J Gerontol 47 B202 B208 1430849
Kemnitz JW Roecker EB Weindruch R Elson DF Baum ST 1994 Dietary restriction increases insulin sensitivity and lowers blood glucose in rhesus monkeys Am J Physiol 266 E540 E547 8178974
Zhang Y Herman B 2002 Ageing and apoptosis Mech Ageing Dev 123 245 260 11744038
Weindruch R Walford RL 1988 The retardation of aging and disease by dietary restriction Springfield (Illinois) Charles C. Thomas 436 p.
Jolly CA 2005 Diet manipulation and prevention of aging, cancer, and autoimmune disease Curr Opin Clin Nutr Metab Care 8 382 387 15930962
Mattson MP 2000 Emerging neuroprotective strategies for Alzheimer's disease: Dietary restriction, telomerase activation, and stem cell therapy Exp Gerontol 35 489 502 10959037
Saffrey MJ 2004 Ageing of the enteric nervous system Mech Ageing Dev 125 899 906 15563936
McCay CM Crowell MF Maynard LA 1935 The effect of retarded growth upon the length of life and upon ultimate size J Nutr 10 63 79
Masoro EJ 1998 Influence of caloric intake on aging and on the response to stressors J Toxicol Environ Health B Crit Rev 1 243 257 9644329
Mattson MP Duan W Guo Z 2003 Meal size and frequency affect neuronal plasticity and vulnerability to disease: Cellular and molecular mechanisms J Neurochem 84 417 431 12558961
Murphy CT McCarroll SA Bargmann CI Fraser A Kamath RS 2003 Genes that act downstream of DAF-16 to influence the lifespan of Caenorhabditis elegans
Nature 424 277 283 12845331
Kenyon C 2005 The plasticity of aging: Insights from long-lived mutants Cell 120 449 460 15734678
Braeckman BP Houthoofd K Vanfleteren JR 2001 Insulin-like signaling, metabolism, stress resistance and aging in Caenorhabditis elegans
Mech Ageing Dev 122 673 693 11322992
Kaeberlein M Powers RW Steffen KK Westman EA Hu D 2005 Regulation of yeast replicative life span by Tor and Sch9 in response to nutrients Science In press.
Meissner B Boll M Daniel H Baumeister R 2004 Deletion of the intestinal peptide transporter affects insulin and TOR signaling in Caenorhabditis elegans
J Biol Chem 279 36739 36745 15155758
Vellai T Takacs-Vellai K Zhang Y Kovacs AL Orosz L 2003 Genetics: Influence of TOR kinase on lifespan in C. elegans
Nature 426 620
Kapahi P Zid BM Harper T Koslover D Sapin V 2004 Regulation of lifespan in Drosophila by modulation of genes in the TOR signaling pathway Curr Biol 14 885 890 15186745
Sharp ZD Bartke A 2005 Evidence for down-regulation of phosphoinositide 3-kinase/Akt/mammalian target of rapamycin (PI3K/Akt/mTOR)-dependent translation regulatory signaling pathways in Ames dwarf mice J Gerontol A Biol Sci Med Sci 60 293 300 15860463
Lin SJ Defossez PA Guarente L 2000 Requirement of NAD and SIR2 for life-span extension by calorie restriction in Saccharomyces cerevisiae
Science 289 2126 2128 11000115
Rogina B Helfand SL 2004 Sir2 mediates longevity in the fly through a pathway related to calorie restriction Proc Natl Acad Sci U S A 101 15998 16003 15520384
Guarente L Picard F 2005 Calorie restriction—The SIR2 connection Cell 120 473 482 15734680
Kaeberlein M Kirkland KT Fields S Kennedy BK 2004 Sir2-independent life span extension by calorie restriction in yeast PLoS Biol 2 E296 15328540
Kaeberlein M Andalis AA Fink GR Guarente L 2002 High osmolarity extends life span in Saccharomyces cerevisiae by a mechanism related to calorie restriction Mol Cell Biol 22 8056 8066 12391171
Smith JS Brachmann CB Celic I Kenna MA Muhammad S 2000 A phylogenetically conserved NAD+ -dependent protein deacetylase activity in the Sir2 protein family Proc Natl Acad Sci U S A 97 6658 6663 10841563
Landry J Sutton A Tafrov ST Heller RC Stebbins J 2000 The silencing protein SIR2 and its homologs are NAD-dependent protein deacetylases Proc Natl Acad Sci U S A 97 5807 5811 10811920
Imai S Armstrong CM Kaeberlein M Guarente L 2000 Transcriptional silencing and longevity protein Sir2 is an NAD-dependent histone deacetylase Nature 403 795 800 10693811
Kaeberlein M McVey M Guarente L 1999 The SIR2/3/4 complex and SIR2 alone promote longevity in Saccharomyces cerevisiae by two different mechanisms Genes Dev 13 2570 2580 10521401
Sinclair DA Guarente L 1997 Extrachromosomal rDNA circles—A cause of aging in yeast Cell 91 1033 1042 9428525
Michel AH Kornmann B Dubrana K Shore D 2005 Spontaneous rDNA copy number variation modulates Sir2 levels and epigenetic gene silencing Genes Dev 19 1199 1210 15905408
Kennedy BK Austriaco NR Zhang J Guarente L 1995 Mutation in the silencing gene SIR4 can delay aging in S. cerevisiae
Cell 80 485 496 7859289
Couzin J 2004 Scientific community. Aging research's family feud Science 303 1276 1279 14988530
Lin SJ Kaeberlein M Andalis AA Sturtz LA Defossez PD 2002 Calorie restriction extends Saccharomyces cerevisiae life span by increasing respiration Nature 418 344 348 12124627
Lin SJ Ford E Haigis M Liszt G Guarente L 2004 Calorie restriction extends yeast life span by lowering the level of NADH Genes Dev 18 12 16 14724176
Anderson RM Latorre-Esteves M Neves AR Lavu S Medvedik O 2003 Yeast life-span extension by calorie restriction is independent of NAD fluctuation Science 302 2124 2126 14605207
Anderson RM Bitterman KJ Wood JG Medvedik O Sinclair DA 2003 Nicotinamide and PNC1 govern lifespan extension by calorie restriction in Saccharomyces cerevisiae
Nature 423 181 185 12736687
Bitterman KJ Anderson RM Cohen HY Latorre-Esteves M Sinclair DA 2002 Inhibition of silencing and accelerated aging by nicotinamide, a putative negative regulator of yeast sir2 and human SIRT1 J Biol Chem 277 45099 45107 12297502
Landry J Slama JT Sternglanz R 2000 Role of NAD(+) in the deacetylase activity of the SIR2-like proteins Biochem Biophys Res Comm 278 685 690 11095969
Gallo CM Smith DL Jr. Smith JS 2004 Nicotinamide clearance by Pnc1 directly regulates Sir2-mediated silencing and longevity Mol Cell Biol 24 1301 1312 14729974
Kaeberlein M Kennedy BK 2005 Large-scale identification in yeast of conserved ageing genes Mech Ageing Dev 126 17 21 15610758
Kaeberlein M Andalis AA Liszt G Fink GR Guarente L 2004
Saccharomyces cerevisiae SSD1-V confers longevity by a Sir2p-independent mechanism Genetics 166 1661 1672 15126388
Kaeberlein M McDonagh T Heltweg B Hixon J Westman EA 2005 Substrate-specific activation of sirtuins by resveratrol J Biol Chem 280 17038 17045 15684413
Tzagoloff A Myers AM 1986 Genetics of mitochondrial biogenesis Annu Rev Biochem 55 249 285 2427014
de Zamaroczy M Bernardi G 1986 The primary structure of the mitochondrial genome of Saccharomyces cerevisiae
Gene 47 155 177 3549452
Kaeberlein M Kirkland KT Fields S Kennedy BK 2005 Genes determining yeast replicative life span in a long-lived genetic background Mech Ageing Dev 126 491 504 15722108
Williams AC Ramsden DB 2005 Nicotinamide homeostasis: A xenobiotic pathway that is key to development and degenerative diseases Med Hypotheses 65 353 362 15922112
Wallace DC 2001 A mitochondrial paradigm for degenerative diseases and ageing Novartis Found Symp 235 247 266 11280029
Wallace DC Shoffner JM Trounce I Brown MD Ballinger SW 1995 Mitochondrial DNA mutations in human degenerative diseases and aging Biochim Biophys Acta 1271 141 151 7599200
Taylor RW Turnbull DM 2005 Mitochondrial DNA mutations in human disease Nat Rev Genet 6 389 402 15861210
Kirchman PA Kim S Lai CY Jazwinski SM 1999 Interorganelle signaling is a determinant of longevity in Saccharomyces cerevisiae
Genetics 152 179 190 10224252
Winzeler EA Shoemaker DD Astromoff A Liang H Anderson K 1999 Functional characterization of the S. cerevisiae genome by gene deletion and parallel analysis Science 285 901 906 10436161
Sikorski RS Hieter P 1989 A system of shuttle vectors and yeast host strains designed for efficient manipulation of DNA in Saccharomyces cerevisiae
Genetics 122 19 27 2659436
|
0
|
PMC1287957
|
CC BY
|
2021-01-05 07:59:37
|
no
|
PLoS Genet. 2005 Nov 25; 1(5):e73
|
latin-1
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010073
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1643588710.1371/journal.pmed.0030003Research ArticleEcologyInfectious DiseasesEpidemiology/Public HealthStatisticsInfectious DiseasesStatisticsPublic HealthLimits to Forecasting Precision for Outbreaks of Directly Transmitted Diseases Forecast Precision for Disease OutbreaksDrake John M
1
1National Center for Ecological Analysis and Synthesis, Santa Barbara, California, United States of AmericaKulldorff Martin Academic EditorHarvard Medical SchoolUnited States of AmericaE-mail: [email protected]
Competing Interests: The author has declared that no competing interests exist.
Author Contributions: JMD designed the study, analyzed the data, and wrote the paper.
1 2006 22 11 2005 3 1 e314 5 2005 27 9 2005 Copyright: © 2006 John M. Drake.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.
The Difficulties of Predicting the Outbreak Sizes of Epidemics
Background
Early warning systems for outbreaks of infectious diseases are an important application of the ecological theory of epidemics. A key variable predicted by early warning systems is the final outbreak size. However, for directly transmitted diseases, the stochastic contact process by which outbreaks develop entails fundamental limits to the precision with which the final size can be predicted.
Methods and Findings
I studied how the expected final outbreak size and the coefficient of variation in the final size of outbreaks scale with control effectiveness and the rate of infectious contacts in the simple stochastic epidemic. As examples, I parameterized this model with data on observed ranges for the basic reproductive ratio (R
0) of nine directly transmitted diseases. I also present results from a new model, the simple stochastic epidemic with delayed-onset intervention, in which an initially supercritical outbreak (R
0 > 1) is brought under control after a delay.
Conclusion
The coefficient of variation of final outbreak size in the subcritical case (R
0 < 1) will be greater than one for any outbreak in which the removal rate is less than approximately 2.41 times the rate of infectious contacts, implying that for many transmissible diseases precise forecasts of the final outbreak size will be unattainable. In the delayed-onset model, the coefficient of variation (CV) was generally large (CV > 1) and increased with the delay between the start of the epidemic and intervention, and with the average outbreak size. These results suggest that early warning systems for infectious diseases should not focus exclusively on predicting outbreak size but should consider other characteristics of outbreaks such as the timing of disease emergence.
Using a mathematical model, John Drake shows that early warning systems for infectious diseases should not focus exclusively on predicting outbreak size but should consider other characteristics of outbreaks such as the timing of disease emergence.
==== Body
Introduction
The epidemiological responsibility to forecast disease outbreaks is an onerous one. Because of the devastating consequences and high costs of disease, predicting outbreaks is a chief goal for public-health planning and emergency preparedness. Thus, quantitative forecasting and development of early warning systems (EWSs) for disease outbreak is a high priority for research and development [1]. According to the World Health Organization, the primary goals of EWSs are to predict the timing of the outbreak and the magnitude of the outbreak [1]. Intuition suggests that for directly transmissible diseases, the magnitude of the outbreak will be extremely difficult to predict because of the stochastic process of infectious contacts [2,3]. This idea is consistent with the recent finding of Sultan et al. [4] that although the timing of annual meningitis outbreaks in West Africa was highly predictable, the final outbreak size varied greatly from year to year. Here I study fundamental limits to forecast precision for (eventually) controlled outbreaks, first theoretically, then using nine well-studied infectious diseases as examples. Finally, I consider a new model that more realistically represents actual outbreaks of emerging infections.
The reason that final outbreak size is generally not predictable is that the eventual dynamics of the outbreak are highly sensitive to the seemingly random sequence of infectious contacts and removal of infectious individuals in the early, typically unobserved stages of the outbreak [3]. Clearly, the final size of an outbreak depends on numerous aspects of the social structure of the population, the environment, and disease- or strain-specific characteristics. Among the more important factors are seasonal climate fluctuations, transmissibility and virulence of the pathogen, population dynamics and structure of the host population, physiological and immunological status of potential hosts, and the social networks of contacts between infectious and susceptible individuals [5–8]. Accordingly, the deterministic approach to epidemic modeling regards the spread of infectious diseases as completely determined by the average effects of these factors on the basic reproductive ratio (R
0) together with initial conditions. Deterministic models of epidemics have provided insight into such important topics as the design of vaccination campaigns and the effect of age structure on epidemic dynamics [5]. From the perspective of EWSs, the timing and average severity of outbreaks also might be modeled quite accurately with deterministic models. However, for emerging diseases or for diseases prone to sudden outbreak, numerical predictions of final outbreak size derived from deterministic models will often deviate substantially from the observed outbreak size [3,9].
In contrast, the stochastic theory of epidemics represents the population as a statistical ensemble with constant or regular average properties but probabilistic changes in disease status for individuals. As a result, properties of the ensemble, such as the final epidemic size, are probabilistic as well [10–14]. Thus, stochastic models quantify the likelihood of outbreaks that deviate from the expected final size [9,15]. Such information about the variation in final outbreak size—its predictability—is crucial if disease forecasting is to be relied upon for planning interventions.
The stochastic theory of epidemics can therefore be used to understand the theoretical limits to forecasting precision for disease outbreaks, including EWSs or forecasts based on the developing epidemic curve as case reports accumulate. I studied how precision in the forecasted final outbreak size for transmissible diseases depends on two dynamical features of outbreaks: the contact rate (β) and the rate of removal (γ) in the simple stochastic epidemic. Next, I developed models of forecast precision for nine outbreak-prone diseases (chicken pox, diphtheria, measles, mumps, poliomyelitis, rubella, scarlet fever, smallpox, and whooping cough) and used removal rate as a control parameter to relate intervention effectiveness to final outbreak size and forecast precision. Finally, I developed a new model to understand how delays in implementing interventions affect final outbreak size and forecast prevision.
Methods
Model
The simplest realistic model for outbreaks with a small number of initially infectious individuals is the simple stochastic epidemic with contact rate β and removal rate γ, which do not change appreciably over the time scale of the outbreak [15,16]. This model is a good approximation if the outbreak meets the following criteria, which are reasonable for modern outbreaks that are rapidly controlled. First, we assume that infectious contacts and removal of infectious individuals are approximately independent in time so that the outbreak is Markovian (compare [17–19]). Second, the rate at which infectious individuals are removed from the population exceeds the rate at which infectious contacts occur (β < γ). Finally, the population is sufficiently large that the number of individuals ultimately infected is not more than a negligible fraction of the susceptible population (i.e., per capita transmission rates are approximately independent of the density of infected individuals). Then, the outbreak is a homogeneous birth–death process (the simple stochastic epidemic) with mean (M) and variance (V) of the final outbreak size given by [10]:
and
Properties of the final size distribution for other classes of epidemics can be found in [11–14,17,20].
The solution given by equations 1 and 2 is for an outbreak in which either (1) epidemiological parameters are naturally such that always R
0 < 1, or (2) public health policy is applied consistently so that intervention is constant and under policy conditions R
0 < 1. For many emerging diseases this is not the case. Rather, initially R
0 > 1, but through intervention that was established some measurable time after the outbreak started, the reproductive ratio is reduced below the epidemic threshold (e.g., severe acute respiratory syndrome [SARS]). This case is considerably more complicated and, to my knowledge, no simple formulas have been obtained for the mean and variance of the final outbreak size. However, it is reasonably straightforward to solve the equations computationally, and a range of conditions can be studied. Below, I consider a case that is more applicable to forecasting emerging diseases, the simple stochastic epidemic with delayed-onset intervention in which there is a constant rate of infectious contacts (β) and a removal rate (γ) that depends on the time since the outbreak began. Specifically, at the start of the outbreak the removal rate is some value less than the rate of infectious contacts and remains constant until some intervention is applied a time t*−t
0 later, after which the removal rate is some constant value greater than β, i.e., γ(t) = γ1
I(t≤t
*) + γ2
I(t>t
*)
, where I is an indicator function equal to one if its argument is true and zero otherwise. Then, we can study the size of the outbreak as a function of the control parameter t*, the time at which intervention is initiated.
Analysis
A measure of precision should quantify the relative magnitude of deviations from an expected value. The coefficient of variation
is a measure of forecast precision that can be interpreted as relative dispersion independent of the magnitude of the data [21]. I used the theoretical CV for final outbreak size obtained from equations 1 and 2, which depends only on the ratio γ/β = R
0
−1 and not on the individual parameter values, to study how forecast precision depends on outbreak characteristics and to estimate forecast precision for nine infectious diseases under different levels of control, represented by increasing γ (see Figure S1). This measure assumes β and γ are known exactly. For individual outbreaks, in which β and γ are not precisely known and the model is only an approximation to the structure of the contact process, violations of modeling assumptions such as the Markov assumption and the lack of an explicit incubation period further erode forecast reliability. Thus this measure represents a theoretical upper bound on forecast precision that will not be attainable in practice.
Although every outbreak will be different as a result of evolution of the etiological agent, changes in social behavior, timing, and the ecological and geographical context in which the outbreak starts, many epidemic parameters (most famously R
0), are reasonably conserved across outbreaks of the same disease. Here, I treat the removal rate γ as a control parameter because it is crucially related to interventions, and estimate β, which is assumed to depend on uncontrollable aspects of the outbreak. The variable β, which is the individual rate of infectious contacts [22], is related to the transmission rate (β0) by the equation β = β0
N, where N is total population size or density in the standard theory (e.g., [5]). This quantity is related to the basic reproductive ratio R
0 by the equation:
Where removal results from recovery of the diseased individual, we can estimate γ from the duration of the incubation (τ1), latent (τ2), and infectious (τ3) periods with the equation γ = (τ1+τ2+τ3)−1
. Estimates of R
0 have been obtained for numerous directly transmitted diseases [5]. Assuming these estimates are based on the natural course of the disease (i.e., without direct intervention), we can rearrange this equation and substitute for γ to obtain an estimate of β:
Given that reported values for these variables vary somewhat, we put an upper bound on β by choosing the highest reported value of R
0 and the lowest reported values for the different τs, whereas a lower bound is obtained from the lowest reported value of R
0 and the highest reported values for the different τs. As a central estimate, I used the center of the reported interval for each variable. Estimates of the ranges of these quantities for several directly transmitted diseases were compiled by Anderson and May ([5], Tables 3.1 and 4.1). Using these values, I used equation 4 to estimate plausible ranges of β for nine directly transmitted diseases (Table 1).
Table 1 Estimates of the Range of β for Nine Directly Transmitted Diseases
I also considered the delayed-onset intervention model wherein initially β > γ (the supercritical case in which epidemic occurs with high probability), but after a time t*−t
0 intervention increases the removal rate γ so β < γ (the subcritical case in which the outbreak is brought under control). This model is a more realistic representation of many emerging outbreaks (e.g., SARS, Foot-and-Mouth disease, and Marburg virus). The solution to the simple stochastic epidemic with delayed-onset intervention can be obtained using generating functions for the probability distribution of the size of the outbreak [10]. The variance of the final outbreak size is in terms of a multiple integral, which was evaluated numerically (see Text S1). As an example, I studied two situations with contrasting initial values for R
0. First, I studied the situation with β = 0.5 and γ1 = 0.25 (R
0 = 2). Second, I studied the situation with β = 0.5 and γ1 = 0.45 (R
0 ≍ 1.1). In both cases, γ2 (the removal rate after intervention) was one, so that post-intervention reproductive ratio was 0.5.
Results
The ratio γ/β, the rate of removal compared with the rate of infection, represents the relative effectiveness of interventions. In the simple stochastic epidemic, the relative effectiveness of intervention is always greater than one because we assume that the outbreak is eventually controlled, i.e., the assumption β < γ above. Figure 1 confirms the intuition that final outbreak size declines as the relative effectiveness of intervention is increased. The CV in the final outbreak size, our measure of the imprecision with which the final outbreak size is forecasted, also declines with control effectiveness. As a benchmark, a forecast might be deemed reliable (in principle) where the CV is less than one, which occurs for
. Figures 2 and 3 show plots of the final outbreak size and the CV over the interval of estimated βs for each of nine directly transmitted diseases. It is important to underscore that the intervals in Figures 2 and 3 represent uncertainty about the value of the parameter β, not variation from stochastic fluctuations. Further understanding of these diseases might allow us to reduce this source of uncertainty by obtaining more precise estimates. In contrast, the CV in Figure 3 represents the range of final outbreak sizes that can result from the stochastic infection process for a fixed set of parameters. In principle, no amount of detailed information about transmission or other ensemble epidemic parameters can reduce this uncertainty.
Figure 1 Expected Final Outbreak Size and CV in the Final Outbreak Size as a Function of Intervention Effectiveness
The expected final outbreak size (solid line) and CV in the final outbreak size (dashed line) are shown as a function of intervention effectiveness (the ratio of the removal rate and contact rate γ/β) for the simple stochastic epidemic. The light horizontal line designates the benchmark where CV = 1.
Figure 2 Expected Final Outbreak Size for Nine Directly Transmitted Diseases as a Function of the Removal Rate
The expected final outbreak size (y-axis) for nine directly transmitted diseases is represented as a function of the removal rate (x-axis). Estimates are bounded by minimum and maximum estimates (dashed lines) of the contact rate β based on published estimates of R
0.
Figure 3 CV in Final Outbreak Size as a Measure of Forecast Precision for Outbreaks of Nine Directly Transmitted Diseases as a Function of Removal Rate
The CV in final outbreak size (y-axis) is a measure of forecast precision, shown here for outbreaks of nine directly transmitted diseases as a function of removal rate (x-axis). Estimates are bounded by minimum and maximum estimates (dashed lines) of the infectious contact rate β based on estimates of R
0. The horizontal line indicates CV = 1.
Numerical analysis of the delayed-onset intervention model showed that (1) the average outbreak size increased with the delay between the start of the outbreak and the start of intervention (Figure 4A), and (2) the CV (in our examples) was everywhere greater than one and increased with the time delay between the start of the outbreak and intervention, but at a declining rate (Figure 4B). The first result is straightforward: The delay between initial infection and intervention increases the total number of secondary (tertiary, etc.) infections that are increasing as a multiplicative process. The explanation of the second result is that the CV in outbreak size scales as the square root of the variance in outbreak size and as the inverse of the average outbreak size. As the average outbreak gets larger the CV increases but at a declining rate (Figure 4C). This effect is mediated by the reproductive ratio of the outbreak, so that the outbreak with the lower R
0 had a lower average outbreak size (Figure 4A), but larger CV (Figure 4B and 4C). Thus, in the sense that the CV measures the predictability of the outbreak, we found that subcritical and controlled outbreaks (R
0 < 1 and R
0 close to 1, respectively) were less predictable (have lower CV) than supercritical (R
0 >> 1) outbreaks of comparable size.
Figure 4 Effect of Time Delay until Intervention on Outbreak Size
Effect of time delay until intervention on outbreak size is contrasted for outbreaks with R
0 = 2 (solid lines) and R
0 ≍ 1.1 (dashed lines).
(A) Average outbreak size (y-axis) increases with the number of days until intervention (x-axis).
(B) CV in outbreak size (y-axis) increases with the number of days until intervention (x-axis).
(C) CV in outbreak size (y-axis) increases at a declining rate (i.e., levels off) as the average outbreak size increases (x-axis). Note that the CV in final outbreak size increases faster in the outbreak with lower R
0.
Discussion
Using theoretical models, I found that unless controls are extremely effective, limits to forecast precision result in highly uncertain estimates of final outbreak size. Specifically, for the simple stochastic epidemic (subcritical case), unless the removal rate is greater than approximately 2.41 times the effective contact rate, the CV of final outbreak size will be greater than one. Imprecision in the delayed-onset intervention model was typically even greater.
Reliable forecasts of outbreaks based on initial cases and/or EWSs could potentially save many lives by increasing preparedness for outbreaks when and where they are most likely or most severe. According to the World Health Organization, forecasts will be most useful when they accurately predict the final size of the outbreak [1]. However, the findings reported here suggest that precise predictions may be unattainable because of high variance in the final outbreak size of directly transmissible diseases, even under the (unreasonable) assumption of perfect information about macroscopic epidemic parameters.
This result does not apply to diseases that are not directly transmitted (e.g., vector-borne illnesses) or to diseases in which parameters change as the outbreak progresses (e.g., SARS [23]). Parameters might change for at least two reasons. First, for emerging infections, about which little is known at the start of the outbreak, increasing ability to diagnose and treat infected patients and the dissemination of information to the public will result in increasing the removal rate. Thus, for example, in the 2003 SARS outbreak, the average lag between onset of symptoms and hospital isolation was initially around 6 d but declined to around 2 d by the fourth wk of the outbreak [23,24]. Second, in outbreaks that ultimately infect a large portion of the population, the rate of infectious contacts will decline as the number of cases increases, diluting the susceptible population. These examples represent important violations of modeling assumptions adopted here and are represented by the inhomogeneous [10,22] and general [15,16] stochastic epidemics respectively. Forecasting precision for these situations is an important topic for research.
Generally, these violations of the simple stochastic epidemic must be considered on a case-by-case basis. We studied one realistic example (the simple stochastic epidemic with delayed-onset intervention) in which an initially supercritical outbreak (R
0 > 1) is controlled by public health measures that increase the rate at which infectious individuals are removed from the population to a level ensuring the outbreak will eventually die out. This is a reasonably realistic model for dynamics of emerging infections with a short incubation period. For two representative examples, we found that the average outbreak size scaled approximately exponentially with the delay between the start of the outbreak and the implementation of intervention (note the log scale of the y-axis in Figure 4A), underscoring the importance of rapid intervention. Intuitively, when R
0 was high the average outbreak size increased faster than when R
0 was low. We also found that the CV in the final outbreak size increased with the lag between initial infection and control, but was smaller in the case with high R
0 than in the case with low R
0. Indeed, for the delayed-onset case with relatively high R
0 (R
0 = 2) the CV seemed to level off at a delay of around 15–20 d, although this was not shown in the case with lower R
0 (Figure 4B and 4C), probably because a longer delay would be required to reach such an asymptote.
In conclusion, the fundamental limit to forecasting precision obtained here represents only variation that results from the stochastic contact process and not from uncertainty about the underlying model or parameter values (compare [3]). These sources of uncertainty will further diminish precision. Further, these results underscore that rapidly implementing control measures has value not only for decreasing the final size of the outbreak, which is the primary goal, but also for decreasing variation in the final size of the outbreak, which is information that can be used to tailor control measures and reduce potential losses. Although these limits to forecast precision should lead to interpreting predictions cautiously—whether derived from statistical analysis, epidemic modeling, computer simulation, or expert opinion—they should not hinder the development of greater and more reliable systems for forecasting outbreaks of infectious disease because there are many features of outbreaks that might be reliably predicted.
Supporting Information
Figure S1 CV of Final Outbreak Size as a Function of R
0
I was unable to obtain a simple relation for the coefficient of variation (CV) in the outbreak size of the subcritical simple stochastic epidemic in terms of the basic reproductive ratio R
0. Numerical results confirm that the CV of final outbreak size depends only on the ratio β and γ (i.e., on R
0). This plot represents the information in Figure 1 as a function of R
0. The value
is the value at which CV equals exactly one. In this sense, outbreaks with R
0 ≤ R
0* are predictable while outbreaks with R
0 > R
0* are unpredictable.
(18 KB PDF).
Click here for additional data file.
Text S1 Numerical Methods to Obtain Variance in Outbreak Size in the Delayed-Onset Intervention Model
(102 KB PDF).
Click here for additional data file.
Patient Summary
Background
Early warning systems that are used to look for outbreaks of infectious diseases are important in public-health planning. One of the most important things that such early warning systems try to predict is the final size of the outbreak. However, for diseases transmitted directly from person to person (rather than via a mosquito, for example), the precision with which the final size can be predicted is often very low.
Why Was This Study Done?
This researcher wanted to study how predictable the final outbreak size of an epidemic is if the effectiveness of control measures and the average number of infectious contacts are known.
What Did the Researcher Do and Find?
He developed a mathematical model that took into account the variation in the infectiousness of nine well-studied infectious diseases. He found that for any outbreak that increases slowly, precise forecasts of the final outbreak size will be impossible. This result was especially true for epidemics in which there was a substantial delay in intervention after infection occurred, and the precision of the forecast got worse as the delay between the start of the epidemic and intervention increased, and with the average outbreak size.
What Do These Findings Mean?
These results suggest that early warning systems for infectious diseases should not focus just on trying to predict outbreak size because this estimate may be inaccurate, but rather they should instead try to predict other characteristics of outbreaks. These results will be of use to people trying to plan for infectious disease outbreaks, but will not affect how patients are managed individually.
Where Can I Get More Information Online?
Based in the United States, the Centers for Disease Control and Prevention (CDC) has a Web site that gives background on how the CDC investigates disease outbreaks, along with details of individual diseases:
http://www.cdc.gov
The World Health Organization has interesting information on early warning systems:
http://www.who.int/csr/alertresponse/en/
In the United Kingdom, the Health Protection Agency has a similar function and gives details on investigations of infectious diseases:
http://www.hpa.org.uk/infections/
The research was conducted while the author was a Postdoctoral Associate at the National Center for Ecological Analysis and Synthesis, a Center funded by the National Science Foundation (Grant #DEB-0072909), the University of California, and the Santa Barbara campus. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Citation: Drake JM (2006) Limits to forecasting precision for outbreaks of directly transmitted diseases. PLoS Med 3(1): e3.
Abbreviations
CVcoefficient of variation
EWSearly warning system
SARSsevere acute respiratory syndrome
==== Refs
References
World Health Organization Using climate to predict infectious disease outbreaks: A review 2004 Geneva World Health Organization Available: http://www.who.int/globalchange/publications/oeh0401/en/ . Accessed 5 October 2005
Drake J Fundamental limits to the precision of early warning systems for epidemics of infectious diseases PLoS Med 2005 2 e144 10.1371/journal.pmed.0020144 15916474
Meyers LA Pourbohloul B Newman MEJ Skowronski DM Brunham RC Network theory and SARS: Predicting outbreak diversity J Theor Biol 2005 232 71 81 15498594
Sultan B Labadi K Guegan JF Janicot S Climate drives the meningitis epidemics onset in West Africa Plos Med 2005 2 43 49
Anderson R May R Infectious diseases of humans: Dynamics and control 1991 Oxford (United Kingdom) Oxford University Press 757
Dowell SF Seasonal variation in host susceptibility and cycles of certain infectious diseases Emerg Infect Dis 2001 7 369 374 11384511
Newman MEJ Spread of epidemic disease on networks Phys Rev E 2002 66 (016128)
Dowell SF Whitney CG Wright C Rose CE Schuchat A Seasonal patterns of invasive pneumococcal disease Emerg Infect Dis 2003 9 573 579 12737741
Isham V Assessing the variability of stochastic epidemics Math Biosci 1991 107 209 224 1806114
Kendall D On the generalized birth-and-death process Ann Math Stat 1948 19 1 15
Ludwig D Final size distributions for epidemics Math Biosci 1975 23 33 46
Ludwig D Qualitative behavior of stochastic epidemics Math Biosci 1975 23 47 73
Ball F Nasell I The shape of the size distribution of an epidemic in a finite population Math Biosci 1994 123 167 181 7827418
Ball F O'Neill P The distribution of general final state random variables for stochastic epidemic models J Appl Probab 1999 36 473 491
Daley D Gani J Epidemic modeling 1999 Cambridge Cambridge University Press 213
Bailey N The total size of a general stochastic epidemic Biometrika 1953 40 177 185
Anderson D Watson R On the spread of a disease with gamma-distributed latent and infectious periods Biometrika 1980 67 191 198
Lloyd AL Realistic distributions of infectious periods in epidemic models: Changing patterns of persistence and dynamics Theor Popul Biol 2001 60 59 71 11589638
Lloyd AL Destabilization of epidemic models with the inclusion of realistic distributions of infectious periods Proc R Soc Lond B Biol Sci 2001 268 985 993
Ball F Clancy D The final size and severity of a generalized stochastic multitype epidemic model Adv Appl Probab 1993 25 721 736
Zar J Biostatistical analysis, 4th ed 1999 Upper Saddle River (New Jersey) Prentice Hall 663
Allen L An introduction to stochastic processes with applications to biology 2003 Upper Saddle River (New Jersey) Pearson/Prentice Hall 385
Chowell G Fenimore PW Castillo-Garsow MA Castillo-Chavez C SARS outbreaks in Ontario, Hong Kong and Singapore: The role of diagnosis and isolation as a control mechanism J Theor Biol 2003 224 1 8 12900200
Lipsitch M Cohen T Cooper B Robins JM Ma S Transmission dynamics and control of severe acute respiratory syndrome Science 2003 300 1966 1970 12766207
|
16435887
|
PMC1288026
|
CC BY
|
2021-01-05 10:40:33
|
no
|
PLoS Med. 2006 Jan 22; 3(1):e3
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0030003
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1630041510.1371/journal.pmed.0030005Research ArticleGenetics/Genomics/Gene TherapyOpthalmologyOphthalmologyGenetics
CFH Y402H Confers Similar Risk of Soft Drusen and Both Forms of Advanced AMD CFH Confers Risk of Soft Drusen and AMDMagnusson Kristinn P
1
*Duan Shan
2
3
4
Sigurdsson Haraldur
5
6
Petursson Hjorvar
1
Yang Zhenglin
2
3
7
Zhao Yu
2
3
Bernstein Paul S
2
Ge Jian
4
Jonasson Fridbert
5
6
Stefansson Einar
5
6
Helgadottir Gudleif
5
Zabriskie Norman A
2
Jonsson Thorlakur
1
Björnsson Asgeir
1
Thorlacius Theodora
1
Jonsson Palmi V
8
Thorleifsson Gudmar
1
Kong Augustine
1
Stefansson Hreinn
1
Zhang Kang
2
3
*Stefansson Kari
1
Gulcher Jeffrey R
1
*1DeCODE Genetics, Reykjavik, Iceland2Department of Ophthalmology and Visual Science, Moran Eye Center, University of Utah, Salt Lake City, Utah, United States of America3 Program in Human Molecular Biology and Genetics, Eccles Institute of Human Genetics, University of Utah, Salt Lake City, Utah, United States of America4Zhongshan Ophthalmic Center, Sun Yat-Sen University, Gaung Zhou, China5Department of Ophthalmology, National University Hospital, Reykjavik, Iceland6Faculty of Medicine, University of Iceland, Reykjavik, Iceland7Sichuan Academy of Medical Sciences and Sichuan Provincial People's Hospital, Chengdu, China8Department of Geriatrics, National University Hospital, Reykjavik, IcelandLotery Andrew Academic EditorSouthampton General HospitalUnited Kingdom*To whom correspondence should be addressed. E-mail: [email protected] (KPM), E-mail: [email protected] (KZ), E-mail: [email protected] (JRG)
Competing Interests: Some authors receive financial compensation from DeCODE Genetics. S. Duan, Z. Yang, P. S. Bernstein, J. Ge, N. A. Zabriskie, and K. Zhang declare that they have no competing interests.
Author Contributions: K. P. Magnusson, K. Zhang, K. Stefansson, and J. R. Gulcher designed the study. H. Sigurdsson, P. S. Bernstein, F. Jonasson, E. Stefansson, G. Helgadottir, N. A. Zabriskie, P. V. Jonsson, and K. Zhang enrolled and phenotyped patients. K. P. Magnusson, S. Duan, H. Petursson, Z. Yang, Y. Zhao, J. Ge, T. Jonsson, A. Björnsson, T. Thorlacius, G. Thorleifsson, A. Kong, H. Stefansson, and K. Zhang performed genotyping and analyzed the data. K. P. Magnusson, H. Petursson, H. Stefansson, K. Zhang, K. Stefansson, and J. R. Gulcher wrote the paper.
1 2006 29 11 2005 3 1 e523 5 2005 27 9 2005 Copyright: © 2006 Magnusson 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.
Pinpointing the Earliest Defects in Age-Related Macular Degeneration
Background
Age-related macular degeneration (AMD) is the most common cause of irreversible visual impairment in the developed world. The two forms of advanced AMD, geographic atrophy and neovascular AMD, represent different pathological processes in the macula that lead to loss of central vision. Soft drusen, characterized by deposits in the macula without visual loss, are considered to be a precursor of advanced AMD. Recently, it has been proposed that a common missense variant, Y402H, in the Complement Factor H (CFH) gene increases the risk for advanced AMD. However, its impact on soft drusen, GA, or neovascular AMD—or the relationship between them—is unclear.
Methods and Findings
We genotyped 581 Icelandic patients with advanced AMD (278 neovascular AMD, 203 GA, and 100 with mixed neovascular AMD/GA), and 435 with early AMD (of whom 220 had soft drusen). A second cohort of 431 US patients from Utah, 322 with advanced AMD (244 neovascular AMD and 78 GA) and 109 early-AMD cases with soft drusen, were analyzed. We confirmed that the CFH Y402H variant shows significant association to advanced AMD, with odds ratio of 2.39 in Icelandic patients (p = 5.9 × 10−12) and odds ratio of 2.14 in US patients from Utah (p = 2.0 × 10−9) with advanced AMD. Furthermore, we show that the Y402H variant confers similar risk of soft drusen and both forms of advanced AMD (GA or neovascular AMD).
Conclusion
Soft drusen occur prior to progression to advanced AMD and represent a histological feature shared by neovascular AMD and GA. Our results suggest that CFH is a major risk factor of soft drusen, and additional genetic factors and/or environmental factors may be required for progression to advanced AMD.
A common missense variant, Y402H, in the Complement Factor H gene is associated strongly with soft drusen, a precursor of advanced age-related macular degeneration
==== Body
Introduction
Age-related macular degeneration (AMD) includes a wide range of phenotypes. Early AMD is characterized mainly by the presence of soft drusen in the macula without visual loss, while advanced AMD is characterized by geographic atrophy (GA or dry AMD) and neovascular AMD (wet AMD) with visual loss. Despite the rising prevalence of AMD as a result of increasing life expectancy, its underlying pathogenesis is poorly understood and there are limited treatment options available. Nutritional supplements and antagonists of vascular endothelial growth factor have been reported to decrease visual loss in neovascular AMD somewhat [1]. Drusen are comprised of small yellowish, extracellular deposits of lipid, protein, and cellular debris, formed beneath the retinal pigment epithelium (RPE), a tissue that underlies the photoreceptor cells. Biochemical analysis of drusen have indeed resulted in identification of complement components and inflammatory modulators [2–8]. Soft drusen and pigmentary abnormalities of the RPE are considered to be an early indication of risk of developing advanced AMD. GA is a consequence of the degeneration of the photoreceptor cells and the RPE. Neovascular AMD is characterized by abnormal growth of capillaries from the choroid and by subsequent exudation of fluid, lipid, and blood.
Several genome-wide linkage scans for AMD including ours (unpublished data) have found suggestive linkage on chromosome 1q [9–15]. The addition of markers within the linkage peak led to recent reports by five groups that there is an association between a common missense variant (Y402H) in CFH and AMD in the United States [3,16–19]. The Y402H allele was present in at least 60% of AMD patients with risk ratios of between 2.0 and 3.5 for one risk allele, and with risk ratios of between 3.3 and 7.4 for carriers of two risk alleles. It has been postulated that the Y402H variant may lead to decreased binding to CRP and heparin and therefore less inhibition of the complement pathway, causing overactivity and deposition of the complement pathway proteins [20]. CFH protein has been detected in choriocapillaris and within soft drusen [3].
However, the question of how the Y402H allele contributes to the different subtypes of AMD has not been properly addressed, owing either to insufficient clinical information or to sample size, as neovascular AMD dominates the patient cohorts in most previously reported studies [3,16–19].
In order to investigate the association between CFH and AMD we performed genotype–phenotype correlations on different clinical subtypes of early and advanced AMD in US and European populations.
Methods
Patients
This study was approved by the Data Protection Authority of Iceland and the National Bioethics Committee of Iceland, and the Institutional Review Board of the University of Utah. All participants signed written informed consent prior to participation in the study. All personal identifiers associated with blood samples, medical information, and genealogy were encrypted. For samples from Iceland, encryption was carried out by the Data Protection Authority, using a third-party encryption system [21]. The Icelandic cohort recruited by DeCODE Genetics has detailed phenotypic information for 2,220 individuals, ie 1,112 patients with neovascular AMD, GA, or early AMD, and 1,108 of their unaffected relatives. Probands were recruited from a list of 2,840 consecutive patients diagnosed with AMD or early AMD at the University Eye Clinic, Reykjavik, or listed in the Icelandic Registry for the Blind during the years 1980–2001, together with relatives. Population controls were not related to the AMD cohort and did not include any first- or second-degree relative pair. A second control group (longevous controls) included 171 unrelated individuals, aged 90 y or older, who had no signs of advanced AMD, diagnosed based on their ability to see fine detail, including print, as assessed in Section D1 of the Minimum Data Set of the Resident Assessment Instrument [22]. Since the prevalence of AMD increases dramatically with age, this group represents healthy “supercontrols” for AMD. A second sample of AMD patients were recruited in Utah at the Moran Eye Center of the University of Utah, and were age-matched with controls with normal eye examinations (individuals aged 60 y or older, with no drusen or RPE changes).
All participants in both populations went through a standard examination protocol and visual-acuity measurements. Slitlamp biomicroscopy of the fundi using 90-diopter lenses was performed. A pair of stereoscopic color fundus photographs (50°) were taken, centered on the fovea using a Topcon fundus camera (Topcon TRV-50VT, Topcon Optical Company, Tokyo, Japan) by a trained ophthalmic photographer. Grading was carried out using a standard grid classification suggested by the International Age-Related Maculopathy Epidemiological Study Group for age-related maculopathy and AMD [23]. All abnormalities in the macula were recorded, including type, size, and number of drusen as well as the presence of hyperpigmentation and hypopigmentation, together with advanced AMD.
Genotyping
The Icelandic cohort that was genotyped included 581 patients with advanced AMD and 435 patients with early AMD, and allele frequencies were compared to that of either 891 population controls or 171 longevous healthy controls (Table 1). The Utah cohort of 244 patients with neovascular AMD, 78 patients with GA, and 109 patients with early AMD with soft drusen was genotyped, and allele frequencies were compared to 203 age-matched healthy controls. A TaqMan assay (Applied Biosystems, Foster City, California, United States) was performed on a 384-well GeneAmp PCR System 9700 (Applied Biosystems) used for PCR to genotype the Icelandic cohort. A direct DNA- sequencing method was used on an ABI 3100 genetic analyzer (Applied Biosystems) to genotype the Utah cohort.
Table 1 Association between Subphenotypes of AMD and CFH Y402H Variant in the Icelandic Cohort
Data Analysis
For the single-marker association of the CFH Y402H variant (rs1061170), we used Fisher's exact test to calculate one-sided p-values for the at-risk allele. As the patient cohort was recruited as families for a linkage analysis, we also repeated the test for association, correcting for the relatedness of the patients by extending a variance-adjustment procedure described previously [24] for sib-ships to apply to general familial relationships. Using the variance-adjustment procedure, the variance of the test statistic is adjusted to take into account the decrease in the effective sample size resulting from the fact that genotypes of relatives are not independent. Both unadjusted and adjusted p-values are presented for comparison. We calculate the odds ratio (OR) of the frequency of the at-risk allele as OR = p/(1−p)/s/(1−s), where p and s are the frequency of the at-risk allele in the patients and in the controls, respectively. In the case of population controls and assuming the multiplicative model, in which the risks of the two alleles of the single-nucleotide polymorphism a person carries multiply [25,26], this corresponds to an estimate of the relative risk of the mutation compared to the wild-type. Specifically, with population controls and the multiplicative model, it can be shown through Bayes' Rule that the OR, as defined above, corresponds to Risk(CT)/Risk(TT) = Risk(CC)/Risk(CT), where C is the mutated allele and T the wild-type, and Risk is the probability of disease given the genotype.
On the basis of the frequency of the at-risk allele and the relative risk, we calculate the population-attributable risk or the reduction in the number of disease cases if the at-risk allele was removed from the population, again assuming the multiplicative model. Confidence intervals of relative risks and ORs were based on the variance-adjusted tests for association, assuming a log-normal distribution. To avoid confusion and to be consistent, we report the results as OR when using healthy controls and as relative risk when using population controls.
Results
In agreement with previous reports [3–19], the Y402H allele confers an OR of 2.32 when comparing Icelandic patients with neovascular AMD to healthy controls. Based on the comparison with population controls, the relative risk of the mutation is estimated to be 1.99 with a corresponding estimated population-attributable risk of 0.48. The Y402H variant also contributes to GA, with OR of 2.27. The patient group with mixed GA/neovascular AMD gave similar results with OR of 2.92. Thus, the Y402H allele contributes equally to GA and neovascular AMD in Icelandic patients with advanced AMD (Table 1).
The comparable association to neovascular AMD and GA was replicated in the Utah cohort, giving ORs of 2.17 and 2.05, respectively (Table 2). Therefore, we conclude that the CFH variant contributes equally to GA and neovascular AMD in our European and US cohorts.
Table 2 Association between Subphenotypes of AMD and CFH Y402H Variant in the Utah Cohort
Furthermore, the Y402H variant contributes to soft drusen in early AMD, with similar ORs in the Icelandic and Utah study groups of 2.52 and 2.10, respectively (see Tables 1 and 2). In contrast, the variant does not show significant association to pigmentary changes found in early AMD. In Iceland we observed significant association (p = 0.01) to hard drusen but with a lower OR (1.57) (Table 1). A significant difference in CFH Y402H allele frequencies was observed when patients with soft drusen were compared with an unrelated set of patients with hard drusen (p = 0.011). This CFH variant also confers increased risk although this is not significant when comparing hard drusen to controls in the Utah cohort (see Table 2).
We also typed the CFH Y402H allele in four ethnically diverse populations from the International HapMap project [27]; Caucasians, residents of Utah with ancestry from northern and western Europe (59), Yorubians, residents of Nigeria (57), Japanese (31), and Chinese (44). The allele frequencies for CFH Y402H in Caucasians and Africans were similar, 39% versus 30.7%, while they were much lower in Asians—8.1% in the Japanese and 6.8% in the Chinese.
Discussion
We have shown that the CFH Y402H allele confers significant risk to neovascular AMD, GA, and soft drusen in early AMD in US and European Caucasian populations. The effect of the CFH Y402H allele on soft drusen in early AMD is similar to its effect on the advanced forms of AMD, neovascular AMD, and GA. Advanced AMD is considered to be one disease with two different end-stage lesions, i.e., choroidal neovasculariztion and GA. Leaky choroidal neovascular blood vessels between Bruch's membrane and RPE are seen typically in neovascular AMD, while GA is characterized by RPE atrophy and overlying photoreceptor loss. Soft drusen located between the RPE and Bruch's membrane are usually precursors of both forms of advanced AMD. The 5-y incidence of soft drusen in the Reykjavik Eye Study, a population-based epidemiological study of individuals aged 50 y and older [28,29], was found to be similar to that of the Beaver Dam Eye Study in the US [30], and the Blue Mountains Eye Study in Australia [31]. Neovascular AMD outnumbers GA by approximately three to one in both the Beaver Dam Eye Study and the Blue Mountains Eye Study. Similar figures were seen in most other Caucasian populations. In the Icelandic population under consideration here, however, GA was found to outnumber neovascular AMD by three to one, as reported in the Reykjavik Eye Study [28,29].It is an open question whether the high GA/neovascular AMD ratio in Iceland is due to genetic or environmental factors that increase the risk of GA or decrease the risk of neovascular AMD. Consumption of fresh fish and fishliver oil with omega-3 polyunsaturated fatty acids among Icelanders is among the highest in the world [28]. Interestingly, Seddon et al. [32] reported a trend for decreasing odds of neovascular AMD with increasing amounts of omega-3 and fresh-fish intake.
The discovery of CFH as an important AMD gene, contributing to the common form of AMD and its confirmation by several groups, is a major advance towards understanding the genetic risk and pathogenesis of AMD. It is apparent that this single common variant confers similar risk in all of the US populations tested and, furthermore, our result for the association of the CFH variant to advanced AMD in a European population is comparable to that in the US. However, the previously reported studies had not adequately addressed the effect of the CFH variant on GA or early AMD. We therefore tested the reported variant in the CFH region for association to the subtypes of AMD in both of our cohorts.
Functionally, CFH is thought to aid in keeping the complement pathway of the innate immune system in check. It is tempting to postulate that a hypothetical lower activity of complement H protein with the histidine variant may lead to increased inflammation that would contribute to the neovascular form of AMD as suggested before [17]. Alternatively, others have suggested that it may have a direct role in soft-drusen formation, which may also be linked to inflammation [3]. Our results, showing that the CFH variant contributes equally to GA and to neovascular AMD, would tend to refute the first hypothesis and lend support to the second.
Given that the protein component of soft drusen includes members of the complement system (including Complement H), we tested the effect of the CFH variant on risk of soft drusen without advanced AMD. Interestingly, this CFH variant confers risk of soft drusen with similar OR (2.52) as with both forms of advanced AMD, even before the fundoscopic findings and visual loss fulfil the criteria for advanced AMD in two independent cohorts. Conversely, there is little or no impact of the CFH variant on other features such as pigmentary changes and hard drusen. Therefore, it appears that the Y402H variant in CFH contributes to the increased risk of advanced AMD largely or entirely through its impact on the development of soft drusen as a precursor of advanced AMD.
This observation may not be surprising given that soft drusen is comprised, in part, of complement proteins including CFH and its binding partner, complement 3b [20]. However, many more elderly patients develop soft drusen than those who ultimately progress to advanced AMD. Numerous epidemiological studies have shown that the prevalence of soft drusen is two to three times greater than that of advanced AMD [33]. The prevalence of soft drusen in Caucasian populations increases with age at the same rate as AMD. Most patients with soft drusen are without any visual symptoms for decades, and only a fraction of individuals who have soft drusen will eventually progress to AMD with visual-acuity loss. For example in the Beaver Dam eye study, only 14% of the patients with soft drusen developed AMD over a 10-y period [30]. Interestingly, in persons of African origin living in United States and Barbados, prevalence of early AMD and soft drusen is slightly lower than in whites, but advanced AMD is rare [34,35]. The allele frequency of the Y402H is less (0.31 versus 0.39) in African Americans than in Caucasians. Therefore, the prevalence discrepancy between the early and advanced AMD in African Americans is consistent with our hypothesis that CFH Y402H causes soft-drusen formation, but it is not sufficient for progression to advanced AMD. To substantiate this hypothesis further, it will be interesting to correlate the prevalence of soft drusen with advanced AMD in Asian populations. Our analysis of the CFH Y402H variant demonstrated that its frequency is less in Asian populations: 0.08 in the Japanese samples and 0.07 in the Chinese. Indeed, AMD in Asians is considered to be infrequent, but careful genotype–phenotype correlation studies are needed in non-Caucasian populations.
Significant genetic influence in early AMD has been demonstrated in a classical twin study comparing concordance of 226 monozygotic and 280 dizygotic twin pairs, with soft drusen and multiple hard drusen showing strong genetic influences with heritability of 57% and 81%, respectively [36]. This is comparable to the heritability of advanced AMD [37]. However, we show that the CFH variant is a risk factor for soft drusen, but not for hard drusen or pigmentary changes, per se. Therefore, there may be other genes that influence the appearance of hard drusen and pigmentary changes. In addition, the difficulty in explaining the difference in the ratio of the prevalences in neovascular AMD and GA across populations through the CFH variant alone supports the notion that additional genetic or environmental factors contribute to the pathogenesis. It is likely that there are other important genes, yet to be found, that contribute to the risk of advanced AMD, particularly among those who already have soft drusen.
Supporting Information
Accession Numbers
The LocusLink (http://www.ncbi.nlm.nih.gov/entrez) accession number for the gene discussed in this paper, CFH, is ID 3075. The OMIM (http://www.ncbi.nlm.nih.gov/entrez) identification number for AMD, ARMD1, is 603075; and for the CFH gene is 134370.
Patient Summary
Background
The commonest cause of poor eyesight in later life in the developed world is known as age-related macular degeneration (AMD). The macula is the central part of the retina (the film-like membrane at the back of the eye) which is the most sensitive and important for sharp central vision. An early sign of AMD is what are called “drusen”—deposits of protein, fat, and cells—which doctors can see in the back of the eye. There are two types of advanced AMD: so-called “wet” or neovascular AMD (neovascular means “new vessel”) and “dry” or geographic atrophy AMD (atrophy means to waste away). Wet AMD occurs when abnormal, fragile blood vessels grow under the macula behind the retina. These blood vessels often leak blood and fluid, which lift the macula. Dry AMD occurs as the light-sensitive cells in the macula (the rods and cones) break down.
Why Was This Study Done?
Although this disease is common, little is understood about why it occurs, and current treatments have limited efficacy. Previous studies have suggested that a gene in a particular part of Chromosome 1 is linked to the chance of getting AMD. The responsible gene is Complement Factor H (CFH), which codes for a protein that is involved in keeping one part of the immune system in check. A variant of CFH has been previously shown to be present more frequently in people with advanced AMD compared to normal controls. These investigators wanted to go further, to find out whether this variant was more linked to the wet or to the dry type of AMD and to early AMD.
What Did the Researchers Do and Find?
They looked at the variant of CFH in two groups of patients with various types of AMD, 1,118 from Iceland and 431 from Utah, and compared the results with people without AMD from the same ethnic groups and age. As had been shown before, they found that one variant of this gene occurred more frequently in the wet form of AMD. However, they report two new observations. First, the variant of CFH also confers risk for the dry form of AMD and second, the variant confers similar risk to drusen in the early form of AMD.
What Do These Findings Mean?
It appears that this gene variant is important early on in the development of AMD—which makes sense as the protein for which this gene codes is involved in keeping the immune system under control. The particular variant found here may not be as efficient as the normal one—that is, it makes it more likely that inflammation will develop in the eye. These findings do not have any immediate implications for treatment, but they suggest that there are other genes that cause the severe forms of AMD with blindness.
Where Can I Get More Information Online?
Here are several Web sites with information on macular degeneration.
MedlinePlus:
http://www.nlm.nih.gov/medlineplus/ency/article/001000.htm
National Institutes of Health Senior Health:
http://nihseniorhealth.gov/agerelatedmaculardegeneration/toc.html
National Eye Institute:
http://www.nei.nih.gov/health/maculardegen/armd_facts.asp
Prevent Blindness America:
http://www.preventblindness.org/eye_problems/amdFAQ.html
Foundation Fighting Blindness:
http://www.blindness.org/MacularDegeneration/
We thank the participating AMD patients and their families. We also thank DeCODE core facilities for their contributions to this work and staff at the ophthalmology clinic at the National University Hospital, Reykjavik: Ingimundur Gislason, Thordur Sverrisson, Gudmundur Viggosson, and Helga Halblaub. The authors acknowledge the following grant support (to K. Zhang): National Institutes of Health (R01EY14428, R01EY14448, core P30EY014800, and GCRC M01-RR00064), Foundation Fighting Blindness, the Ruth and Milton Steinbach Fund, Ronald McDonald House Charities, the Macular Vision Research Foundation, Knights Templar Eye Research Foundation, Grant Ritter Fund, American Health Assistance Foundation, the Karl Kirchgessner Foundation, Val and Edith Green Foundation, and the Simmons Foundation. The funding agencies had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Citation: Magnusson KP, Duan S, Sigurdsson H, Petursson H, Yang Z, et al. (2006) CFH Y402H confers similar risk of soft drusen and both forms of advanced AMD. PLoS Med 3(1): e5.
Abbreviations
AMDage-related macular degeneration
GAgeographic atrophy
ORodds ratio
RPEretinal pigment epithelium
==== Refs
References
Fine SL Martin DF Kirkpatrick P Pegaptanib sodium Nat Rev Drug Discov 2005 4 187 188 15770779
Hageman GS Luthert PJ Victor Chong NH Johnson LV Anderson DH An integrated hypothesis that considers drusen as biomarkers of immune-mediated processes at the RPE-Bruch's membrane interface in aging and age-related macular degeneration Prog Retin Eye Res 2001 20 705 732 11587915
Hageman GS Anderson DH Johnson LV Hancox LS Taiber AJ A common haplotype in the complement regulatory gene factor H (HF1/CFH) predisposes individuals to age-related macular degeneration Proc Natl Acad Sci U S A 2005 102 7227 7232 15870199
Mullins RF Russell SR Anderson DH Hageman GS Drusen associated with aging and age-related macular degeneration contain proteins common to extracellular deposits associated with atherosclerosis, elastosis, amyloidosis, and dense deposit disease Faseb J 2000 14 835 846 10783137
Mullins RF Aptsiauri N Hageman GS Structure and composition of drusen associated with glomerulonephritis: Implications for the role of complement activation in drusen biogenesis Eye 2001 15 390 395 11450763
Anderson DH Mullins RF Hageman GS Johnson LV A role for local inflammation in the formation of drusen in the aging eye Am J Ophthalmol 2002 134 411 431 12208254
Johnson LV Leitner WP Staples MK Anderson DH Complement activation and inflammatory processes in drusen formation and age related macular degeneration Exp Eye Res 2001 73 887 896 11846519
Crabb JW Miyagi M Gu X Shadrach K West KA Drusen proteome analysis: An approach to the etiology of age-related macular degeneration Proc Natl Acad Sci U S A 2002 99 14682 14687 12391305
Klein ML Schultz DW Edwards A Matise TC Rust K Age-related macular degeneration. Clinical features in a large family and linkage to chromosome 1q Arch Ophthalmol 1998 116 1082 1088 9715689
Weeks DE Conley YP Tsai HJ Mah TS Rosenfeld PJ Age-related maculopathy: An expanded genome-wide scan with evidence of susceptibility loci within the 1q31 and 17q25 regions Am J Ophthalmol 2001 132 682 692 11704029
Majewski J Schultz DW Weleber RG Schain MB Edwards AO Age-related macular degeneration—A genome scan in extended families Am J Hum Genet 2003 73 540 550 12900797
Seddon JM Santangelo SL Book K Chong S Cote J A genomewide scan for age-related macular degeneration provides evidence for linkage to several chromosomal regions Am J Hum Genet 2003 73 780 790 12945014
Schmidt S Scott WK Postel EA Agarwal A Hauser ER Ordered subset linkage analysis supports a susceptibility locus for age-related macular degeneration on chromosome 16p12 BMC Genet 2004 5 18 15238159
Abecasis GR Yashar BM Zhao Y Ghiasvand NM Zareparsi S Age-related macular degeneration: A high-resolution genome scan for susceptibility loci in a population enriched for late-stage disease Am J Hum Genet 2004 74 482 494 14968411
Iyengar SK Song D Klein BE Klein R Schick JH Dissection of genomewide-scan data in extended families reveals a major locus and oligogenic susceptibility for age-related macular degeneration Am J Hum Genet 2004 74 20 39 14691731
Edwards AO Ritter R Abel KJ Manning A Panhuysen C Complement factor H polymorphism and age-related macular degeneration Science 2005 308 421 424 15761121
Haines JL Hauser MA Schmidt S Scott WK Olson LM Complement factor H variant increases the risk of age-related macular degeneration Science 2005 308 419 421 15761120
Klein RJ Zeiss C Chew EY Tsai JY Sackler RS Complement factor H polymorphism in age-related macular degeneration Science 2005 308 385 389 15761122
Zareparsi S Branham KE Li M Shah S Klein RJ Strong association of the Y402H variant in complement factor H at 1q32 with susceptibility to age-related macular degeneration Am J Hum Genet 2005 77 149 153 15895326
Giannakis E Jokiranta TS Male DA Ranganathan S Ormsby RJ A common site within factor H SCR 7 responsible for binding heparin, C-reactive protein and streptococcal M protein Eur J Immunol 2003 33 962 969 12672062
Gulcher JR Kristjansson K Gudbjartsson H Stefansson K Protection of privacy by third-party encryption in genetic research in Iceland Eur J Hum Genet 2000 8 739 742 11039572
Morris JN Hawes C Fries BE Phillips CD Mor V Designing the national resident assessment instrument for nursing homes Gerontologist 1990 30 293 307 2354790
Bird AC Bressler NM Bressler SB Chisholm IH Coscas G An international classification and grading system for age-related maculopathy and age-related macular degeneration. The International ARM Epidemiological Study Group Surv Ophthalmol 1995 39 367 374 7604360
Risch N Teng J The relative power of family-based and case-control designs for linkage disequilibrium studies of complex human diseases I. DNA pooling Genome Res 1998 8 1273 1288 9872982
Terwilliger JD Ott J A haplotype-based “haplotype relative risk” approach to detecting allelic associations Hum Hered 1992 42 337 346 1493912
Falk CT Rubinstein P Haplotype relative risks: An easy reliable way to construct a proper control sample for risk calculations Ann Hum Genet 1987 51 227 233 3500674
Consortium TIH The International HapMap Project Nature 2003 426 789 796 14685227
Jonasson F Arnarsson A Sasaki H Peto T Sasaki K The prevalence of age-related maculopathy in iceland: Reykjavik eye study Arch Ophthalmol 2003 121 379 385 12617709
Jonasson F Arnarsson A Peto T Sasaki H Sasaki K 5-year incidence of age-related maculopathy in the Reykjavik Eye Study Ophthalmology 2005 112 132 138 15629833
Klein R Klein BE Tomany SC Meuer SM Huang GH Ten-year incidence and progression of age-related maculopathy: The Beaver Dam eye study Ophthalmology 2002 109 1767 1779 12359593
Mitchell P Smith W Attebo K Wang JJ Prevalence of age-related maculopathy in Australia. The Blue Mountains Eye Study Ophthalmology 1995 102 1450 1460 9097791
Seddon JM Rosner B Sperduto RD Yannuzzi L Haller JA Dietary fat and risk for advanced age-related macular degeneration Arch Ophthalmol 2001 119 1191 1199 11483088
Klein R Peto T Bird A Vannewkirk MR The epidemiology of age-related macular degeneration Am J Ophthalmol 2004 137 486 495 15013873
Friedman DS Katz J Bressler NM Rahmani B Tielsch JM Racial differences in the prevalence of age-related macular degeneration: The Baltimore Eye Survey Ophthalmology 1999 106 1049 1055 10366070
Leske MC Wu SY Hyman L Hennis A Nemesure B Four-year incidence of macular changes in the Barbados Eye Studies Ophthalmology 2004 111 706 711 15051202
Hammond CJ Webster AR Snieder H Bird AC Gilbert CE Genetic influence on early age-related maculopathy: A twin study Ophthalmology 2002 109 730 736 11927430
Seddon JM Cote J Page WF Aggen SH Neale MC The US twin study of age-related macular degeneration: Relative roles of genetic and environmental influences Arch Ophthalmol 2005 123 321 327 15767473
|
16300415
|
PMC1288033
|
CC BY
|
2021-01-05 10:40:32
|
no
|
PLoS Med. 2006 Jan 29; 3(1):e5
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0030005
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0030038SynopsisGenetics/Genomics/Gene TherapyOphthalmologyGeneticsOphthalmologyPinpointing the Earliest Defects in Age-Related Macular Degeneration Synopsis1 2006 29 11 2005 3 1 e38Copyright: © 2006 PLoS Medicine.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.
CFH Y402H Confers Similar Risk of Soft Drusen and Both Forms of Advanced AMD
==== Body
In the developed world, age-related macular degeneration (AMD) is the most common cause of blindness in later life—in the United States, for example, it affects around 15 million people. Early signs of AMD in the retina are pigmentation and soft drusen deposits of protein, fat, and cellular debris. Advanced AMD was previously classified into wet and dry types; dry AMD, now known as geographic atrophy (GA) or atrophic AMD, occurs as the light-sensing cells (photoreceptors) in the macula break down. Wet AMD, now known as neovascular or exudative AMD, is caused when abnormal, fragile blood vessels grow under the macula, underneath the retina. These blood vessels often leak blood, lipid, and fluid, which lift the macula.
Late (sight-threatening) AMD is found in about 2% of all people over 50 years of age, and the incidence of the disease rises with age, occurring in 0.7%–1.4% of people aged 65–75 years, and in 11%–19% of people over 85 years of age. The neovascular form can rapidly lead to severe blindness, whereas the atrophic form progresses more slowly. Although age is the main risk factor for AMD, hypertension, smoking, and a family history of AMD also increase risk of developing the disease.
Previous genetic linkage studies have suggested that a locus on the long arm of Chromosome 1 was involved in AMD's pathogenesis. Further studies, then, refined these analyses and showed that a variant in one gene, Complement Factor H
(CFH), was present more frequently in people with advanced AMD than in normal controls. A paper in
PLoS Medicine now takes this genetic analysis further, asking whether this same variant is also associated with early AMD.
The investigators, from Iceland and the US, looked at two cohorts of patients with advanced and early AMD, and compared them with controls. They confirmed previous work on the association of the
CFH variant with advanced AMD, but furthermore showed that the same variant was associated with soft drusen and also equally with both forms of advanced AMD. The implications of these findings are that
CFH would seem to have a role early in the development of AMD, and that other genes or environmental factors are likely to determine which patients will progress to late AMD, and if so, which type of AMD. This role would fit in with what is known about CFH. CFH is a serum glycoprotein that controls the function of the alternative complement pathway and acts as a cofactor with factor I (C3b inactivator). Family syndromes have been described in which deficiency of CFH leads to spontaneous activation of the alternative pathway. The variation associated with AMD causes a milder phenotype, but may attenuate the complement inhibitory function of CFH, making complement attack of retinal pigmented epithelial and choroidal cells via the alternative pathway more likely.
