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
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0044316002358PerspectivesCorrespondenceOtotoxicity Fechter Laurence D. Pouyatos Benoit Loma Linda VA Medical Center, Loma Linda, California, E-mail:
[email protected] authors declare they have no competing financial interests.
7 2005 113 7 A443 A444 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
The January 2005 issue of EHP provided a much-needed overview of the prevalence of environment noise and its effects on health (Chepesiuk 2005; Manuel 2005; Schmidt 2005). Indeed, noise is pervasive and its adverse health effects are among the most common occupational injuries. Your consideration of noise-induced damage is especially welcome, given the strong focus of EHP on overexposure to chemical agents relative to overexposure to physical stimuli. Curiously absent from the discussion, however, was a review of the evidence that has accumulated over the past two decades concerning the ability of chemical agents to produce hearing impairment directly (ototoxicity) and to interact with noise exposure yielding either additive or synergistic impairment of the auditory apparatus. Research on such processes has received support in the United States from multiple agencies, including the National Institute of Environmental Health Sciences, the National Institute for Deafness and Other Communication Disorders, the National Institute for Occupational Safety and Health, and the U.S. Environmental Protection Agency.
Occupational epidemiology studies have demonstrated noise–chemical interactions in the workplace, and laboratory animal models have been effective in identifying ototoxicants, establishing dosimetry, identifying targets of toxicity, and determining the mechanisms for such ototoxicity. For example, occupational epidemiologic studies of Morata et al. (1997) demonstrated an excess risk of developing hearing loss among workers exposed to mixed solvents (mainly toluene) plus noise among printers compared with noise-exposed referent subjects or non-exposed matched controls. Similar studies have subsequently been published for styrene-exposed workers in the reinforced plastic industry (Morata et al. 2002; Sliwinska-Kowalska et al. 2003).
In laboratory animals, the pioneer experiments on the ototoxicity of solvents were initiated by Pryor and Rebert in the 1980s (e.g., Pryor et al. 1987; Rebert et al. 1983). Since these early studies, the ability of chemicals to directly disrupt auditory function has been established for trichloroethylene (Crofton et al. 1993; Fechter et al. 1998), toluene (Campo et al. 1999; Crofton et al. 1994; Johnson 1993), ethyl benzene (Cappaert et al. 2001), and styrene (Campo et al. 2001), among other agents. In addition, Lataye et al. (2001, 2003) have nicely identified the route by which solvents enter the cochlea and the pattern of damage that they produce in the inner ear.
Using developmental models, Rice and Gilbert (1992) demonstrated that methyl mercury exposure could impair auditory function in young primates. Also, hearing impairments have been reported for lead-exposed children (Osman et al. 1999; Schwartz and Otto 1987, 1991). Crofton and colleagues (Crofton et al. 1999, 2000; Lasky et al. 2002) demonstrated the ability of polychlorinated biphenyls to disrupt the development of the cochlea in rats by disrupting thyroid function.
In this laboratory, we have demonstrated that a series of chemical contaminants with potential to disrupt intrinsic antioxidant pathways or to enhance reactive oxygen species (ROS) generation can produce permanent hearing loss in the presence of noise. These agents include carbon monoxide (Fechter et al. 1987, 1988, 2000), hydrogen cyanide (Fechter et al. 2002), and acrylonitrile (Fechter et al. 2003; Pouyatos et al. 2005). This research provided evidence that intense noise can initiate ROS generation, resulting in cochlear damage. We hypothesized that even moderate noise levels, including noise close to permissible workplace exposure levels, may initiate ROS formation but that these are normally contained by antioxidant pathways. However, in the presence of pro-oxidant chemical agents, we demonstrated that even mild noise can yield oxidative stress leading to the death of sensory receptor cells for sound, the outer hair cells, and subsequent permanent impairment of auditory function (Fechter et al. 2000, 2002, 2003; Pouyatos et al. 2005). It is striking, although not surprising, that the auditory system is vulnerable to a range of chemical agents that initiate toxic processes that have been more fully studied in the brain and other organ systems.
The existing evidence has clear implications for both environmental and occupational health, and it highlights the continuing need for research on the issue. In Europe, the scientific information available has influenced public health policy. In February 2003, the European Parliament and the Council of the European Union (2003) published Directive 2003/10/EC on minimum safety requirements regarding the exposure of workers to noise. Ultimately, an increase in the awareness of the ototoxic potential of chemicals should improve preventive efforts and help reduce the risk of hearing loss.
==== Refs
References
Campo P Lataye R Loquet G Bonnet P 2001 Styrene-induced hearing loss: a membrane insult Hear Res 154 1-2 170 180 11423228
Campo P Loquet G Blachere V Roure M 1999 Toluene and styrene intoxication route in the rat cochlea Neurotoxicol Teratol 21 4 427 434 10440486
Cappaert NL Klis SF Muijser H Kulig BM Smoorenburg GF 2001 Simultaneous exposure to ethyl benzene and noise: synergistic effects on outer hair cells Hear Res 162 1-2 67 79 11707353
Chepesiuk R 2005 Decibel hell: the effects of living in a noisy world Environ Health Perspect 113 A35 A41
Crofton KM Ding D Padich R Taylor M Henderson D 2000 Hearing loss following exposure during development to polychlorinated biphenyls: a cochlear site of action Hear Res 144 1-2 196 204 10831878
Crofton KM Rebert CS Lassiter TL 1994 Solvent-induced ototoxicity in rats: an atypical selective mid-frequency hearing deficit Hear Res 80 25 30 7852200
Crofton KM Rice DC 1999 Low-frequency hearing loss following perinatal exposure to 3,3′,4,4′,5-pentachlorobiphenyl (PCB 126) in rats Neurotoxicol Teratol 21 3 299 301 10386834
Crofton KM Zhao X 1993 Mid-frequency hearing loss in rats following inhalation exposure to trichloroethylene: evidence from reflex modification audiometry Neurotoxicol Teratol 15 6 413 423 8302243
European Parliament and the Council of the European Union 2003. Directive 2003/10/EC of the European Parliament and the Council of 6 February 2003 on the Minimum Health and Safety Requirements Regarding the Exposure of Workers to the Risks Arising from Physical Agents (Noise). Available: http://europa.eu.int/eur-lex/pri/en/oj/dat/2003/l_042/l_04220030215en00380044.pdf [accessed 27 May 2005].
Fechter LD Chen GD Johnson DL 2002 Potentiation of noise-induced hearing loss by low concentrations of hydrogen cyanide in rats Toxicol Sci 66 1 131 138 11861980
Fechter LD Chen GD Rao D Larabee J 2000 Predicting exposure conditions that facilitate the potentiation of noise-induced hearing loss by carbon monoxide Toxicol Sci 58 2 315 323 11099644
Fechter LD Klis SFL Shirwany NA Moore TG Rao D 2003 Acrylonitrile produces transient cochlear function loss and potentiates permanent noise-induced hearing loss Toxicol Sci 75 1 117 123 12832658
Fechter LD Liu Y Herr DW Crofton KM 1998 Trichloroethylene ototoxicity: evidence for a cochlear origin Toxicol Sci 42 1 28 35 9538045
Fechter LD Thorne PR Nuttall AL 1987 Effects of carbon monoxide on cochlear electrophysiology and blood flow Hear Res 27 1 37 45 3583935
Fechter LD Young JS Carlisle L 1988 Potentiation of noise induced threshold shifts and hair cell loss by carbon monoxide Hear Res 34 1 39 47 3403384
Johnson AC 1993 The ototoxic effect of toluene and the influence of noise, acetyl salicylic acid, or genotype. A study in rats and mice Scand Audiol Suppl 39 1 40 8171264
Lataye R Campo P Barthelemy C Loquet G Bonnet P 2001 Cochlear pathology induced by styrene Neurotoxicol Teratol 23 1 71 79 11274877
Lataye R Campo P Pouyatos B Cossec B Blachere V Morel G 2003 Solvent ototoxicity in the rat and guinea pig Neurotoxicol Teratol 25 1 39 50 12633735
Lasky RE Widholm JJ Crofton KM Schantz SL 2002 Perinatal exposure to Aroclor 1254 impairs distortion product otoacoustic emissions (DPOAEs) in rats Toxicol Sci 68 2 458 464 12151642
Manuel J 2005 Clamoring for quiet: new ways to mitigate noise Environ Health Perspect 113 A47 A49
Morata TC Johnson AC Nylen P Svensson EB Cheng J Krieg EF 2002 Audiometric findings in workers exposed to low levels of styrene and noise J Occup Environ Med 44 9 806 814 12227672
Morata TC Fiorini AC Fischer FM Colacioppo S Wallingford KMM Krieg EF 1997 Toluene-induced hearing loss among rotogravure printing workers Scand J Work Environ Health 23 4 289 298 9322820
Pouyatos B Gearhart C Fechter LD 2005 Acrylonitrile potentiates hearing loss and cochlear damage induced by moderate noise exposure in rats Toxicol Appl Pharmacol 204 1 46 56 15781293
Osman K Pawlas K Schutz A Gazdzik M Sokal JA Vahter M 1999 Lead exposure and hearing effects in children in Katowice, Poland Environ Res 80 1 1 8 9931221
Pryor GT Rebert CS Howd RA 1987 Hearing loss in rats caused by inhalation of mixed xylenes and styrene J Appl Toxicol 7 1 55 61 3611598
Rebert CS Sorenson SS Howd RA Pryor GT 1983 Toluene-induced hearing loss in rats evidenced by the brainstem auditory-evoked response Neurobehav Toxicol Teratol 5 1 59 62 6856010
Rice DC Gilbert SG 1992 Exposure to methyl mercury from birth to adulthood impairs high-frequency hearing in monkeys Toxicol Appl Pharmacol 115 1 6 10 1631895
Schmidt CW 2005 Noise that annoys: regulating unwanted sound Environ Health Perspect 113 A43 A44
Sliwinska-Kowalska M Zamyslowska-Szmytke E Szymczak W Kotylo P Fiszer M Wesolowski W 2003 Ototoxic effects of occupational exposure to styrene and co-exposure to styrene and noise J Occup Environ Med 45 1 15 24 12553175
Schwartz J Otto D 1987 Blood lead, hearing thresholds, and neurobehavioral development in children and youth Arch Environ Health 420 3 153 160 3606213
Schwartz J Otto D 1991 Lead and minor hearing impairment Arch Environ Health 46 5 300 305 1953038
|
16002358
|
PMC1257659
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A443-A444
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.113-a443
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0044816002360EnvironewsForumMolecular Biology: The Shape of Food Allergenicity Spivey Angela 7 2005 113 7 A448 A448 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Every year, food allergies cause about 30,000 visits to emergency rooms and an estimated 150 deaths. The culprits are known; only eight foods—milk, eggs, peanuts, tree nuts, fish, shellfish, soybeans, and wheat—cause 90% of all allergic food reactions. But why do those foods cause allergies while others don’t? A study in the January 2005 Journal of Allergy and Clinical Immunology suggests that the answer may lie partly in three-dimensional protein structures that are common to many different plants that cause allergies.
Scientists once thought that any protein could potentially become an allergen. In the current study, however, using a computer program to categorize 129 common plant food allergens, structural biologist John Jenkins of the British Institute of Food Research (IFR) and colleagues found that 65% of these proteins fell into just four structural families. The study used the protein families defined by Pfam, a database of protein structures housed at the Wellcome Trust Sanger Institute in the United Kingdom.
The results suggest that certain protein structures contribute to plants’ allergenicity, says coauthor Clare Mills, head of the allergy research team at IFR. The next step is finding out which structures contribute, and how they do so.
Some of these common structures may make a protein very stable, and thus hard to digest. For instance, one of the four dominant families identified in this study, the cupin family, has barrel-shaped sections (the family gets its name from cupa, a Latin word meaning “barrel”). This shape makes the proteins very stable, Mills says, adding, “If a protein is resistant to digestion, there’s more of it available for the immune antibodies to attack.”
The authors also analyzed surface structures in proteins that are cross-reactive. One family of proteins, the Bet v 1 homologues, showed an unusual conservation of surface shapes across different plants. The scientists studied the family closely to learn more about that conservation and how it underlies the allergic cross-reactivity between birch pollen and plant foods such as apples and celery.
“Generally, proteins change quite a lot on their surface when you go across different species,” Mills says. “But the Bet v 1 family is unusual. Although some of the amino acid residues changed [from the major birch pollen allergen Bet v 1 to the related apple allergen Mal d 1], the shape of the molecule was very much the same.”
According to Mills, the degree of change in the surface of the allergenic protein appears to correlate with the degree of allergic symptoms that people experience. A Bet v 1–related allergen, Api g 1, is found in celery, but its surface shape is altered more from Bet v 1 than that of the apple allergen. Similarly, people with birch pollen allergy can have cross-reactions to celery, but less often than they do to apples.
Mills and colleagues are conducting similar bioinformatics analysis of proteins in pollen and food allergens of animal origin to find out if these also show structural similarities. Although Mills says “it’s not a focus of our research to come up with an in silico method of looking for allergens,” she does say that categorizing proteins into structural families may also help in evaluating the potential allergenicity of proteins found in genetically modified foods. Many people are concerned that these engineered foods may introduce novel proteins that humans are unable to digest.
Stephen Howell, director of the Plant Sciences Institute at Iowa State University, agrees that the study suggests an additional parameter to be considered in evaluating novel proteins for allergenicity. Although new proteins introduced by genetic engineering are already tested extensively, he says that more knowledge can only help inform and improve that testing.
Richard Goodman, a research professor of food science and technology at the University of Nebraska–Lincoln, says that, in addition to bioinformatics tools, researchers may also need to use nuclear magnetic resonance spectroscopy or crystallography to examine tiny differences in surface structure to fully understand protein structures’ role in allergenicity. Allergy is a complicated condition that depends on the amount of allergen present in a food, how often a person has been exposed to it, how many immune cells react to the allergen, and how strongly the cells react. “But,” Goodman says, “this study does indicate that there might be more predictability to this than once thought.”
Degrees of separation. In the above comparison of proteins from apples (left) and celery (right) to that from birch pollen, red areas show where surface structure has been conserved across the proteins. Given these two proteins’ relative structural similarity to that of birch pollen, people allergic to birch pollen are more likely to also be allergic to apples than to celery.
|
16002360
|
PMC1257660
|
CC0
|
2021-01-04 23:41:04
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A448
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.113-a448
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a00451EnvironewsForumEHPnet: UNEP.Net Freshwater Portal Dooley Erin E. 7 2005 113 7 A451 A451 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Even though 70% of the Earth’s surface is covered with water, little of that is freshwater. Today, one-third of the world’s population lives in countries with moderate or high water stress, a fact that leads many experts to proclaim that water may possibly be the primary cause of international tensions and the foremost threat to environmental health in the twenty-first century. The United Nations Environment Programme (UNEP) has set up a Freshwater Portal, located online at http://freshwater.unep.net/, as a centralized resource for anyone looking to learn more about freshwater use, resources, and scarcity. The fully searchable site is part of UNEP’s United Nations Environment Network, which aims to bring specialized environmental science communities together under one umbrella.
The issues associated with freshwater resources are wide-ranging. Many surface water sources are shrinking, as population growth fuels desertification and overuse of resources. Several large rivers now run dry for at least part of the year, and lakes are shrinking. Groundwater, too, is being affected by pollution, salinization, and overuse. Overpumping of water is causing large areas to sink, including almost 60,000 square miles in China. Infrastructure also is in crisis in many areas. In 2002, only 52% of people worldwide were connected to water systems, and only 30% were connected to sanitation services. Each year more than 5 million people die from water-related diseases, and diarrheal diseases are the leading cause of death in children.
The Freshwater Portal has been indexed by nine key issues. These include water scarcity, irrigated agriculture, water and sanitation, water quality, groundwater, transboundary water management, water and ecosystems, floods and droughts, and urban water. For each key issue, UNEP has collected relevant reports, background papers, websites, and other resources. For example, the Water Scarcity section includes links to papers on balancing water uses, managing water within agriculture, and the relationship between water scarcity and poverty. The Water and Sanitation section provides links to two reports as well as to the World Bank Water and Sanitation Program, which guides international efforts to build infrastructure in this area. And the Groundwater section links to a global overview of ground-water conditions, which goes on to detail best management practices for this resource.
The site is also cross-indexed by resource type. Visitors can view all case studies/best practices documents, for example, or go directly to conference proceedings covering multiple topic areas.
Similar portals on other topics are accessible from the top of the homepage. Visitors can choose other Thematic Portals, such as Climate Change or Urban Environment, and can also select from Regional Portals to view information specific to the Arctic, Europe, or Latin America. The homepage also highlights global water assessments and announcements of recent documents, statements, and meetings. Also included is a link to Earthprint.com, where visitors can purchase freshwater publications produced by UNEP and other international organizations.
|
0
|
PMC1257661
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A451
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0045216002361EnvironewsNIEHS NewsGlobal Collaboration Gives Greater Voice to African Journals Tillett Tanya 7 2005 113 7 A452 A454 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Peer-reviewed journals are a vital source of information exchange for researchers and clinicians in the medical and environmental health arena. The timely publication of credible research is integral to advancing the realm of knowledge in any given topic area. Many Northern Hemisphere journals have access to a range of resources to aid in timely publication: a database of potential peer reviewers, adequate staff and tools to produce each issue, and the funds to ensure publication. However, journals in developing regions such as sub-Saharan Africa usually don’t have these same resources. As a result, publication can be erratic, and important information may not reach the people who need it most. Now a multinational partnership of editors from science and medical journals, including EHP, is working to help African journals gain a greater voice on the scientific publishing stage.
Impediments to Publication
Considering that many of the least-developed, lowest-income countries in the world are located in sub-Saharan Africa, it’s not surprising that many assets that are taken for granted in the offices of well-established Northern Hemisphere journals are unavailable to journals in that region. Currently, most African medical journals are funded by academic institutions or professional organizations that usually have only extremely limited funding available. Although it is common for the journals to sell subscriptions and advertisement space, only a small amount of income is generated from these sources due to lack of subscribers and advertisers. Most of the journals have minimal staff to coordinate both the review and production processes, and access to technological tools and other equipment and supplies to facilitate the work is not always possible.
The credibility of published research is heavily dependent upon the peer review process, yet the average African journal has only a limited number of manuscript reviewers. Many things factor into this circumstance: difficulty maintaining reviewer anonymity, cultural mores that discourage criticism of more senior researchers, reviewer conflict-of-interest issues, and the lack of an established system for academic or professional recognition of reviewer input.
All these factors together can lead to an irregular publication schedule, which—with the perception that publication in Northern Hemisphere journals will facilitate more rapid career development—can deter local scientists and clinicians from publishing in their national journals. Therefore, manuscript submission rates are low, and research quality cannot be ensured.
Another difficulty facing African journals is dissemination of their content to other parts of the world. Most information published in African journals never leaves its home borders because these journals are largely not included in major bibliographic databases like the National Library of Medicine’s (NLM) MEDLINE. Such databases have criteria—ranging from quality of content to production quality—that must be met before a journal is accepted for indexing. Only 31 of the 4,900 journals indexed by MEDLINE are from Africa, while more than 4,300 are Northern Hemisphere journals.
Partnerships for Change
To combat the problems faced by journals in Africa, a group of African and Northern Hemisphere medical journal editors met with representatives of other interested organizations at an October 2002 World Health Organization workshop, and created the Forum for African Medical Editors (FAME). Around the same time, representatives from the NLM and the John E. Fogarty International Center (FIC) were discussing how more journals in Africa might be upgraded for acceptance into MEDLINE. In September 2003, respresentatives from the NLM, the FIC, the NIEHS, and nine journals met at the offices of the British Medical Journal (now known officially as BMJ) to discuss ideas for partnerships. The meeting attendees created the African Medical Journal Editors Partnership Program.
Gerald Keusch, assistant provost for global health at Boston University and former director of the FIC, initiated and participated in the 2003 meeting. As Keusch recounts, the participants deemed the most logical answer to solving the limited capacity dilemma of the African journals to be partnering them with counterparts in the United States and the United Kingdom.
Thus, several partnerships were established at this meeting: African Health Sciences (Uganda) was partnered with BMJ, Ghana Medical Journal was paired with The Lancet, Malawi Medical Journal was matched with the Journal of the American Medical Association (now officially known as JAMA), and Mali Médical was partnered with the American Journal of Public Health (AJPH) and EHP. All the participating African journals are in countries where active NIH research is ongoing, and the five partnering Northern Hemisphere journals are interested in promoting public health in developing nations around the world. The FIC, the NLM, and the NIEHS contributed funds to start the program. A contract was then awarded to the Council of Science Editors (CSE) to manage the funds for this pilot project [also see EHP 112: A858 – A860 (2004)].
Why participate in this partnership? For the African journals, the reasons are fairly obvious. For research publications, being connected to the global information network is key to keeping abreast of and benefiting from the very latest information. A partnership like this one would help stabilize the African journals’ connection to today’s information exchange systems and would aid in the compilation and dissemination of public health information in the developing African countries. This, says Keusch, would allow information of a local nature to be more readily accessed locally, and also would allow the evolution of quality publications that all could take pride in.
The Northern Hemispheres journals will benefit too. Thomas J. Goehl, EHP’s editor-in-chief, says that, in addition to participating for purely altruistic reasons, the Northern Hemisphere journals also see this as a unique opportunity to help disseminate research and public health information being generated in Africa. Because developing nations are often exposed to toxicants and pathogens in much greater concentrations and numbers than developed nations, research opportunities abound to learn the basic biology of pathologies that plague both Africa and other countries around the world, he says.
Goehl adds that by creating methods to help developing nations gather and distribute the knowledge they have, industrialized nations can consider those medicinal and traditional procedures when improving their own scientific and medical methods. Thus, a true partnership in research between the developing and industrialized world could ultimately benefit everyone.
However, says Goehl, above all we should be motivated by a duty to protect humanity. “I believe health care should be considered a basic human right,” he says. “If this is true, then we have a moral imperative to provide adequate health care for all people. I hope EHP can contribute to this goal by strengthening journals that are needed to disseminate relevant and credible information to practicing medical professionals.”
Off to a Good Start
Since forming the partnership, the member journals have been involved in several activities. Two training meetings have been held in Africa for local editors and reviewers to develop more effective editorial guidelines for their journals, sharpen writing skills, and improve current manuscript managing processes.
To date, all but one African journal partner have visited the offices of their respective Northern Hemisphere partners. Visits usually last 5–10 days to allow the editors time to get a true sense of how their partner journals operate, and also to give them the opportunity to see how they could adapt operations to fit their own journals’ needs. In the near future, staff from the Northern Hemisphere journals will visit their African counterparts to assist in onsite capacity building.
Supplies needed to facilitate production also are on the way. The NLM has sponsored site visits to each African journal by local technical experts. These experts identified employee training needs and desktop publishing functionality, made hardware and software recommendations, and assessed the journals’ Internet connectivity.
One partner, the Ghana Medical Journal, has already received full equipment upgrades. David Indome, site manager for the technical upgrades, is pleased with how well the transition is going there, and sees the technical overhaul as a change that will help the journal operate much more efficiently. Before the upgrade, the journal was using one computer for all aspects of journal production. The computer, an older model, used severely outdated word processing and virus software. With the upgrade, the staff now has a scanner and two computers with more current word processing, bibliography, desktop publishing, and virus software. “Thanks to the hardware and software upgrade, the Ghana Medical Journal can now prepare articles much faster than before because they are using programs that are easier, quicker, and more convenient,” says Indome. Reliable Internet connection also allows for much easier interaction between editors, reviewers, and authors.
According to Julia Royall, NLM’s chief of international programs and administrator of the equipment and training portion of the partnership, these technical upgrades will help get all of the journals operating at a more efficient level so they can meet the criteria for inclusion in MEDLINE.
EHP, AJPH, and Mali Médical
EHP and AJPH are also actively working with their partner journal, Mali Médical. Operating in Mali’s capital, Bamako, Mali Médical was established in 1975 and published up to four issues per year through 1998. Editor-in-chief Siaka Sidibé has been at the helm since 1996, and since 1998 the journal has consistently published four issues each year.
Sidibé says areas needing immediate attention include computer equipment, Internet access, and training for the editors and reviewers. Another challenge facing the journal is printing costs. Although articles are edited and laid out on a strict schedule, the actual timing of printing an issue can vary due to funding issues. Mali Médical currently has a volunteer editorial board of seven and a staff of three volunteer editors and one paid part-time support person. The journal currently is not indexed in any of the major bibliographic databases.
The week of 24 September–1 October 2004, Goehl, Sidibé, and AJPH editor-in-chief Mary E. Northridge met at the EHP offices in Research Triangle Park, North Carolina, to develop a plan of action for the African journal. Of the four African partner journals, Mali Médical is the only one from a francophone nation. With EHP and AJPH both being English-language journals, language is a challenge—especially in some training efforts—but has not dampened the enthusiasm and resolve of the three partners.
Mali Médical will soon be equipped with hardware and software, and staff will be trained in their use. Through funding from the partnership, a full-time managing editor will be hired for desktop publishing and office administration. The journal also has a new website, http://www.malimedical.org/, which is currently hosted by EHP. The web-site contains articles archived from 2003 to the present, and will eventually be housed on Mali Médical’s own server.
Once all the basic necessities for efficient management are gathered, the journal can begin to focus on raising its recognition. Efforts to this end will include co-publication of research articles in EHP and AJPH, as well as exploration of online manuscript submission and peer review, says Sidibé.
Looking Ahead
The members of the partnership are forging ahead with their goals. Representatives from each partner journal, the FIC, the NLM, and the CSE met during the May 2005 CSE annual meeting to review the first year’s activities and plan for the upcoming year. All agreed that much progress has been made.
All of the tasks defined by the partnership are being addressed, and the African partners have embraced the project. With funding assistance from the FIC, the NLM, and the NIEHS, the four African journals will copublish review articles on several “neglected” diseases, illnesses that generally affect poor people in poor countries and thus may not garner as much research attention from more affluent nations. These articles will appear in both English and French in each journal’s September 2005 (or equivalent) issue. With help from FAME, they have also been able to develop training sessions for editors and research paper writers.
The partners also refined certain original tasks set for the program. For example, instead of training African journal staff in a breadth of skills, a consensus was reached to develop focused training that concentrates on one aspect of the publication process (such as manuscript handling, or marketing and public relations). This concentrated training will help the African journals obtain skills and strengthen their operations at a more accelerated pace.
The participants also agreed they must emphasize public relations to secure more funding. More funding, in turn, will help the African partners achieve and maintain a more regular publication schedule.
Three of the African journals cannot yet be abstracted in MEDLINE, but they have learned they can still submit their content to PubMed Central for archiving, allowing their articles to be read worldwide. Participation in PubMed Central is open to any life sciences journal that meets NLM’s standards for scientific and editorial quality of its content and technical quality of its digital files. Having stable websites could help the journals publish online on time, increasing their chances acceptance into MEDLINE.
The editors themselves will obtain additional help from ScholarOne, a provider of web-based applications to improve the workflow for scholarly journals. ScholarOne has offered software and training services free of charge for five years for each African journal, so the journals can set up and maintain their own online manuscript submission and review systems. SPI Publisher Services has also offered its services free of charge for five years. This company will convert each journal’s files to XML format, a flexible web-site code required for MEDLINE and PubMed Central that allows for more sophisticated website navigation.
The African Medical Journal Editors Partnership Program began with one good idea shared by editors thousands of miles apart. Through the dedication and enthusiasm of the partners and their supporters, great strides have been made in just a short time. Goehl thinks even greater strides are yet to come. “It is a committed one-on-one partnership,” he says, “that is the key to the success of journal capacity building in the developing world.”
|
0
|
PMC1257662
|
CC0
|
2021-01-04 23:41:04
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A452-A454
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a00454EnvironewsNIEHS NewsBeyond the Bench: Promoting Health in Texas Colonias Tillett Tanya 7 2005 113 7 A454 A455 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Sometimes the best educational resources can be found in your own backyard. Staff at the Community Outreach and Education Program (COEP) of the NIEHS Center for Environmental and Rural Health at Texas A&M University recognize how true that can be. Concerned with the health of residents in local colonias (poor, unincorporated neighborhoods along the Texas–Mexico border whose residents are largely of Mexican origin), COEP staff have created a dynamic program that trains promotoras—residents of the colonias—to serve as a link between communities and health educators.
The effort began six years ago as a way to educate colonia residents near Laredo about pesticides and other hazards. These residents lived near farm fields, and pesticides were showing up in house dust and hand-rinse samples. The program now includes a comprehensive environmental health educational program reaching colonia residents in areas bordering Laredo, McAllen, and Bryan.
The promotoras are crucial to the success of the colonia outreach program because they are better able to establish a dialogue with community members. Families then often feel freer to express their concerns with regards to environmental health and more receptive to health education materials. The promotoras usually live in the same neighborhood they provide outreach to, lending a sense of trust and familiarity to their interactions with other residents.
Promotoras are recruited through the Center for Housing and Urban Development at Texas A&M University and through the South Texas Association of Promotoras. According to COEP director Carmen Sumaya, it takes a special person to be a promotora. Although they vary in age from 25 to 60, with consequent variations in life experience, all promotoras are natural community advocates. These volunteers are willing to devote the time to absorb environmental health information and make it relevant to their neighbors so they can take something useful back to their communities.
Melly Tamez, a Laredo promotora, is a dedicated community advocate who sees a need for such information in her community and works to provide it. “The people in my neighborhood need help in health issues, and I want to improve the quality of their lives by instructing them on how to avoid environmental health risks. I take great pride in my role as promotora,” says Tamez.
The first phase of the program involves COEP staff and the promotoras meeting with families at local community centers so residents can identify any environmental health concerns they have. The second phase involves training the promotoras to use flip charts and other materials to convey relevant environmental health information (for example, safe drinking water and food safety practices). The promotoras’ feedback is important at this stage because they can help clarify how health messages can be presented most effectively. In the third phase, the promotoras schedule and conduct visits with their neighbors to provide culturally relevant environmental health education.
In a typical visit, two promotoras come to the home of a resident who has invited at least two neighbors to participate. While one promotora presents environmental health information to the adults, the other engages the children of the household with coloring books and other fun activities. After the presentation, the promotoras answer any questions the residents might have and schedule follow-up visits for one and three months later.
To date, about 15 promotoras have been trained by the COEP. Sumaya acknowledges their value to the COEP’s environmental health campaign. “Promotoras open the door for us to the people of the colonias. They play a pivotal role in the success of our outreach programs,” she says.
Neighborly advice. A program along the Texas–Mexico border trains community members to be health advocates.
|
0
|
PMC1257663
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A454-A455
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a00455EnvironewsNIEHS NewsHeadliners: Reproductive Health: Effects of Organochlorine Compounds on Menstrual Cycles Phelps Jerry 7 2005 113 7 A455 A455 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Windham GC, Lee D, Mitchell P, Anderson M, Petreas M, Lasley B. 2005. Exposure to organochlorine compounds and effects on ovarian function. Epidemiology 16:182–190.
Over the past 20–30 years, environmental health scientists have expressed increasing concern about endocrine disruptors, chemicals that appear to disrupt hormonal activity in humans and animals. Research has shown that women exposed at various life stages to endocrine disruptors may have increased risk of menstrual cycle irregularities, infertility, endometriosis, autoimmune disorders, and cancers of the reproductive system. Now NIEHS grantee Gayle C. Windham of the Department of Health Services in Oakland, California, and colleagues have found that the pesticide DDT and its metabolite DDE were associated with menstrual length differences in a population of immigrant women from Southeast Asia.
DDT was one of the first chemicals to be shown to have adverse endocrine effects. In wild birds, especially those high on the food chain, DDT was linked with weakened eggshells, which caused large drops in the numbers of some species of raptors including the bald eagle. DDT was shown to interfere with the deposition of calcium as the developing egg passes through the bird’s uterus. For this and other reasons, its use was banned in the United States in 1972.
The California researchers studied 50 Laotian women of reproductive age currently residing in the San Francisco Bay area. The team examined serum samples for suspected endocrine disruptors including DDT, DDE, 4 other chlorinated pesticides, and 10 polychlorinated biphenyls. They found that serum samples from all the women in the study had detectable concentrations of DDT and DDE, with mean levels higher than typical of U.S. women.
Menstrual cycle length was approximately four days shorter for women with the highest DDT and DDE levels compared to women with the lowest levels. With each doubling of serum DDE (though not DDT), cycle length decreased by a little more than one day. Also, as DDE level increased, progesterone metabolite levels decreased. There was no significant association between polychlorinated biphenyl levels and changes in cycle length or hormone levels.
These results indicate an effect of DDT exposure on ovarian function and menstrual cycle length, potentially contributing to problems with fertility, pregnancy, and other aspects of reproduction. The findings need to be duplicated because of the small size of the study population, but they do suggest that DDT exposure may be an important factor in reproductive problems. These human health effects also have implications for the continued use of DDT and similar compounds in other parts of the world.
|
0
|
PMC1257664
|
CC0
|
2021-01-04 23:41:04
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A455
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0045616002362EnvironewsFocusPaving Paradise: The Peril of Impervious Surfaces Frazer Lance 7 2005 113 7 A456 A462 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Paved surfaces are quite possibly the most ubiquitous structures created by humans. In the United States alone, pavements and other impervious surfaces cover more than 43,000 square miles—an area nearly the size of Ohio—according to research published in the 15 June 2004 issue of Eos, the newsletter of the American Geophysical Union. Bruce Ferguson, director of the University of Georgia School of Environmental Design and author of the 2005 book Porous Pavements, says that a quarter of a million U.S. acres are either paved or repaved every year. Impervious surfaces can be concrete or asphalt, they can be roofs or parking lots, but they all have at least one thing in common—water runs off of them, not through them. And with that runoff comes a host of problems.
Globally, it is a little more difficult to judge the square mileage of impervious surfaces. “We can extrapolate from the United States to a degree,” says Ferguson, “but there are too many variables to judge accurately.” The United States has a lot of automobiles, and compared to many other countries, Americans tend to build more (and wider) roads, more (and bigger) parking lots, more (and more expansive) shopping centers, and larger houses (with accompanying larger roofs). He says, “The United States might be on a par with Europe, but we’d be very different from India, for example, or any country where large numbers of the populace live in smaller, scattered villages, mostly without paved roads, parking lots, and the like.”
According to the nonprofit Center for Watershed Protection, as much as 65% of the total impervious cover over America’s landscape consists of streets, parking lots, and driveways—what center staff refer to as “habitat for cars.” Says Roger Bannerman, a researcher with the State of Wisconsin Department of Natural Resources: “You see some truly insane things in this country. I’ve seen subdivisions with streets that are thirty to forty feet wide. That’s as wide as a two-lane highway. Most developers are going back to a twenty-five- to twenty-eight-foot width, but you can still see these huge streets.”
Upon these automotive habitats fall a variety of substances, and thereby hangs the rest of the tale. Impervious surfaces collect particulate matter from the atmosphere, nitrogen oxides from car exhaust, rubber particles from tires, debris from brake systems, phosphates from residential and agricultural fertilizers, and dozens of other pollutants. “On a parking lot, for example, we have demonstrated buildups of hydrocarbons, bacterial contamination, metals from wearing brake linings, and spilled antifreeze,” says Ferguson.
On a road of open-graded aggregate (stone), much of that material would seep down into the pavement and soil, and the community of microorganisms living there would begin a rapid breakdown process. But pollutants can’t penetrate an impervious surface, and the rapid flow of rainwater off of impervious surfaces means these pollutants end up in the water. “So then,” says Ferguson, “not only do you have too much water, all moving too fast, you have polluted water that kills fish and makes water unfit for drinking or recreation.”
When Water Has Nowhere to Go
Areas across the country are being impacted by the growth in coverage by impervious surfaces. In Maryland, for example, when watershed imperviousness exceeds 25%, only hardier reptiles and amphibians can thrive, while more pollution-sensitive species are eliminated, according to a 1999 Maryland Department of Natural Resources report titled From the Mountains to the Sea. Watershed imperviousness exceeding 15% results in streams that are impossible to rate “good,” states the report, and even 2% imperviousness can affect pollution-sensitive brook trout.
The 1.1-million-acre Chesapeake Bay watershed, one of the most diverse and delicate ecosystems in the world, is now being impacted by the 400,000 acres of impervious surfaces in Maryland. The Great Lakes, the streams and rivers of the Pacific Northwest, the Everglades of Florida—all are being impacted in one or more ways by runoff from streets, parking lots, and rooftops.
Bannerman has spent the last 30 years studying stream flow and the effect of urbanization on watersheds, including the depletion of groundwater reserves. “Not allowing the rainfall to infiltrate back into the aquifer is a very serious issue,” he says. “If that happens, you lose the base flow [the portion of water derived from underground sources] for streams, and you lose the wetlands fed by springs. It’s a complete disruption of the hydrologic cycle.”
Bannerman cites the example of Lake Wingra, a 1.3-square-kilometer lake in Madison. “A hundred years ago,” he says, “this lake was fed by around thirty-five separate springs. But today, because the lake is now almost entirely surrounded by urban areas, there are only four streams feeding the lake. Local organizations have gotten active in trying to restore the lake’s water quality, but it’s not the same lake it was a hundred years ago.” Lake Wingra now suffers from algal blooms caused by overfertilization, beach closures due to bacterial contamination, turbidity, and drying of surrounding wetlands.
Bruce Wilson, a research scientist with the Minnesota Pollution Control Division, is midway through a satellite survey of impervious surface area in that state. What Wilson has seen thus far is enough to cause significant concern about the state’s growth and development, and the impact of impervious surfaces on the water system.
“Impervious surfaces are impacting the lakes and streams on a number of fronts,” he says. “Velocity of runoff is a big one. Water runs off of these surfaces so rapidly, it creates mini-tsunamis that can cause serious, even irreparable, harm to the stream ecosystem. . . . And of course, the ability to recharge the groundwater system is being impacted. If you get into a twenty- to thirty-percent drop in infiltration [into the aquifer], which means a loss of base flow, the impact on streams being fed by surface water gets magnified still further.”
Another big problem for urban areas is the flash flooding that can occur when heavy rains fall over a city, according to hydrometeorologist Matt Kelsch, an authority on urban flash flooding with The University Corporation for Atmospheric Research in Boulder. Since runoff from an acre of pavement is about 10–20 times greater than the runoff from an acre of grass, Kelsch says impervious surfaces can quickly trigger devastating floods that can produce a host of their own environmental health hazards.
“In urban areas, anywhere from thirty to forty percent of the rainfall runs right into whatever stream is in the area, and in heavily urbanized areas it can be more than fifty percent,” he explains (by comparison, he says, the amount of runoff in subsaturated woodlands is often less than 5%). “If the water overflows the stream banks, it’s going to seek the path of least resistance. In most cases, that’s going to be the roadways.”
In many desert areas, Kelsch says, engineers take advantage of the natural topography, building houses at higher elevations and installing roads that lead up to residential areas. What this does is make the roads far more dangerous. More than 50% of the fatalities in flash floods occur on roads, according to Kelsch.
Floods are often given numerical designations such as “hundred year flood,” meaning such a flood happens once every 100 years (or has a 1% chance of occurring in any given year). The Federal Emergency Management Agency maintains a national list of flood zones and maps of impacted areas. The problem, says Kelsch, is that we’ve changed the playing field. “A couple of factors come into play,” he says. “First, this is still a pretty new country, so most places haven’t been developed long enough to know about the historical risk of a devastating flood. Secondly, when we urbanize an area, we alter the historical frequency of these events. The more we develop an area, the more rainfall we put as runoff directly into streams that have evolved to handle only a fraction of that runoff, and the more that happens, the greater the likelihood of a catastrophic flood.” Several such floods hit New Orleans in the 1980s, and three hit St. Paul–Minneapolis between 1990 and 2001.
Heat Islands and the Stream
Wilson is also studying the “heat island” impact on Minnesota’s trout streams, an impact he says evidence and experience suggest is significant. Impervious surfaces, particularly roads and parking lots, are generally dark, and thus heat-absorbing, so they heat the rainwater as it hits. A sudden thunderstorm striking a parking lot that has been sitting in hot sunshine (where surface temperatures of 120°F are not unheard of) can easily yield a 10°F increase in rainfall temperature. And that heated water isn’t coming off just one parking lot or one street, but more likely several, all adding heated water to a stream or river.
Many aquatic organisms, at different stages of their lives, are vulnerable to even small increases in water temperature. “I’ve seen trout streams in Wisconsin and elsewhere in the Midwest lose whole populations because of—at least in part—the rise in temperature caused by runoff from impervious surfaces,” Wilson says. “Increased temperature also decreases the water’s ability to hold oxygen, which has a further detrimental effect on the aquatic life.” Warm temperatures can cause a variety of problems for fish, including decreased egg survival, retarded growth of fry and smolt, increased susceptibility to disease, and decreased ability of young fish to compete for food and to avoid predation. Especially affected are species that require cold water throughout most stages of their lives, such as trout and salmon.
Eventually, given no additional changes, the temperatures would drop, but in the interim the impact on wildlife could be serious. Oregon is one state that is examining the science of water temperature effects on stream life. Oregon standards for optimal salmon and trout rearing and migration call for water temperatures of 64.4°F. According to a 2004 report by the Oregon Independent Multidisciplinary Science Team, which advises the state government on scientific matters related to the Oregon salmon and watershed management, studies have shown that adult salmon begin to die off at temperatures of 69.8–71.6°F, and some species of trout at slightly higher temperatures. Although young salmon can survive slightly higher temperatures, the impact on their growth and survival rate is well documented.
Impact of Building Materials
Not yet as well documented is the impact of pollutants released into stormwater runoff by building and paving materials themselves. Asphalt is one concern, as it contains coal tar pitch, a recognized human carcinogen, as well as polycyclic aromatic hydrocarbons (PAHs) including benzo[a]pyrene, another carcinogen. Another potential source of pollution is wood used for utility poles, play structures, and other structures that has been treated with chromated copper arsenate (CCA; a substance now being phased out due to health concerns), pentachlorophenol, or creosote. According to a paper presented at the 2004 Annual Water Resources Conference by Melinda Lalor, a professor of environmental engineering at the University of Alabama at Birmingham, in 1987 the United States alone produced some 11.9 million cubic meters of CCA-treated wood, 1.4 million cubic meters of pentachlorophenol-treated wood, and 2.8 million cubic meters of creosote-treated wood. And structures, once built from such materials, are intended to last a long time. The health risks of arsenic and chromium are well known, and while copper is not generally a human health risk, low concentrations of certain ionic forms of this metal are toxic to marine flora and fauna.
“In general,“ says Lalor, “pollutant level tends to vary depending upon the age of the material, and the harshness of the environment to which it is exposed. As material ages and is exposed to high levels of sunlight, temperature extremes, chemicals in the environment such as salt from roads, and so on, leaching out will increase.”
If the pollutant source is a coating, then pollution levels decrease with age, but can still have a significant impact, she says. “If you look at the asphalt used in a parking lot, the top coat is quite toxic. So if you have a heavy rain [soon] after the parking lot goes in, it’s not unusual to see fish kills downstream.”
Lalor cites research published in volume 35, issue 9 (1997) of Water Science and Technology showing that stormwater from roofs and streets contributed 50–80% of the cadmium, copper, lead, and zinc measured in Swiss combined sewer system flows. Polyester roofing materials shed the highest concentrations of metals, followed by tile roofs, then flat gravel roofs. The Swiss researchers also found PAHs and organic halogens in the roof runoff.
The chemicals released can have a significant impact on environmental and potentially human health. “Some materials, such as metals, are especially toxic to fauna at various stages of their life cycle,” says Lalor, “while some organics, particularly petroleum-based organics, can function as pseudoestrogens. So while they may not cause death, they can trigger a significant disruption in the physiology of the organisms exposed to these pollutants.”
According to Lalor, although there are mandated tests for urban stormwater discharge, there are currently no tests mandated for building materials to determine their potential for toxics release. “If a community wants to develop around their drinking water source, they should know about release potential from building materials so they can carefully select those with which they build,” she says. “We don’t yet have the science to support it, but it would be a positive step to be able to go to a builder and say, ‘Look, here’s a list of twelve building material alternatives that would be most environmentally benign for this site and these conditions.’”
Lalor says New Zealand has been the leader in this sort of study, and that nation is preparing to put regulations in place regarding building materials and environmental impact. But such studies haven’t been elevated to a high enough priority in the United States to build the science we need for setting new policies. She adds, “We need to address the entire life cycle of building materials, from what goes into their creation, to the impact of construction on the environment, to the impact of whatever might leach out during their lifetime, to the end-of-life disposal issues.”
The Promise of Porous Pavements
Despite the overwhelming body of evidence supporting the negative relationship between impervious surfaces and the environment, no one would seriously suggest that we stop paving streets or building parking lots. What, then, are the options?
According to Ferguson, there are nine different families of porous pavement materials. Some of these materials are already well known in the United States; they include open-jointed pavers that can be filled with turf or aggregate, “soft” paving materials such as wood mulch and crushed shell, and traditional decking.
Other families include porous concretes and asphalts being developed by engineers and landscape architects. Ferguson says these materials use the same components and manufacturing processes as conventional impervious materials, “and as a general rule, carry the same health and environmental issues. . . . Same chemicals, same energy costs to manufacture, but far different benefits in its use.” These new formulations still provide solid, safe surfaces for foot and vehicle traffic, but also allow rainwater to percolate down into subsurface soils.
The porosity of porous asphalt is achieved simply by using a lower concentration of fine aggregate than in traditional asphalt; it can be mixed at a conventional asphalt plant. Under the porous asphalt coating is a bed of clean aggregate. Importantly, this aggregate is all of the same size, which maximizes the void spaces between the rocks, allowing water to filter through. A layer of geotextile fabric beneath this bed lets water drain into the soil and keeps soil particles from moving up into the stone.
Porous asphalt was actually developed more than 30 years ago, according to Ferguson, but it didn’t pan out at that time. Part of the problem, he believes, was—and continues to be—the low level of federally funded research. “Back in the early eighties, when porous pavement was new, the Environmental Protection Agency [EPA] was really interested, especially in porous asphalt,” he says. “But one of the problems with porous asphalt back then was that on a hot day, the binder softened and migrated down to a cooler layer. That released the surface aggregate and clogged the lower layer.” According to Ferguson, the EPA became discouraged and discontinued studies.
Since then, however, porous asphalt technology has been improved by French, Belgian, and Irish researchers, Ferguson says. During the late 1980s and early 1990s, they discovered that adding polymer fibers and liquid polymers to the asphalt prevented the binder from draining down through the aggregate. “Today, even though [porous asphalt] started out here, what we’re using has been imported back from Europe,” he says.
Ferguson says porous pavement constitutes only a minute fraction of all the paving done each year in the United States. “However,” he continues, “the rate of growth of porous paving, on a percentage basis, is very high, primarily because of public concern about and legal requirements for urban stormwater management. This growth is happening both in the big asphalt and concrete industries, and in the smaller industries that supply competing materials such as concrete blocks and plastic geocells.”
One argument against pervious surfaces in high-traffic areas is that they’re not as durable as their impervious ancestors. That, says Ferguson, is simply not true. “I’ve seen pervious pavement in good shape in places like Minnesota and Alaska, where you have tremendous climatic extremes,” he says. “In Georgia and Oregon, it’s now routine to resurface highways by putting a layer of pervious asphalt over the impervious surface below. That way, water drains laterally below the surface, giving you better traction and visibility.” Although the major advantage to this practice is highway safety, rather than reinfiltration of the water into the groundwater, it still allows for more water to return to the groundwater table than would be the case with an impervious surface, where it merely evaporates back into the atmosphere.
Some pervious surfaces have the additional benefit of allowing pollutants to come into contact with microbes beneath the surface. According to Ferguson, these naturally occurring microbial communities thrive on the large surface area of the pervious pavements’ internal pores and break down contaminants (particularly petroleum by-products) before they can leach down into the water supply.
“Coventry University scientists did a study recently, where they applied oil to a lab mockup of a porous road surface,” he says. “They dumped far more used oil on the surface than you’d ever find accumulating on a parking lot, and none of it reached the soil layer below”—instead, microbes digested it all. The Coventry team, led by Christopher J. Pratt, published an overview of their work in the November 2004 issue of the Quarterly Journal of Engineering Geology and Hydrogeology.
Other Ways of Controlling Runoff
Approaches to dealing with the spread of impervious surfaces go beyond changing the building material itself. Kelsch says a return to more reasonable street width is one measure, and many communities are increasing their number of green areas as a means of allowing rainfall to infiltrate back into the ground.
For urban areas with nearby lakes, Bannerman says construction of “rain gardens” is becoming a popular method that homeowners and businesses can use to help control stormwater runoff. Such gardens are designed with dips in the center to capture water, which then can slowly filter into the ground rather than run off into the storm sewer. Ideally these gardens are situated next to a hard surface such as a sidewalk or driveway, and are planted with hardy native species that can thrive without chemical fertilizers or pesticides.
Ponding basins like those used in Fresno are another option. This city of just over half a million in Southern California’s San Joaquin Valley gets less than 12 inches of rain annually and draws most of its water from underground aquifers and the nearby Kings and San Joaquin rivers. Beginning in the late 1960s, the city started constructing several ponding basins—large basins where stormwater can settle, then drain down through the soil. Water systems manager Lon Martin says the city had two goals in establishing these ponding basins: “First was to keep stormwater runoff from flooding the city and from going into the rivers, potentially causing water quality problems. Secondly, the city has begun a program of intentional aquifer recharging.”
To date, he says, the city has connected nearly 80 of the possible 150 ponding basins to its groundwater recharge system. Recharge from stormwater is one part of the equation, but the city also takes its May–October water allotment from the two rivers, diverts the water to these basins, and then allows gravity to pull the water down through the sandy loam soil into the aquifer.
Green roofs, another method of controlling rainwater runoff, are just what the name implies: roofs planted with all types of vegetation. Also known as “eco-roofs,” these surfaces can be either extensive (lighter in weight, relying on a few inches of soil and using plants like herbs, grasses, and wild-flowers) or intensive (much heavier, with a 12-inch soil depth that can accommodate trees and shrubs). According to the nonprofit Earth Pledge Foundation, green roofs can absorb nearly 75% of the rainfall that lands on them, and they can also reduce the urban heat island effect.
Green roofs perform several roles, one of which is water harvesting, or basically catching rainwater for use elsewhere. “This water is cleaner than that off the pavement,” Ferguson says. “[Water harvesting] is now being practiced in areas where water is less available, such as the Southwest or the Pacific Northwest, with their dry summers. . . . [It] can be a valuable tool in areas where water is scarce.”
In Germany, approximately 10% of the buildings have green roofs, and the city of Tokyo recently mandated that usable rooftop space of greater than 1,000 square meters atop new buildings must be 20% green. Green roofs are also found in North American cities including Chicago, Toronto, and Portland, Oregon.
Beyond Imperviousness
Recognizing the environmental health threat of impervious surfaces as well as other point sources of pollution, the EPA established a stormwater permitting program under the National Pollutant Discharge Elimination System. Phase I of the stormwater program, promulgated in 1990, required permits for separate stormwater systems serving communities of 100,000 or more people, and for stormwater discharges associated with industrial and construction activity involving at least five acres. Phase II, promulgated in 1999, addressed remaining issues and urban areas of fewer than 100,000 people, as well as smaller construction sites and retail, commercial, and residential activities.
But further change will require a shift in how we think about runoff. Bannerman says, “What we’ve begun to do, and must continue to do, is to get away from the idea that rain is wastewater—something to get rid of, to pass along to our neighbors downstream. We need to keep it where it falls, and the way to keep it is to get it back into the ground.”
For flash floods, Kelsch says, “there is no solution. Flooding is going to happen, in spite of everything we can do. What we need to do is what we can to lessen the impact of the inevitable. That means building out of flood plains, and increasing the amount of rainwater we send back into the aquifers while decreasing the amount we discharge into streams.” Building design and use of permeable paving materials will help, he says, but we need to realize these aren’t total solutions. Further, he adds, “If we get stuck in the mindset that we have to have a solution, we may not do anything. And that will make the problem still worse.”
Impervious Cover of Various Land Uses
Impervious to change? Despite community efforts, Wisconsin’s Lake Wingra still suffers the effects of its urban surroundings including algal blooms, bacterial contamination, and turbidity.
Awash in toxicants. Chemicals used in paved surfaces can be toxic to fish, wildlife, and possibly humans.
Breaking through barriers. Porous pavements come in many forms. Parking spaces in Columbus, Ohio (top left) are made of recycled clay aggregate. Shoppers at the Mall of Georgia, the largest mall in the U.S. Southeast, can park in a turf overflow lot (bottom left). The spaces between open-jointed pavers at Ontario’s Sunset Beach Park lakefront access lot (above) admit water and prevent pollution of Lake Wilcox.
On top of the problem. The green roof atop Chicago City Hall contains more than 100 plant species that absorb stormwater and reduce the ambient air temperature by as much as 7–8ºF compared to a nearby tar roof.
|
16002362
|
PMC1257665
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A456-A462
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.113-a456
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0046416002363EnvironewsSpheres of InfluenceCombined Sewer Systems: Down, Dirty, and Out of Date Tibbetts John 7 2005 113 7 A464 A467 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
When combined sewer systems were introduced in 1855, they were hailed as a vast improvement over urban cesspool ditches that ran along city streets and spilled over when it rained. These networks of underground pipes were designed to dry out streets by collecting rainwater runoff, domestic sewage from newly invented flush toilets, and industrial waste-water all in the same pipe. Waste- and stormwater was then discharged directly into waterways; in the early twentieth century, sewage treatment plants were added to clean the wastewater before it hit streams. Combined sewer systems were—and still are—a great idea, with one catch: when too much stormwater is added to the flow of raw sewage, the result is frequently an overflow. These combined sewer overflows (CSOs) have become the focus of a debate regarding the best techniques to manage growing volumes of sewage and stormwater runoff in many older U.S. communities.
In dry weather, a combined sewer system sends a town’s entire volume of waste-water to a sewage plant, which treats and discharges it into a waterway. Rain and snowmelt, however, can fill up a combined sewer. The sewers have been specifically designed with escape overflow pipes so that the mixture of sewage and stormwater doesn’t back up into buildings, including homes. The resulting CSO dumps raw sewage into lakes, rivers, and coastal waters, potentially harming public health and the environment.
In April 1994, the U.S. Environmental Protection Agency (EPA) issued the CSO Control Policy, the national framework for control of CSOs, through the National Pollutant Discharge Elimination System permitting program. This policy mandated that communities dramatically reduce or eliminate their CSOs, and the agency began working with municipalities to improve antiquated sewage systems so they could reach Clean Water Act goals. Under this policy, communities with combined sewer systems must establish a short-term plan to control these discharges as well as a long-term control plan.
The EPA’s mandate on CSOs leaves communities with two basic options, according to Joan B. Rose, a public health microbiologist at Michigan State University. Communities with CSOs can build separate underground pipes for sewage and stormwater. Or they can keep their combined pipes and somehow build more capacity. “But if they shut down [combined sewer systems],” she says, “communities must find a way to store or treat peak flows when it rains.”
When CSOs Occur
About 40 million people in 32 states live in cities with combined sewer systems; most of these systems are found in Maine, New York, Pennsylvania, West Virginia, Ohio, Indiana, Michigan, and Illinois. CSOs are a major water pollution concern for 772 cities, according to the EPA’s 2004 Report to Congress: Impacts and Control of CSOs and SSOs [sanitary sewer overflows, which are associated with another type of sewer system].
Although some major cities like New York City and Philadelphia have combined sewer systems, most communities with CSO problems have fewer than 10,000 people, according to the EPA report. One reason lies in the economy of scale: larger municipalities are more likely to have sufficient tax base and water users to finance remedies to CSO problems.
CSOs annually result in an estimated 850 billion gallons of untreated wastewater and stormwater being discharged into U.S. waterways, according to the EPA report. Thanks to the CSO Control Policy, this is an improvement over figures in the agency’s 2001 report on the same topic, which put the figure at 1.3 trillion gallons per year.
CSOs flood waterways with contaminants including microbial pathogens, suspended solids, chemicals, trash, and nutrients that deplete dissolved oxygen. Microbial pathogens and toxics can be present in CSOs at levels that pose risks to human health. CSOs can therefore lead to contamination of drinking water supplies, beaches, and shellfish beds.
The EPA’s 2004 report indicated that the agency has limited information on the extent of human health impacts of CSOs. One health effect the agency can quantify comes from regularly monitored coastal and Great Lakes beaches. Using data from these locations, the EPA estimates about 3,500–5,500 gastrointestinal illnesses each year are caused by CSO and SSO pollution of swimming waters.
According to Rose, current estimates hold that microbial pathogens in U.S. public drinking water supplies sicken hundreds of thousands of people each year, though most of these waterborne illnesses are mild, disappearing after a few days. It’s difficult, moreover, to trace sources of these illnesses. Many outbreaks in the United States go unreported, and in most outbreaks the pathogen is not identified. CSOs may or may not be to blame.
However, several reports and studies, including one from the 22 November 2002 Morbidity and Mortality Weekly Report, demonstrate that there has been an increase in waterborne disease outbreaks in the United States over the past few years. Rose says these community outbreaks are correlated with rainfall as well as associated overflows and leaks in public sewer systems. Sensitive populations—the elderly, the very young, and those with existing health problems—are most vulnerable to waterborne enteric microorganisms. These populations make up about 20% of the U.S. public.
A Controversial Alternative
In most municipal treatment plants, waste-water usually goes through a two-step process before it is discharged into lakes, rivers, and coastal waters. Large solids are removed first during primary treatment—mechanical screens remove large debris, and sedimentation tanks remove sludge (solids that sink) and scum (elements that rise to the top). In the second step, wastewater is first routed to tanks with activated microbes that break down organic materials and remove some pathogens and more of the remaining solids. This biological treatment can improve the effectiveness of disinfection, which is often the second part of this step—chlorine is used to kill bacteria and other remaining pathogens before the water is released.
This treatment process is the most effective way to ensure that effluent is clean. It has become the required standard for wastewater treatment under the Clean Water Act. But many plants have smaller biological treatment capacities than primary treatment capacities. Biological treatment facilities are expensive, and many communities have outgrown systems that were built 30–40 years ago. Moreover, these facilities can be delicate. Large waste-water flows into biological treatment units, such as those following heavy rains, can wash the microorganisms from the tanks. The units must then be shut down until the microbial population replenishes itself.
“Blending,” or “bypassing,” is one engineering technique that many sewerage operators have used to handle peak flows. During wet weather, utilities route a portion of peak wastewater flows around the biological treatment units, then combine the rerouted flows with the portion of wastewater that went through biological treatment. After blending, the effluent is usually disinfected and discharged into water bodies.
For decades, environmental permitting agencies in some states have allowed sewerage operators to use this technique in an effort to avoid CSOs. Recently, however, blending has become a controversial practice debated by the wastewater industry, environmentalists, and public health scientists.
The wastewater treatment industry argues that bypassing biological treatment for a portion of the water is a significant improvement over releasing completely untreated wastewater, which is what happens when combined sewers overflow. “If the choice is between raw sewage getting sent into waterways and wastewater getting sent to the treatment facility, most people would rather that the wastewater get treatment,” says Alexandra Dunn, general counsel of the National Association of Clean Water Agencies, a trade group representing more than 300 utilities.
However, critics of blending say the process allows for a higher concentration of pollutants to be released into water bodies, potentially sickening more people. When blending wastewater rather than fully treating it, utilities are less effective at removing microbial pathogens, says Charles Haas, an environmental engineer at Drexel University. “It’s much more difficult to disinfect poorly treated wastewater than fully treated wastewater, and I would consider primary-treated wastewater as poorly treated,” he says. Many solids may still remain in primary-treated wastewater, and viruses, parasites, and bacteria within those solids are protected from disinfection, he adds.
“When the wastewater industry talks about blending versus CSOs, they argue that blending is better in terms of protecting public health,” Rose says. “But I don’t think they have the data on pathogens—viruses and parasites—to prove that. Much more research is needed on wastewater and on treatment to control pathogen risks.”
Toward Better Blending
Nancy Stoner, director of the Clean Water Project at the Natural Resources Defense Council, says that current blending policy as outlined in the CSO Control Policy is chaotic and poorly regulated: “A lot of treatment plants do blending [almost anytime it rains], and some are given the leeway by states to do blending, while other treatment plants are not given the leeway but do it anyway.” In recent years, sewerage operators have sought guidance from the federal government on the blending issue.
In November 2003, the EPA proposed a new federal policy that would have authorized municipal sewage plants around the country to blend wastewater in certain circumstances and under certain conditions—for example, only during periods of heavy rain or snowmelt, and only if plants were already meeting effluent standards required for permitting. The EPA said that its proposed policy was already common practice in many communities.
During the public comment period, the EPA received about 98,000 comments, and the proposal was not warmly embraced by environmentalists. “EPA’s proposal would put more partially treated sewage into the environment,” says Stoner. “The solution to overflows is not to bypass, but to fix the leaky sewer systems.”
Congress reacted, too. On 3 March 2005, members of the House of Representatives introduced bipartisan legislation, the “Save Our Waters From Sewage Act,” to block the EPA blending proposal. The legislation called for amendments to the Clean Water Act to “prohibit a publicly owned treatment works from diverting flows to bypass any more of its treatment facility unless the bypass is unavoidable . . . [and] there is not a feasible alternative to the bypass.”
On 19 May 2005, the EPA announced that it would not finalize the sewage blending policy as proposed in November 2003. “Blending is not a long-term solution,” said Benjamin Grumbles, assistant administrator for the Office of Water, in a press release at the time. “Our goal is to reduce overflows and increase treatment of waste-water to protect human health and the environment.” The agency also said it will continue to review policy and regulatory alternatives to create feasible approaches to treat wastewater.
The Cost of Fixing Systems
According to the EPA’s 2000 Clean Watersheds Needs Survey, it would cost about $50.6 billion over the next 20 years to reduce the nation’s CSO volume by 85%. In recent years, some communities with CSOs have increased sewer rates to raise funds to upgrade their infrastructure. But it’s difficult for many localities to pay for large-scale sewer and water treatment facilities without federal and state aid, says Dunn. Some relatively prosperous communities such as Grand Rapids, Michigan, are in the process of installing separate stormwater pipes, she adds, but this is not feasible for large, financially distressed cities such as Detroit.
Municipalities, sewerage operators, public health scientists, and environmentalists are calling for more federal funding to replace aging pipes and upgrade treatment systems. But federal spending for sewerage infrastructure is actually falling. During fiscal year 2005, Congress cut $250 million from the Clean Water State Revolving Fund (CWSRF), which provides low-cost loans primarily for sewerage infrastructure upgrades. In President Bush’s fiscal year 2006 budget proposal, this fund faces a further one-third reduction.
Even in the best of fiscal times, the CWSRF, distributed among 50 states, cannot address municipal needs to borrow for CSO projects and repay on favorable terms. “In most cases,” says Dunn, “the total allocation to a state per year could be used by one city alone for one phase of its project.” For example, she says, the first phase of the CSO program in place at Narragansett Bay, Rhode Island, will cost $250 million, which could use up all of Rhode Island’s annual portion of the CWSRF.
Despite falling federal aid, communities still—as mandated by the CSO Control Policy—must establish and find a way to implement long-term control plans that will provide for full compliance with the Clean Water Act, including significant reduction of CSOs. Communities are in various stages of developing and implementing their long-term plans.
Some cities such as Boston, Chicago, and Atlanta have built deep storage tunnels to hold stormwater overflows, says Chris Hornback, regulatory director of the National Association of Clean Water Agencies. Eventually, the extra wastewater can be treated at a flow that works for a particular wastewater plant. “Building these storage tunnels is a simple, straightforward process, but it costs hundred and hundreds of millions of dollars, and at some point you reach a break point in cost,” he says.
Environmentalists call for less costly methods of reducing stormwater runoff and CSOs. Such methods, says Stoner, include better means of trapping storm-water before it reaches sewers and putting it into the ground instead. Installing rain gardens, permeable pavements, roof gardens, or even just grassy swales or ditches along roadways can be beneficial for a number of reasons: soil and vegetation provide filtration, groundwater supplies are replenished, and overland stormwater flows are diminished. Such methods are mostly low-tech and cost-effective. [For more information on reducing runoff, see “Paving Paradise: The Peril of Impervious Surfaces,” p. A456 this issue.]
Even so, low-impact techniques alone will not be enough to fully control the CSO problem, according to the EPA’s 2004 report. Environmentalists and municipalities agree that, contrary to the current trend, the answer will depend on greater federal investment in wastewater infrastructure around the country.
|
16002363
|
PMC1257666
|
CC0
|
2021-01-04 23:41:04
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A464-A467
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.113-a464
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0046816002364EnvironewsInnovationsOutsmarting Olfaction: The Next Generation of Mosquito Repellents Schmidt Charles W. 7 2005 113 7 A468 A471 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
The world’s most dangerous animal weighs about two milligrams and pursues its human prey at speeds of barely a mile per hour. Surprised? Don’t be. The dubious honor belongs to the lowly mosquito—a fragile creature whose bite infects millions with lethal diseases, such as malaria, dengue, and West Nile encephalitis. For centuries, humans have slathered on insect repellents to deter the buzzing menace—the first recorded repellents were documented by Herodotus around 400 B.C. But these products have always been far from optimal. Even DEET (N,N′-diethyl-n-toluamide), the world’s most popular and efficacious repellent, has numerous shortcomings: it can require frequent applications, it must be applied to all exposed body parts, and it won’t protect against some dangerous mosquito species, including Anopheles albimanus, the chief malaria vector in Central America.
Today, the need for more effective repellents is increasingly urgent, experts say. According to the World Health Organization, global climate change is expanding mosquitoes’ range, heightening the risk of disease for millions of additional people. The Centers for Disease Control and Prevention notes that dengue and West Nile virus are both moving from developing countries towards the United States, where concerns over mosquito exposure are rising. Malaria—which by various estimates kills between 1 million and 3 million people worldwide each year—is also a growing problem in many regions. This is in part because Plasmodium, the mosquito-borne parasite that causes malaria, is fast becoming resistant to existing treatments, such as chloroquine.
Many experts believe that better repellents could help to control mosquito-borne diseases. These next-generation repellents must be effective against anopheline mosquitoes that carry malaria. They should also be cheap and nontoxic, and should last long enough to protect humans as they sleep, when they are most vulnerable.
Where will these repellents come from? The answers, scientists increasingly say, will be found in genomic research. According to this view, knowledge of the genes and proteins that mosquitoes rely on to sense their environment could lead to new repellents that directly interfere with the insects’ ability to detect human beings.
A Sense of the Future
Today, genomic information about mosquitoes is accruing rapidly. Scientists with the International Anopheles Genome Project, a consortium based at the Pasteur Institute in Paris, have already decoded the genome of An. gambiae, the dominant malaria vector in Africa. Another genome, for Aedes aegypti, a vector for both yellow fever and dengue, is now being sequenced with funding from the National Institute of Allergy and Infectious Diseases. Using these data sets, scientists are identifying the genes that control mosquito sensory systems, including the olfactory system.
Mosquitoes rely on smell to guide them towards mates, food, and of course, sources of blood meals. The process is highly specific. For example, only female mosquitoes are attracted to blood sources, which they require as nourishment for their eggs. (Mosquitoes’ main nutriment comes from other sources, mainly plant nectar.) Olfaction is also species-specific. An. gambiae, for instance, prefers to bite humans. “You could be in a room full of cows, and that mosquito will find you and bite you,” says Laurence Zwiebel, an associate professor of biology at Vanderbilt University. Because this species has evolved to target humans, it will stay indoors where it can get to them more easily, he adds.
To better understand the molecular biology of mosquito olfaction, consider the following scenario: You’re asleep in the summertime, in a warm room with an open window. Your body and the bacteria that reside on your skin are giving off a molecular cloud of human-specific odorants. These molecules drift through the air and soon reach a mosquito perched on the wall. In an instant, odorant-binding proteins (OBPs) located on the creature’s antennae bind the molecules and transport them towards receptors located on the surfaces of the olfactory neurons. The odorant–receptor linkage activates a cue in the mosquito’s nervous system, which alerts the insect to your presence. Moments later, the mosquito lands on your unprotected shoulder and begins to feed.
OBPs are now widely viewed as one of the most likely targets for next-generation repellents. Compounds that interfere with these proteins could block mosquitoes’ ability to smell and thus detect humans. “This really is a unique approach,” says Leslie Vosshall, head of the Laboratory of Neurogenetics and Behavior at Rockefeller University. “Since we now know quite a bit about the basic workings of insect olfactory systems, we can focus our search for new repellents that interrupt proteins known to be important for smell. We’re optimistic that this rational approach to repellent design will uncover new compounds that are safer and more effective than those that are currently available.”
Targeting Olfactory Proteins
Vosshall has found that the function of a wide range of odorant receptors in insects depends on a single type of protein “coreceptor” that facilitates the binding of the OBP–odorant complex to the receptor. Working with fruit flies, she has shown that mutations in this protein, which in fruit flies is called Or83b, knocks out the insect’s sense of smell altogether. The protein, she says, is highly conserved across many different species; in mosquitoes, it now goes by the name GPRor7 (until recently, it was known as AgOR7). In recent studies, Vosshall generated a strain of fruit flies that lacked the Or83b gene and hence the ability to smell. This finding, remarkable in its own right, was rendered even more so by a subsequent discovery: when she replaced the missing Or83b gene with the GPRor7 gene obtained from mosquitoes, the fruit flies’ olfactory system was restored to a normal state. This means GPRor7 substituted for Or83b, even though mosquitoes and fruit flies diverged more than 250 million years ago.
Based on these findings, Vosshall suggests that GPRor7 is a plausible target for a mosquito repellent. Knock out this protein, she says, and a mosquito will be unable to smell anything—humans included. Her findings are published in the 22 February 2005 issue of Current Biology.
Zwiebel and his colleagues are taking a different tack. Instead of targeting a single protein that regulates mosquitoes’ entire sense of smell, he’s homing in on the olfactory proteins that bind human odorants only. In the 15 January 2004 issue of Nature, Zwiebel published results showing that one olfactory receptor in An. gambiae—a protein known as Or1—preferentially binds 4-methylphenol, one of roughly 300 compounds found in human sweat. Moreover, the Or1 receptor was found only in female mosquitoes—further evidence of the intricate specificity in the insect’s olfactory machinery.
Today, Zwiebel and his colleagues are striving to identify other human-specific odorants with the goal of creating repellents that inactivate groups of these targets simultaneously. This approach lessens the likelihood that mosquitoes will develop resistance to the repellents, he says. “It’s almost impossible for resistance mutations to appear in four to five receptors concurrently, and they would not make their way into the mosquito population,” he explains. “What we really want is a toolbox of behaviorally disruptive olfactory compounds,” he adds. “This would allow us to fine-tune repellent blends for specific geographic regions, where mosquito populations and olfactory mechanisms might vary.”
The future repellents mark an additional departure because they could be incorporated into time-release systems that put active ingredients into the air. Thus, there would be no need to apply the repellents to the body; instead, they would protect a living space in its entirety. This is important for those mosquitoes—An. gambiae in particular—that prefer indoor environments and sleeping humans.
Scratching the Surface
Experts agree that no new repellent will necessarily be a panacea for mosquito control. Andrew Spielman, a professor of tropical public health at the Harvard University School of Public Health and author of the book Mosquito: A Natural History of Our Most Persistent and Deadly Foe, suggests that all repellents pose a fundamental dilemma: mosquitoes deterred from one protected person will simply flock in greater numbers to another who is unprotected. “What drives the force of disease transmission is the biting rate,” he explains. “If half the people are being bitten by all the mosquitoes, then those people and any mosquitoes that feed on them will wind up being terrifically infected.” But Spielman acknowledges that repellents can be very useful—“I’m just not sure about who specifically is going to benefit from them,” he says.
When posed with this question, Zwiebel responds that health officials must take steps to ensure that no one is left unprotected. “That’s a challenge we need to accept,” he says. “We need to make sure these products are universally available and economically affordable.”
In the end, new repellents will be just part of a broader strategy to control mosquitoes, a strategy that Spielman says must also incorporate better environmental management, housing improvements, and greater use of insecticide-treated bed nets. Meanwhile, it’s hard to say which type of rationally designed repellent is likely to emerge first. Will it target one protein, as Vosshall suggests it could, or will it target several, as Zwiebel says it must? “At the end of the day,” Zwiebel says, “everyone is going forward with the best of intentions, and we just have to see what comes out.”
Tuned in to scents. Mosquitoes “smell” with their antennae.
Reception conservation. Leslie Vosshall and colleagues (left) developed a fruit fly strain lacking the Or83b gene. The mutation means the fruit flies’ odorant receptors (above, stained red) are not properly localized in the sensory hairs, and the flies do not respond when exposed to banana odor. When GPRor7, the mosquito equivalent of Or83b, is put into the mutant fruit flies, the odorant receptors return to the sensory hairs, and the flies now respond to banana odor.
The hair of the mosquito that bit you? The antenna of the female Anopheles mosquito (above right and inset) bristles with various olfactory sensillae used for prey detection, flight direction, and egg laying. Laurence Zwiebel (above left, with mosquitoes) is targeting human-specific olfactory receptors in these mosquitoes.
==== Refs
Suggested Reading
Jones WD Nguyen TA Kloss B Lee KJ Vosshall LB 2005 Functional conservation of an insect odorant receptor gene across 250 million years of evolution Curr Biol 15 4 R119 R121 15723778
Justice RW Biessmann H Walter MF Dimitratos SD Woods DF 2003 Genomics spawns novel approaches to mosquito control BioEssays 25 1011 1020 14505368
Spielman A D’Antonio M 2001. Mosquito: A Natural History of Our Most Persistent and Deadly Foe. New York, NY: Hyperion Books.
|
16002364
|
PMC1257667
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A468-A471
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.113-a468
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a00473EnvironewsScience SelectionsQuestioning Lead Standards: Even Low Levels Shave Points off IQ Adler Tina 7 2005 113 7 A473 A474 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
The maximum blood lead concentration deemed acceptable for children has declined over the years, from 60 micrograms per deciliter (μg/dL) in 1970 to the present-day level of 10 μg/dL, first established in 1991. In the last several years, however, researchers have begun to suspect that even lower concentrations may impair cognition. Now a reevaluation of data from seven international longitudinal studies involving 1,333 children confirms this suspicion [EHP 113:894–899].
The studies—conducted in Boston, Cincinnati, Cleveland, Rochester (New York), Port Pirie (Australia), Mexico City, and Yugoslavia—originally looked at children known to be at risk for lead poisoning, such as those living near lead smelters or in deprived urban settings. Therefore, the majority of the participants had blood lead levels far higher than the averages currently being reported in the developed world. The mean blood lead concentration for the entire group peaked at 17.8 μg/dL at age 2.5 years, and declined to 9.4 μg/dL between ages 5 and 7. Only 18% of the children had maximal blood lead levels of less than 10 μg/dL, and 8% had maximal blood lead levels of less than 7.5 μg/dL.
Most of the children took IQ tests when they were between almost 5 and 7 years of age; the Boston children were tested at age 10. The current team calculated, across the seven studies, how much of the difference in IQ scores was related to lead alone by controlling for other factors that influence IQ scores, including child birth weight, birth order, prenatal exposure to tobacco smoke and alcohol, and mother’s IQ.
On a population basis, an increase in blood lead level from 2.4 to 10 μg/dL at the time of testing was associated with a decrease of 3.9 IQ points. At lower blood lead levels, a small increase in blood lead made a bigger difference in IQ than the same size increase did at higher concentrations. A blood lead level of 20 μg/dL was associated with scoring about 1.9 points lower on tests of IQ compared with a blood lead level of 10 μg/dL. The difference in IQ shrank to 1.1 points when comparing a blood lead level of 20 μg/dL with a concentration of 30 μg/dL.
To determine if the data from one particular study drove the final results, the team removed the findings for one site at a time and recalculated the results. It became clear that no single study was driving the results of the pooled analysis.
Consistent with a study published in the May 2005 issue of EHP, blood lead level at the time of IQ testing was generally a stronger predictor of effects on IQ than was—as previously believed—blood lead level at age 2. The individual-level effect on IQ is difficult to determine, however, and may depend in part on the child’s social environment.
In the United States, about 2–3% of children have a blood lead concentration above 10 μg/dL, but in some cities, such as Rochester, 1 in 5 children have elevated blood lead. These new findings, along with those from previous human and animal studies, point to the importance of eliminating nonessential uses of lead and lowering allowable levels of lead in air emissions, house dust, soil, water, and consumer products.
The test of time. More studies are confirming the surprise finding that blood lead concentration at the time of IQ testing, not peak level, is a better predictor of IQ effects.
|
0
|
PMC1257668
|
CC0
|
2021-01-04 23:41:04
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A473-A474
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a00474EnvironewsScience SelectionsTap Water and Trihalomethanes: Flow of Concerns Continues Hood Ernie 7 2005 113 7 A474 A474 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Trihalomethanes (THMs) are the result of a reaction between the chlorine used for disinfecting tap water and natural organic matter in the water. At elevated levels, THMs have been associated with negative health effects such as cancer and adverse reproductive outcomes. Now a study by government and academic researchers adds to previous evidence that dermal absorption and inhalation of THMs associated with everyday tap water use can result in significantly higher blood THM concentrations than simply drinking the water does [EHP 113:863–870]. The results of this exposure assessment study could serve as a guide for future epidemiologic investigations exploring the potential connection between THMs in tap water and adverse health effects.
The researchers recruited seven healthy participants aged 21–30 years to spend two 24-hour periods (usually one week apart) in one of two study residences. The study residences were typical ranch-style homes, one located in North Carolina, the other in Texas. The North Carolina house was served with a water supply higher in THMs than that of the Texas house.
Over the total two days, each participant performed 14 activities using tap water. These included drinking a hot beverage prepared with tap water (THM-free bottled water was consumed except when drinking was part of a test activity), washing their hands, showering, washing dishes both by hand and in a dishwasher, and washing clothes in a washing machine. The water use activities were rigidly scheduled and controlled for exposure time and water temperature.
The team took baseline measurements of the THMs in ambient indoor air, cold tap water, and subjects’ blood and exhaled breath just before and just after each activity. The ratio between pre- and post-activity measurements illustrated the impact of each activity on participants’ blood and exhaled breath THM concentrations; twofold or greater deviation from baseline was established as meaningful.
Relatively high pre- to postactivity ratios were observed for several of the activities. For example, blood concentrations rose 5- to 15-fold as a result of showering in the North Carolina participants, and rose approximately 5-fold in the Texas subjects. The results confirm that showering and bathing are important sources of THM exposure; they also provide evidence that other THM exposure scenarios, such as washing dishes by hand and being exposed to a cohabitant’s shower steam, may also be important.
Although an apparent dose–response relationship was discovered, the authors emphasize that public health implications should not be inferred from their findings, partly due to the small number of subjects. Their purpose was to shed light on which water use activities should be considered in the context of an epidemiologic study and to establish some practical approaches for future investigations. Noting the wide range in blood THM concentrations among the subjects in this and other studies in response to similar levels of THM exposure, subsequent exposure assessment research is being conducted on the possibility that genetic variation may play a role in individuals’ susceptibility to absorption of THMs.
Are shower buffs in hot water? Household uses of hot tap water such as showering and dish washing result in greater THM absorption than simply drinking the water.
|
0
|
PMC1257669
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A474
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a00475AnnouncementsNIEHS Extramural UpdateThe Superfund Basic Research Program—A Time of Change 7 2005 113 7 A475 A475 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
It is a time of new beginnings for the Superfund Basic Research Program, a university-based grants program established in 1987. While maintaining the program’s premise of supporting basic research for practical application to address the problems associated with the nation’s hazardous waste sites, the program continues to evolve and develop new approaches to address these concerns. An important component of achieving the program’s goals is the routine recompetition of the program. We are pleased to announce that as a result of the most recent recompetition, NIEHS has made awards to nine programs. Included in these nine awards are one new grantee, Brown University, and eight grantees that are continuing participants in the program. These grantees include Boston University, Dartmouth College, Duke University, Texas A & M University, the University of Arizona, the University of California at Davis, the University of California at San Diego, and the University of Kentucky. These nine awards in addition to the eleven existing grantees comprise the current Superfund Basic Research Program.
The new program has a strong emphasis on interdisciplinary and system approaches to addressing the complex issues associated with hazardous waste sites. Grantees are using integrated research models to understand how contaminants are transformed as they move through soils, sediments, and groundwater and how they interact with ecosystems and ultimately affect human health. These robust research efforts are augmented with graduate and postdoctoral training in interdisciplinary environmental research and with community outreach activities. Each grantee is also required to engage proactively in research translation to other federal agencies, industry, and the community at large.
For interested investigators who are not already part of this vital and exciting program but would like to be, please contact program staff about future opportunities. We anticipate that there will be annual solicitations beginning in the summer of 2006. It is not too early to start the planning process! We are available to assist you as you begin to conceptualize a future program at your university.
Contact
William Suk, PhD, Director | (919) 541-0797
Claudia Thompson, PhD | (919) 541-4638
Beth Anderson | (919) 541-4481
Superfund Program sites.
Students participating in a phytoremediation project as part of the University of Washington program.
|
0
|
PMC1257670
|
CC0
|
2021-01-04 23:41:04
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A475
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a00476AnnouncementsFellowships, Grants, & AwardsFellowships, Grants, & Awards 7 2005 113 7 A476 A477 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Stem Cells and Cancer
The clonal nature of most malignant tumors is well established. Experiments spanning several decades have shown, however, that as many as one million murine or human tumor cells are required to transplant a new tumor from an existing one. Two theories have been developed to account for the observation that apparently not every tumor cell is a tumor initiating cell (T-IC). The stochastic theory predicts that every tumor cell can form an entirely new tumor; however, entry into the cell cycle is a stochastic event with low probability. Alternatively, tumor cells may exist in a hierarchical state in which only a small number of cells possess tumor initiating potential. If the stochastic model is correct, then tumor cells are biologically homogeneous and genetic or epigenetic programs that allow for tumorigenesis are operative in the majority of cells that comprise a tumor. The hierarchical model, however, predicts the tumor cells possess a functional heterogeneity and that quantitatively the cells capable of tumorigenesis are a relatively minor population among the bulk of tumor cells.
Recent data from both hematologic malignancies and solid tumors have suggested that there are only minor populations of cells in each malignancy that are capable of tumor initiation. These T-ICs have the functional properties of tumor stem cells. They appear to be capable of asymmetric division and self renewal, are only a minor faction among the bulk of more differentiated cells in the tumor, and can reconstitute all the cell types in the tumor of origin.
Currently, tumor stem cells have been isolated and characterized in several hematologic malignancies and some solid tumors. One of the first tumors in which a stem cell was identified was acute myeloid leukemia (AML). In this disease, the frequency of the leukemic stem cell (LSC) was approximately 1 per million AML blasts, establishing that not every AML cell had T-IC capacity. A CD34 positive/CD38 negative cell fraction representing 0.1 to 1 percent of the tumor cells possessed all the leukemia initiating activity in the NOD/SCID transplantation model. By contrast, the major fraction of the CD34 positive/ CD38 positive cells and the majority of CD34 negative cells, which comprise most of the cells in the tumor, could not initiate leukemia. A multiple myeloma stem cell has also been characterized. Multiple myeloma cell lines and primary patient derived cells express the cell surface marker syndecan-1 (CD138). A population of cells representing less than 5 percent of the cells in the bulk population of multiple myeloma cells was found to be CD138 negative, and possessed in vitro clonogenic potential. These cells also engrafted successfully into Non-obese Diabetic/Severe Combined Immunodeficient (NOD/SCID mice), whereas CD138 positive cells did not engraft.
Recently, a mammary carcinoma stem cell has been isolated from primary mammary carcinomas using four cell surface markers (CD44; CD24; a mammary tumor marker, and epithelial specific antigen). The tumor initiating capacity of the cells was again verified in an in vivo NOD/SCID engraftment assay. The mammary tumor stem cells represented only 2 percent of the unfractionated bulk tumor cells.
Finally, a putative brain tumor stem cell has also been isolated. These cells appear to be between 0.3 to 25 percent of the cells in the brain tumors examined. They are positive for the neural stem cell marker CD133 and have a marked capacity for self renewal and differentiation. Transplantation of these putative neural tumor stem cells into the forebrains of NOD/SCID mice yields tumors phenotypically resembling the tumors from which the stem cells were isolated.
Isolation of tumor stem cells from a larger spectrum of solid tumors and characterization of markers for such cells will be important in understanding how general the role of tumor stem cells are in the pathogenesis of cancer.
In addition to isolating and characterizing tumor stem cells themselves, it is also important to identify the genes and proteins that facilitate the self renewal phenotype that characterize all stem cells. The proteins involved in self renewal in embryonic and adult stem cells appear to be subverted in tumorigenesis to allow the tumor-initiating cells to maintain self renewal capacity. Two families of proteins related to self renewal, the polycomb gene Bmi-1 and the Wnt signaling pathway proteins, have been related to the maintenance of the tumor stem cell phenotype.
The polycomb genes have an essential role in embryogenesis, regulation of the cell cycle, and lymphopoieisis. These genes are transcriptional repressors that are essential for the silencing of other families of genes. Deletion of polycomb gene Bmi-1 in mice results in a progressive loss of all hematopoietic lineages. This loss results from the inability of the Bmi-1(−/ −) stem cells to self renew. Introducing genes known to produce AML into Bmi-1(−/ −) hematopoeitic stem cells (fetal liver cells), induced AML with normal kinetics; however, the Bmi-1(−/ −) leukemic stem cells from primary recipients were unable to produce AML in secondary recipients. These results demonstrate that Bmi-1 is also required for self renewal of leukemic stem cells in the murine AML model.
Another group of genes involved in self renewal are those involved in the Wnt signal transduction cascade. The Wnt protein binds to a receptor called Frizzled and activates cell fate decisions during tissue development. Inhibitors of Wnt signaling produce inhibition of hematopoietic stem cell growth in vitro and reduce hematopoietic reconstitution in vitro. Activation of Wnt signaling in hematopoeitic stem cells leads to increased expression of Hox B4 and Notch-1 genes previously implicated in self renewal of hematopoietic stem cells. The Wnt signaling pathway is involved in both hematopoietic malignancy and colon carcinoma. Although the Wnt ligands themselves are only rarely involved in tumorigenesis, mutations mimicking Wnt receptor (Frizzled) activation induce a set of genes associated with repression of differentiation and potentiation of self-renewal. In general, these mutations involve Wnt signal transduction proteins including the activation of beta-catenin and inactivation of the (APC) adenomatosis polyposis coli protein. In myeloid leukemia, nonphosphorylated beta-catenin accumulates in granulocyte-macrophage progenitors as they progress toward leukemia. These normally more committed progenitors can thus acquire self-renewal properties. A similar accumulation of unphosphorylated beta-catenin has also been observed in multiple myeloma cells. In colon cancer, the APC gene is mutated early in the development of about 90 percent of the colon carcinomas. Similarity in gene expression patterns between populations of colon cancer cells and colon epithelial stem cells has also been observed by DNA microarray analysis. It is possible that mutations in the Wnt signaling pathway maintain the program of stem cell genes in the transcriptionally active state.
Additional research on the important genes and proteins that function to maintain the stem cell phenotype could contribute to increasing the number of specific targets for which cancer therapeutic agents can be designed.
Recent studies of Helicobacter infection in mice have suggested that bone marrow hematopoietic stem cells can contribute to repopulating the gastric mucosal epithelium. These cells can progress, with time, to metaplasia, dysplasia, and cancer. The possibility of stable fusion between the bone marrow derived cells and the gastric mucosa was eliminated, suggesting the possibility of transdifferentiation of hematopoietic stem cells. Research aimed at establishing whether or not stem cells have the capacity to transdifferentiate is also encouraged.
This funding opportunity is intended to promote research on all aspects of tumor stem cell biology, and on the genes and proteins responsible for the tumor stem cell phenotype. Research studies on the characterization of tumor stem cells from the broad spectrum of solid and liquid tumors not already examined, on markers potentially shared by tumor stem cells and normal stem cells, and on the biochemical and molecular regulation of normal and tumor stem cell function are encouraged. Such research can and should include research on in vivo assays for the functional identification of such cells. Studies of the genes regulating self renewal, and studies of regulation of stem cell division by the stem cell niche and/or microenvironment are also encouraged. Investigators working on the cell and molecular biology of embryonic stem cells, adult stem cells, and tumor stem cells are encouraged to apply for support under this funding opportunity. The following questions illustrate areas of high interest, but other relevant innovative projects are also encouraged: 1) What governs the proliferation rate of normal and tumor stem cells? 2) Can oncogenes and their associated mutations affect asymmetric versus symmetric divisions in stem cells? 3) Stem cell quiescence versus growth must ultimately be understood in terms of progression through the cell cycle. Which stem cell-specific genes alter the cell cycle pathway proteins? 4) Do tumor stromal cells constitute a unique tumor stem cell niche? Does the tumor stromal niche act as a constituent of a feedback mechanism with tumor stem cells to control their growth? 5) Are the phenotypes of invasion and metastasis uniquely connected to the tumor stem cell phenotype? 6) Are normal resident adult tissue stem cells a special target for carcinogenic insults? 7) Can new and/or better markers and assays for the isolation and enrichment of tumor stem cells be developed? 8) Can new and/or better in vivo functional assays to identify tumor initiating cells (e.g., engraftment of leukemic stem cells into immunodeficient NOD/SCID mice) be developed? 9) How do changes to stem cells or their environment due to aging affect formation of tumor stem cells or alter their properties?
This funding opportunity will use the R01 and R21 award mechanisms. As an applicant, you will be solely responsible for planning, directing, and executing the proposed project.
This funding opportunity uses just-in-time concepts. It also uses the modular as well as the non-modular budget formats (see http://grants.nih.gov/grants/funding/modular/modular.htm). Specifically, if you are submitting an application with direct costs in each year of $250,000 or less, use the modular budget format described in the PHS 398 application instructions. Otherwise, follow the instructions for non-modular research grant applications.
The PHS 398 application instructions are available at http://grants.nih.gov/grants/funding/phs398/phs398.html in an interactive format. Applicants must use the currently approved version of the PHS 398. For further assistance, contact GrantsInfo at 301-435-0714, (telecommunications for the bearing impared: TTY 301-451-0088) or by e-mail: [email protected].
Applications must be prepared using the most current PHS 398 research grant application instructions and forms. Applications must have a D&B Data Universal Numbering System (DUNS) number as the universal identifier when applying for Federal grants or cooperative agreements. The D&B number can be obtained by calling 866-705-5711 or through the web site at http://www.dnb.com/us/. The D&B number should be entered on line 11 of the face page of the PHS 398 form. The complete version of this PA is available at http://grants/guide/pa-files/PAR-05-086.html.
Contact: R. Allan Mufson, Cancer Immunology Hematology Branch, Division of Cancer Biology, National Cancer Institute, 6130 Executive Blvd, EPN Rm 5062, Bethesda, MD 20892 USA, Rockville, MD 20852 USA (for express/courier service) 301-496-7815, Fax: 301-480-2844, E-mail:
[email protected]; Jill Carrington, Systems Branch, Musculoskeletal Biology, Biology of Aging Program, National Institute on Aging, National Institutes of Health, 7201 Wisconsin Ave, Suite 2C231, MSC 9205, Bethesda, MD 20892 USA, 301-496-6402, fax: 301-402-0010, e-mail:
[email protected]. Reference: PA No. PA-05-086
Secondary Analyses in Obesity, Diabetes, Digestive and Kidney Diseases
The specific objectives of this announcement on Secondary Analyses In Obesity, Diabetes, Digestive And Kidney Diseases are to support the following: (a) research on secondary analyses of data related to the epidemiology of disease areas of NIDDK; (b) preliminary projects using secondary analysis that could lead to subsequent applications for individual research awards; (c) rapid analyses of new databases and experimental modules to inform the design and content of future studies; (d) the archiving of data sets to be made publicly available for research purposes related to disease areas of NIDDK, including both epidemiological studies and multi-center clinical trials.
Research that employs analytic techniques that demonstrate or promote methodological advances in patient-oriented and epidemiologic research is also of interest. International comparative analyses are encouraged. Applications that are innovative and high risk with the likelihood for high impact would be especially encouraged.
Patient-oriented and epidemiologic research projects, particularly multicenter projects, typically generate data with potential utility beyond the specific hypotheses and questions for which they were designed. Often data are not fully analyzed, especially when unexpected research questions emerge after the end of the project's funding period. Analyzing such existing data sets can therefore provide a cost-effective means to test specific hypotheses that have not been adequately examined. The further analysis of existing research data may also be prompted by a need to confirm new findings or to aid in the development of new research questions.
Applicants may conduct secondary analyses using data from a variety of sources. These would include investigator-initiated research activities, cooperative agreements, and contracts or from other public or private sources. Sources may be large, nationally representative data sets such as those of the National Center for Health Statistics or smaller, regional or locally based data sets. Also appropriate for secondary analyses are relevant cross-sectional and longitudinal survey data collected by federal, state, and local government agencies. Secondary data analyses of these data may serve as an economical alternative to expensive and time-consuming new data collection projects. Applicants may also secure access to other data sets that may or may not be in the public domain, such as those collected under research grant funds, sponsored by private entities, or originally collected for purposes other than research, such as health care administrative data sets.
In addition to the examination of specific research hypotheses, existing data sets may also be used to cross-validate exploratory analyses in ongoing studies, to test complex statistical models, and in special circumstances to provide comparison groups for experimental studies. Moreover, secondary analysis is appropriate for many types of data, including qualitative information, and may also cover the integration of quantitative and qualitative data.
NIDDK has established a repository for the archiving of data sets, as well as genetics and tissues from NIDDK sponsored clinical trials and epidemiological studies (http://pubnts06.rti.org/niddk/home.do). Applicants are encouraged to consider the research opportunities available in this NIDDK resource.
A major interest of NIDDK is supporting secondary data analyses in the causes, burden, natural history, and treatment and medical care of overweight and obesity, including analyses of behavioral/environmental factors that may be predictive of long term weight maintenance or prevention of weight gain. Other specific subject areas are restricted to those on which NIDDK conducts research, which include diabetes and endocrine and metabolic diseases; digestive diseases and nutritional disorders, including eating disorders; and kidney, urological, and hematological diseases. All data analyses must concern patient-oriented or epidemiologic research designed to elucidate the etiology, incidence, prevalence, natural history, pathophysiology, or response to therapy of the above-mentioned disorders.
This program announcement addresses several areas considered to be high priority for liver disease research as delineated in the recently published Trans-NIH Action Plan for Liver Disease Research (http://liverplan.niddk.nih.gov), specifically in the areas of fatty liver disease, viral hepatitis, drug- and toxicant-induced liver disease, autoimmune liver disease, pediatric liver disease, liver transplantation, complications of liver disease, and gallbladder and biliary disease.
This mechanism can also be used for the merging of secondary data sets with other data sets. For example, if allowed by informed consent, databases could be matched with hospital data sets or vital statistics to assess longer-term morbidity and/or mortality of patient groups. Meta-analysis, in which results from multiple studies may be compared or combined, is encouraged only if patient level data from the original studies are combined.
Support for the creation of publicly archived data sets involved in secondary analyses would be for information that could be applied to the subject areas of interest. Plans for archiving must include adequate dataset documentation and explanation so that it can be used by researchers not associated with the original study. Provision for easy accessibility of archived datasets is required.
Up to 25% of the direct costs of the grant may be spent on acquiring new information that would be incorporated into the database and would significantly strengthen the analysis. Such information may, for example, be derived from laboratory testing of stored specimens or comparisons of a measurement against a criterion standard to validate the measurement in the database. Conversion of data from paper records to electronic form would also be considered up to the 25% limit. Obtaining such new information must serve the purpose of the secondary data analysis and should not be considered for any other reason.
This funding opportunity will use the NIH Exploratory/Developmental (R21) award mechanism. As an applicant, you will be solely responsible for planning, directing, and executing the proposed project.
This funding opportunity uses just-in-time concepts. It also uses the modular budget format described in the PHS 398 application instructions (see http://grants.nih.gov/grants/funding/modular/modular.htm).
The PHS 398 application instructions are available at http://grants.nih.gov/grants/funding/phs398/phs398.html in an interactive format. Applicants must use the currently approved version of the PHS 398. For further assistance contact GrantsInfo at 301-435-0714 (telecommunications for the hearing impaired: TTY 301-451-0088) or by e-mail: [email protected].
Applications must be prepared using the most current PHS 398 research grant application instructions and forms. Applications must have a D&B Data Universal Numbering System (DUNS) number as the universal identifier when applying for Federal grants or cooperative agreements. The D&B number can be obtained by calling 866-705-5711 or through the web site at http://www.dnb.com/us/. The D&B number should be entered on line 11 of the face page of the PHS 398 form. The complete version of this PA is available at http://grants.nih.gov/grants/guide/pa-files/PA-05-091.html.
Contact: James E. Everhart, Epidemiology and Clinical Trials Branch, Division of Digestive Diseases and Nutrition, National Institute of Diabetes and Digestive and Kidney Diseases, 6707 Democracy Blvd, Rm 655, Bethesda MD 20892-5450 USA, 301-594-8878, fax: 301-480-8300, e-mail: [email protected]; Catherine C. Cowie, Type 1 Diabetes Clinical Trials Program, Division of Diabetes, Endocrinology, and Metabolism, National Institute of Diabetes and Digestive and Kidney Diseases, 6707 Democracy Blvd, Rm 691, Bethesda, MD 20892-5460 USA, 301-594-8804, fax: 301-480-3503, e-mail: [email protected]; Paul W. Eggers, Epidemiology and U.S. Renal Data System, Division of Kidney, Urologic, and Hematologic Diseases, National Institute of Diabetes and Digestive and Kidney Diseases, 6707 Democracy Blvd, Rm 615, Bethesda, MD 20892-5458 USA, 301 594-8305, fax: 301-480-3510, e-mail:
[email protected]. Reference: PA No. PA-05-091
|
0
|
PMC1257671
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A476-A477
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0442a16002357PerspectivesCorrespondenceUrinary Creatinine and Arsenic Metabolism Gamble Mary V. Liu Xinhua Department of Environmental Health Sciences, Department of Biostatistics, Mailman School of Public Health, Columbia University, New York, New York, E-mail:
[email protected] authors declare they have no competing financial interests.
7 2005 113 7 A442 A442 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Urinary creatinine is almost universally employed to adjust concentrations of urinary analytes for variations in hydration status. In the February 2005 issue of EHP, Barr et al. used data from the Third National Health and Nutrition Examination Survey (NHANES III) to establish reference ranges for urinary creatinine for specific age and demographic categories (Barr et al. 2005). They reported that the significant predictors of urinary creatinine concentrations include age, sex, race/ethnicity, body mass index, and fat-free mass. Although these indicators have been known for many years, the unintentional adjustment for these covariates when urinary metabolites are expressed per gram creatinine can have profound effects on the interpretation of data. The fact that these effects are often under-appreciated or even unnoticed renders this paper highly relevant to exposure assessment and well worth revisiting. In our studies of arsenic methylation and one-carbon metabolism, we have noted several additional complications when expressing urinary arsenic as micrograms per gram creatinine. Note that one-carbon metabolism refers to the folate-dependent biochemical pathway responsible for methylation of DNA, arsenic, and hundreds of other substrates.
Our study in Bangladesh on 1,650 adults revealed that urinary creatinine concentrations are significantly correlated with plasma folate concentrations—particularly among males, who had a higher prevalence of folate deficiency than females in Bangladesh (Gamble et al. 2005). Although this association had not been previously reported, it is not surprising considering that the formation of creatine from methylation of guanidino-acetate accounts for approximately 75% of all folate-dependent transmethylation reactions (Mudd and Poole 1975) and that creatine is the direct precursor of creatinine. In some analyses, adjusting urinary arsenic for creatinine obscured correlations between folate and arsenic metabolism. In other analyses, correlations between folate and arsenic/creatinine were due in part to the associations between folate and creatinine. Correct interpretation of the data would not be possible without considering the impact of the correlation between urinary creatinine and plasma folate. As did Barr et al. (2005), we decided to include urinary creatinine in the statistical models as a separate independent variable. However, because of the intimate link between creatine metabolism and one-carbon metabolism, inclusion of urinary creatinine in some models resulted in overcontrolling for the effects of folate and homocysteine, our variables of interest. Thus, expression of total urinary arsenic per gram creatinine runs the risk of confounding relationships between total urinary arsenic and arsenic metabolism. Adjusting for the specific gravity of urine was not useful because it is so highly correlated with urinary creatinine.
In summary, we concur with Barr et al. (2005) that urinary creatinine should be included in multiple regression models as a separate independent variable; in addition, the role of one-carbon metabolism as a predictor of urinary creatinine should also be considered in interpreting results. Specifically, we routinely test if urinary creatinine itself is a predictor of the outcomes of interest.
==== Refs
References
Barr DB Wilder LC Caudill SP Gonzalez AJ Needham LL Pirkle JL 2005 Urinary creatinine concentrations in the U.S. population: implications for urinary biologic monitoring measurements Environ Health Perspect 113 192 200 15687057
Gamble MV Ahsan H Liu X Factor-Litvak P Ilievski V Slavkovich V 2005 Folate and cobalamin deficiencies and hyperhomocysteinemia in Bangladesh Am J Clin Nutr 81 1372 15941889
Mudd SH Poole JR 1975 Labile methyl balances for normal humans on various dietary regimens Metabolism 24 721 735 1128236
|
16002357
|
PMC1257672
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A442a
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.113-a442a
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0442b16002357PerspectivesCorrespondenceUrinary Creatinine: Barr et al. Respond Barr Dana B. Caudill Samuel P. Jones Robert L. Pfeiffer Christine M. Pirkle James L. National Center for Environmental Health Centers for Disease Control and Prevention, Atlanta, Georgia, E-mail:
[email protected] Lynn C. Needham Lance L. Agency for Toxic Substances and Disease Registry, Atlanta, GeorgiaThe authors declare they have no competing financial interests.
7 2005 113 7 A442 A443 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Although individual predictors of urinary creatinine such as sex, body mass index, and age have been reported, no single research endeavor has documented the predictors in one study population as thoroughly as we reported in our recent article (Barr et al. 2005). The large volume of data available in the Third National Health and Nutrition Examination Survey (NHANES III; 1988–1994) [Centers for Disease Control and Prevention (CDC) 2003a] was ideal for examining and documenting these predictors. To date, our study provides the most concrete data in the literature demonstrating creatinine variation in diverse populations and the factors contributing to this variation. We agree with Gamble and Liu that although many research articles have recognized differences in creatinine concentrations within their study populations, few have attempted to correct for this variation. Our analysis of urinary creatinine concentration data in a large, representative segment of the U.S. population was intended to highlight the problems that can be encountered when routinely correcting urinary analyte concentrations for creatinine; however, Gamble and Liu point out in their letter yet another complication that may be encountered when evaluating urinary concentrations of chemicals that undergo a folate-mediated single-carbon metabolism. We are grateful that they alerted us of the possible complication of evaluating data for chemicals such as arsenic. Because folate is routinely measured in the ongoing NHANES cycles and speciated arsenic measurements have begun in the same samples, the role of one-carbon metabolism should certainly be considered in interpreting results for arsenic and other similarly metabolized chemicals for future editions of the CDC’s National Report on Human Exposure to Environmental Chemicals (CDC 2001, 2003b).
==== Refs
References
Barr DB Wilder LC Caudill SP Gonzalez AJ Needham LL Pirkle JL 2005 Urinary creatinine concentrations in the U.S. population: implications for urinary biologic monitoring measurements Environ Health Perspect 113 192 200 15687057
CDC 2001. National Report on Human Exposure to Environmental Chemicals. Atlanta, GA:National Center for Environmental Health, Centers for Disease Control and Prevention. Available: http://www.cdc.gov/nceh/dls/report/ [accessed 21 September 2002].
CDC 2003a. National Health and Nutrition Examination Survey. Hyattsville, MD:National Center for Health Statistics, Centers for Disease Control and Prevention. Available: http://www.cdc.gov/nchs/nhanes.htm [accessed 5 June 2003].
CDC 2003b. Second National Report on Human Exposure to Environmental Chemicals. Atlanta, GA:National Center for Environmental Health. Available: http://www.cdc.gov/exposurereport/2nd/www.cdc.gov/exposurereport [accessed 5 June 2003].
|
0
|
PMC1257673
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A442b-A443
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0444a16002359PerspectivesCorrespondenceBioremediation Monitoring Aust Steven D. Chemistry and Biochemistry Department, Utah State University, Logan, UT, E-mail:
[email protected] author declares he has no competing financial interests.
Editor’s note: In accordance with journal policy, Ganey and Boyd were asked whether they wanted to respond to this letter, but they chose not to do so.
7 2005 113 7 A444 A444 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
In their article published in the February issue of EHP, Ganey and Boyd (2005) made some excellent points about the potential pitfalls of simply assaying for the disappearance of an environmental pollutant during or as a result of bioremediation. This is important because it would be wrong to leave a metabolite that might pose as much or even more risk then the original chemical of interest.
Ganey and Boyd (2005) used the bio-remediation of polychlorinated biphenyls (PCBs) as an example, which was an excellent choice. However, the subject of metabolism of the PCB bioremediation metabolites should also be considered. As chlorines are removed by bioremediation, the less-chlorinated products could be more readily metabolized by many species exposed to the bioremediated material. That is, less-heavily chlorinated products (or intermediates) of bioremediation may be less toxic because of shorter half-lives due to metabolism. This phenomenon can be exemplified by work we conducted years ago at Michigan State University. We showed that 3,4,3′,4′-tetrabromobiphenyl was less toxic than 3,4,5,3′,4′,5′-hexabromobiphenyl, even though it was bound at higher affinity by the dioxin receptors because it was more readily metabolized and eliminated (Millis et al. 1985).
Commercial preparations contain few or no strictly coplanar PCB or polybrominated biphenyl congeners. This fact does not seem to be appreciated, and the impression is sometimes given that those very toxic congeners are in the environment. In fact, the coplanar polyhalogenated biphenyls probably receive way too much attention, most likely because they were used rather extensively in research; however, they were used only as model toxic congeners. The synthesis of strictly coplanar halogenated biphenyls (i.e., 3,4,3′,4′-PCB) is much different from that of the commercial preparations (which was by simple halogenation of biphenyl). Phenyl is strongly ortho-para directing, leading to non-coplanar halogenated biphenyls. The initial para and/or ortho halogenation makes for an even stronger ortho-para directive. Thus, the major components will be non-coplanar halobiphenyls. Only very small amounts of single ortho halobiphenyls can be found in commercial mixtures, and these mixtures are quite ineffective in eliciting effects associated with binding by the dioxin receptor.
==== Refs
References
Ganey PE Boyd SA 2005 An approach to evaluation of the effect of bioremediation on biological activity of environmental contaminants: dechlorination of polychlorinated biphenyls Environ Health Perspect 113 180 185 15687055
Millis CD Mills RA Sleight SD Aust SD 1985 Toxicity of 3,4,5,3′,4′,5′-hexabrominated biphenyl and 3,4,3′,4′-tetrabrominated biphenyl Toxicol Appl Pharmacol 78 88 95 2994254
|
16002359
|
PMC1257674
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A444a
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.113-a444a
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0444b16002359AnnouncementsErrataErrata 7 2005 113 7 A444 A444 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
In “Seasick Lungs: How Airborne Algal Toxins Trigger Asthma Symptoms” [
Environ Health Perspect 113:A324 (2005)], the accompanying photograph of Karenia brevis should have been credited to Daniel Baden/University of North Carolina at Wilmington.
“Linking Toenail Arsenic Content to Cutaneous Melanoma” [
Environ Health Perspect 113:A377 (2005)] should have clarified that Laura E. Beane Freeman received NIEHS funding while at the University of Iowa, before she joined the National Cancer Institute.
EHP regrets the errors.
|
0
|
PMC1257675
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A444b
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a00449a16018112EnvironewsForumDrinking Water: NAS Reports on Perchlorate Safety Dahl Richard 7 2005 113 7 A449 A449 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
A National Academy of Sciences (NAS) panel has issued a final report on the health implications of perchlorate ingestion, recommending a reference dose of 0.0007 milligrams per kilogram (mg/kg) body weight. But the debate over the health risks posed by the chemical, used by the Department of Defense as a rocket fuel additive, is far from over.
Perchlorate compounds have been used since the early 1900s, and environmental perchlorate contamination was first seen in 1985 in wells at California Superfund sites. Since then, perchlorate has been found in 35 states. In May 2004 the U.S. Environmental Protection Agency (EPA) estimated that more than 11 million Americans were drinking water from public supplies containing at least 4 parts per billion (ppb) perchlorate.
Scientists agree that perchlorate can interfere with the production of thyroid hormone since it competes for the uptake of iodide by the thyroid gland. But beliefs about what level of exposure constitutes a health risk vary widely. The Council on Water Quality (CWQ), a chemical and aerospace industry group, often cites a drinking water cutoff of 245 ppb. In contrast, California recommends that drinking water contain no more than 6 ppb perchlorate, and Massachusetts recommends that pregnant women and children not consume water with more than 1 ppb perchlorate.
The broad disagreement, coupled with the prospect of massive cleanup costs—estimated by some to be in the billions—prompted the government to ask the NAS for guidance. According to panel chair Richard B. Johnston, Jr., associate dean for research development at the University of Colorado School of Medicine in Denver, the 15-member group used as its starting point a September 2002 EHP study headed by Monte A. Greer of Oregon Health & Science University. This study was partially funded by the Perchlorate Study Group, an organization created by the Department of Defense and some of its contractors.
The Greer study concluded there was no inhibition of iodide uptake by the thyroid at 0.007 mg/kg body weight. The panel applied a 10-fold uncertainty factor to that figure to derive its own reference dose. “We took what we feel is the most conservative end point,” says Johnston. “It’s way short of any kind of harm.” Five weeks after the NAS made its report public, the EPA responded by adopting the NAS dose level and translating it into a drinking water equivalent level of 24 ppb.
But environmental groups have voiced heated disagreement with the NAS findings. Gina Solomon, a senior scientist at the Natural Resources Defense Council, says the report relied too heavily on a study she calls statistically flawed because of the small number of subjects (just 37). “As the effect [of perchlorate ingestion] gets more subtle, the size of the study group needs to be bigger to see if there’s an effect there or not,” she says.
Further, she says, the report suffers from tunnel vision: “[The NAS] should have been looking at the big picture on perchlorate, and they didn’t do that. The result was that their final report hinged entirely on one controversial industry study.”
Johnston responds that the panel also relied on four other clinical studies as well as several epidemiologic and perchlorate worker studies, all of which supported the Greer findings. And James Strock, a former secretary of the California Environmental Protection Agency who now works with the CWQ, says the NAS findings will provide state and federal regulators “a rare opportunity to promulgate regulations in a transparent manner, working simultaneously from information collected and considered by a world-class panel of experts.”
Johnston agrees, however, that more research will be helpful, especially on perchlorate’s effects on sensitive populations, such as pregnant women, nursing mothers, and infants. A study at Texas Tech University, published 1 April 2005 in Environmental Science & Technology, found that perchlorate levels in 36 samples of breast milk from nursing mothers in 18 states averaged 10.5 ppb, meaning the mothers were ingesting far more than 24 ppb. The study raises the possibility that some infants may be ingesting perchlorate at levels exceeding NAS and EPA safe doses.
Meanwhile, the controversy continues to play out, as described in an upcoming EHP commentary (doi:10.1289/ehp.8254, scheduled for publication in September 2005 and available in draft form at http://dx.doi.org/). Although the EPA has adopted the 24 ppb figure as an “official reference dose,” it’s not yet an enforceable standard, and Solomon says states are left to their own devices. “Some are following the EPA lead, and others are following the California lead,” she says. “This means that consumers in some states will be drinking water with higher levels of perchlorate than consumers in other states. And that’s unfortunate.”
|
16018112
|
PMC1257676
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A449a
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.113-1257676
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0449a16018112EnvironewsForumThe Beat Dooley Erin E. 7 2005 113 7 A449 A451 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
“Cabin Fever” Fears Unfounded
Ever wonder how many infectious organisms are riding along with you in the cabin of a commercial airliner? According to a literature review in the 12 March 2005 Lancet, although airliner cabins are a suitable milieu for the spread of disease, the environmental control systems used by commercial aircraft appear to restrict the movement of airborne microbes. The review found that proper ventilation within cabins with one air exchange reduces concentrations of airborne organisms by 63%. Data models from a study of tuberculosis transmission aboard aircraft showed that doubling cabin ventilation rates reduced the infection risk by half. The paper’s authors point out that the fear of catching an infectious disease from a fellow passenger is greater than the actual risk.
Betting on Biomass
The U.S. Department of Energy has unveiled a $2.85 million Biomass Surface Characterization Laboratory within the National Renewable Energy Laboratory in Golden, Colorado. Dedicated in March 2005, the new lab is designed to give scientists the means to make significant breakthroughs in the development of biomass as a viable energy source. The facility features the most advanced research tools to study biomass-to-energy processes at the atomic and molecular levels. One area the laboratory will explore is the creation of new technologies for biorefineries, which will produce bio-based transportation fuels and various other products the way petroleum refineries do today.
Turning Up the Heat Watch
According to National Weather Service (NWS) data, excessive heat is the leading weather-related cause of death, with at least 1,500 excess deaths from heat-related causes during the average U.S. summer. The NWS has been testing its Heat/Health Watch Warning System to provide the public with five days’ advance notice of excessive heat events. Now the NWS has announced that the system, which has become a model for others worldwide, will be expanded to include every U.S. city with a population exceeding 500,000. In Philadelphia, the first city to implement the system, 117 lives were saved over three years.
Japan Revs Up Idling Law
Since 1997 Japan has required that drivers of commercial vehicles turn off their engines when they were going to be stopped for more than just a few moments—for example, at curbs and stoplights. Now, thanks to studies proving the measures’ effectiveness in reducing carbon dioxide emissions and saving fuel, Japanese leaders plan to spread the movement to include private car owners. The Japanese Environment Agency calculates that if all 68.6 million cars owned in Japan idled for one less minute per day, more than 225,000 fewer tons of carbon dioxide would be emitted and 350 million liters of fuel would be saved.
Africa Forms Waste Institute
Despite international conventions to control the importation and transboundary movement of hazardous waste, African nations still struggle with huge problems of pesticide dumps and illicit trade in hazardous waste. Now 10 nations have signed an agreement under the auspices of the Basel Convention to establish an African institute to deal with waste issues. The institute will be hosted by South Africa, and will be legally established once five states ratify the agreement to create it. The institute will develop training programs for the environmentally sound management of hazardous and other wastes, as well as facilitate the transfer of technologies in this area. Work is under way to set up similar institutes for other areas of Africa.
The Lawn and Short of Mower Pollution
Not all lawn mowers are created equal, nor is all lawn mower pollution. From traditional gas-powered mowers to electric models, the amount of pollution produced varies significantly. A life cycle analysis done by University of Florida engineers confirms that gas-powered mowers produce more smog-forming pollution than their electric-powered counterparts. However, significantly more carcinogens and other toxicants may come from the manufacture and disposal of the batteries used in cordless electric mowers. Corded electric mowers, whose lifetime pollution consists of power plant emissions, were deemed least polluting of those tested.
|
0
|
PMC1257677
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A449b-A451
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0450aEnvironewsForumAllergies: Ionizing Air Cleaners Zapped Hood Ernie 7 2005 113 7 A450 A450 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Ionizing air cleaners—those staples of infomercials and splashy magazine ads—are not only ineffective at removing contaminants from indoor air, but also may emit enough ozone to be a health concern. The effects may be even greater in people with respiratory problems, who make up 80% of the buyers of such devices. Those are the conclusions reached in tests of the units described in the May 2005 Consumer Reports (CR).
CR tested five units (including the top-selling Ionic Breeze from The Sharper Image) and confirmed results reported in October 2003 rating most of the air cleaners “poor” at removing dust and tobacco smoke from the indoor environment. This time around, pollen was added as well, with similarly disappointing results. The cleaners were also tested for generation of ozone, a respiratory irritant. The results showed that some of the least effective models also emitted potentially harmful ozone levels.
“We felt that it was particularly important to notify our subscribers that these air cleaners not only don’t remove particulates from the air, but they also put ozone into it,” says Jeff Asher, vice president and technical director of Consumers Union, the publisher of CR.
There is no regulatory standard for ozone emission by air cleaners; manufacturers claim to adhere to a voluntary standard of 50 parts per billion (ppb), a limit established by the Food and Drug Administration for medical devices. CR used Underwriters Laboratories Standard 867 to measure the units’ ozone levels from 2 inches away in a sealed polyethylene room. All five machines failed that test.
To more accurately reflect actual use conditions, CR also tested the devices in an open laboratory, from distances of 2 inches and 3 feet. Two units failed this this test; the other three (including the Ionic Breeze) produced levels of 26–48 ppb at 2 inches and 2–18 ppb at 3 feet—still high enough by CR’s estimation to be of concern. “The levels were not what I would call of great imminent risk,” says Asher, “but it was of significant risk in the sense of being in an indoor environment, where we just don’t need more ozone.”
The Sharper Image, which unsuccessfully sued CR over its 2003 report, has fired back, assailing the magazine’s credibility. In a 6 April 2005 press statement, CEO Richard Thalheimer called the article “irresponsible in the way it casually and unscientifically speculates about public health and safety. . . . We continue to emphatically disagree with Consumer Union’s methods in evaluating the Ionic Breeze.”
But health and engineering experts find the CR results troubling. “These levels make these devices inappropriate to use for asthmatic patients and for patients with respiratory disease,” says Peyton Eggleston, interim director of the Johns Hopkins Children’s Center.
Richard Shaughnessy, an environmental engineer at the University of Tulsa who has researched air cleaners for many years, concurs, pointing out that “not only are people with respiratory illnesses and asthma the population targeted by most of these air cleaners, they’re also the ones who are most likely to be adversely affected in terms of exposure to small amounts of ozone.”
Buyer beware. Ionizing air cleaners don’t always live up to manufacturer claims and may emit harmful levels of ozone.
|
0
|
PMC1257678
|
CC0
|
2021-01-04 23:41:04
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A450a
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0450bEnvironewsForumPolicy: Framing a Chemical Future Dahlz Richard 7 2005 113 7 A450 A450 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Spurred by recent developments abroad to design new approaches to chemical management, the University of Massachusetts Low-ell Center for Sustainable Production sponsored a two-day conference in April 2005 to stimulate similar productive changes in the United States. The event attracted a mix of some 170 environmentalists, government officials, academics, product representatives, and chemical industry representatives.
The conference aimed to initiate the kind of multi-stakeholder dialogue that the European Commission created in the late 1990s. The European effort has resulted in a draft regulation called REACH (Registration, Evaluation and Authorisation of Chemicals), which calls for manufacturers and importers to identify and report the properties of the substances they use and sell. Other international actions, such as the 2002 World Summit on Sustainable Development, which set a goal of achieving sound global management of chemicals by 2020, have also heightened the need for better U.S. chemical policies. Discussions begun at this workshop may eventually lead to public policy supporting a safer and more competitive U.S. chemical industry.
Speakers opened the meeting with talks on the current thinking of chemical policy as well as specific policies and industrial protocols. Then participants broke out into workshops on various subjects. One group focused on the promotion of innovative industry initiatives, “green chemistry,” and alternative materials. Others dealt with improving information flows in supply chains and beyond, integration of U.S. and global chemical initiatives, and incorporation of improved chemicals management into business processes.
One theme repeated throughout the conference was the need for improved communication about what’s in the products that people buy, and greater transparency of the people who make the products. “The thing that I really noticed about [participant feedback] was the importance that people placed on information—the flow of information, the access to information, who’s responsible, where it’s stored,” said center director Kenneth Geiser.
A second repeating theme was support for green chemistry, chemical processes that reduce or eliminate the use and generation of hazardous substances in economically viable ways. Paul Anastas, director of the Green Chemistry Institute in Washington, DC, said he noticed a recognition that such scientific innovations “can be both economically profitable and environmentally preferable.”
Geiser said there also was nearly universal agreement by participants that the current U.S. chemicals policy, largely embodied in the Toxic Substances Control Act of 1976, is outdated. “Almost everyone I talked to felt that the current chemicals policy system needs overhauling,” he said. “It’s interesting that the business folks felt the same way; it’s not working for them, either.”
The Lowell Center will compile a report on the conference in the next few months. Geiser believes the conference’s goal has been met. “I think we created an enthusiasm for moving forward,” he said. “That was pretty much what we wanted.”
|
0
|
PMC1257679
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A450b
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0472aEnvironewsScience SelectionsMore Concerns for Farmers: Neurologic Effects of Chronic Pesticide Exposure Barrett Julia R. 7 2005 113 7 A472 A472 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Although there is considerable evidence that pesticides are neurotoxic, most research has focused on the short- and long-term consequences of acute high-level exposure such as that seen during industrial accidents or food contamination. To date, little has been known about the effects of chronic moderate exposure such as that experienced by farmers and other workers who regularly use agricultural pesticides. Now, a recent analysis of data collected in the Agricultural Health Study (AHS) links chronic moderate pesticide exposure to neurologic symptoms affecting both the central and peripheral nervous systems [EHP 113:877–882]. According to the research team, increases in such symptoms may be an early indicator of impaired neurological function.
The AHS, an ongoing study sponsored by the NIEHS, the National Cancer Institute, and the U.S. Environmental Protection Agency, furnished a rich data source for the researchers to investigate possible links. Between 1993 and 1997, approximately 20,000 private pesticide applicators (primarily farmers) in Iowa and North Carolina enrolled in the AHS and completed two questionnaires on demographic characteristics, lifestyle, medical history (including neurologic symptoms), and pesticide use. The current analysis focused on 18,782 of these individuals, white men aged 18–75 years who provided complete symptom information.
The 23 symptoms in the analysis included headache, dizziness, depression, limb weakness, poor balance, difficulty concentrating, and vision difficulties. In addition to the symptom information, participants detailed how long and how frequently they used any of 50 pesticides, including insecticides, herbicides, fungicides, and fumigants. They also indicated whether they had ever experienced pesticide poisoning or high-exposure incidents such as accidental skin contact with a large amount of pesticide.
To define cumulative exposure, the researchers calculated lifetime days of use from the number of years and the number of days per year that the applicators had used each pesticide. The team considered two measures of symptoms: the absolute number and the presence of 10 or more. To control for confounding by pesticide poisoning or high-exposure incidents, the researchers conducted analyses with and without those data from affected individuals. They also considered potential effects from pesticide use within the past year.
For pesticides overall, applicators with the most (more than 500) cumulative lifetime days of pesticide use reported more symptoms than those with the fewest lifetime days of use. The relationship between cumulative exposure and symptoms was strongest with insecticides; applicants with the most lifetime days of use were 2.5 times more likely to have 10 or more symptoms as applicators who had never used insecticides. Within the insecticide category, relationships with symptoms were strongest for organophosphates and organochlorines. Neither recent use nor a history of poisoning or high-exposure incident affected the results.
The results of this study extend previous research demonstrating a link between chronic moderate pesticide exposure and a range of cognitive, sensory, and motor symptoms. The AHS is unusually robust due to its large size and its wealth of detailed exposure information. The results of this analysis provide substantial evidence that neurologic symptoms may be increased by even moderate insecticide exposure, and that cumulative exposure may be as important as recent exposure, although more work is needed to understand the pathology underlying the reported symptoms.
Farm field fallout. A recent analysis shows that even moderate chronic pesticide use can result in neurologic symptoms among farmers and other applicators.
|
0
|
PMC1257680
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A472a
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0472bEnvironewsScience SelectionsBlocking Brain Development: How PCBs Disrupt Thyroid Hormone Brown Valerie J. 7 2005 113 7 A472 A473 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Polychlorinated biphenyls (PCBs) have long been known to alter growth and development in animals and humans, and are suspected of interfering with the action of thyroid hormone (TH) in humans. Much less is known about which congeners of this large chemical family may have such action and how they interfere with TH. Now, in an in vitro study using human brain stem cells, a team of German and Californian researchers shows how a specific PCB congener, PCB-118, can interrupt normal TH function and cause the premature differentiation of one type of brain cell [EHP 113:871–876].
Many animal studies, primarily in rats, suggest that PCBs can profoundly affect fetal brain development, which itself is highly dependent on the availability and amount of TH, principally from the mother. PCBs are known to lower circulating blood levels of TH by increasing TH metabolism and binding to TH transport proteins; the exact mechanisms and effects of TH disruption by PCBs are unclear, however. Some epidemiologic evidence suggests a link between PCB exposure during fetal development and subsequent cognitive problems in children, such as lowered overall IQ, attention and motor deficits, and impaired impulse control. The suspicion that these problems may be related to PCBs’ effects on the timing and type of brain cell differentiation led to the current study.
Stem cells known as normal human neural progenitor (NHNP) cells develop into three types of brain cells: neurons, which receive and transmit electrical signals via axons and synapses; astrocytes, which manage neurons’ surrounding environment; and oligodendrocytes, which produce myelin, the fatty sheath that insulates axons. TH is known to control the timing of oligodendrocyte differentiation.
The research team exposed NHNP cells to two PCB congeners—PCB-118 and PCB-126—and observed the effects on cell differentiation. They found that cells exposed to PCB-118 prematurely turned into oligodendrocytes. This finding suggests strongly that PCBs may facilitate the binding of coactivator proteins to cellular TH receptors; these proteins then mimic the action of TH. Paradoxically, a surplus of oligodendrocytes early in brain development may lead to a dearth later, because if there are proportionally fewer neurons, many oligodendrocytes cannot wrap an axon or reproduce. In that case, they undergo apoptosis, or programmed cell death. The end result would likely be a drop in the total number of brain cells.
The current study is consistent with an earlier rat study published in the April 2004 issue of EHP, which found that PCBs interfered directly with fetal TH receptor signaling, as opposed to reducing circulating maternal TH levels. Like many other studies, that study used Aroclor 1254, a commercial mixture of several PCB congeners, and did not determine the effects of each congener. The current study compared two PCB congeners that differ somewhat in their toxic equivalence (TEQ). The results showed that the lower-TEQ congener (PCB-118) acted via the TH pathway while the higher-TEQ congener (PCB-126) did not. The apparent toxicity of a PCB congener with a low TEQ also suggests that the TEQ system may not be a reliable measure of all types of toxicity.
|
0
|
PMC1257681
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A472b-A473
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0488a16002365AnnouncementsBook ReviewNanotechnology: Environmental Implications and Solutions Oberdörster Eva Eva Oberdörster earned her Ph.D. at Duke University and completed her postdoctoral work at Tulane University. She is currently on the faculty at Southern Methodist University in the Department of Biology, where her most recent research focuses on ecotoxicology of engineered nanoparticles.7 2005 113 7 A488 A488 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
By Louis Theodore and Robert G. Kunz
Hoboken, NJ:John Wiley & Sons, 2005. 378 pp. ISBN: 0-471-69976-4, $99.95 cloth
This book, written mainly for engineering students, gives an excellent summary of traditional environmental pollution issues. Ten chapters cover current legislation regarding environmental pollutants; an overview of chemistry and current nanotechnology processes; air, water and solid-waste issues; multimedia analysis; both health and hazard risk assessment; ethical considerations; and concluding remarks on future trends.
Nanotechnology is emerging in a wide variety of applications, yet unfortunately very little is known about the environmental implications of engineered nanomaterials, of possible ways to handle engineered nano-materials as environmental pollutants, or of how nanomaterials behave in the environment when used for remediation. Therefore, such a book will increase awareness of potential problems, but may disappoint those who expect dramatic revelations about nanoparticles as pollutants.
Due to the current paucity of data, the book gives a well-written overview of environmental toxicology of traditional toxicants, with only a bare mention of nanomaterials in introductions to each chapter. The exception is Chapter 2, which covers basic chemistry and various processes used in making nanosized materials. A few analytical tools currently used to study nanomaterials are discussed, followed by a good overview of the current and near-future uses of nanotechnology. Other chapters present traditional monitoring methods and pollution control methods in great detail, and are useful for teaching. However, the authors do not delve into the current literature showing that nanosized materials do not necessarily behave like bulk materials in water and soils and that it would be prudent to consider and develop new treatment or containments strategies.
The long section on current pollution-related regulations implies that laws already exist that apply to engineered nanomaterials. However, the book does not mention several recent rulings stating that the nano-sized materials are not considered different from bulk (until or unless they receive their own Chemical Abstracts Service numbers), and therefore most of the legislation does not apply to these materials. Although Occupational Safety and Health Administration legislation is mentioned, the impressive ongoing effort by scientists from the National Institute for Occupational Safety and Health to develop monitoring tools and toxicity data for nanosized materials is absent from this section.
Because much of the information currently known about nano-sized particles comes from air pollution and inhalation toxicology, the section on air pollution is more robust than other chapters in terms of nano-implications. But the section contains some inaccuracies. The authors assume that nanosized materials behave as gases—which is true only for very small particles (~ < 5 nm). When particle concentrations increase, they can quickly aggregate and behave like larger-sized particles. There is bare mention of the importance of particle chemistry, surface chemistry, concentration, shape, and size in nanoparticles or engineered nanomaterial behavior as air pollutants. Other inaccuracies include the statement that there is no information on deposition of nanosized particles in the respiratory tract, and repeated misuse of terms such as “particulate” versus “particle.” However, these inaccuracies do not distract from the otherwise thorough discussions of pollution issues that introduce the interested engineer to ecotoxicology. The water and solid-waste sections of the book focus on traditional toxicants as case studies and give a detailed description of current waste-water and solid-waste treatment. With few toxicity or exposure data available, the risk assessment and hazard assessment sections focus on traditional toxicants as case studies—a good foundation for future engineers dealing with the emerging risks and hazards of nanotechnology. A chapter on ethics presents several fictional scenarios, useful in classroom discussions on environmental ethics and the role of environmental engineers in monitoring and responsible whistle blowing.
Overall, the book’s good overview of traditional (bulk) toxicants serves as background for potential nanosized material problems. Given the scarcity of data about nanoecotoxicology, very little information now exists on environmental implications of nanotechnology. Such information should become available over the next 5 years, contributing to a second edition of this book.
|
0
|
PMC1257682
|
CC0
|
2021-01-04 23:41:04
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A488a
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institue of Environmental Health Sciences ehp0113-a0488b16002365AnnouncementsNew BooksNew Books 7 2005 113 7 A488 A488 Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright.
==== Body
Apoptotic Pathways as Targets for Novel Therapies in Cancer and Other Diseases
Marek Los, Spencer B. Gibson, eds.
New York:Springer-Verlag, 2005. 396 pp. ISBN: 0-387-23384-9, $149
BioIndustry Ethics
David Finegold, Cecile Bensimon, Abdallah Daar, Margaret Eaton, Beatrice Godard, Bartha Knoppers, Jocelyn Mackie, Peter Singer
Burlington, MA:Elsevier, 2005. 384 pp. ISBN: 0-12-369370-5, $34.95
Chemistry and Safety of Acrylamide in Food
Mendel Friedman, Don Mottram, eds.
New York:Springer-Verlag, 2005. 475 pp. ISBN: 0-387-23920-0, $169
Chernobyl: Catastrophe, Consequences and Solutions
Jim Smith, Nicholas A. Beresford
New York:Springer-Verlag, 2005. 200 pp. ISBN: 3-540-23866-2, $119
Encyclopedic Reference of Immunotoxicology
Hans-Werner Vohr, ed.
New York:Springer-Verlag, 2005. 730 pp. ISBN: 3-540-44172-7, $299
Environmental Health: From Global to Local
Howard Frumkin, ed.
San Francisco, CA:Jossey-Bass, 2005. 1,250 pp. ISBN: 0-7879-7383-1, $65
Facilitating Interdisciplinary Research
Committee on Facilitating Interdisciplinary Research, National Academy of Sciences, National Academy of Engineering, Institute of Medicine
Washington, DC:National Academies Press, 2005. 332 pp. ISBN: 0-309-09435-6, $42
Handbook of Isoelectric Focusing and Proteomics
David Garfin, Satinder Ahuja
Burlington, MA:Elsevier, 2005. 380 pp. ISBN: 0-12-088752-5, $215
Integrative Physiology in the Age of Genomics and Proteomics
Wolfgang Walz
Totowa, NJ:Humana Press, 2005. 280 pp. ISBN: 1-58829-315-7, $99.50
Intelligent Bioinformatics: The Application of Artificial Intelligence Techniques to Bioinformatics Problems
Edward Keedwell, Ajit Narayanan
Hoboken, NJ:John Wiley & Sons, 2005. 294 pp. ISBN: 0-470-02175-6, $110
Metabolome Analyses: Strategies for Systems Biology
Seetharaman Vaidyanathan, George G. Harrigan, Royston Goodacre, eds.
New York:Springer-Verlag, 2005. 405 pp. ISBN: 0-387-25239-8, $129
Methods in Community-Based Participatory Research for Health
Barbara A. Israel, Eugenia Eng, Amy J. Schultz, Edith A. Parker, eds.
Hoboken, NJ:John Wiley & Sons, 2005. 528 pp. ISBN: 0-7879-7562-1, $60
Molecular Biology: Understanding the Genetic Revolution
David Clark
Burlington, MA:Elsevier, 2005. 816 pp. ISBN: 0-12-175551-7, $94.95
Molecular Diagnostics
George Patrinos, Wilhelm Ansorge, eds.
Burlington, MA:Elsevier, 2005. 488 pp. ISBN: 0-12-546661-7, $99.95
Proteomics in Cancer Research
Daniel C. Liebler, Emanuel F. Petricoin, Lance A. Liotta, eds.
Hoboken, NJ:John Wiley & Sons, 2005. 202 pp. ISBN: 0-471-44476-6, $150
Statistical Methods in Molecular Evolution
Rasmus Nielsen, ed.
New York:Springer-Verlag, 2005. 520 pp. ISBN: 0-387-22333-9, $89.95
The p53 Tumor Suppressor Pathway and Cancer
Gerard P. Zambetti, ed.
New York:Springer-Verlag, 2005. 245 pp. ISBN: 0-387-24135-3, $139
|
0
|
PMC1257683
|
CC0
|
2021-01-04 23:41:03
|
no
|
Environ Health Perspect. 2005 Jul; 113(7):A488b
|
utf-8
|
Environ Health Perspect
| 2,005 |
nan
|
oa_comm
|
==== Front
Ann Gen PsychiatryAnnals of General Psychiatry1744-859XBioMed Central London 1744-859X-4-151614455110.1186/1744-859X-4-15ReviewPhobic memory and somatic vulnerabilities in anorexia nervosa: a necessary unity? Myslobodsky Michael [email protected] Howard University and Cerebral Brain Disorder Branch, NIMH, NIH, Bethesda, MD 20892-1379, USA2005 6 9 2005 4 15 15 26 3 2005 6 9 2005 Copyright © 2005 Myslobodsky; licensee BioMed Central Ltd.2005Myslobodsky; 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.
Anorexia nervosa is a clinically significant illness that may be associated with permanent medical complications involving almost every organ system. The paper raises a question whether some of them are associated with premorbid vulnerability such as subcellular ion channel abnormalities ('channelopathy') that determines the clinical expression of the bodily response to self-imposed malnutrition. Aberrant channels emerge as a tempting, if rather speculative alternative to the notion of cognitively-driven neurotransmitter modulation deficit in anorexia nervosa. The concept of channelopathies is in keeping with some characteristics of anorexia nervosa, such as a genetically-based predisposition to hypophagia, early onset, cardiac abnormalities, an appetite-enhancing efficacy of some antiepileptic drugs, and others. The purpose of this article is to stimulate further basic research of ion channel biophysics in relation to restrictive anorexia.
==== Body
Introduction
Anorexia nervosa (AN) is an intractable illness with a high long-term mortality that affects 1% to 3.7% of the young women [1]. The death rate of patients with AN is up to 30 times greater than that of age-matched normal women. About 20% of AN patients remain chronically disabled [2]. Despite its grave complications, the lack of exact pathophysiology and organic definition, denies AN an independent nosological grade or a status of a neuropsychiatric problem. The various theories that have been proposed to explain the cause or origin of AN can be divided into two major schools of thought, socio-cultural and biological. Until very recently, AN was categorized as a disease of psychogenic origin (e.g., a consequence of unresolved conflicts of the individual psychosocial development) [3]. Many subscribed to the cultural paradigm inasmuch as it was rendered secure from experimental scrutiny. Not surprisingly, pharmacotherapeutic options in AN continue to be limited. For years, the disorder was so refractory that even 'heroic' efforts such as lobotomy, once reserved for treating schizophrenia were attempted [4]. Kennedy and Goldbloom [5] maintained in their review of 1991 that there was little, if any role for pharmacotherapy in AN. Over a decade later and more than 200 studies on the topic, the majority of patients stay refractory to the currently available drugs [6,7]. Therefore, alternative approaches toward AN pathophysiology has to be explored. A small proportion of individuals developing AN and the commonality and influence of socio-cultural pressures only emphasize the need for identifying the vulnerable population.
In keeping with this goal, the present article looks at the possibility that AN is associated with intrinsic vulnerability of potassium channels. It is to these channels and to their kinetics that cerebral cells owe their membrane potentials and the many characteristics that control local and distant changes in multiple organ systems. The following provides a selective perspective dealing mainly with the apamin-sensitive small-conductance calcium (Ca2+)-activated K+ channels (SK channels) [8] as they might be related to the cognitive and somatic manifestations of AN.
"Laparophobia" and K+ channels
Young women with AN are recognized for commonly manifesting greater levels of general inhibition, loneliness, and social phobia of being corpulent. Also, fears associated with inadequacies of sexual life are found for 80% of the AN patients even if their initial stages of heterosexual development seemed normal [9]. Problems with sexuality may trigger the onset of AN [10]. In tardive AN (developing after the age 30), the feared sexuality emerged as being of apparent etiological significance, much as it is with earlier onset AN [11]. Therefore, elsewhere, AN was represented as anticipatory anxiety of stoutness and frank fat-phobia ('laparophobia') [12,13] mitigated by the efforts to obtain alternative (non-appetitive) reward, such as exemplified by the paradoxical euphoric state and hyperactivity,[14] symptoms of denial, lack of concern, and alleged satisfaction with their state [15-17].
In theory, the acquisition of fear could be reduced to plasticity changes associated with Ca2+ influx through N-methyl-D-aspartate (NMDA)-receptor channels in response to environmentally or cognitively elicited alarm. Ca2+ ions control a host of neuronal functions, such as transmitter release, excitability and synaptic plasticity. In laboratory environment, Ca2+ effects are reproduced by repetitive input stimulation that elicits a long-lasting increment of synaptic strength known as long-term potentiation (LTP), a widely known model of learning and memory processes (e.g. in the hippocampus or amygdala), with which a great number of neurophysiologic memory studies are performed. In many neuronal cells, intracellular Ca2+ fluxes are increased during and after an action potential that activate K+ channels thereby producing long-lasting changes of conductance and thus lasting membrane hyperpolarization. Therefore, it is conceivable that activity-dependent changes of excitability could be achieved via nonsynaptic mechanisms. Mechanistically, K+ channels are defined as the pore-forming primary transmembrane proteins that initiate cellular polarization by allowing K+ flux down a concentration gradient. Calcium-sensitive K+ conductances are known to play a major role in the modulation of NMDA-induced bursting and the spike afterhyperpolarization, so that dysfunctional K+ channels may contribute in the scenario of a wide range of cognitive aberrations and AN ideation. Ultimately, the process of learning and the strength of associations will be determined by the intrinsic morphology and kinetics as well as the density and distribution profile of ion channels that are embedded in the same membrane of the dendrites and the cell body, which accommodate NMDA receptors [18]. They play distinct physiological tasks from fine tuning membrane excitability in response to sensory input, modulating locomotion and emotional behavior to the induction of synaptic plasticity entailed in memory and cognition, temporally integrated with circadian cues, as well as in the antinociceptive effect [19].
Three subtypes of slow Ca2+-activated K+ (SK) channels (SK1, SK2, SK3) set off by submicromolar intracellular Ca2+ concentrations have been cloned, which differed in their pharmacology and kinetics [8,20-22]. One of the interesting features of SK channels is that normally they reduce neuronal excitability, whereas being blocked by the peptidergic honeybee toxin, apamin, they robustly augment neuronal excitability. In hippocampal CA1 neurons, SK channels contribute to the afterhyperpolarization, affecting neuronal excitability, regulating synaptic plasticity and memory [23]. Using field potential recording in the CA1-region of rat hippocampal slices, Behnisch et al. [24] showed that LTP induced by a single 100 Hz tetanization was intensified by extracellular application of apamin in a concentration range of 1–200 nM. These changes in the sensitivity to apamin were hypothesized to serve a marker of memory state and potentially, memory disorders. In fact, intracerebroventricular injection of apamin appears to improve retention of an odor pair association in rats [25]. Likewise, apamin increased neuronal excitability and facilitated the induction of synaptic plasticity at Schaffer collateral synapses and hippocampal-dependent learning [26]. Mice treated with apamin demonstrated accelerated hippocampal-dependent spatial and nonspatial memory encoding. They required fewer trials to learn the location of a hidden platform in the Morris water maze and less time to encode object memory in an object-recognition task compared with saline-treated mice [27]. Blank et al. [28] found recently that SK3 channel transcript and protein were more abundant in hippocampi from aged mice (22–24 months) compared to hippocampi from young mice (4–6 months). They showed that age-related decrement in trace fear conditioning (a hippocampus-dependent learning task) is correlated with elevated expression of SK channels of the SK3 type in the hippocampus as well as with reduced LTP. The effect was reversed when treated with SK3 antisense oligonucleotides. The authors further suggested that increased hippocampal expression of SK3 channels in aged mice may represent a mechanism that contributes to age-dependent decline in learning and memory and synaptic plasticity. In the hippocampus, SK3 was found predominantly in the terminal field of the mossy fibers and in fine varicose fibers, thereby suggesting their presynaptic localization. Using high-resolution immunofluorescence analysis, one study [18] found that the SK3 clusters were precisely colocalized with the presynaptic marker synapsin and at close range (0.4–0.5 mum) from NMDA-receptors and PSD-95, but rarely associated with GABAA-receptor clusters. This arrangement is consistent with a view that SK3 is a presynaptic channel in excitatory hippocampal synapses, with no preference for glutamatergic or GABAergic postsynaptic neurons, and is probably involved in regulating neurotransmitter release.
The responsibility of SK3 channels for the medium, and possibly the slow components of an afterhyperpolarization current make them a candidate for dysregulation of cellular homeostasis in a variety of systems. They are abundantly present in brain regions implicated in AN, such as the hypothalamus, [29] the limbic system [30] and midbrain regions [31,32] The cardinal role of SK channels in regulating burst firing and rhythmic oscillatory activity appears to be needed for coding for reward-related events by dopaminergic midbrain neurons. The afterhyperpolarization in dopaminergic ventral tegmental area neurons was shown to determine their sensitivity to ethanol reward [33] and by extension, they could be as responsive to some corporeal and extracorporeal rewards. Perhaps, this mechanism may be relevant for locomotion reward in hyperactive AN patients [14]. The Ca2+-dependent activation of K+ channels may be among other regulators of pacemaker activity of interneurons that govern the quasi-periodic repetition of group activities and EEG synchronization. The excitability of fast-spiking prefrontal interneurons may be regulated by dopamine via a voltage-independent 'leak' K+ current and an inwardly rectifying K+ current, thereby modulating pyramidal cell excitability [34]. It remains to be explored which neurons express their distinct subsets of SK channel subunits in specific areas and how they are related to changes of cellular functions translated to the diverse clinical features of AN along its course and co-morbidities.
Somatic and cognitive aberrations: Two in one?
The literature is definitive about an increased risk of diseases in AN. However, whereas psychological and cognitive deficits are conceived of as a component of the AN syndrome, somatic abnormalities are rather attributed to self-imposed malnutrition than to general vulnerability associated with AN [2]. Only infrequently are, cognitive and somatic alterations discussed together albeit in the context of a more specific shortage, such as caused by insufficiency in poly-unsaturated fatty acids that could cause also cognitive abnormalities [35]. Studies that obtain retrospective histories of disorders that occur prior to the age of onset of AN are uncommon, and are limited to psychopathological findings [36]. Assuming that there is a role for SK channels in AN, the question is what somatic manifestations would parallel neuropsychiatric abnormalities?
Medical comorbidity of AN
Of the three SK channels, SK1 and SK2 are predominantly expressed in the nervous system, in cortical pyramidal cells; [37] the basal ganglia and limbic system; [26,38,39] dopaminergic (DA) midbrain neurons; [40] and supraoptic neurosecretory cells [41]. The SK3 protein is seen more diffusely scattered. It is expressed primarily in phylogenetically older brain regions [42], and is also distributed in some peripheral neurons as well as diverse bodily tissues [8]. That wide distribution of SK channels and the fundamental role of K+ currents in controlling membrane excitability of diverse organ systems pose a question of their role in regulation and dysregulation of the functional state of bodily tissues, at least of ectodermal origin, concurrently with that of the central nervous system.
It is telling that AN is associated with an amazing rate of cardiac abnormalities (in up to 86% of the patients) such as electrocardiographic abnormalities, reduced left ventricular mass, a small heart on the chest X-ray, impaired myocardial performance, and others.[43,44] Mitral valve prolapse (MVP) is another common somatic signs of AN.[45,46] Abnormality of the mitral apparatus may be primary or more benign secondary that emerge as a consequence of reduced or abnormal ventricular dimensions due to weight lost [47]. It is not always certain what form of MVP are registered in AN. The possibility of primary MVP abnormality cannot be ruled out since it may be associated with some subtle generalized disorder coupled with peculiar skeletal abnormalities (e.g. pectum excavatum, scoliosis, loss of normal kyphosis of the thoracic spine), somewhat elongated arms and decreased breast mass [48] The breast develops from the anlage of ectodermal cells along the primitive mammary ridges 'milk lines' during the sixth week of gestation. Certain abnormalities of the growing breast such as breast asymmetry (difference of its form, position or volume), hypoplasia of one breast are common finding in normal adolescents [49].
Cardiac arrhythmias and the lengthening of the QT interval are frequently associated with AN.[2] Recurrent syncope and sudden death typically occur in AN during exercise or emotional upset, [50,51] so that it is more likely to be attributed to metabolic aberrations associated with malnutrition, dehydration, or hypoglycemia, or socially-triggered emotional distress. The neurogenic mechanisms are clearly implicated in many cases of cardiac arrhythmia and sudden death in AN. Critchley et al. [52] demonstrated the role of mental and physical stress challenges in a group of 10 out-patients attending a cardiological clinic. Using H2(15)O PET, they obtained a robust positive relationship between right-lateralized asymmetry in midbrain activity and proarrhythmic abnormalities of cardiac repolarization during stress. However, mental and physical stress merely exposed the presence of enhanced cardiac arrhythmic vulnerability such as deficient myocardial repolarization [53]. A greater risk for the development and recurrence of coronary heart disease in women with psychiatric disorders such as depression, panic disorder, and generalized anxiety disorder although commonly attributed to psychosocial factors [54] might also be contributed by proarrhythmic state unmasked by psychosocial stressors. Mutations in genes encoding sodium or potassium channels were shown to underlie the 'Long QT syndrome,' causing arrhythmias [50,55]. It is a frequent cause of syncope and unexpected death in children and young adults, mostly women. Although the syndrome is an autosomal-dominant genetic disorder of cardiac electrical repolarization, the QT interval at presentation is normal about 10% of the time and just borderline prolonged in another 30%, so that premorbid vulnerability may be difficult to establish, and it is hardly looked for as a sign anticipating AN.
A remarkable aspect of the cardiomyocytes is their property for 'memory' of the signal transduction mechanisms and cardiac repolarization. A case in point is a persistent or 'remembered' T-wave on ECG during periods of the previous abnormal QRS complex in the sequence of altered ventricular pacing and manifested during sinus rhythm. This repolarization contributed to by specific K+ channels [56,57] was provocatively labeled as "cardiac memory" [58]. We do not know whether or not "memory" of somatic cells marks a problem that remains latent in manner of 'functional teratogenesis' until triggered centrally (e.g. [52]). Nor do we know to what extent environmental stressors could unmask general SK channels abnormalities of alveolar epithelial cells in the lung, mesenteric and pulmonary arteries, vascular smooth muscles, genitourinary and gastrointestinal smooth muscle cells. Alkon [59] not only set out the question, he also came up with a theory that aberrant channels must represent a systemic disorder in Alzheimer's disease that involves not just the brain but other tissues such as skin, blood, and olfactory mucosa, as well. A change in channel activity leading to a cataract brings together such distant disorders as schizophrenia and myotonic dystrophy [60]. Both carry an increased risk of cataract, regardless of whether it was due to abnormal gene expression or followed drug intervention.
The skin, including its specialized forms such as the retina derives their origin from the same progenitors around the third ventricle. Cutaneous and mucocutaneous changes are well documented to be among the early diagnostic pointers to AN. They include xerosis (71%), cheilitis (76%), bodily hypertrichosis (62%), periungual erythema (48%), gingival changes (37%), and nail changes (29%) [61] along with the thinner body hair and abundant pilosebaceous glands.[62,63] Some skin signs are part of 'vasospastic syndromes' with a range of manifestations from cold intolerance, altered thermoregulation to physical stimuli or emotional stress to Raynaud's phenomenon, [64] the redness, itching, and burning of the skin, particularly fingers, toes, heels, nose, and ears exposed to cold known as perniosis [65,66]. Vasospasm could trigger acute severe exacerbations due to thrombosis inasmuch as the platelets obtained from patients with AN or severe peripheral vascular disease appear to be hyperaggregable [67].
The eye is frequently involved in the vasospastic syndrome, and ocular manifestations of microvascular dysfunction include alteration of conjunctival vessels, corneal edema, retinal arterial and venous occlusions, and others [66]. A high incidence of ocular involvement in the form of episcleral capillary aneurysms and subconjunctival hemorrhages, reduced mean tear production, conjunctival squamous metaplasia [68] is another example of vasospasm in AN. The pathophysiology of 'vasospastic syndromes' is obscure. The failure of the endothelium-derived hyperpolarizing factor may be one of the players. The latter is operationally defined as the hyperpolarization and associated relaxation remaining after the inhibition of the synthesis of NO synthase and prostaglandins (ref. [69] for review) that is possibly another word for K+ channels [70].
Significant osteoporosis affects over half of all women with AN [71,72]. Occasionally, fragile bones due to osteoporosis could lead even to fracture of ribs [73] and the sternum [74]. The mechanisms of bone loss in this condition are poorly understood. Although a low estrogen level is implicated, administration of estrogen alone has not been shown to prevent bone loss. Grinspoon et al.[72] hypothesized that administration of bone trophic hormone, insulin-like growth factor I (IGF-I), a nutritionally dependent hormone [75] that stimulates osteoblast function and collagen synthesis would increase bone turnover in young women with AN. They did obtain hypothesized increased markers of bone turnover in severely osteopenic women. However, IGF-1 may also be prenatally programmed. Infants whose mothers were exposed to peak sunshine during their first trimester were born significantly heavier than infants whose mothers experienced low levels of sunshine during the same period. Tustin et al. [76] attributed facilitated prenatal growth to high levels of IGF-1 due to sunshine exposure during early gestation. Epidemiological studies suggest an association between weight in infancy and skeletal size and the risk of osteoporosis in adulthood. A significant association was established between birth weight and adult bone mineral content at the lumbar spine and femoral neck [77]
In sum, prospective studies of children at risk for AN are missing to establish the presence of somatic anomalies long before the onset of eating disorder. However, the presence of the foregoing aberrations during intrauterine life, though subtle, may be fated to affect adult health trajectories. Collectively, these changes were designated over 40 years ago as "reproductive casualty." [78] Nowadays, the latter latent 'functional teratogenesis' is actively discussed as part of the "Barker hypothesis" that postulates that a number of dysfunctions undergo programming during embryonic and fetal life; that individuals with a completely normal phenotype at birth, may acquire diverse disorders in adolescence or adulthood [79-81].
Is calorie restriction all bad?
Many physicians encounter AN at the bedside, with the syndrome comprised of gonadal failure (a low estrogen state and amenorrhea), cardiovascular abnormalities, osteopenia, thinning of skin and skeletal muscle wasting that are typical of human aging. Therefore AN would be expected to be a harbinger of the premature frailty, increased susceptibility to aging-related disorders and decreased longevity. As it happens, AN is not totally harmful and in special circumstances may even be beneficial. The overall cancer incidence among women with AN identified in the population-based Danish Psychiatric Case Register during 1970–1993 was slightly reduced by a factor of 0.80 (95% confidence interval 0.52–1.18) below that of the general population.[82] In a sample of 7303 Swedish women hospitalized for AN prior to age 40 years (the Swedish Registries from 1965 to1998) there was a more robust decrease in breast cancer incidence compared with the general female population of comparable age. AN developing prior to the first birth followed by a subsequent pregnancy was associated with an even more pronounced reduction in risk [83].
How can this be? Although deficient transmembrane potassium traffic in some cells is troubling, it appears to be associated with an intriguing gain in having reduced physiological and pathological proliferation capacity and thus a diminished oncogenic potential [84]. A population-based retrospective cohort study of 208 Rochester residents who were monitored for up to 63 years since admission for AN found that long-term survival in AN patients did not differ from that expected for the community [85]. We have yet to learn whether or not AN mitigates malignancies in tissues other than breasts that are vulnerable to aging (e.g., colon, bladder) and reduces neuronal loss in neurodegenerative disorders. The role of caloric restriction in increasing longevity was repeatedly demonstrated in laboratory animals and lower organisms. Assuming that all essential nutrients are acquired, AN might be conceived of as the cheapest investment into defenses against infections and cancer, in general. However, AN-style starvation carries unacceptably high risk when a wider range of outcome variables is considered [86].
Side Effects of Anorectic Drugs: Pathophysiology Ex-Juvantibus
Using a preparation of isolated rat lungs, Belohlavkova and colleagues [87] compared the inhibitory effect of ritanserin, an antagonist of 5-HT2 receptors, on fenfluramine- and 5-HT-induced vasoconstriction. As expected, both 5-HT and fenfluramine caused significant increases in perfusion pressure. Ritanserin at a dose (10-7 mol/l) inhibited >80% of the response to 5-HT and reduced the response to fenfluramine by approximately 50%. A higher ritanserin dose (10-5 mol/l) completely abolished the responses to 5-HT but had no more inhibitory effect on the responses to fenfluramine. However, a pharmacological blockade of voltage-gated K+ channel activity (by 4-aminopyridine) markedly potentiated the pulmonary vasoconstrictor response to fenfluramine but was without effect on the reactivity to 5-HT. Clearly, the pulmonary vasoconstrictor response to fenfluramine was only partly mediated by 5-HT receptors, inasmuch as the vasoconstrictor potency of the drug was elevated when the K+-channel activity was reduced or altered in transgenic (SK3-T/T) mice [88]. Fenfluramine was widely employed for the treatment of obesity and abandoned after it was noticed to cause the development of valvular heart disease, hypertension, stroke and digital or mesenteric ischemia [89-91]. Consequently, the possibility was entertained that some individuals may have had intrinsically low activity of K+ channels ("channelopathy"), which, as Belohlavkova and colleagues [87] hypothesized, was not functionally obvious under usual conditions but may have become exposed by anorectic drugs. Although their findings suggest the presence of microvascular dysfunction of unknown origin the term channelopathy alludes to a group of disorders that include, other than congenital long QT syndrome mentioned above, also cyclic vomiting syndrome, neuromyotonia, episodic ataxia, abdominal migraine, and migraine headaches as well as many others have been mapped to chromosomal regions that are rich in ion channel genes [92-94]. This hunch, though highly speculative encourages henceforth to explore, whether or not AN shares some of its pathophysiology with channelopathies.
Is Anorexia a "Channelopathy"?
Aberrant channels emerge as a tempting, if rather speculative alternative to the notion of synaptic modulation deficit in AN. There are several requirements to suspect the presence of anomalous channels in a given disorder:
• Symptoms implicating specific ion channel genes
• The timing of disease onset or deterioration in childhood or adolescence
• Episodic character of manifestations
• Involvement of more than one organ or system
Chandy et al. [95] found that the second (3-prime) CAG repeat was highly polymorphic in control individuals, with alleles ranging in size from 12 to 28 repeats. They tested for an association between the longer alleles of SK3 and these neuropsychiatric disorders. There was a statistically significant overrepresentation of longer alleles in schizophrenia patients and a similar albeit nonsignificant trend in bipolar disorder patients, thereby suggesting that mild variations in the length of the polyglutamine repeats might produce subtle alterations in channel function, and in neuronal behavior. Other groups [96,97], however, did not confirm this finding. The most persuasive evidence indicating an association between inherited disorders of ion channels and AN is the discovery of the gene encoding for the SK channel.[98,99] They support the assumption of a common vulnerability for 'functional psychoses' that may include AN.
Although infrequently (5%), AN may be associated with epilepsy [100]. Epileptiform abnormalities in the EEG are infrequently recorded in AN patients,[101] although their rate is likely to be significantly underestimated since EEG is not routinely examined in eating disorders. In early comprehensive studies of EEG in behavioral disorders of childhood, eating disorders were not even mentioned [102]. However, the absence of clear indices of epileptiform abnormalities is not a critical violation of the channelopathy criteria. The notion of channelopathy may well be expanded into the territory of neurodegenerative disorders, such as Alzheimer's disease, [59,103] which is infrequently associated with epileptiform phenomena. On the other hand, neurologic AN complications comprise the majority of those consistent with the presence of channelopathies: neuromuscular abnormalities (45%); generalized muscle weakness (43%); peripheral neuropathies (13%); headaches (6%); syncope (4%); diplopia (4%), and movement disorders (2%). All these complications have episodic character. Also, local cellular epileptiform activity in the limbic system, such as plateau-bursting type action potentials or global Ca2+ signal (typical for some endocrine-cell-type), may be recorded with no overt manifestations of episodic behavioral disturbances. Thus, a relative deficit of SK channels or even their genetically engineered absence [104] may not yield overt phenotypic outcome, as SK3 overexpression would [28]. Nonetheless, it would subtly increase excitability, reduce the threshold for the induction of synaptic plasticity, and facilitates amygdala or hippocampus-dependent memory. On the background of abnormal temporolimbic machinery such changes of molecular plasticity may conceivably cause hormonal effects to be exaggerated or idiosyncratic, which would set a stage for phobias and obsessive-compulsive symptoms [105].
Cellular excitability changes may be also associated with a coerced movement of water to maintain osmolarity during cellular activity [106]. Water is transported by the aquaporins, a family of membrane proteins that function as water channels in many tissues including neurons, glial cells, astrocytic foot processes near or in direct contact with blood vessels and others. Therefore, a loss of K+ homeostasis in the presence of sustained neuronal activation may follow that of aberrant water fluxes. The mechanism underlying the functional coupling between water transport and K+ has yet to be elucidated. It was noticed, however, that a lasting compromise of cell volume constancy could contribute to a buildup of K+ in the extracellular space and ultimately, set a stage leading to a chronically enhanced excitability and even epileptogenicity [107,108].
One might further posit that estrogen represents additional factor modulating excitatory neurotransmission (apparently via NMDA/AMPA receptors) in the hippocampus [109] Using whole-cell recordings in hypothalamic slices from ovariectomized female guinea pigs, Kelly et al. [110] showed that estrogen (17β-estradiol, E2) robustly augments the efficacy of α1-adrenergic receptor agonists in inhibiting SK currents in preoptic GABAergic neurons. An association between susceptibility for eating disorders and the gene encoding for β-estrogen receptor [99] suggests that a specific group of individuals would have increased AN severity specifically related to changes in hormonal profile. In sum, the notion of AN as a channelopathy is in keeping with the following characteristics of the disorder: its early onset, genetic liability as well as episodic somatic disorders such as cardiac and autonomic abnormalities. As is shown below, the efficacy of some antiepileptic drugs is apparently also consistent with this hypothesis.
Epilogue
The physiological and pathophysiological effects of K+ channels on cerebral and extracerebral functions are numerous. Their role suggests a paradigm of "channelopathy" that articulated a way of simplifying and explaining otherwise seemingly unrelated somatic and neuropsychological findings in AN. If we are to understand the pathophysiology of the disorder, knowledge of the triggers of instabilities in ion fluxes in specially designed prospective studies may be mandatory.
The major limitation of researching the problem is in the populations selected because of multiple AN phenotypes. All patients entered in the studies cited above are either referred for their severe dieting or somatic manifestations consequent to it; many of them are interesting cases reported for their unusual presentation, such as nausea, vomiting, abdominal pain, electrolyte disturbances, sleep disorders, orthostasis and others [111,112].
The basis for choosing a conceptual model of AN, other than its simplicity, is the capacity of the model to provide a common denominator, for both psychopathological profiles and somatic manifestations of the disorder as well as to suggest therapeutic choices. SK3 channels may be potential therapeutic targets for regulating brain excitability as well as alleviating somatic disorders associated with AN. Several selective ligands are already being explored for their ability to block SK channel or facilitate SK channel opening [21,113]. Somewhat facetiously, Iversen[114] admonished that "on average it takes around 30 years for a new scientific discovery to find its way to a new generally available therapy" (p. 1539). Although new technologies may greatly facilitate the progress of identifying potential therapeutic targets, some caution need to be exercised [115]. The "thirty year rule" may still apply in the area of ion channel biophysics. With this in mind, the presence of abnormal neuronal excitability in AN, behooves the research clinicians to the fact that achieving membrane stabilization, reducing action-potential firing, and controlling Ca2+ fluxes in the neural and non-neural tissues can be accomplished by using drugs activating a GABAA receptor complex that results in opening of the Cl- channel and influx of Cl- ions thereby leading to hyperpolarization of the neuronal membranes. However, studies on GABAA-mediated inhibition in eating behaviors have yielded both orexigenic and anorexic effects depending on drug used. Another anticonvulsant to consider is ketamine [116]. In patients with a long history of eating disorder that were resistant to several forms of treatment, Mills and colleagues [117] attempted infusions of 20 mg per hour ketamine for 10 h along with 20 mg twice daily nalmefene as opioid antagonist. Nine of 15 patients examined in the study showed prolonged remission when treated with two to nine ketamine-infusion sessions at intervals of 5 days to 3 weeks. Admittedly, ketamine has limited therapeutic potential because of its adverse psychotomimetic side-effects. In this regard, memantine (1-amino-3, 5-dimethyl-adamantane), an uncompetitive NMDA receptor antagonist may be a better alternative by producing symptomatological improvement under conditions of tonic NMDA receptor activation. Preclinical studies showed that memantine can reduce the behavioral deficits produced by chronic stress [118] and enhance antidepressive effects of fluoxetine given in subtherapeutic doses [119].
Abbreviations
AMPA, amino-3-hydroxy-5-methyl-4-isoxazol propionate; AN, restrictive anorexia nervosa; DA, dopamine; GABA, γ-aminobutyric acid; 5-HT, 5-hydroxytryptamine; IGF-I, insulin-like growth factor I; K+, potassium; LTP, long-term potentiation; NMDA, N-methyl-D-aspartate; SK, slow potassium channel.
Contributors
Michael Myslobodsky is the sole contributor to this review.
Funding
No financial assistance was received for the writing of this paper.
Competing interests
The author(s) declare that they have no competing interests.
==== Refs
Deering S Eating disorders: recognition, evaluation, and implications for obstetrician/gynecologists Prim Care Update Ob Gyns 2001 8 31 35 11164350 10.1016/S1068-607X(00)00067-6
Ravaldi C Vannacci A Ricca V [Cardiac complications of anorexia nervosa] [in Italian] Recenti Prog Med 2003 94 267 270 12793099
Brumberg JJ Fasting Girls: The Emergence of Anorexia Nervosa as a Modern Disease is one of them 1988 Harvard University Press
Morgan J Crisp A Use of Leucotomy for Intractable Anorexia Nervosa: A Long-Term Follow-Up Study Int J Eat Disord 2000 27 249 258 10694710 10.1002/(SICI)1098-108X(200004)27:3<249::AID-EAT1>3.0.CO;2-K
Kennedy S Goldbloom D Current perspectives on drug therapies for anorexia nervosa and bulimia nervosa Drugs 1991 41 367 377 1711444
Agras W Brandt H Bulik C Dolan-Sewell R Fairburn C Halmi K Herzog D Jimerson DCKA Kaye WH le Grange D Lock J Mitchell JE Rudorfer MV Street LL Striegel-Moore R Vitousek KM Walsh BT Wilfley DE. Report of the National Institutes of Health workshop on overcoming barriers to treatment research in anorexia nervosa Int J Eat Disord 2004 35 509 521 15101067 10.1002/eat.10261
Sodersten P Bergh C Ammar A Anorexia nervosa: towards a neurobiologically based therapy Eur J Pharmacol 2003 480 67 74 14623351 10.1016/j.ejphar.2003.08.093
Stocker M Ca(2+)-activated K+ channels: molecular determinants and function of the SK family Nature Rev Neurosci 2004 5 758 770 15378036 10.1038/nrn1516
Raboch J Faltus F Sexuality of Women with Anorexia-Nervosa Acta Psychiatrica Scandinavica 1991 84 9 11 1927571
Schmidt U Tiller J Blanchard M Andrews B Treasure J Is there a specific trauma precipitating anorexia nervosa? Psychological Medicine 1997 27 523 530 9153673 10.1017/S0033291796004369
Joughin NA Crisp AH Gowers SG Bhat AV The Clinical-Features of Late Onset Anorexia-Nervosa Postgrad Med J 1991 67 973 977 1775420
Myslobodsky M Laparophobia: a cognitive perspective on appetite control in anorexia nervosa Obes Rev 2005 6 217 28 16045637 10.1111/j.1467-789X.2005.00188.x
Myslobodsky M The Fallacy of Mother's Wisdom: A Critical Perspective on Health Psychology 2004 New Jersey: World Scientific Publishing
Hebebrand J Exner C Hebebrand K Holtkamp C Casper R Remschmidt H Herpertz-Dahlmann B Klingenspor M Hyperactivity in patients with anorexia nervosa and in semistarved rats: evidence for a pivotal role of hypoleptinemia Physiol Behav 2003 79 25 37 12818707 10.1016/S0031-9384(03)00102-1
Guelfi J C M [Anorexia nervosa. A clinical and psychometric study] Ann Med Psychol (Paris) 1983 141 381 403 6651083
Casper R Heller W 'La douce indifference' and mood in anorexia nervosa: neuropsychological correlates Prog Neuropsychopharmacol Biol Psychiat 1991 15 15 23 10.1016/0278-5846(91)90037-2
Casper RC Behavioral activation and lack of concern, core symptoms of anorexia nervosa? Int J Eating Dis 1998 24 381 393 10.1002/(SICI)1098-108X(199812)24:4<381::AID-EAT5>3.0.CO;2-Q
Obermair GJ Kaufmann WA Knaus HG Flucher BE The small conductance Ca2+-activated K+ channel SK3 is localized in nerve terminals of excitatory synapses of cultured mouse hippocampal neurons European Journal of Neuroscience 2003 17 721 731 12603262 10.1046/j.1460-9568.2003.02488.x
Ocana M Cendan C Cobos E Entrena J Baeyens J Potassium channels and pain: present realities and future opportunities Eur J Pharmacol 2004 500 203 219 15464034 10.1016/j.ejphar.2004.07.026
Sah P Faber ESL Channels underlying neuronal calcium-activated potassium currents Progress in Neurobiology 2002 66 345 353 12015199 10.1016/S0301-0082(02)00004-7
Liegeois JF Mercier F Graulich A Graulich-Lorge F Scuvee-Moreau J Seutin V Modulation of small conductance calcium-activated potassium (SK) channels: A new challenge in medicinal chemistry Current Medicinal Chemistry 2003 10 625 647 12678783 10.2174/0929867033457908
Tzounopoulos T Stackman R Enhancing synaptic plasticity and memory: A role for small-conductance Ca2+-activated K+ channels Neuroscientist 2003 9 434 439 14678575 10.1177/1073858403259282
Cabezas-Fernendez C Martin-Montiel ED Buno-Buceta W Cellular mechanisms regulating neuronal excitability: Functional implications and in epilepsy Revista De Neurologia 2003 36 738 743 12717653
Behnisch T Reymann KG Inhibition of apamin-sensitive calcium dependent potassium channels facilitate the induction of long-term potentiation in the CA1 region of rat hippocampus in vitro Neurosci Lett 1998 25 91 94 9774157 10.1016/S0304-3940(98)00612-0
Fournier C Kourrich S Soumireu-Mourat B Mourre C Apamin improves reference memory but not procedural memory in rats by blocking small conductance Ca2+-activated K+ channels in an olfactory discrimination task Behav Brain Res 2001 121 81 93 11275286 10.1016/S0166-4328(00)00387-9
Bond CT Herson PS Strassmaier T Hammond R Stackman R Maylie J Adelman JP Small conductance Ca2+-activated K+ channel knock-out mice reveal the identity of calcium-dependent afterhyperpolarization currents J Neurosci 2004 24 5301 5306 15190101 10.1523/JNEUROSCI.0182-04.2004
Stackman RW Hammond RS Linardatos E Gerlach A Maylie J Adelman JP Tzounopoulos T Small conductance Ca2+-activated K+ channels modulate synaptic plasticity and memory encoding J Neurosci 2002 22 10163 10171 12451117
Blank T Nijholt I Kye M Radulovic J Spiess J Small-conductance, Ca2+-activated K+ channel SK3 generates age-related memory and LTP deficits Nat Neurosci 2003 6 911 912 12883553 10.1038/nn1101
Bosch M Kelly M Ronnekleiv O Distribution, neuronal colocalization, and 17beta-E2 modulation of small conductance calcium-activated K(+) channel (SK3) mRNA in the guinea pig brain Endocrinology 2002 143 1097 107 11861537 10.1210/en.143.3.1097
Obermair G Kaufmann W Knaus H BE F The small conductance Ca2+-activated K+ channel SK3 is localized in nerve terminals of excitatory synapses of cultured mouse hippocampal neurons Eur J Neurosci 2003 17 721 731 12603262 10.1046/j.1460-9568.2003.02488.x
Stocker M Pedarzani P Differential distribution of three Ca2þ-activated Kþ channel subunits, SK1, SK2, and SK3, in the adult rat central nervous system Mol Cell Neurosci 2000 15 476 493 10833304 10.1006/mcne.2000.0842
Wolfart J Neuhoff H Franz O Roeper J Differential expression of the small-conductance, calcium-activated potassium channel SK3 is critical for pacemaker control in dopaminergic midbrain neurons J Neurosci 2001 21 3443 3456 11331374
Brodie M McElvain M Bunney E Appel S Pharmacological reduction of small conductance calcium-activated potassium current (SK) potentiates the excitatory effect of ethanol on ventral tegmental area dopamine neurons J Pharmacol Exp Ther 1999 290 325 333 10381795
Gorelova N Seamans JK Yang CR Mechanisms of dopamine activation of fast-spiking interneurons that exert inhibition in rat prefrontal cortex J Neurophysiol 2002 88 3150 3166 12466437
Ayton A Dietary polyunsaturated fatty acids and anorexia nervosa: is there a link? Nutr Neurosci 2004 7 1 12 15085553 10.1080/1028415042000194621
Deep A Nagy L Weltzin T Rao R Kaye W Premorbid onset of psychopathology in long-term recovered anorexia nervosa nt J Eat Disord 1995 17 291 297
Yamada SI Takechi H Kanchiku I Kita T Kato N Small-conductance Ca2+-dependent K+ channels are the target of spike-induced Ca2+ release in a feedback regulation of pyramidal cell excitability J Neurophysiol 2004 91 2322 2329 14695351 10.1152/jn.01049.2003
Faber ESL Sah P Physiological role of calcium-activated potassium currents in the rat lateral amygdala J Neurosci 2002 22 1618 1628 11880492
Vogalis F Storm JF Lancaster B SK channels and the varieties of slow after-hyperpolarizations in neurons European Journal of Neuroscience 2003 18 3155 3166 14686890 10.1111/j.1460-9568.2003.03040.x
Wolfart J Roeper J Selective coupling of T-type calcium channels to SK potassium channels prevents intrinsic bursting in dopaminergic midbrain neurons J Neurosci 2002 22 3404 3413 11978817
Greffrath W Magerl W Disque-Kaiser U Martin E Reuss S Boehmer G Contribution of Ca2+-activated K+ channels to hyperpolarizing after-potentials and discharge pattern in rat supraoptic neurones Journal of Neuroendocrinology 2004 16 577 588 15214861 10.1111/j.1365-2826.2004.01204.x
Sailer CA Hu H Kaufmann WA Trieb M Schwarzer C Storm JF Knaus HG Regional differences in distribution and functional expression of small-conductance Ca2+-activated K+ channels in rat brain J Neurosci 2002 22 9698 9707 12427825
Moodie D Anorexia and the heart. Results of studies to assess effects Postgrad Med J 1987 81 46 48
Alvin P Zogheib J Rey C Losay J [Severe complications and mortality in mental eating disorders in adolescence. On 99 hospitalized patients] [Article in French] Arch Fr Pediatr 1993 50 755 762 8060204
Frolich J von Gontard A Lehmkuhl G Pfeiffer E Lehmkuhl U Pericardial effusions in anorexia nervosa European Child & Adolescent Psychiatry 2001 10 54 57 11315536 10.1007/s007870170047
Cooke R Chambers J Anorexia nervosa and the heart Br J Hosp Med 1995 54 313 317 8556209
Levine H Isner J Salem D Primary versus secondary mitral valve prolapse: clinical features and implications Clin Cardiol 1982 5 371 375 7049478
Rosenberg C Derman G Grabb W Buda A Hypomastia and mitral-valve prolapse. Evidence of a linked embryologic and mesenchymal dysplasia N Engl J Med 1983 309 1230 1232 6633572
Onesti M Mezzana P Martan A Scuderi N Breast asymmetry: a new vision of this malformation Acta Chir Plast 2004 46 8 11 15274471
Vincent G The molecular genetics of the long QT syndrome: genes causing fainting and sudden death Annual Rev Med 1998 49 263 274 9509262 10.1146/annurev.med.49.1.263
Buchowski K Pardo J Ringel R Guarda A Inducible syncope in anorexia nervosa: two case reports Int J Eat Disord 2004 35 359 362 15048952 10.1002/eat.10262
Critchley H Taggart P Sutton P Holdright D Batchvarov V Hnatkova K Malik M Dolan R Mental stress and sudden cardiac death: asymmetric midbrain activity as a linking mechanism Brain 2005 128 75 85 15496434 10.1093/brain/awh324
Roche F Estour B Kadem M Millot L Pichot V Duverney D Gaspoz J Barthélémy JC Alteration of the QT rate dependence in anorexia nervosa Pacing Clin Electrophysiol 2004 27 1099 1104 15305959 10.1111/j.1540-8159.2004.00591.x
Bankier B AB L Psychiatric disorders and coronary heart disease in women – a still neglected topic: review of the literature from 1971 to 2000 Psychother Psychosom 2002 71 133 140 12021555 10.1159/000056280
Roberts R R B Genetics and arrhythmias Annu Rev Med 2003 54 257 267 12525675 10.1146/annurev.med.54.073002.182112
Ker I Webb EC Ker JA Bekker PA The heart remembers: observations of cardiac memory in the Dorper sheep heart Onderstepoort Journal of Veterinary Research 2003 70 299 305 14971732
Patberg KW Rosen MR Molecular determinants of cardiac memory and their regulation Journal of Molecular and Cellular Cardiology 2004 36 195 204 14871546 10.1016/j.yjmcc.2003.11.015
Geller JC Rosen MR Persistent T-Wave Changes after Alteration of the Ventricular Activation Sequence – New Insights into Cellular Mechanisms of Cardiac Memory Circulation 1993 88 1811 1819 8403326
Alkon DL Ionic conductance determinants of synaptic memory nets and their implications for Alzheimer's disease Journal of Neuroscience Research 1999 58 24 32 10491569 10.1002/(SICI)1097-4547(19991001)58:1<24::AID-JNR4>3.0.CO;2-N
Rhodes JD Collison DJ Duncan G The human lens expresses calcium activated SK channels linking cataract, schizophrenia and myotonic dystrophy Invest Ophthalmol Vis Sci 2003 44 U211 U211 10.1167/iovs.03-0013
Hediger C Rost B Itin P Cutaneous manifestations in anorexia nervosa Schweiz Med Wochenschr 2000 120 565 575 10842772
Savvas M Treasure J Studd J Fogelman I Moniz C Brincat M The effect of anorexia nervosa on skin thickness, skin collagen and bone density Br J Obstet Gynaecol 1989 96 1392 1394 2620050
Strumia R Varotti E Manzato E Gualandi M Skin signs in anorexia nervosa DERMATOLOGY 2001 203 314 317 11752819 10.1159/000051779
Rustin M Foreman J Dowd P Anorexia nervosa associated with and worsening of chilblains J R Soc Med 1990 83 495 496 2231576
White K Rothe M Milanese A Grant-Kels J Perniosis in association with anorexia nervosa Pediatr Dermatol 1994 11 1 5 8170840
Flammer J Pache M Resink T Vasospasm, its role in the pathogenesis of diseases with particular reference to the eye Prog Retin Eye Res 2001 20 319 349 11286896 10.1016/S1350-9462(00)00028-8
Mikhailidis D Barradas M Jeremy J Gracey L Wakeling A Dandona P Heparin-induced platelet aggregation in anorexia nervosa and in severe peripheral vascular disease Eur J Clin Invest 1985 15 313 319 3938401
Gilbert J Weiss J Sattler A Koch J Ocular manifestations and impression cytology of anorexia nervosa Ophthalmology 1990 97 1001 1007 2402408
Sandow S Factors, fiction and endothelium-derived hyperpolarizing factor Clin Exp Pharmacol Physiol 2004 31 563 570 15479161 10.1111/j.1440-1681.2004.04048.x
Busse R Edwards G Feletou M Fleming I Vanhoutte P Weston A EDHF: Bringing the concepts together Trends Pharmacol Sci 2002 23 374 380 12377579 10.1016/S0165-6147(02)02050-3
LaBan M Wilkins J Sackeyfio A RS T Osteoporotic stress fractures in anorexia nervosa: etiology, diagnosis, and review of four cases Arch Phys Med Rehabil 1995 76 884 887 7668964 10.1016/S0003-9993(95)80558-3
Grinspoon S Baum H Lee K Anderson E Herzog D Klibanski A Effects of short-term recombinant human insulin-like growth factor I administration on bone turnover in osteopenic women with anorexia nervosa J Clin Endocrinol Metab 1996 81 3864 3870 8923830 10.1210/jc.81.11.3864
McAnarney E Greydanus D Campanella V Hoekelman R Rib fractures and anorexia nervosa J Adolesc Health Care 1983 4 40 43 6841238 10.1016/S0197-0070(83)80227-7
Olmos J Pesquera C Amado J Riancho J Gonzalez-Macias J Pathologic fracture of the sternum in a patient with anorexia nervosa [Article in Spanish] Rev Clin Esp 1990 186 23 25 2320765
Caregaro L Favaro A Santonastaso P Alberino F Di Pascoli L Nardi M Favaro S Gatta A Insulin-like growth factor 1 (IGF-1), a nutritional marker in patients with eating disorders Clin Nutr 2001 20 251 257 11407872 10.1054/clnu.2001.0397
Tustin K Gross J Hayne H Maternal exposure to first-trimester sunshine is associated with increased birth weight in human infants Developmental Psychobiology 2004 45 221 230 15549686 10.1002/dev.20030
Gale C Martyn C Kellingray S Eastell R Cooper C Intrauterine programming of adult body composition J Clin Endocrinol Metab 2001 86 267 272 11232011 10.1210/jc.86.1.267
Pasamanick B Knobloch H Brain damage and reproductive casualty Am J Orthopsychiatry 1960 30 98 305
Barker DJP Mothers, babies and disease in later life 1994 London: BMJ Publishing
Lau C Rogers J Embryonic and fetal programming of physiological disorders in adulthood Birth Defects Res C Embryo Today 2004 72 300 312 15662709 10.1002/bdrc.20029
Plagemann A Fetal programming and functional teratogenesis: on epigenetic mechanisms and prevention of perinatally acquired lasting health risks Journal of Perinatal Medicine 2004 32 297 305 15346812 10.1515/JPM.2004.055
Mellemkjaer L Emborg C Gridley G Munk-Jorgensen P Johansen C Tjonneland A Kjaer SK Olsen JH Anorexia nervosa and cancer risk Cancer Causes & Control 2001 12 173 177 11246846 10.1023/A:1008974414116
Michels K Ekbom A Caloric restriction and incidence of breast cancer JAMA 2004 29 1226 1230 15010444 10.1001/jama.291.10.1226
Pardo L Voltage-gated potassium channels in cell proliferation Physiology 2004 19 285 292 15381757 10.1152/physiol.00011.2004
Korndorfer SR Lucas AR Suman VJ Crowson CS Krahn LE Melton LJ Long-term survival of patients with anorexia nervosa: A population-based study in Rochester, Minn Mayo Clinic Proceedings 2003 78 278 284 12630579
Vitousek KM Gray JA Grubbs KM Caloric restriction for longevity: I. Paradigm, protocols and physiological findings in animal research European Eating Disorders Review 2004 12 279 299 10.1002/erv.594
Belohlavkova S Simak J Kokesova A Hnilickova O Hampl V Fenfluramine-induced pulmonary vasoconstriction: role of serotonin receptors and potassium channels Appl Physiol 2001 91 755 761
Taylor MS Bonev AD Gross TP Eckman DM Brayden JE Bond CT Adelman JP Nelson MT Altered expression of small-conductance Ca(2+)-activated K(+) (SK3) channels modulates arterial tone and blood pressure Circ Res 2003 93 124 131 12805243 10.1161/01.RES.0000081980.63146.69
McDaniel S Platoshyn O Yu Y Sweeney M Miriel V Golovina V Krick S Lapp BR Wang JY Yuan JX. Anorexic effect of K+ channel blockade in mesenteric arterial smooth muscle and intestinal epithelial cells Appl Physiol 2001 91 2322 2333
Michelakis E Weir E Anorectic drugs and pulmonary hypertension from the bedside to the bench Am J Med Sci 2001 321 292 299 11307870 10.1097/00000441-200104000-00009
Michelakis E Anorectic drugs and vascular disease: the role of voltage-gated K+ channels Vascul Pharmacol 2002 38 51 59 12378823 10.1016/S1537-1891(02)00126-X
Ptacek L Channelopathies: ion channel disorders of muscle as a paradigm for paroxysmal disorders of the nervous system Dig Dis Sci 1999 44 94S 96S 10490046
Ptacek LJ Ligand-gated ion channelopathies – Mutations in different genes causing one disease Neurology 2000 55 1429 1430 11094094
Lawson K Is there a role for potassium channel openers in neuronal ion channel disorders? Expert Opin Investig Drugs 2000 9 2269 2280 11060806 10.1517/13543784.9.10.2269
Chandy K Fantino E Wittekindt O Kalman K Tong L Ho T Gutman GA Crocq MA Ganguli R Nimgaonkar V Morris-Rosendahl DJ Gargus JJ. Isolation of a novel potassium channel gene hSKCa3 containing a polymorphic CAG repeat: a candidate for schizophrenia and bipolar disorder? Mol Psychiatry 1998 3 32 37 9491810 10.1038/sj.mp.4000353
Austin CP Holder DJ Ma L Mixson LA Caskey CT Mapping of hKCa3 to chromosome 1q21 and investigation of linkage of CAG repeat polymorphism to schizophrenia Molec Psychiat 1999 4 1999
Wittekindt O Jauch A Burgert E Scharer L Holtgreve-Grez H Yvert G Imbert G Zimmer JH M R Macher J-P Chiaroni P van Calker DC M-A Morris-Rosendahl DJ The human small conductance calcium-regulated potassium channel gene (hSKCa3) contains two CAG repeats in exon 1, is on chromosome 1q21.3, and shows a possible association with schizophrenia Neurogenetics 1998 1 259 265 10732800 10.1007/s100480050038
Koronyo-Hamaoui M Danziger Y Frisch A Stein D Leor S Laufer N Carel CFS Minoumi M Apter A Goldman B Barkai G Weizman A Gak E Association between anorexia nervosa and the hsKCa3 gene: a family-based and case control study Mol Psychiatry 2002 7 82 85 11803450 10.1038/sj.mp.4001931
Ben-Dor D Laufer N Apter A Frisch A Weizman A Heritability, genetics and association findings in anorexia nervosa Israel J Psychiatry & Sciences 2002 39 262 270
Patchell R Fellows H Humphries L Neurologic complications of anorexia nervosa Acta Neurol Scand 1994 89 111 116 8191873
Bridgers S Epileptiform abnormalities discovered on electroencephalographic screening of psychiatric inpatients Arch Neurol 1987 44 312 316 2881532
Stevens J Sachdev K V M Behavior disorders of childhood and the electroencephalogram Arch Neurol 1968 18 160 177 5636070
de Silva HA Aronson JK Grahame-Smith DG Jobst KA Smith AD Abnormal function of potassium channels in platelets of patients with Alzheimer's disease Lancet 1998 352 1590 1593 9843105 10.1016/S0140-6736(98)03200-0
Bond CT Sprengel R Bissonnette JM Kaufmann WA Pribnow D Neelands T Storck T Baetscher MJ J Maylie J Knaus H-G Seeburg PHA J P Respiration and parturition affected by conditional overexpression of the Ca(2+)-activated K(+) channel subunit, SK3 Science 2000 289 1942 1946 10988076 10.1126/science.289.5486.1942
Herzog A Psychoneuroendocrine aspects of temporolimbic epilepsy – Part I. Brain, reproductive steroids, and emotions Psychosomatics 1999 40 95 101 10100430
Schwartzkroin PA Baraban SC Hochman DW Osmolarity, ionic flux, and changes in brain excitability Epilepsy Research 1998 32 275 285 9761327 10.1016/S0920-1211(98)00058-8
Amiry-Moghaddam M Williamson A Palomba M Eid T de Lanerolle NC Nagelhus EA Adams ME Froehner SC Agre P Ottersen OP Delayed K+ clearance associated with aquaporin-4 mislocalization: Phenotypic defects in brains of alpha-syntrophin-null mice Proc Natl Acad Sci U S A 2003 100 13615 13620 14597704 10.1073/pnas.2336064100
Eid T Lee TSW Thomas MJ Amiry-Moghaddam M Bjornsen LP Spencer DD Agre P Ottersen OP de Lanerolle NC Loss of perivascular aquaporin 4 may underlie deficient water and K+ homeostasis in the human epileptogenic hippocampus Proc Natl Acad Sci U S A 2005 102 1193 1198 15657133 10.1073/pnas.0409308102
Morgan MA Schulkina J Pfaff DW Estrogens and non-reproductive behaviors related to activity and fear Neurosci Biobehav Rev 2004 28 55 63 15036933 10.1016/j.neubiorev.2003.11.017
Kelly MJ Ronnekleiv OK Ibrahim N Lagrange AH Wagner EJ Estrogen modulation of K+ channel activity in hypothalamic neurons involved in the control of the reproductive axis Steroids 2002 67 447 456 11960620 10.1016/S0039-128X(01)00181-7
Misra M Aggarwal A Miller KK Almazan C Worley M Soyka LA Herzog DB Klibanski A Effects of anorexia nervosa on clinical, hematologic, biochemical, and bone density parameters in community-dwelling adolescent girls Pediatrics 2004 114 1574 1583 15574617 10.1542/peds.2004-0540
Andersen A Unmasking a current medical pretender: anorexia nervosa in gastrointestinal practice Eur J Gastroenterol Hepatol 2004 16 1123 1125 15489569 10.1097/00042737-200411000-00006
Blank T Nijholt I Kye M Spiess J Small conductance Ca2+-activated K+ channels as targets of CNS drug development Curr Drug Targets CNS Neurol Disord 2004 3 161 167 15180477 10.2174/1568007043337472
Iversen L GABA pharmacology – what prospects for the future? Biochem Pharmacol 2004 68 1537 1540 15451396 10.1016/j.bcp.2004.06.039
Higgs G Molecular genetics: the Emperor's clothes of drug discovery? Drug Discov Today 2004 9 727 729 15450232 10.1016/S1359-6446(04)03143-5
Myslobodsky MS Golovchinsky V Mintz M Ketamine – Convulsant or Anti-Convulsant Pharmacol Biochem Behav 1981 14 27 33 7465607 10.1016/0091-3057(81)90099-X
Mills I Park G Manar A Merriman R Treatment of compulsive behaviour in eating disorders with intermittent ketamine infusions Qjm-Monthly J Assoc Physicians 1998 91 493 503
Ossowska G Klenk-Majewska B G S The effect of NMDA antagonists on footshock-induced fighting behavior in chronically stressed rats J Physiol Pharmacol 1997 48 127 135 9098833
Rogoz Z Skuza G Maj J Danysz W Synergistic effect of uncompetitive NMDA receptor antagonists and antidepressant drugs in the forced swimming test in rats Neuropharmacology 2002 42 1024 1030 12128003 10.1016/S0028-3908(02)00055-2
|
16144551
|
PMC1260012
|
CC BY
|
2021-01-04 16:39:08
|
no
|
Ann Gen Psychiatry. 2005 Sep 6; 4:15
|
utf-8
|
Ann Gen Psychiatry
| 2,005 |
10.1186/1744-859X-4-15
|
oa_comm
|
==== Front
BMC BiochemBMC Biochemistry1471-2091BioMed Central London 1471-2091-6-201619427910.1186/1471-2091-6-20Research ArticleComparison of Mycobacterium tuberculosis isocitrate dehydrogenases (ICD-1 and ICD-2) reveals differences in coenzyme affinity, oligomeric state, pH tolerance and phylogenetic affiliation Banerjee Sharmistha [email protected] Ashok [email protected] RaviPrasad [email protected] Vishwa Mohan [email protected] Seyed E [email protected] Centre for DNA Fingerprinting and Diagnostics, ECIL Road, Nacharam, Hyderabad, 500076, India2 Central JALMA Institute for Leprosy, Tajganj, Agra 282001, India3 Jawaharlal Nehru Centre for Advanced Scientific Research (JNCASR), Jakkur, Bangalore 560012, India2005 29 9 2005 6 20 20 8 2 2005 29 9 2005 Copyright © 2005 Banerjee et al; licensee BioMed Central Ltd.2005Banerjee 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
M.tb icd-1 and M.tb icd-2, have been identified in the Mycobacterium tuberculosis genome as probable isocitrate dehydrogenase (ICD) genes. Earlier we demonstrated that the two isoforms can elicit B cell response in TB patients and significantly differentiate TB infected population from healthy, BCG-vaccinated controls. Even though immunoassays suggest that these proteins are closely related in terms of antigenic determinants, we now show that M.tb icd-1 and M.tb icd-2 code for functional energy cycle enzymes and document the differences in their biochemical properties, oligomeric assembly and phylogenetic affiliation.
Results
Functionally, both M.tb ICD-1 and ICD-2 proteins are dimers. Zn+2 can act as a cofactor for ICD-1 apart from Mg+2, but not for ICD-2. ICD-1 has higher affinity for metal substrate complex (Km (isocitrate) with Mg++:10 μM ± 5) than ICD-2 (Km (isocitrate) with Mg++:20 μM ± 1). ICD-1 is active across a wider pH range than ICD-2, retaining 33–35% activity in an acidic pH upto 5.5. Difference in thermal behaviour is also observed with ICD-2 being active across wider temperature range (20°C to 40°C) than ICD-1 (optimum temperature 40°C). The isozymes are NADP+ dependent with distinct phylogenetic affiliations; unlike M.tb ICD-2 that groups with bacterial ICDs, M.tb ICD-1 exhibits a closer lineage to eukaryotic NADP+ dependent ICDs.
Conclusion
The data provide experimental evidence to show that the two open reading frames, Rv3339c (ICD-1) and Rv0066c (ICD-2), annotated as probable ICDs are functional TCA cycle enzymes with identical enzymatic function but different physio-chemical and kinetic properties. The differences in biochemical and kinetic properties suggest the possibility of differential expression of the two ICDs during different stages of growth, despite having identical metabolic function.
==== Body
Background
The central metabolic pathways in bacteria, especially in E.coli, have been extensively studied to understand the physiology of the organisms under altered carbon sources [1]. One of the key regulatory enzymes in the universal tri-carboxylic acid energy cycle is the isocitrate dehydrodenase that allosterically regulates the conversion of oxidative decarboxylation of D-isocitrate to α-ketoglutarate and CO2 in presence of a cofactor [2]. This rate-limiting step is the first NADPH yielding reaction of the TCA cycle [2]. Isocitrate dehydrogenase belongs to a family of enzymes that exhibits diversity with regard to amino acid composition, cofactor specificity, metal ion requirement and oligomeric state. NADP-linked ICDs have been purified and studied from a variety of eukaryotes and prokaryotes with detail investigations on their subunit composition and kinetic properties [3-11]. ICD from different organisms has been phylogenetically affiliated to three subfamilies [12]. Majority of the bacterial ICDs fall into subfamily I that includes archaeal and bacterial NADP dependent ICDs.
M. tb genome has two isoforms of isocitrate dehyrogenase, Rv3339c (ICD-1) and Rv0066c (ICD-2), both annotated as probable isocitrate dehydrogenase based on homology with other enzymes of the ICD family [13]. The two isoenzymes are share only ~14% identity at amino acid level. Earlier, we have pointed to a very unusual property of this TCA cycle enzyme demonstrating that the two isoforms can elicit B cell response in TB patients and significantly discriminate healthy, BCG-vaccinated controls from different groups of TB-infected population when compared to PPD and control antigen M.tb HSP 60 [14]. Although the two isoforms have primarily similar antigenic properties, little is known about their enzymatic properties. We now document differences in their biochemical properties, subunit composition and phylogenetic association. Our study provides experimental evidence to show that the two ORFs are TCA cycle enzymes with identical enzymatic function but different physio-chemical and kinetic properties.
Results
Expression, purification and quantification of M.tb ICD-1 and ICD-2
The over-expressed N-terminal His-tagged M.tb ICD-1 and M.tb ICD-2 were purified on a Nickel affinity column to 95% and 90% homogeneity, respectively (Figure 1, inset). The molecular sizes of the recombinant proteins M.tb ICD-1 and M.tb ICD-2 were determined by SDS-PAGE analyses and were found to be ~ 49 kDa and ~ 86 kDa respectively. The purification was carried out under native conditions for both the proteins from soluble fractions with an yield of 3.25 mg for M.tb ICD-1 and 10.21 mg for M.tb ICD-2 per 500 ml start culture [14].
Figure 1 M. tb Rv3339c (ICD-1) and Rv0066c (ICD-2) are expressed at the protein level. Antibody response to M. tb ICD-1 and ICD-2 was determined by ELISA. Sample representation of ten M. tb patient sera out of 125 showing high immunogenic response when compared with healthy controls. Y axis represents absorbance values at 492λ and X axis represents random patient sera tested for M. tb ICD-1 and ICD-2 antigenic response. 1 – 10, antigen ICD-1; 11 – 20, antigen ICD-2; 21 – 25, Healthy controls for ICD-1; 26 – 3, Healthy controls for ICD-2. Inset: Affinity purification of M. tb ICD-1 and M. tb ICD-2. The different lanes are: lanes 1: M. tb ICD-1; lane M: protein molecular size markers (200 kDa, 116 kDa, 97 kDa, 66 kDa, 45 kDa, 31 kDa and 21.5 kDa); lanes 2: M. tb ICD-2.
The ORFs encoding hypothetical protein M. tb ICD-1 (Rv3339c) and the probable ICD2 (Rv0066c) are functionally expressed as evident from serological evidences
The M. tb icd-1 and M. tb icd-2, both are annotated as the probable isocitrate dehydrogenase based on the sequence homology with ICD-family of enzymes. Serological evidences reveal the presence of antibody titers against both the purified proteins ICD-1 and ICD-2 in the infected sera samples [14]. Figure 1 is a sample representation of ten patients each for ICD-1 (sample 1–10) and ICD-2 (sample 11–20) with their respective healthy controls (Figure 1; 21–25, control reactions for ICD-1 and 26–30, control reactions for ICD-2) suggesting that both the ORFs encoding ICDs are expressed at the protein level as evident from antibodies in TB infected patient sera.
Biochemical characterization reveals differences between ICD-1 and ICD-2 in terms of pH and heat stability
In order to determine the kinetic parameters, the stabilizing components in the reaction like pH, temperature, salt requirement, metal ion components and coenzyme specificity were tested for each enzyme for optimal activity.
Figure 2 shows relative percentage of activity retained by the enzymes with respect to different pH. While the optimum pH for both isoenzyme is 7.5, M. tb ICD-1could tolerate a wider pH range retaining ~30% activity in acidic pH 5.5 to ~ 90% activity in alkaline pH 9.5 (Figure 2). In contrast, M. tb ICD-2 could retain only 10.22% of activity at a pH of 5.5, which gradually decreased to only 3.4% at pH 4, and less than 40% at 9.5 (Figure 2). These results clearly suggest that M. tb ICD-1 is active across a wider range of pH than M. tb ICD-2.
Figure 2 M. tb ICD-1 was active across a wider range of pH than M. tb ICD-2. Activity of M. tb ICD-1 and M. tb ICD-2 was tested as a function of pH (4 to 10). The buffers used for the experiment were: 30 mM Na-acetate buffer (pH 4.0 to pH 5.5), 20 mM phosphate buffer (pH 5.7 to pH 7), 30 mM imidazole buffer (pH 6 to pH 7) and 20 mM Tris buffer (pH 7.5 to pH 10).
Optimum temperature and thermostability of M. tb ICD-1 and M.tb ICD-2 were studied by varying the temperature of the reaction from 20°C to 70°C with incubation time of 30 minutes for each reaction. The optimum temperature for the activity of M. tb ICD-1 is ~40°C (Figure 3) while M. tb ICD-2 shows almost similar activity across 20°C to 40°C. Thermal stability of the two enzymes, however, varied. M. tb ICD-1 retained ~40% activity till 60°C, where M. tb ICD-2 was only ~5% active at that temperature (Figure 3). Thermal inactivation remained irreversible after incubation at 65°C for half an hour for M. tb ICD-1. M. tb ICD-2 could be renatured upon slow cooling till 55°C where partial activity was restored. Complete denaturation of the protein was observed at about 60°C (Figure 3). Similarity in thermal behaviour of the two enzymes is apparent, however fine differences could be registered.
Figure 3 ICD-1 and ICD-2 exhibit differential activity as a function of temperature. The enzyme activity was assayed at different temperatures (20°C to 70°C).
The decrease in the activity at higher salt concentration indicates involvement of ionic interactions during catalysis of M. tb ICD-1 and ICD-2
We determined the effect of NaCl on the stability as well as activity of the enzymes. The absence of NaCl in the reaction buffer did not affect the activity of M. tb ICD-1 significantly but M. tb ICD-2 showed higher activity in presence of upto 200 mM of NaCl as compared when no NaCl was added in the reaction. Presence of salt in the reaction buffer probably maintains the integrity of the enzymes. Higher concentrations of salt (above 200 mM) proved to be detrimental for the activities of both the enzymes (Figure 4a and 4b). The purified recombinant proteins, ICD-1 and ICD-2 degraded less at room temperature when dialysed against atleast 100 mM NaCl. The decrease in the activity at higher salt concentration indicates involvement of ionic interactions during catalysis. The reversibility of the inactivation of the enzyme activities after removal of excess salt has not been tested as the enzymes had a tendency to precipitate upon long exposure to room temperature or even 4°C during dialysis.
Figure 4 Activity of M. tb ICDs decrease at higher salt concentration. Activity of the enzymes ICD-1 (a) and ICD-2 (b) was measured by reduction of NADP at different concentrations of NaCl. The enzymes were active in presence of 200 mM NaCl above which the activity rapidly decreases.
M. tb ICD-1 and M. tb ICD-2 are NADP-dependent and have differential metal cofactors requirement
The coenzyme specificity of M. tb ICD-1 and M. tb ICD-2 was confirmed by checking the activity with both NADP+ and NAD+ (Figure 5a and 5b). The activity curves indicate that M. tb ICDs are NADP+ – dependent members of the isocitrate dehydrogenase family and shows no activity whatsoever in presence of NAD+.
Figure 5 The M. tb ICDs are NADP+ dependent. Coenzyme specificity of M. tb ICD-1(a) and M. tb ICD-2 (b) was assayed in presence of NADP+ and NAD+ as coenzymes. The activity curves clearly indicate that M. tb ICD-1 as well as M. tb ICD-2 are NADP+ dependent enzymes.
The two enzymes were tested for metal ion requirement with respect to four divalent metal ions, namely, Mg++, Zn++ and Mn++ and Ca++. It was apparent that M. tb ICD-1 accepts both Mg++and Zn++ as divalent metal ion cofactor but shows no activity in presence of either Mn++ or Ca++ (Figure 6a). This is unlike M. tb ICD-2, which accepts only Mg++ as metal ion and shows no activity with either Zn++, Ca++ or Mn++ (Figure 6b). The saturation kinetics indicates a complete saturation at 10 mM of metal ion for both Mg++ and Zn++ (data not shown).
Figure 6 Comparative rate curves for the enzyme activities in presence of metal ions. a) Rate curves of M. tb ICD-1 and b) Rate curves of M. tb ICD-2 in presence of Mg++, Zn++, Ca++ and Mn++ as cofactors. The details of the reactions are described in the method section.
Kinetic parameters and feedback inhibition of M. tb ICDs
The basic enzyme kinetics parameters Km, Vmax and Kcat for DL-isocitrate and NADP+ were determined for both ICD-1 and ICD-2 (Table 1). Km (isocitrate) in presence of either Mg++ or Zn++ for M. tb ICD-1 was calculated to be 10 μM ± 5 and 22 μM ± 7, respectively. For M. tb ICD-2, the value is 20 μM ± 1 in presence of Mg++ suggesting that M. tb ICD-1 has higher affinity for Mg-substrate complex than
Table 1 Kinetic parameters for M. tb ICD-1 and M. tb ICD-2 with respect to Mg++ and Zn++
Kinetic parameters With MgCl2 With ZnCl2
M. tb ICD-1
Km(isocitrate) 10 μM ± 5 22 μM ± 7
Vmax(isocitrate) 380 μM NADPH/min 190 μM NADPH/min
Kcat (isocitrate) 3.8 μM NADPH/min/pM enzyme 1.9 μM NADPH/min/pM enzyme
Km(NADP+) 125 μM ± 5 -
Vmax(NADP+) 400 μM NADPH/min -
Kcat (NADP+) 4 μM NADPH/min/pM enzyme -
M. tb ICD-2
Km(isocitrate) 20 μM ± 1 No activity
Vmax(isocitrate) 371.3 μM NADPH/min
Kcat (isocitrate) 37.13 μM NADPH/min/pM enzyme
Km(NADP+) 19.6 μM ± 6
Vmax(NADP+) 374 μM NADPH/min
Kcat (NADP+) 37.4 μM NADPH/min/pM enzyme
Ki(NADPH) 0.46 × 10-5 M.
Zn-substrate and also more readily binds to metal-substrate complex than M.tb ICD-2. The observation that these enzymes showed no activity upon pre-incubation with isocitrate and NADP+ in absence of metal ions suggests that they are unable to utilize free isocitrate as a substrate. Km (NADP) values in presence of Mg++ for M.tb ICD-1 and M. tb ICD-2 are 125 μM ± 5 and 19.6 μM ± 6, respectively. Competitive inhibition was observed with NADPH versus NADP+for M. tb ICD-2 (Figure 7). The mean inhibitor constant, Ki, was calculated to be 0.46 × 10-5 M. Competitive inhibition of M.tb ICD-2 by NADPH indicates the regulation of this enzyme by feedback mechanism (Figure 7). Since M. tb ICD-1 showed very poor affinity for NADP+ (125 μM ± 5), similar feedback inhibition was not observed.
Figure 7 Lineweaver Burk plot for competitive inhibition of M. tb ICD-2.
Oligomeric assemblies of M. tb ICDs
Size exclusion chromatography was performed to check the oligomeric assembly of the recombinant M.tb ICD-1 and M.tb ICD-2. The chromatogram for M.tb ICD-1 showed two distinct peaks indicating presence of two oligomeric species in the solution (Figure 8). The first peak corresponds to a mass of ~200 kDa (tetrameric) while the major peak showed migration of the molecule as a mass of ~100 kDa (dimeric). The results show M.tb ICD-1 to exist in either minor tetrameric or major dimeric state. In order to evaluate the functional significance of the two oligomeric states in catalysis of M.tb ICD-1, each fraction was collected separately and checked for the activity. While the dimeric fraction showed complete activity, the tetrameric fraction displayed a rather insignificant activity (Figure 8, inset). A few dimeric species that might have originated due to distintegration of the tetrameric form may account for the insignificant activity of the collected tetrameric fraction. The monomeric form was previously checked for the activity and was found to be inactive (data not shown). Thus, the oligomeric functional form of the M.tb ICD-1 is a dimer.
Figure 8 M. tb ICD-1 is a homodimer as evident from gel filtration analysis. The elution profiles of recombinant M. tb ICD-1 on a Superdex-200 HR 10/30 column showed two peaks corresponding to tetrameric (200 KDa) and dimeric (100 KDa) state of the protein. Inset: Comparative activity of the collected fractions.
The gel filtration analysis were performed with high performance column Superose™ 6 10/300 GL for higher resolution using an automated chromatographic workstation (BioRad BioLogic Duoflow™) to confirm the oligomeric state of M. tb ICD-2. The experiments were performed with the concentrated protein (~400 μg) under both low (100 mM NaCl) and high (1 M NaCl) salt conditions. The chromatogram showed a peak corresponding to a tetramer (~320 KDa) under low salt condition, which was dissociated into a dimer (~180 KDa) at a high salt concentration of 1 M, but not into a monomer (Figure 9, lower panel). The broadening of the ICD-2 peak under low salt concentration at higher elution volume points to the existence of lower oligomeric species. Sharpness of the peak under 1 M salt corresponding to a dimeric size provides strong evidence that the most stable oligomeric state is a dimer.
Figure 9 Originally annotated as a monomer, M. tb ICD-2 exists as a homodimer. The elution profiles of recombinant M.tb ICD-2 on high performance column SuperoseTM 6 10/300 column. The lower panel of the chromatogram shows a peak corresponding to a tetramer (~320 KDa) under low salt condition, which was dissociated into a dimer (~180 KDa) at a high salt concentration of 1 M. The collected elution fractions run on 7% SDS-PAGE are shown under respective peaks. The upper panel represent chromatographic peaks corresponding to the protein markers; Thyroglobulin (669 KDa), Apoferritin (443 KDa), BSA (66 KDa) and Carbonic anhydrase (29 KDa). Inset: The calibration curve plotted as Ve/Vo versus log of molecular mass for calculating molecular weights of the oligomeric assembly.
These conclusions were further supported by UV (figure 10a) and chemical crosslinking data (figure 10b). It was observed that UV crosslinking gave a distinct dimeric species. However, chemical crosslinking with glutaraldehyde was more efficient and it showed bands corresponding to both dimer and trimer forms of M. tb ICD2.
Figure 10 Crosslinking profiles of M. tb ICD2. A. UV crosslinking profile, B. Chemical crosslinking using glutaraldehyde, C. Glutaraldehyde crosslinking with substrate (Isocitrate) and coenzyme (NADP). UV crosslinking profile clearly shows a band corresponding to a homodimeric species. Chemical crosslinking with glutaraldehyde in presence and absence of NADP and Isocitrate was more efficient and showed bands corresponding to both dimer and trimer forms of M.tb ICD 2. D. The gel filtration assay under an intermediate (500 mM) salt concentration revealed a peak representing a mixture of trimeric and dimeric species. The collected elution fractions were run on 7% SDS-PAGE. D (Inset): The calibration curve plotted as Ve/Vo versus log of molecular mass for calculating molecular weights of the oligomeric assembly.
We also performed chemical crosslinking in presence and absence of coenzyme (NADP) and substrate (Isocitrate). A band corresponding to a trimer was consistently present with the dimeric band (figure 10c). Since the gel filtration data under low salt condition showed a tendency of forming lower oligomeric states, we further explored if under any salt condition a trimeric species that was observed in crosslinking could be obtained. Therefore the gel filtration assay was repeated under an intermediate (500 mM) salt concentration. Figure 10d shows a peak representing a mixture of trimeric and dimeric species under this buffer condition. Therefore, we conclude that M. tb ICD-2 is not a monomeric protein under any conditions tested in this study. Eventhough it can exist in higher oligomerc forms, dimer represented the most stable form that could not be further dissociated into monomers in our buffer condition. The above result equivocally demonstrate that M. tb ICD-2 is not a monomer and exists in higher oligomeric state in contrary to what has been reported in the annotation of H37Rv genome.
M. tb ICD-1 and M. tb ICD-2 have different phylogenetic affiliations
The ICD-1 and ICD-2 were aligned at the protein sequence level, with the corresponding sequences from a range of organism from both prokaryotes and eukaryotes, after a BLAST search (data not shown, but available upon request). Based on the sequence alignment using CLUSTAL, phylogenetic analyses were carried out. Neighbour Joining rooted trees with Thermotoga maritima isocitrate dehydrogenase as outgroup were constructed for both ICD-1 and ICD-2. The confidence was assessed by bootstrap analysis as described in Materials and Methods. The results reveal a closer relationship of the functionally conserved residues of M. tb ICD-1 with eukaryotic NADP+ dependent isocitrate dehydrogenases (Figure 11). The closest prokaryotic neighbour of M. tb ICD-1 was Bifidobacterium longum. The NADP-dependent isocitrate dehydrogenases of the following prokaryotes were found to cluster with M. tb ICD-1: Spingobium yanoikuyae, Caulobacter crescentes, Agrobacterium tumefacians, Brucella sp., Sinnorhizobium meliloti etc (Figure 11). M. tb ICD-1 showed no similarity with any of the mycobacterial ICDs in the protein database, including the second isoform of Mycobacterium tuberculosis, M. tb ICD-2. The phylogenetic tree of M. tb ICD-2 however, is a total contrast of M. tb ICD-1, where it clusters with other NADP+ dependent ICDs of gram-ve bacteria (Figure 12).
Figure 11 M. tb ICD 1 is closer to eukaryotic NADP+ ICDs. A 50% consensus bootstrap neighbour joining tree of M. tb ICD-1 with Thermotoga maritima as outgroup. The numerical values represent the confidence assessed by bootstrap analysis. The tree depicts a closer lineage of M. tb ICD-1 with eukaryotic NADP dependent isocitrate dehydrogenases. The amino acid sequence alingnment of ICDs from these organisms showed more than 65% identity with all the major catalytically important residues conserved (refer text).
Figure 12 M. tb ICD-2 is closer to prokaryotic NADP+ ICDs. Rooted neighbour joining phylogenetic tree of M. tb ICD-2 with Thermotoga maritima as outgroup. The numerical values represent the confidence assessed by bootstrap analysis. ICD-2 clusters with NADP+ dependent ICDs of gram-ve bacteria. M. tb ICD-2 has closest homology with Mycobacterium leprae showing 85.4% similarity at protein level. Other close neighbours are human pathogens like Pseudomonas aeruginosa (65%), Vibrio cholerae (60%), Neisseria menengitidis (59%) etc.
Discussion
Our results demonstrate for the first time experimentally that the two M. tb ORFs Rv3339c and Rv0066c code for functional TCA cycle enzyme that can catalyze the conversion of D-isocitrate to α-ketoglutarate and CO2 in presence of NADP as cofactor and differ in their biochemical properties. Interestingly, an altogether different functionality of these two isoforms in immune recognition was evident from our earlier work [14]. We have biochemically characterized the two isoforms of M. tb ICDs and established the differences between them.
Isocitrate dehydrogenase is an important regulatory enzyme of TCA cycle that has been intensively studied in both prokaryotes and eukaryotes (3–11). It lies at the branchpoint between glyoxylate shunt and citric acid cycle in prokaryotes where the switchover from TCA cycle to glyoxylate shunt depends upon the alteration in the biochemical parameters of isocitrate dehydrogenase. Isocitrate lyase of glyoxylate shunt pathway has much lower affinity for the isocitrate and cannot compete with isocitrate dehydrogenase for the substrate under normal conditions. It has been reported that phosphorylation of ICD controls the flux of isocitrate between the Krebs cycle and the glyoxylate pathway [15,16]. In E.coli, where glyoxylate bypass and citric acid cycle operate concurrently, the activity of a single, functional isocitrate dehydrogenase is closely monitored [17,18]. In M. tb, however, glyoxylate bypass is observed inside macrophages where C2 substrate is the main carbon source [19]. The occurrence of two isoforms of ICD in M. tb genome with the possibility of each having characteristic biochemical properties is interesting under such circumstances.
M. tb ICD-1 and ICD-2 follow a first order reaction and exhibit typical saturation kinetics. Km (isocitrate) value for M. tb ICD-1 clearly indicates a high affinity of this enzyme for isocitric acid as compared to M. tb ICD-2. Km (isocitrate) of some of the known NADP+ dependent ICD has been presented in Table 2 for reference and comparison [3-11]. Several other probable substrates for M. tb ICD-1 with close structural similarity were tested. Two such substrates were aspartate and glutamate. Both the compounds have close similarity with isocitric acid. However, the poor activity of the enzyme with these substrates confirmed that the enzyme is specific to only isocitrate as a substrate (data not shown).
Table 2 Comparison of NADP+-dependent isocitrate dehydrogenases from different organisms.
Organism Km(isocitrate) Km(NADP+)
Blastocladiella emersonii 20 μM 10 μM
Chlorobium limicola 45 μM ± 13 27 μM ± 10
Bacillus substilis 5.9 μM ± 0.9 14.5 μM ± 2.2
Escherichia coli 4.9 μM ± 0.2 19.6 μM ± 3.6
Synechocystis sp PCc 6803 59 μM 12 μM
Pyrococcus furiosus - 4400 μM
Aeropyrum pernix - 30 μM
Thermotoga moritima - 55.2 μM
Aeropyrum fulgidus - 30 μM
Mycobacterium phlei (ATCC-354) 74 μM 53 μM
M. tb ICD-1 10 μM ± 5 125 μM ± 5
M. tb ICD-2 20 μM ± 1 19.6 μM ± 6
Beef liver NADP+-IDH 1.7 μM 7.3 μM
Rat liver (cytosolic) 9.7 μM ± 2.9 11.5 μM ± 0.2
Porcine heart NADP+ IDH - 5 μM ± 0.19
The fact that M. tb ICD-1 could tolerate a broad range of both pH and temperature (Figure 2, 3) than M. tb ICD-2 indicates its robustness. The difference in the pH tolerance helps to postulate the possibility of differential expression of the two isoforms with ICD-1 being expressed during stationary phase when the intracellular pH is expected to vary over a wider range than log phase.
Km (NADP+) of M. tb ICD-1 showed poor affinity for NADP+ as compared to M. tb ICD-2 and other known NADP-dependent isocitrate dehydrogenases (Table 2). The poor affinity of M. tb ICD-1 to NADP+ warranted an investigation on whether dual co-enzyme specificity occurs in M. tb ICD-1 as reported in some archaeal bacteria [12]. We, therefore, compared the enzymatic activity of both the enzymes in presence of NADP+ and NAD+ (Figure 5a and 5b). It can be clearly seen that M. tb ICD-1 as well as M. tb ICD-2 accepts NADP+ and not NAD+ as a proton acceptor.
The homodimeric state of M. tb ICD-1 is the functionally active species, even though residual activity was noticed in tetrameric fraction which could be a reflection of the presence of a few dimeric species as a consequence of disintegration of the tetrameric forms. The chromatogram peak for M. tb ICD-2 corresponding to a tetramer (~320 KDa) under low salt condition, which was dissociated into a dimer (~180 KDa) at a high salt concentration of 1 M, but not into a monomer (Figure 9) provided a strong evidence that the most stable form of M. tb ICD-2 is a dimer. An intermediary trimeric form was observed in chemical crosslinking assays (Figure 10b and 10c), both in presence and absence of coenzyme NADP and substrate isocitric acid. The data were consistent with the gel filtration profile under an intermediate (500 mM) salt concentration (Figure 10d) where it showed a peak representing a mixture of trimeric and dimeric species. We therefore could conclude that M. tb ICD-2 is not a monomeric protein. Our result indicates M. tb ICD-2 exists in different higher oligomeric states which may follow the following equilibria: [tetramer] ⇔ [trimer] ⇔ [dimer]. However, the physiological relevance of the different oligomers could not be concluded from our experiments.
Earlier attempt to trace the evolution of ICDs to understand the adaptive role of isocitrate dehydrogenase in intracellular persistence of this pathogen by Steen et al [12] does not place M. tb ICD-2 phylogenetically. Proximity of the two M. tb ICDs with other isocitrate dehydrogenases was determined. Our results on phylogenetic analysis of M. tb ICD-1 revealed a closer relationship with eukaryotic NADP+ dependent ICDs (Figure 11) with more than 65% identity with that of Glycine max, Sus scrofa, Bos and Homo sapiens. M. tb ICD-1, indeed, is correctly placed in subfamily II that includes eukaryotic NADP dependent ICDs and a single bacterial ICD (Sphingomonas yanoikuyae) [12]. With NADP+ dependent isocitrate dehydrogenase of Sphingomonas yanoikuyae, M. tb ICD-1 has more than 65% similarity at primary structure level. Phylogenetic analysis of M. tb ICD-2 showed that the classical nomenclature applies to ICD-2 and it can be placed in subfamily I, the closest being M. leprae (Figure 12). The closest bacterial relative of M. tb ICD-1 as inferred by our study is NADP dependent isocitrate dehydrogenase of Bifidobacterium longum (Figure 11). Bifidobacterium sp. are gram positive, anaerobic, natural components of human intestinal microbiota [20]. This might be argued as a case of horizontal transfer or lateral transfer of gene amongst unrelated organisms across the boundaries of phylogenetic domains. Horizontal transfer of genes is a common occurrence in nature and accounts for almost 10–50% of genes in bacteria [21,22].
Conclusion
Several ORFs have been characterized since deciphering of M. tb genome [23-25]. Our data represent conclusive proof that the two ORFs, Rv3339c and Rv0066c, are functional TCA cycle enzyme and represent the first attempt to characterize these important members of the TCA cycle of Mycobacterium tuberculosis. Our studies conclusively reveal that both ICD-1 and ICD-2 are NADP+ dependent members of ICD family with the former having closer homology with eukaryotic ICDs and latter with prokaryotes. ICD-1 is a homodimer, while ICD-2 annotated as a monomer, exists in higher oligomeric forms, the dimer being the most stable. The two isoforms differ in their affinity for coenzyme NADP as represented by their Km(NADP) values (Table 1) and also with respect to pH tolerance and thermostability.M. tb ICD-2 is a more efficient enzyme as inferred by comparing Vmax(NADP)/Km(NADP) ratios for the two enzymes but M. tb ICD-1 is more robust in terms of pH tolerance and thermostability. The possibilities of differential expression of these two isoforms during different stages and conditions of growth cannot be ruled out even though the two isoforms have identical enzymatic function.
Methods
Cloning and purification of M. tb ICDs
The 1.230 kb (ICD-1) and 2.238 kb (ICD-2) long ORF was amplified from H37Rv genomic DNA and overexpressed in the pRSET-A/E.coli BL-21 (DE3) expression system as described earlier [14]. The overexpressed his-tagged recombinant protein was purified by Ni2+-nitrilotriacetate affinity chromatography.
Enzyme linked immunosorbent assays (ELISA)
ELISAs were performed with purified recombinant proteins, as described earlier [14], to check the B cell immune response in TB patient sera as evidence to the in vivo expression of the proteins.
Dehydrogenase kinetics/biochemical assays
Dehydrogenase activity was measured spectrophotometrically by monitoring the time dependent reduction of NADP+ to NADPH at 25°C in Unicam UV/Vis spectrometer at 340 nm, the absorbance maximum of NADPH. The standard assay solution contained 20 mM triethanolamine chloride buffer pH 7.5, 2 mM NADP+, 0.03 mM DL-isocitrate, 10 mM MgCl2/10 mM ZnCl2, 100 mM NaCl and 10 -100 pM of the enzyme in a final volume of 400 μL. Environmental parameters for the enzymes were measured by altering the pH of the buffer (range 4 – 10), temperature (20 – 65°C), concentration of substrate (0.01 mM – 0.18 mM), cofactor (0.1 – 2 mM), metal ion (Mg++, Zn++, Mn++; 1 – 12.5 mM) and salt (100 mM – 500 mM) requirement (as indicated in the respective figure legends. The pH dependence of the enzyme was measured using the following buffers: 30 mM Na-acetate buffer (pH 4.0 to pH 5.5), 20 mM phosphate buffer (pH 5.7 to pH 7), 30 mM imidazole buffer (pH 6 to pH 7) and 20 mM Tris buffer (pH 7.5 to pH 10). The cofactor specificity was checked with both NADP+ and NAD+. The heat denaturation was studied in 20 mm triethanolamine chloride buffer pH 7.5 in presence of 1% bovine serum albumin. Enzyme aliquotes were placed in tubes and incubated for 30 minutes in a water bath set at the required temperature (20°C – 65°C). After heating, aliquotes were immediately placed on ice and then assayed for remaining enzyme activity.
The kinetic analysis was carried out at 25°C and pH 7.5 in presence of either Mg++ or Zn++. Km was determined by altering the concentration of either the substrate or the coenzyme. The substrate concentration gradient varied from 0.01 mM to 0.75 mM, while NADP+ concentration was taken from 0.1 mM to 2 mM. The values were plotted as V vs S for calculating Km and Vmax for this first order reaction. The results were counter checked by double inverse Lineweaver-Burk plot. Competitive inhibition was observed with reduced NADP (NADPH) versus NADP to estimate inhibitor constant, Ki. Standard Km analysis was performed followed by repeating the assay with NADPH. Two concentrations of the inhibitor were tested, 0.002 mM and 0.005 mM. The uninhibited run provided the value of Km for the reaction and the inhibited run provided the apparent Km (Kmapp) for the reaction. Ki for the competitive inhibition was calculated by the formula (Km) (I)/Kmapp – Km.
Size exclusion chromatography
Size exclusion chromatography was performed at room temperature using FPLC equipped with Superdex-200 HR 10/30 column (Amersham Pharmacia Biotech) and SuperoseTM 6 10/300 GL (BioRad BioLogic Duo-flow™). Calibration of the columns were performed using protein molecular-mass standards for gel-filtration (Sigma, USA) as described elsewhere [26]. The void volume (Vo) was determined by running Blue Dextran on the column. The calibration curve was plotted as Ve/Vo versus log of molecular mass. A 2.4 mg/ml (for ICD-1) and 1.06 mg/ml (for ICD-2) concentrations of recombinant proteins were used for all gel filtration experiments. The columns were equilibrated with three bed volumes of the elution buffer prior to each run. Protein elution was monitored at A280.
UV and chemical crosslinking
UV crosslinking assays were performed to check the oligomeric assembly of M. tb ICD2. 5 μg of protein per reaction, in TrisCl buffer pH 8, was taken and exposed to UV in a UVP CL-1000 Ultraviolet crosslinker for 1 to 10 minutes at the rate of 1600 Joules/minute and fractioned later on 10% SDS-PAGE along with similar amount of untreated M. tb ICD2 as control. For chemical crosslinking, the protein was equilibrated in 20 mM phosphate buffer pH7.8. 10 mM Glutaraldehyde was used for all the chemical crosslinking reactions. The protein samples were incubated at 37°C, with or without glutaraldehyde. The reaction was stopped at different time points (10' or 20') using SDS loading dye containing 400 mM glycine and subsequently fractionated on 7% SDS-PAGE. 1 mM of either NADP or Isocitrate was used wherever required.
Sequence alignment and phylogenetic analysis
The amino acid sequence of M.tb ICD-1 and M.tb ICD-2 were compared against the NCBI protein database [27]. The sequences with the BLAST score upto e-153 or 65% identity were selected for construction of the phylogenetic tree. The sequences were aligned using CLUSTAL program. Manual alignment was done by Jalview [28] wherever required. The sequence alignment is available on request from the authors. Rooted phylogenetic tree were constructed using the software MEGA3 [29] using the amino acid sequence of Thermotoga maritima isocitrate dehydrogenase as outgroup. The confidence was assessed by bootstrap analysis (thousand replicates using default parameters).
List of abbreviations
ICD, isocitrate dehydrogenas; M. tb, Mycobacterium tuberculosis; TCA, tricarboxylic acid; NADP, nicotinamide adenine dinucleotide phosphate; BSA, Bovine Serum Albumin; NAD, nicotinamide adenine dinucleotide; NADPH, reduced nicotinamide adenine dinucleotide phosphate; bp, base pair(s); kb, kilobase pair(s); kDa, kilo Daltons, pM, pico molar; μM, micro molar.
Authors' contributions
SB has contributed in conception and designing of all experiments, biochemical assays and in-silico analysis, interpretation of data and writing the draft manuscript. AN carried out the biochemical assays and, participated in the sequence alignment and phylogenetic analysis. RP did the cloning and purification of proteins. VMK was involved in overall designing of this study, critical analysis of the data and also provided the biological material and quality control. SEH have been involved in complete supervision and guiding the work, providing the intellectual inputs, revising the draft version and preparing the final version of the manuscript.
Acknowledgements
This work was supported by research grants to SEH from the Council of Scientific and Industrial Research (CSIR) and Department of Biotechnology, Government of India. S B was supported by a Senior Research Fellowship from CSIR. We thank Dr. Ranjan Sen and Dr. Shekhar C Mande, Centre for DNA Fingerprinting and Diagnostics for helpful discussions. We wish to thank Sriramana, Laboratory of Molecular Genetics, Bibhusita, Jisha, Nancy and Irfan of Transcriptional Biology Laboratory, CDFD for their help and co-operation.
==== Refs
Holms H Flux analysis: a basic tool of microbial physiology Adv Microb Physiol 2001 45 271 340 11450111
Stryer L Biochemistry 1995 4 W. H. Freeman and Co., USA
Kanao T Kawamura M Fukui T Atomi H Imanaka T Characterization of isocitrate dehydrogenase from the green sulfur bacterium Chlorobium limicola, A carbon dioxide-fixing enzyme in the reductive tricarboxylic acid cycle Eur J Biochem 2002 269 1926 1931 11952794
Muro-Pastor MI Florencio FJ Purification and properties of NADP-isocitrate dehydrogenase from the unicellular cyanobacterium Synechocystis sp Eur J Biochem 1992 203 99 105 1730247
Reeves HC Daumy GO Lin CC Houston M NADP+-specific isocitrate dehydrogenase of Escherichia coli. I. Purification and characterization Biochim Biophys Acta 1972 258 27 39 4400493
Vasquez B Reeves HC NADP-specific isocitrate dehydrogenase from Escherichia coli. V. Multiple forms of the enzyme Biochim Biophys Acta 1981 660 16 22 7023543
Chen R Yang H A highly specific monomeric isocitrate dehydrogenase from Corynebacterium glutamicum Arch Biochem Biophys 2000 383 238 245 11185559
Balamir A NADP-linked isocitrate dehydrogenase from beef liver: a new method of purification and the effect of metal ion cofactor on its stability Biochem Med 1983 29 194 206 6860318
Jennings GT Minard KI McAlister-Henn L Expression and mutagenesis of mammalian cytosolic NADP+-specific isocitrate dehydrogenase Biochemistry 1997 36 13743 13747 9354646
Soundar S Danek BL Colman RF Identification by mutagenesis of arginines in the substrate binding site of the porcine NADP-dependent isocitrate dehydrogenase J Biol Chem 2000 275 5606 5612 10681542
Dhariwal KR Venkitasubramanian TA NADP-specific isocitrate dehydrogenase of Mycobacterium phlei ATCC 354: purification and characterization J Gen Microbiol 1987 133 2457 2460 3448153
Steen IH Madern D Karlstrom M Lien T Ladenstein R Birkeland NK Comparison of isocitrate dehydrogenase from three hyperthermophiles reveals differences in thermostability, cofactor specificity, oligomeric state, and phylogenetic affiliation J Biol Chem 2001 276 43924 43931 11533060
Cole ST Brosch R Parkhill J Garnier T Churcher C Harris D Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence Nature 1998 393 537 544 9634230
Banerjee S Nandyala A Podili R Katoch VM Murthy KJ Hasnain SE Mycobacterium tuberculosis (Mtb) isocitrate dehydrogenases show strong B cell response and distinguish vaccinated controls from TB patients Proc Natl Acad Sci USA 2004 101 12652 12657 15314217
Kornberg HL The role and control of the glyoxylate cycle in Escherichia coli Biochem J 1966 99 1 11 5337756
Singh SK Miller SP Dean A Banaszak LJ LaPorte DC Bacillus subtilis Isocitrate Dehydrogenase. a substrate analogue for Escherichia coli isocitrate dehydrogenase kinase/phosphatase J Biol Chem 2002 277 7567 7573 11751849
Holms WH Control of flux through the citric acid cycle and the glyoxylate bypass in Escherichia coli Biochem Soc Symp 1987 54 17 31 3332993
Holms H Flux analysis and control of the central metabolic pathways in Escherichia coli FEMS Microbiol Rev 1996 19 85 116 8988566
McKinney JD Honer zu BK Munoz-Elias EJ Miczak A Chen B Chan WT Swenson D Persistence of Mycobacterium tuberculosis in macrophages and mice requires the glyoxylate shunt enzyme isocitrate lyase Nature 2000 406 735 738 10963599
Xu J Gordon JI Honor thy symbionts Proc Natl Acad Sci USA 2003 100 10452 10459 12923294
Schnarrenberger C Martin W Evolution of the enzymes of the citric acid cycle and the glyoxylate cycle of higher plants. A case study of endosymbiotic gene transfer Eur J Biochem 2002 269 868 883 11846788
Lorencz MG Wackernagel W Bacterial gene transfer by natural genetic transformation in the environment Microbial Reviews 1994 58 563 602
Choudhary RK Mukhopadhyay S Chakhaiyar P Sharma N Murthy KJR Katoch VM Hasnain SE PPE antigen Rv2430c of Mycobacterium tuberculosis induces a strong B-Cell response Infection Immunity 2003 71 6338 6343 14573653
Chakhaiyar P Nagalakshmi Y Aruna B Murthy KJR Katoch VM Hasnain SE Regions of high antigenicity within the hypothetical PPE major polymorphic tandem repeat open-reading frame, Rv2608 show a differential humoral response and a low T cell response in various categories of patients with tuberculosis J Infect Dis 2004 190 1237 1244 15346333
Chakhaiyar P Hasnain SE Defining the mandate of tuberculosis research in a postgenomic era Med Princ Pract 2004 13 177 184 15181320
Chauhan R Mande SC Characterization of the Mycobacterium tuberculosis H37Rv alkyl hydroperoxidase AhpC points to the importance of ionic interactions in oligomerization and activity Biochem J 2001 354 209 215 11171096
Altschul SF Gish W Miller W Myers EW Lipman DJ Basic local alignment search tool J Mol Biol 1990 215 403 410 2231712
Clamp M Cuff J Searle SM Barton GJ The Jalview Java alignment editor Bioinformatics 2004 20 426 437 14960472
Kumar S Tamura K Nei M Integrated software for Molecular Evolutionary Genetics Analysis and sequence alignment Brief Bioinform 2004 5 150 163 15260895
|
16194279
|
PMC1260013
|
CC BY
|
2021-01-04 16:26:24
|
no
|
BMC Biochem. 2005 Sep 29; 6:20
|
utf-8
|
BMC Biochem
| 2,005 |
10.1186/1471-2091-6-20
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1181617152610.1186/1471-2407-5-118Research ArticlePre-exenterative chemotherapy, a novel therapeutic approach for patients with persistent or recurrent cervical cancer Lopez-Graniel Carlos [email protected] Rigoberto [email protected] Lucely [email protected] Aaron [email protected] David [email protected] Jose [email protected] Jesus [email protected] Myrna [email protected] Rocio [email protected] la Garza Jaime [email protected] Alfonso [email protected] Division of Surgery, Instituto Nacional de Cancerología, Mexico2 Division of Clinical Research, Instituto Nacional de Cancerología, Mexico3 Department of Pathology, Instituto Nacional de Cancerología, Mexico4 Department of CT scan, Instituto Nacional de Cancerología, Mexico5 Unidad de Investigación Biomédica en Cáncer. Instituto Nacional de Cancerología/Instituto de Investigaciones Biomédicas, UNAM, Mexico2005 19 9 2005 5 118 118 12 1 2005 19 9 2005 Copyright © 2005 Lopez-Graniel 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
Most cervical cancer patients with pelvic recurrent or persistent disease are not candidates for exenteration, therefore, they only receive palliative chemotherapy. Here we report the results of a novel treatment modality for these patients pre-exenterative chemotherapy- under the rational that the shrinking of the pelvic tumor would allow its resection.
Methods
Patients with recurrent or persistent disease and no evidence of systemic disease, considered not be candidates for pelvic exenteration because of the extent of pelvic tumor, received 3-courses of platinum-based chemotherapy. Response was evaluated by CT scan and bimanual pelvic examination; however the decision to perform exenteration relied on the physical findings. Toxicity to chemotherapy was evaluated with standard criteria. Survival was analyzed with the Kaplan-Meier method.
Results
Seventeen patients were studied. The median number of chemotherapy courses was 4. There were 9 patients who responded to chemotherapy, evaluated by bimanual examination and underwent pelvic exenteration. Four of them had pathological complete response. Eight patients did not respond and were not subjected to surgery. One patient died due to exenteration complications. At a median follow-up of 11 months, the median survival for the whole group was 11 months, 3 months in the non-operated and 32 months in those subjected to exenteration.
Conclusion
Pre-exenterative chemotherapy is an alternative for cervical cancer patients that are no candidates for exenteration because of the extent of the pelvic disease. Its place in the management of recurrent disease needs to be investigated in randomized studies, however, its value for offering long-term survival in some of these patients with no other option than palliative care must be stressed.
==== Body
Background
Cervical cancer continues to be an important health burden with a yearly incidence of almost half a million new cases in the world and a mortality rate of about 50% [1]. Currently, locally advanced disease is treated with concurrent cisplatin-based chemoradiation [2]. However, approximately in 25% of all patients treated for cervical carcinoma, the tumor will progress or recur locally [3,4], being the most common site of recurrence the pelvis. Thus, local relapse continues to be a significant problem for these patients, as tumor persistence or local recurrence in an irradiated pelvis indicates a very dismal prognosis [5,6].
Recurrent disease can be treated by a) chemoradiation if the primary disease was approached with surgery; b) palliative chemotherapy if recurrence is considered unresectable and the primary disease was treated with radiation or chemoradiation; and c) pelvic exenteration for selected cases with small, central disease even if primarily treated with chemoradiation or radiation. Pelvic exenteration involves en bloc resection of bladder, genital tract, and rectum; it was first described by Brunschwig in 1948 [7]. This procedure has curative potential in almost half of patients undergoing this procedure [6] and it is commonly reserved for only the small subgroup of recurrent disease patients who meet the "standard" criteria for exenteration (small, central tumors). However, most pelvic recurrences do show a diffuse growth pattern fixed to one or both pelvic side walls. These fixed recurrences are felt at physical examination as "pelvic fibrosis" with or without a dominant mass. Thus, pelvic fibrosis, is an ominous finding significantly related to nodal disease and fixation to pelvic side wall [8].
Due to these facts, the vast majority of recurrent cervical cancer patients are left with no curative options, therefore it is important to search for other therapeutic alternatives in patients that are not "standardly" considered for the classical exenterative procedure. The introduction of high-dose-rate intraoperative radiation therapy (HDR-IORT) combined with radical surgical resection has widened the scope of patients who may be offered surgery [9], however, this form of radiation delivery is not widely available. In addition, despite this modality of treatment provides a reasonable local control rate in patients who have failed prior surgery and/or definitive radiation, only those with complete gross resection at completion of surgery appear to benefit from this radical approach in the salvage setting [10].
Hokel et al., have recently described the laterally extended endopelvic resection (LEER) as a novel surgical salvage therapy to a selected subset of patients with locally advanced and recurrent cervical carcinoma involving the pelvic side wall. This consists in extending the lateral resection plane of pelvic exenteration to the medial aspects of the lumbosacral plexus, sacrospinous ligament, acetabulum, and obturator membrane to allow for resection with disease-free margins [11]. With this salvage approach, they have reported a 5-year survival probability of 46% for those patients considered only for palliation with current treatment options. Although these results are highly encouraging, severe postoperative complications occur in almost half of patients and the procedure is limited to tumors sized <5 cm with a recurrence-free interval from primary radiation treatment of >5 months, and to recurrences that do not involve the larger sciatic foramen; all forms of parietal pelvic side wall disease are not suited for this procedure [12].
Currently, a combination of cisplatin and paclitaxel has shown better response rate and progression- free survival than single agent cisplatin hence, combination chemotherapy as been regarded as the standard of care in patients to be treated with systemic palliative chemotherapy [13]. Chemotherapy however, as a definitive treatment for recurrent cervical cancer has solely a palliative role, with responses that are at best partial and of short duration, as a consequence, almost all patients eventually show progression and die from their disease. Because objective responses are seen in almost a third of these patients, we reasoned that a "local" consolidation would potentially render some of these responding patients free of disease. These observations prompted us to evaluate in a pilot study, a treatment modality we have called "pre-exenterative chemotherapy" in patients with "fixed" pelvic recurrence in the aimed to shrinking the pelvic recurrent tumor to then attempt, then, a "standard" pelvic exenteration.
Methods
All patients had histologically proven persistent or recurrent cervical carcinoma to primary radiation or chemoradiation. At the pelvic examination -under no anesthesia-, these patients were felt to have pelvic fibrosis and diagnosed as having a recurrence of diffuse infiltrative growth pattern (with or without a dominant mass) fixed or not to one or both pelvic side walls. Consequently, these patients were considered by the gynecologist team of our Institution (C L-G, A G-E, GM) to be unsuitable for pelvic exenteration regardless of the CT scan findings. Patients also had to meet the following inclusion criteria: Aged between 18 and 70 years; ECOG performance 0–1; adequate hematological, hepatic and renal functions as determined by: hemoglobin equal or higher than 10 g/L, leukocyte count higher than 4000/mm3, and a platelet count of at least 100 000/mm3, total bilirubin less than 1.5 times the normal upper limit (NUL), transaminases less than 1.5 times NUL, and normal levels of creatinine in serum; a normal posteroanterior chest X-ray as well as having the correspondent informed consent. The exclusion criteria included: severe systemic or uncontrolled disease (infection, central nervous system, metabolic, etc) that precluded the use of chemotherapy and further exenteration; concomitant treatment with any other experimental drug; mental illness and previous or concomitant malignancies except non-melanoma skin cancer. The study was approved by the Institutional Regulatory Boards.
Pre-exenterative chemotherapy
Chemotherapy was administered in an outpatient setting. Diverse chemotherapy schedules based on cisplatin or carboplatin were used as follows: Carboplatin AUC 5, d1, paclitaxel 135 mg/m2, d1 and gemcitabine 800 mg/m2 d1&8 (2 patients); carboplatin AUC 5–6, d1, and paclitaxel 135 mg/m2, d1 (3 patients); carboplatin AUC 5–6, d1, and 5FU 1 gr/m2 d1-5, (3 patients); cisplatin 100 mg/m2 d1, and 5FU 1 g/m2 d1-5, (4 patients); and cisplatin 100 mg/m2 d1 and gemcitabine 1 g/m2 d1&d8 (5 patients). Courses were administered every three weeks for a maximum of 6 courses. Conventional antiemetic therapy and ancillary medications were used during drug treatment. Chemotherapy was stopped in cases of disease progression or prohibitive toxicity.
Response and toxicity to pre-exenterative chemotherapy
Objective evaluation of response to chemotherapy using standard response criteria was not the primary objective of this study as this would have required that all patients had a well-defined and measurable mass. Instead a response to chemotherapy was defined when the pelvic disease was felt less fixed and/or the "fibrosis" was felt softer by the same team of gynecologists that performed the pre-chemotherapy evaluation. Toxicity to chemotherapy was evaluated according to the NCI Common Toxicity Criteria.
Pelvic exenteration
After pre-exenterative chemotherapy, patients were evaluated by the same team of gynecologists through pelvic examination (CT scan was not mandatory). However, the decision to proceed or not with the surgical procedure relied only on pelvic examination and was based on whether there was or not response as evaluated with above described criteria. The other criterion for no exenteration was a worsening of the general clinical condition of the patient regardless of the pelvic examination. Patients were followed every three months after completion of all treatment.
Survival
Overall survival was evaluated using the Kaplan-Meier method and was considered from the date of the diagnosis of the persistent or recurrent disease until the date of death of last visit.
Results
From May 1999 to March 2003, 17 patients were studied in this pilot trial. Baseline characteristics of patients (at diagnosis of their primary disease) are shown in Table 1. The mean age of patients was 43.3 years and all, but two, were squamous histology. FIGO stage distribution was as follows: one patient was IB1, four were stage IB2, five IIB, and seven, stage IIIB. Nine received radiation alone as the definitive treatment of their primary disease, four were treated with radiation plus extrafacial complementary hysterectomy, and four patients received chemoradiation with weekly cisplatin. A complete clinical response was achieved in 13 patients after the primary treatment, three had persistent disease and one progressed. All cases accrued in this study had local pelvic relapse and the median time to progression after primary treatment was 16 months (9–120) in the 13 cases that had complete response, whereas the time to progression for the persistent or progressive disease cases was 4 months (range 2–7 months).
All patients had histological confirmation of their recurrent disease. The clinical status at entering the study is shown in Table 2. All patients complained of pelvic pain. At physical pelvic exam the disease was felt as fixed to the pelvic wall in all cases, 5 (29%) unilaterally and 12 (71%) bilaterally. This was accompanied by unilateral leg edema in six cases, hydronephrosis in three (18%) and both findings: edema and hydronephrosis in three cases (18%).
Table 3 depicts the overall treatment received by the patients. The median number of cycles delivered was four (range 2–6 cycles). Evaluation of response following the aforementioned subjective criteria, performed by bimanual pelvic examination, was achieved in nine patients and these underwent the exenterative procedure. Among the eight patients not exenterated, three showed progression alone, one had clinical deterioration with no change at pelvic examination and four had progression and clinical deterioration. Objective response was also evaluated using classical criteria in measurable disease (complete, no evidence of disease, partial, >50 reduction in the product of the two longest perpendicular diameters of the measurable lesion; no change or stable, <50% decrease or <25% increase, and progressive disease >25% increase). According to this, within the 8 non-operated patients, only four had pre and post-chemotherapy CT scans, three had no response and one had progression. These data correlated well with that registered in the physical examination. On the contrary, in the nine operated patients, five patients had pre and post-chemotherapy evaluation, and all five had partial response. This, also correlates with that perceived in the clinical examination. It is remarkable that within the operable patients, in no case an objective complete response was observed despite four cases had a pathological complete response. Figure 1 shows that in the three cases with pathological complete response that had pre and post chemotherapy CT scan there was residual tumor after chemotherapy.
Chemotherapy was well tolerated. The most common side effect were nausea/vomiting grade 1 and 2, mild to moderate anemia was present in half of patients; all patients presented leukopenia and neutropenia which were grade 3 in five and three patients respectively. There were no episodes of infection or bleeding (Table 4).
In regard to the exenterative procedure, a total infraelevator exenteration was done in eight cases and one had anterior supraelevator exenteration. This patient was the one with positive surgical margins in the vaginal border. It is worth mentioning that, in this case, the transoperative frozen section of the vaginal margin was reported negative; however, the definitive histological analysis showed disease. The definitive histological analysis of the surgical specimens showed a complete pathological response in 4 cases, a residual disease ≤2 cm in four cases, and one case with a residual measuring 8 cm. In seven patients, the urinary diversion consisted of an ileocolonic conduit and an ileal conduit in two cases. Colostomy was done in the eight cases undergoing total exenteration, (Table 5). Regarding surgical morbidity, the mean surgical time was 6.3 hours (range 4.3–8); the mean of bleeding was 1860 mL (range 600–6000 mL). All patients required at least one unit of red blood cells being the mean number of units 3.4 (range 1–6). The mean hospital stay was 11.7 days (range 6–41), and the mean stay in the intensive care unit was 1.8 days (0–12 days). Among the perioperative and post-operative complications, one patient (11%) presented intestinal occlusion that resolved with non-operative measures, one had massive bleeding during the surgery (11%), there was one case with urinary fistula (11%) and two cases showed a perineal infection (22%). One patient (11%) died at day 120 post-exenteration due to sepsis. This patient was one of the four with a pathological complete response (Table 6).
All patients not subjected to exenteration showed disease progression and died within the ensuing months, being the median survival of only 3 months. The status of the operated patients is as follows: patient 1: Path CR, alive without disease at 62 months, patient 2: Path CR, alive without disease 59 months; patient 3: Residual of 2 cm, local and regional recurrence at 7 months post-exenteration, patient 4: Residual of 2 cm, alive without disease 52 months, patient 5: residual of 2 cm, local recurrence at 10 months, patient 6: residual of 2 cm, local recurrence at 7 months, patient 7: Path CR died at four months from surgical complications, patient 8: Path CR, died at 20 months from liver recurrence; patient 9: Residual of 8 cm, alive without disease 13 months. Thus, four of the nine operated patients are alive without disease. Median survival in the intention to treat was 11 months, being 3 versus 32 in the non-operated versus those that underwent exenteration (Figures 2 and 3).
Discussion
Although pelvic exenteration plays a definitive role in the management of recurrent cervical carcinoma, its impact in terms of the proportion of cervical cancer patients who benefit from such radical procedure has remained unchanged because it continues to be indicated in only very selected patients with small central pelvic recurrences. This fact, along with better medical support such as routine use of prophylactic heparin, antibiotics, nutritional support, and routine postoperative monitoring, have reduced the morbidity from pelvic exenteration [14].
In order to increase the proportion of patients in whom this salvage therapy could be attempted, we developed the modality of "pre-exenterative chemotherapy" under the rationale that systemic chemotherapy would allow the obtaining free surgical margins in patients undergoing the "standard" supra or infraelevator pelvic exenteration operation in situations where the extent of pelvic disease predicts that negative surgical margins would unlikely be obtained. The results of this pilot study demonstrate the feasibility of this approach, as nine (53%) out of the 17 patients included in this trial underwent pelvic exenteration obtaining disease-free margins in all but one case; four of them are alive without disease.
The clinical characteristics of the patients included in this study are remarkable in the sense that all of them were considered no suitable for pelvic exenteration according to standard criteria by the team of gynecologists of our Institution. This special subgroup of patients with recurrent disease is better defined if we look at their clinical characteristics: 13 of them relapsed at a median time of only 16 months whereas four were refractory to primary treatment and progressed within two to seven months All of them complained of pelvic pain, five had unilateral leg edema, three presented hydronephrosis and three cases had both signs. Both the short disease-free interval and the presence of one or more of the typical triad of signs and symptoms are either contraindications or factors predicting a very poor outcome after exenteration in most of the reported series [15-20].
Selecting the true candidates for pelvic exenteration is a difficult clinical dilemma in patients with recurrent cervical cancer after radiation therapy. Despite very thorough preoperative investigation, inoperable disease is discovered at the time of laparotomy in up to 50% of cases [21]. CT scanning is still one of the most extensively used diagnostic tool, however it may be difficult to differentiate recurrence from postoperative and post-radiation fibrosis [22,23]. MRI has been regarded superior to CT scan in visualization of the tumor and parametrial invasion in primary tumors [24]; dynamic contrast-enhanced subtraction MRI may differentiate between recurrent tumor and benign conditions [25]. However, when MRI has been used for determining surgical elegibility for pelvic exenteration its accuracy has been of 83% [26]. The difficulties encountered by the common imaging methods for evaluating the extent of disease, such as CT scan and MRI [27], have led to propose laparoscopy to select candidates to undergo the procedure [28,29], which proved to be effective as it may spare unnecessary laparotomy in half of the candidates patients [28].
It must be stressed, however, that all these imaging and laparoscopy efforts to predict resectability and avoid aborted exenterations are done in the setting of "classic indications" of pelvic exenterations, where the ultimate goals are increasing the efficacy of the procedure in terms of disease control and decreased morbidity and mortality. However, our approach in aimed at increasing the proportion of patients in whom this salvage therapy could be attempted under the rationale that systemic chemotherapy would allow obtaining free surgical margins in situations where the extent of pelvic disease predicts that negative surgical margins would unlikely be obtained and therefore exenteration could not be offered to these patients.
These considerations led us to rely on bimanual pelvic examination, which is a subjective test, as our principal criterion for deciding to perform the exenteration (as long as there was no regional or systemic disease evaluated by CT scan). We acknowledge that it would had been very valuable to have an objective pre and post-chemotherapy evaluation of the response by a CT scan, RMI, and/or PET scan in all the cases, however, only nine cases had CT scan pre and post therapy. Notwithstanding it is interesting to notice that although all our patients were "felt" by physical exam to have side wall fixation, the CT scan confirmed it only in six cases based on the criterion of having a less than 3 mm separation of the tumor from the pelvic muscles and/or vascular encasement [30]. It is important also to notice that in the four cases that were not candidates for exenteration after chemotherapy and had pre and post-chemotherapy CT scans, there was no response in three and progression in one, matching closely the findings of bimanual pelvic examinations. In contrast, the five cases that underwent exenteration and were felt to have responded by physical examination, with the criteria used, had partial responses according to the standard WHO criteria suggesting that, after all, CT scan can be a reliable method for evaluating the response to chemotherapy in the setting of pelvic recurrences in a previously irradiated site.
Nevertheless, an important observation is the fact that in three out of the four cases that achieved a pathological complete response, the CT scan was clearly positive for the presence of tumor. This is shown in figure 1, standing out patient 3 (Figure 1e–f) in whom the residual mass after chemotherapy (1e) measured 9 × 5 cm. This finding might suggest that exenteration could be useful after any degree of response to chemotherapy because the actual response to chemotherapy could be of greater magnitude than predicted by CT scan.
The use of chemotherapy in the palliative setting of persistent or recurrent pelvic disease, particularly in a patient who has received definitive radiation or chemoradiation treatment has very limited value. In a review on results of 190 advanced or recurrent disease patients treated with 14 different chemotherapy protocols, the overall response rate was 20.0% (4.2% complete response; 15.8% partial response), with a median response duration of 4.8 months [31]. In a recent phase III study comparing cisplatin versus cisplatin paclitaxel, the response rate in the subgroup with pelvic disease revealed onjective responses in 14(21%) of 66 patients treated with cisplatin alone and in 17(33%) of 52 patients treated with the combination, however, median survival was the same, 8.8 months and 9.7 months, respectively [13]. The response reported here using a platinum-based scheme was 55% (partial responses) in the nine patients with pre and post-chemotherapy evaluation by CT scan, however it was only 29% taking into account the 17 patients evaluated. This response rate as well as the observed toxicity, is within the range expected but no assumptions can be made on the efficacy of any of the schemes used.
So far there is information of the efficacy of chemotherapy in terms of pathological response in recurrent or advanced cervical cancer because chemotherapy is only used as a palliative measure and surgery is commonly not affered after chemotherapy. Here we demonstrate a pathological complete response rate of 44% in the nine patients treated (23.5% taking into the 17 patients), which is higher than obtained in neoadjuvant trials in locally advanced cervical cancer utilizing platinum-based schemes with newer drugs, such as gemcitabine [32,33], vinorelbine [34], paclitaxel [35,36], and irinotecan [37]. It is worthwhile noticing that the neoadjuvant trials with lower complete response rates were those that subjected more patients to surgery. The two studies with the lowest complete responses, 16% and 17% operated 89% and 95% of patients, respectively [35,36], whereas in the trial with 37.5% of complete response, the surgery rate was only 52% [34]. These data may explain our 44% of pathological complete response rates, since only 52.9% of our patients underwent surgery.
A noticeable finding of the present report is that the median survival of 11 months compares favorably with studies using systemic chemotherapy in the palliative setting, ranging from 6 to 10 months [38,39]; however, we must stress that half of patients (the non-operated) had a median survival of only 3 months which suggest that the patient population of patients had indeed very unfavorable clinical characteristics. Of outmost importance is the fact that from the operated patients four achieved pathological complete response despite having tomographic evidence of residual tumor and the median survival for these 9 patients taken to exenterative surgery was 32 months. The fact that there was "gross persistent" disease in the CT scan after chemotherapy support our view that the overestimation of pelvic disease, either by pelvic examination and/or imaging methods, hinder offering a potentially curative surgery to a huge proportion of patients with pelvic recurrence of cervical carcinoma. On these bases, we are just to start a randomized study comparing pre-exenterative chemotherapy and exenteration versus palliative chemotherapy alone in patients with pelvic disease that do not meet the criteria for pelvic exenteration.
Conclusion
The therapeutic modality here reported, called pre-exenterative chemotherapy, is a therapeutic alternative for cervical cancer patients with recurrent or persistent disease limited to the pelvis not usually considered candidates for "classical" pelvic exenteration. Its value in the management of recurrent disease needs to be confirmed in a randomized phase III study.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
R D-V, J U-D, JC and RB participated in data collection and analysis; LC, DC, C L-G, A G-E and A D-G managed the patients; MC, and J de la G critically read and participated in the manuscript; additionally C L-G and A D-G conceived and wrote the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank Elizabeth Robles for her support in the execution of the study.
Figures and Tables
Figure 1 Patients with complete pathological response post-chemotherapy. CT scans of 3 patients showing residual pelvic mass after chemotherapy. Images a,c,e show CT scans pre-treatment, and images b,d,f are the post-treatment control studies. Notably, a patient (images e-f) shows a residual post-chemotherapy pelvic mass measuring 9 × 5 cm.
Figure 2 Overall survival in the intention to treat; four out of 17 are alive for a median survival of 11 months.
Figure 3 Survival in the operated and non-operated patients; median survival was 3 versus 32 months respectively. This difference is highly significant.
Table 1 Characteristics of patients
Number 17
Age 43.3 (29–55)
Histology
Squamous 15
Adenocarcinoma 2
FIGO Stage at diagnosis
1B1 1
1B2 4
IIB 5
IIIB 7
Primary Treatment
RT alone* 9
RT alone + Adj Hyst 4
Chemoradiation** 4
Response to Primary Treatment
Complete response 13
Persistence 3
Progression 1
Months to treatment failure
Recurrence (13 pts) 16 (9–120)
progression (4 pts) 4 (2–7)
* Radiation: 50Gy of external radiation plus brachytherapy to achieve at least 85Gy to point A. ** Six weekly applications of cisplatin at 40 mg/m2 during external radiation.
Table 2 Clinical status of patients at entering the study
Sign/Symptom Number (%)
Pelvic pain 17 100
Fixation to pelvic side wall*
Unilateral 5 29
Bilateral 12 71
Ipsilateral leg edema 6 36
Hydronefrosis 3 18
Leg edema/hydronephrosis 3 18
* As determined by bimanual pelvic examination.
Table 3 Overall treatment
Median number of cycles 4
Exenterated 9
No exenterated 8
Reason for no exenteration
Progression 3
Clinical deterioration 1
Both 4
Table 4 Toxicity to chemotherapy. (expressed by patient).
Toxicity Grades (%)
0 1 2 3 4
Nausea/Vomiting 0 10 6 1 0
Diarrhea 13 2 2 0 0
Neuropathy 15 1 1 0 0
Anemia 6 4 4 3 0
Leukopenia 0 3 9 5 0
Granulocytopenia 0 6 8 3 0
Thrombopenia 12 2 0 3 0
Table 5 Surgical data and pathological response
Exenterated 9
Total infraelevator 8
Anterior supraelevator 1
Pathological Response
Complete 4
Partial 5
≤ 2 cm residual 4
8 cm residual 1
Surgical margins
Negative 8
Positive 1
Ileocolonic conduit 7
Ileal conduit 2
Colostomy 8
Table 6 Surgical morbidity Mean Range
Surgical time 6.3 4.3–8
Bleeding 1860 mL 600–1600
Units transfused 3.4 1–6
Hospital stay 11-7 6–41
Intensive Care Unit stay 1.8 0–12
Complication (events)
Intestinal occlusion 1 (11%)
Massive bleeding 1 (11%)
Urinary fistula 1 (11%)
Perineal infection 2 (22%)
Death 1 (11%)*
The patient died (fistula and infection) at 4 months post-exenteration. This patient had pathological complete response.
==== Refs
Parkin DM Bray F Ferlay J Pisani P Estimating the world cancer burden: Globocan 2000 Int J Cancer 2001 4 153 156 10.1002/ijc.1440
Dueñas-Gonzalez A Cetina L Mariscal I de la Garza J Modern management of locally advanced cervical carcinoma Cancer Treat Rev 2003 29 389 399 12972357 10.1016/S0305-7372(03)00068-9
Lopez MJ Spratt JS Exenterative pelvic surgery J Surg Oncol 1999 72 102 114 10518108 10.1002/(SICI)1096-9098(199910)72:2<102::AID-JSO12>3.0.CO;2-1
Hockel M Knapstein P The combined operative and radiotherapeutic tratment of recurrent tumors infiltrating the pelvic wall Gynecol Oncol 1992 46 20 28 1634136 10.1016/0090-8258(92)90189-P
Leitao MM Chi DS Recurrent cervical cancer Curr Treat Options Oncol 2002 3 105 111 12057073
Friedlander M Guidelines for the treatment of recurrent and metastatic cervical cancer Oncologist 2002 7 342 347 12185296
Brunschwig A Complete excision of pelvic viscera for advanced carcinoma Cancer 1948 1 177 183
Miller B Morris M Rutledge F Mitchell MF Atkinson EN Burke TW Wharton JT Aborted exenterative procedures in recurrent cervical cancer Gynecol Oncol 1993 50 94 99 8349169 10.1006/gyno.1993.1170
del Carmen MG McIntyre JF Goodman A The role of intraoperative radiation therapy (IORT) in the treatment of locally advanced gynecologic malignancies Oncologist 2000 5 18 25 10706646 10.1634/theoncologist.5-1-18
Gemignani ML Alektiar KM Leitao M Mychalczak B Chi D Venkatraman E Barakat RR Curtin JP Radical surgical resection and high-dose intraoperative radiation therapy (HDR-IORT) in patients with recurrent gynecologic cancers Int J Radiat Oncol Biol Phys 2001 50 687 694 11395237 10.1016/S0360-3016(01)01507-3
Hockel M Laterally extended endopelvic resection: Surgical treatment of infrailiac pelvic wall recurrences of gynecology malignancies Am J Obstet Gynecol 1999 180 306 312 9988791
Hockel M Laterally extended endopelvic resection. Novel surgical treatment of locally recurrent cervical carcinoma involving the pelvic side wall Gynecol Oncol 2003 9 369 377 10.1016/S0090-8258(03)00502-X
Moore DH Blessing JA McQuellon RP Haler HT Cella D Benda J Miller DS Olt G King S Boggess JF Rocereto TF Phase III study of cisplatin with or without paclitaxel in stage IVB, recurrent, or persistent squamous cell carcinoma of the cervix: a gynecologic oncology group study J Clin Oncol 2004 22 3113 3119 15284262 10.1200/JCO.2004.04.170
Goldberg JM Piver MS Hempling RE Aiduk C Blumenson L Recio FO Improvements in pelvic exenteration: factors responsible for reducing morbidity and mortality Ann Surg Oncol 1998 5 399 406 9718168
Rutledge FN Smith JP Wharton JT O'Quinn AG Pelvic exenteration: analysis of 296 patients Am J Obstet Gynecol 1977 129 881 892 930972
Symmonds RE Pratt JH Webb MJ Exenterative operations: experience with 198 patients Am J Obstet Gynecol 1975 121 907 918 1115180
Stanhope CR Symmonds RE Palliative exenteration – what, when and why? Am J Obstet Gynecol 1985 152 12 16 2581447
Matthews CM Morris M Burke TW Gershenson DM Wharton JT Rutledge FN Pelvic exenteration in the elderly patient Obstet Gynecol 1992 79 773 777 1565364
Estape R Angioli R Surgical management of advanced and recurrent cervical cancer Semin Surg Oncol 1999 16 236 241 10225302 10.1002/(SICI)1098-2388(199904/05)16:3<236::AID-SSU8>3.0.CO;2-J
Torres Lobaton A Bastida Blanco A Marquez Acosta G Hernandez Aten D Roman Bassaure E Rojo Herrera G Pelvic exenteration for cancer of the uterine cervix (prognostic factors) Ginecol Obstet Mex 1994 62 189 193 8063185
Haas T Buchsbaum HJ Lifshitz S Nonresectable recurrent pelvic neoplasm. Outcome in patients explored for pelvic exenteration Gynecol Oncol 1980 9 177 181 7372190 10.1016/0090-8258(80)90025-6
Halpin TF Frick HC Munnell EW Critical points of failure in the therapy of cancer of the cervix: a reappraisal Am J Obstet Gynecol 1972 114 755 764 4633572
Walsh JW Amendola MA Hall DJ Tisnado J Goplerud DR Recurrent carcinoma of the cervix: CT diagnosis AJR 1981 136 117 1222 6779557
Kim SH Choi BI Lee HP Kang SB Choi YM Han MC Kim CW Uterine cervical carcinoma; comparison of CT and MR findings Radiology 1990 175 45 51 2315503
Kinkel K Ariche M Tardivon AA Spatz A Castaigne D Lhomme C Vanel D Differentiation between recurrent tumor and benign conditions after treatment of gynecologic pelvic carcinoma: value of dynamic contrast-enhanced subtraction MR imaging Radiology 1997 204 55 63 9205223
Popovich MJ Hricak H Sugimura K Stern JL The role of MR imagin in determining surgical elegibility for pelvic exenteration AJR Am J Roentgenol 1993 160 525 31 8430546
Jeong YY Kang HK Chung TW Seo JJ Park JG Uterine cervical carcinoma after therapy: CT and MR imaging findings Radiographics 2003 23 969 981 12853673
Zeisler H Joura EA Moeschl P Maier U Koelbl H Preoperative evaluation of tumor extension in patients with recurrent cervical cancer Acta Obstet Gynecol Scand 1997 76 474 477 9197452
Kohler C Tozzi R Possover M Schneider A Explorative laparoscopy prior to exenterative surgery Gynecol Oncol 2002 86 311 315 12217753 10.1006/gyno.2002.6764
Pannu HK Corl FM Fishman EK Evaluation of cervical cancer: Spectrum of disease Radiographics 2001 21 1155 1168 11553823
Brader KR Morris M Levenback C Levy L Lucas KR Gershenson DM Chemotherapy for cervical carcinoma: Factors determining response and implications for clinical trial design J Clin Oncol 1998 16 1879 1884 9586904
Dueñas-Gonzalez A Gonzalez EA Lopez-Graniel C Reyes M Mota A Munoz D Solorza G Hinojosa LM Guadarrama R Florentino R Mohar A Melendez J Maldonado V Chanona J Robles E de la Garza J A phase II study of gemcitabine and cisplatin combination as induction chemotherapy for untreated locally advanced cervical carcinoma Ann Oncol 2001 12 541 547 11398890 10.1023/A:1011117617514
Duenas-Gonzalez A Lopez-Graniel C Gonzalez A Gomez E Rivera L Mohar A Chanona G Trejo-Becerril C de la Garza J Induction chemotherapy with gemcitabine and oxiplatin for locally advanced cervical carcinoma Am J Clin Oncol 2003 26 22 25 12576919 10.1097/00000421-200302000-00005
Pignata S Silvestro G Ferrari E Selvaggi L Perrone F Maffeo A Frezza P Di Vagno G Casella G Ricchi P Cormio G Gallo C Iodice F Romeo F Fiorentino R Fortuna G Tramontana S Phase II study of cisplatin and vinorelbine as first-line chemotherapy in patients with carcinoma of the uterine cervix J Clin Oncol 1999 17 756 760 10071263
Zanetta G Lissoni A Pellegrino A Sessa C Colombo N Gueli-Alletti D Mangioni C Neoadjuvant chemotherapy with cisplatin, ifosfamide and paclitaxel for locally advanced squamous-cell cervical cancer Ann Oncol 1998 9 997 980
Duenas Gonzalez A Lopez-Graniel C Gonzalez Enciso A Cetina L Rivera L Mariscal I Montalvo G Gomez E de la Garza J Chanona G Mohar A A phase II study of multimodality treatment for locally advanced cervical cancer: neoadjuvant carboplatin and paclitaxel followed by radical hysterectomy and adjuvant cisplatin chemoradiation Ann Oncol 2003 14 1278 84 12881393 10.1093/annonc/mdg333
Sugiyama T Nishida T Kumagai S Fujiyoshi K Okura N Yakushiji M Umesaki N Combination therapy with irinotecan and cisplatin as neoadjuvant chemotherapy in locally advanced cervical cancer Br J Cancer 1999 81 95 98 10487618 10.1038/sj.bjc.6690656
Hogg R Friedlander M Role of systemic chemotherapy in metastatic cervical cancer Expert Rev Anticancer Ther 2003 3 234 240 12722882 10.1586/14737140.3.2.234
Tambaro R Scambia G Di Maio M Pisano C Barletta E Iaffaioli VR Pignata S The role of chemotherapy in locally advanced, metastatic and recurrent cervical cancer Crit Rev Oncol Hematol 2004 52 33 44 15363465
|
16171526
|
PMC1260014
|
CC BY
|
2021-01-04 16:03:05
|
no
|
BMC Cancer. 2005 Sep 19; 5:118
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-118
|
oa_comm
|
==== Front
BMC Clin PatholBMC Clinical Pathology1472-6890BioMed Central London 1472-6890-5-91620738410.1186/1472-6890-5-9Technical AdvanceConventional liquid-based techniques versus Cytyc Thinprep® processing of urinary samples: a qualitative approach Piaton Eric [email protected]ÿnel Jacqueline [email protected] Karine [email protected] Marie-Claude [email protected] Michèle [email protected] INSERM U.407/Université Claude Bernard Lyon 1, Faculté de Médecine Lyon Sud, 69495 Pierre Bénite Cedex, France2 Laboratoire de Cytopathologie, Hôpital Edouard Herriot, Place d'Arsonval, 69437 Lyon Cedex 03, France3 Laboratoire d'Histologie, CHRU de Saint-Etienne, Hôpital Nord, 42055 Saint-Etienne Cedex 2, France2005 6 10 2005 5 9 9 20 4 2005 6 10 2005 Copyright © 2005 Piaton et al; licensee BioMed Central Ltd.2005Piaton et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The aim of our study was to objectively compare Cytyc Thinprep® and other methods of obtaining thin layer cytologic preparations (cytocentrifugation, direct smearing and Millipore® filtration) in urine cytopathology.
Methods
Thinprep slides were compared to direct smears in 79 cases. Cytocentrifugation carried out with the Thermo Shandon Cytospin® 4 was compared to Thinprep in 106 cases, and comparison with Millipore filtration followed by blotting was obtained in 22 cases. Quality was assessed by scoring cellularity, fixation, red blood cells, leukocytes and nuclear abnormalities.
Results
The data show that 1) smearing allows good overall results to be obtained, 2) Cytocentrifugation with reusable TPX® chambers should be avoided, 3) Cytocentrifugation using disposable chambers (Cytofunnels® or Megafunnel® chambers) gives excellent results equalling or surpassing Thinprep and 4) Millipore filtration should be avoided, owing to its poor global quality. Despite differences in quality, the techniques studied have no impact on the diagnostic accuracy as evaluated by the rate of abnormalities.
Conclusion
We conclude that conventional methods such as cytocentrifugation remain the most appropriate ones for current treatment of urinary samples. Cytyc Thinprep processing, owing to its cost, could be used essentially for cytology-based molecular studies.
==== Body
Background
More than 50,000 new cases of urothelial carcinoma, which represents 90% of bladder cancer cases are diagnosed annually in Europe and in North America [1]. About 70% of bladder urothelial carcinomas are superficial (TNM stage pTa-1) and may be viewed, diagnosed and treated by cystoscopy aided by biopsies and transurethral resection [2].
Despite it is recognized as the biological standard for the diagnosis and follow up of bladder tumors, urinary cytology has a mean sensitivity of about 50% and it is hampered by a large amount of non-diagnostic samples [3]. Although urinary cytology detects about 80% of aggressive, high grade (G3) urothelial tumors, some results remain falsely negative, particularly in patients having had TUR or bacillus Calmette-Guérin immunotherapy. In urology practice, cystoscopy is commonly combined with urinary cytology, particularly in the search for high grade wherever its location in the urinary tract.
Liquid-based cytology (LBC) has been developed as a replacement to cytocentrifugation and/or smearing, owing to cell recovery capabilities and better cell preservation. Some LBC methods use a filtration process and a computer-assisted thin-layer deposition of cells (Cytyc Thinprep® supplied by Cytyc Corp., Boxborough, MA), whereas others are based on a sedimentation process (AutoCyte® PREP supplied by TRiPath Imaging, Burlington, NC). In the urine, the use of Cytyc Thinprep 2000 results in increased cellularity and marked reduction of debris, red blood cells (RBC) and crystals [4-7].
However, optimization of cell capture and fixation as well as thin-layer deposition of cells can be achieved by other methods than LBC, particularly while using modern cytocentrifugation methods [7]. In our experience based on 2500 specimens/year for 15 years, and provided specific requirements are followed, direct smears and cytocentrifugation with the Shandon Cytospin® 4 (Thermo Electron Corp., Waltham, MA) produce highly satisfying cytological specimens.
Accordingly, the aim of our study was 1) to objectively analyze the quality of urine samples processed by a body of conventional thin-layer methods as compared with Cytyc Thinprep LBC and 2) to verify if differences noted have an impact on diagnostic accuracy.
Methods
The study population was composed of 224 urine samples taken in patients with symptoms suggesting bladder cancer (gross hematuria, micturition disorders, chronic urinary infection) in 89 cases (39.7%), or followed after transurethral resection for bladder urothelial carcinoma in 135 cases (60.3%).
Urinary samples were taken after cystoscopy in 157 cases (63.8%), and after simple micturition in other cases. All samples were immediately fixed with 50% ethanol (V/V) or with a 20% Polyethyleneglycol 1500 (Merck, Darmstadt, Germany) solution in 50% ethanol (1/3 fixative and 2/3 urine).
Urine samples were sent to the laboratory and separated into two aliquots after homogeneization. One of the aliquots was processed according to the Thinprep LBC recommendations, and the other was processed according to a smear method, by cytocentrifugation or by filtration.
Cytyc Thinprep* processing
The Thinprep 2000 automaton allows thin-layer cell preparations to be provided thanks to a filtration process: after the TransCyt® filter has been plunged into the sample, it rotates at a high speed and facilitates cell and mucus dispersion. A vacuum is then applied to the filter, which collects cells on a 5 μm porosity membrane. A software program allows a homogeneous deposition of cells until saturation. The TransCyt filter is then inverted and a positive pressure allows cells to adhere to an electronegative slide. After insertion of another TransCyt filter and of another slide, the whole procedure may be repeated until the entire sample has been treated.
The urine samples studied were processed according to instructions for non mucoid fluids: samples were mixed with a Cytolyt® solution containing methanol, mucolytic and hemolytic agents and were then centrifuged at 600 G for 10 minutes. After discarding the supernatant, the cell pellet was mixed with a PreservCyt® solution and treated by the Thinprep 2000 processor. Thinprep slides were used in all cases.
Smearing on coated slides
Comparison of LBC with smears was made in 79 cases. After centrifugation at 600 G for 10 minutes and careful removing of the supernatant, the cell pellet was aspirated and smeared on a thin coating layer (Glycerin/Albumin according to Mallory, Bayer Diagnostics, Puteaux, France) previously deposited on two Superfrost® Plus slides (Menzel-Gläser, Braunschweig, Germany). Slides were immediately fixed with a Cell-Fixx® (Thermo Electron Corp., Waltham, MA) spray and allowed to dessicate at room temperature (RT) for at least 1 hour before Papanicolaou staining.
Cytocentrifugation methods
Comparison of LBC with cytocentrifugation was made using the Thermo Shandon Cytospin® 4 in 106 cases. After centrifugation at 600 G for 10 minutes, hypocellular urine samples (< 20 μl cell pellets) were cytocentrifuged with sample chambers up to 0.5 ml. Conversely, urine samples with a large pellet were treated with large volume sample chambers.
The Cytospin system uses centrifugation and fluid absorption principles and allows deposition of a thin layer of cells on round or rectangular areas. The deposition process needs that sample chambers are placed and locked into stainless steel Cytoclip® assembly devices. In order to test various types and qualities of sample chambers we used:
1) three years' old round reusable, autoclavable chambers designed for samples up to 0.5 ml (TPX® chambers with a cell deposition area of 6 mm diameter, allowing 28 mm2 to be screened) in 44 cases,
2) round disposable chambers designed for samples up to 0.5 ml (single Cytofunnel® with a cell deposition area of 6 mm diameter, allowing 28 mm2 to be screened) in 31 cases,
3) large volume disposable chambers designed for samples up to 6 ml (Megafunnel® chambers with a cell deposition area of 21 × 24 mm, allowing 294 mm2 to be screened) in 31 cases.
Two slides of 28 mm2 screening area (for 1 ml of urine), and one slide of 294 mm2 screening area (for 6 ml of urine) were prepared for each specimen studied.
Specially marked coated Cytoslides® provided by Thermo Shandon were used. Although not necessary, slides processed with TPX sample chambers had an additional treatment with a drop of glycerine/albumin deposited on the sample area.
Millipore filtration methods
LBC was compared with Millipore filtration followed by blotting of cells on various slides in 39 cases, in order to test the adhesiveness to various types of commercially available coated slides. Urine was filtered through Magna® MCE nitrocellulose membrane filters, pore size 5 μm, diameter 25 mm placed in a Swinnex® device attached to a 60 ml Luer-Lock® syringe (Bioblock Scientific, Illkirch, France).
After complete filtration and removal of the membrane filter, the blotting was first performed on Polysine® slides (Menzel-Gläser, Braunschweig, Germany) in 8 cases, but the adhesiveness obtained was too impaired for allowing continuation of the assays. We then used Cytyc Thinprep slides in 9 cases, but finally we chose Superfrost® Plus slides and Snowcoat X-tra® slides (Surgipath Europe Ltd, Peterborough, England) equally for the 22 remaining cases.
Using these procedures, the resulting cell deposition area is 25 mm diameter, allowing about 491 mm2 to be screened.
Smears were stained with a hypochromic Papanicolaou stain [8] before analysis.
Analysis of morphologic criteria
A single pathologist (EP) compared conventional and LBC slides using an Olympus BHS microscope. Slides were placed side by side and were analyzed under Plan × 10, Plan × 40 and Oil PlanApo x63 objectives. The global quality of slides was assessed by scoring cellularity, cell fixation, number of RBC, leukocytes and degenerative changes of urothelial cells. The presence of cell groups and clusters was also measured. Special attention was paid to altered cellular features potentially indicating malignant transformation – increased N/C ratio, nuclear hyperchromatism, irregular nuclear shape, prominent nucleoli and mitoses – as previously described [9].
All cellular features were coded from 0 to xxx according to their degree of abnormality.
Urothelial cells were recognized as malignant, high-grade, when they showed increased N/C ratio, nuclear hyperchromatism and markedly irregular nuclear borders or prominent nucleoli. They were recognized as neoplastic, low-grade, when they formed papillary fronds demonstrating increased N/C ratio and slightly irregular nuclear shape, or where numerous elongated cells with slight nuclear abnormalities could be evidenced, as described in the literature [9,10].
Cytological results were categorized as positive or negative for urothelial tumor cells, whatever their grade. Normal, inflammatory, reactive and degenerative conditions of urothelial cells were considered as negative, as well as urothelial atypias of undetermined significance.
Numerical data were analyzed using paired series Chi-square test or Fisher's exact test, when appropriate, and a probability level of 0.05 was regarded as significant.
Results
Using the scoring system as described in the Materials and Methods section, and considering a 0–3 scale, mean and standard deviations as well as the statistical significance of each parameter are shown in Figure 1,2, 3, 4, 5.
Figure 1 Comparison of smears versus Thinprep slides (mean values, standard deviations and statistical significance).
Figure 2 Comparison of cytocentrifugation using reusable TPX chambers versus Thinprep slides (mean values, standard deviations and statistical significance).
Figure 3 Comparison of cytocentrifugation using disposable Cytofunnels (for samples up to 0.5 ml) versus Thinprep slides (mean values, standard deviations and statistical significance).
Figure 4 Comparison of cytocentrifugation using disposable Megafunnels (for samples up to 6 ml) versus Thinprep slides (mean values, standard deviations and statistical significance).
Figure 5 Comparison of Millipore filtration followed by blotting of cells on coated slides versus Thinprep slides (mean values, standard deviations and statistical significance).
Differences noted concern global quality (cellularity and fixation combined) on the one hand, number of RBC and leukocytes on the other hand. Surprisingly, we found that smears allowed obtaining a global quality superimposable to that of Cytyc Thinprep slides (Figure 6). More precisely, the cellularity scores obtained by smears and LBC were 1.97 ± 0.86 versus 1.96 ± 0.68, respectively (p = ns), whereas values for fixation were 2.58 ± 0.48 versus 2.50 ± 0.59 (p = ns).
Figure 6 Group of low grade Urothelial tumour cells obtained after centrifugation and smearing. Papanicolaou stain, x 400.
Cytocentrifugation with 3 years' old reusable sample chambers resulted in significant decrease in both cellularity and fixation quality, whereas cytocentrifugation with disposable sample chambers (whatever the type of chamber used) allowed obtaining the better results (Figure 7).
Figure 7 Group of low grade urothelial tumour cells together with normal superficial cells obtained after liquid based (Thinprep) processing. Papanicolaou stain, x 630.
Millipore filtration resulted in impaired cell preservation even after blotting of cells on coated slides and careful fixation.
The concentration of RBC was significantly decreased after LBC treatment of samples in all circumstances except in Millipore* filtration using 5 μm porosity membranes. Similar comments may be done about leukocytes.
Whatever the technique studied, the search for cell groups and atypias gave results identical than those of LBC except for smears which showed a slightly higher percentage (p = 0.01). However, the values obtained were not strikingly different.
Discussion
As far back as the late 'seventies, authors have attempted to compare cytocentrifugation with other methods such as filtration [11,12]. In those preliminary studies, Millipore filtration was found to give better cell recovery and better morphologic details than cytocentrifugation. However the methods used (reusable sample chambers) was suboptimal: a significant cell loss can be attributed to the roughness of sample chamber walls secondary to repeated cleaning [13].
Waiting the 'nineties was necessary for obtaining comparisons between the Cytyc Thinprep LBC and other methods, with some contradictory results. Many of the studies, published as abstracts of the 40th and 41st Annual Scientific Meetings of the International Academy of Cytology, were not transformed into full length articles [4,5,14,15].
Except for one study which showed processing time and cost several times greater for Cytyc Thinprep LBC than for polycarbonate membrane filtration [15], most series recognize advantages in using LBC. In a recent study comparing cytocentrifugation to Cytyc Thinprep, Cytospin preparations were found superior to LBC in terms of cytomorphologic details and preservation of architectural patterns [16]. However the advantage of LBC concerning cleaner background was noted.
Cytocentrifugation and LBC are not the only available methods for improving diagnostic accuracy: potentially interesting results were previously shown by Albright and Frost [17]. Using a simple density gradient to separate atypical cells from normal cells after fixation with the Saccomanno method, the authors were able to enrich up to 20-fold the atypical and cancer cell fraction. To our knowledge however, these results have not been resumed at a later date.
A more recent study assessed the quality and cost of AutoCyte PREP versus cytocentrifugation of urine specimens in a general laboratory setting [18]. It was shown that the Cytospin method, despite longer preparation time, had 1) shorter screening time, 2) higher number of diagnostic cells, 3) better fixation and staining quality than the AutoCyte PREP. Additionally, the Cytospin method was found 7 times less expensive than the AutoCyte* PREP method.
Concerning conventional methods, the values obtained in our series show that despite differences in quality, the techniques studied have no impact on the diagnostic accuracy as evaluated by the rate of abnormalities (nuclear features and cell groups). About each technique studied, the following comments may be done:
1. Smearing allows obtaining good overall results for the lowest cost. However the longer screening time renders the method suboptimal. Additionally the glycerine/albumin coating used renders slides useless for immunocytochemistry or other molecular studies,
2. Cytocentrifugation with reusable chambers should be avoided if annual renewal cannot be guaranteed,
3. Millipore filtration followed by blotting of cells on coated slides should be avoided, owing to poor global quality and high cost,
4. Cytocentrifugation using disposable chambers (Cytofunnels or Megafunnel chambers) gives excellent results equalling or surpassing LBC if one considers cellularity, fixation and the comfort for screening.
Concerning cost-efficacy comparisons, it has been shown that the monthly cost of the two most efficient methods (Cytocentrifugation with disposable chambers and Cytyc Thinprep LBC) is strikingly different: there is a 92.8% to 154.5% increased cost for LBC versus cytocentrifugation with disposable Megafunnels and Cytofunnels, respectively [19].
However in our opinion, one must consider not only the diagnostic performance and cost, but also the ultimate goal of technical improvements provided by LBC. LBC aims primarily to provide reproducible and well preserved material for additional techniques such as immunocytochemistry, fluorescence in situ hybridization (FISH) and other types of molecular analyses. It has been shown that Thinprep-processed samples allowed efficient recovery of the DNA, RNA and proteins related to the p53 tumor suppressor gene [20].
Conclusion
We conclude that Cytyc Thinprep LBC, despite its cost, may still be considered as a technical progress for cytology-based molecular studies. To an economical point of view and taking into account the value of a meticulous technique, cytocentrifugation with disposable chambers remains the technical standard for current treatment of urinary samples.
List of abbreviations
LBC: liquid-based cytology technique
RBC: red blood cells
RT: room temperature
TUR: transurethral resection
Competing interests
The author(s) declare that they have no competing interest.
Authors' contributions
EP planned the study and prepared the manuscript.
JF, KH and MCR performed the liquid-based techniques (cytocentrifugations and Thinprep processing) as well as smears and Millipore filtrations.
EP and MC performed the cytopathologic evaluations and the statistical analysis All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Sözen S Eskicorabci S Özen H Urinary markers for urothelial cancer BJU Int 2003 92 531 533 14511027 10.1046/j.1464-410X.2003.04427.x
Saad A Hanbury DC McNicholas TA Boustead GB Woodman AC The early detection and diagnosis of bladder cancer: a critical review of the options Eur Urol 2001 39 619 633 11464050 10.1159/000052519
Bastacky S Ibrahim S Wilczinski SP Murphy WM The accuracy of urinary cytology in daily practice Cancer 1999 87 118 128 10385442 10.1002/(SICI)1097-0142(19990625)87:3<118::AID-CNCR4>3.0.CO;2-N
Pondo A Cobbs BC Gupta PK Use of Thinprep in urine cytology (Abstract) Acta Cytol 1992 37 584
Papillo JL Lapen D Cell yield. Thinprep vs. cytocentrifuge Acta Cytol 1992 38 33 36 8291353
Anagnostopoulou I Spathi H Rammou-Kinnia R Karakitsos P Gianni I Georgoulakis J Kittas C Comparative study of Thinprep and conventional voided urine cytology (Abstract) Cytopathology 2000 11 373
Wright RG Halford JA Evaluation of thin-layer methods in urine cytology Cytopathology 2001 12 306 313 11722510 10.1046/j.1365-2303.2001.00341.x
Mouriquand J Differential nucleolar staining affinity with a modified Papanicolaou procedure Stain Technol 1981 56 215 219 6171053
Murphy WM Current status of urinary cytology in the evaluation of bladder neoplasms Hum Pathol 1990 21 886 896 2203672 10.1016/0046-8177(90)90171-Z
Layfield LJ Elsheikh TM Fili A Nayar R Shidham V Review of the state of the art and recommendations of the Papanicolaou Society of Cytopathology for urinary cytology procedures and reporting Diagn Cytopathol 2004 30 24 30 14696141 10.1002/dc.10401
Barrett DL King EB Comparison of cellular recovery rates and morphologic detail obtained using membrane filter and cytocentrifuge techniques Acta Cytol 1976 20 174 180 1065177
Marwah S Devlin D Dekker A A comparative cytologic study of 100 urine specimens processed by the slide centrifuge and membrane filter technique Acta Cytol 1978 22 431 434 281854
Piaton E Faÿnel J Ranchin MC Hutin K Comparative study of conventional techniques and liquid-based processing of urine samples for the evaluation of bladder neoplasms (Abstract) Acta Cytol 2002 46 207
Beech DP Allbee A Atanasoff PE Brahm CL Moore TL Bell DA A comparison of voided urine samples processed by the Cytyc Thinprep processor and the Shandon Cytospin II (Abstract) Acta Cytol 1992 36 583
Goellner JR Plagge AM St Germain KM Johnson DA McClain SL Moltaji H Cytyc Thinprep processor vs. polycarbonate membrane filtration: a comparative study in urine cytology (Abstract) Acta Cytol 1993 37 800
Nassar H Ali-Fehmi R Madan S Use of Thinprep monolayer technique and cytospin preparation in urine cytology: a comparative analysis Diagn Cytopathol 2003 28 115 118 12619090 10.1002/dc.10245
Albright CD Frost JK Centrifugal separation of carcinoma or atypical cells in voided urine Virchows Arch B Cell Pathol 1992 62 45 53
Zardawi IM Duncan J Evaluation of a centrifuge method and thin-layer preparation in urine cytology Acta Cytol 2003 47 1038 1042 14674075
Piaton E Hutin K Faÿnel J Ranchin MC Cottier M Cost-efficiency analysis of modern cytocentrifugation methods versus liquid-based (Cytyc Thinprep) treatment of urinary samples J Clin Pathol 2004 57 1208 1212 15509686 10.1136/jcp.2004.018648
Tisserand P Fouquet C Marck V Mallard C Fabre M Vielh P Soussi T Thinprep-processed fine-needle samples of breast are effective material for RNA- and DNA-based molecular diagnosis. Application to p53 mutation analysis Cancer (Cancer Cytopathol) 2003 99 223 232
|
16207384
|
PMC1260015
|
CC BY
|
2021-01-04 16:27:55
|
no
|
BMC Clin Pathol. 2005 Oct 6; 5:9
|
utf-8
|
BMC Clin Pathol
| 2,005 |
10.1186/1472-6890-5-9
|
oa_comm
|
==== Front
BMC Dev BiolBMC Developmental Biology1471-213XBioMed Central London 1471-213X-5-221620738110.1186/1471-213X-5-22Research ArticleComparison of the gene expression profile of undifferentiated human embryonic stem cell lines and differentiating embryoid bodies Bhattacharya Bhaskar [email protected] Jingli [email protected] Youngquan [email protected] Takumi [email protected] Josef [email protected] Sandii N [email protected] Xianmin [email protected] Thomas C [email protected] Mahendra S [email protected] Raj K [email protected] Laboratory of Molecular Tumor Biology, Division of Cellular and Gene Therapies, Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, MD 20892, USA2 Laboratory of Neuroscience, National Institute on Aging, Baltimore, Maryland 21224, USA3 Bresagen Inc., Athens, GA, USA4 National Institute of Drug Abuse, National Institutes of Health, Bethesda, MD 20892, USA2005 5 10 2005 5 22 22 28 4 2005 5 10 2005 Copyright © 2005 Bhattacharya et al; licensee BioMed Central Ltd.2005Bhattacharya et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The identification of molecular pathways of differentiation of embryonic stem cells (hESC) is critical for the development of stem cell based medical therapies. In order to identify biomarkers and potential regulators of the process of differentiation, a high quality microarray containing 16,659 seventy base pair oligonucleotides was used to compare gene expression profiles of undifferentiated hESC lines and differentiating embryoid bodies.
Results
Previously identified "stemness" genes in undifferentiated hESC lines showed down modulation in differentiated cells while expression of several genes was induced as cells differentiated. In addition, a subset of 194 genes showed overexpression of greater than ≥ 3 folds in human embryoid bodies (hEB). These included 37 novel and 157 known genes. Gene expression was validated by a variety of techniques including another large scale array, reverse transcription polymerase chain reaction, focused cDNA microarrays, massively parallel signature sequencing (MPSS) analysis and immunocytochemisty. Several novel hEB specific expressed sequence tags (ESTs) were mapped to the human genome database and their expression profile characterized. A hierarchical clustering analysis clearly depicted a distinct difference in gene expression profile among undifferentiated and differentiated hESC and confirmed that microarray analysis could readily distinguish them.
Conclusion
These results present a detailed characterization of a unique set of genes, which can be used to assess the hESC differentiation.
==== Body
Background
Embryonic stem cells (hESC) have been isolated from multiple species [1-4] including non-human primates [2] and humans [3,4]. Currently, over hundred different Human embryonic stem cell (hESC) lines have been established [3-7]. hESC populations grow as tightly compacted colonies of undifferentiated cells on mouse [3,4] or human [6] feeders or as colonies in feeder-free conditions using matrix and conditioned medium [8]. hESC has been shown to differentiate in vitro and in vivo to form derivates of all three germ layers. In vitro differentiation can be induced by the process of embryoid body (hEB) formation, which involves aggregating the cells and preventing separation by plating on a non-permissive substrate. Cell to cell interaction and addition of differentiation agents such as retinoic acid (RA) induces differentiation into derivatives of all three germ layers (mesoderm, ectoderm and endoderm) [7,9-12].
hEB can subsequently be induced to undergo further differentiation to generate a variety of cell types, including hematopoietic [13], neuronal [14,15], myogenic [16] and cardiac muscle cells [17,18]. Thus, hEB represent an early stage in the process of lineage specification and should differ from hESC or their more differentiated progeny in their profile of gene expression.
Several different methods have been developed that can be used to assess the process of differentiation. Subtractive hybridization [19,20], differential display polymerase chain reaction (DD-PCR) [21], representational difference analysis (RDA) [22], analysis of expressed sequence tag (EST) [23] and serial analyses of gene expression (SAGE) [24] are but a few commonly used techniques. Perhaps the most commonly used however, is gene array (microarray) [25-27]. Microarrays have been used by several investigators to assess the undifferentiated hESC state [28,29] and provide a data set of useful information. We for example utilized a large-scale oligonucleotide based array to identify a set of 92 genes that are highly upregulated in six hESC lines when compared against human universal reference RNA derived from mature tissues [28]. This set of "stemness genes" along with additional novel genes identified has served to assess the state of undifferentiated cells. However, currently no similar data set is available for genes that may be used to define the embryoid body stage of hESC differentiation and no comparisons between the undifferentiated and differentiated populations have been performed utilizing microarray technology. A recent study has characterized gene expression in embryoid bodies by massively parallel signature sequencing (MPSS) and suggested that several candidate genes specific to hEB may exist [30]. MPSS analysis however, is expensive and unavailable for most laboratories. Therefore, an alternate readily available and economical assay is needed to characterize embryoid bodies and compare them with available datasets on hESC. Microarray studies of hEB's offer the possibility of such an assay that can be used for routine assessment of the state of ES cell differentiation.
To determine if microarrays (or gene arrays) can be used to distinguish between hESC and hEB and to identify candidate markers of the process of differentiation we have compared the gene expression pattern of undifferentiated hESC and differentiated hEB derived from them using a large-scale oligonucleotide based arrays. The expression of selected genes was confirmed by another large-scale array, reverse transcriptase-polymerase chain reaction (RT-PCR), comparison with an expressed sequence tag (EST) enumeration database of ESC [23], MPSS data from embryoid bodies [30] and immunocytochemistry. Our results show that microarray studies can readily distinguish between hESC and hEB and can be used to identify markers of the embryoid body stage.
Results
Microarray detects differences between hESC and hEB derived from them
To assess alteration in gene expression in hESC and hEBs, cells were cultured and induced to form EBs (Fig. 1). Total RNA was harvested from undifferentiated hESCs and hEBs derived from them and a time course of gene expression was performed to assess downregulation of known ES cell specific genes. Day 13 of differentiation was chosen as a time point for subsequent analysis where there is a clear downregulation of known hESC markers and an upregulation of some early markers of differentiation (Fig. 2). The expression of known undifferentiated hESC markers including Oct4, Nanog and Esg1 showed reduction in expression in hEB at day 13 while known markers of differentiation (SOX1, Nestin, and GATA4) showed a marked increase in expression. Several genes (Sox2, TERT, BCRP, Cx43, and Rex-1) did not show detectable change between hESC and hEB cells. For microarray studies each sample was compared to human universal reference RNA (mixture of total RNAs from a collection of adult human cell lines, chosen to represent a broad range of expressed genes and both male and female donors are represented) to maintain uniformity and allow comparisons across samples. cDNA from BG02 and pooled samples of hESC were labeled with Cy5 and huURNA with Cy3, and ~17,000 oligonucleotide arrays were hybridized. Similarly, cDNA from hEB derived from differentiated BG02-ESC (day 13 BG02 EB, day 21 BG02 EB) and pooled WiCell EB (day 13) was labeled with Cy5 and huURNA with Cy3 and arrays were hybridized and data analyzed. Since the process of differentiation is relatively stochastic and cell lines may behave differently as they differentiate, data from different experiments (with different samples) was not pooled and reported as the expression from single hybridization. Each sample was analyzed in duplicate obtained from two different independent cultures.
Figure 1 Phase differentiation of ES and EB. Phase pictures of a human ES cell clone grown under feeder-free conditions (A) and a human EB of day 14 (B).