Fundus images of A, normal macula; B, macula with confluent soft drusen C, macula with dry AMD D, macula with wet AMD
Ultimately, work such as this that dissects out genetic risk factors for diseases are of most value when they suggest, as here, a pathway for targeting, in a disease with few treatment options once the disease is established.
|
0
|
PMC1288034
|
CC BY
|
2021-01-05 10:40:32
|
no
|
PLoS Med. 2006 Jan 29; 3(1):e38
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0030038
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1630041410.1371/journal.pbio.0040006Research ArticleBiophysicsCell BiologyMolecular Biology/Structural BiologyPhysiologyBiochemistryIn VitroMammalsFunctional Amyloid Formation within Mammalian Tissue Amyloid Functions in Biopolymer SynthesisFowler Douglas M
1
Koulov Atanas V
2
Alory-Jost Christelle
2
Marks Michael S
3
Balch William E
2
Kelly Jeffery W [email protected]
1
1 Department of Chemistry and The Skaggs Institute of Chemical Biology, The Scripps Research Institute, La Jolla, California, United States of America2 Department of Cell Biology and the Institute for Childhood and Neglected Diseases, The Scripps Research Institute, La Jolla, California, United States of America3 Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, United States of AmericaWeissman Jonathan Academic EditorUniversity of California, San FranciscoUnited States of America1 2006 29 11 2005 29 11 2005 4 1 e618 10 2005 31 10 2005 Copyright: © 2006 Fowler 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.
The Unfolding of Amyloid's True Colors
Amyloid is a generally insoluble, fibrous cross-β sheet protein aggregate. The process of amyloidogenesis is associated with a variety of neurodegenerative diseases including Alzheimer, Parkinson, and Huntington disease. We report the discovery of an unprecedented functional mammalian amyloid structure generated by the protein Pmel17. This discovery demonstrates that amyloid is a fundamental nonpathological protein fold utilized by organisms from bacteria to humans. We have found that Pmel17 amyloid templates and accelerates the covalent polymerization of reactive small molecules into melanin—a critically important biopolymer that protects against a broad range of cytotoxic insults including UV and oxidative damage. Pmel17 amyloid also appears to play a role in mitigating the toxicity associated with melanin formation by sequestering and minimizing diffusion of highly reactive, toxic melanin precursors out of the melanosome. Intracellular Pmel17 amyloidogenesis is carefully orchestrated by the secretory pathway, utilizing membrane sequestration and proteolytic steps to protect the cell from amyloid and amyloidogenic intermediates that can be toxic. While functional and pathological amyloid share similar structural features, critical differences in packaging and kinetics of assembly enable the usage of Pmel17 amyloid for normal function. The discovery of native Pmel17 amyloid in mammals provides key insight into the molecular basis of both melanin formation and amyloid pathology, and demonstrates that native amyloid (amyloidin) may be an ancient, evolutionarily conserved protein quaternary structure underpinning diverse pathways contributing to normal cell and tissue physiology.
The authors show that native Pmel17 amyloid found in mammalian melanosomes accelerates melanin synthesis.
==== Body
Introduction
Proteins typically adopt a well-defined three-dimensional structure, but can misfold and form aggregates with a specific cross-β sheet fold called amyloid [1–4]. The multistep process of amyloidogenesis is linked to a number of diseases, including many resulting in neurodegeneration [5–7]. Nonpathogenic amyloid has not been detected in higher organisms and was unexpected because of the toxicity associated with its formation. We have discovered an abundant mammalian amyloid structure that functions in melanosome biogenesis, challenging the current view that amyloid in mammals is always cytotoxic.
Melanosomes are highly abundant mammalian cellular organelles generated in developmentally specialized cells including melanocytes and retinal pigment epithelium (RPE) [8,9] that reside in the skin and eyes. Melanosome maturation has been demonstrated to require the formation of detergent-insoluble, lumenal Pmel17 fibers [10–12], which are believed to function in polymerization of intermediates in the synthesis of the tyrosine-based polymer melanin [13,14]. Melanin serves as one of nature's chemical defenses against pathogens, toxic small molecules, and UV radiation, and is present in most eukaryotic phyla ranging from fungi to insects and humans [9,15]. The functional requirement for Pmel17 in pigmentation is also well established. In mice, a point mutation in the Pmel17/silver locus results in a progressive loss of pigmentation, apparently through loss of melanocyte viability [16–19]. Mutations in Pmel17 orthologs in chicken and zebrafish also result in hypopigmentation [20,21]. Melanosome biogenesis utilizes the secretory and endocytic pathways to direct furin-like, proprotein-convertase-mediated proteolytic processing of the transmembrane glycoprotein Pmel17 [10] in an acidic post-Golgi compartment, yielding a 28-kDa transmembrane fragment (Mβ) and an 80-kDa lumenal fragment (Mα) [12]. Mβ is degraded, whereas Mα self-assembles into fibers that form the core of mature melanosomes [8,10].
Herein we show that fibers in isolated mammalian melanosomes, consisting of Mα, have an amyloid structure. This conclusion is based on the binding of dyes that fluoresce upon interacting with a cross-β sheet structure and on our ability to reconstitute Pmel17 amyloid formation in vitro as demonstrated by a variety of biophysical techniques. The rapidity of recombinant Pmel17 fibrilization is unprecedented, consistent with a process optimized by evolution for function and to avoid the toxicity of pathological amyloidogenesis. Moreover, we have shown that reconstituted Pmel17 amyloid accelerates melanin formation in vitro, apparently by serving as a scaffold that templates the polymerization of highly reactive melanin precursors, probably influencing the resulting structure of melanin as well. Importantly, Mα amyloid may also mitigate the toxicity associated with melanin synthesis by sequestering and minimizing diffusion of highly reactive, toxic melanin precursors out of the melanosome. The utilization of the amyloid fold for a major cellular activity in mammals demonstrates that sequence and folding pathway evolution can harness this ancient structure for physiological purposes.
Results
As a first test to examine whether Mα fibers have an amyloid fold, we prepared a highly enriched melanosome fraction from homogenates of the RPE and choroidal layers from bovine eyes [22] (Figure 1). This ex vivo sampling of melanosomes enables a variety of experiments unavailable in the context of whole tissue while maintaining a high degree of physiological relevance. As expected, purified melanosomes all contained Mα (Figure 1B and 1C). The amyloid content of the melanosomes was interrogated using the amyloid-selective fluorophores thioflavin S and Congo red. These molecules preferentially bind to amyloid over other types of protein aggregates, and their fluorescence upon binding is commonly used in the laboratory and the clinic to diagnose the presence of amyloid in vitro and in vivo [23–25]. Strikingly, greater than 95% of the structures in the melanosome fraction bound thioflavin S and Congo red (Figure 1E and 1G), strongly suggesting that melanosomes contain Mα amyloid. Congo red birefringence was not observed from the stained melanosomes because their thickness is an order of magnitude smaller than the requisite 10-μm thickness achieved in tissue sections. Nevertheless, Congo red fluorescence has been shown to be as effective as birefringence in the diagnosis of amyloid diseases [23].
Figure 1 Purified Melanosomes Stain with Amyloidophilic Dyes
Melanosomes were isolated from bovine RPE and choroid and visualized using transmission electron microscopy (A; scale bar = 1 μm), differential interference contrast microscopy (DIC) (B, D, and F; scale bars = 10 μm), indirect immunofluorescence using a Pmel17-specific antibody (C), or the thioflavin S (E) or Congo red (G) amyloidophilic fluorophores. Images (B) and (C), (D) and (E), and (F) and (G) are paired.
To substantiate that thioflavin S and Congo red binding-induced fluorescence reflect the existence of Mα amyloid, we took advantage of the fact that amyloid fibers resist moderate detergent extraction [5]. When melanosomes were extracted with 1% Triton-X 100 to remove their membranes, thioflavin S–stained particle clusters recovered in the detergent-insoluble fraction showed nearly exclusive overlap with Pmel17 antibody fluorescence (Figure 2), suggesting that Mα is a component of the observed amyloid structures within melanosomes. Thioflavin S did not stain residual melanin-containing dense granules (observed by differential interference contrast microscopy) lacking Mα (Figure 2, red arrowheads in insets), demonstrating that the melanin polymer is not responsible for fluorophore binding in intact melanosomes. Bovine melanosomes lose both Mα and thioflavin S binding upon boiling with 10% sodium dodecyl sulfate (data not shown), consistent with the denaturation of the Mα amyloid fiber under these conditions. Because boiling in 10% sodium dodecyl sulfate does not alter the covalent structure of melanin, these results provide further evidence that thioflavin S is specific for the Mα amyloid component of melanosome granules.
Figure 2 Pmel17 and Thioflavin S Fluorescence Overlap in the Detergent-Insoluble Melanosome Fraction
A 1% Triton-X 100 detergent-insoluble fraction was prepared from purified melanosomes and visualized using differential interference contrast microscopy (DIC) (A), indirect immunofluorescence using a Pmel17-specific antibody (B), or thioflavin S fluorescence (C). Arrows denote Pmel17-containing insoluble clusters of variable size. Asterisks indicate the enlarged cluster shown in the lower righthand corner of each panel. In the insets, the large white arrowheads denote Pmel17-positive structures (shown in [B]) that directly overlap with thioflavin S staining (shown in [C]); the red arrowhead in (A) denotes a residual dense melanin-containing granule lacking Pmel17 (shown in [B]) that does not stain with thioflavin S (shown in [C]).
The data suggest that the Mα fibers found in melanosomes within mammalian cells have an amyloid fold. Consistent with this, we found that Mα spontaneously self-assembles into amyloid in vitro. The non-glycosylated, 442-residue lumenal fragment of Pmel17, referred to as recombinant Mα (rMα; without carbohydrate) was reconstituted from Escherichia coli inclusion bodies. Purification of rMα by gel filtration in 8 M guanidinium hydrochloride (GdmCl) was required to preserve an unfolded, nonaggregated state (Figure S1). Dilution of rMα into nondenaturing buffers resulted in exceedingly rapid amyloidogenesis (within 3 s, rate-limited by the time required for mixing) over a broad pH range as monitored by thioflavin T fluorescence (Figure 3A) and endpoint Congo red analysis (Figure 3A, inset). To ensure that rMα amyloidogenesis kinetics were not altered by the presence of GdmCl-resistant Mα seeds, stock rMα solutions purified by gel filtration in 8 M GdmCl were subjected to ultracentrifugation (500,000 g for 1 h) before the top 90% of supernatant was removed and subjected to chaotrope-dilution-initiated amyloidogenesis. Ultracentrifugation did not decrease the rate of amyloidogenesis or alter the concentration of the rMα stock. Other highly amyloidogenic, natively unfolded proteins such as Aβ and α-synuclein do not form amyloid within the dead time of mixing upon transfer from denaturing to nondenaturing conditions (Figure 3B). In fact, rMα amyloidogenesis is at least four orders of magnitude faster than that of Aβ or α-synuclein under identical physiological conditions at room temperature (Figure 3A and 3B). We are unaware of any other protein nearly so amyloidogenic [26], consistent with a functional amyloid fold optimized by evolution to avoid the formation of toxic intermediates prominent in pathogenic amyloidogenesis [7].
Figure 3 rMα Rapidly Forms Thioflavin T– and Congo Red–Positive Fibers under Nondenaturing Conditions
(A) rMα samples (in 8 M GdmCl to preserve an unfolded, nonaggregated state) were diluted by manual mixing to start an amyloid fiber formation time course (monitored by thioflavin T fluorescence) at varying pHs: pH 7.4 (black line, triangles), pH 6.0 (dark grey line, circles), and pH 4.85 (light grey line, squares); control (thioflavin T buffer) (black line, white diamonds ). The inset bar graph reflects endpoint Congo red binding of equimolar amounts of deposits of Mα formed at pH 7.4 (dark grey), Aβ 1–40 fibers associated with Alzheimer disease (light grey), and control (Congo red buffer) (black).
(B) rMα (Pmel) forms thioflavin T (ThT)–positive aggregates at least four orders of magnitude faster than either α-synuclein (α-Syn) or Aβ when all three polypeptides are diluted from 8 M GdmCl into physiological buffer (error bars represent the standard deviation of triplicate samples).
(C) Transmission electron micrograph of typical rMα amyloid fibers with an average diameter of 10 nm, formed under nondenaturing conditions.
A variety of structural techniques were employed to confirm that rMα aggregates possess an amyloid structure. Aggregate morphology was examined by electron microscopy, revealing fibers with an average diameter of 10 nm, typical of amyloid formed in vitro (Figure 3C) [27]. X-ray powder diffraction of rMα aggregates revealed a very strong reflection at 4.6 Å and a weaker reflection at 10 Å (Figure 4A), consistent with the amyloid cross-β sheet quaternary structure [28]. The presence of β-sheet structure is further supported by the far-UV circular dichroism (CD) spectrum of soluble rMα amyloid formed at low concentrations (Figure 4B) and the attenuated total reflectance Fourier transform infrared (FT-IR) spectrum of insoluble rMα amyloid, which are both fully consistent with known amyloid spectra (Figure 4C) [29]. Both methods indicate approximately 50% β-sheet content, suggesting that rMα amyloid may contain folded domains external to the fiber structure. The Ure2p prion amyloid is known to have a central amyloid core with an attached globular domain [30]. Our biophysical data revealed that the Mα fibers required for melanosome biogenesis [10,12] are amyloid.
Figure 4 rMα Fibers Have a Cross-β Sheet Structure
(A) X-ray powder diffraction of lyophilized rMα fibers formed in vitro exhibit a very strong reflection at 4.6 Å and a strong reflection at 10 Å, which is expected of an amyloid cross-β sheet structure.
(B) The far-UV CD spectra of soluble Mα aggregates formed at low concentrations to avoid precipitation support a predominantly β-sheet structure. Mα aggregates are approximately 11% α-helix, 32% β-sheet, 23% β-turn, and 33% disordered, based on curve fitting with a basis set of 43 soluble proteins. Since β-sheet content is estimated using a set of proteins not composed of cross-β sheet structures, the potential error in the estimate cannot be determined.
(C) The attenuated total reflectance FT-IR spectrum of aggregated rMα in the solid state supports a β-sheet-rich structure. Peaks in the amide III (top left, upper curve) and I (top right, upper curve) regions were identified using Fourier self-deconvolution (top left and right, middle curve) and confirmed by second derivative analysis (top left and right, bottom curve). Peak assignments are listed, and were used to fit the original spectrum using fixed Gaussian peaks at the assigned positions (bottom). Peaks assigned to β-sheet regions of the spectrum accounted for a large percentage of the total area in the amide I and III regions.
rMα amyloid was used to further investigate the native function of Mα amyloid fibers in melanogenesis. The monomeric precursor for melanin polymerization, indole-5,6-quinone (DHQ), is one of the terminal products of a series of oxidation steps initiated by the action of the type I transmembrane enzyme tyrosinase on the substrate tyrosine [15]. Melanin is thought to consist of DHQ and other intermediates polymerized upon a template of Mα fibers within the maturing melanosome (Figure 5A and 5B). To test this possibility, we employed an in vitro assay utilizing tyrosinase, 3,4-dihydroxyphenylalanine (DOPA), and rMα amyloid that recapitulates melanin formation within the melanosome (Figure 5C and 5D). A time course reveals that rMα amyloid hastens the formation of insoluble melanin when added to the melanization assay (Figure 5C), resulting in more melanin per unit time. rMα amyloid also affords more than 2.2-fold more melanin after 20 h than an equivalent amount of collagen IV, an α-helical fiber (Figure 5D). Interestingly, DHQ shares a core that is isostructural with the benzothiazole substructure of the amyloidophilic fluorophore thioflavin T (Figure 5A, box), which might account for its affinity for amyloid fibers. Recent studies have shown that thioflavin T binds in a regular fashion parallel to the amyloid fiber axis [31]. Binding of DHQ in an analogous fashion may be what enables Mα amyloid to concentrate and organize reactive DHQ or analogous reactive melanin precursors along the Mα fiber, templating their efficient covalent polymerization. Because the Mα fibers are completely buried during the process of melanogenesis they are unlikely to function as catalysts. Strikingly, α-synuclein and Aβ amyloid enhance the yield of melanin formation in a manner comparable to rMα amyloid in our in vitro melanogenesis assay (Figure 5D). Apparently, the cross-β sheet structure of amyloid, shared between Mα, Aβ, and α-synuclein fibers, functions specifically to template the synthesis of melanin in vitro. We suggest that this process occurs in vivo on Mα fibers within melanosomes.
Figure 5 Amyloid, Including rMα, Specifically Accelerates Melanin Synthesis
(A) In melanosomes, assembly of activated melanin precursors, generated by tyrosinase, occurs along Pmel17 fibers. The boxed portion of (A) illustrates the amyloid-binding dye thioflavin T and the activated melanin precursor DHQ, which possess similar core structures. This suggests an explanation for the ability of Pmel17 to concentrate and organize melanin precursors, thereby enabling melanogenesis.
(B) In vivo, melanosome maturation is a four-step process (I–IV) in which initial formation of the Pmel17 fibrillar matrix (II) enables subsequent melanin polymerization along the Pmel17 fibers (III) (Adapted with permission from [16].)
(C) A time course of melanin synthesis in vitro shows that insoluble rMα amyloid increases the amount of insoluble melanin formed per unit time (grey line) versus a control reaction lacking rMα (black line).
(D) Melanin synthesis after 20 h was also evaluated in the presence of insoluble rMα amyloid, α-synuclein amyloid, Aβ amyloid, and collagen IV α-helical fibers. The melanin precursor D,L-DOPA was incubated in the presence of the enzyme tyrosinase and the amyloid of interest at room temperature. Melanin content of each reaction condition was measured by pelleting insoluble melanin, dissolving it in 1 M NaOH, and measuring the absorbance at 350 nm. Supernatant melanin content was equal for all samples.
In (C) and (D) error bars represent the standard deviation between triplicate samples.
Discussion
The discovery of amyloid as a prominent structure in eukaryotic cells now adds the amyloid fold to the repertoire of structures used in normal mammalian cell physiology. We provide a number of lines of in vitro and ex vivo evidence to support this conclusion. Ex vivo melanosomes exhibit selective environment-sensitive thioflavin T and Congo red fluorescence and have detergent-resistant properties expected of amyloid. rMα assembles faster than any known polypeptide into amyloid fibers, at least four orders of magnitude faster than the Aβ and α-synuclein peptides associated with Alzheimer and Parkinson disease, consistent with an evolved sequence that adopts an amyloid structure. The structure produced spontaneously by Mα in vitro exhibits the characteristic X-ray fiber diffraction, CD and FT-IR spectra, and fibrillar morphology expected of amyloid. Finally, rMα amyloid fibers can hasten melanin formation in vitro by serving as a template for DHQ polymerization, thereby recapitulating the fibers' putative native function [16].
Given the toxic nature of amyloid and its precursors in both intracellular and extracellular contexts, it would be expected that Mα fibrillogenesis be highly regulated to avoid damage to the cell. Indeed, full-length Pmel17 is synthesized and trafficked to early melanosomes as a transmembrane protein incapable of self-assembly. Only when sequestered in the specialized early melanosome compartment is the amyloidogenic fragment, Mα, released by proteolysis. The rapid self-assembly of Mα in combination with its membrane sequestration presumably minimizes the toxicity usually associated with amyloidogenesis. Other proteins may also be involved in the initiation and regulation of Mα amyloidogenesis [32], as is the case for functional E. coli and Salmonella extracellular curli fibers [33], extracellular spider silk fibers [34], Sup35 amyloid in yeast [35], and, potentially, CPEB prions [36].
While functional amyloidogenesis exhibits some similarities to pathogenic amyloid formation, it also displays striking differences. In gelsolin amyloid disease, proteolysis of mutant gelsolin during secretion by the proprotein convertase furin leads to slow, unregulated extracellular pathogenic gelsolin amyloid deposition [37]. Mα amyloid formation is also initiated by proprotein convertase activity, but the product is a functional amyloid structure confined to a membrane-delimited compartment. The rapidity of rMα amyloidogenesis is likely important, as this may preclude the formation of toxic, diffusible intermediates that could compromise cellular integrity [5,7,11]. These key differences in packaging and assembly appear to enable the usage of amyloid as a major intracellular structure for normal function. Study of functional Mα amyloid is likely to provide critical insights into the pathological basis for important misfolding diseases including Huntington, Parkinson, and Alzheimer disease, where the biological contexts and folding constraints differentiating normal from pathological folds are currently not appreciated.
Mα fibers in melanosomes serve to bind and orient highly reactive melanogenic precursors, hastening their polymerization and likely influencing the resulting melanin structure. Another apparently important function of Mα amyloid is to prevent cytotoxicity associated with the process of melanin polymerization, and hence melanosome biogenesis. Highly reactive, uncharged hydrophobic melanin precursor compounds would be expected to diffuse across the membrane and enter the cytoplasm if they were not sequestered by Mα fibers, upon which they polymerize. Large excesses of melanogenic precursors have been shown to produce severe cytotoxic effects in melanizing cells [38], and Pmel17 mutations leading to minimal Mα amyloid fiber formation result in reduced melanocyte viability [16–19]. These observations can be explained by the leakage of toxic melanogenic intermediates from the melanosome as a result of insufficient sequestration by Mα amyloid. Hence, the ability of Mα fibers to bind and concentrate these reactive precursors appears to protect the cell against the toxicity that can result from melanosome biogenesis.
The discovery that a major mammalian biosynthetic pathway utilizes a cross-β sheet structure for function establishes the amyloid fold as a key protein structural motif utilized by a wide variety of organisms from prokaryotes to humans. Melanin polymer chemistry plays a wide variety of roles in an array of organisms—it is involved in pigmentation and cytoprotection in higher eukaryotes, it is critical for stress responses and structural stability in plants, and it is an integral component of the insect immune system [15]. Mα amyloid has a critical role in melanin formation in humans, and is the first example to our knowledge of an amyloid that functions in a chemical reaction, pointing the way towards the discovery of amyloid in other important functional roles. It is now apparent that the amyloid fold has been selected multiple times during evolution for a variety of functions. Given the propensity of most polypeptides to form amyloid in vitro [5], the usage of amyloid in biology may be as common as other canonical protein folds. This contrasts with the current view that there is evolutionary pressure against amyloidogenesis. We suggest that the amyloid fold is a fundamental protein structural motif with unique properties that is capable of performing a wide variety of functions. We propose the general name amyloidin for functional amyloid, with the expectation that the number and diversity of structures of this type will continue to grow.
Materials and Methods
Immunofluorescence
Bovine melanosomes were fixed in methanol at −20 °C and blocked with 5% BSA/2% normal goat serum in TBS for 10 min at room temperature. Melanosomes were incubated with a chicken polyclonal anti-Pmel17 antibody, GP100 (Zymed, San Francisco, California, United States), for 1 h at room temperature at a dilution of 1:150. The secondary antibody (goat anti-chicken IgG rhodamine, Molecular Probes, Eugene, Oregon, United States) was used at a dilution of 1:200 for 1 h at room temperature. Melanosomes were then washed and mounted with PBS and imaged.
Staining with thioflavin S and Congo red
Thioflavin S and Congo red staining were carried out as previously described [25]. Briefly, for thioflavin S, purified bovine melanosomes were thawed, washed once with PBS, and stained for 1 h in a 1% (w/v) solution of thioflavin S in water. Melanosomes were then washed twice with 80% ethanol and once with PBS. For Congo red, melanosomes were thawed, washed once with PBS, and stained using the alkaline Congo red method [25]. Melanosomes were then washed twice with absolute ethanol. Detection of amyloidin encapsulated in granules requires incubation for longer than typical times reported for extracellular pathogenic amyloid.
Melanosome purification and extraction
Melanosomes were isolated from the RPE and choroid layers of bovine eyes by sucrose density ultracentrifugation as previously described [22] with minor modifications. The RPE cell layer was collected in 0.25 M sucrose buffer (10 mM Tris [pH 7.4], 65 mM NaCl, 2 mM MgCl2, protease inhibitor cocktail [Sigma, St. Louis, Missouri, United States]) and disrupted by Dounce homogenization. The homogenate was centrifuged at 2,000 g at 4 °C for 10 min to obtain a postnuclear supernatant. The postnuclear supernatant was layered onto a sucrose step gradient (0.75 M/1.5 M/2 M sucrose) and centrifuged at 85,000 g for 1 h. The melanosome-rich fraction was collected from the 2 M layer of the gradient and washed in 0.25 M sucrose buffer. To prepare the detergent-insoluble fraction, isolated melanosomes were resuspended in extraction buffer (150 mM NaCl, 100 mM Tris, 0.1% NaN3). Triton-X 100 was added from a 10% stock solution to a final concentration of 1%. The suspension was shaken at 4 °C for 2 h. The insoluble fraction was collected by centrifugation at 10,000 g for 1 min and washed three times with extraction buffer.
rMα expression and purification
The lumenal fragment of Pmel17, rMα, consisting of amino acids 25–467 was subcloned into a pET3c vector and expressed in BL21-DE3 E. coli. Shaken cultures (1 l) were grown at 37 °C to OD600 = 0.5 in the presence of 270 μM ampicillin and then induced with 1 μM IPTG for 4 h. Cells were collected via centrifugation at 4 °C, resuspended in TBS, and frozen at −80 °C. The resuspended pellet was thawed and the cells were lysed by probe sonication. rMα formed inclusion bodies that were collected by centrifugation, and washed by resuspension followed by centrifugation twice in washing buffer (1.5 M NaCl, 50 mM KH2PO4/K2HPO4 [pH 7.4], 1% Triton-X 100) and then in TBS. The inclusion body pellet was dissolved in extraction buffer (8 M GdmCl, 50 mM KH2PO4/K2HPO4 [pH 7.4], 100 mM KCl, 5 mM EDTA) by magnetic stirring at 4 °C for 48 h. The resulting solution was centrifuged, filtered through a 0.22 μM cellulose acetate filter, and then frozen at −80 °C. After thawing, the solution was gel filtered using extraction buffer with a Superdex 200 26/60 column. Purified rMα fractions were assayed via SDS-PAGE and Western blot using the GP100 anti-Pmel17 antibody. rMα was concentrated using a 3-kDa MWCO Centricon filter (Millipore, Bedford, Massachusetts United States) and stored at room temperature.
Thioflavin T binding assays
Thioflavin T binding kinetics were assayed using a Cary Eclipse fluorimeter (Varian, Palo Alto, California, United States). Assay buffer (50 mM KH2PO4/K2HPO4, 100 mM KCl, 5 mM EDTA, 20 μM thioflavin T) at the appropriate pH was placed in a stirred cuvette. The solution was excited at 440 nm and data were collected at 485 nm. After data collection had been initiated, an aliquot of concentrated rMα in extraction buffer was rapidly added to a final concentration of 10 μM. Data shown represent several normalized traces that have been averaged. Stagnant thioflavin T binding kinetics were assayed using an Aviv (Lakewood, New Jersey, United States) ATF105 fluorimeter. Stock solutions of Aβ, α-synuclein, and rMα in 8 M GdmCl, 50 mM KH2PO4/K2HPO4 (pH 7.4), and 100 mM KCl were diluted to a final concentration of 10 μM in assay buffer at pH 7.4. The solutions were allowed to aggregate for various amounts of time, whereupon thioflavin T was added from a stock solution to a final concentration of 20 μM. The samples were excited at 440 nm and data were collected at 485 nm. Control experiments were performed to ensure that guanidine-resistant seeds were not affecting Mα amyloidogenesis rates using stock rMα solutions in 8 M GdmCl that were centrifuged for 1 h at 500,000 g. These controls gave identical time courses to experiments in which the centrifugation step was not performed.
Congo red binding assay
An aliquot of concentrated rMα was added to assay buffer (50 mM KH2PO4/K2HPO4, 100 mM KCl, 5 mM EDTA). The solution was vortexed and allowed to incubate for 5 min, whereupon Congo red was added from a concentrated stock solution to a final concentration of 20 μM. Absorbance spectra were recorded using a Hewlett-Packard (Palo Alto, California, United States) 8453 spectrometer. Units of Congo red bound were calculated using the following equation: OD540/25,295 − OD477/46,306 [39].
Fluorescence microscopy
Images were captured using a Zeiss (Oberkochen, Germany) Axiophot epi-fluorescence microscope attached to an Axiocam digital camera using the following filter configurations: Pmel17 antibody (excitation 545 nm, emission 560–625 nm), thioflavin S (excitation 436 nm, emission > 455 nm), and Congo red (excitation 530–585 nm, emission > 600 nm).
CD secondary structure analysis
CD spectra were collected on an Aviv 202A CD spectrometer. Concentrated rMα in extraction buffer was diluted into filtered, de-ionized water to a final protein concentration of 3 μM (336 mM GdmCl, 2.1 μM KH2PO4/K2HPO4, 4.2 μM KCl, 0.21 μM EDTA). Ten spectra were averaged and background corrected before analysis of secondary structure components using averages of the outputs of Cdsstr, SELCON, and CONTIN algorithms with a basis set of 43 soluble proteins [40–42].
FT-IR secondary structure analysis
Concentrated rMα in extraction buffer was dialyzed versus filtered, deionized water. Approximately 100 μL of this 50 μM rMα aggregate solution was deposited on a Ge-attenuated total reflectance infrared cell and allowed to dry under flowing nitrogen. Three thousand scans were acquired on a Nicolet Magna 550 FT-IR (Thermo Electron, Madison, Wisconsin, United States) using dried dialysis buffer as a background. The spectra were smoothed using a nine-point Savitsky-Golan algorithm, and peaks were identified using Fourier self-deconvolution as well as a second-derivative analysis. Assignments in the amide I and amide III regions were made based on literature precedent [29,43]. Spectral sections corresponding to the amide I (1,588–1,806 cm−1) and amide III (1,198–1,330 cm−1) regions were fit using Gaussian peaks with fixed positions. Gross estimates of secondary structure were made based on the relative size of the various peaks.
X-Ray powder diffraction
Concentrated rMα in extraction buffer was dialyzed versus MilliQ water. Aggregates were lyophilized, producing a fine powder. Powder diffraction of approximately 1.0 mg of rMα in quartz capillaries was recorded using a 6-kW Bruker (Madison, Wisconsin, United States) Direct Drive Rotating anode X-ray generator with a Xenocs (Sassenage, France) focusing mirror (50 kV × 100 mA, 0.3 × 3 mm focus, 0.5 mm slits, copper target) and a Mar 345-mm IP scanner. The distance from sample to scanner was 250 mm and CuKα radiation (1.5418 Å) was utilized.
Electron microscopy of rMα fibers
rMα fibers were generated by diluting (from concentrated 8 M GdmCl, 50 mM KH2PO4/K2HPO4 [pH 7.4], 100 mM KCl stock) rMα into 125 mM CH3COOH/CH3COOK buffer (pH 5.0) at a final concentration of 10 μM and allowing it to stand at room temperature for 24 h. rMα aggregates were adsorbed onto 200-mesh carbon-coated copper grids (Electron Microscopy Sciences, Hatfield, Pennsylvania, United States), stained with 1% aqueous uranyl acetate (Electron Microscopy Sciences), and visualized with a Philips (New York, New York, United States) CM100 transmission electron microscope.
Synthetic melanogenesis
The ability of rMα to enhance melanogenesis was evaluated using D,L-DOPA (Sigma) and tyrosinase (Calzyme, San Louis Obispo, California, United States). rMα (0.5 mg) was added to 1.0 ml of fresh assay buffer (5.0 mM D,L-DOPA, 125 mM CH3COOH/CH3COOK buffer [pH 5.0]) from a concentrated stock solution (8 M GdmCl, 50 mM KH2PO4/K2HPO4 [ph 7.4], 100 mM KCl). Aβ was purchased (SynPep, Dublin, California, United States) and rendered seed-free by dissolving in water, sonicating for 15 min, adjusting the pH to 10.5 using 100 mM NaOH, sonicating for 15 min, filtering through a 0.22-μm syringe filter, and finally filtering through a 10-kDa MWCO concentrator (Centricon). The resulting solution (400 μM Aβ) was mixed in equal volumes with a solution of 600 mM NaCl, 300 mM NaPO4 (pH 7.5), and 0.04% NaN3 and allowed to form amyloid with rocking at 37 °C for 4 d. α-Synuclein was expressed and purified from E. coli using a procedure adapted from Lashuel et al. [44]. Cultures were grown to OD600 = 0.5 and then induced for 12 h with 1 mM IPTG. Cells were harvested and lysed using probe sonication at 4 °C, and the supernatant collected. Streptomycin sulfate (1% [w/v]) was added to the supernatant, which was then stirred at 4 °C for 60 min. The precipitated proten was removed by centrifugation, NH4SO4 (0.129 g/ml) was added, and the solution was stirred at 4 °C for 60 min. The supernatant was decanted and the pellet resuspended in one-tenth of the culture volume of 10 mM Tris buffer (pH 7.4). The protein was then purified on a source Q column (buffer A, 10 mM Tris [pH 7.4]; buffer B, A + 1 M NaCl). α-Synuclein-containing fractions were concentrated to one-tenth of their original volume and further purified using a Superdex 200 26/60 column with 100 mM (NH4)2CO3. The purified α-synuclein was then lyophilized and stored at −80 °C until used. Lyophilized α-synuclein was dissolved in 25 mM MES buffer (pH 6.0) and induced to form amyloid by rocking at 37 °C for 48 h. The amyloid nature of the α-synuclein and Aβ aggregates was tested by far-UV CD (which showed β-sheet) and TEM or AFM (which showed fibers). α-Synuclein or Aβ amyloid was collected by centrifugation (16,000 g for 15 min) and resuspended in assay buffer at a final concentration of 0.5 mg/ml. Collagen IV (BD Biosciences, Franklin Lakes, New Jersey, United States) was rendered insoluble by lyophilization and resuspended in assay buffer at a final concentration of 0.5 mg/ml. These solutions were vortexed, and 10 μg of tyrosinase was added. The solutions were allowed to react at room temperature for varying periods of time, after which they were centrifuged at 16,000 g for 15 min. The pellets (insoluble rMα, α-synuclein, collagen IV, and melanin products) were resuspended in 125 mM CH3COOH/CH3COOK buffer (pH 5.0) and centrifuged again. The pellets were then resuspended in 1 M NaOH and then heated to 60 °C for 5 min and vortexed to effect dissolution. Absorbance spectra were recorded at 350 nm.
Supporting Information
Figure S1 The Intrinsic Tryptophan Fluorescence (Excitation 295 nm) of rMα Was Measured in 8 M GdmCl and in Nondenaturing Buffer
rMα tryptophan emission in nondenaturing buffer is significantly blue-shifted with respect to rMα tryptophan emission in 8 M GdmCl, most likely owing to aggregation-induced burial and shielding of the tryptophan residues from the aqueous buffer. The red-shifted data indicate that rMα is unfolded in 8 M GdmCl, consistent with observations using gel filtration chromatography.
(102 KB TIF).
Click here for additional data file.
Accession Numbers
The Swiss-Prot (http://www.ebi.ac.uk/swissprot/) accession numbers for the gene products discussed in this paper are Pmel17 (P40967) and type I transmembrane enzyme tyrosinase (P14679).
The work was supported by the National Institutes of Health (EY11606 to WEB, AG18917 to JWK/WEB, DK46335 to JWK, and AR041855/EY014919 to MSM), the Skaggs Institute of Chemical Biology, and the Lita Annenberg Hazen Foundation. We would like to thank Malcolm Wood, Marilyn Leonard, and Jeanne Matteson for their excellent technical assistance, as well as Xiaoping Dai in Ian Wilson's laboratory for help acquiring the X-ray powder diffraction data.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. WEB and JWK conceived and designed the experiments. DMF, AVK, and CA performed the experiments. DMF, AVK, and CA analyzed the data. MSM, WEB, and JWK wrote the paper.
Citation: Fowler DM, Koulov AV, Alory-Jost C, Marks MS, Balch WE, et al. (2006) Functional amyloid formation within mammalian tissue. PLoS Biol 4(1): e6.
Abbreviations
CDcircular dichroism
FT-IRFourier transform infrared
DHQindole-5,6-quinone
DOPA3,4-dihydroxyphenylalanine
GdmClguanidinium chloride
rMαrecombinant Mα
RPEretinal pigment epithelium
==== Refs
References
Petkova AT Leapman RD Guo Z Yau WM Mattson MP Self-propagating, molecular-level polymorphism in Alzheimer's beta-amyloid fibrils Science 2005 307 262 265 15653506
Chen S Berthelier V Hamilton JB O'Nuallain B Wetzel R Amyloid-like features of polyglutamine aggregates and their assembly kinetics Biochemistry 2002 41 7391 7399 12044172
Sekijima Y Wiseman RL Matteson J Hammarstrom P Miller SR The biological and chemical basis for tissue-selective amyloid disease Cell 2005 121 73 85 15820680
Tanaka M Chien P Yonekura K Weissman JS Mechanism of cross-species prion transmission: An infectious conformation compatible with two highly divergent yeast prion proteins Cell 2005 121 49 62 15820678
Dobson CM Protein chemistry: In the footsteps of alchemists Science 2004 304 1259 1262 1261 15166354
Cohen FE Kelly JW Therapeutic approaches to protein-misfolding diseases Nature 2003 426 905 909 14685252
Caughey B Lansbury PT Protofibrils, pores, fibrils, and neurodegeneration: Separating the responsible protein aggregates from the innocent bystanders Annu Rev Neurosci 2003 26 267 298 12704221
Marks MS Seabra MC The melanosome: membrane dynamics in black and white Nat Rev Mol Cell Biol 2001 2 738 748 11584301
Hearing VJ The melanosome: The perfect model for cellular responses to the environment Pigment Cell Res 2000 13 Suppl 23 34 11041354
Berson JF Theos AC Harper DC Tenza D Raposo G Proprotein convertase cleavage liberates a fibrillogenic fragment of a resident glycoprotein to initiate melanosome biogenesis J Cell Biol 2003 161 521 533 12732614
Kelly JW Balch WE Amyloid as a natural product J Cell Biol 2003 161 461 462 12743097
Berson JF Harper DC Tenza D Raposo G Marks MS Pmel17 initiates premelanosome morphogenesis within multivesicular bodies Mol Biol Cell 2001 12 3451 3464 11694580
Chakraborty AK Platt JT Kim KK Kwon BS Bennett DC Polymerization of 5,6-dihydroxyindole-2-carboxylic acid to melanin by the Pmel17/Silver locus protein Eur J Biochem 1996 236 180 8617263
Lee ZH Hou L Moellmann G Kuklinska E Antol K Characterization and subcellular localization of human Pmel17/Silver, a 100-kDa (pre)melanosomal membrane protein associated with 5,6-dihydroxyindole-2carboxylic acid (DHCIA) converting activity J Invest Dermatol 1996 106 605 8617992
Land EJ Ramsden CA Riley PA Packer L Sies H Quinone chemistry and melanogenesis Quinones and quinone enzymes, part A 2004 New York Academic Press 88 109
Raposo G Marks MS The dark side of lysosome-related organelles: Specialization of the endocytic pathway for melanosome biogenesis Traffic 2002 3 237 248 11929605
Silvers WK The coat colors of mice: A model for mammalian gene action and interaction 1979 New York Springer-Verlag 379
Spanakis E Lamina P Bennett DC Effects of the developmental colour mutations silver and recessive spotting on proliferation of diploid and immortal mouse melanocytes in culture Development 1992 114 675 680 1618134
Quevedo WC Fleischmann RD Dyckman J Seiji M Premature loss of melanocytes from hair follicles of light (B lt) and silver (si) mice Phenotypic expression in pigment cells 1981 Tokyo Tokyo University Press 177 184
Schonthaler HB Lampert JM von Lintig J Schwarz H Geisler R A mutation in the silver gene leads to defects in melanosome biogenesis and alterations in the visual system in the zebra fish mutant fading vision Dev Biol 2005 284 421 436 16024012
Kerje S Sharma P Gunnarsson U Kim H Bagchi S The Dominant white, Dun and Smoky color variants in chicken are associated with insertion/deletion polymorphisms in the Pmel17 gene Genetics 2004 168 1507 1518 15579702
Scott G Zhao Q Rab3a and SNARE proteins: Potential regulators of melanosome movement J Invest Dermatol 2001 116 296 304 11180007
Linke RP Highly sensitive diagnosis of amyloid and various amyloid syndromes using Congo red fluorescence Virchows Arch 2000 436 439 448 10881737
Klunk WE Pettegrew JW Abraham DJ Quantitative evaluation of Congo red binding to amyloid-like proteins with a beta-pleated sheet conformation J Histochem and Cytochem 1989 37 1273 1281 2666510
Westermark GT Johnson KH Westermark P Staining methods for identification of amyloid in tissue Methods Enzymol 1999 309 3 25 10507013
Chiti F Stefani M Taddei N Ramponi G Dobson CM Rationalization of the effects of mutations on peptide and protein aggregation rates Nature 2003 424 805 808 12917692
Lashuel HA Hartley D Petre BM Walz T Lansbury PT Neurodegenerative disease: Amyloid pores from pathogenic mutations Nature 2002 418 291
Eanes ED Glenner GG Physical and chemical properties of amyloid fibers. I. X-ray diffraction studies on amyloid filaments J Histochem Cytochem 1968 16 673 677 5723775
Fink AL Khurana SR Oberg KA Singh BR Determination of secondary structure in protein aggregates using attenuated total reflectance FTIR Infrared analysis of peptides and proteins: Principles and application 2000 Washington (D. C.) American Chemical Society 132 144
Baxa U Taylor KL Wall JS Simon MN Cheng N Architecture of Ure2p prion filaments: The N-terminal domains form a central core fiber J Biol Chem 2003 278 43717 43727 12917441
Krebs MR Bromley EH Donald AM The binding of thioflavin-T to amyloid fibrils: Localisation and implications J Struct Biol 2005 149 30 37 15629655
Hoashi T Watabe H Muller J Yamaguchi Y Vieira WD MART-1 is required for the function of the melanosomal matrix protein Pmel17/GP100 and the maturation of melanosomes J Biol Chem 2005 280 14006 14016 15695812
Chapman MR Robinson LS Pinkner JS Roth R Heuser J Role of Escherichia coli curli operons in directing amyloid fiber formation Science 2002 295 851 855 11823641
Kenney JM Knight D Wise MJ Vollrath F Amyloidogenic nature of spider silk Eur J Biochem 2002 269 4159 4163 12180993
Shorter J Lindquist S Hsp104 catalyzes formation and elimination of self-replicating Sup35 prion conformers Science 2004 304 1793 1797 15155912
Si K Lindquist S Kandel ER A neuronal isoform of the aplysia CPEB has prion-like properties Cell 2003 115 879 891 14697205
Chen CD Huff ME Matteson J Page L Phillips R Furin initiates gelsolin familial amyloidosis in the Golgi through a defect in Ca(2+) stabilization EMBO J 2001 20 6277 6287 11707399
Pawelek JM Lerner AB 5,6-Dihydroxyindole is a melanin precursor showing potent cytotoxicity Nature 1978 276 626 628 723948
Klunk WE Pettegrew JW Abraham DJ Two simple methods for quantifying low-affinity dye-substrate binding J Histochem Cytochem 1989 37 1293 1297 2666512
Johnson WC Analyzing protein circular dichroism spectra for accurate secondary structures Proteins 1999 35 307 312 10328265
Provencher SW Glockner J Estimation of globular protein secondary structure from circular dichroism Biochemistry 1981 20 33 37 7470476
Sreerama N Woody RW A self-consistent method for the analysis of protein secondary structure from circular dichroism Anal Biochem 1993 209 32 44 8465960
Singh BR Singh BR Basic aspects of the technique and applications of infrared spectroscopy of peptides and proteins Infrared analysis of peptides and proteins: Principles and application 2000 Washington (D. C.) American Chemical Society 2 37
Lashuel HA Petre BM Wall J Simon M Nowak RJ Alpha-synuclein, especially the Parkinson's disease-associated mutants, forms pore-like annular and tubular protofibrils J Mol Biol 2002 322 1089 1102 12367530
|
16300414
|
PMC1288039
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2006 Jan 29; 4(1):e6
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040006
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpmedplosmedPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0040008SynopsisBiophysicsCell BiologyMolecular Biology/Structural BiologyPhysiologyBiochemistryIn VitroMammalsThe Unfolding of Amyloid's True Colors Synopsis1 2006 29 11 2005 29 11 2005 4 1 e8Copyright: © 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 Amyloid Formation within Mammalian Tissue
==== Body
What do neurodegenerative diseases and suntans have in common? Scientists at the Scripps Research Institute have found an intriguing molecular connection. In neurodegenerative disorders such as Alzheimer and Parkinson disease, proteins aggregate into specific fibrous structures (called a cross-β sheet) to form insoluble plaques known as amyloid. Because amyloid accumulation can be highly toxic to cells and organisms, often leading to neurodegeneration, therapeutic strategies for treating such protein-conformation disorders involve targeting and reducing amyloid formation and accumulation. But in a new study in PLoS Biology, Douglas Fowler, Atanas Koulov, and colleagues present evidence that the amyloid structure may play a normal role in mammalian cells. Amyloid fibrils are present in melanin-producing cells in great abundance, the authors show, where they help synthesize the sunburn-fighting pigment, melanin.
A native, nontoxic mammalian amyloid structure, generated by the protein Pmel17, helps accelerate synthesis of melanin
These pigments are synthesized in organelles called melanosomes, which reside in specialized skin cells (melanocytes) and the eyes (retinal pigment epithelium), to produce and traffic pigments for coloration, ultraviolet protection, and chemical detoxification. Melanosome biogenesis proceeds via a specialized pathway, related to the pathway producing a broad range of “housekeeping” organelles, including lysosomes, known for engulfing and cleaving, or lysing, proteins. Fowler, Koulov, and colleagues isolated melanosomes from retinal pigment epithelium taken from cows' eyes, and probed them for different protein compositions. Though the authors suspected that melanosomes might contain amyloids (based on previous reports that melanosome proteins resisted denaturation, a property of most amyloid fibers), they were surprised to find the organelle loaded with fibrillar amyloids. They visualized the amyloids primarily by using fluorescent molecules exhibiting selective binding to the characteristic amyloid cross-β sheet conformation, a fluorescent microscopy method long used by pathologists to diagnose protein-conformation disorders.
Which protein contributed to the alarming abundance of the amyloid structure in melanosomes? Several clues pointed to the glycoprotein Pmel17, a critical component of melanosome biogenesis, according to genetic and biochemical data. During melanosome biogenesis, Pmel17 lyses into two fragments, one called Mα that is sequestered into a membrane-bound compartment of the melanosome and another that is degraded. After confirming that the amyloids were comprised of Mα fibers, the authors tried to make Mα fragments fold into amyloid in a test tube. They showed that a purified, nonaggregated Mα (which they called recombinant rMα) folds into amyloids remarkably quickly. When Fowler et al. compared the rate at which rMα forms amyloids to that of other well-known amyloids—Aβ and α-synuclein, which are implicated in Alzheimer and Parkinson disease, respectively—they found that rMα amyloid production was at least four orders of magnitude faster. The authors offer the intriguing hypothesis that by rapidly folding into the amyloid cross-β sheet structure, Mα avoids generating the toxic intermediates that are very common in pathogenic amyloid formation.
Finally, Fowler et al. satisfy a burning question: are Mα amyloid fibers serving a function in melanin synthesis? After reconstituting components of the melanin biosynthethic pathway in vitro, they showed that adding rMα results in a 2-fold increase in melanin production (as does adding other amyloids like Aβ and α-synuclein). Perhaps more importantly, the Mα amyloid fibrils bind and orient the highly reactive organic melanin precursors, mitigating the cellular toxicity observed when Mα amyloid production is halted by mutation.
The authors also raise the intriguing idea that, given the propensity for many proteins to form amyloid fibrils, this conformation may be another physiologically important protein fold found in cells. To differentiate the biologically functional amyloid from pathogenic amyloids, the authors suggest using the term “amyloidin.” Although the common involvement of amyloids between melanin synthesis and protein conformation disorders is most surprising, future research into the differences between amyloid formation in these processes may hold the key for understanding diseases including Huntington, Parkinson, and Alzheimer disease. Because melanosome biogenesis is a tightly regulated process, a deeper understanding of the mechanisms that allow the Pmel17 Mα fragment to avoid the toxic stage of amyloid formation could provide considerable insight into which aspects are missing when proteins misfold. —Jami Milton Dantzker
|
0
|
PMC1288040
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2006 Jan 29; 4(1):e8
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040008
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1381625091510.1186/1471-2407-5-138Research ArticleChance mechanisms affecting the burden of metastases Kendal Wayne S [email protected] Division of Radiation Oncology, The Ottawa Hospital Regional Cancer Centre, 503 Smyth, Ottawa, Ontario, K1H 1C4 Canada and The Ottawa Hospital Research Institute, Ottawa, Ontario, K1H 8L6 Canada2005 26 10 2005 5 138 138 10 7 2005 26 10 2005 Copyright © 2005 Kendal; 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 burden of cancer metastases within an individual is commonly used to clinically characterize a tumor's biological behavior. Assessments like these implicitly assume that spurious effects can be discounted. Here the influence of chance on the burden of metastasis is studied to determine whether or not this assumption is valid.
Methods
Monte Carlo simulations were performed to estimate tumor burdens sustained by individuals with cancer, based upon empirically derived and validated models for the number and size distributions of metastases. Factors related to the intrinsic metastatic potential of tumors and their host microenvironments were kept constant, to more clearly demonstrate the contribution from chance.
Results
Under otherwise identical conditions, both the simulated numbers and the sizes of metastases were highly variable. Comparable individuals could sustain anywhere from no metastases to scores of metastases, and the sizes of the metastases ranged from microscopic to macroscopic. Despite the marked variability in the number and sizes of the metastases, their respective growth times were rather more narrowly distributed. In such situations multiple occult metastases could develop into fully overt lesions within a comparatively short time period.
Conclusion
Chance can have a major effect on the burden of metastases. Random variability can be so great as to make individual assessments of tumor biology unreliable, yet constrained enough to lead to the apparently simultaneous appearance of multiple overt metastases.
==== Body
Background
The burden of metastatic disease can be an important determinant in cancer management [1]. With many of the common epithelial cancers, successful resection of the primary tumor may be all that is required for cure, provided that there are no metastases. If the numbers of metastases are limited, it may be possible to resect them along with the primary tumor for curative intent [2-4]. Moreover, adjuvant therapy may cure individuals who might otherwise succumb, provided that the metastatic burden is small and their disease is responsive to therapy [5,6].
In those individuals who have had their metastatic disease resected, the number of metastases removed tends to have an adverse correlation with survival [3,7-10]. In addition, the total volume of resected metastases may provide an even stronger prognostic indicator [11]. Clinicians have often considered the number of metastases sustained by an individual to be a biological measure of intrinsic tumor behavior [12-16]. In keeping with this view, both clinicians [17] and experimentalists [18-22] have come to equate the metastatic potential of tumors with the numbers of organ metastases.
There is, however, a critical distinction between human clinical observations and animal experiments: Animal experiments can be repeated under more or less uniform conditions; human observations are usually isolated and more heterogeneous. Even so, controlled animal experiments tend to exhibit considerable heterogeneity in the numbers of hematogenous metastases [18,21,23]. In some cases the observed experimental heterogeneity can be attributable to pre-existent variant cells within tumor cell populations [18,24], but even within groups of syngeneic animals treated with the same tumor cells, the numbers of resultant metastases can be quite variable [25,26]. The metastatic heterogeneity, evident with these animal studies, is also evident with human studies [27]. This pattern can be attributed to physiologic heterogeneities in regional organ blood flow and, combined with counting (Poisson) statistics, the two processes can be used to explain the disparities in the numbers of haematogenous metastases evident to individuals under otherwise similar conditions [28,29].
Human autopsy studies have also revealed that the sizes of metastases within individuals can be quite variable [17,30]. In individuals with multiple metastases the frequency distribution for the sizes of their metastases approximates a lognormal form, which presumably relates to a normal distribution for the growth times of metastases, provided that the growth of the secondary deposits is approximately exponential. These normally distributed growth times can, in turn, be explained by the summation of the times required for the multiple sequential steps deemed necessary for tumor growth and metastasis [31].
On this basis, both the number of metastases and the sizes of metastases can manifest random influences, in addition to being affected by the tumor's metastatic potential and the host microenvironment. Since both the numbers of metastases and the total tumor burden have been represented as measures of tumor biology [11-16], it would be important to determine whether or not these additional spurious influences might obscure the biological assessments of malignant behavior.
The central point to be illustrated here is that biologically similar tumors may give rise to highly variable numbers or volumes of metastases. Because of this, the clinically apparent metastatic burden may not provide a good measure of the biological characteristics of a tumor. Animal experiments have shown that the metastatic burden does indeed reflect the biological properties of a tumor, but spurious influences are commonly evident within such experiments [18,21,23]. For this reason animal experiments must be repeated a number of times in order to ascertain the biological properties and to account for chance variations. Clinicians are not able to do this when assessing the biological potential of an individual's cancer. Clinical decisions may be made on the basis of the evident burden of disease, without knowledge as to the degree that chance might have influenced the clinical presentation. For this reason it would be important to demonstrate the potential influence that chance might have on the metastatic burden.
To this end models for the number distribution [29] and size distribution [31] of hematogenous metastases will be utilized here to simulate the metastatic burden within a hypothetical series of identical cancer patients. Since it was not possible to control all the biological and temporal variables in actual human clinical series, simulation was chosen as the most practical means to assess the heterogeneity in metastatic burden that could be attributable to chance.
Methods
Monte Carlo simulations were performed for the numbers of hematogenous metastases within individuals where the intrinsic biological conditions of the tumor and the host microenvironment were presumed to remain constant by maintaining the same mean and variance for the number of metastases in each simulation. Granted, this approach represents a simplification of the influence of tumor growth factors, and their receptors, the production of angiogenic factors, the capacity for motility and invasion, or the capacity for aggregation and deformability which all may affect a tumor's intrinsic capacity to metastasise and proliferate [32]. The interplay of pararcrine and endocrine growth factors, neovascularization, platelet aggregation and the action of immune cells and their products are additional factors that contribute to the host microenvironment [32] were similarly treated. The precise way that these factors might exert their influences goes beyond the aim of this manuscript, and a phenomenological approach was taken instead. To this end the Poisson negative binomial (PNB) distribution can accurately represent the variations in the numbers of hematogenous metastases sustained by individuals under otherwise identical conditions [29], and it was thus used to represent the frequency distribution for the numbers of metastases (see the Additional file 1 for details regarding the PNB distribution). In this model, variations in the numbers of metastases can be attributed to both random chance and to heterogeneity in regional organ blood flow [28,33].
The simulation parameters were chosen to reflect a mean number of about 8 metastases, and a variance of 57 metastases2. This was within the range of observation of controlled metastasis experiments of the murine B16 F10 melanoma in lung [29], and well within the range of the much more heterogeneous clinical data from humans draw from lung and liver metastases secondary to sarcomas, lung carcinomas, and various gastrointestinal carcinomas [27]. This choice of parameters also afforded a sizeable fraction of cases with no metastases and was made with the view that the mean number of resultant metastases was in itself not critical, since the aim of the simulation was to illustrate the variability in the numbers of metastases that might occur, under otherwise constant intrinsic biological conditions.
Admittedly the choice of parameters here was arbitrary, and designed to illustrate (not to prove) the potential variability in the numbers of metastases. In simulations such as these, the end results are predetermined by the choice of models and parameters. However, in order to provide realistic results, the models and parameters used were chosen to emulate laboratory and clinicopathological observation as closely as possible. Because the human data reflected populational heterogeneity, with respect to growth rates of primary tumors and their metastases and the rates of metastatic dissemination, as well as differences in times of the onset of the primary tumor relative to the assessment of the numbers of metastases of each individual's disease, these human data could serve only as an outside bounds for the potential ranges of these parameters. Controlled experiments from the murine B16 F10 melanoma were extrapolated upon to provide somewhat closer bounds for the parameter choices.
A second set of Monte Carlo simulations was performed to yield the size distribution of metastases expected within individuals. Each metastasis was assumed to grow in accordance with a stochastic pure birth process [34] (see the Additional file 1). The growth times for the metastases within any individual were further assumed to be normally distributed, in accordance with deductions from previous human autopsy studies [31]. According to this model, the size variations of the metastases could be essentially attributed to random chance.
The growth times for the metastases, as defined from the establishment of each nascent (and growing) metastasis to the surgical removal of the primary tumor, was arbitrarily assumed to have a mean and standard deviation (SD) of 16 ± 1 volume doublings. After resection of the primary tumor, the metastases were assumed to grow for an additional 12 doublings before they were counted and sized. This sequence was chosen to emulate the clinical sequence of resection of a primary tumor, performed in an individual with no clinically detectable metastases, followed by restaging after a set time interval. As well, and as a first approximation, the growth rates of the metastases were assumed to be the same as that of the primary tumor.
These two simulations were combined into a third simulation, designed to provide the number and sizes of metastases that could be anticipated within a group of similar individuals. Fifty simulations were performed where the biological properties intrinsic to the primary tumor and the host microenvironment were assumed constant.
A further calculation was performed, based upon the size distribution of 4000 simulated metastases, in order to determine the proportion of clinically detectable metastases at various time intervals from the surgical resection of the primary tumor. The growth parameters were the same as in the previous calculations, and the clinical detection threshold was arbitrarily set at 109 cells (~1 cm3).
Results
Figure 1 provides the results of the simulation for the numbers of hematogenous metastases. About 14% of the simulated cases sustained no metastases; the majority had 6 or fewer metastases, whereas a small minority exhibited as many as 40 metastases. The numbers of metastases sustained by individuals under identical conditions were thus quite variable, but they were consistent with the available clinical and pathological studies [27].
Simulations were also performed for the size distribution of metastases in this hypothetical series of patients. A total of 4,000 simulations yielded a mean metastasis size of 4 × 108 cells (range 2 × 104 to 8 × 109 cells; SD 8 × 108 cells). Figure 2 provides the frequency histogram for the logarithm of the number of cells per metastasis, as determined from the 4,000 simulations. The histogram approximated a normal distribution, and was consistent with the available data from human autopsy studies [17,31].
The final simulations were designed to emulate a hypothetical case series of 50 cancer patients. As before, the intrinsic biological properties of host and tumor were assumed identical, the primary tumors were removed at comparable times, and the resultant metastases were enumerated 12 volume-doubling times later.
Figure 3 provides the results from this simulated case series: Fig. 3a gives the numbers of simulated metastases. Six cases sustained no metastases; the majority of cases sustained less than 10 metastases; three cases sustained more than 20 metastases. Figure 3b provides the respective sizes of the metastases from each case. Most cases had metastases that consisted of fewer than 109 cells. Both the numbers and sizes of the metastases were highly variable, despite the identical biological conditions prerequisite to the simulations.
From Fig. 3b it can be seen that the majority of the simulated metastases would have been so small as to be effectively occult, given a detection threshold of 109 cells. However, only a small number of additional volume doublings would have been required for many of these occult metastases to become overt.
To better demonstrate the kinetics of the transition from occult to overt metastases, the proportion of 4,000 simulated metastases that had reached a detection threshold of 109 cells was plotted versus the time interval from removal of the primary tumor. Figure 4 illustrates this graph. After a latent period of about a dozen doubling times, the first metastases would have appeared. With about an additional 3 volume doubling times beyond this, a majority of the remaining metastases would then have become detectable, as had been surmised from Fig. 3b.
The appearance of multiple overt metastases within a relatively short time period could be directly related to the growth time distribution for multiple metastases. As previously noted, this distribution can be approximated by the normal distribution [31]. The narrower this distribution is, the more likely that multiple metastases would seem to appear within a short time period.
Discussion
How well do these simulations represent reality? The answer to this question depends, in part, upon the validity of the two models the simulations were based upon, one for the frequency distribution of the number of hematogenous metastases [29], the other for the size distribution of metastases [31]. The frequency distribution of metastases was derived from observations of thousands of humans and animals afflicted with cancer [25-27,29], and it was mechanistically explained in terms of the influence of physiological variations in regional organ blood on counting statistics [28,29,33]. The size distribution for hematogenous metastases was derived from detailed individual human autopsy studies of thousands of metastases, which agreed well with the lognormal distribution [17,30,31]. It was mechanistically based on the hypothesis of normally distributed growth times for exponentially growing metastases [31]. Both models seemed well grounded on observation, and they both had plausible mechanisms.
As a further requirement for the simulations to be realistic, the underlying models needed to yield results that were within the range of experience. Human data were not available for the numbers of hematogenous metastases sustained by syngeneic hosts, injected with cells from the same tumor and assessed at the same time in the natural history of the disease; but murine data were [18,21,26]. The simulations for the numbers of metastases were thus parameterised so as to fit within the observed range of the murine B16 F10 melanoma data. A mean number of 8 ± 8 (SD) metastases was chosen, and with a variance to mean ratio of σ2/μ~8, results which were similar to observations from clones of the B16 F10 murine melanoma in experimental metastasis assays with age matched syngenic mice [25]. These simulated results also fell within the bounds of the more heterogeneous clinical and pathological observations from humans [27].
With regards to the parameterisation for the size distribution of metastases, the simulated volumes (mean ± SD, 0.4 ± 0.8 cm3), and their coefficient of variation (σ/μ~2), fitted into the range observed from individual human autopsy studies [17,30]. The predictions from both of these Monte Carlo simulations were thus consistent with laboratory and human data.
What insights did these simulations provide into metastasis? The conventional model for hematogenous metastasis involves multiple sequential steps – the primary tumor must produce cells capable of metastasis, these cells must intravasate, exfoliate, successfully traverse the circulation system, arrest in a target organ favourable to growth, extravasate, form micrometastases, induce angiogenesis, and then proliferate [35,36]. At points along this pathway, tumor cells may also die or become dormant. The eventual burden of metastases presumably reflects the kinetic balances inherent to each step, as well as the relative time periods spent in dormancy and proliferation. The biological properties of the tumor cells, the host environment, and the biophysical aspects of the transport processes involved likely would affect metastasis, as would random events [37].
Much has been said and written about metastasis being a nonrandom process [38-43]. This dictum seems most appropriate when restricted to the organ predilections contingent to the seed and soil hypothesis [32,44], and to the anatomic pathways of regional metastasis [45,46]. However, the influence of random chance in metastasis has not as thoroughly been deliberated. To this end, the simulations preformed here showed that both the numbers and sizes of metastases within an individual could be affected in a major way by chance mechanisms.
At the same time the spurious underlying nature of metastasis has been well apparent to experimentalists. In the conventional experimental metastasis assay, and despite efforts to keep conditions constant, when syngeneic age-matched animals are injected with equal sized aliquots of the same tumor cell suspension, the resultant numbers of lung metastases can be quite variable [18-22]. Biologists have come to rely upon statistical tests to compare the metastatic potentials of different tumor variants; the median or mean number of metastases per group would then provide an index for the metastatic potential of the tumor, and the random variability could be quantitated by the variance of the number of metastases per each group [25,26]. This variability can be significant and it would be reasonable to postulate a similar degree of variability to be associated with individual human cases.
The real issue is that the burden of metastasis can be attributed partly to the biological properties of the tumor and partly to chance events. To understand how to use the observed metastatic burden in clinical practice, one must understand the relative importance of the deterministic and random components. Clinicians cannot rely upon multiple replicate observations, and their measured variance, to assess the random component as can be done in the animal laboratory. We are left to extrapolate from the laboratory experiments where the numbers of metastases in otherwise identical circumstances can vary considerably. The parameters of the simulations performed here were chosen to emulate human and animal observations as closely as possible. One obviously cannot prove that there exists a large random component to the metastatic burden by these means; however, one can provide a plausible example for the potential degree of variability that might be encountered in clinical situations.
Chance also seems to have a role in the relatively sudden appearance of multiple metastases, in persons who have previously been followed without apparent disease. Clinical experience indicates that it is not unusual for multiple metastases to seem to appear within a relative short time interval [47-52]. From the simulations presented here, such apparently synchronous emergence of multiple metastases within an individual could be attributable to a relatively compact distribution for their growth times. Admittedly there are also cases where multiple asynchronous metastases can manifest over a longer time course; in these cases presumably this distribution would not be so compact.
The aim of this study was to examine the variability in the metastatic burden, under circumstances where the metastatic potential of a tumor and the conditions within the host for tumor metastasis and growth could be considered constant. Earlier in this article, the significant variability in the numbers of experimental and spontaneous metastases amongst similarly treated animals was alluded to [18,21,26,53]. Because of this variability, laboratory scientists have come to rely upon multiple measurements from identically treated animals in order to assess the metastatic potential of tumors. Indeed, the variability can be such that animals from a treatment group might sustain multiple metastases while others from the same group remained metastasis-free. Similar disparities, if observed in human clinical series, might be construed as to represent the differing biological behavior of individual tumors [12-16]. However in the light of these experimental data, and the simulations provided here, these variations could reasonably be attributed to spurious events, rather than to inherent biological differences. For this reason it would appear to be difficult to draw inferences regarding the biological behavior of a tumor, based upon individual assessments of tumor burden.
Conclusion
The burden of metastases sustained by an individual can presumably be influenced by the intrinsic biology of the tumor, the host microenvironment, the time available for tumor growth, and by the interaction of chance events with these processes. The origins of these chance events are not entirely clear; this could include both physical and biological processes. Nevertheless, such events could lead to a wide variation in number and sizes of metastases, under otherwise controlled conditions. The growth times of individual metastases are also affected by chance. But here the variability is more constrained, and consequently multiple occult metastases within an individual may relatively suddenly become clinically overt.
These particular kinetic properties of metastases, although influenced by tumor and host biology, can thus also be affected by chance, to a degree sufficient to potentially obscure the underlying biology. Clinicians will commonly make an assessment of the nature of a particular individual's cancer based upon the apparent burden of metastases. This simulation demonstrates that stochastic elements could potentially have a major influence on the burden of metastasis, to the point that it would seem unreliable, in any individual, to base assessments as to the biological potential of the cancer on the metastatic burden.
List of abbreviations
PNB, Poisson negative binomial; PGF, probability generating function; CDF, cumulative distribution function; SD, standard deviation.
Competing interests
The author(s) declare that they have no competing interests.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional File 1
The distribution of the numbers of metastases.
Click here for file
Figures and Tables
Figure 1 Frequency distribution for the numbers of metastases. Frequency distributions for the number of hematogenous metastases can be predicted based upon random counting statistics, modulated by regional organ blood flow heterogeneity. In this model the biological properties of the tumor cells and the host environment were otherwise assumed constant. The simulation was executed with a mean of 8 ± 7.6 (SD) metastases, in keeping with observations compiled from murine and human tumors.
Figure 2 Size distribution for metastases. A model for the growth of metastases was based upon a stochastic birth process where the growth times of the individual colonies were normally distributed. Four thousand simulations were performed. The resultant frequency histogram for the logarithm of the number of cells per metastasis is shown here, and it approximated a normal distribution (solid line). This was equivalent to a lognormal distribution for the sizes of metastases.