Figure 2 RT-PCR of some marker genes in pooled samples of ES and EB RNA. RT-PCR analysis of marker genes in Pooled ES and pooled EB cell lines. Total RNA derived from both the ES and EB cell lines (WiCell lines) were subjected to RT-PCR analysis as previously described (34). G3PDH mRNA amplified from these samples served as an internal control. The primers used are listed in supplementary table (Table 7S). The thermocycler conditions used for amplification were 94°C, 4 minutes hot start,94°C, 30 sec; 60°C, 30 sec and 72°C, 1 minute. Ten microliter amplification products were resolved in 2% agarose gel, stained with ethidium bromide (EtBr), and visualized in a transilluminator and photographed.
Overall microarray results between technical replicates were similar and representative images from each experiment are available (see Additional file-1, 2, 3, 4, 5). Microarray analysis showed that 11,000 of the ~17,000 features present on the array were detectable above background at a ≥ 150 minimum intensity and target pixels of one standard deviation (SD) above background (≥30) cutoffs. hEB and hESC expressed approximately 2400 to 3000 genes. Pooled WiCell ESC and BG02 ESC showed over expression of 2471 and 2843 genes at ≥ 2 fold respectively, compared to huURNA (Supplementary Table-1Sa and 1Sb, see additional file-6 and 7). As these cells differentiated to hEB, the number of total genes that were detected remained similar to hESC (supplementary Table 2S, see additional file-8).
For some experiments ESC samples were pooled as we were interested in identifying differences between ESC and EBs that would be common to multiple lines. Further, as EB formation is variable we felt pooling may allow us to focus on large differences which would not be lost in the averaging process. Once we obtained results we then tested if this was true by using cell line provided by a different provider and propagated under different culture condition. In addition, since the purpose of day 21 differentiation was to study which genes persisted and which were modulated as a result of further differentiation we did not use day 21 pooled EBs but focused on results from a single line
We have previously shown that six hESC lines (BG02, BG01, GE01, GE09, TE06 and PES cell lines from GE01, GE07 and GE09) express 92 genes in common at ≥ 3 fold levels when compared with huURNA (28). We therefore examined the expression of these genes and in addition all genes overexpressed in BG02 and pooled hESCs (Table 1/see additional file-9) (supplementary Table 1Sa and 1Sb, see additional file 6 and 7). Eighty-seven of 92 genes were also detected in the present sample confirming the quality of the array hybridization and the fidelity of the samples used. In addition, previous study identified several early markers of differentiation, that were present at low levels in hESC and when hESC cells were differentiated, these genes were upregulated and up regulation was confirmed by EST enumeration technique [28]. The present study confirmed these earlier observations by microarray experiments (Table2/see additional file 9). These differentiation genes included Keratin 8, Keratin 18, ACTC, and TUBB5. When we examined our current results, we found that these genes were also up regulated in hEB cells derived from BG02 and WiCell lines by microarray studies (Table 2/ see additional file 9). These results confirm the EST enumeration data and further provide support for the quality of hEBs used.
Overall the technical replicates, the testing of the expression of known genes and the ability to detect expected changes in the current samples confirmed the suitability of the current data set for additional analysis.
Modulation of "stemness genes by ES cell differentiation
Previously, we identified a set of 92 genes are expressed at high levels in most hESC and can be detected by microarray and EST enumeration [28]. To test if changes in their expression could be used to monitor differentiation, we examined the relative levels of these genes as ES cell differentiated. Eighty-seven of the 92 genes expressed in all six hESC lines were also over expressed in day 13 BGO2-EB and day 13 pooled WiCell EB compared to HuURNA (data not shown). However, as a result of differentiation of the BG02-ESC line 77 out of 92 genes were down modulated in BG02 EB cells (Table 1/ see additional file 9). Among these, known ES cell markers e.g., POU5F1, GTCM-1, LEFTB, Galanin, GJA1, TDGF1, SFRP2, FABP5 and Lin-28 showed a marked decline in their expression compared to undifferentiated ES cells. Similarly, other ES cell specific markers such as CER1 and DNMT3B also showed marked decline in the expression in day 13 BG02 EB compared to ESC (Table1/ see additional file 9). However, three zinc finger proteins did not show any significant change and down modulation of Numatrin, C20orf129 and Laminin receptor was modest (Table1/ see additional file 9).
Twelve novel genes that were overexpressed in all 6 hESC lines tested [28], were all down modulated by differentiation. Among them MGC27165 (IFITM1), GSH1, PPAT, KIAA1573, TD-60, C20orf168, ARL8 showed a marked decline in fold expression in BG02 EB compared to undifferentiated BG02 (Table 1/ see additional file 9) while changes in other genes was modest.
Pooled EB samples of ES cells (WA01, WA07 and WA09) showed a similar gene expression profile as observed in day 13 BG02-EB. 53 out of 92 "stemness" genes were down modulated compared to undifferentiated WiCell ESC (Table 1/ see additional file 9). These 53 genes were a subset of the 77 that were markedly downmodulated in BG02 derived EB compared to undifferentiated BG02-ESC (see above and Table 1/ see additional file 9). POU5F1, LEFTB, FABP5, Galanin, GJA1, SFRP2, Nanog and TDGF1 ES specific genes showed ≥ 3 times lower expression in pooled WiCell EB compared to pooled WiCell ESC (Table 1/ see additional file 9). In addition, other ES cell specific markers like DNMT3B, CER1, and SOX2 also showed down modulation in pooled WiCell EB. Novel genes such as PPAT, MGC27165, GSH1, ARL8, and KIAA1573 showed marked down modulation in pooled EB compared to pooled ES cells. However, zinc finger protein Znf257, laminin receptor, C20orf1 and C20orf129 showed modest decline (Table 1/ see additional file 9).
Both pooled WiCell EB and day 13 BG02-EB showed an upregulation of genes that we had previously identified as being markers of differentiation that were present in detectable levels in undifferentiated hESC lines (Table 2/ see additional file 9). These included early differentiation markers e.g., KRT8, KRT18, TUBB5 and ACTC. Overall, the pattern of differentiation appeared similar with BG02 when compared to embryoid bodies derived from WiCell ESC.
Among 92 genes, thirteen additional novel genes have been identified previously in hESC [28]. However, only 3 of the thirteen genes were present on this array and their expression was analyzed. Our results show that Dppa4 was expressed in BG02 ES and pooled WiCell ES, however, its expression level was eight fold higher in day 13 BG02 EB compared to huURNA and down modulated to 2 fold at day 21 BG02-EB. Similarly, pooled WiCell hESC cells showed 10 fold over expression of Dppa4 compared to HuURNA, however upon differentiation to EB the expression was decreased. In sharp contrast, claudin-6 showed down modulation of expression from pooled WiCell ESC to EB by about 10 fold but was not present in detectable levels in either BG02 ES or BG02 EBs.
In PEB some genes that showed down modulation in BG02 EBs were in fact up regulated compared to ESC. The reason for this reverse pattern is not known. Although we did not check whether these were also up regulated in day 21 pooled EBs, our future studies are focused on addressing these issues. It is possible that a mixture of three hESC lines behaved differently when they were mixed and differentiated.
Overall, these data confirm that seventy-seven of the ninety-two genes we had identified as hESC specific could be used to assess the differentiation state of ES cells as they differentiate. A large number of these genes show a dramatic down modulation as early as Day 13 and this is seen in both WiCell and Bresagen (BG02) cell lines and can be readily assessed by microarray analysis.
Genes that are upregulated as ES cells differentiated to form embryoid bodies
To analyze genes that were upregulated as hESC differentiated, embryoid bodies were prepared as previously described and their quality analyzed as described above. Samples, which showed a clear downmodulation of ES cell specific genes, were subjected to microarray studies and their overall gene expression pattern compared with that of the undifferentiated population. Each individual experiment was analyzed separately and differences in gene expression observed were then compared. When BG02 hESC were compared with embryoid bodies derived from them a total of 333 new genes were expressed at ≥ 3 fold higher levels in day 13 BG02-EB compared to undifferentiated BG02 ESC (Supplementary Table 3S, see additional file-8). These genes included many genes known to be upregulated in EB cells and confirmed the quality of the array and the hybridization (see supplementary data). In addition, we identified numerous additional genes whose expression in hEBs had not been documented previously.
Since we were interested in identifying a core set of genes that are upregulated as ES cells differentiated we compared the pattern of gene upregulation in a second distinct population of ES cells (WiCell lines) grown under different conditions using protocols described previously [7,31,32]. Out of 333 genes that were upregulated in day 13 BG02 EB 194 genes were also upregulated at ≥ 3 fold in WiCell derived EB (pooled EB) (Table 3/ see additional file 9). The remaining 139 genes were also upregulated but they did not meet the rigorous three fold cutoff criteria. The expression of these 139 genes by other techniques will be subject of confirmation in future studies.
The common subset of genes, which showed upregulation at ≥ 3 fold in both BG02 EB and pooled WiCell EBs were classified by hierarchical clustering. As shown in Fig 3, all listed genes (a snap shot of 194 genes) were upregulated in EBs derived either from BG02 or WiCell lines and the expression levels can be readily distinguished from those in undifferentiated ESC. These results clearly show a marked difference in gene expression profile in differentiated compared to undifferentiated ES.
Figure 3 Hierarchical clustering of a set of genes over expressed in EB but not in ES. Hierarchical clustering of genes overexpressed in EB but not in ESC. All 194 genes were clustered; however, a snap shot of some genes is shown due to space limitation. A set of genes showing overexpression at ≥ 3 fold cluster together in BG02-EB and pooled EB but differ from BG02-ESC and pooled ESC. Color indicates the relative expression levels of each gene, with red indicating higher expression, green indicating negative expression and black representing absent expression. The 10 genes as indicated by the arrows can be considered as marker for EB as they are either negatively expressed or absent in ESC in most cases but overexpressed in both the EBs.
More detailed examination of the genes that were differentially expressed in ESC and EBs indicated multiple signaling pathways are altered. BG02 EB showed over expression of Keratin 19, Profilin-1, Fibronectin 1, HAND1, COL1A2, ZAK, COL4A2, BIRC7, NID2, TUBB5, TMSB4X, PLP2, ENO1 and COL5A2 (Supplementary table 4S, see additional file-8). Pooled WiCell EB also showed upregulation of similar genes including HAND1, KRT19, Fibronectin 1, Profilin1, TMSB4X, Vimentin, Enolase, PLP2, COL4A2, CAPN1, NID2, IVL, ZAK, SPTA1 and COL1A2, which are related to cell differentiation or cytoskeleton (Supplementary Table 5S/see additional file 8). Several genes related to cell signaling, cell growth, cell cycle and metabolic activities were uniquely identified (see supplementary table 3S or see additional file-8) and (Table 3/ see additional file 9). For example, Glypican-3, member of the glypican-related integral membrane proteoglycan family (GRIPS) contains a core protein anchored to the cytoplasmic membrane via a glycosyl phosphatidylinositol linkage and plays a role in the control of cell division and growth regulation was over expressed. Calreticulin (CALR), which can inhibit androgen receptor and retinoic acid receptor transcriptional activities in vivo, as well as retinoic acid-induced neuronal differentiation was over expressed by 5 fold. Cyclin-dependent kinase inhibitor 1C (CDKN1C), an inhibitor of several G1 cyclin/CDK complexes (Cyclin-E-CDK2, Cyclin-D2-CDK4 and Cyclin-A-CDK2) and to lesser extent of the mitotic cyclin-B-CDC2, was over expressed in EBs. Finally, 26 hypothetical, 2 zinc finger proteins and 9 unknown genes whose functions are still to be determined were also over expressed in EBs (Supplementary Table 3S and 6S or see additional file-8) and (Table 4/ see additional file 9).
Overall, these results indicate that 194 genes identified as upregulated in two different EB cell lines may serve as early and sensitive markers to monitor ES cell differentiation.
Confirmation of gene expression profile by microarray, MPSS, EST-enumeration, RT-PCR and immunohistochemical analyses
To provide independent verification of the results we utilized three different strategies. We compared gene expression patterns obtained using microarray studies with the MPSS data set generated by our laboratory using the pooled ES and EBs derived from them [30]. In addition, we prepared duplicate samples of BG02 ES and BG02 derived embryoid bodies and subjected them to a microarray analysis using a large scale oligonucleotide array based on a different set of oligonucleotides that were commercially available (Agilent, Foster city, CA). Finally we examined a subset of genes that were not validated by either of these methods by RT-PCR. Of the 194 genes, which were uniquely over expressed at ≥ 3 fold in both BG02 and WiCell cell derived EBs 148 genes showed higher expression in an MPSS analysis of WiCell samples (data now shown and reference [33]. Overall, comparison of microarray results with MPSS showed a high concordance in gene expression profile. For example, known ES cell specific markers including such as POU5F1, Galanin, DNMT3B, GJA1, LEFTB and TDGF1 showed higher expression in undifferentiated BG02 and PES cells compared to differentiated BG02 or PEB by both microarray and MPSS analyses (Table 5/ see additional file 9). Similarly, early ES differentiation markers e.g., KRT8, KRT18, KRT19, ACTC and EB specific genes such as Vimentin, AFP, HAND1, and COL1A2 showed higher expression in EBs compared to ESC by both microarray and EST enumeration (Table 6 and Table 7/ see additional file 9).
When fold expression of 8 out of 9 unknown genes identified by microarray was compared with tpm level detected by MPSS, a similar pattern of gene expression was observed (Table 8/ see additional file 9). All of these genes showed higher expression in BG02-EB and WiCell derived EB compared to BG02-ES and WiCell ESC samples. Six of these genes were also confirmed by RT-PCR analysis (Fig. 5). The reliability of microarray results was further confirmed by RT-PCR analysis of 10 genes for hypothetical proteins, 2 for zinc finger proteins, 3 for unknown proteins and 7 genes that were highly expressed in embryoid bodies (Fig. 4, 5, 6). This similarity in the results confirmed the reliability of the microarray studies.
Figure 4 RT-PCR analysis of some overexpressed genes and novel genes in PEB and EB derived from BG02 compared to ES. RT-PCR analysis of novel genes in PEB and BG02 EB compared to respective ESC. Total RNA derived from both the ES and EB cell lines (BG02 and WiCell lines) were used and RT-PCR performed as described in Fig. 2 legend. G3PDH mRNA served as an internal control.
Figure 5 RT-PCR analysis of some novel genes over expressed in PEB and EB derived from BG02 compared to ES. RT-PCR analysis of six novel genes in EB confirmed by MPSS. Total RNA derived from both the ES and EB cell lines (BG02 and WiCell lines) were subjected to RT-PCR analysis as described in Fig. 2 legend. G3PDH mRNA amplified from these samples served as an internal control.
Figure 6 RT-PCR analysis of some distinctly over expressed genes in EB. RT-PCR analysis of some distinctly overexpressed genes in EB as identified by microarray analysis. Marked overexpression in EB is clearly documented in the figure. Total RNA derived from both the ES and EB cell lines (BG02 and WiCell lines) were subjected to RT-PCR analysis as described in Fig. 2 legend. G3PDH mRNA amplified from these samples served as an internal control.
As forty-six of the 194 genes (Table 9/ see additional file 9) identified by microarray showed no significant difference by MPSS analysis, it was concluded that microarray and MPSS assays may be different in sensitivity or there was variability in the production of embryoid bodies. To address this issue and to determine if the differences observed by microarray were reliable, we used a second microarray platform to examine gene expression in the same samples. Interestingly 33 of the 46 genes showed overexpression in WiCell derived EBs using Agilent human 22 k oligo-arrays (Table 9/ see additional file 9). This suggested that different methods have different sensitivities and it is important to use multiple methods to confirm expression.
The expression of the remaining 13 genes detected as overexpressed by microarray studies using a custom built microarray but not by MPSS or by Agilent microarrays was tested by RT-PCR analysis (Fig 4). Of the thirteen genes, 9 were confirmed by RT-PCR analysis further confirming the differential sensitivity of various arrays and other large-scale analytical methods.
Gene expression profile of Day 21 BG02-EB
Our results identified a large set of genes that are differentially modulated as cells differentiate to form embryoid bodies over a period of two weeks. To examine whether the same set of genes could be used to assess differentiation of hESC at twenty-one day of differentiation, we examined gene expression using RNA prepared from Day 21 embryoid bodies. For these studies, undifferentiated hESC (BG02-ESC) cells were used to generate EBs by a brief exposure to collagenase IV and small clusters of cells were obtained by scraping with a pipette. ES cells were differentiated for 13 and 21 days, harvested to prepare total RNA and analyzed by hybridization to microarrays. We found that a majority of ESC specific genes that were down modulated as cells differentiated for thirteen days were further down modulated as a result of differentiation to day 21 BG02-EB (data not shown). Among 194 genes overexpressed in day 13 BG02-EB and PEB, thirty-three genes showed a decrease in fold expression at day 21 BG02-EB. However, other than COL1A2, which was down modulated from 27 fold to 3 fold, none showed any marked decrease. The remaining genes were either not expressed or did not show any change in the expression (data not shown).
We previously reported that eight early differentiation marker genes were expressed in hESC, which were further upregulated in hEBs as determined by EST enumeration [28]. Therefore, in this study we examined their expression at day 21 of differentiation. For this experiment gene expression between BG02 ES was compared with day 13 BG02-EB and day 21 BG02-EB. We found that six of these genes were markedly down modulated in day 21 BG02-EB (Table 10/ see additional file 9) compared to day 13EB and PEB. The other two genes showed minor or no significant change.
In addition, expression of 11 known EB specific genes that were shown to be overexpressed in both day 13 BG02-EB and pooled EB (supplementary table 4S and 5S or see additional file-8) also examined in day 21 BG02-EB. Interestingly, all of them were down modulated in day 21 BG02-EB (Supplementary table 4S or see additional file-8)
Additionally, we examined the status of genes that were upregulated in day 21EB but not in ES or day 13 BG02-EB. Genes related to cytokeratin and hair keratin (e.g., KRT17, KRT20, and IVL), which are responsible for structural integrity showed higher expression in day 21EB compared to BG02ES or day 13EB. In addition, genes related to mature tissues were exclusively upregulated in day 21EB but not in ES and day 13 BG02-EB (Supplementary Table 4S or see additional file-8) indicating that additional differentiation markers can be used to distinguish early vs. late EBs.
While most genes followed an expected pattern of change in expression in day21 EB, Nanog showed a reverse pattern of expression. On day 13 its expression level was significantly reduced compared to BG02-ES, however, on day 21EB an increased expression was observed compared to day 13 BG02-EB. Levels were almost similar to those in BG02-ESC samples (data not shown). Downmodulation of Nanog in day 13-EB was confirmed by RT-PCR analysis (Fig 7) though its upregulation in Day 21 samples did not match the RT-PCR results (see discussion). The downregulation in nanog followed the downregulation of Oct 3/4 and TERT and other ES specific genes (Figure 7 and data not shown). In addition, although Sox 2 gene showed a decrease in expression in pooled EB compared to pooled ESC but it didn't show any change in expression pattern from BG02ES to day 13 BG02-EB and day 21 BG02-EB (data not shown). Expression of Sox-2 gene was also confirmed by RT-PCR and immunocytochemistry analyses (Fig.7 and Fig. 8). Both analyses demonstrated that Sox2 is expressed in day13 and day 21EBs. Oct3/4 protein expression was used as a control that did not show expression in Sox-2 expressing cells confirming that the Sox-2 expression represented induction in a newly differentiated population rather than in persistent ESC.
Figure 7 RT-PCR analysis of some ES and EB specific genes. RT-PCR analysis of ES and EB specific genes in day 14 and day 21 BG02-EB and BG02-ES samples. Total RNA derived from both the ES and EB cell lines (BG02-ES and EB) were subjected to RT-PCR analysis as described in Fig. 2 legend. G3PDH mRNA amplified from these samples served as an internal control.
Figure 8 Immunocytochemistry of Oct-3/4 and Sox-2 in ES and EB. Immunocytochemistry of Oct3/4 and Sox-2 in ES and EB. Human neural progenitor cells derived from ES cells were stained with various antibodies. (A) All neural progenitor cells are Oct3/4 negative and (B) most of these cells are positive for Sox2. DAPI staining shows every cell in (A) and (B).
These results indicate that KRT8, KRT18, TUBB5, ACTC, SERPINH1, TUBB4, KRT19, HAND1, FN1, ENO1, COL1A2, COL5A2, COL4A2 and other identified markers may serve as a indicators of early and late stages of differentiation as they were further markedly down regulated on day 21 compared to day 13 of differentiation. In contrast, genes such as KRT17, KRT20, IVL, NPHP3, CAPN1, and CNTN6 are markers of later stage of differentiation as they are overexpressed in day 21 compared to day 13 of differentiation. Thus microarray studies can distinguish between ES cells and embryoid bodies as well as between early and late stage embryoid bodies.
Discussion
Our results show that large-scale oligonucleotide based microarrays can be used to distinguish between undifferentiated hESC and differentiating hEBs. Two sets of markers can be distinguished in a set of 'stemness" genes that are present at higher levels in undifferentiated cells whose levels are reduced as cells differentiate and a complementary set of genes that are absent or present at low levels in hESC and are upregulated as cells differentiate. The set of genes that are upregulated as cells differentiate include known as well as unknown genes. This pattern of gene expression was confirmed by a variety of independent means including microarray using a second independent array set, comparison with MPSS using similar samples and by RT-PCR. The combined set of upregulated and down regulated genes will serve as a sensitive indicator of the state of hESC and the novel sets of genes identified are likely candidates that may participate in regulating the process of differentiation.
In our previous study we demonstrated over expression of a set of 92 genes in six undifferentiated ESC cell [28]. Expression of almost all of the genes was confirmed in the BG02-ES and pooled hESC used in the present study. The present study further refines the set of stemness genes to identify those that are rapidly down-regulated as ES cells differentiate. We show that 77 of these are down modulated when differentiated to embryoid bodies using undifferentiated BG02-ES. Further differentiation of BG02-ES to day 21 BG02-EB showed consistency in down modulation of many of these. This refined subset of previously identified stemness genes can be used to monitor the transition of undifferentiated pluripotent human embryonic stem cells to differentiated embryoid bodies. Analyzing the pattern of gene expression, we would suggest that ES specific genes such as Lin-28, PSIP2, PITX2, DNMT3B, and Galanin whose downregulation is seen in both lines and confirmed by EST scan and microarray represent good initial candidates to assess the ES cell state. Furthermore, LEFTB and CER1, inhibitors of nodal signaling, which were downmodulated as ES cells differentiated (data not shown) in addition to other members of the TGFβ signaling family may be sensitive indicators of ESC differentiation. Oct3/4, Sox-2 and nanog in contrast while expressed by ES cells may not be as good in assessing ES cell to embryoid body differentiation. While Oct-3/4 expression declined markedly in both BG02 EBs and pooled ES derived EBs compared to ES the decline was slow and not as rapid as that of other genes. This may be because Oct3/4 expression persists in germ cells that are derived from ESC [31]. Likewise our analysis of nanog showed variable results and there was a discrepancy between the microarray and PCR results and we noted an increase in nanog levels in Day 21 BG02-EB samples by microarray. The reason for this difference is not clear. It is possible that microarray is detecting one of the eleven processed pseudogenes for nanog and that its expression on day 21 represents crosshybridization. Alternatively, nanog expression may be increased as a result of differentiation. This conclusion is supported by a recent report that nanog is expressed in mature tissues [32]. We have confirmed that nanog expression is indeed present in some germ cells and early neuronal populations (data not shown). These tests were performed by RT-PCR using multiple primers and double label immunocytochemistry (results not shown). Future studies will examine expression of nanog in EBs using specific antibody and its function in differentiated cells. At present we recommend that nanog expression should be confirmed by PCR or immunocytochemistry and it should not be included as an initial indicator of the ES cell state and differentiation.
In addition to confirming the downregulation of hESC specific genes our study identified a set 333 genes that were uniquely overexpressed at ≥ 3 fold in day 13 BG02 EB compared to undifferentiated BG02 ES cell lines. Out of these 333 genes 194 genes also showed ≥ 3 fold over expression in EB samples prepared from WiCell lines. This significant similarity of unique genes in BG02 and WiCell line derived EBs suggest that this subset of genes can be classified as EB specific (induced as ESC differentiate) and that embryoid bodies derived from different ES cells maintain significant similarity with each other. These EB specific genes like the ES cell specific genes may be useful in distinguishing EBs from ES cells.
A detailed analysis of upregulated genes in EB indicated that a) differentiation genes, b) cell signaling, cellular process and cell cycle genes, c) cytoskeleton or cell motility genes and d) metabolism and DNA and RNA related genes were modulated. In addition, 27 hypothetical genes, 8 unknown genes and 2 zinc finger genes were upregulated in BG02 derived EB and WiCell lines derived EB. The status of upregulated genes identified by microarray studies was confirmed by MPSS analysis. Among 194 genes, 148 genes showed higher expression in EB by MPSS analysis. Overexpression of 33 genes in EBs compared to HuURNA was confirmed using alternate large-scale array. The expression of a subset of the remaining 13 genes was confirmed by RT-PCR analysis. These results suggest that microarray analysis is a useful complement to MPSS and may identify a overlapping but non-identical set of genes. Among the genes that were upregulated in both EB samples is HAND1 (heart and neural crest derivatives 1). HAND1 was dramatically up regulated (40 fold compared to 6 fold in undifferentiated ES cells). HAND1 belongs to basic helix-loop-helix family of transcription factor, may be required for early trophoblast differentiation as well as the differentiation of heart tissue [33].
Additionally we noted that several markers of early differentiation such as keratin, actin and beta tubulin, which are present at low levels in ESC, were dramatically upregulated as EBs differentiated. We suggest that these markers can serve as sensitive indicators of the process of differentiation and when coupled with the down regulation of ESC markers may reliably distinguish the state of ES cell cultures. Future experiments will assess the sensitivity of a combination of markers identified in this study to temporally profile the process of differentiation. A recent study (34) has compared gene expression of 24-marker genes between EBs differentiated with eight different growth factors and hESC. Similar to our observation by microarray analysis, this study showed over expression of genes related to ectoderm differentiation such as Keratin-8, Keratin-18 and Keratin-19 when differentiated by retinoic acid. A highly significant upregulation of mesodermal related genes e.g., HAND1 was also observed which confirms our observation.
Our results further show that many genes that are induced or upregulated as EBs differentiate remain elevated at Day 21 of differentiation. These genes represent good markers for distinguishing differentiated cells from ESC. In addition we noted that several genes showed differential expression when levels of expression in Day 13 and Day 21 BG02-EB's was compared. Thirty-three of the 194 genes upregulated at Day 13 were down modulated at day21 EB. Other genes such as NPHP3, CAPN1, and CNTN6 were up regulated in day 21 BG02-EB's when compared to day 13 EB. These later appearing genes may reflect additional derivatives that appear as EBs undergo further differentiation. These later appearing markers along with markers that are down regulated can be used to distinguish early from late EBs.
A recent study has characterized differences in gene expression between mouse and human ESC (35). This study identified differences between mouse and human ESC and reported that differences were species specific rather than arising from differing culture conditions. In our current study, while the expression of many genes was similar in rodent and human cells significant differences were found. For example the expression of vimentin, eomesodermin, SSEA4, AFP, IL6ST and HEB was found in hESC but not in mouse ES cells.
The large oligonucleotide arrays have allowed us to build a comprehensive data set that includes hundreds of genes (both novel and unknown) that are either upregulated or downregulated as hESC differentiate. The list of genes has been validated by testing on different cells lines, testing using different oligonucleotide sequences to different regions of the genes and testing arrays used in different formats. Our results show clearly that while small differences exist as different techniques are used the core set of markers is quite robust, the differences in expression quite large and are readily detectable even using a relatively insensitive method such as a microarray. Our results show further that restricting the analysis to only this set of genes as would be necessary in focused array does not reduce the sensitivity of the assessment. Indeed, based on our results we would suggest that an order of magnitude fewer genes may be sufficient if they are appropriately selected from the lists. We have recently developed a focused array that includes genes that are down regulated as hESC differentiate as well as those that are induced as cells differentiate [36].
Conclusion
In conclusion, a similar pattern of gene expression profile was observed in two differentiated embryoid body samples derived from different embryonic stem cell lines. A consistent and marked downmodulation of most of the "stemness" genes was observed in embryoid bodies indicating that relative levels of these genes can be used to assess the transition of ES cells to embryoid bodies. In addition, we show that 194 unique genes are over expressed in EB and that this subset serves as a complement to the previously characterized "stemness genes" in assessing the degree of differentiation of ES lines and differentiating embryoid bodies. Further validation and confirmation of these data with MPSS and RT-PCR documents the usefulness of high throughput microarray technology and identifies an additional set of previously unknown genes that are likely important in regulating the process of differentiation. These EB enriched genes in combination with downmodulated stemness genes may serve as biomarkers to monitor the transition of undifferentiated human embryonic stem cells to differentiated embryoid bodies.
Methods
Isolation and growth of ES cells and differentiation into EB
BG02 ESC were derived at Bresagen Inc., Athens, GA (37) and pooled samples (WiCell lines: WA01, WA07 and WA09, also termed as GE01, GE07 and GE09 respectively) were derived at Wisconsin Alumni Research Foundation, Wicell Research Institute, Madison, WI. (38). ESC were maintained on inactivated mouse embryonic fibroblasts (MEF) feeder cells in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 15% fetal bovine serum (FBS), 5% knockout serum replacement (KSR), 2 mM non-essential amino acids, 2 mM L-glutamine, 50 μg/ml Penn-Strep (all from Invitrogen, Corporation, Carlsbad, CA), 0.1 mM β-mercaptoethanol (Specialty Media), and 4 ng/ml of basic fibroblast growth factor (bFGF, Sigma). These cell lines were maintained as previously described [3,7,39]. Cells were passaged by incubation in cell dissociation buffer (Invitrogen), dissociated, and then seeded at about 20000 cells/cm2. hESC harvested for studies were found to be free of MEF as previously described [30,35].
Embryoid body outgrowths (hEB) were prepared from BG02 cells and pooled samples of ESC (pooled EB) as described [23,30]. Briefly, confluent plates of undifferentiated hESC were used to generate hEBs by a brief exposure to collagenase IV and small clusters of cells were obtained by scraping with a pipette. Cell clusters were resuspended in differentiation medium (KO-DMEM supplemented with glutamine, NEAA, and BME as described for the undifferentiated ESC [8], with 20% FBS in place of 20% serum replacement (SR) and no preconditioning by MEFs) and transferred to individual wells of low adhesion 6-well plates (Costar). After 4 days in suspension, cells were transferred to tissue culture 6-well plates pre-coated with gelatin. Cells were harvested for the preparation of total RNA from BG02 EB on day 13 and day 21 and from pooled EB on day 14. Based on morphological evaluation undifferentiated hESC were also identified in certain regions of differentiating colonies even after 21 days of in vitro culture, suggesting that hESC can undergo several cell divisions in "differentiation promoting" culture conditions while maintaining their pluripotent phenotype [40].
Microarray studies
High quality oligonucleotide glass arrays were produced containing a total of 16,659 seventy-mer oligonucleotides chosen from 750 bases of the 3' end of each ORF (Operon Inc. Valencia, CA). The array includes probes for 2121 hypothetical proteins and 18-expressed sequence tags (ESTs) and spans approximately 50% of the human genome (Operon Inc., Valencia, CA). The arrays were produced in house by spotting oligonucleotides on poly-L-lysine coated glass slides by Gene Machines robotics (Omnigrid, San Carlos, CA). We have followed the MIAME (minimum information about a microarray experiment) guidelines for the presentation of our data [41].
i) Probe preparation
Total RNA was isolated from hESC lines derived pellet by using Trizol reagent (Invitrogen). Total human universal RNA (huURNA) isolated from a collection of adult human tissues to represent a broad range of expressed genes from both male and female donors (BD Biosciences, Palo Alto, CA) served as a universal reference control in the competitive hybridization. Labeled cDNA probes were produced as described [42]. Briefly, 5 μg of total RNA was dissolved in 12 μl of DEPC water and incubated at 70°C for 5 minutes along with 1 μl of aminoallyl-oligo dT (5' amino-modified) primer and quickly chilled for 3 minutes. Then, 2 μl 10× first strand buffer, 1.5 μl SSII enzyme (Stratagene, La Jolla, CA), 1.5 μl 20× aminoallyl dUTP and 2 μl of 0.1 M DTT were added and incubated for 90 minutes at 42°C. After incubation, volume of the reaction mixture was raised to 60 μl with 40 μl of DEPC water.
cDNA was purified by MinElute column (Qiagen, Valencia, CA). 300 μl of Binding buffer PB was added to the coupled cDNA, and the mixture applied to the MinElute column, and centrifuged for 1 min at max speed. After discharging the flow-through, 600 μl of washing buffer PE was added to the column, and centrifuged for 1 min at max speed. The flow-through was discharged and the washing repeated. Then the columns were placed into a fresh eppendorf tube and 15 μl elution buffer added to the center of the membrane, incubated for 1 min at room temperature, centrifuged for 1 min at max speed and probe collected. The probe was dried in speed-vac for 16 minutes. Finally, 5 μl of 2× coupling buffer and 5 μl Cy3 and Cy5 dye mixed into the control (huURNA) and experimental cDNAs (huES cell-derived) and incubated at room temperature in dark for 90 minutes. After incubation, the volume was raised to 60 μl by 50 μl DEPC water and then cDNA was purified by MinElute column and eluted with 13 μl elution buffer by centrifugation.
ii) Hybridization
For hybridization, 36 μl hybridization mixture [26 μl cDNA mixture, 1 μl (10 μg) COT-1 DNA, 1 μl (8–10 μg) poly(dA), 1 μl yeast tRNA (4 μg), 6 μl 20× SSC and 1 μl 10% SDS] was pre-heated at 100°C for 2 minutes and cooled for 1 minute. Total volume of probe was added on the array and covered with cover slip (22 mm × 40 mm). Slides are placed in hybridization chamber and 20 μl water was added to far end of slide (to maintain humidity), and incubated overnight (10–16 hours) at 65°C. Slides were then washed for 2 minutes each into 2× SSC, 1× SSC and 0.1× SSC and spin-dried.
iii) Data filtration, normalization, and analysis
Microarray slides were scanned in both Cy3 (532 nm) and Cy5 (635 nm) channels using Axon GenePix 4000B scanner (Axon Instruments, Inc., Foster City, CA) with a 10-micron resolution. Scanned microarray images were exported as TIFF files to GenePix Pro 3.0 software for image analysis. The raw images were collected at 16-bit/pixel resolutions with 0 to 65,535 count dynamic range. The area surrounding each spot image was used to calculate a local background and subtracted from each spot before Cy5:Cy3 ratio calculation. The average of the resulting total Cy3 and Cy5 signal gave a ratio that was used to normalize the signals. Each microarray experiment was globally normalized to make the median value of the log2-ratio equal to zero. The normalization process corrects for dye bias, PMT (Photo multiplier tube) voltage imbalance, and variations between channels in the amounts of the labeled cDNA probes hybridized. The data files representing the differentially expressed genes were then created.
For advanced data analysis, data files (in gpr format) and image (in jpeg format) were imported into mAdb (microarray database), and analyzed by software tools provided by National Institutes Health Center for Information Technology. Spots with confidence interval of 99 (>3 fold) with at least 150-fluorescence intensity in either channel or 30 μm spot size were considered as good quality spots for analysis. These advanced filters prevented the potential effect of the poor quality spots in data analysis.
RT-PCR analysis
Total RNA was isolated from both ES and EB cell pellets using the RNeasy Qiagen mini protocol and kit (Qiagen, Valencia, CA). cDNA was synthesized using 1 μg of total RNA in a 20 μl reaction. Superscript II (Invitrogen, Carlsbad, CA) and oligo (dT)12–18 primers were used according to the manufacturer's instructions (Invitrogen). The list of primers used for RT-PCR was described in supplemental table 6S (see additional file 8) and annealing conditions are described previously [35].
MPSS analysis and EST enumeration
We compared microarray results with published MPSS and EST enumeration databases as previously described [30,23]. In brief, MPSS was performed using RNA samples from feeder free cultures of differentiated (day 14 pooled EB derived from WiCell lines) and undifferentiated (passage 35–45) WA01, WA07, and WA09 lines (pooled ESC: pooled samples of ESC). The mRNA was converted to cDNA and digested with DpnII. The last DpnII site and the downstream 16 bases were cloned into Megaclone vectors and their sequences (signatures) were determined according to the MPSS protocol [43,44]. A total of 2,403,315 and 2,591,008 sequences were read for pooled ESC and pooled EB respectively, and 36,498 and 26,599 unique signatures were identified. The abundance for each signature was converted to transcripts per million (tpm) for the purpose of comparison between samples. Signatures at an abundance of less than 4 tpm in at least one of samples or those that were not detected in at least two runs across multiple samples were removed and a total of 24, 229 (pooled ES) and 17,970 (pooled EB) sequences were analyzed further.
EST frequency counts of genes expressed in EB samples were performed as described [23]. Briefly, cDNA libraries of hESC lines (WiCell WA01, WA07, and WA09) grown in feeder free conditions and that of EB's derived from the same cell lines were constructed and submitted for EST sequencing. The EST sequences were assembled into overlapping sequence assemblies and mapped to the UniGene database of non-redundant human transcripts. Expression levels were assessed by counting the number of ESTs for a particular gene that was derived from the undifferentiated hESC and comparing them to the number of ESTs derived from the EB samples. Statistical significance was determined using the Fisher Exact Test [45] using a p-value of p≤ 0.05.
MPSS signature classification and annotation
To generate a complete, annotated human signature database, we extracted all the possible signatures ("virtual signatures") from the human genome sequence (hg16, July 2003, Golden Path, UCSC) and the human Unigene sequences (Unigene build #163). The extracted virtual signatures were classified and ranked according to their positions and the presence of polyA tail and polyA signal features in the source sequence as described [46].
Immunocytochemistry
Neural cells derived from hESC were processed for immunocytochemistry as described previously [47]. The human Oct3/4 (1:1000) and Sox2 (1:1000) antibodies were obtained form R&D Systems. Bis-benzamide (1:1000, Sigma) was used to identify the nuclei. The secondary antibodies anti-mouse IgG 568 (1:500) and anti-goat IgG 568 (1:500) were purchased from Molecular Probes. Images were captured on an Olympus fluorescence microscope.
List of abbreviations
hESC-Human embryonic stem cells, EB-embryoid bodies, hEB-Human embryoid bodies, PES-Pooled ES or Wicell derived ES, PEB-Pooled EB, huURNA-Human Universal Reference RNA, MPSS-massively parallel signature sequencing
Authors' contributions
Authors from FDA/CBER conceived the idea of examining gene expression between ESC and EB cells, performed all microarray hybridizations, performed data analysis, interpreted results, drafted the manuscript and negotiated for acceptance and final publication. Authors from NIH/NIA helped design the study, performed RT-PCR and IHC analyses. NIH/NIDA authors performed RT-PCR analysis to confirm expression of some of the genes. Authors from Bresagen Inc. provided stem cell lines to NIA authors and RNA was extracted at NIA. All authors have read and approved the final manuscript.
Supplementary Material
Additional File 1
Jpeg files of all five representative images from five different ES and EB samples (cy5 labeled) hybridized against Human Universal Reference RNA (HuURNA) (cy3 labeled). BG02-ES.
Click here for file
Additional File 2
Jpeg files of all five representative images from five different ES and EB samples (cy5 labeled) hybridized against Human Universal Reference RNA (HuURNA) (cy3 labeled). Day21-BG02-EB.
Click here for file
Additional File 3
Jpeg files of all five representative images from five different ES and EB samples (cy5 labeled) hybridized against Human Universal Reference RNA (HuURNA) (cy3 labeled). Day13-BG02-EB.
Click here for file
Additional File 4
Jpeg files of all five representative images from five different ES and EB samples (cy5 labeled) hybridized against Human Universal Reference RNA (HuURNA) (cy3 labeled). Pooled-EB.
Click here for file
Additional File 5
Jpeg files of all five representative images from five different ES and EB samples (cy5 labeled) hybridized against Human Universal Reference RNA (HuURNA) (cy3 labeled). Pooled-ES.
Click here for file
Additional File 6
Supplementary file -1Sa that represents excel file representation of ≥ 2 fold over expressed genes in BG02-ESC samples hybridized against HuURNA
Click here for file
Additional File 7
Supplementary file -1Sb that represents excel file representation of ≥ 2 fold over expressed genes in Pooled-ESC samples hybridized against HuURNA
Click here for file
Additional File 8
Represents Supplementary table 2S, 3S, 4S, 5S and 6S in word format with heading and legends mentioned separately in each table
Click here for file
Additional File 9
Represents the tables 1–10 in word format with heading and legends mentioned separately in each table
Click here for file
Acknowledgements
We thank Drs. Brenton McCright and Robert Duncan for manuscript review and helpful comments. We also thank Drs. Jing Han for general support and Dr. Amy X Yang for providing oligonucleotide microarrays.
==== Refs
Chen LR Shiue YL Bertolini L Medrano JF BonDurant RH Anderson GB Establishment of pluripotent cell lines from porcine preimplantation embryos Theriogenology 1999 52 195 212 10734388 10.1016/S0093-691X(99)00122-3
Thomson JA Kalishman J Golos TG Durning M Harris CP Becker RA Hearn JP Isolation of a primate embryonic stem cell line Proc Natl Acad Sci U S A 92 7844 7848 1995 Aug 15 7544005
Amit M Carpenter MK Inokuma MS Chiu CP Harris CP Waknitz MA Itskovitz-Eldor J Thomson JA Clonally derived human embryonic stem cell lines maintain pluripotency and proliferative potential for prolonged periods of culture Dev Biol 2000 15 271 278 11071754 10.1006/dbio.2000.9912
Reubinoff BE Pera MF Fong CY Trounson A Bongso A Embryonic stem cell lines from human blastocysts: somatic differentiation in vitro Nat Biotechnol 2000 18 399 404 10748519 10.1038/74447
Amit M Itskovitz-Eldor J Derivation and spontaneous differentiation of human embryonic stem cells J Anat 2002 200 225 232 Review 12033726 10.1046/j.1469-7580.2002.00032.x
Richards M Tan SP Tan JH Chan WK Bongso A The transcriptome profile of human embryonic stem cells as defined by SAGE Stem Cells 2004 22 51 64 14688391 10.1634/stemcells.22-1-51
NIH Human Embryonic Stem Cell Registry NIH Web site Accessed June, 2004
Xu C Inokuma MS Denham J Golds K Kundu P Gold JD Carpenter MK Feeder-free growth of undifferentiated human embryonic stem cells Nat Biotechnol 2001 19 971 974 11581665 10.1038/nbt1001-971
Carpenter MK Rosler E Rao MS Characterization and differentiation of human embryonic stem cells Cloning Stem Cells 2003 5 79 88 12713704 10.1089/153623003321512193
Bagutti C Hutter C Chiquet-Ehrismann R Fassler R Watt FM Dermal fibroblast-derived growth factors restore the ability of beta (1) integrin-deficient embryonal stem cells to differentiate into keratinocytes Dev Biol 231 321 333 2001 Mar 15 11237462 10.1006/dbio.2000.0149
Carpenter MK Inokuma MS Denham J Mujtaba T Chiu CP Rao MS Enrichment of neurons and neural precursors from human embryonic stem cells Exp Neurol 2001 172 383 397 11716562 10.1006/exnr.2001.7832
Chen LR Shiue YL Bertolini L Medrano JF BonDurant RH Anderson GB Establishment of pluripotent cell lines from porcine preimplantation embryos Theriogenology 52 195 212 1999 Jul 15 10734388 10.1016/S0093-691X(99)00122-3
Wang L Menendez P Shojaei F Li L Mazurier F Dick JE Cerdan C Levac K Bhatia M Generation of hematopoietic repopulating cells from human embryonic stem cells independent of ectopic HOXB4 expression J Exp Med 201 1603 14 2005 May 16; Epub 2005 May 9. 15883170 10.1084/jem.20041888
Schulz TC Palmarini GM Noggle SA Weiler DA Mitalipova MM Condie BG Directed neuronal differentiation of human embryonic stem cells BMC Neurosci 4 27 2003 Oct 22 14572319 10.1186/1471-2202-4-27
Rao MS Conserved and divergent paths that regulate self-renewal in mouse and human embryonic stem cells Dev Biol 275 269 86 2004 Nov 15; Review 15501218 10.1016/j.ydbio.2004.08.013
Khoo ML McQuade LR Smith MS Lees JG Sidhu KS Tuch BE Growth and Differentiation of Embryoid Bodies Derived from Human Embryonic Stem Cells: Effect of Glucose and Basic Fibroblast Growth Factor Biol Reprod 2005 Aug 3 16079311
Reppel M Boettinger C Hescheler J Beta-adrenergic and muscarinic modulation of human embryonic stem cell-derived cardiomyocytes Cell Physiol Biochem 2004 14 187 96 15319521 10.1159/000080326
Segev H Kenyagin-Karsenti D Fishman B Gerecht-Nir S Ziskind A Amit M Coleman R Itskovitz-Eldor J Molecular analysis of cardiomyocytes derived from human embryonic stem cells Dev Growth Differ 2005 47 295 306 16026538 10.1111/j.1440-169X.2005.00803.x
Yokota N Mainprize TG Taylor MD Kohata T Loreto M Ueda S Dura W Grajkowska W Kuo JS Rutka JT Identification of differentially expressed and developmentally regulated genes in medulloblastoma using suppression subtraction hybridization Oncogene 23 3444 3453 2004 Apr 22 15064731 10.1038/sj.onc.1207475
Wentzensen N Wilz B Findeisen P Wagner R Dippold W von Knebel Doeberitz M Gebert J Identification of differentially expressed genes in colorectal adenoma compared to normal tissue by suppression subtractive hybridization Int J Oncol 2004 24 987 994 15010839
Pelton TA Sharma S Schultz TC Rathjen J Rathjen PD Transient pluripotent cell population during primitive ectoderm formation: correlation of in vivo and in vitro pluripotent cell development Journal of Cell Science 2002 115 329 339 11839785
Yates KE Forbes RL Glowacki J New chondrocyte genes discovered by representational difference analysis of chondroinduced human fibroblasts Cells Tissues Organs 2004 176 41 53 14745234 10.1159/000075026
Brandenberger R Wei H Zhang S Lei S Murage J Fisk GJ Li Y Xu C Fang R Guegler K Rao MS Mandalam R Lebkowski J Stanton LW Transcriptome characterization elucidates signaling networks that control human ES cell growth and differentiation Nat Biotechnol 2004 22 707 716 15146197 10.1038/nbt971
Richards M Tan SP Tan JH Chan WK Bongso A The transcriptome profile of human embryonic stem cells as defined by SAGE Stem Cells 2004 22 51 64 14688391 10.1634/stemcells.22-1-51
Hu JG Fu SL Zhang KH Li Y Yin L Lu PH Xu XM Differential gene expression in neural stem cells and oligodendrocyte precursor cells: A cDNA microarray analysis J Neurosci Res 2004 1 637 646 15499592 10.1002/jnr.20317
Rao RR Calhoun JD Qin X Rekaya R Clark JK Stice SL Comparative transcriptional profiling of two human embryonic stem cell lines Biotechnol Bioeng 88 273 286 2004 Nov 5 15493035 10.1002/bit.20245
Georgantas RW 3rdTanadve V Malehorn M Heimfeld S Chen C Carr L Martinez-Murillo F Riggins G Kowalski J Civin CI Microarray and serial analysis of gene expression analyses identify known and novel transcripts overexpressed in hematopoietic stem cells Cancer Res 64 4434 4441 2004 Jul 1 15231652 10.1158/0008-5472.CAN-03-3247
Bhattacharya B Miura T Brandenberger R Mejido J Luo Y Yang AX Joshi BH Ginis I Thies RS Amit M Lyons I Condie BG Itskovitz-Eldor J Rao MS Puri RK Gene expression in human embryonic stem cell lines: unique molecular signature Blood 2004 103 2956 2964 15070671 10.1182/blood-2003-09-3314
Tanaka TS Kunath T Kimber WL Jaradat SA Stagg CA Usuda M Yokota T Niwa H Rossant J Ko MS Gene Expression Profiling of Embryo-Derived Stem Cells Reveals Candidate Genes Associated With Pluripotency and Lineage Specificity Genome Res 2002 12 1921 1928 12466296 10.1101/gr.670002
Brandenberger R Khrebtukova I Thies RS Miura T Jingli C Puri R Vasicek T Lebkowski J Rao M MPSS profiling of human embryonic stem cells BMC Dev Biol 4 1 16 2004 Aug 10 15005800
Niwa H Miyazaki J Smith AG Quantitative expression of Oct-3/4 defines differentiation, dedifferentiation or self-renewal of ES cells Nat Genet 2000 24 372 376 10742100 10.1038/74199
Hart HA Hartley L Ibrahim M Robb L Identification, cloning and expression analysis of the pluripotency promoting Nanog genes in Mouse and Human Developmental Dynamics 2004 230 187 198 15108323 10.1002/dvdy.20034
Knofler M Meinhardt G Vasicek R Husslein P Egarter C Molecular cloning of the human Hand1 gene/cDNA and its tissue-restricted expression in cytotrophoblastic cells and heart Gene 224 77 86 1998 Dec 11 9931445
Schuldiner M Yanuka O Itskovitz-Eldor J Melton DA Benvenisty N Effects of eight growth factors on the differentiation of cells derived from human embryonic stem cells PNAS 97 11307 11312 2000 Oct 10 11027332 10.1073/pnas.97.21.11307
Ginis I Luo Y Miura T Thies S Brandenberger R Gerecht-Nir S Amit M Hoke A Carpenter MK Itskovitz-Eldor J Rao MS Differences between human and mouse embryonic stem cells Dev Biol 2004 269 360 380 15110706 10.1016/j.ydbio.2003.12.034
Yang AX Mejido J Luo Y Zheng X Schwartz C Wu T Thies RT Bhattacharya B Han J Freed W Rao MS lPuri RK Development of a Focused Microarray to Assess Human Embryonic Stem Cell Differentiation Stem Cells and Development 2005 14 270 284 15969622 10.1089/scd.2005.14.270
Mitalipova M Calhoun J Shin S Wininger D Schulz T Noggle S Venable A Lyons I Robins A Stice S Human embryonic stem cell lines derived from discarded embryos Stem Cells 2003 21 521 526 12968106 10.1634/stemcells.21-5-521
Xu C Inokuma MS Denham J Golds K Kundu P Gold JD Carpenter MK Feeder-free growth of undifferentiated human embryonic stem cells Nat Biotechnol 2001 19 971 4 11581665 10.1038/nbt1001-971
Thomson JA Itskovitz-Eldor J Shapiro SS Waknitz MA Swiergiel JJ Marshall VS Jones JM Embryonic stem cell lines derived from human blastocysts Science 1998 282 1145 1147 9804556 10.1126/science.282.5391.1145
Draper JS Pigott C Thomson JA Andrews PW Surface antigens of human embryonic stem cells: change upon differentiation in culture J Anat 2002 200 249 258 12033729 10.1046/j.1469-7580.2002.00030.x
Brazma A Hingamp P Quackenbush J Sherlock G Spellman P Stoeckert C Aach J Ansorge W Ball CA Causton HC Gaasterland T Glenisson P Holstege FC Kim IF Markowitz V Matese JC Parkinson H Robinson A Sarkans U Schulze-Kremer S Stewart J Taylor R Vilo J Vingron M Minimum information about a microarray experiment (MIAME)-toward standards for microarray data Nat Genet 2001 29 365 371 11726920 10.1038/ng1201-365
Risinger JI Maxwell GL Chandramouli GV Jazaeri A Aprelikova O Patterson T Berchuck A Barrett JC Microarray analysis reveals distinct gene expression profiles among different histologic types of endometrial cancer Cancer Res 2003 63 6 11 12517768
Brenner S Johnson M Bridgham J Golda G Lloyd DH Johnson D Luo S McCurdy S Foy M Ewan M Roth R George D Eletr S Albrecht G Vermaas E Williams SR Moon K Burcham T Pallas M DuBridge RB Kirchner J Fearon K Mao J Corcoran K Gene expression analysis by massively parallel signature sequencing (MPSS) on microbead arrays Nat Biotechnol 2000 18 1021
Brenner S Williams SR Vermaas EH Storck T Moon K McCollum C Mao JI Luo S Kirchner JJ Eletr S DuBridge RB Burcham T Albrecht G In vitro cloning of complex mixture of DNA on microbead: physical separation of differentially expressed cDNA Proc Natl Acad Sci USA 2000 97 1665 1670 10677516 10.1073/pnas.97.4.1665
Siegel S Castellan N Non parametric Statistics for the Behavioral Sciences 1988 Columbus, OH: McGraw-Hill
Miura T Luo Y Khrebtukova I Brandenberger R Zhou D Thies RS Vasicek T Young H Lebkowski J Carpenter MK Rao MS Monitoring early differentiation events in human embryonic stem cells by massively parallel signature sequencing and expressed sequence tag scan Stem Cells Dev 2004 13 694 715 15684837 10.1089/scd.2004.13.694
Mayer-Proschel M Kalyani AJ Mujtaba T Rao MA Isolation of lineage-restricted neuronal precursors from multipotent neuroepithelial stem cells Neuron 1997 19 773 785 9354325 10.1016/S0896-6273(00)80960-5
|
16207381
|
PMC1260016
|
CC BY
|
2021-01-04 16:30:31
|
no
|
BMC Dev Biol. 2005 Oct 5; 5:22
|
utf-8
|
BMC Dev Biol
| 2,005 |
10.1186/1471-213X-5-22
|
oa_comm
|
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-481619119810.1186/1471-2148-5-48Research ArticlePhylogenetic analysis of eIF4E-family members Joshi Bhavesh [email protected] Kibwe [email protected] Dennis L [email protected] Rosemary [email protected] Center of Marine Biotechnology, Suite 236 Columbus Center, 701 E. Pratt Street, Baltimore, MD 21202, USA2005 28 9 2005 5 48 48 27 5 2005 28 9 2005 Copyright © 2005 Joshi et al; licensee BioMed Central Ltd.2005Joshi 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
Translation initiation in eukaryotes involves the recruitment of mRNA to the ribosome which is controlled by the translation factor eIF4E. eIF4E binds to the 5'-m7Gppp cap-structure of mRNA. Three dimensional structures of eIF4Es bound to cap-analogues resemble 'cupped-hands' in which the cap-structure is sandwiched between two conserved Trp residues (Trp-56 and Trp-102 of H. sapiens eIF4E). A third conserved Trp residue (Trp-166 of H. sapiens eIF4E) recognizes the 7-methyl moiety of the cap-structure. Assessment of GenBank NR and dbEST databases reveals that many organisms encode a number of proteins with homology to eIF4E. Little is understood about the relationships of these structurally related proteins to each other.
Results
By combining sequence data deposited in the Genbank databases, we have identified sequences encoding 411 eIF4E-family members from 230 species. These sequences have been deposited into an internet-accessible database designed for sequence comparisons of eIF4E-family members. Most members can be grouped into one of three classes. Class I members carry Trp residues equivalent to Trp-43 and Trp-56 of H. sapiens eIF4E and appear to be present in all eukaryotes. Class II members, possess Trp→Tyr/Phe/Leu and Trp→Tyr/Phe substitutions relative to Trp-43 and Trp-56 of H. sapiens eIF4E, and can be identified in Metazoa, Viridiplantae, and Fungi. Class III members possess a Trp residue equivalent to Trp-43 of H. sapiens eIF4E but carry a Trp→Cys/Tyr substitution relative to Trp-56 of H. sapiens eIF4E, and can be identified in Coelomata and Cnidaria. Some eIF4E-family members from Protista show extension or compaction relative to prototypical eIF4E-family members.
Conclusion
The expansion of sequenced cDNAs and genomic DNAs from all eukaryotic kingdoms has revealed a variety of proteins related in structure to eIF4E. Evolutionarily it seems that a single early eIF4E gene has undergone multiple gene duplications generating multiple structural classes, such that it is no longer possible to predict function from the primary amino acid sequence of an eIF4E-family member. The variety of eIF4E-family members provides a source of alternatives on the eIF4E structural theme that will benefit structure/function analyses and therapeutic drug design.
==== Body
Background
The recruitment of mRNAs to the ribosomal apparatus is a key step in the regulation of translation initiation. For the majority of eukaryotic mRNAs, recruitment is dependent upon the activity of the translation initiation factor eIF4E. eIF4E binds to the 5'-m7GTP-cap structure of mRNAs and to the initiation factor eIF4G (reviewed [1-3]). Through the interaction of the eIF4E:eIF4G complex with ribosome bound factor eIF3, the 40 S ribosomal subunit is positioned at the 5'-end of the mRNA. Subsequently, the 40 S ribosomal subunit scans the mRNA (5'-3') for the translational start codon prior to 60 S binding and formation of the first peptide bond.
The crystal structures of eIF4Es from Mus musculus and Homo sapiens and the solution structure of eIF4E from Saccharomyces cerevisiae, in each case bound to cap-analogues, show that each consists of an eight-stranded β-sheet supported by three α-helices forming the palm and back of a 'cupped' hand [4-6]. Two conserved aromatic Trp residues (Trp-56 and Trp-102 for H. sapiens eIF4E) grasp the aromatic guanine residue of the cap-structure through 'π'-bond interactions [4,5]. Similar interactions of aromatic amino acid residues with the guanine nucleotide of cap-analogues are seen in the structures of other cap-binding proteins which appear to have evolved independently such as the vaccinia virus 2'-O-methyltransferase, VP39, and of the nuclear cap-binding protein subunit, CBP20, suggesting a common evolutionary theme for methylguanosine/nucleotide-interaction [7]. Hydrogen bonds to the guanine base from a Glu residue (Glu-103 of H. sapiens eIF4E) and the adjacent peptide bond stabilize the interaction of the cap-analogue to the protein. A third Trp residue (Trp-166 of H. sapiens eIF4E) interacts with the N7-methyl moiety of the cap-structure. Sequence comparisons of mammalian eIF4E with eIF4Es from plants and S. cerevisiae, coupled with deletion analyses of eIF4Es from S. cerevisiae and Danio rerio, suggest that the N- and C-termini of eIF4E are dispensable for translation and that the core of eIF4E represented by ~170 amino acids (from His-37 to His-200 in H. sapiens eIF4E) is sufficient for binding to the cap-structure and to eIF4G and 4E-BPs [8,9]. However, the N- and C-termini may be involved in the regulation of eIF4E-activity [10,11] or affect the stability of the protein.
eIF4E-activity is regulated by the actions of eIF4E-binding proteins or 4E-BPs which share sequence similarity with the eIF4E-binding domain within the N-terminal region of eIF4G (reviewed [12]). 4E-BPs act as competitive inhibitors of eIF4E-eIF4G interaction [13-15]. Crystal structures of mouse eIF4E bound to 4E-BPs and fragments of eIF4G show that both proteins interact with eIF4E via a common mechanism involving a sequence with the consensus YxxxxLΦ (where Φ is a hydrophobic residue) [16]. Hyper-phosphorylation of 4E-BPs occurs following stimulation of the Akt/FRAP/TOR signal transduction pathway and results in a reduced affinity for eIF4E [17-19].
Studies of M. musculus eIF4E bound to either a fragment of eIF4G or 4E-BP1 have revealed that His-37, Pro-38, Val-69, Trp-73, Leu-131, Glu-132, and Leu-135 (numbers for H. sapiens eIF4E) interact with the eIF4E-binding regions within eIF4G and 4E-BPs [16]. Val-69 and Trp-73 are within a conserved sequence of the consensus (S/T)V(e/d)(e/d)FW (where the acidic residues are not completely conserved). Substitution of a non-aromatic amino acid for Trp-73 has been shown to disrupt the ability of eIF4E to interact with eIF4G and 4E-BPs [20,21]. Substitution of a Gly residue in place of Val-69 creates an eIF4E variant that binds still binds 4E-BP1 but has a reduced capacity to interact with both eIF4G and 4E-BP2 [21].
eIF4E is ubiquitously expressed and is generally isolated from cell extracts using m7GTP-affinity matrices. Use of such matrices led to the conclusion that in mammalian cells, eIF4E was represented by a single polypeptide of ~25 kDa. Similar chromatographic resolution of proteins from plant cell extracts suggested that plants differ from mammalian cells in that they contain two different but related proteins termed plant eIF4E and eIF(iso)4E, or p26 and p28 (in reference to their apparent molecular weights as judged by SDS-PAGE) [22,23]. A gene encoding eIF4E from S. cerevisiae was isolated and shown by southern analyses and gene disruption studies to be the sole eIF4E gene in that organism [24], a conclusion confirmed by the availability of the sequence of the complete genome. S. cerevisiae lacking a functional eIF4E-gene can be rescued by exogenous expression of mammalian eIF4E showing that S. cerevisiae and mammalian eIF4Es are structurally and functionally comparable [24]. Overall, these findings suggested that, with the exception of plants, organisms contain a single gene that encodes eIF4E.
Growing evidence from genome/EST sequencing projects has revealed that many organisms contain multiple genes encoding proteins that have sequence similarity to recognized, or prototypical, eIF4E proteins such as mammalian eIF4E (reviewed in [25-27]). Consequently, the translation factor eIF4E and its relatives comprise a family of structurally related proteins within a particular organism. To distinguish the recognized vertebrate eIF4E from its relatives, vertebrate eIF4E has since been renamed eIF4E-1 [28] (or eIF4E-1A [9]). The functions of the eIF4E-related proteins are not yet understood. Some may act as translation factors and stimulate global mRNA recruitment, or specifically the recruitment of a subset of mRNAs [29,30]. Others may possess only partial activities when compared to prototypical eIF4Es [26,31] and thus act as inhibitors of mRNA recruitment. Lack of detection of these eIF4E related proteins in fractions derived from cell extracts resolved by m7GTP-affinity chromatography may reflect a variety of underlying causes. The proteins may be expressed ordinarily at low levels, at specific developmental times, or in restricted and untested tissues [9,26,30,32]. Alternatively, they may be unresolved from eIF4E-1 in fractionation by standard polyacrylamide gel electrophoresis. Conversely, they may fail to interact stably with [26], or recognize structures that differ from, the m7GTP-cap-structure [28,33,34] preventing isolation using standard m7GTP-affinity resins.
At the time of writing, BLAST search of the NCBI GenBank NR database using the amino acid sequence of M. musculus eIF4E-1 as a probe recovers <100 unique cDNA sequences of eIF4E-family members with expected values below 14. Only some of these are recognized as eIF4E-family members in the GenBank database. Additional eIF4E-family member sequences can be uncovered from genomic sequences, although these predicted sequences are subject to errors arising from less than adequate predictions of intron/exon boundaries. Through mining of the GenBank dbEST database and assembling sequences of overlapping cDNA fragments to derive consensus cDNA sequences, as well as performing reiterative searches, we have been able to extend the number of identified complete or partial cDNA sequences encoding eIF4E-family members to 379 (derived from 204 taxonomic species). A further 32 eIF4E-family members from 26 additional species can be predicted from the genomic sequences of organisms known to lack or contain few introns in genes transcribed from RNA polymerase II promoters. The sequences of identified eIF4E-family members have been deposited in an internet-accessible database designed for sequence analyses of eIF4E-family members [35]. Analyses of the sequences suggest that the eIF4E-structure has been duplicated numerous times during evolution producing new forms of the protein that may serve other tasks or regulate the activities of the prototypical translation factor.
Results and discussion
Definition of the amino acid core of an eIF4E-family member
Alignment of the complete amino acid sequences of the bonafide translation initiation factors M. musculus eIF4E-1, D. rerio eIF4E-1A, D. melanogaster eIF4E-1, T. aestivum eIF4E and eIF(iso)4E, Schizosaccharomyces pombe eIF4E1, and S. cerevisiae eIF4E, suggests the presence of an evolutionarily conserved "core" region (Figure 1). The region stretches approximately 160–170 residues from His-37 to His-200 of H. sapiens and M. musculus eIF4E-1. Evidence supporting the designation of this region as a functional core comes from deletion analyses [8,9]. The variant of D. rerio eIF4E-1A, eIF4E-1A(Δ1–33), lacking sequences N-terminal to within one amino acid of the conserved core region, is able to rescue the growth of S. cerevisiae lacking a functional eIF4E-gene [9]. Also, the variant of M. musculus eIF4E-1, eIF4E-1(Δ1–35), lacking N-terminal residues up to Lys-36, can bind both the cap-structure, the N-terminal fragment of eIF4G, and 4E-BPs in vitro [16]. Similarly, the S. cerevisiae eIF4E variants eIF4E(Δ1–29/Δ207–213) and eIF4E(Δ201–213) remain active in their ability to rescue the growth of S. cerevisiae lacking a functional eIF4E-gene and to bind the cap-structure in vitro [8]. As evident in Figure 1, regions N- and C-terminal to the defined core are not conserved in all bonafide eIF4E proteins. These regions may be involved in the regulation of eIF4E-activity [10,11] or may affect the stability of the protein. The N-terminal residues of S. cerevisiae eIF4E interact have been found to interact with eIF4G and stabilize the interaction [10]. The C-terminal residues of bonafide eIF4Es from Metazoa and S. pombe, but not of Viridiplantae possess a phosphorylatable Ser residue (Ser-209 of H. sapiens eIF4E-1) which in mammalian eIF4E has been shown to change the binding affinity of eIF4E for the cap structure [11,36]. The consensus sequence of the conserved core region suggests characteristics for a protein to be defined as a member of the eIF4E-family. Aromatic residues Trp, Phe, and His show a distinctive pattern across from N- to C-terminus summarized by H(x5)W(x2)W(x8–12)W(x9)F(x5)FW(x20)F(x7)W(x10)W(x9–12)W(x34–35)W(x32–34)H.
Figure 1 An alignment of the amino acid sequences of selected established eIF4E-family members. An alignment of the complete amino acid sequences of H. sapiens eIF4E-1, M. musculus eIF4E-1, X. laevis eIF4E-1A, D. rerio eIF4E-1A, D. melanogaster eIF4E-1a, T. aestivum eIF4E and eIF(iso)4E, S. pombe eIF4E1, and S. cerevisiae eIF4E. eIF4E-family members with names in blue indicate that the sequence was estimated or verified using genomic sequence data. A sequence of identity is shown with aromatic residues boxed in red. Black and grey shading: conserved amino acids identical in all or similar in greater than 75 % of the sequences shown, respectively. Yellow shading: His-residues that border the conserved core region of an eIF4E-family member. Blue shading: regions of the respective eIF4E-family member that have been shown to be dispensable for eIF4E-function in vitro. Residues in green: positions of residues equivalent to Trp-56, Trp-102, Glu-103 and Trp-166 of H. sapiens and M. musculus eIF4E-1 that directly interact with the cap-structure. Residues in purple: identity with respect to residues Val-69 and Trp-73 of M. musculus eIF4E-1 that interact with eIF4G and 4E-BPs and are found within a region of eIF4E-family members possessing the concensus (S/T)VxxFW (as indicated). Numbers to the right of the sequences indicate the positions of residues from the N-terminal Met.
Acquisition of nucleotide sequences encoding putative eIF4E-family members
In order to obtain nucleotide coding sequences representing an accurate description of the repertoire of functional genes encoding eIF4E-family members within an organism it was decided that some evidence should support the expression of the eIF4E-family member for which sequence is obtained. Consequently, for the most part, sequences of expressed sequence tags (ESTs) were acquired. In general, direct use of genome sequences was avoided due to the possibility of including pseudogenes and the possible inaccuracy with which intron/exon boundaries can be predicted. However, where sufficient EST data to verify nucleotide sequences encoding the core region of an eIF4E-family member was absent, it was considered reasonable to use genome sequences for confirmation. Furthermore, the use of genome sequences was considered valid for organisms whose genomes are known to lack, or contain few, introns in genes transcribed by RNA polymerase II such as some Protista and yeasts. In such cases, genome sequences were used only if sequences indicated that indeed no introns were present in the gene representing an eIF4E-family member and that expressed cDNAs could be identified in the same, or closely related organisms.
Expressed nucleotide sequences encoding putative eIF4E-family members were acquired from GenBank NR and dbEST databases by using the nucleotide and amino acid sequences encoding M. musculus eIF4E-1, eIF4E-2, and eIF4E-3, T. aestivum eIF4E and eIF(iso)4E, A. thaliana nCBP, C. elegans IFE-1, 2, 3, 4, and 5, and S. cerevisiae eIF4E as probes for BLAST searches. Sequences encoding putative eIF4E-family members were easily identified by comparison of computed translations and the consensus pattern for the conserved core region described above. The retrieved eIF4E-related sequences were used to re-probe the databanks to retrieve further sequences of overlapping cDNA fragments from the same species or to obtain sequences from additional species. The process of iteration was continued to obtain sequences encoding more eIF4E-family members. Genomic sequences from organisms known to contain few introns in genes transcribed from RNA pol II promoters were also probed in a similar manner. In all 2,383 nucleotide sequences were collected representing nucleotide sequences from 230 species. The statistics of sequence acquisitions and alignments are presented in Table 1 and 2.
Table 1 Overall statistics of the dataset of nucleotide sequences encoding eIF4E-family members
Databank from which sequences were acquired Number of sequences Number of species Number of eIF4E-family members
GenBank1 dbEST2 2,237 191 356
GenBank NR3 80 32 59
GenBank Genomic4 53 33 42
Genome Projects5 13 1 1
Total6 2,383 230 411
1GenBank: the NIH genetic sequence database. 2GenBank database of EST sequences (dbEST). 3mRNA sequences derived from the GenBank database (excluding those within the dbEST databank and any predicted from genomic sequences). 4Predicted mRNA sequences derived from genomic sequences deposited in the GenBank database. 5Sequences predicted from genomic DNA assemblies not yet submitted to GenBank. 6The total dataset (accessible via the internet at "The eIF4E-family member database" [35]).
Table 2 Identification and verification of nucleotide sequences encoding eIF4E-family members
Region within the coding sequences of an eIF4E-family member Number of eIF4E-family members
Identified Verified4
Start Codon1 278 155
Stop Codon2 259 149
Complete coding sequence 200 105
Sequence encoding the entire core region3 243 120 (142 >90%)
Total dataset (any region identified) 411 NA
1Initiation of translation was assumed to occur at ATG codons. 2The first of any of UGA, UAG, and UAA codons 3' and in frame with core regions were assumed to be translational stop codons. 3Core regions defined as regions including codons representing equivalents of H. sapiens eIF4E-1 residues His-37 to His-200. 4Number of eIF4E-family members for which the sequence or region indicated could be verified by two or more sequence reads.
Nucleotide sequences encoding an eIF4E-family member from a particular species were aligned to produce complete or partial consensus cDNA sequences. In many cases the initiation and/or stop codons could not be accurately identified because of a lack of overlapping clones to remove sequence errors or a lack of clones representing full length cDNA products. However, sufficient sequence information was usually available to identify the complete core regions. Analyses of consensus sequences encoding 220 representative eIF4E-family members from 118 species of eukaryotes are presented.
The eIF4E-family of proteins
Dissection of the dataset by taxonomic criteria is presented in Table 3. The majority of acquired sequences represented eIF4E-family members from Metazoa (Animalia), Viridiplantae and Fungi. Comparison of the number of eIF4E-family members defined within a particular taxonomic division and the number of species the members represented suggests that many organisms contain two or more eIF4E-family members. A radial view of a cladogram derived from an alignment of the nucleotide sequence representing conserved core regions of selected eIF4E-family members from >100 species of Viridiplantae, Metazoa, Fungi, and Protista is presented in Figure 2. Based on branching and clustering, the eIF4E-family can be separated taxonomically into eight sub-groups consisting of 1) the metazoan eIF4E-1 and nematode IFE-3-like; 2) plant eIF4E and eIF(iso)4E-like; 3) fungal eIF4E-1-like; 4) metazoan eIF4E-2-like; 5) plant nCBP-like; 6) fungal nCBP/eIF4E-2-like; 7) metazoan eIF4E-3-like; and 8) a set of atypical eIF4E-family members found in certain protists. Additional eIF4E-family members have been found that fail to fall into any of these subgroups. These are not included in the analysis in Figure 2, but some are discussed in a later section. An alignment of representative members from the eight sub-groups are presented in Figure 3A and comparisons of identities and similarities between the amino acid sequences representing the core regions of selected eIF4E-family members from each of the eight sub-groups are presented in Figure 3B. Variations at residues equivalent to H. sapiens Trp-56 and at Trp-43 of eIF4E-1 (shown in green in Figure 3A) provide a convenient means to categorize the non-protist eIF4E-family members into three classes on a structural basis. With the exception of Trp-56, residues of H. sapiens and M. musculus eIF4E-1 that have been shown to directly interact with the mRNA-cap structure (Trp-56, Trp-102, Glu-103, and Trp-166) are identical in all classified eIF4E-family members for which corresponding residues could be identified.
Table 3 Dissection of dataset with respect to taxonomic divisions
Taxonomic division No. of eIF4E-family members identified1 No. of species represented2 eIF4E-family members/species
Metazoa (Animalia) 186 (186) 89 (89) 2.09
Viridiplantae 155 (155) 83 (83) 1.87
Fungi 42 (18) 36 (16) 1.17
Protista 28 (20) 22 (16) 1.27
Total 411 (379) 230 (204) 1.78
1The number of eIF4E-family members for which nucleotide coding sequences were identified. In parentheses: the number of eIF4E-family members identified excluding those based only on genomic sequences. 2The number of species (organisms) represented for which nucleotide sequences encoding an eIF4E-family member could be identified. In parentheses: the number of species represented excluding those for which genomic sequences were the only source for identification of an eIF4E-family member.
Figure 2 A radial cladogram describing the overall relationship of selected eIF4E-family members from multiple species. The topology of a neighbor-joining tree visualized in radial format derived from an alignment of nucleotide sequences representing the conserved core regions of the indicated eIF4E-family members. The full names of the species represented and the accession numbers for cDNA sequences used to derive consensus core sequences can be found within supplementary data to this publication. Alignments of cDNA sequences to derive consensus core sequences can be obtained and verified at the "eIF4E-family member database" [35]. eIF4E-family member names in black or red indicate whether or not the complete sequence of the conserved core region of the member could be predicted from consensus cDNA sequence data, respectively. eIF4E-family member names in blue indicate that genomic sequence data was used to either verify or determine the nucleotide sequence representing the core region of the member. The shape of a 'leaf' indicates the taxonomic kingdom from which the species containing the eIF4E-family member derives: Metazoa (diamonds); Fungi (squares); Viridiplantae (triangles); and Protista (circles); respectively. The color of a 'leaf' indicates the sub-group of the eIF4E-family member: metazoan eIF4E-1 and IFE-3-like (red); fungal eIF4E-like (gold); plant eIF4E and eIF(iso)4E-like (green); metazoan eIF4E-2-like (cyan); plant nCBP-like (blue); fungal nCBP/eIF4E-2-like (purple); metazoan eIF4E-3-like (pink); atypical eIF4E-family members from some protists(white). eIF4E-family members within structural classes Class I, Class II, and Class III are indicated. Bootstrap values of greater than 60 % derived from 50,000 tests are shown.
Figure 3 Comparison of the conserved cores of eIF4E-family members from different taxonomic sub-groups. A. An alignment of amino acid sequences representing the conserved core regions of the indicated eIF4E-family members. Sequence names are highlighted to indicate structural class: Class I in blue; Class II in green; and Class III in red. Atypical eIF4E-family members that could not be accurately classified based on similarity to other structural class members are shown with sequences names in black. Symbols to the left indicate the taxonomic sub-group of the eIF-4E-family member (as described in the legend to Figure 2). Residues highlighted within the amino acid alignment represent: identity with respect to residues Trp-43, Trp-56, Trp-102, Glu-103, and Trp-166 within H. sapiens eIF4E-1 (green); identity within the conserved (S/T)VxxFW consensus region containing amino acids equivalent to Val-69 and Trp-73 of H. sapiens eIF4E-1(purple); identity with His-residues equivalent to those that border the core region of H. sapiens eIF4E-1 (shaded in yellow). Variations at residues equivalent to Trp-43 and Trp-56 of H. sapiens eIF4E-1 are indicated as follows: Tyr/Phe-shaded in blue with white text; Cys-shaded in red with white text. Residues shaded in black or grey within the alignment indicate amino acids that are identical in all sequences or similar in greater than 85% of the sequences, respectively. Numbers to the right of the alignment represent distances of amino acids with respect to the predicted N-terminal Met residue. B. Identities and similarities (based on a PAM 250 matrix [58]) between the amino acid sequences representing the core regions of selected eIF4E-family members from each of the eight sub-groups.
Class I eIF4E-family members
Structural Class I of the eIF4E-family members include of members of sub-groups 1, 2, and 3 (Figure 2). cDNAs encoding members of structural Class I can be identified in species from Viridiplantae, Metazoa, and Fungi. As judged from completed genomes, many protists also encode Class I-like family members (data not shown). The Class I family includes orthologues of the prototypical eIF4Es described for H. sapiens (eIF4E-1), M. musculus (eIF4E-1), T. aestivum (eIF4E and eIF(iso)4Es), and S. cerevisiae (eIF4E) (reviewed [1,3]). Comparisons of the amino acid sequences representing the core regions of selected members from each of the sub-groups 1, 2, and 3, reveal that they share ~35–40 % identity and ~60–65% similarity with one another (Figure 3B). Alignments and the relationships of selected representative Class I eIF4E-family members are shown in Figures 4, 5, and 6. All identified members of structural Class I (subgroups 1, 2 and 3) possess Trp residues at the equivalents of Trp-43 and Trp-56 of H. sapiens eIF4E-1 and include metazoan eIF4E-1s plant eIF4E(p26) and eIF(iso)4E(p28), and fungal eIF4E. Examination of sequences of members from each of the sub-groups reveals evidence of gene duplications.
Figure 4 Comparison of the conserved core regions of selected Class I eIF4E-family members from Viridiplantae. A. An alignment of the amino acid sequences representing the 'core' regions of Class I eIF4E-family members from the indicated species of Viridiplantae and of eIF4E-1 from H. sapiens. Amino acid residues within the alignment are highlighted as described in the legend to Figure 3A with the exception that residues shaded in grey indicate similar amino acids in more than 90% of the sequences shown. Numbers to the right of the alignment represent distances of amino acids with respect to the N-terminal Met residue (black) or, for eIF4E-family members for which the N-terminal Met could not be predicted, from the first residue shown (red). B. A phylogram constructed by neighbor-joining derived from alignments of nucleotide sequences representing the core regions of the indicated Class I-family members. Bootstrap values greater than 70% derived from 50,000 tests are shown to indicate supported nodes. For A and B: names of eIF4E-family members are highlighted to indicate taxonomic divisions: Eudicotyledons (blue), Liliopsida (green), Bryopsida (purple), Coniferopsida (red), Stem Magnoliophyta (cyan), Magnoliids (orange), Chlorophyceae (black), Mammalia (white on black). Names of family members and residues shaded in cyan indicate evidence that a gene-duplication occurred prior to speciation.
Figure 5 Comparison of the conserved core regions of selected Class I eIF4E-family members from Metazoa. A. An alignment of the amino acid sequences representing the 'core' regions of Class I eIF4E-family members from the indicated species of Metazoa. Amino acid residues within the alignment are highlighted as described in the legend to Figure 3A with the exception that residues shaded in grey indicate similar amino acids in more than 95% of the sequences shown. Numbers to the right of the alignment represent distances of amino acids with respect to the N-terminal Met residue (black) or, for eIF4E-family members for which the N-terminal Met could not be predicted, from the first residue shown (red). B. A phylogram constructed by neighbor-joining derived from alignments of nucleotide sequences representing the core regions of the indicated Class I-family members. Bootstrap values greater than 70% derived from 50,000 tests are shown to indicate supported nodes. For A and B: names of eIF4E-family members highlighted in blue indicate that genomic sequence from the indicated species was employed to verify and predict the amino acid sequence of the eIF4E-family member.
Figure 6 Comparison of the conserved core regions of Class I eIF4E-family members from species of Nematoda. A. An alignment of amino acid sequences representing the conserved core regions of Class I eIF4E-family members from the species of Nematoda indicated and of H. sapiens eIF4E-1. Amino acid residues within the alignment are highlighted as described in the legend to Figure 3A with the following exceptions: residues shaded in black indicate amino acids identical in all eIF4E-family members with respect to regions that could be predicted; residues shaded in grey indicate amino acids identical in all eIF4E-family members from nematoda with respect to regions that could be predicted that differ from equivalent residues in H. sapiens eIF4E-1. Numbers to the right of the alignment represent distances of amino acids with respect to the N-terminal Met residue (black) or, for eIF4E-family members for which the N-terminal Met could not be predicted, from the first residue shown (red). B. A phylogram constructed by neighbor-joining derived from an alignment of nucleotide sequences representing the conserved core regions of the eIF4E-family members indicated. Bootstrap values greater than 70% derived from 50,000 tests are shown to indicate supported nodes. For A and B: names of eIF4E-family members in red indicate that only a portion of the conserved core region could be predicted.
Evidence supporting the presence of two distinct Class I sub-group 2 eIF4E-family members represented by viridiplantae eIF4E and eIF(iso)4E, can be found in species from the viridiplantae classes Liliopsida and Eudicotyledons. Sequences representing either one of eIF4E or eIF(iso)4E can be identified in other viridiplantae classes suggesting that the genes arose from an earlier duplication event. The two forms are closely related in sequence (Figure 4A and 4B) and each possesses all the activities attributed to mammalian eIF4E-1 in vitro [23,37]. However, expression of A. thaliana eIF(iso)4E in S. cerevisiae lacking a functional endogenous eIF4E-gene results in slower growth relative to similar expression of A. thaliana eIF4E [38]. In addition, levels of expression of A. thaliana eIF4E and eIF(iso)4E differ in various A. thaliana tissues; eIF4E is expressed ubiquitously (with the exception of tissues in the zone of specialization of the root); eIF(iso)4E is expressed more abundantly in developing tissues [38]. Subtle differences in their relative activities can be inferred from the requirements for each in potyvirus infected cells. Both plant eIF4E and eIF(iso)4E bind to potyviral genome-linked proteins (Vpgs) [39]. However, strains of A. thaliana lacking eIF(iso)4E, or of P. sativum carrying variants of eIF4E which lack cap-binding ability lose susceptibility to potyvirus infection [40,41].
In certain plant species from Eudicotyledons, Liliopsida and Coniferospida, multiple forms of eIF4E and eIF(iso)4E can be found. Instances of apparent gene duplication can be seen in species such as Zea mays and Triticum aestivum with respect to eIF4E (p26) and eIF(iso)4E (p28). However, there is no evidence to support the hypothesis that these duplications occurred prior to speciation. In contrast, evidence suggesting gene duplication prior to speciation can be found (compare eIF4E-family member names and residues shaded in cyan in Figure 4A and 4B) in the Solanaceae, in which Lycopersicon esculentum (tomato) and Solanum tuberosum (potato) both have two forms of eIF4E(p26) (A and B), and in Salicaceae which have two forms of eIF(iso)4E(p28) (A and B).
Evidence for gene duplication of Class I eIF4E-family member genes eIF4E-1 can also be found in Metazoa with respect to Chordata, Insecta and Nematoda. Three eIF4E-1 sub-family members can be identified from the zebrafish Danio rerio, termed eIF4E-1A, eIF4E-1B [9] and eIF4E-1C (Figure 5A and 5B). Orthologues of the gene encoding eIF4E-1B, but not that of eIF4E-1C, can be found in almost all species above Actinoptergyii for which sequence has been acquired. Both eIF4E-1A and eIF4E-1B possess similar levels of identity when compared to mammalian eIF4E-1s and possess all known residues required for interaction with the cap-structure, eIF4G, and 4E-BPs. eIF4E-1A, like mammalian eIF4E-1, is expressed ubiquitously and can restore the growth of S. cerevisiae lacking a functional eIF4E-gene [9]. It can bind to the cap-structure, eIF4G, and 4E-BP in vitro [9]. Conversely, eIF4E-1B is expressed only during early embryogenesis and in the gonads and muscles of adult fish, and is unable to complement yeast lacking eIF4E. Furthermore, eIF4E-1B is unable to bind to the cap-structure, eIF4G, or 4E-BP in vitro.
Drosophila melanogaster has a total of six genes, all of which are expressed, encoding Class I eIF4E-family members eIF4E-1a-f (also termed eIF4E-1, 4, 5, 3, 7, and 6, respectively [42]). There is also evidence of a seventh Class I eIF4E-family member (termed eIF4E-2 in [42]) that arises from alternate splicing of primary transcripts and shares the same core sequence as eIF4E-1a. Four of the genes share exon/intron structure in their carboxy-terminal regions and form a cluster in the genome. All Class I eIF4Es from D. melanogaster bind to cap-analogue. Furthermore, all of them, except eIF4E-1f (eIF4E-6 in [42]), which has a truncated carboxy-terminal domain, are able to interact with D. melanogaster eIF4G or 4E-BP. The expression of each has been shown to vary throughout the life cycle of the fly. Examination of both expressed sequences and partial or complete genome sequences of insect species has not revealed a similar repertoire of Class I eIF4E-family members outside the genus of Drosophila (data not shown).
An alignment of Class I eIF4E-family members from nematodes is presented in Figure 6A. Although only a single Class I eIF4E orthologue can be found in Ascaris suum, many nematodes express more than one Class I family member (Figure 6B). Four of the five C. elegans eIF4E-family members (termed IFEs for initiation factor of elegans), are Class I members. With respect to activities, IFE-3 corresponds to mammalian eIF4E-1 and binds only to mono-methylated cap-structures. However, in nematodes, a proportion of the mRNAs possesses a tri-methyl-cap arising from the post-transcriptional addition of a tri-methyl-cap containing spliced leader RNA (SLRNA) to the 5' end of a transcribed mRNA [43,44]. The translation of such trans-spliced mRNAs in C. elegans is thought to be mediated by IFE-1, 2, and 5 since they, unlike IFE-3, interact with both mono- and tri-methylated cap-structures [28,33]. IFE-1, 2, and 5 possess more similarity to IFE-3 in sequence than to Class I family members from other phyla of Metazoa suggesting they arose from gene-duplications of a progenitor IFE-3 (Figure 2 and 6B). Evidence in support of this hypothesis comes from recent studies of the only identified IFE-3-like protein from the nematode Ascaris suum. A. suum IFE-3 (also termed eIF4E-3) can bind and stimulate the translation of mRNAs possessing mono- or tri-methylated cap structures [45] in vitro. Furthermore, identified sequences from some nematodes, such as the parasitic Haemanchus contortus suggest that they express single form of eIF4E similar to IFE-3 and a single form related to IFE-1, -2 or -5. No direct relatives of IFE-1, -2 and -5-like proteins have been found in any taxonomic group other than Nematoda.
Multiple Class I eIF4E-family members can also be found in the fission yeast Schizosaccharomyces pombe. S. pombe expresses two eIF4E-family members that share 52 % identity termed eIF4E1 and eIF4E2 [46]. Unlike eIF4E1, expression of eIF4E2 is not essential. However, levels of eIF4E1 and eIF4E2 vary with growth temperature and at higher temperatures eIF4E2 is more abundant than eIF4E1. Both eIF4E1 and eIF4E2 can bind the cap-structure with similar affinity in vitro but eIF4E1 has a 100-fold greater affinity for S. pombe eIF4G than eIF4E2 despite the fact that both share all known amino acids required for interaction with eIF4G. No evidence supporting the expression of two genes directly related to those for S. pombe eIF4E1 and eIF4E2 has been found.
Class II eIF4E-family members
Structural Class II members include eIF4E-2-family members from Metazoa (sub-group 4) and nCBP-family members from Viridiplantae (sub-group 5). Class II eIF4E-family members can also be recognized in pathogenic fungi from the sub-phylum Pezizomycotina, including Coccidioides posadasii and Sclerotinia schleroiorum (sub-group 6, Figure 2) but are absent in the model ascomycetes, S. cerevisiae and S. pombe. An alignment and the relationships of representative Class II eIF4E-family members is presented in Figure 7A and 7B, respectively. Comparisons of amino acid sequences representing the core regions of selected sub-group 4, 5, and 6 eIF4E-family members show that they share approximately 50 % identity and 70–80 % similarity with one another (Figure 3B). The members also posses ~30–35 % identity and 60–65 % similarity to Class I eIF4E-family members of sub-groups 1, 2, and 3. Like Class I eIF4E-family members, Class II eIF4Es also share a structural core of approximately ~160–170 amino acids. All identified members of this class differ from Class I eIF4E-family members in that they possess a hydrophobic residue such as Tyr, Phe, or Leu, but not Trp, in the position equivalent to Trp-56 of H. sapiens eIF4E-1. All identified Class II members from Metazoa and Viridiplantae also contain a substitution at the position equivalent to Trp-43 of H. sapiens eIF4E-1. However, this substitution has not so far been seen in the few Class II members identified in Fungi.
Figure 7 Comparison of the conserved core regions of selected Class II eIF4E-family members. A. An alignment of amino acid sequences representing the conserved core regions of the Class II eIF4E-family members from the taxonomic species indicated and of H. sapiens eIF4E-1. Amino acid residues within the alignment are highlighted as described in the legend to Figure 3A with the exception that residues shaded in grey indicate identical amino acids in greater than 84% of the sequences shown. Numbers to the right of the alignment represent distances of amino acids with respect to the N-terminal Met residue (black) or, for eIF4E-family members for which the N-terminal Met could not be predicted, from the first residue shown (red). B. A phylogram constructed by neighbor-joining derived from an alignment of nucleotide sequences representing the conserved core regions of the eIF4E-family members indicated. Bootstrap values greater than 70% derived from 50,000 tests are shown to indicate supported nodes. For A and B: names of eIF4E-family members in red indicate that only a portion of the conserved core region could be predicted.
Studies have shown the Class II eIF4E-related proteins eIF4E-2A (H. sapiens and M. musculus; also termed eIF4E-2, 4EHP or 4E-LP), IFE-4, (C. elegans), D. melanogaster eIF4E-2 (eIF4E-8 in [42]), and nCBP (A. thaliana), like the mammalian translation factor eIF4E-1, all bind the m7GTP-cap structure [26,29,31,42]. nCBP from A. thaliana differs from H. sapiens and Mus musculus eIF4E-2A and D. melanogaster eIF4E-2 in that A. thaliana nCBP can interact with eIF4G and participate in productive translation [29], whereas H. sapiens and M. musculus eIF4E-2A and D. melanogaster eIF4E-2 cannot [26,42]. Consistent with this observation, M. musculus eIF4E-2A and D. melanogaster eIF4E-2 cannot substitute for S. cerevisiae eIF4E in a strain lacking a functional eIF4E-gene [26,42]. Recent studies have shown that H. sapiens and M. musculus eIF4E-2A can interact with 4E-BPs but to a lesser degree than mammalian eIF4E-1 [26,47]. Although mammalian eIF4E-2A mRNA appears to be expressed in all tissues, the levels of M. musculus eIF4E-2A protein are ~10-fold lower than eIF4E-1 [31].
Given that metazoan eIF4E-2 cannot itself partake in protein synthesis due to its inability to interact with eIF4G, growing evidence suggests a regulatory role for metazoan eIF4E-2 family members. Like mammalian eIF4E-2A, D. melanogaster eIF4E-2 (eIF4E-8 in [42]) appears to expressed at much lower levels than the major Class I form, eIF4E-1a (eIF4E-1 in [42]), although it is present at all stages of the life cycle. Mutants of D. melanogaster that express a markedly reduced level of eIF4E-2 show defects in anterior-posterior axis formation during early embryogenesis [48]. Development of the anterior-posterior axis in D. melanogaster embryos is dependent on the distribution of the maternal effect genes which include bicoid and caudal. In the oocyte, caudal mRNA is evenly distributed, whereas bicoid mRNA is restricted to the anterior of the cell. Translation of bicoid mRNA is activated upon fertilization resulting in a gradient of bicoid protein decreasing toward the posterior of the embryo. Through interaction of bicoid with a region within the 3' UTR of caudal mRNA, the translation of caudal mRNA is inhibited resulting in an opposing gradient of caudal expression. Evidence suggests that D. melanogaster eIF4E-2 binds specifically to a region of bicoid resembling the eIF4E-binding region within eIF4G. This suggests that inhibition of caudal mRNA translation is due to sequestration of the caudal mRNA into a inactive 'circular' complex with which eIF4E-1 and ribosomes cannot interact. Such a mechanism of translational regulation through eIF4E-2 may not be restricted to D. melanogaster. The nematode Class II representative, IFE-4, is expressed in C. elegans in pharyngeal and tail neurons, body wall muscle, spermatheca and vulva, suggesting a special use [30]. Reduction of IFE-4 expression by RNA-interference or introduction of a null mutation produces a pleiotropic phenotype that includes an egg laying defect. Microarray analyses of mRNAs translated in the absence of IFE-4 expression suggest that IFE-4 is required for translation of a subset of mRNAs [28,30]. In mammals, expression from the H. sapiens eIF4E-2A gene (EIF4EL3) is upregulated following conversion of primary solid tumors to associated metastases [49] further suggesting a regulatory role for this protein.
Evidence for the expression of two distinct sub-forms of Class II eIF4E-family members can be recognized in the Actinopterygii (D. rerio, O. mykiss, and T. rubripes) and Amphibia (A. mexicanum, X. laevis and X. tropicalis) (see species containing both eIF4E-2A and eIF4E-2B in Figure 2, and data not shown). The actinopterygian and amphibian sub-forms termed eIF4E-2A represent orthologues of mammalian eIF4E-2A. The sub-forms termed eIF4E-2B are ~85–90 % identical within the core regions to eIF4E-2A sub-forms from the same species. Examination of the genomes of H. sapiens and M. musculus fails to reveal genes corresponding to a mammalian eIF4E-2B sub-form.
Class III eIF4E-family members
An alignment of identified Class III proteins (sub-group 7 in Figure 2) and their relationships is presented in Figure 8A and 8B. Class III members share ~25–30% identity and ~45–55% similarity with members from Class I and Class II (Figure 3B and data not shown). Structural Class III members possess a Cys (in Vertebrata) or Tyr residue at the position equivalent to Trp-56 of H. sapiens eIF4E-1. Unlike Class II eIF4E-family members from Viridiplantae and Metazoa, but similar to all Class I members and Class II members from Fungi, all identified members of Class III possess a Trp residue equivalent to Trp-43 of H. sapiens eIF4E-1. The Class III members recognized have only been identified in Metazoa. They are well represented in chordates from tunicates, agnathans, jawed fish and higher vertebrates. However, Class III eIF4Es are sporadically represented elsewhere in Metazoa, being found in Cnidaria (Hydra) and in some molluscs (Crassostrea virginica) and some insects and arachnids. No evidence supporting the expression of Class III eIF4E-family members was found in nematodes. In Insecta, Class III eIF4E-family members can be found in Pterygota, the winged insects, with representatives from Hemiptera (sharpshooter, Homalodisca coagulata) and Hymenoptera (honey bee, Apis mellifera). However, genes encoding Class III eIF4E-family members are absent in Diptera such as D. melanogaster or Anopheles gambiae. In Arachnida, a Class III eIF4E-family member can be identified in the Parasitiforms (brown ear tick, Rhiphicephalus appendiculatus). The origins of Class III may result from a early duplication of a Class II gene as suggested in Figure 2A. A fuller picture of its phylogenetic distribution and evolutionary relationships will depend on the accumulation of more sequence data from more non-chordate metazoa. However, it seems possible that a progenitor Class III eIF4E-gene evolved early in metazoan evolution, but was subsequently lost in some groups.
Figure 8 Comparison of the conserved core regions of selected Class III eIF4E-family members. A. An alignment of amino acid sequences representing the conserved core regions of the Class III eIF4E-family members from the taxonomic species indicated and of H. sapiens eIF4E-1. Amino acid residues within the alignment are highlighted as described in the legend to Figure 3A. Numbers to the right of the alignment represent distances of amino acids with respect to the N-terminal Met residue (black) or, for eIF4E-family members for which the N-terminal Met could not be predicted, from the first residue shown (red). B. A phylogram constructed by neighbor-joining derived from an alignment of nucleotide sequences representing the conserved core regions of the eIF4E-family members indicated. Bootstrap values greater than 70% derived from 50,000 tests are shown to indicate supported nodes. For A and B: names of eIF4E-family members in red indicate that only a portion of the conserved core region could be predicted.
Class III eIF4E-family members from Vertebrata possess a non-aromatic Cys residue at the position equivalent to Trp-56 of H. sapiens eIF4E-1. Since Trp-56 together with Trp-102 of H. sapiens eIF4E-1 partakes in π-bond stacking interactions with the guanine base of the cap-structure, the ability of vertebrate eIF4E-3 to interact with the cap structure would be unexpected. However, studies with M. musculus eIF4E-3 have shown that the protein does interact with the cap-structure in vitro suggesting that stacking of only one aromatic residue is sufficient for cap-interaction [26]. However, the interaction of M. musculus eIF4E-3 with the cap-structure is weaker than that of either mammalian eIF4E-1 or eIF4E-2. Furthermore, M. musculus eIF4E-3 is less able to distinguish between 7-methylated and non-methylated GTP. M. musculus eIF4E-3 can interact with eIF4G but not with 4E-BPs suggesting that it may participate in translation. However, M. musculus eIF4E-3 is unable to rescue the growth of S. cerevisiae lacking a functional eIF4E-gene. The weaker interaction of M. musculus eIF4E-3 with the cap-structure relative to eIF4E-1 suggests that this protein may be involved in sequestration of eIF4G resulting in inhibition of cap-dependent translation. The distribution of eIF4E-3 in adult mice differs from that of both eIF4E-1 and eIF4E-2, with highest levels of expression in skeletal muscle.
eIF4E-family members from some species of Protista show extension or compaction relative to Class I, II, and III eIF4E-family members
The amino acid sequences representing the core-regions or complete sequences of some of the identified eIF4E-family members from several unicellular eukaryotes are presented in Figure 9A and 9B. Examination of the sequences reveal that they differ significantly from typical Class I, Class II, or Class III eIF4E-family members.
Figure 9 eIF4E-family members from some species of Protista show extension or compaction. A. An alignment of amino acid sequences representing the conserved core regions of eIF4E-family members from Alveolata, Stramenopiles, the Haptophyceae E. huxleyi and of H. sapiens eIF4E-1, and M. musculus eIF4E-2A and eIF4E-3. Green boxes indicate amino acids extensions relative to Class I, II, or III eIF4E-family members from other species. B. An alignment of the complete predicted amino acid sequences of predicted eIF4E-family members from C. merolae, G. theta nucleomorph, and E. cuniculi, and from H. sapiens eIF4E-1, and M. musculus eIF4E-2A and eIF4E-3. Residues shaded in light blue indicate regions N- and C- terminal to the conserved core of the respective eIF4E-family member. Residues shaded in greenindicate variations at positions equivalent to Val-69 and Trp-73 of H. sapiens eIF4E-1. For both A and B: amino acid residues within the alignment are highlighted as described in the legend to Figure 3A with the exception that residues shaded in grey indicate amino acids similar in greater than 80% (A) or 70% (B) of the sequences shown. Numbers to the right of the alignments represent distances of amino acids with respect to the N-terminal Met residue (black) or, for eIF4E-family members for which the N-terminal Met could not be predicted, from the first residue shown (red). eIF4E-family members for which names are shown in red indicate that only a portion of the core region for that member could be estimated. eIF4E-family members for which names are shown in blue indicate that sequences were predicted using genomic sequence data.
eIF4E-family members of the Alveolata, Stramenopiles, and Haptophyta (presented as those of sub-group 8 in Figure 2) possess ~20 % identity and ~40 % similarity with respect to the core regions of eIF4E-family members from Class I, II and III (Figure 3B, and data not shown). Members of sub-group 8 possess either a Trp or Tyr residue equivalent to Trp-56 of H. sapiens eIF4E-1 and a Trp at the residue equivalent to H. sapiens Trp-43 (Figure 9A). Consequently, sub-group 8 members have Class I, Class II (fungal), or Class III-like signatures. However, members of sub-group 8 also possess extended stretches of 12–15 amino acids between residues equivalent to Trp-73 and Trp-102 of H. sapiens eIF4E-1, and 4–9 amino acids between residues equivalent to Trp-102 and Trp-166. Such stretches, the purpose of which are not known, are not seen in any family members of Class I, II and III and suggest that sub-group 8 members have specialized functions relative to those of Class I, II and III eIF4Es. Consequently, subgroup 8 members could be considered a fourth Class of eIF4E-family member.
A clue as to the possible role of extended sequences in the basic structure of eIF4E arises from the studies of eIF4E-family members from the trypanosome Leishmania major. Four intron-less eIF4E-family member genes can be identified from the known genomic sequences of L. major (data not shown). The sequences of two of these members (Leish4E-1 and Leish4E-2) are also presented in Figure 9A. Leish4E-1 and Leish4E-2 possess a Class I-like signature (Trp-residues at the equivalent of H. sapiens eIF4E-1 residues 43 and 56). Like family members of sub-group 8, both Leish4E-1 and Leish4E-2 contain extended amino acid stretches between structural units of the core, although the positions or lengths of the extensions differ from those found in sub-group 8 eIF4Es. In Trypanosomatidae, polycistronic pre-mRNA transcripts are processed to generate monocistronic RNAs which are further modified by the addition of capped spliced leader (SL) RNA. Unlike the tri-methylated SLRNAs from nematodes, the SLRNA of Trypanosomatidae possess mono-methylated cap-structures [50] that are further modified on the first four transcribed nucleotides by addition of 2'-O methyl groups resulting in all mRNAs containing so-called cap-4 structures [59]. The presence of cap-4 and/or the nucleotide sequence of the SLRNA is thought to be required for efficient recruitment of trans-spliced mRNAs into polysomes [51]. Studies in vitro have shown that L. major, LeishIF4E-1, which contains two areas of extended sequence between the structural units of the core, binds both m7GTP and the cap-4 structure with similar affinities [34]. This is in contrast to mammalian eIF4E-1 which possesses a 5-fold greater affinity for m7GTP compared to cap-4. Although trans-splicing has not been demonstrated in Alveolata, Stramenopiles, or Haptophyta, the presence of extensions within the core regions may signify that these eIF4E-family members, like Leish4E-1, recognize more complex cap-structures, specific modifications, or other sequences close to the cap-structure of a mRNA. This is of particular interest because many representatives of these sub-kingdoms are parasites or infectious agents.
Although reliant only on genomic data, specialization of a different nature can be seen in eIF4E-family members from the microsporidian, Encephalitozoon cuniculi, and of the algal endosymbiont of the cryptophyte Guillardia theta (Figure 9B). The diminutive genome of the G. theta algal endosymbiont (~0.55 Mb [52]) has undergone extreme compaction relative to the genomes of other Rhodophyta such as Cyanidioschyzon merolae which has a genome of ~16.5 Mb [53]. The E. cuniculi genome is also highly compacted at ~2.9 Mb [54]. Consistent with compaction, predicted Class I-like eIF4E-family members from both possess short N-and/or C-termini relative to eIF4E-family members from other species.
Trp-73 of H. sapiens eIF4E-1 has been shown to be involved in the interaction of eIF4E-1 with eIF4G and 4E-BPs. With the exception of D. melanogaster eIF4E-1d, all members of the three defined non-protist structural classes of eIF4E-family members possess a Trp-residue equivalent to Trp-73 of H. sapiens eIF4E-1. Examination of the eIF4G-binding regions of eIF4E-family members from both E. cuniculi and the G. theta nucleomorph reveals differences relative to H. sapiens eIF4E-1. In both cases, residues equivalent to Trp-73 of H. sapiens eIF4E-1 are substituted by a non-aromatic Leu residue. As discussed earlier, such a substitution in H. sapiens eIF4E-1 has been shown to impair the ability of eIF4E-1 to interact with either eIF4G or with 4E-BPs [20,21]. In the case of the G. theta algal endosymbiont this variation may not be remarkable since the genome of the nucleomorph appears to lack any identifiable sequence encoding eIF4G-like or 4E-BP-like proteins. Since the genome of the G. theta endosymbiont has been so severely compacted, it is hypothesized that the genes encoding complex cellular functions in the endosymbiont, such as protein synthesis, are limited to the minimal set needed to accomplish the function. Although the endosymbiont has its own mRNAs with 5'-caps and poly(A) tails, elongation and release factors, its genome only encodes a subset of translational initiation factors: eIF1, eIF1A, eIF4A, eIF2 (all subunits, although the alpha subunit is truncated), eIF4E, eIF6 and poly(A) binding protein. The genes for many initiation factors thought to be essential for cannot be identified including eIF4B, eIF5, or the scaffold protein eIF3 (any subunit).
Conclusion
eIF4E, a translational initiation factor found only in eukaryotes has a unique alpha/beta fold that is considered to have no homologues outside the eukaryotes, as determined by sequence comparison or structural analyses [55]. The expansion of sequenced cDNAs and genomic DNAs from organisms of all taxonomic kingdoms has significantly altered our picture of eIF4E which must now be considered to be one member of a structurally related family of eIF4E-like proteins.
Evolutionarily it seems that a single early eIF4E gene has undergone multiple gene duplications generating multiple structural classes. The functions of each member of each structural class remain to be completely understood. However, it is no longer possible to predict eIF4E-function from the primary amino acid sequence of an eIF4E-family member, as exemplified by the functional diversity of examples mentioned here and recently reviewed elsewhere [27]. The ancestral gene of eIF4E has also provided a blueprint for the generation of related proteins with specialized functions found only in certain taxonomic groups. The C. elegans IFE-1, -2 and -5 appear to result from gene duplications that occurred within early nematodes to give rise to a specialized sub-class that recognizes alternate cap structures. No direct relatives of these tri-methyl-cap binding proteins can be found in any other phyla. The extended eIF4Es of certain protists, like L. major have evolved independently to fulfill a similar function. These eIF4E variants seem likely to provide a rich source of variations on the eIF4E structural theme that will provide unique opportunities for structure/function studies and therapeutic drug design. As more sequence data becomes available and more eIF4E-family members are tested for their activities both in vitro and in vivo our understanding of the origins and functions of individual members will advance. The data provided here have been deposited in an internet accessible database for online access to assembled sequences encoding eIF4E-family members [35]. The site has been developed to allow easy searches, as well as sequence comparisons and other analyses.
Methods
Acquisition of cDNA sequences encoding eIF4E-family members
The nucleotide sequences encoding M. musculus eIF4E-1, eIF4E-2, and eIF4E-3, T. aestivum eIF4E and eIF(iso)4E, A. thaliana nCBP, C. elegans IFE-1, 2, 3, 4, and 5, and S. cerevisiae eIF4E were used to probe GenBank (NR), and dbEST databases for homologous cDNA sequences from other species through use of the BLAST 2.0 software package. In an iterative process, the retrieved sequences were used to re-probe the databanks to obtain further sequences of overlapping cDNA fragments from the same organism or to obtain related sequences from additional species. For budding yeasts and some protists, which lack, or possess few numbers of, introns in genes transcribed from RNA polymerase II promoters, genomic sequences encoding eIF4Es were also acquired from the GenBank database. The expression of many of these genomic sequences have been verified by the presence of at least one EST sequence for each.
Derivation of consensus cDNA sequences encoding eIF4E-family members
Overlapping nucleotide sequences encoding an eIF4E-family member from a particular species were aligned to produce complete or partial consensus cDNA sequences. In most cases, sequences within 3'-UTRs were used to verify that EST sequences described the same eIF4E-family member. However, due to usage of alternative splicing and/or of alternative polyadenylation sites, 3'-UTRs sometimes differed in sequence. Sequences were not considered verified unless a minimum of two sequences representing overlapping cDNA fragments confirmed the assignment of a single nucleotide. Where multiple nucleotides assignments for a particular base in a consensus sequence were supported by multiple sequences these variations were considered to represent polymorphisms or variations in strains. To allow for such variations, information about the strain used for development of the cDNA library from which an EST was recovered (provided by some submitters to GenBank databanks) was utilized to select a subset of sequences. Where strain information was not available, or where strain information still suggested multiple possibilities for an assignment, the assignment of a base supported by the majority EST sequences was chosen. In almost all cases these variations led to either no change at the amino acid level due to codon-degeneracy, or a single amino acid variation. Furthermore, such variations were either confined to regions encoding segments N-terminal of the core of the eIF4E-family member or had no affect on the prediction of amino acids in positions related to those known to be involved in the interaction of mammalian eIF4E-1 with the cap-structure, eIF4G, and 4E-BPs.
Alignments and analyses of sequences representing eIF4E-family members
Since alignments of mammalian eIF4E-1s, plant eIF4E and eIF(iso)4E and S. cerevisiae eIF4E suggest the presence of an evolutionarily conserved core region and because deletion analyses of S. cerevisiae eIF4E and D. rerio eIF4E-1A suggest that the N-termini and C-termini are dispensable with respect to cap-binding, eIF4G and 4E-BP interaction, sequences representing the core region of an eIF4E-family member were used for analyses unless otherwise stated. Amino acid alignments were performed using ClustalW (version 1.8) software and adjusted as necessary [56]. Alignments of the nucleotide sequences representing the core regions of eIF4E-family members (~480–510 nucleotides per member) were generated by reverse translation of amino acid alignments, and substitution of the factual nucleotide sequences. Concensus phylogenetic trees were generated from nucleotide alignments using the "neighbor-joining" and "boot-strapping" algorithms within Mega 3.0 [57]. In some cases, only a part of the nucleotide sequence representing the core region of an eIF4E-family member could be identified. Where indicated, partial sequences were included in the analyses. Unless stated, names assigned to eIF4E-family members were based on designations previously applied by investigators in the field of translational control.
Analyses of the consensus cDNA sequences of 220 representative eIF4E-family members from 118 species are presented. The predicted amino acid sequences, accession numbers of all sequences used to derive consensus cDNA sequences, and full names of all the species from which they were derived, are supplied in the form of Additional File 1. Sequence alignments and derived concensus cDNA sequences corresponding to all eIF4E-family members identified are available through the 'eIF4E-family member database [35].
Authors' contributions
BJ conceived of the study, designed and constructed the eIF4E-family member database [35], acquired and aligned sequences, analyzed the sequences for their phylogenetic relationships, and drafted the manuscript. KL participated in database management and sequence alignment. DM participated in the analyses of sequences for their phylogenetic relationships. RJ conceived of the study, participated in its design and coordination, and helped draft the manuscript.
Supplementary Material
Additional File 1
List of species names, predicted amino acid sequences, and accession numbers.
Click here for file
Acknowledgements
This work was supported by NSF grant MCB-0134013 to R. J. and B. J. K. L. was funded in part by REU awards from NSF and in part by the NOAA-EPP-funded Living Marine Resources Cooperative Science Center.
==== Refs
Gingras AC Raught B Sonenberg N eIF4 initiation factors: effectors of mRNA recruitment to ribosomes and regulators of translation Annu Rev Biochem 1999 68 913 963 10872469 10.1146/annurev.biochem.68.1.913
Hershey JWB Merrick WC Sonenberg N, Hershey JWB, Mathews MB Pathway and mechanism of initiation of protein synthesis Translational control of gene expression 2000 Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY 33 88
von der Haar T Gross JD Wagner G McCarthy JE The mRNA cap-binding protein eIF4E in post-transcriptional gene expression Nat Struct Mol Biol 2004 11 503 511 15164008 10.1038/nsmb779
Marcotrigiano J Gingras AC Sonenberg N Burley SK Cocrystal structure of the messenger RNA 5' cap-binding protein (eIF4E) bound to 7-methyl-GDP Cell 1997 89 951 961 9200613 10.1016/S0092-8674(00)80280-9
Matsuo H Li H McGuire AM Fletcher CM Gingras AC Sonenberg N Wagner G Structure of translation factor eIF4E bound to m7GDP and interaction with 4E-binding protein Nat Struct Biol 1997 4 717 724 9302999 10.1038/nsb0997-717
Tomoo K Shen X Okabe K Nozoe Y Fukuhara S Morino S Ishida T Taniguchi T Hasegawa H Terashima A Sasaki M Katsuya Y Kitamura K Miyoshi H Ishikawa M Miura K Crystal structures of 7-methylguanosine 5'-triphosphate (m(7)GTP)- and P(1)-7-methylguanosine-P(3)-adenosine-5', 5'-triphosphate (m(7)GpppA)-bound human full length eukaryotic initiation factor 4E: biological importance of the C-terminal flexible region Biochem J 2002 362 539 544 11879179 10.1042/0264-6021:3620539
Hsu PC Hodel MR Thomas JW Taylor LJ Hagedorn CH Hodel AE Structural requirements for the specific recognition of an m7G mRNA Cap Biochemistry 2000 39 13730 13736 11076512 10.1021/bi000623p
Vasilescu S Ptushkina M Linz B Muller PP McCarthy JE Mutants of eukaryotic initiation factor eIF-4E with altered mRNA cap binding specificity reprogram mRNA selection by ribosomes in Saccharomyces cerevisiae. J Biol Chem 1996 271 7030 7037 8636134 10.1074/jbc.271.12.7030
Robalino J Joshi B Fahrenkrug SC Jagus R Two zebrafish eIF4E family members are functionally divergent and are differentially expressed J Biol Chem 2004 279 10532 10541 14701818 10.1074/jbc.M313688200
Gross JD Moerke NJ von der Haar T Lugovskoy AA Sachs AB McCarthy JE Wagner G Ribosome loading onto the mRNA cap is driven by conformational coupling between eIF4G and eIF4E Cell 2003 115 739 750 14675538 10.1016/S0092-8674(03)00975-9
Scheper GC van Kollenburg B Hu J Luo Y Goss DJ Proud CG Phosphorylation of eukaryotic initiation factor 4E markedly reduces its affinity for capped mRNA J Biol Chem 2002 277 3303 3309 11723111 10.1074/jbc.M103607200
Richter JD Sonenberg N Regulation of cap-dependent translation by eIF4E inhibitory proteins Nature 2005 433 477 480 15690031 10.1038/nature03205
Pause A Belsham GJ Gingras AC Donze O Lin T-A Lawrence JCJ Sonenberg N Insulin-dependent stimulation of protein synthesis via a novel regulator of cap function Nature 1994 371 762 767 7935836 10.1038/371762a0
Lin T-A Kong X Haystead TAJ Pause A Belsham G Sonenberg N Lawrence JCJ PHAS-I as a link between MAP-kinase and translational initiation Science 1994 266 544 547
Poulin F Gingras AC Olsen H Chevalier S Sonenberg N 4E-BP3, a new member of the eukaryotic initiation factor 4E-binding protein family J Biol Chem 1998 273 14002 14007 9593750 10.1074/jbc.273.22.14002
Marcotrigiano J Gingras AC Sonenberg N Burley SK Cap-dependent translation initiation in eukaryotes is regulated by a molecular mimic of eIF4G Mol Cell 1999 3 707 716 10394359 10.1016/S1097-2765(01)80003-4
Takata M Ogawa W Kitamura T Hino Y Kuroda S Kotani K Klip A Gingras AC Sonenberg N Kasuga M Requirement for Akt (protein kinase B) in insulin-induced activation of glycogen synthase and phosphorylation of 4E-BP1 (PHAS-1) J Biol Chem 1999 274 20611 20608 10400692 10.1074/jbc.274.29.20611
Gingras AC Kennedy SG O'Leary MA Sonenberg N Hay N 4E-BP1, a repressor of mRNA translation, is phosphorylated and inactivated by the Akt(PKB) signaling pathway Genes Dev 1998 12 502 513 9472019
Yang D Brunn GJ Lawrence JC Mutational analysis of sites in the translational regulator, PHAS-I, that are selectively phosphorylated by mTOR FEBS Lett 1999 453 387 390 10405182 10.1016/S0014-5793(99)00762-0
Pyronnet S Imataka H Gingras AC Fukunaga R Hunter T Sonenberg N Human eukaryotic translation initiation factor 4G (eIF4G) recruits mnk1 to phosphorylate eIF4E EMBO J 1999 18 270 279 9878069 10.1093/emboj/18.1.270
Ptushkina M von der Haar T Karim MM Hughes JM McCarthy JE Repressor binding to a dorsal regulatory site traps human eIF4E in a high cap-affinity state EMBO J 1999 18 4068 4075 10406811 10.1093/emboj/18.14.4068
Lax S Fritz W Browning K Ravel J Isolation and characterization of factors from wheat germ that exhibit eukaryotic initiation factor 4B activity and overcome 7-methylguanosine 5'-triphosphate inhibition of polypeptide synthesis Proc Natl Acad Sci 1985 82 330 333 3855554
Browning KS Lax SR Ravel JM Identification of two messenger RNA cap binding proteins in wheat germ J Biol Chem 1987 262 11228 11232 2440886
Altmann M Muller PP Pelletier J Sonenberg N Trachsel H A mammalian translation initiation factor can substitute for its yeast homologue in vivo J Biol Chem 1989 264 12145 12147 2663851
Lasko P The Drosophila melanogaster genome: translation factors and RNA binding proteins J Cell Biol 2000 150 F51 56 10908586 10.1083/jcb.150.2.F51
Joshi B Cameron A Jagus R Characterization of mammalian eIF4E family members Eur J Biochem 2004 271 2189 2203 15153109 10.1111/j.1432-1033.2004.04149.x
Hernandez G Vazquez-Pianzola P Functional diversity of the eukaryotic translation initiation factors belonging to eIF4 families Mech Dev 2005 122 865 876 15922571 10.1016/j.mod.2005.04.002
Keiper BD Lamphear BJ Deshpande AM Jankowska-Anyszka M Aamodt EJ Blumenthal T Rhoads RE Functional characterization of five eIF4E isoforms in Caenorhabditis elegans J Biol Chem 2000 275 10590 10596 10744754 10.1074/jbc.275.14.10590
Ruud KA Kuhlow C Goss DJ Browning KS Identification and characterization of a novel cap-binding protein from Arabidopsis thaliana J Biol Chem 1998 24 10325 10330 9553087 10.1074/jbc.273.17.10325
Dinkova TD Keiper BD Korneeva NL Aamodt EJ Rhoads RE Translation of a small subset of Caenorhabditis elegans mRNAs is dependent on a specific eukaryotic translation initiation factor 4E isoform Mol Cell Biol 2005 25 100 113 15601834 10.1128/MCB.25.1.100-113.2005
Rom E Kim HC Gingras AC Marcotrigiano J Favre D Olsen H Burley SK Sonenberg N Cloning and characterization of 4EHP, a novel mammalian eIF4E-related cap-binding protein J Biol Chem 1998 273 13104 13109 9582349 10.1074/jbc.273.21.13104
Amiri A Keiper BD Kawasaki I Fan Y Kohara Y Rhoads RE Strome S An isoform of eIF4E is a component of germ granules and is required for spermatogenesis in C. elegans Development 2001 128 3899 3912 11641215
Jankowska-Anyszka M Lamphear BJ Aamodt EJ Harrington T Darzynkiewicz E Stolarski R Rhoads RE Multiple isoforms of eukaryotic protein synthesis initiation factor 4E in Caenorhabditis elegans can distinguish between mono- and trimethylated mRNA cap structures J Biol Chem 1998 273 10538 10542 9553113 10.1074/jbc.273.17.10538
Yoffe Y Zuberek J Lewdorowicz M Zeira Z Keasar C Orr-Dahan I M J-A Stepinsk iJ Darzynkiewicz E Shapira M Cap-binding activity of an eIF4E homolog from Leishmania RNA 2004 10 1764 1775 15388875 10.1261/rna.7520404
The eIF4E-Family Member Database
Zuberek J Jemielity J Jablonowska A Stepinsk J Dadlez M Stolarski R Darzynkiewicz E Influence of electric charge variation at residues 209 and 159 on the interaction of eIF4E with the mRNA 5' terminus Biochemistry 2004 43 5370 5379 15122903 10.1021/bi030266t
Aliyeva E Metz AM Browning KS Sequences of two expressed sequence tags (EST) from rice encoding different cap-binding proteins Gene 1996 180 221 223 8973371 10.1016/S0378-1119(96)00418-0
Rodriguez CM Freire MA Camilleri C Robaglia C The Arabidopsis thaliana cDNAs coding for eIF4E and eIF(iso)4E are not functionally equivalent for yeast complementation and are differentially expressed during plant development Plant J 1998 13 465 473 9680993 10.1046/j.1365-313X.1998.00047.x
Leonard SPD Wittmann S Daigneault N Fortin MG Laliberte JF Complex formation between potyvirus VPg and translation eukaryotic initiation factor 4E correlates with virus infectivity J Virol 2000 74 7730 7737 10933678 10.1128/JVI.74.17.7730-7737.2000
Duprat A Caranta C Revers F Menand B Browning KS Robaglia C The Arabidopsis eukaryotic initiation factor (iso)4E is dispensable for plant growth but required for susceptibility to potyviruses Plant J 2002 32 927 934 12492835 10.1046/j.1365-313X.2002.01481.x
Lellis AD Kasschau KD Whitham SA Carrington JC Loss-of-susceptibility mutants of Arabidopsis thaliana reveal an essential role for eIF(iso)4E during potyvirus infection Curr Biol 2002 12 1046 1051 12123581 10.1016/S0960-9822(02)00898-9
Hernandez G Altmann M Sierra JM Urlaub H Diez Del Corral R Schwartz P Rivera-Pomar R Functional analysis of seven genes encoding eight translation initiation factor 4E (eIF4E) isoforms in Drosophila Mech Dev 2005 122 529 543 15804566 10.1016/j.mod.2004.11.011
Blumenthal T Trans-splicing and polycistronic transcription in Caenorhabditis elegans Trends Genet 1995 11 132 136 7732590 10.1016/S0168-9525(00)89026-5
Stover NA Steele RE Trans-spliced leader addition to mRNAs in a cnidarian Proc Natl Acad Sci 2001 98 5693 5698 11331766 10.1073/pnas.101049998
Lall S Friedman CC Jankowska-Anyszka M Stepinski J Darzynkiewicz E Davis RE Contribution of trans-splicing, 5'-leader length, cap-poly(A) synergism, and initiation factors to nematode translation in an Ascaris suum embryo cell-free system J Biol Chem 2004 279 45573 45585 15322127 10.1074/jbc.M407475200
Ptushkina M Berthelot K von der Haar T Geffers L Warwicker J McCarthy JE A second eIF4E protein in Schizosaccharomyces pombe has distinct eIF4G-binding properties Nucleic Acids Res 2001 29 4561 4569 11713305 10.1093/nar/29.22.4561
Tee AR Tee JA Blenis J Characterizing the interaction of the mammalian eIF4E-related protein 4EHP with 4E-BP1 FEBS Lett 2004 564 58 62 15094042 10.1016/S0014-5793(04)00313-8
Cho PF Poulin F Cho-Park YA Cho-Park IB Chicoine JD Lasko P Sonenberg N A new paradigm for translational control: inhibition via 5'-3' mRNA tethering by Bicoid and the eIF4E cognate 4EHP Cell 2005 121 411 423 15882623 10.1016/j.cell.2005.02.024
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
Perry KL Watkins KP Agabian N Trypanosome mRNAs have unusual "cap 4" structures acquired by addition of a spliced leader Proc Natl Acad Sci 1987 84 8190 8194 3120186
Zeiner GM Sturm NR Campbell DA The Leishmania tarentolae spliced leader contains determinants for association with polysomes J Biol Chem 2003 278 38269 38275 12878606 10.1074/jbc.M304295200
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
Matsuzaki M Misumi O Shin-I T Maruyama S Takahara M Miyagishima SY Mori T Nishida K Yagisawa F Nishida K Yoshida Y Nishimura Y Nakao S Kobayashi T Momoyama Y Higashiyama T Minoda A Sano M Nomoto H Oishi K Hayashi H Ohta F Nishizaka S Haga S Miura S Morishita T Kabeya Y Terasawa K Suzuki Y Ishii Y Asakawa S Takano H Ohta N Kuroiwa H Tanaka K Shimizu N Sugano S Sato N Nozaki H Ogasawara N Kohara Y Kuroiwa T Genome sequence of the ultrasmall unicellular red alga Cyanidioschyzon merolae 10D Nature 2004 428 653 657 15071595 10.1038/nature02398
Katinka MD Duprat S Cornillot E Metenier G Thomarat F Prensier G Barbe V Peyretaillade E Brottier P Wincker P Delbac F El Alaoui H Peyret P Saurin W Gouy M Weissenbach J Vivares CP Genome sequence and gene compaction of the eukaryote parasite Encephalitozoon cuniculi Nature 2001 414 450 453 11719806 10.1038/35106579
Aravind L Koonin EV Eukaryote-specific domains in translation initiation factors: implications for translation regulation and evolution of the translation system Genome Res 2000 10 1172 1184 10958635 10.1101/gr.10.8.1172
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
Kumar S Tamura K Nei M MEGA: Molecular Evolutionary Genetics Analysis software for microcomputers Comput Appl Biosci 1994 10 189 191 8019868
Dayhoff MO Schwartz RM Orcutt BC Dayhoff MO A model of evolutionary change in proteins. matrices for detecting distant relationships Atlas of protein sequence and structure 1978 National Biomedical Research foundation: Washington, D.C 345 358
Bangs JD Crain PF Hashizume T McCloskey JA Boothroyol JC Mass spectrometry of mRNA cap 4 from trypanosomatids reveals two novel nucleosides J Biol Chem 1992 267 9805 9815 1349605
|
16191198
|
PMC1260017
|
CC BY
|
2021-01-04 16:29:16
|
no
|
BMC Evol Biol. 2005 Sep 28; 5:48
|
utf-8
|
BMC Evol Biol
| 2,005 |
10.1186/1471-2148-5-48
|
oa_comm
|
==== Front
BMC Fam PractBMC Family Practice1471-2296BioMed Central London 1471-2296-6-421622129910.1186/1471-2296-6-42Study ProtocolFrequent attenders in general practice: problem solving treatment provided by nurses [ISRCTN51021015] Schreuders B [email protected] Oppen P [email protected] Marwijk HWJ [email protected] JH [email protected] WAB [email protected] Department of General Practice, VU University Medical Center, Amsterdam, the Netherlands2 Department of Psychiatry, VU University Medical Center, Amsterdam, the Netherlands3 Institute for Research in Extramural Medicine, VU University Medical Center, Amsterdam, the Netherlands2005 12 10 2005 6 42 42 11 8 2005 12 10 2005 Copyright © 2005 Schreuders et al; licensee BioMed Central Ltd.2005Schreuders 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
There is a need for assistance from primary care mental health workers in general practice in the Netherlands. General practitioners (GPs) experience an overload of frequent attenders suffering from psychological problems. Problem Solving Treatment (PST) is a brief psychological treatment tailored for use in a primary care setting. PST is provided by nurses, and earlier research has shown that it is a treatment at least as effective as usual care. However, research outcomes are not totally satisfying. This protocol describes a randomized clinical trial on the effectiveness of PST provided by nurses for patients in general practice. The results of this study, which currently being carried out, will be presented as soon as they are available.
Methods/design
This study protocol describes the design of a randomized controlled trial to investigate the effectiveness and cost-effectiveness of PST and usual care compared to usual care only.
Patients, 18 years and older, who present psychological problems and are frequent attenders in general practice are recruited by the research assistant. The participants receive questionnaires at baseline, after the intervention, and again after 3 months and 9 months. Primary outcome is the reduction of symptoms, and other outcomes measured are improvement in problem solving skills, psychological and physical well being, daily functioning, social support, coping styles, problem evaluation and health care utilization.
Discussion
Our results may either confirm that PST in primary care is an effective way of dealing with emotional disorders and a promising addition to the primary care in the UK and USA, or may question this assumption. This trial will allow an evaluation of the effects of PST in practical circumstances and in a rather heterogeneous group of primary care patients. This study delivers scientific support for this use and therefore indications for optimal treatment and referral.
==== Body
Background
In primary care the prevalence of psychological problems (e.g. depression, anxiety, stress, somatization, unexplained or functional symptoms) ranges from 30% to 70%. Patients with these complaints, symptoms or disorders frequently visit their general practitioner [1] and only 3% of all patients are referred to a specialist. This implies that mental health care is a core activity in primary care [2]. For many of these complaints and symptoms no evidence-based treatment is available [3]. There is a clear need for an effective treatment for common mental disorders in primary care.
Problem Solving Treatment(PST) in primary care
In 1971 D'Zurilla and Goldfried published a theory in which problem-solving was defined as a cyclic process in five stages: problem orientation; problem definition; generation of alternative solutions; decision making, and solution implementation and was called problem solving therapy [4]. Since then, problem solving therapy has been applied for a wide range of psychological problems in all kinds of areas. In 1995 Gath and Mynors-Wallis conducted an experiment based on a basic form of PST in primary care. This is strictly protocollized and based on the principles of cognitive behavioral therapy (CBT) [5]. The treatment is brief and focuses on practical skill-building. It consists of a maximum of six sessions, each of which contains seven steps of problem-solving (see Figure 1) which are applied in a systematic manner towards problem resolution. The rationale is that it increases the patient's understanding of the relationship between everyday problems and psychological symptoms. The goal of PST is to help patients to regain control of their lives.
Figure 1 The seven stages of problem solving treatment.
There is evidence that PST can be an effective way of helping patients, and in particular patients with depression, to deal with psychological problems. One earlier study showed the superiority of PST over placebo but no superiority over amitryptiline [6]. A second study showed equal results in clinical outcomes between patients who received PST and patients who received usual care from their GP [6]. When community nurses provided PST the results were the same as for usual care from the GP, but the economic evaluation was more positive for the PST group [7]. Patients with minor depression who received PST showed the same improvement as patients who received a placebo, but their symptoms improved somewhat more rapidly than those of patients who received a placebo during the latter treatment. Patients with dysthymia who received PST and paroxetine showed significantly more improvement than patients who received a placebo [8]. Compared to other GP interventions there is good evidence PST is effective for major depression [1].
PST provided by nurses as a potential option
Patients with psychological problems need more time than is available in general practice. The usual 10-minute consultation with a GP is generally too short to explain and explore these psychological problems. To complicate matters more, these problems are often interwoven with physical issues such as fatigue and sleeplessness. Furthermore, patients are ambiguous in presenting their symptoms [9]. Given this fact, in combination with the high prevalence of psychological problems in primary care, treating these patients will result in a shift of tasks to nurses. Especially nurses who are skilled in working with psychiatric patients, may become indispensable in primary care [10]. Nurses can be successfully trained in the techniques of PST and can provide effective PST for primary care patients [11]. Recent results show that a CBT protocol for panic-disorder can adhered by a therapist with minimal of or no CBT experience [12].
There are several issues which stimulate further investigation. First, PST may be the way forward in the Netherlands, where GPs have a heavy workload and patients need better tailored collaborative forms of care, focused on self-help and education.
There are still very few nurses working in Dutch general practice and although preliminary experiments are taking place to enhance and define the role of nurses in primary care, PST could be a welcome innovation in their task profile. So, innovative projects in primary care in the Netherlands are needed.
Secondly, there is a lack of research outcomes on the effectiveness of talking treatment for anxiety symptoms in patients. In this trial, patients with depressive as well as anxiety symptoms will be included. Only one study has reported substantially better outcomes for primary care patients with panic disorder, who received CBT and pharmacotherapy from a therapist with minimal or no CBT experience, like a nurse, than patients with usual care only from their GP [12].
Third, PST in primary care could prevent or stimulate a referral to secondary care for patients with complaints which cannot be treated in primary care. This would also stimulate better tailored collaborative forms of care and prevent the deterioration of complaints.
The primary aim of the present trial is therefore to investigate whether PST for patients with psychological problems provided by nurses in primary care, is effective.
Methods
Design
A randomized, controlled trial is being carried out to evaluate the effects of PST. At least 160 primary care patients will be included; 80 will receive usual care and PST and 80 will receive usual care only. At baseline, after the intervention and after 3 and 9 months the patients will be asked to fill in a questionnaire, and at baseline and after 9 months they will be asked to cooperate in a (diagnostic) telephonic interview. Primary outcome is the reduction of symptoms, and other outcomes measured are improvement in problem solving skills, psychological and physical well being, daily functioning, social support, coping styles, problem evaluation and health care utilization.
The Medical Ethics Committee of the VU Medical Center in Amsterdam approved the study design.
Study population
The study population will consist of Dutch-speaking adults (18+) who visited a participating GP more than three times in the last six months. To asses whether psychological problems are present, the General Health Questionnaire 12 item version (GHQ-12) will be used for screening [13]. At random we visited the participating GP practices to ask patients to fill in our screening questionnaire while they where waiting to see their GP. If they had a score negative score on more then three out of twelve questions (indicating the presence of psychological problems) and if they were willing to participate, they were included. Patients were excluded from the study if they: received any treatment in mental health care; suffered from a severe (psychical) disease or personality-disorder; accepted no other explanation for their complaints than a somatic rationale; and patients with an non-consistent medication for anxiety or depression. Patients with severe drug addictions, suicidal wishes or mental retardation were excluded. An external researcher conducted block-randomization, so the allocation was concealed.
Intervention: usual care and PST provided by nurses
Consistent with earlier research on PST training skills [11], the nurses in this RCT were trained for two days by experienced supervisors who were also members of the original Oxford research group[6]. The nurses were closely supervised by means of video and audiotapes. A CBT supervisor will carry out supervision after the training, for one hour every three weeks. The nurses will deliver audiotapes and PST protocol forms to the supervisor. Consistent with advice in earlier research [11] before the nurses started treating patients in the trial, they treated four patients to practice their problem-solving skills after the training. The patients are also seen by their GP for general health management if necessary.
Usual care: health management provided by the GP
The consult is intended to be as natural as possible so that the GP will not influence the quality of the usual care provided. Many GPs use the guidelines issued by the Dutch College of General Practitioners [14]. The guidelines for psychological complaints such as anxiety and depression describe management options as (anti-anxiety of anti-depressant) medication and/or 'watchful waiting' if a referral seems unnecessary [15].
Outcome measures
Primary outcome: reduction of symptoms measured with the HADS
The Hospital Anxiety and Depression Scale (HADS, [16]) is used to monitor symptom levels of anxiety and depression in the study population. The questionnaire consists of 14 items to which answers can be given on a 4-point scale (0–3). The HADS is considered to be unbiased by coexisting general medical conditions, and changes in HADS scores can therefore be used to calculate an objective effect size of the treatment provided (calculations described in 'Sample size'). In the Dutch validation of the HADS [16] the primary care patients have a mean of 6.2 (SD 3.8)for anxiety and 3.7(SD 3.4) for depression with a total mean of 9.9 (SD 6.1). Reliability for these patients is a Cronbach's alpha of .82 for the total score. The HADS is found to perform well in assessing the severity of symptoms [17].
Secondary outcomes: reduction of symptoms measured with the PHQ
The Patient Health Questionnaire (PHQ) is designed to facilitate the diagnosis of common mental disorders in primary care patients [18]. The PHQ is a self report version of the Primary Care Evaluation of Mental Disorders (PRIME-MD). The questions do not only focus on mood disorders but also about functional impairments, life stressors and specific events (such as menstruation, pregnancy and childbirth). Its diagnostic validity is good, and patients feel comfortable filling in the questionnaire [19]. There is a 15-item questionnaire for men and a 16-item questionnaire for women, and the scoring range varies. We consider that a decrease in the score on this questionnaire after the intervention represents a reduction in mood disorders, functional impairments, life stressors and distress about specific events.
Improvement in problem-solving skills
The Social Problem Solving Skills-Revised (SPSI-R, 15) is a 52-item, self-report inventory, which is designed by D'Zurilla to measure problem-solving skills [20]. The SPSI-R consists of five factors: 1) positive problem orientation (PPO), 2) negative problem orientation(NPO), 3) rational problem solving (RPS), 4) impulsivity/carelessness style (ICS) and, 5) avoidance style(AS). Alphas for these five scales range from .76 to .92 [21] an test-retest reliability ranges from .72 to .88 [20].
Psychological and physical well-being
This will be measured with the Short Form-36 (SF-36) which contains 36 questions and standardized response options and relating to eight different areas (multi-item): physical functioning, role limitations due to psychical health problems, bodily pain, general health perceptions, vitality, social functioning, role limitations due tot emotional problems, and general mental health [22]. The mean alpha for reliability in the general population is good, as well as validity which makes the SF-36 a practical instrument for use in the general population.
Social support
The Social Support Inventory is a questionnaire which comprises 20 descriptions of social support pertaining to emotional support, informative support, social companionship, or instrumental support. Together these items give an overall representation of satisfaction with social support (the perceived adequacy). It is a reliable and brief measurement instrument which is not influenced by psychological distress [23].
Coping-styles
The Ways of Coping Questionnaire (WAYS) (the Dutch adaption is called the VOMS: Vragenlijst over Omgaan met Situaties) is based on the Lazarus and Folkman transactional coping theory of [18]. It measures coping processes, not coping dispositions or styles. The WAYS can assess and identify thoughts and actions that individuals use to cope with stressful encounters in everyday life. The WAYS measures eight coping factors: confrontive coping, distancing, self-controlling, seeking social support, accepting responsibility escape-avoidance, planful problem solving, and positive reappraisal.
Rumination
Actual scientific reports suggest that rumination is a significant, and probably prognostic, factor for depression. The rumination scale (RRS) measures the extend to which people ruminate [24]. Rumination is seen as a coping style and characterizes depressive mood. The reliability is good and its validity is satisfying [25,26].
Problem evaluation
This will be assessed with a brief, qualitative questionnaire about medical outcomes, the care, the illness and the treatment of the patients, as experienced by the patients, to complement all the quantitative questionnaires. We chose the PSYCHLOPS(also known in the literature as MYMOP [27,28]) to evaluate the problems patients experience and the progress they make over time.
Health care utilization
The Trimbos/iMTA questionnaire for Costs associated with Psychiatric Illness (Tic-P) is used to measure the amount of health care received by the patients and to register sickness absence from work [29] Furthermore, we chose the EQ-5D (or Euroqol, [30]) because this is a standardized measurement instrument for a wide range of health conditions which provides a simple descriptive profile and a single index value for health status. The EQ-5D (or EuroQol) was originally designed to complement other instruments such as the SF-36, and it is administered to assess a patient's general health status, in 5 dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression[1]. Because each of the five dimensions can be sub-divided into 3 levels a total of 243 health states can be assessed. Using the Dolan model (1997) the total score will be expressed in utilities [31]. The official Dutch translation of the Euroqol will be administered [31,32]. Incremental cost-effectiveness ratios will be calculated, in which the difference in costs between intervention subjects and control subjects will be divided by the difference in effects between both groups. Incremental cost-utility ratios will also be calculated in which the difference in costs between the two groups will be divided by the difference in QALYs gained between the two groups.
Power/analysis
Randomization takes place at patient level. To evaluate the effects of the randomization, descriptive statistics will be used to compare the baseline measurements of the two groups. If necessary, differences between baseline variables on relevant characteristics (such as baseline HADS score) will be entered as covariates in the analysis.
To detect a clinically relevant difference between interventions (effect size of 0.4) with the primary outcome measure HADS (power of .90 and an alpha of .05) 130 completers are needed. In both conditions there will be 65 completers. We estimate the drop-out rate to be on 20%, so we therefore need 160 participants. If 20 practices cooperate, we will need to include 8 patients in every practice. We expect a non-response of 50%, so we will need to screen 16 patients in every practice. In a sample of patients with mixed symptoms of anxiety and depression, Cropper et al. [33]observed a mean overall HADS score of 6.25 at baseline. In the Dutch validation [16] the primary care patients have a mean on anxiety of 6.2 (SD 3.8) and 3.7(SD 3.4) for depression with a total mean of 9.9 (SD6.1). We consider a standardized mean difference (SMD) of 0.4 (p = 0,05) on the primary outcome (HADS) to represent a relevant improvement in the PST group versus the usual care group[34].
Analysis
Linear regression models will be used to examine differences in investigate on the HADS. Scores will be entered in a repeated measure design (GLM), and (optional) covariates will be differences at baseline level. Repeated measures with several independent variables will be used to investigate differences in improvement in all secondary outcome measures between groups. The analyses will be performed on a per protocol basis ('completers'), as well as according to the 'intention-to-treat' principle. Trend analyses will also allow 'last observation carried forward' analyses. To assess whether protocol deviations have caused bias, the results of the intention-to-treat analyses will be compared with analyses of the PST group, including the completers.
Sample size
A GP in the Netherlands has an average of 80 consultations a week (children excluded). Six of the patients who consult their GP had done so more than three times in the previous six months and were 'frequent attenders'. With a response of 50%, three patients a week per general practice will be sufficient, but: only one third of them will meet the inclusion criteria. This implies that 1 patient can be included per practice per week. To include 160 patients with a screening in two practices per week, will take approximately one and a half years.
Arguments for publishing a design
The primary goal of presenting the design of this study before the results are available is to offer the reader the opportunity to consider the methodological quality of this study more critically and this is also a benefit for caregivers, because this extensive information provides more insight into the practical applications of the study intervention.
Publication also prevents publication-and-analysis bias. Trials that lead to adverse or negative results are less likely to be submitted for publication [35]. This can be avoided by publishing a priori the design of a study and the plans for analysis. Not only will the researchers be more inclined to publish the results, but transparency will also be increased and, in any case, data can be requested from the researcher for inclusion in a systematic review.
Discussion
Our results may either confirm that PST in primary care is an effective way of dealing with emotional disorders and a promising addition to the primary care in the UK and USA, or may question this assumption. This trial will allow an evaluation of the effects of PST in practical circumstances and in a rather heterogeneous group of primary care patients.
Strengths and limitations
Many methodological requirements for a high quality trial are met [36]. Allocation is concealed through block-randomization by an external researcher. Recruitment of responders will be in the GP's waiting room, without the GP knowing where and how randomization will take place. The methodology used in this trial will overcome concerns of selection bias. The relevance to the Dutch primary care seems sufficient and the generalisability of our sample to patients in everyday practice seems high. The sample size is large, when compared to other trials on psychological problems in general practice [37]. Additionally, whether the outcome is negative or positive, the project will give a clearer understanding of who might or might not benefit from this treatment.
Another strength of this study is the chosen primary outcome, the self-report HADS. The HADS is a well-known questionnaire to measure reduction of symptoms of anxiety and depression. Many previous trials in psychiatry have relied on assessments of the therapist. Another strength is the combination of quantitative assessments and a qualitative process evaluation. If the psychiatric symptoms decline more in the intervention group than in the usual care group, we can therefore safely attribute this to PST treatment. If not, we will explore mechanisms This will provide useful information about implementing the intervention. A practical strength of the study is that the intervention can take place without disturbing care as usual, both in the study as in future implementation.
Some limitations of the present study need to be addressed. It is possible that patients are less motivated to attend PST treatment since screening of all participants takes place in the waiting room of the GP's. This may be lead to a higher dropout percentage in the PST-condition in comparison with the dropout percentage of the CAU.
This study may be criticized because of the lack of a control group without treatment. However, patients are always free to visit their general practitioners. Furthermore, naturalistic studies of the longitudinal course of anxiety and depression indicate that patients who have these symptoms longer than three months are suffering from chronic symptoms [38,39].
Other potential criticism concerns the suitability of PST for primary care. Although PST seems suitable for primary care, provision of PST by trained nurses will not always be available in daily practice of small primary care practices.
Conclusion
PST is a short psychological treatment for use in general practice. This study delivers scientific support for this use and therefore indications for optimal treatment and referral. Study completion is anticipated for January 2006, with results available in May 2006.
Competing interests
The author(s) declare they have no competing interests.
Authors' contributions
PvO and HvM developed the design of the randomized clinical trial and participated in writing the article. WS and JS advised on the content of the article. BS conducts the research and wrote the article. All authors provided comments on the drafts and have read and approved the final version.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The Health Research and Development Council (ZONMW) in the Netherlands funded this project.
==== Refs
Huibers MJH Beurskens AJHM Blijenberg G Van Schayk CP The effectiviness of psychosocial interventions delivered by general practitioners: a Cochrane review. the cochrane library 2003
Cardol M Van Dijk L De Jong JD De Bakker DH Westert GP Huisartsenzorg: wat doet de poortwachter ? Tweede Nationale Studie naar ziekten en verrichtingen in de huisartspraktijk 2005 Utrecht, Bilthoven:NIVEL, Rijksinstituut voor Volksgezondheid en Milieu 1 219
Wilkinson P Mynors-Wallis L Problem-solving therapy in the treatment of unexplained physical symptoms in primary care: a preliminary study J Psychosom Res 1994 38 591 598 7990067 10.1016/0022-3999(94)90056-6
D'Zurilla TJ Goldfried MR Problem solving and behavior modification J Abnorm Psychol 1971 78 107 126 4938262
van den Hout JH Kerckhoffs-Hanssen MR Vlaeyen JWS The effects of problem-solving therapy in mental health care and behavioural medicine: a systematic review. To solve or not to solve? Effects of problems solving therapy and graded activity in non-specific low back pain 2002 Maastricht 33 58
Mynors-Wallis L Problem-solving treatment: evidence for effectiveness and feasibility in primary care Int J Psychiatry Med 1996 26 249 262 8976466
Mynors-Wallis L Davies I Gray A Barbour F Gath D A randomised controlled trial and cost analysis of problem-solving treatment for emotional disorders given by community nurses in primary care Br J Psychiatry 1997 170 113 119 9093498
Barrett JE Williams JWJ Oxman TE Frank E Katon W Sullivan M Hegel MT Cornell JE Sengupta AS Treatment of dysthymia and minor depression in primary care: a randomized trial in patients aged 18 to 59 years J Fam Pract 2001 50 405 412 11350703
Kessler D Bennewith O Lewis G Sharp D Detection of depression and anxiety in primary care: follow up study BMJ 2002 325 1016 1017 12411363 10.1136/bmj.325.7371.1016
Marks I Innovations in mental health care delivery. British Journal of Psychiatry 1992 160 589 597 1591570
Mynors-Wallis L Davies I Gray A Barbour F Gath D A randomised controlled trial and cost analysis of problem-solving treatment for emotional disorders given by community nurses in primary care. British Journal of Psychiatry 1997 113 119 9093498
Roy-Byrne PP Craske MG Stein MB Sullivan G Bystritsky A Katon W Golinelli D Sherbourne CD A Randomized Effectiveness Trial of Cognitive-Behavioral Therapy and Medication for Primary Care Panic Disorder Arch Gen Psychiatry 2005 62 290 298 15753242 10.1001/archpsyc.62.3.290
Koeter MWJ Ormel J General Health Questionnaire. Nederlandse bewerking: handleiding 1991 Swets, test services.
Braspenning JCC F.G. S Grol RPTM Kwaliteit in beeld. Medisch-technisch handelen huisartsen doorgelicht. Medisch Contact 2004
Van Marwijk HWJ Bijl D Van Gelderen M De Haan M E VWB Dutch College of General Practioners Guideline Depression, first revision [NHG-Standaard Depressieve stoornis (depressie). Eerste herziening. In Dutch]. Huisarts Wet 2003 614 623
Spinhoven PH Ormel J Sloekers PPA Kempen JIGM Speckens AEM Hemert AM A validation study of the hospital anxiety and depression scale (in different groups of) Dutch subjects. Psychological medicine 1997 363 370 9089829 10.1017/S0033291796004382
Bjelland I Dahl AA Haug TT Neckelmann D The validity of the Hospital Anxiety and Depression Scale. Journal of Psychosomatic Research 2002 Volume 52 69 77 11832252 10.1016/S0022-3999(01)00296-3
Taylor SE Coping and stress Health psychology 1995 3 third Singapore, Mc Graw-Hill 219 293
Spitzer RL Kroenke K Williams JB Validation and utility of a self-report version of PRIME-MD: the PHQ primary care study. Primary Care Evaluation of Mental Disorders. Patient Health Questionnaire JAMA 1999 282 1737 1744 10568646 10.1001/jama.282.18.1737
D'Zurilla TJ Chang EC Nottingham EJ Faccini L Social problem-solving deficits and hopelessness, depression, and suicidal risk in college students and psychiatric inpatients J Clin Psychol 1998 54 1091 1107 9840781 10.1002/(SICI)1097-4679(199812)54:8<1091::AID-JCLP9>3.0.CO;2-J
Nezu AM Nezu CM Problem Solving Therapy Journal of Psychothrapy Integration 2001 11 187 205 10.1023/A:1016653407338
Aaronson NK Muller M Cohen PD Essink-Bot ML Fekkes M Sanderman R Sprangers MA Te Velde A Verrips E Translation, validation, and norming of the Dutch language version of the SF-36 Health Survey in community and chronic disease populations J Clin Epidemiol 1998 51 1055 1068 9817123 10.1016/S0895-4356(98)00097-3
Timmerman IHG Emanuel-Zuurveen ES Emmelkamp PMG The social support Inventory (SSI): A brief scale to asses perceived adequacy of social support Clinical Psychology & Psychotherapy 2002 7 401 410 10.1002/1099-0879(200011)7:5<401::AID-CPP253>3.0.CO;2-I
Nolen-Hoeksema S Parker LE Larson J Ruminative coping with depressed mood following loss J Pers Soc Psychol 1994 67 92 104 8046585 10.1037/0022-3514.67.1.92
Raes F Hermans D Eelen P Rumineren bij depressie, of: hoe stilstaan bij depressie eigenlijk achteruit gegaan is. Gedragstherapie 2003 147 163
Raes F Hermans D Eelen P De Nederlandstalige versie van de Ruminative Response Style (RRS-NL) en de Rumination on Sadness Scale (RSS-NL). Gedragstherapie 2003 97 104
Paterson C Measuring outcomes in primary care: a patient generated measure, MYMOP, compared with the SF-36 health survey BMJ 1996 312 1016 1020 8616351
Paterson C Seeking the patient's perspective: a qualitative assessment of EuroQol, COOP-WONCA charts and MYMOP Qual Life Res 2004 13 871 881 15233501 10.1023/B:QURE.0000025586.51955.78
Hakkaart-Van Roijen L Trimbos/iMTA questionnaire for Costs associated with Psychiatric Illness (Tic-P). Institute for Medical Technology assessment 2002
Kind P Spilker B The Euroqol Instrument: an index of health-related quality of life. Quality of life and pharmacoeconomics in clinical trials 1996 Philadelphia, Lipincott-Ravwen Publishers 191 201
Dolan P Modeling valuations for EuroQol health states. Med Care 1997 1095 1108 9366889 10.1097/00005650-199711000-00002
Van der Roer N Van Tulder MW Barendse JM Van Mechelen W Franken WK Ooms AC De Vet HCW Cost-effectiveness of an intensive group training protocol compared to physiotherapy guideline care for sub-acute and chronic low back pain: design of a randomised controlled trial with an economic evaluation. [ISRCTN45641649] BMC Musculoskeletal Disorders 2004 5 15560843
Cropper M A G Mc Ewan GD Munt DM Rushbook LA al SV A double-blind comparative study of alprazolam and dothiepin hydrochloride in the treatment of anxiety associated with depression. Pharmatherapeutica 1987 76 82 3310018
Altman DG Practical statistics for medical research. Chapman & Hall 1991 1 London
Easterbrook PJ Berlin JA Gopalan R Matthews DR Publication bias in clinical research Lancet 1991 337 867 872 1672966 10.1016/0140-6736(91)90201-Y
D M KF S DG A The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomized trials. J Am Podiatr Med Assoc 2005 437 442
C C M D K F V G al. PB Antidepressant drugs and generic counselling for treatment of major depression in primary care: randomised trial with patient preference arms. BMJ 2005 772 775 16195299
Bijl RV van Zessen G Ravelli A Psychiatric morbidity among adults in The Netherlands: the NEMESIS-Study. II. Prevalence of psychiatric disorders. Netherlands Mental Health Survey and Incidence Study Ned Tijdschr Geneeskd 1997 141 2453 2460 9555131
Spijker J de Graaf R Bijl RV Beekman ATF Ormel J Nolen WA Duration of major depressive episodes in the general population: results from The Netherlands Mental Health Survey and Incidence Study (NEMESIS) Br J Psychiatry 2002 181 208 213 12204924 10.1192/bjp.181.3.208
|
16221299
|
PMC1260018
|
CC BY
|
2021-01-04 16:29:13
|
no
|
BMC Fam Pract. 2005 Oct 12; 6:42
|
utf-8
|
BMC Fam Pract
| 2,005 |
10.1186/1471-2296-6-42
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1401620971410.1186/1471-2164-6-140Research ArticleThe distribution of SNPs in human gene regulatory regions Guo Yongjian [email protected] D Curtis [email protected] School of Computational Sciences, George Mason University, Manassas, VA 20110 USA2 Virginia Bioinformatics Institute, Bioinformatics Facility I (0477), Virginia Tech, Blacksburg, VA 24060 USA2005 6 10 2005 6 140 140 16 3 2005 6 10 2005 Copyright © 2005 Guo and Jamison; licensee BioMed Central Ltd.2005Guo and Jamison; 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
As a result of high-throughput genotyping methods, millions of human genetic variants have been reported in recent years. To efficiently identify those with significant biological functions, a practical strategy is to concentrate on variants located in important sequence regions such as gene regulatory regions.
Results
Analysis of the most common type of variant, single nucleotide polymorphisms (SNPs), shows that in gene promoter regions more SNPs occur in close proximity to transcriptional start sites than in regions further upstream, and a disproportionate number of those SNPs represent nucleotide transversions. Additionally, the number of SNPs found in the predicted transcription factor binding sites is higher than in non-binding site sequences.
Conclusion
Current information about transcription factor binding site sequence patterns may not be exhaustive, and SNPs may be actively involved in influencing gene expression by affecting the transcription factor binding sites.
==== Body
Background
Genetic variation has been found to be a ubiquitous phenomenon, and forms the genetic basis for species diversity. Currently, there are sequence variant data accumulated for humans [1], mouse [2], soybean [3] and other organisms. With the completion of Human Genome Project [4], the study of genetic variation has become one of the keystones in biomedical research, not only because it affects an individual's anthropometric characteristics but also because it influences risk of disease and response to environmental challenges [5]. The information derived from the study of variation not only deepens our understanding of human genes and evolution, but also brings benefits to the identification and treatment of human genetic diseases.
There are several different types of genetic sequence variants, including single nucleotide polymorphisms (SNPs), small deletion and insertion polymorphisms (indels), micro-satellite markers, and polymorphic insertion elements such as retrotransposons [6]. Because the most common variants are SNPs, the term is often abused as a synonym for genetic sequence variation. However, here we restrict its usage to the formal SNP definition: a single base change at a single position.
As a result of high-throughput genotyping methods, millions of human SNPs have been reported in recent years. To more efficiently study those with significant biological functions, a practical method is to concentrate efforts on SNPs located in genomic regions with important functions. There have been several studies focusing on evaluating how SNPs impact phenotype. For example, Ramensky et al. [7] have applied several rules to predict biological effects amino acid substitutions made by non-synonymous coding SNPs. Clifford et al. [8] have explored non-synonymous coding SNPs effects on protein function through exploring those introducing amino acid alterations in protein motif regions.
Another genome region important to gene function are gene regulatory regions. Through binding of specific transcription factors, gene promoters are directly involved in gene transcription initiation and regulation. Thus sequence variation in gene promoters may alter transcription factor identification and binding, which in turn can influence gene expression and effect biological impacts. For example, it has been found that one allele of the HLA-G gene (-725G), whose products inhibits maternal anti-fetal immune response, is highly associated with increasing risk for miscarriage [9]. One possible explanation is that the SNP falls within the binding site of interferon response factor-1 (IRF-1), affecting IRF-1 binding and down-regulating transcription of the HLA-G gene.
Here we analyze the distribution of SNPs in human gene regulatory regions. Putative transcription factor binding sites in the gene promoters were computationally derived and compared to previously identified SNPs. The results show that SNPs have differential distribution characteristics both in gene regulatory regions and in transcription factor binding sites when compared to the entire genomic SNP population.
Results
The build 33 release of the genome sequence contains 545 contig sequences mapped across all of the Homo sapiens chromosomes. 19,741 gene sequences were extracted from NCBI RefSeq database, which were linked to contig sequences by 15,803 LocusLink entries. While most loci have only one gene, some have two or more forming the structure of a gene cluster. In addition, sequence version discrepancies between the databases lead to the removal of approximately 3000 genes. The loss of these genes does not create any bias, as the errors in the affected sequence records were randomly distributed across the databases with no discernible pattern. Thus the final sequence data consisted of derived promoter regions for 16,429 genes.
TFD contains 3,749 mammalian transcription factor binding site sequences, with approximately 2,000 of them being found at least once in the derived gene promoter sequences. More than 1,700 binding sites were not found, presumably because they are either non-human, located outside the immediate upstream gene regions, or are binding sites for trans-regulatory elements. The number of predicted binding sites per promoter follows a normal distribution.
SNP distribution in gene promoters
Over 35,000 SNPs were found in the gene regulatory regions. More than 99.8% of them are two state alleles. The nucleotide substitution rates of promoter SNPs with two alleles are shown in Table 1. For comparison, substitution rates for all SNPs in the dbSNP database are also listed. In both sets of SNPs, transition class SNPs account for approximately two-thirds of the total number. This is not surprising, as the mismatches made by substitution within the same chemical group is thermodynamically more stable [10]. The remaining one-third SNPs are roughly evenly distributed across the four transversion types.
Table 1 Nucleotide substitution rate. Nucleotide substitution rates for SNPs in promoter regions (P) and all SNPs genomewide (A) show roughly similar rates for transition and transversion substitution, with a slight increase in CG transversions as would be expected from CG rich regions.
Type Substitution Frequency (P) Frequency (A)
Purine A ↔ G 31.55% 33.10%
Pyrimidine C ↔ T 30.99% 33.10%
Purine ↔ Pyrimidine C ↔ G 12.75% 8.93%
Purine ↔ Pyrimidine A ↔ C 9.43% 8.77%
Purine ↔ Pyrimidine G ↔ T 9.33% 8.82%
Purine ↔ Pyrimidine A ↔ T 5.94% 7.42%
For SNPs in promoter regions, G/C substitution is higher than other nucleotide substitutions in the transversion type, which corresponds to the higher GC content in gene promoters. The distributions of SNPs of different types are plotted in Figure 1, which demonstrates that promoter SNPs are not evenly distributed in the promoter range of [-2000, -1], with more SNPs are observed in the region close to the transcriptional start site. While the numbers of SNPs of all types are all increased in the region near transcriptional start site, the number of SNPs of transversion type shows the largest increase.
Figure 1 Distributions of SNPs in different categories in gene promoter regions. Each symbol represents the average number of SNPs found in a 20 bp bin across approximately fourteen thousand promoter sequences with a full length of 2000. This composite statistic shows the SNP density increases in proximity to the transcriptional start site.
Transcription factor binding site redundancy
Data redundancy of transcription factor binding site motifs was evaluated by comparing binding site sequence similarity, which was divided into three categories: exact coverage, partial coverage and single nucleotide difference. The results of these three categories are shown in Table 2. There are 310 pairs of binding sites with exact sequence coverage such that the sequences can be represented using more generalized patterns. There are 1,114 pairs with partial coverage properties that can be regarded as a subsequence of other binding sites. Finally, 833 pairs are found with sequences that differ only at a single nucleotide position. Potential data redundancy of the binding site sequences makes it uninformative to study individual binding sites in the partial coverage and single nucleotide difference categories, so these motif pairs were merged into single motifs.
Table 2 Transcription factor binding site sequence similarity. Examples of the three classes of degeneracy in binding site sequence motifs. For motif pairs with exact coverage, the less restrictive motifs were used. Motif pairs in the partial coverage and single difference categories were merged into a composite motif.
Category Pair Number Binding Site Examples
Sequence Site Name
Exact Coverage 310 YYCCGCCC
CCCCGCCC EARLY-SEQ1
(Sp1)-TK.1
Partial Coverage 1,114 TGGNNNNNNGCCA
YTGGCANNNTGCCAR NFI_CS3
TGGCA_RS
Single Difference 833 TKCTGATTGTYTMM
TKCTGATTGGYTMM E-alpha_Y_box
NF-Y_CS
Transcription factor binding sites distribution in gene promoters
The distribution of binding sites in the gene promoter regions is not uniform, as shown in Figure 2. In the range of -2000 to -400 upstream the frequency is fairly constant. However the frequency increases in the regions closer to the transcriptional start site, reaching two to three times higher in the -100 to -50 range than that in the further sequence regions. This suggests most transcription factor binding sites occur within 250 bp of the initiation site. Beyond the transcriptional start site the frequency of binding sites decreases dramatically, with the frequency value dropping below that of the entire upstream sequence.
Figure 2 Distribution of transcription factor binding sites in gene promoter regions. The x-axis is the position index of gene promoter sequence, and the y-axis is the occurrence frequency of the binding sites across all promoters. Only genes with full length promoter sequences were included in calculation (approximately 14,000 sequences). A portion of the downstream sequence relating to transcriptional start site was also included. The position with index of 0 specifies the transcriptional start sites.
Changing the expectation value from 0.1 to 0.01 causes the overall transcription factor binding site frequency to become smaller. The decrease is uniform across the entire sequence range of -2000 to 200, as fewer binding site sequence patterns fulfilling the match criteria are detected. This uniformity suggests there is no issue of selective sensitivity.
Transcription factor binding sites in random sequence datasets simulating promoter sequence in the range of -2000 to -1 were predicted using an expect value 0.1, and the results are shown in Figure 3. There is a measurable difference in the occurrence of transcription factor binding sites in the real promoter sequence dataset and the randomly generated sequence dataset. Three-fold more binding sites are detected in the random sequences than in the real promoter sequences, suggesting a non-random nucleotide distribution in real promoter sequences.
Figure 3 Comparison of transcription factor binding site distributions in random sequence datasets and real promoter sequence dataset. Each curve represents the occurrence frequency of predicted binding sites in different data sets of comparable size. "Real Random Seq" is a data set of completely random sequences in which the emission probabilities of A, C, G and T were equal and uniform across the entire 2 kb. "Adjusted Random Seq" is a data set of random sequences generated with the adjusted emission probabilities of A, C, G and T according to that in the corresponding position at the real promoter sequence. "Real Seq" is the real promoter sequence dataset. An expectation value of 0.1 was used for detecting transcription factor binding sites in these datasets.
The patterns of transcription factor binding site occurrence in the two random sequence datasets are different. For completely random sequences, the occurrence appears to be flat across the whole range. However, in the frequency-based random sequences an increased appearance is observed in the sequence range close to transcriptional start sites similar to the pattern in the real promoter dataset. As the only difference between these two random datasets is the nucleotide emission probabilities at each sequence position, it suggests that transcription factor binding sites are biased by sequence composition. As GC content is enriched in the regions close to transcriptional start site [11], the fact that more transcription factor binding sites are observed in the promoter region near transcriptional start sites indicates that the binding site sequences are GC skewed.
SNP distribution in transcription factor binding sites
For 34,858 promoter SNPs mapped to 13,723 promoter regions, 2,078 (5.9%) of them are located in the predicted transcription factor binding sites, and 1,969 (5.6%) of them have alleles introducing new binding sites. The overlap of these two SNP sets has 243 members, indicating that their alleles may have different effects on transcription factor binding.
The nucleotide frequency of SNPs falling inside and outside transcription factor binding sites is shown in Table 3. The frequency of SNP distribution in the overall gene promoter region is 0.13%. However, the frequency increases to 0.20% (p = 2.2 × 10-16) in the predicted transcription factor binding site regions.
Table 3 Promoter SNPs distribution on predicted transcription factor binding sites. Frequency of SNPs appearing in promoter regions and in putative transcription factor binding sites. The frequency of SNPs is significantly higher in putative binding sites than in the surrounding sequence (p = 2.2 × 10-16).
Categories SNPs Nucleotides Frequency
Inside a binding site 3,804 1,890,176 0.20%
Outside the binding site 31,054 25,555,824 0.12%
Total 34,858 27,446,000 0.13%
A set of 293 experimentally derived transcription factor binding sites (associated with 85 genes) drawn from TRANSFAC were linked to predicted transcription factors used for this study. Of these, 13 contained SNPs, as shown in Table 4. The total number of nucleotides organizing the 293 TF binding sites is 5131. With 13 SNPs, the frequency of SNPs in the TF binding site sequence is 0.25%, a value not significantly different from the frequency of promoter SNPs in the putative TF binding sites (p-value = 0.348).
Table 4 Promoter SNPs distribution on experimental transcription factor binding sites. Genes which have SNPs in experimentally determined transcription factor binding sites. The frequency is not significantly different from the frequency of promoter SNPs in the putative TF binding sites (p-value = 0.348).
Gene Accession ID SNP Accession ID SNP Position TF Binding Site Start TF Binding Site End Binding Site Name TRANSFAC Binding Site ID
NM_000384 rs9282608 -149 -155 -134 ApoB-site [25] R03692
NM_000384 rs13306199 -82 -86 -61 AF-1-k-apoB [26] R01612
NM_005252 rs4645852 -84 -85 -78 c-fos.5 [27] R00470
NM_002467 rs4645940 -1827 -1834 -1820 c-myc-PUR [28] R04301
NM_002467 rs13250910 -322 -326 -310 NHE [29] R01804
NM_002690 rs2307158 -83 -93 -75 Sp1-human-beta-polymerase [30] R00287
NM_002690 rs2307155 -55 -72 -51 CREB-beta-polymerase [31] R00288
NM_000805 rs9889551 -100 -102 -95 gastrin-negative-element [32] R02031
NM_000176 rs10482604 -708 -713 -691 GRFE [33] R03301
NM_000600 rs13447445 -64 -80 -64 NF-kappaB-IL-6 [34] R01634
NM_000208 rs1864010 -494 -511 -486 Sp1-insulin_receptor [35] R03287
NM_002123 rs2854271 -106 -117 -99 HLA-DQB1-Xbox [36] R03695
NM_001063 rs8177185 -617 -623 -599 transferrin-undefined-site [37] R01453
Discussion
Gene promoters for this study were derived using the human genome contig sequences and the RefSeq gene sequences, which are synthesis products of NCBI GenBank [12] records and other information sources. One concern is whether the gene sequences have reliable transcriptional start sites. The traditional experimental methods for identifying transcriptional start sites, such as S1 nuclease mapping [13], primer extension [14] and 5' RACE [15] are technically difficult and not always reliable. Consequently, the Eukaryotic Promoter Database [16], which collects the data from literatures, has only several hundred data entries.
In contrast, the Database of Transcriptional Start Sites [17] uses the full length cDNA library created by oligo-capping [18] to capture the longest mRNA transcript, and has accumulated more than 400,000 sequence entries. These cDNA sequences have been compared with the gene sequences in the RefSeq database to find the target genes and used to study the difference of the transcriptional start sites in the two datasets. According to DBTSS, for 8397 full length cDNA sequences whose corresponding RefSeq gene have been identified and compared, more than 85% of them have transcriptional start site differences equal to or less than 50 nucleotides. Thus the RefSeq gene sequences are relatively reliable in defining the transcriptional start sites.
Although more than 85% of RefSeq gene sequences have almost exact transcriptional start sites, it is possible that some of the derived upstream sequences may not be real gene promoters. The impact of false promoters on the observations is minimal however, as the statistics are calculated using composite statistics. The use of composite statistics also reduces potential bias introduced by highly studied genes, where there might be a greater number of SNPs simply due to a greater number of sequenced individuals, and bias due to the greater interest in the 200 bp immediately upstream of the transcription start site. The normal distribution of binding sites per promoter also argues against any bias due to frequently studied genes, because if there were such study bias we would expect a bimodal distribution with the frequently studied genes having more SNPs forming one mode, and the less studied genes with fewer SNPs forming the second.
Transcription factor binding site prediction algorithms tend to over-predict sites. However, correlation of experimentally determined sites with the predicted sites showed no significant deviation in the number of SNPs falling within confirmed or predicted transcription factor binding sites. The nucleotide frequency of SNPs in experimental transcription factor binding site sequence is 0.25%, which is comparable with the 0.20% nucleotide frequency of promoter SNPs in the putative TF binding sites, and significantly higher than the 0.13% overall nucleotide frequency in the promoter sequences used in this study. Thus we are reasonably certain that the observations reflect a real phenomenon.
When transcription factor binding sites are mapped on gene promoter regions, the anchoring transcriptional start site forms a separating point as shown in Figure 2. Upstream of the transcriptional start site, occurrence frequency of binding sites goes up, whereas the frequency is reduced downstream of the start site. It is clear that SNP distribution in gene promoter sequences is not even. SNPs occur more often in regions close to transcriptional start site than further regions, as is shown Figure 1. This conforms to other observations that the SNP distribution in human chromosomes is not uniform [19].
In general, SNP frequency is directly related to the evolutionary pressure on the target genome regions. For example, more SNPs are accumulated in repetitive sequence, introns and pseudogenes, as the evolution pressure in these regions is relatively low compared to functional gene sequence regions. Given this hypothesis, it is difficult to explain why more SNPs are observed in the sequence region close to transcriptional start site, as the region is important to the initiation of gene transcription, and sequence alteration has potentials to influence gene expression.
One possible explanation is that higher SNP frequency is related to important functions of the regions close to transcription start sites. The accumulation of SNPs in the human genome is like a snapshot of human evolutionary history in which genes, especially those with specific functions, are under continuous natural selection pressure and alteration by mutation, genetic drift and gene flow. As a result, the expression pattern of a gene may be changed. While some genes become totally inactive, others experience expression level alteration. It is possible that SNPs occurring in gene promoter regions play an important role in such scenario, so that the higher frequency of SNPs close to transcriptional start site is related to subtle alteration of gene expression which results in population diversity.
Nucleotide changes caused by SNPs can be classified as transitions and transversions. Generally, transitions are more conservative, because the substituted nucleotides belong to the same chemical group, purine or pyrimidine. In contrast, SNPs resulting in transversions involve nucleotide substitution across purine and pyrimidine chemical groups. The different chemical structures normally limit the occurrence of SNPs of transversion type, however such SNPs have increased frequency in regions close to transcriptional start site as shown in Figure 1. Comparison of nucleotide substitution rates for SNPs in gene promoter regions versus the entire genome also demonstrates that transversion SNPs are increased in promoter regions. It is notable that SNPs with C/G substitution are significantly increased, suggesting that SNPs with unfavorable structure substitution are indeed selected for in the gene promoter regions close to transcriptional start site.
One explanation for the increased SNPs of transversion type may lie in the gene transcription mechanism. During gene transcription, the DNA double helix needs to be opened so that one strand can be used as the template for producing mRNA. In the DNA double helix structure, there are two hydrogen bonds between A and T base pairs and three hydrogen bonds between G and C base pairs, thus opening a G-C base pair requires more energy than an A-T base pair opening does. GC-rich regions often require formation of protein-DNA complexes to facilitate helix opening. Involvement of proteins allows tighter control mechanisms on gene transcription so that gene expression can more accurately respond to environmental conditions. An increased frequency of SNPs with C/G substitutions may be evolutionarily selected for to maintain higher GC content and allowing changes of transcription regulation based on sequence alteration.
It is curious that there are more SNPs in binding site regions than in normal sequence regions. This is contradictory to the hypothesis that transcription factor binding sites are highly conserved because they are important to gene transcription regulation. It is possible that the observations are an artifact and the identified putative transcription factor binding sites are not real binding sites for human genes. As the binding site sequence patterns used are in the mammalian category of the TFD database, some transcription factor binding sites may belong to organisms other than human. However, the false positive binding sites would be simply normal sequence regions in gene promoter which should have the lower genomic SNP frequency, biasing the frequency downward rather than upward.
A second potential artefact is quality control in the transcription factor binding site data used. While data redundancy may exist as shown in Table 2, it is possible that some binding site sequence patterns are not exhaustive. Thus, identified SNPs in the binding sites may have no influence on the binding site integrity and therefore will not affect transcription factor binding and gene expression. This hypothesis cannot be ruled out computationally and will require experimental results to confirm if a SNP found in a transcription factor binding site can exert a real effect on gene expression. If not, the sequence pattern of the binding site needs to be expanded. However, the sequence patterns used for prediction represent consensus sequences drawn from experimental data, and are probably relatively stable and exhaustive.
A third explanation for increased observation of SNPs in the binding sites is that SNPs are involved in altering gene expression during evolution through affecting the binding site sequences. Effects of SNPs in the binding site is likely not simple. While some binding sequence changes made by SNPs may totally interrupt gene expression, others may only influence the level of expression. Considering that gene transcription is a complex process involving many transcription factors, a single position change may not influence all of them. In evolution, the requirement for adjusting certain gene expression level to certain environmental factors forms a natural selection on gene regulation, on which SNPs occurring in transcription factor binding site have direct effect. Therefore, the fact that more SNPs are in the binding sites may demonstrate diverse requirements for different gene expression under different condition.
Conclusion
A biological explanation for increased observation of SNPs in the binding sites may be that SNPs are involved in altering gene expression during evolution by affecting the binding site sequences. The effects of SNPs in the binding site is likely not simple. While some binding sequence alterations made by SNPs may totally interrupt gene expression, others may only influence the level of expression. Considering that gene transcription is a complex process involving many transcription factors, a single position change may not influence all of them.
In evolution, the requirement for adjusting certain gene expression level to certain environmental factors forms a natural selection on gene regulation, on which SNPs occurring in transcription factor binding site have direct effect. Therefore, the fact that more SNPs are in the transcription factor binding sites may demonstrate diverse requirements for different gene expression under different condition. Thus, promoter SNPs may be an active factor in natural selection on gene transcription.
Methods
Data from four different sources (LocusLink, Genome, TFD, and dbSNP) were used to generate the promoter SNP data set. At each step, a number of steps to clean and normalize the data were undertaken. Figure 4 shows the general steps and data sources used in our investigation.
Figure 4 Flow chart of analysis steps used to generate data set. A high-level view of the methods used to generate the data sets used in this study. Canonical data sources are shown in bold, and general steps are shown in italics. Parenthetical numbers indicate the number of unique data elements in the data set.
Gene promoter identification
Gene promoter sequences were calculated using human genome contig sequences (build 33) from NCBI and gene location information from LocusLink. For single gene loci, the gene was mapped onto its target contig using the scope value, reverse complementing the contig sequence as necessary. The 2 kb sequence immediately upstream of the transcriptional start site was then extracted and used as the promoter sequence. For genes located closer than 2 kb to the start of contig sequence, the length of the extracted upstream sequence was limited to the available sequence. Loci having more than one gene suggest that genes are co-transcribed, or gene products resulting from alternative splicing. Therefore, the start site of the first gene in the locus was used to derive the upstream sequence, which was used as the promoter sequence for all of the genes in the locus. The few conflicts from data discrepancies across different NCBI databases were resolved by forcing agreement on sequence version in the data extraction pipeline.
SNP distribution in gene promoters
Promoter SNPs were mapped using the SNP location information from dbSNP [6]. Non-unique and loosely mapped SNPs accounted for only 5.5% for the total number of SNPs and were excluded. SNPs were then classified according to the substitution type as either transitions, which includes nucleotide substitution within the purine (A/G) or pyrimidine (C/T) group, or as transversions, which represents substitution across the groups. The general distribution pattern of SNPs was then computed.
Data redundancy of transcription factor binding sites
Transcription factor binding site sequences were acquired from the Transcription Factor Database (TFD) [20]. The database contains more than three thousand mammalian binding site sequence entries encoded using the nomenclature of International Union of Pure and Applied Chemistry (IUPAC) [21]. Inspection of the sequence data revealed that some sequences differ at only one or two nucleotides, and some are subsequences of other sequences. To evaluate possible data redundancy, the binding site sequences were mutually compared and the sequence pairs were grouped into three categories: exact coverage (a complete match), partial coverage (one is a subsequence of the other) and single nucleotide difference (differing at an internal nucleotide). Binding sites in the partial coverage and single nucleotide difference categories were merged.
Transcription factor binding site identification and distribution in gene promoters
Transcription factor binding sites were predicted using a suffix tree algorithm [22,23]. The IUPAC codes were fully expanded and placed into a suffix tree, which allows the exact promoter sequence string to be searched against the binding site motifs while preserving the degeneracy of the binding sites. The probability of the binding site motif appearing by chance was computed using an expectation value given by the equation , where k is the number of possible nucleotides at position i.
The distribution of binding sites in gene promoters was examined in order to identify the relation between occurrence of binding sites and location in gene promoter regions. Calculations were performed only upon the full length (2 K bp) promoters sequence. A 200 bp portion of the gene subsequence just downstream of the transcriptional start site was also included with the promoter sequences. Expectation values of 0.1 and 0.01 were used as the control for binding site significance.
To evaluate the chance appearance of transcription factor binding sites, we computed their occurrences in two HMM-generated random datasets that simulate the promoter sequence range of [-2000, -1]. The first dataset was created by assigning equal an emission probability to all four nucleotides at each position, while for the second dataset the emission probability of nucleotide was determined according to the observed nucleotide probabilities from the real promoter dataset. Both simulated data sets contained the same number of sequences as the real promoter set.
SNP distribution in putative binding sites
Integrity of transcription factor binding sites was analyzed by comparing the SNP distributions within and outside the putative transcription binding sites. The transcription factor binding site data set contained both the binding sites directly identified from the canonical sequence, and the binding sites "recovered" by substituting previously identified SNP into the sequence.
SNP distribution in experimentally derived binding sites
Experimentally derived transcription factor binding sites were retrieved from TRANSFAC database 6.0 [24], the eukaryotic transcription factors database. To evaluate whether SNPs occur in the experimentally derived TF binding sites, we linked the predicted TF binding sites with the binding sites in TRANSFAC using RefSeq literature references. The initial linkage was then confirmed using binding site sequence patterns, site locations and the order of the binding sites in gene promoter.
Authors' contributions
YG conceived of the study and carried out the coding and statistical analysis. DCJ participated in its design and coordination, and helped draft the manuscript.
==== Refs
Smigielski EM Sirotkin K Ward M Sherry ST dbSNP: a database of single nucleotide polymorphisms Nucleic Acids Res 2000 28 352 355 10592272 10.1093/nar/28.1.352
Zimdahl H Nyakatura G Brandt P Schulz H Hummel O Fartmann B Brett D Droege M Monti J Lee YA Sun Y Zhao S Winter EE Ponting CP Chen Y Kasprzyk A Birney E Ganten D Hubner N A SNP map of the rat genome generated from cDNA sequences Science 2004 303 807 14764869 10.1126/science.1092427
Zhu YL Song QJ Hyten DL Van Tassell CP Matukumalli LK Grimm DR Hyatt SM Fickus EW Young ND Cregan PB Single-nucleotide polymorphisms in soybean Genetics 2003 163 1123 1134 12663549
Collins FS Morgan M Patrinos A The Human Genome Project: lessons from large-scale biology Science 2003 300 286 290 12690187 10.1126/science.1084564
Sachidanandam R Weissman D Schmidt SC Kakol JM Stein LD Marth G Sherry S Mullikin JC Mortimore BJ Willey DL Hunt SE Cole CG Coggill PC Rice CM Ning Z Rogers J Bentley DR Kwok PY Mardis ER Yeh RT Schultz B Cook L Davenport R Dante M Fulton L Hillier L Waterston RH McPherson JD Gilman B Schaffner S Van Etten WJ Reich D Higgins J Daly MJ Blumenstiel B Baldwin J Stange-Thomann N Zody MC Linton L Lander ES Altshuler D A map of human genome sequence variation containing 1.42 million single nucleotide polymorphisms Nature 2001 409 928 933 11237013 10.1038/35057149
Sherry ST Ward M Sirotkin K dbSNP-database for single nucleotide polymorphisms and other classes of minor genetic variation Genome Res 1999 9 677 679 10447503
Ramensky V Bork P Sunyaev S Human non-synonymous SNPs: server and survey Nucleic Acids Res 2002 30 3894 3900 12202775 10.1093/nar/gkf493
Clifford RJ Edmonson MN Nguyen C Buetow KH Large-scale analysis of non-synonymous coding region single nucleotide polymorphisms Bioinformatics 2004 20 1006 1014 14751981 10.1093/bioinformatics/bth029
Ober C Aldrich CL Chervoneva I Billstrand C Rahimov F Gray HL Hyslop T Variation in the HLA-G promoter region influences miscarriage rates Am J Hum Genet 2003 72 1425 1435 12721954 10.1086/375501
Kiewitz C Tummler B Similar profile of single nucleotide substitution types in bacteria and human genetic disease Genome Letters 2002 1 111 114 10.1166/15373050260506700
Kel-Margoulis OV Tchekmenev D Kel AE Goessling E Hornischer K Lewicki-Potapov B Wingender E Composition-sensitive analysis of the human genome for regulatory signals In Silico Biol 2003 3 145 171 12954097
Benson DA Karsch-Mizrachi I Lipman DJ Ostell J Wheeler DL GenBank Nucleic Acids Res 2003 31 23 27 12519940 10.1093/nar/gkg057
Wong HC Schnepf HE Whiteley HR Transcriptional and translational start sites for the Bacillus thuringiensis crystal protein gene J Biol Chem 1983 258 1960 1967 6296116
Selvanayagam CS Tsai SY Tsai MJ Selvanayagam P Saunders GF Multiple origins of transcription for the human placental lactogen genes J Biol Chem 1984 259 14642 14646 6209277
Wu T Ikezono T Angus CW Shelhamer JH Characterization of the promoter for the human 85 kDa cytosolic phospholipase A2 gene Nucleic Acids Res 1994 22 5093 5098 7800505
Cavin Perier R Junier T Bucher P The Eukaryotic Promoter Database EPD Nucleic Acids Res 1998 26 353 357 9399872 10.1093/nar/26.1.353
Suzuki Y Yamashita R Nakai K Sugano S DBTSS: DataBase of human Transcriptional Start Sites and full-length cDNAs Nucleic Acids Res 2002 30 328 331 11752328 10.1093/nar/30.1.328
Suzuki Y Yoshitomo-Nakagawa K Maruyama K Suyama A Sugano S Construction and characterization of a full length-enriched and a 5'-end-enriched cDNA library Gene 1997 200 149 156 9373149 10.1016/S0378-1119(97)00411-3
Balasubramanian S Harrison P Hegyi H Bertone P Luscombe N Echols N McGarvey P Zhang Z Gerstein M SNPs on human chromosomes 21 and 22 -- analysis in terms of protein features and pseudogenes Pharmacogenomics 2002 3 393 402 12052146 10.1517/14622416.3.3.393
Ghosh D Object-oriented transcription factors database (ooTFD) Nucleic Acids Res 2000 28 308 310 10592257 10.1093/nar/28.1.308
Dybkaer R International Union of Pure and Applied Chemistry and International Federation of Clinical Chemistry. IUPAC Section of Clinical Chemistry; Commission on Quantities and Units in Clinical Chemistry and IFCC Committee on Standards, Expert Panel on Quantities and Units; Approved Recommendation (1978). Quantities and Units in Clinical Chemistry Clin Chim Acta 1979 96 157F 183F 476957 10.1016/0009-8981(79)90065-2
Weiner P Linear pattern matching algorithms. 1973 1 11
McCreight EM A space-economical suffix tree construction algorithm. Journal of Algorithms 1976 23 262 272
Wingender E Chen X Fricke E Geffers R Hehl R Liebich I Krull M Matys V Michael H Ohnhauser R Pruss M Schacherer F Thiele S Urbach S The TRANSFAC system on gene expression regulation Nucleic Acids Res 2001 29 281 283 11125113 10.1093/nar/29.1.281
Paulweber B Levy-Wilson B The mechanisms by which a human apolipoprotein B gene enhancer and reducer interact with the promoter are different in cultured cells of hepatic and intestinal origin J Biol Chem 1991 266 24161 24168 1660893
Metzger S Halaas JL Breslow JL Sladek FM Orphan receptor HNF-4 and bZip protein C/EBP alpha bind to overlapping regions of the apolipoprotein B gene promoter and synergistically activate transcription J Biol Chem 1993 268 16831 16838 8344962
Fisch TM Prywes R Roeder RG c-fos sequence necessary for basal expression and induction by epidermal growth factor, 12-O-tetradecanoyl phorbol-13-acetate and the calcium ionophore Mol Cell Biol 1987 7 3490 3502 3119989
Bergemann AD Johnson EM The HeLa Pur factor binds single-stranded DNA at a specific element conserved in gene flanking regions and origins of DNA replication Mol Cell Biol 1992 12 1257 1265 1545807
Postel EH Berberich SJ Flint SJ Ferrone CA Human c-myc transcription factor PuF identified as nm23-H2 nucleoside diphosphate kinase, a candidate suppressor of tumor metastasis Science 1993 261 478 480 8392752
Englander EW Wilson SH Protein binding elements in the human beta-polymerase promoter Nucleic Acids Res 1990 18 919 928 2315044
Kedar PS Widen SG Englander EW Fornace AJJ Wilson SH The ATF/CREB transcription factor-binding site in the polymerase beta promoter mediates the positive effect of N-methyl-N'-nitro-N-nitrosoguanidine on transcription Proc Natl Acad Sci USA 1991 88 3729 3733 1827204
Wang TC Brand SJ Islet cell-specific regulatory domain in the gastrin promoter contains adjacent positive and negative DNA elements J Biol Chem 1990 265 8908 8914 2187875
Leclerc S Xie BX Roy R Govindan MV Purification of a human glucocorticoid receptor gene promoter-binding protein. Production of polyclonal antibodies against the purified factor J Biol Chem 1991 266 8711 8719 2026589
Ray A LaForge KS Sehgal PB On the mechanism for efficient repression of the interleukin-6 promoter by glucocorticoids: enhancer, TATA box, and RNA start site (Inr motif) occlusion Mol Cell Biol 1990 10 5736 5746 2233715
Araki E Murakami T Shirotani T Kanai F Shinohara Y Shimada F Mori M Shichiri M Ebina Y A cluster of four Sp1 binding sites required for efficient expression of the human insulin receptor gene J Biol Chem 1991 266 3944 3948 1995641
Hasegawa SL Boss JM Two B cell factors bind the HLA-DRA X box region and recognize different subsets of HLA class II promoters Nucleic Acids Res 1991 19 6269 6276 1956787
Shakhov AN Collart MA Vassalli P Nedospasov SA Jongeneel CV Kappa B-type enhancers are involved in lipopolysaccharide-mediated transcriptional activation of the tumor necrosis factor alpha gene in primary macrophages J Exp Med 1990 171 35 47 2104921 10.1084/jem.171.1.35
|
16209714
|
PMC1260019
|
CC BY
|
2021-01-04 16:32:48
|
no
|
BMC Genomics. 2005 Oct 6; 6:140
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-140
|
oa_comm
|
==== Front
BMC Med GenetBMC Medical Genetics1471-2350BioMed Central London 1471-2350-6-361620217210.1186/1471-2350-6-36Technical AdvanceSequence diversity within the HA-1 gene as detected by melting temperature assay without oligonucleotide probes Graziano Claudio [email protected] Massimo [email protected] Cecilia [email protected] Pier Luigi [email protected] Berardino [email protected] Human Genetics Unit, Department of Clinical Physiopathology, University of Florence, Viale Gaetano Pieraccini 6, 50139 Firenze, Italy2 Current address: U.O. e Cattedra di Genetica Medica, Policlinico S. Orsola-Malpighi, Via Massarenti 9, 40138 Bologna, Italy2005 4 10 2005 6 36 36 28 2 2005 4 10 2005 Copyright © 2005 Graziano 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 minor histocompatibility antigens (mHags) are self-peptides derived from common cellular proteins and presented by MHC class I and II molecules. Disparities in mHags are a potential risk for the development of graft-versus-host disease (GvHD) in the recipients of bone marrow from HLA-identical donors. Two alleles have been identified in the mHag HA-1. The correlation between mismatches of the mHag HA-1 and GvHD has been suggested and methods to facilitate large-scale testing were afterwards developed.
Methods
We used sequence specific primer (SSP) PCR and direct sequencing to detect HA-1 gene polymorphisms in a sample of 131 unrelated Italian subjects. We then set up a novel melting temperature (Tm) assay that may help identification of HA-1 alleles without oligonucleotide probes.
Results
We report the frequencies of HA-1 alleles in the Italian population and the presence of an intronic 5 base-pair deletion associated with the immunogeneic allele HA-1H. We also detected novel variable sites with respect to the consensus sequence of HA-1 locus. Even though recombination/gene conversion events are documented, there is considerable linkage disequilibrium in the data. The gametic associations between HA-1R/H alleles and the intronic 5-bp ins/del polymorphism prompted us to try the Tm analysis with SYBR® Green I. We show that the addition of dimethylsulfoxide (DMSO) during the assay yields distinct patterns when amplicons from HA-1H homozygotes, HA-1R homozygotes, and heterozygotes are analysed.
Conclusion
The possibility to use SYBR® Green I to detect Tm differences between allelic variants is attractive but requires great caution. We succeeded in allele discrimination of the HA-1 locus using a relatively short (101 bp) amplicon, only in the presence of DMSO. We believe that, at least in certain assets, Tm assays may benefit by the addition of DMSO or other agents affecting DNA strand conformation and stability.
==== Body
Background
Acute graft-versus-host disease (aGvHD) is still a major cause of morbidity after allogeneic HLA-identical bone marrow transplantation, occurring in 10–60% of patients receiving matched sibling allograft, depending on prophylaxis regimen. These figures turned out to be even higher in the case of unrelated matched allograft [1]. Recent studies emphasize the involvement of mHags disparities in the development of aGvHD [2-4].
Among known autosomal mHags, only HA-1 has been implicated as a cause of aGvHD in humans [2]. HA-1 is a nonapeptide from a protein encoded by a gene termed KIAA0223 (GenBank accession no. D86976), a polymorphic gene that has two known alleles differing at positions 500 and 504 of the cDNA sequence, resulting in a single aminoacid change. The HA-1H allele encodes histidine at position 3 of the peptide, is recognized by HLA-A*0201-restricted cytotoxic T cells and is only expressed by cells of haematopoietic origin [5]. Its allelic counterpart, HA-1R, encodes arginine at position 3. HLA-A*0201 molecules have low affinity for the HA-1R peptide and the complex does not generate a detectable immune response [5]. HA-1 disparity can thus be defined as the presence of HA-1H in the recipient but not in the donor, because in such cases T cells in the transplanted donor marrow respond to mHags in the recipient.
Four different DNA-based strategies have been described so far to perform HA-1 allelotyping. They rely on sequence specific primer (SSP) PCR [6], restriction fragment length polymorphism (RFLP) PCR [7], reference strand mediated conformation analysis (RSCA) [8], and allele specific fluorescence-labelled probes [9].
Methods
HA-1 allelotyping
We performed HA-1 typing through allele-specific PCR as described by Wilke et al [6], with minor modifications. Two different primer sets (designed on both strands) were used: each set contained a common external primer and specific primers for allele HA-1H and HA-1R (Figure 1).
The two common primers were used to generate a fragment of an expected length of 486 bp containing the polymorphic sites. Relative positions of primer pair sets A and B designed to perform allele-specific PCR typing at the HA-1 locus are reported in figure 1. Forty ng of genomic DNA was used in 50 μL of a reaction mixture containing Applied Biosystems PCR Buffer with 1.5 mmol/L MgCl2, 15 pmol of each primer, 0.8 mmol/L dNTPs, and 2 units of AmpliTaq polymerase (Applied Biosystems). Cycling conditions were according to Wilke et al [6]. Amplicons were purified with the QIAquick Purification Kit (Qiagen). Sequencing was performed on the ABI PRISM 310 Genetic Analyzer according to the protocol provided by the manufacturer.
Melting curve analysis
The primers used to amplify a 101-bp fragment were: forward, 5'-CTTCGCTGAGGGCCTTGAG-3' and reverse, 5'-CCTTGGGTCTGGCTCTGTCTT-3'. The reactions were performed in a total volume of 100 μL containing 50 μL of SYBR® Green PCR Master Mix (Applied Biosystems), a 300/300 mmol/L forward/reverse primer combination and 50 ng of genomic DNA. The cycling conditions were: 94°C for 45 sec, 60°C for 45 sec and 72°C for 45 sec in 33 cycles. After PCR, DMSO at 5% or 10% was added to the reaction tubes and the total volume was divided in 4 wells of an optical reaction plate. The dissociation protocol was performed in accordance to the default thermal profile of the ABI 5700 Sequence Detection System (Applied Biosystems), which included, after a 15 sec hold at 95°C and a 20 sec hold at 60°C, a slow ramp (20 min) from 60°C to 95°C. The fluorescence was monitored in real time (every 3 sec) for each sample. The melting peaks were calculated by the ABI 5700 software as the negative derivative of fluorescence with respect to the temperature (-dF/dT vs T).
Results and discussion
We report our experience with melting temperature (Tm) analysis, SSP-PCR, and direct sequencing in detecting the HA-1 polymorphism in a cohort of Italian subjects, which allowed us to disclose further polymorphic variations surrounding the two HA-1 alleles and, in particular, an intronic 5-bp deletion, strongly associated with HA-1H allele.
We sampled 131 unrelated subjects from the Italian population, obtaining a genotypic distribution of 23 HA-1H homozygotes, 41 HA-1R homozygotes, and 67 heterozygotes. The resulting allelic frequencies were 0.43 for allele HA-1H and 0.57 for allele HA-1R. The observed genotypes conformed to the Hardy-Weinberg law (X2 = 0.23, not significant). Furthermore, our allele frequency estimates are associated with a standard error of 0.03 and are in line with those reported by Tseng et al [7] for North American Caucasians.
HA-1 typing was performed by allele-specific PCR as described by Wilke et al [6], with minor modifications (Figure 1). Occasionally, we have obtained ambiguous typing results using this approach. The same difficulties have been mentioned by Kreiter et al [9]. These findings prompted us to sequence the entire region. In comparison with the sequence retrieved from GenBank under accession number AF092537, which corresponds to the HA-1R, we identified an intronic 5-bp deletion in the first case examined. Therefore, we decided to analyse a random sample from our population by sequencing the HA-1 genomic region in a total of 36 subjects. The results are summarised in Table 1. Twenty-seven of 72 chromosomes (37.5%) showed the 5-bp deletion. Further polymorphic variation was detected at c.509+65 and at c.509+130. Four of 72 chromosomes (5.6%) showed G to A transitions at those intronic sites. Haplotypes A and B are present in the Italian population at comparable frequencies. They differ at the sites defining the HA-1R/H polymorphism as well as at c.509+213 where the pentanucleotide TTTAT is missing in haplotype B. This deletion was also identified by us in subjects from Sardinia, Germany and China, indicating that it is widespread in human populations, possibly underlying its ancient origin. We deposited this sequence under GenBank accession number AF236756. In the meanwhile, the same ins/del polymorphism was reported by Aróstegui et al [8], which also described the occurrence of "del" with HA-1H allele and that of "ins" with HA-1R. Our data showed that these gametic associations are in nearly complete linkage disequilibrium and that further variation is present within the HA-1 gene. Haplotype C is by far less frequent than haplotypes A and B. It is characterised by the presence of G to A transitions at c.509+65 and c.509+310 in a haplotype A context. A relatively recent gene conversion event, which may have occurred on this milieu, is the most parsimonious path we can envisage. The same ancestral haplotype A may have harboured a silent exonic transition (c.427C→T) to generate haplotype E. Whether this variation is polymorphic or merely a rare mutation remains to be verified on larger samples. We also found a single chromosome with the HA-1R allele associated with the intronic deletion (haplotype D). This chromosome may well have raised from a crossing-over in a HA-1R/H heterozygote.
In order to make HA-1 typing easier and faster, we tested the Tm approach to define HA-1 alleles. As reported, the ability of Tm analysis to discriminate different PCR products mainly depends on the length of the amplicon and the GC content [11,12]. Both amplification and Tm analysis are automated processes that can be performed on fluorescence-detection equipped thermal cyclers. Double-stranded DNA products are detected by the SYBR® Green I nucleic acid dye in real-time during PCR. At the end of the reaction the temperature at which the transition double/single strand DNA occurs is revealed by following SYBR® Green I emission decay. Different PCR products may determine specific dissociation profiles [11,12].
As a matter of fact, knowledge of the gametic associations between HA-1R/H alleles and the intronic 5-bp ins/del polymorphism might allow a straightforward Tm analysis combining H/R allele specific primer and the presence of the 5-bp deletion in a single amplicon. However, the fragment would be too large to allow correct typing (data not shown). In this respect, our results are in keeping with the criticisms raised by von Ahsen et al [13] towards Tm assays based on SYBR® Green I melting curves [14,15].
We tried to overcome those limitations and designed a pair of primers to amplify a 101-bp fragment including only the polymorphic positions in the coding sequence of HA-1 gene. Quality controls included the assessment of consistency in testing and re-testing the same specimens using independently generated PCR products. The Tm analysis with SYBR® Green I was routinarily carried out in quadruplicate and turned out to allow discrimination of the H/R alleles by adding DMSO before the dissociation protocol (Figure 2, panel C). As depicted in panels A and B of Figure 2, the heterozygous pattern is clearly distinguishable from the two homozygous curves because of a shoulder, most likely due to the presence of heteroduplexes [12]. In our experience we obtained allele discrimination only by adding DMSO. It is well known that DMSO and other reagents such as formamide and glycerol can affect conformation and stability of nucleic acid strands. Indeed, we found useful to add DMSO in order to discriminate between H/R alleles and we believe that similar methods could find application in Tm analysis of certain PCR products where ins/del polymorphisms and/or SNPs are present.
Conclusion
We detected mutations in novel sites relative to a consensus sequence of HA-1 locus. We showed that these variations are arranged onto few rare haplotypes possibly generated through recombination/gene conversion events from two major ancestral haplotypes. The observed sequence diversity, besides HA-1R/H, involves either a synonymous codon or intronic changes making them unlikely direct determinants of allogeneic recognition by cytotoxic T cells.
Probe-free, Tm assays most often require confirmation by other DNA mutation detection assays. However, we propose that Tm-based methods may be successfully applied to find experimental conditions, such as the use of DMSO described here, for genotyping ins/del and/or single nucleotide polymorphisms.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
C.G. carried out SSP-PCR allelotyping, performed sequence analyses and helped in drafting the manuscript. M.G. carried out the Tm assay, participated in the sequence alignment and helped in drafting the manuscript. C.M. participated in SSP-PCR allelotyping and in sequence analyses. P.L.M. participated in the design of the study and performed the statistical analysis. B.P. conceived the study, participated in its whole design and coordination and edited the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was supported by Fondazione Rulfo, Parma and by funds ex MURST 60%. C.G. and M.G. are the recipients of post-doc fellowships of the University of Florence.
Figures and Tables
Figure 1 Relative positions of primer pair sets A and B designed to perform allele-specific PCR typing at the HA-1 locus. A(1) is the forward primer common to site-specific primers A(2) and A(3) designed with the 3'OH-end at c.500C/T. B(4) is the reverse primer common to site-specific primers B(5) and B(6) designed with the 3'OH-end at c.504A/G. Primer combinations A(1) + A(2) and B(4) + B(5) are specific for HA-1H. Primer combinations A(1) + A(3) and B(4) + B(6) are specific for HA-1R.
Figure 2 Tm analysis with SYBR® Green I of 101-bp PCR products to discriminate the H/R alleles. Panel A: Melting profile without DMSO. Panel B: Melting profile with DMSO 5%. Panel C: Melting profile with DMSO 10%. Heterozygotes are depicted as thick lines, the heterozygous pattern is distinguishable from the two homozygous curves because of the shoulder, most likely due to the presence of heteroduplexes.
Table 1 HA-1 haplotypes found in 72 random chromosomes of the Italian population.
Name N c.427C/T c.500C/T§ c.504A/G§ c.509+65G/A c.509+130G/A c.509+213 ins/del TTTAT
Aa 39 C T G G G ins
Bb 27 C C A G G del
C 4 C T G A A ins
D 1 C T G G G del
E 1 T T G G G ins
§ HA-1R is defined by c.500T and c.504G; HA-1H is defined by c.500C and c.504A (6)
a This haplotype corresponds to the sequence reported under GenBank accession number AF092537
b This haplotype corresponds to the sequence reported under GenBank accession number AF236756
==== Refs
Ringdén O Deeg HJ Ferrara JLM, Deeg HJ, Burakoff SJ Clinical spectrum of graft-versus-host disease Graft-versus-host disease 1997 New York: Marcel Dekker 525 559
Goulmy E Schipper R Pool J Blokland E Falkenburg JH Vossen J Gratwohl A Vogelsang GB van Houwelingen HC van Rood JJ Mismatches of minor histocompatibility antigens between HLA-identical donors and recipients and the development of graft-versus-host disease after bone marrow transplantation N Engl J Med 1996 334 281 285 8532022 10.1056/NEJM199602013340501
Tseng LH Lin MT Hansen JA Gooley T Pei J Smith AG Martin EG Petersdorf EW Martin PJ Correlation between disparity for the minor histocompatibility antigen HA-1 and the development of acute graft-versus-host disease after allogeneic marrow transplantation Blood 1999 94 2911 2914 10515896
Tait BD Maddison R McCluskey J Deayton S Heatley S Lester S Bardy P Szer J Grigg A Spencer A Schwarer A Holdsworth R Clinical relevance of the minor histocompatibility antigen HA-1 in allogeneic bone marrow transplantation between HLA identical siblings Transplantation Proc 2001 33 1760 1761 10.1016/S0041-1345(00)02816-5
den Haan JM Meadows LM Wang W Pool J Blokland E Bishop TL Reinhardus C Shabanowitz J Offringa R Hunt DF Engelhard VH Goulmy E The minor histocompatibility antigen HA-1: a diallelic gene with a single amino acid polymorphism Science 1998 279 1054 1057 9461441 10.1126/science.279.5353.1054
Wilke M Pool J den Haan JMM Goulmy E Genomic identification of the minor histocompatibility antigen HA-1 locus by allele-specific PCR Tissue Antigens 1998 52 312 317 9820596
Tseng L-H Lin M-T Martin PJ Pei J Smith AG Hansen JA Definition of the gene encoding the minor histocompatibilty antigen HA-1 and typing for HA-1 from genomic DNA Tissue Antigens 1998 52 305 311 9820595
Arostegui JI Gallardo D Rodriguez-Luaces M Querol S Madrigal JA Garcia-Lopez J Granena A Genomic typing of minor histocompatibility antigen HA-1 by reference strand mediated conformation analysis (RSCA) Tissue Antigens 2000 56 69 76 10958358 10.1034/j.1399-0039.2000.560109.x
Kreiter S Wehler T Landt O Huber C Derigs H-G Hess G Rapid identification of minor histocompatibility antigen HA-1 subtypes H and R using fluorescence-labeled oligonucleotides Tissue Antigens 2000 56 449 452 11144294 10.1034/j.1399-0039.2000.560509.x
Nickerson DA Taylor SL Weiss KM Clark AG Hutchinson RG Stengard J Salomaa V Vartiainen E Boerwinkle E Sing CF DNA sequence diversity in a 9.7-kb region of the human lipoprotein lipase gene Nature Genet 1998 19 233 240 9662394 10.1038/907
Ririe KM Rasmussen RP Wittwer CT Product differentiation by analysis of DNA melting curves during the polymerase chain reaction Anal Biochem 1997 245 154 160 9056205 10.1006/abio.1996.9916
Lipsky RH Mazzanti CM Rudolph JG Xu K Vyas G Bozak D Radel MQ Goldman D DNA melting analysis for detection of single nucleotide polymorphisms Clin Chem 2001 47 635 644 11274012
von Ahsen N Oellerich M Schutz E Limitations of genotyping based on amplicon melting temperature Clin Chem 2001 47 1331 1332 11427477
Marziliano N Pelo E Minuti B Passerini I Torricelli F Da Prato L Melting temperature assay for a UGT1A gene variant in Gilbert syndrome Clin Chem 2000 46 423 425 10702534
Pirulli D Boniotto M Puzzer D Spano A Amoroso A Crovella S Flexibility of melting temperature assay for rapid detection of insertions, deletions, and single-point mutations of the AGXT gene responsible for type 1 primary hyperoxaluria Clin Chem 2000 46 1842 1844 11067824
|
16202172
|
PMC1260020
|
CC BY
|
2021-01-04 16:03:33
|
no
|
BMC Med Genet. 2005 Oct 4; 6:36
|
utf-8
|
BMC Med Genet
| 2,005 |
10.1186/1471-2350-6-36
|
oa_comm
|
==== Front
BMC MicrobiolBMC Microbiology1471-2180BioMed Central London 1471-2180-5-551620213810.1186/1471-2180-5-55Research ArticleIn-house nucleic acid amplification tests for the detection of Mycobacterium tuberculosis in sputum specimens: meta-analysis and meta-regression Flores Laura L [email protected] Madhukar [email protected] John M [email protected] Lee W [email protected] Divisions of Infectious Diseases and Epidemiology, School of Public Health, University of California, Berkeley. CA 94720. USA2 Division of Molecular Biomedicine, CINVESTAV-IPN, Mexico DF, Mexico3 Division of Pulmonary & Critical Care Medicine, San Francisco General Hospital, San Francisco, CA 94110. USA2005 3 10 2005 5 55 55 10 5 2005 3 10 2005 Copyright © 2005 Flores 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
More than 200 studies related to nucleic acid amplification (NAA) tests to detect Mycobacterium tuberculosis directly from clinical specimens have appeared in the world literature since this technology was first introduced. NAA tests come as either commercial kits or as tests designed by the reporting investigators themselves (in-house tests). In-house tests vary widely in their accuracy, and factors that contribute to heterogeneity in test accuracy are not well characterized. Here, we used meta-analytical methods, including meta-regression, to identify factors related to study design and assay protocols that affect test accuracy in order to identify those factors associated with high estimates of accuracy.
Results
By searching multiple databases and sources, we identified 2520 potentially relevant citations, and analyzed 84 separate studies from 65 publications that dealt with in-house NAA tests to detect M. tuberculosis in sputum samples. Sources of heterogeneity in test accuracy estimates were determined by subgroup and meta-regression analyses. Among 84 studies analyzed, the sensitivity and specificity estimates varied widely; sensitivity varied from 9.4% to 100%, and specificity estimates ranged from 5.6% to 100%. In the meta-regression analysis, the use of IS6110 as a target, and the use of nested PCR methods appeared to be significantly associated with higher diagnostic accuracy.
Conclusion
Estimates of accuracy of in-house NAA tests for tuberculosis are highly heterogeneous. The use of IS6110 as an amplification target, and the use of nested PCR methods appeared to be associated with higher diagnostic accuracy. However, the substantial heterogeneity in both sensitivity and specificity of the in-house NAA tests rendered clinically useful estimates of test accuracy difficult. Future development of NAA-based tests to detect M. tuberculosis from sputum specimens should take into consideration these findings in improving accuracy of in-house NAA tests.
==== Body
Background
Accurate and early diagnosis of tuberculosis (TB) is a critical step in the management and control of TB. Because conventional tests for TB have several limitations, nucleic acid amplification (NAA) tests have emerged as promising alternatives. The polymerase chain reaction (PCR) is the best-known and most widely used NAA test. NAA tests are categorized as commercial (kit-based) or in-house ("home-brew"). In-house tests are those assays where the investigators design their own protocols. In-house tests are commonly used in developing countries where commercial kits may not be affordable. The accuracy of NAA tests for TB has been extensively studied since the early 1990s. Since hundreds of studies have evaluated the accuracy of NAA tests, it is now possible to evaluate their overall performance using meta-analysis methods and determine which study design or test-related characteristics are associated with higher diagnostic accuracy.
Several systematic reviews and meta-analyses have been published in the past few years on the accuracy of NAA tests for pulmonary and extra-pulmonary TB [1,5]. Because these meta-analyses and reviews have synthesized data from over 200 primary studies, and because their results are highly consistent with each other, they provide us with the best available evidence on the accuracy of NAA tests. The following are the main findings of the meta-analyses and reviews: most of the studies on NAA tests reported very high estimates of specificity, for both pulmonary and extra-pulmonary TB [1-5]. Sensitivity estimates, in contrast, have been much lower and highly variable (heterogeneous) [1-5]. Sensitivity estimates have been lower in paucibacillary forms of TB (smear-negative pulmonary and extra-pulmonary TB), and higher in smear-positive pulmonary TB [2,3]. Another striking result is the widespread lack of consistency in accuracy estimates across studies – studies have reported highly variable estimates of test accuracy [1-5]. For example, in our previous meta-analysis on NAA tests for tuberculous meningitis, sensitivity estimates varied between 0 – 100% [3]. In general, almost all the meta-analyses have demonstrated that the sensitivity and specificity of in-house PCR assays have been more variable and inconsistent than commercial tests [2,3,5].
Why do studies on in-house PCR assays produce such highly variable estimates of sensitivity and specificity? Is the variability due to differences in study design or to differences in assay characteristics and laboratory techniques? Are there specific study design features and assay characteristics that yield higher estimates of accuracy? Answers to these questions might help to identify features of NAA tests that maximize accuracy. However, these questions are difficult to address in individual studies. Meta-analytic methods, on the other hand, are well suited to explore the issue of why studies produce variable results. By synthesizing data from multiple studies and increasing the power of analyses, meta-analyses are able to employ techniques that help to identify sources of heterogeneity in study findings. In this meta-analysis, we reviewed 65 published studies on in-house NAA tests for pulmonary TB. The main objective of our meta-analysis was to determine factors associated with heterogeneity as well as higher accuracy estimates of accuracy in studies that evaluated in-house PCR for the diagnosis of pulmonary TB.
Results
Study selection
By searching multiple databases and sources we identified 2520 potentially relevant citations on NAA tests for tuberculosis. After screening titles and abstracts, 434 English and Spanish articles were selected for full-text review and 129 articles reported inclusion of sputum specimens tested by an in-house PCR assay. Sixty-one articles were then excluded mainly because data were not separately provided for sputum samples (sputum and other clinical specimens were analyzed together). Also, three articles were excluded because real time PCR was used and the number was insufficient to place them in a separate category. A total of 65 articles [10-74], were included in the final analysis. Four articles were in Spanish [19,41,46,53]. Thirteen articles reported evaluations of more than one NAA test against a common reference standard [11,13,14,20,28,36,47,54,60,62,68,69,72]. Each such test comparison was counted as a separate study. Thus, the total number of test comparisons (hereafter referred to as studies) was 84.
Characteristics of included studies
The summary characteristics of the included studies are shown in Table 1. The average sample size of the included studies was 149 (range 14 to 727). Our data, as seen in Table 1, were affected by the poor quality of reporting in the primary studies. Fifty-five of 84 (65.5%) studies did not report blinded interpretation of NAA test independent of the reference standard, while only 29 of 84 (34.5%) reported single or double blinding of NAA test and the reference standard. Most of the studies, 60 of 84 (72%), were cross-sectional whereas 24 (28%) were case-control studies. The studies differed greatly in terms of laboratory characteristics. Fifty-four of 84 (65%) studies used IS6110 as amplification target by itself or in combination with other targets, and 30 studies (35%) used other targets (e.g. MPB64, 38 kDa). The studies were categorized as those using any chemical method for DNA extraction (including phenol-chloroform) and those in which any physical or mechanical extraction method was used. Sixty-eight of 84 (81%) studies reported a simple PCR protocol (including multiplex PCR), whereas 16 (19%) studies used a nested or seminested PCR protocol. Lastly, 49 of 84 (58%) studies used UV transillumination of an electrophoretic gel, and 35 (42%) used DNA hybridization probes to detect the amplification products.
Overall diagnostic accuracy
When all 84 studies were evaluated together, the sensitivity estimates ranged from 9.4% to 100%, and specificity estimates ranged from 5.6% to 100%. Both measures were highly heterogeneous (P < 0.001 for test of heterogeneity). Figure 1 shows the overall accuracy of PCR in a summary receiver operating characteristic (SROC) plot. The symmetric curve shows a trade off between sensitivity and specificity. The area under the SROC curve was 97%, and summary DOR was 159.4, indicating high accuracy. However, the significant heterogeneity in sensitivity and specificity estimates precluded the determination of clinically useful summary measures.
Exploration of heterogeneity
In order to identify factors associated with heterogeneity, we performed stratified (subgroup) analyses. Table 2 presents the study quality and assay factors assessed and their effect on the estimated summary Diagnostic Odds Ratio (DOR). As seen in Table 2, studies that did not report the use of blinding produced a DOR nearly 2.5 times higher than studies that reported blinded interpretation of index test and reference standard results. Studies with PCR tests that used a nested protocol had almost 2 times higher DOR than those using a regular PCR protocol. The use of IS6110 target in comparison with those using any other target for amplification showed a DOR 1.7 times higher than studies that used other targets. A similar result was obtained when studies that used UV transillumination of a gel were compared to those that used a probe for detection. Studies that used chemical reagents for DNA extraction produced DOR estimates that were about 1.12 times greater than studies that used physical methods, indicating that the use of chemicals (including phenol chloroform) does not significantly improve the test accuracy. Only five studies reported the analysis on smear negative samples. When stratified by smear status, no major difference was seen in the DOR. But this result may be due to the small number of studies reporting only smear negative samples for diagnosis.
As in our previous meta-analyses [2,3], none of the stratified analyses for DOR results fully explained the significant heterogeneity across studies in this review; the statistical tests for heterogeneity were significant even within the various strata (data not shown). Therefore, a meta-regression analysis was performed (Table 3) to simultaneously evaluate multiple covariates in the same analysis. The outcome of the regression analysis is reported as the Relative Diagnostic Odds Ratios (RDOR).
As shown in Table 3, studies that used IS6110 as target of amplification, and studies that used nested PCR methods produced RDOR that were significantly higher than those that used other amplification targets or PCR methods. We present the SROC curves for these subgroups in figures 3A and 3B for target and amplification technique, respectively, to show the trade off between sensitivity and specificity. Blinding, detection technique and smear status showed a slightly higher RDOR but they were not statistically significant in the final regression model. Chemical-based DNA extraction did not produce a significant RDOR, indicating that the use of any chemical reagent for DNA extraction did not substantially affect diagnostic accuracy. No difference was seen in DOR in those studies that used phenol-chloroform versus any other DNA extraction method (data not shown).
Discussion
Principal findings
Diagnostic methods and, therefore, control of tuberculosis would be greatly improved by the standardization and application of nucleic acid amplification tests. Our meta-analytical review of 84 in-house PCR-based studies to detect M. tuberculosis in sputum samples showed a summary receiver operator characteristic (SROC) of 97%, indicating an overall high accuracy of these tests. However, because of significant heterogeneity in sensitivity and specificity estimates, clinically meaningful estimates of accuracy could not be derived. Our analysis showed substantial variability in specificity and sensitivity estimates, and it is clear that in-house PCR tests produce highly inconsistent estimates of diagnostic accuracy. Heterogeneity was clearly evident in the results and could not been explained fully even after stratified analyses. Variability in study design, study quality, and differences in thresholds (cut-points) across studies might account for some of the observed heterogeneity. Nevertheless, the meta-regression analysis highlighted some variables that do appear to yield higher accuracy estimates. The use of IS6110 as the amplification target, and the use of a nested PCR protocol appear to enhance accuracy. It is therefore worth considering the inclusion of these elements in in-house PCR protocols. Our analyses also suggest that the methods used for DNA extraction and signal detection were not critical.
Clinical implications
Because of the observed heterogeneity in sensitivity and specificity, it is difficult to determine clinically useful estimates of accuracy. On the other hand, our findings have some relevance for the clinical microbiology laboratory. Our results suggest that the use of IS6110 target sequence in the protocol, and the use of nested PCR methods appear to significantly increase the diagnostic accuracy of PCR. In our previous meta-analysis on NAA test for tuberculous pleuritis, we found that NAA tests that used IS6110 targets produced DOR estimates 2.5 times higher than tests that used other target sequences [48]. Lack of blinding has been found to be associated with higher accuracy in previous meta-analyses [49,53]. Nevertheless we did not find a significant effect in our meta-regression model. Our stratified analyses, however, did show that unblinded studies were associated with higher summary DOR than blinded studies. Previous empiric research [37] and our earlier meta-analyses [48,49] suggest that studies that use a case-control design tend to overestimate diagnostic accuracy. Surprisingly, study design had little impact on diagnostic accuracy in our current analyses. It is possible that laboratory factors (such as target sequence and amplification technique) had a much stronger impact on accuracy than study design features in our analyses.
Limitations of the review
In our review, we found only five studies reporting the analysis of smear negative specimens. Therefore, we could not determine the effect of smear status on accuracy of PCR. Since clinical sputum specimens frequently include smear-negative samples, the conclusions made in this meta-analysis may not apply to studies that included a large number of smear-negative samples. The accuracy estimates for smear-negative specimens have mostly been derived from studies on commercial kits, which have shown high specificity but lower and variable sensitivity [53]. The US Food and Drug Administration (FDA) approved the use of specific commercial kits initially only for smear positive samples, and recently for smear negative specimens [75]. Our review also excluded more recent studies that used other protocols for the detection of amplified DNA, such as real time PCR. We found only three such studies, and hence they could not be subject to meta-analyses. In the future, such methods may prove to enhance NAA test accuracy.
Implications for research
One test characteristic significantly associated with increased accuracy was the use of IS6110 as a target of amplification. IS6110 is present in the M. tuberculosis genome, usually as multiple copies, which helps to increase the sensitivity of a PCR test. Potential problem with using this target is that some strains from certain parts of the world lack this insertion sequence [76]. A possible solution may be to use more than one target. However, we found that multiplex PCR did not contribute to increase the diagnostic accuracy.
Conclusion
In summary, this meta-analytical review of various protocols for PCR-based diagnosis of pulmonary TB identified a few factors associated with improved diagnostic accuracy, and others that did not make a substantial difference. Future development of NAA-based tests to detect M. tuberculosis from sputum specimens should take into consideration these test characteristics as a way to improve accuracy of in-house NAA tests to diagnose pulmonary TB.
Methods
Identification of studies
We searched the following databases: PUBMED (1985–2002), EMBASE (1988–2002), Web of Science (1990–2002), BIOSIS (1993–2002), Cochrane Library (2002; Issue 2), and LILACS (1990–2002). All searches were up to date as of August 2002. The PubMed search was repeated in March 2004, to cover recent literature. The Journal of Clinical Microbiology, a high-yield journal with respect to diagnostic studies was separately searched (1992–2003). The search terms used included "tuberculosis", "mycobacterium tuberculosis", "nucleic acid amplification techniques", "direct amplification test", "polymerase chain reaction", "ligase chain reaction", "molecular diagnostic techniques", "sensitivity and specificity", "accuracy", or "predictive value". Citations were searched from multiple databases as well as obtained from experts in the field and from manufacturers of commercial tests. Reference lists from primary and review articles were searched. English and Spanish articles were selected for final full-text review. Conference abstracts were excluded because they universally contained inadequate data to permit evaluation. This criterion had been used and reported in previous papers [3].
Study eligibility
Our search strategy aimed to include all available studies on in-house NAA tests for direct detection of M. tuberculosis in sputum specimens. To be included in the meta-analysis, a study should have: 1) included at least one comparison of an in-house PCR with an appropriate reference standard (i.e. culture), for detection of M. tuberculosis complex; 2) provided sufficient information on sensitivity and specificity; 3) provided enough information to judge methodological quality of the study.
The following studies were excluded from the review: 1) case reports; 2) evaluation of NAA tests on animal specimens; 3) studies on use of PCR assays for typing of strains; 4) studies on use of PCR assays for determining drug resistance; 5) studies on use of PCR for detection of only non-tuberculosis mycobacteria and 6) studies using only commercial NAA kits.
Data extraction
The final analyses included all available studies on in-house PCR tests for direct detection of M. tuberculosis in sputum specimens. Two reviewers (LLF and MP) determined study eligibility independently. After study selection, data were extracted from each included study using a standardized data extraction form.
The final set of English and Spanish articles was assessed by one reviewer (LLF) and a sample of these was assessed by a second reviewer (MP) to check accuracy in data extraction. For each study, the following quality criteria were scored as met or not: 1) independent comparison of NAA test against reference standard; 2) cross-sectional versus case-control study design; 3) blinded (single or double) interpretation of test and reference standard results. The test methodology criteria included: 1) species identification, 2) methodology used for DNA extraction 3) type of PCR performed (nested-seminested vs. regular, including multiplex), 4) amplification target (IS6110 vs. any other target) 5) method of detection of the final product (ultra-violet (UV) transillumination of an electrophoretic gel vs. use of labeled probes for DNA hybridization), 6) measures taken to avoid contamination and 7) inclusion of positive and negative controls in the assay. If no data on the above criteria were reported in the primary studies, we contacted the authors of the studies for such information. For the purposes of analyses, responses coded as "not reported" were grouped together with "not met".
Since discrepant analysis (where discordant results between test and culture results are resolved, post-hoc, using clinical data) may be a potential source of bias in NAA test assessments, we preferentially included unresolved data where available.
Meta-analysis methods
We used standard methods recommended for meta-analyses of diagnostic test evaluations [6]. Data were analyzed using Stata (version 8) and Meta-Disc (version 1.1) software. Our analyses focused on the following measures of diagnostic accuracy: sensitivity (true positive rate [TPR]), specificity (1-false positive rate [FPR]), and diagnostic odds ratio (DOR). Sensitivity is the proportion positive test results among those with the target disease. Specificity is the proportion negative test results among those without the disease. The DOR is a single indicator of test accuracy [7] that combines the data from sensitivity and specificity into a single number. The DOR of a test is the ratio of the odds of positive test results in the diseased relative to the odds of positive test results in the non-diseased. The value of a DOR ranges from 0 to infinity, with higher values indicating better discriminatory test performance (higher accuracy). A DOR of 1.0 indicates that a test does not discriminate between patients with the disorder and those without it. DOR values lower than 1.0 suggest improper test interpretation (a greater proportion of negative test results in the group with disease). Mathematically, the DOR can be computed using any of the following equations [7]:
DOR = (TP/FN) / (FP/TN)
DOR = [sensitivity/(1 - specificity)] / [(1 - specificity)/specificity]
DOR = Positive likelihood ratio / Negative likelihood ratio
Each study in the meta-analysis contributed a pair of numbers: TPR and FPR. Since these measures are correlated and vary with the thresholds (cut points for determining test positives) employed in the individual studies, it is customary to analyze TPR and FPR proportions as pairs, and to also explore the effect of threshold on study results [6]. We summarized the joint distribution of sensitivity and specificity using the Summary Receiver Operating Characteristic (SROC) curve [8]. Unlike a traditional ROC plot that explores the effect of varying thresholds on sensitivity and specificity in a single study, each data point in the SROC space represents a separate study. The SROC curve is obtained by fitting a symmetric regression curve to pairs of TPR and FPR. The SROC curve and the area under it present an overall summary of test performance, and display the trade off between sensitivity and specificity. A shoulder-like ROC curve suggests that variability in thresholds employed could, in part, explain variability in study results. It also suggests a common, homogeneous underlying DOR that does not change with the diagnostic threshold. The area under the SROC curve is a global measure of overall test accuracy. An area under the curve of 100% indicates perfect discriminatory ability. Heterogeneity in meta-analysis refers to a high degree of variability in study results [9]. Such heterogeneity could be due to variability in thresholds (cut-points), disease spectrum, test methods, and study quality across studies [9]. In the presence of significant heterogeneity, pooled, summary estimates from meta-analyses are not meaningful. We investigated heterogeneity using a meta-regression analysis. The meta-regression analysis was an extension of the SROC model [8,9]. In this unweighted linear regression model, studies (and not patients or specimens) were the units of analysis. The DOR was the outcome (dependent) variable. The independent variables were the covariates that might be associated with the variability in the DOR. Based on our previous meta-analyses [1-3], the following covariates were specified a priori as potential sources of variability: study design, blinded interpretation of NAA test and reference standard, type of PCR test, target sequence amplified, use of probes to detect amplification products, and use of phenol-chloroform for DNA extraction. The results of the meta-regression model are expressed as relative diagnostic odds ratios (RDOR) [7,9]. An RDOR of 1.0 indicates that a particular covariate (e.g. blinded study design) does not affect the overall DOR. An RDOR >1.0 indicates that studies with a particular characteristic (e.g. those that employed a specific target sequence in the PCR) have a higher DOR than studies without this characteristic. For a RDOR <1.0, the reverse holds.
Authors' contributions
LLF designed the study, carried out the study selection, data abstraction, analysis and drafted the manuscript. MP participated in study selection, checked data abstraction and analysis. JC participated in the design of the study and critically reviewed the manuscript. LWR provided input in study design and critically reviewed the manuscript.
Acknowledgements
LLF is supported by the Fogarty Tuberculosis supplementary grant (TW00905-S1) and the Consejo Nacional de Ciencia y Tecnologia (CONACyT), National Council of Science and Technology, Mexico (scholarship number 129617). MP acknowledges the support of the National Institutes of Health (NIH), Fogarty AIDS International Training Program (1-D43-TW00003-16). LLF and MP were also partially supported by NIH/NIAID (RO1 AI 34238).
Figures and Tables
Figure 1 Summary Receiver Operative Curve (SROC) for all studies. Each solid circle represents each study in the meta-analysis. The regression line summarizes the overall diagnostic accuracy. Area under the curve (AUC) = 0.97.
Figure 2 Effect of Significant Test Characteristics on Summary ROC curves: ROC curves for comparison among targets used in the in-house PCR assay (A) and different amplification techniques employed in the assays (B).
Table 1 Study characteristics and methodological quality of included studies
Study characteristic Frequency [N = 84 studies]
Study design
Cross sectional 60 (71%)
Case-control 24 (29%)
Blinded assessment of NAA test and reference standard results
Double or single blinded 29 (34%)
Unblinded or Unknown 55 (66%)
Target sequence for amplification
IS6110 54 (64%)
Other target sequences 30 (36%)
Use of chemical-based DNA extraction method
Any chemical method 55 (66%)
Physical methods 29 (34%)
Amplification technique
Nested or semi-nested PCR 16 (19%)
Regular or multi-plex PCR 68 (81%)
Detection of amplified products
Probe-based method 49 (58%)
Gel-UV method 35 (42%)
Table 2 Stratified analysis: effect of study and test characteristics on summary diagnostic odds ratios
Subgroup (Number of studies) Summary DOR (95% CI)
Study design Case-control (24) 134.4 (65.2 – 213.1)
Cross-sectional (60) 171.1 (106.7 – 274.2)
Blinding Single or double blinded (29) 90.2 (46.2 – 176.2)
Unblinded or NR (55) 221.2 (137.7 – 355.2)
DNA extraction method Chemical (55) 153.7 (90.1 – 262.4)
Physical (29) 171.7 (96.2 – 306.6)
Amplification method PCR/multiplex (68) 139.3 (89.2 – 217.7)
Nested PCR (16) 266.6 (140.8 – 504.6)
Target sequence IS6110 (54) 236.1 (152.9 – 364.6)
Other target (30) 77.4 (38.2 – 156.9)
Detection method Probe (49) 130.6 (78.4 – 217.6)
Gel-UV (35) 215.4 (115.1 – 403.2)
Sputum smear status Pos/Both/NR (78) 161.5 (107.5 – 242.6)
Negative (6) 128.6 (33.4 – 495.1)
Table 3 Meta-regression analysis to determine sources of heterogeneity
Covariate Coefficient P-value Relative diagnostic odds ratio (RDOR) 95% confidence interval
Intercept 3.248 0.0052 ---- ----
Threshold (S) - 0.072 0.4522 ---- ----
Case control vs. cross-sectional design - 0.292 0.5219 0.75 (0.30 – 1.85)
Blinded vs. unblinded studies 0.441 0.2795 1.55 (0.69 – 3.48)
Chemical vs. other DNA extraction methods - 0.209 0.5877 0.81 (0.38 – 1.74)
IS6110 vs. other target sequences 1.055 0.0074 2.87 (1.34 – 6.16)
Nested vs. regular amplification protocols 1.196 0.0135 3.31 (1.29 – 8.49)
Probe vs. gel-UV detection methods 0.157 0.7222 1.17 (0.49 – 2.81)
Sputum smear positive/both vs negative 0.242 0.8146 1.27 (0.16 – 9.88)
Intercept: constant term in the model
S: indicator of threshold (logit TPR+logit FPR); TPR: true positive rate; FPR: false positive rate RDOR: relative diagnostic odds ratio (obtained by exponentiating the model coefficients)
==== Refs
Flores LL M. tuberculosis in clinical samples from patients with pulmonary tuberculosis: a systematic review and meta-analysis Master's in Public Health Thesis 2004 University of California, Berkeley, Department of Infectious Diseases. Spring
Pai M Flores LL Hubbard A Riley LW Colford JM Nucleic acid amplification in the diagnosis of tuberculous pleuritis: a systematic review and meta-analysis BMC Infect Dis 2004 4 6 15102325 10.1186/1471-2334-4-6
Pai M Flores LL Pai N Hubbard A Riley LW Colford JM Diagnostic accuracy of nucleic acid amplification tests for tuberculous meningitis: a systematic review and meta-analysis Lancet Infect Dis 2003 3 633 643 14522262 10.1016/S1473-3099(03)00772-2
Piersimoni C Scarparo C Relevance of commercial amplification methods for direct detection of Mycobacterium tuberculosis complex in clinical samples J Clin Microbiol 2003 41 5355 5365 14662911 10.1128/JCM.41.12.5355-5365.2003
Sarmiento OL Weigle K Alexander J Weber DJ Miller WC Assessment by meta-analysis of PCR for diagnosis of smear-negative pulmonary tuberculosis J Clin Microbiol 2003 41 3233 3240 12843069 10.1128/JCM.41.7.3233-3240.2003
Pai M McCulloch M Enanoria W Colford JM Systematic reviews of diagnostic test evaluations: What's behind the scenes? Evid Based Med 2004 9 101 103 10.1136/ebm.9.4.101
Glas AS Lijmer JG Prins MH Bonsel GJ Bossuyt PM The diagnostic odds ratio: a single indicator of test performance J Clin Epidemiol 2003 56 1129 1135 14615004 10.1016/S0895-4356(03)00177-X
Littenberg B Moses L Estimating diagnostic accuracy from multiple conflicting reports: a new meta-analytic method Med Decis Making 1993 13 313 321 8246704
Lijmer JG Bossuyt PM Heisterkamp SH Exploring sources of heterogeneity in systematic reviews of diagnostic tests Stat Med 2002 21 1525 537 12111918 10.1002/sim.1185
Altamirano M Kelly MT Wong A Bessuille ET Black WA Smith JA Characterization of a DNA probe for detection of Mycobacterium tuberculosis complex in clinical samples by polymerase chain reaction J Clin Microbiol 1992 30 2173 2176 1500529
Amicosante M Richeldi L Trenti G Paone G Campa M Bisetti A Saltini C Inactivation of polymerase inhibitors for Mycobacterium tuberculosis DNA amplification in sputum by using capture resin J Clin Microbiol 1995 33 629 630 7751368
Andersen AB Thybo S Godfrey-Faussett P Stoker NG Polymerase chain reaction for detection of Mycobacterium tuberculosis in sputum Eur J Clin Microbiol Infect Dis 1993 12 922 927 8187787 10.1007/BF01992166
Aslanzadeh J de la Viuda M Fille M Smith WB Namdari H Comparison of culture and acid-fast bacilli stain to PCR for detection of Mycobacterium tuberculosis in clinical samples Mol Cell Probes 1998 12 207 211 9727196 10.1006/mcpr.1998.0174
Beige J Lokies J Schaberg T Finckh U Fischer M Mauch H Lode H Kohler B Rolfs A Clinical evaluation of a Mycobacterium tuberculosis PCR assay J Clin Microbiol 1995 33 90 95 7535316
Brisson-Noel A Gicquel B Lecossier D Levy-Frebault V Nassif X Hance AJ Rapid diagnosis of tuberculosis by amplification of mycobacterial DNA in clinical samples Lancet 1989 2 1069 1071 2572798 10.1016/S0140-6736(89)91082-9
Chaiprasert A Prammananan T Samerpitak K Pattanakitsakul S Yenjitsomanus P Jearanaisilawong J Charoenratanakul S Direct detection of M. tuberculosis in sputum by reamplification with 16S rRNA-based primer Asia-Pacific Journal of Molecular Biology and Biotechnology 1996 4 250 259
Choi YJ Hu Y Mahmood A Clinical significance of a polymerase chain reaction assay for the detection of Mycobacterium tuberculosis Am J Clin Pathol 1996 105 200 204 8607445
Clarridge JE 3rdShawar RM Shinnick TM Plikaytis BB Large-scale use of polymerase chain reaction for detection of Mycobacterium tuberculosis in a routine mycobacteriology laboratory J Clin Microbiol 1993 31 2049 2056 8370729
Codina G Gonzalez-Fuente T Martin N DNA amplification with PCR for tuberculosis diagnosis. [Spanish] Enfermedades Infecciosas y Microbiologia Clinica 1992 10 277 280 1390997
Cohen RA Muzaffar S Schwartz D Bashir S Luke S McGartland LP Kaul K Diagnosis of pulmonary tuberculosis using PCR assays on sputum collected within 24 hours of hospital admission Am J Respir Crit Care Med 1998 157 156 161 9445294
Dar L Sharma SK Bhanu NV Broor S Chakraborty M Pande JN Seth P Diagnosis of pulmonary tuberculosis by polymerase chain reaction for MPB64 gene: an evaluation in a blind study Indian J Chest Dis Allied Sci 1998 40 5 16 9727278
Del Portillo P Murillo LA Patarroyo ME Amplification of a species-specific DNA fragment of Mycobacterium tuberculosis and its possible use in diagnosis J Clin Microbiol 1991 29 2163 2168 1939567
Del Portillo P Thomas MC Martinez E Maranon C Valladares B Patarroyo ME Lopez MC Multiprimer PCR system for differential identification of mycobacteria in clinical samples J Clin Microbiol 1996 34 324 328 8789008
Down JA O'Connell MA Dey MS Walters AH Howard DR Little MC Keating WE Zwadyk P Haaland PD McLaurin DA Cole G Detection of Mycobacterium tuberculosis in respiratory specimens by strand displacement amplification of DNA J ClinMicrobiol 1996 34 860 865
Eisenach KD Sifford MD Cave MD Bates JH Crawford JT Detection of Mycobacterium tuberculosis in sputum samples using a polymerase chain reaction Am Rev Respir Dis 1991 144 1160 1163 1952448
Folgueira L Delgado R Palenque E Noriega AR Detection of Mycobacterium tuberculosis DNA in clinical samples by using a simple lysis method and polymerase chain reaction J Clin Microbiol 1993 31 1019 1021 8463383
Forbes BA Hicks KE Direct detection of Mycobacterium tuberculosis in respiratory specimens in a clinical laboratory by polymerase chain reaction J Clin Microbiol 1993 31 1688 1694 8349744
Garcia-Quintanilla A Garcia L Tudo G Navarro M Gonzalez J Jimenez de Anta MT Single-tube balanced heminested PCR for detecting Mycobacterium tuberculosis in smear-negative samples J Clin Microbiol 2000 38 1166 1169 10699014
Gengvinij N Pattanakitsakul SN Chierakul N Chaiprasert A Detection of Mycobacterium tuberculosis from sputum specimens using one-tube nested PCR Southeast Asian J Trop Med Public Health 2001 32 114 125 11485071
Ginesu F Pirina P Sechi LA Molicotti P Santoru L Porcu L Fois A Arghittu P Zanetti S Fadda G Microbiological diagnosis of tuberculosis: a comparison of old and new methods J Chemother 1998 10 295 300 9720468
Gunisha P Madhavan HN Jayanthi U Therese KL Polymerase chain reaction using IS6110 primer to detect Mycobacterium tuberculosis in clinical samples Indian J Pathol Microbiol 2001 44 97 102 11883143
Hashimoto T Suzuki K Amitani R Kuze F Rapid Detection of Mycobacterium tuberculosis in Sputa by the Amplification of IS6110 Internal Medicine 1995 34 605 610 7496067
Ikonomopoulos JA Gorgoulis VG Zacharatos PV Kotsinas A Tsoli E Karameris A Panagou P Kittas C Multiplex PCR assay for the detection of mycobacterial DNA sequences directly from sputum In Vivo 1998 12 547 552 9827366
Ikonomopoulos JA Gorgoulis VG Zacharatos PV Manolis EN Kanavaros P Rassidakis A Kittas C Multiplex polymerase chain reaction for the detection of mycobacterial DNA in cases of tuberculosis and sarcoidosis Modern Pathology 1999 12 854 862 10496593
Kambashi B Mbulo G McNerney R Tembwe R Kambashi A Tihon V Godfrey-Faussett P Utility of nucleic acid amplification techniques for the diagnosis of pulmonary tuberculosis in sub-Saharan Africa Int J Tuberc Lung Dis 2001 5 364 369 11334256
Kearns AM Freeman R Steward M Magee JG A rapid polymerase chain reaction technique for detecting M. tuberculosis in a variety of clinical specimens J Clin Pathol 1998 51 922 924 10070334
Kocagoz T Yilmaz E Ozkara S Kocagoz S Hayran M Sachedeva M Chambers HF Detection of Mycobacterium tuberculosis in sputum samples by polymerase chain reaction using a simplified procedure J Clin Microbiol 1993 31 1435 1438 8314982
Kolk AH Schuitema AR Kuijper S van Leeuwen J Hermans PW van Embden JD Hartskeerl RA Detection of Mycobacterium tuberculosis in clinical samples by using polymerase chain reaction and a nonradioactive detection system J Clin Microbiol 1992 30 2567 2575 1400955
Kox LF Jansen HM Kuijper S Kolk AH Multiplex PCR assay for immediate identification of the infecting species in patients with mycobacterial disease J Clin Microbiol 1997 35 1492 1498 9163468
Kox LF Rhienthong D Miranda AM Udomsantisuk N Ellis K van Leeuwen J van Heusden S Kuijper S Kolk AH A more reliable PCR for detection of Mycobacterium tuberculosis in clinical samples J Clin Microbiol 1994 32 672 678 8195377
Laniado-Laborin R Enriquez-Rosales ML Licea Navarroll AF Diagnosis of tuberculosis by detection of Mycobacterium tuberculosis with a noncommercial polymerase chain reaction. [Spanish] Revista del Instituto Nacional de Enfermedades Respiratorias 2001 14 22 26
Maher M Glennon M Martinazzo G Turchetti E Marcolini S Smith T Dawson MT Evaluation of a novel PCR-based diagnostic assay for detection of Mycobacterium tuberculosis in sputum samples J Clin Microbiol 1996 34 2307 2308 8862607
Manjunath N Shankar P Rajan L Bhargava A Saluja S Shriniwas Evaluation of a polymerase chain reaction for the diagnosis of tuberculosis Tubercle 1991 72 21 27 1882442 10.1016/0041-3879(91)90020-S
Mitarai S Oishi K Fukasawa M Yamashita H Nagatake T Matsumoto K Clinical-Evaluation of Polymerase Chain-Reaction DNA Amplification Method for the Diagnosis of Pulmonary Tuberculosis in Patients with Negative Acid-Fast Bacilli Smear Tohoku Journal of Experimental Medicine 1995 177 13 23 8693483
Miyazaki Y Koga H Kohno S Kaku M Nested polymerase chain reaction for detection of Mycobacterium tuberculosis in clinical samples J Clin Microbiol 1993 31 2228 2232 8370757
Moran Moguel MC Hernandez DA Pena-Montes de Oca PM Gallegos Arreola MP Flores-Martinez SE Montoya-Fuentes H Figuera LE Villa-Manzanares L Sanchez-Corona J [Detection of Mycobacterium tuberculosis by polymerase chain reaction in a selected population in northwestern Mexico] Rev Panam Salud Publica 2000 7 389 394 10949900
Mustafa AS Chugh TD Abal AT Polymerase chain reaction targeting of single- and multiple-copy genes of mycobacteria in the diagnosis of tuberculosis Nutrition 1995 11 665 669 8748247
Nolte FS Metchock B McGowan JE JrEdwards A Okwumabua O Thurmond C Mitchell PS Plikaytis B Shinnick T Direct detection of Mycobacterium tuberculosis in sputum by polymerase chain reaction and DNA hybridization J Clin Microbiol 1993 31 1777 1782 8349753
Noordhoek GT Kaan JA Mulder S Wilke H Kolk AH Routine application of the polymerase chain reaction for detection of Mycobacterium tuberculosis in clinical samples J Clin Pathol 1995 48 810 814 7490312
Pao CC Yen TS You JB Maa JS Fiss EH Chang CH Detection and identification of Mycobacterium tuberculosis by DNA amplification J Clin Microbiol 1990 28 1877 1880 2121786
Patnaik M Liegmann K Peter JB Rapid detection of smear-negative Mycobacterium tuberculosis by PCR and sequencing for rifampin resistance with DNA extracted directly from slides J Clin Microbiol 2001 39 51 52 11136747 10.1128/JCM.39.1.51-52.2001
Rodriguez JC Fuentes E Royo G Comparison of two different PCR detection methods: Application to the diagnosis of pulmonary tuberculosis Apmis 1997 105 612 616 9298099
Rodriguez JC Fuentes E Royo G Detection of Mycobacterium tuberculosis in sputum by PCR and revealed chemoluminescence. [Spanish] Enfermedades Infecciosas y Microbiologia Clinica 1997 15 186 189 9312276
Rossetti ML Jardim SB Moura AR Oliveira H Zaha A Improvement of Mycobacterium tuberculosis detection in clinical samples using DNA purified by glass matrix J Microbiol Methods 1997 28 139 146 10.1016/S0167-7012(97)00978-0
Savic B Sjobring U Alugupalli S Larsson L Miorner H Evaluation of polymerase chain reaction, tuberculostearic acid analysis, and direct microscopy for the detection of Mycobacterium tuberculosis in sputum J Infect Dis 1992 166 1177 1180 1402031
Seethalakshmi S Korath MP Kareem F Jagadeesan K Detection of Mycobacterium tuberculosis by polymerase chain reaction in 301 biological samples – a comparative study J Assoc Physicians India 1998 46 763 766 11229242
Soini H Agha SA El-Fiky A Viljanen MK Comparison of amplicor and 32-kilodalton PCR for detection of Mycobacterium tuberculosis from sputum specimens J Clin Microbiol 1996 34 1829 1830 8784603
Soini H Skurnik M Liippo K Tala E Viljanen MK Detection and identification of mycobacteria by amplification of a segment of the gene coding for the 32-kilodalton protein J Clin Microbiol 1992 30 2025 2028 1500509
Sritharan V Barker RH A simple method for diagnosing M. tuberculosis infection in clinical samples using PCR Mol Cell Probes 1991 5 385 395 1791860
Su WJ Tsou AP Yang MH Huang CY Perng RP Clinical experience in using polymerase chain reaction for rapid diagnosis of pulmonary tuberculosis Chin Med J (Taipei) 2000 63 521 526
Tan MF Ng WC Chan SH Tan WC Comparative usefulness of PCR in the detection of Mycobacterium tuberculosis in different clinical specimens J Med Microbiol 1997 46 164 169 9060877
Tansuphasiri U Chinrat B Rienthong S Evaluation of culture and PCR-based assay for detection of Mycobacterium tuberculosis from sputum collected and stored on filter paper Southeast Asian J Trop Med Public Health 2001 32 844 855 12041562
Thierry D Brisson-Noel A Vincent-Levy-Frebault V Nguyen S Guesdon JL Gicquel B Characterization of a Mycobacterium tuberculosis insertion sequence, IS6110, and its application in diagnosis J Clin Microbiol 1990 28 2668 2673 2177747
Thierry D Chureau C Aznar C Guesdon JL The detection of Mycobacterium tuberculosis in uncultured clinical specimens using the polymerase chain reaction and a non-radioactive DNA probe Mol Cell Probes 1992 6 181 191 1383698 10.1016/0890-8508(92)90015-P
Totsch M Schmid KW Brommelkamp E Stucker A Puelacher C Sidoroff G Mikuz G Bocker W Dockhorn-Dworniczak B Rapid detection of mycobacterial DNA in clinical samples by multiplex PCR Diagn Mol Pathol 1994 3 260 264 7866636
van Helden PD Toit R Jordaan A Taljaard B Pitout J Victor T The use of the polymerase chain reaction test in the diagnosis of tuberculosis S Afr Med J 1991 80 515 516 1948472
Verma A Rattan A Tyagi JS Development of a 23S rRNA-based PCR assay for the detection of mycobacteria Indian J Biochem Biophys 1994 31 288 294 8002010
Victor T du Toit R van Helden PD Purification of sputum samples through sucrose improves detection of Mycobacterium tuberculosis by polymerase chain reaction J Clin Microbiol 1992 30 1514 1517 1624570
Walker DA Taylor IK Mitchell DM Shaw RJ Comparison of polymerase chain reaction amplification of two mycobacterial DNA sequences, IS6110 and the 65kDa antigen gene, in the diagnosis of tuberculosis Thorax 1992 47 690 694 1440462
Weekes KM Pearse MJ Sievers A Ross BC d'Apice AJ The diagnostic use of the polymerase chain reaction for the detection of Mycobacterium tuberculosis Pathology 1994 26 482 486 7892054 10.1080/00313029400169232
Williams DL Spring L Gillis TP Salfinger M Persing DH Evaluation of a polymerase chain reaction-based universal heteroduplex generator assay for direct detection of rifampin susceptibility of Mycobacterium tuberculosis from sputum specimens Clin Infect Dis 1998 26 446 450 9502469
Wilson SM McNerney R Nye PM Godfrey-Faussett PD Stoker NG Voller A Progress toward a simplified polymerase chain reaction and its application to diagnosis of tuberculosis J Clin Microbiol 1993 31 776 782 8463386
Yoon KH Cho SN Lee TY Cheon SH Chang J Kim SK Chong Y Chung DH Lee WY Kim JD Detection of Mycobacterium tuberculosis in clinical samples from patients with tuberculosis or other pulmonary diseases by polymerase chain reaction Yonsei Med J 1992 33 209 216 1292244
Yuen KY Chan KS Chan CM Ho BS Dai LK Chau PY Ng MH Use of PCR in routine diagnosis of treated and untreated pulmonary tuberculosis J Clin Pathol 1993 46 318 322 8496388
CDC Nucleic Acid Amplification Tests for Tuberculosis MMWR 2000 49 593 594
Das S Paramasivan CN Lowrie DB Prabhakar R Narayanan PR IS6110 Restriction Fragment Length Polymorphism Typing of Clinical Isolates of Mycobacterium tuberculosis from Patients with Pulmonary Tuberculosis in Madras, South India Tuber Lung Dis 1995 76 550 554 8593378 10.1016/0962-8479(95)90533-2
|
16202138
|
PMC1260021
|
CC BY
|
2021-01-04 16:03:41
|
no
|
BMC Microbiol. 2005 Oct 3; 5:55
|
utf-8
|
BMC Microbiol
| 2,005 |
10.1186/1471-2180-5-55
|
oa_comm
|
==== Front
BMC OphthalmolBMC Ophthalmology1471-2415BioMed Central London 1471-2415-5-241621266710.1186/1471-2415-5-24Research ArticlePostoperative IOP prophylaxis practice following uncomplicated cataract surgery: a UK-wide consultant survey Zamvar Usha [email protected] Baljean [email protected] Department of Ophthalmology, Stirling Royal Infirmary, Livilands, Stirling, UK2 Princess Alexandra Eye Pavilion, Chalmers Street, Edinburgh, UK2005 7 10 2005 5 24 24 25 11 2004 7 10 2005 Copyright © 2005 Zamvar and Dhillon; 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 order to minimise postoperative intraocular pressure (IOP) rise, after routine uncomplicated cataract surgery, prophylaxis may be adopted. Currently, there are no specific guidelines in this regard resulting in wide variation in practice across the UK. We sought to document these variations through a questionnaire survey.
Methods
A questionnaire was sent to all consultant ophthalmic surgeons in the UK.
Results
62.6% of surgeons did not use any IOP lowering agents. 37.4% surgeons routinely prescribed some form of medication. The majority (86.8%) used oral diamox. 20.6% of surgeons said they based their practice on evidence, 43.3% on personal experience, and 17.6% on unit policy. Surprisingly, among the two groups of surgeons (those who gave routine prophylaxis, and those who did not) the percentages of surgeons quoting personal experience, unit policy, or presence of evidence was strikingly similar. The timing of the first postoperative IOP check varied from the same day to beyond 2 weeks. Only 20.2% of surgeons had ever seen an adverse event related to IOP rise; this complication is thus very rare.
Conclusion
This survey highlights a wide variation in the practice and postoperative management of phacoemulsification cataract surgery. What is very striking is that there is a similar proportion of surgeons in the diametrically opposite groups (those who give or do not give routine IOP lowering prophylaxis) who believe that there practice is evidence based. The merits of this study suggests that consideration must be given to drafting a uniform guideline in this area of practice.
==== Body
Background
Phacoemulsification and intra-ocular lens implantation (PhIOL) is one of the most cost-effective, elective surgical interventions. In order to minimise postoperative intraocular pressure (IOP) rise, prophylaxis may be adopted. Currently, there are no specific guidelines for prophylaxis in uncomplicated cataract surgery. We studied current prophylaxis practice in a UK-wide survey which showed wide variation in prophylaxis practice.
Methods
We conducted a pilot, self-administered postal-based survey of the Scottish ophthalmic consultants. The results of this survey suggested variation in prophylaxis practice for IOP rise, duration until 1st IOP monitoring, and management of elevated IOP following uncomplicated phacoemulsification with intraocular lens implantation with no coexisting comorbidity. This prompted us to extend our enquiry across the U.K. We obtained the names and addresses of ophthalmic consultants working across the United Kingdom, from the Royal College of Ophthalmologists, London. A questionnaire (Appendix 1) was sent to all the consultants between April and July 2003. The data were entered on a spreadsheet using Microsoft Excel. Data was analysed using SPSS 11(SPSS Inc, Chicago, IL).
Results
The questionnaire was sent to 834 ophthalmic surgeons, and 515 (61.7%) responded. No reminders were sent. Ten did not perform cataract surgery and their responses were excluded from further analysis.
Routine use of IOP prophylaxis following uncomplicated phacoemulsification and lens implant surgery
Of the 505 surgeons, who performed cataract surgery, 316 (62.6%) did not use any IOP lowering agent following uncomplicated phacoemulsification, and lens implant surgery. The remaining 189 (37.4%) routinely prescribed some form of medication for IOP prophylaxis.
Of the 189 surgeons who used some form of prophylaxis, oral diamox was used by 164 (86.8%), and a topical agent was used by 21 (11.1%). The remaining 4 (2.1%) surgeons used both.
Of the 168 surgeons who used oral diamox, 113 (67.2%), gave only one dose, 41 (24.4%) gave two doses, 9 (5.3%) gave three doses, and 5 (2.9%) gave four or more doses. Of the 25 surgeons who used topical agents, 22 (88%) gave only one dose, and 1 (4%) surgeon each gave 3, 5, and 6 doses.
Basis of IOP prophylaxis practice
We questioned the surgeons about the basis of their practice using a forced choice selection of a) based on evidence, b) personal experience, or c) as a matter of unit policy. Surgeons were allowed to select more than one option. Two hundred and ten (41.5%) surgeons did not reply, 104 (20.6%) said their practice was based on evidence, 219 (43.3%) said their practice was based on personal experience, and unit policy was the basis for 89 (17.6%). We analysed further the basis of practice, according to whether routine prophylaxis was used or not, and which drug was used. A total of 316 surgeons did not prescribe IOP lowering agent routinely. Of these, 209 surgeons (66.1%) chose not to reply when asked about the basis of their practice. Of the 107, who chose to reply, 42 (39.2%) said their practice was based on evidence, 74 (69.1%) said their practice was based on personal experience, and 22 (20.5%) on unit policy. One hundred and eighty nine surgeons routinely used some form of IOP lowering agent. Of these only one surgeon (0.5%) did not reply, to the question regarding the basis of their practice. Of the 188 who chose to reply, 62 (32.9%) said their practice was based on evidence, 124 (65.9%) said their practice was based on personal experience, and 43 (22.8%) on unit policy. The difference in the proportion of surgeons in the two groups choosing not to reply to this question is significant (p < 0.001). Among the respondents, the percentages of surgeons quoting personal experience, unit policy, or presence of evidence is strikingly similar (figure 1).
Time of first postoperative IOP check
The timing of the first postoperative IOP check varied. Fifty-five surgeons (10.9%) reported the first IOP check was carried out on the same day, 150 (29.7%) on the first postoperative day, 105(20.8%) by the first week, 136 (26.9%) at 2 weeks, and 48 (9.5%) beyond 2 weeks. Nine surgeons (1.8%) said they never check IOP routinely. There was no significant difference in the timing of the first postoperative check between the users and non-users of IOP prophylaxis.
To assess whether the timing of postoperative IOP measurement has any impact on the incidence of reported adverse events, we looked at the following:
a) the various time points at which postoperative IOP measurements were made amongst the two groups of surgeons (those who give routine prophylaxis versus those who don't). (Table 1)
b) the relation between the timing of first postoperative IOP check and the reported experience with adverse events. (Table 2)
There was no statistically significant association between the timing of first IOP check, and the practice of giving routine prophylaxis. Similarly, there was no statistically significant association between the timing of first IOP check, and the reported experience with adverse events.
Adverse event related to postoperative intraocular pressure elevation
One hundred and two surgeons (20.2%) had seen an adverse event related to IOP rise, compared to 396 (78.4%) who had not. 7 (1.4%) surgeons did not answer this question.
Of the 189 surgeons who routinely use IOP prophylaxis, 31 (16%) had encountered an adverse event in their practice, and 157 (83%) had not. Of the 316 surgeons who did not routinely use any prophylaxis, 71 (22%) had seen an adverse event, compared to 239 (75.6%) who had not. The difference was not significant (p = 0.10). A variety of adverse events were reported including corneal oedema, central retinal vein occlusion, ocular pain, optic neuropathy.
Discussion
Cataract extraction is one of the most commonly performed and successful surgical procedures. This survey highlights a wide variation in the practice in the postoperative management of phacoemulsification cataract surgery.
Raised intraocular pressure is one of the most common complication following cataract surgery, requiring specific treatment [1-6]. Many treatment strategies of blunting acute post-op IOP spikes have been proposed, including use of prophylactic intracameral cholinergic agents and topical and systemic antiglaucoma medication [7-10]. However, this does not seem to eliminate spikes and many have found their effect to be negligible compared to a placebo [11-14]. The current literature on medical prophylaxis is conflicting [15-18] Most of the antiglaucoma agents used to prevent or lessen IOP increase postoperatively have limitations. 87% of responders who use IOP prophylaxis prefer oral Diamox over the topical agents. Oral Diamox or, Acetazolamide, a systemic sulphonamide inhibitor of carbonic anhydrase enzyme, reduces the flow of aqueous humor, thereby lowering the IOP. Acute urinary retention amongst men with prostatic enlargement and falls amongst the elderly may also occur with oral Diamox. Less serious side effects include thirst, drowsiness, polyuria and paraesthesia. This may result in accidents in elderly patients who have just undergone ocular surgery. More severe adverse reactions include fatal aplastic anaemia, sulfaallergy cross sensitivity and acid base disturbance [21].
Iopidine and Timoptol are the most common topical agents used for post-op IOP prophylaxis as shown in the survey. A number of clinical trials studying the effect of pre and post-operative use of Apraclonidine and Timoptol in reducing post-op rise have shown variable results[8,9,12,14,17,18].
Our survey also demonstrates a wide variation in the timing of the first IOP check. Only 10.9 % of our responders check IOP on the day of surgery. These patients do not visit the hospital on the first postoperative day, which is very convenient for them, and for overall majority of patients, visual outcome is not compromised when routine next day review is omitted after phacoemulsification surgery [19,20].
This reflects the relatively low frequency of severe IOP elevation one day postoperatively, the self limiting nature of IOP spikes and the tolerance of a healthy eye.
This survey also showed that only 20.2% surgeons had ever encountered an adverse event related to IOP rise. The vast majority of surgeons (78.4%) had never encountered one. An adverse event related to IOP rise is rare [20]. Assuming that each surgeon performs 400 cataract operations per year, and that the 20.2% surgeons who had encountered the complications see it twice a year, the incidence of this complication would be 0.1%. When the incidence of any complication is this low, it may be difficult to organise a randomised controlled trial to show if the use of IOP lowering prophylaxis is effective or not, as the sample size would run in tens of thousands. For example, in this case, we would need 700,000 patients to be randomised for a study with a power of 80%, to show a difference of 20% in the incidence of adverse event related to IOP rise. This is clearly an impossible task. Randomised controlled trials have been conducted to assess the role of IOP prophylaxis [7-10,12-18], but they have used the IOP level as a surrogate marker. There may be fluctuations in the IOP levels, and a statistically significant difference between the two groups, but whether this actually results in a change in the incidence of an adverse event related to IOP rise is unproven.
It was a weakness of our questionnaire that we did not ask the surgeons about their annual cataract surgical volumes, and the number of complications they had encountered. Another point of this study that we would like to highlight is that a high proportion of surgeons not prescribing routine IOP prophylaxis chose not to give a reason for the basis of their practice.
Conclusion
In summary, this survey shows a very wide variation in practice regarding postoperative management of patients undergoing phacoemulsification with intraocular lens implant. What is very striking is that there is a similar proportion of surgeons in the diametrically opposite groups (those who give or don't give routine IOP lowering prophylaxis) who believe that their practice is evidence based. Personal experience was cited by a large percentage of surgeons in each group. Practice of medicine is not necessarily evidence based. Reasons include [22] clinical experience, over-reliance on surrogate outcomes, ritual and mystique. Our survey adds another explanation: interpretation of the evidence in different ways, perhaps to fit with one's clinical practice.
Whilst the authors would not wish to be prescriptive in post PhIOL prophylaxis, the merits of this study suggests further consideration might be given to drafting a uniform guideline in this area of practice.
Appendix 1:
1) Do you routinely give any intraocular pressure lowering agent to your patients (without co-morbidities) following uncomplicated, phacoemulsification and lens implant surgery?
a) Yes
b) No
2) If Yes, which one of the following?
a) Topical medication (Please specify the name of the drug)
b) Oral Diamox- 250 mg/500 mg
3) At what stage following the surgery, do the patients receive the above medication?
a) at the end of the surgical procedure
b) in the recovery area
c) on the ward
d) at home (after discharge from hospital)
4) How many doses are given?
a) one
b) two
c) three
5) What is the basis of your practice? (you may tick more then one)
a) Based on evidence
b) Based on personal experience
c) Unit policy
6) Do you give subconjunctival injection at the end of the procedure?
a) yes
b) no
7) If yes, which of the following ?
a) subconj antibiotic
b) subconj steroid
c) both
8) When do the patients have their intraocular pressure checked for the first time after surgery?
a) Few hours after surgery
b) First postoperative day
c) 1 week postop
d) 2 weeks postop
9) Have you come across a sight threatening condition caused by raised postoperative IOP?
a) Yes (please specify)
b) No
10) How would you treat a patient with significantly raised IOP (> 30 mm Hg) within the first 24 hours following surgery?
a) paracentesis through the 2nd port
b) oral diamox
c) I/V diamox
11) What post-op medication do you give to your patients and for how long?
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Details of contributors and guarantor:
BD had the original idea, UZ and BD drafted the questionnaire, UZ tabulated the replies, performed the statistical analysis and wrote the original manuscript, UZ and BD contributed to the final manuscript. BD is the guarantor.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Figures and Tables
Figure 1 Basis of practice. EB: Evidence Based PE: Personal Experience UP: Unit Policy
Table 1 Timing of first IOP check Surgeons give routine prophylaxis (n = 189) Surgeons do not give routine prophylaxis (n = 316)
Same Day 22 (11.6%) 33 (10.4%)
1st postop day 52 (27.5%) 98 (31.0%)
By 1st week 35 (18.5%) 70 (22.1%)
At 2 weeks 59 (31.2%) 77 (24.3%)
Beyond 2 weeks 17 (8.9%) 31 (9.8%)
Never 4 (2.1%) 5 (1.5%)
No reply 0 (0%) 2 (0.6%)
Legend: This table shows the timing of the first postoperative IOP check in the two groups of surgeons (those who routinely give or don't give prophylaxis). There is no statistical significant difference between the two groups.
Table 2 Timing of first postop IOP check No. of surgeons (n = 503) Surgeons who have experienced adverse events
Same day 55 (10.9%) 15 (27.2%)
1st post-op day 150 (29.7%) 25 (16.6%)
By first week 105 (20.8%) 20 (19.0%)
At 2 weeks 136 (26.9%) 36 (26.4%)
Beyond 2 weeks 48 (9.5%) 6 (12.5%)
Never 9 (1.8%) 0 (0%)
Legend: Relation between timing of first IOP check and reported experience with adverse events. There is no significant trend noticed.
==== Refs
Tan JH Newman DK Klunker C Watts SE Burton RL Phacoemulsification cataract surgery: is routine review necessary on the first postoperative day? Eye 2000 14 53 5 10755100
Tufail A Foss AJ Hamilton AM Is the first day postoperative review necessary after cataract extraction? Br J Ophthalmol 1995 79 646 8 7662627
Whitefield L Crowston J Little BC First day follow up for routine phacoemulsification? Br J Ophthalmol 1996 80 148 50 8814746
Cohen VM Demetria H Jordan K Lamb RJ Vivian AJ First day postoperative review following uncomplicated phacoemulsification Eye 1998 12 634 6 9850255
Dinakara S Desai SP Raj PS Is the first postoperative day review necessary following uncomplicated phacoemulsification surgery? Eye 2000 14 364 6 11027001
Desai P Minassian DC Reidy A National cataract surgery survey 1997–98: a report of the results of the clinical outcomes Br J Ophthalmol 1999 83 1336 40 10574810
Zohdy GA Rogers ZA Lukaris A Sells M Roberts-Harry TJ A comparison of the effectiveness of dorzolamide and acetazolamide in preventing postoperative intraocular pressure rise following phacoemulsification surgery J R Coll Surg Edinb 1998 43 344 6 9803110
Araie M Ishi K Effects of apraclonidine on intraocular pressure and blood-aqueous barrier permeability after phacoemulsification and intraocular lens implantation Am J Ophthalmol 1993 116 67 71 8101046
Fry LL Comparison of the postoperative intraocular pressure with Betagan Betoptic, Timoptic, Iopidine, Diamox, Pilopine Gel, and Miostat J Cataract Refract Surg 1992 18 14 9 1346539
Scherer WJ Mielke DL Tidwell PE Hauber FA Effect of Latanoprost on intraocular pressure following cataract extraction J Cataract Refract Surg 1998 24 964 7 9682119
Barak A Desatnik H Ma-Naim T Ashkenazi I Neufeld A Melamed S Early postoperative intraocular pressure pattern in glaucomatous and non glaucomatous patients J Cataract Refract Surg 1996 22 607 11 8784635
Bomer TG Lagreze WD Funk J Intraocular pressure rise after phacoemulsification with posterior chamber lens implantation: effect of prophylactic medication, wound closure, and surgeons' experience Br J Ophthalmol 1995 79 809 13 7488598
Rainer G Menapace R Schmetterer K Findl O Georgopoulos M Vass C Effect of Dorzolamide and latanoprost on intraocular pressure after small incision cataract surgery J Cataract Refract Surg 1999 25 1624 9 10609207 10.1016/S0886-3350(99)00260-6
Sterk CC Renzenbrink-Bubberman AC van Best JA The effect of 1% apraclonidine on intraocular pressure after cataract surgery Ophthalmic Surg Lasers 1998 29 472 5 9640568
Ruiz RS Wilson CA Musgrove KH Prager TC Management of intraocular pressure after cataract extraction Am J Ophthalmol 1987 103 487 91 3551618
Byrd S Singh K Medical control of intraocular pressure after cataract surgery J Cataract Refract Surg 1998 24 1493 7 9818340
Lai JS Chua JK Leung AT Lam DS Latanoprost versus timolol gel to prevent ocular hypertension after phacoemulsification and intraocular lens implantation J Cataract Refract Surg 2000 26 386 91 10713234 10.1016/S0886-3350(99)00364-8
Duperre J Grenier B Lemire J Mihalovits H Sebag M Lambert J Effect of timolol vs acetazolamide on sodium hyaluronate-induced rise in intraocular pressure after cataract surgery Can J Ophthalmol 1994 29 182 6 7994673
Tinley CG Frost A Hakin KN McDermott W Ewings P Is visual outcome compromised when next day review is omitted after phacoemulsification surgery? A randomised controlled trial Br J Ophthalmol 2003 87 1350 5 14609832 10.1136/bjo.87.11.1350
Thirumalai B Baranyovits PR Intraocular pressure changes and the implications on patient review after phacoemulsification J Cataract Refract Surg 2003 29 504 7 12663014 10.1016/S0886-3350(02)01481-5
Fraunfelder FT Meyer SM Bagby GC JrDreis MW Hematologic reactions to carbonic anhydrase inhibitors Am J Ophthalmol 1985 100 79 81 4014383
Doust J Del Mar C Why do doctors use treatments that do not work? BMJ 2004 328 474 5 14988163 10.1136/bmj.328.7438.474
|
16212667
|
PMC1260022
|
CC BY
|
2021-01-04 16:03:50
|
no
|
BMC Ophthalmol. 2005 Oct 7; 5:24
|
utf-8
|
BMC Ophthalmol
| 2,005 |
10.1186/1471-2415-5-24
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-1021620737910.1186/1471-2458-5-102Research ArticleIs relatively young age within a school year a risk factor for mental health problems and poor school performance? A population-based cross-sectional study of adolescents in Oslo, Norway Lien Lars [email protected] Kristian [email protected] Brit [email protected] Sonja [email protected] Espen [email protected] Institute of General Practice and Community Medicine, University of Oslo, PO Box 1130, Blindern, Oslo 0318, Norway2 The Norwegian Public Health Institute, PO Box 4404 Nydalen, 0403 Oslo, Norway3 Regional Centre for Child and Adolescent Psychiatry, Region East and South, 23 Tasen, 0801 Oslo, Norway2005 5 10 2005 5 102 102 11 11 2004 5 10 2005 Copyright © 2005 Lien et al; licensee BioMed Central Ltd.2005Lien 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
Several studies have shown that children who are relatively young within a school year are at greater risk for poorer school performance compared with their older peers. One study also reported that relative age within a school year is an independent risk factor for emotional and behavioral problems. The objective of this study was to test the hypothesis that relatively younger adolescents in the multiethnic population of Oslo have poorer school performance and more mental health problems than their relatively older classmates within the same school year.
Methods
This population-based cross-sectional study included all 10th-grade pupils enrolled in 2000 and 2001 in the city of Oslo. The participation rate was 88%. Of the 6,752 pupils in the study sample, 25% had a non-Norwegian background. Mental health problems were quantified using the abbreviated versions of Symptom Check List-25 (SCL-10) and the Strength and Difficulties Questionnaire (SDQ). Information on school performances and mental health problems were self-reported. We controlled for confounding factors including parental educational level, social support, gender, and ethnicity.
Results
The youngest one-third of pupils had significantly lower average school grades than the middle one-third and oldest one-third of their classmates (p < 0.001). Of the mental health problems identified in the questionnaires, the groups differed only on peer problems; the youngest one-third reported significantly more problems than the middle and oldest groups (p < 0.05). Age within a school year and gender showed significant interactions with total SDQ score, SDQ peer problems score, SDQ pro social score, and SCL-10 score. After stratifying for gender, the peer problem scores differed significantly between age groups only among boys. The SCL-10 score was significant, but only in girls and in the opposite direction to that expected, with the oldest pupils having significantly higher scores than the other two groups (p < 0.05).
Conclusion
In adolescents from a multicultural city in Norway, relative age within a school year significantly influenced academic performance. In contrast to data from Great Britain, relative age within a school year was not an important risk factor for mental health problems in adolescents in Oslo.
==== Body
Background
Teachers and parents claim that the youngest pupils in school classes perform less well and have more conduct and hyperactivity problems than their older classmates [1]. These allegations have been confirmed in several studies of school performance [2], special educational needs [3,4], learning difficulties [5], and academic performance [6]. Different theories have been proposed to explain this effect. One explanation is that a season-of-birth effect is linked to prenatal exposure to infections [7], similar to the effect identified for schizophrenia and major depression [8]. Another rationalization is that pupils who are relatively younger have less preschool experience and are therefore less well equipped to meet the demanding expectations at school [9]. A third explanation is the age-position effect, which maintains that the oldest are more socially advantaged, mature, and satisfied at school than relatively younger pupils within the same school year [5].
The explanation of a biological season-of-birth effect has been discounted [5] since the phenomenon of poorer performance in relatively younger pupils seems to occur irrespective of different cut-off dates for school enrolment in different countries [10] and geographical regions [11]. The poorer performance in relatively younger pupils is more likely to be related to limited preschool experience or an age-position effect. In Norway and most western countries where all children born in the same calendar year enter school at the same time, the oldest and youngest pupils may differ in age by nearly one full year.
A recent study reported that the effect of relative age of entry into school is also an independent risk factor for mental health problems [11]. The study population was in England, Wales, and Scotland, which have different cut-off dates for school entry: 1 September for England and Wales, and 1 March for Scotland. Children born in summer (May-August) in England and Wales, and in winter (November-February) in Scotland were disadvantaged. The study included the self-report part of the Strength and Difficulties Questionnaire (SDQ) for the age group 11–15 years. Our study was based on the same questionnaire and a similar age group, making these studies directly comparable.
Parental educational level is strongly related to their children's school performance, and is associated with mental health problems [12-14]. A recent study from Norway showed that social support is one of the psychosocial variables that have proven to be consistently associated with mental health during adolescence [15]. Even if research studies have demonstrated that minority youth report the same level of mental health as their majority peers, some groups seems to be at increased risk of mental health problems [16,17]. Several studies has found that girls perform better than boys at school, and that girls have more mental health problems [3,18]. Even if these factors do not relate systematically to seasonal birth, such a relationship cannot be completely ruled out, and the effect of children's ages at school entry should be adjusted for these factors. We controlled for these factors, and examined the possible interaction effects of ethnicity, gender, and age at entry into school on our main outcome variables.
The objective of this study was to explore, in a multicultural population, the association between relative age at school entry, school performance, and mental health problems in Norwegian adolescents. We hypothesized that pupils who are relatively younger when they enter school have lower average school grades and more mental health problems than their older classmates. We controlled for parental educational level, ethnicity, and perceived social support. We designed our study to be similar to other studies to compare our data with those of previous studies in this area.
Methods
Sample
The Oslo Health Study 2000–2001 was conducted as a joint collaboration between the Norwegian Institute of Public Health, the University of Oslo, and the Municipality of Oslo. The part of the study that examined the adolescents in the sample encompassed all pupils in 10th grade (15–16 years old) included in the class lists for each school in Oslo County during the spring terms (March-June) of 2000 and 2001.
All pupils completed two questionnaires during two school sessions. A project assistant was present in the classroom to instruct the students and to perform the practical parts of the survey, such as distributing and collecting the questionnaires. Questionnaires were left at the school for students absent from school on the day of the survey. The school was contacted if the questionnaires were not returned after a while. Students who had not completed the questionnaires after some months were sent questionnaires to complete at home and return in a prepaid return envelope.
A total of 7,343 pupils, representing 88.3% of the 8,316 eligible subjects, answered at least one question. The participation rates were 86.1% in boys and 90.6% in girls. Of the 7,343 pupils, 590 (8.0%) were excluded because they had entered school either one year later or one year earlier than their classmates. The final number of participants whose questionnaires were analyzed was 6,752, comprising two cohorts. The gender, age, and month-of-birth distributions of the participants were representative for the Norwegian population born in 1985–86. All data were self-reported by the pupils.
Variables
The 10-item Symptoms Check List (SCL-10) was used as to measure psychological distress. SCL-10 [19] has approximately the same sensitivity and specificity for detecting psychological symptoms or global distress as the more widely used SCL-25 [20,21] and correlates highly (r = 0.97) with the 25-item version [22]. The 10 items included in the short version are:
Suddenly scared for no reason
Feeling fearful
Faintness, dizziness, or weakness
Feeling tense or keyed up
Blaming yourself for things
Difficult falling asleep, staying asleep
Feeling hopelessness about the future
Feeling blue
Feeling everything is an effort
Feeling of worthlessness
Each item is rated on a scale from 1 (not at all) to 4 (extremely). The distribution of the SCL-10 data for our participants was highly skewed, and the data were transformed logarithmically to approximate a normal distribution.
The Strength and Difficulties Questionnaire (SDQ) is a recently developed questionnaire to assess mental health problems in children and adolescents aged 4–16 years. Its reliability and validity are generally satisfactory [23,24]. The questionnaire has 25 items and scores are classified into five subscales: emotional symptoms, conduct problems, hyperactivity, peer problems, and prosocial behavior. The first four subscales are summed to give the total difficulties score. Each of the SDQ items was scored 1 to 3, with the options "not correct", "partly correct", and "completely correct".
Average grade
The participants were asked to fill in the most recent grade recorded in their school record book in Mathematics, Written Norwegian, English, and Social Science. An average grade score was calculated from these four grades.
Season of birth
We divided the calendar year into three parts. Pupils born in January to April were labeled the "oldest group"; those born in May to August, the "middle group"; and those born in September to December, the "youngest group". The terms "oldest", "middle" and "youngest" refer to relative age according to the starting cut-off age of school in Norway, which is 1 January.
Social support
The questions on social support were formulated as four positive statements on the pupil's perception: (1) of attachment, (2) that his or her opinions were valued, (3) that he or she was helped or supported, and (4) that he or she felt appreciated for each of the social-network items of family, friends, class, and teacher. The scoring alternatives were "completely agree", "partly agree", "partly disagree", and "completely disagree". The scores for each of the social-network variables were summed into one score for each of the items. The four items were then summed into one variable for total social support. The median was used to dichotomize perceived social support into high and low values [25]. Data were missing for 1.5% of the responses for social support from family, 1.4% for support from friends, 3.9% for support from the class, and 4.5% for support from teachers.
Parental education level
Statistics Norway registers data on the education levels of all residents in Norway using the Norwegian Educational Standard (NUS) coding system [26]. The registry data were linked to the questionnaire data, providing us with information on the parents' educational levels. Because only a small number of parents had no formal education or only primary education, this group was classified with those parents with secondary schooling as their highest educational level. Two other groups were defined according to their level of university education: less than ("lower university/college") or more than ("higher university/college") four years of university education. Information on parental education levels was missing for 19% of the participants, and these missing values were included as a separate category for this variable in the analyses.
Ethnicity
"Norwegian" was defined as having one or two Norwegian parents, and "minority" was defined as both parents born outside Norway. Forty-nine (0.7%) of the pupils did not give any information about their parents' countries of birth.
Height and menarche
The questions asked were "How tall are you now?" and "Have you begun to menstruate?", which was answered "yes" or "no".
Missing data for the dependent variables
Seventy individuals (1.0%) had missing data for one or more of the SCL-10 items and were excluded from the analysis. For total SDQ scores and SDQ subscores, the proportion of missing values varied from 225 (3.3%) to 239 (3.6%); 439 (6.9%) of the participants did not provide complete information about grades. A larger proportion of pupils classified in the minority group, and more boys than girls, were excluded because of missing data. The proportion of missing data did not vary systematically by month of birth.
Statistical methods
SPSS for Windows version 11.0 was used in the statistical analysis. Frequencies were observed and cross-table statistics were tested with Pearson's χ2 tests. Analysis of variance (ANOVA) was applied to each of the mental health and grade outcome variables as dependent variables, and age at school entry as an independent variable. To determine whether season of birth was confounded by ethnicity, gender, or parental education level, and to test for their possible interaction effects, these variables were also entered as predictors. Tukey post hoc tests were used to locate any significant differences identified in the ANOVA. Pupils with missing data for any of the variables were excluded from the analysis, with the exception of the variable "parental education level". In the analysis of average grades, data from 735 subjects (10.9%) were removed from the multivariate analysis because of missing data, and in the analysis of the other response variables, data from 6.1%–6.5% of subjects were removed because of missing data.
Ethics
The study protocol was reviewed the Regional Committee for Medical Research Ethics and approved by the Norwegian Data Inspectorate. The study was conducted in full accordance with the World Medical Association Declaration of Helsinki.
Results
The population characteristics did not differ significantly between the three relative age groups (Table 1). To test for differences in physical maturity levels, we calculated the average height of boys and the number who had not reached menarche among girls (all based on self report). The average height of the boys differed significantly between groups: average height was 176.9 cm (range, 175.2–178.5 cm) in the oldest group, 175.8 cm (175.3–176.3 cm) in the middle group, and 174.5 cm (173.8–175.2 cm) in the youngest group (F = 4.3, p = 0.01). The percentage of girls who had not reached menarche also differed significantly between groups: 1.8% of the girls in the oldest group, 2.0% of the middle group, and 3.9% of the youngest group had not yet reached menarche (Pearson's χ2, p = 0.03).
Table 1 Population characteristics and social support in different relative age groups within a school year
Oldest third (n = 2219) Middle third (n = 2160) Youngest third (n = 2373)
Year of birth
1984 cohort 45.1 48.1 47.3
1985 cohort 50.8 49.9 49.0
Gender
Female 49.9 51.9 49.9
Ethnicity
Minority groups 22.8 24.4 25.4
Parental educational level
Higher university 16.9 17.7 17.8
Lower university 26.6 26.0 27.9
Secondary 37.4 37.9 36.5
Missing 19.0 18.5 17.8
Social support
Family low 43.8 41.5 41.6
Class low 36.5 36.0 35.5
Teachers low 41.5 41.4 39.2
Friends low 43.1 41.5 40.9
None of the differences between the groups was significant (p > 0.05) by Pearson's χ2 test.
The eight continuous dependent variables were all SDQ scores (conduct problems, hyperactivity, peer problems, emotional symptoms, prosocial behavior, total SDQ score), logarithm-transformed SCL-10 score, and average school grade. Table 2 shows these scores in the three age groups adjusted for social support, ethnicity, gender, and parental education. The SDQ peer problem scores and average grade differed significantly between age groups: the oldest group scored highest on average grade and lowest on SDQ peer problems, and the youngest group scored lowest on average grade and highest on SDQ peer problems. Of the peer problem questions in the SDQ questionnaire, the age groups differed most on the item "being generally liked".
Table 2 Adjusted average scores and SD of total SDQ, five SDQ subscores, SCL-10 score, and grade in different relative age groups within a school year
Oldest third Middle third Youngest third p
Mean SD Mean SD Mean SD
Total SDQ score 9.81 4.85 9.74 4.88 9.78 4.85 0.81
SDQ Emotional 2.64 2.20 2.53 2.18 2.56 2.20 0.44
SDQ Hyperactivity 3.54 2.01 3.62 2.02 3.55 2.03 0.48
SDQ Conduct problems 2.12 1.61 2.15 1.59 2.15 1.63 0.84
SDQ Peer problems 1.49 1.47 1.49 1.51 1.58* 1.53 0.01
SDQ Pro social 7.45 1.80 7.37 1.82 7.35 1.84 0.97
SCL-10 score 0.79 0.51 0.78 0.51 0.76* 0.49 0.03
Average grade 4.01* 0.79 3.95 0.80 3.92 0.79 <0.001
Scores were adjusted for ethnicity, gender, parental education, and social support from family, class, teachers, and friends. * indicates the group that differed significantly from each other by Tukey's post hoc test (p = 0.05).
The groups differed significantly on the SCL-10 score, but in a direction opposite to that expected in that the youngest group scored lower than the other groups. The Tukey post hoc test showed that the average grade was significantly higher in the oldest group than in the other groups, but that the middle and youngest groups did not differ. The average SDQ peer problem score was significantly higher in the youngest group than in the other two groups (Table 2).
We used ANOVA to examine the interaction between relative age at entry into school and ethnicity, gender, parental education, and social support from family, friends, class, and teachers for each of the dependent variables. The interaction between gender and relative age was significant for total SDQ score (p = 0.04), SDQ peer problems score (p = 0.01), SDQ prosocial score (p = 0.05), and SCL-10 score (p = 0.03).
Because of the significant interaction between gender and age at entry into school, we stratified the data by gender (Table 3). The differences between age groups in the SDQ peer problem scores were significant only in boys, and in the SCL-10 scores only in girls, but in the opposite directions to those hypothesized. Average grade differed between the relative age groups in both boys and girls. The post hoc tests showed that SDQ peer problems and average grades differed between the youngest and oldest groups among boys, and that average grades differed significantly between all groups in girls. None of the other response variables differed significantly when analyzed for boys and girls separately.
Table 3 Adjusted average scores for total SDQ, five SDQ subscores, SCL-10 score, and grade in boys and girls in different relative age groups within a school year
Boys Girls
Oldest Middle Youngest p Oldest Middle Youngest p
Total SDQ score 9.01 8.90 9.31 0.07 10.61 10.56 10.37 0.35
SDQ Emotional 1.74 1.67 1.77 0.24 3.55 3.35 3.27 0.07
SDQ Hyperactivity 3.43 3.42 3.44 0.95 3.68 3.78 3.61 0.19
SDQ Conduct problems 2.34 2.24 2.32 0.39 1.91 2.01 1.97 0.46
SDQ Peer problems 1.52 * 1.56 1.74 * <0.001 1.47 1.42 1.42 0.70
SDQ Pro social 6.96 6.81 6.92 0.41 7.90 7.95 7.84 0.46
SCL-10 score 0.64 0.62 0.62 0.62 0.94 0.94 0.89* 0.003
Average grade 3.90 * 3.87 3.82 * 0.009 4.14 * 4.06 * 4.03 * <0.001
Scores were adjusted for ethnicity, gender, parental education, and social support from family, class, teachers, and friends. * indicates the groups that differed significantly from each other by Tukey's post hoc test (p = 0.05).
Effect size was calculated as the difference in average score between the youngest and oldest group divided by the pooled standard deviations for the response variables that differed significantly between groups. The effect size for boys and girls together was 0.06 for SDQ peer problems, 0.11 for average grades, and 0.06 for logarithm-transformed SCL-10 scores. In boys, the effect size was 0.14 for SDQ peer problems and 0.10 for average grade. In girls, the effect size was 0.04 for average grade and 0.09 for logarithm-transformed SCL-10 scores.
Discussion
This study confirms the data from several other studies showing that children who are relatively young within their school year perform less well than their older classmates [2-6]. The relationship between relative age and school performance seems to be linear, at least for school performance. However, the effect size was small for both boys and girls, according to Cohen's classification [27]. To our knowledge, this is the first study to demonstrate the effect of relative age at entry into school on school performance in Norwegian adolescents.
Of the mental health problems studied, only the SDQ peer problems in boys differed significantly between the three age groups; the scores were highest in the youngest group, although the effect size was small. This might indicate that the youngest pupils in the class experienced difficulty being accepted by their peers. Further studies are required to determine whether the feeling of lack of acceptance among the relatively young pupils influences school performance, and whether boys are more likely to experience peer problems than are girls.
Trends in both emotional subscores in the SDQ and SCL-10 went in the direction opposite to that expected in girls, because girls in the oldest group had higher scores on emotional problems than their younger class mates (although the effect size was small). Our data differ from those of other studies of adolescents, which reported more emotional problems, including self-reported mental health problems, in the youngest age groups [11]. The explanation for these differences might be a more rapid increase in emotional distress in girls than in boys during mid-adolescence [28-30], an effect that may have been more prominent in our older participants (15–16 years) than in the 11–15-year-old pupils studied by Goodman et al. [11].
It is possible that a systematic effect of age at entry into school in our sample was obscured by a stronger age effect. Although the relationship between relative age at entry into school and mental health problems persisted in all age groups in the study by Goodman el al. [11], we believe that this effect wanes with increasing age because the relative age difference in a class becomes less important as the adolescents mature. This might explain why our study showed that relative age had less effect on mental health problems in a sample of 15- and 16-year-old adolescents than in the British study, which examined a wider and younger age range.
Another possible explanation for the discrepancy between our results and those from Great Britain is the differences in the school systems. In the past in Norway, children entered school in their seventh year, but now enter at age six. In Britain, children start school at age four or five. In Norway, there is no formal evaluation of performance before the eighth grade, when the pupils are graded for the first time. In Britain, students are evaluated and choose their course of study earlier than in Norway.
The British school system may place greater demands on pupils than the Norwegian system, at least at earlier ages. This pressure might produce greater differences in mental health status between the younger, less mature pupils and their older classmates in a British setting compared with those in a Norwegian setting [3,9,18]. However, this cannot explain why the relatively older girls in our study reported more emotional distress than their younger peers. One explanation might be the rapid increase in emotional distress among girls during mid-adolescence, as discussed earlier. Another possible explanation is that Norway has a stronger season-of-birth effect than countries located further south, which might manifest as elevated depressive symptom levels among people born during the late winter [8,31]. These hypotheses require further investigation.
Our conclusion about systematic gender differences in the effects of age at entry into school rests on two of 64 possible significant interaction effects (age at entry with each of the covariates for eight outcome measures), and only one was significant at the 0.001 level. We urge caution when drawing conclusions from these relationships.
The strength of our study was the high response rate (88.3%) from all 15- to 16-year-old adolescents in Oslo for two consecutive years in a multicultural environment. Our study might have had selection problems, but these are unlikely to have affected our observations. Two validated questionnaires were used to assess mental health problems and we controlled for known possible confounders in the data analysis.
An important limitation of this study was that all effect measures were based on self-report, and we were unable to include a third-party evaluation by teachers, parents, or diagnostic interviews. Another limitation is that we do not know the extent to which differences in peer or emotional problems affect the average pupil in everyday life, because we did not include the impact part of the SDQ. The study would have benefited by the inclusion of younger age groups to examine whether the effects of relative age within the school year differ between primary school and junior high school, from which our sample was taken.
One of the problems incurred when calculating many main effects and interactions in a single study is the increased risk of identifying significant results by chance. We believe that conclusions should be drawn with care from our data. We were also unable to account for the pupils who did not participate because they were absent temporarily or had dropped out of school. We chose to exclude pupils who had begun school later or earlier than normal, which might have restricted the range of outcome variables somewhat, although we think it unlikely that such a (moderate) restriction in range affected the results.
Conclusion
Our study shows that relatively young age within a school year has a small but significant effect on school performance in adolescents. The effect of relative age on mental health was weak and significant only for peer problems among boys and emotional problems among girls (opposite to the predicted direction). More studies are required to confirm the relatively weak associations demonstrated between relative age within a school year and school performance and mental health problems. The results from this study do not show effect sizes large enough to justify changes in the way children are enrolled in school in Norway.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
LL carried out the statistical analyses and drafted the manuscript. KT took part in the statistical analyses and commented on the drafts. SH and BO took part in writing the methods section and commented on the drafts. EB took part in planning the study, participated in its design and coordination, and commented on the drafts. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Data collection was conducted as part of the Oslo Health Study 2000–2001 in collaboration with the National Health Screening Service of Norway, now the Norwegian Institute of Public Health.
==== Refs
Peagam E From May to December: the birth-month factors in referrals for special education for children with emotional and behavioural difficulties Therapeutic Care and Education 1992 1 101 111
Jinks P An investigation into the effect of date of birth on subsequent school performance Educational Research 1964 6 220 225
Wilson G The effects of season of birth, sex and cognitive abilities on the assessment of special educational needs Educational Psychology 2000 20 153 166
Wallingford E Prout HT The relationship of season of birth and special education referral Psychology in School 2000 37 379 387 10.1002/1520-6807(200007)37:4<389::AID-PITS9>3.0.CO;2-K
Gledhill J Ford T Goodman R Does season of birth matter? The relationship between age within the school year (season of birth) and educational difficulties amongst a representative general population sample of children and adolescents (aged 5–15) in Great Britain Research in Education 2002 68 41 47
Russel R Startup M Month of birth and academic achievement Journal of Personal Individual Experience 1986 7 839 846 10.1016/0191-8869(86)90082-6
Orme J Intelligence and season of birth British Journal of Medical Psychology 1962 35 233 234 13940368
Castrogiovanni P Iapichino S Pacchierotti C Pieraccini F Season of birth in psychiatry. A review Neuropsychobiology 1988 37 175 181 9648124 10.1159/000026499
Sharp C Hutchison D Whetton C How do season of birth and length of schooling affect children's attainment at key stage 1? Educational Research 1994 36 107 121
Berglund G A note on intelligence and season of birth British Journal of Psychology 1967 58 147 151 5625754
Goodman R Gledhill J Ford T Child psychiatric disorder and relative age within school year: cross sectional survey of large population sample British Medical Journal 2003 327 472 475 12946967
Nica-Udangiu S Ionescu R Nica-Udangiu L Popovici E State of mental health and social and familial background as explanatory factors of poor university performance Neurologie et Psychiatrie 1981 19 35 43 7194501
World Health Organization Mental Health: New understanding new hope. The burden of mental and behavioural disorders The world health report 2001 2002 WHO
Bijl RV Cuijpers P Smit F Psychiatric disorders in adult children of parents with a history of psychopathology Social Psychiatry and Psychiatric Epidemiology 2002 37 7 19 11924749 10.1007/s127-002-8208-8
Haavet O Saugstad O Strand J Positive factors associated with promoting health in low-risk and high-risk populations of 15- and 16-year-old pupils in Oslo, Norway Acta Paediatrica 2005 94 345 351 16028654
Berg-Kelly K Eriksson J Adaptation of adopted foreign children at mid-adolescence as indicated by aspects of health and risk taking–a population study European Child and Adolescent Psychiatry 1997 6 199 206 9442998
Oppedal B Roysamb E Heyerdahl S Ethno-cultural group, acculturation, and psychiatric problems in young immigrants Journal of Child Psychology and Psychiatry and Allied Disciplines
West P Sweeting H Fifteen, female and stressed: changing patterns of psychological distress over time Journal of Child Psychology and Psychiatry 2003 44 399 414 12635969 10.1111/1469-7610.00130
Strand BH Dalgard OS Tambs K Rognerud M Measuring the mental health status of the Norwegian population: a comparison of the instruments SCL-25, SCL-10, SCL-5 and MHI-5 (SF-36) Nordic Journal of Psychiatry 2003 57 113 118 12745773 10.1080/08039480310000932
Lipman RS Covi L Shapiro AK The Hopkins Symptom Checklist (HSCL)–factors derived from the HSCL-90 Journal of Affective Disorders 1979 1 9 24 162184 10.1016/0165-0327(79)90021-1
Hesbacher P Rickels R Morris R Newman H Rosenfeld M Psychiatric illness in family practice Journal of Clinical Psychiatry 1980 41 6 10 7351399
Tambs K Moderate effects of hearing loss on mental health and subjective well-being: results from the Nord-Trondelag Hearing Loss Study Psychosomatic Medicine 2004 66 776 782 15385706 10.1097/01.psy.0000133328.03596.fb
Goodman R Psychometric properties of the strengths and difficulties questionnaire Journal of the American Academy of Child and Adolescent Psychiatry 2001 40 1337 1345 11699809 10.1097/00004583-200111000-00015
Goodman R Meltzer H Bailey V The Strengths and Difficulties Questionnaire: a pilot study on the validity of the self-report version International Review of Psychiatry 2003 15 173 177 12745329 10.1080/0954026021000046137
Ystgaard M Life stress, social support and psychological distress in late adolescence Social Psychiatry and Psychiatric Epidemiology 1997 32 277 283 9257518 10.1007/BF00789040
Statistics Norway Den individbaserte utdanningsstatistikken (Individual-based educational statistics) Norwegian Official Statistics NOS C645 2001
Cohen J Statistical Power Analysis for the Behavioral Sciences 1988 Hillsdale, NJ: Erlbaum
Graber JA Lewinsohn PM Seeley JR Brooks-Gunn J Is psychopathology associated with the timing of pubertal development? Journal of the American Academy of Child and Adolescent Psychiatry 1997 36 1768 1776 9401339 10.1097/00004583-199712000-00028
Angold A Erkanli A Silberg J Eaves L Costello EJ Depression scale scores in 8–17-year-olds: effects of age and gender Journal of Child Psychology and Psychiatry and Allied Disciplines 2002 43 1052 1063
Costello EJ Mustillo S Erkanli A Keeler G Angold A Prevalence and development of psychiatric disorders in childhood and adolescence Archives of General Psychiatry 2003 60 837 844 12912767 10.1001/archpsyc.60.8.837
Martin R Foels P Clanton G Moon K The season of birth is related to child retention rates, achievements and rate of diagnosis of LD Journal of Learning Disabilities 2004 37 307 317 15493403
|
16207379
|
PMC1260023
|
CC BY
|
2021-01-04 16:28:57
|
no
|
BMC Public Health. 2005 Oct 5; 5:102
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-102
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-1051621266910.1186/1471-2458-5-105Research ArticleEvaluation of school absenteeism data for early outbreak detection, New York City Besculides Melanie [email protected] Richard [email protected] Farzad [email protected] Don [email protected] Communicable Disease, New York City Department of Health and Mental Hygiene, New York, NY, USA2 Mathematica Policy Research, Inc, Cambridge, MA, USA3 Epidemiology and Surveillance, New York City Department of Health and Mental Hygiene, New York, NY, USA2005 7 10 2005 5 105 105 15 4 2005 7 10 2005 Copyright © 2005 Besculides et al; licensee BioMed Central Ltd.2005Besculides 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
School absenteeism data may have utility as an early indicator of disease outbreaks, however their value should be critically examined. This paper describes an evaluation of the utility of school absenteeism data for early outbreak detection in New York City (NYC).
Methods
To assess citywide temporal trends in absenteeism, we downloaded three years (2001–02, 2002–03, 2003–04) of daily school attendance data from the NYC Department of Education (DOE) website. We applied the CuSum method to identify aberrations in the adjusted daily percent absent. A spatial scan statistic was used to assess geographic clustering in absenteeism for the 2001–02 academic year.