Figure 3 Fifty simulations for the numbers and sizes of metastases. In these simulations the primary tumors were presumed successfully resected from each host. All biological properties of the tumors and hosts were assumed constant. The only sources for deviation were random chance and heterogeneities in regional organ blood flow. In each simulation there were no clinically detectable metastases at the time of surgery, and the individuals were restaged after a time to allow for 12 volume doublings. a. Simulated numbers of metastases. Using the same parameterization as employed in Fig. 1, six cases sustained no metastases whereas the remaining cases sustained anywhere from one to just under forty metastases. b. Simulated sizes of metastases. The resultant sizes of metastases from each case are plotted here semi-logarithmically. The right axis provides the numbers of cells per metastasis; the left gives the corresponding diameters of the metastases. Many of these metastases would likely be below the threshold of conventional clinical detection (arbitrarily defined here as at 109 cells). Given additional time for growth the subclinical metastases would have also become overt.
Figure 4 Detection of metastases. The percentage of clinically detectable metastases relative to the total number of metastases within a patient is plotted here versus the time interval from the removal of the primary tumor to detection. This plot was based upon the 4000 simulations cells. A latent period provided in Fig. 2, and the detection threshold was set at 109 was apparent, where the metastases remain clinically occult, followed by a period where the proportion of overt metastases rapidly increased.
==== Refs
Withers HR Suwinski R Radiation dose response for subclinical metastases Semin Radiat Oncol 1998 8 224 228 9634499
Cady B McDermott WV Major hepatic resection for metachronous metastases from colon cancer Ann Surg 1985 201 204 209 3970601
Vogt-Moykopf I Bulzebruck H Kryza S Kruschinski H Branscheid D Schirren J Results in surgery of pulmonary metastases Chirurgie 1992 118 263 271 1341280
Patchell RA Tibbs PA Walsh JW Dempsey RJ Maruyama Y Kryscio RJ Markesbery WR Macdonald JS Young B A randomized trial of surgery in the treatment of single metastases to the brain N Engl J Med 1990 322 494 500 2405271
Early breast cancer trialists' collaborative group Systemic treatment of early breast cancer by hormonal, systemic or immune therapy: 133 randomized trials involving 31,000 recurrences and 24, 000 deaths among 75,000 women Lancet 1992 339 1 15 1345950
Early breast cancer trialists' collaborative group Polychemotherapy for early breast cancer: an overview of the randomized trials Lancet 1998 352 930 942 9752815 10.1016/S0140-6736(98)03301-7
Taylor M Forster J Langer B Taylor BR Greig PD Mahut C A study of prognostic factors for hepatic resection for colorectal metastases The American Journal of Surgery 1997 173 467 471 9207156 10.1016/S0002-9610(97)00020-2
Lagerwaard FJ Levendag PC Nowak PJ Eijkenboom WM Hanssens PE Schmitz PI Identification of prognostic factors in patients with brain metastases: a review of 1292 patients Int J Radiat Oncol Biol Phys 1999 43 795 803 10098435 10.1016/S0360-3016(98)00442-8
Weber SM Jarnagin WR DeMatteo RP Blumgart LH Fong Y Survival after resection of multiple hepatic colorectal metastases Ann Surg Oncol 2000 7 643 650 11034240 10.1007/s10434-000-0643-3
Mutsaerts ELAR van Ruth S Zoetmulder FAN Rutgers EJT Hart AAM van Coevorden F Prognostic factors and evaluation of surgical management of hepatic metastases from colorectal origin: a 10-year single-institute experience Journal of Gastrointestinal Surgery 2005 9 178 186 15694813 10.1016/j.gassur.2004.06.005
Ercolani G Grazi GL Ravaioli M Cescon M Gardini A Varotti G Del Gaudio M Nardo B Cavallari A Liver Resection for Multiple Colorectal Metastases: Influence of Parenchymal Involvement and Total Tumor Volume, vs Number or Location, on Long-term Survival Arch Surg 2002 137 1187 1192 12361432 10.1001/archsurg.137.10.1187
Ishii T Ikegami N Hosoda Y Koide O Kaneko M The biological behaviour of gastric cancer J Pathol 1981 134 97 115 6265613 10.1002/path.1711340202
Bakalakos EA Kim JA Young DC Martin EW Jr Determinants of Survival following Hepatic Resection for Metastatic Colorectal Cancer World Journal of Surgery 1998 22 399 405 9523523 10.1007/s002689900404
Cady B Jenkins RL Steele GD JrLewis WD Stone MD McDermott WV Jessup JM Bothe A Lalor P Lovett EJ Lavin P Linehan DC Surgical margin in hepatic resection for colorectal metastasis: a critical and improvable determinant of outcome Ann Surg 1998 227 566 571 9563547 10.1097/00000658-199804000-00019
Allen PJ Coit DG The role of surgery for patients with metastatic melanoma Curr Opin Oncol 2002 14 221 226 11880715 10.1097/00001622-200203000-00014
Allen PJ Coit DG The Surgical Management of Metastatic Melanoma Ann Surg Oncol 2002 9 762 770 12374659 10.1245/ASO.2002.02.009
Yamanami H Chiba R Kobari M Matsuno S Takahashi T Total number and size distribution of hepatic metastases of carcinoma Anal Quant Cytol Histol 1999 21 216 226 10560494
Fidler IJ Kripke ML Metastasis results from preexisting variant cells within a malignant tumor Science 1977 197 893 895 887927
Kripke ML Gruys E Fidler IJ Metastatic heterogeneity of cells from an ultraviolet light-induced murine fibrosarcoma of recent origin Cancer Res 1978 38 2962 2967 679204
Fidler IJ Cifone MA Properties of metastatic and nonmetastatic cloned subpopulations of an ultraviolet-light-induced murine fibrosarcoma of recent origin Am J Pathol 1979 97 633 648 507195
Fidler IJ Gruys E Cifone MA Barnes Z Bucana C Demonstration of multiple phenotypic diversity in a murine melanoma of recent origin J Natl Cancer Inst 1981 67 947 956 6944560
Welch DR Neri A Nicolson GL Comparison of 'spontaneous' and 'experimental' metastasis using rat 13762 mammary adenocarcinoma metastatic cell clones Invasion Metastasis 1983 3 65 80 6677622
Poste G Tzeng J Doll J Greig R Rieman D Zeidman I Evolution of tumor cell heterogeneity during progressive growth of individual lung metastases Proc Natl Acad Sci U S A 1982 79 6574 6578 6959137
Heppner GH Tumor heterogeneity Cancer Res 1984 44 2259 2265 6372991
Kendal WS Frost P Experimental metastasis: a novel application of the variance-to-mean power function J Natl Cancer Inst 1987 79 1113 1115 3479636
Kendal WS Clustering of murine lung metastases reflects fractal nonuniformity in regional lung blood flow Invasion Metastasis 1999 18 285 296
Kendal WS Lagerwaard FJ Agboola O Characterization of the frequency distribution for human hematogenous metastases: evidence for clustering and a power variance function Clin Exp Metastasis 2000 18 219 229 11315095 10.1023/A:1006737100797
Kendal WS A stochastic model for the self-similar heterogeneity of regional organ blood flow Proc Natl Acad Sci U S A 2001 98 837 841 11158557 10.1073/pnas.021347898
Kendal W A Frequency Distribution for the Number of Hematogenous Organ Metastases J Theor Biol 2002 217 203 218 12202114 10.1006/jtbi.2002.3021
Douglas JRS Significance of the size distribution of bloodborn metastases Cancer 1971 27 379 390 5541952
Kendal WS The size distribution of human hematogenous metastases J Theor Biol 2001 211 29 38 11407889 10.1006/jtbi.2001.2326
Fidler IJ The pathogenesis of cancer metastasis: the 'seed and soil' hypothesis revisited Nat Rev Cancer 2003 3 453 458 12778135 10.1038/nrc1098
Bassingthwaighte JB King RB Roger SA Fractal nature of regional myocardial blood flow heterogeneity Circ Res 1989 65 578 590 2766485
Feller W An introduction to probability theory and its applications 1968 1 New York: John Wiley & Sons
Fidler IJ 7th Jan Waldenstrom Lecture. The biology of human cancer metastasis Acta Oncol 1991 30 668 675 1659837
Chambers AF Groom AC MacDonald IC Dissemination and growth of cancer cells in metastatic sites Nat Rev Cancer 2002 2 563 572 12154349 10.1038/nrc865
Weiss L Random and nonrandom processes in metastasis, and metastatic inefficiency Invasion Metastasis 1983 3 193 207 6588046
Fidler IJ Biological behavior of malignant melanoma cells correlated to their survival in vivo Cancer Res 1975 35 218 224 1109790
Hart IR Fidler IJ Role of organ selectivity in the determination of metastatic patterns of B16 melanoma Cancer Res 1980 40 2281 2287 7388794
de la Monte SM Moore GW Hutchins GM Nonrandom distribution of metastases in neuroblastic tumors Cancer 1983 52 915 925 6871832
Nicolson GL Organ specificity of tumor metastasis: role of preferential adhesion, invasion and growth of malignant cells at specific secondary sites Cancer Metastasis Rev 1988 7 143 188 3293836 10.1007/BF00046483
Fidler IJ Critical factors in the biology of human cancer metastasis: twenty-eighth G.H.A. Clowes memorial award lecture Cancer Res 1990 50 6130 6138 1698118
Radinsky R Fidler IJ Regulation of tumor cell growth at organ-specific metastases In Vivo 1992 6 325 331 1520835
Paget S The distribution of secondary growths in cancer of the breast. 1899 Cancer Metastasis Rev 1989 8 98 101 2673568
Weiss L Bronk J Pickren JW Lane WW Metastatic patterns and target organ arterial blood flow Invasion Metastasis 1981 1 126 135 7188382
Nathanson SD Insights into the mechanisms of lymph node metastasis Cancer 2003 98 413 423 12872364 10.1002/cncr.11464
Sasano S Yamamoto H Otsuka T Imura Y Nishii N Nitta S Pulmonary multiple metastases of ovarian granulosa cell tumor 15 years after initial diagnosis Jpn J Thorac Cardiovasc Surg 2000 48 655 658 11080955
King DT Gurevitch AW Hirose FM Multiple cutaneous metastases of a scapular chondrosarcoma Arch Dermatol 1978 114 584 586 646373 10.1001/archderm.114.4.584
Mochizuki R Gunji Y Takayama W Miyazaki S Makino H Matsushita K Miyauchi H Chiba S Ochiai T A case of the 5-year survivor of ascending colon cancer associated with synchronous multiple liver metastasis and peritoneal dissemination successfully treated with combination therapy of systemic and hepatic arterial infusion chemotherapy Gan To Kagaku Ryoho 2004 31 1662 1664 15553676
Dalerba P Ricci A Russo V Rigatti D Nicotra MR Mottolese M Bordignon C Natali PG Traversari C High homogeneity of MAGE, BAGE, GAGE, tyrosinase and Melan-A/MART-1 gene expression in clusters of multiple simultaneous metastases of human melanoma: implications for protocol design of therapeutic antigen-specific vaccination strategies Int J Cancer 1998 77 200 204 9650552 10.1002/(SICI)1097-0215(19980717)77:2<200::AID-IJC5>3.0.CO;2-U
Dhar-Munshi S Ameen M Wilson RS Simultaneous metastases of cutaneous malignant melanoma to conjunctiva and choroid Br J Ophthalmol 2000 84 930 931 10979653 10.1136/bjo.84.8.928b
Neto AG Deavers MT Silva EG Malpica A Metastatic tumors of the vulva: a clinicopathologic study of 66 cases Am J Surg Pathol 2003 27 799 804 12766584 10.1097/00000478-200306000-00011
Harris JF Chambers AF Hill RP Ling V Metastatic Variants are Generated Spontaneously at a High Rate in Mouse KHT Tumor PNAS 1982 79 5547 5551 6957880
Jørgensen B The Theory of Exponential Dispersion Models 1997 Chapman & Hall, London
Particle data group Mathematical tools or statistics, Monte Carlo, group theory Phys Lett B 2004 592 275 297
|
16250915
|
PMC1289278
|
CC BY
|
2021-01-04 16:03:08
|
no
|
BMC Cancer. 2005 Oct 26; 5:138
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-138
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1391625314610.1186/1471-2407-5-139Research ArticleThe expression of HSP60 and HSP10 in large bowel carcinomas with lymph node metastase Cappello Francesco [email protected] Sabrina [email protected] Francesca [email protected] Fabio [email protected]à Lorenzo [email protected] Tommaso E [email protected] Felicia [email protected] Giovanni [email protected] Sezione di Anatomia Umana, Dipartimento di Medicina Sperimentale, Università degli Studi di Palermo, Italy2 Reparto di Anatomia Patologica, Ospedale "Civico", Palermo, Italy2005 28 10 2005 5 139 139 31 3 2005 28 10 2005 Copyright © 2005 Cappello et al; licensee BioMed Central Ltd.2005Cappello 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 involvement of Heat Shock Proteins (HSP) in cancer development and progression is a widely debated topic. The objective of the present study was to evaluate the presence and expression of HSP60 and HSP10 in a series of large bowel carcinomas and locoregional lymph nodes with and without metastases.
Methods
82 Astler and Coller's stage C2 colorectal cancers, of which 48 well-differentiated and 34 poorly-differentiated, were selected along with 661 lymph nodes, including 372 with metastases and 289 with reactive hyperplasia only, from the same tumours. Primitive tumours and both metastatic and reactive lymph nodes were studied; specifically, three different compartments of the lymph nodes, secondary follicle, paracortex and medullary sinus, were also analysed. An immunohistochemical research for HSP60 and HSP10 was performed and the semiquantitative results were analysed by statistical analysis to determine the correlation between HSPs expression and 1) tumour grading; 2) degree of inflammation; 3) number of lymph nodes involved; 4) lymph node compartment hyperplasia. Moreover, western blotting was performed on a smaller group of samples to confirm the immunohistochemical results.
Results
Our data show that the expression of HSP60, in both primary tumour and lymph node metastasis, is correlated with the tumoral grade, while the HSP10 expression is not. Nevertheless, the levels of HSP10 are commonly higher than the levels of HSP60. In addition, statistical analyses do not show any correlation between the degree of inflammation and the immunopositivity for both HSP60 and HSP10. Moreover, we find a significant correlation between the presence of lymph node metastases and the positivity for both HSP60 and HSP10. In particular, metastatic lymph nodes show a higher percentage of cells positive for both HSP60 and HSP10 in the secondary follicles, and for HSP10 in the medullary sinuses, when compared with hyperplastic lymph nodes.
Conclusion
HSP60 and HSP10 may have diagnostic and prognostic significance in the management of this tumour and their overexpression in tumoral cells may be functionally related to tumoral progression. We hypothesise that their expression in follicular and medullary cells of lymph nodes may be induced by formation of metastases. Further studies based on these observations could lead to a better understanding of the HSPs involvement in colorectal cancer progression, as well as other neoplasms.
==== Body
Background
Heat shock proteins (HSP) are a family of molecules that are highly conserved during evolution and involved in many cellular functions, such as protein folding. Consequently, their alteration may have multiple pathophysiologic effects and the number of papers studying their expression in normal and pathologic conditions is constantly increasing [1-3]. In particular, the role of a number of HSPs, such as HSP27, -70, -72 and -90, during carcinogenesis has already been widely investigated, in vivo and in vitro, in many conditions, such as lung [4], breast [5], esophageal [6] and ovarian [7] cancer, as well as osteosarcoma [8], and lymphoblastic leukemia [9]. The data obtained in these studies seems to suggest that this group of HSPs may be useful as tools in the management of primitive neoplasms. Some articles have also suggested a possible relationship between HSP expression and lymph node metastasis formation [10-15].
HSP60 and HSP10 are two chaperones that interact in a two-step folding mechanism in the mitochondria of prokaryotic and eukaryotic cells [16]. In addition, these proteins may be involved in other cellular functions, such as mediating specific tumour signals, but these roles are not yet well understood [3]. In the last few years, our research group has evaluated the presence and expression of HSP60 and HSP10 in a series of carcinogenetic models, such as the "dysplasia-carcinoma" sequences of uterine exocervix [17,18], large bowel [18,19] and prostate [20]. These data have highlighted that these chaperones are overexpressed during the carcinogenetic steps; in particular, they accumulate in the cytoplasm of dysplastic and neoplastic cells, and their levels of expression increases in the sequence leading from dysplasia towards carcinoma. We have hypothesised that HSP60 and HSP10 might be considered as new diagnostic and prognostic tools for these cancers [21,22], being involved in the molecular steps of carcinogenesis, analogously to what has already been demonstrated with other tumours [23-28].
HSP10 was recently shown to be selectively expressed by myelocyte and megakaryocyte precursors in normal human bone marrow [29]. This feature disappears during lineage maturation, and it was hypothesised that HSP10 might have another role during differentiation and/or proliferation of those normal cellular lineages apart from the co-chaperonin one, although the obtained results could not explain this selective expression.
In view of these factors, in the present work the presence and expression of HSP60 and HSP10 were studied in a series of advanced large bowel carcinomas (LBC) with lymph node metastases. In particular, we investigated whether their expression was dependent on the grade of differentiation of the tumour and presence of regional lymph node metastasis. Moreover, we analysed the significance of the data to determine the correlation between HSP expression and both degree of inflammation and number of lymph nodes involved. Finally, three different compartments of each reactive lymph node, the secondary follicles (SF), the paracortex (PC) and the medullary sinuses (MS), were examined; lymph nodes with metastases (MLN) were compared with lymph nodes with reactive hyperplasia (HLN) only, to determine any differences in HSP60 and HSP10 expression.
Methods
Immunohistochemistry
82 LBC of Astler and Coller's stage C2, with locoregional node metastases, were collected. 48 well-differentiated (G1) carcinomas were compared with 34 poorly-differentiated (G3) ones. The specimens were formalin-fixed and paraffin-embedded. From each case, a 5 micra section of both tumour infiltrating the bowel wall (Ti) and lymph node with metastasis (Ni) were obtained. 10 specimens of normal colonic mucosae were selected, from which 5 micra sections were obtained as controls.
From all the tumours collected, 661 lymph nodes were selected and divided in two groups; the first comprising of 372 lymph nodes with metastases (MLN) and the second of 289 lymph nodes with reactive hyperplasia (HLN) only. A 5 micra section was obtained from each sample.
An immunostaining for HSP60 (monoclonal, SIGMA, H4149, 1:400), and HSP10 (Polyclonal, StressGen, SPA-110, 1:400) was performed on the Ti, Ni, MLN and HLN sections, using an avidin-biotin complex kit (LSAB2, DAKO, Cat. No. K677). A non-immune serum was run concurrently as negative control. 3-3'-diaminobenzidine (DAB chromogen solution, DAKO, Cat. No. K3467) was used as develop chromogen. Nuclear counterstaining was obtained using hematoxylin (DAKO, Cat. No. S2020).
Three independent observers (F.C., F.R. and L.M.) examined the specimens and performed a semiquantitative analysis to evaluate the percentage of positive cells in 10 HPF. The mean of the triplicate observation data was used for statistical analyses. Moreover, we semiquantified the degree of inflammation (infiltrating lymphocytes) in each tumoral specimen on a scale of 0–3 + (from absent to strongly positive).
We analysed the significance of the data using the Student "t" test (P < 0.05). A one-way "analysis of variance" (ANOVA) was used to determine the correlation between HSP expression and 1) tumour grading; 2) degree of inflammation; 3) number of lymph nodes involved; 4) lymph node compartment hyperplasia.
Western blotting
Biopsies from specimens of Ti were frozen in liquid nitrogen for molecular biology. Specifically, 10 G1 and 10 G3 specimens of both tumours and metastatic lymph nodes were collected randomly, along with 4 biopsies of normal colonic mucosae as controls.
20 μg of total tissue extracts in each lane and a protein marker (Kaleidoscope prestained standard, Bio-Rad, Cat. No 1610324) were separated by electrophoresis on denaturing 15% polyacrylamide slab gel (SDS-PAGE) and transferred to a nitrocellulose membrane (Nitrocell Paper, Bio-Rad, Cat. No 1620115). After 1 hour at room temperature (RT) with a blocking buffer (5% low-fat dried milk in TBST: 50 mM Tris-HCl pH 7,5, 150 mM NaCl, 0,1 % Tween-20) under gentle shaking, the membrane was incubated with anti-HSP60 primary antibody (monoclonal mouse, SIGMA, H4149, 1:1000) overnight at 4°C. After washings, the membrane was incubated with HRP-conjugated secondary antibody (anti-mouse, Pierce, 1:10000, Cat. No 31432) for 1 hour at RT with shaking and the specific binding was detected using a chemiluminescent substrate (SuperSignal West Pico Chemiluminescent Substrate, Pierce, Cat. No 34080) for autoradiography.
The same membrane was stripped with a stripping buffer (Restore TM Western Blot Stripping Buffer, Pierce, Cat. No 21059) and incubated with anti-HSP10 primary antibody (polyclonal rabbit, StressGen, SPA-110, 1:2000), following the procedures described above (secondary antibody: anti-rabbit, Pierce, 1:20000 Cat. No 31462).
Results
HSP60 and HSP10 positivity in primitive versus metastatic tumours of different grade
HSP60 was present in 40 out of 48 (83.3%) Ti G1 (fig. 1a,c) and in 32 out of 34 (94.1%) Ti G3 LBC (fig. 1b,d), while HSP10 was present in all examined specimens of Ti of both G1 (fig. 1e,g) and G3 (fig. 1f,h) LBC. In these specimens, both molecules were present in the cytoplasm of neoplastic cells, and they were also rarely present in some inflammatory elements scattered in the stroma. In particular, statistical analyses, we did not find any correlation between the degree of inflammation and the immunopositivity for HSP60 (p > 0.05) and HSP10 (p > 0.1).
Figure 1 HSP60 positivity in carcinoma (arrow) but not in normal tissues (arrowhead) in specimens from both well differentiated (a) and poorly differentiated (b) tumours (Magnification: 10×). A higher magnification (40×) shows that the positivity is diffuse into cytoplasm of tumoral cells of both G1 (c) and G3 (d) specimens; few positive interstitial (inflammatory) cells were scattered in the interposed stroma. HSP10 is also diffusely expressed by tumoral cells of both G1 (e) and G3 (f) LBC (Magnification: 10×). A higher magnification (40×) of both G1 (g) and G3 (h) shows that the positivity is mainly localised in the cytoplasm of neoplastic elements.
In addition, the percentage of HSP60 positive cells was higher in the G3 group (mean: 70%) compared with the G1 (mean: 35%) (fig. 2a), while a similar number of HSP10 positive elements were present in both groups (mean: respectively 74% and 75%) (fig. 2b). Normal epithelium above the infiltrating neoplasms resulted commonly negative or with few scattered positive elements (fig. 1b), similarly to what observed in the biopsies of normal colonic mucosae (data not shown). Statistical analyses showed that the difference between the number of HSP60 positive cells in G1 and G3 LBC was significant (p < 0.0005), while statistic difference was not found in HSP10 positivity (p > 0.05).
Figure 2 Graphics showing comparison between data obtained by quantitative analyses of HSP60 (a) and HSP10 (b) positive tumoral cells in G1 and G3 Ti and Ni.
Fig. 2c shows that 28 out of 48 Ni of G1 LBC (58%) were positive for HSP60 (fig. 3a), compared to 26 out of 34 Ni of G3 CRC (76%) (fig. 3b). Fig. 2d shows that all Ni of both G1 (fig. 3c) and G3 (fig. 3d) LBC were positive to HSP10. Statistical analyses showed that the difference between the number of HSP60 positive Ni in G1 and G3 LBC was significant (p < 0.01), while statistic difference was not found for HSP10 positivity (p > 0.05). Both HSP60 and HSP10 positivity was often co-localised within vascular (fig. 3e) and nervous (fig. 3f) structures invaded by neoplastic tissue in both Ti and Ni.
Figure 3 Infiltrated lymph nodes from G1 (a) and G3 (b) LBC show tumoral glands positive for HSP60. Metastases also show glands positive for HSP10 in both G1 (c) and G3 (d) carcinomas (Magnification: 40×). HSP60 positivity shows vascular (e) and neural (f) invasion by cancer (Magnification: 10×).
The results of the immunoblotting analyses were comparable to the immunohistochemical data (fig. 4). The quantity of HSP60 was higher in G3 specimens of both Ti and Ni, when compared to G1. HSP10 was present in a similar amount in all examined specimens. Specimens of normal colonic tissue were commonly under the threshold of detectability for both HSP60 and HSP10.
Figure 4 Western blot analyses for the research of HSP60 and HSP10 in tissue extracts of normal colonic mucosa (1), Ti G1 (2), Ti G3 (3), Ni G1 (4) and Ni G3 (5).
HSP60 and HSP10 positivity in metastatic versus hyperplastic lymo nodes
We selected 661 lymph nodes from all the tumours (mean: 8.1; range: 5–12; S.D. 2.2) and divided in two groups; the first comprising of 372 MLN and the second 289 HLN. Firstly, we examined the presence of a statistic correlation between HSPs expression and number lymph node involved by the disease. We found the presence of a significant correlation between the presence of metastases and the positivity for both HSP60 (p < 0.005) and HSP10 (p < 0.001) in lymph nodes.
Subsequently, a semiquantitative analysis on the immunohistochemical observations in the lymph node compartments of HLN was performed. Table 1 summarizes these results. In SF cells of the HLN group, only 5% were positive for HSP60 and 13% for HSP10. On the contrary, we found an increase in the number of HSP60 (28%) and HSP10 (35%) positive cells in SF of MLN (fig. 5a). We also found a great increase in the number of HSP10 positive cells in MS cells (38%) of MLN when compared to the HLN group (3%). There was no significant difference in the number of HSP60 positive cells between both HLN and MLN groups in MS (fig. 5b), similarly to the number of cells positive to both chaperones in PC of both HLN and MLN groups (fig. 5c). Statistical analyses showed a significant difference between the number of HSP60 positive cells in SF of the HLN and MLN groups (p < 0.0003).
Table 1 Mean percentages and ranges of immunopositive cells
HLN PART MEAN RANGE MLN PART MEAN RANGE
SF 5% 0–8% SF 28% 7–41%
HSP60 PC 2% 0–4% HSP60 PC 2% 0–4%
MS 1% 0–2% MS 3% 0–6%
SF 13% 4–22% SF 35% 18–58%
HSP10 PC 5% 2–8% HSP10 PC 9% 4–19%
MS 3% 1–5% MS 38% 22–65%
Mean percentages and ranges of immunopositive cells in hyperplastic lymph nodes (HLN), left side, and metastatic lymph node (MLN), right side. SF: secondary follicles; PC: paracortex; MS: medullary sinuses. See text for more details.
Figure 5 Diagrams showing the differences of the mean number of positive cells between HLN and MLN in the lymph node compartments: a) secondary follicles; b) paracortex; c) medullary sinuses.
Analogously, the number of HSP10 positive cells in both SF (p < 0.001) and MS (p < 0.0001) presented a significant difference. The positivity for both HSP60 (fig. 6a) and HSP10 (fig. 6b) in the cells of all reactive lymph node compartments was commonly localised in the cytoplasm.
Figure 6 High magnifications (100×) of immunostaining for HSP60 (a) and HSP10 (b) show cytoplasmic positivity.
Discussion
Although HSPs were first defined as proteins induced by environmental and pathophysiologic stress, they are also implicated in protein-protein interactions, such as folding, translocation, and prevention of inappropriate protein aggregation. Recently, other functions concerning their pivotal roles during cancer development and progression have been suggested (30).
In a study conducted on a series of esophageal squamous cell carcinomas, the overexpression of HSP70 was correlated with lymph node metastasis, and lymphatic vessel invasion, and the authors suggested that HSP70 expression might be used to assess the clinical outcome after surgery [12]. A reduced expression of HSP70 and HSP40 has also been associated with a lower histopathologic differentiation in a series of gastric carcinomas [31]. In addition, Hwang et al. [13] have demonstrated that the expression of HSP70 and HSP110 was increased in highly metastatic colorectal cancer cell lines, but not in weak metastatic cells, suggesting that the expression of these HSPs is highly correlated with the advanced clinical stages and positive lymph node involvement. In multivariate analyses concerning the type, grade, stage of the tumour, invasion of lymphatics, blood vessels and nerves as well as lymph nodes, in 36 pancreatic adenocarcinomas, HSP70 immunoexpression was found to be an independent prognostic factor [32]. Finally, in an immunohistochemical study on 102 esophageal squamous cell carcinoma specimens Kawanishi et al. [33] suggested that the expression of HSP27 and HSP70 was frequently reduced and therefore it should be considered an independent prognostic factor in this disease.
In their study on primary invasive ductal carcinomas of the breast with lymph node metastasis, Storm et al. [11] found that HSP27 might confer cytoprotection for metastatic cells, and they postulated that HSP27 overexpression is associated with reduced disease-free survival in breast carcinomas. Contrastingly, Tetu et al. [34] did not find any predictive role for HSP27 in the outcome in node-positive breast carcinomas.
Piselli et al. [35] recently performed a cytofluorimetric analysis on human pancreatic adenocarcinoma cells, both grown in vitro and collected ex vivo from primary tumours or lung metastases of tumour-engrafted SCID mice; they were the first to demonstrate an HSP60 surface expression on metastatic cells, but this expression was not correlated with metastasization. In a multivariate analysis on a series of metastatic breast cancers, Schneider et al. [36] demonstrated that neither HSP27 nor HSP60 expression was able to exclude axillary node invasion completely. Finally, Ito et al. [37] studied the expression of HSP27, -60, -70 and -90 in 24 squamous cell carcinomas of the tongue using immunohistochemistry, finding that HSP immunoexpression might change during tumorigenesis, but there was no correlation between HSP staining and survival period, clinical stage, lymph node metastasis, histological grade or p53 immunostaining.
As far as we are aware, the present work is the first study reporting the expression of HSP60 and HSP10 in a series of LBC with lymph node metastases, and the immunolocalisation of these molecules in the different compartments of reactive lymph nodes.
The presence and expression of HSP60 and HSP10 in some carcinogenic models, in particular, both pre-tumoral (dysplastic) and neoplastic lesions of large bowel, as well as uterine exocervix and prostate gland has been investigated previously [17-20]. These experiments showed that the level of these two strictly related mitochondrial chaperonins increases from normal through dysplastic towards neoplastic tissues. These proteins resulted diffusely localised in the cytoplasms of dysplastic and neoplastic cells. Moreover, few scattered inflammatory elements were occasionally positive at stromal level. Considering this overexpression, it was hypothesised that these molecules might have different functions during cancer development, apart from the mitochondrial regeneration during normal cell proliferation. Nevertheless, the exact nature of this role is still not understood.
In the present paper, the expression of HSP60 in Ti and Ni was found to be dependent on the tumoral grade, while the expression of HSP10 was not. The number of HSP60 positive tumoral cells in Ti of G3 LBC was higher than in G1, and the number of HSP60 positive Ni in G3 LBC was higher than in G1. Immunohistochemical results were confirmed by immunoblotting analysis. Therefore, we postulate a prognostic significance of these data. HSP10 was strongly positive in all examined specimens, and these results may have a diagnostic utility. Interestingly, a higher positivity for HSP60, but not for HSP10, was correlated with the presence of lymph node metastasis and this data may have a histopathologic value.
Although normal epithelia periodically regenerate cellular elements by mitosis of basal cells, the present work shows that HSP60 and HSP10, in normal epithelia, are under the antibody detection threshold for immunohistochemical analyses, while neoplastic elements show a strong cytoplasmic expression of these proteins. Many papers have already shown the involvement of the HSP60/HSP10 complex in preventing the activation of the apoptotic machinery. Samali et al [38] were the first to demonstrate that pro-caspase-3 is present in the mitochondrial fraction of Jurkat T cells in a complex with the chaperon proteins HSP60 and HSP10 and that the release of mitochondrial HSP may also accelerate caspase activation in the cytoplasm of intact cells. These data are in accordance with the findings of Xanthoudakis et al. [39] who showed that ATP-dependent 'foldase' activity of HSP60 may induce pro-caspase-3 maturation in Jurkat cells stimulated to undergo apoptosis by a Fas-independent pathway and that this represents an important regulatory event in apoptotic cell death. Lin et al. [40] suggested that overexpression of the combination of HSP60 and HSP10 and of HSP60 or HSP10 individually may protect myocytes against apoptosis in an in vitro model of ischemia/reperfusion injury. More recently they have demonstrated that HSP10 overexpression may inactivate Raf, ERK, and p90Ribosomal kinase (p90RSK), suggesting that only HSP10 is involved in the complex mechanisms that protect myocytes against simulated ischemia and reoxygenation induced death.
We could postulate that both HSP60 and HSP10 are up-regulated in cancer for extramitochondrial functions, i.e. in the block of the apoptotic machinery that usually takes place during cancer development and progression. Although HSP60 and HSP10 should be functionally correlated, HSP10 is present in a higher number of specimens and with a higher expression than HSP60. This result may indicate a different function of HSP10 inside the cytoplasm of tumoral cells. In a recent study, where the expression of HSP10 and HSP60 was investigated in a series of normal human bone marrows, similar data were found [29]. Interestingly, a selective preference of HSP10 for myeloid and megakaryocytic precursors was discovered [29]. Therefore, other roles of HSP10 apart from the co-chaperonin one during bone marrow cell proliferation and differentiation of normal cells could be hypothesised. This is backed up by other studies on this co-chaperonin [41,42].
In the present study, the presence and localisation of HSP60 and HSP10 in a series of human lymph nodes were evaluated. Lymph node functioning is crucial for an individual's survival. Normal lymph nodes are very small structures, not always clinically detectable in human body. They react to antigens by uptaking and processing them, eventually destroying them. Reacting lymph nodes enlarge due to immunologic stimulation that drives their hyperplasia. Examining a histological section of hyperplastic lymph node, different compartments can be distinguished: 1) the follicles, where precursors of plasma cells and memory B cells are formed; 2) paracortex, where specific cellular response takes place, generating antigen-specific T cells; 3) medullar sinuses, where the lymph carried from afferent to efferent lymphatic vessels is cleared by macrophages [43-45]. Commonly, hyperplasia involves all lymph node structures; as already demonstrated, in a study on a wide series of HLN and MLN from axilla, the most common pattern of hyperplasia is the mixed type (66.5% of 996 HLN and 68.6% of 4711 MLN). In particular, hyperplasia mainly involved both follicles and paracortex area (22.2% of MLN and 19.2% of HLN), although often sinuses were also found enlarged [46]. A fundamental step during the examination of an enlarged lymph node is the distinction between a reactive hyperplasia and a neoplasm. The latter may be of primary (lymphoma) or secondary (metastasis) type. In an investigation on a series of 1159 lymph nodes from breast cancer, micrometastases (involving less that 25% of the lymph nodal volume) have been shown often to be present also in very small nodes (less that 2 mm in diameter) [47]. Unfortunately, the detection of metastases in a lymph node is difficult if they are small in dimension. Several studies have shown that a number of histologically negative lymph nodes may present, at a retrospective analysis, a micrometastasis revealed only immunohistochemically or with immunofluorescence techniques; the presence of micrometastasis in the lymph node has great prognostic and therapeutic implications [48-53].
In this study, the number of SF cells positive for both HSP60 and HSP10 increased significantly in MLN, when compared to HLN. We postulate that this increase may be related to formation of metastases. As a consequence, although cytokeratin staining is a much easier way to detect micrometastasis, the statistically significant increment in the MLN group of the number of HSP60 and HSP10 positive elements in secondary follicles could also be considered diagnostic to predict the presence of lymph node metastases. Therefore, a lymph node with germinal centre presenting a strong staining for HSP60 and/or HSP10 without any apparent metastasis should be examined in detail, since this observation may reflect an increased likelihood of finding a micrometastasis.
Since MS of MLN showed a significantly higher amount of HSP10, but not HSP60, positive cells, when compared to the MS of HLN, we could assume that HSP10 in this site is under unknown stimulation inducing its overexpression, for functional roles, i.e. cell proliferation. The role of HSP60 during proliferation and differentiation of eukaryotic cells has already been demonstrated [54-56]. Moreover, HSP60 and HSP10, working together, could protect mitochondrial function and prevent apoptotic cell death [37,39] although some studies have shown that these molecules do not always act as a single functional unit in vivo [41,57].
Conclusion
Many papers have been focusing on the role of some HSPs to predict cancer progression [11-13,58]. We found particularly interesting the paper of Storm [11], who first showed the expression of HSP27 in metastatic lymph nodes to confer cytoprotection for metastatic cells of breast carcinoma and its association with the reduced disease-free survival. In addition, a selective expression of HSP70 in the germinal centres of HLN has been demonstrated, although the meaning of this overexpression is not understood [59].
Our researches showed an overexpression of HSP60 and HSP10 in LBC with lymph node metastases. Both molecules could be useful for histopathologic diagnosis of this neoplasm, as well as to better assess the prognosis. We could also assume that both proteins are involved in LBC progression, i.e. exercising an antiapoptotic effect. We hypothesise that the increased expression of HSP60 and HSP10 in reactive lymph node cells is due to their role in proliferation of normal cells. Recently it has been shown that HSP60 activates macrophages and T-cell, and this could also be a possible explanation for its overexpression in lymph nodes with metastatic cancer upregulation [60]. Based on these considerations, it could be postulated that paracrine factors, such as cytokines, may induce an up-regulation of these molecules, but the exact role of the overexpression remains to be confirmed.
Serial sectioning could be used to further study if any of the HLN population that present HSP60 and HSP10 staining in the range of MLN contain occult metastases; it could also be interesting to compare HLN from LBC specimens with non-neoplastic setting ones, i.e. resections of active inflammatory bowel disease, to confirm whether the former could be used as proper controls. Finally, the selective overexpression of HSP10 in MS of MLN could support the hypothesis that this molecule, following unknown biological stimulations, acts independently from HSP60.
In conclusion, our study could add new data to the classic classification of G1 or G3, since both HSP60 and HSP10 positivity may help to detect more aggressive tumors. Moreover, the comparison of the expression levels of these chaperones with other predictors of survival, as grading, number of metastatic lymph nodes and degree of tumor infiltrating lymphocytes may be useful in colon cancer management.
Abbreviations
HSP: Heat Shock Proteins
LBC: Large Bowel Carcinomas
SF: Secondary Follicles
PC: Paracortex
MS: Medullary Sinuses
MLN: Lymph Nodes with Metastases
HLN: Reactive Hyperplasia
G1: Well-Differentiated Tumours
G3: Poorly-Differentiated Tumours
Ti: Specimens of Tumour Infiltrating the Bowel Wall
Ni: Specimens of Lymph Node with Metastasis
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
FC designed the study, examined the immunohistochemical results, performed a semiquantitative analysis and drafted the manuscript;
SD collected of specimens, carried out the immunohistochemistry and carried out the Western blotting analyses;
FR collected of specimens, examined the immunohistochemical results and performed a semiquantitative analysis;
FB participated in the design of the study and in the draft of the manuscript, performed the statistical analysis and examined the Western blotting results ;
LM participated in the collection of specimens, examined the immunohistochemcal results and performed a semiquantitative analysis;
FF participated in the and examination of the Western blotting results and in the draft of the manuscript;
GZ coordinated the design and execution of the study.
All authors red 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 MIUR ex-60% funds of Professor G. Zummo, Prof. F. Farina and Dr. F. Cappello.
==== Refs
Wataba K Saito T Fukunaka K Ashihara K Nishimura M Kudo R Over-expression of heat shock proteins in carcinogenic endometrium Int J Cancer 2001 91 448 456 11251965 10.1002/1097-0215(200002)9999:9999<::AID-IJC1077>3.0.CO;2-F
Rashmi R Kumar S Karunagaran D Ectopic expression of HSP70 confers resistance and silencing its expression sensitizes human colon cancer cells to curcumin-induced apoptosis Carcinogenesis 2004 25 179 187 14604899 10.1093/carcin/bgh001
Mosser DD Morimoto RI Molecular chaperones and the stress of oncogenesis Oncogene 2004 23 2907 2918 15077153 10.1038/sj.onc.1207529
Zhong L Peng X Hidalgo GE Doherty DE Stromberg AJ Hirschowitz EA Antibodies to HSP70 and HSP90 in serum in non-small cell lung cancer patients Cancer Detect Prev 2003 27 285 290 12893076 10.1016/S0361-090X(03)00097-7
O'Neill PA Shaaban AM West CR Dodson A Jarvis C Moore P Davies MP Sibson DR Foster CS Increased risk of malignant progression in benign proliferating breast lesions defined by expression of heat shock protein 27 Br J Cancer 2004 90 182 188 14710227 10.1038/sj.bjc.6601449
Lambot MA Peny MO Fayt I Haot J Noel JC Overexpression of 27-kDa heat shock protein relates to poor histological differentiation in human oesophageal squamous cell carcinoma Histopathology 2000 36 326 330 10759946 10.1046/j.1365-2559.2000.00858.x
Elpek GO Karaveli S Simsek T Keles N Aksoy NH Expression of heat-shock proteins HSP27, HSP70 and HSP90 in malignant epithelial tumour of the ovaries APMIS 2003 111 523 530 12780528 10.1034/j.1600-0463.2003.1110411.x
Uozaki H Ishida T Kakiuchi C Horiuchi H Gotoh T Iijima T Imamura T Machinami R Expression of heat shock proteins in osteosarcoma and its relationship to prognosis Pathol Res Pract 2000 196 665 673 11087053
Lauten M Beger C Gerdes K Asgedom G Kardinal C Welte K Schrappe M Expression of heat-shock protein 90 in glucocorticoid-sensitive and -resistant childhood acute lymphoblastic leukaemia Leukemia 2003 17 1551 1556 12886242 10.1038/sj.leu.2403027
Ciocca DR Clark GM Tandon AK Fuqua SA Welch WJ McGuire WL Heat shock protein HSP70 in patients with axillary lymph node-negative breast cancer: prognostic implications J Natl Cancer Inst 1993 85 570 574 8455204
Storm FK Mahvi DM Gilchrist KW Heat shock protein 27 overexpression in breast cancer lymph node metastasis Ann Surg Oncol 1996 3 570 573 8915490
Noguchi T Takeno S Shibata T Uchida Y Yokoyama S Muller W Expression of heat shock protein 70 in grossly resected esophageal squamous cell carcinoma Ann Thorac Surg 2002 74 222 226 12118763 10.1016/S0003-4975(02)03641-X
Hwang TS Han HS Choi HK Lee YJ Kim YJ Han MY Park YM Differential, stage-dependent expression of HSP70, HSP110 and Bcl-2 in colorectal cancer J Gastroenterol Hepatol 2003 18 690 700 12753152 10.1046/j.1440-1746.2003.03011.x
Noguchi T Wada S Takeno S Moriyama H Kimura Y Uchida Y Lymph node metastasis could be predicted by evaluation of macrophage infiltration and HSP70 expression in superficial carcinoma of the esophagus Oncol Rep 2003 10 1161 1164 12883674
Zuo DS Dai J Bo AH Fan J Xiao XY Significance of expression of heat shock protein90alpha in human gastric cancer World J Gastroenterol 2003 9 2616 2618 14606110
Richardson A Schwager F Landry SJ Georgopoulos C The importance of a mobile loop in regulating chaperonin/ co-chaperonin interaction: humans versus Escherichia coli J Biol Chem 2001 276 4981 4987 11050098 10.1074/jbc.M008628200
Cappello F Bellafiore M Palma A Marciano V Martorana G Belfiore P Martorana A Farina F Zummo G Bucchieri F Expression of 60-kD heat shock protein increases during carcinogenesis in the uterine exocervix Pathobiology 2002 70 83 88 12476033 10.1159/000067304
Cappello F Bellafiore M David S Anzalone R Zummo G Ten kilodalton heat shock protein (HSP10) is overexpressed during carcinogenesis of large bowel and uterine exocervix Cancer Lett 2003 196 35 41 12860287 10.1016/S0304-3835(03)00212-X
Cappello F Bellafiore M Palma A David S Marciano V Bartolotta T Sciume C Modica G Farina F Zummo G Bucchieri F 60KDa chaperonin (HSP60) is over-expressed during colorectal carcinogenesis Eur J Histochem 2003 47 105 110 12777205
Cappello F Rappa F David S Anzalone R Zummo G Immunohistochemical evaluation of PCNA, p53, HSP60, HSP10 and MUC-2 presence and expression in prostate carcinogenesis Anticancer Res 2003 23 1325 1331 12820390
Cappello F HSP60 and HSP10 as diagnostic and prognostic tools in the management of exocervical carcinoma Gynecol Oncol 2003 91 661 14675699 10.1016/j.ygyno.2003.08.009
Pomara G Cappello F RE: Heat shock proteins: their role in urological tumors J Urol 2003 170 927 928 12913740 10.1097/01.ju.0000080564.23671.32
Hsu PL Hsu SM Abundance of heat shock proteins (HSP89, HSP60, and HSP27) in malignant cells of Hodgkin's disease Cancer Res 1998 58 5507 5513 9850087
Schneider J Jimenez E Marenbach K Romero H Marx D Meden H Immunohistochemical detection of HSP60-expression in human ovarian cancer. Correlation with survival in a series of 247 patients Anticancer Res 1999 19 2141 2146 10470162
Laad AD Thomas ML Fakih AR Chiplunkar SV Human gamma delta T cells recognize heat shock protein-60 on oral tumor cells Int J Cancer 1999 80 709 714 10048972 10.1002/(SICI)1097-0215(19990301)80:5<709::AID-IJC14>3.0.CO;2-R
Trieb K Gerth R Windhager R Grohs JG Holzer G Berger P Kotz R Serum antibodies against the heat shock protein 60 are elevated in patients with osteosarcoma Immunobiology 2000 201 368 376 10776793
Lebret T Watson RW Molinie V O'Neill A Gabriel C Fitzpatrick JM Botto H Heat shock proteins HSP27, HSP60, HSP70, and HSP90: expression in bladder carcinoma Cancer 2003 98 970 977 12942564 10.1002/cncr.11594
Pignatelli D Ferreira J Soares P Costa MJ Magalhaes MC Immunohistochemical study of heat shock proteins 27, 60 and 70 in the normal human adrenal and in adrenal tumors with suppressed ACTH production Microsc Res Tech 2003 61 315 323 12768547 10.1002/jemt.10341
Cappello F Tripodo C Farina F Franco V Zummo G HSP10 selective preference for myeloid and megakaryocytic precursors in normal human bone marrow Eur J Histochem 2004 49 261 265 15590416
Papp E Nardai G Soti C Csermely P Molecular chaperones, stress proteins and redox homeostasis Biofactors 2003 17 249 257 12897446
Isomoto H Oka M Yano Y Kanazawa Y Soda H Terada R Yasutake T Nakayama T Shikuwa S Takeshima F Udono H Murata I Ohtsuka K Kohno S Expression of heat shock protein (HSP) 70 and HSP 40 in gastric cancer Cancer Lett 2003 198 219 228 12957361 10.1016/S0304-3835(03)00305-7
Sagol O Tuna B Coker A Karademir S Obuz F Astarcioglu H Kupelioglu A Astarcioglu I Topalak O Immunohistochemical detection of pS2 protein and heat shock protein-70 in pancreatic adenocarcinomas. Relationship with disease extent and patient survival Pathol Res Pract 2002 198 77 84 11928868 10.1078/0344-0338-00190
Kawanishi K Shiozaki H Doki Y Sakita I Inoue M Yano M Tsujinaka T Shamma A Monden M Prognostic significance of heat shock proteins 27 and 70 in patients with squamous cell carcinoma of the esophagus Cancer 1999 85 1649 1657 10223556 10.1002/(SICI)1097-0142(19990415)85:8<1649::AID-CNCR2>3.0.CO;2-V
Tetu B Brisson J Landry J Huot J Prognostic significance of heat-shock protein-27 in node-positive breast carcinoma: an immunohistochemical study Breast Cancer Res Treat 1995 36 93 97 7579511 10.1007/BF00690189
Piselli P Vendetti S Vismara D Cicconi R Poccia F Colizzi V Delpino A Different expression of CD44, ICAM-1, and HSP60 on primary tumor and metastases of a human pancreatic carcinoma growing in scid mice Anticancer Res 2000 20 825 831 10810361
Schneider J Pollan M Ruibal A Jimenez E Lucas AR Nunez MI Sanchez J Tejerina A Histologic grade and CD44 are independent predictors of axillary lymph node invasion in early (T1) breast cancer Tumour Biol 1999 20 319 30 10567878 10.1159/000030096
Ito T Kawabe R Kurasono Y Hara M Kitamura H Fujita K Kanisawa M Expression of heat shock proteins in squamous cell carcinoma of the tongue: an immunohistochemical study J Oral Pathol Med 1998 27 18 22 9466730
Samali A Cai J Zhivotovsky B Jones DP Orrenius S Presence of a pre-apoptotic complex of pro-caspase-3, HSP60 and HSP10 in the mitochondrial fraction of jurkat cells EMBO J 1999 18 2040 2048 10205158 10.1093/emboj/18.8.2040
Xanthoudakis S Roy S Rasper D Hennessey T Aubin Y Cassady R Tawa P Ruel R Rosen A Nicholson DW HSP60 accelerates the maturation of pro-caspase-3 by upstream activator proteases during apoptosis EMBO J 1999 18 2049 2056 10205159 10.1093/emboj/18.8.2049
Lin KM Lin B Lian IY Mestril R Scheffler IE Dillmann WH Combined and individual mitochondrial HSP60 and HSP10 expression in cardiac myocytes protects mitochondrial function and prevents apoptotic cell deaths induced by simulated ischemia-reoxygenation Circulation 2001 103 1787 1792 11282911
Lin KM Hollander JM Kao VY Lin B Macpherson L Dillmann WH Myocyte protection by 10 kD heat shock protein (HSP10) involves the mobile loop and attenuation of the Ras GTP-ase pathway FASEB J 2004 18 1004 1006 15059967
Yang EC Guo J Diehl G DeSouza L Rodrigues MJ Romaschin AD Colgan TJ Siu KW Protein expression profiling of endometrial malignancies reveals a new tumor marker: chaperonin 10 J Proteome Res 2004 3 636 643 15253447 10.1021/pr049975z
Hsu SM Cossman J Jaffe ES Lymphocyte subsets in normal human lymphoid tissues Am J Clin Pathol 1983 80 21 30 6190394
Dardick I Dardick AM Morphometry of normal human lymphoid tissues. Nuclear parameters for comparative studies of lymphoma Arch Pathol Lab Med 1984 108 190 196 6546504
van den Oord JJ de Wolf-Peeters C Desmet VJ The composite nodule. A structural and functional unit of the reactive human lymph node Am J Pathol 1986 122 83 91 2934988
Cappello F Bellafiore M Palma A Marciano V Zummo G Farina F Bucchieri F Study of axillary lymph node asymmetry in a female population J Anat 2001 199 617 620 11760893 10.1017/S002187820100841X
Cappello F Barresi E Martorana A Breast cancer and lymph node metastasis: relation of lymph node dimensions and metastasis Pathologica 2000 92 5 8 10842993
Gonzalez Bosquet J Keeney GL Mariani A Webb MJ Cliby WA Cytokeratin staining of resected lymph nodes may improve the sensitivity of surgical staging for endometrial cancer Gynecol Oncol 2003 91 518 525 14675670 10.1016/j.ygyno.2003.08.026
Jiao X Eslami A Ioffe O Kwong KF Henry M Zeng Q Refaely Y Burrows W Gamliel Z Krasna MJ Immunohistochemistry analysis of micrometastasis in pretreatment lymph nodes from patients with esophageal cancer Ann Thorac Surg 2003 76 996 999 14529974 10.1016/S0003-4975(03)00658-1
Lara JF Young SM Velilla RE Santoro EJ Templeton SF The relevance of occult axillary micrometastasis in ductal carcinoma in situ: a clinicopathologic study with long-term follow-up Cancer 2003 98 2105 2113 14601079 10.1002/cncr.11761
Munakata S Aihara T Morino H Takatsuka Y Application of immunofluorescence for intraoperative evaluation of sentinel lymph nodes in patients with breast carcinoma Cancer 2003 98 1562 1568 14534870 10.1002/cncr.11689
Juretzka MM Jensen KC Longacre TA Teng NN Husain A Detection of pelvic lymph node micrometastasis in stage IA2-IB2 cervical cancer by immunohistochemical analysis Gynecol Oncol 2004 93 107 111 15047221 10.1016/j.ygyno.2003.11.033
Lentz SE Muderspach LI Felix JC Ye W Groshen S Amezcua CA Identification of micrometastases in histologically negative lymph nodes of early-stage cervical cancer patients Obstet Gynecol 2004 103 1204 1210 15172853
Meinhardt A Wilhelm B Seitz J Expression of mitochondrial marker proteins during spermatogenesis Hum Reprod Update 1999 5 108 119 10336016 10.1093/humupd/5.2.108
Bethke K Staib F Distler M Schmitt U Jonuleit H Enk AH Galle PR Heike M Different efficiency of heat shock proteins (HSP) to activate human monocytes and dendritic cells: superiority of HSP60 J Immunol 2002 169 6141 6148 12444117
Flohe SB Bruggemann J Lendemans S Nikulina M Meierhoff G Flohe S Kolb H Human heat shock protein 60 induces maturation of dendritic cells versus a Th1-promoting phenotype J Immunol 2003 170 2340 2348 12594256
Dubaquie Y Looser R Funfschilling U Jeno P Rospert S Identification of in vivo substrates of the yeast mitochondrial chaperonins reveals overlapping but non-identical requirement for HSP60 and HSP10 EMBO J 1998 17 5868 5876 9774331 10.1093/emboj/17.20.5868
Cuezva JM Chen G Alonso AM Isidoro A Misek DE Hanash SM Beer DG The bioenergetic signature of lung adenocarcinomas is a molecular marker of cancer diagnosis and prognosis Carcinogenesis 2004 25 1157 1163 14963017 10.1093/carcin/bgh113
Leopardi O Naughten W Giannulis I Mirra M Frigo B HSP70 is selectively over-expressed in the blast cells of the germinal centres and paracortex in reactive lymph nodes Histopathology 2001 39 566 571 11903573 10.1046/j.1365-2559.2001.01293.x
Breloer M More SH Osterloh A Stelter F Jack RS Bonin A Macrophages as main inducers of IFN-gamma in T cells following administration of human and mouse heat shock protein 60 Int Immunol 2002 14 1247 1253 12407015 10.1093/intimm/dxf090
|
16253146
|
PMC1289279
|
CC BY
|
2021-01-04 16:39:12
|
no
|
BMC Cancer. 2005 Oct 28; 5:139
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-139
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1451627113910.1186/1471-2407-5-145Research ArticleTransformation of human bronchial epithelial cells alters responsiveness to inflammatory cytokines Loewen Gregory M [email protected] Erin [email protected] Frédéric [email protected] Dongfeng [email protected] Jihnhee [email protected] Sameera [email protected] Sei-Ichi [email protected] Heinz [email protected] Departments of Medicine, Roswell Park Cancer Institute, Buffalo, NY 14263, USA2 Departments of Molecular and Cell Biology, Roswell Park Cancer Institute, Buffalo, NY 14263, USA3 Departments of Pathology, Roswell Park Cancer Institute, Buffalo, NY 14263, USA4 Departments of Biostatistics, Roswell Park Cancer Institute, Buffalo, NY 14263, USA5 Departments of Cancer Genetics, Roswell Park Cancer Institute, Buffalo, NY 14263, USA2005 4 11 2005 5 145 145 22 7 2005 4 11 2005 Copyright © 2005 Loewen et al; licensee BioMed Central Ltd.2005Loewen 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
Inflammation is commonly associated with lung tumors. Since inflammatory mediators, including members of the interleukin-6 (IL-6) cytokine family, suppress proliferation of normal epithelial cells, we hypothesized that epithelial cells must develop mechanisms to evade this inhibition during the tumorigenesis. This study compared the cytokine responses of normal epithelial cells to that of premalignant cells.
Methods
Short-term primary cultures of epithelial cells were established from bronchial brushings. Paired sets of brushings were obtained from areas of normal bronchial epithelium and from areas of metaplastic or dysplastic epithelium, or areas of frank endobronchial carcinoma. In 43 paired cultures, the signalling through the signal transducer and activator of transcription (STAT) and extracellular regulated kinase (ERK) pathways and growth regulation by IL-6, leukemia inhibitory factor (LIF), oncostatin M (OSM), interferon-γ (IFNγ) or epidermal growth factor (EGF) were determined. Inducible expression and function of the leukemia inhibitory factor receptor was assessed by treatment with the histone deacetylase inhibitor depsipeptide.
Results
Normal epithelial cells respond strongly to OSM, IFNγ and EGF, and respond moderately to IL-6, and do not exhibit a detectable response to LIF. In preneoplastic cells, the aberrant signaling that was detected most frequently was an elevated activation of ERK, a reduced or increased IL-6 and EGF response, and an increased LIF response. Some of these changes in preneoplastic cell signaling approach those observed in established lung cancer cell lines. Epigenetic control of LIF receptor expression by histone acetylation can account for the gain of LIF responsiveness. OSM and macrophage-derived cytokines suppressed proliferation of normal epithelial cells, but reduced inhibition or even stimulated proliferation was noted for preneoplastic cells. These alterations likely contribute to the supporting effects that inflammation has on lung tumor progression.
Conclusion
This study indicates that during the earliest stage of premalignant transformation, a modified response to cytokines and EGF is evident. Some of the altered cytokine responses in primary premalignant cells are comparable to those seen in established lung cancer cell lines.
==== Body
Background
Lung epithelial cells are consistently exposed to many irritants and pathogens. Excessive exposure can lead to inflammatory conditions even though effective mechanisms are in place to contain and remove harmful components [1]. Epithelial damage results in tissue repair. Chronic injury and repeated cycles of tissue repair in presence of an inflammatory reaction may provide conditions that are conducive for selection of cells with enhanced proliferation and/or reduced sensitivity to signals for growth arrest and differentiation [2]. An environment that favors tumorigenesis is created when genetic and epigenetic changes enhance proliferation, reduce differentiation and/or attenuate apoptotic reactions [3]. Changes in epithelial morphology and proliferation may result in decreased autofluorescence that is grossly detectable with autofluorescence bronchoscopy [4]. A step-wise progression has been hypothesized to precede frank malignancy [5,6], and recent autofluorescence bronchoscopy studies have confirmed the malignant potential of metaplasia and dysplasia of the bronchial epithelium [7-9]. Direct visualization of these changes has made it possible to better understand the role of inflammation in lung carcinogenesis.
Inflammation has been reported to contribute to the development of cancer [1,2,10,11], and IL-6 cytokines, such as oncostatin M (OSM), actually arrest growth of cultured epithelial [12] and other cell types [13]. We hypothesized that members of the interleukin-6 (IL-6) cytokine family may contribute to the step-wise progression by providing growth-stimulatory activity. We also hypothesized that the transformed premalignant cells escape the inhibitory activity of cytokines as a function of the transformation process and that these transformed cells have altered cytokine responsiveness. These alterations should include reduced signaling through growth-suppressing pathways and/or enhanced signaling through growth-promoting pathways.
IL-6 cytokines are recognized by receptors that belong to the group of hematopoietin receptors [14]. Signal transduction is communicated by receptor-associated Janus protein tyrosine kinases that phosphorylate the receptor subunits. The signaling proteins are recruited to the tyrosine phosphorylated receptors, include signal transducer and activator of transcription-3 (STAT3), the protein tyrosine phosphatase SHP-2 and the adaptor Shc, which link to the RAS-MAPK-ERK pathway [14,15]. The magnitude of these immediate signaling reactions is a measure for the level of receptor activation in treated cells, and this is particularly true for the tyrosine phosphorylation of STAT3 and dual phosphorylation of ERK1/2 [16].
At the present time, very little is known about: (a) the response pattern of normal, non-immortalized human lung epithelial cells to inflammatory mediators, (b) the individual variation of the response patterns, and (c) the effects that premalignant transformation has on the responsiveness. Our study was designed to determine the response of bronchial epithelial cells from normal epithelium and abnormal lesions to inflammatory mediators and IL-6-type cytokines, and to define the effects of these cytokines on signaling and cell growth regulation. We developed techniques for obtaining proliferating epithelial cell cultures from bronchoscopy brushings and we used those cultures for functional screening.
Methods
Patient population
Patients with tobacco exposure who were enrolled in an institutional lung cancer screening trial underwent autofluorescence bronchoscopy using the LIFE system (LIFE systems, Xillix, Vancouver BC) in concert with low-dose spiral CT of the chest [17]. Informed consent was obtained and procedures were performed in accordance with an IRB-approved protocol.
Bronchoscopy and cytologic brushing
Bronchoscopy was performed under monitored anesthesia care. The airways were examined with white light and for autofluorescence. The appearance of the bronchial mucosa was classified as normal or abnormal. Bronchial epithelial brushings (area of ~25 mm2) and biopsies were taken from one normal and one to three abnormal sites. Cytology samples from the brushings were directly smeared onto glass slides and fixed in 95% ethanol. Residual tissue on the cytology brush was then used to establish cell cultures. Diagnoses of metaplasia, grades of dysplasia or non-invasive carcinoma in situ were made strictly based on the criteria established by the World Health Organization [18].
Primary epithelial cell cultures from bronchial brushing
The cytology brush with adherent tissue was incubated in 5 ml of trypsin (1/125 dilution in PBS; Invitrogen, Carlsbad, CA) at 37°C for 5 min. Dissociated cells were collected by centrifugation, resuspended in serum- and calcium-free keratinocyte medium containing human recombinant EGF, bovine pituitary extract, cholera toxin, and mycostatin (Invitrogen), and plated into one 3.5-cm diameter, collagen type-1-coated dish. The number of adherent cells after 24 h incubation ranged from <1 to ~20 × 103. Medium was replaced every 4 days. When successful (generally with >5 × 103 cells initially present), proliferating epithelial cells reached 80% confluence (~1 × 106 cells) in 2–3 weeks. Subcultures of the first passage (in 24-well culture plate) were used to determine the cytokine response patterns and cellular homogeneity by immunostaining for expression of cytokeratin (class AE1/3 pancytokeratin). Cultures from the second passage were used for thymidine incorporation. Cells from the third passage were used for confirming consistency of cytokine response profile and spectral karyotyping (SKY) [19]. Of note is that all primary epithelial cultures showed limited life span and ceased to divide after 4 to 7 passages.
Isolation of macrophages, fibroblasts, and type II epithelial cells
Human residual lung tissue was obtained from Roswell Park Cancer Institute Tissue Procurement Service under an IRB-approved protocol. Macrophages were mechanically extracted from minced lung tissue, purified by Histopaque gradient centrifugation (Sigma Chemical Co., St. Louis, MO) and plated at a density of 3 × 105 cells/cm2 in RPMI-1640 containing 10% FCS. After 45 min, adherent cells (>95% macrophages) were incubated in medium (1 ml/1 × 106 cells) containing 1 μg/ml LPS. Conditioned medium was collected after 16 h. The concentrations of cytokines were determined by multiplex immunobead flow cytometry (Luminex Inc., Austin, TX). To obtain fibroblasts, macrophage-depleted lung tissue pieces were incubated with RPMI-1640 containing 10% FCS, antibiotics and mycostatin. After 7–10 days, fibroblasts had grown out of the pieces to ~50% confluence. These were selectively released by digestion with trypsin for 2–5 min at room temperature. Fibroblasts from passage 2 were used for analysis. To obtain type II epithelial cells, macrophage-depleted residual lung tissue pieces were incubated with 5 volumes of trypsin for 15 min at 37°C. Trypsin digestion was repeated. Cells released during the second digestion were cultured as described for bronchial epithelial cells. The homogeneity of epithelial cell population was determined by immunostaining of first passage cultures for surfactant expression (Dako, Captine, CA) [20].
Established lung cell lines
Bronchial epithelial cell lines immortalized with the human papilloma virus E6E7 gene, HBE4 and HBE137 (ATCC, Manassas, VA), were cultured like primary bronchial epithelial cells. The non-small cell lung cancer cell line A549 was grown in DMEM containing 10% FCS, and the NCI lines H23, H125, H358, H441, H522 (ATCC) and ADLC [21] were cultured in RPMI-1640 medium containing 10% FCS.
Cytokine treatments for analysis of signalling
Cells were seeded into 24-well plates. When the cultures reached ~90% confluence, they were incubated for 2 h in serum- and factor-free RPMI-1640 (to reduce signalling effects by serum growth factors or EGF), followed by treatment for 15 min with the same medium containing 100 ng/ml recombinant IL-6, OSM (Amgen Corporation, Seattle, WA), LIF (Wyeth Pharmaceuticals, Cambridge MA), EGF (Gibco-Invitrogen Carlsbad, CA), 100 units/ml IFNγ (Roche Applied Science, Indianapolis, IN), 500 ng/ml insulin or 0.1 μM phorbol myristic acid (Sigma, St. Louis, MO). Cells were washed with PBS and lysed with RIPA buffer containing 0.1 mM orthovanadate and 1:100 diluted protease inhibitor cocktail (Calbiochem, San Diego, CA). LIF receptor (LIFR) expression was induced by treating epithelial cells with growth medium containing 20 nM of the histone deacetylase inhibitor depsipeptide FR901228 (NCI) [22].
Growth analysis
Cells were seeded at 5% confluence (~1–5 × 103 primary epithelial cells, HBE135 and fibroblasts) or 1% confluence (~0.5–4 × 103 cells from established lines) into 24-well culture plates using the appropriate complete growth medium. Replicate cultures were treated with complete growth medium containing serially diluted cytokines or conditioned medium of LPS-activated macrophages. At day 3, the media were replaced and at day 6, the cells were released by trypsin digestion and the number of viable (trypan blue dye-excluding) cells counted in a hemocytometer. In each series, the mean number of cells recovered for cultures treated with growth medium alone (ranged from 0.5 to 4 × 105) was used as used as internal reference (defined as 100%).
Thymidine incorporation
Epithelial cells were seeded with growth medium into 24-well culture plates (~5 × 104 cells per well). After 24 h, duplicate cultures were treated with growth medium containing serially diluted OSM or conditioned medium from LPS-activated macrophages. Twenty-four hours later, 1 μCi of [3H]thymidine (Amersham Biosciences) was added to each culture and incubation of the cells continued for an additional 16 h. Cells were released by trypsin and collected onto filter paper by a cell harvester (Tomtec, Hamden, CT). The amount of incorporated tritium was measured by a scintillation counter (Trilux microbeta, Turku Finland). The mean of the net values of the duplicate wells was expressed relative to the mean incorporation determined for the control cultures in each of the series, which was defined as 100%.