Results
Moderate increases in absenteeism were observed among children during peak influenza season. Spatial analysis detected 790 significant clusters of absenteeism among elementary school children (p < 0.01), two of which occurred during a previously reported outbreak.
Conclusion
Monitoring school absenteeism may be moderately useful for detecting large citywide epidemics, however, school-level data were noisy and we were unable to demonstrate any practical value in using cluster analysis to detect localized outbreaks. Based on these results, we will not implement prospective monitoring of school absenteeism data, but are evaluating the utility of more specific school-based data for outbreak detection.
==== Body
Background
Public health agencies nationwide are developing surveillance systems using non-traditional data sources to enhance outbreak detection capacity. The New York City Department of Health and Mental Hygiene (DOHMH), for example, conducts daily monitoring of ambulance dispatch calls, emergency department (ED) visits, pharmacy sales and worker absenteeism [1-3]. Each of these "syndromic" surveillance systems relies on non-specific, pre-diagnostic data. Having multiple, complementary systems may increase or decrease detection performance. Findings from one data stream can support or contradict findings from another, providing additional evidence on which to decide whether a syndromic signal requires a public health response.
One data source that has been considered for use in syndromic surveillance is school absenteeism records. Reports of elevated school absenteeism made to the Milwaukee Department of Health during the 1993 Cryptosporidium outbreak preceded the recognition of the outbreak [4]. Monitoring of school absenteeism has also been incorporated into the Electronic Surveillance System for the Early Notification of Community-based Epidemics (ESSENCE II) program [5].
Prior to committing public health resources to establishing and maintaining a new syndromic surveillance system, however, it is important to evaluate its potential usefulness. Resources are limited, and even simple systems require staff time and equipment to process data, conduct the analyses and respond to signals. Using a recently developed framework for evaluating early outbreak detection systems as a guide [6], this paper evaluates whether daily monitoring of school absenteeism data is useful for early outbreak detection in New York City considering the use of other syndromic systems. We use three influenza seasons to test the utility of the data. The seasons varied in severity, with 2001–02 characterized as severe, 2002–03 as mild, and 2003–04 as moderate.
Methods
There are approximately 1.1 million students enrolled in the New York City (NYC) public school system. At each of the 1,160 schools, absences are recorded on a "bubble sheet" during a designated homeroom class period. Bubble sheets are forwarded to the school's administrative offices and scanned into a local database that automatically transmits the data to a central database at the NYC Department of Education (DOE). Preliminary data are available by noon each day, however these are subject to correction over the next few days for students who, for example, arrive late. Reason for absence is not recorded.
For the purposes of this evaluation, we downloaded attendance data for the 2001–02, 2002–03, and 2003–04 academic years from the DOE website [7]. Data consisted of the daily percent present (i.e., not absent) aggregated by 'Community School Districts,' which include elementary and middle school children (kindergarten through 8th grade), and 'High School' (9th through 12th grade). We calculated the median daily absentee rates separately among elementary/middle and high school students over the three-year period, and examined whether the absentee rates differed between the two groups. We hypothesized that absenteeism would be significantly higher among the older group of students, making age group-specific analyses necessary. We also assessed whether absenteeism varied by day of week and whether it was higher on days scheduled for parent teacher conferences, state exams, or half days, so that we could control for such days. Statistical significance of differences was measured using the Wilcoxon-signed rank test.
To identify days on which prospective surveillance would have indicated a statistically significant overall increase in absenteeism – or 'signal' – and specifically to determine whether the system could provide timely indications of community-wide influenza, we analyzed daily data from October 1, 2001 through June 25, 2004 retrospectively. The analysis mimicked prospective monitoring, specifically, all data from September 15, 2001 up to and including the day of analysis were used in each days analysis, but no future data. Separate analyses were carried out for elementary/middle and high school students.
There are many reasons why students are absent from school that are unrelated to illness and we wanted to control for these reasons whenever possible. Several steps were therefore taken to minimize the number of false positive signals that would normally result from a time-series analysis on data with many extreme data points (i.e., days with increased absenteeism). We removed extreme data points with a known explanation for absenteeism (e.g., Halloween) from the analysis because they were uninformative. Next, we adjusted the observed daily percent absent using a linear regression model, based on an existing DOHMH ambulance dispatch surveillance model [1]. The percent absent was modeled as a linear function of day-of-week (parameterized as 4 dummy variables with Tuesday as reference) and whether or not the day was a scheduled low attendance day (clerical half-days, parent-teacher conference days, and state exam days). We compared daily percentages with a 14 day baseline using a modified cumulative sums (CuSum) method (8). The two modifications were: 1) We eliminated from the baseline any day on which the residual from the regression model was more than two times the standard error of all residuals. This reduced the influence of extreme, uninformative data points on the baseline mean and standard deviation. 2) We terminated CuSum signals if the percent absent returned to within 0.5 standard deviations of the baseline mean. This reduced the number of multi-day signals that were due to a spike in absenteeism on one day that was extreme enough to cause signals for two or three consecutive days. The daily percent absent was plotted along with any CuSum C1, C2 or C3 signal [8], the daily number of emergency department (ED) patients ages 5–17 complaining of fever or influenza-like illness from an existing ED surveillance system [2], and the weekly number of influenza A and B isolates identified at NYC reference laboratories.
To assess geographic clustering in absenteeism we obtained a more detailed dataset from the DOE, which consisted of the daily number of students registered and absent by school and grade during the 2001–02 academic year. Spatial clustering by school location was assessed using a modified, purely spatial scan statistic [9,10] with a 30-day baseline, 1-day maximum temporal window and 20% (of registered students) maximum spatial window. More than one significant spatial cluster per day was possible and clusters of absenteeism could occur at a single school or several schools in a contiguous geographic area. We evaluated whether this method would have detected the sole gastrointestinal school outbreak reported during the 2001–02 school year. Analyses were carried out using SAS version 8.0 (SAS Institute, Cary, NC) and SaTScan version 4.0.3 (available free at [11]).
Results
Extreme, uninformative data points
Many extreme uninformative data points were removed from the dataset including 42 days just prior to or after a holiday, all 39 days during the first week and last two weeks of school, 7 snow days, and Halloween. In all, 99 (18%) of 538 elementary/middle school days and 104 (19%) of 541 high school days were removed.
Citywide analysis
The median, unadjusted, daily percent absent over the three-year period was 7.3% (range 4.4%-38.9%) among elementary/middle school children and 17.8% (range 3.3%-63.2%) among high school students (Wilcoxon p < 0.01) (see Figure 1). Absenteeism was higher on Mondays and Fridays (median 15.8%), compared to Tuesdays (15.3%), Wednesdays (14.2%) and Thursdays (14.0%) (Wilcoxon p < 0.01). Absenteeism was also more than twice as high on days scheduled for parent-teacher conferences, New York State Regents exams, or half-days (30.4%) compared to other days (13.9%) (Wilcoxon p < 0.01).
Figure 1 Unadjusted daily percent absent among elementary/middle school and high school students in the New York City school system, September 2001 through June 2004.
On 55 (8.0%) of 439 elementary/middle school days and on 52 (8.4%) of 437 high school days, the number of absences observed significantly exceeded the number expected (i.e., signaled) (see Figures 2, 3, 4). On many of these days, low attendance appeared to be unrelated to illness. Despite the removal of extreme, uninformative days, eleven (20%) signals among elementary/middle school children, and 24 (46%) signals among high school students occurred during late May and June as the school year came to a close. An additional 5 (9%) elementary/middle school signals and 14 (27%) high school signals occurred one day prior to or after vacation periods.
Figure 2 Adjusted daily percent absent among elementary/middle school and high school students in the New York City school system for the 2001–02 school year, adjusted for day-of-week and clerical days. Plotted against Emergency Department visits for fever and flu-like illness among children age 5–17 and the number of influenza A and influenza B isolates identified at three World Health Organization reference virology laboratories in New York City.
Figure 3 Adjusted daily percent absent among elementary/middle school and high school students in the New York City school system for the 2002–03 school year, adjusted for day-of-week and clerical days. Plotted against Emergency Department visits for fever and flu-like illness among children age 5–17 and the number of influenza A and influenza B isolates identified at three World Health Organization reference virology laboratories in New York City.
Figure 4 Adjusted daily percent absent among elementary/middle school and high school students in the New York City school system for the 2003–04 school year, adjusted for day-of-week and clerical days. Plotted against Emergency Department visits for fever and flu-like illness among children age 5–17 and the number of influenza A and influenza B isolates identified at three World Health Organization reference virology laboratories in New York City.
Increases in absences coincided with four community-wide outbreaks of influenza-like illness. In the first of these outbreaks, statistically significant increases in absenteeism were detected among elementary/middle school children on five days and among high school students on two days between December 10–20, 2001. This period coincided with the early stages of community-wide influenza A in New York City as evidenced by the number of influenza isolates at World Health Organization reference virology laboratories in NYC and increases in ED visits for fever/flu syndromes.
A second series of signals occurred among elementary/middle school children between March 5–13, 2002, and coincided with increases in positive influenza B isolates at reference laboratories and a season-high peak in ED visits for fever/flu and respiratory syndromes among young children. No increase in absenteeism was observed among high school students at this time (see Figure 2).
During the 2002–03 school year, a third series of signals among elementary/middle school students was observed between January 22–28, 2003. This coincided with a peak in positive influenza A isolates and ED visits for fever/flu and respiratory syndromes among children. Although sporadically elevated during this period, increases in absences among high school students were not statistically significant.
In November 2003, increases in ED visits for fever and flu-like syndrome and positive influenza A isolates at reference laboratories marked an unusually early start to community-wide influenza in New York City. School absenteeism increased exponentially during this period, with seven citywide signals among elementary/middle school students and eight among high school students from December 3–22, 2003.
Geographic clustering analysis
Application of the spatial scan statistic identified 790 geographic areas with statistically significant clustering of absences (p ≤ 0.01) among elementary/middle school children. On average, there were three significant clusters per day with a median of five schools involved in each cluster (mean 32). The mean number of excess absences (observed – expected) in significant clusters was 287 and the mean relative risk (RR = observed/expected) was 2.1. The mean number of excess absences per school in significant clusters was 15.
Spatial cluster analysis detected an increase in absenteeism at one elementary school (school A) during a previously reported gastrointestinal outbreak. SaTScan ranks clusters from most to least significant and several schools or one school can be involved in each cluster. On two consecutive days (November 15 – 16, 2002), school A was the only school in the most significant cluster detected in the city. The number of excess absences and RR of the clusters involving school A on the 15th (235 excess, 2.6 RR) and 16th (152 excess, 3.6 RR) were in the top 99th percentile of all clusters seen throughout the year. However, excesses in absenteeism of this magnitude were not rare; 59 schools involved in clusters had single day excesses of 200 absences or more during the year, and these signals did not stand out among the 788 other significant clusters detected during the academic year.
Among high school students during 2001–02 there were 1,018 significant (p ≤ 0.01) spatial clusters detected, a mean of four per day. The median number of schools involved in each cluster was two (mean 9, SD 18). On average there were 236 excess cases in each cluster (RR 2.1). The median excess number of students absent in each school involved in a cluster was 15 (mean 44, SD 97). There were no known localized outbreaks among high school students during the study period to test the ability of the system to detect such outbreaks.
Discussion
School absenteeism data are collected nationwide and experience with their use in disease surveillance is likely to be of interest to many jurisdictions. Correlating peaks in daily time series from these data with positive influenza isolates is one evidenced-based approach for evaluating the ability to detect acute infectious disease outbreaks. During the 2001–02 public school year we detected moderate increases in student absenteeism associated with peak influenza A activity. Winter holiday recess likely limited continued school transmission of influenza and complicated the interpretation of signals and our ability to detect a stronger relationship. Such gaps in data limit the use of school absenteeism for tracking citywide outbreaks during these periods. We also detected a sustained increase in absenteeism among elementary/middle school students, but not high school students, during the peak of the 2001–02 influenza B season. This is consistent with the known greater susceptibility of young children to influenza B [12] and with increases in ED visits for influenza-like illness among young children but not among teenagers during this period. The relatively mild 2002–03 influenza season was associated with only a small increase in absenteeism among elementary/middle school students.
The unusually early 2003–04 influenza season afforded us the opportunity to evaluate the sensitivity of school absenteeism data to detect community-wide illness while school was in session. In this year, influenza was also severe among children because the strain of influenza circulating at the time was one that children had minimal prior cross immunity to. We observed a nearly two-fold increase in absenteeism during the peak in influenza activity. Signals generated from school absentee data, therefore, appear to require a more severe influenza epidemic among children that does not overlap with school holidays to yield a clear signal. The school absenteeism signals from the 2003–04 influenza season would have been unlikely however to have added to our existing surveillance knowledge as the ED syndromic surveillance system detected increases in influenza-like illness with earlier notification and greater specificity. The same holds true for all outbreaks identified with the school absenteeism data presented.
Spatial cluster analysis of school-level data for 2001–02 was less useful for outbreak detection. While a signal was detected during a known outbreak of gastrointestinal illness, there were several other days when similar increases were seen that appeared not to be associated with a recognized outbreak. Perhaps more importantly, routine investigations of the high number of spatial signals detected is impractical for most local or state public health agencies, given competing priorities and limited staff resources. Although some of the false positives in absenteeism at the school level could have been due to real illness, many other factors were undoubtedly involved. Given the retrospective nature of this analysis, however, we cannot confirm this, as we did not investigate the causes of signals. The magnitude of the increases detected in school absenteeism during influenza season fall far below increases detected by other syndromic surveillance systems in use by the DOHMH. Dramatic and sustained spikes in ED visits for fever and respiratory complaints and ambulance dispatch calls for fever/flu have been observed at the start of community-wide influenza outbreaks in each of the three years that this system has been in operation. Typically these influenza signals occur days to weeks before traditional influenza surveillance [1,2]. In contrast, analysis of school absenteeism data with the methods we used did not clearly differentiate between illness-related absenteeism and non-illness related absences.
Conclusion
Monitoring school absenteeism data in New York City for outbreak detection is appealing for several reasons. Data are population-based, non-confidential, available in close to real time and can be retrieved by linking to a single central database.
Absenteeism may also provide insight into patterns of less severe illness among children who may not seek medical care. In theory, it can potentially capture illness earlier than other surveillance systems. However, in practice the data have several limitations: (1) they are non specific (i.e., the reason for absence is not known), (2) data are not collected throughout the year due to weekends, holidays and vacations, and (3) a large percentage of the days for which data are collected are uninformative, even with statistical adjustment. School absenteeism data would be more useful if they were accompanied by information about the reason students are absent.
Public health researchers should continue to seek out new sources of data to support outbreak detection, but should evaluate their usefulness before investing resources to incorporate them into routine practice. This evaluation suggests that currently available New York City school absenteeism data are too non-specific to be useful for early outbreak detection. Although it is possible that additional filtering and statistical modeling could reduce further the noise in these data, it is impractical to implement this in the public health setting and would be a poor investment given the limited information gained. Therefore, prospective surveillance of school absenteeism will not be implemented in New York City at this time. We are currently evaluating the utility of monitoring chief complaints reported by students visiting the school nurse.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
M Besculides, R Heffernan, and D Weiss conceived of the study. M Besculides and R Heffernan conducted the analysis and D Weiss and F Mostashari provided input on analysis. M Besculides led the writing and R Heffernan assisted with the writing. All authors helped interpret findings and review drafts of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors would like to thank Dr. Melissa Marx for reviewing the manuscript and providing editorial advice.
==== Refs
Mostashari F Fine A Das D Adams J Layton M Use of ambulance dispatch data as an early warning system for community-wide influenza like illness, New York City J Urban Health 2003 80 i43 9 12791778
Heffernan R Mostashari F Das D Karpati A Kuldorff M Weiss D Syndromic surveillance in public health practice, New York City Emerg Infect Dis 2004 10 858 64 15200820
Das D Mostashari F Weiss D Balter S Heffernan R Monitoring Over-the-Counter Pharmacy Sales for Early Outbreak Detection-New York City 2003 Presented at the National Syndromic Surveillance Conference
Proctor ME Blair KA Davis JP Surveillance data for waterborne illness detection: an assessment following a massive waterborne outbreak of Cryptosporidium infection Epidemiol Infect 1998 120 43 54 9528817 10.1017/S0950268897008327
The Electronic Surveillance System for the Early Notification of Community-Based Epidemics Website Accessed November 2003
Buehler JW Hopkins RS Overhage JM Sosin DM Tong V CDC Working Group Framework for evaluating public health surveillance systems for early detection of outbreaks: recommendations from the CDC Working Group MMWR Recomm Rep 53 1 11 2004 May 7 15129191
The New York City Department of Education Website Accessed July 2004
Hutwagner L Thompson W Seeman GM Treadwell T The bioterrorism preparedness and response Early Aberration Reporting System (EARS) J Urban Health 2003 80 i89 96 12791783
Kulldorff M Prospective time periodic geographical disease surveillance using a scan statistic J of the Royal Statistical Society 2001 A164 61 72 10.1111/1467-985X.00186
Kulldorff M A spatial scan statistic Communication in Statistics Theories and Methods 1997 26 1481 96
SatScan Website
Monto AS Sullivan KM Acute respiratory illness in the community. Frequency of illness and the agents involved Epidemiol Infect 1993 110 145 60 8432318
|
16212669
|
PMC1260024
|
CC BY
|
2021-01-04 16:28:57
|
no
|
BMC Public Health. 2005 Oct 7; 5:105
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-105
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-991620215110.1186/1471-2458-5-99Study ProtocolELSID-Diabetes study-evaluation of a large scale implementation of disease management programmes for patients with type 2 diabetes. Rationale, design and conduct – a study protocol [ISRCTN08471887] Joos Stefanie [email protected] Thomas [email protected] Marc [email protected] Michel [email protected] Sabine [email protected] Jochen [email protected] Petra [email protected] Joachim [email protected] Department of General Practice and Health Services Research, University of Heidelberg, Voßstr. 2, D-69115 Heidelberg, Germany2 Centre for Quality of Care Research, Radboud University Medical Centre Nijmegen, P.O. Box 9101, 6500 HB Nijmegen, The Netherlands3 Department of General Practice, University of Frankfurt, Theodor- Stern-Kai 7, D-60590 Frankfurt am Main, Germany4 AQUA-Institute for Applied Quality Improvement and Research in Health Care, Weender Landstr. 11, D-37073 Goettingen, Germany2005 4 10 2005 5 99 99 4 8 2005 4 10 2005 Copyright © 2005 Joos et al; licensee BioMed Central Ltd.2005Joos 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
Diabetes model projects in different regions of Germany including interventions such as quality circles, patient education and documentation of medical findings have shown improvements of HbA1c levels, blood pressure and occurrence of hypoglycaemia in before-after studies (without control group). In 2002 the German Ministry of Health defined legal regulations for the introduction of nationwide disease management programs (DMP) to improve the quality of care in chronically ill patients. In April 2003 the first DMP for patients with type 2 diabetes was accredited. The evaluation of the DMP is essential and has been made obligatory in Germany by the Fifth Book of Social Code. The aim of the study is to assess the effectiveness of DMP by example of type 2 diabetes in the primary care setting of two German federal states (Rheinland-Pfalz and Sachsen-Anhalt).
Methods/Design
The study is three-armed: a prospective cluster-randomized comparison of two interventions (DMP 1 and DMP 2) against routine care without DMP as control group. In the DMP group 1 the patients are treated according to the current situation within the German-Diabetes-DMP. The DMP group 2 represents diabetic care within ideally implemented DMP providing additional interventions (e.g. quality circles, outreach visits). According to a sample size calculation a sample size of 200 GPs (each GP including 20 patients) will be required for the comparison of DMP 1 and DMP 2 considering possible drop-outs. For the comparison with routine care 4000 patients identified by diabetic tracer medication and age (> 50 years) will be analyzed.
Discussion
This study will evaluate the effectiveness of the German Diabetes-DMP compared to a Diabetes-DMP providing additional interventions and routine care in the primary care setting of two different German federal states.
==== Body
Background
Diabetes is a major and growing health care problem. The number of diabetes patients is expected to increase globally from 135 to 300 million between 1995 and 2025 [1], the vast majority will have type 2 diabetes [2]. The quality of care for diabetic patients in Germany was like in many western countries criticized for more than two decades. The Expert Committee of the Government therefore recommended in 2003 diabetes as a priority area [3]. Epidemiological data from primary care was mostly lacking, although recently published data showed that performance of practices and diabetes control in patients could be better than suggested [4-6]. Quality of care is expected to improve within disease management programs (DMP) by the implementation of evidence-based clinical practice, by means of guidelines, quality circles, educational meetings, outreach visits, patient education and by improving coordination among different health care providers [7,8]. Consequently, clinical guidelines and DMPs for primary care have been developed to improve quality and cost-effectiveness of health care for chronic conditions such as diabetes.
In the reform law of 2002 the German Ministry of Health defined a complicated process for the introduction of DMPs and on 27 February 2003 the Federal Insurance Office accredited the first DMP for type 2 diabetes [9]. Several patient- and provider-oriented interventions within the German-style DMP aiming at decreasing mortality and morbidity of patients with type 2 diabetes reducing micro- and macrovascular complications and increasing quality of life of diabetes patients. The underlying criteria therefore have been developed by the newly formed Coordinating Committee and have gone through a tough process of certification before they have been regulated by law by the Ministry of Health in 2002 [10]. The German DMP structure is tightly linked to financial incentives for the sickness funds, i.e. it is linked with the risk structure compensation (= RSC) which was introduced to compensate for differences in the risk structure of the insured population. [9]. Besides the requirement of a comprehensive documentation and the obligation to provide guideline-oriented healthcare the statutory health insurances are obliged to evaluate DMPs by the Fifth Book of Social Code [9,11].
However, until now valid scientific data showing the effectiveness of the German Diabetes-DMP are missing. Recently published data of diabetes model projects in the region of Nordrhein, Sachsen-Anhalt and Baden-Wuerttemberg show improvements of HbA1c levels, blood pressure and occurrence of hypoglycaemia [12-14]. However, the duration of these model projects was too short to draw conclusions concerning clinical endpoints (i.e. a cardiovascular events, amputation rate) and bias because of the preferred inclusion of highly motivated patients can not be excluded [15].
Methods/Design
Aim and design of the study
The aim of the study is to assess the effectiveness of DMP compared to routine care in the primary care setting of two different German federal states (Rheinland-Pfalz and Sachsen-Anhalt). Since DMPs are liable to an implementation process ongoing at the moment which is differing considerably depending on several criteria (e.g. region of Germany, rural/urban area, health insurance) three groups will be observed: a "routine care group without DMP" (= control group; CG), a "DMP real group" (= DMP 1) and a group of patients participating in an (within the scope of the German law regulations) optimally in practices implemented DMP (= DMP 2).
The study is designed as a prospective cluster-randomized comparison of the two intervention groups (DMP 1 and DMP 2) against routine care as control group (Fig 1). The cluster randomization was chosen because this has optimal internal validity (absence of confounders) while avoiding contamination of interventions associated with patient randomisation.
Figure 1 Study design.
Scientific hypotheses
• The German-Diabetes-DMP (DMP 1) is more effective than routine diabetes care (CG)
• An optimally implemented form of the Diabetes-DMP (DMP 2) is more effective than the German-Diabetes-DMP (DMP 1).
Sample size calculation
Sample size calculations for cluster randomized trials differ from sample size calculations for common RCTs requiring a larger number of patients. Based on the main outcome parameters we performed a power calculation with the Cluster Randomization Sample Size Calculator ver.1.02 of the University of Aberdeen on the basis of an ICC of 0,05. Considering possible drop-outs a sample size of 200 GPs will be required (each GP including 20 patients).
Selection of participants
500 GPs in Rheinland-Pfalz and 500 GPs in Sachsen-Anhalt participating in the Diabetes-DMP will be recruited by an invitation letter giving information about the study project. From those who will accept to participate, 100 GPs will be randomized in the DMP group 1 and 100 in the DMP group 2. All patients with a type 2 diabetes out of these practices insured at the general regional health funds (Allgemeine Ortskrankenkassen = AOK) and participating in the DMP will be included in the study then. Incomplete participation in the interventions will be an exclusion criterion in the DMP group 2.
For the control group 4000 AOK patients out of practices not participating in the DMP will be detected through diabetes tracer medication and age (> 50 years) and will be matched. Participation in a DMP offered by another health fund is an exclusion criterion in the control group.
Data collection and analysis
All data will be provided by the AOK. Data of the DMP groups are retrieved by the DMP documentation forms which have to be completed routinely every 3 to 6 months during the scheduled DMP visits of participating patients. The information about the tracer medication identifying patients of the control group is also provided by the AOK. The data reporting will follow the CONSORT recommendations for cluster randomized trials [16].
Data analysis will be performed according to the intention-to-treat (ITT) method. For missing data the 'last observation carried forward' -method will be used.
Outcome-parameter
Outcome parameters were defined considering best available evidence, legal regulations and feasibility. According to our two scientific hypotheses two groups of outcome parameters were defined: "outcome parameters I" for the comparison of the DMP groups and "outcome parameters II" for the comparison of DMP and routine care. This differentiation of the outcome parameters is essential because in the control group availability of patient data will be limited to process indicators only. The outcome parameters are displayed in Table 1.
Table 1 Primary and secondary outcome parameters
OUTCOME PARAMETERS I OUTCOME PARAMETERS II
DMP 1 vs. DMP 2 DMP 1 vs. CG
Primary outcome parameter Primary outcome parameter
Proportion of patients achieving target values for HbA1c and RR according to the legal regulations (10) Proportion of patients with prescriptions for antidiabetic, antihypertensive and lipid-lowering drugs
Secondary outcome parameters Secondary outcome parameters
Proportion of patients with prescriptions for antidiabetic, antihypertensive and lipid-lowering drugs ----
Proportion of patients with referrals to ophthalmologists, specialists for diabetology and diabetic feet Proportion of patients with referrals to ophthalmologists, specialists for diabetology and diabetic feet
Proportion of patients referred to a patient education training for diabetes and hypertension Proportion of patients referred to a patient education training for diabetes and hypertension
Proportion of patients with severe complications (amputation, dialysis etc.) Proportion of patients with severe complications (amputation, dialysis etc.)
Proportion of patients with > 2 hospitalizations in the last 6 months Proportion of patients with > 2 hospitalizations in the last 6 months
Consultation rate Consultation rate
Days of incapacity to work Days of incapacity to work
Mean differences of HbA1c, RR, BMI and glomerular filtration rate ---
SCORE risk chart (RR, cholesterol, smoking status, age, gender) ---
Drop out rate from the DMP ---
Intervention
The minimum requirements for the Diabetes-DMP are regulated by law and defined in the RSAV [10]. According to the legal regulations the intervention in the DMP1 group comprises consultations at 3- or 6-months intervals. During these consultations a detailed diabetes-specific anamnesis and physical examination incl. taking blood pressure and an analysis of HbA1c are carried out. Furthermore agreements are made concerning further treatment, e.g. target values for HbA1c and blood pressure and participation on patient education programs for diabetes or hypertension. All medical findings as well as the current medication have to be documented within structured, standardized documentation sheets at each consultation. If required a referral to a specialist (e.g. ophthalmologist) will be arranged. Furthermore the GPs get a special diabetes-handbook including current, evidence-based information about diabetes therapy. The GPs participating in the DMP1 group will receive no additional intervention.
There are major difficulties introducing new evidence into general practice requiring comprehensive implementation support at different levels (patients, doctors, practice team) [17,18]. Therefore, in the DMP 2 group implementation support at the level of doctors and practice team will be provided. In addition to the clinical interventions at patients level in the DMP group 1 several components aiming at optimizing the implementation process are provided in this group (= implementation interventions) (Table 2).
Table 2 Implementation Interventions in DMP 2 group
Interventions Description
Interactive quality circle meetings During these meetings all aspects of evidence based treatment of diabetes in a primary care setting will be discussed (2 × per year).
Educational meetings for medical assistants During these meetings medical assistants will be supported in finding individual strategies for optimal implementation of the DMP in their practices (2 × per year).
Outreach visits During these meetings individual problems within the implementation process of the DMP will be discussed with the GP and the assistant team (1 × per year)
Homepage with "best practice" examples Detailed information for the praxis team about Diabetes-DMP incl. case studies via internet (electronic individual feedback)
The control group represents routine care without DMP. There will be no intervention at all.
Timeframe of the study
The study team will start to invite GPs at the end of September 2005. After receiving a written declaration of their willingness to participate in the study and to accept random assignment to the different groups GPs will be randomized.
Patient inclusion and pre-data collection will start in 2005. The observation period will be 24 months.
Description of risks
Since the interventions aim at the evidence based improvement of skills of GPs and practice teams serious risks or undesired effects for patients are not to be expected. There are no specific risks related to the study.
Patient informed consent
Previous to DMP participation patients already receive written and oral information about the content and extent of the DMP by their treating GPs, i.e. about potential benefits for their health and potential risks. Furthermore, patients are informed that DMP data including medical data will be analyzed. In case of acceptance they have to sign a special DMP participation form.
Ethical principles and vote of the ethics committee
The study is being conducted in accordance with medical professional codex and the Helsinki Declaration as of 1996 as well as the German Federal Data Security Law (BDSG). DMP participation of patients is voluntary and can be cancelled at any time without provision of reasons and without negative consequences for their future medical care.
The study protocol was approved by the ethics committee of the University of Heidelberg in April 2005 (Approval-Nr. 098-2005). Furthermore the evaluation of the DMP is regulated by law in the Fifth Book of Social Code (§137 f. Abs. 4 SGB V).
Data security/disclosure of original documents
The patient names and all other confidential information fall under medical confidentiality rules and are treated according to German Federal Data Security Law (BDSG).
Data management will be performed by the AQUA-Institute, Goettingen. All study related data are stored on a protected central server. Only direct members of the internal study team can access the respective files. Intermediate and final reports are stored at the office of the Department of General Practice and Health Services Research at the Heidelberg University Clinic.
Discussion
There are specific difficulties in evaluating the effectiveness of complex interventions such as disease management programs [19], particularly, in our case that implementation process has already started [15]. At the moment it is not conceivable to which phase of the implementation process the DMP actually has preceded. However, the DMP group 2 represents an ideal implementation of the DMP within the scope of the German law regulation.
To adapt the evaluation design to the real conditions of diabetes care in Germany at the moment, but also to consider the ongoing implementation process we have chosen the design of a prospective randomized-controlled comparison of the two DMP groups embedded in the quasi-experimental design of a controlled before-after study with a blinded control group. The quasi-experimental design is predetermined because allocation to the control group can not be perfomed randomly and because baseline values for patients participating in the DMP have not been documented.
A fully experimental design with a randomized control group was rejected by the authors, because this could create an artificial care situation. That is informing patients and GPs of the control group about the study and probably asking for completing any questionnaires would introduce an enormous bias.
According to the theoretical model for design and evaluation of complex interventions by Campbell et al the presented study can be assigned to phase III and IV (15, 20). However, pilot projects have shown that an observation period of 24 months will be too short to show significant differences in severe clinical endpoints (i.e. amputations, diabetic renal insufficiency, cardio-vascular events). Therefore it was decided to have a combination of HbA1c and blood pressure as primary endpoint in the randomized part of the trail and prescriptions in the quasi-experimental part of the trial.
Competing interests
The authors declare that they have no competing interests. A contract between the authors and the funder (who might have an interest to show the effectiveness of DMP in this context) ensures that the authors have the full scientific responsiblility and will publish the results in any case.
Authors' contributions
SJ, TR, JS and MW conceived the study and draft the manuscript. SL, MH, JG and PKK participated in designing the study. 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
The study is funded by the AOK, the general regional health funds (= Allgemeine Ortskrankenkassen), Germany.
==== Refs
King H Aubert RE Herman WH Global burden of diabetes, 1995-2025: prevalence, numerical estimates, and projections Diabetes Care 1998 21 1414 1431 9727886
Amos AF McCarty DJ Zimmet P The rising global burden of diabetes and its complications: estimates and projections to the year 2010 Diabet Med 1997 14 Suppl 5 S1 85 9450510
Sachverständigenrat zur Begutachtung der Entwicklung im Gesundheitswesen Finanzierung, Nutzerorientierung und Qualität 2003
T U Szecsenyi J S HD K The Sinsheim Diabetes Study
A Representative Cross-Sectional Study on the Quality of the Care of Type-2-Diabetics in General Practices
[Eine repräsentative Querschnittstudie zur Versorgungsqualität von Typ-2-Diabetikern in der Hausarztpraxis] Z Allg Med 2004 80 497 502
Rothenbacher D Ruter G Saam S Brenner H [Management of patients with type 2 diabetes. Results in 12 practices of general practitioners] Dtsch Med Wochenschr 2002 127 1183 1187 12035113 10.1055/s-2002-31940
Rothenbacher D Ruter G Saam S Brenner H Younger patients with type 2 diabetes need better glycaemic control: results of a community-based study describing factors associated with a high HbA1c value Br J Gen Pract 2003 53 389 391 12830567
Renders CM Valk GD Griffin S Wagner EH Eijk JT Assendelft WJ Interventions to improve the management of diabetes mellitus in primary care, outpatient and community settings Cochrane Database Syst Rev 2001 CD001481 11279717
Weingarten SR Henning JM Badamgarav E Knight K Hasselblad V Gano AJ Ofman JJ Interventions used in disease management programmes for patients with chronic illness-which ones work? Meta-analysis of published reports BMJ 2002 325 925 12399340 10.1136/bmj.325.7370.925
Busse R Disease management programs in Germany's statutory health insurance system Health Aff (Millwood ) 2004 23 56 67 15160803 10.1377/hlthaff.23.3.56
Bundesausschuß G Anforderungen an die Ausgestaltung von Disease-Management-
Programmen für Patienten mit Diabetes mellitus Typ 2 2002
M M U K Demands on the Evaluation of Complex Healthcare Programs - Disease Management in Germany Z Arztl Fortbild Qualitatssich 2005 99 179 184 15999580
A E M S Improved Diabetes Care in Saxony -Anhalt-results of the Evaluation of the Diabetes Model project (First Quarter 2001-Last Quarter 2002) Z Arztl Fortbild Qualitatssich 2005 99
G B HK S Evaluation of the Pilot Project for the Gradula Area-Wide Outpatient Medical Care of Patients with Diabetes Mellitus in Southern Wuerttemberg Z Arztl Fortbild Qualitatssich 2005 99 185 189 15999581
L A W H J O G B Modernes Diabetesmanagement in der ambulanten Versorgung Ergebnisse der wissenschaftlichen Begleitung der Diabetesvereinbarungen in der Kassenärztlichen Vereinigung Nordrhein 2002 Köln, Deutscher Ärzte-Verlag
Gerlach FM Beyer M Szecsenyi J Raspe H [Evaluation of disease management programs--current deficits, demands and methods] Z Arztl Fortbild Qualitatssich 2003 97 495 501 14611145
Campbell MK Elbourne DR Altman DG CONSORT statement: extension to cluster randomised trials BMJ 2004 328 702 708 15031246 10.1136/bmj.328.7441.702
Grol R Grimshaw J From best evidence to best practice: effective implementation of change in patients' care Lancet 2003 362 1225 1230 14568747 10.1016/S0140-6736(03)14546-1
Grol R WMEM Improving Patient Care The implementation of change in clinical practice, 2005 Oxford, Elsevier
Øvretveit J Evaluating health interventions 2000 London, Open University Press
Campbell M Fitzpatrick R Haines A Kinmonth AL Sandercock P Spiegelhalter D Tyrer P Framework for design and evaluation of complex interventions to improve health BMJ 2000 321 694 696 10987780 10.1136/bmj.321.7262.694
|
16202151
|
PMC1260025
|
CC BY
|
2021-01-04 16:28:57
|
no
|
BMC Public Health. 2005 Oct 4; 5:99
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-99
|
oa_comm
|
==== Front
BMC Plant BiolBMC Plant Biology1471-2229BioMed Central London 1471-2229-5-211620217410.1186/1471-2229-5-21DatabasePineappleDB: An online pineapple bioinformatics resource Moyle Richard L [email protected] Mark L [email protected] Jonni [email protected] David J [email protected] José R [email protected] Plant Genetic Engineering Laboratory, Department of Botany, University of Queensland, QLD 4072, Brisbane, Australia2 Institute of Molecular Biosciences, University of Queensland, QLD 4072, Brisbane, Australia3 Department of Biological Science, University of Waikato, Hamilton, New Zealand2005 5 10 2005 5 21 21 9 6 2005 5 10 2005 Copyright © 2005 Moyle 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
A world first pineapple EST sequencing program has been undertaken to investigate genes expressed during non-climacteric fruit ripening and the nematode-plant interaction during root infection. Very little is known of how non-climacteric fruit ripening is controlled or of the molecular basis of the nematode-plant interaction. PineappleDB was developed to provide the research community with access to a curated bioinformatics resource housing the fruit, root and nematode infected gall expressed sequences.
Description
PineappleDB is an online, curated database providing integrated access to annotated expressed sequence tag (EST) data for cDNA clones isolated from pineapple fruit, root, and nematode infected root gall vascular cylinder tissues. The database currently houses over 5600 EST sequences, 3383 contig consensus sequences, and associated bioinformatic data including splice variants, Arabidopsis homologues, both MIPS based and Gene Ontology functional classifications, and clone distributions. The online resource can be searched by text or by BLAST sequence homology. The data outputs provide comprehensive sequence, bioinformatic and functional classification information.
Conclusion
The online pineapple bioinformatic resource provides the research community with access to pineapple fruit and root/gall sequence and bioinformatic data in a user-friendly format. The search tools enable efficient data mining and present a wide spectrum of bioinformatic and functional classification information. PineappleDB will be of broad appeal to researchers investigating pineapple genetics, non-climacteric fruit ripening, root-knot nematode infection, crassulacean acid metabolism and alternative RNA splicing in plants.
==== Body
Background
In terms of commercial production, pineapple [Ananas comosus (L.) Merrill] is the third most important tropical fruit after banana and mango. Pineapple fruits are classified as non-climacteric, as there is no respiratory burst or spike in ethylene production during ripening and exogenous application of ethylene does not rapidly accelerate fruit ripening. Much has been learnt about the control of fruit ripening in climacteric fruit using tomato as a model system. In particular, manipulation of genes involved in the ethylene biosynthetic pathway and a MADS box transcription factor have led to altered ripening characteristics [1-5]. Conversely, almost nothing is known of how non-climacteric fruit ripening is controlled. Efforts to identify genes controlling non-climacteric fruit ripening are hampered by the small number of non-climacteric fruit gene sequences available for study. Thus, as a first step toward understanding the molecular basis of non-climacteric fruit ripening in pineapple, an EST sequence project has been initiated to isolate expressed sequences from mature green unripe and yellow ripened pineapple fruits [6].
Many crop species, including pineapple, are susceptible to root-knot nematode infection. Crop losses due to nematode infections are estimated to be more than 100 billion dollars each year [7]. Additionally, the toxic soil fumigants used to control nematodes are becoming increasingly banned in many countries. Understanding the molecular mechanisms governing the nematode-plant interaction is of utmost importance in developing alternative strategies for the control of nematode infection. As such, we have constructed EST sequencing libraries from pineapple root and gall vascular cylinder tissue infected with the root-knot nematode Meloidogyne javanica. The vascular cylinder contains the giant cell structures that the nematode feeds upon, and can be dissected from the root cortex and stripped of nematodes with relative ease. Sequencing EST clones from such libraries is a first step toward isolating and identifying gene sequences involved in the plant-nematode interaction.
The collection of EST sequence information requires accurate gene annotation as well as dedicated platforms for storage, processing, curation and data retrieval. Ideally, collected sequence information should be easily accessed, and presented in a user-friendly format that provides the tools to mine the data efficiently. PineappleDB was developed to provide the research community with access to a curated and searchable bioinformatics resource housing fruit, root and gall derived EST sequences, contig sequences, clone annotations, functional classification and Gene Ontology information – all via a user-friendly web interface. PineappleDB will be a valuable resource of broad appeal to researchers studying pineapple genetics, crassulacean acid metabolism, non-climacteric fruit ripening, alternative RNA splicing in plants and root-knot nematode infection.
Construction and content
Database architecture
The database was developed using MySQL 4.0, and implemented on a server running RedHat 9.0. The web interface uses cgi scripts written in Perl 5.8.1. Perl scripts were also used for data processing and uploading into the database. Field descriptions and a full database schema are provided on the help page of the website.
PineappleDB web interface
The pineapple bioinformatic resource can be accessed through a web interface [8]. The introductory page contains information about the pineapple bioinformatic resource, access to the pineapple EST database, lists of contigs containing full-length coding sequences, alternatively spliced clones, putative nematode sequences, and links to pineapple related web pages.
The pineapple EST sequence database can be searched by cloneID, contigID, text, or by sequence homology to either individual EST sequences or contig consensus sequences. Both BLASTN and TBLASTX searches can be performed against the pineappleDB with multiple searches also possible. The search output contains clone and contig information, including sequence, putative identification, a link to the nearest homologue in the NCBI nr database, length of homologous sequence and percentage identity of nearest match, a link to information on the closest homologue in Arabidopsis, MIPS based functional classifications, Gene Ontology information, and the presence of splice variants (Fig. 1). Links to all EST clones clustering within the same contig, and the distribution of the EST's across the fruit and root libraries are also listed in the search output.
PineappleDB will be periodically upgraded as annotation, functional classification, and GO information are updated.
Utility
EST sequencing and bioinformatic analysis pipeline
Users may refer to the pineappleDB flow diagram on the homepage of the website for an overview of the bioinformatics pipeline. In total, 7296 clones from five libraries were 5' end sequenced. The libraries were constructed from uninfected root tips (~2 cm), dissected vascular cylinders of galls from early infection (1–4 weeks post infection), dissected vascular cylinders of galls from late infection (5–10 weeks post infection), mature green fruit and mature yellow fruit. Over 75% of clones returned an average Phred20 score of more than 700 bp. Raw sequences were manually edited for sequence quality and trimmed of plasmid contaminant and polyA tail in the sequence viewer program Chromas v2.13 (Technelysium). 5861 edited sequences were retrieved at an average read length of 769 bp. The 1615 clones with poor sequence quality and/or yielding less than 150 bp of insert sequence were eliminated from further bioinformatic analysis.
The 408 green fruit, 1140 yellow fruit, 343 root tip, 1298 early infection and 2461 late infection edited EST sequences were clustered into 3383 contigs, using SeqMan sequence assembly software (DNASTAR Inc. Madison, USA) and key parameters of minimum 90% match over at least 45 bp overlap. All edited EST sequences have been submitted to the GenBank dbEST [GenBank: CO730741-CO732287, DT335767-DT339792] [9]. Each sequence was assigned a putative identification by BLASTX alignment of contig consensus sequences to the GenBank non-redundant (nr) protein database [10]. Those clones that did not retrieve a BLASTX match better than the 10-20 E-value cut-off were annotated as an undiscovered sequence. The sequences were also BLASTX searched against MATDB to retrieve putative MIPS based functional classifications for clones with a homologous gene from the model plant organism Arabidopsis thaliana [11]. GO classifications were obtained by downloading GO results from a multiple search of the TAIR Arabidopsis resource [12]. A semi-automated process of parsing BLAST hits and manually curating the putative annotations resulted in a spreadsheet of information including cloneID, contig number, number of clones in each contig, nearest BLASTX match, accession number of match, length of match, percent similarity, putative annotation and functional classifications.
Identification of full length clones
Contig consensus sequences containing a polyA tail were analyzed for open-reading frames using EditSeq sequence analysis software (DNASTAR Inc. Madison, USA) and were BLASTX searched against the GenBank nr protein database [13]. Contig consensus sequences were identified as containing a putative full length coding sequence by alignment to known full-length protein coding sequences, and/or by the presence of stop codons upstream of a significant open reading frame. A list of the putative full length coding sequences identified is present in the pineapple bioinformatic resource.
Identification of splice variants
Edited clone sequences generally assembled into contigs with 97–100% homology. However, the contig assembly report occasionally revealed incidences where some clones clustered with between 90–97% homology or that some clones did not cluster into existing contigs due to homology somewhat below the 90% threshold. An inspection of these clone sequences and contigs revealed the presence of apparently unspliced intron sequence, and/or the absence of exon sequence in some of the clones. A comparative analysis to other clone sequences within the contig alignment and to homologous protein coding sequences in GenBank verified that 120 clones contain an apparent "mis-splicing" event. The putative splice variant clones containing un-spliced intron sequence and/or missing spliced exon sequence are listed in the pineapple bioinformatic resource. The presence or absence of a putative splice variant is also reported in contig/clone search outputs.
Identification of putative nematode sequences
Despite precautions to remove nematodes from root tissues prior to library construction, it was anticipated that there would be some contamination of the pineapple gall libraries with nematode derived sequences. All contig consensus sequences containing root and gall EST's were BLASTN searched against the GenBank dbEST and BLASTX searched against the GenBank nr database. Matches to known nematode sequences were manually inspected and 77 contigs identified as containing a putative nematode sequence. All contigs containing putative nematode sequence are listed within the online pineapple bioinformatic resource.
Conclusion
The pineapple EST sequencing project was initiated as a first step toward identifying genes involved in and the molecular basis of non-climacteric fruit ripening and the nematode-plant interaction. The online pineapple bioinformatics resource was developed to house EST sequence information and associated bioinformatic data in a user-friendly format. PineappleDB can be freely accessed via the internet, and currently contains BLAST and text search tools to efficiently mine the dataset for clones and contigs of interest. The resulting search outputs contain comprehensive information on the clone and contigs including cloneID, contig number, number of clones in each contig, nearest BLASTX match, accession number of match, length of match, percent similarity, putative annotation, splice variants, MIPS based functional classifications, Gene Ontology classifications, and the distribution of clones from each library (fig. 1). Links are also provided to other clones within the same contig, the GenBank BLASTX nearest neighbour, and to homologous coding sequences from the model organism Arabidopsis thaliana.
PineappleDB houses the first reported collection of EST sequences isolated from pineapple. PineappleDB will grow as more EST sequence information becomes available. Furthermore, we have initiated a pineapple microarray project and it is anticipated that gene expression data will be incorporated into the online pineapple bioinformatics resource in the future. The EST database will periodically be upgraded as annotation, functional classification, and gene ontology information is updated.
Availability and requirements
The PineappleDB resource can be accessed via
Contact: Dr. José R Botella at [email protected]
Authors' contributions
*The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint first authors. RM was responsible for data collection, the bioinformatic pipeline and manuscript preparation. MC developed the online database and undertook batch BLAST processes. JR-K contributed to the fruit EST editing and identification of full length coding sequences. DJF participated in the conception, design and co-ordination of the study and helped complete the manuscript. JRB designed, supervised and coordinated the project.
Acknowledgements
The pineapple bioinformatic resource was funded in part by an ARC-Linkage grant with Golden Circle Limited. M.L.C. was funded by the Australian Research Council Special Research Centre for Functional and Applied Genomics.
Figures and Tables
Figure 1 Example of contig search output information from PineappleDB.
==== Refs
Ben Amor M Flores B Latchs A Bouyazen M Pech JC Inhibition of ethylene biosynthesis by antisense ACC oxidase RNA prevents chilling injury in Charentais cantaloupe melons Plant Cell Environ 1999 22 1579 1586 10.1046/j.1365-3040.1999.00509.x
Hadfield KA Dang T Guis M Pech JC Bouzayen M Bennett AB Characterization of ripening-regulated cDNAs and their expression in ethylene-suppressed charentais melon fruit Plant Physiol 2000 122 977 983 10712562 10.1104/pp.122.3.977
Theologis A Oeller PW Wong LM Rottmann WH Gantz DM Use of a tomato mutant constructed with reverse genetics to study fruit ripening, a complex developmental process Dev Genet 1993 14 282 295 8222344 10.1002/dvg.1020140406
Vrebalov J Ruezinsky D Padmanabhan V White R Medrano D Drake R Schuch W Giovannoni J A MADS-box gene necessary for fruit ripening at the tomato ripening-inhibitor (rin) locus Science 2002 296 343 346 11951045 10.1126/science.1068181
Adams-Phillips L Barry C Giovannoni J Signal transduction systems regulating fruit ripening Trends Plant Sci 2004 9 331 338 15231278 10.1016/j.tplants.2004.05.004
Moyle R Fairbairn DJ Ripi J Crowe M Botella JR Developing pineapple fruit has a small transcriptome dominated by metallothionein J Exp Bot 2005 56 101 112 15520025
Abad P Favery B Rosso MN Castagnone-Sereno P Root-knot nematode parasitism and host response: molecular basis of a sophisticated interaction Mol Plant Path 2003 4 217 224 10.1046/j.1364-3703.2003.00170.x
PineappleDB: The Online Pineapple Bioinformatics Resource
Boguski MS Lowe TM Tolstoshev CM dbEST--database for "expressed sequence tags" Nat Genet 1993 4 332 333 8401577 10.1038/ng0893-332
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
Schoof H Ernst R Nazarov V Pfeifer L Mewes HW Mayer KF MIPS Arabidopsis thaliana Database (MAtDB): an integrated biological knowledge resource for plant genomics Nucleic Acids Res 2004 32 D373 6 14681437 10.1093/nar/gkh068
The Arabidopsis Information Resource
National Center for Biotechnology Information []
|
16202174
|
PMC1260026
|
CC BY
|
2021-01-04 16:03:52
|
no
|
BMC Plant Biol. 2005 Oct 5; 5:21
|
utf-8
|
BMC Plant Biol
| 2,005 |
10.1186/1471-2229-5-21
|
oa_comm
|
==== Front
Biomed Eng OnlineBioMedical Engineering OnLine1475-925XBioMed Central London 1475-925X-4-561620216410.1186/1475-925X-4-56Book ReviewReview of "Computational Modeling of Genetic and Biochemical Networks" edited by James M. Bower and Hamid Bolouri Dubitzky Werner [email protected] School of Biomedical Sciences, University of Ulster, Coleraine, Co. Londonderry, BT52 1SA, UK2005 4 10 2005 4 56 56 Bower JM and Bolouri H , editors.
Computational Modeling of Genetic and Biochemical Networks .
Cambridge, Massachusetts; London, England: The MIT Press . 2001 . 390 pages, ISBN Number: 0262024810, $35 . 19 9 2005 4 10 2005 Copyright © 2005 Dubitzky; licensee BioMed Central Ltd.2005Dubitzky; 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
Since the coming-of-age of the scientific discipline called bioinformatics it has become increasingly clear that computers in biology will be important not only for managing and analyzing biological and biomedical data but also for modeling and simulation of life processes and systems. The field concerned with this aspect of computers in biology has become known as computational biology. Computational Modeling of Genetic and Biochemical Networks, edited by James M. Bower and Hamid Bolouri, is a text that deals exclusively with computational biology concepts and applications. The book can be seen as an advocate of this increasingly important discipline as it presents a range of problems and methodologies that demonstrate that biology can be approached systematically and systemically by fully characterizing – through a combination of theory, simulation and experiment – entire biological systems such as metabolism, signal transduction and gene regulation, their interacting biochemical networks, or even higher levels of biological organization. The book is organized into two parts: Part I. Modeling Genetic Networks and Part II. Modeling Biochemical Networks.
The first part of the book is concerned with models of gene regulation, including protein-DNA and DNA-DNA interactions. Five chapters are devoted to this topic, by reviewing several tools and methods that are currently available for unraveling genetic regulatory networks at various levels of resolution and abstraction. They include logical and probabilistic approaches, which are applied to problems including prokaryotic and eukaryotic systems. Chapter 1 provides an excellent introduction into the basics of gene regulation and reviews a range of methods and tools that have been used to model gene-regulatory mechanisms and systems. Against this background, Chapters 2 to 5 discuss different studies and methodologies in detail.
The second part of the book tackles protein interactions produced by gene regulation. It first considers interactions among few molecules and then goes on to present models aimed at understanding reactions and diffusion by large numbers of molecules. This part has six chapters. Although an overview chapter, like Chapter 1 in Part I is missing, collectively, the introductions of the six chapters provide an interesting and comprehensive overview of the different biological systems and their background and the relevant modeling methodologies and tools. The biological systems considered in this part include cell cycle regulation, signaling pathways, and excitable membranes and synaptic interactions.
Overall, the volume provides an excellent and broad overview of computational biology and the methodologies and tools needed to model and simulate complex biological systems. One message the book conveys is that tackling computational biology problems requires significant effort and considerable knowledge of mathematics, information technology (IT) and biology. While a great effort is made to cover the relevant background in biology, theory and IT, it could nevertheless be difficult to follow some of the detailed discussions. This is a consequence of presenting a wide range of biological problems and formal and computational methods.
|
0
|
PMC1260027
|
CC BY
|
2021-01-04 16:37:34
|
no
|
Biomed Eng Online. 2005 Oct 4; 4:56
|
utf-8
|
Biomed Eng Online
| 2,005 |
10.1186/1475-925X-4-56
|
oa_comm
|
==== Front
Malar JMalaria Journal1475-2875BioMed Central London 1475-2875-4-491619754110.1186/1475-2875-4-49ResearchEffect of larval crowding on mating competitiveness of Anopheles gambiae mosquitoes Ng'habi Kija R [email protected] Bernadette [email protected] Gamba [email protected] Bart GJ [email protected] Gerry F [email protected] Heather M [email protected] Ifakara Health Research and Development Centre (IHRDC), P. O. Box 53, Ifakara, Tanzania2 University of Dar es Salaam, P. O. Box 35064 Dar es Salaam, Tanzania3 International Atomic Energy Agency (IAEA), Agency's Laboratories Seibersdorf, Seibersdorf A-2444, Austria4 Laboratory of Entomology. P.O. Box 8031, 6700 EH, Wageningen University, Wageningen, The Netherlands5 Department of Public Health and Epidemiology, Swiss Tropical Institute, Basel, Switzerland2005 30 9 2005 4 49 49 17 8 2005 30 9 2005 Copyright © 2005 Ng'habi et al; licensee BioMed Central Ltd.2005Ng'habi 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 success of sterile or transgenic Anopheles for malaria control depends on their mating competitiveness within wild populations. Current evidence suggests that transgenic mosquitoes have reduced fitness. One means of compensating for this fitness deficit would be to identify environmental conditions that increase their mating competitiveness, and incorporate them into laboratory rearing regimes.
Methods
Anopheles gambiae larvae were allocated to three crowding treatments with the same food input per larva. Emerged males were competed against one another for access to females, and their corresponding longevity and energetic reserves measured.
Results
Males from the low-crowding treatment were much more likely to acquire the first mating. They won the first female approximately 11 times more often than those from the high-crowding treatment (Odds ratio = 11.17) and four times more often than those from the medium-crowding treatment (Odds ratio = 3.51). However, there was no overall difference in the total number of matings acquired by males from different treatments (p = 0.08). The survival of males from the low crowding treatment was lower than those from other treatments. The body size and teneral reserves of adult males did not differ between crowding treatments, but larger males were more likely to acquire mates than small individuals.
Conclusion
Larval crowding and body size have strong, independent effects on the mating competitiveness of adult male An. gambiae. Thus manipulation of larval crowding during mass rearing could provide a simple technique for boosting the competitiveness of sterile or transgenic male mosquitoes prior to release.
==== Body
Background
Mosquitoes within the Anopheles gambiae species complex are the most important vectors of malaria in sub-Saharan Africa [1-3]. The infective bite of these mosquitoes is in large part responsible for the more than 500 million clinical attacks of malaria reported worldwide each year, resulting in more than one million deaths [4,5]. Currently the two most widely implemented vector control strategies are indoor residual insecticide spraying and insecticide-treated bednets (ITNs), both of which have proven effective in the reduction of malaria transmission in some areas [6-11]. However, multiple insecticide resistance is emerging amongst the major malaria vectors An. gambiae [12] and Anopheles funestus [13], and there are complications associated with introduction, distribution and proper use of ITNs [14,15] that indicate these strategies alone may not be sufficient to eliminate malaria transmission. New tools aimed at stopping malaria development in humans are promising, but the development of an efficacious antigen for vaccine production is slow, and parasite resistance to locally available drugs is increasing whilst new drugs that are effective are often unaffordable [16].
One promising new control prospect is the possibility of rendering wild vector populations less susceptible to infection by releasing mosquitoes that are genetically modified to resist infection [17-19], or sterile males that will mate with wild females and stop them from reproducing [20]. In the case of a genetically modified mosquito (GMM) strategy, malaria could be reduced by fixing a resistance gene in vector populations, [21-23], and in the case of sterile male release, malaria could be cut by a collapse in the vector population due to a high frequency of unviable matings. Any such release of sterile or GM mosquitoes should consist only of males [20,24] because this sex does not blood feed, and thus they will not increase the number or nature of mosquito bites per person at release sites. The success or failure of a GMM or sterile programme will depend largely on whether released males can successfully compete for mates against wild males [25,26]. Current evidence from laboratory experiments suggests that GMMs have reduced competitiveness and are generally out-competed in the presence of unmodified laboratory-reared males [27-29]. Operationally, the consequences of releasing males with poor competitiveness are dire. For example, the general failure of mosquito control programmes launched in the 1970s that aimed to reduce vector populations by releasing sterile males can be largely attributed to their poor mating competitiveness [20,24], and to a lesser extent, the dispersal of fertile males into control areas. In the case of GMM, some argue that even if modified males have lower fitness than the wild type, refractory genes will still spread provided they are linked to an efficient genetic drive mechanism [30]. However, such a drive mechanism could only act if insemination occurs in the first place [24], which it may not if GMM competitiveness is very low. Furthermore, no efficient genetic drive mechanism has yet been identified for Anopheles, and even assuming one is, there are doubts about whether it could be tightly linked to a potentially costly resistance gene [23,31,32]. The enhancement of male competitiveness thus remains crucial for successful gene introduction. Gaining an understanding of the ecological factors that govern Anopheles mating biology in general, and promote male competitiveness in particular, will increase the chances of success of future GMM and sterile male-based control efforts [24-26,33].
One ecological factor known to have a great influence on the life-history of adult Anopheline, Culicines and Aedes mosquitoes is the density at which larvae develop [34-37]. In nature, larvae of An. gambiae hatch and grow in a range of aquatic habitats [38]. In the absence of predators and pathogens, the number of larvae in a particular habitat and the amount of food available to them determines the number of adults that emerge from a habitat [39,40], their survival [37,40] and body size [34,41,42]. Crowded larvae are thought to be at a disadvantage because they are faced with greater competition for food [40], and are exposed to higher levels of toxic waste products, crowding chemicals and physical interference from other larvae [43-45].
Whereas the importance of larval density to female Anopheline and Aedes mosquitoes has been broadly investigated [34,46,47], no doubt prioritized because of their direct role in disease transmission, little is known about its consequences for male mosquito vigour. Of the few known studies (in Anopheles and Aedes sp.) that have considered how larval density could influence male development [41,48,49] their focus has been on the effect of food limitation, not that of chemical or physical interference.
Here the effect of larval crowding on the mating competitiveness of adult male An. gambiae was investigated. The focus was specifically on the effects of crowding in larval habitats, not on food limitation, which was controlled for by providing each larva with an equal amount of food per unit time. Crowding was prioritized for study because, space rather than food was believed to be the biggest limiting factor when mass-producing transgenic or sterile mosquitoes for field release. In addition to conducting mating assays, the teneral reserves of males from different crowding conditions was also quantified to test if any observed differences in competitiveness could be explained by energetic limitation. Energy reserves influence mosquito behavioural activities such as swarming and feeding [50,51], and may vary in response to larval crowding. In addition to testing the effect of larval crowding on mating competitiveness, it was also examined whether it influences male longevity, as this is another potential determinant of male lifetime reproductive fitness.
Methods
Rearing
An. gambiae sensu stricto from a colony at the Ifakara Health Research and Development (IHRDC), Tanzania, were used in this study. This colony was established from a wild population located near Njage village in 1996. First instar An. gambiae s.s. larvae were obtained from colony cages and assigned randomly to density treatments of 100, 200 and 300 larvae per rearing tray (37 × 14 × 13 cm). Each tray was filled with 1 L of water and supplied with fish food (Tetramin®). In each tray, 0.2 mg of Tetramin® was added for each larva, thus 20 mg, 40 mg or 60 mg was added to the low-, medium- and high-crowding treatment trays respectively, each day. Trays were inspected visually twice a day for the presence of pupae. Once detected, pupae were collected, counted and held individually in vials to allow for emergence. Batches of males from all three larval treatments that emerged on the same day were compared against one another in mating trials using females from the low-crowding treatment.
Marking
From the time of emergence, males were pooled according to crowding treatment and held in separate cages. On the second day after emergence, cohorts of adult males from two of the three rearing conditions were marked with green or pink fluorescent dusts respectively. One group was left unmarked. Marking treatments were alternated between crowding treatments across trials to ensure no systematic bias in performance due to dusting. Furthermore, pilot studies where males from the same crowding condition were marked with different colours revealed no effect of dust presence or colour on mating performance.
Mating experiments
On the third day after emergence, 30 males (10 males from each crowding treatment) were put together in one cage (15 × 15 × 10 cm). The cage was exposed to natural light a few hours before dusk. Observation of the cage began approximately 10 minutes prior to dusk. One or two males were observed to initiate the swarming process, just above a black disc (a swarm marker) that was placed on the bottom of the cage, with most of the remaining males joining the swarm after a few minutes. Once swarming was underway, 10 females from the low crowding condition were added to the cage (making a 3:1 male to female ratio). These females were simultaneously released into the cage using an aspirator. Mating activity was observed with a low-watt red light bulb. Pairs observed to form copula were immediately aspirated out of the cage and put together into a holding cup. On each evening of experiments, observation of mating was confined to an interval of 40–45 minutes. Observation ceased when all males had stopped swarming. At the close of the swarming session, unmated females were removed from the cage. The following morning, a fluorescent lamp was used to identify the larval rearing environment of each mated male.
Females observed to have copulated with males were blood-fed on the morning following mating and moved into individual vials. Wet filter paper was placed on the bottom of these vials to act as an oviposition site. After five days in individual holding tubes, all eggs laid by mated females were collected and counted. Wing lengths of both males and females that mated, as well as a sub-sample of those from males that did not, were measured under a dissecting microscope.
Quantification of energy reserves
Batches of newly emerged males from each larval crowding regime were killed by shaking and transferred individually into glass test tubes for the quantification of lipids, sugars and glycogen. Once in tubes, mosquitoes were crushed using a glass rod. One hundred micro-litres (μl) of 2% sodium sulphate (which adsorbs glycogen) and 600 μl of a 1:2 chloroform-methanol mixture (which dissolves lipids and sugars respectively) were added to each tube. Tubes were then covered and incubated for 24 hrs at room temperature. For each batch of males that was analysed, one blank was prepared by adding the same chemicals to a tube that had no mosquito. Lipids, glycogen and sugars of each male, were then quantified using a colorimetric technique adapted for mosquito analysis [52].
Longevity of unmated males
In a separate series of experiments, males emerging from each larval rearing regime were denied access to females but held in cages to monitor their longevity. These males were provided with a 10% glucose solution for sustenance until death. All dead males were removed and counted daily.
Statistical analyses
The main aim of statistical analyses was to test for differences in the mating competitiveness, energy reserves, and longevity of An. gambiae males reared under different crowding conditions. Three analyses were conducted to assess mating competitiveness. First, analysis was restricted only to the first male to mate in each of 28 trials. The first male to mate was considered to be the fittest in the group (the first place 'winner'), and used a chi-square test to examine how larval crowding treatment influenced a male's probability of being a winner. Secondly, to test whether the total number of copulations in all nights was influenced by larval crowding treatment, a chi-square test was again used. Finally, the order in which males mated during a night (1st, 2nd, 3rd etc.) was examined whether was influenced by larval crowding treatment. For this, the analysis was restricted to data from the 14 trials (out of 28) where at least five matings occurred in a night. Males that mated were given a rank that corresponded to the order in which they mated during the trial (e.g. 1st to mate got '1', etc). A Kruskal-Wallis test was then used to test the relationship between larval crowding treatment and mating rank (dependent variable). General Linear Models (GLM) were used to test whether larval crowding treatment influenced male wing length, or the abundance of lipids, glycogen and sugars they had on emergence. GLM were also used to test whether the number of eggs laid by a female was influenced by the larval crowding condition of the male that inseminated her. Finally, Kaplan-Maier survival analysis was used to test whether the survival of males depended on the crowding condition under which they were reared. All statistical analyses were done using the SPSS for windows and SAS system for Windows (version 8).
Results
Mating competitiveness
A total of 1,120 An. gambiae mosquitoes were used in 28 nights of mating experiments (280 females and 840 males). Restricting consideration to the first male to mate, we observed that males from low crowding environments were much more likely to succeed (χ22 = 13.61, p = 0.01, Figure. 1). Males from the low crowding treatment won approximatelly11 times (Odds ratio [95% CI] = 11.17, [2.7–50]) more often than those from the high crowding treatment, while those from the medium crowding condition won approximately 4 times more often (O.R [95% CI] = 3.51, [0.9–16.7]). Analysis of all copulations (not just the first) in all 28 nights trials showed no statistically significant difference in mating frequency between males from different crowding treatments (χ22 = 4.99, p = 0.08), however there was a trend towards a higher mating frequency at low crowding condition, similar to that demonstrated in the 'first-to-mate' analysis (Figure 2). In the subset of 14 trials where at least five males mated, there was a weak tendency for males from the least crowded larval condition to mate before those from more crowded conditions, but it was not statistically significant (χ22 = 5.09, p = 0.08, Figure. 3).
Figure 1 Frequency at which males from high, medium and low crowding conditions were the 'first-to-mate' in 28 nights of mating trials. The error bars represent the standard error as estimated from the binomial distribution.
Figure 2 Proportion of total matings in 28 nights of trials going to males from low, medium and high larval crowding treatments. Error bars are the standard error as estimated from the binomial distribution (n = 133).
Figure 3 Distribution of mating ranks of males from low, medium and high crowding treatments as observed in the 14 mating trials in which at least 5 matings occurred. One circle represents two observations and the dark line in each treatment gives the median mating rank. Overlapping lines have the same mating rank, but have been spaced to indicate the number of observations per rank.
The average size of male mosquitoes did not vary significantly between larval crowding treatments (F2,397 = 2.43, p = 0.09, mean body size: 2.75 ± 0.23 mm, 2.79 ± 0.12 mm and 2.79 ± 0.12 mm for low-, medium- and higher-crowding conditions, respectively). However, of those that were measured (n = 398), males who successfully obtained a female were larger than those that did not (F1,397 = 6.97, p = 0.01, mean body size: 2.82 ± 0.02 mm and 2.76 ± 0.01 mm respectively, Figure. 4). There was no difference between the body size of males who mated first, and those who mated later in the evening (F1,115 = 1.79, p = 0.18, but both groups were larger than males who did not mate, Figure. 4).
Figure 4 Body sizes (as indexed by wing length) of males who were the first to mate, who mated but were not the first, and that did not mate at all. Bars with the same number of asterisks (*) are not statistically different, but bars with differing numbers are.
Only 15 out of 52 mated and subsequently blood-fed females oviposited their eggs. Amongst this subset, no association was found between egg batch size and paternal crowding condition (F2,12 = 0.67, p = 0.53) or maternal wing length (F1,11 = 1.98 p = 0.19). Additionally, there was no association between the probability that females would oviposit and the larval crowding condition of her mate (χ22 = 0.91, p = 0.63).
Male teneral reserves and longevity
Pooling all treatments, the mean amounts of teneral reserves in newly emerged males were 14.24 (± 1.34) μg of lipids, 1.34 (± 0.71) μg of sugars and 7.96 (± 0.39) μg of glycogen. There was no evidence that larval crowding conditions influenced the abundance of these reserves in newly emerged adult males (lipids: F2,66 = 1.36, p = 0.26, sugars: F2,66 = 2.16, p = 0.12 and glycogen: F2,66 = 2.12, p = 0.13, Figure. 5).
Figure 5 The mean mass of lipids, glycogen and sugar in newly emerged An. gambiae s. s. males reared in low, medium and high larval crowding conditions.
The survival of 132 male An. gambiae was observed (Nlow = 44, Nmedium = 37 and Nhigh = 51). The survival of adult males varied in response to the crowding conditions under which they were reared (Log-rank = 10.79, df = 2, p < 0.01, Figure 6), with the median survival of males equaling 21, 25 and 26 days for low-, medium- and high-crowding conditions, respectively. Males from low larval crowding conditions had poorer survival than those from medium (Log-rank = 7.14, df = 1, p < 0.01) and higher crowding treatment (Log-rank 8.14, df = 1, p < 0.01). The survival of males from medium and higher crowding conditions did not differ (Log-rank = 0.12, df = 1, p = 0.73).
Figure 6 Survival of adult male An. gambiae s.s., from low, medium and high larval crowding treatments.
Discussion
This study shows that larval crowding influences the mating competitiveness of male An. gambiae mosquitoes. Results from 28 replicated experiments indicates that males reared under low crowding conditions are eleven times more likely to be the first in a swarming group to obtain a female than those reared at high crowding conditions. However, when all matings were considered (not just the first in each night), there was no evidence that the frequency of copulations obtained by males varied in response to larval crowding conditions.
Thus, this study has shown that larval crowding conditions influences a male's chance of beating his competitors in order to obtain the first female, but not his chance of getting a female in general. What does this say about the role of larval crowding as a determinant of male fitness? It was proposed, that a male mosquito's ability to obtain the first available female is more likely to reflect their lifetime reproductive potential than their success in eventually getting a mate; especially under controlled laboratory conditions. There are several reasons for this hypothesis. The first is that during mating, male An. gambiae implant a mating plug in females which presents a temporary physical barrier to further insemination [53-56]. Presuming females do not leave the swarm as soon as they are mated, males who hesitate may find themselves at a greater risk of encountering unreceptive females than those who mated first. Secondly, mating in An. gambiae is thought to be confined to a 15–20 min period [57-59] around dusk. Within this period, some males have been observed to return to the swarm after they have mated and continue seeking females [57]. Those who obtain the first females that enter the swarm are more likely to have sufficient time to return to the swarm after mating to look for additional females than those who mate later in the night. In our study, males were removed from the mating arena as soon as they obtained mate, so we could not test whether earlier maters were also more likely to mate repeatedly during the evening or not. However, this possibility is worth further study. A third reason for believing that those males who mated first have the highest mating competitiveness is that in nature, males are exposed to predation risks from insect predators, such as dragonflies, while swarming [60,61]. Those who mate first can leave the swarm and escape this risk, or even if they remain in the swarm, will have had the advantage of passing on their genes before being preyed upon. If predators we could have been introduced into the laboratory experimental cages, the ones that mated first might have exhibited an additional survival advantage. The final reason for hypothesizing that males who were the first to mate in our experiments would be the most competitive in nature is that the conditions under which male Anopheles compete for females in the field are much more intense than those created here. For example, while here the ratio created experimentally was 3 males to each female, in the field, males outnumber females at the mating site, in the range of 10:1 up to 600:1 [62,63]. Under such skewed conditions where males dramatically outnumber females, it is extremely important for a male to seize a female at the earliest opportunity, as there is no guarantee another female will turn up before the end of the evening. Thus, any factor that was believed to increases a males chance of being the 'first-to-mate', as the study demonstrated with larval crowding, will be strongly correlated with their lifetime reproductive successes under natural conditions.
Several possible mechanisms that could explain the differences in mating patterns between crowding treatments were evaluated. The first was body size, which influenced the total number of males that mated, with larger males being more likely to obtain a mate than smaller ones. A similar finding was reported in Anopheles freeborni [57], whereas no size-dependency for mating was observed in An. gambiae by Charlwood [66]. Although body size influenced mating, in general it did not explain treatment-associated differences in males who were the first-to-mate. This is because there was no difference in body size between males who mated first or later and no systematic difference in body size between crowding treatments. To conclude, both body size and larval crowding can independently influences male mating success, and that the effect of the latter is not exclusively driven by variation in the former trait.
The amount of teneral reserves in males did not differ between crowding treatments, and thus could not explain this differences in mating success. Eliminating these possibilities, it was hypothesized that the observed differences in mating success between crowding conditions could be due to the detrimental effects of chemicals [43,45] and/or waste products that are released in crowded conditions, with larvae grown in dense conditions suffering more from exposure than those at low crowding.
When held at high density, some mosquito larvae release 'crowding chemicals' that retard the growth of their conspecifics [43]. This phenomenon has been recorded for Aedes aegypti, but not for Anopheles. In Aedes, chemical growth retardants are released by larvae when their density increases, even if each larva receives a constant ration of food [45]. There is, however, a certain food ration threshold above which no chemicals are produced regardless of the number of individuals [44,45]. When food rations are below this threshold, however, the release of these chemicals may regulate the number of adults that emerge [43]. Although the presence of such chemicals was not assayed here, its existence would explain why in the absence of food limitation, mosquitoes grown in highly crowded conditions performed poorer than those from low crowding. The mechanism through which such chemical factors could have influenced mating success is not clear, as it was not associated with between-treatment variation in body size or teneral reserves. Thus, it was assumed that exposure to these chemical factors may have led to subtle differences in size, behaviour or physiology not detected here (i.e. changes in male flight ability or reaction time) that ultimately influenced mating competitiveness. Further experiments are required to confirm whether such chemical factors exist in Anopheles, and how they operate.
Larval crowding also influenced the survival of An. gambiae adult males (Figure 4). Whereas males from low crowding conditions were generally the first to mate and, thus, probably the most competitive for mates, they also had the poorest survival. This observation suggests the existence of an energetic trade-off between reproduction and survival in male Anopheles, such as has been observed in other insects [64]. In male An. gambiae, such a trade-off could arise because males that are the first to mate are those that are the most active, and spend more time flying and swarming than those with lower mating success. As flying is energetically costly [50,53], an increased tendency to do so may lead to both; an enhanced mating competitiveness and reduced long-term survival, as we observed in males from the low crowding condition here. As the proportion of time that males from different crowding conditions were flying in this experiment was not observed, it remains unknown if differential activity could explain the between-group variation in mating success and survival. Further study is required to measure whether flight activity is linked to mating success, and it whether influences the rate at which a male's energy reserves and longevity decrease.
The reduced survival of males from low crowding conditions may not necessarily compromise their long-term reproductive fitness. The benefits of being the first to mate during the early part of their adult life, as discussed above, may compensate for having a reduced number of mating opportunities in the longer term due to poorer survival. If so, our findings are consistent with the theoretical claim that longevity may not be a reliable measure of male reproductive fitness [64,65]. Further experiments in which males are given multiple opportunities to mate during their natural life are required to confirm whether being the first to mate on any given evening is indeed the best predictor of male mosquito lifetime reproductive success. Ideally these experiments would be carried in larger semi-field systems [33], as well as in natural populations, so realistic costs of activity (i.e. exposure to predation, energetic drain) can be incorporated.
Conclusion
These novel findings have direct application to genetic control strategies for malaria that seek to reduce transmission by releasing sterile or malaria-refractory Anopheles males. The reported poor competitive success of transgenic male mosquitoes [27-29] could be enhanced by rearing males in conditions of low crowding and high food abundance. This could create a cohort of highly competitive yet relatively short-lived males for release. Ideally, transgenic males should be both highly competitive and long-lived. However, should an energetic trade-off exist between their competitiveness and longevity as suggested here, we argue it would be more useful to focus on increasing their short-term mating competitiveness by methods such as those discussed here.
To increase the competitiveness of mass-reared males, it is advocated: 1) to maintain males at low densities and/or regular changing of rearing water to avoid the build-up of crowding of chemicals that might result in disadvantaged males, and 2) to supply larvae with sufficient amounts of food. This finding, therefore, may help to overcome some of the mating-related hurdles that impeded early genetic control trials [24]. This proposes that, the fitness of all current genetically modified Anopheles constructs [17,19] be re-assayed after under ideal larval conditions in order to show how substantially ecological manipulation could increase their mating success relative to the wild type.
Authors' contributions
KN and BJ were directly involved in the experimental work. KN, HF and GK developed the experimental design. HF helped in logistics, advised statistical analysis of the data and supervised manuscript preparation. BGJK obtained funding for this work. GK, BGJK and GN provided comments on the manuscript prior to submission.
Acknowledgements
We would like to thank the entomology team at the IHRDC for their invaluable assistance. We thank Dr. Ana Rivero for her advice regarding energy reserve analysis, and Dr. Tom Smith and Dr. Nick Colegrave for discussion of statistical analysis. We are also thankful to Dr. Mark Benedict for helpful comments during manuscript preparation. This research is funded by a VIDI grant (no. 864.03.004) awarded by the Dutch Scientific Organisation (NWO) to Bart G.J. Knols, and carried out at the IHRDC.
==== Refs
White GB Anopheles gambiae complex and disease transmission in Africa Trans R Soc Trop Med Hyg 1974 68 279 301 10.1016/0035-9203(74)90035-2
Gillies MT DeMeillon B The Anophelinae of Africa South of the Sahara (Ethiopian zoogeographical region) 1968 54 Johannesburg: South African Institute for Medical Research
Kiszewski A Mellinger A Spielman A Malaney P Sachs SE Sachs J A global index representing the stability of malaria transmission Am J Trop Med Hyg 2004 70 486 498 15155980
Hay IS Guerra AC Tatem JA Atkinson MP Snow WR Urbanization, malaria transmission and disease burden in Africa Nature Rev 2005 3 81 90 10.1038/nrmicro1069
Snow RW Guerra CA Noor AM Myint HY Hay SI The global distribution of clinical episodes of Plasmodium falciparum malaria Nature 2005 434 214 217 15759000 10.1038/nature03342
Alten B Caglar SS Simsek FM Kaynas S Effect of insecticide-treated bednets for malaria control in Southeast Anatolia-Turkey J Vect Ecol 2003 28 97 107
Hawley WA ter Kuile FO Steketee RS Nahlen BL Terlouw DJ Gimnig JE Shi YP Vulule JM Alaii JA Hightower AW Kolczak MS Kariuki SK Phillips-Howard PA Implications of the western Kenya permethrin-treated bed net study for policy, programme implementation, and future research Am J Trop Med Hyg 2003 68 168 173 12749501
Wiseman V Hawley WA ter Kuile FO Phillips-Howard PA Vulule JM Nahlen BL Mills AJ The cost-effectiveness of permethrin-treated bed nets in an area of intense malaria transmission in western Kenya Am J Trop Med Hyg 2003 68 161 167 12749500
ter Kuile FO Terlouw DJ Phillips-Howard PA Hawley WA Friedman JF Kolczak MS Kariuki SK Shi YP Kwena AM Vulule JM Nahlen BL Impact of permethrin-treated bed nets on malaria and all-cause morbidity in young children in an area of intense perennial malaria transmission in western Kenya: cross-sectional survey Am J Trop Med Hyg 2003 68 100 107 12749492
Curtis CF Should the use of DDT be revived for malaria vector control? Biomedica 2002 22 455 461 12596442
Roberts DR Manguin S Mouchet J DDT house spraying and re-emerging malaria Lancet 2000 356 330 332 11071203 10.1016/S0140-6736(00)02516-2
Yawson AE McCall PJ Willson MD Donnelly MJ Species abundance and insecticide resistance of Anopheles gambiae in selected areas of Ghana and Burkina Faso Med Vet Entomol 2004 18 372 377 15642004 10.1111/j.0269-283X.2004.00519.x
Brooke BD Kloke G Hunt RH Koekemoer LL Temu EA Taylor ME Small G Hemingway J Coetzee M Bioassay and biochemical analyses of insecticide resistance in southern African Anopheles funestus (Diptera: Culicidae) Bull Entomol Res 2001 91 265 273 11587622
Curtis C Maxwell C Lemnge M Kilama WL Steketee RW Hawley WA Bergevin Y Campbell CC Sachs J Teklehaimanot A Ochola S Guyatt H Snow RW Scaling-up coverage with insecticide-treated nets against malaria in Africa: who should pay? Lancet Infect Dis 2003 3 304 307 12726981 10.1016/S1473-3099(03)00612-1
Lengeler C Insecticide-treated bed nets and curtains for preventing malaria Cochrane Database Syst Rev 2004 2 15106149
WHO/UNICEF The African Malaria Report 2003 The African Malaria Report 2003 120
Ito J Ghosh A Moreira LA Wimmer EA Jacobs-Lorena M Transgenic anopheline mosquitoes impaired in transmission of a malaria parasite Nature 2002 417 452 455 12024215 10.1038/417452a
Moreira AL Ito J Ghosh A Devenport M Zieler H Abraham EG Crisanti A Nolan T Catteruccia F Jacobs-Lorena M Bee venom phospholipase inhibits malaria parasite development in transgenic mosquitoes J Biol Chem 2002 277 40839 40843 12167627 10.1074/jbc.M206647200
James AA Blocking malaria parasite invasion of mosquito salivary glands J Exp Biol 2003 206 3817 3821 14506217 10.1242/jeb.00616
Benedict MQ Robinson AS The first releases of transgenic mosquitoes: an argument for the sterile insect technique Trends Parasitol 2003 19 349 355 12901936 10.1016/S1471-4922(03)00144-2
Kiszewski AE Spielman A Spatially explicit model of transposon-based genetic drive mechanisms for displacing fluctuating populations of anopheline vector mosquitoes J Med Entomol 1998 35 584 590 9701949
Mshinda H Killeen GF Mukabana WR Mathenge E Mboera LEG Knols BGJ Development of genetically modified mosquitoes in Africa Lancet Infect Dis 2004 4 264 265 15120341 10.1016/S1473-3099(04)01000-X
Tabachnick WJ Reflections on the Anopheles gambiae genome sequence, transgenic mosquitoes and the prospect for controlling malaria and other vector borne diseases J Med Entomol 2003 40 597 606 14596272
Ferguson FM John B Ng'habi K Knols BGJ Addressing the sex imbalance in knowledge of vector biology Trends Evol Ecol 2005 20 202 209 10.1016/j.tree.2005.02.003
Knols BG JrNjiru BN Mathenge EM Mukabana WR Beier JC Killeen GF Malariasphere: A greenhouse-enclosed simulation of a natural Anopheles gambiae (Diptera: Culicidae) ecosystem in Western Kenya Malar J 2002 1 19 12537599 10.1186/1475-2875-1-19
Scott TW Takken W Knols BG JrBoete C The ecology of genetically modified mosquitoes Science 2002 298 117 119 12364785 10.1126/science.298.5591.117
Catteruccia F Godfray HC Crisanti A Impact of genetic manipulation on the fitness of Anopheles stephensi mosquitoes Science 2003 299 1225 1227 12595691 10.1126/science.1081453
Irvin N Hoddle MS O'Brochta DA Carey B Atkinson PW Assessing fitness costs for transgenic Aedes aegypti expressing the GFP marker and transposase genes PNAS 2004 101 891 896 14711992 10.1073/pnas.0305511101
Moreira LA Wang J Collins FH Jacobs-Lorena M Fitness of anopheline mosquitoes expressing transgenes that inhibit Plasmodium development Genetics 2004 166 1337 1341 15082552 10.1534/genetics.166.3.1337
Boëte C Koella JC A theoretical approach to predicting the success of genetic manipulation of malaria mosquitoes in malaria control Malar J 2002 1 3 12057019 10.1186/1475-2875-1-3
Gould F Schliekelman P Population genetics of autocidal control and strain replacement Ann Rev Entomol 2004 49 193 217 14651462 10.1146/annurev.ento.49.061802.123344
Riehle MA Srinivasan P Moreira CK Jacobs-Lorena M Towards genetic manipulation of wild mosquito populations to combat malaria: advances and challenges J Exp Biol 2003 206 3809 3816 14506216 10.1242/jeb.00609
Knols BG JrNjiru BN Mukabana RW Mathenge EM Killeen GF Scott TW, Takken W Contained semi-field environments for ecological studies on transgenic African malaria vectors Ecology of transgenic mosquitoes 2003 Wageningen: Wageningen University and Research Centre 99 106
Lyimo EO Takken W Koella JC Effect of rearing temperature and larval density on larval survival, age at pupation and adult size of Anopheles gambiae Entomol Exp Appl 1992 63 265 271 10.1007/BF00192775
Koella JC Lyimo EO Variability in the relationship between weight and wing length of Anopheles gambiae J Med Entomol 1996 33 261 264 8742532
Agnew P Haussy C Michalakis Y Effects of density and larval competition on selected life history traits of Culex pipiens quinquefasciatus (Diptera: Culicidae) J Med Entomol 2000 37 732 735 11004786
Hawley W The effect of larval density on adult longevity of a mosquito, Aedes sierrensis : Epidemiological consequences J Anim Ecol 1985 54 955 964
Fillinger U Sonye G Killeen G Knols BG JrBecker N The practical importance of permanent and semipermanent habitats for controlling acquatic stages of Anopheles gambiae sensu lato mosquitoes: Operational observations from a rural town in western Kenya Trop Med Inter Health 2004 9 1274 1289 10.1111/j.1365-3156.2004.01335.x
Gimnig JE Ombok M Kamau L Hawley WA Characteristics of larval anopheline (Diptera: Culicidae) habitats in Western Kenya J Med Entomol 2001 38 282 288 11296836
Ye-Ebiyo Y Pollack RJ Kiszewski A Spielman A Enhancement of development of larval Anopheles arabiensis by proximity to flowering maize (Zea mays) in turbid water and when crowded Am J Trop Med Hyg 2003 68 748 752 12887038
Schneider JR Morrison AC Astete H Scott TW Wilson ML Adult size and distribution of Aedes aegypti (Diptera: Culicidae) associated with larval habitats in Iquitos, Peru J Med Entomol 2004 41 634 642 15311454
Gama RA Alves Kde C Martins RF Eiras AE Resende MC Effect of larvae density on adult size of Aedes aegypti reared under laboratory condictions Revis Soc Bras Med Trop 2005 38 64 66
Roberts D Overcrowding of Culex sitiens (Diptera: Culicidae) larvae: Population regulation by chemical factors or mechanical interference J Med Entomol 1998 35 665 669 9775589
Reisen WK Intraspecific competition in Anopheles stephensi Liston Mosq News 1975 35 470 480
Moore CG Whitacre DM Competition in mosquitoes. 2. Production of Aedes aegypti larval growth retardant at various densities and nutritional levels Ann Entomol Soc Am 1972 65 915 918
Mercer DR Effects of larval density on the size of Aedes polynesiensis adults (Diptera: Culicidae) J Med Entomol 1999 36 702 708 10593069
Armbruster P Hutchinson RA Pupal mass and wing length as indicators of fecundity in Aedes albopictus and Aedes geniculatus (Diptera: Culicidae) J Med Entomol 2002 39 699 704 12144308
Gomulski LM Larval density, adult size and mating competitiveness in the mosquito Anopheles gambiae Trans R Soc Trop Med Hyg 1985 79 276 277
Koenraadt CJM Takken W Cannibalism and predation among larvae of the Anopheles gambiae complex Med Vet Entomol 2003 17 61 66 12680927 10.1046/j.1365-2915.2003.00409.x
Briegel H Physiological bases of mosquito ecology J Vect Ecol 2003 28 1 11
Holliday-Hanson ML Yuval B Washino RK Energetics and sugar-feeding of field-collected anopheline females J Vect Ecol 1997 22 83 89
Van Handel E Assay of lipids, glycogen and sugars in individual mosquitoes: correlations with wing length in field collected Aedes vexans J Am Mosq contr Assoc 1988 4 549 550
Charlwood JD Jones MDR Mating behaviour in the mosquito, Anopheles gambiae s.l. Close range and contact behaviour Physiol Entomol 1979 4 111 120
Craig GB Jr Mosquitoes: Female monogamy induced by male accessory gland substance Science 1967 156 1499 1501 5611023
Bryan JH Results of consecutive matings of female Anopheles gambiae Sp B with fertile and sterile males Nature 1968 218 489 5649708
Giglioli MEC Mason GF The mating plug in anopheline mosquitoes Proc R Soc London Series A 1966 41 123 129
Yuval B Wekesa JW Washino RK Effect of body size on swarming behaviour and mating success of male Anopheles freeborni (Diptera: Culicidae) J Insect Behav 1993 6 333 342 10.1007/BF01048114
Charlwood JD Pinto J Sousa CA Madsen H Ferreira C do Rosario VE The swarming and mating behaviour of Anopheles gambiae s.s. (Diptera: Culicidae) from Sao Tome Island J Vect Ecol 2002 27 178 183
Charlwood JD Thompson R Madsen H Observations on the swarming and mating behaviour of Anopheles funestus from southern Mozambique Malar J 2003 2 2 12636875 10.1186/1475-2875-2-2
Downes JA The swarming and mating flight of Diptera Ann Rev Entomol 1969 14 271 298 10.1146/annurev.en.14.010169.001415
Subramanian KA Nature watch: When dragons fly Resonance 2002 68 78
Charlwood JD Jones MDR Mating behaviour in the mosquito, Anopheles gambiae s.l. II. Swarming behaviour Physiol Entomol 1980 5 315 320
Reisen WK Aslam Y Siddiqui TF Observation on the swarming and mating of some Pakistan mosquitoes in nature Ann Entomol Soc Am 1977 70 988 995
Hunt J Brooks R Jonnions MD Smith MJ Bentsen CL Bussiere LF High-quality male field crickets invest heavily in sexual display but die young Nature 2004 432 1024 1027 15616562 10.1038/nature03084
Hansen TF Price DK Good genes and old age: Do old mates provide superior genes? J Evol Biol 1995 8 759 778 10.1046/j.1420-9101.1995.8060759.x
Charlwood JD Pinto J Sousa CA Ferreira C Do Rosario VE Male size does not affect mating success of Anopheles gambiae in Sao Tome Med Vet Entomol 2002 16 109 111 11963975 10.1046/j.0269-283x.2002.00342.x
|
16197541
|
PMC1260028
|
CC BY
|
2021-01-04 16:24:13
|
no
|
Malar J. 2005 Sep 30; 4:49
|
utf-8
|
Malar J
| 2,005 |
10.1186/1475-2875-4-49
|
oa_comm
|
==== Front
Part Fibre ToxicolParticle and Fibre Toxicology1743-8977BioMed Central London 1743-8977-2-81620970410.1186/1743-8977-2-8ReviewPrinciples for characterizing the potential human health effects from exposure to nanomaterials: elements of a screening strategy Oberdörster Günter [email protected] Andrew [email protected] Ken [email protected] Vincent [email protected] Julie [email protected] Kevin [email protected] Janet [email protected] Barbara [email protected] Wolfgang [email protected] David [email protected] Stephen [email protected] Nancy [email protected] David [email protected] Hong [email protected] report from the ILSI Research Foundation/Risk Science Institute Nanomaterial Toxicity Screening Working Group [email protected] Department of Environmental Medicine, University of Rochester, 601 Elmwood Avenue, P.O. Box EHSC, Rochester, NY 14642, USA2 Project on Emerging Nanotechnologies, Woodrow Wilson International Center for Scholars, 1300 Pennsylvania Avenue, N.W., Washington, DC 20004-3027, USA3 MRC/University of Edinburgh Centre for Inflammation Research, ELEGI Colt Laboratory Queen's Medical Research Institute, 47 Little France Crescent, Edinburgh EH16 4TJ, UK4 Pathology and Physiology Research Branch, Health Effects Laboratory Division, National Institute for Occupational Safety and Health, 1095 Willowdale Road, Morgantown, WV 26505, USA5 Risk Science Institute, ILSI Research Foundation, International Life Sciences Institute, One Thomas Circle, N.W., Suite 900, Washington, DC 20005-5802, USA6 Center for Biological and Environmental Nanotechnology, MS-63, P.O. Box 1892, Rice University, Houston, TX 77251-1892, USA7 Respiratory/Inhalation Toxicology, Central Product Safety, Procter & Gamble Company, PO Box 538707, Cincinnati, OH 45253-8707, USA8 Office of Research and Development, United States Environmental Protection Agency, Ariel Rios Building, Mail Code: 8722F, 1200 Pennsylvania Avenue, N.W., Washington, DC 20460, USA9 Project on Emerging Nanotechnologies, Woodrow Wilson International Center for Scholars, 1300 Pennsylvania Avenue, N.W., Washington, DC 20004-3027, USA10 Institute for Inhalation Biology & Focus Network: Aerosols and Health, GSF National Research Centre for Environment and Health, Ingolstadter Landstrasse 1, 85764 Neuherberg, Munich, Germany11 Risk Assessment Division, Office of Pollution Prevention & Toxics, United States Environmental Protection Agency, 7403M, 1200 Pennsylvania Avenue, N.W., Washington, DC 20460, USA12 Center for Chemical Toxicology and Research Pharmacokinetics, College of Veterinary Medicine, North Carolina State University, 4700 Hillsborough Street, Raleigh, NC 27606, USA13 DuPont Haskell Laboratory for Health and Environmental Sciences, P.O. Box 50, 1090 Elkton Road, Newark, DE 19714-0050, USA14 Department of Chemical Engineering, University of Rochester, Gavett Hall 253, Rochester, NY 14627, USA2005 6 10 2005 2 8 8 3 10 2005 6 10 2005 Copyright © 2005 Oberdörster et al; licensee BioMed Central Ltd.2005Oberdörster 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.
The rapid proliferation of many different engineered nanomaterials (defined as materials designed and produced to have structural features with at least one dimension of 100 nanometers or less) presents a dilemma to regulators regarding hazard identification. The International Life Sciences Institute Research Foundation/Risk Science Institute convened an expert working group to develop a screening strategy for the hazard identification of engineered nanomaterials. The working group report presents the elements of a screening strategy rather than a detailed testing protocol. Based on an evaluation of the limited data currently available, the report presents a broad data gathering strategy applicable to this early stage in the development of a risk assessment process for nanomaterials. Oral, dermal, inhalation, and injection routes of exposure are included recognizing that, depending on use patterns, exposure to nanomaterials may occur by any of these routes. The three key elements of the toxicity screening strategy are: Physicochemical Characteristics, In Vitro Assays (cellular and non-cellular), and In Vivo Assays.
There is a strong likelihood that biological activity of nanoparticles will depend on physicochemical parameters not routinely considered in toxicity screening studies. Physicochemical properties that may be important in understanding the toxic effects of test materials include particle size and size distribution, agglomeration state, shape, crystal structure, chemical composition, surface area, surface chemistry, surface charge, and porosity.
In vitro techniques allow specific biological and mechanistic pathways to be isolated and tested under controlled conditions, in ways that are not feasible in in vivo tests. Tests are suggested for portal-of-entry toxicity for lungs, skin, and the mucosal membranes, and target organ toxicity for endothelium, blood, spleen, liver, nervous system, heart, and kidney. Non-cellular assessment of nanoparticle durability, protein interactions, complement activation, and pro-oxidant activity is also considered.
Tier 1 in vivo assays are proposed for pulmonary, oral, skin and injection exposures, and Tier 2 evaluations for pulmonary exposures are also proposed. Tier 1 evaluations include markers of inflammation, oxidant stress, and cell proliferation in portal-of-entry and selected remote organs and tissues. Tier 2 evaluations for pulmonary exposures could include deposition, translocation, and toxicokinetics and biopersistence studies; effects of multiple exposures; potential effects on the reproductive system, placenta, and fetus; alternative animal models; and mechanistic studies.
==== Body
1.0 Introduction
The rapid proliferation of many different engineered nanomaterials presents a dilemma to regulators regarding hazard identification. The screening strategy developed by the International Life Sciences Institute Research Foundation/Risk Science Institute (ILSI RF/RSI) Nanomaterial Toxicity Screening Working Group is an effort to make a significant contribution to the initial hazard identification process for nanomaterial risk assessment.
Engineered nanomaterials are commonly defined as materials designed and produced to have structural features with at least one dimension of 100 nanometers or less. Such materials typically possess nanostructure-dependent properties (e.g., chemical, mechanical, electrical, optical, magnetic, biological), which make them desirable for commercial or medical applications. However, these same properties potentially may lead to nanostructure-dependent biological activity that differs from and is not directly predicted by the bulk properties of the constituent chemicals and compounds. This report outlines the elements of a toxicological screening strategy for nanomaterials as the first step – i.e., hazard identification – in the risk assessment process. Both in vitro and in vivo methodologies were considered in the development of the screening strategy.
Engineered nanomaterials encompass many forms and are derived from numerous bulk substances. Nanoparticles form a basis for many engineered nanomaterials, and are currently being produced in a wide variety of types for a variety of applications; fullerenes (C60 or Bucky Balls), carbon nanotubes (CNT), metal and metal oxide particles, polymer nanoparticles and quantum dots are among the most common.
Engineered nanomaterials are presenting new opportunities to increase the performance of traditional products, and to develop unique new products. "The ability to create unusual nanostructures such as bundles, sheets, and tubes holds promise for new and powerful drug delivery systems, electronic circuits, catalysts, and light-harvesting materials." [1]
Many current efforts are predominantly focused on using relatively simple nanostructured materials such as metal oxide nanoparticles and carbon nanotubes in applications such as high performance materials, energy storage and conversion, self-cleaning surface coatings and stain-resistant textiles. Research into more complex nanomaterials is anticipated to lead to applications such as cellular-level medical diagnostics and treatment and advanced electronics. However, as nanotechnology blurs traditionally rigid boundaries between scientific disciplines, a rapid growth in unanticipated applications is to be expected over the next years and decades.
As new nanotechnology-based materials begin to emerge, it will be essential to have a framework in place within which their potential toxicity can be evaluated, particularly as indicators suggest traditional screening approaches may not be responsive to the nanostructure-related biological activity of these materials.
Several national and international organizations are currently developing standard definitions for comment terms in nanomaterial science including the International Association of Nanotechnology's Nomenclature and Terminology Subcommittee and the American National Standards Institute Nanotechnology Standards Panel (ANSI-NSP). The following key definitions are used throughout this document.
1. Nanoparticle
A particle with at least one dimension smaller than 100 nm including engineered nanoparticles, ambient ultrafine particles (UFPs) and biological nanoparticles.
2. Engineered/Manufactured Nanoparticle
A particle engineered or manufactured by humans on the nanoscale with specific physicochemical composition and structure to exploit properties and functions associated with its dimensions. Engineered nanoparticles include particles with a homogeneous composition and structure, compositionally and structurally heterogeneous particles (for instance, particles with core-shell structures) and multi-functional nanoparticles (for instance, 'smart' nanoparticles being developed for medical diagnostics and treatment).
3. Nanomaterial
A material having a physicochemical structure on a scale greater than typically atomic/ molecular dimensions but less than 100 nm (nanostructure), which exhibits physical, chemical and/or biological characteristics associated with its nanostructure.
4. Nanostructured Particle
A particle with a physicochemical structure on a scale greater than atomic/molecular dimensions but less than 100 nm, which exhibits physical, chemical and/or biological characteristics associated with its nanostructure. A nanostructured particle may be much larger than 100 nm. For example, agglomerates of TiO2 nanoparticles that are significantly larger than 100 nm in diameter may have a biological activity determined by their nanoscale sub-structure. Other examples include zeolites, meso-porous materials and multifunctional particulate probes.
5. Agglomerate/Aggregate
The terms "agglomerate" and "aggregate" are used differently and even interchangeably in different fields. In the context of this report, the term "agglomerate" is used exclusively to describe a collection of particles that are held together by both weak and strong forces, including van der Waals and electrostatic forces, and sintered bonds. In this document, the term is used interchangeably with 'aggregate'. However, the importance of understanding how the binding forces of an agglomerate affect the dispersibility of the component particles under different conditions – in essence how easily the agglomerate de-agglomerates – is noted.
6. Nanoporous Material
A material with particles that are larger than 100 nm may have significant structuring on the nanometer size scale, thereby providing properties based upon this smaller structuring that may be toxicologically relevant (e.g., dramatically increased surface area as compared to the bulk). Nanoporous materials, such as zeolites, are a significant class of materials which have porosity in the sub-100 nm size range but whose primary particles may be large.
2.0 Objectives and Scope
The objective of the ILSI RSI Nanomaterial Toxicity Screening Working Group, which was convened in February 2005, was to identify the key elements of a toxicity screening strategy for engineered nanomaterials. The group considered potential effects of exposure to nanomaterials by inhalation, dermal, oral, and injection routes; discussed how mechanisms of nanoparticle toxicity may differ from those exhibited by larger particles of the same chemical; and identified significant data needs for designing a robust screening strategy.
The elements of a screening strategy for nanomaterials presented by the Nanomaterial Toxicity Screening Working Group include an evaluation of the physicochemical characteristics and dose metrics; acellular assays; in vitro assays for lung, skin, and mucosal membranes; and in vivo assays for lung, skin, oral, and injection exposures.
This project was funded by the U.S. Environmental Protection Agency Office of Pollution Prevention and Toxics through a cooperative agreement with the ILSI Research Foundation/Risk Science Institute. It was an outgrowth of another project under the same cooperative agreement that proposed strategies for short-term toxicity testing of fibers [2]. Among the principal conclusions of the latter project were the importance of the physicochemical characterization of the fibers, the value of subchronic (1–3 month) rat inhalation exposure studies, and the typically key role in fiber toxicity of biopersistence of inhaled fibers in the lung and of chronic inflammation leading to cell proliferation and interstitial fibrosis.
3.0 Literature Survey
The potential for human and ecological toxicity associated with nanomaterials and ultrafine particles is a growing area of investigation as more nanomaterials and products are developed and brought into commercial use. To date, few nanotoxicology studies have addressed the effects of nanomaterials in a variety of organisms and environments. However, the existing research raises some concerns about the safety of nanomaterials and has led to increased interest in studying the toxicity of nanomaterials for use in risk assessment and protection of human health and the environment. A new field of nanotoxicology has been developed to investigate the possibility of harmful effects due to exposure to nanomaterials [3]. Nanotoxicology also encompasses the proper characterization of nanomaterials used in toxicity studies. Characterization has been important in differentiating between naturally occurring forms of nanomaterials, nano-scale byproducts of natural or chemical processes, and manufactured (engineered) nanomaterials. Because of the wide differences in properties among nanomaterials, each of these types of nanoparticles can elicit its own unique biological or ecological responses. As a result, different types of nanomaterials must be categorized, characterized, and studied separately, although certain concepts of nanotoxicology based on the small size, likely apply to all nanomaterials.
As materials reach the nanoscale, they often no longer display the same reactivity as the bulk compound. For example, even a traditionally inert bulk compound, such as gold, may elicit a biological response when it is introduced as a nanomaterial [4]. New approaches for testing and new ways of thinking about current materials are necessary to provide safe workplaces, products, and environments as the manufacturing of nanomaterials and products increases and, as a result, exposure to nanomaterials increases. The diverse routes of exposure, including inhalation, dermal uptake, ingestion, and injection, can present unique toxicological outcomes that vary with the physicochemical properties of the nanoparticles in question.
The earliest studies investigating the toxicity of nanoparticles focused on atmospheric exposure of humans and environmentally relevant species to heterogeneous mixtures of environmentally produced ultrafine particulate matter (having a diameter <100 nm). These studies examined pulmonary toxicity associated with particulate matter deposition in the respiratory tract of target organisms [5-15]. Epidemiological assessments of the effects of urban air pollution exposure focusing on particulate matter produced as a byproduct of combustion events, such as automobile exhaust and other sources of urban air pollution, showed a link in test populations between morbidity and mortality and the amount of particulate matter [16-19]. Some researchers have found an increased risk of childhood and adult asthma correlated to environmental exposure to ultrafine particulate matter in urban air [20-22]. However, other research does not indicate the same correlation [23-25].
Laboratory-based studies have investigated the effects of a large range of ultrafine materials through in vivo exposures using various animal models as well as cell-culture-based in vitro experiments. To date, animal studies routinely show an increase in pulmonary inflammation, oxidative stress, and distal organ involvement upon respiratory exposure to inhaled or implanted ultrafine particulate matter [7,11,26-30]. Tissue and cell culture analysis have also supported the physiological response seen in whole animal models and yielded data pointing to an increased incidence of oxidative stress, inflammatory cytokine production, and apoptosis in response to exposure to ultrafine particles [31-37]. These studies have also yielded information on gene expression and cell signaling pathways that are activated in response to exposure to a variety of ultrafine particle species ranging from carbon-based combustion products to transition metals. Polytetrafluoroethylene fumes in indoor air pollution are nano-sized particles, highly toxic to rats [38]. They elicit a severe inflammatory response at low inhaled particle mass concentrations, suggestive of an oxidative injury [39-41].
In contrast to the heterogeneous ultrafine materials produced incidentally by combustion or friction, manufactured nanomaterials can be synthesized in highly homogenous forms of desired sizes and shapes (e.g., spheres, fibers, tubes, rings, planes). Limited research on manufactured nanomaterials has investigated the interrelationship between the size, shape, and dose of a material and its biological effects, and whether a unique toxicological profile may be observed for these different properties within biological models.
Typically, the biological activity of particles increases as the particle size decreases. Smaller particles occupy less volume, resulting in a larger number of particles with a greater surface area per unit mass and increased potential for biological interaction [42-46]. Recent studies have begun to categorize the biological response elicited by various nanomaterials both in the ecosystem and in mammalian systems. Although most current research has focused on the effect of nanomaterials in mammalian systems, some recent studies have shown the potential of nanomaterials to elicit a phytotoxic response in the ecosystem. In the case of alumina nanoparticles, one of the US market leaders for nano-sized materials, 99.6% pure nanoparticles with an average particle size of 13 nm were shown to cause root growth inhibition in five plant species [46].
Toxicological studies of fibrous and tubular nanostructures have shown that at extremely high doses these materials are associated with fibrotic lung responses and result in inflammation and an increased risk of carcinogenesis. Single-walled carbon nanotubes (SWCNT) have been shown to inhibit the proliferation of kidney cells in cell culture by inducing cell apoptosis and decreasing cellular adhesive ability. In addition, they cause inflammation in the lung upon instillation [26,33,47-49]. Multi-walled carbon nanotubes (MWCNT) are persistent in the deep lung after inhalation and, once there, are able to induce both inflammatory and fibrotic reactions [47].
Dermal exposure to MWCNT has been modeled through cell culture and points to the nanoparticles' ability to localize within and initiate an irritation response in target epithelial cells [50]. Proteomic analysis conducted in human epidermal keratinocytes exposed to MWCNT showed both increased and decreased expression of many proteins relative to controls. These protein alterations suggested dysregulation of intermediate filament expression, cell cycle inhibition, altered vesicular trafficking/exocytosis and membrane scaffold protein down-regulation [50,51]. In addition, gene expression profiling was conducted on human epidermal keratinocytes exposed to SWCNT that showed a similar profile to alpha-quartz or silica. Also, genes not previously associated with these particulates before from structural protein and cytokine families were significantly expressed [52]. Dosing keratinocytes and bronchial epithelial cells in vitro with SWCNT has been shown to result in increases in markers of oxidative stress [50,53,54].
Charge properties and the ability of carbon nanoparticles to affect the integrity of the blood-brain barrier as well as exhibit chemical effects within the brain have also been studied. Nanoparticles can overcome this physical and electrostatic barrier to the brain. In addition, high concentrations of anionic nanoparticles and cationic nanoparticles are capable of disrupting the integrity of the blood-brain barrier. The brain uptake rates of anionic nanoparticles at lower concentrations were greater than those of neutral or cationic formulations at the same concentrations. This work suggests that neutral nanoparticles and low concentration anionic nanoparticles can serve as carrier molecules providing chemicals direct access to the brain and that cationic nanoparticles have an immediate toxic effect at the blood-brain barrier [55,56].
Tests with uncoated, water soluble, colloidal C60 fullerenes have shown that redox-active, lipophilic carbon nanoparticles are capable of producing oxidative damage in the brains of aquatic species [55]. The bactericidal potential of C60 fullerenes was also observed in these experiments. This property of fullerenes has possible ecological ramifications and is being explored as a potential source of new antimicrobial agents [57-59].
Oxidative stress as a common mechanism for cell damage induced by nano- and ultrafine particles is well documented; fullerenes are model compounds for producing superoxide. A wide range of nanomaterial species have been shown to create reactive oxygen species both in vivo and in vitro. Species which have been shown to induce free radical damage include the C60 fullerenes, quantum dots, and carbon nanotubes [30,60-66]. Nanoparticles of various sizes and chemical compositions are able to preferentially localize in mitochondria where they induce major structural damage and can contribute to oxidative stress [65].
Quantum dots (QDs) such as CdSe QDs have been introduced as new fluorophores for use in bioimaging. When conjugated with antibodies, they are used for immunostaining due to their bright, photostable fluorescence.
To date, there is not sufficient analysis of the toxicity of quantum dots in the literature, but some current studies point to issues of concern when these nanomaterials are introduced into biological systems. Recently published research indicates that there is a range of concentrations where quantum dots used in bioimaging have the potential to decrease cell viability, or even cause cell death, thus suggesting that further toxicological evaluation is urgently needed [67,68]. While it is well known that bulk cadmium selenide (CdSe) is cytotoxic, it has been suggested that CdSe quantum dots are cytocompatible, and safe for use in whole animal studies. This postulate is based in part on the use of protecting groups around the CdSe core of the quantum dot. These coatings have been shown to be protective, but their long-term stability has not been evaluated thoroughly. Recent studies exploring the cytotoxicity of CdSe-core quantum dots in primary hepatocytes as a liver model found that these quantum dots were acutely toxic under certain conditions. The cytotoxicity correlates with the liberation of free Cd2+ ions due to deterioration of the CdSe lattice. These data suggest that quantum dots can be rendered nontoxic initially for in vivo use when appropriately coated. However, the research also highlights the need to further explore the long-term stability of the coatings used, both in vivo and exposed to environmental conditions [69].
The range of approaches and methods used to reach conclusions regarding the effects of manufactured nanomaterials and ultrafine particles has led to different results. This inconsistency indicates a need for standardized tests in order to get comparable results in screening nanomaterials for potential adverse effects. As the field of nanotoxicology continues to grow, standard toxicology tests will aid those entering the field and allow for better comparisons and conclusions in determining the toxic effects of nanomaterials.
4.0 Elements of a screening strategy for nanomaterials
While the nanostructure-dependent properties of many engineered nanomaterials may place them in the category of potential hazards, the direct risk they present to human health will depend on the probability of exposures occurring, and the extent to which materials entering the body exhibit behavior associated with their nanostructure. Figure 1[70]; Biokinetics of Nano-sized Particles; While many uptake and translocation routes have been demonstrated, others still are hypothetical and need to be investigated. Largely unknown are translocation rates as well as accumulation and retention in critical target sites and their underlying mechanisms. These as well as potential adverse effects will be largely dependent on physicochemical characteristics of the surface and core of nano-sized particles. Both qualitative and quantitative changes in nano-sized particles biokinetics in a disease or compromised organism need also to be considered.
Figure 1 Biokinetics of Nano-sized Particles. While many uptake and translocation routes have been demonstrated, others still are hypothetical and need to be investigated. Largely unknown are translocation rates as well as accumulation and retention in critical target sites and their underlying mechanisms. These as well as potential adverse effects will be largely dependent on physicochemical characteristics of the surface and core of nano-sized particles. Both qualitative and quantitative changes in nano-sized particles' biokinetics in a diseased or compromised organism need also to be considered. Reproduced with permission from Environmental Health Perspectives.
In many cases, nanostructured materials will be components of large-scale products such as nano-composites, surface coatings and electronic circuits, and the potential for direct exposure will be negligible. However, if nanostructured materials may enter the body, toxicity screening strategies are required to ascertain the potential risk they present.
Nanoparticles are an obvious form of engineered nanomaterial presenting a significant exposure potential, because they can be readily deposited in the lungs or on the skin, and potentially translocate within the body. However, agglomerates of nanoparticles from a few hundred nanometers to a few micrometers in diameter may also be inhaled, ingested or deposited on the skin, and may have the potential to express toxicity associated with their nanostructure. Similarly, it is conceivable that nanostructured particles of a few micrometers in diameter and below (such as fragments of a nano-composite or a nanostructured surface coating) may exhibit nanostructure-dependent biological properties. In each of these cases, exposure potential exists for materials in air and in liquid suspensions or slurries.
In this section, three key aspects of toxicity screening strategies are addressed: characterization of nanomaterials, in vitro screening strategies and in vivo screening strategies (covering inhalation, dermal, ingestion, and injection exposure routes). Screening strategies are developed around nanoparticles, but are relevant to all engineered nanomaterials that are capable of entering the body through inhalation, ingestion, dermal penetration, or injection and expressing biological activity which is associated with their nanostructure.
4.1 Physicochemical Characterization
4.1.1 Introduction
Unlike gases, liquids and many solid materials, the desirable properties of engineered nanomaterials closely depend on size, shape and structure (both physically and chemically) at the nanoscale. Similarly, there is a strong likelihood that biological activity will depend on physicochemical parameters not usually considered in toxicity screening studies. Although quantitative toxicity studies on engineered nanomaterials are still relatively sparse, published data on fullerenes, single walled carbon nanotubes, nanoscale metal oxides such as TiO2 and nanometer-diameter low solubility particles, support the need to carefully consider how nanomaterials are characterized when evaluating potential biological activity [62,71-75]. Respirable fibers present perhaps the closest analogy to a material that is not fully characterized by mass and chemical composition alone. However, the diversity and complexity of nanomaterials suggests that the level of characterization appropriate to toxicity screening tests will be commensurately more sophisticated.
Until the mechanistic associations between nanomaterial characteristics and toxicity are more fully understood, it will be necessary to ensure that all nanomaterial characteristics that are potentially significant are measured or can be derived in toxicity screening tests. In particular, in as far as it is possible; it is desirable to collect sufficient information to allow retrospective interpretation of toxicity data in the light of new findings. In this context, identifying a set of characterization criteria for nanomaterial toxicity screening presents a significant challenge. Clearly, the ideal of characterizing every possible aspect of a test material, while laudable, is impractical. In this document, we have therefore focused on the context under which characterization takes place and the minimum set of characterization parameters we consider essential within that context. Essential parameters have been supplemented with those considered desirable and those considered of interest but optional within a screening study. The two overarching characterization contexts discussed are human exposure studies and in vitro/in vivo studies. In the case of the latter, we consider material characterization after administration, characterization at the point of administration and characterization of the bulk material as produced or supplied. The relative importance of characterizing dose against different physical metrics during inhalation exposures is also discussed. Recommendations are subsequently made on physicochemical characterizations for nanomaterial toxicity screening tests and characterization methods capable of providing the recommended information.
4.1.2 Framework for Material Characterization
Material characterization for toxicity screening studies is most appropriately considered in the context of the studies being undertaken. Requirements for in vitro and in vivo screening studies will differ according to the material delivery route or method. Additionally, understanding human exposures in the context of developing appropriate screening studies will present a further set of characterization requirements. Four screening study contexts are proposed, and characterization recommendations are developed within these contexts:
• Human exposure characterization
• Characterization of material following administration
• Characterization of administered material
• Characterization of as-produced or supplied material
Human Exposure Characterization
Where exposure to a specific material is known to occur or is anticipated, exposure studies are desirable in developing and selecting appropriate toxicity screening tests. At present engineered nanomaterials are predominantly at the research or pre-production stage, and there are relatively few environments where exposures are known to occur. However, if commercialization of products using nanomaterials develops as anticipated, the potential for exposure is likely to increase dramatically over the coming decade. Therefore, estimates of future use and potential human exposures should be considered in the development of toxicity screening.
Nanomaterial Characterization after Administration
Characterizing delivered nanomaterial after administration in a test system or model provides the highest quality of data on dose and material properties that are related to observed responses, but this is limited by current methodological capabilities. Characterization after administration is particularly advantageous where the possibility of physicochemical changes in the material before and after administration exists. Examples of potential changes include aggregation state, physisorption or chemisorption of biomolecules and biochemically-induced changes in surface chemistry. In addition, possible physicochemical changes as a result of nanomaterial interactions with the surrounding biological systems such as rapid dissolution of water- or lipid soluble fractions of the nanomaterial need to be carefully considered. While characterization after administration is considered an ideal to work towards, it is recognized that in many cases, characterization at the point of administration will be a more realistic and feasible option. It is also recognized that in many cases, characterization at the point of administration will be essential for the intercomparison of studies, irrespective of whether characterization after administration is carried out.
Characterization of Administered Material
Characterization of administered material in toxicity screening studies is fundamental. This approach addresses potential physicochemical changes between the bulk material and the administered material (such as agglomeration state) and allows more robust causal associations between the material and observed responses to be developed. However, given the strong sensitivity of many nanomaterial properties to their local environment, it should be noted that biologically relevant changes in the physicochemical nature of a nanomaterial between administration and deposition may have a significant impact on observed responses in some instances.
Characterization of As-Produced or Supplied Material
Characterization of nanomaterials as-produced or as-supplied represents the most direct approach to obtaining physicochemical information and may provide useful baseline data on the material under test. Most engineered nanomaterials have a functionality based on their physicochemistry. It is therefore likely that information of relevance to toxicity screening studies will be available from suppliers or producers in many cases. However, due to the current lack of accepted nanomaterial characterization standards, it is strongly recommended that wherever possible, independent characterization of test nanomaterials be conducted.
Characterization of supplied nanomaterial may not appropriately represent physicochemical properties of the material during or following administration. For this reason exclusive reliance on this approach is discouraged, and is only recommended where characterization of material during or after administration is clearly not feasible.
4.1.3 Key Characteristics
Previous studies of asbestos and other fibers have shown that the dimension, durability and dose (the three D's) of fibrous particles are key parameters with respect to their pathogenicity. In general, fibers with a smaller diameter will penetrate deeper in the lungs. Long fibers (longer than the diameter of alveolar macrophages) stimulate macrophages to release inflammatory mediators and will only be cleared slowly. In addition to fiber length, chemical factors play an important role in fiber durability and biopersistence; fibers with high alkali or alkali earth oxide contents and low contents of Al2O3, Fe2O3, TiO2 tend to have low durability and hence low biopersistence [76]. On the other hand, studies of mineral particles have demonstrated that the toxic and carcinogenic effects are, in some cases, related to the surface area of inhaled particles and their surface activity [77,78]. Particle surface characteristics are considered to be key factors in the generation of free radicals and reactive oxygen species formation and in the development of fibrosis and cancer by quartz (crystallized silica) [77].
The unusual properties of nanomaterials are predominantly associated with their nanometer-scale structure, size and structure-dependent electronic configurations and an extremely large surface-to-volume ratio relative to bulk materials. Particles in the nanosize range can deposit in all regions of the respiratory tract including the distal lungs. Due to their small size, nanoparticles may pass into cells directly through the cell membrane or penetrate between or through cells and translocate to other parts of the body. Limited data have suggested possible translocation of inhaled nanoparticles to the nervous system and other organs/tissues [79-81].
The size of nanoparticles alone may not be the critical factor determining their toxicity; the overall number and thus the total surface area may also be important. As a particle decreases in size, the surface area increases (per unit mass only; if you normalize to number of particles, the surface area decreases) and a greater proportion of atoms/molecules are found at the surface compared to those inside. Thus, nanoparticles have a much larger surface area per unit mass compared with larger particles. The increase in the surface-to-volume ratio results in the increase of the particle surface energy which may render them more biologically reactive.
Nano-scale materials are known to have various shapes and structures such as spheres, needles, tubes, plates, etc. Nanoporous materials are materials with defined pore-sizes in the nanometer range. The effects of the shape on the toxicity of nanomaterials are unknown. The shape of nanomaterials may have effects on the kinetics of deposition and absorption in the body. The results of a recent in vitro cytotoxicity study appear to suggest that single-wall nanotubes are more toxic than multi-wall nanotubes [82].
Chemical composition is another important parameter for the characterization of nanomaterials, which comprise nearly all substance classes, e.g., metal/metal oxides, compounds, polymers as well as biomolecules. Some nanomaterials can also be a combination of the above components in core-shell or other complex structures. Dependent on the particle surface chemistry, reactive groups on a particle surface will certainly modify the biological effects. Under ambient conditions, some nanoparticles can form aggregates or agglomerates. These agglomerates have various forms, from dendritic structure to chain or spherical structures. To maintain the characteristics of nanoparticles, they are often stabilized with coatings or derivative surface to prevent agglomeration. The properties of nanoparticles can be significantly altered by surface modification and the distribution of nanoparticles in the body strongly depends upon the surface characteristics. Changes of surface properties by coating of nanoparticles to prevent aggregation or agglomeration with different types and concentrations of surfactants have been shown to change their body distribution and the effects on the biological systems significantly [83,84].
Therefore, it is recommended that the following physicochemical properties of the test materials should be characterized:
• Size distribution
• Agglomeration state
• Shape
• Crystal structure
• Chemical composition – including spatially averaged (bulk) and spatially resolved heterogeneous composition
• Surface area
• Surface chemistry
• Surface charge
• Porosity
4.1.4 Dose Metrics
In any toxicity screening study, careful consideration should be given to the metric used to quantify dose. Although response may be associated with a wide range of physicochemical characteristics, measuring dose against a physical metric of mass, surface area or particle number for a well-characterized material will enable quantitative interpretation of data. Appropriate selection of the dose metric will depend on the hypothesized parameter most closely associated with anticipated response or the metric which may be most accurately measured. It is strongly recommended that in all cases, sufficient information is collected to enable dose against all three primary physical metrics to be derived. This may be achieved where the relationships between nanomaterial mass, surface area and particle number concentration are known, or where measurements of particle size distribution are made that enable derivation of all three dose metrics. Where nanomaterials are administered in a liquid medium, such as in the technique of intratracheal instillation or pharyngeal aspiration, the nature and amount of material within the suspension should be fully characterized before delivery in terms of number, surface area and mass concentration. Inhalation studies present additional challenges of measuring dose over time, and require both on-line (time resolved) and off-line analysis.
Off-line mass concentration measurements using filter-based methods offer continuity with standard inhalation studies and are recommended as an essential component of inhalation nanomaterial screening tests. Likewise, on-line mass concentration measurements are recommended as an essential component of inhalation studies. Gravimetric and/or chemical analysis of filter samples will provide the most accurate characterization of exposure in many cases when compared to off-line surface area and number concentration analyses. With appropriate additional information, such measurements may be used to calculate aerosol surface area or number concentration. However, the diameter cubed relationship between particle size and mass can lead to large errors when transforming from mass to number concentration if the size distribution is broad or there are small numbers of excessively large particles present. On-line mass-concentration measurements using instruments such as the Tapered Element Oscillating Microbalance (TEOM®) potentially offer high precision and good accuracy [85], although they are susceptible to errors where the sampled aerosol contains volatile components. On-line photometric mass concentration methods are generally good for monitoring the temporal stability of aerosol and providing a real-time indication of mass exposure, although they are relatively insensitive to particles smaller than approximately 0.5 μm in diameter [86]. However, in general more appropriate methods should be used for providing real-time measurements of number and surface-area exposure [85,87-89].
Aerosol size distribution measurements enable reasonably good calculation of exposure against all three physical metrics, if parameters such as particle shape and density are known. Off-line size distribution measurement methods such as Transmission Electron Microscopy (TEM) analysis offer detailed information on this distribution but are extremely time consuming, and frequently limited by the collection techniques and, in the case of TEM analysis, inference of 3-dimensional structure from 2-dimensional images. On-line size measurement techniques such as Differential Mobility Analysis [90] are capable of measuring aerosol size distribution with a time resolution of tens of seconds. Aerosol number concentration between given particle diameters is easily derived from aerosol size distribution measurements, although interpretation of such data in terms of mass or surface-area dose requires additional information on particle characteristics such as shape and density. It is recommended that for each nanoparticle type, size distribution measurement techniques be validated against TEM analysis.
Off-line surface area characterization is possible using isothermal gas-adsorption, although techniques suited to filter samples need to be employed. There is also some possibility that the surface area of the collected material will differ from that of the airborne material due to compaction and surface occlusion. However, the extent to which this may occur is not well understood. Published studies have shown a good correlation between off-line surface area measurement and biological response [91], suggesting that errors associated with collection and subsequent analysis can be small. This holds particularly for insoluble particles; ideally surface area measurements would be required on the insoluble core of a nanomaterial after its water-and/or lipid soluble compounds have been dissolved from the particle surface. Aerosol diffusion charging has been shown to provide a measure of surface area on-line where the charging rate is low [89], and a small number of aerosol diffusion chargers are commercially available. These devices have been shown to measure aerosol surface area well for particles smaller than 100 nm in diameter [87]. At larger diameters, measured surface area progressively underestimates aerosol surface area. In particular, the surface area of porous particle structures as well as that of highly aggregated particles will generally not be determined. Data have been published on a particular aerosol diffusion charger indicating that it provides a measure of aerosol surface area dose in the lungs, as opposed to aerosol surface area exposure [92]. While on-line aerosol surface area measurements are desirable during inhalation exposure studies, uncertainties associated with current techniques suggest caution when interpreting such measurements.
Number concentration may be measured on-line with relative ease using instruments such as Condensation Particle Counters [88]. Although it is not clear how biologically relevant number concentration is as a dose metric, the ease with which such measurements are made and their value in tracking temporal changes in exposure lead to their being recommended as essential in inhalation studies.
Table 1 summarizes recommendations for measuring exposure during inhalation studies.
Table 1 Recommendations for measuring exposure during inhalation studies
Metric Measurement Recommendation
Mass – off-line E (coupled with on-line)
Mass – on-line E
Size distribution – off line E
Size distribution – on line E/D
Surface area – off line O
Surface area – on line O
Number – off line N
Number – on line E
E: These measurements are considered to be essential.
D: These measurements are considered to provide valuable information, but are not recommended as essential due to constraints associated with complexity, cost and availability.
O: These measurements are considered to provide valuable but non-essential exposure information.
N: These measurements are not considered to be of significant value to inhalation studies.
4.1.5 Characterization Prioritization
In developing recommendations on material characterizations for nanomaterial toxicity screening studies, three specific factors have been taken into consideration: the context within which a material is being evaluated, the importance of measuring a specific parameter within that context, and the feasibility of measuring the parameter within a specific context. Recommendations on off-line material characterizations for nanomaterial toxicity screening studies are presented in Table 2.
Table 2 Recommendations on material characterization
Characterization (Off-line) Human exposure Toxicity Screening Studies
Supplied material Administered material Material in vivo/in vitro
Size distribution (primary particles) E (Combine with agglomeration state) E D D
Shape E E O O
Surface area D E D O
Composition E E O O
Surface chemistry D E D D/O
Surface contamination D N D N
Surface charge – suspension/solution O E E O
Surface charge – powder (use bio fluid surrogate) O E N O
Crystal structure O E O O
Particle physicochemical structure E E D D
Agglomeration state E N E D
Porosity D D N N
Method of production E E -- --
Preparation process -- -- E --
Heterogeneity D E E D
Prior storage of material E E E --
Concentration E -- E D
E: These characterizations are considered to be essential.
D: These characterizations are considered to provide valuable information, but are not recommended as essential due to constraints associated with complexity, cost and availability.
O: These characterizations are considered to provide valuable but non-essential information.
N: These characterizations are not considered to be of significant value to screening studies.
In addition, recommendations have been made on recording information on nanomaterial production, preparation, storage, heterogeneity, and agglomeration state. To enable retrospective interpretation of toxicity data and replication of tests, it is strongly recommended that all information on the production and processing of nanomaterials be recorded. Fully documenting storage time and conditions (including temperature, humidity, exposure to light and atmosphere composition) is essential, as physicochemical changes may take place over time. If possible, the physicochemical stability of samples over time should be demonstrated. Where a test material is a heterogeneous mixture of different components, information is required on the relative abundance of the different components, and whether associations in the bulk material are maintained in the administered material, or whether different components are preferentially administered with specific delivery mechanisms.
The agglomeration state of a nanomaterial during and following administration may have a significant impact on its biological activity. Agglomeration state at different structure scales should be characterized, including primary (primary particles), secondary (primary particle agglomerates and self-assembled structures) and tertiary (assemblies of secondary structures) scales. Ideally, agglomeration state in the biological environment following administration should be evaluated. If possible, some insight into the binding forces within agglomerates (e.g. relatively weak van der Waals forces or relatively strong sintered bonds) should be obtained. Material agglomeration or de-agglomeration in different liquid media should also be investigated where possible.
Characterization of material as administered is recommended as the highest priority, supplemented by characterization after in vitro or in vivo administration where possible, and followed in order of preference by characterization of the material as produced or supplied. Recommended characterizations in Table 2 reflect both this hierarchy and the feasibility of making measurements within the respective contexts.
4.1.6 Analysis Methods
Many analytical techniques, both established and developmental, are available for characterizing the nanomaterial properties listed in Table 2. Table 3 lists some of the more widely available techniques and relates them to the nanomaterial characteristics of interest to toxicity screening studies. Techniques have been categorized with respect to their applicability to specific material characteristics. In general, the table is self-explanatory, and further information on each technique can be obtained from a wide range of sources. A number of techniques are only suitable for materials in certain forms, or specific classes of materials. For instance, while Transmission Electron Microscopy is capable of providing a wealth of information on nanoparticles and is considered a gold standard for evaluating particle size distribution and shape, dry (or in the case of cryo-TEM, frozen liquid-encapsulated) well-dispersed samples that are sufficiently robust to withstand high vacuums are required. Similarly, techniques such as Infrared (IR) spectroscopy are particularly sensitive to surface organic compounds, but are less useful for quantifying inorganic surface chemistry. In a number of cases, a complex technique such as TEM can be used to validate a characterization method that is more practical to use on a routine basis.
Table 3 Applicability of a range of analytical techniques to providing specific physicochemical information on engineered nanomaterials, in the context of toxicity screening studies
Analytical technique
Transmission Electron Microscopy (TEM) Scanning Electron Microscopy (SEM) X-Ray Diffraction (XRD) X-ray Photon Spectroscopy (XPS) Auger Spectroscopy (AES) Secondary Ion Mass Spectrometry (SIMS) Scanning Probe Microscopy Dynamic Light Scattering (DLS) Zeta potential Size Exclusion Chromatography Analytical Ultracentrifugation Differential Mobility Analysis (DMA) Isothermal Adsorption (e.g. BET) Spectroscopic techniques (UV vis, IR, Raman, NMR) Elemental analysis (eg ICP-MS/AA etc)
Physicochemical Characteristic Size distribution (primary particles) ▲ ● ● ● ● ● ● ●
Shape ▲ ● ●
Surface area ● ◇ ◇ ◇ ◇ ●
Composition ● ● ● ▲ ● ▲
Surface chemistry ● ● ● ◇
Surface contamination ● ●
Surface charge – suspension/solution ▲
Surface charge – powder (use bio fluid surrogate) ▲
Crystal structure ● ◇ ▲
Particle physicochemical structure ▲ ●
Agglomeration state ▲ ● ● ●
Porosity ◇ ● ●
Heterogeneity ▲ ● ◇
Other applicable techniques are available that have not been listed.
▲Highly applicable
● Capable of providing information in some cases
Capable of providing information in some cases, with validation from more accurate/applicable techniques
◇ Capable of providing qualitative or semi-quantitative information
Given the wide range of analytical techniques available in many disciplines associated with nanotechnology, multidisciplinary collaborations with research and analysis groups offering state of the art nanomaterial characterization capabilities are strongly recommended when carrying out nanomaterial toxicity screening studies.
4.1.7 Research Gaps
1. The development of viable in vivo nanomaterial (including nanoparticles) detection techniques.
2. The development and production of inexpensive real-time monitoring instruments and methods for aerosol mass concentration (low concentrations, nanoscale particles), surface area concentration and size distribution.
3. The development of standardized, well characterized nanomaterial samples.
4. The development of radio-labeled nanomaterial samples, and samples that can be tracked and detected through neutron-activation.
5. The development of more advanced surface chemistry characterization techniques, in particular techniques capable of detecting and speciating biological molecules on the surface of nanoparticles and nanomaterials.
6. The development of electron microscopy techniques for biologically-relevant nanoscale analysis.
4.1.8 Recommendations
1. All nanomaterial physicochemical characteristics that are potentially significant should be measured or be derivable in toxicity screening tests.
2. Characterization of nanomaterial as administered is strongly recommended, supplemented by characterization following administration where it is technically feasible and practicable. Characterization of the bulk material as-produced or supplied to the exclusion of the above is not recommended, except where more appropriate measurements are not feasible.
3. It is recommended that independent characterizations of nanomaterials (beyond information provided by producers and suppliers) are carried out where possible.
4. It is recommended that the following physicochemical properties of nanomaterials should be characterized in the context of toxicity screening tests: Particle size distribution, agglomeration state, particle shape, crystal structure, chemical composition (bulk and spatial), surface area, surface chemistry, surface charge, and porosity.
5. It is recommended that in all cases, sufficient information be collected to enable derivation of the delivered dose against all three primary physical metrics (number, surface area and mass concentration).
6. Off-line mass concentration measurements using filter-based methods are recommended as an essential component of inhalation nanomaterial screening tests. In addition, off-line measurement of aerosol size distribution is recommended.
7. On-line mass concentration and number measurements are recommended as an essential component of inhalation studies.
8. Multidisciplinary collaborations between research and analysis groups offering state of the art nanomaterial characterization capabilities are strongly recommended.
9. It is recommended that information on nanomaterial production, preparation, storage, heterogeneity and agglomeration state be recorded for all nanomaterial toxicity screening studies.
10. It is recommended that nanomaterial preparation methods are fully documented, including the selection of appropriate dispersion media, methods of dispersion in the medium and agglomeration state within the medium. Specific preparation techniques are not recommended, as these will depend on the material and test protocols being used. However, caution is advised when using ultrasonic agitation to disperse materials, as at high energies the method may be sufficiently aggressive to alter the material characteristics (see section 4.3.1.1).
4.2 In vitro Testing Methods
4.2.1 Introduction
Before considering the application of specific in vitro testing methods to the assessment of the toxicity of nanomaterials, there are several generic issues that should be noted.
1) Advantages and disadvantages In general in vitro techniques are seen as an important adjunct to in vivo studies. These studies allow specific biological pathways to be tested under controlled conditions, as well as isolation of pathways that is not feasible in vivo; e.g., it is difficult to discriminate in vivo whether complement activation has a role in any pro-inflammatory effects of particles. The complement system can be isolated in vitro, and its potential role investigated. There are, of course, well-documented problems with in vitro approaches, including lack of validation against in vivo adverse effects, dosimetry mismatch, over-simplicity, non-involvement of the complete inflammatory response, etc.
2) Control particles It is important, in view of the above, that adequate positive and negative control particles are included in all experiments. This at least allows the test particle to be bench-marked against particles of known toxicity. These can include standard crystalline silica (quartz; e.g, Min-U-Sil or DQ12) as a known cytotoxic particle and fine TiO2 as an inert particle.
3) Expression of dose Toxicity and other responses should be expressed in relation to a range of dose metrics depending on the material and the dose metric data that are available (see Section 4.1).
4) Adsorption of proteins by nanoparticles The large surface area of nanoparticles means that they are capable of adsorbing proteins. Nanoparticles of various types have been reported to adsorb key proteins such as albumin [93], fibronectin and TGF-β [94]. This may confound endpoints that rely on the measurement of a protein as the protein may be produced but may also remove from the supernatant onto the nanoparticle surface by adsorption, providing a false-negative.
The in vitro tests that are presented will be divided into portal of entry toxicity and target organ toxicity. The potential target cells and associated appropriate endpoints will be described. Finally, research gaps and recommendations will be identified.
4.2.2 Portals of Entry
4.2.2.1 Lungs
The lungs represent a potential target for any airborne particles, and many in vitro models for the lung exist. Particles deposit on the airway or alveolar epithelium and encounter mucus or epithelial lining fluid. They may then interact with macrophages, which may result in their clearance, or they may enter the interstitium where they may make contact with fibroblasts and endothelial cells or cells of the immune system.
The Epithelium
The epithelium is the first barrier that confronts particles that deposit in either the conducting airways or the alveolar region. Therefore, both bronchial and alveolar epithelial cells should be considered as target cells for in vitro studies. Endpoints for detecting nanoparticle effects could include toxicity measurements, such as LDH release, for necrosis or various cytokine expression (IL-8, MCP-1 etc), [91,95] and activation of inflammation-related transcription factors such as NF-κB and AP-1[96,97]. Oxidative and nitrosative stress are dominant mechanistic hypotheses for cell damage and activation caused by pathogenic particles. These can be monitored by measuring oxidative stress using dichlorofluorescein [98] or oxidized glutathione as endpoints [99] and nitrosated proteins as a measure of active nitrogen species [100]. Responses to particle-induced oxidative/nitrosative stress can include up-regulation of anti-oxidant genes [101] such as superoxide dismutase and glutathione peroxidase, and so these can also be measured. Proliferative effects of nanoparticles can be assessed using a variety of assays including bromo-deoxyuridine incorporation [102].
If cancer is an endpoint that is under consideration, then direct measures of genotoxicity can be quantified by methods that include COMET assay and 8-hydroxy-deoxyguanosine measurement [103,104]. The translocation of nanoparticles across the epithelium could be an important discriminator of harmfulness and, although there are few publications specifically addressing transfer of particles across the epithelium in vitro, these should be developed and could contribute to understanding the factors that regulate translocation.
Macrophages
Macrophages play a key role in the cellular response to particles that deposit in the lungs. Macrophages could be affected by nanoparticles in various ways that can be studied in vitro through a variety of assays. Cellular cytotoxicity could be measured using conventional methods, such as lactate dehydrogenase release. Macrophage activation occurs following phagocytosis of a number of pathogenic particles leading to release of cytokines (tumour necrosis factor alpha (TNFα), interleukin-6 (IL-6) etc) and nuclear transfer of inflammation-related transcription factors nuclear factor kappa B (NF-κB) and activator protein 1(AP-1). Macrophages undergo an oxidative burst (OB) on phagocytosis of particles [105] and the extent of this in response to nanoparticles could be investigated. Nitric oxide (NO) may also be produced, in response to particles [106] and in the presence of superoxide radical peroxynitrite, a highly toxic species, can be produced [107]. If the OB or NO production is exaggerated, there could be 'bystander' injury to epithelial cells whilst diminished OB/NO production could mean impaired microbicidal activity that allows infection. Another key macrophage function reported to be impaired by nanoparticles is phagocytosis, [108] and so the effect of test nanoparticles on this function could be considered. The cytoskeleton is key to normal cell functioning and could be targeted by nanoparticles and so could be investigated.
Endothelial cells
Although these are found in the lungs, they are considered a part of the cardiovascular system and are dealt with below.
Fibroblasts
Fibroblasts are found in the interstitium and are liable to be affected by any particle that gains access to this site. At least two important modes of response could be activated by nanoparticle/fibroblast interactions and both modes constitute relevant endpoints for in vitro testing: 1) Pro-inflammatory effects, measured by cytokine/chemokine gene expression (TNFα; etc); or 2) fibrogenic responses activated either by direct stimulation of fibroblast growth or extra-cellular matrix secretion by the nanoparticle, or by autocrine stimulation following nanoparticle-stimulated release from the fibroblasts of growth factors such as transforming growth factor beta and platelet-derived growth factor.
The Immune System
Immunopathological effects could be envisaged if particles interact with lymphocytes, or as a consequence of their predilection for entering the interstitium, they modulate dendritic cell function. The effects of nanoparticles on immunological functions including antigen presentation by macrophages and dendritic cells and the subsequent effects on immune responses in vitro are relevant endpoints and appropriate tests should be designed.
Co-Cultures
In addition to monocultures of lung cells, co-cultures such as epithelial cells/macrophages or epithelial cells/endothelial cells may more closely represent the in vivo situation, and so such studies are encouraged.
Lung Slices
Methodology to culture whole lung tissue slices is available, such that multiple pulmonary cell types can be exposed in vitro in the same configuration as they occur in vivo.
Cell Lines vs. Freshly-Derived Cells
If possible, freshly-derived primary cells should be used. Where cell lines are used, these should preferably not be cancer cells. Where cancer cells are used, the endpoint response under study should be carefully compared to non-cancer cells to ensure that, for that endpoint, the fact that the cell is a cancer cell does not greatly modify the response compared to a non-cancer cell.
Whole Heart-Lung Preparation
The Langendorff heart-lung preparation may provide the opportunity to study the behavior of nanoparticles under highly controlled conditions. In this model the exsanguinated heart and lungs are maintained by perfusion and so transport between the lungs and the vascular space can be studied in the absence of blood [109].
4.2.2.2 Skin
Skin or the integument is the largest organ of the body and is unique because it is a potential route for exposure to nanoparticles during their manufacture and also provides an environment within the avascular epidermis where particles could potentially lodge and not be susceptible to removal by phagocytosis [110]. What are the toxicological consequences of "dirty" nanoparticles (catalyst residue) becoming lodged in the epidermis? In fact, it is this relative biological isolation in the lipid domains of the epidermis that has allowed for the delivery of drugs to the skin using lipid nanoparticles and liposomes. Larger particles of zinc and titanium oxide used in topical skin-care products have been shown to be able to penetrate the stratum corneum barrier of rabbit skin with highest absorption occurring from water and oily vehicles [111]. This could also apply to manufactured nanoparticles. Can nanoparticles gain access to the epidermis after topical exposure, the first step in a toxicological reaction? Exposure to metallic nanoparticles, whose physical properties would allow them to catalyze a number of biomolecular interactions, potentially could produce adverse toxicological effects. More information is required regarding the efficiency of decontamination of nanoparticles from skin since solubilization and dilution, the two hallmarks of post-exposure decontamination, might be less efficacious for these solid structures.
Research should address the effects of dermal and systemic exposure to a number of types of nanoparticles in the skin. The skin is a primary route of potential exposure to toxicants, including novel nanoparticles. However, there is no information on whether particles are absorbed across the stratum corneum barrier or whether systemically administered particles can accumulate in dermal tissue. Nanoparticles may traverse through the stratum corneum layers at varying rates due to particle size or become sequestered within the epidermis to increase their exposure time to viable epidermal keratinocytes.
Nanomaterials are difficult to obtain in large quantities; therefore, it is best to conduct in vitro tests to estimate in vivo starting doses for toxicity testing [112]. At least three or four concentrations with controls should be used in all in vitro systems. These data would provide a preliminary, but relevant, assessment of both systemic exposure after topical administration as well as cutaneous hazard after both topical and systemic exposure, two essential components of any risk assessment.
Cell Culture
Human epidermal keratinocyte (HEK) monolayers can be affected by nanoparticle interactions. It has already been shown that changes in biomarkers of viability and toxicity can occur with exposure to multi-wall carbon nanotubes [50]. Cytotoxicity endpoints should be evaluated: 1) cell viability-metabolic markers such as mitochondrial reduction of tetrazolium salts into insoluble dye (MTT), 2) decreased cell viability-membrane markers like neutral red uptake into cell lysosomes, trypan blue exclusion and cell attachment/cell detachment, and 3) pro-inflammatory cytokine affects measured by TNFα, IL-8, IL-6, IL-10, or IL-1β. Genomics and proteomics assays could be used to explore the mechanism behind the toxicity. However, caution must be taken when using carbon black or any other material as a control because complications may occur. Carbon can adsorb the viability dyes, such as neutral red, and interfere with the absorption spectra. False positives will occur. The type of carbon black used is extremely important. For instance, ultrafine carbon black has been utilized in inhalation studies but dosing in cell culture gives different results, especially when conducting viability and cytokine assays.
Three dimensional skin cell cultures are also available commercially. They have shown to be able to predict irritation but may significantly overestimate absorption or penetration [113-116]. Assays listed above can be used but may not be applicable with nanomaterials due to adsorption.
Flow-through Diffusion Cell Studies
Diffusion cell system consists of flow-through diffusion blocks each containing multiple Teflon cells perfused by a constant temperature circulator through a Silastic oxygenator, an automatic fraction collector, and a desiccant. Circular fresh skin from pigs (pig skin mimics human skin and eliminates the extreme variability seen with random source human skin) or humans are placed epidermal – side up in Teflon flow-through diffusion cell. Compound containing nanoparticles is dosed on the epidermal side whilst the dermal side in each cell is bathed with receptor fluid at a set flow rate. The perfusate is collected at defined intervals up to 24 hrs and nanoparticles flux in the perfusate can be assessed by radioactivity counting, fluorescence, or UV detection. The skin surface can then be swabbed to remove non-absorbed surface particles and then tape stripped to remove a stratum corneum sample to assess nanoparticle penetration into this outermost epidermal layer. Serial sectioning of the skin can also be carried out [117,118].
Isolated Perfused Porcine Skin Flap (IPPSF)
The isolated perfused porcine skin flap (IPPSF) would be an ideal model to study the absorption and toxicity of nanomaterials. The IPPSF has an intact functional microcirculation, a viable epidermis and dermis and can be well controlled. A single-pedicle, axial pattern tubed skin flap is obtained from the abdomen of pigs following surgical creation of the flap perfused primarily by the caudal superficial epigastric artery and its associated paired venae commitantes. The IPPSF is transferred to the perfusion apparatus that is a custom designed temperature and humidity-regulated chamber. Nanomaterials can be topically dosed to the skin surface and perfusate samples collected over an eight hour period and assessed for nanoparticle flux [119-121].
Other acute toxicity in vitro assays are available but are used to test corrosives (rat transcutaneous electrical resistance (TER), commercially available EPISKIN, Epiderm and Corrositex) and irritation (EPISKIN, and Epiderm). However, the major traditional endpoint for skin toxicity is using the cell viability assay MTT reduction that has been shown to be unpredictable with nanomaterials due to marker interactions with nanoparticles.
4.2.2.3 Mucosa
Mucosa is the moist tissue that lines particular organs and body cavities throughout the body, including the nasal cavity, oral cavity, lungs, vagina and gastrointestinal tract. Potentially one of the most important portals of entry for nanoparticle exposure (excluding the nasal cavity and lung, which has been detailed above) is the gastrointestinal tract. Either accidental or intentional exposure via oral administration to the GI tract can lead to significant exposures. Efficient uptake of nanoparticles via the GI tract has been well documented in oral feeding studies and gavage studies using particles ranging from 10 nm to 500 nm [122-124]. In these studies nanoparticles translocated through the mucosal lining and epithelial barrier of the intestine and were associated with the GALT (gastroinstetinal associated lymphatic tissue) and circulatory system within as little as 60 minutes time [125].
Intestinal epithelium can be studied using a variety of methods including immortalized cell-lines and tissue constructs. An example of immortalized cell-lines used to study the uptake of materials across the intestinal epithelial barrier include Caco cells, which have been used in many pharmaceutical studies to determine intestinal permeability [126].
These assays could be adapted for use in in vitro translocation rate studies or for developing a mechanistic understanding of the translocation process. IEC-6 and IEC-18 cell lines have been used extensively in mechanistic studies of the intestinal epithelial lining as well and may represent useful tools for nanoparticle research [127-129]. These cells have been used to measure the activation of various signal pathways after toxicant exposure, as well as cytokine and ROS/RNS release [130,131].
Dependent upon the specific application, vaginal and oral-lining exposure may be possible although it is unlikely that these, in general, would represent significant portal of entry exposure routes. However, there are a variety of cell-lines and tissue constructs or models available for study of translocation and impact of nanoparticle exposure through these routes [132-134].
4.2.3 Cellular Assays
Study of target organs distal to the site of deposition pre-supposes that there is translocation and redistribution of nanoparticles away from the portals of entry in the lungs, skin or gut. As discussed above, potential target organs include blood, endothelium, neural tissue, heart, kidney, liver, and spleen.
4.2.3.1 Endothelium
The endothelium is represented by a thin layer of cells lining the vasculature throughout the body. It has been demonstrated that ultrafine particles or nanoparticles may have a wide range of effects on the endothelium. In vitro cultures may be useful in elucidating mechanistic information about transport across the alveolo-capillary or blood-brain barrier and on endothelial cell effects. Cultured endothelial cells are well suited to determining the effects nanoparticles may have on RNS production which has been demonstrated to play a significant role in the homeostasis of the vasculature [135-137].
4.2.3.2 Blood
In vitro studies using fractionated blood products (isolated red blood cells, platelets, leukocytes, or serum with complement) can be utilized in evaluating the effect on circulating blood. Activation of platelets, red blood cell interactions, production of ROS/RNS, cytokine/chemokine release from leukocytes, and complement activation are relevant endpoints to evaluate for nanoparticles. It has been demonstrated that nanoparticles have the ability to enter the circulatory system once translocation from site of entry has occurred [138,139].
4.2.3.3 Spleen
The spleen is a major site of immune processing and lymphoid maturation, and accumulation of particles in the spleen may have consequences for immune responses and immunopathology. Spleen cells can be isolated and studied for the effects of nanoparticles in vitro. Endpoints could include antigen processing and immune responsivity in vitro, markers of lymphoid cell differentiation, and functional aspects such as dendritic cell function and lymphocyte proliferation.
4.2.3.4 Liver
The liver is a complex organ and is structurally and functionally heterogeneous. The liver is the major site for biotransformation and defense against foreign materials and xenobiotics. It is an integral structure having two separate blood supplies, many different cell types, and many different functions. Liver injury, due to nanomaterials, may be characterized based on histologic lesions, such as inflammation or necrosis. Injury to the liver may also be characterized at the molecular level. Some of the most common mechanisms of hepatocellular injury are via the cytochrome P450 metabolic pathways. The liver can excrete materials into the bile; therefore, the biliary system may be exposed as well. In vivo there are reports that a variety of different toxins cause hepatocellular injury by a range of different mechanisms such as cytochrome P450 activation, alcohol dehydrogenase activation, membrane lipid peroxidation, protein synthesis inhibition, disruption of calcium homeostasis, and activation of pro-apoptotic receptor enzymes. Every effort should be made to use human derived cells for in vitro assays, because these studies could be used to predict toxicity in humans. However, there is a considerable human variability in enzyme function.
Primary Human Hepatocytes
Primary human hepatocyte cultures are available commercially with well-characterized metabolic profiles and a full complement of metabolizing enzymes. Availability of human cells is limited due to the increase in demand for liver transplantations.
Isolated Perfused Liver
This complicated model system would be suitable for the study of nanomaterials, because it is the closest model that mimics in vivo and would allow for a detailed characterization of particle distribution within the organ.
Liver Slices
Modern techniques of precision slicing have allowed liver slices to become a good model, because it retains the normal tissue organization which may be particularly critical for nanoparticle studies.
Collagen Sandwich Cultures
In this model system, the structural and functional integrity is retained for several days. The bile canaliculi are well preserved, and release of the enzymes alanine aminotransferase and aspartate aminotransferase can be evaluated.
In general, using these in vitro systems, hepatic metabolism can be studied using isolated hepatocytes and cell lines and evaluated for changes in CYP450. Microsomes may be used in screening nanomaterials for metabolism using LC/MS to identify metabolite formation. Subcellular fractions, liver slices and whole liver homogenates may be used to evaluate liver function and toxicity. To study the effects of nanomaterials on hepatic function specific endpoints, such as enzyme systems, mitochondrial function, albumin synthesis, cell detachment, gene and / or protein expression, and membrane damage should be considered. Mechanistically distinct endpoints could be utilized, such as cell morphology, viability, membrane damage, alamar blue metabolism, ATP content, covalent protein binding, peroxisomal proliferation, and GSH content. Other biomarker identifications, such as transcription and proteomic profiling should be studied. However, biomarkers may have limitations because immortalized cell lines are genotypically and phenotypically different from the organ itself. Also, hepatocyte cell cultures represent a single cell system and will only provide information on events that directly affect the cell itself. Some of these test systems may be able to predict the toxicity of nanomaterials as long as the assumptions and limitations are realized [140,141].
4.2.3.5 Nervous System
Central Nervous System
In vitro systems to study the effects of particles on the nervous system could include culture of neurons and addition of nanoparticles to determine effects on neuronal function. Endpoints could include ROS/RNS production, apoptosis, metabolic status, effects on the action potential and ion regulation in general. Microglial cells are a type of macrophage found in the brain, and they may be involved in handling any nanoparticle that gets to the brain. The responses of microglial cells to nanoparticles should be studied along the lines of those described for macrophages in the lung section. Other cells that could be studied for effects of nanoparticles are astrocytes, glial cells that have a number of important roles that influence the behavior of neurons, and oligodendrocytes which provide support to axons by producing the myelin sheath, which insulates the axons.
Peripheral Nervous System
The skin and the other portals of entry and target organs will have a nerve supply and the skin for example, has sensory nerves that are present near the surface of the body. Nanoparticles have the potential to gain access to these nerves and be transported or affect them in a number of ways. This could be studied by using neuron culture of peripheral nerves and studying the effect of nanoparticles for various relevant endpoints (e.g., dorsal root ganglion neurons). In the autonomic nervous system both sympathetic and para-sympathetic neurons can be cultured, and effects of nanoparticles on their viability, metabolism, electrical activity and ionic homeostasis could be studied.
4.2.3.6 Heart
Cardiac function could be altered by nanoparticles that find their way into the heart muscle from the microcirculation. Cardiomyocytes can be cultured and effects on their general viability, ionic homeostasis and metabolism could be ascertained. Additionally, cardiomyocytes beat with regular rhythm in vitro, and the effects on this could be measured and any effects examined as to mechanism.
4.2.3.7 Kidney
The kidney is a major filtering system to eliminate toxicants from the bloodstream and it has been demonstrated that nanoparticles can be excreted via the kidney. Whether there are adverse effects of nanoparticles on the kidneys is unknown but this can be evaluated using in vitro techniques. Permeability assays used in pharmaceutical studies can be evaluated for use in measuring translocation and penetration through the renal tubules. Effects on the epithelial tubules and vasculature can be evaluated with existing cell culture techniques. A variety of endpoints can be evaluated, including signal transduction response, oxidative stress, cellular viability, ion channel flux, modulation in the release of growth factors and proteinases as important indicators to renal homeostasis. Several models exist that may be useful to investigate nanoparticle-kidney interactions including renal tissue slices to evaluate translocation, oxidative stress, signal transduction responses and toxicity [142,143]. Immortalized cell-lines are an inexpensive alternative to kidney slice models and may provide mechanistic information on the cytotoxicity of nanomaterials. Cells derived from isolated glomeruli, distal tubule/ collecting ducts, proximal tubule or proximal nephrons have been well characterized and are commercially available [144,145]. An example is the use of HEK-293 cells (human embryonic kidney cells) to evaluate cytotoxicity of chemicals [146]. MDKK cells and LLC-PK1 cells have been used extensively in in vitro mechanistic studies and can be utilized to evaluate effects of nanoparticles [147,148].
4.2.4 Non-Cellular Assays
Durability
The ability of a particle to persist contributes to its ability to accumulate as dose. In fiber toxicology, there are well-documented dynamic and static protocols for assessing this property of durability in vitro, using Gambles balanced salt solution [149]. These fiber protocols could be modified to allow measurement of durability of nanoparticles in vitro.
Complement Activation
The complement system is a protein cascade that has evolved to detect foreign, mostly microbial, surfaces. It is, however, activated by asbestos [150] and by carbon nanoparticles [151]. Their high surface per unit mass and surface activity may mean that other types of nanoparticles might be potent at activating the complement system. This might modify the response by opsonising the particles (C3b) or causing inflammation by the production of anaphylatoxin (C5a). Studies on the ability of nanoparticles to activate the complement system are therefore warranted.
Adsorptive Properties
The large surface area of nanoparticles means that they can adsorb proteins [152,153]. Adsorption of different proteins might occur with different nanoparticle surfaces, and this could modify how they are handled by macrophages and other cells. This could therefore be a focus of study.
Free Radical Production
Most, probably all, pathogenic particles generate free radicals in cell-free systems, and this ability to cause oxidative stress contributes to their ability to initiate inflammation, and cause cell injury and genotoxicity [95,154-157]. The free radicals can arise as a consequence of stable radicals at the particulate surface (quartz) [158], redox cycling of ionic transition metal via the Fenton reaction (e.g. welding fume), [159] or by unknown surface mechanisms (nanoparticle carbon black) [160]. The ability of particles to generate free radicals can be assessed by a range of assays, including plasmid DNA scission [161], electron paramagnetic spin resonance, [162] and 8-OH-dG production in 'naked' DNA or a DCFD assay for in vitro ROS production [163].
Computational Toxicology
In addition to establishing screening methods mentioned above, efforts to assess risk associated with engineered nanomaterials or other environmental stressors should include a collection of new technologies called computational toxicology. Computational toxicology is defined as the application of mathematical and computer models and molecular biology approaches to improve prioritization of data requirements and risk assessments for environmental protection.
This approach involves four areas:
- computational chemistry which refers to physical-chemical-mathematical modeling at the molecular level and includes topics such as quantum chemistry, force fields, molecular mechanics, molecular simulations, molecular modeling, molecular design, and cheminformatics;
- molecular biology which allows for the characterization of genetic constituency and the application of wide coverage technologies, such as genomics, proteomics, and metabonomics, to provide the key indicators of cellular and organismal response to stressor input;
- computational biology or bioinformatics, which involves the development of molecular biology databases and the analysis of the data;
- systems biology which refers to the application of mathematical modeling and reasoning to the understanding of biological systems and the explanation of biological phenomena.
Computational toxicology is designed to increase the capacity to prioritize, screen, and evaluate materials by enhancing the ability to predict their toxicity. In addition to the "omics," quantitative structure-activity relationships (QSARs) developed in physical organic chemistry should be evaluated as to whether they can aid in predicting the structure-property relationship of nanomaterials. These multidisciplinary models could then be considered in a source-to-outcome continuum from environmental release through entire concentration, exposure concentration, target organelles, early biological effects, and adverse outcome.
While finding the relationship between structure and toxicity of nanomaterials using computational toxicology is likely a long time away, it is well to keep these models in mind while developing screening methods and to use current methods to validate and inform their development.
4.2.5 Research Gaps
There is a paucity of data on the effects of nanoparticles on these different target cells and their respective endpoints in vitro. We therefore identify an urgent research need to obtain more information about nanoparticles in all of these systems. We do, however, identify some pressing needs and these include:
1. In vitro assays need to be used to determine important parameters that drive the toxicity and translocation potential of nanoparticles e.g. size, surface area, surface reactivity, etc.
2. Decisions have to be made regarding the most appropriate and useful in vitro endpoints and their relative utility and importance (e.g. translocation, generation of ROS, cytokine release, cytotoxicity). A ranking of these in vitro assays in order of relevance and utility should be attempted.
3. In vitro data should be used to develop a paradigm for nanoparticle toxicity that predicts the toxicity based on measurement of in vitro parameters; when mature, this paradigm could be critically tested in vivo.
4. Toxicokinetic data should be used to select the target cells and systems that are appropriate and to select plausible dose levels to use in the in vitro assays.
5. The effect of nanoparticle form (e.g. singlet particles or aggregates, use of surfactant) should be determined and the nanoparticle dose should be characterised as much as possible, e.g. regarding surface area, metals, etc.
6. There is a pressing need to prepare and choose appropriate benchmark materials for in vitro testing.
7. How do we interpret in vitro results without appropriate mechanistic information from in vivo models?
4.2.6 Recommendations
Table 4 presents available in vitro systems for portal of entry testing. Table 5 presents available in vitro systems for other potential target organs.
Table 4 Available in vitro systems for portal of entry testing
Portal of Entry Cell/Tissue Type Effect Endpoint Research Gap
Lung Epithelium Toxicity Trypan blue, LDH, apoptosis
Inflammation Gene expression, oxidative stress, signal transduction pathways
Translocation Transfer of nanoparticles across membranes Translocation process
Carcinogenesis Genotoxicity, comet assay, 8OHdG, hprt assay, proliferation assay
Macrophages Toxicity Trypan blue, LDH, apoptosis
Chemotaxis Chemotaxis assay Recognition/Activation/Phagocytosis Process
Phagocytosis Particle uptake into cells, cytoskeletal staining
Inflammation Gene expression, oxidative stress, signal transduction pathways Additional markers?
Immune Cells Immune response Cytokine profile, adjuvant effects Additional markers?
Endothelium Inflammation Adhesion molecules, oxidative stress Additional markers?
Coagulation Von Willebrand factor, tissue factor
Fibroblasts Inflammation Oxidative stress, cytokine profile, gene expression profile
Fibrosis Collagen synthesis, cell proliferation
Lung slices Inflammation Oxidative stress, signal transduction pathway, immuno-histopathology
Translocation Particles across membranes Translocation process
Fibrosis Collagen synthesis
Skin Cell systems (e.g. HEK) Cytotoxicity Inflammation Cell viability – MTT, neutral red, Cytokine profile
Flow-through diffusion systems Absorption
Isolated Skin Flap Model Absorption, Cytotoxicity, Inflammation Glucose utilization, any other markers depending on end points (cytokine profiles, histopath, etc.)
Mucosa Intestinal epithelium (GI tract) Cytotoxicity Cell viability – MTT, neutral red, trypan blue Apoptosis
Inflammation Cytokine profile, oxidative stress, signal transduction pathway
Translocation Permeability assays
GALT Inflammation Cytokine profile, oxidative stress
Immune response Adjuvant effects Additional markers?
Buccal epithelium (oral cavitiy) Cytotoxicity Cell viability – MTT, neutral red, trypan blue Apoptosis
Inflammation Cytokine profile, oxidative stress, signal transduction pathway
Translocation Permeability assays
Vaginal epithelium (reproductive system) Cytotoxicity Cell viability – MTT, neutral red, trypan blue Apoptosis
Inflammation Cytokine profile, oxidative stress, signal transduction pathway
Translocation Permeability assays
Table 5 Available in vitro systems for potential target organs
Target Organ Cell/Tissue Type Effect Endpoint Research Gap
Endothelium Endothelial cells (e.g. HUV-EC-C) Cytotoxicity Cell viability – MTT, neutral red, trypan blue Apoptosis
Homeostasis Oxidative stress, gene expression profile Additional markers?
Translocation Permeability assays
Blood Red blood cells, platelets, bone marrow (megakaryocytes) Inflammation/Immune response Platelet activation
Cytokine/chemokine release from leukocytes
Oxidative stress
Complement activation
RBC/particle interactions Markers?
Liver Hepatocytes Toxicity Cell viability – MTT, neutral red, trypan blue Apoptosis
Kupffer cells Inflammation Cytokine profile, oxidative stress, signal transduction pathway, gene expression
Coagulation Von Willebrand factor, tissue factor
Isolated perfused liver slices Translocation, distribution Histopathology Translocation process
Liver slices Toxicity studies Cytotoxicity, P450 assay, ATP assays, GSH content Additional markers? Genomics, Proteomics?
Collagen sandwich cultures Toxicity studies Cytotoxicity, P450 assay, ATP assays, GSH content Additional markers? Genomics, Proteomics?
Spleen Lymphocytes Immune response Cytokine profile
Central and peripheral nervous system Neuronal cells Toxicity Cytotoxicity – Trypan blue, LDH Apoptosis
Inflammation Cytokine profile, oxidative stress, signal transduction pathway, gene expression
Translocation Gene expression, microscopic examination
Astroglial, Microglial cells Inflammation Cytokine profile, oxidative stress, signal transduction pathway, gene expression
Heart Cardiomyocytes Toxicity Cytotoxicity – Trypan blue, LDH Apoptosis
Inflammation Cytokine profile, oxidative stress, signal transduction pathway, gene expression
Function Beat – rhythm testing
Kidney Cell (e.g. HK-2, MDCK, LCC-PK1) Toxicity Cytotoxicity – Trypan blue, LDH Apoptosis
Inflammation Cytokine profile, oxidative stress, signal transduction pathway, gene expression
Translocation Permeability assays Additional markers?
Kidney slices Toxicity Cytotoxicity – Trypan blue, LDH Apoptosis
Inflammation Cytokine profile, oxidative stress, signal transduction pathway, gene expression
Translocation Permeability assays Additional markers?
1. In vitro tests are recommended as they provide a rapid and relatively inexpensive way to assess the potential toxicity of nanoparticles; there are, however, well-documented drawbacks of the in vitro assays such as their relative simplicity and the high doses commonly utilized.
2. We recommend that "benchmark" particle controls be utilized in all studies such as crystalline silica and respirable TiO2.
3. Non-cellular tests including nanoparticle durability, complement activation, adsorption and free radical production can all yield valuable data on potential harmfulness of nanoparticles; computational toxicology may also make a contribution.
4. Attention should be given to the potential confounding effect of adsorption of proteins or assay constituents onto the nanoparticles surface.
5. Various cell-based systems are available with varying benefits and drawbacks, including single cell cultures of cell lines and freshly-derived cells, co-cultures, organ cultures (e.g. tracheal explants) and heart-lung preparations.
6. The lung is a key target organ and so lung epithelial cells, macrophages, immune cells and fibroblasts represent key cells for nanoparticle effects with specific regard to inflammation, immunopathology, fibrosis, genotoxicity, microbial defense and clearance.
7. Skin represents a target for nanoparticles, especially from nanoparticles in cosmetics and a number of in vitro test systems are recommended including keratinocyte culture, Flow-through Diffusion Cell, and Isolated Perfused Porcine Skin Flap (IPPSF).
8. Mucosa, the moist tissue that lines the nasal cavity, oral cavity, lungs, vagina and gastrointestinal tract also represents a potential target for nanoparticles and various in vitro systems are available for testing and should be utilized.
9. The tendency of nanoparticles to gain access to the vasculature means that endothelium and components of the blood are potential targets of nanoparticles and these can be studied in vitro and we urge that this pathway receive special attention.
10. The spleen, kidney, heart and liver will be target organs for bloodborne nanoparticles and we advise that a number of in vitro test systems are available to model effects in these organs.
11. Transfer of nanoparticles to the brain and interactions with the autonomic nervous system in the lungs have been reported and we strongly recommend that in vitro models be utilized to study the impact of nanoparticles in these important neural cells.
4.3 In vivo Assays
The following section details a two tier approach to in vivo assays. In Vivo assays are presented for pulmonary, oral, dermal and injection exposures. Tier 1 evaluations are presented for all routes of exposure and Tier 2 evaluations are presented for pulmonary exposures. Tier 1 Evaluations include markers of damage, oxidant stress, and cell proliferation. The Tier 2 evaluation for pulmonary exposures includes deposition, translocation, and biopersistence studies; effects of multiple exposures; potential effects on the reproductive system, placenta, and fetus; alternative animal models; and mechanistic studies. The section concludes with identification of research gaps and a summary of principal recommendations related to in vivo testing of nanomaterials.
4.3.1 Pulmonary Exposure – Tier 1
Currently little information is available regarding airborne levels of nanomaterials generated during production and processing or quantities which may be aerosolized into the environment. However, due to their small size, aerosolization of respirable nanomaterials is likely, either as singlet or as aggregated particles and exposure by the inhalation route is a concern [48]. The following is a template which is recommended for the evaluation of possible adverse effects on the lung and other organ systems of pulmonary exposure to nanomaterials. A critical step in in vivo testing of nanomaterials is the characterization of the test material as described in Section 4.1.
4.3.1.1 Exposure Method
Inhalation
Inhalation is the preferred method of exposure of the respiratory tract for hazard identification and to obtain dose-response data. Physicochemical characterization of the generated aerosol is essential. Of particular interest is information regarding the particle size distribution of the aerosolized nanomaterial, i.e., singlet nanoparticles vs. aggregates of primary particles. The generated aerosol must be well controlled for particle size and concentration, and attempts should be made to reproduce human exposure conditions for a specific airborne nanomaterial. A critical barrier to conducting inhalation studies with nanomaterials is that the amount of material is often limited. Intratracheal inhalation uses less material than whole body or nose-only inhalation exposures, but still requires more than may be available. Under such constraints, pulmonary exposure by intratracheal instillation, pharyngeal or laryngeal aspiration is acceptable for hazard identification. It has to be kept in mind that the upper respiratory tract will not be targeted.
Intratracheal Instillation
Intratracheal instillation of nanomaterial suspended in an appropriate vehicle is considered an acceptable method for pulmonary exposure to evaluate the relative toxicity of the test material [164]. Efforts should be made to disaggregate the nanomaterial suspended in vehicle. Nanoparticles vary significantly in their dispersibility; given the lack of other general techniques for disaggregating such particles, vortexing plus sonication is recommended. However, both probe and bath sonication may generate significant localized heat and pressure, disrupting surface coatings which were intentionally used to impart specific characteristic to the nanoparticles. Where sonication must be used, bath sonication is recommended. In all cases, the specific preparation technique (duration and power of sonication) should be reported. The characterization of the suspended particles should define the exposure material with respect to vehicle and degree of vortexing and sonication of the sample. Suspension of nanomaterial in a serum or surfactant-containing vehicle is sometimes employed to assist disaggregation of the nanoparticles. However, because these substances would adhere to the particle surface, the effect of such surface coatings on the biological activity of the particle is an issue which must be evaluated.
Pharyngeal and Laryngeal Aspiration
Pharyngeal aspiration has been shown to be an effective exposure method which results in a relatively even distribution of particles throughout the lungs [165]. A concern with the pharyngeal aspiration technique for pulmonary exposure is the unintentional aspiration of food particles from the oral cavity during this procedure. Therefore, food should be withheld the night before the aspiration exposure. Furthermore, a naïve group should be added in addition to the vehicle control to evaluate any pulmonary effects of aspiration alone [165]. As an alternative to pharyngeal aspiration exposure and to avoid contamination with materials from the oral cavity, laryngeal aspiration may be used, particularly in rats.
4.3.1.2 Design
Animal Model
Currently, a large database exists using rats or mice to evaluate the pulmonary toxicity of particles. Therefore, for comparability to other toxicology studies, the use of the rat or mouse model is preferred. However, other animal models may be preferred to evaluate specific endpoints and would be acceptable.
Gender
At this time, there is no information concerning gender specific pulmonary sensitivity to nanomaterials. Therefore, there is no recommendation as to a preferred gender for these studies.
Dosimetry
Since mass may not be the proper dose metric for comparing the toxicity of fine vs. ultrafine particles [166,167], characterization of the test material should also include surface area per mass and particle number per mass. For practical purposes, dose could be monitored as mass delivered/animal or mass inhaled/animal and then be converted easily to a surface area or particle number dose as necessary, provided the correlation between these three particle parameters is available.
Benchmark Material
To place any pulmonary response to exposure to a given nanomaterial in perspective, results should be compared to those for particles of well-defined toxicity. Such benchmark materials could include nano-sized TiO2, carbon black, or crystalline silica. These benchmark materials should be characterized for surface area and particle number per mass, as well as for particle size and with respect to chemical purity and crystallinity to allow comparisons to be made using a variety of dose metrics.
Exposure Concentration
It is recommended that a minimum of three exposure levels be used. Information regarding the actual anticipated exposure levels in humans would be useful in determining the exposure concentration range to be evaluated. However, such information for nanoparticles is often lacking. In all cases, similar exposure concentrations of the test and benchmark materials should be used, and the various dose metrics discussed above should be considered when choosing the exposures for the benchmark and test materials. It is recommended that the highest concentration chosen should exhibit toxicity with the benchmark material.
Exposure Duration
For intratracheal instillation or pharyngeal aspiration, a single exposure to the nanomaterial is sufficient for Tier 1 studies. Caveat: consider high dose and bolus effect! For inhalation, a two week exposure is recommended, although shorter exposures, perhaps at higher concentrations, should be done if this mimics human exposures.
Pulmonary Parameters
Pulmonary responses should be monitored 24 hours to 28 days post-exposure. A suggested time course could be 24 hrs, 1 week and 28 days post-exposure.
4.3.1.3 Pulmonary Endpoints
Inhalation Studies
The degree/intensity and duration of pulmonary inflammation and cytotoxic effects following nanoparticle exposures are important endpoints for assessing the toxicity of a test nanoparticle.
1. Bronchoalveolar lavage (BAL) damage markers – BAL profile. This method samples the cells and fluid from the bronchoalveolar space and allows the assessment of inflammation by quantification of cell numbers and types and components of the fluid phase. In addition, considerable extra information can be gained by various ex vivo manipulations of the BAL cells, e.g., gene expression, phagocytic potential, etc. Other BAL damage markers include BAL lactate dehydrogenase levels (as a measure of cytotoxicity), BAL protein levels (increases in BAL fluid protein concentrations generally are consistent with enhanced permeability of vascular proteins into the alveolar regions, indicating a breakdown in the integrity of the alveolar-capillary barrier), and BAL alkaline phosphatase levels (as a measure of Type 2 alveolar epithelial cell toxicity). Methodologies for cell counts, differentials, and pulmonary biomarkers in lavaged fluids have previously been described [168,169].
2. Oxidative stress markers – ROS/RNS. Reactive oxygen and nitrogen species have been implicated in DNA damage and induction of inflammatory cytokines and growth factors. Acellular BAL fluid levels of gluthathione, total antioxidants, or nitrate/nitrite (a measure of nitric oxide production), lipid peroxidation of lung tissue, or ex vivo measurement of ROS/RNS from BAL cells can be employed to monitor oxidant generation and oxidant stress. Methodologies for oxidative stress markers have been described [170,171].
3. Histopathology – Description of the general effects of treatments on the lungs should include endpoints such as presence of dust-laden macrophages, cellular infiltrates and hyperplastic changes in the epithelium. It is recommended that the entire respiratory tract be evaluated for adverse pathological effects. This would include the upper respiratory tract – the nose, larynx and upper airways; the lower respiratory tract and lymph nodes; and the pleural region. Histopathological observations in a Tier 1 process would focus primarily on inflammatory responses and the development of fibrosis. Fibrosis can be determined in lung tissue by specific staining of collagen in histopathological slides, or by qualitative and quantitative histopathology.
4. Cell proliferation – Increased cell division plays a key role in pathological responses and can be determined in epithelial or mesothelial cells by uptake of labeled nucleotide precursors, such as tritiated thymidine or BrdU. Recommended experiments are designed to measure the effects of particle exposures on airway and lung parenchymal cell turnover in rats following exposures. Groups of particulate-exposed rats and corresponding controls can either be pulsed or implanted subcutaneously with minipumps containing 5-bromo-2'deoxyuridine (BrdU) dissolved in a sodium bicarbonate buffer solution. Methodologies for cell proliferation studies have previously been described [168,169]. An alternative method to BrDU staining is the PCNA staining method. Proliferating cell nuclear antigen (PCNA) is a nuclear protein associated with cell proliferation and has been used to discriminate via immunohistochemistry proliferating cells in numerous tumor types including those found in the lung [172-174].
Intratracheal Instillation or Pharyngeal/Laryngeal Aspiration Studies
As discussed above, inhalation is the most physiologically relevant and therefore preferred method of pulmonary exposure for hazard identification and to obtain dose-response data. However, both intratracheal instillation of nanomaterials suspended in an appropriate vehicle and pharyngeal or laryngeal aspiration (with appropriate caveats) are considered to be acceptable methods for pulmonary exposure to evaluate the relative toxicity of the test material. Similar to studies using aerosol exposures, the following pulmonary endpoints should be evaluated in a Tier 1 testing strategy approach for assessing lung hazards to nanoparticles:
1. BAL damage markers
2. Oxidative stress markers
3. Histopathology
4. Cell proliferation
4.3.1.4 Other Organ Endpoints
Exposure to nanoparticles via the respiratory tract includes a high probability of translocation to other organs and tissues – depending on nanoparticle size and surface chemistry. Although translocation rates may be very low, localization at sensitive sub-cellular sites (e.g., mitochondria) could result in adverse responses directly induced by the nanoparticle. Alternatively, or in addition, potential oxidative stress and inflammatory responses elicited by nanoparticles in the respiratory tract may result in the release of mediators which can lead to indirect secondary effects in extra-pulmonary organ systems. Thus, it is essential to include an evaluation of potential effects in remote organs and tissues, such as liver, spleen, bone marrow, heart, kidney, and CNS, in the Tier 1 evaluation.
Histopathological examination of extra-pulmonary tissues should be mandatory; however, this alone may only show significant effects following longer-term or very high exposures. Therefore, consideration should also be given to determining organ-specific endpoints, such as acute phase proteins and coagulation factors, for effects on the cardiovascular system, immune response assays for effects on the spleen, and immunohistochemical staining for dopaminergic neurons in brain sections to evaluate neurogenic effects. Additional functional assays (e.g., measurement of heart rate variability) may be considered, but since they require specific equipment and expertise, these are not mandatory for Tier 1 studies.
4.3.2 Pulmonary Exposure – Tier 2
Research results showing that nanoparticles can translocate from the portal of entry, the respiratory tract, via different pathways to other organs/tissues makes them uniquely different from larger-sized particles in that they may induce direct adverse responses in remote organs. In particular, such responses may be initiated through the interaction of nanoparticles with sub-cellular structures following endocytosis by different target cells. For this reason, special attention needs to be given to recognizing such effects, which in the healthy organism are probably very subtle initially, even not detectable, but could have serious consequences in a compromised organism or a compromised organ. Examples are effects of anthropogenic ultrafine particles in asthmatics, in people with cardiovascular diseases, in the elderly and very young. Complementary Tier 2 studies can be used to obtain more data for hazard identification as an initial step for risk assessment. Tier 2 studies will provide additional information to either characterize further effects seen in Tier 1 studies or to obtain new data using specific models of susceptibility. Ideally, studies should be performed using inhalation exposures as a first choice, in particular if a positive response was seen in Tier 1 studies when intratracheal instillation or pharyngeal/laryngeal aspiration was used. If insufficient amounts of material are available, multiple low dose (1–10 μg/kg body weight) exposures of the respiratory tract using the above-mentioned non-inhalation methods can be applied, i.e., dosing once or twice/week for 4 weeks with 2–3 months follow-up.
4.3.2.1 Animal Models
The absence of an effect of nanoparticles in a normal animal model does not imply that there will be no effect in a model which exhibits enhanced susceptibility. Increased susceptibility can be due to a number of factors, including age, disease, altered organ function, genetic polymorphism. Respective animal models include exposures of senescent, transgenic and knockout animals and animals with compromised organ systems (e.g., hypertension; diabetes models; immuno-compromised, infectivity models). In general, susceptibility models include compromised functions of the respiratory tract, CNS, cardiovascular system (dysfunction of endothelial cells, platelets), bone marrow, and/or kidney. It is essential that the models used are relevant to the human disease state and that a respective animal model has been validated in the peer-reviewed literature. In addition to obtaining information for identifying nanoparticle hazards, Tier 2 studies will also provide data on underlying mechanisms which can be used in concert with the mechanistic in vitro studies.
4.3.2.2 Multiple Exposures
As indicated above and under Tier 1 studies, repeated inhalation exposures are the first choice for realistic dosing. Information about the physicochemical nature of airborne nanoparticles at the workplace, or anticipated exposure of the general public (consumer), is essential to mimic the same for animal exposures. Issues include: are the nanoparticles aggregated or singlets; what is the diameter in the airborne state; what is known about other chemical characteristics (see discussion in this document, Section 4.1.)? Generation and monitoring of airborne nanoparticles for inhalation exposures requires special equipment and expertise. As an alternative to inhalation, intratracheal instillation, oropharyngeal or laryngeal aspiration can be used. However, unless the nanoparticles are coated and made "soluble" in physiological solutions, artifacts due to aggregation may occur when using these non-inhalation methods. In addition, the upper respiratory tract is circumvented by the non-inhalation methods, thereby eliminating potential neuronal nanoparticle translocation to the CNS. To address this concern, one could expose the animal to nanoparticles by nasal instillation. Techniques for tracking the uptake of nanoparticles by sensory neurons would be similar to those discussed in the section concerning deposition, translocation, and biopersistence (Section 4.3.2.3). Furthermore, the impact of coating for purposes of "solubilizing" the nanoparticles has to be carefully considered in cases where anticipated human exposure is to the uncoated material, since cellular uptake, translocation and effects will be affected by the surface coating.
Daily repeated inhalation exposures over four weeks are suggested for the Tier 2 studies, with an up to 3-month post-exposure observation period, including interim post-exposure sacrifice days. With respect to the non-inhalation methods of exposure, dosing can occur 1–2 times/week over a 4-week period, followed by the post-exposure observation period. Special attention needs to be given to the selection of exposure concentrations (inhalation) and doses (instillation, aspiration). Knowledge about anticipated human exposure will be extremely valuable, and use of predictive particle deposition models (e.g., MPPD model) can be used to determine realistic exposure concentrations/doses.
The Tier 2 studies are aimed at obtaining additional information with respect to the biokinetics of the nanomaterials following exposure of the respiratory tract (see below, "Deposition, translocation and biopersistence studies"), the stability of nanoparticles in the organ system (e.g., in vivo change of surface chemistry, bioavailability of core material), and the potential acute and sub-acute effects in the mammalian organism, including genomic and proteomic evaluation. Other endpoints are the same as listed under Tier 1 studies for the respiratory tract.
4.3.2.3 Deposition, Translocation and Biopersistence Studies
Exposure to airborne nanoparticles via the inhalation route leads to deposition in the various compartments of the respiratory tract according to probabilities dependent on three important parameter groups: aerodynamic and thermodynamic nanoparticle properties, breathing pattern, and the three-dimensional geometry and structure of the respiratory tract. Deposition probability of nanoparticles below a thermodynamic diameter of 500 nm increases with decreasing size because of the increasing diffusion velocity leading to an increased deposition in the small airways and the alveoli, in particular. Below 20 nm, the location of deposition of nanoparticles changes to the upper respiratory tract because of their even higher diffusion velocity [175]. Currently existing computer codes provide a first estimate of deposition probabilities in the various regions of the respiratory tract which may require modification if there are indications that nanoparticles may undergo changes of their aero- and/or thermodynamic properties not being considered in those codes [176,177].
Once deposited, insoluble nanoparticles undergo clearance mechanisms specific to the region of the respiratory tract; i.e., at all regions nanoparticles will interact with proteins of the epithelial lining fluid potentially forming complexes which are likely to affect their subsequent metabolic fate and biokinetics [178]. On the epithelium of conducting airways, mucociliary clearance provides a rapid transport, to the larynx for further transport into the gastro-intestinal tract and excretion. Note, however, that there also occurs long-term retention in human airways for a fraction of nanoparticles which increases with the decreasing size of nanoparticles giving rise to cellular uptake. On the epithelium of the alveolar region there is no rapid transport so that phagocytosis by free phagocytes can occur with subsequent slow clearance to the larynx, but because of the limited capability of macrophages to recognize nanoparticles, endocytotic processes and trans-cellular transport by other cells like epithelial type I + II cells, become prominent together with para-cellular transport mechanisms across tight junctions under inflamed conditions [179,180]. These mechanisms result in translocation of nanoparticles into the interstitium, lymphatic drainage and possible translocation across endothelial cells into capillaries. This access to the blood circulation provides accumulation and possible adverse reactions in secondary target organs such as the cardiovascular system, liver, spleen, bone marrow, central nervous system, endocrine organs and interaction with endothelial cells, platelets and immuno-competent cells in the circulation. Besides the direct effects of translocated nanoparticles on secondary organs, indirect effects may occur as well triggered by interactions of nanoparticles at their site of retention in the respiratory tract with adjacent biological systems like cells, fluids, proteins and extracellular matrix. Subsequent cell activation can lead to release of cytokines and other mediators which subsequently diffuse into the circulation to induce adverse responses in secondary target organs. Because the underlying mechanisms of nanoparticle translocation and accumulation or mediator response in secondary target organs are not fully understood, the determination of nanoparticle kinetics should be a high priority.
The scenario described above relates to biopersistent nanoparticles which maintain their particulate state; however, nanoparticles even when they may be insoluble in water may not persist as a solid particle in cells and body fluids but may fully or partially (e.g., surface coating) disintegrate/dissolve, so that eventually a biopersistent core with different particle properties/toxicities is retained. Because of the high diversity of emerging nanoparticles, studies of their biopersistence should be regarded of high priority.
Methodologies for such biopersistence studies include:
• Radio-labeled or fluorescently or magnetically tagged monitoring in lungs and various organs
Radio-labeling of nanoparticles specifically with a radio-isotope of one of the contextual chemical elements of the particle matrix is the gold standard for studying translocation kinetics allowing for time-efficient, extremely high sensitivity and specificity measurements, particularly when they aim to account for a balance of the entire distribution in the body and in excretions [79,181]. Also fluorescent or magnetic labeling provides a powerful means of high sensitivity and specificity to determine translocated fractions in various target organs [181,182]. Extreme care should be taken; however, to assure that any label stays firmly with the nanoparticles, otherwise the results will be severely flawed. Also, evidence is required that the process of labeling does not modify the nanoparticle in its function and its surface since surface is the predominant interacting substrate with biological systems.
• If impossible to tag: EM to monitor deposition and fate
Labeling of nanoparticles will not be possible in every case. In this event tracking of nanoparticles by electron microscopy may be a suitable alternative to monitor the fate of electron-dense nanoparticles in the organism. The method of choice for evaluation is a quantitative morphometric approach, which is by far preferable over qualitative spotting images, which may lead to a wrong interpretation. In addition, it is strongly recommended to search for adequate alternate labeling techniques, which may not be obvious at the first glance but would still provide a feasible option, for example use of confocal microscopy with fluorescently labeled nanoparticles.
• Chemical analysis may be possible
If the matrix of the nanoparticle provides a characteristic chemical element or compound, chemical analysis of this characteristic provides another strong alternative to track the fate of nanoparticles. Importantly, contamination and possible endogenous background levels require careful distinction as well as the determination of the lower level of sensitivity of the analytical approach.
4.3.2.4 Genomics and Proteomics
Nanomaterials distributing to different tissues of the body following deposition in the respiratory tract can potentially affect multiple cellular functions, and it will be difficult to determine with conventional assays what changes and adverse effects may have occurred. Use of genomic and proteomic analyses should be considered, the former providing information about specific mechanisms at the molecular level (e.g., oxidative stress) and the latter linking this to the expression of proteins resulting in effects at the cellular and tissue level. Results from such analyses are needed to help in the interpretation of responses. Analysis of the results of these assays requires the input of bioinformatics which will help in the interpretation of elicited responses. Together, genomic and proteomic studies represent an effective strategy combining hypothesis-forming and hypothesis-driven research which is needed in the assessment of nanoparticle risks within the framework of a multidisciplinary team approach.
4.3.2.5 Effects on the Reproductive system, Placenta, and Fetus
The studies to assess Tier 2 reproductive effects following pulmonary exposures to nanoparticles should follow protocols similar to the OECD Guideline 422 for Testing of Chemicals (Combined Repeated Dose Toxicity Study with the Reproduction/ Developmental Toxicity Screening Test – adopted 3-22-1996). The test substance should be administered in gradual doses to several groups of male and female rats. Males should be dosed for a minimum of 4 weeks (which includes a minimum of 2 weeks prior to mating during the mating period and approximately 2 weeks post mating). Given the limited pre-mating dosing period in males, fertility may not be a particularly sensitive indicator of testicular toxicity and should be concomitant with a detailed histopathological analysis of the male gonads to assess impact on fertility and spermatogenesis.
Females should be dosed throughout the study – including 2 weeks prior to mating (with the objective of covering a minimum of 2 estrus cycles), the variable time to conception, the duration of pregnancy, and a minimum of 4 days after delivery, up to and including the day before the scheduled sacrifice. The duration of gestation should be recorded and is calculated from day 0 of pregnancy. Each litter should be examined as soon as possible after delivery to establish the number and sex of pups, stillbirths, live births, runts (pups that are significantly smaller than corresponding control pups), and the presence of gross abnormalities.
Live pups should be counted and sexed and litters weighed within 24 hours of parturition (day 0 or 1 post-partum) and on day 4 post-partum. In addition to the observations on parent animals, any abnormal behavior of the offspring should be recorded.
4.3.3 Oral Exposure – Tier 1
It is possible that during the life of a nanomaterial (production, application, disposal, etc) it may appear in the water supply or be inadvertently ingested. If this is a concern, the effects of oral exposure to the nanomaterial should be investigated. Exposure should be by a single gavage at a dose which would represent the worse case human exposure. As with pulmonary exposure, for oral exposure the physical and chemical properties of the test material should be characterized in the form delivered to the test animal. Rats or mice are the recommended model system. There is no preference concerning gender for such studies. The feces should be collected for four days post-exposure, and the amount of nanomaterial eliminated vs. retained should be determined. Particularly GALT, mesenteric lymph nodes and liver should be analyzed for the presence of nanoparticles. If absorption of the nanomaterial from the gastrointestinal tract is near zero, then the systemic effects of oral exposure to that nanomaterial need not be evaluated. However, if significant absorption of the nanomaterial is evident, evaluation of systemic toxicity is recommended using histology and functional assays as described for the various organ systems after pulmonary exposure.
4.3.4 Injection – Tier 1
Some nanomaterials are being evaluated as drug delivery systems. In such a case, the potential toxicity of this nanomaterial after injection should be evaluated. Rats or mice are the recommended model system. There is no preference concerning gender for these studies. If possible, a tagged nanoparticle should be injected and its distribution to various organs (liver, spleen, heart, bone marrow, kidney, and lung) and elimination in the feces should be monitored for a week post-exposure. Histology and functional assays (e.g., mitochondrial function) of the various organ systems should be implemented as described following pulmonary exposure (Section 4.3.1.4).
4.3.5 Skin Exposure – Tier 1
For skin absorption of nanomaterials, the most appropriate animal model should be used. Rats are most common but rabbits, guinea pigs and pigs are also used to assess toxicity and irritation. The rat and pig are recommended as the animal of choice. The rat being small and already having an established database in the field of toxicology by other routes of exposure could be used because the amount of nanomaterials needed would be less for this small species. Frequently, the domestic pig is utilized in absorption studies because the skin is anatomically, physiologically and biochemically similar to that of humans. Twenty-four hours prior to dosing with nanomaterials, the area (10% of the body surface) on the back should be clipped to remove hair. Three doses at log intervals (mg/cm2) plus controls (vehicle, no material controls and a positive control) should be applied to both normal skin and abraded skin to mimic how humans are exposed. The material should be applied in an occluded fashion because nanomaterials, unlike many chemicals, will not be absorbed into the skin immediately. The occlusion device (site protection) used to prevent the nanomaterial from falling off should be attached to the surface of the skin by non-irritating tape. At least 4–6 animals per group/ dose plus controls and vehicle controls should be utilized over the duration of 24 hrs. At 0.5, 1, 2, 4, 8, and 24 hrs, the treatment sites should be scored for erythema and edema using the Draize test scores. Skin biopsies should be taken for transmission electron microscopy that would identify cellular changes as well as localize the penetrated particles within the skin. Light microscopy could be utilized to assess the morphological alterations that could occur due to the acute toxicity of the nanomaterials but this will not detect nanomaterial localization. For repeated exposures, nanomaterials should be applied daily for 5 or 7 days and could continue until 28 days. This could depend on the type and the amount of nanomaterials that are available. If a 28 day study is planned, then daily clinical observations should be conducted. At termination, hematology, clinical chemistry, evaluation of local lymph nodes and an immunotoxicology battery of tests should be performed. Standard full necropsy exam (liver, kidney, etc) should also be conducted [183,184]. The specific goal of the study, dermal absorption or irritation, will dictate the specific study design (e.g. duration of study, samples collected).
4.3.6 Research Gaps
Significant research gaps exist; the first four included in the following list pertain to general informational needs on production, use, and exposure to nanomaterials that would be helpful in the design of toxicity test.
1. What is being made and in what quantities in the nanotechnology industry?
2. What exposure levels are likely in the workplace?
3. What is/are the likely route(s) of exposure?
4. What are occupational vs. environmental exposures?
5. Radio-labeled particles are needed for investigation of deposition, translocation, and biopersistence. This requires specific lab that can work with and detect labeled materials; labeling not feasible for many materials.
6. A source of reference nanomaterials should be available to researchers.
4.3.7 Recommendations
1. Studies involving in vitro (non-cellular and cellular), pulmonary, oral, injection and dermal exposure are recommended in Tier 1 testing.
2. It is essential that exposures in animal models be relevant to human exposures, when known, for production, use, and disposal.
3. It is recommended that exposure route be relevant to anticipated human exposures for production, use, and disposal.
4. It is recommended that the test material be fully characterized, preferably as delivered to the animal.
5. Following pulmonary exposure, recommended endpoints to be measured include organ-specific markers of inflammation, oxidant stress and cell proliferation (e.g., mitochondrial) and histopathology in the lung as well as measurement of damage to non-pulmonary organs.
6. Studies involving a) use of susceptible models, b) multiple exposures, c) evaluation of deposition, translocation and biopersistence, d) reproductive effects, and e) mechanistic genomic and proteomic techniques may be considered for Tier 2 testing.
5.0 Conclusion
Engineered nanomaterials presenting a potential risk to human health include those capable of entering the body and exhibiting a biological activity that is associated with their nanostructure. Nanomaterial-based products such as nanocomposites, surface coatings and electronic circuits are unlikely to present a direct risk as exposure potential will be low to negligible. Nanomaterials that are most likely to present a health risk are nanoparticles, agglomerates of nanoparticles, and particles of nanostructured material (where the nanostructure determines behavior). In each of these cases, exposure potential exists for materials in air and in liquid suspensions or slurries.
Recognizing the early stage of understanding of the potential toxicity of nanomaterials and that little knowledge exists regarding specific nanomaterial characteristics which may be indicators of toxicity, the elements of a screening strategy outlined in this document include a significant research component. The range and extent of recommended testing reflect this developing state of knowledge. The elements of a screening strategy are clear, but the detailed approach will evolve and become more focused and selective as the results of these early-stage screening/research studies become available. A more thorough discussion of the 'elements' presented and the development of a more robust and detailed strategy will only be possible as knowledge increases. Elements of a nanotoxicity testing strategy which have been detailed in previous sections are summarized below.
Physicochemical Characterization
Appropriate physicochemical characterization of nanomaterials used in toxicity screening tests is essential, if data are to be interpreted in relation to the material properties, inter-comparisons between different studies carried out, and conclusions drawn regarding hazard. The dependence of nanomaterial behavior on physical and chemical properties places stringent requirements on physicochemical characterization and includes assessing a range of properties, including particle size distribution, agglomeration state, shape, crystal structure, chemical composition, surface area, surface chemistry, surface charge and porosity. Precise requirements will differ for in vivo and in vitro studies, and according to the material delivery route or method. In addition, characterizing human exposures introduces a third set of requirements.
A wide range of analytical methods are available that are applicable to nanomaterials, and multidisciplinary collaborations are encouraged to ensure appropriate methods are adopted. Particular consideration should be given to the use of Transmission Electron Microscopy, which in many cases can be considered the gold standard of nanoparticle characterization. In addition, information on nanomaterial production, preparation, storage, heterogeneity and agglomeration state should be recorded in all cases. Characterization of nanomaterials after administration in vitro or in vivo is considered the ideal in screening studies, although it currently presents significant analytical challenges. Characterization of the material as administered is therefore recommended for most screening tests. Characterization of the nanomaterial solely as produced or supplied is only considered appropriate where the previous two approaches are not viable. In all screening studies, dose should be evaluated against appropriate metrics. The three principal physical metrics of interest are mass, surface area and number concentration of particles: given current uncertainty over the relevance of each, it is important that all three are measured or derivable in any given study.
In Vitro Testing Methods
In vitro tests of toxicity yield data rapidly and can provide important insights and confirmations of the mechanism of in vivo effects. We recommend that a wide range of in vitro tests be applied to the key research questions relating to the potential hazard associated with nanoparticles exposure. A wide range of in vitro approaches exist that can be matched to specific questions relating to different aspects of nanoparticles toxicity. Non-cellular tests can provide information on aspects such as biopersistence, free radical generation by particle surfaces and activation of humoral systems such as the complement system; computational toxicology methods may also be useful. Cell-based systems can comprise cell lines and freshly derived primary cells in monocultures or co-cultures. Organ cultures and heart/lung preparations are also potentially useful for studying nanoparticles effects and translocation. We recommend that the in vitro tests should reflect the different portals of entry and target organs that nanoparticles could impact and these include lung, skin, mucosal membranes, endothelium, blood, spleen, liver, nervous system, and heart. As always, care should be taken in interpreting data obtained from in vitro systems because of the high doses normally used in vitro and the impact of a bolus effect. There should be inclusion of appropriate benchmark particles to contextualize the results of in vitro assays. We recommend vigilance for artifactual effects peculiar to nanoparticles caused by their large adsorptive surface which can deplete cell products or assay constituents and thereby confound assay results. In addition to the utilization of existing test systems we suggest that new assays may be developed, for example to study transit of nanoparticles across cell layers.
In Vivo Testing Methods
For in vivo testing of nanomaterials, two tiers of studies are discussed. Tier 1 studies would involve pulmonary, oral, injection, and dermal exposure as would be relevant to the human exposure(s) of concern. A critical initial step in in vivo testing is full characterization of the test material. Endpoints of concern for pulmonary exposure involve organic-specific markers of inflammation, oxidant stress, and cell proliferation and histopathology in the lung as well as measurement of damage to non-pulmonary organs. Tier 2 pulmonary exposure studies are recommended but not mandated. These studies would provide useful information for a complete risk assessment of a nanomaterial. Tier 2 studies include: 1) use of susceptible models, 2) effects of multiple exposures, 3) deposition, translocation and biopersistence studies, 4) evaluation of reproductive effects, and 5) mechanistic studies employing genomic and proteomic techniques.
The testing strategy for in vivo studies would be of greatest value for hazard identification and risk assessment if exposure dose, route of exposure, and particle characteristics closely modeled those of human exposure. Therefore, an understanding of the life cycle of a given nanomaterial, i.e., exposure during production, upon use, and environmentally, is a critical research need.
6.0 Competing interests
The author(s) declare that they have no competing interests.
7.0 Authors' contributions
Günter Oberdörster served as the Chair of the Nanomaterial Toxicity Screening Working Group. Andrew Maynard served as the Chair of the physiochemical characteristics sub-group, Ken Donaldson served as the Chair of the in vitro testing methods sub-group, and Vincent Castranova served as the Chair of the in vivo assays sub-group. Julie Fitzpatrick managed this project for the ILSI Research Foundation/Risk Science Institute (Table 6).
Table 6 The Working Group members are listed below:
Günter Oberdörster (Chair) University of Rochester
Kevin Ausman Rice University
Janet Carter Procter & Gamble Company
Vincent Castranova National Institute for Occupational Safety & Health
Ken Donaldson University of Edinburgh (UK)
Julie Fitzpatrick ILSI Research Foundation/Risk Science Institute
Barbara Karn U.S. Environmental Protection Agency and Woodrow Wilson International Center for Scholars
Wolfgang Kreyling GSF National Research Centre for Environment and Health (Germany)
David Lai U.S. Environmental Protection Agency
Andrew Maynard Woodrow Wilson International Center for Scholars
Stephen Olin ILSI Research Foundation/Risk Science Institute
Nancy Monteiro-Riviere North Carolina State University
David Warheit DuPont Haskell Laboratory for
Health and Environmental Sciences
Hong Yang University of Rochester
8.0 Acknowledgements
This project was funded by the U.S. Environmental Protection Agency Office of Pollution Prevention and Toxics through Cooperative Agreement X-82916701 with the ILSI Research Foundation/Risk Science Institute.
The ILSI Research Foundation/Risk Science Institute would like to thank the Woodrow Wilson International Center for Scholars, Project on Emerging Nanotechnologies – created in partnership with the Pew Charitable Trusts – for hosting a meeting to bring this report to the attention of the scientific and regulatory communities.
David Mustra, U.S. Environmental Protection Agency, contributed significantly to Section 3.0, Literature Survey.
==== Refs
Hood E Nanotechnology: looking as we leap Environ Health Perspect 2004 112 A740 A749 15345364
Bernstein D Castranova V Donaldson K Fubini B Hadley J Hesterberg T Testing of fibrous particles: short-term assays and strategies Inhal Toxicol 2005 17 497 537 16040559
Donaldson K Stone V Tran CL Kreyling W Borm PJ Nanotoxicology Occup Environ Med 2004 61 727 728 15317911 10.1136/oem.2004.013243
Goodman CM McCusker CD Yilmaz T Rotello VM Toxicity of gold nanoparticles functionalized with cationic and anionic side chains Bioconjug Chem 2004 15 897 900 15264879 10.1021/bc049951i
Cheng YS Hansen GK Su YF Yeh HC Morgan KT Deposition of ultrafine aerosols in rat nasal molds Toxicol Appl Pharmacol 1990 106 222 233 2256113 10.1016/0041-008X(90)90242-M
Kelly JT Asgharian B Nasal molds as predictors of fine and coarse particle deposition in rat nasal airways Inhal Toxicol 2003 15 859 875 12872177
Baggs RB Ferin J Oberdorster G Regression of pulmonary lesions produced by inhaled titanium dioxide in rats Vet Pathol 1997 34 592 597 9396140
Bermudez E Mangum JB Wong BA Asgharian B Hext PM Warheit DB Pulmonary responses of mice, rats, and hamsters to subchronic inhalation of ultrafine titanium dioxide particles Toxicol Sci 2004 77 347 357 14600271 10.1093/toxsci/kfh019
Donaldson K MacNee W Potential mechanisms of adverse pulmonary and cardiovascular effects of particulate air pollution (PM10) Int J Hyg Environ Health 2001 203 411 415 11556145 10.1078/1438-4639-00059
Donaldson K Brown D Clouter A Duffin R MacNee W Renwick L The pulmonary toxicology of ultrafine particles J Aerosol Med 2002 15 213 220 12184871 10.1089/089426802320282338
Ferin J Oberdorster G Penney DP Pulmonary retention of ultrafine and fine particles in rats Am J Respir Cell Mol Biol 1992 6 535 542 1581076
Ferin J Pulmonary retention and clearance of particles Toxicol Lett 1994 72 121 125 8202922 10.1016/0378-4274(94)90018-3
Kuschner WG Wong H D'Alessandro A Quinlan P Blanc PD Human pulmonary responses to experimental inhalation of high concentration fine and ultrafine magnesium oxide particles Environ Health Perspect 1997 105 1234 1237 9370520
Oberdorster G Pulmonary effects of inhaled ultrafine particles Int Arch Occup Environ Health 2001 74 1 8 11196075 10.1007/s004200000185
Oberdorster G Oberdorster E Oberdorster J Nanotoxicology: an emerging discipline evolving from studies of ultrafine particles Environ Health Perspect 2005 113 823 839 16002369
MacNee W Donaldson K How can ultrafine particles be responsible for increased mortality? Monaldi Arch Chest Dis 2000 55 135 139 10949875
Oberdorster G Gelein RM Ferin J Weiss B Association of particulate air pollution and acute mortality: involvement of ultrafine particles? Inhal Toxicol 1995 7 111 124 11541043
Wichmann HE Spix C Tuch T Wolke G Peters A Heinrich J Daily mortality and fine and ultrafine particles in erfurt, germany part I: role of particle number and particle mass Res Rep Health Eff Inst 2000 5 86 11918089
Zhu Y Hinds WC Krudysz M Kuhn T Froines J Sioutas C Penetration of freeway ultrafine particles into indoor environments Journal of Aerosol Science 2005 36 303 322 10.1016/j.jaerosci.2004.09.007
Donaldson K Gilmour MI MacNee W Asthma and PM10 Respir Res 2000 1 12 15 11667958 10.1186/rr5
Peters A Wichmann HE Tuch T Heinrich J Heyder J Respiratory effects are associated with the number of ultrafine particles Am J Respir Crit Care Med 1997 155 1376 1383 9105082
Wong GW Lai CK Outdoor air pollution and asthma Curr Opin Pulm Med 2004 10 62 66 14749608 10.1097/00063198-200401000-00011
Pekkanen J Timonen KL Ruuskanen J Reponen A Mirme A Effects of ultrafine and fine particles in urban air on peak expiratory flow among children with asthmatic symptoms Environ Res 1997 74 24 33 9339211 10.1006/enrs.1997.3750
Penttinen P Timonen KL Tiittanen P Mirme A Ruuskanen J Pekkanen J Ultrafine particles in urban air and respiratory health among adult asthmatics Eur Respir J 2001 17 428 435 11405521 10.1183/09031936.01.17304280
Penttinen P Timonen KL Tiittanen P Mirme A Ruuskanen J Pekkanen J Number concentration and size of particles in urban air: effects on spirometric lung function in adult asthmatic subjects Environ Health Perspect 2001 109 319 323 11335178
Warheit DB Laurence BR Reed KL Roach DH Reynolds GA Webb TR Comparative pulmonary toxicity assessment of single-wall carbon nanotubes in rats Toxicol Sci 2004 77 117 125 14514968 10.1093/toxsci/kfg228
Zhang Q Kusaka Y Zhu X Sato K Mo Y Kluz T Comparative toxicity of standard nickel and ultrafine nickel in lung after intratracheal instillation J Occup Health 2003 45 23 30 14605425 10.1539/joh.45.23
Heyder J Takenaka S Long-term canine exposure studies with ambient air pollutants Eur Respir J 1996 9 571 584 8730022 10.1183/09031936.96.09030571
Zhang Q Kusaka Y Sato K Nakakuki K Kohyama N Donaldson K Differences in the extent of inflammation caused by intratracheal exposure to three ultrafine metals: role of free radicals J Toxicol Environ Health A 1998 53 423 438 9537280 10.1080/009841098159169
Zhou YM Zhong CY Kennedy IM Leppert VJ Pinkerton KE Oxidative stress and NFkappaB activation in the lungs of rats: a synergistic interaction between soot and iron particles Toxicol Appl Pharmacol 2003 190 157 169 12878045 10.1016/S0041-008X(03)00157-1
Barlow PG Donaldson K MacCallum J Clouter A Stone V Serum exposed to nanoparticle carbon black displays increased potential to induce macrophage migration Toxicol Lett 2005 155 397 401 15649623 10.1016/j.toxlet.2004.11.006
Brown DM Donaldson K Borm PJ Schins RP Dehnhardt M Gilmour P Calcium and ROS-mediated activation of transcription factors and TNF-alpha cytokine gene expression in macrophages exposed to ultrafine particles Am J Physiol Lung Cell Mol Physiol 2004 286 L344 L353 14555462 10.1152/ajplung.00139.2003
Cui D Tian F Ozkan CS Wang M Gao H Effect of single wall carbon nanotubes on human HEK293 cells Toxicol Lett 2005 155 73 85 15585362 10.1016/j.toxlet.2004.08.015
Hetland RB Cassee FR Refsnes M Schwarze PE Lag M Boere AJ Release of inflammatory cytokines, cell toxicity and apoptosis in epithelial lung cells after exposure to ambient air particles of different size fractions Toxicol In Vitro 2004 18 203 212 14757111 10.1016/S0887-2333(03)00142-5
Ramage L Guy K Expression of C-reactive protein and heat-shock protein-70 in the lung epithelial cell line A549, in response to PM10 exposure Inhal Toxicol 2004 16 447 452 15204760 10.1080/08958370490439614
Stearns RC Paulauskis JD Godleski JJ Endocytosis of ultrafine particles by A549 cells Am J Respir Cell Mol Biol 2001 24 108 115 11159043
Stone V Tuinman M Vamvakopoulos JE Shaw J Brown D Petterson S Increased calcium influx in a monocytic cell line on exposure to ultrafine carbon black Eur Respir J 2000 15 297 303 10706495 10.1034/j.1399-3003.2000.15b13.x
Oberdorster G Gelein RM Ferin J Weiss B Association of particulate air pollution and acute mortality: involvement of ultrafine particles? Inhal Toxicol 1995 7 111 124 11541043
Johnston CJ Finkelstein JN Gelein R Baggs R Oberdorster G Characterization of the early pulmonary inflammatory response associated with PTFE fume exposure Toxicol Appl Pharmacol 1996 140 154 163 8806881 10.1006/taap.1996.0208
Johnston CJ Finkelstein JN Mercer P Corson N Gelein R Oberdorster G Pulmonary effects induced by ultrafine PTFE particles Toxicol Appl Pharmacol 2000 168 208 215 11042093 10.1006/taap.2000.9037
Srivastava R McKinnon JT Todd P Particulates from PTFE degradation in terrestrial and microgravity Aviat Space Environ Med 1999 70 505 510 10332948
Cassee FR Muijser H Duistermaat E Freijer JJ Geerse KB Marijnissen JC Particle size-dependent total mass deposition in lungs determines inhalation toxicity of cadmium chloride aerosols in rats. Application of a multiple path dosimetry model Arch Toxicol 2002 76 277 286 12107645 10.1007/s00204-002-0344-8
Oberdorster G Significance of particle parameters in the evaluation of exposure-dose-response relationships of inhaled particles Inhal Toxicol 1996 8 73 89 11542496
Huang M Khor E Lim LY Uptake and cytotoxicity of chitosan molecules and nanoparticles: effects of molecular weight and degree of deacetylation Pharm Res 2004 21 344 353 15032318 10.1023/B:PHAM.0000016249.52831.a5
Warheit DB Nanoparticles: Health impacts? Materials Today 2004 7 32 35 10.1016/S1369-7021(04)00081-1
Yang L Watts DJ. Particle surface characteristics may play an important role in phytotoxicity of alumina nanoparticles Toxicology Letters 2005 Corrected Proof.
Shvedova AA Kisin ER Mercer R Murray AR Johnson VJ Potapovich AI Unusual inflammatory and fibrogenic pulmonary responses to single walled carbon nanotubes in mice Am J Physiol Lung Cell Mol Physiol 2005 15951334
Maynard AD Baron PA Foley M Shvedova AA Kisin ER Castranova V Exposure to carbon nanotube material: aerosol release during the handling of unrefined single-walled carbon nanotube material J Toxicol Environ Health A 2004 67 87 107 14668113
Lam CW James JT McCluskey R Hunter RL Pulmonary toxicity of single-wall carbon nanotubes in mice 7 and 90 days after intratracheal instillation Toxicol Sci 2004 77 126 134 14514958 10.1093/toxsci/kfg243
Monteiro-Riviere NA Nemanich RJ Inman AO Wang YY Riviere JE Multi-walled carbon nanotube interactions with human epidermal keratinocytes Toxicol Lett 2005 155 377 384 15649621 10.1016/j.toxlet.2004.11.004
Monteiro-Riviere NA Wang YY Hong SM Inman AO Nemanich RJ Tan J Proteomic analysis of nanooparticle exposure in human keratonicyte cell culture The Toxicologist 2005 84 2183
Cunningham MJ Magnuson SR Falduto MT Gene expression profiling of nanoscale materials using a systems biology approach The Toxicologist 2005 84 9
Muller J Huaux F Moreau N Misson P Heilier JF Delos M Respiratory toxicity of multi-wall carbon nanotubes Toxicology and Applied Pharmacology Corrected Proof.
Shvedova AA Castranova V Kisin ER Schwegler-Berry D Murray AR Gandelsman VZ Exposure to carbon nanotube material: assessment of nanotube cytotoxicity using human keratinocyte cells J Toxicol Environ Health A 2003 66 1909 1926 14514433
Oberdorster E Manufactured nanomaterials (fullerenes, C60) induce oxidative stress in the brain of juvenile largemouth bass Environ Health Perspect 2004 112 1058 1062 15238277
Lockman PR Koziara JM Mumper RJ Allen DD Nanoparticle surface charges alter blood-brain barrier integrity and permeability J Drug Target 2004 12 635 641 15621689 10.1080/10611860400015936
Yamakoshi Y Sueyoshi S Miyata N [Biological activity of photoexcited fullerene] Kokuritsu Iyakuhin Shokuhin Eisei Kenkyusho Hokoku 1999 50 60 10859936
Yamakoshi YN Yagami T Sueyoshi S Miyata N Acridine Adduct of [60]Fullerene with Enhanced DNA-Cleaving Activity J Org Chem 1996 61 7236 7237 11667639 10.1021/jo961210q
Yamakoshi Y Umezawa N Ryu A Arakane K Miyata N Goda Y Active oxygen species generated from photoexcited fullerene (C60) as potential medicines: O2-* versus 1O2 J Am Chem Soc 2003 125 12803 12809 14558828 10.1021/ja0355574
Avogbe PH yi-Fanou L Autrup H Loft S Fayomi B Sanni A Ultrafine particulate matter and high-level benzene urban air pollution in relation to oxidative DNA damage Carcinogenesis 2005 26 613 620 15591089 10.1093/carcin/bgh353
Beck-Speier I Dayal N Karg E Maier KL Schumann G Schulz H Oxidative stress and lipid mediators induced in alveolar macrophages by ultrafine particles Free Radic Biol Med 2005 38 1080 1092 15780766 10.1016/j.freeradbiomed.2005.01.004
Brown DM Wilson MR MacNee W Stone V Donaldson K Size-dependent proinflammatory effects of ultrafine polystyrene particles: a role for surface area and oxidative stress in the enhanced activity of ultrafines Toxicol Appl Pharmacol 2001 175 191 199 11559017 10.1006/taap.2001.9240
Derfus AM Chan WCW Bhatia SN Probing the Cytotoxicity of Semiconductor Quantum Dots Nano Letters 2004 4 11 18 10.1021/nl0347334
Shiohara A Hoshino A Hanaki K Suzuki K Yamamoto K On the cyto-toxicity caused by quantum dots Microbiol Immunol 2004 48 669 675 15383704
Li N Sioutas C Cho A Schmitz D Misra C Sempf J Ultrafine particulate pollutants induce oxidative stress and mitochondrial damage Environ Health Perspect 2003 111 455 460 12676598
Sayes CM Gobin AM Ausman KD Mendez J West JL Colvin VL Nano-C(60) cytotoxicity is due to lipid peroxidation Biomaterials 2005 26 7587 7595 16005959 10.1016/j.biomaterials.2005.05.027
Lovric J Bazzi HS Cuie Y Fortin GR Winnik FM Maysinger D Differences in subcellular distribution and toxicity of green and red emitting CdTe quantum dots J Mol Med 2005 83 377 385 15688234 10.1007/s00109-004-0629-x
Shiohara A Hoshino A Hanaki K Suzuki K Yamamoto K On the cyto-toxicity caused by quantum dots Microbiol Immunol 2004 48 669 675 15383704
Oberdorster G Gelein RM Ferin J Weiss B Association of particulate air pollution and acute mortality: involvement of ultrafine particles? Inhal Toxicol 1995 7 111 124 11541043
Oberdorster G Oberdorster E Oberdorster J Nanotoxicology: an emerging discipline evolving from studies of ultrafine particles Environ Health Perspect 2005 113 823 839 16002369
Oberdorster G Toxicology of ultrafine particles: in vivo studies Phil Trans Roy Soc London Series 2000 358 2719 2740
Warheit DB Laurence BR Reed KL Roach DH Reynolds GA Webb TR Comparative pulmonary toxicity assessment of single-wall carbon nanotubes in rats Toxicol Sci 2004 77 117 125 14514968 10.1093/toxsci/kfg228
Donaldson K Stone V Gilmour PS Brown DM MacNee W Ultrafine particles: mechanisms of lung injury Phil Trans Roy Soc London Series 2000 358 2741 2749
Shvedova AA Castranova V Kisin ER Schwegler-Berry D Murray AR Gandelsman VZ Exposure to carbon nanotube material: assessment of nanotube cytotoxicity using human keratinocyte cells J Toxicol Environ Health A 2003 66 1909 1926 14514433
Sayes C Fortner J Guo W Lyon D Boyd A Ausman K The Differential Cytotoxicity of Water-Soluble Fullerenes American Chemical Society 2004 4 1881 1887
Searl A A review of the durability of inhaled fibers and options for the design of safer fibers Ann Occup Hyg 1994 38 839 855
Fubini B Surface reactivity in the pathogenic response to particulates Environ Health Perspect 1997 105 1013 1020 9400693
Oberdorster G Ferin J Lehnert BE Correlation between particle size, in vivo particle persistence, and lung injury Environ Health Perspect 1994 102 173 179 7882925
Kreyling WG Semmler M Erbe F Mayer P Takenaka S Schulz H Translocation of ultrafine insoluble iridium particles from lung epithelium to extrapulmonary organs is size dependent but very low J Toxicol Environ Health A 2002 65 1513 1530 12396866 10.1080/00984100290071649
Semmler M Seitz J Erbe F Mayer P Heyder J Oberdorster G Long-term clearance kinetics of inhaled ultrafine insoluble iridium particles from the rat lung, including transient translocation into secondary organs Inhal Toxicol 2004 16 453 459 15204761 10.1080/08958370490439650
Oberdorster G Sharp Z Atudorei V Elder A Gelein R Kreyling W Translocation of inhaled ultrafine particles to the brain Inhal Toxicol 2004 16 437 445 15204759 10.1080/08958370490439597
Jia G Wang H Yan L Wang X Pei R Yan T Cytotoxicity of carbon nanomaterials: single-wall nanotube, multi-wall nanotube, and fullerene Environ Sci Technol 2005 39 1378 1383 15787380 10.1021/es048729l
Araujo L Lobenberg R Kreuter J Influence of the surfactant concentration on the body distribution of nanoparticles J Drug Target 1999 6 373 385 10342385
Kirchner C Liedl T Kudera S Pellegrino T Munoz JA Gaub HE Cytotoxicity of colloidal CdSe and CdSe/ZnS nanoparticles Nano Lett 2001 5 331 338 15794621 10.1021/nl047996m
Baltensperger UE Weingartner E Burtscher H Keskinen J Baron PA, Willeke K Dynamic Mass and Surface Area Measurement Aerosol Measurement Principles, Techniques, and Applications 2001 New York: John Wiley and Sons 387 418
Gebhart J Baron P, Willeke K Optical Direct-Reading Techniques: Light Intensity Systems Aerosol Measurement Principles, Techniques, and Applications 2001 New York: John Wiley and Sons 419 454
Ku BK Maynard A Comparing aerosol surface-area measurement of monodisperse ultrafine silver agglomerates using mobility analysis, transmission electron microscopy and diffusion charging J Aerosol Sci 2005 36 1108 1124 10.1016/j.jaerosci.2004.12.003
Cheng Y-S Baron P, Willeke K Condensation Detection and Diffusion Size Seperation Techniques Aerosol Measurement Principles, Techniques and Applications 2001 New York: John Wiley and Sons 569 601
Keller A Fierz M Siegmann K Siegmann HC Filippov A A Surface Science with Nanosized Particles in Carrier Gas J Vac Sci Technol A 2001 19 1 8 10.1116/1.1339832
Flagan R Baron P, Willeke K Electrical Techniques Aerosol Measurement: Principles, Techniques and Applications 2001 New York: John Wiley and Sons 537 568
Fubini B Aust AE Bolton RE Borm PJ Brunch J Ciapetto G Non-animal tests for evaluating the toxicology of solid xenobiotics Atla-Alternatives to Laboratory Animals 1998 26 579 617
Wilson WE Han HS Stanek J Turner JH Pui DYH The fuchs surface area, as measured by charge acceprance of atmospheric particles, may be a useful indicator of the surface area of particles deposite in the lungs Madrid Spain, Pergamon
Brown DM Stone V Findlay P MacNee W Donaldson K Increased inflammation and intracellular calcium caused by ultrafine carbon black is independent of transition metals or other soluble components Occup Environ Med 2000 57 685 691 10984341 10.1136/oem.57.10.685
Kim H Liu X Kobayashi T Kohyama T Wen FQ Romberger DJ Ultrafine carbon black particles inhibit human lung fibroblast-mediated collagen gel contraction Am J Respir Cell Mol Biol 2003 28 111 121 12495939 10.1165/rcmb.4796
Gilmour PS Rahman I Donaldson K MacNee W Histone acetylation regulates epithelial IL-8 release mediated by oxidative stress from environmental particles Am J Physiol Lung Cell Mol Physiol 2003 284 L533 L540 12573991
Schins RPF Donaldson K Nuclear factor kappa-B activation by particles and fibres Inhal Toxicol 2000 12 317 326 10.1080/08958370050165193
Timblin CR Guthrie GD Janssen YW Walsh ES Vacek P Mossman BT Patterns of c-fos and c-jun proto-oncogene expression, apoptosis, and proliferation in rat pleural mesothelial cells exposed to erionite or asbestos fibers Toxicol Appl Pharmacol 1998 151 88 97 9705890 10.1006/taap.1998.8450
Stringer B Kobzik L Environmental particulate-mediated cytokine production in lung epithelial cells (A549): role of preexisting inflammation and oxidant stress Journal of Toxicology & Environmental Health 1998 55 31 44 9747602 10.1080/009841098158601
Barrett EG Johnston C Oberdorster G Finkelstein JN Silica-induced chemokine expression in alveolar type II cells is mediated by TNF-alpha-induced oxidant stress Am J Physiol 1999 276 L979 L988 10362723
Hess DT Matsumoto A Kim SO Marshall HE Stamler JS Protein S-nitrosylation: purview and parameters Nat Rev Mol Cell Biol 2005 6 150 166 15688001 10.1038/nrm1569
Janssen YM Van Houten B Borm PJ Mossman BT Cell and tissue responses to oxidative damage Lab Invest 1993 69 261 274 8377469
Quinlan TR BeruBe KA Marsh JP Janssen YW Taishi P Leslie KO Patterns of inflammation, cell-proliferation, and related gene-expression in lung after inhalation of chrysotile asbestos American Journal Of Pathology 1995 147 728 739 7677184
Schins RP Mechanisms of genotoxicity of particles and fibers Inhal Toxicol 2002 14 57 78 12122560 10.1080/089583701753338631
Schins RP Knaapen AM Cakmak GD Shi T Weishaupt C Borm PJ Oxidant-induced DNA damage by quartz in alveolar epithelial cells Mutat Res 2002 517 77 86 12034310
Brown DM Roberts NK Donaldson K Effect of coating with lung lining fluid on the ability of fibres to produce a respiratory burst in rat alveolar macrophages Toxicology In Vitro 1998 ITRO 15 24 10.1016/S0887-2333(97)00093-3
Chao CC Park SH Aust AE Participation of nitric-oxide and iron in the oxidation of dna in asbestos-treated human lung epithelial-cells Archives Of Biochemistry And Biophysics 1996 326 152 157 8579364 10.1006/abbi.1996.0059
Coffey MJ Phare SM Peters-Golden M Interaction between nitric oxide, reactive oxygen intermediates, and peroxynitrite in the regulation of 5-lipoxygenase metabolism Biochim Biophys Acta 2002 1584 81 90 12385890
Renwick LC Donaldson K Clouter A Impairment of alveolar macrophage phagocytosis by ultrafine particles Toxicol Appl Pharmacol 2001 172 119 127 11298498 10.1006/taap.2001.9128
Meiring JJ Borm PJ Bagate K Semmler M Seitz J Takenaka S The influence of hydrogen peroxide and histamine on lung permeability and translocation of iridium nanoparticles in the isolated perfused rat lung Part Fibre Toxicol 2005 2 3 15982423 10.1186/1743-8977-2-3
Monteiro-Riviere NA Riviere JE Marquardt H, Schafer S, McCellan R, Welsch F Skin Toxicology Toxicology 1999 San Diego: Academic Press 439 457
Lansdown ABG Taylor A Zinc and titanium oxides: promising UV-absorbers but what influences do they have on the intact skin? Int J Cosmetic Science 1997 19 167 172 10.1046/j.1467-2494.1997.171712.x
National Institute of Health Guidance Document on Using In Vitro Data to Estimate In Vivo Starting Doses for Acute Toxicity 2001
Qiao GL Brooks JD Riviere JE Pentachlorophenol dermal absorption and disposition from soil in swine: effects of occlusion and skin microorganism inhibition Toxicol Appl Pharmacol 1997 147 234 246 9439719 10.1006/taap.1997.8288
Regnier M Ortonne JP Darmon M Histological defects of chronic benign familial pemphigus expressed in tissue culture Arch Dermatol Res 1990 281 538 540 2322013 10.1007/BF00412741
Regnier M Asselineau D Lenoir MC Human epidermis reconstructed on dermal substrates in vitro: an alternative to animals in skin pharmacology Skin Pharmacol 1990 3 70 85 2078346
Monteiro-Riviere NA Inman AO Snider TH Blank JA Hobson DW Comparison of an in vitro skin model to normal human skin for dermatological research Microsc Res Tech 1997 37 172 179 9144629 10.1002/(SICI)1097-0029(19970501)37:3<172::AID-JEMT2>3.0.CO;2-Q
Bronaugh RL Stewart RF Methods for in vitro percutaneous absorption studies IV: The flow-through diffusion cell J Pharm Sci 1985 74 64 67 3981421
Chang SK Riviere JE Percutaneous absorption of parathion in vitro in porcine skin: effects of dose, temperature, humidity, and perfusate composition on absorptive flux Fundam Appl Toxicol 1991 17 494 504 1794653 10.1016/0272-0590(91)90200-N
Riviere JE Bowman KF Monteiro-Riviere NA Dix LP Carver MP The isolated perfused porcine skin flap (IPPSF). I. A novel in vitro model for percutaneous absorption and cutaneous toxicology studies Fundam Appl Toxicol 1986 7 444 453 3781134 10.1016/0272-0590(86)90094-1
Riviere JE Monteiro-Riviere NA The isolated perfused porcine skin flap as an in vitro model for percutaneous absorption and cutaneous toxicology Crit Rev Toxicol 1991 21 329 344 1741948
Monteiro-Riviere NA Kemppainen BW, Reifenrath WG Specialized Technique: Isolated Perfused Porcine Skin Flap Methods for Skin Absorption 1990 Boca Raton: CRC Press 175 189
Jani PU Florence AT McCarthy DE Further histological evidence of the gastrointestinal absorption of polystyrene nanospheres in the rat International Journal of Pharmaceutics 1992 84 245 252 10.1016/0378-5173(92)90162-U
Jani PU McCarthy DE Florence AT Titanium dioxide (rutile) particle uptake from the rat GI tract and translocation to systemic organs after oral administration International Journal of Pharmaceutics 1994 105 157 168 10.1016/0378-5173(94)90461-8
Powell JJ Harvey RS Ashwood P Wolstencroft R Gershwin ME Thompson RP Immune potentiation of ultrafine dietary particles in normal subjects and patients with inflammatory bowel disease J Autoimmun 2000 14 99 105 10648120 10.1006/jaut.1999.0342
Moghimi SM Hunter AC Murray JC Long-circulating and target-specific nanoparticles: theory to practice Pharmacol Rev 2001 53 283 318 11356986
Zhou Y Yokel RA The Chemical Species of Aluminum Influences Its Paracellular Flux across and Uptake into Caco-2 Cells, a Model of Gastrointestinal Absorption Toxicol Sci 2005 87 15 26 15933224 10.1093/toxsci/kfi216
Kishikawa H Miura S Yoshida H Hirokawa M Nakamizo H Higuchi H Transmural pressure induces IL-6 secretion by intestinal epithelial cells Clin Exp Immunol 2002 129 86 91 12100026 10.1046/j.1365-2249.2002.01895.x
Quaroni A Isselbacher KJ Cytotoxic effects and metabolism of benzo[a]pyrene and 7, 12-dimethylbenz[a]anthracene in duodenal and ileal epithelial cell cultures J Natl Cancer Inst 1981 67 1353 1362 6273638
Li M Hurren R Zastawny RL Ling V Buick RN Regulation and expression of multidrug resistance (MDR) transcripts in the intestinal epithelium Br J Cancer 1999 80 1123 1131 10376961 10.1038/sj.bjc.6690475
Smith JM Johanesen PA Wendt MK Binion DG Dwinell MB CXCL12 activation of CXCR4 regulates mucosal host defense through stimulation of epithelial cell migration and promotion of intestinal barrier integrity Am J Physiol Gastrointest Liver Physiol 2005 288 G316 G326 15358596 10.1152/ajpgi.00208.2004
Urayama S Musch MW Retsky J Madonna MB Straus D Chang EB Dexamethasone protection of rat intestinal epithelial cells against oxidant injury is mediated by induction of heat shock protein J Clin Invest 1998 102 1860 1865 9819372
Fichorova RN Anderson DJ Differential expression of immunobiological mediators by immortalized human cervical and vaginal epithelial cells Biol Reprod 1999 60 508 514 9916021
Kubilus J Cannon CL Klausner M Lonardo EC Human tissue model for vaginal irritation studies The Toxicologist 2002 66 378
Kubilus J Breyfogle B Sheasgreen J Hayden P Wertz P Dale B Characterization of New Buccal and Gingival Epithelial Tissue MatTek Corporation web page 2004 1 2
O'donnell VB Free radicals and lipid signaling in endothelial cells Antioxid Redox Signal 2003 5 195 203 12716479 10.1089/152308603764816550
Fleming I Busse R NO: the primary EDRF J Mol Cell Cardiol 1999 31 5 14 10072711 10.1006/jmcc.1998.0839
Harrison DG Treasure CB Mugge A Dellsperger KC Lamping KG Hypertension and the coronary circulation. With special attention to endothelial regulation Am J Hypertens 1991 4 454S 459S 1910638
Nemmar A Hoylaerts MF Hoet PH Dinsdale D Smith T Xu H Ultrafine particles affect experimental thrombosis in an in vivo hamster model Am J Respir Crit Care Med 2002 166 998 1004 12359661 10.1164/rccm.200110-026OC
Oberdorster G Sharp Z Atudorei V Elder A Gelein R Lunts A Extrapulmonary translocation of ultrafine carbon particles following whole-body inhalation exposure of rats J Toxicol Environ Health A 2002 65 1531 1543 12396867 10.1080/00984100290071658
Worth AP Balls M Alternative (non-animal) methods for chemicals testing: current status and future prospects 2002 71 82
Dambach DM Andrews BA Moulin F New technologies and screening strategies for hepatotoxicity: use of in vitro models Toxicol Pathol 2005 33 17 26 15805052 10.1080/01926230590522284
Vickers AE Rose K Fisher R Saulnier M Sahota P Bentley P Kidney slices of human and rat to characterize cisplatin-induced injury on cellular pathways and morphology Toxicol Pathol 2004 32 577 590 15603542 10.1080/01926230490508821
Ricardo SD Bertram JF Ryan GB Reactive oxygen species in puromycin aminonucleoside nephrosis: in vitro studies Kidney Int 1994 45 1057 1069 8007575
Ryan MJ Johnson G Kirk J Fuerstenberg SM Zager RA Torok-Storb B HK-2: an immortalized proximal tubule epithelial cell line from normal adult human kidney Kidney Int 1994 45 48 57 8127021
Raychowdhury MK Ibarra C Damiano A Jackson GR JrSmith PR McLaughlin M Characterization of Na+-permeable cation channels in LLC-PK1 renal epithelial cells J Biol Chem 2004 279 20137 20146 14982925 10.1074/jbc.M311946200
Chou CC Riviere JE Monteiro-Riviere NA The cytotoxicity of jet fuel aromatic hydrocarbons and dose-related interleukin-8 release from human epidermal keratinocytes Arch Toxicol 2003 77 384 391 12851741 10.1007/s00204-003-0461-z
Rustom R Wang B McArdle F Shalamanova L Alexander J McArdle A Oxidative stress in a novel model of chronic acidosis in LLC-PK1 cells Nephron Exp Nephrol 2003 95 e13 e23 14520010 10.1159/000073019
Panneerselvam K Freeze HH Mannose enters mammalian cells using a specific transporter that is insensitive to glucose J Biol Chem 1996 271 9417 9421 8621609 10.1074/jbc.271.16.9417
Eastes W Potter RM Hadley JG Estimation of dissolution rate from in vivo studies of synthetic vitreous fibers Inhal Toxicol 2000 12 1037 1054 11015142 10.1080/08958370050164644
Warheit DB George G Hill LH Snyderman R Brody AR Inhaled asbestos activates a complement-dependent chemoattractant for macrophages Laboratory Investigation 1985 52 505 514 3990243
Barlow PG Donaldson K Maccallum J Clouter A Stone V Serum exposed to nanoparticle carbon black displays increased potential to induce macrophage migration Toxicol Lett 2005 155 397 401 15649623 10.1016/j.toxlet.2004.11.006
Brown DM Wilson MR MacNee W Stone V Donaldson K Size-dependent proinflammatory effects of ultrafine polystyrene particles: a role for surface area and oxidative stress in the enhanced activity of ultrafines Toxicol Appl Pharmacol 2001 175 191 199 11559017 10.1006/taap.2001.9240
Kim H Liu X Kobayashi T Kohyama T Wen FQ Romberger DJ Ultrafine carbon black particles inhibit human lung fibroblast-mediated collagen gel contraction Am J Respir Cell Mol Biol 2003 28 111 121 12495939 10.1165/rcmb.4796
McNeilly JD Heal MR Beverland IJ Howe A Gibson MD Hibbs LR Soluble transition metals cause the pro-inflammatory effects of welding fumes in vitro Toxicol Appl Pharmacol 2004 196 95 107 15050411 10.1016/j.taap.2003.11.021
Brown DM Donaldson K Borm PJ Schins RP Dehnhardt M Gilmour P Calcium and ROS-mediated activation of transcription factors and TNF-alpha cytokine gene expression in macrophages exposed to ultrafine particles Am J Physiol Lung Cell Mol Physiol 2004 286 L344 L353 14555462 10.1152/ajplung.00139.2003
Marano F Boland S Bonvallot V Baulig A Baeza-Squiban A Human airway epithelial cells in culture for studying the molecular mechanisms of the inflammatory response triggered by diesel exhaust particles Cell Biol Toxicol 2002 18 315 320 12240962 10.1023/A:1019548517877
Nel AE Diaz-Sanchez D Li N The role of particulate pollutants in pulmonary inflammation and asthma: evidence for the involvement of organic chemicals and oxidative stress Curr Opin Pulm Med 2001 7 20 26 11140402 10.1097/00063198-200101000-00004
Schins RP Knaapen AM Cakmak GD Shi T Weishaupt C Borm PJ Oxidant-induced DNA damage by quartz in alveolar epithelial cells Mutat Res 2002 517 77 86 12034310
McNeilly JD Heal MR Beverland IJ Howe A Gibson MD Hibbs LR Soluble transition metals cause the pro-inflammatory effects of welding fumes in vitro Toxicol Appl Pharmacol 2004 196 95 107 15050411 10.1016/j.taap.2003.11.021
Wilson M Lightbody JH Donaldson K Stone V Oxidative Interactions between ultrafine particles and metals in vitro and in vivo 2002
Gilmour PS Brown DM Lindsay TG Beswick PH MacNee W Donaldson K Adverse health effects of PM10 particles: involvement of iron in generation of hydroxyl radical Occupational & Environmental Medicine 1996 53 817 822 8994401
Schins RP Knaapen AM Cakmak GD Shi T Weishaupt C Borm PJ Oxidant-induced DNA damage by quartz in alveolar epithelial cells Mutat Res 2002 517 77 86 12034310
Leanderson P Soderkvist P Tagesson C Axelson O Formation of DNA adduct 8-hydroxy-2'-deoxyguanosine induced by man-made mineral fibres IARC Sci Publ 1988 422 424 3198230
Driscoll KE Howard BW Carter JM Janssen YM Mossman BT Isfort RJ Mitochondrial-derived oxidants and quartz activation of chemokine gene expression Adv Exp Med Biol 2001 500 489 496 11764986
Rao GV Tinkle S Weissman DN Antonini JM Kashon ML Salmen R Efficacy of a technique for exposing the mouse lung to particles aspirated from the pharynx J Toxicol Environ Health A 2003 66 1441 1452 12857634 10.1080/15287390306417
Oberdorster G Pulmonary effects of inhaled ultrafine particles Int Arch Occup Environ Health 2001 74 1 8 11196075 10.1007/s004200000185
Tran CL Buchanan D Cullen RT Searl A Jones AD Donaldson K Inhalation of poorly soluble particles. II. Influence Of particle surface area on inflammation and clearance Inhal Toxicol 2000 12 1113 1126 11114784 10.1080/08958370050166796
Warheit DB Carakostas MC Hartsky MA Hansen JF Development of a short-term inhalation bioassay to assess pulmonary toxicity of inhaled particles: comparisons of pulmonary responses to carbonyl iron and silica Toxicol Appl Pharmacol 1991 107 350 368 1847247 10.1016/0041-008X(91)90215-Z
Warheit DB Hansen JF Yuen IS Kelly DP Snajdr SI Hartsky MA Inhalation of high concentrations of low toxicity dusts in rats results in impaired pulmonary clearance mechanisms and persistent inflammation Toxicol Appl Pharmacol 1997 145 10 22 9221819 10.1006/taap.1997.8102
Fubini B Hubbard A Reactive oxygen species (ROS) and reactive nitrogen species (RNS) generation by silica in inflammation and fibrosis Free Radic Biol Med 2003 34 1507 1516 12788471 10.1016/S0891-5849(03)00149-7
Shi X Ding M Chen F Wang L Rojanasakul Y Vallyathan V Reactive oxygen species and molecular mechanism of silica-induced lung injury J Environ Pathol Toxicol Oncol 2001 20 85 93 11570677
Carter JM Driscoll KE The role of inflammation, oxidative stress, and proliferation in silica-induced lung disease: a species comparison J Environ Pathol Toxicol Oncol 2001 20 33 43 11570672
Fontanini G Pingitore R Bigini D Vignati S Pepe S Ruggiero A Growth fraction in non-small cell lung cancer estimated by proliferating cell nuclear antigen and comparison with Ki-67 labeling and DNA flow cytometry data Am J Pathol 1992 141 1285 1290 1361306
Ettlin RA Perentes E Kolopp M Lardelli P Arnold J Karamitopoulou E Overview of quantitative methods in toxicologic pathology In Vivo 1993 7 315 324 8218973
Kreyling W Semmler M Moller W Health effects of ultrafine particles Journal of Aerosol Science 2004 35 1155 1156 Ref Type: Journal (Full)
Asgharian B Miller FJ, Subramaniam RP Dosimetry software to predict particle deposition in humans and rats 1999
ICRP Annals of the ICRP: Human Respiratory Tract Model for Radiological Protection. Pergamon 1994
Kreyling W Scheuch G Clearance of particles deposited in the lungs 2000 New York, Basel: Marcel Dekker Inc
Rothen-Rutishauser BM Kiama SG Gehr P A three-dimensional cellular model of the human respiratory tract to study the interaction with particles Am J Respir Cell Mol Biol 2005 32 281 289 15640437 10.1165/rcmb.2004-0187OC
Kiama SG Cochand L Karlsson L Nicod LP Gehr P Evaluation of phagocytic activity in human monocyte-derived dendritic cells J Aerosol Med 2001 14 289 299 11693840 10.1089/089426801316970240
Kreyling WG Blanchard JD Godleski JJ Haeussermann S Heyder J Hutzler P Anatomic localization of 24- and 96-h particle retention in canine airways J Appl Physiol 1999 87 269 284 10409585
Moller W Barth W Kohlhaufl M Haussinger K Stahlhofen W Heyder J Human alveolar long-term clearance of ferromagnetic iron oxide microparticles in healthy and diseased subjects Exp Lung Res 2001 27 547 568 11597117 10.1080/019021401753181827
U.S.Environmental Protection Agency Health Effects Test Guidelines OPPTS 8702500 Acute Dermal Irritation 1998
U.S.Environmental Protection Agency Health Effects Test Guidelines OPPTS 8701200 Acute Dermal Toxicity 1998
|
16209704
|
PMC1260029
|
CC BY
|
2021-01-04 16:38:30
|
no
|
Part Fibre Toxicol. 2005 Oct 6; 2:8
|
utf-8
|
Part Fibre Toxicol
| 2,005 |
10.1186/1743-8977-2-8
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-701611532010.1186/1743-422X-2-70HypothesisReplicative homeostasis II: Influence of polymerase fidelity on RNA virus quasispecies biology: Implications for immune recognition, viral autoimmunity and other "virus receptor" diseases Sallie Richard [email protected] Suite 35, 95 Monash Avenue, Nedlands, Western Australia, 6009, Australia2005 22 8 2005 2 70 70 31 7 2005 22 8 2005 Copyright © 2005 Sallie; licensee BioMed Central Ltd.2005Sallie; 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.
Much of the worlds' population is in active or imminent danger from established infectious pathogens, while sporadic and pandemic infections by these and emerging agents threaten everyone. RNA polymerases (RNApol) generate enormous genetic and consequent antigenic heterogeneity permitting both viruses and cellular pathogens to evade host defences. Thus, RNApol causes more morbidity and premature mortality than any other molecule. The extraordinary genetic heterogeneity defining viral quasispecies results from RNApol infidelity causing rapid cumulative genomic RNA mutation a process that, if uncontrolled, would cause catastrophic loss of sequence integrity and inexorable quasispecies extinction. Selective replication and replicative homeostasis, an epicyclical regulatory mechanism dynamically linking RNApol fidelity and processivity with quasispecies phenotypic diversity, modulating polymerase fidelity and, hence, controlling quasispecies behaviour, prevents this happening and also mediates immune escape. Perhaps more importantly, ineluctable generation of broad phenotypic diversity after viral RNA is translated to protein quasispecies suggests a mechanism of disease that specifically targets, and functionally disrupts, the host cell surface molecules – including hormone, lipid, cell signalling or neurotransmitter receptors – that viruses co-opt for cell entry. This mechanism – "Viral Receptor Disease (VRD)" – may explain so-called "viral autoimmunity", some classical autoimmune disorders and other diseases, including type II diabetes mellitus, and some forms of obesity. Viral receptor disease is a unifying hypothesis that may also explain some diseases with well-established, but multi-factorial and apparently unrelated aetiologies – like coronary artery and other vascular diseases – in addition to diseases like schizophrenia that are poorly understood and lack plausible, coherent, pathogenic explanations.
==== Body
Introduction
1.1 Global impact of RNA polymerases
Many of the world's population suffer from acute and chronic viral infection. The two common types of chronic viral hepatitis (CVH), hepatitis B (HBV) and C (HCV) are major causes of death and morbidity; conservative estimates suggest 400 million people are persistently infected with HBV, while HCV may infect a further 200 million. Annually, in excess of two million people will die from cirrhosis or liver cancer caused by CVH, and many more suffer chronic ill health as result. During the 20 years since the human immunodeficiency virus (HIV) was identified, perhaps 40 million people have become infected worldwide and each year about a million die from resulting immunodeficiency and consequent opportunistic infections, particularly tuberculosis, and other complications. Poor countries bear a disproportionate burden of disease caused by these viruses that further exacerbate poverty through pervasive economic disruption and diversion of limited resources to healthcare and disease control. Emerging viral pathogens including West Nile virus (WNV), the SARS coronavirus, endemic viruses like Murray Valley, Japanese, and other encephalitis viruses, Dengue and yellow fever, and seasonal influenza, hepatitis A (HAV) and E (HEV) cause enormous further morbidity and mortality, while pandemic outbreaks of virulent influenza strains remain a constant threat. Together, these viruses probably kill more people every ten days than the Boxing Day Tsunami. RNA viral infections, including Foot and Mouth, Bovine Viral Diarrhea Virus (BVDV) and Hog Cholera Virus (HChV), cause similar devastation of animal populations with enormous economic consequences.
RNA polymerases generate massive genetic variability of RNA viruses and retroviruses that circulate within infected hosts as vast populations of closely related, but genetically distinct, molecules known as quasispecies. After translation, this genetic variability causes near-infinite antigenic heterogeneity, facilitating viral evasion of host defences. Tuberculosis, malaria and other cellular pathogens also express broad cell-surface antigenic heterogeneity, generated by DNA-dependent RNApol. Thus, RNA polymerases probably cause more morbidity and premature mortality in man, and other animals, and greater economic loss, than any other molecule.
1.2 RNA viruses and immune control
Despite a depressing global epidemiology that strongly suggests otherwise, the immune system is thought to "control" viruses. What practical meaning does "immune control" have for the individual? There is no argument for HBV, and other viruses, high affinity antibody, generated by prior vaccination or other exposures and directed against neutralizing epitopes, will prevent HBV infection (excepting vaccine escape mutations [1,2]), in part by blocking viral ligand interaction with cell receptors, or that most patients exposed to HBV develop neutralizing antibodies (HBsAb), clear HBsAg from serum, and will normalize liver function long term. However, even patients who develop robust immune responses to HBV, defined by high-affinity antiHBsAb and specific antiviral cytotoxic T cell (CTL) responses, will have both "traces of HBV [3] ... many years after recovery from acute hepatitis" [3] and transcriptionally active HBV demonstrable in peripheral blood mononuclear cells (PBMCs) [4]. Furthermore, occult HBV is detected in liver tissue of patients with isolated antiHBc (i.e. HBsAg/HBsAb negative) [5] and in patients with HBsAg-negative hepatocellular carcinoma [6] suggesting, at least some patients, HBV in may persist irrespective of any immune responses, implying long term latency and low level basal replication may be a survival/reproductive strategy for HBV.
For most patients, acute HCV or HIV infection results in life-long viral persistence. Although many patients develop immunological responses, including specific antibody and CTL reactivity to various viral antigens, these responses have little discernible impact on either HCV or HIV replication that occurs essentially unchecked at rates estimated between 1010 and 1012 virions per day [7,8], indefinitely, while progressive destruction of liver or immune cells proceeds, commonly resulting in cirrhosis or liver cancer (for HCV) or death from immune deficiency (for HIV). Evidence that prior HCV infection confers no protective immunity against heterologous HCV infection in humans [9] or chimpanzees [10] or against either homotypic [11] or heterotypic [12] human reinfection, confirmation that active HCV infection persists long after either apparent spontaneous [13] or treatment-induced [14] viral clearance, or that vaccines causing specific antiviral B and T cell responses fail to protect against infection in animals [15], and that antibodies to HCV envelope protein E2 are only detected in animals with persistent infection [16,17], further undermines the potency of "immune control" and suggests, at least for patients with HCV, the definition of "control" may need to broadened significantly.
Based on observations that stronger specific CD4/CD8 immune responses with T-helper (TH1) cytokine profiles are found more frequently in patients with self limiting viral infections than those who develop chronic viral carriage [18,19] it is thought ability to mount robust adaptive immune responses predicts viral clearance while failure to do so results in chronic viral carriage [20]. However, detailed and very painstaking studies, albeit in small numbers of chimpanzees [21] and patients following antiviral therapy [22], have failed to demonstrate any relationship between T cell responses and viral clearance. Although development of TH1and other immune responses are certainly temporally and, probably, causally related to reduced viral replication and viral clearance the assumed direction of causality (immune response -> reduced viral replication), is not proved by the fact those responses develop, post hoc ergo propter hoc, as comforting a conclusion as it may be to reach.
The first part of this paper explores the impact of RNApol fidelity on quasispecies behaviour, specifically in mediating immune avoidance during acute HCV infection. We suggest the primary event causing reduction in viral replication is inhibition of RNApol processivity by variant viral proteins, specifically envelope and envelope-related proteins. We also suggest that immune responses to viruses are thwarted initially by broad antigenic diversity generated by low RNApol fidelity but develop, when they do, after viral replication falls (because of reduced RNApol processivity) and polymerase fidelity increases – linked events that occur because of replicative homeostasis – thus restricting antigenic diversity sufficiently to permit focused immune recognition. We further suggest immune responses strategically exploit replicative homeostasis to force viruses to reveal critical dominant antigenic epitopes, facilitating progressively more focused immune responses. The second part explores the ineluctable consequence of viral RNA quasispecies: That is, translation of RNAs into protein quasispecies with a spectrum of phenotypes and unpredictable properties, among which may be disruption of the cell surface receptors that viruses co-opt for cell entry. This innate property of viral quasispecies may explain a wide variety of diseases apart from viral autoimmunity.
2. Immunological, viral and biochemical kinetics following acute viral hepatitis
Acute HCV and HBV infection have characteristic kinetics of viral replication, adaptive immune responses, and cause predictable tissue injury, reflected in elevated serum aminotransferases. These kinetic and transaminase responses are summarized schematically for patients with persistent infection (figure 1) [23]. Initial HCV replication is very rapid and viral load increases exponentially until about week 4, at which point viraemia increases more slowly, and asymptotically, towards ~107 genome equivalents (geq)/ml by weeks 7–8 (these kinetics alone suggesting competitive inhibition of RNApol). This exponential increase of viral RNA in serum reflects explosive dissemination of virus in tissues, detectable by in-situ hybridisation throughout hepatocytes, including the nuclei, within days of infection [24]. Viral replication declines rapidly from weeks 10–11 to weeks 14–16 falling by 102–3 geq/ml but lower level (~105 geq/ml) fluctuating replication persists, generally indefinitely, thereafter. By contrast, neither HBV DNA nor HBV antigens are detectable in either serum or liver for 4–7 weeks post infection [25,26]. Elevation of alanine aminotransferase (ALT), reflecting hepatocyte injury, is typically much greater for HBV than HCV, peaks about two weeks after replication of either virus declines. Fluctuating transaminase elevation – mirroring fluctuating viraemia in HCV infection [27] – often persists indefinitely. This kinetic profile contains three paradoxes:
Figure 1 Viral replication, immunological and tissue injury kinetics following acute HCV and HBV infection. Data summated from Figure 1 [29] and modified to represent typical patients with chronic viral persistence. Note: a) High level HCV replication for 6–8 weeks prior to any immune responses, b) onset of humoral immune response well after down-regulation of viral replication [34], and c) transaminase peaks occurs ~ 2weeks later.
2.1 The replicative kinetic paradox
This has been described in detail previously, and relates to the replicative kinetics of HCV, HIV and HBV [28] and other viruses causing persistent infection. Briefly, and specifically for HCV, if immune functions are responsible for falling viral replication seen between point A to point B (figure 2), then the immunological clearance forces at point A must exceed the viral expansive forces (proposition 1). At points B to D (or any point between), where equilibrium develops, immune and viral forces must be equal, by definition (proposition 2). As viral concentration and, therefore, viral forces fall between points A and B to D by 102–3 geq/ml (observation 1), the immune forces must also fall by >102–3 between A and B to D for equilibrium to develop (proposition 3). There is no evidence this occurs, and very considerable evidence that immune force(s), as judged by development of specific cytotoxic T cell and antibody responses, are increasing during this time [29] (observation 2, proposition 4). Antecedent propositions (1–3) and (observation 2, proposition 4) are self-contradictory and incompatible with the conclusive belief that immune responses cause HCV replication to fall, hence either (a); the well-documented and multiply repeated observations of viral kinetics and adaptive immune responses are incorrect or (b); falling HCV replication beginning week 10 is not caused by host factors. Simply put, if immune or other host defences are able to clear virus at point A, why should they falter at B when less then 1% of initial viral load and antigenic diversity remain?
Figure 2 Paradoxical HCV replication kinetics. If host immune clearance forces (Ic, black arrows) reduce viral replication acutely (point A), then they must exceed viral expansive forces (Ve, grey arrows) at that point. At equilibrium (e.g. points B through D), viral concentrations (—) and, therefore, viral forces, have fallen by 102–3 hence, immune forces Ic must fall by >102–3 from A to B for equilibrium to develop. There is no evidence this happens.
2.2 Temporal tissue injury (aminotransferase) paradox
Both HBV and HCV are non-cytolytic and viral clearance from hepatocytes, as well as hepatocyte injury, thought to be immune mediated. However, for both HBV and HCV the brisk fall in viral replication following acute infection precedes the peak of transaminase rise by at least two weeks (figure 1). If falling viral replication is due to adaptive immune responses causing hepatocyte lysis the transaminase peak should either precede or be coincident with falling replication. This temporal relationship is also inconsistent with the belief immune factors cause the falling replication seen during acute HCV or HBV, and is analagous to non-cytolytic reductions of viral replication observed for both HBV and lymphocytic choriomeningitis virus (LCV) experimentally, that suggested either [unspecified] antiviral mechanisms are operative [30,31], or that auto-inhibition of RNApol by viral mechanisms (replicative homeostasis) occurs [28]. However, if other non-cytopathic host anti-viral mechanism(s) are responsible, the kinetic paradox implies their potency falls significantly between points A and B.
2.3 The Hepatitis C "early replication" paradox
Hepatitis C replication kinetics and their relationship to immune responses are well documented [32,33] but reveal an unexplained paradox. Despite high level viral replication, adaptive cellular immune responses to HCV are completely undetectable for at least 7–10 weeks [33] after infection, while humoral responses are rarely detected before 12–14 weeks [34], and in some patients [35], and some chimpanzees [36], are never detected at all. An exhaustive and very careful review of the clinical and experimental data relating adaptive immune response and HCV replication kinetics has been published recently [29]. Seeking to rationalize the enigma posed by a complete lack of immune responses to HCV replication of ~106–7 geq/ml at week 6 but [variable] immune responses to replication at ~105 geq/ml after week 14, the authors conclude "..[the data]...appear[s] to be consistent with the interpretation that HBV and HCV are ignored by the adaptive immune system for about 2 months after primary infection" and "[in HCV].. the adaptive response seems to really ignore for several weeks a substantial quantity of virus (at least 106 copies/ml)..". This is certainly an accurate synthesis of an extensive and highly complex literature but does it make any sense?
If adaptive immune responses really ignore high level HCV replication for two months, as suggested, then the following mechanism(s) are implied: a) an accurate mechanism for prompt detection of infection; b) A timing mechanism; c) A trigger mechanism for immune responses independent of any viral factor (given levels of virus are greater before immune recognition than afterwards the trigger for immune response must be either non-viral or falling (!) viraemia); and, as cytomegalovirus (CMV)-specific CD4(+) T cell responses arise within 7 days of CMV infection [37]; d) A mechanism allowing the immune system to differentiate HCV from CMV and other viruses (and reasons to do so). While possible, this seems unusually inelegant and pointlessly counterproductive, especially as events soon after infection probably determine whether virus is cleared or chronic infection develops. It is much more likely that adaptive cellular or humoral immune responses do not develop in the first 6–7 weeks of HCV infection simply because the virus isn't "seen". Why should HCV replicating at 106–7 geq/ml at week 6 be invisible to the immune system but visible when replicating at 105 geq/ml long term? Dissection of this problem requires explicit analysis of what is being measured and how.
3.1 Hepatitis C: measurement and detection
Assay of HCV RNA and detection of HCV by immune responses measure two quite different things. Quantitation of HCV is typically performed by branch-chain cDNA assay (bDNA) or quantitative PCR (qPCR) using probes or primers complementary to conserved 5'untranslated (5'UTR) HCV RNA sequences. Immune responses to HCV typically "measures" envelope proteins translated from envelope-encoding RNA (EeRNA) sequences and are directed at specific antigenic amino acid sequences and polypeptide conformations, not total viral envelope protein concentrations. While concentrations of 5'UTR RNA will be proportional to EeRNA concentrations in any given sample, they may not be identical for two reasons; i) RNA transcription may prematurely terminate making 5'UTR RNAs relatively more prevalent than EeRNAs and ii) HCV 5'UTR is highly conserved, while EeRNA s are less constrained, making hybridization efficiencies of PCR primers or bDNA probes greater for 5'UTR RNAs than for the population of EeRNAs, causing relative under-estimation of true envelope RNA concentration1. Nonetheless, as 5'UTR HCV RNA concentrations will be proportional to EeRNA concentration, the question remains; why should envelope proteins translated from EeRNA sequences present at concentrations corresponding to ~105 5'UTR geq/ml at 16 weeks be visible immunologically, but envelope proteins derived from EeRNA sequences corresponding to ~106–7 5'UTR geq/ml at 4–6 weeks remain unseen? Quasispecies biology, specifically variable RNApol fidelity, replicative homeostasis, and sequence-specific requirements for both genetic and immunological detection suggest an answer.
4.0 Quasispecies biology: Generation of genomic and phenotypic diversity
RNA viruses replicate by copying antigenomic templates, a process catalysed by RNApol, an enzyme lacking fidelity or proof reading function [38-41]. Theoretically, an RNA viral genome like HCV (about 9200 bases) could assume any of 49200 (about 8.95 × 105538) possible sequence combinations exceeding, by some margin, population estimates of protons in the known universe (about 1080), meaning the potential complexity of RNA viral quasispecies is infinite, for all practical purposes. An RNApol fidelity rate of 10-5 errors per base copied predicts at least one and as many as 10 (estimated for HIV) [39] genomic mutations will be introduced during each cycle of replication. Furthermore, as HCV replication results in synthesis of ~1012 virions per person per day [8], on average, mutations will develop at each genomic locus ~107 times/day, while the probability any two genomes synthesized consecutively will be identical is about 10-6. The sum effect is inexorable accumulation of genomic mutations – that, by itself, should threaten replicative fitness because of Muller's ratchet [42] – and progressive dilution of wild-type genomes (figure 3), processes that make long-term stability of RNA virus quasispecies highly paradoxical [43]. As argued previously, a combination of selective genomic replication and variable RNApol fidelity, both mediated by replicative homeostasis, act together to prevent RNA quasispecies extinction [28].
Figure 3 Simplified, two dimensional clade diagram of hyperdimensional viral RNA protein sequence-space. Because of RNApol (P) infidelity and Müller's ratchet, mutations () are introduced into each RNA template synthesized, and progressively accumulate, resulting in an RNA quasispecies with sequence progressively divergent from consensus sequence. Translation results in a spectrum of proteins (, , , etc.) with properties that also vary progressively from wild-type sequence () to highly variant proteins (, , etc.). Some RNAs will be so abnormal that translation or replication fails or is truncated (), while others will code for grossly defective proteins ( , etc.).
The phenotypic consequences of viral quasispecies biology may be more important. Progressive divergence of genomic RNA sequences away from wild-type sequences caused by RNApol infidelity generates a massive population of closely related, but genetically distinct, RNA molecules (figure 3), an effect operative at all scales from each open reading frame (ORF) to whole virus species. A quasispecies of ORF RNAs has but one inevitable outcome; translation of a quasispecies of viral proteins with a vast and highly variable spectrum of phenotypes, some subtly nuanced, others grossly defective. Furthermore, mutations that create new, or obliterate pre-existing, start or stop codons in a significant proportion of RNAs, will cause translation of highly unusual and heterogeneous proteins, particularly during high-level viral replication, a phenomenon that may explain HBeAg. Viral quasispecies cannot, and will not, produce homogeneous proteins with predictable and consistent phenotypic and antigenic properties.
4.1 Quasispecies biology: Frequency distribution of genomic and phenotypic diversity
While RNApol infidelity will cause progressive divergence of copied sequences away from wild-type or consensus sequences, the probability of any particular sequence arising will fall dramatically with increasing genetic distance from that consensus sequence (figure 4), allowing conceptual representation of the resulting genomic (and consequent phenotypic) diversity as a frequency distribution curve, with increasingly variant sequences surrounding a 'centre of gravity of replication', formed by wild-type sequences. Viral quasispecies occupy hyperdimensional sequence-spaces, hence any physical representation is necessarily simplified, but because mutation away from wild-type sequences is equally probable in all directions, variant RNA and protein frequencies will be normally distributed and the standard deviation (SD, σ) – insofar as 'normal' or 'standard' can be applied to a hyperdimensional space – of that distribution will be a function of RNApol fidelity; if RNApol is completely faithful, the RNAs and proteins will be monoclonal and σ = 0; if RNApol has no fidelity, RNA will be synthesised randomly, and all RNA and consequent protein sequences will arise with equally probability, therefore σ = ∞. While viral RNA and related protein sequences are theoretically unconstrained (at least before any consideration of functionality), the sequence specificities of any reagents used in their detection (bDNA probes, PCR primers, mAbs etc) are not, by definition, and their specificity and the efficiency with which they detect variant molecules will fall progressively the further those variant sequences are from the consensus sequence. A zone of 'reagent specificity' may therefore be defined probably encompassing wild type and some variant sequences, but there will exist some RNA sequences and corresponding proteins of any quasispecies that are undetectable with these sequence-specific reagents. A threshold of detection of any assay (including immune detection) may similarly be defined; RNA or protein sequences present at concentrations below this conceptual level being undetectable by that particular assay. The HCV "early replication" paradox now partially resolves; the 5'UTR sequences are both highly conserved and common to virtually all RNAs in the quasispecies, therefore, the 5'UTR concentration – that is, the common measure of HCV viraemia – corresponds to the area under the frequency distribution curve. By contrast, envelope RNA sequences (and related envelope proteins) are not so constrained and their relevant concentrations (i.e. whether or not that RNA or protein sequence is detectable) corresponds to the frequency of that specific sequence in the quasispecies and that, in turn, depends on RNApol fidelity; if RNApol fidelity is low, the frequency or concentration of any particular RNA or protein sequence will also be low and may be below the detection threshold, while increasing RNApol fidelity may increase sequence frequency [i.e. the concentration of specific proteins] above detection threshold. But why should specific EeRNA sequence frequencies – in other words, HCV RNApol fidelity – increase after week 8, facilitating adaptive immune responses? Viral autoregulation, specifically replicative homeostasis, provides an answer.
Figure 4 Two-dimensional representation of hyperdimensional RNA (or corresponding protein) frequency distribution curve (scale arbitrary) with conceptual centre of gravity of replication (wild type, green) and variant sequences (blue), zone of reagent specificity (red shading) and threshold of detection (TOD) of any assay. As mutations ( , ) accumulate and RNA sequence progressively diverges from consensus sequence (0) the probability of that RNA sequence and corresponding protein (e.g. envelope, Env.) arising falls rapidly. Standard deviation (σ) of frequency distribution is proportional to RNApol fidelity.
5.0 Co-evolutionary adaptation
Interactions among species, whether between humming birds and flowering plants, primitive viroids and prokaryotic cells or HCV and man, results in an unremitting process of adaptation and responsive counter-adaptation – in effect, a molecular arms race – for each species just to maintain ecological parity. The price of survival for a species is continual evolution. Survival, for viruses, requires cell entry, a precondition long antedating necessity to evade more complex host defenses, including interferons and other cytokines and adaptive immune responses, while for cells, and complex cellular organisms, cell wall defenses, including receptor polymorphisms, form a principal barrier against viral invasion. Viral survival – effectively meaning RNApol survival – on an evolutionary timescale, as argued previously [28,44], requires control of mutation and replication rates in a manner adaptively responsive to constantly changing biota and this implies dynamic linkage of RNApol fidelity and processivity with quasispecies phenotypic and antigenic diversity, meaning an autoregulatory linkage – Replicative homeostasis – between RNApol fidelity and processivity and envelope proteins, as argued previously [28]. By definition, evolutionary co-adaptation occurs in response to adaptations in locally prevalent interacting species. Natural selection for beak variation(s) in Darwin's finches occurs as a consequence of concrete survival benefits these variations – mediating, for example, enhanced food harvesting interactions with other variable plant or animal species – confer to individual Galapagos Island birds, rather than any inexorable hypothetical 'improvement' in beak function for finches in general. If a species is widely distributed in space, but population mixing is slow or incomplete, locally prevalent interactions with other species will vary and regional genetic variations will arise and be maintained, hence progressive divergence from the original genotype (speciation) may result. For viruses, and their hosts, genetic variations – reflected in viral genotype and cell surface polymorphisms and resulting disease susceptibilities – would be predicted, and are observed [45-50], to have frequencies that vary geographically.
5.1 Enzymatic Autoregulation
Consider the following; An enzyme (E) functioning in a closed system synthesizes either product A or B that both interact with E to influence output such that A:E interactions cause production of B, while B:E interactions produce A. Irrespective of starting conditions (excluding substrate exhaustion and product inhibition), an equilibrium will eventually develop (Figure 5) with the relative concentrations of A:B determined by the relative association constants (K) of A:E (KA:E) and B:E (KA:B) and the velocity (ν) of production of A from B:E (νA) and B from A:E (νB). Removal or addition of either A or B will alter equilibrium conditions but not the fact equilibrium is reached; if A is removed, for example, the increased frequency of B:E interactions will cause compensatory increased A synthesis; in this sense enzymatic autoregulation occurs. Intuitive analysis suggests that enzymes acting in a milieu of increasing concentrations of inhibitory molecules become progressively less processive until reduced enzyme output is insufficient to further inhibit enzyme activity, and an equilibrium state is reached. Considering viral replication, if alteration of RNApol fidelity causes synthesis of either wild-type or variant RNA sequences (simplified, as a continuum between these two must exist) that are subsequently translated into either wild-type or variant polypeptides that then interact with RNApol such that wild-type: RNApol are high affinity interactions that induce rapid, low fidelity RNApol replication while variant protein: RNApol interactions are low affinity and cause high fidelity RNApol replication at low rate then an equilibrium will eventually develop. Hence, as relative concentrations of wild-type and variant viral proteins vary, alteration of both processivity and fidelity of RNApol results, permitting viruses to adaptively respond to environmental changes, including immune recognition and reaction to evolving cell receptors. Stable, highly reactive equilibria not only develop as a result of RNApol/envelope interactions and viral autoregulation, there is no option but for this to occur.
Figure 5 Autoregulation of a simple enzyme system: If enzyme E produces either A () or B () and product:enzyme interactions occur such that A:E produce B while B:E favour A, then high initial concentrations of A (or B) will cause rapid synthesis of B (or A). Equilibrium ultimately develops irrespective of starting conditions.
5.2 Co-evolutionary adaptation: Cell-surface polymorphisms
Generation and maintenance of polymorphisms, that is, replacement of existing genes – that, by operational Darwinian definition, have proved their functionality and evolutionary fitness by surviving to reproduce – with variant genes (polymorphisms) of uncertain functionality, fitness or overall compatibility within an organism, is an evolutionary strategy that will only be sustained on a geological timescale if new polymorphisms confer survival benefits to organisms that exceeds the risks and metabolic costs of generating and sustaining those polymorphisms. For primitive cells, lacking functional humoral, cellular or cytokine defense mechanisms, development of cell-surface protein polymorphisms is an obvious adaptive strategy to thwart invasion by primitive viruses. Like other adaptive strategies, cell-surface polymorphisms are strongly selected for, and have been highly conserved over deep time, and are found in all organisms from primitive prokaryotic cells [51] and thermophilic bacteria [52] through to plants [53] as well as mammalian cells, strongly suggesting a critical evolutionary function. The lock and key hypothesis, for which there is very considerable evidence [54-57], first proposed by JBS Haldane [58], contends polymorphisms arise, and are maintained, as protection against cellular parasitism, particularly by viruses2. While DNA-encoded protein polymorphisms form necessary defenses against viral access, they may not be sufficient; a quasispecies of cells (e.g. the liver) expressing similar and static receptor variations renders those cells vulnerable to sustained attack from any virus that successfully invades any one cell, and further dynamic modification of cell receptors, triggered by viral infection and mediated at the transcriptional level by modulation of DNA dependent RNA polymerase fidelity in nearby uninfected cells, by a mechanism similar to replicative homeostasis would seem possible.
6.0 Problems of Detection
A clear, unambiguous band at the "C" position on a sequencing gel, causes "cytosine" to be assigned to that genetic locus. But does this certitude reflect reality, at least for viral RNA quasispecies? Direct PCR sequencing is an "averaging" procedure revealing the most frequent nucleotide at any particular locus. However, nucleic acids and proteins cannot express 'an average', and discrete quanta of specific nucleotides or amino acids are present at every locus. A typical clinical serum sample, containing 4 × 105 geq/ml HCV and mutating at 10-5 substitutions/base, will contain examples of each possible nucleotide at every locus, but most variations will remain undetected during sequencing or any other method of quasispecies analysis. Analysis of cloned DNA gives cleaner data than PCR sequencing but if 100 clones (and multiple HCV quasispecies clones are highly unlikely to be identical) provides definitive sequence, would we process the 101st to reveal different and, potentially, critical sequence variations? And if we did, how would we recognise its importance? Is important sequence likely to be present at frequencies of < 1%? Infectious virions containing, presumably, full-length functional genome and corresponding wild-type proteins, are often outnumbered by ~6 × 104:1 in serum by defective and non-infectious particles [53] that presumably do not, suggesting that important genetic sequence and associated phenotype may occasionally be extremely rare. How the immune system recognizes uncommon, nondescript, but important protein sequences in a featureless background of similar molecules is a non-trivial problem for which replicative homeostasis may suggest a solution.
7.0 Replicative Homeostasis
Replicative homeostasis, described in detail elsewhere [28,44], is an epicyclic mechanism of viral autoregulation that results when viral proteins, notably envelope (Env), influence RNApol fidelity and processivity. The predicted consequences of replicative homeostasis for rates of intracellular viral replication and mutation, cellular expression of viral proteins and immunological responses occurring because of replicative homeostasis is represented schematically (figures 6, 7). During early viral replication in a naive cell devoid of inhibitory molecules (panel A, a), high affinity wild- type envelope:polymerase interactions predominate, causing rapid low-fidelity polymerase activity resulting in rapid synthesis of variant viral RNAs and subsequently proteins, hence causing a broad spectrum of viral proteins to be expressed on the cell surface, each at concentrations below the threshold of immune detection (TOD). RNApol infidelity ensures synthesis of variant viral RNAs and proteins predominates early, hence variant protein molecules progressively accumulate within cells relative to wild-type viral molecules (Panels B-D) and increasing the probability of variant viral envelope:RNApol interactions. Variant viral envelope:RNApol interactions causing progressive inhibition of RNA polymerase processivity and increasing RNApol fidelity, reducing diversity of viral RNAs synthesized and progressively restricting viral protein diversity expressed on the cell surface (panels b to d), increasing cell-surface concentrations of individual viral proteins above the threshold of detection (panels C, c) at which point a polyclonal immune response develops. Development of low-affinity polyclonal blocking antibodies, restricting cellular egress of viral proteins, further increasing intracellular concentrations of variant envelope proteins, still further increasing the probability of variant viral envelope:RNApol interactions and inexorably further restricting antigenic diversity increasing relative expression of wild-type proteins thus further exposing these epitopes to immune surveillance and facilitating specific high-affinity immune responses, including cytotoxic T cell responses, (D,d) to wild-type proteins. Thus, the immune responses can strategically utilize replicative homeostasis to force viruses to reveal important and dominant wild-type epitopes, but those responses develop initially as a consequence of restriction of RNApol fidelity that occur because of replicative homeostasis. High-affinity responses further deplete intracellular concentrations of wild-type proteins, progressively reducing wild-type envelope:RNApol interactions, greatly reducing RNApol processivity to the point of viral latency (E,e), caused by variant viral envelope:RNApol interactions.
Figure 6 Dynamic progression of RNApol functional properties, processivity () and fidelity () predicted by replicative homeostasis. Initial state (A, corresponding to panel A, Figure 7): in a newly infected cell, high-affinity wild-type:RNApol interactions will predominate resulting in high RNApol processivity but low fidelity causing high-level viraemia with broad virus phenotypic spectrum, maximizing cell tropism. Intracellular accumulation of variant viral proteins (B, c.f. panel B, Figure 7) reduces RNApol processivity but increases fidelity reducing viral RNA synthesis and consequently, viraemia before a dynamic, fluctuating equilibrium (C, c.f. panel C or D, Figure 7) develops in which inhibition of RNApol by variant viral proteins is balanced by increases in RNApol fidelity (with consequent synthesis of wild-type viral products tending to cause high RNApol processivity).
Figure 7 Conceptual progression of intracellular viral replication events, including variable RNApol fidelity and processivity, restriction of antigenic diversity and immune recognition under influence of Replicative homeostasis. Panels (A->E) changing frequency distribution of viral RNA and protein quasispecies, panels (a->e) cellular events. Initial state (panels A,a) viral replication occurring in cells devoid of molecular inhibitors of RNApol high affinity wild-type envelope (Enve, green): RNApol interactions predominate, causing rapid low-fidelity viral RNA synthesis and, consequently, a broad spectrum of viral proteins expressed on cell surface at concentrations below TOD. As variant viral proteins accumulate within cells (panel b) and variant viral envelope: RNApol interactions increase, RNApol fidelity increases while processivity decreases, restricting the distribution of viral RNA and proteins, reducing antigenic display on cells. As variant viral envelope: RNApol predominate (panel c), the frequency distribution of expressed viral proteins is restricted so the individual concentration of some proteins increases beyond TOD, allowing immune recognition and polyclonal, low affinity antibodies to develop, blocking cellular egress of viral proteins, further increasing variant viral envelope: RNApol interactions, thus immune responses force viruses to reveal wild-type epitopes by restricting antigenic diversity. High affinity responses once developed (panel d) preferentially reduce intracellular concentration of wild-type viral proteins further increasing variant viral envelope: RNApol interactions still further restricting RNApol processivity to the point of viral latency (panel e).
8.0 Discussion
The hepatitis C "early replication" paradox now resolves completely when considered in the context of replicative homeostasis; initial high level HCV replication (due to high RNApol processivity) remains immunologically undetectable for 6–8 weeks, or more, because of low RNApol fidelity causing a broad spectrum of HCV envelope proteins each expressed on cell surfaces at concentrations below the threshold of detection even while viraemia, reflected in concentrations of 5'UTR RNA common to each RNA species, are present at 106–7 geq/ml. As replication progresses, intracellular accumulation of variant viral proteins increase RNApol fidelity but decrease processivity (replicative homeostasis), downregulating HCV replication and reducing viraemia but restricting antigenic diversity and increasing expression of HCV envelope proteins to beyond the threshold of immune detection. Furthermore, the temporal tissue injury (aminotransferase) paradox also resolves in this light: Focussed immune recognition (including cytotoxic T cell responses) doesn't develop until after viral antigenic diversity is restricted by replictive homeostasis the transaminase peak would not be expected until after viral replication falls due to autoinhibition of RNApol processivity. Varying expression of viral proteins by modulating RNApol fidelity to facilitate immune escape would seem a useful evolutionary adaptation that might be retained by more complex organisms, including cellular pathogens like tuberculosis and malaria, to optimize their stability within hosts.
This mechanism of immune avoidance might also explain maternal-foetal tolerance. The human foetus maintains a stable parasitic existence during gestation (and, I expect, to University age and beyond) that is tolerated despite normal maternal immune responsiveness in general and lack of specific tolerance to paternal antigens in particular, a situation made more problematic as expressed foetal antigens are predominantly of paternal origin [54]. While immunological isolation of foetal tissue by the placental trophoblastic layer [55], and placental display of HLA-G [56], probably contribute to foetal stability in the face of a potentially robust immune attack, neither mechanism would explain persistence of viable foetal nucleated red blood cells within the maternal circulation [57] in quantities sufficient to permit clinical prenatal diagnosis [58]. Is it possible foetal tolerance is mediated by regulating the fidelity of foetal DNA dependent RNA transcriptases to ensure any cell-surface antigens are expressed heterogeneously and at levels below the threshold of maternal immune responsiveness?
9.0 Autoimmunity
For many classical autoimmune disorders, including primary biliary cirrhosis [59], multiple sclerosis, and rheumatoid arthritis, convincing epidemiological evidence [60], including cases clustering [61,62], strongly suggests these diseases are triggered by infectious agents in genetically predisposed individuals. In others, such as diabetes mellitus, tantalizing epidemiological [63], clinical [64] and laboratory [65] evidence has implicated enteroviruses, but has suggested viral-triggered autoimmune processes, rather than cytolytic destruction of pancreatic beta-cells [66]. Similar circumstantial evidence exists for myocarditis, demyelinating diseases, myositis and other post infectious inflammatory disorders. When MacFarlane Burnet wrote autoimmunity arises from "inability to distinguish self from non-self" HBV, HCV, HIV and other viruses, now established to cause diseases with clear autoimmune features were unknown. Viral infections, particularly hepatitis C – and its treatment with interferon – are associated with many varied autoimmune phenomena [67], and thyroid disease [68-70], diabetes mellitus [71,72], membranous, membranoproliferative and cryoglobulinemic glomerulonephritis, vasculitis and peripheral neuropathy [73], and autoimmune gastritis [74] are all very well documented, although the mechanism(s) are unknown and causality is certain. Classical serological markers of autoimmunity, including rheumatoid factor, antinuclear antibodies (ANA), anticardiolipin, antithyroid, anti-liver/kidney/microsomal antibodies (anti-LKM), as well as HCV/anti-HCV immune complex formation and mixed essential cryoglobulinemia are common accompaniments of chronic HCV infection [73], raising the obvious question of whether all "autoimmunity" has a viral basis. Indeed, Zinkernagel's pragmatic and subtly anticipatory; "If we know the infection, we call the disease immunopathologically mediated; if we do not recognize or know it, we call the disease autoimmune [75]" fully reflects recent explosive growth of information and the deeper questions this information poses.
10.0 Virus receptor disease
RNA virus quasispecies biology, specifically the generation of RNA quasispecies by RNApol, and translation of these immensely variable RNAs into protein quasispecies, suggests an immediate solution to the problem of viral autoimmunity and, by extension, to autoimmunity in general, as well as suggesting a unifying hypothesis to explain other diseases known to have multi-factorial aetiologies that include inflammatory components – such as coronary artery disease – in addition to other diseases – including schizophrenia and some forms of depression – that currently lack rational and coherent pathogenic explanations.
Viruses are known to co-opt cell surface molecules, including lectins, hormone receptors and cell signaling molecules, to access cells. Receptors, and other cell surface molecules, identified as "viral receptors"or to specifically interact with viral proteins include prostaglandins, catecholamines and acetylcholine receptors [76], serotonergic neurotransmitters (5HT) [77], endothelial cell glycoproteins [78], insulin-like growth factor (IGF-IR) and its major signaling molecules insulin receptor substrates IRS-1 [79] and IRS-2 [80], epidermal growth factor (EGF) [81], neurotrophin receptor [82], thyroid hormone receptor TRalpha1 [83], an immunoglobulin protein superfamily [84], low density lipoprotein (LDL) receptors [85,86], transferrin receptor (TfR) [87], asialoglycoprotein receptor (ASGP-R) [88,89], and angiotensin-converting enzyme 2 [90], to cite biologically diverse examples. Of necessity, some receptor affinity studies have used cloned viral protein ligands, an artificial situation that cannot approach the phenotypic complexity of RNA viral protein quasispecies. Nonetheless, variable virus receptor affinities [91,92], evolutionary adaptation of receptor affinity [93], emergence of escape variants with altered receptor affinities [94], temporal alteration of receptor usage [92] and capacity to exploit alternative entry pathways [95] have all been confirmed, suggesting viruses are capable of generating highly plastic ligands with very broad receptor affinities.
If a virus co-opts a receptor for cell entry, then wild-type envelope (consensus sequence) epitopes, coded for by wild-type RNA sequences, will probably form the common viral ligand. However, any viral RNA quasispecies also contain a vast spectrum of RNAs derived from, and similar to, envelope open reading frame (ORF) consensus sequence, but variant from it. As the envelope ORF quasispecies sequences progressively diverge from wild-type, the quasispecies of envelope proteins translated from these variant ORFs will also, and inexorably, diverge in sequence, structure and biological function from wild-type envelope sequence proteins. Some of these envelope proteins will be functionally identical, but others, and probably the vast majority, will range from subtly different to grossly abnormal, either due to major differences of sequence and/or chemical or steric amino acid incompatibility, or because of premature introduction of stop codons. Even minor amino acid differences, as sickle cell anaemia illustrates, and has been confirmed specifically for viral receptor usage [96,97], may catastrophically alter a proteins' function with respect to co-opted viral receptors, with some having no binding affinity, while others will bind strongly and act as agonists, antagonists or competitive inhibitors of normal receptor function. Variant and defective viruses, and their polypeptides, will be in vast molar excess compared to wild-type [53] but will exhibit similarly high antigenic variability, permitting escape from immune and other scavenger mechanisms. As many variant viral polypeptides will bind tightly to "self" receptors, but contain immunogenic non-self motifs, a polymorphic (because variant viral proteins will themselves be highly polymorphic due to the quasispecies process) immune response, apparently directed against "self" antigens, but actually targeting virus protein-receptor complexes virtually indistinguishable from normal cell receptors, will result causing apparent 'autoimmune' tissue damage.
This mechanism suggests an explanation for common autoimmune phenomena. If a virus enters cells because wild-type envelope motifs interact with insulin, insulin receptor substrate [79,80], TSH or related molecules [83], or acetylcholine [76] receptors, many variant envelope polypeptides, generated by envelope ORF quasispecies RNAs, would have similar receptor binding affinity, but may effectively disrupt receptor function, predictably causing impaired glucose tolerance or diabetes mellitus, thyroid dysfunction, or myasthenia gravis with secondary resistance to, and elevation of, the normal hormone ligand (insulin, TSH etc.). The expected consequences disruption of receptor function by variant viral proteins might explain many common biochemical pathologies; For example, what effect would chronic blockade of parathyroid (PTH) receptors by viral proteins have on PTH levels, the parathyroid glands, or bone?
Leptin is a 16Kda protein hormone secreted by adipocytes and carried across the blood-brain barrier by a rate-limiting transporter to act on hypothalamic receptors [98] where, among other functions, it regulates thyrotropin-releasing hormone (trh) genes and upregulates alpha-melanocyte-stimulating hormone and other anorexigenic neuropeptides [99] important to appetite-regulation and energy balance [100]. Leptin also regulates a broad spectrum of other processes and behaviours including thermogenesis, blood pressure and immune function. s=Serum leptin concentrations and leptin resistance, are independent markers of obesity, weight gain, systemic hypertension [101], diabetes mellitus [102], obstructive sleep apnoea [103] and myocardial infarction [104], while polymorphisms of the leptin gene are associated with insulin resistance [105] and long-term risk of developing diabetes mellitus [102]. Predictably, variant envelope proteins generated by envelope ORF RNA quasispecies from viruses utilizing leptin receptors for cell access would have similar receptor affinity, but exhibit non-physiological leptin antagonist or agonist properties, thus disrupting leptin receptor function, altering energy regulation, and causing either excess caloric intake unrestrained by satiety responses, or inappropriate satiety signals with pathologically reduced caloric intake. As clear evidence exists for viral disruption of leptin function [106] and virus-associated weight gain in humans [107] and monkeys [108], is it possible the global epidemics of type II diabetes mellitus, insulin resistance, hyperlipidaemia and obesity now prevalent [109-116], are just that; epidemics fundamentally caused by viruses that co-opt insulin or leptin or other associated receptors for cell access and generate protein quasispecies that disrupt receptor function? Could it also be that ethnically based epidemics of obesity, diabetes mellitus, hypertension and reno-vascular disease (the 'metabolic syndrome'), as seen in PIMA Indians, Nauruans and Australian Aborigines [115] have developed not primarily because of exposure to "Western" foods and lifestyles – that, after all, are all-pervasive without necessarily having so dramatic an effect on other groups – but because of chronic or recurrent exposure to viruses, or genotypes of viruses to which their particular repertoire of receptor polymorphisms confer no protection? Or that anorexia nervosa develops, in some patients, when variant viral proteins with aberrant leptin-agonist function arise during the course of viral infection, as the temporal relationship between infection and disease onset, very clearly documented in one study [117], suggests.
Cardiovascular disease, the leading cause of premature death and disability in most western countries, has a well-established multi-factorial basis involving a complex interplay between genetic predisposition, environmental and personal risk factors – including systemic hypertension, diabetes mellitus, hyperlipidaemia, obesity and cigarette smoking – and more recently recognized mechanisms, including endothelial dysfunction [118], vascular inflammation [119] and leptin levels [104]. Systemic hypertension, diabetes mellitus and hyperlipidaemia have long-established, but complex, patterns of inheritance, a situation further compounded by evidence receptor polymorphisms – including those of angiotensin II type 1 receptor [120], IRS-1 gene [121] and low density lipoprotein receptor (LDLR) [122] – both confer disease susceptibility and have regionally variable prevalences [123,124].
The flaviviradae – including HCV – as a family, and the rous sarcoma virus, utilize low density lipoprotein receptors to enter cells [85,125], while angiotensin II [90], insulin receptor substrates (IRS1 and IRS 2) [79], and endothelial cell glycoproteins [78] and other receptors widely distributed in vascular tissues are known to be permissive for virus cell entry establishing, in principle and in fact, viral-protein receptor affinity relevant to cardiovascular diseases. Viruses accessing cells through these receptors will generate a quasispecies of variant proteins capable of disrupting receptor function potentially causing hyperlipidaemia, hypertension, hyperglycaemia and endothelial dysfunction, as well as immune-mediated endothelial cell damage, thus establishing the necessary and sufficient conditions and a chain of events that potentially link viruses and vascular diseases, including myocardial infarction. This hypothesis exists at the confluence of established risk factors for coronary artery disease, including genetic susceptibility, polymorphisms predisposing to hypertension [126-128], diabetes [126] and hypercholesterolaemia and substantial new data implicating vascular inflammation [119,129], endothelial dysfunction [119,130], leptin dysregulation [104] and viral infection [131,132] in the pathogenesis of vascular disease. Furthermore, this final common pathway can account for that small, but significant, group of patients with vascular diseases but no clinically identifiable risk factors, as well as the non-random co-incidence of depression and coronary artery disease [133] (as discussed below) in addition to the anti-inflammatory action of HMG-CoA reductase inhibitors (statins) [134], and their effect in lowering cardiovascular mortality independent of cholesterol reduction [135]; if statins compete with variant viral proteins for HMG-CoA reductase receptor binding, and displace immunologically attractive molecules, inflammatory responses directed at viral product, but involving endothelial cell receptors, will be ameliorated (figure 8).
Figure 8 Cell receptor (R) and normal ligand (L; insulin, PTH, leptin etc.) relationship (1; unbound, 2; activated), receptor permissive for virus cell entry (3) or blocked by polymorphism (Rp, 4). Receptor blockade by variant viral envelope proteins (green E, 5), blockade by antigenic envelope proteins stimulating "autoimmune"response apparently directed against self receptors (E, 6), competitive displacement of antigenic proteins by drug (D, e.g. statin, aspirin) abrogating immune response (7).
Human immunodeficiency virus HIV-associated dementia (HIVD) occurs in 15% of HIV-infected adult patients, and as a major cause of dementia in the young represents "proof of principle" of virus-caused dementia, raising the possibility other forms of virus related dementia exist. Although highly active antiretroviral therapy (HAART) has reduced the incidence of HIV-D by 40–50% [136], it remains a major cause of morbidity and the pathogenesis poorly understood. Direct cytopathic effects of HIV or other viruses are unlikely, while active replication of virus, high-level viral protein expression [137], and increased viral envelope sequence-diversity in blood and brain [138] are all important, clearly indicating viral proteins are pathogenically important. The clinical features of HIVD, including psychomotor slowing, apathy, and altered gait and posture, strongly suggest a subcortical dementia with involvement of the basal ganglia and striatal dopamine receptor pathways. Schizophrenia, depression and bipolar affective disorder, and anorexia nervosa are highly prevalent, chronic conditions of unknown aetiology that cause enormous morbidity and generate significant health care costs. Each of these disorders have well documented, albeit regionally variable, associations with receptor – including dopamine – polymorphisms [124,139-143], as well as epidemiological evidence that viral infections are aetiologically important, either directly or as precipitating events [117,144-147], although other sero-epidemiological studies [148] and work directly seeking viral nucleic acids in patients with schizophrenia have proved negative [149]. If a virus, or viruses, use dopamine, acetylcholine [76], neurotrophin [82], serotonergic (5-HT)[77], or other neuro-transmitter receptors to access cells (and, given RNA virus quasispecies biology, it would be surprising if some didn't), then the RNA quasispecies will generate a quasispecies of variant polypeptides potentially reactive to these receptors. While it is difficult to imagine what effect perfusing a functional human brain with a solution of antigenic, inflammatory polypeptides that bind to, and are variably disruptive of, critical neurotransmitter receptor function, might have on cognition, perception, behaviour, attention span, abstract thought, fine motor or emotional control, it is unlikely to be beneficial. In this context, the well-documented cognitive abnormalities – unrelated to depression – found in patients with early HCV and HIV infection [150-152] are unsurprising.
12.0 Virus Receptor Disease: Conclusions
Virus receptor disease (VRD) is quite distinct from either immune complex deposition disease due to deposition of macromolecules in tight vascular arcades, or from disease related to altered cell tropisms and is also completely independent of the primary site of viral replication; both non-inflammatory receptor blockade and immune-mediated inflammation directed at viral protein-receptor complexes could cause pathology of tissues non-permissive for and remote from the primary site(s) of viral replication with "autoimmune" damage to the liver, pancreas, brain, skin or lungs arising, for example, from chronic small intestinal virus infection. Viral quasispecies biology predicts VRD will have other characteristics. First, due to replicative homeostasis, the ratio of wild type to variant viral proteins of the quasispecies will both fluctuate with time and will alter dramatically after initial infection; if wild-type proteins are dominantly agonist in function with respect to their receptor, variant proteins, most likely, will predominantly exhibit antagonist function (and vice versa). Furthermore, the net effect of viral proteins (because of viral autoregulation) will fluctuate initially between receptor agonist and antagonist function, before becoming predominantly antagonistic, thus providing a possible explanation for transient thyrotoxicosis during early thyroiditis (before hypothyroidism supervenes), for hypoglycaemia seen during early insulin-receptor antibody-mediated insulin resistance [153], and for the contradictory functions ascribed to HIV nef [154]. A corollary of fluctuating phenotypic dominance of viral protein quasispecies is that receptor affinity of these proteins will also fluctuate, and any resulting inflammation may vary in both intensity and anatomical distribution over time. Second, because viruses utilize alternate receptors for cell access, apparently homogeneous disease processes could result from multiple different viruses. Similarly, because virus quasispecies produce a broad spectrum of protein phenotypes, and the receptor polymorphisms permissive for cell entry for specific viruses will be variably distributed in host populations, pathology of widely variable tissues in different individuals could result from the same virus. Third, as evolutionary co-adaptation results in progressive genetic co-divergence of interacting species, the receptor polymorphisms predisposing to (or protecting against) infection by any particular virus, and resulting VRD, and the common viruses causing them, would be predicted to vary geographically, an expectation multiply confirmed for disease associated polymorphisms. As a corollary this suggests individuals migrating from regions where hosts and virus strains are stably co-adaptated to other areas, where different viruses are prevalent, might experience increased rates of VRD – beaks optimally adapted for finch survival on the Galapagos may be a liability elsewhere – a prediction again amply confirmed [155-157];.
Finally, if immune mechanisms are unable to clear RNA viruses like HCV and do not cause the reduced viral replication seen during acute infection, are they any more likely to be effective against other RNA viruses? Is it possible that self-limiting infections like influenza and SARS also autoregulate their replication, and, like HCV or HBV, become partially dormant, yet remain transcriptionally active, in the face of an active and powerful immune response? PCR amplification of influenza RNA from convalescent samples makes this readily testable, while the documented relationship of influenza to myocardial infarction [132] and juvenile rheumatoid arthritis [61] makes the question important. If confirmed, the well-documented seasonality of some depressive illnesses [158] and schizophrenia, [146] and increased rates of schizophrenia during influenza epidemics [144], and the increased incidence of both depression [146] and schizophrenia [144,145] following in-utero exposure to influenza may be more rationally explained.
Footnotes
1. If quantitative PCR (qPCR) assays of both 5'UTR and envelope RNAs are performed serially, and data expressed as [5'UTR RNA]/[Env RNA] for each sample, then a numerical expression describing changing quasispecies complexity over time may be obtained.
2.In case prescient genius is unappreciated, Haldane formulated the "lock and key" hypothesis on the basis of protein polymorphisms, defined by gel electrophoresis, and some general musing about predation and evolutionary struggle, two decades before the nature of DNA was elucidated.
Conflict of interests
I have no pecuniary interests, whatever, in this work and do not stand to gain financially or otherwise from it.
Acknowledgements
I thank my wife Sophie J Coleman, and sons Matt and Tim, for everything important, my parents Dick and Janet for extraordinary opportunity, and some great physician-teachers – that most noble vocation – of the University of Western Australia Medical School; Professors Mike McCall, Dick Joske, Bill Reed, Bill Musk, Peter Pullan, Michael Quinlan, Dick Lefroy and Ted Haywood. Special thanks to Karl Ruckriegel for turning back-of-envelope sketches into first-class graphics. Any remaining lack of clarity is my fault.