Western blot analysis
Replicate aliquots of cell lysates, containing 10 μg of protein, were electrophoresed on 7.5% polyacrylamide gels [16]. Extracts from paired samples were co-analyzed. The proteins were transferred to protean membranes (Schleicher & Schuell, Keene, NH). Immediately after transfer, the membranes were stained with Ponceau red to verify equal loading and membrane-transfer of proteins. In each experimental series, two replicate blots were horizontally cut at the ~60 kD size position; the upper sections were used to probe for phosphorylated and total STAT3 and the lower sections for phosphorylated ERK and total ERK. Separate blots were used to probe for phosphorylated and total STAT1. In cases where larger higher amounts of cells were recovered, extracts were also analyzed for receptor proteins. The membranes were reacted with antibodies to phospho-specific forms or ERK1/2, STAT1, STAT3 and EGFR (Cell Signalling Technology, Inc., Beverly, MA) and total forms of ERK1/2, STAT1, and STAT3, LIFR (Santa Cruz, Santa Cruz, CA). The membranes were incubated with the appropriate peroxidase-conjugated secondary antibodies (ICN Biomedical, Aurora, OH) and the antibody binding was visualized by enhanced chemiluminescence reaction (Amersham Biosciences Piscataway, NJ). In each experimental series, immunoblots were exposed to X-ray films for various lengths of time (1 sec to 30 min) to obtain images that are in the linear range of signal detection [23]. For pictorial presentation of immunoblot data, the signals for either total STAT3 or ERK1/2 served as loading controls.
Densitometric analysis
Immunoblots were digitalized and quantified with Image Quant TL Software (Amersham Biosciences Piscataway, NJ). The net pixel value for each protein band that lied within the linear range of detection was normalized to the co-analyzed standard. To compare the responses between normal and abnormal epithelial cell cultures from individual donors, as well the responses among different donors, the OSM-induced phosphorylation of STAT1, STAT3 and ERK1/2 in the normal cell cultures of each paired set was used as an internal reference (defined as 100). Similarly, the magnitude of increased or decreased signalling in cells from paired cultures was defined by the ratio with the values obtained for the cells derived from the bronchial sites initially identified as being normal. Since the level of phosphorylated STAT3 in LIF-treated normal cells was generally undetectable, we arbitrarily used 1% as lowest value for calculating fold changes. Thus, the calculated values for LIF effects may represent underestimates. Basal levels of signalling and cytokine responses of cell lines were determined by the level of phosphorylated STAT3 and ERK1/2 and normalized to that values obtained for the same cells treated with OSM (defined as 100%).
Statistical evaluation
The inferences for all the hypothesis tests are based on the significance level of 0.05. All statistical analyses were performed in an exploratory manner. For each combination of biomarker (P-STAT1, P-STAT3, and P-ERK) and treatment (one control and 5 cytokines), the relative increase in phosphorylation in the abnormal cells compared with its matched normal cells from the same patient was subjected to the exact sign test with paired samples. A decrease or increase of DNA syntheses within a treatment (OSM or macrophage factors) by different doses of the treatment was tested using the exact sign test with paired samples and the exact Page's L test. The DNA synthesis values of normal and metaplastic cell cultures in paired samples were compared using the exact Wilcoxon two-sample test.
Results
Transformed lung epithelial cell lines have grossly abnormal patterns of cytokine responsiveness
We predicted that if oncogenic transformation of lung epithelial cells is associated with altered responsiveness to inflammation and reduced growth-inhibition, then the most prominent deviations from the regulatory phenotype should be seen in advanced lung cancer cells. Therefore, as first step, we determined if responsiveness to members of the IL-6 family (IL-6, OSM and LIF) were detectable in established lines of malignant human non-small cell lung carcinoma cells. The responsiveness to the cytokines was defined by measurable phosphorylation of STAT and ERK. This is a treatment reaction that is a measure for the level of active cytokine receptors and downstream signaling pathways in the target cells [14,16]. Normal type II epithelial cells and pulmonary fibroblasts were included in the analyses to gauge cell type-specific differences in signaling reactions (Fig. 1).
Figure 1 Cytokine response of lung cells. A. Primary cultures of the normal type II epithelial cells and fibroblasts from residual lung tissue and the indicated cell lines were treated for 15 min with factors listed at the bottom. Extracts were analyzed by immunoblotting for phosphorylated STAT3 and ERK-1/-2. The level of immunodetectable total ERK-1/-2 indicates protein loading. B. The quantitative values (mean and SD) for the phosphorylation of ERK-1/-2 (open bars) and STAT3 (closed bars) by treatments with LIF, OSM, IL-6 and EGF were determined in 5 independent experiments (expressed relative to the OSM effect, equal 100%). C. Extracts from control treated cultures from A were analyzed for the immunodetectable level of total STAT3 and ERK-1/-2. D. Type II epithelial cells were treated for the indicated length of time with OSM (100 ng/ml) and the level of phosphorylated STAT3 and ERK-1/-2 was determined by immunoblotting.
Phosphorylation of STAT3 and ERK was sufficient for identifying responsiveness to cytokines in specific cell types (Fig. 1A shows representative immunoblots and Fig. 1B presents the quantitative data from independently performed experiments involving 5 separate matched cultures of type II epithelial cells and fibroblasts). In brief, normal type II epithelial cells showed the following features (Fig. 1A, top panel): OSM strongly activated phosphorylation of STAT3 and ERK, while IL-6 was less effective, and LIF did not produce any detectable response. EGF stimulated phosphorylation of ERK in the range of OSM, while insulin was minimally effective. Fibroblasts exhibited a prominent LIF response and a several-fold higher activation of ERK relative to STAT3 by IL-6 cytokines (Fig. 1A, bottom panel), in contrast to normal epithelial cells. The maximal level of receptor action in epithelial cells was reached after 15 min agonist treatment (Fig. 1D, example of OSM on normal type II epithelial cells).
Deviations from the normal regulatory phenotype of epithelial cells were already evident in immortalized bronchial epithelial cells, HBE4 (Fig. 1A) and HBE137 (not shown). These cell lines responded to IL-6 cytokines like normal epithelial cells, but showed a detectable STAT3 signaling by LIF and a ~2-fold higher ERK activation by OSM and IL-6. More profound differences were detected in carcinoma lines (Fig. 1A and not shown). In carcinoma cell lines, the major trends included a strong STAT3 activation by LIF (ADLC, H125), an increased (ADLC, H23) or decreased STAT3 activation by IL-6 (H125, H324), a decreased ERK activation by EGF (H522), and a treatment-independent, constitutive activation of the ERK pathway (H23). Despite the substantial changes in response patterns, STAT and ERK signaling in response to OSM was consistently high in all analyzed lung cancer cell lines. Moreover, the cell-type and line specific differences in STAT3 and ERK signaling were not due different expression of the signaling proteins as demonstrated by the comparable expression level of total STAT3 and ERK proteins among the cell types (Fig. 1C).
Cell line-specific effect of cytokines on proliferation
To relate cytokine signaling with effects on cell proliferation, representative lung cell types were cultured for 6 days in growth medium containing in addition increasing concentrations of OSM, IL-6 or LIF (Fig. 2). In all cell types, OSM reduced in a dose-dependent manner proliferation, with maximal effect at a concentration of 20 to 100 ng/ml. Maximal inhibition varied among the cell types and ranged from >80% (normal primary epithelial cells) to ~40% (HBE137, A549, fibroblasts) and ~20% (H23). IL-6 and LIF did not appreciably alter proliferation of epithelial or fibroblastic cells. The comparative analyses also indicated that immortalization of epithelial cells with the E6E7 gene (HBE4 or HBE137) or constitutive activation of ERK pathway (H23) correlated with a lower growth inhibition by OSM.
Figure 2 Effect of cytokines on growth of lung cell cultures. The cells listed at the top were cultured for 6 days in the presence of serially diluted cytokines or conditioned medium of LPS-activated macrophages (CM MΦ). Dilution 1 for the cytokines was 100 ng/ml and for macrophage medium a 1/10-dilution of the original conditioned medium. The number of viable cells for each culture was expressed relative to the cell number in the control culture (defined as 100%). Values represent means and S.D. of 4 separate cultures.
The same cells were tested for growth in the presence of conditioned medium from lipopolysaccaride (LPS) activated macrophages. Such medium is considered to contain a physiologically relevant mixture of inflammatory mediators (Table 1). The LPS macrophage medium suppressed the proliferation of epithelial cells in a dose-dependent fashion (Fig. 2). Suppression ranged from >80% (primary epithelial cells and A549) to <20% in HBE137 and H23. In contrast, the same treatment led to enhanced proliferation of fibroblasts.
Table 1 Cytokine production of pulmonary macrophages
Cytokine Concentration (ng/ml)*
Minus LPS Plus LPS
IL-1β 0.01 ± 0.01 5.4 ± 1.3
TNFα 0.4 ± 0.3 137 ± 44
IL-6 2.0 ± 0.6 290 ± 113
OSM 0.08 ± 0.01 0.87 ± 0.29
IL-8 27 ± 1 101 ± 26
IL-10 0.16 ± 0.01 8.4 ± 1.5
G-CSF 0.09 ± 0.01 20.3 ± 1.6
IL-12 0.01 ± 0.01 0.18 ± 0.03
TGFβ1 < 0.002 < 0.002
* 2 × 106 macrophages/ml incubated for 16 h with medium containing 1 μg/ml LPS. Mean and S.D., N = 3.
Taken together, these data indicated that cytokine responsiveness and growth regulation by OSM and inflammatory mediators are indeed subject to alterations in lung epithelial cells in part as a function of immortalization and transformation. Characteristic changes in the cellular response patterns may serve as markers for the transformation process or may even contribute functionally to tumorigenesis. One of the key questions is at what stage in the transformation of lung epithelial cells are these changes established. To address this question, we used short-term primary cultures of non-immortalized epithelial cells derived from normal and pathologically distinct stages of premalignant lesions sampled by brushing during bronchoscopy of patients. The experimental approach also allowed us to establish the more basic information of what is the cytokine responsiveness of normal epithelial cells and what is the range among individuals.
Application of primary epithelial cells derived from bronchial brushing
Epithelial cell cultures were established from the brushed bronchial epithelium. Biopsies of the same sites were processed for pathological and cytological evaluation. The pathological findings (normal, metaplasia, dysplasia or non-invasive carcinoma in situ) were subsequently applied in the interpretation of the signaling data (Table 2). No cases of invasive carcinoma or advanced lung cancer were included. From May 2000 to March 2005, 192 separate brushings were taken from 96 patients (EC-1 to EC-96). From these, 113 brushings (59%) yielded sufficient epithelial cells that expanded to cultures suitable for functional analyses. Approximately 50% of sites initially judged to be abnormal by bronchoscopy proved to be normal by histology.
Table 2 Cultures of bronchial epithelial cells used in this study
Number Successful
Patients 96
Brushings (total) 192 113
Paired brushings 89 43
Paired cultures
Normal/Normal 20
Normal/Metaplasia 17
Normal/Dysplasia 2
Normal/Carcinoma 4
Single cultures
Normal 20
Metaplasia 2
Dysplasia 1
Carcinoma 4
Based on immunostaining, every cell preparation consisted of >95% cytokeratin-positive epithelial cells. All cultures derived from normal and abnormal sites showed essentially the same epithelial cell morphology. To determine whether or not the primary cultures demonstrated any gross chromosomal abnormalities, cells were analyzed by SKY. With the exception of one carcinoma, the cells showed normal karyotypes. The carcinoma cell culture carried a chromosome 10 deletion (q11.2-q22) in all metaphase spreads.
Paired epithelial cell cultures indicate the occurrence of altered cytokine responsiveness at early stage of transformation
The cytokine responsiveness of the epithelial cells was determined by the activation of signaling and by the effect on DNA synthesis. Signaling was measured in first passage subcultures by treatment with cytokines and growth factors for 15 min followed by immunodetection of the phosphorylated signaling proteins (representative example in Fig. 3). The level of phosphorylated STAT1, STAT3 and ERK1/2 were quantified and expressed relative to the level of these proteins in OSM-treated normal cultures in each pair (Fig. 4A–F).
Figure 3 Cytokine-specific signalling in bronchial epithelial cells. Paired primary cultures of normal and metaplastic epithelial cells (EC-14) were treated for 15 min with the factors listed at the bottom. The relative levels of phosphorylated STAT1, STAT3, EGFR and ERK, as well as the total STAT3 and ERK, were determined by immunoblotting.
Figure 4 Activation of STAT1, STAT3 and ERK signalling in paired epithelial cell cultures. Paired cultures of epithelial cells, as listed in Table 2, were analyzed for the relative changes in STAT1, STAT2 and ERK signalling mediated by the treatments listed at the bottom. The level of phosphorylation of the signalling proteins was expressed relative to the level determined in the OSM-treated normal epithelial cells in the corresponding pair of cultures (defined as 100). A, C and E represent 20 pairs of the combination normal/normal and B, D and E, represent 23 pairs of the combination of normal/abnormal (metaplasia, dysplasia, carcinoma). G and H, indicate the fold-changes in the level of phosphorylated ERK and STAT3 in the matched of culture pairs of normal/normal (left side panel) and normal/abnormal epithelial cells. The values for the 4 carcinoma cases are shown by open circles.
The analyses of normal bronchial epithelial cells defined the following response pattern. The reaction to cytokine treatment (Fig. 3) was comparable to that of type II epithelial cells (Fig. 1A). Data from 63 separate preparations of normal bronchial epithelial cell cultures indicated that basal level of phosphorylated ERK was consistently low (7.5 ± 5.6 % of OSM level; mean ± SD) and the basal levels of phosphorylated STAT3 and STAT1 were generally low to non-detectable. OSM prominently activated ERK (Fig. 4A &4B) and STAT3 (Fig. 4C &4D). The response was uniformly high among cultures from different individuals. IL-6 response, although quite variable among individuals, was consistently below that of OSM. The level of activated ERK was 37 ± 18% of OSM (Fig. 4A &4B) and activated STAT3 was 63 ± + 27% (Fig. 4C &4D). The two-fold lower activation of ERK relative to STAT3 by IL-6 is due to the difference in signaling by the IL-6 and OSM receptor [16,22]. LIF did not elicit a measurable ERK and STAT3 signaling in any of normal epithelial cell culture. EGF activated ERK to 90 ± 29% of OSM level (Fig. 4A &4B). EGF generally had no measurable effect on STAT3 even when the cells were treated with EGF for longer than 15 minutes (not shown). In all cases, IFNγ did not appreciably activate ERK and STAT3, but yielded highest level of STAT1 phosphorylation (2441 ± 2073% of OSM level (Fig. 4E &4F). The IFNγ response from patient to patient showed the highest variability among the cytokine responses, even though paired cultures from individual patients exhibited highly consistent levels of STAT1 activation by IFNγ.
The reproducibility of detecting comparable patterns in independently derived cell cultures was assessed in those paired cultures in which the samples were initially classified as "abnormal" (or "suspicious") by autofluorescence bronchoscopy but proved normal by subsequent pathological analysis (Fig. 4A,C &4E). When cell cultures were derived from histologically normal areas with abnormal fluorescence, the level and range of signaling reactions elicited by the treatments were essentially identical to those with normal histology and normal fluorescence. ERK activation by OSM was 108 ± 35%, by IL-6 was 31 ± 15%, and by EGF was 85 ± 40%. STAT3 activation by OSM was 97 ± 31% and by IL-6 was 63 ± 27%. The cytokine-induced activation of ERK and STAT3 in paired cultures indicated highly consistent response patterns (Fig. 4G and 4H, left side panels). Confirmed premalignant epithelial cells (metaplasia, dysplasia and carcinoma) exhibited two recurring aberrant phenotypes when compared with normal cells. The first aberrancy was the elevated basal or cytokine-stimulated level of phosphorylated ERK (Fig. 4B and 4G, right panel) and the second was the reactivation of LIFR function (Fig. 4D and 4H, right panel).
In the 23 preneoplastic cell cultures, the basal level of phosphorylated ERK was increased to 22 ± 27% of the OSM value (p = 0.07). Based on two-sample Kolmogorov-Smirnov test, two cases that were cultured from metaplastic foci were above the 95% confidence interval. Preneoplastic cells showed also an overall trend of increased ERK signaling when treated with OSM (133 ± 66%; p = 0.17) or IL-6 (53 ± 38%; p = 0.01) but not with EGF (85 ± 67%; p = 0.46). The ERK signaling by IL-6 and EGF included several cases that lie outside of the 95% confidence interval. For the IL-6 response, 3/23 cases were above and 5/23 cases below the 95% confidence interval, and for EGF response, 4/23 cases were above and 3/23 cases below. STAT3 phosphorylation by OSM and IL-6, and STAT1 phosphorylation by OSM, IL-6 and IFNγ did not show significant differences among the cases tested (Fig. 4C–F), in contrast to ERK phosphorylation.
Among the abnormal epithelial cells with elevated basal ERK phosphorylation (Fig. 4B), the two cell cultures from metaplastic foci in the airway were distinct because the basal ERK phosphorylation that was equivalent to the OSM-treated control culture (Fig. 5A). The increased phosphorylation of ERK, but not of STATs in these metaplastic cells was similar to the regulatory phenotype of NCI H23 tumor cell line (Fig. 1A). Due to the elevated ERK phosphorylation, the response to EGF was obscured, even though a normal or even enhanced level of EGFR could be demonstrated (Fig. 5B). The high basal level phosphorylation of ERK was abrogated by treatment with U0126, an inhibitor for MEK1 (Fig. 5C), suggesting a constitutive activation of the MAPK pathway upstream of ERK in these metaplastic cells.
Figure 5 Deregulated ERK signalling in metaplastic epithelial cells. A. Paired cultures of epithelial cells (EC-85) were analyzed by immunoblotting for cytokine-mediated phosphorylation of STAT1, STAT3 and ERK. Basal level of phosphorylated ERK but not of phosphorylated STATs is elevated. B, Expression of EGFR, STAT3 and ERK in the untreated cultures was determined by immunoblotting. C, Cultures of normal and metaplastic cells were treated for 3 h with serum-free RPMI containing 0.1% carrier DMSO or the same medium with 10 μM U0126. The level of phosphorylated ERK was determined by immunoblotting. D, DNA synthesis was determined in response to treatment with serially diluted OSM and conditioned medium of LPS activated macrophages. One set of cultures of normal and metaplastic cells were also treated for 3 hour with 10 μM U0126 in normal growth medium prior to the addition of [3H]thymidine. The incorporation of [3H]thymidine (mean and range of duplicate cultures) were normalized to the number of cells and expressed relative to the values determined for the control cultures of the normal epithelial cells.
The pair-wise comparison also revealed that 5/23 of the abnormal cases (4 metaplasia and 1 carcinoma) had a LIF response detectable by the activation of ERK and STAT3 (Fig. 3, Fig. 4B,D,G and 4H). Three cases (2 metaplasia and 1 carcinoma) produced a signaling that lied outside of the 95% confidence interval. The carcinoma case also included a prominent reduction of IL-6 receptor function (Fig. 6), which resulted in an overall cytokine response pattern comparable to that of the NCI H125 tumor cell line (Fig. 1A). The changes in signaling reactions correlated with receptor expression. The expression of LIFRα in normal cells was undetectable by immunoblotting whereas expression was observed in the 5 cultures of abnormal cells with LIF response.
Figure 6 Regulated expression of LIFR. Paired cultures of normal and carcinoma cells (EC-9) were incubated with or without 20 nM FR901228 for 6 h and then treated for 15 min with the factors listed at the bottom. The level of phosphorylated and total STAT3 and ERK and total LIFRα were determined by immunoblotting.
The re-expression of LIFR in various cell types is controlled by an epigenetic process that depends on histone acetylation within CpG islands of the LIFR promoter region [22]. To determine whether altered protein acetylation could account for the switch in LIF responsiveness from normal to carcinoma cells, the normal epithelial cells were treated for 6 h with the histone deacetylase inhibitor FR901228. The treatment induced the expression of LIFRα and established a LIF-dependent STAT3 and ERK signaling that was comparable to the level observed in the carcinoma cells (Fig. 6). A similar FR901228 treatment of the carcinoma cells further enhanced LIFR expression and LIF-dependent signaling (Fig. 6). Interestingly, the same treatment reduced he expression of IL-6R in normal cells resulting in a response pattern as found in the corresponding carcinoma cells from the same patient (Fig. 6).
OSM and secreted macrophage factors attenuate proliferation of bronchial epithelial cells
DNA synthesis of the epithelial cells was inhibited by OSM and LPS conditioned medium from of activated macrophages in a dose-dependent fashion (Fig. 7). The data for all cultures identified as normal cells (Fig. 7A) indicated that low concentrations of OSM (≤ 0.1 ng/ml) or LPS conditioned macrophage medium (<10-3 dilution) had no appreciable modulating effects. Only at higher concentrations, suppression of DNA synthesis became evident (p < 0.05). In contrast, several cultures of metaplastic and dysplastic cells exhibited the trend, in which the cells responded to low doses of OSM or conditioned medium from macrophages by a stimulation of DNA synthesis (Fig. 7B). The exact sign test using the paired samples (normal and abnormal) indicated a significant increase of DNA synthesis by low dose (10-3 dilution) of macrophage medium (p = 0.016). The comparison of the data from paired cultures by the Wilcoxon two-sample test also indicated that DNA synthesis in the abnormal epithelial cells was significantly increased at lowest concentration of OSM (p = 0.02) and macrophage medium (p = 0.005) relative to the normal cells. The reduced sensitivity for inhibited DNA synthesis in all cases could be correlated with metaplasia which shows enhanced ERK phosphorylation in response to cytokine treatment (example of dysplastic cells in Fig. 7C). In addition, we noted that in premalignant cells the basal ERK phosphorylation approached the level of OSM-treated cells or EGF-treated cells (Fig. 4B). In addition, the thymidine incorporation by premalignant cells maintained in growth medium was close to two-fold higher than in the corresponding normal cultures (Fig. 5D). As expected, the ERK activity was reduced by treatment in both normal and carcinoma cells with U0126 which correlates with the suppression of DNA synthesis observed (Fig. 5C). High concentrations of OSM or macrophage factors were effective to reduce this intrinsically stimulated DNA synthesis in both normal and premalignant epithelial cell cultures.
Figure 7 Effect of OSM and macrophage factors on DNA synthesis of normal and abnormal epithelial cells. A, Normal epithelial cells (25 cultures) and B abnormal epithelial cells (8 metaplasia and 2 dysplasia) were treated with serially diluted OSM or conditioned macrophage medium and the incorporation of [3H]thymidine determined. All values expressed relative to the untreated cultures in each series. C, Paired cultures of normal and dysplastic epithelial cells (EC-65) were analyzed by immunoblotting for OSM-stimulated ERK phosphorylation and the same cells for inhibition of [3H]thymidine incorporation by OSM and macrophage-factors (mean and range of duplicate values)
Discussion
This study addressed the relationship of inflammation and oncogenic transformation of lung epithelial cells. We sought to identify cellular changes in IL-6 cytokine signaling that correlate with attenuated growth suppression of lung cancer cells. To our knowledge, this is the first report of a functional comparison of cytokine response in short-term primary epithelial cell cultures generated from brushings of normal and abnormal bronchial epithelial sites as characterized by autofluorescence bronchoscopy. We were able to demonstrate for the first time that (a) the response profile of normal epithelial cells and is variable among individuals, and that (b) premalignant epithelial cells already have stable changes in receptor activities for IL-6 cytokines and in signal transduction involving the ERK pathway. Similar changes in the signaling reactions are found in established lung cancer cell lines and these correlate in part with less inhibited cell proliferation.
The interpretation of the results requires the understanding of three underlying issues: (i) the use of primary cell cultures to define the regulatory phenotypes of cells that are representative of premalignant lesion in situ; (ii) the potential mechanism that alters signalling reactions; and (iii) the causal relationship of altered signaling with the pattern of gene expression and growth regulation in abnormal lung cells.
Primary cultures of bronchial epithelial cells as representative of specific lesions
Our hypothesis was that the responsiveness of premalignant epithelial cells to inflammatory mediators and IL-6 cytokines is altered and that these changes permit sustained proliferation in the presence of inflammation. The target of our functional analyses was the proliferating premalignant cells of the central epithelium, in patients who had not yet developed lung cancer. All of the patients have a history of smoking, resulting in the presence of tar deposition and corresponding accumulation of tissue macrophages that harbor the ingested tar particles. We used short-term cultures of proliferating epithelial cells derived from foci of premalignant change as identified by autofluorescence bronchoscopy.
Our goal was to culture abnormal proliferating cells that were representative of a specific lesion identified in the airway with autofluorescence. A tissue sample collected by a focal biopsy provides lesion-specific cellular material for culture, which contains an admixture of epithelial cells, basal cells, fibroblasts and stromal cells [24]. In our collection process, the brush moves across the lesion, collecting a site-specific sample of abnormal cells devoid of non-epithelial cells. When taken as a whole, the cultured cells from the foci of premalignancy in the airway behaved differently than cultured cells from a normal area in the same patient. Our functional analyses (Figs. 3,4,5,6,7) indicated a consistent cytokine response pattern, which has been used as a marker for the phenotype of normal bronchial epithelial cells. The quantitative range of responses observed among preparations likely reflects the individual variation of epithelial cells due to genetic and cellular differentiation differences. We believe that bronchial brushings are optimal for recovering relatively homogenous cultures of proliferating epithelial cells free of macrophages and stromal cells.
As noted above, the cultures of cells obtained from the brushing of abnormal sites may include cells in different stages of transformation as well as some normal epithelial cells. We did find that the relative recovery and growth rate of viable cells from brushings of normal epithelium and from pre-malignant epithelial lesions were comparable in vitro. It is conceivable however that normal cells, if included in the brushing of abnormal sites, could outperform the transformed cell types. An experimental assessment of homogeneity included SKY analysis. In all instances, cells from normal, metaplastic and dysplastic sites proved to be diploid, which is consistent with the findings of Franklin et al. [24]. Since we can not rule out a fractional contribution of normal epithelial cells to the response profile determined for abnormal cells (Fig. 4B,D,F,G and 4H) the results may underestimate the altered phenotype of transformed cells.
We know that regulatory phenotypes may have been influenced by cellular changes introduced by growth ex vivo, and this creates an unavoidable ambiguity of the tissue culture approach. To normalize for this possibility, the comparison of normal and abnormal cells adhered to a strict side-by-side analysis of cultures, which had been generated and maintained under identical conditions. The differences discovered by this approach are considered to describe the effect of the transformation process on cytokines responsiveness (Fig. 4G and 4H) as well as proliferation (Fig. 7).
The sites for brushings representing normal and abnormal epithelium were solely selected based on the visual appearance when viewed during bronchoscopy, under two illumination conditions. For each of the 96 patients, the pathological examination of the sample taken from normal sites confirmed the normal phenotype of the epithelium. In contrast, the examination of the tissue from sites with grossly abnormal fluorescence suggested that only ~50% of the cases involved metaplasia or worse. Epithelial cell cultures from areas of abnormal fluorescence that did not exhibit preneoplasia under histological examination normal represented functionally normal epithelium, despite occasional microscopic evidence of inflammation. The patterns of cytokine signaling and proliferation observed for those cultures were fully compatible with the normal epithelial phenotype (Fig. 4G &4H). It is conceivable that the causes for the abnormal appearance of the epithelial sites involved transient processes in the bronchial environment. But the deviation of such sites from normal epithelium with an altered phenotype could not be detected in vivo with our analytical tools. The manifestation of these processes could be lost during the establishment of primary cell cultures.
Mechanisms of altered responsiveness
In the tissue culture amplification of cells from brushings, the regulatory phenotype represents a stable feature. The epithelial cell responsiveness to cytokines is subject to effective positive and negative feedback mechanisms that could significantly alter the receptor-mediated signaling, depending on cytokine exposure. Reduction of signaling by IL-6 cytokines has been associated with treatment-induced receptor protein down regulation [16,25], expression of signal modifying SOCS proteins [26] or desensitization of gp130 [27]. Enhanced signalling can be the consequence of treatment-induced expression of receptor proteins [16]. All of these reversible regulatory events are presumed to be resolved during in vitro culture period required to establish the primary cells cultures. Treatment of the cells at this point should provide the most accurate reflection of the signaling capability of the cytokine receptor systems and the effect on gene expression and proliferation.
The difference between normal and preneoplastic cell types goes beyond the range of differences observed for duplicate samplings of normal epithelium (Fig. 4), and is ascribed to the stable modification brought about by the transformation process. Possible mechanisms are genetic and epigenetic changes [1,28] that lead to the inactivation of receptor genes such as silencing LIFR by DNA methylation [29] or enhanced expression of LIFR, OSMR, but reduced expression of IL-6R by acetylation of histone and transcription-controlling factors [22]. Whether enhanced signaling through LIFR or IL-6R is also a consequence of gene amplification as noted to be the case for EGFR [30], is not known. Thus far, no example of a duplication of the LIFR gene in transformed cells has been reported.
Aside of altered receptor expression, changes in any of the downstream signal-transducing proteins and enzymes are likely to contribute to the signaling phenotype of the lung cells. For instance, no STAT3 signaling by IL-6 cytokine receptors is found in the prostate PC3 cells because these cells have lost the genes for STAT3 and STAT5 [31]. None of the primary cultures or lung cancer cell lines revealed such a drastic modification of signaling as a function of transformation. A more common observation in lung cancer is the enhanced signaling by IL-6 cytokine receptors towards the STAT and ERK pathway (Figs. 1,3,4,5). While immunoblot analysis did not indicate appreciable changes in the expression levels of these signal-transducing proteins to account for enhanced signaling, further work is needed to investigate two other possible mechanisms. First, enhanced signaling might result from altered level of protein kinases (JAKs, MAPKs), which determine the engagement of the signaling pathways, and secondly, it is possible that phosphatase levels might determine the extent and duration of STAT and MAPK signaling process [14,32].
The constitutive activation of the ERK pathway as well as the enhanced signaling through ERK by both IL-6 cytokines and EGF is the most notable feature in about one quarter of the metaplastic and carcinoma lesions (Fig. 4B &4G). This phenotype, including its U0126 inhibition, points to an aberration of the MAPK pathway that may involve the more commonly found oncogenic mutation of Ras [33]. Alternatively, as already suggested above, a reduced dephosphorylation of ERK by a deficiency in appropriate MAPK phosphatase activities could be involved. These possibilities remain to be identified in cells derived from premalignant lesions.
Altered cytokine receptor signaling as cause for changed cellular response
The profile of gene expression and cellular proliferation are downstream events from the altered signaling by IL-6 cytokines in premalignant epithelial cells. The deregulation of ERK and enhanced proliferation of epithelial cells has been noted in many cancer cell types [23,34], and ERK deregulation appears to also cause the reduced suppression by OSM and inflammatory cytokines (Fig. 5D &7). The mode of action may be twofold: enhancing expression of mitogenic functions (e.g., immediate growth response genes) and the reduced expression of cell cycle arresting proteins (p21 and p27) [14,35]. The precise mode of OSM mediated growth suppression is still unclear. It has been proposed that enhanced STAT3 activity is oncogenic [36]. However, a mutant gp130 that is unable to activate STAT3 also fails to suppress of proliferation, suggesting STAT3 is a potential inhibitor of growth [37]. Studies on OSM action on different breast cancer cell lines cytokines have suggested that the balance between STAT and ERK activation (magnitude and duration) determines growth promotion or growth inhibition [37].
Conclusion
Short-term cultures of bronchial epithelial cells can be established from brushings obtained at bronchoscopy. Analyses of paired cultures of cells derived from normal and abnormal site permits the identification of the responsiveness of the cells to inflammatory mediators. Normal epithelial cells show a highly consistent and strong signaling response to OSM, IL-6, and IFNγ and EGF. Early stage of premalignant transformation is associated with modified activities of signal transduction pathways activated by cytokines. Most prominent alterations include elevated ERK phosphorylation and re-expression and function of LIF receptor. OSM and the factors released by activated pulmonary macrophages suppress proliferation of normal epithelial cells. However, this suppression is significantly reduced in abnormal cells. These cytokine responses in cultured preneoplastic cells are remarkably similar to those seen in established lung cancer cell lines. The data suggest that a change in the signaling reaction to inflammatory mediators as a function of transformation contributes to the capability of lung tumor cells to proliferate in presence of tumor-associated inflammation.
List of abbreviations
EGF, epidermal growth factor; IL-6, interleukin-6; IFNγ, interferon γ LIF, leukemia inhibitory factor; LIFR, leukemia inhibitory factor receptor; ERK, extracellular regulated kinase; LPS, lipopolysaccharides; MAPK, mitogen-activated protein kinase; OSM, oncostatin M; PBS, phosphate buffered saline; SKY, spectral karyotyping; STAT, signal transducer and activator of transcription.
Competing interests
The author(s) have no competing financial or non-financial interests.
Authors' contributions
GML performed all patient-related aspects of the work; ET and FB carried out all cellular and molecular characterizations of the primary cells cultures; DT performed all pathological examinations of biopsies; SR managed the clinical and research data; JY performed all statistical analyses; SM applied SKY analysis to epithelial cell cultures; and HB established the experimental cell system. GML and HB designed the study and drafted 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 the staff of the Tissue Procurement and Flow Cytometry at RPCI for excellent service and Kristin Huntoon for critical reading of the manuscript. This work was supported by NCI grants CA37447 to GML and CA85580 to HB, funds by the Roswell Park Alliance to HB and GML, and Roswell Park Cancer Support Grant CA16056.
==== Refs
Driscoll KE Carter JM Howard BW Hassenbein DG Pepelko W Baggs RB Oberdorster G Pulmonary inflammatory, chemokine, and mutagenic responses in rats after subchronic inhalation of carbon black Toxicol Appl Pharmacol 1996 136 372 380 8619246 10.1006/taap.1996.0045
Ardies CM Inflammation as cause for scar cancers of the lung Integr Cancer Ther 2003 2 238 246 15035887 10.1177/1534735403256332
Philip M Rowley DA Schreiber H Inflammation as a tumor promoter in cancer induction Semin Cancer Biol 2004 14 433 439 15489136 10.1016/j.semcancer.2004.06.006
Lam S Kennedy T Unger M Miller YE Gelmont D Rusch V Gipe B Howard D LeRiche JC Coldman A Gazdar AF Localization of bronchial intraepithelial neoplastic lesions by fluorescence bronchoscopy Chest 1998 113 696 702 9515845
Saccomanno G Archer VE Auerbach O Saunders RP Brennan LM Development of carcinoma of the lung as reflected in exfoliated cells Cancer 1974 33 256 270 4810100
Auerbach O Saccomanno G Kuschner M Brown RD Garfinkel L Histologic findings in the tracheobronchial tree of uranium miners and non-miners with lung cancer Cancer 1978 42 483 489 679150
Bota S Auliac JB Paris C Metayer J Sesboue R Nouvet G Thiberville L Follow-up of bronchial precancerous lesions and carcinoma in situ using fluorescence endoscopy Am J Respir Crit Care Med 2001 164 1688 1693 11719311
Hirsch FR Prindiville SA Miller YE Franklin WA Dempsey EC Murphy JR Bunn PA JrKennedy TC Fluorescence versus white-light bronchoscopy for detection of preneoplastic lesions: a randomized study J Natl Cancer Inst 2001 93 1385 1391 11562389 10.1093/jnci/93.18.1385
Breuer RH Pasic A Smit EF van Vliet E Vonk Noordegraaf A Risse EJ Postmus PE Sutedja TG The natural course of preneoplastic lesions in bronchial epithelium Clin Cancer Res 2005 11 537 543 15701838
Knaapen AM Borm PJ Albrecht C Schins RP Inhaled particles and lung cancer. Part A: Mechanisms Int J Cancer 2004 109 799 809 15027112 10.1002/ijc.11708
Emmendoerffer A Hecht M Boeker T Mueller M Heinrich U Role of inflammation in chemical-induced lung cancer Toxicol Lett 2000 112–113 185 191 10.1016/S0378-4274(99)00285-4
Grant SL Begley CG The oncostatin M signalling pathway: reversing the neoplastic phenotype? Mol Med Today 1999 5 406 412 10462753 10.1016/S1357-4310(99)01540-3
Kortylewski M Heinrich PC Mackiewicz A Schniertshauer U Klingmuller U Nakajima K Hirano T Horn F Behrmann I Interleukin-6 and oncostatin M-induced growth inhibition of human A375 melanoma cells is STAT-dependent and involves upregulation of the cyclin-dependent kinase inhibitor p27/Kip1 Oncogene 1999 18 3742 3753 10391682 10.1038/sj.onc.1202708
Heinrich PC Behrmann I Haan S Hermanns HM Muller-Newen G Schaper F Principles of interleukin (IL)-6-type cytokine signalling and its regulation Biochem J 2003 374 1 20 12773095 10.1042/BJ20030407
Kim H Baumann H Dual signalling role of the protein tyrosine phosphatase SHP-2 in regulating expression of acute-phase plasma proteins by interleukin-6 cytokine receptors in hepatic cells Mol Cell Biol 1999 19 5326 5338 10409724
Blanchard F Wang Y Kinzie E Duplomb L Godard A Baumann H Oncostatin M regulates the synthesis and turnover of gp130, leukemia inhibitory factor receptor alpha, and oncostatin M receptor beta by distinct mechanisms J Biol Chem 2001 276 47038 47045 11602599 10.1074/jbc.M107971200
Loewen G Reid M Tan D Klippenstein D Nava E Natarajan R Mahoney M Bimodality Lung Cancer Screening in High risk Patients Chest 2004 125 163s 164s 15136489 10.1378/chest.125.5_suppl.163S-a
Travis WD Colby TV Corrin B Shimosato Y Brambilla E World Health Organization classification of lung and pleural tumors 1999 Springer Verlag Berlin, Germany
Matsui S. Sait S Jones CA Nowak N Gross KW Rapid localization of transgenes in mouse chromosomes using a combined Spectral Karyotyping/FISH technique Mammalian Genome 2002 13 680 685 12514745 10.1007/s00335-002-2197-0
Tan D Kirley S Li Q Ramnath R Slocum H Brooks J Wu CL Zukerberg L Loss of Cables Protein Expression in Human Non-Small Cell Lung Cancer: A Tissue Microarray Study Human Pathol 2003 34 143 149 12612882 10.1053/hupa.2003.26
Bepler G Koehler A Kiefer P Havemann K Beisenherz K Jaques G Gropp C Haeder M Characterization of the state of differentiation of six newly established human non-small-cell lung cancer cell lines Differentiation 1988 37 158 171 2840315
Blanchard F Kinzie E Wang Y Duplomb L Godard A Held WA Asch BB Baumann H FR90 an inhibitor of histone deacetylases, increases the cellular responsiveness to IL-6 type cytokines by enhancing the expression of receptor proteins Oncogene 1228 21 6264 6277 12214267 10.1038/sj.onc.1205777
Douglas WG Tracy E Tan D Yu J Hicks WL JrRigual NR Loree TR Wang Y Baumann H Development of head and neck squamous cell carcinoma is associated with altered cytokine responsiveness Mol Cancer Res 2004 2 585 593 15498933
Franklin WA Fokvord JM Varella-Garcia M Kennedy T Proudfoot S Cook R Dempsey EC Helm K Bunn PA Miller YE Expansion of bronchial epithelial cell populations by in vitro culture of explants from dysplastic and histologically normal sites Am J Resp Cell Mol Biol 1996 15 297 304
Blanchard F Duplomb L Wang Y Robledo O Kinzie E Pitard V Godard A Jacques Y Baumann H Stimulation of leukemia inhibitory factor receptor degradation by extracellular signal-regulated kinase J Biol Chem 2000 275 28793 28801 10858440 10.1074/jbc.M003986200
Lang R Pauleau AL Parganas E Takahashi Y Mages J Ihle JN Rutschman R Murray PJ SOCS3 regulates the plasticity of gp130 signalling Nat Immunol 2003 4 546 550 12754506 10.1038/ni932
Mahboubi K Kirkiles-Smith NC Karras J Pober JS Desensitization of signalling by oncostatin M in human vascular cells involves cytoplasmic Tyr residue 759 in gp130 but is not mediated by either Src homology 2 domain-containing tyrosine phosphatase 2 or suppressor of cytokine signalling 3 J Biol Chem 2003 278 25014 25023 12724316 10.1074/jbc.M211867200
Huber RM Stratakis DF Molecular oncology – perspectives in lung cancer Lung Cancer 2004 45 S209 S213 15552801 10.1016/j.lungcan.2004.07.973
Blanchard F Tracy E Smith J Chattopadhyay S Wang Y Held WA Baumann H DNA methylation controls the responsiveness of hepatoma cells to leukemia inhibitory factor Hepatology 2003 38 1516 1528 14647063
Hirsch FR Varella-Garcia M Bunn PA JrDi Maria MV Veve R Bremmes RM Baron AE Zeng C Franklin WA Epidermal growth factor receptor in non-small-cell lung carcinomas: correlation between gene copy number and protein expression and impact on prognosis J Clin Oncol 2003 21 3798 3807 12953099 10.1200/JCO.2003.11.069
Clark J Edwards S Feber A Flohr P John M Giddings I Crossland S Stratton MR Wooster R Campbell C Cooper CS Genome-wide screening for complete genetic loss in prostate cancer by comparative hybridization onto cDNA microarrays Oncogene 2003 22 1247 1252 12606952 10.1038/sj.onc.1206247
Morrison DK Davis RJ Regulation of MAP kinase signalling modules by scaffold proteins in mammals Annu Rev Cell Dev Biol 2003 19 91 118 14570565 10.1146/annurev.cellbio.19.111401.091942
Ramakrishna G Sithanandam G Cheng RY Fornwald LW Smith GT Diwan BA Anderson LM K-ras p21 expression and activity in lung and lung tumors Exp Lung Res 2000 26 659 671 11195463 10.1080/01902140150216747
Chang F Steelman LS Lee JT Shelton JG Navolanic PM Blalock WL Franklin RA McCubrey AJ Signal transduction mediated by the Ras/Raf/MEK/ERK pathway from cytokine receptors to transcription factors: potential targeting for therapeutic intervention Leukemia 2003 17 1263 1293 12835716 10.1038/sj.leu.2402945
Klausen P Pedersen L Jurlander J Baumann H Oncostatin M and interleukin 6 inhibit cell cycle progression by prevention of p27kip1 degradation in HepG2 cells Oncogene 2000 19 3675 3683 10951574 10.1038/sj.onc.1203707
Yu H Jove R The STATs of cancer--new molecular targets come of age Nature Rev Cancer 2004 4 97 105 14964307
Lai CF Ripperger J Wang Y Kim H Hawley RB Baumann H The STAT3-independent signalling pathway by glycoprotein 130 in hepatic cells J Biol Chem 1999 274 7793 7802 10075671 10.1074/jbc.274.12.7793
Li C Ahlborn TE Kraemer FB Liu J Oncostatin M-induced growth inhibition and morphological changes of MDA-MB231 breast cancer cells are abolished by blocking the MEK/ERK signalling pathway Breast Cancer Res Treat 2001 6 111 121 11437097 10.1023/A:1010614724664
|
16271139
|
PMC1289280
|
CC BY
|
2021-01-04 16:39:12
|
no
|
BMC Cancer. 2005 Nov 4; 5:145
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-145
|
oa_comm
|
==== Front
BMC Fam PractBMC Family Practice1471-2296BioMed Central London 1471-2296-6-441625313310.1186/1471-2296-6-44DebateA framework to evaluate research capacity building in health care Cooke Jo [email protected] Primary Care and Social Care Lead, Trent Research and Development Unit, formerly, Trent Focus Group, ICOSS Building, The University of Sheffield, 219 Portobello, Sheffield S1 4DP, UK2005 27 10 2005 6 44 44 12 6 2005 27 10 2005 Copyright © 2005 Cooke; licensee BioMed Central Ltd.2005Cooke; 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
Building research capacity in health services has been recognised internationally as important in order to produce a sound evidence base for decision-making in policy and practice. Activities to increase research capacity for, within, and by practice include initiatives to support individuals and teams, organisations and networks. Little has been discussed or concluded about how to measure the effectiveness of research capacity building (RCB)
Discussion
This article attempts to develop the debate on measuring RCB. It highlights that traditional outcomes of publications in peer reviewed journals and successful grant applications may be important outcomes to measure, but they may not address all the relevant issues to highlight progress, especially amongst novice researchers. They do not capture factors that contribute to developing an environment to support capacity development, or on measuring the usefulness or the 'social impact' of research, or on professional outcomes.
The paper suggests a framework for planning change and measuring progress, based on six principles of RCB, which have been generated through the analysis of the literature, policy documents, empirical studies, and the experience of one Research and Development Support Unit in the UK. These principles are that RCB should: develop skills and confidence, support linkages and partnerships, ensure the research is 'close to practice', develop appropriate dissemination, invest in infrastructure, and build elements of sustainability and continuity. It is suggested that each principle operates at individual, team, organisation and supra-organisational levels. Some criteria for measuring progress are also given.
Summary
This paper highlights the need to identify ways of measuring RCB. It points out the limitations of current measurements that exist in the literature, and proposes a framework for measuring progress, which may form the basis of comparison of RCB activities. In this way it could contribute to establishing the effectiveness of these interventions, and establishing a knowledge base to inform the science of RCB.
==== Body
Background
The need to develop a sound scientific research base to inform service planning and decision-making in health services is strongly supported in the literature [1], and policy [2]. However, the level of research activity and the ability to carry out research is limited in some areas of practice, resulting in a low evidence base in these areas. Primary Care, for example, has been identified as having a poor capacity for undertaking research [3-5], and certain professional groups, for example nursing and allied health professionals, lack research experience and skills [5-7]. Much of the literature and the limited research on research capacity building (RCB) has therefore focused on this area of practice, and these professional groups. Policy initiatives to build research capacity include support in developing research for practice, where research is conducted by academics to inform practice decision making, research within or through practice, which encompasses research being conducted in collaboration with academics and practice, and research by practice, where ideas are initiated and research is conducted by practitioners [3,8].
The interventions to increase research capacity for, within, and by practice incorporates initiatives to support individuals and teams, organisations and networks. Examples include fellowships, training schemes and bursaries, and the development of support infrastructures, for example, research practice networks [9-13]. In the UK, the National Coordinating Centre for Research Capacity Development has supported links with universities and practice through funding a number of Research and Development Support Units (RDSU) [14]which are based within universities, but whose purpose is to support new and established researchers who are based in the National Health Service (NHS). However, both policy advisers and researchers have highlighted a lack of evaluative frameworks to measure progress and build an understanding of what works[15,16].
This paper argues for a need to establish a framework for planning and measuring progress, and to initiate a debate about identifying what are appropriate outcomes for RCB, not simply to rely on things that are easy to measure. The suggested framework has been generated through analysis of the literature, using policy documents, position statements, a limited amount of empirical studies on evaluating research RCB, and the experience of one large RSDU based in the UK.
Discussion
The Department of Health within the UK has adopted the definition of RCB as 'a process of individual and institutional development which leads to higher levels of skills and greater ability to perform useful research". (pp1321) [17]
Albert & Mickan cited the National Information Services in Australia [18] who define it as
" an approach to the development of sustainable skills, organizational structures, resources and commitment to health improvement in health and other sectors to multiply health gains many times over.'
RCB can therefore be seen as a means to an end, the end being 'useful' research that informs practice and leads to health gain, or an end in itself, emphasising developments in skills and structures enabling research to take place.
A framework for measuring capacity building should therefore be inclusive of both process and outcome measures [19], to capture changes in both the 'ends' and 'means'; it should measure the ultimate goals, but also measure the steps and mechanisms to achieve them. The notion of measuring RCB by both process and outcome measures is supported within the research networks literature [12,20], and capacity building in health more generally [19,21]. Some argue we should acknowledge 'process as outcome', particularly if capacity building is seen as an end in itself [21]. In this context process measures are 'surrogate' [12], or 'proxy' outcome measures[16]. Carter et al [16]stress caution in terms of using 'proxy' measures in the context of RCB, as there is currently little evidence to link process with outcome. They do not argue against the notion of collecting process data, but stress that evaluation work should examine the relationship of process to outcome. The proposed framework discussed in this paper suggests areas to consider for both process and outcome measurement.
The most commonly accepted outcomes for RCB cited in the literature includes traditional measures of high quality research including publications, conference presentations, successful grant applications, and qualifications obtained. Many evaluations of RCB have used these as outcomes [9,10,22,23]. Some argue that publications in peer reviewed journals are a tall order for the low research skills base in some areas of health care practice [5], and argue for an appropriate time frame to evaluate progress. Process measures in this context could measure progress more sensitively and quickly.
However, using traditional outcomes may not be the whole story in terms of measuring impact. Position statements suggest that the ultimate goal of research capacity building is one of health improvement [17,18,24]. In order for capacity building initiatives to address these issues, outcomes should also explore the direct impact on services and clients: what Smith [25]defines as the social impact of research.
There is a strong emphasis within the primary care literature that capacity building should enhance the ability of practitioners to build their research skills: to support the development of research 'by' and 'with' practice [3,26], and suggests 'added value' to develop such close links to practice. A framework to measure RCB should explore and try to unpack this 'added value', both in terms of professional outcomes,[10] which include increasing professional enthusiasm, and supporting the application of critical thinking, and the use of evidence in practice. Whilst doing research alongside practice is not the only way these skills and attitudes can be developed, it does seem to be an important impact of RCB that should be examined.
The notion of developing RCB close to practice does not necessarily mean that it is small scale just because it is close to the coal face. Obviously, in order for individuals and teams to build up a track record of experience their initial projects may justifiably be small scale, but as individual's progress, they may gain experience to be able to conduct large scale studies, still based on practice problems, working in partnership with others. Similarly networks can support large scale studies as their capacity and infrastructure is developed to accommodate them.
The framework
The framework is represented by Figure 1. It has two dimensions
Figure 1 Research Capacity Building: A Framework for Evaluation.
• Four structural levels of development activity. These include individual, team, organisational, and the network or supra- organisational support level (networks and support units). These are represented by the concentric circles within the diagram.
• Six principles of capacity building. This are discussed in more detail below but include: building skills and confidence, developing linkages and partnerships, ensuring the research is 'close to practice', developing appropriate dissemination, investments in infrastructure, and building elements of sustainability and continuity. Each principle is represented by an arrow within the diagram, which indicates activities and processes that contribute towards capacity building. The arrows cut across the structural levels suggesting that activities and interventions may occur within, and across, structural levels. The arrow heads point in both directions suggesting that principles applied to each structural level could have an impact on other levels.
The framework acknowledges that capacity building is conducted within a policy context. Whilst this paper focuses on measurement at different structural levels, it should be acknowledged that progress and impact on RCB can be greatly nurtured or restricted by the prevailing policy. Policy decisions will influence opportunities for developing researchers, can facilitate collaborations in research, support research careers, fund research directed by practice priorities, and can influence the sustainability and the very existence of supportive infrastructures such as research networks.
The paper will explain the rationale for the dimensions of the framework, and then will suggest some examples of measurement criteria for each principle at different structural levels to evaluate RCB. It is hope that as the framework is applied, further criteria will be developed, and then used taking into account time constraints, resources, and the purpose of such evaluations.
Structural levels at which capacity building takes place
The literature strongly supports that RCB should take place at an individual and organisational level [8,15,27,28]. For example, the conceptual model for RCB in primary care put forward by Farmer & Weston [15] focuses particularly on individual General Practitioners (GPs) and primary care practitioners who may progress from non participation through participation, to become academic leaders in research. Their model also acknowledges the context and organisational infrastructure to support RCB by reducing barriers and accommodating diversity through providing mentorship, collaborations and networking, and by adopting a whole systems approach based on local need and existing levels of capacity. Others have acknowledged that capacity development can be focussed at a team level [11,29]. Jowett et al [30] found that GPs were more likely to be research active if they were part of a practice where others were involved with research. Guidance from a number of national bodies highlights the need for multiprofessional and inter-professional involvement in conducting useful research for practice [3,4,6,31] which implies an appropriate mix of skills and practice experience within research teams to enable this [32]. Additionally, the organisational literature has identified the importance of teams in the production of knowledge [18,33,34].
Developing structures between and outside health organisations, including the development of research networks seems important for capacity building [12,24,34]. The Department of Health in the UK [14] categorizes this supra-organisational support infrastructure to include centres of academic activity, Research & Development Support Units, and research networks.
As interventions for RCB are targeted at different levels, the framework for measuring its effectiveness mirrors this. However, these levels should not be measured in isolation. One level can have an impact on capacity development at another level, and could potentially have a synergistic or detrimental effect on the other.
The six principles of research capacity building
Evaluation involves assessing the success of an intervention against a set of indicators or criteria [35,36], which Meyrick and Sinkler [37] suggest should be based on underlying principles in relation to the initiative. For this reason the framework includes six principles of capacity building. The rationale for each principle is given below, along with a description of some suggested criteria for each principle. The criteria presented are not an exhaustive list. As the framework is developed and used in practice, a body of criteria will be developed and built on further.
Principle 1. Research capacity is built by developing appropriate skills, and confidence, through training and creating opportunities to apply skills
Rationale
The need to develop research skills in practitioners is well established [3,4,6], and can be supported through training [14,26], and through mentorship and supervision [15,24,28]. There is some empirical evidence that research skill development increases research activity [23,38], and enhances positive attitudes towards conducting and collaborating in research [39]. Other studies cite lack of training and research skills as a barrier to doing research [30,31]. The need to apply and use research skills in practice is highlighted in order to build confidence [40]and to consolidate learning.
Some needs assessment studies highlight that research skills development should adopt 'outreach' and flexible learning packages and acknowledge the skills, background and epistemologies of the professional groups concerned [7,15,39,41,42]. These include doctors, nurses, a range of allied health professional and social workers. Developing an appropriate mix of professionals to support health services research means that training should be inclusive and appropriate to them, and adopt a range of methodologies and examples to support appropriate learning and experience [15,31,41]. How learning and teaching is undertaken, and the content of support programmes to reflect the backgrounds, tasks and skills of participants should therefore be measured. For example, the type of research methods teaching offered by networks and support units should reflect a range and balance of skills needed for health service research, including both qualitative and quantitative research methods.
Skills development also should be set in the context of career development, and further opportunities to apply skills to practice should be examined. Policy and position statements [14,26] support the concept of career progression or 'careers escalator', which also enables the sustainability of skills. Opportunities to apply research skills through applications for funding is also important [9,10,22,43,44].
At team and network level Fenton et al [34]suggest that capacity can be increased through building intellectual capacity (sharing knowledge), which enhances the ability to do research. Whilst there is no formal measure for this, an audit of the transfer of knowledge would appear to be beneficial. For example teams may share expertise within a project to build skills in novice researchers [45]which can be tracked, and appropriate divisions of workload through reading research literature and sharing this with the rest of the team/network could be noted.
The notion of stepping outside of a safety zone may also suggest increased confidence and ability to do research. This may be illustrated at an individual level by the practitioner-researcher taking on more of a management role, supervising others, or tackling new methodologies/approaches in research, or in working with other groups of health and research professionals on research projects. This approach is supported by the model of RCB suggested by Farmer and Weston [15] which supports progress from participation through to academic leadership.
Some examples of criteria for measuring skills and confidence levels are give in table 1.
Table 1 Building skills and confidence
Structural level Examples of suggested criteria
Individual • Skills developed (and how)
• Evidence of progressive skill development
• Evidence of confidence building through sharing new skills with others, applying existing skills in new situations, working with other professional groups in research
• Research undertaken
Teams • Skills developed (and how)
• Skill mix of team
• Skill/knowledge transfer tracked- (intellectual capital)
• Evidence of progressive skill development
• Evidence of confidence building through sharing new skills with others, applying existing skills in new situations, working with other professional groups in research
• Research undertaken
Organisational • Evidence of training research needs assessment
• Availability and use of training funds
• Evidence of outreach work undertaken in organisations
• Levels of skills within workforce, and skill mix of the skills across groups
• Evidence of matching novice and experienced researchers
• Research undertaken, funding approved.
Supra organisational (networks and support units) • Provision of flexible learning packages
• Provision of training shaped around the skills, background and needs of differing professional groups
• Examples of knowledge/information transfer (through a variety of mechanisms, including workshops, web-based discussions forums)
• Evidence of outreach work, its take up and use
• Responses to needs based work
• Evidence of secondment opportunities offered and taken up.
Principle 2. Research capacity building should support research 'close to practice' in order for it to be useful
Rationale
The underlying philosophy for developing research capacity in health is that it should generate research that is useful for practice. The North American Primary Care Group [24] defined the 'ultimate goal' of research capacity development as the generation and application of new knowledge to improve the health of individuals and families (p679). There is strong support that 'useful' research is that which is conducted 'close' to practice for two reasons. Firstly by generating research knowledge that is relevant to service user and practice concerns. Many argue that the most relevant and useful research questions are those generated by, or in consultation with, practitioners and services [3,11,24], policy makers [46] and service users [47,48]. The level of 'immediate' usefulness [49] may also mean that messages are more likely to taken up in practice[50]. Empirical evidence suggests that practitioners and policy makers are more likely to engage in research if they see its relevance to their own decision making [31,39,46]. The notion of building research that is 'close to practice' does not necessarily mean that they are small scale, but that the research is highly relevant to practice or policy concerns. A large network of practitioners could facilitate large scale, experimental based projects for example. However, the adoption of certain methodologies is more favoured by practice because of their potential immediate impact on practice [47] and this framework acknowledges such approaches and their relevance. This includes action research projects, and participatory inquiry [31,42]. An example where this more participatory approach has been developed in capacity building is the WeLREN (West London Research Network) cycle [51]. Here research projects are developed in cycles of action, reflection, and dissemination, and use of findings is integral to the process. This network reports high levels of practitioner involvement.
Secondly, building research capacity 'close to practice' is useful because of the skills of critical thinking it engenders which can be applied also to practice decision making [28], and which supports quality improvement approaches in organisations [8]. Practitioners in a local bursary scheme, for example, said they were more able to take an evidence-based approach into their every day practice [9].
Developing a 'research culture' within organisations suggests a closeness to practice that impacts on the ability of teams and individuals to do research. Lester et al [23] touched on measuring this idea through a questionnaire where they explored aspects of a supportive culture within primary care academic departments. This included aspects around exploring opportunities to discuss career progression, supervision, formal appraisal, mentorship, and junior support groups. This may be a fruitful idea to expand further to develop a tool in relation to a health care environment.
Some examples of criteria for measuring the close to practice principle are give in table 2
Table 2 Close to practice
Structural level Examples of suggested criteria
Individuals and teams • Evidence of clinical expertise and 'hunches' within the research questions and projects
• Examples of critical thinking used in practice
• Evidence of patient centred outcome measures in projects, and impact of project on patients' quality of life, including social capital and health gain.
• Use of methodologies that are action orientated
• Use of methodologies that include cost effectiveness approaches
• Evidence on level, and nature, of service user involvement
Organisational • Evidence of informing research questions by gaps in knowledge at an organisational level
• Measurements on a culture where research is 'valued, accepted, encouraged and enjoyed'.
• Evidence of managerial support/involvement on research projects
• Evidence of supporting service user links in research
Supra-organisational (networks and support units) • Evidence of research questions being developed with practice, needs and priorities
• Co-ordination of research programmes between health organisations and university
• Development and use of outcomes measures useful for research and practice
• Development and use of cost effectiveness methodologies
• Action research orientated approaches undertaken
• Development of service user panels
3. Linkages, partnerships and collaborations enhance research capacity building
Rationale
The notion of building partnerships and collaborations is integral to capacity building [19,24]. It is the mechanism by which research skills, and practice knowledge is exchanged, developed and enhanced [12], and research activity conducted to address complex health problems [4]. The linkages between the practice worlds and that of academia may also enhance research use and impact [46].
The linkages that enhance RCB can exist between
• Universities and practice [4,14,43]
• Novice and experienced researchers [22,24,51].
• Different professional groups [2,4,20,34]
• Different health and care provider sectors [4,31,47,52]
• Service users, practitioners and researchers [47,48]
• Researchers and policy makers [46]
• Different countries [28,52]
• Health and industry [53,54]
It is suggested that it is through networking and building partnerships that intellectual capital (knowledge) and social capital (relationships) can be built, which enhances the ability to do research [12,31,34]. In particular, there is the notion that the build up of trust between different groups and individuals can enhance information and knowledge exchange[12]. This may not only have benefits for the development of appropriate research ideas, but may also have benefits for the whole of the research process including the impact of research findings.
The notion of building links with industry is becoming progressively evident within policy in the UK [54] which may impact on economic outcomes to health organisations and the society as a whole[55,56].
Some examples of criteria for measuring linkages and collaborations are given in table 3.
Table 3 Linkages, collaborations and partnerships.
Structural level Examples of suggested criteria
Individual • Who they have worked with: to gain knowledge and to share knowledge
• Evidence of increased number of research partnerships
• Evidence of inter-professional working
Teams • Who the team has worked with: academic and practice
• Network development (work with other teams)
• Evidence of inter-professional and other links
Organisational • Links with universities/RDSUs
• Evidence of joint posts with university
• Evidence of working with other service organisations on research
• Evidence of contribution/memberships to Networks
• Work with funding bodies
Supra-organisational (networks and support units) • Joint posts hosted
• Evidence of research collaboration with practitioners, teams, networks and organisations in health care practice
• Development of links across networks
• International links
4. Research capacity building should ensure appropriate dissemination to maximize impact
Rationale
A widely accepted measure to illustrate the impact of RCB is the dissemination of research in peer reviewed publications, and through conference presentations to academic and practice communities [5,12,26,57]. However this principle extends beyond this more traditional method of dissemination. The litmus test that ultimately determines the success of capacity building is that it should impact on practice, and on the health of patients and comminutes[24] that is; the social impact of research [25]. Smith [25]argues that the strategies of dissemination should include a range of methods that are 'fit for purpose'. This includes traditional dissemination, but also includes other methods, for example, instruments and programmes of care implementation, protocols, lay publications, and publicity through factsheets, the media and the Internet.
Dissemination and tracking use of products and technologies arising from RCB should also be considered, which relate to economic outcomes of capacity building [55]. In the UK, the notion of building health trusts as innovative organisations which can benefit economically through building intellectual property highlights this as an area for potential measurement [56].
Some examples of criteria for measuring appropriate dissemination are given in table 4
Table 4 Appropriate dissemination and impact
Structural level Examples of suggested criteria
Individuals and Teams • Papers in research and practice journals
• Conference presentations
• Applied dissemination of findings
• Evidence of influence on local strategy and planning
Organisational • Ease of access to research undertaken locally
• Seminar programmes relating to research undertaken
• Examples of evidence based practice and applying locally developed knowledge in strategy policy and practice
• Funding to support practitioners and teams to disseminate findings
• Successful applications for intellectual property submitted based on R&D developed in organisation
Supra-organisational (networks and support units) • Papers focussing on health services research, written with practitioners
• Conference presentations at practice- focussed conferences
• Applied dissemination
• Innovative dissemination
• Successful applications for intellectual property submitted based on R&D developed in partnership with health organisations
5. Research capacity building should include elements of continuity and sustainability
Rationale
Definitions of capacity building suggest that it should contain elements of sustainability which alludes to the maintenance and continuity of newly acquired skills and structures to undertake research [18,19]. However the literature does not explore this concept well [19]. This in itself may be partly due problems around measuring capacity building. It is difficult to know how well an initiative is progressing, and how well progress is consolidated, if there are no benchmarks or outcomes against which to demonstrate this.
Crisp et al [19] suggests that capacity can be sustained by applying skills to practice. This gives us some insight about where we might look for measures of sustainability. It could include enabling opportunities to extend skills and experience, and may link into the concept of a career escalator. It also involves utilizing the capacity that has been already built. For example engaging with those who have gained skills in earlier RCB initiatives to help more novice researchers, once they have become 'experts', and in finding an appropriate place to position the person with expertise with the organisation. It could also be measured by the number of opportunities for funding for continued application of skills to research practice.
Some examples of criteria for measuring sustainability and continuity are gibe in table 5
Table 5 Continuity and sustainability
Structural level Examples of suggested criteria
Individual • Successful access to funding for continued application of skills (grants and fellowships)
• Continued contacts with collaborators/linkages
• Examples of continued support and supervision arrangements
Teams • Recognition and matching of skills
• Successful access to funding for continued application of skills
Organisational • Secondment opportunities, available and used
• Local responsive funding access and use
• Recognition and matching of skills
• Examples of continued collaboration
Supra-organisational (networks and support units) • Examples of continued collaboration
• Linked support within career pathways
• Fellowships supported
6. Appropriate infrastructures enhance research capacity building
Rationale
Infrastructure includes structures and processes that are set up to enable the smooth and effective running of research projects. For example, project management skills are essential to enable projects to move forward, and as such should be measured in relation to capacity building. Similarly, projects should be suitably supervised with academic and management support. To make research work 'legitimate' it may be beneficial to make research a part of some job descriptions for certain positions, not only to reinforce research as a core skill and activity, but also to review in annual appraisals, which can be a tool for research capacity evaluation. Information flow about calls for funding and fellowships and conferences is also important. Hurst [42] found that information flow varied between trusts, and managers were more aware of research information than practitioners.
The importance of protected time and backfill arrangements as well as funding to support this, is an important principle to enable capacity building [9, 15, 24, 58]. Such arrangements may reduce barriers to participation and enable skills and enthusiasm to be developed[15]. Infrastructure to help direct new practitioners to research support has also been highlighted[14]. This is particularly true in the light of the new research governance and research ethics framework in the UK [59]. The reality of implementing systems to deal with the complexities of the research governance regulations has proved problematic, particularly in primary care, where the relative lack of research management expertise and infrastructure has resulted in what are perceived as disproportionately bureaucratic systems. Recent discussion in the literature has focused on the detrimental impact of both ethical review, and NHS approval systems, and there is evidence of serious delays in getting research projects started [60]. Administrative and support staff to help researchers through this process is important to enable research to take place [61].
Some examples of criteria for measuring are given in table 6.
Table 6 Infrastructure
Structural level Examples of suggested criteria
Individual • Evidence of project management in projects (objective setting with time scales)
• A description of mentorship and supervision structures
• Research is part of job description and reviewed in annual appraisal
Teams • Evidence of project management in projects
• A description of mentorship and supervision
• Protected time taken
Organisational • Evidence of R&D information dissemination strategies
• Use and availability of protected time
• Evidence of back fill availability and use
• Research is part of annual appraisal for some jobs
• Evidence of help with governance and ethics
Supra-organisational (networks and support units) • The nature of collaborations (co-authorship, order of authorship)
• Organize information exchange events. Description of attendance
Conclusion
This paper suggests a framework which sets out a tentative structure by which to start measuring the impact of capacity building interventions, and invites debate around the application of this framework to plan and measure progress. It highlights that interventions can focus on individuals, teams, organisations, and through support infrastructures like RDSUs and research networks. However, capacity building may only take place once change has occurred at more than one level: for example, the culture of an organisation in which teams and individuals work may have an influence of their abilities and opportunities to do research work. It is also possible that the interplay between different levels may have an effect on the outcomes at other levels. In measuring progress, it should be possible to determine a greater understanding of the relationship between different levels. The framework proposed in this paper may be the first step to doing this.
The notion of building capacity at any structural level is dependent on funding and support opportunities, which are influenced by policy and funding bodies. The ability to build capacity across the principles developed in the framework will also be dependent of R&D strategy and policy decisions. For example, if policy fluctuates in its emphasis on building capacity 'by', 'for' or 'with' practice, the ability to build capacity close to practice will be affected.