==== Refs
Ni F Fang D Gan R Li Z Duan S Xu Z A new immune escape mutant of hepatitis B virus with an Asp to Ala substitution in aa144 of the envelope major protein Res Virol 1995 146 397 407 8834756 10.1016/0923-2516(96)80899-5
Lee KM Kim YS Ko YY Yoo BM Lee KJ Kim JH Hahm KB Cho SW Emergence of vaccine-induced escape mutant of hepatitis B virus with multiple surface gene mutations in a Korean child J Korean Med Sci 2001 16 359 362 11410701
Rehermann B Ferrari C Pasquinelli C Chisari FV The hepatitis B virus persists for decades after patients' recovery from acute viral hepatitis despite active maintenance of a cytotoxic T-lymphocyte response Nat Med 1996 2 1104 1108 8837608 10.1038/nm1096-1104
Michalak TI Pasquinelli C Guilhot S Chisari FV Hepatitis B virus persistence after recovery from acute viral hepatitis J Clin Invest 1994 93 230 239 8282792
Sanchez-Quijano A Jauregui JI Leal M Pineda JA Castilla A Abad MA Civeira MP Garcia de Pesquera F Prieto J Lissen E Hepatitis B virus occult infection in subjects with persistent isolated anti-HBc reactivity J Hepatol 1993 17 288 293 8315257 10.1016/S0168-8278(05)80207-7
Shiota G Oyama K Udagawa A Tanaka K Nomi T Kitamura A Tsutsumi A Noguchi N Takano Y Yashima K Occult hepatitis B virus infection in HBs antigen-negative hepatocellular carcinoma in a Japanese population: involvement of HBx and p53 J Med Virol 2000 62 151 158 11002243 10.1002/1096-9071(200010)62:2<151::AID-JMV5>3.0.CO;2-N
Perelson AS Neumann AU Markowitz M Leonard JM Ho DD HIV-1 dynamics in vivo: virion clearance rate, infected cell life-span, and viral generation time Science 1996 271 1582 1586 8599114
Neumann AU Lam NP Dahari H Gretch DR Wiley TE Layden TJ Perelson AS Hepatitis C viral dynamics in vivo and the antiviral efficacy of interferon-alpha therapy Science 1998 282 103 107 9756471 10.1126/science.282.5386.103
Lai ME Mazzoleni AP Argiolu F De Virgilis S Balestrieri A Purcell RH Cao A Farci P Hepatitis C virus in multiple episodes of acute hepatitis in polytransfused thalassaemic children Lancet 1994 343 388 390 7905553 10.1016/S0140-6736(94)91224-6
Farci P Alter HJ Govindarajan S Wong DC Engle R Lesniewski RR Mushahwar IK Desai SM Miller RH Ogata N Lack of protective immunity against reinfection with hepatitis C virus Science 1992 258 135 140 1279801
Kao JH Chen PJ Wang JT Yang PM Lai MY Wang TH Chen DS Superinfection by homotypic virus in hepatitis C virus carriers: studies on patients with post-transfusion hepatitis J Med Virol 1996 50 303 308 8950686 10.1002/(SICI)1096-9071(199612)50:4<303::AID-JMV4>3.0.CO;2-C
Kao JH Chen PJ Lai MY Chen DS Superinfection of heterologous hepatitis C virus in a patient with chronic type C hepatitis Gastroenterology 1993 105 583 587 8392958
Pham TN MacParland SA Mulrooney PM Cooksley H Naoumov NV Michalak TI Hepatitis C virus persistence after spontaneous or treatment-induced resolution of hepatitis C J Virol 2004 78 5867 5874 15140984 10.1128/JVI.78.11.5867-5874.2004
Radkowski M Gallegos-Orozco JF Jablonska J Colby TV Walewska-Zielecka B Kubicka J Wilkinson J Adair D Rakela J Laskus T Persistence of hepatitis C virus in patients successfully treated for chronic hepatitis C Hepatology 2004 41 106 114 15619235 10.1002/hep.20518
Forns X Payette PJ Ma X Satterfield W Eder G Mushahwar IK Govindarajan S Davis HL Emerson SU Purcell RH Vaccination of chimpanzees with plasmid DNA encoding the hepatitis C virus (HCV) envelope E2 protein modified the infection after challenge with homologous monoclonal HCV Hepatology 2000 32 618 625 10960458 10.1053/jhep.2000.9877
Bassett SE Thomas DL Brasky KM Lanford RE Viral persistence, antibody to E1 and E2, and hypervariable region 1 sequence stability in hepatitis C virus-inoculated chimpanzees J Virol 1999 73 1118 1126 9882313
Prince AM Brotman B Lee DH Ren L Moore BS Scheffel JW Significance of the anti-E2 response in self-limited and chronic hepatitis C virus infections in chimpanzees and in humans J Infect Dis 1999 180 987 991 10479122 10.1086/314973
Woitas RP Lechmann M Jung G Kaiser R Sauerbruch T Spengler U CD30 induction and cytokine profiles in hepatitis C virus core-specific peripheral blood T lymphocytes J Immunol 1997 159 1012 1018 9218624
Lechner F Wong DK Dunbar PR Chapman R Chung RT Dohrenwend P Robbins G Phillips R Klenerman P Walker BD Analysis of successful immune responses in persons infected with hepatitis C virus J Exp Med 2000 191 1499 1512 10790425 10.1084/jem.191.9.1499
Lauer GM Walker BD Hepatitis C virus infection N Engl J Med 2001 345 41 52 11439948 10.1056/NEJM200107053450107
Thomson M Nascimbeni M Havert MB Major M Gonzales S Alter H Feinstone SM Murthy KK Rehermann B Liang TJ The clearance of hepatitis C virus infection in chimpanzees may not necessarily correlate with the appearance of acquired immunity J Virol 2003 77 862 870 12502802 10.1128/JVI.77.2.862-870.2003
Rahman F Heller T Sobao Y Mizukoshi E Nascimbeni M Alter H Herrine S Hoofnagle J Liang TJ Rehermann B Effects of antiviral therapy on the cellular immune response in acute hepatitis C Hepatology 2004 40 87 97 15239090 10.1002/hep.20253
Sallie R 1a F Viral replication, immunological and tissue injury kinetics following acute HCV and HBV infection 2005
Negro F Pacchioni D Shimizu Y Miller RH Bussolati G Purcell RH Bonino F Detection of intrahepatic replication of hepatitis C virus RNA by in situ hybridization and comparison with histopathology Proc Natl Acad Sci U S A 1992 89 2247 2251 1312716
Fong TL Di Bisceglie AM Biswas R Waggoner JG Wilson L Claggett J Hoofnagle JH High levels of viral replication during acute hepatitis B infection predict progression to chronicity J Med Virol 1994 43 155 158 8083663
Berquist KR Peterson JM Murphy BL Ebert JW Maynard JE Purcell RH Hepatitis B antigens in serum and liver of chimpanzees acutely infected with hepatitis B virus Infect Immun 1975 12 602 605 1100525
Ribeiro RM Layden-Almer J Powers KA Layden TJ Perelson AS Dynamics of alanine aminotransferase during hepatitis C virus treatment Hepatology 2003 38 509 517 12883496 10.1053/jhep.2003.50344
Sallie R Replicative Homeostasis: A fundamental mechanism mediating selective viral replication and escape mutation Virol J 2005 2 10 15707489 10.1186/1743-422X-2-10
Bertoletti A Ferrari C Kinetics of the immune response during HBV and HCV infection Hepatology 2003 38 4 13 12829979 10.1053/jhep.2003.50310
Guidotti LG Rochford R Chung J Shapiro M Purcell R Chisari FV Viral clearance without destruction of infected cells during acute HBV infection Science 1999 284 825 829 10221919 10.1126/science.284.5415.825
Guidotti LG Borrow P Brown A McClary H Koch R Chisari FV Noncytopathic clearance of lymphocytic choriomeningitis virus from the hepatocyte J Exp Med 1999 189 1555 1564 10330434 10.1084/jem.189.10.1555
Major ME Mihalik K Fernandez J Seidman J Kleiner D Kolykhalov AA Rice CM Feinstone SM Long-term follow-up of chimpanzees inoculated with the first infectious clone for hepatitis C virus J Virol 1999 73 3317 3325 10074186
Thimme R Oldach D Chang KM Steiger C Ray SC Chisari FV Determinants of viral clearance and persistence during acute hepatitis C virus infection J Exp Med 2001 194 1395 1406 11714747 10.1084/jem.194.10.1395
Chen M Sallberg M Sonnerborg A Weiland O Mattsson L Jin L Birkett A Peterson D Milich DR Limited humoral immunity in hepatitis C virus infection Gastroenterology 1999 116 135 143 9869611
Rapicetta M Dettori S Kondili L Chionne P Ciccaglione A Miceli M Mannella E Persistence of HCV-RNA in a blood donor with negative antibody assays Vox Sang 1999 76 192 193 10341337 10.1159/000031049
Bassett SE Brasky KM Lanford RE Analysis of hepatitis C virus-inoculated chimpanzees reveals unexpected clinical profiles J Virol 1998 72 2589 2599 9525575
Rentenaar RJ Gamadia LE van DerHoek N van Diepen FN Boom R Weel JF Wertheim-van Dillen PM van Lier RA ten Berge IJ Development of virus-specific CD4(+) T cells during primary cytomegalovirus infection J Clin Invest 2000 105 541 548 10683384
Park SG Kim Y Park E Ryu HM Jung G Fidelity of hepatitis B virus polymerase Eur J Biochem 2003 270 2929 2936 12846825 10.1046/j.1432-1033.2003.03650.x
Preston BD Poiesz BJ Loeb LA Fidelity of HIV-1 reverse transcriptase Science 1988 242 1168 1171 2460924
Pugachev KV Guirakhoo F Ocran SW Mitchell F Parsons M Penal C Girakhoo S Pougatcheva SO Arroyo J Trent DW High fidelity of yellow fever virus RNA polymerase J Virol 2004 78 1032 1038 14694136 10.1128/JVI.78.2.1032-1038.2004
Steinhauer DA Domingo E Holland JJ Lack of evidence for proofreading mechanisms associated with an RNA virus polymerase Gene 1992 122 281 288 1336756 10.1016/0378-1119(92)90216-C
Duarte E Clarke D Moya A Domingo E Holland J Rapid fitness losses in mammalian RNA virus clones due to Muller's ratchet Proc Natl Acad Sci U S A 1992 89 6015 6019 1321432
Eigen M Viral quasispecies Sci Am 1993 269 42 49 8337597
Sallie R Replicative homeostasis: a mechanism of viral persistence Med Hypotheses 2004 63 515 523 15288380 10.1016/j.mehy.2004.02.050
Muhlemann K The molecular epidemiology of mumps virus Infect Genet Evol 2004 4 215 219 15450201 10.1016/j.meegid.2004.02.003
Capurso C Solfrizzi V D'Introno A Colacicco AM Capurso SA Capurso A Panza F Interleukin 6-174 G/C promoter gene polymorphism and sporadic Alzheimer's disease: geographic allele and genotype variations in Europe Exp Gerontol 2004 39 1567 1573 15501028 10.1016/j.exger.2004.07.006
Friedlaender JS Sgaramella-Zonta LA Kidd KK Lai LY Clark P Walsh RJ Biological divergences in south-central Bougainville: an analysis of blood polymorphism gene frequencies and anthropometric measurements utilizing tree models, and a comparison of these variables with linguistic, geographic, and migrational "distances" Am J Hum Genet 1971 23 253 270 4996844
Simmonds P Holmes EC Cha TA Chan SW McOmish F Irvine B Beall E Yap PL Kolberg J Urdea MS Classification of hepatitis C virus into six major genotypes and a series of subtypes by phylogenetic analysis of the NS-5 region J Gen Virol 1993 74 2391 2399 8245854
Brown JS Jr A novel mechanism to explain protein abnormalities in schizophrenia based on the flavivirus resistance gene Mol Psychiatry 2001 6 701 711 11673799 10.1038/sj.mp.4000890
Lung FW Tzeng DS Shu BC Ethnic heterogeneity in allele variation in the DRD4 gene in schizophrenia Schizophr Res 2002 57 239 245 12223255 10.1016/S0920-9964(01)00313-9
Ovreas L Daae FL Torsvik V Rodriguez-Valera F Characterization of microbial diversity in hypersaline environments by melting profiles and reassociation kinetics in combination with terminal restriction fragment length polymorphism (T-RFLP) Microb Ecol 2003 46 291 301 12904916 10.1007/s00248-003-3006-3
Yu X Angov E Camerini-Otero RD Egelman EH Structural polymorphism of the RecA protein from the thermophilic bacterium Thermus aquaticus Biophys J 1995 69 2728 2738 8599679
Bundock PC Henry RJ Single nucleotide polymorphism, haplotype diversity and recombination in the Isa gene of barley Theor Appl Genet 2004 109 543 551 15146316 10.1007/s00122-004-1675-z
Haanpaa M Nurmikko T Hurme M Polymorphism of the IL-10 gene is associated with susceptibility to herpes zoster Scand J Infect Dis 2002 34 112 114 11928840 10.1080/00365540110077218
Helminen M Lahdenpohja N Hurme M Polymorphism of the interleukin-10 gene is associated with susceptibility to Epstein-Barr virus infection J Infect Dis 1999 180 496 499 10395868 10.1086/314883
LaFleur C Granados J Vargas-Alarcon G Ruiz-Morales J Villarreal-Garza C Higuera L Hernandez-Pacheco G Cutino-Moguel T Rangel H Figueroa R HLA-DR antigen frequencies in Mexican patients with dengue virus infection: HLA-DR4 as a possible genetic resistance factor for dengue hemorrhagic fever Hum Immunol 2002 63 1039 1044 12392857 10.1016/S0198-8859(02)00682-1
Brouwer KC Lal RB Mirel LB Yang C van Eijk AM Ayisi J Otieno J Nahlen BL Steketee R Lal AA Polymorphism of Fc receptor IIa for IgG in infants is associated with susceptibility to perinatal HIV-1 infection Aids 2004 18 1187 1194 15166534 10.1097/00002030-200405210-00012
Haldane J Symposium sui fattori ecologi e genetici dela specilazione negli animali Supplemnto a la La Ricerca Scientifica: 1949 1949 68 76
Piatak M JrSaag MS Yang LC Clark SJ Kappes JC Luk KC Hahn BH Shaw GM Lifson JD High levels of HIV-1 in plasma during all stages of infection determined by competitive PCR Science 1993 259 1749 1754 8096089
Solter D Differential imprinting and expression of maternal and paternal genomes Annu Rev Genet 1988 22 127 146 3071246 10.1146/annurev.ge.22.120188.001015
Billington WD Maternal immune response to pregnancy Reprod Fertil Dev 1989 1 183 190 discussion 191. 2682805
Sargent IL Maternal and fetal immune responses during pregnancy Exp Clin Immunogenet 1993 10 85 102 8251183
Hamada H Arinami T Kubo T Hamaguchi H Iwasaki H Fetal nucleated cells in maternal peripheral blood: frequency and relationship to gestational age Hum Genet 1993 91 427 432 8314554 10.1007/BF00217766
Steele CD Wapner RJ Smith JB Haynes MK Jackson LG Prenatal diagnosis using fetal cells isolated from maternal peripheral blood: a review Clin Obstet Gynecol 1996 39 801 813 8934032 10.1097/00003081-199612000-00009
Selmi C Ross SR Ansari AA Invernizzi P Podda M Coppel RL Gershwin ME Lack of immunological or molecular evidence for a role of mouse mammary tumor retrovirus in primary biliary cirrhosis Gastroenterology 2004 127 493 501 15300582 10.1053/j.gastro.2004.05.033
McCall M Le Gay Brereton T Dawson A Millingen KS Sutherland JM Tyrer JH Eadie MJ Acheson ED Multiple sclerosis in Australia Proc Aust Assoc Neurol 1968 5 347 352 5707157
Pritchard MH Matthews N Munro J Antibodies to influenza A in a cluster of children with juvenile chronic arthritis Br J Rheumatol 1988 27 176 180 3378122
Prince MI Chetwynd A Diggle P Jarner M Metcalf JV James OF The geographical distribution of primary biliary cirrhosis in a well-defined cohort Hepatology 2001 34 1083 1088 11731995 10.1053/jhep.2001.29760
Dahlquist G Frisk G Ivarsson SA Svanberg L Forsgren M Diderholm H Indications that maternal coxsackie B virus infection during pregnancy is a risk factor for childhood-onset IDDM Diabetologia 1995 38 1371 1373 8582549
Andreoletti L Hober D Hober-Vandenberghe C Fajardy I Belaich S Lambert V Vantyghem MC Lefebvre J Wattre P Coxsackie B virus infection and beta cell autoantibodies in newly diagnosed IDDM adult patients Clin Diagn Virol 1998 9 125 133 9645994 10.1016/S0928-0197(98)00011-7
Gerling I Chatterjee NK Nejman C Coxsackievirus B4-induced development of antibodies to 64,000-Mr islet autoantigen and hyperglycemia in mice Autoimmunity 1991 10 49 56 1660313
Graves PM Norris JM Pallansch MA Gerling IC Rewers M The role of enteroviral infections in the development of IDDM: limitations of current approaches Diabetes 1997 46 161 168 9000690
Eddleston AL Hepatitis C infection and autoimmunity J Hepatol 1996 24 55 60 8836890
Prummel MF Laurberg P Interferon-alpha and autoimmune thyroid disease Thyroid 2003 13 547 551 12930598 10.1089/105072503322238809
Obermayer-Straub P Manns MP Hepatitis C and D, retroviruses and autoimmune manifestations J Autoimmun 2001 16 275 285 11334493 10.1006/jaut.2000.0488
Editorial Autoimmune thyroid disease in interferon-treated patients Lancet 1985 2 100 101 2861513
Caronia S Taylor K Pagliaro L Carr C Palazzo U Petrik J O'Rahilly S Shore S Tom BD Alexander GJ Further evidence for an association between non-insulin-dependent diabetes mellitus and chronic hepatitis C virus infection Hepatology 1999 30 1059 1063 10498660 10.1002/hep.510300416
Bosi E Minelli R Bazzigaluppi E Salvi M Fulminant autoimmune Type 1 diabetes during interferon-alpha therapy: a case of Th1-mediated disease? Diabet Med 2001 18 329 332 11437866 10.1046/j.1464-5491.2001.00492.x
McMurray RW Elbourne K Hepatitis C virus infection and autoimmunity Semin Arthritis Rheum 1997 26 689 701 9062950 10.1016/S0049-0172(97)80005-4
Fabbri C Jaboli MF Giovanelli S Azzaroli F Pezzoli A Accogli E Liva S Nigro G Miracolo A Festi D Gastric autoimmune disorders in patients with chronic hepatitis C before, during and after interferon-alpha therapy World J Gastroenterol 2003 9 1487 1490 12854147
Zinkernagel RM Antiinfection immunity and autoimmunity Ann N Y Acad Sci 2002 958 3 6 12021079
Koschel K Halbach M Rabies virus infection selectively impairs membrane receptor functions in neuronal model cells J Gen Virol 1979 42 627 632 219141
Ceccaldi PE Fillion MP Ermine A Tsiang H Fillion G Rabies virus selectively alters 5-HT1 receptor subtypes in rat brain Eur J Pharmacol 1993 245 129 138 8491253 10.1016/0922-4106(93)90120-X
Godfraind C Havaux N Holmes KV Coutelier JP Role of virus receptor-bearing endothelial cells of the blood-brain barrier in preventing the spread of mouse hepatitis virus-A59 into the central nervous system J Neurovirol 1997 3 428 434 9475114
Del Valle L Wang JY Lassak A Peruzzi F Croul S Khalili K Reiss K Insulin-like growth factor I receptor signaling system in JC virus T antigen-induced primitive neuroectodermal tumors – medulloblastomas J Neurovirol 2002 8 138 147 12491166 10.1080/13550280290101111
Kawaguchi T Yoshida T Harada M Hisamoto T Nagao Y Ide T Taniguchi E Kumemura H Hanada S Maeyama M Hepatitis C virus down-regulates insulin receptor substrates 1 and 2 through up-regulation of suppressor of cytokine signaling 3 Am J Pathol 2004 165 1499 1508 15509521
Marsh YV Eppstein DA Vaccinia virus and the EGF receptor: a portal for infectivity? J Cell Biochem 1987 34 239 245 3497935 10.1002/jcb.240340403
Zocher M Czub S Schulte-Monting J de La Torre JC Sauder C Alterations in neurotrophin and neurotrophin receptor gene expression patterns in the rat central nervous system following perinatal Borna disease virus infection J Neurovirol 2000 6 462 477 11175319
Zuo F Kraus RJ Gulick T Moore DD Mertz JE Direct modulation of simian virus 40 late gene expression by thyroid hormone and its receptor J Virol 1997 71 427 436 8985367
Carson SD Receptor for the group B coxsackieviruses and adenoviruses: CAR Rev Med Virol 2001 11 219 226 11479928 10.1002/rmv.318
Agnello V Abel G Elfahal M Knight GB Zhang QX Hepatitis C virus and other flaviviridae viruses enter cells via low density lipoprotein receptor Proc Natl Acad Sci U S A 1999 96 12766 12771 10535997 10.1073/pnas.96.22.12766
Monazahian M Bohme I Bonk S Koch A Scholz C Grethe S Thomssen R Low density lipoprotein receptor as a candidate receptor for hepatitis C virus J Med Virol 1999 57 223 229 10022791 10.1002/(SICI)1096-9071(199903)57:3<223::AID-JMV2>3.0.CO;2-4
Hueffer K Parker JS Weichert WS Geisel RE Sgro JY Parrish CR The natural host range shift and subsequent evolution of canine parvovirus resulted from virus-specific binding to the canine transferrin receptor J Virol 2003 77 1718 1726 12525605 10.1128/JVI.77.3.1718-1726.2003
Bitzer M Lauer U Baumann C Spiegel M Gregor M Neubert WJ Sendai virus efficiently infects cells via the asialoglycoprotein receptor and requires the presence of cleaved F0 precursor proteins for this alternative route of cell entry J Virol 1997 71 5481 5486 9188621
Saunier B Triyatni M Ulianich L Maruvada P Yen P Kohn LD Role of the asialoglycoprotein receptor in binding and entry of hepatitis C virus structural proteins in cultured human hepatocytes J Virol 2003 77 546 559 12477859 10.1128/JVI.77.1.546-559.2003
Li W Moore MJ Vasilieva N Sui J Wong SK Berne MA Somasundaran M Sullivan JL Luzuriaga K Greenough TC Angiotensin-converting enzyme 2 is a functional receptor for the SARS coronavirus Nature 2003 426 450 454 14647384 10.1038/nature02145
Bergelson JM Modlin JF Wieland-Alter W Cunningham JA Crowell RL Finberg RW Clinical coxsackievirus B isolates differ from laboratory strains in their interaction with two cell surface receptors J Infect Dis 1997 175 697 700 9041347
Matrosovich M Tuzikov A Bovin N Gambaryan A Klimov A Castrucci MR Donatelli I Kawaoka Y Early alterations of the receptor-binding properties of H1, H2, and H3 avian influenza virus hemagglutinins after their introduction into mammals J Virol 2000 74 8502 8512 10954551 10.1128/JVI.74.18.8502-8512.2000
Nielsen L Blixenkrone-Moller M Thylstrup M Hansen NJ Bolt G Adaptation of wild-type measles virus to CD46 receptor usage Arch Virol 2001 146 197 208 11315632 10.1007/s007050170169
Melder DC Pankratz VS Federspiel MJ Evolutionary pressure of a receptor competitor selects different subgroup a avian leukosis virus escape variants with altered receptor interactions J Virol 2003 77 10504 10514 12970435 10.1128/JVI.77.19.10504-10514.2003
Baranowski E Ruiz-Jarabo CM Domingo E Evolution of cell recognition by viruses Science 2001 292 1102 1105 11352064 10.1126/science.1058613
Murrell MT Porotto M Greengard O Poltoratskaia N Moscona A A single amino acid alteration in the human parainfluenza virus type 3 hemagglutinin-neuraminidase glycoprotein confers resistance to the inhibitory effects of zanamivir on receptor binding and neuraminidase activity J Virol 2001 75 6310 6320 11413297 10.1128/JVI.75.14.6310-6320.2001
Rogers GN Paulson JC Daniels RS Skehel JJ Wilson IA Wiley DC Single amino acid substitutions in influenza haemagglutinin change receptor binding specificity Nature 1983 304 76 78 6191220 10.1038/304076a0
Banks WA The many lives of leptin Peptides 2004 25 331 338 15134858 10.1016/j.peptides.2004.02.014
Jequier E Leptin signaling, adiposity, and energy balance Ann N Y Acad Sci 2002 967 379 388 12079865
Bjorbaek C Kahn BB Leptin signaling in the central nervous system and the periphery Recent Prog Horm Res 2004 59 305 331 14749508 10.1210/rp.59.1.305
Shintani M Ikegami H Fujisawa T Kawaguchi Y Ohishi M Katsuya T Higaki J Shimamoto K Ogihara T Leptin gene polymorphism is associated with hypertension independent of obesity J Clin Endocrinol Metab 2002 87 2909 2912 12050272 10.1210/jc.87.6.2909
McNeely MJ Boyko EJ Weigle DS Shofer JB Chessler SD Leonnetti DL Fujimoto WY Association between baseline plasma leptin levels and subsequent development of diabetes in Japanese Americans Diabetes Care 1999 22 65 70 10333905
Phillips BG Kato M Narkiewicz K Choe I Somers VK Increases in leptin levels, sympathetic drive, and weight gain in obstructive sleep apnea Am J Physiol Heart Circ Physiol 2000 279 H234 237 10899061
Soderberg S Ahren B Jansson JH Johnson O Hallmans G Asplund K Olsson T Leptin is associated with increased risk of myocardial infarction J Intern Med 1999 246 409 418 10583712 10.1046/j.1365-2796.1999.00571.x
Chiu KC Chu A Chuang LM Saad MF Association of leptin receptor polymorphism with insulin resistance Eur J Endocrinol 2004 150 725 729 15132731 10.1530/eje.0.1500725
Bernard A Cohen R Khuth ST Vedrine B Verlaeten O Akaoka H Giraudon P Belin MF Alteration of the leptin network in late morbid obesity induced in mice by brain infection with canine distemper virus J Virol 1999 73 7317 7327 10438820
Atkinson RL Dhurandhar NV Allison DB Bowen RL Israel BA Albu JB Augustus AS Human adenovirus-36 is associated with increased body weight and paradoxical reduction of serum lipids Int J Obes Relat Metab Disord 2005 29 281 286 10.1038/sj.ijo.0802830
Dhurandhar NV Whigham LD Abbott DH Schultz-Darken NJ Israel BA Bradley SM Kemnitz JW Allison DB Atkinson RL Human adenovirus Ad-36 promotes weight gain in male rhesus and marmoset monkeys J Nutr 2002 132 3155 3160 12368411
Seidell JC Obesity, insulin resistance and diabetes – a worldwide epidemic Br J Nutr 2000 83 S5 8 10889785
Kohn M Booth M The worldwide epidemic of obesity in adolescents Adolesc Med 2003 14 1 9 12529186
Abelson P Kennedy D The obesity epidemic Science 2004 304 1413 15178768 10.1126/science.304.5676.1413
Thomsen BL Ekstrom CT Sorensen TI Development of the obesity epidemic in Denmark: cohort, time and age effects among boys born 1930–1975 Int J Obes Relat Metab Disord 1999 23 693 701 10454102 10.1038/sj.ijo.0800907
Mokdad AH Serdula MK Dietz WH Bowman BA Marks JS Koplan JP The spread of the obesity epidemic in the United States, 1991–1998 Jama 1999 282 1519 1522 10546690 10.1001/jama.282.16.1519
Zimmet P Alberti KG Shaw J Global and societal implications of the diabetes epidemic Nature 2001 414 782 787 11742409 10.1038/414782a
Zimmet P Boyko EJ Collier GR de Courten M Etiology of the metabolic syndrome: potential role of insulin resistance, leptin resistance, and other players Ann N Y Acad Sci 1999 892 25 44 10842650
Morabia A Costanza MC The obesity epidemic as harbinger of a metabolic disorder epidemic: trends in overweight, hypercholesterolemia, and diabetes treatment in geneva, Switzerland, 1993–2003 Am J Public Health 2005 95 632 635 15798121 10.2105/2004.047877
Park RJ Lawrie SM Freeman CP Post-viral onset of anorexia nervosa Br J Psychiatry 1995 166 386 389 7788133
Heras M Sanz G Roig E Perez-Villa F Recasens L Serra A Betriu A Endothelial dysfunction of the non-infarct related, angiographically normal, coronary artery in patients with an acute myocardial infarction Eur Heart J 1996 17 715 720 8737102
Nystrom T Nygren A Sjoholm A Persistent endothelial dysfunction is related to elevated C-reactive protein (CRP) levels in Type II diabetic patients after acute myocardial infarction Clin Sci (Lond) 2005 108 121 128 15504106
Wang WY Zee RY Morris BJ Association of angiotensin II type 1 receptor gene polymorphism with essential hypertension Clin Genet 1997 51 31 34 9084931
Oh JY Barrett-Connor E Association between vitamin D receptor polymorphism and type 2 diabetes or metabolic syndrome in community-dwelling older adults: the Rancho Bernardo Study Metabolism 2002 51 356 359 11887173 10.1053/meta.2002.29969
Pedersen JC Berg K Normal DNA polymorphism at the low density lipoprotein receptor (LDLR) locus associated with serum cholesterol level Clin Genet 1988 34 306 312 2906577
Bates P Young JA Varmus HE A receptor for subgroup A Rous sarcoma virus is related to the low density lipoprotein receptor Cell 1993 74 1043 1051 8402880 10.1016/0092-8674(93)90726-7
Ji Q Ikegami H Fujisawa T Kawabata Y Ono M Nishino M Ohishi M Katsuya T Rakugi H Ogihara T A common polymorphism of uncoupling protein 2 gene is associated with hypertension J Hypertens 2004 22 97 102 15106800 10.1097/00004872-200401000-00018
Romano-Spica V Mettimano M Ianni A Specchia ML Ricciardi G Savi L Epidemiology of essential hypertension: the role of genetic polymorphism Eur J Epidemiol 2003 18 211 219 12800945 10.1023/A:1023360410810
Castellano M Glorioso N Cusi D Sarzani R Fabris B Opocher G Zoccali C Golin R Veglio F Volpe M Genetic polymorphism of the renin-angiotensin-aldosterone system and arterial hypertension in the Italian population: the GENIPER Project J Hypertens 2003 21 1853 1860 14508191 10.1097/00004872-200310000-00012
Danesh J Whincup P Walker M Lennon L Thomson A Appleby P Gallimore JR Pepys MB Low grade inflammation and coronary heart disease: prospective study and updated meta-analyses BMJ 2000 321 199 204 10903648 10.1136/bmj.321.7255.199
Perticone F Sciacqua A Scozzafava A Ventura G Laratta E Pujia A Federici M Lauro R Sesti G Impaired endothelial function in never-treated hypertensive subjects carrying the Arg972 polymorphism in the insulin receptor substrate-1 gene J Clin Endocrinol Metab 2004 89 3606 3609 15240653 10.1210/jc.2003-032161
Naghavi M Wyde P Litovsky S Madjid M Akhtar A Naguib S Siadaty MS Sanati S Casscells W Influenza infection exerts prominent inflammatory and thrombotic effects on the atherosclerotic plaques of apolipoprotein E-deficient mice Circulation 2003 107 762 768 12578882 10.1161/01.CIR.0000048190.68071.2B
Naghavi M Barlas Z Siadaty S Naguib S Madjid M Casscells W Association of influenza vaccination and reduced risk of recurrent myocardial infarction Circulation 2000 102 3039 3045 11120692
Joynt KE Whellan DJ O'Connor CM Depression and cardiovascular disease: mechanisms of interaction Biol Psychiatry 2003 54 248 261 12893101 10.1016/S0006-3223(03)00568-7
Crisby M Modulation of the inflammatory process by statins Drugs Today (Barc) 2003 39 137 143 12698208 10.1358/dot.2003.39.2.740209
Larose E Ganz P Statins and endothelial dysfunction Semin Vasc Med 2004 4 333 346 15861315 10.1055/s-2004-869590
McArthur JC HIV dementia: an evolving disease J Neuroimmunol 2004 157 3 10 15579274 10.1016/j.jneuroim.2004.08.042
Teo I Veryard C Barnes H An SF Jones M Lantos PL Luthert P Shaunak S Circular forms of unintegrated human immunodeficiency virus type 1 DNA and high levels of viral protein expression: association with dementia and multinucleated giant cells in the brains of patients with AIDS J Virol 1997 71 2928 2933 9060651
Van Marle G Rourke SB Zhang K Silva C Ethier J Gill MJ Power C HIV dementia patients exhibit reduced viral neutralization and increased envelope sequence diversity in blood and brain Aids 2002 16 1905 1914 12351950 10.1097/00002030-200209270-00007
Anttila S Illi A Kampman O Mattila KM Lehtimaki T Leinonen E Association of EGF polymorphism with schizophrenia in Finnish men Neuroreport 2004 15 1215 1218 15129177 10.1097/00001756-200405190-00027
Glatt SJ Faraone SV Tsuang MT Meta-analysis identifies an association between the dopamine D2 receptor gene and schizophrenia Mol Psychiatry 2003 8 911 915 14593428 10.1038/sj.mp.4001321
Fridman C Ojopi EP Gregorio SP Ikenaga EH Moreno DH Demetrio FN Guimaraes PE Vallada HP Gattaz WF Dias Neto E Association of a new polymorphism in ALOX12 gene with bipolar disorder Eur Arch Psychiatry Clin Neurosci 2003 253 40 43 12664313 10.1007/s00406-003-0404-y
Urwin RE Bennetts B Wilcken B Lampropoulos B Beumont P Clarke S Russell J Tanner S Nunn KP Anorexia nervosa (restrictive subtype) is associated with a polymorphism in the novel norepinephrine transporter gene promoter polymorphic region Mol Psychiatry 2002 7 652 657 12140790 10.1038/sj.mp.4001080
Brown AS Begg MD Gravenstein S Schaefer CA Wyatt RJ Bresnahan M Babulas VP Susser ES Serologic evidence of prenatal influenza in the etiology of schizophrenia Arch Gen Psychiatry 2004 61 774 780 15289276 10.1001/archpsyc.61.8.774
Takei N Mortensen PB Klaening U Murray RM Sham PC O'Callaghan E Munk-Jorgensen P Relationship between in utero exposure to influenza epidemics and risk of schizophrenia in Denmark Biol Psychiatry 1996 40 817 824 8896767 10.1016/0006-3223(95)00592-7
Machon RA Mednick SA Huttunen MO Adult major affective disorder after prenatal exposure to an influenza epidemic Arch Gen Psychiatry 1997 54 322 328 9107148
Rott R Herzog S Fleischer B Winokur A Amsterdam J Dyson W Koprowski H Detection of serum antibodies to Borna disease virus in patients with psychiatric disorders Science 1985 228 755 756 3922055
Roos RP Davis K Meltzer HY Immunoglobulin studies in patients with psychiatric diseases Arch Gen Psychiatry 1985 42 124 128 3977540
Sierra-Honigmann AM Carbone KM Yolken RH Polymerase chain reaction (PCR) search for viral nucleic acid sequences in schizophrenia Br J Psychiatry 1995 166 55 60 7894877
Kramer L Bauer E Funk G Hofer H Jessner W Steindl-Munda P Wrba F Madl C Gangl A Ferenci P Subclinical impairment of brain function in chronic hepatitis C infection J Hepatol 2002 37 349 354 12175630 10.1016/S0168-8278(02)00172-1
Forton DM Thomas HC Murphy CA Allsop JM Foster GR Main J Wesnes KA Taylor-Robinson SD Hepatitis C and cognitive impairment in a cohort of patients with mild liver disease Hepatology 2002 35 433 439 11826420 10.1053/jhep.2002.30688
Poutiainen E Iivanainen M Elovaara I Valle SL Lahdevirta J Cognitive changes as early signs of HIV infection Acta Neurol Scand 1988 78 49 52 3176882
Flier JS Bar RS Muggeo M Kahn CR Roth J Gorden P The evolving clinical course of patients with insulin receptor autoantibodies: spontaneous remission or receptor proliferation with hypoglycemia J Clin Endocrinol Metab 1978 47 985 995 263346
Sallie R The paradox of HIV nef function: Resolution of the contradictions Med Hypotheses 2005 65 44 46 15893116 10.1016/j.mehy.2005.02.008
Schooling M Leung GM Janus ED Ho SY Hedley AJ Lam TH Childhood migration and cardiovascular risk Int J Epidemiol 2004 33 1219 1226 15567872 10.1093/ije/dyh221
Cantor-Graae E Pedersen CB McNeil TF Mortensen PB Migration as a risk factor for schizophrenia: a Danish population-based cohort study Br J Psychiatry 2003 182 117 122 12562738 10.1192/bjp.182.2.117
Patel JV Vyas A Cruickshank JK Prabhakaran D Hughes E Reddy KS Mackness MI Bhatnagar D Durrington PN Impact of migration on coronary heart disease risk factors: Comparison of Gujaratis in Britain and their contemporaries in villages of origin in India Atherosclerosis 2005
Arias B Gutierrez B Pintor L Gasto C Fananas L Variability in the 5-HT(2A) receptor gene is associated with seasonal pattern in major depression Mol Psychiatry 2001 6 239 242 11317230 10.1038/sj.mp.4000818
|
16115320
|
PMC1260030
|
CC BY
|
2021-01-04 16:38:59
|
no
|
Virol J. 2005 Aug 22; 2:70
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-70
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2241615688710.1186/1471-2105-6-224Methodology ArticleVersatile and declarative dynamic programming using pair algebras Steffen Peter [email protected] Robert [email protected] Faculty of Technology, Bielefeld University, Postfach 10 01 31, 33501 Bielefeld, Germany2005 12 9 2005 6 224 224 24 5 2005 12 9 2005 Copyright © 2005 Steffen and Giegerich; licensee BioMed Central Ltd.2005Steffen and Giegerich; 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
Dynamic programming is a widely used programming technique in bioinformatics. In sharp contrast to the simplicity of textbook examples, implementing a dynamic programming algorithm for a novel and non-trivial application is a tedious and error prone task. The algebraic dynamic programming approach seeks to alleviate this situation by clearly separating the dynamic programming recurrences and scoring schemes.
Results
Based on this programming style, we introduce a generic product operation of scoring schemes. This leads to a remarkable variety of applications, allowing us to achieve optimizations under multiple objective functions, alternative solutions and backtracing, holistic search space analysis, ambiguity checking, and more, without additional programming effort. We demonstrate the method on several applications for RNA secondary structure prediction.
Conclusion
The product operation as introduced here adds a significant amount of flexibility to dynamic programming. It provides a versatile testbed for the development of new algorithmic ideas, which can immediately be put to practice.
==== Body
Background
Dynamic Programming is an elementary and widely used programming technique. Introductory textbooks on algorithms usually contain a section devoted to dynamic programming, where simple problems like matrix chain multiplication, polygon triangulation or string comparison are commonly used for the exposition. This programming technique is mainly taught by example. Once designed, all dynamic programming algorithms look similar: They are cast in terms of recurrences between table entries that store solutions to intermediate problems, from which the overall solution is constructed via a more or less sophisticated case analysis. However, the simplicity of these small programming examples is deceiving, as this style of programming provides no abstraction mechanisms, and hence it does not scale up well to more sophisticated problems.
In biological sequence analysis, dynamic programming algorithms are used on a great variety of problems, such as protein homology search, gene structure prediction, motif search, analysis of repetitive genomic elements, RNA secondary structure prediction, or interpretation of data from mass spectrometry [1-3]. The higher sophistication of these problems is reflected in a large number of recurrences – sometimes filling several pages – using more complicated case distinctions, numerous tables and elaborate scoring schemes. Hence, implementing a novel dynamic programming algorithm is a cumbersome task and requires extensive testing, while the resulting programs are difficult to re-use on related problems.
However, these difficulties are alleviated somewhat by a certain programming discipline: We may organize a dynamic programming algorithm such that the typical dynamic programming recurrences describe the problem decomposition and case analysis, but are completely separated from the intended optimization objective. Neither the initialization values for trivial problems, nor the scoring and objective functions, nor the required number of answers, not even the data type of the result must be visible in the recurrences. In this setting, one can simply exchange one encapsulated scoring scheme by another – including ones that do not solve optimizations problems, but compute other types of useful information about the search space. The new technique proposed here is a "product" operation on scoring schemes, creating a new scheme from two given ones. This product uses a non-trivial way to combine the two objective functions. Given some standard scoring schemes and the product operation, we can perform a remarkable variety of applications, such as optimizations under multiple objective functions, alternative solutions and backtracing, holistic search space analysis, ambiguity checking, and more, without additional programming effort, and without creating a need of debugging.
Overview
We set the stage for our exposition with a condensed review of the "algebraic" approach to dynamic programming. We also introduce the individual scoring schemes that will be used in products later. We then introduce and discuss our definition of the product operation. From there, we proceed with a series of examples demonstrating the versatile use of products. The new product operation has been implemented and made available via the Bielefeld Bioinformatics Server [4], where the reader may run the examples presented in this paper, as well as his or her own ones. In our own, real-world programming projects, the product operation has become indispensable, and we report from our experience in the implementation of several recent bioinformatics tools.
Algebraic dynamic programming by example
For our presentation, we need to give a short review of the concepts underlying the algebraic style of dynamic programming (ADP): trees, signatures, tree grammars, and evaluation algebras. We strive to avoid formalism as far as possible, and give an exemplified introduction here, sufficient for our present concerns. See [5] for a complete presentation of the ADP method. As a running example, we use the RNA secondary structure prediction problem. We start with a simple approach resulting in an ADP variant of Nussinov's algorithm [6] and move on to a more elaborate example to permit the demonstration of our new concepts. Nothing of the new ideas presented here is specific to the RNA folding problem. Products can be applied to all problems within the scope of algebraic dynamic programming, including pairwise problems like sequence alignment [5].
RNA secondary structure prediction
While today the prediction of RNA 3D structure is inaccessible to computational methods, its secondary structure, given by the set of paired bases, can be predicted quite reliably. Figure 1 gives examples of typical elements found in RNA secondary structure, called stacking regions (or helices), bulge loops, internal loops, hairpin loops and multiple loops.
Figure 1 Typical elements found in RNA secondary structure.
The first approach to structure prediction was proposed by Nussinov in 1978 and was based on the idea of maximizing the number of base pairs [6]. Today's algorithms are typically based on energy minimization.
ADP methodology
When designing a dynamic programming algorithm in algebraic style, we need to specify four constituents:
• Alphabet: How is the input sequence given?
• Search space: What are the elements of the search space and how can they be represented?
• Scoring: Given an element of the search space, how do we score it?
• Objective: Given a number of scores, which are the ones we are interested in?
In the following, we will work through these steps for the RNA secondary structure prediction problem.
Alphabet
The input RNA sequence is a string over = {a, c, g, u}. is called the alphabet and * denotes the set of sequences over of arbitrary length. ε denotes the empty string. In the following, we denote the input sequence with w ∈ *.
Search space
Given the input sequence w, the search space is the set of all possible secondary structures the sequence w can form. In the ADP terminology, the elements of the search space for a given input sequence are called candidates. Our next task is to decide how to represent such candidates. Two possible ways are shown in Figure 2. The first variant is the well-known dot-bracket notation, where pairs of matching parentheses are used to denote pairing bases. The second variant, the tree representation, is the one we use in the algebraic approach.
Figure 2 Two candidates in the search space for the best secondary structure for the sequence gucaugcaguguca.
Such a tree representation of candidates is quite commonly used in RNA structure analysis, but not so in other applications of dynamic programming. To appreciate the scope of the ADP method, it is important to see that such a representation exists for any application of dynamic programming (see appendix).
In our example, the trees are constructed using four different node labels. Each label represents a different situation, which we want to distinguish in the search space and in the eventual scoring of such a candidate. A node labeled pair represents the paring of two bases in the input sequence. The remaining nodes right, split and nil represent unpaired, branching and empty structures. It is easy to see that each tree is a suitable representation of the corresponding dot-bracket string. Also note that in each of the example trees, the original sequence can be retrieved by collecting the leaves in a counter-clockwise fashion. This is what we call the yield of the tree. The yield function y maps candidate trees back onto their corresponding sequences.
The next important concept is the notion of the signature. The signature describes the interface to the scoring functions needed in our algorithm. We can derive the signature for our current example by simply interpreting each of the candidate trees' node labels as a function declaration:
The symbol Ans is the abstract result domain. In the following, Σ denotes the signature, TΣ the set of trees over the signature Σ.
With the concepts of yield and signature we are now prepared to give a first definition of the search space: Given an input sequence w and a signature Σ, the search space P(w) is the subset of trees from TΣ, whose yield equals w. More formally, P(w) = {t ∈ TΣ|y(t) = w}.
This would suffice as a very rough description of the search space. In general, we want to impose more restrictions on it, for example, we want to make sure, that the operator pair is only used in combination with valid base pairs. For this purpose we introduce the notion of tree grammar. Figure 3 shows grammar nussinov78, origin of our two example trees. This grammar consists of only one nonterminal, s, and one production with four alternatives, one for each of the four function symbols that label the nodes. Z denotes the axiom of the grammar. The symbols base and empty are terminal symbols, representing an arbitrary base and the empty sequence. The symbol basepairing is a syntactic predicate that guarantees that only valid base pairs can form a pair-node.
Figure 3 Tree grammar nussinov78. Terminal symbols in italics.
Our refined definition of the search space is the following: Given a tree grammar over Σ and and a sequence w ∈ *, the language described by is = {t|t ∈ TΣ, t can be derived from the axiom via the rules of }. The search space spawned by w is .
From the language theoretic viewpoint, is the set of all parses of the sequence w for grammar .
The method we use for constructing the search space is called yield parsing, a solved problem that need not concern us here.
Scoring
Given an element of the search space as a tree t ∈ , we need to score this element. In our example we are only interested in counting base pairs, so scoring is very simple: The score of a tree is the number of pair-nodes in t. For the two candidates of Figure 2 we obtain scores of 3 (t1) and 4 (t2). To implement this, we provide definitions for the functions that make up our signature Σ:
In mathematics, the interpretation of a signature by a concrete value set and functions operating thereon is called an algebra. Hence, scoring schemes are algebras in ADP. Our first example is the algebra bpmax for maximizing the number of base pairs. The subscript bpmax attached to the function names indicates, that these definitions are interpretations of the function under this algebra. In the following, we will omit these subscripts.
The flexibility of the algebraic approach lies in the fact that we don't have to stop with definition of one algebra. Simply define another algebra and get other results for the same search space. We will introduce a variety of algebras for our second, more elaborate example in Section In-depth search space analysis.
Objective
The tree grammar describes the search space, the algebra the scoring of solution candidates. Still missing is our optimization objective. For this purpose we add an objective function h to the algebra which chooses one or more elements from a list of candidate scores. An algebra together with an objective function forms an evaluation algebra. Thus algebra bpmax becomes:
A given candidate t can be evaluated in many different algebras; we use the notation ε(t) to indicate the value obtained from t under evaluation with algebra ε.
Given that yield parsing constructs the search space for a given input, all that is left to do is to evaluate the candidates in a given algebra, and make our choice via the objective function h. For example, candidates t1 and t2 of Figure 2 are evaluated by algebra bpmax in the following way:
Definition 1 (Algebraic dynamic programming)
• An ADP problem is specified by a signature Σ over , a tree grammar over Σ, and a Σ-evaluation algebra ε with objective function h.
• An ADP problem instance is posed by a string w ∈ *. Its search space is the set of all its parses, .
• Solving an ADP problem is computing h{ε(t) | t ∈ } in polynomial time and space with respect to |w|.
In general, Bellman's Principle of Optimality [7] must be satisfied in order to achieve polynomial efficiency.
Definition 2 (ADP formulation of Bellman's Principle) An evaluation algebra satisfies Bellman's Principle, if for each k-ary function f in Σ and all answer lists z1,..., zk, the objective function h satisfies
h([f(x1,..., xk) | x1 ← z1,..., xk ← zk]) = h([f(x1,..., xk) | x1 ← h(z1),..., xk ← h(zk)])
as well as
h( z ++ z' ) = h( h(z) ++ h(z') )
where ++ denotes list concatenation, and ← denotes list membership.
Bellman's Principle, when satisfied, allows the following implementation: As the trees that constitute the search space are constructed by the yield parser in a bottom up fashion, rather than building them explicitly as elements of TΣ, for each function symbol f the evaluation function fε is called. Thus, the yield parser computes not trees, but their evaluations. To reduce their number (and thus to avoid exponential explosion) the objective function may be applied at an intermediate step where a list of alternative answers has been computed. Due to Bellman's Principle, the recursive intermediate applications of the objective function do not affect the final result.
As an example, consider the following two candidates (represented as terms) in the search space for sequence aucg:
Since algebra bpmax satisfies Bellman's Principle, we can apply the objective function h at intermediate steps inside the evaluation of candidates t3 and t4:
Given grammar and signature, the traditional dynamic programming recurrences can be derived mechanically to implement the yield parser. In the sequel, we shall use the name of a grammar as the name of the function that solves the dynamic programming problem at hand. Naturally, it takes two arguments, the evaluation algebra and the input sequence.
In-depth search space analysis
Note that the case analysis in the Nussinov algorithm is redundant – even the sequence ' aa' is assigned the two trees Right (Right Nil 'a') 'a' and Split (Right Nil 'a') (Right Nil 'a'), which actually denote the same structure.
In order to study also suboptimal solutions, a non-redundant algorithm was presented in [8]. Figure 4 shows the grammar wuchty98. Here the signature has 8 function symbols, each one modeling a particular structure element, plus the list constructors (nil, ul, cons) to collect sequences of components in a unique way. Nonterminal symbol strong is used to capture structures without isolated (unstacked) base pairs, as "lonely pairs" are known to be energetically unstable. Purging them from the search space decreases the number of candidates considerably. This grammar, because of its non-redundancy, can also be used to study combinatorics, such as the expected number of feasible structures of a sequence of length n. This algorithm, as implemented in RNAsubopt [8], is widely used for structure prediction via energy minimization. The thermodynamic model is too elaborate to be presented here, and we will stick with base pair maximization as our optimization objective for the sake of this presentation. Figure 5 shows four evaluation algebras that we will use with grammar wuchty98. We illustrate their use via the following examples, where g(a,w) denotes the application of grammar g and algebra a to input w. Table 1 summarizes all results for an example sequence.
Figure 4 Tree grammar wuchty98. Terminal symbols in italics.
Figure 5 Four evaluation algebras for grammar wuchty98. Arguments a and b denote bases, (i, j) represents the input subword wi + 1 .... wj, and s denotes answer values. Function dots(r) in algebra pretty yields a string of r dots ('.').
Table 1 Applications of grammars wuchty98 and nussinov78 with different individual algebras on input w = cgggauaccacu.
Application Result
wuchty98(enum, w) [Str (Ul (Bl (0,1) (Sr 'g' (Hl 'g' (3,10) 'c') 'u'))), Str (Ul (Bl (0,2) (Sr 'g' (Hl 'g' (4,10) 'c') 'u'))), Str (Cons (Bl (0,1) (Sr 'g' (Hl 'g' (3,7) 'c') 'c')) (Ul (Ss (9,12)))), Str (Cons (Bl (0,2) (Sr 'g' (Hl 'g' (4,7) 'c') 'c')) (Ul (Ss (9,12)))), Str (Ul (Ss (0,12)))]
wuchty98(pretty, w) [".((.......))", "..((......))", ".((....))...", "..((...))...", "............"]
wuchty98(bpmax, w) [2]
wuchty98(count,w) [5]
nussinov78(count,w) [9649270]
wuchty98(enum,w): The enumeration algebra enum yields unevaluated terms. By convention, function symbols are capitalized in the output. Since the objective function is identity, this call enumerates all candidates in the search space spawned by w. This is mainly useful in program debugging, as it allows us to inspect the search space actually traversed by our program.
wuchty98(pretty, w): The pretty-printing algebra pretty yields a dot-bracket string representation of the same structures as the above.
wuchty98(bpmax,w): The base pair maximization algebra is bpmax, such that this call yields the maximal number of base pairs that a structure for w can attain. Here the objective function is maximization, and it can be easily shown to satisfy Bellman's Principle. Similarly for grammar nussinov78.
wuchty98(count,w): The counting algebra count has as its objective function summation, and εcount(t) = 1 for all candidates t. Hence, summing over all candidate scores gives the number of candidates.
However, the evaluation functions are carefully written such that they satisfy Bellman's Principle. Thus, [length(wuchty98(enum,w))] == wuchty98(count,w), where the right-hand side is polynomial to compute, while the left-hand side typically is exponential due to the large number of answers returned by wuchty98(enum,w). Technically, the result of wuchty98(count,w) is a singleton list, hence the [...].
nussinov78(count,w): This computes (using an analogous version of the counting algebra not shown here) the number of structures considered by the Nussinov algorithm, which, in contrast to the above, is much larger than the size of the search space.
These examples show analyses achieved by individual algebras. We now turn to what can be done by their combination.
Results and discussion
In this section we first introduce and discuss our definition of the product operation. From there, we proceed with a series of examples demonstrating its usage.
The product operation on evaluation algebras
We define the product operation as follows:
Definition 3 (Product operation on evaluation algebras) Let M and N be evaluation algebras over Σ. Their product M***N is an evaluation algebra over Σ and has the functions
fM***N((m1, n1)...(mk, nk)) = (fM(m1,..., mk), fN(n1,..., nk)) for each f in Σ,
and the objective function
Above, ∈ denotes set membership and hence ignores duplicates. In contrast, ← denotes list membership and respects duplicates. Implementing set membership may require some extra filtering effort, but when the objective function hM, which computes L, does not produce duplicates anyway, it comes for free. We illustrate the application of the product operation to algebras bpmax and count:
Here, each function calculates a pair of two result values. The first is the result of algebra bpmax, the second is the result of algebra count. The interesting part is the objective function h. It receives the list of pairs as input, with each pair consisting of the candidate's scores for the first and for the second algebra. In the first step the objective function of algebra bpmax (max) is applied to the list of the first pair elements. The result is stored in L. In this example, L holds only one element, namely the maximum base pair score of the input list. In general, L holds many elements or may be empty. For each element of L, a new intermediate list is constructed that consists of all corresponding right pair elements of the input list. This intermediate list is then applied to the objective function of the second algebra (here: summation). Finally, the result of h is constructed by combination of all elements from L with their corresponding result for the second algebra stored in r. This computes the optimal number of base pairs, together with the number of candidate structures that achieve it.
One should not rely on intuition alone for understanding what M***N actually computes. For any tree grammar and product algebra M***N, their combined meaning is well defined by Definition 1, and the view that a complete list of all candidates is determined first, with hM***N applied only once in the end, is very helpful for understanding. But does (M***N, w) actually do what it means? The implementation works correctly if and only if Bellman's Principle is satisfied by M***N, which is not implied when it holds for M and N individually! Hence, use of product algebras is subject to the following Proof obligation: Prove that M***N satisfies Bellman's Principle (Definition 2).
Alas, we have traded the need of writing and debugging code for a proof obligation. Fortunately, there is a theorem that covers many important cases:
Theorem 1 (1) For any algebras M and N, and answer list x, (idM* * * idN)(x) is a permutation of x.
(2) If hM and hN minimize with respect to some order relations ≤M and ≤n, then hM***N minimizes with respect to the lexicographic ordering (≤M, ≤N).
(3) If both M and N minimize as above and both satisfy Bellman's Principle, then so does M***N.
Proof. (1) According to Definition 3, the elements of x are merely re-grouped according to their first component. For this to hold, it is essential that L is treated as a set. (2) follows directly from Definition 3. (3) In the case of minimization, Bellman's Principle is equivalent to (strict) monotonicity of the functions fM and fN with respect to ≤M and ≤N, and this carries over to the combined functions (trivially) and the lexicographic ordering (because of (2)). □
In the above proof, strict monotonicity is required only if we ask for multiple optimal, or for the k best solutions rather than a single, optimal one [9].
Theorem 1 (1) justifies our peculiar treatment of the list L as a set. It states that no elements of the search space are lost or get duplicated by the combination of two algebras. Theorem 1 (2,3) say that *** behaves as expected in the case of optimizing evaluation algebras. This is very useful, but not too surprising. A surprising variety of applications arises when *** is used with algebras that do not do optimization, like enum, count, and pretty.
The proof obligation is met for all the applications studied below. A case where the proof fails is, for example, wuchty98(count***count,w), which consequently delivers no meaningful result.
Implementing the product operation
The algebraic style of dynamic programming can be practiced within any decent programming language. It is mainly a discipline of structuring our dynamic programming algorithms, the perfect separation of problem decomposition and scoring being the main achievement. When using the ASCII notation for tree grammars proposed in [5], the grammars can be compiled into executable code. Otherwise, one has to derive the explicit recurrences and implement the corresponding yield parser. Care must be taken to keep the implementation of the recurrences independent of the result data type, such that they can be run with different algebras, including arbitrary products.
All this given, the extra effort for using product algebras is small. It amounts to implementing the defining equations for the functions of M***N generically, i.e. for arbitrary evaluation algebras M and N over the common signature Σ. In a language which supports functions embedded in data structures, this is one line per evaluation function, witnessed by our implementation in Haskell (available for download). In other languages, abstract classes (Java) or templates (C++) can be used. It is essential to provide a generic product implementation. Otherwise, each new algebra combination must be hand-coded, which is not difficult to do, but tedious and error-prone, and hence necessitates debugging. A generic product, once tested, guarantees absence of errors for all combinations.
Efficiency discussion
Before we turn to the uses of ***, a word on computational efficiency seems appropriate. Our approach requires to structure programs in a certain way. This induces a small (constant factor) overhead in space and time. For example, we must generically return a list of results, even with analyses that return only a single optimal value. Normally, each evaluation function is in O(1), and when h returns a single answer, asymptotic efficiency is solely determined by the tree grammar [5]. This asymptotic efficiency remains unaffected when we use a product algebra. Each table entry now holds a pair of answers, each of size O(1). Things change when we employ objective functions that produce multiple results, as the size of the desired output can become exponential in n, and then it dominates the overall computational effort. For example, the optimal base pair score may be associated with a large number of co-optimal candidates, especially when the grammar is ambiguous. Thus, if using *** makes our programs slower (asymptotically), it is not because of an intrinsic effect of the product operation, but because we decide to do more costly analyses by looking deeper into the search space.
The only exception to this rule is the situation where objective function hM produces duplicate values, which must be filtered out, as described with Definition 3. In this case, a non-constant factor proportional to the length of answer lists is incurred.
The concrete effect of using product algebras on CPU time and space is difficult to measure, as the product algebra runs a more sophisticated analysis than either single one. For an estimation, we measure the (otherwise meaningless) use of the same algebra twice. We compute wuchty98(bpmax,w) and compare to wuchty98(bpmax***bpmax,w). The outcome is shown in Table 2. For input lengths from 200 to 1600 bases, the product algebra uses 9.57% to 21.34% more space and is 18.97% to 29.46% slower than the single algebra.
Table 2 Measuring time and space requirements of the product operation. All results are for a C implementation of wuchty98, running on a 900 MHz Ultra Sparc 3 CPU under Sun Solaris 10. The space requirements were measured using a simple wrapper function for malloc, that counts the number of allocated bytes. Times were measured with gnu time.
|w| wuchty98(bpmax,w) wuchty98(bpmax***bpmax,w) %
time (sec) 200 0.58 0.69 + 18.97
space (MB) 200 1.88 2.06 + 9.57
time (sec) 400 4.65 6.02 + 29.46
space (MB) 400 4.60 5.37 + 16.74
time (sec) 800 52.04 65.54 + 25.94
space (MB) 800 15.61 18.77 + 20.24
time (sec) 1600 590.72 725.03 + 22.74
space (MB) 1600 59.85 72.62 + 21.34
Applications of product algebras
We now turn to applications of product algebras. Table 3 summarizes all results of the analyses discussed in the sequel, for a fixed example RNA sequence.
Table 3 Example applications of product algebras with grammar wuchty98 on input w = cgggauaccacu.
Application Result
wuchty98(bpmax***count,w) [(2,4)]
wuchty98(bpmax***pretty,w) [(2,".((.......))"), (2,"..((......))"), (2,".((....))..."), (2,"..((...))...")]
wuchty98(bpmax***enum,w) [(2, Str (Ul (Bl (0,1) (Sr 'g' (Hl 'g' (3,10) 'c') 'u')))), (2, Str (Ul (Bl (0,2) (Sr 'g' (Hl 'g' (4,10) 'c') 'u')))), (2, Str (Cons (Bl (0,1) (Sr 'g' (Hl 'g' (3,7) 'c') 'c')) (Ul (Ss (9,12))))), (2, Str (Cons (Bl (0,2) (Sr 'g' (Hl 'g' (4,7) 'c') 'c')) (Ul (Ss (9,12)))))]
wuchty98(bpmax***(enum***pretty,w) [(2,(Str (Ul (Bl (0,1) (Sr 'g' (Hl 'g' (3,10) 'c') 'u'))), ".((.......))")), (2, (Str (Ul (Bl (0,2) (Sr 'g' (Hl 'g' (4,10) 'c') 'u'))), "..((......))")), (2,(Str (Cons (Bl (0,1) (Sr 'g' (Hl 'g' (3,7) 'c') 'c')) (Ul (Ss (9,12)))), ".((....))...")), (2, (Str (Cons (Bl (0,2) (Sr 'g' (Hl 'g' (4,7) 'c') 'c')) (Ul (Ss (9,12)))), "..((...))..."))]
wuchty98(shape***count, w) [("_ [_]", 2), ("_ [_]_", 2), ("_",1)]
wuchty98(bpmax(5)***shape, w) [(2,"_ [_]"), (2,"_ [_]_"), (0,"_")]
wuchty98(bpmax(5)***(shape***count), w) [(2, ("_[_]", 2)), (2, ("_[_]_", 2)), (0, ("_",1))]
wuchty98(shape***bpmax, w) [("_[_]", 2), ("_[_]_", 2), ("_",0)]
wuchty98(bpmax***pretty', w) [(2,".((....))...")]
wuchty98(pretty***count, w) [(".((.......))",1), ("..((......))",1), (".((....))...",1), ("..((....))...",1), ("............",1)]
Application 1: Backtracing and co-optimal solutions
Often, we want not only the optimal answer value, but also a candidate which achieves the optimum. We may ask if there are several optimal candidates. If so, we may want to see them all, maybe even including some near-optimal candidates. The traditional technique is to store a table of intermediate answers and backtrace through the optimizing decisions made [1]. This backtracing can become quite intricate to program if we ask for more than one candidate. We can answer these questions without additional programming efforts using products:
wuchty98(bpmax***count,w) computes the optimal number of base pairs, together with the number of candidate structures that achieve it.
wuchty98(bpmax***enum,w) computes the optimal number of base pairs, together with all structures for w that achieve this maximum, in their representation as terms from TΣ.
wuchty98(bpmax***pretty,w) does the same as the previous call, producing the string representation of structures.
wuchty98(bpmax***(enum***pretty),w) does both of the above.
To verify all these statements, apply Definition 3, or visit the ADP web site and run your own examples. It is a nontrivial consequence of Definition 3 that the above product algebras in fact give all co-optimal solutions. Should only a single one be desired, we would use enum or pretty with a modified objective function h that retains only one element.
Note that our substitution of backtracing by a "forward" computation does not affect asymptotic runtime efficiency. With bpmax***enum, for example, the algebra stores in each table entry the optimal base pair count, together with the top node of the optimal candidate(s) and pointers to its immediate substructures, which reside in other table entries. Hence, even if there should be an exponential number of co-optimal answers, they are represented in polynomial space, because subtrees are shared. Should the user decide to have them all printed, exponential effort is incurred, just as with a backtracing implementation.
Application 2: Holistic search space analysis
Abstract shapes were recently proposed in [10] as a means to obtain a holistic view of an RNA molecule's folding space, avoiding the explicit enumeration of a large number of structures. Bypassing all relevant mathematics, let us just say that an RNA shape is an object that captures structural features, like the nesting pattern of stacking regions, but not their size. We visualize shapes by strings alike dot-bracket notation, such as _[_[_]], where _ denotes an unpaired region and [together with the matching ] denotes a complete helix of arbitrary length. This is achieved by the following shape algebra. Here, function shMerge appends two strings and merges adjacent characters for unpaired regions (_). The function nub eliminates duplicates from its input list.
Together with a creative use of ***, the shape algebra allows us to analyze the number of possible shapes, the size of their membership, and the (near-) optimality of members, and so on. Let bpmax(k) be bpmax with an objective function that retains the best k answers (without duplicates).
wuchty98(shape***count,w) computes all the shapes in the search space spawned by w, and the number of structures that map onto each shape.
wuchty98(bpmax(k)***shape,w) computes the best k base pair scores, together with their candidates' shapes.
wuchty98(bpmax(k)***(shape***count),w)
computes base pairs and shapes as above, plus the number of structures that achieve this number of base pairs in the given shape.
wuchty98(shape***bpmax,w) computes for each shape the maximum number of base pairs among all structures of this shape.
Application 3: Optimization under lexicographic orderings
Theorem 1 is useful in practice as one can test different objectives independently and then combine them in one operation. A simple case of using two orderings would be the following: Assume we have a case with a large number of co-optimal solutions. Let pretty' be pretty with h = min.
wuchty98(bpmax***pretty',w) computes among several optimal structures the one which comes alphabetically first according to its string representation.
Of course, there are more meaningful uses of a primary and a secondary optimization objective. For lack of space, we refrain from introducing another optimizing algebra here.
Application 4: Testing ambiguity
Dynamic programming algorithms can often be written in a simpler way if we do not care whether the same solution is considered many times during the optimization. This does not affect the overall optimum. A dynamic programming algorithm is then called redundant or ambiguous. In such a case, the computation of a list of near-optimal solutions is cumbersome, as it contains duplicates whose number often has an exponential growth pattern. Also, search space statistics become problematic – for example, the counting algebra speaks about the algorithm rather than the problem space, as it counts evaluated, but not necessarily distinct solutions. Tree grammars with a suitable probabilistic evaluation algebra implement stochastic context free grammars (SCFGs) [2]. The frequently used statistical scoring schemes, when trying to find the answer of maximal probability (the Viterbi algorithm, cf. [2]), are fooled by the presence of redundant solutions. In principle, it is clear how to control ambiguity [11]. One needs to show unambiguity of the tree grammar in the language theoretic sense (not the associated string grammar – it is always ambiguous, else we would not have an optimization problem), and the existence of an injective mapping from TΣ to a canonical model of the search space. However, the proofs involved are not trivial. Rather, one would like to implement a check for ambiguity that is applicable for any given tree grammar, but this may be difficult or even impossible, as the problem is closely related to ambiguity of context free languages, which is known to be formally undecidable.
Recently, Dowell and Eddy showed that ambiguity really matters in practice for SCFG design, and they suggest a procedure for ambiguity testing [12]. This test uses a combination of Viterbi and Inside algorithms to check whether the (putatively optimal) candidate returned by Viterbi has an alternative derivation. A more complete test is the following, and due to the use of ***, it requires no implementation effort:
The required homomorphism from the search space to the canonical model may be coded as another evaluation algebra. In fact, if we choose the dot-bracket string representation as the canonical model, algebra pretty does exactly this. We can test for ambiguity by testing injectivity of pretty – by calling wuchty98(pretty***count,w) on a number of inputs w. If any count larger than 1 shows up in the results, we have proved ambiguity. This test is strictly stronger than the one by Dowell and Eddy, which detects ambiguity only if it occurs with the (sampled) "near-optimal" predictions. This and other test methods are studied in detail in [13].
Limitations of the product operation
The above applications demonstrate the considerable versatility of the algebra product. In particular, since a product algebra is an algebra itself, we can work with algebra triples, quadruples, and so on. All of these will be combined in the same fashion, and here we reach the limits of the product operation. The given definition of *** is not the only way needed to combine two algebras. In abstract shape analysis [10], we use three algebras mfe (computing minimal free energy), shape and pretty. A shape representative structure is the structure of minimal free energy within the shape. Similarly to the above, wuchty98(shape***(mfe***pretty),w) computes the representatives of all shapes. However, computing only the k best shape representatives requires minimization within and across shapes, which neither mfe***shape nor shape***mfe can achieve. Hence, a hand-coded combination of the three algebras is necessary for this particular analysis.
Conclusion
We hope to have demonstrated that the evaluation algebra product as introduced here adds a significant amount of flexibility to dynamic programming. We have shown how ten meaningful analyses with quite diverse objectives can be obtained by using different products of a few simple algebras. The techniques we have shown here pertain not just to RNA folding problems, but to all dynamic programming algorithms that can be formulated in the algebraic style.
The benefits from using a particular coding discipline do not come for free. There is some learning effort required to adapt to a systematic approach and to abandon traditional coding habits. After that, the discipline pays back by making programmers more productive. Yet, the pay-off is hard to quantify. We therefore conclude with a subjective summary of our experience as bioinformatics toolsmiths. After training a generation of students on the concepts presented here, we have enjoyed a boost in programming productivity. Four bioinformatics tools have been developed using this approach -pknotsRG [14], RNAshapes [10], RNAhybrid [15] and RNAcast [16]. The "largest" example is the pseudoknot folding grammar, which uses 47 nonterminal symbols and a 140-fold case distinction. The techniques described here have helped to master such complexity in several ways:
• The abstract formulation of dynamic programming recurrences in the form of grammars makes it easy to communicate and refine algorithmic ideas.
• Writing non-ambiguous grammars for optimization problems allows us to use the same algorithm for mathematical analysis of the search space.
• Ambiguity checking ensures us that such analyses are correct, that probabilistic analyses are not fooled by redundant recurrences, and that near-optimal solutions when reported do not contain duplicates.
• enum algebras allow for algorithm introspection – we can obtain a protocol of all solution candidates the program has seen, a quite effective program validation aid.
• Using pretty and enum algebras in products frees us from the tedious programming of backtracing routines.
• The use of the product operation allows us the create new analyses essentially with three key strokes – and a proof obligation that must be met.
This has created a good testbed for the development of new algorithmic ideas, which can immediately be put to practice.
Methods
The new product operation has been implemented and made available via the Bielefeld Bioinformatics Server [4], where the reader may run the examples presented in this paper, as well as his or her own ones.
Appendix: "Reverse engineering" of dynamic programming algorithms
To support the claim that a tree representation of candidates always exists, we show how such a representation can be found by a sort of reverse engineering of a given DP algorithm not formulated in the ADP framework. The reasoning applied is called symbolic evaluation in computer science terminology.
Assume that the algorithm returns a score of (say) X1. The last step in the computation of this score value must have been the application of some function f1. This function corresponds to a particular problem decomposition. Let X1 = f1(a, b,..., X2, X3,...). Here a, b,... come from the input problem and are parameters for the local score contribution, while X2, X3,... are scores from subproblems. For example, in pairwise sequence alignment, f1 may correspond to replacing one amino acid by another, adding the PAM score of mutating a to b to the score of the remaining alignment, X2. There are no additional Xi in this case. With RNA folding, f1 may correspond to a multiloop, with a, b being the closing base pair, and X2 corresponding to the first stem within the multiloop, and X3 to a succession of the other stems. This includes DP algorithms that consider RNA pseudoknots. f1 may correspond to composing a pseudoknot from two crossing helices corresponding to X2 and X3, as is apparent in the pseudoknot folding program pknotsRG [14], which was developed with the ADP method. Unless the DP algorithm has been written very systematically, the actual operations that implement f1 may be scattered over various places in the code.
Whatever the meaning and implementation of f1 is – we record the formula F1 = f1(a, b,..., X2, X3,...), and recursively apply the same consideration to the subproblem scores Xi, using the appropriate functions fi to obtain formulas Fi which we substitute in F1 for the corresponding Xi, and so on. This eventually results in a large formula F1 that contains no more Xi and hence depends only on the input parameters. We have recorded, as the symbolic expression F1, the complete computation of the score X1. We have not recorded how this formula was found – it does not at all reflect the control structure of the algorithm. As any formula can be naturally considered as a tree, we have found a tree representation of the optimal candidate. Clearly, all suboptimal candidates have similar representations, obtained in the same way.
The reverse engineering of a known DP algorithm is not only a theoretical possibility, but also an instructive educational exercise, and helpful in analyzing and improving upon previous work.
Authors' contributions
Both authors cooperated closely in developing the work presented here, and in writing the manuscript.
Acknowledgements
We are grateful to Sebastian Oehm for a careful reading of this manuscript. Morris Michael and Marco Rüther compiled the showcase of classical dynamic programming algorithms for the ADP web site.
==== Refs
Gusfield D Algorithms on Strings, Trees, and Sequences 1997 Cambridge University Press
Durbin R Eddy S Krogh A Mitchison G Biological Sequence Analysis 1998 Cambridge University Press
Bafna V Edwards N On de novo interpretation of tandem mass spectra for peptide identification Proceedings of the seventh annual international conference on Computational molecular biology 2003 ACM Press 9 18
Algebraic Dynamic Programming homepage
Giegerich R Meyer C Steffen P A Discipline of Dynamic Programming over Sequence Data Science of Computer Programming 2004 51 215 263
Nussinov R Pieczenik G Griggs J Kleitman D Algorithms for loop matchings SIAM J Appl Math 1978 35 68 82
Bellman R Dynamic Programming 1957 Princeton University Press
Wuchty S Fontana W Hofacker IL Schuster P Complete Suboptimal Folding of RNA and the Stability of Secondary Structures Biopolymers 1999 49 145 165 10070264
Morin TL Monotonicity and the principle of optimality J Math Anal Appl 1982 86 665 674
Giegerich R Voss B Rehmsmeier M Abstract Shapes of RNA Nucleic Acids Res 2004 32 4843 4851 15371549
Giegerich R Explaining and Controlling Ambiguity in Dynamic Programming Proc Combinatorial Pattern Matching 2000 Springer LNCS 1848 46 59
Dowell RD Eddy SR Evalution of lightweight stochastic context-free grammars for RNA secondary structure prediction BMC Bioinformatics 2004 5 71 15180907
Reeder J Steffen P Giegerich R Effective ambiguity checking in biosequence analysis BMC Bioinformatics 2005 6 153 15967024
Reeder J Giegerich R Design, implementation and evaluation of a practical pseudoknot folding algorithm based on thermodynamics BMC Bioinformatics 2004 5 104 15294028
Rehmsmeier M Steffen P Höchsmann M Giegerich R Fast and effective prediction of microRNA/target duplexes RNA 2004 10 1507 1517 15383676
Reeder J Giegerich R Consensus Shapes: an alternative to the Sankoff Algorithm for RNA consensus structure prediction Bioinformatics 2005 21 3516 23 [Advance access] 16020472
|
16156887
|
PMC1261154
|
CC BY
|
2021-01-04 16:27:27
|
no
|
BMC Bioinformatics. 2005 Sep 12; 6:224
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-224
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2251615689610.1186/1471-2105-6-225Methodology ArticlePathway level analysis of gene expression using singular value decomposition Tomfohr John [email protected] Jun [email protected] Thomas B [email protected] Department of Biostatistics and Bioinformatics and Center for Bioinformatics and Computational Biology, Institute for Genome Sciences and Policy, Duke University, Durham, North Carolina 27708, USA2005 12 9 2005 6 225 225 27 1 2005 12 9 2005 Copyright © 2005 Tomfohr et al; licensee BioMed Central Ltd.2005Tomfohr 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
A promising direction in the analysis of gene expression focuses on the changes in expression of specific predefined sets of genes that are known in advance to be related (e.g., genes coding for proteins involved in cellular pathways or complexes). Such an analysis can reveal features that are not easily visible from the variations in the individual genes and can lead to a picture of expression that is more biologically transparent and accessible to interpretation. In this article, we present a new method of this kind that operates by quantifying the level of 'activity' of each pathway in different samples. The activity levels, which are derived from singular value decompositions, form the basis for statistical comparisons and other applications.
Results
We demonstrate our approach using expression data from a study of type 2 diabetes and another of the influence of cigarette smoke on gene expression in airway epithelia. A number of interesting pathways are identified in comparisons between smokers and non-smokers including ones related to nicotine metabolism, mucus production, and glutathione metabolism. A comparison with results from the related approach, 'gene-set enrichment analysis', is also provided.
Conclusion
Our method offers a flexible basis for identifying differentially expressed pathways from gene expression data. The results of a pathway-based analysis can be complementary to those obtained from one more focused on individual genes. A web program PLAGE (Pathway Level Analysis of Gene Expression) for performing the kinds of analyses described here is accessible at .
==== Body
Background
Gene expression microarrays provide a snapshot of the expression levels of thousands of genes within a cell or tissue sample. A persistent challenge is to interpret this data: to identify key genes or patterns of expression associated with some condition and so to gain valuable clues about the biological processes related to that condition.
While a variety of methods have been developed to identify significant changes in the expression of individual genes [1-4], another useful perspective can be gained by viewing expression data at the level of groups of related genes. One approach along these lines identifies similarities, such as shared pathways or GO annotations [5], between genes previously identified in an individual gene analysis [6,7]. A potential problem is that this approach relies on the individual genes within a category of interest to stand out. Modest but consistent changes in the expression of a group of related genes could be missed if relatively few of the individual genes appear significant.
A promising alternative focuses at the outset on identifying significantly differently expressed groups of genes from a collection of predefined sets of genes (e.g., pathways and complexes) [8,9]. The usefulness of such an approach was strikingly demonstrated by Mootha et al. [9] in a study of gene expression profiles of muscle in type 2 diabetics (DM2). As reported by them, no single gene showed up as significant in a comparison between DM2s and subjects with normal glucose tolerance (NGT). Their 'gene-set enrichment analysis' (GSEA), however, uncovered a set of genes involved in oxidative phosphorylation as being significantly downregulated in DM2 vs. NGT.
In this article we present a new pathway based approach to the analysis of gene expression that, while similar in spirit to GSEA, has a number of potential advantages. Briefly, GSEA involves ranking all the genes (for example, by significance level in a two-group comparison) and then calculating an 'enrichment score' (ES) for each pathway that depends on the rankings of its member genes. Our method instead begins by translating gene expression levels into pathway 'activity' levels, which are derived from singular value decompositions (SVD). The activity levels are used for making comparisons and in general can be used in the same kinds of applications as gene expression levels.
We demonstrate the approach using the same expression data analyzed by Mootha et al. [9] in their study of type 2 diabetes, and also with expression data from airway epithelia of smokers and non-smokers [10]. Our analysis leads us to conclusions similar to those obtained using GSEA in the diabetes set, but overall appears to perform better in identifying differentially expressed pathways in comparisons between smokers and non-smokers.
The results presented in this article, including statistics for pathways and colormaps of expression profiles, were obtained using a web program we have developed called PLAGE (Pathway Level Analysis of Gene Expression) [11].
Results and discussion
Outline of the method
In the next two sections we analyze gene expression data from skeletal muscle of type 2 diabetics and airway epithelia of different types of smokers. Here we give a brief overview of our approach. A more detailed description is given in the Methods section.
The method is outlined in Fig. 1. The analysis is based on a predefined collection of pathways (e.g., sets of genes coding for proteins involved in specific metabolic or signaling pathways). We use a collection of about 400 pathways obtained from the KEGG (Kyoto Encyclopedia of Genes and Genomes) and Biocarta websites [12,13].
Figure 1 Outline of pathway level analysis of gene expression.
The main goal is to determine, based on the gene expression data, which (if any) of the pathways are associated with some variable of interest such as disease status. To address this, we start by calculating activity levels for each pathway within the samples (in this article, each sample is the gene expression profile in a tissue sample from one individual).
We define the activity level in terms of the first eigenvector, 'metagene', in the singular value decomposition (SVD) of the matrix of expression levels Y (Fig. 2A). The expression matrix is restricted, however, to include only those genes within one predefined pathway at a time. This restriction is one of the main differences from previous applications of SVD (e.g., [14,15]) to gene expression analysis.
Figure 2 Pathway activity levels. Schematic illustration of our approach to quantifying the activity level of a pathway. (A) A colormap of the expression levels for the genes in a hypothetical pathway after standardizing the expression levels to have zero mean and unit variance over samples. This represents the matrix Y described in the text. (B) The main component of the variation in the expression matrix depicted in (A). This representation is determined by the activity levels c and weights w (see Methods) associated with the first metagene in the singular value decomposition (SVD) of Y . The activity level in a sample (one column of the expression matrix) can be thought of as specifying a location in the range of expression profiles shown in (C). Positive activity levels here indicate relatively high (low) expression for genes with positive (negative) weight. For example, the expression profile (column) furthest to the left in the expression matrix is in the high positive region of the range of expression profiles. The colormaps in (A) and (B) show the samples divided into two hypothetical groups (e.g., case samples and control samples). We note, however, that the matrix Y contains expression values for all samples: the activity levels are determined by performing SVD using expression data from all samples without regard to how the samples are classified.
As a gene expression value represents the level of expression of a gene in some sample, the activity level represents the 'level' of the first metagene in a sample. The first metagene is simply a vector of weights, one weight for each gene, and a positive activity level indicates the relatively high (low) expression of genes with positive (negative) weight; a negative activity level indicates the reverse. The activity level in a given sample can be thought of as specifying the position of an expression profile (one column of the expression matrix) in a range of possible profiles as shown in Fig. 2C.
The main motivation for using the first metagene from SVD to define the activity level is that the weights (first metagene) and associated activity levels together capture the main component of the variation in the full expression matrix Y (Fig. 2B depicts the main component of the variation for the expression matrix in Fig. 2A). Generally, the higher metagenes may also contain meaningful structure but in this article we focus only on the first component of the variation. It may be useful in the future to devise a scheme for extending the analysis to higher metagenes.
Once the activity levels are determined, they can be used in the same kinds of applications as gene expression levels. For example, we might ask which pathways have activity levels that are significantly higher in samples from a case group (e.g., diabetic) than those from a control group. In this article, we mainly perform simple two-group comparisons using t statistics (see Methods) but, in principle, any statistical model for the level of expression of individual genes is adaptable to one for activity levels.
Type 2 diabetes
The data set analyzed by Mootha et al. [9] contains gene expression profiles in muscle tissue for each of 17 type 2 diabetics (DM2), 17 subjects with normal glucose tolerance (NGT), and 9 with impaired glucose tolerance (IGT). The gene expression data is available at the Whitehead Institute Center for Genome Research website [16] along with phenotype data including, for example, ages, body size measurements, blood glucose levels after oral glucose tolerance test (OGTT), and other information. More details can be found in Ref. [9].
The calculations of activity levels were done using all 43 gene expression profiles. Using the t statistic to compare mean activity levels identified no pathways showing apparently different expression between any of the groups DM2, IGT, and NGT. Specifically, the pathway with the highest significance level, 'Activation of cAMP-dependent protein kinase, PKA (protein kinase A)' (Biocarta) was found in comparing DM2 and NGT (upregulated in DM2); the significance level, however, was calculated to be only p = 0.4.
We then looked for correlation between activity levels and potentially more informative variables including blood glucose levels after oral glucose tolerance test (OGTT) and VO2max (a measure of maximum oxygen utilization). The most significant pathway identified here was Oxidative Phosphorylation (KEGG); the Oxidative Phosphorylation activity level was found to be significantly negatively correlated with blood glucose concentration 1 hour after OGTT (Pearson correlation r = -0.52 and p = 0.03). Other pathways found to be correlated with glucose levels 1 hour after OGTT include, in order of significance, Biocarta 'Activation of cAMP-dependent protein kinase, PKA (protein kinase A)' (r = +0.47, p = 0.06), and KEGG 'ATP synthesis' (r = -0.43, p = 0.17). The genes in the ATP synthesis pathway, however, are entirely contained within the genes of the KEGG oxidative phosphorylation pathway. The results from these comparisons are summarized in Table 1. A more detailed statistical analysis of the relationship between glucose levels, oxidative phosphorylation, and diabetic status is presented in Fig. 3 and refines the result obtained from the initial analysis.
Table 1 Pathways correlated with a type 2 diabetic phenotype. The table shows p-values for the three pathways most correlated with blood glucose concentration as measured two hours after an oral glucose tolerance test (OGTT). r is the Pearson correlation. Also shown are p-values, generally indicating low significance levels, for these pathways determined from t-statistic comparisons between DM2 (type 2 diabetic) and NGT (normal glucose tolerance).
Comparison
(genes in data set/total genes in pathway) Pathway glucose after OGTT DM2 vs. NGT
(96/123) Oxidative phosphorylation 0.031 (r = -0.517) 0.565 (down in DM2)
(6/6) Activation of cAMP-dependent protein kinase 0.062 (r = +0.474) 0.395 (up in DM2)
(35/40) ATP synthesis 0.166 (r = -0.432) 0.855 (down in DM2)
Figure 3 Negative correlation between oxidative phosphorylation and blood glucose levels after OGTT. Scatter plot of blood glucose levels 2 hours after OGTT vs. oxidative phosphorylation activity levels. The three subject groups – type 2 diabetic (DM2), normal glucose tolerance (NGT), and impaired glucose tolerance (IGT) – are distinguished by color; solid lines show the first principal component for each group independent of the others. Group means are shown in black squares. The inset shows the 95% confidence intervals for the linear correlation coefficients for each group. Negative correlation between glucose levels and oxidative phosphorylation reaches statistical significance only within DM2 subjects.
The connection between blood glucose levels and oxidative phosphorylation (and ATP synthesis) seems biologically reasonable. Oxidative phosphorylation makes up the last few steps in the series of reactions leading to the synthesis of ATP from the oxidation of glucose. Unusually high blood glucose levels imply a relatively low activity of this pathway. Protein kinase A (PKA) facilitates the breakdown of glycogen into glucose in skeletal muscle cells [17]; an elevated PKA pathway in skeletal muscle of subjects exhibiting a type 2 diabetic phenotype may then possibly be interpreted as a cellular reaction to glucose starvation.
We note that our results differ slightly from those of Mootha et al. [9] who, using GSEA, found a significant (p = 0.029) downregulation of oxidative phosphorylation genes in diabetics compared to non. In contrast, we find reasonable significance only in the level of correlation with blood glucose levels after OGTT. This and other differences with GSEA are discussed in a comparison below.
Effects of smoking on airway epithelia
A recent study examined the effects of smoking on gene expression in airway epithelia [10]. Expression data were obtained from a large number of subjects including former and current smokers, and those who have never smoked. The study identified genes differentially expressed between the different groups and some of the general functional categories represented by these genes. A pathway based analysis can be complementary to the kind already given by drawing attention to groups of genes involved in more specific cellular processes.
We obtained the gene expression data for the smoking study from the Airway Gene Expression Database (AGED) [18]. The data consist of gene expression profiles from 75 subjects including 34 current smokers, 18 former smokers, and 23 subjects who have never been smokers. The data set has undergone some preprocessing steps including normalization and filtering for genes detected on the microarrays. More details can be found at the AGED website [18] and in Ref. [10].
We identified pathways most differentially expressed in two-group comparisons using t statistics (see Methods). Table 2 shows the top ranking pathways from these comparisons and p-values. The top of Fig. 4 shows a colormap of the activity levels for these pathways in the different samples. The bottom of Fig. 4 shows a colormap for the expression levels of the genes in the KEGG glutathione metabolism pathway; colormaps for the other pathways can be viewed at the PLAGE website [11].
Table 2 Top pathways identified in comparisons between current (C), former (F), and never (N) smokers. Pathways identified as differentially expressed with p < 0.05 using pathway activity levels in comparisons between smokers and non-smokers. The p-values were determined using 10,000 random permutations as described in the Methods section. p < 0.0001 means no pathway in any of the 10,000 permutations showed higher significance. We note that the change (up or down) is determined, somewhat arbitrarily, by the average expression level captured by the first metagene. Specifically, the pathway is called 'up' if the average of ∑icjwi is greater in the first group (e.g., C in 'C vs. N') than in the second. A given pathway, however, will typically have some genes with higher and some with lower mean expression in one group as compared to another.
Comparison (genes in data set/total genes in pathway) Pathway p change
C vs. N (14/45) gamma-Hexachlorocyclohexane degradation <0.0001 down
(15/39) Prostaglandin and leukotriene metabolism <0.0001 up
(11/24) O-Glycans biosynthesis <0.0001 up
(6/21) Pentose and glucuronate interconversions <0.0001 up
(24/34) Glutathione metabolism <0.0001 up
(3/12) Lectin Induced Complement Pathway 0.0004 down
(11/19) Chaperones modulate interferon Signaling Pathway 0.0006 up
(6/15) TACI and BCMA stimulation of B cell immune responses. 0.0044 down
(3/6) Tetrachloroethene degradation 0.0054 up
(3/6) FXR and LXR Regulation of Cholesterol Metabolism 0.0062 down
(4/7) TSP-1 Induced Apoptosis in Microvascular Endothelial Cell 0.0065 down
(16/28) Galactose metabolism 0.0067 down
(13/20) Biosynthesis of steroids 0.0164 up
(7/11) Map Kinase Inactivation of SMRT Corepressor 0.0274 down
(25/68) Nicotinate and nicotinamide metabolism 0.0279 up
(4/14) Classical Complement Pathway 0.0314 down
(6/19) Complement Pathway 0.0364 down
(9/13) Nucleotide sugars metabolism 0.0369 up
(3/3) Degradation of the RAR and RXR by the proteasome 0.0396 down
F vs. C (3/12) Lectin Induced Complement Pathway 0.0068 up
(13/20) Biosynthesis of steroids 0.0083 down
(24/34) Glutathione metabolism 0.0403 down
F vs. N (14/45) gamma-Hexachlorocyclohexane degradation 0.0321 down
Figure 4 Expression profiles in airway epithelia of current (C), former (F), and never (N) smokers. Top: colormap of pathway activity levels for the highest ranking pathways in the comparison between current smokers and never smokers. Bottom: colormap for genes in the KEGG glutathione metabolism pathway. Glutathione is an important anti-oxidant known to be increased in the lungs of smokers. The genes with the highest weights in this pathway, GCLM and GCLC, encode the subunits of glutamate cysteine ligase (GCL), the rate-limiting enzyme in the synthesis of glutathione [24].
We also performed one-way ANOVA using pathway activity levels and found the top pathways from this analysis (those with the smallest p-values) to be virtually the same as those identified using t statistics.
The pathway most significantly differentially expressed in both current and former smokers as compared to never smokers is 'gamma-Hexachlorocyclohexane degradation' (KEGG). Genes in this pathway and corresponding weights from SVD are given in Table 3. Gamma-Hexachlorocyclohexane, also known as lindane, is a toxic insecticide and the connection with smoking is not at first clear. Several genes in the gamma-Hexachlorocyclohexane degradation pathway, however, have known associations with smoking. The gene with the highest absolute weight, CYP2A6, plays a key role in the metabolism of nicotine and a number of studies have indicated an important relationship between CYP2A6 and smoking [19,20]. For example, individuals with inactive CYP2A6 alleles have been reported to be at lower risk for becoming dependent on cigarettes [21]. Our analysis highlights the potential importance of a cellular process in connection with smoking and further, by the high absolute weight assigned to CYP2A6, points to the key role played by CYP2A6 in this process.
Table 3 Weights for genes in two pathways that show an association with smoking status. SVD weights for genes in pathways identified as significantly differentially expressed between current and never smokers. The gene with highest absolute weight in the gamma-Hexachlorocyclohexane degradation pathway, CYP2A6, plays a key role in nicotine metabolism and has been linked to nicotine dependence [21]. GALNT3, a gene with relatively high weight in the O-Glycans biosynthesis pathway, initiates mucin-type O-glycosylation [22], suggesting a connection with the increased sputum production observed in smokers. The overall sign of the weights has here been chosen so that a positive weight implies relatively higher expression in current as compared to never smokers.
Pathway gene weight
gamma-Hexachlorocyclohexane degradation CYP2A6 -0.42
CYP2F1 -0.38
CYP1B1 0.38
CYP2A7 -0.38
ALPL 0.28
CYP2B6 -0.27
CYP4B1 -0.27
CYP2J2 -0.23
PON2 0.19
ACP5 0.15
ACP2 0.15
ACP1 0.15
CYP2C9 0.09
CYP1A2 -0.02
O-Glycans biosynthesis GALNT7 0.50
GALNT1 0.42
GALNT3 0.42
B4GALT4 0.38
GALNT6 0.28
B4GALT1 0.23
GALNT2 0.21
C1GALT1 0.20
OGT -0.15
B4GALT3 0.11
B4GALT2 0.05
Other pathways upregulated in current vs. never smokers include metabolism of prostaglandins and leukotrienes (associated with pain response and inflammation), O-Glycans biosynthesis, and glutathione metabolism.
Genes from the O-Glycans pathway are listed in Table 3. Several of these are also within the list of genes identified by Spira et al. [10] The present analysis, however, reveals more directly the functional relationship shared by the genes and draws special attention to the role of this pathway in the physiological response to smoking. O-linked glycosylation plays an important role in the production of proteoglycans, some of which are constituents of mucus [17]. In particular, one of the O-Glycans genes GALNT3 initiates mucin-type O-glycosylation [22]. It therefore seems plausible that the elevated activity of the O-Glycans pathway is linked to the increased sputum production observed in smokers [23].
Glutathione is an important antioxidant known to be increased in the lungs of smokers [24]. A colormap of the expression levels of the genes in the glutathione metabolism pathway is shown in Fig. 4; the higher expression of many of these genes in the current smokers is clear. The increased expression of glutathione metabolism genes in smokers was also noted by Spira et al. [10]. Interestingly, the two genes with the highest weights are GCLM and GCLC; these genes encode the subunits of glutamate cysteine ligase (GCL), which is the rate-limiting enzyme in the synthesis of glutathione [24].
Comparison with gene-set enrichment analysis
We used GSEA to identify significant pathways in comparisons between the different groups of smokers. The ranking of genes required to evaluate enrichment scores was done, following Ref. [9], using the signal to noise ratio (the absolute value of the difference of the means of the two groups divided by the sum of the within-group standard deviations). For this analysis we used our collection of pathways and complexes. In comparing current and never smokers, the gene-set with the highest enrichment score (ES) was determined to be the KEGG ribosome genes with an ES of 162. However, using 1000 random permutations as described in Ref. [9] to evaluate the significance of this ES yielded a p-value of only 0.17. Comparing former and never smokers, GSEA also finds the ribosome gene-set to have the highest ES (304) and with a reasonably high significance level of p = 0.047 (1000 permutations). Finally, the comparison between current and former smokers identifies 'Lectin Induced Complement Pathway' (Biocarta) as the pathway with the highest ES (128) but the significance level is very low (p = 0.52).
We also used GSEA with our collection of gene-sets to compare DM2 and NGT in the diabetes data discussed above. Here the ribosome gene-set is found to have the highest ES (278) and p = 0.073 (1000 permutations). Oxidative phosphorylation has the next highest ES (261) with p = 0.092.
It is interesting that in three of our four comparisons, the ribosome gene-set is found to have the highest ES. We noticed that the expression levels for the ribosome genes are very strongly correlated. This can be seen very clearly in colormaps of the expression levels in both the diabetes and smoking data sets (not shown) and is also implied by an unusually large first eigenvalue. For example, in the smoking data set the first metagene for ribosome accounts for over fifty percent of the total variation, = 0.53. In comparison, this fraction is only 0.23 for Oxidative Phosphorylation (which has 87 genes vs. the ribosome's 80 genes present in the data set) and 0.05 for uncorrelated gaussian noise obtained from averaging over 1000 simulated data sets of the same size as the ribosome. These observations lead us to speculate that GSEA may not effectively distinguish between gene sets that show a consistent difference between groups of samples and sets with genes that are merely strongly coordinately expressed. In the method we have proposed, correlated expression patterns influence the form of the first metagene and the pathway activity levels, but this is treated separately from the question of whether there is a consistent difference between groups.
To summarize, our approach suggests a number of potentially interesting pathways and hypotheses from comparisons between smokers and non-smokers. The results of our analysis seem biologically reasonable and are more consistent with the findings of Spira et al. [10] than those obtained by GSEA. As a specific example, we identified glutathione metabolism genes as having significantly (p < 0.0001) higher expression levels in current vs. never smokers. This feature was also noted by Spira et al. [10] in their original analysis and seems evident in the colormap shown in Fig. 4. GSEA, however, does not find this or any other pathway to be differentially expressed below p = 0.05 between current and never smokers.
Conclusion
We have introduced a method for analyzing gene expression data in terms of a collection of predefined pathways and complexes. In summary, the approach quantifies the level of activity of each pathway within each sample and uses the activity levels as the basis for making comparisons. Mootha et al. [9] previously demonstrated that looking at expression data in terms of predefined pathways can provide valuable insights not easily attainable by methods more focused on individual genes and we feel that the results in this article reinforce their view.
The main applications we have presented are simple two-group comparisons. We suggest that pathway activity levels may also be useful in other areas, for example, as variables in models for gene regulation (e.g., Bayesian networks [25]), for clustering and classification [15,26], and in computational efforts to discover novel gene-pathway associations. One possibility is to use SVD on the activity levels themselves to determine 'metapathway' signatures for the status of some disease following the approach described in Ref. [15]. There are a number of potential advantages to be gained in these and other cases. For one, features defined by groups of genes will tend to be more robust in the face of variation at the level of individual genes. In addition, the features here (pathways) are based on a large amount of prior biological knowledge and so are potentially more directly biologically meaningful.
Methods
Gene sets
As in GSEA, the basis for our analysis is a classification of genes into known pathways and complexes. Here, we use about 400 pathways and complexes characterized on the Biocarta and KEGG (Kyoto Encyclopedia of Genes and Genomes) websites [12,13]. The KEGG pathways mainly include processes related to metabolism and biosynthesis (e.g., Fatty acid metabolism and Ubiquinone biosynthesis). Those on Biocarta cover a wider variety of cellular processes including a large number of immune signaling pathways (Toll-like receptor pathway, B-cell receptor signaling pathway, complement pathway), as well as metabolic and biosynthetic pathways. Both collections also include complexes such as the T-cell receptor and ribosome and broad gene categories such as cytokines. The KEGG and Biocarta pathways continue to expand and evolve. The versions used for this article are available at the PLAGE website [11].
Pathway activity levels
Our analysis starts by quantifying, in each sample, the level of 'activity' of each pathway. As described in the Results and discussion section, we define the activity level of a pathway in a given sample as the 'level' of expression of a certain metagene in that sample, i.e., the level of the first metagene from the SVD (e.g., Ref [27]) of the matrix of expression levels. In this section we provide the details and motivation behind this choice.
We begin by standardizing the gene expression levels to have zero mean and unit variance over samples. For each pathway, we form a matrix Y (rows = genes, columns = samples) containing the standardized expression levels from all samples but for the genes in that pathway only. We write the singular value decomposition of Y as
Y = WDC. (1)
Here the columns of the matrix W are the orthonormal (W⊤W = I, the identity matrix) eigenvectors or metagenes of Y, D is a diagonal matrix containing the associated eigenvalues, and each column of C is a vector of coefficients for one of the samples indicating the level of each metagene in the sample. The rows of C are also orthonormal (CC⊤ = I). Assume the eigenvalues are ordered from highest to lowest going down the diagonal of D. The first metagene w – that associated with the largest eigenvalue – is then the first column of W. We write its eigenvalue as λ and the associated coefficients (first row of C) as cj.
The activity level of a pathway in a given sample j is taken as the coefficient cj for the first metagene. It follows also from the orthonormality of the columns of W and rows of C that
That is, the activity level cj can also be regarded (up to a non-essential scale factor) as a weighted sum of the standardized expression levels of the individual genes, the weights being given by the first metagene w.
One motivation for using the first metagene in SVD is that the resulting combination of activity levels and weights specifies an optimal approximation to the matrix Y (i.e., accounts for the main component of the variation in the data). Specifically, assume the following statistical model for the expression levels
yij = αiχj + εij (3)
where the vector χ is constrained to have unit norm and the εij are independent Gaussian random variables. The estimates for α and χ that minimize the sum of the squared errors are just the first metagene scaled by its eigenvalue, λw, and the associated vector of activity levels c, respectively. The approximation of a set of expression profiles using the first metagene is illustrated in Fig. 2.
A useful fact about the first eigenvalue is that its square is a measure of the amount of variation accounted for by the first metagene. Specifically, with ng = number of genes and ns = number of samples, the total amount of variation in the data is (recall, the expression levels are standardized so that ) and the variation remaining after subtracting off the profile described by the first metagene is ∑ij(yij - λwicj)2 = ng(ns - 1) - λ2.
Evaluating significance
In this article, we mainly perform pairwise comparisons to identify pathways for which the mean activity level in one group (e.g., diabetic) is significantly different from that in the other (non-diabetic). To accomplish this we calculate a t statistic for each pathway:
where A and B are labels for the groups, nA is the number of samples in group A, and μA and VA are the mean and variance of the activity level in A (similarly for group B). To determine a fair measure of significance in an analysis of this kind, it is important to account for the fact that we are testing a large number of hypotheses [28]. For this purpose, we perform a large number (10,000 for this article) of comparisons by randomly permuting the sample labels and for each permutation recording the t statistic for the most significant pathway (the maximum t-statistic) identified using the permuted labels. Our p-values are computed with reference to the maximum t-statistic, i.e., we calculate a p-value by taking the fraction of the maximum t-statistics that exceed the t-statistic.
This basic approach was also used by Mootha et al. [9] to determine the significance of the ES (enrichment score) for a pathway except that the comparison there is with maximum enrichment scores. Generally, other statistics can be used in similar fashion. For example, to evaluate the significance of the level of correlation of activity levels with blood glucose concentration in the diabetes data, we determined the minimum and maximum (Pearson) correlations between activity levels and randomly permuted glucose concentration values in each of 10,000 permutations. p-values were obtained as the fraction of these extremal correlation values that were stronger (higher if r > 0 and lower if r < 0) than the value obtained using the correct ordering of glucose levels.
Availability and requirements
The SVD-based pathways analysis method has been implemented in a web program [11] called PLAGE (pathway level analysis of gene expression). PLAGE will run through standard web browsers.
Authors' contributions
JT, JL, and TBK all contributed to the development of the method and the writing of the manuscript. JT wrote the web program.
Acknowledgements
We gratefully acknowledge the financial support of the NIH through the Duke University Center for Translational Research (5 P30 AI051445-03) and of the NSF through a grant to David Bird (NCSU; DBI 0077503) as well as the Duke Center for Bioinformatics and Computational Biology. We also thank two anonymous reviewers for many helpful suggestions.
==== Refs
Tusher VG Tibshirani R Chu G Significance analysis of microarrays applied to the ionizing radiation response Proc Natl Acad Sci USA 2001 98 5116 21 11309499 10.1073/pnas.091062498
Efron B Tibshirani R Storey JD Tusher V Empirical Bayes analysis of a microarray experiment J Amer Statist Assoc 2001 96 1151 1160 10.1198/016214501753382129
Baldi P Long AD A Bayesian framework for the analysis of microarray expression data: regularized t-test and statistical inferences of gene changes Bioinformatics 2001 17 509 19 11395427 10.1093/bioinformatics/17.6.509
Kepler TB Crosby L Morgan KT Normalization and analysis of DNA microarray data by self-consistency and local regression Genome Biol 2002 3 research0037 12184811 10.1186/gb-2002-3-7-research0037
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Harris MA Hill DP Issel-Tarver L Kasarskis A Lewis S Matese JC Richardson JE Ringwald M Rubin GM Sherlock G Gene ontology: tool for the unification of biology. The Gene Ontology Consortium Nat Genet 2000 25 25 9 10802651 10.1038/75556
Doniger SW Salomonis N Dahlquist KD Vranizan K Lawlor SC Conklin BR MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-expression profile from microarray data Genome Biol 2003 4 R7 12540299 10.1186/gb-2003-4-1-r7
Zeeberg BR Feng W Wang G Wang MD Fojo AT Sunshine M Narasimhan S Kane DW Reinhold WC Lababidi S Bussey KJ Riss J Barrett JC Weinstein JN GoMiner: a resource for biological interpretation of genomic and proteomic data Genome Biol 2003 4 R28 12702209 10.1186/gb-2003-4-4-r28
Kurhekar MP Adak S Jhunjhunwala S Raghupathy K Genome-wide pathway analysis and visualization using gene expression data Pac Symp Biocomput 2002 462 73 11928499
Mootha VK Lindgren CM Eriksson KF Subramanian A Sihag S Lehar J Puigserver P Carlsson E Ridderstrale M Laurila E Houstis N Daly MJ Patterson N Mesirov JP Golub TR Tamayo P Spiegelman B Lander ES Hirschhorn JN Altshuler D Groop LC PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes Nat Genet 2003 34 267 73 12808457 10.1038/ng1180
Spira A Beane J Shah V Liu G Schembri F Yang X Palma J Brody JS Effects of cigarette smoke on the human airway epithelial cell transcriptome Proc Natl Acad Sci USA 2004 101 10143 8 15210990 10.1073/pnas.0401422101
Pathway Level Analysis of Gene Expression
Kyoto Encyclopedia of Genes and Genomes
Biocarta
Alter O Brown PO Botstein D Singular value decomposition for genome-wide expression data processing and modeling Proc Natl Acad Sci USA 2000 97 10101 6 10963673 10.1073/pnas.97.18.10101
West M Blanchette C Dressman H Huang E Ishida S Spang R Zuzan H Olson JAJ Marks JR Nevins JR Predicting the clinical status of human breast cancer by using gene expression profiles Proc Natl Acad Sci USA 2001 98 11462 7 11562467 10.1073/pnas.201162998
Whitehead Institute Center for Genome Research website
Alberts B Johnson A Lewis J Raff M Roberts K Walter P Molecular Biology of the Cell 2002 New York: Garland
Airway Gene Expression Database
Sellers EM Tyndale RF Mimicking gene defects to treat drug dependence Ann NY Acad Sci 2000 909 233 46 10911933
Ariyoshi N Miyamoto M Umetsu Y Kunitoh H Dosaka-Akita H Sawamura Y Yokota J Nemoto N Sato K Kamataki T Genetic polymorphism of CYP2A6 gene and tobacco-induced lung cancer risk in male smokers Cancer Epidemiol Biomarkers Prev 2002 11 890 4 12223434
Tyndale RF Sellers EM Variable CYP2A6-mediated nicotine metabolism alters smoking behavior and risk Drug Metab Dispos 2001 29 548 52 11259349
Gu C Oyama T Osaki T Li J Takenoyama M Izumi H Sugio K Kohno K K Y Low expression of polypeptide GalNAc N-acetylgalactosaminyl transferase-3 in lung adenocarcinoma: impact on poor prognosis and early recurrence Br J Cancer 2004 90 436 42 14735190 10.1038/sj.bjc.6601531
Dwyer TM Cigarette smoke-induced airway inflammation as sampled by the expired breath condensate Am J Med Sci 2003 326 174 8 14557729 10.1097/00000441-200310000-00004
Neurohr C Lenz AG Ding I Leuchte H Kolbe T Behr J Glutamate-cysteine ligase modulatory subunit in BAL alveolar macrophages of healthy smokers Eur Respir J 2003 22 82 7 12882455 10.1183/09031936.03.00080403
Friedman N Inferring cellular networks using probabilistic graphical models Science 2004 303 799 805 14764868 10.1126/science.1094068
Eisen MB Spellman PT Brown PO Botstein D Cluster analysis and display of genome-wide expression patterns Proc Natl Acad Sci USA 1998 95 14863 8 9843981 10.1073/pnas.95.25.14863
Strang G Introduction to Linear Algebra 2003 Cambridge: Wellesley
Dudoit S Shaffer JP Boldrick JC Multiple hypothesis testing in microarray experiments Stat Sci 2003 18 71 103 10.1214/ss/1056397487
|
16156896
|
PMC1261155
|
CC BY
|
2021-01-04 16:27:27
|
no
|
BMC Bioinformatics. 2005 Sep 12; 6:225
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-225
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2301617152810.1186/1471-2105-6-230Research ArticleChromosomal clustering of a human transcriptome reveals regulatory background Vogel Jan H [email protected] Heydebreck Anja [email protected] Antje [email protected] Silke [email protected] Cardiovascular Genetics Group, Department of Vertebrate Genomics, Max Planck Institute for Molecular Genetics, Ihnestrasse 73, 14195 Berlin, Germany2 Department of Computational Molecular Biology, Max Planck Institute for Molecular Genetics, Ihnestrasse 73, 14195 Berlin, Germany2005 19 9 2005 6 230 230 17 5 2005 19 9 2005 Copyright © 2005 Vogel et al; licensee BioMed Central Ltd.2005Vogel 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
There has been much evidence recently for a link between transcriptional regulation and chromosomal gene order, but the relationship between genomic organization, regulation and gene function in higher eukaryotes remains to be precisely defined.
Results
Here, we present evidence for organization of a large proportion of a human transcriptome into gene clusters throughout the genome, which are partly regulated by the same transcription factors, share biological functions and are characterized by non-housekeeping genes. This analysis was based on the cardiac transcriptome identified by our genome-wide array analysis of 55 human heart samples. We found 37% of these genes to be arranged mainly in adjacent pairs or triplets. A significant number of pairs of adjacent genes are putatively regulated by common transcription factors (p = 0.02). Furthermore, these gene pairs share a significant number of GO functional classification terms. We show that the human cardiac transcriptome is organized into many small clusters across the whole genome, rather than being concentrated in a few larger clusters.
Conclusion
Our findings suggest that genes expressed in concert are organized in a linear arrangement for coordinated regulation. Determining the relationship between gene arrangement, regulation and nuclear organization as well as gene function will have broad biological implications.
==== Body
Background
To understand the global regulatory network underlying specific transcriptomes, several distinct aspects have to be considered [1,2]; (A) the genomic organization of those transcripts [3], (B) their regulation by general and specific transcription factors, (C) the influence of epigenetic effects such as e.g. histone modifications [4], (D) the local environment [5,6], and (D) the functional role of the transcripts as well as their protein products as nodes of the network. In the present report, we show for the first time for a human transcriptome that there is a relationship between the genomic organization, transcriptional regulation and functional role.
It has long been known that transcriptional regulation is related to chromosomal gene order; prokaryotic operons are the best known example [7]. For lower eukaryotes coexpressed adjacent genes were first described in Saccharomyces cerevisiae [8,9]. For a part of those gene pairs, a common transcriptional activation was proposed through a shared upstream activating sequence, which occurs in the promoter region of one of the two genes. Furthermore, correlated triplets, but not quadruples, were found to occur more often than expected in yeast. Reports for Caenorhabditis elegans [10], Drosophila melanogaster [11,12], Homo sapiens and Mus musculus [13-15] showed coexpression of co-localized genes in higher eukaryotes, and reports of particular gene cluster such as the human β-globin locus [16], the interleukin-13 gene locus [17] and others [18-20] indicate the association to regulated chromatin domains. However, on a global scale only few insights into the molecular mechanisms of the transcriptional regulation of clustered genes have been gathered so far, and data about small clusters of adjacent genes have only been partially analyzed. Beside the finding of housekeeping gene clusters throughout the human genome [14], no evidence for a functional correlation of clustered genes had been shown. However, coexpressed genes in general (regardless of their localization) appear to function in similar biological processes [21].
In this paper we describe the chromosomal co-localization in adjacent pairs of a large proportion of the human transcriptome in heart in the context of their expression dynamics, their transcriptional regulation and their function in shared biological processes. We examined the cardiac transcriptome using data from our previous genome-wide array analysis [22] and found profound evidence for a significant clustering of more than 37% of those genes located mainly in pairs or triplets. A significant proportion of these clustered genes have common putative transcription factor binding sites within their promoter regions and share common biological functions.
Results
To characterize genomic organization of the cardiac transcriptome, we investigated a set of 3.172 heart-expressed genes (HXP) identified in our previous study that represent the cardiac transcriptome based on the analysis of 55 human heart samples [22]. This gene set reflects all genes continuously transcribed in the analyzed heart samples. Thus, we focused on the information whether or not a gene and herewith a particular genomic region is transcribed at all and considered expressed neighboring genes as coexpressed gene clusters regardless of the expression levels of individual genes. We assigned the position of this HXP set with regard to the whole human genome as represented by Ensembl compared to the HU2 gene set represented on the arrays. In order to reflect the actual adjacency of genes on the chromosomes, we defined gene neighbors according to their Ensembl annotation, rather than using the HU2 gene set as a basis (Fig. 1).
Figure 1 Gene localization based on Ensembl, HU2 and HXP. Genes marked with * illustrate an adjacent gene pair in reference to HU2 and HXP, but not referring to the Ensembl gene set (131 cases). Therefore, only genes like those marked with # are noted as clusters throughout.
Chromosomal distribution and gene clusters
First, we analyzed the overall chromosomal distribution of HXP and observed no overrepresentation of HXP on specific chromosomes (Chi-Square-Test, p = 0.2). On average each chromosome contained a proportion of 18% HXP genes out of the overall analyzed dataset. Upon closer analysis of the distinct localization of HXP genes, we observed small groups of physically adjacent genes along the chromosomes throughout the genome. In Figure 2 the chromosomal distribution of the overall HXP genes and neighboring coexpressed HXP genes is represented. We calculated the number of gene clusters made up of two to five physically adjacent HXP genes and measured the statistical significance of this local clustering by comparing the numbers of HXP gene clusters with a random distribution obtained by 100,000 permutations of HU2. We observed a significantly higher number of adjacent gene pairs (881 genes, p = 0.01) and gene triplets (307 genes, p = 0.02) in HXP than would be expected for a random distribution, whereas the number of quadruples and quintuples did not differ significantly (Table 1). In total, we found 1,179 HXP genes to be locally clustered. Further, we analyzed whether there was any prevalence regarding gene orientation within these clusters compared to a random distribution obtained by 10,000 permutations. Besides an enrichment of co-oriented gene pairs within clusters of size ≥ 2 (p = 0.03), we observed no bias in the numbers of anti-oriented gene pairs in clusters ≥ 2 (p = 0.2) as well as for co- and anti-oriented gene triplets in clusters ≥ 3 (p = 0.3 and p = 0.6, respectively).
Figure 2 Genomic organization of the overall and clustered heart-expressed genes. The chromosomal gene order of clustered genes is represented in red, whereas the non-clustered genes are shown in gray, both appear to distribute without notable prevalence among the chromosomes.
Table 1 Numbers of detected heart-expressed gene clusters of different sizes.
Cluster-size Number of gene cluster ENSG p-value Z-score
2 480 881 0.01 3.89
3 79 207 0.02 3.24
4 15 57 0.05 2.68
5 1 5 0.7 0.07
ENSG represents the number of unique Ensembl genes in all clusters of the given size. The p-values and Z-scores result from a comparison with a random distribution obtained by 100,000 permutations of the HU2 dataset.
Gene clusters and housekeeping functionality
Previously, it had been suggested that housekeeping genes are arranged in clusters along the genome [14]. Therefore, we assessed the tissue expression of our HXP gene set and the subset of coexpressed adjacent genes in 79 human tissues, for which the expression information of protein-coding genes had been recorded by the GNF Symatlas [15] (Fig. 3). We observed a slightly bimodal distribution for the overall HXP gene set, with a major peak corresponding to expression in 79 tissues. This distribution differed from the one observed for coexpressed adjacent genes. Here, the majority of genes showed an expression in a distinct number of tissues, not reflecting a housekeeping-like expression profile.
Figure 3 Tissue expression of overall and clustered HXP genes. The numbers of tissue expression of clustered heart-expressed genes are presented filled black. Of the 79 analyzed tissues, the expression profile of clustered genes shows a broad panel of tissue expression but only very few clustered genes are expressed in the majority of tissues. Thus we concluded that clustered coexpressed genes are mainly non-housekeeping genes.
Transcriptional regulation of gene clusters
We extended our analysis to determine to what extent these gene clusters are regulated by common transcription factors. For this purpose, we identified the putative transcription factor binding sites (TFBS) for the HXP set using the CORG database [23]. CORG is based on TFBS in non-coding regions conserved between human and mouse (see Methods), and enabled us to identify binding sites of 276 distinct binding site models in the promoter regions of a total of 1,777 HXP genes. For the remaining HXP genes no conserved non-coding region could be identified.
Taking into account that several models can represent binding sites of one transcription factor, the identified TFBS pertained to only 216 distinct transcription factors. Within the HXP set with assigned TFBS, 501 genes belonged to chromosomally clustered gene pairs and for 171 pairs, binding sites for both genes were predicted [see Additional file 1]. We observed the largest number of common predicted transcription factors for the antisense-sense gene pair consisting of the HOOK2 protein and transcription factor JUNB, which share binding sites for 64 transcription factors. The mean number of observed common transcription factors predicted to regulate adjacent genes was 4.3. There were 23 genes in the HXP set with binding sites for more than 75 different transcription factors, which we consider to be outliers.
Finally, we tested the significance of the observed number of gene pairs potentially regulated by at least one common transcription factor by comparing this number to the numbers of gene pairs with commonly assigned transcription factors seen in 10,000 random permutations of HXP genes. Regardless of whether the calculation was based on the number of distinct TFBS or distinct transcription factors, the common transcriptional regulation was significantly greater within the observed gene clusters than expected from the random distribution (p = 0.02 and p = 0.03, respectively). This significance was not influenced by genes defined as outliers with regard to their large number of common TFBS. Furthermore, the identified common transcription factors belonged to a broad panel of transcription factors classes without prevalence of particular families. Taking the gene orientation into account again, we found a homogenous distribution of strand orientation within observed gene pairs regulated by common transcription factors. An example of a gene cluster regulated by common transcription factors is given in Figure 4.
Figure 4 Example of four adjacent coexpressed genes. Shown is the gene cluster distribution on human chromosome 2 with an example of four adjacent genes regulated in part by common transcription factors and sharing gene ontology categories. Coexpressed gene clusters are shown in red. Each coexpressed gene pair ABI2 – RAPH1; RAPH1 – CD28 and CD28 – CTLA4 shares GO terms as indicated. Genes of the triplet RAPH1 – CD28 – CTLA4 have transcription factor binding sites for AP3 and EV1 in common. Furthermore, each gene pair RAPH1 – CD28 and CD28 – CTLA4 shares additional TFBS. Genes are marked in green, arrows indicate the strand orientation, promotor regions and transcription factors are colored in yellow.
Functional relationship of gene clusters
Further, we analyzed to what extent heart-expressed genes organized as adjacent pairs are involved in the same biological process using the Gene Ontology (GO) database as a source of annotations of biological functions. Naturally, our analysis was limited to those genes annotated with GO terms that were around 60% of all HXP genes (1,921 genes) including those located in 176 gene clusters. A total of 2,158 different GO terms could be assigned to the overall HXP set with 1,241 GO terms mapping to genes located in clusters. To focus on non-generic functional annotations, we calculated the number of GO terms shared within pairs of adjacent HXP genes for the 5th and 6th level of granularity of the GO hierarchy and compared these to a random distribution generated from HXP genes to account for the tissue-specificity of the analyzed dataset. We observed a significant enrichment of shared GO terms within pairs of adjacent genes for both levels of granularity (p << 0.0001 for each). Only eight of the 96 and 45 pairs sharing the 5th and 6th level of functional classification are conspicuous in terms of having originated by gene duplications [see Additional file 2] (for an example see Fig. 4).
Expression pattern of adjacent genes
Finally, we attempted to determine whether genes located close to each other show a correlation of their expression levels within our previously analyzed patient cohort, which consisted of 5 cardiac phenotypes with characteristic expression profiles [22]. First, we built correlation maps for visual inspection of the overall HXP gene set. To construct such matrices, we sorted the HXP genes of each chromosome according to their arrangement on the chromosome and calculated the correlation of expression of each possible gene pair using Pearson's correlation coefficient. We observed groups of coexpressed genes (cor ≥ 0.5) on several chromosomes. As an example, the correlation matrix of human chromosome 10 shows areas of coexpressed genes between genes 7–10 and 27–30 [see Additional file 3]. Correlation maps provide a gross overview of the coexpression of genes located nearby each other. Therefore, we further analyzed the coexpression of gene pairs with respect to their distance on two different scales: the base pair distance and the number of genes located between pairs. For neither of these scales do we observe significant coexpression of co-localized gene pairs.
Discussion
Our data suggest that a large proportion of the cardiac transcriptome in human is linearly arranged in small groups of adjacent genes such that the genes within each group tend to be regulated by the same transcription factors and appear to share granular biological processes. Even though the numbers of shared transcription factors and shared GO terms are small, they are considerable larger than what could be expected by chance. These findings provide powerful evidence that gene clustering plays a potentially important role in concerted gene regulation through the location of regulatory elements with respect of the regulated genes.
We focused our analysis on the information whether or not a gene and herewith a particular genomic region is transcribed at all in the human heart and did not consider rates of transcription. The proportion of genes arranged in clusters exceeded the number reported previously, which could be explained by different definitions of coexpression, as other reports defined coexpression mainly based on the correlation coefficient of continuous valued expression levels or considered only highly expressed genes [13]. Creating an expression neighborhood through the localization of regulatory elements would be an efficient means to increase regional gene activity, which has been shown to be influenced by the local concentration of regulatory proteins [24-26]. In addition, such regulatory elements may influence the expression status within a chromosomal region e.g. by spreading of histone modifications. Further support for concerted regulation of clustered genes is provided by our observation of functional relatedness between clustered genes. Apart from the identification of housekeeping genes arranged in clusters such relatedness has not been reported in human [14]. However without considering gene localization, it has been shown recently that coexpressed genes are often functionally related [21]. Further studies will be required to show if there is an evolutionary constraint upon the disruption of co-localized genes. So far, sequencing projects are still in process and a comparison between close relatives such as human and for example mouse would not be sufficient due to their high synteny. From our analysis, we propose that duplication events alone are not sufficient as an explanation.
In the past some reports suggested that tissue-specific genes, e.g. genes specifically expressed in human skeletal muscle and adipose tissue, are preferentially located on certain chromosomes [27,28]. For the cardiovascular transcriptome in human, a disparate chromosomal distribution has been reported with enrichment on chromosomes 17, 19 and 22 [29,30]. With the present knowledge of genome annotation that reveals the inhomogeneous chromosomal distribution of the human genome, we cannot confirm such observations for the cardiac transcriptome.
Finally, we analyzed correlation of expression levels of genes located in clusters, as it has been suggested for the human transcriptome and other organism. By using the base pair distance as well as the gene distance between gene pairs, we failed to observe any significant correlation between co-localization and expression levels. Recently, it has been proposed that such correlation could be influenced by the probe localization on the array [31], which was corrected for in our primary array analysis. Further, we took into account that our dataset was based on different cardiac phenotypes caused by multiple factors, which could increase background noise and reduce the ability to recognize distinct coexpression of genes in our sample collection.
Conclusion
In summary, we provide evidence that the linear arrangement of genes expressed in concert is due to coordinated regulation by common transcription factors. We suggest that determining the relationship between nuclear organization and gene arrangement will lead to a deeper understanding of how transcriptomes, dedicated to a particular cellular function or fate, are controlled. Here, meta-analysis of the large-scale transcriptome array data beginning to appear in public repositories, could build the basis for the discovery of the nodes between gene regulation and nuclear organization. Those analyses could provide insights into a regulatory constraint such that genes localized in clusters tend to be coregulated throughout several tissues. Furthermore, such a regulatory constraint may be a crucial factor in the development of human diseases caused by partial deletions or insertions of chromosomal units separating genes regulated in concert.
Methods
HXP dataset
The data composition and the classification of expressed genes was done using the Human Unigene II clone set containing 74.695 IMAGE clones [22]. The genomic localizations of the clones were determined via Ensembl and CrossMatch sequence comparison. In summary, 40.416 clones were assigned to 16.260 Ensembl genes, which resulted in 67% coverage of the Ensembl dataset version 11.31.1. In a previous array study, we identified the cardiac transcriptome by hybridisation experiments of 55 cardiac samples using arrays containing the above IMAGE clones [22]. The finally defined heart-expressed subset (HXP) of IMAGE-clones contained 3.172 Ensembl genes represented by 4.167 clones. This gene set refers to the 15% highest expressed genes in at least 4 heart samples whose natural log ratios had a standard deviation across the samples of at least 0.5, and were obtained out of the overall set of human genes (HU2) after normalization of expression levels for array position and averaging over duplicates. The analyzed samples belonged to four categories, (1) the normal right/left atrial and ventricular samples, (2) right atrial and ventricular samples obtained from patients with Tetralogy of Fallot, (3) right atrial samples of patients with ventricular septal defect, and (4) right atrial samples of patients with atrial septal defect. For the further analysis, we excluded the Y chromosome since the dataset was composed of samples from males and females. Functional categories of these genes were assigned using the Gene Ontology classification.
GFN Symatlas
To analyze the distribution of tissue expression of the HXP dataset, we used the microarray gene expression information for 79 human tissues in the GNF Symatlas dataset [15]. This dataset contained almost 34,000 probe sets with 'present', 'marginal', or 'absent' calls for each probe set in each tissue. We considered a probe set expressed if it had a 'present' or 'marginal' call. If probe sets with different expression calls had the same chromosomal location, we considered the 'present' or 'marginal' call in case where one of the probe sets had an 'absent' call. Probe Sets with ambiguous location information were excluded, such that the resulting dataset consisted of 10,715 Ensembl genes with distinct chromosomal locations. Based on the Ensembl gene IDs we could map 1,600 HXP genes including 183 clustered genes to expression information represented by the GNF Symatlas.
Physical annotation of gene distances
To measure the distance of a pair of genes, we used the number of base pairs between them as well as defining the distance of a pair of genes with respect to the amount of Ensembl genes located between them (Fig. 1). A pair of genes was defined as adjacent if there are no other genes in the Ensembl dataset that lie between the two different gene loci. By using these different scales, we were able to distinguish between gene pairs located very close to each other in a chromosomal region with high gene density, as well as describing the relationship of a gene pair independent of its physical localization on the chromosome. Groups or pairs of directly neighboring genes that we analyzed are referred to as gene clusters. The definition of 'neighborhood' refers to the genes included in the Ensembl dataset.
Chromosomal distribution of heart-expressed genes
The number of genes on each chromosome was calculated for the HXP and HU2 dataset. The two distributions were compared using the Chi-square test. The numbers of detected gene clusters were compared to a randomly built dataset based on the genes in the HU2 dataset (100,000 permutations). Furthermore, we used the hypergeometrical distribution to assess whether genes located close to each other share similar expression levels (cor ≥ 0.65) more often than genes located further apart.
Identification of regulatory transcription factors
The identification of putative regulatory transcription factors binding sites was done using the TRANSFAC database. To reduce the rate of false positives among putative binding sites, we first filtered our data by searching for conserved, non-coding upstream regions of orthologous genes in human and mouse as annotated in the Comparative Regulatory Genomics database CORG [23]. CORG is based on the assumption that high levels of sequence conservation in non-coding upstream regions of orthologous genes are likely to reflect common regulatory elements.
Identification of similar gene expression levels
To determine similarity in gene expression for pairs of genes, we calculated the Pearson correlation coefficient as well as the Euclidian distance of their expression values across all 55 previously analyzed tissues [22]. To assess the potential relationship of gene localization and similar expression, we used distance-correlation plots and correlation matrices. Correlation matrices give a rough overview of similar expression levels of genes located on the same chromosome by displaying color-coded correlation coefficients (see for example web supplement Fig. 1S).
Authors' contributions
JHV acquired the data and performed the main analysis of data. AvH has made substantive intellectual contribution to concept and interpretation of data. AP participated in analysis of data and drafting of article. SS conceived the project, managed and participated in analysis and interpretation of data.
Supplementary Material
Additional File 1
Transcription factors shared by gene clusters. Represented are clustered heart-expressed genes with their HUGO gene names, Ensembl gene IDs, strand orientations, transcriptional start site distances and transcription factors for which binding sites could be predicted. The shared transcription factors are indicated in italic.
Click here for file
Additional File 2
Gene Ontology terms shared by gene clusters. Shown are clustered heart-expressed genes annotated with Ensembl gene IDs, HUGO gene names and shared Gene Ontology (GO) terms of layer 4 and 5. The GO term IDs and descriptions are given.
Click here for file
Additional File 3
Correlation matrix of human chromosome 10. The matrix of Pearson correlation coefficients between the expression profiles of heart-expressed genes is shown color-coded, with genes being arranged according to their order on the chromosome. Gene pairs with similar expression levels are depicted in blue, anti-correlated pairs are shown in red.
Click here for file
Acknowledgements
We thank Raffaello Galli and Christoph Dieterich for technical assistance, and Sarah Teichmann for helpful discussions. This work was supported by a grant from The Max-Planck-Society for the Advancement of Science.
==== Refs
van Driel R Fransz PF Verschure PJ The eukaryotic genome: a system regulated at different hierarchical levels J Cell Sci 2003 116 4067 4075 12972500 10.1242/jcs.00779
Hurst LD Pal C Lercher MJ The evolutionary dynamics of eukaryotic gene order Nat Rev Genet 2004 5 299 310 15131653 10.1038/nrg1319
Hershberg R Yeger-Lotem E Margalit H Chromosomal organization is shaped by the transcription regulatory network Trends Genet 2005 21 138 142 15734572 10.1016/j.tig.2005.01.003
Grewal SI Moazed D Heterochromatin and epigenetic control of gene expression Science 2003 301 798 802 12907790 10.1126/science.1086887
Brown KE Chromatin folding and gene expression: new tools to reveal the spatial organization of genes Chromosome Res 2003 11 423 433 12971719 10.1023/A:1024966424909
Stein GS Lian JB Montecino M Stein JL van Wijnen AJ Javed A Pratap J Choi J Zaidi SK Gutierrez S Harrington K Shen J Young D Pockwinse S Nuclear microenvironments support physiological control of gene expression Chromosome Res 2003 11 527 536 12971727 10.1023/A:1024943214431
Lawrence JG Shared strategies in gene organization among prokaryotes and eukaryotes Cell 2002 110 407 413 12202031 10.1016/S0092-8674(02)00900-5
Cho RJ Campbell MJ Winzeler EA Steinmetz L Conway A Wodicka L Wolfsberg TG Gabrielian AE Landsman D Lockhart DJ Davis RW A genome-wide transcriptional analysis of the mitotic cell cycle Mol Cell 1998 2 65 73 9702192 10.1016/S1097-2765(00)80114-8
Cohen BA Mitra RD Hughes JD Church GM A computational analysis of whole-genome expression data reveals chromosomal domains of gene expression Nat Genet 2000 26 183 186 11017073 10.1038/79896
Roy PJ Stuart JM Lund J Kim SK Chromosomal clustering of muscle-expressed genes in Caenorhabditis elegans Nature 2002 418 975 979 12214599
Boutanaev AM Kalmykova AI Shevelyov YY Nurminsky DI Large clusters of co-expressed genes in the Drosophila genome Nature 2002 420 666 669 12478293 10.1038/nature01216
Spellman PT Rubin GM Evidence for large domains of similarly expressed genes in the Drosophila genome J Biol 2002 1 5 12144710 10.1186/1475-4924-1-5
Caron H van Schaik B van der Mee M Baas F Riggins G van Sluis P Hermus MC van Asperen R Boon K Voute PA Heisterkamp S van Kampen A Versteeg R The human transcriptome map: clustering of highly expressed genes in chromosomal domains Science 2001 291 1289 1292 11181992 10.1126/science.1056794
Lercher MJ Urrutia AO Hurst LD Clustering of housekeeping genes provides a unified model of gene order in the human genome Nat Genet 2002 31 180 183 11992122 10.1038/ng887
Su AI Wiltshire T Batalov S Lapp H Ching KA Block D Zhang J Soden R Hayakawa M Kreiman G Cooke MP Walker JR Hogenesch JB A gene atlas of the mouse and human protein-encoding transcriptomes Proc Natl Acad Sci USA 2004 101 6062 6067 15075390 10.1073/pnas.0400782101
Schubeler D Francastel C Cimbora DM Reik A Martin DI Groudine M Nuclear localization and histone acetylation: a pathway for chromatin opening and transcriptional activation of the human beta-globin locus Genes Dev 2000 14 940 950 10783166
Yamashita M Ukai-Tadenuma M Kimura M Omori M Inami M Taniguchi M Nakayama T Identification of a conserved GATA3 response element upstream proximal from the interleukin-13 gene locus J Biol Chem 2002 277 42399 42408 12205084 10.1074/jbc.M205876200
Choudhary SK Wykes SM Kramer JA Mohamed AN Koppitch F Nelson JE Krawetz SA A haploid expressed gene cluster exists as a single chromatin domain in human sperm J Biol Chem 1995 270 8755 8762 7721781 10.1074/jbc.270.15.8755
Kalmykova AI Nurminsky DI Ryzhov DV Shevelyov YY Regulated chromatin domain comprising cluster of co-expressed genes in Drosophila melanogaster Nucleic Acids Res 2005 33 1435 1444 15755746 10.1093/nar/gki281
Lawson GM Knoll BJ March CJ Woo SL Tsai MJ O'Malley BW Definition of 5' and 3' structural boundaries of the chromatin domain containing the ovalbumin multigene family J Biol Chem 1982 257 1501 1507 6276388
Lee HK Hsu AK Sajdak J Qin J Pavlidis P Coexpression analysis of human genes across many microarray data sets Genome Res 2004 14 1085 1094 15173114 10.1101/gr.1910904
Kaynak B von Heydebreck A Mebus S Seelow D Hennig S Vogel J Sperling HP Pregla R Alexi-Meskishvili V Hetzer R Lange PE Vingron M Lehrach H Sperling S Genome-wide array analysis of normal and malformed human hearts Circulation 2003 107 2467 2474 12742993 10.1161/01.CIR.0000066694.21510.E2
Dieterich C Wang H Rateitschak K Luz H Vingron M CORG: a database for Comparative Regulatory Genomics Nucleic Acids Res 2003 31 55 57 12519946 10.1093/nar/gkg007
DeKoter RP Singh H Regulation of B lymphocyte and macrophage development by graded expression of PU.1 Science 2000 288 1439 1441 10827957 10.1126/science.288.5470.1439
Lundgren M Chow CM Sabbattini P Georgiou A Minaee S Dillon N Transcription factor dosage affects changes in higher order chromatin structure associated with activation of a heterochromatic gene Cell 2000 103 733 743 11114330 10.1016/S0092-8674(00)00177-X
Wallin JJ Gackstetter ER Koshland ME Dependence of BSAP represser and activator functions on BSAP concentration Science 1998 279 1961 1964 9506950 10.1126/science.279.5358.1961
Bortoluzzi S Rampoldi L Simionati B Zimbello R Barbon A d'Alessi F Tiso N Pallavicini A Toppo S Cannata N Valle G Lanfranchi G Danieli GA A comprehensive, high-resolution genomic transcript map of human skeletal muscle Genome Res 1998 8 817 825 9724327
Gabrielsson BL Carlsson B Carlsson LM Partial genome scale analysis of gene expression in human adipose tissue using DNA array Obes Res 2000 8 374 384 10968729
Barrans JD Ip J Lam CW Hwang IL Dzau VJ Liew CC Chromosomal distribution of the human cardiovascular transcriptome Genomics 2003 81 519 524 12706110 10.1016/S0888-7543(03)00008-9
Dempsey AA Pabalan N Tang HC Liew CC Organization of human cardiovascular-expressed genes on chromosomes 21 and 22 J Mol Cell Cardiol 2001 33 587 591 11181026 10.1006/jmcc.2000.1335
Kluger Y Yu H Qian J Gerstein M Relationship between gene co-expression and probe localization on microarray slides BMC Genomics 2003 4 49 14667251 10.1186/1471-2164-4-49
|
16171528
|
PMC1261156
|
CC BY
|
2021-01-04 16:27:27
|
no
|
BMC Bioinformatics. 2005 Sep 19; 6:230
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-230
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2321617657610.1186/1471-2105-6-232SoftwareEXPANDER – an integrative program suite for microarray data analysis Shamir Ron [email protected] Adi [email protected] Amos [email protected] Chaim [email protected] Israel [email protected] Roded [email protected] Yosef [email protected] Ran [email protected] School of Computer Science, Sackler Faculty of Exact Sciences. Tel Aviv University, Tel Aviv 69978 Israel2 The David and Inez Myers Laboratory for Genetic Research, Department of Human Genetics, Sackler School of Medicine. Tel Aviv University, Tel Aviv 69978, Israel2005 21 9 2005 6 232 232 5 7 2005 21 9 2005 Copyright © 2005 Shamir et al; licensee BioMed Central Ltd.2005Shamir 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 microarrays are a prominent experimental tool in functional genomics which has opened the opportunity for gaining global, systems-level understanding of transcriptional networks. Experiments that apply this technology typically generate overwhelming volumes of data, unprecedented in biological research. Therefore the task of mining meaningful biological knowledge out of the raw data is a major challenge in bioinformatics. Of special need are integrative packages that provide biologist users with advanced but yet easy to use, set of algorithms, together covering the whole range of steps in microarray data analysis.
Results
Here we present the EXPANDER 2.0 (EXPression ANalyzer and DisplayER) software package. EXPANDER 2.0 is an integrative package for the analysis of gene expression data, designed as a 'one-stop shop' tool that implements various data analysis algorithms ranging from the initial steps of normalization and filtering, through clustering and biclustering, to high-level functional enrichment analysis that points to biological processes that are active in the examined conditions, and to promoter cis-regulatory elements analysis that elucidates transcription factors that control the observed transcriptional response. EXPANDER is available with pre-compiled functional Gene Ontology (GO) and promoter sequence-derived data files for yeast, worm, fly, rat, mouse and human, supporting high-level analysis applied to data obtained from these six organisms.
Conclusion
EXPANDER integrated capabilities and its built-in support of multiple organisms make it a very powerful tool for analysis of microarray data. The package is freely available for academic users at
==== Body
Background
Gene expression microarrays are a prominent experimental tool in functional genomics. They have revolutionized biological research by providing genome-wide snapshots of transcriptional networks that are active in the cell. This opens the opportunity for gaining global, systems-level understanding of cellular processes. Microarray platforms for measuring the expression levels of most or all genes of an organism are available for a variety of organisms ranging from yeast to human. Experiments that use this technology typically generate overwhelming volumes of data, unprecedented in biological research, which makes the task of mining meaningful biological knowledge out of the raw data a major challenge. Hence, exploitation of gene expression data is fully dependent on the availability of advanced data analysis and statistical tools. Many algorithms and software tools for analysis of microarray data were developed in recent years, including sophisticated methods for signal extraction and array normalization [1,2], clustering [3,4], and statistical identification of over-represented functional categories [5] and promoter motifs [6,7]. At present, of special need are integrative software packages that provide users with a set of algorithms collectively covering the whole range of steps in microarray data analysis, thereby significantly boosting the analysis flow and the researcher's ability to deduce meaningful biological conclusions from the overwhelming volume of recorded data. Here we present the EXPANDER program suite for gene expression data analysis.
Implementation
EXPANDER (EXPression ANalyzer and DisplayER), initially developed as a clustering tool [8], has been redesigned as a 'one-stop shop' tool for analysis of the data. EXPANDER 2.0 integrates methods and algorithms that collectively cover different steps of the data analysis, ranging from the initial steps of normalization and filtering, through module detection by clustering and biclustering, to high-level analysis of functional enrichment and of promoter cis-regulatory elements. EXPANDER serves as the major platform in which we integrate various gene expression analysis algorithms that were developed in our lab, including CLICK for clustering [9], SAMBA for biclustering [10], PRIMA for promoter elements analysis [7], and TANGO for GO functional enrichment analysis (manuscript in preparation). In addition, EXPANDER implements various visualization utilities that accompany each of the analysis modules. Four basic design principles instructed us in the implementation of the package: First, the analysis flow should be highly streamlined. Second, although some of the modules are based on highly complicated algorithms, their use should be kept simple and results should be presented in an intuitive manner. Third, data analysis is expected to be done iteratively, allowing users to examine different parameter settings and clustering algorithms – therefore, special effort was put on efficient implementation of the algorithms. Forth, users should be freed from the burden of compiling annotation data required for the analysis. Therefore, EXPANDER not only implements the analysis algorithms, but also supplies users with all necessary annotation and sequence data.
EXPANDER is available with genome-wide pre-processed functional Gene Ontology (GO) and promoter sequences data files for yeast, worm, fly, rat, mouse and human, supporting high-level analysis of data obtained from these organisms. EXPANDER supports analysis of both relative and absolute expression level datasets, the former generated by cDNA microarrays and the latter by, e.g., Affymetrix oligonucleotide arrays. The main utilities provided by EXPANDER and the major algorithms implemented in it are described in the Results section below. Figure 1 gives a high level summary of EXPANDER's analysis flow and of the main algorithms implemented in each analysis step.
Figure 1 A high level summary of EXPANDER's microarray data analysis flow and of the main algorithms implemented in each analysis step.
EXPANDER is implemented in Java. Most of the algorithms it runs were implemented in C. EXPANDER versions for Windows and UNIX are freely available for academic users.
Results
In this section we describe the main analysis modules implemented in EXPANDER, and present a case analysis that demonstrates the strength of this package in deriving biological conclusions out of massive gene expression datasets.
Normalization
The goal of this pre-processing step is the removal of technical biases among the analyzed chips. Currently, the default normalization scheme applied by Affymetrix software is the global scaling, which multiplies all intensities measured in a chip by a constant factor to bring the average/median intensity level in each chip to a predefined fixed level. However, several studies pointed out that global scaling is too naïve in many cases, and that more sophisticated normalization procedures accounting, e.g., for intensity-dependent bias, are required [11,12]. We implemented in EXPANDER two such methods: non-linear regression and quantiles equalization as described in [1]. Normalization of cDNA arrays requires intensity levels measured in both red and green channels. EXPANDER expects log ratios (Red/Green) as input when analyzing dual channels data. Therefore, normalization schemes in EXPANDER are available at this stage to one-channel datasets. Several novel normalization schemes are not yet implemented in EXPANDER (e.g., Variance Stabilizing Normalization (VSN) [13], Li-Wong invariant set normalization [14]). Users can load EXPANDER with data that were normalized using external tools.
Filtering utilities
EXPANDER provides several commonly-used filtering options based on fold-change factors, minimal variation criteria, or choosing the n most variant genes, allowing the user to focus downstream analysis on the set of genes that show sufficient variation across the measured conditions.
Cluster analysis
Clustering algorithms applied to gene expression data partition the genes into distinct groups according to their expression patterns over the probed biological conditions. Such partition should assign genes with similar expression patterns to the same cluster (keeping the homogeneity merit of the clustering solution) while retaining the distinct expression pattern of each cluster (ensuring the separation merit of the solution). Cluster analysis eases the interpretation of the data by reducing its complexity and revealing the major patterns that underlie it. EXPANDER implements a few of the most widely used clustering algorithms – SOM [4], K-means [15], and hierarchical clustering [3], as well as CLICK, a graph theoretic based algorithm developed in our lab. CLICK is described in detail in [16] and it was demonstrated to outperform other algorithms according to several figures of merit [9]. When computing a clustering solution, EXPANDER also specifies its homogeneity and separation measures, enabling the user to compare the merits of different solutions. Several displays for patterns (Fig. 2) and matrices (Fig. 3) are provided for the visualization of clustering solutions.
Figure 2 All-patterns display of a clustering solution. Each graph represents a specific cluster. The X-axis represents the measured conditions. The Y-axis represents (standardized) expression levels. Each cluster is represented by the mean expression pattern over all the genes assigned to it. Error bars denote ± 1 standard deviation. Clicking within a cell opens a window that lists the genes that are assigned to the cluster.
Figure 3 Matrix displays. (A) Unclustered expression matrix display. Each row corresponds to a gene, and each column to a biological sample. The color of the (i, j) cell in the matrix indicates the expression level of the ith gene in the jth sample. Green represents below-average expression level; Red represents above-average expression level (color scheme can be adjusted by the user). (B) The same dataset as in A, with genes ordered according to a clustering solution. Horizontal white lines separate the different clusters. (C) Unclustered similarity matrix display. The color of the (i, j) cell in the matrix represents the similarity between the expression patterns of the ith and the jth genes over all the samples (hence the matrix is symmetric). Red represents high similarity, and green represents low similarity. (D) Same as in C, with genes ordered on both axes according to a clustering solution. Clusters appear as distinct red blocks along the matrix diagonal, and similar clusters are manifested by off-diagonal reddish blocks.
Bicluster analysis
As expression data accumulate, and profiles over hundreds of different biological conditions are readily available, clustering becomes too restrictive. Clustering algorithms globally partition genes into disjoint sets according to the overall similarity in their expression patterns, i.e., they search for genes that exhibit similar expression levels over all the measured conditions. Such requirement is appropriate when analyzing small to medium size datasets from one or a few related experiments or when analyzing time-series data, as it provides statistical robustness and produces results that are easily visualized and comprehended. Yet, when larger datasets are analyzed, a more flexible approach is frequently advantageous. A bicluster (or a module) is defined as a set of genes that exhibit significant similarity over a subset of the conditions (Fig 4a). A biclustering algorithm can dissect a large gene expression dataset into a collection of biclusters, where genes or conditions can take part in more than one bicluster. A set of biclusters can thus characterize a combined, multifaceted gene expression dataset [10]. An enhanced version of our biclustering algorithm, called SAMBA (Statistical-Algorithmic Method for Bicluster Analysis) is integrated in EXPANDER and is the preferable partition-analysis approach for large heterogeneous datasets that encompass dozens of conditions (Fig 4b). SAMBA 2.0 can handle datasets with thousands of conditions profiled over entire genomes. For technical description of the SAMBA algorithm see [10,17]. Briefly, the algorithm first transforms gene expression data into a weighted bipartite graph (with genes and conditions as its two parts) and then applies a statistical scoring scheme and a combinatorial algorithm to identify heavy subgraphs in the bipartite graph. Each such heavy subgraph represents a bicluster. SAMBA operates in three phases: in the first step bicluster seeds are detected, then each seed is optimized to a locally optimal bicluster, and finally a non redundant subset of the locally optimized biclusters is selected. SAMBA 2.0 contains a new implementation of the first step in which efficient hashing techniques are now utilized, thereby significantly improving running time. It also features a new redundancy filtering algorithm (step 3) that optimizes the total likelihood of a set of biclusters using a probabilistic model that generalizes the single bicluster model.EXPANDER allows the user to tune SAMBA's performance by selecting among several multi-level discretization schemes based on the numerical characteristics of the analyzed dataset. Another important tunable parameter controls the stringency of the redundancy-filtering algorithm.
Figure 4 Bicluster analysis. (A) A bicluster corresponds to a submatrix defined by row and column subsets. Both subsets are not known in advance. After reordering the original data matrix, it can be seen as the rectangle with the yellow border. (B) EXPANDER summarizes bicluster analysis results in a table that lists the dimensions (numbers of genes and conditions) of the biclusters identified and their scores. Clicking on a row in this table pops-up a window with the submatrix view of the selected bicluster. Below the table there are two examples of biclusters identified in a dataset comprising some 1,000 genes measured across over 70 conditions in human cells. Row and column labels are gene and condition names for the bicluster, respectively.
Functional enrichment analysis
After identifying the main co-expressed gene groups in the data (either by clustering or biclustering), one of the major challenges is to ascribe them to some biological meaning. To assist the researcher in this task, EXPANDER contains a statistical analysis module that seeks specific functional categories that are significantly over-represented in the analyzed gene groups, with respect to a given background set of genes. In addition to pointing to possible biological roles for distinct gene sets, such analysis was demonstrated to be very helpful in assigning putative functional roles to uncharacterized genes [10,18]. EXPANDER is provided with pre-compiled functional annotation files for six organisms: yeast (S. cerevisiae), worm (C. Elegans), fly (D. melanogaster), rat (R. norvegicus), mouse (M. musculus) and human, releasing the user from the burden of compiling such annotation information. These annotation files, compiled based on data provided by the Gene Ontology (GO) consortium [19] and the central databases for these organisms, associate genes with GO functional categories.
A major challenge in identifying cases of over-represented GO categories is obtaining a good estimation of statistical significance for each case, that takes multiple testing into account (hundreds of categories are typically tested for enrichment). What complicates this task is the hierarchical tree-like structure of the ontology, which induces strong dependencies among GO categories. Thus, standard methods for accounting for multiple testing, which assume independent tests (e.g., Bonferroni, False Discovery Rate) are far too stringent. EXPANDER uses the TANGO (Tool for ANalysis of GO enrichments) algorithm for coping with this problem (Tanay et al., in preparation). Briefly, TANGO repeatedly shuffles genes to compute an empirical distribution of maximum p-values for functional enrichment obtained across a random sample of clusters that maintain the same size characteristics of the analyzed clusters. TANGO uses this empirical distribution to determine thresholds for significant enrichment on the true clusters. Another problem that stems from the strong dependencies among GO categories is the high level of redundancy in the reported enriched categories, which often include both parent and child nodes associated with highly overlapping set of genes. TANGO filters out such redundant categories by performing conditional enrichment tests that ensure that all the reported enriched categories are statistically significant even after taking into account the enrichment of their related nodes in the tree. An example for the visualization of TANGO results is shown in Figure 5.
Figure 5 GO functional enrichment analysis. (A) Enriched GO categories identified by TANGO in the analyzed gene groups (clusters or biclusters) are displayed as bar diagrams; each corresponding to a specific gene group (i. e., cluster or bicluster). In these diagrams, GO categories are color-coded, and the height of a bar represents the statistical significance (-log10(p-value)) of the observed enrichment for its corresponding category. The percentage of genes in the group assigned to the enriched category is indicated above the bar. (B) Clicking on a bar pops-up a window that lists the group's genes that are associated with the corresponding GO category. In this window, genes are linked to central annotation DBs (SGD [25] for yeast, WormBase [26] for worm, FlyBase [27] for fly, and Entrez Gene [28] for human, mouse and rat) where detailed gene descriptions can be found for in-depth analysis.
Cis-regulatory element analysis
Microarray measurements provide snapshots of cellular transcriptional programs that take place in the examined biological conditions. These measurements do not, however, directly reveal the regulatory networks that underlie the observed transcriptional activity, i.e. the transcription factors (TFs) that control the transcription of the responding genes. Computational promoter analysis can shed light on the regulators layer of the network. Based on the assumption that genes that exhibit similar expression pattern over multiple conditions are likely to be controlled by common regulators and, therefore, share common cis-regulatory elements in their promoter regions, several algorithms have been developed to identify over-represented cis-elements in promoters of co-expressed genes. Such computational approaches successfully delineated transcriptional networks in organisms ranging from yeast to human [7,15]. EXPANDER provides such promoter analysis utility by integrating our PRIMA (PRomoter Integration in Microarray Analysis) tool which is described in detail in [7]. In short, given target sets and a background set of promoters, PRIMA performs statistical tests aimed at identifying transcription factors whose binding site signatures are significantly more prevalent in any of the target sets than in the background set. Typically, sets of co-expressed genes identified using either cluster or bicluster analysis serve as target sets, and the entire collection of promoters of genes present on the microarray serves as the background set. In its stand-alone version, an execution of PRIMA typically takes several hours to complete. To facilitate the computations of PRIMA from within EXPANDER, we added a preprocessing phase, which decreased the running time to just a few minutes on a standard PC. The preprocessing phase is run by us on occasions of major updates to genome sequence assemblies of the supported organisms (typically, once every few months). It generates promoter-fingerprints file per organism. These fingerprints files map computationally-identified high scoring putative binding sites ('hits') of all TFs to the entire set of promoters in the organisms. In the version integrated in EXPANDER, PRIMA loads the hits data from the fingerprints files rather than scanning promoter sequences de-novo on each run, thereby drastically reducing the running time. This improvement greatly enhanced the flexibility of PRIMA, enabling its execution in an iterative way, in which results obtained by different clustering solutions can be routinely compared. EXPANDER provides genome-wide pre-processed promoter fingerprints files for the six organisms that are we currently support (yeast, worm, fly, mouse, rat and human). The integration of PRIMA into EXPANDER allows the user to both identify the major expression patterns in his/her data (by applying EXPANDER's cluster analysis module), and points to transcription factors that underlie the transcriptional alterations observed in the clusters (Fig 6).
Figure 6 Promoter cis-regulatory elements enrichment. (A) Transcription factors (TFs) whose DNA binding site signatures are over-represented in promoters of the genes assigned to the clusters are displayed in bar diagrams. Like the display for the GO analysis (Fig. 5), each diagram corresponds to a specific gene group (cluster or bicluster), TFs are color-coded and identified by the accession number of their binding site model in TRANSFAC DB. The statistical significance of the observed enrichment for a TF is represented by the height of its bar (-log10(p-value)). The TF enrichment factor, which is the ratio between the prevalence of the TF hits in the gene group and in the background set of promoters, is indicated above the bar. (B) Clicking on a specific bar pops-up a window that lists the genes in the group whose promoters were found to contain a hit for that TF. In this window, genes are linked to central annotation DB of the analyzed organism as specified in the legend of Figure 5.
Demonstration of EXPANDER's capabilities
To demonstrate the utility of the EXPANDER package, we applied it to a very large dataset published recently by Murray et al [20]. This study recorded expression profiles in several human cell lines exposed to various stressful conditions. The authors integrated these data with a dataset in which expression profiles were measured throughout the progression of the cell cycle [21]. The combined dataset contains expression data for 36,825 probes measured over 174 conditions. The analysis of such complex dataset poses a daunting bioinformatics challenge. Murray et al. used the Cluster/TreeView tool [3] to hierarchically cluster this dataset, and by visual inspection of the resulting tree defined the main clusters in the data. A second "adoption step" was then applied, in which each main cluster adopted genes whose expression pattern resembled the cluster's mean pattern. Overall, 23 clusters containing 1245 distinct genes were reported. Biological meaning was assigned to the clusters by inspection of their expression profiles and of the genes they contain. No promoter analysis was reported.
As we noted above, when analyzing large datasets, biclustering becomes more appropriate than clustering. Therefore, we subjected this dataset to bicluster analysis using SAMBA. We first replaced missing entries with 0 (which corresponds to 'no change' in log-transformed data) and then scanned the dataset for probes whose expression was changed by at least 2-fold in at least 7 conditions. Some 10% of the clones (3,392) passed this filtering. We applied SAMBA to the union of these genes and the 1245 genes analyzed by Murray et al. The union contains 3892 genes. SAMBA identified 155 biclusters on this filtered dataset. (These biclusters can overlap – genes can be assigned to several biclusters – but are not redundant: a pruning step removes highly overlapping biclusters.) The identified biclusters reveal the major expression patterns that underlie this intricate dataset. Next, we aimed to assign biclusters with putative functional meaning, and to identify major TFs that regulate the transcriptional responses captured by them. To this goal, we applied the TANGO and PRIMA modules (both were run with default parameters).
The purpose of this exercise is not to apply in-depth biological analysis of stress responses in human cells, but to demonstrate the strength and agility of EXPANDER in analysis of complex microarray datasets. Therefore we only briefly summarize some of the major biclusters identified in the dataset, along with their putative biological roles and transcriptional regulators that were computationally discovered by EXPANDER. The major biclusters identified are listed in Table 1 and some of them are presented in Additional file 1. In agreement with Murray et al., we found that most of the transcriptional responses to stressful conditions were agent- and cell-type- specific (for example, bicluster #1 represents 145 genes that were activated only in Hela cells exposed to heat shock; bicluster #24 represents over 100 genes that were activated only in fibroblasts exposed to DDT). In addition, some biclusters correspond to more general stress responses that were induced by multiple agents and in different cell lines (for example, bicluster #53 contains 51 genes that were down-regulated in response to both oxidative stress and heat shock in Hela cells and in response to heat shock in fibroblasts). As pointed by Murray et al, analyzing together the stress data and the cell-cycle data allows the distinction between genes that respond directly to the stress agents and those whose change can be explained by differences in the fraction of cells in the different phases of the cell cycle due to activation of cell cycle checkpoints after exposure to damaging agents. Indeed, bicluster #106 is enriched for DNA replication genes, up-regulated in S-phase time points in the cell-cycle dataset, and down-regulated in fibroblasts exposed to either DDT or Menadione, probably reflecting an arrest of these cells in early G1 or G2 phase. Similarly, biclusters #40 and #104 show the down-regulation of mitotic genes in several cell lines and in response to various stress agents, probably reflecting the reduction in the fraction of cells undergoing mitosis in these stressed cell populations.
Table 1 Major biclusters identified in the test case analysis of the human stress data set.
Bicluster number Num of Conditions Num of Genes Enriched GO (GOid, p-val) Enriched TF binding site signatures (TRANSFAC id, p-val) Comments
106 9 79 DNA Replication (GO:0006260, 5.3 × 10-9) E2F (M00918, 1.3 × 10-7) Down-regulation of DNA replication genes in fibroblasts exposed to DDT or Menadione.
40 33 74 Mitosis (GO:0007067, 9.3 × 10-19) NF-Y (M00287, 6.7 × 10-22) IRF-7 (M00453, 9.5 × 10-5) Down-regulation of mitotic genes in response to various stresses.
104 41 13 Mitosis (GO:0007067, 3.3 × 10-10) NF-Y (M00287, 3.4 × 10-9) Down-regulation of mitotic genes in response to various stresses.
16 5 89 Carboxylic acid metabolism (GO:0019752, 3.4 × 10-8) --- Genes activated in Hela cells in response to Tunicamycin and Menadione
1 6 145 Response to unfolded protein (GO:0006986, 1.2 × 10-7) --- Genes activated in Hela cells in response to heat shock
9 7 142 Response to unfolded protein (GO:0006986, 7.3 × 10-9) AP-2alpha (M00469, 5.6 × 10-4) Genes activated in K562 cells in response to heat shock
111 10 24 Response to unfolded protein (GO:0006986, 1.5 × 10-7) --- Genes that are activated by heat shock but repressed by crowding in Hela cells
24 6 105 Transcription corepressor (GO:0003714, 1.5 × 10-6) HIF-1 (M00797, 6.9 × 10-4) Genes activated in fibroblasts in response to DDT
27 8 200 --- ---- Genes activated in fibroblasts in response to oxidative stress (H2O2)
61 9 134 --- ---- Genes that are repressed by crowding in fibroblasts.
53 22 51 ---- N-Myc(M00055, 2.7 × 10-6) Genes that are repressed in both Hela cells and fibroblasts.
89 9 115 --- AP-4 (M00005, 2.1 × 10-4) Genes repressed in Hela cells in response to various stresses.
123 10 31 --- NFkB (M00051, 7.1 × 10-4) Genes activated in Hela cells in response to DDT.
In several biclusters PRIMA identified significant enrichment for binding site signatures of TFs that are known to control the respective biological processes (e.g., over-representation of E2F binding site in bicluster #106, which is enriched for DNA replication genes; enrichment of NF-Y binding site in bicluster #40, which is enriched for mitotic genes). In other biclusters PRIMA suggests novel links between TFs and stress responses (e.g., over-representation of N-Myc binding site in bicluster #53, which contains genes that are repressed by different stresses).
Some of EXPANDER's salient advantages are evident from the above analysis: The biclustering module, which is unique to EXPANDER among packages for microarray data analysis, allows systematic detection of the major expression patterns in highly complex datasets. Biclusters provide higher resolution gene groups, some encompassing many conditions but most covering relatively small subsets and thus focusing on specific phenomena. Functional enrichment and promoter analyses are done in a streamlined and integrated fashion, and so most of the expert's effort can be devoted to biological interpretation. Last, analysis of microarray data requires experimentation with various filtering thresholds and algorithmic parameters settings; therefore it is of high importance that the analysis modules will require relatively short running time. EXPANDER was designed to meet this requirement. A full analysis iteration, which includes biclustering, functional enrichment and promoter analyses applied to the above massive dataset that we used as an example, takes some 15 mins on a standard PC.
Comparison with other tools
Several integrative packages for the analysis of gene expression data were are available, among them are INCLUSive [22], Expression-Profiler, GEPAS [23], TIGR's Multiple Experiment Viewer, and ArrayPipe [24]. EXPANDER has several advantages over extant packages. While some of the integrative packages are designed as web portals that provide links to independent programs, where, in some cases, the outputs are sent to the user by e-mail and not always in a format directly compatible with subsequent analysis steps, in EXPANDER the analysis flow is inherently streamlined and straightforward. In addition, EXPANDERs' strength lies in the advanced algorithms it uniquely provides: CLICK for clustering, SAMBA for biclustering, TANGO for identification of GO enrichment, and PRIMA for the identification of enriched TF binding site signatures. The synergism that stems from the integration of these algorithms into one package grants EXPANDER with very powerful analytical capabilities. Another feature that distinguishes EXPANDER is its built-in support for genome-wide analysis of data obtained from six major research organisms.
Conclusion
Designed as a 'one-stop shop' for gene expression data analysis, EXPANDER provides algorithms covering main analysis steps including (1) the initial process of normalization and filtering for removing biases and focusing downstream analysis on responding genes in the dataset; (2) clustering and biclustering to discover the main expression patterns in the data; (3) high-level functional enrichment analysis; and (4) promoter cis-element analysis to gain insights on the biological meaning of the identified expression patterns and to point to transcriptional regulators that underlie them. These integrated capabilities provided by EXPANDER and its built-in support of multiple organisms make it a very powerful tool for analysis of microarray data. Although some of the analysis modules implemented in EXPANDER are based on sophisticated algorithms, their execution remains simple and intuitive.
We will routinely post on EXPANDER's website updated GO annotation and promoter fingerprint files for all the supported organisms. EXPANDER's users will be notified of such updates. We will continue to maintain and expand EXPANDER to keep it as an integrative suite that provides state-of-the-art algorithms and visualization utilities for analysis of microarray data. We will also expand the group of organisms supported by the package according to the availability of appropriate information and data.
Availability and requirements
• Project name: EXPANDER
• Project home page:
• Operating system(s): Windows, UNIX
• Programming language: Java for the envelope and C for most of the algorithms.
• Other requirements: Java 1.4 or higher
• License: free for non-commercial users.
• Any restrictions to use by non-academics: License needed.
Abbreviations
EXPANDER – EXPression ANalyzer and DisplayER
SOM – Self Organizing Maps
CLICK – CLuster Identification via Connectivity Kernels
SAMBA – Statistical-Algorithmic Method for Bicluster Analysis
TANGO – Tool for ANalysis of GO enrichments
PRIMA – PRomoter Integration in Microarray Analysis
GO – Gene Ontology
PWM – Position weight matrix
TF – Transcription factor
Authors' contributions
R. Sharan developed CLICK. AT and R. Sharan developed SAMBA, and AT with IS improved and enhanced it. CL and RE developed PRIMA. AT developed TANGO. AM coded and integrated the complete package and the visualization methods. RE and CL created the annotation and promoter files for the six species. RE performed the use-case data analysis and wrote the manuscript with R. Shamir. R. Shamir, with help from YS and RE, conceived, designed and led the project.
Supplementary Material
Additional File 1
Examples of 13 major biclusters identified on the human cells stress-response dataset. Enriched GO categories and TFBS signature found in these biclusters are summarized in Table 1. In each matrix, rows and columns correspond, respectively, to genes and conditions that participate in the bicluster. Labels of the conditions follow this convention: the cell line is indicated first (Hela for Hela cells, WI for WI38 fibroblasts, or K for K-562 cells), followed by an indication for the stress agent (HS – heat shock, DDT, Ox – H2O2 oxidative stress, Crd – crowding, Men – Menadione, Tun – Tunicamycin, CC – cell cycle data measured in Hela cells synchronized using double thymidine, and CCb – cell cycle data in Hela cells synchronized using thymidine-nocodazole). The last number in the label indicates the time point.
Click here for file
Acknowledgements
This study was supported in parts by research a grant from the Ministry of Science and Technology, Israel. R. Elkon is a Joseph Sassoon Fellow. A. Tanay is supported in part by a scholarship in Complexity Science from the Yeshaia Horvitz Association. R. Sharan is supported by an Alon fellowship.
==== Refs
Bolstad BM Irizarry RA Astrand M Speed TP A comparison of normalization methods for high density oligonucleotide array data based on variance and bias Bioinformatics 2003 19 185 193 12538238 10.1093/bioinformatics/19.2.185
Irizarry RA Hobbs B Collin F Beazer-Barclay YD Antonellis KJ Scherf U Speed TP Exploration, normalization, and summaries of high density oligonucleotide array probe level data Biostatistics 2003 4 249 264 12925520 10.1093/biostatistics/4.2.249
Eisen MB Spellman PT Brown PO Botstein D Cluster analysis and display of genome-wide expression patterns Proc Natl Acad Sci U S A 1998 95 14863 14868 9843981 10.1073/pnas.95.25.14863
Tamayo P Slonim D Mesirov J Zhu Q Kitareewan S Dmitrovsky E Lander ES Golub TR Interpreting patterns of gene expression with self-organizing maps: methods and application to hematopoietic differentiation Proc Natl Acad Sci U S A 1999 96 2907 2912 10077610 10.1073/pnas.96.6.2907
Al-Shahrour F Diaz-Uriarte R Dopazo J FatiGO: a web tool for finding significant associations of Gene Ontology terms with groups of genes Bioinformatics 2004 20 578 580 14990455 10.1093/bioinformatics/btg455
Aerts S Thijs G Coessens B Staes M Moreau Y De Moor B Toucan: deciphering the cis-regulatory logic of coregulated genes Nucleic Acids Res 2003 31 1753 1764 12626717 10.1093/nar/gkg268
Elkon R Linhart C Sharan R Shamir R Shiloh Y Genome-wide in silico identification of transcriptional regulators controlling the cell cycle in human cells Genome Res 2003 13 773 780 12727897 10.1101/gr.947203
Sharan R Maron-Katz A Shamir R CLICK and EXPANDER: a system for clustering and visualizing gene expression data Bioinformatics 2003 19 1787 1799 14512350 10.1093/bioinformatics/btg232
Sharan R Elkon R Shamir R Cluster analysis and its applications to gene expression data Ernst Schering Res Found Workshop 2002 83 108 12061008
Tanay A Sharan R Kupiec M Shamir R Revealing modularity and organization in the yeast molecular network by integrated analysis of highly heterogeneous genomewide data Proc Natl Acad Sci U S A 2004 101 2981 2986 14973197 10.1073/pnas.0308661100
Schuchhardt J Beule D Malik A Wolski E Eickhoff H Lehrach H Herzel H Normalization strategies for cDNA microarrays Nucleic Acids Res 2000 28 E47 10773095 10.1093/nar/28.10.e47
Yang YH Dudoit S Luu P TP. S Normalization for cDNA Microarray Data Technical Report Department of Statistics, University of California at Berkeley 2000
Huber W von Heydebreck A Sultmann H Poustka A Vingron M Variance stabilization applied to microarray data calibration and to the quantification of differential expression Bioinformatics 2002 18 Suppl 1 S96 104 12169536
Kel AE Kel-Margoulis OV Farnham PJ Bartley SM Wingender E Zhang MQ Computer-assisted identification of cell cycle-related genes: new targets for E2F transcription factors J Mol Biol 2001 309 99 120 11491305 10.1006/jmbi.2001.4650
Tavazoie S Hughes JD Campbell MJ Cho RJ Church GM Systematic determination of genetic network architecture Nat Genet 1999 22 281 285 10391217 10.1038/10343
Sharan R Shamir R CLICK: a clustering algorithm with applications to gene expression analysis Proc Int Conf Intell Syst Mol Biol 2000 8 307 316 10977092
Tanay A Sharan R Shamir R Discovering statistically significant biclusters in gene expression data Bioinformatics 2002 18 Suppl 1 S136 44 12169541
Wu LF Hughes TR Davierwala AP Robinson MD Stoughton R Altschuler SJ Large-scale prediction of Saccharomyces cerevisiae gene function using overlapping transcriptional clusters Nat Genet 2002 31 255 265 12089522 10.1038/ng906
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Harris MA Hill DP Issel-Tarver L Kasarskis A Lewis S Matese JC Richardson JE Ringwald M Rubin GM Sherlock G Gene ontology: tool for the unification of biology. The Gene Ontology Consortium Nat Genet 2000 25 25 29 10802651 10.1038/75556
Murray JI Whitfield ML Trinklein ND Myers RM Brown PO Botstein D Diverse and specific gene expression responses to stresses in cultured human cells Mol Biol Cell 2004 15 2361 2374 15004229 10.1091/mbc.E03-11-0799
Whitfield ML Sherlock G Saldanha AJ Murray JI Ball CA Alexander KE Matese JC Perou CM Hurt MM Brown PO Botstein D Identification of genes periodically expressed in the human cell cycle and their expression in tumors Mol Biol Cell 2002 13 1977 2000 12058064 10.1091/mbc.02-02-0030.
Coessens B Thijs G Aerts S Marchal K De Smet F Engelen K Glenisson P Moreau Y Mathys J De Moor B INCLUSive: A web portal and service registry for microarray and regulatory sequence analysis Nucleic Acids Res 2003 31 3468 3470 12824346 10.1093/nar/gkg615
Herrero J Vaquerizas JM Al-Shahrour F Conde L Mateos A Diaz-Uriarte JS Dopazo J New challenges in gene expression data analysis and the extended GEPAS Nucleic Acids Res 2004 32 W485 91 15215434
Hokamp K Roche FM Acab M Rousseau ME Kuo B Goode D Aeschliman D Bryan J Babiuk LA Hancock RE Brinkman FS ArrayPipe: a flexible processing pipeline for microarray data Nucleic Acids Res 2004 32 W457 9 15215429
Christie KR Weng S Balakrishnan R Costanzo MC Dolinski K Dwight SS Engel SR Feierbach B Fisk DG Hirschman JE Hong EL Issel-Tarver L Nash R Sethuraman A Starr B Theesfeld CL Andrada R Binkley G Dong Q Lane C Schroeder M Botstein D Cherry JM Saccharomyces Genome Database (SGD) provides tools to identify and analyze sequences from Saccharomyces cerevisiae and related sequences from other organisms Nucleic Acids Res 2004 32 D311 4 14681421 10.1093/nar/gkh033
Chen N Harris TW Antoshechkin I Bastiani C Bieri T Blasiar D Bradnam K Canaran P Chan J Chen CK Chen WJ Cunningham F Davis P Kenny E Kishore R Lawson D Lee R Muller HM Nakamura C Pai S Ozersky P Petcherski A Rogers A Sabo A Schwarz EM Van Auken K Wang Q Durbin R Spieth J Sternberg PW Stein LD WormBase: a comprehensive data resource for Caenorhabditis biology and genomics Nucleic Acids Res 2005 33 (Database Issue) D383 9 15608221
Drysdale RA Crosby MA Gelbart W Campbell K Emmert D Matthews B Russo S Schroeder A Smutniak F Zhang P Zhou P Zytkovicz M Ashburner M de Grey A Foulger R Millburn G Sutherland D Yamada C Kaufman T Matthews K DeAngelo A Cook RK Gilbert D Goodman J Grumbling G Sheth H Strelets V Rubin G Gibson M Harris N Lewis S Misra S Shu SQ FlyBase: genes and gene models Nucleic Acids Res 2005 33 (Database Issue) D390 5 15608223
Maglott D Ostell J Pruitt KD Tatusova T Entrez Gene: gene-centered information at NCBI Nucleic Acids Res 2005 33 (Database Issue) D54 8 15608257
|
16176576
|
PMC1261157
|
CC BY
|
2021-01-04 16:27:28
|
no
|
BMC Bioinformatics. 2005 Sep 21; 6:232
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-232
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2331617658510.1186/1471-2105-6-233Research ArticleGeneRank: Using search engine technology for the analysis of microarray experiments Morrison Julie L [email protected] Rainer [email protected] Desmond J [email protected] David R [email protected] Bioinformatics Research Centre, University of Glasgow, Glasgow, UK2 Department of Mathematics, University of Strathclyde, Glasgow, UK3 Institute of Biomedical and Life Sciences, Glasgow, UK2005 21 9 2005 6 233 233 19 4 2005 21 9 2005 Copyright © 2005 Morrison et al; licensee BioMed Central Ltd.2005Morrison 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
Interpretation of simple microarray experiments is usually based on the fold-change of gene expression between a reference and a "treated" sample where the treatment can be of many types from drug exposure to genetic variation. Interpretation of the results usually combines lists of differentially expressed genes with previous knowledge about their biological function. Here we evaluate a method – based on the PageRank algorithm employed by the popular search engine Google – that tries to automate some of this procedure to generate prioritized gene lists by exploiting biological background information.
Results
GeneRank is an intuitive modification of PageRank that maintains many of its mathematical properties. It combines gene expression information with a network structure derived from gene annotations (gene ontologies) or expression profile correlations. Using both simulated and real data we find that the algorithm offers an improved ranking of genes compared to pure expression change rankings.
Conclusion
Our modification of the PageRank algorithm provides an alternative method of evaluating microarray experimental results which combines prior knowledge about the underlying network. GeneRank offers an improvement compared to assessing the importance of a gene based on its experimentally observed fold-change alone and may be used as a basis for further analytical developments.
==== Body
Background
Since its launch in 1998, the Google search engine has all but monopolized page searches on the world-wide web [1]. The basis of this astonishing success is the PageRank algorithm developed by Google founders Larry Page and Sergey Brin [2], which allows efficient and stable prioritization of search results. Here we show how the basic idea of PageRank can be transferred quite intuitively to the analysis of gene expression datasets in molecular biology. We modify PageRank appropriately to produce a new algorithm, GeneRank, and explore its limits and potential for the analysis of both synthetic and real-world data.
Just as the original PageRank is stable against the artificial inflation of a web page's rank by web designers, we hope that GeneRank may obtain a more robust ranking of genes in (typically very noisy [3]) microarray experiments. While PageRank uses hyperlinks between web pages to achieve this end, we combine the expression measurements with external information, such as functional annotations, protein interaction data or previous experimental results.
Data sharing techniques have been successfully implemented previously using, for example, GO annotations [4,5] or protein-protein interactions [6-8] to define the network connectivity. Justification for this is given by the observation that connected genes are more likely to be co-expressed [6]. Additional advantages include the possibility for GO terms to provide insight into the process of co-expression [4] and for functional classification to be made for previously unannotated genes [7,9]. These examples serve to demonstrate the value and feasibility of combining data from different sources. Not only is the analysis of a microarray experiment likely to be more robust when prior information is included, but networks rather than single genes can be identified as important, and further biological inferences can be drawn about the data with the aid of this added information.
Our aim in this work is to use connectivity data to produce a prioritization of the genes in a microarray experiment that is less susceptible to variation caused by experimental noise than one based on expression levels alone. This is achieved using GeneRank, a customised version of the PageRank algorithm.
Results and discussion
The algorithm
The algorithm on which we base our method of microarray experiment analysis was originally devised for assessing the importance of web pages in search engine results. We show here that its formulation allows for a simple and intuitive extension for our application. The PageRank algorithm, used by the successful search engine Google [10], is based on the premise that a web page should be highly ranked if other highly ranked pages contain hyperlinks to it. This idea naturally extends to analysing the results of a microarray experiment, where we would like a gene to be highly ranked if it is linked to other highly ranked genes, even if its own position is lower, e.g., due to measurement variability. We can think of this as the "vote of confidence" principle (See Figure 1). Here we would hope that the relative ranking of the gene with little or no differential expression will be boosted by the PageRanking process. In some cases there may be an even stronger biological interpretation. For example, suppose the gene with low differential expression is a transcription factor that controls the expression of all genes connected to it. The transcription factor itself may be "activated" by the experimental treatment but not change its expression – but its target genes will. Hence, GeneRank should be able to highlight the transcription factor among the results.
Figure 1 The Vote of Confidence Principle. Just as a the PageRank of a web page will be high if it is linked to other highly ranked pages, we hope that the relative ranking of a gene will be increased if it is linked to other highly differentially expressed genes.
The original PageRank algorithm also has a random walk interpretation where the ranks correspond to the invariant measure of a teleporting random walk on the web. This is equivalent to saying that the rank of a web page is proportional to the time spent at the web page whilst surfing the web. This idea can also be intuitively extended to ranking genes, where the rank of a gene is proportional to the amount of time a biologist should spend looking at a gene whilst analysing the experimental results.
As with the original algorithm, we require a network or graph to allow us to calculate a rank for each entity in the network. With the original algorithm, nodes represent web pages and a link exists between two nodes if one page contains a hyperlink to the other. This results in a directed graph. In our case, we define an undirected graph where a node represents a gene and the edges can be defined by some other "previous knowledge". For our purpose, we use either Gene Ontology annotations [11] or expression profile correlation coefficients. In addition to the network structure, we require a vector of expression changes, experimentally observed in a microarray experiment, as input for the algorithm. Other measures of differential expression could also be used, e.g. p-values, which are suitably transformed to ensure genes which are highly changed have a large input value. We shall only consider the use of expression changes.
For each gene, the expression level vector contains the value for its expression change in the experiment under consideration. The algorithm, GeneRank, also uses a free parameter, d, in the range [0..1] that controls the weighting of expression change to connectivity used in the calculations. If d = 0, the ranking returned is based solely on the absolute value of the expression fold-change for that gene. For d = 1 we return the ranking based on connectivity. By setting d in the range [0..1] we interpolate between these two extremes. One advantage of our approach is that we apply the algorithm to the entire network, i.e. we do not require a pre-defined threshold of important genes. We simply require an experimentally determined expression change and some connectivity information for each gene, and based on this GeneRank will provide a re-ordering of genes in terms of their apparent importance. While "importance" may seem a rather vague concept in a biological context, we will show how it can be assessed objectively in the Testing the Algorithm section below. For full details of the algorithm and the random walk interpretation, the reader is referred to the Methods section.
Data
In addition to the gene expression data, GeneRank uses a network or graph as input. We use the absolute value (the algorithm requires positive expression change values) of the gene expression data as a weight for each node in the graph and define the network connectivity by some other criterion. We use either Gene Ontology annotations or correlation coefficients, but there are many other possibilities, e.g., metabolic networks, transcription factor networks, or protein-protein interactions. We also used synthetic networks with controlled topological features for evaluation purposes. The three types of network were constructed as follows.
GO networks
Genes are connected if they share an annotation defined by the Gene Ontology. This defines three networks, one for each of the GO sections; Biological Process, Cellular Component and Molecular Function. We do not use the acyclic directed graph associated with the Gene Ontology, but assign leaf nodes as the annotations for each gene. A yeast diauxic shift experiment [12] was used to define the expression change vector. Data was chosen from the 20.5 hour time point when expression changes were largest. The mean degree, , and the clustering coefficient C [13], are given in Table 1 for each of the three networks. These are global graph properties which can be used to compare network topologies.
Table 1 Network Parameters
Network Parameters
Network k C
Synthetic Network 1 40 0.0918
Synthetic Network 2 28 0.1034
Synthetic Network 3 40 0.0804
Biological Process 39 0.8636
Cellular Component 44 0.9461
Molecular Function 47 0.9444
Correlation Coefficient 155 0.5326
Correlation coefficient networks
A yeast stress data set consisting of 156 microarray experiments under a wide range of stress conditions was used to construct these networks. This data set is discussed in [14]. We randomly removed 15 experiments from the data set and each was used as the expression change vector. The correlation coefficient [15] was calculated for each gene pair using the reported expression changes for the remaining 141 experiments. Edges were defined in the network for pairs of genes with correlation r > 0.5. This data was taken from the Stanford Microarray Database [16]. Values for and C are given in Table 1.
Synthetic networks
To allow control over the network structure, synthetic networks were defined with 1000 genes. The genes were split into two sets, A and B, where the genes in set A ("changed genes") were allocated an expression change drawn from a N(2,1) distribution and the expression of the set B genes ("unchanged genes") were drawn from a N(0,1) distribution. Unless otherwise stated, |A| = 100 and |B| = 900 (the sizes of sets A and B). Edges were randomly assigned between genes with probabilities pA, pAB and pB, where these are the probabilities of two set A genes being connected, a set A gene being connected to a set B gene and two set B genes being connected, respectively. The clustering coefficient and the mean degree of the nodes depend on the values of pA, pAB and pB. Representative examples are listed in Table 1.
Synthetic network 1 is the standard case with equal expected degree across both sets and |A| = 100. Synthetic network 2 is the case where EdegA = 1.5EdegB. Also the relative connectivity () in both cases is 1 since this is where optimal results are observed (see relative expected degree section below). Synthetic network 3 is the case where |A| = 500 and again we have an equal expected degree across boths sets. Here the relative connectivity is 0.2618, again where optimal results are observed for this network (see Relative set size section below). The expected degree of a node in A is Edeg(A) = (|A| - 1)pA + |B|pAB and for set B is Edeg(B) = |A|pAB + (|B| - 1)pB.
To justify drawing the expression levels for sets A and B from N(2,1) and N(0,1) distributions, respectively, we compare the expression changes in a sample synthetic network to those in a yeast diauxic shift data set [12]. By measuring the mean and variance of the expression changes for every possible size of set A and B in both networks, we see that our method of assigning synthetic expression produces a distribution of values very similar to that observed in the diauxic shift experiment (see Figure 2).
Figure 2 Estimating μ. To justify drawing the expression of the set A genes from a N(2,1) distribution and the set B genes from a N(0,1) distribution, we compute the mean and variance for every possible size of set A and B in a synthetic network and compare this with an experimental data set.
Testing the algorithm
Synthetic networks
Since we are trying to improve the ranking of genes produced in microarray experiment, we need to quantify the quality of the ranking produced by the algorithm. In the case of synthetic data, we know that all genes in set A should be ranked above the genes in set B, which gives us a basis for comparing the ranking from experimentally observed fold-changes with the re-ordered ranking produced by GeneRank. To quantify GeneRank results, we used the Area Under the Receiver Operating Characteristic Curve (AUC) [17,18]. This value describes how well the ranked list discriminates between genes in set A and set B. It will have a value of 0.5 if sets A and B are randomly mixed and a value of 1 if they are perfectly separated, i.e. if all the genes of set A ("true changed genes") appear at the very top of the list. We emphasise that with d = 0 the algorithm is equivalent to ranking on pure fold-change.
The construction of synthetic networks allows us to obtain full control over the network structure. We experimented with various network parameters.
Relative connectivity and expression-connectivity weighting
We measure the relative connectivity as since this is the expected number of connections in A divided by the expected number of connections between sets A and B. The expression-connectivity weighting parameter d is tested at 0.05 intervals in the range [0.05..0.95]. We vary both relative connectivity and expression-connectivity weighting in all tests along with one other variable. All results shown are averaged over 5 runs of the experiment and the AUC is calculated for each combination in the parameter space.
Relative expected degree
We carried out a number of tests where Edeg(A) = αEdeg(B), for 1 ≤ α ≤ 1.5. Here Edeg(A) and Edeg(B) denote the expected degree of a gene in sets A and B, respectively. Results are given in Figure 3. In all cases we see an improvement achieved by the algorithm compared to pure fold-change ranking. The effect is only slight when the expected degree of every node is equal throughout the network (α = 1), and increases for larger values of α. In addition, we never dramatically decrease the quality of the results for low values of d (d ≤ 0.8). Also, the range of d for which we see an approximately constant high AUC increases as the relative connectivity increases.
Figure 3 Varying the relative expected degree between sets A and B. Here we are varying the expected degrees of sets A and B. The GeneRank algorithm provides slight improvement over pure expression based ranking when the expected degree of boths sets is equal. As the expected degree of set A becomes larger compared to that of set B, the improvement observed over expression ranking increases. We are measuring the AUC of the ranking (max = 1), averaged over 5 experiments. The lines on the diagrams indicate constant values of the AUC. The black cross indicates where the maximum AUC occurs, and hence shows for what values of d and relative connectivity the method works best.
A number of observations can be made from the experimental results where Edeg(A) > Edeg(B):
• The maximum AUC achieved by the algorithm increases as the difference between Edeg(A) and Edeg(B) increases. For the case where α = 1.5, we come particularly close to the maximum value of 1, (0.98).
• As the difference between Edeg(A) and Edeg(B) increases, we see less distortion of results even for unsuitably high values of d. This is consistent with the fact that increasing d gives connectivity a greater influence in the ranking.
• The maximum AUC achieved in each case occurs at larger values of d as the difference between the expected degrees increases. Again this is consistent with a high value of d corresponding to a greater weighting of connectivity on the algorithm.
• The improvement by the algorithm over expression change ranking is greater when the difference between the expected degree of both sets is greater.
To summarise these findings, a higher expected degree of set A genes compared with set B results in the algorithm producing a higher AUC, and hence more accurate results.
Relative set size
Four cases were investigated, where |A| = 50,100, 200 and 500. In each case |B| = 1000 - |A|, i.e. between 5 and 50% of the genes were defined as "differentially expressed". Networks were constructed to have equal expected degree across all nodes. Results are given in Figure 4. We see that the performance of the algorithm varies with |A|. The best results (highest AUC) are achieved at |A| = 100. In addition, an improvement over expression change ranking is observed for |A| = 50 and |A| = 200. As expected, with |A| = 500, where half of the total network is in set A, the algorithm performs most poorly and generally fails to give an improvement.
Figure 4 Varying the relative sizes of sets A and B. Our previous tests used |A| = 100. Now |A| is varied and the AUC is again calculated, averaged over 5 experiments. The maximum AUC for each test is shown by the black cross. The expected degree of sets A and B are equal in all cases. The only case where the algorithm produces a deterioration of the expression based ranking is where A = 500, which is half of the total network size. In all other cases, the algorithm increases the AUC compared to that achieved by pure expression ranking. The highest AUC is achieved in the case where |A| = 100.
These results on synthetic networks suggest that for certain network structures GeneRank can achieve a significant improvement over ranking based on pure differential expression. The relative expected degree of sets A and B has a considerable effect on performance. In cases where the algorithm performs well (Edeg(A) > Edeg(B)), the optimal results occur when 0.75 ≤ d ≤ 0.85. It is curious, but probably pure coincidence, that the value d = 0.85 is reportedly used by Google [1]. At such a high value of d, the algorithm is giving significant weight to connectivity information, as is appropriate considering the high level of experimental noise in the simulated expression data.
However, in our tests the quality of the results generally decreases for values of d beyond ≈ 0.85, showing that we do require some expression change information to make the best possible interpretation. It appears that although optimal results arise when there is some expression considered in the ranking, a major contributory factor to the success of the algorithm is the high relative degree of the genes that are differentially expressed. This structure is certainly not present in all biological networks. In the next section we explore whether suitable real networks can be identified.
GO networks
As described earlier, we construct the GO networks by defining an edge between two genes if they share an annotation allocated by the Gene Ontology Consortium. This allows us to construct three networks, one for each section defined by the Gene Ontology: Biological Process, Cellular Component and Molecular Function.
An initial test combined the real network connectivity with synthetic expression changes. We ordered the genes based on expression change in the yeast diauxic shift experiment and allocated the top 300 down-regulated genes to set A. We know from the synthetic network testing that it is preferable for the algorithm to have a higher expected degree of set A compared to set B. This is the case with each of the three networks where the set A genes were chosen in this way. Synthetic expression was assigned as before to the genes in sets A and B. This allowed us to quantify the results produced by the algorithm. We used the algorithm to find the AUC for 0.05 ≤ d ≤ 0.95. Results for the three networks are shown in Figure 5, and in each case we see that the algorithm is able to improve on expression change ranking. In particular, with values of 0.05 ≤ d ≤ 0.5 we increase the AUC for all cases, and for general use of the algorithm we would therefore suggest d = 0.5 would be an appropriate choice for d. This increase of AUC is only slight in the case of the Molecular Function network, but is more dramatic for the Cellular Component network. Also, on average, the AUC is high for the Cellular Component network, and is higher for all values of d than is obtained for the other networks.
Figure 5 Combining real connectivity with synthetic expression. The network connectivity for each of the three GO networks: Biological Process, Cellular Component and Molecular Function were combined with synthetic expression data. In each case, the 300 most down-regulated genes which were defined using the real yeast diauxic shift data were allocated expression from a N(2,1) distribution and the remainder of the genes were given expression from a N(0,1) distribution. Again we measure the AUC for the ranking. In all cases an improvement over expression ranking is observed for lower values of d although for the Molecular Function network this change is slight. Applying the algorithm to the Cellular Component network achieves the highest overall AUC. The results begin to decrease in quality for d > 0.55, except for the Cellular Component network where the results decrease for d ≈ 0.65.
To check if GeneRank produces a gene ranking which is robust to noise we conducted a further experiment using the GO networks. Real experimental data were used throughout. The Cellular Component network was used in this experiment. For each of the top 200 genes sorted by differential expression, we set its expression change to 0 in turn (i.e., defined it as "unchanged") and determined if the GeneRank algorithm was able to pick up this anomaly and consequently move the gene towards its original place in the ranked list. The premise here is that its connections to other highly changed genes will boost the artificially altered gene in the ranking. The same experiment was done for 200 randomly selected genes. The results are given in Figure 6.
Figure 6 GO networks: testing the 'boosting' ability of the algorithm. An experiment was carried out to assess how well the algorithm is able to increase the relative ranking of a gene based on its connections to other highly changed genes. The top 200 most changed genes are set in turn to have a differential expression of 0. If the ranking were based on pure differential expression only, each gene would appear at the bottom of the list. By PageRanking, we raise the position of the gene closer to its original ranking. The same effect is not observed when a random 200 genes are chosen. The majority of these genes will not have connections to other highly changed genes. The blue line represents the original expression ranked position, the red circles show the original GeneRanked position and the blue asterisks show the modified GeneRanked position.
To quantify the results we calculated the quality index B as
where alt_PR is the GeneRanked position after the expression of the gene has been artificially altered, orig_exp_rank is the original expression-based position in the list, and alt_exp_rank is the expression-based position after the differential expression has been set to 0.
In the case where we are altering a gene in the top 200, a 'boosting' effect is observed and the ranked position after the fold-change has been moved towards the original ranked position. We can observe groups of genes which are boosted to the same level (shown by 'lines' of blue asterisks). It is likely that these genes are a completely connected subgraph, which results in all genes being given the same ranking. Altering genes which were originally ranked in the top 200, we achieve B = 0.7728. This effect is not observed for a random set of genes (B = 0.4165).
Correlation coefficient networks
Using the correlation coefficient network defined by the stress data set [14] we carried out the same experiment to check the robustness of the ranking produced by the algorithm. Here we have one network with different expression change vectors to be used as input to the algorithm. Each of the 15 experiments in the data set, which were not used in the network construction, was used as the input expression vector in turn. Results for six representative experiments are shown in Figure 7. The expression change of each gene was set to 1 in turn, i.e. defining it as "only slightly changed".
Figure 7 Correlation Coefficient networks: testing the 'boosting' ability of the algorithm. The same experiment as in Figure 6 was carried out on the correlation coefficient networks. In each case the network connectivity is identical but the expression change vector, used as input to the algorithm, is randomly chosen to be one experiment from the stress data set. The x-axis represents the top 200 genes when ranked using expression change information. We calculate a 'boosting' measure to quantify how much we increase the relative rank of each gene after it has been altered. In this case, each gene was changed to have expression change 1. The large values for B (max = 1) indicate that the algorithm achieves a high level of 'boosting' due to the connections that the altered genes have to highly changed genes. The blue line represents the original expression ranked position, the red circles show the original GeneRanked position and the blue asterisks show the modified GeneRanked position.
Again we calculate a value of our quality index B as described above. As a result of the high degree of the top 200 genes, we see that a high level of 'boosting' is achieved, as demonstrated by the high values of B. In other words, we are able to significantly raise the position of the altered gene in comparison with the ranked position that would have been observed had the standard expression change ranking been used.
Conclusion
The purpose of this work was to explore the possibility of adapting the PageRank algorithm, used by Google in assessing the importance of web pages, for the task of prioritizing the 'importance' of genes in a microarray experiment. Our new algorithm, GeneRank, allows connectivity and expression data to be combined to produce a more robust and informed summary of an experiment, compared to the standard procedure of basing the importance of a gene on its measured expression change. Although we use expression change values as expression data, this is not restricted, and some other means of capturing the expression information may also be used. GeneRank can be justified theoretically and has been tested on synthetic data, experimental data and a combination of both. The algorithm has a single parameter, d, that controls the relative weighting of expression and connectivity information. A value d = 0 ranks genes based on pure expression information and a value d = 1 ranks on pure connectivity degree. The optimal value of d is data-dependent, but based on our results we suggest d = 0.5 for general use. With d = 0.5 we observed no deterioration and generally an improvement over ranking based on pure expression change in the case where we combined real connectivity information with synthetic expression changes. GeneRank is simple to implement, gives a principled approach to combining different data types, and is a novel instance of applying search engine technology to this important task. We note that GeneRank results are not designed to replace the actual expression measurements, but should be used alongside the results with additional biological knowledge, to draw attention to unusual structures within the data. For example, a gene which is not viewed as important from the microarray results alone but is highly ranked in the GeneRank results, should be given further biological consideration.
While the improvement of gene rankings upon application of GeneRank is already significant in the examples presented, it may become even more so once comprehensive high-quality biological network information becomes available. Of particular interest in that respect will be transcriptional regulatory networks, such as are now being generated by technologies like ChIP-chip (see [19-21] for early examples using yeast as a model organism). As discussed above, the information encoded in such regulatory networks will be intuitively amenable to GeneRank analysis. It will also re-introduce an element of directedness into the network, moving it even closer to the original PageRank application.
Methods
The original algorithm
We summarize the basic PageRank algorithm which was developed by Larry Page and Sergey Brin at Stanford University [2] and forms the basis of the successful search engine Google. Further details may also be found in [1,22].
PageRank assigns a measure of relevance or importance to each web page, allowing Google to return high-quality pages in response to a user query. The algorithm is designed to be robust to methods of deception, where web page designers attempt to artificially boost the PageRank of their page by altering the local link structure. Robustness follows from the recursive nature of the algorithm, where a page is highly ranked if it is linked to by other highly ranked pages. A link from page i to page j is regarded as a "vote of confidence" for page j from page i. The algorithm views the web as a directed graph G(V, E), where the N nodes V are the web pages and the edges E represent the links between pages. This information can be stored in an adjacency matrix, W ∈ RN × N, where wij = 1 if there is a link from page i to page j and wij = 0 otherwise. We define degi := to be the degree (more precisely, the out-degree) of the ith page. Suppose we have assigned an initial ranking r[0] ∈ N. The PageRank algorithm proceeds iteratively, updating the ranking for the jth page from to according to the formula
Here denotes the ranking of page j at the nth iteration and d ∈ (0,1) is a fixed parameter. The value d = 0.85 appears to be used by Google [1,2]. We see from (1) that the rank of a page depends on the rank of all pages that link to it. Scaling by 1/degi in the summation ensures that each page has equal influence in the voting procedure. Each page gets a rank of 1 - d automatically and also gets d times the votes given by other pages. Iterating to convergence in (1) is equivalent to solving for r ∈ N in the linear system
(I - dWT D-1)r = (1 - d)e, (2)
where I is the identity matrix, WT is the transpose of W, D = diag(degi) and e ∈ N has all ei = 1. Applying PageRank is equivalent to applying the Jacobi iteration [23] to (2), and convergence to a unique solution r is guaranteed under the condition
ρ(dWTD-1) < 1, (3)
where ρ(·) denotes the spectral radius. The convergence condition (3) holds for any 0 <d < 1.
A random walk interpretation
The PageRanking process has an alternative interpretation in terms of a random walk [1,2,22]. Suppose that a random walker is currently at page i. On the next step the walker
teleports: with probability 1 - d moves to a new page, chosen uniformly over all web pages, or,
surfs: with probability d moves to a page that is linked to from page i; in this case each page j such that wij = 1 is equally likely to be chosen as the destination.
The PageRank vector r, when normalised so that its components sum to one, corresponds to the invariant measure for this process. In other words, rj is the long-time proportion of visits made to page j. A further interpretation based on mean hitting times rather than invariant measures is given in [24]. The biological implication of the random walk interpretation is discussed in the description of the algorithm in the Results and Discussion section.
The modified algorithm: GeneRank
The PageRank idea translates intuitively to the analogous situation of gene expression analysis. Instead of producing a ranked list of web pages, we produce a ranked list of genes. PageRank views hyperlinks as votes of confidence, so we similarly allow functional connections to boost rank. Just as PageRank counts votes from a highly-ranked page as more influential than votes from a lowly-ranked page, we will allow connections to genes with high differential expression to carry greater significance than connections to genes with low differential expression. Figure 1 gives a graphical view of the concept.
PageRank gives each web page a rank of (1 - d) "for free". We will adapt this to give each gene a rank of (1 - d)exi, where exi is the absolute value of the expression change for gene i. Letting denote the ranking of gene j after the nth iteration, we take initial ranking r[0] = ex/||ex||1, where ||·||1 denotes the vector 1-norm. Then we let
Here W ∈ N × N is the connectivity matrix for the gene network, so wij = wji = 1 if genes i and j are connected and wij = wji = 0 otherwise.
We remark that this iteration may also be motivated from the viewpoint of personalised PageRanking [1,2], where teleporting jumps in the random walk process are biased towards a user's preferred locations – here, we are biasing according to expression level.
The iteration (4) corresponds to Jacobi on the system
(I - dWT D-1)r = (1 - d)ex, (5)
and, because the iteration matrix has not been altered, the condition that convergence is guaranteed for all 0 <d < 1 continues to hold. Since W is symmetric as the network is undirected, we could replace WT by W. This is unlike the original algorithm, where a directed network is used.
In summary, the GeneRank algorithm is finding the customised ranking vector r defined by the linear system (5). A Matlab implementation of the algorithm is available in the additional file geneRank.m The random walk interpretation carries through to this more general setting. If the teleporting step is re-defined so that the destination gene is not chosen uniformly over the whole set, but rather is chosen with probability proportional to absolute expression level, then r in (5), suitably scaled, is the invariant measure. Overall, we have a true generalization of PageRank in the sense that (a) the algorithm has both "vote of confidence" and "random walk" interpretations and (b) for the case where all exi = 1 we recover the original PageRank algorithm.
It is trivial to check that with the choice d = 0 the system (5) has solution r = ex. In this case the genes are ranked purely on expression level. We will now study the other extreme, where d = 1, and show that this case may be regarded as ranking purely on connectivity.
For d = 1, the iteration (4) becomes
and the system (5) for the corresponding fixed point becomes
(I - WT D-1)r = 0. (7)
First, we show that the sum of the rankings is preserved by the iteration. From (6),
Also, it is clear from (6) that the iteration preserves the nonnegativity of the initial ranks; that is, ≥ 0. Next, we note that deg/||deg||1 is a fixed point of (6). To see this, put r[n-1] = deg/||deg||1 in the right-hand side of (6) to obtain
Now, we observe that ρ(WTD-1) ≤ ||WTD-1||1 = 1, and hence all eigenvalues of WTD-1 are less than or equal to 1 in modulus. Because W is symmetric, we have WTD-1deg = WTe = deg, showing that there is at least one eigenvector, deg, corresponding to eigenvalue 1. Suppose now that λ = 1 is a simple eigenvalue of WTD-1 and that r* with ||r*||1 = 1 is another solution of (7). Then
So r* - deg/||deg||1 is an eigenvector of WTD-1 corresponding to eigenvalue 1. It follows that r* - deg/||deg||1 must be a multiple of deg and hence r* = ± deg/||deg||1 . We may summarize our findings in the following result.
Result If the eigenvalue λ = 1 of WTD-1 is simple, then r = deg/||deg||1 is the unique solution of (7) that satisfies the required constraints ||r||1 = 1 and ri ≥ 0.
Overall, we conclude that the extremal parameter values d = 0 and d = 1 represent ranking by pure expression level and pure degree, respectively, and hence by changing the value of d we may interpolate between these two extremes.
Abbreviations
ROC – Receiver Operating Characteristic
AUC – Area Under ROC Curve
GO – Gene Ontology
Authors' contributions
JLM implemented the algorithm, performed the experiments, and drafted the manuscript. RB provided biological datasets and interpretation, and participated in the experimental design. DRG contributed to study coordination and continuous discussions. DJH first conceived of the application of PageRanking to biological data, and participated in study design and supervision. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
The Matlab GeneRank implementation. The file contains a Matlab implementation of the GeneRank algorithm. The file requires a matrix describing the network connectivity and a vector of expression changes for each gene. The output is the vector of rankings for each gene.
Click here for file
Additional File 2
A Matlab .mat file containing sample GO networks and expression change vectors. This file can be loaded into Matlab using the command load G0_matrix. This will load three matrices (w_All, w_Up and w_Down) and three expression change vectors (expr_data, expr_dataUp and expr_dataDown) into the current workspace. These matrices were constructed using the all three sections of the Gene Ontology, where a link is present between two genes if they share a GO annotation. Only genes which are up-regulated are included in w_Up and only down-regulated in w_Down. The GeneRank algorithm should be used with the corresponding matrix and expression change vector, e.g. the command ranking = geneRank(w_Up, expr_dataUp,d) would be used to calculate the ranking of the up-regulated genes in the experiment.
Click here for file
Acknowledgements
JLM was supported by a Synergy scholarship, a jointly supervised studentship between the universities of Strathclyde and Glasgow.
RB was supported by a BBSRC grant (17/G17989) to Anna Amtmann and a personal research fellowship from the Caledonian Research Foundation..
DJH was supported by a Fellowship from the Royal Society of Edinburgh/Scottish Executive Education and Lifelong Learning Department and by the EPSRC grant GR/562383/01.
==== Refs
Langville AN Meyer CD Deeper inside PageRank Internet Mathematics 2005 1 335 380
Page L Brin S Motwani R Winograd T The PageRank citation ranking: bringing order to the web Tech rep Stanford Digital Library Technologies Project 1998
Tilstone C Vital statistics Nature 2003 424 610 612 12904757 10.1038/424610a
Hanash S Creighton C Making sense of microarray data to classify cancer The Pharmacogenomics Journal 2003 3 308 311 14597962 10.1038/sj.tpj.6500209
Breitling R Amtmann A Herzyk P Graph-based iterative Group Analysis enhances microarray interpretation BMC Bioinformatics 2004 5 100 15272936 10.1186/1471-2105-5-100
Ideker T Ozier O Schwikowski B Siegel AF Discovering regulatory and signalling circuits in molecular interaction networks Bioinformatics 2002 18 S233 S240 12169552
Tornow S Mewes H Functional modules by relating protein interaction networks and gene expression Nucleic Acids Research 2003 31 6283 6289 14576317 10.1093/nar/gkg838
Yamanishi Y Vert JP Kanehisa M Protein network inference from multiple genomic data: a supervised approach Bioinformatics 2004 20 i363 i370 15262821 10.1093/bioinformatics/bth910
Kemmeren P van Berkurn NL Vilo J Bijma T Donders R Brazma A Holstege FC Protein Interaction Verification and Functional Annotation by Integrating Analysis of Genome-Scale Data Molecular Cell 2002 9 1133 1143 12049748 10.1016/S1097-2765(02)00531-2
Google
Gene Ontology
DeRisi JL Iyer VR Brown PO Exploring the metabolic and genetic control of gene expression on a genomic scale Science 1997 278 680 686 9381177 10.1126/science.278.5338.680
Zhou H Scaling exponents and clustering coefficients of a growing random network Phy Rev E Stat Nonlin Sofa Matler Phys 2002 66 016125
Gasch AP Spellman PT Kao CM Carmel-Harel O Eisen MB Storz G Botstein D Brown PO Genomic expression programs in the response of yeast cells to environmental changes Molecular Biology of the Cell 2000 11 4241 4257 11102521
Weisstein EW Correlation Coefficient From MathWorld – A Wolfram Web Resource
Stanford Microarray Database
Bamber D The area above the ordinal dominance graph and the area below the receiver operating characteristic graph Journal of Mathematical Psychology 1975 12 387 415 10.1016/0022-2496(75)90001-2
Gribskov M Robinson NL Use of receiver operating characteristic (ROC) analysis to evaluate sequence matching Computer and Chemistry 1996 20 25 33 10.1016/S0097-8485(96)80004-0
Ren B Robert F Wyrick JJ Aparicio O Jennings EG Simon I Zeitlinger J Schreiber J Hannett N Kanin E Volkert TL Wilson CJ Bell SP Young RA Genome-wide location and function of DNA binding proteins Science 2000 290 2306 2309 11125145 10.1126/science.290.5500.2306
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
Horak CE Luscombe NM Qian J Bertone P Piccirrillo S Gerstein M Snyder M Complex transcriptional circuitry at the G1/S transition in Saccharomyces cerevisiae Genes & Development 2002 16 3017 3033 12464632 10.1101/gad.1039602
Higham DJ Taylor A The sleekest link algorithm Inst of Math and Its Appl (IMA) Mathematics Today 2003 39 192 197
Golub GH Van Loan CF Matrix computations 1996 third The John Hopkins University Press
Higham DJ Google PageRank as mean playing time for pinball on the reverse web Applied Mathematics Letters
|
16176585
|
PMC1261158
|
CC BY
|
2021-01-04 16:27:27
|
no
|
BMC Bioinformatics. 2005 Sep 21; 6:233
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-233
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2361618803710.1186/1471-2105-6-236SoftwareSIMPROT: Using an empirically determined indel distribution in simulations of protein evolution Pang Andy [email protected] Andrew D [email protected] Paulo AS [email protected] Elisabeth RM [email protected] Ontario Cancer Institute, University Health Network, Toronto, Ontario, Canada2 Dept. Medical Biophysics, University of Toronto, Toronto, Ontario, Canada3 Cold Spring Harbor Laboratory, 1 Bungtown Road, Cold Spring Harbor, NY 11724 USA2005 27 9 2005 6 236 236 29 4 2005 27 9 2005 Copyright © 2005 Pang et al; licensee BioMed Central Ltd.2005Pang 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
General protein evolution models help determine the baseline expectations for the evolution of sequences, and they have been extensively useful in sequence analysis and for the computer simulation of artificial sequence data sets.
Results
We have developed a new method of simulating protein sequence evolution, including insertion and deletion (indel) events in addition to amino-acid substitutions. The simulation generates both the simulated sequence family and a true sequence alignment that captures the evolutionary relationships between amino acids from different sequences. Our statistical model for indel evolution is based on the empirical indel distribution determined by Qian and Goldstein. We have parameterized this distribution so that it applies to sequences diverged by varying evolutionary times and generalized it to provide flexibility in simulation conditions. Our method uses a Monte-Carlo simulation strategy, and has been implemented in a C++ program named Simprot.
Conclusion
Simprot will be useful for testing methods of analysis of protein sequence families particularly alignment methods, phylogenetic tree building, detection of recombination and horizontal gene transfer, and homology detection, where knowing the true course of sequence evolution is essential.
==== Body
Background
Protein evolution has been largely modelled by considering the amino acid substitution process. There have been few statistical studies of the processes of insertion and deletion. Thorne et al. (1991) [2] described a theoretical parametric model that has been used to model the processes of insertion and deletion of single amino acids. The model has been extended, and others developed, to include the consideration of longer indels ([3-5]), however a model based on actual sequences may be more realistic and therefore preferable.
The study Benner, Cohen and Gonnet (1993) [6] is therefore a landmark one. The distribution of indels length was empirically determined from the alignment of conserved proteins with less than 100 PAM units of sequence divergence. This limit on the range of divergence was established in order to reduce both the redundancy of indel events counted, and the numbers of indels that resulted from independent overlapping events. In that study and in a more recent update [7], the estimate for the indel length distribution fit to a Zipfian distribution.
The study of Qian and Goldstein (2001) [1] on the other hand, derived an empirical distribution for the length of indels from a database of protein alignments sharing no more than 25% sequence identity. The distribution in that case fit a linear combination of 4 exponential functions. We call this function the Qian-Goldstein distribution and the Zipfian distributions found by [7] and [6], the Benner distributions. The Qian-Goldstein distribution is more applicable to protein sequence comparisons with long sequence divergence whereas the Benner distributions are more applicable to sequences of lower divergence. The Qian-Goldstein distribution was derived for the determination of realistic gap insertion and deletion penalties that are generally used in alignment algorithms. These affine gap penalties are used to mimic the fact that although insertions and deletions are rare events, they often involve more than one amino acid. That observation reflects the fact that some regions of protein sequence and structure are able to tolerate sections of insertion or deletion.
The evolutionary processes of mutation and subsequent natural selection determine the occurrence of substitutions, insertions and deletion. The specifics of the processes are difficult to model accurately since they are determined by many factors at all context levels (i.e. the population, the genome, the cell, and particularly the DNA and protein sequence and structure). However general protein evolution models are useful as they can help determine the baseline expectations for the evolution of sequences, and they have been extensively used for the computer simulation of artificial sequence data sets.
Two freely available programs that generate sets of sequences by Monte Carlo simulation of evolution are Seq-Gen [8,9], and Rose [10]. Seq-Gen generates sequences using a given evolutionary tree, making substitutions according to a specified model. Several models of amino acid substitution are available, including the popular PAM [11] and JTT models [12]. Additionally, Seq-Gen allows rates of evolution to vary between sites according to the gamma model developed by Yang [13]. Seq-Gen only considers substitutions and does not simulate the processes of insertion and deletion. On the other hand, the Rose program does simulate insertions and deletions, along with substitutions, but has the disadvantage of not allowing for different rates of evolution at different sites. The user determines the distribution of indel length used by Rose software. That distribution is then fixed and does not depend on evolutionary time (i.e. branch length in the tree); only the frequency of indels is determined by the branch length separating the ancestral and daughter sequences.
The empirically derived distribution of Qian-Goldstein [1] was obtained using a subset of structural sequence alignments corresponding to highly diverged sequences in the database. The distribution as such is limited to models of proteins corresponding to this set of circumstances. Although the Qian-Goldstein distribution is fixed with respect to evolutionary time, it has the property of being easily parameterized. We generalized the model so that it applies to proteins with variable sequence divergence and show that this generalized distribution may be comparable to the Benner distribution [7] at shorter evolutionary distances. We implemented our generalized Qian-Goldstein distribution in a new program for the simulation of protein sequences (Simprot). Like earlier programs, Simprot allows for several models of amino acid substitution, and permits gamma distributed sites rates according to the Yang [13] model. By incorporating our parameterized Qian-Goldstein model for indels, the user has flexibility to modify the distribution and obtain longer/shorter or more/less frequent insertions and deletions. Simprot is the first program to simulate protein sequence evolution with the additional capability of being able to simulate indels with a variable length distribution. Additionally, Simprot allows the protein sequence to be segmented such that the different segments can evolve with distinct sets of parameters and tree.
Results
Parametrization of the Qian-Goldstein indel length distribution
The empirically derived Qian-Goldstein distribution [1] (equation 8 in that paper) is given by
QG(n)=1.027×10−2e−n/0.96 + 3.031×10−3e−n/3.13 + 6.141×10−4e−n/14.3 (1) + 2.090×10−5e−n/81.7.
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@C3D8@
This function describes the frequency of an indel, of any length n > 0, as a fraction of the average length of the protein sequence. The model accurately describes a data set of aligned sequences with less than 25% sequence identity. The total frequency of indels is estimated by ∑n > 0QG(n), which converges rapidly to 0.0238. This value is close to the observed frequency of indels (0.030) that was found by Qian and Goldstein in database they analyzed [1].
As mentioned above, the dataset used to infer Equation 1 was highly diverged, so we may assume it accurately applies to sequences of large divergence. We will therefore assume that the Qian-Goldstein applies at an evolutionary distance c, a parameter to which evolutionary time t will be scaled. This allows us to define QG'(n, t, c) = QG(n) for n > 0 and t = c.
However this only defines the QG' function at one evolutionary time point, t = c. It is necessary to define the expected distribution of observed indel lengths for all evolutionary times. The Qian-Goldstein distribution describes the observed length frequency after a large amount of divergence, but it does not describe the actual distribution of the expected rate of fixation in the population of insertion and deletion mutations (the rate of indel occurrence). This is because a single observed indel may have been the result of several actual events. Even if the length distribution for indel occurrences were known, a Markov model for the process of insertion and deletion would need to be established and used to derive the expected distribution of observed indels for any given degree of divergence. Additional empirical data is needed to derive the expected distribution of observed indel lengths scaled to other divergence times.
In the absence of additional empirical data, we must make some assumptions about the insertion and deletion processes to derive the indel length distribution for all evolutionary time.
1. We assume that the length of indels will increase with evolutionary time as larger indels are more easily tolerated and smaller ones overlap. We therefore expect that shorter indels arise over smaller divergence times and that larger indels are the result of independent but contiguous events. We design the distribution such that it is has the property that the limit as time goes to 0 for the expected frequency of all indels (>1), is also 0. This assumes that the instantaneous rate of an indel involves only a single amino acid, which is unlikely (see for example [14]). The assumption is only approximatively true even if the mutation process created only single amino acid indels because multiple mutations may be fixed by natural selection and genetic drift. The effect on the indel model will be that the lengths of indels may be underestimated for very low sequence divergence.
2. We design the distribution such that it is time-reversible. This makes the assumptions that the probability of insertions is equal to the frequency of deletions and that these have equal length distributions. Data from DNA genome level comparisons [15,16] indicate these assumption are not necessarily true, but the effects of this on the long range evolution in proteins is not clear. The Qian-Goldstein and Benner distributions assume time reversibility since the direction of events was not known for the protein sequences they analyzed. Time-reversibility is a desirable mathematical property that is often used in sequence analysis programs for alignment and phylogeny.
3. We assume that the observed indel length distribution keeps its original form as a sum of four exponential terms at any fixed time point, and not just for time t = c. This is consistent with the assumption in the original Qian-Goldstein distribution, which fits four exponential terms. Using a function of this form allows us to scale the exponential in each term separately.
4. There are still many ways to introduce the time parameter t into the function. Our third assumption was then to chose a simple linear scaling of the exponents of the function with time. We found this scaling to give reasonable results when we compare the Benner distributions which were obtained at shorter time scales (see below).
With these assumptions, we then define the scaled QG' function for n > 0 as
QG'(n,t,c)=1.027×10−2e−nc/0.96t + 3.031×10−3e−nc/3.13t + 6.141×10−4e−nc/14.3t (2) + 2.090×10−5e−nc/81.7t.
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@22A6@
To turn equation 2 into a probability distribution (which sums to 1), we must divide the function by the sum of all values for n > 0 such that that
GQGc(n,t)=QG'(n,t,c)∑n=1∞QG'(n,t,c). (3)
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGhbWrcqWGrbqucqWGhbWrdaWgaaWcbaGaem4yamgabeaakiabcIcaOiabd6gaUjabcYcaSiabdsha0jabcMcaPiabg2da9maalaaabaGaemyuaeLaem4raCKaei4jaCIaeiikaGIaemOBa4MaeiilaWIaemiDaqNaeiilaWIaem4yamMaeiykaKcabaWaaabmaeaacqWGrbqucqWGhbWrcqGGNaWjcqGGOaakcqWGUbGBcqGGSaalcqWG0baDcqGGSaalcqWGJbWycqGGPaqkaSqaaiabd6gaUjabg2da9iabigdaXaqaaiabg6HiLcqdcqGHris5aaaakiabc6caUiaaxMaacaWLjaWaaeWaaeaacqaIZaWmaiaawIcacaGLPaaaaaa@58E4@
We call GQG the Generalized Qian-Goldstein distribution. GQG is a scaled version of the QG function that describes the probability distribution of indels of length n (conditional on n > 0) at any evolutionary time t and assuming an evolutionary scale factor c.
In Figure 1 the distribution of indel lengths is shown plotted for varying values of c and t. In figure 2, we compare the GQG distribution (with parameters c = 3 and PAM 50, which are very appropriate) with the data from [7] which was obtained from sequence comparisons of PAM < 100. The striking fit of the GQG distribution to data of much lower sequence divergence indicates that our scaling of the original QG distribution is appropriate.
Figure 1 The GCG distribution of indel length is determined by the evolutionary distance for a given evolutionary scale factor c. The expected frequency of indels of given lengths are plotted. In a. the distribution is shown for different evolutionary distances (as labelled next to the corresponding lines). In b. the evolutionary distance is fixed and the GCG length distribution is plotted for different evolutionary scale factor values (as labelled next to the corresponding lines).
Figure 2 Comparison of the GQG distribution with the data obtained from the study [7] for protein sequences with less than 100 PAM sequence divergence. The parameters of the GQG disribution are set to the default c = 3 and t = PAM 50. These values were chosen simply because they seemed reasonable, not to maximize the fit of the curve to the data. The striking fit indicates that our scaling of the QG distribution is appropriate to model indels at lower levels of sequence divergence.
Once defined in this way, the GQG does not give the frequency of indels (only their length distribution). The rates for the assumed four independent poisson processes for the appearance of indels can be combined into a single instantaneous rate z. The frequency of indels defines p such that
p = 1 - e-zt/c. (4)
We define the indel frequency rate p as a parameter from which z can be calculated. Figure 3 shows the frequency of indels as z is increased for different values of the parameter c such that p = 0.03 (the observed Qian-Goldstein frequency).
Figure 3 The indel probability of the GCG distribution is determined by the indel rate z (x-axis) and the evolutionary scale factor c (labelled next to the corresponding line). This probability can be set by the user to influence the number of indels present in the final alignment.
By introducing parameters in the distribution, we allow a large amount of flexibility in the generation of indels and on their lengths. The indel frequency parameter p can modify the indel frequency and the evolutionary scale factor c parameter can be used to independently modify the distribution of indel lengths. Larger values of p will yield more indels and smaller values of will yield fewer indels. Larger values of c will yield shorter indels and smaller values of c will yield larger indels.
The original impetus for the estimation of the indel distribution was to derive gap insertion (γI) and gap extension (γE) penalties for use in alignment programs [1]. We used the formulas from [1] to derive an approximation for the natural log odds penalties for gaps
γE≈GQGc(3,t)−GQGc(1,t)2GQGc(2,t) (5)
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqaHZoWzdaWgaaWcbaGaemyraueabeaakiabgIKi7oaalaaabaGaem4raCKaemyuaeLaem4raC0aaSbaaSqaaiabdogaJbqabaGccqGGOaakcqaIZaWmcqGGSaalcqWG0baDcqGGPaqkcqGHsislcqWGhbWrcqWGrbqucqWGhbWrdaWgaaWcbaGaem4yamgabeaakiabcIcaOiabigdaXiabcYcaSiabdsha0jabcMcaPaqaaiabikdaYiabdEeahjabdgfarjabdEeahnaaBaaaleaacqWGJbWyaeqaaOGaeiikaGIaeGOmaiJaeiilaWIaemiDaqNaeiykaKcaaiaaxMaacaWLjaWaaeWaaeaacqaI1aqnaiaawIcacaGLPaaaaaa@547B@
γI≈log(p1−eγE)+2γE. (6)
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqaHZoWzdaWgaaWcbaGaemysaKeabeaakiabgIKi7kGbcYgaSjabc+gaVjabcEgaNnaabmaabaWaaSaaaeaacqWGWbaCaeaacqaIXaqmcqGHsislcqWGLbqzdaahaaWcbeqaaiabeo7aNnaaBaaameaacqWGfbqraeqaaaaaaaaakiaawIcacaGLPaaacqGHRaWkcqaIYaGmdaWgaaWcbaGaeq4SdC2aaSbaaWqaaiabdweafbqabaaaleqaaOGaeiOla4IaaCzcaiaaxMaadaqadaqaaiabiAda2aGaayjkaiaawMcaaaaa@486B@
Implementation
We have implemented the Generalized Qian-Goldstein distribution in a program called Simprot to simulate protein sequence evolution. Given a bifurcating phylogenetic tree, children sequences inherit the sequence of their parent with modification due to mutation events. The number of mutations expected depends on the length of evolutionary time that separates the child from the parent and their type is determined by the chosen models. Substitutions are made according to the user-chosen substitution model. Insertions and deletions are made according to the GQG model described above. The user determines the values of the evolutionary scale factor c which controls the indel length distribution, and the indel frequency rate p which determines their frequencies. The shape parameter for the gamma model of [13] distribution of evolutionary rates is also determined by the user.
The parameters for the models are input via an interface screen (available through the Web, or by download for local Windows and Unix/Linux systems). The locally installed versions allow several input screens such that the resulting simulated protein sequences will consist of segments each evolving according to its own set of parameters and tree.
The program generates sequences according to the chosen indel and substitution models and outputs the alignments of sequences from the terminal branches. When several protein segments have been selected, the sequences are appropriately fused into single sequences by matching the names of the terminal taxons in the input tree files. The gap opening and gap extension penalties corresponding to the input parameters and time t = c for each protein segment is an additional output provided by the program for user reference.
Evolution
Each protein segment is simulated independently. Simprot parses the given tree file into a tree structure to use as a guide in simulating evolution. It then generates a random amino acid sequence of given length r at the root of the tree according to the equilibrium frequency of amino acids in the substitution model. Each amino acid site is assigned a rate of evolution based on the gamma distribution. The program then recursively generates mutations on the protein sequence at each of the tree nodes. There are two types of mutations: insertion/deletion and substitution. Indels are performed before substitutions at each tree node.
Number of indels
Simprot assumes a Poisson process for insertion and deletions and thus the expected frequency of indels (of any length) in a sequence is
p = 1 - e-zt/c, (7)
where z is the indel probability and t is the branch length to the daughter sequence which is scaled by the evolutionary scale factor. For each amino acid site, a uniformly distributed random number is picked to check whether it is lower than the expected frequency. The number of times this happens over the entire sequence becomes the number of indels that will be performed.
Indel positions
Indel sites are chosen according to their rate of evolution as given by the gamma distribution. This means that sites more likely to substitute will also be more likely to have an insertion or deletion.
Indel length
To determine the length of an indel after choosing to create one, the cumulative distribution function (CDF) of the indel-length probabilities for n > 0 as determined by the GQG model is evaluated using Eq. 3. A cap on the indel length is also applied. Indels must be shorter than the maximum indel length or 5% of the sequence length (whichever is smaller).
Indel type
Simprot chooses between insertion and deletion with equal probability. If insertion is chosen, an amino acid sequence of the indel length is generated (according to the same amino acid frequency distribution that generated the root sequence) and inserted before the indel position. If deletion is chosen, the indel length of amino acids are deleted beginning with the current position. If the length of the indel is greater than the number of amino acids in the sequence following the current position, additional amino acids are deleted towards the start of the sequence.
The probability of amino acids being inserted and deleted is the same so that the length of the sequences should remain approximately the same. The sequence is updated after each indel event and all indels are performed before substitutions.
Substitutions
Once all indels have been performed at a given node, Simprot performs substitutions of the individual amino acid according to the evolutionary substitution model. Currently the models implemented are PAM, JTT and PMB. The substitution probabilities are calculated from the previously calculated eigenvalue decomposition of the probability matrix. This strategy, first used by Felsenstein in the Phylip package [17] facilitates computation of the substitution probabilities for any branch length. The model considers the probability of all amino acid substitutions for a given branch length times the evolutionary rate at the site (as determined by the gamma model). As the program traverses the tree, the descendant nodes inherit the mutations generated.
Alignment
A copy of the "true" sequence alignment is also produced for the generated sequence family. At each node, the locations of insertions and deletions are maintained relative to the sequence at the parent node. This correspondence is called the "gapped sequence" because gap characters (-) are inserted in copies of both the current sequence and the parent sequence to represent the correspondence. After the sequence family has been generated, a recursive traversal rebuilds the true alignment using the gapped sequences. The procedure makes use of the fact that, for any node in the tree, the true alignments are known for the sequences in the left and right subtrees from this node, and the gapped sequences can be used to align these two true alignments, producing the true alignment for all sequences below the root. This procedure requires only a linear traversal of the tree, and therefore imposes no significant additional cost of computation. Simprot outputs the aligned sequences from the leaves of the tree in Fasta and Phylip format. It also creates a file of the set of unaligned protein sequences. If the protein is segmented, the files for the segments are merged into the final alignment.
Conclusion
While the process of amino acid substitution has been extensively studied and modelled, there has been relatively little study of the insertion-deletion process in protein coding sequences [18]. The model we propose may not fit all proteins but it has the properties of being based on an empirically derived distribution, and being flexible so as to allow a user to test many conditions. We plan to use additional empirical data of the frequency and distribution of indels in proteins to refine our model in subsequent releases of Simprot. The alignments generated by Simprot will be useful for testing methods of analysis of protein sequence families. It will be particularly useful for the development of new alignment methods, phylogenetic tree building, detection of recombination and horizontal gene transfer, and homology detection, where knowing the true course of sequence evolution is essential.
Availability and requirements
Project name: Simprot
Project home page:
Operating systems: Linux, Windows 95 or later (local installation)
Programming language: C++
License: University of Illinois/NCSA Open Source License
Any restrictions to use by non-academics: no
List of abbreviations
PMB probability matrix from Blocks, JTT Jones Taylor Thorton, PAM Percent Accepted Mutation, GQG Generalized Qian-Goldstein distribution, CDF cumulative distribution function.
Authors' contributions
AP implemented the GQG distribution in Simprot and helped draft the manuscript. ADS implemented the indel and substitution processes in Simprot and helped draft the manuscript. PASN implemented the gap penalties, created the GUI interfaces and provided comments on the manuscript. ERMT derived the GQG distribution, supervised the project and approved the final manuscript.
Acknowledgements
Simprot has been modified and recreated many times over the years and we thank all who have contributed to it: Shalini Veerassamy, Thomas Lui, Zhuozhi Wang, and Ginny Li. We thank Alex Kondrashov and another anonymous reviewer for their comments and suggestions for improving the manuscript. We thank CIHR and Genome Canada for funding.
==== Refs
Qian B Goldstein RA Distribution of Indel lengths Proteins 2001 45 102 4 11536366 10.1002/prot.1129
Thorne JL Kishino H Felsenstein J An evolutionary model for maximum likelihood alignment of DNA sequences J Mol Evol 1991 33 114 124 1920447 10.1007/BF02193625
Thorne JL Kishino H Felsenstein J Inching toward reality: an improved likelihood model of sequence evolution J Mol Evol 1999 34 3 16 1556741 10.1007/BF00163848
Metzler D Statistical alignment based on fragment insertion and deletion models Bioinformatics 2003 19 490 499 12611804 10.1093/bioinformatics/btg026
Miklos I Lunter GA Holmes I A Long Indel model for evolutionary sequence alignment Mol Biol Evol 2004 21 529 40 14694074 10.1093/molbev/msh043
Benner SA Cohen MA Gonnet GH Empirical and structural models for insertions and deletions in the divergent evolution of proteins J Mol Biol 1993 229 1065 82 8445636 10.1006/jmbi.1993.1105
Chang MS Benner SA Empirical analysis of protein insertions and deletions determining parameters for the correct placement of gaps in protein sequence alignments J Mol Biol 2004 341 617 31 15276848 10.1016/j.jmb.2004.05.045
Rambaut A Grassly NC Seq-Gen: an application for the Monte Carlo simulation of DNA sequence evolution along phylogenetic trees Comput Appl Biosci 1997 13 235 8 9183526
Grassly NC Adachi J Rambaut A PSeq-Gen: an application for the Monte Carlo simulation of protein sequence evolution along phylogenetic trees Comput Appl Biosci 1997 13 559 60 9367131
Stoye J Evers D Meyer F Rose: generating sequence families Bioinformatics 1998 14 157 163 9545448 10.1093/bioinformatics/14.2.157
Dayhoff MO Schwartz RM Orcutt BC Dayhoff MO A model of evolutionary change in proteins Atlas of Protein Sequence and Structure 1978 5 National Biomedical Research Foundation 345 352
Jones DT Taylor WR Thornton JM The rapid generation of mutation data matrices from protein sequences Computer Applications in the Biosciences 1992 8 275 282 1633570
Yang Z Maximum likelihood estimation of phylogeny from DNA sequences when substitution rates differ over sites Mol Biol Evol 1993 10 1396 1401 8277861
Kondrashov AS Rogozin IB Context of deletions and insertions in human coding sequences Hum Mutat 2004 23 177 85 14722921 14722921 10.1002/humu.10312
Ogurtsov A Aleksey Y Sunyaev S Kondrashov AS Indel-based evolutionary distance and mouse-human divergence Genome Res 2004 14 1610 6 15289479 10.1101/gr.2450504
Denver D Morris K Lynch M Thomas WK High mutation rate and predominance of insertions in the Caenorhabditis elegans nuclear genome 2004 430 679 82 15295601
Felsenstein J PHYLIP (phylogeny inference package) version 3.6.3 2002 Available via the web
Thorne JL Models of protein sequence evolution and their applications Curr Opin Genet Dev 2000 10 602 605 11088008 10.1016/S0959-437X(00)00142-8
|
16188037
|
PMC1261159
|
CC BY
|
2021-01-04 16:27:28
|
no
|
BMC Bioinformatics. 2005 Sep 27; 6:236
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-236
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2371619119010.1186/1471-2105-6-237SoftwareMeasuring similarities between transcription factor binding sites Kielbasa Szymon M [email protected] Didier [email protected] Hanspeter [email protected] Institute for Theoretical Biology, Humboldt University, Invalidenstraße 43, D-10115 Berlin, Germany2 Unité de Chronobiologie Théorique, Université Libre de Bruxelles, CP 231, Campus Plaine, Bvd du Triomphe, B-1050 Bruxelles, Belgium2005 28 9 2005 6 237 237 22 11 2004 28 9 2005 Copyright © 2005 Kielbasa et al; licensee BioMed Central Ltd.2005Kielbasa 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
Collections of transcription factor binding profiles (Transfac, Jaspar) are essential to identify regulatory elements in DNA sequences. Subsets of highly similar profiles complicate large scale analysis of transcription factor binding sites.
Results
We propose to identify and group similar profiles using two independent similarity measures: χ2 distances between position frequency matrices (PFMs) and correlation coefficients between position weight matrices (PWMs) scores.
Conclusion
We show that these measures complement each other and allow to associate Jaspar and Transfac matrices. Clusters of highly similar matrices are identified and can be used to optimise the search for regulatory elements. Moreover, the application of the measures is illustrated by assigning E-box matrices of a SELEX experiment and of experimentally characterised binding sites of circadian clock genes to the Myc-Max cluster.
==== Body
Background
In order to dissect the complex machinery of transcriptional control computational tools are widely used [1]. Candidate binding sites of known transcription factors are located by consensus sequence search or binding scores calculated from position weight matrices (PWMs) [2]. These matrices are derived from position frequency matrices (PFMs) obtained by aligning binding sites for a given transcription factor. PFMs contain the observed nucleotide frequencies at each position of the alignment. A popular collection of eukaryotic PFMs is given by the Transfac database [3]. Furthermore, an open-access database, Jaspar [4], has been compiled recently.
On-line tools are available to calculate high-scoring binding sites on the basis of these matrix collections [5-7]. For a given transcription factor these programs predict many binding sites (on average every 1000 bp) implying a high excess of false positives [1]. The situation is even worse if hundreds of different binding profiles are studied in parallel leading to multiple testing issues. Often these predictions overlap as a result of similarities of transcription factor binding profiles.
First steps to overcome the flood of false positive signals are accurate predictions of promoter regions and enhancers [8-10]. Phylogenetic footprinting [11-13], correlation with gene expression data [14,15] or analysis of cooperative binding of multiple transcription factors [16] allow to reduce the amount of false positives by at least an order of magnitude. Another helpful strategy is the a priori reduction of the number of matrices to be considered. However, a user-defined preselection of a few matrices is highly subjective and might hide novel interactions of several transcription factors. Therefore, in this paper we combine two objective criteria to measure similarities of transcription factor binding site profiles. These measures allow to construct groups of similar profiles. Representative matrices of the groups may be chosen and constitute a reduced and unbiased list of independent profiles for searching binding sites.
Similarities in the collections of matrices may arise from several sources:
1. Identical transcription factors are represented by different matrices. This appears, e.g., due to the distinct nomenclature in Transfac and Jaspar (for example the TATA-binding protein is referred as TATA in Transfac and as TBP in Jaspar) or due to the availability of matrices obtained with different methods (see for example Transfac matrices SRF_01 and SRF_Q6) or stringency criteria (see for example AP1_Q2 and AP1_Q6).
2. Factors within one family are represented by similar matrices due to the conserved structure of DNA-binding domains [17]. For example, both ATF and CREB matrices belong to the same bZIP family and recognise the TGACGT consensus sequence.
3. There might be so far undetected similarities of different transcription factor binding sites. Such similarities can point to a possible cross-talk between different regulatory pathways (see our discussion of E-box binding sites below).
4. It might be difficult to distinguish matrices for which only a few binding sites are known.
In order to identify similar matrices we combine two similarity measures. The first one is based on the χ2 distance of position frequencies of PFMs. The other utilizes scores from the corresponding position weight matrices (PWMs) – we calculate for a given pair of binding profiles the scores along a test DNA sequence and take the corresponding Pearson correlation coefficient as a similarity measure. Although related similarity measures have been already studied individually [15,17-21], our combined approach applied to the Transfac matrices reveals that the two selected measures capture different properties of the matrices and therefore the measures complement each other. Moreover, since for many matrices only a few experimentally verified binding sites are available we take into account these small sample sizes in both measures. The application of the measures is illustrated by mapping CLOCK-BMAL1 binding sites of circadian clock genes to the Myc-Max family.
Implementation
Databases
A commonly used database of experimentally verified transcription factor binding sites is Transfac [3]. The release from May 2004 provides 694 position frequency matrices (PFMs) covering vertebrates, plants, insects and fungi. Recently, a publicly available Jaspar database [4] was compiled with 108 PFMs associated mainly to vertebrates. For our large-scale statistical analysis we discarded all matrices with inconsistencies, for example matrices, where the number of sites aligned to construct the matrix (sample size) could not be determined. Furthermore, we excluded rather poor matrices with a length below 6 bases or a sample size below 5. After these consistency checks and filtering steps we arrived at 637 different matrices for Transfac and 103 matrices for Jaspar. All the matrices can be characterized by their length, the sample size, and the information content [22] (Tab. 1).
Table 1 Properties of Transfac and Jaspar matrices: We removed matrices for which the sample size was normalized to 100 and no information about the actual number of samples was available, as well as matrices of length below 6 or sample size below 5.
Property Transfac Jaspar
Number of original matrices 694 108
Number of matrices after filtering 637 103
Min length 6 6
Max length 30 30
Median length 12 11
Min sample size 5 6
Max sample size 389 389
Median sample size 18 23
Min information content 3.6 5.7
Max information content 44.3 26.2
Median information content 12.8 11.6
χ2 distance D between position frequency matrices
For each possible overlap (of at least 6 bases) of two PFMs we count the number of corresponding columns which are statistically independent. This task can be addressed by the homogeneity test using the χ2 measure with 3 degrees of freedom. The application of PFMs for the characterization of binding sites implies that the nucleotide positions are regarded as independent. Even though statistical dependencies between positions are known [23-25] the assumption of independent positions is a rather good approximation [1,26]. In the following we denote by fb,i and gb,i the entries of the overlapping parts of the two frequency matrices to be compared. The index i refers to the base position along the matrices and b enumerates the four nucleotides A, C, G and T. The χ2 distance at the position i is then given by:
χ2=∑b=A,C,G,T(Ng,ifb,i−Nf,igb,i)2Nf,iNg,i(fb,i+gb,i)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@6A6E@
where Nf,i = ∑bfb,i and Ng,i = ∑bgb,i are the sample sizes of the matrices columns at position i. If χ2 exceeds the threshold of χth2
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqaHhpWydaqhaaWcbaGaeeiDaqNaeeiAaGgabaGaeGOmaidaaaaa@3248@ (p = 0.05) = 7.81 the null hypothesis that the base counts in both columns are from the same distribution is rejected with a p-value of 0.05. In order to simplify the analysis we simply count the number of significantly different positions. The example in Fig. 1 shows that for an appropriate alignment (with shift = 3) of the two matrices all χ2-values are below the χth2
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqaHhpWydaqhaaWcbaGaeeiDaqNaeeiAaGgabaGaeGOmaidaaaaa@3248@ threshold and hence no column appears to be different. Although the counts in some columns look quite different the limited sample size allows no statistically significant discrimination.
Figure 1 CREB versus ATF matrices: The distance D is computed for each possible alignment between the two matrices. For each aligned column, we calculated the χ2 scores. D is then the number of χ2 values which exceed the threshold = 7.81 For shift= 0, the two matrices are not properly aligned, D = 7. For shift= 3, the two matrices are properly aligned, D = 0.
Obviously, the number of significantly different columns depends on the relative position of both matrices. In our algorithm we study all possible alignments with a minimum overlap of 6 bases and containing at least 75% of the information content of each matrix. We calculate the minimal number of different positions among these alignments. We call this number D and interpret it as the distance between the compared matrices. Fig. 1 illustrates that for a correct alignment of the ATF and CREB a distance D = 0 is obtained whereas other alignments lead to statistically significant different columns.
An advantage of the distance measure we use in comparison to earlier studies [15,17,19,20] is the emphasis on the limited sample size of many matrices. Only few binding sites, such as those recognized by the Sp1 factor, are characterized by hundreds of experimentally verified sites. The more common sample size is around 15–20 (see Tab. 1) and, thus, it is much more difficult to distinguish matrices. The χ2 measure leading to the distance D takes into account the limited sample size in a statistically well defined manner. The proposed measure could be generalized by allowing gaps, using the sum of scores or by taking the number of possible shifts into account. Since we studied in this paper only rather strong similarities our simple discrete threshold D ≤ 1 was sufficient.
Correlation C between position frequency matrices scores
The information on experimentally verified binding sites stored in PFMs can be exploited to predict novel sites. For this purpose position weight matrices (PWMs) can be constructed from the counts fb,i in the following manner [1,27]. First, the probability pb,i of a base b at a given position i is given by:
pb,i=fb,i+sbNi+∑b′=A,C,G,Tsb′
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGWbaCdaWgaaWcbaGaemOyaiMaeiilaWIaemyAaKgabeaakiabg2da9maalaaabaGaemOzay2aaSbaaSqaaiabdkgaIjabcYcaSiabdMgaPbqabaGccqGHRaWkcqWGZbWCdaWgaaWcbaGaemOyaigabeaaaOqaaiabd6eaonaaBaaaleaacqWGPbqAaeqaaOGaey4kaSYaaabeaeaacqWGZbWCdaWgaaWcbaGafmOyaiMbauaaaeqaaaqaaiqbdkgaIzaafaGaeyypa0JaeeyqaeKaeiilaWIaee4qamKaeiilaWIaee4raCKaeiilaWIaeeivaqfabeqdcqGHris5aaaaaaa@4D8D@
where Ni = ∑b' fb',i denotes the sample size at the position i leading to the relative frequency fb,iNi
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaadaWcaaqaaiabdAgaMnaaBaaaleaacqWGIbGycqGGSaalcqWGPbqAaeqaaaGcbaGaemOta40aaSbaaSqaaiabdMgaPbqabaaaaaaa@347B@. This estimator is modified using pseudo-counts sb. As suggested earlier [28] we choose sb = Ni4
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaadaWcaaqaamaakaaabaGaemOta40aaSbaaSqaaiabdMgaPbqabaaabeaaaOqaaiabisda0aaaaaa@3078@, i.e. the pseudo-count is proportional to the standard deviation of the counted frequencies. Such a choice of relatively large pseudo-counts has a pronounced effect on PWMs with a small sample size. Due to the pseudo-counts the estimated probabilities are strictly positive even if zeros appear in the PFM. From the estimated probabilities pb,i we obtain the weights wb,i as follows:
wb,i=log2pb,irb,
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWG3bWDdaWgaaWcbaGaemOyaiMaeiilaWIaemyAaKgabeaakiabg2da9iGbcYgaSjabc+gaVjabcEgaNnaaBaaaleaacqaIYaGmaeqaaOWaaSaaaeaacqWGWbaCdaWgaaWcbaGaemOyaiMaeiilaWIaemyAaKgabeaaaOqaaiabdkhaYnaaBaaaleaacqWGIbGyaeqaaaaakiabcYcaSaaa@4134@
where rb refers to the a priori probability to find a base b in the DNA sequence. Consequently, the weights wb,i represent log-likelihood ratios to find a base b at a position i. Finally, the score Sk around the position k of a test DNA sequence is a sum of the weights corresponding to bases observed in the DNA sequence at the subsequent positions starting from the position k. The sum Sk is computed for each position k of the matrix along the DNA sequence. High positive scores Sk indicate locations in the test DNA sequence with strong binding affinities whereas zero or negative scores are found elsewhere (Fig. 2).
Figure 2 Comparison of ATF and CREB matrices: Correlation C of ATF and CREB scores along a test DNA sequence. Left: first 30 scores for ATF (solid line) and CREB (dashed line). Right: scores for ATF versus scores for CREB. Only the first 200 scores are plotted, but the full length of the test DNA sequence is 10000 bases. Upper (shift = 0): the matrices are not properly aligned (C = 0.068). Lower (shift = 3): the matrices ATF and CREB are properly aligned and both reveal a binding site at position 20 (C = 0.881).
This widely used technique of score calculation leads immediately to the second similarity measure (similar in spirit to the method used in [18], but modified to take into account the sample sizes of compared matrices). For two given matrices f and g we can directly obtain the corresponding scores Skf
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGtbWudaqhaaWcbaGaem4AaSgabaGaemOzaygaaaaa@30BC@ and Skf
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGtbWudaqhaaWcbaGaem4AaSgabaGaemOzaygaaaaa@30BC@ along all positions k in a given test DNA sequence. If the weight matrices are highly similar we expect positive peaks at nearly the same positions, i.e. a prediction of nearly the same set of binding sites. In order to quantify the similarity of both matrices we calculate the Pearson correlation coefficient along a test sequence. Here we also consider all possible relative shifts between two PWMs (with a minimum overlap of 6 bases) and then take the maximum correlation coefficient as the similarity measure C of the two matrices. We have found, that the correlation coefficients do not depend strongly on the value of the pseudo-counts and reflect mainly the relevant rare peaks.
In this paper we take as the test DNA sequence a random sequence with equidistributed bases. For specific applications it might be appropriate to use other test sequences such as upstream regions of the genes of interest.
Sensitivity and specificity
Sensitivity and specificity of different methods for measuring similarities of profiles recognized by transcription factors were assessed as follows: since large sets of experimentally verified similar matrix pairs are not available, artificial sets were prepared. A representative initial matrix (either ATF or CREB) was resampled to construct a set of matrices. On average we probed the initial matrix 18 times (which corresponds to the median sample size of Transfac matrices). In order to study varying sample sizes for each generated matrix the number of samples was randomly chosen out of the range from 13 to 21. All the matrices generated this way should be classified as similar to each other. A set with matrices dissimilar to each other was prepared by random shuffling of the contents of the initial matrix. The nucleotide counts at each position were randomly reordered as well as the order of the positions. Additionally, we take into account different lengths of the matrices. Both sets were extended with random columns and the number of added columns was chosen randomly from zero to half of the length of the initial matrix. In the analysis, sensitivity was defined as the fraction of resampled matrices which were correctly identified as similar matrices. Specificity was defined as the fraction of random matrices which were identified as dissimilar. Six methods quantifying similarity of profiles were compared. The D (chi2th) and C (corr) functions were calculated as introduced above. Another score was defined as a sum of χ2 obtained for each compared columns (chi2sum). Three other methods (introduced in [15,17,20]) calculate a total sum over all compared columns of: Euclidian distance (ned), column-column correlation (ccc) and scalar product of columns (sp).
Results and discussion
In this paper two similarity measures of matrices are studied. The first quantifies for a given pair of matrices the number of significantly different columns D. The other represents the correlation C of binding sites scores along a DNA sequence for each of the given matrices.
Comparison of both similarity measures
For the Transfac library we analyze whether the pairs of matrices with small distances D and high correlation coefficients C coincide, i.e. for what matrices the two measures give consistent results. Fig. 3 shows histograms of correlation coefficients C for matrices with distances D = 0, 1, 2. It turns out that there are many pairs of matrices with D = 0 and large values of C (see the right peak in the upper panel of Fig. 3). For such matrices the differences between their columns are negligible and predicted binding sites are essentially identical.
Figure 3 Combinations of both measures: Histograms of the correlation C of the scores vectors obtained for different values of the distance D (number of significantly different columns according to the χ2 test). These data have been calculated for the Transfac matrices.
There are, however, also many pairs of matrices with D = 0 and relatively small correlation coefficients C (see the left peak in the upper panel of Fig. 3). These pairs refer mainly to matrices with a low information content and/or small sample size. In such cases the differences between columns are not statistically significant (many Ns in both consensus sequences) but their scores along a test DNA sequence correlate only weakly. For example, matrices V$STAT4_01 and V$MEF2_01 (see Transfac) are characterised by sample sizes N = 6, N = 5 respectively and have a distance D = 0 but a correlation C = 0.20.
There are also cases with a high correlation coefficient but with a distance D > 2. Such a situation appears for large matrices for which only a part is informative. For example matrices V$GR_01 and V$PR_01 (see Transfac) have a length of 27, but only six positions constitute the core sequence (TGTTCT). Among the others positions three are significantly different, leading to a distance D = 3 but these differences affect the correlation C only weakly (C = 0.92).
Several alternative measures have been proposed. We assessed the sensitivity and the specificity of these measures, as described in methods. The results of the comparison are presented in the supplemental Fig. 4. Both the our correlation measure and the column-to-column similarity give (for an appropriate threshold) a high specificity and sensitivity. However, in some cases, as illustrated above, adding a second criteria is useful to discard pairs involving large matrices for which only a part is informative. The D measure defined here can be used for this purpose. Both introduced measures quantify different properties and complement each other. Although alternative choices of measures might have been done, the advantage of using the correlation C is its implicit normalisation (the results do not depend much on the length and the sample size of the matrices) and the advantage of the distance D is its easy interpretation (number of different columns). Therefore, in the following, we focus on the most similar matrices based on the distance D and correlation C measures.
Clusters of similar matrices
Here we study the matrices of both Jaspar and Transfac databases. We consider pairs of matrices for which D ≤ 1 and C ≥ 0.8 as highly similar. These stringent thresholds were chosen to identify the most obvious similarities and they imply that the matrices are almost indistinguishable from a statistical point of view and that their scores along DNA sequences are strongly correlated. We verified that for all these pairs of matrices both similarity measures select the same relative shift of the corresponding matrices.
Fig. 4 shows an overview of all such matrices. Even though details of these clusters are only readable in the supplementary material (Fig. 1) the graph reveals interesting properties: The connecting lines visualizing high similarity join Jaspar matrices (ellipses) with Transfac matrices (boxes) in many cases. Consequently, our technique allows an automatic "alignment" of these collections of matrices. This is not a trivial task since the naming conventions used in the databases is different, and thus finding matrices corresponding to each other requires expert knowledge. We find that 84 matrices from Jaspar have counterparts in Transfac with D ≤ 1 and C ≥ 0.8. Another 16 matrices have somewhat smaller similarities D ≤ 3 and C ≥ 0.6. Only the Jaspar matrices P_HMG-1, P_HMG-IY and V_Ghlf, have no obvious "partners" in Transfac. A complete list of Transfac-Jaspar matrix pairs with high similarities is provided in the supplementary material (Tab. 1). Lists for other thresholds or other sets of matrices can be calculated through our web interface [29].
Figure 4 Graph showing similar matrices: Transfac matrices are indicated in white boxes, Jaspar matrices are indicated in gray ellipses. An edge is drawn between two matrices when D ≤ 1 and C ≥ 0.8. An enlarged version of this figure is available in the supplementary material (Fig. S1).
In addition to the edges between Transfac and Jaspar matrices there are many clusters containing multiple Transfac or Jaspar matrices. These clusters reflect pronounced similarities in the matrix collections. There are for example, matrices of the same transcription factor with different degrees of stringency (see for instance AP1 matrices). Moreover, different transcription factors of certain families have almost identical binding motifs (see for example Myc-Max, USF and ARNT). A complete list of all clusters is provided in the supplementary material (Tab. S2). An interesting collection of structural classes of transcription factors has been compiled recently by Sandelin and Wasserman [17]. Consistent with their results we find also clusters of the ETS family (see cluster 2 in Tab. S2, also enlarged in Fig. 5b), bHLH transcription factors (cluster 15), and REL family (cluster 5).
Figure 5 Clusters of similar matrices: Transcription factor families (a) GATA and (b) ETS.
In Fig. 5 we present enlargements of two selected clusters representing the GATA (panel a) and ETS (panel b) transcription factors family. The high similarity of these matrices cannot be directly noticed by inspection of names or consensus sequences. Furthermore, subgroups might be detected using our statistical approach. For example, the GATA cluster reveals that the Jaspar matrix has particularly high similarity to the Transfac entries GATA1_02, GATA3_01 and GATA6_01, but less similarities to other members of the GATA class. The clusters visualized in Fig. 4 and Fig. 5 can be exploited to reduce the number of matrices. Highly similar matrices match a DNA sequence either both or not at all. Therefore, one could construct "consensus matrices" as in [17] or one might select representative matrices in each cluster. In this way the number of overlapping predictions in the search for transcription factor binding sites can be decreased [17].
Mapping of novel matrices to databases
A careful inspection of the clusters found automatically by our similarity analysis might reveal unexpected similarities pointing to possible cross-talks of different signaling cascades on the level of transcriptional regulation. As an example we discuss the regulation of circadian clock genes and cell cycle control [30,31]. In both processes bHLH transcription factors bind as dimers to E-boxes. The corresponding Myc-Max cluster appeared already in Fig. 4 (the largest cluster). In the mammalian circadian clock the CLOCK-BMAL1 dimer regulates clock genes such as Per1, Per2, Per3, Cry1 and Cry2. We found no matrix in Transfac or Jaspar describing explicitly the binding sites of CLOCK-BMAL1. Consequently, we constructed such matrices ourselves in two different ways. On one hand we collected 9 experimentally verified binding sites from 7 different clock genes [32-36]. On the other hand, we took from a SELEX experiment 10 sequences with high affinities to the CLOCK-BMAL1 dimer [37].
Both matrices are visualized in Fig. 6a. Details of the matrix construction are given in the supplementary material (Tab. S3). Both matrices contain the E-box consensus motif CACGTG but differ in the flanking regions.
Figure 6 Mapping of CLOCK-BMAL1 matrices: (a) CLOCK-BMAL1 matrices based on experimentally characterised binding sites of clock genes and from a SELEX study (see Tab. S3 of the supplementary material for the list of these binding sites). (b) Mapping of CLOCK-BMAL1 matrices on E-box matrices. These matrices have been selected from the Transfac database and include MYC, MAX, ARNT, MYOD, USF, TAL1/E47 (see [35] for a review on E-box transcription factors). An edge is drawn when D ≤ 1 and C ≥ 0.8.
Fig. 6b shows that these novel matrices have highly similar counterparts in Transfac (NMYC, MYC, USF). Consequently, cross-talk of the circadian clock with cell cycle regulation and tumor genesis can be expected at the level of transcriptional control. Indeed, the success of chronotherapies and recent detailed studies on cross-talk underline the dependence of circadian rhythms with tumor growth [38]. Also in the process of liver regeneration a pronounced effect of the circadian clock on cell cycle control has been found [39]. This example illustrates that a careful SELEX experiment combined with a mapping of the resulting matrix to known matrices can reveal possible functions of the corresponding transcription factor.
Conclusion
Understanding gene regulation in higher eukaryotes is still challenging and current computational algorithms suffer from a large amount of false positive predictions [1,40]. In particular, mutually dependent position frequency matrices in databases such as Transfac or Jaspar lead to predictions of binding sites which overlap, what may be misinterpreted as a cluster of binding sites. Consequently, a careful pre-selection of matrices is essential. On one hand, expert knowledge can be used to select a subset of candidate matrices for the analysis of upstream regions. Such a selection is, however, subjective and novel combinations of transcription factor binding sites might be missed. On the other hand, for large scale computational studies, it is useful to have an automatic tool to detect similar matrices. Therefore, we introduce in this paper a method combining two independent similarity measures to compare position frequency matrices. This approach can be used to quantify similar matrices, to map the entries of different databases, and to cluster matrices.
The first similarity measure used in our approach is based on a χ2 test. In contrast to earlier approaches based on normalized frequencies [15,17,20] we take into account the small sample size of many matrices. We count the number of significantly different matrix columns which defines the distance D. In this paper we focus on highly similar matrices with D ≤ 1. In forthcoming studies the χ2 measure might be taken directly to calculate distances of matrices in more detail.
The second measure is related to the primary application of position weight matrices – the prediction of binding sites in uncharacterized DNA sequences. We calculate for two matrices of interest the scores along a test DNA sequence and derive the Pearson correlation coefficient C of these vectors. Thus large values of C indicate that both matrices predict essentially the same binding sites. In this paper we take a 10000 bp long random sequence with equiprobable and independent bases as the test DNA sequence. However, the measure can be easily adapted also to other test sequences such as sets of promoter regions.
Our combined similarity measure was first used to map the Jaspar matrices to the Transfac database automatically. Then, requiring rather strong similarity (D ≤ 1, C ≥ 0.8) we identified similar matrices present in these databases and constructed clusters of almost indistinguishable matrices. By choosing only one representative matrix for each cluster it is possible to construct smaller sets of matrices as input of binding site prediction algorithms. Consequently, this approach decreases the number of overlapping binding site predictions. Moreover, such a reduced set constitutes a better input for methods predicting close occurrences of different binding sites (e.g. [16]). In order to eliminate false signals further, approaches such as phylogenetic footprinting [1,12,13], transcriptional profiling [14], ChIP on chip experiments [41,42] or modeling cis-regulatory modules need to be combined with a preselection of independent matrices. Our combined technique can be used to predict cross-talk on the level of transcriptional control. As an illustration we discuss the cluster of E-box binding bHLH transcription factors. Since circadian clock genes are regulated by a binding site quite similar to the Myc-Max motif, a strong interdependence of circadian regulation and cell cycle control is expected and is indeed known empirically for decades in connection with chronotherapies or liver regeneration.
Finally we use the similarity measures to assign newly derived matrices to known factors. To illustrate this application, we map an E-box matrix obtained from SELEX experiments with the CLOCK-BMAL1 dimer to the Myc-Max cluster. Thus the possible function of poorly characterized transcription factors can be predicted using affinity measurements combined with a comparison of the resulting matrix to database matrices.
Availability
The method is available through a web interface at .
Authors' contributions
SK, DG and HH designed the study. SK and DG were involved in programming and SK set up the web interface. SK, DG and HH interpreted the results and drafted the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Correspondence between Jaspar and Transfac matrices: For each Jaspar matrix similar (D ≤ 1 and C ≥ 0.8) Transfac matrices are listed. 84 Jaspar matrices have at least one corresponding Transfac matrix.
Click here for file
Additional File 2
Clusters of similar (D ≤ 1 and C ≥ 0.8) Jaspar and Transfac matrices.
Click here for file
Additional File 3
Binding sites for Clock-Bmal1: Experimentally characterized binding sites for Clock-Bmal1 in clock genes and in selected sequences (SELEX experiment).
Click here for file
Additional File 4
Comparison of different measures: specificity and sensitivity are determined as described in the "Methods" section of the paper for various thresholds of the different similarity measures. Specificity is defined as the fraction of the number of resampled matrices (TP on y-axis) found as similar. Sensitivity is defined as the fraction of the number of randomized matrices (TP on x-axis) found as dissimilar. Curves: "corr": correlation of scores along a DNA sequence, i.e. our score C (thresholds = 0.99, 0.95, 0.9, 0.8, 0.7...); "chi2th": our chi2 measure D (thresholds = 0, 1..8); "chi2sum": sum of column chi2 distances; "ned": normalized euclidian distance; "ccc": column-column correlation; "sp": column scalar product.
Click here for file
Acknowledgements
The authors thank N. Blüthgen, M. Swat and M. Futschik for discussions and critical reading of the manuscript. SzMK is supported by the German Federal Ministry of Education and Research (BMBF) and the German Research Foundation (DFG). DG is Chargé de Recherches du Fonds National Belge de la Recherche Scientifique.
==== Refs
Wasserman W Sandelin A Applied bioinformatics for the identification of regulatory elements Nat Rev Genet 2004 5 276 87 15131651
Stormo G DNA binding sites: representation and discovery Bioinformatics 2000 16 16 23 10812473
Matys V Fricke E Geffers R Gossling E Haubrock M Hehl R Hornischer K Karas D Kel A Kel-Margoulis O Kloos D Land S Lewicki-Potapov B Michael H Munch R Reuter I Rotert S Saxel H Scheer M Thiele S Wingender E TRANSFAC: transcriptional regulation, from patterns to profiles Nucleic Acids Res 2003 31 374 8 12520026
Sandelin A Alkema W Engstrom P Wasserman W Lenhard B JASPAR: an open-access database for eukaryotic transcription factor binding profiles Nucleic Acids Res 2004 32 D91 4 14681366
Quandt K Frech K Karas H Wingender E Werner T MatInd and MatInspector: new fast and versatile tools for detection of consensus matches in nucleotide sequence data Nucleic Acids Res 1995 23 4878 84 8532532
Kel A Gossling E Reuter I Cheremushkin E Kel-Margoulis O Wingender E MATCH: A tool for searching transcription factor binding sites in DNA sequences Nucleic Acids Res 2003 31 3576 9 12824369
Frith M Fu Y Yu L Chen J Hansen U Weng Z Detection of functional DNA motifs via statistical over-representation Nucleic Acids Res 2004 32 1372 81 14988425
Werner T Fessele S Maier H Nelson P Computer modeling of promoter organization as a tool to study transcriptional coregulation FASEB J 2003 17 1228 37 12832287
Ohler U Stemmer G Harbeck S Niemann H Stochastic segment models of eukaryotic promoter regions Pac Symp Biocomput 2000 380 91 10902186
Davuluri R Grosse I Zhang M Computational identification of promoters and first exons in the human genome Nat Genet 2001 29 412 7 11726928
Wasserman W Palumbo M Thompson W Fickett J Lawrence C Human-mouse genome comparisons to locate regulatory sites Nat Genet 2000 26 225 8 11017083
Dieterich C Cusack B Wang H Rateitschak K Krause A Vingron M Annotating regulatory DNA based on man-mouse genomic comparison Bioinformatics 2002 18 S84 90 12385988
Ureta-Vidal A Ettwiller L Birney E Comparative genomics: genome-wide analysis in metazoan eukaryotes Nat Rev Genet 2003 4 251 62 12671656
Bussemaker H Li H Siggia E Regulatory element detection using correlation with expression Nat Genet 2001 27 167 71 11175784
Hughes J Estep P Tavazoie S Church G Computational identification of cis-regulatory elements associated with groups of functionally related genes in Saccharomyces cerevisiae J Mol Biol 2000 296 1205 14 10698627
Frith M Li M Weng Z Cluster-Buster: Finding dense clusters of motifs in DNA sequences Nucleic Acids Res 2003 31 3666 8 12824389
Sandelin A Wasserman W Constrained binding site diversity within families of transcription factors enhances pattern discovery bioinformatics J Mol Biol 2004 338 207 15 15066426
Haverty P Frith M Weng Z CARRIE web service: automated transcriptional regulatory network inference and interactive analysis Nucleic Acids Res 2004 32 W213 6 15215383
Hannenhalli S Levy S Predicting transcription factor synergism Nucleic Acids Res 2002 30 4278 84 12364607
Pietrokovski S Searching databases of conserved sequence regions by aligning protein multiple-alignments Nucleic Acids Res 1996 24 3836 45 8871566
Schones DE Sumazin P Zhang MQ Similarity of position frequency matrices for transcription factor binding sites Bioinformatics 2005 21 307 313 15319260
Schneider T Stormo G Gold L Ehrenfeucht A Information content of binding sites on nucleotide sequences J Mol Biol 1986 188 415 31 3525846
King O Roth F A non-parametric model for transcription factor binding sites Nucleic Acids Res 2003 31 e116 14500844
Zhou Q Liu J Modeling within-motif dependence for transcription factor binding site predictions Bioinformatics 2004 20 909 16 14751969
Barash Y Elidan G Friedman N Kaplan T Modeling dependencies in protein-DNA binding sites Proceedings of the seventh annual international conference on Computational molecular biology 2003 ACM Press New York, NY, USA 28 37
Benos P Bulyk M Stormo G Additivity in protein-DNA interactions: how good an approximation is it? Nucleic Acids Res 2002 30 4442 51 12384591
Hertz G Stormo G Identifying DNA and protein patterns with statistically significant alignments of multiple sequences Bioinformatics 1999 15 563 77 10487864
Fickett J Quantitative discrimination of MEF2 sites Mol Cell Biol 1996 16 437 41 8524326
wmCompare
Cardone L Sassone-Corsi P Timing the cell cycle Nat Cell Biol 2003 5 859 61 14523398
Matsuo T Yamaguchi S Mitsui S Emi A Shimoda F Okamura H Control mechanism of the circadian clock for timing of cell division in vivo Science 2003 302 255 9 12934012
Darlington T Wager-Smith K Ceriani M Staknis D Gekakis N Steeves T Weitz C Takahashi J Kay S Closing the circadian loop: CLOCK-induced transcription of its own inhibitors per and tim Science 1998 280 1599 603 9616122
Gekakis N Staknis D Nguyen H Davis F Wilsbacher L King D Takahashi J Weitz C Role of the CLOCK protein in the mammalian circadian mechanism Science 1998 280 1564 9 9616112
Chen W Baler R The rat arylalkylamine N-acetyltransferase E-box: differential use in a master vs. a slave oscillator Brain Res Mol Brain Res 2000 81 43 50 11000477
Munoz E Brewer M Baler R Circadian Transcription. Thinking outside the E-Box J Biol Chem 2002 277 36009 17 12130638
Vallone D Gondi S Whitmore D Foulkes N E-box function in a period gene repressed by light Proc Natl Acad Sci U S A 2004 101 4106 11 15024110
Hogenesch J Gu Y Jain S Bradfield C The basic-helix-loop-helix-PAS orphan MOP3 forms transcriptionally active complexes with circadian and hypoxia factors Proc Natl Acad Sci U S A 1998 95 5474 9 9576906
Filipski E King V Li X Granda T Mormont M Liu X Claustrat B Hastings M Levi F Host circadian clock as a control point in tumor progression J Natl Cancer Inst 2002 94 690 7 11983758
Schibler U Circadian rhythms. Liver regeneration clocks on Science 2003 302 234 5 14551421
Kielbasa S Blüuthgen N Sers C Schäfer R Herzel H Prediction of Cis-Regulatory Elements of Coregulated Genes Genome Informatics 2004 15 117 124 15712115
Ren B Robert F Wyrick J Aparicio O Jennings E Simon I Zeitlinger J Schreiber J Hannett N Kanin E Volkert T Wilson C Bell S Young R Genome-wide location and function of DNA binding proteins Science 2000 290 2306 9 11125145
Martone R Euskirchen G Bertone P Hartman S Royce T Luscombe N Rinn J Nelson F Miller P Gerstein M Weissman S Snyder M Distribution of NF-kappaB-binding sites across human chromosome 22 Proc Natl Acad Sci U S A 2003 100 12247 52 14527995
|
16191190
|
PMC1261160
|
CC BY
|
2021-01-04 16:27:26
|
no
|
BMC Bioinformatics. 2005 Sep 28; 6:237
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-237
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2391619119510.1186/1471-2105-6-239Methodology ArticleOptimized between-group classification: a new jackknife-based gene selection procedure for genome-wide expression data Baty Florent [email protected] Michel P [email protected]ère Guy [email protected] Aedín C [email protected] Martin H [email protected] Pulmonary Gene Research, University Hospital Basel, CH-4031 Basel, Switzerland2 Laboratoire de Biométrie et de Biologie Évolutive, UMR CNRS 5558, Université Claude Bernard Lyon 1, 43 blvd du 11 Novembre, 1918, 69622 Villeurbanne Cedex, France3 Bioinformatics Conway Institute, University College Dublin, Ireland2005 28 9 2005 6 239 239 10 6 2005 28 9 2005 Copyright © 2005 Baty et al; licensee BioMed Central Ltd.2005Baty 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
A recent publication described a supervised classification method for microarray data: Between Group Analysis (BGA). This method which is based on performing multivariate ordination of groups proved to be very efficient for both classification of samples into pre-defined groups and disease class prediction of new unknown samples. Classification and prediction with BGA are classically performed using the whole set of genes and no variable selection is required. We hypothesize that an optimized selection of highly discriminating genes might improve the prediction power of BGA.
Results
We propose an optimized between-group classification (OBC) which uses a jackknife-based gene selection procedure. OBC emphasizes classification accuracy rather than feature selection. OBC is a backward optimization procedure that maximizes the percentage of between group inertia by removing the least influential genes one by one from the analysis. This selects a subset of highly discriminative genes which optimize disease class prediction. We apply OBC to four datasets and compared it to other classification methods.
Conclusion
OBC considerably improved the classification and predictive accuracy of BGA, when assessed using independent data sets and leave-one-out cross-validation.
Availability
The R code is freely available [see Additional file 1] as well as supplementary information [see Additional file 2].
==== Body
Background
Gene expression microarrays enable the simultaneous measurement of the expression levels of thousands of genes. Supervised classification of gene expression data aims to identify combinations of genes which give the best discrimination of groups of samples specified in advance. For such methods, which are classically used in disease class prediction, the identification of a subset of discriminating genes can be critical [1,2]. Indeed, a large proportion of genes are generally non-informative in terms of disease class prediction. A gain in classification and prediction performance can be expected when predictors are built upon a subset of highly discriminating genes [3,4].
Several algorithms capable of selecting a subset of predictive genes were recently proposed [5]. These methods include a genetic algorithm [6], maximum difference subset algorithm (MDSS) [7], support vector machines [8,9], a shrunken centroids technique [2,10] and several which use of discriminant functions [11].
However, two issues remain: 1) different subsets of genes may provide comparable optimal discriminations [1]; 2) it is generally difficult to determine the optimal number of genes for discrimination [12,13]. This number may vary according to the number of individuals in the training set, the number of groups to discriminate and the method used for classification and prediction. Dolédec and Chessel [14] developed a supervised classification approach, Between Group Analysis (BGA), which was recently applied to microarray data [15]. The authors specified several key features of BGA that make it a method of choice for sample classification and class prediction. In BGA, all genes participate in the discrimination. Consequently, no gene selection step is required. On the other hand, BGA calculates group means and is therefore sensitive to outliers. Our objective was to improve the robustness of BGA by optimizing the number of discriminating genes supporting the analysis.
In this study, we propose a new jackknife-based algorithm – optimized between-group classification (OBC) – that produces a selection of the most robust discriminating genes in order to improve the accuracy of disease class prediction. The criterion optimized in OBC is the percentage of between group inertia (% BG inertia). OBC is applied to BGA but it could also be associated with other supervised methods. We tested the efficiency of OBC on four datasets using independent test sets and leave-one-out cross-validation (LOOCV). We compared our approach to different classification methods.
Results
Outline of the OBC algorithm
OBC can be described in three steps (Figure 1). These steps are detailed below. Each dataset used in this study was systematically split into a training set and a test set. OBC was applied exclusively to the training set.
Figure 1 Overall description of OBC. Three steps are required to perform OBC optimization. In the first pre-selection step, n most discriminating genes are selected by performing a BGA on the training set with the whole set of genes. In the second step, a jackknife optimization is performed on the initial subset of genes and the least influential genes in terms of % BG inertia are removed successively. This second step is iteratively computed, decrementing the genes down to 5. Finally in the third step, the optimal subset of genes is identified (subset with the best classification accuracy and the best stability).
Pre-selection of discriminating genes – step 1
In the first step, pre-selection of a few hundred most discriminating genes is made. This is to reduce the number of calculations and computational resources in step 2 (below). This initial set of discriminating genes is obtained from a BGA of the whole training set (including all genes). Genes with the highest scores on BGA discriminating axes, those located at extremities of BGA axes, are collected. For datasets where samples are grouped into 2 categories (binary categorization), we selected an equal amount of genes at each end of the single discriminating BGA axis. For datasets with more than two categories, we chose genes projected at the periphery of each pair of discriminating axes using a "peeling" function (successive 2D convex hulls).
Jackknife optimization – step 2
This second step of the algorithm is cpu and time consuming. Due to computational limitations, the number of pre-selected genes should be in the order of a few hundreds (the optimization of 150 genes and 24 samples required 1 h 50 min on a Pentium 4 2.66 GHz computer). Strategies to reduce calculation time are discussed below.
Classification accuracy by LOOCV
The performance of the subsets of predictive genes was assessed using LOOCV. To perform LOOCV, a sample is removed from the dataset and a BGA is performed on the remaining samples. The excluded sample is projected on to the BGA and classified. This is iteratively performed until all samples have been subjected to cross-validation. The percentage of samples correctly classified by cross-validation is calculated. This parameter measures the prediction accuracy of the subset of genes.
Optimization criterion
The objective of OBC is to improve the discrimination efficiency of BGA, by excluding genes which contribute least to the % BG inertia of samples. OBC uses a jackknife iteration to maximize the between group inertia while minimizing the within group inertia. The inertia decomposition can be described as follow.
Let us suppose N the number of samples (xi is the ith sample and wi its weight), dist(xi, xj) the squared Euclidean distance between two samples xi and xj, K the number of groups (Gk is the kth group) and Nk the number of individuals in the kth group. Potentially distances other than Euclidean could be used. In a preliminary analysis we found that the Euclidean distance performs similarly to Manhattan distance. Therefore, given its relative ease of implementation, we use Euclidean distance throughout our analyses. By using a weighted pair-group average calculation, the total inertia can be decomposed into within group inertia (Eq. 1) and between group inertia (Eq. 2). The algorithm aims to maximize the percentage of between group inertia, i.e. the ratio of the between group inertia to the total inertia (Eq. 3).
% BG inertia = BG inertia/(WG inertia + BG inertia) (3)
Measurement of the contribution of each individual gene using jackknifing
We assessed sequentially the influence of each gene in the remaining gene subset using a jackknife procedure. In jackknife analysis, we remove a gene, perform a BGA on the dataset and calculate the % BG inertia. If we remove a gene which positively contributes to the between group discrimination, the % BG inertia decreases and vice versa. By comparing the % BG inertia before and after removing a given gene, one can assess the influence of this gene. In addition, we assess the stability of the % BG inertia during jackknife (described later).
Jackknife approaches have been previously used in the context of gene selection [7,16]. As an example, Lyons-Weiler et al. (2003) [7] used jackknifing to reduce the false positive rate of a gene set. In the present study, we used jackknifing to progressively eliminate the least discriminative genes from a subset of genes.
Backward optimization
At each step of the algorithm, the gene that contributes least to the % BG inertia is removed from the dataset. Another jackknife procedure is then performed with the remaining genes. This backward optimization algorithm reduces the number of genes from a large subset (typically a few hundreds of genes) to a minimal subset (fixed to minimum of 5).
Stability and robustness of the optimization – variance of % BG inertia and Monte-Carlo permutation test
The variance of % BG inertia was used as a measure of the stability of the optimization. By jackknifing a subset of n genes, we obtain n values of % BG inertia. The range of variation of these values is the variance of % BG inertia. During backward optimization the number of genes included in the classifier gets smaller, and the effect of the jackknife perturbation measured by the variance of % BG inertia tends to increase. If this variance is high, the robustness and the stability of the prediction model is low. Consequently, low variance of % BG inertia is preferable.
Throughout the optimization, the statistical significance of BGA is evaluated with a Monte-Carlo permutation test.
Identification of the optimal subset of genes for disease class prediction – step 3
The optimal subset of genes are identified with the aid of the summary diagram which summarizes the results of the algorithm. The optimal subset of genes should have both high LOOCV prediction accuracy and stability (i.e. minimal variance of % BG inertia). If optimization of these two parameters resulted in a range of near optimal solutions, we chose subsets with fewer genes and higher % BG inertia. Importantly, although we calculate prediction accuracy of the independent test set, these results were never taken into account in OBC, as this would result in over-training.
Application of OBC to sarcoidosis data
Between group analysis
Standard BGA was applied to the whole sarcoidosis training data set. The biplot representation shows that BGA separated the three phenotypes with no overlap (Figure 2, panel A). The first axis separated the healthy controls from the sarcoidosis patients. The second axis separates the two stages of sarcoidosis. The efficiency of classification of new samples was measured using LOOCV. Seventy-five percent of the 24 samples were classified correctly. However, we observed discrepancies in classification accuracy between the three phenotypes. All healthy controls, 6 out of 7 stage I, but none of the stage II/III sarcoidosis patients were correctly re-classified. When we tried to predict the classification of a blind test of 8 follow-up patients, using this BGA of the whole set of genes, only 50% of these test samples were correctly classified. Four out of five patients, which recovered 6 months after they were diagnosed with a stage I sarcoidosis, were classified in the healthy group. All of the patients still suffering from active sarcoidosis stage II/III (n = 3) were incorrectly classified.
Figure 2 Optimized between-group classification applied to sarcoidosis data. In panel A, 24 individuals (solid circles) in the training set (H: healthy controls, SI: sarcoidosis stage I, SII: sarcoidosis stage II/III) and 8 individuals (empty circles) in the test set (283, 286, 287, 289 and 290 as H; 282, 284 and 285 as SII) are classified by a standard BGA using the whole set of genes. Panel B shows the different parameters of OBC as a function of the number of genes used in the analysis: the percentage of between group inertia (solid line), the percentage of good cross-validation (dashed line) and the variance of between group inertia (dot-dashed line). For indication, the percentage of test samples correctly predicted is represented by a dotted line. This parameter was not used in optimization of the training model. The vertical line shows the optimal number of genes. In panel C, the 105 most discriminating genes (initial subset) are located at the periphery of the biplot (black crosses) and the 58 optimal genes are highlighted (circled crosses). In panel D, 8 test-samples are classified using a BGA based on the 58 optimal genes.
Optimized between group classification
We selected the 105 most discriminating genes in this initial BGA, using the above mentioned peeling procedure (Figure 2, panel C). OBC was applied on this subset of genes. The least influential genes in terms of % BG inertia were removed one by one.
Figure 2 (panel B) shows the evolution of classification parameters; % BG inertia, % correct classification in LOOCV, and variance of % BG inertia. During the optimization process, the % BG inertia increased when the number of genes decreased until it reached an optimum, then it decreased when the number of genes fell below this optimum threshold. The percentage of correct classification in LOOCV was stable in a range of 20–70 genes. When the number of genes further decreased, it started to oscillate. The variance of % BG inertia was very low for subsets of more than 58 genes. This parameter increased considerably for subsets fewer than 57 genes. Finally, the dotted line represents the evolution of percentage of test sets correctly classified (this parameter was not considered during optimization).
The subset of genes with the best cross-validation efficiency and least variable % BG inertia was judged to be the optimal subset. Therefore, this was a subset of 58 genes (Figure 2, panel C). The accuracy of LOOCV obtained using this optimized subset of genes was clearly improved since 96% of samples were correctly classified (100, 80 and 100% respectively in sarcoidosis stage I, stage II/III and healthy controls). Figure 2 (panel D) shows the projection of 8 follow-up samples predicted by this subset of classifiers. These predictions were also improved since 2/3 of sarcoidosis stage II/III were correctly associated to their group, whereas 4/5 of patients in remission from a stage I sarcoidosis were classified as healthy. Patient 283, who was mis-classified, clinically recovered from a sarcoidosis stage I. It is possible that signals of gene activity specific to stage I sarcoidosis be still detectable in this patient.
Application of OBC to tumour data
Between group analysis
BGA was applied to the whole tumour training set [20]. BGA clearly separated the 4 different types of tumours with no overlap (Figure 3, panel A). Based upon the complete set of 2308 genes, the LOOCV showed that 93% of the 63 samples from the training set were correctly cross-validated and 19/20 of the test sets were correctly predicted. The most discriminating genes associated with the different groups were identified at the periphery of the BGA biplot (Figure 3, panel C).
Figure 3 Optimized between-group classification applied to tumour data. In panel A, 63 samples (solid circles) of the training set (BL: Burkitt's lymphoma, EWS: Ewing's sarcoma, NB: neuroblastoma, RMS: rhabdomyo sarcoma) and 25 samples (empty circles) of the test set (7, 15 and 18 as BL-NHL; 2, 6, 12, 19, 20 and 21 as EWS; 1, 8, 14, 16, 23 and 25 as NB; 4, 10, 17, 22 and 24 as RMS; 3, 5, 9, 11 and 13 as control samples that do not belong to one of the 4 groups) are classified by the standard BGA based on the whole set of genes. Panel B shows the different parameters of OBC as a function of the number of genes used in the analysis: the percentage of between group inertia (solid line), the percentage of good cross-validation (dashed line) and the variance of between group inertia (dot-dashed line). For indication, the percentage of test samples correctly predicted is represented by a dotted line. This parameter was not used in optimization of the training model. The vertical line shows the optimal number of genes. In panel C, the 245 most discriminating genes are represented with small crosses and the 90 optimal genes are highlighted (circled crosses). In panel D, the 25 test-samples are classified using a BGA based on the 90 optimal genes.
Optimized between-group classification
From the initial BGA, the 245 most discriminating genes were selected. We applied the optimization algorithm to this initial subset. We used the optimization diagram to determine the optimal subset of genes. As shown in diagram Figure 3, panel B, there was a range of near optimal solutions (high % of correct cross-validation and low variance of % BG inertia). We decided to choose an optimal subset of 90 genes for which the accuracy, the stability and the % BG inertia were high.
The results of BGA using the 90 optimal genes are plotted in Figure 3 (panel C). The accuracy of LOOCV, of BGA on the 63 training samples using the 90 optimal genes, increased to 100%. All 20/20 test sets were correctly classified (Figure 3 panel D).
Stability of OBC and test of significance
The stability of OBC was controlled by monitoring the evolution of variance of % BG inertia. This parameter was of great importance as it monitored whether the classification was overly influenced by a few genes.
The Monte-Carlo permutation test was constantly significant for the different datasets (estimated p-value = 0.001). This result suggests that our method is robust.
Sensitivity and specificity
We built confusion matrices from the results obtained from LOOCV and classification of independent test sets. Then, we calculated the sensitivity and specificity of BGA and OBC for each disease category. Sensitivity measures the proportion of individuals correctly classified for a given disease class (true positives). Specificity measures the proportion of individuals that do not belong to the class and which are not classified in this class (true negatives). The sensitivities and specificities of OBC vs. standard BGA are summarized in Tables 1 and 2.
Table 1 Sensitivity and specificity of OBC compared with standard BGA in sarcoidosis dataset.
LOOCV Test set
H SI SII H SI SII
OBC Sensitivity 1 1 0.8 0.8 - 0.67
Specificity 0.92 1 1 1 0.5 1
BGA Sensitivity 1 0.86 0 0.8 - 0
Specificity 0.83 0.82 0.95 1 0.75 1
Table 2 Sensitivity and specificity of OBC compared with standard BGA in tumour dataset.
LOOCV Test set
EWS NB RMS BL EWS NB RMS BL
OBC Sensitivity 1 1 1 1 1 1 1 1
Specificity 1 1 1 1 1 1 1 1
BGA Sensitivity 0.87 1 0.9 1 1 0.83 1 1
Specificity 0.975 0.98 0.93 1 1 1 0.93 1
Figure 4 shows the sensitivity as a function of 1 – specificity of BGA with and without optimization (black dots and white dots, respectively), when applied to the sarcoidosis and tumour datasets (panels A and B, respectively). The results of LOOCV and classification of independent test sets are shown in the left and right plots respectively.
Figure 4 Analysis of sensitivity and specificity. The sensitivity and specificity of OBC (solid circles) were compared to standard BGA (empty circles). The prediction accuracy of OBC when applied to the sarcoidosis dataset was assessed using (A) LOOCV (left panel) and classification of the independent dataset (right panel). OBC was also applied to the tumour dataset and tested using (B) LOOCV (left panel) and classification of the independent dataset (right panel). Arrows show the improvement of sensitivity and specificity obtained with OBC compared to the standard BGA.
We observed an improvement in prediction sensitivity and specificity of both the sarcoidosis (Figure 4A) and tumour datasets (Figure 4B) when OBC was applied in LOOCV and independent test sample cross validation.
Comparison with other algorithms
We compared OBC with three other recently described gene selection methods: the GA/KNN algorithm [6], maximal margin linear programming (MAMA) [17] and nearest shrunken centroid [10]. Results (Table 3) show that OBC outperforms these approaches in terms of accuracy of LOOCV and classification of independent test sets. Comparisons between BGA and other supervised classification methods [15] report that BGA outperforms or performs with similar effectiveness.
Table 3 Comparison of the accuracy of OBC with different classification methods.
Dataset Method % correct LOOCV % correct prediction of independent test samples
Sarcoidosis OBC 96% 75%
BGA 75 50
GA/KNN 92 62.5
MAMA 67 62.5
Shrunken centroids 791 62.51
Tumour OBC 100 100
BGA 92.6 95
GA/KNN 100 95
MAMA 98 76
Shrunken centroids 1002 902
1Threshold of 2.801
2Threshold of 2.459
OBC applied to datasets with binary categorization
Colon cancer dataset
We assessed the prediction accuracy of OBC when applied to the colon cancer data set, which contains two categories of tumor samples. We applied OBC optimization to the 100 most discriminating genes. Results of LOOCV, show an increase of accuracy from 85% for standard BGA to 94% for OBC (based on 20 optimized genes). We investigated the sensitivity and specificity of OBC classification prediction when applied to independent test data. We built 26 pairs of training sets/test sets by randomly splitting the complete data set of 62 samples into training sets of 40 samples and test sets of 22 samples. OBC produced an improvement in both the sensitivity (83% to 87%) and specificity (87% to 91%) of prediction.
Leukemia dataset
We compared the prediction accuracy of BGA and OBC using LOOCV of the whole dataset. The percentage of samples correctly predicted in LOOCV was 90% for BGA and 99% for OBC (based on 40 optimized genes). Similarly to the colon cancer data analysis, we built 24 pairs of training sets/test sets by randomly splitting the whole dataset into 50 training and 22 test samples. Application of OBC to the leukemia dataset improved the sensitivity and specificity of test set classification. When OBC was applied, the sensitivity of classification was improved for both ALL (97% to 99%) and AML (91% to 92%). The specificity of prediction of both ALL and AML samples was also improved (respectively, 91% to 92% and 97% to 99%).
Discussion
Selection of genes that optimize disease class prediction is a significant and difficult challenge in microarray data analysis. Most discriminative functions require more cases than variables which is not realistic in the context of microarray experiments. A further challenge is the considerable amount of noise in microarray data. BGA can be applied to complete datasets without prior gene selection and performs comparably or outperforms several other approaches [15]. We showed that an optimized gene selection considerably improves the predictive power of BGA. Our jackknife-based algorithm tests the robustness of BGA discriminating genes and progressively excludes weaker discriminators. As a consequence, it optimizes the performance of BGA and reduces the number of discriminating genes.
The OBC algorithm presented here might be time consuming depending on the size of the initial subset of genes. Increasing the number of genes in the initial dataset ensures that more potentially discriminative genes are present in the analysis. However, the time required for the optimization process increases significantly. We assessed the percentage of gain in % BG inertia obtained by increasing the number of genes in the initial subset. This number depends on the dataset and in particular the number of groups to discriminate. The optimal number of genes of OBC starting genes is around 100 for the sarcoidosis data and 150–200 in the tumour data (Figure 5).
Figure 5 Number of genes included in the initial subset. This plot shows the maximum % BG inertia reached by the optimization procedure as a function of the number of genes present in the initial subset of genes (top curve: tumour data; bottom curve: sarcoidosis data). The dashed lines delimit the optimal size of the initial subset of genes for both datasets (above which the gain in % BG inertia is lower).
In OBC, the choice of the initial subset of genes from which the algorithm starts remains critical and alternative procedures might be used. For example, the genetic algorithm proposed by Li et al. (2001) [6] could be associated to OBC and might provide some improvements in performance.
Different options could be considered to speed up the algorithm. We considered removing more than one least influential gene at a time in the jackknife optimization. The execution time would decrease proportionally to the number of genes removed at each step. For example, if we removed 10 % least influential genes from the subset of genes at each step, we could greatly increase the speed of execution of the algorithm. With this, it would be possible to include a few thousands of genes in the initial subset of genes. The decision on how many genes to remove per jackknife cycle is a trade off between testing more combinations of genes (and therefore testing more efficiently gene-gene interactions) and including more genes in the analysis. On the other hand, the numerous tasks performed during the optimization could be split into several jobs, which could be potentially computed in parallel by a cluster of processors/computers. Finally another solution would be to rewrite the computationally demanding parts of the algorithm in a more efficient computer language like C.
We decided to choose a backward optimization procedure as this seemed to be more adapted for taking possible gene-gene interactions into account. The prediction power of a single gene might be negligible in itself while it might be preponderant when associated with one or a few other genes. Removing a gene that jointly participates with other genes to the group discrimination will have an impact, which is measurable by a backward approach, whereas no evidence might be found by using a forward optimization.
Our results show that an improvement in discriminative and predictive power of BGA can be achieved by reducing the number of predictors in the analysis to a small subset of highly discriminative genes. These genes contribute to improve the % BG inertia. In this study, two criteria were used to define the optimal subset of genes: a positive criterion, the percentage of correct classification by LOOCV and a negative criterion the variance of % BG inertia. When searching for the subset of genes where both criteria were optimized, we generally found a range of near optimal solutions. In the sarcoidosis dataset, the size of the optimal subset of genes was around 60, whereas in the tumour dataset, subsets including around 90 genes were found to be optimal. By using a method that associates a genetic algorithm with the k-nearest neighbors technique (GA/KNN) on a lymphoma dataset, Li et al. (2001) [6] concluded that using only a few discriminating genes may not be reliable, whereas using too many genes will add noise to the classification. They suggested 50–200 genes would give an optimal result which is in agreement with our study.
Conclusion
We propose OBC, a novel jackknife-based backward optimization algorithm, which improves both the classification and predictive power of BGA. Our algorithm tended to outperform alternative classification techniques. In the future, OBC could be used as a decision making-tool for disease class prediction based on gene expression data in various clinical situation. Future developments will include the application of the algorithm to different supervised methods.
Methods
Data sets
Sarcoidosis data
The gene expression study was carried out on 12 healthy controls (H), 7 sarcoidosis stage I patients (SI) and 5 sarcoidosis stage II/III patients (SII). This dataset was published previously and details can be found in [18]. These 24 samples correspond to the sarcoidosis training set. In addition, 6 months later, 8 follow-up chips were done for some of the sarcoidosis patients. Among these patients, 3 still had active sarcoidosis stage II/III and 5 were recovered from sarcoidosis stage I. These 8 supplementary samples correspond to the sarcoidosis test set. The expression level of 12626 probe sets was measured with Affymetrix' GeneChip® (HG-U95Av2). The complete dataset and the raw files have been deposited in NCBIs Gene Expression Omnibus (GEO) [19], and are accessible through GEO Series accession number GSE1907.
Tumour data
This dataset was published by Khan et al. (2001) [20]. The authors measured the expression of 6567 genes in four types of small round blue cell tumours (NB: neuroblastoma; RMS: rhabdomyo sarcoma; BL: Burkitt's lymphoma; EWS: Ewing's sarcoma). A filtered dataset containing the expression level of 2308 genes is publicly accessible [21]. The whole dataset contained 88 samples split into a training set (63 samples) and a test set (25 samples).
Colon cancer data
This colon cancer dataset was studied by Alon et al. (1999) [22]. It contained 62 samples obtained from 40 tumor samples and 22 control samples. Gene expression profiles were analyzed using Affymetrix' microarrays containing more than 6500 genes. This dataset was randomly split into training sets and test sets (40 and 22 samples, respectively). This dataset is available as a Bioconductor data package [23].
Leukemia data
The leukemia dataset [24] contained 72 samples from patients having two types of acute leukemia. Among the 72 patients, 47 had acute lymphoblastic leukemia (ALL) and 25 had acute myeloid leukemia (AML). Samples were obtained from bone marrow or peripheral blood. Gene expression profiling was analyzed with Affymetrix' microarrays containing 7159 probe sets. This dataset was randomly split into training sets and test sets (50 and 22 samples, respectively). The dataset is available as a Bioconductor data package [25].
Software and statistical analysis
The OBC algorithm was written in R (version 1.9.1), an open-source statistical software [26]. The algorithm is freely available [see Additional file 1] and further information can be find as well [see Additional file 2]. Some specific R packages were used in this study: the Bioconductor packages for microarray analysis [27]; ADE4 [28] and MADE4 [29] for multivariate analysis. The sarcoidosis dataset was normalized using the vsn algorithm [30].
Between group analysis
BGA is a particular extension of conventional ordination methods such as principal component analysis (PCA) or correspondence analysis (COA) where groups of samples are specified in advance [14]. The association of COA with BGA is particularly powerful, as COA has been shown to have several advantages over PCA in analysis of gene expression data [31,32]. In order to simplify the notations in the paper, the acronym BGA refers to the between-group correspondence analysis.
The between group analysis of the statistical triplet (X, Q, D) – where X is a data table of n rows (samples) and p columns (variables), Q is a p × p diagonal matrix containing the variable weights and D is a n × n diagonal matrix containing the sample weights – given the class indicator f, is the analysis of the triplet (G, Q, Dw) where G is the table of the means of X per group and Dw is the diagonal matrix of group weights [28]. Let us consider K the number of specified groups, a typical BGA yields K - 1 discriminating axes that ordinate the groups of sample by maximizing the between group variance (see [15] for mathematical details). Linear discriminant analysis is a related method which aims to maximize the percentage of variance explained by the grouping but which has different constraints and which cannot be applied to tables where the number of variables exceeds the number of samples [33].
Genes and samples ordinated by BGA can be projected on discriminating axes and visualized simultaneously on a biplot. The most discriminating genes are projected at the extremity of each axis whereas less informative genes are projected near the origin of each axis.
Authors' contributions
FB developed the algorithm, performed the analysis and wrote the paper in the team led by MHB. MHB supervised the study and provided substantial methodological input and was involved in drafting of the manuscript. MPB gave technical advice and biological input AC and GP provided important support regarding the refinements of the algorithm and BGA. All authors read and approved the manuscript.
Supplementary Material
Additional File 1
R code of the OBC algorithm.
Click here for file
Additional File 2
Further description of the sarcoidosis and tumour data. This files gives details about the optimal subset of genes obtained after OBC.
Click here for file
Acknowledgements
We would like to thank the reviewers whose comments were very stimulating and helped considerably to improve this paper. This study was sponsored by the Krebsliga beider Basel.
==== Refs
Li L Pedersen LG Darden TA Weinberg CR Class prediction and discovery based on gene expression data Genome Information Systems and Technology 2001
Yeung KY Bumgarner RE Multiclass classification of microarray data with repeated measurements: application to cancer Genome Biol 2003 4 R83 14659020 10.1186/gb-2003-4-12-r83
Li W Sun F Grosse I Extreme value distribution based gene selection criteria for discriminant microarray data analysis using logistic regression J Comput Biol 2004 11 215 226 15285889 10.1089/1066527041410445
Tan Y Shi L Tong W Wang C Multi-class cancer classification by total principal component regression (TPCR) using microarray gene expression data Nucleic Acids Res 2005 33 56 65 15640445 10.1093/nar/gki144
Xiong M Jin L Li W Boerwinkle E Computational methods for gene expression-based tumor classification Biotechniques 2000 29 1264 8 1270 11126130
Li L Weinberg C Darden T Pedersen L Gene selection for sample classification based on gene expression data: study of sensitivity to choice of parameters of the GA/KNN method Bioinformatics 2001 17 1131 42 11751221 10.1093/bioinformatics/17.12.1131
Lyons-Weiler J Patel S Bhattacharya S A classification-based machine learning approach for the analysis of genome-wide expression data Genome Res 2003 13 503 512 12618382 10.1101/gr.104003
Brown M Grundy W Lin D Cristianini N Sugnet C Furey T Ares M Haussler D Knowledge-based analysis of microarray gene expression data by using support vector machines Proc Natl Acad Sci USA 2000 97 262 7 10618406 10.1073/pnas.97.1.262
Furey T Cristianini N Duffy N Bednarski D Schummer M Haussler D Support vector machine classification and validation of cancer tissue samples using microarray expression data Bioinformatics 2000 16 906 14 11120680 10.1093/bioinformatics/16.10.906
Tibshirani R Hastie T Narasimhan B Chu G Diagnosis of multiple cancer types by shrunken centroids of gene expression Proc Natl Acad Sci USA 2002 99 6567 72 12011421 10.1073/pnas.082099299
Dudoit S Fridlyand J Speed TP Comparison of discrimination methods for the classification of tumors using gene expression data J Amer Stat Assoc 2002 97 77 87 10.1198/016214502753479248
Li W Yang Y Lin S, KF Johnson E How many genes are needed for a discriminant microarray data analysis? Methods of microarray data analysis 2002 Kluwer Academic 137 150
Ein-Dor L Kela I Getz G Givol D Domany E Outcome signature genes in breast cancer: is there a unique set? Bioinformatics 2005 21 171 178 15308542 10.1093/bioinformatics/bth469
Doledec S Chessel D Rhytmes saisonniers et composantes stationnelles en milieu aquatique. I – Description d'un plan d'observation complet par projection de variables Acta Oecologica Oecologia Generalis 1987 8 403 426
Culhane AC Perrière G Considine EC Cotter TG Higgins DG Between-group analysis of microarray data Bioinformatics 2002 18 1600 1608 12490444 10.1093/bioinformatics/18.12.1600
Zhang H Yu CY Singer B Cell and tumor classification using gene expression data: construction of forests Proc Natl Acad Sci USA 2003 100 4168 4172 12642676 10.1073/pnas.0230559100
Antonov AV Tetko IV Mader MT Budczies J Mewes HW Optimization models for cancer classification: extracting gene interaction information from microarray expression data Bioinformatics 2004 20 644 652 15033871 10.1093/bioinformatics/btg462
Rutherford RM Staedtler F Kehren J Chibout SD Joos L Tamm M Gilmartin JJ Brutsche MH Functional genomics and prognosis in sarcoidosis–the critical role of antigen presentation Sarcoidosis Vasc Diffuse Lung Dis 2004 21 10 18 15127969
Gene Expression Omnibus
Khan J Wei JS Ringner M Saal LH Ladanyi M Westermann F Berthold F Schwab M Antonescu CR Peterson C Meltzer PS Classification and diagnostic prediction of cancers using gene expression profiling and artificial neural networks Nat Med 2001 7 673 679 11385503 10.1038/89044
Small round blue cell tumours dataset
Alon U Barkai N Notterman DA Gish K Ybarra S Mack D Levine AJ Broad patterns of gene expression revealed by clustering analysis of tumor and normal colon tissues probed by oligonucleotide arrays Proc Natl Acad Sci USA 1999 96 6745 6750 10359783 10.1073/pnas.96.12.6745
Colon cancer dataset
Golub TR Slonim DK Tamayo P Huard C Gaasenbeek M Mesirov JP Coller H Loh ML Downing JR Caligiuri MA Bloomfield CD Lander ES Molecular classification of cancer: class discovery and class prediction by gene expression monitoring Science 1999 286 531 537 10521349 10.1126/science.286.5439.531
Leukemia dataset
R Development Core Team R: A language and environment for statistical computing 2004 R Foundation for Statistical Computing, Vienna, Austria [ISBN 3-900051-00-3].