In terms of developing a science of RCB, there is a need to capture further information on issues of measuring process and outcome data to understand what helps develop 'useful' and 'useable' research. The paper suggests principles whereby a number of indicators could be developed. The list is not exhaustive, and it is hoped that through debate and application of the framework further indicators will be developed.
An important first step to building the science of RCB should be debate about identifying appropriate outcomes. This paper supports the use of traditional outcomes of measurement, including publications in peer reviewed journals and conference presentations. This assures quality, and engages critical review and debate. However, the paper also suggests that we might move on from these outcomes in order to capture the social impact of research, and supports the notion of developing outcomes which measure how research has had an impact on the quality of services, and on the lives of patients and communities. This includes adopting and shaping the type of methodologies that capacity building interventions support, which includes incorporating patient centred outcomes in research designs, highlighting issues such as cost effectiveness of interventions, exploring economic impact of research both in terms of product outputs and health gain, and in developing action oriented, and user involvement methodologies that describe and demonstrate impact. It also may mean that we have to track the types of linkages and collaborations that are built through RCB, as linkages that are close to practice, including those with policy makers and practitioners, may enhance research use and therefore 'usefulness'. If we are to measure progress through impact and change in practice, an appropriate time frame would have to be established alongside these measures.
This paper argues that 'professional outcomes' should also be measured, to recognize how critical thinking developed during research impacts on clinical practice more generally.
Finally, the proposed framework provides the basis by which we can build a body of evidence to link process to the outcomes of capacity building. By gathering process data and linking it to appropriate outcomes, we can more clearly unpack the 'black box' of process, and investigate which processes link to desired outcomes. It is through adopting such a framework, and testing out these measurements, that we can systematically build a body of knowledge that will inform the science and the art of capacity building in health care.
Summary
• There is currently little evidence on how to plan and measure progress in research capacity building (RCB), or agreement to determining its ultimate outcomes.
• Traditional outcomes of publications in peer reviewed journals, and successful grant applications may be the easy and important outcomes to measure, but do not necessarily address issues to do with the usefulness of research, professional outcomes, the impact of research activity on practice, or on measuring health gain.
• The paper suggests a framework which provides a tentative structure by which measuring the impact of RCB could be achieved, shaped around six principles of research capacity building, and includes four structural levels on which each principle can be applied.
• The framework could be the basis by which RCB interventions could be planned, and progress measured. It could act as a basis of comparison across interventions, and could contribute to establishing a knowledge base on what is effective in RCB in healthcare
Competing interests
The author(s) declare that they have no competing interests.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
My warm thanks go to my colleagues in the primary care group of the Trent RDSU for reading and commenting on earlier drafts of this paper, and for their continued support in practice.
==== Refs
Muir Gray JA Evidence-based Healthcare. How to make health policy and management decisions 1997 Edinburgh, Churchill Livingstone
Department of Health Research and Development for a First Class Service 2000 Leeds, DoH
Mant D National working party on R&D in primary care. Final Report. 1997 London, NHSE South and West.
Department of Health Strategic review of the NHS R&D Levy (The Clarke Report) 1999 , Central Research Department, Department of Health 11
Campbell SM Roland M Bentley E Dowell J Hassall K Pooley J Price H Research capacity in UK primary care British Journal of General Practice 1999 49 967 970 10824340
Department of Health Towards a strategy for nursing research and development 2000 London, Department of Health
Ross F Vernon S Smith E Mapping research in primary care nursing: Current activity and future priorities NT Research 2002 7 46 59
Marks L Godfrey M Developing Research Capacity within the NHS: A summary of the evidence 2000 Leeds, Nuffield Portfolio Programme Report.
Lee M Saunders K Oak trees from acorns? An evaluation of local bursaries in primary care. Primary Health Care Research and Development 2004 5 93 95 10.1191/1463423604pc197xx
Bateman H Walter F Elliott J What happens next? Evaluation of a scheme to support primary care practitioners with a fledgling interest in research. Family Practice 2004 21 83 86 14760051 10.1093/fampra/cmh118
Smith LFP Research general practices: what, who and why? British Journal of General Practice 1997 47 83 86 9101690
Griffiths F Wild A Harvey J Fenton E The productivity of primary care research networks British Journal of General Practice 2000 50 913 915 11141879
Fenton F Harvey J Griffiths F Wild A Sturt J Reflections from organization science of primary health care networks. Family Practice 2001 18 540 544 11604380 10.1093/fampra/18.5.540
Department of Health Research Capacity Development Strategy 2004 London, Department of Health
Farmer E Weston K A conceptual model for capacity building in Australian primary health care research. Australian Family Physician 2002 31 1139 1142 12516520
Carter YH Shaw S Sibbald B Primary care research networks: an evolving model meriting national evaluation. British Journal of General Practice 2000 50 859 860 11141868
Trostle J Research Capacity building and international health: Definitions, evaluations and strategies for success. Social Science and Medicine 1992 35 1321 1324 1462171 10.1016/0277-9536(92)90035-O
Albert E Mickan S Closing the gap and widening the scope. New directions for research capacity building in primary health care Australian Family Physician 2002 31 1038 10341
Crisp BR Swerissen H Duckett SJ Four approaches to capacity building in health: consequences for measurement and accountability Health Promotion International 2000 15 99 107 10.1093/heapro/15.2.99
Ryan Wyke S The evaluation of primary care research networks in Scotland British Journal of General Practice 2001 154 155
Gillies P Effectiveness of alliances and partnerships for health promotion Health Promotion International 1998 13 99 120 10.1093/heapro/13.2.99
Pitkethly M Sullivan F Four years of TayRen, a primary care research and development network Primary Care Research and Development 2003 4 279 283 10.1191/1463423603pc167oa
Lester H Carter YH Dassu D Hobbs F Survey of research activity, training needs. departmental support, and career intentions of junior academic general practitioners British Journal of General Practice 1998 48 1322 1326 9747550
North American Primary Care Research Group What does it mean to build research capacity? Family Medicine 2002 34 678 684 12455253
Smith R Measuring the social impact of research BMJ 2001 323 528 11546684 10.1136/bmj.323.7312.528
Sarre G Capacity and activity in research project (CARP): supporting R&D in primary care trusts 2002
Del Mar C Askew D Building family/general practice research capacity Annals of Family Medicine 2004 2 535 540
Carter YH Shaw S Macfarlane F Primary Care research team assessment (PCRTA): development and evaluation Occasional paper (Royal College of General Practitioners) 2002 81 1 72
Jowett S Macleod J Wilson S Hobbs F Research in Primary Care: extent of involvement and perceived determinants among practitioners for one English region British Journal of General Practice 2000 50 387 389 10897537
Cooke J Owen J Wilson A Research and development at the health and social care interface in primary care: a scoping exercise in one National Health Service region Health and Social Care in the Community 2002 10 435 4444 12485130 10.1046/j.1365-2524.2002.00395.x
Raghunath AS Innes A The case of multidisciplinary research in primary care Primary Care Research and Development 2004 5 265 273
Reagans RZER Networks, Diversity and Productivity: The Social Capital of Corporate R&D Teams. Organisational Science 2001 12 502 517 10.1287/orsc.12.4.502.10637
Ovretveit J Evaluating Health Interventions 1998 Buckingham, Open University 10351264
Meyrick J Sinkler P An evaluation Resource for Healthy Living Centres 1999 London, Health Education Authority 10352950
Hakansson A Henriksson K Isacsson A Research methods courses for GPs: ten years' experience in southern Sweden. British Journal of General Practice 2000 50 811 812 11127172
Bacigalupo B Cooke J Hawley M Research activity, interest and skills in a health and social care setting: a snapshot of a primary care trust in Northern England. Health and Social Care in the Community
Kernick D Evaluating primary care research networks - exposing a wider agenda British Journal of General Practice 2001 51 63 11271882
Owen J Cooke J Developing research capacity and collaboration in primary care and social care: is there enough common ground? Qualitative Social Work 2004 3 398 410 10.1177/1473325004048022
Hurst Building a research conscious workforce Journal of Health Organization and management 2003 17 373 384 14628490
Gillibrand WP Burton C Watkins GG Clinical networks for nursing research International Nursing Review 2002 49 188 193 12243596 10.1046/j.1466-7657.2002.00124.x
Campbell J Longo D Building research capacity in family medicine: Evaluation of the Grant Generating Project Journal of Family Practice 2002 51 593 12160486
Cooke J Nancarrow S Hammersley V Farndon L Vernon W The "Designated Research Team" approach to building research capacity in primary care Primary Health Care Research and Development
Innvaer S Vist G Trommald M Oxman A Health policy- makers' perceptions of their use of evidence: a systematic review Journal of Health Services Research and Policy 2002 7 239 244 12425783 10.1258/135581902320432778
NHS Service Delivery Organisation NHS Service Delivery and Against National R&D programme, National listening exercise 2000 London, NHS SDO
Hanley J Bradburn S Gorin M Barnes M Evans C Goodare HB Involving consumers in research and development in the NHS: briefing notes for researchers. 2000 Winchester, Consumers in NHS Research Support Unit,
Frenk J Balancing relevance and excellence: organisational responses to link research with decision making Social Science and Medicine 1992 35 1397 1404 1462179 10.1016/0277-9536(92)90043-P
National Audit Office. An international review on Governments' research procurement strategies. A paper in support of Getting the evidence: Using research in policy making. 2003 London, The Stationary Office.
Thomas P While A Increasing research capacity and changing the culture of primary care towards reflective inquiring practice: the experience of West London Research Network (WeLReN) Journal of Interprofessional Care 2001 15 133 139 11705009 10.1080/13561820120039865
Rowlands G Crilly T Ashworth M Mager J Johns C Hilton S Linking research and development in primary care: primary care trusts, primary care research networks and primary care academics Primary Care Research and Development 2004 5 255 263 10.1191/1463423604pc201oa
Davies S R&D for the NHS- Delivering the research agenda: ; London. 2005 National Coordinating Centre for Research Capacity Development
Department of Health. Best Research for Best Health: A New National Health Research Strategy. The NHS contribution to health research in England: A consultation 2005 London, Department of Health
Buxton M Hanney S Jones T Estimating the economic value to societies of the impact of health research: a critical review Bulletin of the World Health Organisation 2004 82 733 739
Department of Health. The NHS as an innovative organisation: A framework and guidance on the management of intellectual property in the NHS 2002 London, Department of Health
Sarre G Trent Focus Supporting research and development in primary care organisations: report of the capacity and activity in research project (CARP) 2003
Department of Health Research Governance Framework for Health and Social Care 2001 London, Department of Health.
Hill J Foster N Hughes R Hay E Meeting the challenges of research governance. Rheumatology 2005 44 571 572 15757969 10.1093/rheumatology/keh579
Shaw S Developing research management and governance capacity in primary care organizations: transferable learning from a qualitative evaluation of UK pilot sites Family Practice 2004 21 92 98 14760053 10.1093/fampra/cmh120
|
16253133
|
PMC1289281
|
CC BY
|
2021-01-04 16:29:13
|
no
|
BMC Fam Pract. 2005 Oct 27; 6:44
|
utf-8
|
BMC Fam Pract
| 2,005 |
10.1186/1471-2296-6-44
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-1001626908710.1186/1471-2334-5-100Case ReportEpidural abscess caused by Streptococcus milleri in a pregnant woman Lampen Russell [email protected] Gonzalo [email protected] Division of Infectious Diseases, Department of Internal Medicine, Virginia Commonwealth University Medical Center, Richmond VA, USA2 Quality Health Care, Department of Internal Medicine, Virginia Commonwealth University Medical Center, Richmond VA, USA2005 3 11 2005 5 100 100 25 4 2005 3 11 2005 Copyright © 2005 Lampen and Bearman; licensee BioMed Central Ltd.2005Lampen and Bearman; 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
Bacteria in the Streptococcus milleri group (S. anginosus, S. constellatus, and S. intermedius) are associated with bacteremia and abscess formation. While most reports of Streptococcus milleri group (SMG) infection occur in patients with underlying medical conditions, SMG infections during pregnancy have been documented. However, SMG infections in pregnant women are associated with either neonatal or maternal puerperal sepsis. Albeit rare, S. milleri spinal-epidural abscess in pregnancy has been reported, always as a complication of spinal-epidural anesthesia. We report a case of spinal-epidural abscess caused by SMG in a young, pregnant woman without an antecedent history of spinal epidural anesthesia and without any underlying risk factors for invasive streptococcal disease.
Case presentation
A 25 year old pregnant woman developed neurological symptoms consistent with spinal cord compression at 20 weeks gestation. She underwent emergency laminectomy for decompression and was treated with ceftriaxone 2 gm IV daily for 28 days. She was ambulatory at the time of discharge from the inpatient rehabilitation unit with residual lower extremity weakness.
Conclusion
To our knowledge, this is the first reported case of a Streptococcus milleri epidural abscess in a healthy, pregnant woman with no history of epidural anesthesia or invasive procedures. This report adds to the body of literature on SMG invasive infections. Treatment of SMG spinal-epidural abscess with neurologic manifestations should include prompt and aggressive surgical decompression coupled with targeted anti-infective therapy.
==== Body
Background
Streptococcus milleri group (SMG) bacteria are associated with localized abscess formation, most notably in the liver and brain [1]. SMG bacteremia frequently results from occult abscesses, endocarditis, or an underlying gastro-intestinal malignancy [2]. Spinal epidural abscess with SMG are rare, but have been described in individuals with previous epidural anesthesia or malignancy. We report a spinal epidural abscess in a healthy, pregnant woman.
Case presentation
A 25 year old woman (G2P1) at 20 weeks gestation was in her usual state of good health until 2 weeks prior to hospital presentation. She developed progressive inter-scapular pain followed by lower extremity paresthesias. On the day of admission she noted severe weakness in both legs which developed over the course of 3–4 hours, resulting in the inability to ambulate. She denied fever, rigors, nausea, emesis, or diarrhea prior to admission. She also reported urinary hesitancy for 2–3 days prior to admission, with loss of bladder control on the day of admission. There was no loss of bowel control. At the time of hospital presentation, she was awake, alert, oriented, and not in acute cardiopulmonary distress. The vital signs were stable with a blood pressure of 134/82 mmHg, temperature 97.8°F, pulse 76/minute, and a respiratory rate of 18/minute. She was unable to stand or ambulate. Physical exam revealed ascending sensory deficits to the T4 region below her breasts. Ankle reflexes were absent bilaterally. Patellar reflexes were 1/4 bilaterally and Babinski reflexes were up-going bilaterally. Initial laboratory data revealed a leukocytosis of 18,600/μL (normal 3,700–9,700/μL) hemoglobin of 10.4 gm/dL (normal 12.0–15,0 gm/dl), and electrolytes were within normal limits. Blood cultures were not obtained at time of admission.
An MRI of the thoracic spine, performed at the time of presentation, revealed an irregular mass-like density predominantly at the T1 and T2 levels, with a hyper-intense and serpiginous contour demonstrated in the epidural space (Figure 1). Spinal cord volume loss due to compression was noted in the upper thoracic segments. Owing to the spinal cord compression and associated neurologic symptoms, an emergent laminectomy was performed. Operative tissue samples were sent for pathology and microbiologic analysis. Histopathology revealed nonspecific chronic, active inflammation. Gram stain of the specimens was negative. Operative wound cultures grew Streptococcus milleri on the 3rd post-operative day. Two sets (2 anaerobic and 2 aerobic collection bottles) of blood cultures obtained upon receiving the results of the wound cultures and prior to initiating antibiotic therapy were negative. Vaginal cultures were not obtained either prior or post-operatively to determine if the patient was colonized with SMG.
Figure 1 multi-planar, multi-sequence MRI imaging of the thoracic spine. Pre and post contrast sagittal and axial T1, and turbo T2 sequences acquired. An irregular masslike density predominantly T1 and T2 hyperintense with a serpiginous contour is demonstrated in epidural location, as a mass surrounding the thoracic cord. The lesion extends from the T1 level distally through the lower thoracic spine.
The patient was treated with ceftriaxone 2 gm IV daily for 28 days. An MRI of the abdomen and pelvis revealed no focal fluid collections or abscesses. A transesophageal echocardiogram revealed no valvular pathology. She was discharged to home after 21 days and was able to ambulate with the assistance of a walker. At the time of hospital discharge, partial improvement in the sensory deficit was noted. Six weeks after initial therapy, residual sensory and motor deficits persisted, requiring the patient to ambulate with the aid of a walker. The patient gave birth to a healthy baby boy by vaginal delivery. Ten months after the initial presentation, the patient was ambulating without a walker and had near complete recovery of both the motor and sensory deficits.
Discussion
Spinal epidural abscess is a rare condition that occurs even more infrequently in pregnancy. Hunter and colleagues described a case of S. aureus spinal epidural abscess secondary to posterior vertebral osteomyelitis in a previously healthy, 27 year old pregnant woman [3]. Surgical decompression and antibiotic therapy with intravenous methicillin and gentamicin resulted in prompt recovery and improvement of a mild neurologic deficit.
In another case report, a 22 year old woman presented with infrascapular back pain, paresthesias of the thighs, difficulty on ambulation, and fever 6 days post partum [4]. The delivery had been unremarkable and the patient had not received epidural analgesia. Myelography revealed a complete block at the level of the 4th thoracic vertebrae. An emergent, upper thoracic laminectomy was performed. The causative agent was penicillin sensitive S. aureus. Despite surgical decompression and systemic antibiotics, the neurologic deficits failed to reverse.
While S. aureus has been implicated in spontaneous epidural abscess during pregnancy, S. milleri group bacterial epidural abscess has to date only been described following epidural anesthesia [5,6]. The first case described by Gelfand et al [5] involved a 31 year old woman who had a lumbar epidural catheter placed for anesthesia during vaginal delivery. She returned to the hospital 11 days later with signs and symptoms of spinal compression. A laminectomy was performed and she received a 21 day course of appropriate antibiotics.
A second case recently reported by Schroeder and colleagues[6] described an epidural abscess caused by SMG that occurred following the placement of an epidural catheter for anesthesia in an uncomplicated full term vaginal delivery. The catheter was in place for 6 hours intrapartum and was removed promptly following delivery. The patient developed signs of spinal cord compression on post-partum day 5 and was found to have an epidural abscess by MRI. Laminectomy was performed and the patient was treated with ceftriaxone and clindamycin for 4 weeks.
Although rare, S. milleri has been previously implicated in post-partum spinal-epidural abscesses. However, prior cases were associated with the placement of an invasive, epidural catheter. This likely served as the portal of entry into the spinal-epidural space.
Streptococcus milleri Group (SMG) organisms were previously known as Streptococcus anginosus or Streptococcus milleri-anginosus group. They are gram-positive cocci distinguished by their microaerophilic growth requirements, their formation of colonies <0.5 mm in diameter, and by the presence of a distinct caramel-like odor when cultured [1]. SMG organisms are commensals of the oral cavity and of the gastrointestinal tract. SMG organisms are notorious causes of pyogenic, invasive infections, and have been found in head and neck abscesses, bacteremia with endocarditis, liver abscess, thoracic empyema, brain abscess, and spinal epidural abscess [1]. Although unusual, infective endocarditis by SMG organisms has been reported [1]. Patients with underlying medical conditions, such as cirrhosis, diabetes mellitus, and malignancies, are predisposed to invasive infections with Streptococcus milleri [2].
Colonization of the vagina has also been noted [1]. In a revew of 214 fetal necropsies from second term spontaneous abortions, MacGowen found 40 cases of chorioamnionitis or intrauterine pneumonia; SMG were implicated in 5 of these infections. In two of the five cases, maternal vaginal swabs taken 24 hours prior to delivery showed profuse growth of S. milleri [7]. In a similar study that reviewed clinical specimens of neonates suspected of neonatal infection, S. milleri was found in 7.9% of the 2,510 neonates examined [8]. While these studies demonstrate the pathogenic potential and extent of vaginal colonization with S. milleri in prenatal and neonal infections, it remains unknown if pregnancy increases vaginal colonization.
Spinal epidural abscesses in the non-pregnant population due to S. milleri are also uncommon [5,9-11]. All reported cases involve patients with either a malignancy, a recent epidural catheter, or spinal surgery [1,3,5,6] Our case was unusual as this patient was a healthy, young woman without history of prior surgical intervention, malignancy, or chronic illness. To our knowledge, this is the first reported case of a Streptococcus milleri epidural abscess in a healthy, pregnant woman. Previous reports of pregnancy related SMG infections are associated with neonatal sepsis or maternal puerperal sepsis but not spinal epidural abscess [3].
The etiology of the patient's spinal epidural abscess is unclear. She may have been transiently bacteremic with S. milleri from either an oral, gastrointestinal, or vaginal source with consequent seeding of the central nervous system. Apart from a pelvic examination performed early in the first trimester of pregnancy, no other invasive procedures were reported.
Our decision to treat for 28 days with ceftriaxone was consistent with prior reports of successful treatment for SMG spinal-epidural abscess in non-pregnant women [1]. Emergent surgical decompression and drainage is needed if there is evidence of neurological impairment [12]. For spinal-epidural abscesses without motor or sensory deficits, antibiotic therapy alone may be considered [12]. Significant neurological improvement is unlikely when surgical decompression is delayed greater than 24 hours.
Conclusion
We report a case of spinal-epidural abscess caused by SMG in a young, pregnant woman. The patient had no underlying medical conditions, including cirrhosis, diabetes mellitus, and malignancies, and had not undergone any prior invasive procedures. As such, no risk factors for invasive disease with SMG were identified.
While most reports of SMG infection occur in patients with underlying medical conditions, SMG infections during pregnancy have also been documented. However, reports of SMG infections in pregnancy have typically been associated with either neonatal sepsis or maternal puerperal sepsis, rather than spinal epidural abscess. This case report adds to the literature on pregnancy related SMG infections.
Despite the absence of comorbidities in this case, the degree and rapidity of clinical deterioration was noteworthy and manifested as spinal cord volume loss in the upper thoracic segments with concurrent sensory and motor deficits of the bilateral lower extremities. Given the potential severity of SMG spinal-epidural infections, prompt surgical laminectomy and intravenous antibiotic therapy is of paramount importance.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Both RL and GB were involved in the care of the patient and in the writing of the entire manuscript. Both authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Written consent was obtained from the patient for the publication of the study.
==== Refs
J G Occurrence and Pthogenicity of the Streptococcus milleri Group. Reviews of Infectious Diseases 1988 10 257 284 3287560
R B M BL N LZ Clinical significance of bacteremia involving the "Streptococcus milleri" group: 51 cases and review. Clinical Infectious Diseases 1998 27 385 387 9709892
JC H MD R TK T JC P Spinal Epidural Abcess. Aust NZ Journal of Surgery 1977 47 672-4
CG M R M Puerperal spinal epidural abscess. Lancet 1973 7803 608 609
M G Bakhtian B S Spinal sepsis due to Streptococcus milleri: two cases and review. Review of Infectous Diseases 1991 13 559 563
TH S WA K E N U H K U Spinal epidural abscess--a rare complication after epidural analgesia for labour and delivery. Br J Anaesth 2004 6 896 898
MacGowan AP Terry PB Streptococcus milleri and second trimester abortion J Clin Pathol 1987 40 292 293 3558861
Raymond J Bergeret M Francoual C Chavinie J Gendrel D Neonatal infection with Streptococcus milleri Eur J Clin Microbiol Infect Dis 1995 14 799 801 8536729 10.1007/BF01690996
K G R D P K Acute spinal epidural abscess caused by Streptococcus milleri. Journal of Infection 1988 16 304 3397581
E M RM F B VB G M B D B D Streptococcus milleri spondylodiskitis. Journal of Rheumatology 1990 17 1421 1423 2254907
D S P L E W K G Infections due to Lancefield group F and related Streptococci (S.milleri, S.anginosus). Medicine 1981 60 197 207 7231153
D W P K D R Medical management of spian epidural abscesses: case report and review. Clinical Infectious Diseases 1992 15 22 1617070
|
16269087
|
PMC1289282
|
CC BY
|
2021-01-04 16:28:16
|
no
|
BMC Infect Dis. 2005 Nov 3; 5:100
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-100
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-941625962310.1186/1471-2334-5-94Research ArticleComparison of the systemic inflammatory response syndrome between monomicrobial and polymicrobial Pseudomonas aeruginosa nosocomial bloodstream infections Marra Alexandre R [email protected] Gonzalo ML [email protected] Richard P [email protected] Michael B [email protected] Department of Infectious Diseases, Universidade Federal de São Paulo, São Paulo, Brazil2 Department of Internal Medicine, Medical College of Virginia Campus, Virginia Commonwealth University, Richmond, Virginia, USA2005 31 10 2005 5 94 94 15 6 2005 31 10 2005 Copyright © 2005 Marra et al; licensee BioMed Central Ltd.2005Marra 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
Some studies of nosocomial bloodstream infection (nBSI) have demonstrated a higher mortality for polymicrobial bacteremia when compared to monomicrobial nBSI. The purpose of this study was to compare differences in systemic inflammatory response and mortality between monomicrobial and polymicrobial nBSI with Pseudomonas aeruginosa.
Methods
We performed a historical cohort study on 98 adults with P. aeruginosa (Pa) nBSI. SIRS scores were determined 2 days prior to the first positive blood culture through 14 days afterwards. Monomicrobial (n = 77) and polymicrobial BSIs (n = 21) were compared.
Results
78.6% of BSIs were caused by monomicrobial P. aeruginosa infection (MPa) and 21.4% by polymicrobial P. aeruginosa infection (PPa). Median APACHE II score on the day of BSI was 22 for MPa and 23 for PPa BSIs. Septic shock occurred in 33.3% of PPa and in 39.0% of MPa (p = 0.64). Progression to septic shock was associated with death more frequently in PPa (OR 38.5, CI95 2.9–508.5) than MPa (OR 4.5, CI95 1.7–12.1). Maximal SIR (severe sepsis, septic shock or death) was seen on day 0 for PPa BSI vs. day 1 for MPa. No significant difference was noted in the incidence of organ failure, 7-day or overall mortality between the two groups. Univariate analysis revealed that APACHE II score ≥20 at BSI onset, Charlson weighted comorbidity index ≥3, burn injury and respiratory, cardiovascular, renal and hematologic failure were associated with death, while age, malignant disease, diabetes mellitus, hepatic failure, gastrointestinal complications, inappropriate antimicrobial therapy, infection with imipenem resistant P. aeruginosa and polymicrobial nBSI were not. Multivariate analysis revealed that hematologic failure (p < 0.001) and APACHE II score ≥20 at BSI onset (p = 0.005) independently predicted death.
Conclusion
In this historical cohort study of nBSI with P. aeruginosa, the incidence of septic shock and organ failure was high in both groups. Additionally, patients with PPa BSI were not more acutely ill, as judged by APACHE II score prior to blood culture positivity than those with MPa BSI. Using multivariable logistic regression analysis, the development of hematologic failure and APACHE II score ≥20 at BSI onset were independent predictors of death; however, PPa BSI was not.
==== Body
Background
P. aeruginosa is an important nosocomial BSI pathogen with a high associated mortality [1]. Although the frequency of Gram-negative sepsis has diminished over the last 20 years, the incidence of polymicrobial nBSI infection has increased [2]. In addition, the mortality associated with nosocomial bloodstream infections, particularly those occurring in the intensive care setting, is greater than that of community acquired BSI [3].
Prior studies of nosocomial bloodstream infection (nBSI) have reported a higher associated mortality with polymicrobial nBSI than with monomicrobial nBSI [4]. These studies compared polymicrobial infections with monomicrobial infections caused by diverse pathogens [5]. As such, few investigators have analyzed the clinical significance of polymicrobial versus monomicrobial BSI with a specific pathogen [6]. In addition, a major difficulty in interpreting prior studies is both the marked heterogeneity of variables and the varied definitions of polymicrobial infection [4,6,7]. Little information exists about the systemic inflammatory response in polymicrobial BSI [4,5].
The purpose of this study was to evaluate and compare the inflammatory response, clinical course, and outcomes of monomicrobial and polymicrobial nosocomial BSI due to Pseudomonas aeruginosa.
Methods
Setting
The Virginia Commonwealth University Medical Center (VCUMC) is an 820-bed tertiary care facility in Richmond, Virginia. The hospital houses 9 intensive care units (ICUs), including pediatric ICUs and a burn unit. Approximately 30,000 patients are admitted annually.
Study design
Using the Surveillance and Control of Pathogens of Epidemiological Importance (SCOPE) database of bloodstream infections occurring at 49 U.S. hospitals [1], we identified all patients with a diagnosis of nBSI due to P. aeruginosa at VCUMC from January 1, 1996 through December 31, 2003. Patients were considered to have had BSI due to P. aeruginosa if ≥ 1 blood culture was positive for this organism. Each patient was only included once at the time of the first BSI. Bacteremia was defined as polymicrobial if microorganisms other than P. aeruginosa were recovered from the blood culture within a 24 h period. If the bloodstream isolate was a potential skin contaminant (e.g., diphtheroids, Propionibacterium spp, Bacillus spp, coagulase-negative staphylococci, or micrococci), the presence of an intravascular catheter and the initiation of targeted antimicrobial therapy were required for the diagnosis, as well as at least 1 of the following findings: temperature of >38.0°C or < 36.0°C, chills, and or systolic blood pressure of <90 mmHg. Clinical data were concurrently collected by infection control practitioners using a standardized case report form. The data collected routinely included age, gender, location of the patient (ward vs. ICU), clinical service, duration of hospitalization prior to onset of BSI, predisposing clinical conditions, and bloodstream pathogen. Predisposing clinical conditions were required to be present prior to BSI and included neutropenia (defined as an absolute neutrophil count <500/μl), peritoneal or hemodialysis, and central venous catheters. Sources of secondary BSI were identified by cultures obtained from distant sites that yielded the same pathogen. Adverse outcomes that occurred during the hospital stay were recorded. These included organ failure and mortality (7-day and overall hospital). The clinical condition of each patient was classified daily according to systemic inflammatory response syndrome (SIRS) criteria [SIRS, sepsis, severe sepsis or septic shock] and APACHE II scores from two days prior to the first positive blood culture through 14 days afterwards [8,9]. The severity of underlying disease for each patient was classified using the Charlson weighted comorbidity index [10]. Patients who had nosocomial BSI due to monomicrobial P. aeruginosa (MPa) were compared to patients who had nosocomial BSI due to polymicrobial P. aeruginosa (PPa) nBSI.
Definitions
The patient's physiological conditions prior to the BSI and on the day of BSI were assessed using the APACHE II score [9]. The clinical condition of each patient during the bloodstream infection was classified daily as SIRS, sepsis, severe sepsis or septic shock using criteria previously published by the American College of Chest Physicians / Society of Critical Care Medicine (ACCP/SCCM) [8]. Systemic Inflammatory Response Syndrome (SIRS) was defined as two or more of the following: (a) temperature >38°C or <36°C, (b) heart rate >90 beats per minute, (c) respiratory rate >20 breaths per minute or a PaCO2 <32 mmHg, or (d) white blood cell count >12 × 109/L or <4 × 109/L or the presence of more than 10% immature neutrophils.
Sepsis was defined as SIRS associated with P. aeruginosa isolated from at least one blood culture. Severe sepsis was associated with organ dysfunction, hypotension or systemic manifestations of hypoperfusion. Septic shock was defined as sepsis associated with hypotension unresponsive to intravenous fluid challenge or the need for a vasopressor agent. The presence of organ system failure at the time of BSI and during the clinical course was assessed using the criteria described by Fagon [11]. Nosocomial infection and sources of infection were defined according to Centers for Disease Control and Prevention (CDC) criteria [12]. Adequate empiric antimicrobial treatment was defined as therapy administered within 24 hours after blood culture samples were obtained that included the administration of any antimicrobial agent to which the P. aeruginosa and the other co-pathogens were susceptible [13], except when a susceptible aminoglycoside was used in conjunction with another antimicrobial to which the organisms were resistant or when a susceptible aminoglycoside was used alone.
Microbiological methods
Blood cultures (each consisting of aerobic and anaerobic bottles) were processed at the VCUMC clinical laboratory using the BACTEC® 9240 blood culture system (Becton Dickinson, Sparks MD).
Statistical analysis
For continuous variables, mean values were compared using two sample t-tests for independent samples. Differences in proportions were compared using a Chi-square test or Fisher's exact test when appropriate. Mean values were reported ± 1 SD. All tests of significance are two-tailed. When collinearity was identified, the variable with the greatest measure of association was included in the multivariate analysis. Odds ratios were calculated for all variables. Ninety five percent confidence intervals were calculated for all odd ratios. Variables found to be significant in univariate analysis were then entered into a multivariate model. Alpha was set at 0.05. All statistical analyses were done using the Statistical Package for the Social Sciences software (SPSS, Chicago, IL, USA).
Results
Study population and patient characteristics
A total of 160 nosocomial P. aeruginosa BSIs were identified at VCUMC during the eight-year study period. Of these, 19 clinically significant episodes of BSI (11.9%) were identified in pediatric patients (<18 years of age), 92 episodes were monomicrobial, and 49 episodes were polymicrobial BSI. Fifteen monomicrobial and 12 polymicrobial BSIs had incomplete medical records. In 16 cases co-pathogens were recovered from the blood culture >24 hours after the isolation of Pseudomonas aeruginosa. The remaining 77 monomicrobial and 21 polymicrobial BSIs caused by P. aeruginosa were included in the analysis.
Proportions and means for the different variables in the two groups are listed in Table 1. There were no significant differences in age or gender between the two groups (p = 0.71 and p = 0.72, respectively). Burn injuries were more commonly seen in cases of polymicrobial P. aeruginosa BSI vs. monomicrobial cases (42.9% vs. 10.4%, p = 0.002). Underlying malignancy was more commonly seen in the monomicrobial P. aeruginosa BSI cohort (24.7% vs. 4.8%, p = 0.064). No statistically significant differences were observed in the proportion of patients with diabetes mellitus or gastrointestinal complications between the two BSI groups. No difference was also observed in Charlson scores ≥3 between the two groups (19.0% for PPa vs. 29.9% for MPa, p = 0.32). Although the majority of patients acquired P. aeruginosa BSI in the intensive care unit setting, no statistically significant differences were noticed between the two comparison groups (90.5% for PPa vs. 81.8% for MPa, p = 0.51). A central venous catheter was present in more than three-quarters of both groups (90.5% in PPa vs. 83.1% in MPa, p = 0.51). Twenty-seven percent of the MPa BSI patients received total parental nutrition compared to 14.3% of PPa BSI patients (p = 0.22). A greater proportion of PPa BSI patients (38.1%) received blood transfusions compared with 26.0% of MPa BSI patients; however, this was not statistically significant (p = 0.28). More than half of patients in both groups needed ventilatory support (71.4% in PPa vs. 61.0% in MPa, p = 0.38) prior to the onset of BSI. There was no difference in the use of any class of antimicrobials prescribed prior to Pseudomonas BSI between the two groups (85.7% in PPa vs. 84.4% in MPa, p = 0.88).
Table 1 Patient characteristics and outcomes, stratified by polymicrobial infection.
VARIABLES Polymicrobial (n = 21) Monomicrobial (n = 77) P
N % N %
Demographic characteristics
Age >60 years 7 33.3 29 37.7 0.71
Male gender 17 81.0 46 59.7 0.72
Mean LOS prior to nBSI (days) ± SD (range) 26 ± 28.6 (5–121) - 33 ± 44.7 (2–323) - 0.38
Mean hospital stay (days) ± SD (range) 46.9 ± 32.0 (9–128) - 69.0 ± 75.4 (5–415) - 0.05
ICU stay 19 90.5 63 81.8 0.51
Underlying conditions
Charlson score ≥3 4 19.0 23 29.9 0.32
Burn injury 9 42.9 8 10.4 0.002
Diabetes mellitus 5 23.8 17 22.1 1.0
Neoplasia 1 4.8 19 24.7 0.064
Neutropenia 0 - 10 13.0 0.11
Gastrointestinal diseases 4 19.0 16 20.8 1.0
Therapeutics
Mechanical ventilation 15 71.4 47 61.0 0.38
Central venous line 19 90.5 64 83.1 0.51
Hemodialysis 3 14.3 12 15.6 1.0
TPN 3 14.3 21 27.3 0.22
Transfusion 8 38.1 20 26.0 0.28
Prior antibiotics 18 85.7 65 84.4 0.88
Conditions related to the clinical course
APACHE II score ≥20 at BSI onset 15 71.4 50 64.9 0.58
Mean time to appropriate antimicrobial therapy (days) 3.40 - 1.7 - 0.029
Inadequate antibiotic therapy 18 85.7 37 48.1 0.002
Imipenem resistant P. aeruginosa 6 28.6 20 26.0 0.81
Outcomes
Respiratory failure 15 71.4 58 75.3 0.72
Cardiovascular failure 7 33.3 30 39.0 0.64
Renal failure 11 52.4 28 36.4 0.18
Hematologic failure 6 28.6 26 33.8 0.65
Hepatic failure 2 9.5 10 13.0 1.0
7-day mortality 6 28.6 16 20.8 0.56
Overall mortality 8 38.1 37 48.1 0.42
The mean interval between hospitalization and the onset of BSI did not differ between the monomicrobial and polymicrobial groups (26 ± 28.6 days vs. 33 ± 44.7 days, p = 0.38). However, the overall hospital mean length of stay was significantly longer for MPa BSI (69.0 ± 75.4 vs. 46.9 ± 32.0 days, p = 0.05).
Microbiological features
78.6% of BSIs were caused by monomicrobial P. aeruginosa infection (MPa) and 21.4% by polymicrobial P. aeruginosa infection (PPa). The most frequent pathogens associated with polymicrobial P. aeruginosa BSI were coagulase-negative staphylococci (16.6%), Acinetobacter baumannii (16.6%), Staphylococcus aureus (12.5%) and Candida albicans (12.5%) (Table 2). Polymicrobial infection with more than 2 organisms was seen in 19.1% of PPa BSI cases. No statistically significant differences were observed in the proportion of imipenem resistance in P. aeruginosa isolates between the two groups (28.6% for PPa vs. 26.0% for MPa, p = 0.81).
Table 2 Characteristics of 26 co-pathogens isolated in 21 cases of polymicrobial P. aeruginosa BSI.
Microorganisms Polymicrobial BSI cases (n = 21)
N %
Number of agents (associated with P. aeruginosaBSI)
1 17 80.9
2 3 14.3
3 1 4.8
Agents (n = 26)
CNS 4 15.4
Staphylococcus aureus* 3 11.5
Enterococcus faecalis 1 3.8
Enterococcus faecium** 2 7.7
Streptococcus pneumoniae 1 3.8
Acinetobacter baumannii 4 15.4
Burkholderia cepacia 2 7.7
Enterobacter cloacae 1 3.8
Klebsiella pneumoniae 3 11.5
Klebsiella oxytoca 1 3.8
Serratia marcescens 1 3.8
Candida albicans 3 11.5
CNS = coagulase-negative staphylococci
*Two methicillin-resistant S. aureus
**One vancomycin-resistant E. faecium
Clinical course
Septic shock occurred in 33.3% of PPa and in 39.0% of MPa (p = 0.64). Progression to septic shock was associated with death more frequently in PPa (OR 38.5, CI95 2.9–508.5) than MPa (OR 4.5, CI95 1.7–12.1). Maximal SIR (severe sepsis, septic shock or death) was seen on day 0 for PPa BSI vs. day 1 for MPa (Figure 1). Median APACHE II scores on the day of BSI were 23 in the PPa group and 22 in the MPa group. There was also no difference in APACHE II scores at 14 days post-BSI diagnosis between the MPa and PPa groups (Figure 2). Appropriate empiric antimicrobials were begun within 24 hours in 59.6% of MPa and in 14.3% of PPa (p = 0.002). In addition the time to adequate therapy was twice as long for patients with PPa infection (3.4 days vs. 1.7 days, p = 0.029). No significant difference was noted in the incidence of organ failure, 7-day mortality, or overall mortality between the two groups as seen in table 1.
Figure 1 Systemic inflammatory response over time in patients with P. aeruginosa nBSI stratified by polymicrobial infection.
Figure 2 Mean APACHE II scores in patients with P. aeruginosa nBSI stratified by polymicrobial infection.
Univariate analysis revealed that APACHE II score ≥20 at BSI onset, Charlson score, burn injury, and respiratory, cardiovascular, renal and hematologic failure were associated with death (table 3). Age, malignancy, diabetes mellitus, hepatic failure, gastrointestinal complications, inappropriate empiric antimicrobial therapy, infection with imipenem-resistant P. aeruginosa, and polymicrobial infection were not significant predictors of mortality on univariate analysis. Using logistic regression analysis, the following variables were independent predictors for death (table 3): hematologic failure (OR 16.9; CI95 3.9–73.2) and APACHE II score ≥20 at BSI onset (OR 9.7; CI95 1.9–47.9).
Table 3 Risk factors for in-hospital mortality in patients with P. aeruginosa nosocomial bloodstream infection.
VARIABLES Died (n = 45) Recovered (n = 53) Univariate analysis Multivariate analysis
N % N % OR CI 95% OR CI 95%
Age >60 years 18 40.0 18 34.0 1.3 0.6–2.9
Burn injury 12 26.7 5 9.4 3.5 1.1–10.8 3.2 0.6–18.2
Diabetes mellitus 10 22.2 12 22.6 0.9 0.4–2.5
Gastrointestinal complication 6 13.3 14 26.4 0.4 0.1–1.2
Neoplasia 11 24.4 9 17.0 1.6 0.6–4.2
APACHE II score ≥20 at BSI onset 41 91.1 24 45.3 12.7 3.9–39.5 9.7 1.9–47.9
Charlson score ≥3 17 37.8 10 18.9 2.6 1.0–6.5 2.7 0.7–10.1
Respiratory failure 39 86.7 34 64.2 3.6 1.3–10.1 1.4 0.3–7.2
Cardiovascular failure 26 57.8 11 10.8 5.2 2.1–12.7 2.6 0.7–9.3
Renal failure 25 55.6 14 26.4 3.5 1.5–8.1 1.3 0.4–4.2
Hematologic failure 26 57.8 6 11.3 10.7 3.8–30.2 16.9 3.9–73.2
Hepatic failure 8 17.8 4 7.5 2.6 0.7–9.5
Inadequate antibiotic therapy 26 57.8 29 54.7 1.1 0.7–1.7
Imipenem resistant P. aeruginosa 16 35.6 10 18.9 2.37 0.9–5.9
Polymicrobial infection 8 17.8 13 24.5 0.7 0.2–1.8
Discussion
As polymicrobial BSI with P. aeruginosa is associated with high mortality, we decided to investigate whether the systemic inflammatory response with polymicrobial P. aeruginosa BSI is more intense than the systemic inflammatory response associated with monomicrobial P. aeruginosa BSI. During the study period, we found one-fifth of cases were polymicrobial. Since our intention was to analyze the 14-day period after P. aeruginosa BSI for the intensity of the systemic inflammatory response, we adopted a more conservative definition of polymicrobial BSI (all organisms isolated in the 24 hour period following the first culture positive for P. aeruginosa) rather than the more standard definition (organisms isolated within 48 hours) [14] in order to minimize misclassification of serial monomicrobial infections as polymicrobial infections. It should be noted that the proportion of Pseudomonas BSI cases that were polymicrobial was lower at our hospital than in the SCOPE hospitals overall, where 32.7% of the 1,256 nosocomial Pseudomonas BSI cases were polymicrobial.
It is also important to note that 19.1% of the patients with PPa BSI had 2 or 3 co-pathogens associated with P. aeruginosa. Coagulase negative staphylococci (CNS) were found in 15.4% of the PPa BSIs. It is often difficult to determine the pathogenic role of coagulase-negative staphylococci when these organisms are isolated in blood cultures. To avoid this, we utilized more strict criteria for classification of skin flora as pathogens. In addition, in only one of the four polymicrobial BSIs in which coagulase-negative staphylococci (CNS) was a co-pathogen was a third organism also found. However, in this particular case, multiple blood cultures yielded both Pseudomonas and CNS, which leads us to believe that both organisms played a pathogenic role.
Acinetobacter baumannii was found in 15.4% of the polymicrobial cases. A previous study found that Acinetobacter spp. may be a co-pathogen in P. aeruginosa polimycrobial BSI [6]. Although both pathogens are non-fermentative Gram-negative bacteria causing nosocomial infections, principally in intensive care units, they show different epidemiologic characteristics [1].
Interestingly, no significant difference was noted between the MPa BSI and PPa BSI cohorts with respect to gastrointestinal complications, such as bowel perforation or gastrointestinal procedures. As such, a gastrointestinal source for the etiology of the polymicrobial BSI is not clearly defined. However, burn injury was more common in PPa BSI (p = 0.002). The loss of the natural cutaneous barrier to infection likely led to microbial colonization and subsequent invasion of the bloodstream [15]. Of note was that malignant conditions were more commonly associated with MPa BSI than PPa BSI.
No difference in APACHE II scores was noted between the two comparison groups during 14 days of follow up. On analysis of severe sepsis, septic shock and death, no statistically significant differences were observed between the MPa and PPa groups. A previous study by Aliaga et al. showed that patients with polymicrobial infection involving P. aeruginosa were worse clinically and developed shock more frequently [6]. However, the severity of underlying diseases chosen by these authors was only evaluated by the McCabe classification. In our study, the Charlson weighted comorbidity index and serial APACHE II scores were used to assess the patients' severity of illness (figure 2).
Our study also demonstrated the difficulty in choosing empiric antimicrobial treatment for polymicrobial infections at the time P. aeruginosa was isolated. This difficulty is even greater if organisms such as VRE or Candida are co-pathogens where it is likely that therapy will be inadequate until all pathogens have been identified [13]. This at least partially explains the difference observed in inadequate empiric treatment in patients with PPa compared to patients with MPa (85.7% vs. 48.1%, p = 0.025). It is also important to note that there was no difference in imipenem resistance between the PPa and MPa groups (p = 0.81).
By univariate analysis several organ dysfunctions (respiratory, cardiovascular, renal and hematologic) were associated with death, but age, hepatic failure, inappropriate antimicrobial therapy, imipenem resistance and polymicrobial infection were not. The clinical belief that polymicrobial BSI portends a worse prognosis than monomicrobial BSI was not demonstrated in our study. Other studies, however, have associated polymicrobial infection with higher mortality [4,5]. Another surprising finding was that the mean length of hospital stay was higher for the monomicrobial group than the polymicrobial group. Thus, our findings suggest that the outcome of P. aeruginosa BSI is more closely related to the underlying physiological response to sepsis than it is to polymicrobial infection. However, it must be emphasized that because of the relatively small sample size of our study, a difference of at least 28% between outcome events in the monomicrobial and polymicrobial groups would be necessary to detect a statistically significant difference in SIRS, organ failure or mortality rate. In the SCOPE hospitals overall, the crude mortality of polymicrobial P. aeruginosa BSI (42.4%) was significantly higher than that seen with monomicrobial P. aeruginosa BSI (34.8%), (p = 0.01). Thus, the smaller sample size in our study could have led to a type II error in our finding of no significant difference in mortality between mono- and polymicrobial BSI.
In conclusion, in patients with P. aeruginosa nBSI, one-fifth of cases are polymicrobial, the incidence of septic shock and organ failure is high, patients with PPa BSI are not more severely ill prior to infection than those with MPa BSI, and APACHE II score ≥20 at BSI onset and the development of hematologic failure are independent predictors of death.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
ARM participated in the design of the study, collected the data and performed the statistical analysis. GMLB participated in the design of the study and performed the statistical analysis. RPW participated in the design of the study and coordination. MBE conceived of the study, and participated in its design and coordination. 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 CAPES – Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (Brasília, Brazil).
==== Refs
Wisplinghoff H Bischoff T Tallent SM Seifert H Wenzel RP Edmond MB Nosocomial bloodstream infections in US hospitals: analysis of 24,179 cases from a prospective nationwide surveillance study Clin Infect Dis 2004 39 309 317 15306996 10.1086/421946
Martin GS Mannino DM Eaton S Moss M The epidemiology of sepsis in the United States from 1979 through 2000 New Engl J Med 2003 348 1546 1554 12700374 10.1056/NEJMoa022139
Blot S Vandewoude K De Bacquer D Colardyn F Nosocomial bacteremia caused by antibiotic-resistant Gram-negative bacteria in critically ill patients: clinical outcome and length of hospitalization Clin Infect Dis 2000 34 1600 1606 12032895 10.1086/340616
Pittet D Li N Wenzel RP Association of secondary and polymicrobial nosocomial bloodstream infections with higher mortality Eur J Clin Microbiol Infect Dis 1993 12 813 819 8112351 10.1007/BF02000400
Pittet D Li N Woolson RF Wenzel RP Microbiological factors influencing the outcome of nosocomial bloodstream infections: a 6-year validated, population-based model Clin Infect Dis 1997 24 1068 1078 9195059
Aliaga L Mediavilla JD Llosá J Miranda C Rosa-Fraile M Clinical significance of polymicrobial versus monomicrobial bacteremia involving Pseudomonas aeruginosa Eur J Clin Microbiol Infect Dis 2000 19 871 874 11152313 10.1007/s100960000392
Laupland KB Zygun DA Davies HD Church DL Louie TJ Doig CJ Population-based assessment of intensive care unit-acquired bloodstream infections in adults: Incidence, risk factors, and associated mortality rate Crit Care Med 2002 30 2462 2467 12441755 10.1097/00003246-200211000-00010
American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis Crit Care Med 1992 20 864 874 1597042
Knaus WA Draper EA Wagner DP Zimmerman JE APACHE II: a severity of disease classification system Crit Care Med 1985 13 818 829 3928249
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
Fagon JY Chastre J Novara A Medioni P Gibert C Characterization of intensive care unit patients using a model based on the presence or absence of organ dysfunctions and/or infection: the ODIN model Intensive Care Med 1993 19 137 144 8315120 10.1007/BF01720528
Garner JS Jarvis WR Emori TB Horan TC Hughes JM CDC definitions for nosocomial infections Am J Infect Control 1988 16 128 140 2841893 10.1016/0196-6553(88)90053-3
Ibrahim EH Sherman G Ward S Fraser VJ Kollef MH The influence of inadequate antimicrobial treatment of bloodstream infections on patient outcomes in the ICU setting Chest 2000 118 9 11 10893351 10.1378/chest.118.1.146
Weinstein MP Murphy JR Reller LB Lichtenstein KA The clinical significance of positive blood cultures: a comprehensive analysis of 500 episodes of bacteremia and fungemia in adults. II. Clinical observations, with special reference to factors influencing prognosis Rev Infect Dis 1983 5 54 70 6828812
Mayhall CG The epidemiology of burn wound infections: then and now Clin Infect Dis 2003 37 543 550 12905139 10.1086/376993
|
16259623
|
PMC1289283
|
CC BY
|
2021-01-04 16:28:16
|
no
|
BMC Infect Dis. 2005 Oct 31; 5:94
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-94
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-961625964310.1186/1471-2334-5-96Research ArticleHigh carriage rate of high-level penicillin-resistant Streptococcus pneumoniae in a Taiwan kindergarten associated with a case of pneumococcal meningitis Lauderdale Tsai-Ling [email protected] Wei Yang [email protected] Ming Fang [email protected] I Fei [email protected] Yu Chen [email protected] Kai Sheng [email protected] I-Wen [email protected] Christine C [email protected] Division of Clinical Research, National Health Research Institutes, Zhunan, Taiwan2 Kaohsiung Municipal United Hospital, Kaohsiung, Taiwan3 Department of Pediatrics, Veterans General Hospital-Kaohsiung, Kaohsiung, Taiwan4 National Yang-Ming University, Taipei, Taiwan2005 1 11 2005 5 96 96 4 8 2005 1 11 2005 Copyright © 2005 Lauderdale et al; licensee BioMed Central Ltd.2005Lauderdale 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 Taiwan19F-14 Streptococcus pneumoniae clone and its variants are being found with increasing frequency in the Asia-Pacific region. A 5-year old child with S. pneumoniae meningitis caused by a high-level penicillin resistant strain (MIC = 4 μg/ml) was admitted to a hospital in southern Taiwan. We carried out a study to determine the potential source of this strain.
Methods
Nasopharyngeal cultures were obtained from all children attending the same kindergarten as the index case. To determine their relatedness all isolates were compared by serotype, antimicrobial susceptibility profile and pulsed field gel electrophoresis (PFGE).
Results
A high proportion of the children including the index case (32/78, 41.0%) carried S. pneumoniae in their nasopharynx (NP). The most common serotype was 19F (13/32, 40.6%). The PFGE types of the 19F serotype isolates obtained from the patient's blood, CSF and NP were identical and were related to 11 other serotype 19F NP isolates including 10 that were indistinguishable from the Taiwan19F-14 clone. All 14 isolates had similar high-level penicillin and multi-drug resistance. The serotypes of the other 19 NP isolates included 6A (2), 6B (10), 23F (5), 9V (1) and 3 (1). The overall rate of penicillin resistance in these S. pneumoniae from these children was 87.5% (28/32), with an MIC50 of 2 and MIC90 of 4 ug/ml. In addition, multi-drug resistant-isolates (isolates resistant to 3 different classes of antimicrobials) accounted for 87.5% (28/32) of all isolates.
Conclusion
The high carriage rate of high-level penicillin- and multi-drug- resistant S. pneumoniae in a kindergarten associated with a case of pneumococcal meningitis emphasizes the need for restraint in antibiotic use and consideration of childhood immunization with conjugate pneumococcal vaccine to prevent the further spread of resistant S. pneumoniae in Taiwan.
==== Body
Background
There has been an alarming increase in recent years in the prevalence of penicillin- resistant S. pneumoniae and pneumococcal meningitis caused by penicillin non-susceptible S. pneumoniae. Thus far pneumococcal meningitis caused by high-level penicillin resistant strains (MIC = 4 ug/ml) accounts for only a small portion of the cases reported from various countries around the world [1-6]. A case of pneumococcal meningitis caused by a high-level penicillin resistant S. pneumoniae 19F recently occurred in a 5-year old boy in southern Taiwan. The same microorganism was isolated from his CSF, blood and nasopharynx. We decided to determine whether the source of his infection might be kindergarten children with whom he had close contact. This was based on the knowledge that spread of multi-drug-resistant clones of S. pneumoniae occurs in the day-care center and kindergarten settings [7-10]. The most common clones include Spanish 23F, Taiwan 23F and 19F.
Nasopharyngeal surveillance cultures were performed on all the children who attended the same kindergarten as the patient. All the isolates of S. pneumoniae were characterized by susceptibility to a variety of antimicrobial drugs, serotype and pulsed field gel electrophoresis (PFGE) profiles.
Case report
A 5-year-old boy with a two-day history of fever, vomiting and poor intake was brought to the emergency room of Veterans General Hospital-Kaohsiung, Taiwan in April 2002. He had cough for 2 days before the onset of fever and had visited a pediatric clinic, where oral antibiotics were prescribed. No otitis media or sinusitis was noted upon admission. He also had no prior hospitalizations or any major systemic illness in the past. On admission, his vital signs consisted of a temperature of 38.6°C (ear), pulse rate 108/min, respiratory rate 60/min and blood pressure of 134/87 mmHg. The physical examination revealed a drowsy child with nuchal rigidity and a positive Kernig sign. The Glascow Coma Scale was E1V1 M6. Chest radiography was normal. A lumbar puncture revealed an opening pressure of > 400 mmHg. The total white blood cell count of the CSF was 66/mm3, with 28% neutrophils, 68% lymphocytes, and 4% monocytes. The glucose and protein were 11 mg/dL and 596 mg/dL, respectively. Numerous gram-positive cocci in pairs were seen on microscopic examination. CSF and blood and nasopharyngeal cultures were positive for S. pneumoniae. The C-reactive protein was 12.3 mg/dL (normal <1 mg/dL). The immunoglobulin profile was within normal limits. The patient was intubated. Cefotaxime 200 mg/Kg/D and vancomycin 60 mg/Kg/D were administered immediately after lumbar puncture was performed. Dexamethasone 0.6 mg/Kg/D was also administered prior to the parenteral antibiotics and continued for 4 days. The isolates obtained from blood, nasopharynx and CSF demonstrated an MIC to penicillin of 4 μg/mL. A follow-up lumbar puncture was performed 72 hours after admission. The CSF glucose was 52 mg/dL and protein was 292 mg/dL with a lowered opening pressure. Vancomycin and cefotaxime were administered for a total of 15 days. He became afebrile on the 7th day. He was gradually weaned from the ventilator and fully recovered without neurological sequelae.
Methods
Nasopharyngeal cultures
Surveillance cultures of nasopharynx were performed on all the children who attended the same kindergarten as the patient. Nasopharyngeal culture was also performed on both of the parents and the younger sibling of the index case. Specimens were collected by a single investigator using a cotton swab placed 1–1.5 cm into the nasopharynx. The specimens were immediately inoculated on a 5% sheep blood plate (Becton Dickinson Microbiology System, Cockeysville, MD). All plates were incubated for 24–48 hours at 37°C in 5% carbon dioxide. S. pneumoniae was identified by typical colonial appearance, α-hemolysis, and gram stain. Confirmatory tests included optochin sensitivity and bile solubility tests (Becton Dickinson). All isolates were frozen at -70°C in tryptic soy broth for further analysis.
Antimicrobial susceptibility testing
Minimum inhibitory concentrations were determined using the broth micro-dilution method following the guidelines of Clinical and Laboratory Standards Institute (formerly NCCLS) (CLSI/NCCLS) [11]. A final inoculum of 5 × 105 CFU/ml in Mueller-Hinton broth containing 2–5% lysed horse blood was used to inoculate the Sensititre STPF3 standard plate (Trek Diagnostics, East Essex, England). This device contained the following concentrations of antimicrobials (μg/ml): amoxicillin/clavulanic acid (2–16), cefepime (0.12–2), ceftriaxone (0.03–2), cefotaxime (0.12–4), cefuroxime (0.5–4), chloramphenicol (2 – 16), erythromycin (0.25–2), gatifloxacin (0.5–8), Gemifloxacin (0.03–0.5), levofloxacin (0.5–16), linezolide (0.25–4), meropenem (0.25–2), penicillin (0.03–8), moxifloxacin (0.25–8), tetracycline (0.5–8), trimethoprim/sulfamethoxazole (SXT) (0.5–4), and vancomycin (0.5–4). Interpretive criteria were based on those indicated in CLSI/NCCLS document M100-S14 [12]. In calculating resistance percentages, the 3 isolates from the patient were counted as one.
Serotyping
Serogrouping and serotyping of S. pneumoniae were performed by the Quelling reaction using Omni serum, followed by pool, group, then factor serum (Statens Serum Institut, Copenhagen, Denmark).
Pulsed field gel electrophoresis (PFGE)
Molecular typing of the genomic DNA was performed by PFGE. Preparation of DNA plugs and subsequent digestion by SmaI was performed following previously published protocols [13]. After staining with ethidium bromide, restriction fragments were imaged with an IS-1000 Digital Imaging System (Alpha Innotech Corporation, San Leandro, CA). PFGE patterns were analyzed using CHEF Mapper XA interactive software (version 1.2, Bio-Rad). International clones defined by the Pneumococcal Molecular Epidemiology Network (Spain23F-1, Taiwan23F-15, Taiwan19F-14, Spain6B-2) were used for comparison [14]. Cluster analysis was performed and dendrograms were prepared by the unweighted pair group method with arithmetic averages with the Jaccard coefficient. PFGE pulsotypes were assigned to clusters of isolates based on the published criteria [15].
Results
S. pneumoniae was isolated from the nasopharynx of 32 (41.0%) of the 78 children who attended the same kindergarten, including the index case (Table 1). The age of the children in the current study ranged from 4 to 6.5 years old. None of the children in the kindergarten including the index case had received conjugate pneumococcal vaccine. No S. pneumoniae was found in the nasopharynx of the parents and the younger sibling of the index case. A total of 34 isolates were available for further analysis including the CSF, blood and nasopharyngeal isolates from the index case and 31 isolates from the nasopharynx of the other children attending the kindergarten. The most common serotype isolated from the nasopharynx was 19F (13/32, 40.6%). This was followed by serotypes 6B (10/32, 31.3%,) and 23F (5/32, 15.6%). There were two isolates of serotype 6A and one isolate each of serotypes 9V and 3. The isolates of S. pneumoniae recovered from CSF, blood and nasopharynx of the index case were identical by PFGE (Figure 1, A2 pulsotype). Ten of the 12 19F isolates from the other children were identical to the Taiwan19F-14 clone (Figure 1 A1 pulsotype, Figure 2 lanes 2 and 3). The isolates from the index case differed from the Taiwan19F-14 clone by 4 bands, indicating it was a variant of the Taiwan19F-14 clone (Figure 1 A2 and A1 pulsotypes, Figure 2 lanes 1 and 3).
Table 1 Distribution of serotypes, PFGE patterns and antimicrobial susceptibility profiles of 34 isolates of S. pnuemoniae isolates from 32 children attending a kindergarten in Kaohsiung, Taiwan
Isolatea Serotype PFGE typeb MIC (ug/ml) of:c
PEN AUG FRX CRO FTX FEP MEM CHL ERY LEV LID SXT
P151a 19F A2 4 4 >4 2 2 4 1 8 >2 1 1 >4
P154a 19F A2 4 4 >4 2 2 4 1 8 >2 1 1 >4
P156a 19F A2 4 4 >4 2 2 4 1 4 >2 1 1 >4
P174 19F A1 2 ≤2 4 1 1 0.5 0.5 8 >2 1 2 4
P176 19F A1 2 ≤2 4 1 1 1 0.5 4 >2 2 2 4
P183 19F A1 2 ≤2 >4 1 1 1 0.5 8 >2 1 2 4
P186 19F A1 2 ≤2 4 1 1 1 0.5 8 >2 1 2 4
P192 19F A1 2 ≤2 4 1 1 1 0.5 4 >2 2 2 4
P198 19F A1 2 ≤2 >4 1 1 1 0.5 8 >2 2 2 4
P199 19F A1 2 ≤2 >4 1 1 1 0.5 8 >2 1 2 4
P200 19F A1 4 ≤2 >4 1 1 1 0.5 4 >2 1 2 4
P203 19F A1 2 ≤2 >4 1 1 2 0.5 8 >2 2 2 4
P204 19F A1 4 ≤2 >4 1 1 1 0.5 4 >2 1 2 4
P179 19F 4 ≤2 >4 1 1 1 0.5 4 >2 1 2 4
P190 19F 4 ≤2 >4 2 2 2 0.5 4 >2 1 1 >4
P177 6B B1 4 ≤2 >4 1 1 1 0.5 4 >2 1 1 >4
P178 6B B1 2 ≤2 >4 1 1 1 0.5 4 >2 1 2 >4
P180 6B B1 4 4 >4 1 1 1 0.5 4 >2 1 2 4
P181 6B B1 4 4 >4 1 1 1 0.5 4 >2 1 1 >4
P185 6B B1 4 ≤2 >4 1 1 1 0.5 4 >2 1 2 >4
P194 6B B1 2 ≤2 >4 1 1 1 0.5 4 >2 1 2 >4
P195 6B B1 2 ≤2 >4 1 1 1 0.5 4 >2 1 2 >4
P201 6B B1 4 ≤2 >4 1 1 1 0.5 4 >2 1 2 >4
P182 6B 2 ≤2 4 1 0.5 1 ≤0.25 4 >2 1 2 >4
P197 6B ≤0.03 ≤2 ≤0.5 ≤0.06 ≤0.12 ≤0.12 ≤0.25 4 ≤0.25 ≤0.5 1 ≤0.5
P189 23F C1 2 ≤2 >4 1 1 1 ≤0.25 8 >2 1 2 ≤0.5
P205 23F C1 2 ≤2 >4 1 1 1 0.5 8 >2 1 2 4
P175 23F 4 ≤2 >4 2 1 2 0.5 8 >2 1 2 4
P188 23F 4 ≤2 >4 2 1 2 0.5 8 >2 1 2 4
P196 23F 2 ≤2 >4 2 1 1 0.5 8 >2 1 2 >4
P191 6A D1 ≤0.03 ≤2 ≤0.5 ≤0.06 ≤0.12 ≤0.12 ≤0.25 4 >2 1 1 4
P202 6A D1 0.06 ≤2 ≤0.5 ≤0.06 ≤0.12 ≤0.12 ≤0.25 8 >2 1 1 4
P184 9V 4 4 >4 2 2 2 0.5 8 >2 2 1 >4
P187 3 ≤0.03 ≤2 ≤0.5 ≤0.06 ≤0.12 ≤0.12 ≤0.25 4 ≤0.25 2 1 ≤0.5
a Isolates P151, P154, and P156 were from blood, nasopharynx, and CSF of the meningitis patient, respectively. All other isolates were from nasopharynx of children attending the same kindergarten as the patient.
b PFGE type, Pulse field gel electrophoresis pulsotype; see Figure 1.
c AUG, amoxicillin/clavulanic acid; CHL, chloramphenicol; CRO, ceftriaxone; ERY, erythromycin; FEP, cefepime; FRX, cefuroxime; FTX, cefotaxime; LEV, levofloxacin; LID, Linezolid; MEM, meropenem; PEN, penicillin; SXT, trimethoprim/sulfamethoxazole. All isolates were resistant to tetracycline (MIC > 8 ug/ml) except isolate P187, which was susceptible (MIC ≤ 0.5 ug/ml). All isolates were susceptible to vancomycin (MIC ≤ 0.5 ug/ml).
Figure 1 Dendrogram of 34 S. pneumoniae pediatric isolates (isolates starting with P) based on PFGE results. Reference strains: 27336, R6; ATCC 700669, Spain23F-1; ATCC 700670 Spain 6B-2; ATCC 700905 Taiwan19F-14; ATCC 700906 Taiwan23F-15.
Figure 2 PFGE fingerprint patterns of SmaI restriction digest of serotype 19F S. pneumoniae isolates. M, lambda ladder molecular size markers (shown in kbp); lane 1, meningitis patient isolates; lane 2, ATCC700905 (Taiwan19F-14); lanes 3 – 5, other serotype 19F nasopharyngeal isolates from children attending the same kindergarten as the patient. Numbers at the bottom of the gel indicate the number of isolates with the same pattern.
All serotype 19F isolates (15 isolates from 13 children) were high-level penicillin (MIC ≥ 2 ug/ml) and multi-drug resistant. The three isolates from the case exhibited similar antibiogram as the other 12 serotype 19F nasopharyngeal isolates from his kindergarten contacts (Table 1). Although the 19F S. pneumoniae isolates exhibited similar antibiograms, the isolates from the patient were more resistant with higher amoxicillin/clavulanic acid, cefotaxime and cefepime MIC. In addition, based on meningitis interpretive criteria, isolates obtained from the index case were resistant to cefotaxime (MIC 2 ug/ml), cefepime (MIC 4 ug/ml), and ceftriaxone (2 ug/ml). The serotypes isolated from the other 19 children also exhibited high-level penicillin resistance except for serotypes 6A and a single strain of 6B, and 3. The overall rate of highly penicillin resistant S. pneumoniae (MIC 2 – 4 μg/ml) isolated from the nasopharynx of these 32 children was 87.5% (28/32). In addition, 13 (40.6%) of these 32 isolates had a penicillin MIC of ≥ 4 ug/ml. The rate of resistance to erythromycin, cefuroxime, trimethoprim/sulfamethoxazole and tetracycline were also high; at 93.8% (30/32), 87.5% (28/32), 90.6% (29/32) and 96.9% (31/32), respectively. Multi-drug resistant-isolates (isolates resistant to 3 different classes of antimicrobials) accounted for 87.5% (28/32) of all isolates. Most isolates remained susceptible to amoxicillin/clavulanic acid (81.25%, 26/32) and cefotaxime (84.4%, 27/32) despite the high prevalence of penicillin resistance. All isolates demonstrated universal susceptibility to all fluoroquinolones including gatifloxacin, gemifloxacin and moxifloxacin (data not shown) in addition to levofloxacin. All isolates were also uniformly susceptible to linezolid and vancomycin (Table 1).
Discussion
Nasopharyngeal carriage of S. pneumoniae in day-care center attendees has been well documented [9,10,16,17]. Studies have found the carriage rates to range from 44%–65% at day-care centers, with one study reporting 53% of isolates highly resistant to penicillin, and another study reporting 13–19% of multi-drug resistance in serotype 14 isolates [16,17]. Taiwan has an extremely high carriage rate of penicillin-resistant S. pneumoniae among children attending day care centers and kindergartens [8]. The current report demonstrates an even higher nasopharyngeal carrier rate of penicillin resistant strains of 87.5% compared to 71.5% previously reported by our group [8]. In addition, 40.6% of the isolates in the present study had a penicillin MIC of ≥ 4 ug/ml, the highest percent of isolates with such high MICs reported in the literature [1-4,6]. The absence of resistance to fluoroquinolones is probably related to their uncommon use by pediatricians.
It is unclear why 41.0% of the children in the Kindergarten carried a variety of strains of S. pneumoniae yet only one child developed bacterial meningitis with serotype 19F. The child did not have history of recurrent infection and his immunoglobulin profiles were normal. It is unlikely that he had a congenital or acquired immunodeficiency or asplenia that predisposed him to invasive pneumococcal infection. Further studies are needed to determine if the invading strain is more virulent than the others. We have no ready explanation why the S. pneumoniae serotype 19F isolated from the index case exhibited a different albeit possibly related PFGE pattern from those isolated from his kindergarten contacts. The nasopharyngeal swab was collected at the same time as the CSF and blood cultures. We do not know how long the patient was colonized before the onset of his disease. However, it is well recognized that nasopharyngeal colonization precedes pneumococcal infection and studies have shown that S. pneumoniae strains recovered from the CSF of the meningitis patients to be the same as the strains carried in the nasopharynx of patients [18,19].
The Taiwan19F clone is the most common serotype that causes pneumococcal diseases in Asian countries [20]. A recent study of S. pneumoniae isolates from patients with meningitis in Japan found high-level penicillin resistance to be associated with the presence of multiple (three) abnormal penicillin binding protein (pbp) genes. Serotype 19F was among those showing the greatest high-level penicillin resistance [6]. Another report from Japan found that the most prevalent serotype in a major Japanese medical center to be 19F, the majority of which were Taiwan19F-14 clone and its variants [21]. High-level antibiotic resistant strains of the Taiwan19F clone and its variants are also relatively common in invasive and non-invasive pneumococcal infections in New Zealand [13,22].
The other serotypes isolated from children in the current study were similar to those encountered in our previous surveillance studies of day care centers and kindergartens in Kaohsiung, in which serotype 23F was the most common, followed by 19F, 6B, 6A, and 14 [23]. We were surprised by the absence of serotype 14, a serotype commonly found in nasopharyngeal and invasive isolates in children ≤ 5 years old, especially those ≤ 2 years old [3,4,24]. The differences between the current and previous studies in the distribution of serotypes may be accounted for by differences in the ages of the children. The children in the current study were older (4 – 6.5 years) compared to 2 months to 7 years in the prior study.
Conclusion
S. pneumoniae and Neisseria meningitidis became the leading causes of meningitis in the United States and Netherlands after the widespread use of H. influenzae type b vaccine [5]. With the introduction of conjugate pneumococcal vaccine, there is now considerable evidence showing substantial decreases in invasive pneumococcal diseases and concomitant penicillin resistance in vaccine and vaccine-related serotypes in children who were vaccinated [25,26]. Although we did not find S. pneumoniae carriage in the parents and the young sibling of the index case, spread of S. pneumoniae and antibiotic-resistant serotypes from day-care center attendees to their siblings has been reported [17,27]. In addition, prior antibiotic use has been shown to contribute to increased S. pneumoniae colonization and diseases [16,17,28]. In February 2001 the Bureau of National Health Insurance in Taiwan implemented a new rule restricting antimicrobial prescription for acute upper respiratory infections in ambulatory patients. This has significantly decreased antimicrobial consumption in this country. Nevertheless, there are still an excessive number of ambulatory patient visits for respiratory infections resulting in antimicrobials being prescribed [29]. It is hoped that a combination of restraint in antibiotic use and implementation of childhood immunization with conjugate pneumococcal vaccine can reduce the burden of pneumococcal illness and multidrug resistant strains in Taiwan and other countries.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
TLL supervised the molecular study and antimicrobial susceptibility testing, carried out the analysis, and prepared the final manuscript. WYL assisted in specimen and data collection of the children attending the kindergarten, and identification and shipment of specimen. MFC, IFH, YCL, and KSH assisted in taking care of the patients and specimen collection. IWH carried out the molecular and antimicrobial susceptibility testing. CCC conceived and coordinated the study, carried out the specimen collection and analysis, and prepared the draft manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank Jui-Fen Lai for her assistance in serotyping. We also wish to thank Dr. Calvin M. Kunin for his helpful review of the manuscript. This work was supported in part by an intramural grant from the National Health Research Institutes (CL-093-PP-01) and a grant from the Veterans General Hospital – Kaohsiung, Taiwan (VGHKS92-81).
==== Refs
de Andrade AL Pimenta FC Laval CA de Andrade JG Guerra ML Brandileone MC Invasive pneumococcal infection in a healthy infant caused by two different serotypes J Clin Microbiol 2004 42 2345 6 15131230 10.1128/JCM.42.5.2345-2346.2004
Di Fabio JL Castaneda E. Agudelo CI De La HF Hortal M Camou T Echaniz-Aviles G Noemi M Barajas C Heitmann I Hormazabal JC Brandileone MC Dias V Regueira M Ruvinski R Corso A Lovgren M Talbot JA De Quadros C Evolution of Streptococcus pneumoniae serotypes and penicillin susceptibility in Latin America, Sireva-Vigia Group, 1993 to 1999. PAHO Sireva-Vigia Study Group. Pan American Health Organization Pediatr Infect Dis J 2001 20 959 967 11642630
Doit C Loukil C Geslin P Bingen E Phenotypic and genetic diversity of invasive pneumococcal isolates recovered from French children J Clin Microbiol 2002 40 2994 8 12149364 10.1128/JCM.40.8.2994-2998.2002
Pantosti A D'Ambrosio F Tarasi A Recchia S Orefici G Mastrantonio P Antibiotic susceptibility and serotype distribution of Streptococcus pneumoniae causing meningitis in Italy, 1997–1999 Clin Infect Dis 2000 31 1373 9 11096005 10.1086/317502
Swartz MN Bacterial meningitis – a view of the past 90 years N Engl J Med 2004 351 1826 8 15509815 10.1056/NEJMp048246
Ubukata K Chiba N Hasegawa K Kobayashi R Iwata S Sunakawa K Antibiotic susceptibility in relation to penicillin-binding protein genes and serotype distribution of Streptococcus pneumoniae strains responsible for meningitis in Japan, 1999 to 2002 Antimicrob Agents Chemother 2004 48 1488 94 15105095 10.1128/AAC.48.5.1488-1494.2004
Barnes DM Whittier S Gilligan PH Soares S Tomasz A Henderson FW Transmission of multidrug-resistant serotype 23F Streptococcus pneumoniae in group day care: evidence suggesting capsular transformation of the resistant strain in vivo J Infect Dis 1995 171 890 6 7706816
Chiou CC Liu YC Huang TS Hwang WK Wang JH Lin HH Yen MY Hsieh KS Extremely high prevalence of nasopharyngeal carriage of penicillin-resistant Streptococcus pneumoniae among children in Kaohsiung, Taiwan J Clin Microbiol 1998 36 1933 7 9650939
Reichler MR Allphin AA Breiman RF Schreiber JR Arnold JE McDougal LK Facklam RR Boxerbaum B May D Walton RO The spread of multiply resistant Streptococcus pneumoniae at a day care center in Ohio J Infect Dis 1992 166 1346 53 1431252
Dagan R O'Brien KL Modeling the association between pneumococcal carriage and child-care center attendance Clin Infect Dis 2005 40 1223 6 15825021 10.1086/428585
Clinical and Laboratory Standards Institute (formerly NCCLS) Methods for Dilution Antimicrobial Susceptibility Tests for Bacteria That Grow Aerobically, Approved Standard – Sixth Edition. M7-A6 2003 Clinical and Laboratory Standards Institute. Wayne, PA
Clinical and Laboratory Standards Institute (formerly NCCLS) Performance standards for antimicrobial susceptibility testing, 14th informational supplement. M100-S14 2004 Clinical and Laboratory Standards Institute. Wayne, PA
Bean DC Ikram RB Klena JD Molecular characterization of penicillin non-susceptible Streptococcus pneumoniae in Christchurch, New Zealand J Antimicrob Chemother 2004 54 122 9 15150169 10.1093/jac/dkh271
McGee L McDougal L Zhou J Spratt BG Tenover FC George R Hakenbeck R Hryniewicz W Lefevre JC Tomasz A Klugman KP Nomenclature of major antimicrobial-resistant clones of Streptococcus pneumoniae defined by the pneumococcal molecular epidemiology network J Clin Microbiol 2001 39 2565 71 11427569 10.1128/JCM.39.7.2565-2571.2001
Tenover FC Arbeit RD Goering RV Mickelsen PA Murray BE Persing DH Swaminathan B Interpreting chromosomal DNA restriction patterns produced by pulsed-field gel electrophoresis: criteria for bacterial strain typing J Clin Microbiol 1995 33 2233 9 7494007
Boken DJ Chartrand SA Goering RV Kruger R Harrison CJ Colonization with penicillin-resistant Streptococcus pneumoniae in a child-care center Pediatr Infect Dis J 1995 14 879 84 8584316
Craig AS Erwin PC Schaffner W Elliott JA Moore WL Ussery XT Patterson L Dake AD Hannah SG Butler JC Carriage of multidrug-resistant Streptococcus pneumoniae and impact of chemoprophylaxis during an outbreak of meningitis at a day care center Clin Infect Dis 1999 29 1257 64 10524972 10.1086/313451
Bogaert D De Groot R Hermans PW Streptococcus pneumoniae colonisation: the key to pneumococcal disease Lancet Infect Dis 2004 4 144 54 14998500 10.1016/S1473-3099(04)00938-7
de Andrade AL Pimenta FC Brandileone MC Laval CA Guerra ML de Andrade JG Di Fabio JL Genetic relationship between Streptococcus pneumoniae isolates from nasopharyngeal and cerebrospinal fluid of two infants with Pneumococcal Meningitis J Clin Microbiol 2003 41 3970 2 12904432 10.1128/JCM.41.8.3970-3972.2003
Song JH Jung SI Ko KS Kim NY Son JS Chang HH Ki HK Oh WS Suh JY Peck KR Lee NY Yang Y Lu Q Chongthaleong A Chiu CH Lalitha MK Perera J Yee TT Kumarasinghe G Jamal F Kamarulzaman A Parasakthi N Van PH Carlos C So T Ng TK Shibl A High prevalence of antimicrobial resistance among clinical Streptococcus pneumoniae isolates in Asia (an ANSORP study) Antimicrob Agents Chemother 2004 48 2101 7 15155207 10.1128/AAC.48.6.2101-2107.2004
Kasahara K Maeda K Mikasa K Uno K Takahashi K Konishi M Yoshimoto E Murakawa K Kita E Kimura H Clonal dissemination of macrolide-resistant and penicillin-susceptible serotype 3 and penicillin-resistant Taiwan 19F-14 and 23F-15 Streptococcus pneumoniae isolates in Japan: a Pilot Surveillance Study J Clin Microbiol 2005 43 1640 1645 15814978 10.1128/JCM.43.4.1640-1645.2005
Bean DC Klena JD Characterization of major clones of antibiotic-resistant Streptococcus pneumoniae in New Zealand by multilocus sequence typing J Antimicrob Chemother 2005 55 375 8 15681581 10.1093/jac/dki001
Chiou CC McEllistrem MC Novel penicillin-, cephalosporin-, and macrolide-resistant clones of Streptococcus pneumoniae serotypes 23F and 19F in Taiwan which differ from international epidemic clones J Clin Microbiol 2001 39 1144 7 11230444 10.1128/JCM.39.3.1144-1147.2001
Hausdorff WP Feikin DR Klugman KP Epidemiological differences among pneumococcal serotypes Lancet Infect Dis 2005 5 83 93 15680778
Bogaert D Hermans PW Adrian PV Rumke HC de Groot R Pneumococcal vaccines: an update on current strategies Vaccine 2004 22 2209 20 15149779 10.1016/j.vaccine.2003.11.038
Kaplan SL Mason EO JrWald ER Schutze GE Bradley JS Tan TQ Hoffman JA Givner LB Yogev R Barson WJ Decrease of invasive pneumococcal infections in children among 8 children's hospitals in the United States after the introduction of the 7-valent pneumococcal conjugate vaccine Pediatrics 2004 113 443 9 14993532 10.1542/peds.113.3.443
Givon-Lavi N Fraser D Porat N Dagan R Spread of Streptococcus pneumoniae and antibiotic-resistant S. pneumoniae from day-care center attendees to their younger siblings J Infect Dis 2002 186 1608 14 12447737 10.1086/345556
Radetsky MS Istre GR Johansen TL Parmelee SW Lauer BA Wiesenthal AM Glode MP Multiply resistant pneumococcus causing meningitis: its epidemiology within a day-care centre Lancet 1981 2 771 3 6116903 10.1016/S0140-6736(81)90184-7
Ho M Hsiung CA Yu HT Chi CL Yin HC Chang HJ Antimicrobial usage in ambulatory patients with respiratory infections in Taiwan, 2001 J Formos Med Assoc 2004 103 96 103 15083239
|
16259643
|
PMC1289284
|
CC BY
|
2021-01-04 16:28:16
|
no
|
BMC Infect Dis. 2005 Nov 1; 5:96
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-96
|
oa_comm
|
==== Front
BMC Med GenetBMC Medical Genetics1471-2350BioMed Central London 1471-2350-6-381626289110.1186/1471-2350-6-38Research ArticleAging syndrome genes and premature coronary artery disease Low Adrian F [email protected]'Donnell Christopher J [email protected] Sekar [email protected] Brendan [email protected] Claudia U [email protected] Stanley Y [email protected] Patrick T [email protected] Calum A [email protected] Cardiovascular Research Center and Cardiology Division, Massachusetts General Hospital, and Harvard Medical School, Boston, MA, USA2 Cardiology Division, Massachusetts General Hospital, and Harvard Medical School, Boston, MA, USA3 The Framingham Heart Study, Framingham MA, and the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, USA2005 31 10 2005 6 38 38 12 2 2005 31 10 2005 Copyright © 2005 Low 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
Vascular disease is a feature of aging, and coronary vascular events are a major source of morbidity and mortality in rare premature aging syndromes. One such syndrome is caused by mutations in the lamin A/C (LMNA) gene, which also has been implicated in familial insulin resistance. A second gene related to premature aging in man and in murine models is the KLOTHO gene, a hypomorphic variant of which (KL-VS) is significantly more common in the first-degree relatives of patients with premature coronary artery disease (CAD). We evaluated whether common variants at the LMNA or KLOTHO genes are associated with rigorously defined premature CAD.
Methods
We identified 295 patients presenting with premature acute coronary syndromes confirmed by angiography. A control group of 145 patients with no evidence of CAD was recruited from outpatient referral clinics. Comprehensive haplotyping of the entire LMNA gene, including the promoter and untranslated regions, was performed using a combination of TaqMan® probes and direct sequencing of 14 haplotype-tagging single nucleotide polymorphisms (SNPs). The KL-VS variant of the KLOTHO gene was typed using restriction digest of a PCR amplicon.
Results
Two SNPs that were not in Hardy Weinberg equilibrium were excluded from analysis. We observed no significant differences in allele, genotype or haplotype frequencies at the LMNA or KLOTHO loci between the two groups. In addition, there was no evidence of excess homozygosity at the LMNA locus.
Conclusion
Our data do not support the hypothesis that premature CAD is associated with common variants in the progeroid syndrome genes LMNA and KLOTHO.
==== Body
Background
CAD is the most common cause of death in the developed world and an increasingly important cause of mortality in the developing world. The dominant pathophysiologic paradigm in CAD is that observed in a relatively rare inherited form of atherosclerosis; Familial Hypercholesterolemia caused by mutations in the LDL receptor[1]. In this condition a primary endocytic abnormality leads, at least partly though excess elevations in LDL cholesterol, to focal endothelial injury, and chronic inflammatory lesions of the arterial wall. Ultimately through plaque rupture and thrombosis there is episodic end-artery occlusion. Importantly, while the later phases of this process appear to be relevant to many forms of CAD, the earliest injury is unknown in most forms of vascular disease precluding truly preventative strategies. Recently the identification of Mendelian forms of coronary disease has suggested that other major pathways may contribute to each stage of the process[2,3]. Family history is a major risk factor for premature CAD, but the genetic contributions to common forms of the disease are unknown[4]. The high prevalence of CAD in older age groups suggests that some forms of atherosclerosis are an integral part of the aging process, and inherited premature aging (or progeroid) syndromes are associated with extensive vascular disease[5]. The causal genes underlying two such rare Mendelian forms of aging recently have been identified.
Discrete mutations in the LMNA gene have been demonstrated to cause a range of inherited syndromes including a variant of Emery-Dreifus muscular dystrophy, dilated cardiomyopathy with conduction disease and Charcot-Marie-Tooth disease[6]. The mechanisms of this remarkable pleiotropy are unknown, but have been attributed to discrete functions of different lamin domains in individual tissues. At least two of the syndromes caused by lamin mutations include premature CAD. Some lamin mutations result in Dunnigan's partial lipodystrophy in which CAD is a prominent feature, especially in females. These individuals suffer from an unusual form of insulin resistance, with morphologic abnormalities including hemifacial loss of subcutaneous adipose tissue, as well as hypertension, dyslipidemia and vascular disease. Patients in such families also have elevated CRP levels, and lower leptin and adiponectin levels[7]. A second aging disorder, Hutchison-Gilford Progeria Syndrome (HGPS), is the result of recurrent de novo mutations of a single nucleotide in exon 11 of the LMNA gene have been shown to cause the progeroid disorder)[8,9]. HGPS is characterized by extreme 'aging' in multiple tissues with most affected individuals dying from atherosclerotic vascular disease in their late teens. In many cases of HGPS there also appeared to be germ-line loss of the second LMNA allele, suggesting that somatic mutations at this locus might lead to more common forms of aging, possibly in a tissue-restricted manner. Given the extreme forms of vascular disease seen in these two laminopathies, and the strong association of the metabolic syndrome with CAD we explored the role of LMNA variation in premature coronary disease.
The second gene implicated in progeria is KLOTHO encoding a membrane protein of unknown function sharing homology with beta-glucosidases. Targeted deletion of the Klotho locus in mice results in reduced longevity, vascular disease, osteoporosis and chronic lung disease[10]. Recent work has demonstrated impaired angiogenesis and vasculogenesis in these Klotho-deficient mice[11]. A role for the ortholog of Klotho in human aging was suggested by the finding that a specific KLOTHO allele (KL-VS) which changes amino-acid sense is underrepresented in older age groups. This finding was reproduced in three ethnically distinct groups[12]. Individuals homozygous for the KL-VS allele were 2.6-fold less likely to survive to 65 years of age or greater. A role for the human KLOTHO gene in vascular disease also has been suggested by work demonstrating that the same KL-VS allele is associated with increased risk of occult atherosclerosis in a high-risk sample consisting of siblings of individuals with premature CAD. This effect of the KL-VS allele was evident even after adjustment for known risk factors [13]. We also explored the role of the KL-VS allele in our cohort of subjects with premature CAD.
Previous studies of these two aging genes largely have been confined to rare kindreds or cohorts with direct evidence of CAD. We tested the hypothesis that common variants at the LMNA and KLOTHO loci were associated with angiographically-defined premature CAD in a case-control association study.
Methods
Subjects
Between 1999 to 2003, we attempted to recruit serial patients with premature CAD presenting to the Massachusetts General Hospital were recruited. Inclusion criteria were; documented acute coronary syndrome (myocardial infarction or unstable angina) confirmed on coronary angiogram; and age ≤ 50 years for males or ≤ 55 years for females. Exclusion criteria were: logistic difficulties precluding enrollment, history of familial dyslipidemia, type I diabetes, endstage renal disease, a history of recent trauma, sepsis, or previous thoracic irradiation. All patients gave written informed consent prior to study enrollment. A structured interview and physical examination was performed during the initial hospitalization. This included a detailed medical history, careful documentation of cardiovascular risk factors, current and past medications and a comprehensive family history using an instrument which we have previously validated. Hypertension was defined as previous antihypertensive use or a documented untreated systolic blood pressure >140 mmHg or diastolic>90 mmHg. Hypercholesterolemia was defined as an LDL cholesterol >130 mg/dl. Diabetes mellitus was defined as the use of diet, oral hypoglycemic agents or insulin to control blood glucose. Smoking was defined as the regular use of tobacco at any stage in the previous decade. A blood sample for nucleic acid extraction was obtained at enrollment. In parallel, over the same time period a control population was recruited from outpatient referral clinics at Massachusetts General Hospital. These control subjects were free from symptoms suggestive of CAD or ECG abnormalities, and all had undergone transthoracic echocardiography to exclude the presence of subclinical structural heart disease.
Genetic analyses
LMNA genotyping
To generate representative haplotypes at the LMNA locus, we first assembled in silico 35,374 base pairs of reference sequence encompassing the entire coding region of the gene including sequence reported by Lin et al[14], consensus finished sequence and trace data from the region from the Human Genome Project[15]. Additional direct sequencing of the relevant genomic regions was performed where necessary. Sequence assembly and analysis was performed using the aid of Vector NTI version 8 (InforMax™). SNPs spanning the LMNA gene were selected from dbSNP [16] and the published literature. Selection was based on genomic coverage and where possible on polymorphism information content. In addition, all common (defined as those present in ≥ 5% of the population) SNPs changing amino acid sense were included.
A combination of TaqMan® based assays and direct sequencing was used to type the SNPs in an initial subset of the study cohort. Details of probes and primers are available in Table 5. Allelic discrimination using TaqMan® was performed using 5 ng of sample DNA in a 25 μL reaction containing 12.5 μL TaqMan® Universal PCR Mix (Applied Biosystems), 300 nM primers, 200 nM TaqMan® MGB probes (Applied Biosystems). Reaction conditions consisted of preincubation at 50°C for 2 minutes, 95°C for 10 minutes, then cycling for 40 cycles of 95°C, 15 seconds; 60°C, 1 minute. Amplifications were performed in an ABI Prism 7000 machine (Applied Biosystems) for continuous fluorescence monitoring. Direct cycle sequencing was used to type one series of closely linked SNPs as detailed in Table 5.
For each LMNA SNP the allele frequencies were defined and testing for Hardy Weinberg equilibrium was performed. Haplotypes at the LMNA locus were defined using the modified estimation-maximization algorithm implemented in the software package Haploview [17]. Following confirmation that specific SNPs did not segregate independently but were in linkage disequilibrium with each other, the methods of expectation-maximization-based haplotype frequency estimation and permutation-based hypothesis testing were performed as previously described [18].
KLOTHO genotyping
The KL-VS allele was typed using modifications of the published conditions[12]. Sample DNA was amplified by PCR (sense primer 5'-GCCAAAGTCTGGCATCTCTA-3'; antisense primer 5'-TTCCATGATGAACTTTTTGAGG-3') under the following conditions: 95°C for 2 minutes, followed by 35 cycles of 94°C for 30 seconds, 60°C for 30 seconds, and 72°C for 1 minute, followed by a 20 minute 72°C final extension. PCR products were then digested with MaeIII (Roche) at 55°C for 16 hours and electrophoretically separated on a 2% agarose gel. The KL-VS allele is characterized by diagnostic MaeIII restriction fragments of 265 and 185 bp respectively. Allele frequencies were determined by gene counting.
Statistical analyses
Continuous variables are presented as means ± SD. Baseline characteristics were compared using the Student's unpaired t test for continuous data and the Chi-square or Fisher's exact test for categorical data. Single-locus tests of association between either SNP allele frequencies or SNP genotype frequencies and case-control status were carried out using standard contingency Chi-square tests. Based on the method of Chapman and Nam and assuming an allele frequency of 0.20, our study design has 93% power to detect a difference in allele frequency of 1.8 times or more at α = 0.05 [19].
Results
Subjects
During the study period we enrolled 295 subjects with premature CAD and 145 controls. The premature CAD cohort, as expected, exhibited a higher proportion with diabetes mellitus, hypertension, and smoking (see Table 1). In addition, the premature CAD cohort was younger and more likely to be obese. Self-reported ethnicity was similar between the two cohorts.
LMNA genotypes
Two SNPs (rs3204564 and rs536857) that were not in Hardy Weinberg equilibrium were not included in subsequent analyses. Median spacing of the SNPs was 1,206 bp apart with a range of 146 bp to 7,063 bp (SNPs rs568035 and rs568036 are adjacent to each other). Haplotype analysis revealed the presence of 6 haplotypes, with the most frequent occuring at an overall frequency of 43.8% (see Figure 1 and Table 2). Haplotype allele frequencies did not differ significantly between patients and controls. Subsequently each individual SNP was tested independently for association with premature CAD. No significant difference in allele frequency for any SNP in the LMNA gene was observed between the control and patient groups (see Table 3). In analyses adjusting for age, BMI, gender, hypertension, diabetes mellitus, statin use, and smoking history we did not observe any significant associations.
KLOTHO genotypes
Previous studies have documented the heterozygote carrier frequency for the KL-VS allele to be between 20 to 30% and the homozygosity frequency to be between 1 to 4% [12]. The allele distribution is very similar in our patient and control cohorts, and no significant difference in allele frequencies was observed between these groups (see Table 4). The frequency of the KL-VS allele was 14.0% in patients with premature CAD compared to 18.2% in the control population (p = 0.09).
Discussion
We hypothesized that common variation at the LMNA or KLOTHO loci might result in typical forms of coronary disease in the absence of progeroid syndromes. We tested not only the hypothesis that alleles of LMNA or KLOTHO would be associated with premature CAD (a population enriched for inherited contributions), but also pre-specified that such premature vasculopathy might be associated with an excess of homozygosity at the LMNA locus. Our data suggest that, within the constraints of the current study, there is no association between common variation in the aging genes LMNA or KLOTHO and rigorously defined premature CAD. Further, there was no evidence of excess homozygosity at the LMNA locus, rendering a somatic "two-hit" mechanism at this locus less likely as a potential cause of CAD. These data contradict previous findings in smaller studies and emphasize the difficulties intrinsic to such genetic association studies.
Possible explanations for contradictory findings
Genetic association studies relating common or "complex" phenotypes in large patient cohorts may be the only method capable of unraveling small population-wide genetic effects, but these studies prove difficult to reproduce and are of limited utility in defining causation [20-24]. Several intrinsic limitations of genetic association approaches contribute to the disparity between our results and previous studies of the LMNA and KLOTHO loci including the low prior probability of any observed association, population stratification, and varying degrees of linkage disequilibrium with neighbouring genes [20,25]. One of the most difficult potential confounders is underlying etiologic heterogeneity, magnified by the relatively low resolution of many traditional clinical phenotypes[24]. Clearly, not all CAD is caused by the same mechanism, and there is variation in the biologic behaviour of the various syndromes, ranging from occult chronic ischemia through to ischemic sudden death. The heritable basis for each of these components in the "spectrum" of CAD may be quite distinct. This phenotypic heterogeneity remains an important issue despite our rigorous use of coronary angiographic diagnoses. Our control population was older and had a male predominance, both known cardiovascular risk factors. The patient population had a higher prevalence of diabetes mellitus, hypertension, and smoking. This is not unusual as most patients with CAD already have documented cardiovascular risk factors [26], and it is conceivable that many 'risk factors' are actually manifestations of subclinical forms of vascular disease. By excluding patients presenting with undifferentiated chronic stable CAD we hoped to minimize the phenotypic heterogeneity. Nevertheless, it remains possible that the progeroid genes we have studied are associated with a particular subset of CAD, but not with premature disease presenting as acute coronary syndromes.
Additional factors also may explain our findings. Our study would not have detected somatic mutations present only in the vessel walls. While this is relatively unlikely given the common progenitors shared by hematologic and endothelial lineages, mutations restricted to more differentiated cells may not be detectable. The association between the KLOTHO gene and premature atherosclerosis seen in the study by Arking, et al.[13] may reflect a chance relationship between reduced survival from other causes and a common trait, but by studying acute syndromes we may have selected a distinct subset of disease in which the effects of KLOTHO have been diluted.
Study limitations
Our study has several intrinsic limitations. The control population did not undergo invasive clinical testing to definitively exclude CAD, but nonetheless had extensive non-invasive evaluations including echocardiography. The genotype frequencies observed suggest that the current study is adequately powered to detect a risk ratio of 1.8 or more [22], but would be unlikely to detect smaller population wide effects or large effects from rare alleles. The contributions of rare alleles would be better addressed using a family based strategy[27].
Future genetic studies in CAD
Association studies remain controversial and our current study demonstrates some of the problems encountered with this approach. Despite rigorous phenotyping, detailed haplotyping and adequate power to detect a genetic effect of similar magnitude to that seen in previous studies of CAD, we did not see any significant differences in genetic architecture between our study and control populations. We have therefore demonstrated that genes associated with the progeroid syndrome are not likely to have a major effect on the development of premature atherosclerosis, despite a clear biological rationale.
In spite of the heterogeneity of CAD, insights from rare Mendelian variants have proven broadly applicable. The identification of new pathways through such familial forms of CAD will contribute to our understanding of all forms of vascular disease[2,3]. Linkage based family analyses are likely to yield more robust results than association studies and we believe increasingly will be used in future studies on the genetics of CAD. Some disorders undoubtedly result from more common ancient alleles, and understanding the basic haplotype structure of the human genome will facilitate their identification[28]. However, the genetic dissection of common conditions will require much more finely textured phenotypes than those traditionally employed in clinical medicine[24].
Conclusion
Common variants in aging syndrome genes previously implicated in CAD are not associated with rigorously defined premature acute CAD. This negative finding may reflect the specific phenotypes tested and highlights one of the major limitations of genetic association studies, the phenotypic heterogeneity of most 'common' diseases.
Abbreviations
CAD-coronary artery disease
LMNA-lamin A/C
HGPS-Hutchison Gilford Progeria Syndrome
SNP-single nucleotide polymorphism
Conflict of interest
The author(s) declare that they have no competing interests.
Authors' contributions
AFL was involved in the study design, carried out the molecular genetic studies, participated in the sequence analysis and drafted the manuscript. SK, PTE, CUC, SYS, BE and CAM, participated in the recruitment of subjects. CAM and COD participated in the design of the study and performed the statistical analysis. CAM and COD conceived of the study, and 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
AFL is supported by the National Medical Research Council of Singapore. PTE is supported by a K-23 award from the NIH(HL-71632).
Figures and Tables
Figure 1 Linkage disequilibrium plot demonstrates the inheritance of tagged SNPs as a single block within the LMNA gene.
Table 1 Baseline demography of study cohorts. Values are presented as number (percentage) unless otherwise indicated.
Baseline Characteristics PCAD Controls P-value
Number 295 145
Age at enrollment, years 47.4 ± 7.1 54.3 ± 10.8 <0.001
Male gender 199 (68.2) 115 (79.3)
BMI 30.2 ± 6.6 27.1 ± 4.4 <0.001
Cardiovascular risk factors
Diabetes mellitus 55 (18.8) 7 (4.86) <0.001
Hypertension 112 (38.2) 15 (10.4) <0.001
Hypercholesterolemia 104 (35.4) 30 (38.0) 0.67
Statin use 104 (35.4) 30 (38.0) 0.67
Smoking 113 (38.4) 8 (12.5) <0.001
Table 2 Distribution of major haplotype blocks in LMNA among patients with premature CAD and the control population.
Haplotype Frequency
Haplotype i ii iii iv v vi vii viii ix x xi Overall PCAD* Controls p-value
I C G G A T G C G A C G 0.44 0.43 0.45 0.55
II C G G A T G T G A C G 0.30 0.29 0.29 0.85
III T G G A T G C G A C A 0.13 0.13 0.14 0.82
IV T T A G C A C C G T G 0.05 0.05 0.07 0.25
V T G G A T G C G A C G 0.05 0.05 0.04 0.41
VI T G G G T A C C A T G 0.02 0.02 0.01 0.23
SNP ids; i-rs2485662;ii-rs517606;iii-rs593987;iv-rs528636;v-rs508641;vi-rs553016;vii-rs4641;viii-rs7339;ix-rs568036;x-rs568035;xi-rs6669212.
*PCAD: Group with premature coronary artery disease
2 SNPs (rs3204564 and rs536857) that were not in Hardy Weinberg equilibrium were excluded.
Table 3 Frequency of major SNP allele in LMNA among patients with premature CAD and the control population.
SNP
Major Allele
Freq in PCAD*
Freq in Controls
Minor Allele
p-value
rs955383 A 0.78 0.76 G 0.46
rs2485662 C 0.73 0.74 T 0.75
rs517606 G 0.94 0.93 T 0.55
rs593987 G 0.94 0.93 A 0.62
rs528636 A 0.92 0.92 G 0.92
rs508641 T 0.95 0.93 C 0.35
rs553016 G 0.93 0.92 A 0.89
rs4641 C 0.70 0.71 T 0.81
rs7339 G 0.93 0.92 C 0.90
rs568036 A 0.95 0.93 G 0.35
rs568035 C 0.92 0.92 T 1.00
rs6669212 G 0.86 0.86 A 1.00
* PCAD: Group with premature coronary artery disease
2 SNPs (rs3204564 and rs536857) that were not in Hardy Weinberg equilibrium were excluded.
Table 4 Genotype frequency of KLVS in cohort
Genotype
PCAD*
Frequency
Controls
Frequency
Chi-square
p-value
WW 216 73.5% 95 66.4% 2.321 0.13
WK 73 24.8% 44 30.8% 1.731 0.19
KK 5 1.7% 4 2.8% 0.574 0.45
Total 294 143
* PCAD: Group with premature coronary artery disease
Table 5 Analysis of SNPs in LMNA. Details of primers and probes used in TaqMan® analysis and PCR conditions for direct sequencing.
SNPs typed by TaqMan®
SNP ID Position Forward Primer (5'-3') Reverse Primer (5'-3') TaqMan® probe (VIC) TaqMan® probe (FAM)
rs955383 2,533 GGCCAGGAGTTTTAGACCAAATT TGAGTAGCTGAGACTATAGGCTATCATG ATAGCAAGgCTCCGTC TAGCAAGaCTCCGTCTC
rs2485662 3,971 ACTACCTTCTTTCTGGCTGAAACCAG GGGGAAGCAGGGCTGGG AGACCCAAtATTGGCT ACCCAAcATTGGCT
rs517606 9,827 CGAACTCCTAGGCTCAAGTAATCC GGGCTACAGACTAGAAAGAGACAGAGA CTGGGATgTATAGGCA CTGGGATtTATAGGCATGA
rs593987 16,890 TTGCTGTGCTGGTGCCTTT GAGTTGGCACTTGCAAATGTGA AGCCgGACTTCCT AGCCaGACTTCCTTG
rs528636 19,123 TCTAAATTCTGAGAGCCTCCTAGTACA GCAACTTAGATTCCGAGCTCCTT AGCAGCCaTTAGC AGCAGCCgTTAGC'
rs508641 20,642 GGAGAGAGAGGGAAAAGCATTCA TGACTCAGGGCTCAGGAACGT ACGGGGtAGAGCT ACGGGGcAGAGCT
rs553016 27,366 GCTTGGGACTCTGGGGAG CTTCCACACCAGGTCGGTA TCCCATCgCCACCCA TCCCATCaCCACCCA
rs4641 28,037 CGAGGATGAGGATGGAGATGA TCAGCGGCGGCTACCA TCCATCACCACCAcGTG CCATCACCACCAtGTG
rs7339 29,479 CAGAACTGCAGCATCATGTAATCTG GGGTTATTTTTCTTTGGCTTCAAG CCTGCACgTCATGG CCTGCACcTCATGG
rs3204564 30,139 AGGTGGAAGAAGGGAGAAGAAAG GCACCCCACTTGGCTTCA CCTAGCTTTAgACCCTGG CCTAGCTTTAaACCCTGG
rs536857 31,790 CACTGTGGGCTGGGGAACAC CCTCAGCCCTCCTCCTCAAGAG AGCAGGCaACGTT AGCAGGCgACGTT
SNPs typed by direct sequencing
SNP ID Position Primers(5'-3') PCR conditions
rs568036 30,669 These SNPs are contained in the amplicon amplified by the primers 5'-ACTGCATCCTCCTGCTCATT-3' and 5'-GGCTCCTACTTGGCCTAACA-3' 95°C for 2 min, followed by 35 cycles of 94°C for 30 s,60°C for 30 s, and 72°C for 90 s, followed by a 20 min 72°C final extension.
rs568035 30,670
rs6669212 30,816
==== Refs
Rader DJ Cohen J Hobbs HH Monogenic hypercholesterolemia: new insights in pathogenesis and treatment J Clin Invest 2003 111 1795 1803 12813012 10.1172/JCI200318925
Wang L Fan C Topol SE Topol EJ Wang Q Mutation of MEF2A in an inherited disorder with features of coronary artery disease Science 2003 302 1578 1581 14645853 10.1126/science.1088477
Wang Q Rao S Shen GQ Li L Moliterno DJ Newby LK Rogers WJ Cannata R Zirzow E Elston RC Topol EJ Premature myocardial infarction novel susceptibility locus on chromosome 1P34-36 identified by genomewide linkage analysis Am J Hum Genet 2004 74 262 271 14732905 10.1086/381560
Marenberg ME Risch N Berkman LF Floderus B de Faire U Genetic susceptibility to death from coronary heart disease in a study of twins N Engl J Med 1994 330 1041 1046 8127331 10.1056/NEJM199404143301503
Tegos TJ Kalodiki E Sabetai MM Nicolaides AN The genesis of atherosclerosis and risk factors: a review Angiology 2001 52 89 98 11228092
Burke B Stewart CL Life at the edge: the nuclear envelope and human disease Nat Rev Mol Cell Biol 2002 3 575 585 12154369 10.1038/nrm879
Hegele RA Kraw ME Ban MR Miskie BA Huff MW Cao H Elevated serum C-reactive protein and free fatty acids among nondiabetic carriers of missense mutations in the gene encoding lamin A/C (LMNA) with partial lipodystrophy Arterioscler Thromb Vasc Biol 2003 23 111 116 12524233 10.1161/01.ATV.0000047460.27435.B8
De Sandre-Giovannoli A Bernard R Cau P Navarro C Amiel J Boccaccio I Lyonnet S Stewart CL Munnich A Le Merrer M Levy N Lamin a truncation in Hutchinson-Gilford progeria Science 2003 300 2055 12702809 10.1126/science.1084125
Eriksson M Brown WT Gordon LB Glynn MW Singer J Scott L Erdos MR Robbins CM Moses TY Berglund P Dutra A Pak E Durkin S Csoka AB Boehnke M Glover TW Collins FS Recurrent de novo point mutations in lamin A cause Hutchinson-Gilford progeria syndrome Nature 2003 423 293 298 12714972 10.1038/nature01629
Kuro-o M Matsumura Y Aizawa H Kawaguchi H Suga T Utsugi T Ohyama Y Kurabayashi M Kaname T Kume E Iwasaki H Iida A Shiraki-Iida T Nishikawa S Nagai R Nabeshima YI Mutation of the mouse klotho gene leads to a syndrome resembling ageing Nature 1997 390 45 51 9363890 10.1038/36285
Shimada T Takeshita Y Murohara T Sasaki K Egami K Shintani S Katsuda Y Ikeda H Nabeshima Y Imaizumi T Angiogenesis and vasculogenesis are impaired in the precocious-aging klotho mouse Circulation 2004 110 1148 1155 15302783 10.1161/01.CIR.0000139854.74847.99
Arking DE Krebsova A Macek M SrMacek M JrArking A Mian IS Fried L Hamosh A Dey S McIntosh I Dietz HC Association of human aging with a functional variant of klotho Proc Natl Acad Sci U S A 2002 99 856 861 11792841 10.1073/pnas.022484299
Arking DE Becker DM Yanek LR Fallin D Judge DP Moy TF Becker LC Dietz HC KLOTHO allele status and the risk of early-onset occult coronary artery disease Am J Hum Genet 2003 72 1154 1161 12669274 10.1086/375035
Lin F Worman HJ Structural organization of the human gene encoding nuclear lamin A and nuclear lamin C J Biol Chem 1993 268 16321 16326 8344919
Lander ES Linton LM Birren B Nusbaum C Zody MC Baldwin J Devon K Dewar K Doyle M FitzHugh W Funke R Gage D Harris K Heaford A Howland J Kann L Lehoczky J LeVine R McEwan P McKernan K Meldrim J Mesirov JP Miranda C Morris W Naylor J Raymond C Rosetti M Santos R Sheridan A Sougnez C Stange-Thomann N Stojanovic N Subramanian A Wyman D Rogers J Sulston J Ainscough R Beck S Bentley D Burton J Clee C Carter N Coulson A Deadman R Deloukas P Dunham A Dunham I Durbin R French L Grafham D Gregory S Hubbard T Humphray S Hunt A Jones M Lloyd C McMurray A Matthews L Mercer S Milne S Mullikin JC Mungall A Plumb R Ross M Shownkeen R Sims S Waterston RH Wilson RK Hillier LW McPherson JD Marra MA Mardis ER Fulton LA Chinwalla AT Pepin KH Gish WR Chissoe SL Wendl MC Delehaunty KD Miner TL Delehaunty A Kramer JB Cook LL Fulton RS Johnson DL Minx PJ Clifton SW Hawkins T Branscomb E Predki P Richardson P Wenning S Slezak T Doggett N Cheng JF Olsen A Lucas S Elkin C Uberbacher E Frazier M Initial sequencing and analysis of the human genome Nature 2001 409 860 921 11237011 10.1038/35057062
dbSNP
Barrett JC Fry B Maller J Daly MJ Haploview: analysis and visualization of LD and haplotype maps Bioinformatics 2004
Fallin D Cohen A Essioux L Chumakov I Blumenfeld M Cohen D Schork NJ Genetic analysis of case/control data using estimated haplotype frequencies: application to APOE locus variation and Alzheimer's disease Genome Res 2001 11 143 151 11156623 10.1101/gr.148401
Chapman DG Nam JM Asymptotic power of chi square tests for linear trends in proportions Biometrics 1968 24 315 327 5683872
Cardon LR Bell JI Association study designs for complex diseases Nat Rev Genet 2001 2 91 99 11253062 10.1038/35052543
Hegele RA SNP judgments and freedom of association Arterioscler Thromb Vasc Biol 2002 22 1058 1061 12117714 10.1161/01.ATV.0000026801.56080.14
Botstein D Risch N Discovering genotypes underlying human phenotypes: past successes for mendelian disease, future approaches for complex disease Nat Genet 2003 33 228 237 12610532 10.1038/ng1090
Terwilliger JD Haghighi F Hiekkalinna TS Goring HH A bias-ed assessment of the use of SNPs in human complex traits Curr Opin Genet Dev 2002 12 726 734 12433588 10.1016/S0959-437X(02)00357-X
Colhoun HM McKeigue PM Davey Smith G Problems of reporting genetic associations with complex outcomes Lancet 2003 361 865 872 12642066 10.1016/S0140-6736(03)12715-8
Freedman ML Reich D Penney KL McDonald GJ Mignault AA Patterson N Gabriel SB Topol EJ Smoller JW Pato CN Pato MT Petryshen TL Kolonel LN Lander ES Sklar P Henderson B Hirschhorn JN Altshuler D Assessing the impact of population stratification on genetic association studies Nat Genet 2004 36 388 393 15052270 10.1038/ng1333
Greenland P Knoll MD Stamler J Neaton JD Dyer AR Garside DB Wilson PW Major risk factors as antecedents of fatal and nonfatal coronary heart disease events Jama 2003 290 891 897 12928465 10.1001/jama.290.7.891
Zhao LP Hsu L Davidov O Potter J Elston RC Prentice RL Population-based family study designs: an interdisciplinary research framework for genetic epidemiology Genet Epidemiol 1997 14 365 388 9271710 10.1002/(SICI)1098-2272(1997)14:4<365::AID-GEPI3>3.0.CO;2-2
Gabriel SB Schaffner SF Nguyen H Moore JM Roy J Blumenstiel B Higgins J DeFelice M Lochner A Faggart M Liu-Cordero SN Rotimi C Adeyemo A Cooper R Ward R Lander ES Daly MJ Altshuler D The structure of haplotype blocks in the human genome Science 2002 296 2225 2229 12029063 10.1126/science.1069424
|
16262891
|
PMC1289285
|
CC BY
|
2021-01-04 16:03:33
|
no
|
BMC Med Genet. 2005 Oct 31; 6:38
|
utf-8
|
BMC Med Genet
| 2,005 |
10.1186/1471-2350-6-38
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-1141626290510.1186/1471-2458-5-114Research ArticleBlood pressure patterns in rural, semi-urban and urban children in the Ashanti region of Ghana, West Africa Agyemang Charles [email protected] William K [email protected] Ellis [email protected] Marc A [email protected] Institute of Health Policy and Management, Erasmus Medical Center, Rotterdam, The Netherlands2 Institute for Medical Technology Assessment, Erasmus Medical Center, Rotterdam, The Netherlands3 School of Medical Sciences, Kwame Nkrumah University of Ghana, Kumasi, Ghana2005 1 11 2005 5 114 114 18 5 2005 1 11 2005 Copyright © 2005 Agyemang et al; licensee BioMed Central Ltd.2005Agyemang 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
High blood pressure, once rare, is rapidly becoming a major public health burden in sub-Saharan/Africa. It is unclear whether this is reflected in children. The main purpose of this study was to assess blood pressure patterns among rural, semi-urban, and urban children and to determine the association of blood pressure with locality and body mass index (BMI) in this sub-Saharan Africa setting.
Methods
We conducted a cross-sectional survey among school children aged 8–16 years in the Ashanti region of Ghana (West-Africa). There were 1277 children in the study (616 boys and 661 females). Of these 214 were from rural, 296 from semi-urban and 767 from urban settings.
Results
Blood pressure increased with increasing age in rural, semi-urban and urban areas, and in both boys and girls. The rural boys had a lower systolic and diastolic blood pressure than semi-urban boys (104.7/62.3 vs. 109.2/66.5; p < 0.001) and lower systolic blood pressure than urban boys (104.7 vs. 107.6; p < 0.01). Girls had a higher blood pressure than boys (109.1/66.7 vs. 107.5/63.8; p < 0.01). With the exception of a lower diastolic blood pressure amongst rural girls, no differences were found between rural girls (107.4/64.4) and semi-urban girls (108.0/66.1) and urban girls (109.8/67.5). In multiple linear regression analysis, locality and BMI were independently associated with blood pressure in both boys and girls.
Conclusion
These findings underscore the urgent need for public health measures to prevent increasing blood pressure and its sequelae from becoming another public health burden. More work on blood pressure in children in sub-Saharan African and other developing countries is needed to prevent high blood pressure from becoming a major burden in many of these countries.
==== Body
Background
High blood pressure has been identified as one of the leading causes of cardiovascular disease and premature mortality in the world [1]. In traditional African societies, high blood pressure, once rare, [2] is rapidly becoming a major public health burden [3-6]. The recent data show prevalence rates as high as 33% in some communities [3,4]. The increasing prevalence of hypertension is well reflected in the increasing stroke and cardiovascular disease morbidity and mortality [7-9].
In children, blood pressure tracking patterns confirm that persistent blood pressure elevation may be related to hypertension in adulthood [10,11]. The emerging data also suggest that primary hypertension is detectable and occurs commonly in the young [12]. In addition, the presence of elevated blood pressure in childhood has been linked with left ventricular hypertrophy [13]. As a result, in most western countries assessment and management of blood pressure in childhood is strongly recommended to promote improved cardiovascular health in adulthood [12]. Many epidemiological studies in various countries have been conducted to determine normal standard reference levels for age, sex, and body size [12,14]. However, in sub-Saharan African countries, blood pressure data on children and adolescents are very scarce. In Ghana for example, apart from the blood pressure profiles made in the 1970s on 5–12 year-olds in the Accra region, no other published data are available [14]. It is also unclear whether the recent rapid increases in blood pressure and prevalence of hypertension in adults [4,5,15] are reflected in children. Given the fast health transition towards non-communicable diseases and changes in lifestyle associated with urbanization [1,16], there is an urgent need for research on blood pressure in children so that appropriate cost-effective interventions can be introduced early in life to prevent the double burden of diseases in adulthood. The main purpose of this study was to assess blood pressure patterns among children in rural, semi-urban and urban settings in Ghanaian and to determine the association of blood pressure with locality and BMI in this Sub-Saharan African setting.
Methods
Study area
Ghana is located on West Africa's Gulf of Guinea, only a few degrees north of the Equator with a total area of 238,540 square kilometres. It borders Côte d'Ivoire to the west, Burkina Faso to the north, Togo to the east and the Gulf of Guinea to the south (Figure 1). According to the 2000 census, the total population was about 18,800,000 with annual growth rate of 2.4%. The literacy rate is 74.8%. The predominant religion is Christianity (69%), followed by Islam (15.6%), traditional religions (8.5%), and other religions (6.9%). The life expectancy in 2001 was 56.2 years for men and 59.3 years for women. The GNP per capita in 2002 was US $1,900. Data for this study were collected in the Ashanti region, a region found near the centre of the country. It covers an area of 24,390 square kilometres representing 10.2% of the land area of Ghana. The region produces most of the country's cocoa, minerals and timber.
Figure 1 Map of Ghana.
Study design
Data were collected in seven primary schools among healthy children between the ages of 8 to 16 years in the cool season (August–September 2004). Four schools in the urban regional capital (Kumasi) were randomly selected from the schools' lists, two schools in semi-urban and one school in rural setting. Rural refers here to a village without electricity and main water supply, where the main occupation is subsistence farming. The sub-urban villages have electricity and main water supply, and the main occupation is subsistence farming. The schools were visited prior to the data collection in order to obtain permission from the relevant school principals as well as from the children. Following the local rules, the village chiefs and elders were also contacted in advance to obtain their permission. Because only physical measurements were made, only verbal informed consent was sought from the children and their guardians before measurements were taken. Data collection took place during normal school hours. In each school, all children age 8–16 years were included except for one big school in the regional capital where every other class was included. None of the children in the schools refused to participate in this study. Height was measured without shoes with a measuring tape to the nearest 0.5 cm. Weight was measured to the nearest 0.1 kg after removal of shoes, jackets, heavier clothing and pocket contents (using an Electronic Korona Profimed scale, Germany). Body mass index (BMI) was calculated as weight (kg) divided by height (m2). Children were classified as being overweight according to the BMI-for-age cut-off points corresponding to an adult BMI of 25 kg/m2 [17]. Blood pressure and pulse were measured in the morning with a validated oscillometric automated digital blood pressure device (Omron M5-I monitor). Using appropriate cuff sizes, two readings with one-minute interval were taken on the right arm with the child in a seated position after at least five minutes rest. The mean of the two readings was used for analysis. The same trained staff made blood pressure measurements in all locations. In each school, prior to blood pressure measurements in children, all the teachers including the head-teacher had their blood pressure measured in front of the children to allay apprehension. The Committee on Human Research Publication and Ethics, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana approved the study protocol.
Data analysis
Age specific mean systolic and diastolic blood pressure levels were determined for rural, semi-urban and urban groups. The association between blood pressure and age was examined using linear regression analysis. Multiple linear regression analysis enabled age-adjusted comparisons of systolic and diastolic blood pressure levels to be made between gender and locality. Multiple linear regression analyses were also performed separately for boys and girls to assess the independent contribution of locality and BMI to systolic and diastolic blood pressure after adjustment for other factors associated with blood pressure in univariate analyses, including age and resting heart rate. All statistical analyses were performed using SPSS for Windows version 11.5 (SPSS Inc. Chicago, USA).
Results
Table 1 shows the characteristics of the study population, anthropometrics, and blood pressure levels. There were 1277 participants in the study (616 boys and 661 females). Of these, 214 were from rural, 296 from semi-urban and 767 from urban settings.
Table 1 Characteristics of the population, anthropometrics and blood pressure levels by gender and locality
Sex Boys Girls
Boys (n = 616) Girls (n = 661) Rural (n = 111) Semi-urban (n = 143) Urban (n = 362) Rural (n = 103) Semi-urban (n = 153) Urban (n = 405)
Age (y) 12.9 (2.3) 13.0 (2.1) 12.7 (0.2) 11.8 (0.2)** 13.4 (0.2)*** 12.6 (0.2) 12.5 (0.2) 13.2 (0.1)**
Height 147.2 (0.6) 148.8 (1.6) 141.5 (1.2) 137.8 (1.2)* 150.7 (0.7)*** 142.2 (1.1) 142.0 (1.0) 149.5 (0.6)***
Weight 38.8 (0.4) 41.2 (0.4)*** 34.8 (0.9) 32.1 (0.8)* 42.7 (0.5)*** 36.1 (1.0) 36.4 (0.8) 44.3 (0.5)***
BMI 17.7 (0.1) 18.8 (0.1)*** 17.1 (0.2) 16.5 (0.2)** 18.4 (0.2)*** 17.6 (0.3) 17.7 (0.3) 19.5 (0.2)***
Overweight % 3.1 6.4*** 0.0 1.4 4.6* 4.9 1.3 8.6
Pulse Rate 78.7 (0.5) 85.3 (0.5)*** 82.5 (1.1) 78.3 (1.2)* 77.7 (0.7)*** 88.9 (1.0) 79.5 (1.0)*** 86.7 (0.7)
Systolic BP# 107.5 (0.4) 109.1 (0.4)** 104.7 (1.1) 109.2 (1.0)*** 107.6 (0.6)** 107.4 (1.0) 108.0 (0.8) 109.8 (0.5)
Diastolic BP# 63.8 (0.4) 66.7 (0.3)*** 62.3 (0.7) 66.5 (0.7)*** 63.1 (0.4) 64.6 (0.9) 66.1 (0.7) 67.5 (0.5)**
Results are mean (SE); # Age adjusted. BMI = Body mass index; BP = Blood pressure; *P < 0.05, **P < 0.01, ***P < 0.001
Blood pressure levels
The systolic and diastolic blood pressure increased with increasing age in both boys and girls (figure 2a and 2b). This trend was also seen in rural, semi-urban and urban settings (Table 2). In a simple regression analysis, systolic and diastolic blood pressure increased by 3.0 mmHg and 1.0 mmHg per year respectively for rural children, and 2.0 mmHg and 0.6 mmHg per year respectively for semi-urban children, and 2.4 mmHg and 1.0 mmHg for urban children. All regression coefficients were statistically significant (p < 0.001). The age adjusted mean systolic and diastolic BP levels were significantly lower in boys than in girls.
Figure 2 a and b: Systolic and diastolic blood pressure by age and gender (results are shown as mean and SE).
Table 2 Mean (SE) Systolic and diastolic blood pressure (mm Hg) by age group and locality
Rural (n = 213) Semi-urban (n = 296) Urban (658)
Age (y) n Systolic BP Diastolic BP n Systolic BP Diastolic BP n Systolic BP Diastolic BP
8 18 86.0 (2.4) 51.0 (1.7) 32 96.3 (1.5) 59.8 (1.4) 28 97.9 (1.6) 59.9 (1.4)
9 13 94.1 (2.7) 57.8 (1.8) 24 103.5 (1.5) 65.8 (1.5) 25 98.6 (1.4) 59.8 (1.8)
10 20 98.4 (2.3) 62.2 (1.7) 27 101.4 (2.1) 64.0 (1.5) 29 97.4 (2.3) 58.4 (2.1)
11 31 100.4 (1.9) 61.5 (1.5) 34 104.3 (1.6) 66.0 (1.6) 26 103.2 (1.9) 66.8 (1.6)
12 29 103.5 (2.6) 64.4 (1.9) 34 108.7 (2.1) 66.4 (1.5) 68 102.6 (1.1) 60.7 (0.9)
13 30 105.6 (1.7) 63.5 (1.4) 32 109.4 (1.7) 68.4 (1.2) 140 105.7 (1.0) 61.5 (1.0)
14 33 107.5 (2.0) 62.2 (1.0) 48 111.9 (1.8) 67.6 (1.2) 197 109.6 (1.2) 63.9 (0.6)
15 22 114.5 (2.6) 65.9 (1.7) 37 111.2 (1.8) 65.3 (1.4) 104 115.9 (1.6) 66.4 (1.1)
16 18 115.7 (2.3) 68.4 (1.9) 28 113.4 (1.7) 66.9 (1.5) 41 115.1 (1.4) 66.4 (1.2)
P value <0.0001 <0.001 <0.0001 <0.0001 <0.0001 <0.0001
P value for linear regression
Rural versus semi-urban and urban
As table 1 shows, the age adjusted mean systolic and diastolic blood pressures were significantly lower in rural boys than they were in semi-urban boys. Compared with urban boys, rural boys had a significantly lower age adjusted mean systolic blood pressure but a similar diastolic blood pressure. Among girls, the age adjusted mean systolic and diastolic blood pressure levels were lower in rural girls compared with urban girls, although only the diastolic blood pressure difference was statistically significant. No significant differences were found between rural girls and semi-urban girls except for a lower resting heart rate in semi-urban girls.
Table 3 shows that boys living in a rural area had lower systolic and diastolic blood pressure levels than other boys while BMI was positively associated with systolic and diastolic blood pressure. Among girls, rural locality was independently associated with lower diastolic blood pressure while BMI was positively associated with both systolic and diastolic blood pressure.
Table 3 Multiple regression analysis of factors associated with systolic and diastolic blood pressure
Systolic blood pressure Diastolic blood pressure
Boys Beta SE p-value Beta SE p-value
Rural locality -3.53 1.16 0.003 -1.71 0.84 0.043
BMI 1.63 0.24 <0.0001 0.60 0.17 <0.0001
Age 1.64 0.24 <0.0001 0.53 0.17 0.002
Heart rate 0.13 0.04 0.001 0.06 0.03 0.016
R2 0.28 0.08
Girls
Rural locality -0.07 1.09 0.325 -2.06 0.93 0.026
BMI 1.03 0.17 <0.0001 0.69 0.14 <0.0001
Age 1.68 0.22 <0.0001 0.54 0.19 0.005
Heart rate 0.16 0.03 <0.0001 0.13 0.03 <0.0001
R2 0.25 0.12
BMI = Body Mass Index
Discussion
Key findings
Blood pressure increased with age in rural, semi-urban and urban areas, and in both boys and girls. Blood pressure levels were lower in the rural population than in the semi-urban and urban populations. Locality and BMI were independently associated with blood pressure in both boys and girls.
Discussion of key findings
The increase in blood pressure with age is consistent with previous reports in Ghana [4,5,14]. This was not only seen in semi-urban and urban settings but also in the rural setting. Less than half a century ago, ancestral African populations living traditional lives showed a lower mean blood pressure with little or no increase with increasing age, and low prevalence of hypertension [2]. However, in this present study, both systolic and diastolic blood pressure increased with increasing age (3 mmHg and 1.0 mmHg per year of age) in rural children. These findings seem to suggest that the protective effect against high blood pressure in rural settings in sub-Saharan Africa is fading. One possible explanation for these blood pressure trends in rural setting may be changes in lifestyles amongst these societies [18,19]. These findings do not bode well for the future, especially at a time when urbanization and westernization are proceeding at a faster rate [16]. In 2003 for example, stroke and cardiovascular disease were the 6th and the 7th most common causes of death in the Ashanti region of Ghana [8]. Increased blood pressure is a major contributing factor for stroke and cardiovascular diseases [7].
The lower blood pressure level found among rural school children is also consistent with the adult studies in Ghana [4,5] and other reports in sub-Saharan Africa [20,21]. For example, in Cappuccio et al's study, blood pressure was generally lower amongst rural dwellers than amongst semi-urban dwellers in Ghana [4]. In the present study, rural and urban differences in diastolic blood pressure still remained amongst both boys and girls even after adjustments for potential confounding factors. The reasons for these differences are unclear and further studies are needed to identify other factors that may contribute to the differences we observed. Boys had more favourable blood pressure profiles than girls. These gender differences in blood pressure patterns are consistent with earlier findings in the Greater Accra region of Ghana in both children [14] and adult [5] studies but contrast with findings of Cappuccio and colleagues in the Ashanti region of Ghana [4]. This is surprising given that Cappuccio et al's study was conducted in the same region as ours. Nonetheless, gender differences in blood pressure are generally inconsistent among African origin populations. In our recent report on blood pressure levels in ethnic minority children in the UK [22], blood pressure levels were generally more favorable for girls than for boys of African descent, while for adults, blood pressure patterns were more favourable for men than for women [23].
The strong and independent association between BMI and blood pressure is worrisome, especially for females in urban Ghana. Although the mechanisms by which BMI may lead to hypertension are poorly understood, it is now generally recognised that high BMI significantly increases the risk of hypertension. The impact of increasing BMI on high blood pressure has been clearly demonstrated in several populations [24,25]. Cooper and colleagues showed that the prevalence of high blood pressure increased as BMI increased across African descent populations [26]. In our study, both urban boys and girls had a higher BMI and were more likely to be overweight compared to their rural counterparts. Sinaiko and colleagues' prospective study showed that increases in weight and BMI in childhood were significantly associated with an increased risk of high blood pressure and other cardiovascular diseases in adulthood [27]. If the increasing frequency of overweight children is left unchecked while urbanization and westernization continue, it may lead to an increase in hypertension and other cardiovascular risks in future generations of Ghanaians [21]. In addition, the multiple regression models explained only 8% to 28% of the variance in systolic and diastolic blood pressures. This indicates that more work is needed to identify the other factors that contribute to an increase in blood pressure among children in this sub-Saharan African setting.
Limitations
Like many population-based surveys, our blood pressure level was based on an average of two measurements at a single visit. A more precise estimate of blood pressure level would be obtained by multiple measurements obtained during several visits. Also, evidence suggests that during puberty blood pressure increases more rapidly, with a significant gender difference in the age of onset [28]. In the present study, pubertal status was not assessed and this may affect our study results. One other possible limitation is the use of blood pressure measurement techniques such as an automated oscillometric device, as opposed to auscultatory mercury manometers in children. However, the Omron M5-I device we used in this study has been shown to be a valid instrument for use amongst children [29]. Despite these potential shortcomings, the findings from this study are consistent with those from adult studies [4,5,15]. Our results may well be representative of the Ashanti Region of Ghana as a whole and may be used to formulate local health policy. Also, because this study provides a comprehensive assessment of children in three different localities, its results should serve as a wake-up call for sub-Saharan Africa countries and other developing countries to step up cost-effective measures early in life to prevent the double burden of diseases in adulthood.
Conclusion and implications
This study shows an increase in blood pressure with age among these sub-Saharan African children in both rural and urban settings. This increase in blood pressure corresponds with the increasing prevalence of hypertension reported among adults. Blood pressure was positively associated with BMI in both boys and girls. These findings underscore the urgent need for public health measures to prevent high blood pressure and its sequelae from becoming another public health burden. In view of the scarcity of resources in Ghana as well as in many other developing nations, activities aimed at controlling increasing blood pressure in children have to compete with many other pressing health needs. Nevertheless, the long-term health problems that can result from increased blood pressure in children can be considerable. It is therefore important that measures be taken to reduce the risks of these problems and thereby optimize the health outcomes for future generations. Reducing the mean population blood pressure level by even as little as 2–3 mmHg could have a major impact in reducing associated morbidity and mortality [30]. A small pilot study of a nutritional education program in one village in Ghana resulted in the reduction in mean systolic and diastolic blood pressure by 6.4/4.5 mmHg within four weeks [31]. With careful implementation, such cost-effective measures may lead to an important reduction in blood pressure especially among city children, thereby sparing the next generation from high blood pressure related complications [31,32]. More work on blood pressure in children in sub-Saharan African and other developing countries is desperately needed, since high blood pressure is becoming a major public health issue in many countries.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
CA and MAB were responsible for study concept and design. CA and EO were responsible for data collection. CA, MAB and WKR were responsible for analysis and interpretation of data. CA drafted the manuscript and all were involved in critical revision of the manuscript. Statistical expertise was provided by WKR.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We are very grateful to Professor Raj Bhopal (University of Edinburgh) and Dr Karien Stronks (Academic Medical Center, Amsterdam), Drs Ank de Jonge (University of Nijmegen) for providing advice that helped to improve the earlier version of this paper. We thank Dr van Montfrans and Dr Lizzy Brewster for useful advice prior to data collection. We thank all the nurses, the school children and the teachers for making this study possible. We are also indebted to the two reviewers for providing comments that helped to improve the earlier version of this paper. This project was supported by the Health Research and Development Council of the Netherlands (ZonMw).