Gentleman RC Carey VJ Bates DM Bolstad B Dettling M Dudoit S Ellis B Gautier L Ge Y Gentry J Hornik K Hothorn T Huber W Iacus S Irizarry R Leisch F Li C Maechler M Rossini AJ Sawitzki G Smith C Smyth G Tierney L Yang JY Zhang J Bioconductor: open software development for computational biology and bioinformatics Genome Biol 2004 5 R80 15461798 10.1186/gb-2004-5-10-r80
Chessel D Dufour AB Thioulouse J The ade4 package – I: One-table methods R News 2004 4 5 10
Culhane AC Thioulouse J Perrière G Higgins DG MADE4: an R package for multivariate analysis of gene expression data Bioinformatics 2005 21 2789 2790 15797915 10.1093/bioinformatics/bti394
Huber W von Heydebreck A Sultmann H Poustka A Vingron M Variance stabilization applied to microarray data calibration and to the quantification of differential expression Bioinformatics 2002 18 S96 104 12169536
Fellenberg K Hauser NC Brors B Neutzner A Hoheisel JD Vingron M Correspondence analysis applied to microarray data Proc Natl Acad Sci USA 2001 98 10781 10786 11535808 10.1073/pnas.181597298
Wouters L Gohlmann HW Bijnens L Kass SU Molenberghs G Lewi PJ Graphical exploration of gene expression data: a comparative study of three multivariate methods Biometrics 2003 59 1131 1139 14969494 10.1111/j.0006-341X.2003.00130.x
Perrière G Thioulouse J Use of correspondence discriminant analysis to predict the subcellular location of bacterial proteins Comput Methods Programs Biomed 2003 70 99 105 12507786 10.1016/S0169-2607(02)00011-1
|
16191195
|
PMC1261161
|
CC BY
|
2021-01-04 16:27:27
|
no
|
BMC Bioinformatics. 2005 Sep 28; 6:239
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-239
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2401619428110.1186/1471-2105-6-240Research ArticleSequence variation in ligand binding sites in proteins Magliery Thomas J [email protected] Lynne [email protected] Department of Molecular Biophysics & Biochemistry, Yale University, P.O. Box 208114, New Haven, CT 06520-8114, USA2 Present address: Department of Chemistry and Department of Biochemistry, The Ohio State University, 100 W. 18th Ave., Columbus, OH 43210, USA3 Department of Chemistry, Yale University, New Haven, CT, USA2005 30 9 2005 6 240 240 15 3 2005 30 9 2005 Copyright © 2005 Magliery and Regan; licensee BioMed Central Ltd.2005Magliery and Regan; 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 recent explosion in the availability of complete genome sequences has led to the cataloging of tens of thousands of new proteins and putative proteins. Many of these proteins can be structurally or functionally categorized from sequence conservation alone. In contrast, little attention has been given to the meaning of poorly-conserved sites in families of proteins, which are typically assumed to be of little structural or functional importance.
Results
Recently, using statistical free energy analysis of tetratricopeptide repeat (TPR) domains, we observed that positions in contact with peptide ligands are more variable than surface positions in general. Here we show that statistical analysis of TPRs, ankyrin repeats, Cys2His2 zinc fingers and PDZ domains accurately identifies specificity-determining positions by their sequence variation. Sequence variation is measured as deviation from a neutral reference state, and we present probabilistic and information theory formalisms that improve upon recently suggested methods such as statistical free energies and sequence entropies.
Conclusion
Sequence variation has been used to identify functionally-important residues in four selected protein families. With TPRs and ankyrin repeats, protein families that bind highly diverse ligands, the effect is so pronounced that sequence "hypervariation" alone can be used to predict ligand binding sites.
==== Body
Background
The central challenge of the post-genomic era is to determine the structures and functions of thousands of newly-identified putative proteins [1,2]. Elucidating how proteins carry out their functions in diverse contexts and in organisms from all three domains of life is both fundamentally important to understanding biological function and critical for engineering new functions into novel proteins. Sequence conservation alone can be used to structurally categorize many proteins or putative proteins [3]. Additionally, catalytic sites in enzymes can sometimes be identified from conserved surface motifs [4-8]. In contrast, sites with poor sequence conservation have been largely ignored, because they are assumed to be of little structural or functional importance [9].
Sequence alignment of proteins with similar structures has shown that as sequence identity increases, average backbone variation decreases [10]. Within a particular protein family, residues in the hydrophobic core are the most conserved, implying they play a key role in specifying the fold [11]. In contrast, solvent-exposed residues tend to be variable, with mutations having little deleterious effect on overall structure or stability [12]. Consequently, conservation of surface residues is commonly taken to be an indication of functional importance [13]. This idea can be used to identify active-site residues from a collection of proteins that perform the same function, but it is not applicable to families of proteins that use a common scaffold to bind diverse ligands. Rather, we might hypothesize that such binding sites will be composed of positions that are variable.
Recently, we used a statistical free energy (SFE) approach [14,15] to understand better the role of conserved residues in defining tetratricopeptide repeat (TPR) motifs, which are thought to commonly mediate protein-protein interactions [16]. Strikingly, we found that the ligand-binding site of the motif can readily be identified by sequence hypervariation of positions proximal to the ligand, as evidenced by very low statistical free energies separating those positions from a position-independent reference state. Here, we examine this observation in more detail and demonstrate that specificity-determining residues in TPRs, ankyrin (Ank) repeats, Cys2His2 zinc fingers (Zifs), and PDZ domains can be identified from sequence variation.
By analyzing protein families with exceptional biochemical and biophysical characterization, we show that, when the ligand repertoire is highly diverse for a particular family, the binding site can be identified from sequence hypervariation alone. However, even when the ligands have significant features in common, sequence variation can be used to "dissect" binding sites to identify specificity-determining residues. We demonstrate this sequence variation using probabilistic and information theory approaches closely tied to the mathematics of covariation, which are more suitable for this type of analysis than SFEs or Shannon entropies. Statistical identification of specificity-determining residues will greatly facilitate the engineering proteins with novel functions and targets.
Results & discussion
The TPR binding site
The TPR is a common 34 amino-acid protein motif that occurs in arrays, most frequently with three contiguous repeats [17]. Although TPR domains are thought to mediate protein-protein interactions, only a few examples have been well characterized. The large number of known TPR sequences, nearly 10,000 in Pfam [18], makes this motif an excellent target for statistical analysis. Using several mathematical approaches, we calculated the separation of the amino acid distribution at each position in TPRs from a position-independent reference state, amino acid usage in all proteins in yeast (Figure 1a). Note that we have performed this calculation on all of the 34 positions in the TPR motif; TPR domains are made up of tandem repeats of the TPR motif.
Figure 1 Relative entropy analysis of 6,887 canonical-length (34 aa) TPR repeats. (a) The relative entropy values are shown for each TPR position, with secondary structure indicated (cylinders represent helices and lines represent loops). Arrows indicate the positions of the seven most variable residues. These values are mapped onto the co-crystal structures of HOP-TPR1/Hsc70 peptide (b) and HOP-TPR2A/Hsp90 peptide (c), with the TPR domains rendered in spheres and the ligands in sticks. Two views from 180° rotation of each molecule are shown. The concave, ligand binding surfaces, left, are clearly more variable than the convex, solvent exposed surfaces, right. A small insertion in TPR2A is colored grey. (d) Views of the concave binding surfaces as in (c), but only those residues known to contact the ligand from co-crystal structures are colored [19]. Rendered from PDB entries 1ELW and 1ELR using PyMOL.
When relative entropy values are mapped onto the ligand-bound co-crystal structures of two different three-TPR domains (the TPR1 and TPR2A domains from Hsp-Organizing Protein, HOP [19]), it is immediately apparent that the concave, peptide-binding face of the TPR domains is more variable (i.e., more like the reference state) than the convex, solvent-exposed face (Figure 1b and 1c). For clarity, the concave surfaces in Figure 1d are depicted with coloration of only ligand-binding residues independently identified from crystallographic analysis [19]. For both TPR domains, every residue in contact with the ligand peptide is in a position with a small relative entropy (blue or green in the figures), indicating small differences from the reference distribution.
This hypervariation is a consequence of each TPR having evolved to bind a different specific ligand (or portion of a ligand). When TPR proteins are considered collectively, the binding positions are statistically randomized to an extent that is dictated by the repertoire of amino acids required to perform the range of binding functions. In contrast, solvent-exposed residues in general mutate in a stochastic fashion throughout evolution, slowly reverting or "drifting" toward a "neutral" distribution. The lack of structural or functional importance of solvent-exposed residues results in little selective pressure against mutation, but the extent of randomization is limited by evolutionary time and subtle factors such as overall protein solubility. As a result of the high diversity of the ligands of TPR repeats, the binding surface is more variable than the solvent-exposed surface. Specifically, we predict that the ligand-binding residues that show the most sequence variation are the positions that determine ligand specificity.
Positions 2, 5, 9, 12, 13, 33 and 34 show the most sequence variation in TPRs (i.e., they have the lowest relative entropies, ≤ 0.30). These seven residues all lie on the same face of the motif, and they are the residues that are exposed on the concave face of TPR domains. In fact, the TPR-peptide co-crystal structures show that residues in positions 2, 5, 6, 9, 12, and 13 are used by HOP TPR1 and TPR2A to bind their ligands [19]. Since few TPR-ligand structures have been solved, it remains to be seen whether or not other TPR domains utilize positions 33 and 34. Their spatial proximity to the other binding residues suggests that this is likely.
It is not surprising that some of the positions in contact with the ligand peptides are more biased than others (such as position 6). Some positions used for binding may also have other restrictions (such as structural restrictions) that limit the repertoire of amino acids allowed, and some positions may be important for binding affinity but not for specificity (which is to say they may bind a feature that is common among all ligands). For example, position 6 is modestly conserved in TPRs overall (it is frequently Asn) and is more buried than the other binding positions. The position-6 residues make contacts to the backbone of the ligand peptides here. It is worth noting that a position-6 Asn in the PEX5 C-terminal TPR domain also appears to make contact to the peptide backbone of the unrelated peroxisomal targeting signal peptide [20].
Several of the spheres in TPR2A are grey, indicating that they correspond to non-canonical positions and were therefore not calculated. At present, our analysis does not consider the effect of insertions and deletions. In the future, one could imagine including "deletion" as another "amino acid," so that site occupancy would contribute to the variation score.
Measuring sequence variation
The use of metrics that measure the difference from a position-independent reference distribution is key to our observation, because (1) it is not clear how mere lack of conservation is related to variability, and (2) the likelihood of mutation from one residue to another is affected by factors such as genetic code bias and the greater difficulty of accommodating bulky or reactive amino acids. Here, we take the reference state to be amino acid usage in all open reading frames in Saccharomyces cerevisiae [21], which is independent of position but accounts for genetic code bias and amino acid chemistry. Using SFE calculations, we have previously demonstrated that using this reference state gives virtually indistinguishable results from other position-independent reference states such as amino acid usage in all proteins in the Pfam database [16].
We originally noted that the ligand binding site of TPRs was identified by sequence variation using statistical free energies [16]. SFEs are essentially a measure of the difference between amino acid distributions, relating the "probability" of observing a particular distribution to thermodynamic importance based on the exponential relationship given by the Boltzmann law.
PP(ref)=eΔGstatkT* [1]
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaadaWcaaqaaiabdcfaqbqaaiabdcfaqnaaBaaaleaacqGGOaakcqWGYbGCcqWGLbqzcqWGMbGzcqGGPaqkaeqaaaaakiabg2da9iabdwgaLnaaCaaaleqabaWaaSaaaeaacqqHuoarcqWGhbWrdaWgaaadbaGaem4CamNaemiDaqNaemyyaeMaemiDaqhabeaaaSqaaiabdUgaRjabdsfaunaaCaaameqabaGaeiOkaOcaaaaaaaGccaWLjaGaaCzcaGqabiab=TfaBjab=fdaXiab=1faDbaa@4846@
(See Methods for an explanation of kT*.) This approach was introduced by Lockless & Ranganathan; however, their formalism for SFEs does not explicitly calculate the probability of observing a particular positional distribution [14]. Instead it uses the root-mean-square of the binomial probabilities of observing each amino acid, over all twenty amino acids x. That is,
ΔGstat=kT*∑x(lnPxPx(ref))2 [2]
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqqHuoarcqWGhbWrdaWgaaWcbaGaem4CamNaemiDaqNaemyyaeMaemiDaqhabeaakiabg2da9iabdUgaRjabdsfaunaaCaaaleqabaGaeiOkaOcaaOWaaOaaaeaadaaeqbqaamaabmaabaGagiiBaWMaeiOBa42aaSaaaeaacqWGqbaudaWgaaWcbaGaemiEaGhabeaaaOqaaiabdcfaqnaaBaaaleaacqWG4baEcqGGOaakcqWGYbGCcqWGLbqzcqWGMbGzcqGGPaqkaeqaaaaaaOGaayjkaiaawMcaamaaCaaaleqabaGaeGOmaidaaaqaaiabdIha4bqab0GaeyyeIuoaaSqabaGccaWLjaGaaCzcaGqabiab=TfaBjab=jdaYiab=1faDbaa@5301@
where Px is given by
Px=N!nx!(N−nx)!fxnx(1−fx)N−nx [3]
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGqbaudaWgaaWcbaGaemiEaGhabeaakiabg2da9maalaaabaGaemOta4KaeiyiaecabaGaemOBa42aaSbaaSqaaiabdIha4bqabaGccqGGHaqidaqadaqaaiabd6eaojabgkHiTiabd6gaUnaaBaaaleaacqWG4baEaeqaaaGccaGLOaGaayzkaaGaeiyiaecaaiabdAgaMnaaDaaaleaacqWG4baEaeaacqWGUbGBdaWgaaadbaGaemiEaGhabeaaaaGcdaqadaqaaiabigdaXiabgkHiTiabdAgaMnaaBaaaleaacqWG4baEaeqaaaGccaGLOaGaayzkaaWaaWbaaSqabeaacqWGobGtcqGHsislcqWGUbGBdaWgaaadbaGaemiEaGhabeaaaaGccaWLjaGaaCzcaGqabiab=TfaBjab=ndaZiab=1faDbaa@5462@
Here, N is the total number of sequences, nx is the number of sequences with amino acid x at the given position, and fx is the expected frequency of x from the reference state. This "vector" formalism for estimating the overall probability is empirically quite effective, but we speculated that a metric more tied to the mathematics of covariation would be more rigorous for our approach.
Since both of the following are true,
∑xnx=N and ∑xfx=1
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaadaaeqbqaaiabd6gaUnaaBaaaleaacqWG4baEaeqaaaqaaiabdIha4bqab0GaeyyeIuoakiabg2da9iabd6eaojaaykW7caaMc8UaaGPaVlaaykW7caaMc8UaeeyyaeMaeeOBa4MaeeizaqMaaGPaVlaaykW7caaMc8UaaGPaVlaaykW7cqqGGaaidaaeqbqaaiabdAgaMnaaBaaaleaacqWG4baEaeqaaaqaaiabdIha4bqab0GaeyyeIuoakiabg2da9iabigdaXaaa@5238@
the probability of observing a particular distribution is simply given by the multinomial probability,
Pmult=N!∏xnx!∏xfx [4]
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGqbaudaWgaaWcbaGaemyBa0MaemyDauNaemiBaWMaemiDaqhabeaakiabg2da9maalaaabaGaemOta4KaeiyiaecabaWaaebuaeaacqWGUbGBdaWgaaWcbaGaemiEaGhabeaakiabcgcaHaWcbaGaemiEaGhabeqdcqGHpis1aaaakmaarafabaGaemOzay2aaSbaaSqaaiabdIha4bqabaGcdaahaaWcbeqaaiabd6gaUnaaBaaameaacqWG4baEaeqaaaaaaSqaaiabdIha4bqab0Gaey4dIunakiaaxMaacaWLjaWexLMBbXgBcf2CPn2qVrwzqf2zLnharyGvLjhzH5wyaGabbiaa=TfacaWF0aGaa8xxaaaa@563C@
As expected, the ΔGstat values are very closely related (R2 = 0.89) to the ln Pmult values for the 34 positions in the TPR motif (Figure 2a). Note, however, that the values of ΔGstat and ln Pmult are dependent upon the total number of sequences N (since it is much less likely to observe a particular amino acid 200 times out of 1000 than 20 out of 100, if you are expecting it only 5% of the time).
Figure 2 Measuring differences in distributions. (a) Lockless & Ranganathan statistical free energies versus the logarithm of the multinomial probability for each of the 34 sites in TPRs. (b) Relationship of SFEs to sequence (Shannon) entropy for TPR sites. (c) Relationship of logarithm of multinomial probabilities to sequence entropy (circles) and relative entropy (squares).
Recently, Dekker et al. suggested that SFEs are merely a measure of sequence (Shannon) entropy (Hi), which implicitly measures how a distribution varies from equal usage [22]. (This is because the maximum entropy arises from a distribution with equal usage.)
Hi=−∑xpxlnpx [5]
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGibasdaWgaaWcbaGaemyAaKgabeaakiabg2da9iabgkHiTmaaqafabaGaemiCaa3aaSbaaSqaaiabdIha4bqabaGccyGGSbaBcqGGUbGBcqWGWbaCdaWgaaWcbaGaemiEaGhabeaaaeaacqWG4baEaeqaniabggHiLdGccaWLjaGaaCzcamXvP5wqSXMqHnxAJn0BKvguHDwzZbqegyvzYrwyUfgaiqqacaWFBbGaa8xnaiaa=1faaaa@4C11@
Here, px is the proportion of sequences with amino acid x at position i. While it is true that ΔGstat is correlated with Hi (Figure 2b), this correlation is a consequence of the reference state of ΔGstat calculations being fairly close to equal usage of amino acids. It is also affected by the fact that ΔGstat values are not based on a rigorous measure of overall probability.
Plotting the multinomial probabilities associated with the distributions at the 34 positions in TPRs against sequence entropy and relative entropy clearly demonstrates that ln Pmult measures the same thing as relative entropy, but something different from sequence entropy (Figure 2c). (The difference becomes more dramatic the more the reference state deviates from equal usage.) In fact, relative entropy D(p||f) is an information theory approach to measuring the "distance" between distributions, given by,
D(p||f)=∑xpxlnpxfx [6]
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGebarcqGGOaakcqWGWbaCcqGG8baFcqGG8baFcqWGMbGzcqGGPaqkcqGH9aqpdaaeqbqaaiabdchaWnaaBaaaleaacqWG4baEaeqaaOGagiiBaWMaeiOBa42aaSaaaeaacqWGWbaCdaWgaaWcbaGaemiEaGhabeaaaOqaaiabdAgaMnaaBaaaleaacqWG4baEaeqaaaaaaeaacqWG4baEaeqaniabggHiLdGccaWLjaGaaCzcamXvP5wqSXMqHnxAJn0BKvguHDwzZbqegyvzYrwyUfgaiqqacaWFBbGaa8Nnaiaa=1faaaa@5410@
It can be shown, using the Stirling approximation for factorials, that multinomial probability is in fact linearly related to relative entropy by the number of sequences (see Supplemental Material).
One significant advantage of relative entropy over multinomial probability and ΔGstat is that relative entropy is independent of the total number of sequences N. Since the intention of the SFE approach is to estimate the significance associated with an amino acid distribution relative to a reference state, we submit that relative entropies are the most convenient way to do this. Relative entropies combine the sample-size independence and ease of calculation of Shannon entropies with the reference-state correction of Lockess & Ranganathan's method, while at the same time measuring that correction in a mathematically-rigorous way.
Other methods, in addition to Shannon entropy and the Lockless & Ranaganthan method, have been suggested for scoring residue conservation, including metrics that account for residue properties such as size or hydrophobicity [23]. There is evidence that binding sites have unique compositional preferences, which may suggest alternative reference states for our method [24]. It will be interesting to examine how attention to property variation may improve our method of dissecting binding sites in structural families.
Effects of sample size
In order to determine how many sequences are required to identify binding residues in TPRs by this method, subsets of various sizes were randomly selected from the 6,887 TPR sequences. In Figure 3a, the average relative entropy values from subsets (5 each) with approximately 6887, 3444, 1722, 861, 430, 215 and 108 randomly-selected sequences are shown for all positions. The overall pattern is evident with as few as about 100 sequences, and there is virtually no difference between subsets with 400 or more sequences. Values from the five random subsets of each size are shown for the seven lowest relative entropy positions in the full data set (Figure 3b). Again, there is essentially no discernable difference down to as few as 400 sequences, and there is not significant variability until one examines fewer than 200 sequences. We therefore expect that this analysis is applicable to protein families with as few as about 200–400 sequences.
Figure 3 Effects of sample size. (a) Average relative entropy associated with each of the 34 positions in TPRs with random subsets of various sizes. Each cluster of bars represents one position in the TPR motif. The cluster is composed of bars, left to right, from sets with approximately 6887, 3444, 1722, 861, 430, 215 and 108 sequences. Each bar is the average of five subsets of the same size (except 6887, since there is only one set this size – all sequences). (b) Relative entropies associated with five randomly chosen subsets of various sizes for the seven positions most like the reference state. Each cluster of bars represents one position. The individual bars show the calculated relative entropies for subsets of the same sizes as in (a) (five of each size).
Ankyrin repeats: comparison with experiment
A corresponding analysis of ankyrin (Ank) repeats, another experimentally well-characterized protein-binding motif, clearly confirms our prediction that low relative entropies can be used to identify specificity-determining residues when the repertoire of ligand is highly variable. Anks are helix-turn-helix-loop motifs, which bind their ligands with residues in the loops and on the surface of the helical array proximal to the loops [25]. Figure 4 shows the relative entropies from over 15,000 Ank repeats mapped onto the co-crystal structure of mouse GA binding protein β1 with the GABPα ligand bound [26]. Again, our analysis dramatically reveals that the residues known to form the binding site are among the most variable; the positions most like the reference state (relative entropy ≤ 0.39) are 2, 3, 5, 13, 14, 17, 32, and 33.
Figure 4 Relative entropy analysis of canonical positions in 15,497 Ank repeats. (a) The positional relative entropies are shown with secondary structural elements noted (grey arrows are β-strands). Blue and green arrows indicate the most variable positions; asterisks (*) indicate positions mutated by the Plückthun lab to alter Ank-domain specificity. (b) The location of the binding site in a single Ank repeat in the loop and proximal α-helical surface is labeled. (c) The 4-Ank domain from GABPβ1 (spheres) is shown bound to the ligand GABPα (ribbons) in two views from 180° rotation. Again, the binding surface is evident from the low relative entropies. Note that some non-binding surface-exposed positions, particularly turn residues, are conserved due to their importance in defining the Ank fold. Some positions in GABPβ1 do not map onto the canonical Ank sequence and are colored grey. Rendered using PyMOL from PDB entry 1AWC.
Significantly, the functional importance of the residues we have identified has already been verified experimentally. Plückthun and colleagues considered four Ank-ligand co-crystal structures and chose key interacting residues in Ank repeats from those whose solvent accessibility was most changed upon ligand binding: positions 2, 3, 5, 13, 14, and 33. Ank repeat domains which bind to different ligands, for example maltose binding protein and two kinases, have been selected from a library of Ank repeat domains in which only these positions were randomized [27,28]. This result confirms our proposal that these are the specificity-determining residues. In fact, the crystal structure of one of the selected ankyrin proteins that binds MBP directly demonstrates the role of these residues in binding. In addition, position 32 is close in space to these residues, and may well participate in binding for some Ank proteins. It is not clear why position 17, which lies in the turn between the motif's helices, is more variable than other non-binding surface-exposed positions.
Dissecting binding sites when ligands have conserved features
We also analyzed Cys2His2 zinc fingers (Zifs) in the same way. In contrast to TPR and Ank repeats, which are stabilized by burial of hydrophobic residues, Zifs are mostly stabilized by ligation of zinc(II) and binding to DNA. Not surprisingly, then, positions that ligate the zinc ion (Cys-10, Cys-7, His+7, His+11) and a subset of positions that contact the DNA (e.g., Tyr-12, Lys-5, Phe-3, Arg+9) are highly biased (where -1 is the position immediately before the α-helix and the consensus residue is listed). Residues with low relative entropy (≤0.5) are essentially in two patches, near the end of the end of the helix buried in the major groove and on the solvent-exposed surface distal to the DNA. When one considers only the positions that are in contact with the ligand DNA, the residues with the lowest relative entropies (blue and green spheres in Figure 5) are positions -2, -1, 1, 2, 3, 5 and 6. Extensive phage-display selection work has shown that positions -1, 1, 2, 3, 5 and 6 are critical to specificity for the target DNA sequences studied [29,30]. In contrast, positions that contact the DNA but have no effect on specificity, such as basic residues that make contacts to the phosphate backbone, are essentially invariant (orange and red spheres in Figure 5).
Figure 5 Relative entropy analysis of canonical positions in 28,442 C2H2 zinc fingers. (a) Positions in the graph are shown in the order found in Pfam and numbered by convention (where -1 is the residue N-terminal to the α-helix). Note that the y-axis scale is different from Figs. 1 and 2 due to the almost invariant zinc(II)-binding residues (-10, -7, +7 and +11). Blue and green arrows indicate the seven predicted specificity-determining positions. (b) The middle zinc finger of Zif268 bound to DNA (purple) is shown, with the Zn(II) atom as a pink sphere [43]. (c) The residues in contact with the DNA from all three zinc fingers of Zif268 are rendered in spheres. The DNA-binding positions group into variable, specificity-determining positions (blue and green spheres) projecting into the major groove of the DNA, and conserved positions that enhance affinity to DNA but do not affect specificity (orange and red spheres). Rendered with PyMOL from PDB entry 1A1I.
Note that in this case the specificity-determining residues are not necessarily the most variable residues in the motif; they are the most variable residues in the motif that are in contact with the ligand. Thus, for Zifs, sequence hypervariation is not sufficient to identify the binding site, but statistical analysis together with a sample structure reveals specificity-determining positions without further characterization. Apparently, the repertoire of amino acids needed to bind DNA in the major groove is less diverse than that needed for the range of binding functions exhibited by TPR or Ank repeat domains.
A similar phenomenon is observed for PDZ domains, whose peptide ligands have highly conserved elements. PDZ domains are ubiquitous globular protein-protein interaction domains. It is thought that most PDZ domains bind to the C-terminus of target proteins, typically making contact to the carboxyl-terminal four to five residues. PDZ domains can be categorized into two classes (I and II) that bind to consensus sequences X-(S/T)-X-(V/I/L)-CO2- and X-Φ-X-Φ-CO2-, respectively (where Φ represents a hydrophobic residue and X represents an arbitrary residue) [31]. Since our method identifies residues that vary with the repertoire of ligands (i.e., specificity-determining residues), we would expect that positions that bind to the terminal carboxylate and C-terminal hydrophobic side-chain (P0) will be highly biased (i.e., conserved); positions that bind to the alcohol or hydrophobic residue at P-2 will be biased; and positions that bind to P-1 and P-3 will be highly variable. In Figure 6b, we have highlighted the positions in the example PDZ domain 3 from PSD95 that have been identified from NMR studies and X-ray co-crystal structures to be involved in binding to the terminal four residues of the ligand (323–328, 339, 340, 342, 372, 373, 376, 379 and 380) [32]. We also included positions 318, 322, 329 and 331, which are within 5 Å of the ligand peptide (KQTSV-CO2-) in the example structure (computed with DeepView [33]).
Figure 6 Relative entropy analysis of canonical positions in 2,751 PDZ domains. (a) Positions in the graph are shown in the order found in Pfam and with the same numbering. Only positions with greater than 50% occupancy were calculated. The eight variable binding positions are marked with arrows, and the corresponding residue number in PSD95-3 is listed. (b) The structure of PSD95-3 with its cognate ligand peptide, KQTSV [44]. Note that atoms are missing from the ligand lysine sidechain due to lack of electron density in the X-ray data. The structurally-determined binding residues (see text) are colored, and the eight predicted specificity-determining positions are labeled with residue numbers as in (a). Rendered with PyMOL from PDB entry 1BE9.
The eight most variable residues in this group are 331, 342, 376, 373, 339, 328, 340 and 326. The most variable residue, 331 (here, a glutamate), contacts the P-4 lysine, which is variable among PDZ ligands and whose effect on specificity has not been extensively examined. Positions 326, 328, 339, 340 and 342 interact with the variable P-1 and P-3 positions. Interestingly, computational redesign of PDZ domain specificity confirms the central importance of these residues in specificity determination. Reina et al. were able to change the specificity of PSD95-3 from KQTSV to KITWV and KRTEV (retaining ligand class, since P0 and P-2 are the same) [34]. In the first case, residues 326, 339, 340, 342, 380 and 397 were mutated. In the second case, residues 326, 328, 339, 340, 342 and 397 were mutated. Note that position 397 is outside of the canonical PDZ domain, and so was not examined in our work; position 380 was mutated to improve the stability of the domain, not its binding specificity.
Positions 373 and 376 contact the P-2 position, which is either an alcohol or a hydrophobic residue depending upon class. The identity of position 372 is known to be highly correlated with the ligand class, because it is typically occupied by a polar residue (often histidine) for class I ligands and by a hydrophobic residue for class II ligands [31]. As expected, position 372 displays intermediate variability (yellow in the figure). We hypothesize that positions 373 and 376 are much more variable than 372 and 380 because they are further away from the P-2 threonine; in fact, position 373 is farther than 5 Å away. These residues are likely more important for binding class II ligands in which P-2 is hydrophobic and therefore generally larger. P0 is a hydrophobic residue in virtually every known PDZ ligand, and it invariantly presents the carboxylate terminus of the peptide. Not surprisingly, then, the residues that it contacts (322–325, 327) are highly biased. In fact, position 324 is a glycine in 97% of PDZ domains, and the turn in which it lies hosts the carboxylate.
It is worth noting that, in contrast to the other three examples presented above, this calculation was carried out on a whole domain instead of a repeat motif. It is also worth noting that the binding site, as in the case of zinc fingers, could not be easily predicted from relative variability of sequence alone due to commonalities among the ligands that result in conserved elements of the binding sites. However, in combination with an example co-crystal structure, the specificity-determining positions can again be inferred from sequence variation, and the inference matches closely what has been derived from extensive biochemical characterization and engineering.
The meaning of relative entropy values
In our previous study of TPRs by SFE analysis, we empirically demonstrated that for a particular sample size and scaling, levels of sequence variation (ΔGstat) could be usefully grouped as such: 0–1.25, hypervariable or no bias; 1.25–2.5, slight bias; 2.5–5.0, significant bias; 5.0–10.0, dramatic bias; 10+, restriction to a small subset of amino acids [16]. Regression analysis between SFE and relative entropy values for all TPR positions suggests that these values correspond to relative entropies of approximately 0.3, 0.5, 0.9 and 1.5. For convenience, we therefore chose 0.0–0.3, 0.3–0.6, 0.6–0.9, 0.9–1.2 and 1.2+ as bins for coloration of the figures in this publication. The examples in this study suggest that "normal" surface positions typically exhibit relative entropies in the range of 0.3–0.6, and that specificity-determining positions typically have relative entropies less than 0.5. The overlap of these values highlights the difficulty in using this approach as a purely predictive algorithm: only when the repertoire of ligands is extremely diverse (as with TPRs and ankyrin repeats) is there a clear distinction between ligand-binding residues and surface residues in general. We are in the process of a much broader application of this procedure to all of the families in the Pfam database, which we will use to refine the meaning of the relative entropy values in protein families overall (M. Gerstein, T. Gianoulis, T.J.M. and L.R., unpublished work).
Conclusion
The notion that positions that bind diverse ligands will be variable among a family of proteins seems fairly obvious, but this approach has not yet been utilized as a general strategy. One notable precedent is seen in the original studies of antigen-binding sites in antibodies, which were identified as variable regions when the amino acid sequences of antibodies were first determined [35]. Various family-based approaches have been applied to the prediction of functional residues, typically analyzing sequence variability from collections of proteins with similar function (and therefore emphasizing the functional importance of conserved residues). For example, "evolutionary trace" and related methods divide multiply-aligned sequences into subfamilies, typically by phylogeny, comparing patterns of conservation among evolutionarily-related subfamilies and often mapping onto 3D structure [13,36-42]. Basically, these methods posit that positions that are conserved in all sequences are important for structure, and positions that are conserved within subfamilies (but vary among the sub-types) are important for function (i.e., the function of the proteins in the subfamily).
Here we show how analysis of sequence variability can be enlarged to understand functional variability in whole families of proteins with similar structures. If one collects proteins with the same structure and diverse functions, then structural positions will be conserved and functional positions will vary, and the degree of variation will be related to the degree of variation among the ligands or substrates. In the case of repeat motifs such as TPRs and ankyrins, where structural elements are further divorced from functional conservation by ignoring how the motifs are arranged in domains, the degree of variability is sometimes so profound that it alone can be used to predict the binding site. When the ligands have commonalities, then it becomes more difficult to predict the binding site from variation alone. In that case, as with PDZ domains and Zifs, variation in combination with an example structure still reveals specificity-determining binding positions, which is critical information for re-engineering specificity. (The corollary to this argument is that the pattern of variation of known binding residues will suggest the pattern of variation in the ligands.)
Analysis of overall variance among structurally-related families provides complementary information to methods that analyze variance among evolutionarily-related subfamilies, which have proven very powerful in recent years. A major challenge for these evolutionary trace methods is accurate functional sub-typing, particularly when family members have diverged very significantly. Our method avoids functional sub-typing and, rather, benefits from increased functional divergence of family members (since it results in increased variation among functional positions). Further attention to variable residues in families overall therefore stands to improve exiting methods of functional prediction.
There are hundreds of binding scaffolds with sufficient examples known to permit this type of statistical analysis. The use of rigorous measures of how amino acid distributions differ improves significantly upon conservation alone as a means of identifying important residues within a protein family (this has been reviewed recently [23]). The rapid identification of specificity-determining positions will be useful for the design of proteins with altered binding specificity. The predictions of specificity-determining residues in Ank repeat proteins, Zifs and PDZ domains agree strikingly well with results from considerable structural and biochemical work, and therefore provide a guide for re-engineering binding specificity by design even for protein families lacking extensive characterization. Moreover, knowledge of the specificity-determining residues can be incorporated into evolutionary trace methods to develop a comprehensive view of residues critical for function.
Methods
Sequences
Aligned sequences of TPRs, Ank repeats, C2H2 Zifs and PDZ domains were downloaded from Pfam [18]. TPRs of non-canonical length (i.e., not 34 amino acids) were discarded, and only canonical positions were considered with Ank repeats, Zifs and PDZ domains (i.e., ignoring low-occupancy positions from insertions and deletions). All calculations were carried out in Microsoft Excel 2003 on a Dell Latitude C640 with a 2.2 GHz Intel Mobile Pentium 4 processor. Factorials were computed from the Stirling approximation.
Statistical Free Energies
The SFEs associated with each amino acid were determined from application of the Boltzmann law [1], where k is the Boltzmann constant, kT* is an arbitrary energy unit (since the "temperature" of the ensemble T* is not necessarily related to T for conventional systems), and Pref is the probability associated with a hypothetical site with amino acid usage as in the reference state. The binomial probability Px of the observation of nx sequences with amino acid x was calculated from [3], where fx is the frequency of the amino acid in a reference state, all ORFs in Saccharomyces cerevisiae. The SFEs associated with the observed frequencies of the 20 amino acids at each site can be thought of as elements of a 20-dimensional vector. The scalar length of this vector, the root-mean-square average for all amino acids [2], is therefore taken to be the statistical free energy, ΔGstat, that separates the observed positional amino acid distribution from the reference state. For comparison to Lockless & Ranganathan [14], the ΔGstat values were arbitrarily divided by 100. However, the Ranganathan group normalizes the number of sequences to 100, and we have shown that ΔGstat is proportional to N for large N. Therefore, the ΔGstat values we calculate are 10-fold larger than those calculated by the Ranganathan method for the same amino acid distribution, since we have normalized to N = 1000.
Other metrics of distribution difference
Multinomial probabilities [4], sequence (Shannon) entropies [5] and relative entropies [6] were calculated as described above. For sequence and relative entropy calculations, the frequencies were calculated as:
px=nx+1N+1 [7]
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGWbaCdaWgaaWcbaGaemiEaGhabeaakiabg2da9maalaaabaGaemOBa42aaSbaaSqaaiabdIha4bqabaGccqGHRaWkcqaIXaqmaeaacqWGobGtcqGHRaWkcqaIXaqmaaGaaCzcaiaaxMaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaeHbwvMCKfMBHbaceeGaa83waiaa=DdacaWFDbaaaa@46FA@
so that ln px was always defined (and is valid as long as N is large). The values used for fx are listed in our previous paper [16].
List of abbreviations
Ank, ankyrin; HOP, Hsp-organizing protein; MBP, maltose binding protein; SFE, statistical free energy; TPR, tetratricopeptide repeat protein; Zif, zinc finger
Authors' contributions
T.J.M. performed all calculations and analysis, and drafted the manuscript. L.R. aided in interpretation of the data and manuscript preparation, and provided support.
Supplementary Material
Additional File 1
The relationship between multinomial probability and relative entropy is derived.
Click here for file
Acknowledgements
We thank A. López Cortajarena, T. Kajander, S. Mochrie, J. Venkatraman and C.G.M. Wilson (Yale) and S.S. Licht (MIT) for critical reading of this manuscript and insightful comments. Special thanks to M. Gerstein (Yale) for helpful suggestions on calculating distribution differences. T.J.M. was an N.I.H. Postdoctoral Fellow (GM065750). This work was supported in part by N.I.H. grants GM49146 and GM62413 (L.R.).
==== Refs
Teichmann SA Murzin AG Chothia C Determination of protein function, evolution and interactions by structural genomics Current Opinion in Structural Biology 2001 11 354 363 11406387 10.1016/S0959-440X(00)00215-3
Thornton JM From genome to function Science 2001 292 2095 2097 11408660 10.1126/science.292.5524.2095
Gough J Karplus K Hughey R Chothia C Assignment of homology to genome sequences using a library of hidden Markov models that represent all proteins of known structure Journal of Molecular Biology 2001 313 903 919 11697912 10.1006/jmbi.2001.5080
Bork P Koonin EV Protein sequence motifs Current Opinion in Structural Biology 1996 6 366 376 8804823 10.1016/S0959-440X(96)80057-1
Whisstock JC Lesk AM Prediction of protein function from protein sequence and structure Quarterly Review of Biophysics 2003 36 307 340 10.1017/S0033583503003901
Yao H Kristensen DM Mihalek I Sowa ME Shaw C Kimmel M Kavraki L Lichtarge O An accurate, sensitive, and scalable method to identify functional sites in protein structures Journal of Molecular Biology 2003 326 255 261 12547207 10.1016/S0022-2836(02)01336-0
Ma B Elkayam T Wolfson H Nussinov R Protein-protein interactions: structurally conserved residues distinguish between binding sites and exposed protein surfaces Proceedings of the National Academy of Sciences of the United States of America 2003 100 5772 5777 12730379 10.1073/pnas.1030237100
Jones S Thornton JM Searching for functional sites in protein structures Current Opinion in Chemical Biology 2004 8 3 7 15036149 10.1016/j.cbpa.2003.11.001
Mihalek I Res I Lichtarge O A family of evolution-entropy hybrid methods for ranking protein residues by importance Journal of Molecular Biology 2004 336 1265 1282 15037084 10.1016/j.jmb.2003.12.078
Chothia C Lesk AM The relation between the divergence of sequence and structure in proteins EMBO Journal 1986 5 823 826 3709526
Bashford D Chothia C Lesk AM Determinants of a protein fold. Unique features of the globin amino acid sequences Journal of Molecular Biology 1987 196 199 216 3656444 10.1016/0022-2836(87)90521-3
Bowie JU Reidhaar-Olson JF Lim WA Sauer RT Deciphering the message in protein sequences: tolerance to amino acid substitutions Science 1990 247 1306 1310 2315699
Lichtarge O Bourne HR Cohen FE An evolutionary trace method defines binding surfaces common to protein families Journal of Molecular Biology 1996 257 342 358 8609628 10.1006/jmbi.1996.0167
Lockless SW Ranganathan R Evolutionarily conserved pathways of energetic connectivity in protein families Science 1999 286 295 299 10514373 10.1126/science.286.5438.295
Shortle D Propensities, probabilities and the Boltzmann hypothesis Protein Science 2003 12 1298 1302 12761401 10.1110/ps.0306903
Magliery TJ Regan L Beyond consensus: statistical free energies reveal hidden interactions in the design of a TPR motif J Mol Biol 2004 343 731 745 15465058 10.1016/j.jmb.2004.08.026
D'Andrea LD Regan L TPR proteins: the versatile helix Trends in Biochemical Sciences 2003 28 655 662 14659697 10.1016/j.tibs.2003.10.007
Sonnhammer EL Eddy SR Durbin R Pfam: a comprehensive database of protein domain families based on seed alignments Proteins 1997 28 405 420 9223186 10.1002/(SICI)1097-0134(199707)28:3<405::AID-PROT10>3.0.CO;2-L
Scheufler C Brinker A Bourenkov G Pegoraro S Moroder L Bartunik H Hartl FU Moarefi I Structure of TPR domain-peptide complexes: critical elements in the assembly of the Hsp70-Hsp90 multichaperone machine Cell 2000 101 199 210 10786835 10.1016/S0092-8674(00)80830-2
Gatto GJJ Geisbrecht BV Gould SJ Berg JM Peroxisomal targeting signal-1 recognition by the TPR domains of human PEX5 Nature Structural Biology 2000 7 1091 1095 11101887 10.1038/81930
Kanehisa M Goto S Kawashima S Nakaya A The KEGG databases at GenomeNet Nucleic Acids Research 2002 30 42 46 11752249 10.1093/nar/30.1.42
Dekker JP Fodor A Aldrich RW Yellen G A perturbation-based method for calculating explicit likelihood of evolutionary co-variance in multiple sequence alignments Bioinformatics 2004 20 1565 1572 14962924 10.1093/bioinformatics/bth128
Valdar WS Scoring residue conservation Proteins 2002 48 227 241 12112692 10.1002/prot.10146
Villar HO Kauvar LM Amino acid preferences at protein binding sites FEBS Lett 1994 349 125 130 8045288 10.1016/0014-5793(94)00648-2
Sedgwick SG Smerdon SJ The ankyrin repeat: a diversity of interactions on a common structural framework Trends in Biochemical Sciences 1999 24 311 316 10431175 10.1016/S0968-0004(99)01426-7
Batchelor AH Piper DE de la Brousse FC McKnight SL Wolberger C The structure of GABPalpha/beta: an ETS domain- ankyrin repeat heterodimer bound to DNA. Science 1998 279 1037 1041 9461436 10.1126/science.279.5353.1037
Binz HK Amstutz P Kohl A Stumpp MT CBriand C Forrer P Grutter MG Pluckthun A High-affinity binders selected from designed ankyrin repeat protein libraries Nature Biotechnology 2004 22 575 582 15097997 10.1038/nbt962
Binz HK Stumpp MT Forrer P Amstutz P Pluckthun A Designing repeat proteins: well-expressed, soluble and stable proteins from combinatorial libraries of consensus ankyrin repeat proteins Journal of Molecular Biology 2003 332 489 503 12948497 10.1016/S0022-2836(03)00896-9
Wolfe SA Nekludova L Pabo CO DNA recognition by Cys(2)His(2) zinc finger proteins Annual Review of Biophysics & Biomolecular Structure 2000 29 183 212 10.1146/annurev.biophys.29.1.183
Pabo CO Peisach E Grant RA Design and selection of novel Cys(2)His(2) zinc finger proteins Annual Review of Biochemistry 2001 70 313 340 11395410 10.1146/annurev.biochem.70.1.313
Songyang Z Fanning AS Fu C Xu J Marfatia SM Chishti AH Crompton A Chan AC Anderson JM Cantley LC Recognition of unique carboxyl-terminal motifs by distinct PDZ domains Science 1997 275 73 77 8974395 10.1126/science.275.5296.73
Wiedemann U Boisguerin P Leben R Leitner D Krause G Moelling K Volkmer-Engert R Oschkinat H Quantification of PDZ domain specificity, prediction of ligand affinity and rational design of super-binding peptides J Mol Biol 2004 343 703 718 15465056 10.1016/j.jmb.2004.08.064
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
Reina J Lacroix E Hobson SD Fernandez-Ballester G Rybin V Schwab MS Serrano L Gonzalez C Computer-aided design of a PDZ domain to recognize new target sequences Nat Struct Biol 2002 9 621 627 12080331
Hilschmann N Craig LC Amino acid sequence with Bence-Jones proteins Proceedings of the National Academy of Sciences of the United States of America 1965 53 1403 1409 5324619
Sali A Overington JP Johnson MS Blundell TL From comparisons of protein sequences and structures to protein modelling and design Trends Biochem Sci 1990 15 235 240 2200167 10.1016/0968-0004(90)90036-B
Hannenhalli SS Russell RB Analysis and prediction of functional sub-types from protein sequence alignments J Mol Biol 2000 303 61 76 11021970 10.1006/jmbi.2000.4036
Innis CA Shi J Blundell TL Evolutionary trace analysis of TGF-beta and related growth factors: implications for site-directed mutagenesis Protein Eng 2000 13 839 847 11239083 10.1093/protein/13.12.839
Landgraf R Xenarios I Eisenberg D Three-dimensional cluster analysis identifies interfaces and functional residue clusters in proteins J Mol Biol 2001 307 1487 1502 11292355 10.1006/jmbi.2001.4540
del Sol Mesa A Pazos F Valencia A Automatic methods for predicting functionally important residues J Mol Biol 2003 326 1289 1302 12589769 10.1016/S0022-2836(02)01451-1
Folkertsma S van Noort P Van Durme J Joosten HJ Bettler E Fleuren W Oliveira L Horn F de Vlieg J Vriend G A family-based approach reveals the function of residues in the nuclear receptor ligand-binding domain J Mol Biol 2004 341 321 335 15276826 10.1016/j.jmb.2004.05.075
Berezin C Glaser F Rosenberg J Paz I Pupko T Fariselli P Casadio R Ben-Tal N ConSeq: the identification of functionally and structurally important residues in protein sequences Bioinformatics 2004 20 1322 1324 14871869 10.1093/bioinformatics/bth070
Elrod-Erickson M Benson TE Pabo CO High-resolution structures of variant Zif268-DNA complexes: implications for understanding zinc finger-DNA recognition Structure 1998 6 451 464 9562555 10.1016/S0969-2126(98)00047-1
Doyle DA Lee A Lewis J Kim E Sheng M MacKinnon R Crystal structures of a complexed and peptide-free membrane protein-binding domain: molecular basis of peptide recognition by PDZ Cell 1996 85 1067 1076 8674113 10.1016/S0092-8674(00)81307-0
|
16194281
|
PMC1261162
|
CC BY
|
2021-01-04 16:27:27
|
no
|
BMC Bioinformatics. 2005 Sep 30; 6:240
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-240
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2421620212910.1186/1471-2105-6-242Methodology ArticleClustering protein sequences with a novel metric transformed from sequence similarity scores and sequence alignments with neural networks Ma Qicheng [email protected] Gung-Wei [email protected] Richard [email protected] Joseph D [email protected] NR [email protected] Biomedical Computing, Genome and Proteome Sciences, Novartis Institutes for BioMedical Research, Inc. Cambridge, MA 02139 USA2005 3 10 2005 6 242 242 12 4 2005 3 10 2005 Copyright © 2005 Ma et al; licensee BioMed Central Ltd.2005Ma 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 sequencing of the human genome has enabled us to access a comprehensive list of genes (both experimental and predicted) for further analysis. While a majority of the approximately 30000 known and predicted human coding genes are characterized and have been assigned at least one function, there remains a fair number of genes (about 12000) for which no annotation has been made. The recent sequencing of other genomes has provided us with a huge amount of auxiliary sequence data which could help in the characterization of the human genes. Clustering these sequences into families is one of the first steps to perform comparative studies across several genomes.
Results
Here we report a novel clustering algorithm (CLUGEN) that has been used to cluster sequences of experimentally verified and predicted proteins from all sequenced genomes using a novel distance metric which is a neural network score between a pair of protein sequences. This distance metric is based on the pairwise sequence similarity score and the similarity between their domain structures. The distance metric is the probability that a pair of protein sequences are of the same Interpro family/domain, which facilitates the modelling of transitive homology closure to detect remote homologues. The hierarchical average clustering method is applied with the new distance metric.
Conclusion
Benchmarking studies of our algorithm versus those reported in the literature shows that our algorithm provides clustering results with lower false positive and false negative rates. The clustering algorithm is applied to cluster several eukaryotic genomes and several dozens of prokaryotic genomes.
==== Body
Background
Clustering of protein sequences from different organisms has been used to identify orthologous and paralogous protein sequences, to find protein sequences unique to an organism, and to derive the phylogenetic profile for a cluster of protein sequences. These are some of the essential components of a comparative genomics study of protein sequences across several genomes.
The methods of clustering protein sequences can be either domain-based or family-based. All the clustering methods start with an all-against-all pairwise protein sequence similarity search. The domain-based clustering methods organize the protein sequence universe into domain clusters where domains are the structural units of proteins, e.g., COG [1], ProDom [2], and Picasso [3] (Figure 1). A multidomain protein may belong to multiple domain clusters.
Figure 1 The schematic view of family-based clustering. Figure 1 illustrates a typical example of the clustering of three protein families denoted by the three oval outlines. Family I consists of protein sequences 1 and 2. Family II consists of protein sequences 3, 4, and 5. Family III consists of protein sequence 6 and 7. Domain A is common to families 1 and 2 while Domain B is common to families 2 and 3.
Clusters of Orthologous Groups (COGs) find triangles of mutually consistent genome-specific best hits from distant organisms without specifying a fixed similarity cut-off, thus accommodating both fast evolving and slow evolving genes. It then merges triangles which share a common edge. Each COG cluster is further analyzed manually to eliminate false positives caused by multidomain proteins so that each COG cluster represents a domain.
ProDom is based on the assumption that short protein sequences are single domain proteins. It first sorts all the protein sequences according to their lengths. It then undergoes a repetitive process: during each iteration, ProDom chooses as the query sequence the current shortest protein sequence or its internal repeat unit if it has internal repeats, searches the whole protein sequence set with PSI-BLAST [4], builds the sequence profile, and masks segments covered by the sequence profile for multidomain proteins or removes the single domain proteins completely covered by the sequence profile. The process iterates until there is no sequence left in the protein sequence set.
Picasso merges pairwise sequence alignments from the initial all-against-all pairwise sequence similarity searches into multiple sequence alignments of closer homologs, and later hierarchically merges these multiple sequence alignments into representative sequence profiles of remote homologs by profile-profile comparisons. The representative sequence profiles may contain sequences of different domain structures, but share at least one domain. Picasso then cuts domains within the representative sequence profiles into individual domain clusters based on the concept of overlapping maximal clusters proposed in SYSTERS [5]. Maximal clusters are clusters not fully contained in any other clusters. Two maximal clusters may have not only the overlapping set of neighbour members but also the unique set of neighbour members to these two maximal clusters. Thus these two overlapping maximal clusters must be of different domain structures sharing at least one domain which corresponds to the overlapping set of neighbour members. Then these two overlapping maximal clusters must undergo domain-cutting to be split into individual domains corresponding to closed neighbours, where no member has any neighbour outside of the cluster, from multiple alignments. However, since it is still a challenging problem to precisely pinpoint the structure domain border based on primary sequence information [6,7], the performance of the clustering algorithm will be determined by the accuracy of domain demarcations.
Family-based clustering methods group protein sequences into families, which contain a group of evolutionarily related proteins that share similar domain architecture (see Figure 1), e.g., CluSTr [8], SYSTERS, ProClust[9,10], PROTONET [11]/ProtoMap [12], and MCL[13]. CluSTr clusters protein sequence with the single linkage algorithm using the Z-score as the metric.
SYSTERS uses each protein sequence in the dataset as a seed sequence and applies the single linkage algorithm with a stringent threshold. Thus, each seed sequence has a cluster associated with it. It then merges all the clusters to maximal clusters. The maximal clusters could be either separate maximal clusters corresponding to single domain protein clusters or overlapping maximal clusters representing clusters having multiple domains, but sharing at least one domain.
ProClust uses a different metric to detect whether the aligned two proteins have similar domain structures. The metric value, which scales from 0 to 1, is the ratio of the raw score of the sequence alignments to the raw score of one of those two sequences aligned to itself. Thus the metric value between two sequences is directional. It assumes that the metric is symmetric if two aligned sequences have similar domain structure and non-symmetric otherwise. It then represents each sequence as a vertex and represents the metric value above the threshold as a directional edge in a directed graph. Each strongly connected component corresponds to a cluster [9]. It was later improved by building Profile-HMMs for all clusters having more than 20 sequences and merging two clusters A and B into a cluster corresponding to a SCOP superfamily if the average E-value from searching all the sequences in the cluster A against the profile-HMM of the cluster B is below the threshold[10].
PROTONET applies the hierarchical clustering of protein sequences based on the their pairwise similarity E-values, but adopts different rules for merging clusters: arithmetic mean, geometric mean, and harmonic mean. However, different families may have different levels of sequence conservation. It is not appropriate to choose one E-value threshold. And at the level of higher E-value, it may merge two clusters of different domain structures, but sharing one domain. However, different families may have different levels of sequence conservation. It is not appropriate to choose one E-value threshold. And at the level of higher E-value, it may merge two clusters of different domain structures, but sharing one domain.
Transitive homology detection methods have been proposed in the Intermediate Sequence Search, ISS [14,15], and [16]. It works by searching the query sequence against the database with a conservative threshold to find the closely homologous sequences and using these homologous sequences as seeds to search the database to find remotely homologous sequences with a less conservative threshold. The method has been shown to be close to the profile based methods and better than a direct pairwise homology search [17]. But, it is a challenge to quantify the indirect, transitive homology as opposed to using the E-value for quantifying direct pairwise sequence homology.
The Markov cluster (MCL) [13] algorithm has been successfully applied to clustering protein sequences. MCL represents protein sequences as nodes on a graph where similar proteins are connected by edges weighted according to their BLASP E-Value. The MCL algorithm works through a series of iterative random walks across the graph and inflations of the edge weights that gradually strengthens the connections between very similar nodes and weakens the connections between less similar nodes. MCL makes no explicit use of protein domain architecture but does leverage transitive homologies in the random walk phase of the algorithm.
Compared to the hierarchical clustering family based clustering method, e.g., PROTONET [11], our method can take advantage of the transitive homologue closure by the third intermediate sequence to detect remote homologues at the superfamily level. Compared to single linkage based methods, e.g., CluSTr [8], our method avoids the problem of merging two unrelated multi-domain cluster sharing a common domain. Compared to the iterative clustering method, e.g., SYSTERS [5], our method generates clusters where each sequence belongs to only one cluster.
Results and discussion
Benchmarking
In order to test the performance of CLUGEN, we selected all Swissprot [18] sequences with an InterPro [19] annotation, which resulted in 41480 sequences from 1598 InterPro families. The criteria we used to select sequences are that more than one member database from the InterPro annotation have the same superfamily or domain assignment and that the aligned region of the Swissprot sequence with respect to either Profile or hidden Markov model is longer than 30 amino acids. The benchmarking dataset is available on request. Figure 2 shows the InterPro superfamily/domain size distribution in the benchmarking dataset. There are 102 singleton families, that is families that consist of only one sequence. The largest family is IPR000276, the Rhodopsin-like G-Protein Coupled Receptor (GPCR) family which comprised of 1058 protein sequences.
Figure 2 shows the distribution of InterPro family size. Figure 2 shows the distribution of the InterPro families used in the benchmarking dataset based upon the number of members in each family. There are 102 singleton InterPro families, and the largest InterPro family in the benchmarking dataset is Rhodopsin-like GPCR superfamily which has 1058 protein sequences in the benchmarking dataset.
Performance measure
We measure CLUGEN's performance by sensitivity, specificity, and goodness. A protein sequence is a false positive (FP) if it is misclassified to a certain InterPro superfamily/domain and a true positive (TP) otherwise. A protein sequence in a certain InterPro superfamily/domain is a false negative (FN) if it is not classified to that InterPro superfamily/domain (Figure 3). Let Nfp and Ntp denote the number of false positives and the number of true positives with respect to a cluster. Let N fn denote the number of false negatives with respect to a InterPro superfamily/domain.
Figure 3 Definition of various clustering parameters. Figure 3 illustrates the mapping of three generated clusters denoted by oval outlines differentiated by different colors into a InterPro family denoted by a rectangle. The cluster can be mapped to an InterPro family only if more than 50% cluster members belong to that InterPro family; and is declared as a orphan cluster otherwise. Protein sequences outside the rectangle are false positives. Protein sequences within both the oval outline and the rectangle are true positives. Protein sequences wholly within the grey rectangle are false negatives.
Specificity: The specificity of a cluster is defined as Ntp / (Ntp + Nfp).
Sensitivity: The sensitivity of an InterPro superfamily/domain is defined as Ntp / (Ntp + Nfn ).
Goodness: The goodness of an InterPro superfamily/domain is a measure of how well a cluster corresponds to the mapped InterPro superfamily/domain and is defined below (Equation 1) where N denotes the number of generated clusters associated with that InterPro superfamily/domain. The Area Under the ROC Curve (AUC) has been shown to be a better evaluation measure than accuracy within the context of binary classification, where the negative dataset is clearly defined. However, we cluster protein sequences into 1598 interpro families simultaneously. As a result, using the AUC as a measure of performance is not the appropriate metric here. Instead, we adopt as the "goodness " the set relative measure as defined in [12]. In order to decrease the goodness value when a large number of clusters is associated with an InterPro superfamily/domain, a penalty of (N-1) is applied in the numerator of the equation.
Ideally specificity, sensitivity, and goodness should be 100%.
Equation 1:
Goodness=∑i=1NNtpi−N+1∑i=1NNtpi+∑i=1NNfpi+Nfn
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGhbWrcqWGVbWBcqWGVbWBcqWGKbazcqWGUbGBcqWGLbqzcqWGZbWCcqWGZbWCcqGH9aqpdaWcaaqaamaaqahabaGaemOta40ccqWG0baDcqWGWbaCdaWgaaqaaiabdMgaPbqabaGccqGHsislcqWGobGtcqGHRaWkcqaIXaqmaSqaaiabdMgaPjabg2da9iabigdaXaqaaiabd6eaobqdcqGHris5aaGcbaWaaabCaeaacqWGobGtliabdsha0jabdchaWnaaBaaabaGaemyAaKgabeaakiabgUcaRmaaqahabaGaemOta40ccqWGMbGzcqWGWbaCdaWgaaqaaiabdMgaPbqabaGccqGHRaWkcqWGobGtliabdAgaMjabd6gaUbqaaiabdMgaPjabg2da9iabigdaXaqaaiabd6eaobqdcqGHris5aaWcbaGaemyAaKMaeyypa0JaeGymaedabaGaemOta4eaniabggHiLdaaaaaa@66DA@
Overall performance
We evaluated CLUGEN at several threshold values. Table 1 lists shows the specificity, sensitivity, and goodness as well as the number of generated clusters and the number of orphan clusters as a function of the different threshold values respectively. A cluster can be mapped to an InterPro family only if more than 50% of the cluster members belong to that InterPro family; and is declared as an orphan cluster otherwise. At one extreme of the spectrum, each cluster is a singleton cluster consisting of only one protein sequence. Thus both specificity and sensitivity are 100%. But the goodness value is very low, the reciprocal of the size of the family. Clustering based on more stringent threshold values, e.g. 0.5, generates a larger number of smaller clusters causing a smaller number of false positives, also resulting in a low goodness value. As the threshold values become less stringent, small clusters merge at different levels into larger clusters, therefore a smaller number of larger clusters are generated. At the threshold of 0.2, there are 1706 clusters resulting in a specificity of 97.1%, sensitivity 98.6%, goodness value of 78.2%, and the number of orphan clusters is 201. As can be seen from table 1, the threshold value is a compromise of sensitivity, specificity, goodness and the number of orphan clusters. Ideally, we would like the clustering results to produce as many clusters as there should be and as few orphan clusters as possible.
Table 1 Specificity, sensitivity, goodness, cluster number, and orphan cluster values at different cutoff values on the benchmarking dataset.
cutoff specificity sensitivity goodness cluster number Number of orphan clusters
0.20 97.11% 98.60% 78.20% 1706 201
0.22 97.37% 98.70% 78.00% 1742 180
0.25 97.61% 98.70% 77.60% 1786 157
0.29 97.85% 98.70% 76.90% 1837 133
0.33 98.06% 98.90% 76.30% 1896 107
0.40 98.43% 99.00% 75.00% 1972 79
0.50 98.70% 99.10% 72.60% 2073 59
For a basis of comparison we also applied the MCL [13] algorithm to the same test dataset with various inflation values. Results are depicted in Figures 4 and 5. At higher specificities, the sensitivity of both methods increases. This is expected because higher specificities are achieved via stricter thresholds that result in more clusters overall and fewer large clusters. In the extreme case one could place each test sequence in its own cluster of size 1 and achieve 100% sensitivity and 100% specificity but with a low goodness score. This trade-off between sensitivity, specificity, and goodness is clearly evident in Figure 4; as specificity increases, sensitivity increases whereas goodness decreases.
Figure 4 Specificity, sensitivity, and goodness on the benchmarking dataset. Sensitivity and specificity for CLUGEN and MCL at various specificities. At higher specificities, the sensitivity of both methods increases, whereas the goodness of both methods decreases. This is expected because higher specificities are achieved via stricter parameter thresholds that more clusters overall and fewer large clusters. Performance for both methods is comparable in this range with CLUGEN performing better at lower specificities and MCL performing better at higher specificities.
Figure 5 The number of generated clusters and orphan clusters on the benchmarking dataset. Total clusters and orphan clusters for clugen and MCL at various specificities. With stricter parameter thresholds, overall specificity and the total number of clusters increases for both methods. The larger number of small clusters at higher specificities leads to a reduction in the number of orphan clusters in both methods.
In Figure 5 we see additional tradeoffs between specificity and overall performance. As specificity increases the number of orphan clusters decreases. This improvement in performance comes with an increase in the total number of clusters. Once again the extreme case of one sequence per cluster guarantees no orphan clusters at the cost of many non-informative clusters. Ideally one wishes to strike a balance reducing the number of orphan clusters while not drastically increasing the total number of clusters.
The overall performances of MCL and CLUGEN are fairly similar, with CLUGEN demonstrating a clear advantage at specificities below 0.98. CLUGEN's sensitivity and goodness are better at specificities below 0.98, whereas MCL's goodness is slightly better at specificities greater than 0.98. The number of total clusters and orphan clusters generated by both methods are comparable at specificities below 0.98. CLUGEN tends toward fewer orphan clusters at the cost of more total clusters at higher specificities.
Analysis of some CLUGEN generated clusters
In this section, we will give examples of some successfully generated clusters with one-to-one correspondence to specific InterPro families, some clusters which have false positives, and some which have false negatives.
As previously outlined, 41480 sequences with Interpro superfamily annotation (1598 clusters) were clustered using our algorithm. This results in a total of 1972 clusters. Overall, there are 1199 clusters that have been correctly mapped with one-to-one correspondence to 1199 out of 1598 InterPro superfamilies /domains. There are 79 orphan clusters. Some correctly clustered large protein superfamily/domain examples are: 507 Cytochrome P450 proteins are correctly clustered into family IPR001128 without false positives and false negatives; 398 large chain ribulose bisphosphate carboxylase proteins are correctly clustered into family IPR000685 without false positives and false negatives; 290 short-chain dehydrogenase/reductase SDR proteins are clustered into family IPR002198 without false positives and false negatives. Table 2 shows top 50 InterPro superfamily/domains that have been mapped to clusters with one-to-one correspondence.
Table 2 Top 50 InterPro superfamily/domains that have been mapped to clusters with one-to-one correspondence
InterPro family/Domain ID Type Number of proteins in the benchmark dataset Description
IPR001128 Family 507 Cytochrome P450
IPR000685 Family 398 Ribulose bisphosphate carboxylase, large chain
IPR002198 Family 290 Short-chain dehydrogenase/reductase SDR
IPR004000 Family 255 Actin/actin-like
IPR002423 Family 226 Chaperonin Cpn60/TCP-1
IPR001023 Family 221 Heat shock protein Hsp70
IPR002085 Family 181 Zinc-containing alcohol dehydrogenase superfamily
IPR000173 Family 177 Glyceraldehyde 3-phosphate dehydrogenase
IPR001175 Family 169 Neurotransmitter-gated ion-channel
IPR000910 Family 169 HMG1/2 (high mobility group) box
IPR001353 Family 147 20S proteasome, A and B subunits
IPR000894 Family 141 Ribulose bisphosphate carboxylase, small chain
IPR000298 Family 135 Cytochrome c oxidase, subunit III
IPR001019 Family 135 Guanine nucleotide binding protein (G-protein), alpha subunit
IPR000568 Family 134 H+-transporting two-sector ATPase, A subunit
IPR001400 Family 133 Somatotropin hormone
IPR000883 Family 131 Cytochrome c oxidase, subunit I
IPR001364 Family 131 Hemagglutinin, HA1/HA2 chain
IPR00970 Family 130 Secreted growth factor Wnt protein
IPR001664 Family 127 Intermediate filament protein
IPR000847 Domain 127 Bacterial regulatory protein, LysR
IPR001659 Family 124 Phycobilisome protein
IPR001694 Family 123 Respiratory-chain NADH dehydrogenase, subunit 1
IPR001811 Family 119 Small chemokine, interleukin-8 like
IPR000215 Family 118 Proteinase inhibititor I4, serpin
IPR001926 Family 114 Pyridoxal-5'-phosphate-dependent enzyme, beta subunit
IPR000515 Family 113 Binding-protein-dependent transport systems inner membrane component
IPR001424 Family 112 Copper/Zinc superoxide dismutase
IPR001804 Family 111 Isocitrate/isopropylmalate dehydrogenase
IPR001691 Domain 109 Glutamine synthetase, catalytic domain
IPR000934 Domain 105 Metallophosphoesterase
IPR001189 Family 105 Manganese and iron superoxide dismutase
IPR001041 Domain 105 Ferredoxin
IPR001099 Family 104 Naringenin-chalcone synthase
IPR001450 Domain 102 4Fe-4S ferredoxin, iron-sulfur binding domain
IPR001427 Family 102 Pancreatic ribonuclease
IPR000484 Family 100 Photosynthetic reaction centre protein
IPR000954 Family 98 Aminotransferase class-III
IPR001576 Family 93 Phosphoglycerate kinase
IPR000230 Family 93 Ribosomal protein S12, bacterial and chloroplast form
IPR002068 Domain 91 Heat shock protein Hsp20
IPR001750 Domain 90 NADH/Ubiquinone/plastoquinone (complex I)
IPR000836 Domain 90 Phosphoribosyltransferase
IPR001993 Family 90 Mitochondrial substrate carrier
IPR001236 Family 85 Lactate/malate dehydrogenase
IPR002210 Family 83 Papillomavirus major capsid L1 (late) protein
IPR001395 Family 81 Aldo/keto reductase
IPR000943 Family 80 Sigma-70 factor
IPR002226 Family 80 Catalase
IPR001766 Domain 80 Fork head transcription factor
We also conducted a detailed analysis of clusters that had false negatives/false positives in order to understand the areas in which the clustering algorithm could be further improved. The following is a description of errors encountered in clustering algorithms with specific reference to the data from our method.
Errors from low-complexity and coiled-coil regions
The first type of error is due to the presence of low complexity sequences with repetitive sequence patterns or sequences with coiled-coil structures, since we mask the low complexity regions and coiled-coil regions before the all-against-all pairwise similarity searches. As an example, the InterPro family IPR000533, Tropomyosins, which regulate muscle contraction, are alpha-helical proteins that form a coiled-coil. There are 25 tropomyosin sequences in the benchmarking dataset, among which 24 tropomyosin sequences are false negative sequences and appear in the following cluster along with members of IPR002699 ATP synthase subunit D.
Errors from short sequences or from an abundance of certain amino acid type in the sequences
Short sequences with less than 70 amino acids could also cause false positives in the clustering results. Cluster 1259 which is mapped to InterPro family 003019, the metallothionein superfamily, consists of 125 short protein sequences with 68 amino acids on average in length among which 34 false positive protein sequences are from InterPro family IPR001762, Disintegrin, and 34 false positive protein sequences are from InterPro family IPR000877, Bowman-Birk serine protease inhibitor. The reason these families cluster together is that metallothioneins are small proteins with high content of cysteine residues, while disintegrins and Bowman-Birk serine protease inhibitors are also short cysteine-rich protein sequences. This suggests that a more stringent threshold should be applied to cluster short protein sequences which are rich in a particular amino acid.
Similar domain structures in different superfamilies
Sequences that belong to different superfamilies but share similar domain structures may also cluster incorrectly in some cases. For example, 1039 out of 1058 sequences correctly cluster into the IPR000276 Rhodopsin-like G-protein coupled receptors family. But 17 out of 19 sequences from the IPR000276 Rhodopsin-like G-protein coupled receptors which have Cysteine-rich N-terminal regions are mistakenly clustered into the InterPro 000372 which is annotated as a cysteine-rich flanking region N-terminal. Similarly, members of the IPR001878 CCHC Zinc finger domain have been incorrectly clustered into the cluster 1008 which is mapped to the InterPro family 000981 Neurohypophysical hormone because they share the two cysteine residues and other surrounding weak motifs.
Conclusion
This paper describes a novel method for the clustering of protein sequences based on a new metric derived from prediction using neural networks and further utilizing the metric to model the transitive sequence homologue to detect the remote homologue. Good performance with respect to the InterPro protein sequence database has been achieved on the benchmarking dataset.
Clustering of sequences has many applications in target discovery and gene functionalization. One can identify in silico, antimicrobial drug targets by examining clusters without any eukaryotic member sequence in it. These proteins could be potentially selective targets for infectious diseases due to the absence of any appreciable homolog in the human proteome. Such computationally derived targets need to be further validated using experimental data derived from gene expression profiling and proteomics experiments [20]. Another application is to predict the function for prokaryotic proteins of unknown function by phylogenetic profiling [21], where the phylogenetic profile for a cluster is a vector of binary values, with 1 meaning the presence of a genome in that cluster and 0 otherwise. The assumption here is that genes with the same phylogenetic profile could have the same function.
Method
Feature extraction
After we mask the low complexity regions and the coiled-coil regions and carry out the all-against-all pairwise sequence similarity searches, we extract four sets of features to represent the homology between any given pair of sequences. The first two sets of input features detect the homology of two aligned sequences, the last two sets of input features test whether two aligned sequences have similar domain structures. We use neural networks to map input features to a new metric, a probability value which scales from 0 to 1 and is interpreted as the likelihood that two sequences are of the same homologous superfamily.
The first input feature is the log scale of the pairwise E-value.
The raw score, from which the E-value of the two aligned sequences is derived, is calculated by summing up the log score of each position in the alignment, which assumes that each position is independent of the other. However, in practice, it has been shown that two consecutive positions in the alignments are quite correlated [22]. To model the correlation between two consecutive positions in the alignment, we adopt the concept of the 2-gram encoding method [23]. Ideally, hydrophobic regions of one sequence should align with the hydrophobic regions of the other sequence, and hydrophilic regions should align with each other as well. Each position in the alignment could fall into one of four categories: residue identity denoted by A1, hydrophobic similarity denoted by A2, hydrophilic similarity denoted by A3, and everything else denoted by A4. Let Lena denote the total length of the alignment and Occuri,j denote the number of occurrences of i and j, where i is immediately followed by j, with i and j denoting any one of A1, A2, A3, or A4 respectively. Let freqi,j denote frequency of i,j, which is equal to Occuri,j/(Lena-1). Thus the second set of input features consists of freqi,j values of the alignment positions, which consists of 16 input feature values for a pair of aligned sequences since each of the two consecutive positions could be one of four possible values.
Each of the two aligned sequences is separated into three segments, the unaligned N-terminal region, the aligned region, and the unaligned C-terminal region, by the beginning position of the alignment, denoted by begini, and the end position of the alignment, denoted by endi with respect to sequencei (Figure 6). Let Leni denote the length of sequencei. Then lengths of three segments of sequencei are begini -1, endi - begini +1, and Leni - endi, respectively. If we normalize the length of each of three segments within an aligned sequencei by dividing the length of each segment by Leni, we get a vector of three values, Segi1 = (begini-1)/Leni, Segi2 = (endi - begini +1)/Leni, and Segi3 = (Leni - endi)/Leni. Intuitively, if the two aligned sequences have similar domain structures, the alignment will divide the two aligned sequences i and j in a similar proportion, and the linear correlation coefficient, LCC1 defined by Equation 2, between these two vectors tend to be close to 1. So the third set of input features include LCC1, Segi1, Segi2, Segi3, Segj1, Segj2, and Segj3.
Figure 6 A schematic view of a pairwise alignment. Figure 6 shows a pairwise alignment between two aligned sequences. The aligned regions of the two sequences are highlighted. Their boundaries are pinpointed by the arrows.
Equation 2:
LCC1=13∑k=1N(Segik−13)(Segjk−13)13∑k=1N(Segik−13)213∑k=1N(Segjk−13)2
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@7C1C@
The fourth and final input feature is to measure the overlap between two neighbor sets of aligned sequences, where the neighbor set, Seti of sequencei, is defined as the set of protein sequences that sequencei hits when sequencei is used as the query sequence. One straightforward method to measure the overlap is to use the ratio of the cardinality of the intersection of two neighbor sets to the cardinality of the union of two neighbor sets.
Here we propose another method to measure the overlap. Specifically, if we represent the neighbor set of sequencei as Vectori, the value of the kth element of Vectori is the log E-value, Logik between sequencei and its kth neighbor in Seti. However, Vectori and Vectorj for two aligned sequences, sequencei and sequencej, may be of different dimensions since the cardinalities of Seti and Setj may be different. We make these two vectors have the same dimension by adding the log E-value threshold to Vectori whenever the sequencei has no corresponding neighbor in the neighbor set, Setj of the other aligned sequencej. Thus the last input feature is the linear correlation coefficient LCC2 between Vectori and Vectorj defined by Equation 3. Intuitively, the more similar the domain structure two aligned sequences have, the more similar neighbor sets they will have, and the closer to 1 the linear correlation coefficient will be.
Equation 3:
LCC2=1N∑k=1N(Logik−Logi)(Logjk−Logj)1N∑k=1N(Logik−Logi)21N∑k=1N(Logjk−Logj)2
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@8ABD@
where N is the dimension of Vectori and Vectorj and Logi and Logj are the mean values and are defined by Logi=1N∑k=1NLogik
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGmbatcqWGVbWBcqWGNbWzdaWgaaWcbaGaemyAaKgabeaakiabg2da9maalaaabaGaeGymaedabaGaemOta4eaamaaqahabaGaemitaWKaem4Ba8Maem4zaC2aaSbaaSqaaiabdMgaPjabdUgaRbqabaaabaGaem4AaSMaeyypa0JaeGymaedabaGaemOta4eaniabggHiLdaaaa@42BC@, and Logj=1N∑k=1NLogjk
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGmbatcqWGVbWBcqWGNbWzdaWgaaWcbaGaemOAaOgabeaakiabg2da9maalaaabaGaeGymaedabaGaemOta4eaamaaqahabaGaemitaWKaem4Ba8Maem4zaC2aaSbaaSqaaiabdQgaQjabdUgaRbqabaaabaGaem4AaSMaeyypa0JaeGymaedabaGaemOta4eaniabggHiLdaaaa@42C0@.
To summarize, the first input feature is the log scale of the pairwise E-value. The second input feature consists of 16 feature values, which are frequency values of the alignment positions and the third input feature includes 7 values, which relate to the details of the alignments. And finally, the fourth input feature includes 1 feature value which measures the overlap of the neighbour sets to make a total of 25 input features which will be used to train the neural network as described below.
Neural networks
After we represent the sequence homology between a pair of sequences by the set of 25 input features, we train the neural network using the training data. Each homologous pair of sequences is labeled as 1 if they belong to the same InterPro superfamily or the same domain if they are single domain proteins, and 0 otherwise. We selected as large a number of sequences as possible to train the neural networks to avoid overfitting the data. In all, we selected 27844 homologous sequence pairs as the positive training set and 29999 non-homologous sequence pairs as the negative training set.
The neural network we use is a fully connected feed-forward back propagation neural network and has one hidden layer with sigmoid activation functions (see Figure 7). The output layer of the neural network has one output unit. The output value is bounded between 0 and 1. The network is trained with the scaled conjugate gradient algorithm [24] implemented in MATLAB [25].
Figure 7 The architecture of the neural network. Figure 7 demonstrates the architecture of the neural network. The neural network is actually fully connected, but not shown in the figure for simplicity, and has three layers. The first layer is the input layer consisting of 25 input features. The hidden layer in the middle has 4 nodes. The output layer has one output node.
Given the large amount of training data relative to the number of the weights in the neural network, the neural network is unlikely to overfit. It may however underfit if there are not a sufficient number of weights. If the training data are smaller relative to the number of weights in the neural network, measures should be taken to avoid the overfitting problem and the cross-validation method should be used to choose the best model. Clearly in this study, such is not the case.
We used a split-sample approach in which the validation set is not used during training, but is used to select the best model. After the neural network is trained, it is validated on the validation dataset containing 250597 homologous pairs of sequences and 30000 non-homologous pairs. Different numbers of hidden layer nodes have been tested. The ultimate goal is not to determine if any two proteins sequences belong to the same Interpro family, but to cluster all sequences in Interpro families as accurately as possible. We selected the model with the smallest number of weights and smallest error on the validation set. Thus, we chose 4 hidden layer nodes such that the neural network has the least number of hidden units and the best performance on the validation dataset with a specificity of 94.18% and a sensitivity of 91.81%.
Modeling the transitive homology
The neural network is then used to calculate the metric value for any pair of protein sequences that hit to each other below the E-value that was used as a cutoff. The metric value, P(A,B), for protein sequences A and B is interpreted as the likelihood that these two protein sequences belong to the same InterPro superfamily or have the same single domain. The value P(A,B) is replaced by P(A,C)P(C,B) if there exists a sequence C such that P(A,C)P(C,B) is larger than the current value of P(A,B). This transformation takes advantage of the transitive homology of sequences A and B through the intermediate sequence C, assuming that protein sequences A and C and protein sequences B and C are independently homologous. Figure 8 illustrates the transitive homology between sequence a and sequence b through the third sequence c. The E-values between sequence a and sequence c, sequence c and sequence b, as well as sequence a and sequence b are 0.01, 0.005, 20 respectively. P(a,c), P(c,b), and P(a,b) are 0.8, 0.9, and 0.2 respectively. The homology between sequence a and sequence b cannot be detected with their direct E-value. However, the value of P(a,b) is assigned to 0.72 because of the transitive sequence homology.
Figure 8 Figure 8 illustrates the transitive homology between sequence a and sequence b through the third sequence c. The homology between sequence a and sequence b can be detected with P(a,b) = 0.72 by the transitive sequence homology.
Hierarchical average linkage clustering
Once the metric value for every pair of protein sequences is calculated, the hierarchical average linkage clustering method is applied to cluster the protein sequences in the new metric space using the geometric mean as the merging rule.
Hierarchical average linkages uses the Unweighted Pair-Group Average (UPGA) [26] to measure the distance between clusters. Let Di, i = 1,2, ... n. denote the protein sequences contained in Cluster D and let Ej, j = 1,2, ..., m denote the protein sequence contained in Cluster E. The geometric mean distance G between Cluster D and Cluster E is defined as Equation 4:
Equation 4:
G=∏i=1,j=1i=n,j=mP(Di,Ej)
MathType@MTEF@5@5@+=feaafeart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGhbWrcqGH9aqpdaqeWbqaaiabdcfaqjabcIcaOiabdseaeTGaeeyAaKMccqGGSaalcqWGfbqrliabdQgaQPGaeiykaKcaleaacqWGPbqAcqGH9aqpcqaIXaqmcqGGSaalcqWGQbGAcqGH9aqpcqaIXaqmaeaacqWGPbqAcqGH9aqpcqWGUbGBcqGGSaalcqWGQbGAcqGH9aqpcqWGTbqBa0Gaey4dIunaaaa@49A8@
The hierarchical average linkage clustering works in an iterative process: it begins with each protein sequence as a singleton cluster; during each iteration, it finds two clusters with the lowest metric value, then joins these two clusters into a new cluster, and updates the metric value between this new cluster and all others. This process iterates until the current lowest metric value exceeds the threshold.
Authors' contributions
QM carried out the design and implementation of the method and wrote the manuscript. JDS compared the performances between CLUGEN and MCL. GWC and RC participated in the project. NRN directed and participated in the project and prepared the figures in the manuscript. All authors involved in reviewing and revising the manuscript and approved the final manuscript.
Acknowledgements
We thank Dr. Xia Yang for comments and feedback. We also acknowledge Mischa Reinhardt and Darrell Ricke for a critical reading of this manuscript.
==== Refs
Tatusov RL Koonin EV Lipman DJ A genomic perspective on protein families Science 1997 278 631 637 9381173
Gouzy J Corpet F Kahn D Whole genome protein domain analysis using a new method for domain clustering Comput Chem 1999 23 333 340 10404623
Heger A Holm L Picasso: generating a covering set of protein family profiles Bioinformatics 2001 17 272 279 11294792
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
Krause A Vingron MA set-theoretic approach to database searching and clustering Bioinformatics 1998 14 430 438 9682056
George RA Heringa J SnapDRAGON: a method to delineate protein structural domains from sequence data J Mol Biol 2002 316 839 851 11866536
Nagarajan N Yona G A multi-expert system for the automatic detection of protein domains from sequence information Proceedings of the seventh annual international conference on Computational molecular biology 2003 Berlin, Germany 224 234
Kriventseva EV Fleischmann W Zdobnov EM Apweiler R CluSTr: a database of clusters of SWISS-PROT+TrEMBL proteins Nucleic Acids Res 2001 29 33 36 11125042
Bolten E Schliep A Schneckener S Schomburg D Schrader R Clustering protein sequences–structure prediction by transitivehomology Bioinformatics 2001 17 935 941 11673238
Pipenbacher P Schliep A Schneckener S Schonhuth A Schomburg D Schrader R ProClust: improved clustering of protein sequences with an extended graph-based approach Bioinformatics 2002 S182 191 12386002
Sasson O Linial N Linial M The metric space of proteins-comparative study of clustering algorithms Bioinformatics 2002 S14 21 12169526
Yona G Linial N Linial M ProtoMap: automatic classification of protein sequences, a hierarchy of protein families, and local maps of the protein space Proteins 1999 37 360 378 10591097
Enright AJ Van Dongen S Ouzounis CA An efficient algorithm for large-scale detection of protein families Nucleic Acids Res 2002 30 1575 84 11917018
Teichmann SA Chothia C Church GM Park J Fast assignment of protein structures to sequences using the intermediate sequence library PDB-ISL Bioinformatics 2000 16 117 124 10842732
Park J Teichmann SA Hubbard T Chothia C Intermediate sequences increase the detection of homology between sequences J Mol Biol 1997 273 349 354 9367767
Gerstein M Measurement of the effectiveness of transitive sequence comparison, through a third 'intermediate' sequence Bioinformatics 1998 14 707 714 9789096
Park J Karplus K Barrett C Hughey R Haussler D Hubbard T Chothia C Sequence comparisons using multiple sequences detect three times as many remote homologues as pairwise methods J Mol Biol 1998 284 1201 1210 9837738
Boeckmann B The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003 Nucleic Acids Res 2003 31 365 370 12520024
Zdobnov EM Apweiler R InterProScan – an integration platform for the signature-recognition methods in InterPro Bioinformatics 2001 17 847 848 11590104
Cole ST Comparative mycobacterial genomics as a tool for drug target and antigen discovery Eur Respir J Suppl 2002 36 78s 86s 12168750
Pellegrini M Marcotte EM Thompson MJ Eisenberg D Yeates TO Assigning protein functions by comparative genome analysis: protein phylogenetic profiles Proc Natl Acad Sci U S A 1999 96 4285 4288 10200254
Pei J Grishin NV AL2CO: calculation of positional conservation in a protein sequence alignment Bioinformatics 2001 17 700 712 11524371
Wang TJ Ma Q Shasha D Wu C New techniques for extracting features from protein sequences IBM Systems Journal, Special Issue on Deep Computing for the Life Sciences 2001 40 426 441
Bishop CM Neural Networks for Pattern Recognition 1995 Oxford University Press, New York, New York
Hanselman DC Mastering MATLAB 5: A comprehensive tutorial and reference 1998 Prentice Hall, Upper Saddle River, New Jersey
Hastie T Tibshirani R Friedman J The Elements of Statistical Learning 2001 Springer, New York
|
16202129
|
PMC1261163
|
CC BY
|
2021-01-04 16:27:27
|
no
|
BMC Bioinformatics. 2005 Oct 3; 6:242
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-242
|
oa_comm
|
==== Front
PLoS Comput BiolPLoS Comput. BiolpcbiplcbploscompPLoS Computational Biology1553-734X1553-7358Public Library of Science San Francisco, USA 1624470410.1371/journal.pcbi.0010049plcb-01-05-04Research ArticleStructural Evolution of the Protein Kinase–Like Superfamily Structural Evolution of KinasesScheeff Eric D 1*Bourne Philip E 121 San Diego Supercomputer Center, University of California, San Diego, California, United States of America
2 Department of Pharmacology, University of California, San Diego, California, United States of America
Thornton Janet EditorEuropean Bioinformatics Institute, United Kingdom* To whom correspondence should be addressed. E-mail: [email protected] 2005 21 10 2005 8 9 2005 1 5 e4917 3 2005 8 9 2005 Copyright: © 2005 Scheeff and Bourne.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 protein kinase family is large and important, but it is only one family in a larger superfamily of homologous kinases that phosphorylate a variety of substrates and play important roles in all three superkingdoms of life. We used a carefully constructed structural alignment of selected kinases as the basis for a study of the structural evolution of the protein kinase–like superfamily. The comparison of structures revealed a “universal core” domain consisting only of regions required for ATP binding and the phosphotransfer reaction. Remarkably, even within the universal core some kinase structures display notable changes, while still retaining essential activity. Hence, the protein kinase–like superfamily has undergone substantial structural and sequence revision over long evolutionary timescales. We constructed a phylogenetic tree for the superfamily using a novel approach that allowed for the combination of sequence and structure information into a unified quantitative analysis. When considered against the backdrop of species distribution and other metrics, our tree provides a compelling scenario for the development of the various kinase families from a shared common ancestor. We propose that most of the so-called “atypical kinases” are not intermittently derived from protein kinases, but rather diverged early in evolution to form a distinct phyletic group. Within the atypical kinases, the aminoglycoside and choline kinase families appear to share the closest relationship. These two families in turn appear to be the most closely related to the protein kinase family. In addition, our analysis suggests that the actin-fragmin kinase, an atypical protein kinase, is more closely related to the phosphoinositide-3 kinase family than to the protein kinase family. The two most divergent families, α-kinases and phosphatidylinositol phosphate kinases (PIPKs), appear to have distinct evolutionary histories. While the PIPKs probably have an evolutionary relationship with the rest of the kinase superfamily, the relationship appears to be very distant (and perhaps indirect). Conversely, the α-kinases appear to be an exception to the scenario of early divergence for the atypical kinases: they apparently arose relatively recently in eukaryotes. We present possible scenarios for the derivation of the α-kinases from an extant kinase fold.
Synopsis
Most proteins have distinct three-dimensional structures that determine much of their functional capability. Proteins that are related usually have similar structures, owing to their shared genetic heritage and (often) similar function. Hence, one can speak of “families” of proteins that at one time all shared a common ancestor gene, but have diverged over eons of evolution into distinct forms with similar but altered sequences. In some cases, this sequence divergence can occur to the point that the structures of the proteins actually begin to change, forming “superfamilies” of distantly related proteins. Traditionally, events in protein evolution are investigated through the construction of evolutionary trees based on similarity between protein sequences. However, at the superfamily level sequence similarity weakens to the point that building accurate trees becomes much more problematic. This work attempts to address this problem by integrating structural similarity information into the analysis. Because protein structure changes much more slowly than sequence, structural similarity provides powerful signals about the relationships between proteins. When this new form of tree is considered alongside other evolutionary information, the authors are able to provide a supportable history for much of the evolution of the important protein kinase–like superfamily.
Citation:Scheeff ED, Bourne PE (2005) Structural evolution of the protein kinase–like superfamily. PLoS Comput Biol 1(5): e49.
==== Body
Introduction
A protein superfamily has been defined as a group of proteins that share structure, sequence, and functional features that strongly suggest they are all derived from the same common ancestor protein [1]. However, because protein sequences are highly degenerate, protein superfamily relationships are often not detectable from sequence information alone [2,3]. Protein superfamily relationships often have become apparent when structures of proteins were solved experimentally, only to reveal surprising structural similarities with known structures (e.g., [4]). Hence, structural information provides the gateway through which superfamily-level relationships may be studied. The Structural Classification Of Proteins (SCOP) database classifies proteins hierarchically, based on a tiered class, fold, superfamily, and family system [1]. The superfamilies within the SCOP database are divided up into distinct families of more closely related proteins. Protein families usually display clear sequence similarity and highly similar structures. Hence the “protein landscape” contains families of closely related proteins that share distant common ancestry with other families, forming superfamilies.
The Ser/Thr and Tyr protein kinases are a family of proteins that act as important arbiters of signal transduction in eukaryotes [5–7], and many prokaryotes [8–11]. With the determination of the first protein kinase structure [12], it became possible to place the distinctive protein kinase catalytic core motif into a structural context. The determination of additional kinase structures enforced the notion that the basic fold of the protein kinase catalytic core was structurally well conserved, and had been reused across long evolutionary timescales in a largely intact form [13].
The protein kinases exert control over their protein targets by covalent modification of a Ser, Thr, or Tyr residue with the γ-phosphate group cleaved from ATP. All of the typical protein kinases (TPKs) share a common catalytic core consisting of a small, mostly β-sheet, N-terminal subdomain and a larger, mostly α-helical, C-terminal subdomain [13] (Figure 1). The ATP binding pocket sits in a cleft between these two subdomains, which can rotate into “open” and “closed” conformations depending on ATP binding and the activation state of the molecule [14–16]. The residues involved in the phosphotransfer reaction sit at the outside edge of the ATP binding region and are highly conserved [13,17].
Figure 1 Two Views of the Structure of PKA [70]
The structure consists of two subdomains: a small, primarily β-sheet N-terminal subdomain, and a larger, primarily helical C-terminal subdomain. ATP and metal ions are bound in the cleft between the two subdomains. The small left-side view depicts PKA in the “standard” orientation used by the authors when the structure was initially solved [12], and in many subsequent publications. The larger view on the right side depicts PKA in an “open-book” format that makes structural features in the two subdomains easier to compare between families. The open-book view is achieved by rotating the standard view 90° about the vertical axis, then splitting the two subdomains at the linker region and rotating each 90° in opposite directions about the horizontal axis. Helical secondary structures (both α-helices and 3–10 helices) are depicted as cylinders, and β-strands are depicted as arrows. Elements are labeled according to the standard conventions for PKA. Some secondary structure (particularly 3–10 helices) is not labeled in the standard PKA convention, and so is unlabeled here. One structure (Helix 1) was named by us (see text). Underlined labels belong to helical structures; non-underlined labels belong to β-strands. Secondary structure elements are colored according to their conservation status in the overall superfamily as follows: yellow, elements are part of the “universal core” seen in all kinases in the superfamily; orange, elements are present in more than two, but not all, of the kinases in the superfamily; purple, elements seen only in this family, but inserted within in the portion of the chain forming the universal core; blue, elements seen only in this family, and connected to the N- or C-terminal ends of the universal core. A bound pseudosubstrate inhibitor (PKI) is present in the structure [12], and depicted in gray. This inhibitor likely describes the binding location of actual substrates of PKA. The bound ATP molecule is rendered as a ball-and-stick model, while the bound Mg ions are rendered as gray spheres. The ATP and Mg ions are duplicated in mirror image and shown interacting with both the N- and C- terminal subdomains in the open-book rendering. The most critical and highly conserved residues in PKA (and the broader superfamily) are shown as ball-and-stick models in green, and labeled according to the standard PKA numbering scheme. In addition, the glycine-rich loop is also depicted in green, though individual glycine residues are not shown. The loop that forms the linker region between the subdomains is depicted in red. Other loops within the universal core are shown in white, except for loops linking purple regions (which are shown in purple), and loops outside of the universal core (shown in blue). Key loops described extensively in the text are labeled. For increased clarity, residues 300–350 have been removed from the C-terminus of PKA. This loop region is unique to PKA, and would have been colored blue if present in the figure. Molecular renderings in this figure were created with MOLSCRIPT [90].
With the acceleration in the rate of deposition to the Protein Data Bank (PDB) [18], a large complement of sequence-divergent TPK structures have become available, and make a more comprehensive structural study of this family possible. Additionally, several structures of distant TPK relatives have become available [19–24]. These atypical kinases (AKs) are phosphotransferases that clearly share homology with the TPK catalytic core, but do not conserve all of the usual kinase motifs, and modify the initial notions of the “essential” fold characteristics of protein kinase–like phosphotransferases. While they are termed “atypical” relative to the TPKs, the AKs often represent relatively large families of important proteins (an overview of the structures of the catalytic cores of the AKs is provided in Figure 2, and summary information is provided in Table 1).
Figure 2 Views of Structural Representatives from Six Families in the Kinase-Like Superfamily Other Than the TPKs
Structures are shown in an open-face view, and using the same conventions as used for PKA in Figure 1. ATP and metal ions are shown in mirror image where available in the structure. Similar to Figure 1, secondary structural elements are colored according to their conservation status in the overall superfamily as follows: yellow, elements are part of the “universal core” seen in all kinases in the superfamily; orange, elements are present in more than two, but not all, of the kinases in the superfamily; red, elements shared between only two families; purple, elements seen only in this family, but inserted within in the portion of the chain forming the universal core; blue, elements seen only in this family, and connected to the N- or C-terminal ends of the universal core. Secondary structural elements are labeled according to the standard conventions for the individual structure. As in Figure 1, the glycine-rich loop is rendered in green and the loop forming the linker region is rendered in red. For clarity, the conserved residues shown in Figure 1 are not rendered in these structures, though in most cases they are similar. Structures shown are as follows: (A) aminoglycoside phosphotransferase (APH(3′)-IIIa [24]); (B) CK (CKA-2 [23]); (C) ChaK [20]; (D) PI3K [21]; (E) AFK [22]; and (F) PIPKIIβ [19]. Molecular renderings in this figure were created with MOLSCRIPT [90].
Table 1 Kinase Structures Included in the Representative Set
Figure 3 Enhanced Sequence Alignment Derived from the Structural Alignment of Kinase Representatives
Abbreviated names of kinase representatives are provided with the gray box at the left-hand side of the figure (see Table 1 for more information on structures). The name is followed by the PDB ID [18] for the structure used in the alignment. The number in parenthesis following the PDB ID is the residue number of the first residue shown in the alignment. The sequences of the six AKs are clustered at the top of the alignment, followed by the sequence of PKA, which is highlighted. The alignment is annotated for key structural features using the JOY software [78]. Secondary structure is represented using the following conventions: light-gray box, β-strand; medium-gray box, 3–10 helix; dark-gray box, α-helix. Residue characteristics are represented using the following conventions: uppercase, solvent inaccessible; lowercase, solvent accessible; bold, hydrogen bond to main chain amide; underline, hydrogen bond to main chain carbonyl; tilde, hydrogen bond to other side-chain; italic, positive Φ; breve, cis-peptide. Residues that are highly conserved within the TPK family and some AKs are highlighted in boxes for the sequences where the conservation applies. The residue(s) seen at these positions are shown in uppercase above the boxes. The letter O stands for general hydrophobicity, but not a specific residue type. Residues that are more weakly conserved in the TPKs but are also conserved in many other AK families are noted with a lowercase letter above the appropriate alignment columns. Selected residues of interest that are conserved only within the TPKs are depicted using the same conventions above, but with gray lettering (depiction of residues conserved only in the TPKs is not exhaustive, i.e., only residues discussed in the text are highlighted above the alignment. Generally, this is done in structural regions unique to the TPKs). Secondary structures are labeled with the nomenclature used for PKA [12]. Sequence representing unresolved portions of the structure is not shown by JOY. In key portions of the alignment, this sequence is added back in and shown in light gray.
Figure 3 (continued) Enhanced Sequence Alignment Derived from the Structural Alignment of Kinase Representatives
Figure 3 (continued) Enhanced Sequence Alignment Derived from the Structural Alignment of Kinase Representatives
Figure 3 (continued) Enhanced Sequence Alignment Derived from the Structural Alignment of Kinase Representatives
The aminoglycoside phosphotransferase APH(3′)-IIIa is a kinase that phosphorylates several aminoglycoside antibiotics at the 3′ and/or 5′′ hydroxyl, inactivating them [25]. Though the structure of this enzyme has clear similarities to that of the TPKs, it also has distinct structural motifs, particularly in the C-terminal subdomain [4] (Figure 2).
Choline kinase (CK) participates in the pathway that eventually produces phosphatidylcholine, an important constituent of cell membranes that can be cleaved to produce a variety of second messengers [26]. The available structure is of choline kinase isoform A-2 (CKA-2) from Caenorhabditis elegans [23]. This structure has a very large and complex C-terminal domain, with features distinct from those of the TPKs (Figure 2).
Channel kinase (ChaK) is a protein kinase domain that is an integral part of a transient receptor potential channel. ChaK is a representative of the α-kinase family, a small but important kinase family that has no detectable sequence similarity to the TPKs [27]. The α-kinases are so named because they appear to phosphorylate residues within α-helices [28], as opposed to the loop-type regions targeted by the TPKs [29]. ChaK has a relatively similar N-terminal subdomain to that of the TPKs, but its C-terminal domain is extensively modified [20] (Figure 2).
Phosphoinositide 3-kinases (PI3Ks) phosphorylate various forms of phosphatidylinositol (PI) at the 3-hydroxyl position. The available PI3K structure [21] is that of PI3Kγ, a “class IB” PI3K that preferentially phosphorylates phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2], creating phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P3] [30]. PI(3,4,5)P3 is an important second messenger that activates a variety of pathways in cells [31]. Relative to the TPKs, PI3K has a somewhat “flat-faced” architecture, with a more open active-site region (Figure 2). This structure allows it (in concert with accessory domains) to interact directly with the plasma membrane and phosphorylate PI in situ [21].
Actin-fragmin kinase (AFK) is a Thr protein kinase that has been isolated from the slime mold Physarum polycephalum, and at present has been detected in only this one organism. It phosphorylates actin when it is bound to the protein fragmin, helping to render control over actin polymerization [32]. Though this enzyme is clearly homologous to the TPKs, it has a modified N-terminal subdomain and an extensively modified C-terminal subdomain (Figure 2). The modifications in the C-terminal domain produce a flattened substrate binding region that allows for binding to the target actin molecule [22].
Type IIβ phosphatidylinositol phosphate kinase (PIPKIIβ) phosphorylates phosphatidylinositol 5-phosphate (PI5P) at the 4-hydroxyl position to generate PI(4,5)P2. PI(4,5)P2 is an important second messenger in cells [33], and can be further phosphorylated by PI3K as described above. The enzyme forms a homodimer that displays a highly flat-faced architecture with large patches of positively charged residues. This structure appears to allow PIPKIIβ to interact directly with the cell membrane, phosphorylating PI5P in situ [19]. PIPKIIβ is a structurally divergent enzyme that is not actually within the protein kinase–like superfamily as defined by SCOP. PIPKIIβ has almost no sequence similarity, and weak structural similarity, to the protein kinase–like superfamily. For this reason, it is in a different fold grouping in the SCOP hierarchy (d.143.1, as opposed to d.144.1). However, a careful study has linked this structure to the protein kinase–like superfamily through comparative structure analysis [34]. Cheek et al. have provided a comprehensive classification for all kinases, including the many superfamilies without any evolutionary relationship to the protein kinase–like superfamily (when the term “kinase” is used in this work, it refers specifically to members of the protein kinase–like superfamily) [35,36]. Unlike SCOP, they have placed the PIPK family within the same fold group as the kinase superfamily. Also, PIPKIIβ appears to share a similar catalytic mechanism to that of the kinases. Therefore, it is considered in this work, as an example of an evolutionarily ambiguous structural relationship.
We sought to use the structures of these AKs and the TPKs to determine a true “essential” kinase fold that is seen in all members of the kinase superfamily, as well as shared structural characteristics between the various families. We encoded these structural characteristics into a phylogenetic character matrix. We then combined this information with a structure-based sequence alignment in a unified Bayesian phylogenetic analysis [37,38]. Such an approach has been used previously for sequence data combined with morphological data, to determine relationships between species [39]. Also, discrete structural and sequence motif characters have been used previously to study fold-level relationships between protein structures [40]. However, to our knowledge, our study is the first in which the nuanced information available in a full-length sequence alignment is combined with structural characters in a unified analysis. Use of these two complementary sources of data allowed us to make rational phylogenetic predictions with high confidence, despite the very low sequence similarity inherent in superfamily-level comparisons. The results provide considerable insight into the development of the various kinases in the superfamily from a common ancestor. In addition, our approach offers a new and broadly applicable approach to the study of protein superfamily evolution.
Results/Discussion
Selection of a Representative Kinase Structure Set
The large number of kinase structures available necessitated the selection of a representative set of non-redundant structures for structural alignment. We used a rigorous framework based on both sequence and structural criteria to select the most representative structures within the superfamily. Our criteria were guided primarily by the structure classification provided by the SCOP database [1] (see Materials and Methods for details of our selection criteria). The resulting set of structures constituted 25 TPKs and the six AKs described in the introduction (Table 1).
Structural Alignment and Analysis of the Superfamily
Creation of a highly accurate alignment using sequence information alone is difficult for the TPKs and impossible if the other superfamily members are included [41,42]. Therefore, in order to provide an overview of the structural and sequence features of the superfamily, we created an alignment of the structures based on their structural features. Although automated structure alignment methods are available [43], their accuracy is limited, and the ideal alignment of structures is often ambiguous [44,45]. Therefore, to ensure a highly accurate alignment the structures were aligned manually, using an automated multiple structure alignment as a starting point (see Materials and Methods).
Analysis of the aligned structures and sequences produced several key themes. First, the kinases all share a universal conserved core section, which roughly describes the region constituting the ATP binding pocket and locations of residues involved in the phosphotransfer reaction. Second, the conserved region, while mostly maintained in terms of its overall secondary structures, is often modified substantially in terms of the spatial placement of the structural elements. Third, the kinases generally have distinctive structural elements joined to both the N- and C-terminal ends of the universal core region. In addition, many also have substantial insertions that occur within conserved structural elements in the universal core region. In most cases, these structural insertions have absolutely no spatial similarity between families, though there are intriguing exceptions. Fourth, though the sequence similarity between families is very low, a small group of residues shows remarkable conservation across the entirety of the superfamily. Many of these residues have been previously recognized as highly important for proper activity in the TPKs [13,17]. Hence, it appears that the all of the kinases utilize a similar mechanism for phosphotransfer. The overall impression that emerges is one of a superfamily that has assiduously retained its basic function, but simultaneously has been heavily modified over the course of evolution to phosphorylate a variety of targets, interact with a range of partner proteins, and respond to different regulatory mechanisms.
Phylogenetic Analysis of the Kinase Superfamily
Traditionally, molecular phylogenies are constructed as trees based on sequence similarity, coupled to an underlying theory of sequence evolution [46]. The extreme sequence divergence seen in the kinase superfamily (and in superfamilies in general) makes such determinations problematic. Therefore, in order to postulate an evolutionary history for the kinase superfamily, we constructed a phylogenetic tree using a Bayesian method [38,39] to integrate the sequence and structural data into a single analysis. This combined phylogenetic model provides higher reliability than a model produced using sequence or structural information alone.
Bayesian analysis was carried out using Markov Chain Monte Carlo as implemented in the program MrBayes [38,47]. The sequence alignment presented in Figure 3 was used as the input alignment. Because this sequence alignment was generated from a high-quality structural alignment, one difficulty normally posed when building trees for distantly related sequences—aligning them accurately—was eliminated. Hence, the only limitation on phylogenetic inference was the inherent sequence degradation at the superfamily level.
Structural data were incorporated as a 20-column character matrix, containing the 20 distinctive structural characteristics described below (Table 2). Converting these characteristics into a character matrix allowed for much of the structural information from our comparative analysis to be quantitatively evaluated in MrBayes. These two datasets were simultaneously evaluated in MrBayes as “mixed” data, allowing for the creation of a single tree that provided maximum agreement with both (Figure 4; see Materials and Methods for detailed information).
Table 2 Distinctive Structural Characters Used in the Construction of a Phylogeny for the Kinases
Figure 4 Proposed Phylogeny for the Kinase-Like Superfamily, Based on a Unified Bayesian Analysis of Both the Sequence Alignment in Figure 3 and the Structural Character Matrix in Table 2
Structures are labeled by their PDB IDs, followed by the abbreviated name of the structure. TPKs are to the left of the figure, and are labeled with their group membership. TPKs labeled with a black asterisk are classified differently in our tree compared with the classification produced by Manning et al. [7]. The AKs are highlighted with an orange oval. Major branches are labeled with their posterior probabilities. Gray ovals represent areas of doubt in the tree, based on the tree itself and other aspects of our analysis (see text). The left-hand oval represents uncertainty as to the closest TPK relative to the AKs; it is unclear where precisely the AKs should link to the TPKs (note that this uncertainty does not include the branching of most of the TPK groups in this region, as these are generally well supported). The right-hand oval represents uncertainty as to the proper placement of ChaK and PIPKIIβ. These kinases are difficult to place with high confidence because of their extreme divergence. They are labeled with red asterisks to denote the speculative nature of the current placement (see text).
Selection of Structural Characters for Phylogenetic Analysis
Because protein structure is much more conserved than protein sequence over the course of evolution, it is possible to determine the likely relationships between proteins through comparative structure analysis. Structures that have similar features are likely to share a closer evolutionary relationship, especially if the features are uncommon in protein structures in general [34,40,48,49]. Based on our structural alignment, we undertook a careful comparative analysis of the structures in the superfamily to isolate distinctive structural characters seen in only one or more structures in the superfamily, but not all.
The majority of characters collected were in the universal core of the kinases, as this is the most conserved portion between the different families in the superfamily. This region represents a functional “cassette” responsible for the essential kinase functions of ATP binding and phosphotransfer. Almost all sequence and structure changes within this cassette during evolution would be expected to be deleterious to proper kinase function. Hence, in the most parsimonious scenario, any successful changes in the region would likely occur only once, and then be reused by progeny kinases. Therefore, similarities (and differences) seen within the universal core are expected to be more significant than similarities in other parts of the structures.
In addition, characters were collected for structures outside of the universal core shared by only a subset of the superfamily. Since these sorts of structures are further from the functional core, they can be expected to change more quickly than those within the core. Therefore, to be included, these sorts of structures had to be substantial and distinctive, as opposed to the more subtle structural differences accepted in the core. Finally, a subset of characters specific only to the TPKs was collected. Because there is more than one structure available for this family, this information was used to help improve the phylogenetic analysis within the highly diverse TPK family.
Since sequence motif information is inherently present in the sequence alignment (and this was included in the analysis), the presence/absence of particular sequence motifs was generally not included in the character matrix. However, specific modifications involving sequence that had special structural or functional implications were included, since in many cases the critical importance of these changes is not sufficiently expressed within the sequence data.
We provide a brief summary of each of the characters included in the analysis, and their importance to the structure and function of the enzymes. For the sake of economy, when secondary structural elements that form the universal core are named generically, we use the conventions used for protein kinase A (PKA) [12] (and many other TPKs) and use uppercase to denote this standardized nomenclature (e.g., “Helix C”). When elements from specific structures are discussed, the corresponding element names for these structures (where different from those for PKA) are provided in lowercase. Conversion of this scheme to that used for the other kinase families is available in the labeling of elements in Figures 1 and 2. Similarly, the residue numbers for generic residue positions are based on the residues and numbers for PKA. In cases where a residue number is provided that is specific to a structure, it is followed by the residue number for the comparable residue in PKA in parentheses (e.g., “Q1767(L172)”). Comparable residues for any other structure in the set may then be retrieved from the alignment provided in Figure 3. The characters are presented in approximate N-terminal to C-terminal order.
1: Ion pair analogous to K72-E91 in PKA.
In all of the kinases, a very highly conserved lysine (K72) or arginine residue is present in Strand 3, facing the binding pocket. In most of the structures with bound ATP, K72 interacts with the α and β phosphates of the ATP molecule, helping to stabilize them in the proper conformation for phosphotransfer [15]. The position of K72 is stabilized by the formation of an ion pair with a glutamic acid residue (E91) in Helix C. By linking Helix C to Strand 3, the Lys-Glu ion pair also helps to stabilize the overall fold of the N-terminal subdomain. Some of the AKs have conservative substitutions at either of these positions (Figure 3). In others, such as PI3K [21] and ChaK [20], the negatively charged residue at E91 may play a diminished role, or form an ion pair with K72 only when the kinase is in an active conformation. Such conformational shifts are seen in the TPKs, wherein the K72-E91 ion pair is broken by movement of Helix C when the kinase is in an inactive state [15,50]. The one distinctive exception is seen in PIPKIIβ, which retains K72 but lacks a clear replacement for E91. D156(H87) in PIPKIIβ may fulfill the role of E91 in PKA [19], but unlike the other kinases, a negative charge has been completely removed from position E91 in PIPKIIβ.
2: α-Helix B.
Between Strand 3 and Helix C, most of the kinases have a short loop structure. However, the AGC group of TPKs (Table 1) and the aurora-2 kinase [51] share the distinctive α-Helix B at this location (Figure 3). This helix is not seen in any of the other TPKs. Remarkably, however, it is seen in ChaK, where it is the same length, though it is shifted spatially from what is seen in the AGC kinase PKA (Figures 1–3). Hence, the conservation of Helix B in ChaK is surprising, particularly given its distinctive structure.
3: Kink in α-Helix C.
In PIPKIIβ, helix 4 (Helix C) contains a distinctive kink not seen in any of the other kinases (Figure 2). This kink requires some reorganization of the ATP binding pocket and allows for interaction of the N-terminal subunit with the highly modified shape of the C-terminal subunit (see characters below). The kink also appears to play a role in the lack of a K72-E91 ion pair (character 1) in this structure, because it places the region of the helix where the required Glu residue would reside far from K150(K72).
4: Kink in Strand 4.
Most kinases in the superfamily have a distinctive kink near the beginning of Strand 4. This kink modifies the placement and architecture of much of the hydrophobic pocket formed by Strand 4, Helix C, and Helix E. ChaK, PI3K, and AFK are the exceptions, and contain a straightened (and/or shortened) Strand 4 (strand 9 in ChaK; strand 6 in PI3K), which changes the architecture in this region of the core. This change results in the requirement for a gap within the Strand 4 region when aligning these structures with others in the superfamily (Figures 2 and 3).
5: Helical structure in the area of α-Helix D.
Helix D appears just after the linker region in the TPKs (Figure 1). In most of the AKs, helical structures are present in this region, though they are not always superposable, and some are 3–10 helices rather than α-helices. However, ChaK is distinctive in that it completely lacks this element (Figure 2).
6: Orientation of α-Helix E.
Helix E stabilizes the ATP binding pocket through its interactions with Strands 7 and 8. In most of the kinases, it is oriented at approximately 45° to these elements, but in PIPKIIβ, helix 6(Helix E) is approximately parallel to them, a major reorganization of the supporting structure of the catalytic core (Figure 2).
7: Key conserved histidine at H158.
Helix E (helix D in CKA-2; helix 4 in APH(3′)-IIIa) also contains a conserved histidine residue, H158, which is shared only between the TPKs and the APH and CK families. Remarkably however, H158 is not conserved in the tyrosine kinase group within the TPKs. H158 forms a hydrogen bond with D220 and in so doing, participates in a hydrogen-bond network that links together Helices E, F, and the crossing loops in the catalytic region of these kinases (see below and Figure 5). Hence, in the conservation of this interaction, the APH and CK families display a closer relationship to the Ser/Thr TPKs than do the tyrosine kinases (it should be noted that H158, while conserved in APHs, is less conserved than it is in the Ser/Thr TPKs and CKs, and may be of somewhat reduced importance in this family).
Figure 5 Shared Hydrogen-Bonding Networks between Distantly Related Structures in the Kinase-Like Superfamily
Colors and nomenclature for secondary structural elements are identical to those provided in Figure 2. Structures shown are the C-terminal subdomains of four structures: (A) PKA [70]; (B) CKA-2 [23]; (C) PI3K [21]; and (D) AFK [22]. For clarity, some portions of structures are omitted. Residues involved in the shared hydrogen-bond networks are shown in a ball-and-stick rendering. For clarity, side-chains are omitted for residues that only participate in the network via backbone interactions. Residues involved directly in catalysis or metal binding are shown with light-green stick regions in the ball-and-stick rendering. Metal atoms, when present, are shown as gray spheres. ATP (or ATP analog), when present, is shown in a line rendering. Hydrogen bonds are shown in cyan. The orientation of the structures is similar but not identical (structures were rotated somewhat to make H-bond contacts more visible). Molecular renderings in this figure were created with MOLSCRIPT [90].
8: Large helical insertion between Helix E and Strand 6.
Two of the kinases, CKA-2 and APH(3′)-IIIa, contain a distinctive insertion immediately after Helix E (helix D in CKA-2; helix 4 in APH(3′)-IIIa). The shared insertion consists of two interacting helices, linked by a short loop containing a small helix (Figure 2). In both kinases, these insertions effectively replace the Activation/P+1 Loop of the TPKs (see character 14). Though they do not align perfectly (Figure 3), the striking similarity of these elements, and their absence in all other kinases, suggests that they are a product of relatively close common ancestry between CKs and APHs.
9: Structure underlying the catalytic region.
The Catalytic Region of many of the kinase families is supported by complex hydrogen-bond networks that stabilize the architecture of the active site. There are distinctive similarities in these networks that suggest relatively close evolutionary relationships between some families. The TPKs, CKA-2, and APH(3′)-IIIa all share an H-bond network centered around a highly conserved His or Tyr residue at position Y164, which usually forms a hydrogen bond with the backbone carbonyl of position T183, just after the end of Strand 8 (strand 11 in CKA-2). This interaction is significant, because D184 is highly conserved, and interacts with a magnesium atom in the active site that is important for ATP interaction and the phosphotransfer reaction [13]. In addition, this region is the area in which a “crossing loops” structure is formed, where the catalytic loop and the loop between Strands 8 and 9 cross. This type of motif is unusual in protein structures, and is one of the hallmarks of the kinase superfamily [34]. The Y164-T183 hydrogen bond is also a part of a larger conserved H-bond network shared by the APH, CK, and TPK families. This network includes H158 in Helix E (character 7) and D220 in Helix F (helix G in CKA-2; helix 5 in APH(3′)-IIIa), and essentially ties together the catalytic region in these kinases (Figure 5).
In AFK and PI3K, the H-bond to the backbone of position T183 is instead made by an arginine residue at position L167 (Figures 3 and 5). This Arg residue effectively replaces, from a location three positions down the chain, the function of Y164. Thus, these two structures share a distinctive interaction at the center of their catalytic regions that replaces a conserved interaction seen in many of the other kinases. Further, these two kinases both lack the extended H-bond network seen in the three families above.
ChaK and PIPKIIβ do not have any of the H-bonding patterns seen in the other two groups. They each use unique underlying structures to stabilize their catalytic regions.
10: Architecture of the catalytic region.
Between a highly conserved Asp (important for catalysis) at position D166 and Strand 7 the backbone in most of the kinases adopts a structure commonly called the “catalytic loop.” In most structures containing the element, this “loop” actually consists partly of a short 3–10 helix. Two structures, PIPKIIβ and ChaK, lack the catalytic loop completely, and instead have an approximately linear connection between D166 and Strand 7 (strand 10 in PIPKIIβ; strand 13 in ChaK; Figures 2 and 3).
11: Insertion in the catalytic region.
Following the Arg residue at position L167, AFK contains an insert that loops away from the catalytic region and interacts with the C-terminal subdomain. This element is unique to AFK (Figure 2).
12: Asp residue at 171, or apparent compensation for its absence.
In those structures containing the 3–10 helix (or a loop in a similar conformation), the last position of the helix contains a highly conserved asparagine residue, N171. This important residue is responsible for interaction with a magnesium ion, which in turn interacts with the phosphate groups of ATP [13]. It also participates in the H-bond network discussed above (see “Structure underlying the catalytic region”), further increasing its importance (Figures 3 and 5).
In the two kinases lacking the helical element, there is an interesting divergence in compensation for the lack of N171. In ChaK, the next position down the chain, Q1767(L172) is the highly similar residue glutamine. Remarkably, the longer side-chain of this glutamine is angled such that the amide group is in a similar location in space to the amide group of N171 in the other structures. Conversely, in PIPKIIβ there is no obvious compensation for the loss of N171, and since ATP is not present in this structure it is unclear how PIPKIIβ interacts with ATP without N171. Hence, ChaK is more similar to the rest of the kinases in this area of the structures, and this is reflected in our matrix (Table 2).
13: Similar direct hydrophobic link between catalytic region and Helix E.
In the structures of the TPKs, APH(3′)-IIIa, and CKA-2, conserved hydrophobic residues (L167 and L172) flank the 3–10 helix and face into the hydrophobic core. They interact directly with each other, as well as a conserved hydrophobic residue at I150 in Helix E (helix D in CKA-2; helix 4 in APH(3′)-IIIa). Though many other kinase families have conserved hydrophobic residues at these positions (Figure 3) and have a clear hydrophobic pocket, this distinctive link is specific to the TPK, APH, and CK families. These interactions are important because they form a direct link between the Catalytic Loop and Helix E, stabilizing the conformation of the Catalytic Loop.
14: Nature of structure linking Strand 9 and Helix F.
The region immediately following Strand 9 is termed the “Activation Loop” in the TPKs, because many TPKs are regulated by phosphorylation of residues in this loop [15,52–54]. All of the TPKs in our set have a substantial activation loop (Figure 3). The loop immediately following the Activation Loop is often termed the “P+1 loop” in the TPKs, because it interacts with residues in the substrate protein chain one position (and beyond) from the actual residue targeted for phosphorylation [29]. The P+1 loop is followed by the distinctive APE (or similar) motif in most TPKs. Beginning at P207 in the motif there is a conserved helix, which we term Helix 1 to avoid conflict with the standard TPK naming scheme. The last residue in the APE motif, E208, is highly conserved within the TPKs. It forms an ion pair with an arginine residue, R280, further down the chain. R280 is located in a loop between Helices H and I. Hence, the effect of the ion pair is to hold the C-terminal subdomain together. This ion pair is retained in all TPKs except the CK1 group (see character 17). However, in terms of overall architecture, all the TPKs have a similar structure in the Helix 1 region (and the rest of the C-terminal subunit).
None of the AKs share a similar structure to TPKs in the Activation Loop region (Figure 2). Most structures have a markedly shortened loop relative to that seen for the activation/P+1 loops in the TPKs, and the structures are distinct in most families (accurate analysis of the Activation Loop regions of many of the AKs is difficult because they are not resolved in the experimental structures). The exceptions are CKA-2 and APH(3′)-IIIa, which share a distinctive short and highly twisted β-sheet in the Activation Loop region formed by Strands 6 and 9 (strands 9 and 12 in CKA-2; Figure 2). This structure allows for an extremely short “Activation Loop,” the shortest within the superfamily.
15: Positioning of Helix F.
Helix F, which follows the various loop structures, constitutes the last region of structural similarity shared by all of the kinases, though the similarity in this region drops off rapidly. It could be argued that in some cases, this helix superposes so poorly between superfamily structures that it should not be considered part of the “universal core.” However, it is present with an approximately similar orientation in all structures, and in most cases seems to have a similar role: stabilization of the backbone of the Catalytic Loop. However, the manner in which this stabilization is achieved is highly variable.
An exception to this variability is seen between the TPKs, APH(3′)-IIIa, and CKA-2. In these three families, Helix F (helix G in CKA-2; helix 5 in APH(3′)-IIIa) is maintained in a highly similar orientation and is readily superposable (Figures 1 and 2). More significantly, the families share an aspartate residue, D220, that is highly conserved in the three families. This residue forms hydrogen bonds with the backbone amides of Y164 and R165 and (with the exception of the tyrosine kinases; see character 7 above) the side-chain of H158. Hence, a network of residues and contacts that is responsible for the specific geometry of the most conserved regions of the kinase fold has been carefully conserved in these three kinase families.
Though Helix F can be superposed relatively well between the TPK, APH, and CK families, it is much more variable in the four remaining families, and is only weakly superposable. The large helical insertion into the Activation Loop of AFK pushes helix 8 (Helix F) into an angled position, such that it tilts away from the catalytic loop. The space opened by this translocation is filled by the insertion seen in the middle of the catalytic loop in this structure (character 11 and Figure 2). In PI3K, helix 7 (Helix F) is shortened such that a loop region interacts with much of the catalytic loop, partly replacing the role of Helix F in other structures (Figure 2). In ChaK, helix E (Helix F) is shortened and tilted away from the catalytic loop to the point that it appears to play no direct role in stabilizing this element. PIPKIIβ has a structure that is more similar to what is seen in Helix F in the TPKs, except that the orientation of helix 8 (Helix F) relative to strands 10 and 12 (Strands 7 and 8) is nearly parallel, rather than an approximate 45° angle as seen in the TPKs (Figure 2).
16: Structural similarities in C-terminal subunit, following the universal core.
Though Helix F represents the end of the universal core shared by all kinases in the superfamily, many of the kinases have additional structure beyond this point, and there are shared substructures between some families that argue for a closer evolutionary relationship. All of the TPKs share superposable Helices G, H, and I (Figures 1 and 3). However, none of the other kinase families contain these structures.
APH(3′)-IIIa and CKA-2 share two superposable helices in their C-terminal subunits along with a very similar overall topology. CKA-2 follows helix G (Helix F) with a small β-sheet and a small helix, which APH(3′)-IIIa lacks. However, the helix that follows is superposable between the structures. After this helix, CKA-2 has an additional two helices, while APH(3′)-IIIa has an irregular loop structure. However, the overall path of the chain is identical between the two structures, and they share another superposable helix in the likely substrate binding region. The chain of APH(3′)-IIIa terminates at the end of this helix, while CKA-2 adds an additional two helices (Figure 2).
AFK and PI3K have differing structures in the area of Helix F (helix 8 in AFK; helix 7 in PI3K). However, immediately following this region the two structures share a set of similar helical elements. The first of these helices interacts with Helix E (helix 6 in both AFK and PI3K), and superposes well between the two structures. The second and third of these helices superpose only weakly. However, they are in approximately similar orientations, and together with the first helix form a motif that is distinct within the superfamily. After the third helix, PI3K has two additional helices, which are not seen in AFK (Figure 2).
The C-terminal subdomain structure of ChaK is completely novel, and not shared by any other kinase in the superfamily. Remarkably, a zinc finger [55] forms the center of the subdomain and links all the major elements together [20]. The zinc coordination links helices D and E (Helices E and F) and the final terminal helix, which each provide one of the coordinating histidine or cysteine residues. The final coordinating cysteine is provided by the loop linking helix E and the final helix.
The C-terminal subdomain of PIPKIIβ contains essentially no additional structure beyond helix 8 (Helix F).
17: Ion pair analogous to E208-R280 in PKA (TPKs only).
In CK1, the APE sequence in Helix 1 (described above) is replaced with the motif SIN (which is conserved within the CK1 group). This motif essentially fills the roles of APE in the first two positions, but at position N188(E208), an asparagine residue replaces the glutamate seen in other TPKs, and hence no ion pair is formed. CK1 also does not contain a positively charged residue that correlates to R208 in the other TPKs (Figure 3). However, it substitutes a new ion pair that the other TPKs lack. Residue E202(W222) from Helix F forms an ion pair with residue R261(L273) from Helix H. Thus, the linkage between different regions of the C-terminal subdomain is essentially retained, albeit with a pair of residues that are novel with respect to the rest of the TPKs. The substitution of APE with SIN (and a different ion pairing) may have implications for the evolution of CK1 relative to the other TPKs, given the strict conservation of the E208-R280 ion pair in these structures. However, the overall structure of the C-terminal subdomain of CK1 is still very similar to that for the other TPKs.
18: Extensive helical insertions between Helix G and Helix H (TPKs only).
The CMGC group of TPKs contains distinctive helical insertions between Helix G and Helix H. These insertions are variable in position and helix length, but they are much more extensive than the small insertions occasionally seen in other families. Interestingly, CK2 also contains these insertions (Figure 3).
19: Insertion between R280 and Helix I (TPKs only).
The AGC kinases share a distinctive insertion between R280 and Helix I (Figure 3).
20: Helix I structure (TPKs only).
Helix I often actually consists of two shorter helices joined by a linker. In most cases, the first helix is an α-helix, and the second is a 3–10 helix (Figure 3). This split helix structure is dominant for Ser/Thr kinases, while Tyr kinases have a single long Helix I. Interestingly, three Ser/Thr kinases share the Tyr kinase–like architecture for Helix I. One of these is TGFβR1 from the tyrosine kinase–like (TKL) group, so the structural similarity is unsurprising. However the other two kinases, CK1 and the bacterial kinase PknB, do not have an obvious reason to display this similarity to the Tyr kinases.
Comparing the Phylogenetic Analysis with Other Data
We interrogated our phylogenetic model against the backdrop of species distribution of the families. We utilized the pre-computed results available in PFAM [56] to survey the presence or absence of the kinase families corresponding to structures in our set in the three superkingdoms of life (Table 3). These species representation data also fit well with other lines of inquiry (see below). We also created superpositions of selected structures based on our alignment to provide root mean square deviation (RMSD) values as a general estimate of structural similarity (Table 4). These were helpful in augmenting our own qualitative knowledge of structural similarities seen between the families, and their likely significance.
Table 3 Phylogenetic Distribution of Kinase Families within the Superfamily, According to the Pfam Resource
Table 4 RMSD and Number of Aligned Residues from Representative Kinase Structure Alignments, When Superposed Based on the Alignment Presented in Figure 3
Finally, we compared our tree with a tree made using only sequence information and a more traditional distance-based method of phylogenetic inference, to provide a comparative benchmark (Figure 6; see Materials and Methods for details of the tree construction). Although this tree did not utilize structural information, it still could take advantage of the highly accurate sequence alignment. However, this tree demonstrates the difficulty inherent in using sequence information alone to discern superfamily-level relationships. While the tree is able to successfully cluster groups of similar proteins out at the edges with acceptable confidence, the center of the tree suffers from low bootstrap values, and thus is somewhat speculative in these areas (we report branches with bootstrap values of < 50% of replicates as speculative based on the results of benchmarking studies [57,58]). Interestingly, comparison with the tree produced with MrBayes reveals a large degree of overlap. Areas of agreement between the two trees provide additional supporting evidence for the validity of the results.
Figure 6 Conventional Distance-Based Phylogenetic Tree of the Kinase-Like Superfamily, Based Only on the Sequence Alignment from Figure 3
This tree did not explicitly incorporate structural information, and is provided for purposes of comparison with the Bayesian tree presented in Figure 4. Structures are labeled by their PDB IDs, followed by the abbreviated name of the structure. The AKs are highlighted by orange ovals. Bootstrap values are provided for major branches. Some branches are too short for values to fit; these are marked with red letters that correspond to the following values: a, 199; b, 170; c, 101; d, 141. Branches highlighted in gray were not supported by bootstrap values above 500, and should be considered speculative (if based only on this tree data) [57,58]. Many of the core relationships within the superfamily cannot be resolved with confidence using the conventional sequence-based approach.
However, we believe that the MrBayes tree is much more reliable than the conventional tree, given the explicit addition of structural information. Review of Bayesian trees generated using only the sequence information or structural information (Figures S1 and S2) demonstrated that neither dataset alone was capable of producing a resolved tree. When compared with the tree generated by sequence alone, the tree incorporating structural information (presented in Figure 4) provided several concrete benefits. First, the combined tree resolved polytomies (“star trees” at particular nodes) seen in the sequence-only tree. Second, the combined tree provided higher branch confidence values for many branches (in Bayesian trees, branch confidence values are estimated as posterior probabilities, which are generally interpreted as the probability that a branch is correct, provided that the evolutionary model and priors are correct [37,59]). Third, where branch changes occurred in the combined tree, the net effect was generally to produce a tree with better agreement with the structural observations (i.e., the use of structural characters in the analysis produced the desired effect). We discuss the various data within the context of the implications for structural evolution of the kinases.
How Did the Various Kinases Evolve?
The TPKs appear to be ancient but display remarkable conservation of sequence and structural features.
Against the backdrop of the AKs, the TPKs can be seen to be a remarkably well-conserved family of enzymes, given their high level of duplication and broad distribution within the three superkingdoms of life (Table 3) [6,8–11,60]. It would appear that the TPK core structure, once arrived at, has required little modification in order to switch to many different protein substrates. The TPKs share not only a highly similar core cassette but also a large amount of distinctive substrate binding and stabilizing structure in their C-terminal regions. In addition, they contain numerous sequence motifs that are extremely well conserved, even though some (such as APE) appear to play a primarily structural role that would seem to be replaceable with other sequences and structures. Consistent with these observations, the TPKs form a relatively tight cluster in our phylogenetic tree, with clear subsections representing the Ser/Thr kinases and Tyr kinases, and the TKL group as an intermediate between the two (Figure 4).
Interestingly, our tree also places the one bacterial TPK structure available, PknB [61], in its own distinct group near the center of the tree, in the middle of the radiation of the TPKs (Figure 4). This location is consistent with a scenario in which an ancestor of the TPKs arose before the radiation of the three superkingdoms of life, with the other TPKs in our tree developing separately in eukaryotes. Leonard et al. have conducted an in-depth sequence-based study that placed the PknB kinase within the “Pkn2” group of kinases in bacteria, and noted that of the prokaryotic kinases, the Pkn2 group is the most closely related to the TPKs seen in eukaryotes [8]. Pkn2 kinases are not seen in archaea, and Leonard et al. suggested that this indicates that the Pkn2 group was horizontally transferred into bacteria from eukaryotes shortly after the divergence of the three superkingdoms of life. Thus, some of the eukaryotic-like TPKs seen in bacteria could be the result of an early horizontal transfer event. Our tree would also be consistent with this scenario. It should be noted that any scenario for the development of TPKs in bacteria must place them into the bacterial lineage very early in evolution, given their very broad distribution in this superkingdom [8,9,11,60], and results of codon bias and G/C content studies [62].
Manning et al. have produced a tree for the all TPKs in the human genome, using sequence information only [7]. As our tree had the benefit of a potentially more accurate sequence alignment, as well as the inclusion of structural features, we sought to compare our results with theirs. The two trees display a high level of agreement, though some differences are evident. Interestingly, where our tree differs substantially, we are often able to offer structural arguments suggesting that our tree is more likely to be correct.
In terms of the overall tree architecture of the various TPK groups, our tree is nearly identical to that by Manning et al., with the exception that their tree places the STE group kinases closer to the TKL and TK groups than the CK1 group. Our tree places the CK1 group closer to TKL/TK than STE, with a very high posterior probability (Figure 4). As noted above, the TK, TKL, and CK1 groups share a similar Helix I structure that is changed in all other eukaryotic TPKs in our set (Table 2, character 20, and Figure 3).
We also classify two specific kinases differently than Manning et al. The first, CK2, is classified by Manning et al. as “other” and placed near the root of the CMGC group on their tree. Our tree instead places CK2 well within the CMGC group, with a high posterior probability on the major branch separating the group from the rest of the TPKs (Figure 4). As described above, CK2 also contains the distinctive helical insertions between Helices G and H, insertions otherwise only seen within members of the CMGC group (Table 2, character 18). Finally, our conventional tree also places CK2 well within the CMGC group, with a reasonably strong bootstrap value for the major branch (Figure 6). We submit that CK2 should be considered fully a member of the CMGC group. The other kinase for which our classification differs is cell cycle checkpoint kinase (Chk1). Manning et al. classify this kinase as a member of the CAMK group, placing it near the root of the group. Our tree classifies this kinase as “other,” and the separated CAMK group has a very high posterior probability on its main branch, indicating that the rest of the CAMK group is very sequence distinct from Chk1 (Figure 4). Our conventional tree also separates Chk1 from the CAMK group, with a strong bootstrap value separating the CAMK group from Chk1 and the rest of the TPK family (Figure 6). However, in this case there is no direct structural argument for the placement of Chk1 in or out of the CAMK group. Therefore, we remove Chk1 from the CAMK group for purposes of our analysis, but do not necessarily argue for its reclassification.
The TPK that forms the closest link with the AKs is difficult to determine.
The AKs form a distinct phyletic group (see below), but the TPK that constitutes the closest link to the AKs is difficult to verify with a high degree of certainty. Our tree places Chk1 in this position, with a moderate posterior probability (Figure 4). Chk1 does seem to potentially be a good candidate, as it is widely distributed in eukaryotes, and is a key player in the critical (and presumably ancient) cellular response to DNA damage, as well as cell cycle control [63].
However, there is no compelling structural evidence linking Chk1 to the AKs. Only two of our structural characters show partial representation in both the TPKs and AKs, thus providing structural information as to possible TPK/AK links (characters 2 and 7; see above and Table 2). These two characters do not directly link Chk1 to the AKs. Chk1 also does not show any tendency toward lower RMSD values when aligned to the AKs, relative to other TPKs (Table 4). Hence, the linking of Chk1 to the AKs is done primarily through sequence, which can be unreliable at this level of divergence.
Given this level of doubt in the analysis, it is not surprising that our conventional tree instead presents CK1 as being the closest link (Figure 6). Bootstrap support is very weak for the link, but as with Chk1, CK1 does have some characteristics that make it attractive as the link to the AKs. CK1 is the only kinase to replace the APE motif with a SIN motif, and in the process lose the distinctive E208-R280 ion pair seen in other TPKs (see above). As the AKs obviously lack this ion pair as well, CK1 could be seen as a more “primitive” kinase. Given the very broad distribution of the CK1 group in eukaryotes [6], the ion pair switch appears likely to have occurred shortly after the separation of eukaryotes into a distinct superkingdom. CK1 also has a variety of other sequence peculiarities that cause it to be placed in a unique location on our phylogenetic trees, intermediate between the Ser/Thr kinases and Tyr kinases (Figures 4 and 6). Hence, CK1 likely represents an ancient group of TPKs.
However, we are not aware of any confirmed case of a CK1-like kinase in prokaryotes, indicating that CK1-like kinases are limited to eukaryotes. BLAST searches by us against all bacterial genomes revealed that the 50 highest scoring hits (BLAST E-values from 2 × 10−14 to 1 × 10−8) maintained the usual APE motif seen in the rest of the TPKs (or similar motifs seen in the TPKs, such as SPE). Further, the changes seen in CK1 are relatively minor compared with differences between the TPKs and the AKs, and our structural analysis did not indicate any direct evidence that the CK1 group should be considered closely linked to the AKs. Though CK1 is missing the APE motif, it still has a P+1 loop and Helix 1 structure that are very similar to the other TPKs (Figure 3). CK1 also does not align to the AKs with lower RMSD or more aligned positions, relative to the other TPKs (Table 4).
The examples of Chk1 and CK1 illustrate the difficulty in determining the specific TPK that constitutes the closest link to the AKs. Though Chk1 appears to be the strongest candidate at this time for the closest link to the AKs, we believe that such links will remain speculative in the absence of new kinase structures that might provide additional insights.
The AKs Form a Distinct Group
There is strong evidence that the AKs form a separate phyletic group, and that this group has an ancient origin, probably evolving as early as the TPKs. This is in contrast to an alternate scenario where the TPKs developed first and then the AKs arose via intermittent divergence from various TPKs. An ancient origin for the AKs is supported by our tree, which separates the AKs from the TPKs completely, with a very high posterior probability on the separating branch (Figure 4). Three of the families, the PI3Ks, CKs, and APHs, are broadly distributed in eukaryotes and seen in many bacteria, similar to the pattern seen in the TPKs (two AK families, the PIPKs and α-kinases, are not so broadly distributed and have a more puzzling origin; see next section). This is the opposite pattern from what would be expected if these AKs had diverged intermittently, in which case they would appear in only a subset of organisms. These three AK families traverse the entirety of the AK portion of the tree, helping to establish its ancient origin. Further, as mentioned in the previous section, only two of our structural characters indicated that specific AK families might have closer relationships with specific TPK groups. In other words, most of the AKs do not appear to simply represent different modifications of extant TPK structures.
Within the AKs, the CKs and APHs can be most closely linked with the TPKs. These three families share distinctive structure and sequence motifs within their core cassettes that stabilize the geometry of the catalytic residues and the crossing loops (see structure analysis above, and Table 2). Also, it has been shown that APH(3′)-IIIa has some protein kinase activity [64], providing a functional link between the APHs and TPKs.
As stated previously, CKA-2 and APH(3′)-IIIa also share a remarkable amount of additional structure within their C-terminal subdomains (Figure 2). This structure is seen in two different sections of the protein chain, extensive in length, superposable, and not seen in any other member of the superfamily. These observations argue compellingly that the CK and APH families are relatively closely related, and the most closely related within the superfamily. Accordingly, our phylogenetic tree places APH(3′)-IIIa and CKA-2 close together, though with considerable evolutionary distance after their split (Figure 4). It would appear that choline and APHs shared a similar common ancestor. This common ancestor, in turn, shared a relatively close common ancestor with the TPKs. Whether the common ancestor looked more like a TPK or the APHs/CKs is unknown.
The TPK/APH/CK cluster can be linked to PI3K and AFK partly by establishing a major evolutionary split in the superfamily based on the structure of the core cassette. Most of the families within the superfamily have a short 3–10 helix (or a loop in nearly this conformation) in the middle of their catalytic loop regions. In all of these structures, the third position of this 3–10 helix contains a highly conserved asparagine residue, N171, which is responsible for binding a metal ion. In addition, this 3–10 helix is nearly immediately preceded by the most highly conserved reside in the superfamily, D166 (Figure 3). Given the critical importance of this region of the kinases, modifications would be expected to be extremely rare. Indeed, this motif is highly resistant to alteration, as a broad assortment of kinases in the superfamily, despite large changes in substrate and supporting structures, have carefully retained it (Figures 1 and 2). AFK does contain an insertion between D166 and N171, demonstrating that such insertions can occur. However, the insertion in AFK changes the orientation of these residues very little, indicating that in this one case the insertion was acceptable precisely because it did not change the essential structure of the catalytic loop. However, ChaK and PIPKIIβ lack this element, instead using an approximately linear chain structure (with compensation in ChaK for the loss of N171, and no obvious compensation in PIPKIIβ; Figures 2 and 3). Thus, it seems reasonable that AFK and PI3K should be grouped relatively closely to the TPK/APH/CK cluster, despite more extensive structural divergence between these structures.
Though AFK is a protein kinase, and can be linked to the TPK/APH/CK cluster, it appears to be more closely related to PI3K than to the TPKs. Though the structural evidence for this linkage is weaker than that linking together the TPK/APH/CK cluster, it remains persuasive. First, though PI3K and AFK share a similar crossing loop structure to that seen in the TPK/APH/CK cluster, the specific residue motifs are changed. Instead of using a histidine or tyrosine residue at Y164 to form a hydrogen bond with the backbone of T183 in the other loop, AFK and PI3K both use an arginine residue at L167 to form this interaction (Figure 5). This interaction is shared by only these two structures. In addition, AFK and PI3K do not conserve an aspartate residue at D220 (seen in all other kinases containing the 3–10 helix motif in their catalytic loop) and the larger network of interactions that are seen in conjunction with this residue (Figure 5).
If structures outside of the conserved core are considered, AFK and PI3K have three similar helices in their C-terminal subdomains, one of which is highly superposable. The other two are weakly superposable, but not seen in any other structures in the superfamily (Figure 2). The net effect of the overall structure of both AFK and PI3K is that the enzyme is flat-faced [21,22]. As AFK is seen in only one species (Table 3), and PI3K is seen in many, a scenario in which PI3K and AFK evolved from a common ancestor might require that AFK evolve from a kinase similar to PI3K. Such a scenario is quite plausible, as even present-day PI3K has some protein kinase activity [65,66] (and enzymes can change their substrate specificity relatively easily over long evolutionary timescales [67]). In addition, a small family of Ser/Thr protein kinases has been identified that contain a catalytic domain highly similar to that seen in PI3K. These phosphoinositide 3-kinase related kinases (PIKKs) demonstrate that the PI3K catalytic domain can be readily modified to phosphorylate protein targets exclusively [68]. However, as with PI3K, these kinases do not share obvious sequence similarity with AFK. AFK may thus represent an alternate modification of a lipid kinase to become a pure protein kinase. Alternately, both AFK and PI3K may have independently converged upon the observed structural similarities as a result of the requirement to be flat-faced. However, our phylogenetic tree also shows AFK and PI3K to share a common ancestor, with relatively high posterior probability (Figure 4).
PIPKIIβ and ChaK are Highly Divergent Kinase Structures, Both from the Rest of the Superfamily and from Each Other
Though ChaK and PIPKIIβ can be distinguished from other kinases in the superfamily based on their lack of a 3–10 helix in their catalytic loops, this does not mean they have any clear similarity to each other that would suggest a close evolutionary link. Indeed, these two kinases do not share any distinctive structure or sequence motifs, and appear no more similar to each other than to the 3–10 helix containing group. RMSD values and number of aligned positions between the two structures are no better than those for comparison of ChaK and PIPKIIβ with the rest of the superfamily (Table 4). Both kinases share an approximately linear catalytic region, but the way in which this structure is achieved is quite different. ChaK has short strands 13 and 14 (Strands 7 and 8), coupled to a novel structure of strands 12 and 15 (Strands 6 and 9) that avoids the use of a crossing loops in the C-terminal subdomain. PIPKIIβ uses elongated strands 10 and 12 (Strands 7 and 8), lacks Strands 6 and 9, and has crossing loops (Figure 2).
Though SCOP does not place PIPKIIβ in the same superfamily as the other kinases, a comparative study has linked this structure to the protein kinase–like superfamily [34]. Our analysis does not suggest any reason to doubt this linkage, but it does indicate that PIPKIIβ is the most divergent kinase in our set. For example, PIPKIIβ displays substantial changes in ion pair patterns and orientation of secondary structural elements (see analysis above and Table 2).
Since ChaK and PIPKIIβ are highly dissimilar, it follows that that they should not be considered close relatives. Both ChaK and PIPKIIβ have been suggested to provide possible links between the protein kinase–like superfamily and two other superfamilies containing mostly metabolic enzymes: the SAICAR synthase and ATP-grasp superfamilies [20,34]. In the case of PIPKIIβ, our analysis does not contradict this possibility. PIPKIIβ is extremely structurally distant from the rest of the superfamily (Table 4), and conserves only the most minimal set of residues related to ATP binding and catalysis, as well as a few hydrophobic residues that form shared hydrophobic cores (Figure 3). We attempted to place PIPKIIβ on our phylogenetic tree, both in an effort to illuminate its origins, and provide a possible outgroup for the tree. Remarkably, the tree places the origin of the PIPKs in the middle of the AKs. This region could be a likely “origin” point for the kinases, where an ancestral kinase diverged to form the AKs, as well as the TPKs (Figure 4). Thus, the phylogenetic tree results are consistent with a very distant relationship between PIPKs and the rest of the kinase superfamily. However, given the weak structural evidence for the location of PIPKs on the tree, this link should be considered speculative (while PIPKIIβ has many distinct structural features, most do not provide informative characters in our matrix for purposes of placing branches). Consideration of species distribution of the PIPKs indicates that they appear to be restricted to the eukaryotes (Table 3). This observation suggests that PIPKs are a more recent arrival into the arsenal of kinases, perhaps developed by eukaryotes in response to a heightened requirement for more complex signaling networks. However, if the PIPKs are a relatively recent invention, this precludes a role for them as a direct link between the SAICAR synthase and/or ATP-grasp folds and protein kinase–like superfamily. However, it does not preclude the possibility that the PIPKs and the kinase superfamily share a very distant common ancestor (which was not necessarily functionally a kinase). The PIPKs share notable structural similarity with the SAICAR synthetase family, leading them to be grouped within this superfamily in the SCOP database [1]. We speculate that the PIPKs may have become kinases through derivation from an ancient non-kinase fold, perhaps a protein similar to SAICAR synthetase. Hence, they may have become kinases through a process of “convergent divergence” with the rest of the kinase superfamily. In such a scenario, the PIPKs would have converged upon the same kinase activity that had already been discovered much earlier by their distant relatives in the rest of the kinase superfamily.
Though ChaK has also been suggested as a possible link between the kinase superfamily and the ATP-grasp superfamily [20], our results, as well as the work of others [27], cast considerable doubt upon this hypothesis. Consideration of the species distribution of α-kinases indicates that they are only narrowly distributed in eukaryotes, appearing primarily in metazoans, and completely absent from green plants (Table 3). This data suggests that the α-kinases appeared relatively recently in evolution, and thus they are precluded from being a direct link between two ancient and widely distributed superfamilies. Presumably, the α-kinases were derived from an extant kinase. However, determining the closest relative to the α-kinases is difficult because of the extremely divergent sequence and structure of ChaK.
Our Bayesian tree places ChaK well within the AKs, closest to PI3K and AFK. Though the posterior probability is relatively low for the branch separating these three families, it is high for the branch separating the three families and the PIPKs from the rest of the superfamily (Figure 4). This would suggest that the closest known structural relative to the α-kinases may be the PI3K family (since AFK apparently evolved recently and is narrowly distributed, it is precluded as a possible source protein for the derivation of the α-kinases). PI3K and ChaK do share a distinctive straightened Strand 4 (strand 6 in PI3K; strand 9 in ChaK, Table 2), but otherwise they do not have any clear structural similarity that would argue for a link. RMSD values for superpositions between these two proteins are unremarkable relative to the rest of the superfamily (Table 4).
Our conventional tree provides a completely contradictory scenario, but there are reasons to consider it as another plausible possibility. Not only does ChaK appear to radiate from the TPKs, it appears to radiate specifically from the AGC group, with rapid mutational events placing it at a great eventual distance from this group (Figure 6). Though bootstrap support for this origin for ChaK is weak, it is surprisingly strong compared with many other branches, especially given the extreme rearrangements in this structure. Remarkably, searches against the PDB with combinatorial extension (CE) [69] reveal that the strongest structural matches to ChaK are several PKA structures, members of the AGC group of TPKs (strongest match: PDB ID: 1CDK [70], CE Z-score = 4.1, CE RMSD = 4.1Å). By contrast, PI3K does not display such close structural similarity to ChaK (CE Z-score = 3.5, CE RMSD = 4.6Å). Further supporting an AGC group origin for ChaK is the presence of α-Helix B, a structure that is a distinctive feature of the AGC kinases (Figures 1–3 and Table 2).
We speculate that the α-kinases were developed to provide a novel signaling capacity useful to more complex eukaryotic organisms. Given the rapid divergence of the α-kinase family from the rest of the kinase superfamily, and the high level of sequence similarity within the α-kinase family [27], we suggest that the most likely scenario for the creation of the α-kinase family is a single catastrophic genetic event. This event could have perhaps taken the form of deletion of much of the C-terminal end of an extant kinase gene, or fusion of a kinase gene with another gene. While such an event would usually not lead to a functional kinase, this mutation would have produced a kinase that had the novel capability to phosphorylate α-helices.
If the α-kinases were derived from a TPK, it is possible that they contain a zinc finger because this was the way that a functional fold was “rescued” after severe modification of the c-terminal subdomain. It is intriguing that the zinc coordination site in the α-kinases is partly formed by a histidine residue, H1751(F154) in helix D (Helix E) of ChaK. Though H1751 does not structurally align with the conserved H158 seen in the AGC kinases (it is one turn up the helix from H158; Figure 3), it is possible that the presence of a highly conserved histidine in this region of the structure provided part of the initial zinc coordination site in the first α-kinase. Afterward, the location of the helix may have shifted in the α-kinase structure, or the histidine could have been replaced in a point mutation by H1751. Apparently, the first α-kinase underwent a period of rapid sequence change, perhaps to optimize its stability and function. Regardless of the source protein, this process would have led to its distinctive structure and great sequence distance from the TPKs and other AKs (Figures 2, 4, and 6)
Conclusion
The kinase superfamily provides an interesting example of the types of changes seen in proteins over long evolutionary timescales. Lesk and Chothia were the first to perform an in-depth study of protein structure evolution [71]. They described a gradual evolutionary drift of sequence and structure in the globins, but with careful maintenance of the heme binding pocket essential to function.
The changes seen in the kinases are more severe at both the structure and sequence level. It would appear that a major driving force for these large structural changes is the diversity of substrates that kinases from the superfamily must recognize and phosphorylate. Kinase superfamily members phosphorylate an amazing array of targets, from small molecules such as choline (CK) [23], to loop-type regions of proteins (the TPKs) [29], to α-helices (α-kinases) [28], to membrane-bound phosphoinositides (the lipid kinases) [19,21]. The structural changes between families, particularly in the C-terminal subunit, allow for such interactions to take place. In other cases, structural changes have allowed the kinases in the superfamily to partner with accessory domains important to activity and/or regulation (e.g., [21]).
The kinases have been adapted for so many purposes that, in the end, all they have in common is the essential kinase function, and the fold required to carry it out. The large structural shifts seen outside of this region have obliterated sequence similarity outside of the universal core. Even within the core, notable structure and sequence changes have occurred, considering the direct role of this region in the essential function of these enzymes. However, where changes occur to the core that would affect function of the enzyme, there is generally clear compensation for the lost structures and residues, such that function is retained. This sort of plasticity has been previously noted in larger-scale studies of protein evolution [67,72]. The net effect of these sorts of changes is a very low degree of sequence similarity at the superfamily level, even within the core. With such weak sequence similarity between superfamily members, it will not be surprising if other proteins join the superfamily once their structures are solved. A number of divergent kinases have already been identified for which structures are not yet available [35,36].
In this study, we have sought to provide a framework for understanding the development of the kinase superfamily from a common ancestor. By incorporating structural information into our phylogenetic analysis, we have been able to provide a coherent scenario for the evolution of the kinases, with strong support for most of our predictions. Though some areas of kinase structural evolution are still in doubt, we believe the framework provided here will be valuable as structures for more members of the superfamily become available. We expect that many of these structures will be able to provide additional insights into the structural evolution of this rich and expanding superfamily.
Materials and Methods
Construction of the representative set of kinase structures.
We utilized the classification scheme provided by the SCOP [73] and ASTRAL [74] resources (version 1.65) as a guideline for structure selection. To produce a representative set from the SCOP/ASTRAL domains, the sequences for all structures in superfamily d.144.1 (“protein kinase–like”) were clustered via the single-linkage method using BLASTCLUST [75], such that no structure in any cluster could be aligned to a structure in any other cluster with sequence identity ≥ 45%. A single structure was then chosen from each cluster as the structural representative for that group. The choice of a 45% identity cutoff was based on the observation that sequences can be aligned with high accuracy above ~40% identity based on sequence information alone [41,42,76]. Hence, alignments between representative structures from each cluster were likely to benefit from the use of structural information, while structure-based alignment within a cluster would be unlikely to surpass the accuracy achievable with standard sequence alignment techniques. In addition, this filtration ensured that all structures included in the alignment would be evolutionarily divergent, and thus provide interesting information about structural and sequence conservation in the superfamily.
Representative structures were manually selected from each sequence cluster based on the following cascading tests: (1) Structures were favored if they were bound to ATP or an ATP analog, or if (for TPKs) they were in a “closed” conformation [14,16]. Structures bound to ATP (or “closed”) were more informative because their ATP interactions could be studied, they tended to have fully resolved loop regions, and they were easier to align and compare. (2) Higher-resolution structures were favored. (3) Structures with wild-type sequences were favored over structures with experimental sequence mutations.
As discussed in the Introduction, the structure of PIPKIIβ was also added to the set of structures, even though it is not a member of the same SCOP fold group as the other kinases (d.143.1, as opposed to d.144.1). New kinases are constantly being added to the PDB; this representative set was kept unchanged for the duration of the study to maintain the tractability of the dataset.
Structural alignment of kinase representatives.
The representative kinase structures were first aligned using a variant of the CE method [69] modified to provide progressive multiple alignments of protein structures. Using this alignment as a starting point, the alignment was then completely overhauled manually, starting at the N-termini of the proteins and following the structural trace through to the C-termini. No regions were ignored or skipped (i.e., even loops were carefully considered and aligned). The alignment was constructed with the primary aim of maximizing the aligned positions between structures, provided that there was a rational basis for the alignment. This meant, for example, that secondary structural elements could be aligned even if they diverged spatially upon rigid body superposition. We also sometimes used transitive alignments to align portions of structures. This meant that when two elements were distant spatially between a pair of structures, a third structure was considered that provided a “bridge” between the first and second structures. The element could be aligned to the bridge structure for both structures, providing a rational alignment between an otherwise difficult-to-align structural pair. At all times, the alignment was guided by direct visual inspection of the structures, using the CE alignment viewing software [77] and other structure viewers as appropriate. In addition, sequence and structure alignments previously published by kinase experts were used as a guideline [13,17]. Finally, many of the initial publications reporting the structures in the representative set provided alignments to other kinases (see Table 1 for citations). These alignments were also considered where appropriate. Structures were aligned with the goal of providing an optimal alignment between each structure and all other structures in the set, as opposed to one or two other structures (e.g., the closest relative of the structure in question). This process was painstaking, but yielded an extremely high-quality alignment of the protein kinase–like superfamily that considered both structural and functional features. It should be noted that aligning structures with the goal of creating an optimal multiple alignment will, in many cases, produce slightly suboptimal alignments between any given pair of structures (this occurs because often there must be a “compromise” when pairwise alignments of shared structures are not consistent with each other). In practice, this is an issue only in ambiguous regions; the key highly conserved regions can be aligned optimally throughout the superfamily. However, our bias toward maximal alignment of positions and the issue of pairwise suboptimality resulted in relatively high RMSD values (Table 4). Alignments of equivalent segments with an automated method such as CE will often produce lower RMSD, but with fewer aligned positions. However, automated methods such as CE must limit their alignments of ambiguous regions to avoid alignment errors. When creating manual alignments, this limitation is removed. We believe the alignment to be of sufficient quality to serve as a “gold standard” for studying the kinases (and for benchmarking protein structure alignment methods as well). The alignment is available in several formats for download from http://www.sdsc.edu/pb/kinases.
Analysis of the structure and sequence alignment.
The resulting residue equivalences from the manual alignment were used to produce both superpositions of the kinase structures and a corresponding sequence alignment. The sequence alignment was annotated and analyzed using the JOY software [78], which maps structural features onto sequence alignments. In order to standardize the classification of secondary structures, the DSSP [79] method as implemented in sstruc [80] in the JOY software was used as the final arbiter of secondary structure classification (Figure 3).
Analysis of residue conservation was achieved initially by careful visual inspection of the alignment. Conservation at sequence positions within each family was confirmed through the use of Consurf-HSSP [81] conservation data provided through the PDBsum database [82]. Further confirmation as to specific aspects of residue conservation (i.e., conservation of a specific residue to identity, or conservation of a specific property) was accomplished through survey of the family alignments provided in the Pfam database (where available) [56].
Analysis of the structures was performed with molecular viewing software, augmented with the JOY annotation results. The Chimera software [83] was used to create superpositions of structures based on the manual alignment (Table 4). Residues of particular interest were evaluated for hydrogen-bond interactions and other contacts via the CSU server [84].
Phylogenetic tree construction.
The structure-based sequence alignment presented in Figure 3 was used as the basis of all sequence-based portions of the phylogenetic analysis (one TPK structure, Pak1 [85], has a non-wild-type K299R(K72R) substitution, which was reverted to a Lys in our sequence alignment when performing phylogenetic analysis). The tree presented in Figure 4 was constructed using Bayesian phylogenic inference in the program MrBayes [47]. A combined analysis was performed, using both the sequence alignment and the structural characters matrix in Table 2 as “mixed” data [38]. Structural characters were submitted to MrBayes as morphological (“standard”) characters. The characters were modeled as unordered (e.g., a character could change directly from 0 to 2 without having to pass through 1). Both the sequence data and morphology data were modeled with an independent gamma distribution of substitution rates, using the default approximation of four rate classes for each. MrBayes offers a wide selection of model priors for amino acid substitution, and ideally the best-fitting priors should be chosen for the final analysis. Preliminary runs with MrBayes using a mixture of model priors (using the option aamodelprior = mixed in the command prset) demonstrated conclusively that priors based on the substitution rates from the BLOSUM matrices [86] provided the best fit to the sequence alignment data (they had, by far, the highest posterior probability in the analysis). Therefore, the BLOSUM model was used to provide substitution priors for the amino acid sequence portion of the data. Morphological characters were modeled using the default substitution prior for “standard” characters provided in MrBayes. All other settings used in MrBayes were the defaults for the software. The simulation was run for 2,000,000 generations, with tree sampling every 100 generations, for a total of 20,000 trees. At the completion of the run, the “average standard deviation of split frequencies” (a metric in MrBayes to determine convergence of the simulation) was ~0.0084, well below the recommended maximum of 0.1 (MrBayes documentation). A tree was generated using the default methodology and the recommended “burnin” (discarding) of the first 25% of samples (i.e., the tree was generated using the final 15,000 of 20,000 samples). A file containing the input alignment, run settings, and instructions for replication of the MrBayes results is available at http://www.sdsc.edu/pb/kinases.
In order to ascertain the influence of the morphology and sequence datasets on the resulting mixed tree, similar runs were made with MrBayes on the sequence and morphology datasets alone. These runs used identical parameter settings to those for the mixed model for the corresponding datasets (except that they were run for a smaller number of generations). The sequence-only tree was run for 300,000 generations, after which the standard deviation of split frequencies was ~0.037. The structural characters-only tree was run for 500,000 generations, after which the standard deviation of split frequencies was ~0.011. Both runs were processed using the same procedures as above. The resulting trees are provided in Figures S1 and S2, and demonstrate that each of the two methods alone was unable to produce a resolved tree.
Trees produced in PHYLIP [87] used only the sequence alignment data (derived from the structure alignment), and did not consider the structural characters. The alignment was first subjected to bootstrapping via the SEQBOOT program (with default settings), producing 1,000 replicates. Sequence distances were then estimated for each replicate in the program PROTDIST. Since tests with MrBayes indicated that the BLOSUM-based model provided the best fit to the alignment data, distances between sequences were estimated using the PMB model of residue substitution, which is based on the BLOSUM matrices [88]. Substitution rates were modeled as following a gamma distribution, with α = 2.15 (the correct value for α was estimated using a preliminary run of MrBayes with the BLOSUM priors). Trees were constructed for each bootstrap replicate using the Fitch-Margoliash method [46] in the program FITCH. Finally, a single consensus tree was built from the resulting trees in the program CONSENSE, using the default “majority rule (extended)” mode (this method places branches in the final tree when they are seen in > 50% of the input trees; it then places branches with lower representation if they are consistent with the current branches, using cascading selection for highest bootstrap values). Branch lengths were estimated for the resulting tree using the original alignment to determine distances in PROTDIST. These branch lengths were then applied to the consensus tree using FITCH. A copy of the input alignment and instructions for replication of the results is available at http://www.sdsc.edu/pb/kinases.
Supporting Information
Figure S1 Phylogenetic Tree Made with MrBayes, Using Only the Structure-Based Sequence Alignment in Figure 3 as Input
Structures are labeled using a pseudo-ASTRAL ID code, in which positions 2–5 provide the PDB ID code, and the last position provides the specific chain from the PDB file (if applicable). Posterior probabilities are provided to the right of each resolved branch. Numerous polytomies are visible as horizontal branches that are not subdivided by internal branches. Where branches are resolved, posterior probabilities are usually lower than those for the tree in Figure 4. This figure and Figure S2 were created using TreeView [89].
(50 KB TIF)
Click here for additional data file.
Figure S2 Phylogenetic Tree Made with MrBayes, Using Only the Structural Characters Provided in Table 2
Structures are labeled using a pseudo-ASTRAL ID code, in which positions 2–5 provide the PDB ID code, and the last position provides the specific chain from the PDB file (if applicable). Posterior probabilities are provided to the right of each resolved branch. Numerous polytomies are visible as horizontal branches that are not subdivided by internal branches. Though the structural characters provided key information that significantly improved the tree in Figure 4, they are inadequate to discern relationships by themselves, particularly for the TPKs.
(40 KB TIF)
Click here for additional data file.
Accession Numbers
The Protein Data Bank (http://www.rcsb.org/pdb/) accession numbers for proteins discussed in this paper are AFK (1CJA), APH(3′)-IIIa, (1J7U), ChaK (1IA9), CKA-2 (1NW1), Pak1 (1F3M), PI3K (1E8X), PIPKIIβ (1BO1), and PKA (1CDK).
We thank Ilya Shindyalov for assistance with the CE software, Russell Doolittle for helpful discussions, and John Huelsenbeck for helpful discussions and assistance with the MrBayes software. We also thank Natarajan Kannan and Andrew F. Neuwald for helping us to detect an error in our structural alignment between APH(3′)-IIIa/CKA-2 and the TPKs. This work was supported in part from the National Institute of General Medical Sciences (NIGMS) (grant 1GM63208).
Competing interests. The co-author of this manuscript is the editor-in-chief of PLoS Computational Biology.
Author contributions. EDS and PEB conceived and designed the experiments. EDS performed the experiments, analyzed the data, and wrote the paper.
A previous version of this article appeared as an Early Online Release on September 8, 2005 (DOI: 10.1371/journal.pcbi.0010049.eor).
Abbreviations
AFKactin-fragmin kinase
AKatypical kinase
ChaKchannel kinase
Chk1cell cycle checkpoint kinase
CKcholine kinase
CKA-2choline kinase isoform A-2
PDBProtein Data Bank
PIphosphatidylinositol
PI3Kphosphoinositide 3-kinase
PIPKIIβtype IIβ phosphatidylinositol phosphate kinase
PKAprotein kinase A
RMSDroot mean square deviation
SCOPStructural Classification Of Proteins
TKLtyrosine kinase–like
TPKtypical protein kinase
==== Refs
References
Murzin AG Brenner SE Hubbard T Chothia C 1995 SCOP: A structural classification of proteins database for the investigation of sequences and structures J Mol Biol 247 536 540 7723011
Muller A MacCallum RM Sternberg MJ 1999 Benchmarking PSI-BLAST in genome annotation J Mol Biol 293 1257 1271 10547299
Chothia C Lesk AM 1986 The relation between the divergence of sequence and structure in proteins Embo J 5 823 826 3709526
Hon WC McKay GA Thompson PR Sweet RM Yang DS 1997 Structure of an enzyme required for aminoglycoside antibiotic resistance reveals homology to eukaryotic protein kinases Cell 89 887 895 9200607
Hanks SK 2003 Genomic analysis of the eukaryotic protein kinase superfamily: A perspective Genome Biol 4 111 12734000
Manning G Plowman GD Hunter T Sudarsanam S 2002 Evolution of protein kinase signaling from yeast to man Trends Biochem Sci 27 514 520 12368087
Manning G Whyte DB Martinez R Hunter T Sudarsanam S 2002 The protein kinase complement of the human genome Science 298 1912 1934 12471243
Leonard CJ Aravind L Koonin EV 1998 Novel families of putative protein kinases in bacteria and archaea: Evolution of the “eukaryotic” protein kinase superfamily Genome Res 8 1038 1047 9799791
Kennelly PJ 2002 Protein kinases and protein phosphatases in prokaryotes: A genomic perspective FEMS Microbiol Lett 206 1 8 11786249
Kennelly PJ 2003 Archaeal protein kinases and protein phosphatases: Insights from genomics and biochemistry Biochem J 370 373 389 12444920
Shi L Potts M Kennelly PJ 1998 The serine, threonine, and/or tyrosine-specific protein kinases and protein phosphatases of prokaryotic organisms: A family portrait FEMS Microbiol Rev 22 229 253 9862122
Knighton DR Zheng JH Ten Eyck LF Ashford VA Xuong NH 1991 Crystal structure of the catalytic subunit of cyclic adenosine monophosphate-dependent protein kinase Science 253 407 414 1862342
Taylor SS Radzio-Andzelm E 1994 Three protein kinase structures define a common motif Structure 2 345 355 8081750
Cox S Radzio-Andzelm E Taylor SS 1994 Domain movements in protein kinases Curr Opin Struct Biol 4 893 901 7712293
Huse M Kuriyan J 2002 The conformational plasticity of protein kinases Cell 109 275 282 12015977
Sowadski JM Epstein LF Lankiewicz L Karlsson R 1999 Conformational diversity of catalytic cores of protein kinases Pharmacol Ther 82 157 164 10454194
Hanks SK Hunter T 1995 Protein kinases 6. The eukaryotic protein kinase superfamily: Kinase (catalytic) domain structure and classification Faseb J 9 576 596 7768349
Berman HM Westbrook J Feng Z Gilliland G Bhat TN 2000 The Protein Data Bank Nucleic Acids Res 28 235 242 10592235
Rao VD Misra S Boronenkov IV Anderson RA Hurley JH 1998 Structure of type IIbeta phosphatidylinositol phosphate kinase: A protein kinase fold flattened for interfacial phosphorylation Cell 94 829 839 9753329
Yamaguchi H Matsushita M Nairn AC Kuriyan J 2001 Crystal structure of the atypical protein kinase domain of a TRP channel with phosphotransferase activity Mol Cell 7 1047 1057 11389851
Walker EH Perisic O Ried C Stephens L Williams RL 1999 Structural insights into phosphoinositide 3-kinase catalysis and signalling Nature 402 313 320 10580505
Steinbacher S Hof P Eichinger L Schleicher M Gettemans J 1999 The crystal structure of the Physarum polycephalum actin-fragmin kinase: An atypical protein kinase with a specialized substrate-binding domain Embo J 18 2923 2929 10357805
Peisach D Gee P Kent C Xu Z 2003 The crystal structure of choline kinase reveals a eukaryotic protein kinase fold Structure (Camb) 11 703 713 12791258
Burk DL Hon WC Leung AK Berghuis AM 2001 Structural analyses of nucleotide binding to an aminoglycoside phosphotransferase Biochemistry 40 8756 8764 11467935
Walsh C 2000 Molecular mechanisms that confer antibacterial drug resistance Nature 406 775 781 10963607
Exton JH 1994 Phosphatidylcholine breakdown and signal transduction Biochim Biophys Acta 1212 26 42 8155724
Drennan D Ryazanov AG 2004 Alpha-kinases: Analysis of the family and comparison with conventional protein kinases Prog Biophys Mol Biol 85 1 32 15050379
Ryazanov AG Pavur KS Dorovkov MV 1999 Alpha-kinases: A new class of protein kinases with a novel catalytic domain Curr Biol 9 R43 45
Pinna LA Ruzzene M 1996 How do protein kinases recognize their substrates? Biochim Biophys Acta 1314 191 225 8982275
Domin J Waterfield MD 1997 Using structure to define the function of phosphoinositide 3-kinase family members FEBS Lett 410 91 95 9247130
Cantley LC 2002 The phosphoinositide 3-kinase pathway Science 296 1655 1657 12040186
De Corte V Gettemans J Waelkens E Vandekerckhove J 1996 In vivo phosphorylation of actin in Physarum polycephalum . Study of the substrate specificity of the actin-fragmin kinase Eur J Biochem 241 901 908 8944781
Doughman RL Firestone AJ Anderson RA 2003 Phosphatidylinositol phosphate kinases put PI4,5P2 in its place J Membr Biol 194 77 89 14502432
Grishin NV 1999 Phosphatidylinositol phosphate kinase: A link between protein kinase and glutathione synthase folds J Mol Biol 291 239 247 10438618
Cheek S Ginalski K Zhang H Grishin NV 2005 A comprehensive update of the sequence and structure classification of kinases BMC Struct Biol 5 6 15771780
Cheek S Zhang H Grishin NV 2002 Sequence and structure classification of kinases J Mol Biol 320 855 881 12095261
Huelsenbeck JP Ronquist F Nielsen R Bollback JP 2001 Bayesian inference of phylogeny and its impact on evolutionary biology Science 294 2310 2314 11743192
Ronquist F Huelsenbeck JP 2003 MrBayes 3: Bayesian phylogenetic inference under mixed models Bioinformatics 19 1572 1574 12912839
Nylander JA Ronquist F Huelsenbeck JP Nieves-Aldrey JL 2004 Bayesian phylogenetic analysis of combined data Syst Biol 53 47 67 14965900
Aravind L Anantharaman V Koonin EV 2002 Monophyly of class I aminoacyl tRNA synthetase, USPA, ETFP, photolyase, and PP-ATPase nucleotide-binding domains: Implications for protein evolution in the RNA Proteins 48 1 14 12012333
Sauder JM Arthur JW Dunbrack RL Jr. 2000 Large-scale comparison of protein sequence alignment algorithms with structure alignments Proteins 40 6 22 10813826
Elofsson A 2002 A study on protein sequence alignment quality Proteins 46 330 339 11835508
Eidhammer I Jonassen I Taylor WR 2000 Structure comparison and structure patterns J Comput Biol 7 685 716 11153094
Feng ZK Sippl MJ 1996 Optimum superimposition of protein structures: Ambiguities and implications Fold Des 1 123 132 9079372
Godzik A 1996 The structural alignment between two proteins: Is there a unique answer? Protein Sci 5 1325 1338 8819165
Fitch WM Margoliash E 1967 Construction of phylogenetic trees Science 155 279 284 5334057
Huelsenbeck JP Ronquist F 2001 MRBAYES: Bayesian inference of phylogenetic trees Bioinformatics 17 754 755 11524383
Pastore A Lesk AM 1990 Comparison of the structures of globins and phycocyanins: Evidence for evolutionary relationship Proteins 8 133 155 2235993
Murzin AG 1998 How far divergent evolution goes in proteins Curr Opin Struct Biol 8 380 387 9666335
Xu W Doshi A Lei M Eck MJ Harrison SC 1999 Crystal structures of c-Src reveal features of its autoinhibitory mechanism Mol Cell 3 629 638 10360179
Cheetham GM Knegtel RM Coll JT Renwick SB Swenson L 2002 Crystal structure of aurora-2, an oncogenic serine/threonine kinase J Biol Chem 277 42419 42422 12237287
Krupa A Preethi G Srinivasan N 2004 Structural modes of stabilization of permissive phosphorylation sites in protein kinases: Distinct strategies in Ser/Thr and Tyr kinases J Mol Biol 339 1025 1039 15178245
Johnson LN Noble ME Owen DJ 1996 Active and inactive protein kinases: Structural basis for regulation Cell 85 149 158 8612268
Nolen B Taylor S Ghosh G 2004 Regulation of protein kinases; controlling activity through activation segment conformation Mol Cell 15 661 675 15350212
Krishna SS Majumdar I Grishin NV 2003 Structural classification of zinc fingers: Survey and summary Nucleic Acids Res 31 532 550 12527760
Bateman A Coin L Durbin R Finn RD Hollich V 2004 The Pfam protein families database Nucleic Acids Res 32 D138 141 14681378
Erixon P Svennblad B Britton T Oxelman B 2003 Reliability of Bayesian posterior probabilities and bootstrap frequencies in phylogenetics Syst Biol 52 665 673 14530133
Hillis DM Bull JJ 1993 An empirical test of bootstrapping as a method for assessing confidence in phylogenetic analysis Syst Biol 42 182 192
Ronquist F 2004 Bayesian inference of character evolution Trends Ecol Evol 19 475 481 16701310
Ponting CP Aravind L Schultz J Bork P Koonin EV 1999 Eukaryotic signalling domain homologues in archaea and bacteria. Ancient ancestry and horizontal gene transfer J Mol Biol 289 729 745 10369758
Ortiz-Lombardia M Pompeo F Boitel B Alzari PM 2003 Crystal structure of the catalytic domain of the PknB serine/threonine kinase from Mycobacterium tuberculosis J Biol Chem 278 13094 13100 12551895
Han G Zhang CC 2001 On the origin of Ser/Thr kinases in a prokaryote FEMS Microbiol Lett 200 79 84 11410353
Chen Y Sanchez Y 2004 Chk1 in the DNA damage response: Conserved roles from yeasts to mammals DNA Repair (Amst) 3 1025 1032 15279789
Daigle DM McKay GA Thompson PR Wright GD 1999 Aminoglycoside antibiotic phosphotransferases are also serine protein kinases Chem Biol 6 11 18 9889150
Foukas LC Shepherd PR 2004 Phosphoinositide 3-kinase: The protein kinase that time forgot Biochem Soc Trans 32 330 331 15046601
Stack JH Emr SD 1994 Vps34p required for yeast vacuolar protein sorting is a multiple specificity kinase that exhibits both protein kinase and phosphatidylinositol-specific PI 3-kinase activities J Biol Chem 269 31552 31562 7989323
Todd AE Orengo CA Thornton JM 2001 Evolution of function in protein superfamilies, from a structural perspective J Mol Biol 307 1113 1143 11286560
Abraham RT 2004 PI 3-kinase related kinases: ‘Big' players in stress-induced signaling pathways DNA Repair (Amst) 3 883 887 15279773
Shindyalov IN Bourne PE 1998 Protein structure alignment by incremental combinatorial extension (CE) of the optimal path Protein Eng 11 739 747 9796821
Bossemeyer D Engh RA Kinzel V Ponstingl H Huber R 1993 Phosphotransferase and substrate binding mechanism of the cAMP-dependent protein kinase catalytic subunit from porcine heart as deduced from the 2.0 A structure of the complex with Mn2+ adenylyl imidodiphosphate and inhibitor peptide PKI(5–24) Embo J 12 849 859 8384554
Lesk AM Chothia C 1980 How different amino acid sequences determine similar protein structures: The structure and evolutionary dynamics of the globins J Mol Biol 136 225 270 7373651
Todd AE Orengo CA Thornton JM 2002 Plasticity of enzyme active sites Trends Biochem Sci 27 419 426 12151227
Lo Conte L Ailey B Hubbard TJ Brenner SE Murzin AG 2000 SCOP: A structural classification of proteins database Nucleic Acids Res 28 257 259 10592240
Brenner SE Koehl P Levitt M 2000 The ASTRAL compendium for protein structure and sequence analysis Nucleic Acids Res 28 254 256 10592239
Altschul SF Madden TL Schaffer AA Zhang J Zhang Z 1997 Gapped BLAST and PSI-BLAST: A new generation of protein database search programs Nucleic Acids Res 25 3389 3402 9254694
Vogt G Etzold T Argos P 1995 An assessment of amino acid exchange matrices in aligning protein sequences: The twilight zone revisited J Mol Biol 249 816 831 7602593
Shindyalov IN Bourne PE 2001 A database and tools for 3-D protein structure comparison and alignment using the Combinatorial Extension (CE) algorithm Nucleic Acids Res 29 228 229 11125099
Mizuguchi K Deane CM Blundell TL Johnson MS Overington JP 1998 JOY: Protein sequence-structure representation and analysis Bioinformatics 14 617 623 9730927
Kabsch W Sander C 1983 Dictionary of protein secondary structure: Pattern recognition of hydrogen-bonded and geometrical features Biopolymers 22 2577 2637 6667333
Smith DK 1989 SSTRUC [computer program] Department of Biochemistry and Molecular Biology, University College London
Glaser F Rosenberg Y Kessel A Pupko T Ben-Tal N 2005 The ConSurf-HSSP database: The mapping of evolutionary conservation among homologs onto PDB structures Proteins 58 610 617 15614759
Laskowski RA Chistyakov VV Thornton JM 2005 PDBsum more: New summaries and analyses of the known 3D structures of proteins and nucleic acids Nucleic Acids Res 33 D266 268 15608193
Pettersen EF Goddard TD Huang CC Couch GS Greenblatt DM 2004 UCSF Chimera—A visualization system for exploratory research and analysis J Comput Chem 25 1605 1612 15264254
Sobolev V Sorokine A Prilusky J Abola EE Edelman M 1999 Automated analysis of interatomic contacts in proteins Bioinformatics 15 327 332 10320401
Lei M Lu W Meng W Parrini MC Eck MJ 2000 Structure of PAK1 in an autoinhibited conformation reveals a multistage activation switch Cell 102 387 397 10975528
Henikoff S Henikoff JG 1992 Amino acid substitution matrices from protein blocks Proc Natl Acad Sci U S A 89 10915 10919 1438297
Felsenstein J 2004 PHYLIP (Phylogeny Inference Package), version 3.62 [computer program]. Department of Genome Sciences, University of Washington, Seattle Available: http://evolution.genetics.washington.edu/phylip.html . Accessed 22 September 2005.
Veerassamy S Smith A Tillier ER 2003 A transition probability model for amino acid substitutions from blocks J Comput Biol 10 997 1010 14980022
Page RD 1996 TreeView: An application to display phylogenetic trees on personal computers Comput Appl Biosci 12 357 358 8902363
Kraulis PJ 1991 Molscript—A program to produce both detailed and schematic plots of protein structures J Appl Crystallogr 24 946 950
Yang J Cron P Good VM Thompson V Hemmings BA 2002 Crystal structure of an activated Akt/protein kinase B ternary complex with GSK3-peptide and AMP-PNP Nat Struct Biol 9 940 944 12434148
Lodowski DT Pitcher JA Capel WD Lefkowitz RJ Tesmer JJ 2003 Keeping G proteins at bay: A complex between G protein-coupled receptor kinase 2 and Gbetagamma Science 300 1256 1262 12764189
Biondi RM Komander D Thomas CC Lizcano JM Deak M 2002 High resolution crystal structure of the human PDK1 catalytic domain defines the regulatory phosphopeptide docking site Embo J 21 4219 4228 12169624
Mayans O van der Ven PF Wilm M Mues A Young P 1998 Structural basis for activation of the titin kinase domain during myofibrillogenesis Nature 395 863 869 9804419
Tereshko V Teplova M Brunzelle J Watterson DM Egli M 2001 Crystal structures of the catalytic domain of human protein kinase associated with apoptosis and tumor suppression Nat Struct Biol 8 899 907 11573098
Goldberg J Nairn AC Kuriyan J 1996 Structural basis for the autoinhibition of calcium/calmodulin-dependent protein kinase I Cell 84 875 887 8601311
Owen DJ Noble ME Garman EF Papageorgiou AC Johnson LN 1995 Two structures of the catalytic domain of phosphorylase kinase: An active protein kinase complexed with substrate analogue and product Structure 3 467 482 7663944
Meng W Swenson LL Fitzgibbon MJ Hayakawa K Ter Haar E 2002 Structure of mitogen-activated protein kinase-activated protein (MAPKAP) kinase 2 suggests a bifunctional switch that couples kinase activation with nuclear export J Biol Chem 277 37401 37405 12171911
Chen P Luo C Deng Y Ryan K Register J 2000 The 1.7 A crystal structure of human cell cycle checkpoint kinase Chk1: Implications for Chk1 regulation Cell 100 681 692 10761933
Bax B Carter PS Lewis C Guy AR Bridges A 2001 The structure of phosphorylated GSK-3beta complexed with a peptide, FRATtide, that inhibits beta-catenin phosphorylation Structure (Camb) 9 1143 1152 11738041
Schulze-Gahmen U De Bondt HL Kim SH 1996 High-resolution crystal structures of human cyclin-dependent kinase 2 with and without ATP: Bound waters and natural ligand as guides for inhibitor design J Med Chem 39 4540 4546 8917641
Xie X Gu Y Fox T Coll JT Fleming MA 1998 Crystal structure of JNK3: A kinase implicated in neuronal apoptosis Structure 6 983 991 9739089
Nolen B Yun CY Wong CF McCammon JA Fu XD 2001 The structure of Sky1p reveals a novel mechanism for constitutive activity Nat Struct Biol 8 176 183 11175909
Niefind K Guerra B Pinna LA Issinger OG Schomburg D 1998 Crystal structure of the catalytic subunit of protein kinase CK2 from Zea mays at 2.1 A resolution Embo J 17 2451 2462 9564028
Xu RM Carmel G Sweet RM Kuret J Cheng X 1995 Crystal structure of casein kinase-1, a phosphate-directed protein kinase Embo J 14 1015 1023 7889932
Huse M Chen YG Massague J Kuriyan J 1999 Crystal structure of the cytoplasmic domain of the type I TGF beta receptor in complex with FKBP12 Cell 96 425 436 10025408
Till JH Becerra M Watty A Lu Y Ma Y 2002 Crystal structure of the MuSK tyrosine kinase: Insights into receptor autoregulation Structure (Camb) 10 1187 1196 12220490
Hubbard SR 1997 Crystal structure of the activated insulin receptor tyrosine kinase in complex with peptide substrate and ATP analog Embo J 16 5572 5581 9312016
Stamos J Sliwkowski MX Eigenbrot C 2002 Structure of the epidermal growth factor receptor kinase domain alone and in complex with a 4-anilinoquinazoline inhibitor J Biol Chem 277 46265 46272 12196540
Boeckmann B Bairoch A Apweiler R Blatter MC Estreicher A 2003 The SWISS-PROT protein knowledgebase and its supplement TrEMBL in 2003 Nucleic Acids Res 31 365 370 12520024
|
16244704
|
PMC1261164
|
CC BY
|
2021-01-05 09:18:23
|
no
|
PLoS Comput Biol. 2005 Oct 21; 1(5):e49
|
utf-8
|
PLoS Comput Biol
| 2,005 |
10.1371/journal.pcbi.0010049
|
oa_comm
|
==== Front
BMC BiolBMC Biology1741-7007BioMed Central London 1741-7007-3-201618803210.1186/1741-7007-3-20Research ArticleThe sequence of rice chromosomes 11 and 12, rich in disease resistance genes and recent gene duplications The Rice Chromosomes 11 and 12 Sequencing Consortia* [email protected] Waksman Institute, Rutgers University, Piscataway, New Jersey 088542005 27 9 2005 3 20 20 11 7 2005 27 9 2005 Copyright © 2005 Messing and The Rice Chromosomes 11 and 12 Sequencing Consortia*; licensee BioMed Central Ltd.2005Messing and The Rice Chromosomes 11 and 12 Sequencing Consortia*; 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
Rice is an important staple food and, with the smallest cereal genome, serves as a reference species for studies on the evolution of cereals and other grasses. Therefore, decoding its entire genome will be a prerequisite for applied and basic research on this species and all other cereals.
Results
We have determined and analyzed the complete sequences of two of its chromosomes, 11 and 12, which total 55.9 Mb (14.3% of the entire genome length), based on a set of overlapping clones. A total of 5,993 non-transposable element related genes are present on these chromosomes. Among them are 289 disease resistance-like and 28 defense-response genes, a higher proportion of these categories than on any other rice chromosome. A three-Mb segment on both chromosomes resulted from a duplication 7.7 million years ago (mya), the most recent large-scale duplication in the rice genome. Paralogous gene copies within this segmental duplication can be aligned with genomic assemblies from sorghum and maize. Although these gene copies are preserved on both chromosomes, their expression patterns have diverged. When the gene order of rice chromosomes 11 and 12 was compared to wheat gene loci, significant synteny between these orthologous regions was detected, illustrating the presence of conserved genes alternating with recently evolved genes.
Conclusion
Because the resistance and defense response genes, enriched on these chromosomes relative to the whole genome, also occur in clusters, they provide a preferred target for breeding durable disease resistance in rice and the isolation of their allelic variants. The recent duplication of a large chromosomal segment coupled with the high density of disease resistance gene clusters makes this the most recently evolved part of the rice genome. Based on syntenic alignments of these chromosomes, rice chromosome 11 and 12 do not appear to have resulted from a single whole-genome duplication event as previously suggested.
==== Body
Background
Rice (Oryza sativa) is a major staple food and is consumed by nearly half the world's population. It accounts for more than 21% of global human per capita energy and 15% of per capita protein. In the past few decades, although rice production has doubled due to the introduction of high yielding varieties/hybrids and improved cultivation practices, it is still insufficient to cope with ever-increasing global demand, which is expected to increase at the rate of about 1% per annum [1]. At the same time, inappropriate natural resource use, along with biotic and abiotic stress pressure, is casting a shadow on rice production [2]. Access to the rice genome sequence will enable identification of genes responsible for traits and alleles that will be essential to meet the growing demands of food production in the coming years. Towards this end, whole-genome shotgun-based draft sequences of the indica and japonica subspecies of rice were reported previously [3-5] while the International Rice Genome Sequencing Project (IRGSP), using a clone-by-clone approach, focused on generating a high quality, finished sequence of the rice genome [6]. Indeed, access to the rice genome has served as a catalyst for investigations on comparative genomics, functional genomics, map-based gene cloning and molecular breeding in rice [7,8].
Rice is one of the cereals in the Poaceae family, which collectively provides the largest source of calories for human consumption. Within the cereals are larger genome species such as maize, wheat, millet and sorghum [9]. Synteny in the cereals has been reported previously using molecular markers [10] and recently at a higher resolution using sequences available from the rice, sorghum, maize, and wheat genomes [3,11,14]. Thus, access to a complete, high-resolution rice genome sequence will facilitate research on other cereals with larger, partially sequenced genomes.
One of the emerging features of plant genomes appears to be the recent generation of gene copies that have diverged in regulation and function [15]. A common pathway for such a mechanism in plants is whole-genome duplication (WGD), which occurred for instance in maize as little as 5 million years ago [16]. Other mechanisms involve tandem gene amplifications and segmental duplications. In rice, chromosomes 11 and 12 provide such an example. In addition, several genes of agronomic importance, such as blast, bacterial blight, virus and insect resistance, photoperiod-sensitive male-sterility as well as salt tolerance, have been mapped on these two chromosomes [17]. Here, we report the in-depth analysis of chromosomes 11 and 12 of a japonica cultivar of rice with genetic lengths estimated to be 118.6 cM and 110.1 cM, respectively [18]. We have generated high-quality finished sequences for these two chromosomes, annotated the chromosomes for genes and other features, and used these data to examine novel features such as the organization of disease resistance genes. We also asked whether chromosomes 11 and 12 are likely to have resulted from a WGD event by examining duplication events between the two chromosomes through alignments of mapped genes of wheat along the rice chromosomes.
Results and discussion
General features of chromosomes 11 and 12
A total of 255 and 269 BAC/PAC clones were sequenced from chromosomes 11 and 12, respectively. Chromosome 11 is slightly longer than chromosome 12, with 28.4 Mb and 27.5 Mb, respectively, of sequence identified for them (Table 1, Fig. 1). Excluding the telomeres, a few physical gaps remain on the two pseudomolecules; six and one on chromosomes 11 and 12, respectively (Accession Numbers: chromosome 11: DP000010, chromosome 12: DP000011). This was due to a difficulty in obtaining clones that span these regions, especially the centromeres and telomeres, which present technical challenges in sequencing highly repetitive tracts of the genome.
Table 1 Statistics of rice chromosomes 11 and 12
Feature Statistics
Chr11 Chr12
Total number of BACs/PACs 255 269
Total BAC length (Mb) 35.5 33.5
Total nonoverlapping sequence (Mb) 28.4 27.5
Short arm (Mb) 12.0 12.0
Long arm (Mb) 16.4 15.5
Integrated genetic markers 283(loci) 269(loci)
G+C Content
Overall 42.8 43.0
Exons 51.6 51.8
Introns 38.7 38.8
Intergenic regions 40.8 41.0
Total number of genesa 4,286(3,148) 4,169(2,845)
Known/Putative genes 2,364(55.2%) 2,395(57.4%)
Expressed genes 328(7.7%) 336(8.1%)
Hypothetical genes 1,594(37.2) 1,438(34.5%)
Gene density (kb)b 6.6(9.0) 6.6(9.7)
Average gene length (bp)c 2,603 2,589
Total number of gene models 4,436 4,355
Average exon size (bp) 350 339
Average intron size (bp) 378 371
Average number exons per gene model 4.1 4.2
a Total number of genes includes the TE-related genes and the non-TE-related genes.
b Gene density is reported for all genes and non-TE-related genes (in parentheses).
c Average length of the genes is determined using the length of from the start codon to stop codon of the longest isoform.
Figure 1 Display of features on rice chromosomes 11 and 12. Using a false-color display, we plotted features present on rice chromosomes 11 and 12. Select genetic markers are plotted with their cM positions noted; physical gaps are plotted with the centromere gap noted in red; gene density (GD) is plotted in two tracks (all genes and non-TE-related genes); expression density is determined by aligment to ESTs; transposable element (TE) density is plotted with a separate track for MITEs; tRNAs are the transfer RNAs, and the CP and MT represent chloroplast and mitochondrial insertions.
Of the 4,286 and 4,169 genes identified on chromosomes 11 and 12, respectively, 3,148 and 2,845 could be annotated as non-transposable element related (non-TE; Fig. 1). A similar percentage of genes on the two chromosomes could be assigned a putative function or annotated as encoding an expressed protein, leaving a similar percentage annotated as encoding a hypothetical protein (Table 1). With respect to domain composition, there was a striking difference in representation of some Pfam domains between the chromosomes, suggesting the presence of genes coding for different suites of proteins on the two chromosomes. As shown in Tables 2 and 3, there was a large enrichment of proteins on chromosome 11 (166 proteins) containing a leucine-rich repeat (LRR) domain, which is involved in protein-protein recognition and is a hallmark of disease resistance genes (see below). Besides the LRR-domain, there are two other domains common to disease resistance genes, protein kinase and NB-ARC domains, which were also enriched on chromosome 11 relative to chromosome 12; 106 versus 74 proteins containing protein kinase domains and 102 versus 49 proteins with NB-ARC domains on chromosome 11 versus 12, respectively. Based on alignments by a rice transcript, 1,235 (39%) and 1,221 (43%) of non-TE-related genes on chromosomes 11 and 12, respectively, were active (Additional figure 10 [see Additional file 1]). Of these, 952 and 980 non-TE-related genes from chromosomes 11 and 12, respectively, could be aligned with a FL-cDNA. In addition, 152 non-TE-related genes aligned with ESTs from other monocots on each chromosome. For 26 genes from chromosome 11 and 23 from chromosome 12, only non-monocot evidence was available for expression (Additional figure 10 [see Additional file 1]). A ready set of mutants is available for 800 and 845 genes on chromosomes 11 and 12, respectively, as evidenced by a flanking sequence tag (FST) in or within 500 bp of the transcription unit.
Table 2 Predominant Pfam domains within the chromosome 11 predicted rice proteome. All non-TE related proteins were searched using the Hmmpfam program and Pfam domains above the trusted cutoff were parsed out. Only the top 20 Pfam domains are listed.
Pfam accession Numbers of matched proteins Pfam common name
PF00560 166 Leucine Rich Repeat
PF00069 106 Protein kinase domain
PF00931 102 NB-ARC domain
PF00646 78 F-box domain
PF00023 32 Ankyrin repeat
PF00097 29 Zinc finger, C3HC4 type (RING finger)
PF01535 23 PPR repeat
PF00098 21 Zinc knuckle
PF05699 19 hAT family dimerisation domain
PF06654 19 Protein of unknown function (DUF1165)
PF07197 19 Protein of unknown function (DUF1409)
PF04578 14 Protein of unknown function, DUF594
PF00036 13 EF hand
PF00076 13 RNA recognition motif. (a.k.a. RRM, RBD, or RNP domain)
PF03018 13 Dirigent-like protein
PF00067 12 Cytochrome P450
PF00400 12 WD domain, G-beta repeat
PF00651 12 BTB/POZ domain
PF00704 12 glycosyl hydrolase, family 18
PF00106 11 oxidoreductase, short chain dehydrogenase/reductase family
Table 3 Predominant Pfam domains within the chromosome 12 predicted rice proteome. All non-TE related proteins were searched using the Hmmpfam program and Pfam domains above the trusted cutoff were parsed out. Only the top 20 Pfam domains are listed.
Pfam accession Numbers of matched proteins Pfam common name
PF00069 74 Protein kinase domain
PF00560 60 Leucine Rich Repeat
PF00646 51 F-box domain
PF00931 49 NB-ARC domain
PF00098 28 Zinc knuckle
PF01535 28 PPR repeat
PF03578 27 HGWP repeat
PF00097 25 Zinc finger, C3HC4 type (RING finger)
PF05699 22 hAT family dimerisation domain
PF00036 18 EF hand
PF00249 16 Myb-like DNA-binding domain
PF00067 14 Cytochrome P450
PF00076 13 RNA recognition motif. (a.k.a. RRM, RBD, or RNP domain)
PF00004 11 ATPase, AAA family
PF00010 11 Helix-loop-helix DNA-binding domain
PF00400 11 WD domain, G-beta repeat
PF02892 11 BED zinc finger
PF06654 11 Protein of unknown function (DUF1165)
PF07197 11 Protein of unknown function (DUF1409)
PF00023 10 Ankyrin repeat
A similar fraction of chromosomes 11 and 12 could be classified as repetitive (29.5% and 31.6%, respectively (Additional tables 5 and 6 [see Additional file 1]), which with the exception of miniature inverted repeat transposable elements (MITEs) were enriched in the centromeric and pericentromeric regions of the two chromosomes (Fig. 1). As noted previously [19], MITEs, which constitute 18.6 and 16.2% of the repetitive sequences on chromosomes 11 and 12, respectively, were enriched in the euchromatic arms of the two chromosomes (Fig. 1).
A similar number of non-TE-related proteins with putative homologs in model species were present in chromosomes 11 and 12 (Fig. 2). The highest number of putative homologs was seen in Arabidopsis, with 60.9% and 59.6% of the proteins from chromosomes 11 and 12, respectively, having a potential homolog in Arabidopsis (E value cutoff 10-5). Similar ranges of potential homologs were seen between the two chromosomes and the other model organisms, ranging from 7–30.6% in chromosome 11 and 10–33.6% in chromosome 12.
Figure 2 Presence of potential homologs in rice chromosomes 11 and 12 with other model organisms. BLASTP was used to search the proteomes of the listed model organisms and with the non-TE-related proteins from chromosomes 11 and 12. The number of proteins with matches at the designated E-value cutoffs are plotted for each model organism.
A high percentage of rice genes could be aligned with genomic assemblies from two other cereals, maize and sorghum. Of the 2,405 maize assemblies with a best hit on rice chromosome 11 or 12, 1,879 overlap with non-TE-related genes while 160 overlap with TE-related genes (Additional table 7 [see Additional file 1]). The remaining 366 assemblies align to intergenic regions, which might include genes or TEs that are degenerated and not recognized by our bioinformatics methods. Interestingly, for sorghum, the occurrence of alignments was similar to maize except that there was a higher incidence of alignment to TE or intergenic regions, which could be due the fact that the sorghum dataset includes only methyl-filtrated and not Cot-enriched sequences [20-22]. The 1,469 non-TE-related genes that aligned to both maize and sorghum genomic assemblies suggest that these genes might predate the divergence of the Panicoideae, estimated at 50 mya [23].
Disease resistance genes in rice chromosomes 11 and 12
Disease resistance genes (R-like genes) conferring resistance to viral, bacterial, fungal and nematode pathogens have been grouped into five classes on the basis of their encoded protein products [24]. For instance, a genome-wide study identified such typical protein domains in the R-like genes of Arabidopsis [25]. Major classes of conserved domains include leucine zipper (LZ), coiled coil (CC), nucleotide binding site (NBS), leucine-rich repeat (LRR), protein kinase, Toll-IL-IR homology region (TIR) and trans-membrane (TM) as well as miscellaneous R-genes. Apart from the genes containing these typical domains, one maize disease resistance gene, HM1, encodes a reductase that detoxifies HC-toxin [26]. We analyzed the sequences of chromosomes 11 and 12 to identify the type and distribution of the disease resistance as well as downstream defense response genes (Fig. 3). We identified 201 R-like gene models in chromosome 11 (Additional table 8 [see Additional file 1]), which is 4.5% of the total number of gene models predicted for this chromosome. Of these, 73 (36.3%) have homology to the NBS-LRR class of R-like genes and 38 show homology to the LRR-TM-like genes. The long arm of rice chromosome 11 (11L) contains almost twice (132 genes) the number of R-like genes compared to the short arm (68 genes). We also identified 17 downstream defense response genes including glucanases, chitinases and thaumatin-like proteins. Most of these R-like genes and defense response-like genes are present in large clusters of tandem arrays indicating their origin by duplication from a few ancestral genes (Fig. 4). Large clusters of genes (31, 35 and 37 members) are present between positions 30–40 cM (between 5,658,160 – 7,920,402 bp), 80–90 cM (between 20,001,952 – 22,632,644 bp) and 110–119 cM (between 26,213,144 – 28,180,239 bp) of chromosome 11, respectively. Nucleotide positions are based on the pseudomolecules of TIGR release 3 [27]. A detailed depiction of duplicated genes in the interval 112–119 cM on chromosome 11 is shown in Fig. 5 and a cladogram of these genes is shown in Additional figure 11 [see Additional file 1], confirming that the genes within a cluster generally belong to the same clade. A tandem array of 12 Xa21-like genes is present on the short arm of chromosome 11 at position 19 cM (between 3,516,492 – 3,665,074 bp), while the actual Xa21 gene is present on the long arm of chromosome 11 [28]. A large cluster of 14 defense response genes, 12 of which are chitinases, is present in tandem at 116.2 cM (between 28,056,455 – 28,122,601 bp). Several other R-like genes are also arranged in similar but small clusters.
Figure 3 Frequency distribution of different categories of R-like genes and defense response genes on rice chromosomes 11 and 12. The miscellaneous category includes genes showing high homology with putative disease resistance proteins and defense response category includes genes showing high homology to chitinases, glucanases and thaumatin-like proteins. (LZ, leucine zipper; NBS, nucleotide binding site; LRR, leucine rich repeat; TM, trans-membrane).
Figure 4 Distribution pattern of resistance genes and defense response genes on rice chromosomes 11 (A) and 12 (B). Each gene category is color coded and plotted on the rice chromosome bar with respect to its cM position. Width of the vertical colored bars is proportional to the number of genes located at that position.
Figure 5 Plot of a portion of the rice chromosome 11 showing tandem arrays of disease resistance and defense response genes between positions 112.0 – 119.0 cM. The category of genes is color-coded and the arrowheads depict their direction. The numbers indicate cumulative number of all the genes predicted by TIGR on chromosome 11. The scale is based on number of genes such that the space occupied by one arrowhead corresponds to one gene, genes in the gap between arrowheads do not match with R-like genes and large gaps of unmatched genes are marked by a double slash (//).
On chromosome 12, 88 gene models showed homology to R-like genes (Fig. 3 Additional table 8 [see Additional file 1]), which is 2.0% of the total number of gene models predicted for this chromosome. However, 50 of these 88 gene models belonged to the miscellaneous category including viral resistance, Verticillium wilt resistance, BLB resistance genes and those containing LRR motif but without NBS, CC or LZ motifs. On chromosome 12, only 18 gene models showed homology to the NBS-LRR category. Although the R-like genes and defense response genes were distributed throughout the chromosome, they were present in clusters (Fig. 4). A large cluster of 23 miscellaneous resistance genes was found between positions 40–50 cM (between 5,858,249 and 10,009,727 bp) and a cluster of seven defense response genes was found at position 107.4 cM (between 26,803,729 and 26,870,016 bp). The previously described blast resistance gene Pita [29] was present near the centromere of chromosome 12. The total number of R-like genes in chromosome 12 was less than half of the number in chromosome 11, even though the size and total number of genes in these two chromosomes are similar. This can be attributed to the enrichment of NBS-LRR, LZ-NBS-LRR and LRR-TM-like genes in chromosome 11, which was 3–4 times higher than chromosome 12 (Fig. 3). The difference in the number of R-like genes between the two chromosomes also has an impact on the extent of tandem gene arrays. Chromosome 11 has a total of 924 genes (29%) and chromosome 12 has 684 genes (24%) that are duplicated at least once within a short distance (Fig. 6).
Figure 6 Array sizes of tandemly repeated genes on rice chromosomes 11 and 12. Total tandemly amplified genes on 11 are 924 or 29% of the total genes; total tandemly amplified genes on 12 are 684 or 24% of the total genes.
A previous report on the analysis of R-like genes in the entire rice genome revealed that most of the R-like genes (24.98%) are present on chromosome 11 [30]. Another report found more than 25% of the physically mapped NBS-encoding genes on chromosome 11 and claimed that approximately 20% of the NBS-LRR genes in the Nipponbare genome were predicted to be pseudogenes [31]. Comparison of the total number of genes containing the LRR domain (837) in the finished genome sequence [6] with the number of such genes on chromosomes 11 (166 genes) and 12 (60 genes) in the present study (Tables 2 and 3) shows a similar percentage (27%) of R-like genes to be on these two chromosomes. However, the previous total estimate of 536 R-like genes is significantly lower than the one from the finished genome (837), indicating that previous draft sequences did not provide sufficient information for capturing all candidate genes for disease resistance and, in addition, that gene finding methods are not directly comparable. Complete sequence information is also important for accurate sequence alignments of R-like genes with cloned disease resistance genes from rice (Pita, Xa21, Pib) Arabidopsis (RPM1), tomato, (Cf2, Cf9), barley (Mla1) and wheat (Lr10). The R-like gene and defense response gene hot spots in these two chromosomes will be invaluable for future mapping and cloning of disease resistance genes from naturally occurring disease resistant rice lines. However, manual re-annotation of these R-like genes, as done in case of Arabidopsis [25], and functional validation of the candidate resistance genes, would be important for drawing practical benefits from this information.
Recent duplication on chromosomes 11 and 12
Chromosomes 11 and 12 harbor duplicated regions at the distal ends of their short arms as determined by physical and genetic mapping [32]. Based on previous versions of the rice genome sequence, it became evident that segmental duplications occur throughout the genome [5,7,33]. Here, we focused on segmental duplications between chromosome 11 and 12 only because the finished sequences of these chromosomes permit us to conduct an analysis of the organization of their genes in a detail that was not possible before. At ≥50% coverage and ≥80% identity and excluding TE-related gene models, 546 gene models were identified as duplicated. A subset of these, 350 from chromosome 11 and 352 from chromosome 12, was found to be unique between these chromosomes, excluding repeated gene models. Based on the chromosomal locations of these unique genes, the maximum extent of duplication was confined within the first 3 Mb of both chromosomes (Fig. 7). Possibly, there could be a difference between japonica and indica varieties, as previous comparisons identified a second duplication between chromosomes 11 and 12 in the indica [5] but not the japonica [33] species. Here, most of the duplicated gene models (98%) were found in the same orientation on both the chromosomes (Fig. 8a,b). The 3 Mb region was further analyzed to find the number of duplicated genes at ≥30, ≥40, ≥50, ≥60, ≥70, ≥80 and ≥90% coverage. The numbers of duplicated genes at ≥30% coverage for chromosomes 11 and 12 were 287 and 304, while at ≥90% coverage, the numbers of duplicated genes were 109 and 113, respectively (Additional table 9 [see Additional file 1]). Using alignment to ESTs and FL-cDNAs, a total of 132 (chromosome 11) and 145 (chromosome 12) gene models were expressed; of these, a total of 90 expressed gene models were common to both chromosomes. Interestingly, 42 and 55 gene models were expressed but their homoeologous copies on the respective chromosome were not expressed [Additional table 10; see Additional file 1]. Although it is possible that all ESTs or cDNAs are not represented in the database, it may be that variation in expression reflects diverged expression of one of the homoeologous copies. Divergence of regulation of gene expression after gene amplification has also been reported in maize [15,34].
Figure 7 Chromosome 12 sequences were used as query against a database of chromosome 11 sequences using MegaBLAST as described under Methods. At ≥50% coverage and >80% identity, the frequency distribution of unique duplicate gene models is plotted over the length of chromosomes 11 and 12. Based on the chromosomal locations of these unique genes, duplications were identified throughout the length of both the chromosomes. The maximum extent of duplication, however, was found to be confined within the first 2 Mb region of both chromosomes.
Figure 8 Gene duplication between rice chromosomes 11 and 12 over the whole length (a) and in the first 3 Mb region (b). Blue and red lines connect duplicate gene models in the same and opposite orientation, respectively.
To determine the time of the segmental duplication, the full-length coding regions of a subset of homoeologous genes that had been genetically mapped and were distributed over the entire length of the duplication were aligned and pair-wise analyzed for their nucleotide substitution rates (Table 4). By applying a codon likelihood model, synonymous (Ks) and non-synonymous (Ka) substitution rates were compared. The values for each gene pair varied 18-fold for Ks and 556-fold for Ka, indicating that divergence rates differed significantly among genes. We therefore applied the χ2 homogeneity test to the estimates of divergence time for the rice homoeologs [16]. The χ2 value was highly significant (χ2 = 191.2, P < 0.001), implying that these linked genes diverged within the same time interval. Based on the speciation of rice about 50 mya [23], we calculate the average divergence time for the segmental duplication to be 7.7 mya. We explain the difference from previously published estimates of 25 and 21 mya [3,5] by the sequence accuracy of finished versus whole-genome shotgun sequences [6].
Table 4 Nucleotide substitution analysis of a subset of genes in the duplication of chromosomes 11 and 12. A high confidence gene set has been selected for the calculation of synonymous (Ks) and non-synonymous (Ka) nucleotide substitutions. The genes are presented with their Ks/Ka values according to their chromosomal positions.
Gene Number gene putative function chrom11 start chrom11 stop chrom12 start chrom12 stop Ks se of Ks Ka se of Ka my
1 WRKY family transcription factor 749015 750212 788607 789687 0.16 0.037 0.055 0.013 12.30769
2 WRKY family transcription factor 758715 762245 801426 804997 0.04 0.017 0.015 0.006 3.076923
3 WRKY family transcription factor 781473 782513 816140 817037 0.023 0.017 0.007 0.004 1.769231
4 WRKY family transcription factor 786625 787866 823302 824540 0.077 0.02 0.04 0.007 5.923077
5 alpha-hydroxynitrile lyase 811241 812335 851943 854138 0.0196 0.034 0.0118 0.007 1.507692
6 polyneuridine aldehyde esterase-like 812960 814118 855275 856433 0.025 0.0147 0.009 0.0065 1.923077
7 glycosyl hydrolase family 819185 821808 857627 859498 0.02567 0.009 0.02 0.0053 1.974615
8 scarecrow 1119113 1121579 1042889 1045809 0.0865 0.03 0.0087 0.0026 6.653846
9 glutathione S = transferase T3 1165521 1166165 1096622 1097382 0.327 0.116 0.155 0.04 25.15385
10 apyrase GS52 1174941 1177791 1106432 1109216 0.027 0.01 0.019 0.005 2.076923
11 apyrase S-type 1221864 1225463 1129908 1132441 0.084 0.04 0.027 0.016 6.461538
12 no apical meristem 1233158 1234682 1138382 1139934 0.094 0.026 0.017 0.004 7.230769
13 no apical meristem 1240175 1241314 1146147 1147453 0.32 0.15 0.073 0.0155 24.61538
14 40S robosomal protein S16 1284724 1285173 1168928 1169349 0.093 0 0.003 0 7.153846
15 exostosin family 1288082 1291244 1172080 1174695 0.098 0.02 0.0135 0.004 7.538462
16 cdc45-like protein 1305369 1307183 1188143 1189957 0.068 0.02 0.0037 0.0017 5.230769
17 myb protein homolog 1316546 1317575 1199788 1200856 0.0535 0 0.0115 0 4.115385
18 homoserine dehydrogenase-like protein 1328599 1332331 1215892 1219509 0.0339 0.01 0.01 0.0037 2.607692
19 receptor-like protein kinase 1537568 1540056 1472686 1475196 0.1 0.036 0.02 0.0054 7.692308
20 receptor-loke protein kinase 1545094 1547508 1490358 1492844 0.081 0.021 0.011 0.0027 6.230769
21 60S acidic ribosomal protein 1645643 1647855 1601129 1603328 0.071 0.019 0.0054 0.0027 5.461538
22 calcium dependent protein kinase 1691610 1693844 1647299 1649413 0.0279 0.01 0.00077 0.00078 2.146154
23 YVH 1 protein-tyrosine phosphatase 1696442 1698793 1651813 1654064 0.034 0.012 0.012 0.00436 2.615385
24 hydroxymethylglutaryl-CoA lyase 1715582 1718088 1671209 1673936 0.0757 0.016 0.022 0.0053 5.823077
25 cytochrome P450-like protein 1766039 1767592 1715187 1716737 0.176 0.039 0.0138 0.003 13.53846
26 zinc finger protein-like 1791802 1794503 1731264 1733959 0.1596 0.0386 0.0132 0.0039 12.27692
27 cell death associated protein 1802302 1804168 1744300 1746170 0.1235 0.036 0.0204 0.00572 9.5
28 steroid sulfotransferase 1911460 1912476 1836949 1837788 0.1936 0.078 0.202 0.049 14.89231
29 cyt p450 protein 1997769 2000557 1905242 1908061 0.02 0.015 0.0088 0.0006 1.538462
The number of unique duplicated gene models for the remaining portion (~24–25 Mb) of chromosomes 11 and 12 was 114 (32.6%) and 102 (29%), respectively, with a high degree of rearrangement. Within this region, there were 163 and 189 gene models at ≥30% coverage, while at ≥90% coverage, there were only 31 genes for both the chromosomes (Additional table 10 [see Additional file 1]). Interestingly, only 18 (15.8%) and 14 (13.7%) genes (chromosome 11 and 12, respectively) were found to be expressed based on homology to an EST or cDNA, in contrast to 56 and 58% in the first 3 Mb region of both the chromosomes. Out of these, nine expressed genes were common to both chromosomes. Also, there were nine and five expressed genes of chromosomes 11 and 12 that had non-expressed counterparts. This analysis supports the view that the first 3 Mb duplication of chromosome 11 and 12 is of recent origin. However, the loss of order as well as the low percentage of expressed genes, on each chromosome or in duplicate gene pairs, does not support a duplication of the remaining regions of chromosomes 11 and 12, unless it is very ancient in origin and signatures of duplication have largely disappeared.
Rice-wheat synteny
To investigate further whether rice chromosomes 11 and 12 were related to each other and arose by a WGD event, we compared each gene model predicted from these two chromosomes against 586,577 wheat ESTs. Comparison of bin-mapped wheat ESTs with earlier versions of rice genome sequence provided a glimpse of the structural similarities between these two important cereal genomes [11,13]. However, the present study involves comparative analysis of the finished version of rice chromosomes 11 and 12, which allowed a higher resolution analysis with wheat and a comparative analysis between rice chromosomes 11 and 12. A gene-by-gene comparison of sequence homology of all the predicted rice gene models with the wheat ESTs revealed that 1,588 (35.8%) and 2,220 (51%) of the gene models in chromosomes 11 and 12, respectively, have significant homology with the wheat ESTs at a bit cut-off score of 100. Of these, 416 gene models (26.2%) from chromosome 11 and 552 gene models (24.86%) from chromosome 12 showed significant similarity with bin-mapped wheat EST contigs, a number that is much larger than the levels reported earlier for chromosomes 11 and 12 [13]. A complete list of the matching genes along with their annotated functions is provided in Additional tables 11 and 12 [see Additional file 1].
Out of the 416 gene models from rice chromosome 11, 338 (81.3%) mapped to single wheat homoeologous groups, with the maximum number (35.2%) located on group 4 chromosomes of wheat. The distribution of rice genes that mapped to the wheat chromosome groups 1, 2, 3, 5, 6 and 7 was 9.2%, 9.5%, 12.4%, 15.4%, 8.9% and 9.5%, respectively (Fig. 9 Additional figures 12 and 13 [see Additional file 1]). The majority of rice chromosome 11 gene models mapped to group 4 chromosomes of wheat indicating their common origin. Many of these genes were clustered in the distal region of the long arm of wheat chromosome 4A, but the same gene models mapped to the short arms of wheat chromosomes 4B and 4D, reflecting significant rearrangements and a dynamic state of gene organization in homoeologous chromosomes of wheat and rice chromosome 11. Similar to chromosome 11, a high percentage of chromosome 12 gene models (441, 79.9%) mapped to single wheat homoeologous groups. Of these, the maximum percentage (30.6%) mapped to group 5 chromosomes of wheat whereas the remainder were distributed almost uniformly on the other six wheat homoeologous groups (Fig. 9 Additional figures 12 and 13 [see Additional file 1]).
Figure 9 Syntenic mapping of rice genes from rice chromosomes 11 and 12 to wheat. Distribution of the rice gene homologues from rice chromosomes 11 (blue bars) and 12 (yellow bars) on to the wheat chromosomes of seven homoeologous groups.
A comparative distribution of rice gene homologs from chromosomes 11 and 12 to the seven wheat homoeologous groups clearly indicates that the two rice chromosomes have different origins and, apart from the recent 3 Mb duplication, there is not much in common between the two chromosomes. This contradicts earlier observations that the whole of the rice chromosomes 11 and 12 may have evolved as a result of chromosome duplication via polyploidy about 70 mya, i.e. before the divergence of cereals [7]. Furthermore, these results are also consistent with the alignment of the genetic maps of cereal genomes [10], where rice aligns with two homoeologous regions of maize.
Conclusion
With the completion of the sequences of rice chromosomes 11 and 12, we were able to identify 289 R-like and 28 defense response-like genes, accurately date a 3 Mb recent duplication between the two chromosomes, and show significant synteny between these two chromosomes and wheat chromosome groups 4 and 5, respectively. Rice chromosome 11 has several large clusters of fast evolving disease resistance and defense response genes that have originated by the process of tandem duplication and subsequent divergence under the selective pressure of rice pathogens. This sequence and annotation will be an essential resource for the engineering of rice for tolerance to biotic and abiotic stresses to accommodate growing production needs.
Methods
Sequencing and bioinformatic methods
Bacterial artificial chromosome and P1-derived artificial chromosome (PAC) clones from chromosomes 11 and 12 were sequenced to 8–10× coverage using shotgun DNA sequencing [35,36] and standard high-throughput methods [6]. Gaps and ambiguities were resolved using re-sequencing, alternative chemistries, PCR and primer walking [6]. Pseudomolecules or virtual contigs for chromosomes 11 and 12 were constructed as described previously [27]. The pseudomolecules and annotation are available in GenBank under the accession numbers DP000010 and DP000011. Genes were identified using the ab initio gene finder FGENESH [37] and gene model structure was improved with EST and full-length cDNA (FL-cDNA) evidence using the Program to Assemble Spliced Alignments [38]. Genes were annotated for function as described [27]. Transposable element-related (TE) genes were identified using either TBLASTN similarity to known repetitive elements in the TIGR Oryza Repeat Database [39] or the presence of repetitive element-related Pfam domains [40]. Flanking sequence tags (FSTs) from various insertion mutagenesis projects were aligned with the two pseudomolecules using flast [41] with a cutoff of 95% identity over 80% of the FST length. Transfer RNAs were identified using tRNA-ScanSE [42]. Organellar insertions were determined using BLASTN with the rice mitochondrion and chloroplast genome sequences using a cutoff of 95% identity (Fig. 1). Sequenced genetic markers (4,619 sequences [18,43]) were searched against the pseudomolecules with flast; a cutoff criterion of ≥95% identity over ≥90% length of the marker was used to align the markers to the chromosome sequence. Repetitive sequences were identified, classified, and quantitated on the pseudomolecules using RepeatMasker [44] with the TIGR Oryza Repeat Database [39] and a default RepeatMasker cut-off score of 225.
Non-TE-related genes with rice transcript evidence were identified by searching chromosomes 11 and 12 against the TIGR rice gene index (Release 16 [45]). To identify genes with FL-cDNA support, we searched 33,678 FL-cDNAs available from the KOME database [46]. Genes with an alignment of ≥95% identity and a minimum of 50% length of the alignment covered by a gene index sequence were considered supported by a rice EST or FL-cDNA. Seven monocot gene indices [47] were also used to demonstrate support of expressed genes (wheat release 9.0, maize release 15.0, barley release 9.0, sorghum release 8.0, rye release 3.0, onion release 1.0 and sugarcane release 2.1). Cutoff criteria used with the non-rice monocot gene indices were ≥70% identity over 80% of the length of the gene index sequence and a minimum of 50% of the gene model coverage by the gene index sequence. For the analysis of the type and distribution of the disease resistance and defense response genes, coding sequences from the chromosomes 11 and 12 gene models were used in a BLASTX search with the nr database of NCBI [48] and the top hits were extracted in Excel files. The BLASTX output was then searched manually using auto filters against 24 different keywords/phrases known to represent R-like and defense response genes, and categorized into five main classes as follows: (i) NBS-LRR (matching with NBS-LRR, but not with LZ-NBS-LRR and LRR, CC-NBS-LRR, Pib, Pita, I2C, Rp1-d8, T10RGA, LR10, Mla1 and rust resistance), (ii) LZ-NBS-LRR (matching with LZ-NBS-LRR, but not with NBS-LRR, CC-NBS-LRR, LRR and RPM1), (iii) LRR-TM (matching with Xa21, serine/threonine kinases and Cf2/Cf5 resistance), (iv) miscellaneous category (matching with disease resistance, viral resistance, Yr10, Verticillium wilt resistance, LRR, but not with NBS-LRR, CC-NBS-LRR, LZ-NBS-LRR and bacterial leaf blight resistance) and (v) defense response genes (matching with glucanases, chitinases and thaumatin).
Maize genomic assemblies (243,807 total [Release 4.0, Feb. 23, 2004]) were downloaded from TIGR [20,49]. Sorghum genomic reads (593,969) were downloaded from NCBI and processed using the Lucy program [50] to remove vector and low quality sequences. The remaining "good" reads (504,458) were assembled into 163,908 clusters as described [20]. The assemblies from sorghum and maize were searched against the TIGR rice pseudomolecules [Release 3.0, January 2005] using the BLASTZ program [51]. Homologs were identified in other model organisms by searching the non-TE-related proteins from chromosomes 11 and 12 against the E. coli, Synechocystis, yeast, Drosophila, C. elegans, human and Arabidopsis predicted proteomes using BLASTP.
Chromosome 11–12 duplication analysis was performed using the predicted gene models. Chromosome 12 sequences were used as query against a database of chromosome 11 sequences using MegaBLAST [52]. Only non-TE gene models at ≥50% coverage were used for further analysis. To compare the expression pattern of duplicated genes, a homology search was performed against the NCBI EST and KOME cDNA databases using the unique duplicated non-TE gene models at ≥90% coverage and identity. Wheat-rice synteny was determined using BLASTN of the chromosome 11 and 12 gene model sequences against Triticum aestivum sequences obtained from dbEST. Search parameters were as described previously [12]. Matching gene models were compared with wheat contigs containing bin-mapped ESTs [53] and plotted on the 21 wheat chromosomes.
List of abbreviations used
bacterial artificial chromosome (BAC), flanking sequence tag (FST), leucine-rich repeat (LRR), million years ago (mya), nucleotide binding site (NBS), resistance gene (R-gene), transposable element (TE), whole-genome duplication (WGD).
Authors' contributions
Consortia members are listed in order of Mb of DNA sequences produced. CRB (TIGR), NKS (IIRGS-IARI), AT (IIRGS-UDSC), and JM (PGIR) conducted the coordination of data analysis and manuscript conception.
At GENOSCOPE, EP, AC, JW, MS, and FQ provided data analysis or intellectual input while NC, ND, GO, SS, AD, LC, BS, PW and CS contributed to sequence/assembly aspects of the project.
At IIRGS-IARI, AD, IAG contributed in physical mapping and identification of BACs; KG, AS, MY, RD, SP, MR, PKS and VS in shotgun cloning, template preparation; SS, AB, AP, KS, HS, SCS and SDM in sequencing, VD and AKP in sequence assembly and annotation, KB in wheat-rice synteny. TRS, TM and NKS contributed in all activities and provided intellectual inputs in analysis.
At IIRGS-UDSC, SR, AM, AKB, AG, VG, DK, VR, SV, AK, PK, SS contributed to sequence, assembly and data analysis. PK, JPK, AKT provided intellectual input for strategic plan and participated in manuscript preparation. AKT is also Coordinator of IIRGS.
At TIGR, QY, SO, JL, WZ, AW, HL, JH, BH, JW, SLS, OW, CF, and CRB provided data analysis or intellectual input while KMJ, MK, LO, TT, DF, JB, BW, SJ, MR, HV, LT, SVA, ML, TU, TF, VZ, SI, JH and ARV contributed to sequence/assembly aspects of the project.
At AGI, WW, DK and CM contributed to shotgun library construction and clone management. YY, TR, JC, KC, HRK, DS and RAW contributed to shotgun sequencing, sequence assembly and finishing aspects of this project.
At CSHL, MK and LS provided sequence assembly and finishing, LN, RP and TZ production sequencing, LP, AO and SD contributed to data analysis and annotation, and WRM directed the research.
At PGIR, EL contributed to clone management, EL and BT to shotgun library construction and sequence/assembly, CD, GF and JM to data analysis, and JM to writing the manuscript; JM is the corresponding author.
At RGP, JW, HM and TM contributed to physical map construction, Fiber-FISH analysis, and physical map extension, NM carried out data analysis and TS directed the research.
At UW, WJ and JJ contributed to gap analysis using fiber-FISH.
Supplementary Material
Additional file 1
Additional tables (5–12) and figures (10–13) are presented in the Supporting On-line Data, which is in pdf. Each table and figure is annotated with a legend.
Click here for file
Acknowledgements
*Consortia members:
Genoscope: Nathalie Choisne1, Nadia Demange1, Gisela Orjeda1, Sylvie Samain1, Angélique D'Hont2, Laurence Cattolico1, Eric Pelletier1, Arnaud Couloux1, Béatrice Segurens1, Patrick Wincker1, Claude Scarpelli1, Jean Weissenbach1, Marcel Salanoubat1, Francis Quétier1; Indian Initiative for Rice Genome Sequencing (IIRGS), Indian Agricultural Research Institute: Nagendra K Singh3, Trilochan Mohapatra3, Tilak R Sharma3, Kishor Gaikwad3, Archana Singh3, Vivek Dalal3, Subodh K Srivastava3, Anupam Dixit3, Ajit K Pal3, Irfan A Ghazi3, Mahavir Yadav3, Awadhesh Pandit3, Ashutosh Bhargava3, K Sureshbabu3, Rekha Dixit3, Harvinder Singh3, Suresh C Swain3, Sumita Pal3, M Ragiba3, Pradeep K Singh3, Vibha Singhal3, Sangeeta D Mendiratta3, Kamlesh Batra3; Indian Initiative for Rice Genome Sequencing (IIRGS), University of Delhi South Campus: Saurabh Raghuvanshi4, Amitabh Mohanty4, Arvind K. Bharti4,8, Anupama Gaur4, Vikrant Gupta4, Dibyendu Kumar4, Ravi Vydianathan4, Shubha Vij4, Anita Kapur4, Parul Khurana4, Sulabha Sharma4, Paramjit Khurana4, Jitendra P. Khurana4, Akhilesh K. Tyagi4†; The Institute for Genomic Research (TIGR): Qiaoping Yuan5, Shu Ouyang5, Jia Liu5, Wei Zhu5, Aihui Wang5, Haining Lin5, John Hamilton5, Brian Haas5, Jennifer Wortman5, Kristine M. Jones5, Mary Kim5, Larry Overton5, Tamara Tsitrin5, Douglas Fadrosh5, Jayati Bera5, Bruce Weaver5, Shaohua Jin5, Shivani Johri5, Matt Reardon5, Hue Vuong5, Luke Tallon5, Susan Van Aken5, Matthew Lewis5, Teresa Utterback5, Tamara Feldblyum5, Victoria Zismann5, Stacey Iobst5, Joseph Hsiao5, Aymeric R. de Vazeille5, Steven L. Salzberg5, Owen White5, Claire Fraser5, and C. Robin Buell5; Arizona Genomics Institute: Yeisoo Yu6, Teri Rambo6, Jennifer Currie6, Kristi Collura6, Hye Ran Kim6, Diana Stum6, Wenming Wang6, Dave Kudrna6, Christopher Mueller6, Rod A. Wing6; Cold Spring Harbor Laboratory: Melissa Kramer7, Lori Spiegel7, Lidia Nascimento7, Raymond Preston7, Theresa Zutavern7, Lance Palmer7, Andrew O'Shaughnessy7, Sujit Dike7, W. Richard McCombie7; Plant Genome Initiative at Rutgers: Joachim Messing8, Charles Du8,11 Galina Fuks8, Eric Linton8,12 Bahattin Tanyolac8,13; Rice Genome Research Program: Jianzhong Wu9, Nobukazu Namiki9, Hiroshi Mizuno9, Takashi Matsumoto9, Takuji Sasaki9; University of Wisconsin: Weiwei Jin10, Jiming Jiang10
Consortia members' affiliation:
1Genoscope, 2 rue Gaston Crémieux CP5706, 91057 Evry-Cedex (France); 2UMR PIA, Cirad-Amis, TA40-03 avenue Agropolis, 34398 Montpellier Cedex 05 (France); 3National Research Centre on Plant Biotechnology, Indian Agricultural Research Institute, New Delhi-110 012, India; 4Department of Plant Molecular Biology, University of Delhi South Campus, New Delhi-110021, India; 5The Institute for Genomic Research, 9712 Medical Center Dr, Rockville, MD 20850; 6Arizona Genomics Institute, The University of Arizona, Tucson, AZ 85750; 7Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 11724; 8Plant Genome Initiative at Rutgers, Waksman Institute, Rutgers University, Piscataway, New Jersey 08854; 9Rice Genome Research Program, National Institute of Agrobiological Sciences, 1–2 Kannondai, 2-Chome, Tsukuba, Ibaraki 305-8602, Japan; 10University of Wisconsin, Department of Horticulture, Madison, WI 53706, USA.
Consortia members' present addresses:
11Montclair State University, Montclair, New Jersey 07043; 12Plant Biology Labs, Michigan State University, East Lansing, MI 48824; 13Department of Bioengineering, Ege University, Izmir 35100, Turkey.
Consortia members' sponsors:
The IIRGS acknowledges the Department of Biotechnology, Government of India, for financial assistance, and the Indian Council of Agricultural Research, New Delhi, for support. Work on rice chromosome 11 at TIGR was made possible by grants from the U. S. Department of Agriculture Cooperative State Research, Education, and Extension Service (C.R.B. 99-35317-8275, 2003-35317-13173), the National Science Foundation (C.R.B. DBI998282; DBI0321538), and the U. S. Department of Energy (C.R.B. DE-FG02-99ER20357). We acknowledge the assistance of the TIGR Sequencing Facility, the TIGR Informatics Department, the TIGR IT Group, and the J. Craig Venter Joint Technology Center. Funding for work on rice chromosome 11 was provided by a grant from the U. S. Department of Agriculture Cooperative State Research, Education, and Extension Service to R.A.W and W.R.M. (#2002-35317-12414). R.A.W. acknowledges the assistance of the AGI Sequencing, Physical Mapping, BAC/EST Resource Centers and the Arizona Genomics Computational Laboratory. W.R.M. acknowledges the CSHL sequencing and informatics groups. Work at the PGIR was supported by Rutgers, The State University of New Jersey. We thank the Japanese Ministry of Agriculture, Forestry and Fishery for genetic markers, which were utilized to identify seed BACs for sequencing.
==== Refs
Khush GS Origin, dispersal, cultivation and variation of rice Plant Mol Biol 1997 35 25 34 9291957 10.1023/A:1005810616885
Peng S Huang J Sheehy JE Laza RC Visperas RM Zhong X Centeno GS Khush GS Cassman KG Rice yields decline with higher night temperature from global warming Proc Natl Acad Sci U S A 2004 101 9971 9975 15226500 10.1073/pnas.0403720101
Goff SA Ricke D Lan TH Presting G Wang R Dunn M Glazebrook J Sessions A Oeller P Varma H A draft sequence of the rice genome (Oryza sativa L. ssp. japonica) Science 2002 296 92 100 11935018 10.1126/science.1068275
Yu J Hu S Wang J Wong GK Li S Liu B Deng Y Dai L Zhou Y Zhang X A draft sequence of the rice genome (Oryza sativa L. ssp. indica) Science 2002 296 79 92 11935017 10.1126/science.1068037
Yu J Wang J Lin W Li S Li H Zhou J Ni P Dong W Hu S Zeng C The Genomes of Oryza sativa: a history of duplications PLoS Biol 2005 3 e38 15685292 10.1371/journal.pbio.0030038
International-Rice-Genome-Sequencing-Project The map-based sequence of the rice genome Nature 2005 436 793 800 16100779
Paterson AH Bowers JE Chapman BA Ancient polyploidization predating divergence of the cereals, and its consequences for comparative genomics Proc Natl Acad Sci U S A 2004 101 9903 9908 15161969 10.1073/pnas.0307901101
Sasaki T Matsumoto T Antonio BA Nagamura Y From mapping to sequencing, post-sequencing and beyond Plant Cell Physiol 2005 46 3 13 15659433 10.1093/pcp/pci503
Arumuganathan K Earle ED Nuclear DNA content of some important plant species Plant Mol Biol Reporter 1991 9 208 218
Gale MD Devos KM Comparative genetics in the grasses Proc Natl Acad Sci U S A 1998 95 1971 1974 9482816 10.1073/pnas.95.5.1971
Sorrells ME La Rota M Bermudez-Kandianis CE Greene RA Kantety R Munkvold JD Miftahudin Mahmoud A Ma X Gustafson PJ Comparative DNA sequence analysis of wheat and rice genomes Genome Res 2003 13 1818 1827 12902377
Singh NK Raghuvanshi S Srivastava SK Gaur A Pal AK Dalal V Singh A Ghazi IA Bhargav A Yadav M Sequence analysis of the long arm of rice chromosome 11 for rice-wheat synteny Funct Integr Genomics 2004 4 102 117 15085449 10.1007/s10142-004-0109-y
La Rota M Sorrells ME Comparative DNA sequence analysis of mapped wheat ESTs reveals the complexity of genome relationships between rice and wheat Funct Integr Genomics 2004 4 34 46 14740255 10.1007/s10142-003-0098-2
Lai J Ma J Swigonova Z Ramakrishna W Linton E Llaca V Tanyolac B Park YJ Jeong OY Bennetzen JL Gene loss and movement in the maize genome Genome Res 2004 14 1924 1931 15466290 10.1101/gr.2701104
Swigonova Z Bennetzen JL Messing J Structure and evolution of the r/b chromosomal regions in rice, maize and sorghum Genetics 2005 169 891 906 15489523 10.1534/genetics.104.034629
Swigonova Z Lai J Ma J Ramakrishna W Llaca V Bennetzen JL Messing J Close split of sorghum and maize genome progenitors Genome Res 2004 14 1916 1923 15466289 10.1101/gr.2332504
Khush GS Brar DS Kush GS, Brar DS, Hardy B Rice genetics from Mendel to functional genomics Rice Genetics 2001 IV Los Banos, Philippines: IRRI 3 25
Harushima Y Yano M Shomura A Sato M Shimano T Kuboki Y Yamamoto T Lin SY Antonio BA Parco A A high-density rice genetic linkage map with 2275 markers using a single F2 population Genetics 1998 148 479 494 9475757
The-Rice-Chromosome-10-Sequencing-Consortium In-depth view of structure, activity, and evolution of rice chromosome 10 Science 2003 300 1566 1569 12791992 10.1126/science.1083523
Whitelaw CA Barbazuk WB Pertea G Chan AP Cheung F Lee Y Zheng L van Heeringen S Karamycheva S Bennetzen JL Enrichment of gene-coding sequences in maize by genome filtration Science 2003 302 2118 2120 14684821 10.1126/science.1090047
Palmer LE Rabinowicz PD O'Shaughnessy AL Balija VS Nascimento LU Dike S de la Bastide M Martienssen RA McCombie WR Maize genome sequencing by methylation filtration Science 2003 302 2115 2117 14684820 10.1126/science.1091265
Bedell JA Budiman MA Nunberg A Citek RW Robbins D Jones J Flick E Rholfing T Fries J Bradford K Sorghum genome sequencing by methylation filtration PLoS Biol 2005 3 e13 15660154 10.1371/journal.pbio.0030013
Gaut BS Evolutionary dynamics of grass genomes New Phytologist 2002 154 15 28 10.1046/j.1469-8137.2002.00352.x
Hulbert SH Webb CA Smith SM Sun Q Resistance gene complexes: evolution and ugtilization Annu Rev Phytopathol 2001 39 285 312 11701867 10.1146/annurev.phyto.39.1.285
Meyers BC Kozik A Griego A Kuang H Michelmore RW Genome-wide analysis of NBS-LRR-encoding genes in Arabidopsis Plant Cell 2003 15 809 834 12671079 10.1105/tpc.009308
Johal GS Briggs SP Reductase activity encoded by the HM1 disease resistance gene in maize Science 1992 258 985 987 1359642
Yuan Q Ouyang S Wang A Zhu W Maiti R Lin H Hamilton J Haas B Sultana R Cheung F The institute for genomic research Osa1 rice genome annotation database Plant Physiol 2005 138 18 26 15888674 10.1104/pp.104.059063
Song WY Wang GL Chen LL Kim HS Pi LY Holsten T Gardner J Wang B Zhai WX Zhu LH A receptor kinase-like protein encoded by the rice disease resistance gene, Xa21 Science 1995 270 1804 1806 8525370
Bryan GT Wu KS Farrall L Jia Y Hershey HP McAdams SA Faulk KN Donaldson GK Tarchini R Valent B tA single amino acid difference distinguishes resistant and susceptible alleles of the rice blast resistance gene Pi-ta Plant Cell 2000 12 2033 2046 11090207 10.1105/tpc.12.11.2033
Zhou T Wang Y Chen JQ Araki H Jing Z Jiang K Shen J Tian D Genome-wide identification of NBS genes in japonica rice reveals significant expansion of divergent non-TIR NBS-LRR genes Mol Genet Genomics 2004 271 402 415 15014983 10.1007/s00438-004-0990-z
Monosi B Wisser RJ Pennill L Hulbert SH Full-genome analysis of resistance gene homologues in rice Theor Appl Genet 2004 109 1434 1447 15309302 10.1007/s00122-004-1758-x
Wu J Kurata N Tanoue H Shimokawa T Umehara Y Yano M Sasaki T Physical mapping of duplicated genomic regions of two chromosome ends in rice Genetics 1998 150 1595 1603 9832535
Guyot R Keller B Ancestral genome duplication in rice Genome 2004 47 610 614 15190378 10.1139/g04-016
Song R Llaca V Linton E Messing J Sequence, regulation, and evolution of the maize 22-kD alpha zein gene family Genome Res 2001 11 1817 1825 11691845
Messing J Crea R Seeburg PH A system for shotgun DNA sequencing Nucleic Acids Res 1981 9 309 321 6259625
Vieira J Messing J The pUC plasmids, an M13mp7-derived system for insertion mutagenesis and sequencing with synthetic universal primers Gene 1982 19 259 268 6295879 10.1016/0378-1119(82)90015-4
Salamov AA Solovyev VV Ab initio gene finding in Drosophila genomic DNA Genome Res 2000 10 516 522 10779491 10.1101/gr.10.4.516
Haas BJ Delcher AL Mount SM Wortman JR Smith RK JrHannick LI Maiti R Ronning CM Rusch DB Town CD Improving the Arabidopsis genome annotation using maximal transcript alignment assemblies Nucleic Acids Res 2003 31 5654 5666 14500829 10.1093/nar/gkg770
Ouyang S Buell CR The TIGR Plant Repeat Databases: a collective resource for the identification of repetitive sequences in plants Nucleic Acids Res 2004 32 D360 363 14681434 10.1093/nar/gkh099
Bateman A Birney E Cerruti L Durbin R Etwiller L Eddy SR Griffiths-Jones S Howe KL Marshall M Sonnhammer EL The Pfam protein families database Nucleic Acids Res 2002 30 276 280 11752314 10.1093/nar/30.1.276
Yuan Q Liang F Hsiao J Zismann V Benito MI Quackenbush J Wing R Buell R Anchoring of rice BAC clones to the rice genetic map in silico Nucleic Acids Res 2000 28 3636 3641 10982886 10.1093/nar/28.18.3636
Lowe TM Eddy SR tRNAscan-SE: a program for improved detection of transfer RNA genes in genomic sequence Nucleic Acids Res 1997 25 955 964 9023104 10.1093/nar/25.5.955
Causse MA Fulton TM Cho YG Ahn SN Chunwongse J Wu K Xiao J Yu Z Ronald PC Harrington SE Saturated molecular map of the rice genome based on an interspecific backcross population Genetics 1994 138 1251 1274 7896104
RepeatMasker
Quackenbush J Cho J Lee D Liang F Holt I Karamycheva S Parvizi B Pertea G Sultana R White J The TIGR Gene Indices: analysis of gene transcript sequences in highly sampled eukaryotic species Nucleic Acids Res 2001 29 159 164 11125077 10.1093/nar/29.1.159
Kikuchi S Satoh K Nagata T Kawagashira N Doi K Kishimoto N Yazaki J Ishikawa M Yamada H Ooka H Collection, mapping, and annotation of over 28,000 cDNA clones from japonica rice Science 2003 301 376 379 12869764 10.1126/science.1081288
Monocot-Gene-Index
National-Center-for-Biotechnology-Information
Maize-Genomic-Survey-Sequences
Chou HH Holmes MH DNA sequence quality trimming and vector removal Bioinformatics 2001 17 1093 1104 11751217 10.1093/bioinformatics/17.12.1093
Schwartz S Kent WJ Smit A Zhang Z Baertsch R Hardison RC Haussler D Miller W Human-mouse alignments with BLASTZ Genome Res 2003 13 103 107 12529312 10.1101/gr.809403
Zhang Z Schwartz S Wagner L Miller W A greedy algorithm for aligning DNA sequences Comput Biol 2000 7 203 214 10.1089/10665270050081478
Mapped-Wheat-ESTs
|
16188032
|
PMC1261165
|
CC BY
|
2021-01-04 16:39:15
|
no
|
BMC Biol. 2005 Sep 27; 3:20
|
utf-8
|
BMC Biol
| 2,005 |
10.1186/1741-7007-3-20
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1151616805610.1186/1471-2407-5-115Research ArticleRadiosensitising effect of electrochemotherapy with bleomycin in LPB sarcoma cells and tumors in mice Kranjc Simona [email protected] Maja [email protected] Alenka [email protected] Marjeta [email protected] Gregor [email protected] Institute of Oncology Ljubljana, Department of Experimental Oncology, Zaloska 2, SI-1000 Ljubljana, Slovenia2 Jozef Stefan Institute, Jamova 39, SI-1000 Ljubljana, Slovenia2005 16 9 2005 5 115 115 14 4 2005 16 9 2005 Copyright © 2005 Kranjc 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
Bleomycin is poorly permeant but potent cytotoxic and radiosensitizing drug. The aim of the study was to evaluate whether a physical drug delivery system – electroporation can increase radiosensitising effect of bleomycin in vitro and in vivo.
Methods
LPB sarcoma cells and tumors were treated either with bleomycin, electroporation or ionizing radiation, and combination of these treatments. In vitro, response to different treatments was determined by colony forming assay, while in vivo, treatment effectiveness was determined by local tumor control (TCD50). Time dependence of partial oxygen pressure in LPB tumors after application of electric pulses was measured by electron paramagnetic oxyimetry.
Results
Electroporation of cells in vitro increased radiosensitising effect of bleomycin for 1.5 times, in vivo radiation response of tumors was enhanced by 1.9 fold compared to response of tumors that were irradiated only. Neither treatment of tumors with bleomycin nor application of electric pulses only, affected radiation response of tumors. Application of electric pulses to the tumors induced profound but transient reduction of tumor oxygenation. Although tumor oxygenation after electroporation partially restored at the time of irradiation, it was still reduced at the level of radiobiologically relevant hypoxia.
Conclusion
Our study shows that application of electric pulses to cells and tumors increases radiosensitising effect of bleomycin. Furthermore, our results demonstrate that the radiobiologically relevant hypoxia induced by electroporation of tumors did not counteract the pronounced radiosensitising effect of electrochemotherapy with bleomycin.
==== Body
Background
Bleomycin (BLM) is a glycopeptide antibiotic that is used in treatment of variety of human neoplasms, particularly lymphomas, squamous cell carcinoma and germ-cell tumors [1,2]. New understandings of mechanisms of its cytotoxicity have led to development of more potent and toxic therapeutic analogues [1]. Besides cytotoxic, BLM has also radiosensitising effect that was demonstrated in cells, experimental tumors as well as in the clinic [3-8].
Cytotoxicity of BLM once inside the cell is very high, but limited due to the poor uptake into the cells, which was shown to be receptor mediated endocytosis [9]. It has been shown that electroporation, a physical method of drug delivery into the cells and tissues, induces transient permeabilization of the cell membrane, thus enabling facilitated transport of BLM and other cytotoxic drugs like cisplatin into the cells, resulting in their increased cytotoxicity [10-12]. This combined treatment of chemotherapeutic drugs with hampered transport through the cell membrane and electroporation in vivo was termed electrochemotherapy (ECT). Its effectiveness in local tumor control was demonstrated in numerous animal tumor models and in several clinical studies in cancer patients with different tumors, malignant melanoma, sarcomas and carcinomas [13-15]. In addition, it was demonstrated that application of electric pulses to the tumor as well as normal tissues reduces blood flow and oxygenation [16-19].
The role of electroporation in radiosensitisation of tumors with cisplatin has already been shown. It was demonstrated that electroporation increased radiosensitising effect of cisplatin in two murine tumor models; EAT carcinoma and LPB sarcoma [20,21]. Feasibility of this approach was shown also in human adenocarcinoma skin metastases [22]. Furthermore, it was demonstrated that electroporation is effective in potentiating antitumor effectiveness of hypoxia-selective drug tirapazamine combined with tumor irradiation [23].
The aim of this study was to demonstrate the feasibility of this approach with another radiosensitising drug, BLM, which has proven its high effectiveness in ECT protocol. In addition, tumor oxygenation was measured after application of electric pulses to the tumors in order to determine the impact of reduced oxygenation in tumors on radiation response.
Methods
Cell line, tumors and animals
Murine sarcoma LPB cells, syngeneic to C57Bl/6 mice were used in the experiments. The cells were grown in Eagle minimum essential medium (EMEM) supplemented with 10% fetal calf serum (FCS) (Sigma, Chemical Co., St. Louis, MO, USA). Cells were routinely subcultured twice per week and maintained in a humidified atmosphere with 5% CO2 at 37°C. Inbred C57Bl/6 mice were purchased from the Institute of Pathology, University of Ljubljana (Slovenia). Mice were maintained at 21°C with natural day/night light cycle in a conventional animal colony. In the experiments mice of both sexes were used and distributed evenly between the groups. The average weight of mice subjected to treatment protocol was 22 g for female and 25 g for male mice.
Solid subcutaneous tumors were induced dorsolaterally by the injection of 1.3 × 106 viable tumor cells in EMEM supplemented with 2% FCS prepared from cell culture in vitro. The tumors reached approximately 40 mm3 in volume in 10–12 days. Then, the mice were marked, divided randomly into experimental groups and subjected to specific experimental protocol. Treatment protocols were approved by the Ministry of Agriculture, Forestry and Food of the Republic of Slovenia No. 323-02-237/01, and are in compliance with the standards required by the UKCCCR guidelines.
Drug
A stock solution (3 mg/ml) of BLM (Blenoxane, Bristol Myers Squibb Co., Princeton, NJ, USA) was prepared in phosphate buffered saline. Final concentration of BLM (0.28 μg/ml) was freshly prepared in EMEM before each experiment in vitro. In vivo experiments were performed using BLM at the dose of 0.5 mg/kg, which was daily prepared in 0.9% NaCl solution.
Irradiation of cells and tumors
For irradiation an X-ray unit Darpac 2000 (Gulmay Medical Ltd, Shepperton, UK), operated at 220 kV, 10 mA, and with 0.55 mm Cu and 1.8 mm Al filtration was used. In experiments in vitro, cells (1 × 106 cells/ml EMEM) were irradiated in low attachment plates at a dose rate 2 Gy/min with graded doses (2–8 Gy) and thereafter plated in Petri dishes for colony forming assay. In experiments in vivo, tumors were irradiated at a dose rate 2.1 Gy/min with single graded doses (5–50 Gy). Mice were put into a holder for 6 mice on the X-ray unit with the apertures for the irradiation of the tumors; the rest of the body of the mice was protected by lead block, and by the lead holder for the mice. In the holders, mice were restrained, but not anaesthetized during irradiation. To ensure uniform dose distribution through the tumor volume, the tumors were exposed to irradiation by two opposing treatment fields through each of which 50% of the dose was delivered [21].
Electrochemotherapy of cells and tumors
For electroporation, an electric pulse generator Jouan GHT 1287 (St. Herblain, France) delivering 8 square wave electric pulses of amplitude over distance ratio of 1200 V/cm, duration 100 μs at 1 Hz was used. In experiments in vitro 90 μl of cell suspension (2.2 × 107 cells/ml) was mixed with 10 μl of BLM at a concentration of 0.28 μg/ml, as described previously [20]. Briefly, one half of the mixture was exposed to electric pulses and the other half served as a control for BLM treatment alone. Thereafter the cells were incubated for 5 min at room temperature in low attachment plates, diluted and plated on Petri dishes for colony forming assay. The survival of cells treated with ECT was normalized to electroporation treatment alone.
In in vivo experiments, ECT of tumors was performed as described previously [20]. Briefly, 3 min after intravenous injection of BLM (injection volume 150 μl), electric pulses were applied to the tumors using plate electrodes with 8 mm distance between them. Electrodes were placed on the skin overlying the tumor at the opposing margins and electric pulses delivered in two sets of 4 pulses in opposed directions by rotating the electrodes for 90°.
Study design in vitro
To determine whether electroporation increases radiosensitising effect of BLM in vitro, LPB cells were exposed to electric pulses in the presence of BLM and placed in low attachment plates for 5 min. Thereafter, 1 ml of EMEM was added and after additional 5 min cells were irradiated and plated in Petri dishes (Figure 1A). The survival of cells treated with different treatment combinations that included irradiation, were normalized to the appropriate control group, in order to demonstrate the interaction between the treatments on cell's survival. Survival of cells after electroporation and irradiation was normalized to the effect of electroporation, survival of cells after combination of BLM and irradiation was normalized to the effect of BLM, whereas when the cells were irradiated after ECT, cell survival data were normalized to the effect of ECT. All the data were pooled from 3 independent experiments performed in triplicates. The effect of treatments was assessed by comparison of IC90 values (drug concentration required to reduce cell survival for 90%). Enhancement factor (EF) was calculated on the basis IC90 values.
Study design in vivo
To determine whether application of electric pulses increases radiosensitising effect of BLM, ECT was combined with local tumor irradiation with 20 min interval between the treatments (Figure 1B). Antitumor effectiveness of ECT combined with irradiation was evaluated by comparing the effects to single treatments or to the other treatment combinations: control untreated tumors, tumors treated with BLM, electroporation or irradiation only, tumors treated by ECT, and tumors treated with combination of BLM or electroporation and irradiation. Tumors were irradiated with single doses (5, 10, 15, 20, 25, 30, 35, 40, 45, 50 Gy). Each experimental group consisted of 3–7 mice and the data were pooled from 3 independent experiments.
Assessment of tumor response
The tumor volume was determined by measuring three orthogonal tumor diameters (e1, e2 and e3) with Vernier calliper. Tumor volume was calculated by the formula V = π × e1 × e2 × e3/6. Tumor regression and regrowth was followed until the tumors grew up to 350 mm3, and then the animals were sacrificed. If the tumors regressed after therapy, animals were checked for the presence of the tumor in the irradiation field at 4–5 day intervals up to 100 days. The animals were considered cured if they were tumor free at day 100. Radiation dose needed to control 50% of irradiated tumors (TCD50) was used to determine response of tumors (included were 9–21 mice per irradiation dose). Dose response curves were computed by the logit method of analysis [24]. Enhancement factor (EF) was calculated based on TCD50 values.
EPR oximetry measurements
EPR oximetry was used to measure partial oxygen pressure (pO2) in normal muscle, subcutaneous tissue, untreated LPB tumors as well as in the tumors subjected to electric pulses, as described previously [25]. Briefly, EPR oximetry is based on the fact that molecular oxygen is paramagnetic with two unpaired electrons and has a very rapid relaxation rate. It provides an effective relaxation mechanism to other paramagnetic species via Heisenberg spin exchange interaction. Consequently, the presence of oxygen influences both, spin-spin and spin-lattice relaxation times of the paramagnetic probe. Therefore, an increase in oxygen concentration increases the EPR spectra line-width, decreases the resolution of hyperfine structure, and decreases the microwave power at which the saturation of the EPR absorption lines occurs. All these parameters can be measured by EPR. Therefore, 3 small crystals of the oxygen sensitive paramagnetic probe lithium phtalocyanine (LiPc, 15 – 40 μm in diameter) were inserted into tumor (one in centre and one in periphery) and one in selected normal tissue (subcutaneous tissue or skeletal muscle) one day before the treatment. Immediately after the treatment, the EPR spectra were recorded continuously for 60 min and thereafter at each selected time point for 15 min. The mice were anaesthetized by intraperitoneal injection of a mixture of Domitor (1.0 mg/kg body weight; Pfizer GmbH, Karlsruhe, Germany) and 10% ketamine (75.0 mg/kg body weight; Veyx-Pharma GmbH, Schwarzenborn, Germany). During the anesthesia, warm air was used to keep the body temperature as close as possible to 37°C with variations of up to 0.5°C during single measurement. The measurements were performed on Varian E-9 EPR spectrometer, with a custom-made low frequency microwave bridge operating at 1.1 GHz and an extended loop resonator (11 mm in diameter), both designed by Professor T. Walczak (Darmouth Medical School, Hanover, NH). Typical spectrometer settings were: modulation frequency, 100 kHz; modulation amplitude not more than one-third of the peak-to-peak line-width, and scan range, 2 mT. The line width of the EPR spectra reflects the pO2, on the site of the paramagnetic probe and was determined from the calibration curve presented elsewhere, as the changes in the EPR spectrum can be calibrated with known concentration of oxygen [25].
Statistical analysis
All data were tested for normality of distribution. The statistical differences between the treatment groups were assessed by a t-test after one-way ANOVA was performed and fulfilled. SigmaStat statistical software (SPSS inc.) was used for statistical analysis. P levels of less than 0.05 were taken as significant.
Results
Radiosensitisation of LPB cells in vitro
In order to determine the radiosensitising effect of BLM, LPB cells exposed to BLM alone or ECT were irradiated with graded X-ray doses up to 8 Gy (Figure 2). Exposure of cells either to BLM, electroporation or ECT statistically significantly increased radiation response of LPB cells. Increase in radiation response was the lowest when the cells were exposed to BLM (EF = 1.19), probably due to low BLM concentration and a very short exposure time. On the other hand, the radiosensitivity of LPB cells was greatly enhanced when the cells were pretreated with ECT (EF = 1.53). When cells were exposed to electric pulses prior to irradiation the enhancement factor was 1.25. Therefore, radiosensitising effect of ECT might not be due to just increased BLM delivery into the cells by electroporation, but also to radiosensitisation by electroporation of cells (Figure 2). Surviving fraction of LPB cells after pertinent control treatments BLM alone, electric pulses alone and ECT is presented in Table 1.
Radiosensitisaton of LPB tumors in vivo
Radiosensitising effect of ECT was evaluated also on subcutaneous LPB tumors in mice. As endpoints for evaluation of antitumor effectiveness were used tumor growth delay and local tumor control (TCD50 assay). Neither treatment of animals with BLM alone nor application of electric pulses to the tumors prior to irradiation of tumors had any effect on local tumor control, as observed by TCD50 values (Table 2, Figure 3). However, ECT of tumors statistically significantly increased radiation response of the tumors (Table 2, Figure 3). TCD50 value was reduced from 23.1 Gy in mice that were treated with irradiation only, to 12.4 Gy in mice that were treated with ECT prior to irradiation. Since the enhancement factor was 1.9, it is evident that electroporation of tumors significantly contributed to radiosensitisation of tumors with BLM, specifically because combined treatment of ECT with irradiation was statistically significantly more effective than treatment of tumors that were irradiated in combination with electroporation or BLM.
Antitumor effectiveness of pertinent control groups BLM alone, electric pulses alone and ECT is shown in Table 1. ECT of tumors had good antitumor effect and some of the tumors were cured (8.7%). Treatment with BLM only or application of electric pulses to the tumors had minimal effect on tumor growth and no tumor cures were obtained after either of the treatments.
Side effects associated with application of electric pulses were instantaneous contractions of muscles located beneath the site of treatment, which disappeared immediately after each pulse. Irradiation alone or combined with BLM, electric pulses and ECT resulted in hair loss, but none of the treatments induced skin desquamation.
Tissue oxygenation
In order to determine whether electroporation of tumors can induce radiobiologically relevant hypoxia, that could reduce radiosensitising effect of ECT, pO2 was measured in the tumors after exposure to electric pulses. To determine whether reduction in tumor oxygenation is confined to the electroporated area (tumor), pO2 was measured in the same animal also in the normal tissues (skeletal muscle, subcutaneous tissue) that were not exposed to electric pulses. Application of electric pulses to the tumors statistically significantly reduced pO2 in the tumors (Figure 4). Five min after application of electric pulses to the tumors pO2 was lowered in the centre of the tumors down by 75% and in the periphery of the tumors down by 50%. Immediately thereafter, tumors started to reoxygenate; however 6 h after electroporation the oxygen level was still at 70% of pretreatment level and even after 24 h was not completely restored (Figure 4). Specifically, at 20 min after electroporation of the tumors, i.e. at the time of tumor irradiation, in the tumor centre as well as in the tumor periphery there was still reduced tumor oxygenation, at the level of radiobiologically relevant hypoxia (4.6 ± 0.6, 6.1 ± 0.3, respectively). In normal tissues; skeletal muscle and subcutaneous tissue pO2 was not affected by application of electric pulses to the tumors (Figure 4). In addition, pO2 was measured in untreated LPB tumors at the same time points to evaluate the whether tumor growth can affect tumor oxygenation. As shown on Figure 4, pO2 in untreated tumors did not change throughout the 24 h measurement.
Discussion
This study shows that application of electric pulses to cells and tumors increases radiosensitising effect of BLM. As already demonstrated, electroporation of tumors increases BLM uptake into the tumor cells [10] and therefore this might be the principal mechanism of the increased radiosensitivity. Furthermore, our results demonstrate that the radiobiologically relevant hypoxia induced by electroporation of tumors did not counteract the pronounced radiosensitising effect of ECT.
Radiosensitising effect of BLM was demonstrated in many in vitro and in vivo studies, as well as in clinical trials [3,5-8,26,27]. Good potentiation of radiation response was found in some in vitro studies, even without use of drug delivery systems [4], which is in agreement with our results. Although with different concentration of BLM and incubation time used, the potentiation of radiation response was similar in the study of Leith et al and ours (enhancement factor 1.25 and 1.19, respectively). In vivo studies on mice demonstrated potentiation of radiation response with BLM when it was given in high doses up to 100 mg/kg [5,6]. At these doses radiation response was potentiated either after single irradiation dose or in fractionated regime up to 1.23-fold [5,6,8]. Our results showed that radiation response in vivo was not affected by BLM. The discrepancy in the response between our and other studies could be attributed to the much lower dose of BLM (0.5 mg/kg), which was used in our study.
Electroporation is one of the drug delivery systems which was shown to be effective in potentiation of BLM and cisplatin cytotoxicity [14]. In vivo ECT of tumors in experimental systems as well as in treatment of cancer patients is feasible and effective for local tumor control [13,15]. In this system low doses of drugs are needed, because electroporation of tumors in vivo increases drug accumulation in tumor cells from 2-fold for cisplatin and 4-fold for BLM [10,11]. Drug doses needed for effective local tumor control in ECT protocol are so low that they have minimal or no antitumor effectiveness without subsequent application of electric pulses to the tumors.
In order to reduce drug dosage for effective radiosensitisation of tumors several drug delivery systems can be used. Drug delivery systems, such as incorporation of the drug into liposomes or other vehicles and local drug administration were used to increase delivery of cisplatin to tumors [28-31], however to our knowledge there is no reports using drug delivery systems for in vivo radiosensitisation of tumors with BLM. Combination of ECT with cisplatin and irradiation has already been tested in cells, tumors in mice and a patient. The results demonstrated that ECT increased radiation response of cells and tumors, and that the predominant underlying mechanism was increased cisplatin accumulation in the cells of tumors [20-22]. The potentiation of radiation response by ECT with cisplatin was 1.4 compared to tumors that were treated with combined cisplatin and irradiation treatment, and 1.6 compared to tumors that were irradiated only [20]. In the present study, the potentiation of radiation response by ECT with BLM was 1.9 fold compared to tumors that were irradiated only and those that were concomitantly treated with BLM, since BLM only had no radiosensitising effect at this low drug dose. As discussed in our previous studies on ECT with cisplatin and irradiation, several mechanisms could be responsible for potentiation of radiation response of ECT with BLM [20,21].
First, electroporation induces transient permeability of cell membrane and therefore enables increased drug accumulation in the cells. For BLM as well as for cisplatin it was shown that electroporation of cells or tumors causes increased drug accumulation in tumor cells in vitro as well as in tumors [10,11,32]. Therefore, increased BLM accumulation in the tumor cells may consequently result in better radiosensitisation of cells and tumors.
Second, application of electric pulses to the tissues, either normal tissues or tumors results in temporary, but reversible reduction of tissue perfusion [16-19,23,33]. The effect was more pronounced in tumors compared to normal tissues, where it was shown that electroporation induced profound reduction of tumor blood flow and reduced tumor oxygenation [17-20,23]. In order to verify whether this kind of electric pulses induce hypoxia also in LPB tumor model, used in this study, tissue oxygenation was measured. It was shown that electroporation of tumors induced instant, but reversible tumor hypoxia, which was not restored at the time of tumor irradiation. Since the induced tumor hypoxia was radiobiologically relevant [34] it could have an effect on tumor response to irradiation. However, the curability of tumors treated with the electroporation combined with tumor irradiation was the same as the curability of tumors that were irradiated only. In contrast, our results in vitro demonstrated that electroporation of cells enhanced radiation response. Therefore some other mechanisms may affect the radiation response of tumors. It is already known that reactive oxygen species are formed after electroporation of the cells in vitro [35,36]. Reactive oxygen species are known to contribute to radiation damage to the cells. Indeed, we have observed in this and previous studies that electroporation of cells in vitro in normoxic conditions predisposed them to radiation damage [20,21]. In vivo on solid tumors, the radiation response was dependent on tumor type. Electroporation enhanced radiation response in EAT carcinoma, but not in LPB sarcoma [20,21]. Therefore, in this study, the observed radiation response of tumors that were exposed to electric pulses could be ascribed to, both radiobiologically relevant tumor hypoxia and induction of reactive oxygen species, the effects that counteract each other.
Consequently, in the case of ECT combined with irradiation, the electroporation-induced generation of reactive oxygen species in the tumor cells may additionally contribute to DNA damage induced by BLM and irradiation. However, since the data on reactive oxygen species were obtained on cells in vitro further studies in vivo are needed to demonstrate the presence and role of reactive oxygen species in tumors after electroporation. This question could be addressed also by prolonging the interval between the BLM or ECT and tumor irradiation, so as to ensure restitution of the tumor oxygenation before tumor irradiation is performed. Furthermore, this problem might be evaluated also by using anti-oxidant drugs in combination with electroporation and irradiation.
In conclusion, results of our study show that ECT with BLM greatly increased radiation response of LPB tumors. ECT is already used in treatment of patients with cutaneous and subcutaneous tumor nodules [13-15]. Recently, clinically certified generators of electric pulses came on the market. This will enable broader application of ECT in the clinics, also in combination with radiotherapy.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SK participated in the design of the study, performed the experiments, and drafted the manuscript. MC participated in design of the study and critically revised the manuscript. AG participated in design of the study and experiments. MS helped to draft the manuscript and participated in analysis and interpretation of the data. GS conceived the study, participated in design of the study and coordination, and critically revised the draft.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was supported by the Ministry of Higher Education, Science and Technology of the Republic of Slovenia.
Figures and Tables
Figure 1 Schematic illustration of the experimental setting for the in vitro (A) and in vivo (B) study.
Figure 2 Radiation survival curves of LPB cells exposed to different irradiation doses only and to a combination with bleomycin (BLM; 0.28 μg/ml), electroporation and electrochemotherapy (ECT). Cell survival was determined using colony forming assay. Values are AM ± SEM (n = 9).
Figure 3 Radiation dose response curves for local tumor control of LPB sarcoma tumors. Tumors were treated with electrochemotherapy (ECT) 20 min prior to irradiation or in different combinations of irradiation either with bleomycin (BLM) given intravenously (0.5 mg/kg), or with application of 8 electric pulses to the tumors (electroporation – irradiation; EP-IR)). Groups consisted at least of 9 mice per irradiation dose.
Figure 4 Time dependence of partial oxygen pressure in normal tissues and LPB sarcoma tumors after application of electric pulses to the tumors. pO2 measured in untreated LPB tumors at the same time points is also shown. Partial oxygen pressure was measured by EPR oximetry. Values are AM ± SEM (n = 7–10).
Table 1 Surviving fraction of the cells in vitro, tumor doubling time and complete responses of the tumors treated with BLM, electroporation and ECT.
In vitro
In vivo
Group n Sfa n DT (days)b CR (n)c
Control 9 1 16 2.3 ± 0.1 0
BLM 9 0.83 ± 0.03 17 4.4 ± 0.5 0
EPd 9 0.81 ± 0.02 13 4.9 ± 0.8 0
ECT 9 0.69 ± 0.03 23 19.6 ± 2.2 2
a Surviving fraction of LPB cells
b Tumor doubling time of the tumors that regrew after the treatment (AM ± SEM)
c Tumor cures were determined 100 days after the treatment
d Electroporation
Table 2 Comparison between tumor curability dose (TCD50), confidence interval (95%) and EF for irradiation alone (IR) or combined with electroporation (EP-IR), BLM (BLM-IR), and ECT (ECT-IR) of LPB sarcoma tumors.
Group TCD50 Confidence interval (95-%) EF
IR 23.1 22.6–23.6
EP-IR 22.1 21.7–22.7 1.0
BLM-IR 22.8 22.4–23.4 1.0
ECT-IR 12.4 11.9–13.0 1.9
==== Refs
Chen J Stubbe J Bleomycins: towards better therapeutics Nat Rev Cancer 2005 5 102 112 15685195 10.1038/nrc1547
Haskell CH Haskell CH Drugs used in cancer chemotherapy Cancer Treatment 1990 Philadelphia: Saunders WB 63 65
Djeridane M Oudard S Escoffre-Barbe M Lacote-Thierry L Desablens B Briere HJ Dib M Cassasus P Ghandour C Lamy T Lejeune F Simon M Traulle C Vigier M Maisonneuve H Breiere J Colonna P Andrieu JM Treatment of patients with advanced or bulky Hodgkin's disease with 12-week doxorubicin, bleomycin, vinblastine, and dacarbazine-like chemotherapy regimen followed by extended-field full-dose radiotherapy Cancer 2002 95 2169 2179 12412171 10.1002/cncr.10932
Leith JT Lee ES Leite DV Glicksman AS Enhanced X ray sensitivity of human colon tumour cells by combination of N-methylformamide with chemotherapeutic agents Int J Radiat Oncol Biol Phys 1986 12 1423 1427 2428792
Lelieveld P Scoles MA Brown JM Phil D Kallman RF The effect of treatment in fractionated schedules with the combination of X-irradiation and six cytotoxic drugs on the RIF-1 tumour and normal mouse skin Int J Radiat Oncol Biol Phys 1985 11 111 121 4038500
Molin J Sogaard PE Overgaard J Experimental studies on the radiation-modifying effect of bleomycin in malignant and normal mouse tissue in vivo Cancer Treat Rep 1981 65 583 589 6166372
Smid L Budihna M Zakotnik B Soba E Strojan P Fajdiga I Zargi M Oblak I Dremelj M Lesnicar H Postoperative concomitant irradiation and chemotherapy with mitomycin C and bleomycin for advanced head-and-neck carcinoma Int J Radiat Oncol Biol Phys 2003 56 1055 1062 12829141 10.1016/S0360-3016(03)00207-4
Teicher BA Herman TS Holden SA Combined modality therapy with bleomycin, hyperthermia and radiation Cancer Res 1988 48 6291 9297 2460220
Pron G Mahrour N Orlowski S Tounekti O Poddevin B Belehradek JJr Mir LM Internalisation of the bleomycin molecules responsible for bleomycin toxicity: a receptor-mediated endocytosis mechanism Biochem Pharmacol 1999 57 45 56 9920284 10.1016/S0006-2952(98)00282-2
Belehradek J JrOrlowski S Ramirez LH Pron G Poddevin B Mir LM Electropermeabilisation of cells in tissues assessed by the qualitative and quantitative electroloading of bleomycin Biochem Biophys Acta 1994 1190 155 163 7509192
Cemazar M Miklavcic D Scancar J Dolzan V Golouh R Sersa G Increased platinum accumulation in SA-1 tumour cells after in vivo electrochemotherapy with cisplatin Br J Cancer 1999 79 1386 1391 10188880 10.1038/sj.bjc.6690222
Jaroszeski MJ Dang V Pottinger C Hickey J Gilbert R Heller R Toxicity of anticancer agents mediated by electroporation in vitro AntiCancer Drugs 2000 11 201 208 10831279 10.1097/00001813-200003000-00008
Gothelf A Mir LM Gehl J Electrochemotherapy: results of cancer treatment using enhanced delivery of bleomycin by electroporation Cancer Treat Rev 2003 29 371 387 12972356 10.1016/S0305-7372(03)00073-2
Mir LM Therapeutic perspectives of in vivo cell electropermeabilization Bioelectrochemistry 2001 53 1 10 11206915 10.1016/S0302-4598(00)00112-4
Sersa G Cemazar M Rudolf Z Electrochemotherapy: advantages and drawbacks in treatment of cancer patients Cancer Ther 2003 1 133 142
Gehl J Skovsgaard T Mir LM Vascular reactionts to in vivo electroporation: characterization and consequences for drug and gene delivery Biochem Biophys Acta 2002 1569 51 58 11853957
Sersa G Cemazar M Parkins CS Chaplin DJ Tumour blood flow changes induced by application of electric pulses Eur J Cancer 1999 35 672 677 10492645 10.1016/S0959-8049(98)00426-2
Sersa G Cemazar M Miklavcic D Chaplin DJ Tumour blood flow modifying effect of electrochemotherapy with bleomycin Anticancer Res 1999 19 4017 4022 10628347
Sersa G Krzic M Sentjurc M Ivanusa T Beravs K Kotnik V Coer A Swartz HM Cemazar M Reduced blood flow and oxygenation in SA-1 tumours after electrochemotherapy with cisplatin Br J Cancer 2002 87 1047 1054 12434299 10.1038/sj.bjc.6600606
Kranjc S Cemazar M Grosel A Scancar J Sersa G Electroporation of LPB sarcoma cells in vitro and tumours in vivo increases radiosensitizing effect of cisplatin Anticancer Res 2003 23 275 282 12680224
Sersa G Kranjc S Cemazar M Improvement of combined modality therapy with cisplatin and radiation using electroporation of tumours Int J Radiat Oncol Biol Phys 2000 46 1037 1041 10705027 10.1016/S0360-3016(99)00464-2
Sersa G Cemazar M Rudolf Z Fras AP Adenocarcinoma skin metastases treated by electrochemotherapy with cisplatin combined with radiation Radiol Oncol 1999 33 291 296
Maxim PG Carson JJ Ning S Knox SJ Boyer AL Hsu CP Benaron DA Walleczek J Enhanced effectiveness of radiochemotherapy with tirapazamine by local application of electric pulses to tumours Radiat Res 2004 162 185 193 15387146
Okunief FP Morgan D Niemerko A Suit HD Radiation dose response of human tumours Int J Radiat Oncol Biol Phys 1995 32 1227 1237 7607946 10.1016/0360-3016(94)00475-Z
Sentjurc M Cemazar M Sersa G EPR oximetry of tumours in vivo in cancer therapy Spectrochim Acta Part A 2004 60 1379 1385
Lin JC Ho ES Jan JS Yang CH Liu FS High complete response rate of concomitant chemoradiotherapy for locally advanced squamous cell carcinoma of the uterine cervix Gynecol Oncol 1996 61 101 108 8626094 10.1006/gyno.1996.0105
Tseng CJ Chang CT Lai CH Soong YK Hong JH Tang SG Hsueh S A randomized trial of concurrent chemoradiotherapy versus radiotherapy in advanced carcinoma of the uterine cervix Gynecol Oncol 1997 66 52 58 9234921 10.1006/gyno.1997.4721
Begg AC Deurloo MJ Kop W Bartelink H Improvement of combined modality therapy with cisplatin and irradiation using intratumoural drug administration in murine tumours Radiother Oncol 1994 31 129 137 8066192 10.1016/0167-8140(94)90393-X
Harrington KJ Rowlinson-Busza G Syrigos KN Vile RG Uster PS Peters AM Stewart JS Pegylated liposome-encapsulated doxorubicin and cisplatin enhance the effect of radiotherapy in a tumour xenograft model Clin Cancer Res 2000 6 4939 4949 11156255
Ning S Yu N Brown DM Kanekal S Knox SJ Radiosensitization by intratumoural administration of cisplatin in a sustained-release drug delivery system Radiother Oncol 1999 50 215 223 10368046 10.1016/S0167-8140(98)00134-0
Yapp DT Lloyd DK Zhu J Lehnert SM The potentiation of the effect of radiation treatment by intratumoural delivery of cisplatin Int J Radiat Oncol Biol Phys 1998 42 413 420 9788424 10.1016/S0360-3016(98)00236-3
Engstrom P Salford LG Persson BRR Jaroszeski MJ, Heller R, Gilbert R Distribution of bleomycin in a rat model Electrochemotherapy, Electrogenetherapy, and transdermal drug delivery 2000 New Jersey: Humana Press
Gehl J Geertsen PF Efficient palliation of hemorrhaging malignant melanoma skin metastases by electrochemotherapy Melanoma Res 2000 10 585 589 11198481 10.1097/00008390-200012000-00011
Molls M Vaupel P Blood perfusion and microenvironment of human tumours 2000 Berlin: Springer-Verlag
Bonnafous P Vernhes M Teissie J Gabriel B The generation of reactive-oxygen species associated with long-lasting pulse-induced electropermeabilisation of mammalian cells is based on a non-destructive alteration of the plasma membrane Biochem Biophys Acta 1999 1461 123 134 10556494
Gabriel B Teissie J Spatial compartmentation and time resolution of photooxidation of a cell membrane probe in electropermeabilized Chinese hamster ovary cells Eur J Biochem 1995 228 710 718 7737168 10.1111/j.1432-1033.1995.tb20314.x
|
16168056
|
PMC1261257
|
CC BY
|
2021-01-04 16:03:05
|
no
|
BMC Cancer. 2005 Sep 16; 5:115
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-115
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1201617153210.1186/1471-2407-5-120Technical AdvanceUsefulness of PKH fluorescent labelling to study leukemic cell proliferation with various cytostatic drugs or acetyl tetrapeptide – AcSDKP Boutonnat Jean [email protected] Anne-Marie [email protected] Jean-Pierre [email protected] Jérôme [email protected] Johanna [email protected] Magali [email protected] Josiane [email protected] Xavier [email protected] Pierre-Emmanuel [email protected] Laboratoire de Dynamique Cellulaire, E.P.H.E, IFRT 130, CNRS UMR 5525, 38706 La Tronche cedex, France2 Département d'Hématologie-Oncologie, Hôtel Dieu, 75004 Paris, France3 Institut de Chimie des Substances Naturelles, CNRS, avenue de la Terrasse, 91198, Gif-sur-Yvette, France4 Département de Langue, Université Joseph Fourier, 38706 La Tronche cedex, France2005 20 9 2005 5 120 120 15 3 2005 20 9 2005 Copyright © 2005 Boutonnat 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
PKH67 labelling was compared for classical proliferation assessment (using S phase evaluation) to analyse the cell proliferation of 29 AML patients treated or not with various drugs. Among these drugs, the effect of tetrapeptide AcSDKP or AcSDKP-NH2 on AML cells, stimulated or not by cytokines, was also evaluated in order to determine (i) if AcSDKP was able to inhibit blast cell proliferation as it inhibits haematopoietic progenitors (ii) if AcSDKP-NH2 was more stable than AcSDKP with FBS.
Methods
For PKH labeling, cells were suspended in Diluent C, and rapidly admixed with PKH67 solution at 20 μM PKH67. Staining was stopped by addition of FBS.
Results
A good correlation between PKH67 labelling and bromodeoxyuridine incorporation was obtained first with 6/9 patients for control cells, then for 11/17 AML patients treated with classical antileukemic drugs (among whom 4 were also treated with AcSDKP). The effect of AcSDKP was also studied on 7 patients. The discrepancy between both methods was essentially due to an accumulation of cells into different cycle phases measured by BrdUrd incorporation secondary to drug action and PKH67 labelling which measured the dynamic proliferation. This last method allows identifying resistant cells which still proliferate. AcSDKP or AcSDKP-NH2 induced a decrease of leukemic cell proliferation in 5/7 patients when cytokines were added (in order to stimulate proliferation) one day after tetrapeptide AcSDKP or AcSDKP-NH2. No effect on proliferation was noted when cytokines were added to AcSDKP-NH2.
Conclusion
PKH67 labelling method is a powerful tool for cell proliferation assessment in patients with AML, even in cells treated by various drugs.
==== Body
Background
The successful treatment of acute myeloid leukaemia (AML) is frequently impeded by the development of resistance to a wide spectrum of cytotoxic drugs and by cell proliferation. Daunorubicin (DNR), Cytarabine (AraC), Etoposide (VP16), Mitoxantrone (Mitox), and Amsacrin (AMSA) are used in the treatment of AML and can induce drug resistance [1]. Various methods are available to assess leukemic cell proliferation. Common methods for proliferation assessment, such as bromodeoxyuridine (BrdUrd) incorporation, are correlated to treatment outcome [2,3]. BrdUrd is an analogue of thymidine and can be incorporated specifically into DNA instead of thymidine. BrdUrd incorporation was described, in literature, as a reference technique for cell proliferation evaluation but is often difficult to standardize [3]. Evaluation of cell distribution in each phase could also be determined by monoparameter analysis after stoichiometric DNA labelling using propidium iodide (PI) [4]. These two methods require cell fixation and cell permeabilization whereas PKH dye labelling can be performed on living cells.
PKH (from the author who developped these dyes: Paul Karl Horan) are vital lipophilic, fluorescent, membrane intercalating dyes [5]. They contain two long alkyl chains, which allow a strong anchorage in the lipid bilayer. When labelled cells divide, the resulting daughter cells receive half the label, reducing the fluorescence intensity to one-half that the parent cells. As a consequence, the proliferation of labelled cells is correlated to a decrease in fluorescence [6,7]. Drugs such as DNR, a fluorescent molecule, do not interfere with PKH67 staining, when a delay (3 hours minimum) between PKH67 labelling and DNR incubation is respected [8].
The tetrapeptide acetyl-N-Ser-Asp-Lys-Pro (AcSDKP) isolated from bone marrow was identified as a physiological regulator of hematopoietic stem cell proliferation [9]. It inhibits the proliferation of normal haematopoietic stem cells and early progenitors in vivo as well as in vitro [10-12]. However, the AcSDKP role on cell proliferation has been discussed. Some authors have reported that AcSDKP has no effect on the proliferative status of leukemic progenitors [11] and therefore may selectively prevent the cycle initiation of normal stem cells.
Recent studies have reported that AcSDKP is inactivated by foetal bovine serum (FBS). It is hydrolyzed in blood by the soluble angiotensin-I converting enzyme (ACE) [13]. A new AcSDKP (AcSDKP-NH2) was developed to increase its stability against ACE degradation in FBS and blood. Therefore, it was interesting to know if this analogue also shared common properties with AcSDKP on the proliferation status of leukemic cells.
The aim of this study was to compare the proliferation of 29 AML cells from patients treated or not with cytostatic drugs using two methods: i) dye dilution method using PKH67 ii) or DNA content. The AcSDKP or AcSDKP-NH2 effect on cell proliferation was analyzed.
Methods
Reagents
Ficoll, PKH67 and Diluent C were given by Sigma-Aldrich (St Quentin Fallavier, France). Daunorubicin (DNR), Aracytine (AraC) and Amsacrine (AMSA) were purchased from Roger-Bellon (Neuilly sur Seine, France). Mitoxantrone (Mitox) was given from Léderlé (Rungis, France). VP16 was supplied by Pierre-Fabre (Castres, France). BrdUrd and anti-BrdUrd were purchased from Roche-Diagnostic (Meylan, France). Cell growth medium, fetal bovine serum (FBS), phosphate-buffered saline (PBS), and cellulose syringe filters (0.45 mm) were purchased from Invitrogen (Cergy-Pontoise, France).
Recombinant granulocyte-macrophage colony stimulating factor (GM-CSF) and granulocyte colony stimulating factor (G-CSF) were used at a concentration of 20 ng/mL. Interleukin 3 (IL3) was used at 50 ng/mL. The stem cell factor (SCF) concentration was 25 ng/mL, and recombinant human ertyhropoietin (EPO) was used at 1.5 UI/mL. All growth factors were purchased from Biosource (California, USA).
The synthetic tetrapeptide (Acetyl-N-Ser-Asp-Lys-Pro) or AcSDKP-NH2 was generously donated by J. Bignon and J.Thierry respectively (Institut de Chimie des Substances Naturelles, CNRS, France).
AML cells sampled from patients
Leukemic cells were obtained from peripheral blood or bone marrow samples in 29 AML patients, separated on a gradient of density (Ficoll). Only the patients samples with more than 50% of living blasts were included in the experiments. The 29 patients were studied as follow: 9 using BrudUrd incorporation and PKH67 labelling without drug treatment, 17 using BrudUrd incorporation and PKH67 labelling in the presence of drug or not; among the 18 patients, 4 were studied along with the remaining 3 using PKH67 labelling and DNA content analysis in order to determine AcSDKP or AcSDKP-NH2 effect.
Cells were cultured in RPMI-1640 medium supplemented with 10% FBS, 2 mM glutamine, 100 UI/mL penicillin, and 100 ng/mL streptomycin, at 37°C in a humidified atmosphere with 5% CO2.
Analysis were performed every day, as long as the percentage of viable cells was higher or equal to 50%. Viable and necrotic cells were identified, using flow cytometry with forward and side scatter parameters.
PKH67 labelling
PKH67 labelling was performed as described previously [14]. Briefly, 107 cells were suspended in 1.0 mL of Diluent C, and stained by rapidly admixing with a 20 μM working PKH67 solution, prepared by diluting 20 μl of 10-3 M ethanolic dye stock in 1.0 mL of Diluent C immediately prior to staining. Final staining concentration was therefore 10 μM PKH67 and 5 × 106 cells/mL). Staining was stopped after 3 minutes by addition of 2 mL of FBS and cells were washed 3 times with 5 mL of RPMI-1640 containing 10% FBS. For each sample, an aliquot of blasts was fixed with 2% PFA at D0 and kept at +4°C in order to maintain the original fluorescence.
BrdUrd incorporation
107 cells were cultured in complete medium for each sample. After 24 hours, 5 μM BrdUrd were incorporated in control cells for 15 minutes at 37°C. After elimination of excess BrdUrd by 2 washings, cells were fixed by 70° ethanol and maintained at +4°C during 24 hours. Revelation of BrdUrd molecules was made by treating cells with hydrochloric acid 4N during 15 minutes at room temperature. After several washings, 20 μL of Ac anti-BrdUrd [15] were added to control cells for 30 minutes at room temperature. Ac anti-BrdUrd was revealed by Fab'2 (antibody coupled with FITC, 0.4 μg/mL) for 30 minutes at room temperature. Cells were labeled simultaneously with 10 μg/mL PI and treated with RNase (1 mg/mL) during 30 minutes at room temperature, then analyzed using a FACSCalibur (BD Biosciences, le Pont de Claix, France) to measure DNA content (G0/1 and G2+M phases).
Evaluation of blast proliferation following drug treatment
Blasts were collected and labeled with PKH67. Cells were treated during four days with one or the combination of several drugs after one day of culture: DNR, AraC, AMSA, VP16, and Mitox, at a concentration of 10-6 M. These concentrations are commonly used in therapy.
AcSDKP effect on proliferation of cells (stimulated or not by cytokines)
2.106 labelled with PKH67 were seeded in 25 cm2 culture dishes numbered as described below:
Sample 1: PKH67 labelled cells without the mixture of cytokines in complete medium at 37°C.
Sample 2: PKH67 labelled cells were cultured with cytokines in complete medium at 37°C (G-CSF, IL3, SCF, GM-CSF, and EPO).
Sample 3: cells with cytokines and 10-9 M AcSDKP-NH2 in complete medium at 37°C.
Sample 4: cells without cytokines and 10-5 M AcSDKP in complete medium at 37°C.
Sample 5: cells without cytokines and 10-9 M AcSDKP in complete medium at 37°C.
Sample 6: cells without cytokines and 10-9 M AcSDKP-NH2 in complete medium at 37°C.
Sample 7: cells without cytokines and 10-9 M AcSDKP-NH2 in complete medium at 37°C. After a day of culture, cells contained in samples 4, 5, and 6 were washed and cytokines were added to the complete medium.
The tetrapetide AcSDKP concentrations used upper are known to be active.
DNA content analysis
The cell cycle was evaluated with the Cycle Test™ kit (BD-Biociences, Le Pont de Claix, France). Briefly, cells from patients were incubated with trypsin in a spermine tetrahydrochloride detergent buffer for 10 min at room temperature. Trypsin inhibitor and ribonuclease A were added for 10 min without washing. Finally, PI was added and incubated for 10 min, then cells were immediately analyzed by flow cytometry and distribution of cells in each phase was evaqluated using Modfit software (Verity software). Since both tetrapeptide could induce an increase of cells in G0/1 phase with the Cycle Test™ kit was used because it was easier than Brdu incorporation to evaluate few modification in cell cycle repartition.
Flow cytometry analysis
Mean fluorescence intensity per cell was measured using a FACSCalibur flow cytometer (BD Biosciences, Le Pont de Claix, France) equipped with an air-cooled argon ion laser emitting 15 mW at 488 nm and a photodiode laser emitting 10 mW at 635 nm. PKH67 and FITC fluorescence were collected with a 530 ± 30 nm band-pass filter; PI fluorescence was collected with a 585 ± 44 nm band-pass filter. 50,000 events were acquired with the CellQuest software (BD Biosciences, Le Pont de Claix, France). Electronic compensation settings for FITC and PI were FL1-FL2 = 10% and FL2-FL1 = 35%; and FL1-FL2 = 8% and FL2-FL1 = 30% for PKH67/PI. Kolmogorov Sminov statistical test (K/S test) was used to point out difference between PKH fluorescence histograms. Differences between histograms were considered as statistical significant when p was under 0.001.
Results
We compared blast proliferation with two methods: a reference method (BrdUrd incorporation) and a dye dilution method (PKH67 labelling), in order to validate the use of PKH for proliferation assessment of living AML cells. Table 1 shows the cell cycle distribution and the mean PKH67 fluorescence of cells from 9 patients. Fluorescence ratio is, PKH67 fluorescence at D0 divided by PKH67 fluorescence at D4 (D0/D4). Three different groups could be described according to the proliferation rate determined by PKH67 decrease and BrdUrd incorporation i) first group with proliferation corresponding to PKH67 decrease and an S phase superior to 4% ii) second group with low S phase (1%) and PKH67 ratio under 1.2 and iii) a third group with a discordance between S phase and PKH67 decrease. Using the Kolmogorov Sminov test we showed that with a PKH67 fluorescence ratio at 1.21 there was a significant difference between histograms [p < 0.001 for patients (Lio., Pi., Tr., Gr.)] but not below this value. We considered that cells proliferated when PKH67 fluorescence ratio was superior to 1.21. In the first group of patients, (Lio., Pi., Tr., Gr.) PKH67 fluorescence ratios were 2.96, 1.37, 1.39, 1.29 respectively (table 1) and showed a cell proliferation correlated to the S phase fraction (30% to 4%) (Figure 1a).
In the second group, patients Dun and Br (Figure 1b) showed a low decrease of PKH67 fluorescence, with a fluorescence ratio (1.15 and 1.16) respectively at D0/D4. The cell cycle analysis showed that 1% of the cells was in S phase (Figure 1b).
In the third group, the S phase ranged from 8 to 11% but the PKH67 ratio ranged from 1.06 to 1.16. These 3 patients showed a discrepancy between PKH67 fluorescence, revealing an interruption of proliferation and S phase (Figure 1c).
Evaluation of blast proliferation treated with various drugs (Table 2)
In this part of experimentation, 17 patients were analysed. According to the initial results shown in Table 1 and the K/S test, three groups were identified: (i) cells from 4 patients which still underwent proliferation even after drug treatment (ii) cells from 6 patients which were more or less sensitive to one or more drugs associated with interruption of proliferation, and (iii) 7 patients with no spontaneous proliferation.
Cells from patient Gi presented a moderate spontaneous proliferation (PKH67 ratio at 1.4 and 8% of cells in S phase) (Figure 2). When cells were treated with AraC, DNR, or both no decrease of PKH67 ratio was noted. Small modifications of cell cycle phase distribution were seen essentially with DNR or DNR + AraC with an accumulation in G2+M phase.
For the cells of patient Lh, a decrease of the PKH67 ratio compared to the control (3.7) was noted for AraC (2.32), DNR (2.09), or combination of both (1.92) but cells still underwent proliferation. When cells were treated with AraC a moderate accumulation was seen in the S phase (39%) and an accumulation in G2+M phase for DNR (30%) or DNR+ AraC (24%). For both patients, Da and Go, no accumulation was noted in one of the cell cycle phases but with AraC a moderate decrease of proliferation kinetics was pointed out by a decrease of PKH67 fluorescence ratio (1.4 and 1.3 respectively) compared to the control (1.6).
In the second group Sat, Le, Fo (figure 3), and Ko cells had spontaneous proliferation but each drug interrupted proliferation. More often AraC induces an accumulation in G1 or S phase and the others drugs, Amsacrine, daunorubicine, or VP16 induced an accumulation in G2+M. When drugs were combined, accumulation in one cell cycle phase was not obvious. For patients Sa and Na, treatment with AraC induced a decrease of proliferation shown by a PKH67 fluorescence ratio decrease, 1.4 and 2.5 respectively compared to 3 for the control. As for other patients in the group, DNR interrupted the proliferation with a PKH ratio under 1.2. In the third group no spontaneous proliferation was noted "a fortiori" under treatment.
The evaluation of AcSDKP on blast proliferation stimulated or not by cytokines (Table 3)
We used the PKH67 assay and Cycle test from BD to assess proliferation on blasts incubated with cytokines, AcSDKP or AcSDKP-NH2.
In this experiment, two groups of patients could be individualized: three patients underwent spontaneous proliferation (Be, Dur, Ko), and the four others (Pa, Ph, Al, Av) underwent proliferation only after cytokine stimulation.
The proliferation of cells from patient Be was not more stimulated by adding cytokines compared to the control and no effect of AcSDKP or AcSDKP-NH2 was noted on PKH67 fluorescence ratio or on the cell cycle phase. Cells obtained from Dur patient were stimulated by cytokines and an increase of proliferation was noted according to the increase of PKH67 fluorescence ratio (1.5 compared to the control 1.2) and no effects of AcSDKP or AcSDKP-NH2 was noted on cell cycle phase or PKH67 fluorescence ratio. Cells from patient Ko were stimulated by cytokines but if AcSDKP or AcSDKP-NH2 was added a day before cytokines (sample 3) cell proliferation was interrupted along with a decrease of the S phase associated with an increase of G0/1 phase and a decrease of PKH67 fluorescence ratio (1 for sample 4, 1.12 for sample 5, 1.11 for sample 6 compared to the control 1.49 sample 2).
In the second group, cells from patients Pa and Ph (Figure 4) underwent proliferation after cytokine stimulation but if AcSDKP or AcSDKP-NH2 was added a day before adding cytokines a slow decrease of proliferation was noted (0.2 in the fluorescence ratio in sample 4,5,6 compared to the ratio of sample 2 or 3). No obvious difference was noted in the proportion of S phase in the various samples. For the two last patients (Al and Av) AcSDKP or AcSDKP-NH2 which was added one day before adding cytokines interrupted the proliferation with a PKH67 fluorescence ratio and a percentage of S phase close to 1.
Discussion
This study had for aim to compare cell proliferation of 29 AML patients by two methods i) S phase evaluation and DNA content for proliferation assessment and ii) PKH67 labelling to monitor proliferation of living blast treated or not cytostatic drugs or AcSDKP a regulator of stem cells.
The proliferation assessment, using BrdUrd incorporation and PKH67 labelling, was carried out on 9 AML patients, 7 of whom had been recently diagnosed and the other 2 were relapsing. In order to compare results, the K/S test was applied to point out statistical difference between PKH67 fluorescence histograms. Results were significant when the ratio of the PKH67 histograms fluorescence was above 1.2. In the first group of patients, the proliferation was highly significant identified by BrdUrd incorporation and by a decrease of PKH67 fluorescence. These results show a good correlation between the S phase percentage and proliferation rate. In the second group, no objective proliferation was shown either with PKH67 labelling or S phase. In the third group a discrepancy was noted between the S phase evaluation and proliferation using PKH67 labelling with high S phase and no significant proliferation with PKH67 labelling, The discrepancy between the two methods can be explained by the fact that BrdUrd incorporation associated to the measure of the DNA content, allows to estimate cell distribution in various phases like a snapshot, as opposed to PKH67 which measures cell division. The evaluation of dynamic proliferation, using BrdUrd incorporation, would require using the "pulse – chase" method which is time-consuming and difficult to set-up [16].
The correlation between PKH67 labelling and the rate of S phase cells had to be validated when cells are cultured with cytostatic drugs. Indeed out of 17 AML patients, treated with drugs, 11 showed a good correlation between the S phase percentage of cells and the decrease of PKH67 fluorescence. In the first group of patients (4 cases), a significant proliferation of cells was observed when cells were treated with various drugs. This proliferation of cells treated by DNR, VP16, Mitox could be due to ATP binding cassette proteins such as Pgp, MRP, and BCRP [17-19]. Pgp, MRP, BCRP are known to extrude DNR, VP16, and Mitox. The proliferation of cells treated by AraC could be explained by spliced deoxycytidine kinase [20]. F. Lacombe et al., [2], reported that a minimum of 3% of cells in phase S was necessary to show a significant difference, in a study of DNA synthesis inhibition of by AraC. These results are correlated to our results obtained with the K/S test. AraC, a drug often used in therapy [21], was described in literature as an antimetabolite involving the S phase accumulation of cells [22,23].
In the second group, discrepancies were seen between the percentage of S phase and the PKH67 fluorescence ratio in the presence of AraC, in two cases. The S phase percentage was superior to 20% with a BrdUrd/FITC and PI biparametric analysis. The PKH67 fluorescence histogram did not reveal any cellular proliferation. In these cases, cells seemed to accumulate in phase S. The same profile was observed when cells were treated with the DNR+AraC mixture. Both drugs were in competition and only AraC dominated, because an accumulation of cells was obtained in phase S and not in phase G2+M. We did not observe cumulative effects between these two drugs. VP16 and AMSA are inhibitors of topo-isomerases II and we observed a low decrease of PKH67 fluorescence in 11 cases, thus a low rate of proliferation correlated to the S phase percentage of cells. We did not observe any modification of cell cycle distribution in the presence of these drugs. These two drugs were reported as inducing cell accumulation in G2+M [24,25].
We thus studied a possible effect of the tetrapeptide AcSDKP or the amide AcSDKP (which was supposed to be more stable) on proliferation of AML blasts, previously stimulated or not by cytokines. AcSDKP isolated from bone marrow was identified as a physiological regulator of cell proliferation [9]. AcSDKP is known to inhibit cell proliferation of normal haematopoietic in vivo and in vitro [10,26]. However, it was described as being ineffective on the proliferation of leukemic cells [27] and thus it could selectively prevent the cell cycle initiation. It was also described as a protector of normal haematopoietic human cells, against the toxic effects of drugs, and the effects of radiotherapy [28-31]. It has no effect on the growth and DNA synthesis of HL60 leukemic cells [27]. The biological properties of AcSDKP and the absence of antiproliferative activity on leukemic cells suggest that possible therapeutic applications, such as the protection of hematopoietic cells, could be used in association with chemotherapy. The Institute of Chemistry of Natural Substances (CNRS, Gif-sur-Yvette, France.) developed a tetrapeptide amide to limit degradation by an ACE enzyme when cells are cultured with FCS or blood [32]. We demonstrated that there was no difference between the tetrapeptide amide and the natural tetrapeptide. This could be explained by the fact that the medium used for the culture of blastic cells contained only 10% of FCS. AcSDKP had no effect on the proliferation of stimulated cells in 5 patients [27], as well as on leukemic HL60 and K562 cells lines, and on CML cells [31,32]. We observed a low rate of, or an inhibition of proliferation in 3 patients when cytokines were added one day after AcSDKP. M Smeets et al. [33] reported that "noncycling" progenitors, whether normal or leukemic, presented a relatively increased level of MDR protein expression. We probably modified the expression of resistant proteins when we activated the proliferation of these cells, by a decrease of resistant protein expression. P Te Boekhorst et al., [34] showed the existence of a relationship between the Pgp protein and a high fraction of cells in phase S. They showed that high S phase was frequently associated with the expression of multidrug resistance proteins and poor prognosis, in acute myeloid leukaemia.
Conclusion
Our study more frequently demonstrated a good correlation between PKH 67 fluorescence decrease and proportion of S phase obtained by BrdUrd incorporation. We observed a discrepancy when cells were blocked in one of the cell cycle phases and demonstrated the power of PKH67 labelling to follow proliferation even when cells were treated with drugs. This method could be applied to sort proliferating cells growing with drugs and to determine their chemoresistant protein profiles. We also demonstrated the absence of benefit of the NH2 AcSDKP compared to AcSDKP since both are able to modify blast cell proliferation which prevents using them in therapy.