==== Refs
Murray CJ Lopez AD Mortality by cause for eight regions of the world: Global Burden of Disease Study Lancet 1997 349 i1269 1276 ii1436–1442 10.1016/S0140-6736(96)07493-4
Shaper AG Wright DH Kyobe J Blood pressure and body build in three nomadic tribes of northern Kenya East Afr Med J 1969 46 273 81 5345965
Edwards R Unwin N Mugusi F Whiting D Rashid S Kissima J Aspray TJ Alberti KG Hypertension prevalence and care in an urban and rural area of Tanzania J Hypertens 2000 18 145 52 10694181 10.1097/00004872-200018020-00003
Cappuccio FP Micah FB Emmett L Kerry SM Antwi S Martin-Peprah R Phillips RO Plange-Rhule J Eastwood JB Prevalence, detection, management, and control of hypertension in Ashanti, West Africa Hypertension 2004 43 1017 22 15037552 10.1161/01.HYP.0000126176.03319.d8
Amoah AG Hypertension in Ghana: a cross-sectional community prevalence study in greater Accra Ethn Dis 2003 13 310 5 12894954
van der Sande MA Milligan PJ Nyan OA Rowley JT Banya WA Ceesay SM Dolmans WM Thien T McAdam KP Walraven GE Blood pressure patterns and cardiovascular risk factors in rural and urban gambian communities J Hum Hypertens 2000 14 489 496 10962516 10.1038/sj.jhh.1001050
Walker RW McLarty DG Kitange HM Whiting D Masuki G Mtasiwa DM Machibya H Unwin N Alberti KG Stroke mortality in urban and rural Tanzania. Adult Morbidity and Mortality Project Lancet 2000 355 1684 7 10905244 10.1016/S0140-6736(00)02240-6
Thorogood M Connor MD Lewando-Hundt G Tollman S Ngoma B SASPI Project Team Secondary prevention of stroke – results from the Southern Africa StrokePrevention Initiative (SASPI) study Bull World Health Organ 2004 82 503 8 15500283
Connor MD Thorogood M Casserly B Dobson C Warlow CP SASPI Project Team Prevalence of stroke survivors in rural South Africa: results from the Southern Africa Stroke Prevention Initiative (SASPI) Agincourt field site Stroke 2004 35 627 32 14963282 10.1161/01.STR.0000117096.61838.C7
Bao W Threefoot SA Srinivasan SR Berenson GS Essential hypertension predicted by tracking of elevated blood pressure from childhood to adulthood: the Bogalusa Heart Study Am J Hypertens 1995 8 657 65 7546488 10.1016/0895-7061(95)00116-7
Lauer RM Clarke WR Childhood risk factors for high adult blood pressure: the Muscatine Study Pediatrics 1989 84 633 41 2780125
National High Blood Pressure Education Program Working Group on High Blood Pressure in Children and Adolescents The fourth report on the diagnosis, evaluation, and treatment of high blood pressure in children and adolescents Pediatrics 2004 114 555 76 15286277 10.1542/peds.114.2.S2.555
Daniels SR Loggie JM Khoury P Kimball TR Left ventricular geometry and severe left ventricular hypertrophy in children and adolescents with essential hypertension Circulation 1998 97 1907 11 9609083
Blankson JM Larbi EB Pobee JO Blood pressure levels of African children J Chronic Dis 1977 30 735 43 591602 10.1016/0021-9681(77)90002-9
Agyemang C Bruijnzeels E Owusu-Dabo W Factors associate with hypertension awareness, treatment and control in Ghana, West Africa J Hum Hypertens 1 5 2005; Aug 25
Cappuccio FP Epidemiologic transition, migration and cardiovascular disease Int J Epidemiol 2004 33 387 8 15082646 10.1093/ije/dyh091
Cole TJ Bellizzi MC Flegal KM Dietz WH Establishing a standard definition for child overweight and obesity worldwide: international survey BMJ 2000 320 1240 1243 10797032 10.1136/bmj.320.7244.1240
Kaufman JS Owoaje EE James SA Rotimi CN Cooper RS Determinants of hypertension in West Africa: contribution of anthropometric and dietary factors to urbanrural and socioeconomic gradients Am J Epidemiol 1996 143 1203 18 8651219
Whelton SP Chin A Xin X He J Effect of aerobic exercise on blood pressure: a metaanalysis of randomized, controlled trials Ann Intern Med 2002 136 493 503 11926784
Obika LF Adedoyin MA Olowoyeye JO Pattern of paediatric blood pressure in rural, semi-urban and urban communities in Ilorin, Nigeria Afr J Med Sci 1995 24 371 7
Proctor MH Moore LL Singer MR Hood MY Nguyen US Ellison RC Risk profiles for non-communicable diseases in rural and urban schoolchildren in the Republic of Cameroon Ethn Dis 1996 6 235 43 9086313
Agyemang C Bhopal R Bruijnzeels M Do variations in blood pressures of South Asian, African and Chinese descent children reflect those of the adult populations in the UK? A review of cross-sectional data J Hum Hypertens 2004 18 229 37 15037871 10.1038/sj.jhh.1001658
Agyemang C Bhopal R Is the blood pressure of people from African origin adults in the UK higher or lower than that in European origin white people? A review of cross-sectional data J Hum Hypertens 2003 17 523 34 12874609 10.1038/sj.jhh.1001586
Kaufman J Barkey N Hypertension in Africa: an overview of prevalence rates and causal risk factors Ethn Dis 1993 3 S83 101 8087029
Johnson JL Heineman EF Heiss G Hames CG Tyroler HA Cardiovascular disease risk factors and mortality among black women and white women aged 40–64 years in Evans County, Georgia Am J Epidemiol 1986 123 209 20 3946371
Cooper R Rotimi C Ataman S McGee D Osotimehin B Kadiri S Muna W Kingue S Fraser H Forrester T Bennett F Wilks R The prevalence of hypertension in seven populations of West African origin Am J Public Health 1997 87 160 8 9103091
Sinaiko AR Donahue RP Jacobs DR JrPrineas RJ Relation of weight and rate of increase in weight during childhood and adolescence to body size, blood pressure, fasting insulin, and lipids in young adults. The Minneapolis Children's Blood Pressure Study Circulation 1999 99 1471 6 10086972
El Assaad MA Topouchian JA Asmar RG Evaluation of two devices for self-measurement of blood pressure according to the international protocol: the Omron M5-I and the Omron 705IT Blood Press Monit 2003 8 127 33 12900590
Voors AW Foster TA Frerichs RR Webber LS Berenson GS Studies of blood pressures in children, ages 5–14 years, in a total biracial community: the Bogalusa heart study Circulation 1976 54 319 327 939029
Whelton PK Epidemiology of hypertension Lancet 1994 344 101 6 7912348 10.1016/S0140-6736(94)91285-8
Cappuccio FP Plange-Rhule J Phillips RO Eastwood JB Prevention of hypertension and stroke in Africa Lancet 2000 356 677 8 10968456 10.1016/S0140-6736(05)73821-6
Kerry SM Emmett L Micah FB Martin-Peprah R Antwi S Phillips RO Plange-Rhule J Eastwood JB Cappuccio FP Rural and semi-urban differences in salt intake, and its dietary sources, in Ashanti, West Africa Ethn Dis 2005 15 33 39 15720047
|
16262905
|
PMC1289286
|
CC BY
|
2021-01-04 16:28:59
|
no
|
BMC Public Health. 2005 Nov 1; 5:114
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-114
|
oa_comm
|
==== Front
Environ HealthEnvironmental Health1476-069XBioMed Central London 1476-069X-4-241626289210.1186/1476-069X-4-24ResearchAsbestosis in an asbestos composite mill at Mumbai: A prevalence study Murlidhar V [email protected] Vijay [email protected] Department of Surgery, LTM Medical College, 1st Floor, College building, Sion, Mumbai 400 022, India2 Occupational Health and Safety Centre, Gokuldas Pasta Road, Neelkant Apts, Dadar (E), Mumbai 400 014, India2005 31 10 2005 4 24 24 26 5 2005 31 10 2005 Copyright © 2005 Murlidhar and Kanhere; licensee BioMed Central Ltd.2005Murlidhar and Kanhere; 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
Of an estimated 100000 workers exposed to asbestos in India, less than 30 have been compensated. The reasons for such a small number are: refusal by management sponsored studies to grant medical certifications to workers suffering from occupational diseases, lack of training for doctors in diagnosis of occupational lung diseases, deliberate misdiagnosis by doctors of asbestosis as either chronic bronchitis or tuberculosis and the inherent class bias of middle class doctors against workers. The aim of the study was to identify workers suffering from Asbestosis (parenchymal and pleural non-malignant disease) among the permanent workers of the Hindustan Composites Factory and assess their disability and medically certify them, whereupon they could avail of their basic rights to obtain compensation and proper treatment.
Methods
The study was conducted by the Occupational Health and Safety Centre and the Workers' Union. Asbestosis was diagnosed if they had an occupational history of asbestos exposure for at least 15 years and showed typical radiographic findings.
Results
Of 232 workers in the factory, 181 participated in the survey. 22% of them had asbestosis. All the asbestos affected workers had at least 20 years of exposure. 7% had rhonchi, 34% had late basal inspiratory rates, 82% had more than 80% of Forced Expiratory Volume in the first second (FEV1)/Forced Vital capacity (FVC) ratio and 66% had FVC less than 80% of the predicted value. On radiology 7% had only pleural disease, 10% had both pleural and parenchymal disease and 82% had only parenchymal disease. The association of pleural disease with chest pain was statistically significant.
Conclusion
We found the prevalence of asbestosis among exposed workers to be less than that anticipated for the number of years of exposure due to "Healthy Worker Effect". We suggest that all affected asbestos workers (including those who have been forced to leave) in India be medically certified and compensated. We also recommend better control of asbestos use in India. We also implore the management to provide all information about the work process and its hazards, conduct medical checkups as mandated by law and give the medical records to the workers.
==== Body
Background
There are an estimated 100000 people exposed to asbestos at work in India [1-3]. Many Indian studies have been conducted to estimate the prevalence of Asbestosis in India [4-6]. But less than 30 workers have been compensated even though legislations for compensations in the form of the Workmen's Compensation Act (WC Act) and the Employees State Insurance Act (ESI Act) were enacted in 1923 and 1948 respectively [7]. The reasons for such a small number are: refusal by management sponsored studies to grant medical certifications to workers suffering from occupational diseases, lack of training for doctors in diagnosis of occupational lung diseases, deliberate misdiagnosis by doctors of Asbestosis as either chronic bronchitis or tuberculosis and the inherent class bias of middle class doctors against workers [7-9]. The aim of the study was to identify from among the permanent workers of the Hindustan Composites Factory, those suffering from Asbestosis (parenchymal and pleural non-malignant disease) and assess their disability so that they could be medically certified and avail of their basic rights to obtain compensation and proper treatment.
The case study, Hindustan Composites, then called Asbestos Magnesia and Friction Material (AM and FM) was established in Sewri, Mumbai in the year 1949. In 1956, with a change of name to Hindustan Ferodo, it was shifted to its present site in Ghatkopar, Mumbai. In 1990, the ownership changed and the company is now called Hindustan Composites. In 2003, the management declared a lockout that lasted for seven and a half months. The workers' union struggled during this time and after an agreement to reduce the workforce, production began in June 2004. The company started production again with 232 permanent workers and around 100 contract (temporary) workers. 110 workers were coerced to resign under the so-called Voluntary Retirement Scheme (VRS).
The workers union of the Hindustan Composites Asbestos Factory, the Krantikari Kamgar Union, approached the Occupational Health and Safety Centre (OHSC), Mumbai in March 2004 to study the prevalence of Asbestosis amongst its workers and get medical certification of affected workers.
The Occupational Health and Safety Centre (OHSC) [10], established in 1988, is a voluntary organization working with workers and unions. The OHSC has in the past conducted studies on occupational diseases and has helped more than thousand workers to claim compensation for Byssinosis, Noise-induced hearing loss, Occupational asthma, Acid burns, Radiation injuries and other occupational injuries [10-14].
Work Process
The following section describes the factory processes that lead to asbestos production. The management refused permission to conduct the survey inside the factory premises and did not cooperate in the survey conducted outside the factory's gates. They had not given the workers any information regarding the health effects of asbestos and other hazardous materials used in the production process. The workers were not involved in any of the decision-making processes involved in procuring the raw material and in the work process. Hence, the following information regarding the production process is from the knowledge that the workers have gathered during their years of work.
There are four main departments in the factory: Asbestos Textiles, Compressed Asbestos Fiber (CAF), Brake and Clutch Lining (BCL) and Goods Receiving Section (GRS). There are other allied departments such as Maintenance, Security etc. Asbestos Textiles, Compressed Asbestos Fiber (CAF), Brake and Clutch Lining (BCL) are the "dusty sections" where the exposure to asbestos dust is the highest.
White asbestos is brought to the factory in bags. It is in the form of minute fibers that appears like white powder or lumps. It enters the process in the factory at two stage points.
At one of the stage points, Asbestos is mixed with a viscose material and then sent to the Carding Section. In the Carding Section, slivers or wicks are produced. A sliver is a loose, thin continuous fiber ready to be drawn and twisted. A wick is a piece of cord or tape. These are transferred to the Frame Section. Here, yarn is produced by plying/winding of 2, 3 or 4 fibers together. From the Frame Section the material is sent to the Plating Section to produce rope, to the Weaving Section to make cloth or rolls, and to the BCL department. In the BCL department, the above material is mixed with varnish, heated and cooled. Then, surface grinding takes place.
At the other stage point, Asbestos fiber is mixed with resins or rubber and is processed in a mixer with spikes. The rotating mixer is heated. The soft material produced is sent between two rolls/calendars. One roll is heated and the other is chilled. Asbestos sheets are produced when this material comes out of the rolls.
Methods
Worker-activists and one of the authors, V. Murlidhar (hereinafter VM), discussed the modalities and necessary steps in diagnosis of Asbestosis. The study was done from 10th of November 2004 to 13th of November 2004 at the gates of the factory. Workers were informed and came as per their convenience for the check up. The response from the workers was very good and out of the 232 permanent workers, 181 attended the camp. The contract workers have an insecure tenure and were apprehensive. As a result, they did not present for the examination.
One of the authors (VM) was present when the questionnaire was filled for every worker. The questionnaire prepared by us was similar to that used in previous published studies on surveys of occupational lung diseases [12]. A detailed occupational history of exposure to asbestos was taken as per standard guidelines [15]. Symptoms of dyspnoea, chest pain and chest tightness were recorded. We physically examined workers for clubbing, end-inspiratory basal crackles and rhonchi, which have prognostic significance [16,17]. Smoking score was defined as number of cigarettes smoked per day times the number of smoking years.
Lung function tests (PFTs) were performed, as in previous studies of occupational lung disease by the OHSC [12]. The lung function tests were carried out using a Wright ventilometer (VM1) which gives digital readings for Forced Vital capacity (FVC), Forced Expired volume in the first second (FEV1), their ratio and Peak flow rate (PEF).
Three readings of PFT were taken and the consistent value chosen. Prediction equations for Indian subjects were used to calculate the expected Forced Expiratory Vital capacity in the First Second (FEV1) and Forced Vital capacity (FVC values) [18]. The equations used were as follows:
Females
FEV1: 0.0274 H – 0.0103 A – 1.995
FVC: 0.0370 H – 0.0070 – 3.197
Males
FEV1: 0.0396 H – 0.0212 A – 3.130
FVC: 0.0603 H – 0.0136 – 4.488
H= Height in cms
A= Age in years
Chest radiographs were taken as per prescribed specifications and classified with due regard to quality as per the International Classification of Radiographs of Pneumoconiosis (ILO classification), [19,20] also see additional file 1: Chest radiograph classification. The radiographs were studied and classified by one of the authors (VM). The ILO classification profusion score of 1/1 and greater was considered positive for diagnosis of Asbestosis, which is the 1986 guideline laid down by the American Thoracic Society [19].
High Resolution Computerized Tomography (HRCT) was done in only one patient who was suspected to have lung cancer. The other suspected patient of pleural cancer opted not to be investigated further.
The data was fed into EPI6 for MSDOS statistical software. The statistical test used when appropriate was the Chi Square test.
Diagnostic Criteria for Asbestosis
Asbestosis was diagnosed if they had an occupational history of asbestos exposure for at least 15 years and showed typical radiographic findings.
Clinical examination findings were used only as prognostic indicator for treatment. Pulmonary function testing was used for impairment assessment as per standard criteria, [13,21] also see additional file 2: Pulmonary Impairment assessment guidelines.
Results
Asbestosis Distribution
The workers of Textile, BCL and CAF departments (which have higher dust exposure) constituted 81% of the workers examined, while workers in Maintenance and other ancillary department (which have lesser dust exposure) accounted for the remaining 19% (Table I).
Table 1 Distribution of workers among departments: n = 181
Department Dusty Sections Other Sections Total
Textile BCL CAF Maintenance Other
Number of Workers 93 18 37 11 22 181
Number of Workers affected by Asbestosis 21(23%)a 7(39%)a 6(16%)a 3(27%)a 4(18%)a 41(23%)a
a Number of workers affected by Asbestosis (% age of total number of workers in the department)
There were 41(23%) Asbestosis cases among the workers. Of them 34(82%) were from the dusty sections (Table I). Only 7(18%) were from the other departments but it was not statistically significant.
All workers had at least 20 years of exposure, 51% had an exposure of more than 30 years. All workers were above 40 years of age, 52% of the workers were older than 50 years of age. The mean age was 54 years.
Symptoms and Signs
Only 4% of the total workers reported history suggestive of chronic bronchitis. Of the total workers, 77% were non-smokers, 13% had a smoking score of less than 100 and only 8% had a smoking score of more than 100. 20 percent of the Asbestosis-affected workers gave history of smoking. Among the non-smokers, 33% (11) had FEV1 less than 80% of the predicted value.
Half of the workers reported mild dyspnoea grade 1 and only 3% reported grade 2 dyspnoea. Of these 55% had FVC less than 80% of the predicted value.
Mild chest pain or tight chest was reported by 50% of the workers. The chest discomfort was not continuous in anyone. The chest pain symptoms were reported for less than 5 years in 39%, and only 11% had the symptom for more than 5 years. Only 2% of the total workers had pulmonary tuberculosis, which was treated fully.
On clinical examination of the Asbestosis cases, none had clubbing, only 7% had rhonchi which were occasional and 34% had late inspiratory basal crackles. Two of the Asbestosis affected workers had pleural/lung tumors in addition to having Parenchymal Asbestosis. One of them declined further investigation, the other case was investigated by HRCT and lung biopsy, and lung cancer (T3, N2, and M0) was proved. Another worker died of cardio pulmonary failure within weeks of the completion of the study and certification as Asbestosis. The average disability percent of the affected workers was 32%.
Lung Function Tests (FEV1, FVC, and FEV1/FVC)
Of the 181 workers tested, 62% had abnormal FVC (less than 80% of predicted), 15% had abnormal FEV1 (less than 80% of predicted), 79% had more than 80% of FEV1/FVC.
Of the 41 Asbestosis affected workers, 66% had FVC less than 80% of predicted, 37% had FEV1 less than 80% of predicted and 82% had more than 80% of FEV1/FVC. (Table II)
Table 2 Lung Function in the Asbestosis cases. n = 41
Pulmonary Function Percentage Predicted FVC Percentage Predicted FEV1 FEV1/FVC %
< 80% > 80% <80 % > 80% < 80% 80–90% > 90%
Number of workers 27(5)a 14(2)a 15(3)a 26(4)a 7(2)a 19(2)a 15(3)a
Percentage 66% 34% 37% 63% 18% 46% 36%
a Number of workers checked (Number of Smokers)
Chest X-ray
Pleural disease was identified in 17% of the affected workers, parenchymal disease in 92% of the affected workers (82% had only parenchymal disease) and 7% had only pleural disease. (Table III)
Table 3 Radiological findings of Asbestosis cases: n = 41
Pleural Disease Parenchymal disease
All Pleural Only Pleural Pleural and Parenchymal All Parenchymal s/sa s/ta t/ta
7(17%)b 3(7%)b 4(10%)b 38 (92%)b 27(65%)b 8(19%)b 3(7%)b
a- s/s, s/t, t/t are classification of opacities as per ILO criteria [20] b- Number of cases (%age)
Of the 7 cases with pleural disease 5 reported chest pain where as only 1(one) with parenchymal disease reported chest pain. This is statistically significant (p = 0.05).
Discussion
It is estimated that 6000 workers are directly exposed and nearly 100000 workers indirectly exposed to asbestos [1]. The prevalence rate of Asbestosis in our study was 23%, which is less than the expected prevalence among workers exposed to asbestos for more than 20 years [16,22]. Many studies reported a prevalence of above 70% among workers exposed to asbestos for more than 20 years [16,19,22].
The reasons for the lower prevalence found in our study are many. The primary reason is the "healthy worker effect". Many affected workers have been forced to leave the company or to take voluntary retirement (VRS). Some may even have died due to the disease. Hence, the workers remaining in the factory are relatively healthy workers. As in most industries, workers who are casual or temporary do the hazardous jobs. However, since their livelihood is at stake, they would not come for the survey. Hence, exclusion of the casual workers in whom the prevalence rate would probably be higher is another reason for the lower prevalence in our study. A chest film clearly showing the characteristic signs of Asbestosis in the presence of a compatible history of exposure is adequate for the diagnosis of the disease. Further imaging procedures like an HRCT are not required [19]. Nevertheless, a high resolution CT (HRCT) might have picked up more cases of parenchymal Asbestosis [23,24]. Financial constraints limited the number of workers who could undergo HRCT. It was also beyond our capacity to identify workers who had left their jobs or had retired. These could have given us a truer prevalence rate.
Nearly half of the workers reported dyspnoea and 55% of them had FVC less than 80% of the predicted value. 11–17% reduction in the FVC has been reported in asbestos workers who report dyspnoea [25]. Even report of mild dyspnoea is important, as has been indicated in another study where 33% of asbestos workers reporting mild dyspnoea had diminished FVC [26]. In our study 34% had basal rates, which increases the risk of asbestos-related mortality [16,19].
In our study parenchymal changes are more common that pleural changes. This is similar to another Indian study on asbestos prevalence where they found that parenchymal changes are six times more common than pleural changes in Asbestosis [4]. In our study, there was a statistically significant association of chest pain with pleural changes on radiographs. This is consistent with another study wherein nearly half of the workers with pleural disease reported chest pain [26]. Rapidly progressive or severe chest pain should raise clinical suspicion of malignancy [19].
Physical findings in Asbestosis include basilar rates, often characterized by end-inspiratory crackles (rates). In some cases of advanced Asbestosis, finger clubbing may be present [16,19]. Physical findings of crackles, clubbing, or cyanosis are associated with increased risk for asbestos-related mortality [16,17]. In our study nearly 40% of Asbestosis workers had either rates or rhonchi. However, physical findings have low sensitivity and hence limited clinical utility [19].
Tuberculosis (TB) was noted in 2% of cases. It is important to clearly diagnose TB, since Asbestosis in India has been misreported as TB in the past [7]. We also found two cases of pulmonary/pleural tumors. This is important because not a single case of occupational lung cancer has been compensated in India, to date.
The lung function impairments in Asbestosis affected workers in our study were typical: majority of the workers 66% had a restrictive pattern with diminished FVC, and 37% had an obstructive pattern of decreased FEV1. As with other interstitial lung diseases, the classic finding of Asbestosis is a restrictive impairment as in our study. Mixed restrictive and obstructive impairment is also frequently seen; in contrast, isolated obstructive impairment is unusual [19]. Most workers affected by Asbestosis (82%) also had FEV1/FVC ratio above 80%. This increase in ratio has been noted in the past [27]. Demonstration of functional impairment is not required for the diagnosis of a nonmalignant asbestos-related disease, but where present should be documented as part of the complete evaluation [19]. It contributes to the diagnosis in defining the activity of the diseases and the resulting impairment can be quantified for compensation purposes [19,21].
It is reported that nearly 50% of Asbestosis workers have a FVC less than 80% of the predicted value. Hence, we tabulated the ventilatory impairments using this criteria [19]. Another reason for using the 80% of expected FVC and FEV1 criteria for tabulation lay in the use of this cut-off limit by the assessments of impairments due to pulmonary disease [21].
It is never enough to emphasize that institutional deficiencies within the medical system and the industry management are the primary contributors to occupational diseases like Asbestosis. Disability assessment is an important responsibility of the physician, yet, it is not routinely taught in medical schools [11]. Despite compensation being a legal right in India, the affected worker cannot hope to gain compensation without certifications of the resultant impairment as an occupational disease [11]. As per Indian Law, it is mandatory for the management to give detailed information about the work processes and the effects of the hazardous processes on one's health to workers in their local language. This was not done by the management. Such an act is indubitably unethical [8]. Forcing the management to pay compensation to workers will induce the owners to employ the safety measures and precautions that are mandated by Indian Law.
Conclusion
We found the prevalence of Asbestosis among exposed workers to be less than that expected for years of exposure. This is mainly due to the "healthy worker effect" (i.e., most of the affected workers in past years have either died or have been removed from employment due to lockouts). Some have opted for early retirement coerced by the management.
There are less than 30 cases of Asbestosis compensated in India among the 100000 exposed workers. Many must have died of the disease or of lung or pleura cancer. Workers concerned with asbestos are to be medically checked by the management every year while continuing in such a job and after he has ceased to work in such a job. This is a specific responsibility of the occupier of any factory having any hazardous process. All the workers who have left a job involving asbestos, under VRS or otherwise, need to be medically checked, once a year at the very least. This is their legal right.
The diagnosis of Asbestosis, in particular, imposes a duty on the doctor to inform the patient that he or she has a disease that is work-related. The duty extends to reporting the disease and to inform the patient that he or she may have legal or adjudication options for compensation. The role of the physician in this compensation process includes performing an objective evaluation of impairment consistent with the rules of the specific compensation system [10,11,13-15].
We recommend better control of asbestos use in India. We also recommend that the management give all information regarding the hazardous processes and their medical records after conducting the mandatory annual medical checkups as is mandated by law.
List of abbreviations
AM and FM: Asbestos Magnesia and Friction Material
BCL: Brake and Clutch Lining
CAF: Compressed Asbestos Fiber
GRS: Goods Receiving Section
CAF: Compressed Asbestos Fiber
ESI Act: Employees State Insurance Act
FVC: Forced Vital capacity
FEV1: Forced Expiratory Volume in the first second
HRCT: High Resolution Computerized Tomography
ILO classification: International Classification of Radiographs of Pneumoconiosis
OHSC: Occupational Health and Safety Centre
PFTs: Lung function tests
PEF: Peak flow rate
TB: Tuberculosis
VRS: Voluntary Retirement Scheme
WC Act: Workmen's Compensation Act
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
VM conceived the study, supervised all aspects of its implementation, and led the writing of the manuscript. VK assisted with the study and contributed to the design and analysis of all study components. All authors helped to interpret findings and review drafts of the manuscript.
Supplementary Material
Additional File 1
Chest radiograph classification format as per ILO classification. Description of Data: This table is the specified format for recording chest radiographs as per the ILO guidelines for classification of pneumoconiosis.
Click here for file
Additional File 2
Impairment Assessment guidelines used for calculating disability of affected workers. Description of Data: These are the guidelines used for assessment of respiratory impairment.
Click here for file
Acknowledgements
The authors are thankful to the team members of OHSC- P.S. Malvadkar and Ms. Surekha Sakpal for assistance in conducting lung function tests and to the office bearers and activists of the Krantikari Kamgar Union and the workers of the Hindustan Composites Company for their full cooperation in this study. We are thankful to Dr. Anirudha Badade for facilitating radiography of the workers. We are also thankful to Dr Veena Murlidhar, Vasanthi Venkatesh and Dr Nobojit Roy for helping with proof reading of the article
==== Refs
Sharma DC Indian groups demand ban on asbestos Lancet 2002 359 1414 11978351 10.1016/S0140-6736(02)08400-3
International Commission on Occupational Health Asbestos and Public Health - the International Dimension
Joshi TK Gupta RK Asbestos-related morbidity in India International Journal of Occupational and Environmental Health 2003 9 249 253 12967161
Dave SK Bhagia LJ Mazumdar PK Patel GC Kulkarni PK Kashyap SK The correlation of chest radiograph and pulmonary function tests in asbestos miners and millers Indian Journal of Chest Diseases and Allied Sciences 1996 38 81 89 8822641
Dave SK Ghodasara NB Mohanrao N Patel GC Patel BD The relation of exposure to asbestos and smoking habit with pulmonary function tests and chest radiograph Indian Journal of Public Health 1997 41 16 24 9567522
Mukherjee AK Rajmohan HR Dave SK Rajan BK Kakde Y Rao SR An environmental survey in chrysotile asbestos milling processes in India American Journal for Industrial Medicine 1992 22 543 551
Mudur G Asbestos poisoning was covered up by doctors, claims health team BMJ 2003 327 248 12896923 10.1136/bmj.327.7409.248-c
Murlidhar V Occupational health physicians: unwilling or unable to practise ethically Indian Journal of Medical ethics 2002 10 26
Nemery B Occupational respiratory disease Natl Med J India 1995 8 199 201 7549847
Occupational Health and Safety Centre
Murlidhar V Leach M, Scoones I and Wynne B Demystifying Occupational and Environmental Health Experience from India. Science and Citizens Globalisation and the challenge of engagement 2005 London, UK, Zed Books 130 141
Murlidhar V Murlidhar VJ Kanhere V Byssinosis in a Bombay textile mill Natl Med J India 1995 8 204 207 7549849
Murlidhar V Kanhere V Occupational noise-induced hearing loss: the first two cases compensated in India Natl Med J India 1998 11 150 9707710
Murlidhar V Occupational diseases among textile workers of Mumbai NMJI 2002 15 181 10.1002/jnm.449
Committee on Occupational and Environmental Health of the American Lung Association of San Diego and Imperial Counties Taking the occupational history Ann Intern Med 1983 99 641 651 6357019
Markowitz SB Marabia A Lilis R Miller A Nicholson WJ Levin S Clinical predictors of mortality from asbestosis in the North American Insulator Cohort Am J Respir Crit Care Med 1997 156 101 108 9230732
Murphy RLJ Gaensler EA Holford SK Del Bono EA Epler G Crackles in the early detection of asbestosis Am Rev Respir Dis 1984 129 375 379 6367573
Kamat SR Tyagi NK Rashid SSA Lung function in Indian adult subjects Lung India 1982 1 11 21
American Thoracic Society Diagnosis and Initial Management of Nonmalignant Diseases Related to Asbestos Am J Respir Crit Care Med 2004 170 691 715 15355871 10.1164/rccm.200310-1436ST
International Labour Office International Classification of Radiographs of Pneumoconioses 2003 Geneva, Switzerland, International Labour Organization
Murlidhar V Murlidhar V Kanhere V Evaluation of Impairments and Disabilities 2005 Mumbai, Bhalani Publishing House, Mumbai 16262892
Weill H Waggenspack C Bailey W Ziskin M Rossiter C Radiographic and physiologic patterns among workers engaged in manufacture of asbestos cement products J Occup Med 1973 15 248 252 4348280
Aberle DR Gamsu G Ray CS High-resolution CT of benign asbestos related diseases: clinical and radiographic correlation Am J Roentgenol 1988 151 883 891 3263018
Genevois PA De Vuyst P Dedeire S Cosaert J Vande Weyer R Sturyven J Conventional and high-resolution CT in asymptomatic asbestos-exposed workers. Acta Radiol 1994 35 226 229 8192957
Brodkin CA Barnhart S Anderson G Checkoway H Omenn GS Rosenstock L Correlation between respiratory symptoms and pulmonary function in asbestos-exposed workers. Am Rev Respir Dis 1993 148 32 37 8317811
Yates DH Browne K Stidolph PN Neville E Asbestos-related bilateral diffuse pleural thickening: natural history of radiographic and lung function abnormalities. Am J Respir Crit Care Med 1996 153 301 306 8542134
Braunwald E Harrison’s Principles of Internal Medicine 2001 15th New York, McGraw-Hill
|
16262892
|
PMC1289287
|
CC BY
|
2021-01-04 16:36:33
|
no
|
Environ Health. 2005 Oct 31; 4:24
|
utf-8
|
Environ Health
| 2,005 |
10.1186/1476-069X-4-24
|
oa_comm
|
==== Front
Immunome ResImmunome Research1745-7580BioMed Central London 1745-7580-1-41630575710.1186/1745-7580-1-4DatabaseAntiJen: a quantitative immunology database integrating functional, thermodynamic, kinetic, biophysical, and cellular data Toseland Christopher P [email protected] Debra J [email protected] Helen [email protected] Shelley L [email protected] Martin J [email protected] Kelly [email protected] Irini A [email protected] Pingping [email protected] Channa K [email protected] Darren R [email protected] Edward Jenner Institute for Vaccine Research, High Street, Compton, Berkshire, RG20 7NN, UK2005 6 10 2005 1 4 4 17 6 2005 6 10 2005 Copyright © 2005 Toseland et al; licensee BioMed Central Ltd.2005Toseland 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.
AntiJen is a database system focused on the integration of kinetic, thermodynamic, functional, and cellular data within the context of immunology and vaccinology. Compared to its progenitor JenPep, the interface has been completely rewritten and redesigned and now offers a wider variety of search methods, including a nucleotide and a peptide BLAST search. In terms of data archived, AntiJen has a richer and more complete breadth, depth, and scope, and this has seen the database increase to over 31,000 entries. AntiJen provides the most complete and up-to-date dataset of its kind. While AntiJen v2.0 retains a focus on both T cell and B cell epitopes, its greatest novelty is the archiving of continuous quantitative data on a variety of immunological molecular interactions. This includes thermodynamic and kinetic measures of peptide binding to TAP and the Major Histocompatibility Complex (MHC), peptide-MHC complexes binding to T cell receptors, antibodies binding to protein antigens and general immunological protein-protein interactions. The database also contains quantitative specificity data from position-specific peptide libraries and biophysical data, in the form of diffusion co-efficients and cell surface copy numbers, on MHCs and other immunological molecules. The uses of AntiJen include the design of vaccines and diagnostics, such as tetramers, and other laboratory reagents, as well as helping parameterize the bioinformatic or mathematical in silico modeling of the immune system. The database is accessible from the URL: .
B Cell EpitopesMHC-peptide bindingT Cell EpitopesMHCTCRAntibodiesVaccines
==== Body
Introduction
There is a vast, and ever increasing, volume of important information that has accumulated from decades of experimental analysis within immunology. This will only become compounded as high-throughput techniques begin to impinge upon the immunological biosciences. The only efficient way for this information to be properly utilized requires the development of databases that store it and systems that use it. Although the type of data archived may alter from case to case, nonetheless the creation, use, and manipulation of databases containing biologically important information is the most crucial feature of current bioinformatics, both as it supports the genomic and post-genomic revolutions and as a discipline in its own right. There is nothing new in developing databases focusing on immunology: many spotlighting the in-depth sequence analysis of individual immunomacromolecules have existed for some time [1]. Functional or epitope-orientated databases are a more recent development. Examples include the now defunct MHCPEP database [2], FIMM [3], SYFPEITHI [4], the HIV sequence database [5], the HLA ligand database [6], the EPIMHC database [7], and the MHCBN database [8].
An epitope is any molecular structure that can be recognised by the immune, or other biological, system. Epitopes, or the antigen from which they are derived, can be composed of protein, carbohydrate, lipid, nucleotide, or a combination thereof. It is through recognition of foreign, or non-self, epitopes that the immune system can identify and, hopefully, destroy pathogens. Hitherto, peptide epitopes have been the best studied, and have, traditionally, have been categorized as either T cell or B cell epitopes. T cell epitopes are peptides presented to the cellular arm of the immune system via the MHC-peptide-TCR complex. B cell epitopes represent surface regions of an antigen that are bound by soluble or membrane-bound antibodies. If this region of a protein antigen is comprised of residues distally separated within the primary structure, and brought into local proximity by protein folding, then it is termed a discontinuous or conformational B cell epitope. Linear or continuous B cell epitope residues are sequential in both primary structure and thus as a region on the proteins' surface. Such epitopes are predominantly identified by antigen-specific antibody cross-reactivity with peptides.
There is a need to create a databank for the wider disciplines of immuno-vaccinologists, which can act as a central repository and resource. Our aim is to complement other databanks [2-8] and thus we have developed AntiJen, a computational information resource for immunology and vaccinology that integrates quantitative kinetic, thermodynamic and biophysical data, with functional and cellular information. AntiJen v2.0, a development of our earlier database system JenPep [9,10], contains functional data on T cell and B cell epitopes. Moreover, the B cell archive is now sub-divided into linear and conformational epitopes. These epitopes form the basis of the humoral immune response and, unlike T cell epitopes, methods of prediction are often inaccurate [11]. A more in-depth B cell epitope archive should aid the development of prediction strategies. Antigen recognition by the Major Histocompatibility Complex (MHC) is vital to T cell activation hence, the inclusion of thermodynamic data on the binding of peptides to MHC molecules and T Cell Receptor (TCR) binding to peptide-MHC (pMHC) complexes. This data is complemented by kinetic data based on the same molecular interactions. Data on antigen processing and presentation is also included in AntiJen. Binding data derived from peptide interactions with the Transporter Associated with Antigen Processing (TAP transporter) are included in the archive. Additionally, quantitative specificity data from position-specific peptide libraries is included. AntiJen also incorporates thermodynamic data on protein-protein interactions, within an immunological context, such as co-receptor and superantigen binding, plus interactions with the MHC. All of these interactions are, potentially, key factors for the successful computational design of vaccines.
AntiJen also contains biophysical data, including diffusion coefficients and cell surface copy numbers, on a variety of immunological molecules. Such data provides insight into the number of target receptors, which is an important, if under explored, component of binding between cells. Indeed, the number of molecules expressed on the membrane can alter depending on disease. The final addition to the databank focuses upon antigen binding to antibodies. One key innovation is a greatly increased compendium of experimental conditions, which, in conjunction with a greatly enhanced search capacity, consolidates our databases as a unique, value-added data source, fostering developments within both in silico immunology and the wider community of immunovaccinology. The database is available from the URL: .
Database development
Relative to the database system used for JenPep [9,10], the interface to AntiJen is entirely new, having been completely rewritten. AntiJen has been designed and implemented using postgreSQL, a system comprising a relational database and database server, and has thus established increased database robustness, creating an improved infrastructure for foreseeable issues of data storage and data growth. Data within AntiJen is structured into twenty-four normalised tables. Each is category specific and holds either statistical or experimental data. Additional tables accommodate the keyword data – which powers our protein-orientated antigen search and allows integration of the BLAST search – and there is also a structural data table to accommodate links to external structural databases. The user interface consists of a series of HTML forms. The search requests from these forms target PERL scripts integrated with SQL which in turn query the database.
Database content
Compared to its progenitor JenPep [9,10], the data archived in AntiJen v2.0 has grown considerably in depth (additional data types such as experimental conditions), breadth (addition of new data to existing databases), and scope (addition of extra sub-databases containing novel kinds of information). Additions to AntiJen have been derived from exhaustive searching of the primary literature, to give a dataset of > 31,000 entries. AntiJen v2.0 now consists of 11 sub-databases; details of the different databases are given in Table 1. The relative sizes of the databases and the growth from JenPep are summarised in Table 2.
Table 1 AntiJen sub-databases and content.
DATABASE CONTENT
T Cell Epitopes Contains T Cell epitope peptides (known binders).
B Cell Epitopes Contains B Cell linear and conformational epitope peptides.
MHC-Peptide Binding data relating to antigenic peptides and MHC interactions.
TCR Binding data relating to antigenic peptides – TCR – MHC interactions.
TAP Binding data relating to antigenic peptides and TAP interactions.
Kinetics Kinetic binding data for MHC peptide interactions.
IPPI Binding data for a collection of immunological protein interactions.
Diffusion Coefficient Collection of Diffusion and Friction coefficients for surface peptides.
Copy Number Number/Abundance of cell surface molecules.
Peptide Libraries Relative binding data for antigenic peptide amino acid substitutions.
Antibody-Peptide A variety of antibodies known to bind proteins.
Table 2 Size of AntiJen relative to JenPep. The number of peptides for each category in the AntiJen database is given, distinguishing between class I and class II categories, where appropriate. Growth versus JenPep 1 and 2, the progenitors of AntiJen, is included. For certain data categories, most obviously TAP binding data, re-evaluation of the quality of data within JenPep has seen it decrease rather than increase, however the expansion of the data is clearly seen.
DATABASE JenPep v1.0 JenPep v2.0 AntiJen v1.0 AntiJen v2.0
Class 1 Class 2 Total Class 1 Class 2 Total Class 1 Class 2 TOTAL Class 1 Class 2 TOTAL
T cell epitotes 1266 795 2061 2060 1158 3218 2247 1578 3825 2402 1585 4158
MHC peptide binding 3196 2652 5848 6411 5925 12336 6853 7772 14625 7304 8114 15454
TAP peptide binding 432 441 408 1106
B cell epitotes 816 1295 3541
TCR – peptide-MHC 49 375 124 594 527 253 782
MHC peptide kinetics 704 243 947 897 294 1150
IPPI 805 2675
Copy Number 161 243 414
Diffusion coefficients 759
Peptide Libraries 897
Antibody 395
AntiJen contains both generic and dataset-specific data. For each entry, we record the peptide sequence (eg. YTSDYFISY) of the epitope using the standard one-letter code, its length (9 in this case), and, by linking to the sequence database Swiss-Prot or NCBI , the antigen to which the peptide sequence most closely matches (in the case of YTSDYFISY: C-ests-1 (p54), SWISS-PROT code P41156). The description of the antigen is, wherever possible, obtained directly from the literature. AntiJen is also linked to PUBMED. This allows us to record the original citation associated with the data. For example, for YTSDYFISY, we cite: Journal of Immunol 1994 volume 152 pages 3913–3924, PUBMED ID 8144960. For the T cell epitope, MHC ligand, and TCR-pMHC complex categories, we also record, for each peptide, the MHC restriction in terms of the host species, class (class I vs. class II), and, where the data is available, the serotype and allele. For YTSDYFISY, these data would be human, class I, HLA-A1, and A*0101.
Entries within AntiJen are, in turn, linked to external databases, which enables further in-depth cross referencing. As we have said, protein sequence identifiers, which may be the source of an antigenic peptide or immunological co-receptor, link directly to details in the Swiss-Prot database [12] or the NCBI protein database. Journal references can be viewed via a link to the PUBMED database , and thus to full literature references, where available. AntiJen also links to structural data, currently derived from the MPID database [13] and the Protein Data Bank [14]. The database aims to provide access to background data where available.
Allele identifiers serve as a link to the IMGT/HLA database at the European Bioinformatics Institute [15]. Inherent variability in the way MHC alleles are named within the primary literature prevents us from unambiguously standardizing nomenclature within AntiJen. HLA nomenclature follows that of the HLA Informatics Group . An allele is named using a defined pattern. For example, for HLA-A*0101: the HLA-A refers to the HLA locus; the first 01 to the serologically recognized A1 antigen and the final 01 to the individual HLA allele protein sequence. AntiJen stores the antigen classification (i.e. HLA-A1) and, when available, the specific allele. We have often encountered problems with the nonstandard allele reporting. A 4-digit HLA name necessarily implies the two digit serological antigen, a two digit classification clearly does not imply a specific allele.
During database compilation, a sequence search allows us to identify the protein from which an epitope sequence originates. However, because epitopes are generally short, their sequences may be present in several potential antigens: in orthologues, paralogues, or in totally unrelated sequences. As epitopes are processed from whole proteins via a complex proteolytic pathway, one can use the sequence context to infer preferred proteasomal or endoplasmic protease cleavage patterns, but not if its context is defined incorrectly. Moreover, assuming that AntiJen is used subsequently to assign the antigenic status of proteins, wrongly identifying particular proteins as antigens can lead to the percolation of annotation errors [16,17].
AntiJen is, where possible, a quantitative database archiving continuous measures of binding. This is a fundamental feature of several sub databases, such as the MHC ligand and pMHC-TCR databases. The binding of an immunological macromolecule to a peptide or other biomacromolecule is quantified as are other receptor-ligand interactions:
R+L↔RL
Here R is the receptor (an MHC or TCR), L the ligand (peptide or pMHC), and RL, the receptor-ligand complex (pMHC or pMHC-TCR complex). The rate of the forward reaction is proportional to [L] [R]. The rate of the reverse reaction is proportional to [RL] as no other species are involved in dissociation. At equilibrium, the forward and reverse rates are equal, and so using kon and koff as the respective constants:
kon[R][L] = koff[RL]
Rearranging:
Here KD is the equilibrium dissociation constant, which represents the concentration of ligand that occupies 50% of the equilibrium receptor population, and KA is the equivalent association constant.
Experimentally, the measurement of equilibrium dissociation constants is often addressed using radio-ligand binding assays. Saturation radio-ligand binding assays measure equilibrium binding, at a range of peptide concentrations, to establish affinity (Ka) and receptor number (Bmax). Competitive binding experiments determine binding at a single labelled ligand concentration in the presence of a range of concentrations of unlabelled ligand. AntiJen records a hierarchy of different binding measures in its different sub-databases. Equilibrium constants are the most dependable and sit atop this hierarchy. Next come IC50 values, which can be obtained from a competitive radio-ligand or fluorescence assay. These are the Fmost commonly reported binding measures.
Values obtained from radio-ligand or fluorescence methods may be significantly different. IC50 values for a peptide may vary between experiments depending on the intrinsic affinity and concentration of the standard radiolabelled reference peptide, as well as the intrinsic affinity of the test peptide. IC50 values vary with the equilibrium dissociation constant, at least within a single experiment. In practice, the variation in IC50 is often small enough that values can be compared between experiments. For the peptide discussed above, YTSDYFISY, the radiolabelled IC50 value recorded in AntiJen is 5.3 nM. BL50 values are also obtained from a peptide binding assay and are commonly reported. They are the half maximal binding levels calculated from mean fluorescence intensities of peptides binding to MHCs bound on the surface of RMA-S or T2 cells. Cells, pre-incubated with peptides, are labelled with a fluorescent monoclonal antibody. An overview of the thermodynamic and kinetic binding data within AntiJen is given in Table 3.
Table 3 AntiJen Thermodynamic and Kinetic Data. An overview of the 6 AntiJen databases that provide binding data. It must also be noted that several of the databases contain additional data not present in any of the other databanks.
MHC-Peptide Kinetics IPPI TAP pMHC-TCR Antibody TOTAL
IC50 8562 0 247 1000 0 4 9813
Kon 0 188 563 0 157 87 995
Koff 0 146 610 0 150 101 1007
KD 359 156 1143 16 227 70 1971
Ka 65 0 37 0 28 132 262
t1/2 0 207 72 0 148 0 427
AntiJen also now contains experimental conditions, such as temperature and pH. A summary of this data is given in Table 4. The accuracy of data depends greatly upon the experimental method used. The grouping of data with respect to specific experimental techniques allows a more thorough assessment of training sets. Figure 1 shows the distribution of data for each type of analysis with respect to each database. The MHC Kinetics and TAP databases highlight the problems outlined above. The kinetics database contains data determined from over 14 methods while the TAP database is derived from 4 methods, with radiolabelled assays accounting for 80% of the data.
Table 4 Experimental conditions and associated information archived in AntiJen. Number of recorded experimental conditions stored within the AntiJen database. For each condition (temperature, pH, etc.) we show here the number of entries within a particular sub-database. [Standard] is the concentration of labelled standard peptide in an assay. Likewise, [competitor] is the concentration of competitor peptide within a competition assay. [peptide] is the concentration of peptide in a kinetic experiment. The Method category refers to a standard procedure used to perform a particular assay. Differences in the number of recorded data, relative to figures in Table 1, arise primarily from the omission of key details from particular papers. Archiving of experimental conditions is on-going.
DATABASE TOTAL pH Temperature [standard] Stand. peptide seq. [competitor] Method [peptide]
MHC Binding 15454 6679 9831 10893 12796 5007 1251
MHC Kinetics 1150 677 1101 1149 606
TAP Binding 1106 22 243 1092 1101 86 981
TCR-pMHC 782 426 632 668
IPPIa 2675 726 1371 2600
Copy Number 414 183 278 414
Peptide Libraries 897 897 897
Diffusion Coefficient 759 321 668 736
Antibody 395 119 115 372
Figure 1 The distribution of experimental methods applied within each database. The number of different experimental methods and the abundance of data relating to the method is shown within the figures. The 'OTHERS' category refers to methods for which there is a relatively small number of entries.
The compilation process has highlighted the considerable inconsistency within the immunological literature regarding the recording of such fundamental data. AntiJen contains, however, a direct, verbatim transcription of data from the primary literature. As such, we do not attempt, as a matter of policy, the comprehensive and retrospective correction of potential errors. To undertake such correction would only compound any errors, introducing the kind of percolating inconsistencies so much a feature of other database systems [16,17]. Further inaccuracies may stem from our logistic inability to verify data independently, therefore we must trust those values reported in the literature.
Subsidiary Databases in AntiJen
The AntiJen database contains a number of sub-databases. Each of these contains data on different aspects of the biological function and/or biophysical properties of different classes of immunomacromolecule. We describe the nature and content of each sub database below.
B Cell Epitopes
Epitopes are the principal chemical moieties recognized by the immune system. Although the importance of non-peptide epitopes, such as carbohydrates and lipids, is now increasingly well understood, peptidic B cell and T cell epitopes remain the principal tools by which the intricacy of the immune response can be explored. B cell epitopes are regions of the surface of a protein, or other biomacromolecule, recognized by soluble or membrane-bound Antibody molecules. In developing AntiJen, we have discarded the contents of our previous B cell archive and constructed one de novo. It contains an entirely new data set with a substantially different data structure. There are two forms of epitopes: linear and discontinuous. A linear B cell epitope is composed of a single stretch of sequential residues. A discontinuous B cell epitope is composed of sequentially separate residues brought into close proximity in a conformationally-dependent arrangement. The data we archive is primarily focused upon linear epitopes. This is due to the far greater amount of experimental data available for linear epitopes, which reflects both the relatively facile experiments needed to identify them and an implicit belief in their utility as potential vaccine epitopes. By contrast, discontinuous epitopes are thought to be more prevalent within folded proteins, but are far more challenging to determine experimentally. The archive catalogues the sequence of binding peptides, and also gives the length and source: TTGDVIASS, a 9 amino acid peptide from Escherichia coli non-fimbrial adhesion. Residues identified as important in binding to the antibody are recorded. This may correspond to a whole peptide or a subsequence, such as TTGDVI in the above example. The peptides are also categorized in terms of their relative observed immunodominance. Antibodies host organism and isotype are recorded. The current B cell epitope archive contains 3,541 epitopes.
T Cell Epitopes
T cell epitopes are short peptides bound by major histocompatibility complexes (MHC) and subsequently recognized by T cells. Epitopes recognized by both CD4+ and CD8+ T cells are included in the database. Such epitopes can be identified in many different ways. However, this diversity of measurement imposes a certain need for consistency, necessitating the requirement for recording a range of different experimental methods. The archive has expanded to include 4,158 entries. The entries contain the epitopes, ranging in length from 4 to 38 amino acids, peptide information, detailing the source, with links to Swiss-Prot and the corresponding MHC restriction data such as Serotype, Allele and Class. Additionally, the peptides are categorized in to groups such as Allergens, Bacterial, Cancer, Human, Viral and Self peptides.
MHC – Peptide binding
AntiJen continues to archive quantitative data on the thermodynamics of peptide interactions [18,19], and it has expanded in number and content, with additions such as experimental conditions, plus specific Standard and Competitor peptide concentrations used in the assays. The current archive contains 15,454 entries. The sequence of the binding peptide, along with the source, plus relevant MHC restriction data is recorded. The restriction alleles currently include those from Human, Mouse, Rat, Rhesus Monkey, Cotton-top Tamarin, and Chimpanzee. AntiJen contains IC50 values, binding affinity measurements from competitive binding assays, for which the standard and competitor peptides and concentrations are recorded, plus BL50 values, calculated from peptide stabilizing assays. Where possible, antibodies and the concentrations used to calculate BL50 values are archived. Additionally, but on a somewhat smaller scale, equilibrium association (KA) and dissociation (KD) constants are recorded for peptide-MHC interaction. Melting temperatures (Tm) and signal wavelength are also recorded; this is the temperature and wavelength at which 50% of the MHC protein is denatured as measured by circular dichroism. AntiJen also records so-called Weak/Non-binders. This indicates that the peptide has been tested in an MHC restriction assay and has been found to exhibit a binding affinity, i.e. an IC50 value > 10,000 nM for a radio-ligand assay, so low that it can be categorized as inactive.
pMHC-T Cell Receptor interaction
The TCR sub-database contains 782 entries, which records thermodynamic and kinetic binding data for the interaction of peptide-MHC (pMHC) complexes with TCRs. Different MHCs exhibit a distinct selectivity for certain peptide sequences. T cell receptors, in their turn, also exhibit different affinities for peptide-MHC complexes. The entries contain epitope, peptide source and MHC restriction data, as described above, plus TCR structure information, located at the MMBD database . Furthermore, any mutations are noted and a designated name for the TCR is archived. In addition, the peptides are recorded as either agonists or antagonists. The binding data is given as equilibrium constants (KD), EC50 values, rate of association (Kon), rate of dissociation (Koff), association constant (KA) and the half-life (t1/2) of the TCR-peptide interaction.
TAP Binding
This dataset contains binding data for the interactions between peptides and the TAP transporter, one of the principle steps in antigen presentation. As with the peptide-MHC database, the data is established from competitor binding experiments, based on labelled assays. Therefore, standard and competitor peptide sequences and their concentrations are recorded. The binding data is given as IC50 and KD values. The database currently contains 1,106 entries, with peptides from Human, Rat and Mouse sources. Based on IC50 values > 10,000 nM, the peptides are categorized as weak/non-binders. The entries have increase in number from the level found in JenPep, although several entries were removed in an effort to increase the accuracy and consistency of the archive (Table 2).
Peptide-MHC Kinetics
AntiJen's kinetics sub-database, which contains 1,150 entries, mostly relates to Class I MHC data. It records measurements for forward and reverse rate constants for complexation events. This complements the thermodynamic measurements on peptide-MHC binding described above. The data currently focuses upon both the half-lives of binding interactions, as well as association and dissociation rate constant values (Kon and Koff) for the recorded epitopes. Additionally, concentrations of the peptide, MHC and TAP are archived. The half-life for radioisotope labelled β2-microglobulin dissociation from an MHC class I complex, as measured at 37°C, is also archived. This is a kinetic measurement rather than a thermodynamic one, although it is often assumed that the greater the half-life the stronger the peptide-MHC complex. The half-life (t1/2) equals:
Here the t1/2 corresponds to the dissociation of the MHC-β2 microglobulin complex rather than the kinetics of the protein-ligand interaction, but is still peptide dependent, as well as kinetic in nature.
Immunological Protein-Protein Interactions
The immune system is built around protein interactions therefore we developed another new sub-database which deals with Immunological Protein-Protein Interactions (IPPI). This archive contains 2,675 entries based on a variety of binding data, such as Kon and Koff rates, for a range of macromolecules implicated in physiological or pathological interactions, as well as KD, KA and IC50 values. The molecules include receptors such as CD4 or CD8 molecules, superantigens and other microbial virulence factors, cytokine receptors and cell adhesion molecules. The entries detail both protein partners involved in the binding interaction, with links provided to the NCBI-Entrez database. Additional data for MHC receptors is archived, whereby the reactive epitope is recorded and the co-receptors are categorized into viral, bacterial and self peptides. MHC data outlined in the previous databases is given, where appropriate including any mutations to the MHC.
Antibody – Protein Binding
AntiJen also contains a further sub-database, which comprises thermodynamic data relating to antibody-antigen binding. The dataset contains 395 entries for antigen proteins and antibodies, mostly derived from viral and mammalian sources. Reported values were obtained using radiolabelled assays and BIAcore analysis. This archive should aid in the selection of antibodies and peptides for in vitro studies. The entries list the antibodies and the binding/kinetic data, consisting of KD and KA values and to a lesser extent Kon, Koff and IC50 values.
Peptide Libraries
This archive further complements our MHC binding databases by indicating the relative contribution of residues within peptide libraries to MHC binding. 897 entries contain quantitative specificity data derived from position specific peptide libraries [20]. This catalogues the relative effect on affinity, in the form of IC50, log relative SD50 and log SI values, of all substitutions, at all peptides positions, against a random sequence backdrop. All of the libraries relating to a known peptide binder are designated a name within AntiJen, this usually consists of the author and year of publication. The archive contains the core peptide, along with the mutation position and the substituted amino acid. The corresponding MHC data is given as mentioned above.
Diffusion Co-efficients
To further increase the range of data archived, AntiJen also contains 759 records of cellular biophysical data, in the form of diffusion co-efficients, recorded as cm2s-1, for a diversity of cell surface molecules, including MHCs (Mouse and Human), viral peptides and other receptors [21]. The molecules are either chemically or fluorescently labelled and then measured using one of two methods: Single Particle Tracking (SPT) or Fluorescence Photobleaching Recovery (FPR). SPT monitors the lateral motion of a labelled molecule while FPR measures the rate of subsequent infiltration from a photobleached section of the membrane. Friction co-efficient data is also given, which measures of the velocity and force applied to an antibody-coated bead. Records contain the cell or cell type where diffusion is occurring, the name of the diffusing protein along with the form of labelling applied. Furthermore, specific experimental data is given such as antibody bead size. In this case the diffusion of the beads is monitored. Increasingly, data relating to photobleaching is included, such as beam power, bleaching duration, pre- and post-bleach time, etc..
Copy Numbers
The final sub-database contains 414 measures of cell surface populations of different molecules, called cell surface copy numbers hereafter. This database focuses on an array of molecules, including Class I MHCs (22) and TCRs [23]. The entries are given as number and type of MHC molecules, number of MHC-peptide complexes or abundance of peptides associated with each MHC serotype, generally defined by mass spectroscopy. The entries list the cell type, the antibody bound to the MHC and, if appropriate, the binding epitope. This only applies to the number of MHC-peptide complexes and the abundance of peptides associated with each MHC.
Searching the database
Search mechanisms within AntiJen are significantly improved and allow either a detailed or a broad search from a simple user interface. From our experience with JenPep, we recognize that accessibility to the data in a user friendly manner is a vital requirement, and have improved our current search mechanisms and developed new search interfaces. Two different search mechanisms are available. One is based on BLAST [24] and the other is a bespoke system, allowing several alternative searches. Within a typical search, the user-entered search criteria are carried from an HTML form to a category specific PERL/SQL script, which performs the database queries.
The BLAST search allows querying of a peptide or nucleotide sequence against the proteins contained in AntiJen; all entries containing data within AntiJen which are relevant to a protein sequence are linked via the BLAST output. A local database of protein sequences is searched with BLASTP or BLASTX [24] using the BLOSUM64 matrix. All BLAST control variables are fixed. An HTML Front-End (Figure 2), where a sequence can be entered or uploaded, connects to a web server-based PERL/CGI scripts, which interacts with BLAST. An annotated version of the default BLAST output is produced and links to AntiJen entries via SWISS-PROT [12] accession codes, which act as a query within a Keyword search. This allows AntiJen entries to be viewed directly from the BLAST output.
Figure 2 Overview of the different search methods within AntiJen. The example search is focused upon an MHC ligand. The MHC ligand data can be searched directly (A) from a link on the AntiJen homepage, a broad search and specific search is available. A search for the epitope AMALLRLPLV, has one hit (D), this leads to the entry (G). All of the other sub-databases can be searched in this manner from the homepage. The two other searches are more generalised. A Keyword search (B) carries out a broad search on the whole database, for the criteria – Bacteria. This search gives 139 hits (E) and all of the sub-database entries can be selected from this output. The final search method is a BLAST search (C). The peptide (or nucleotide) sequence is queried against a local protein database. The output (F) provides links to the sub-databases relative to the protein.
At present, peptide string, keyword, and protein name index searches are available within the bespoke system, which allows querying of individual peptide sequences or at the level of whole protein antigens. An overview of the AntiJen search systems is given in Figure 2 and 3. Epitopes, MHC, TCR, TAP peptide binding and kinetics databases can all be searched using sequence strings. The search protocol first returns an epitope list and a count of epitope matches. Subsequently, experimental criteria can be accessed for each selected epitope. Peptides can be searched using an amino acid orientated query: a sparse peptide string, similar in form to a peptide binding motif [3] or a PROSITE pattern [25], is used to identify all matching sequences. See Figure 4. Alternately, a list of protein antigens within AntiJen can be searched using keywords; the thermodynamic binding data such as MHC-peptide, TCR-pMHC and TAP, plus the B and T cell archives, related to the search criteria can then be selected from the matches based upon the SWISS-PROT accession codes [12], displaying all corresponding entries in the database.The other search method, allows the IPPI, diffusion co-efficients, peptide libraries, antibody-protein and copy number sub-databases to be searched using an index method, within a user-friendly HTML drop-down menu. Each of these methods can also be moderated using subsidiary search filters, data size ranges, and result presentation alternatives, such as peptide length or IC50 values. Minimum and maximum values can be used to restrict results, as can selection of MHC restriction alleles.
Figure 3 Searchable database types within AntiJen. The database contains 3 types of searchable sub database: a set of Antigens searchable by keyword, various databases of functional and thermodynamic data searchable by peptide sequence, and a database of immunological protein-protein interactions searchable through an index. Peptide sequence searches can be explicit or "motif" based. Searches can also be focussed by setting the value ranges for properties, such as IC50 etc, recorded in the databases. Currently there is a link between all the thermodynamic and kinetic databases and between them and the database of antigens. Only antigens with data in one of the other sub-databases are included in the antigen database. Links to external databases are also indicated. The BLAST search provides an overall search of the databases except the Protein-Protein Interactions archive.
Figure 4 Sparse peptide sequence search. Examples of the two search types available in AntiJen: (Search 1) a substring query and (Searches 2 and 3) PROSITE-like sparse queries allowing sets of variable (indicated by asterisks) and alternative (encased in square brackets) amino acids in the search. In our example, each query is an extension of the previous search. The initial search type returns a single hit, which is a weak binder. By introducing variable amino acid positions within the query string, the second query permits access to a larger data set with 39 hits being returned. The third search utilises both variable and alternative amino acid request functionality and returns 1996 hits. The number of entries returned can be reduced by specifying the epitope length, limiting IC50 values and restricting by one MHC allele. This search is constrained to peptides of amino acid length 9, which returns 280 hits. Constraining further by MHC allele HLA-A*0201, reduces this to 120. An additional constraint using the IC50 data filter, and requesting values below 500 only (the epitope range), reduces this again to 60.
Discussion
The name given to our new database, AntiJen, reflects a shift from a peptide orientated database structure, which was inherent within our earlier JenPep database, to one which can properly balance its focus on both protein antigens and isolated peptides. As such, it represents an important, integrated, immunological data resource. AntiJen now provides broad insight into both T cell and B cell mediated antigen recognition. In addition, through the auspices of the IPPI sub-database, the database also throws light on co-stimulation by co-receptors and gives important insights into the innate immune response. Our approach to protein-protein interactions, focusing on measured affinities, complements other methods, such as the Yeast-2-Hybrid system, which, while giving greater volumes of data, has problems of accuracy [26]. This, while not devoid of experimental artifact, gives a usefully different perspective on cataloguing protein-protein interactions. The further addition of weak binding notation to the MHC-peptide and TAP provides a greater overview of the nature of antigenic epitopes. This is further improved by the addition of the peptide libraries database, whereby key peptide residues can be highlighted. New databases have expanded the breadth of AntiJen to include biophysical data such as diffusion co-efficients and cellular data such as abundance of molecules. The antibody-antigenic protein sub-database will also provide a key resource for in vivo and in vitro studies, aiding in the selection of antibodies and peptide/protein targets.
AntiJen distinguishes itself from the other specific binding databases [2-8] in several ways. Firstly, more data is recorded; our MHC-peptide database contains over 2,000 more entries than MHCPEP [2] and 10,000 more entries than EPIMHC [7]. Additionally, we have not restricted our archive to only high binders or to a specific category, as seen in EPIMHC and the HIV sequence databases [5]. Furthermore, AntiJen is currently a curated database, which is constantly expanding.
Most obviously, AntiJen is useful in the design of epitope and subunit vaccines. Additionally, AntiJen is helpful in the design of clinical diagnostics and other laboratory reagents, such as the selection of peptides for tetramer design. AntiJen is also useful in the parameterization of mathematical models in theoretical immunology [27]. The redevelopment of the database has focused not only on content, but also on infrastructure. The current system, based on epitope string, keyword and index searches, along with an overall BLAST search, plus the redesigned HTML interface, leads to much greater accessibility and usability. Finally, the database acts as a repository of quantitative, continuous data, for the development of data-driven in silico predictive models, such as prediction of epitopes and MHC binding [18,19,28,29] through QSAR modeling.
Future work
Future tasks in the development of AntiJen, fall into two principle categories: eliminating deficiencies, errors, and inconsistencies within the database and simultaneously reinforcing it by expanding its depth, breadth, and scope. We also need to monitor updates within external databases, so that any alterations are mirrored within the archive. Like all other such repositories, AntiJen is prone to both systematic and random errors within the data accumulation process. User feedback and our interactions with immunologists will hopefully address persisting errors. Deficiencies in our database include our current inability to encode chemically or post-translationally modified peptides, non-natural MHC mutants and non-amino acid peptidomimetic MHC ligands. Additionally, it would also be interesting to complement our existing data on TAP binding with information on antigen presentation pathways, such as proteasomal and cathepsin cleavage patterns. Moreover, the compilation of B cell or antibody epitope data is an area ripe for robust development. Linear and conformational B cell epitopes are very much larger in number than our current compilation, leaving us scope to greatly increase recorded epitopes.
Conclusion
The development of a database is always a work in progress. Not simply because the easily accessible literature is typically always increasing, but also because of the desire to capture as much of the existing, but hidden, literature, as possible. In the post-genomic era, the database has formed the bedrock and language of bioinformatics; increasingly databases are coming to underpin our modern understanding of biology as a whole. Traditionally, databases have arisen as a response to need, answering the individual and idiosyncratic questions posed by biologists. However, the history of bioinformatics databases has shown the extraordinarily diverse ways in which archived data can be used.
In creating AntiJen, we were motivated partly by our desire, and the desire of collaborators, to use the data within it to build predictive in silico models [16,17,28,29], and partly by a more altruistic desire to generate a useful, integrated database system with a quantitative focus. AntiJen has many potential uses throughout the immunological discipline, from immunoinformaticans to experimental immunologists and vaccinologists. By increasing the degree to which data is machine readable and web accessible, we open up new, and previously unthought-of, avenues for the bioinformatic exploration of immunological data.
AntiJen is a primary data resource, amongst the most complete of its kind, yet, like SWISSPROT [12] or GenBank [30] decades ago, it is still relatively small and offers much scope for improved annotation. We see the database as a foundation from which to consolidate, through time, thus achieving a comprehensive resource of immunological data.
Acknowledgements
We acknowledge the helpful assistance of Ms Christianna Zygouri. We should also like to thank Dr P Borrow and Prof V Brusic for helpful discussions. The Edward Jenner Institute for Vaccine Research wishes to thank its sponsors: GlaxoSmithKline, the Medical Research Council, the Biotechnology and Biological Sciences Research Council, and the UK Department of Health.