Abreviations
PKH: Paul Karl Horan, BrdUrd: bromodeoxyuridine, PI: propidium iodide, AcSDKP: tetrapeptide acetyl-N-Ser-Asp-Lys-Pro, NH2AcSDKP : tetrapeptide acetyl-N-Ser-Asp-Lys-Pro amide, DNR: Daunorubicin, AraC: Aracytine, AMSA: Amsacrine, Mitox: Mitoxantrone, VP16: Etoposide.
Competing interests
The author(s) declare that they have no competing interest.
Authors' contributions
A-M F acquired and analyzed part of the data. J Bi, and J W provided drugs. M B wrote the first draft of the manuscript. J Bo, J-P M, and X R conceived and the designed the study and provided guidance to all aspects of this project. English revision of the maniuscript was done by P-E C. 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 the "Ligue Nationale contre le Cancer, Comité de Savoie & Haute-Savoie", GEFLUC and ESPOIR societies.
Figures and Tables
Figure 1 Comparative flow cytometry analyses of percentage of cells in S phase, using BrdUrd incorporation and PKH67 labelling {PKH fluorescence intensity (a.u.) at Day 0 / PKH fluorescence intensity (a.u.) at Day 4}. Figure 1a patient Pi, Figure 1b patient Dun, Figure 1c patient Ga.
Figure 2 Comparative flow cytometry analyses of percentage of cells in S phase using BrdUrd incorporation and PKH67 ratio in patients Gi., treated with DNR, AraC, and a mixture of DNR+AraC after 4 days.
Figure 3 Comparative flow cytometry analyses of percentage of cells in S phase using BrdUrd incorporation and decrease of PKH67 fluorescence ratio in patient Fo., treated with DNR, AraC, AMSA, and a mixture of DNR+AMSA+AraC after 4 days.
Figure 4 Comparative flow cytometry analyses using PKH67 fluorescence ratio (PKH®) and cell cycle of patient Ph. treated with cytokines and AcSDKP after 4 days of culture.
Table 1 Comparison between BrdUrd incorporation asociated with DNA content (% of cell in cell cycle) at day 4 and PKH67 labelling ratio {PKH fluorescence intensity (a.u.) at Day 0 by PKH fluorescence intensity at Day 4} in 9 AML patients.
Patients Cell cycle Cell proliferation
G0/1% S% G2+M % D0 (a.u.) D4 (a.u.) PKH ratio
Lio
47 30 23 579 194 2.96
Pi
72 10 18 1966 1424 1.37
Tr
66 11 23 1851 1324 1.39
Gr
76 4 20 1988 1324 1.21
Dun
64 1 35 4821 4138 1.16
Br Relapse
76 1 23 3960 3438 1.15
Li
69 11 20 3020 2834 1.07
Ga
81 9 10 1995 1853 1.06
To
69 8 23 1444 1257 1.16
Table 2 Comparison between BrdUrd incorporation and PKH67 labelling in 17 AML patients treated with various drugs (DNR, AraC, VP16, Mitox, and AMSA). Percentage distribution of patient cells in each cell cycle phase obtained by BrdUrd incorporation and PKH67 labelling ratio at day 4 (PKH fluorescence intensity at Day 0 by PKH fluorescence intensity at Day 4).
GI
Control AraC DNR AraC + DNR
G0/1 89 88 74 72
S 8 6 8 4
G2+M 3 6 18 24
PKH ratio 1,4 1,35 1,37 1,25
Lh
Control AraC DNR AraC + DNR
G0/1 57 53 54 53
S 33 39 16 23
G2+M 10 8 30 24
PKH ratio 3,7 2,32 2,09 1,93
Da
Control AraC VP16 Mitox AraC + VP16 + Mitox
G0/1 74 76 66 73 71
S 7 10 10 12 17
G2+M 19 14 24 15 12
PKH ratio 1,6 1,4 1,4 1,5 1,4
Go
Control AraC VP16 Mitox AraC + VP16 + Mitox
G0/1 60 64 85 71 75
S 16 10 4 8 2
G2+M 24 26 11 21 23
PKH ratio 1,6 1,3 1,6 1,6 1,3
Sat
Control AraC VP16 Mitox AraC + VP16 + Mitox
G0/1 82 87 89 89 89
S 14 7 4 1 1
G2+M 4 6 7 10 10
PKH ratio 1,7 1,1 1,1 1,1 1,1
Le
Control AraC VP16 Mitox AraC + VP16 + Mitox
G0/1 70 61 56 77 77
S 14 19 16 2 1
G2+M 16 20 28 21 22
PKH ratio 2,3 1,12 1,07 1,05 1,05
Fo
Control AraC AMSA DNR DNR+AraC+AMSA
G0/1 81 75 90 71 90
S 14 20 2 6 2
G2+M 5 5 8 23 8
PKH ratio 1,7 1,1 1,18 1,18 1,18
Ko
Control AraC DNR VP16 AraC+DNR+VP16
G0/1 71 80 70 74 76
S 13 2 8 4 3
G2/M 16 18 22 22 21
PKH ratio 1,5 1,05 1,13 1,04 1,06
Sa
Control AraC DNR AraC+DNR
G0/1 56 71 76 76
S 32 10 9 5
G2+M 12 19 17 19
PKH ratio 3 1,4 1,1 1,12
Na Control AraC DNR AraC+DNR
G0/1 65 74 57 73
S 31 21 1 23
G2+M 4 5 42 4
PKH ratio 3 2,5 1,13 1,11
Pa
Control AraC DNR AraC+DNR
G0/1 89 90 87 89
S 3 1 1 1
G2+M 8 9 12 10
PKH ratio 1,18 1,1 1,1 1,1
La
Control AraC DNR AraC+DNR
G0/1 83 85 60 80
S 11 7 3 13
G2+M 6 8 37 7
PKH ratio 1,02 1,01 1,1 1,1
Luu
Control AraC DNR AraC+DNR
G0/1 86 82 86 85
S 1 1 2 3
G2+M 13 17 12 12
PKH ratio 1,1 1,12 1,13 1,12
Av
Control AraC DNR AraC+DNR
G0/1 68 65 80 68
S 20 16 13 15
G2+M 12 19 7 17
PKH ratio 1,09 1,07 1,08 1,1
Ph
Control AraC DNR AraC+DNR
G0/1 74 75 77 79
S 1 2 3 1
G2+M 25 23 20 20
PKH ratio 1,1 1,03 1,05 1,07
Li
Control DNR
G0/1 70 73
S 11 7
G2+M 19 20
PKH ratio 1,07 1,08
To
Control DNR
G0/1 92 73
S 5 5
G2+M 3 22
PKH ratio 1,19 1,18
Table 3 Comparison of blast proliferation by cell cycle analysis using PI labelling and PKH67 labelling (PKH fluorescence intensity (a.u.) at Day 0 / PKH fluorescence intensity (a.u.) at Day 4) in 7 AML patients stimulated or not by mixture of cytokines and incubated with a tetrapeptide AcSDKP (amide or not). Sample 1: PKH67 labelled cells cultured without the mixture of cytokines. Sample 2: PKH67 labelled cells cultured with the mixture of cytokines. Sample 3: cells with cytokines and AcSDKP-NH2 in 10-9 M. Sample 4: cells incubated with AcSDKP in 10-5 M to which cytokines were added after one day of culture. Sample 5: cells incubated with AcSDKP in 10-9 M in which cytokines were added after one day of culture. Sample 6: cells incubated with AcSDKP-NH2 in 10-9 M to which cytokines were added after one day of culture. Sample 7: cells without cytokines and AcSDKP-NH2 in 10-9 M.
Patients Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6 Sample 7
Be
PKH ratio 1.24 1.25 1.26 1.27 1.28 1.3 1.26
G0/1 82 88 78 83 80 81 86
S 9 8 10 8 9 7 8
G2+M 9 4 12 9 11 12 6
Dur
PKH ratio 1.2 1.5 1.51 1.49 1.48 1.51 1.25
G0/1 87 78 82 81 81 81 90
S 6 11 8 9 8 8 6
G2+M 7 11 10 10 11 11 4
Ko
PKH ratio 1.51 1.49 1.50 1 1.12 1.11 1
G0/1 90 87 90 94 97 96 96
S 7 6 7 1 1 2 1
G2+M 3 7 3 5 2 2 3
Patients Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6 Sample 7
Pa
PKH ratio 1.18 2.01 2.1 1.8 1.75 1.75 1.13
G0/1 86 73 70 71 70 73 90
S 3 17 17 18 19 19 5
G2+M 11 10 13 11 11 8 5
Ph
PKH ratio 1.18 1.5 1.5 1.31 1.32 1.33 1.11
G0/1 85 78 82 83 76 83 87
S 2 12 12 9 11 9 2
G2+M 13 10 6 8 13 8 11
Al
PKH ratio 1 1.37 1.25 1.01 1.09 1.07 1
G0/1 95 83 89 93 94 91 92
S 1 8 5 1 1 1 1
G2+M 4 9 6 6 5 8 7
Av
PKH ratio 1.1 1.6 2 1.1 1.09 1.02 1
G0/1 92 66 63 90 89 90 94
S 1 16 17 1 1 1 1
G2+M 7 18 20 9 10 9 5
==== Refs
Schaich M Illmer T Aulitzky W Bodenstein H Clemens M Neubauer A Repp R Schakel U Soucek S Wandt H Ehninger G Intensified double induction therapy with high dose mitoxantrone, etoposide, m-amsacrine and high dose ara-C for elderly acute myeloid leukemia patients aged 61–65 years Haematologica 2002 87 808 815 12161356
Lacombe F Belloc F Dumain P Detection of cytarabine resistance in patients with acute myelogenous leukaemia using flow cytometry Blood 1994 84 716 723 8043860
Dolbeare F Gratzner HG Pallavicini MG Gray JW Flow cytomertic mesurement of total DNA content and incorporated bromodeoxyuridine Proc Natl Acad Sci U S A 1983 80 5573 5577 6577444
Lacombe F Belloc F Flow cytometry study of cell cycle, apoptosis and drug resistance in acute leukaemia Hematol Cell Ther 1996 38 495 504 9030962 10.1007/s00282-996-0495-9
Horan PK Slezak SE Jensen BD Cellular proliferation history by fluorescent analysis Cytometry 1988 203 213
Yamamura Y Rodriguez N Schwartz A Eylar E Bagwell B Yano N A new flow cytometric method for quantitative assessment of lymphocyte mitogenic potentials Cell Mol biol 1995 41 S121 S132 8574140
Boutonnat J Barbier M Muirhead K Mousseau M Grunwald D Ronot X Seigneurin D Response of chemosensitive and chemoresistant leukemic cell lines to drug therapy: simultaneous assessment of proliferation, apoptosis, and necrosis Cytometry 2000 42 50 60 10679743 10.1002/(SICI)1097-0320(20000215)42:1<50::AID-CYTO8>3.0.CO;2-C
Barbier M Morjani H Muirhead KA Ronot X Boutonnat J Optimization of PKH67 labeling conditions for the study of daunorubicin-treated leukemic cells Oncol Rep 2005 14 553 560 16012744
Lenfant M Wdzieczak-Bakala J Guittet E Prome JC Sotty D Frindel E Inhibitor of hemotopoietic pluripotent stem cell proliferation: purification and determination of its strucutre Proc Natl Acad Sci USA 1989 86 779 782 2915977
Jackson JD Ozerol E Yan Y Ewel C Talmadge JE Activity of Acetyl-Ser-Asp-Lys-Pro (AcSDKP) on human hemotopoietic progenitors in short-term and long-term murine bone marrow cultures J Hematother Stem Cell Res 2000 9 489 496 10982247 10.1089/152581600419152
Bonnet D Cesaire R Lemoine F Aoudjhane M Najman A Guigon M The tetrapeptide AcSDKP, an inhibitor of the cell-cycle status for normal human hematopoietic progenitors, has no effect on leukemic cells Exp Hematol 1992 20 251 255 1544396
Cashman JD Eaves AC Eaves CJ Tetrapeptide AcSDKP specifically blocks the cycling of primitive normal but not leukemic progenitors in long term culture evidence for an indirect mechanism Blood 1994 84 1534 1542 8068944
Rieger KJ Saez-Servent N Papet MP Wdzieczak-Bakala J Morgat JL Thierry J Voelter W Lenfant M Isnard F Ferero JM Involvement of human plasma angiotensin I-converting enzyme in the degradation of the haemoregulatory peptide N-acetyl-seryl-aspartyl-lysyl-proline C R Acad Sci III 1992 315 545 550 1300237
Boutonnat J Barbier M Seigneurin D Muirhead K Mousseau M Rousselle C Ronot X The use of PKH26 and 67 for cell proliferation assessment in chemosensitive and chemoresistant cells Eur Micros Anal 1998 56 17 19
Gratzner HG Monoclonal antibody to 5-bromo- and 5-iododeoxyuridine: A new reagent for detection of DNA replication Science 1982 218 474 475 7123245
Begg A McNally NJ Shrieve DC Kärcher H A method to measure the duration of DNA synthesis and the potential doubling time from a single sample Cytometry 1985 6 620 626 4064842 10.1002/cyto.990060618
Jones PM Georges AM The Abc transporter structures and mechanism: perspective on recent research Cell Mol Life Sci 2004 61 682 699 15052411 10.1007/s00018-004-4272-z
Van der Pol MA Broxterman HJ Pater JM Feller N van der Maas M Weijers GW Scheffer GL Allen JD Scheper RJ van Loevezijn A Ossenkoppele GJ Schuurhuis GJ Function of the ABC transporters, P-glycoprotein, multidrug resistance protein and breast cancer resistance protein, in minimal residual disease in acute myeloid leukemia Haematologica 2003 88 134 147 12604403
Plasschaert SL Van Der Kolk DM De Bont ES Vellenga E Kamps WA De Vries EG Breast cancer resistance protein (BCRP) in acute leukemia Leuk Lymphoma 2004 45 649 654 15160935 10.1080/10428190310001597928
Veuger MJ Honders MW Spoelder HE Willemze R Barge RM Inactivation of deoxycytidine kinase and overexpression of P-glycoprotein in AraC and daunorubicin double resistant leukemic cell lines Leuk Res 2003 27 445 453 12620296 10.1016/S0145-2126(02)00224-2
Parisi E Draznin J Stoopler E Schuster SJ Porter D Sollecito TP Acute myelogenous leukemia: advances and limitations of treatment Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2002 93 257 263 11925533
Hamada A Kawaguchi T Nakano M Clinical pharmacokinetics of cytarabine formulations Clin Pharmacokinet 2002 41 705 718 12162758
Lacombe F Belloc F Dumain P Puntous M Cony-Makhoul P Bernard P Boisseau MR Reiffers J Detection of the resistance to Ara-C blast cells in acute myeloid leukemia using flow cytometry Bull Cancer 1995 82 349 356 7626842
Poot M Hiller KH Heimpel S Hoehn H Distinct patterns of cell cycle disturbance elicited by compounds interfering with DNA topoisomerase I and II activity Exp Cell Res 1995 218 326 330 7537696 10.1006/excr.1995.1162
Ishiyama K Satoh S Igarashi Y Kumagai H Yahagi A Sasaki H Flow cytometric analysis of the cell cycle of the leukemic cell lines treated with etoposide and cytosine arabinoside Tohoku J Exp Med 1994 174 95 107 7900157
Guigon M Bonnet D Lemoine F Kobari L Parmentier C Mary JY Najman A Inhibition of human bone marrow progenitors by the synthetic tetrapeptide AcSDKP Exp Hematol 1990 18 1112 1115 2209766
Coutton C Guigon M Bohbot A Ferrani K Oberling F Photoprotection of normal human hemotopoietic progenitors by the tetrapeptide N-AcSDKP Exp Hematol 1994 22 1076 1080 7925774
Aidoudi S Guigon M Drouet V Caen JP Han ZC The tetrapeptide AcSDKP reduces the sensibility of murine CFU-MK and CFU-GM progenitors to aracytine in vitro and in vivo Int J Hematol 1998 68 145 155 9803673 10.1016/S0925-5710(98)00048-6
Bogden AE Carde P De Paillette ED Moreau JP Tubiana M Frindel E Amelioration of chemotherapy-induced toxicity by cotreatment with AcSDKP, a tetrapeptide inhibitor of hematopoietic stem cell proliferation Am NY Acad Sci 1991 628 126 139
Watanabe T Brown GS Kelsey LS Yan Y Jackson JD Ewel C Kessinger A Talmage JE In vivo protective effects of tetrapeptide AcSDKP with or without granulocyte colony-stimulationg factor on murine progenitor cell following sublethal irradiation Exp Hematol 1996 24 713 721 8635527
Lauret E Dumentil D Miyanomae T Stainteny F Further studies on the biological activities of the CFU-S inhibitory tetrapeptide AcSDKP. II. Unresponsiveness of isolated adult rat hepatocytes, 3T3, FDC-P2 and K562 cell lines to AcSDKP. Possible involvement of intermediary cell(s) in the mecanism of AcSDKP action Exp Hematol 1989 17 1081 1086 2583252
Bonnet D Cesaire R Lemoine F Aoudjhane M Najman A Guigon M The tetrapeptide AcSDKP, an inhibitor of the cell-cycle status for normal human hematopoietic progenitors, has no effect on leukemic cells Exp Hematol 1992 20 251 255 1544396
Smeets ME Raymakers RA Vierwinden G Pennings AH Wessels H de Wite T Triggering noncycling hematopoietic progenitors and leukemic blasts to proliferate increases anthracycline retention and toxicity by downregulating multidrug resistance Blood 1999 94 2414 2423 10498614
Boekhorst PA Lowenberg B Van Kapel J Nooter K Sonnevld P Multidrug resistant cells with high proliferative capacity determine response to therapy in acute myeloid leukaemia Leukaemia 1995 9 1025 1031
|
16171532
|
PMC1261258
|
CC BY
|
2021-01-04 16:03:04
|
no
|
BMC Cancer. 2005 Sep 20; 5:120
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-120
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1221617659010.1186/1471-2407-5-122Technical AdvanceA method to estimate cell cycle time and growth fraction using bromodeoxyuridine-flow cytometry data from a single sample Eidukevicius Rimantas [email protected] Dainius [email protected] Ramunas [email protected] Nijole [email protected] Vita [email protected] Mykolas [email protected] Willem Den [email protected] Faculty of Mathematics and Informatics, Vilnius University, Naugarduko 24, 03225 Vilnius, Lithuania2 Institute of Oncology, Vilnius University, Santariškių 1, 08660 Vilnius, Lithuania3 Institute of Immunology, Vilnius University, Moletų pl. 29, 08409 Vilnius, Lithuania4 Department of Pathobiology, Utrecht University, P.O. Box 80158, 3508 TD Utrecht, The Netherlands2005 22 9 2005 5 122 122 28 12 2004 22 9 2005 Copyright © 2005 Eidukevicius 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
Presently available flow cytometric methods of bromodeoxyuridine (BrdUrd) labelling do not provide information on the cell cycle time (TC) and the growth fraction (GF). In this paper, we describe a novel and simple method to estimate TC and GF from flow cytometric analysis of a single tumour sample after BrdUrd labelling.
Methods
The proposed method is based on two assumptions: (1) the number of labelled cells traversing the cell cycle per unit time is constant and (2) the total number of labelled cells is constant throughout the cycle, provided that cells produced after division are excluded. The total numbers of labelled divided G1 cells, labelled divided S cells, labelled undivided S cells, and labelled undivided G2 cells were obtained for DNA histograms of BrdUrd-positive cells in a collected sample. These cell numbers were used to write equations to determine the durations of cell cycle phases, TC and GF. To illustrate the application of the proposed formulae, cell cycle kinetic parameters were analysed in solid SL2 tumours growing in DBA/2 mice and in human T-leukaemia Jurkat cells in culture.
Results
The suitability of the proposed method for estimating durations of the cell cycle phases, TC and GF was demonstrated. TC in SL2 tumours was found to be relatively constant at 4 and 10 days after tumour implantation (20.3 ± 1.1 h and 21.6 ± 0.9 h, respectively). GF in tumours at day 10 was lower than GF at day 4 (54.2 ± 7.7% vs. 79.2 ± 5.9%, p = 0.0003). Approximate values of TC and GF of cultured Jurkat cells were 23.9 h and 79.3%, respectively.
Conclusion
The proposed method is relatively simple and permits estimation of the cell cycle parameters, including TC and GF, from a single tumour sample after labelling with BrdUrd. We have shown that this method may be useful in preclinical studies, allowing estimation of changes in GF during growth of murine tumours. Experiments with human Jurkat cells suggest that the proposed method might also prove suitable for measurement of cell kinetics in human tumours. Development of suitable software enabling more objective interpretation of the DNA profile in this method would be desirable.
==== Body
Background
Uncontrolled proliferation is regarded as one of the most important traits of malignancy. Therefore, a considerable effort has been directed to the evaluation of the parameters of cell kinetics in tumours. A significant improvement in cell kinetic studies was achieved by the introduction of the technique of flow cytometric analysis of cells labelled with bromodeoxyuridine (BrdUrd) and the relative movement (RM) method [1]. The RM method allows estimation of the duration of S phase (TS) and the labelling index (LI) from a single sample. From these parameters, the potential doubling time (Tpot) can be calculated. Tpot is defined as the time to double the number of proliferating cells in the absence of cell loss [1]. This derived cell kinetic parameter has been postulated to be a predictor of a tumour's proliferative capability and has been widely studied in clinic in an attempt to find its prognostic significance and the potential value in predicting treatment outcome [2]. However, more recent evidence has not confirmed the existence of association between Tpot and disease outcome measures, whereas information on clinical significance of LI and TS remains ambiguous [3-5]. Therefore, new methods of evaluating cell kinetics in tumours are required.
Cell production rate is determined by the cell cycle time (TC), i.e. the time interval between cell divisions, and the growth fraction (GF), i.e. the proportion of cells engaged in the cell cycle. Thus, TC and GF can be regarded as the most important parameters in evaluating cell production in tumours. However, neither the original RM method [1] nor its more recent modification [6] provide enough data to calculate the duration of all cell cycle phases, consequently the duration of the whole cycle, and GF of a tumour.
TC and GF can be estimated by calculating the percent BrdUrd labelling in tumour samples vs. time after injection of BrdUrd. However, the disadvantage of this method is that subsequent tumour samples have to be taken at intervals during several days after the administration of BrdUrd [7]. This method does not allow a measurement of cell kinetics in an individual tumour at a particular time point and is useless in a clinical setting.
Currently, proliferation-related antigen Ki-67 is widely used to measure GF in tumours and normal tissues. The Ki-67 antigen is considered to be present in the nuclei of cells in the G1, S and G2 phases of the cell cycle as well as in mitosis, but not in quiescent cells (cells in G0) [8]. Although the relationship between Ki-67 protein expression and cell proliferation is established, there is some evidence that expression of Ki-67 antigen may be minimal in late G1 and early S phase cells [9]. The accessibility of Ki-67 epitopes may vary during the cell cycle [10]. In addition, not all cells containing Ki-67 antigen may be actively proliferating [11]. Thus, detection of proliferative markers (such as Ki-67) may not correspond in every case to the theoretically defined term of GF. Obviously, determination of proliferative markers does not provide information about TC.
In this paper, we describe a novel and simple method to estimate TC and GF from flow cytometric analysis of a single tumour sample after BrdUrd labelling.
Methods
Mice and tumours
Female DBA/2 mice at the age of 8–12 weeks were obtained from the local breeding facility at the Institute of Immunology, Vilnius, Lithuania. SL2, a spontaneously arisen DBA/2 – derived lymphoma was maintained by weekly intraperitoneal passage in DBA/2 mice. Solid SL2 tumours for cell proliferation analysis were induced by subcutaneous injection of 107 SL2 cells on the chest of mice and analysed after 4 and 10 days.
Experimental research on animals has been conducted according to recommendations of the Lithuanian Ethics Committee for the Laboratory Animal Use.
Cell culture
Human acute T-cell leukaemia Jurkat cells were cultured in RPMI 1640 medium supplemented with 10% foetal bovine serum. The cultures were incubated at 37°C in a humidified atmosphere with 5% CO2.
BrdUrd labelling
The mice were injected i.p. with 1 ml of 2 mg/ml BrdUrd (5-bromo-2'-deoxyuridine, Sigma, St Louis, MO, USA) solution in physiological saline. The half-life of BrdUrd in rodent blood is only about 15 min. Therefore, a single i.p. injection acts as a pulse label [12]. At 10 h after BrdUrd injection the tumours were dissected and cut into slices of about 1 mm thickness. To avoid sampling errors, several slices were cut both from central and peripheral parts of a tumour. Tumour slices were fixed in 70% ethanol, and stored at +4°C for one or two days until staining.
For in vitro labelling of Jurkat cells, BrdUrd was dissolved in phosphate-buffered saline (PBS) and added to culture medium at a final concentration of 10 μM. After incubation for 30 min at 37°C, the cells were rinsed twice with RPMI 1640 medium and the cultures further incubated at 37°C. Samples for analysis were taken at 8, 10, 12 and 16 h after the BrdUrd pulse, the cells were washed with PBS and fixed in 70% ice-cold ethanol.
Flow cytometric staining and measurement
Slices of solid SL2 tumours were cut in fragments of about 1 mm3 and incubated in 0.4 mg/ml pepsin solution in 0.1 N HCl at 37°C with continuous agitation to produce a nuclear suspension. Isolation of nuclei was revealed by microscopic observation after 30 to 60 min of incubation. The nuclear suspension was filtered through a 48 μm nylon mesh. Nuclear suspensions of SL2 cells and suspensions of fixed Jurkat cells were washed and incubated in 2 N HCl for 30 min at room temperature for DNA denaturation. The acid solution was then neutralised with 0.1 M Na2B4O7 (pH 8.5) and the nuclear or cell suspensions were washed twice in PBS. Fifty μl of PBS containing 0.5% Tween 20 and 20 μl of FITC-conjugated mouse anti-BrdUrd monoclonal antibody (Becton Dickinson, Heidelberg, Germany) was added to the pellet containing 106 nuclei and incubated for 30 min at room temperature in the dark. After washing twice in PBS, 1 ml of a solution containing 5 mg/l propidium iodide (Sigma, St Louis, MO, USA) in PBS was added to the pellet and incubated for 30 min at 37°C in the dark.
The cellular DNA content and the amount of incorporated BrdUrd were simultaneously measured using a FACSort flow cytometer (Becton Dickinson, Heidelberg, Germany). Green fluorescent light emission (FITC = BrdUrd incorporation) was collected in the FL1 detector and red fluorescence (propidium iodide = DNA content) was collected in FL2 (for SL2 cells) or FL3 (for Jurkat cells) detector. Data from 15,000 nuclei per sample were acquired as dot plots of BrdUrd labelling vs. DNA content using LYSYS II software (Becton Dickinson, Heidelberg, Germany). Doublets were excluded by using the doublet discrimination mode. WinMDI version 2.8 software was used to analyse the acquired data.
Mathematical formulae for estimation of duration of cell cycle phases and GF in BrdUrd-labelled tumours
We propose that the total number of cells Sl labelled during the BrdUrd pulse can be estimated after any time interval shorter than the duration of the cell cycle as
where Slu – labelled undivided S cells, G2lu – labelled undivided G2 cells, G1ld – labelled divided G1 cells, and Sld – labelled divided S cells (1/2 is introduced to exclude cells produced after division). Theoretical DNA distributions of BrdUrd-labelled cells at different post-labelling times are graphically represented in Figure 1. The equation (1) is based on the assumption that labelled cell mortality or exit from the cell cycle during the period of measurement is negligible. If this is the case, the total number of cells labelled during the BrdUrd pulse remains constant throughout the cycle, provided that cells produced after division are excluded. Mitotic cells are not included in the analysis, since they lack the nuclear membrane and the cell membrane is disrupted by the nuclear isolation medium used for preparation for flow cytometric analysis [13]. Taking into account that the mitotic time is relatively short and the number of cells in mitosis can be regarded as negligible, the omission of mitotic cells may not have a considerable impact on estimation of the total number of BrdUrd-labelled cells.
Dot plots of BrdUrd labelling vs. DNA content were used to gate BrdUrd-positive cells (Figure 2A). The gate settings were adjusted on the control profiles, so as reproducibly to distinguish BrdUrd-positive fluorescence. G1ld, Sld, Slu and G2lu compartments in each DNA histogram of BrdUrd-positive cells were determined manually and numbers of cells in each of these compartments were obtained using statistics option of WinMDI 2.8 software (Figure 2B). These cell numbers were used in the equations of the proposed method. Instead of cell numbers, cell percentages over the total number of cells measured per sample can also be used.
The proposed method requires that considerable proportions of cells be found in Slu and Sld compartments. Slu and Sld compartments can be distinguished in the DNA histogram of BrdUrd-positive cells, provided that the time interval t after the BrdUrd pulse is shorter than the duration of the S phase, but sufficient for the labelled cells to re-enter a new S phase, i.e. if + < t < TS (, and TS are durations of G2, G1 and S phases of the cell cycle, respectively). Sld compartment was separated from Slu compartment by the channel corresponding to the lowest number of events (Figure 2C).
The number of labelled cells traversing S phase per unit time can be estimated as
The number of labelled cells traversing G2 phase, mitosis, G1 phase and re-entering S phase per unit time can be estimated as
The labelled cells begin to enter G1 phase after time and the number of cells traversing G1 phase and re-entering S phase per unit time can be estimated as
The labelled cells begin to re-enter S phase after time + and the number of cells re-entering S phase per unit time can be estimated as
Estimation of durations of cell cycle phases in the proposed method is based on the assumption that the number of labelled cells traversing the cell cycle per unit time is constant. One of the models illustrating this assumption is the rectangular age distribution of proliferating cells described by Steel [14]. In this case, ν1 = ν2 = ν3 = ν4 and
The equation (1) and sequentially the equalities of the equation (6) can be used to determine TS, and :
GF can be estimated using the classical formula reported by Steel [14]:
where LI is the labelling index of the whole population and LIP is the labelling index of proliferating cells. If the number of labelled cells traversing the cell cycle per unit time is constant, we have
where TC is the duration of the cell cycle (TC = + TS + ). The LI can be calculated by the formula proposed by Johansson et al. [13]:
where nt is the total number of BrdUrd-positive and BrdUrd-negative cells measured per sample, G2 is the total number of G2 phase cells and G2ul is the number of unlabelled G2 cells. In the proposed method, t < TS and G2ul is negligible. Thus, the total number of G2 phase cells is approximately equal to G2lu.
Formulae for estimation of durations of cell cycle phases based on a decreasing exponential cell age distribution hypothesis are presented in the Additional file 1.
RM approach
In addition to using the proposed formulae, TS values were estimated also using the classical RM method [1], here referred to as RM (linear), and the more recently proposed RM method, based on a cubic fit to the RM [6], here referred to as RM (cubic). The latter method also allows calculation of . Using the RM (linear) method, the relative mean DNA content RM(t) of the moving cohort of BrdUrd-labelled cells in relation to that of G1 and G2 cells, was calculated using the formula
where and are the mean DNA content of cells in G1 and G2 phases, respectively, FLU is the mean DNA content of labelled undivided cells (Slu and G2lu) and t is post-labelling time. TS was calculated from the equation
Using the RM (cubic) method, TS was calculated from the equation
In this equation, ν is defined as
where flu(t) and fld(t) are fractions of labelled undivided and labelled divided cells, respectively, at post-labelling time t. In our case, flu(t) is the fraction of Slu and G2lu cells, and fld(t) – the fraction of G1ld and Sld cells, respectively, in the total number of cells nt measured per sample.
An assumed exponential growth rate c was estimated as
where
was calculated from the equation
= t + ln(1 - fld(t)/2)/c (19)
Statistics
Pearson's correlation coefficient was used to determine the correlation between the TS or values obtained using the proposed method and the RM approach. Differences between the GF values at day 4 and day 10 were analysed by the Wilcoxon rank sums test.
Results
To illustrate application of the proposed formulae, duration of cell cycle phases and GF were analysed in solid SL2 tumours at 4 and 10 days after implantation. , TS, and TC values estimated using the proposed method and TS and values in the same tumours estimated using the RM approach are given in Table 1. A significantly positive correlation between the TS values obtained using the proposed method and the RM approach was observed (Figures 3A and 3B). In addition, a significantly positive correlation was observed between the values obtained with the proposed method and with the RM (cubic) method (Figure 3C).
GF in solid SL2 tumours at 4 days and 10 days after implantation, estimated using the proposed method, are shown in Figure 4. GF in solid SL2 tumours at 10 days was significantly lower than GF at 4 days after implantation.
Representative dot plots of BrdUrd content vs. DNA content of Jurkat cells at different post-labelling times are shown in Figure 5. Samples of Jurkat cells taken at 8 h after the BrdUrd pulse were not suitable for analysis with the proposed method, because there were no labelled cells in the Sld compartment, i.e. the labelled cells had not yet entered the new S phase. Interval of 12 h appeared to be the most appropriate for cell kinetic analysis of Jurkat cells with the proposed method, because there were considerable proportions of cells both in the Sld compartment and in the Slu compartment. At 10 h, the number of labelled cells in the Sld compartment was still rather low, whereas at 16 h, only the small number of cells remained in the Slu compartment. Estimates of cell cycle phases and GF in Jurkat cell line obtained at 10 h, 12 h and 16 h after BrdUrd labelling are presented in Table 2.
Estimates of durations of cell cycle phases based on rectangular age distribution hypothesis and corresponding estimates obtained by calculations based on a decreasing exponential cell age distribution hypothesis are compared in supplementary Table 1 [see Additional file 1]. Numerical examples illustrating that the estimates of durations of cell cycle phases are not very sensitive to erroneous setting (+/-5%) of regions of interest in the DNA-BrdUrd plots are presented in supplementary Table 2 [see Additional file 2].
Discussion
The proposed method is based on two assumptions. The first assumption, i.e., the labelled cell mortality or exit from the cell cycle during the period of measurement is negligible, can be regarded as reasonably simple and generally acceptable. On the other hand, it is quite difficult to obtain reliable estimates of cell death rates in bivariate BrdUrd/DNA cytometry. A method for estimating the rate of cell death in G2/M phase of exponentially growing cell populations has only recently been published [15].
The second assumption is that the number of labelled cells traversing the cell cycle per unit time is constant. Although rather simplistic, this assumption does not appear to induce a great extent of error, as illustrated by corresponding estimates obtained by calculations based on a decreasing exponential cell age distribution hypothesis.
Our results illustrate, that the proposed method can be successfully used to obtain kinetic estimates of the cell cycle and GF in a murine tumour in vivo and in human tumour cells in vitro. The finding of a decrease in GF with the "age" of murine tumours shows that information retrieved with the proposed method may be biologically relevant. A good correlation was found between the TS values of SL2 tumours obtained using our proposed method and the generally accepted RM (linear) or RM (cubic) methods. The values of SL2 tumours obtained with the proposed method correlated with the values obtained with the RM (cubic) method.
The TS values obtained with the proposed method had a narrower spread (11.6 to 14.7 h) compared to the TS values in the same tumours obtained with the RM (linear) and the RM (cubic) method (12.6 to 17.6 h and 13.5 to 18.9 h, respectively). Thus, the TS values of SL2 tumours obtained with the proposed method are more compatible with the notion that TS may be a species-specific constant [16]. The TS values obtained using the proposed method were slightly lower than those obtained using the RM methods. This difference can be explained by the fact that in the RM approach, the moving cohort of BrdUrd-labelled cells includes some G2 cells, which actually are not moving in the histogram. Thus, the RM of the cohort of BrdUrd-labelled cells is slower than actual traversing of cells through the S phase. Therefore, we think that our proposed formulae provide more accurate estimation of TS.
Similarly to other methods of analysis of cell kinetics using DNA profiles, the proposed method is subject to errors of profile interpretation. Using a fitting programme to extract more accurate data would be preferable. To our knowledge, however, currently available fitting programmes (including ModFit LT™, version 3.1) do not allow the separation of Slu and Sld compartments, the essential procedure in the proposed method. Thus, none of the currently available software can by applied for analysis of the DNA profile in a way required for the proposed method. On the other hand, the identification of a single channel with the minimum event count to separate Slu and Sld compartments was feasible in all DNA histograms analysed and, therefore, can be regarded as quite robust procedure.
Both the RM approach and the proposed method entail the problem of separation of diploid tumour cells from normal intratumoural cells. In the proposed method, admixture of normal cells does not affect the determination of TC, because this parameter is estimated using only BrdUrd-labelled cells, and proliferation of normal intratumoural cells is usually negligible. However, the proposed method may underestimate GF in diploid tumours, if procedures to separate normal cells from tumour cells are not used. The possible solutions to tackle this problem have been summarized by Antognoni P et al. [2] and are mainly based on using cytokeratin markers to separate epithelial tumour cells from intratumoural inflammatory cells and fibroblasts. In addition, it is recommended to present cell kinetic data from diploid and aneuploid tumours separately [17]. SL2 was proved to be an aneuploid (nearly tetraploid) tumour with the DNA index of 1.8 (data not shown). Thus, admixture of normal cells in our analysis can be ruled out.
In experiments with SL2 tumour, PI fluorescence was collected in the FL2 detector and compensation for the spectral overlap with FITC fluorescence was used. Generally, however, it is recommended to use FL3 (as done in experiments with Jurkat cells) rather than FL2 to overcome any crossover with the FITC in FL1 [12]. Running the samples at a higher FL3 voltage such that more of the scale is used would facilitate detection of hypodiploid DNA peaks.
In the RM approach, BrdUrd is injected usually 4 to 8 h prior to biopsy or surgical removal of a tumour [18]. The proposed method requires that considerable proportions of cells be found in each of the four compartments to be analysed (Slu, G2lu, G1ld, and Sld). This condition is fulfilled only if the time interval from the BrdUrd pulse to the time of measurement is longer than duration of G2 and G1 phases but still shorter than duration of S phase. Obviously, this cannot be assured in advance. Thus, the proposed method has to be tailored for a particular tumour type and the optimal time of measurement has to be found by trial and error. As an example of this trial and error procedure, the finding of an optimal time for measurement of cell kinetics in Jurkat cells is described in this paper.
Measurements with Jurkat cells also show, that the proposed method may be sensitive to intercell variability in phase transit times. If the time of measurement is too short and only the fastest cells enter Sld compartment (e.g. 10 h in case of Jurkat cells), the duration of G1 phase may be underestimated. On the other hand, if the time of measurement is too long and only the slowest cells remain in the Slu compartment (e.g. 16 h in case of Jurkat cells), the duration of S and other cell cycle phases may be overestimated. Thus, the time for measurement of cell kinetics with the proposed method has to be selected carefully. To minimize the effect of intercell variability in phase transit times, it is advisable to use the same time interval from the BrdUrd pulse to the time of measurement in all samples to be compared.
Measurements with cultured Jurkat cells presented in this paper suggest that the proposed method might be also suitable for studies on human tumours. It seems that the suitability of the proposed method for clinical studies could only be compromised if the G1 phase in human tumours in vivo was very long. Obviously, this can only be determined by testing the proposed method in clinical studies after infusion of BrdUrd to patients.
Conclusion
The proposed method is relatively simple and permits estimation of cell cycle parameters, including TC and GF, from a single tumour sample after labelling with BrdUrd. We have shown that this method may be useful in preclinical studies, allowing estimation of changes in GF during growth of murine tumours. Experiments with human Jurkat cells suggest that the proposed method might also prove suitable for measurement of cell kinetics in human tumours. Development of suitable software enabling more objective interpretation of the DNA profile in this method would be desirable.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RE conceived the mathematical formulae and contributed to elaboration of the manuscript. DC coordinated the study and wrote the manuscript. RJ analysed the data and participated in writing of the manuscript. NK performed flow cytometric analysis. VP contributed to elaboration of the manuscript. MM coordinated animal experiments, ethical guidelines. WDO critically appraised 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
Calculations of durations of cell cycle phases based on hypothesis of decreasing exponential cell age distribution. The Additional file 1 includes supplementary Table 1 – Durations of cell cycle phases in murine solid SL2 tumours calculated using rectangular or decreasing exponential cell age distribution
Click here for file
Additional file 2
Supplementary Table 2 – Consequences of erroneous measure of percentages of cells in regions of interest (ROI) in the DNA-BrdUrd plot on the estimates of cell cycle phase durations and GF
Click here for file
Figures and Tables
Figure 1 Theoretical DNA distributions of BrdUrd-labelled cells vs. cell number at different post-labelling times. A – immediately after BrdUrd pulse, i.e. at t = 0, B – at 0 < t <, C – at < t < TS and < t < + , and D – at + < t < TS. , TS and are durations of G1, S and G2 phases of the cell cycle, respectively. Horizontal lines dividing G1ld and Sld bars indicate that half of the labelled divided cells are excluded from calculations.
Figure 2 Example of estimation of SL2 tumour cell numbers within different compartments of the cell cycle 10 h after in vivo labelling with BrdUrd. A. Flow cytometric dual parameter dot plot of BrdUrd content (log FITC fluorescence) vs. DNA content (propidium iodide fluorescence) with a gate set for BrdUrd positive cells. B. DNA histogram analysis of BrdUrd positive cells; M1 indicates labelled divided G1 cells (G1ld); M2 indicates labelled divided S cells (Sld); M3 indicates labelled undivided S cells (Slu); M4 indicates labelled undivided G2 cells (G2lu). C. The same DNA histogram as in B, shown at higher magnification; Sld compartment is separated from Slu compartment by the channel corresponding to the lowest number of events (indicated by dotted vertical line). Total number of cells within each of these markers is obtained using statistics option of WinMDI 2.8 software.
Figure 3 Correlation between cell cycle phase duration values obtained with the proposed method and with the RM approach for mouse solid SL2 tumours. A. Correlation between TS values obtained with the proposed method and with the RM (linear) method. B. Correlation between TS values obtained with the proposed method and with the RM (cubic) method. C. Correlation between values obtained with the proposed method and with the RM (cubic) method.
Figure 4 GF in solid SL2 tumours at 4 days and at 10 days after tumour implantation (mean ± SD). The difference between GF at day 4 and day 10 is statistically significant (p = 0.0003).
Figure 5 Representative dot plots of BrdUrd content (log FITC fluorescence) vs. DNA content (propidium iodide fluorescence) of Jurkat cells at different post-labelling times. Jurkat tumour cells were labelled with BrdUrd for 30 min and then fixed after 8 h (A), 10 h (B), 12 h (C) and 16 h (D). A. Labelled cells have not yet entered the Sld compartment. B. Labelled cells start entering the Sld compartment. C. Considerable proportions of labelled cells are present both in the Sld and in the Slu compartment. D. Labelled cells are leaving the Slu compartment. The optimal time for measurement of cell kinetics in Jurkat cells with the proposed method appears to be 12 h.
Table 1 Estimation of durations of cell cycle phases in murine SL2 tumours using the proposed method and the RM methods *
Tumour "age" Method TG1 TS TG2 TC
4 days (n = 10) Proposed 5.1 ± 0.4 12.6 ± 1.0 2.6 ± 0.5 20.3 ± 1.1
RM (linear) N.A. 14.1 ± 1.5 N.A. N.A.
RM (cubic) N.A. 14.8 ± 1.6 0.7 ± 0.4 N.A.
10 days (n = 9) Proposed 5.4 ± 0.5 12.9 ± 0.6 3.3 ± 0.5 21.6 ± 0.9
RM (linear) N.A. 14.4 ± 1.1 N.A. N.A.
RM (cubic) N.A. 15.8 ± 3.3 1.4 ± 0.6 N.A.
* Values are given in h (mean ± SD)
N.A. = not available
Table 2 Estimation of durations of cell cycle phases and GF in Jurkat cell line using the proposed method at different times of measurement *
Hours after BrdUrd labelling TG1 TS TG2 TC GF (%)
10 5.2 11.1 4.7 21.0 78.9
12 9.3 12.2 2.4 23.9 79.3
16 14.2 16.3 1.0 31.5 53.3
* Values obtained from individual measurements, given in h
==== Refs
Begg AC McNally NJ Shrieve DC Karcher H A method to measure the duration of DNA synthesis and the potential doubling time from a single sample Cytometry 1985 6 620 626 4064842 10.1002/cyto.990060618
Antognoni P Terry NHA Richetti A Luraghi R Tordiglione M Danova M The predictive role of flow cytometry-derived tumor potential doubling time (Tpot) in radiotherapy: open questions and future perspectives Int J Oncol 1998 12 245 256 9458346
Begg AC Haustermans K Hart AA Dische S Saunders M Zackrisson B Gustaffson H Coucke P Paschoud N Hoyer M Overgaard J Antognoni P Richetti A Bourhis J Bartelink H Horiot JC Corvo R Giaretti W Awwad H Shouman T Jouffroy T Maciorowski Z Dobrowsky W Struikmans H Rutgers D Wilson GD The value of pretreatment cell kinetic parameters as predictors for radiotherapy outcome in head and neck cancer: a multicenter analysis Radiother Oncol 1999 50 13 23 10225552 10.1016/S0167-8140(98)00147-9
Haustermans K Fowler JF Is there a future for cell kinetic measurements using IdUrd or BdUrd? Int J Radiat Oncol Biol Phys 2001 49 505 511 11173147 10.1016/S0360-3016(00)01506-6
Dobrowsky W Dobrowsky E Wilson GD In vivo cell kinetic measurements in a randomized trial of continuous hyperfractionated accelerated radiotherapy with or without mitomycin C in head-and-neck cancer Int J Radiat Oncol Biol Phys 2003 55 576 582 12573744 10.1016/S0360-3016(02)04121-4
White RA Meistrich ML Pollack A Terry NH Simultaneous estimation of TG2+M, TS, and Tpot using single sample dynamic tumor data from bivariate DNA-thymidine analogue cytometry Cytometry 2000 41 1 8 10942890 10.1002/1097-0320(20000901)41:1<1::AID-CYTO1>3.0.CO;2-3
Zätterström UK Johansson M Källén A Baldetorp B Oredsson S Wennerberg J Killander D Comparison of BrdUrd and [3H]TdR incorporation to estimate cell proliferation, cell loss, and potential doubling time in tumor xenografts Cytometry 1992 13 872 879 1459003 10.1002/cyto.990130810
Scholzen T Gerdes J The Ki-67 protein: from the known and the unknown J Cell Physiol 2000 182 311 322 10653597 10.1002/(SICI)1097-4652(200003)182:3<311::AID-JCP1>3.0.CO;2-9
Bruno S Darzynkiewicz Z Cell cycle dependent expression and stability of the nuclear protein detected by Ki-67 antibody in HL-60 cells Cell Prolif 1992 25 31 40 1540682
Barbareschi M Girlando S Mauri FM Forti S Eccher C Mauri FA Togni R Dalla Palma P Doglioni C Quantitative growth fraction evaluation with MIB1 and Ki67 antibodies in breast carcinomas Am J Clin Pathol 1994 102 171 175 8042584
Van Oijen MG Medema RH Slootweg PJ Rijksen G Positivity of the proliferation marker Ki-67 in noncycling cells Am J Clin Pathol 1998 110 24 31 9661919
Wilson GD Ormerod MG Analysis of DNA – measurement of cell kinetics by the bromodeoxyuridine/anti-bromodeoxyuridine method Flow Cytometry: A Practical Approach 2000 3 New York: Oxford University Press 159 177
Johansson MC Johansson R Baldetorp B Oredsson SM Comparison of different labelling index formulae used on bromodeoxyuridine-flow cytometry data Cytometry 1998 32 233 240 9667513 10.1002/(SICI)1097-0320(19980701)32:3<233::AID-CYTO10>3.0.CO;2-O
Steel GG Growth kinetics of tumours 1977 Oxford: Oxford University Press
White RA Asmuth DM Lu Y Wang N Li XD Reece L Pollard RB Nokta M Leary JF Terry NH Estimating cell death in G(2)M using bivariate BrdUrd/DNA flow cytometry Cytometry 2005 66 32 40 15915505 10.1002/cyto.a.20147
Rew DA Wilson GD Cell production rates in human tissues and tumours and their significance. Part II: clinical data Eur J Surg Oncol 2000 26 405 417 10873364 10.1053/ejso.1999.0907
Terry NH Peters LJ The predictive value of tumor-cell kinetic parameters in radiotherapy: considerations regarding data production and analysis J Clin Oncol 1995 13 1833 1836 7636524
Baisch H Otto U Hatje U Fack H Heterogeneous cell kinetics in tumors analyzed with a simulation model for bromodeoxyuridine single and multiple labeling Cytometry 1995 21 52 61 8529471 10.1002/cyto.990210111
|
16176590
|
PMC1261259
|
CC BY
|
2021-01-04 16:03:05
|
no
|
BMC Cancer. 2005 Sep 22; 5:122
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-122
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1231616228010.1186/1471-2164-6-123DatabaseXenDB: Full length cDNA prediction and cross species mapping in Xenopus laevis Sczyrba Alexander [email protected] Michael [email protected] Ali H [email protected] Robert [email protected] Curtis R [email protected] FSU College of Medicine, Department of Biomedical Sciences, 1269 W. Call Street, Tallahassee, FL 32306, USA2 AG Praktische Informatik, Technische Fakultät, Universität Bielefeld, D-33594 Bielefeld, Germany3 The Rockefeller University, Laboratory of Molecular Vertebrate Embryology, 1230 York Avenue, New York, NY 10021, USA2005 14 9 2005 6 123 123 5 5 2005 14 9 2005 Copyright © 2005 Sczyrba et al; licensee BioMed Central Ltd.2005Sczyrba 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
Research using the model system Xenopus laevis has provided critical insights into the mechanisms of early vertebrate development and cell biology. Large scale sequencing efforts have provided an increasingly important resource for researchers. To provide full advantage of the available sequence, we have analyzed 350,468 Xenopus laevis Expressed Sequence Tags (ESTs) both to identify full length protein encoding sequences and to develop a unique database system to support comparative approaches between X. laevis and other model systems.
Description
Using a suffix array based clustering approach, we have identified 25,971 clusters and 40,877 singleton sequences. Generation of a consensus sequence for each cluster resulted in 31,353 tentative contig and 4,801 singleton sequences. Using both BLASTX and FASTY comparison to five model organisms and the NR protein database, more than 15,000 sequences are predicted to encode full length proteins and these have been matched to publicly available IMAGE clones when available. Each sequence has been compared to the KOG database and ~67% of the sequences have been assigned a putative functional category. Based on sequence homology to mouse and human, putative GO annotations have been determined.
Conclusion
The results of the analysis have been stored in a publicly available database XenDB . A unique capability of the database is the ability to batch upload cross species queries to identify potential Xenopus homologues and their associated full length clones. Examples are provided including mapping of microarray results and application of 'in silico' analysis. The ability to quickly translate the results of various species into 'Xenopus-centric' information should greatly enhance comparative embryological approaches.
Supplementary material can be found at .
==== Body
Background
Following the publication of the first automated cDNA sequencing study in 1991 demonstrating the utility of large scale random clone cDNA sequencing approaches [1], there has been a rapid and accelerating growth of such Expressed Sequence Tags (EST). The initial study of 600 partial human sequences has grown to more than 20.0 × 106 while more than 30 organisms have more than 100,000 sequences. To make sense of the resulting sequence, a variety of bioinformatic approaches have been developed to identify protein coding sequences and domains [2-4] and generate 'unigene' sets based on agglomerative clustering methods [5,6]. Clustering EST sequences is a widely used method for analyzing the transcriptome of a genome. Especially for organisms whose genome is not (yet) sequenced, the EST data is a valuable source of information. While enormously useful, most current analysis tools result in the loss of significant biological information such as alternatively spliced transcripts and polymorphisms [7-18]. Alternative splicing in particular plays important roles during both development and in the mature organism [7-15]. Moreover, most EST based approaches appear to overestimate the number of unique sequences compared to gene predictions based on whole genome sequencing efforts [19-22].
There are different approaches for EST clustering; the most commonly used being (1) each cluster represents a distinct gene, alternative transcripts of the same gene are grouped together into the same cluster. UniGene is one approach that uses this gene-based strategy [23-27]. (2) Alternative transcripts are represented by distinct clusters. Using genome assembly tools like CAP3 [28] or Phrap [29,30] results in such a clustering, as these tools cannot (and are not designed to) handle the kinds of differences in the EST sequences. (3) STACK [6] groups ESTs based on their tissue source first, and clusters are then generated for each tissue separately. Our approach first generates gene-oriented clusters and then attempts to generate separate contigs which potentially correspond to alternative transcripts.
The underlying principle for each of these approaches is a pairwise comparison of all sequences to identify common subsequences of a given length and identity that is subsequently used to group sequences into clusters. The types of pairwise comparisons result in a runtime that is quadratic in the number of sequences to be compared. To achieve better running times, most tools try to identify promising pairs of sequences by applying word-based algorithms, which consider the frequency of common words in each pair of sequences [31]. In any case these approaches have to compare all possible pairs of sequences, resulting in a running time that grows quadratically with the number of sequences. We have implemented a pipeline for rapid processing and clustering of EST data, based on enhanced suffix arrays [32-34]. Compared to other methods it reduces the running time tremendously. While we focus on generating gene-based clusters, we also assembled each cluster separately using CAP3 to generate consensus sequences for further analyses. Liang et al. evaluated Phrap, CAP3, TA-EST and TIGR Assembler and found in their analysis that CAP3 consistently out-performed the other programs [35]. We therefore chose CAP3 for cluster assembly.
All sequence and clustering information obtained with our approach was stored in a relational database system. To allow for extensive queries, GenBank annotations were incorporated including the library source, tissue type, cell type and developmental stage. Results of all sequence analyses performed on the consensus sequences were stored in the database. This way, comparative queries could be answered to identify e.g. full length clones, sequences unique to X. laevis, or shared between Xenopus and another organism. The comparative query also allows the identification of the set of Xenopus sequences most related to a set from another organism. Thus, the XenDB database is designed to address a critical issue facing many researchers: the comparison of genomic studies in one organism and their application to studies in another model organism. This task is faced by many laboratories attempting to extract the information gained in human, mouse, fly and worm microarray and library sequencing studies which often consist of large tables of genes.
While other databases such as UniGene [36] or TIGR Gene Indices [37] also provide collections of clustered ESTs, the unique batch functionality of mapping results from other organisms to Xenopus laevis and retrieving their potential full length clones was not available before. Moreover, our implementation is specifically designed and focused on relating Xenopus sequence data to the major model organisms. Thus, one can search for the Xenopus homologue directly using the human or mouse protein.
Construction and Content
Sequence sources and cleanup
350,468 Sequences were downloaded from GenBank release 138 and stored in a relational database using the open source ORDBMS PostgreSQL. The following divisions were included: Vertebrate Sequences (VRT, 5,506 sequences), EST (344,747 sequences) and High Throughput cDNA (HTC, 215 sequences). 228,496 sequences were annotated as 5' ESTs and 116,122 as 3' ESTs. 245,415 different cDNA clones were represented in the data set, out of which 92,463 had both 5' and 3' sequences. Entries annotated as being genomic sequences were excluded from the analysis. To enhance the usability and search capabilities of the database, complete GenBank entries were incorporated. Annotations including but not limited to library source, tissue type, cell type and developmental stage were extracted directly from GenBank entries (feature: source, qualifiers: clone_lib, tissue_type, cell_type and dev_stage). Unfortunately, the sequences are not very well annotated in GenBank. 34% of the sequences do not have a tissue type assigned and 36% have no developmental stage information. Distributions of tissue types, developmental stages and clone libraries are shown in supplemental files [see additional files 2, 3 and 4 respectively].
197,888 ESTs (57.4% of the EST sequences) had information about high quality start or end of sequencing reads. This information was used to trim sequences according to high quality regions to insure best sequence quality. Vector sequence was downloaded from GenBank and VectorDB [38] and the sequence masked using the program Vmatch [39] developed by Stefan Kurtz. Vmatch is based on a novel sequence index (enhanced suffix arrays, [32-34]), allowing for the rapid identification of similarities in large sequence sets. ESTs were trimmed to eliminate vector sequence located at either the 5' or 3' end (6678 ESTs, 1.9% of total sequence set). In some cases, additional non vector sequence preceded or followed known vector sequence. If such non-vector sequence was less than 20 bases long, it was trimmed from the EST together with the vector sequence. ESTs that had vector sequences left after trimming were discarded completely. Repetitive elements were obtained from Repbase [40] and GenBank and masked using RepeatMasker [41]. In addition, if hits against ribosomal RNA and mitochondrial sequences were found in the downloaded sequence set, the corresponding sequences were removed. The availability of complete mitochondrial genomic and ribosomal sequences makes the inclusion of these sequences unnecessary while masking was performed to minimize possible clustering errors arising from these common sequences. Sequences that had less than 100 consecutive bases left after cleanup were discarded completely (21,039 sequences, 6.0%). The resulting sequence set consisted of 317,242 sequences (90.5%) with an average length of 536 bases (see Table 1).
Table 1 Summary of Xenopus EST cleanup and clustering.
Total number of ESTs and cDNAs 350,468
Number of distinct clones 245,415
Number of good sequences 317,242
Average trimmed EST length (bp) 536
Number of 3' EST sequences 116,122
Number of 5' EST sequences 228,496
Clones with 5' and 3' sequences 92,463
Number of clusters 25,971
Number of singletons 40,877
Number of CAP3 contigs 31,353
Number of CAP3 singletons 4,801
Average CAP3 contig length (bp) 1,045
Max. cluster size (no. of ESTs) 6,332
Average cluster size (no. of ESTs) 10.6
Cluster sizes: # EST
4,097 – 8,192 1
2,049 – 4,096 1
1,025 – 2,048 2
513 – 1,024 15
257 – 512 35
129 – 256 116
65 – 128 414
33 – 64 973
17 – 32 1,755
9 – 16 2,974
5 – 8 4,571
3 – 4 6,444
2 8,670
Clustering and assembly of tentative contig sequences
The cleaned X. laevis EST sequence set was grouped into gene specific clusters using Vmatch. Vmatch preprocesses the EST sequences into an index structure: an enhanced suffix array. This data structure has been shown to be as powerful as suffix trees, with the advantage of a reduced space requirement and reduced processing time. Further on, enhanced suffix arrays have been shown to be superior to other matching tools for a variety of applications [33,42,43]. For a detailed introduction of enhanced suffix arrays see Abouelhoda et al. [34]. Briefly, the index efficiently represents all substrings of the sequences and allows the solution of matching tasks, in time independent of the size of the index (unlike BLAST). Vmatch was chosen for the following reasons: (1) At first, there was no clustering tool available which could handle large data sets efficiently, and which was documented well enough to allow a detailed replication and evaluation of existing clusters. (2) Second, Vmatch identifies similarities between sequences rapidly, and it provides additional options to cluster a set of sequences based on these matches. Furthermore, the Vmatch output provides information about how the clusters were derived. Due to the efficiency of Vmatch, we were able to perform the clustering for a wide variety of parameters on the complete sequence set (see below). This allowed us to study the effect of the parameter choice on the clustering. Moreover, in the future, the efficiency will allow us to more frequently update the data set. A longer term goal of the project is to generate a data set that maintains the different alleles in this pseudotetraploid animal as separate entries. The clustering approach has been integrated into an analysis pipeline which can be applied to other organisms that often receive less attention from the bioinformatics community.
The database sequences were clustered according to the matches found in a self comparison of the index. Initially each database sequence is put into its own cluster. Then all pairs of matches are generated and each pair is evaluated to possibly form single linkage clusters. To identify matching sequences, Vmatch first computes all maximal exact matches of a given minimal length (seeds) between all sequences. These seeds are extended in both directions allowing for matches, mismatches, insertions, and deletions using the X-Drop alignment strategy as described previously. This greedy alignment strategy was developed for comparing highly similar DNA sequences that differ only by sequencing errors, or by equivalent errors from other sources [44].
In an attempt to objectively define appropriate clustering criteria, we took advantage of the speed of the Vmatch clustering approach to systematically vary the relevant parameters (overlap length, % identity, seedlength and X-drop value). It was hypothesized that the 'correct' parameters would be revealed as an abrupt change in the curve on the resulting graph. An example of such an analysis showing the effect of varying the overlap length and % identity is presented in supplemental materials [see additional file 1]. Here a number of conclusions become apparent. First, at this level of resolution (~30 independent clusterings), a distinct point indicating the 'correct' parameter does not become readily apparent. Second, the collapse of the cluster set to few clusters containing every larger numbers of individual sequences serves as a reminder that all sequences (regardless of species) can be considered part of a single cluster. Finally, as the length overlap decreased, we observed the formation of 'superclusters' containing >10,000 sequences clearly derived from multiple gene families. These problem of 'superclusters' diminished at an overlap length of ~135 (data not shown, and not apparent in additional file 1). These clusters appear to be due to the presence of undefined repetitive elements, chimeric sequences and possibly transposed elements. Studies on the nature of the clustered sequences and the effects of parameter variation are ongoing.
For the current data set, we tried to select parameters which mimic the parameters that were probably used for generating the UniGene clusters. Unfortunately, the algorithm used for constructing the UniGene clusters is not sufficiently documented to allow complete reproduction. We selected parameters designed to produce a stringent clustering of the available sequences. For the described data set, sequences were clustered when a pairwise match of at least 150 nucleotides and 98% identity was found (seedlength = 33, X-Drop = 3). The construction of the enhanced suffix array took 33 minutes on a SUN UltraSparc III (900 MHz) CPU. Clustering took another 17 minutes. This resulted in 25,971 clusters containing 276,365 sequences (87.11% of the input set) and 40,877 singletons (12.89%). The average cluster size was 10.6 (std. dev 51.8) sequences. The distribution of cluster sizes is shown in Table 1. 22,834 clusters were composed of ESTs only, 61 clusters of mRNA sequences (VRT and HTC divisions) only and 3,076 clusters of both mRNAs and ESTs. Among the singletons are 4262 sequences which contain less than 150 nt (after sequence cleanup described above) and would therefore be incapable of being joined in a cluster. Less than 25% of these sequences have a significant match against NR database and less than 2% of the sequences match full length cDNA criteria described below.
Next, a consensus sequence was generated for each cluster using CAP3 [28]. The aim of this approach was to both refine the number of clusters and to improve the overall sequence quality. This latter aim simplifies the design of oligonucleotide probes. The 25,971 clusters produced 31,353 tentative contig (TC) sequences (avg. length: 1,045 bp, std. dev: 729 bp) and 4,801 singlets (avg. length: 664 bp, std. dev: 424 bp). The longest TC was 13,130 bp (DNA-dependent protein kinase catalytic subunit, accession: [Genbank:AB016434]), while the smallest TC was 154 bases long. Here, it became obvious that CAP3 is a genome assembly program not designed to assemble EST clusters containing potential splice variants: CAP3 assembly subsequently split a fraction of the clusters into separate contigs and singletons. On average, a cluster was split into 1.2 (std. dev 3.0) TCs and 1.8 (std. dev 11.3) singlets by CAP3. As illustrated in Table 1, the average length of the sequences increased from 536 bp (average for input ESTs) to 1,045 bp (average for CAP3 contig sequences) which was lower than the average length for previously characterized Xenopus full length sequences (sequences selected as full length by XGC had an average length of 2,115 bp).
There are many genes whose transcript is significant longer than 2× the current state of the art sequencing run of ~1000 bp. This means that 5' and 3' sequences derived from a >2 kb transcript are unable to be joined without sequence from incomplete cDNA clones which provide a source of nested deletions. Sequences from both ends can be linked by annotation, and this has been done by a variety of clustering approaches including NCBI UniGene which uses a double linkage rule. Non-overlapping 5' and 3' ESTs are assigned to the same cluster if clone IDs are found that link at least two 5' ends from one cluster with at least two 3' ends from another cluster and the two clusters are merged. We have examined the effect of double linkage joining using the clone annotation. In this analysis, 17,588 clusters were stable and the total number of clusters was reduced from 25,971 to 21,249. Most of the joined clusters (3,122) were created from two clusters while three clusters were combined 456 times. While the number of clusters is decreased by this joining, our overall analysis is not affected. Potential full length clones selected as part of the P5P group (see below) are also unaffected by annotation linkage. We provide the identity of clusters 'linked by annotation' as part of the XenDB output.
Sequence analysis
We have performed a variety of sequence comparisons at the protein level including translation analysis. The sequences of cluster TCs and all singletons were subject to extensive BLASTX [45] and FASTY [46] homology searches vs. the non-redundant protein database (NR) from NCBI and the proteomes of five major model organisms using the high throughput analysis pipeline of the Genlight system [47] Proteome sets for H. sapiens, M. musculus and R. norvegicus were obtained from the International Protein Index [48,49]. The IPI provides a top-level guide to the main databases: Swiss-Prot, TrEMBL, RefSeq and Ensembl. It curates minimally redundant yet maximally complete sets of the indexed organisms. C. elegans and D. melanogaster protein sequences were retrieved from the UniProt database [50]. UniProt proteome sets are solely derived from Swiss-Prot and TrEMBL entries. Additionally, all available protein sequences for X. laevis and X. tropicalis were extracted from GenBank. additional file 5 provides an overview of the downloaded data sets. Performing separate comparisons allows a search for matching sequences based on the identity of any gene known from each species as well as query for genes which have matches in some but not all databases. We believe that this will aid in the discovery and analysis of conserved and unique genes. In addition to these databases, we have included BLASTX searches in the KOG database and have used the results to functionally classify the Xenopus sequences. All sequences resulting from the clustering and assembly processes were compared to these protein sets using BLASTX with an E-value cutoff of 1.0e-6. ESTs are often of low sequence quality, and sequencing errors can still exist in the assembled TC sequences. Therefore, all analyses against the protein databases were also done using FASTY (E-value cutoff: 1.0e-6) a version of FASTA that compares a DNA sequence to a protein sequence database, translates the DNA sequence in three forward (or reverse) frames and allows (in contrast to BLASTX) for frame shifts, maximizing the length of the resulting alignments.
Identification of chimeric sequences
A significant issue in EST clustering methods is the presence of chimeric sequence which inappropriately joins unrelated genes into a single cluster. While the number of chimeric sequences is estimated at less than 1% [51,52], their presence has disproportionate effects on the clustering outcome. To identify potential chimeric sequences, we analyzed the FASTY hits in the protein NR database and applied the following simple procedure: Matches of at least 100 bp in length were mapped back to the TC sequences to identify the regions that are covered by a match. If two matches overlap, the region will be extended accordingly. If after the mapping two clearly separated regions remain, the TC is flagged as potential chimera (see Figure 3).
Figure 3 Identification of chimeric TCs: Matches of at least 100 bp in length were mapped back to the TC sequences to identify the regions that are covered by a match (yellow boxes). If two matches overlap, the region will be extended accordingly. If after the mapping two clearly separated regions remain as shown here, the TC is flagged as potential chimera.
Examination of the identified chimeric sequences reveals three major classes. In the first, two distinct FASTY hits can be identified which do not overlap and are in opposite orientation. In the second, the second identified FASTY hit matches retroviral or transposable element related sequences. This suggests the possibility that these may reflect real transcripts in which a mobile element has been inserted into the genome. A close evaluation of such sequences may provide some insights into the evolutionary history of various populations of Xenopus. The final class of potential chimeric sequences identified contains short predicted or hypothetical proteins. This class may in fact not be chimeric at all but may reflect errors in protein coding prediction methods.
The described procedure identified 113 potential chimeric TCs (0.3% of the 33,034 sequences with matches against the protein NR database), which are flagged in the database as such. We do not eliminate these potential chimeras, as they don't significantly affect the results of the sequence analyses done later on, which are mainly based on the best hit only. In fact, the analysis underestimates the number of full length sequences, as some chimeras cover two full length protein matches. A complete identification of chimeric sequences is practically impossible without a comparison to the underlying genome sequence. And even then, polycistronic transcripts which may exist cannot be separated from chimeras perfectly [53].
Definitions
In the subsequent analyses we were interested in three kinds of information: (1) Full Length Orf containing COntigs (FLOCOs), (2) Full Length Insert containing CLones (FLICLs), and (3) Predicted 5' (P5P) sequences. The result of the clustering and CAP3 analysis generates a set of tentative contig sequences (TC). FLOCOs are defined as TC sequences that have an (almost) full length hit against a known protein. These sequences are especially useful for gene identification. Full length insert containing clones, FLICLs, were predicted. Such clones are distinguished by sequence homologies corresponding to the amino terminal part of a protein but are not restricted at the carboxy-terminus. These sequences are derived from clones which are predicted to carry a full length insert (see below), though the full length sequence has not been determined, usually because of single pass EST sequencing from the 5' end. Finally, we identified sequences that we call P5P for which sequence similarity did not extend through the amino-terminal end of the protein but whose length was sufficient to include a full length coding sequence of a similarly sized protein.
Identification of Full Length Orf containing COntigs (FLOCOs)
We were especially interested in full length hits of the TC sequences vs. known proteins. For this purpose, BLASTX and FASTY hits were categorized into four classes, representing the quality of the full length matches (see Figure 1): (1) Matches cover 100% of the sequence of a known protein. Additionally, the matched protein sequence has to begin with the conserved methionine and has to end at a conserved STOP codon. (2) Matches covering 100% of the sequence of a known protein. Additionally, the matched protein sequence has to include the initial methionine. (3) Matches capable of covering 100% of the matched protein sequence with no additional constraints. (4) Matches that cover the protein over almost its full length, allowing the match to start or end maximal ten amino acids after/before the start or end of the protein.
Figure 1 Full length clone selection (top) and TC categories (bottom). ESTs derived from different clones were clustered and assembled. The CAP3 contig was compared to protein databases using BLASTX and FASTY and hits categorized in 4 categories. Class 1 hits had to match the whole protein sequence and start with an ATG in the TC and M in the protein and the hit had to end at a STOP codon. Class 2 hits had to match the whole protein sequence, start with an ATG in the TC and M in the protein. Class 3 had to match the full protein sequence (without further restrictions), class 4 had to cover the protein over almost its full length, allowing the match to start or end maximal 10 ten amino acids after/before the start or end of the protein. Predicted 5' TCs (P5P) had to have enough sequence to fill up the missing 5' end of the protein sequence. Clone selection: Clone A and B were discarded because of missing IMAGE id. Clone 54321 does not span 5' end of protein match. Clone 21345 was selected as most 5' clone fulfilling the requirements.
Table 2 shows the number of identified FLOCOs using BLASTX. 3,942 TCs were Class 1 hits in the non-redundant protein database. As the stringency of the full length definition was relaxed, the number of TCs characterized as full length increases to 5,050 (Class 2), 7,792 (Class 3) and 12,389 (Class 4) TCs respectively. As EST sequences have many sequencing errors, and even the assembly of clusters can not correct all of these, FASTY comparisons were done for the same data set (Table 3). This way, the length of the resulting alignments could be maximized. A comparison of Table 2 and Table 3 shows the effect of frame shift corrections obtained by FASTY. The number of TCs having Class 1 hits could be increased to 5,139 while the less stringent categories increased similarly by an average of 20%. The effect of frameshift correction can clearly be seen in Figure 2. Table 4 and Table 5 show the average lengths of TCs for each of the four categories. Here, the average length of the TCs is 2,210 bp for Class 1 TCs having FASTY matches against X. laevis, corresponding very well to already known Xenopus proteins. Overall, the average length decreases with lower quality categories as expected, especially for Class 4, where the alignment can miss 20 amino acids on both ends of the matching protein. The only exceptions are Drosophila and C. elegans, where the average length increases for Class 4 sequences.
Table 2 Number of X. laevis TCs with full length BLASTX hits in the non-redundant protein database (NCBI), five model organisms, and available X. laevis and X. tropicalis proteins, determined by BLASTX. Lower quality categories include sequences from higher, more stringent categories.
Class Protein NR Human Mouse Rat Fruitfly C. elegans X. laevis X. tropicalis
1 3,942 1,760 1,765 1,455 219 140 2,918 495
2 5,050 2,067 2,076 1,736 311 233 3,104 541
3 7,792 2,647 2,919 2,592 392 283 3,898 590
4 12,389 5,587 5,841 3,078 2,071 1,856 5,024 1,033
P5P 15,870 13,942 14,179 13,113 8,425 8,117 9,227 4,334
Table 3 Number of X. laevis TCs with full length FASTY hits in the non-redundant protein database (NCBI), five model organisms, and available X. laevis and X. tropicalis proteins, determined by FASTY. Lower quality categories include sequences from higher, more stringent categories.
Class Protein NR Human Mouse Rat Fruitfly C. elegans X. laevis X. tropicalis
1 5,139 2,347 2,337 1,930 268 190 3,862 660
2 6,243 2,692 2,671 2,248 383 296 4,119 721
3 9,576 3,528 3,774 3,374 473 357 4,967 796
4 14,094 6,467 6,701 6,341 2,249 1,918 5,701 1,241
P5P 15,651 13,578 13,954 13,085 8,108 7,746 9,055 4,159
Figure 2 Comparison of a BLASTX alignment with corresponding full length FASTY alignment, as generated by the Genlight system. Blue boxes in (a) indicate open reading frames, green boxes start and red boxes stop codons, respectively. The assembled TC sequence has a frameshift at position 1150 from frame 1 to 3, generating two distinct HSPs in the BLASTX alignment (b). FASTY clearly corrects this frameshift and generates a full length alignment (c).
Table 4 Average length of X. laevis TCs for different BLASTX full length TC categories.
Class Protein NR Human Mouse Rat Fruitfly C. elegans X. laevis X. tropicalis
1 1984 1835 1805 1788 1620 1541 2171 1743
2 1831 1806 1776 1775 1541 1391 2120 1697
3 1630 1813 1775 1834 1560 1429 1981 1693
4 1393 1680 1675 496 1638 1640 1879 1660
Table 5 Average length of X. laevis TCs for different FLASTY full length TC categories.
Class Protein NR Human Mouse Rat Fruitfly C. elegans X. laevis X. tropicalis
1 2007 1888 1859 1843 1659 1575 2210 1807
2 1837 1856 1821 1819 1563 1440 2152 1774
3 1553 1790 1772 1804 1569 1441 2019 1768
4 1329 1683 1673 1664 1611 1563 1910 1703
Comparing the numbers of full length sequences in Table 2 and Table 3, the matches in human, mouse, rat and X. laevis are in general agreement (2619 full length sequences for Class 1 on average). What is striking is the deviation of both the number of full length TCs as well as the average length of TCs having matches against Drosophila and C. elegans: only 268 and 190 full length sequences with average lengths of 1659 and 1575 bp for Drosophila and C. elegans in Class 1, respectively. Only within the Class 4 category there are 2,249 and 1,918 TCs with average lengths of 1,611 bp and 1,563 bp, respectively. A possible explanation for this difference is the divergence of the vertebrate species from these invertebrate model systems.
Selection of putative Full Length Insert containing CLones (FLICLs)
Often, biologists are interested in identifying a full length clone for further study and this desire has been met by the establishment of a number of the Gene Collections (the Mammalian Gene Collection [54], the Xenopus Gene Collection [55] and the Zebrafish Gene Collection [56]). We have extended our analysis described above to select potential full length insert containing clones (FLICLs) that are available through the IMAGE consortium and provide a simple yet powerful search tool to rapidly match homologous genes of interest to their Xenopus counterparts. The Gene Collections are an NIH initiative that supports the production of cDNA libraries, clones and 5'/3' sequences to provide a set of full-length (ORF) sequences and cDNA clones of expressed genes for a variety of model systems.
Since the average length of the characterized full length vertebrate protein is 1,400 bases and the average sequence length of a TC is 1,045 bases, many sequences which are full length will not be detected by the previous approach and will contain sequence gaps of approximately 350 bases. To identify additional clones that potentially carry a full length insert, we queried the database for sequence matches which were sufficiently long to include the start methionine but which did not have sufficient homology to be detected by the previous methods Thus, a sequence with a query start position (Startq) which is greater than the subject start site (Starts) is potentially a full length open reading frame (hereafter referred to as P5P, predicted 5 prime). Clearly, the value of such a prediction decreases as the values of Startq increases and the predictive value increases with lower values of Starts. Full length clones predicted by this method are subject to 3' truncations due to mispriming in poly(A) rich regions rather than at the polyA tail. Such regions would be characterized by the presence of the amino acid lysine (codons AAA, AAG) or asparagine (codons AAU, AAC).
Best FASTY hits were extracted for TCs from all four full length categories as well as the P5P categories as described above. For TCs matching these categories, the most 5' EST contributing to the CAP3 contig sequence was selected. In addition, the selected clone had to span the amino-terminal end of the FASTY protein match. Finally, to ensure the ready availability of the clones and therefore the utility of the analysis, the selected clone had to be available through the IMAGE consortium. See Figure 1 for an illustration of 5' clone selection. The P5P criteria selected 15,651 potential full length insert containing clones out of which 10,500 are distinct IMAGE clones, which represents an additional 1,557 sequences compared to Class 4. Two examples of such predicted protein coding sequences are presented in Figure 4. We have mapped these clones to 7,782 distinct clusters. To assess the quality of the FL prediction method, we compared our set to the IMAGE clone set selected by the Xenopus Gene Collection (XGC, [55]) for full length sequencing. As of April 2004 the XGC had selected 10,482 IMAGE clones for sequencing. Our analysis selected 3,152 IMAGE clones that were identical to clones selected by the XGC. Of the remaining 7,348 clones from our set, 4,866 selected IMAGE clones were found in an identical cluster as 4,465 XGC selected clones (note that some of these clones are in the same cluster). In addition, 1,154 XGC clones did not have sequence available to be included in our analysis. The remaining 1,711 IMAGE clones selected for sequencing by XGC are not found in our predicted set while 2,482 clones were unique to our set. In an effort to examine why the 1,711 sequences selected for sequencing were not identified as full length, we compared the startq and starts values as described above. Using the P5P prediction criteria described above, we identify 107 XGC selected IMAGE clones that we predict are not full length but have an alternative clone which we predict is full length. Though final confirmation of the results requires additional sequencing, our method appears to be successful at identifying full length sequences and distinguishing non-full length sequences identified by an independent method. The FL clones are labeled in the XenDB web interface (see below), allowing a rapid identification of potential FL clones for a gene of interest.
Figure 4 Two examples of TCs derived from clones predicted to have a full length insert (P5P). The start positions in the hit suggest that the unmatched amino-terminal protein sequence is not well conserved between X. laevis and the matched organisms, here rabbit (top) and human (bottom), but the open reading frames (blue boxes) indicate that the clones the sequences were derived from do actually contain a full length insert. (Screenshots of the results were generated by the Genlight system.)
Due to the large number of sequences, we are unable to examine each sequence individually. Since the analysis depends on the overall degree of conservation among the sequences, such an approach will not be as successful on weakly conserved genes. In general, it seems likely that decreasing e-values correspond to higher quality predictions. On a global basis, the results need to be carefully considered, as an independent assessment of the distribution of conservation among the ensemble of sequences is not available.
Gene Ontology prediction and Functional Classification
The Gene Ontology (GO) project [57] is an ongoing international collaborative effort to generate consistent descriptions of gene products using a set of three controlled vocabularies or ontologies: biological processes, cellular components, and molecular functions. The GO vocabulary allows consistent searching of databases using uniform queries. The availability of such vocabularies can be critical to the interpretation of high through put approaches such as microarrays. Based on FASTY homologies with both mouse and human sequence, we have mapped GO annotations to the Xenopus sequences. Of the 30,683 TCs with matches to mouse (29,971) or human IPI sequences (29,963), 19,721 TCs have been assigned putative GO annotations. Among the 10,500 potential full length ORF containing IMAGE clones, 6,886 have been assigned GO annotations.
The non-redundant X. laevis data set was then classified based on their homology to known proteins from the KOG [58] database (BLASTX 1.0e-5 E-value cutoff, best hit selection). KOGS are euKaryotic clusters of Orthologous Groups. KOG includes proteins from 7 eukaryotic genomes: C. elegans, D. melanogaster, H. sapiens, A. thaliana, S. cerevisiae, S. pombe, E. cuniculi.17,624 sequences (67.3%) had a hit against the KOG database and could be assigned a functional category.
Identification of conserved genes not found in major model organisms
To identify additional genes within the dataset that are not found by comparison to protein sets of the major model organisms and to assess the extent of diverged or non conserved sequences, open reading frames of 600 nucleotides or longer were selected from the clustered data set for analysis. 219 sequences that did not have any hit in the previous analyses were identified (188 TCs representing 178 clusters and 31 singlets). We further restricted the number of sequences by re-running the BLASTX and FASTY analysis with E-value cutoffs of 0.01. 111 sequences (91 TCs representing 87 clusters consisting of an average of 6 ESTs per cluster and 19 singlets) without any significant similarity in protein databases could be identified and these were examined by TBLASTN against the human, mouse and 'others' EST databases (22.7 million sequences total). Signal peptides were identified by SignalP [59] as well as transmembrane domains by TMHMM [60,61]. Results are presented in Table 6. The analysis identified 46 sequences with similarity to other organisms (E<0.01) with 11 sequences matching chicken (Gallus gallus), 10 sequences matching zebrafish (Danio rerio) and 6 sequences matching the rainbow trout (Oncorhynchus mykiss). Three of the sequences matched human sequences with less significance than the cutoff used above (i.e. 1.0e-6). Among the sequences with highly significant BLAST hits were two matches to the eastern tiger salamander (Ambystoma tigrinum tigrinum) and one to the rainbow trout (Oncorhynchus mykiss). A surprising match was to barley (Hordeum vulgare, E = 9.0e-35) which was the only plant represented among these hits. The remaining 65 sequences did not have significant homology to existing public database sequences. For 7 sequences both signal peptide cleavage sites and transmembrane domains could be identified. Another 15 sequences had either a signal peptide cleavage site or a transmembrane domain. These 22 sequences are potentially novel membrane proteins.
Table 6 Xenopus Long Open Reading Frames (>= 600 nt) without homology to major model organism protein sequences. ORF sequences were compared to all available EST data using TBLASTN. The 46 sequences shown here have homologies to ESTs from other organisms (E < 0.01). For each TC, the number of ESTs in the TC and the accession, SignalP and TMHMM results, and description and E-value of the best hit is shown. Additionally (not shown here), both signal peptides and transmembrane domains could be predicted in: clSignal peptides only in: cl4857_sin8, cl11312_sin2, cl11866_ctg2, cl14117_ctg1, cl16548_ctg1, cl19372_ctg2; Transmembrane domains only in: cl3994_ctg1, vimsin144578, cl18799_ctg1, cl18978_ctg1, cl18978_ctg2, cl25690_ctg1, cl23256_ctg1.
Contig/ORF #ESTs Accession SignalP TM Description (best hit) E-value
cl9703_ctg1_1 53 CN060851 Ambystoma tigrinum tigrinum 5.00E-112
cl15798_ctg1_1 4 CN061938 Ambystoma tigrinum tigrinum 6.00E-53
cl9914_ctg1_1 11 BX864357 Oncorhynchus mykiss 4.00E-45
vmsin143901_1 1 CK600275 Rattus norvegicus 4.00E-43
cl10823_ctg1_1 3 CA471690 Danio rerio 5.00E-39
cl2563_ctg2_1 10 AV913994 Hordeum vulgare subsp. Vulgare 9.00E-35
cl1723_ctg1_1 12 BU129000 Gallus gallus 2.00E-34
vmsin213651_1 1 CD218114 Gallus gallus 7.00E-30
cl15560_ctg1_1 3 CK871392 Danio rerio 2.00E-24
cl10197_ctg1_1 3 CA975598 Danio rerio 2.00E-22
cl11603_ctg1_1 4 AJ456928 Gallus gallus 7.00E-22
cl2506_ctg1_1 7 BJ494402 Oryzias latipes 3.00E-19
vmsin144578_1 1 BU241764 Gallus gallus 1.00E-17
cl24411_ctg1_1 2 BW379961 Ciona intestinalis 2.00E-16
cl25096_ctg1_1 2 BX269216 Gallus gallus 5.00E-16
vmsin117573_1 1 BU114361 Gallus gallus 3.00E-14
vmsin141365_1 1 BI385350 Amphioxus Branchiostoma fl. 3.00E-14
vmsin275700_1 1 CN024469 Danio rerio 4.00E-14
cl5895_ctg1_1 12 BX870166 Oncorhynchus mykiss 6.00E-14
cl18998_ctg1_1 2 BW156550 Ciona intestinalis 8.00E-13
cl5042_ctg1_1 14 CN316430 Danio rerio 3.00E-12
cl9402_ctg2_1 2 AJ448952 Gallus gallus 2.00E-11
cl19097_ctg1_1 4 CN023422 yes Danio rerio 2.00E-09
cl4943_ctg1_1 4 AJ450094 Gallus gallus 5.00E-09
cl19576_ctg1_1 2 BX862425 Oncorhynchus mykiss 4.00E-08
vmsin9176_1 1 CO051215 Leucoraja erinacea 1.00E-07
cl5371_ctg1_1 9 CD295994 Strongylocentrotus purpuratus 5.00E-07
cl10375_ctg1_1 9 BU133150 yes Gallus gallus 7.00E-07
cl3127_ctg2_1 19 CF577195 Saccharum sp. 2.00E-05
vmsin5140_1 1 BM265659 Danio rerio 6.00E-05
cl15473_ctg1_1 8 CN502421 Danio rerio 7.00E-05
cl3097_ctg2_1 30 CA374396 Oncorhynchus mykiss 8.00E-05
cl9923_ctg1_1 2 DAA01768 Lytechinus variegatus 1.00E-04
cl15340_ctg1_1 14 CN180033 yes 1 Danio rerio 3.00E-04
cl11246_ctg1_1 6 BX302229 Oncorhynchus mykiss 5.00E-04
cl4857_ctg3_1 7 BF718744 Homo sapiens 6.00E-04
cl18267_ctg1_1 2 AAS58046 Babesia bovis 0.001
cl5917_ctg1_1 6 CN004343 Canis familiaris 0.002
cl9934_ctg1_1 4 CD740019 yes 2 Gallus gallus 0.002
cl3233_ctg1_3 12 CN506386 yes 5 Danio rerio 0.003
cl22258_ctg1_1 2 BM485921 Gallus gallus 0.004
cl14723_ctg1_1 3 BF037758 Homo sapiens 0.005
cl5206_ctg1_1 9 BG166355 Homo sapiens 0.005
cl5199_ctg2_1 8 BM627372 yes 1 Anopheles gambiae 0.006
vmsin18077_1 1 BX877871 yes 1 Oncorhynchus mykiss 0.007
cl5686_ctg1_1 2 BG783827 Strongylocentrotus purpuratus 0.008
Utility
User interface
The results of the analyses described above have been incorporated into an SQL database amenable to complex queries. The database can be accessed through a user friendly web based interface (XenDB). XenDB allows individual and batch queries using Xenopus accession, GI, and XenDB, UniGene and TIGR cluster IDs. In addition, the user can query the Xenopus sequence hits using any protein accession/GI number both singly and in batch mode. This allows a rapid identification of Xenopus TCs and their corresponding clones with hits to given protein sequences. The output of various queries displays the matching Xenopus cluster(s) and links to a web page as presented in Figure 5. For each cluster, links to the best hit for a number of model organisms are provided as well as links to the assembly result, consensus sequence generated by CAP3, and visual alignments of all FASTY results. GenBank accession numbers for each EST in the cluster and whether the corresponding clone has been identified as full length are provided. Additionally, for each TC the COG and KOG classification, as well as the GO terms are available.
Figure 5 Cluster view of the XenDB Web interface. Best FASTY hits to NR protein database, five model organisms and Xenopus proteins are shown on top. Gene Ontologies (GO) are based on best human and mouse IPI hits, functional categories on hits to COG and KOG databases. Below, additional information for each EST in the cluster is shown, such as accession, UniGene and TGI id, clone, cell and tissue type. Clones predicted not to be full length are colored red. Links to CAP3 assembly and TC sequence are provided.
The analysis and database system provides a very powerful tool which will enable the Xenopus community to take advantage of a number of technical and experimental advances. We have selected a couple of examples to illustrate possible types of queries. In considering the results, it is important to bear in mind that these examples can be combined to further refine the sequence set. In the first example, we sought to identify all the genes of a known type or class. In the second example, we wished to identify the set of Xenopus sequences which best matched a set of genes from another species identified using the CGAP database administered by the National Cancer Institute (NCI) [62,63]. A final example demonstrates the ability of the system to translate results identified by microarray technologies, or other related high throughput technologies, to identify likely Xenopus homologues.
Homeobox gene identification
Homeobox containing proteins are a very important group of transcriptional regulators that play key roles in developmental processes. They can be divided into a 'complex' and a 'dispersed' super class representing the homeotic genes and the large number of homeodomain containing proteins dispersed (and diverged) within the genome [64]. The homeotic (Hox) genes play key roles in the anterior-posterior patterning of both vertebrate and invertebrate embryos and in Xenopus are often used as markers of anterior-posterior development. [65-67]. The vertebrate homeotic genes are organized into four clusters arranged in the same order in which they are expressed in the anterior-posterior axis [64]. Of the 39 vertebrate Hox genes, we have identified 28 homologs in Xenopus laevis, while 19 are present in the protein database (Table 7). For those sequences not identified, we sought to determine whether they had been identified in the genome of Xenopus tropicalis. To do so, we used TBLASTX, provided as a tool on the Xenopus tropicalis website [68] to search for the missing sequences. Strong matches were identified for all of the remaining Hox genes except HoxD12. Using the BLASTN tool on the genome site, we confirmed that the gene order was conserved within each scaffold (data not shown). Interestingly, we were unable to identify HoxD12 within the predicted region though both HosxD11 and HoxD13 were recognized.
Table 7 Homeobox genes in X. laevis: for each HOX gene the corresponding cluster and TC is shown, as well as the most 5' clone in the assembly and the protein accession number, if available. When X. laevis genes were not identified, an identifier corresponding X. tropicalis sequence is provided.
IPI Accession Description Xenopus cluster/contig FASTY e-value BLASTX e-value FL Clone Protein Accession
IPI00027694 HOX-A1. cluster:4123 contig:1 4.0e-85 1.99E-99 5536792 AAH44984
IPI00012049 HOX-A2. cluster:7495 contig:1 4.1e-130 7.64E-145 3556495 AAG30508
IPI00012050 HOX-A3. cluster:10945 contig:1 6.5e-91 2.89E-111 4683538 AAH41731
IPI00020926 HOX-A4. fgenesh.C_1023000005
IPI00302291 HOX-A5. cluster:25739 contig:1 6.9e-44 1.27E-38
IPI00010742 HOX-A6. fgenesh.C_1023000003
IPI00010743 HOX-A7. cluster:3210 contig:1 5.8e-40 1.17E-64 XL071e19 AAA49753
IPI00010744 HOX-A9. vm_singlet:264323 1.2e-33 3.48E-29
IPI00010731 HOX-A10. fgenesh.C_1487000003
IPI00010754 HOX-A11. cluster:6499 contig:1 7.2e-42 Was C11 XL088b06
IPI00305850 HOX-A13. vm_singlet:174355 3.8e-57 1.22E-97
IPI00294724 HOX-B1. fgenesh.C_2225000001
IPI00027261 HOX-B2. fgenesh.C_2225000002
IPI00027259 HOX-B3. fgenesh.C_2225000003
IPI00014540 HOX-B4. cluster:22503 contig:1 1.2e-27
IPI00012514 HOX-B5. vm_singlet:57425 8.5e-35 3.92E-59
IPI00015075 HOX-B6. cluster:2339 contig:1 6.2e-42 2.52E-72 XL098k02
IPI00172584 HOX-B7. cluster:1985 singlet:1 2.6e-65 8.16E-77 4201615 P04476
IPI00014536 HOX-B8. cluster:16406 contig:1 2.8e-28 9.90E-43
IPI00014539 HOX-B9. cluster:8543 contig:1 4.0e-30 1.05E-50 XL069k06 P31272
IPI00030703 HOX-B10. cluster:24736 contig:1 5.6e-48 6.95E-74
IPI00295561 HOX-C4. fgenesh.C_202000010
IPI00022893 HOX-C5. vm_singlet:33065 1.5e-41 6.14E-32
IPI00015921 HOX-C6. cluster:9871 singlet:1 4.2e-93 3.16E-109 4202432 P02832
IPI00010756 HOX-C8. cluster:11257 contig:1 5.2e-95 9.74E-118 XL045l21 AAB71818
IPI00010757 HOX-C9. fgenesh.C_202000007
IPI00020947 HOX-C10. cluster:3243 contig:1 1.3e-51 1.63E-127 4970594 AAO25534
IPI00011610 HOX-C11. fgenesh.C_202000005
IPI00010758 HOX-C12. vm_singlet:240042 2.4e-46 2.75E-22
IPI00010759 HOX-C13. cluster:21388 contig:1 2.0e-80 5.86E-89 XL064e01
IPI00001551 HOX-D1. cluster:9419 contig:1 2.5e-50 1.68E-65 3475513 AAA49745
IPI00215882 HOX-D3. cluster:4099 contig:1 2.6e-114 3.48E-121 4684054
IPI00012390 HOX-D4. cluster:21685 contig:1 7.1e-67 7.99E-83 5571854 AAQ95789
IPI00008481 HOX-D8. cluster:11793 contig:1 5.8e-62 2.08E-74 5543040 AAH60408
IPI00292734 HOX-D9. cluster:13847 contig:1 6.5e-38 5.28E-55 XL045k22 CAC44973
IPI00292735 HOX-D10. cluster:6503 contig:1 3.8e-135 3.97E-143 4032032 CAC44974
IPI00305856 HOX-D11. fgenesh.C_1333000003
IPI00018803 HOX-D12. missing
IPI00018806 HOX-D13. cluster:13386 contig:1 1.7e-93 2.17E-112 3399571 AAO25535
Homologue identification from the Cancer Genome Anatomy Project (CGAP)
A second example takes advantage of the CGAP database [69] administered by the National Cancer Institute (NCI). This database and resource incorporates a large number of interconnected modules aimed at gene expression in cancer. Among the modules are a Serial Analysis of Gene Expression (SAGE) database [70,71]. The SAGE approach counts polyadenylated transcripts by sequencing a short 14 bp tag at the genes 3'end and is a quantitative method to examine gene expression [70]. Another module is the Digital Gene Expression Displayer (DGED) which distinguishes statistical differences in gene expression between two pools of libraries [72]. Each method generates tables of genes based on a wide variety of selection criteria. As would be expected, the source for the vast majority of the available data comes from either human or mouse thus demanding a tool to cross match the results in Xenopus.
For this particular example, we selected a tissue based query (DGED) derived from SAGE data in which we sought a set of genes that might include potential markers for glial or astrocyte fates. For this query, we selected all brain, cortex, cerebellum and spinal cord libraries excluding any libraries derived from cell lines. This yielded 58 potential libraries. From this we selected any library labeled as a glioblastoma for pool A and libraries labeled astrocytoma for pool B while excluding the remaining libraries (which included medulloblastomas, ependymomas, etc.). We did not distinguish between cancer grades. This limited the total number of libraries to six glioblastoma and nine astrocytoma libraries containing 487,197 and 863,610 SAGE tags each, respectively. Submission of the query resulted in the identification of 395 tags with a 2× expression factor and a 0.05 significance factor (default CGAP query values). These 395 tags represented 308 different sequences (180 were >2 fold higher in glioblastoma and 128 were >2 fold higher in astrocytoma) which corresponded to 278 proteins in the public database (115 glioblastoma, 163 astrocytoma) and were matched using the batch GenBank accession module available online in XenDB to 100 and 142 Xenopus sequences, respectively. (In the interests of space we have not included the extended table but provide the saved DGED query [see additional file 6] and the two text files [see additional files 7 and 8] that can be uploaded to the XenDB database). The results table includes links to the matching cluster and TC, the e-value and rank and whether a full length clone has been identified. The contig web link leads to additional information including the consensus analysis, the top FASTY hits to five model organisms and links to the Xenopus EST sequences in the TC (Figure 5). Among the genes identified are vimentin (15×, P = 0.01) and sox10 (7.6×, P = 0.03), genes previously established as markers of glial and oligodendrocyte fate respectively [73-75] as well as genes downstream of the Notch signalling pathway, known to be important for glia formation [76]. Thus the system developed and presented here allows 'in silico' based tools established for the study and analysis of other organisms, particularly human and mouse, to be easily and rapidly applied to the Xenopus model system.
Homologues of Drosophila eye development genes
In the final example, we take advantage of the database to perform a comparative analysis of microarray expression data. In many instances, the outcome of an array type experiment is a variety of tables listing regulated genes and the associated expression changes. Currently, there are few published Xenopus array studies available [77-85] while there exist extensive databases of expression for a variety of model organisms. The NCBI maintains a common database, the Gene Expression Omnibus [86] which contains data from over 15,000 samples including 337 Human, 92 mouse and 12 Drosophila experiments (average 25 samples/experiment). Based on an ongoing interest in eye development, we selected a recent paper by Michaut and co-workers in the Gehring lab which examined gene expression changes induced by ectopic expression of the eyeless gene (ey/Pax-6) in Drosophila imaginal disks [87]. The development of the eye is evolutionarily conserved among both vertebrates and invertebrates [88,89]. Many important insights into eye development have come from studies in Drosophila which has defined a genetic cascade of evolutionarily conserved regulatory factors [90]. One such factor is Pax-6/eyeless which is capable of inducing ectopic eyes on both flies [91] and vertebrates [92]. In the Michaut study, 371 eye-induced genes are detected using two different oligonucleotide based array platforms (Affymetrix and Hoffmann-LaRoche) and 73 are discussed in detail within the text (Michaut et al., Table 1, 2). To identify likely homologues of these genes in Xenopus, GenBank accession numbers were obtained from the NCBI Gene Expression Omnibus ([93], accession # GSE271) and used to query the XenDB database to identify 47 potential homologues of the Drosophila Pax6/ey regulated genes and included 32 predicted full length sequences (Table 8). As these sequences are available from commercial sources, they can be readily obtained and tested using the various experimental approaches available to Xenopus such as gain of function studies by microinjection.
Table 8 Xenopus matches to Pax6/ey Regulated Genes identified by Michaut et al.
# Cluster Ctg FL clone Protein Accession Description E-value DM Rank All Rank
1 21344 1 YES AAA19592 Lola protein short isoform 3.90E-10 42 508
2 21344 1 YES AAA19593 Lola protein long isoform 7.00E-10 67 553
3 22774 NO AAA21879 atonal protein 1.20E-17 3 21
4 5646 1 YES AAA28528 fasciclin II 4.30E-28 5 61
5 3838 1 NO AAA28723 eyes absent 1.80E-118 1 49
6 10868 1 YES AAB61239 bunched gene product 2.00E-21 6 39
7 BJ063320 NO AAC46506 Dachshund 1.60E-16 8 44
8 10334 1 YES AAC47196 Lozenge 2.90E-56 4 77
9 7019 1 YES AAD38602 scratch 4.40E-35 15 83
10 4763 2 YES AAD38642 BcDNA.GH11415 2.90E-146 3 15
11 16925 1 YES AAD38646 BcDNA.GH11973 8.20E-14 1 14
12 18882 1 NO AAD52845 Pebble 7.40E-62 2 14
13 3666 1 YES AAF24476 Sticky ch1 1.70E-11 3 56
14 7799 2 YES AAF48990 CG12238-PA 1.70E-22 3 64
15 19264 1 YES AAF55415 CG5407-PA 6.30E-198 3 10
16 5529 1 YES AAF57639 CG15093-PA 2.20E-45 1 24
17 CD327522 NO AAK06753 roughoid/rhomboid-3 1.10E-29 8 26
18 22774 NO AAK14073 DNA-binding transcription factor 8.60E-10 11 158
19 1415 445 YES AAL86442 slamdance 5.50E-70 26 194
20 BU911996 NO AAN74533 transcription factor fruitless 7.90E-10 28 459
21 CD329851 NO BAA78210 white protein 2.20E-36 17 54
22 21321 1 YES CAA33450 glass protein 2.20E-45 21 1739
23 2426 1 YES CAA38746 neurotactin 2.40E-24 103 706
24 9209 1 YES CAA52934 Drosophila cyclin E type I 2.50E-56 2 25
25 18485 1 NO CAA76941 UNC-13 protein 2.70E-165 1 14
26 17438 1 NO NP_523928 CG7525-PA 8.70E-24 101 1508
27 570 1 YES NP_524354 CG4236-PA 0 1 17
28 BI349728 NO NP_573095 CG9170-PA 2.70E-17 1 7
29 1761 1 NO NP_609033 CG9536-PA 1.20E-21 1 6
30 12008 1 YES NP_609545 CG14946-PA 9.10E-25 8 63
31 440 2 YES NP_610108 CG8663-PA 5.70E-17 5 90
32 9019 1 YES NP_611013 CG11798-PA 1.40E-07 156 2411
33 10147 2 YES NP_648269 CG5653-PA 1.90E-16 5 48
34 3752 1 YES NP_649919 CG9427-PA 4.10E-13 1 32
35 20081 1 YES NP_725617 CG5522-PF 7.10E-49 1 18
36 2636 2 YES NP_729075 CG10625- 1.70E-28 16 1185
37 8386 YES O18381 Eyeless protein 3.90E-70 7 75
38 11614 1 YES P00528 Tyrosine-protein kinase Src64B 4.30E-152 3 150
39 4073 1 NO P10181 Homeobox protein rough 3.10E-14 13 165
40 919 NO P20483 String protein (Cdc25-like 3.30E-40 3 43
41 1777 1 YES P36872 Twins protein (PR55) 0 2 41
42 9517 1 YES P48554 Ras-related protein Rac2 1.00E-109 1 22
43 7661 1 YES Q01070 E(spl) mgamma 5.50E-19 5 52
44 7661 1 YES Q01071 E(spl) mdelta 1.10E-15 7 63
45 4146 2 YES Q23989 Villin-like protein quail 6.30E-23 9 138
46 10061 1 YES Q27324 Derailed protein 1.20E-45 23 400
47 14903 1 YES Q27350 Sine oculis protein 3.90E-87 1 20
Sequences without significant homology
48 O77459 transcription factor Ken 60 NP_651346 CG11849-PA
49 AAF46666 CG10527-PA 61 Q23997 Chitinase-like protein DS47 precursor
50 NP_728586 CG9134-PA 62 AAD09748 Gasp precursor
51 NP_609450 CG17124-PA 63 AAF63503 SP2523
52 CG140595 Zea mays genomic 64 AAF47412 CG13897-PA
53 NP_570064 CG10803-PA 65 AAL27368 zinc finger C2H2 protein sequoia
54 NP_650785 CG5835-PA 66 NP_730444 CG32209: CG32209-PB
55 AAF51847 CG11370-PA 67 NP_723827 CG18507-PA
56 AAG46059 SKELETOR 68 NP_611728 CG13532-PA
57 AAN61340 BcDNA:GH10711 69 NP_651343 CG13651-PA
58 NP_729183 CG10121-PA 70 NP_995997 CG12605-PA
59 AAO39528 RE22242p 71 NP_610067 CG9335-PA
Discussion
Comparative approaches to important biological problems have resulted in enormous progress in the past decades. The advent of genomic and proteomic approaches has led to a torrent of data in many organisms and has demanded increasingly sophisticated bioinformatic approaches to organize and manage the information. We have developed an integrated information resource with a user-friendly interface powered by an automated clustering pipeline which will allow researchers to take advantage of the wealth of knowledge available in the public domain.
Comparison to human and mouse
Human and mouse are the best studied vertebrate organisms at the molecular level. In addition to the well publicized genome projects, both have extensive EST collections. This has led to the prediction and characterization of 44,775+ human sequences and 36,182 mouse sequences [94]. As vertebrate development is well conserved, it is important to assess the extent to which the Xenopus EST project has identified the known vertebrate genes. At the same time, one would like to identify any genes that are unique to Xenopus. Most gene prediction programs rely on homology thus eliminating this approach to unique gene identification. Sequences without significant homology could arise from incomplete sequencing that does not extend into the coding region. Results of the human genome project suggest that this would not be the case for a majority of the sequences analyzed in this report. The average 5' UTR in humans is 240 bp and the 3' UTR is 400 bp [95]. Sequencing reactions with current technologies yield readable sequence of 700 bases on average. Therefore, at least some subset of sequences would yield their protein sequence to analysis. An alternative origin of non-homologous sequences would be unspliced or improperly spliced transcripts. This possibility is also minimized by the utilization of polyA tails for RNA selection and reverse transcription priming using oligo(dT). A final, obvious and expensive approach is to select non-homologous sequences for full length double stranded sequencing. Sequence without errors more easily yields the desired open reading frame in even the simplest bioinformatic programs.
Sequences without hits
A class of sequences includes those without significant BLAST hits. In our analysis we have used a cutoff e-value of 10e-6. This of course is necessarily arbitrary, since as mentioned above it is not known what the exact level of similarity is between any given sequence pair. Based on this value, we remain with 43,753 sequences that neither have a BLASTX nor a FASTY hit to a known model organism sequence. The lack of similarity could be due to significant divergence of the sequence, the lack of an appropriate homologue in the public dataset, sequencing errors inherent in EST data or due to the presence of non-coding, presumably regulatory sequences, in the EST clone set. These unmatched sequences mirror the situation in the UniGene set for both mouse and human with greater than 3 and 4 × 106 EST sequences in 76,000 and 106,000 clusters respectively while fewer than 25,000 coding sequences have been recognized [21,94,96]. The source of these discrepancies are currently unclear, but may arise from non coding RNA (ncRNA)[97], micro RNA precursors [98], incompletely or unspliced transcripts [99]. In particular, ncRNAs are a likely source for a large fraction of the discrepancy based on estimates of a 10-fold greater number of non-coding transcription units than protein coding genes [100]. It has been estimated that >95% of transcription is non-coding [101]. Much of the analysis and identification of ncRNA relies on the availability of genomic sequence which is currently unavailable for X. laevis and incomplete for X. tropicalis, the highly homologous diploid species.
Completeness of Xenopus EST set
We have compared all the Xenopus sequences to the human and mouse protein sets to identify conserved proteins. An obvious question is how complete is the Xenopus EST set and what percentage of genes have been identified assuming that the vast majority of protein coding sequences have been evolutionarily conserved. Of the ~40,000 sequences in the IPI databases, 9,225 human and 7,664 mouse sequences do not have a strong match (E < 1.0e-6). Thus, there is a considerable effort remaining to develop a complete Xenopus protein coding set. In the course of our analysis we note the high degree of similarity between the allotetraploid laevis and diploid tropicalis Xenopus species which depended on the length of the matching sequence. For sequences covering >= 95% of the query, there was an average of 94% identity while the average identity dropped to 91% and 88% as the coverage dropped to 90 and 80% respectively. This conservation may allow sequences from both species to be combined to generate a more complete set.
It is well known that the outcome of clustering methods on a large scale depends on the variety of involved parameters. A systematic comparison between UniGene or TIGR Gene Indices and our results turns out to be extremely difficult, mainly because the underlying sequence sets differ as well due to different sequence cleanup and masking approaches. To maximize the utility and usability of our analysis, we have incorporated UniGene and TGI information into our dataset and provide simple tools for identifying the related UniGene and TGI identifier.
Future prospects
Both the clustering and consensus generation approaches are very rapid: 50 minutes for clustering on a single 900 MHz SPARC-CPU and a few hours for assembly on a cluster of 20 heterogeneous SPARC-based machines with 450 to 900 MHz. We therefore have achieved the design goal of being able to frequently update this aspect of the analysis. The subsequent comparative sequence analysis requires significantly greater resources and time (several weeks on same cluster of heterogeneous workstations). The analysis described above is performed by various PERL based scripts developed during the course of our analysis which will allow updates and application to other model systems. We are currently working on a tool to compare clusters over time which will allow the sequence analysis described below to be performed on the restricted set of modified/new clusters rather than to the entire ensemble. The effect of CAP3 consensus generation is that a given cluster can be split into several separate TC sequences, usually due to low sequence quality or differences in the UTR regions of the sequences. The UTR end splitting is likely due to the differences between the in-paralogs in this allotetraploid species. We believe that such information will be of value to those researchers interested in a variety of evolutionary questions, examples of which will be discussed below. The difference in ploidy makes Xenopus laevis distinct from all of the other organisms for which similar analysis have been performed.
As with all ongoing high throughput sequencing efforts, certain aspects of the results change in proportion to the total number of sequences. As noted above, a complete gene set for Xenopus will require additional sequencing. The generation of tetra, octo and dodecaploid species of Xenopus between 80 and 10 million years ago [102] offers opportunities in the field of evolutionary biology. For example, comparisons of 3' UTR regions between in-paralogs of Xenopus laevis and their counterpart diploid tropical species may improve statistical models of molecular evolution. At the genome level, the potential availability of genome data from the polyploid species may provide insight into questions of chromosome segregation and silencing. The selection of Xenopus as a model organism by the NIH and the establishment of the Trans-NIH Xenopus Initiative [103] have directly led to the support of EST and genome sequencing efforts. Among the priorities identified is the establishment and funding of a Xenopus Database [104] which will integrate sequence, expression and other Xenopus data. We hope to be able to update the results described here on a regular basis and contribute to the community effort.
Conclusion
One of the primary goals of the effort was to provide a resource of gene-oriented EST clusters and transcript oriented TCs, enriched with various information from heterogeneous sources, that would be of value to the biology community and the Xenopus community in particular. Using the XenDB system, the biologist can identify sequences of interest using simple gene name queries, accessions, or gene ontologies. The identified sequences have been mapped to public resources like NCBI's UniGene and TIGR Gene Indices and a consensus sequence prepared. In addition, we have identified publicly available IMAGE clones that maximizes the 5' sequence to provide a full length construct when possible. These clones are available from IMAGE consortium providers.
Availability and requirements
Sequence availability, XenDB database and results display
The database and associated files are freely accessible through the XenDB website: . The GenBank accession numbers and FASTA formatted files of the masked and clipped input sequences, as well as the TC sequences and results of the example applications (see below) can be downloaded. Additionally, the list of full length clones is available to researchers interested in performing genome-wide studies. Programs, scripts and database dumps are available from the authors upon request. The XenDB database should be cited with the present publication as a reference.
List of abbreviations used
EST: Expressed Sequence Tag, ORDBMS: Object Relational Database Managemant System, TC: tentative contig sequence, KOG: clusters of euKaryotic Orthologous Groups, GO: Gene Ontology, VRT: Vertebrate Sequences, HTC: High Throughput cDNA, XGC: Xenopus Gene Collection, MGC: Mammalian Gene Collection, ZGC: Zebrafish Gene Collection, FL: Full Length, IPI: International Protein Index, CGAP: Cancer Genome Anatomy Project, DGED: Differential Gene Expression Database, SAGE: Serial Analysis of Gene Expression, ncRNA: non-coding RNA, TGI: TIGR Gene Index
Authors' contributions
A.S. developed and implemented the Vmatch based clustering pipeline. M.B. contributed his high throughput sequence analysis system Genlight. A.S. and M.B. developed the XenDB database schema, performed the post clustering data analyses and contributed to the manuscript. A.H.B. provided supervision and guidance on the development of the project design goals and the interpretation of analysis output with regard to biological significance. R.G. provided supervision and guidance on the development of the clustering pipeline and provided essential infrastructure. C.R.A. provided advice and guidance on the development of the clustering pipeline, the incorporation of analysis into the database and performed and interpreted the various queries presented and wrote a significant portion of the manuscript.
Supplementary Material
Additional File 2
Table S1, Distribution of EST sequences in the analysis based on the annotated tissue source for the preparation of the library. (NOTE: annotations are imported directly from GenBank entries and are dependent on the original annotation.)
Click here for file
Additional File 3
Table S2: The 20 most abundant developmental stage annotations in the X. laevis data set as annotated in GenBank: Distribution of EST sequences in the analysis based on the annotated developmental stage of the source library. (NOTE: annotations are imported directly from GenBank entries and are dependent on the original annotation.)
Click here for file
Additional File 4
Table S3: The 30 most abundant Clone Libraries in the X. laevis data set as determined by the GenBank annotation. (NOTE: annotations are imported directly from GenBank entries and are dependent on the original annotation.)
Click here for file
Additional File 1
Figure S1, Effect of Parameter Variation on EST Clustering: Masked and trimmed EST sequences were clustered using the Vmatch algorithm using different overlap length and percentage identity values. The total number of clusters (blue) and the number of singletons (red) are plotted against the minimal overlap length. Values were plotted at different percentage identities (squares 98%, stars 96%, circles 94%).
Click here for file
Additional File 5
Table S4: Sizes of protein sets used for sequence analysis of clustered sequences.
Click here for file
Additional File 6
file containing the SAGE database query used in the glioblastoma and astrocytoma analysis.
Click here for file
Additional File 7
File containing protein accession numbers of SAGE glioblastoma genes for upload to XenDb system
Click here for file
Additional File 8
File containing protein accession numbers of SAGE astrocytoma genes for upload to XenDb system
Click here for file
Acknowledgements
The authors thank Jan Reinkensmeier for his help in setting up the XenDB Web pages, Alin Vonika, Trent Clarke and Stefan Kurtz for comments on the manuscript. The FSU School of Computation Science and Information Technology and FSU Supercomputing Facility provided computing resources. CRA was supported by an FSU Research Foundation Program Enhancement Grant.
==== Refs
Adams MD Kelley JM Gocayne JD Dubnick M Polymeropoulos MH Xiao H Merril CR Wu A Olde B Moreno RF . Complementary DNA sequencing: expressed sequence tags and human genome project Science 1991 252 1651 1656 2047873
Zhang MQ Computational prediction of eukaryotic protein-coding genes Nat Rev Genet 2002 3 698 709 12209144 10.1038/nrg890
Henderson J Salzberg S Fasman KH Finding genes in DNA with a Hidden Markov Model J Comput Biol 1997 4 127 141 9228612
Besemer J Borodovsky M Heuristic approach to deriving models for gene finding Nucleic Acids Res 1999 27 3911 3920 10481031 10.1093/nar/27.19.3911
Pontius JU Wagner L Schuler GD UniGene: a unified view of the transcriptome The NCBI Handbook 2003 Bethesda, MD, National Center for Biotechnology Information 21-1 21-12
Christoffels A van Gelder A Greyling G Miller R Hide T Hide W STACK: Sequence Tag Alignment and Consensus Knowledgebase Nucleic Acids Res 2001 29 234 238 11125101 10.1093/nar/29.1.234
Mironov AA Fickett JW Gelfand MS Frequent alternative splicing of human genes Genome Res 1999 9 1288 1293 10613851 10.1101/gr.9.12.1288
Ladd AN Cooper TA Finding signals that regulate alternative splicing in the post-genomic era Genome Biol 2002 3 reviews0008 12429065 10.1186/gb-2002-3-11-reviews0008
Lipscombe D Pan JQ Gray AC Functional diversity in neuronal voltage-gated calcium channels by alternative splicing of Ca(v)alpha1 Mol Neurobiol 2002 26 21 44 12392054 10.1385/MN:26:1:021
Stamm S Signals and their transduction pathways regulating alternative splicing: a new dimension of the human genome Hum Mol Genet 2002 11 2409 2416 12351576 10.1093/hmg/11.20.2409
Venables JP Alternative splicing in the testes Curr Opin Genet Dev 2002 12 615 619 12240623 10.1016/S0959-437X(02)00347-7
Roberts GC Smith CW Alternative splicing: combinatorial output from the genome Curr Opin Chem Biol 2002 6 375 383 12023119 10.1016/S1367-5931(02)00320-4
Oklu R Hesketh R The latent transforming growth factor beta binding protein (LTBP) family Biochem J 2000 352 Pt 3 601 610 11104663 10.1042/0264-6021:3520601
Tarone G Hirsch E Brancaccio M De Acetis M Barberis L Balzac F Retta SF Botta C Altruda F Silengo L Retta F Integrin function and regulation in development Int J Dev Biol 2000 44 725 731 11061437
Klint P Claesson-Welsh L Signal transduction by fibroblast growth factor receptors Front Biosci 1999 4 D165 D177 9989949
Chevreux B Pfisterer T Drescher B Driesel AJ Muller WE Wetter T Suhai S Using the miraEST assembler for reliable and automated mRNA transcript assembly and SNP detection in sequenced ESTs Genome Res 2004 14 1147 1159 15140833 10.1101/gr.1917404
Kota R Rudd S Facius A Kolesov G Thiel T Zhang H Stein N Mayer K Graner A Snipping polymorphisms from large EST collections in barley (Hordeum vulgare L.) Mol Genet Genomics 2003 270 24 33 12938038 10.1007/s00438-003-0891-6
Useche FJ Gao G Harafey M Rafalski A High-throughput identification, database storage and analysis of SNPs in EST sequences Genome Inform Ser Workshop Genome Inform 2001 12:194-203. 194 203 11791238
Nekrutenko A Reconciling the numbers: ESTs versus protein-coding genes Mol Biol Evol 2004 21 1278 1282 15034132 10.1093/molbev/msh125
Wang JP Lindsay BG Leebens-Mack J Cui L Wall K Miller WC DePamphilis CW EST clustering error evaluation and correction Bioinformatics 2004 20 2973 84 15189818 10.1093/bioinformatics/bth342
Genome-Consortium Finishing the euchromatic sequence of the human genome Nature 2004 431 931 945 15496913 10.1038/nature03001
Ewing B Green P Analysis of expressed sequence tags indicates 35,000 human genes Nat Genet 2000 25 232 234 10835644 10.1038/76115
Wheeler DL Church DM Edgar R Federhen S Helmberg W Madden TL Pontius JU Schuler GD Schriml LM Sequeira E Suzek TO Tatusova TA Wagner L Database resources of the National Center for Biotechnology Information: update Nucleic Acids Res 2004 32 D35 D40 14681353 10.1093/nar/gkh073
Wheeler DL Church DM Federhen S Lash AE Madden TL Pontius JU Schuler GD Schriml LM Sequeira E Tatusova TA Wagner L Database resources of the National Center for Biotechnology Nucleic Acids Res 2003 31 28 33 12519941 10.1093/nar/gkg033
Schuler GD Pieces of the puzzle: expressed sequence tags and the catalog of human genes J Mol Med 1997 75 694 698 9382993 10.1007/s001090050155
Schuler GD Boguski MS Stewart EA Stein LD Gyapay G Rice K White RE Rodriguez-Tome P Aggarwal A Bajorek E Bentolila S Birren BB Butler A Castle AB Chiannilkulchai N Chu A Clee C Cowles S Day PJ Dibling T Drouot N Dunham I Duprat S East C Hudson TJ . A gene map of the human genome Science 1996 274 540 546 8849440 10.1126/science.274.5287.540
Boguski MS Schuler GD ESTablishing a human transcript map Nat Genet 1995 10 369 371 7670480 10.1038/ng0895-369
Huang X Madan A CAP3: A DNA sequence assembly program Genome Res 1999 9 868 877 10508846 10.1101/gr.9.9.868
Ewing B Green P Base-calling of automated sequencer traces using phred. II. Error probabilities Genome Res 1998 8 186 194 9521922
Phrap sequence assember website 2005 Laboratory of Phil Green, HHMI Genome Sciences Department, University of Washington
Burke J Davison D Hide W d2_cluster: a validated method for clustering EST and full-length cDNAsequences Genome Res 1999 9 1135 1142 10568753 10.1101/gr.9.11.1135
Abouelhoda MI Ohlebusch E Kurtz S Optimal exact string matching based on suffix arrays Proceeding of the Ninth International Symposium on String Processing and Information Retieval Lecture Notes in Computer Science 2002 2476 Springer Verlag 31 43
Abouelhoda MI Kurtz S Ohlebusch E The Enhanced Suffix Array and its Applications to Genome Analysis Proceedings of the Second Workshop on Algorithms in Bioinformatics Lecture Notes in Computer Science 2002 2452 Springer Verlag 449 463
Abouelhoda MI Kurtz S Ohlebusch E Replacing Suffix Trees with Enhanced Suffix Arrays Journal of Discrete Algorithms 2004 2 53 86 10.1016/S1570-8667(03)00065-0
Liang F Holt I Pertea G Karamycheva S Salzberg SL Quackenbush J An optimized protocol for analysis of EST sequences Nucleic Acids Res 2000 28 3657 3665 10982889 10.1093/nar/28.18.3657
Wheeler DL Barrett T Benson DA Bryant SH Canese K Church DM DiCuccio M Edgar R Federhen S Helmberg W Kenton DL Khovayko O Lipman DJ Madden TL Maglott DR Ostell J Pontius JU Pruitt KD Schuler GD Schriml LM Sequeira E Sherry ST Sirotkin K Starchenko G Suzek TO Tatusov R Tatusova TA Wagner L Yaschenko E Database resources of the National Center for Biotechnology Information Nucleic Acids Res 2005 33 D39 D45 15608222 10.1093/nar/gki062
Quackenbush J Cho J Lee D Liang F Holt I Karamycheva S Parvizi B Pertea G Sultana R White J The TIGR Gene Indices: analysis of gene transcript sequences in highly sampled eukaryotic species Nucleic Acids Res 2001 29 159 164 11125077 10.1093/nar/29.1.159
Vector Database Website 2005
The Vmatch large scale sequence analysis software website 2005
Jurka J Repbase update: a database and an electronic journal of repetitive elements Trends Genet 2000 16 418 420 10973072 10.1016/S0168-9525(00)02093-X
Smit A Green P Repeat Masker Website and Server 2005
Beckstette M Strothmann D Homann R Giegerich R Kurtz S PoSSuMsearch: Fast and Sensitive Matching of Position Specific Scoring Matrices Using Enhanced Suffix Arrays In Proceedings of the German Conference on Bioinformatics (GCB 2004) GI Lecture Notes in Informatics, 53:53-64
Kruger J Sczyrba A Kurtz S Giegerich R e2g: an interactive web-based server for efficiently mapping large EST and cDNA sets to genomic sequences Nucleic Acids Res 2004 32 W301 W304 15215398 10.1093/nar/gkh586
Zhang Z Schwartz S Wagner L Miller W A greedy algorithm for aligning DNA sequences J Comput Biol 2000 7 203 214 10890397 10.1089/10665270050081478
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
Pearson WR Wood T Zhang Z Miller W Comparison of DNA sequences with protein sequences Genomics 1997 46 24 36 9403055 10.1006/geno.1997.4995
Beckstette M Mailänder JT Marhöfer RJ Sczyrba A Ohlebusch E Giegerich R Selzer PM Genlight: Interactive high-throughput sequence analysis and comparative genomics Journal of Integrative Bioinformatics Yearbook Bioinformatics 2004 Hofestädt R 2004 8 Magdeburg, IMBio, Informationsmanagement in der Biotechnologie e.V. 79 94
European Bioinformatics Institute International Protein Index Website 2005
Kersey PJ Duarte J Williams A Karavidopoulou Y Birney E Apweiler R The International Protein Index: an integrated database for proteomics experiments Proteomics 2004 4 1985 1988 15221759 10.1002/pmic.200300721
Apweiler R Bairoch A Wu CH Barker WC Boeckmann B Ferro S Gasteiger E Huang H Lopez R Magrane M Martin MJ Natale DA O'Donovan C Redaschi N Yeh LS UniProt: the Universal Protein knowledgebase Nucleic Acids Res 2004 32 (Database issue) D115 D119 14681372 10.1093/nar/gkh131
Aaronson JS Eckman B Blevins RA Borkowski JA Myerson J Imran S Elliston KO Toward the development of a gene index to the human genome: an assessment of the nature of high-throughput EST sequence data Genome Res 1996 6 829 845 8889550
Hillier LD Lennon G Becker M Bonaldo MF Chiapelli B Chissoe S Dietrich N DuBuque T Favello A Gish W Hawkins M Hultman M Kucaba T Lacy M Le M Le N Mardis E Moore B Morris M Parsons J Prange C Rifkin L Rohlfing T Schellenberg K Marra M . Generation and analysis of 280,000 human expressed sequence tags Genome Res 1996 6 807 828 8889549
Komar AA Hatzoglou M Internal ribosome entry sites in cellular mRNAs: The mystery of their existence J Biol Chem 2005 15749702
The Mammalian Gene Collection 2005
The Xenopus Gene Collection 2005
The Zebrafish Gene Collection 2005
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Harris MA Hill DP Issel-Tarver L Kasarskis A Lewis S Matese JC Richardson JE Ringwald M Rubin GM Sherlock G Gene ontology: tool for the unification of biology. The Gene Ontology Consortium Nat Genet 2000 25 25 29 10802651 10.1038/75556
Koonin EV Fedorova ND Jackson JD Jacobs AR Krylov DM Makarova KS Mazumder R Mekhedov SL Nikolskaya AN Rao BS Rogozin IB Smirnov S Sorokin AV Sverdlov AV Vasudevan S Wolf YI Yin JJ Natale DA A comprehensive evolutionary classification of proteins encoded in complete eukaryotic genomes Genome Biol 2004 5 R7 14759257 10.1186/gb-2004-5-2-r7
Bendtsen JD Nielsen H von Heijne G Brunak S Improved prediction of signal peptides: SignalP 3.0 J Mol Biol 2004 340 783 795 15223320 10.1016/j.jmb.2004.05.028
Krogh A Larsson B von Heijne G Sonnhammer EL Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes J Mol Biol 2001 305 567 580 11152613 10.1006/jmbi.2000.4315
Sonnhammer EL von Heijne G Krogh A A hidden Markov model for predicting transmembrane helices in protein sequences Proc Int Conf Intell Syst Mol Biol 1998 6 175 182 9783223
Lash AE Tolstoshev CM Wagner L Schuler GD Strausberg RL Riggins GJ Altschul SF SAGEmap: a public gene expression resource Genome Res 2000 10 1051 1060 10899154 10.1101/gr.10.7.1051
Strausberg RL Buetow KH Greenhut SF Grouse LH Schaefer CF The cancer genome anatomy project: online resources to reveal the molecular signatures of cancer Cancer Invest 2002 20 1038 1050 12449737 10.1081/CNV-120005922
Gehring WJ Affolter M Burglin T Homeodomain proteins Annu Rev Biochem 1994 63 487 526 7979246 10.1146/annurev.bi.63.070194.002415
Cox WG Hemmati-Brivanlou A Caudalization of neural fate by tissue recombination and bFGF development 1995 121 4349 4358 8575335
Wright CV Morita EA Wilkin DJ De Robertis EM The Xenopus XIHbox 6 homeo protein, a marker of posterior neural induction, is expressed in proliferating neurons Development 1990 109 225 234 1976504
Isaacs HV Pownall ME Slack JM Regulation of Hox gene expression and posterior development by the Xenopus caudal homologue Xcad3 EMBO J 1998 17 3413 3427 9628877 10.1093/emboj/17.12.3413
JGI Xenopustropicalis Web Site 2005
Cancer Genome Anatomy Project 2005
Velculescu VE Zhang L Vogelstein B Kinzler KW Serial analysis of gene expression Science 1995 270 484 487 7570003
Boon K Osorio EC Greenhut SF Schaefer CF Shoemaker J Polyak K Morin PJ Buetow KH Strausberg RL De Souza SJ Riggins GJ An anatomy of normal and malignant gene expression Proc Natl Acad Sci U S A 2002 99 11287 11292 12119410 10.1073/pnas.152324199
Lal A Lash AE Altschul SF Velculescu V Zhang L McLendon RE Marra MA Prange C Morin PJ Polyak K Papadopoulos N Vogelstein B Kinzler KW Strausberg RL Riggins GJ A public database for gene expression in human cancers Cancer Res 1999 59 5403 5407 10554005
Kuhlbrodt K Herbarth B Sock E Hermans-Borgmeyer I Wegner M Sox10, a novel transcriptional modulator in glial cells J Neurosci 1998 18 237 250 9412504
Yoshida M Intermediate filament proteins define different glial subpopulations J Neurosci Res 2001 63 284 289 11170178 10.1002/1097-4547(20010201)63:3<284::AID-JNR1022>3.0.CO;2-6
Yoshida M Colman DR Glial-defined rhombomere boundaries in developing Xenopus hindbrain J Comp Neurol 2000 424 47 57 10888738 10.1002/1096-9861(20000814)424:1<47::AID-CNE4>3.0.CO;2-5
Gaiano N Fishell G The role of notch in promoting glial and neural stem cell fates Annu Rev Neurosci 2002 25 471 490 12052917 10.1146/annurev.neuro.25.030702.130823
Konig R Baldessari D Pollet N Niehrs C Eils R Reliability of gene expression ratios for cDNA microarrays in multiconditional experiments with a reference design Nucleic Acids Res 2004 32 e29 14966261 10.1093/nar/gnh027
Crump D Werry K Veldhoen N Van Aggelen G Helbing CC Exposure to the herbicide acetochlor alters thyroid hormone-dependent gene expression and metamorphosis in Xenopus Laevis Environ Health Perspect 2002 110 1199 1205 12460798
Munoz-Sanjuan I Bell E Altmann CR Vonica A Brivanlou AH Gene profiling during neural induction in Xenopus laevis: regulation of BMP signaling by post-transcriptional mechanisms and TAB3, a novel TAK1-binding protein development 2002 129 5529 5540 12403722 10.1242/dev.00097
Tran PH Peiffer DA Shin Y Meek LM Brody JP Cho KW Microarray optimizations: increasing spot accuracy and automated identification of true microarray signals Nucleic Acids Res 2002 30 e54 12060692 10.1093/nar/gnf053
Altmann CR Bell E Sczyrba A Pun J Bekiranov S Gaasterland T Brivanlou AH Microarray-based analysis of early development in Xenopus laevis Dev Biol 2001 236 64 75 11456444 10.1006/dbio.2001.0298
Arima K Shiotsugu J Niu R Khandpur R Martinez M Shin Y Koide T Cho KW Kitayama A Ueno N Chandraratna RA Blumberg B Global analysis of RAR-responsive genes in the Xenopus neurula using cDNA microarrays Dev Dyn 2005 232 414 431 15614783 10.1002/dvdy.20231
Peiffer DA von Bubnoff A Shin Y Kitayama A Mochii M Ueno N Cho KW A Xenopus DNA microarray approach to identify novel direct BMP target genes involved in early embryonic development Dev Dyn 2005 232 445 456 15614776 10.1002/dvdy.20230
Shin Y Kitayama A Koide T Peiffer DA Mochii M Liao A Ueno N Cho KW Identification of neural genes using Xenopus DNA microarrays Dev Dyn 2005 232 432 444 15614765 10.1002/dvdy.20229
Chung HA Hyodo-Miura J Kitayama A Terasaka C Nagamune T Ueno N Screening of FGF target genes in Xenopus by microarray: temporal dissection of the signalling pathway using a chemical inhibitor Genes Cells 2004 9 749 761 15298682 10.1111/j.1356-9597.2004.00761.x
Edgar R Domrachev M Lash AE Gene Expression Omnibus: NCBI gene expression and hybridization array data repository Nucleic Acids Res 2002 30 207 210 11752295 10.1093/nar/30.1.207
Michaut L Flister S Neeb M White KP Certa U Gehring WJ Analysis of the eye developmental pathway in Drosophila using DNA microarrays Proc Natl Acad Sci U S A 2003 100 4024 4029 12655063 10.1073/pnas.0630561100
Glaser T Walton DS Maas RL Genomic structure, evolutionary conservation and aniridia mutations in the human PAX6 gene Nat Genet 1992 2 232 239 1345175 10.1038/ng1192-232
Gehring WJ Ikeo K Pax 6: mastering eye morphogenesis and eye evolution Trends Genet 1999 15 371 377 10461206 10.1016/S0168-9525(99)01776-X
Gehring WJ The genetic control of eye development and its implications for the evolution of the various eye-types Int J Dev Biol 2002 46 65 73 11902689
Halder G Callaerts P Gehring WJ Induction of ectopic eyes by targeted expression of the eyeless gene in Drosophila [see comments] Science 1995 267 1788 1792 7892602
Chow RL Altmann CR Lang RA Hemmati-Brivanlou A Pax6 induces ectopic eyes in a vertebrate development 1999 126 4213 4222 10477290
The NCBI Gene Expression Omnibus 2005
Su AI Wiltshire T Batalov S Lapp H Ching KA Block D Zhang J Soden R Hayakawa M Kreiman G Cooke MP Walker JR Hogenesch JB A gene atlas of the mouse and human protein-encoding transcriptomes Proc Natl Acad Sci U S A 2004 101 6062 6067 15075390 10.1073/pnas.0400782101
Waterston RH Lindblad-Toh K Birney E Rogers J Abril JF Agarwal P Agarwala R Ainscough R Alexandersson M An P Antonarakis SE Attwood J Baertsch R Bailey J Barlow K Beck S Berry E Birren B Bloom T Bork P Botcherby M Bray N Brent MR Brown DG Brown SD Bult C Burton J Butler J Campbell RD Carninci P Cawley S Chiaromonte F Chinwalla AT Church DM Clamp M Clee C Collins FS Cook LL Copley RR Coulson A Couronne O Cuff J Curwen V Cutts T Daly M David R Davies J Delehaunty KD Deri J Dermitzakis ET Dewey C Dickens NJ Diekhans M Dodge S Dubchak I Dunn DM Eddy SR Elnitski L Emes RD Eswara P Eyras E Felsenfeld A Fewell GA Flicek P Foley K Frankel WN Fulton LA Fulton RS Furey TS Gage D Gibbs RA Glusman G Gnerre S Goldman N Goodstadt L Grafham D Graves TA Green ED Gregory S Guigo R Guyer M Hardison RC Haussler D Hayashizaki Y Hillier LW Hinrichs A Hlavina W Holzer T Hsu F Hua A Hubbard T Hunt A Jackson I Jaffe DB Johnson LS Jones M Jones TA Joy A Kamal M Karlsson EK Karolchik D Kasprzyk A Kawai J Keibler E Kells C Kent WJ Kirby A Kolbe DL Korf I Kucherlapati RS Kulbokas EJ Kulp D Landers T Leger JP Leonard S Letunic I Levine R Li J Li M Lloyd C Lucas S Ma B Maglott DR Mardis ER Matthews L Mauceli E Mayer JH McCarthy M McCombie WR McLaren S McLay K McPherson JD Meldrim J Meredith B Mesirov JP Miller W Miner TL Mongin E Montgomery KT Morgan M Mott R Mullikin JC Muzny DM Nash WE Nelson JO Nhan MN Nicol R Ning Z Nusbaum C O'Connor MJ Okazaki Y Oliver K Overton-Larty E Pachter L Parra G Pepin KH Peterson J Pevzner P Plumb R Pohl CS Poliakov A Ponce TC Ponting CP Potter S Quail M Reymond A Roe BA Roskin KM Rubin EM Rust AG Santos R Sapojnikov V Schultz B Schultz J Schwartz MS Schwartz S Scott C Seaman S Searle S Sharpe T Sheridan A Shownkeen R Sims S Singer JB Slater G Smit A Smith DR Spencer B Stabenau A Stange-Thomann N Sugnet C Suyama M Tesler G Thompson J Torrents D Trevaskis E Tromp J Ucla C Ureta-Vidal A Vinson JP Von Niederhausern AC Wade CM Wall M Weber RJ Weiss RB Wendl MC West AP Wetterstrand K Wheeler R Whelan S Wierzbowski J Willey D Williams S Wilson RK Winter E Worley KC Wyman D Yang S Yang SP Zdobnov EM Zody MC Lander ES Initial sequencing and comparative analysis of the mouse genome Nature 2002 420 520 562 12466850 10.1038/nature01262
Morey C Avner P Employment opportunities for non-coding RNAs FEBS Lett 2004 567 27 34 15165889 10.1016/j.febslet.2004.03.117
Bartel DP MicroRNAs: genomics, biogenesis, mechanism, and function Cell 2004 116 281 297 14744438 10.1016/S0092-8674(04)00045-5
Gupta S Zink D Korn B Vingron M Haas SA Genome wide identification and classification of alternative splicing based on EST data Bioinformatics 2004 20 2579 2585 15117759 10.1093/bioinformatics/bth288
Yelin R Dahary D Sorek R Levanon EY Goldstein O Shoshan A Diber A Biton S Tamir Y Khosravi R Nemzer S Pinner E Walach S Bernstein J Savitsky K Rotman G Widespread occurrence of antisense transcription in the human genome Nat Biotechnol 2003 21 379 386 12640466 10.1038/nbt808
Mattick JS Non-coding RNAs: the architects of eukaryotic complexity EMBO Rep 2001 2 986 991 11713189 10.1093/embo-reports/kve230
Sammut B Marcuz A Pasquier LD The fate of duplicated major histocompatibility complex class Ia genes in a dodecaploid amphibian, Xenopus ruwenzoriensis Eur J Immunol 2002 32 1593 1604 12115642 10.1002/1521-4141(200206)32:6<1593::AID-IMMU1593>3.0.CO;2-6
Trans-NIH Xenopus Initiative Website 2005
Xenbase Xenopus Web Resource Website 2005
|
16162280
|
PMC1261260
|
CC BY
|
2021-01-04 16:32:45
|
no
|
BMC Genomics. 2005 Sep 14; 6:123
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-123
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1251616228810.1186/1471-2164-6-125Research ArticleComparative genomic analysis reveals a novel mitochondrial isoform of human rTS protein and unusual phylogenetic distribution of the rTS gene Liang Ping [email protected] Jayakumar R [email protected] Lei [email protected] John J [email protected] Bruce J [email protected] Department of Cancer Genetics, Roswell Park Cancer Institute, Buffalo, USA2 Department of Pharmacology and Therapeutics, Roswell Park Cancer Institute, Buffalo, USA2005 14 9 2005 6 125 125 20 4 2005 14 9 2005 Copyright © 2005 Liang et al; licensee BioMed Central Ltd.2005Liang 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 rTS gene (ENOSF1), first identified in Homo sapiens as a gene complementary to the thymidylate synthase (TYMS) mRNA, is known to encode two protein isoforms, rTSα and rTSβ. The rTSβ isoform appears to be an enzyme responsible for the synthesis of signaling molecules involved in the down-regulation of thymidylate synthase, but the exact cellular functions of rTS genes are largely unknown.
Results
Through comparative genomic sequence analysis, we predicted the existence of a novel protein isoform, rTS, which has a 27 residue longer N-terminus by virtue of utilizing an alternative start codon located upstream of the start codon in rTSβ. We observed that a similar extended N-terminus could be predicted in all rTS genes for which genomic sequences are available and the extended regions are conserved from bacteria to human. Therefore, we reasoned that the protein with the extended N-terminus might represent an ancestral form of the rTS protein. Sequence analysis strongly predicts a mitochondrial signal sequence in the extended N-terminal of human rTSγ, which is absent in rTSβ. We confirmed the existence of rTS in human mitochondria experimentally by demonstrating the presence of both rTSγ and rTSβ proteins in mitochondria isolated by subcellular fractionation. In addition, our comprehensive analysis of rTS orthologous sequences reveals an unusual phylogenetic distribution of this gene, which suggests the occurrence of one or more horizontal gene transfer events.
Conclusion
The presence of two rTS isoforms in mitochondria suggests that the rTS signaling pathway may be active within mitochondria. Our report also presents an example of identifying novel protein isoforms and for improving gene annotation through comparative genomic analysis.
==== Body
Background
The rTS (ENOSF1) gene, a member of the enolase family, was initially identified in Homo sapiens by the discovery of an RNA with extensive complementarity to the mRNA for the DNA biosynthetic enzyme thymidylate synthase[1,2]. The rTS gene was later shown to code for two proteins (rTSα and rTSβ) through alternative RNA splicing [2,3]. The mRNA for rTSα is complementary to thymidylate synthase mRNA, while the mRNA for rTSβ is not [2,3]. The rTSβ protein is the major protein product of the rTS gene and its expression is associated with the down-regulation of thymidylate synthase protein as cultured cells enter growth arrest [2]. Expression of rTSβ correlates with the production of small molecules that appear to mediate the down-regulation of thymidylate synthase protein by a novel intercellular signaling mechanism [2]. Overproduction of rTSβ occurs in some cells resistant to inhibitors of thymidylate synthase or dihydrofolate reductase, indicating a role for the rTS gene in folate and nucleotide metabolism, as well as anticancer drug resistance [2-6].
While the specific function(s) of the rTS gene products are currently under investigation, we now report a new rTS protein isoform and its association with mitochondria. The existence of this new isoform, rTSγ, was first predicted using a computational comparative genomic sequence analysis approach and was then verified experimentally. This unexpected observation suggests that rTS may have functions in addition to intercellular signaling.
Results
A conserved extended protein N-terminus can be deduced from all available rTS genes
Comprehensive analysis of all available database sequences revealed that rTS genes demonstrate an atypical phylogenetic distribution. rTS exists only in a few groups of eubacteria, two fungal lineages (Ascomycota and Basidiomycota), and most animal species from insects to mammals. Among bacterial rTS orthologous genes, several were annotated with a longer N-terminus based on a start codon located further upstream. These proteins include NP_355739.1 (Agrobacterium tumefaciens str. C58), NP_540624.1(Brucella melitensis 16M), NP_639408.1 (Xanthomonas campestris pv. campestris str. ATCC 33913), NP_669902.1 (Yersinia pestis KIM), NP_828458.1 (Streptomyces avermitilis MA-4680), CAD61030.1 (Arthrobacter ilicis), and ZP_00227861.1 (Kineococcus radiotolerans SRS30216), while many other proteins, including NP_405150.1 (Yersinia pestis CO92), NP_437232.1 (Sinorhizobium meliloti), NP_533476.1 (Agrobacterium tumefaciens str. C58), NP_744975.1 (Pseudomonas putida KT2440), ZP_00213853.1 (Burkholderia cepacia R18194), ZP_00281771.1 (Burkholderia fungorum LB400), AAM39023.1 (Xanthomonas axonopodis pv. citri str. 306), and YP_070105 (Yersinia pseudotuberculosis) were annotated with an N-terminus equivalent to that of human rTSβ. Therefore, we determined whether an equivalent extended N-terminus could be predicted in the human rTS gene. Previously, all available human rTS genomic sequences appeared to contain a sequence gap immediately upstream of the start codon of rTSβ, and the published 5'-end of the rTS mRNAs was originally determined by RACE (Rapid Amplification of cDNA Ends) analysis of cloned sequences [2]. Thus, a longer N-terminal was not predicted initially, and not expected based upon the existing experimental evidence. However, at the time we started this analysis, one BAC clone, RP11-778P8 [AC021474.1], was found to contain the sequence covering an extended exon region as well as the rest of the gene, although its sequence was in a status of unordered fragments. In addition, a GenBank entry for the human rTS gene by Dolnick and Su [AF305057] contains the complete 5'-end upstream sequence. Analysis of this sequence by GenScan [7] predicts a start codon upstream of the start codon for rTSβ, yielding an extended N-terminal that is 27 amino acids longer than rTSβ (Fig. 1A). A sequence comparison between the extended human rTS protein region and the bacterial rTS proteins possessing a longer N-terminus revealed a high level of sequence similarity in the extended region. Therefore, we reasoned that a longer N-terminus may exist in all rTS orthologous genes, and this was confirmed by the presence of a possible equivalent extension of the N-terminus in all available rTS genomic sequences (Fig. 1A–E). We named this potential new isoform rTSγ to distinguish it from rTSβ and rTSα. During our preparation of this manuscript, NCBI released a new RefGene entry for human rTSβ [NM_202758 and NP_059982.1]. This entry, dated Dec-20-2004, predicted a different N-terminus and was subsequently replaced by another entry [NM_017512 and NP_059982.2] on 02-March-2005, which has a deduced protein product identical to that described in this report. While all available animal rTS genes, including those from human, three fish species (fresh and salt water pufferfish and zebrafish), the basal chordate, Ciona intestinalis, and the invertebrate Anopheles gambiae, share the same intron/exon boundary position at least for the first exon/intron junction (Fig. 1A&B), the fungal rTS genes seem to vary in gene structure making them different from the animal genes, as well as from each other (Fig. 1C&D).
Figure 1 Prediction of extended N-termini for rTS genes. The genomic sequences and the predicted protein translations for the N-terminal regions from four different species, Homo sapiens, Anopheles gambiae, Ustilago maydis, and Sinorhizobium meliloti, are shown in panels A to D, respectively. A double underlined "ATG" indicates the predicted start codon for the extended N-terminus, while the "ATG" with single underline indicates the start codon of the isoform with a shorter N-terminus. Grey highlights indicate the canonical "GT....AG" intron motif.
In addition to the genomic sequences, EST sequences for rTS genes containing sufficient 5'-end sequences were also identified for a few more animal species, including cow, rat, frog, multiple bony fish species, sea squirt, beetles and mites. Multiple sequence alignment analysis revealed that the extended N-termini of rTS genes are conserved from bacteria to human (Fig. 2). Therefore, we believe that the extended N-terminal region represents an ancestral form of the rTS gene products.
Figure 2 Sequence conservation within the extended N-terminal regions among all available rTS genes. The N-terminal section of all available rTS protein sequences were analyzed using Clustal X followed by boxshading using the BOXSHADE server [32]. The arrow indicates the start methionine in the short form of rTS proteins. Black boxed residues are identical, grey boxed residues are similar. The species name of each sequence is indicated by the last two letters in the sequence ID (ag, Anopheles gambiae; ai, Arthrobacter ilicis; am, Apis mellifera; an, Aspergillus nidulans; at, Agrobacterium tumefaciens; bc, Burkholderia cepacia; bf, Burkholderia fungorum; bm, Brucella melitensis; bt, Bos tauras; cf, Canis familiaris; ci, Ciona intestinalis; cn, Cryptococcus neoformans; dr, Danio rerio; fr, Takifugu rubripes; gz, Gibberella zeae; hs, Homo sapien; kr, Kineococcus radiotolerans; mg, Magnaporthe grisea; nc, Neurospora crassa; pp, Pongo pygmaeus; ppu, Pseudomonas putida KT2440; ps, Polaromonas sp. JS666; pt, Pan troglodytes; rx, Rubrobacter xylanophilus; sa, Streptomyces avermitilis; sm, Sinorhizobium meliloti; ss, Silicibacter sp. TM1040; tr, Takifugu radiatus; um, Ustilago maydis; xa, Xanthomonas axonopodis; xc, Xanthomonas campestris; xl, Xenopus laevis; yp, Yersinia pestis). The accession numbers of the sequences are provide in the last column of the alignment. Sequences labeled with * are those annotated with a shorter N-terminus, while those labeled with # are the ones containing extra sequences at the N-terminus (see complete sequences in Additional file 1).
The extended N-terminus contributes a mitochondrial signal in human rTSγ protein
During our search for potential new motifs and/or functions contributed by the extended N-terminal region of rTS, we found that this extended sequence was predicted to contain a mitochondrial signal. As shown in Table 1, all available programs predict a strong mitochondrial signal for rTSγ, but not for rTSβ, suggesting that the mitochondrial signal is conferred by the extended N-terminal sequence.
Table 1 Predicted mitochondrial signal for rTSβ and γ proteins
Program rTSγ rTSβ Reference
Mitoprot 0.9382 0.0117 [22]
TargetP 0.719 0.077 [23]
PSORTII 26.1% 8.7% [24]
MITOPRED 92.3% 0 [25,26]
rTSγ protein exists and is associated with mitochondria in human cells
We addressed the existence of the rTSγ isoform and its possible association with mitochondria experimentally. Initially, a cytosolic fraction and an organellar pellet (including mitochondria and lysosomes) were prepared from CCRF-CEM human cells, and their proteins were resolved by electrophoresis and analyzed for rTS protein expression (Fig. 3A). The results indicate that two rTS proteins with apparent molecular mass of 52.7 ± 1.8 and 47.6 ± 0.7 kDa (approximately corresponding to the difference in the predicted molecular mass of the rTSβ and γ isoforms, respectively) are present. There was a preferential distribution of the higher molecular mass species in the organellar fraction as compared to the cytosolic fraction. The presence in the organellar fraction of both lysosomal (LAMP-1) and mitochondrial (MnSOD) marker proteins, however, did not allow us to conclude that rTS proteins are present in the mitochondria, rather than in the lysosomes. To resolve this, we partially separated lysosomes from mitochondria using a 5–20% iodixanol gradient [8,9](Fig. 3B). The protein profile of the iodixanol gradient shows a peak centered on fractions 14–15 with a pronounced shoulder in fractions 10–12. GDH activity appears in fractions 12–18 with a peak at fraction 15, indicating the presence of mitochondria in these fractions. The shoulder in the protein profile (fractions 10–12) that lacks GDH activity suggests the presence of non-mitochondrial organelles, including lysosomes. This suggestion was confirmed by Western blotting of the gradient fractions for MnSOD and LAMP-1 markers (Fig. 3C). Analysis of the distribution of rTSβ in the gradient indicates that both the rTSβ and γ species co-localize with MnSOD, but not LAMP-1, conclusively demonstrating their presence in mitochondria, but not lysosomes.
Figure 3 Localization of rTSβ and rTSγ to mitochondria. A. Presence of rTSβ in both cytosol and organelle fractions. Five μg of protein from the organelle (lane 1), cytosol (lane 2), or 5 μg each of the organelle plus cytosol (lane 3) were resolved by gel electrophoresis, blotted and probed for the presence of the indicated proteins. B. Gradient fractionation of mitochondria and lysosomes. The crude organellar suspension from CCRF-CEM cells was fractionated on a linear iodixanol gradient as described. Individual fractions were collected and assayed for GDH activity and protein content. —, Density (g/ml) of fractions from a mock 5–20% iodixanol gradient; -◊-, GDH activity; -○-, Protein content. C. Western blot. Five μl of each fraction was denatured and processed for Western blotting as described.
Discussion
Functional implications for the rTS gene based on its mitochondrial location
It has recently been determined that rTSβ is an enzyme responsible for the synthesis of small signaling molecules involved with the down-regulation of thymidylate synthase as cells enter growth arrest in vitro [6]. The signaling associated with rTSβ was also shown to act intercellularly [6]. Sequence analysis of the rTS gene suggested the presence of a mitochondrial leader sequence that would be expected to increase the length of rTSβ from 416 to 443 amino acids in rTSγ with an expected change in molecular mass from 46,892 to 49,742 Da. Based upon the prediction of a mitochondrial leader sequence, we evaluated whether there is a mitochondrial form of the rTS protein. The results shown in Fig. 3 indicate that this is the case, with two species being present in the mitochondria. These two protein species have apparent molecular masses of 52.7 ± 1.8 and 47.6 ± 0.7 kDa. The smaller species corresponds well to the predicted size expected for rTSβ with the mitochondrial leader sequence cleaved off, while the larger species differs by 3 kDa from the predicted molecular mass for the isoform with the leader sequence. The amounts of rTSγ and rTSβ in the combined cytoplasmic and mitochondrial fractions differ from that expected to result from combining equal amounts of each fraction. The increased abundance of rTSβ, relative to rTSγ in the combined fraction may be the result of in situ cleavage of the rTSγ mitochondrial leader sequence by cytosolic proteases, which has been observed to occur in yeast cytosol [10], although transfer of protein may also contribute to the change in signal strength as the MnSOD signal is weaker in the lane with the combined fractions than in the lane loaded with just the organelle protein. Although the migration of proteins in SDS-PAGE gives only approximate estimation of molecular masses, there is a possibility that other post-translational modifications may contribute to this discrepancy. The co-localization of rTSβ and rTSγ with GDH and MnSOD after subcellular fractionation strongly suggests that the extended N-terminal sequence serves as the leader sequence for mitochondrial location of the protein and that it is likely this is partially cleaved to generate the β isoform, once rTSγ is transported into the mitochondria. The presence of the rTSβ protein in mitochondria raises the question of what role this enzyme and the rTS signaling pathway may play there. Mitochondria are a major site for folate metabolism in mammalian cells [11]. Thymidylate synthase is also found within mitochondria [12], despite the absence of a canonical mitochondrial leader sequence, and the relationship of rTS signaling to thymidylate synthase and folate metabolism may ultimately provide the explanation for this phenomenon. Recent evidence indicates that treatment of cells with an rTSβ signaling mimic can affect the cytoskeleton and cause down-regulation of TYMS [10,13]. The co-localization of rTS and thymidylate synthase in the mitochondria may indicate that rTSβ, in addition to its role in intercellular signaling, also provides intracellular signals that regulate thymidylate synthase levels in the mitochondria.
Since the extended N-termini of rTS are conserved from bacteria to human, we believe that the rTSγ form of the rTS gene represents the ancestral form of this gene, while rTSβ, which seems to be the predominantly expressed form in the cytosol, represents an isoform that appeared later during evolution and came to co-exist with the ancestral isoform, at least in Homo sapiens. Based on a recent study which shows that suboptimal AUG codons can support translation via leaky scanning and reinitiation [14], the co-existence of protein products translated using different AUG codons in the same reading frame may not be a rare phenomena. In fact, the starting AUG codons for rTSγ and rTSβ are both qualified as optimal start codons according to recent studies of translation initiation in mammalian and plant genes [14-16], providing an explanation for the co-existence of rTSγ and rTSβ. A recent study suggests that many alternative splicing forms cause differential subcellular localization, especially in targeting either peroxisomes or mitochondria [17], and our data serves as evidence supporting such a notion. An interesting subject for future studies will be to determine when the shorter rTSβ isoform appeared during evolution.
Phylogenetic distribution and origin of rTS gene
rTS is highly conserved, being found in a variety of species, ranging from bacteria to human. Unlike other enolase genes with a wider phylogenetic distribution, or the thymidylate synthase (TYMS) gene, which is ubiquitous in organisms from all three kingdoms and is highly conserved, rTS demonstrates an unusual phylogenetic distribution. As shown in Fig. 4, the presence of rTS is limited to a few groups of eubacteria including α-, β-, and γ-proteobacteria, actinobacteria, some fungi, and animal species spanning the phylogenetic range from insects to mammals. Among the vertebrate species for which draft genome sequences and/or a large number of EST sequences are available, we were able to identify rTS sequences from all species with the exception of chicken and mouse [see a complete list of deduced amino acid sequences in Additional file 1]. Furthermore, we observed in all vertebrate species with sufficient genome sequences including human, chimpanzee, rat, and fugu fish, that the TYMS-rTS gene pair is part of a large conserved gene synteny among vertebrates (data not shown). While the absence of rTS sequence in chicken may be due to the insufficiency of available genome sequences or ESTs, we were puzzled by the failure to retrieve any rTS sequences in mouse, considering the fact that the mouse genome sequence is now fairly complete and its EST data is quite comprehensive. However, we recently obtained preliminary data showing expression of mouse rTS protein and mRNA (data not shown). Another unusual observation is that the rTS gene does not seem to be present in E. coli while the rTS gene is present in another enterobacterium,Yersinia pestis. Similarly, although the rTS gene is present in Anopheles gambiae and several other insect species, but is not found in Drosophila. So far, no rTS gene has been identified in any plant or yeast species despite the fact that complete genome sequences and comprehensive EST sequences are available for a number of species in these lineages. We also find no rTS sequences from Caenorhabditis elegans or other worm genome sequences. The same is true for all archaebacteria and many eubacteria lineages. These observations suggest that the rTS gene originated as a bacterial gene which was either horizontally transferred into certain animal and fungal lineages, or alternatively, was lost in all the lineages that do not contain rTS genes. An interesting observation is that, while the rTS branches for the animal lineage and the fungal lineage show a topology that agrees well with commonly accepted tree of life for these species, the branches for the bacterial lineages conflict with the commonly accepted tree for the bacterial lineages (Fig. 4). The latter is demonstrated by the fact that species from the same lineage do not always group together, while species from different bacterial lineages do cluster together in many cases (Fig. 4). For example, one β-proteobactial rTS sequence (rTS_bc) is grouped with α-proteobacterial sequences, while the other two β-proteobacterial sequences cluster with a γ-proteobacterial sequence and the three rTS sequences from the High G+C gram positive bacteria are located on three different branches, which all consist of sequences from multiple groups (Fig. 4A). Although not all nodes for the bacterial groups are supported by a high bootstrap value, many nodes are supported with high confidence. This type of unexpected phylotree shown by the rTS sequences suggests the possibility of horizontal gene transfer among the bacteria species, which is common [19-21]. The fact that most of the bacteria and fungi that have a rTS gene are human and/or animal pathogens (Aspergillus nidulans, Brucella suis, Brucella melitensis, Burkholderia cepacia Burkholderia fungorum, Cryptococcus neoformans, Yersinia pestis, Silicibacter sp), plant pathogens (Agrobacterium tumefaciens, Gibberella zeae, Xanthomonas axonopodis, Xanthomonas campestris., Ustilago maydis, Magnaporthe grisea), or in one instance a plant symbiont (S. meliloti), adds weight to the hypothesis of horizontal gene transfer events. In the case of plant pathogens, the plants can serve as a mediator between a bacterial donor and an animal acceptor. This may also suggest that the rTS gene is required to create and/or maintain a certain type of host-pathogen relationship. Therefore, a better understanding of the biological function of the rTS gene may provide new insights for disease control related to these bacterial pathogens.
Figure 4 Unusual phylogenetic distribution of the rTS gene. The sequences of the same set of proteins displayed in Fig. 2 were used to generate a neighbor-joining tree using ClustalX. The tree was displayed in TreeView program, while the taxonomic labels of the species were added manually afterward. In addition to the extension of the N-terminus as indicated in Fig. 2, adjustments to the currently annotated exon boundaries for a few rTS genes were necessarily made based on the best match between their genomic sequences and the human rTSγ protein sequence (see detailed modifications in supplemental materials). Bootstrap values that are over 700 (1000 trials) are shown at the nodes.
In addition to the discovery of this novel mitochondrial isoform of rTS, our report also represents a good illustration of the use of comparative genomic analysis for identifying novel protein isoforms of genes and for improving existing gene predictions. Through this study, we have made a comprehensive collection of rTS orthologous sequences and have made a list of suggested modifications to the existing gene annotation (see detailed modifications in the supplemental materials). In addition, by searching all annotated human genes, we identified a few other genes in which evolutionarily conserved alternative upstream AUG codons exist (similar situation as rTSβ). These genes include SELL [NM_000655], SPCS1 [NM_014041.1], NT5C3 [NM_016489], and SDN1 [NM_014390] (data not shown).
Conclusion
In summary, through comparative genomic analyses we revealed an unusual phylogenetic distribution of the rTS gene and identified a novel mitochondria isoform of this gene and verified it experimentally. A mitochondrial location of rTS protein and phylogenetic distribution of this gene provide us with new information that will assist in elucidating its function.
Methods
Identification of an extended N-terminal region for all available rTS genes
To search for all available rTS orthologous sequences, we first queried all protein sequences deposited in the NCBI non-redundant protein database (nr) by BLAST search [21]. Among all identified rTSβ orthologous protein sequences, the ones with an N-terminus equivalent to human rTSβ were identified and their corresponding genomic sequences, including sufficient 5'-end upstream sequences for analysis, were retrieved. Extended N-termini were deduced by extending the start codon to the next available "ATG" upstream of the start codon used by the existing annotation. In addition to the rTS orthologous genomic sequences, we also searched the NCBI EST database by performing a TBLASTN search with the extended human rTSβ protein sequence to collect rTS cDNA sequences from additional species. To predict any potential new function or cellular location contributed by the extended N-terminus, we analyzed the human rTSγ and rTSβ protein sequences with Mitoprot [22], TargetP [23], PSORTII [24], and Mitopredict [25,26].
Subcellular fractionation of CCRF-CEM cells
The human T-lymphoblastic cell line CCRF-CEM was cultured as described [27]. Subcellular fractions of CCRF-CEM cells were isolated essentially as described [28], except that protease (0.5 mM Pefabloc) and phosphatase inhibitors (1 mM sodium metavanadate and 1 mM NaF) were included in all buffers. All steps were performed at 0 – 4°C. Briefly, 1-liter of CCRF-CEM cells (~3.5 × 105 cells/ml) was centrifuged at 1000 × g for 5 min and the pellet was washed with iced 0.9% NaCl, and then suspended in 5 pellet volumes of ice-cold hypotonic buffer [30] and allowed to swell for 5 min on ice. The suspension was homogenized in an ice-cold 7-ml glass Dounce homogenizer with 15 strokes of the tight pestle to obtain > 95% cell disruption. The homogenate was immediately made up to 250 mM sucrose and centrifuged at 1000 × g for 5 min. The pellet was washed once with 2.5 original pellet volumes of cold isotonic buffer (1 mM Na2-EDTA, 250 mM sucrose; pH 6.9). The two supernatants were combined to generate the post-nuclear supernatant (PNS). The PNS was centrifuged at 17,000 × g, 15 min to generate a cytosolic fraction (supernatant) and an organellar pellet containing both mitochondria and lysosomes. The organellar pellet was suspended in 1 ml of HES (20 mM HEPES-NaOH, 1 mM Na2-EDTA, 250 mM sucrose; pH 7.4). Subcellular fractions were assayed in duplicate for activity of the cytosolic enzyme lactate dehydrogenase (LDH) [30] and mitochondrial matrix enzyme glutamate dehydrogenase (GDH) [31] to ensure that fractionation was successful. To separate mitochondria and lysosomes, this organellar suspension was made up to 30% (w/v) iodixanol [8,32] in a final volume of 2.2 ml and placed on a 5–20% linear iodixanol gradient (9 ml), centrifuged at 70,000 × g for 1.5 hr, and collected as 0.5-ml fractions, which were then diluted with 0.5 ml ice-cold HES and centrifuged at 30,000 × g for 15 min. The resulting organelle pellets were suspended in 200 μl of isotonic buffer (as above, except that NaF was 50 mM). Protein concentration was determined using the BioRad (Hercules, CA) protein assay kit.
Western blotting
Proteins were resolved by denaturing gel electrophoresis using 10% polyacrylamide and transferred to PVDF membranes essentially as described [4]. The primary antibodies used were: D3 (mouse monoclonal to rTSβ), LAMP-1 (mouse monoclonal to lysosome-associated membrane protein-1; Santa Cruz Biotechnology), and MnSOD (rabbit polyclonal to manganese superoxide dismutase; Stressgen). Secondary antibodies consisted of horseradish peroxidase conjugated F(ab')2 fragments and were obtained from Jackson ImmunoResearch Laboratories. Probed blots were imaged using West Pico Dura chemiluminescent reagent (Pierce) and X-OMAT AR X-ray film (Kodak). Pre-stained protein molecular weight markers (BioRad) were included during electrophoresis to allow determination of apparent molecular weights of detected antigens. All experiments were repeated at least twice with similar results.
Authors' contributions
PL conceived the project, performed most of the computational analysis, and drafted the complete manuscript; BJD conceived the project and carried out the western blot experiments; JRN and JJM performed the subcellular fractionation and related enzyme assays; LS contributed to the collection and annotations of rTS sequences.
Supplementary Material
Additional File 1
Deduced rTS protein sequences based on existing DNA sequences and modifications made to existing rTS protein sequences.
Click here for file
Acknowledgements
The authors would like to thank Dr. John M. Graham, JG Research Consultancy, UK, for his suggestions and advice for the preparation of iodixanol gradients. This work is partially supported by a developmental fund from Roswell Park Cancer Institute (PL) and NIH grants CA101515 (PL), EB002116 (BJD), CA43500 (JJM), and CA16056 (RPCI).
==== Refs
Babbitt PC Hasson MS Wedekind JE Palmer DR Barrett WC Reed GH Rayment I Ringe D Kenyon GL Gerlt JA The enolase superfamily: a general strategy for enzyme-catalyzed abstraction of the α-protons of carboxylic acids Biochemistry 1996 35 16489 16501 8987982 10.1021/bi9616413
Dolnick BJ Cloning and characterization of a naturally occurring antisense RNA to human thymidylate synthase mRNA Nucleic Acids Res 1993 21 1747 1752 8493092
Dolnick BJ Black AR Winkler PM Schindler K Hsueh CT rTS gene expression is associated with altered cell sensitivity to thymidylate synthase inhibitors Adv Enzyme Regul 1996 36 165 180 8869746 10.1016/0065-2571(95)00009-7
Dolnick BJ Black AR Alternate splicing of the rTS gene product and its overexpression in a 5-fluorouracil-resistant cell line Cancer Res 1996 56 3207 3210 8764108
Dolnick BJ Lu K Yin MB Rustum YM Recent advances in the study of rTS proteins. rTS expression during growth and in response to thymidylate synthase inhibitors in human tumor cells Adv Enzyme Regul 1997 37 95 109 9381988 10.1016/S0065-2571(96)00007-6
Dolnick BJ Angelino NJ Dolnick R Sufrin JR A novel function for the rTS gene Cancer Biol Ther 2003 2 364 369 14508106
Burge C Karlin S Prediction of complete gene structures in human genomic DNA J Mol Biol 1997 268 78 94 9149143 10.1006/jmbi.1997.0951
Ford T Graham J Rickwood D Iodixanol: a nonionic iso-osmotic centrifugation medium for the formation of self-generated gradients Anal Biochem 1994 220 360 366 7978279 10.1006/abio.1994.1350
Graham J Ford T Rickwood D The preparation of subcellular organelles from mouse liver in self-generated gradients of iodixanol
4 Anal Biochem 1994 220 367 373 7978280 10.1006/abio.1994.1351
Zhou J Bai Y Weiner H Proteolysis prevents in vivo chimeric fusion protein import into yeast mitochondria. Cytosolic cleavage and subcellular distribution J Biol Chem 1995 270 16689 16693 7622479 10.1074/jbc.270.28.16689
Wagner C Symposium on the subcellular compartmentation of folate metabolism J Nutr 1996 126 1228S 1234S 8642461
Samsonoff WA Reston J McKee M O'Connor B Galivan J Maley G Maley F Intracellular location of thymidylate synthase and its state of phosphorylation J Biol Chem 1997 272 13281 13285 9148948 10.1074/jbc.272.20.13281
Dolnick RY Wu Q Angelino NJ Stephanie LV Chow K Sufrin JR Dolnick BJ Enhancement of 5-Fluorouracil Sensitivity by an rTS Signaling Mimic in H630 Colon Cancer Cell Cancer Res 2005 65 5917 5924 15994970 10.1158/0008-5472.CAN-05-0431
Wang XQ Rothnagel JA 5'-untranslated regions with multiple upstream AUG codons can support low-level translation via leaky scanning and reinitiation
3 Nucleic Acids Res 2004 32 1382 1391 14990743 10.1093/nar/gkh305
Kozak M Pushing the limits of the scanning mechanism for initiation of translation Gene 2002 299 1 34 12459250 10.1016/S0378-1119(02)01056-9
Lukaszewicz M M F Jerouville B Stas A Boutry M In vivo evaluation of the context sequence of the translation initiation codon in plants Plant Sci 2000 154 89 98 10725562 10.1016/S0168-9452(00)00195-3
Nakao M Barrero RA Mukai Y Motono C Suwa M Nakai K Large-scale analysis of human alternative protein isoforms: pattern classification and correlation with subcellular localization signals
1 Nucleic Acids Res 2005 33 2355 2363 15860772 10.1093/nar/gki520
Philippe H Douady CJ Horizontal gene transfer and phylogenetics Curr Opin Microbiol 2003 6 498 505 14572543 10.1016/j.mib.2003.09.008
Koonin EV Makarova KS Aravind L Horizontal gene transfer in prokaryotes: quantification and classification Annu Rev Microbiol 2001 55 709 742 11544372 10.1146/annurev.micro.55.1.709
Gogarten JP Doolittle WF Lawrence JG Prokaryotic evolution in light of gene transfer Mol Biol Evol 2002 19 2226 2238 12446813
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
Claros MG Vincens P Computational method to predict mitochondrially imported proteins and their targeting sequences Eur J Biochem 1996 241 779 786 8944766 10.1111/j.1432-1033.1996.00779.x
Emanuelsson O Nielsen H Brunak S von HG Predicting subcellular localization of proteins based on their N-terminal amino acid sequence J Mol Biol 2000 300 1005 1016 10891285 10.1006/jmbi.2000.3903
Nakai K Horton P PSORT: a program for detecting sorting signals in proteins and predicting their subcellular localization Trends Biochem Sci 1999 24 34 36 10087920 10.1016/S0968-0004(98)01336-X
Guda C Guda P Fahy E Subramaniam S MITOPRED: a web server for the prediction of mitochondrial proteins Nucleic Acids Res 2004 32 W372 W374 15215413
Guda C Fahy E Subramaniam S MITOPRED: a genome-scale method for prediction of nucleus-encoded mitochondrial proteins Bioinformatics 2004 20 1785 1794 15037509 10.1093/bioinformatics/bth171
McCloskey DE McGuire JJ Russell CA Rowan BG Bertino JR Pizzorno G Mini E Decreased folylpolyglutamate synthetase activity as a mechanism of methotrexate resistance in CCRF-CEM human leukemia sublines J Biol Chem 1991 266 6181 6187 2007575
McGuire JJ Russell CA Balinska M Human cytosolic and mitochondrial folylpolyglutamate synthetase are electrophoretically distinct. Expression in antifolate-sensitive and -resistant human cell lines J Biol Chem 2000 275 13012 13016 10777604 10.1074/jbc.275.17.13012
Abmayr SMWJL Preparation of nuclear and cytoplasmic extracts from mammalian cells Current protocols in molecular biology 1993 2 Wiley: New York 12.11.11 12.11.19
Storrie B Madden EA Isolation of subcellular organelles Methods Enzymol 1990 182 203 225 2156127
Schmidt E Bergemeyer HU Glutamate dehydrogenase - UV assay Methods in enzymatic analysis 1974 2 New York, Academic Press, Inc 650 656
BOXSHADE server 2005
|
16162288
|
PMC1261261
|
CC BY
|
2021-01-04 16:32:46
|
no
|
BMC Genomics. 2005 Sep 14; 6:125
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-125
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1301617152410.1186/1471-2164-6-130Research ArticleCharacterization of the global profile of genes expressed in cervical epithelium by Serial Analysis of Gene Expression (SAGE) Pérez-Plasencia Carlos [email protected] Gregory [email protected]ázquez-Ortiz Guelaguetza [email protected] José [email protected] Hugo [email protected] Alfredo [email protected]ña-Sanchez Patricia [email protected] Mauricio [email protected] Laboratorio de Oncología Genómica, Unidad de Investigación Médica en Enfermedades Oncológicas, Hospital de Oncología, CMN Siglo XXI-IMSS, Mexico2 John Hopkins University, School of Medicine, Baltimore, MD, USA3 Unidad de Investigación Médica en Enfermedades Autoinmunes, Hospital de Especialidades, CMN Siglo XXI-IMSS México2005 19 9 2005 6 130 130 31 5 2005 19 9 2005 Copyright © 2005 Pérez-Plasencia et al; licensee BioMed Central Ltd.2005Pérez-Plasencia 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
Serial Analysis of Gene Expression (SAGE) is a new technique that allows a detailed and profound quantitative and qualitative knowledge of gene expression profile, without previous knowledge of sequence of analyzed genes. We carried out a modification of SAGE methodology (microSAGE), useful for the analysis of limited quantities of tissue samples, on normal human cervical tissue obtained from a donor without histopathological lesions. Cervical epithelium is constituted mainly by cervical keratinocytes which are the targets of human papilloma virus (HPV), where persistent HPV infection of cervical epithelium is associated with an increase risk for developing cervical carcinomas (CC).
Results
We report here a transcriptome analysis of cervical tissue by SAGE, derived from 30,418 sequenced tags that provide a wealth of information about the gene products involved in normal cervical epithelium physiology, as well as genes not previously found in uterine cervix tissue involved in the process of epidermal differentiation.
Conclusion
This first comprehensive and profound analysis of uterine cervix transcriptome, should be useful for the identification of genes involved in normal cervix uterine function, and candidate genes associated with cervical carcinoma.
==== Body
Background
One of the most frequent malignancies in women worldwide is the Uterine Cervical Carcinoma (CC), both in incidence and mortality and the first cause of death among the Mexican female population [1]. High-risk human papillomavirus (HPV) persistent infection is considered the most important risk factor associated with the development of this tumor [2,3]. Although HPV is a mandatory cause for CC, it is not sufficient to trigger all the changes required for its development [4].
A number of recent studies about gene expression profiles in in vitro HPV-infected cultured keratinocytes and from (CC) clinical samples have provided an initial notion of the changes in gene expression induced by HPV and in early CC [5-10]. Moreover, some studies have compared normal versus tumor-induced gene expression in cervical samples with the aim to identify potential tumor markers of clinical value [11-13].
At present, there are reports of genes expressed by keratinocytes derived from a normal human epidermis and from mouse uterus carried out by Serial Analysis of Gene Expression (SAGE) [14-17]. However, no such study exists for human cervix. Therefore, the aim of our study was to describe the first compendium of expressed genes in normal cervical epithelium, which is composed mainly by keratinocytes strongly influenced by hormones. To achieve this we used SAGE, which is capable of producing an accurate molecular picture of cervical tissue based on expressed genes, as the main methodology. As SAGE is not dependent on preexisting databases of expressed genes, it provides an unbiased view of gene expression profiles within the mRNA populations [18]. SAGE allows the simultaneous quantitative and qualitative analysis of thousands of gene transcripts based on two principles: first, 14 mers are sufficient to uniquely identify 95% of cell transcripts [19]; and second, cloning of these 14 bp tags serially with the insertion of a restriction enzyme recognition sequence as an anchor, the throughput is considerably increased. To obtain a catalog of expressed genes and their relative frequencies we performed database analysis to relate each tag to its corresponding gene [20]. As an important drawback of SAGE is that a large amount of messenger RNA (2.5–5 μg polyA RNA) is required, and our tissue supply was limited (a punch biopsy) we employed the MicroSAGE protocol in RNA thereof [21]. The present report describes a partial transcriptome of a sample derived from normal cervical epithelium used to construct a SAGE library with 30,418 sequenced tags.
Results and discussion
SAGE library derived from one normal uterine ectocervical sample
Our Sage library was obtained from ectocervical tissue from a 38 year old healthy woman with active sexual life, not taking any hormonal therapy, nor any other drug that could potentially alter cervical physiology, we designated this as SAGE_cervix_normal_B_1. Histological analysis of this sample by H&E revealed normal ectocervical tissue, approximately 80% epithelium and 20% stroma without evidence of glands. There were minimal inflammatory infiltrates in the periphery of the sample, considered normal for this type of tissue.
The SAGE library yielded 30,418 sequenced tags, which was used to generate a table, which represents genes expressed in normal human cervix. For a complete list of the expressed genes, please visit the SAGEmap website, [22,23]. The derived catalog of expressed genes represents the first attempt to generate a comprehensive and profound analysis of the cervical epithelium expression profile. The wealth of information obtained allows detection of genes involved in normal epithelium physiology, as well as possible target genes of HPV infection. In general, tag frequency in a typical SAGE experiment follows a normal distribution [24,25]. Table 1 summarizes the general statistics of this library. As seen there is a normal distribution, where only a limited number of tags were either highly expressed or at an extremely low frequency (4.4 and 4.9%, respectively). Tags with a frequency of 1 were not considered for quantitative purposes, because these are likely to represent artifacts of sequencing or of the SAGE procedure [26].
Table 1 SAGE data general statistics.
Frequency distribution a Tags Individual Genesb
>200 tags 1,348 (4.4%) 5 (0.11%)
100–200 tags 2,371 (7.8%)c 21 (0.5%)d
20–100 tags 5,536 (18.19%) 80 (1.8%)
5–19 tags 6,349 (20.8%) 575 (13.2%)
2 – 4 tags 13,316 (43.7%) 3,050 (70.4%)
1 tag 1,498 (4.9%) 609 (14.1%)
Total 30,418 (100%) 4,340 (100%)
No of unique tags 8,062
Tags with SAGE database matches 5,255
No. of different transcripts matched 4,340
No of poorly characterized transcripts 1,215
Genes with known functione 2,453
aCalculation of the frequency distributon of a given tag was based on total tags sequenced in library.
bSome genes have more than one tag, hence we search for individual genes in a database created in SQL language.
c,dPercentage of tags or genes in frequency group.
ebased on FatiGO Datamining website [31]
Representativity of the data
According to Zhang et al. [27], a study of SAGE data mining analysis of 300,000 tags, 75% of mRNA consists of transcripts expressed at more than five copies per cell, and, in general, transcripts are expressed at a range from one to 5,300 copies per cell. With this in mind, our ~30,000-tag library, represents 10% of the total tags analyzed by Velculescu et al. The most frequently represented tag in the current report had a frequency of 515 (16,930 tags per million TPM). An estimate of such data indicates that this gene tag has an expression level of ~5,150 copies per cell, similar to what is observed in digital northerns of other top expressed tags in SAGE libraries (Figure 1A). We have to keep in mind, however, that in certain tissues, some genes are expressed at much higher levels, such as growth hormone, with 149,630 TPM in pituitary gland [28]. Because SAGE analysis represents a qualitative and quantitative assay of messenger RNA abundance not biased by cloning or polymerase chain reaction efficiency [29], our data provide an estimate of the genes normally expressed by normal uterine cervix.
Figure 1 A) Comparison of most expressed tags among different SAGE libraries. Normalized expression levels (TPM) are similar between libraries with different total sequenced tags, indicating comparable messenger abundance among top expressed genes. Expression levels were obtained from SAGEmap website . aTPM: Tags per million. Normalization to compare libraries with different numbers of sequenced tags. TPM is obtained by the following formula [(Tag frequency)(1000,000)/Total No. of sequenced tags]. bDN: digital northern, indicating gene expression level for a specific gene in a library. B. Graphical representation of expression levels (TPM) for three constitutive genes in several normal (N) and tumoral (T) tissues. Brain tissue libraries: SAGE BB542 whitematter (N) and SAGE Brain medulloblastoma B 98 04 P117 (T). Breast: SAGE Breast normal organoid (N) and B SAGE Breast carcinoma epithelium AP DCIS6 (T). Gastric: SAGE normal gastric body epithelial (N) and SAGE Hiroshima GC W246T (T). Liver: SAGE normal liver (N) and SAGE Liver cholangiocarcinoma B K2D (T). Kidney: SAGE Duke Kidney (N) and SAGE_Kidney_carcinoma_B_D2 (T). Colon: SAGE NC2 (N) and SAGE Tu98 (T). Prostate: SAGE PR317 normal prostate (N) SAGE PR317 prostate tumor (T). Lung: SAGE normal lung (N) and SAGE Lung adenocarcinoma MD L10 (T). Expression levels are indicated as tags per million.
Among the most frequently expressed tags in our library (Table 2), some corresponded to ubiquitously expressed transcripts (GRIN2C, FTH1, GNS, RPLP2, RPL21). The presence of this type of genes is a common result in SAGE experiments with an expected heterogeneity in their expression levels [14,15,17,19], indicating a possible role as housekeeping genes (Figure 1B). In a report Velculescu et al., by means of data base analysis of SAGE libraries, found that ~1,000 genes are present in all normal or tumor tissues analyzed with over five copies per cell [30]. Hence, this list of genes identified by data mining is termed minimal transcriptome (i. e., the set of genes expressed by every cell), which represents genes constitutively expressed. In supplementary information of Velculescu's work [30] a search for the minimal transcriptome in our library, indicates >95% of housekeeping genes (data not shown), further validating the cervical library.
Table 2 Top 20 expressed genes in normal cervical tissue.
Tag sequence Tags TPMa UniGene ID Gene Cluster name Biological Functionb
TACCTGCAGA 515 16930 Hs.416073 S100A8 S100 calgranulin A Regulation of cell cycle progression and differentiation.
TAGGTTGTCT 356 11703 Hs.374596 TPT1 Tumor protein, translationally-controlled 1 Unknown
TTTCCTGCTC 276 9073 Hs.139322 SPRR3 Small proline-rich protein 3 Cross-linked envelope protein of keratinocytes
GAGGGAGTTT 201 6607 Hs.523463 RPL27A Ribosomal protein L27a Component of the ribosomal 60S subunit
GTGACCACGG 188 6180 Hs.436980 GRIN2C Glutamate receptor, N-methyl D-aspartate 2C Ionotropic glutamate receptor
GTGGCCACGG 184 6049 Hs.112405 S100A9 S100 calcium binding protein A9 (calgranulin B) Regulation of cell cycle progression and differentiation.
GGGCTGGGGT 173 5687 Hs.425125; Hs.90436 RPL29; SPAG7 Ribosomal protein L29; Sperm associated antigen 7 Component of the ribosomal 60S subunit.
GCATAATAGG 168 5523 Hs.381123 RPL21 Ribosomal protein L21 Component of the ribosomal 60S subunit.
TCAGATCTTT 161 5292 Hs.446628 RPS4X Ribosomal protein S4, X-linked Component of the ribosomal 40S subunit.
GTTGTGGTTA 155 5095 Hs.99785 FLJ21245 Unknown
GGATTTGGCC 151 4964 Hs.437594 RPLP2 Ribosomal protein, large P2 Component of the ribosomal 60S subunit.
TTGGGGTTTC 143 4701 Hs.448738 FTH1 Ferritin, heavy polypeptide 1 Important for iron homeostasis, stores iron in a soluble, nontoxic, readily available form.
TTGGTCCTCT 138 4536 Hs.381172 RPL41 Rribosomal protein L41 Component of the ribosomal 60S subunit.
TGCACGTTTT 130 4273 Hs.265174 RPL32 Ribosomal protein L32 Component of the ribosomal 60S subunit.
ACAAAGCATT 128 4208 Hs.369982 IGFBP5 Insulin-like growth factor binding protein 5 IGF-binding proteins prolong the half-life of the IGFs and have been shown to either inhibit or stimulate the growth promoting effects of the IGF on cell culture.
AGGGCTTCCA 122 4010 Hs.401929 RPL10 Ribosomal protein L10 Component of the ribosomal 60S subunit.
CCACTGCACT 114 3747 Hs.107003 CCNB1IP1 Cyclin B1 interacting protein 1 Functions in progression of the cell cycle through G(2)/M.
GGCAAGCCCC 107 3517 Hs.148340 RPL10A ribosomal protein L10a Component of the ribosomal 60S subunit.
CTGGGTTAAT 105 3451 Hs.334534 GNS Glucosamine (N-acetyl)-6-sulfatase Lysosomal enzyme found in all cells. It is involved in the catabolism of heparin, heparan sulphate, and keratan sulphate.
TGCACTTCAA 103 3386 Hs.62886 SPARCL1 SPARC-like 1 (mast9, hevin) Calcium ion binding
aTPM: Tags per million; [(Tag frequency)(1000,000)/Total No of sequenced tags].
bBiological function obtained from SOURCE, at
Spectrum of genes expressed by normal cervical tissue
To obtain better knowledge of the functional categories of global gene expression profile, we employed the Fatigo Data mining website [31,32]. Figure 2 shows the distribution of expressed genes by functional categories defined by the Gene Ontology Consortium. As seen, the most frequent individual transcripts correspond to genes involved in maintenance and basic metabolism. On the other hand, genes corresponding to other processes such as cell growth regulation, morphogenesis, cell differentiation, or death were not as frequently expressed.
Figure 2 Functional categories assigned to individual genes identified in normal cervical SAGE library. Genes can be assigned in different functional categories. aThe percentage was calculated with 3,764 initial genes from which 2,720 genes had Gene Ontology classification.
Top expressed non-ubiquitous genes in normal cervical tissue mainly correspond to epithelial growth and differentiation
It was important to distinguish which non-ubiquitous genes were predominantly expressed in normal cervix. As seen in table 2, genes related with epithelial differentiation and squamous architectural maintenance are abundantly represented in our library. These include S100A8, S100A9, and SPRR3, that belong to a complex of genes that are subject to coordinate regulation during keratinocyte differentiation. This complex has been called the epidermal differentiation complex (EDC) and is located on the 1q21 chromosome [33,34]. These genes share spatial and temporal expression and interrelated functions and are grouped in three related gene families: cornified envelope precursor proteins (involucrin, loricrin, and the small proline-rich proteins [SPRRs]); intermediate filament-associated proteins (profilaggrin and trichohyalin), and calcium binding proteins (the S100As) [reviewed in [35]]. Approximately 30 genes belonging to the EDC are clustered together in a 200 Mb region, from which there are 20 genes expressed the in cervical SAGE library (Table 3).
Table 3 Genes belonging to 1q21 epidermal differentiation complex (EDC) expressed in cervical tissue
TAG Sequence TAGS TPMa UniGene ID Gene ID Gene name
TACCTGCAGA 515 16930 Hs.416073 S100A8 S100 calcium binding protein A8
TTTCCTGCTC 276 9073 Hs.139322 SPRR3 Small proline-rich protein 3
GTGGCCACGG 184 6049 Hs.112405 S100A9 S100 calcium binding protein A9
GATCAGGCCA 18 591 Hs.275243 S100A6 S100 calcium binding protein A12
GATCTCTTGG 17 558 Hs.38991 S100A2 S100 calcium binding protein A2
AGCAGATCAG 15 493 Hs.400250 S100A10 S100 calcium binding protein A10
CGTGGGACAC 12 394 Hs.110196 NICE-1 (C1orf42) Chromosome 1 open reading frame 42
CAGGCCCCAC 12 394 Hs.417004 S100A11 S100 calcium binding protein A11
ATGTGTAACG 8 263 Hs.81256 S100A4 S100 calcium binding protein A4
GAGCAGCGCC 7 230 Hs.112408 S100A7 S100 calcium binding protein A7
ATGATCCCTG 7 230 Hs.355542 SPRR2A Small proline-rich protein 2A
TTGTGATGTA 7 230 Hs.85844 TPM3 Tropomyosin 3
TTCCCTTACC 6 197 Hs.244349 LCE3D Late cornified envelope 3D
GTCAGGGGAT 5 164 Hs. 12341 ADAR1 ADAR Adenosine deaminase, RNA-specific
CCCTTGAGGA 5 164 Hs.1076 SPRR1B Small proline-rich protein 1B (cornifin)
CCCAGATGAT 4 131 Hs.7854 SLC39A1 Solute carrier family 39 (zinc transporter), member 1
AACCCTAAAA 2 65 Hs.75117 ILF2 Interleukin enhancer binding factor 2, 45 kDa
GCAAATTTGA 2 65 Hs.6396 JTB Jumping translocation breakpoint
CAAGGATCTA 2 65 Hs.355906 NICE-3 (C1orf43) Chromosome 1 open reading frame 43
CAAGGATCTA 2 65 Hs.490551 UBAP2L (NICE-4) Ubiquitin associated protein 2-like
AGCCACTGCA 2 65 Hs.516439 IVL Involucrin
aTPM: Tags per million.
End point RT-PCR analysis confirms expression of genes detected by SAGE
It was important to confirm the expression of some EDC representative genes in different normal cervical tissues by a different technique. For this, we chose end point reverse transcriptase polymerase chain reaction (RT-PCR) analysis. Figure 3 shows the expression of five EDC genes in HPV negative tissue samples with no histopathologic lesion. As expected, the majority of cases expressed these genes. However, there were some differences in the level of expression among the different normal samples. This could be due to the fact that samples were taken on different days of the menstrual cycle (hormonal influence) or to unknown physiological differences among biological systems.
Figure 3 Expression of genes clustered in 1q21, in normal cervical tissues. One hundred nanograms of total RNA purified of each sample was used in one RT-PCR reaction with gene specific primers; then one tenth of each RT-PCR reaction was subjected to agarose gel electrophoresis. MW: molecular weight marker; C1–C6 six different normal cervical samples
Minor expression of fibroblast-related genes in cervical tissue
The gene expression catalog reported on here was obtained from a heterogeneous population of cells composed mainly of epithelial keratinocytes in dissimilar differentiation stages (basale, spinosum and granulosum strata). Nevertheless, these tissues also contain fibroblasts associated with connective, besides other minor cell populations. To know which genes are related to fibroblasts, we compared a SAGE library derived from neonatal foreskin primary fibroblasts (Agnes Baross, British Columbia Genome Sciences Centre). We found 923 gene tags shared by both libraries, which could due to the presence of fibroblasts in the Cervix SAGE library (supplementary information). Shared genes with known biological function reveal that processes as signal transduction, regulation of transcription and cell adhesion are mainly involved. We consider important to identify minor contributions to global gene expression profile in a heterogeneous cell population; however, it is important to note that unknown differences between cervical and neonatal foreskin fibroblasts could exist.
Conclusion
To our knowledge, this is the first effort to achieve a global profile of gene expression in normal cervical tissue. This was accomplished by means of a methodology that produced an accurate catalog of expressed genes in this tissue. Analysis of gene expression revealed genes involved in keratinocyte differentiation. These genes have not been detected in cervical epithelium by traditional methodologies such as RT-PCR or in situ hybridization. Although our SAGE library was derived from a single donor, the majority of samples analyzed expressed the genes selected, indicating reproducibility in human samples. SAGE methodology is a complex and expensive analysis mainly due to the great sequencing efforts required to achieve SAGE libraries. Nevertheless, the overwhelming information derived from these justifies the effort and provides better knowledge of cervical biology and physiology. In a near future, it could also provide an insight of cervical physiology or HPV infection and in other pathologies affecting cervical tissue.
Methods
Tissues
Normal cervices were obtained from women with negative Pap smears, confirmed by histopathological analysis, attending at the Dysplasia Clinic at General Hospital of Mexico, SS who had been subjected to hysterectomy due to uterine myomatosis. All patients were in reproductive age and none of them received hormonal therapy or contraceptives. All the described procedures were evaluated and approved by the local ethics committee of the Mexican Institute of Social Security. Written informed consent was obtained from all the patients. All tissue samples were longitudinally divided in three sections, the central part was snapped frozen in liquid nitrogen and stored at -70°C until nucleic acid extraction, and the other two were fixed overnight in 70% ethanol and were paraffin embedded at the Department of Pathology, Oncology Hospital, National Medical Center SXXI, Mexico. Serial sections from these fractions stained by Haematoxilin/ Eosin were inspected for representativity of the tissue.
HPV detection and typing
Genomic DNA was extracted from the phenol phase left by the TRIzol reagent (Gibco BRL, USA) RNA isolation protocol and amplified by PCR with MY11/MY09 primers [36] (Table 4). PCR products were separated by electrophoresis on 1% agarose gel. Only HPV negative samples were included in this study.
Table 4 Oligonucleotides sequences used in this work.
Gene Sense (5'→3') Antisense (5'→3') Annealing temperature (°C)a Product size (bp) Reference
HPV* GCMCAGGGWCATAAYAATGG CGTCCMARRGGAWACTGATC 55 450 [36]
S100 A8 ATGCCGTCTACAGGGATGAC ACGCCCATCTTTATCACCAG 58 160 This paper
S100 A9 TCAGCTGGAACGCAACATAGA TCAGCTGCTTGTCTGCATT 56 205 This Paper
SPRR3 TTCCACAACCTGGAAACACA TTCAGGGACCTTGGTGTAGC 55 174 This paper
NICE-3 ACGGCTATGAAACAGCCCGCTA GCACATTGCAACTGACTGGCTT 57 330 This paper
NICE-4 ACGGAATCCAATGAGGAAGGCA TCAGTATTGGCTGGCTCTGCAT 57 294 This paper
GAPDH CATCTCTGCCCCCTCTGCTGA GGATGACCTTGCCCACAGCCT 60 205 [38]
aTm was calculated using primerquest program from [39]; however, it was necessary to adjust Tm in some cases.
Micro SAGE protocol
Micro SAGE was performed according to Datson et al. [21] with minor modifications, by means of the Invitrogen's I-SAGE kit (Invitrogen, San Diego, CA USA). RNA isolation was done in TRIzol according to manufacturer's instructions. Five μg of total RNA was used as input material. A heating step was introduced at 65°C for 10 minutes followed by 2 minutes on ice to allow a better separation of concatenamers [37]. Products greater than 300 bp and smaller than 2,000 bp were excised, extracted and cloned in the SphI site of pZero vector. Clones were selected and screened for inserts by PCR. Cervix library was sequenced by Agencourt through SAGE sequencing service (CGAP collaboration, GR). Sequence files were analyzed with the SAGE300 software [18,20], which identifies the anchoring enzyme sites and extracts the two tags flanked by NlaIII site. Gene identity and UniGene cluster assignment of each SAGE tag was obtained by using the tag-to-gene "reliable" map, from SAGEmap NCBI site [22,23]. The tags extracted were uploaded to SAGEmap and corresponding accession numbers were retrieved using the H. sapiens NCBI-GenBank database.
Reverse Transcription-Polymerase Chain Reaction (RT-PCR) analysis
Total RNA was extracted from six normal cervical tissues using TRIzol, quantified by densitometric analysis and its quality evaluated by denaturing gel electrophoresis. Contaminiating DNA was digested and removed with Rnase-free Dnase (Promega). Expression analysis was performed using 100 ng total RNA in a RT-PCR reaction (Access RT-PCR System, Promega). The mRNA was reverse-transcribed at 48°C for 45 min. After an initial denaturation at 94°C for 2 minutes, the double stranded cDNA synthesized was amplified for 40 cycles with denaturation at 94°C for 30 seconds, annealing at 54–60°C for 1 minute and extension at 70°C for 2 minutes with specific oligonucleotides (Table 4) in a Perkin Elmer 480 Thermocycler.
Sense and antisense sequence of oligonucleotides for S100 A8 and 9, SPRR3, NICE-3 and -4 genes were designed with the program Primerquest [38]. GAPDH gene expression was used as an internal control.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
C.P.P. carried out the microSAGE protocol, real time RT-PCR validations, the bioinformatics analysis and writing the manuscript. G.R. provided sequencing of SAGE library. J.M. helped to write the manuscript and participated in discussions. H.G. A.H. and P.P.S. helped for the bioinformatics analysis and database comparisons. M.S. is the principal investigator and was involved in the conceptualization, design and writing of the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional data file 1
Tags shared between fibroblast and cervix. Tags founded in SAGE_cervix_normal_B_1(30418 tags) BJ dermal fibroblasts (57573 tags) and SAGE_cervix_normal_B_1 (30418 tags). Ubiquous tags were deleted in both libraries. BJ dermal fibroblast library was derived from neonatal foreskin primary fibroblasts cultured in Ham's F10 medium supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 ug/ml streptomycin. Library was developed by Agnes Baross at British Columbia Genome Sciences Centre. Columns are: Gene TAG, Cluster ID and Gene Name
Click here for file
Acknowledgements
This work was partially supported by grants of CONACyT (F7114 and 34686-M, MS), and FOFOI-IMSS (MS). For the sequencing of SAGE library, funding was provided by the National Cancer Institute's Cancer Genome Anatomy Project (NIH 23XS073 and 24XS070, GR). During this work CPP, GVO, PPS were recipients of CONACyT, DGEP-UNAM and IMSS fellowships.
This work was submitted in partial fulfilment of the requirements for the D. Sc. degree in for PPC at DOCTORADO EN CIENCIAS BIOMEDICAS, UNIVERSIDAD NACIONAL AUTONOMA DE MEXICO.
==== Refs
Mexican Ministry of Health Registro Histopatológico de Neoplasias Malignas, Compendio de mortalidad y morbilidad 1998 Secretaría de Salud, México
Schiffman MH Bauer HM Hoover RN Glass AG Cadell DM Rush BB Scott DR Sherman ME Kurman RJ Wacholder S Epidemiologic evidence showing that human papillomavirus infection causes most cervical intraepithelial neoplasias J Natl Cancer Inst 1993 85 958 964 8388478
Villa L Human papillomaviruses and cervical cancer Adv Cancer Res 1997 71 321 341 9111869
Franco EL Rohan TE Villa LL Epidemiologic evidence, human papillomavirus infection as a necessary cause of cervical cancer J Natl Cancer I 1999 91 506 511 10.1093/jnci/91.6.506
Ruutu M Peitsaro P Johansson B Syrjanen S Transcriptional profiling of a human papillomavirus 33-positive squamous epithelial cell line which acquired a selective growth advantage after viral integration Int J Cancer 2002 100 318 326 12115547 10.1002/ijc.10455
Duffy CL Phillips SL Klingelhutz AJ Microarray analysis identifies differentiation-associated genes regulated by human papillomavirus type 16 E6 Virology 2003 314 196 205 14517073 10.1016/S0042-6822(03)00390-8
Thomas JT Oh ST Terhune SS Laimins LA Cellular changes induced by low-risk human papillomavirus type 11 in keratinocytes that stably maintain viral episomes J Virol 2001 75 7564 7571 11462028 10.1128/JVI.75.16.7564-7571.2001
Garner-Hamrick PA Fostel JM Chien WM Banerjee NS Chow LT Broker TR Fisher C Global effects of human papillomavirus type 18 E6/E7 in an organotypic keratinocyte culture system J Virol 2004 78 9041 9050 15308700 10.1128/JVI.78.17.9041-9050.2004
Toussaint-Smith E Donner DB Roman A Expression of human papillomavirus type 16 E6 and E7 oncoproteins in primary foreskin keratinocytes is sufficient to alter the expression of angiogenic factors Oncogene 2004 23 2988 2995 14968115 10.1038/sj.onc.1207442
Nees M Geoghegan JM Hyman T Frank S Miller L Woodworth CD Papillomavirus type 16 oncogenes downregulate expression of interferon-responsive genes and upregulate proliferation-associated, NF-kappaB-responsive genes in cervical keratinocytes J Virol 2001 75 4283 4296 11287578 10.1128/JVI.75.9.4283-4296.2001
Chen Y Miller C Mosher R Zhao X Deeds J Morrissey M Bryant B Yang D Meyer R Cronin F Gostout BS Smith-McCune K Schlegel R Identification of cervical cancer markers by cDNA and tissue microarrays Cancer Res 2003 63 1927 1935 12702585
Shim C Zhang W Rhee CH Lee JH Profiling of differentially expressed genes in human primary cervical cancer by complementary DNA expression array Clin Cancer Res 1998 4 3045 3050 9865919
Vazquez-Ortiz G Pina-Sanchez P Vazquez K Duenas A Taja L Mendoza P Garcia JA Salcedo M Overexpression of cathepsin f, matrix metalloproteinases 11 and 12 in cervical cancer BMC Cancer 2005 5 68 15989693 10.1186/1471-2407-5-68
van Ruissen F Jansen BJ de Jongh GJ Zeeuwen PL Schalkwijk J A partial transcriptome of human epidermis Genomics 2002 79 671 678 11991716 10.1006/geno.2002.6756
van Ruissen F Jansen BJ de Jongh GJ van Vlijmen-Willems IM Schalkwijk J Differential gene expression in premalignant human epidermis revealed by cluster analysis of serial analysis of gene expression (SAGE) libraries FASEB J 2002 16 246 248 11772949
Jansen BJ van Ruissen F de Jongh G Zeeuwen PL Schalkwijk J Serial analysis of gene expression in differentiated cultures of human epidermal keratinocytes J Invest Dermatol 2001 116 12 22 11168793 10.1046/j.1523-1747.2001.00218.x
Larose M St-Amand J Yoshioka M Belleau P Morissette J Labrie C Raymond V Labrie F Transcriptome of mouse uterus by serial analysis of gene expression (SAGE): comparison with skeletal muscle Mol Reprod Dev 2004 68 142 148 15095334 10.1002/mrd.20065
Velculescu VE Zhang L Zhou W Vogelstein J Basrai MA Bassett DE JrHieter P Vogelstein B Kinzler KW Characterization of the yeast transcriptome Cell 1997 88 243 251 9008165 10.1016/S0092-8674(00)81845-0
Dinel S Bolduc C Belleau P Boivin A Yoshioka M Calvo E Piedboeuf B Snyder EE Labrie F St-Amand J Reproducibility, bioinformatic analysis and power of the SAGE method to evaluate changes in transcriptome Nucleic Acids Res 2005 33 e26 15716308 10.1093/nar/gni025
Velculescu V Zhang L Vogelstein B Kinzler K Serial analysis of gene expression Science 1995 270 484 487 7570003
Datson N Perk-de Jong J van den Berg M de Kloet E Vreugdenhil E MicroSAGE: a modified procedure for serial analysis of gene expression in limited amounts of tissue Nucleic Acids Res 1999 27 1300 1307 9973618 10.1093/nar/27.5.1300
SAGEmap NCBI
Lash AE Tolstoshev CM Wagner L Schuler GD Strausberg RL Riggins GJ Altschul SF SAGEmap: a public gene expression resource Genome Res 2000 10 1051 1060 10899154 10.1101/gr.10.7.1051
Audic S Claverie JM The Significance of Digital Gene Expression Profiles Genome Res 1997 7 986 995 9331369
Ruijter JM Van Kampen AH Baas F Statistical evaluation of SAGE libraries: consequences for experimental design Physiol Genomics 2002 11 37 44 12407185
Yamamoto M Wakatsuki T Hada A Ryo A Use of serial analysis of gene expression (SAGE) technology J Immunol Methods 2001 250 45 66 11251221 10.1016/S0022-1759(01)00305-2
Zhang L Zhou W Velculescu VE Kern SE Hruban RH Hamilton SR Vogelstein B Kinzler KW Gene expression profiles in normal and cancer cells Science 1997 276 1268 1272 9157888 10.1126/science.276.5316.1268
Nishida Y Yoshioka M St-Amand J The top 10 most abundant transcripts are sufficient to characterize the organs functional specificity: evidences from the cortex, hypothalamus and pituitary gland Gene 2005 344 133 141 15656980 10.1016/j.gene.2004.09.007
Polyak K Riggins GJ Gene Discovery Using the Serial Analysis of Gene Expression Technique: Implications for Cancer Research J Clin Oncol 2001 19 2948 2958 11387368
Velculescu VE Madden SL Zhang L Lash AE Yu J Rago C Lal A Wang CJ Beaudry GA Ciriello KM Cook BP Dufault MR Ferguson AT Gao Y He TC Hermeking H Hiraldo SK Hwang PM Lopez MA Luderer HF Mathews B Petroziello JM Polyak K Zawel L Kinzler KW Analysis of human transcriptomes Nat Genet 1999 23 387 388 10581018 10.1038/70487
Fatigo Data mining
Al-Shahrour F Díaz-Uriarte R Dopazo J FatiGO: a web tool for finding significant associations of Gene Ontology terms to groups of genes Bioinformatics 2004 20 578 580 14990455 10.1093/bioinformatics/btg455
Marenholz I Volz A Ziegler A Davies A Ragoussis I Korge BP Mischke D Genetic analysis of the epidermal differentiation complex (EDC) on human chromosome 1q21: chromosomal orientation new markers a 6-Mb YAC contig Genomics 1996 37 295 302 8938441 10.1006/geno.1996.0563
Elder JT Zhao X Evidence for local control of gene expression in the epidermal differentiation complex Exp Dermatol 2002 11 406 412 12366693 10.1034/j.1600-0625.2002.110503.x
Williams RE Broad S Sheer D Ragoussis J Subchromosomal Positioning of the Epidermal Differentiation Complex (EDC) in Keratinocyte Lymphoblast Interphase Nuclei Exp Cell Res 2002 272 163 175.36 10.1006/excr.2001.5400
Resnick RM Cornelissen MT Wright DK Eichinger GH Fox HS ter Schegget J Manos MM Detection and typing of human papillomavirus in archival cervical cancer specimens by DNA amplification with consensus primers J Natl Cancer Inst 1990 82 1477 1484 2167986
Kenzelmann M Muhlemann K Substantially enhanced cloning efficiency of SAGE (Serial Analysis of Gene Expression) by adding a heating step to the original protocol Nucleic Acids Res 1999 27 917 918 9889294 10.1093/nar/27.3.917
Hsu EM McNicol PJ Guijon FB Paraskevas M Quantification of HPV-16 E6-E7 transcription in cervical intraepithelial neoplasia by reverse transcriptase polymerase chain reaction Int J Cancer 1993 55 397 401 7690739
Primerquest program
|
16171524
|
PMC1261262
|
CC BY
|
2021-01-04 16:32:46
|
no
|
BMC Genomics. 2005 Sep 19; 6:130
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-130
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1351617658610.1186/1471-2164-6-135Research ArticleConstruction and validation of a Bovine Innate Immune Microarray Donaldson Laurelea [email protected] Tony [email protected] Christian [email protected] Ylva [email protected] Antonio [email protected] Sean [email protected] YongHong [email protected] Keren [email protected] Ross [email protected] CSIRO Livestock Industries, Queensland Bioscience Precinct, 306 Carmody Rd., St Lucia 4067, QLD, Australia2 Co-operative Research Centre for Innovative Dairy Products, Level 1, 84 William St, Melbourne, 3000, VIC, Australia2005 22 9 2005 6 135 135 23 6 2005 22 9 2005 Copyright © 2005 Donaldson et al; licensee BioMed Central Ltd.2005Donaldson 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
Microarray transcript profiling has the potential to illuminate the molecular processes that are involved in the responses of cattle to disease challenges. This knowledge may allow the development of strategies that exploit these genes to enhance resistance to disease in an individual or animal population.
Results
The Bovine Innate Immune Microarray developed in this study consists of 1480 characterised genes identified by literature searches, 31 positive and negative control elements and 5376 cDNAs derived from subtracted and normalised libraries. The cDNA libraries were produced from 'challenged' bovine epithelial and leukocyte cells. The microarray was found to have a limit of detection of 1 pg/μg of total RNA and a mean slide-to-slide correlation co-efficient of 0.88. The profiles of differentially expressed genes from Concanavalin A (ConA) stimulated bovine peripheral blood lymphocytes were determined. Three distinct profiles highlighted 19 genes that were rapidly up-regulated within 30 minutes and returned to basal levels by 24 h; 76 genes that were up-regulated between 2–8 hours and sustained high levels of expression until 24 h and 10 genes that were down-regulated. Quantitative real-time RT-PCR on selected genes was used to confirm the results from the microarray analysis. The results indicate that there is a dynamic process involving gene activation and regulatory mechanisms re-establishing homeostasis in the ConA activated lymphocytes. The Bovine Innate Immune Microarray was also used to determine the cross-species hybridisation capabilities of an ovine PBL sample.
Conclusion
The Bovine Innate Immune Microarray has been developed which contains a set of well-characterised genes and anonymous cDNAs from a number of different bovine cell types. The microarray can be used to determine the gene expression profiles underlying innate immune responses in cattle and sheep.
==== Body
Background
Microarray technology is a transcript profiling strategy that allows simultaneous measurement of expression of large numbers of genes in a sample. The expression of thousands of genes can be rapidly monitored in different biological samples allowing the identification of differentially expressed genes. These data, often in conjunction with pre-existing knowledge of specific biochemical pathways and networks, enable a greater understanding of the molecular differences that contribute to the functional specialisation of specific biological samples. Microarrays also have the capacity to identify novel gene networks.
Whilst there are many sources and types of comprehensive microarrays useful for applications with mouse and human samples, microarrays specifically designed for use with samples from production animals, particularly ruminants are not widely available. Some studies have used human or murine microarrays for applications with tissues from livestock production animals [1,2]. However, there is only an average of 86% nucleotide sequence identity between transcripts from cattle and either human or mouse transcripts suggesting that cross-species hybridisations may provide relatively restricted information [1]. Recently, specialised or focused bovine cDNA microarrays have been reported, which are suitable for studies with specific tissues or physiological states. These microarrays provide an excellent tool for examination of gene expression in a specific tissue (eg. muscle) but their general availability is limited [3-9]. There are reports of a relatively small bovine immune-endocrine cDNA microarray representing 167 genes [4] and a third generation immune gene cDNA microarray constructed from bovine leukocytes which contains 1250 genes [10]. Both of these microarrays contain only a limited representation of the many immune related genes, based on surveys of the murine and human scientific literature. Recently, a relatively comprehensive bovine cDNA microarray containing over 18,000 unique transcripts was announced but its general availability is unclear [11]. A bovine Affymetrix microarray has been released although the corresponding gene annotations are limited and the technology is still relatively expensive [12].
There is considerable interest in the identification of bovine and ovine genes that contribute to the relative resistance or susceptibility to disease. This is emphasised by the lack of effective therapeutic strategies for a number of diseases, the costs associated with existing treatments and the range of diseases that need to be considered. For one livestock disease alone, mastitis in dairy cows, it is estimated that economic losses amount to 1.8 billion dollars per annum in the USA, despite considerable management and therapeutic interventions [13]. Mastitis is caused by a wide range of gram negative and gram positive bacteria that in some instances have developed resistance to antibiotic treatment [14-16]. Many other diseases of cattle are also of considerable economic and medical importance eg. Leptospirosis and Johne's disease [17].
One strategy to efficiently counter the variety of infective agents in livestock is to enhance their broad spectrum innate immune resistance, either by marker assisted selective breeding to enrich for advantageous alleles, or active modulation of pivotal proteins that increase broad disease resistance mechanisms. The biological feasibility of these approaches is highlighted by animal breeds that are inherently more resistant to some forms of diseases or parasites as well as specific physiological states that highlight disease susceptibility [18,19]. In addition, mouse models clearly indicate that different strains can show highly variable responses to bacterial challenge [20]. The success of this strategy requires the identification of genes that contribute to resistance mechanisms and to the pathology of disease.
A bovine innate immune cDNA microarray has been constructed to allow identification of genes involved in responses of cattle and sheep to disease challenges. It consists of 1480 defined innate immune related genes selected on the basis of their function in mammals and 5376 anonymous clones selected from subtracted and normalised cDNA libraries constructed from a range of 'challenged' bovine epithelial and leukocyte cells. The former group of genes were identified from a variety of sources, primarily the mammalian literature, reporting genes involved in innate immunity. The latter group included genes produced by immune cells and epithelial cells, recognising the important role of both cell types in innate immunity [21]. The microarray was validated for technical reproducibility and sensitivity using Concanavalin A (ConA) activation of bovine peripheral blood lymphocytes. ConA is a T-cell mitogenic lectin that is often used to model lymphocyte activation responses and has been extensively used in human and mouse studies [22]. In addition, it is demonstrated that the microarray is useful for applications with ovine samples, as a result of the relatively high sequence identity between ovine and bovine transcripts (96 ± 2.4%) [23].
The Bovine Innate Immune Microarray will be useful for monitoring gene expression changes as a response to infection in a wide range of bovine and ovine tissues. This knowledge may allow the development of strategies that exploit these genes and the proteins they encode, leading to enhanced disease resistance in an individual or animal population. There is also increasing evidence for the involvement of some innate immune system components in normal physiological regulatory processes and hence the microarray will also contribute to an understanding of these processes [24].
Results
In this study "signal" refers to the background subtracted mean fluorescence intensity and "element" refers to the DNA amplicon printed onto the microarray slide surface.
Preparation of microarray elements
The Bovine Innate Immune Microarray contains 16,128 individually printed spots. These were made up of 1480 characterised bovine and ovine candidate genes identified by manual literature searches; clones generated from subtracted and normalised cDNA libraries derived from 'challenged' bovine cells and cell lines; and a series of positive and negative control elements (Table 1). The majority of the elements were printed in duplicate adjacent to each other, in a configuration of 48 sub-arrays set out in 4 columns and 12 rows. One 384-well plate of control elements was printed in duplicate at both the top and bottom of each sub-array.
Table 1 Source of elements contributing to the Bovine Innate Immune Microarray
Origin of Microarray Probe Number of Elements
Literature candidates – Bovine 1310
Literature candidates – Ovine 170
Vectors, primers and blanks 8
Lucidea controls set 23
PBL – ConA 576
PBL – Control 960
BoMac – Control 480
BoMac – LPS 480
MAC-T (collagen) – Control 480
MAC-T (collagen) – LPS 480
MAC-T – Control 480
MAC-T – LPS 480
bMEC – Control 480
bMEC – LPS 480
Total 6887
Plasmid clones corresponding to the majority of the candidate genes were identified in available cDNA libraries including; Meat Animal Research Center (MARC) 1–5 libraries [25,26]. and CSIRO Livestock Industries ovine and bovine cDNA libraries [27,28] (see Additional file 1). Candidate genes and controls not present in the cDNA libraries were amplified using gene specific primers (see Additional file 2).
Subtracted, normalised cDNA libraries were constructed using stimulated and unstimulated populations of immune and epithelial cell types, including: bovine peripheral blood lymphocytes (PBL); primary bovine mammary epithelial cells (bMEC); an immortalised bovine macrophage cell line (BoMAC); and an immortalised bovine mammary epithelial cell line (MAC-T). Bovine PBLs were stimulated with Concanavalin A (ConA), whilst the bMEC, BoMAC and MAC-T cell lines were stimulated with lipopolysaccharide (LPS).
The cDNA libraries were made using the Clontech PCR-Select cDNA subtraction protocol as described in the Methods section. The average insert length across all libraries constructed was 450 bp. The effectiveness of library normalisation was initially evaluated by DNA sequence analysis. Fifty clones from each library were sequenced and annotated by BLASTN and BLASTX searches of the Genbank Non-Redundant and Human RefSeq databases. The level of redundancy was calculated by comparing the number of unique gene sequences to the total number of clones sequenced. The average level of redundancy of the subtracted and normalised cDNA libraries was 26.6% with an inter-library redundancy rate of 27.4%. The redundancy primarily consisted of clone duplicates rather than many clones representing a small number of abundant transcripts. Therefore, the maximum number of genes represented on the microarray is approximately 5400.
Analysis of microarray printing
The printed microarray elements were visualised by hybridisation with Panomer™ 9 random oligodeoxynucleotide, Alexa Fluor® 532 conjugate (Molecular Probes, Invitrogen). The slides were visually inspected for irregular or missing spots. The depletion of some spots due to vaporisation or inconsistent pin deposition was noted and taken into account during hybridisation analysis.
Fragments, 500 bp in length, from bovine β-actin and GAPDH were printed in a dynamic range of 25, 50, 100, 200 and 400 pg/spot, (assuming an average deposition of 1 nl printing solution per spot; Rob Moore, personal communication). The signal reported from these elements was averaged across spotted replicates and showed an increase in fluorescence intensity in proportion to the amount of DNA in each spot, up to 200 pg (Fig. 1). However, the signal intensity reached a plateau in the elements containing 200–400 pg per spot, a phenomenon that may be due to stearic hindrance or fluorescence quenching [29]. Using the signal data reported by the dynamic range of β-actin and GAPDH elements as a standard curve, the average quantity of DNA printed for each element was 297 ± 52 pg/spot.
Figure 1 Effect of DNA element quantity on signal from fluorescently labelled random oligonucleotides. The fluorescence intensities of Panomer™ 9 random oligodeoxynucleotides, Alexa Fluor® 532 conjugate (Molecular Probes, Invitrogen) were measured after hybridisation to β-actin(◆) and GAPDH (■) control elements. These elements were spotted in a gradient of 25, 50, 100, 200 or 400 pg DNA per spot. Error bars represent one standard deviation of the mean.
Analyses of DNA element quantity, length and position within the target transcript
A minimum quantity of spotted DNA is required to ensure that spot morphology and reproducibility are optimal. In addition, the spotted DNA needs to be in excess over the target DNA to ensure that the fluorescence signal is not limiting and proportional to the hybridised target DNA. Thus, spotted DNA quantity theoretically should not influence the reported signal. The β-actin and GAPDH control elements, which were printed in a range from 25–400 pg/spot, were used to measure the effect of spot DNA quantity on signal intensity. Data were taken from microarray slides hybridised with labelled cDNA from bovine PBLs stimulated with ConA. The signals from replicate control elements were averaged. β-actin and GAPDH elements report stable and reproducible signal when printed at a quantity of 50–200 pg/spot (Fig. 2(a)). Elements with a DNA quantity of 25 pg/spot have reduced signal strength, presumably because the DNA in the spot was limiting. Elements with a DNA quantity of 400 pg/spot also have reduced signal strength, which may be caused by competitive self hybridisation of the element DNA or possibly stearic hindrance of the fluorescent dyes. Therefore, elements containing between 50–200 pgDNA/spot should report accurate and reproducible signal using labelled cDNA from biological samples.
Figure 2 Effect on the signal reported from DNA elements of varying quantity, length and position within the target transcript. The mean background corrected signal intensity was measured for each β-actin and GAPDH control element on a microarray hybridised with labelled cDNA from ConA stimulated PBLs. (a) Signal reported by β-actin (◆) and GAPDH (■) elements as a function of the quantity of spotted DNA. The error bars denote one standard deviation of the mean. (b) Signal reported by β-actin (◆) and GAPDH (■) elements with DNA lengths ranging from 200 to 1500 bp. (c) Signals reported by β-actin (black) and GAPDH (unshaded) elements of constant DNA length but positioned at the 3'-end, mid-region or 5'-end of the respective target transcripts.
Figure 2(b) shows the effect of varying the length of the spotted β-actin and GAPDH cDNAs. Bovine β-actin and GAPDH elements ranging in length from 200 to 1500 bp from a common 3'-end were generated as described in the Methods section. These elements were spotted onto the microarray at a concentration of 200 pg/spot. Signal data were taken from slides hybridised with labelled cDNA from ConA stimulated bovine PBLs. The GAPDH elements displayed similar signals for all DNA lengths. However, the signal intensity reported from the β-actin control elements increases as the length of the element DNA increased. The shorter 200 and 500 bp fragments contain DNA corresponding to the 3'-UTR of the bovine β-actin gene, whilst the longer control elements include part of the coding region of the β-actin transcript. Therefore, the increased signal reported from the longer elements may be due to an increasing proportion of cross-hybridisation with other members of the actin superfamily, which will bind to the more conserved region of the element DNA.
The elements printed on the Bovine Innate Immune Microarray are a mixture of cDNAs from both the 3' and 5' end of the expressed transcript. To explore the effect this may cause on the reported signal, distinct 400–500 bp elements from the 3'-end, middle and 5'-end of the bovine β-actin and GAPDH transcripts were used as positional controls. These elements were printed at a constant concentration of 200 pg/spot. The labelled cDNA from ConA-stimulated bovine PBLs was produced using both oligo-dT and random hexamer priming. Figure 2(c) demonstrates that the signals for both the β-actin and GAPDH control elements are independent of the position of spotted DNA with respect to the target transcript. Therefore, by using both random hexamer and oligo-dT to generate the labelled cDNA targets, there is no discernible bias between elements from the 3' or 5' ends of the target transcripts.
Microarray reproducibility
Six microarrays hybridised with replicate cDNA samples were used to study slide-to-slide data reproducibility. The variation introduced by using different microarray slides for each hybridisation was determined by calculating the correlation of signal ratio between each slide. cDNAs from ConA stimulated PBLs and unstimulated PBLs were labelled with either Cy3 or Cy5 and hybridised to six independent microarray slides (Fig. 3(a)). An example of the correlation of signal ratio between two slides is presented in figure 3(b). For each element, the signal ratio was calculated by comparing the signal from ConA stimulated PBLs to the signal from unstimulated PBLs. The reproducibility of the signal ratio was determined by calculating the correlation co-efficient between data from each microarray in a pair-wise manner (Table 2). The mean correlation co-efficient was 0.88. The reproducibility of the signal ratio increases to a correlation co-efficient of 0.93, when comparing data from slides with the same dye orientation.
Figure 3 Slide to slide reproducibility of the Bovine Innate Immune Microarray. (a) Schematic diagram of the experimental design. Each arrow represents one microarray slide with the arrow direction indicating the cDNA labelling from Cy5 to Cy3-labelled cDNA. Bovine PBLs were cultured for 24 h with or without ConA (5 μg/ml). (b) An example of slide-to-slide reproducibility depicted as a scatter plot of the signal ratio on slide 1 vs slide 3. The signal ratio was calculated by dividing the background corrected signal for ConA stimulated PBLs by the background corrected signal for unstimulated PBLs.
Table 2 Correlation coefficients of the signal ratio1 data sets reported from six replicate micrroarrays2
ConA-Cy3 ConA-Cy5
Slide 1 Slide 2 Slide 3 Slide 4 Slide 5 Slide 6
ConA-Cy3 Slide 1 1
Slide 2 0.919 1
Slide 3 0.943 0.924 1
ConA-Cy5 Slide 4 0.828 0.800 0.847 1
Slide 5 0.896 0.870 0.909 0.933 1
Slide 6 0.806 0.789 0.840 0.929 0.920 1
Average Ratio Correlation (all data) 0.88
Average Ratio Correlation (ConA = Cy3) 0.93
Average Ratio Correlation (ConA = Cy5) 0.93
1 The signal ratio was calculated by dividing ConA signal divided by Unstimulated signal for each element
2 The schematic diagram of experimental design is depicted in figure 3
Microarray limit of detection
The Lucidea Universal Scorecard RNA mix (Amersham Bioscience, UK) was included on one slide to assess the limit of detection of the Bovine Innate Immune Microarray. The Lucidea RNA mix contains specific transcripts in a range of concentrations. The Lucidea RNA mix (4 μl) was added to sample total RNA (20 μg) to produce a concentration gradient of 0.1, 0.3, 1, 3, 10, 30, 100, 1000 and 3000 pg of Lucidea transcripts per μg of sample RNA. Each Lucidea transcript bound to a specific Lucidea calibration control element present on the microarray. The limit of detection of the microarray was then determined in pg of Lucidea transcript per μg of total RNA. The Lucidea RNA mix was included in both the Cy3 and Cy5 labelling reactions to determine whether there is any difference in hybridisation sensitivity between the two Cy dyes.
The signal reported from each of the Lucidea calibration control elements was averaged across replicates (Fig. 4). The Lucidea transcript present at a concentration of 0.1 pg/μg of total RNA, had a detectable log2 signal of 6.83 ± 0.24 in the Cy5 channel and 8.93 ± 0.40 in the Cy3 channel when hybridised to the microarray. At a concentration of 1 pg/μg of total RNA, transcripts showed no signal intensity differences between the Cy3 and Cy5 dye channels. Therefore, we have set the limit of detection of the Bovine Innate Immune Microarray at 1 pg/μg of sample total RNA, as this is the lowest detectable starting template which produces a reliable signal, independent of Cy dye label.
Figure 4 Estimation of the limit of signal detection for the Bovine Innate Immune Microarray. The Lucidea Universal RNA mix was used in both Cy3 and Cy5 labelling reactions with specific transcripts present at a concentration range of 0.1 to 3000 pg per μg of total sample RNA. The mean background corrected signal, reported from each corresponding Lucidea Calibration control element printed on the microarray is plotted as log2 (signal) in Cy3 (■) and Cy5 (◆) dye channels. Error bars represent one standard deviation from the mean.
Analysis of the time course of activation of PBLs with ConA
Labelled cDNAs were generated from bovine PBLs that had been cultured with 5 μg/μl ConA for time periods of 0, 0.5, 2, 4, 8 or 24 h. cDNA produced from the bovine PBLs collected at each time point was used in a hybridisation mix with cDNA from unstimulated cells. A dye swap microarray was also included for each sample (Fig. 5(a)). An MA plot is used to depict the Cy5:Cy3 signal ratio (M = log2 [Cy5/Cy3]) and total signal intensity (A = 1/2 * ([log2 Cy5]+[log2Cy3])) for each spot on the microarray [30]. Figure 5(b) depicts the MA plots for the five microarray slides hybridised with Cy5-labelled cDNA from stimulated PBLs and Cy3-labelled cDNA from unstimulated cells (MA plots for the dye swap experiments are not shown). The MA plots show an increase in the differences of gene expression between the stimulated/unstimulated samples as the time course progresses. In this example, elements which have a signal ratio above zero have greater signal intensity (and therefore higher transcript abundance) in the ConA-activated sample compared to the unstimulated sample. As the time course progressed, an increasing number of elements were showing higher transcript abundances in the cDNA from ConA-activated PBLs. In addition, a smaller population of elements is showing decreased transcript abundance in the ConA treated sample.
Figure 5 Time course of ConA activation of bovine peripheral blood lymphocytes. (a) Schematic diagram of the experimental design. Each arrow represents one microarray slide with the arrow direction indicating the cDNA labelling from Cy5 to Cy3-label. Bovine PBLs were stimulated with ConA (5 μg/ml) for 0.5, 2, 4, 8 or 24 h. cDNA from the treated cells were compared to cDNA from unstimulated cells. (b) MA plots of microarray data from the ConA activation time course (dye swap replicates are not shown). Labelled cDNA from bovine PBLs treated with ConA (Cy5) were compared to labelled cDNA from unstimulated cells (Cy3). The X-axis shows the total signal intensity for each element present on the microarray (calculated as 1/2*((log2(Cy5) + log2(Cy3))). Y-axis shows the log2 (signal ratio) (log2(Cy5/Cy3)). Elements with a log2 (signal ratio) greater than zero represent transcripts which are more abundant in the ConA activated PBL sample.
A mixed model ANOVA analysis of the data revealed that 252 (4.3%) of the microarray elements were differentially expressed in response to ConA and the sensitivity of this experiment was calculated to be 80 transcripts per million by the method described in Reverter et al. (2005) [31]. This level of sensitivity is in the same range as SAGE gene expression data. Annotation of the differentially expressed gene list, including sequencing of the relevant clones from the anonymous cDNA libraries, found that the 252 elements represented 105 non-redundant transcripts (Fig. 6).
Figure 6 Clusters of differentially expressed genes from PBLs stimulated with ConA. 252 elements representing 109 unique transcripts were found to be differentially expressed in bovine PBLs during the time course of ConA stimulation. Ten primary K-means clusters were grouped according to similarity into three general profiles; Induction 1, Induction 2, and Suppression 1. These are depicted by the average signal from the elements within the cluster. Y-axis is the background corrected mean signal. The gene symbol for each transcript within the cluster is listed.
Using the default K-means clustering analysis in GeneSpring 6.1 software, (Silicon Genetics, Redwood City, CA, USA) the differentially expressed elements were grouped into 10 primary clusters (data not shown). Several of the primary clusters were very similar and contained clones of the same gene. Consequently, the primary clusters were condensed by grouping clusters with similar expression patterns into three basic profiles. As shown in figure 6 the first cluster, Induction 1, represents 19 transcripts that were up-regulated within 30 minutes of stimulation with ConA but which returned to the base line by 24 h. Induction 2 (76 transcripts) represents transcripts that were up-regulated in response to ConA, with an expression peak at 8 h and sustained expression levels until 24 h. The final cluster, Suppression 1, represents 10 transcripts which are down-regulated in response to ConA.
The Induction 2 cluster of differentially expressed genes contains 76 genes and is the largest of the three clusters. Nine of those genes encode chemokine ligands or receptors. A number of secreted growth factors and cytokines were also present in this group (ie CSF1, IFNG, LTA). The cluster also contains 11 heat shock related proteins (chaperonins) and 11 transcription factors. Six genes in the same expression profile are involved in regulating differentiation and apoptosis while a further 8 genes are involved in intracellular signalling or cell cycling.
The 19 genes in Induction 1, which are up-regulated early, include components of the Toll signalling pathway (TRAF1, TRAF5, NFKBIA), the TNF axis (TNF, TNFRSF18) and genes regulating phosphorylation switches (DUSP1, DUSP2, ZAP70 and INPP1). Up-regulation of several transcription factors was also observed (REL, MYC, NR4A1, NR4A3, EGR1, EGR2, JUN) as well as mRNA for the glycoprotein CD69, which is a cellular marker for early activation of lymphocytes.
The majority of genes identified as being differentially expressed (95/105) were more abundant in PBLs stimulated with ConA. However, ten transcripts were down-regulated (Suppression 1) including 4 that encode extracellular matrix proteins (THBS, ITGA5, M160, CSRCR) and 2 genes involved in G-protein signalling (RASGRP2 and RGS14).
qRT-PCR validation of microarray expression profiles
Quantitative real-time RT-PCR (qRT-PCR) was used to independently evaluate the differential expression profile of representative genes from each of the three main profiles. The same total RNA samples were used in both the microarray and qRT-PCR assays to accurately validate the data which was generated using the microarray. Assays were designed for tumour necrosis factor alpha (TNF), interferon gamma (IFNG), chemokine (C-C motif) ligand 3-like 1 (CCL3L1), interleukin 2 receptor alpha (IL2RA), tumor necrosis factor receptor super family member 9 (TNFRSF9) and thrombospondin (THBS). qRT-PCR assays were also performed for the reference genes, acidic ribosomal protein large P0 (RPLPO) and Glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Primer information is listed in Table 3. To compare the data generated by microarray analysis with data from qRT-PCR, the qRT-PCR data was first normalised to the reference gene RPLPO, then the log2 (fold change) was calculated for each time point relative to the gene expression in unstimulated cells.
Table 3 Oligonucleotide primer sequences for qRT-PCR validation of microarray results
Gene name (Symbol) Accession number Forward Reverse Amplicon size (bp)
Acidic Ribosomal Protein Large P0 (RPLP0) NM_001012682 CAACCCTGAAGTGCTTGACAT AGGCAGATGGATCAGCCA 220
Glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH) AF077815 CCTGGAGAAACCTGCCAAGT GCCAAATTCATTGTCGTACCA 226
Tumour Necrosis Factor alpha (TNF) NM_173966 CTGGTTCAGACACTCAGGTCCT GAGGTAAAGCCCGTCAGCA 183
Interferon gamma (IFNG) NM_174086 GTGGGCCTCTCTTCTCAGAA GATCATCCACCGGAATTTGA 234
Chemokine (C-C motif) ligand 3-like 1 (CCL3L1) NM_174511 GGTCTTCTCGGCACCATTT CCAGGTCGGTGATGTATTCC 209
Interleukin 2 receptor alpha (IL2RA) NM_174358 ACCATGATAAACTGCGACTGC GGTTGGTAAGAAAGTTCCACTCC 513
Tumor Necrosis Factor Receptor Super Family Member 9 (TNFRSF9) NM_001561 AAATCCTGCAGTGATCGTGTCC CTTCTTCAGCAGCCCTGGAAT 189
Thrombospondin (THBS) NM_174196.1 CCAATCCTGACCAGAAGGAC TGGCGTACAACCCAGTTAGG 202
Figure 7 depicts the log2 (fold change) calculated from both microarray data and qRT-PCR data for TNF, TNFRSF9, IL2RA, CCL3L1 and THBS. Each plot shows a similar pattern of gene expression across the time course, for both the microarray and qRT-PCR data. However, the log2 (fold change) calculated by qRT-PCR consistently shows a greater magnitude of change compared to the microarray data. This may be due to the larger dynamic range of signal intensities that can be detected using qRT-PCR [32].
Figure 7 Validation of gene expression profiles by quantitative real time RT-PCR (qRT-PCR). The three gene expression profiles discovered by microarray analysis were validated by qRT-PCR with genes representative of each cluster. The log2 (fold change) was calculated for both microarray data (◆) and qRT-PCR data (■) at each treatment time point relative to that in unstimulated cells. TNF represents the gene expression profile of Induction 1; TNFRSF9, IL2RA and CCL3L1 represent the Induction 2 profile and THBS represents the Suppression 1 profile.
TNF was selected as a representative of the Induction 1 profile. The qRT-PCR data accurately corresponds to the general profile calculated for this cluster using the microarray data, i.e. transcript expression levels were markedly increased after 30 minutes of ConA stimulation and the gene expression level had almost returned to the basal level by 24 h. TNFRSF9, IL2RA and CCL3L1 represent elements from the Induction 2 profile. Again the qRT-PCR data confirm the gene expression profile deduced from the microarray data. These transcripts were up-regulated by 2–4 h, gene expression peaked at 8 h and was sustained until the 24 h time point. However, the expression of CCL3L1 returned to base levels at 24 h.
THBS represents elements of Suppression 1 where transcript expression levels are down- regulated in response to ConA stimulation. The qRT-PCR data confirm this pattern of expression. The log2 (fold change) calculated from the microarray data is sustained at a level of -2 at 24 h, while the log2 (fold change) calculated from the qRT-PCR data indicates that THBS expression continues to decline at 24 h. The reason for this difference is not clear but could be due to the greater sensitivity of qRT-PCR.
Cross-species validation
To determine the hybridisation efficiency of ovine transcripts to the Bovine Innate Immune Microarray, a direct comparison of gene expression in equivalent ovine and bovine samples was undertaken. PBLs were extracted from each species. The cells were then cultured for 24 h without stimulation, to avoid any expression differences which may occur due to different species-specific responses to ConA. Ovine, bovine and a 50:50 mixture of ovine and bovine labelled cDNA were applied to the Bovine Innate Immune Microarray in an All Pairs design using dye swaps for each comparison (Fig. 8(a)).
Figure 8 Cross-species hybridisation of an ovine sample to the Bovine Innate Immune Microarray. (a) Schematic diagram of the experiment which used unstimulated ovine and bovine PBLs cultured for 24 h. Each arrow represents one microarray slide where the arrow direction indicates the cDNA labelling from Cy5 to Cy3-label. (b) Scatter plots of microarray data as log2 (signal). The upper and middle panels show the variation observed when comparing signals from an identical cDNA source which was labelled with both Cy3 and Cy5 dye; bovine/bovine (upper panel) and ovine/ovine (middle panel). The lower panel shows a scatter plot of the log2 (signal) from ovine PBL cDNA compared to the log2 (signal) from bovine PBL cDNA.
Labelled cDNA from the bovine sample produced significant signal from 9010 elements (56% of the microarray), while labelled cDNA from the ovine sample produced significant signal from 8444 elements (52% of the microarray). Assuming that the same repertoire of genes is expressed in both species, then the Bovine Innate Immune Microarray detects 94% of the transcripts from unstimulated ovine PBLs. However, only one animal from each species was examined in this experiment, more biological replicates of each species will be required before assumptions can be made regarding species specific gene expression differences.
Figure 8(b) depicts the signal correlation between bovine cDNA labelled with Cy3 and Cy5 (Upper Panel), ovine cDNA labelled with Cy3 and Cy5 (Middle Panel) and bovine labelled cDNA and ovine labelled cDNA (Lower Panel). The correlation coefficients of the signals from a hybridisation using only bovine or only ovine cDNAs are 0.95 and 0.92, respectively. The correlation coefficient for the ovine and bovine comparison is 0.87. Therefore, the ovine genes being detected by the microarray have a very similar relative expression pattern to the corresponding bovine genes.
Discussion
The Bovine Innate Immune Microarray was constructed using the wealth of information pertaining to innate immune related genes from the mouse, human, bovine and ovine literature and complementing this with anonymous genes derived from bovine cells that were 'challenged' to produce innate immune responses. Taking into account the level of redundancy across the cDNA libraries the microarray contains up to 5400 unique bovine genes. Therefore, the microarray should have the potential to identify pathways in bovine cells and tissues activated by a range of immune stimuli. There is also a growing body of evidence that implicates many immune-related genes in normal physiological processes eg. tissue remodelling [33]. The microarray may therefore have wider applications in assessing changes in gene expression accompanying natural physiological and developmental processes. The inclusion of a variety of control elements provides useful tools to assess the reproducibility and sensitivity of the microarray, two of the generally limiting factors of this technology.
The reproducibility of the microarray was ascertained by comparing samples from bovine PBLs with ConA activated PBLs using replicates and dye swaps. The raw signals and signal ratios reported from six independent microarrays were highly reproducible. The average correlation co-efficient of the signal ratio across the six microarrays was 87.7%. Therefore the slide-to-slide reproducibility of the Bovine Innate Immune Microarray is high. The lower limit of signal detection on the Bovine Innate Immune Microarray was measured using 'spike in' RNA transcripts of known quantities from the commercial Lucidea Microarray Scorecard (Amersham Biosciences). This allowed calculation of an empirical lower limit of detection of 1 pg/μg of total RNA.
Interpretation of transcript profiling data assumes that changes in mRNA expression are mirrored by corresponding changes in encoded protein quantity and therefore activity. While this may not always be true it is reasonable to expect this for the many genes particularly with immune related functions which require rapid responses of cells to external stimuli and their rapid return to a resting state in the absence of the stimuli. A number of studies have also shown significant correlation between mRNA and protein expression levels [34,35].
A number of alternative transcript profiling techniques such as EST frequency analysis, Massive Parallel Signature Sequencing (MPSS) [36] and Serial Analysis of Gene Expression (SAGE) [37,38] have potential for much greater depth of coverage compared with microarrays. The former techniques reveal that in most tissues and cells the majority of transcripts are expressed at low levels. Thus, these techniques can provide information on the abundance of transcripts expressed below the sensitivity limit of a microarray, the inference being that microarrays report the transcriptional activities of components of pathways but not the entire pathway or gene network. EST frequency analysis, MPSS and SAGE however, have other limitations such as relative high cost and low sample throughput. Microarray technology provides a compromise that allows assessment of activated gene pathways in the context of relatively large numbers of samples.
The Bovine Innate Immune Microarray is also suitable for use with ovine samples. The relatively high sequence identity of ovine and bovine orthologous transcripts (mean identity of 96 ± 2.4%)[23] and the use of long cDNAs on the microarray underlie this ability. Intriguingly, cross-species microarrays investigating gene expression in comparable tissues may reveal species-specific differences that underlie functional differences. The functional differences in the innate immune responses of comparable tissues in related species are likely to be the result of subtle quantitative differences in large numbers of genes that contribute to the dynamic balance in the overall system. These differences may be ascertained by screening a population of animals from each species.
The power of a microarray can be enhanced by the experimental design [39]. Inclusion of a time course or dose response also enhances the ability of microarray data to identify differentially expressed genes and provides a means to cluster genes behaving in a concordant manner. Microarray analysis was performed on bovine PBLs that had been subjected to activation with ConA over a 24 h period. Analysis of the time course data has identified three major clusters of differentially expressed genes. The challenge is to link groups of genes in these clusters with the known physiological outcomes of lymphocyte activation i.e. secretion of intercellular signalling proteins such as cytokines, chemokines and growth factors, activation of intracellular signalling cascades, cytoskeletal, cell surface and extracellular matrix reorganisation, apoptosis and proliferation. Clearly, the approach is to use known information pertaining to human and murine innate immune pathways to allow an understanding of the pathways activated in bovine cells. Hence, the following discussion has utilised existing information pertaining to murine and human genes to enable interpretation of the data from the bovine PBLs activated with ConA.
Induction 1
TNF axis and Toll pathway
The first cluster, Induction 1, contains genes that were rapidly up-regulated within 30 minutes of stimulation with ConA and subsequently returned to their basal levels at 24 h. This cluster contained 19 transcripts including TNF, a potent extracellular signalling factor that is a pivotal proinflammatory cytokine as well as a multifunctional protein regulating cell proliferation, differentiation and apoptosis)[40]. The cluster encodes components of intracellular signalling pathways known to be involved in responses of cells to immune challenges. In particular, the TNF Receptor Associated Factors, TRAF1, TRAF2 and TNFRSF18 associate with the TNF receptor superfamily. TRAF1 and TRAF2 regulate gene expression in activated lymphocytes by mediating signal transduction from TNF receptors via MAPK8/JNK and NF-κB pathway [41]. TNFRSF18 has been implicated in anti-apoptotic signals via TRAF2, which is thought to be involved in protection of lymphocytes against activation-induced cell death [42,43]. NFKBIA, an inhibitor of NF-κB, was also present in the cluster suggesting that after only 30 minutes stimulation of the PBLs with ConA, homeostatic mechanisms were activated. Indeed, the up-regulation of NFKBIA at this time may be an important mechanism that returns the expression levels of the genes in this cluster to basal levels at 24 h.
Phosphorylation switches
DUSP1 and 2, Dual Specificity Phosphatases 1 and 2, can reverse MAP kinase activation by dephosphorylating phosphotyrosine and phosphothreonine residues [44]. MAP kinase is a central regulatory component of a cascade of signalling proteins that ultimately leads to the activation of many transcription factors [44]. DUSP1 and 2 are also probably dampening the overall stimulatory response and may be contributing to the homeostasis characteristic of this cluster. ZAP70, a protein tyrosine kinase, is associated with the T-cell receptor and plays a key role in NF-κB activation within T-cells [41]. Clearly the changes in the expression levels of DUSP1 and DUSP2, ZAP70 and INPP1 (Inositol Polyphosphate-1-Phosphatase) are consistent with the phosphorylation switches that are known to underlie activation of many of the pathways in activated lymphocytes [45].
Transcription factors
A number of transcription factors are also components of this cluster and these are likely to orchestrate the changes in gene expression that accompany activation of PBLs. This group includes REL, MYC, NR4A1, NR4A3, EGR1, EGR2 and JUN. REL encodes a subunit of the NF-κB complex that controls the expression of a wide range of cytokines and other genes associated with lymphocyte activation [41]. NR4A1 (Nuclear Receptor subfamily 4, group A, member 1) has similarities to the steroid and thyroid hormone receptor families and has been implicated as a regulator of apoptosis in murine thymocytes [46]. The expression of this gene in murine macrophages has been reported to show rapid but transient induction by several mitogens [47], a response identical to that described for this cluster. The role of the related gene NR4A3 is not clear. EGR1 and EGR2 in murine lymphocytes are also induced by mitogenic stimuli but their precise functional roles are not well defined [48]. Interestingly, the transcription factors JUN and FOS cooperate in promoting gene transcription and both EGR1 and EGR2 have structural and functional similarities with FOS [49]. The latter gene is a member of the Suppression 1 cluster and is down-regulated during ConA mediated PBL activation. These observations suggest that the JUN/FOS axis of the MAP kinase pathway is a key regulatory point in bovine lymphocyte activation by ConA.
Glycoprotein
CD69, another member of this cluster, is one of the earliest inducible cell surface glycoproteins during murine lymphoid activation and is involved in regulating lymphocyte proliferation [50].
In summary, the genes in the Induction 1 cluster are largely regulatory in function (with the exception of TNF which can be classed as an intercellular effector) with many involved in controlling lymphocyte activation, apoptosis and gene expression. Some of these genes may be over-expressed at this early time point to limit the activation response and prevent over-stimulation of the PBLs, with consequent detrimental effects on themselves as well as by-stander cells.
Induction 2
The Induction 2 cluster (76 transcripts) is markedly up-regulated and sustained at 24 hours of stimulation with ConA. A number of strong functional themes are evident in this cluster.
Cytokines
The first theme pertains to a group that encodes three pleiotropic and very potent cytokines, CSF1 (Colony Stimulating Factor 1), IFNG (Interferon γ) and LTA (Lymphotoxin α precursor; or TNFβ). These proteins mediate a wide range of immuno-stimulatory, differentiation and proliferative responses of immune cells and are key effector outputs from activated lymphocytes [51]. LTA may also have a role in apoptosis through the TNFR-associated death domain pathway)[40].
Chemokines
Chemokines are major effectors secreted from activated lymphocytes. This group consists of chemokine ligands CCL1, CCL20, CCL3, CCL3L1, CCL4L, CXCL10 and three chemokine receptors, IL2RA, CCR4 and BLR1. Chemokines are typically chemo-attractants that regulate cell trafficking and have fundamental immuno-regulatory roles including involvement in inflammatory processes [51].
Heat shock related
The heat shock related proteins (chaperonins) may be required to facilitate correct protein folding in the transcriptionally activated PBLs or to stimulate NF-κB via the Toll-like receptors [52]. Genes such as HSPD1, HSPA4 and HSPA8 have been directly implicated in innate immune responses of immune cells [52].
Intracellular signalling
The fourth group consists of six genes that have a role in intracellular signalling (RGS1, GBP1, GBP3, GBP5, DUSP5 and PTPN11). RGS1 (Regulator of G-protein Signalling 1) is a regulator of G-protein signalling and may play a role in attenuation of the RAS mediated signal that promotes PBL activation. GBP1, 3 and 5 are IFNG induced guanylate binding proteins [53]. Although their exact roles are unclear they may also have capacity to regulate the RAS signalling pathway. DUSP5 (Dual specificity phosphatase 5) inactivates members of the Mitogen Activated Protein Kinase (MAPK) family that promote lymphocyte activation via NF-κB and therefore may be involved in attenuation of this signal. Although PTPN11 (Protein tyrosine Phosphatase Non-Receptor type) also regulates the activity of the MAPK pathway it is not known whether this promotes or inhibits the pathway [54].
Differentiation and apoptosis regulators
Another group of genes may also be classed as intracellular signalling factors but have additional, specific roles in regulating differentiation and apoptosis [55]. This group consists of GADD45B, CFLAR, CASP7, BCL2, BCL2A1 and IER3. Presumably the function of the proteins encoded by these genes is to maintain the dynamic balance between cell proliferation, differentiation and apoptosis. Many of these genes are induced in stressed or activated immune cells.
Nuclear proteins
A number of genes have been identified that encodes nuclear proteins that play a role in maintenance of the nuclear lamina (LMNB1), bind to the nuclear matrix (SRRM1) or are involved in the assembly of ribosomes (NPM1 and NCL), suggesting major alterations in the organisation and function of the nucleus of activated cells [56]. Consistent with this there was also over-expression of a histone (H2AFZ) and a histone deacetylase (HDAC5) in this group.
Transcription factors
A functionally diverse array of 11 transcription factors are represented in this cluster (TRIP12, HIF1A, BATF, APEX1, ATF3, CEBPG, IRF1, SLAMF1, PARK7, RAN, RBM13 and NFKB1). Notable amongst these is NFKB1 a pivotal transcription factor regulating the expression of a large number of immune related genes (see above) and CEBPG which cooperates with FOS to regulate the transcription of genes containing PRE-1 enhancer elements [41,57]. (see the earlier discussion on FOS). Clearly the ConA activated lymphocytes have enhanced transcriptional activity and this, in conjunction with increases in the chaperonins, is consistent with an enhanced role in the production of secreted cytokines and chemokines, the primary effectors secreted from these cells.
Cell surface proteins
This cluster also contains a group of genes encoding proteins bound to the cell surface (ICOS, SCARB1, SLAMF1, ANXA9, C1QBP and CD53). Three of these genes, ICOS, SLAMF1 and CD53 encode proteins that play important roles in cell-cell signalling and the regulation of cell proliferation suggesting that they may reflect the induction of proliferation in the ConA activated PBLs [58].
Suppression 1
Extracellular matrix
There are relatively few genes in the Suppression 1 cluster but one theme is the suppression of extracellular matrix proteins involved in anchoring cells to other cells and to the extracellular matrix. This is consistent with the induction of a proliferative response in activated PBLs. The proteins encoded by these genes include: THBS (Thrombospondin), an adhesive glycoprotein mediating cell-cell and cell-matrix interactions; ITGA5 (Integrin Alpha chain A5), an adhesive component of the extracellular matrix; and M160 (CD163 Antigen B), a member of the scavenger receptor (SRCR) superfamily.
Signalling
A second theme relates to the suppression of two genes whose encoded proteins are involved in G-protein signalling. RASGRP2 (RAS guanyl releasing protein 2 isoform 1) activates small GTPases involved in intracellular signal transduction such as RAS and its counter regulatory factor RAP1 [59]. The latter two proteins are components of the MAPK pathway, which is intimately involved in lymphocyte activation [44]. Suppression of RASGRP2 may be attenuating the RAS signalling pathway and hence lymphocyte activation (i.e. promoting homeostasis) by directly decreasing the activity of RAS or by activating RAP1 which in turn inhibits RAS. Consistent with this RGS14 (Regulator of G-protein Signalling 14) attenuates the signalling activity of G-proteins such as RAS [60].
Apoptosis regulator
This cluster also contains IL7R (Interleukin 7 Receptor) whose functions involve the inhibition of apoptosis during differentiation and activation of lymphocytes. The suppression of the expression of this gene may be facilitating apoptosis in the activated PBLs.
Overall, the differentially expressed genes in ConA activated bovine PBLs reflect the emphasis on the expression of a range of secreted intercellular signalling proteins, which are regulated by a dynamic balance between intracellular proteins promoting activation and proliferation and those that modulate or attenuate this process. Many of the genes identified in the clusters have also been noted as inducible genes in murine and human immune cells thereby attesting to the functional similarities of the immune systems of mammalian species and the capabilities of the microarray.
In conclusion, the Bovine Innate Immune Microarray has revealed details of the many gene networks that are activated in a model of bovine lymphocyte activation and it provides a powerful tool for examining the innate immune responses in bovine and ovine tissues challenged with bacteria (eg. mastitis), parasites (eg. intestinal Helminths) and a wide range of viruses. The microarray also has the potential for examining the roles of innate immune related genes in normal physiological processes such as the tissue remodelling occurring during mammary tissue involution.
Methods
Generation of the Bovine Innate Immune Microarray
DNA probes printed on the Bovine Innate Immune Microarray came from two main sources: innate immune candidate genes identified from the literature and anonymous cDNA clones generated by subtractive normalisation of transcripts from a variety of 'challenged' bovine cells.
(i) Selection of innate immune candidate genes
An extensive search of the human, mouse and bovine immunobiology literature and databases was undertaken to create a set of well characterised candidate genes based on their function and evidence of expression during immune responses [61-64]. Plasmid clones corresponding to some of the candidate genes were identified in available cDNA libraries including: Meat Animal Research Center (MARC) 1–5 libraries [25,26]; and in-house ovine and bovine cDNA libraries that had been used for the production of ESTs [27,28]. (see Additional file 1). Selected clones were streaked on LB agar plates (100 μg/ml Ampicillin) and grown overnight at 37°C. Individual colonies from each clone were transferred into wells of 96-well plates containing 200 μl Terrific broth (12 g/l bacto-tryptone, 24 g/l bacto-yeast extract, 0.4% glycerol, 17 mM KH2PO4, 72 mM K2HPO4, 100 μg/ml Ampicillin) and grown overnight at 37°C. Replicates of the libraries were made from the overnight cultures and both replicate and master plates were stored in 40% glycerol at -80°C. Plasmid cDNA clones were not available for all of the selected candidate genes and therefore gene-specific primers were designed based on publicly available bovine, ovine and/or human sequences. If no bovine or ovine sequence information was available primers were designed in areas that were conserved across known mammalian species. Primer design and optimisation were carried out using Primer 3 [65]. Primer information is listed in Additional file 2. Database searches, alignments and sequence analyses were performed with the aid of the ANGIS [66], NCBI [67] and IBISS [68,69]. databases. Bovine transcripts that had not been annotated were identified by BLAST searches against the orthologous human or mouse sequences [70].
(ii) Cell lines used for subtracted and normalised cDNA libraries
Primary bovine mammary epithelial cells (bMEC) were kindly donated by Dr Paul Sheehy (Sydney University, NSW) [21]. An immortalised bovine mammary epithelial cell line (MAC-T) was provided by Dr Kevin Nicholas (University of Melbourne, VIC) [71], while an immortalised bovine macrophage cell line (BoMac) was donated by Dr Timothy Doran (CSIRO Livestock Industries, VIC) [72]. Peripheral blood lymphocytes (PBL) were isolated from blood collected from healthy Hereford cattle and healthy Merino sheep housed at the Queensland Department of Primary Industries, Animal Research Institute (Yeerongpilly, Qld). Subtracted and normalised cDNA libraries were constructed using stimulated and unstimulated populations of these cell types. Bovine PBL were stimulated with ConA while bMEC, BoMAC and MAC-T cell lines were stimulated with LPS.
(iii) Cell culture conditions
bMEC were established on collagen type 1 (calf skin) (Sigma Chemical Co., St. Louis, MO) in a composite Medium 199/Hams F12 medium (Invitrogen, Carlsbad, CA) supplemented with 45 mM sodium bicarbonate, 4 mM sodium acetate, 18 mM Hepes, 20% horse serum, 5% foetal calf serum (FCS), 100 U/ml penicillin, 100 μg/ml streptomycin, 100 μg/ml kanamycin (Invitrogen), 500 μg/ml insulin, 100 μg/ml cortisol and 1 μg/ml EGF (mouse) (Sigma Chemical Co.). MAC-T were cultured in Dulbecco's modified Eagles's medium (DMEM) supplemented with 5% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin and 100 μg/ml kanamycin (Invitrogen). BoMac were established on a FCS treated flask in RPMI 1640 supplemented with 10% FCS, 4 mM sodium pyruvate, 50 mM HEPES, 100 U/ml penicillin, 100 μg/ml streptomycin and 100 μg/ml kanamycin (Invitrogen). All cell lines were cultured at 37°C in a humidified atmosphere of 5% CO2.
(iv) Collection and culture of pripheral blood lymphocytes (PBLs)
Peripheral blood lymphocytes from healthy adult Hereford cattle were isolated from defibrinated peripheral whole blood. Briefly, whole blood was centrifuged at 500 g for 30 min at 20°C. The buffy coat was isolated and layered over a Ficoll-Paque gradient (Amersham Pharmacia Biotech, England), followed by centrifugation at 500 g for 30 min at 20°C. PBLs were collected at the interface and washed twice in RPMI 1640 (Invitrogen, Carlsbad, CA) by centrifugation at 500 g for 10 min at 20°C. The PBLs were resuspended in RPMI 1640 supplemented with 10% FCS, 4 mM sodium pyruvate, 50 mM HEPES, 100 U/ml penicillin, 100 μg/ml streptomycin and 100 μg/ml kanamycin (Invitrogen) and cultured in 150 cm2 flasks (Sarstedt, Nümbrecht, Germany) with or without ConA (Sigma Chemical Co., St. Louis, MO) at a final concentration of 5 μg/ml. The cells were harvested for subsequent RNA extraction after incubation for 24 h.
(v) Stimulation of cells in culture
bMEC, MAC-T and BoMac were cultured in 150 cm2 flasks (Sarstedt, Nümbrecht, Germany), as previously described, until 80% confluence. In addition, MAC-T cultures were grown on collagen type 1 (calf skin) (Sigma Chemical Co.) in DMEM/10% FCS media supplemented with 500 μg/ml insulin, 100 μg/ml cortisol and 1 μg/ml EGF (mouse) (Sigma Chemical Co.). All cultures were then washed once in PBS before subsequent addition of fresh media with or without LPS (phenol extracted from Escherichia coli serotype O55:B5; Sigma Chemical Co. Cat# L2880) at a final concentration of 50 μg/ml. After incubation of the cells for 24 h, they were washed with PBS before being harvested using a cell scraper for subsequent RNA extraction.
(vi) RNA isolation, cDNA synthesis and cDNA library production
Total RNA was extracted from cells using an RNeasy Midi Kit (QIAGEN, Basel, Switzerland). The total RNA prepared from each sample was treated twice with DNase I (on column, QIAGEN DNase I and after elution (Ambion DNase I)) to minimise the presence of genomic DNA. mRNA was purified from each sample using a MicroPoly(A) Pure mRNA Puification Kit (Ambion). Total RNA and mRNA was quantified by spectrophotometric measurements at 260 nm and 280 nm and its purity and integrity verified by the OD260/OD280 ratio (>1.8) and by visualisation on a denaturing gel. mRNA purity was also analysed on an Agilent Bioanalyser (Agilent). cDNA synthesis was undertaken with 2 μg of isolated mRNA per sample using MMLV Superscript III reverse transcriptase (Invitrogen) and the Clontech cDNA synthesis primer from the PCR-Select cDNA Subtraction Kit (Clontech). The PCR-Select cDNA Subtraction Kit (Clontech) was used to generate subtracted, normalised cDNA pools for each of the different cell types and activation states, which were subsequently cloned into pGEM-T (Invitrogen) and cultured in XL1-blue E. coli. The cDNA libraries were grown overnight on LB agar plates (100 μg/ml ampicillin). Colonies were picked at random by hand from each library and were transferred into wells of 96-well plates that contained 200 μl Terrific broth and grown overnight at 37°C. The number of colonies picked from each library is listed in Table 1. Replicates of the libraries were made from the overnight cultures and both replicate and master plates were stored in 40% glycerol at -80°C.
(vii) Validation of subtracted and normalised cDNA libraries
A random selection of 50 clones from each cDNA library was sequenced to ensure adequate library quality. Clones were sequenced using M13 universal forward primer and the ABI Prism® BigDye terminator sequencing mix 3.1 (Applied Biosystems, USA). The ESTs were screened for vector and E. coli sequence contamination and quality clipped using the bioinformatic Staden package 'pregap4' with a cut-off of no more than 10 ambiguous bases in any window of 100 bases [73,74]. The clipped EST sequences have been submitted to the Genbank EST database accession numbers DT319147-DT319651[67]. Annotations were added to the clipped EST sequences by comparing them to the Genbank non-redundant and Human RefSeq nucleotide and amino acid databases using BLASTN and BLASTX [75,70]. Gene names were assigned to transcripts only if the Expect score (E-value) confidence limit was less than e-10. The level of redundancy within each cDNA library and across the cDNA libraries was also analysed by comparing the number of unique gene sequences against the number randomly selected sequenced clones.
(viii) Control elements
A panel of control elements, both positive and negative, were generated for incorporation into the microarray. Additional control elements were based on β-actin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) included size range products, representations of different regions of these transcripts and different printing quantities. These elements were used to monitor the efficiency of printing and to determine if any of the element characteristics has an effect the fluorescent signal produced by probe hybridisation. The Lucidea Universal Scorecard (Amersham Bioscience, UK) was included on the microarray and used to monitor the efficiency of probe production, probe hybridisation and microarray scanning. Details of each product are shown in Table 4. Control elements were verified by sequencing.
Table 4 Summary of bovine and S. aureus control elements
Control Element Name Reference Accession Number3 Start Position (bp) Finish Position (bp)
β-actin 200 AY141970 1675 1804
β-actin 5001 AY141970 1479 1804
β-actin 750 AY141970 1177 1804
β-actin 1000 AY141970 914 1804
β-actin 1500 AY141970 576 1804
β-actin 2000 AY141970 25 1804
β-actin 3'-region2 AY141970 1273 1822
β-actin Mid-region2 AY141970 784 1273
β-actin 5'-region2 AY141970 49 575
GAPDH 200 TC289232 1111 1347
GAPDH 5801 TC289232 771 1347
GAPDH 750 TC289232 575 1347
GAPDH 1350 TC289232 1 1347
GAPDH 3'-region2 TC289232 409 876
GAPDH 5'-region2 TC289232 59 429
S. aureus GAPDH2 NC_002745 832860 833305
S. aureus 30S ribosomal protein2 NC_002745 1513891 1514573
1 Denotes elements printed in a dynamic range of 25, 50 100, 200 and 400 pg per spot. All other elements were printed at 200 pg per spot.
2 Control elements not available in cDNA libraries were amplified using gene specific primers. Primer sequences are listed in Additional file 2.
3 Genbank sequence (AY or NC) or TIGR cluster (TC).
β-actin control elements included a range of fragments of 200, 500, 750, 1000, 1500 and 1900 bp in size, with each fragment originating from the 3'-end of the transcript. Three fragments, each of approximately 500 bp, spanning three independent regions of the β-actin transcript were also generated. GAPDH elements were similarly generated including a size range of 200, 580, 750 and 1350 bp and three fragments of approximately 400 bp representing unique regions of the transcript.
The control element DNA concentration was monitored by measuring the absorbance at 260 nm and 280 nm to achieve a final printing concentration of 200 ng/μl. The printing concentrations of 500 bp β-actin and GAPDH fragments (indicated in Table 4) were prepared in a dynamic range of 25, 50, 100, 200 and 400 ng/μl. Therefore, each element printed in this range contains 25 – 400 pg of DNA per spot, assuming a pin deposition volume of 1 nl for MicroSpot 2500 pins (Rob Moore, personal communication).
The Staphylococcus aureus genes, GAPDH and 30S ribosomal protein S1 were specifically amplified from S. aureus DNA and included on the microarray. Negative control elements were also included to monitor the signal from printing buffer, plasmid vector sequence. The plasmid vector controls include DNA from pGEM-T (Promega, USA) and pCMVsport6 (Clontech). Flanking vector sequence amplified in conjunction with cDNA clone insert sequence was also incorporated as a negative control element. Sequence information for the all control elements are shown in Table 4.
(ix) Production of microarray elements
E. coli lysates were prepared by adding 20 μl of overnight cell culture mix to 180 μl of water followed by heating at 95°C for 15 min. DNA inserts from the MARC cDNA libraries and the subtracted and normalised cDNA libraries were amplified by adding 5 μl of the cell lysate directly into PCR master mix (50 μM dNTP, 0.15 μM forward primer, 0.15 μM reverse primer, 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2 and 0.6 U of Taq F2 DNA polymerase (Fisher Biotech) in a total volume of 65 μl in 96 well plates. MARC1-4.Fwd (5'AGGAAACAGCTATGACCAT3') and MARC1-4.Rev (5'GTTTTCCCAGTCACG ACG3') primers were used for amplification from the MARC cDNA libraries. Clontech.P1 (5'TGCAGCGGCCGCCCGGGCAGGT3') and Clontech.P2R (5'AGCGTGGTCGCGGCCG AGGT3') primers were used for amplification from the anonymous cDNA libraries. Following a pre-heat step of 94°C for 4 min, PCR was performed in a PE 9700 Thermocycler (Perkin Elmer) using the following conditions: 94°C for 30 s, 50°C for 30 s and 72°C for 2 min (35 cycles) followed by 94°C for 30 s, 50°C for 30 s and 72°C for 5 min (1 cycle). A 5 μl sample of each PCR reaction was analysed by agarose gel electrophoresis (96 well gel, BioRad). Greater than 97% of reactions amplified a single PCR product with an average length of 450 bp. Clones with either no insert or multiple inserts were also prepared for printing onto the microarray and the specific elements annotated with this information. None of the elements containing more than one PCR product were included in the analyses. Element DNA was not screened for regions of low complexity or SINE or LINE elements.
To prepare the amplified cDNA for printing, 60 μl of isopropanol was added to each well, the plates were then inverted 5 times and incubated at -20°C for 10 min. After the precipitation step, plates were centrifuged at 4500 g for 60 min, washed with 70% ethanol and air-dried. Amplicons were resuspended in 20 μl of high purity water and transferred to one of nineteen 384-well microarray source plates using a BioMek 2000 liquid handling robot. An additional 384-well microarray source plate containing the control elements and the Lucidea Scorecard elements (Amersham Biosciences, UK) was also produced for a total of 20 microarray source plates, which were again allowed to air dry.
(x) Microarray printing
Microarray printing was performed at the Machines for Genes Laboratory, CSIRO Livestock Industries (Geelong, Vic). Of the 20 384-well source plates, 19 were printed in duplicate side-by-side, whilst the 384-well control plate was printed in duplicate at both the top and bottom of each printed block, giving a total of 16,128 printed elements. Elements were spotted on Corning UltraGAPS slides using a BioRobotics MicroGrid II TAS using MicroSpot 2500 pins. These pins produce a spot size of approximately 100 μm diameter with a pitch of 220 μm between spots. Each amplified DNA fragment was printed in duplicate adjacent to each other. The pin configuration was arranged in 4 columns and 12 rows yielding 48 subarrays of 19 × 19 spots. Elements were printed in a 150 mM sodium phosphate printing buffer (pH 8.5). The spots were then fixed to the slide by baking for two hours at 80°C. To ensure DNA spotting was uniform across the slides, samples were taken from each batch of microarrays and probed with Panomer™ 9 random oligodeoxynucleotide conjugated with Alexa Fluor® 532 (Molecular Probes, Invitrogen). Reproducibility was visually assessed between spot duplicates and between slides within a printing batch.
cDNA synthesis and labelling
20 μg of total RNA per dye channel per array was reverse transcribed with Superscript III (Invitrogen) in the presence of 2-aminoallyl-dUTP (Sigma Chemical Co.) using both oligo-dT18 (2 μg) and pd(N)6 random hexamer (1 μg) (Amersham Bioscience, UK) to prime cDNA synthesis. First strand cDNAs were purified using the QIAGEN PCR purification kit (QIAGEN) and subsequently labelled using n-hydroxysuccinate (NHS)-derivatized Cy3 and Cy5 dyes (Amersham Biosciences, UK). Labelled cDNAs were purified to remove unincorporated dyes then dried to ~1.0 μl in a vacuum desiccator.
Microarray hybridisation
Microarray hybridisation was performed essentially as per Lehnert et al. 2004 with some modifications [76]. Microarray slides were treated in a pre-hybridisation solution (4 × standard saline citrate (SSC), 0.1% N-lauroyl sarcosine (SDS), 50% formamide) for 30 min at room temperature. Labelled cDNAs were resuspended in 60 μl (final volume) of hybridisation buffer (15.5 μg Human CotI DNA (Invitrogen), 20 μg PolyA (Sigma), 4 × SSC, 0.1% SDS, 50% formamide), pre-warmed to 95°C for 3 min then held at 44°C until applied to the slides. Hybridisation was performed in the dark at 44°C for 16 h in sealed ArrayIt™ Hybridization Cassettes (TeleChem International, Inc.) submerged in a waterbath. Following hybridisation, three washes were applied to the slides: 2 × SSC, 0.1% SDS for 15 min, pre-warmed to 44°C; 0.2 × SSC for 15 min; and 0.06 × SSC for 5 min. Slides were washed once in high purity water to remove any remaining salt and dried in a centrifuge at 40 g for 5 min.
Data acquisition
Dried slides were scanned immediately using GenePix™ 4000 array scanner. GenePix™Pro software version 5.0 (Axon Instruments Inc.) was then used to process array images, align spots, integrate robot-spotting files with the microarray image, and to export reports of spot intensity data. Slides were visually examined and spots with irregular morphology were excluded from data analysis. The final report was retrieved as raw spot intensities in tab-delimited files, compatible with Microsoft Excel and VCE analysis programs. Raw data for the microarray experiments reported herein are stored in the GenEx database and can be accessed on request [77].
Microarray statistical analysis
Statistical analysis of the microarray data was undertaken using background corrected mean signal intensities from each dye channel. Raw data were subjected to a series of quality measures before being included in further analyses. In brief, each data point was required to be flagged "present" in the GenePix 6.0 software, have a mean to median signal ratio greater than 0.85 and have a signal to noise ratio greater than zero. A mixed-model ANOVA was applied to the data as this type of analysis allows full utilisation of the information available, with multiple factors and a hierarchy of sources of variation [78-82]. The model predicts the percentage of differentially expressed elements by calculating the proportion of variation arising from the sample treatment, as opposed to the variation introduced by experimental effects [3]. Restricted maximum likelihood (REML) estimates of variance components and best linear unbiased predictions (BLUP) were obtained using the VCE software [83]. Differentially expressed genes were identified using the EMMIX software for model-based clustering using mixtures of normal distributions [84,85]. Elements determined to be differentially expressed were further analysed in GeneSpring 6.1 (Silicon Genetics, Redwood City, CA) by a K-means clustering analysis using default parameters.
qRT-PCR validation
Quantitative real-time PCR (qRT-PCR) of selected transcripts was used to validate the expression profile that was observed in the microarray analyses. Transcripts were selected as representatives of a cluster showing a specific expression profile. qRT-PCR was performed using the same total RNA sample as was used in microarray analysis to ensure the results from the two technologies could be compared directly. Primer pairs for each selected transcript were designed from clone sequences using Primer3 software as described above. The measurements were performed using the Sybr Green system in an ABI Prism 7900 Sequence Detection System (PE Applied Biosystems, Foster City, CA) [21]. Briefly, a constant amount of cDNA (derived from 5 ng of total RNA) was used for each qRT-PCR measurement and four technical replicates were performed for each gene, one of which was a reference gene. This allows quantification of the target gene in one sample relative to that in another (the calibrator) using the "2-ΔΔCt method" of calculating fold change in gene expression [86]. The procedure relies on a common and constant reference gene in all samples. The reference gene RPLP0 was used for all calculations as it showed no change in expression upon treatment of cells. For all qRT-PCR measurements, the abundance of each transcript was measured relative to that of RPLP0. In order to present the relative changes in gene expression measured in qRT-PCR in the same scale as that found in microarray analysis the quantity of a transcript in one biological sample or state relative to that in a calibrating sample was expressed as log2 (fold change) in gene expression. The calibrator used in this experiment was unstimulated bovine PBLs.
The log2 (fold change) was calculated for both the microarray and qRT-PCR data. Microarray data were calculated as log2 (signal at time t) minus log2 (signal at time 0). qRT-PCR log2 (fold change) was calculated as Ct time 0 minus Ct time t, without normalisation against reference genes. Using this method a log2 (fold change) of 2 is equal to a doubling of transcript expression relative to the reference sample, in this case, unstimulated cells.
Microarray reproducibility and sensitivity
Microarray reproducibility was determined by comparing the signal from six independent slides where the hybridisation samples were identical. PBLs from a healthy adult Hereford steer (ARI, Yeerongpilly), were isolated and cultured for 24 h both with and without ConA (5 μg/ml). Total RNA was extracted and 160 μg of RNA from each treatment was used to produce 2-aminoallyl-labelled cDNA. Equal amounts of cDNA from each treatment were labelled with Cy3 and Cy5 to facilitate dye swap experiments. The hybridisation mixture containing Cy3-labelled cDNA from ConA stimulated PBLs and Cy5-labelled cDNA from unstimulated PBLs was split across three independent slides. The reciprocal hybridisation mixture was also split across three slides. A schematic diagram of the experimental design is shown in figure 3(a).
The signal ratios of ConA signal/unstimulated signal were calculated for all elements on each microarray. The correlation co-efficients between the set of signal ratios from each microarray were determined in a pair-wise manner. The mean correlation co-efficient was determined and represents the mean correlation in signal ratios observed between data from any two microarray slides. The mean correlation co-efficient was also determined for the three microarrays hybridised with Cy3-labelled cDNA from ConA stimulated PBLs and Cy5-labelled cDNA from unstimulated PBLs, the average correlation was also determined for the microarrays where the samples were labelled with opposite dyes.
The Lucidea Universal Scorecard RNA mix (Amersham Bioscience, UK) was included on one slide to assess the limit of detection of the Bovine Innate Immune Microarray. Individual transcripts within the Lucidea RNA mix are present in a range of concentrations. The Lucidea RNA mix was included in the cDNA synthesis reaction at 4 μl per 20 μg of sample total RNA. This generated a concentration gradient of specific Lucidea transcripts of 0.1, 0.3, 1, 3, 10, 30, 100, 1000 and 3000 pg per μg of sample total RNA. The Lucidea RNA mix was included in both the Cy3 and Cy5 labelling reactions.
Time course of differential gene expression during ConA stimulation of PBLs
The Bovine Innate Immune Microarray was used to analyse gene expression profiles of transcripts in bovine PBLs in response to stimulation with ConA over a time course of 24 h. PBLs were isolated from a healthy adult Hereford steer (ARI, Yeerongpilly). Transcript expression in bovine PBLs stimulated with ConA (5 μg/ml) for 0.5, 2, 4, 8 and 24 h was determined using control PBLs (unstimulated) as the common reference. Dye swaps were included for each of the five time points. The microarray experimental design is depicted in figure 5(a).
Analysis of cross-species hybridisation
A direct comparison of gene expression in equivalent ovine and bovine samples was undertaken. PBLs were isolated from a healthy adult Hereford steer and a healthy adult Merino sheep (ARI, Yeerongpilly). Total RNA was extracted from un-stimulated ovine and bovine PBLs, reverse transcribed using Superscript III, and labelled with Cy3 or Cy5 as described above. Ovine, bovine and a 1:1 mixture of ovine and bovine labelled cDNA were applied to the Bovine Innate Immune Microarray in an All-Pairs design using dye swaps for each comparison as depicted in figure 8(a).
Authors' contributions
LD was responsible for PBL cDNA library construction, microarray experiments, Genespring analysis, and manuscript preparation; LD and TV made the BoMAC, MAC-T and PMEC cDNA libraries; CG maintained cell culture lines; CG, YS and TV designed qRT-PCR assays; CG, YS, TV and LD amplified elements for microarray printing; TR performed statistical analysis; SM provided bioinformatics support; YW and KB developed the original microarray hybridisation protocol; RT conceived the study, and participated in its design, coordination, analysis and writing. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Defined Gene. This file lists the details of the bovine and ovine candidate genes and controls selected for inclusion on the Bovine Innate Immune Microarray. The list includes Genbank accession numbers and BlastN results for each sequence.
Click here for file
Additional File 2
Gene Specific Primers. This file lists the primer sequences for candidate genes and controls selected for inclusion on the Bovine Innate Immune Microarray. The list includes the Genbank accession numbers for the bovine, human or ovine sequence used as a basis for primer design for each gene.
Click here for file
Acknowledgements
We would like to acknowledge Rob Moore (CSIRO Livestock Industries, Geelong) for printing the Innate Immune Microarray, Queensland Department of Primary Industries Animal Research Institute, (Yeerongpilly), for supply of bovine and ovine blood and CSIRO Livestock Industries colleagues for their critique and input into the candidate gene list. This work was supported by the Co-operative Research Centre for Innovative Dairy Products.
==== Refs
Chitko-McKown CG Fox JM Miller LC Heaton MP Bono JL Keen JE Grosse WM Laegreid WW Gene expression profiling of bovine macrophages in response to Escherichia coli O157:H7 lipopolysaccharide Dev Comp Immunol 2004 28 635 45 15177116 10.1016/j.dci.2003.10.002
Moody DE Zou Z McIntyre L Cross-species hybridisation of pig RNA to human nylon microarrays BMC Genomics 2002 3 27 12354330 10.1186/1471-2164-3-27
Moser RJ Reverter A Kerr CA Beh KJ Lehnert SA A mixed-model approach for the analysis of cDNA microarray gene expression data from extreme-performing pigs after infection with Actinobacillus pleuropneumoniae J Anim Sci 2004 82 1261 71 15144065
Tao W Mallard B Karrow N Bridle B Construction and application of a bovine immune-endocrine cDNA microarray Vet Immunol Immunopathol 2004 101 1 17 15261689 10.1016/j.vetimm.2003.10.011
Burton JL Madsen SA Yao J Sipkovsky SS Coussens PM Agger JF Toft N An Immunogenomics Approach to Understanding Periparturient Immunosuppression and Mastitis Susceptibility in Dairy Cows Acta Veterinaria Scandinavica, Supplementum 2003 71 88
Wang Z Dooley TP Curto EV Davis RL VandeBerg JL Cross-species application of cDNA microarrays to profile gene expression using UV-induced melanoma in Monodelphis domestica as the model system Genomics 2004 83 588 99 15028282 10.1016/j.ygeno.2003.10.007
Liu HC Cheng HH Tirunagaru V Sofer L Burnside J A strategy to identify positional candidate genes conferring Marek's disease resistance by integrating DNA microarrays and genetic mapping Anim Genet 2001 32 351 9 11736805 10.1046/j.1365-2052.2001.00798.x
Coussens PM Colvin CJ Rosa GJ Perez Laspiur J Elftman MD Evidence for a novel gene expression program in peripheral blood mononuclear cells from Mycobacterium avium subsp. paratuberculosis-infected cattle Infect Immun 2003 71 6487 98 14573671 10.1128/IAI.71.11.6487-6498.2003
Everts RE Band MR Liu ZL Kumar CG Liu L Loor JJ Oliveira R Lewin HA A 7872 cDNA microarray and its use in bovine functional genomics Vet Immunol Immunopathol 2005 105 235 45 15808303 10.1016/j.vetimm.2005.02.003
Coussens PM Pudrith CB Skovgaard K Ren X Suchyta SP Stabel JR Heegaard PM Johne's disease in cattle is associated with enhanced expression of genes encoding IL-5, GATA-3, tissue inhibitors of matrix metalloproteinases 1 and 2, and factors promoting apoptosis in peripheral blood mononuclear cells Vet Immunol Immunopathol 2005 105 221 34 15808302 10.1016/j.vetimm.2005.02.009
Suchyta SP Sipkovsky S Kruska R Jeffers A McNulty A Coussens MJ Tempelman RJ Halgren RG Saama PM Bauman DE Boisclair YR Burton JL Collier RJ DePeters EJ Ferris TA Lucy MC McGuire MA Medrano JF Overton TR Smith TP Smith GW Sonstegard TS Spain JN Spiers DE Yao J Coussens PM Development and testing of a high-density cDNA microarray resource for cattle Physiol Genomics 2003 15 158 64 13130080
Affymetrix
National Mastitis Council Current Concepts of Bovine Mastitis 1996 Madison, WI: National Mastitis Council
Erskine RJ Eberhart RJ Hutchinson LJ Spencer SB Campbell MA Incidence and types of clinical mastitis in dairy herds with high and low somatic cell counts J Am Vet Med Assoc 1988 192 761 5 3281923
Sutra L Poutrel B Virulence factors involved in the pathogenesis of bovine intramammary infections due to Staphylococcus aureus J Med Microbiol 1994 40 79 89 8107066
Aungier SP Austin FH A study of the efficacy of intramammary antibiotics in the treatment of clinical mastitis Br Vet J 1987 143 88 90 3548878
Axford RFE Breeding for disease resistance in farm animals 2000 Wallingford: CABI Publishers
Burton JL Erskine RJ Immunity and mastitis. Some new ideas for an old disease Vet Clin North Am Food Anim Pract 2003 19 1 45v 12682934 10.1016/S0749-0720(02)00073-7
Mwangi EK Stevenson P Ndung'U JM Stear MJ Reid SW Gettinby G Murray M Studies on host resistance to tick infestations among trypanotolerant Bos indicus cattle breeds in east Africa Ann N Y Acad Sci 1998 849 195 208 9668465
Wells CA Ravasi T Faulkner GJ Carninci P Okazaki Y Hayashizaki Y Sweet M Wainwright BJ Hume DA Genetic control of the innate immune response BMC Immunol 2003 4 5 12826024 10.1186/1471-2172-4-5
Strandberg YC Gray C Vuocolo T Donaldson LA Tellam R Lipopolysaccharide and lipoteichoic acid induce different innate immune responses in bovine mammary epithelial cells Cytokine 2005 15882946
Jubert AV Hersh EM McBride CM Responses of appendix and peripheral blood lymphocytes to stimulation with various mitogens Surg Gynecol Obstet 1973 136 567 73 4540199
McEwan JC Crawford AM What the bovine genome project means to New Zealand Proceedings of the New Zealand Society of Animal Production 2004 64 86 91
Zhao L Hart S Cheng J Melenhorst JJ Bierie B Ernst M Stewart C Schaper F Heinrich PC Ullrich A Robinson GW Hennighausen L Mammary Gland Remodeling Depends on gp130 Signaling through Stat3 and MAPK J Biol Chem 2004 279 44093 100 15292206 10.1074/jbc.M313131200
Smith TP Grosse WM Freking BA Roberts AJ Stone RT Casas E Wray JE White J Cho J Fahrenkrug SC Bennett GL Heaton MP Laegreid WW Rohrer GA Chitko-McKown CG Pertea G Holt I Karamycheva S Liang F Quackenbush J Keele JW Sequence evaluation of four pooled-tissue normalized bovine cDNA libraries and construction of a gene index for cattle Genome Res 2001 11 626 30 11282978 10.1101/gr.170101
Sonstegard TS Capuco AV White J Van Tassell CP Connor EE Cho J Sultana R Shade L Wray JE Wells KD Quackenbush J Analysis of bovine mammary gland EST and functional annotation of the Bos taurus gene index Mamm Genome 2002 13 373 9 12140684 10.1007/s00335-001-2145-4
Adelson DL Cam GR DeSilva U Franklin IR Gene expression in sheep skin and wool (hair) Genomics 2004 83 95 105 14667813 10.1016/S0888-7543(03)00210-6
Wang YH McWilliam SM Barendse W Kata SR Womack JE Moore SS Lehnert SA Mapping of 12 bovine ribosomal protein genes using a bovine radiation hybrid panel Anim Genet 2001 32 269 73 11683713 10.1046/j.1365-2052.2001.00791.x
Haugland RP Handbook of Fluorescent Probes and Research Chemicals 1996 6 Eugene, OR: Molecular Probes
Dudoit S Yang YH Callow MJ Speed TP Statistical Methods for Identifying Differentially Expressed Genes in Replicated cDNA Microarray Experiments Statistica Sinica 2002 12 111 139
Reverter A McWilliam SM Barris W Dalrymple BP A rapid method for computationally inferring transcriptome coverage and microarray sensitivity Bioinformatics 2005 21 80 89 15308544 10.1093/bioinformatics/bth472
Heid CA Stevens J Livak KJ Williams PM Real time quantitative PCR Genome Res 1996 6 986 94 8908518
Stein T Morris JS Davies CR Weber-Hall SJ Duffy MA Heath VJ Bell AK Ferrier RK Sandilands GP Gusterson BA Involution of the mouse mammary gland is associated with an immune cascade and an acute-phase response, involving LBP, CD14 and STAT3 Breast Cancer Res 2004 6 R75 91 14979920 10.1186/bcr753
Efrat S Kaempfer R Control of biologically active interleukin 2 messenger RNA formation in induced human lymphocytes Proc Natl Acad Sci U S A 1984 81 2601 5 6609365
Gygi SP Rochon Y Franza BR Aebersold R Correlation between protein and mRNA abundance in yeast Mol Cell Biol 1999 19 1720 30 10022859
Brenner S Johnson M Bridgham J Golda G Lloyd DH Johnson D Luo S McCurdy S Foy M Ewan M Roth R George D Eletr S Albrecht G Vermaas E Williams SR Moon K Burcham T Pallas M DuBridge RB Kirchner J Fearon K Mao J Corcoran K Gene expression analysis by massively parallel signature sequencing (MPSS) on microbead arrays Nat Biotechnol 2000 18 630 4 10835600 10.1038/76469
Velculescu VE Zhang L Vogelstein B Kinzler KW Serial analysis of gene expression Science 1995 270 484 7 7570003
Boon K Riggins GJ SAGE as a strategy to isolate cancer-related genes Methods Mol Biol 2003 222 463 79 12710705
Churchill GA Fundamentals of experimental design for cDNA microarrays Nat Genet 2002 32 490 5 12454643 10.1038/ng1031
Aggarwal BB Signalling pathways of the TNF superfamily: a double-edged sword Nat Rev Immunol 2003 3 745 56 12949498 10.1038/nri1184
Weil R Israel A T-cell-receptor- and B-cell-receptor-mediated activation of NF-kappaB in lymphocytes Curr Opin Immunol 2004 16 374 81 15134788 10.1016/j.coi.2004.03.003
Gurney AL Marsters SA Huang RM Pitti RM Mark DT Baldwin DT Gray AM Dowd AD Brush AD Heldens AD Schow AD Goddard AD Wood WI Baker KP Godowski PJ Ashkenazi A Identification of a new member of the tumor necrosis factor family and its receptor, a human ortholog of mouse GITR Curr Biol 1999 9 215 8 10074428 10.1016/S0960-9822(99)80093-1
Nocentini G Giunchi L Ronchetti S Krausz LT Bartoli A Moraca R Migliorati G Riccardi C A new member of the tumor necrosis factor/nerve growth factor receptor family inhibits T cell receptor-induced apoptosis Proc Natl Acad Sci U S A 1997 94 6216 21 9177197 10.1073/pnas.94.12.6216
Farooq A Zhou MM Structure and regulation of MAPK phosphatases Cell Signal 2004 16 769 79 15115656 10.1016/j.cellsig.2003.12.008
Cannons JL Schwartzberg PL Fine-tuning lymphocyte regulation: what's new with tyrosine kinases and phosphatases? Curr Opin Immunol 2004 16 296 303 15134778 10.1016/j.coi.2004.03.011
Calnan BJ Szychowski S Chan FK Cado D Winoto A A role for the orphan steroid receptor Nur77 in apoptosis accompanying antigen-induced negative selection Immunity 1995 3 273 82 7552993 10.1016/1074-7613(95)90113-2
Kim SO Ono K Tobias PS Han J Orphan nuclear receptor Nur77 is involved in caspase-independent macrophage cell death J Exp Med 2003 197 1441 52 12782711 10.1084/jem.20021842
Pawlinski R Pedersen B Kehrle B Aird WC Frank RD Guha M Mackman N Regulation of tissue factor and inflammatory mediators by Egr-1 in a mouse endotoxemia model Blood 2003 101 3940 7 12543866 10.1182/blood-2002-07-2303
Murphy LO MacKeigan JP Blenis J A network of immediate early gene products propagates subtle differences in mitogen-activated protein kinase signal amplitude and duration Mol Cell Biol 2004 24 144 53 14673150 10.1128/MCB.24.1.144-153.2004
Ziegler SF Ramsdell F Alderson MR The activation antigen CD69 Stem Cells 1994 12 456 65 7804122
Moser B Wolf M Walz A Loetscher P Chemokines: multiple levels of leukocyte migration control Trends Immunol 2004 25 75 84 15102366 10.1016/j.it.2003.12.005
Prohaszka Z Fust G Immunological aspects of heat-shock proteins-the optimum stress of life Mol Immunol 2004 41 29 44 15140573 10.1016/j.molimm.2004.02.001
Wynn TA Nicolet CM Paulnock DM Identification and characterization of a new gene family induced during macrophage activation J Immunol 1991 147 4384 92 1753106
Fragale A Tartaglia M Wu J Gelb BD Noonan syndrome-associated SHP2/PTPN11 mutants cause EGF-dependent prolonged GAB1 binding and sustained ERK2/MAPK1 activation Hum Mutat 2004 23 267 77 14974085 10.1002/humu.20005
Willis S Day CL Hinds MG Huang DC The Bcl-2-regulated apoptotic pathway J Cell Sci 2003 116 4053 6 12972498 10.1242/jcs.00754
Srivastava M Pollard HB Molecular dissection of nucleolin's role in growth and cell proliferation: new insights FASEB J 1999 13 1911 22 10544174
Davydov IV Bohmann D Krammer PH Li-Weber M Cloning of the cDNA encoding human C/EBP gamma, a protein binding to the PRE-I enhancer element of the human interleukin-4 promoter Gene 1995 161 271 5 7665092 10.1016/0378-1119(95)00271-7
Okamoto N Tezuka K Kato M Abe R Tsuji T PI3-kinase and MAP-kinase signaling cascades in AILIM/ICOS- and CD28-costimulated T-cells have distinct functions between cell proliferation and IL-10 production Biochem Biophys Res Commun 2003 310 691 702 14550257 10.1016/j.bbrc.2003.09.065
Clyde-Smith J Silins G Gartside M Grimmond S Etheridge M Apolloni A Hayward N Hancock JF Characterization of RasGRP2, a plasma membrane-targeted, dual specificity Ras/Rap exchange factor J Biol Chem 2000 275 32260 7 10918068 10.1074/jbc.M006087200
Kimple RJ Kimple ME Betts L Sondek J Siderovski DP Structural determinants for GoLoco-induced inhibition of nucleotide release by Galpha subunits Nature 2002 416 878 81 11976690 10.1038/416878a
Coussens PM Nobis W Bioinformatics and high throughput approach to create genomic resources for the study of bovine immunobiology Vet Immunol Immunopathol 2002 86 229 44 12007889 10.1016/S0165-2427(02)00005-3
Hernandez A Karrow N Mallard BA Maillard JC Pinard van der Laan MH Evaluation of Immune Responses of Cattle as a Means to Identify High or Low Responders and Use of a Human Microarray to Differentiate Gene Expression Genetics, Selection, Evolution 2003 35 S67 S81
Schwerin M Czernek-Schafer D Goldammer T Kata SR Womack JE Pareek R Pareek C Walawski K Brunner RM Application of disease-associated differentially expressed genes–mining for functional candidate genes for mastitis resistance in cattle Genet Sel Evol 2003 35 S19 34 12927078
Wells CA Ravasi T Sultana R Yagi K Carninci P Bono H Faulkner G Okazaki Y Quackenbush J Hume DA Lyons PA Continued discovery of transcriptional units expressed in cells of the mouse mononuclear phagocyte lineage Genome Res 2003 13 1360 5 12819134 10.1101/gr.1056103
Primer3
Australian National Genomic Information Service
National Centre for Biotechnology Information
IBISS: Interactive Bovine In Silico SNP Database
Hawken RJ Barris WC McWilliam SM Dalrymple BP An interactive bovine in silico SNP database (IBISS) Mamm Genome 2004 15 819 27 15520884 10.1007/s00335-004-2382-4
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 402 9254694 10.1093/nar/25.17.3389
Huynh HT Robitaille G Turner JD Establishment of bovine mammary epithelial cells (MAC-T): an in vitro model for bovine lactation Exp Cell Res 1991 197 191 9 1659986 10.1016/0014-4827(91)90422-Q
Stabel JR Stabel TJ Immortalization and characterization of bovine peritoneal macrophages transfected with SV40 plasmid DNA Vet Immunol Immunopathol 1995 45 211 20 7676607 10.1016/0165-2427(94)05348-V
Staden Package WWW site
Staden R The Staden sequence analysis package Mol Biotechnol 1996 5 233 41 8837029
Benson DA Karsch-Mizrachi I Lipman DJ Ostell J Wheeler DL GenBank: update Nucleic Acids Res 2004 D23 6 14681350 10.1093/nar/gkh045
Lehnert SA Wang YH Byrne KA Development and application of a bovine cDNA microarray for expression profiling of muscle and adipose tissue Australian Journal of Experimental Agriculture 2004 44 1127 1133 10.1071/EA03238
CSIRO Livestock Industries Gene Expression database
Reverter A Barris W McWilliam S Byrne KA Wang YH Tan SH Hudson N Dalrymple BP Validation of alternative methods of data normalization in gene co-expression studies Bioinformatics 2005 21 1112 20 15564293 10.1093/bioinformatics/bti124
Reverter A Wang YH Byrne KA Tan SH Harper GS Lehnert SA Joint analysis of multiple cDNA microarray studies via multivariate mixed models applied to genetic improvement of beef cattle J Anim Sci 2004 82 3430 9 15537761
Kerr MK Martin M Churchill GA Analysis of variance for gene expression microarray data J Comput Biol 2000 7 819 37 11382364 10.1089/10665270050514954
Wolfinger RD Gibson G Wolfinger ED Bennett L Hamadeh H Bushel P Afshari C Paules RS Assessing gene significance from cDNA microarray expression data via mixed models J Comput Biol 2001 8 625 37 11747616 10.1089/106652701753307520
Cui X Churchill GA Statistical tests for differential expression in cDNA microarray experiments Genome Biol 2003 4 210 12702200 10.1186/gb-2003-4-4-210
VCE4
McLachlan GJ Bean RW Peel D A mixture model-based approach to the clustering of microarray expression data Bioinformatics 2002 18 413 22 11934740 10.1093/bioinformatics/18.3.413
McLachlan GL Peel D Basford KE Adams A The EMMIX software for the fitting of mistures of normal and t-components Journal of Statistical Software 2004 4
Livak KJ Schmittgen TD Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method Methods 2001 25 402 8 11846609 10.1006/meth.2001.1262
|
16176586
|
PMC1261263
|
CC BY
|
2021-01-04 16:32:46
|
no
|
BMC Genomics. 2005 Sep 22; 6:135
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-135
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-711616475710.1186/1471-2334-5-71Research ArticleEffects of type and level of training on variation in physician knowledge in the use and acquisition of blood cultures: a cross sectional survey Parada Jorge P [email protected] David N [email protected] Gordon D [email protected] Kevin B [email protected] Midwest Center for Health Services and Policy Research, Hines VA Hospital, Hines, IL, USA2 Department of Medicine-Loyola University Medical Center, and the Stritch School of Medicine-Loyola University Chicago, Maywood, IL, USA3 Department of Medicine John Stroger Hospital of Cook County and Rush Medical College, Chicago, IL, USA4 Center for Healthcare Studies, and the Division of the General Medicine Feinberg School of Medicine, Northwestern University, Chicago, IL, USA2005 15 9 2005 5 71 71 28 3 2005 15 9 2005 Copyright © 2005 Parada et al; licensee BioMed Central Ltd.2005Parada 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
Blood culture (BCX) use is often sub-optimal, and is a user-dependent diagnostic test. Little is known about physician training and BCX-related knowledge. We sought to assess variations in caregiver BCX-related knowledge, and their relation to medical training.
Methods
We developed and piloted a self-administered BCX-related knowledge survey instrument. Expert opinion, literature review, focus groups, and mini-pilots reduced > 100 questions in multiple formats to a final questionnaire with 15 scored content items and 4 covariate identifiers. This questionnaire was used in a cross-sectional survey of physicians, fellows, residents and medical students at a large urban public teaching hospital. The responses were stratified by years/level of training, type of specialty training, self-reported practical and theoretical BCX-related instruction. Summary scores were derived from participant responses compared to a 95% consensus opinion of infectious diseases specialists that matched an evidence based reference standard.
Results
There were 291 respondents (Attendings = 72, Post-Graduate Year (PGY) = 3 = 84, PGY2 = 42, PGY1 = 41, medical students = 52). Mean scores differed by training level (Attending = 85.0, PGY3 = 81.1, PGY2 = 78.4, PGY1 = 75.4, students = 67.7) [p ≤ 0.001], and training type (Infectious Diseases = 96.1, Medicine = 81.7, Emergency Medicine = 79.6, Surgery = 78.5, Family Practice = 76.5, Obstetrics-Gynecology = 74.4, Pediatrics = 74.0) [p ≤ 0.001]. Higher summary scores were associated with self-reported theoretical [p ≤ 0.001] and practical [p = 0.001] BCX-related training. Linear regression showed level and type of training accounted for most of the score variation.
Conclusion
Higher mean scores were associated with advancing level of training and greater subject-related training. Notably, house staff and medical students, who are most likely to order and/or obtain BCXs, lack key BCX-related knowledge. Targeted education may improve utilization of this important diagnostic tool.
==== Body
Background
"Of all the microbiologic procedures performed in the laboratory, few are as important as the prompt recovery of microorganisms from the blood."[1] – J.A. Washington, editor of Infectious Disease Clinics.
Per year, more than 200,000–250,000 blood stream infections occur in the United States, [2-5] and are the 10th leading cause of death in the United States. [6] Bloodstream infection is associated with crude mortality rates as high as 50% in certain populations. [6-9] Bacteria enter the bloodstream indirectly via the lymphatic system with extravascular infections and directly with intravascular infections, and may present as transient (as with procedures or manipulation of infected tissues or mucosal surfaces or as with meningitis, osteomyelitis or pneumonia), intermittent (as with an undrained abscess) or continuous (as with endocardidtis and endovascular infections) bacteremia. Thus, the yield of blood cultures are related to the underlying infectious process and may be of limited utility at times. However, clinical management of infectious diseases depends on the accurate identification of the causal microorganism and its antimicrobial susceptibility, unusual organisms may be identified that may not be adequately treated by routine empirical coverage. [10] Because blood cultures (BCXs) represent the "gold standard" for the diagnosis of blood stream infections, their timely and appropriate utilization play a pivotal role in antimicrobial therapy. [11]
Blood cultures are a highly user-dependent diagnostic test. Optimal BCX yield – highest sensitivity and highest specificity – critically depends not only on the nature of underlying infectious process, but also on technique and timing of specimen acquisition. Proper aseptic technique has been shown to decrease the rate of contaminants. [12,13] The timing of BCX acquisition in relation to rigors/fevers and antibiotic administration also impacts on BCX yield. [13] The volume of blood, number and type of BCX bottles collected, likewise impact on sensitivity/specificity. [1,12-14]
Blood culture use is often sub-optimal. Standards published by the American Society of Microbiology indicate that the rate of contaminant BCXs should not exceed 3%. [15] Nevertheless, many teaching hospitals have rates that exceed 6%. [15,16] It has been estimated that each contaminant BCX adds $4,500 in additional cost (due to additional diagnostic testing, increased length of stay, unnecessary medication use and associated adverse events). [15] In addition, failures to obtain BCXs represent failed opportunities for guiding antimicrobial therapy. For example, 43% of empiric vancomycin courses in our institution lacked appropriate cultures, impeding efforts to streamline subsequent antimicrobial therapy. [17] Rapid identification and susceptibility testing of bacterial BCX isolates is also associated with more timely and cost-effective antibiotic therapy in hospitalized patients. [18]
Duration of medical training has been shown to impact resource utilization and diagnostic test use. [19] In teaching institutions like ours, inexperienced house officers and medical students are responsible for most BCX ordering and acquisition. The purpose of this study was to assess the associations between the level or type of physician training and of BCX-related knowledge among a wide range of physicians and physicians in training. We also sought to determine whether self-reported BCX-related training is a reliable indicator of measurable BCX-related knowledge.
Methods
This is a cross sectional study using a self-administered survey instrument.
To develop a self-administered survey instrument to assess BCX-related knowledge and training, we began with a preliminary questionnaire containing > 100 evidence-based questions in multiple formats. Focus groups and mini-pilots were used to ensure face and content validity, and to guide an informal process of item reduction.
The prototype survey instrument used 32 knowledge-related multiple-choice and Likert scale questions. The BCX-knowledge questions covered various domains related to BCX use, including clinical indications for obtaining blood cultures, blood culture acquisition procedure, aseptic technique, optimal volume of blood for culture, number and timing of blood cultures, clinical indicators of BCX contamination, effect of antibiotic use on BCX utility, and usefulness of routine anaerobic cultures. All questions involved qualitative rather than quantitative judgments, thereby preventing bias in responses of caregivers for different patient populations (e.g., internists versus pediatricians).
We included four additional items to identify covariates of BCX-related knowledge. These included level of training, type of specialty training, and self-reported theoretical and practical BCX-related instruction. Sub-categories within each covariate group are delineated in Table 1.
Table 1 Selected Characteristics of Physicians.
Category of Training Number (Percent)
Level of Training* Attending 22 (9.2)
≥ PGY3 115 (48.1)
PGY2 19 (7.9)
PGY1 27 (11.3)
Student 17 (7.1)
Specialty/Type of Training† Infectious Diseases 19 (7.9)
Medicine 20 (8.4)
Emergency Medicine 72 (24.7)
Surgery 84 (28.9)
Family Practice 42 (14.4)
Obstetrics-Gynecology 41 (14.1)
Pediatrics 52 (17.9)
Self-Reported Theoretical Some 130 (44.7)
BCX-Related Instruction† Little 96 (33.0)
None 65 (22.3)
Self-Reported Practical Some 156 (53.6)
BCX-Related Instruction† Little 73 (25.1)
None 62 (21.3)
* Total number = 291, students included
† Total number = 239, students excluded from total
We recruited infectious diseases specialists from a public teaching hospital, a private university medical center, and Chicago-area infectious disease specialists in private practice to guide item reduction and to establish a reference standard. This convenience sample of 23 infectious diseases physicians was recruited at a local infectious diseases conference, and their responses were used to establish an expert opinion reference standard. This reference standard was meant to establish a clinically useful approximation of optimal BCX-related knowledge and was established by tabulating the question-by-question infectious diseases specialists' responses of greatest frequency. To overcome issues related to any discordance in expert opinion, only responses with high concordance (95%) responses were included to establish the final consensus opinion reference standard. Thus, of the 32 BCX-knowledge items, 15 (47%) demonstrated a 95% consensus opinion by infectious diseases specialists and were retained in the final questionnaire [see Additional file 1]. In an item-by-item analysis of the 15 items, agreement with the reference standard was scored as 1+ (plus one) and disagreement as 1- (minus one). We summed these items to provide an overall summary score for each subject that was converted to a 0–100 scale.
We defined an optimal knowledge standard as responses matching evidence-based responses derived from our previous literature review. Of note, we found that both the expert opinion and consensus opinion reference standards matched the independently determined evidence-based optimal knowledge standard for all 15 items in the final survey instrument.
We administered the survey to a convenience sample of attending physicians, house officers and medical students of a large urban public teaching hospital. Most questionnaires were administered at conferences of the target services; specialties that use BCX infrequently (e.g. radiology, anesthesiology, psychiatry, dermatology, etc.) were not offered survey participation. Survey participation was voluntary and anonymous and the number of persons who did not complete or return the study is unknown. Respondents were not permitted to look up answers, and on average the survey questionnaire took < 5 minutes to complete. All responders were classified by level of training (medical student, interns, junior residents, senior residents and fellows, and attending staff), as well as type of specialty training (medicine, emergency medicine, surgery, obstetrics-gynecology, pediatrics, family practice and infectious diseases). Of note, interns, junior and senior residents, as well as fellows were classified according to their post graduate year (PGY) of training, as PGY1, PGY2, PGY3, or PGY ≥ 4.
Data were entered into a computerized database and analyzed using SPSS® version 11.1 for Windows (SPPS Inc., Chicago, IL). Univariate analysis was conducted for each covariate. First we examined the mean item scores for each of the 15 scored items in the survey. Then, paired ANOVAs were used to examine the effect of level of training, type of specialty training, and self-reported theoretical and practical BCX-related instruction on the mean summary scores. Linear regression modeling was used to analyze the independent effects of these covariates on summary scores. Post Hoc testing included Bonferoni adjustment for multiple comparisons.
Approval for this study was granted by the Human Subjects subcommittee of the Institutional Review Boards of the study hospital and affiliate university. Participant confidentiality was maintained throughout the study. All members of the research team report no financial conflicts of interest with study participation.
Results
Two hundred and ninety respondents met inclusion criteria and appropriately completed the survey. Table 1 provides a breakdown of responder characteristics by level of training, type of specialty training, and self-reported BCX instruction.
Level of training was found to be strongly and positively associated with BCX-related knowledge, with a difference of more than 17 points in the mean scores of attending physicians and medical students (85 versus 67.7 points, respectively; p < 0.001), and intermediate mean scores for resident physicians that improved with each additional year of training (Table 2). Infectious diseases physicians' scores of BCX-related knowledge were far higher than those of physicians from other specialties (p < 0.001), among which internal medicine physicians scored the highest (Table 2). Higher levels of self-reported theoretical and practical BCX-related instruction were similarly correlated with higher scores for BCX-related knowledge (Table 2).
Table 2 Physician BCX Knowledge Summary Scores by Level of Training, Self-Reported Theoretical or Practical Blood Culture Related Training, and Type of Specialty Training.
Category Mean Score* Score STD† Range P-value
Level of Training
Attending‡ 85.00 12.00 53.33–100 —
≥ PGY3 81.10 11.20 40.00–100 0.04
PGY2 78.40 14.10 20.00–100 0.01
PGY1 75.40 10.83 53.33–100 ≤ 0.001
Student 67.70 15.63 26.67–100 ≤ 0.001
Specialty/Type of Training
ID‡ 96.07 6.07 80.00–100 — (≤ 0.001)§
Medicine 81.73 12.37 20.00–100 ≤ 0.001 (—)
Emergency Med 79.63 9.30 66.67–100 ≤ 0.001 (0.483)
Surgery 78.53 13.50 40.00–100 ≤ 0.001 (0.233)
Family Practice 76.47 9.47 53.33–100 ≤ 0.001 (0.095)
Ob-Gyn 74.40 9.50 60.00–100 ≤ 0.001 (0.015)
Pediatrics 74.00 8.33 60.00–100 ≤ 0.001 (0.008)
Students 67.70 15.63 26.67–100 ≤ 0.001 (≤ 0.001)
Self-Reported Theoretical Instruction
Some‡ 81.38 13.53 20.00–100 —
Little 78.26 12.77 53.33–100 0.081
None 73.33 14.95 26.67–100 ≤ 0.001
Self-Reported Practical Instruction
Some‡ 80.90 13.14 20.00–100 —
Little 78.72 13.32 53.33–100 0.241
None 72.13 14.78 26.67–100 0.001
* Scale of Mean Summary Scores = 0–100
† STD = Standard Deviation
‡ Comparison group for the paired ANOVA (see P-values), ID = Infectious diseases
§ (P-value using Medicine as the comparison group)
Level of training and type of specialty training accounted for most of the variability in scores when all covariates are entered into general linear regression modeling (Table 3). After controlling for level of training and type of specialty training, self-reported theoretical and practical BCX-related instruction largely ceased to account for statistically significant changes in scores. In contrast, after controlling for level of training, statistically significant differences in scores persisted for all services as compared to infectious diseases (p < 0.001, Table 2, infectious diseases specialists as the comparison group). General linear regression modeling revealed that the categories infectious diseases, obstetrics-gynecology, pediatrics, PGY1 and student accounted for most of the variability in scores (Table 3).
Table 3 Multivariate Analysis of Physician BCX Knowledge Summary Scores in Model Including Level of Training, Type of Specialty Training, and Self-Reported Theoretical and Practical Blood Culture Related Training.
Category P-value 95% Confidence Interval of Difference in Scores with Reference Group
Level of Training
Attending* — —
> PGY3 0.181 -2.385, 0.452
PGY3 0.808 -1.641, 1.279
PGY2 0.081 -2.805, 0.162
PGY1 0.016 -3.379, -0.350
Student < 0.001 -6.271, -3.224
Specialty/Type of Training
Infectious Diseases < 0.001 1.691, 5.305
Medicine* — —
Emergency Medicine 0.255 -2.867, 0.762
Surgery 0.180 -2.609, 0.492
Family Practice 0.099 -3.440, 0.299
Ob-Gyn 0.005 -4.405, -0.780
Pediatrics 0.015 -3.991, -0.439
Self-Reported BCX-related Training
Practical Training 0.347 -0.327, 0.925
Theoretical Training 0.529 -0.430, 0.835
* Excluded due to co-linearity
Examined item-by-item, all 15 questions were found to both match an evidence-based optimal knowledge standard and covered all previously identified relevant domains of BCX-related knowledge. Figure 1 details the item-by-item variability in services' responses to the 15 questions. Many items demonstrated similar high scores among respondents. Seven items had marked variability in responses; with the greatest number of incorrect answers they contributed the most strongly to the variability in overall scores.
Figure 1 Item-by-Item Performance (Mean Score) by Type of Specialty Training/Service (Note: Seven questions (qn) – 6, 8, 9, 10, 12, 13 and 15 – accounted for the greatest variability in responses).
Discussion
Our data suggests that substantial variation exists among physicians in BCX-related knowledge and that this knowledge is significantly associated with the level and type of training. Apart from infectious diseases physicians, we identified important deficiencies in BCX-related knowledge among all specialties and at all levels of training. Of particular concern was the finding that scores of BCX-related knowledge were lowest among interns and medical students who are most often responsible for ordering and obtaining blood specimens for testing. In addition, approximately one half of respondents reported little or no practical or theoretical instruction in BCX utilization (Table 1). It is notable that fellows and attendings with infectious diseases specialty training had far higher mean summary scores than all other providers (p < 0.001 for all comparisons). It appears that there may be a specific body of knowledge that is commonly acquired during even early stages of infectious diseases training. Together, these results suggest that educational interventions to increase BCX-related knowledge may be an effective strategy for improving utilization of this important diagnostic test.
These data also suggests that self-reported instruction in practical and theoretical BCX-related knowledge correlate with overall BCX-related knowledge. This effect is largely explained by other factors like level of training and type of specialty training. However, at the lower training levels (student and intern), perceived BCX instruction remain a statistically significant predictor of score and may be a reliable independent measure of overall BCX knowledge. This may have implications for medical education as this suggests a simple direct query about perceived BCX instruction may help identify trainees who may benefit from additional education in BCX use.
In addition, these data show that physicians at higher levels of training (attendings) have greater BCX-related knowledge than physicians at lower levels of training and medical students. Admittedly, this conclusion seems intuitively logical, yet to our knowledge this is the first study that specifically investigated this relationship in this area. Furthermore, it might not have been surprising to find a very different relationship; one where those most directly involved with BCX acquisition (interns and residents) scored higher than students (lacking experience) or attendings (distanced from direct experience). These findings highlight an interesting paradox: the providers most directly involved ordering and obtaining BCXs (house staff and medical students) are the least knowledgeable about their indications and use. There appears to be a clear mismatch of these providers and those who are most knowledgeable about BCXs (attendings). We hypothesize this may contribute to sub-optimal BCX use, and wonder if better education early in training may improving knowledge and perception of BCX utility. As seven of the fifteen questions were found to have much greater variability in their responses (Figure 1), it is possible that an even more user friendly, reduced item, questionnaire may be deployable. These questions focused on the timing of specimen acquisition with regard to symptoms, number of cultures, and volume of blood for optimal blood culture yield, as well querying the respondents' understanding of the underlying infectious processes' likelihood of producing high grade bacteremia (i.e., endovascular infections versus pneumonia or cellulits) and when most blood cultures convert from negative to positive.
We believe these findings show that an area-specific (BCX-related) knowledge survey can be used to identify and target providers who might most benefit from educational interventions. In addition, measures of BCX-related knowledge could be used to assess if BCX-related knowledge predicts appropriateness of BCX acquisition behavior. Most importantly, this model might be extended to other areas of medical training.
A number of study limitations are worth noting. This is a single centre study, so generalizability of these findings remains unproven. In our approach to data analysis, some infectious diseases attending responders served for both to establish the reference standard and were subsequently included in the database. Understandably, if scores were derived from a comparison to this reference standard, the high scores of the infectious diseases service can be construed to be a self-fulfilling prophecy. However, a comparison of reference standard responses (be it expert opinion or consensus opinion standards) to evidence-based optimal knowledge showed all reference standard responses matched evidence-based responses. Therefore, the reference standard does not introduce a bias away from evidence-based optimal responses. However, we believe it does serve to establish a clinically based practical scoring system. An examination of infectious diseases fellows' responses supports this view. Infectious diseases fellows' responses were not used to establish the reference standard. Yet, the infectious diseases fellows' mean summary score was 90.5. This is higher than any other category in any service or any other level of training (except infectious diseases attendings). We believe this reflects that certain medical training does lead to increased BCX-related knowledge. This increased knowledge is reflected in higher mean summary scores and a closer approximation of optimal evidence-based knowledge. Finally, the survey was administered to a convenience sample primarily captured at conferences of target services, and unfortunately the participant response rates are unknown. We recognize the potential bias that may be associated with selective sampling of responders, and while we are encouraged by the strength of our findings, expanded meticulous surveying at multiple centers are needed to confirm these findings. We also recognize differential motivation of respondents may influence our findings. Infectious disease trained respondents perhaps felt more compelled to perform on this survey than others. However, most physicians are competitive individuals and we expect that those disinclined to perform well would also be disinclined to complete the survey. We know all infectious disease participants complete and return the survey while not all participants from other groups did. Thus, it is more likely that a selection bias towards a subpopulation of more motivated and higher scoring participants would be amongst non-infectious disease trained respondents.
Additional work is recommended. It would be important to investigate the link between BCX-related knowledge and actual physician behavior with regard to appropriate BCX acquisition. In addition, of great importance in this age of significantly decreased physician phlebotomies would be to broaden the focus to include non-physician phlebotomy and nursing staff which more and more commonly obtain BCXs on physicians orders. Appropriate blood culture acquisition could improve the sensitivity and specificity of BCXs. Improved diagnostic outcomes could be achieved through increased indicated phlebotomy, reduced unnecessary phlebotomy, as well as improved aseptic technique and decreased number of contaminants. Potentially, this could contribute to reductions in empiric antibiotic use, cost of treatment, and unnecessary use of hospital resources (including syringes, culture bottles, laboratory staff and equipment) and shorten hospital length of stay.
Finally, it there is the opportunity to extend the concepts of this work to other areas of medical care. This approach can be applied to other medical domains, as well as non-physician health care providers, to help better understand the relationship of professional training and the use of diagnostic or therapeutic interventions; or to screen persons for targeted educational interventions.
Conclusion
Our data suggests that, as expected, level of training and type of specialty training is related to BCX-related knowledge. However, it appears that knowledge is not similar across physician specialties. Specifically, average knowledge scores of providers from medical, surgical and emergency medicine services were higher than those of providers from pediatrics, obstetrics-gynecology, and family medicine. At the higher levels of training, where the type of specialty training exerts greater impact, self-reported BCX-related instruction does not act as an independent predictor of BCX-related knowledge. However, at the lower training levels (student and intern), perceived BCX instruction may be a reliable independent measure of overall BCX knowledge. As such, it may serve as a screen to identify trainees who might benefit from targeted educational interventions. Improvement in knowledge deficits around BCX use may provide opportunities for better use of this important diagnostic tool, and ultimately improved patient outcomes.
Abbreviations
BCX – Blood Cultures
PGY – Post Graduate Year
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
All authors were responsible for study conception and design. JPP, GDS, DNS undertook the literature review. JPP collected the data and was responsible for the data management and analysis. All authors contributed to the interpretation of the study findings. JPP wrote the first draft of the paper and all the authors contributed to further drafts and critical review of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional File 1
Survey Instrument. The BCX-related Knowledge Survey Instrument (includes the 15 scored items, and the four co-variate identifiers).
Click here for file
Acknowledgements
We would like to thank Drs. Hem Aggarwal and Rose Johnson for their valuable aid in data collection and processing, as well as Dr. Mehr Iqbal for her aid in editing the manuscript.
==== Refs
Washington JA II Blood cultures: principles and technique Mayo Clinic Proc 1975 50 91 8
Washington JA Ilstrup DM Blood cultures: issues and controversies Rev Infect Dis 1986 8 792 802 3538319
Endimiani A Tamborini A Luzzaro F Lombari G Toniolo A A two-year analysis of risk factors and outcomes in patients with bloodstream infection Jpn J Infect Dis 2003 56 1 7 12711818
Pittet D Wenzel RP Nosocomial bloodstream infections: secular trends in rates, mortality, and contribution to total hospital deaths Arch Intern Med 1995 155 1177 84 7763123 10.1001/archinte.155.11.1177
Beekmann SE Diekema DJ Chapin KC Doern GV Effects of rapid detection of bloodstream infections on length of hospitalization and hospital charges J Clin Microbiol 2003 41 3119 25 12843051 10.1128/JCM.41.7.3119-3125.2003
Arbo M Snydman D Influence of blood culture results on antibiotic choice in the treatment of bacteremia Arch Intern Med 1994 154 2641 5 7993147
Hanon FX Monnet DL Sorensen TL Molbac K Petersen G Schonheyder H Survival of patients with bacteremia in relation to initial empirical antimicrobial treatment Scand J Infect Dis 2002 34 520 8 12195878 10.1080/00365540110080827
Nathwani D Davey P France AJ Phillips G Orange G Parratt D Impact of an infection consultation service for bacteremia on clinical management and use of resources Q J M 1996 89 789 97
Leibovici L Konisberger H Pitlik S Samara Z Drucker M Bacteremia and fungemia of unknown origin in adults Clin Infect Dis 1992 14 436 43 1445517
Mandell Principles and Practice of Infectious Diseases 2000 Churchill Livingstone, Inc Copyright ©
Aronson MD Bor DH Blood cultures Ann Int Med 1987 106 246 53 3541726
Weinstein MP Current blood culture methods and systems: clinical concepts, technology, and interpretation of results Clin Infect Dis 1996 23 40 6 8816127
Smith-Elekes S Weinstein MP Blood cultures Inf Dis Clin North Am 1993 7 221 34
Weinstein MP Reller LB Murphy JR Lichtenstein KA The clinical significance of positive blood cultures: a comprehensive analysis of 500 episodes of bacteremia and fungemia in adults: I. laboratory and epidemiologic observations Rev Infect Dis 1983 5 35 53 6828811
Bates DW Goldman L Lee TH Contaminant blood cultures and resource utilization: the true consequences of false-positive results JAMA 1991 265 365 69 1984535 10.1001/jama.265.3.365
Strand CL Wajsbort RR Sturmann K Effect of iodophor versus iodine tincture skin preparation on blood culture contamination rate JAMA 1993 269 1004 6 8429580 10.1001/jama.269.8.1004
Schiff GD Wisniewski M Bult J Parada JP Aggarwal H Schwartz DN Improving inpatient antibiotic prescribing: insights from participation in a national collaborative Jt Comm J Qual Improv 2001 27 387 402 11480200
Trenholme GM Kaplan RL Karakusis PH Stine T Fuhrer J Landau W Levin S Clinical impact of rapid identification and susceptibility testing of bacterial blood culture isolates J Clin Microbiol 1989 27 1342 5 2473995
Mazzuca SA Cohen SJ Clark CM Jr Evaluating clinical knowledge across years of medical training J Med Ed 1981 56 83 90
|
16164757
|
PMC1261264
|
CC BY
|
2021-01-04 16:28:16
|
no
|
BMC Infect Dis. 2005 Sep 15; 5:71
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-71
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-731617151910.1186/1471-2334-5-73Research ArticleGeneration of human antibody fragments recognizing distinct epitopes of the nucleocapsid (N) SARS-CoV protein using a phage display approach Flego Michela [email protected] Bonito Paola [email protected] Alessandro [email protected] Silvia [email protected] Alessandra [email protected] Felicia [email protected] Antonio [email protected] Maurizio [email protected] Department of Drug Research and Evaluation, Istituto Superiore di Sanità, Rome, Italy2 Department of Infectious Parasitic and Immune-mediated Diseases, Istituto Superiore di Sanità, Rome, Italy2005 19 9 2005 5 73 73 18 4 2005 19 9 2005 Copyright © 2005 Flego et al; licensee BioMed Central Ltd.2005Flego 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 acute respiratory syndrome (SARS)-CoV is a newly emerging virus that causes SARS with high mortality rate in infected people. Successful control of the global SARS epidemic will require rapid and sensitive diagnostic tests to monitor its spread, as well as, the development of vaccines and new antiviral compounds including neutralizing antibodies that effectively prevent or treat this disease.
Methods
The human synthetic single-chain fragment variable (scFv) ETH-2 phage antibody library was used for the isolation of scFvs against the nucleocapsid (N) protein of SARS-CoV using a bio panning-based strategy. The selected scFvs were characterized under genetics-molecular aspects and for SARS-CoV N protein detection in ELISA, western blotting and immunocytochemistry.
Results
Human scFv antibodies to N protein of SARS-CoV can be easily isolated by selecting the ETH-2 phage library on immunotubes coated with antigen. These in vitro selected human scFvs specifically recognize in ELISA and western blotting studies distinct epitopes in N protein domains and detect in immunohistochemistry investigations SARS-CoV particles in infected Vero cells.
Conclusion
The human scFv antibodies isolated and described in this study represent useful reagents for rapid detection of N SARS-CoV protein and SARS virus particles in infected target cells.
==== Body
Background
The widespread diffusion and mortality of severe acute respiratory syndrome (SARS), caused by a new Coronavirus (SARS-CoV) has threatened the entire world [1] and has urged the scientific community to develop molecular and serological tests which can assist for rapid detection of SARS cases and implementation of control measures [2]. Effective prophylaxis and antiviral therapies are urgently needed in the event of re-emergence of the highly contagious and often fatal SARS infection.
Like other known Coronaviruses, SARS-CoV is an enveloped virus containing four structural proteins, namely the membrane (M), envelope (E), spike (S) and the nucleocapsid (N) proteins [1]. The N protein is a 423 amino acid, predicted phospho-protein of 46 kDa, which shares little homology with other members of the Coronavirus genus [3,4]. Several studies have found that N protein is highly immunogenic, thus antibody response in patients with SARS is directed most frequently and predominantly to the nucleocapsid. It has also been found that anti N antibodies are detected early and with high specificity during infection [5]. Therefore, the generation of N-specific reagents, in particular monoclonal antibodies (mAbs) could be of assistance in studies of viral replication and antiviral activity, as well as in the diagnosis of SARS-CoV infection at various disease stages. It has also been found that SARS patients show clinical improvement if they are treated with serum from previously infected patients suggesting that passive immunotherapy could be developed for the treatment of this infectious disease [6]. However, the traditional approach to generate mAbs to SARS-CoV has presented difficulties for several reasons including safety concerns in handling SARS particles [7]. To overcome these limitations, we applied a very effective and safe in in vitro procedure based on human scFvs selection from the large synthetic ETH-2 phage antibody library [8]. Here, we report the identification, production, and epitope characterization of human scFv antibodies specifically recognizing distinct N protein domains. These recombinant mAbs were found to bind selectively and with good affinity to distinct N protein epitopes, and were suitable to specifically identify SARS particles in Vero infected cells.
Methods
SARS-CoV antigens
Nucleocapsid (N) protein of SARS-CoV (residues: 1–422) and its fragments N1 (residues: 1–117), N2 (residues: 110–340) and N3 (residues: 333–422) expressed in E. coli by DNA recombinant technology as reported by Carattoli, Di Bonito et al [9]. These polypeptides have proved to react with specific sera of SARS-CoV infected patients.
ETH-2 antibody phage library
The synthetic human recombinant antibodies library (ETH-2) consists of a large array (more than 109 antibody combination) of scFv polypeptides displayed on the surface of M13 phage. It was built by random mutagenesis of the CDR3 of only three antibody germline gene segments (DP47 for the heavy chain, DPK22 and DPL16 for the light chain). Diversity of the heavy chain was created by randomizing four to six positions replacing the pre-existing positions 95 to 98 of the CDR3. The diversity of the light chain was created by randomizing six positions (91 to 96) in the CDR3 [8].
Isolation of phage antibodies from ETH-2 library
An aliquot of the ETH-2 library, containing 1013 cfu phage, was used to isolate specific human antibodies in scFv format to SARS-CoV N protein. Immunotubes (Nunc Maxisorp; Denmark) were coated overnight (ON) at room temperature (RT) with 10 μg/ml of recombinant purified SARS-CoV N protein in PBS. After panning, phages were eluted with 1 ml of 100 mM triethylamine and the solution was immediately neutralized by adding 0.5 ml of 1 M Tris-HCl pH 7.4. Eluted phages were used to infect log phase TG1 E. coli bacteria and amplified for the next round of selection. Briefly, 50 ml of 2 × YT with 100 μg/ml ampicillin (2 × YTA medium) and glucose 1% were inoculated with enough bacterial suspension to yield an OD600 nm≅ 0.05–0.1. The culture was grown up to OD600 nm 0.4–0.5 and infected with M13 K07 helper phage in a ratio of around 20:1 phage/bacteria. The rescued phages were concentrated by precipitation with PEG 6000 and used for next round of panning (usually three to recover antigen-specific antibody phages from the ETH-2 library). For soluble scFv preparation, individual colonies were grown in 96 flat bottomed wells (Nunc) for 2 hours at 37°C in 180 μl 2 × YTA medium and glucose 0.1% in 96 well plates and inducted with 50 μl 2 × YTA medium and 6 mM IPTG. The following day the plates were spinned down at 1800 g for 10 minutes and the supernatants containing soluble scFvs were recovered and tested for specificity.
ELISA
96 well ELISA-plates (Nunc Maxisorp) were coated ON at RT with 0.5 μg of antigen (N protein or its N1, N2, N3 fragments) in PBS. The following day a blocking solution, consisting of 2% non fat dry milk in PBS (MPBS), was added; plates were washed with PBS containing 0.05% Tween 20 (TPBS) and incubated for 2 hours at RT with 50 μl of supernatants containing soluble scFv antibody, anti-Flag M2 antibody (1.6 μg/ml Sigma-Aldrich; MO, USA) and anti-mouse HRP conjugated antibody (1.6 μg/ml Dako; Denmark). The reaction was developed using 3, 31-5, 51-tetramethylbenzidin BM blue, POD-substrate soluble (Roche Diagnostics; IN, USA) and stopped by adding 50 μl of 1 M sulfidric acid. The reaction was detected with an ELISA reader (Biorad; CA, USA), and the results were expressed as A (absorbance) = A(450 nm)-A(620 nm).
Western blotting
1 μg of SARS-CoV N, N1, N2 and N3 proteins were loaded onto 12% SDS-PAGE and transferred to a nitrocellulose membrane using standard procedures. The membrane was blocked in 4% MPBS ON at RT. Blotted proteins were incubated for 2 hours with supernatant containing soluble scFvs, washed with 0,05% TPBS and incubated again with 5 μg/ml anti-Flag M2 mouse antibody (Sigma-Aldrich) in 2% MPBS. After an additional incubation for 1 hour at RT in presence of 5 μg/ml of a goat anti-mouse antibody HRP-conjugated (Dako), the reaction was developed and visualized with a chemiluminescence detection kit (Pierce; IL, USA).
Immunocytochemical determination of SARS-CoV particles
BIOCHIP slides (Euroimmun, Luebeck, Germany) coated with SARS-CoV infected Vero cells were used for immunocytochemistry determination of virus particles by phage or scFv specific antibodies. Slides were treated for 10 minutes with peroxidase block solution (Dako En Vision System HRP), washed in TPBS buffer and incubated for 1 hour at RT with 1012 phage antibody particles (resuspended in 25 μl of TPBS) or scFv containing supernatant (25 μl). As control, BIOCHIP slides were incubated with an irrelevant anti-glucose oxidase phage or scFv antibodies [10]. After extensive washings with TPBS, the slides were incubated for 1 hour at RT with an anti-M13 secondary mouse antibody (5 μg/ml in 20 μl TPBS) (Amersham) or with an anti-FLAG secondary mouse antibody (5 μg/ml in 20 μl TPBS) (Sigma-Aldrich) for labelling phage and scFv reactive antibodies, respectively. After washings, the slides were treated for 30 minutes at RT with cromogen solutions (Dako) and inspected with a light microscope.
DNA characterization and sequences
Plasmid DNA from individual bacterial colonies of MA2.D5, MA2.D7 and MA2.E3 clones was digested with specific endonucleases and CDR3 regions sequenced with an automated DNA sequencer (M-Medical/Genenco, Pomezia, Italy) using fdseq1 (5'-GAA TTT TCT GTA TGA GG-3') and pelBback (5'-AGC CGC TGG ATT GTT ATT AC-3') primers.
Results and discussion
Isolation of human scFv antibodies
To isolate specific scFv antibodies, an aliquot of the human synthetic ETH-2 library containing approximately 1 × 1012 cfu phages was introduced for panning into Maxisorp immunotubes coated with purified SARS-CoV N protein as antigen. Non-specifically absorbed phages were removed by washing. Bound phages were eluted, amplified and used for the next cycle of panning as described elsewhere [8,10]. By using this protocol [10], phage antibody populations specifically recognizing the N protein of SARS-CoV were isolated after only three rounds of selection. Plating on agar of TG1 phage antibody-infected cells, allowed growth of individual phagemid clones. Soluble scFvs derived from IPTG inducted colonies, were screened by ELISA and several of them proved to be specific for N protein (Figure 1).
Figure 1 N protein specific scFvs antibodies determined by ELISA. IPTG inducted bacterial supernatants of individual colonies from the third round of the ETH-2 library selection on N protein were tested in 96-well microtiter plates coated with antigen or glucose oxidase (negative control). ELISA readings higher than three fold above negative controls were scored as positive reactions. ELISA values of the scFv clones against N protein (panel A) and glucose oxidase (panel B) are shown.
Epitope recognition
N-positive scFv clones were analysed by ELISA for reactivity with N1, N2 and N3 protein fragments (Figure 2). Three distinct classes of scFv antibodies recognizing either the intact N protein only or also N2 or N3 fragments were identified. ScFv antibodies recognizing N protein only but none of its fragments (for instance the MA2D5 scFv) were likely directed against epitopes encompassing two adjacent protein fragments or conformational epitopes expressed only on the integral protein. The absence of scFv antibodies reacting with the N1 fragment somewhat matches the lower antigenicity of this polypeptide, as compared to immunodominant N2 and N3 fragments, [9], despite the reported reactivity of several linear synthetic epitopes of the N1 region with SARS sera [11]. However, a relatively low number of scFv clones were here tested for a sound conclusion on this aspect.
Figure 2 N protein domain recognition by specific scFv antibodies. The scFvs MA2.E3, MA2.D5 and MA2.D7 antibody clones showing distinct N protein recognition patterns in an ELISA (A, A', A") were also analyzed for different epitope recognition by western blot. While all scFv antibodies react with a 46–48 kDa band corresponding to the MW of the N protein (B, B', B"), the scFvs MA2.E3 and MA2.D7 also react with a 28–30 KDa and 12–14 KDa band (B, B") corresponding to the MW of N2 and N3 protein fragments, respectively.
SARS-CoV particles identification
Three of the most reactive scFv antibody clones named MA2.D5, MA2.D7 and MA2.E3 each one representing the different classes of epitope recognition in the N protein domains were further assessed to verify their biochemical patterns, reactivity with SARS-CoV infected cells, and DNA sequences of their encoding genes.
Western blotting analysis (Figure 2) shows that all three antibodies react with a 46 kDa band corresponding to the predicted molecular mass of the intact N protein [3]. As expected by previous ELISA indications, the scFvs MA2.E3 and MA2.D7 recognize also a 28–30 kDa and 12–14 kDa bands respectively, corresponding to the predicted molecular mass of N protein fragments N2 and N3 [9].
The molecular analysis of the phage antibodies MA2.D5, MA2.D7 and MA2.E3 shows scFv antibodies with a molecular weight of about 27 kDa and the integrity of the genes encoding for the cognate scFvs displayed on M13 phage (data not show). VH and VL gene sequences, shows that each single scFv antibody clone possess a unique DNA sequence encoding for the CDR3 region (Figure 3); this uniqueness is consistent with the distinct epitope recognition patterns of the three scFvs noted above. Immunocytochemical investigations show that MA2.D5 antibody specifically recognizes SARS-CoV particles in infected Vero cells (Figure 4). The other two antibodies MA2.D7 and MA2.E3, though genuinely reacting with N protein and its N2 and N3 fragments in ELISA and western blotting (see above) did not prove to be useful for SARS-CoV determination in infected cells.
Figure 3 Molecular genetics characteristics of the scFv antibodies. The nucleotide composition, and the corresponding amino acid sequences and residue position in the CDR3 region of the selected scFv antibodies MA2.D5, MA2.D7 and MA2.E5 are reported. A schematic representation of the scFv antibodies displayed on M13 phage as pIII fusion proteins is depicted.
Figure 4 SARS-CoV detection by MA2.D5 scFv antibody. The immunocytochemical detection of SARS-CoV particles in infected monkey Vero cells by the MA2.D5 antibody is shown in A (soluble scFv protein) and B (phage displaying scFv antibody). In C and D the reactivity of Vero cells with an irrelevant scFv and phage antibody are shown.
Conclusion
The usefulness of the library as a tool for generating monoclonal antibodies against viral pathogens [12-14] including SARS-CoV [15,16] has been tested and showing that phage antibodies may recognize viral proteins used as antigens. In particular, a rodent library has proved to be very effective for the isolation of murine scFvs exhibiting high specificity against SARS-CoV E protein, which is recognized as a 31 kDa protein in western blot studies [15]. Interestingly, one of these clones (A17) isolated by selecting the rodent library against the 31 kDa E protein cross-reacts in ELISA with a well-defined fragment of the 46–48 kDa N protein. Recently, human mAbs were identified by biopanning of a synthetic library on SARS-CoV lysate; the isolated scFvs expressed on phages have proved to recognize N protein only in ELISA while no evidences for N protein recognition in western blot studies or in SARS infected cells were shown [16]. The efficient display of recombinant antibodies on filamentous phage has been also assessed by our findings showing that soluble human scFvs selected from the ETH-2 library may yield agents capable of specifically recognizing SARS-CoV particles in infected Vero cells as well as the intact 46–48 kDa SARS-CoV N protein and its N2 (28–30 kDa) and N3 (12–14 kDa) fragments in western blot and ELISA investigations (Figure 2 and 4). To our knowledge, the human scFvs directed against N protein, by us isolated and characterized, are the first antibodies so far described which are able to detect SARS-CoV N protein in different routinary laboratory techniques such as ELISA, Western blot and immunocytochemistry. Hence, these human mAbs represent excellent candidates for future development of serological diagnosis of SARS-CoV infections since high functionality (all clones express soluble antibodies in bacteria), specificity (all clones recognize the 46–48 kDa N protein) and quality (all clones give good signals in ELISA, in western blot as well and stain SARS-CoV infected cells).
Furthermore, the genes encoding for the selected MA2.D5, MA2.D7 and MA2.E3 antibodies have been isolated and sequenced, thus facilitating various molecular approaches including site direct mutagenesis to maturate binding affinity and construct whole recombinant immunoglobulins for SARS-CoV studies and applications. To this latter regard, another immediate use of human scFv directed to N protein is through intracellular expression as a novel strategy of gene therapy aimed at knockout the replicative cycle of SARS-CoV in infected cells. It is thought that intracellular expression of scFv antibodies may be superior over alternative methods for gene inactivation such as anti-sense RNA and dominant negative mutants because of its high specificity [17,18].
Nevertheless, phage display approach make feasible several strategies for the isolation of neutralizing human mAbs to provide an immediate treatment for emergency prophylaxis of SARS-CoV [19] while vaccines and new drugs are underway. For example, immune library may be constructed and screened on SARS-CoV proteins, using B-cells from convalescent SARS patients as a genetic source of specific VH and VL. In alternative, synthetic peptides mimicking immunodominant epitopes in S or M SARS-CoV proteins [20] may be used as a substrate antigens for the identification of specific scFvs from synthetic human phage antibody library.
List of abbreviations
SARS, Severe Acute Respiratory Syndrome; CoV, Coronavirus; N, Nucleocapsid; scFv, single chain fragment variable; ETH-2, synthetic scFv antibody phage library; CDR, complementarity determining regions; mAbs, monoclonal antibodies.
Competing interests
The author(s) declare that they have not competing interest.
Authors' contributions
MF carried out phage antibodies and scFvs selection from all different N-antigens, participated in the genetic molecular, biochemical and immunohistochemical characterization of scFvs specific for distinct epitopes of N protein.
PDB and FG carried out isolation, expression, production and purification of nucleocapsid protein from SARS-CoV.
AA participated in phage antibodies selection and carried out immunoassay and biochemical characterization of scFvs against N protein and its polypeptide fragments.
ACr firstly, ideated and developed a genetic molecular model to isolate, identify and express SARS-CoV N protein and its fragments.
AC promotes the genetic molecular study of SARS-CoV N protein, participated in the design of the study, supervisioned the experiments and critically revised the manuscript. MC conceived of the study, promotes the approach with phage library to select specific scFv human antibodies against SARS-CoV protein. Furthermore, MC participated in the design and coordination of the research and drafting the manuscript.
All authors have read and approved the final version of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We are grateful to Mrs. S. Tocchio for editorial assistance. This work was supported by a grant from Ministero della Salute, Italy, under contract for "Progetto Speciale Lotta alla SARS". MF is supported by an AIDS grant from the Italian Ministry of Health.
==== Refs
Peiris JS Guan Y Yuen KY Severe acute respiratory syndrome Nat Med 2004 88 97 10.1038/nm1143
Bermingham A Heinen P Iturriza-Gomara M Gray J Appleton H Zambon MC Laboratory diagnosis of SARS Philos Trans R Soc Lond B Biol Sci 2004 359 1083 1089 15306394 10.1098/rstb.2004.1493
Timani KA Ye L Ye L Zhu Y Wu Z Gong Z Cloning, sequencing, expression, and purification of SARS-associated coronavirus nucleocapsid protein for serodiagnosis of SARS J Clin Virol 2004 30 309 312 15163419 10.1016/j.jcv.2004.01.001
Surjit M Liu B Kumar P Chow VT Lal SK The nucleocapsid protein of the SARS coronavirus is capable of self-association through a C-terminal 209 amino acid interaction domain Biochem Biophys Res Commun 2004 317 1030 1036 15094372 10.1016/j.bbrc.2004.03.154
Leung DT Tam FC Ma CH Chan PK Cheung JL Niu H Tam JS Lim PL Antibody response of patients with severe acute respiratory syndrome (SARS) targets the viral nucleocapsid J Infect Dis 2004 190 379 386 15216476 10.1086/422040
Burnouf T Radosevich M Treatment of severe acute respiratory syndrome with convalescent plasma Hong Kong Med J 2003 4 309 310 12904626
Xiong S Wang YF Zhang MY Liu XJ Zhang CH Liu SS Qian CW Li JX Lu JH Wan ZY Zheng HY Yan XG Meng MJ Fan JL Immunogenicity of SARS inactivated vaccine in BALB/c mice Immunol Lett 2004 95 139 143 15388253 10.1016/j.imlet.2004.06.014
Viti F Nilsson F Demartis S Huber A Neri D Design and use of phage display libraries for the selection of antibodies and enzymes Methods Enzymol 2000 326 480 505 11036659
Carattoli A Di Bonito P Grasso F Giorgi C Blasi F Niedrig M Cassone A Recombinant protein-based ELISA and immuno-cytochemical assay in diagnosis of Sars J Med Virol 2005 15834868
Ascione A Flego M Zamboni S De Cinti E Dupuis ML Cianfriglia M Isolation and characterization of the human monoclonal antibodies C10 in single-chain fragment variable (scFv) format to glucose oxidase from Aspergillus niger Hybrid Hybridomics 2004 23 380 384 15684666
Wang J Wen J Li J Yin J Zhu Q Wang H Yang Y Qin E You B Li W Li X Huang S Yang R Zhang X Yang L Zhang T Yin Y Cui X Tang X Wang L He B Ma L Lei T Zeng C Fang J Yu J Wang J Yang H West MB Bhatnagar A Lu Y Xu N Liu S Assessment of Immunoreactive Synthetic Peptides from the Structural Proteins of Severe Acute Respiratory Syndrome Coronavirus Clin Chem 2003 49 1989 1996 14633869 10.1373/clinchem.2003.023184
Veiga E De Lorenzo Fernandez LA Neutralization of enteric coronaviruses with Escherichia coli cells expressing single-chain Fv-autotransporter fusions J Virol 2003 77 13396 13398 14645594 10.1128/JVI.77.24.13396-13398.2003
Hayhurst A Happe S Mabry R Koch Z Iverson BL Georgiou G Isolation and expression of recombinant antibody fragment to the biological warfare pathogen Brucella melitensis J Immunol Methods 2003 276 185 196 12738372 10.1016/S0022-1759(03)00100-5
Sapats SI Heine HG Trinidad L Gould GJ Foord AJ Doolan SG Prowse S Ignjatovic J Generation of chiken single chain antibody variable fragments (scFv) that differentiate and neutralize infectious bursal disease virus (IBDV) Arch Virol 2003 148 497 515 12607101 10.1007/s00705-002-0931-2
Liu H Ding YL Han W Liu MY Tian RY Yang SL Gong Y Recombinant scFv antibodies against E protein and N protein of severe acute respiratory syndrome virus Acta Biochim Biophys Sin (Shanghai) 2004 36 541 547 15295646
Liu ZX Yi GH Qi YP Liu YL Yan JP Qian J Du EQ Ling WF Identification of single-chain antibody fragments specific against SARS-associated coronavirus from phage-displayed antibody library Biochem Biophys Res Commun 2005 329 437 444 15737606 10.1016/j.bbrc.2005.02.003
Cattaneo A Biocca S The selection of intracellular antibodies Trends Biotechnol 1999 17 115 121 10189716 10.1016/S0167-7799(98)01268-2
Jannot CB Beerli RR Mason S Gullick WJ Hynes NE Intracellular expression of a single chain antibody directed to the EGFR leads to growth inhibition of tumour cells Oncogene 1996 13 275 282 8710366
Sui J Li W Roberts A Matthews LJ Murakami A Vogel L Wong SK Subbarao K Farzan M Marasco WA Evaluation of human monoclonal antibody 80R for immunoprophylaxis of severe acute respiratory syndrome by an animal study, epitope mapping, and analysis of spike variants J Virol 2005 79 5900 6 15857975 10.1073/pnas.0307140101
Zhong X Yang H Guo ZF Sin WY Chen W Xu J Fu L Wu J Mak CK Cheng CS Yang Y Cao S Wong TY Lai ST Xie Y Guo Z B-cell responses in patients who have recovered from severe acute respiratory syndrome target a dominant site in the S2 domain of the surface spike glycoprotein J Virol 2005 79 3401 3048 15731234 10.1128/JVI.79.6.3401-3408.2005
|
16171519
|
PMC1261265
|
CC BY
|
2021-01-04 16:28:16
|
no
|
BMC Infect Dis. 2005 Sep 19; 5:73
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-73
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-751618535210.1186/1471-2334-5-75Research ArticleDog filariosis in the Lazio region (Central Italy): first report on the presence of Dirofilaria repens Scaramozzino Paola [email protected] Simona [email protected] Paolo Michele [email protected] Marcello [email protected] Francesco [email protected] Gabriella [email protected] Istituto Zooprofilattico Sperimentale delle Regioni Lazio e Toscana, Via Appia Nuova 1411, 00178 Rome, Italy2 Parasitology Section, Department of Public Health Science, University of Rome 'La Sapienza' Piazza Aldo Moro, Rome, Italy2005 26 9 2005 5 75 75 5 5 2005 26 9 2005 Copyright © 2005 Scaramozzino et al; licensee BioMed Central Ltd.2005Scaramozzino 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
Epidemiological investigations were carried out in the Lazio Region to assess the status of canine filariosis and to evaluate the actual risk for veterinary and medical public health.
Methods
Since August 2001 to June 2003, a total of 972 canine blood samples, collected in public kennels and from private owners animals of the 5 Provinces of the Region, were tested. The presence of filarial parasites was evaluated by microscopy and bio-molecular techniques; the species identification was performed by means of the same diagnostic tools.
Results
A total of 17/972 (1.75%; 95%CI 1.06%–2.85%) blood samples were parasitized by D. repens,13 out them drawn by dogs resident in the Province of Roma, and 4 in the other provinces. Multivariate analysis was performed in order to evaluate the association between filariosis and risk factors. The origin from coastal territories seems to be a significant risk factor to acquire the infection.
Conclusion
This is the first report of canine filariosis in the Lazio Region, where D. repens was before reported only in foxes. The risk of human zoonotic infection is stressed, and the absence of other filarial species is discussed
==== Body
Background
Filarial nematodes described in dogs are: Dirofilaria immitis, D. repens, Acanthocheilonema reconditum, A. dracunculoides and Cercopitifilaria grassi (Order: Spirurida, Superfamily: Filarioidea, Family: Onchocercidae). The most prevalent species are D. immitis, D. repens, and A. reconditum, that show a different geographical distribution: cosmopolitan for A. reconditum and D. immitis, restricted to the Europe, Middle East, Asia and Africa for D. repens. D. immitis is responsible for heartworm disease, whereas the other species produce subcutaneous or splanchnic infections. Furthermore, in areas where dog filarioses are endemic, at least D. immitis and D. repens are recognized as etiological agent of zoonotic infections in humans.
Canine heartworm disease is regarded as one of the most dangerous threat for the dog health, but it is an emerging sanitary problem also for cats. This dirofilariosis is endemic-hyperendemic in the Northern Italy (the Po Valley is the largest endemic area), with prevalence rates ranging from 22 to 68% (even 80% where animals are not receiving chemoprophylaxis) [1-5]. Similar high prevalence is reported from other countries in Southern Europe [6,7], where the incidence rate observed during the last decades has been increasing, with a northward spread of the infection [5]. This trend is confirmed, in Italy, by recent surveys that found endemic some previously disease-free areas [4]. Different is the pattern of heartworm infection in Central and Southern Italy, where much lower infection rates (13%) [8] are observed, or its presence is reported only occasionally [9,10]. As far the Lazio region, up to now there have been no reports about autochtonous infections, despite the presence of competent vectors belonging to the Culex and Aedes genera.
D. repens is considered scarcely pathogenic and therefore its distribution is less studied. In Europe the parasite has been reported in Bulgary (1%) [11], in Switzerland (1.6%) [12], in Greece (12–37%) [13], in France (1.36%) [14] and in the mediterranean side of Spain with infection rates ranging from 5.1% to 84.6% [15,16]. As far as Italy the species is reported with increasing prevalence from northern to southern regions. The parasite has been evidenced recently in most regions of central Italy (Toscana, Umbria, and Campania) with prevalences ranging from 2 to 21%. In detail infection rates reported for canine subcutaneous dirofilariosis in aformentioned regions are 21.1%; 6% and 2% respectively [17,8,10]. In the Lazio region D. repens has been reported only in foxes in the 60ies [18]; since then, no additional data on its geographical distribution and its presence among the dog population are available. The present study is, therefore, aimed to assess the status of canine filariosis in this region, considering in particular the public health risk in the city of Rome, where the relationship between dog and human populations is extremely tight. The role of some potential risk factors has been also investigated.
Methods
Study area
The Lazio Region is 17,207 km large and its territory is divided among 5 Provinces: Roma, Viterbo, Rieti, Latina and Frosinone. It is bordered by the Tyrrhenian Sea to the West and by the Apennine mountains (ca. 2,000 m a.s.l.) to the East. Landscape is mainly hilly, with coastal plains only taking about 20% of the territory. Some of the pre-existing natural marshy lands have been dried at the time of anti-malaria campaigns. Climate is classified as Mediterranean or sub-tropical, with dry summer and mild winters.
There are no official data about the canine population, but it is estimated to be 400,000 only in the Province of Roma, about 200.000 of which in the capital.
Sampling protocol
Since August 2001 to June 2003 a sampling protocol on canine population of the Lazio Region was carried out in two phases. In the first sampling phase, aimed to test the dog population of the Roma Province, two sub-samples were defined: (i.) ownership animals (n = 320) and (ii.) dogs from 6 public kennels (n = 352). The former (i.) was calculated to single out at least one positive at a 95% confidence level, assuming 1% of expected prevalence. In order to define a strategy to sample ownership dogs sera, the city was subdivided in six study areas, lying both in the urban internal territory (n = 4) and along the coast (n = 2). For each area, one private veterinary practice was involved in the survey. Veterinarians were asked to bleed dogs randomly chosen among the ones brought to clinics for routinely control and were also asked to collect basic information data. The second sub-sample (ii.) was obtained by a convenience sampling of a variable number (15–91) of dogs from each kennel.
Only animals older than 6 month, not recently treated by antihelmintics, have been included in the study. Dogs were bled before 9:00 in the morning or after 5:00 in the afternoon, to assure sampling times that fit in with the microfilariae circadian rhythm.
In the second phase, in order to extend the study and to acquire some more information on the distribution of canine filariosis in the entire Region, a supplementary convenience sampling was performed in four kennels, each located in one of the remaining 4 Provinces of the Lazio, and each keeping dogs caught in the one's territory. In every one of the kennels, which were lacking of insect proof nets, 7 to 97 dogs were tested.
Data about age, sex, hair size (short/long), living habitat (kennel/private house), province of residence, area of residence (urban/rural, coast/hill/mountain) for each dog were collected and computerized.
Further samples (about 170) coming from routinely diagnostic activity of the Istituto Zooprofilattico Sperimentale di Roma were included in the final data set.
Altogether, a total of 972 dog blood samples were tested in this study.
Filarial infection was detected by microscopic and bio-molecular techniques.
Microscopic analysis
EDTA blood samples were kept at 4°C until microscopic examination, performed usually within 2–3 days. Blood was processed according to the modified Knott method
[19]. Microfilariae were looked for at low magnification (200X) and then observed at 400X and 1000X in order to identify the species on the basis of morphological features [20]. Samples were then stored at -20°C for the successive analysis.
Bio-molecular analysis
Samples positive to microscopy were examined individually, to confirm species specificity of microscopic identification, whereas negative samples were examined on pools of 4 samples each. DNA extraction was achieved by the QIAamp DNA blood extraction kit (Qiagen). A first PCR-based analysis was performed according to the protocol previously developed [21], using "filarial" specific ribosomal primers named S2-S16 [22]. The amplification give rise to a product of about 400 bp for most filarial species, whereas D. repens yields an additional fragment of about 350 bp. Amplification products were excised from agarose gel, purified with the Nucleo Spin ® Extract kit (Macherey-Nagel), and analysed for sequencing by M-Medical Srl. Sequences comparison was achieved by CLUSTAL analysis [23]. Moreover, PCR positive samples were further tested using primers specific for D. repens and D. immitis [21], to confirm the identification of filarial species. Amplification products expected are of 325-bp and a ladder consisting of multiples of 175-bp for D. repens, and a minor 348-bp fragment with a major 747-bp fragment for D. immitis.
Data analysis
The prevalence of filariosis and the confidence interval of the estimates at 95%confidence level, based on the results of microscopic and molecular analysis, were calculated for the whole sample and for each Province of the dog origin. Only for the dog population coming from the Roma Province, whose data collection procedures were well controlled, more accurate and uniformly distributed on the territory, the complete data set was available. Univariate logistic linear regression analysis at 95% confidence level was performed in order to estimate the effect of each of the following variables on the risk of filariosis: distance from the coastline (<20 Km; >20 Km); urbanization level of the origin area (rural/urban); breeding in kennel/private house; dog age (6–12; 13–36; 37–120; > 120 months); hair length. To investigate the possibility of confounding in the observedunivariate associations, we conducted multivariate logistic regressions with all candidate independent variables. The Odds Ratios (OR) and the 95% confidence intervals (95%CI) were calculated. A P-value of <0.05 was considered significant. All analyses were conducted using SPSS® 10.0 (SPSS Inc., 1999).
Results
Among the overall blood samples collected in the Lazio Region (n = 972), 17 were positive for Dirofilaria by modified Knott method, giving a prevalence of 1.75% (95%CI 1.06%–2.85%). Dirofilariosis prevalences observed in each Province were: 13/730 (95%CI 1.78%; 0.99%–3.11%) in Roma, 1/97 (1.03%) in Latina, 2/46 (4.35%) in Rieti, and 1/18 (12.5%) in Viterbo. No one of the 73 animals tested in the Frosinone Province were parasitized.
Filarial species identified by microscopy in all positive samples was D. repens. Molecular tools confirmed the number of positive and negative dogs and the species identification, as demonstrate by the pattern shown by all positive samples. In fact, the application of primers for "filarial parasites", and the sequencing of amplified products analysed by Clustal, indicated a sequence similarity of 99.8% with D. repens. The first identification was then tested by specific primers for D. repens and D. immitis [21]: all positive samples were amplified only by D. repens primers, so confirming the sequence analysis. An example is reported in fig. 1.
Figure 1 PCR. Amplification products with specific primers for D. repens shown by 6 positive blood samples (lanes 1–6); negative (7) and positive (8) controls, molecular marker 100 bp (M).
The results regarding the Roma province are summarized in tab. 1. A more detailed analysis concerning dogs found positive showed that 12 (among 393) came from coastal areas (9 from extra-urban and 3 from urban territory) and 1 (among 337) from the inner-land (urban territory) (fig. 2). Four positive dogs came from small kennels in the two areas along the coast; the other 9 lived in private houses. All the dogs (n = 91) tested from the largest public kennel within Rome municipality and located in the urban area, which hosts stray dogs coming from the entire city, were negatives.
Table 1 Province of Rome: tested dogs (n° positive) on the basis of distance from the coastline, age class and sex.
distance from the coastline ≤ 20 km
age class (months) female male nd* f+m+nd
1(< 13) 10 (0) 12 (0) - 22 (0)
2 (13–36) 26 (0) 28 (0) 2 (0) 56 (0)
3 (37–120) 56 (1) 75 (5) 2 (1) 133 (7)
4 (> 120) 4 (0) 17 (0) - 21 (0)
nd* 41 (0) 65 (5) 55 (0) 161(5)
all age classes coastal 137 (1) 197 (10) 59 (1) 393 (12)
> 20 km
1(< 13) 12 (0) 17 (0) - 29 (0)
2 (13–36) 34 (0) 47 (0) - 81 (0)
3 (37–120) 68 (0) 73 (0) - 141 (0)
4 (> 120) 3 (0) 6 (1) - 9 (1)
nd* 7 (0) 21 (0) 49 (0) 77 (0)
all age classes inner land 124 (0) 164 (1) 49 (0) 337 (1)
total (coastal + inner land) 261 (1) 361 (11) 108 (1) 730 (13)
*Not determined
Figure 2 Dirofilariosis in the Lazio Region Distribution of dirofilariosis in the Lazio region (in the mainframe). Prevalence data in the Province of Rome respect to the distance to the coast line (< 20 km; > 20 km). In pale grey tested negative municipalities, in dark grey the positive municipalities
Since a large number of dogs were of unknown age (238/730) and 5 of them were positive, the effect of age on the risk for dirofilariosis was not evaluable, although all the remaining positive dogs (n = 8) were > 36 months of age. Even if the number of microfilaraemic dogs was higher in rural environment (9/374; 2.4%) than in urban areas (4/343; 1.2%), no effect of the urbanization level of the origin area was observed at the univariate logistic regression analysis. Similarly no effect was observed with reference neither to breeding in kennel/private house nor to hair length.
On the contrary, the origin from coastal territories (inner 20 km from coast line) represents a significant risk factor for canine dirofilariosis in the Roma Province. In fact, the risk for dirofilariosis was more than 10 fold higher in dogs coming from coastal areas than from the inner-land (OR 10.58; 95%CI 1.37 to 81.82)
Moreover, the risk for dirofilariosis turned out higher in males than in female dogs (OR 8.19; 95%CI 1.04 to 62.50). Nevertheless, at the multivariate analysis the effect of sex was lost, because of the confounding effect of the distance from the coastline. Then, adjusting for the other variables in the multivariate model, dogs more at risk in the Roma Province were the ones living within 20 Km to the coastline (OR 10.43; 95%CI 1.23 to 88.71).
Discussion
In the present study canine filariosis was detected for the first time in the Lazio Region. Its prevalence is higher along the seacoast territory and, inner land, at the boundary with the Umbria Region. The etiological agent of the infections was D. repens, species before reported in the Region only in foxes [18]. The absence of notification of this filarial worm in dogs, in spite of its report in wild animals and in dogs living in the neighbouring Regions [8,10] could be due to the recent passage of the parasite from wild to domestic environment, or could be related to the low pathogenicity of the nematode that usually make difficult, also to trained veterinarians, to suspect the infection, unless during specific survey.
D. repens is the species more prevalent also in the canine population of the bordering Tuscany [17], and it is well represented in the adjacent Umbria and Campania Regions, where D. immitis and A. reconditum, respectively, are the predominant filarial nematodes [8,10]. Infection rates observed, ranging from 1.03 to 12.5%, are in general agreement with those reported in the aforementioned Regions, even if data regarding Provinces other than Roma have to be considered only indicative, because coming from only one kennel per province and few other extra samples.
About the risk factors evidenced, some of the our data fit in with the conclusions of previous studies (no influence of hair length and sex), [4,8,10], whereas the influence of the age of the dog, not evidenced in this study, has been considered by the Authors cited above in a discordant way. However, the low prepatence times observed in experimental infections suggest that higher infection rates in adult individuals are simply related to the longer exposition times. Unfortunately, the typology of the dog population tested (kennels and urban dogs) hampered the investigation on other potential risk factors like the use of the animals (hunting practice?), the presence of a nocturnal shelter and the number of hours spent on open air. Anyway, low infection rates evidenced would have made difficult such an analysis.
The large confidence intervals of the OR at the multivariate analysis is an expression of the uncertainty of the risk association between dirofilariosis and coastal provenience of dogs evidenced in this study. This result is likely due to the low number of positve dogs among the hundreds tested in this survey. However, the higher prevalence of dirofilariosis in dogs living near the coast was already detected [24]. The apparent higher risk for those animals, can't be explained by a different Culicidae fauna, since the known vector species, Cx. pipiens and Ae. albopictus, are both widespread in the territory studied. The difference should be recognized in climatic conditions, that can favour the fast development of large mosquito population and, over all, can be more suitable for a fast development of larval stages of D. repens in mosquitoes and for their transmission [25], as suggested to explain also the higher infection rates by D. immitis along river valley [26].
This survey failed to evidence D. immitis, A. reconditum and A. dracunculoides. The absence of D. immitis as an autochthonous etiological agent of infection is indirectly confirmed by the results of the post mortem examination of about 550 dogs coming from the province of Roma in the period 2001–2003, regularly inspected in the hearth cavities, which proved all negative (Pathology Department of the Istituto Zooprofilattico Sperimentale, pers. comm.). Nevertheless, recent studies showed that canine filariosis by both species is increasing its geographic spread. In particular, the evident progression along river valleys recorded in the Umbria Region [8] and confirmed by the moderate prevalence of dirofilariosis detected in our study in the Rieti and Viterbo provinces, bordering to that region, is a threat of an imminent introduction of cardiopulmonary dirofilariosis also in the Lazio Region and in the Roma territory. In fact, even if laboratory data suggested that infection by D. repens may play a protective role against infection by D. immitis, so "defending" the dog from more pathogenic species [27], the recent report of D. immitis in mosquitoes of the Region caught in the urbanized area [28] is challenging.
Dogs of the Region turned out also A. reconditum and A. dracunculoides free. The first species, largely present in the adjacent Regions, and the second, observed in foxes of the province of Roma [29], were undetected by microscopy and "filarial" primers during this survey. Their absence could be explained by the sampled dog population, practically lacking of hunting and shepherd dogs, probably more exposed to the arthropod vectors.
Finally, the survey carried out, reporting the presence of the slightly pathogen D. repens in the dog population, reassures on the sanitary status of the animals but, at the same time, is alarming for humans that could be infected by zoo-antropophilic vector mosquitoes like C. pipiens and Ae. albopictus. In fact, D. repens is the species to date recognized in human infections reported in Italy. The zoonotic impact of this parasitic infection has been recently evaluated [30]: more than one hundred of human infections have been reported in 5 years from the whole country. Therefore, a specific education program in preventing and detecting the disease in dogs as well as in humans, for citizen, doctors and veterinary, has to be encouraged.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PS partecipated at the design of the study and drafted the manuscript. SG carried out the molecular tests. MDP contribute to acquire the samples and to carry out microscopic examinations. MS designed the sampling and performed the statistical analysis. FS coordinated the working group and revised critically the manuscript. GC helped to draft the manuscript and revised it. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Financial support was provided by Italian Ministry of Health (Ricerca corrente 2000). Authors are very grateful to Miss Anna Maria Cosentino, to Veterinary Health Service (ASL RMD), to Dr. Paolo Fulignati and Emma Caputo for the clinical and laboratory diagnosis kindly provided.
==== Refs
Rossi L Pollono F Meneguz PG Gribaudo L Balbo T An epidemiological study of canine filarioses in north-west Italy: what has changed in 25 years? Vet Res Commun 1996 20 308 315 8865572 10.1007/BF00366536
Poglayen G Martini M Bomben L Roda R An updating of the occurrence of canine heartworm disease in northern Italy Vet Res Commun 1996 20 303 307 8865571 10.1007/BF00366535
Petruschke G Rossi L Pollono F Genchi C Sulle filariosi canine in Provincia di Varese e nel basso Canton Ticino Parassitologia 1998 40 133
Pietrobelli M Soldano F Frangipane di Regalbono A Bandera C Filariosi cardio-polmonare del cane in Friuli-Venezia Giulia Obiettivi e documenti veterinari 1998 1 63 68
Genchi C Kramer LH Prieto G Simón F, Genchi C Epidemiology of canine and feline Dirofilariasis: a global view Heartworm infection in humans and animals 2001 Ediciones Universidad Salamanca 121 133
Doby JM Couatarmanac'h A Aznar C Filarioses canines par Dirofilaria immitis (Ledy, 1856) et D. repens (Raillet & Henry 1911) dans l'ouest de la France Bulletin Société Française Parasitologie 1986 4 229 233
Ortega-Mora LM Gomez-Bautista M Rojo-Vasquez F Rodenas A Guerriero J Survey of the prevalence of canine filariasis in Spain Prev Vet Med 1991 2 63 68
Piergili Fioretti D Diaferia M Grelloni V Maresca G Canine filariosis in Umbria: an update of the occurrence one year after the first observation of autochtonous foci Parassitologia 2003 45 79 83 15267001
Pennisi MG Furfaro N La filariosi del cane in Sicilia Obiettivi e documenti veterinari 1997 26 30
Cringoli G Rinaldi L Veneziano V Capelli G A prevalence survey and risk analysis of filariosis in dogs from the Mt. Vesuvius area of southern Italy Vet parasitol 2001 102 243 52 11777604 10.1016/S0304-4017(01)00529-5
Kanev I Kamenov I Gauchev G Prelezov P Tzvetkov Y Toncheva V Halacheva M Georgieva D Vuchev D Tanchev T Slavova G Dirofilaria repens and Dirofilaria immitis in animals and humans in Bulgaria VII EMOP Abstracts, Parassitologia 1996 38 1 2
Bucklar H Scheu U Mossi R Deplazes P Breitet sich in der Sudschweiz die Dirofilariose beim Hund aus? Schweiz arch tierheilkd 1998 140 255 260 9646715
Himonas C Canine filariasis caused by Dirofilaria repens in Greece VII EMOP Abstracts, Parassitologia 1996 38 1 2
Chauve C Importance in France of the infestation by Dirofilaria (Nochtiella) repens in dogs VII EMOP Abstracts, Parassitologia 1996 38 1 2
Lucientes J Castillo JA Present status of Dirofilaria repens in Spain VII EMOP Abstracts, Parassitologia 1996 38 1 2
Cancrini G Allende E Favia G Bornay F Anton F Simon F Canine dirofilariosis in two cities of southeastern Spain, Vet parasitol 2000 92 81 86 10936548 10.1016/S0304-4017(00)00270-3
Marconcini A Magi M Incidenza della filariosi canina nella Maremma Grossetana Annuali Facoltà Medicina Veterinaria, Pisa 1991 44 153 156
Mantovani A Tsaglis A Ricerche sulla presenza di Dirofilaria repens nelle volpi di alcune province dell'Italia centrale Parassitologia 1965 7 123 126
Rawlings CA Saunders WB Hearthworm in dogs and cats 1986 212 218
Euzeby J Resultats du diagnostic hématologique. Critères généraux de diagnose des microfilaires Diagnostic Expérimental des helminthoses animales Livre 1 1981 Informations Techniques des Services Vétérinaires; Ministère de l'Agriculture Paris 303 311
Favia G Lanfrancotti A della Torre A Cancrini G Coluzzi M A PCR-based method to identify D. repens and D. immitis Parasitology 1996 113 567 571 8939053
Xie X Bain O Williams SA Molecular phylogenetic studies on filarial parasites based on 5S ribosomal spacer sequences Parasite 1994 1 141 151 9140481
Thompson JD Gibson TJ Plewniak F Jeanmougin F Higgins DG The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools Nucleic Acids Res 1997 25 4876 4882 9396791 10.1093/nar/25.24.4876
Barriga O Myron G Handbook series in zoonoses. Section C: Parasitic Zoonoses 1982 II Schultzs, Boca Raton, Florida, CRC Press
Cancrini G Sun Yanchang Della Torre A Coluzzi Influenza della temperatura sullo sviluppo larvale di Dirofilaria repens in diverse specie di zanzare Parassitologia 1988 38
Genchi C Vezzoni A Di Sacco B Baroni G La dirofilariosi cardiopolmonare del cane: epidemiologia e profilassi Veterinaria 1991 5 83 90
Genchi C Solari Basano F Bandi C Di Sacco B Venco L Vezzoni A Cancrini G Factors influencing the spread of heartworms in Italy: interaction between Dirofilaria immitis and Dirofilaria repens Proceeding Heartworm Symposium 1995 American Heartworm Society 65 71
Cancrini G Romi R Gabrielli S Toma L Di Paolo M Scaramozzino P First finding of D. repens in a natural population of Aedes albopictus Med vet entomol 2003 17 448 451 14651660 10.1111/j.1365-2915.2003.00463.x
Cancrini G Filariasi in Vulpes vulpes della Provincia di Roma Parassitologia 1986 28 210 211
Pampiglione S Rivasi F Human dirofilariasis due to Dirofilaria (Nochtiella)repens :an update of world literature from 1995 to 2000 Parassitologia 2000 42 231 254 11686084
|
16185352
|
PMC1261266
|
CC BY
|
2021-01-04 16:28:15
|
no
|
BMC Infect Dis. 2005 Sep 26; 5:75
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-75
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-761618535310.1186/1471-2334-5-76Research ArticleUse of a nested PCR-enzyme immunoassay with an internal control to detect Chlamydophila psittaci in turkeys Van Loock Marnix [email protected] Kristel [email protected] Trudy O [email protected] Guido [email protected] Bruno M [email protected] Daisy [email protected] Department of Biosystems, Catholic University of Leuven, Kasteelpark Arenberg 30, 3001 Heverlee, Belgium2 Department of Molecular Biotechnology, Ghent University, Coupure Links 653, 9000 Gent, Belgium3 Department of Health and Human Services, National Center for Infectious Diseases, Centres for Disease Control and Prevention, Public Health Service, Atlanta, Georgia 30333, USA4 Department of Virology, Parasitology and Immunology, Ghent University, Salisburylaan 133, 9820 Merelbeke; Belgium2005 26 9 2005 5 76 76 3 6 2005 26 9 2005 Copyright © 2005 Van Loock et al; licensee BioMed Central Ltd.2005Van Loock 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
Laboratory diagnosis of Chlamydophila psittaci, an important turkey respiratory pathogen, is difficult. To facilitate the diagnosis, a nested PCR-enzyme immunoassay (PCR-EIA) was developed to detect the Cp. psittaci outer membrane protein A (ompA) gene in pharyngeal swabs.
Methods
The fluorescein-biotin labelled PCR products were immobilized on streptavidin-coated microtiter plates and detected with anti-fluorescein peroxidase conjugate and a colorimetric substrate. An internal inhibition control was included to rule out the presence of inhibitors of DNA amplification. The diagnostic value of the ompA nested PCR-EIA in comparison to cell culture and a 16S-rRNA based nested PCR was assessed in pharyngeal turkey swabs from 10 different farms experiencing respiratory disease.
Results
The sensitivity of the nested PCR-EIA was established at 0.1 infection forming units (IFU). Specificity was 100%. The ompA nested PCR-EIA was more sensitive than the 16S-rRNA based nested PCR and isolation, revealing 105 out of 200 (52.5%) positives against 13 and 74 for the latter two tests, respectively. Twenty-nine (23.8%) out of 122 ompA PCR-EIA negatives showed the presence of inhibitors of DNA amplification, although 27 of them became positive after diluting (1/10) the specimens in PCR buffer or after phenol-chloroform extraction and subsequent ethanol precipitation.
Conclusion
The present study stresses the need for an internal control to confirm PCR true-negatives and demonstrates the high prevalence of chlamydiosis in Belgian turkeys and its potential zoonotic risk. The ompA nested PCR-EIA described here is a rapid, highly sensitive and specific diagnostic assay and will help to facilitate the diagnosis of Cp. psittaci infections in both poultry and man.
==== Body
Background
Avian chlamydiosis is caused by the obligate intracellular Gram-negative bacterium Chlamydophila psittaci (formerly Chlamydia psittaci). Currently, seven genotypes of Cp. psittaci are known to infect birds [1-3]. Avian chlamydiosis in birds is usually systemic and occasionally fatal. The clinical signs vary greatly in severity and depend on the species, age of the bird and the strain of Cp. psittaci. Avian chlamydiosis can produce lethargy, hyperthermia, abnormal excretions, nasal and eye discharges, and reduced egg production. Mortality rates range up to 30% [4]. Avian chlamydiosis occurs worldwide, with the incidence and distribution varying greatly with the species of bird and the serotype of the chlamydial organism. In the past, chlamydiosis in turkeys was thought to be limited to the United States and to free-ranging flocks. Most outbreaks in US turkeys were explosive, involving one or more flocks [5-10]. Nowadays, the increase in confinement-rearing of turkeys and the prevention of wild birds flying in and out the turkey houses seems to contribute to a decrease of severe outbreaks. Probably, the situation is comparable to the one in Europe where, at present, Cp. psittaci is nearly endemic in Belgian, German and probably French turkeys [11-13]. However, devastating, explosive outbreaks with high mortality rates occur occasionally, whereas present outbreaks are mostly characterized by respiratory signs without mortality [4]. Nevertheless, Cp. psittaci causes important economical losses as a primary pathogen and trough it's pathogenic interaction with other respiratory pathogens like the avian pneumovirus (APV) and Ornithobacterium rhinotracheale (ORT) [13]. Cp. psittaci is also a threat to public health as this zoonotic agent can infect humans and precautions should be taken when handling infected birds or contaminated materials [14-17]. Human infections are common following handling or processing of infected turkeys or ducks [2,7,8,18]. Most infections are through inhalation of infectious aerosols and subsequently processing plant employees, farm workers, veterinarians and poultry inspectors are at risk. However, personnel who were employed to further process turkey meat could also become infected [19].
Thus, diagnosis is essential. In contrast to cell culture and serology, antigen detection methods like micro-immunofluorescence and PCR provide a more rapid, specific and sensitive alternative for identification of Cp. psittaci infection. However, currently described PCR assays for birds use either labour intensive and/or insensitive post PCR detection methods. A PCR-enzyme immunoassay (PCR-EIA) would circumvent this problem. At the moment, we are not aware of a nested PCR- enzyme immunoassay (PCR-EIA) for demonstrating Cp. psittaci infection, although the method has been successfully used for C. pneumoniae detection in human respiratory specimens [20,21].
The objective of the present study was to develop and evaluate a rapid and simple EIA for semi-quantitative detection of the amplified Cp. psittaci outer membrane protein A (ompA) gene, included with an internal inhibition control to eliminate possible false positive results during field sample analysis.
Methods
Specimens
In the fall of 2001, 200 fattening turkeys from 10 different farms in Belgium (8 farms) or in Northern France (2 farms) were examined at slaughter for the presence of Cp. psittaci. All turkeys had been vaccinated against Newcastle disease (NCD) (Nobilis® ND LaSota; Intervet International, Boxmeer, The Netherlands) and in 7 out of 10 farms turkeys had also been vaccinated against APV (Nobilis® RTV; Intervet International). Farmers provided information about clinical symptoms throughout the rearing period. All farms had experienced one or more periods of respiratory disease.
Pharyngeal swabs were collected from 20 ad randomly selected turkeys on each turkey farm. Of each turkey there was taken 1 sample by using cotton tipped aluminium shafted swabs (Fiers, Kuurne, Belgium) in Cp. psittaci transport medium [22] consisting of: 0.2 M sucrose (VWR International, Haasrode, Belgium); 0.015 M Na2HPO4 (VWR International), 0.01 M NaH2PO4 (VWR International) and 20% inactivated foetal calf serum (Integro, Leuvenheim, The Netherlands). Swabs were shaken vigorously for 1 hour and centrifuged (10 min, 2790 × g, 4°C). One millilitre of supernatant was provided with 1% streptomycin sulphate (10 mg/ml; Invitrogen), 2% vancomycin (5 mg/ml; Glaxo Smith Kline) and 1.6% fungizone (250μl/ml; Invitrogen) and subsequently used for Cp. psittaci isolation.
Generation of the internal control
The internal inhibition control was constructed starting from the pcDNA1 vector in which the ompA gene of a Cp. psittaci serovar D strain 92/1293 was inserted (Fig 1 &2) [23]. First, a fragment of 231 bp of the ompA gene was amplified using primers ML-BbrpI-F01 and ML-Bbrp1-R01 (table 1), which provided a BbrpI restriction site at their 5' end for subsequent cloning. The PCR reaction was performed using Pfu DNA polymerase in 50 μl reactions containing dNTP's (0.2 μM final concentration), Pfu buffer (10x), ML-BbrpI-F01 and ML-Bbrp1-R01 (0.5 μM final concentration), DMSO (7.5 %), Pfu DNA Polymerase (2.5 U, Stratagene, La Jolla, USA) and 2 μg plasmid DNA. After an initial denaturation at 95°C for 1 minute, 30 cycles of 30 seconds at 95°C, 30 seconds at 58°C and 1 minute at 72°C, followed by a final elongation at 72°C for 10 min, were performed. PCR products were subjected to electrophoresis on a 1.2% agarose gel stained with ethidium bromide and photographed under UV illumination. Size was determined using Smart Ladder (Eurogentec, Seraing, Belgium). PCR products were purified using the Qiaquick PCR Purification kit (Qiagen) and ligated into the pPCR-script™ Amp SK (+) vector (Stratagene, La Jolla, USA), as described by the manufacturer. Next, the vector was transformed into Epicurian coli XL10-Gold Kan ultracompetente cells (Stratagene) using heat shock. Clones were selected on Luria-Bertoni (LB) medium containing ampicilin (100 μg/ml) and grown in microtiter plates for 2 hours. The presence of the insert was confirmed by PCR clone analysis. Therefore, 5 μl of each clone was subjected to PCR in a 50 μl reaction mixture containing 50 mM KCl, 20 mM Tris-HCl (pH 8.3), 2 mM MgCl2, 0.1% Tween20, 200 μM each dNTP, 1.25 μM of each inner ompA primer (table 1) and 0.1 U SuperTaq polymerase (5 U/μl). After an initial denaturation at 95°C for 5 minutes, 20 cycles of one min at 95°C, two minutes at 55°C and three minutes at 72°C, with a final extension at 72°C for 5 min, were performed. To ensure PCR accuracy, the construct was sequenced using the ABI PRISM Bigdye™ Terminator Cycle Sequencing Ready Reaction Kit (ABI, Foster City, USA), following the manufacturers' manual. Sequencing samples were analyzed on the ABI PRISM 377 DNA sequencer (Perking Elmer).
Figure 1 Generation of the internal control using primers ML-Bbrp1-F01 and ML-Bbrp1-R01.
Figure 2 Location of the outer and inner primers in the ompA gene.Numbering according to ompA sequences in Genebank
Table 1 Oligonucleotides used in this study
Oligonucleotide Length (bp) Sequence (5'-3')
Sense outer 16 rRNA 18 ACG GAA TAA TGA CTT CGG
Anti-sense outer 16S rRNA 18 TAC CTG GTA CGC TCA ATT
Sense inner 16S rRNA 21 ATA ATG ACT TCG GTT GTT ATT
Anti-sense inner 16S rRNA 20 TGT TTT AGA TGC CTA AAC AT
ML-BbrpI-F01 29 GCC ACG TGC GTC TGC AAC ACT CAA ATA TC
ML-BbrpI-R01 28 GGC ACG TGC AGT TGT AAG AAG TCA GAG T
Sense outer ompA 21 CCT GTA GGG AAC CCA GCT GAA
Anti-sense outer ompA 22 GGT TGA GCA ATG CGG ATA GTA T
Fluorescein-sense inner ompA 17 GCA GGA TAC TAC GGA GA
Biotin-antisense inner ompA 18 GGA ACT CAG CTC CTA AAG
Positive clones were grown overnight in 4 ml LB medium containing ampicilin (100 μg/ml) and subjected to the Qiaprep Spin Miniprep kit (Qiagen) to obtain purified plasmid. Using BbrpI, the 231 bp fragment was cut out of the pPCR-script™ Amp SK (+) vector and ligated into the dephosphorylated Bbrp1 site, situated within the ompA gene of the pcDNA1::MOMP vector [23]. Thus, the Bbrp1 restriction site is located within the target sequence of the inner primers and subsequently, nested PCR amplification of the inhibition control resulted in a PCR product of 703 bp (fig. 1). E. coli strain MC1061/P3 was transfected by electroporation (Gene Pulser, Bio-Rad). Again, selected clones were subjected to PCR clone analysis to asses the presence of the insert and its sequence was determined by the dideoxy chain terminating method, as described above.
Preparation of positive control DNA
Cp. psittaci strains genotype A to F strains were propagated in cycloheximide-treated BGM cells, as described elsewhere [24]. Bacteria were harvested at approximately 72 hours by disrupting the cells by subsequent freezing and thawing, followed by sonication and differential centrifugation (Urografin 76%). Purified elementary bodies were pelleted, washed, resuspended in sucrose-phosphate-glutamate buffer, and stored in aliquots at -70°C. For determination of bacterial titres, BGM monolayers grown on glass cover slips (Chlamydia Trac Bottles, International Medical) were infected and stained by the IMAGEN™ direct immunofluorescence assay [24]. Inclusion forming units (IFU) were determined by counting the numbers of inclusions cultured in duplicate in Chlamydia Trac Bottles using 10-fold serial dilutions of purified EBs. Cp. psittaci titres were expressed as IFU per millilitre and IFU quantitated in this manner were used as the positive-control DNA in PCR assays.
OmpA nested PCR-EIA
Clinical specimens included pharyngeal swabs from turkeys taken at slaughter. These specimens as well as positive control DNA were prepared for ompA nested PCR-EIA by the STD DNA extraction method, performed as followed: Cp. psittaci suspensions or turkey specimens were pelleted at 13,000 x g, resuspended in 198 μl STD buffer (0.01 M Tris-HCl [pH 8.3], 0.05 M KCl, 0.0025 M MgCl2.6H20, 0.5% Tween20) and 2 μl proteinase K (20 mg/ml stock solution; Sigma Chemical Co.). The specimens were incubated at 56°C for one hour and subsequently heated at 100°C for 10 min. Samples for both PCR's were prepared in a class II laminar flow hood, and amplification and analysis of PCR products were performed in separate locations.
The nested ompA PCR-EIA was developed using external and internal primers (table 1) generating a biotin-fluorescein dual labelled internal PCR product of 472 bp. Both inner and outer sense primers are located within the first conserved domain (CD1) of ompA, whereas the inner anti-sense primer is located in CD3 and the outer anti-sense primers overlaps CD4 and variable domain 4 (fig. 2). First round PCR occurred in 50 mM KCl, 20 mM Tris-HCl (pH 8.3), 2 mM MgCl2, 0.1% Tween20, 200 μM each dNTP, 1.25 μM each external primer (table 1) and 0.1 U SuperTaq polymerase (5 U/μl). After an initial denaturation at 95°C for 5 minutes, 20 cycles of one min at 95°C, two min at 59°C and three min at 72°C, with a final extension at 72°C for 5 min were performed. Second round amplification was performed under similar conditions, using labelled internal primers (each 10 μM; table 1), adapted annealing temperature (47°C) and adapted number of cycles (25). Subsequently, nested PCR generated a fluorescein/biotin dual-labelled product of 472 bp. To minimize false-positive results, each step of the nested PCR was performed in physically separated places.
To allow colorimetric detection of the ompA PCR products, 50 μl of the PCR product diluted 1/10 in PBS supplemented with 3% BSA was transferred in duplicate to streptavidin-coated microtiter plates (2 μg/well for 3 hours at 37°C) and incubated at 37°C for one hour. Non-specific binding places were blocked overnight (4°C) with 5% BSA in PBS. Subsequently, the plates were washed twice with PBS and incubated (1 hour, 37°C) with a horseradish peroxidase labelled anti-fluorescein antibody (Invitrogen), diluted 1/1000 in PBS supplemented with 3% BSA. Following incubating and washing with PBS, the ABTS substrate solution (2,2' azino-di-3-ethylbenzothiazoline sulphonate, KPL) was added to the wells. Absorbencies were read at 450 nm after incubating for 30 minutes at 37°C (TiterTek MultiskanR Plus, MKII, TechGen Internatonal). Three positive controls consisting of serial 10-fold dilutions of PCR products generated from 5 IFU of Cp. psittaci and five negative controls (water) were included in each assay. Results were positive if the absorbance exceeded the cut off value of the mean of negative controls plus three times the standard deviation. Nested PCR-EIA negative samples were re-tested after adding 10 ng internal inhibition control and visualized by gel electrophoresis to assess possible inhibition.
The sensitivity of the PCR was evaluated by testing 10-fold serial dilutions of DNA extracted from purified elementary bodies of Cp. psittaci strains 92/1293 [2]. Next, diagnosis of six reference strains of Cp. psittaci serovars A-F was assessed (table 2). The specificity was determined by testing DNA extracted from other bacterial species commonly found in the avian respiratory tract and avian respiratory tract tissue, originating from Cp. psittaci negative specific pathogen free turkeys (CNEVA, Ploufragan, France). The following micro-organisms were tested for cross-reactivity in the PCR assay with both the first- and second-step PCR primers: Acinetobacter species, Aspergillus flavus, Candida albicans, Enterococcus faecelis, Escherichia coli, Klebsiella species, Mycobacterium avium, Mycoplasma gallisepticum, Mycoplasma meleagridis, Ornithobacterium rhinotracheale, Pasteurella species, Proteus mirabilis, Pseudomonas species, Salmonella enteritidis, Salmonella gallinarum, Salmonella pullorum, Staphylococcus species, Streptococcus species and Xanthomonas maltophila. A Cp. psittaci positive control of 1 IFU was included in every test to verify that the PCR was working.
Table 2 Cp. psittaci reference strains
Strain Year Country Host Serovar Reference
VS 1 1985 USA, Georgia Amazona sp. A [34]
CP3 1957 USA, Calofornia Columba livia B [35]
GD 1960 Germany Anas platyrhynchos C [36]
NJ1 1954 USA, New Jersey Meleagris gallopavo D [37]
MN 1934 USA, California Homo sapiensa E [38]
VS225 1991 USA, Texas Parakeet F [39]
a Probably originated from birds, isolated in ferrets from human [39].
16S-rRNA nested PCR – gel electrophoresis
The performance of the ompA nested PCR-EIA was compared to those of isolation and another nested PCR, targeting the 16S rRNA gene [25]. Cp. psittaci suspensions or turkey specimens were prepared for PCR by the QiaAmp Blood kit (Qiagen Inc., Chatworth, Califormia) adapted by [25]. Genus-specific first-step primers and species-specific second step primers generated PCR products of 436 bp and 127 bp, respectively (table 1). Amplification products were visualized by gel electrophoresis (1.5% Nusieve GTG agarose, FMC Bioproducts, Rockland, Maine). PCR negatives were spiked with 5 IFU of Cp. psittaci to control for the presence of inhibitors. The limit of detection of this 16S-rRNA-based PCR was 5 IFU as previously described [25]. Samples for both PCR's were prepared in a class II laminar flow hood, and amplification and analysis of PCR products were each performed in separate locations.
Comparison to Cp. psittaci isolation
Pharyngeal swabs were examined for the presence of viable Cp. psittaci by isolation in cycloheximide-treated Buffalo Green Monkey (BGM) cells. Swabs were shaken at 4°C for 1 hour and centrifuged (10 minutes, 2790 × g, 4°C). The supernatant was used for Cp. psittaci isolation in BGM cells and subsequent identification using the IMAGEN™ direct immunofluorescence assay (DakoCytomation, Denmark), as previously described [24]. All inoculated monolayers were stained at 6 days post inoculation. Inclusion-negative cultures were passaged once. After adding an equal volume of sucrose phosphate glutamate (SPG; [25]) and freezing at -80°C, cultures were thawed, cell suspensions were sonicated and centrifuged once (2000 x g). Supernatant was inoculated in duplicate onto new BGM monolayers as described elsewhere [24]. Staining was performed at 3, and if negative at 6 days post inoculation.
Validation
A specimen was considered positive if culture positive. In addition, a culture-negative, but 16S rRNA-based PCR positive specimen was considered to be a true positive only if it could be verified by ompA-based PCR.
Results
Development of the nested PCR-EIA
Optimizing PCR conditions was performed using STD extracted DNA of Cp. psittaci serovar D strain 92/1293. Initial PCR with temperature gradients were performed with either inner or outer primer sets separately to determine optimal annealing temperature for both primer sets. The optimal annealing temperature for outer and inner primer sets was determined at 59°C and 47°C, respectively (data not shown). Next, optimal primer dilutions were tested to obtain a single band as nested PCR product, after visualization on 1.2% agarose gel (fig. 3). Hereto, external primers were used at 0.625 μM and internal primers at 10 μM. First and second round PCR amplification with the outer and inner primers resulted in PCR products of 872 bp and 472 bp, respectively. After amplification biotinylated PCR products were immobilized to streptavidin-coated microtiter wells and detected with anti-fluorescein peroxidase conjugate and a colorimetric substrate. Next, optimal enzyme immunoassay conditions were realised, among others by diluting the dual labelled PCR product 1/10 in dilution buffer (PBS + 3% BSA + 2% IgG free horse serum). All incubation steps and reaction components of this EIA were optimized prior to use with pharyngeal swabs.
Figure 3 Different conditions observed in the nested PCR-EIA analysis of Cp. psittaci in turkey field samples. Lane 1: molecular marker (BenchTop 1 kb DNA ladder; Promega); lane 2: Cp. psittaci positive sample, showing one band (472 bp) diagnostic for Cp. psittaci and a band (703 bp) for the internal inhibition control; lane 3: Cp. psittaci negative sample, showing only the internal inhibition control band; lane 4: a sample with inhibitory substances lacking both the Cp. psittaci-specific band and the internal inhibition control band; lane 5: negative control, free from Cp. psittaci DNA and internal control DNA.
Sensitivity
Following definition of optimal reagent and reaction conditions, sensitivity and specificity of the ompA nested PCR was determined. The STD DNA-extraction was performed on 108 IFU and tenfold dilutions of the purified DNA were subjected to the nested PCR and visualised on a 1.2% agarose gel. This resulted in a final nested PCR product of 472 bp and a detection limit of 10-2 IFU (fig. 4A). Subjecting the dual-labelled nested PCR product to the EIA, resulted also in a detection limit of 10-2 IFU (fig. 4B) and there was a linear relationship between the measured absorbance and the tenfold dilution series. However, when the tenfold dilution series of Cp. psittaci elementary bodies was made prior to the STD DNA-extraction, sensitivity decreased one log to 10-1 IFU. Similarly, when field samples, which tested negative by nested PCR, were spiked with the serial tenfold dilutions, sensitivity decreased to 10-1 IFU.
Figure 4 (A) Nested PCR-EIA analysis on a tenfold serial dilution of Cp. psittaci strain 92/1293. (B) Visualisation of the tenfold serial dilution on a 1.2% agarose gel following nested PCR. 103 IFU (lane 1) until 10-2 IFU (lane 6). Lane 7 shows the PCR results on a negative control. Lane 8: Phage lambda PstI fragments as size marker.
Amplification of chlamydial DNA and the internal inhibition control was achieved with the ompA inner and outer primer sets, as the additional DNA fragment for the inhibition control was inserted within the target sequence of the ompA inner primer set. Therefore, the nested PCR amplification of Cp. psittaci cultures or inhibitory substance-free clinical specimens, which were positive for Cp. psittaci exhibited two bands on ethidium-stained agarose gel electrophoresis: one band (472 bp) diagnostic for Cp. psittaci and a control band (703 bp) for the internal inhibition control (fig 3). Inhibitory substance-free clinical specimens negative for Cp. psittaci contained only the control band, indicating that no detectable inhibitors were present and that biochemical conditions were optimal for PCR amplifications. When inhibitory substances were present in field samples, no bands were detected on ethidium-stained agarose gel. Adding 10 ng of inhibition control to the nested PCR mixture was determined as the optimal condition to assess inhibition in field samples (fig 3).
Specificity
The nested PCR-EIA was able to detect al 6 tested Cp. psittaci reference strains (table 2), whereas strains of Chlamydia trachomatis, Chlamydophila pneumoniae, Chlamydophila abortus and Chlamydophila felis remained undetected. Furthermore, avian respiratory tract tissue originating from Cp. psittaci negative specific-pathogen-free turkeys (CNEVA, Ploufragan, France) and a wide range of non-chlamydial bacteria were tested and showed no cross-reactivity with: Acinetobacter species, Aspergillus flavus, Candida albicans, Enterococcus faecelis, Escherichia coli, Klebsiella species, Mycobacterium avium, Mycoplasma gallisepticum, Mycoplasma meleagridis, Ornithobacterium rhinotracheale, Pasteurella species, Proteus mirabilis, Pseudomonas species, Salmonella enteritidis, Salmonella gallinarum, Salmonella pullorum, Staphylococcus species, Streptococcus species and Xanthomonas maltophila. A Cp. psittaci positive control of 1 IFU was included in every test to verify that the PCR was working.
Analysis of pharyngeal specimens
Two hundred turkeys from 10 different farms in Belgium (8 farms) or in Northern France (2 farms) were examined at slaughter for the presence of Cp. psittaci. All farms had experienced one or more periods of respiratory disease. All samples have been analysed for the presence of Cp. psittaci by isolation in BGM cells, nested PCR-EIA and 16S rRNA nested PCR (table 3). The pharyngeal swabs were inoculated onto cycloheximide-treated BGM cells. Cp. psittaci was isolated from 54 specimens (27%) after the first inoculation and from 20 additional samples (10%) following one passage. Thus, 74 out of 200 (37%) specimens revealed to be culture positive. They were all confirmed as culture positive by ompA nested PCR-EIA analysis of the infected BGM monolayers. One hundred and twenty six samples remained negative, notwithstanding an additional 6 days passage on BGM cells. OmpA nested PCR-EIA was able to detect chlamydial DNA successfully in 105 on 200 (52.5%) pharyngeal swabs. However, 29 out of 122 (23.8%) PCR-EIA negatives clearly demonstrated inhibition showing no internal control band on the agarose gel. Seventeen out of 29 samples containing inhibitors became positive after prior 1/10 dilution in PCR buffer. Specimens that still showed inhibition were subjected to phenol-chloroform extraction and ethanol precipitation to further purify the DNA and were retested. Ten tested positive for Cp. psittaci, but 2 of 29 specimens continued to show inhibition. Thus, finally 105 out of 200 (52.5%) pharyngeal swabs tested positive by the ompA nested PCR-EIA. Surprisingly, the 16S rRNA-based PCR could only confirm 13 out of 105 ompA PCR positives revealing a total of 6.5% positive turkeys. Nine of these 13 positives could be confirmed by isolation while the remaining 4 were negative by isolation but positive by ompA PCR, indicating that they were true positives.
Table 3 Results of nested PCR's on turkey pharyngeal specimens compared to isolation as a reference test
Isolation N OmpA PCR 16S rRNA PCR
positive negative positive negative
Negatives 126 31 95 4 122
Positives 74 74 0 9 65
Total 200 105 95 13 187
Referring to isolation as a reference, all 10 examined farms were Cp. psittaci positive at slaughter. The same was true when looking at the ompA nested PCR results. Although apparently less sensitive, the 16S rRNA PCR detected Cp. psittaci in 50% of the examined farms.
Discussion
A nested PCR-EIA based on the detection of the ompA gene was developed and evaluated for the diagnosis of chlamydiosis in turkeys. Nested PCR resulted in a 5' fluorescein and 3' biotin labelled ompA fragment of 472 bp which was subsequently detected in an enzyme immunoassay. Although the designed inner and outer anti-sense primers showed few nucleotide mismatches as compared to the ompA sequences of Cp. psittaci genotype C, D and F reference strains, amplification of Cp. psittaci genotypes A to F strains consistently resulted in the anticipated nested PCR-EIA product. The nested PCR-EIA was 100% specific as all Cp. psittaci genotypes were detected, but no C. trachomatis, C. pneumoniae, C. abortus or C. felis DNA. Additionally, no cross-reactivity was observed with other bacterial respiratory pathogens commonly found in the avian respiratory tract or with turkey respiratory tract DNA.
Nested PCR was chosen in order to obtain high sensitivity and specificity. Amplification of internal control DNA helped us in confirming true-negative PCR results by ruling out the presence of inhibitors of DNA amplification. Adding 10 ng of the internal inhibition control did not comprise sensitivity, as 10-2 IFU of all Cp. psittaci genotype reference strains was detected. However, when field samples, which tested negative in the nested PCR-EIA were spiked with the tenfold DNA dilutions, sensitivity decreased to 10-1 IFU, probably due to the higher amount of background DNA.
In the present study, 200 commercial turkeys, originating from 10 different farms in Belgium (8 farms) or Northern France (2 farms) were sampled at slaughter to examine for the presence of Cp. psittaci. Isolation in BGM cells revealed 74 (37%) positives. Application of the ompA nested PCR-EIA on pharyngeal DNA could confirm all culture positive results. Moreover, the ompA nested PCR-EIA detected 31 additional positives, resulting in a total of 105 (52.5%) Cp. psittaci positive turkeys. However, 29 (23.8%) of the 105 PCR-EIA positives were initially negative by the EIA and during retesting, when the internal control was added to the PCR mix, they demonstrated inhibition lacking the internal control band on an EtBr stained agarose gel. Yet, 17 samples became positive after prior 1/10 dilution of the specimen in PCR buffer. Moreover, after phenol-chloroform extraction and ethanol precipitation all but two of the sample became positive. Those two samples could not be diagnosed by the ompA nested PCR-EIA, as inhibitory substances could not be removed. Culture and 16S rRNA nested PCR for those 2 samples were also negative. However as the latter two tests have shown to be less sensitive and consequently, the presence or absence of Cp. psittaci in these 2 samples cannot be conclude. The present results clearly demonstrate the importance of using an internal control to help identify true-negatives when examining turkey pharyngeal swabs, as inhibition of DNA amplification seem to occur rather frequently in these specimens. Notwithstanding the presence of polymerase inhibitors, pharyngeal swabs still remain the first choice for sampling live birds. Pharyngeal specimens are preferred as cloacal shedding of Cp. psittaci is intermittent and, in contrast, the respiratory tract appears to be the last system to be cleared of infection. Furthermore, pathogenesis of Cp. psittaci revealed that lateral nasal glands can be infected for a extended period [26]. Secretions of these glands function to keep the mucosa moist and drainage of infected secretions into the pharyngeal cavity can serve as source for Cp. psittaci. Also secretions from the lung are expelled into the pharyngeal area [27].
Surprisingly, the 16S rRNA-based PCR could only confirm 13 out of 105 ompA nested PCR-EIA positives revealing only 6.5% positive turkeys. Discrepant results were probably not due to different extraction methods, as 30 16S-rRNA negative samples remained negative even after using the STD extraction method as for the ompA - based nested PCR.
The PCR-EIA turned out to be more sensitive than isolation in cell culture and more sensitive than the 16S rRNA-based nested PCR. The 16S-rRNA PCR primers have already been shown to be sensitive (5 IFU) and specific [25]. The ompA nested PCR-EIA is approximately 50 times more sensitive than the 16S-rRNA based PCR. The sensitivity of the nested PCR-EIA was also superior to isolation of Cp. psittaci in cell culture, which is in agreement with other reports [28-30]. Moreover, the nested PCR-EIA is easy, rapid, and less labour-intensive than isolation and non-viable Chlamydiaceae can be detected, due to the relative high stability of DNA. This allows less stringent demands on collection, transportation and storage of the samples, making the nested PCR-EIA an ideal diagnostic method for monitoring turkey flocks during processing.
Referring to isolation as a reference, all 10 examined farms were Cp. psittaci positive at slaughter. The same was true when looking at the ompA nested PCR results. Although apparently less sensitive, the 16S rRNA PCR still detected Cp. psittaci in 50% of the examined farms. Results are in concordance with previous reports, demonstrating the high prevalence of Cp. psittaci in Belgian turkeys [13]. Public health is here of concern, as poultry workers, veterinary surgeons and slaughterhouse employees are at risk of becoming infected by this zoonotic agent [9,15,16,19,31-33]. In the present study, 37% of the turkeys were still shedding infectious Cp. psittaci, when transported to the slaughterhouse posing a threat to human health. Thus, the presented ompA nested PCR-EIA will help to facilitate the diagnosis of Cp. psittaci infections in both poultry and man.
Competing interests
The author(s)declare that they have no competing interests
Authors' contributions
Marnix Van Loock and Kristel Verminnen have made substantial contributions to conception, design, acquisition of data, analysis and interpretation of data. Trudy Messmer made substantial contributions to acquisition of data, analysis and interpretation. Guido Volckaert, Bruno Goddeeris and Daisy Vanrompay have been involved in revising the manuscript critically for important intellectual content.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was supported by project S6037-section 2 grant from the Belgian Ministry of Public Health and Intervet International N.V. (Boxmeer, The Netherlands).
R. Pensaert, A. Doop, Ph. Deloddere and D. Gilliaert are acknowledged for their assistance in the slaughterhouse (Volys-Star, Lendelede, Belgium). N. Boon and W. Verstraete (Department of Biochemical and Microbial Technology, Ghent University) and E. de Graef and F. Haesebrouck (Department of Veterinary Bacteriology and Mycology, Ghent University) for the bacterial strains used for specificity testing.
==== Refs
Andersen AA Serotyping of Chlamydia psittaci isolates using serovar-specific monoclonal antibodies with the microimmunofluorescence test J Clin Microbiol 1991 29 707 711 1890172
Vanrompay D Ducatelle F Haesebrouck F Hendrickx W Primary pathogenicity of an European isolate of Chlamydia psittaci from turkey poults Vet Microbiol 1993 38 103 113 8128594 10.1016/0378-1135(93)90078-L
Geens T Van Loock M Kaleta E Magnino S Andersen AA Vanrompay D Deák J Serological and molecular characterization of Chlamydophila psittaci strains using serovar-specific monoclonal antibodies, OmpA RFLP analysis and OmpA sequencing Proceedings of the 5th meeting of the European society for Chlamydia research: Hungary 2004 Budapest: University of Szeged 302 1–4 September 2004
Andersen AA Vanrompay D Saif YM Chlamydiosis Diseases of Poultry 2003 11 Iowa: Iowa State University Press 863 879
Hall CF Glass SE Grimes JE Moore RW An epidemic of ornithosis in Texas turkeys in 1974 Southwest Vet 1975 28 19 21
Page LA Derieux WT Cutlip RC An epornitic of fatal chlamydiosis (ornithosis) in South Carolina turkeys J Am Vet Med Ass 1975 166 175 178 1112744
Anderson DC Stoesz PA Kaufmann AF Psittacosis outbreak in employees of a turkey-processing plant Am J Epidmiol 1978 107 140 148
Chalmers ZSK Oriel D, Schachter J, Taylor-Robinson D, Ward ME Duck ornithosis and related zoonosis in the UK Chlamydial infections 1986 Cambridge: Cambridge University Press 345 348
Newman JA Chlamydia spp. Infection in turkey flocks in Minnesota J Am Vet Med Assoc 1989 195 1528 1530 2599937
Grimes JE Wyrick PB Calnek BW, Barnes HJ, Beard CW, Reid M, Yoder HW Jr Chlamydiosis (ornithosis) Diseases of poultry 1991 9 Ames: Iowa State University Press 311 325
Vanrompay D Butaye P Sayada C Ducatelle R Haesebrouck F Characterization of avian Chlamydia psittaci strains using omp1 restriction mapping and serovar-specific monoclonal antibodies Res Microbiol 1997 148 327 333 9765811 10.1016/S0923-2508(97)81588-4
Hafez HM Sting R Jodas S Stadler A Hafez HM, Mazaheri A Chlamydia psittaci infections in meat turkey: investigations on the interaction with other avian infectious agents Proceedings of the 1st international symposium on turkey diseases 1998 Giessen, Germany: Verlag der DVG Service GmbH 208 217
Van Loock M Geens T De Smit L Nauwynck H Van Empel P Naylor C Hafez HM Goddeeris BM Vanrompay D Key role of Chlamydophila psittaci on Belgian turkey farms in association with other respiratory pathogens Vet Microbiol 2005 107 91 101 15795081 10.1016/j.vetmic.2005.01.009
Moroney JF Guevara R Iverson C Chen FM Skelton SK Messmer TO Plikaytis B Williams PO Blake P Butler JC Detection of chlamydiosis in a shipment of pet birds, leading to recognition of an autbreak of clinically mild psittacosis in humans Clin Infect Dis 1998 26 1425 1429 9636874
Soni R Seale JP Young IH Fulminant psittacosis requiring mechanical ventilation and demonstrating serological cross-reactivity between Legionella longbeachae and Chlamydia psittaci Respirology 1999 4 203 205 10382241 10.1046/j.1440-1843.1999.00176.x
Gosbell IB Ross AD Turner IB Chlamydia psittaci infection and reinfection in a veterinarian Aust Vet 1999 7 511 513
Everett KDE Chlamydia and Chlamydiales: more than meets the eye Vet Microbiol 2000 75 109 126 10889402 10.1016/S0378-1135(00)00213-3
Hinton DG Shipley A Galvin JW Harkin JT Brunton RA Chlamydiosis in workers at duck farm and processing plant Aust Vet J 1993 70 174 176 8343087
Hedberg K White KE Forfang JC Korlath JA Friendshuh KA Hedberg CW MacDonald KL Osterholm MT An outbreak of psittacosis in Minnesota turkey industry workers: implications for modes of transmission and control Am J Epidemiol 1989 130 569 577 2764001
Wilson PA Phipps J Samuel D Saunders NA Development of a simplified polymerase chain reaction-enzyme immunoassay for detection of Chlamydia pneumoniae J Appl Bacteriol 1996 80 431 438 8849645
Jantos CA Roggendorf R Wuppermann FN Hegemann JH Rapid detection of Chlamydia pneumoniae by PCR-enzyme immunoassay J Clin Microbiol 1998 36 1890 1894 9650931
Spencer WN Johnson FW Simple transport medium for the isolation of Chlamydia psittaci from clinical material Vet Rec 1983 113 535 536 6364542
Vanrompay D Cox E Mast J Goddeeris B Volckaert G High-level expression of Chlamydia psittaci major outer membrane protein in COS cells and in skeletal muscles of turkeys Infect Immun 1998 66 5494 5500 9784562
Vanrompay D Van Nerom A Ducatelle R Haesebroeck F Evaluation of five immunoassays for detection of Chlamydia psittaci in cloacal and conjunctival specimens from turkeys J Clin Microbiol 1994 32 1470 1474 8077391
Messmer TO Skelton SK Moroney JF Daugharty H Fields BS Application of a nested, multiplex PCR to psittacosis outbreaks J Clin Microbiol 1997 35 2043 2046 9230378
Arizmendi F Grimes JE Comparison of the Gimenez staining method and antigen detection ELISA with culture for detecting Chlamydiae in birds J Vet Diag Invest 1995 7 400 401
Andersen AA Comparison of pharyngeal, fecal and cloacal samples for the isolation of Chlamydia psittaci from experimentally infected cockatiels and turkeys J Vet Diagn Invest 1996 8 448 450 8953529
Tong CYW Donnelly C Harvey G Sillis M Multiplex polymerase chain reaction for the simultaneous detection of Mycoplasma pneumoniae and Chlamydia psittaci in respiratory samples J Clin Path 1999 52 257 263 10474515
Sheehy N Markey B Gleeson M Quinn PJ Differentiation of Chlamydia psittaci and Chlamydia pecorum strains by species-specific PCR J Clin Microbiol 1996 34 3175 3179 8940467
Hewinson RG Griffiths PC Bevan BJ Kirwan SES Field ME Woodward MJ Dawson M Detection of Chlamydia psittaci DNA in avian clinical samples by polymerase chain reaction Vet Microbiol 1997 54 155 166 9057259 10.1016/S0378-1135(96)01268-0
Harris JW Zoonotic human chlamydiosis of avian origin: a review with particular reference to epidemiology and control Wild Poultry Sci J 1983 39 5 23 10.1079/WPS19830001
Stanford CF Connolly JH Ellis WA Smyth ET Coyle PV Montgomery WI Simpson DI Zoonotic infections in Northern Ireland farmers Epidemiol Infect 1990 105 565 570 2123457
Andrews BE Major R Palmer SR Ornithosis in poultry workers Lancet 1981 1 632 634 6110862 10.1016/S0140-6736(81)91552-X
Andersen AA Van Deusen RA Production and partial characterization of monoclonal antibodies to four Chlamydia psittaci isolates Infect Immun 1988 56 2075 2079 3294184
Page LA Bankowski RA Factors affecting the production and detection of ornithosis antibodies in infected turkeys Am J Vet Res 1960 21 971 978 13732028
Illner VF Zur Frage der Uebertragung des Ornithosevirus durch das Ei Monatsh Veterinärmed 1960 17 116 117
Page LA Baranowski RA Investigation of a recent ornithosis epornitic in Californian turkeys Am J Vet Res 1959 20 941 945
Francis T JrMagill TP An unidentified virus proceeding acute meningitis and pneumonitis in experimental animals J Exp Med 1938 68 147 163 10.1084/jem.68.2.147
Andersen AA Two new serovars of Chlamydia psittaci from North American birds Diagn Invest 1997 9 159 164
|
16185353
|
PMC1261267
|
CC BY
|
2021-01-04 16:28:16
|
no
|
BMC Infect Dis. 2005 Sep 26; 5:76
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-76
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-801620214110.1186/1471-2334-5-80Case ReportPersistence of lipoatrophy after a four-year long interruption of antiretroviral therapy for HIV1 infection: case report Parruti Giustino [email protected] Giuseppe Marani [email protected] Unit of Infectious Diseases, Ospedale Civile "Spirito Santo", Via Fonte Romana 8, 65126 Pescara, Italy2005 3 10 2005 5 80 80 14 3 2005 3 10 2005 Copyright © 2005 Parruti and Toro; licensee BioMed Central Ltd.2005Parruti and Toro; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
HIV-infected patients on long-term highly active antiretroviral therapy often present peculiar patterns of fat redistribution, referred to as lipodystrophy. In spite of recent investigations, it is not known whether and to what extent the main features of lipodystrophy – that is lipoatrophy of peripheral fat at face, limbs and buttocks, as well as fat accumulation at breasts, abdomen and the dorso-cervical region – can be reversible once clinically manifest.
Case presentation
A 35 year old Caucasian HIV infected female developed severe diffuse lipodystrophy while on highly active antiretroviral therapy. A remarkable increase of breast size, fat accumulation at waist, and a fat pad on her lumbar spine were paralleled by progressive and disfiguring lipoatrophy of face, limbs and buttocks. The patient decided to interrupt her therapy after 20 months, with a stably suppressed viremia and a CD4 lymphocyte count >500/μL. She could carry on a safe treatment interruption for longer than 4 years. Most sites of fat accumulation switched to nearly normal appearance, whereas lipoatrophy was substantially unchanged at all affected sites.
Conclusion
our observation provides pictorial evidence that lipoatrophy may not be reversible even under ideal circumstances. Therefore, strategies to prevent lipoatrophy should be considered when defining therapeutic regimens for HIV infected patients, especially those at high risk.
==== Body
Background
HIV-infected patients on long-term highly active antiretroviral therapy (HAART) often present peculiar patterns of fat redistribution, involving both visceral and peripheral adipose tissue, referred to as lipodystrophy [1,2]. This complex condition was recognized shortly after the introduction of HAART; to date, however, it has been poorly elucidated in terms of pathogenesis [3,4]. Its main features are lipoatrophy of peripheral fat at face, limbs and buttocks, and/or fat accumulation at breasts, abdomen and the dorso-cervical region [3-5].
Complex methods for objective measurements of fat deposits on affected sites involve the use of Dual-energy X-ray absorptiometry, Magnetic Resonance Tomography or CT scans. These methods, however, are costly and not always accessible in routine clinical practice. Moreover, they still lack adequate standardization, so that the diagnosis of lipodystrophy more frequently relies on concordant patient's and physician's evaluation [3-5].
Cross sectional cohort studies indicate that lipodystrophy affects some 50% of patients continually treated with first-generation antiretroviral drugs for at least 18–24 months. Combined or severe abnormalities, however, develop only in 5 to 20% of affected patients [3-6], more frequently in older patients, females, patients with more advanced HIV disease and a longer exposure to antiretroviral drugs, especially stavudine and/or indinavir [6-8]. As to the evolution of body fat abnormalities, evidence has been gathered that once body changes become clinically evident, they generally tend to persist, improving only in a minority of cases [9].
Several strategies have been evaluated in clinical trials, with the aim of controlling such a stigmatizing side-effect. Switching to regimens including Nucleoside Reverse Transcriptase Inhibitors other than stavudine and/or a Non-Nucleoside Reverse Transcriptase Inhibitor instead of a Protease Inhibitor yielded objectively measured, significant increases in subcutaneous fat [10-12]; these, however, were not recorded as significant improvements by the affected patients, even after a long follow-up. At the same time, sites of fat accumulation were generally unmodified in parallel observations [10-12]. Some drugs are presently under evaluation for lipodystrophy, including metformine and rosiglitazone, with inconclusive results as yet [3,5]. Finally, it is not clear whether and to what extent treatment interruptions may be helpful once lipodystrophy is clinically manifest.
Here we report on the evolution of HAART-induced severe combined lipodystrophy in one patient who carried out a very long lasting spell of therapy interruption.
Case presentation
A 35 year old Caucasian female was found to be infected with HIV1 in 1989, when she tested after heterosexual exposure. She was put on zidovudine in 1995, when her CD4 lymphocyte count steadily declined. She took the drug intermittently until 1997, when she was still asymptomatic, with CD4 lymphocytes <100/μL and a high HIV1 viral load (>500 000 copies/mL). In July 1997, she started her first line HAART regimen including stavudine 30 mg twice a day (bid), lamivudine 150 mg bid and indinavir 800 mg 3 times a day. At that time her body weight was stable, her body shape was lean and preserved, with a body mass index (BMI) of 18.3. Adherence and response to treatment were optimal; she reached undetectable viremia (< 400 copies/mL) and a >200/μL CD4 lymphocyte gain by the 12th week of treatment. Since April 1998, the patient complained about remarkable breast size increase, fat accumulation at waist, and a fat pad grown over her lumbar spine. The pad was 7 × 10 × 3 cm in size by July, 1998. At that time indinavir was replaced with ritonavir/saquinavir 400/400 mg bid due to renal stones. Rapidly progressive, severe lipoatrophy at limbs, buttocks and face ensued in the following months, together with a further increase in size of her fat pads. HIV1 viremia remained constantly undetectable; metabolic parameters were normal throughout follow-up, with an isolated abnormal value of triglycerides (306 mg/dL) in August 1998. Adherence to treatment was full until April 2000, when the CD4 lymphocyte count was >500/μL. The patient was offered a simplified regimen, which she refused, preferring a sharp interruption of HAART. At that time, she did not allow us to take pictures. After interrupting HAART, HIV1 viremia rebounded to lower than baseline, remaining <100 000 copies/mL (mode, 25 000 copies/mL) thereafter. CD4 lymphocyte counts declined very slowly; the patient remained asymptomatic, regularly attending her follow-up visits. Her metabolic parameters, including triglycerides, remained within the normal range at all further checks. She decided to carry on without any treatment for over 4 years, until July 2004, when the CD4 lymphocyte count had dropped to 84/μL. Pads of fat accumulation at waist had reverted to nearly normal appearance; the lumbar pad was no longer appreciable; breasts were still enlarged, although to a lesser extent (Fig. 1a). At variance, lipoatrophy was still well evident; the patient admitted partial improvement at upper extremities, whereas she found no improvement at all for buttocks (Fig. 1b) and legs (Fig. 1c and 1d). Her BMI had dropped to 17.3. The patient was offered a new HAART regimen including nevirapine 200 mg bid, tenofovir diproxil 300 mg once a day and lamivudine 150 mg bid; adherence was full until her last visit, on December 23th, 2004, with a CD4 lymphocyte count of 250/μL and a viral load of 1250 copies/mL.
Figure 1 Patient's body habit details at the end of treatment interruption. 1a. Appearance of breasts. Insert. Appearance of patient's breasts before HAART. 1b. Persistent and marked lipoatrophy at buttocks. 1c. Lipoatrophy at legs, front view. 1d. Lipoatrophy of left calf, back view
Our observation provides pictorial evidence that severe combined lipodystrophy in our patient was only partially reversible, even under the ideal condition of a long lasting and safe interruption of HAART. Recent reports indicate that the loss of CD4 lymphocytes during treatment interruptions is frequently rapid in patients with a Nadir CD4 lymphocyte count <200/μL and a high rate of cell gain under treatment, irrespective of the level of CD4 lymphocyte count at the interruption of HAART [13,14]. In spite of that, our patient could safely carry her interruption on for over 4 years, remaining asymptomatic throughout the period, with >200 CD4 T-lymphocytes/μL until late in 2003. She did eat a light and balanced diet; she reported taking L-carnitine (5 grams/daily) since June, 2000, until June, 2001, whereas she denied taking any other drug or supplement possibly interfering with lipid metabolism thereafter. Under these extremely favourable circumstances, sites of lumbar and waist fat accumulation reverted to nearly normal appearance, whereas breasts showed partial improvement (Fig. 1a). Such a clear-cut reversal of fat accumulation has not been reported in controlled trials investigating the efficacy of therapy switches [10-12], probably because of persistent interference with lipid metabolism caused by the new antiretroviral drugs [3-5].
Severe lipoatrophy at buttocks and legs, instead, was unchanged both on patient's and on our judgment (Fig. 1b, 1c and 1d). Although it is plausible that instrumental measurements at the affected sites might have detected partial improvements, our observation suggests that the loss of peripheral adipose tissue caused by HAART [3-5,15] may not be fully reversible after treatment interruption, even in the long run.
Conclusion
Lipodystrophy represents a major drawback of antiretroviral therapy for HIV infection. Its clinical and psychological impact may ultimately jeopardize HAART efficacy in many cases [3,5]. At the time of therapy interruption, our patient declared she would prefer dying rather than living with such disfiguring changes of her body shape. Recent studies systematically addressed the impact of severe lipodystrophy on patients' quality of life, documenting that body habit changes challenge compliance with therapy, as well as social relationships, performance of daily activities, sexuality and self-esteem, especially in young women [16].
As a growing body of clinical evidence suggests that newer HAART regimens may be less toxic on the adipose tissue, causing a clear-cut lower rate of body shape changes in the long run [17], prevention of lipodystrophy should become a key issue when tailoring individual regimens, particularly for patients at high risk of developing fat abnormalities on the basis of the available data.
List of abbreviations used
μL = microliter
mL = milliliter
bid = bis in die
mg = milligrams
highly active antiretroviral therapy = HAART
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
GP was in charge of the patient since 1997 at the AIDS Unit in Pescara General Hospital and carried out most of her follow-up visits; report of this case was conceived during a conversation with GMT, who also favoured presentation of the case at the Glasgow meeting in 2004. In depth discussion of pertinent medical literature led to an agreed description of the case and to the draft of manuscript. Both authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We are sincerely indebted with Mrs Loredana Puglielli, chief nurse in our Unit, for long lasting helpful assistance with this patient.
Written consent was obtained from the patient for publication of the present report.
==== Refs
Carr A Samaras K Burton S Law M Freund J Chisholm DJ Cooper DA A syndrome of peripheral lipodystrophy, hyperlipidaemia and insulin resistance in patients receiving HIV protease inhibitors AIDS 1998 12 F51 F58 9619798 10.1097/00002030-199807000-00003
Carr A Samaras K Thorisdottir A Kaufmann GR Chisholm DJ Cooper DA Diagnosis, prediction, and natural course of HIV-1 protease-inhibitor-associated lipodystrophy, hyperlipidaemia, and diabetes mellitus: a cohort study Lancet 1999 353 2093 9 10382692 10.1016/S0140-6736(98)08468-2
Carr A HIV lipodystrophy: risk factors, pathogenesis, diagnosis and management AIDS 2003 S141 148 Review 12870540
Tershakovec AM Frank I Rader D HIV-related lipodystrophy and related factors Atherosclerosis 2004 174 1 10 15135244 10.1016/S0021-9150(03)00246-6
Koutkia P Grinspoon S HIV-Associated lipodystrophy: Pathogenesis, Prognosis, Treatment, and Controversies Annu Rev Med 2004 55 303 17 14746523 10.1146/annurev.med.55.091902.104412
Lichtenstein KA Ward DJ Moorman AC Delaney KM Young B Palella FJ JrRhodes PH Wood KC Holmberg SD Clinical assessment of HIV-associated lipodystrophy in an ambulatory population AIDS 2001 15 1389 1398 11504960 10.1097/00002030-200107270-00008
Saves M Raffi F Capeau J Rozenbaum W Ragnaud JM Perronne C Basdevant A Leport C Chene G Factors Related to Lipodystrophy and Metabolic Alterations in Patients with Human Immunodeficiency Virus Infection Receiving Highly Active Antiretroviral Therapy Clin Infect Dis 2002 34 1396 405 11981737 10.1086/339866
Joly V Flandre P Meiffredy V Leturque N Harel M Aboulker JP Yeni P Increased risk of lipoatrophy under stavudine in HIV-1-infected patients: results of a substudy from a comparative trial AIDS 2002 16 2447 2454 12461419 10.1097/00002030-200212060-00010
Galli M Ridolfo AL Adorni F Cappelletti A Morelli P Massetto B Piazza M Gianelli E Vaccarezza M Gervasoni C Moroni M Correlates of risk of adipose tissue alterations and their modification over time in HIV-1 infected women treated with antiretroviral therapy Antiviral Therapy 2003 8 347 354 14518704
Carr A Workman C Smith DE Hoy J Hudson J Doong N Martin A Amin J Freund J Law M Cooper DA Abacavir substitution for nucleoside analogs in patients with HIV lipoatrophy: a randomized trial JAMA 2002 288 207 215 12095385 10.1001/jama.288.2.207
Moyle GJ Baldwin C Langroudi B Mandalia S Gazzard BG A 48-Week, Randomized, Open-Label Comparison of Three Abacavir-Based Substitution Approaches in the Management of Dyslipidemia and Peripheral Lipoatrophy JAIDS 2003 33 22 28 12792351
Martin A Smith DE Carr A Ringland C Amin J Emery S Hoy J Workman C Doong N Freund J Cooper DA Reversibility of lipoatrophy in HIV-infected patients 2 years after switching from a thymidine analogue to abacavir: the MITOX Extension Study AIDS 2004 18 1029 1036 15096806 10.1097/00002030-200404300-00011
Mussini C Bugarini R Perno CF Antinori A Borghi V Bertoli A D'Arrigo R Bedini A Mongiardo N Cossarizza A Esposito R Kinetics of CD4 cells after discontinuation of antiretroviral therapy in patients with virological failure and a CD4 cell count greater than 500 cells /microl AIDS 2002 16 1551 1553 12131194 10.1097/00002030-200207260-00013
Boschi A Tinelli C Ortolani P Moscatelli G Morigi G Arlotti M CD4+ cell count guided treatment interruptions in chronic HIV-infected patients with good response to HAART AIDS 2004 18 2381 2389 15622314
Roche R Poizot-Martin I El Yazidi CM Compe E Gastaut JA Torresani J Planells R Effects of antiretroviral drug combinations on the differentiation of adipocytes AIDS 2002 16 13 20 11741158 10.1097/00002030-200201040-00003
Blanch J Rousaud A Martínez E De Lazzari E Milinkovic A Peri JM Blanco JL Jaen J Navarro V Massana G Gatell JM Factors Associated with Severe Impact of Lipodystrophy on the Quality of Life of Patients Infected with HIV-1 Clin Infect Dis 2004 38 1464 1470 15156486 10.1086/383573
Gallant JE Staszewski S Pozniak AL DeJesus E Suleiman JM Miller MD Coakley DF Lu B Toole JJ Cheng AK 903 Study Group. Efficacy and safety of tenofovir DF vs stavudine in combination therapy in antiretroviral-naive patients: a 3-year randomized trial JAMA 2004 292 191 201 15249568 10.1001/jama.292.2.191
|
16202141
|
PMC1261268
|
CC BY
|
2021-01-04 16:28:15
|
no
|
BMC Infect Dis. 2005 Oct 3; 5:80
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-80
|
oa_comm
|
==== Front
BMC PediatrBMC Pediatrics1471-2431BioMed Central London 1471-2431-5-381617657910.1186/1471-2431-5-38Research ArticleChildren in reviews: Methodological issues in child-relevant evidence syntheses Cramer Kristie [email protected] Natasha [email protected] Virginia [email protected] Lisa [email protected] Katrina [email protected] George [email protected] Terry P [email protected] Alberta Research Centre for Child Health Evidence, Department of Pediatrics, University of Alberta, Edmonton, Alberta, Canada2 Department of Medicine, Division of Nephrology, University of Alberta, Edmonton, Alberta, Canada3 Center for Population and Evidence Based Medicine, University of Texas-Houston Health Sciences Center, Houston, Texas, USA4 Department of Pediatrics and Child Health, The Children's Hospital at Westmead, Westmead, New South Wales, Australia5 School of Child and Adolescent Health, University of Cape Town, Cape Town, South Africa2005 21 9 2005 5 38 38 25 4 2005 21 9 2005 Copyright © 2005 Cramer et al; licensee BioMed Central Ltd.2005Cramer 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 delivery of optimal medical care to children is dependent on the availability of child relevant research. Our objectives were to: i) systematically review and describe how children are handled in reviews of drug interventions published in the Cochrane Database of Systematic Reviews (CDSR); and ii) determine when effect sizes for the same drug interventions differ between children and adults.
Methods
We systematically identified all of the reviews relevant to child health in the CDSR 2002, Issue 4. Reviews were included if they investigated the efficacy or effectiveness of a drug intervention for a condition that occurs in both children and adults. Information was extracted on review characteristics including study methods, results, and conclusions.
Results
From 1496 systematic reviews, 408 (27%) were identified as relevant to both adult and child health; 52% (213) of these included data from children. No significant differences were found in effect sizes between adults and children for any of the drug interventions or conditions investigated. However, all of the comparisons lacked the power to detect a clinically significant difference and wide confidence intervals suggest important differences cannot be excluded. A large amount of data was unavailable due to inadequate reporting at the trial and systematic review level.
Conclusion
Overall, the findings of this study indicate there is a paucity of child-relevant and specific evidence generated from evidence syntheses of drug interventions. The results indicate a need for a higher standard of reporting for participant populations in studies of drug interventions.
==== Body
Background
Health care decisions for individual patients are influenced by the availability of evidence that pertains most directly to the patient. However, health care providers are often faced with a paucity of evidence for specific patient groups such as children and youth, and thus must rely on evidence of questionable applicability as the basis for their health care decisions.
In the case of children, where there is a recognized gap in research evidence [1-4], health care providers frequently extrapolate evidence derived from adult studies to guide their decision-making [5]. To illustrate, several European surveys found 29–72% of the drugs prescribed to children are unlicensed or off-label [6-8]. In the United States 80% of the new drugs approved between 1984 and 1989 had no indication for use in children [9]. As a result, children may be less likely than adults to receive health care that is based on research in children.
This is unsettling given that it is widely recognized children and adults differ [10] in terms of their physiology and biology as well as the developmental and disease processes they experience [11,12]. Children and adults may also differ with respect to their response to therapies. For example, selective serotonin reuptake inhibitors have been found to be effective for treating depression in adults, whereas the evidence suggests these drugs are associated with an increased risk of suicidal behaviour in children [13]. Children treated with glucocorticosteroids for long periods of time have been found to be at risk of growth retardation, whereas this risk does not exist in adults [14]. Further, phenobarbital has a sedative effect in adults but may cause paradoxical hyperactivity in children [15]. These examples illustrate the potential risks of treating children without child-specific evidence. Nevertheless until pediatric research is conducted, health care providers are faced with a serious dilemma. If they deny children treatments known to be effective in adults, they may deny them effective treatments. Alternatively, if they treat children and/or youth with untested interventions they may be using treatments that differ in effect or are even harmful.
Health care providers increasingly turn to evidence syntheses to guide their clinical decisions. Systematic reviews represent the most rigorous and comprehensive synthesis of information on a specific clinical question. The Cochrane Collaboration is an international organization that promotes rigorous methodological standards in the preparation and maintenance of systematic reviews and supports the use of evidence by ensuring systematic reviews are available through the Cochrane Database of Systematic Reviews (CDSR). The CDSR presents a sample of high quality systematic reviews upon which to base methodological investigations [16]. We sought to systematically identify all child-relevant reviews published in the CDSR and describe whether child specific evidence was available. We then examined whether adults and children differed with respect to their response to different therapies.
Methods
Systematic review identification
One investigator reviewed the titles and abstracts of all 1496 complete systematic reviews published in the CDSR 2002, Issue 4 and identified relevant reviews in which: the efficacy or effectiveness of a drug intervention was investigated and the condition investigated occurs in both children and adults. Reviews of conditions judged by a pediatrician on our research team (TK) to be rare in children were excluded. If there was any uncertainty about the relevance of a review a pediatrician from our research team was consulted.
Data extraction
Characteristics of each systematic review were extracted using a standard data collection form. Information was extracted on review characteristics such as study inclusion criteria, number of included trials, results, and author's conclusions (complete list in Appendix A).
Descriptive analysis
Each systematic review was categorized as "adult", "child", or "mixed" based on the trials the review authors intended to include and the trials that were actually included. If the review authors did not specify the population, the intended review type was categorized as 'not specified'. If the ages of the trial participants could not be determined, the actual review type was categorized as 'uncertain'.
Quantitative data analysis: Effect size differences between children and adults
Numerical results were analyzed in Stata 7.0 and S-plus 6.0. The primary outcome for our analysis was the primary outcome investigated in the review. If the systematic review authors did not specify a primary outcome, the first outcome listed by the review authors was used. Effect sizes (e.g., relative risks and standardized mean differences) and their corresponding standard errors were calculated for each trial included in each review using the data as presented in the review. Adults were defined as 18 years and older and children were defined as 18 years and younger.
Results were combined separately for each review using a random effects model [17]. For dichotomous results, summary relative risks (RR) were calculated separately for children and adults. Then, ratios of RRs (i.e., child RR divided by adult RR) were estimated using meta-regression [18] to summarize the relationship between child and adult RRs (i.e., to investigate potential effect size differences between adults and children). The natural logarithm of the RR responses is regressed on a child indicator variable; the exponentiated estimated coefficient is the ratio of RR. For drug interventions intended to prevent an adverse outcome, a ratio of RRs less than 1 indicates that children experienced more benefit from the drug intervention than adults. For drug interventions intended to bring about a beneficial outcome, a ratio of RRs less than 1 indicates children experienced less benefit from the drug intervention than adults. A ratio of RR of 0.75 or 1.33 was considered a "small" but clinically relevant difference and 0.5 or 2 was considered to be a "moderate" difference. Ninety-five percent confidence intervals (95% CI) were calculated for each summary statistic.
For reviews reporting continuous variables, standardized mean differences (SMD) were calculated separately for adults and children. Differences of SMDs (child SMD minus adult SMD) were calculated using meta-regression to summarize the relationship between child and adult SMDs (i.e., to investigate potential effect size differences between adults and children). The SMD responses are regressed on a child indicator variable; the estimated coefficient is the difference in SMD. A difference in SMDs of ± 0.2 was considered a "small" but clinically relevant difference [19] and of ± 0.5 was considered to be a "moderate" difference. Ninety-five percent confidence intervals (95% CI) were calculated for each summary statistic.
Heterogeneity was quantified for the overall summary estimate using the I2 statistic [20,21], that indicates the percent variability due to between-study variability as opposed to within-study variability. An I2 value greater than 50% was considered moderately large.
Results
Description of child relevant reviews
Of the 1496 completed reviews in the CDSR, 403 were considered relevant to child health. Five of the relevant reviews investigated both treatment and preventive interventions for the same condition; each of these was counted as two independent reviews. Therefore, 408 reviews published by 34 different Cochrane Collaborative Review Groups (list available on request) were identified and evaluated.
Intended review types
In 16% (67/408) of reviews, the authors intended to only include adult trials; 11% (45/408) intended to only include child trials; 38% (153/408) intended to include both; and 35% (143/408) did not specify age criteria.
Thirty-six percent (24/67) of the adult review authors defined individuals 14 and older as adults, 43% (29/67) did not provide a definition for "adult", and the remaining 21% defined adults as 18 and older. Four percent (2/45) of the child review authors defined children as 19 years old and younger, 9% (4/45) did not provide a definition for "children", and the remaining 87 % (39/45) defined children as 18 years old and younger. The majority (89%) of the mixed review authors stated they planned to include children and adults but they did not define "children" or "adults". The remaining 11% (16/153) stated they would include individuals greater than one month.
Actual review types
Fifty-five percent (37/67) of the intended adult reviews actually included adult trials; 80% (36/45) of the intended child reviews actually only included child trials; and 63% (96/153) of the intended mixed reviews actually included both adult and child trials and/or mixed trials. The types of trial participants were could not be determined in 35% (50/143) of the not specified reviews (Table 1).
Table 1 Intended review type versus actual types of included trials
Intended Review Type
Adult Reviews
N = 67 (%) Child Reviews
N = 45 (%) Mixed Reviews
N = 153 (%) Not specified Reviews
N = 143 (%) Total
N = 408 (%)
Actual type of included trials Adult only 37 (55) 0 (0) 17 (11) 24 (17) 78 (19)
Child only 0 (0) 36 (80) 8 (5) 6 (4) 50 (12)
Mixed 9 (13) 7 (16) 96 (63) 51 (36) 163 (40)
Uncertain 18 (27) 2 (4) 24 (17) 50 (33) 94 (23)
No trials 3 (4) 0 (0) 8 (5) 12 (8) 23 (6)
Characteristics of the reviews
Table 2 outlines the characteristics of the reviews by intended review type. In 6% (23/408) of the reviews, no relevant studies were found. The remaining 94% (385/408) included a median of 8 (interquartile range (IQR) 4–15) trials. The number of participants included in each review could be determined in 91% (352/385) of the reviews; they included a median of 681 (IQR 210–1670) participants.
Table 2 Characteristics of the systematic reviews by intended review type
Intended Review Type
Variable Adult Reviews
N = 67 Child Reviews
N = 45 Mixed Reviews
N = 153 Not Specified Reviews
N = 143 Total
N = 408
Funding, N (%)
External 26 (39) 23 (51) 67 (44) 89 (62) 205 (50)
Internal 56 (84) 24 (53) 92 (60) 93 (65) 265 (65)
No funding 6 (9) 10 (22) 45 (29) 22 (15) 83 (20)
Languages of studies, N (%)
English only 1 (1) 0 0 3 (2) 4 (1)
English and non-English (no language restrictions) 34 (51) 16 (36) 53 (35) 54 (38) 157 (38)
Not specified 32 (48) 29 (64) 100 (65) 86 (22) 247 (61)
Included study designs, N (%)
RCT 47 (70) 30 (67) 91 (60) 82 (57) 250 (61)
CCT 20 (30) 15 (33) 62 (40) 60 (42) 157 (38)
Other 0 0 0 1 (0.7) 1 (0.2)
Median number of included trials, Median (Interquartile range) 9 (3–14) 10 (5–13) 8 (3–12) 7 (4–17) 8 (4–15)
Median number of participants, Median (Interquartile range) 813 (263–1836) 718 (313–1712) 551 (235–1474) 776 (305–2151) 681 (210–1670)
Number of reviews where the authors planned a subgroup analysis by age, N 2 10 30 10 52
Number of reviews where authors conducted a subgroup analysis by age, N 2 6 20 3 31
Number of reviews where authors found a difference in effect between adults and children, N 1 0 0 0 1
Thirteen percent (52/408) of the review authors planned to conduct a subgroup analysis based on age. Sixty percent (31/52) of these conducted the planned subgroup analysis and 3% (1/31) of these found a significant effect size difference by age. Only 11 of these 52 reviews specified the age groups they planned to compare: older adults to younger adults (1), children under 5 to adults and children over 5 (1), children and adults to older adults (1), children less than 12 to adolescents and adults older than 12 (1), children (0–12 years old) to adolescents (12–18) (2), and adults (18+) to children (<18)(5). None of these comparisons were significant. One unplanned subgroup analysis by age found a significant difference but the age used for the subgroup analysis was not reported.
Quantitative data analysis: Effect size differences between children and adults
Only 9% (37/408) of the reviews included enough data to allow for an investigation of effect size differences between adults and children. Figure 5 depicts the reasons that data from the majority of the reviews could not be used for this investigation. Thirty-seven percent of the reviews had no comparative data; the remaining 54% of reviews could not be used because of insufficiencies in the reporting of participant characteristics from the included trials.
Figure 5 Flow of data loss.
None of these reviews reported separate child and adult data collected from the same trial, which is the most likely to provide valid information as both adults and children would have been exposed to the same experimental procedures [22]. As a result, our analyses were exclusively based on between study comparisons where the adults and children may have been exposed to different experimental procedures [22]. Among the 37 reviews, a median of two studies (IQR 1 to 5) per review were omitted from the meta-regression because these studies either only presented collapsed adult and child data or they did not define the age ranges of the participants (Figure 5).
For the main comparison in each review, 24% (9/37) had useable data for only one child and one adult trial. Because at least 3 data points are required, data from these meta-analyses could not be analyzed using meta-regression and precision around these estimates is absent from the meta-metagraphs (Figures 1, 2, 3, 4).
Figure 1 Meta-metagraph comparing child and adult relative risk summaries. * I2 for the overall relative risk is greater than 50%. ** the overall relative risk was statistically significant. † an event in the original data indicated benefit; the data has been manipulated so that an event indicates harm.
Figure 2 Meta-metagraph comparing child and adult relative risk summaries, continued. * I2 for the overall relative risk is greater than 50%. ** the overall relative risk was statistically significant. † an event in the original data indicated benefit; the data has been manipulated so that an event indicates harm.
Figure 3 Meta-metagraph comparing child and adult relative risk summaries, continued. * I2 for the overall relative risk is greater than 50%. ** the overall relative risk was statistically significant. † an event in the original data indicated benefit; the data has been manipulated so that an event indicates harm.
Figure 4 Meta-metagraph comparing child and adult SMD summaries. * I2 for the overall relative risk is greater than 50%. ** the overall relative risk was statistically significant. † a greater value in the original data indicated less benefit; the data has been manipulated so that a greater value indicates more benefit.
The age definitions of adults and children were not consistent across reviews. A sensitively analysis was therefore conducted in which children were defined as 14 years and younger and adults were defined as 14 years and older. There were mostly negligible differences between the different age cutoff estimates.
When we compared the magnitude of effect between adults and children, only one (1/37) estimate was significant; this favored adults [23]. This estimate was significant for the 14-year cut off (14 studies, 947 patients) but not for the 18-year cutoff (8 studies, 1405 patients). In this review any artemisinin drug was compared to standard treatment (e.g., quinoline drugs) for severe malaria as measured by mortality.
Four of the 37 meta-analyses measured therapeutic benefit instead of prevention. To create a consistent definition for direction in the meta-metagraphs, RR for prevention instead of therapy was measured for these four meta-analyses. When the inverse was used one of these meta-analyses [24] showed a significant difference between adults and children for both age cutoffs; the results favored children [25]. This meta-analysis compared add-on lamotrigine to placebo in drug resistant partial epilepsy as measured by treatment success; we measured treatment failure. However, if the outcome in this meta-analysis is left as a therapeutic outcome (as presented in the review) the ratio of RRs for both age cutoffs is not significant. Four studies from this meta-analysis (538 patients) were included in the 18-year subgroup comparison and ten studies (797 patients) were included in the 14-year subgroup comparison. Ten studies (425 patients) were omitted from the 18-year subgroup comparison and nine studies (403 patients) were omitted from the 14-year subgroup comparison due to undefined age ranges.
Aside from the two meta-analyses discussed above, a significant difference between adult and child responses to therapy was not found in the majority (35/37) of the meta-analyses. However, in all but one of the comparisons [26] contained "small" effect sizes within the plausible values of the 95% confidence limits. Of the comparisons that used the 18-year cut off, 59% (17/29) had point estimates that were larger than the "small" effect size cutoff and 28% (8/29) had point estimates that were larger than the "moderate" effect size cutoff. Of the comparisons that used the 14-year cut off, 63% (12/19) had point estimates that were larger than the "small" effect size cutoff and 11% (2/19) had point estimates that were larger than the "moderate" effect size cutoff. In addition, of the meta analyses that included enough data points to be analyzed using meta-regression (N = 25), 84% (21/25) at the 18 year cut off and 79% (11/14) at the 14 year cut off had 95% confidence intervals that contained "moderate" effect sizes.
Discussion
This study provides support for the speculation there is too little child health evidence available from systematic reviews published in the CDSR. Although all 408 of the reviews evaluated here were on topics relevant to children, only about half intended (48%) or actually (52%) included data from children. Even when reviews included evidence generated from children, they frequently did not distinguish between children and adults in the analysis. Of the reviews that planned to include both children and adults (39%) only 20% planned a subgroup analysis by age. When data were available to investigate effect size differences between adults and children, the vast majority of comparisons lacked the power to detect "small" but clinically important differences.
Of particular interest was the finding that substantial variability occurred among the reviews with respect to how children, adolescents, and adults were defined. Some of the reviews defined adults as 14 and older whereas others defined them as 19 and older. The participants were often only described as "adults", "children", or "school-aged children". Furthermore, several reviews did not provide any information about the ages of the participants in the included studies. For example, 18% of the mixed reviews did not report the ages of the participants in any of the included studies and 55% only reported the ages of the participants in some of the included studies. Perhaps this is because ages were not reported at the trial level. We reviewed the included studies in a random sub-sample of the reviews (N = 40) and found when age characteristics were not reported in the review, a median of 72% of the trial authors also did not report age characteristics in the trial (Figure 5). Poor reporting decreases the amount of data available for a quantitative subgroup analysis in evidence reviews and limits the generalizability of results by health care professionals. The CONSORT statement for reporting of RCTs delineates appropriate reporting of baseline demographics [27]. Similarly, for meta-analyses QUOROM, MOOSE and the Cochrane handbook for systematic reviews of interventions provide guidelines for appropriate reporting of study characteristics including a description of the participant populations [28-30].
Our comparisons lacked the power to detect "small" and most "moderate" clinically significant differences. Several factors contribute to this lack of power: 1) too few studies; 2) trial authors did not report age characteristics of the participants, decreasing the number of useable studies; 3) systematic review authors did not report age data decreasing the number of useable studies; and 4) segregated adult and child data was not presented in the mixed trials. Until there is clear evidence whether there are or are not differences between adults and children, separate analyses need to be conducted, particularly where there is biological plausibility for a difference.
Our study was limited by the available data. Because the definitions for adult and child varied among the reviews, standard definitions for child (0–18) and adult (18+) could not be used consistently to categorize the reviews. Instead, the categories were based on the review author's definitions, resulting in less precise categories. In addition, since we did not review psychosocial and educational interventions the results of this study cannot be applied to these interventions. We also did not include reviews of rare conditions because including reviews of these conditions (frequently recognized as different in children and adults) had the potential to bias the results in favour of our hypothesis. Finally, because we only reviewed and summarized systematic reviews published in the CDSR, inferences generated from this study can only be applied to how children are treated in reviews published in the CDSR.
Since within-study evidence was not available for our comparisons, these comparisons are limited by potential between-study confounders [22]. Factors other than the one of interest may be responsible for any of the observed effect size differences, hence these inferences are to be viewed as preliminary, requiring confirmation [22] by trial level child-adult comparisons (i.e., within study comparisons). A potential solution is found in systematic reviews that use individual patient data (IPD). IPD adds more power by controlling for between study confounders and by enabling more sensitive modelling of the age-effect relationship [31]. Standard definitions of "child", "adult", and "adolescent" need to be developed and used in research studies, although caution is indicated to avoid misclassification bias from inappropriate age groupings. Very real biological and physiological differences may be masked by lumping individuals (e.g., adolescents and adults) into the same category.
In addition to including a checklist item for baseline demographics or study characteristics, guidelines developed to improve the quality of reporting studies such as CONSORT, QUOROM, MOOSE, and the Cochrane handbook for systematic reviews of interventions need to specifically suggest standards for reporting participant ages [27-30]. As well, to ensure adequate reporting, investigators need to be aware of and adhere to these accepted checklists and guidelines. Researchers planning to conduct studies that include both adults and children need to consider and investigate the differences between adults and children. We recommend that an a priori subgroup analysis by age be included, particularly when there is biological plausibility for differences.
Conclusion
Our analysis of systematic reviews did not exclude important differences in effect sizes between adults and children. Our research supports the need for better reporting in studies of drug interventions. When combining evidence, researchers need to be aware of the potential differences between the groups they are combining, especially when known physiological and biological differences exist. There is a need to define children, adults, and adolescents and to determine when it is appropriate and necessary to do subgroup analysis by age.
Competing interests
The author(s) declare they have no competing interests.
Authors' contributions
All authors contributed towards the conception and design of the study and the interpretation of the data. They also read, edited and approved the final manuscript. KC and TK identified relevant studies. KC, LH and KW participated in the data extraction. NW conceived of and conducted the statistical analysis. KC conducted the descriptive analysis. KC and NW drafted the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional File 1
Appendix A. This is appendix A for the manuscript that provides details of the data extracted from each systematic review included in the study.
Click here for file
Acknowledgements
This project was supported by a research contract provided by the Agency for Healthcare Research and Quality (AHRQ). We would like to thank Michelle Tubman and Christine Tyrrell for their assistance extracting data. As well we would like to thank Lisa Tjosvold for her assistance with the Cochrane Database of Systematic Reviews and data extraction.
==== Refs
Myers RP Poynard T Interferon for interferon nonresponding and relapsing patients with chronic hepatitis C Cochrane Database of Systematic Reviews 2002 CD003617
Panpanich R Garner P Antibiotics for treating scrub typhus Cochrane Database of Systematic Reviews 2002 CD002150
Soares-Weiser K Brezis M Leibovici L Antibiotics for spontaneous bacterial peritonitis in cirrhotics Cochrane Database of Systematic Reviews 2002 CD002232
Olliaro P Mussano P Amodiaquine for treating malaria Cochrane Database of Systematic Reviews 2002 CD000016
Gilbert RE See SE Jones LV Stanford MS Antibiotics versus control for toxoplasma retinochoroiditis Cochrane Database of Systematic Reviews 2002 CD002218
Sipe J Dunn L Aciclovir for Bell's palsy (idiopathic facial paralysis) Cochrane Database of Systematic Reviews 2002 CD001869
Lafuente-Lafuente C Melero-Bascones M Active chest compression-decompression for cardiopulmonary resuscitation Cochrane Database of Systematic Reviews 2002 CD002751
Casimiro L Brosseau L Milne S Robinson V Wells G Tugwell P Acupuncture and electroacupuncture for the treatment of RA Cochrane Database of Systematic Reviews 2002 CD003788
Wen ZH Gardener E Wang YP Nitrates for Achalasia Cochrane Database of Systematic Reviews 2002 CD002299
Johansen HK Gotzsche PC Amphotericin B lipid soluble formulations versus amphotericin B in cancer patients with neutropenia Cochrane Database of Systematic Reviews 2002 CD000969
Roos YBWEM Rinkel GJE Vermeulen M Algra A van Gijn J Antifibrinolytic therapy for aneurysmal subarachnoid haemorrhage Cochrane Database of Systematic Reviews 2002 CD001245
Elphick H Southern K Antifungal therapies for allergic bronchopulmonary aspergillosis in people with cystic fibrosis Cochrane Database of Systematic Reviews 2002 CD002204
Mwandumba HC Squire SB Fully intermittent dosing with drugs for treating tuberculosis in adults Cochrane Database of Systematic Reviews 2002 CD000970
Meremikwu M Logan K Garner P Antipyretic measures for treating fever in malaria Cochrane Database of Systematic Reviews 2002 CD002151
Hirst A Sloan R Benzodiazepines and related drugs for insomnia in palliative care Cochrane Database of Systematic Reviews 2002 CD003346
Busch A Schachter CL Peloso PM Bombardier C Exercise for treating fibromyalgia syndrome Cochrane Database of Systematic Reviews 2002 CD003786
Lee A Cooper MC Craig JC Knight JF Keneally JP Effects of nonsteroidal anti-inflammatory drugs on post-operative renal function in normal adults Cochrane Database of Systematic Reviews 2002 CD002765
Normand SLT Tutorial in biostatistics meta-analysis: formulating, evaluating, combining, and reporting Stat Med 18 321 59 10070677 10.1002/(SICI)1097-0258(19990215)18:3<321::AID-SIM28>3.3.CO;2-G
Cohen J Statistical Power Analysis for the Behavioural Sciences 1988 revised Hillsdale; New Jersey: Lawrence Erlbaum Associates
Higgins JPT Thompson Quantifying heterogeneity in a meta-analysis Stat Med 2002 21 1538 1558 10.1002/sim.1186
Higgins JPT Thompson SG Deeks JJ Altman DG Measuring inconsistency in meta-analyses BMJ 2003 327 557 560 12958120 10.1136/bmj.327.7414.557
Oxman AD Guyatt GH A consumer's guide to subgroup analyses Ann Intern Med 1992 116 78 84 1530753
McIntosh HM Olliaro P Artemisinin derivatives for treating severe malaria Cochrane Database of Systematic Reviews 2002 CD000527
Ramaratnam S Marson AG Baker GA Lamotrigine add-on for drug-resistant partial epilepsy Cochrane Database of Systematic Reviews 2002 CD001909
Deeks JJ Issues in the selection of a summary statistic for meta-anlysis of clinical trials with binary outcomes Stat Med 2002 21 1575 1600 12111921 10.1002/sim.1188
Marson AG Kadir ZA Hutton JL Chadwick DW Gabapentin add-on for drug-resistant partial epilepsy Cochrane Database of Systematic Reviews 2002 CD001415
Altman DG Schulz KF Moher D Egger M Davidoff F Elbourne D Gotzsche PC Lang T The Revised CONSORT Statement for Reporting Randomized Trials: Explanation and Elaboration Ann Intern Med 2001 134 663 694 11304107
Stroup DF Berlin JA Morton SC Olkin I Williamson GD Rennie D Moher D Becker BJ Sipe TA Thacker SB Meta-analysis of Obervational Studies in Epidemiology JAMA 2000 283 2008 2011 10789670 10.1001/jama.283.15.2008
Moher D Cook DJ Eastwood S Olkin I Rennie D Stroup DF Improving the quality of reports of meta-analysis of randomized controlled trials: the QUOROM statement Lancet 1999 354 1896 1900 10584742 10.1016/S0140-6736(99)04149-5
Green S Higgins J Cochrane Handbook for Systematic Reviews of Interventions 4.2.5, Section 7.5
Stewart LA Tierney JF To IPD or not to IPD? Eval Health Prof 2002 25 76 97 11868447 10.1177/0163278702025001006
|
16176579
|
PMC1261269
|
CC BY
|
2021-01-04 16:31:06
|
no
|
BMC Pediatr. 2005 Sep 21; 5:38
|
utf-8
|
BMC Pediatr
| 2,005 |
10.1186/1471-2431-5-38
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623197610.1371/journal.pmed.0020332Research in TranslationCancer BiologyGenetics/Genomics/Gene TherapyMolecular Biology/Structural BiologyOtherPharmacology/Drug DiscoveryOncologyOpthalmologyPediatricsOphthalmologyOncologyPediatricsGeneticsUse of Preclinical Models to Improve Treatment of Retinoblastoma Research in TranslationDyer Michael A *Rodriguez-Galindo Carlos Wilson Matthew W Michael A. Dyer is in the Department of Developmental Neurobiology, Carlos Rodriguez-Galindo is in the Department of Hematology Oncology, and Matthew W. Wilson is in the Department of Surgery, Division of Ophthalmology, St. Jude Children's Research Hospital, Memphis, Tennessee, United States of America. Matthew W. Wilson and Michael A. Dyer are also in the Department of Ophthalmology, University of Tennessee Health Science Center, Memphis, Tennessee, United States of America.
Competing Interests: The authors declare that no competing interests exist.
*To whom correspondence should be addressed. E-mail: [email protected] 2005 25 10 2005 2 10 e332Copyright: © 2005 Dyer 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.
Retinoblastoma: Teacher of Cancer Biology and Medicine
Dyer and colleagues examine the most promising preclinical models that recapitulate the molecular, genetic, and cellular features of retinoblastoma.
==== Body
Retinoblastoma is a rare childhood cancer of the retina. Approximately one in 20,000 children are affected worldwide; only 250 to 300 new cases are reported in the United States each year. Enucleation (see Glossary) is usually the treatment of choice for children with unilateral disease. Treatment of children with bilateral retinoblastoma is more challenging, as eye and vision preservation become priorities. Historically, bilateral retinoblastoma was treated with a combination of laser therapy, cryotherapy, and radiotherapy. Today, patients with bilateral retinoblastoma first receive upfront chemotherapy to reduce the tumor burden, and then undergo aggressive focal therapies. This approach has increased the rate of eye salvage and decreases or delays the use of radiation therapy.
Glossary
BAC-CGH: Bacterial artificial chromosome-comparative genome hybridization. Large overlapping regions of the human and mouse genomes have been incorporated into bacterial artificial chromosomes. These DNA fragments are then spotted onto glass slides in an array, and can be used to hybridize two genomic samples labeled with different fluorochromes. If a particular genomic region is amplified or deleted in one sample, it will be revealed as an increase or decrease in the intensity of the fluorescence of that sample on the array. This approach is useful for identifying genomic changes in cancer cells as they progress from preneoplastic lesions to metastatic cancer.
Chx10: Most if not all retinal progenitor cells express the Chx10 homeobox gene. Mutations in the Chx10 gene cause microphthalmia in both humans and mice, indicating that the Chx10 protein is an important regulator of retinal progenitor cell proliferation. In addition, Chx10 is expressed in mature bipolar neurons and Müller glia of the mouse and human retina.
Chx10-Cre;RbLox/Lox;p107−/− genetic background: Mice with conditional inactivation of Rb in retinal progenitor cells during retinal development. The p107 gene is deleted in the entire mouse, so the retinal cells are lacking both Rb and p107. These mice develop retinoblastoma by several months of age.
Cre transgenic lines: A variety of transgenic mouse lines have been generated that express Cre recombinase under the control of tissue-specific promoters. For example, the Chx10 promoter drives Cre expression in retinal progenitor cells, bipolar cells, and Müller glia.
Cre: A bacteriophage P1 protein that rapidly catalyzes site-specific recombination between two LoxP sites that are made up of a 34-bp DNA sequence.
Drug-efflux transporters: Some chemotherapeutic drugs are substrates for drug-efflux transporters that are expressed in the brain, and as a result, exhibit negligible uptake in the brain. If cancer cells express high levels of drug-efflux transporters, there may be an impact on how those cells respond to certain classes of chemotherapeutic agents.
Epigenetic programs: Genetic changes that result in alterations in gene expression are key features of many types of cancer. However, there are many examples of nongenetic changes in gene expression that contribute to cancer initiation or progression. These nongenetic or epigenetic changes can result from alterations in chromosomal structure or from condensation as a result of DNA methylation or histone organization.
Enucleation: Surgical removal of the eye from its orbit. Children with retinoblastoma who undergo enucleation often receive a prosthetic eye.
Exon-specific knock-in mouse: A genetically engineered mouse strain that encodes a single amino acid substitution in a single exon of the Rb gene. As a result of the exon-specific knock-in mutation, a normally spliced mRNA is produced that encodes a full-length Rb protein with a single amino acid substitution.
Rb
null alleles: An allele of Rb that is mutated and makes no functional protein. This is in contrast to some point-mutant alleles of Rb that retain partial function.
Rb,p107
conditional knock-out mice: The conditional allele of Rb called RbLox, contains two small DNA sequences called LoxP sites, which are recognition elements for the Cre recombinase. When Cre recombinase is introduced into the cell, it goes into the nucleus and deletes the DNA between the two LoxP sites, leading to a conditional knock-out tissue. There is not currently a p107Lox allele available so to inactivate both Rb and p107 in the developing retina, RbLox/−;p107−/− mice were crossed to a transgenic mouse line expressing Cre recombinase in the developing retina. When Rb and p107 are knocked out in the developing retina, retinoblastoma forms.
Rb−/−
mice:
Rb knock-out mice carry a disrupted copy of the mouse ortholog of the human RB1 tumor-suppressor gene. Rb−/− mice die in utero around day 13.5 of gestation due to hematopoietic and neural defects.
RbLox: A genetically engineered mouse line with two LoxP sites flanking one or more critical exons of the mouse Rb gene. When Cre recombinase is expressed in cells with a RbLox allele, the DNA between the LoxP sites is deleted, leading to a null allele of Rb.
RbLox/−;p53Lox/−;p107−/−
mice: These mice are similar to the RbLox/−;p107−/− mice described above, except that they also contain a conditional allele of the p53 tumor suppressor. Conditional inactivation of Rb, p53 and p107 in the developing retina leads to aggressive invasive bilateral retinoblastoma in 100% of mice.
SNP chips: Single nucleotide polymorphisms are associated with particular genomic regions. Oligonucleotide arrays, or chips, have been developed that allow researchers to compare two DNA samples, and determine if there are amplifications or deletions.
SV40
large T oncogene: A viral oncogene from the Simian Virus 40 genome. Large T protein binds and inactivates the Rb and p53 tumor-suppressor proteins, and leads to deregulated cell proliferation.
Two-photon live imaging: Standard laser confocal imaging is not suitable for many in vivo, or live, imaging applications because the light cannot penetrate deep enough into the tissue, and the energy of the light that is required may kill the cells being studied. Two-photon imaging combines the energy of two photons that are half the energy and twice the wavelength on a single cell. The overall result is that images can be obtained much deeper in the tissue because of the longer wavelength, and the cellular toxicity is reduced because of the lower energy of the light being used.
Unfortunately, recent advances in targeted chemotherapy have not benefited patients with retinoblastoma because our knowledge is limited regarding the signaling pathways affected following the inactivation of the retinoblastoma gene RB1. Also, retinoblastoma clinical trials take years to complete because so few patients are available. Finally, pharmaceutical companies have little financial incentive to develop therapies for childhood cancers because pediatric patients represent only 1% of all patients with cancer. Therefore, preclinical models that recapitulate molecular, genetic, and cellular features of retinoblastoma are essential for identifying the most promising new therapies.
For decades, St. Jude Children's Research Hospital has brought together leading researchers and clinicians to tackle some of the most debilitating childhood diseases. This article is based on our recent symposium, “Retinoblastoma: From Bench to Bedside,” at which ophthalmologists, pediatric oncologists, and cancer researchers from around the world gathered to review the current status of retinoblastoma management, and to discuss the most effective ways to use the preclinical retinoblastoma models developed last year [1–5].
Landmark Studies in Retinoblastoma Research
After studying retinoblastoma's pattern of inheritance susceptibility in families with a history of the disease, Alfred Knudson (Fox Chase Cancer Center) proposed a “two-hit” model to explain how a putative tumor suppressor could be inherited as a dominant trait that relied on inactivation of both alleles of the gene [6]. His colleagues studied the genetic lesions in children with heritable retinoblastoma and cloned the first human tumor-suppressor gene RB1 [7], supporting Knudson's two-hit model. These important discoveries generated great enthusiasm for the possibility of improved therapies for retinoblastoma, as exemplified in the statement by David Abramson (Memorial Sloan-Kettering Cancer Center) in 1986: “Within fifteen years, at the outside, we'll be able to stop retinoblastoma before it begins. I'm so sure of that that I've already given the drug a name. I call it retino-revert. We'll be able to diagnose a child prenatally and start giving retino-revert to the mother to prevent retinoblastoma from growing as the fetus is developing” [8]. However, we are no closer today to identifying an effective targeted chemotherapy for retinoblastoma than we were when this statement was made. This slow progress can be attributed, in part, to the lack of preclinical models that recapitulate the disease.
The first mouse model of retinoblastoma was generated in 1990 by ectopically expressing the SV40 large T oncogene in the developing retina [9]. The limitation of this model is that the SV40 T antigen may disrupt pathways that are not deregulated in human retinoblastoma. Thus, treatments that are effective in this model may have little clinical impact. In 1992, genetically engineered mouse strains carrying an inactivated Rb1 gene were generated by three groups [10–12]. Unlike children with germline RB1 mutations, Rb1 mice did not develop retinoblastoma. In 1998, chimeric mouse studies demonstrated that p107, a gene related to RB1, was involved in preventing retinal tumor formation in mice [13]. Although these studies provided important clues about the resistance of mice to retinoblastoma, chimeric mice cannot be used as a preclinical model because of considerable interanimal variability and low efficiency of chimeric mouse generation.
New Preclinical Models of Retinoblastoma
Childhood central nervous system (CNS) tumors arise during fetal or early childhood development. A major hurdle in the development of preclinical retinoblastoma models has been our limited knowledge about normal neural development. In other words, we have greater understanding of the molecular genetics of retinoblastoma than of the normal developmental neurobiology of the retina. Over the past ten years, Connie Cepko (Harvard University) and many others have advanced the field of retinal development, so that we now have a context in which to understand the cancer's biology [14,15].
Retinoblastoma research marked 2004 as an important year. Researchers combined developmental neurobiology with cancer biology to elucidate retinoblastoma formation, and those findings are now being moved directly into the clinic [1,5,16]. Advances were made in three key areas. First, the role of Rb protein in normal retinal development was elucidated. The Rb gene regulates two distinct processes in the developing mouse retina: rod photoreceptor development and progenitor cell proliferation (Schweers B, Donovan SL, Gray J, Zhang J, Martins R, et al., unpublished data) [5,17]. In Rb−/− mice, p107 partially compensates for Rb in the latter process but not the former, indicating that the two genes have both shared roles and unique roles (Schweers B, Donovan SL, Gray J, Zhang J, Martins R, et al., unpublished data) [5]. This work is an extension of Julien Sage's (Stanford University) work on p107 compensation after Rb inactivation in mouse embryonic fibroblasts [18].
The second advance was the generation of a mouse carrying targeted deletions of Rb and p107 in the developing retina [1,19]. By several weeks of age, the eyes of Rb,p107 conditional knock-out mice filled with immature, proliferating retinal cells, which gave rise to retinoblastoma [1]. Shortly after these studies were published, two other groups published similar findings using Cre transgenic lines with broader expression patterns in the retina (Pax6-Cre) [3] and in the developing CNS (Nestin-Cre) [2]. Unfortunately, neither Pax6-Cre nor Nestin-Cre mice can be used as preclinical models because of low penetrance, late onset, and nonautonomous effects (cell nonautonomous effects are phenotypic changes that occur in the tissue or cell being studied as a result of a change in the environment, rather than loss of the gene of interest in the tissue or cell itself). The importance of the p53 pathway in retinoblastoma was shown in mice with retinas lacking Rb, p107, and p53 [1]. These mice developed bilateral retinoblastoma with 100% penetrance and a short latency; thus, they are ideal for testing new chemotherapeutic drugs (Figure 1). Also, their mosaic pattern of Cre expression in Chx10-Cre mice is a more appropriate model of human retinoblastoma than Pax6-Cre or Nestin-Cre (discussed in [1,20]).
Figure 1 p53 Prevents Invasive, Aggressive Retinoblastoma in Mice
Rb was conditionally inactivated in retinal progenitor cells by using the Chx10-Cre transgenic mouse line and the RbLox allele. On a p107-deficient genetic background, these mice develop retinoblastoma with a penetrance of approximately 60%. However, the disease rarely progresses to invasive retinoblastoma with ocular hypertrophy, and the mice rarely become moribund. In contrast, mice lacking both copies of p53, Rb, and p107 (Chx10-Cre;RbLox/−;p53Lox/−;p107−/−) in their retinal progenitor cells develop aggressive, invasive retinoblastoma. More importantly, penetrance is 100% as bilateral retinoblastoma with a short latency (100.3 ± 42.3 days), which is ideal for testing new chemotherapeutic drugs. Interestingly, mice with one copy of wild-type Rb also develop invasive, aggressive retinoblastoma, but with a longer latency than do Chx10-Cre;RbLox/−;p53Lox/−;p107−/− mice. Preliminary studies indicated that these tumors have lost heterozygosity at the Rb locus; thus, Chx10-Cre;RbLox/−;p53Lox/−;p107−/− mice are the only mouse model of retinoblastoma that recapitulates this feature of human retinoblastoma.
The third advance in retinoblastoma research followed the generation and characterization of two other preclinical models. This work has not been published but was presented at the symposium “Retinoblastoma: From Bench to Bedside”; the Dyer laboratory and St. Jude have made these models directly available to the retinoblastoma research community. Studies on these animal models were combined with comprehensive cell culture, molecular biology, and pharmacokinetic studies to investigate drug efficacy for the treatment of retinoblastoma [16]. Analyses of five drugs and their combinations revealed that topotecan combined with carboplatin is a feasible alternative to the current triple-drug therapy (i.e., etoposide, vincristine, and carboplatin). These preclinical studies may provide useful information to complement the ongoing RET-5 clinical protocol at St. Jude (C.R. and M.W.W.) focused on the use of topotecan-combination chemotherapy.
Strengths and Limitations of Current Preclinical Models of Retinoblastoma
An orthotopic xenograft model has been developed for retinoblastoma, which relies on injecting human retinoblastoma cells into the eyes of newborn rats (Figure 2) [16]. At the symposium, there was great enthusiasm for the orthotopic xenograft model developed at St. Jude [16], and with new cell lines or primary tumors directly injected into the eyes of newborn rats, this xenograft model could be improved further (Figure 2). Enthusiasm for the clonal focal model involving the E1A 13S retrovirus in p53-deficient mice [16] was dampened by the use of an oncogene (Schweers B, Donovan SL, Gray J, Zhang J, Martins R, et al., unpublished data) [4,21] that could disrupt pathways that are not inactivated in human tumors. Briefly, a replication-incompetent retrovirus expressing the E1A 13S oncogene is injected into the subretinal space of p53-deficient newborn mice. Within a few weeks, these mice develop focal clonal retinoblastoma (Schweers B, Donovan SL, Gray J, Zhang J, Martins R, et al., unpublished data) [4,16,21]. The approach using Cre retrovirus injected in the eyes of newborn RbLox/−;p53Lox/−;p107−/− mice (discussed in [4]) was considered at the symposium to be superior to that using the E1A oncogene; however, the Cre-retrovirus model relies on p53 inactivation, which is not mutated in human tumors. One suggestion at the symposium was to identify the genetic lesion that disrupts the p53 pathway in primary human tumors, and then genetically engineer mice with the same mutation.
Figure 2 A Developmentally Appropriate Orthotopic Retinoblastoma Xenograft Model
(A) Injection of 1,000 cultured human retinoblastoma cells into the vitreal cavity of newborn rat eyes leads to retinoblastoma by two weeks of age. The rats do not require immunosuppression because they are immunonaive for the first 24 hours after birth. This is an ideal model for testing new drugs and for studying the retinoblastoma cancer stem cell properties.
(B) Human retinoblastoma cells have been genetically engineered to express the luciferase reporter gene. A two-week-old rat that received an intravitreal injection of Y79-LUC cells at birth is shown. Y79-LUC cells are retinoblastoma cells that express the firefly luciferase gene under the control of a constitutive promoter (see [16]). At two weeks, the rat received an intraperitoneal injection of the luciferase substrate (luciferin). The fluorescence is directly proportional to the tumor volume and the number of cells in the tumor. By using the Xenogen imaging system, we can follow individual tumors as they grow and respond to chemotherapy.
Two additional concerns about the current models were discussed. One is the use of Rb null alleles. Children with retinoblastoma often have a mutated RB1 allele and a null allele. David Goodrich (Roswell Park Cancer Institute) and colleagues have created an exon-specific knock-in mouse with a single amino acid change found in human retinoblastoma (Sun H, Yanjie C, Zhang X, Schweers B, Dyer MA, et al., unpublished data). The knock-in allele was crossed with the Chx10-Cre;RbLox/Lox;p107−/− and Chx10-Cre;RbLox/Lox;p107–/–;p53Lox/Lox genetic backgrounds. The single amino acid change at residue 654 is a low-penetrance allele of RB1 in humans; it will be interesting to determine if mice have a similar phenotype. The second concern is that models with two germline mutations do not recapitulate human retinoblastoma. In sporadic retinoblastoma, acute mutations are believed to occur at both RB1 loci; in heritable retinoblastoma, one allele has a germline disruption and the other undergoes acute inactivation. This concern is strengthened by Sage's finding that the timing of Rb1 inactivation (chronic versus acute) influences p107 compensation [18]. In these experiments, p107 compensation refers to the upregulation of p107 gene and protein expression when Rb expression is lost. By combining the knock-in point-mutant Rb allele described above with Cre-expressing retroviruses, we can more accurately recapitulate human retinoblastoma.
Another topic extensively discussed was the role of the p53 pathway in retinoblastoma. Clearly, p53 inactivation accelerates retinoblastoma in mice and leads to more aggressive, invasive tumors (see Figure 1) [1]. However, whether the p53 pathway is inactivated in human retinoblastoma is unknown. Several possible mechanisms were presented for the inactivation of the p53 pathway when the p53 gene is intact, as in retinoblastoma. It is particularly important that p53 is intact because it may provide targets for chemotherapy. Comprehensive analysis of this pathway in retinoblastoma should be a key focus of future research.
Using Preclinical Data in Support of Clinical Trials
One of the major obstacles to testing new chemotherapeutic combinations over the past 10–12 years has been the lack of a preclinical model that faithfully recapitulates the human disease. We are now able to merge our understanding and research of the intraocular physiology and drug kinetics with the unique biology of retinoblastoma (both in its initiation and in its temporal progression) in the same model. Researchers are now given the opportunity to answer some of the critical questions that the new era of conservative management has posed. The preclinical models have been successfully used to test new drugs and combinations [1,16]. However, more is still expected from these models, as many relevant questions have been left unanswered by clinical trials. More information regarding new agents, correct combinations and schedules, and route of administration (systemic or periocular) is needed.
It is clear that one of the major hurdles for future progress is suboptimal communication between the different clinical and translational researchers. This hurdle is compounded by the fact that there is still substantial disagreement among treating physicians about the objectives of the treatment and the means (treatment regimens) to achieve them. A simple question such as what constitutes the standard of care for a child with bilateral retinoblastoma is not easy to answer. For example, carboplatin as a single-drug regimen can provide ocular salvage that is as good as more aggressive regimens that also incorporate etoposide and vincristine, but the efficacy of both regimens depends on the stage of the intraocular disease and the appropriate use of aggressive focal treatments by an expert team.
It is not surprising then that single-drug carboplatin therapy is used by some institutions, while others use the two-drug carboplatin–vincristine combination, the more standard three drug combination that incorporates etoposide, or a modification of the latter that adds cyclosporine to inhibit drug-efflux transporters that might be involved in drug resistance. Because of the limited number of patients with retinoblastoma that are eligible for chemotherapy clinical trials, a proper comparison of those treatment regimens may not be possible.
While these questions are being proposed, new ideas are being brought into the equation. For example, the unique anatomy and physiology of the eye creates an environment where cancer remains well contained until advanced stages. In this context, increasing the intraocular concentration of antineoplastic agents through modulation of their intraocular pharmacokinetics, and developing new methods of direct intraocular delivery of drugs will become priorities. Even more challenging are efforts to design drugs that treat (or prevent the development of) retinoblastoma in utero, or to generate different mouse models that recapitulate the different human RB1 mutations. Gene therapy is also an interesting option, but this therapy has its own unique challenges, such as spread of the virus and secondary toxicity to the surrounding normal retina. How to incorporate these innovations into current regimens will become a major challenge in the future. Here again, the information gathered in the preclinical models will be pivotal to the successful translation of these innovations into the frontline treatments.
Input from clinicians and pharmacologists is essential for designing experiments that will be clinically relevant. Pharmocokinetics, side effects, dose, and schedule are essential considerations when performing preclinical studies for retinoblastoma. For this reason, a retinoblastoma working group that includes pharmacologists, pathologists, neurobiologists, ophthalmologists, and pediatric oncologists, such as the one at St. Jude is essential for effectively moving new treatments tested in the lab into clinical trials. Just as the 20th century saw improved outcomes and decreased morbidity for retinoblastoma, so may the 21st century, with targeted chemotherapy and new methods of drug delivery into the eye. In this new era of translational research, experts from several different fields must work together in the delicate process of identifying and resolving the challenges posed by this unique malignancy. Just as retinoblastoma has been at the forefront of many of the advances in cancer biology over the past several decades, retinoblastoma researchers are in a unique position to serve as the example for translational research (see [22]).
Retinoblastoma Provides a Unique Window into Tumor Growth and Death
Advances in retinoblastoma treatment are, in part, due to the unique environment of the eye––a transparent system that permits direct visualization of the disease (Figure 3). A clear cornea, lens, and vitreous are inherent to successful management, e.g., transparency is imperative for precise delivery of focal therapies. Imaging modalities (i.e., ultrasonography, computed tomography, and magnetic resonance imaging) can be used to confirm the diagnosis but not to assess the response to treatment. Only with indirect ophthalmoscopy and scleral depression can the entire retina be examined. Also, tumor recurrences smaller than 1 mm and early vitreous seeding are easily appreciated by the trained observer; this resolution is not attainable by other imaging techniques.
Figure 3 The Eye Is a Transparent System That Permits Direct Visualization of Retinoblastoma
(A) Untreated retinoblastoma involving the macula of the left eye. The translucent cornea and lens allow detailed visualization. The tumor is an amelanotic, partially calcified mass that has broken through the overlying retina. Dilated arterioles infiltrate the tumor.
(B) The same tumor after being treated with chemotherapy. The mass has completely calcified, and the caliber of the overlying retinal vessels has diminished. A focal area of atrophied retinal pigmented epithelium surrounds the lesion.
Future treatment regimens will continue to rely on the unique properties of the eye; however, they will exploit the eye's permeability rather than transparency. A substantial body of work has already begun to document the efficacy of periocular chemotherapy [23,24]. The volume, concentration, and toxicity of agents are limiting current endeavors; however, investigators are exploring both new drugs and new vehicles of delivery to achieve efficacious ocular concentrations.
Future Directions: Targeted Chemotherapy
To more efficiently target chemotherapy to retinoblastoma and move away from broad-spectrum agents, we need to answer three fundamental questions: what is the disease's cell-of-origin, how does it expand, and what genetic events occur after RB1 inactivation?
Typically, the cancer cell of origin is inferred from the expression of cell type–specific markers or other morphologic features. In CNS cancers, marker expression is difficult to interpret because the cell types are complex. Neurobiologists define normal cell populations by marker expression, dendritic and axonal morphology, and location within the CNS. These features are often lost or disrupted in CNS tumor cells; therefore, it is difficult to determine their cell of origin. The use of Cre-expressing retroviruses and two-photon live imaging will allow us to follow individual tumor clones as they expand beyond single-infected retinal progenitor cells, and help to identify the retinoblastoma cell of origin in mouse models (discussed in [4]). Identifying the retinoblastoma cell of origin is essential because different cells may rely on entirely different pathways to drive proliferation and tumor expansion. Verification that the cell of origin in mouse retinoblastoma is the same as that in human retinoblastoma is essential for successful use of the model in screening new drugs targeted to deregulated pathways in that cell.
Second, we need to determine if retinoblastoma expands by a cancer stem cell mechanism [25,26] or by expansion of all cells in the tumor. A tumor can expand from a small population of cancer stem cells, even if it did not arise from a normal stem cell. This is of particular importance in retinoblastoma because no stem cells are present where tumors arise in the neural retina [27,28]. The xenograft model (see Figure 2) is ideally suited for determining if retinoblastoma expands by a stem cell mechanism. Human and mouse tumors could be fractionated, serially transplanted into the developing eye, and followed to determine which cells reconstitute the entire tumor. Then, by efficiently targeting those cells, we could halt tumor progression without dramatically altering tumor volume. If all cells expand, then targeted chemotherapy will need to focus on halting the expansion of all cells in the tumor.
Third, we must identify the secondary genetic events that follow RB1 inactivation. Processes such as escape from apoptosis (programmed cell death), deregulated growth, and telomere maintenance (capping of the ends of chromosomes) occur in most (if not all) cancers [29], but not all retinoblastomas achieve these changes through common genetic mechanisms. Indeed, the retinoblastoma cell of origin may be a heterogeneous population with diverse epigenetic programs, which may hinder the identification of common pathways for targeting chemotherapy. Molecular genetic analysis of primary human retinoblastomas and mouse retinoblastomas by gene expression microarrays, BAC-CGH, and single-nucleotide-polymorphism (SNP) chips provides a good starting point. Information on deregulated pathways in retinoblastoma can then be exploited to target chemotherapy and, ultimately, to halt tumor progression without the side effects associated with broad-spectrum chemotherapy.
Conclusion
Developing new treatments for a complex disease such as retinoblastoma requires expertise in pediatric oncology, ophthalmology, pharmacology, and developmental neurobiology. The open forum at the “Retinoblastoma: From Bench to Bedside” symposium provided a unique opportunity for experts in each of these areas to discuss the best way to take advantage of recently developed preclinical models of retinoblastoma in order to achieve the common goal of saving vision in children with this devastating cancer.
While there is disagreement about the best treatment, there was a broad consensus about the importance of preclinical models to test new therapies prior to clinical trials. By combining preclinical models with analysis of primary human tumors, we hope to move away from broad-spectrum chemotherapy and its associated side effects, and more effectively target the pathways that are deregulated in retinoblastoma.
For further information, please see http://www.stjude.org/retinoblastoma, http://www.stjude.org/dyer, and http://www.cure4kids.net.
Citation: Dyer MA, Rodriguez-Galindo C, Wilson MW (2005) Use of preclinical models to improve treatment of retinoblastoma. PLoS Med 2(10): e332.
Abbreviations
CNScentral nervous system
St. JudeSt. Jude Children's Research Hospital
==== Refs
References
Zhang J Schweers B Dyer MA The first knockout mouse model of retinoblastoma Cell Cycle 2004 3 952 959 15190215
MacPherson D Sage J Kim T Ho D McLaughlin ME Cell type-specific effects of Rb deletion in the murine retina Genes Dev 2004 18 1681 1694 15231717
Chen D Levine-bar I Vanderluit JL Slack RS Agochiya M Cell-specific effects of RB or RB/p107 loss on retinal development implicate an intrinsically death-resistant cell-of-origin in retinoblastoma Cancer Cell 2004 5 539 551 15193257
Dyer MA Bremner R The search for the retinoblastoma cell of origin Nat Rev Cancer 2005 5 91 101 15685194
Zhang J Gray J Wu L Leone G Rowan S Rb regulates proliferation and rod photoreceptor development in the mouse retina Nat Genet 2004 36 351 360 14991054
Knudson A Mutation and cancer: Statistical study of retinoblastoma Proc Natl Acad Sci U S A 1971 68 820 823 5279523
Friend SH Bernards R Rogelj S Weinberg RA Rapaport JM A human DNA segment with properties of the gene that predisposes to retinoblastoma and osteosarcoma Nature 1986 323 643 646 2877398
Angier N Natural obsessions: The search for the oncogene 1988 Boston Houghton Mifflin 394
O'Brien JM Marcus DM Bernards R Carpenter JL Windle JJ A transgenic mouse model for trilateral retinoblastoma Arch Ophthalmol 1990 108 1145 1151 1696469
Jacks T Fazeli A Schmitt EM Bronson RT Goodell MA Effects of an Rb mutation in the mouse Nature 1992 359 295 300 1406933
Lees E Faha B Dulic V Reed SI Harlow E Cyclin E/cdk2 and cyclin A/cdk2 kinases associate with p107 and E2F in a temporally distinct manner Genes Dev 1992 6 1874 1885 1398067
Clarke AR Maandag ER van Roon M van der Lugt NM van der Valk M Requirement for a functional Rb-1 gene in murine development Nature 1992 359 328 330 1406937
Robanus-Maandag E Dekker M van der Valk M Carrozza ML Jeanny JC p107 is a suppressor of retinoblastoma development in pRb-deficient mice Genes Dev 1998 12 1599 609 9620848
Livesey FJ Cepko CL Vertebrate neural cell-fate determination: Lessons from the retina Nat Rev Neurosci 2001 2 109 118 11252990
Dyer MA Cepko CL Regulating proliferation during retinal development Nat Rev Neurosci 2001 2 333 342 11331917
Laurie NA Gray JK Zhang J Leggas M Relling M Topotecan combination chemotherapy in two new rodent models of retinoblastoma Clin Cancer Res 2005 In press
Donovan SL Dyer MA Developmental defects in Rb-deficient retinae Vision Res 2004 44 3323 3333 15536000
Sage J Miller AL Perez-Mancera PA Wysocki JM Jacks T Acute mutation of retinoblastoma gene function is sufficient for cell cycle re-entry Nature 2003 424 223 228 12853964
Rowan S Cepko CL Genetic analysis of the homeodomain transcription factor Chx10 in the retina using a novel multifunctional BAC transgenic mouse reporter Dev Biol 2004 In press
Schweers BA Dyer MA New genetic tools to study retinal development in vivo Vis Neurosci 2005 In press
Dyer MA Mouse models of childhood cancer of the nervous system J Clin Pathol 2004 57 561 576 15166259
Knudson A Retinoblastoma: Teacher of cancer biology and medicine PLoS Med 2005 2 e349 10.1371/journal.pmed.0020349 16231982
Hayden BH Murray TG Scott IU Cicciarelli N Hernandez E Subconjunctival carboplatin in retinoblastoma: Impact of tumor burden and dose schedule Arch Ophthalmol 2000 118 1549 1554 11074812
Van Quill KR Dioguardi PK Tong CT Gilbert JA Aaberg TM Subconjunctival carboplatin in fibrin sealant in the treatment of transgenic murine retinoblastoma Ophthalmology 2005 112 1151 1158 15885791
Lapidot T Sirard C Vormoor J Murdoch B Hoang T A cell initiating human acute myeloid leukaemia after transplantation into SCID mice Nature 1994 367 645 648 7509044
Pardal R Clarke MF Morrison SJ Applying the principles of stem-cell biology to cancer Nat Rev Cancer 2003 3 895 902 14737120
Tropepe V Coles BL Chiasson BJ Horsford DJ Elia AJ Retinal stem cells in the adult mammalian eye Science 2000 287 2032 2036 10720333
Coles BL Angenieux B Inoue T Del Rio-Tsonis K Spence JR Facile isolation and the characterization of human retinal stem cells Proc Natl Acad Sci U S A 2004 101 15772 15777 15505221
Vogelstein B Kinzler KW Cancer genes and the pathways they control Nat Med 2004 10 789 799 15286780
|
16231976
|
PMC1261510
|
CC BY
|
2021-01-05 11:13:37
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e332
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020332
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623197710.1371/journal.pmed.0020339PerspectivesBioengineeringAllergy/ImmunologyDermatologyOncologyGeneral MedicineVaccinesBiotechnologyCancer BiologyCell BiologyImmunologyInfectious DiseasesMolecular Biology/Structural BiologySystems BiologyVirologyImmunology and AllergyOncologyDermatologyCancer Vaccines: The Next Generation of Tools to Monitor the Anticancer Immune Response PerspectiveNestle Frank O *Tonel Giulia Farkas Arpad Frank O. Nestle, Giulia Tonel, and Arpad Farkas are in the Department of Dermatology, University Hospital Zurich, Zurich, Switzerland.
Competing Interests: The authors declare that no competing interests exist.
*To whom correspondence should be addressed. E-mail: [email protected] 2005 25 10 2005 2 10 e339Copyright: © 2005 Nestle 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.
Marked Differences in Human Melanoma Antigen-Specific T Cell Responsiveness after Vaccination Using a Functional Microarray
Test Your Knowledge: Ten Questions about Melanoma
Nestle and colleagues discuss a new immunomonitoring tool, described by Chen et al. in PLoS Medicine, that may give insight into the vaccine-induced anticancer immune response.
==== Body
The promise of a cancer vaccination to fight tumors using the patient's own immune system is at the same time fascinating and challenging. Since cancer typically strikes at an older age, there is no evolutionary pressure on the immune system to develop successful anticancer effectors. This lack of evolutionary pressure to fight cancer is an apparent contrast to infectious diseases, for which a reasonable argument can be made that the immune system has evolved to fight pathogens. These assumptions might explain the relative success of vaccine development in the area of infectious diseases, and the long history of failed attempts in cancer vaccination.
And yet there is overwhelming evidence from laboratory studies that specific effectors of the immune system are able to recognize and destroy cancer cells. Soluble effectors of the immune system, such as antibodies, are now reliable weapons in the fight against lymphomas [1]. The graft-versus-leukemia effect (immunological rejection of leukemia cells following bone marrow transplantation) strongly indicates the power of the immune system to control certain tumors [2]. Thus, the immune system, even though primarily evolved to fight pathogens, clearly has the means to fight cancer.
Challenges of Cancer Vaccination
What are the prerequisites for a successful anticancer immune response? Before destruction of a tumor is achieved by the host immune system, a simplified four-step scenario has to take place: (1) activation and expansion of tumor-specific effector cells, (2) migration of effector cells to the tumor site, (3) effector-cell recognition of tumor cells, and (4) effector-cell destruction of tumor cells. In addition, a quantitative balance between the number of tumor cells and effector cells needed to destroy the tumor has to be taken into account. All these steps need to be investigated and controlled for successful cancer vaccination.
Cytotoxic T cells are among the best-investigated effector cells of the immune system. There is strong evidence that the presence of intratumoral T cells correlates with improved clinical outcome in certain human cancers during the natural immune response against tumors [3]. Dendritic cells are among the most powerful activators of tumor-specific helper cells and cytotoxic T cells [4]. Expansion of human tumor antigen-specific helper cells and cytotoxic T cells in peripheral blood has been demonstrated using antigen-pulsed dendritic cells as well as intracutaneous peptide immunization in the presence or absence of adjuvant [5–7].
Specific effectors of the immune system are able to recognize and destroy cancer cells.
There is, however, limited insight into the magnitude, breadth, and molecular nature of the induced immune responses. There are also currently no means to discriminate between protective and nonprotective (curative and noncurative) T cell responses. In fact, we are not certain about the frequency of tumor antigen-specific T cells that is necessary for tumor destruction. While viral protection models suggest that a high frequency (number) of vaccine-induced specific effectors is necessary, alternative hypotheses favor the generation of low-frequency vaccine-induced responses, which might in turn affect pre-existing antitumor-specific T cells [8]. Therefore, new developments in the area of monitoring and understanding the tumor-specific immune response, in combination with small innovative pilot vaccine trials, are needed.
Monitoring of the Cancer Immune Response
There are many ways to assess a cancer-specific immune response, including monitoring (1) direct cytotoxicity of effectors, as measured by chromium release assays (see Glossary), (2) cytokine release from effector cells, as assessed by flow cytometry or enzyme-linked immunosorbent assay techniques, (3) T cell receptor (TCR) specificities, as assessed by MHC-peptide multimers, (4) clonal composition of the T cell response via CDR (complementarity-determining region) 3 spectratyping, and (5) T cell degranulation via cell surface exposure of cluster designation (CD)107 [9–11].
Glossary
Cluster designation (CD) 107: A lysomal membrane protein that translocates to the cell surface during the killing process.
CDR3 spectratyping: CDRs of TCRs are the parts of these molecules that determine their specificity and make contact with specific ligands. Spectratyping defines certain types of DNA gene segments that constitute the CDR.
Chromium release assay: Assay for cytotoxic activity of killer cells.
Enzyme-linked immunosorbent assay (ELISA): A serological assay in which a bound antigen or antibody is detected by a linked enzyme that converts a colorless substrate into a colored product.
Flow cytometry: Analysis of biological material by detection of light-absorbing or fluorescing properties of cells.
HLA-A2-Immunoglobulin dimers: Dimers of human leukocyte antigen domains fused to an immunoglobulin scaffold.
MHC-peptide multimers: MHC-peptide multimers detect vaccine-specific T cells.
Multiparametric microarray platform: Platform to assess multiple parameters at once on a small glass chip.
NK cells: NK cells are large, granular non-T, non-B lymphocytes that kill certain tumor cells.
NKT cells: NKT cells are lymphocytes that share features of both T cells and NK cells.
TCR specificities: Specific TCRs recognizing MHC-peptide complexes.
Many of these techniques are useful, but fail to fully assess the functional complexity of an anticancer T cell response in a comprehensive manner. Analyzing T cell specificities using MHC-peptide multimers has revolutionized the field of cancer vaccination, but does not provide insights into T cell function. Detection of interferon-γ production by tumor-specific T cells gives some functional insights, but lacks information about TCR specificities and does not cover the broad spectrum of potential effector cytokines. Indeed, studies of chronic viral infection models suggest the relevance of multiple cytokines such as interferon γ, interleukin 2, and tumor necrosis factor α for effector T cell fitness [12].
A New Study of a Novel Immunomonitoring Tool
A new study by Chen et al. in the current issue of PLoS Medicine provides a novel approach to these scientific issues, and increases our insight into the vaccine-induced anticancer immune response using a novel immunomonitoring tool [13]. The tool combines analysis of TCR specificities with detection of cytokine production in a multiparametric microarray platform. The authors array HLA-A2-immunoglobulin dimers loaded with the peptide of interest with cytokine-capturing antibodies on three-dimensional substrates composed of microscope slides coated with a polyacrylamide gel. This allows the comprehensive analysis of multiple T cell specificities and functional outcomes. Similar platforms were recently used to investigate antigen-specific T cell clones [14]. The current study, however, provides the first comprehensive analysis of vaccine-induced antitumor immune responses in patients with cancer.
One of the most striking findings is the marked variation of responses toward well-defined peptide vaccines. Variation of the response was seen both at the level of a single patient and, independently, at the level of the specific antigen. Thus, no patient or antigen-specific functional response pattern was observed. Even though the current study does not allow definitive conclusions about a link between specific cytokine secretion profiles and clinical outcomes, it appears as if both interferon γ and tumor necrosis factor α were relevant for tumor clearance, as indicated by prolonged recurrence-free periods in patients with such a cytokine profile.
Implications of the Study
The study raises several questions. Are cytokine signatures present in certain subpopulations of effector T cells, especially those successful in tumor rejection? Are cytokine signatures predictive of the clinical outcome? It will be interesting to test T cell subpopulations, especially those derived from secondary lymphoid tissues and the tumor site. Are there cytokine signatures in response to pathogens or pathogen-specific vaccines, and how do these signatures differ from those induced by cancer vaccines? Antibody responses might be detected by protein microarrays [15]. Are there ways to array and functionally analyze other components of the immune system such as NKT cells, NK cells, or granulocytes?
Taken together, the current study establishes the fundamentals for future application of high-throughput multiparametric platforms that simultaneously capture antigen-specific T cells and detect secreted products in the analysis of tumor and, potentially, pathogen-specific immune responses.
Citation: Nestle FO, Tonel G, Farkas A (2005) Cancer vaccines: The next generation of tools to monitor the anticancer immune response. PLoS Med 2(10): e339.
Abbreviations
CDRcomplementarity-determining region
MHCmajor histocompatibility complex
NKnatural killer
TCRT cell receptor
==== Refs
References
Rastetter W Molina A White CA Rituximab: Expanding role in therapy for lymphomas and autoimmune diseases Annu Rev Med 2004 55 477 503 14746532
Bleakley M Riddell SR Molecules and mechanisms of the graft-versus-leukaemia effect Nat Rev Cancer 2004 4 371 380 15122208
Zhang L Conejo-Garcia JR Katsaros D Gimotty PA Massobrio M Intratumoral T cells, recurrence, and survival in epithelial ovarian cancer N Engl J Med 2003 348 203 213 12529460
Nestle FO Farkas A Conrad C Dendritic-cell-based therapeutic vaccination against cancer Curr Opin Immunol 2005 17 163 169 15766676
Cerundolo V Hermans IF Salio M Dendritic cells: A journey from laboratory to clinic Nat Immunol 2004 5 7 10 14699398
Schuler-Thurner B Schultz ES Berger TG Weinlich G Ebner S Rapid induction of tumor-specific type 1 T helper cells in metastatic melanoma patients by vaccination with mature, cryopreserved, peptide-loaded monocyte-derived dendritic cells J Exp Med 2002 195 1279 1288 12021308
Gilliet M Kleinhans M Lantelme E Schadendorf D Burg G Intranodal injection of semimature monocyte-derived dendritic cells induces T helper type 1 responses to protein neoantigen Blood 2003 102 36 42 12560234
Germeau C Ma W Schiavetti F Lurquin C Henry E High frequency of antitumor T cells in the blood of melanoma patients before and after vaccination with tumor antigens J Exp Med 2005 201 241 248 15657293
Speiser DE Immunological techniques: Ex vivo characterization of T cell-mediated immune responses in cancer Curr Opin Immunol 2005 17 419 422 15955683
Chen DS Davis MM Cellular immunotherapy: Antigen recognition is just the beginning Springer Semin Immunopathol 2005 27 119 127 15834723
Hernandez-Fuentes MP Warrens AN Lechler RI Immunologic monitoring Immunol Rev 2003 196 247 264 14617209
Wherry EJ Ahmed R Memory CD8 T-cell differentiation during viral infection J Virol 2004 78 5535 5545 15140950
Chen DS Soen Y Stuge TB Lee PP Weber JS Marked differences in human melanoma antigen-specific T cell responsiveness after peptide vaccination using a functional microarray PLoS Med 2005 2 e265 10.1371/journal.pmed.0020265 16162034
Stone JD Demkowicz WE Stern LJ HLA-restricted epitope identification and detection of functional T cell responses by using MHC-peptide and costimulatory microarrays Proc Natl Acad Sci U S A 2005 102 3744 3749 15728728
Bacarese-Hamilton T Gray J Crisanti A Protein microarray technology for unraveling the antibody specificity repertoire against microbial proteomes Curr Opin Mol Ther 2003 5 278 284 12870438
|
16231977
|
PMC1261511
|
CC BY
|
2021-01-05 11:13:40
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e339
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020339
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623197810.1371/journal.pmed.0020342PerspectivesGenetics/Genomics/Gene TherapyPharmacology/Drug DiscoveryHematologyHematology (including Blood Transfusion)GeneticsDrugs and adverse drug reactionsPharmacology and toxicologyPharmacogenetics in the Management of Coumarin Anticoagulant Therapy: The Way Forward or an Expensive Diversion? PerspectivesGreaves Mike Mike Greaves is Professor of Haematology at the School of Medicine, University of Aberdeen, Aberdeen, United Kingdom. E-mail: [email protected]
Competing Interests: I have been a paid expert advisor, on an ad hoc basis, to AstraZeneca (London, United Kingdom), a company developing thrombin inhibitors. I have no financial interest in this company's success (I have no shares in the company, nor have I received grants or awards from the company).
10 2005 25 10 2005 2 10 e342Copyright: © 2005 Mike Greaves.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 C1173T Dimorphism in the VKORC1 Gene Determines Coumarin Sensitivity and Bleeding Risk
Greaves discusses a new study in PLoS Medicine that takes us closer to being able to adopt a pharmacogenetic approach to reduce bleeding risk from coumarin therapy.
==== Body
Warfarin and related coumarin-based anticoagulants are the mainstay of pharmacological management for the long-term prevention of thromboembolism. It is estimated that over 1 million residents in the United Kingdom take warfarin; in excess of 20 million prescriptions for the anticoagulant are dispensed each year in the United States of America. Despite recent advances in the development of novel, alternative, oral antithrombotics it is likely that coumarins will be used widely for at least the next decade.
Mechanism of Action
Coumarins inhibit the posttranslational carboxylation of glutamate residues on proteins dependent on vitamin K, including coagulation factors II, VII, IX, and X. This carboxylation process is essential for their biological activity and requires the reduced form of vitamin K. Coumarins inhibit vitamin K epoxide reductase, the enzyme responsible for the recycling of vitamin K. The vitamin K epoxide reductase complex 1 (VKORC1) gene has been identified only recently [1] and this has permitted study of the effect of polymorphisms on sensitivity to coumarins.
The metabolism of warfarin depends principally upon the hepatic microsomal enzyme P450 2C9 (Cyp2C9), which catalyses degradation of the more potent S enantiomer to inactive metabolites.
Variations in the Anticoagulant Effect
Coumarins have very few side effects. However, by far the most common unwanted effect, abnormal bleeding, may be lethal. Overall, the rate of life-threatening bleeding is around two per 100 patient years and more minor bleeding is common [2]. This is despite the widespread adoption of the International Normalised Ratio (INR) as the method for standardisation of the prothrombin time, the coagulation assay used to measure the anticoagulant effect of warfarin.
Individual patients on warfarin are within the target INR range for only 50%–70% of the time.
For most clinical indications, dosing is aimed at achieving a target INR of 2.5 (range 2.0–3.0), which represents a level of anticoagulation associated with an optimal relationship between antithrombotic efficacy and bleeding risk. However, even with the best available management, individual patients on warfarin are within the target INR range for only 50%–70% of the time, on average. Although the reasons for this imprecision of dose response are often not immediately apparent, important contributing factors include co-medication with interacting drugs [3] and incomplete adherence. In addition, the dietary content of vitamin K has a measurable effect on the INR in patients taking warfarin [4].
A Genetic Component to Coumarin Sensitivity
In addition to this variation in dose-response within individuals, the inter-individual dose range required to achieve and maintain the target INR is exceptionally wide—for example, in the case of warfarin, a maintenance dose of 1 to >10 mg daily. Population studies suggested a genetic component to coumarin sensitivity; for example, people of Chinese origin are more sensitive, and African Americans less sensitive, than those of European ancestry [5,6]. This suggestion has now been confirmed; a significant proportion of the variation in warfarin sensitivity, including the variation between ethnic groups, can be accounted for by polymorphisms within the recently identified VKORC1 gene that influence transcriptional regulation [7]. This finding complements the earlier observation that variations in the rate of coumarin metabolism, due to the functional effects of polymorphisms in Cyp2C9, also contribute to individual coumarin sensitivity [8], as may polymorphisms in the genes for coagulation factors II and VII [9].
These findings raise the question of whether advances in understanding of the pharmacogenetics of oral anticoagulant therapy could beneficially influence clinical practice and patient safety by facilitating individualised coumarin dosing—hence reducing bleeding event rates. Answers are just beginning to emerge. In a retrospective study, Higashi et al. found that variant alleles of Cyp2C9 are associated with increased risk of above average INRs and longer duration to achieve stable dosing (compared to wild-type) in patients treated with warfarin [10]. Furthermore, patients with a variant genotype had increased risk of serious or life-threatening bleeding, although numbers of events were small, dictating caution in interpretation.
A New Study
Additional evidence of the potential for a pharmacogenetic approach to reduce bleeding risk from coumarin therapy is presented in the current issue of PLoS Medicine [11]. Reitsma and colleagues used a case-control study design to examine the effects of the C1173T polymorphism in intron 1 of the VKORC1 gene on dose requirement and occurrence of severe bleeding in subjects treated with the long-acting coumarin phenprocoumon or short-acting acenocoumarol. A significant effect of genotype on the dose required to achieve target INR was confirmed. In addition, among users of phenprocoumon, but not among users of acenocoumarol, carriers of at least one T allele appeared to have an increased risk of bleeding as well as greater sensitivity to coumarin. However, counterintuitively, the quality of INR control appeared to be better in subjects on acenocoumarol.
These intriguing findings, along with those of Higashi and colleagues, suggest that further prospective studies of the utility of a pharmacogenetic approach to improving the safety of oral anticoagulation with warfarin are justified. Such studies will of necessity be large and should include cost-effectiveness analyses. They should be implemented, but the question arises whether their importance may be undermined by the licensing for long-term use of alternative, novel oral anticoagulants that have much more predictable dose-response characteristics, such as direct thrombin inhibitors.
Citation: Greaves M (2005) Pharmacogenetics in the management of coumarin anticoagulant therapy: The way forward or an expensive diversion? PLoS Med 2(10): e342.
Abbreviations
Cyp2C9hepatic microsomal enzyme P450 2C9
INRInternational Normalised Ratio
VKORC1vitamin K epoxide reductase complex 1
==== Refs
References
Li T Chang CY Jin DY Lin PJ Khvorova A Identification of the gene for vitamin K epoxide reductase Nature 2004 427 541 544 14765195
Palareti G Leali N Coccheri S Poggi M Manotti C Bleeding complications of oral anticoagulant treatment: An inception-cohort, prospective collaborative study (ISCOAT). Italian Study on Complications of Oral Anticoagulant Therapy Lancet 1996 348 423 428 8709780
Panneerselvam S Baglin C Lefort W Baglin T Analysis of risk factors for over-anticoagulation in patients receiving long-term warfarin Brit J Haematol 1998 103 422 424 9827914
Lubetsky A Dekel-Stem E Chetrit A Lubin F Halkin H Vitamin K intake and sensitivity to warfarin in patients consuming regular diets Thrombos Haemostas 1999 81 396 399
Yu HC Chan TY Critchley JA Woo KS Factors determining the maintenance dose of warfarin in Chinese patients QJM 1996 89 127 135 8729554
Dang MT Hambleton J Kayser SR The influence of ethnicity on warfarin dosage requirement Ann Pharmacother 2005 39 1008 1012 15855242
Rieder MJ Reiner AP Gage BF Nickerson DA Eby CS Effect of VKORC1 haplotypes on transcriptional regulation and warfarin dose New Engl J Med 2005 352 2285 2293 15930419
Taube J Halsall D Baglin T Influence of cytochrome P-450 CYP2C9 polymorphisms on warfarin sensitivity and risk of over-anticoagulation in patients on long-term treatment Blood 2000 96 1816 1819 10961881
D'Ambrosio RL D'Andrea G Cappucci F Chetta M Di Perna P Polymorphisms in factor II and factor VII genes modulate oral anticoagulation with warfarin Haematologica 2004 89 1510 1516 15590403
Higashi MK Veenstra DL Kondo LM Wittkowski AK Srinouanprachanh SL Association between CYP2C9 genetic variants and anticoagulant-related outcomes during warfarin therapy JAMA 2002 287 1690 1698 11926893
Reitsma P Van der Heijden J Groot A Rosendaal F Büller H A C1173T dimorphism in the VRKORC1 gene determines coumarin sensitivity and bleeding risk PLoS Med 2005 2 e312 10.1371/journal.pmed.0020312 16201835
|
16231978
|
PMC1261512
|
CC BY
|
2021-01-05 10:38:16
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e342
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020342
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623197910.1371/journal.pmed.0020346Correspondence and Other CommunicationsEpidemiology/Public HealthHIV/AIDSHealth education (including prevention and promotion)HIV Infection/AIDSInfectious DiseasesSexual HealthSexually transmitted diseasesThe Tenofovir Pre-Exposure Prophylaxis Trial in Thailand: Researchers Should Show More Openness in Their Engagement with the Community CorrespondenceChua Arlene Ford Nathan Wilson David Cawthorne Paul
1
1Médecins Sans FrontièresBangkokThailandE-mail: [email protected]
Competing Interests: The authors have declared that no competing interests exist.
10 2005 25 10 2005 2 10 e346Copyright: © 2005 Chua 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 Abandoned Trials of Pre-Exposure Prophylaxis for HIV: What Went Wrong?
We Must Not Let Protestors Derail Trials of Pre-Exposure Prophylaxis for HIV
==== Body
Two recent articles in PLoS Medicine [1,2] criticise the role played by activists in raising concerns about the tenofovir trial for HIV prophylaxis. We fully support the fact that activism should be based on informed opinion, rather than speculation, unwarranted criticism, overreaction, or sensationalising facts [1], and believe that in Thailand, the concerns raised by activists are entirely legitimate.
The key community groups that have expressed concerns about the tenofovir trial in Thailand are the Thai Drug Users Network (TDN) and the Thai AIDS Treatment Advocacy Group (
TTAG), which are described in Box 1. These community groups, which can justifiably claim to represent Thai drug users, are well informed about the trial, but their objective concerns have been ignored by the trial investigators. Contrary to the assertion of Joep Lange [2] “that the investigators did consult intensely with community groups concerned”, TDN and
TTAG were not consulted about the trial design and conduct until a very late stage, after several attempts to engage with the investigators had been rebutted. TDN and
TTAG had attempted to constructively engage with the investigators since October 2004; they confined their statements of concern to private letters and meetings with the investigators, until the matter was made public in a Lancet editorial in March 2005 [3].
Box 1. Community Groups Representing IDUs in Thailand
Thai Drug Users Network
TDN, Thailand's only drug users' group, was established in December 2002 in response to the health and human rights crisis facing drug users in Thailand, particularly injectors. TDN's mission is to promote the basic human rights of people who use drugs, in order for them to be able to live with dignity in Thai society. TDN undertakes peer-driven HIV prevention, care, and support for IDUs, has supported the Thai government National Harm Reduction Working Group's activities, and has provided technical input to United Nations Office on Drugs and Crime and World Health Organization consultations. Projects are jointly implemented with other organisations, such as the project Preventing HIV and Increasing Care and Support for IDUs in Thailand, which is funded by the Global Fund to Fight AIDS, Tuberculosis, and Malaria (GFATM).
Thai AIDS Treatment Action Group
TTAG was founded in December 2002 by the founding chairman of the Thai Network of People Living with HIV/AIDS (PLHA) in order to promote leadership and advocacy among PLHA.
TTAG's mission is to promote equal access to AIDS prevention and treatment for all individuals through policy advocacy and coalition building, and by strengthening the capacity of people living with HIV/AIDS to advocate for their human rights. Projects include Preventing HIV and Increasing Care and Support for IDUs in Thailand, which is funded by GFATM, and the Mekong Region Treatment Preparedness Initiative.
Key Objections of TDN and
TTAG to the Tenofovir Pre-Exposure Prophylaxis Trial in Thailand in Its Present Form
Absence of community consultation during the trial design and lack of meaningful consultation during its implementation.
Best current prophylactic methods unavailable to trial participants.
No commitment by trial sponsors to promote the safety of trial participants when accessing services.
No commitment by the researchers to work, after the trial, with the Thai Ministry of Public Health towards price reductions of tenofovir.
Principle Recommendations
Urgently establish a committee, chaired by the US Centers for Disease Control and Prevention, to address key HIV prevention, treatment, and care issues for Bangkok IDUs in the context of the trial. Members should include two TDN representatives, and Thai Red Cross, government, and nongovernmental representatives.
Involve TDN in trial outreach and education, including curriculum development.
Develop partnerships to ensure the safety of trial participants when accessing services; for example, the Bangkok Metropolitan Authority should host police training workshops on harm reduction, and TDN should be involved in these activities.
Commit to supporting TDN in efforts to ensure at least two years of post-trial tenofovir to trial participants, and to working with the Thai Ministry of Public Health towards price reductions.
One major concern about the trial was the failure to provide sterile needles and syringes. Singh and Mills assert that this is consistent with Thai government policy [1]. Long prison terms and death sentences are the norm for drug-related offences [4], and Thai police, who have wide discretionary powers, still occasionally use possession of needles as evidence to arrest suspected drug users. Thus, although needles and syringes are available over the counter from most retail pharmacies, intravenous drug users (IDUs) are afraid to purchase them and, indeed, are often afraid to use services known to be provided for drug users. There is, however, no law or policy forbidding the distribution of clean needles and syringes, and preventing the investigators from doing so.
In fact, the situation in Thailand is improving, with the National Harm Reduction Working Group, chaired by the Ministry of Public Health, taking steps to increase activities in this domain. In 2004, at the 15th International AIDS Conference in Thailand, the prime minister said, “We are now implementing a harm reduction program to reduce the risk of HIV infection among injecting drug users…the program…will be conducted through concerted collaboration among solo UN agencies, government bodies and non-governmental organizations including the Drug User Network” [5].
The reason clean injection materials are not distributed within the trial is because the United States government, who sponsors the trial, bans federally funded organisations (including the Centers for Disease Control and Prevention, who are overseeing the trial) from supporting needle and syringe exchange.
Irrespective of whether a needle exchange exists in Thailand, or what the policies of the trial's funders are regarding needles and syringes, investigators have a duty to respect the Helsinki Declaration requirement that “benefits, risks, burdens and effectiveness of a new method should be tested against those of the best current prophylactic, diagnostic and therapeutic methods” [6].
The HIV/AIDS community in Thailand is not naive about the ethics of clinical trials: many have been directly or indirectly affected by previous AIDS drug trials in Thailand that have raised ethical concerns [7,8]. Nevertheless, TDN and
TTAG have, from the beginning, made it clear that they support the development of innovative prevention tools to reduce the burden of global HIV, and would like this trial to go ahead.
We believe that the disagreements surrounding the tenofovir trial in Thailand would have been avoided if the investigators had set out to engage the community more openly, and if the wealth of established knowledge among community members could have contributed enormously to the success of the trial design and implementation. TDN and
TTAG have made recommendations (see Box 1) that represent a constructive way for this trial to move forward. Mechanisms that ensure systematic involvement of legitimate representatives of the affected community as partners in research are the only way to ensure that future trials will proceed in a more productive way.
Citation: Chua A, Ford N, Wilson D, Cawthorne P (2005) The tenofovir pre-exposure prophylaxis trial in Thailand: Researchers should show more openness in their engagement with the community. PLoS Med 2(10): e346.
==== Refs
References
Singh J Mills E The abandoned trials of pre-exposure prophylaxis for HIV: What went wrong? PLoS Med 2005 2 e234 10.1371/journal.pmed.0020234 16008507
Lange J We must not let protestors derail trials of pre-exposure prophylaxis for HIV PLoS Med 2005 2 e248 10.1371/journal.pmed.0020248 16008501
[Anonymous] The trials of tenofovir trials Lancet 2005 365 1111 15799093
Human Rights Watch Thailand—Not enough graves: The war on drugs, HIV/AIDS, and violations of human rights 2004 New York Human Rights Watch Available: http://www.hrw.org/reports/2004/thailand0704/thailand0704.pdf . Accessed 31 August 2005
Shinawatra T Opening address 15th International AIDS Conference; 2004 July 11-–16 July 2004 Bangkok, Thailand International Aids Society, Thai Ministry of Public Health Available: http://www.kaisernetwork.org/health_cast/uploaded_files/071104_ias_opening.pdf . Accessed 6 September 2005
Jintarkanon S Nakapiew S Tienudom N Suwannawong P Wilson D Unethical clinical trials in Thailand: A community response Lancet 2005 367 1617 1618
Suwanjandee J Wilson D Helsinki declaration and Thailand Lancet 1999 354 343
Wilson D North–South research in developing countries must respond to community's priorities BMJ 1999 319 1496 1497
|
16231979
|
PMC1261513
|
CC BY
|
2021-01-05 10:38:15
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e346
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020346
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623198010.1371/journal.pmed.0020347Correspondence and Other CommunicationsInfectious DiseasesEpidemiology/Public HealthHealth PolicyHIV/AIDSClinical trialsHealth education (including prevention and promotion)Health PolicyHIV Infection/AIDSInfectious DiseasesResponse to Joep M. A. Lange CorrespondenceDitmore Melissa
1
1Research for Sex Work, Chiang Mai, Thailand; Network of Sex Work Projects, Hong Kong, ChinaE-mail: [email protected]
Competing Interests: MD was an inaugural board member of the Network of Sex Work Projects, and is Editor of Research for Sex Work, which is published by the Network of Sex Work Projects. MD has encouraged participation, and recommended specific people for meetings with the Gates Foundation and the Joint United Nations Programme on HIV/AIDS, but has not been employed by these agencies. MD has advocated for participatory approaches to research.
10 2005 25 10 2005 2 10 e347Copyright: © 2005 Melissa Ditmore.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 Must Not Let Protestors Derail Trials of Pre-Exposure Prophylaxis for HIV
==== Body
Joep M. A. Lange, in his PLoS Medicine Perspective [1], neglected the complaints of the sex workers in the tenofovir trials—a lack of attention to their long-term health care, an appalling lack of answers to their questions about side effects, inadequate translation of the trial materials in some places, and, in past experience, a failure to deliver any new drugs to sex workers in the developing world once a trial is complete. Cambodian sex workers were offended by the assertion of trial conductors that they should be the bodies on which tests are conducted for the benefit of the rest of the world, without guarantees of health care for side effects and infections that occur during a trial, and without even receiving answers to their questions about the trial. Most people in developed countries would have been offended if they'd been asked to do the same. The expectation that marginalized populations will accept such ungracious treatment is patently offensive.
Sex workers would like to see research continue, and would like even more to see sex workers have access to effective treatment and prevention. Trial participants in developed countries have been motivated to push for faster development of drugs by the need for treatment. Sex workers have met with the organizers and supporters of the trials—this is extremely cooperative! Some meetings have been good and others have been baldfaced tokenism, some even without language translation. Nothing says “we don't care what you have to say” louder than not translating for someone flown thousands of miles to attend a meeting for two days.
When sex workers and other marginalized people are genuine participants with input at all stages of research, they will be eager research participants. There is a good example of this in the journal Research for Sex Work. There is an update on the tenofovir trials, but the lead article is titled “Cambodian sex workers conduct their own research” [2]. These sex workers were invited to choose a topic and design research that would be useful for them. They did so because they had input into the research at every level. Drug researchers should take note.
Citation: Ditmore M (2005) Response to Joep M. A. Lange. PLoS Med 2(10): e347.
==== Refs
References
Lange J We must not let protestors derail trials of pre-exposure prophylaxis for HIV PLoS Med 2005 2 e248 10.1371/journal.pmed.0020248 16008501
Jenkins C Cambodian sex workers conduct their own research Res Sex Work 2005 2005 3 Available: http://www.researchforsexwork.org . Accessed 31 August 2005
|
16231980
|
PMC1261514
|
CC BY
|
2021-01-05 10:38:17
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e347
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020347
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623198110.1371/journal.pmed.0020348PerspectivesGenetics/Genomics/Gene TherapyMental HealthSchizophrenia and Other Psychotic DisordersGeneticsDoes the CAPON Gene Confer Susceptibility to Schizophrenia? PerspectiveEastwood Sharon L Sharon L. Eastwood is in the Department of Psychiatry, University of Oxford, Oxford, United Kingdom. E-mail: [email protected]
Competing Interests: The author declares that no competing interests exist.
10 2005 25 10 2005 2 10 e348Copyright: © 2005 Sharon L. Eastwood.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.
Increased Expression in Dorsolateral Prefrontal Cortex of CAPON in Schizophrenia and Bipolar Disorder
Eastwood discusses a new study in PLoS Medicine that suggests that overexpression of the CAPON gene, leading to disruption of NMDA receptor function, may be important in the etiology of severe mental illnesses.
==== Body
A major breakthrough towards a greater understanding of the pathophysiology of severe mental illness came with the recent identification of potential susceptibility genes for schizophrenia and bipolar disorder, the two main diagnostic categories of functional psychoses. Although it has been known for many years that schizophrenia and bipolar disorder tend to run in families, and twin and adoption studies have clearly established the heritability of both disorders, no single “gene for schizophrenia” or “gene for bipolar disorder” appears to exist. Instead, the pattern of inheritance suggests that these common mental illnesses are caused by many different genes of small effect, acting together to confer susceptibility.
For schizophrenia, several candidate susceptibility genes have been identified, but for all except one of them, the single-nucleotide polymorphisms found to be associated with the disorder are noncoding. Hence, changes in the expression of the genes (for example, in terms of their splicing or relative abundance), rather than an amino acid alteration, are thought to underlie their genetic association [1]. To investigate this possibility further, studies of gene expression in postmortem brains are required.
A New Gene Expression Study
In this month's PLoS Medicine, Linda Brzustowicz and colleagues present the results of their study of one potential schizophrenia susceptibility gene, CAPON (carboxyl-terminal PDZ ligand of neuronal nitric oxide synthase), using postmortem tissue samples of the dorsolateral prefrontal cortex, a pathological “hotspot” identified in studies of schizophrenia [2].
The gene for CAPON is located on Chromosome 1q22, a locus of interest initially identified from linkage studies of schizophrenia. Two studies, including one by Brzustowicz and colleagues, subsequently detected an association between single-nucleotide polymorphisms within the CAPON gene and the occurrence of schizophrenia [3,4]. The role of CAPON in binding nitric oxide synthase (Figure 1) places it at the centre of regulation of N-methyl-D-aspartate (NMDA) receptor–mediated glutamate neurotransmission, abnormalities of which have long been proposed to be involved in schizophrenia, lending credence to CAPON as a susceptibility gene for the disorder. However, as sequencing of CAPON failed to reveal a coding mutation, Brzustowicz and colleagues went on to examine whether the expression of the gene is altered in schizophrenia.
Figure 1 CAPON Binds Nitric Oxide Synthase, Regulating NMDA Receptor–Mediated Glutamate Neurotransmission
Binding of CAPON results in a reduction of NMDA receptor/nitric oxide synthase (NOS) complexes, leading to decreased NMDA receptor–gated calcium influx and a catalytically inactive nitric oxide synthase. Overexpression of either the full-length or the novel shortened CAPON isoform as reported by Brzustowicz and colleagues is, therefore, predicted to lead to impaired NMDA receptor–mediated glutamate neurotransmission.
(Illustration: Bang Wong, http://www.clearscience.info)
First, using standard techniques, the authors identified a novel CAPON mRNA transcript, and verified that the predicted shortened protein was expressed in the brain. In order to continue on with this work, and determine if CAPON expression was altered in schizophrenia, they used RNA samples isolated from the dorsolateral prefrontal cortex of psychiatrically normal individuals, individuals with schizophrenia, and individuals with bipolar disorder. These samples were provided by the Stanley Medical Research Institute (http://www.stanleyresearch.org/programs/brain_collection.asp). Although the authors' primary interest was schizophrenia, as a condition of use, the Stanley Medical Research Institute required investigators to be blinded to diagnosis, and, hence, the experiments were run on individuals with bipolar disorder as well.
Using quantitative real-time polymerase chain reaction, the authors found that while the mRNA for the full-length CAPON protein was unchanged in either disorder, for both schizophrenia and bipolar disorder, mRNA expression of the novel shortened CAPON isoform was increased compared to the psychiatrically normal controls. Antipsychotic drug treatment received by the patients did not appear to underlie the authors' findings, and genotyping revealed that individuals with one or two copies of alleles previously identified as associated with schizophrenia had higher levels of the novel CAPON isoform mRNA. The authors concluded that overexpression of either CAPON isoform would be expected to disrupt NMDA receptor function, and that the results of their study not only added support to the role of CAPON in schizophrenia, but also implicated the gene in the aetiology of bipolar disorder.
Limitations of the Study
All postmortem studies such as this are limited by the suitability of the material available, and the brain series from which they came. The Stanley Medical Research Institute series used by the authors is well matched for variables known to influence studies of gene expression, and the results of the study are likely to be genuine. However, replication in another brain series should be attempted. In addition, as this was a homogenate-based study of one brain region, future studies should examine other areas, and also use additional techniques (such as in situ hybridization histochemistry) to determine whether expression is altered in all or only specific neuronal populations. As the authors were limited to examining RNA, it will also be interesting to determine if expression of the encoded shortened CAPON protein is similarly increased in schizophrenia and bipolar disorder. Lastly, functional in vitro studies examining the knock-on effects of overexpression of the shortened CAPON protein may be especially useful for understanding the function of this novel isoform and the pathophysiological consequences of the authors' findings.
Conclusion
This study is an example of one approach that may be used to help us begin to understand how genes, such as CAPON, may confer risk for disease susceptibility. The paper also highlights the growing awareness that although, classically, schizophrenia and bipolar disorder have been conceptualised as separate disorders, it may be more appropriate to envisage them as phenotypical extremes of a continuum of disease caused by an overlapping set of genes [5]. Although the findings of this study are not of immediate clinical relevance, a better understanding of the genes and pathways that are involved will ultimately lead to the development of more effective treatments for both disorders.
Citation: Eastwood SL (2005) Does the CAPON gene confer susceptibility to schizophrenia? PLoS Med 2(10): e348.
Abbreviations
CAPONcarboxyl-terminal PDZ ligand of neuronal nitric oxide synthase
NMDAN-methyl-D-aspartate
==== Refs
References
Harrison PJ Weinberger DR Schizophrenia genes, gene expression, and neuropathology: On the matter of their convergence Mol Psychiatry 2005 10 40 68 15263907
Xu B Wratten N Charych EI Buyske S Firestein BL Increased expression in dorsolateral prefrontal cortex of CAPON in schizophrenia and bipolar disorder PLoS Med 2005 2 e263 10.1371/journal.pmed.0020263 16146415
Brzustowicz LM Simone J Mohseni P Hayter JE Hodgkinson KA Linkage disequilibrium mapping of schizophrenia susceptibility to the CAPON region of chromosome 1q22 Am J Hum Genet 2004 74 1057 1063 15065015
Zheng Y Li H Qin W Chen W Duan Y Association of the carboxyl-terminal PDZ ligand of neuronal nitric oxide synthase gene with schizophrenia in the Chinese Han population Biochem Biophys Res Commun 2005 328 809 815 15707951
Craddock N O'Donovan MC Owen MJ The genetics of schizophrenia and bipolar disorder: Dissecting psychosis Med Genet 2005 42 193 204
|
16231981
|
PMC1261515
|
CC BY
|
2021-01-05 10:38:16
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e348
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020348
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623198210.1371/journal.pmed.0020349PerspectivesCancer BiologyGenetics/Genomics/Gene TherapyMolecular Biology/Structural BiologyOtherOncologyOpthalmologyPediatricsOncologyPediatricsOphthalmologyRetinoblastoma: Teacher of Cancer Biology and Medicine PerspectiveKnudson Alfred Alfred Knudson is at the Fox Chase Cancer Center, Philadelphia, Pennsylvania, United States of America. E-mail: [email protected]
Competing Interests: The author declares that no competing interests exist.
10 2005 25 10 2005 2 10 e349Copyright: © 2005 Alfred Knudson.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.
Use of Preclinical Models to Improve Treatment of Retinoblastoma
Retinoblastoma attracts the interest of oncologists as well as cancer biologists. Knudson discusses why retinoblastoma has been a model for guiding our understanding of cancer genes.
==== Body
Despite its low incidence, retinoblastoma has attracted the great interest of not just pediatric oncologists, but also cancer biologists and oncologists who treat cancer in adults. Retinoblastoma reminds us that rare conditions with a hereditary component may be rare for a reason—the genes associated with them are important biologically. It is hardly surprising then that retinoblastoma continues to be a subject of interest to a large group of clinicians and scientists, as described by Dyer and colleagues in this issue of PLoS Medicine [1].
A Success Story
For the pediatric oncologist, the treatment of retinoblastoma has been one of the success stories of the past century, with a cure rate in the United States of 95% [2,3]. The clinicians' concerns are now those children who do not survive, those who lose one or both eyes, and those who sustain delayed effects, including late death from a second cancer.
For cancer biologists, including myself, retinoblastoma has been a model for guiding our understanding of cancer genes. For me, it was a simplifying condition to study: a child could inherit a mutated gene that could produce a cancer even in a newborn, so such a gene must surely be important for cancer research. Indeed, the retinoblastoma gene RB1 was the first hereditary cancer gene and the first tumor-suppressor gene to be cloned, and it has proven to be biologically important. The RB protein was shown to be a principal controller of the cell cycle in all human tissues. Furthermore, RB protein is inactivated by specific proteins produced by DNA tumor viruses, thereby demonstrating the oncogenic mode of action of their already known transforming factors. Showing this inactivation was, historically, critically important in resolving conflicting notions of viral versus genetic oncogenesis, just as the RNA tumor viruses directed investigators to proto-oncogenes.
Lessons for Other Cancers
Pediatric oncologists treating retinoblastoma continue to address problems of general interest to all oncologists. The discovery that RB1 is a gene whose mutations can also predispose a person to sarcomas, which in turn can be initiated by ionizing radiation, has implications for other cancers, and pediatrics has led the way in modifying therapy to reduce the probability of such a consequence. The use of chemotherapy and focal treatment to reduce the need for bilateral enucleation (surgical removal of the eye from its orbit) and/or radiation provides another example of the amelioration of late effects. The incidences of both late mortality and blindness have been reduced.
Retinoblastoma has been a model for guiding our understanding of cancer genes.
Following the cloning of RB1, more than 40 hereditary cancer genes have been cloned, including the gene for familial adenomatous polyposis, APC. As with retinoblastoma, APC was shown to be somatically mutant in a large majority of nonhereditary colon cancers, thereby stimulating interest in the study of hereditary cancer for the purpose of illuminating the biology of all cancer. However, the common carcinomas are often preceded by benign precursor lesions, such as the colonic polyp, whose diagnosis and removal can be an important preventive measure, especially since there is often a considerable time interval between formation of the precursor and its malignant transformation.
Making Further Clinical Progress
Further clinical progress may depend on a deeper understanding of the mechanisms of transformation in cells with defective or absent RB protein. Still unanswered is the following question: why is RB1 a retinoblastoma gene? Knowledge of control of the cell cycle by RB protein does not explain the protein's tissue specificity or its importance for retinal development. Following the cloning of RB1, knockout mice were produced, enabling the production and study of homozygotes. The homozygous state is lethal to embryos, with developmental defects in multiple tissues, including the brain. If heterozygous humans did not also develop tumors, we might know RB1 as a recessive, developmental, lethal gene. Can the developmental defects be explained by what is known about control of the cell cycle by RB protein? Apparently not.
Now Dyer and colleagues, and other researchers [1] have discovered that RB protein interacts with other proteins that have separate developmental roles. A mutant retinal tissue progenitor cell is apparently arrested along its developmental pathway, and the arrested daughter cells continue to divide indefinitely. This discovery has stimulated a search for separate agents affecting either differentiation or development, one such combination being carboplatin and topotecan.
Even if a cure is usually accomplished with the application of current knowledge, there remain two problems. The first problem is the lethal progression of a few cases. The second is the problem of late effects, especially the difficult effect of second cancers in hereditary cases. Since these problems are also issues for other cancers, retinoblastoma assumes a special importance, with a seemingly better opportunity to identify secondary events in tumors. The rapid growth of retinoblastoma may facilitate recognition of rate-limiting steps in progression, especially with the availability of a good animal model. Since RB1 and other functionally related genes are somatically mutant in so many cancers, study of them should be widely relevant.
A New Role for Pediatric Oncologists?
Can pediatric oncologists participate directly in reducing the burden of cancer in adults? Is there a possibility of prevention? I have serious doubts that much could be done to prevent cancer in children, but intervention may be feasible in hereditary conditions that cause cancer in adults. Thus, both familial adenomatous polyposis and neurofibromatosis type 1 have incidences more than twice that of retinoblastoma. Successful intervention in childhood might possibly lead to delay or prevention of later malignancies. Any success would also be relevant to prevention of nonhereditary cancer, which would be especially important for colon cancer since most cases involve somatic mutations in the APC gene. Pediatric oncologists may find themselves in a new leadership role.
Citation: Knudson A (2005) Retinoblastoma: Teacher of cancer biology and medicine. PLoS Med 2(10): e349.
==== Refs
References
Dyer MA Rodriguez-Galindo C Wilson M Use of preclinical models to improve treatment of retinoblastoma PLoS Med 2005 2 e332 10.1371/journal.pmed.0020332 16231976
Eng C Li FP Abramson DH Ellsworth RM Wong FL Mortality from second tumors among long-term survivors of retinoblastoma J Natl Cancer Inst 1993 85 1121 1128 8320741
Byrne J Fears TR Whitney C Parry DM Survival after retinoblastoma: Long-term consequences and family history of cancer Med Pediatr Oncol 1995 24 160 165 7838037
|
16231982
|
PMC1261516
|
CC BY
|
2021-01-05 10:39:16
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e349
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020349
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623198210.1371/journal.pmed.0020349PerspectivesCancer BiologyGenetics/Genomics/Gene TherapyMolecular Biology/Structural BiologyOtherOncologyOpthalmologyPediatricsOncologyPediatricsOphthalmologyRetinoblastoma: Teacher of Cancer Biology and Medicine PerspectiveKnudson Alfred Alfred Knudson is at the Fox Chase Cancer Center, Philadelphia, Pennsylvania, United States of America. E-mail: [email protected]
Competing Interests: The author declares that no competing interests exist.
10 2005 25 10 2005 2 10 e349Copyright: © 2005 Alfred Knudson.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.
Use of Preclinical Models to Improve Treatment of Retinoblastoma
Retinoblastoma attracts the interest of oncologists as well as cancer biologists. Knudson discusses why retinoblastoma has been a model for guiding our understanding of cancer genes.
==== Body
Despite its low incidence, retinoblastoma has attracted the great interest of not just pediatric oncologists, but also cancer biologists and oncologists who treat cancer in adults. Retinoblastoma reminds us that rare conditions with a hereditary component may be rare for a reason—the genes associated with them are important biologically. It is hardly surprising then that retinoblastoma continues to be a subject of interest to a large group of clinicians and scientists, as described by Dyer and colleagues in this issue of PLoS Medicine [1].
A Success Story
For the pediatric oncologist, the treatment of retinoblastoma has been one of the success stories of the past century, with a cure rate in the United States of 95% [2,3]. The clinicians' concerns are now those children who do not survive, those who lose one or both eyes, and those who sustain delayed effects, including late death from a second cancer.
For cancer biologists, including myself, retinoblastoma has been a model for guiding our understanding of cancer genes. For me, it was a simplifying condition to study: a child could inherit a mutated gene that could produce a cancer even in a newborn, so such a gene must surely be important for cancer research. Indeed, the retinoblastoma gene RB1 was the first hereditary cancer gene and the first tumor-suppressor gene to be cloned, and it has proven to be biologically important. The RB protein was shown to be a principal controller of the cell cycle in all human tissues. Furthermore, RB protein is inactivated by specific proteins produced by DNA tumor viruses, thereby demonstrating the oncogenic mode of action of their already known transforming factors. Showing this inactivation was, historically, critically important in resolving conflicting notions of viral versus genetic oncogenesis, just as the RNA tumor viruses directed investigators to proto-oncogenes.
Lessons for Other Cancers
Pediatric oncologists treating retinoblastoma continue to address problems of general interest to all oncologists. The discovery that RB1 is a gene whose mutations can also predispose a person to sarcomas, which in turn can be initiated by ionizing radiation, has implications for other cancers, and pediatrics has led the way in modifying therapy to reduce the probability of such a consequence. The use of chemotherapy and focal treatment to reduce the need for bilateral enucleation (surgical removal of the eye from its orbit) and/or radiation provides another example of the amelioration of late effects. The incidences of both late mortality and blindness have been reduced.
Retinoblastoma has been a model for guiding our understanding of cancer genes.
Following the cloning of RB1, more than 40 hereditary cancer genes have been cloned, including the gene for familial adenomatous polyposis, APC. As with retinoblastoma, APC was shown to be somatically mutant in a large majority of nonhereditary colon cancers, thereby stimulating interest in the study of hereditary cancer for the purpose of illuminating the biology of all cancer. However, the common carcinomas are often preceded by benign precursor lesions, such as the colonic polyp, whose diagnosis and removal can be an important preventive measure, especially since there is often a considerable time interval between formation of the precursor and its malignant transformation.
Making Further Clinical Progress
Further clinical progress may depend on a deeper understanding of the mechanisms of transformation in cells with defective or absent RB protein. Still unanswered is the following question: why is RB1 a retinoblastoma gene? Knowledge of control of the cell cycle by RB protein does not explain the protein's tissue specificity or its importance for retinal development. Following the cloning of RB1, knockout mice were produced, enabling the production and study of homozygotes. The homozygous state is lethal to embryos, with developmental defects in multiple tissues, including the brain. If heterozygous humans did not also develop tumors, we might know RB1 as a recessive, developmental, lethal gene. Can the developmental defects be explained by what is known about control of the cell cycle by RB protein? Apparently not.
Now Dyer and colleagues, and other researchers [1] have discovered that RB protein interacts with other proteins that have separate developmental roles. A mutant retinal tissue progenitor cell is apparently arrested along its developmental pathway, and the arrested daughter cells continue to divide indefinitely. This discovery has stimulated a search for separate agents affecting either differentiation or development, one such combination being carboplatin and topotecan.
Even if a cure is usually accomplished with the application of current knowledge, there remain two problems. The first problem is the lethal progression of a few cases. The second is the problem of late effects, especially the difficult effect of second cancers in hereditary cases. Since these problems are also issues for other cancers, retinoblastoma assumes a special importance, with a seemingly better opportunity to identify secondary events in tumors. The rapid growth of retinoblastoma may facilitate recognition of rate-limiting steps in progression, especially with the availability of a good animal model. Since RB1 and other functionally related genes are somatically mutant in so many cancers, study of them should be widely relevant.
A New Role for Pediatric Oncologists?
Can pediatric oncologists participate directly in reducing the burden of cancer in adults? Is there a possibility of prevention? I have serious doubts that much could be done to prevent cancer in children, but intervention may be feasible in hereditary conditions that cause cancer in adults. Thus, both familial adenomatous polyposis and neurofibromatosis type 1 have incidences more than twice that of retinoblastoma. Successful intervention in childhood might possibly lead to delay or prevention of later malignancies. Any success would also be relevant to prevention of nonhereditary cancer, which would be especially important for colon cancer since most cases involve somatic mutations in the APC gene. Pediatric oncologists may find themselves in a new leadership role.
Citation: Knudson A (2005) Retinoblastoma: Teacher of cancer biology and medicine. PLoS Med 2(10): e349.
==== Refs
References
Dyer MA Rodriguez-Galindo C Wilson M Use of preclinical models to improve treatment of retinoblastoma PLoS Med 2005 2 e332 10.1371/journal.pmed.0020332 16231976
Eng C Li FP Abramson DH Ellsworth RM Wong FL Mortality from second tumors among long-term survivors of retinoblastoma J Natl Cancer Inst 1993 85 1121 1128 8320741
Byrne J Fears TR Whitney C Parry DM Survival after retinoblastoma: Long-term consequences and family history of cancer Med Pediatr Oncol 1995 24 160 165 7838037
|
16231983
|
PMC1261520
|
CC BY
|
2021-01-05 10:38:17
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e362
|
latin-1
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020362
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623198210.1371/journal.pmed.0020349PerspectivesCancer BiologyGenetics/Genomics/Gene TherapyMolecular Biology/Structural BiologyOtherOncologyOpthalmologyPediatricsOncologyPediatricsOphthalmologyRetinoblastoma: Teacher of Cancer Biology and Medicine PerspectiveKnudson Alfred Alfred Knudson is at the Fox Chase Cancer Center, Philadelphia, Pennsylvania, United States of America. E-mail: [email protected]
Competing Interests: The author declares that no competing interests exist.
10 2005 25 10 2005 2 10 e349Copyright: © 2005 Alfred Knudson.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.
Use of Preclinical Models to Improve Treatment of Retinoblastoma
Retinoblastoma attracts the interest of oncologists as well as cancer biologists. Knudson discusses why retinoblastoma has been a model for guiding our understanding of cancer genes.
==== Body
Despite its low incidence, retinoblastoma has attracted the great interest of not just pediatric oncologists, but also cancer biologists and oncologists who treat cancer in adults. Retinoblastoma reminds us that rare conditions with a hereditary component may be rare for a reason—the genes associated with them are important biologically. It is hardly surprising then that retinoblastoma continues to be a subject of interest to a large group of clinicians and scientists, as described by Dyer and colleagues in this issue of PLoS Medicine [1].
A Success Story
For the pediatric oncologist, the treatment of retinoblastoma has been one of the success stories of the past century, with a cure rate in the United States of 95% [2,3]. The clinicians' concerns are now those children who do not survive, those who lose one or both eyes, and those who sustain delayed effects, including late death from a second cancer.
For cancer biologists, including myself, retinoblastoma has been a model for guiding our understanding of cancer genes. For me, it was a simplifying condition to study: a child could inherit a mutated gene that could produce a cancer even in a newborn, so such a gene must surely be important for cancer research. Indeed, the retinoblastoma gene RB1 was the first hereditary cancer gene and the first tumor-suppressor gene to be cloned, and it has proven to be biologically important. The RB protein was shown to be a principal controller of the cell cycle in all human tissues. Furthermore, RB protein is inactivated by specific proteins produced by DNA tumor viruses, thereby demonstrating the oncogenic mode of action of their already known transforming factors. Showing this inactivation was, historically, critically important in resolving conflicting notions of viral versus genetic oncogenesis, just as the RNA tumor viruses directed investigators to proto-oncogenes.
Lessons for Other Cancers
Pediatric oncologists treating retinoblastoma continue to address problems of general interest to all oncologists. The discovery that RB1 is a gene whose mutations can also predispose a person to sarcomas, which in turn can be initiated by ionizing radiation, has implications for other cancers, and pediatrics has led the way in modifying therapy to reduce the probability of such a consequence. The use of chemotherapy and focal treatment to reduce the need for bilateral enucleation (surgical removal of the eye from its orbit) and/or radiation provides another example of the amelioration of late effects. The incidences of both late mortality and blindness have been reduced.
Retinoblastoma has been a model for guiding our understanding of cancer genes.
Following the cloning of RB1, more than 40 hereditary cancer genes have been cloned, including the gene for familial adenomatous polyposis, APC. As with retinoblastoma, APC was shown to be somatically mutant in a large majority of nonhereditary colon cancers, thereby stimulating interest in the study of hereditary cancer for the purpose of illuminating the biology of all cancer. However, the common carcinomas are often preceded by benign precursor lesions, such as the colonic polyp, whose diagnosis and removal can be an important preventive measure, especially since there is often a considerable time interval between formation of the precursor and its malignant transformation.
Making Further Clinical Progress
Further clinical progress may depend on a deeper understanding of the mechanisms of transformation in cells with defective or absent RB protein. Still unanswered is the following question: why is RB1 a retinoblastoma gene? Knowledge of control of the cell cycle by RB protein does not explain the protein's tissue specificity or its importance for retinal development. Following the cloning of RB1, knockout mice were produced, enabling the production and study of homozygotes. The homozygous state is lethal to embryos, with developmental defects in multiple tissues, including the brain. If heterozygous humans did not also develop tumors, we might know RB1 as a recessive, developmental, lethal gene. Can the developmental defects be explained by what is known about control of the cell cycle by RB protein? Apparently not.
Now Dyer and colleagues, and other researchers [1] have discovered that RB protein interacts with other proteins that have separate developmental roles. A mutant retinal tissue progenitor cell is apparently arrested along its developmental pathway, and the arrested daughter cells continue to divide indefinitely. This discovery has stimulated a search for separate agents affecting either differentiation or development, one such combination being carboplatin and topotecan.
Even if a cure is usually accomplished with the application of current knowledge, there remain two problems. The first problem is the lethal progression of a few cases. The second is the problem of late effects, especially the difficult effect of second cancers in hereditary cases. Since these problems are also issues for other cancers, retinoblastoma assumes a special importance, with a seemingly better opportunity to identify secondary events in tumors. The rapid growth of retinoblastoma may facilitate recognition of rate-limiting steps in progression, especially with the availability of a good animal model. Since RB1 and other functionally related genes are somatically mutant in so many cancers, study of them should be widely relevant.
A New Role for Pediatric Oncologists?
Can pediatric oncologists participate directly in reducing the burden of cancer in adults? Is there a possibility of prevention? I have serious doubts that much could be done to prevent cancer in children, but intervention may be feasible in hereditary conditions that cause cancer in adults. Thus, both familial adenomatous polyposis and neurofibromatosis type 1 have incidences more than twice that of retinoblastoma. Successful intervention in childhood might possibly lead to delay or prevention of later malignancies. Any success would also be relevant to prevention of nonhereditary cancer, which would be especially important for colon cancer since most cases involve somatic mutations in the APC gene. Pediatric oncologists may find themselves in a new leadership role.
Citation: Knudson A (2005) Retinoblastoma: Teacher of cancer biology and medicine. PLoS Med 2(10): e349.
==== Refs
References
Dyer MA Rodriguez-Galindo C Wilson M Use of preclinical models to improve treatment of retinoblastoma PLoS Med 2005 2 e332 10.1371/journal.pmed.0020332 16231976
Eng C Li FP Abramson DH Ellsworth RM Wong FL Mortality from second tumors among long-term survivors of retinoblastoma J Natl Cancer Inst 1993 85 1121 1128 8320741
Byrne J Fears TR Whitney C Parry DM Survival after retinoblastoma: Long-term consequences and family history of cancer Med Pediatr Oncol 1995 24 160 165 7838037
|
16231984
|
PMC1261521
|
CC BY
|
2021-01-05 10:38:16
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e363
|
latin-1
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020363
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623198210.1371/journal.pmed.0020349PerspectivesCancer BiologyGenetics/Genomics/Gene TherapyMolecular Biology/Structural BiologyOtherOncologyOpthalmologyPediatricsOncologyPediatricsOphthalmologyRetinoblastoma: Teacher of Cancer Biology and Medicine PerspectiveKnudson Alfred Alfred Knudson is at the Fox Chase Cancer Center, Philadelphia, Pennsylvania, United States of America. E-mail: [email protected]
Competing Interests: The author declares that no competing interests exist.
10 2005 25 10 2005 2 10 e349Copyright: © 2005 Alfred Knudson.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.
Use of Preclinical Models to Improve Treatment of Retinoblastoma
Retinoblastoma attracts the interest of oncologists as well as cancer biologists. Knudson discusses why retinoblastoma has been a model for guiding our understanding of cancer genes.
==== Body
Despite its low incidence, retinoblastoma has attracted the great interest of not just pediatric oncologists, but also cancer biologists and oncologists who treat cancer in adults. Retinoblastoma reminds us that rare conditions with a hereditary component may be rare for a reason—the genes associated with them are important biologically. It is hardly surprising then that retinoblastoma continues to be a subject of interest to a large group of clinicians and scientists, as described by Dyer and colleagues in this issue of PLoS Medicine [1].
A Success Story
For the pediatric oncologist, the treatment of retinoblastoma has been one of the success stories of the past century, with a cure rate in the United States of 95% [2,3]. The clinicians' concerns are now those children who do not survive, those who lose one or both eyes, and those who sustain delayed effects, including late death from a second cancer.
For cancer biologists, including myself, retinoblastoma has been a model for guiding our understanding of cancer genes. For me, it was a simplifying condition to study: a child could inherit a mutated gene that could produce a cancer even in a newborn, so such a gene must surely be important for cancer research. Indeed, the retinoblastoma gene RB1 was the first hereditary cancer gene and the first tumor-suppressor gene to be cloned, and it has proven to be biologically important. The RB protein was shown to be a principal controller of the cell cycle in all human tissues. Furthermore, RB protein is inactivated by specific proteins produced by DNA tumor viruses, thereby demonstrating the oncogenic mode of action of their already known transforming factors. Showing this inactivation was, historically, critically important in resolving conflicting notions of viral versus genetic oncogenesis, just as the RNA tumor viruses directed investigators to proto-oncogenes.
Lessons for Other Cancers
Pediatric oncologists treating retinoblastoma continue to address problems of general interest to all oncologists. The discovery that RB1 is a gene whose mutations can also predispose a person to sarcomas, which in turn can be initiated by ionizing radiation, has implications for other cancers, and pediatrics has led the way in modifying therapy to reduce the probability of such a consequence. The use of chemotherapy and focal treatment to reduce the need for bilateral enucleation (surgical removal of the eye from its orbit) and/or radiation provides another example of the amelioration of late effects. The incidences of both late mortality and blindness have been reduced.
Retinoblastoma has been a model for guiding our understanding of cancer genes.
Following the cloning of RB1, more than 40 hereditary cancer genes have been cloned, including the gene for familial adenomatous polyposis, APC. As with retinoblastoma, APC was shown to be somatically mutant in a large majority of nonhereditary colon cancers, thereby stimulating interest in the study of hereditary cancer for the purpose of illuminating the biology of all cancer. However, the common carcinomas are often preceded by benign precursor lesions, such as the colonic polyp, whose diagnosis and removal can be an important preventive measure, especially since there is often a considerable time interval between formation of the precursor and its malignant transformation.
Making Further Clinical Progress
Further clinical progress may depend on a deeper understanding of the mechanisms of transformation in cells with defective or absent RB protein. Still unanswered is the following question: why is RB1 a retinoblastoma gene? Knowledge of control of the cell cycle by RB protein does not explain the protein's tissue specificity or its importance for retinal development. Following the cloning of RB1, knockout mice were produced, enabling the production and study of homozygotes. The homozygous state is lethal to embryos, with developmental defects in multiple tissues, including the brain. If heterozygous humans did not also develop tumors, we might know RB1 as a recessive, developmental, lethal gene. Can the developmental defects be explained by what is known about control of the cell cycle by RB protein? Apparently not.
Now Dyer and colleagues, and other researchers [1] have discovered that RB protein interacts with other proteins that have separate developmental roles. A mutant retinal tissue progenitor cell is apparently arrested along its developmental pathway, and the arrested daughter cells continue to divide indefinitely. This discovery has stimulated a search for separate agents affecting either differentiation or development, one such combination being carboplatin and topotecan.
Even if a cure is usually accomplished with the application of current knowledge, there remain two problems. The first problem is the lethal progression of a few cases. The second is the problem of late effects, especially the difficult effect of second cancers in hereditary cases. Since these problems are also issues for other cancers, retinoblastoma assumes a special importance, with a seemingly better opportunity to identify secondary events in tumors. The rapid growth of retinoblastoma may facilitate recognition of rate-limiting steps in progression, especially with the availability of a good animal model. Since RB1 and other functionally related genes are somatically mutant in so many cancers, study of them should be widely relevant.
A New Role for Pediatric Oncologists?
Can pediatric oncologists participate directly in reducing the burden of cancer in adults? Is there a possibility of prevention? I have serious doubts that much could be done to prevent cancer in children, but intervention may be feasible in hereditary conditions that cause cancer in adults. Thus, both familial adenomatous polyposis and neurofibromatosis type 1 have incidences more than twice that of retinoblastoma. Successful intervention in childhood might possibly lead to delay or prevention of later malignancies. Any success would also be relevant to prevention of nonhereditary cancer, which would be especially important for colon cancer since most cases involve somatic mutations in the APC gene. Pediatric oncologists may find themselves in a new leadership role.
Citation: Knudson A (2005) Retinoblastoma: Teacher of cancer biology and medicine. PLoS Med 2(10): e349.
==== Refs
References
Dyer MA Rodriguez-Galindo C Wilson M Use of preclinical models to improve treatment of retinoblastoma PLoS Med 2005 2 e332 10.1371/journal.pmed.0020332 16231976
Eng C Li FP Abramson DH Ellsworth RM Wong FL Mortality from second tumors among long-term survivors of retinoblastoma J Natl Cancer Inst 1993 85 1121 1128 8320741
Byrne J Fears TR Whitney C Parry DM Survival after retinoblastoma: Long-term consequences and family history of cancer Med Pediatr Oncol 1995 24 160 165 7838037
|
16231985
|
PMC1261522
|
CC BY
|
2021-01-05 10:40:29
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e364
|
latin-1
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020364
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020376CorrectionInfectious DiseasesMicrobiologyOtherPharmacology/Drug DiscoveryEpidemiology/Public HealthHealth PolicyInfectious DiseasesMedicine in Developing CountriesDrugs and adverse drug reactionsPublic HealthInternational healthHealth PolicyMicrobiologyCorrection: A Breakthrough in R&D for Neglected Diseases: New Ways to Get the Drugs We Need CorrectionMoran Mary 10 2005 25 10 2005 2 10 e376Copyright: © 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.
A Breakthrough in R&D for Neglected Diseases: New Ways to Get the Drugs We Need
==== Body
In PLoS Medicine, volume 2, issue 9, DOI: 10.1371/journal.pmed.0020302
In the third paragraph of the section “Neglected-Disease R&D Activity,” the sentence “Nearly one-third of these projects are at the clinical trial stage, including seven drugs now in Phase III trials…” should read instead “Nearly one-third of these projects are at the clinical trial stage, including six drugs now in Phase III trials….”
Citation: (2005) Correction: A breakthrough in R&D for neglected diseases: New ways to get the drugs we need. PLoS Med 2(10): e376.
|
0
|
PMC1261526
|
CC BY
|
2021-01-05 10:39:16
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e376
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020376
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020389CorrectionEpidemiology/Public HealthGeriatricsHealth PolicyMental HealthStatisticsPsychiatryChronic Disease ManagementDementiaGeriatric MedicineHealth PolicyPublic HealthCorrection: The Incidence of Dementia in England and Wales: Findings from the Five Identical Sites of the MRC CFA Study CorrectionMatthews Fiona Brayne Carol 10 2005 25 10 2005 2 10 e389Copyright: © 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.
The Incidence of Dementia in England and Wales: Findings from the Five Identical Sites of the MRC CFA Study
==== Body
In PLoS Medicine, volume 2, issue 8, DOI: 10.1371/journal.pmed.0020193
Some of the data presented in the Abstract and Results were incorrect.
The third sentence of the Methods and Findings section in the Abstract should read as follows: “Incidence rates rise with age, particularly above the age of 75 y, from 6.7 (95% confidence interval, 3.8–12.4) per 1,000 person years at age 65–69 y to 68.5 (95% confidence interval, 52.5–88.1) per 1,000 person years at age 85 y and above.” The fifth sentence should read as follows: “Hence, it is estimated that approximately 163,000 new cases of dementia occur in England and Wales each year.”
In Results, under “Combined Incidence Analysis,” the third sentence should read as follows: “The population burden of these rates equates to approximately 163,000 new occurring dementia cases each year in England and Wales (95% confidence interval [CI] 96,000 to 272,000).”
Citation: (2005) Correction: The incidence of dementia in England and Wales: Findings from the five identical sites of the MRC CFA study. PLoS Med 2(10): e389.
|
0
|
PMC1261527
|
CC BY
|
2021-01-05 10:38:17
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e389
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020389
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020390CorrectionInfectious DiseasesInfectious DiseasesMalariaMedicine in Developing CountriesCorrection: Estimation of the Total Parasite Biomass in Acute Falciparum Malaria from Plasma PfHRP2 CorrectionDondorp Arjen M Desakorn Varunee Pongtavornpinyo Wirichada Sahassananda Duangjai Silamut Kamolrat Chotivanich Kesinee Newton Paul N Pitisuttithum Punnee Smithyman A. M White Nicholas J Day Nicholas P. J 10 2005 25 10 2005 2 10 e390Copyright: © 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.
Estimation of the Total Parasite Biomass in Acute Falciparum Malaria from Plasma PfHRP2
==== Body
In PLoS Medicine, volume 2, issue 8, DOI: 10.1371/journal.pmed.0020204
There are two data errors in the Results, under “Patient Characteristics”: of the patients with severe disease, 47 patients (28%) had cerebral malaria, and ten patients (6%) had an admission Hct below 20%.
Likewise, under “Estimated Total Parasite Biomass in Relation to Other Markers of Severity,” the text should indicate that 47 patients had cerebral malaria, and in Table 2, the number of patients with GCS < 11 should be 47.
Citation: (2005) Correction: Estimation of the total parasite biomass in acute falciparum malaria from plasma PfHRP2. PLoS Med 2(10): e390.
|
0
|
PMC1261528
|
CC BY
|
2021-01-05 10:38:18
|
no
|
PLoS Med. 2005 Oct 25; 2(10):e390
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020390
|
oa_comm
|
==== Front
Biomed Eng OnlineBioMedical Engineering OnLine1475-925XBioMed Central London 1475-925X-4-541617429010.1186/1475-925X-4-54ResearchRegional differences in APD restitution can initiate wavebreak and re-entry in cardiac tissue: A computational study Clayton Richard H [email protected] Peter [email protected] Department of Computer Science, University of Sheffield, Regent Court, 211 Portobello Street, Sheffield S1 4DP, UK2 Departments of Cardiology and Cardiothoracic Surgery, University College Hospital, 16-18 Westmoreland Street, London W1G 8PH, UK2005 20 9 2005 4 54 54 15 7 2005 20 9 2005 Copyright © 2005 Clayton and Taggart; licensee BioMed Central Ltd.2005Clayton and Taggart; 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
Regional differences in action potential duration (APD) restitution in the heart favour arrhythmias, but the mechanism is not well understood.
Methods
We simulated a 150 × 150 mm 2D sheet of cardiac ventricular tissue using a simplified computational model. We investigated wavebreak and re-entry initiated by an S1S2S3 stimulus protocol in tissue sheets with two regions, each with different APD restitution. The two regions had a different APD at short diastolic interval (DI), but similar APD at long DI. Simulations were performed twice; once with both regions having steep (slope > 1), and once with both regions having flat (slope < 1) APD restitution.
Results
Wavebreak and re-entry were readily initiated using the S1S2S3 protocol in tissue sheets with two regions having different APD restitution properties. Initiation occurred irrespective of whether the APD restitution slopes were steep or flat. With steep APD restitution, the range of S2S3 intervals resulting in wavebreak increased from 1 ms with S1S2 of 250 ms, to 75 ms (S1S2 180 ms). With flat APD restitution, the range of S2S3 intervals resulting in wavebreak increased from 1 ms (S1S2 250 ms), to 21 ms (S1S2 340 ms) and then 11 ms (S1S2 400 ms).
Conclusion
Regional differences in APD restitution are an arrhythmogenic substrate that can be concealed at normal heart rates. A premature stimulus produces regional differences in repolarisation, and a further premature stimulus can then result in wavebreak and initiate re-entry. This mechanism for initiating re-entry is independent of the steepness of the APD restitution curve.
==== Body
Background
Understanding the mechanisms that initiate and sustain malignant ventricular arrhythmias is an important research problem because ventricular tachycardia and fibrillation (VT and VF) are a notable cause of premature death, and remain an important public health problem in the industrialised world. Recent attention has focussed on restitution, the influence of an abrupt change in cycle length on action potential duration (APD). The dynamic behaviour of APD in response to cycle length changes has been shown in theoretical, experimental and computational studies to be a major determinant of wavefront stability [1,2]. An APD restitution curve with a slope > 1 can result in the initiation and subsequent instability of re-entrant arrhythmias [3], although other important mechanisms of initiation and instability have also been identified [4].
Regional differences in electrophysiological properties are a characteristic finding in the hearts of patients with cardiac pathology. Regional differences in repolarisation are often described as dispersion, and vulnerability to arrhythmias has been shown to depend on dispersion of repolarisation in both experimental [5,6] and computational studies [7-10]. This conceptual link between dispersion of repolarisation and vulnerability to re-entry may make a tacit assumption that APD dispersion is a static property of the tissue, resulting from underlying heterogeneity in electrophysiology. However, APD is a dynamic property of cardiac tissue, and is reduced at short diastolic interval (DI). The APD restitution curve is a model of the dynamical behaviour of cardiac tissue. One important consequence of this dynamic behaviour is that dispersion of repolarisation can be produced in electrophysiologically homogenous tissue if the DI is spatially non-uniform [11,12], resulting in alternans, wavebreak and re-entry.
Regional differences in both static and dynamic APD are a property of normal cardiac tissue. Regional differences in APD restitution have been documented, both at different locations within the ventricular wall [13-15], between left and right ventricle [16], and in chronically ischaemic hearts [17]. Experimental optical mapping of the ventricular epicardial surface has shown that regional differences in APD restitution can be exposed by a closely coupled premature stimulus [18], and that dispersion of repolarisation produced in this way increases vulnerability to VF [16,19,20].
Figure 1(a) illustrates a simplified caricature of cardiac tissue with two neighbouring regions, each with the APD restitution curve shown in Figure 1(b). In the first region (R1) the APD restitution curve is flat across a wide range of DI with a steep decline at short DI, whereas in the second region (R2) the APD restitution has a shallower slope. If we assume that the conduction velocity (CV) restitution of regions R1 and R2 is the same, then pacing the tissue at a long cycle length, and hence with a long DI, will elicit a similar APD in each region as indicated in Figure 1(b). A premature stimulus (S2) with a short DI will, however, elicit a different APD in R1 and R2, as indicated in Figure 1(b). An additional premature stimulus (S3) could then encounter a long APD in R1 and a short APD in R2. If the S2S3 interval is sufficiently short, then this stimulus could be blocked in R1, resulting in wavebreak at the R1R2 boundary. This is illustrated in Figure 1(c), which shows a cartoon of action potentials in R1 and R2. Wavebreak is a precursor to re-entry, and so a sequence of three stimuli (S1, S2, S3) delivered from the same site would initiate wavebreak and then re-entry in tissue with this type of regional difference in APD restitution.
Figure 1 Idealised arrangement of cardiac tissue with regional differences in APD restitution. (a) Two-dimensional tissue sheet, with regions R1 and R2 outlined in red and blue respectively, and stimulus electrode at the bottom. (b) Cartoon of APD restitution of regions R1 (red) and R2 (blue), showing APDs in each region resulting from stimulus at long and short DI. (c) Cartoon of action potentials elicited in R1 and R2 with S1 S2 S3 stimulus protocol.
The behaviour of cardiac tissue with regional differences in APD restitution has not been studied in detail. The novel aspect of this study was therefore to examine the idealised situation shown in Figure 1, and to investigate how wavebreak and re-entry can be produced in tissue with regional differences in APD restitution. A wider aim of this study was to assess whether vulnerability to arrhythmias could be predicted from knowledge of the APD restitution properties of the tissue in a particular heart. Computational models of cardiac tissue are becoming a valuable experimental tool for testing and proposing hypotheses because tight control over tissue geometry and electrophysiology enables some problems to be dissected into their component parts. We therefore used a simplified computational model of cardiac tissue in which the APD restitution of R1 and R2 could be varied in a controlled manner.
Methods
Computational model
We simulated action potential propagation in a 2D sheet of isotropic cardiac tissue with membrane voltage Vm described by the monodomain equation
where Cm is specific membrane capacitance, D a diffusion coefficient and Iion current flow though the cell membrane per unit area. A great many cell models have been developed that reproduce the action potential of cardiac cells from different species, and from different parts of the heart [21]. Biophysically detailed models are computationally demanding to solve, and so we elected to use a simplified model of the action potential. For this study the key feature of cardiac electrophysiology was the APD and CV restitution, and this is captured by the 3-variable model described by Fenton and Karma [4,22,23], which we used to describe Iion. Details of the model are given in Appendix 1. We used the four parameter sets given in Table 1[4,22] to give four variants of the model, each with different APD restitution. The first two variants (Steep1 and Steep2) had an APD restitution slope > 1 at short DI, and the second two (Flat1 and Flat2) had an APD restitution slope < 1 at short DI.
Table 1 Parameter values for each variant of the 3-variable model
Parameter Steep1 Steep2 Flat1 Flat2 units
V0 -85 -85 -85 -85 mV
Vfi 15 15 15 15 mV
gfi 4 4 4 4 mS cm-2
td 0.25 0.25 0.25 0.25 ms
tr 50 33 33 33 ms
tsi 45 30 30 30 ms
t0 8.3 12.5 12.5 12.5 ms
tv+ 3.33 3.33 3.33 3.33 ms
tv1- 1000 1250 1250 1250 ms
tv2- 19.2 19.6 19.6 19.6 ms
tw+ 667 870 870 870 ms
tw- 11 41 60 120 ms
uc 0.13 0.13 0.13 0.13 None
uv 0.055 0.04 0.04 0.04 None
ucsi 0.85 0.85 0.85 0.85 None
k 10 10 10 10 None
Numerical methods
We solved the FK equations using a simple explicit Euler scheme, and the nonlinear diffusion equation using a forward time centre space finite difference method with a time step (Δt) of 0.1 ms, a space step (Δ) of 0.25 mm and no-flux boundary conditions at each edge. The specific membrane capacitance was set to 1 μF cm-2, and the diffusion coefficient set to 0.1 mm2 ms-1.
Tissue geometry and stimulus protocol
We examined the initiation of re-entry in 150 × 150 mm 2D tissues with R1 and R2 allocated the steep APD restitution variants Steep1 and Steep2 respectively, and with square and circle configurations of R1. We then repeated the study with R1 and R2 allocated the flat restitution variants Flat1 and Flat2 respectively. For the square configuration, region R1 was defined for x <nx/2 and ny/4 <y <3ny/4, with region R2 elsewhere. This arrangement corresponded to the idealised geometry shown in Figure 1. In the circle configuration we sought to simulate a more physiologically plausible arrangement, and so we allocated R1 to a circular region located in the centre of the tissue, with a radius of 12.5, 25, or 50 mm.
Each tissue sheet was stimulated along its bottom edge (y = 0) by raising the bottom 0.5 mm of tissue above threshold for 2 ms with an S1 S2 S3 stimulation protocol. The initial conditions for the model were imposed as described above to correspond to the state of the model following a period of steady pacing at a long (≥ 500 ms) cycle length. S1 was delivered at the beginning of the simulation, and the S2 S3 interval was varied in steps of 1 ms.
APD and CV restitution
The APD and CV restitution curves for each variant of the model were measured from a thin strip of uniform tissue 75 mm long. Four S1 stimuli were given to one end of the tissue at 500 ms intervals, followed by a premature S2 stimulus. The S1S2 interval was reduced until a propagating action potential could not be elicited. APD was measured to 90% repolarisation; so all measurements of APD in this paper correspond to APD90. We measured DI and APD in the centre of the strip. We also measured the time difference between the action potential upstroke at the stimulus site and the action potential upstroke in the centre of the strip, and used this information to calculate CV restitution. This method underestimates CV slightly, due to the action of the stimulus with the tissue edge.
Results
APD and CV restitution
The APD and CV restitution for each of the four variants of the computational model measured in a thin strip of homogenous tissue are shown in Figure 2. There was little difference between the CV restitution curves. At long DI, the APD of each variant was around 140 ms, a value comparable with that found in guinea pig [24] or rabbit [16] ventricular tissue. At shorter DI, each pair of APD restitution curves diverged, showing that a markedly different APD would be elicited from each of the two variants.
Figure 2 APD and CV restitution measured in a thin strip of uniform tissue. Panels (a) and (b) show APD restitution for the steep and flat variants of the model, and in each case the green line indicates the difference in APD produced at each DI. Panels (c) and (d) show the CV restitution for the steep and flat variants of the model. Model variants with restitution shown in red were allocated to region R1, and those shown in blue allocated to region R2. S3 is blocked in R1, because the tissue is still repolarising from the S2 action potential.
Wavebreak and re-entry
In the tissue sheet model with heterogenous APD restitution, we were able to initiate wavebreak and re-entry over a range of S1 S2 S3 intervals when R1 and R2 had different APD restitution. This was the case when the slope of APD restitution in both R1 and R2 was steep, and also when the slope of APD restitution in both R1 and R2 was relatively flat. We were also able to initiate wavebreak and re-entry for both square and circular configurations of R1. Figure 3 shows an example of wavebreak and re-entry for the square configuration, where both R1 and R2 had steep APD restitution.
Figure 3 (a) Snapshots showing distribution of membrane potential in the 2D model 400, 500, 600, and 700 ms after delivery of S1 stimulus. S1 S2 interval was 200 ms, and S2 S3 interval 120 ms. The colour scheme shows resting tissue as black, and brighter colours show depolarised tissue. Dotted grey lines show boundary between R1 and R2 regions with different APD restitution, and grey arrows show direction of propagation. (b) Recordings of membrane potential from sites indicated by stars in (a). (c) APD restitution curves for R1 and R2, showing DIS1S2, and the APD produced in R1 and R2. See text for details.
Figure 3(a) shows four snapshots of the distribution of membrane voltage within the tissue at different times after the S3 stimulus. The action potentials initiated by each stimulus propagated from the bottom to the top of the virtual tissue; hence the first panel of Figure 3(a) shows action potentials elicited by each of the S1, S2, and S3 stimuli propagating from bottom to top as indicated by the arrows. The dotted grey line indicates the boundary between regions R1 and R2. The S2 action potential has a longer APD in R1 compared to R2, and this results in block of the S3 action potential close to the bottom edge of R1 and the development of wavebreak and then a re-entrant spiral wave, shown in subsequent panels of Figure 3(a).
The two stars in each panel of Figure 3(a) indicate points from which simulated transmembrane potentials were recorded from the model, and the time series of these are shown in Figure 3(b). The top trace shows the recording from R1 and the bottom trace the recording from R2. The APD restitution of each region is shown in Figure 3(c). The first (S1) action potential elicited APDs of 139 and 142 ms in R1 and R2 respectively, corresponding to the flat regions of the restitution curves in Figure 2. The second (S2) action potential had a DI (DIS1S2) of 58 and 61 ms in R1 and R2, and the resulting APDs were 120 and 61 ms respectively as indicated by the dashed lines in Figure 3(c). The third (S3) stimulus was delivered 120 ms after the second, and a propagating action potential was initiated in the lower half of the virtual tissue as shown in the first panel of Figure 3(a), with a DI (DIS2S3) of 58 ms. Figure 3(b) shows that this action potential arrived at the R1 R2 boundary before repolarisation was complete, and so was blocked. In R2 repolarisation was complete, and so the S3 action potential continued to propagate with a wavebreak at the boundary between R1 and R2.
Re-entry
In all of the simulations the wavebreak formed by block in R1 curled around the top edge of R1 and re-entered this region as it recovered, and so wavebreak always resulted in at least one cycle of re-entry. The re-entrant spiral waves tended to break up in tissue with steep APD restitution and tended to remain stable in tissue with flat APD restitution, although the simulations only extended for a few cycles of re-entry following initiation. This behaviour is illustrated in Figure 4, and also in the additional movie files.
Figure 4 Snapshots showing the behaviour of re-entry following wavebreak. The colour scheme is the same as in Figure 3, with brighter colours showing depolarised tissue. (a) Immediate break up of re-entry for model with steep APD restitution. (b) Initially stable re-entry in model with flat APD restitution. Movies of these simulations are included in additional files 1 and 2.
Size and shape of heterogeneity
The overall behaviour of virtual tissue with a circular region R1 was similar to that of virtual tissues with a rectangular region R1. For an S1 S2 interval of 200 ms, wavebreak and re-entry were initiated when the S2 S3 interval was greater than 101 ms. The upper limit of the S2 S3 interval that resulted in wavebreak was 150 ms for and R1 radius of 12.5 mm, 161 ms for radius 25 mm, and 158 ms for radius 50 mm. This upper limit compared with a value of 161 ms for the square configuration.
An example of the initiation of re-entry for R1 radii of 12.5 and 50 mm, and an S2 S3 interval of 120 ms is shown in Figure 5 for the configuration with steep APD restitution. As in Figure 3, the boundary between R1 and R2 is indicated by a grey dotted line. In the top panel, action potentials resulting from the S2 and S3 stimuli are shown, and the prolonged action potential in R1 can be seen. For an R1 radius of 12.5 mm (Figure 5(a)), the effect of electrotonic current flow during repolarisation resulted in a smaller region with longer APD. Although this region was large enough to initiate wavebreak and 1 cycle of re-entry, the two re-entrant waves were blocked during their second cycle, and re-entry terminated. For the larger scale heterogeneity (Figure 5(b)), re-entry persisted and broke up into multiple wavelets close to the boundary between R1 and R2. Additional files 1,2,3,4show movies of the simulations shown in Figure 5.
Figure 5 Example snapshots showing wavebreak and re-entry in simulated tissue with circular heterogeneity, and steep APD restitution. The colour scheme is the same as in Figure 3, with brighter colours showing depolarised tissue. The boundary of R1 is shown by the grey dotted line. Each panel shows consecutive snapshots of membrane voltage following stimuli with S1 S2 interval of 200 ms, and S2 S3 interval of 120 ms. (a) Wavebreak and sustained re-entry for R1 with radius of 50 mm. (b) Wavebreak and one cycle of re-entry for R1 with radius of 12.5 mm. Movies of these simulations are included in additional files 3 and 4.
From these findings we concluded that a circular R1 region behaves in a similar way to a square region, and that the size of the circle is small enough for electrotonic effects to become important. This latter observation is in agreement with other studies that have examined electrotonic effects with static differences in APD [10,25].
Stimulus protocol
For the square configuration of R1 and R2, and for a range of S1 S2 intervals, we measured the range of S2 S3 intervals that produced wavebreak and re-entry, and designated this the vulnerable area. Figure 6 shows the S2 S3 interval plotted against the S1 S2 interval for simulated tissue with both steep and flat APD restitution, with the vulnerable area that elicited wavebreak and re-entry shown in grey. The shape of each vulnerable area was different. For steep APD restitution, re-entry could be initiated over a range of S1 S2 intervals of 70 ms, and a range of S2 S3 intervals of up to 75 ms. In contrast, for flat APD restitution, re-entry could be initiated over a much longer range of S1 S2 intervals of 150 ms, but the range of S2 S3 intervals was much shorter, with a maximum of 21 ms. Hence, although re-entry could be initiated in tissue with both steep and flat APD restitution, the shape of the APD restitution curve was important for determining the dimensions of the vulnerable area.
Figure 6 Combinations of S1 S2 interval (CLS1S2) and S2 S3 interval (CLS2S3) resulting in wavebreak and re-entry in models with (a) steep APD restitution, and (b) flat APD restitution. Red points show the upper limit of S2 S3 interval that resulted in wavebreak, above this threshold the S3 action potential propagated in region R1 without wavebreak. Blue points show the lower limit of S2 S3 interval that resulted in wavebreak, below this threshold the S3 action potential was blocked in region R2 at the stimulus site. The vulnerable regions where re-entry and wavebreak were initiated are bounded by these upper and lower limits, and are shown shaded grey. The dashed line in (a) indicates the shortest CLS1S2 that produced a propagating action potential.
A clue about these differences can be gleaned from Figure 2. Figure 2a–b indicates the difference in APD between each pair of models. For steep APD restitution, large differences in APD were produced over a narrow range of DI, whereas for flat APD restitution, small differences in APD were produced over a wider range of DI. Hence for the model with steep APD restitution, regional differences in APD could be produced only over a narrow range of S1S2 interval. In contrast, the models with shallow APD restitution produced smaller APD differences, but over a longer range of S1 S2 interval.
Figures 2 and 6 therefore highlight the key mechanism by which wavebreak and re-entry were produced by a two-stage mechanism in tissue with regional differences in APD restitution. First, a premature beat produced regional differences in repolarisation. Second, a further premature beat was partially blocked in the regions of prolonged repolarisation. The differences between the APD restitution in each region, and not their slope, determined the dimensions of the vulnerable area.
Analysis
The broader aim of this study was to assess whether knowledge of restitution properties could be used to predict the vulnerability to re-entry shown in Figure 6. APD and CV restitution are complex properties of cardiac tissue, and the APD and CV of a particular beat are influenced by the pacing history or cardiac memory [26,27], by electrotonic current within the tissue [18], and APD may not always be a monotonic function of DI [27]. For the sake of simplicity in the analysis below, this complexity is noted but not included. In this simplified case, the APD of beat n+1 depends only on the preceding DI according to the iterative relation below.
APDn+1 = f(DIn)
DIn = CLn - APDn (2)
Where f(DI) is the APD restitution curve that gives APD as a function of DI. For the S1 S2 S3 protocol used in this study there are two cycle lengths, CLS1S2, and CLS2S3 with corresponding diastolic intervals
DIS1S2 = CLS1S2 – APDS1
DIS2S3 = CLS2S3 – f (CLS1S2 – APDS1 ) (3)
The second diastolic interval DIS2S3 depends on APDS2, which can be calculated from the APD restitution curve. The third (S3) beat will be blocked if DIS2S3 is less than DImin, where DImin is the shortest DI that results in a propagating beat. If we consider tissue with regions R1 and R2 as shown in Figure 1, and ignore the effects of CV restitution, then the S3 action potential will be blocked in region R1 if
and in region R2 if
For re-entry to be initiated, we require block in region R1, and propagation in region R2. This condition produces a wavebreak, and is fulfilled if
Based on this analysis, if the APD restitution in regions R1 and R2 is identical, then re-entry cannot be initiated with this stimulus protocol. This is reflected in equation (6), where the range of CLS2S3 that result in re-entry is zero if both sides of the inequality are equal. Another consequence of this analysis is that wavebreak will only occur if the left hand side of equation 6 is less than the right hand side, so APD restitution must produce a longer APD in R1 than in R2 for short DI. Our final observation from this equation is that there is no requirement that the slope of any of the restitution curves should be >1, and this is supported by our findings shown in Figure 6.
Figure 7(a) and Figure 7(c) show the upper and lower bounds of the vulnerable region predicted from equation 6, as well as the measurements from the model, for both steep and flat APD restitution. Although the predicted lower bounds (blue lines) agreed well with the observations, the initial prediction of the upper bound (dashed red line) overestimated the range of CLS2S3 resulting in wavebreak, especially for longer CLS1S2.
Figure 7 Predicted upper and lower bounds for initiating wavebreak in models. (a) Predicted upper (red) and lower (blue) bounds for model with steep APD restitution. The modified upper bound (solid red line) used information from the delays shown in (b). (b) Conduction delays measured in heterogenous tissue with two regions of steep APD restitution. Delays are delayS2S3, for S3 action potentials at CLS2S3 close to those that result in block in R1, and each line shows this delay for a different S1 S2 interval. Region R2 is to the left of the dotted grey line, and region R1 to the right. See text for details. (c) Predicted upper (red) and lower (blue) bounds for model with flat APD restitution. The modified upper bound (solid red line) used information from the delays shown in (d). (d) Conduction delays measured in heterogeneous tissue with flat restitution. See text for details.
Although the CV restitution of the models only exerts an effect at short DI (Figure 2), we hypothesised that conduction delays associated with CV restitution accounted for the difference between the predicted and observed upper bound, and we modified equation 4 accordingly.
The delays in this equation result from CV restitution, and depend on the distance between the stimulus site and the border of the heterogeneity. We measured these time delays for combinations of CLS1S2 and CLS2S3 that resulted in propagation of the S3 action potential, and hence corresponded to the upper bound of vulnerability. The first delay delayS1S2 was small except for values of CLS1S2 close to the lower limit, by the second delay delayS2S3, and these delays are plotted for steep and flat restitution in Figure 7(b) and Figure 7(d) respectively. These plots show evidence of very slow conduction in the border zone between R1 and R2. For a shorter CLS2S3, the S3 action potential was blocked in R2 close to the point of slowest conduction, at around 47.5 mm, indicated by a black arrow on Figures 7(b) and 7(d). We fitted a polynomial to the measurements of delayS2S3 at this distance to estimate delayS2S3 for the upper bound of vulnerability as a function of CLS1S2. This estimate, together with measurements of delayS1S2 enabled us to plot equation 7, and this is shown as a solid red line on Figures 7(a) and 7(c). This modified prediction of the upper bound provides a much better fit to the measurements. It is likely that electrotonic interaction, and errors in the polynomial fit could account for the remaining differences.
Thus our analysis of these results shows that information about both APD and CV restitution are necessary to predict the vulnerability of tissue with heterogenous APD restitution. The effect of CV restitution is to introduce delays that reduce vulnerability, and these delays depend on the distance between the stimulus site and the border of the heterogeneity. This dependence was investigated for a model of tissue with a static APD heterogeneity by Panfilov and Vasiev, who showed that the width of the vulnerable period decreases monotonically with increasing distance between stimulus site and border of the heterogeneity [9].
Discussion
The novel finding of this study is that regional differences in APD restitution can act as a potent arrhythmogenic substrate by producing rate dependent regional differences in repolarisation. These regional differences may be concealed at normal heart rates, but exposed by a premature beat. A further premature beat can then be partially blocked by regions of refractory tissue, resulting in wavebreak and re-entry. This finding is important because the exact mechanism by which re-entry and VF are initiated in the human heart remains unclear. As a consequence, it is difficult to identify with precision those patients who are at risk of sudden cardiac death.
The arrhythmogenic effects of regional differences in repolarisation have been studied both experimentally [5,6] and with an early computational model [7]. Regions with longer refractory periods can block a premature stimulus, resulting in re-entry, and a recent computational study has examined this mechanism in detail [10]. In this study, regional differences in repolarisation were presumed to be a static feature resulting from regional pathology. Winfree [28] showed that the interaction of a premature beat with repolarising tissue was another mechanism capable of initiating re-entry, and this idea was later verified experimentally [29]. This approach was significant because it explained how re-entry could be initiated in normal, electrically uniform tissue. Another mechanism capable of producing re-entry in uniform tissue depends on APD and CV restitution. Steep restitution can produce spatial and/or temporal APD alternans leading to wavebreak and unstable re-entry [12,30].
VF and other re-entrant arrhythmias are, however, more prevalent in hearts that are affected by disease processes that augment electrical heterogeneity. For example, regional ischaemia is an effective arrhythmogenic substrate and a recent experimental study has shown that it produces regional differences in APD restitution [17]. Our present study builds on this and other experimental work [18,19].
Mechanism
The key arrhythmogenic mechanism we have investigated is exposure of regional differences in repolarisation by a closely coupled stimulus, followed by regional block of a further closely coupled premature beat. There is no general requirement that for the slopes of the APD restitution curves to be steep, and we have shown that this mechanism produces wavebreak for both steep (slope>1) and shallow (slope < 1) APD restitution. However, the relative steepness of the APD restitution curve in each region is important, as described below.
Importance of stimulus sites
For the sake of clarity all three stimuli were delivered from the same site in our simulations. In this scheme (Figure 1) it was important that the R1 region(s) distal to the stimulus site should have steeper APD restitution than the surrounding R2 tissue, so that the S3 beat was not blocked close to the stimulus site. Other schemes are of course possible, and in real tissue the normal and premature beats may originate from two or more different sites. This would modify the detail of the mechanism we have described, but the core idea would remain the same.
For example, if the R1 region had shallower APD restitution than the R2 region, then delivery of the S1 and S2 stimuli from within R2 would result in a long APD in R2 but a short APD in R1. Thus if the S3 beat originated in R2, it would either propagate normally with a short APD in R1, or be blocked in R2. However, if the S3 beat was to originate in R1, it could be blocked by the longer repolarisation in R2 while propagating in R1, forming a wavebreak and initiating re-entry. More simulations would clarify this, but are beyond the scope of the current study.
Geometry
The simulations in the present study have considered only two regions, R1 and R2, but the overall mechanism is also applicable to tissue with multiple regions with regional differences in APD restitution. Our findings also indicate that the shape of the different regions has a small effect on the initiation of wavebreak and re-entry, but our results for the circular heterogeneity indicate that the size of the regions is important. In a previous computational study we have shown that both size and cell-to-cell coupling determine the potency of static arrhythmogenic heterogeneities, presumably through electrotonic current flow within the tissue [10]. Figure 5 suggests that electrotonic current flow is also important for dynamically induced heterogeneity, and smaller regions produce transient wavebreak but do not support sustained re-entry.
Effect of CV restitution
This paper has focussed on APD restitution rather than CV restitution. The four variants of the cell membrane model used in this study possessed almost identical CV restitution (Figure 2), with significant conduction delays only becoming evident at short DI. However, the analysis given above and detailed in Figure 7 indicates that both APD and CV restitution are important in determining whether wavebreak and re-entry will occur for a given stimulus sequence. This effect is greater for longer S1S2 intervals since APDS2 is longer in both R1 and R2, DIS1S2 is shorter in R1 and R2, and hence the S3 beat is more delayed. A greater delay to the S3 beat at the boundary between R1 and R2 offers more time for the tissue in R1 to repolarise, and reduces the incidence of block. This delay therefore underlies the reduction in the vulnerable region, and explains the overestimation of the vulnerable region by equation 6.
Regional differences in CV restitution as well as APD restitution would add an additional layer of complexity to the behaviours documented here, but this detailed analysis is outside the scope of this discussion.
Predicting vulnerability
A quantitative assessment of vulnerability resulting from regional differences in APD restitution would be a valuable clinical tool. These differences would be exposed as regional differences in repolarisation when the heart is paced at short cycle lengths. A recent computational study has shown that regional differences in repolarisation produce distinctive changes in T wave shape on the electrocardiogram [31]. We would expect that regional differences in APD restitution would produce characteristic rate-dependent changes in T wave shape.
Our preliminary efforts described above and shown in Figure 7 however, suggest that although it may be possible to identify patients at risk using this type of approach, it may be difficult to estimate the extent of vulnerability. The main reason for this is the effect of conduction delays arising from CV restitution. Although the CV restitution curve for a region of tissue may be well characterised, the delays affecting a particular premature beat depend on the spatial relationship between R1, R2, and the stimulus site(s). The effect of CV restitution delays is to prolong the DI between two closely coupled beats, and the extent of the delay depends on the distance between the stimulus site and the recording site [9]. Hence without detailed knowledge of the spatial relationship between regions with altered APD restitution and the site of origin of premature beats, it may be difficult to predict the size of the vulnerable region.
Limitations
This study has several limitations. We used a greatly simplified computational model to represent the dynamical behaviour of tissue, which does not describe the details of current flow through ion channels, pumps, and exchangers in the cell membrane, and it does not attempt to include the effects of intracellular Ca2+ storage and release. Neither does it include the effects of cardiac memory. However, the model does capture the APD and CV restitution of real cells and tissue, and it is these dynamical features that are relevant for the mechanism that was explored in this study. For the sake of simplicity and clarity, we also simulated isotropic 2D sheets with abrupt changes in APD restitution between different regions, yet real ventricular tissue is both anisotropic and 3-dimensional, and spatial changes in APD restitution are likely to be gradual. We only investigated tissues with a limited range of restitution curves. All of these limitations arose from a desire to minimise the computational demands of the study.
In this study we assumed that the APD restitution of ventricular tissue could be described by a monotonic curve where APD depends solely on the preceding DI. However, APD restitution curves recorded from human hearts may have a more complex shape [27]. In addition, there is substantial evidence to suggest that APD restitution may itself be dynamic, and depend on the stimulus history and not simply on the preceding DI [26,32].
Further work with biophysically detailed models of the cardiac cell membrane, a wider range of restitution characteristics, and models of anisotropic 3D tissue will be needed to establish fully the extent to which the findings presented here could be relevant in real cardiac tissue.
Clinical implications
This study has some important clinical implications. In the human heart, APD restitution is flattened in ischaemia [33], and steepened by adrenergic agents [34], suggesting that inhomogeneous sympathetic innervation as a result of nerve sprouting may generate heterogenous APD restitution. Cardiac drugs can also decrease APD restitution slope [2,35]. Evidence from isolated myocytes and tissue preparations also suggest that there are transmural differences in APD restitution [36], and these could act together with the mechanisms described above. These effects could in turn be further modified by additional factors including regional stretch, hypertrophy, and regional remodelling. Detailed experimental and clinical studies are now needed to establish precisely the relative importance of these factors for the arrhythmogenic substrate.
Conclusion
This study used a simplified computational model of cardiac tissue to test the idea that regional differences in APD restitution can be a potent substrate for initiating re-entrant arrhythmias. These regional differences can be concealed at normal heart rates. A two-stage process can produce wavebreak and re-entry. First, regional differences in repolarisation can be produced by a premature beat, and second, these regional differences can then interact with a further premature beat, resulting in wavebreak and re-entry through the well-established mechanism of conduction block. Since the determinant of wavebreak is independent of APD restitution slope, we found that re-entry could be produced in simulated tissue with both steep (slope > 1) and flat (slope < 1) APD restitution.
Authors' contributions
RHC designed the study, wrote the simulation code, ran the simulations, and wrote the manuscript. PT conceived the study, helped interpret the results, and also contributed significantly to the manuscript.
Appendix: Three-variable model
The 3-variable Fenton Karma model has three currents, two inward (depolarising) currents corresponding broadly to Na+ and Ca2+ currents, and a slow outward (repolarising) current corresponding to K+ currents. The membrane voltage Vm was scaled with the resting potential V0 and the Nernst potential of the fast inward current Vfi to give a dimensionless activation variable u that varies between 0 and 1 where
Setting Cm to 1 μF mm-2, the equations of the model were
The currents Jfi, Jsi and Jso had units of ms-1 and were given by
Where Θ denotes the Heaviside step function and Θ(x) is equal to 1 for x ≥ 0 and 0 for x <0.
Supplementary Material
Additional File 1
Figure4a.mpg Wavebreak and re-entry in the model with steep APD restitution. The S1S2 interval was 200 ms, and the S2S3 interval was 120 ms. Movie of simulation depicted in Figure 4a.
Click here for file
Additional File 2
Figure4b.mpg Wavebreak and re-entry in the model with flat APD restitution. The S1S2 interval was 350 ms, and the S2S3 interval was 130 ms. Movie of simulation depicted in Figure 4b.
Click here for file
Additional File 3
Figure5a.mpg Wavebreak and re-entry in the model with steep APD restitution, and circular R1 with radius 50 mm. The S1S2 interval was 200 ms, and the S2S3 interval was 120 ms. Movie of simulation depicted in Figure 5a.
Click here for file
Additional File 4
Figure4a.mpg Wavebreak and re-entry in the model with steep APD restitution, and circular R1 with radius 12.5 mm. The S1S2 interval was 200 ms, and the S2S3 interval was 120 ms. Movie of simulation depicted in Figure 5b.
Click here for file
Acknowledgements
This work was supported by the British Heart Foundation through the award of project grant PG/03/102/15852 to RHC. Computations were done on machines in the Computational Biology Laboratory at the University of Leeds, and we are grateful to Professor Arun Holden for making this facility available. We would also like to thank Flavio Fenton for helpful discussions.
==== Refs
Karma A Spiral breakup in model equations of action potential propagation in cardiac tissue Physical Review Letters 1993 71 1103 1107 10055449 10.1103/PhysRevLett.71.1103
Garfinkel A Kim YH Voroshilovsky O Qu ZL Kil JR Lee MH Karagueuzian HS Weiss JN Chen PS Preventing ventricular fibrillation by flattening cardiac restitution Proc Natl Acad Sci U S A 2000 97 6061 6066 10811880 10.1073/pnas.090492697
Xie F Qu ZL Yang J Baher A Weiss JN Garfinkel A A simulation study of the effects of cardiac anatomy in ventricular fibrillation Journal of Clinical Investigation 2004 113 686 693 14991066 10.1172/JCI200417341
Fenton FH Cherry EM Hastings HM Evans SJ Multiple mechanisms of spiral wave breakup in a model of cardiac electrical activity Chaos 2002 12 852 892 12779613 10.1063/1.1504242
Han J Moe GK Nonuniform recovery of excitability in ventricular muscle Circulation Research 1964 14 44 60 14104163
Kuo CS Munkata K Reddy P Surawicz B Characteristics and possible mechanism of ventricular arrhythmia dependent on the dispersion of action potential durations Circulation 1983 67 1356 1367 6851031
Moe GK Rheinboldt WC Abildskov JA A computer model of atrial fibrillation American Heart Journal 1964 67 200 220 14118488 10.1016/0002-8703(64)90371-0
Sampson KJ Henriquez CS Simulation and prediction of functional block in the presence of structural and functional ionic heterogeneity American Journal of Physiology (Heart and Circulatory Physiology) 2001 281 H2597 H2603 11709428
Panfilov A Vasiev BN Vortex initiation in a heterogeneous excitable medium Physica D 1991 49 107 113 10.1016/0167-2789(91)90200-S
Clayton RH Holden AV Dispersion of cardiac action potential duration and the initiation of re-entry: A computational study Biomedical Engineering OnLine 2005 4 11 15720712 10.1186/1475-925X-4-11
Watanabe MA Fenton FH Evans SJ Hastings HM Karma A Mechanism for discordant alternans Journal of Cardiovascular Electrophysiology 2001 12 196 206 11232619 10.1046/j.1540-8167.2001.00196.x
Qu ZL Garfinkel A Chen PS Weiss JN Mechanisms of discordant alternans and induction of reentry in simulated cardiac tissue Circulation 2000 102 1664 1670 11015345
Gima K Rudy Y Ionic current basis of electrocardiographic waveforms. A model study Circulation Research 2002 90 889 896 11988490 10.1161/01.RES.0000016960.61087.86
Antzelevitch C Nesterenko VV Muzikant AL Influence of transmural repolarization gradients on the electrophysiology and pharmacology of ventricular myocardium. Cellular basis for the Brugada and Long QT syndromes. Philosophical Transactions of the Royal Society of London A 2001 359 1201 1216 10.1098/rsta.2001.0826
Viswanathan PC Shaw RM Rudy Y Effects of I-Kr and I-Ks heterogeneity on action potential duration and its rate dependence - A simulation study Circulation 1999 99 2466 2474 10318671
Banville I Gray RA Effect of action potential duration and conduction velocity restitution on alternans and the stability of arrhythmias Journal of Cardiovascular Electrophysiology 2002 13 1141 1149 12475106 10.1046/j.1540-8167.2002.01141.x
Yuuki K Hosoya Y Kubota I Yamaki M Dynamic and not static change in ventricular repolarisation is a substrate of ventricular ischaemia on chronic ischaemic myocardium Cardiovascular Research 2004 63 645 652 15306220 10.1016/j.cardiores.2004.04.017
Laurita KR Rosenbaum DS Modulation of ventricular repolarization by a premature stimulus. Role of epicardial dispersion of repolarization kinetics demonstrated by optical mapping of the intact guinea pig heart Circulation Research 1996 79 493 503 8781482
Laurita KR Girouard SD Akar FG Rosenbaum DS Modulated dispersion explains changes in arrhythmia vulnerability during premature stimulation of the heart Circulation 1998 98 9851966
Pak HN Hong SJ Hwang GS Lee HS Park SW Ahn JG Ro YM Kim YH Spatial dispersion of action potential duration restitution kinetics is associated with induction of ventricular tachycardia/fibrillation in humans Journal of Cardiovascular Electrophysiology 2004 15 1357 1363 15610278 10.1046/j.1540-8167.2004.03569.x
Noble D Rudy Y Models of cardiac ventricular action potentials: iterative interaction between experiment and simulation Philos Trans R Soc Lond Ser A-Math Phys Eng Sci 2001 359 1127 1142 10.1098/rsta.2001.0820
Fenton F Karma A Vortex dynamics in three-dimensional continuous myocardium with fibre rotation: Filament instability and fibrillation Chaos 1998 8 20 47 12779708 10.1063/1.166311
Clayton RH Holden AV A method to quantify the dynamics and complexity of re-entry in computational models of ventricular fibrillation Physics in Medicine and Biology 2002 47 225 238 11837614 10.1088/0031-9155/47/2/304
Girouard SD Pastore JM Laurita KR Gregory KW Rosenbaum DS Optical mapping in a new guinea pig model of ventricular tachycardia reveals mechanisms for multiple wavelengths in a single reentrant circuit Circulation 1996 93 603 613 8565181
Sampson KJ Henriquez CS Electrotonic influences on action potential duration dispersion in small hearts: a simulation study American Journal of Physiology (Heart and Circulatory Physiology) 2005 289 H350 H360 15734889 10.1152/ajpheart.00507.2004
Hund TJ Rudy Y Determinants of excitability in cardiac myocytes: Mechanistic investigation of memory effect Biophys J 2000 79 3095 3104 11106615
Franz MR The electrical restitution curve revisited: Steep or flat slope - which is better? Journal of Cardiovascular Electrophysiology 2003 14 S140 S147 14760916 10.1046/j.1540.8167.90303.x
Winfree AT Sudden cardiac death: A problem in topology Scientific American 1983 248 144 161 6857229
Frazier DW Wolf PD Wharton JM Tang ASL Smith WM Ideker RE Stimulus induced critical point. Mechanism for electrical initiation of reentry in normal canine myocardium. Journal of Clinical Investigation 1989 83 1039 1052 2921316
Pastore JM Girouard SD Laurita KR Akar FG Rosenbaum DS Mechanism linking T wave alternans to the genesis of cardiac fibrillation Circulation 1999 99 1385 1394 10077525
van Huysduynen BH Swenne CA Draisma HHM Antoni ML van de Vooren H van der Wall EE Schalij MJ Validation of ECG indices of ventricular repolarization heterogeneity: A computer simulation study J Cardiovasc Electrophysiol 2005 16 1097 103 16191120 10.1111/j.1540-8167.2005.40758.x
Cherry EM Fenton FH Suppression of alternans and conduction blocks despite steep APD restitution: Electrotonic, memory, and conduction velocity restitution effects American Journal of Physiology (Heart and Circulatory Physiology) 2004 286 H2332 H2341 14751863 10.1152/ajpheart.00747.2003
Taggart P Sutton PMI Boyett MR Lab M Swanton H Human ventricular action potential during short and long cycles. Rapid modulation by ischaemia. Circulation 1996 94 2526 2534 8921797
Taggart P Sutton PMI Chalabi Z Boyett MR Simon R Elliot D Gill GS Effect of adrenergic stimulation on action potential duration restitution in humans Circulation 2003 107 285 289 12538429 10.1161/01.CIR.0000044941.13346.74
Riccio ML Koller ML Gilmour RF Electrical restitution and spatiotemporal organization during ventricular fibrillation Circulation Research 1999 84 955 963 10222343
Yan GX Shimuzu W Antzelevitch C Characteristics and distribution of M cells in arterially perfused canine left ventricular wedge preparations Circulation 1998 98 1921 1927 9799214
|
16174290
|
PMC1261529
|
CC BY
|
2021-01-04 16:37:35
|
no
|
Biomed Eng Online. 2005 Sep 20; 4:54
|
utf-8
|
Biomed Eng Online
| 2,005 |
10.1186/1475-925X-4-54
|
oa_comm
|
==== Front
Harm Reduct JHarm Reduction Journal1477-7517BioMed Central London 1477-7517-2-171617909010.1186/1477-7517-2-17ReviewHarm reduction-the cannabis paradox Melamede Robert [email protected] Biology Department, 1420 Austin Bluffs Parkway, University of Colorado, Colorado Springs, 80918, USA2 Bioenergetics Institute, 1420 Austin Bluffs Parkway, University of Colorado, Colorado Springs, 80918, USA2005 22 9 2005 2 17 17 19 11 2004 22 9 2005 Copyright © 2005 Melamede; licensee BioMed Central Ltd.2005Melamede; 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.