==== Refs
Brusic V Zeleznikow J Petrovsky N Molecular immunology databases and data repositories J Immunol Methods 2000 238 17 28 10758232 10.1016/S0022-1759(00)00159-9
Brusic V Rudy G Harrison LC MHCPEP, a database of MHC-binding peptides: update 1997 Nucleic Acids Res 1998 26 368 371 9399876 10.1093/nar/26.1.368
Schonbach C Koh JL Flower DR Wong L Brusic V FIMM, a database of functional molecular immunology: update 2002 Nucleic Acids Res 2002 30 226 229 11752300 10.1093/nar/30.1.226
Rammensee H Bachmann J Emmerich NP Bachor OA Stevanovic S SYFPEITHI: database for MHC ligands and peptide motifs Immunogenetics 1999 50 213 219 10602881 10.1007/s002510050595
Kuiken C Korber B Shafer RW HIV sequence databases AIDS Rev 2003 5 52 61 12875108
Sathiamurthy M Hickman HD Cavett JW Zahoor A Prilliman K Metcalf S Fernandez Vina M Hildebrand WH Population of the HLA ligand database Tissue Antigens 2003 61 12 19 12622773 10.1034/j.1399-0039.2003.610102.x
Reche PA Zhang H Glutting JP Reinherz EL EPIMHC: a curated database of MHC-binding peptides for customized computational vaccinology Bioinformatics 2005 21 2140 2141 15657103 10.1093/bioinformatics/bti269
Bhasin M Singh H Raghava GPS MHCBN: a comprehensive database of MHC binding and non-binding peptides 2002
Blythe MJ Doytchinova IA Flower DR JenPep: a database of quantitative functional peptide data for immunology Bioinformatics 2002 18 434 439 11934742 10.1093/bioinformatics/18.3.434
McSparron H Blythe MJ Zygouri C Doytchinova IA Flower DR JenPep: A Novel Computational Information Resource for Immunobiology and Vaccinology J Chem Inf Comput Sci 2003 43 1276 1287 12870921 10.1021/ci030461e
Blythe MJ Flower DR Benchmarking B cell epitope prediction: underperformance of existing methods Protein Sci 2005 14 246 248 15576553 10.1110/ps.041059505
Boeckmann B Bairoch A Apweiler R Blatter MC Estreicher A Gasteiger E Martin MJ Michoud K O'Donovan C Phan I Pilbout S Schneider M The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003 Nucleic Acids Res 2003 31 365 370 12520024 10.1093/nar/gkg095
Govindarajan KR Kangueane P Tan TW Ranganathan S MPID: MHC-Peptide Interaction Database for sequence-structure-function information on peptides binding to MHC molecules Bioinformatics 2003 19 309 310 12538264 10.1093/bioinformatics/19.2.309
Berman HM Westbrook J Feng Z Gilliland G Bhat TN Weissig H Shindyalov IN Bourne PE The Protein Data Bank Nucleic Acids Res 2000 28 235 242 10592235 10.1093/nar/28.1.235
Robinson J Waller MJ Parham P de Groot N Bontrop R Kennedy LJ Stoehr P Marsh SG IMGT/HLA and IMGT/MHC: sequence databases for the study of the major histocompatibility complex Nucleic Acids Res 2003 31 311 314 12520010 10.1093/nar/gkg070
Gilks WR Audit B De Angelis D Tsoka S Ouzounis CA Modeling the percolation of annotation errors in a database of protein sequences Bioinformatics 2002 18 1641 1649 12490449 10.1093/bioinformatics/18.12.1641
Linial M How incorrect annotations evolve – the case of short ORFs Trends Biotechnol 2002 21 298 300 12837613 10.1016/S0167-7799(03)00139-2
Doytchinova IA Flower DR A comparative molecular similarity index analysis (CoMSIA) study identifies an HLA-A2 binding supermotif J Comput Aided Mol Des 2002 16 535 544 12602948 10.1023/A:1021917203966
Doytchinova IA Flower DR Towards the quantitative prediction of T-cell epitopes: coMFA and coMSIA studies of peptides with affinity for the class I MHC molecule HLA-A*0201 J Med Chem 2001 44 3572 3581 11606121
Liu R Enstrom AM Lam KS Combinatorial peptide library methods for immunobiology research Exp Hematol 2003 31 11 30 12543103 10.1016/S0301-472X(02)01008-1
Chakrabarti A Matko J Rahman NA Barisas BG Edidin M Self-association of class I major histocompatibility complex molecules in liposome and cell surface membranes Biochemistry 1992 31 7182 7189 1322696 10.1021/bi00146a022
Lippolis JD White FM Marto JA Luckey CJ Bullock TN Shabanowitz J Hunt DF Engelhard VH Analysis of MHC class II antigen processing by quantitation of peptides that constitute nested sets J Immunol 2002 169 5089 5097 12391225
Marano N Holowka D Baird B Bivalent binding of an anti-CD3 antibody to Jurkat cells induces association of the T cell receptor complex with the cytoskeleton J Immunol 1989 143 931 938 2787352
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
Gattiker A Gasteiger E Bairoch A ScanProsite: a reference implementation of a PROSITE scanning tool Appl Bioinformatics 2002 1 107 108 15130850
Sprinzak E Sattath S Margalit H How reliable are experimental protein-protein interaction data? J Mol Biol 2003 327 919 923 12662919 10.1016/S0022-2836(03)00239-0
Andersen PS Menne C Mariuzza RA Geisler C Karjalainen K A response calculus for immobilized T cell receptor ligands J Biol Chem 2001 276 49125 49132 11592972 10.1074/jbc.M109396200
Guan P Doytchinova IA Flower DR HLA-A3 supermotif defined by quantitative structure-activity relationship analysis Protein Eng 2003 16 11 18 12646688 10.1093/proeng/gzg005
Doytchinova IA Blythe MJ Flower DR Additive method for the prediction of protein-peptide binding affinity. Application to the MHC class I molecule HLA-A*0201 J Proteome Res 2002 1 263 272 12645903 10.1021/pr015513z
Benson DA Karsch-Mizradu I Lipman DJ Ostell J Rapp BA Wheeler DL GenBank Nucleic Acid Res 2000 28 15 18 10592170 10.1093/nar/28.1.15
|
16305757
|
PMC1289288
|
CC BY
|
2021-01-04 16:37:20
|
no
|
Immunome Res. 2005 Oct 6; 1:4
|
utf-8
|
Immunome Res
| 2,005 |
10.1186/1745-7580-1-4
|
oa_comm
|
==== Front
Int Semin Surg OncolInternational seminars in surgical oncology : ISSO1477-7800BioMed Central London 1477-7800-2-231625091110.1186/1477-7800-2-23ResearchExpression of thromboxane synthase, TBXAS1 and the thromboxane A2 receptor, TBXA2R, in human breast cancer Watkins Gareth [email protected] Anthony [email protected] Robert E [email protected] Wen G [email protected] Metastasis & Angiogenesis Research Group, Wales College of Medicine, Cardiff University Cardiff UK2005 26 10 2005 2 23 23 27 9 2005 26 10 2005 Copyright © 2005 Watkins et al; licensee BioMed Central Ltd.2005Watkins 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
Thromboxane synthase (TxS) metabolizes the cyclooxygenase product, prostaglandin H(2), into thromboxanes. Some of the thromboxanes are known to be biologically active on cancer cells. The aim of the study was to investigate the expression of thromboxane synthases, TBXAS1 and the thromboxane A2 receptor, TBXA2R in a cohort of human breast cancer patients and also to assess their potential clinical relevance.
Methods
Human breast tumour tissues (n = 120) and non-neoplastic mammary tissues (n = 32) were studied. Levels of TBXA2R and TBXAS1 transcripts were quantified using quantitative real-time RT-PCR analysis and correlated with clinical/pathological information including nodal status, grade, prognosis and long term survival (median follow-up period 120 months).
Results
Breast tumour tissue expressed higher levels of TBXA2R compared with normal mammary tissues, although the difference was not statistically significant (p = 0.09). There was no difference between tumour and normal tissues for TBXAS1. However, TBXA2R expression was significantly increased in grade 3 tumours(p = 0.006 vs grade 1), while TBXAS1 was significantly reduced in grade 3 tumours (p = 0.026 vs grade 1 tumours). A similar differential expression pattern was seen in tumours from patients with different prognosis, in that patients with predicted poor prognosis had higher, but not statistically different, levels of TBXA2R, and significantly lower levels of TBXAS1 (p = 0.008). Finally, Kaplan-Meier survival analysis has shown that patients with high levels of TBXA2R had significantly shorter disease free survival (103.8 (79.1–128.5) months) compared with those with low levels (123.7 (112.0–135.3)) months, p = 0.043.
Conclusion
Thromboxane synthases are differentially expressed in human breast cancer. While TBXA2R is highly expressed in aggressive tumours and linked with poor prognosis, TBXAS1 is expressed at significantly low levels in high grade tumours and tumour patients with poor prognosis. TBXA2R thus has a significant prognostic value in clinical breast cancer.
==== Body
Background
Prostaglandin metabolites (eicosanoids) are known to be selectively active in regulating functions in cells including cancer cells [1,2]. For example, 12-HETE and 13-HODE have been shown to act as pro- and anti-cancer eicosanoids in a range of cancer cells. Eicosanoids are generated from prostaglandins by specific enzymes, some of which have been shown to be actively involved in the development and progression of cancers [3,4]. We have previously reported aberrant expression of the other groups of prostaglandin enzymes 5-, 12-, 15 LOX and COX-2 in human breast cancer and have demonstrated a distinct pattern of difference with these enzymes [5].
Thromboxane synthase (TxS) metabolizes the cyclooxygenase product, prostanglandin H(2), into thromboxane A(2) (TXA(2)), which can cause vessel constriction, platelet activation, and aggregation. In human prostate cancer, thromboxane synthase has been found to be weakly expressed or absent in normal differentiated luminal or secretory cells, significantly elevated in less differentiated or advanced prostate tumors, and markedly increased in tumors with perineural invasion [6]. Over-expression of the enzyme in prostate cancer cells increased the cellular motility [6]. The same over expression of TBXA2R has been seen in adenocarcinoma and squamous cell carcinoma of the lung [7].
Thromboxane synthase inhibitors have been shown to induce apoptosis in glioma cells [8]. In a colorectal tumour model, cancer cells transduced with TXA(2) synthase cDNA produced faster growing tumours, an effect that can be reversed by TXA inhibitors [9].
In cancer cells, including breast cancer cells, TBXA2R has been shown not to influence the adhesion of cancer cells to matrix proteins [10]. Interestingly, TBXA2R has been shown to influence angiogenesis in a lung tumour model, potentially by affecting the migration of endothelial cells [11]. Thromboxane synthase has been shown to be associated with metastasis of renal cell carcinoma [12]. Increased expression of TXA synthase enzymes is a feature of differentiated monocytoid leukaemia cell lines [13]. Early studies using thromboxane synthase inhibitors in vivo have failed to show any beneficial effects on metastasis or spread to lymph nodes [14]. Thromboxane (TX) synthase inhibitors and a TBXA2R receptor antagonist have been found to inhibit the formation of metastasis from tail vein injected B16a cells, as well as reduce spontaneous metastasis from subcutaneous B16a and Lewis lung carcinoma tumours [15].
In the current study, we have investigated the level of expression of Thromboxane A synthase 1, TBXAS1 (otherwise known as Cyp5 and TXS) and Thromboxane A2 receptor, TBXA2R, in a cohort of human breast cancer patients. In addition, we have analysed the clinical and prognostic relevance of both enzymes with the clinical outcome of the patients over a 10 year period.
Materials and methods
Tissues and patients
The cohort of breast tissues and patients were as previously described [5,16], except that the median follow-up for the patients was 120 months. 120 tumour tissues and 32 normal tissues were included in the current study, in which normal tissues were obtained from the same patients with breast tumour but were away from tumour margins. The normal background tissues were free from tumour involvement as confirmed by histological examination. The pathological data, tumour staging, histological types and the initial prognostic index for the patients were previously described and is given in table 1.
Table 1 Clinical and pathological details of the study cohort
Negative (n =) Positive (n =)
Nodal status n = 65 55
ER status 71 49
Grade Grade 1 Grade 2 Grade 3
n = 23 41 56
Others
Histology Ductal Lobular medullary tubular mucinous
n = 88 14 2 2 4
TNM staging TNM 1 TNM 2 TNM 3 TNM 4
n = 69 40 7 4
Clinical outcome Disease free With Metastasis With local recur. Died of breast Cancer Died of unrelated diseases
n = 81 7 5 20 7
Tissue processing and extraction of RNA and generation of cDNA
Tissue processing was as we previously reported [17]. Frozen tissues were sectioned using a cryostat and stored at -20°C until use. Over 20 frozen sections from the tissues were homogenised in a RNA extraction solution using a hand held homogeniser to extract total RNA. The concentrations of RNA were quantified using a UV spectrophotometer. 1 μg RNA was used to generate cDNA using a commercially available RT kit (AbGene Laboratories, Essex, England, UK).
Quantitative analysis of thromboxane synthases
The level of thromboxane synthases transcripts from the above prepared cDNA was determined using a real-time quantitative PCR, based on the Amplifluor technology [18-20]. Briefly, Primers were designed using the Beacon Designer software (version 2, Palo Alto, California, USA), to amplify regions of human thromboxane synthases that have no significant overlap with other known sequences and that the amplified products span over at least one intron. Primers for TBXA2R (gene bank accession number U27325) were 5'tcagctcctggggatcat'3 and 5'actgaacctgaccgtacaggtttcgcagcactgtct'3 and for TXA synthase (gene bank accession number NM_030984) 5'caggtgttggttgagaactt'3 and 5'actgaacctgaccgtacatgtcacgtaaaaacagaacg'3. To one of the primer was added an additional sequence, known as the Z sequence (5'actgaacctgaccgtaca'3) which is complementary to the universal Z probe (Intergen Inc., Oxford, England, UK). A Taqman detection kit for β-actin was purchased from Perkin-Elmer. The reaction was carried out using the following: Hot-start Q-master mix (Abgene), 10 pmol of specific forward primer, 1 pmol reverse primer which has the Z sequence, 10 pmol of FAM-tagged probe (Intergen Inc.,), and cDNA from approximate 50 ng RNA. The reaction of was carried out using a IcyclerIQ (Bio-Rad) which equipment with a optic unit that allows an real time detection of 96 reactions, using the following condition: 94°C for 12 minutes, 50 cycles of 94°C for 15 seconds, 55°C for 40 seconds and 72°C for 20 seconds. The levels of the transcripts were generated from a standard that were run together with the samples.
Statistical analysis was carried out using Mann-Whitney U test and the Kruskal-Wallis test and survival analysis using Kaplan-Meier survival curve and Univariate analysis (SPSS12).
Results
Mammary tissues differentially expressed both TBXA2R and TBXAS1
Breast tissues expressed both TBXA2R and TBXAS1, although levels of TBXA2R transcripts tended to be higher than that of TBXAS1 (figure 1). When normal mammary tissues and tumour tissues were compared, TBXA2R was found to be higher in tumour tissues than in normal tissues, although this is not statistically significant (p = 0.09). The levels of TBXAS1 transcript were identical between normal and tumour tissues (figure 1 left).
Figure 1 Levels of TBXA2R (left) and TBXAS1 (right) in mammary tissues.
TBXA2R and TBXAS1 transcript levels are differentially linked to tumour differentiation
In the current study, we have observed that there was significant correlation between levels of TBXA2R and TBXAS1 and tumour grade. As revealed in figure 2 (left), grade 2 and grade 3 tumours had significantly higher levels of TBXA2R transcript (p = 0.0158 and p = 0.006 vs grade 1, respectively). Grade 3 tumour had significantly lower levels of TBXAS1 compared with grade 1 tumours (p = 0.026). The difference between grade 1 and grade 2 was not significant.
Figure 2 Levels of TBXA2R (left) and TBXAS1 (right) and their relationship with tumour grade. * p < 0.05 vs grade 1.
When levels were compared between node positive and node negative tumours, no significant difference was seen with either enzyme (5189 ± 2244 vs 13222 ± 10821 p = 0.48, and 33.2 ± 10 vs 18.5 ± 6.3, p = 0.21, for TBXA2R and TBXAS1, respectively).
The levels of both molecules were also analysed against ER status, based on a recent study [21]. As shown in table 2, ER positive tumours had significantly higher levels of TBXA2R compared with ER negative tumours (p=0.0128). No significant difference was seen with TBXAS1 in different ER status. We also compared the two major histological types of breast tumours in the current study, namely ductal and lobular tumours. As shown in table 3, no significant difference was seen between different types.
Table 2 TBXA2R and TBXAS1 levels and ER status
ER (-) ER (+) P value
TBXA2R 2286 ± 1402 21549 ± 16583 p = 0.0128
TBXAS1 16.4 ± 3.6 46.9 ± 18.5 p = 0.12
Table 3 TBXA2R and TBXAS1 levels and relationship with histological types
Ductal Lobular P value
TBXA2R 9363 ± 6542 5541 ± 5337 p = 0.67
TBXAS1 23.1 ± 6.1 22.8 ± 13.9 p = 0.34
Thromboxane synthase levels, prognosis and clinical outcomes
In the current study, we have used two methods in assessing the prognosis and clinical outcome. The first method was using a predictive indicator, the Nottingham Prognostic Index (NPI). Patients were divided into three different prognostic groups, those with predicted good prognosis (NPI1<3.4), moderate (NPI2 3.4–5.4) and poor prognosis (NPI3>5.4). As shown in table 1, patients with predicted poor prognosis had high levels of TBXA2R, however, the difference was not significant. It is noteworthy that the levels of TBXAS1 transcript were significantly lower in patients with predicted poor prognosis (table 4).
Table 4 TBXA2R and TBXAS1 levels and prognosis.
Good Moderate Poor
TBXA2R 5188 ± 2244 2561 ± 1206 (p = 0.30) 37614 ± 35470 (p = 0.38)
TBXAS1 33.2 ± 10 20.8 ± 8.2 (p = 0.086) 12.1 ± 6.7 (p = 0.008)
P values are comparison with patients with good prognosis.
The second method was to compare levels of the enzyme transcript based on the clinical outcomes of the patients, over a 10 year follow-up. This resulted in a sub division of patients into two groups:- those who remained disease free or who developed metastasis or local recurrence, and those who died of breast cancer (excluding non-cancer related deaths). As shown in table 5, although a high level of TBXA2R was seen in patients who died of breast cancer, this was not significant. No significant differences in TBXA2R and TBXAS1 transcript were seen between groups with difference clinical outcomes.
Table 5 TBXA2R and TBXAS1 levels and the clinical outcome.
Disease free With metastasis With local recurrence Patients who died of breast cancer
TBXA2R 3405 ± 1503 8660 ± 3661 (p = 0.11) 12882 ± 12873 (p = 0.52) 38611 ± 32930 (p = 0.40)
TBXAS1 22.3 ± 5.5 10.6 ± 4.3 (p = 0.10) 48 ± 47 (p = 0.62) 25.7 ± 12.4 (p = 0.81)
P values are comparison with patients who remained disease free.
Thromboxane synthase and long term survival
The value of TBXA2R and TBXAS1 in assessing long term survival was analysed using the Kaplan-Meier survival and Univariate analysis. As seen in figure 3A, high levels of TBXA2R transcript was associated with a significantly shorter disease free survival (103.8 (79.1–128.5) months vs 123.7 (112.0–135.3) month, for patients with high and low levels, respectively, p = 0.043). There was no significant correlation between TBXA2R and the overall survival (figure 3B), 111.8 (84–139.3) vs 126.1 (115.5–136.7) months, for patients with high and low TBXA2R, p = 0.43, figure 3B). The comparison for the subgroups with different levels of TBXAS1 returned no significant difference for the overall survival (p = 0.43) and disease free survival (p = 0.412).
Figure 3 TBXA2R and long term disease free (A) and overall (B) survival. High TBXA2R levels were associated with significant shorter disease free survival (A), p = 0.043.
Discussion
The current study has reported that in human breast cancer, transcript levels of thromboxane synthases are aberrant. While TBXA2R is normally expressed in tumours and particularly in aggressive tumours, TBXAS1, however, showed a very different expression pattern from that of TBXA2R. Furthermore, TBXA2R levels are associated with disease free survival.
High levels of TBXA2R and TBXA2Rs have been previously reported in various tumours including prostate, glioma, and melanoma. As a result, previous studies have shown that inhibitors/antagonists to TBXA2R are able to inhibit metastatic spread. In a liver metastasis model from colon cancer, the thromboxane synthase inhibitor, sodium ozagrel at a dose of 15 mg/kg body weight, has been shown to significantly reduce the rate of metastatic tumours in the liver, by 57% [22]. Ketoconazole has been shown to act as an antagonist to both thromboxane synthase and 5-LOX [23]. Ketoconazole has been shown to reduce lung metastasis from melanoma cells in an in vivo model. This together with the current study indicates that TBXA2R, has a potential impact, via its metabolised product TAX2, on the metastatic nature of cancer cells.
Metastasis is a complex biological and clinical phenomenon that involves a number of highly independent yet crucial steps during its development. Thromboxane synthase has been shown to influence a number of steps in the process including cell migration, angiogenesis and apoptosis. Certain eicosanoids (i.e. PGE2) may also stimulate oestrogen synthesis and contribute to the growth of breast tumours, via pathways including that of aromatase [24,25]. In addition, eicosanoids also interplay with other lipids, by way of the peroxisome proliferators activated receptors (PPAR) [26] which are known to be aberrantly expressed in breast cancer [19].
The level of expression of thromboxane synthase in glioma cells appears to correlate with the speed of migration [27]. Similarly, TBXA2R expression has been shown to influence endothelial cell migration and angiogenesis [11]. Nevertheless, anomalies exist as some inhibitors to thrombaxane synthase, such as dazmegrel, failed to impact on the growth and weight of fibrosarcoma [28,29].
The other noteworthy observation that can be derived from the current study is the expression pattern of TBXAS1, which is in clear contrast to that of TBXA2R. TBXAS1 is expressed at significantly low levels in high grade tumours (grade 3) and in patients with predicted poor clinical outcome (prognostic index NPI greater than 5.4). This indicates that metabolites generated by TBXAS1 may play a very different role in cancer to that of TBXA2R. There is very little experimental data available on this aspect. Future studies on the TBXAS1 in vitro and in vivo would be of significant interest. Finally, the current study has shown that ER positive tumours have significantly higher TBXA2R levels than ER negative tumours. Given the ongoing interest in COX-2 inhibitors and aromatase inhibitors [30,31], it would be of interest to take into consideration of TBXA2R, ER status as well as aromatase expression profile, when considering trial design and choice of therapies.
A number of factors are known to regulate the expression of TXA synthases. Factors upregulating thromboxane synthase include phorbol ester [34], Activin A [32], Benzoquinone derivatives [33] which are interesting as some of the compounds are dually active on TBXA2R and 5-LOX. Expression of thromboxane synthase is regulated by cis-elements, trans-activators and potentially by genomic methylation [35]. These early observations have an important bearing when considering the current results.
In conclusion, expression of thromboxane synthases in clinical human breast cancer is aberrant. While TBXA2R is commonly expressed in tumours and particularly in aggressive tumours, TBXAS1, in contrast, showed a very different expression pattern from that of TBXA2R. Furthermore, TBXA2R levels are associated with disease free survival, indicating that the receptor is a prognostic factor in clinical breast cancer.
List of abbreviations
12-HETE: 12-hydroxyeicosatetraenoic acid;
13-HODE: 13- hydroxyoctadecadienoic acid;
COX-2: cyclooxygenase-2;
ER: oestrogen receptor;
LOX: lipoxygenase;
PGE2: prostaglandin E2;
TBXA2R: thromboxane A2 receptor;
TBXAS1: thromboxane A synthase 1;
TXA2: thromboxane A2;
TxS: Thromboxane synthase;
Competing interests
The author(s) declare that they have no competing interest.
Authors' contributions
GW: conducting experiments, data analysis and manuscript preparation.
REM: study design, sample collection and MS preparation.
ADJ: Histological analysis.
WEJ: study design, experimental design, data analysis, MS preparation.
Acknowledgements
WGJ acknowledge Cancer Research Wales and Cancer Research UK for support.
==== Refs
Karmali RA Choi K Otter G Eicosanoids and metastasis. Experimental aspects in Lewis lung carcinoma Cancer Biochem Biophys 1986 9 97 104 3815323
Horrobin DF Nutritional and medical importance of γ-linolenic acid Prog Lipid Res 1992 31 163 94 1334266 10.1016/0163-7827(92)90008-7
Honn KV Timar J Rozhin J Bazaz R Sameni M Ziegler G Sloane BF A lipoxygenase metabolite, 12-(s)-HETE, stimulates protein-kinase c-mediated release of cathepsin-b from malignant-cells Exp Cell Res 1994 214 120 130 7521840 10.1006/excr.1994.1240
Horrobin DF Horrobin DF Essential fatty acids, lipid peroxidation, and cancer 'omega-6 essential fatty acids' 1990 Wiley-Liss, New York 351 378
Jiang WG Douglas-Jones A Mansel RE Level of expression of lipoxygenase-5, 12, and 15 and cyclooxygenase in human breast cancer Prostaglandin Leukotr Ess Fatty Acids 2003 69 275 281 10.1016/S0952-3278(03)00110-8
Nie D Che M Zacharek A Qiao Y Li L Li X Lamberti M Tang K Cai Y Guo Y Grignon D Honn KV Differential expression of thromboxane synthase in prostate carcinoma: role in tumor cell motility Am J Pathol 2004 164 429 39 14742249
Ermert L Dierkes C Ermert M Immunohistochemical expression of cyclooxygenase isoenzymes and downstream enzymes in human lung tumors Clin Cancer Res 2003 9 1604 10 12738712
Kurzel F Hagel Ch Zapf S Meissner H Westphal M Giese A Cyclo-oxygenase inhibitors and thromboxane synthase inhibitors differentially regulate migration arrest, growth inhibition and apoptosis in human glioma cells Acta Neurochir (Wien) 2002 144 71 87 11807649 10.1007/s701-002-8276-9
Pradono P Tazawa R Maemondo M Tanaka M Usui K Saijo Y Hagiwara K Nukiwa T Gene transfer of thromboxane A(2) synthase and prostaglandin I(2) synthase antithetically altered tumor angiogenesis and tumor growth Cancer Res 2002 1;62 63 6 11782360
Kleber A Hartmann RW Jose J No effect of thromboxane A2 on the attachment of tumor cell lines MDA MB 231, DU145, and U937 to the basement membrane in an in-vitro model Cancer Res Clin Oncol 2001 127 751 4
Nie D Lamberti M Zacharek A Li L Szekeres K Tang K Chen Y Honn KV Thromboxane A(2) regulation of endothelial cell migration, angiogenesis, and tumor metastasis Biochem Biophys Res Commun 2000 7;267 245 51 10623605 10.1006/bbrc.1999.1840
Bryant J Correlation between metastatic patterns in renal cell carcinoma and tissue distribution of thromboxane synthetase Acta Oncol 1994 33 708 9 7946454
Sanduja SK Mehta K Xu XM Hsu SM Sanduja R Wu KK Differentiation-associated expression of prostaglandin H and thromboxane A synthases in monocytoid leukemia cell lines Blood 1991 15;78 3178 85 1742485
Stamford IF Melhuish PB Carroll MA Corrigan CJ Patel S Bennett A Survival of mice with NC carcinoma is unchanged by drugs that are thought to inhibit thromboxane synthesis or increase prostacyclin formation Br J Cancer 1986 54 257 63 3755610
Honn KV Inhibition of tumor cell metastasis by modulation of the vascular prostacyclin/thromboxane A2 system Clin Exp Metastasis 1983 1 103 14 6242706 10.1007/BF00121490
Watkins G Douglas-Jones A Bryce R Mansel RE Jiang WG Expression of osteonectin (SPARC) in human breast cancer Prostaglandins Leukotriences and Essential Fatty Acids 2005 72 267 272 10.1016/j.plefa.2004.12.003
Jiang WG Watkins G Lane J Douglas-Jones A Cunnick GH Mokbel M Mansel RE Prognostic value of Rho familty and and rho-GDIs in breast cancer Clinical Cancer Research 2003 9 6432 6440 14695145
Jiang WG Watkins G Fodstad O Douglas-Jones A Mokbel K Mansel RE Differential expression of the CCN family members Cyr61 from CTGF and Nov in human breast cancer Endocrine Related Cancers 2004 11 781 791 10.1677/erc.1.00825
Jiang WG Douglas-Jones A Mansel RE Level of expression of PPAR-gamma and its co-activator (PPAR-GCA) in human breast cancer International Journal of Cancer 2003 106 752 757 10.1002/ijc.11302
Jiang WG Davies G Martin TA Parr C Watkins G Mason MD Mokbel K Mansel RE Molecular targeting of matrilysin and its impact on tumour growth in vivo, the potential implications in breast cancer therapy Clinical Cancer Research 2005 11 6012 6019 16115946 10.1158/1078-0432.CCR-05-0275
Jiang WG Watkins G Douglas-Jones A Mokbel K Mansel RE Fodstad O Expression of Com-1/P8 in human breast cancer and its relevance to clinical outcome and ER status International Journal of Cancer 2005 117 730 737 10.1002/ijc.21221
Yokoyama I Hayashi S Kobayashi T Negita M Yasutomi M Uchida K Takagi H Prevention of experimental hepatic metastasis with thromboxane synthase inhibitor Res Exp Med (Berl) 1995 195 209 15 8525071
Nardone PA Slotman GJ Vezeridis MP Ketoconazole: a thromboxane synthetase and 5-lipoxygenase inhibitor with antimetastatic activity in B16-F10 melanoma J Surg Res 1988 44 425 9 3361885 10.1016/0022-4804(88)90185-0
Zhao Y Agarwal VR Mendelson CR Simpson ER Estrogen biosynthesis proximal to a breast tumor is stimulated by PGE2 via cyclic AMP, leading to activation of promoter II of the CYP19 (aromatase) gene Endocrinology 1996 137 5739 42 8940410 10.1210/en.137.12.5739
Brodie AM Lu Q Long BJ Fulton A Chen T Macpherson N DeJong PC Blankenstein MA Nortier JW Slee PH van de Ven J van Gorp JM Elbers JR Schipper ME Blijham GH Thijssen JH Aromatase and COX-2 expression in human breast cancers J Steroid Biochem Mol Biol 2001 79 41 7 11850206 10.1016/S0960-0760(01)00131-5
Rubin GL Zhao Y Kalus AM Simpson ER Peroxisome proliferator-activated receptor gamma ligands inhibit estrogen biosynthesis in human breast adipose tissue: possible implications for breast cancer therapy Cancer Res 2000 60 1604 8 10749129
Giese A Hagel C Kim EL Zapf S Djawaheri J Berens ME Westphal M Thromboxane synthase regulates the migratory phenotype of human glioma cells Neuro-oncol 1999 1 3 13 11550298 10.1215/S1522851798000143
Chiabrando C Broggini M Castelli MG Cozzi E Castagnoli MN Donelli MG Garattini S Giavazzi R Fanelli R Prostaglandin and thromboxane synthesis by M5076 ovarian reticulosarcoma during growth: effects of a thromboxane synthetase inhibitor Cancer Res 1987 15;47 988 91 3100032
Davies G Martin LA Sacks N Dowsett M Cyclooxygenase-2 (COX-2), aromatase and breast cancer: a possible role for COX-2 inhibitors in breast cancer chemoprevention Ann Oncol 2002 13 669 78 12075734 10.1093/annonc/mdf125
Goss PE Breast cancer prevention – clinical trials strategies involving aromatase inhibitors J Steroid Biochem Mol Biol 2003 86 487 93 14623548 10.1016/S0960-0760(03)00378-9
Vicenzi E Lampugnani MG Bolognese Dalessandro AP Niewiarowska A de Gaetano G Donati MB Dissociation between thromboxane generation and metastatic potential in cells from a murine fibrosarcoma. Studies with a selective thromboxane synthase inhibitor Int J Cancer 1987 15;39 488 91 3557705
Yamashita T Shimizu T Ogata E Activin A/erythroid differentiation factor induces thromboxane A2 synthetic activity in murine erythroleukemia cells J Biol Chem 1991 25;266 3888 92 1995639
Ohkawa S Terao S Terashita Z Shibouta Y Nishikawa K Dual inhibitors of thromboxane A2 synthase and 5-lipoxygenase with scavenging activity of active oxygen species. Synthesis of a novel series of (3-pyridylmethyl)benzoquinone derivatives J Med Chem 1991 34 267 76 1992126 10.1021/jm00105a042
Ihara H Yokoyama C Miyata A Kosaka T Nusing R Ullrich V Tanabe T Induction of thromboxane synthase and prostaglandin endoperoxide synthase mRNAs in human erythroleukemia cells by phorbol ester FEBS Lett 1992 20;306 161 4 1633871 10.1016/0014-5793(92)80990-X
Lee KD Baek SJ Shen RF Multiple factors regulating the expression of human thromboxane synthase gene Biochem J 1996 1;319 783 91 8920981
|
16250911
|
PMC1289289
|
CC BY
|
2021-01-04 16:38:37
|
no
|
Int Semin Surg Oncol. 2005 Oct 26; 2:23
|
utf-8
|
Int Semin Surg Oncol
| 2,005 |
10.1186/1477-7800-2-23
|
oa_comm
|
==== Front
J Autoimmune DisJournal of Autoimmune Diseases1740-2557BioMed Central London 1740-2557-2-81625962210.1186/1740-2557-2-8ResearchLack of correlation between the levels of soluble cytotoxic T-lymphocyte associated antigen-4 (CTLA-4) and the CT-60 genotypes Purohit Sharad [email protected] Robert [email protected] Christin [email protected] Weipeng [email protected] Desmond [email protected] Andy [email protected] Diane [email protected] Yi-Hua [email protected] Jin-Xiong [email protected] Center for Biotechnology and Genomic Medicine, Medical College of Georgia, CA4095 Augusta, GA 309122 Department of Pediatrics, University of Florida, Gainesville FL 32607, USA2005 31 10 2005 2 8 8 16 8 2005 31 10 2005 Copyright © 2005 Purohit et al; licensee BioMed Central Ltd.2005Purohit 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
Cytotoxic T lymphocyte-associated antigen-4 (CTLA-4) plays a critical role in downregulation of antigen-activated immune response and polymorphisms at the CTLA-4 gene have been shown to be associated with several autoimmune diseases including type-1 diabetes (T1D). The etiological mutation was mapped to the CT60-A/G single nucleotide polymorphism (SNP) that is believed to control the processing and production of soluble CTLA-4 (sCTLA-4).
Methods
We therefore determined sCTLA-4 protein levels in the sera from 82 T1D patients and 19 autoantibody positive (AbP) subjects and 117 autoantibody negative (AbN) controls using ELISA. The CT-60 SNP was genotyped for these samples by using PCR and restriction enzyme digestion of a 268 bp DNA segment containing the SNP. Genotyping of CT-60 SNP was confirmed by dye terminating sequencing reaction.
Results
Higher levels of sCTLA-4 were observed in T1D (2.24 ng/ml) and AbP (mean = 2.17 ng/ml) subjects compared to AbN controls (mean = 1.69 ng/ml) with the differences between these subjects becoming significant with age (p = 0.02). However, we found no correlation between sCTLA-4 levels and the CTLA-4 CT-60 SNP genotypes.
Conclusion
Consistent with the higher serum sCTLA-4 levels observed in other autoimmune diseases, our results suggest that sCTLA-4 may be a risk factor for T1D. However, our results do not support the conclusion that the CT-60 SNP controls the expression of sCTLA-4.
==== Body
Background
Effective T cell activation requires a 'costimulation' signal that is mediated through CD28 interacting with B7 family members on antigen presenting cells (APC) [1]. The cytotoxic T lymphocyte associated antigen-4 (CTLA-4) was initially described as a B7 binding protein and a receptor expressed on the surface of activated T cells [2]. It belongs to the immunoglobulin gene superfamily and shares homology with CD28. CTLA-4 has been reported to be an important negative regulator of autoimmune diseases [3,4]. CTLA-4 blockade enhances T cell responses in vitro and in vivo [5,6], augments antitumor immunity [7] and exacerbates autoimmune diseases [8]. Several reports have indicated that CTLA-4 deficient mice show a severe lymphoproliferative disorder and autoimmune disease with early lethality [9,10]. Treatment with anti-CTLA-4 mAb of BDC2.5/NOD mice provoked a rapid onset of diabetes, indicating that a higher CTLA-4 presence was required for suppression of autoimmune phenomenon in these mice [11,12]. Recently, a soluble form of CTLA-4 (sCTLA-4) was found to be expressed constitutively by unstimulated human T cells [13]. Circulating sCTLA-4 protein was found to be present in human serum and is shown to possess an inhibitory effect on mixed leucocyte response [14].
Several studies have demonstrated a genetic association between polymorphisms within or near the CTLA-4 gene and T1D [15-19] as well as other autoimmune diseases [20-24]. This susceptibility locus has been recognized as IDDM12. Our previous studies indicated that CTLA-4 was the only gene contained in the IDDM12 susceptibility interval, suggesting that CTLA-4 is indeed the IDDM12 gene [16]. In a recent report by Ueda et al [25] the susceptibility interval was further narrowed to a 6.1 kb region at the 3' UTR of the CTLA-4 gene and the CT60-A/G single nucleotide polymorphism (SNP) was suggested to be the etiological mutation. The susceptible CT60-G allele was reported to produce a lower amount of soluble CTLA-4 mRNA in the peripheral blood lymphocytes than the disease resistant CT60-A allele. These results suggested that sCTLA-4 may confer protective effect against T1D. If this effect is indeed true, one would predict lower sCTLA-4 in the serum in T1D patients compared to controls. However, the prediction is in direct conflict with the observations in other autoimmune diseases including autoimmune thyroid disease [26], systemic lupus erythematosus [27] and myasthenia gravis [28], in which the serum sCTLA-4 levels are increased in patients compared to controls. The measurement of serum sCTLA-4 protein in a larger sample set is vital in evaluating the potential role of sCTLA-4 in T1D, and to better understand the molecular and functional basis underlying the genetic association between the CTLA-4 gene and T1D.
Methods
Patient sera
The study population consists of 218 subjects from the South-eastern United States. All study subjects were genotyped for HLA-DQB1 and evaluated for three autoantibodies (IA-2A, GADA and IAA) using established methods [29,30] Subjects used in this study are participants in the prospective assessment in newborns for diabetes autoimmunity (PANDA) program. Briefly, PANDA screens newborns from the general population as well as children with a first degree relative with T1D using HLA genotyping. Those subjects with high risk genes are monitored for the appearance of islet autoantibodies and clinical diabetes. Therefore, most of the autoantibody-negative (AbN) subjects also have high risk HLA genes and the AbN group is not randomly selected from the general population. The autoantibody-positive (AbP) subjects have been tested persistently positive for two or more islet autoantibodies. Based on our results from PANDA and previous studies, the AbP group has 70–80% of chance to progress to T1D [31] and indeed represent a very high risk group. Since autoantibody production is one of the hallmarks of autoimmunity, the AbP and T1D group can be combined to assess the impact of autoimmunity on the CTLA-4 levels. Appropriate institutional review boards approved the study design and informed consent was obtained from all subjects.
Assay of sCTLA-4
A sandwich ELISA assay as described by Oaks et al [26] was used to measure the serum sCTLA-4 levels in a total of 218 subjects, including 117 autoantibody-negative (AbN), 19 autoantibody-positive (AbP) and 84 patients with T1D. The 96-well microtiter plates (Pierce Biotechnology, Rockford, IL) were coated with 1.0 ug/ml anti-CTLA-4 monoclonal antibody (clone BNI3; Pharmingen, San Deigo, CA). After blocking, 100 ul of 1:10 diluted serum samples were added to each well and the plates were incubated for 2 hr in a humid chamber at 37°C and then washed to remove unbound material. After washing, 100 ul biotinylated anti-CTLA-4 mAb (1.0 ug/ml, clone AS-33, Antibody Solutions, Palo Alto, CA) was added and the reactions were incubated for another 1 hr at 37°C in a humid chamber. Reactions were developed using streptavidin-peroxidase complex (Biorad, Hercules, CA) and 3,3',5,5'-tetramethy benzidine substrate (Sigma, St. Louis, MO) for 10 min at room temperature, the reaction was terminated with 2N H2SO4 and optical density was read at 450 nm and 630 nm. A standard curve was generated using a dilution series of commercially available CTLA-4-Ig fusion protein (0.125 ng/ml to 10 ng/ml Ancell, Bayport, MN). This assay has a linear range between 0.5 and 10 ng/ml and the vast majority of the samples fall under this range. Each sample was analyzed in duplicate.
Genotyping of 3' untranslated region of CTLA-4 gene
A fragment of 268 bp encompassing the CTLA-4 C/T-60 single nucleotide polymorphism (SNP) in the 3' untranslated region was amplified using the forward primer 5'GCTTCATGAGTCAGCTTTGC3' and reverse primer 5'ATAGGACCACAGGT3'. The amplified PCR products were digested using the 10 units of HpyCH4IV (New England Biolabs) and separated on 3% agarose gels. The C-60 allele yielded two bands of 151 and 103 bp and the T-60 allele yields a band of 268 bp. The genotyping technique for the C/T-60 SNP was further confirmed by DNA sequencing of a subset of samples using a 300XL DNA sequencer (ABI Sciex).
Statistical Analysis
Absorbance values obtained at 450 nm were normalized with the absorbance values at 630 nm. sCTLA-4 levels were log-transformed prior to analysis. Two of the T1D subjects had very high sCTLA4 levels with serum CTLA4 levels were > 12 ng/ml, or 3.5 standard deviations from the mean of 2.57 ng/ml. Further, initial analyses involving analysis of variance (ANOVA) indicated these two subjects' values were outliers. As such, the data for these two subjects were removed from all subsequent analyses. We used linear-mixed model ANOVA (Proc Mixed procedure of SAS) in which plate was included as a random effect to examine differences in sCTLA4 levels. Initially we analyzed phenotypic group (AbN, AbP, and T1D) alone, but subsequently conducted separate analyses adding other factors. These analyses were as follows: (1) factorial ANOVA with phenotypic group and CTLA-4 genotype; (2) factorial ANOVA with phenotypic group and gender; (3) ANOVA with phenotypic group and age as a covariate, and the interaction between age and phenotypic group; (4) ANOVA with phenotypic group and duration of T1D as a covariate, and the interaction between duration of T1D and phenotypic group; and (5) factorial ANOVA with phenotypic group and HLA genotype.
Results
Clinical and demographic information is presented in Table 1. A majority of the subjects in the study were below the age of 20 years (73%) in all three groups. The subjects were tested for IAA, GADA and IA-2A autoantibodies as well as HLA-DQB1 genotypes. HLA-DQB1 genotyping information was available for 96.58, 94.74 and 98.78 percent of patients from the AbN, AbP and T1D groups, respectively. Eighty-eight percent of T1D subjects were diagnosed with T1D before the age of 20, with an average age of 8.8 (range 0.9–41.2). Fifty-seven out of eighty-two T1'1D subjects have a T1D duration of five years and less.
Table 1 Clinical and Demographic characteristics of subjects involved in the study with respect to number of individuals, age of diagnosis, duration of disease, genotype and antibody number.
Variable AbN AbP T1D
Total number 117 19 82
Age (years) 15.1 14.5 15.8
75%ile 22.0 28.5 20.1
90%ile 37.0 37.0 36.0
Age range (0.8–48.6) (0.6–44.0) (0.9–43.7)
Age of Diagnosis (years) 8.8 (0.8–41.2)
Duration of T1D (years) 6.2 (0–41.0)
Age: n (percent)
<20 years 86 (73.51) 14 (73.68) 60 (73.17)
>20 years 31 (26.49) 5 (26.32) 22 (26.83)
Sex: n (percent)
Male 53 (45.30) 9 (47.37) 50 (60.97)
Female 64 (54.70) 10 (52.63) 32 (39.03)
Genotype: n (percent)*
0201/0201 8 (6.84) 0 (0.00) 13 (15.85)
0302/0302 12 (10.26) 4 (21.05) 6 (7.32)
0201/0302 31 (26.50) 5 (26.32) 38 (46.34)
0201/x 21 (17.95) 3 (15.79) 10 (12.20)
0302/x 23 (19.66) 4 (21.05) 11 (13.41)
x/x 18 (15.38) 2 (10.53) 3 (3.66)
Age, age of diagnosis and duration of disease is presented as means (range).
*Genotyping information was not available on one AbP, one T1D and 4 AbN individual(s).
Sex and genotype are presented as number of individuals of individuals (n). Values presented in parenthesis are percentage of total.
We first compared the serum sCTLA-4 levels between the three phenotypic groups (i.e., AbN, AbP and T1D). The protein levels for T1D (mean = 2.24 ng/ml, range = 0–10.1 ng/ml) and AbP (mean = 2.17 ng/ml, range = 0.2–7.7 ng/ml) were slightly higher than that in AbN (mean = 1.69 ng/ml, range = 0.0–11.5 ng/ml) (Table 2), although these differences were not statistically significant.
Table 2 sCTLA-4 levels in AbN, T1D and AbP individuals. Values presented are mean and 95% confidence interval in ng/ml.
Group AbN AbP T1D T1D/AbN AbP/AbN
All subjects (p = 0.58) 1.7 (0.6–3.6) (n = 117) 2.2 (0.5–5.5) (n = 19) 2.2 (0.9–4.6) (n = 82) 1.3 1.3
CTLA-4 genotype
A/A 1.9 (0.7–4.1) (n = 12) 2.5 (n = 1) 1.9 (0.8–4.9) (n = 19) 1.0 1.2
A/G 1.5 (0.5–4.0) (n = 21) 1.7 (0.4–4.2) (n = 7) 1.9 (0.7–3.8) (n = 16) 1.2 1.1
G/G 1.6 (0.7–3.0) (n = 49) 2.5 (0.9–5.4) (n = 9) 2.4 (1.1–4.4) (n = 26) 1.3 1.3
Male (p = 0.53) 1.7 (0.9–2.8) (n = 53) 2.3 (0.9–4.5) (n = 9) 2.3 (1.3–3.7) (n = 51) 1.4 1.3
Female (p = 0.70) 1.7 (0.9–2.8) (n = 64) 2.1 (0.8–4.2) (n = 10) 2.2 (1.2–3.6) (n = 31) 1.3 1.2
DQB1*0201 positive 1.3 (0.6–2.4) (n = 60) 1.8 (0.6–4.0) (n = 8) 2.2 (1.1–3.8) (n = 62) 1.7 1.4
DQB1*0201 negative 2.2 (1.2–3.8) (n = 53) 3.3 (1.4–6.8) (n = 10) 1.9 (0.8–3.8) (n = 19) 0.9 1.5
DQB1*0302 positive 1.6 (0.8–2.9) (n = 66) 2.4 (1.0–4.6) (n = 13) 2.3 (1.2–3.9) (n = 56) 1.4 1.5
DQB1*0302 negative 1.9 (0.9–3.3) (n = 47) 2.6 (0.9–5.9) (n = 5) 1.8 (0.7–3.6) (n = 25) 1.0 1.4
The serum CTLA-4 levels were analyzed after stratification by phenotypic groups (T1D, AbP and AbN) and the CTLA-4 CT-60 SNP genotypes (A/A, A/G and G/G) (Table 2). A mixed model ANOVA using phenotypic group and CT-60 genotypes as factorial fixed effects revealed no differences in sCTLA-4 levels between CTLA-4 genotypes (p = 0.46) or genotype/phenotype interactions (p = 0.82). A similar ANOVA using CT-60 genotypes alone as a fixed effect did not reveal any significant differences in sCTLA-4 levels between CTLA-4 genotypes (p = 0.64).
We then analyzed the data after conditioning on genetic, phenotypic or demographic parameters. Neither gender showed differences in serum CTLA-4 levels between the three phenotypic groups (Table 2). The relationship between sCTLA-4 levels and age differed between the three phenotypic groups (p = 0.022). The sCTLA-4 levels decreased with age in the controls (p = 0.048; Fig. 1). In contrast, sCTLA-4 levels increased with age in both the T1D and AbP groups (Fig. 1), although these relationships were not significant (p > 0.1). This difference in the relationship with age will result in AbN controls having lower sCTLA-4 levels at later ages compared with both AbP and T1D subjects. Serum sCTLA-4 levels in T1D subjects did not show an association with duration of disease (p = 0.4) nor with the age at disease onset (p = 0.6; data not shown).
Figure 1 Relationship of sCTLA-4 with age and phenotypic groups. The lines shown are the estimated line based on the mixed linear model.
The serum CTLA-4 levels were also analyzed after conditioning on the HLA-DQB1 genotypes by using phenotypic group, HLA-DQB1*201, and HLA-DQB1*302 as factorial fixed effects in a mixed model ANOVA. No differences were observed in sCTLA-4 levels between HLA-DQB1*0302 genotypes (p = 0.96), and the three phenotypic groups stratified by HLA-DQB1 genotype did not show any differences (Table 2, p = 0.51). AbN subjects with the DQB1*201 allele tended to have lower sCTLA levels (1.3 ng/ml vs 2.2 ng/ml), although the difference was not significant, a similar trend was observed in AbP group (1.8 ng/ml vs 3.3 ng/ml). The T1D group subjects with and without DQB1*201 allele have a very similar levels of sCTLA-4 (2.2 ng/ml vs 1.9 ng/ml). The differences between the three phenotypic groups for subjects with the DQB1*201 allele were not significantly different from the differences observed for those without the DQB1*201 allele (Table 2; p = 0.13). The main effect of the DQB1*0201 allele in the factorial mixed-model ANOVA was marginally significant (p = 0.08) with serum CTLA-4 levels being lower in individuals with a DQB1*0201 allele (mean = 1.8 ng/ml) than individuals without a DQB1*0201 allele (mean = 2.4 ng/ml). We decided to redo this analysis by combining the AbP and T1D groups for three reasons: (1) sample sizes were small for some of the genotype/phenotype combinations; (2) subjects that are positive for multiple antibodies and with a high risk HLA genotype are much more likely to develop T1D in future; and (3) autoimmunity is the common denominator of the AbP and T1D groups. When the data were analyzed with these two phenotypic groups considered as a single group (AbP + T1D vs. AbN), the main effect of DQB1*0201 allele became significant (p = 0.02), with the remaining effects still not significant. T1D and AbP subjects did show a trend towards having larger sCTLA-4 levels (mean = 2.2 ng/ml) compared to the AbN subjects (mean = 1.3 ng/ml) when only subjects with the DQB1*0201 were considered, however, the difference was not statistically significant (p = 0.15).
Discussion
Type-1 diabetes is marked by the production of pancreatic islet β cell-specific auotantibodies and destruction of the insulin-producing β cells by autoreactive T cells. A role of CTLA-4 in the pathogenesis in T1D and other autoimmune diseases has been well documented. In this study we provide some suggestive evidence that high risk autoantibody positive subjects and T1D patients both have increased levels of sCTLA-4 in serum compared to autoantibody-negative subjects. We observed that larger differences in sCTLA4 levels between T1D/AbP and AbN subjects occur in the older age group. Further, we observed a difference between AbP/T1D subjects and AbN subjects for those carrying the DQB1*0201 allele.
As a negative regulator of T cell activation, blockade of CTLA-4 by monoclonal anti-CTLA-4 antibody provokes a rapid onset of diabetes in BDC2.5/NOD mouse model [11]. Treatment of animals with recombinant CTLA-4Ig molecule delays the onset of T1D and other autoimmune diseases [11,12,32-36]. However, the function and potential role of sCTLA-4 have not been well studied. sCTLA-4 is generated by alternative splicing of CTLA-4 mRNA, which induces a frame shift by deletion of a transmembrane region of CTLA-4 resulting in a native soluble protein [13]. sCTLA-4 is constitutively expressed on nonstimulated T cells and its expression is downregulated after T cell activation [14]. The soluble form of surface proteins is believed, in most cases, to play an inhibitory role due to competition for ligands with their surface counterparts. The finding that the sCTLA-4 expression level remains at sustained levels suggests that sCTLA-4 blocks the B7-mCTLA-4 interaction, thereby enhancing T-cell activation and autoreactivity by inhibiting the induction of anergy [37,38]. Alternatively, sustained sCTLA-4 levels may play a protective role via inhibition of the B7-CD28 interactions. Therefore, the role of sCTLA-4 in autoimmunity may depend on the relative binding affinity of sCTLA-4 to B7.1 and B7.2. This question was indirectly addressed in several autoimmune diseases by comparing the sCTLA-4 levels in the serum of patients and controls. Elevated sCTLA-4 has been reported in organ specific autoimmune thyroid disease [26], systemic lupus erythematosus [27] and myasthenia gravis [28]. These observations suggest that sCTLA-4 may contribute to the development of autoimmune diseases, probably through inhibiting the B7-mCTLA-4 interaction and down-regulation of T cell activation.
We are unaware of any study that has examined sCTLA-4 levels in the serum of T1D patients. A recent study by Ueda et al. [25], suggested that a SNP (CT60-A/G) in the 3' UTR of the CTLA-4 gene may determine the efficiency of the splicing and production of sCTLA-4 mRNA. Based on a small number of subjects, the susceptible G allele was suggested to produce lower amounts of sCTLA-4 mRNA. Based on these observations, the authors concluded that sCTLA-4 expression is the functional basis for the observed genetic association between T1D and the CTLA-4 gene. If this conclusion were correct, T1D patients would be expected to have lower serum CTLA-4 levels. Our results found no evidence to indicate that sCTLA-4 levels are decreased in patients compared to controls. In contrast, our data suggested that the serum sCTLA-4 levels were slightly higher in T1D patients. Our results also indicated that the increased sCTLA-4 levels in T1D patients were not due to the hyperglycemic conditions because the autoantibody positive subjects also had increased serum sCTLA-4 levels. We also directly tested the correlation between sCTLA-4 levels and the CT-60 SNP in all three phenotypic groups (AbN, AbP and T1D) and found no significant correlation in any of these groups. The discrepancies between our study and the previous report [25] may be explained by a number of factors. First, mRNA was studied in the previous report, while serum protein was analyzed in this study. As the biological function of sCTLA-4 is carried out at the protein level, our data is more applicable to the role of sCTLA-4 in T1D pathogenesis. Second, the sample size in the previous report [25] was extremely small and random variation is quite likely. Although the sample size in our study is not very large, it is several times larger than the previous report and is sufficiently powered to detect large differences. As there is no indication of a correlation between sCTLA-4 levels and the CT-60 genotype, it is unlikely that the C/T-60 SNP plays a major role in controlling the expression of sCTLA-4.
Conclusion
Consistent with the observations in other autoimmune diseases in humans as well as data in the NOD mice, our data suggest that sCTLA-4 is potentially a risk factor for the development of T1D. Our data also raises a serious doubt about the conclusion that the expression of sCTLA-4 is controlled by the CT60 SNP in the 3' end of the CTLA-4 gene. Therefore, the functional basis for the genetic association between CTLA-4 and autoimmune diseases as well as the etiological mutation in the CTLA-4 region should be re-considered.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SP and JXS designed the studies, helped with the interpretation and the writing of the manuscript. SP, CC and WZ were primarily involved in carrying out the clinical assessments and the acquisition of data. RP performed the statistical analyses and was involved in preparing the manuscript. AM and DS were responsible for collecting clinical samples and patient evaluation. DH and YH were involved in sample and data collection.
Acknowledgements
This work was supported by grants from the National Institute of Child Health and Development (2RO1HD37800 and 1R21HD050196) to JXS. SP was supported by a JDRF postdoctoral fellowship (JDRF #3-2004-195).
==== Refs
Lenschow DJ Walunas TL Bluestone JA CD28/B7 system of T cell costimulation Annu Rev Immunol 1996 14 233 258 8717514 10.1146/annurev.immunol.14.1.233
Linsley PS Brady W Urnes M Grosmaire LS Damle NK Ledbetter JA CTLA-4 is a second receptor for the B cell activation antigen B7 J Exp Med 1991 174 561 569 1714933 10.1084/jem.174.3.561
Linsley PS Brady W Grosmaire L Aruffo A Damle NK Ledbetter JA Binding of the B cell activation antigen B7 to CD28 costimulates T cell proliferation and interleukin 2 mRNA accumulation J Exp Med 1991 173 721 730 1847722 10.1084/jem.173.3.721
Brunet JF Denizot F Luciani MF Roux-Dosseto M Suzan M Mattei MG Golstein P A new member of the immunoglobulin superfamily – CTLA-4 Nature 1987 328 267 270 3496540 10.1038/328267a0
Walunas TL Bakker CY Bluestone JA CTLA-4 ligation blocks CD28-dependent T cell activation J Exp Med 1996 183 2541 2550 8676075 10.1084/jem.183.6.2541
Krummel MF Allison JP CD28 and CTLA-4 have opposing effects on the response of T cells to stimulation J Exp Med 1995 182 459 465 7543139 10.1084/jem.182.2.459
Leach DR Krummel MF Allison JP Enhancement of antitumor immunity by CTLA-4 blockade Science 1996 271 1734 1736 8596936
Tivol EA Borriello F Schweitzer AN Lynch WP Bluestone JA Sharpe AH Loss of CTLA-4 leads to massive lymphoproliferation and fatal multiorgan tissue destruction, revealing a critical negative regulatory role of CTLA-4 Immunity 1995 3 541 547 7584144 10.1016/1074-7613(95)90125-6
Karandikar NJ Vanderlugt CL Walunas TL Miller SD Bluestone JA CTLA-4: a negative regulator of autoimmune disease J Exp Med 1996 184 783 788 8760834 10.1084/jem.184.2.783
Waterhouse P Penninger JM Timms E Wakeham A Shahinian A Lee KP Thompson CB Griesser H Mak TW Lymphoproliferative disorders with early lethality in mice deficient in Ctla-4 Science 1995 270 985 988 7481803
Luhder F Hoglund P Allison JP Benoist C Mathis D Cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) regulates the unfolding of autoimmune diabetes J Exp Med 1998 187 427 432 9449722 10.1084/jem.187.3.427
Luhder F Chambers C Allison JP Benoist C Mathis D Pinpointing when T cell costimulatory receptor CTLA-4 must be engaged to dampen diabetogenic T cells Proc Natl Acad Sci USA 2000 97 12204 12209 11035773 10.1073/pnas.200348397
Magistrelli G Jeannin P Herbault N Benoit DC Gauchat JF Bonnefoy JY Delneste Y A soluble form of CTLA-4 generated by alternative splicing is expressed by nonstimulated human T cells Eur J Immunol 1999 29 3596 3602 10556814 10.1002/(SICI)1521-4141(199911)29:11<3596::AID-IMMU3596>3.3.CO;2-P
Oaks MK Hallett KM Penwell RT Stauber EC Warren SJ Tector AJ A native soluble form of CTLA-4 Cell Immunol 2000 201 144 153 10831323 10.1006/cimm.2000.1649
Marron MP Raffel LJ Garchon HJ Jacob CO Serrano-Rios M Martinez Larrad MT Teng WP Park Y Zhang ZX Goldstein DR Tao YW Beaurain G Bach JF Huang HS Luo DF Zeidler A Rotter JI Yang MC Modilevsky T Maclaren NK She JX Insulin-dependent diabetes mellitus (IDDM) is associated with CTLA4 polymorphisms in multiple ethnic groups Hum Mol Genet 1997 6 1275 1282 9259273 10.1093/hmg/6.8.1275
Marron MP Zeidler A Raffel LJ Eckenrode SE Yang JJ Hopkins DI Garchon HJ Jacob CO Serrano-Rios M Martinez Larrad MT Park Y Bach JF Rotter JI Yang MC She JX Genetic and physical mapping of a type 1 diabetes susceptibility gene (IDDM12) to a 100-kb phagemid artificial chromosome clone containing D2S72-CTLA4-D2S105 on chromosome 2q33 Diabetes 2000 49 492 499 10868973
Nistico L Buzzetti R Pritchard LE Van der AB Giovannini C Bosi E Larrad MT Rios MS Chow CC Cockram CS Jacobs K Mijovic C Bain SC Barnett AH Vandewalle CL Schuit F Gorus FK Tosi R Pozzilli P Todd JA The CTLA-4 gene region of chromosome 2q33 is linked to, and associated with, type 1 diabetes. Belgian Diabetes Registry Hum Mol Genet 1996 5 1075 1080 8817351 10.1093/hmg/5.7.1075
Donner H Rau H Walfish PG Braun J Siegmund T Finke R Herwig J Usadel KH Badenhoop K CTLA4 alanine-17 confers genetic susceptibility to Graves' disease and to type 1 diabetes mellitus J Clin Endocrinol Metab 1997 82 143 146 8989248 10.1210/jc.82.1.143
Van der Auwera BJ Vandewalle CL Schuit FC Winnock F De Leeuw IH Van Imschoot S Lamberigts G Gorus FK CTLA-4 gene polymorphism confers susceptibility to insulin-dependent diabetes mellitus (IDDM) independently from age and from other genetic or immune disease markers. The Belgian Diabetes Registry Clin Exp Immunol 1997 110 98 103 9353155 10.1111/j.1365-2249.1997.t01-1-512-ce1410.x
Yanagawa T Hidaka Y Guimaraes V Soliman M DeGroot LJ CTLA-4 gene polymorphism associated with Graves' disease in a Caucasian population J Clin Endocrinol Metab 1995 80 41 45 7829637 10.1210/jc.80.1.41
Kemp EH Ajjan RA Husebye ES Peterson P Uibo R Imrie H Pearce SH Watson PF Weetman AP A cytotoxic T lymphocyte antigen-4 (CTLA-4) gene polymorphism is associated with autoimmune Addison's disease in English patients Clin Endocrinol 1998 49 609 613 10.1046/j.1365-2265.1998.00579.x
Kotsa K Watson PF Weetman AP A CTLA-4 gene polymorphism is associated with both Graves disease and autoimmune hypothyroidism Clin Endocrinol 1997 46 551 554 10.1046/j.1365-2265.1997.1710996.x
Huang D Giscombe R Zhou Y Lefvert AK Polymorphisms in CTLA-4 but not tumor necrosis factor-alpha or interleukin 1beta genes are associated with Wegener's granulomatosis J Rheumatol 2000 27 397 401 10685804
Huang D Liu L Noren K Xia SQ Trifunovic J Pirskanen R Lefvert AK Genetic association of Ctla-4 to myasthenia gravis with thymoma J Neuroimmunol 1998 88 192 198 9688341 10.1016/S0165-5728(98)00119-2
Ueda H Howson JM Esposito L Heward J Snook H Chamberlain G Rainbow DB Hunter KM Smith AN Di Genova G Herr MH Dahlman I Payne F Smyth D Lowe C Twells RC Howlett S Healy B Nutland S Rance HE Everett V Smink LJ Lam AC Cordell HJ Walker NM Bordin C Hulme J Motzo C Cucca F Hess JF Metzker ML Rogers J Gregory S Allahabadia A Nithiyananthan R Tuomilehto-Wolf E Tuomilehto J Bingley P Gillespie KM Undlien DE Ronningen KS Guja C Ionescu-Tirgoviste C Savage DA Maxwell AP Carson DJ Patterson CC Franklyn JA Clayton DG Peterson LB Wicker LS Todd JA Gough SC Association of the T-cell regulatory gene CTLA4 with susceptibility to autoimmune disease Nature 2003 423 506 511 12724780 10.1038/nature01621
Oaks MK Hallett KM Cutting edge: a soluble form of CTLA-4 in patients with autoimmune thyroid disease J Immunol 2000 164 5015 5018 10799854
Liu MF Wang CR Chen PC Fung LL Increased expression of soluble cytotoxic T-lymphocyte-associated antigen-4 molecule in patients with systemic lupus erythematosus Scand J Immunol 2003 57 568 572 12791095 10.1046/j.1365-3083.2003.01232.x
Wang XB Kakoulidou M Giscombe R Qiu Q Huang D Pirskanen R Lefvert AK Abnormal expression of CTLA-4 by T cells from patients with myasthenia gravis: effect of an AT-rich gene sequence J Neuroimmunol 2002 130 224 232 12225905 10.1016/S0165-5728(02)00228-X
She JX Bui MM Tian XH Muir A Wakeland EK Zorovich B Zhang LP Liu MC Thomson G Maclaren NK Additive susceptibility to insulin-dependent diabetes conferred by HLA-DQB1 and insulin genes Autoimmunity 1994 18 195 203 7858104
Wasserfall C Atkinson M Jodoin E Schatz D She JX Henderson LO Ellis T Glutamic acid decarboxylase and IA-2 autoantibodies in type 1 diabetes: comparing sample substrates for autoantibody determinations Pediatr Diabetes 2000 1 10 16 15016237 10.1034/j.1399-5448.2000.010103.x
Bingley PJ Bonifacio E Williams AJ Genovese S Bottazzo GF Gale EA Prediction of IDDM in the general population: strategies based on combinations of autoantibody markers Diabetes 1997 46 1701 1710 9356015
McIntosh KR Linsley PS Bacha PA Drachman DB Immunotherapy of experimental autoimmune myasthenia gravis: selective effects of CTLA4Ig and synergistic combination with an IL2-diphtheria toxin fusion protein J Neuroimmunol 1998 87 136 146 9670855 10.1016/S0165-5728(98)00071-X
Davidson A Wang X Mihara M Ramanujam M Huang W Schiffer L Sinha J Co-stimulatory blockade in the treatment of murine systemic lupus erythematosus (SLE) Ann N Y Acad Sci 2003 987 188 98 12727639
Verwaerde C Naud MC Delanoye A Wood M Thillaye-Goldenberg B Auriault C de Kozak Y Ocular transfer of retinal glial cells transduced ex vivo with adenovirus expressing viral IL-10 or CTLA4-Ig inhibits experimental autoimmune uveoretinitis Gene Ther 2003 10 1970 1981 14528321 10.1038/sj.gt.3302101
Kremer JM Westhovens R Leon M Di Giorgio E Alten R Steinfeld S Russell A Dougados M Emery P Nuamah IF Williams GR Becker JC Hagerty DT Moreland LW Treatment of rheumatoid arthritis by selective inhibition of T-cell activation with fusion protein CTLA4Ig N Engl J Med 2003 349 1907 1915 14614165 10.1056/NEJMoa035075
Najafian N Sayegh MH CTLA4-Ig: a novel immunosuppressive agent Expert Opin Investig Drugs 2000 9 2147 2157 not cited in text 10.1517/13543784.9.9.2147
Carreno BM Bennett F Chau TA Ling V Luxenberg D Jussif J Baroja ML Madrenas J CTLA-4 (CD152) can inhibit T cell activation by two different mechanisms depending on its level of cell surface expression J Immunol 2000 165 1352 1356 10903737
Pandiyan P Gartner D Soezeri O Radbruch A Schulze-Osthoff K Brunner-Weinzierl MC CD152 (CTLA-4) determines the unequal resistance of Th1 and Th2 cells against activation-induced cell Death by a mechanism requiring PI3 kinase J Exp Med 2004 199 831 842 15007096 10.1084/jem.20031058
|
16259622
|
PMC1289290
|
CC BY
|
2021-01-04 16:38:03
|
no
|
J Autoimmune Dis. 2005 Oct 31; 2:8
|
utf-8
|
J Autoimmune Dis
| 2,005 |
10.1186/1740-2557-2-8
|
oa_comm
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.