This article examines harm reduction from a novel perspective. Its central thesis is that harm reduction is not only a social concept, but also a biological one. More specifically, evolution does not make moral distinctions in the selection process, but utilizes a cannabis-based approach to harm reduction in order to promote survival of the fittest. Evidence will be provided from peer-reviewed scientific literature that supports the hypothesis that humans, and all animals, make and use internally produced cannabis-like products (endocannabinoids) as part of the evolutionary harm reduction program. More specifically, endocannabinoids homeostatically regulate all body systems (cardiovascular, digestive, endocrine, excretory, immune, nervous, musculo-skeletal, reproductive). Therefore, the health of each individual is dependant on this system working appropriately.
==== Body
Introduction
The concept of harm reduction is at the heart of conflicting international drug policies. The Dutch pioneered this approach. Today most European countries and Canada have embraced the idea that society benefits most when drug policy is designed to help people with drug problems to live better lives rather than to punish them. In contrast, the United States federal policy demands rigid zero tolerance with overwhelming emphasis on incarceration of offenders (the Drug War). Although, seemingly reasonable arguments can be made to support both sides of the dispute, the recent global trend towards harm reduction has resulted from the acknowledgement that drug use has been a part of all societies throughout history and the realization that repressive policies are expensive, ineffective, and often harmful.
A dramatic example of the benefits that can result from a harm reduction approach to drugs is seen with needle exchange programs. While prohibitionists argue that providing clean injection equipment promotes drug use, the facts do not support this contention. For example, the Australian needle exchange program is credited with keeping the HIV/AIDS infection rate very much lower than what is typically found globally . Commonly cited examples of the failed repressive policies championed by the United States are the now repealed alcohol prohibition and the current drug war. Crime, financial support for terrorism, disrespect for the law, and destruction of families, communities, and ecosystems can all be attributed to drug prohibition. Yet, the staggering cost of the drug war, driven by United States policy and taxpayers' money, amounts to many billions of dollars a year.
Cannabis is the third most commonly used drug in the world, following tobacco and alcohol. In the United States, much of the drug war is focused on marijuana (over 700,000 people arrested last year alone). Is there justification for this policy? The gateway theory states marijuana use leads to the use of other drugs, and drives the U.S. policy despite evidence that suggests alcohol and tobacco use may foster the gateway effect [1,2]. In contrast, countries that support harm reduction focus their enforcement and social support efforts on "hard drugs." Consequently, many countries have effectively decriminalized marijuana. Holland, having the most liberalized drug laws, does not have more cannabis users (over age twelve) than do more repressive countries, and the per capita number of heroin users is also lower . The Dutch Ministry of Justice estimates that 0.16% of cannabis users are heroin users. This figure does not support cannabis being a gateway drug. Data from the 2000 National Household Survey on Drug Abuse (U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration) also shows that the vast majority of people who try cannabis do not go on to use hard drugs.
A little explored question is what does harm reduction specifically mean with respect to cannabis consumption? This article will address cannabis harm reduction from a biological perspective. Two directions will be examined: what are the biological effects of cannabis use and what are the social effects that emerge from the biological foundation.
Like many substances that are put into the human body, there can be positive or negative consequences that result from cannabis consumption, depending on amount, frequency, quality, and probably most importantly, the idiosyncratic biochemistry of the user. Prohibitionists concentrate their efforts on the negative effects of cannabis use, while anti-prohibitionists tend to focus on the positive effects. If we assume that both sides have valid arguments, the issue to be resolved is one of balance between the negative and positive effects. Would a policy of tolerance, or prohibition, be more likely to reduce harm overall? Which policy would better serve society as a whole, as well as problematic drug users?
Biological science can be more objectively evaluated than social science. The central theme that will be presented in this article is that appropriate cannabis use reduces biological harm caused by biochemical imbalances, particularly those that increase in frequency with age. Proper cannabis use, as distinguished from misuse, may have significant positive health effects associated with the way cannabis mimics natural cannabinoids. In essence, it is proposed that the endocannabinoid system, selected by 600 million years of evolution, is a central mediator of biological harm reduction through its homeostatic activities. The social implications of cannabis use will be viewed as emerging from the biological platform. Herein lies the paradox of cannabis and harm reduction. Is appropriate use of cannabis better than no use?
The Controversy
Cannabis use can be divided into three categories, recreational, medical, and religious. The latter will not be examined in this article. Some, including those who favor or oppose cannabis use, presume recreational and medical use are the same. On the one side, it is often claimed that any cannabis use is justified by some underlying medical need. On the other side, cannabis use is presumed to have no medical value, with the implication that those who use it are simply "getting stoned." While the former claim may be too extreme, the latter defies current scientific understanding of the biological functions of the endocannabinoids. While many people are reluctant to approve recreational cannabis use, it appears that most people support medical use. The United States Federal Government denies that there is any valid medical use for cannabis, while the National Institute of Drug Abuse (NIDA) provides marijuana on a monthly basis to a few medical users through the compassionate Investigatory New Drug (IND) program of the Food and Drug Administration (FDA). Nevertheless, a number of states, through either legislative action or voter initiative, have approved the use of medical marijuana[3].
Current Federally Approved Medical Marijuana Uses
In order to better assess arguments for and against the medical use of marijuana, the scientific evidence for the health benefits of cannabis will be reviewed below. It should be noted that the federally supplied cannabis users have been receiving and using cannabis for 11 to 27 years with clinically demonstrated effectiveness in the treatment of glaucoma, chronic musculoskeletal pain, spasm and nausea, and spasticity of multiple sclerosis [4]. Furthermore, there is no evidence that these patients have suffered any negative side effects from their cannabis use.
The Endocannabinoid System
Cannabis preparations have been used medically for thousands of years for illnesses such as epilepsy, migraine headaches, childbirth, and menstrual symptoms. However, it is only relatively recently that the active components have been identified and their mechanisms of action have begun to be understood. While delta-9-tetrahydrocannabinol (THC) was first synthesized by Mechoulam in 1967 [5], it was not until 1990 that the cannabinoid receptor was localized in the brain [6] and cloned [7]. Since then, discoveries in the field have proceeded at an ever-increasing pace. The discovery of cannabinoid receptors on cells naturally prompted the search for internal compounds (endogenous ligands) that would activate the receptors since it seemed unlikely that cannabis receptors had evolved so people could partake of cannabis. In 1992, anandamide was discovered [8]. This lipid metabolite was the first ligand of an ever-expanding class of molecules known as endocannabinoids (internal marijuana-like compounds) to be discovered. Endocannabinoid synthesis, degradation, transport, and receptors together form the endocannabinoid system.
The broad therapeutic potential that can result from correctly manipulating the endocannabinoid system is just beginning to be realized[9,10]. In fact, major pharmaceutical companies, and university researchers all around the world are now engaged in the cannabinoid-related research [11]. Their efforts focus on learning how the endocannabinoid system functions, and on how to manipulate it in order to increase or decrease its activity, depending on the illness or condition under consideration. GW Pharmaceuticals in Britain has been developing and testing a plant extract-based product line that is in clinical trials in Britain and Canada [12]. The results thus far have been positive to the extent that Bayer AG has entered into a 25-million-dollar distribution agreement for GW's product, Sativex which has recently been approved in Canada. In contrast, Sanofi Research has developed an antagonist that will inhibit the ability of endocannabinoids to stimulate hunger and thus potentially be useful for weight control.
Evolution of Endocannabinoids
The cannabinoid system appears to be quite ancient [13,14], with some of its components dating back about 600 million years to when the first multicellular organisms appeared. The beginnings of the modern cannabinoid system are found in mollusks [15] and hydra [16]. As evolution proceeded, the role that the cannabinoid system played in animal life continuously increased. It is now known that this system maintains homeostasis within and across the organizational scales of all animals. Within a cell, cannabinoids control basic metabolic processes such as glucose metabolism [17]. Cannabinoids regulate intercellular communication, especially in the immune [18] and nervous systems [19]. In general, cannabinoids modulate and coordinate tissues, organ and body systems (including the cardiovascular [20], digestive [16], endocrine [21], excretory [22,23], immune [18], musculo-skeletal [24], nervous [19], reproductive [25], and respiratory [26] systems). The effects of cannabinoids on consciousness are not well understood, but are well known, and underlie recreational cannabis use. These effects also have therapeutic possibilities [27].
Cannabinoids: Homeostatic Regulators
The homeostatic action of cannabinoids on so many physiological structures and processes is the basis for the hypothesis that the endocannabinoid system is nothing less than a naturally evolved harm reduction system. Endocannabinoids protect by fine-tuning and regulating dynamic biochemical steady states within the ranges required for healthy biological function. The endocannabinoid system itself appears to be up- or down-regulated as a function of need. As will be detailed later in this article, endocannabinoid levels naturally increase in the case of head injury and stroke [28], and the number of cannabinoid receptors increases in response to nerve injury and the associated pain [29]. In contrast, the number of cannabinoid receptors is reduced when tolerance to cannabinoids is induced [30].
Physical Characteristics of Living Systems
To illustrate the multidimensional biochemical balancing act performed by cannabinoids, a variety of endo- and exocannabinoid activities will be reviewed below. In order to appreciate these activities a brief introduction to cell biology may provide the context for this review. All life is dependant upon the maintenance of its dynamic organization through sufficient input of nutrients and removal of wastes. The more complicated an organism is, the more complex the coordination required to accomplish the essential tasks necessary to maintain this vital flow of inputs and outputs. Coordination requires communication. Cells communicate by thousands of different, but specific, receptors on cell surfaces that respond to thousands of different, but also specific, molecules (ligands) that bind to the receptors. A receptor that is bound to its activating ligand causes biochemical changes to occur in the cell. In response to such regulatory signals on the membrane, biochemical regulation within the cell occurs at the level of gene expression as well as at the level of enzyme action and other processes outside the nucleus. Ultimately these changes, through complex biochemical pathways, allow cells to divide, carry out specialized tasks, lie dormant, or die. Any of these cellular activities, when not properly coordinated, can result in illness. Two major categories of disease states are those that result from acute illness commonly caused by infections and those that are age-related. Historically, in the United States, the cause of death has transitioned from being pathogen-induced to age-related. Current scientific literature regarding cannabis indicates that its use is often bad for the former but good for the latter (see Immunology section below).
Cannabinoids and Brain Disorders
Since cannabis' action on the brain is most widely known due to its recreational use, the nervous system will serve as the starting point for examining cannabinoid activity as an example of a natural biological harm reduction system. Numerous disease states associated with the nervous system will be seen as potential targets for cannabinoid-based therapy [31]. The nervous system is composed of nerve and supporting cells. In addition to the role cannabinoids play in a healthy nervous system [32], the regulatory effects of cannabinoids in cases of stroke [28], Parkinson's disease [33], Huntington's disease [34], amyotrophic lateral sclerosis (ALS) [35], Alzheimer's disease [36], glioma (a type of brain tumor), [37] multiple sclerosis [38], seizures[39], and pain [40,41] will be examined.
Cannabinoids and the Healthy Brain
In a healthy individual, cannabinoids play a direct role in neurotransmission of many nerve cell types. They exhibit the unusual property of retrograde transmission, in which the cannabinoid neurotransmitter diffuses backwards across the neural cleft to inhibit the presynaptic action potential [42]. This function essentially regulates the sensitivity of a nerve cell by acting as a feedback mechanism that prevents excessive activity. Some nerve cells die when they are excessively stimulated by excitatory neurotransmitters (excitotoxins) such as glutamate. Cannabinoids can reduce the level of stimulation and protect against this form of cell death [43,44]. In addition to their down-regulatory effect on neurotransmission, cannabinoids play other roles in reducing this type of cell death (biological harm reduction) by regulating the role of interleukin-1 (IL-1, an inflammatory cytokine) and the IL-1 receptor antagonist (IL-1ra) [45]. For example, cannabinoids were shown to modulate the release of IL-1ra thereby protecting against IL-1 assisted cell death [46].
The role of cannabinoids in neurological health and disease goes beyond the prevention of cell death and regulates neuronal differentiation. Cannabinoid receptors are functionally coupled to the fibroblast growth factor receptor (FGF). The FGF receptor, when stimulated, activates lipid catabolism via diacylglerol (DAG) lipase which causes the hydrolysis of DAG to produce 2-arachidonyl glycerol (2AG) [47]. 2AG is an endocannabinoid shown to be important for axon growth and guidance[48]. This function is critical for nerves to innervate their target effectors. The ability to control these fundamental neurological activities, in conjunction with the anti-inflammatory properties of cannabinoids, is likely to have important regenerative health benefits for people suffering from neurological damage as occurs with stroke or injury [28].
Multiple Sclerosis
Both animal and human studies provide strong evidence of the therapeutic potential of cannabinoids to provide relief from a number of neurological disease states [49]. The use of cannabinoids to treat people suffering from multiple sclerosis (MS) is an excellent example of the importance of "medical marijuana" as an agent of harm reduction[50] MS is a neurodegenerative disease in which the immune system attacks components of the nervous system. The axons of many central nervous system (CNS) neurons are surrounded by a myelin sheath that acts much like an insulator around a wire. MS is associated with the degradation of the myelin sheath that leads to loss of axon function and cell death, thus producing the disease symptoms.
Cannabis-based therapies for the treatment of MS can provide symptomatic and true therapeutic relief. On the one hand, cannabinoids help to reduce spasticity in an animal model of MS (chronic relapsing experimental autoimmune encephalomyelitis (CREAE) [51]. However, the involvement of the cannabinoid system in the etiology of MS goes much deeper. MS is in reality an autoimmune disease. In order to appreciate why cannabinoids can have in important role, beyond what has already been mentioned, in treating MS on a mechanistic level [52], a brief introduction to immunology is required.
Cannabinoids and the Immune System
The role of the immune system is simplistically thought of as protecting us from foreign attack. More inclusively, however, the immune system has the biological function of modulating the life, death, and differentiation of cells in order to protect us. The immune system accomplishes these tasks, in part, by balancing two mutually opposed pathways known, respectively, as the "Th1" and "Th2" response. The Th1 immune response is critical for fighting infections caused by specific infectious agents [53]. This function is inhibited by cannabinoids. Thus cannabinoids are important homeostatic modulators of the immune system. While often classified as immune inhibitors, cannabinoids actually promote the Th2 response while they inhibit the Th1 response. Therefore cannabinoids are immune system modulators. A specific cannabinoid receptor (Cb2) [54] is found on most cells of the immune system.
Th1 Immune Response
The Th1 pathway is proinflammatory and functions by inducing the defensive production of free radicals that are vital for fending off pathogens, especially intracellular pathogens, such as those that cause Legionnaire's disease, Leishmania, and tuberculosis. Accordingly, the use of cannabis should be avoided when the Th1 arm of the immune system is needed to fight a particular disease. Although contagion as well as immune suppression may have been involved, a recent study supports this perspective, in that a cluster of new tuberculosis cases was traced to a shared water pipe [55]. Free radical production, inflammation and cell-mediated immunity are characteristic of the Th1 response. The targeting of infectious organisms, or infected cells, by a Th1 immune response results in healthy surrounding cells being exposed to free radicals. Much as if radiation had been applied, there is collateral damage that occurs with a targeted Th1 immune response.
Cannabinoids and Th1 Mediated Auto-Immune Diseases
In contrast to the Th1 immune response, the Th2 immune response promotes the humoral arm of the immune system. It turns down the Th1 response, is characterized by antibody production, and is typically anti-inflammatory. Ideally, the Th1 and Th2 pathways are functionally balanced to optimally meet the survival needs of an organism in its environment. In reality however, many autoimmune diseases, and other age related diseases, are characterized by an excessive Th1-driven immune response at the site of the of the tissue damage involved. Multiple sclerosis, arthritis, Crohn's disease, and diabetes are all diseases that fall into this category.
The therapeutic impact of cannabinoids on these diseases can be dramatic. For example, when rodents were given experimental autoimmune encephalomyelitis (EAE) as an MS animal model and were treated with cannabinoids, the results were profound [56]. In a study that involved both guinea pigs and rats, 98% of the EAE animals that were not treated with THC died. In contrast, greater than 95% of THC-treated animals survived. They had only mild symptoms with a delayed onset or no symptoms at all. The capacity of cannabinoids to down-regulate a spectrum of auto-immune diseases should serve as a warning against the long term use of CB1 inhibitors for weight control. Such drugs are currently in the regulatory pipeline [57] and one of the participants in the clinical trial unexpectedly developed multiple sclerosis [58].
Cannabinoid Actions-Biphasic Responses
The brief interludes into cell biology, neurology, and immunology provide a biological platform for considering how cannabinoids might impact a variety of other disease states. It is important to keep in mind that in its role as a general homeostatic modulator, too much or too little cannabinoid activity can be harmful. Cannabinoid levels or concentration ranges vary as a function of an organism's genetics, the cell types under consideration, and their health and environment. Care must be taken when evaluating the scientific literature on cannabinoids and their effects. Cannabinoids often exhibit biphasic responses [59]. Low doses of cannabinoids may stimulate the Th2 immunological response, whereas high doses may inhibit the Th2 response and shift the balance in favor of a Th1 response. From a harm reduction perspective, these observations demonstrate the critical importance of dose-dependent, disease-dependent, state-dependent, and individually tailored approaches to cannabis therapeutics [60].
The use of cannabinoids in the treatment of Parkinson's disease is an example of a condition where excessive or deficient cannabinoid activity may prove problematic. Parkinson's disease results from the loss of levo-dopamine (L-dopa) producing neurons. In an animal model of Parkinson's disease, L-dopa producing cells are killed with 6-hydroxydopamine. Rats so treated exhibit spontaneous glutamatergic activity that can be suppressed by exo- as well as endocannabinoids [61]. The standard treatment for Parkinson's disease involves L-dopa replacement therapy. Unfortunately, this treatment often results in dyskinesia (abnormal voluntary movements). Recent clinical trials have shown that cannabinoid treatment reduces the reuptake of gamma-aminobutyric acid (GABA) and relieves the L-dopa-induced dyskinesia [33], as well as L-dopa induced rotations in 6-hydroxydopamine-lesioned rats [62]. In contrast to the potential benefits of cannabinoid agonists just cited, using a different animal model, the cannabis antagonist SR141716A reduced reserpine-induced suppression of locomotion [63]. Thus, in this model locomotion was restored by inhibiting the endocannabinoid pathway.
Cannabinoids and Cancer
Possibly the greatest harm-reducing potential afforded by cannabinoids comes from their use by cancer patients. Cannabinoids possess numerous pharmacological properties that are often beneficial to cancer patients. Many people are aware of the anti-emetic and appetite stimulating effects of cannabinoids [64]. A systemic study designed to quantify the efficacy of cannabinoids as an anti-emetic agent examined data from 30 randomized controlled studies that were published between 1975 and 1997 and included 1366 patients who were administered non-smoked cannabis [65]. For patients requiring a medium level of control, cannabinoids were the preferred treatment (between 38% and 90%). This preference was lost for patients requiring a low or a high level of control. Sedation and euphoria were noted as beneficial side effects, whereas dizziness, dysphoria, hallucinations, and arterial hypotension were identified as harmful side effects.
The cancer cell killing [66] and pain relieving properties of cannabinoids are less well known to the general public. Cannabinoids may prove to be useful chemotherapeutic agents [67]. Numerous cancer types are killed in cell cultures and in animals by cannabinoids. For example, cannabinoids kill the cancer cells of various lymphoblastic malignancies such as leukemia and lymphoma [68], skin cancer [69], glioma [70], breast and prostate cancer [71], pheochromocytoma [72], thyroid cancer [73], and colorectal cancer[74]. Since 2002 THC has been used in a clinical trial in Spain for the treatment of glioma [75]. However, not all cancers are the same, and cannabinoid-induced biochemical modifications, while effective in killing the cells of some cancers, as indicated above, can have the opposite effect on the cells of other types of cancer. For example, recent work has shown that the synthetic cannabinoid, methanandamide, can promote the growth of lung cancer cells by a receptor independent pathway that involves the up-regulation of COX2 [76]. Although much has been learned about the therapeutic value of cannabinoid agonists and antagonists in different situations, scientific understanding of how to appropriately modulate the endocannabinoid pathways remains preliminary, with much remaining to be learned.
Cannabinoids and Pain
One area of current research that has begun attracting public interest is the pain relieving potential of cannabinoids, for both cancer [77] and non-cancer patients [78]. Medicine based on cannabis extract has demonstrated positive effects for pain relief [79]. Recently, an intrinsic role for cannabinoids in pain circuitry was discovered: the endocannabinoid AEA was identified as the natural ligand for the vanilloid receptors [80]. Vanilloid receptors, which are ligand-gated cation channels, are primary targets for the treatment of pain [81]. The cannabinoids seem to function in a pathway parallel to the opioid pathway [82] and are thought to exert anti-nociceptic activity at the level of the spinal cord and the brain [83], although they can also act peripherally by inhibiting mast cell degranulation [84]. In recognition of the pain relieving properties of cannabinoids, England [11] and Canada [41] are using cannabis preparations to provide relief to citizens suffering from a variety of disorders. Human trials have established that co-administration of cannabinoids can dramatically lower opioid use and can provide pain relief for neurogenic symptoms where other treatments have failed [85]. Recently, the topical application of the synthetic cannabinoid WIN 55,212-2 significantly enhanced the antinociceptive activity of morphine, opening the door for possible cannabis-induced pain relief with reduced cognitive side effects [86]. The intrinsic role of endocannabinoids in modulating pain is further supported by the up-regulation of the CB1 receptor in rats following nerve damage [29]. Once again, nature has selected cannabinoids to reduce harm.
Smoking and Lung Cancer
Fundamental to any consideration of cannabis-based harm reduction, as a biological phenomenon or as a policy, is how to best administer the drug. Smoking cannabis preparations, in contrast to oral administration [87], has the benefit of rapid action that allows self-titration of the drug's activity [88,89]. Unfortunately, cannabis smoke contains numerous carcinogenic compounds [90]. In fact, cannabis smoke may contain more tars than tobacco smoke [91]. However, despite the fact that cannabis smoke does produce cellular changes that are viewed as precancerous, a major epidemiological study does not find that cannabis smoking is associated with tobacco related cancers [92]. A number of recent studies provide a scientific foundation for the clear relationship between tobacco smoking and lung cancer, a relationship that does not hold true for cannabis smoke (manuscript submitted to HRJ). For example nicotine, acting via nicotine receptors, is critical in the development of tobacco related cancer by inhibiting the death of genetically damaged cells [93]. Tobacco also promotes the development of blood vessels needed to support tumor growth [94] whereas cannabis inhibits tumor vascularization in nonmelanoma skin cancer [69] and glioma [95]. Although conclusions derived from an oft-cited study examining the carcinogenic effects of cannabis, tobacco, and cannabis combined with tobacco claims to show a link between cannabis smoking and head and neck cancer [96]. But these results do not hold up under scrutiny. The study does support a link between tobacco use that is exacerbated by concurrent cannabis use and the development of head and neck cancer. However, the "cannabis use only" group was composed only of two subjects, undermining the statistical relevance of conclusions regarding this group.
Smoking Alternatives
Regardless of whether or not smoking cannabis can cause lung cancer, smoking anything containing partially oxidized hydrocarbons, carcinogens, and irritants a priori, is not healthy and will have negative health consequences. Fortunately, harm-reducing alternatives exist. While often touted as a problem, the availability of high THC cannabis with high levels of THC permits less cannabis to be smoked for therapeutic effects. Additionally, methods of vaporizing the active ingredients of cannabis have been shown to successfully remove most compounds of concern while efficiently delivering the desired ones [97]. These results contrast with a recent Australian study that found that the use of a water pipe, or bong, failed to reduce tars or carbon monoxide delivered to the smoker [98]. GW Pharmaceuticals is developing an oral spray that should prove to be an additional safe and effective alternative delivery system [12] and valuable to medical cannabis users. The company has also identified strains with defined ratios of various cannabinoids for which specific medicinal value will be determined.
Cannabinoids Affect Drug Metabolism
Another important cannabis and harm reduction topic that must be considered is that of how the use of cannabis impacts on the pharmacokinetics of other drugs [99]. A number of drugs are metabolized by the P450 family of isoenzymes, including numerous cannabinoids [100]. Even though cannabinoids stimulate the transcription of P450 (2A and 3C), they also directly inhibit the activity of this enzyme [101]. There are likely to be pros and cons associated with P450 inhibition. P450 activity activates procarcinogens in tobacco smoke to create active cancer-causing mutations [102]. Thus, the inhibition of these enzymes by cannabinoids may minimize some of the negative consequences of smoke inhalation. On the other hand, many pharmaceutical drugs are metabolized by these enzymes. The reduction of the rate of drug metabolism by cannabinoids with pharmokinetic consequences has been shown for cocaine [103], barbiturates [104], opiates [105], alcohol, the antipsychotic haloperidol [106], and others [107].
Thus far, both endo- and exocannabinoids are seen to reduce harm in numerous circumstances. Cannabinoid-based therapies have been especially helpful for the treatment of a variety of neurological and immunological disorders. Yet, we have only scratched the surface of the scientific literature on cannabinoids and their biological effects. Nevertheless, it should be apparent that cannabinoids have enormous medical potential as we learn to manipulate the natural cannabinoid harm reduction system that has evolved in the animal kingdom.
A fundamental question that remains unanswered is how basic, complex biochemical phenomena, as touched on briefly in this article, collectively emerge as substantial contributors to health and behavior. In far-from-equilibrium thermodynamic systems, such as living organisms, there are discontinuities between underlying molecular dynamics and associated emergent macroscopic phenomena [108]. In such systems, small changes (called "perturbations") can amplify with consequences for the organization of the whole system. The cannabinoids help to regulate an amazingly broad range of biochemical events. All of these effects have genetic foundations. As such, natural genetic/biochemical variation in a population can be expected to have significant effects on health and behavior. It should be expected that in a population distribution of cannabinoid levels and sensitivities, as a function of an individual's health/disease status, some individuals would naturally need to increase their cannabinoid activity while others would need theirs lowered. Although the focus of this paper has been to suggest the many circumstances in which higher cannabinoid activity would be beneficial, these circumstances will necessarily differ among individuals with different congenital cannabinoid levels and sensitivities. Therefore, reduced cannabinoid activity would be beneficial under some conditions. A prime example of potential harmful effects of excess cannabinoids is their effects on pregnancy where low levels are needed but high levels are harmful [109].
Behavioral Effects: Self-administration and Reward
The broad homeostatic activities of cannabinoids that have been developed in this article have been rooted in hard science. The extension of these ideas to the psychological and behavioral levels is intrinsically more speculative, but remains consistent with the literature. For years, researchers have looked into the possible addictive qualities of cannabis. The lack of significant reward behavior was indicated by the lack of self-administration in primates. Experiments examining preference in rats demonstrated that low doses of THC could induce place preference but that higher doses produced drug aversion [110], again demonstrating the homeostatic nature of cannabinoids. Self-administration is typical of most psychoactive drugs of abuse. Hence, one could conclude that marijuana has a low potential for abuse.
Some may question the conclusion that cannabis has a low abuse potential since an animal model using squirrel monkeys was recently developed in which self-administration behavior was maintained using THC [111]. Interestingly, and consistent with the notion that the cannabinoid system is a biological homeostatic harm reduction mechanism, the self-administration of THC ranges from 2 to 8 ug/kg and peaks at 4 ug/kg [112]. Thus, in this animal model a controlled dose is chosen. To further put these experiments in perspective, the dose used must be examined more closely. A 1-gram joint of 10% THC content would contain 100 mg of THC. The self-administered dose schedule chosen by the animal of 4 ug/kg would correspond to 360 ug of THC (if absorption was complete, approximately 1/278 of the joint) for a 200-pound human. Similarly, in rats, the intravenous self-administration of the synthetic cannabinoid Win 55,212-2 also occurred in a biphasic manner, with a maximum response occurring at 12 ug/kg[113] The self-regulated, controlled use of low drug doses is not characteristic of addictive drugs of abuse.
Additional cannabinoid involvement in reward behavior is suggested by the increased activity of dopaminergic neurons stimulated with psychoactive cannabinoids [114]. This pathway is shared by other major drugs of abuse including, morphine, ethanol, and nicotine [115]. However, the production of glucocorticoid hormones that are normally produced in response to stress [116], are suppressed by cannabinoids [117]. Are cannabinoids addictive, is pleasure addictive, or is a low stress state addictive?
Cannabinoids and Stress
Stress and reward are complicated components of addictive behavior. How does repeated use of THC influence these states? A recent study examines this question by measuring glucose utilization in different areas of the rat brain following repeated treatment with THC [118]. After 7 and 21 days of THC treatment, THC no longer resulted in reduced glucose utilization in many areas of the brain typically affected by a single THC dose (most cortical, thalamic, and basal ganglia regions). In contrast, glucose utilization in other areas of the brain remained unaltered (nucleus accumbens, mediodorsal thalamus, basolateral amygdala, portions of the hippocampus and median raphe). Thus while the effects of THC on body temperature and locomotor activity become resistant to repeated THC administration, those areas involved in many higher brain functions remain responsive to THC. This differential adaptation to THC administration is consistent with a low addictive potential. The best evidence that demonstrates the absence of an addictive response to cannabis use is the fact that most people who use it do not continue to use it, and stop using it without any effort.
The stress-relieving properties of cannabinoids are an important aspect of their pharmacological activity. An interesting mechanism by which cannabinoids may promote stress relief is through their effects on memory. Cannabinoids control the extinction of painful memories [119]. What a blessing for those suffering from debilitating or life threatening illnesses: cannabinoids may help them to forget their misfortune.
Independent of the direct addictive or non-addictive properties of cannabis, the cannabis-opioid connection will be examined in more detail. Both drug families function (not necessarily exclusively) through biochemical pathways that are regulated by specific receptor-ligand interactions. However, there appears to be, as yet not fully defined, crosstalk between these pathways [120]. For example, CB1 receptor knockout mice are non-responsive to CB1 cannabinoid activities and show reduced addictive effects of opiates [121]. Similarly, Lewis rats showed enhanced sensitivity to morphine self-administration after treatment with the synthetic cannabinoid CP55040 [122]. Examining the cannabis-opioid connection from the other direction, chronic morphine administration results in some down-regulation of cannabinoid receptors along with a significant reduction in 2AG [123]. These results show both positive and negative feedback relationships between the endocannabinoid and opiate systems. They also suggest that cannabinoids might serve to reduce the symptoms of opiate withdrawal [124].
The possibility that cannabinoids could serve as an addiction interrupter was demonstrated in rats where the synthetic cannabinoid agonist Win 55-212,2 reduced intravenous self-administration of cocaine [125]. Similarly, recent studies indicate that THC may facilitate nicotine withdrawal in mice [126] and inhibit alcohol preference in a model of alcoholism [127]. The opposite indications, that blocking cannabinoids receptors could serve as an addition interrupter has also been made [128].
Behavioral Complexity
Behavioral processes and their complexities set humans apart from other animals. Can we simply extrapolate from animal to human behavior? It is one thing to comparatively examine the molecular and cell biology of animals and extrapolate to humans. However, the behavioral repertoire of humans appears to be dramatically enhanced over other animals and is therefore more difficult to connect between the species. Evolutionary relationships show that the cannabinoid receptors are located in the more advanced areas of our brains. Again, any population is always a spread around the average value of any parameter. A subset of the human population will inevitably retain a more primitive behavioral repertoire. Is this subset more susceptible to addictive behavior or psychological problems that could result from cannabis consumption? Has the cannabinoid system been optimized for the regulation of more primitive behavior or, alternatively, is it better optimized for the behavioral flexibility required of modern humans? Indeed, is there any evidence that the cannabinoid system, like our cortical capacity, may enable even greater behavioral flexibility in the more complex societies and altered environments of the future?
Answers to these questions are suggested by the data of human cannabis consumption. Most people who use cannabis in their youth stop using it as their lives progress. Most do so as a natural part of their development. They do so without outside intervention or help. They do so without ever having become heroin users, schizophrenic, or motivationally compromised. These facts indicate that for the majority of people who try marijuana, it is not addictive, does not lead to heroin use, nor is it a trigger for the onset of psychological problems. However, due to the complexity of cannabinoid activities, it is likely that in a small percentage of the population, cannabis use may foster problems. The biology presented in this paper suggests that such individual differences should be expected. We must learn to identify individuals who would be negatively affected by cannabis use; they are the people that an intelligent drug policy would help to identify and assist. In contrast, our policy criminalizes the majority of users and further harms them, perhaps psychologically as well as medically, through its repercussions.
The use of cannabis – and any mind-altering drug – by young developing minds rightfully remains an area of focus and concern. For example, is there a relationship between cannabis use and schizophrenia? Schizophrenia is characterized by distortions of reality, disturbances of language and thought processes, and social withdrawal. Certainly, aspects of cannabis intoxication parallel these symptoms. It is feared that cannabis can precipitate this state [129], especially in susceptible individuals [130]. It has been suggested that schizophrenics (or potential schizophrenics) fall into two categories with respect to cannabis use [131]. One group may find symptomatic relief in the use of cannabis, while the other may actually take the risk of inducing the onset of the disease. The complexities of this issue are illuminated by the unpredictable behavior of interacting complex systems such as the nervous and immune systems, as will be considered below.
In an important recent study, De Marchi et al [132], examined the endocannabinoid levels in healthy volunteers and compared them to that of schizophrenic patients, both before and after successful antipsychotic treatment. Patients suffering with acute disease had significantly higher anandamide levels in their blood than did the normal individuals or patients in clinical remission. Might these elevated cannabinoid levels be contributing to the disease symptoms, and what might be causing them? Cannabinoids act homeostatically across biological subsystems. A possible immune involvement in schizophrenia has long been suspected, and immunological parameters have been implicated in the disease. For example, there is an inverse correlation between schizophrenia and rheumatoid arthritis; an individual generally does not get both illnesses [133]. Interestingly, schizophrenia has been correlated with HLA type, Toxoplasma gonodii infection, and exposure to cats [133]. Toxoplasma gonodii infects brain neurons, and is best controlled with a strong pro-inflammatory immune response. Endocannabinoids modulate the pro-inflammatory TH1 response by up-regulating the anti-inflammatory Th2 response. Hence, it is likely that some individuals idiosyncratically respond to Toxoplasma gonodii infections by producing excess endocannabinoids and suffering the associated abnormal mental state. Antipsychotic drugs have actually improved the outcome of infection with this parasite[134].
Conclusion
Evolution has selected the endocannabinoids to homeostatically regulate numerous biological phenomena that can be found in every organized system in the body, and to counteract biochemical imbalances that are characteristic of numerous damaged or diseased states, in particular those associated with aging. Starting from birth, cannabinoids are present in mother's milk [135], where they initiate the eating process. If the activity of endocannabinoids in the mouse milk is inhibited with a cannabinoid antagonist, the newborn mice die of starvation. As life proceeds, endocannabinoids continuously regulate appetite, body temperature, reproductive activity, and learning capacity. When a body is physically damaged, the endocannabinoids are called on to reduce inflammation, protect neurons [136], regulate cardiac rhythms [137] and protect the heart form oxygen deprivation [20]. In humans suffering from colorectal cancer, endocannabinoid levels are elevated in an effort to control the cancer [74]. They help relieve emotional suffering by reducing pain and facilitating movement beyond the fears of unpleasant memories [119].
While this review is far from complete, it attempts to provide a conceptual overview that supports the endocannabinoid system as being nature's method of harm reduction. There is a pattern to all the cannabinoid-mediated activities described. Many of the biochemical imbalances that cannabinoids protect against are associated with aging. Aging itself is a system-wide movement towards chemical equilibrium (away from the highly regulated far-from-equilibrium state) and as such is an imbalance from which all living organisms suffer. In contrast, the harmful consequences of cannabis use, however exaggerated they often appear to be, are likely to represent significant potential risk for a minority of the population for whom reduced cannabinoid levels might promote mental stability, fertility or more regulated food consumption.
Supplementary Material
Additional File 1
It contains hyperlinks to the text document.
Click here for file
Acknowledgements
I thank Suzanne Stradley, Jenell Forschler, and Carolyn Rogers, graduate students in Laura Fillmore's electronic publishing course: Writing and Publishing Program (Emerson College), for creating the text links used in this article (see additional file 1).
==== Refs
Lindsay GB Rainey J Psychosocial and pharmacologic explanations of nicotine's "gateway drug" function J Sch Health 1997 67 123 126 9130188
Ginzler JA Cochran BN Domenech-Rodriguez M Cauce AM Whitbeck LB Sequential progression of substance use among homeless youth: an empirical investigation of the gateway theory Subst Use Misuse 2003 38 725 758 12747403 10.1081/JA-120017391
Herstek J Watson A Kammer C Pharmaceuticals issue brief: medical marijuana: year end report-2002 Issue Brief Health Policy Track Serv 2002 1 12 12856685
Russo E Mathre ML Byrne A Velin R Bach PJ Sanchez-Ramos J Kirlin KA Chronic Cannabis Use in the Compassionate Investigational New Drug Program:An Examination of Benefits and Adverse Effects of Legal Clinical Cannabis Journal of Cannabis Therapeutics 2002 3 56 10.1300/J175v02n01_02
Mechoulam R Gaoni Y The absolute configuration of delta-1-tetrahydrocannabinol, the major active constituent of hashish Tetrahedron Lett 1967 12 1109 1111 6039537 10.1016/S0040-4039(00)90646-4
Herkenham M Lynn AB Little MD Johnson MR Melvin LS de Costa BR Rice KC Cannabinoid receptor localization in brain Proc Natl Acad Sci U S A 1990 87 1932 1936 2308954
Matsuda LA Lolait SJ Brownstein MJ Young AC Bonner TI Structure of a cannabinoid receptor and functional expression of the cloned cDNA Nature 1990 346 561 564 2165569 10.1038/346561a0
Devane WA Hanus L Breuer A Pertwee RG Stevenson LA Griffin G Gibson D Mandelbaum A Etinger A Mechoulam R Isolation and structure of a brain constituent that binds to the cannabinoid receptor Science 1992 258 1946 1949 1470919
Di M Bisogno T De Petrocellis L Endocannabinoids: new targets for drug development Curr Pharm Des 2000 6 1361 1380 10903398 10.2174/1381612003399365
Cravatt BF Lichtman AH Fatty acid amide hydrolase: an emerging therapeutic target in the endocannabinoid system Curr Opin Chem Biol 2003 7 469 475 12941421 10.1016/S1367-5931(03)00079-6
Baker D Pryce G Giovannoni G Thompson AJ The therapeutic potential of cannabis Lancet Neurol 2003 2 291 298 12849183 10.1016/S1474-4422(03)00381-8
Cannabis-Based Medicines – GW Pharmaceuticals: High CBD, High THC, Medicinal Cannabis – GW Pharmaceuticals, THC:CBD Drugs R D 2003 4 306 309 12952500
Elphick MR Satou Y Satoh N The invertebrate ancestry of endocannabinoid signalling: an orthologue of vertebrate cannabinoid receptors in the urochordate Ciona intestinalis Gene 2003 302 95 101 12527200 10.1016/S0378-1119(02)01094-6
Elphick MR Egertova M The neurobiology and evolution of cannabinoid signalling Philos Trans R Soc Lond B Biol Sci 2001 356 381 408 11316486 10.1098/rstb.2000.0787
Sepe N De Petrocellis L Montanaro F Cimino G Di Marzo V Bioactive long chain N-acylethanolamines in five species of edible bivalve molluscs. Possible implications for mollusc physiology and sea food industry Biochim Biophys Acta 1998 1389 101 111 9461251
De Petrocellis L Melck D Bisogno T Milone A Di Marzo V Finding of the endocannabinoid signalling system in Hydra, a very primitive organism: possible role in the feeding response Neuroscience 1999 92 377 387 10392859 10.1016/S0306-4522(98)00749-0
Guzman M Sanchez C Effects of cannabinoids on energy metabolism Life Sci 1999 65 657 664 10462066 10.1016/S0024-3205(99)00288-X
Yuan M Kiertscher SM Cheng Q Zoumalan R Tashkin DP Roth MD Delta 9-Tetrahydrocannabinol regulates Th1/Th2 cytokine balance in activated human T cells J Neuroimmunol 2002 133 124 131 12446015 10.1016/S0165-5728(02)00370-3
Ralevic V Cannabinoid modulation of peripheral autonomic and sensory neurotransmission Eur J Pharmacol 2003 472 1 21 12860468 10.1016/S0014-2999(03)01813-2
Hiley CR Ford WR Endocannabinoids as mediators in the heart: a potential target for therapy of remodelling after myocardial infarction? Br J Pharmacol 2003 138 1183 1184 12711614 10.1038/sj.bjp.0705155
Brown TT Dobs AS Endocrine effects of marijuana J Clin Pharmacol 2002 42 90S 96S 12412841
Pinto L Izzo AA Cascio MG Bisogno T Hospodar-Scott K Brown DR Mascolo N Di Marzo V Capasso F Endocannabinoids as physiological regulators of colonic propulsion in mice Gastroenterology 2002 123 227 234 12105851 10.1053/gast.2002.34242
Dmitrieva N Berkley KJ Contrasting effects of WIN 55212-2 on motility of the rat bladder and uterus J Neurosci 2002 22 7147 7153 12177210
Grotenhermen F Muller-Vahl K IACM 2nd Conference on Cannabinoids in Medicine Expert Opin Pharmacother 2003 4 2367 2371 14640935 10.1517/14656566.4.12.2367
Habayeb OM Bell SC Konje JC Endogenous cannabinoids: metabolism and their role in reproduction Life Sci 2002 70 1963 1977 12148689 10.1016/S0024-3205(01)01539-9
Calignano A Katona I Desarnaud F Giuffrida A La Rana G Mackie K Freund TF Piomelli D Bidirectional control of airway responsiveness by endogenous cannabinoids Nature 2000 408 96 101 11081515 10.1038/35040576
Rottanburg D Robins AH Ben-Arie O Teggin A Elk R Cannabis-associated psychosis with hypomanic features Lancet 1982 2 1364 1366 6129463 10.1016/S0140-6736(82)91270-3
Nagayama T Sinor AD Simon RP Chen J Graham SH Jin K Greenberg DA Cannabinoids and neuroprotection in global and focal cerebral ischemia and in neuronal cultures J Neurosci 1999 19 2987 2995 10191316
Lim G Sung B Ji RR Mao J Upregulation of spinal cannabinoid-1-receptors following nerve injury enhances the effects of Win 55,212-2 on neuropathic pain behaviors in rats Pain 2003 105 275 283 14499445 10.1016/S0304-3959(03)00242-2
Caberlotto L Rimondini R Hansson A Eriksson S Heilig M Corticotropin-Releasing Hormone (CRH) mRNA Expression in Rat Central Amygdala in Cannabinoid Tolerance and Withdrawal: Evidence for an Allostatic Shift? Neuropsychopharmacology 2003
Croxford JL Therapeutic potential of cannabinoids in CNS disease CNS Drugs 2003 17 179 202 12617697
Wilson RI Nicoll RA Endocannabinoid signaling in the brain Science 2002 296 678 682 11976437 10.1126/science.1063545
Sieradzan KA Fox SH Hill M Dick JP Crossman AR Brotchie JM Cannabinoids reduce levodopa-induced dyskinesia in Parkinson's disease: a pilot study Neurology 2001 57 2108 2111 11739835
Lastres-Becker I de Miguel R De Petrocellis L Makriyannis A Di Marzo V Fernandez-Ruiz J Compounds acting at the endocannabinoid and/or endovanilloid systems reduce hyperkinesia in a rat model of Huntington's disease J Neurochem 2003 84 1097 1109 12603833 10.1046/j.1471-4159.2003.01595.x
Raman C McAllister SD Rizvi G Patel SG Moore DH Abood ME Amyotrophic lateral sclerosis: delayed disease progression in mice by treatment with a cannabinoid Amyotroph Lateral Scler Other Motor Neuron Disord 2004 5 33 39 15204022 10.1080/14660820310016813
Milton NG Anandamide and noladin ether prevent neurotoxicity of the human amyloid-beta peptide Neurosci Lett 2002 332 127 130 12384227 10.1016/S0304-3940(02)00936-9
Guzman M Sanchez C Galve-Roperh I Control of the cell survival/death decision by cannabinoids J Mol Med 2001 78 613 625 11269508 10.1007/s001090000177
Pryce G Ahmed Z Hankey DJ Jackson SJ Croxford JL Pocock JM Ledent C Petzold A Thompson AJ Giovannoni G Cuzner ML Baker D Cannabinoids inhibit neurodegeneration in models of multiple sclerosis Brain 2003 12876144
Wallace MJ Blair RE Falenski KW Martin BR DeLorenzo RJ The endogenous cannabinoid system regulates seizure frequency and duration in a model of temporal lobe epilepsy J Pharmacol Exp Ther 2003
Iversen L Cannabis and the brain Brain 2003 126 1252 1270 12764049 10.1093/brain/awg143
Ware MA Gamsa A Persson J Fitzcharles MA Cannabis for chronic pain: case series and implications for clinicians Pain Res Manag 2002 7 95 99 12185373
Wilson RI Nicoll RA Endogenous cannabinoids mediate retrograde signalling at hippocampal synapses Nature 2001 410 588 592 11279497 10.1038/35069076
Nadler V Mechoulam R Sokolovsky M The non-psychotropic cannabinoid (+)-(3S,4S)-7-hydroxy-delta 6- tetrahydrocannabinol 1,1-dimethylheptyl (HU-211) attenuates N-methyl-D-aspartate receptor-mediated neurotoxicity in primary cultures of rat forebrain Neurosci Lett 1993 162 43 45 8121633 10.1016/0304-3940(93)90555-Y
Hampson AJ Grimaldi M Axelrod J Wink D Cannabidiol and (-)Delta9-tetrahydrocannabinol are neuroprotective antioxidants Proc Natl Acad Sci U S A 1998 95 8268 8273 9653176 10.1073/pnas.95.14.8268
Patel HC Boutin H Allan SM Interleukin-1 in the brain: mechanisms of action in acute neurodegeneration Ann N Y Acad Sci 2003 992 39 47 12794045
Molina-Holgado F Pinteaux E Moore JD Molina-Holgado E Guaza C Gibson RM Rothwell NJ Endogenous interleukin-1 receptor antagonist mediates anti-inflammatory and neuroprotective actions of cannabinoids in neurons and glia J Neurosci 2003 23 6470 6474 12878687
Sugiura T Kondo S Sukagawa A Nakane S Shinoda A Itoh K Yamashita A Waku K 2-Arachidonoylglycerol: a possible endogenous cannabinoid receptor ligand in brain Biochem Biophys Res Commun 1995 215 89 97 7575630 10.1006/bbrc.1995.2437
Williams EJ Walsh FS Doherty P The FGF receptor uses the endocannabinoid signaling system to couple to an axonal growth response J Cell Biol 2003 160 481 486 12578907 10.1083/jcb.200210164
Glass M The role of cannabinoids in neurodegenerative diseases Prog Neuropsychopharmacol Biol Psychiatry 2001 25 743 765 11383976 10.1016/S0278-5846(01)00162-2
Page SA Verhoef MJ Stebbins RA Metz LM Levy JC Cannabis use as described by people with multiple sclerosis Can J Neurol Sci 2003 30 201 205 12945941
Baker D Pryce G Croxford JL Brown P Pertwee RG Huffman JW Layward L Cannabinoids control spasticity and tremor in a multiple sclerosis model Nature 2000 404 84 87 10716447 10.1038/35003583
Baker D Pryce G The therapeutic potential of cannabis in multiple sclerosis Expert Opin Investig Drugs 2003 12 561 567 12665412 10.1517/13543784.12.4.561
Klein TW Newton CA Nakachi N Friedman H Delta 9-tetrahydrocannabinol treatment suppresses immunity and early IFN-gamma, IL-12, and IL-12 receptor beta 2 responses to Legionella pneumophila infection J Immunol 2000 164 6461 6466 10843702
Munro S Thomas KL Abu-Shaar M Molecular characterization of a peripheral receptor for cannabinoids Nature 1993 365 61 65 7689702 10.1038/365061a0
Munckhof WJ Konstantinos A Wamsley M Mortlock M Gilpin C A cluster of tuberculosis associated with use of a marijuana water pipe Int J Tuberc Lung Dis 2003 7 860 865 12971670
Lyman WD Sonett JR Brosnan CF Elkin R Bornstein MB Delta 9-tetrahydrocannabinol: a novel treatment for experimental autoimmune encephalomyelitis J Neuroimmunol 1989 23 73 81 2542370 10.1016/0165-5728(89)90075-1
Fernandez JR Allison DB Rimonabant Sanofi-Synthelabo Curr Opin Investig Drugs 2004 5 430 435 15134285
van Oosten BW Killestein J Mathus-Vliegen EM Polman CH Multiple sclerosis following treatment with a cannabinoid receptor-1 antagonist Mult Scler 2004 10 330 331 15222701 10.1191/1352458504ms1044oa
Berdyshev EV Boichot E Germain N Allain N Anger JP Lagente V Influence of fatty acid ethanolamides and delta9-tetrahydrocannabinol on cytokine and arachidonate release by mononuclear cells Eur J Pharmacol 1997 330 231 240 9253958 10.1016/S0014-2999(97)01007-8
Melamede RJ Indications for Cannabinoids: Autoimmune Diseases Cannabis and Cannabinoids, Pharmacology, Toxicology and Therapeutic Potential 2000
Gubellini P Picconi B Bari M Battista N Calabresi P Centonze D Bernardi G Finazzi-Agro A Maccarrone M Experimental parkinsonism alters endocannabinoid degradation: implications for striatal glutamatergic transmission J Neurosci 2002 22 6900 6907 12177188
Gilgun-Sherki Y Melamed E Mechoulam R Offen D The CB1 Cannabinoid Receptor Agonist, HU-210, Reduces Levodopa-Induced Rotations in 6-Hydroxydopamine-Lesioned Rats Pharmacol Toxicol 2003 93 66 70 12899667 10.1034/j.1600-0773.2003.930202.x
Di Marzo V Hill MP Bisogno T Crossman AR Brotchie JM Enhanced levels of endogenous cannabinoids in the globus pallidus are associated with a reduction in movement in an animal model of Parkinson's disease FASEB J 2000 14 1432 1438 10877836 10.1096/fj.14.10.1432
Mechoulam R Hanu L The cannabinoids: an overview. Therapeutic implications in vomiting and nausea after cancer chemotherapy, in appetite promotion, in multiple sclerosis and in neuroprotection Pain Res Manag 2001 6 67 73 11854768
Tramer MR Carroll D Campbell FA Reynolds DJ Moore RA McQuay HJ Cannabinoids for control of chemotherapy induced nausea and vomiting: quantitative systematic review BMJ 2001 323 16 21 11440936 10.1136/bmj.323.7303.16
Guzman M Sanchez C Galve-Roperh I Cannabinoids and cell fate Pharmacol Ther 2002 95 175 184 12182964 10.1016/S0163-7258(02)00256-5
Parolaro D Massi P Rubino T Monti E Endocannabinoids in the immune system and cancer Prostaglandins Leukot Essent Fatty Acids 2002 66 319 332 12052046 10.1054/plef.2001.0355
McKallip RJ Lombard C Fisher M Martin BR Ryu S Grant S Nagarkatti PS Nagarkatti M Targeting CB2 cannabinoid receptors as a novel therapy to treat malignant lymphoblastic disease Blood 2002 100 627 634 12091357 10.1182/blood-2002-01-0098
Casanova ML Blazquez C Martinez-Palacio J Villanueva C Fernandez-Acenero MJ Huffman JW Jorcano JL Guzman M Inhibition of skin tumor growth and angiogenesis in vivo by activation of cannabinoid receptors J Clin Invest 2003 111 43 50 12511587 10.1172/JCI200316116
Sanchez C Galve-Roperh I Canova C Brachet P Guzman M Delta9-tetrahydrocannabinol induces apoptosis in C6 glioma cells FEBS Lett 1998 436 6 10 9771884 10.1016/S0014-5793(98)01085-0
Melck D De Petrocellis L Orlando P Bisogno T Laezza C Bifulco M Di Marzo V Suppression of nerve growth factor Trk receptors and prolactin receptors by endocannabinoids leads to inhibition of human breast and prostate cancer cell proliferation Endocrinology 2000 141 118 126 10614630 10.1210/en.141.1.118
Wilson RGJ Tahir SK Mechoulam R Zimmerman S Zimmerman AM Cannabinoid enantiomer action on the cytoarchitecture Cell Biol Int 1996 20 147 157 8935159 10.1006/cbir.1996.0019
Portella G Laezza C Laccetti P De Petrocellis L Di Marzo V Bifulco M Inhibitory effects of cannabinoid CB1 receptor stimulation on tumor growth and metastatic spreading: actions on signals involved in angiogenesis and metastasis FASEB J 2003 17 1771 1773 12958205
Ligresti A Bisogno T Matias I De Petrocellis L Cascio MG Cosenza V D'argenio G Scaglione G Bifulco M Sorrentini I Di Marzo V Possible endocannabinoid control of colorectal cancer growth Gastroenterology 2003 125 677 687 12949714 10.1016/S0016-5085(03)00881-3
Blazquez C Gonzalez-Feria L Alvarez L Haro A Casanova ML Guzman M Cannabinoids inhibit the vascular endothelial growth factor pathway in gliomas Cancer Res 2004 64 5617 5623 15313899 10.1158/0008-5472.CAN-03-3927
Gardner B Zhu LX Sharma S Tashkin DP Dubinett SM Methanandamide increases COX-2 expression and tumor growth in murine lung cancer FASEB J 2003
Kehl LJ Hamamoto DT Wacnik PW Croft DL Norsted BD Wilcox GL Simone DA A cannabinoid agonist differentially attenuates deep tissue hyperalgesia in animal models of cancer and inflammatory muscle pain Pain 2003 103 175 186 12749972 10.1016/S0304-3959(02)00450-5
Ware MA Doyle CR Woods R Lynch ME Clark AJ Cannabis use for chronic non-cancer pain: results of a prospective survey Pain 2003 102 211 216 12620613 10.1016/s0304-3959(02)00400-1
Berman J Lee J Cooper M Cannon A Sach J McKerral S Taggart M Symonds C Fishe K Birch R Efficacy of two cannabis-based medicinal extracts for relief of central neuropathic pain from brachial plexus avulsion: results of a randomised controlled trial Anaesthesia 2003 58 938 10.1046/j.1365-2044.2003.03408_3.x
Smart D Gunthorpe MJ Jerman JC Nasir S Gray J Muir AI Chambers JK Randall AD Davis JB The endogenous lipid anandamide is a full agonist at the human vanilloid receptor (hVR1) Br J Pharmacol 2000 129 227 230 10694225 10.1038/sj.bjp.0703050
Caterina MJ Schumacher MA Tominaga M Rosen TA Levine JD Julius D The capsaicin receptor: a heat-activated ion channel in the pain pathway Nature 1997 389 816 824 9349813 10.1038/39807
Walker JM Huang SM Endocannabinoids in pain modulation Prostaglandins Leukot Essent Fatty Acids 2002 66 235 242 12052039 10.1054/plef.2001.0361
Walker JM Hohmann AG Martin WJ Strangman NM Huang SM Tsou K The neurobiology of cannabinoid analgesia Life Sci 1999 65 665 673 10462067 10.1016/S0024-3205(99)00289-1
Rice AS Farquhar-Smith WP Nagy I Endocannabinoids and pain: spinal and peripheral analgesia in inflammation and neuropathy Prostaglandins Leukot Essent Fatty Acids 2002 66 243 256 12052040 10.1054/plef.2001.0362
Wade DT Robson P House H Makela P Aram J A preliminary controlled study to determine whether whole-plant cannabis extracts can improve intractable neurogenic symptoms Clin Rehabil 2003 17 21 29 12617376 10.1191/0269215503cr581oa
Yesilyurt O Dogrul A Gul H Seyrek M Kusmez O Ozkan Y Yildiz O Topical cannabinoid enhances topical morphine antinociception Pain 2003 105 303 308 14499448 10.1016/S0304-3959(03)00245-8
Hall W Degenhardt L Medical marijuana initiatives: are they justified? How successful are they likely to be? CNS Drugs 2003 17 689 697 12873153
Abrams DI Medical marijuana: tribulations and trials J Psychoactive Drugs 1998 30 163 169 9692378
Grotenhermen F Pharmacokinetics and pharmacodynamics of cannabinoids Clin Pharmacokinet 2003 42 327 360 12648025
Novotny M Lee ML Bartle KD A possible chemical basis for the higher mutagenicity of marijuana smoke as compared to tobacco smoke Experientia 1976 32 280 282 1253890 10.1007/BF01940790
Wu TC Tashkin DP Djahed B Rose JE Pulmonary hazards of smoking marijuana as compared with tobacco N Engl J Med 1988 318 347 351 3340105
Sidney S Quesenberry CPJ Friedman GD Tekawa IS Marijuana use and cancer incidence (California, United States) Cancer Causes Control 1997 8 722 728 9328194 10.1023/A:1018427320658
Minna JD Nicotine exposure and bronchial epithelial cell nicotinic acetylcholine receptor expression in the pathogenesis of lung cancer J Clin Invest 2003 111 31 33 12511585 10.1172/JCI200317492
Heeschen C Jang JJ Weis M Pathak A Kaji S Hu RS Tsao PS Johnson FL Cooke JP Nicotine stimulates angiogenesis and promotes tumor growth and atherosclerosis Nat Med 2001 7 833 839 11433349 10.1038/89961
Blazquez C Casanova ML Planas A Del Pulgar TG Villanueva C Fernandez-Acenero MJ Aragones J Huffman JW Jorcano JL Guzman M Inhibition of tumor angiogenesis by cannabinoids FASEB J 2003 17 529 531 12514108
Zhang ZF Morgenstern H Spitz MR Tashkin DP Yu GP Marshall JR Hsu TC Schantz SP Marijuana use and increased risk of squamous cell carcinoma of the head and neck Cancer Epidemiol Biomarkers Prev 1999 8 1071 1078 10613339
Gieringer D Laurent SJ Goodrich S Cannabis Vaporizer Combines Efficient Delivery of THC with Effective Suppression of Pyrolytic Compounds Journal of Cannabis Therapeutics 4 7 27 10.1300/J175v04n01_02
Gowling L Ali R White J Respiratory Harms of Smoked Cannabis 2000 Adelaide Australia: Drug and Alcohol Services Council 11819707
Agurell S Halldin M Lindgren JE Ohlsson A Widman M Gillespie H Hollister L Pharmacokinetics and metabolism of delta 1-tetrahydrocannabinol and other cannabinoids with emphasis on man Pharmacol Rev 1986 38 21 43 3012605
Yamamoto I Watanabe K Narimatsu S Yoshimura H Recent advances in the metabolism of cannabinoids Int J Biochem Cell Biol 1995 27 741 746 7584607 10.1016/1357-2725(95)00043-O
Bornheim LM Everhart ET Li J Correia MA Induction and genetic regulation of mouse hepatic cytochrome P450 by cannabidiol Biochem Pharmacol 1994 48 161 171 8043019 10.1016/0006-2952(94)90236-4
Tretyakova N Matter B Jones R Shallop A Formation of benzo[a]pyrene diol epoxide-DNA adducts at specific guanines within K-ras and p53 gene sequences: stable isotope-labeling mass spectrometry approach Biochemistry 2002 41 9535 9544 12135376 10.1021/bi025540i
Pellinen P Honkakoski P Stenback F Niemitz M Alhava E Pelkonen O Lang MA Pasanen M Cocaine N-demethylation and the metabolism-related hepatotoxicity can be prevented by cytochrome P450 3A inhibitors Eur J Pharmacol 1994 270 35 43 8157080
Deutsch DG Tombler ER March JE Lo SH Adesnik M Potentiation of the inductive effect of phenobarbital on cytochrome P450 mRNAs by cannabidiol Biochem Pharmacol 1991 42 2048 2053 1741782 10.1016/0006-2952(91)90608-8
Cichewicz DL McCarthy EA Antinociceptive synergy between delta(9)-tetrahydrocannabinol and opioids after oral administration J Pharmacol Exp Ther 2003 304 1010 1015 12604676 10.1124/jpet.102.045575
Marchese G Casti P Ruiu S Saba P Sanna A Casu G Pani L Haloperidol, but not clozapine, produces dramatic catalepsy in {Delta}9-THC-treated rats: possible clinical implications Br J Pharmacol 2003
Khanna P Gupta MB Gupta GP Sanwal GG Ali B Influence of chronic oral intake of cannabis extract on oxidative and hydrolytic metabolism of xenobiotics in rat Biochem Pharmacol 1991 41 109 113 1986734 10.1016/0006-2952(91)90017-Y
Prigogine I From Being to Becoming 1980 San Fransisco: W.H. Freeman
Park B McPartland JM Glass M Cannabis, cannabinoids and reproduction Prostaglandins Leukot Essent Fatty Acids 2004 70 189 197 14683692 10.1016/j.plefa.2003.04.007
Lepore M Vorel SR Lowinson J Gardner EL Conditioned place preference induced by delta 9-tetrahydrocannabinol: comparison with cocaine, morphine, and food reward Life Sci 1995 56 2073 2080 7776834 10.1016/0024-3205(95)00191-8
Tanda G Munzar P Goldberg SR Self-administration behavior is maintained by the psychoactive ingredient of marijuana in squirrel monkeys Nat Neurosci 2000 3 1073 1074 11036260 10.1038/80577
Justinova Z Tanda G Redhi GH Goldberg SR Self-administration of Delta(9)-tetrahydrocannabinol (THC) by drug naive squirrel monkeys Psychopharmacology (Berl) 2003 169 135 140 12827345 10.1007/s00213-003-1484-0
Fattore L Cossu G Martellotta CM Fratta W Intravenous self-administration of the cannabinoid CB1 receptor agonist WIN 55,212-2 in rats Psychopharmacology (Berl) 2001 156 410 416 11498718 10.1007/s002130100734
Ameri A The effects of cannabinoids on the brain Prog Neurobiol 1999 58 315 348 10368032 10.1016/S0301-0082(98)00087-2
Gardner EL Addictive potential of cannabinoids: the underlying neurobiology Chem Phys Lipids 2002 121 267 290 12505706 10.1016/S0009-3084(02)00162-7
Marinelli M Piazza PV Interaction between glucocorticoid hormones, stress and psychostimulant drugs Eur J Neurosci 2002 16 387 394 12193179 10.1046/j.1460-9568.2002.02089.x
Di S Malcher-Lopes R Halmos KC Tasker JG Nongenomic glucocorticoid inhibition via endocannabinoid release in the hypothalamus: a fast feedback mechanism J Neurosci 2003 23 4850 4857 12832507
Whitlow CT Freedland CS Porrino LJ Functional consequences of the repeated administration of Delta(9)-tetrahydrocannabinol in the rat Drug Alcohol Depend 2003 71 169 177 12927655 10.1016/S0376-8716(03)00135-2
Marsicano G Wotjak CT Azad SC Bisogno T Rammes G Cascio MG Hermann H Tang J Hofmann C Zieglgansberger W Di Marzo V Lutz B The endogenous cannabinoid system controls extinction of aversive memories Nature 2002 418 530 534 12152079 10.1038/nature00839
De Vries TJ Homberg JR Binnekade R Raaso H Schoffelmeer AN Cannabinoid modulation of the reinforcing and motivational properties of heroin and heroin-associated cues in rats Psychopharmacology (Berl) 2003 168 164 169 12669182 10.1007/s00213-003-1422-1
Ledent C Valverde O Cossu G Petitet F Aubert JF Beslot F Bohme GA Imperato A Pedrazzini T Roques BP Vassart G Fratta W Parmentier M Unresponsiveness to cannabinoids and reduced addictive effects of opiates in CB1 receptor knockout mice Science 1999 283 401 404 9888857 10.1126/science.283.5400.401
Norwood CS Cornish JL Mallet PE McGregor IS Pre-exposure to the cannabinoid receptor agonist CP 55940 enhances morphine behavioral sensitization and alters morphine self-administration in Lewis rats Eur J Pharmacol 2003 465 105 114 12650839 10.1016/S0014-2999(03)01455-9
Vigano D Grazia Cascio M Rubino T Fezza F Vaccani A Di Marzo V Parolaro D Chronic morphine modulates the contents of the endocannabinoid, 2-arachidonoyl glycerol, in rat brain Neuropsychopharmacology 2003 28 1160 1167 12637958
Yamaguchi T Hagiwara Y Tanaka H Sugiura T Waku K Shoyama Y Watanabe S Yamamoto T Endogenous cannabinoid, 2-arachidonoylglycerol, attenuates naloxone-precipitated withdrawal signs in morphine-dependent mice Brain Res 2001 909 121 126 11478928 10.1016/S0006-8993(01)02655-5
Fattore L Martellotta MC Cossu G Mascia MS Fratta W CB1 cannabinoid receptor agonist WIN 55,212-2 decreases intravenous cocaine self-administration in rats Behav Brain Res 1999 104 141 146 11125733 10.1016/S0166-4328(99)00059-5
Balerio GN Aso E Berrendero F Murtra P Maldonado R Delta9-tetrahydrocannabinol decreases somatic and motivational manifestations of nicotine withdrawal in mice Eur J Neurosci 2004 20 2737 2748 15548217 10.1111/j.1460-9568.2004.03714.x
Basavarajappa BS Hungund BL Role of the endocannabinoid system in the development of tolerance to alcohol Alcohol Alcohol 2004 15550443
Le Foll B Goldberg SR Cannabinoid CB1 antagonists as promising new medications for drug dependence J Pharmacol Exp Ther 2004 15525797
Zammit S Allebeck P Andreasson S Lundberg I Lewis G Self reported cannabis use as a risk factor for schizophrenia in Swedish conscripts of 1969: historical cohort study BMJ 2002 325 1199 12446534 10.1136/bmj.325.7374.1199
Patton GC Coffey C Carlin JB Degenhardt L Lynskey M Hall W Cannabis use and mental health in young people: cohort study BMJ 2002 325 1195 1198 12446533 10.1136/bmj.325.7374.1195
Bersani G Orlandi V Kotzalidis GD Pancheri P Cannabis and schizophrenia: impact on onset, course, psychopathology and outcomes Eur Arch Psychiatry Clin Neurosci 2002 252 86 92 12111342 10.1007/s00406-002-0366-5
De Marchi N De Petrocellis L Orlando P Daniele F Fezza F Di Marzo V Endocannabinoid signalling in the blood of patients with schizophrenia Lipids Health Dis 2003 2 5 12969514 10.1186/1476-511X-2-5
Torrey EF Yolken RH The schizophrenia-rheumatoid arthritis connection: infectious, immune, or both? Brain Behav Immun 2001 15 401 410 11782106 10.1006/brbi.2001.0649
Jones-Brando L Torrey EF Yolken R Drugs used in the treatment of schizophrenia and bipolar disorder inhibit the replication of Toxoplasma gondii Schizophr Res 2003 62 237 244 12837520 10.1016/S0920-9964(02)00357-2
Fride E Ginzburg Y Breuer A Bisogno T Di Marzo V Mechoulam R Critical role of the endogenous cannabinoid system in mouse pup suckling and growth Eur J Pharmacol 2001 419 207 214 11426843 10.1016/S0014-2999(01)00953-0
Mechoulam R Spatz M Shohami E Endocannabinoids and neuroprotection Sci STKE 2002 2002 RE5 11972360
Krylatov AV Uzhachenko RV Maslov LN Bernatskaya NA Makriyannis A Mechoulam R Pertwee RG Sal'nikova OM Stefano JB Lishmanov Y Endogenous cannabinoids improve myocardial resistance to arrhythmogenic effects of coronary occlusion and reperfusion: a possible mechanism Bull Exp Biol Med 2002 133 122 124 12428277 10.1023/A:1015574100494
|
16179090
|
PMC1261530
|
CC BY
|
2021-01-04 16:36:50
|
no
|
Harm Reduct J. 2005 Sep 22; 2:17
|
utf-8
|
Harm Reduct J
| 2,005 |
10.1186/1477-7517-2-17
|
oa_comm
|
==== Front
Malar JMalaria Journal1475-2875BioMed Central London 1475-2875-4-421616474810.1186/1475-2875-4-42ResearchConcurrence of Plasmodium falciparum dhfr and crt mutations in northern Ghana Mockenhaupt Frank P [email protected] J Teun [email protected] Teunis A [email protected] Stephan [email protected] Rowland N [email protected] Robert W [email protected] Ulrich [email protected] Institute of Tropical Medicine, Charité – University Medicine Berlin, Spandauer Damm 130, 14050 Berlin, Germany2 Dept. of Medical Microbiology, Radboud University Nijmegen Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands3 Division of Infectious Diseases, Tropical Medicine and AIDS, Academic Medical Centre, PO Box 22660, 1100 DD Amsterdam The Netherlands4 Bernhard-Nocht-Institute for Tropical Medicine, Bernhard-Nocht-Strasse 74, 20359 Hamburg, Germany5 School of Medicine and Health Sciences, University for Development Studies, PO Box TL 1350, Tamale, Northern Region, Ghana2005 15 9 2005 4 42 42 13 5 2005 15 9 2005 Copyright © 2005 Mockenhaupt et al; licensee BioMed Central Ltd.2005Mockenhaupt 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
Both chloroquine (CQ) and sulfadoxine-pyrimethamine (SP) are failing drugs in much of sub-Saharan Africa. Previous findings suggest an association between resistance to CQ and to SP in vivo, in vitro, and on the molecular level.
Methods
In 126 Ghanaian children with uncomplicated malaria, associations between mutations conferring resistance in the Plasmodium falciparum dihydrofolate reductase (dhfr; SP) and chloroquine resistance transporter (crt; CQ) genes, concentrations of residual antimalarial drugs, and gametocyte carriage were examined.
Results
Mutant dhfr alleles and the CQ-resistance allele crt T76 were strongly associated with each other. Isolates exhibiting the dhfr triple mutation seven times more likely also contained crt T76 parasites as compared to isolates without the dhfr triple variant (P = 0.0001). Moreover, both, isolates with the dhfr triple mutation (adjusted OR, 3.2 (95%CI, 1.0–10.4)) and with crt T76 (adjusted OR, 14.5 (1.4–150.8)) were associated with an increased likelihood of pre-treatment gametocytaemia. However, crt T76 did not correlate with gametocytaemia following SP treatment and no selection of crt T76 in SP treatment failure isolates was observed.
Conclusion
These results confirm an association between CQ and SP resistance markers in isolates from northern Ghana. This could indicate accelerated development of resistance to SP if CQ resistance is already present, or vice versa. Considering the enhanced transmission potential as reflected by the increased proportion of isolates containing gametocytes when resistant parasites are present, co-resistance can be expected to spread in this area. However, the underlying mechanism leading to this constellation remains obscure.
==== Body
Background
Chloroquine (CQ) and sulfadoxine-pyrimethamine (SP) are the most frequently used antimalarial drugs in sub-Saharan Africa. With the spread and intensification of drug resistance of Plasmodium falciparum more effective combinatory drug regimes are now introduced in many African countries [1]. However, CQ and SP are cheap and widely available and will most likely be in use for some more years, particularly in home treatment [2-4].
In several East African countries where SP was introduced in response to intense CQ-resistance, the drug has gradually lost efficacy although the pace of this development is subject to controversy [5,6]. Previously, the core mutations linked with resistance to CQ and SP were found to be associated with each other in isolates from southern Ghana [7]. Although CQ and SP are structurally unrelated and the mutations conferring resistance are located on separate chromosomes [8-10], this finding suggests that parasites resistant to CQ may acquire resistance to SP more easily than sensitive ones, or vice versa. Studies in murine models [11] and in vitro [12,13] support this hypothesis. If this was true, the spread of CQ resistance could pave the way for an accelerated development of SP resistance, and thus, bear substantial importance to the health systems of affected regions. In addition, transmission potential may be increased in both SP and CQ resistant parasites [14,15].
Recently, mutations in the P. falciparum dihydrofolate reductase (dhfr) gene in northern Ghana were observed not only to be predictive for SP treatment failure but also to be associated with increased pre-treatment gametocyte carriage [16]. Here, associations between dhfr alleles, the core mutation in the P. falciparum chloroquine resistance transporter gene (crt T76), residual antimalarials, and gametocyte carriage in children with uncomplicated malaria were re-examined.
Methods
P. falciparum isolates were collected from children with uncomplicated malaria participating in a treatment trial in Tamale, Northern Region, Ghana, at the end of the rainy season 2002. The results of this trial and on the dhfr and dihydropteroate synthetase gene (dhps) patterns are described elsewhere [16,17]. In the present report, data from all 126 children with complete follow-up after SP treatment for whom dhfr and dhps were genotyped are analysed. The study protocol was approved by the Ethics Committee, Ministry of Health, Northern Region, and by the Health Research Unit, Ministry of Health, Accra, and parents' informed consent was obtained.
Patients were children aged 6–59 months with uncomplicated P. falciparum monoinfection (≥ 2,000 ≤ 200,000/μL) [18,19]. None had severe malnutrition, a febrile disease other than malaria, "danger signs" [18], or severe malaria [20]. Treatment consisted of a single dose of SP (25/1.25 mg/kg, Fansidar, Roche, Switzerland), and success was monitored according to current guidelines [17,19].
Venous blood was collected into EDTA. Asexual parasites and gametocytes were counted per ≥ 200 and 500 white blood cells, respectively, on Giemsa-stained thick blood films, and densities were calculated based on a putative mean WBC count of 8,000/μL. Gametocyte counts one week following treatment were available for 104 children. Pre-treatment levels of CQ and pyrimethamine in blood were measured by ELISA assays [21] with limits of detection of 10 ng/mL and 25 ng/mL, respectively.
P. falciparum dhfr, dhps, and crt alleles were assessed by restriction fragment length polymorphisms (RFLP) of amplicons generated by nested PCR assays applying hot start procedures (HotStart Taq, Qiagen, Germany). Primers and conditions are described elsewhere [22,23] as are restriction enzymes and RFLP conditions to characterize the codons dhfr, 16, 51, 59, 108, 164; dhps, 436, 437, 540, 581, 613; and crt, 76. Laboratory strains were used as controls for PCR and RFLP assays. Electrophoresis was perfomed on 3% GTG agarose (FMC Bioproducts, USA) gels.
Frequencies were compared by χ2-test or Fisher's exact test, and continuous variables by Student's t-test, Mann-Whitney U-test, or Kruskal-Wallis test as applicable. Logistic regression models were used to adjust for potential confounders of the presence of resistance mutations and to identify risk factors for gametocytaemia.
Results
The median age of the 68 girls and 58 boys was 28.5 months (range, 6–59). Median axillary temperature and geometric mean parasite density were 38.8°C (range, 37.5–41.0) and 36,358/μL (range, 2,068–175,333/μL), respectively. Gametocytes were found in 19% (24/126) of the children at a geometric mean density of 50/μL (16–1,120/μL). CQ in blood was observed in 66% (83/126; median, 32 ng/mL, range, 10–718) but pyrimethamine in only two children (78 & 326 ng/mL).
The dhfr alleles Thr-108, Val-16 and Leu-164 were not detected. Classifying mixed alleles as mutant, dhfr Ile-51 occurred in 56% (70/126), Arg-59 in 65% (82/126), and Asn-108 in 72% (91/126). Triple dhfr mutations (Ile-51/Arg-59/Asn-108) were seen in 47% (59/126), and quadruple (triple dhfr + dhps Gly-437) and quintuple (triple dhfr + dhps Gly-437/Glu-540) variants in 44% (55/126), and 0.8% (1/126), respectively. Crt T76 occurred in 80% (101/126) of the isolates.
The concomitant occurrence of dhfr, dhps, and crt alleles in identical isolates was analyzed. All dhfr mutations were linked with each other. Dhfr Ile-51 was associated with Asn-108 (OR, 29.8; 95%CI, 8.0–161, P < 0.0001) and Arg-59 (OR, 7.7; 95%CI, 3.3–19,5, P < 0.0001). All isolates exhibiting dhfr Arg-59 additionally had Asn-108 (P < 0.0001). Notably, in isolates with dhfr variant parasites, the CQ-resistance allele crt T76 occurred significantly more frequent than in the respective wildtype isolates (Tab. 1). Separating isolates into such with e.g. dhfr codon 108 pure wildtype, wildtype & mutant, and pure mutant, the prevalence of crt T76 was 67.5% (23/35), 82.4% (14/17), and 86.5% (64/74), respectively (χ2trend = 6.1; P = 0.01). Basically the same trend was seen for dhfr codons 51 (P = 0.002) and 59 (P = 0.002, data not shown). Vice versa, in isolates with crt wildtype, with both crt wildtype & T76, and crt T76 alone, the dhfr triple mutation was detected in 16% (4/25), 53% (9/17), and 55% (46/84), respectively.
Table 1 Associations between dhfr and crt mutations in 126 Ghanaian children with uncomplicated malaria
dhfr mutation Prevalence of crt T76 mutation (%) Odds ratio (95%CI) of crt T76 being present
Unadjusted P Adjusted for CQ in blood* P
51
absent 67.9 (38/56) 1 1
present 90.0 (63/70) 4.3 (1.5–12.5) 0.002 5.0 (1.8–13.8) 0.002
59
absent 63.6 (28/44) 1 1
present 89.0 (73/82) 4.6 (1.7–13.0) 0.0007 4.4 (1.7–11.4) 0.002
108
absent 65.7 (23/35) 1 1
present 85.7 (78/91) 3.1 (1.2–8.6) 0.01 3.2 (1.2–8.2) 0.02
triple
absent 68.7 (46/67) 1 1
present 93.2 (55/59) 6.3 (1.9–26.6) 0.0006 7.0 (2.2–22.5) 0.001
dhfr, dihydrofolate reductase gene, crt, chloroquine resistance transporter gene; 95%CI, 95% confidence interval; CQ, chloroquine; *, Potential confounders were tested for in logistic regression models including age, sex, parasite density, dhps alleles, and axillary temperature. Adjusted odds ratios include the presence of chloroquine in blood as the only factor found to be associated (univariate odds ratio, 3.2; 95%CI, 1.2–8.6; P = 0.01).
The prevalence of crt T76 increased with increasing number of any dhfr mutations: it was 63% (20/32) in isolates with pure dhfr wildtype, 74% (26/35) in isolates with single or double mutations, and 93% (55/59) in isolates exhibiting the dhfr triple mutation (χ2trend = 13.1; P = 0.0003). Adjusting for CQ in blood, isolates with dhfr triple mutations revealed a seven-fold increased odds of concomitantly exhibiting crt T76 (Tab. 1). Dhps alleles neither correlated with dhfr alleles nor with crt T76 (data not shown).
Dhfr, dhps, and crt alleles were examined with respect to pre-treatment gametocytaemia. As reported elsewhere [16], children were significantly more likely to harbour gametocytes in the presence of dhfr 51, 59, or triple mutations. This was also true for isolates exhibiting crt T76 (Tab. 2). Gametocytes were not observed in any (0/21) isolate with wildtype alleles for both dhfr and crt, but in 16% (8/50) of isolates exhibiting either dhfr or crt mutations, and in 29% (16/55) of isolates revealing both, dhfr triple and crt T76 (χ2trend = 8.7, P = 0.003). Adjusting for additional factors influencing gametocytaemia, i.e. high parasite density arbitrarily set as >50,000/μL, the presence of CQ in blood, and axillary temperature, both the dhfr triple mutation and crt T76 were associated with an increased likelihood of gametocytaemia (Tab. 2). Dhfr alleles or crt T76 were not associated with gametocyte density (data not shown).
Table 2 Associations with pre-treatment gametocytemia in 126 Ghanaian children with uncomplicated malaria
Factor Prevalence of gametocytemia (%) Odds ratio (95%CI) of gametocytemia
Unadjusted P Multivariate P
CQ in blood
None 27.9 (12/43) 1 1
Present 14.5 (12/83) 0.4 (0.2–1.2) 0.07 0.3 (0.1–1.0) 0.04
Ax. temperature n.a. 0.3 (0.1–0.6) 0.0008 0.2 (0.1–0.5) 0.0009
Parasite density
< 50,000/μL 26.3 (20/76) 1 1
≥ 50,000/μL 8.0 (4/50) 0.2 (0.1–0.8) 0.01 0.2 (0.1–0.8) 0.02
dhfr triple mutation
Absent 10.4 (7/67) 1 1
Present 28.8 (17/59) 3.5 (1.2–10.2) 0.009 3.2 (1.0–10.4) 0.049
crt 76 mutation
Absent 4.0 (1/25) 1 1
Present 22.8 (23/101) 7.1 (1.0–304) 0.04 14.5 (1.4–150.8) 0.02
Ax., axillary; n.a., not applicable
One week after treatment, gametocytes were observed in 63% (66/104) of the children. Gametocytaemia following SP treatment was associated with pre-treatment dhfr mutations, e.g. dhfr Asn-108, OR 3.1; 95%CI, 1.3–7.6. In children harbouring crt T76 parasites, gametocytes were only slightly more frequent (66% (54/82)) than in patients with crt wildtype parasites (55% (12/22), P = 0.3).
Finally, selection of crt T76 in SP treatment failures was examined. No selection was observed: In matched pairs of isolates obtained from 35 children suffering SP treatment failure, crt T76 was present in 86% (30/35) of pre-treatment isolates and in 80% (28/35) of isolates obtained at treatment failure.
Discussion
In this study on P. falciparum isolates from northern Ghana, two major findings are presented. First, SP and CQ resistance markers are strongly associated with each other, independent of residual antimalarials. Second, both are associated with an increased prevalence of gametocytes.
This study has several limitations and particularly the finding of an association between unrelated mutations needs caution in interpretation. Because polyclonal infections predominate in the area [24], the detection of a mutant allele in an isolate does not necessarily mean that all clones carry the mutation. Thus, it cannot be stated whether the linkage between the resistance markers is a true one, i.e. on the chromosomal level, or reflects co-occurrence in individual isolates. The limited number of crt wildtypes also hampered proper testing of a linkage disequilibrium. Although dhfr mutations were significantly more common in the presence of crt T76 this observation needs to be construed with caution since 80% of the isolates in this study harboured the latter variant. Also, it is impossible to comment whether the presence of crt T76 favours the presence of dhfr mutations or vice versa because this requires longitudinal observations. Likewise, the lack of temporal information impairs a clear statement on whether resistance mutations bring about increased gametocytogenesis. Due to methodology the resistance genotype of asexual parasites cannot be separated from that of gametocytes.
Despite these difficulties in drawing firm conclusions, several previous findings support the hypothesis of a linkage disequilibrium between mutations associated with SP resistance and CQ resistance. Early reports from the 1950s and 1960s indicated that in areas or patients with established pyrimethamine-resistance in Nigeria, Burkina Faso, and Venezuela, CQ exhibited a reduced activity (reviewed in [25]). In a murine malaria model, CQ resistance could be induced in pyrimethamine-resistant parasites but not in sensitive ones [11]. In isolates from Cameroon, the in vitro activity of pyrimethamine was ten times lower in CQ resistant than in sensitive P. falciparum [13]. Similar but less pronounced differences have also been observed in other studies [26,27]. In southern Ghana, we previously observed that the dhfr core mutation Asn-108 was three times more likely found in isolates exhibiting crt T76 than in isolates comprising crt wildtype parasites [7]. The reason for this apparent association between resistance to CQ and pyrimethamine or SP is obscure since both drugs have distinct modes of action and resistance to these is determined by mutations on different chromosomes [8-10]. One alluringly simple explanation could be that parasites in the study area have merely become resistant to both drugs, possibly as a result of drug pressure. In this regard, previous, simultaneous or sequential treatment with CQ and SP could have selected for resistant parasites which subsequently persisted for a longer period than the drugs can be detected in blood. However, pyrimethamine was seen in only two children which indicates that it is rarely used in this community and which argues against drug-induced selection for SP resistance. In addition, the results are corrected for the presence of CQ in blood which can be detected for approximately one month after intake [28]. This does not exclude the possibility of selected and persisting parasites but renders it rather unlikely. Alternatively, the association between crt T76 and dhfr mutations could reflect a rapid mutator phenotype with the ability of accelerated resistance to multiple drugs. This hypothesis originates from in vitro studies observing that parasites resistant to common antimalarials acquire resistance to structurally unrelated drugs more rapidly than susceptible strains. The genetic basis of this phenomenon is unknown but suggested to involve an increased frequency of mutations per se and consequently a higher probability of modified proteins which could also include drug targets [12]. In fact, selection for high-grade pyrimethamine resistance in vitro has been shown to enhance the degree of overall genomic polymorphism [29]. In this regard, it is noteworthy that crt T76 was more frequently observed with increasing number of dhfr mutations, i.e. with increasing degree of SP resistance.
In the present study, the association between the resistance markers meets with an increased presence of gametocytes in isolates comprising mutant dhfr or crt alleles. Gametocytaemia seemed to reflect a rather long duration of infection as can be deduced from its low prevalence in the presence of factors suggestive for acute disease, i.e. high body temperature and parasite density, and previous CQ treatment (Tab. 2). Again, an increased frequency of resistance mutations in gametocytaemic children could result from previous drug-related selection as outlined above. However, increased gametocytaemia preceding treatment has also been observed in infections subsequently found to be CQ resistant [30-32]. Hallett et al. [15] reported that in patients with crt T76 parasites, gametocyte density was highly increased one week following CQ treatment. In addition, gametocytes from patients carrying crt T76 parasites produced 38 times higher oocyst burdens in the mosquito as compared to crt wildtype parasites. In Tamale, both residual CQ levels and parasites with the crt T76 mutation are abundant [33], and a clear association of crt T76 and pre-treatment gametocyte prevalence is seen. Only one crt wildtype isolate contained gametocytes impairing a sound analysis of the effect of crt T76 on gametocyte density. However, increased gametocyte production by crt T76 parasites [15] in the presence of residual CQ could partially explain the present finding. The synthesis of both, crt and dhfr mutations being associated and increased gametocytaemia in their presence, gives rise to a grim scenario: Given that CQ resistant parasites have an improved transmission potential [15] the association with SP resistance would contribute to an accelerating spread of resistance to both drugs, particularly in areas where CQ resistance is frequent. In the present study, neither selection of crt T76 in SP treatment failure was observed nor a significantly elevated proportion of post-treatment gametocytaemia among children initially harbouring crt T76 parasites. Although this was not expected and would reflect an extraordinary rapid process, both observations may be influenced by the small sample size.
Conclusion
The present data provide evidence supporting a hypothesis on a connection between resistance to CQ and SP suggesting that both, CQ and SP resistance favour transmission. This needs to be verified by carefully designed longitudinal studies in regions of differing levels of drug resistance and endemicity. Per se, antimalarial treatment must be effective, and more effective than CQ and SP, not only to reduce treatment failures but also the transmission of potential co-resistance. Eventually, as this has been shown to counterbalance enhanced transmission of resistant parasites [15] the results strongly support combinatory treatment including artemisinine-derivatives.
Authors' contributions
FPM, JTB, and UB designed the study. RNO, SE and FPM were responsible for patient recruitment, clinical and parasitological examinations, and PCR assays. TAE measured drug concentrations. JTB and RWS did the gametocyte counts. FPM and JTB wrote the paper with major contributions of the other authors.
Acknowledgements
We thank the children who participated in this study and their parents as well as the members of the Northern Region Malaria Project (NORMAP). Part of this study was funded by the World Health Organization (EPH/CSR, M50-181-9).
==== Refs
World Health Organisation The use of antimalarial drugs Report of a WHO Informal Consultation 17-17 November 2000 WHO/CDS/RBM/200133 2000 World Health Organization: Geneva, Switzerland
Foster SD Pricing, distribution, and use of antimalarial drugs Bull World Health Organ 1991 69 349 363 1893512
Abuaku BK Koram KA Binka FN Antimalarial drug use among caregivers in Ghana Afr Health Sci 2004 4 171 177 15687069
Dzator J Asafu-Adjaye J A study of malaria care provider choice in Ghana Health Policy 2004 69 389 401 15276317 10.1016/j.healthpol.2004.01.005
Plowe CV Kublin JG Dzinjalamala FK Kamwendo DS Mukadam RA Chimpeni P Molyneux ME Taylor TE Sustained clinical efficacy of sulfadoxine-pyrimethamine for uncomplicated falciparum malaria in Malawi after 10 years as first line treatment: five year prospective study BMJ 2004 28 545 14757706 10.1136/bmj.37977.653750.EE
White N Sulfadoxine-pyrimethamine for uncomplicated falciparum malaria: sulfadoxine-pyrimethamine is not working in Malawi BMJ 2004 328 1259 15155511 10.1136/bmj.328.7450.1259
Mockenhaupt FP Eggelte TA Till H Bienzle U Plasmodium falciparum pfcrt and pfmdr1 polymorphisms are associated with the pfdhfr N108 pyrimethamine-resistance mutation in isolates from Ghana Trop Med Int Health 2001 6 749 755 11679121 10.1046/j.1365-3156.2001.00792.x
Wang P Read M Sims PF Hyde JE Sulfadoxine resistance in the human malaria parasite Plasmodium falciparum is determined by mutations in dihydropteroate synthetase and an additional factor associated with folate utilization Mol Microbiol 1997 23 979 986 9076734 10.1046/j.1365-2958.1997.2821646.x
Peterson DS Walliker D Wellems TE Evidence that a point mutation in dihydrofolate reductase-thymidylate synthase confers resistance to pyrimethamine in falciparum malaria Proc Natl Acad Sci USA 1988 85 9114 9118 2904149
Fidock DA Nomura T Talley AK Cooper RA Dzekunov SM Ferig MT Ursos LM Sidhu AB Naude B Deitsch KW Su XZ Wootton JC Roper PD Wellems TE Mutations in the P. falciparum digestive vacuole transmembrane protein PfCRT and evidence for their role in chloroquine resistance Mol Cell 2000 6 861 871 11090624 10.1016/S1097-2765(05)00077-8
Powers KG Jacobs RL Good WC Koontz LC Plasmodium vinckei: production of chloroquine-resistant strain Exp Parasitol 1969 26 193 202 5401455 10.1016/0014-4894(69)90112-X
Rathod PK McErlean T Lee PC Variations in frequencies of drug resistance in Plasmodium falciparum Proc Natl Acad Sci USA 1997 94 9389 9393 9256492 10.1073/pnas.94.17.9389
Ringwald P Eboumbou ECM Bickii J Basco LK In vitro activities of pyronaridine, alone and in combination with other antimalarial drugs, against Plasmodium falciparum Antimicrob Agents Chemother 1999 43 1525 1527 10348789
Mendez F Munoz A Carrasquilla G Jurado D Arevalo-Herrera M Cortese JF Plowe CV Determinants of treatment response to sulfadoxine-pyrimethamine and subsequent transmission potential in falciparum malaria Am J Epidemiol 2002 156 230 238 12142257 10.1093/aje/kwf030
Hallett RL Sutherland CJ Alexander N Ord D Jawara M Drakeley CJ Pinder M Walraven G Targett GA Alloueche A Combination therapy counteracts the enhanced transmission of drug-resistant malaria parasites to mosquitoes Antimicrob Agents Chemother 2004 48 3940 3943 15388456 10.1128/AAC.48.10.3940-3943.2004
Mockenhaupt FP Bousema JT Eggelte TA Schreiber J Ehrhardt S Wassilew N Otchwemah RN Sauerwein RW Bienzle U Plasmodium falciparum dhfr but not dhps mutations associated with sulfadoxine-pyrimethamine treatment failure and gametocyte carriage in northern Ghana Trop Med Int Health 2005 10 901 908 16135198 10.1111/j.1365-3156.2005.01471.x
Mockenhaupt FP Ehrhardt S Dzisi SY Bousema JT Wassilew N Schreiber J Anemana SD Cramer JP Otchwemah RN Sauerwein RW Eggelte TA Bienzle U A randomised, placebo-controlled, and double-blind trial on sulfadoxine-pyrimethamine alone or combined with artesunate or amodiaquine in uncomplicated malaria Trop Med Int Health 2005 10 668 671 15960705 10.1111/j.1365-3156.2005.01444.x
World Health Organization Assessment of therapeutic efficacy of antimalarial drugs for uncomplicated falciparum malaria in areas with intense transmission Document WHO/Mal961077 1996 World Health Organization: Geneva, Switzerland
World Health Organization Monitoring Antimalarial Drug Resistance Document WHO/CDS/CSR/EPH/200217 2002 World Health Organization: Geneva, Switzerland
World Health Organization Severe falciparum malaria Trans R Soc Trop Med Hyg 2000 94 S1 90 11103309 10.1016/S0035-9203(00)90300-6
Witte AM Klever HJ Brabin BJ Eggelte TA van der Kaay HJ Alpers MP Field evaluation of the use of an ELISA to detect chloroquine and its metabolites in blood, urine and breast-milk Trans R Soc Trop Med Hyg 1990 84 521 525 2091344 10.1016/0035-9203(90)90024-9
Duraisingh MT Curtis J Warhurst DC Plasmodium falciparum: detection of polymorphisms in the dihydrofolate reductase and dihydropteroate synthetase genes by PCR and restriction digestion Exp Parasitol 1998 89 1 8 9603482 10.1006/expr.1998.4274
Djimde A Doumbo OK Cortese JF Kayentao K Doumbo S Diourte Y Dicko A Su XZ Nomura T Fidock DA Wellems TE Plowe CV Coulibaly D A molecular marker of chloroquine-resistant falciparum malaria New Engl J Med 2001 344 257 262 11172152 10.1056/NEJM200101253440403
Mockenhaupt FP Ehrhardt S Eggelte TA Markert M Anemana S Otchwemah R Bienzle U Plasmodium falciparum multiplicity correlates with anaemia in symptomatic malaria Trop Med Int Health 2003 8 857 859 14516295 10.1046/j.1365-3156.2003.01117.x
World Health Organisation Chemotherapy of Malaria Tech Rep Ser 1961 226 31 39
Ringwald P Bickii J Basco LK In vitro activity of dihydroartemisinin against clinical isolates of Plasmodium falciparum in Yaounde, Cameroon Am J Trop Med Hyg 1999 61 187 192 10463665
Pradines B Tall A Rogier C Spiegel A Mosnier J Marrama L Fusai T Millet P Panconi E Trape JF Parzy D In vitro activities of ferrochloroquine against 55 Senegalese isolates of Plasmodium falciparum in comparison with those of standard antimalarial drugs Trop Med Int Health 2002 7 265 270 11903989 10.1046/j.1365-3156.2002.00848.x
Eggelte TA Navaratnam V, Payne D Production of monoclonal antibodies against antimalarial drugs for use in immunoassays The validation of chemical and immunological tests for antimalarials in body fluids 1990 Centre for Drug Research International Monograph Series 3, Universiti Sains Malaysia, Penang, Malaysia 35 63
Toteja R Nair L Bhasin VK Genome comparison of progressively drug resistant Plasmodium falciparum lines derived from drug sensitive clone Mem Inst Oswaldo Cruz 2001 96 427 433 11313656 10.1590/S0074-02762001000300025
Hess FI Iannuzzi A Leafasia J Cowdrey D Nothdurft HD von Sonnenburg F Löscher T Rieckmann KH Risk factors of chloroquine resistance in Plasmodium falciparum malaria Acta Trop 1996 61 293 306 8910654 10.1016/0001-706X(96)00011-3
Sokhna CS Trape JF Robert V Gametocytaemia in Senegalese children with uncomplicated falciparum malaria treated with chloroquine, amodiaquine or sulfadoxine + pyrimethamine Parasite 2001 8 243 250 11584755
Sowunmi A Fateye BA Changes in Plasmodium falciparum gametocytaemia in children with chloroquine-sensitive asexual infections Parasite 2003 10 363 369 14710634
Ehrhardt S Mockenhaupt FP Eggelte TA Agana-Nsiire P Stollberg K Anemana SD Otchwemah RN Bienzle U Chloroquine blood concentrations and molecular markers of chloroquine-resistant Plasmodium falciparum in febrile children in northern Ghana Trans R Soc Trop Med Hyg 2003 97 697 701 16117966 10.1016/S0035-9203(03)80106-2
|
16164748
|
PMC1261531
|
CC BY
|
2021-01-04 16:24:13
|
no
|
Malar J. 2005 Sep 15; 4:42
|
utf-8
|
Malar J
| 2,005 |
10.1186/1475-2875-4-42
|
oa_comm
|
==== Front
Malar JMalaria Journal1475-2875BioMed Central London 1475-2875-4-431616805410.1186/1475-2875-4-43ResearchRapid urban malaria appraisal (RUMA) I: Epidemiology of urban malaria in Ouagadougou Wang Shr-Jie [email protected] Christian [email protected] Thomas A [email protected] Penelope [email protected] Diallo A [email protected] Xavier [email protected] Natalie [email protected] Mathieu [email protected] Marcel [email protected] Swiss Tropical Institute (STI), P.O. Box, CH-4002 Basel, Switzerland2 Centre National de Recherche et de Formation sur le Paludisme, Ouagadougou (CNRFP), 01 B.P. 2208, Ouagadougou 01, Burkina Faso3 Ecole Inter-Etats d'Ingénieurs de l'Equipement Rural (EIER), 03 B.P. 7023, Ouagadougou 03, Burkina Faso2005 16 9 2005 4 43 43 10 6 2005 16 9 2005 Copyright © 2005 Wang et al; licensee BioMed Central Ltd.2005Wang 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
Rapid urbanization in sub-Saharan Africa has a major impact on malaria epidemiology. While much is known about malaria in rural areas in Burkina Faso, the urban situation is less well understood.
Methods
An assessment of urban malaria was carried out in Ouagadougou in November -December, 2002 during which a rapid urban malaria appraisal (RUMA) was applied.
Results
The school parasitaemia prevalence was relatively high (48.3%) at the cold and dry season 2002. Routine malaria statistics indicated that seasonality of malaria transmission was marked. In the health facilities, the number of clinical cases diminished quickly at the start of the cold and dry season and the prevalence of parasitaemia detected in febrile and non-febrile cases was 21.1% and 22.0%, respectively. The health facilities were likely to overestimate the malaria incidence and the age-specific fractions of malaria-attributable fevers were low (0–0.13). Peak prevalence tended to occur in older children (aged 6–15 years). Mapping of Anopheles sp. breeding sites indicated a gradient of endemicity between the urban centre and the periphery of Ouagadougou. A remarkable link was found between urban agriculture activities, seasonal availability of water supply and the occurrence of malaria infections in this semi-arid area. The study also demonstrated that the usage of insecticide-treated nets and the education level of family caretakers played a key role in reducing malaria infection rates.
Conclusion
These findings show that determining local endemicity and the rate of clinical malaria cases are urgently required in order to target control activities and avoid over-treatment with antimalarials. The case management needs to be tailored to the level of the prevailing endemicity.
==== Body
Background
An estimated 200 millions people live in urban malaria endemic areas in Africa [1] and a high proportion of clinical admissions in these areas are treated as malaria. Urban malaria poses a major challenge for health care systems in Africa. The impact of urbanization and uncontrolled population growth on malaria endemicity needs to be established [2-5].
Epidemiological profiles and clinical patterns are known to vary between urban, suburban and rural environments in Burkina Faso [6]. Sabatinelli et al. [7]collected blood samples from 2,117 children aged 0–5 years in Ouagadougou and reported prevalence rates of 3.0%, 9.5%, 20.0%, 15.6%, 21.8% and 26.4% in sectors 1, 8, 11, 14, 22 and 23, respectively. These findings showed a gradient of malaria incidence from the centre (lowest risk) to the areas close to an artificial lake (highest prevalence). Dabire [8] further categorized the residence of patients into 1) town centre, 2) the areas across the canals and 3) the shore areas of the artificial lake or dam (barrage), reporting prevalence rates of 13.1%, 25.3% and 31.4%, respectively.
Entomological research in Ouagadougou dates back to the colonial period. The main vectors Anopheles gambiae s.l., Anopheles funestus and Anopheles nili were already identified in Ouagadougou by Le Gac et al., in 1945 [9]. Later, a longitudinal entomological survey [10,11] reported six species of Anopheles sp. mosquitoes with An. gambiae s.l. and An. funestus playing a key role during the dry season. The focality of malaria transmission was also noted. Concerns were also addressed about the breeding of Anopheles sp. vectors in large water reservoirs and in the water supply resources [12].
A standard study protocol Rapid Urban Malaria Appraisal (RUMA) was developed in June 2002 based on a WHO proposal and an Environmental Health Project draft protocol [13,14]. RUMAs were commissioned by Roll Back Malaria (RBM) for three francophone countries (Ivory Coast, Burkina Faso and Benin) and one anglophone country (Tanzania). Each of the four assessment reports provides the following: an overview of the urbanization history, an estimate of the fractions of malaria-attributable fevers, parasite rates for different areas, an outline of health care services and highlights of the "lessons learned" from the survey. A separate overview introduces this work in a wider framework [15].
The study aimed to compile a minimum dataset on urban malaria features in Ouagadougou within a period of six to ten weeks and to display the malaria risk in relation to population settlements, social and health care services, as well as the environment. The study aimed to provide essential information to better plan malaria control interventions.
Methods
Study sites and sample selection
Ouagadougou is the capital of Burkina Faso, situated between latitude 12.0 N-13.0 N and longitude 1.15 E-1.40 E, 300 meters above sea level. To the north, the vegetation thins out into sand dunes as it approaches the Sahara. The total area of Ouagadougou was around 570–655 km2 in 2000 [16]. The annual precipitation is 750 to 900 mm. The rainy season is between June and October, the cold and dry season is between November and January and the hot and dry season is between February and May. The average temperature is approximately 19°C in January and 40°C from April to May. The total population in Ouagadougou was around 1,040,000 inhabitants in 2002 [17,18].
The sanitary administrative structure is not identical to the political administrative structure. There are five administrative districts, which are divided into 30 urban sectors and 17 peripheral villages (Figure 1a). The four sanitary districts are Pissy (sectors 1–12 and 16–19) Kossodo (sectors 13, 23–27), Paul VI (sectors 20–22) and Secteur 30 (sectors 14, 15 and 28–30). Ouagadougou were classified into three different areas (centre, intermediate and periphery), according to their population size, distance to the centre and physical characteristics. The patterns of development and settlement, i.e. commercial, industrial areas, residential areas and natural environments (lakeside, forest or dry areas), were considered. It is a rapid assessment with limited budget; therefore, in each area only one health facility and school were selected for the surveys. Health facilities with a higher volume of outpatients per day were considered for the survey.
Figure 1 a) Map of sanitary districts and sectors, and Anopheles sp. breeding sites. Numbers indicate sectors. b) Map of selected sites for schools and health facility-based surveys. c) Malaria prevalence by sectors of Ouagadougou. Health facilities surveys.
Centre: Due to low attendance in the urban dispensary (Dispensaire Urbain) which was originally selected, two sites were added to speed-up recruitment: Centre de Santé et de Promotion Sociale (CSPS) Paspange sector 12 and CSPS Dapoya sector 3 (Figure 1b). The selected dispensaries and the school, Paspanga C primary school (Figure 1b), are situated between the main dam and a busy commercial centre.
Intermediate areas: Dagnoin A primary school is situated in a poor residential area of sector 29, east of Ouagadougou (Figure 1b). An irrigation canal passes through this area. The missionary hospital St. Camille serves as a government Centre Médical avec Antenne Chirurgicale (CMA), with the highest attendance of patients in town.
Periphery: Kossodo A, B and C primary schools are situated in sector 26 at the north-east border behind the small forest Bois de Boulogne, which is a large farming area near an industrial zone. Centre Médical (CM) Kossodo is just opposite to the schools (Figure 1b). It is the only CM in Kossodo and it serves the majority of the population there.
To maximise case detection, the Centre National de Recherche et de Formation sur le Paludisme, Ouagadougou (CNRFP), situated in a busy area of the sector 4, near the University of Ouagadougou and CHN-YO, was identified as an additional site for the health facility-based fever survey (Figure 1b). It is not a health centre or clinic, but a national laboratory known for receiving self-referred malaria patients.
RUMA Methodology
Review of literature and collection of health statistics
The author surveyed the articles published on urban malaria and the relevant thesis presented in the medical libraries in Ouagadougou. Demographic, health information and routine malaria reports were collected from the CNRFP, the Institut National de la Statistique et de la Démographie (INSD) and the statistics unit of the municipal health department.
Mapping of breeding sites
A rapid entomological survey was conducted with the assistance of CNRFP entomologists and produced a map of Anopheles breeding sites. At the beginning of the rainy season in 2002, there had been an entomological survey in Ouagadougou conducted by CNRFP, to identify mosquito breeding sites. In December, 2002, the entomological team double-checked all water bodies which were identified earlier. Due to time limitations, additional potential breeding sites were not checked. All larvae were collected and transferred to water containers. The containers were filled with the water taken from the breeding sites and were labelled with their location. These were delivered to the entomological laboratory of the CNRFP for hatching. The mature adult mosquitoes were then identified to species level. Geographic coordinates were recorded for all confirmed breeding sites of Anopheles sp. by Global Positioning System (GPS) (Garmin_ eTrex 12 canal GPS). A vector layer of a digital map of Ouagadougou with the locations of all health facilities and schools was provided by the GIS unit of the Ecole Inter-Etats d'Ingénieurs de l'Equipement Rural (EIER).
School parasitaemia surveys
School parasitaemia surveys were carried out to estimate endemicity in different transmission areas (high, medium and low). The school surveys were carried out from November 21 to 30, 2002, at the cold and dry season. Each school was close to the health facility chosen for the fever survey (see below) and was visited by the survey team (sociology students-interviewers, nurses, laboratory technicians and a field assistant). A series of meetings was held with teachers and schoolmasters to explain the purpose and methodology of the survey. Participation was voluntary and parents had to fill out a questionnaire and sign the consent form (See Additional file 1). Only children who returned the consent forms had a blood sample taken and axillary temperature measured. Samples were collected from 200 students aged six to ten years in each school. Children were interviewed with the assistance of schoolteachers regarding their socio-economic situation and malaria infection histories. Both thin and thick blood films were taken on the same slide and stained by Giemsa stain. Based on the assumption that 8000 white blood cells are found in one ml of blood, parasite density in thick smears was defined as the number of parasites per 200 white blood cells. If 200 white blood cells were identified and less than 9 malarial parasites found, the process was continued until 500 white blood cells were identified.
Health facility fever surveys
This methodology aimed to assess malaria prevalence in fever cases and estimate the age-specific fractions of malaria-attributable fevers and improve the case definition and clinical diagnosis of malaria. In urban areas, an estimated 5 % to 50% of fever cases among children under 15 years old were due to malaria. A sample size of 200 in each facility gave an estimate of the proportion of cases with parasites with the following approximate lower 95% confidence limits: for 5%, lower 95% CI: 2; for 50%, lower 95% CI: 46. The survey activities lasted for 18 days from December 1st, 2002. Over this period, 200 fever cases and 200 non-fever controls in each health facility were interviewed (See Additional file 2). About 50% of the sample was aged ≤ 5 years. Outpatients with a history of fever (past 36 hours) or with a measured temperature of ≥ 37.5°C were defined as cases. Controls were recruited from another department of the same clinic without current or past fever, and matched by age and residency. Infants with congenital abnormalities, patients with signs of severe disease, patients returning to the health facility for follow-up visits and patients who were not permanent town residents for more than six months per year were excluded from the survey. After being recruited and giving informed consent, each patient had an axillary temperature measurement and had a blood film taken. The odds ratio (OR) is the proportion of odds of having parasitaemia in fever cases over controls. The formula for the fraction of fever episodes attributable to malaria parasites is: (1-1/Odds Ratio)*P with P being the proportion of fever episodes in which the subjects also had malaria parasites present.
For quality control, 200 slides were re-examined by a senior technician of the CNRFP and then a second time at the Swiss Tropical Institute (STI). The sensitivity, specificity and accuracy rates of readings were 98.7%, 98.2% and 98.6%, which was considered excellent.
Brief description of the health care system
The diversity of malaria fever management has a critical impact on the early diagnosis and treatment of malaria. Planning malaria control in urban areas cannot be achieved without an overview of the distribution, coordination and practices of existing treatment providers. This evaluation involved a meeting with representatives of the CNRFP, EIER and the municipal health department. The senior officers facilitated the exchange of information, in particular through lists and maps of public and private health care providers.
Statistical methods
The data were double-entered and validated in EpiInfo 6.04 (CDC Atlanta, USA, 2001). Data analysis was carried out in Stata 8 (Stata Corp. Texas, USA, 2003). Explanatory variables were age group, sex, residence, education level, axillary temperature, fever duration, bednet usage, urban agricultural activity, water resources and previous malaria infection within one month, as well as whether the patients had visited a rural area. The X2 test was applied to assess associations between categorical variables. Logistic regression was performed to assess the association between binary outcomes and explanatory variables, adjusted for the confounding effects of age groups.
Ethics
The Ethics Committee of the Ministry of Health of Burkina Faso and the research commission of the STI gave approval for the protocol. All the patients gave informed consent. In school surveys, the consent forms and questionnaires were delivered to the parents there days before the survey. Chloroquine (CQ) or amodiaquine (AQ) were paid if the patients presented fever signs.
Results
From 1945 to 2002, 39 papers regarding urban malaria epidemiology in Ouagadougou were published in international journals. The documentation of urban malaria research in Burkina Faso is well organised; most research work was published in thesis of resident doctors reporting and analyzing data from their clinical practices. Previous studies showed that the malaria peak prevalence occurred in September and October (30%), while 80–95% of malaria cases are reported during the rainy season (July to mid-October) and up to six weeks afterwards. The reports of clinical malaria dropped from December onwards to reach the lowest point (3%) in June [8,19]. The overall parasite index was 16% in Ouagadougou in August and September, 1984 [7]. From January to December, 1988, 6,109 outpatients in three health facilities were examined and found that malaria infections were the main reason for fever episodes with a prevalence rate of 33% [20].
Brief description of the health care system
The health care system in Burkina Faso is structured in five hierarchical levels (health post, health centre and district, regional and national hospitals). The public, private and voluntary services in Ouagadougou were heterogeneously distributed by district (Figure 2a). There were 80 public health facilities, 110 private health services (formations sanitaires privées) and 16 religious health facilities (formations confessionnelles). A total of 69 prescription pharmacies (officines pharmaceutiques) and 29 non-prescription drug outlets (Dépôts MEG) for essential medicines were registered in Ouagadougou in 2002 (Figure 2b). Over 60% of the prescription pharmacies were located in Pissy district (centre and south of city). Very few pharmacies were open in the intermediate and periphery areas. There was one dispensary (Centre de Santé et de Promotion Sociale) serving 23,800 inhabitants in Kossodo, 9,100 in Paul VI, 23,100 in Pissy and 12,400 in Secteur 30, respectively. There was no private health facility in Paul VI. This shows that the workload of public health services was heavy. Most of the inhabitants (76.0%, 83.4%, 72.4% and 64.2% in Kossodo, Paul VI, Pissy and Secteur 30, respectively) lived within 0–4 km of a public health facility, which means that the accessibility of services was good.
Figure 2 a) Categories and distribution of health services in Ouagadougou in 2002. Centre Hospitalier National (CHN)-National hospital. Centre Hospitalier Régional (CHR)-Regional hospital. Centre Médical (CM)-Health centre. Centre Médical avec Antenne Chirurgicale (CMA)-Health centre with an operating theatre. Centre de Santé et de Promotion Sociale (CSPS)-Dispensary and reproductive health unit. b) Distribution of pharmacies in Ouagadougou 2002. Dépôts MEG: Drug outlets. Officines pharmaceutiques: Prescription pharmacies.
Although the policy of privatization of healthcare services was launched more than 15 years ago, the private sector in Ouagadougou has not developed as much as in other SSA cities. However, many private paramedical practices and drug outlets are not authorized and registered and, therefore, the number of private health facilities was certainly underestimated. Over a third of all health facilities were owned by the government in 2002.
Results of malaria routine reports
Malaria morbidity and mortality data are reported from each health facility on a seasonal basis. The seasonal reports were available for 1999–2001, but not for 2002. The datasets were missing for the sanitary district Paul VI from October to December, 2001. The annual and seasonal patterns of mild and severe clinical malaria were marked (Figures 3 and 4). The highest incidence rates were reported from July to September and incidence rates went down from October to December. The lowest point was during the cold and dry season, from January to March. There was no big year-to year variation of reporting for simple malaria cases, while there was a sharp increase in reported complicated malaria during the rainy season in 2001. There was also a marked increase in reported cases in Pissy during the 1999 rainy season, for which no clear explanation could be found (Figure 4b).
Figure 3 Reported simple malaria cases in Ouagadougou, by sanitary district, 1999–2002. Adult ≥ 15 years. a) Kossodo b) Pissy c) Paul VI d) Secteur 30
Figure 4 Reported complicated malaria cases in Ouagadougou, by sanitary district, 1999–2002. Adult ≥ 15 years. a) Kossodo b) Pissy c) Paul VI d) Secteur 30
In 2001, there were 203,466 mild malaria cases (30–40% of all consultations) reported among 596,365 consultations in the public health facilities of Ouagadougou (Table 1). There were 20,071 complicated malaria cases reported, which accounted for 1.9–4.9% of total consultations in the different age groups and 10% of all malaria cases. The original seasonal malaria reports were classified by age and gender using the following categories: one year old, one to four years old, five to 15 years old, male and female from 15 years-old. For infants and children under five years, one third of clinical consultations were due to malaria and one third was due to diarrhoea and lower respiratory tract infections. For children aged 5–14 years, malaria (41.4%), skin problems and wounds (14.2%), as well as lower respiratory tract infections (9.9%) were the major causes of consultations.
Table 1 Reported malaria cases and top 3 major causes for clinical consultations in Ouagadougou 2001.
Age category Simple malaria Severe malaria LRTI¥ Skin/wounds Diarrhoea Total consultations
Infants < 1 year 31,430 (32.4%) 3,478 (3.6%) 15,348 (15.8%) 7,765 (8.0%) 16,306 (16.8%) 97,001
Children 1–4 years 58,070 (37.7%) 7,542 (4.9%) 20,592 (13.4%) 12,410 (8.0%) 20,338 (13.2%) 154,196
Children 5–14 years 38,283 (41.4%) 3,765 (4.1%) 9,120 (9.9%) 13,167 (14.2%) 3,424 (3.7%) 92,436
Adults ≥ years 75,683 (29.9%) 5,286 (2.1%) 20,862 (8.3%) 30,095 (11.9%) 10,884 (4.3%) 252,732
Total 203,466 (34.1%) 20,071 (3.4%) 65,922 (11.1%) 63,437 (10.6%) 50,952 (8.5%) 596,365
¥LTRI: Low Tract Respiratory Infection
School parasitaemia surveys
Plasmodium falciparum and Plasmodium malariae were detected. The presence of a malaria infection was found in 285 out of 590 valid samples (48.3 %, 95% CI: 44.2–52.4). The prevalence of parasitaemia was 31.6%, 37.6% and 73.1% in Paspanga C, Dognoin A and Kossodo A, B, C primary schools, respectively. Each school had its own catchments area, although the children attending the same school lived in different districts of Ouagadougou. The children's residence was further classified; the prevalence rates were 24.1%, 38.6% and 68.7% in the centre, intermediate and periphery areas, respectively (Figure 5).
Figure 5 Malaria and fever prevalence in school children by area of residency. School parasitaemia survey. Vertical bars represent 95%CI.
Health facility-based surveys
Overall, 123/437 fever cases (22.0%) and 110/436 (20.1%) of controls were positive. The majority of infections were due to P. falciparum; very few cases of P. malariae and Plasmodium vivax were identified. The difference in parasitaemia rates between facility-based surveys and school surveys was likely to be due to the different age distribution of the sample populations: the mean age was 7.7 years in the school parasitaemia surveys and 19.9 years in the facility-based fever surveys.
In all four cities where RUMA was used, the age groups were classified using following categories: one year old, one to five years old, six to 15 years old and adults from 15 years-old. Table 2 shows that in the above age groups the parasites rates in febrile episodes were 12.1%, 25.9%, 37.1% and 18.0%, respectively, while 14.3%. 14.4%, 34.5% and 19.8 % of controls were parasitaemic. The six to 15 year old group of children was at the highest risk of malaria infection compared to infants and adults. Age-specific differences were only significant in the fever group (Table 2). The OR of having malaria in fever cases varied from 0.82 to 2.07 for different age groups. The estimated fractions of malaria-attributable fevers were low: -0.03, 0.13, 0.04 and -0.02 in the above age groups, respectively. A separate paper will give an overview of the fraction of fever attributed to malaria.
Gradients of malaria prevalence and breeding site mapping
The study population was categorized by their residence rather than the location of health facilities. Malaria infections were nearly equally found in both febrile episodes and afebrile controls in the study (Figure 6). Overall, the centre and intermediate areas of Ouagadougou were at a lower risk of malaria (prevalence: 14.1% and 20.7%) compared to the periphery of Ouagadougou (25.9%). Higher malaria infections were found among the residents from the periphery sectors 18 and 26 (Figure 1c). There was no malaria infection among the residents of sectors 8–10, where the government and business centres are located. The parasite rates in peripheral sectors such as 15–17 and 20 were surprisingly low, but this could be due to a small sample population.
Figure 6 Malaria prevalence in fever cases and control groups by residential areas of patients. Health facility-based surveys. Vertical bars represent 95% CI.
In total, eight sites at the periphery and two sites in the centre were identified where Anopheles sp. larvae production was ongoing (Figure 1a). In the city centre, one Anopheles sp. breeding site was found in sector 8 where the canal passes and one in puddles/pools around artificial lake No. 2 in sector 11. In the periphery, the major open water bodies with productive Anopheles breeding were found in sectors 18, 19, 20, 21, 23, 28 and 30 (Figure 1a). The small temporary breeding sites in household compounds were not checked.
Socio-economic factors
The socio-economic status of study population was heterogeneous: over 70% of those in the city centre and around 50% of those in the intermediate and peripheral areas had at least a primary school level education. The proportion of households with tap water supply in the centre was higher (47.0%) than in the periphery (31.9%). Over 50% of the population relied on public fountains for water supply. There was no big difference in housing materials in the different areas (53.0–63.6% of houses were built with concrete and bricks and 3–5% were built with mud). An average of 42.2% of study population used a bednet the night before survey. The residents of the centre (48.8%) were more likely to sleep under a bednet than those in the intermediate (36.2%) and periphery areas (38.6%). This difference was significant (intermediate areas: OR = 0.59, 95% CI = 0.42–0.84, P < 0.001, periphery: OR = 0.66, 95% CI = 0.50–0.87, P < 0.001). More ITNs were present in households in the city centre (10.4%) than in the intermediate (7.5%) and the periphery areas (7.1%).
A logistic regression model was performed to estimate the association between socio-economic factors and malaria infection, adjusted for the effects of residential areas and age groups (Table 3). The risk of malaria was significantly reduced for people sleeping under a bednet (OR = 0.74, 95% CI = 0.54–1.00, P < 0.05). Neither education levels of the caretakers nor housing material were identified as significant risk factors. Having an urban agricultural land/garden near living compounds was positively associated with a malaria infection (OR = 1.39, 95% CI = 1.01–1.92, P < 0.05). Increased risk of malaria was associated with exposure to open water bodies, like fountains (OR = 1.66, 95% CI = 1.19–2.31, P < 0.005) or streams (OR = 2.80, 95% CI = 1.30–6.04, P < 0.005). Travelling to rural areas within 90 days was not correlated with the presence of parasitaemia.518 (47.5%) patients were treated for malaria within one month prior to the survey: 62.0% were treated at health centres and hospitals, while 35.0 % were self-treated at home or underwent no treatment. Very few people reported purchasing drugs in a pharmacy (2.2%) or using traditional medicine (0.8%).
Table 2 Odds ratio (OR) of having parasitaemia by age groups and fever/control groups. Health facility-based surveys.
Malaria prevalence Fever Controls Fever Controls
Age groups OR 95% CI P value OR 95% CI P value
Infants 0–1 year 7/58 (12.1%) 3/21 (14.3%) 1 - - 1 - -
Children 1–5 years 45/174 (25.9%) 15/104 (14.4%) 2.54 1.08–6.00 <0.05 1.01 0.27–3.86 0.987
Children 6–15 years 23/62 (37.1%) 20/58 (34.5%) 4.30 1.67–11.03 <0.005 3.16 0.83–12.02 0.092
Adults >15 years 48/266 (18.0%) 72/363 (19.8%) 1.60 0.69–3.75 0.276 1.48 0.43–5.18 0.535
Table 3 Socio-economic factors and the risk of malaria infection by logistic regression model. Health facility-based surveys.
Socio-economic factors % OR 95% CI P value
Adjusted for the effects of residential areas and age groups
Education
Primary 23.2% 1 - -
Secondary 33.4% 0.97 0.62–1.49 0.873
Superior 5.2% 0.96 0.44–2.09 0.911
No education 35.5% 1.3 0.85–1.98 0.222
Religious 2.6% 0.74 0.24–2.27 0.594
Housing material
Concrete/brick 58.1% 1 - -
Leaf/mud 4.6% 1.61 0.82–3.19 0.17
Leaf 0.8% 2.13 0.50–9.00 0.304
Others 36.5% 1.45 1.06–1.98 < 0.05
Water supply resource
Tap water 38.1% 1 - -
Well 0.6% 1.58 0.18–13.90 0.68
Fountain 58.1% 1.66 1.19–2.31 < 0.005
Others 3.2% 2.8 1.30–6.04 < 0.005
Living near a garden or agriculture land
No 71.0% 1 - -
Yes 29.0% 1.39 1.01–1.92 < 0.05
Adjusted for the effects of different residential areas
Bednet usage
No use 58.0% 1 - -
Used 42.0% 0.74 0.54–1.00 < 0.05
Without adjusting for residential areas and age groups
Rural exposure within 90 days
No 91.3% 1 - -
Yes 8.7% 1.14 0.70–1.90 0.6
Previous malaria treatment within 30 days with the presence of parasitaemia
No 52.5% 1 - -
Yes 47.5% 1.1 0.82–1.48 0.5
Monitoring of parasite resistance to anti-malarial drugs
The evolution of drug resistance of malaria in Burkina Faso has been described in two urban areas, Ouagadougou and Bobo-Dioulasso and is summarized in Table 4[21,22].
Table 4 Susceptibility of P. falciparum to antimalarials in Burkina Faso.
Year Drugs tested Study sites Urban/Rural Authors Failure rate In vivo
1982–1986 CQ (in vitro & in vivo) Koudougou Urban [36] First case found
1988–1989 CQ Koudougou
Zaghtouli
Dori
Banfora
Fada N'Gourma Urban
Rural
Urban
Rural
Rural [37] 25%
1988 CQ Zaghtouli Rural [38] 18.7%
1989 CQ Dapelgo Rural [38] 20.2%
1982–1991 CQ, SP, quinine, MP (in vitro) Ouagadougou Bobo-Dioulasso Urban [39] 6–15.8 %
1990–1992 CQ, SP (in vitro), quinine, halofantrine hydrochloride MP, Ouagadougou and its neighbouring villages Urban
Rural [22, 40-42] CQ & SP: 8.1–24.4%
Others: 0%
1993 CQ Ouagadougou Urban [43] 25%
1995–1996 CQ, quinine, MP (in vitro) Bobo-Dioulasso Urban [44] CQ: 19–20%
M:2–9.6%
1992–1998 CQ, AQ, quinine, halofantrine MP Ouagadougou Urban [45] AQ:4.3% & 2.2 % in 1997
CQ:8.5% in 1992,
CQ:20% in 1994
H: 7.9% in 1997 (in vitro)
MP: 0% in 1997 (in vitro)
Q: 0.9% in 1995
MP+H:7.6% in 1997
1999–2002 CQ & SP Bobo-Dioulasso Urban [46] CQ:18%
SP: < 1%
AQ: Amodiaquine
CQ: Chloroquine
MP: Mefloquine
SP: Sulfadoxine/pyrimethamine
Discussion and conclusion
RUMA methodology
The RUMA methodology is a cross-sectional study and the results may be different at another time of the year, or even vary between years.
Valuable information was extracted from the existing scientific literature and from health statistics. However, the research highlighted the need to enhance the capacity of municipal health department in collecting, processing, disseminating and using information. Some information needed for the evaluation of the health care system, such as the map with public/private health service providers and the schools had been prepared earlier by the GIS unit of EIER. A malaria map of breeding sites was produced during the cold and dry season mainly due to the help of CNRFP. They conducted an entomological survey in Ouagadougou at the beginning of the rainy season in 2002. Without such help, mapping breeding sites would not be possible in the frame of a RUMA.
Mis-diagnosis of malaria
One-third of all clinical consultations were diagnosed as malaria cases and many of them are likely to be mis-diagnosed. Fever is no longer an indicative sign for the diagnosis of malaria As a result, a review of clinical guideline for the management of fever episodes is necessary in order to reduce over-treatment.
A heterogeneity of clinical signs of severe malaria was observed between the paediatric wards of the main hospital, the Centre Hospitalier National Yalgado Ouédraogo (CHN-YO) and a rural district hospital near Ouagadougou during the rainy season [6]. In town, the age distribution and the clinical spectrum of severe malaria were related to the place of residence of the patients. While Sanou [23] demonstrated that malaria remained a major cause of childhood morbidity and mortality in the main hospital in Ouagadougou during the rainy season 1993–94, the mis-diagnosis of malaria was reported to be close to 34.7% [24]. Dabire in 1990 showed that 63.4% of malaria cases in CHN-YO defined on clinical criteria alone were not parasitaemic [8]. The severity and diversity of malaria symptoms reflect the diversity of local malaria endemicity, leading to difficulties for malaria diagnosis. Presumptive treatment of malaria based uniquely on the fever sign leads to high rates of over-diagnosis and over-treatment. Hence, it is important to introduce diagnosis tests, among which rapid tests are very promising. If physicians then exclude malaria they should check for other causes of fever and initiate an appropriate therapy. All the cases confirmed by the rapid diagnosis test should be treated as malaria since the transmission level is not high.
Parasitaemia results
In the school surveys, the parasitaemia rates were always higher than the febrile episodes since many children have asymptomatic infection. It was also observed that the community prevalence remained high at the beginning of the cold and dry season, while clinical malaria cases diminished quickly in health facilities. The control group had an even higher prevalence of parasitaemia than the fever group. The difference between parasitaemia rates in active and passive case detection was expected. The health facilities received patients from various communities, while school children mostly came from the same areas. This indicated that malaria infections were clustered in certain areas and that there were different levels of malaria transmission in Ouagadougou. Another explanation could be that some feverish patients may have taken paracetamol or another anti-pyretic and hence might not necessarily have presented with fever when visiting the dispensary.
Urban agriculture and Anopheles vector breeding sites
The risk of a malaria infection in Ouagadougou might be associated with seasonal agricultural activities. A Ouagadougou home gardening map produced by Cissé and Gerstl showed that a majority of home gardens were situated at the city's outskirts and along the shores of artificial lakes and canals [25-27]. The irrigated plots along the dam and the water channels create rural enclaves within the city. A study conducted near Ouagadougou described a positive association between irrigated farming, the malariometric and malaria morbidity among children below five years of age [28]. In the health facility-based surveys, it was found that the water supply types (fountains or streams) and the proximity to home gardening fields were associated with malaria infection. This association was noted as well in Uganda [29] and Dar es Salaam [30].
Malaria clusters in Ouagadougou
The parasitaemia prevalence rates in schools varied widely from the city centre to the periphery. The heterogeneity of endemicity within a small distance between the urban centre and the periphery of Ouagadougou was confirmed. Different levels of malaria transmission were previously found to be related to the spatial and temporal distribution of An. gambiae larval breeding sites in Ouagadougou [11,31]. Rossi et al. and Petrarca et al. concluded that higher prevalence rates of malaria occurred in areas where larvae breeding sites were semi-permanent. The most significant semi-permanent breeding sites are along the artificial lake, particularly in sectors 19–23. This was consistent with the observations mentioned above: higher prevalence rates were found along artificial lakes and canals, which are associated with urban agricultural activities.
Awareness and practices of malaria prevention
ITNs coverage was not high, possibly due to economic constraints and availability problems. Annual household expenses for malaria treatment were estimated at 38,398 CFA [24]. The average cost of malaria treatments was 4,929 CFA per person in 1992 [32]. The average hospitalisation costs for complicated malaria have risen up to 21,160 CFA in the paediatric ward of CHN-YO [33]. Because of such high cost, the patients reported taking self-medication before visiting health facilities (35%) or increasingly invested in preventing mosquito bites (42.2%). These results agreed with the findings of local researchers who found 28–36% [7], 30.2% [34] and 50% [19] self-medication.
Some suggestions can be made concerning in-depth research and interventions. Firstly, emphasis urgently needs to be put on the proper diagnosis of "malaria cases", including the use of rapid tests [35]. Secondly, Ouagadougou has a huge artificial lake for its water supply and irrigation. It would be important to explore the malaria impact of current hydro-agriculture systems and urban agriculture activities during the cold and dry season and to find ways to mitigate their malaria potential. Thirdly, the cost for malaria treatment is relatively high in Burkina Faso, which leads to a high proportion of self-medication. The pricing policy in public health facilities should be revised as it should be considered an obstacle to case management and malaria control goals.
List of abbreviations
AQ Amodiaquine
CHN Centres Hospitalier National
CHN-YO Centre Hospitalier National Yalgado Ouédraogo (CHN-YO)
CHR Centre Hospitalier Régional
CM Centre Médical
CMA Centre Médical avec Antenne Chirurgicale
CNRFP Centre National de Recherche et de Formation sur la Paludisme, Burkina Faso
CSPS Centre de Santé et de Promotion Sociale
CQ Chloroquine
INSD Institut National de la Statistique et de la Démographie
EIER Ecole Inter-Etats d'Ingénieurs et de l'Equipement Rural, Burkina Faso
GPS Global Positioning System
GIS Geographic Information System
ITNs Insecticide-Treated Nets
MP Mefloquine
OR Odds Ratio
RUMA Rapid Urban Malaria Appraisal
SP Sulfadoxine/pyrimethamine
STI Swiss Tropical Institute
Authors' contributions
SW participated in the design of the study, conducted the field work, analysed and interpreted data, drafted and revised the manuscript. CL conceived the study, coordinated the field work and revised the manuscript. TS and PV participated in the design and statistical analysis. DD, XP and ES managed and supervised the data collection and laboratory work in the field. MK supervised the production of the breeding site map and the cleaning of this dataset. MT participated in the conception of the work, revised it critically at different stages and gave final approval of the version to be published.
Note
Table 2. Odds ratio (OR) of having parasitaemia by age groups and fever/control groups. Health facility-based surveys.
Supplementary Material
Additional File 1
School survey questionnaire: the questionnaire for school parasitaemia survey.
Click here for file
Additional File 2
Hospital survey questionnaire: the questionnaire for health facility-based survey.
Click here for file
Acknowledgements
The study was funded by the Roll Back Malaria Partnership and STI. The authors are grateful to schools and selected health facilities in Ouagadougou for their warm acceptance and to the team of CNRFP and EIER for the great technical support in the field. We thank the INSD of Burkina Faso for allowing us to access the information. Special thanks go to Dan Anderegg and Dr. Andrei Chirokolava for editing and reviewing this manuscript.
==== Refs
Keiser J Utzinger J Caldas de Castro M Smith TA Tanner M Singer BH Urbanization in sub-saharan Africa and implication for malaria control Am J Trop Med Hyg 2004 71 118 27 15331827
Donnelly MJ McCall PJ Lengeler C Bates I D'Alessandro U Barnish G Konradsen F Klinkenberg E Townson H Trape JF Hastings IM Mutero C Malaria and urbanization in sub-Saharan Africa Malar J 2005 4 12 15720713 10.1186/1475-2875-4-12
Omumbo JA Guerra CA Hay SI Snow RW The influence of urbanisation on measures of Plasmodium falciparum infection prevalence in East Africa Acta Trop 2005 93 11 21 15589793 10.1016/j.actatropica.2004.08.010
Hay SI Guerra CA Tatem AJ Atkinson PM Snow RW Urbanization, malaria transmission and disease burden in Africa Nat Rev Microbiol 2005 3 81 90 15608702 10.1038/nrmicro1069
Robert V Macintyre K Keating J Trape JF Duchemin JB Warren M Beier JC Malaria transmission in urban sub-Saharan Africa Am J Trop Med Hyg 2003 68 169 76 12641407
Modiano D Sirima BS Sawadogo A Sanou I Pare J Konate A Pagnoni F Severe malaria in Burkina Faso: urban and rural environment Parassitologia 1999 41 251 4 10697864
Sabatinelli G Bosman A Lamizana L Rossi P Prevalence of malaria in Ouagadougou and the surrounding rural environment during the period of maximal transmission Parassitologia 1986 28 17 31 3455530
Dabire E Morbidité et mortalité palustres au sein de la pathologie fébrile dans le service de pédiatrie de l'hôpital Yalgado Ouédraogo MSc thesis 1990 Ouagadougou: Université de Ouagadougou 110
Le Gac P Seite P Combescot de Marsaguet G Etude sur le paludisme à Ouagadougou Bulletin de Extique la Societe de Pathologie Extique 1945 217 220
Bosman A Sabatinelli G Lamizana L Further observations on chemoprophylaxis and prevalence of malaria using questionnaire data in urban and rural areas of Burkina Faso Parassitologia 1988 30 257 62 3271989
Rossi P Belli A Mancini L Sabatinelli G A longitudinal entomologic survey on the transmission of malaria in Ouagadougou (Burkina Faso) Parassitologia 1986 28 1 15 3455529
Parent G Ouedraogo A Zagre NM Compaore I Kambire R Poda JN Large dams, health and nutrition in Africa: beyond the controversy Sante 1997 7 417 22 9503500
Warren M Billing P Bendahmane D Wijeyaratne P Malaria in urban and peri-urban areas in sub-Sahara Africa. Environmental Health Project, activity report No.71 1999 Washington, DC 44 10358806
WHO Rapid Urban Malaria Appraisal 2001 Geneva
Wang S-J Lengeler C Smith TA Vounatsou P Cisse G Diallo DA Akogbeto M Mtasiwa D Teklehaimanot A Tanner M Rapid urban malaria appraisal (RUMA) in sub-Saharan Africa Malar J submitted to
Institut National de la Statistique et de la Démographie (INSD) L'enquête Démographique et de Santé du Burkina Faso (EDSBF-II) 2000 Ouagadougou
Municipale de Ouagadougou Commune de Ouagadougou 2004
World Bank Upgrading of low income settlements, country assessment report, Burkina Faso 2002 Ouagadougou 29
Coulibaly SO La part du paludisme dans les affections fébriles en milieu urbain de Ouagadougou. Etude menée dans trois dispensaires de la ville 1989 Ouagadougou: Université de Ouagadougou
Coulibaly CO Guiguemde TR Lamizana L Ouedraogo JB Dabiret E The role of malaria in febrile diseases in the urban environment of Ouagadougou (Burkina Faso, West Africa) Ann Soc Belg Med Trop 1991 71 5 10 2043001
Aouba AE Analyse de la situation de la chimiorésistance du paludisme au Burkina Faso: conséquences thérapeutiques 1992 Ouagadougou:Université de Ouagadougou
Del Nero L Lamizana L Nebie I Pietra V Susceptibility of Plasmodium falciparum to chloroquine and mefloquine in Ouagadougou area (Burkina Faso) Parassitologia 1993 35 1 4 8065814
Sanou I Pare J Traore S Modiano D Kam KL Kabore J Lamizana L Sawadogo SA Guiguemde TR Clinical signs of severe malaria in a pediatric hospital in Ouagadougou Santé 1997 7 13 7
Coulibaly N Etude de l'impact économique du paludisme et du niveau de l'immunité humorale des maladesen zone rurale de Bobo-Dioulasso (Burkina Faso) 1993 Ouagadougou: Université de Ouagadougou
Cissé G Impact sanitaire de l'utilisation d'eaux usés et polluées en agriculture urbaine: Cas du maraîchage à Ouagadougou (Burkina Faso) thèse de doctorat No 1639 1997 Lausanne: Ecole Polytechnique Federale de Lausanne 267
Cissé G Kientga M Ouédraogo B Tanner M Développement du maraîchage autour des eaux de barrage à Ouagadougou: quels sont les risques sanitaires à prendre en compte? Cahiers Agricultures 2002 11 31 38
Gerstl S The economic costs and impact of home gardening in Ouagadougou, Burkina Faso PhD thesis 2001 Basel: University of Basel
Zamane H Indices paludométriques et morbidité palustre chez les enfants de 0 à 59 mois dans les hydroaménagements agricoles de la vallée du Sourou en 1999 (Burkina Faso) 2000 Ouagadougou: Université de Ouagadougou
Staedke SG Nottingham EW Cox J Kamya MR Rosenthal PJ Dorsey G Short report: proximity to mosquito breeding sites as a risk factor for clinical malaria episodes in an urban cohort of Ugandan children Am J Trop Med Hyg 2003 69 244 6 14628938
Caldas de Castro M Yamagata Y Mtasiwa D Tanner M Utzinger J Keiser J Singer BH Integrated urban malaria control: a case study in Dar es Salaam, Tanzania Am J Trop Med Hyg 2004 71 103 17 15331826
Petrarca V Petrangeli G Rossi P Sabatinelli G Antimalarial campaign program in Ouagadougou (Burkina Faso): the Anopheles gambiae complex in the city of Ouagadougou and surrounding villages Ann Ist Super Sanita 1986 22 189 91 3752794
Dajoari M Contribution à l'évaluation de l'impact économique du paludisme chez les travailleurs d'entreprises: Etude menée à l'Office de santé des travailleurs de Bobo-Dioulasso 1992 Ouagadougou: Université de Ouagadougou; 111
Sanon VM Etude du coût financier direct de la prise en charge du paludisme grave en milieu pédiatrique de Ouagadougou 1999 Ouagadougou: Université de Ouagadougou 10028380
Traore S Formes graves de paludisme au service de pédiatrie du centre hospitalier national Yalgado Ouédraogo: Aspects épidémiologiques et cliniques 1994 Ouagadougou: Université de Ouagadougou
Othnigué N Wyss K Tanner Marcel Genton Blaise Urban malaria in the Sahel: Prevalence and seasonality of presumptive malaria and parasitaemia at primary care level in Chad submitted to the journal
Baudon D Devoucoux R Roux J Sondo B Sensitivity of Plasmodium falciparum to chloroquine in the hyperendemic malaria savannah zone of Burkina Faso ex-Upper Volta. Use of in vivo and in vitro tests. Demonstration of a resistant strain in vitro Bull Soc Pathol Exot Filiales 1984 77 658 65 6395988
Pietra V Lamizana L del Nero L Rotigliano G In vivo chemosensitivity of Plasmodium falciparum to chloroquine in Burkina Faso: development of resistance 1988–1990 Parassitologia 1992 34 131 4 1339966
Kabore B Surveillance de l'efficacité therapeutique de la chloroquine a Ouagadougou en 1989–1990 1991 Ouagadougou: Université de Ouagadougou
Guiguemde TR Aouba A Ouedraogo JB Lamizana L Ten-year surveillance of drug-resistant malaria in Burkina Faso (1982–1991) Am J Trop Med Hyg 1994 50 699 704 8024062
Del Nero L Lamizana L Pietra V Rotigliano G A national survey of the prevalence of chloroquine resistant Plasmodium falciparum in Burkina Faso J Trop Med Hyg 1993 96 186 90 8505775
Del Nero L Lamizana L Nebie I Sare S Bougouma L Pietra V In vivo sensitivity of Plasmodium falciparum to halofantrine hydrochloride in Burkina Faso Am J Trop Med Hyg 1994 50 102 6 8304564
Del Nero L Nebie I Soudouem G Pietra V Chloroquine and sulfadoxine/pyrimethamine sensitivity in Burkina Faso. In vivo sensitivity of Plasmodium falciparum to chloroquine and sulfadoxine/pyrimethamine in Burkina Faso Trop Geogr Med 1994 46 8 10 8165742
Nabalma S Relation paludisme grossesse chez la femme enceinte à Ouagadougou: Intérêt d'une méthode de surveillance de chloroquino-résistance de P. falciparum reposant sur l'apposition placentaire couplée au dosage de chloroquinémie 1994 Ouagadougou: Université de Ouagadougou
Ouedraogo A Etude in vivo de l'activité antiplasmodique de l'extrait hydroalcoolique de Gardenia Sokotensis Hutch (Rubiaceae) chez la souris NMRI infestée par Plasmodium bergheio (Burkina Faso) 1998 Ouagadougou: Université de Ouagadougou
Djire AA Analyse de l'évolution de la chimiorésistance du paludisme du Burkina Faso de 1992–1998 1999 Ouagadougou: Université de Ouagadougou
Tinto H Zoungrana E Coulibaly S Ouedraogo J Traore M Guiguemde T Van Marck E D'Alessandro U Chloroquine and sulphadoxine-pyrimethamine efficacy for uncomplicated malaria treatment and haematological recovery in children in Bobo-Dioulasso, Burkina Faso during a 3-year period 1998–2000 Trop Med Int Health 2002 7 925 30 12390597 10.1046/j.1365-3156.2002.00952.x
|
16168054
|
PMC1261532
|
CC BY
|
2021-01-04 16:24:13
|
no
|
Malar J. 2005 Sep 16; 4:43
|
utf-8
|
Malar J
| 2,005 |
10.1186/1475-2875-4-43
|
oa_comm
|
==== Front
Malar JMalaria Journal1475-2875BioMed Central London 1475-2875-4-461617908910.1186/1475-2875-4-46ResearchA randomized trial of artemether-lumefantrine versus mefloquine-artesunate for the treatment of uncomplicated multi-drug resistant Plasmodium falciparum on the western border of Thailand Hutagalung Robert [email protected] Lucy [email protected] Elizabeth A [email protected] Rose [email protected] Alan [email protected] Kaw L [email protected] Pratap [email protected] Thomas [email protected] Nicholas J [email protected] François H [email protected] Shoklo Malaria Research Unit, Mae Sod, Tak Province, Thailand2 Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand3 Centre for Vaccinology and Tropical Medicine, Nuffield Department of Clinical Medicine, John Radcliffe Hospital, Headington, Oxford, UK4 Menzies School of Health Research, Darwin, NT, Australia5 Institut für Tropenmedizin, Berlin, Germany2005 22 9 2005 4 46 46 16 7 2005 22 9 2005 Copyright © 2005 Hutagalung et al; licensee BioMed Central Ltd.2005Hutagalung 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 use of antimalarial drug combinations with artemisinin derivatives is recommended to overcome drug resistance in Plasmodium falciparum. The fixed combination of oral artemether-lumefantrine, an artemisinin combination therapy (ACT) is highly effective and well tolerated. It is the only registered fixed combination containing an artemisinin. The trial presented here was conducted to monitor the efficacy of the six-dose regimen of artemether-lumefantrine (ALN) in an area of multi-drug resistance, along the Thai-Myanmar border.
Methods
The trial was an open-label, two-arm, randomized study comparing artemether-lumefantrine and mefloquine-artesunate for the treatment of uncomplicated falciparum malaria with 42 days of follow up. Parasite genotyping by polymerase chain reaction (PCR) was used to distinguish recrudescent from newly acquired P. falciparum infections. The PCR adjusted cure rates were evaluated by survival analysis.
Results
In 2001–2002 a total of 490 patients with slide confirmed uncomplicated P. falciparum malaria were randomly assigned to receive artemether-lumefantrine (n = 245) or artesunate and mefloquine (n = 245) and were followed for 42 days. All patients had rapid initial clinical and parasitological responses. In both groups, the PCR adjusted cure rates by day 42 were high: 98.8% (95% CI 96.4, 99.6%) for artemether-lumefantrine and 96.3% (95% CI 93.1, 98.0%) for artesunate-mefloquine. Both regimens were very well tolerated with no serious adverse events observed attributable to either combination.
Conclusion
Overall, this study confirms that these two artemisinin-based combinations remain highly effective and result in equivalent therapeutic responses in the treatment of highly drug-resistant falciparum malaria.
==== Body
Introduction
Multi-drug resistance of Plasmodium falciparum is a major health problem in many countries and the number of drugs available, effective and affordable is very limited [1]. Along the Thai-Myanmar border, P. falciparum has developed resistance to almost all available antimalarials [2]. As in tuberculosis and HIV where resistance to drugs is a serious issue, combination therapy has been applied to malaria treatment [3]. The use of antimalarial drug combinations with artemisinin derivatives has been advocated and is now implemented in many countries [4]. An extensive amount of information on efficacy and safety of mefloquine has been reported and reviewed [5]. Artemisinin or Qinghaosu is an extract of the medical plant Qinghao (Artemisia annua), which together with its derivatives, artesunate and artemether are the most active antimalarial compounds to date [6]. The artemisinin derivatives have a rapid onset of therapeutic effect, where a single dose can reduce the parasite biomass by a factor of approximately 104 every 48 hours. In addition, they have a very short terminal elimination half-life of less than 2 hours [7]. Previous studies showed that once-daily administration with artemisinin derivatives produced equivalent cure rates to more frequent administration [8]. A three-day course of artesunate combined with high dose mefloquine has become the standard treatment combination for P. falciparum infections in Thailand [9]. Oral artesunate-mefloquine is the most widely used combination. More recently, a fixed combination of oral artemether-lumefantrine (formerly known as benflumetol) has become available. Artemether is a methyl-ether derivative of artemisinin. Lumefantrine is a racemic fluorine derivative with high blood schizontocidal activity [10]. Both artemisinin combination therapies (ACT) are highly effective and well tolerated [11]. However, resistance to mefloquine and/or to lumefantrine, would compromise both combinations. Therefore it is important to monitor the therapeutic efficacy and thus provide advance warning in case of change. The trial presented here was conducted to monitor the efficacy of the six-dose regimen of artemether-lumefantrine combination given over three days for the treatment of uncomplicated P. falciparum infections in adults and children on the western border of Thailand.
Patients and methods
This study was conducted in the Maela and Mawker Tai malaria clinics of the Shoklo Malaria Research Unit (Mae Sot, Thailand)between July 2001 and June 2002. Patients were recruited from two populations: displaced people of the Karen ethnic minority and migrant workers living on the western border of Thailand. This is an area of low and unstable transmission of Plasmodium vivax and multi-drug-resistant P. falciparum [12]. The trial was an open-label, two-arm, randomized study comparing artemether-lumefantrine andmefloquine-artesunate. This study was approved by the Ethical and Scientific Committees of the Faculty of Tropical Medicine, Mahidol University.
Procedures
Patients >10 kg in weight who had slide-confirmed acute P. falciparum malaria were included in the study, provided that they or their guardians gave fully informed written consent intheir own language, they were not pregnant, they had not received mefloquine in the previous 63 days and there were no other clinical or laboratory signs of severe illness and/or severe and complicated malaria [13]. If they gave written informed consent, they were allocated randomly to receive either the six-dose regimen of artemether-lumefantrine (Coartem® 20/120, Novartis Pharma AG, Basel, Switzerland) or mefloquine (Lariam®, Hoffman-La Roche, Basel, Switzerland) plus artesunate (Guilin Pharmaceutical Factory No.1, Guilin, China). At enrolment (Day 0), a medical history was obtained, a full physical examination was performed and blood was taken for quantitative parasite counts and routine haematology (finger prick blood sample for malaria smear and haematocrit). All information was recorded on a standard case record form. All patients were examined and blood smears were taken daily until they became aparasitaemic, and then weekly for 6 weeks. At each visit a questionnaire on adverse events was completed. A blood smear was also taken from any patient complaining of fever or symptoms compatible with malaria during the follow-up period. Parasite counts were determined on Giemsa-stained thick and thin blood films. The person-gametocyte-weeks were calculated per 1,000 person-weeks after excluding the episodes on admission and during treatment.
Drug regimens
Computerized randomization was in blocks of ten. Patients allocated to artemether-lumefantrine group (ALN) received the tablets at 0 and 8 hours and twice daily for the following 2 days. Artemether-lumefantrine was dispensed as a fixed dose combination tablet. Each tablet contained 20 mg of artemether and 120 mg of lumefantrine. The number of tablets was given according to the body weight. The minimum dosage for patients weighing less than 15 kg was one tablet per dose; patients between 15 and 24 kg received two tablets, those between 25 and 34 kg received three tablets and patients 35 kg and above were treated with four tablets per dose. Patients allocated to artesunate-mefloquine group (MAS3) received artesunate, 4 mg/kg oncedaily for 3 days (day 0 was the first day of treatment), plus mefloquine, 15 mg/kg on day 1 and 10 mg/kg on day 2.
Each patient was given antipyretics and cooled by tepid sponging if the tympanic temperature was equal or above 37.5°C before drug administration. Drug administration was observed in all patients and if vomiting occurred in less than 30 min, administration of the full dose was repeated, if vomiting occurred between 30 and 60 min, half the dose was repeated. Patients treated with artemether-lumefantrine were given a glass of chocolate milk (200 ml) with each dose to increase absorption [14].
Outcome measures
The primary therapeutic outcome measure in this study was the incidence of microscopically and genotypically confirmedrecrudescent infections in both treatment groups by day 42. Parasite genotyping by the polymerase chain reaction (PCR) was used to distinguish recrudescent from newly acquired P. falciparum infections. P. falciparum infections were genotyped for allelic variation in three polymorphic antigen loci, merozoite surface proteins 1 and 2 (MSP-1 and MSP-2) and glutamate rich protein (GLURP), on admission and in case of parasite reappearance [15,16]. Secondary measures were the immediate treatment responses: parasite clearance, fever clearance, incidence of adverse events, and degree of anaemia. The sample size was calculated to detect a difference in failure rates of 7 % between the two regimens with 90% CI and 80 % power assuming a 20% drop out.
Adverse events
Adverse events were symptoms or signs that were not presenton admission and that developed after the start of treatment. All adverse events, including those probably related to malaria, were recording and compared among treatment groups. The rates of early vomiting (<1 h) after each dose and for each drug wererecorded and compared among the groups in the analysis.
Management of recrudescent infections
Patients with uncomplicated recrudescent infections were re-treated with artesunate, 2 mg/kg/day for 7 days; patients >8 years oldalso received doxycycline, 4 mg/kg/day for 7 days.
Statistical analysis
Data were analysed using SPSS for Windows, version 11. Categorical data were compared using the Chi-square test with Yates' correction or by Fisher's exact test, as appropriate. Continuous variables conforming to a normal distribution were compared using Student's t test. Data not normally distributed were log-transformed or compared using the Mann-Whitney U test. The relative risks were calculated using cross-tabulations. The rates of adverse events at three different periods (days 1–2, days 3–7 and days 14–42) were compared among treatment groups. For each of the three periods, the events were counted only once (e.g., if a patient vomited on day one and day two, this was counted as one adverse event). The PCR-adjusted cure rates were evaluated by survival analysis and compared using the log-rank test. Patients for whom PCR genotyping was either inconclusive or missing were censored in the survival analysis on the day of parasite reappearance. For all statistical tests the significance level (p) was set at 0.05.
Results
Four hundred and ninety patients with uncomplicated P. falciparum infections were enrolled between July 2001 and May 2002. The artesunate-mefloquine and artemether-lumefantrine groups included 245 patients each, the age range for all patients was 2–72 years. Baseline characteristics were similar in both groups (Table 1). In total, 484 patients (242 each group) were included in the final evaluation. Six patients (three in each group) were excluded for the following reasons; withdrew consent (1), non-compliance e.g. failure to complete trial treatment course (4), failure to meet protocol criteria (1). Overall compliance was good; around 99% of the patients in the study (481 of 484) were seen at the day seven scheduled visit, 96.1% (465 patients) were seen at day 28 and 93.4% (452 patients) were seen at day 42.
Table 1 Demographic and baseline characteristics
ALN* (n = 245) MAS3** (n = 245)
Maela 100 100
Mawker Tai 145 145
Males, no. (%) 172 (70) 164 (67)
Age, years
Mean (SD) 23.2 (14.6) 23.6 (15.1)
Range 3–70 2–72
Age group, no. (%)
<5 10 (4.1) 8 (3.3)
5–14 81 (33.1) 75 (30.6)
>14 154 (62.9) 162 (66.1)
Weight, kg
Mean (SD) 42.4 (14.5) 42.3 (15.0)
Range 10–77 10–78
Temperature, °C
Mean (SD) 37.7 (1.0) 37.8 (1.1)
Range 35.6–40.5 35.9–41.0
Fevera, no (%) 136 (55.5) 142 (58.0)
Haematocrit, %
Mean (SD) 36.7 (5.9) 36.4 (5.8)
Range 21–52 20–51
Geometric mean (range) 8,047 7,570
parasite count (μl-1) (32–198,789) (16–198,789)
Hepatomegaly, no (%) 48 (19.6) 50 (20.4)
Splenomegaly, no (%) 70 (28.6) 57 (23.3)
* ALN = artemether-lumefantrine, ** MAS3 = artesunate-mefloquine aTympanic temperature ≥ 37.5°C
Clinical and parasitological findings
The initial responses to the two treatment groups were similar. None of the patients developed severe malaria. On admission, 55.0% (133/242) of the patients on ALN and 57.9% (140/242) of the patients in the MAS3 group had a tympanic temperature ≥ 37.5°C. All except three patients had a normal temperature on day 3 (2 in ALN and 1 in MAS3). There was no difference in fever clearance times between the two treatment groups (Figure 1). Parasite clearance times were short and most patients cleared their parasitaemia by day two. Figure 2 shows the percentage of patients with positive slide for asexual P. falciparum in both groups. By day three, four (1.8%) of 227 patients in the artemether-lumefantrine recipients and three (1.3%) of 238 artesunate-mefloquine recipients still had a positive blood film (P > 0.05). Overall, 12.3 % of patients were anaemic (haematocrit <30%) on admission, 10.8% in ALN group and 13.8% in MAS3 group (P = 0.33). The mean (SD) decrease in haematocrit value at day seven from baseline was greater in the group receiving MAS3 than in the ALN group: 9.3% (SD,11.5%; 95% CI, 7.7% to 10.9%) compared with 6.7% (SD, 11.4%; 95% CI, 5.1 to 8.3%) respectively (P = 0.023).
Figure 1 Percentage of patients with fever (temperature > 37.5°C).
Figure 2 Percentage of patients with positive slide for asexual P. falciparum forms.
During the 42-day follow-up period, 27 new P. falciparum infections occurred among artemether-lumefantrine and 24 among artesunate-mefloquine recipients (P > 0.05) (Table 2). The PCR-adjusted cure rates by day 42 were 98.8% (95% CI, 96.4% to 99.6%) in the ALN group and 96.3% (95% CI, 93.1% to 98.0%) in the MAS3 group (P = 0.08). PCR confirmed treatment failures were more likely in children aged below 15 years than in adults (RR, 5.1; 95% CI, 1.4–18.7; P = 0.006). The mean age was 13.6 years (n = 12; SD = 8.5) in patients with treatment failure and 23.7 years (n = 438; SD = 15.2) in successfully treated patients. In this trial, only age group was independently associated with treatment failure, but not other factors e.g. higher parasitaemia (>10,000/μL); anaemia (haematocrit <30%); fever (tympanic temp ≥ 37.5°C) on admission; sites; treatment groups; early vomiting (within one hour following drug administration). The mediantime to recrudescence was comparable for MAS3 group (21 days; n = 9; range, 14–28 days) and ALN group (28 days; n = 3; range, 21–42 days; P > 0.05).
Other parasitological findings
Of 452 patients, 119 (26.3%) had P. vivax parasitaemia detected during follow up. There were significantly fewer cases of vivax malaria in the MAS3 group (29 of 227) than in the ALN group (90 of 225) (P < 0.001). The median time to appearance of P. vivax parasitaemia was longer in the MAS3 group (40 days; range, 13–43 days) than in the ALN group (28 days; range, 14–43 days; P < 0.001).
Twenty patients (8.3%) in the artemether-lumefantrine group and 19 (7.9%) in the artesunate-mefloquine group had gametocytes detected during the first 3 days. All except one patient in ALN group cleared gametocytes within first week after start of treatment. After excluding these, gametocytes developed (between day 7 and 42) in 1.2% (3/241) of ALN group and 1.3 (3/240) of MAS3 group. The person-gametocyte weeks were low and similar: 2.7 (95% CI, 0.6–7.8) per 1000 person-weeks for both groups.
Adverse events
Both treatment regimens were well tolerated. No serious adverse events were reported. Overall, 5/242(2.1%) of the patients vomited one or more doses of medication in the ALN group and 2/242(0.8%) of the MAS3 treated patients (RR, 2.5; 95% CI, 0.5–12.7; P = 0.45). The rates of early vomiting (within one hour) of the drugs were very low (around 2%) and did not differ among groups (one in each group on day 2).
The most commonly reported and possibly drug-related adverse events to both combination therapies were effects on the gastrointestinal (abdominal pain, anorexia, nausea, diarrhoea and late vomiting e.g. >1 h after administration of treatment) and central nervous system (headache, dizziness). Figure 3 shows the proportions of possibly drug-related adverse events of those who did not have those symptoms at admission during follow-up in both groups. Overall, there were less adverse events in ALN group compared to MAS3, though the differences were not statistically significant.
Figure 3 Possibly drug-related adverse events (day 1 – day 42).
Discussion
The loss of affordable effective antimalarial drugs to resistance represents a major threat to the people of malaria endemic countries [1]. Using ineffective drugs with high failure rates kills many and is unacceptable. A clear treatment policy and readiness to use the new, more effective artemisinin-based combination therapies (ACT) are crucial [17]. Along the north-western border of Thailand, where highly multi-drug resistant isolates of P. falciparum are found, artesunate-mefloquine combination therapy (MAS3) is the standard treatment regimen for uncomplicated falciparum malaria [9]. Early diagnosis and treatment with an artemisinin-based drug combination of very high efficacy that reduces gametocyte carriage, has led to a marked decline in the incidence of falciparum malaria and a reversal of the previous trend toward increasing mefloquine resistance [18]. Artemisinin derivatives will ensure rapid clinical and parasitological responses and are remarkably effective, hence clinical deterioration is extremely unusual. To optimize therapy, combination with a slower-acting antimalarial drug is required. Systematic use of ACT would help to delay the emergence of resistance if the drug was used widely. However, the continued use of mefloquine monotherapy or with only 2 days of artesunate in this region, provides persistent selective pressure to continue the evolution of mefloquine resistance, which could diminish the efficacy of the artesunate-mefloquine combination and that of artemether-lumefantrine.
Artemether-lumefantrine has been introduced recently for oral treatment of uncomplicated falciparum malaria. In Thailand, several trials have been conducted with this combination. The six-dose schedule provides sustained blood lumefantrine levels and thus improved cure rates [11]. Lumefantrine is highly lipophilic and the oral bioavailability varies considerably between individuals and increases greatly if the drug is administered after a meal rich in fat [19].
The present trial reconfirmed the efficacy of the six-dose regimen of artemether-lumefantrine given over three days [20]. Both treatments in this study cleared fever and parasitaemia promptly and reliably. Both treatments were well tolerated and highly effective. Importantly 2/3 less P. vivax infections and 12 days longer median time to appearance of P. vivax parasitaemia in the MAS3 group were most probably due to the longer terminal half-life of mefloquine compared to lumefantrine [21,22]. More data on the safety and efficacy of artemether-lumefantrine in very small children and pregnant women are needed.
Authors' contributions
RH carried out the study and analyzed the data. RH, EAA, RMG, PS, TJ, NJW, FN conceived the study, participated in its design and co-ordination and contributed to draft the manuscript. LP, KLT assisted in collection of data. AB performed the PCR experiments. All authors read and approved the final manuscript.
Table 2 Treatment response.
Treatment group ALN* (n = 245) MAS3** (n = 245)
Compliance, no. (%)
Completed day 7 241 (99.6%) 240 (99.2%)
Completed day 28 232 (95.9%) 233 (96.3%)
Completed day 42 225 (93.0%) 227 (93.4%)
Cumulative proportion of patients with clinical failure, no (%)
Day 7 0 (0) 0 (0)
Day 28 13 (5.6) 14 (6.0)
Day 42 27 (12.0) 24 (10.6)
PCR, no.
Novel 23 14
Recrudescent 2 8
Novel + recrudescent 1 1
Indeterminate/missing 1 1
PCR-adjusted cure rates, no. (%)
Day 7 0 (100) 0 (100)
Day 28 2 (99.1) 9(96.1)
Day 42 3 (98.8) 9 (96.3)
* ALN = artemether-lumefantrine, ** MAS3 = artesunate-mefloquine
Acknowledgements
We thank all the staff of the Shoklo Malaria Research Unit (SMRU) for their diligent work. This study was part of the Wellcome Trust-Mahidol University, Oxford Tropical Medicine Research Programme funded by the Wellcome Trust of Great Britain.
==== Refs
White NJ Drug resistance in malaria Br Med Bull 1998 54 703 715 10326295
Nosten F ter Kuile F Chongsuphajaisiddhi T Luxemburger C Webster HK Edstein M Phaipun L Thew KL White NJ Mefloquine-resistant falciparum malaria on the Thai-Burmese border Lancet 1991 337 1140 1143 1674024 10.1016/0140-6736(91)92798-7
White NJ Delaying antimalarial drug resistance with combination chemotherapy Parassitologia 1999 41 301 308 10697872
Looareesuwan S Olliaro P White NJ Chongsuphajaisiddhi T Sabcharoen A Thimasarn K Nosten F Singhasivanon P Supavej S Khusmith S Wylings S Kanyok T Walsh D Leggat PA Doberstyn EB Consensus recommendation on the treatment of malaria in Southeast Asia Southeast Asian Journal of Tropical Medicine and Public Health 1998 29 355 360 9886127
Price R van Vugt M Phaipun L Luxemburger C Simpson J McGready R ter Kuile F Kham A Chongsuphajaisiddhi T White NJ Nosten F Adverse effects in patients with acute falciparum malaria treated with artemisinin derivatives Am J Trop Med Hyg 1999 60 547 555 10348227
Hien TT White NJ Qinghaosu Lancet 1993 341 603 608 8094838 10.1016/0140-6736(93)90362-K
White NJ Assessment of the pharmacodynamic properties of antimalarial drugs in vivo Antimicrob Agents Chemother 1997 41 1413 1422 9210658
Luxemburger C ter Kuile FO Nosten F Dolan G Bradol JH Phaipun L Chongsuphajaisiddhi T White NJ Single day mefloquine-artesunate combination in the treatment of multi-drug resistant falciparum malaria Trans R Soc Trop Med Hyg 1994 88 213 217 8036679 10.1016/0035-9203(94)90303-4
Nosten F Luxemburger C ter Kuile FO Woodrow C Eh JP Chongsuphajaisiddhi T White NJ Treatment of multidrug-resistant Plasmodium falciparum malaria with 3-day artesunate-mefloquine combination J Infect Dis 1994 170 971 977 7930743
Wernsdorfer WH Landgraf B Kilimali VA Wernsdorfer G Activity of benflumetol and its enantiomers in fresh isolates of Plasmodium falciparum from East Africa Acta Trop 1998 70 9 15 9707360 10.1016/S0001-706X(97)00141-1
van Vugt MV Wilairatana P Gemperli B Gathmann I Phaipun L Brockman A Luxemburger C White NJ Nosten F Looareesuwan S Efficacy of six doses of artemether-lumefantrine (benflumetol) in multidrug-resistant Plasmodium falciparum malaria American Journal of Tropical Medicine and Hygiene 1999 60 936 942 10403324
Luxemburger C Thwai KL White NJ Webster HK Kyle DE Maelankirri L Chongsuphajaisiddhi T Nosten F The epidemiology of malaria in a Karen population on the western border of Thailand Transactions of the Royal Society of Tropical Medicine and Hygiene 1996 90 105 111 8761562 10.1016/S0035-9203(96)90102-9
Severe and complicated malaria. World Health Organization, Division of Control of Tropical Diseases Trans R Soc Trop Med Hyg 1990 84 Suppl 2 1 65 2219249
White NJ van Vugt M Ezzet F Clinical pharmacokinetics and pharmacodynamics and pharmacodynamics of artemether-lumefantrine Clin Pharmacokinet 1999 37 105 125 10496300
Brockman A Paul RE Anderson TJ Hackford I Phaiphun L Looareesuwan S Nosten F Day KP Application of genetic markers to the identification of recrudescent Plasmodium falciparum infections on the northwestern border of Thailand American Journal of Tropical Medicine and Hygiene 1999 60 14 21 9988316
Jelinek T Kilian AH Westermeier A Proll S Kabagambe G Nothdurft HD von Sonnenburg F Loscher T Population structure of recrudescent Plasmodium falciparum isolates from western Uganda Trop Med Int Health 1999 4 476 480 10470338 10.1046/j.1365-3156.1999.00428.x
White NJ Pongtavornpinyo W The de novo selection of drug-resistant malaria parasites Proc R Soc Lond B Biol Sci 2003 270 545 554 10.1098/rspb.2002.2241
Nosten F van Vugt M Price R Luxemburger C Thway KL Brockman A McGready R ter Kuile F Looareesuwan S White NJ Effects of artesunate-mefloquine combination on incidence of Plasmodium falciparum malaria and mefloquine resistance in western Thailand: a prospective study Lancet 2000 356 297 302 11071185 10.1016/S0140-6736(00)02505-8
Ezzet F van Vugt M Nosten F Looareesuwan S White NJ Pharmacokinetics and pharmacodynamics of lumefantrine (benflumetol) in acute falciparum malaria Antimicrob Agents Chemother 2000 44 697 704 10681341 10.1128/AAC.44.3.697-704.2000
van Vugt M Looareesuwan S Wilairatana P McGready R Villegas L Gathmann I Mull R Brockman A White NJ Nosten F Artemether-lumefantrine for the treatment of multidrug-resistant falciparum malaria Trans R Soc Trop Med Hyg 2000 94 545 548 11132386 10.1016/S0035-9203(00)90082-8
Karbwang J Na-Bangchang K Clinical application of mefloquine pharmacokinetics in the treatment of P falciparum malaria Fundam Clin Pharmacol 1994 8 491 502 7721226
Ezzet F Mull R Karbwang J Population pharmacokinetics and therapeutic response of CGP 56697 (artemether + benflumetol) in malaria patients Br J Clin Pharmacol 1998 46 553 561 9862244 10.1046/j.1365-2125.1998.00830.x
|
16179089
|
PMC1261533
|
CC BY
|
2021-01-04 16:24:13
|
no
|
Malar J. 2005 Sep 22; 4:46
|
utf-8
|
Malar J
| 2,005 |
10.1186/1475-2875-4-46
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1041616474510.1186/1465-9921-6-104ResearchImmune sensitization of equine bronchus: glutathione, IL-1β expression and tissue responsiveness Matera MG [email protected] L [email protected] A [email protected] A [email protected] C [email protected] M [email protected] Department of Experimental Medicine, Unit of Pharmacology, 2nd University of Naples, Naples, Italy2 Department of Respiratory Medicine, Unit of Pneumology and Allergology, A. Cardarelli Hospital, Naples, Italy3 Department of Veterinary Clinical Science, Faculty of Veterinary Medicine, University of Bologna, Bologna, Italy4 Department of Veterinary Public Health and Animal Pathology, Faculty of Veterinary Medicine, University of Bologna, Bologna, Italy5 Department of Pharmaceutical Sciences, Santissima Annunziata Hospital, Chieti, Italy2005 15 9 2005 6 1 104 104 7 2 2005 15 9 2005 Copyright © 2005 Matera et al; licensee BioMed Central Ltd.2005Matera 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
Increasing clinical epidemiological and experimental evidence indicates that excess of production of reactive oxygen free radicals (ROS) induced by an oxidative stress is involved in the pathogenesis of a number of human airway disorders, as well as equine recurrent airway obstruction. Free-radicals modulate the activation of transcription factors, such as nuclear factor-(NF)-κB and activator protein (AP)-1, in several different cells. This activation leads to expression of many pro-inflammatory cytokines, including interleukin (IL)-1β. We have hypothesized that equine airway sensitization might induce an oxidative stress and increase the ROS production, which in turn might enhance a production of IL-1β and airway hyperresponsiveness.
Methods
We have examined the effect of passive sensitization on IL-1β mRNA expression and electrical field stimulation (EFS)-induced contraction in equine isolated bronchi, and the potential interference of reduced-glutathione (GSH), an antioxidant, with these responses. Bronchi passively sensitized with serum from animals suffering from heaves and having high total level of IgE, and control tissues, either pretreated or not with GSH (100 μM), were used to quantify IL-1β mRNA. Other tissues were used to study the effect of EFS (3–10–25 Hz).
Results
Mean IL-1β mRNA expression was higher in passively sensitized than in control rings. GSH significantly (p < 0.05) reduced the IL-1β mRNA expression only in passively sensitized bronchi. ELF induced a frequency-dependent contraction in both non-sensitized and passively sensitized tissues, with a significantly greater response always observed in sensitized tissues. GSH did not modify the EFS-induced contraction in non-sensitized bronchi, but significantly (p < 0.05) decreased it in passively sensitized tissues.
Conclusion
Our data indicate that the passive sensitization of equine bronchi induces inflammation and hyperresponsiveness. These effects might be due to an oxidative stress because a pretreatment with GSH decreased the increased IL-1β mRNA expression and responsiveness to EFS of passively sensitized bronchi.
equine bronchipassive sensitizationIL-1βreduced-glutathione
==== Body
Introduction
Increasing clinical epidemiological and experimental evidence indicates that excess of production of reactive oxygen free radicals (ROS) is involved in the pathogenesis of a number of airway disorders [1]. When airway cells and tissues are exposed to oxidative stress elicited by environmental pollutants, infections, antigen challenge, inflammatory reactions or decreased level of antioxidants, the enhanced production of ROS can negatively affect several chronic obstructive airway diseases, including human asthma and COPD and equine recurrent airway obstruction (RAO) [2-4]. Horses suffering from RAO have a decreased pulmonary antioxidant capacity, which may render them more susceptible to oxidative challenge [5].
The importance of ROS in inducing RAO is documented by the fact that isoprostanes, which are markers of lipid peroxidation [7], are increased by pulmonary oxidative stress induced by strenuous exercise in heaves-horses and, moreover, epiPGF2α is significantly augmented in these horses during an exacerbation and correlated with airway inflammation and pulmonary function [8,9]. It must be highlighted that experimental airway antigen challenge is associated with immediate formation of ROS, which persists throughout the late asthmatic response [10] and, at least in rats, antigen exposure also increases lipid peroxidation in bronchoalveolar lavage [6].
This resulting oxidative stress may lead to the induction of redox-sensitive transcription factors such as nuclear factor (NF)-κB, hypoxia inducible factor (HIF) and activator protein (AP)-1 in several different cells that leads to an increased expression of pro-inflammatory cytokines, including interleukin (IL)-1β [11,12]. Interestingly, IL-1β plays an autocrine role in altered responsiveness of atopic asthmatic sensitized airway smooth muscle, at least in humans [13].
High levels of NF-κB activity are found during crisis in heaves-susceptible horses [14,15]. Therefore, it is not surprising that exacerbation of the disease in heaves-susceptible horses coincides with increased mRNA expression in BAL cells of the pro-inflammatory cytokines IL-1β and tumor necrosis factor (TNF)-α [16].
Airways that have been passively sensitized with serum from atopic asthmatic patients can be used to study pathophysiological mechanisms that underlie the induction of airway inflammation and hyperresponsiveness [17,18].
We hypothesized that airway sensitization might induce an oxidative stress and increase the ROS production, which in turn might enhance a production of IL-1β and airway hyperresponsiveness. Therefore, this study aimed to evaluate the effect of passive sensitization on IL-1β mRNA expression and on contraction induced by electrical field stimulation (EFS) in equine isolated bronchi and to explore the effect of an antioxidant, the reduced-glutathione (GSH), on these responses.
Materials and methods
Tissue preparations
Four healthy equine male lungs (aged 1.7 ± 0.09 years; weighted kg 390 ± 64.9) were obtained from local abattoir; all animals showed a negative history of heaves. Immediately, after resection, 3rd generation bronchi were excised, cleaned and cut in rings.
Passive sensitization
Tissues were passively sensitized in a random manner, as described by Schmidt et al. [19]. Briefly, equine bronchial smooth muscle rings were rotated overnight at room temperature in tubes containing modified Krebs Henseleit buffer (KH; composition in mM. 118.4 NaCl, 25.0 NaHCO3, 11.7 dextrose, 4.7 KCl, 2.6 CaCl2, 2H2O, 1.19 MgSO4·7H2O and 1.16 KH2PO4 with a cyclo-oxygenase inhibitor; indomethacin; 5 μM; pH 7,4,) in the absence (control rings) or presence of sensitizing serum (sensitized rings). The serum was prepared from the whole blood of animals suffering from heaves, during an exacerbation, with a total level of IgE of 8,095.04 ± 90.9, measured as described by Tizard [20]. Sera were not pooled but were frozen at -20°C in 200-ml aliquots until required.
In order to evaluate the role of ROS in sensitized airways, some of control tissues or sensitized ones were treated with GSH (100 μM). The following morning rings were transferred to 10 ml organ baths aerated with 5% CO2 and 95 O2%, at 37°C; containing modified KH buffer, which was changed every 10 min. The isometric changes in tension were measured with a transducer Fort 10 WPI (Basile, Instruments, Italy)
Lymphocyte proliferation assay
Lymphocyte proliferation assay was carried out in duplicate wells of a 6 well flat bottomed plates (Costar, MA) using horse peripheral blood mononucler cells (PBMCs). For PBMC isolation, blood samples (10 ml) were collected from a healthy horse; cells were then isolated using the standard Ficoll-Hypaque method, and washed three times in PBS. PBMCs were seeded by centrifugation and suspended in RPMI 1640 medium (GIBCO) supplemented by 10% inactivated SFB, and antibiotics (penicillin/ml 200 IU and 150 μg/ml streptomycin) at a final concentration of 2 × 106 cells/ml. For lymphocyte stimulation assays (LSAs), 1 ml of cell suspension along with either 1 ml of ConA (extract from Concanavalia ensiformis, ICN Biochemicals, Cleveland, Ohio, USA) at 10 μg/ml or 1 ml of medium alone were added to 6 well plates The plates were incubated at 37°C and 5% CO2 for 18 or 24 hours. Duplicate cultures were set up to separately measure cytokine production and cellular proliferation. Total RNA was isolated from either stimulated or non stimulated PBMCs cultures using the RNeasy tissue kit (Qiagen Inc., Valencia, CA, USA). RNA concentration was measured by optical density. RNA samples were treated with amplification grade DNAse I (Amersham) to remove any traces of genomic DNA. RT PCR was performed in order to amplify IL1β mRNA using the specific primers FIL-1β 5'GAGGCAGCCATGGCAGCAGTA3' and RIL-1β 5'TGTGAGCAGGGAACGGGTATCTT3' that were designed on the basis of the horse IL1β sequence published in the GENBANK database (D42147). The RT PCR was performed using the Reverse iTtm one step RT-PCR kit (ABgene), briefly, RNA of each sample was reverse transcribed at 47°C for 30' and then incubated at 95°C for 5'. The resulting cDNA was amplified in 35 cycles of denaturation at 94°C for 30 sec, annealing at 58°C for 30 sec and extension at 72°C for 30 sec, followed by a final extension step at 72°C for 7 min. The amplicons were subsequently analyzed on a 1% agarose gel stained with ethidium bromide using Fluor S Multimager (Biorad USA).
Standard DNA preparation
The specific fragment of 257 bp of horse IL1β gene previously amplified from ConA stimulated PBMCs was cloned into the pCR 4/TOPO vector using TOPO cloning kit (Invitrogen) and purified with Turbo Kit (QBIOgene). The recombinant plasmid was linearized upstream the target sequence using the restriction endonuclease PmeI (Fermentas) to avoid the presence of supercoiled plasmid and to simulate more closely the amplification efficiency of genomic DNA. The linearized plasmid was visualized and quantified by electrophoresis in 1% agarose gel containing ethidium bromide using 2-log DNA ladder (New England BioLabs). 10-fold dilutions of the plasmid were made representing 1,14 × 109 copies of DNA/μl of template and stored at -80°C until required. The PCR standard curve was constructed by plotting the plasmid DNA dilutions against the corresponding QPCR threshold cycle value. The threshold was determined using the Auto-Find Threshold function of the Rotor-Gene 3000 (Corbett Research, AU) that scans the range of threshold levels to obtain the best fit of the standard curve through the samples that have been defined as standards.
Expression of mRNA for IL-1β in equine isolated bronchi
RNA was extracted from all mucosal specimens by using RNeasy tissue kit (QIAgen Inc., Valencia, CA, USA). The total RNA was isolated and quantified by spectrophotometry and the ratio 260/280 was estimated. The purified mRNA was incubated for 30 minutes at 37°C with DNAse I (Amersham Biosciences) to eliminate genomic DNA. 50 μg of mRNA was reverse transcribed at 37°C for 60' using the Sensicript reverse transcriptase (Qiagen) and specific reverse primer RIL-1β'. The resulting cDNA was then amplified by real time PCR. The quantitative PCR assay was developed to allow an absolute quantification of the cDNA. The reaction was carried out on the Rotor-Gene 3000 system (Corbett Research, Australia), using QuantiTect SYBR Green PCR kit (Qiagen). Each sample was amplified in a final volume of 20 μl containing 10 μl of master mix, 20 pmol of primers FIL1β and RIL1β and 5 μl of cDNA. Cycling parameters were 15 min at 95°C for the hot start TAQ polymerase activation followed by 45 cycles of 15 s at 94°C, 20 s at 58°C and 20 s at 72°C. Fluorescence data acquisition was carried out at the end of each PCR extension phase. The reaction was performed testing the samples and 5 standard plasmid dilution in duplicate.
Electrical field stimulation studies (EFS)
Tissues were allowed to equilibrate passively for 90 min. During equilibration (2 h) optimal passive tension was determined by gentle stretching of tissue (≅2 mg). The tissue responsiveness was assessed by acetylcholine (ACh) 100 μM; when the response reached a plateau, rings were washed tree times and allowed to equilibrate for 30 min. After full recovery of the tissues, EFS was performed by placing tissues between two wire platinum electrodes, connected to a stimulator 3165 multiplexing pulse booster (Basile Instruments, Italy). Tissues were electrically stimulated for 10 sec (10 V, 0.2 ms) by increasing EFS frequencies (3–10–25 Hz). At the end of the experiments, the wet weight of each tissue was determined.
Analysis of results
Contractile response are expressed as percentage of ACh (100 μM) induced contraction. All values are presented as mean ± SEM. All n values refer to the number of the lungs. Statistical significance was assessed by multifactorial analysis of variance (ANOVA). In presence of a significant overall ANOVA, whenever there was significance, the Tukey's Multiple Comparision test was applied at the 5% level for comparison of the means. A probably level of P < 0.05 was considered as significant for all tests. All data analysis was performed using computer software (GraphPad Prism, CA, USA).
Drugs
The following drugs were used: acetylcholine (ACh), indomethacin and GSH (Sigma, Chem. Co. St Louis, MO).
Results
Baseline characteristics of the bronchial rings
There was no significant difference (P > 0.05) between passively sensitized and non-sensitized bronchi in absence or in presence of a treatment with GSH (100 μM) in wet weight or contraction induced by ACh 100 μM (table 1).
Table 1 Wet weight and contraction induced by ACh (100 μM.) in equine non-sensitized (C) and passively sensitized isolated bronchi, in absence or in presence of a treatment with reduced-glutathione (G; 100 μM). All values are mean ± SEM of four samples
C C+G S S+G
Wet weight mg 285 ± 42.2 295 ± 39 285 ± 30.1 254 ± 27.6
ACh contraction (gr) 15.80 ± 3.4 13.70 ± 0.4 16.40 ± 0.9 13.20 ± 0.5
Effect of passive sensitization and GSH on expression on mRNA for IL-1β
Mean IL-1β mRNA expression in equine passively sensitized bronchi was higher when compared to the control rings. A pretreatment with GSH did not modify the IL-1β mRNA expression in non-sensitized bronchi, whereas it significantly reduced the IL-β mRNA expression in passively sensitized ones (Figure 1 and 2).
Figure 1 Effects of passive sensitization and reduced-glutathione (G; 100 μM) on IL-1β mRNA. Values are mean ± SEM of four samples. C: control rings; S: passively sensitized rings. *P < 0.05 C vs. S; #P < 0.05 C+G vs. S; §P < 0.05 S+G vs. S.
Figure 2 Electrophoresis of IL1b RT-PCR product. Specific 257 bp fragments amplified from PBMCs after 18 and 24 hours stimulation with ConA (lane A and B), lane C no template control, lane M molecular weight marker 100 bp
Effect of passive sensitization and GSH on EFS responses
EFS induced a contraction, frequency dependent, in both non-sensitized and passively sensitized tissues. The magnitude of this contraction was significantly greater in serum-sensitized tissues than in non-sensitized ones at each frequency used. (figure 3). A pretreatment with GSH did not modify the EFS induced contraction in non-sensitized bronchi (figure 3). In passively sensitized equine bronchial rings, a pretreatment with GSH (100 μM) significantly (P < 0.05) decreased contraction at all frequencies used compared to passively sensitized tissues (figure 3).
Figure 3 Effects of passive sensitization and reduced-glutathione (G; 100 μM) on contraction induced by electrical field stimulation in equine isolated bronchi. Values are expressed as percentage of ACh (100 μM) induced contraction and are mean ± SEM of four samples. C: control rings; S: passively sensitized rings. P < 0.05 vs. S.
Discussion
It has been demonstrated that passive sensitization of isolated airway smooth muscles increases the contractile responses to several agonists, such as histamine and leukotriene C4 [19]. This effect is independent of specific IgE and appears to be related to some other factors associated with serum containing high concentrations of total IgE [17,19,21].
The increased reactivity of airways could be due to an increased production of ROS, a term that includes various oxygen free radicals (superoxide anions, hydroxyl radicals, hydrogen peroxide, hypochlorous acid, peroxynitrite and ozone) released from alveolar macrophages, neutrophils and eosinophils when exposed to oxidative stress elicited by inflammatory reactions, infections, environmental pollutants or antigen challenge [22]. In effect, studies carried out in asthmatic patients have demonstrated that antigen challenge enhances the production of ROS. Leukocytes obtained from asthmatic patients generate more ROS compared to cells derived from control subjects [23]. Also neutrophils and monocytes purified from blood of asthmatic patients possess a higher propensity to release ROS than cells obtained from controls [24]. Oxidative stress induced by IgE challenge leads to the activation of genes for many pro-inflammatory cytokines, including IL-1β, which plays an important role in airway hyperresponsiveness and this not only in man but likely even in heaves [2,12,16,25-27].
In the present study, equine passively sensitized isolated bronchi showed a greater expression of IL-1β mRNA than non sensitized tissues. Our result fits with the in vitro documentation that oxidants cause the release of inflammatory mediators such as TNF-α, IL-8 and IL-1-β [26,27]. This is a phenomen that has also been documented in horses. In fact, Giguere et al. [16] demonstrated that the IL-1β mRNA expression was significantly higher after exposure to moldy hay in horses with heaves when compared to values obtained during clinical remission or in healthy controls in one out of two trials.
The sensitized bronchial rings revealed also an increase in the contractile response induced by EFS, which might be related to an altered neural functioning. In human isolated airways, Ichinose et al. [28] demonstrated that the incubation of bronchi with IgE enhanced cholinergic neurotransmission via a M2 dysfunction, without affecting the responses to exogenously applied ACh. This finding is likely related to an indirect mechanism involving inflammatory cell-derived mediators, such as cytokines, and ROS, which can indirectly influence contractile responses related to an alteration of airway nerve function. An increased cholinergic nerve-induced contraction induced by chemical oxidants has been documented in rat tracheal smooth muscle too [29]. On the other hand, ROS also elicit a direct effect on airway smooth muscles, acting either as contractant or relaxant agents. In vitro, it has been documented that exposure of guinea pig tracheal tissues to peroxynitrite and hypochlorous acid causes hyperresponsiveness to histamine and substance P [30,31] and that the oxidative stress increases the contraction of isolated human airway smooth muscles [32]. In agreement with these findings, our results demonstrated that passive sensitization enhances cholinergic nerve-induced contraction, without affecting contractile responses induced by exogenous ACh in equine bronchial rings. In contrast, they did not confirm the previous documentation of Olszewwski et al [33] that in equine trachea, hydrogen peroxide reduced exogenous spasmogen-induced contraction. This discrepancy can partially be explained by the fact that in the study of Olszewski et al [33] only single rather synergistic action of free-radicals in the trachea was examined, whereas our findings have been obtained on isolated bronchi.
In order to confirm the hypothesis that airways inflammation and hyperresponsiveness observed in equine passively sensitized bronchi could be related to an increased production of ROS, we evaluated the effects of a pretreatment with GSH on either non-sensitized or passively sensitized bronchi. We observed that a pretreatment with GSH significantly decreased the IL-1β mRNA expression and the contraction induced in passively sensitized tissues, but did not show any effect in the non-sensitized ones. In order to justify these different responses, we must consider that under normal circumstances ROS are kept under tight control by SOD enzymes [26]. In acute and chronic inflammation, the production of ROS increases at a rate that overwhelms the capacity of the endogenous defense system to remove them [26]. Moreover, GSH is an important component of the lung antioxidant defense [27]. Therefore, a supplementation of this antioxidant agent is necessary to protect lungs from oxidative stress during acute or chronic inflammation. Anyway, there is evidence that also other antioxidant agents, such as N-acetyl-cysteine (NAC), block both in vitro and in vivo the release of inflammatory mediators from epithelial cells and macrophages by a mechanism involving intracellular GSH and decreasing NF-κB activation [34].
In conclusion, our results indicate that passive sensitization of equine bronchi induces airway inflammation and hyperresponsiveness. These effects might be due to an oxidative stress. In fact, the increased IL-1β mRNA expression and responsiveness to EFS of passively sensitized bronchi were decreased by a pretreatment with GSH, which is an antioxidant agent. It is not a surprise that the increase in IL-1β mRNA expression was linked to an increase in airway hyperresponsiveness. In fact, it has been documented that pre-incubation of human isolated small bronchi with IL-1β increased responsiveness to substance P [35] and, moreover, this cytokine induced hyperresponsiveness in sensitized Brown-Norway rats [36]. However, from our study any proper relationship between the increase in IL-1β mRNA expression and the in smooth muscle contractility cannot clearly been seen and we cannot denied that GSH inhibited both in an independent manner. Therefore, further studies will better explain this finding.
==== Refs
MacNee W Oxidative stress and lung inflammation in airways disease Eur J Pharmacol 2001 429 195 207 11698041 10.1016/S0014-2999(01)01320-6
Henricks PAJ Nijkamp FP Reactive oxygen species as mediators in asthma Pulm Pharmacol Ther 2001 14 409 421 11782121 10.1006/pupt.2001.0319
Langen RC Korn SH Wouters EF ROS in the local and systemic pathogenesis of COPD Free Radic Biol Med 2003 35 226 235 12885585 10.1016/S0891-5849(03)00316-2
Art T Kirschvink N Smith N Lekeux P Indices of oxidative stress in blood and pulmonary epithelium lining fluid in horses suffering from recurrent airway obstruction Equine Vet J 1999 31 397 401 10505955
Deaton CM Marlin DJ Smith NC Roberts CA Harris PA Schroter RC Kelly FJ Antioxidant and inflammatory responses of healthy horses and horses affected by recurrent airway obstruction to inhaled ozone Equine Vet J 2005 37 243 249 15892234
Blesa S Cortijo J Mata M Serrano A Closa D Santangelo F Estrela JM Suchankova J Morcillo EJ Oral N-acetylcysteine attenuates the rat pulmonary inflammatory response to antigen Eur Respir J 2003 21 394 400 12661991
Kirschvink N Fievez L Bougnet V Art T Degand G Smith N Marlin D Roberts C Harris P Lekeux P Effect of nutritional antioxidant supplementation on systemic and pulmonary antioxidants status, airway inflammation and lung function in heaves-affected horses Equine Vet J 2002 34 705 712 12455842
Kirschvink N Art T Smith N Lekeux P Effects of exercise and COPD crisis on isoprostane concentration in plasma and bronchoalveolar lavage fluid in horses Equine Vet J Suppl 2001 30 88 91 10659229
Kirschvink N Art T Smith N Lekeux P Bronchoconstrictive properties of inhaled 8-epi-PGF2α in healthy and heaves-susceptible horses Vet Res 2001 32 397 407 11592610 10.1051/vetres:2001132
Sanders SP Zweier JL Harrison SJ Trush MA Rembish SJ Liu MC Spontaneous oxygen radical production at sites of antigen challenge in allergic subjects Am J Respir Crit Care Med 1995 151 1725 1733 7767513
Morcillo EJ Estrela J Cortijo J Oxidative stress and pulmonary inflammation: pharmacological intervention with antioxidants Pharmacol Res 1999 40 393 404 10527653 10.1006/phrs.1999.0549
Bureau F Delhalle S Bonizzi G Fievez L Dogne S Kirschvink N Vanderplasschen A Merville MP Bours V Lekeux P Mechanisms of persistent NF-κB activity in the bronchi of an animal model of asthma J Immunol 2000 165 5822 5830 11067942
Hakonarson H Herrick DJ Serrano PG Grunstein MM Autocrine role of interleukin 1β in altered responsiveness of atopic asthmatic sensitized airway smooth muscle J Clin Invest 1997 99 117 124 9011565
Baeuerle PA Baichwal VR NF-κB as a frequent target for immunosuppressive and anti-inflammatory molecules Adv Immunol 1997 65 111 137 9238509
Bureau F Bonizzi G Kirschvink N Delhalle S Desmecht D Merville MP Bours V Lekeux P Correlation between nuclear factor-κB activity in bronchial brushing samples and lung dysfunction in an animal model of asthma Am J Respir Crit Care Med 2000 161 1314 1321 10764329
Giguere S Viel L Lee E MacKay RJ Hernandez J Franchini M Cytokine induction in pulmonary airways of horses with heaves and effect of therapy of inhaled fluticasone propionate Vet Immunol Immunopathol 2002 85 147 158 11943316 10.1016/S0165-2427(01)00420-2
Watson N Bodtke K Coleman RA Dent G Morton BE Ruhlmann E Magnussen H Rabe KF Role of IgE in hyperresponsiveness induced by passive sensitization of human airways Am J Respir Crit Care Med 1997 155 839 844 9117014
Rabe KF Mechanisms of immune sensitization of human bronchus Am J Respir Crit Care Med 1998 158 S167 S170
Schmidt DT Watson N Dent G Ruhlmann E Branscheid D Magnussen H Rabe KF The effect of selective and non-selective phosphodiesterase inhibitors on allergen- and leukotriene C4-induced contractions in passively sensitized human airways Br J Pharmacol 2000 131 1607 1618 11139438 10.1038/sj.bjp.0703725
Tizard IR Veterinary immunology: an Introduction 2004 7 Philadelphia: Saunders 145 154
Roberts JA Thomson NC The effect of passive sensitization of human bronchial smooth muscle on in vitro sensitivity to histamine Pulm Pharmacol 1989 2 103 105 2520489 10.1016/0952-0600(89)90031-8
Folkerts G Klock J Muijsers RB Nijkamp FP Reactive nitrogen and oxygen species in airway inflammation Eur J Pharmacol 2001 429 521 562 10.1016/S0014-2999(01)01324-3
Vachier I Chanez P Le Doucen C Damon M Descomps B Godard P Enhancement of reactive reactive oxygen species formation in stable and unstable asthmatic patients Eur Resp J 1994 3 1518 1592
Vachier I Damon M Le Doucen C de Paulet AC Chanez P Michel FB Godard P Increased oxygen species generation in blood monocytes of asthmatic patients Am Rev Respir Dis 1992 146 1161 1166 1443865
Lavoie JP Maghni K Desnoyers M Taha R Martin JG Hamid QA Neutrophilic airway inflammation in horses with heaves is characterized by a Th2-type cytokine profile Am J Respir Crit Care Med 2001 164 1410 1413 11704587
Rahman I MacNee W Regulation of redox glutathione levels and gene transcription in lung inflammation: therapeutic approaches Free Radic Biol Med 2000 28 1405 1420 10924859 10.1016/S0891-5849(00)00215-X
Rahman I Oxidative stress and gene transcription in asthma and chronic obstructive pulmonary disease: antioxidant therapeutic targets Curr Drug Targets Inflamm Allergy 2002 1 291 315 14561194 10.2174/1568010023344607
Ichinose M Miura M Tomaki M Oyake T Kageyama N Ikarashi Y Maruyama Y Shirato K Incubation with IgE increases cholinergic neurotransmission in human airways in vitro Am J Respir Crit Care Med 1996 154 1272 1276 8912735
Ohrui T Sekizawa K Yamauchi K Ohkawara Y Nakazawa H Aikawa T Sasaki H Takishima T Chemical oxidant potentiates electrically and acetylcholine-induced contraction in rat trachea: possible involvement of cholinesterase inhibition J Pharmacol Exp Ther 1991 259 371 376 1920123
Sadeghi-Hashjin G Folkerts G Henricks PA Verheyen AK van der Linde HJ van Ark I Coene A Nijkamp FP Peroxynitrite induces airway hyperresponsiveness in guinea pigs in vitro and in vivo Am J Respir Crit Care Med 1996 153 1697 1701 8630623
Murlas CG Murphy TP Lang Z HOCl causes airway substance P hyperresponsiveness and neutral hypoactivity Am J Physiol 1990 258 L361 L368 1694406
Cortijo J Marti-Cabrera M de la Asuncion JG Pallardo FV Esteras A Bruseghini L Vina J Morcillo EJ Contraction of human airways by oxidative stress protection by N-acetylcysteine Free Radic Biol Med 1999 27 392 400 10468214 10.1016/S0891-5849(99)00070-2
Olszewski MA Robinson NE Yu MF Derksen FJ Effects of peroxide on isolated trachealis muscle of horses Am J Vet Res 1995 56 1479 1485 8585660
Blackwell TS Blackwell TR Holden EP Christman BW Christman JW In vivo antioxidant treatment suppresses nuclear factor κB activation and neutrophilic lung inflammation J Immunol 1996 157 1630 1637 8759749
Barchasz E Naline E Molimard M Moreau J Georges O Emonds-Alt X Advenier C Interleukin-1β-induced hyperresponsiveness to [Sar9,Met(O2)11]substance P in isolated human bronchi Eur J Pharmacol 1999 379 87 95 10499376 10.1016/S0014-2999(99)00484-7
Tsukagoshi H Sakamoto T Xu W Barnes PJ Chung KF Effect of interleukin-1β on airway hyperresponsiveness and inflammation in sensitized and nonsensitized Brown-Norway rats J Allergy Clin Immunol 1994 93 464 469 8120273 10.1016/0091-6749(94)90355-7
|
16164745
|
PMC1261534
|
CC BY
|
2021-01-04 16:23:26
|
no
|
Respir Res. 2005 Sep 15; 6(1):104
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-104
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1091617151510.1186/1465-9921-6-109ResearchExpression profiling of laser-microdissected intrapulmonary arteries in hypoxia-induced pulmonary hypertension Kwapiszewska Grazyna [email protected] Jochen [email protected] Stephanie [email protected] Isabel [email protected] Inke R [email protected] Andreas [email protected] Werner [email protected] Rainer M [email protected] Norbert [email protected] Ludger [email protected] Department of Pathology, Justus-Liebig-University Giessen, Germany2 Department of Medical Biometry and Statistics, University at Luebeck, Germany3 Department of Internal Medicine, Justus-Liebig-University Giessen, Germany2005 19 9 2005 6 1 109 109 5 1 2005 19 9 2005 Copyright © 2005 Kwapiszewska et al; licensee BioMed Central Ltd.2005Kwapiszewska 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
Chronic hypoxia influences gene expression in the lung resulting in pulmonary hypertension and vascular remodelling. For specific investigation of the vascular compartment, laser-microdissection of intrapulmonary arteries was combined with array profiling.
Methods and Results
Analysis was performed on mice subjected to 1, 7 and 21 days of hypoxia (FiO2 = 0.1) using nylon filters (1176 spots). Changes in the expression of 29, 38, and 42 genes were observed at day 1, 7, and 21, respectively. Genes were grouped into 5 different classes based on their time course of response. Gene regulation obtained by array analysis was confirmed by real-time PCR. Additionally, the expression of the growth mediators PDGF-B, TGF-β, TSP-1, SRF, FGF-2, TIE-2 receptor, and VEGF-R1 were determined by real-time PCR. At day 1, transcription modulators and ion-related proteins were predominantly regulated. However, at day 7 and 21 differential expression of matrix producing and degrading genes was observed, indicating ongoing structural alterations. Among the 21 genes upregulated at day 1, 15 genes were identified carrying potential hypoxia response elements (HREs) for hypoxia-induced transcription factors. Three differentially expressed genes (S100A4, CD36 and FKBP1a) were examined by immunohistochemistry confirming the regulation on protein level. While FKBP1a was restricted to the vessel adventitia, S100A4 and CD36 were localised in the vascular tunica media.
Conclusion
Laser-microdissection and array profiling has revealed several new genes involved in lung vascular remodelling in response to hypoxia. Immunohistochemistry confirmed regulation of three proteins and specified their localisation in vascular smooth muscle cells and fibroblasts indicating involvement of different cells types in the remodelling process. The approach allows deeper insight into hypoxic regulatory pathways specifically in the vascular compartment of this complex organ.
==== Body
Background
Chronic pulmonary hypertension is associated with structural alterations of the large and small intrapulmonary arteries. Smooth muscle cells, endothelial cells and fibroblasts are involved in this process of vascular remodelling. A set of genes is known to be transcriptionally induced under hypoxic conditions by hypoxia-induced transcription factors (HIF) [1-4] and mice partially deficient for HIF-1α only develop attenuated pulmonary hypertension [5,6]. Several growth factors like PDGF (Platelet derived growth factor), FGF (Fibroblast growth factor) and TGF-β (Transforming growth factor-beta) have been shown to be induced during pulmonary vascular remodelling [7-9]. Finally, regulation of matrix-related genes like procollagens and MMPs (Matrix metalloproteinases) were also described to participate in this process [10,11]. However, a comprehensive set of genes involved in remodelling has not been identified and the time course of gene induction from the initial stimulus up to the structural changes is poorly understood.
Expression arrays can simultaneously determine regulation of a multitude of genes [12-14]. Applying arrays for analysis of hypoxia-induced gene regulation in the lung [13,14], the use of tissue homogenate results inevitably in an averaging of the various expression profiles of the different cell types. As intrapulmonary arteries represent only a minimal portion of the lung tissue (<10 %) the expression profile of this compartment may be largely masked or even lost when using lung homogenates. To overcome this problem, laser-microdissection techniques have been successfully employed and shown to precisely isolate single cells or compartments under optical control [15-17]. Recently, we subjected laser-microdissected intrapulmonary arteries to cDNA array profiling and showed that the expression signature of these isolated arteries differs remarkably from that of lung homogenates [18].
In this study we aimed to identify genes in the vascular compartment that are involved in the development of pulmonary hypertension and the process of lung vascular remodelling in response to hypoxia. Lungs from control mice and those exposed to normobaric hypoxia (FiO2 = 0.1) were excised and used to prepare tissue sections. After laser-microdissection of intrapulmonary arteries, extracted RNA was preamplified and subsequently hybridized to cDNA arrays. To determine the onset of expression changes among different genes and the time course of regulation, hypoxic time periods of 1, 7 and 21 days were selected. For validation of array-based differential gene expression, a subset of genes was independently measured by a combination of laser-microdissection and real-time PCR. Additionally, immunohistochemical analysis was performed for the three selected genes S100A4, CD36 and FKBP1a to determine protein regulation and localisation.
Methods
Lung preparation of mice under hypoxia/normoxia
Lungs were prepared as described previously [18]. All animal experiments were approved by the local authorities (Regierungspräsidium Giessen, no. II25.3-19c20-15(1) GI20/10-Nr.22/2000). In brief, male Balb/cAnNCrlBR mice (Charles River, Sulzfeld, Germany, 20–22 g) were exposed to normobaric hypoxia (inspiratory O2 fraction (FiO2 = 0.1)) in a ventilated chamber. Mice exposed to normobaric normoxia were kept in a similar chamber at a FiO2 of 0.21. After 1, 7 and 21 days, animals were intraperitoneally anesthetized (180 mg sodium pentobarbital/kg body weight), a midline sternotomy was performed, and the lungs were flushed via a catheter in the pulmonary artery (PA) with an equilibrated Krebs Henseleit buffer at room temperature. Afterwards, the airways were instilled with 800 μl prewarmed TissueTek® (Sakura Finetek, Zoeterwoude, The Netherlands). After ligation of the trachea, the lungs were excised and immediately frozen in liquid nitrogen. Preparation of the hypoxic animals was continuously performed in the hypoxic environment.
Laser-assisted microdissection
Microdissection was performed as described in detail previously [18-20]. In brief, cryo-sections (10 μm) from lung tissue were mounted on glass slides. After hemalaun staining for 45 seconds, the sections were subsequently immersed in 70% and 96% ethanol and stored in 100% ethanol until use. No more than 10 sections were prepared at once to reduce the storage time. Intrapulmonary arteries with a diameter of 250–500 μm were selected and microdissected under optical control using the Laser Microbeam System (P.A.L.M., Bernried, Germany) (Figure 1A). Afterwards, the vessel profiles were isolated with a sterile 30 G needle. Needles with adherent vessels were transferred into a reaction tube containing 200 μl RNA lysis buffer.
Figure 1 Intrapulmonary arteries. A) laser-microdissection of small intrapulmonary arteries. 1) The laser cuts along the outer side of the tunica adventitia. 2) A sterile needle is used to isolate the vessel. 3) Needle with adherent vessel is lifted and transferred afterwards to a reaction tube. Magnification × 200. B) Representative intrapulmonary arteries during the process of vascular remodelling. 1) Under normoxic conditions. 2) At day 1 of hypoxia. 3) At day 7 of hypoxia. Smooth muscle cell layer causes vascular thickening. 4) At day 21 of hypoxia. Magnification × 200.
mRNA extraction
Messenger RNA isolation was performed according to the Chomczynski protocol with some modifications as previously described in detail [18]. After washing, RNA was resuspended in 10 μl RNase free H2O, and then subjected to DNase digestion (Ambion, Austin, TX; 1U, 30 min, 37°C). Afterwards, extraction was repeated and RNA was finally resuspended in 4 μl H2O.
cDNA synthesis, amplification, labelling and hybridisation
These steps were performed as described previously [18]. Total RNA was reverse transcribed using the SMART™ PCR cDNA Synthesis Kit (Clontech, Palo Alto, CA). Complementary DNA was purified by the QIAquick™ PCR Purification Kit (Qiagen, Hilden, Germany) and eluted in 45 μl elution buffer (EB). From the eluted cDNA, 2 μl were separated for further determination of the amplification factor. For the PCR-based amplification, the remaining cDNA was mixed with 5 μl 10 × buffer, 1 μl PCR Primer (10 μM), 1 μl dNTP (10 mM) and 1 μl Advantage™ 2 Polymerase Mix. PCR conditions were 95°C for 1 min, followed by 19 cycles with 95°C for 15 s, 65°C for 30 s and 68°C for 3 min. The resulting PCR product was purified using the QIAquick™ columns as described above. Elution buffer (44 μl) was applied twice for elution and 2 μl were used to determine the amplification factor. All incubations were performed with a GeneAmp™ 2400 PCR cycler (PE Applied Biosystems, Foster City, USA).
The purified PCR product was labeled with α-32P dATP using the Atlas SMART™ Probe Amplification Kit (Clontech), purified by QIAquick™ columns, and eluted twice with 100 μl elution buffer. Hybridization was done at 68°C overnight on Mouse 1.2 II Atlas™ cDNA Arrays nylon filters with 1176 spotted cDNAs (Clontech). After washing, filters were exposed to an imaging plate (Fuji Photo Film, Tokyo, Japan). The plate was read with a phosphorimaging system (BAS RPI 1000, Fuji Photo Film).
Analysis of array data
Raw data were collected using the AtlasImage™ 2.0 software (Clontech). Values of spot intensities were adjusted by a global normalization using the sum method provided by the software. The mean global background was calculated, and spots were considered to be present if the spot signal was at least two-fold higher than that.
For changes in transcript abundance, the normalized difference was used as a measure:
Here, IN is given by the adjusted intensity for the normoxia sample and IH by the adjusted intensity for the hypoxia sample, respectively.
For relatively small regulation (2–3 fold), D is comparable to the commonly used log-ratio of the intensities (log2(Q) with Q = IH/IN): D ≈ 0.5•log2(Q). The values of D have a codomain limited between -1 to +1: if either intensity equals 0, log(Q) cannot be determined meaningfully (log(Q) = ± ∞), whereas D gives -1 or +1 in these situations. Between -0.5 and +0.5 (2 fold regulation), both calculation methods give similar results.
The values can be transformed into each other by
The advantage of the normalized difference method over the log-ratio method is that genes with zero values (i.e., "on" and "off" regulation) can be included into further statistical analyses. Additionally, the variation of strongly regulated genes is decreased by expressing the changes as a difference instead of ratios.
In order to screen for relevant genes, the difference of the D values from zero was tested by a two-sided one-sample t-test. Those genes with p values ≤ 0.1 were considered to be potentially regulated genes as real-time PCR confirmed the regulation in >90%.
Relative mRNA quantification by real-time PCR
To confirm the results obtained by nylon membrane hybridization, the regulation of a subset of genes was analyzed by real-time quantitative PCR using the ΔΔ CT method for the calculation of relative changes [21]. Real-time PCR was performed by the Sequence Detection System 7700 (PE Applied Biosystems). PBGD, an ubiquitously as well as consistently expressed gene that is free of pseudogenes was used as reference. For cDNA synthesis, reagents and incubation steps were applied as described previously (18). The reactions (final volume: 50 μl) were set up with the SYBR™Green PCR Core Reagents (Applied Biosystems) according to the manufacturer's protocol using 2 μl of cDNA. The oligonucleotide primer pairs are given in Table 1 (final concentration 200 nM). Cycling conditions were 95°C for 6 min, followed by 45 cycles of 95°C for 20 s, 58°C for 30 s and 73°C for 30 s. Due to the non-selective dsDNA binding of the SYBR™Green I dye, melting curve analysis and gel electrophoresis were performed to confirm the exclusive amplification of the expected PCR product.
Table 1 Primer sequences and amplicon sizes. The primer sets work under identical PCR cycling conditions to obtain simultaneous amplification in the same run. Sequences were taken from GeneBank, Accession numbers are given.
Genbank Accession Primer Sequence (5' → 3') Amplicon Length [bp]
Gene Forward Reverse
PBGD M28664 GGTACAAGGCTTTCAGCATCGC ATGTCCGGTAACGGCGGC 135
Col1a1 U08020 CCAAGGGTAACAGCGGTGAA CCTCGTTTTCCTTCTTCTCCG 124
Col1a2 X58251 TGTTGGCCCATCTGGTAAAGA CAGGGAATCCGATGTTGCC 113
Col3 a1 X52046 TCAAGTCTGGAGTGGGAGG TCCAGGATGTCCAGAAGAACCA 92
CA3 M27796 GACGGGAGAAAGGCGAGTTC CAGGCATGATGGGTCAAAGTG 101
Mgp D00613 GTGGCGAGCTAAAGCCCAA CGTAGCGCTCACACAGCTTG 101
Myl6 U04443 CTTTGAGCACTTCCTGCCCA CCTTCCTTGTCAAACACACGAA 101
Spi3 U25844 TCCTGCCTCAAGTTCTATGAAGC TGTTGATGTGCTGTCGGGAC 82
Cytb245b M31775 TTTCGGCGCCTACTCTATCG TCTGTCCACATCGCTCCATG 101
Bzrp D21207 GAAACCCTCTTGGCATCCG CCTCCCAGCTCTTTCCAGACT 105
Psap U27340 GCAGTGCTGTGCAGAGATGTG TCGCAAGGAAGGGATTTCG 104
Tie2 E08401 GCCGAAACATCCCTCACCT TGGATCTTGGTGCTGGTTCAT 102
PDGFb AF162784 CGCCTGCAAGTGTGAGACAAT CGAATGGTCACCCGAGCTT 105
SRF AB038376 GTCTCCCTCTCGTGACAGCAG CAGTTGTGGGTACAGACGACGT 101
VEGF-R1/FLT1 D88689 GGAGCTTTCACCGAACTCCA TCTCAGTCCAGGTGAACCGC 101
TGF-β1 M13177 GCCCTGGATACCAACTATTGCTT AGTTGGCATGGTAGCCCTTG 127
FGF2 NM_008006 AGCGACCCACACGTCAAACT CGTCCATCTTCCTTCATAGCAAG 104
Tsp1 J05605 ACAGTTGCACAGAGTGTCACTGC CATTCACCATCAGGAACTGTGG 103
CD36 L23108 CCACTGCTTTCAAAAACTGGG GCTGCTGTTCTTTGCCACG 101
CD81 X59047 CCTCAGGCGGCAACATACTC GGCTGCAATTCCAATGAGGT 101
FK506bp1a X60203 CAAGCAGGAGGTGATCCGAG CGGTGGCTCCATAGGCATAG 104
bFGF1 precursor X51893 TACAAGAAAACCACCAACGGC CCAAAAGACCACACATCGCTC 101
Il-9 receptor M84746 GGCAGCAGCGACTATTGCAT ACACAGGAAGGGCCACAGG 115
Cyt cVIIc X52940 GGTTCACGACCTCCGTGGT CATCATAGCCAGCAACCGC 101
Ogn D31951 GACCTGGAATCTGTGCCTCCT ACGAGTGTCATTAGCCTTGCAG 114
Ptbp1 X52101 TGGTGTGGTCAAAGGCTTCA GCAGTTCAATCAGCGCCTG 101
S100A4 D00208 AGGAGCTACTGACCAGGGAGCT TCATTGTCCCTGTTGCTGTCC 103
Hypoxia response element (HRE)
Genes regulated after 1 day of hypoxia treatment were screened for presence of hypoxia response elements (HRE). The consensus sequence chosen for HRE was "BACGTSSK", were B can be T, G or C; S – G or C and K – T or G. Regulated genes from 1 day array results were screened 5,000 bp downstream and upstream from coding sequence for the occurrence of this consensus sequence. Sequences were obtained from (according to accession numbers given for the corresponding features on the nylon arrays).
Biological processes
Accession numbers from genes being regulated in hypoxia conditions were subjected to screening biological processes by using Gene Ontology page, AmiGo:
Immunohistochemistry
Cryo-sections (10 μm thick) from lung tissue were mounted on Superfrost glass slides (R. Langenbrinck, Germany). Slides were dried overnight and stored at -20°C until use. Fixation was performed in acetone (Riedel-de Haen, Seelze) for 10 minutes. All antibodies were diluted in ChemMate™ Antibody Diluent, (Dako, Denmark). Following dilutions of primary antibodies were used: Rabbit polyclonal anti-human S100A4 antibody (Neomarkers, Fremont, CA) – 1:700, rabbit polyclonal anti-human FKBP1a antibody (Abcam, Cambridge, UK) – 1:300, rabbit polyclonal anti-human CD36 (Santa Cruz Biotech, California, USA) – 1:200. S100A4 and CD36 were incubated in a humid chamber overnight, while FK506BP (FKBP1a, FKBP12) was incubated for one hour. Afterwards, the slides were washed 3 × in TBS and incubated with the secondary antibody goat anti-rabbit IgG (Southern Biotech, Eching, Germany) – 1:150 for 40 min. After washing, alkaline phosphatase conjugated anti-goat antibody (Rockland, Gilbertsville, PA) – 1:200, 40 min was applied. Negative controls were performed with the omission of the first antibody.
Results
Animal model: Vascular remodelling
Prolonged exposure to hypoxia results in structural changes of small intrapulmonary arteries in mouse lungs. These changes are mainly characterised by thickening of media layer (proliferation of vascular smooth muscle cells) (Figure 1B).
Array analysis
For each array analysis 30 to 40 vessel profiles (diameter 250–500 μm) were isolated from lung sections of animals kept in hypoxia (FiO2 0.1) and those kept in normoxia for 1, 7, and 21 days. In all cases, four independent hybridization experiments were performed. When comparing exposure to hypoxia against normoxia, 29 genes (19 up/10 down), 38 genes (18 up/20 down), and 42 genes (25 up/17 down) were regulated after 1, 7, and 21 days, respectively with a p-value ≤ 0.1 (Additional files 1, 2 and 3).
Determination of regulation by real-time RT-PCR
For all hypoxic time periods, subsets of genes were selected for independent determination of regulation by real-time RT-PCR using intrapulmonary arteries isolated by laser-microdissection. To confirm the array data, we randomly selected genes from the unified list of genes, but with a certain focus on genes with a regulation factor between 0.5 and 2. Three independent experiments were performed for each gene. Mean ± SEM is presented in the respective columns in additional files 1, 2 and 3. In total, 37 ratios of hypoxic to normoxic expression were determined. From these genes under investigation, 34 (95 %) were clearly confirmed to be up- or down-regulated. Only CD 81 failed to be ascertained at day 7. Although, most of the genes were regulated by less than factor 2 when assessed by array analysis, the vast majority of these regulations were confirmed by real-time PCR (Figure 2).
Figure 2 Comparison of array based time course of expression to that obtained by real-time RT-PCR (red: array; blue: TaqMan). A) Matrix γ-carboxyglutamate protein. B) Procollagen 3 α1. C) Prosaposin.
Growth factor analysis
Among growth factors and receptors that were assumed to be regulated, sequences of PDGF (β-polypeptide), TGF-β1, TSP-2/TSP-1 (sequence homology 77%) and VEGF-R1 (Flt) were immobilized on the applied nylon filter. However, no hybridisation signal was detected for these genes. Therefore, relative mRNA levels of these genes together with FGF-2, Angiopoietin Receptor 2 (TIE2) and Serum Response Factor (SRF) were determined by real-time PCR from laser-microdissection from 1 and 7 days hypoxic/normoxic intrapulmonary arteries (Table 2). All transcripts were detected by real-time RT-PCR. PDGF-B and TSP-1 showed an upregulation after 1 and 7 days of hypoxia, TIE-2, TGF-β and SRF only after 7 days. VEGF-R1 mRNA was increased after 1 day, but decreased after 7 days. FGF-2 was slightly downregulated in hypoxia.
Table 2 Growth factors determined by real-time PCR. Among growth factors and receptors that were described to be regulated, TSP-1, VEGF-R1 (Flt), PDGF-B, Serum Response Factor (SRF), TGF-β 1, Angiopoietin Receptor 2 (TIE2) and FGF-2 were separately determined by relative mRNA quantification after laser-microdissection from 1 and 7 day hypoxic/normoxic intrapulmonary arteries. Mean ± SEM is given from n = 4 independent experiments.
Genes 1 Day Hypoxia 7 Days Hypoxia
Thrombospondin 1 (TSP-1) 4.61 ± 0.79 1.95 ± 0.44
VEGF-R1/FLT1 2.38 ± 0.43 0.61 ± 0.13
PDGF-β 1.41 ± 0.28 2.96 ± 0.82
Serum Response Factor (SRF) 1.09 ± 0.08 1.70 ± 0.29
Transforming Growth Factor β 1 (TGF-β 1) 0.94 ± 0.14 2.10 ± 0.46
Angiopoietin Receptor 2 (TIE2) 0.91 ± 0.09 1.94 ± 0.21
Fibroblast Growth Factor 2 (FGF-2) 0.75 ± 0.14 0.80 ± 0.15
Classification of genes according to biological processes
Genes were grouped in nine classes according to their biological processes:
Organogenesis (angiogenesis, muscle development), cell adhesion/cell organisation, signal transduction, cell growth and/or maintenance (cell cycle, lipid transport, ion transport), immune response (antigen presentation, immune cell activation), proteolysis and peptidolysis, transcription/translation process (DNA packaging and repair, RNA processing, protein biosynthesis), energy metabolism/electron transport (carbohydrate metabolism, lipid catabolism, electron transport, removal of superoxide radicals), unknown (biological processes not known for mouse or human genes).
The sizes of the pie charts in Figure 3 correspond to the contribution of genes involved in one of the biological processes. After 1 day of hypoxia most regulated genes (> 35%) responsible for metabolism, while at later time points this group was less prominent (~20% for 7 and 21 days). With continued exposure to hypoxia the subset of regulated genes responsible for organogenesis (3.5%, 13%, and 9% for 1, 7 and 21 days, respectively) and immune response (0%, 3%, and 7% for 1, 7 and 21 days respectively) was increased.
Figure 3 Gene classification according to biological processes. Significantly regulated genes were grouped according to their biological processes from NCBI, Gene Ontology, AmiGo. A) 1 day hypoxia, B) 7days hypoxia, C) 21days hypoxia.
Genes potentially regulated by hypoxia-inducible transcription factor (HIF) responsive element (HRE)
The genomic context of genes upregulated after 1 day was screened 5,000 bp downstream and upstream from coding sequence for the presence of the HIF-responsive element consensus sequence "BACGTSSK". Among those genes some were carrying HRE (e.g. CD36, and MAD4), while others did not have any (e.g. apolipoprotein D). From 17 different possible variants of HRE, four: CACGTGGT, GACGTGGG, CACGTGCT and TACGTGGG were found to be the most common sequences (47% of all HRE) see Figure 4.
Figure 4 Putative HIF-responsive elements (HRE) of the genes upregulated at day 1. Twenty genes were screened for the presence of the consensus sequence "BACGTSSK" 5000 bp up- and downstream the coding sequence. Aldolase C, a known HIF-responsive gene, was excluded. Fifteen genes were found carrying one or more putative HREs.
Regulation and protein localisation of CD36, S100A4, and FKBP1a
Three genes (CD36, S100A4, and FKBP1a) were selected for further characterisation. From the array data, CD36 showed a mean of 1.1 at day 1 and 0.9 at day 7 (both unregulated), with a remarkable standard deviation. Using real-time RT-PCR, upregulation (2.9 ± 0.56) was observed at day 1 and a slight downregulation at day 7, but also with high deviation (0.7 ± 0.29) (Figure 5A and additional files 2 and 3). On the other hand, the data from the arrays and real-time RT-PCR for S100A4 and FKBP1a showed strong correlation in upregulation during prolonged hypoxia exposure.
Figure 5 Regulation of S100A4, CD36 and FKBP1a on mRNA and protein level. A) Comparison of regulation between laser-microdissected arteries and lung homogenate from 1, 7, and 21 days of hypoxia exposure. (Red: array; blue: TaqMan). B) Immunohistochemical staining of S100A4, CD36 and FKBP1a in the mouse lung.
We also examined whether the expression levels of CD36, S100A4, and FKBP1a could have been detected by real-time RT-PCR using lung homogenate. Interestingly, only S100A4 was significantly regulated at day 7 of hypoxia exposure, while no regulation was observed for any of the other genes at all time points (Figure 5A).
Regulation was then investigated on the protein level by immunohistochemistry (Figure 5B). CD36, S100A4, and FKBP1a showed a similar time course of protein expression as predicted by real-time RT-PCR. S100A4 and CD36 were localised exclusively to smooth muscle cells, whilst FKBP1a expression was restricted to the adventitia. Localisation of S100A4 was confirmed by the co-localisation with anti-alpha smooth muscle actin on serial sections (Figure 6A). After prolonged hypoxic exposure (7 and 21 days) S100A4 was additionally located in neo-muscularised resistance vessels (Figure 6B).
Figure 6 Immunolocalisation of S100A4. A) S100A4 protein (left panel) co-localises with alpha-smooth muscle actin (right panel). B) Small vessels (marked by arrows) are negative for S100A4 under normoxia (left panel) however stain positive for S100A4 after 21 days of hypoxia.
Discussion
cDNA arrays have been shown to be powerful tools for the broad analysis of the transcriptome. The combination with laser-microdissection reveals compartment- or even cell-type specific gene regulation within complex tissues and organs [22-24] that may be masked using tissue homogenate (Figure 5a). Indeed, when comparing tissue homogenates to intrapulmonary arteries, the whole expression profiles differed completely [18]. Thus, the presented study is focusing on microdissected intrapulmonary arteries for the analysis of gene expression underlying hypoxic vascular remodelling.
Technical aspects
Statistical analysis
For measurement of differential gene expression, the ratio of intensities is usually calculated after normalization. For genes with intensity values close to background or even absent in one condition, the ratio cannot be calculated. Consequently, these genes are excluded from statistical analysis although they are obviously regulated. To overcome this problem, the differences of the background-corrected and normalized intensities were used instead of their ratios. However, among the genes measured independently by real-time PCR, 95% were confirmed in regulation (e.g. osteoglycin after 1 d, cytochrome b-245 alpha polypeptide after 21 d).
Technical limitations
A couple of reasons may cause a discrepancy of the results obtained from arrays and real-time PCR:
Filter-based micro arrays have a limited dynamic range. This mainly is due to the fact that images have to be acquired where the intensity information is coded into 16-bit variables [25,26]. Real-time PCR offers a significantly higher dynamic range for detection that is more than 20,000-fold higher than the range of arrays obtained from 16-bit images [27,28]. Additionally, cross-hybridisation on the arrays may reduce the dynamic range or even completely cover differences, especially of low abundant genes [29]. Furthermore, micro arrays with several hundreds or even several thousands of sequences are hybridised at one temperature. As the immobilized sequences may vary a bit in their optimum hybridisation temperature, some labelled products may show suboptimal hybridisation efficiencies at the given temperature. Finally, low-abundant transcripts may not yield enough signal and fail to be detected by array analysis but are easily identified by quantitative RT-PCR. Consequently, both sensitivity and precision limit the ability to detect and identify regulated genes by arrays. Due to these limitations coupled with statistical restrictions, array data should be confirmed by real-time PCR. Following this line, some important genes (i.e., VEGF-R1, TGF-β) known to be involved in the remodelling process [7,30,31] were expected to be regulated in response to hypoxia. As these genes failed to be positive by array analysis, we performed real-time RT-PCR. By this more sensitive technique, the genes were detected throughout and regulation levels could be determined. We conclude that the absence of labelled spots does not necessarily indicate the absence of the gene's mRNA.
Furthermore, utilising nylon filters with 1176 spotted genes some gene subsets were absent, including several interesting candidates in hypoxia induced regulation, e.g., ion channels, some growth and transcription factors. With potential importance for our focus of the remodelling process, we exemplarily analysed some additional genes by real-time PCR (FGF-2, TIE2, Serum Response Factor).
Differential gene expression and time courses
Among the genes with potential regulation, some showed differential expression at one, two or all three different time points. While some genes have already been mentioned to be involved in hypoxia-induced vascular remodelling (e.g. procollagens; [10], many others are shown to be related to this process for the first time. As expected, hypoxia did not turn out to be a dramatic stimulus for expression changes, and only few genes were measured to be upregulated with more than factor two (i.e., procollagens after 7 and 21 days), or to be downregulated to the same extent (i.e., CD36 after 21 days). After 1 day of hypoxia, ion-binding genes (45-kDa calcium-binding protein precursor, S100 calcium binding protein A4, chloride ion current inducer protein) as well as transcription modulating genes (MAD4, poly A binding proteins, and polypyrimidine tract binding protein) were predominantly regulated. FK506 binding protein 1a is well known to be involved in cell cycle regulation [32], but also in contraction-associated Ca2+ release from the sarcoplasmatic reticulum [33]. This may indicate altered ion homeostasis in response to hypoxia as well as transcriptional preparation and initiation of long-term modifications in the vascular cells. Growth stimulus via increased expression of VEGF-R1, TSP-1, and PDGF fits well into this view. Interleukin 9 receptor, a T(H)2-type cytokine receptor, showed increased expression after 1 day, followed by downregulation after 21 days. Interestingly, it was also found to be upregulated in fibroblasts derived from an aortic aneurysm [34]. After 7 days, PDGF and TSP-1 were still increased as compared to controls. Serum responsive factor (SRF), angiopoietin 2 receptor (TIE2), fibroblast inducible secreted protein (FISP, mouse homolog of mda-7/Il-24) and TGF-β joined the upregulated growth and angiopoesis mediators. The production of matrix was apparently increased, as indicated by enhanced expression of fibronectin, matrix gamma carboxyglutamate protein and procollagen subunits. Vasodilator-stimulated phosphoprotein (VASP), a substrate of NO targeted cGMP dependent protein kinase [35] that is involved in fibroblast migration [36] was also upregulated. After 21 days, while the matrix production was still ongoing, reconstruction by proteases (carboxypeptidase E, serine proteinase inhibitor 2.2) additionally occurred.
To identify possible regulation mechanisms, we defined groups of genes exhibiting similar time courses of differential gene expression. Examples of these groups are given in Figure 7. First, we grouped genes that were upregulated throughout all time points. Representatives are FK506 binding protein 1a (12 kDa), prosaposin, fibroblast inducible secreted protein (FISP) and aldolase 3C isoform. In contrast, we found genes that were downregulated throughout (i.e., osteoglycin, cell division cycle 10 homolog, HSP 60, cellular nucleic acid binding protein). Furthermore, some genes were upregulated after 1 day, but strongly decreased afterwards, dropping below the normoxic level (i.e., anti-oxidant protein 1, CD36, interleukin 9 receptor, cathepsin D). Another group showed initial downregulation, but increased afterwards above the normoxic level (i.e., matrix gamma carboxyglutamate protein, procollagen 3α 1 subunit, tubulin alpha 7, small inducible cytokine A21A). Finally, some genes seem to be unregulated at early stages, but were at later stages up- or downregulated ("late response"). Genes belonging to this group are inhibitor of DNA binding 1, cathepsin L precursor, carboxypeptidase E and carbonic anhydrase 3.
Figure 7 Classification of genes with similar regulation pattern. Four representatives each are given. A) Continuous upregulation at day 1, 7, and 21. B) Continuous downregulation at day 1, 7, and 21. C) Primarily upregulated, afterwards decrease under normoxic level (= downregulation). D) Primarily downregulated, afterwards increase over normoxic level (= upregulation). E) Primarily not regulated, afterwards up- or downregulated ("late response").
Even if some of these data vary and may lead to slight changes in the classification of the genes, fairly consistent profiles were noted for many genes. In addition, many time-courses were confirmed by real-time PCR-derived measurements (see Additional files 1, 2, and 3). When directly comparing the array-based regulation profile to that based on real-time PCR (Figure 2 and 5A), excellent correlation was found for matrix gamma-carboxyglutamate protein, procollagen 3α1 subunit, S100 calcium binding protein A4 and FK506 binding protein 1a. The level of prosaposin upregulation when measured by real-time PCR was greater than by arrays at day 21. CD36 varied considerably at day 1 and 7 using both techniques. While array measurements did not allow allocation of this gene definitely to group C or E, relative mRNA quantification indicated primary upregulation and thus inclusion to group C. Overall, the possibility to allocate many genes to one of these five groups supports the hypothesis that these genes may be regulated by common mechanisms and regulatory elements, although not being primarily related.
Most of the genes regulated in array experiments were responsible for metabolism. Hypoxia regulates many genes involved in glycolysis [37-39], lipid pathways [40,41], protein synthesis and degradation [42,43]. The expression of metabolic genes was more pronounced at the early time point (1 day of hypoxia), which might indicate an adaptative response. Moreover, with increased duration of hypoxia more genes responsible for angiogenesis were upregulated. This finding matches perfectly to reports, which demonstrate vascular remodelling after prolonged exposure to hypoxia [44-46].
Due to the potential discrepancy between mRNA and the protein levels, we applied immunohistochemical staining to analyse protein expression. All three investigated proteins (S100A4, CD36 and FKBP1a), showed good correlation to mRNA expression levels. S100A4 and CD36 were localised exclusively to smooth muscle cells, while FKBP1a expression was restricted to the adventitia (Figure 5B). At later time points (7 and 21 days), we additionally found S100A4 in newly muscularized small vessels. Interestingly, approximately 5% of mice overexpressing S100A4 develop spontaneously pulmonary arterial lesions similar to that seen in patients with pulmonary vascular disease [47]. Lawire et al. have recently described that induction of S100A4 by serotonin induces migration of human pulmonary artery SMC [48]. In accordance with these studies, the observed upregulation of S100A4 and localisation to small vessels indicates an ongoing remodelling process stimulated by hypoxia. CD36 has been associated with many processes such as scavenger receptor functions, lipid metabolism, fatty acid transport, angiogenesis, cardiomyopathy and TGF-β activation [49]. Therefore, its higher expression in arteries after 1 day hypoxia exposure may indicate adaptation to low oxygen tension. Another protein, FKBP1a was more abundant in later hypoxia time points and was already shown to be involved in cell cycle regulation and Ca2+ homeostasis [32,33]. Moreover, FKBP1a was found to be activated via ERK-R and AKT pathway leading to the HIF-2α nuclear translocation and subsequent transcription of target genes responsible for increased angiogenesis and proliferation [50].
Genes potentially regulated by hypoxia-inducible transcription factors (HIF)
Alveolar hypoxia leads to vasoconsrtiction of pulmonary arteries. Chronic hypoxia downregulates expression of voltage-gated potassium channels [51], resulting in depolarisation of smooth muscle cells, subsequent Ca2+ influx and increased vasoconstriction. Small intrapulmonary vessels appear to react stronger to oxygen deprivation than larger vessels. This might be due to different expression level of potassium channels on both types of vessels. Supporting this hypothesis, Archer et al. have shown preferential expression of voltage-gated potassium channels in resistance pulmonary arteries [52].
In addition to increased cytoplasmic Ca2+ levels, another important effectors for hypoxic remodelling are hypoxia-inducible transcription factors (HIF) [1-3]. The binding to HIF-responsive elements (HREs) following nuclear translocation results in an increased transcription of the respective genes. Both, the HIF-1α and HIF-2α subunits undergo hypoxia-induced protein stabilisation and bind identical target DNA sequences [53]. After defining a consensus sequence for the HREs [54], several dozen genes have been revealed to possess HREs [3,4]. Moreover, using reporter assays regulation was confirmed to be HIF dependant (i.e., erythropoietin; ref. [55]). Among the genes positively detected on the nylon filters, aldolase C is known to be regulated in a HIF-dependent manner [4] and was upregulated at all time points (Figure 7, group A). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), another HRE-carrying gene, was found to be upregulated at day 7 and 21. However, in arrays from day 1 the GAPDH spot intensity was maximum for both normoxia and hypoxia, and a ratio could not be calculated. We investigated the genes upregulated at 1 day (Additional file 1) for the presence of HRE. From the 21 upregulated genes identified by array analysis, we screened 5000 bp up- and downstream of the coding sequence for the presence of the consensus sequence "BACGTSSK" [54]. Putative HREs were detected in 15 genes (Figure 4). Interestingly, 4 from 17 possible sequence variants that had the highest occurrence were also found in well-known HIF-1 regulated genes (VEGF, EPO, ENO1, and GAPDH). This finding underlines the importance of genes carrying the above mentioned sequences. Respective genes may be HIF-induced, which remains to be confirmed in the future by reporter gene assays or electrophoretic mobility shift analysis. On the other hand, in six upregulated genes no HRE consensus sequences could be found. These genes may be induced by a HIF dependent hypoxia-responsive element not represented by the above given consensus sequence. Alternatively, these genes may be indirectly regulated by another, primarily HIF-induced gene. Additionally, other regulatory pathways may exist to upregulate genes in a hypoxia dependent manner.
Conclusion
Combining laser-microdissection and cDNA array analysis allows a compartment-specific broad gene expression analysis of intrapulmonary arteries in a model of hypoxia-induced pulmonary hypertension. Sets of genes were found to be up- or downregulated at 1, 7 and 21 days of hypoxia reflecting different states of vascular remodelling. According to similar time courses of differential expression, 5 groups were classified indicating common regulation mechanisms. Among the genes upregulated at day 1, several carry putative HIF responsive transcription elements while others do not. This may suggest alternative pathways of hypoxia sensing and downstream gene regulation. Immunohistochemistry confirmed regulation of three proteins and specified their localisation in vascular smooth muscle cells (S100A4, CD36) and fibroblasts (FKBP1a) indicating involvement of the different cells types in the remodelling process. Thus, our approach revealed several new genes involved in the process of hypoxic lung vascular remodelling and allows deeper insight into the underlying mechanisms of the vascular lung compartment.
Authors' contributions
GK: laser-microdissection, arrays, real-time PCR, immunohistochemistry, preparation of the manuscript
JW: analysis of array data and real-time PCR data
SW: laser-microdissection, arrays, real-time PCR
IL: immunohistochemistry, real-time PCR
IRK: advice and discussion of statistical calculation
AZ: advice and discussion of statistical calculation
WS: design of project, discussion of data
RMB: introduction to laser-microdissection, analysis of immunohistochemistry and histopathology
NW: animal model of hypoxia induced pulmonary hypertension, discussion of data
LF: coordination and design of project, preparation of the manuscript
All authors have read and approved the finial manuscript.
Supplementary Material
Additional File 1
List of genes up- or down-regulated at day 1 of hypoxia. For changes in transcript abundance, the normalized difference D was used as a measure (see Methods). The D derived Q(D) is given and compared to the commonly used ratio of the intensities Q = IH/IN. If either intensity equals 0, log2(Q) cannot be determined meaningfully, whereas D gives -1 or +1 in these situations. This allows to include genes with zero values (i.e., "on" and "off" regulation) into further statistical analyses. In order to screen for relevant genes, the difference from zero of the D values was tested by a two-sided one-sample t-test. Those genes with p-values ≤ 0.1 were considered to be potentially regulated as real-time PCR confirmed in >90% the regulation. TaqMan PCR derived ratios are given as mean ± standard error of mean (SEM).
Click here for file
Additional File 2
List of genes up- or down-regulated at day 7 of hypoxia. For changes in transcript abundance, the normalized difference D was used as a measure (see Methods). The D derived Q(D) is given and compared to the commonly used ratio of the intensities Q = IH/IN. If either intensity equals 0, log2(Q) cannot be determined meaningfully, whereas D gives -1 or +1 in these situations. This allows to include genes with zero values (i.e., "on" and "off" regulation) into further statistical analyses. In order to screen for relevant genes, the difference from zero of the D values was tested by a two-sided one-sample t-test. Those genes with p-values ≤ 0.1 were considered to be potentially regulated as real-time PCR confirmed in >90% the regulation. TaqMan PCR derived ratios are given as mean ± standard error of mean (SEM).
Click here for file
Additional File 3
List of genes up- or down-regulated at day 21 of hypoxia. For changes in transcript abundance, the normalized difference D was used as a measure (see Methods). The D derived Q(D) is given and compared to the commonly used ratio of the intensities Q = IH/IN. If either intensity equals 0, log2(Q) cannot be determined meaningfully, whereas D gives -1 or +1 in these situations. This allows to include genes with zero values (i.e., "on" and "off" regulation) into further statistical analyses. In order to screen for relevant genes, the difference from zero of the D values was tested by a two-sided one-sample t-test. Those genes with p-values ≤ 0.1 were considered to be potentially regulated as real-time PCR confirmed in >90% the regulation. TaqMan PCR derived ratios are given as mean ± standard error of mean (SEM).
Click here for file
Acknowledgements
We thank K. Quanz and M. M. Stein for excellent technical assistance, L. Marsh for critical reading of the manuscript, G. Jurat for photographic arrangement, and W.H. Gerlich (Institute of Virology, Justus-Liebig-University Giessen) for using the phosphorimaging system.
==== Refs
Semenza G Signal transduction to hypoxia-inducible factor 1 Biochem Pharmacol 2002 64 993 998 12213597 10.1016/S0006-2952(02)01168-1
Bracken CP Whitelaw ML Peet DJ The hypoxia-inducible factors: key transcriptional regulators of hypoxic responses Cell Mol Life Sci 2003 60 1376 1393 12943226 10.1007/s00018-003-2370-y
Wenger RH Cellular adaptation to hypoxia: O2-sensing protein hydroxylases, hypoxia-inducible transcription factors, and O2-regulated gene expression Faseb J 2002 16 1151 1162 12153983 10.1096/fj.01-0944rev
Semenza GL Oxygen-regulated transcription factors and their role in pulmonary disease Respir Res 2000 1 159 162 11667980 10.1186/rr27
Yu AY Shimoda LA Iyer NV Huso DL Sun X McWilliams R Beaty T Sham JS Wiener CM Sylvester JT Semenza GL Impaired physiological responses to chronic hypoxia in mice partially deficient for hypoxia-inducible factor 1alpha J Clin Invest 1999 103 691 696 10074486
Shimoda LA Manalo DJ Sham JS Semenza GL Sylvester JT Partial HIF-1alpha deficiency impairs pulmonary arterial myocyte electrophysiological responses to hypoxia Am J Physiol Lung Cell Mol Physiol 2001 281 L202 8 11404263
Tuder RM Flook BE Voelkel NF Increased gene expression for VEGF and the VEGF receptors KDR/Flk and Flt in lungs exposed to acute or to chronic hypoxia. Modulation of gene expression by nitric oxide J Clin Invest 1995 95 1798 1807 7706486
Katayose D Ohe M Yamauchi K Ogata M Shirato K Fujita H Shibahara S Takishima T Increased expression of PDGF A- and B-chain genes in rat lungs with hypoxic pulmonary hypertension Am J Physiol 1993 264 L100 6 8447423
Arcot SS Fagerland JA Lipke DW Gillespie MN Olson JW Basic fibroblast growth factor alterations during development of monocrotaline-induced pulmonary hypertension in rats Growth Factors 1995 12 121 130 8679246
Berg JT Breen EC Fu Z Mathieu-Costello O West JB Alveolar hypoxia increases gene expression of extracellular matrix proteins and platelet-derived growth factor-B in lung parenchyma Am J Respir Crit Care Med 1998 158 1920 1928 9847287
Archer S Rich S Primary pulmonary hypertension: a vascular biology and translational research "Work in progress" Circulation 2000 102 2781 2791 11094047
Alizadeh AA Eisen MB Davis RE Ma C Lossos IS Rosenwald A Boldrick JC Sabet H Tran T Yu X Powell JI Yang L Marti GE Moore T Hudson JJ Lu L Lewis DB Tibshirani R Sherlock G Chan WC Greiner TC Weisenburger DD Armitage JO Warnke R Levy R Wilson W Grever MR Byrd JC Botstein D Brown PO Staudt LM Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling Nature 2000 403 503 511 10676951 10.1038/35000501
Beer DG Kardia SL Huang CC Giordano TJ Levin AM Misek DE Lin L Chen G Gharib TG Thomas DG Lizyness ML Kuick R Hayasaka S Taylor JM Iannettoni MD Orringer MB Hanash S Gene-expression profiles predict survival of patients with lung adenocarcinoma Nat Med 2002 8 816 824 12118244
Hoshikawa Y Nana-Sinkam P Moore MD Sotto-Santiago S Phang T Keith RL Morris KG Kondo T Tuder RM Voelkel NF Geraci MW Hypoxia induces different genes in the lungs of rats compared with mice Physiol Genomics 2003 12 209 219 12464684
Fink L Seeger W Ermert L Hanze J Stahl U Grimminger F Kummer W Bohle RM Real-time quantitative RT-PCR after laser-assisted cell picking Nat Med 1998 4 1329 1333 9809560 10.1038/3327
Luo L Salunga RC Guo H Bittner A Joy KC Galindo JE Xiao H Rogers KE Wan JS Jackson MR Erlander MG Gene expression profiles of laser-captured adjacent neuronal subtypes Nat Med 1999 5 117 122 9883850 10.1038/4806
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
Fink L Kohlhoff S Stein MM Hanze J Weissmann N Rose F Akkayagil E Manz D Grimminger F Seeger W Bohle RM cDNA array hybridization after laser-assisted microdissection from nonneoplastic tissue Am J Pathol 2002 160 81 90 11786402
Fink L Stahl U Ermert L Kummer W Seeger W Bohle RM Rat porphobilinogen deaminase gene: a pseudogene-free internal standard for laser-assisted cell picking Biotechniques 1999 26 510 516 10090993
Fink L Kinfe T Seeger W Ermert L Kummer W Bohle RM Immunostaining for cell picking and real-time mRNA quantitation Am J Pathol 2000 157 1459 1466 11073806
Livak KJ Schmittgen TD Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method Methods 2001 25 402 408 11846609 10.1006/meth.2001.1262
Leethanakul C Patel V Gillespie J Pallente M Ensley JF Koontongkaew S Liotta LA Emmert-Buck M Gutkind JS Distinct pattern of expression of differentiation and growth-related genes in squamous cell carcinomas of the head and neck revealed by the use of laser capture microdissection and cDNA arrays Oncogene 2000 19 3220 3224 10918578 10.1038/sj.onc.1203703
Ohyama H Zhang X Kohno Y Alevizos I Posner M Wong DT Todd R Laser capture microdissection-generated target sample for high-density oligonucleotide array hybridization Biotechniques 2000 29 530 536 10997267
Sgroi DC Teng S Robinson G LeVangie R Hudson JRJ Elkahloun AG In vivo gene expression profile analysis of human breast cancer progression Cancer Res 1999 59 5656 5661 10582678
Steward GF Jenkins BD Ward BB Zehr JP Development and testing of a DNA macroarray to assess nitrogenase (nifH) gene diversity Appl Environ Microbiol 2004 70 1455 1465 15006766 10.1128/AEM.70.3.1455-1465.2004
Sharov V Kwong KY Frank B Chen E Hasseman J Gaspard R Yu Y Yang I Quackenbush J The limits of log-ratios BMC Biotechnol 2004 4 3 15113428 10.1186/1472-6750-4-3
Heid CA Stevens J Livak KJ Williams PM Real time quantitative PCR Genome Res 1996 6 986 994 8908518
Wilhelm J Pingoud A Real-time polymerase chain reaction Chembiochem 2003 4 1120 1128 14613102 10.1002/cbic.200300662
Talla E Tekaia F Brino L Dujon B A novel design of whole-genome microarray probes for Saccharomyces cerevisiae which minimizes cross-hybridization BMC Genomics 2003 4 38 14499002 10.1186/1471-2164-4-38
Cai Y Han M Luo L Song W Zhou X Increased expression of PDGF and c-myc genes in lungs and pulmonary arteries of pulmonary hypertensive rats induced by hypoxia Chin Med Sci J 1996 11 152 156 9387399
Botney MD Bahadori L Gold LI Vascular remodeling in primary pulmonary hypertension. Potential role for transforming growth factor-beta Am J Pathol 1994 144 286 295 8311113
Hidalgo M Rowinsky EK The rapamycin-sensitive signal transduction pathway as a target for cancer therapy Oncogene 2000 19 6680 6686 11426655 10.1038/sj.onc.1204091
Bultynck G De Smet P Weidema AF Ver Heyen M Maes K Callewaert G Missiaen L Parys JB De Smedt H Effects of the immunosuppressant FK506 on intracellular Ca2+ release and Ca2+ accumulation mechanisms J Physiol 2000 525 Pt 3 681 693 10856121 10.1111/j.1469-7793.2000.t01-1-00681.x
Tilson MD Fu C Xia SX Syn D Yoon Y McCaffrey T Expression of molecular messages for angiogenesis by fibroblasts from aneurysmal abdominal aorta versus dermal fibroblasts Int J Surg Investig 2000 1 453 457 11341602
Oelze M Mollnau H Hoffmann N Warnholtz A Bodenschatz M Smolenski A Walter U Skatchkov M Meinertz T Munzel T Vasodilator-stimulated phosphoprotein serine 239 phosphorylation as a sensitive monitor of defective nitric oxide/cGMP signaling and endothelial dysfunction Circ Res 2000 87 999 1005 11090544
Krause M Dent EW Bear JE Loureiro JJ Gertler FB Ena/VASP proteins: regulators of the actin cytoskeleton and cell migration Annu Rev Cell Dev Biol 2003 19 541 564 14570581 10.1146/annurev.cellbio.19.050103.103356
Hance AJ Robin ED Simon LM Alexander S Herzenberg LA Theodore J Regulation of glycolytic enzyme activity during chronic hypoxia by changes in rate-limiting enzyme content. Use of monoclonal antibodies to quantitate changes in pyruvate kinase content J Clin Invest 1980 66 1258 1264 7440714
McMurtry IF Rounds S Stanbrook HS Studies of the mechanism of hypoxic pulmonary vasoconstriction Adv Shock Res 1982 8 21 33 6291346
Ouiddir A Planes C Fernandes I VanHesse A Clerici C Hypoxia upregulates activity and expression of the glucose transporter GLUT1 in alveolar epithelial cells Am J Respir Cell Mol Biol 1999 21 710 718 10572068
Hardie DG Hawley SA AMP-activated protein kinase: the energy charge hypothesis revisited Bioessays 2001 23 1112 1119 11746230 10.1002/bies.10009
Jiang XC D'Armiento J Mallampalli RK Mar J Yan SF Lin M Expression of plasma phospholipid transfer protein mRNA in normal and emphysematous lungs and regulation by hypoxia J Biol Chem 1998 273 15714 15718 9624168 10.1074/jbc.273.25.15714
Hochachka PW Buck LT Doll CJ Land SC Unifying theory of hypoxia tolerance: molecular/metabolic defense and rescue mechanisms for surviving oxygen lack Proc Natl Acad Sci U S A 1996 93 9493 9498 8790358 10.1073/pnas.93.18.9493
Tinton S Tran-Nguyen QN Buc-Calderon P Role of protein-phosphorylation events in the anoxia signal-transduction pathway leading to the inhibition of total protein synthesis in isolated hepatocytes Eur J Biochem 1997 249 121 126 9363762 10.1111/j.1432-1033.1997.t01-1-00121.x
Rabinovitch M Gamble W Nadas AS Miettinen OS Reid L Rat pulmonary circulation after chronic hypoxia: hemodynamic and structural features Am J Physiol 1979 236 H818 27 443445
Stenmark KR Mecham RP Cellular and molecular mechanisms of pulmonary vascular remodeling Annu Rev Physiol 1997 59 89 144 9074758 10.1146/annurev.physiol.59.1.89
Rabinovitch M Pathobiology of pulmonary hypertension. Extracellular matrix Clin Chest Med 2001 22 433 49, viii 11590839 10.1016/S0272-5231(05)70282-3
Greenway S van Suylen RJ Du Marchie Sarvaas G Kwan E Ambartsumian N Lukanidin E Rabinovitch M S100A4/Mts1 produces murine pulmonary artery changes resembling plexogenic arteriopathy and is increased in human plexogenic arteriopathy Am J Pathol 2004 164 253 262 14695338
Lawrie A Spiekerkoetter E Martinez EC Ambartsumian N Sheward WJ Maclean MR Harmar AJ Schmidt AM Lukanidin E Rabinovitch M Interdependent Serotonin Transporter and Receptor Pathways Regulate S100A4/Mts1, a Gene Associated With Pulmonary Vascular Disease Circ Res 2005 16002749
Yehualaeshet T O'Connor R Green-Johnson J Mai S Silverstein R Murphy-Ullrich JE Khalil N Activation of rat alveolar macrophage-derived latent transforming growth factor beta-1 by plasmin requires interaction with thrombospondin-1 and its cell surface receptor, CD36 Am J Pathol 1999 155 841 851 10487979
Khatua S Peterson KM Brown KM Lawlor C Santi MR LaFleur B Dressman D Stephan DA MacDonald TJ Overexpression of the EGFR/FKBP12/HIF-2alpha pathway identified in childhood astrocytomas by angiogenesis gene profiling Cancer Res 2003 63 1865 1870 12702575
Smirnov SV Robertson TP Ward JP Aaronson PI Chronic hypoxia is associated with reduced delayed rectifier K+ current in rat pulmonary artery muscle cells Am J Physiol 1994 266 H365 70 8304521
Archer SL Wu XC Thebaud B Nsair A Bonnet S Tyrrell B McMurtry MS Hashimoto K Harry G Michelakis ED Preferential expression and function of voltage-gated, O2-sensitive K+ channels in resistance pulmonary arteries explains regional heterogeneity in hypoxic pulmonary vasoconstriction: ionic diversity in smooth muscle cells Circ Res 2004 95 308 318 15217912 10.1161/01.RES.0000137173.42723.fb
Uchida T Rossignol F Matthay MA Mounier R Couette S Clottes E Clerici C Prolonged hypoxia differentially regulates hypoxia-inducible factor (HIF)-1alpha and HIF-2alpha expression in lung epithelial cells: implication of natural antisense HIF-1alpha J Biol Chem 2004 279 14871 14878 14744852 10.1074/jbc.M400461200
Kvietikova I Wenger RH Marti HH Gassmann M The transcription factors ATF-1 and CREB-1 bind constitutively to the hypoxia-inducible factor-1 (HIF-1) DNA recognition site Nucleic Acids Res 1995 23 4542 4550 8524640
Semenza GL Roth PH Fang HM Wang GL Transcriptional regulation of genes encoding glycolytic enzymes by hypoxia-inducible factor 1 J Biol Chem 1994 269 23757 23763 8089148
|
16171515
|
PMC1261535
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Sep 19; 6(1):109
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-109
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1101617908610.1186/1465-9921-6-110ResearchUncoordinated production of Laminin-5 chains in airways epithelium of allergic asthmatics Amin Kawa [email protected] Christer [email protected]éus Lahja [email protected] Kaoru [email protected] Ismo [email protected] Per [email protected] Department of Medical Sciences, Clinical Chemistry, Uppsala University, Uppsala, Sweden2 Department of Medical Sciences, Respiratory Medicine and Allergology, Uppsala University, Uppsala, Sweden3 Division of Cell Biology, Kihara Institute for Biological Research, Yokohama City University, Yokohama, Japan4 Institute of Biomedicine/Anatomy, University of Helsinki, Helsinki, Finland2005 22 9 2005 6 1 110 110 9 2 2005 22 9 2005 Copyright © 2005 Amin et al; licensee BioMed Central Ltd.2005Amin 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
Laminins are a group of proteins largely responsible for the anchorage of cells to basement membranes. We hypothesized that altered Laminin chain production in the bronchial mucosa might explain the phenomenon of epithelial cell shedding in asthma. The aim was to characterize the presence of Laminin chains in the SEBM and epithelium in allergic and non-allergic asthmatics.
Patients and methods
Biopsies were taken from the bronchi of 11 patients with allergic and 9 patients with non-allergic asthma and from 7 controls and stained with antibodies against the Laminin (ln) chains alpha1-alpha5, beta1-beta2 and gamma1-gamma2.
Results
Lns-2,-5 and -10 were the main Laminins of SEBM. The layer of ln-10 was thicker in the two asthmatic groups while an increased thickness of lns-2 and -5 was only seen in allergic asthmatics. The ln gamma2-chain, which is only found in ln 5, was exclusively expressed in epithelial cells in association with epithelial injury and in the columnar epithelium of allergic asthmatics.
Conclusion
The uncoordinated production of chains of ln-5 in allergic asthma could have a bearing on the poor epithelial cell anchorage in these patients.
==== Body
Background
Asthma is a chronic inflammatory disease of the lungs that may have allergic or non-allergic causes [1-3]. The allergic type of asthma is characterized by the accumulation of eosinophils, mast cells and lymphocytes of the Th2-type in the bronchial mucosa, whereas the non-allergic asthma has a substantial accumulation of neutrophils in addition to eosinophils and mast cells [3]. Structural changes and remodelling of the bronchial mucosa with signs of epithelial injury, subepithelial basement thickening, smooth muscle hypertrophy, increased vascularization and innervation are prominent features of the allergic type of asthma and less prominent in the non-allergic type [3].
Basement membranes (BMs) are built of cell-polymerizing networks of type IV collagens and laminins connected by nidogen/entactin [4,5]. The major role of laminin for epithelial cells is to anchor them to BM for cell differentiation and maintenance of cell function. Laminins are heterotrimeric molecules made up by one α, one β and one γ chain. Until today we know of five α-chains, three β-chains and three γ-chains. These chains combine into at least 14 different Laminins (lns) i.e. lns 1–14. The distribution of these Laminin isoforms varies between tissues, but in most BMs more than one Laminin is present. The chains of laminins have different regions that function by binding to cellular receptor molecules among which the most abundant are integrins, dystroglycan and the recently characterized Lutheran blood group antigen [4,6]. Several studies have shown the fundamental importance of intact Laminins in the BMs, since mutations may give rise to serious diseases such as epidermolysis bullosa in which the anchoring of the skin is grossly impaired [7]. Laminins also interact with many other cells and promote migration and angiogenesis and their functions in tumour invasion is one of the hot research topics of today [4].
The injury of the respiratory epithelium in the bronchi in allergic asthmatics may be one of the mechanisms underlying bronchial hyperresponsiveness which is one of the main features of asthma [8-11]. The mechanisms behind the fragility of the epithelium in allergic asthmatics, i.e. the propensity of the epithelium to shed from its anchorage to the subepithelial basement membrane (SEBM) and basal cells have not been explained. Since one obligatory component in this anchoring process is mediated by Laminins, we hypothesized that uncoordinated production of Laminin chains might contribute to weaken these anchoring forces. Our aim was therefore to describe the presence of the various Laminins in the epithelium and especially SEBM of allergic asthmatics in comparison with non-allergic asthmatics and healthy non-asthmatic controls.
Materials and methods
Subjects
Bronchial biopsies were collected from twenty-nine non-smoking adults divided into the following groups: healthy controls (n = 7), patients with allergic asthma (n = 11) and patients with non-allergic asthma (n = 9) (Table 1). All patients had a clinical asthma diagnosis, current asthma symptoms and increased responsiveness to inhaled methacholine. The allergic asthma patients all had a positive skin prick test (≥ 3 mm) for at least one common allergen (birch, timothy grass (Phleum pratense), mugwort (Artemisia vulgaris), cat, dog, horse, house dust mite (Dermatophagoides pteronyssinus), Cladosporium, and Alternaria.) while the non-allergic asthma patients and the controls all had a negative skin prick test. All patients with allergic asthma were examined outside the birch and grass pollen season (April to August).
Table 1 Patient characteristics (n or median (range))
Healthy control (n = 7) Allergic asthma (n = 11) Non-allergic asthma (n = 9)
Age (yr) 25 (22–43) 37 (29–63) 41 (17–62)
Sex (M/F) 2/5 2/9 2/7
FEV1 (% pred) 98 (71–120) 94 (72–109) 86 (72–97)
FVC (% pred) 98 (78–109) 100 (86–118) 87 (76–96)
Symptom score * 0 (0–1) 2 (0–4) 2 (1–2)
PEF-variability (%) 5 (3–9) 11 (6–22) 10 (5–20)
PC20 (mg/ml) - 2.7 (0.07–32) 8.7 (1.0–32)
Pollen allergy 0 9/11 0
Pet allergy 0 11/11 0
Mite allergy 0 4/11 0
Mould allergy 0 3/11 0
*number of symptoms recorded in a questionnaire during 2 weeks (9)
All but one allergic and one non-allergic patient with asthma were on regular treatment with inhaled glucocorticosteroids (budesonide 200–800 g/day) and inhaled β2-agonists as needed. The average use of inhaled glucocorticosteroids was similar in the two asthma groups. A more detailed description of the study population has been presented in a previous report[3].
Bronchoscopy
The patients were given 10 mg diazepam (Stesolid®, Dumex, Copenhagen, Denmark) orally and 0.5 mg atropine (Atropin, NM Pharma, Stockholm, Sweden) subcutaneously 30 minutes before the investigation. The upper airways were anaesthetized with lidocain hydrochloride (Xylocain, Astra, Södertälje, Sweden). Using a flexible fibre bronchoscope (Olympus P 20D) with a FB 15C 2,0 mm forceps (Olympus), two biopsies were taken from each of three different airway levels in the right lung: (A) in the upper lobe bronchus immediately after the division from the main bronchus, (B) at the division between the middle and lover lobe bronchi and (C) from the main lower lobe divisions. The specimens were immediately examined under a light microscope to ensure the presence of a complete mucosa and fixed as described below. The patients were instructed to take their regular asthma sprays the morning of the bronchoscopy.
The study was conducted in accordance with the Declaration of Helsinki and was approved by the ethics committee at the Faculty of Medicine at the University of Uppsala.
Immunohistochemistry
The expression of different Laminin chains in the epithelium and in the subepithelial basement membrane was studied in frozen sections by the use of monoclonal antibodies and the alkaline phosphate-anti-alkaline phosphatase method (APAAP) visualization system. Mouse monoclonal antibodies (MAbs) against the ln α1 chain (clone 161 EB7)[12], ln α4 chain (clone 168FC10)[13], ln α5 (clone 4C7) [14,15], ln β1 chain (clone 114DG.)[16], ln β2 chain (clone C4, obtained from the hybridoma C4 developed by Joshua Sanes obtained from the Developmental Studies hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242), ln γ1 chain (113BC7), ln γ2 chain (clone D4B5) [17] were produced and characterized as described earlier. MAb against lns α2 and α3 chains (clones Lam M and P3H9-2, respectively) were purchased from Chemicon International, Temecula, California, USA.
The bronchial biopsy specimens were taken from the upper lobe and frozen immediately in melting propane previously cooled in liquid nitrogen and further processed as described in detail previously [3]. Incubation with antibodies to ln α1 chain (diluted 1:4 in PBS), ln α2 chain (diluted 1:500 in PBS), ln α3 chain (diluted 1:1000 in PBS) ln α4 chain (diluted 1:5 in PBS), ln α5 (diluted 1:300 in PBS), ln β1 chain (diluted 1:400 in PBS), ln β2 chain (diluted 1:750 in PBS), ln γ1 chain (diluted 1:400 in PBS) and ln γ2 chain (diluted 1:1000 in PBS) were performed at room temperature in a humid chamber for 20 h and terminated by two rinses in PBS. The bound antibodies were visualized with the alkaline phosphates-anti-alkaline phosphatase method (APAAP kit K670, Dakocytomation, Glostrup, Denmark) using fast red substrate. The sections were counterstained with Mayer's hematoxylin (Merck; Darmstadt, Germany) for two minutes and mounted with Faramount (S 3025, Dako). In the negative controls, the primary antibodies were omitted.
Microscopic Evaluation of Sections
The Leica DMLB microscope (Wetzlar GmbH, Germany) was equipped with a Leica Microsystems digital camera (DC 300F) connected to a PC-computer. The images were captured and saved in the computer for further evaluation using the software package Qwin v2.7. In each biopsy two subsequent sections were evaluated. The thickness of the Laminin layers (in μm) was measured in immunolabeled frozen sections using a X10 objective and the computerized image analysis system after calibration with the aid of a stage micrometer. Measurements were carried out on 100 randomly selected sites per section and the results expressed as the mean of these measurements. The variation in estimation of structural changes between the two microscopic sections varied between 4–8% (% coefficient of variation). All slides were assessed by an observer blinded to the diagnosis of the patient.
Statistics
All statistics were calculated using non-parametric tests. Comparisons between the three groups were initially performed by means of analysis of variance (Kruskal-Wallis test). In case of significance paired group comparisons were performed with the Mann-Whitney U-test. A p-value of <0.05 was regarded as statistically significant.
Results
Ln α-chain
With MAb against ln α1-chain we saw some weak patchy staining of the SEBM in biopsies from patients with allergic asthma, but not in the biopsies of healthy subjects or of subjects with non-allergic asthma. Alfa2-chains were found in the SEBM and were significantly thicker in biopsies from allergic asthmatics as compared to non-allergic asthmatics and healthy controls (table 2). The α3-chain was also found in the SEBM and the layer was significantly thicker in biopsies from allergic asthmatics compared to non-allergic asthmatics and controls (Figure 1, table 2). No staining of the epithelium was discerned with the antibodies against the α3-chain. No staining of the SEBM or epithelium was found with antibodies against α4-chains. The staining of SEBM with antibodies against the α5-chain showed a thicker layer in the allergic than in the non-allergic asthmatics. The layer α5 was also thicker in the non-allergic asthmatics than in the controls (Figure 2).
Table 2 The thickness of various laminin chains in SEBM (μm)
Laminin chain Healthy controls Allergic asthma Non-allergic asthma
α1-chain No staining Patchy, weak staining No staining
α2-chain 1.94 (1.70–2.20) 2.83 (2.50–3.30)**, ‡‡‡ 2.19 (1.8–2.90)
α3-chain 2.46 (1.80–3.10) 3.77 (3.30–4.40)***, ‡‡‡ 2.61 (2.20–3.10)
α4-chain No staining No staining No staining
α5-chain 2.31 (1.90–2.50) 4.10 (3.30–4.60)***, ‡‡‡ 2.86 (2.40–3.50) **
β1-chain 2.13 (1.60–2.80) 4.84 (4.10–5.50)***, ‡‡‡ 3.29 (3.10–3.80) ***
β2-chain 1.93 (1.60–2.60) 2.34 (2.10–2.800)*, ‡‡ 2.04 (1.80–2.300)
γ1-chain 2.47 (1.90–3.00) 4.86 (4.20–5.40) ***, ‡‡‡ 3.49 (2.90–3.90) ***
γ2-chain 2.03 (1.70–2.40) 2.96 (2.60–3.50)**, ‡‡‡ 2.36 (2.00–2.80)
Results are given as medians and interquartile ranges. Differences between groups were calculated by the Mann-Whitney non-parametric test and statistical differences shown in the table. **, *** indicate p < 0.01 and p < 0.001, respectively, in the comparison between either of the two asthma groups with the results of the healthy controls. ‡, ‡‡, ‡‡‡ indicate p < 0.05, <0.01 and 0.001, respectively, in the comparison between the two asthma groups.
Figure 1 Cryostat sections of bronchial biopsies stained with antibodies against the ln α3 chain. Allergic asthma (A), non-allergic asthma (B) and healthy control (C) (original magnification ×170). The comparison of the thickness of SEBM is shown and the significant differences between the groups shown in the figure. Mayer's hematoxylin.
Figure 2 Cryostat sections of bronchial biopsies stained with antibodies against the ln α5 chain. Allergic asthma (A), non-allergic asthma (B) and healthy control (C) (original magnification ×170). The comparison of the thickness of SEBM is shown and the significant differences between the groups shown in the figure. Mayer's hematoxylin.
Ln β-chains
Ln β1-chains were seen in the SEBM and the thickness was significantly higher in biopsies from allergic asthmatics as compared to non-allergics and healthy controls (Table 2). The staining of SEBM with antibodies against the β2-chain showed a slight increase of the thickness in allergic asthmatics as compared to non-allergic asthmatics and controls, but with no difference between non-allergic asthmatics and controls (Table 2).
Ln γ-chains
Staining of the biopsy with antibodies against the ln γ1-chain revealed an increased thickness in the SEBM in both allergic and non-allergic asthma as compared to controls and a much thicker layer in allergic asthmatics as compared to non-allergics (Figure 3). The staining of the bronchial mucosa with antibodies against the γ2-chain revealed a thicker layer in the SEBM in allergic asthmatics as compared to both non-allergic asthmatics and controls (Figure 4, Table 2). The antibodies also stained the epithelium. Thus, as shown in the figure the staining was found both in the apical part of the columnar epithelium and in the basal cells. Staining of the apical part of the columnar epithelium was only found in intact epithelium from allergic asthmatics, whereas staining of the basal cells was seen in all three groups in areas of epithelial injury. A close correlation was also found between the epithelial integrity in the three study groups and the thickness of the ln γ2-chain (figure 5).
Figure 3 Cryostat sections of bronchial biopsies stained with antibodies against the ln γ1 chain. Allergic asthma (A), non-allergic asthma (B) and healthy control (C) (original magnification ×170). The comparison of the thickness of SEBM is shown and the significant differences between the groups shown in the figure. Mayer's hematoxylin.
Figure 4 Cryostat sections of bronchial biopsies stained with antibodies against the ln γ2 chain. Allergic asthma (A), non-allergic asthma (B) and healthy control (C) (original magnification ×170). As shown in the figure epithelial staining was found both in the apical part of the columnar epithelium and in the basal cells. Staining of the apical part of the columnar epithelium was only found in intact epithelium from allergic asthmatics. The comparison of the thickness of SEBM is shown and the significant differences between the groups shown in the figure. Mayer's hematoxylin.
Figure 5 The relationship between epithelial integrity (%) in the bronchial mucosa and the thickness of the ln γ2-chain (μm). Overall there was a negative correlation between the epithelial integrity and the ln γ2-chain thickness (r = -0.73, p < 0.001), whereas no significant correlations were observed within the respective patient group. Diamonds represent allergic asthmatics, squares represent non-allergic asthmatics and closed circles healthy controls.
When combining the information on individual laminin chains, we found that lns-2 (α2β1γ1),-5 (α3β3γ2) and -10 (α5β1γ1) were the main laminins of the SEBM. The layer of ln-10 was thicker in the two asthmatic groups while an increased thickness of lns-2 and -5 was only seen in allergic asthmatics. The staining of ln γ2-chain in the absence of ln α3 in epithelial cells does not fit with any presently known laminin.
Discussion
This study has systematically investigated the presence of most Laminin chains in the epithelium and SEBM of both allergic and non-allergic asthmatics. The primary object was to test the hypothesis that differences in Laminin chain compositions in SEBM might help explain the phenomenon of epithelial fragility and shedding as is typically seen in biopsies of allergic asthmatics [3]. We found several alterations in the Laminin chain composition in the SEBM of allergic asthmatics. Most of these differences seemed quantitative rather than qualitative. Unexpectedly we found distinct qualitative differences with respect to ln-5 chain compositions that may have a bearing on the poorer anchorage of epithelial cells to BM in allergic asthma. The finding of a close correlation between ln γ2-chain deposition and epithelial injury indeed emphasises the close relationship between laminin chain production and epithelial injury as is observed in certain subjects with asthma. However, the relationship does not tell us whether epithelial shedding is a cause of the uncoordinated laminin chain production or whether the uncoordinated production is a consequence of the repair processes induced by epithelial injury.
The unexpected findings were both related to the ln γ2-chain staining. According to the present knowledge the γ2-chain is only found as part of ln-5 (α3β3γ2) [4,18]. In our biopsies we found exclusive staining with the antibodies against the γ2-chain in the epithelium, with no sign of simultaneous staining with antibodies against the α3-chain. The epithelium staining could be a reflection of the fact that airways epithelium is a producer of the ln γ2-chain and that the staining reflects the deposition of non-secreted protein. It was of particular interest that we found accumulation of immunoreactivity in the apical part of intact columnar epithelium in allergic asthmatics, but not in the other two groups, whereas staining of the basal cells were seen in all three groups in areas of epithelial destruction. This staining was observed without any concomitant staining of the complementary ln-5 α3 chain, which suggests an uncoordinated production of the γ2 chain in the epithelium of allergic asthmatics resulting in the intracellular accumulation of the γ2 chain, since the extracellular secretion probably requires the assembly of the heterotrimeric molecule [18]. The staining pattern could also indicate the presence of a hitherto unrecognised Laminin or alternatively that the α3 chain had been proteolytically modified with the loss of the particular epitope recognized by our monoclonal antibodies [19].
The intense staining of the cytoplasm of basal cells in areas of epithelial injury suggests that the basal cells are producers of the ln γ2 chain and probably also of the whole heterotrimeric complex of ln-5, although no staining of the other two chains was observed. Indeed, sole expression of ln γ2 chain has been shown in invading tumour cells[4,20], which shows that uncoordinated production of the three ln-5 chains may take place under certain conditions. It is also of interest that Lappi-Blanco et al. in a recent report found ln γ2 chain expression to be increased in regenerating epithelial cells and also found γ2 chain in basal cells of normal bronchus [21]. These results suggest that our findings of intense staining seen in the basal cells at areas of tissue injury may be a sign of re-epithelialization and repair.
The SEBM showed the presence of mainly three Laminins i.e. ln-5 (α3β3γ2), ln-10 (α5β1γ1) and ln-2 (α2β1γ1). The two former were expected based on earlier findings [4,22], whereas the presence of ln-2 mostly is associated with BMs surrounding tissues such as muscles and nerves [23]. In a previous report we indicated the wide presence of α1-chains in SEBM, which is seemingly contrasted by the present results[24]. Those results, however, were based on the false assumption that the monoclonal antibody 4C7 specifically recognizes α1-chains, which is not the case. The 4C7 antibody only recognizes the α5-chains [14,15]. As was previously found the thickness of the Laminin layer in the SEBM was increased in allergic asthmatics as compared to both non-allergic asthmatics or healthy controls [3]. This difference was most obvious for ln-10, since also the thickness found in non-allergic asthmatics was increased as compared to healthy non-asthmatic controls. This was contrasted by the increased thickness of ln-5 and -2, which was only seen in allergic asthmatics. These differences, therefore suggest qualitative differences in the production of various chains in allergic and non-allergic asthmatics, which may relate to the differences in the inflammatory processes going on in these two diseases. The allergic asthma being eosinophil-mast cell-Th2 driven and the non-allergic asthma being more neutrophil-mast cell driven, although eosinophils are also present at increased amounts in the non-allergic asthma [3].
As mentioned above the primary aim of this work was to test the hypothesis that an imbalance in Laminin chain production might explain the observed epithelial cell loss in allergic asthmatics. This hypothesis is seemingly refuted by our data, since others have shown that ln-5 induces the formation of hemidesmosomes [25], which actually promote stable cell:matrix adhesion. Another interesting property of ln-5 is the biological activity of the proteolytically modified fragments, which might modify cellular behaviours [26]. However, it should also be noted that mutations or modifications of any of the chains of ln-5 are associated with severe disease due to separation of epithelia from the underlying basement membrane [7]. Thus, we cannot exclude any processing of ln-5 in the inflamed tissue of allergic asthma as an explanation of poor anchorage of the epithelial cells in the bronchi to the underlying basement membrane.
Conclusion
We conclude that Laminin chain deposition in the epithelium and SEBM of allergic and non-allergic asthmatics differs in qualitative and quantitative terms and that there is a close relationship between ln γ2-chain deposition and epithelial injury. The uncoordinated production of the chains of ln-5 in the epithelium of allergic asthmatics may be of particular interest, since ln-5 promotes the formation of hemidesmosomes, which promote stable cell matrix adhesion.
Abbreviations
alkaline phosphate-anti-alkaline phosphatase (APAAP), basement membranes (BMs), Laminin (ln), Mouse monoclonal antibodies (Mabs), subepithelial basement membrane; (SEBM),
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Kawa Amin and Lahja Sevéus have done the immunohistochemistry part of the study and also been involved in the evaluation of the data
Christer Janson has been responsible for recruiting the patients and has been involved in the evaluation of the data
Per Venge initiated the study and has been the principal author of the paper
Ismo Virtanen and Kaoru Miyazaki have provided the unique antibodies and also been involved in the evaluation of the data
Acknowledgements
The participation of members of the BHR-group is appreciated.
This study was supported by grants from the Swedish Heart and Lung Foundation, the Care and Allergy Foundation (Vårdalstiftelsen) and the Swedish Allergy and Asthma Foundation.
==== Refs
Holgate ST The epidemic of allergy and asthma Nature 1999 402 B2 B4 10586888 10.1038/35037000
Busse WW Rosenwasser LJ Mechanisms of asthma J Allergy Clin Immunol 2003 111 S799 S804 12618746 10.1067/mai.2003.158
Amin K Ludviksdottir D Janson C Nettelbladt O Bjornsson E Roomans GM Boman G Seveus L Venge P Inflammation and structural changes in the airways of patients with atopic and nonatopic asthma Am J Respir Crit Care Med 2000 162 2295 2301 11112154
Patarroyo M Tryggvason K Virtanen I Laminin isoforms in tumor invasion, angiogenesis and metastasis Semin Cancer Biol 2002 12 197 207 12083850 10.1016/S1044-579X(02)00023-8
Libby RT Champliaud MF Claudepierre T Xu Y Gibbons EP Koch M Burgeson RE Hunter DD Brunken WJ Laminin expression in adult and developing retinae: evidence of two novel CNS laminins J Neurosci 2000 20 6517 6528 10964957
Kikkawa Y Moulson CL Virtanen I Miner JH Identification of the binding site for the Lutheran blood group glycoprotein on laminin alpha 5 through expression of chimeric laminin chains in vivo J Biol Chem 2002 277 44864 44869 12244066 10.1074/jbc.M208731200
McGowan KA Marinkovich MP Laminins and human disease Microsc Res Tech 2000 51 262 279 11054876 10.1002/1097-0029(20001101)51:3<262::AID-JEMT6>3.0.CO;2-V
Fireman P Understanding asthma pathophysiology Allergy Asthma Proc 2003 24 79 83 12776439
Ludviksdottir D Janson C Bjornsson E Stalenheim G Boman G Hedenstrom H Venge P Gudbjornsson B Valtysdottir S Different airway responsiveness profiles in atopic asthma, nonatopic asthma, and Sjogren's syndrome. BHR Study Group. Bronchial hyperresponsiveness Allergy 2000 55 259 265 10753017 10.1034/j.1398-9995.2000.00252.x
Crimi E Spanevello A Neri M Ind PW Rossi GA Brusasco V Dissociation between airway inflammation and airway hyperresponsiveness in allergic asthma Am J Respir Crit Care Med 1998 157 4 9 9445270
Kushima A Motojima S Yamai T Makino S The participation of epithelial desquamation in the increase of bronchial hyperresponsiveness after antigen challenge in patients with bronchial asthma Allergy (Japan) 1990 39 No.12 1581 1589
Virtanen I Gullberg D Rissanen J Kivilaakso E Kiviluoto T Laitinen LA Lehto VP Ekblom P Laminin alpha1-chain shows a restricted distribution in epithelial basement membranes of fetal and adult human tissues Exp Cell Res 2000 257 298 309 10837144 10.1006/excr.2000.4883
Petajaniemi N Korhonen M Kortesmaa J Tryggvason K Sekiguchi K Fujiwara H Sorokin L Thornell LE Wondimu Z Assefa D Patarroyo M Virtanen I Localization of laminin alpha4-chain in developing and adult human tissues J Histochem Cytochem 2002 50 1113 1130 12133914
Engvall E Earwicker D Haaparanta T Ruoslahti E Sanes JR Distribution and isolation of four laminin variants; tissue restricted distribution of heterotrimers assembled from five different subunits Cell Regul 1990 1 731 740 2099832
Tiger CF Champliaud MF Pedrosa-Domellof F Thornell LE Ekblom P Gullberg D Presence of laminin alpha5 chain and lack of laminin alpha1 chain during human muscle development and in muscular dystrophies J Biol Chem 1997 272 28590 28595 9353324 10.1074/jbc.272.45.28590
Virtanen I Lohi J Tani T Korhonen M Burgeson RE Lehto VP Leivo I Distinct changes in the laminin composition of basement membranes in human seminiferous tubules during development and degeneration Am J Pathol 1997 150 1421 1431 9094997
Mizushima H Koshikawa N Moriyama K Takamura H Nagashima Y Hirahara F Miyazaki K Wide distribution of laminin-5 gamma 2 chain in basement membranes of various human tissues Horm Res 1998 50 Suppl 2 7 14 9721586 10.1159/000053118
Tunggal P Smyth N Paulsson M Ott MC Laminins: structure and genetic regulation Microsc Res Tech 2000 51 214 227 11054872 10.1002/1097-0029(20001101)51:3<214::AID-JEMT2>3.0.CO;2-J
Hintermann E Quaranta V Epithelial cell motility on laminin-5: regulation by matrix assembly, proteolysis, integrins and erbB receptors Matrix Biol 2004 23 75 85 15246107 10.1016/j.matbio.2004.03.001
Katayama M Sekiguchi K Laminin-5 in epithelial tumour invasion J Mol Histol 2004 35 277 286 15339047 10.1023/B:HIJO.0000032359.35698.fe
Lappi-Blanco E Kaarteenaho-Wiik R Salo S Sormunen R Maatta M Autio-Harmainen H Soini Y Paakko P Laminin-5 gamma2 chain in cryptogenic organizing pneumonia and idiopathic pulmonary fibrosis Am J Respir Crit Care Med 2004 169 27 33 14500258 10.1164/rccm.200210-1234OC
Virtanen I Laitinen A Tani T Paakko P Laitinen LA Burgeson RE Lehto VP Differential expression of laminins and their integrin receptors in developing and adult human lung Am J Respir Cell Mol Biol 1996 15 184 196 8703474
Patton BL Laminins of the neuromuscular system Microsc Res Tech 2000 51 247 261 11054875 10.1002/1097-0029(20001101)51:3<247::AID-JEMT5>3.0.CO;2-Z
Altraja A Laitinen A Virtanen I Kämpe M Simonsson BG Karlsson SE Håkansson L Venge P Sillastu H Laitinen LA Expression of laminins in the airways in various types of asthmatic patients: A morphometric study Am J Respir Cell Mol Biol 1996 15 482 488 8879182
Michelson PH Tigue M Jones JC Human bronchial epithelial cells secrete laminin 5, express hemidesmosomal proteins, and assemble hemidesmosomes J Histochem Cytochem 2000 48 535 544 10727295
Ghosh S Stack MS Proteolytic modification of laminins: functional consequences Microsc Res Tech 2000 51 238 246 11054874 10.1002/1097-0029(20001101)51:3<238::AID-JEMT4>3.0.CO;2-3
|
16179086
|
PMC1261536
|
CC BY
|
2021-01-04 16:23:26
|
no
|
Respir Res. 2005 Sep 22; 6(1):110
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-110
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1111618535610.1186/1465-9921-6-111ResearchE1A expression dysregulates IL-8 production and suppresses IL-6 production by lung epithelial cells van den Berg Arjen [email protected] Mieke [email protected] Henk M [email protected] René [email protected] Department of Pulmonology, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands2 Department of Experimental Immunology, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands2005 26 9 2005 6 1 111 111 16 6 2005 26 9 2005 Copyright © 2005 van den Berg et al; licensee BioMed Central Ltd.2005van den Berg 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 adenoviral protein E1A has been proposed to play a role in the pathophysiology of COPD, in particular by increasing IL-8 gene transcription of lung epithelial cells in response to cigarette smoke-constituents such as LPS. As IL-8 production is also under tight post-transcriptional control, we planned to study whether E1A affected IL-8 production post-transcriptionally. The production of IL-6 by E1A-positive cells had not been addressed and was studied in parallel. Based on our previous work into the regulation of IL-8 and IL-6 production in airway epithelial cells, we used the lung epithelial-like cell line NCI-H292 to generate stable transfectants expressing either E1A and/or E1B, which is known to frequently co-integrate with E1A. We analyzed IL-8 and IL-6 production and the underlying regulatory processes in response to LPS and TNF-α.
Methods
Stable transfectants were generated and characterized with immunohistochemistry, western blot and flow cytometry. IL-8 and IL-6 protein production was measured by ELISA. Levels of IL-8 and IL-6 mRNA were measured using specific radiolabeled probes. EMSA was used to assess transcriptional activation of relevant transcription factors. Post-transcriptional regulation of mRNA half-life was measured by Actinomycin D chase experiments.
Results
Most of the sixteen E1A-expressing transfectants showed suppression of IL-6 production, indicative of biologically active E1A. Significant but no uniform effects on IL-8 production, nor on transcriptional and post-transcriptional regulation of IL-8 production, were observed in the panel of E1A-expressing transfectants. E1B expression exerted similar effects as E1A on IL-8 production.
Conclusion
Our results indicate that integration of adenoviral DNA and expression of E1A and E1B can either increase or decrease IL-8 production. Furthermore, we conclude that expression of E1A suppresses IL-6 production. These findings question the unique role of E1A protein in the pathophysiology of COPD, but do not exclude a role for adenoviral E1A/E1B DNA in modulating inflammatory responses nor in the pathogenesis of COPD.
==== Body
Background
Chronic obstructive pulmonary disease (COPD) is characterized by chronic inflammation and irreversible airflow obstruction [1], and is associated with cigarette smoking. Not all smokers, however, develop COPD. Hogg and co-workers have put forward the concept that the presence of the adenoviral E1A DNA and protein in airway structural cells, leading to enhanced IL-8 production in response to endotoxin exposure, may be related to the development of COPD. First they showed by PCR analysis that lung tissue from COPD patients contained more E1A DNA than lung tissue from matched non-COPD smokers [2]. Subsequently, and in line with the presence of E1A DNA, E1A protein was found to be expressed in airway and alveolar epithelial cells from COPD patients [3]. Based on the model of genomic integration of adenoviral DNA proposed by Graham [4], it is likely that a second early adenoviral gene, E1B, is also integrated and possibly expressed, but this was not investigated. As E1A interacts with a large number of regulatory proteins and as epithelial cells express inflammatory proteins, it was postulated that E1A protein modifies the expression of these inflammatory proteins. Stable E1A transfectants of alveolar type-II-like A549 cells indeed showed increased IL-8 [5] and ICAM-1 [6] expression specifically in response to LPS, which is a major constituent of cigarette smoke. Similar results were obtained with E1A-transformed bronchial epithelial cells transfected with both the E1A and E1B gene [7]. The in vivo relevance of E1A expression was illustrated with a guinea pig model, showing an enhanced inflammatory response to cigarette smoke in animals with lung tissue containing E1A DNA [8].
The increased in vitro production of IL-8 and ICAM-1 by E1A transfectants to LPS appeared to be dependent on an enhanced transcriptional activity, involving activation of NFκB [5,7]. Our previous studies into the regulation of IL-8 and IL-6 production by lung epithelial-like NCI-H292 cells indicated that, besides transcriptional regulation through NFκB, AP-1 and C/EBP activation, post-transcriptional regulation, exemplified by a modified mRNA degradation, was a major means of regulating IL-8 and IL-6 responses [9-11]. Similar findings were obtained with another lung epithelial cell line, Calu-3 cells, as well as primary bronchial epithelial cells. In fact, lung epithelial cells with a decreased rate of mRNA degradation displayed hyperresponsive IL-8 and IL-6 production, similar to that observed for E1A-expressing A549 cells [5]. Therefore, we hypothesized that, expression of E1A in lung epithelial-like cells may lead to stabilization of IL-8 mRNA paralleled by increased IL-8 production in response to LPS and to TNF-α. In parallel, we analyzed IL-6 production, which was anticipated to be decreased, as E1A is known to inhibit IL-6 transcription [13].
To investigate this, we generated stable transfectants of NCI-H292 cells expressing E1A. As the E1B gene is frequently co-integrated with that of E1A, and as E1B protein modifies E1A functions, we also generated stable E1A- and E1B-transfectants. Stable transfectants of E1B and of the empty vector expressing green fluorescent protein (GFP)-tagged zeocin-resistance protein served as controls.
Materials and methods
Constructs
In order to construct pTracerSV40-ZeocinGFP vectors (Invitrogen, Paisley, UK) expressing E1A, E1B, or both proteins, pAt153-Xho [14] (a kind gift from Dr. Robert Vries, LUMC, Department of Molecular Cell Biology, Leiden, The Netherlands), containing the first 5789 bp of the Ad5 genome was used as donor construct. To construct the vector pTracer-E1A, containing E1A only, pAt153-xho was digested with Sst1 and the 7K fragment, containing E1A and vector sequence, was isolated and ends were blunted with T4 DNA polymerase. Subsequently this fragment was digested with EcoR1 and the 1774 bp fragment was isolated and ligated into EcoR1/EcoRV digested pTracer-SV40.
To construct the pTracer-E1B, containing only E1B (both 19K- and 55K-E1B proteins), pAt153-Xho was digested with HPA1 and APA1 and the isolated fragment was ligated into pTracer-SV40. The vector pTtracer-E1AB was constructed by digestion of pAt153-Xho with EcoR1 and APA1 and subsequent ligation.
All constructs were verified by sequencing (BigDye sequencing kit, ABI, Foster City, CA) and restriction analysis.
Generation of stable clones
Human lung mucoepidermoid carcinoma derived NCI-H292 cells (CRL 1848; American Type Culture Collection (ATCC), Manassas, VA) were grown to 90% confluency in a 75 cm2 culture flask, as described before [11]. Before transfection, growth medium was replaced with 10 ml medium without penicillin and streptomycin. Twenty-five μg of vector DNA was mixed with 60 μl of Lipofectamine 2000 (Invitrogen) in a volume of 3 ml Optimem-1 (Invitrogen) and layered onto NCI-H292 cells resulting in 20–30% GFP-positive cells after 24 h. Stable clones were obtained by selection in medium containing 100 μg/ml Zeocin (Invitrogen), of which 250 μl/well was plated in 48-wells plates at a concentration of 6 cells/ml. Medium was replaced twice a week. After formation of colonies, screening of clones was initially performed by immunohistochemistry in 96-well plates. Clones positive for E1A or E1B were selected and expression of E1A or E1B was confirmed by Western blot. Clones negative for E1A or E1B as determined by immunohistochemistry were discontinued. Monoclonality of clones was tested by flowcytometry analyzing both GFP and E1A.
Cell culture
Clones were cultured and propagated as described before for NCI-H292 cells [11], with the exception that 100 μg/ml Zeocin was added to the medium. Before experiments, cells were cultured one week without Zeocin, and all experiments were performed in Zeocin-free medium.
For cytokine release, 6 × 105 cells were plated and grown overnight in 500 μl in 24-well plates. For isolation of mRNA and nuclear extracts, 30 × 105 cells were plated and grown overnight in 2.5 ml in 6-well plates.
Immunohistochemistry
Cells were fixed with 4% (v/v) paraformaldehyde/0.1% (w/v) Saponin and subsequently incubated with ice-cold methanol for 1 min. To prevent non-specific signals, cells were incubated for 1 h with 5% (w/v) BSA containing 0.1% (w/v) sodiumazide and 0.1% (v/v) H2O2 and 0.1% Saponin. Then, the cells were incubated with primary antibody (M73 (Santa Cruz Biotechnologies, Santa Cruz, CA) for E1A, 1G11 for E1B-19K or 9C10 E1B-55K [15], both kind gifts from Dr. Robert Vries (LUMC, Leiden, Netherlands) diluted 1:500 in PBS/0.5%BSA/0.1% Saponin (PBSAP) overnight at 4°C. Next, cells were washed 3× with PBS/0.1% Saponin and incubated with 1:250 biotinylated goat-anti-mouse-IgG in PBSAP (DakoCytomation Glostrup, Denmark) for 1 h followed by 3 wash steps. Finally, cells were incubated with Streptavidin-Horseradish Peroxidase (HRP, DakoCytomation) (1:250 in PBSAP) for 30 min, washed and developed with AEC staining solution (Vector Laboratories, Burlingame, CA).
E1A FACS
Cells were trypsinized and fixed with 4% paraformaldehyde/0.1% Saponin. To prevent aspecific binding, cells were incubated for 30 min with PBS/5% BSA/0.1% Saponin. Then, cells were incubated with the primary antibody M73 (1:500 in PBSAP) for 30 min and subsequently with biotinylated goat anti-mouse IgG (1:250 in PBSAP) for 30 min. Next, cells were labeled with Streptavidin-Allophycocyanin (DakoCytomation, 1:100 in PBSAP) and analyzed by a FACSCalibur flow cytometer and CellQuest Pro software (BD Biosciences). All incubations were performed at RT on a shaking platform. Between incubations, cells were washed twice with PBSAP.
Western blot
Lysates were prepared by scraping cells (± 5 × 106) in lysis buffer (1% (w/v) NP40, 10 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 1 mM phenylmethylsulphonyl-fluoride (PMSF)). Lysates were cleared by centrifugation at 13,000 g for 15 min. Protein contents in cell lysates were determined using Coomassie Plus protein assay reagent (Pierce, Rockford, IL, USA). Fifty μg of protein/lane was separated by SDS-PAGE under reducing conditions. After transfer to nitrocellulose (Hybond-C, Amersham, Buckinghamshire, UK), blots were blocked with 5% (w/v) non-fat dry milk in TBST (10 mM Tris, 150 mM NaCl and 0.05% (v/v) Tween-20, pH 8.0), and were probed overnight at 4°C with M73 antibody 1:1000 diluted in TBST containing 2.5% non-fat dry milk. Immunoreactive proteins were visualized using HRP-conjugated Ig (Goat-anti-mouse for M73 and 9C10 or Goat-anti-Rat for 1G11) and enhanced chemiluminescence (ECL, Amersham).
Determination of IL-6 and IL-8 protein
Cells were exposed for 8 h to various doses of TNF-α (rhTNF-α, R&D Systems, Minneapolis, MN, USA) or LPS derived from E.Coli K-235 (L2018, Sigma-Aldrich, St. Louis, MO) up to 5 ng/ml and 1 μg, respectively. The amount of IL-6 and IL-8 in culture supernatants was measured by sandwich ELISA, as described before [9,12].
mRNA half-life analysis
Cells were stimulated with TNF-α (5 ng/ml) or LPS (0.1 μg/ml) for one hour before 5 μg/ml actinomycin D (Sigma-Aldrich) was added to block further transcription. Total RNA was extracted with TriZol (Invitrogen) at 0, 40 and 80 minutes after Actinomycin D addition. The amount of IL-6, IL-8 and GAPDH mRNA was determined by dotblotting and hybridization with specific 32P-labeled probes for IL-6, IL-8 and GAPDH, which have been extensively validated for specificity in our samples by Northern blot as described [10,11]. Blots were quantified using a phosphorimager and variable loading was corrected for by expressing mRNA levels relative to that of the housekeeping gene GAPDH. mRNA half-life was calculated using linear regression.
Isolation of nuclear extracts and electrophoretic mobility shift assay (EMSA)
Nuclear extracts were isolated after 1 h stimulation with 5 ng/ml TNF-α and 0.1 μg LPS as described [10,11]. Protein concentrations were measured as described above. Five μg of the nuclear extracts were incubated with 32P-labeled oligonucleotides at 4°C for 1 h and separated on a 4% non-reducing poly-acrylamide gel at slowly increasing voltages (60–220 V). Bands were identified by supershift using 1 μg of antibodies against p65 for NF-κB, c-fos and c-jun for AP-1, and C/EBP-β for C/EBP (Santa Cruz Biotechnology Inc., Santa Cruz, CA) and by competition with cold probe. The intensity of the bands was quantified using a phosphorimager. The following oligonucleotides were used in the EMSA:
NF-κB, 5'-TTGCAAATCGTGGAATTTCCTCTGACATAA-3';
AP-1, 5'-TTAAGTGTGATGACTCAGGTTTAA-3';
C/EBP, 5'-TTAAAGGACGTCACATTGCACAATCTTAATAA-3'.
Construction of siRNA
siRNAs directed against both 12S and 13S E1A mRNA were designed using the Ambion siRNA target finder using accession code AY147066.
The following sequences were selected:
AACTGTATGATTTAGACGTGA (Start position in sequence: 134)
AAGTGAAAATTATGGGCAGTG (Start position in sequence: 599)
AATGCAATAGTAGTACGGATA (Start position in sequence: 671)
AATTTTTACAGTTTTGTGGTT (Start position in sequence: 660)
AATGTATCGAGGACTTGCTTA (Start position in sequence: 800)
AAGATCCCAACGAGGAGGCGG (Start position in sequence: 723, published in [16])
Oligonucleotides were then designed using the Ambion siRNA Template Design Tool . siRNAs were constructed with the Silencer™ siRNA Construction Kit (Ambion, Austin, TX) according to the manufacturer's instructions. As a control for proper synthesis and transfection efficiency, the GAPDH siRNA template included in the kit was used. One to 10 pmol of siRNA was mixed with 0.5 μl Lipofectamine 2000 (Invitrogen) in 50 μl Optimem-1 medium (Invitrogen) according to the manufacturer's instructions and transferred to a well of a 48-wells plate with cells at 30–50% confluency. Assessment of gene silencing was performed 24- or 48 post-transfection by PCR and immunohistochemistry.
E1A and GAPDH PCR
Single strand cDNA was synthesized from total RNA isolated with Trizol (Invitrogen) using 500 ng of oligo(dT)15 and 100 units of Superscript II (Gibco/Brl) in a 50 μl volume. One μl of cDNA was used in the PCR reaction (50 mM KCl, 2 mM MgCl2, 10 mM Tris-HCl (pH 9.0) 200 mM of each dNTP, 0.1% Triton X-100, 200 nM of each primer and 1.25 U of Taq DNA polymerase (Promega, Madison, WI)). For E1A, the PCR conditions were 30 thermal cycles at 94°C for 1 min, 59°C for 1 min and 72°C for 1 min, followed by a final extension at 72°C for 10 min. The following primers were designed using primer 3 and accession code X02996:
5'-GTGACGACGAGGATGAAGA-3' (bp 395–413);
5'-ACGGCAACTGGTTTAATGG-3' (bp 614–632).
This primer set produces a 238 bp product for 12S E1A mRNA and a 375 bp product for 13S E1A mRNA. Unspliced E1A mRNA yields a 495 bp product.
For GAPDH, the PCR conditions were 25 thermal cycles at 94°C for 1 min, 55°C for 1 min and 72°C for 1 min, followed by a final extension at 72°C for 10 min. The following primers were designed using primer 3 and accession code NM_002046:
5'-ATGAAGGTCGGAGTCAACG-3' (bp 86–103);
5'-TGAAGACGCCAGTGGACTC-3' (bp 364–382):
This primer set produces a 296 bp product.
Results
Generation of stable transfectants and expression of target proteins
Constructs of E1A, E1B and E1A plus E1B in the pTracerSV40-Zeo vector were used to generate stable transfectants of NCI-H292 cells. Over 150 zeocin-resistant GFP-positive clones transfected with the E1A construct were generated, but none of these clones expressed E1A at a level detectable by immunohistochemistry (IHC). As expression of E1A leads to apoptosis in non-ras-transformed cells [17,18], it is likely that NCI-H292 cells expressing E1A protein underwent apoptosis. Transfection with E1A plus E1B yielded 144 zeocin-resistant clones, 6 of which (designated AB-clones) were found to express E1A by IHC (Fig 1A). Western blot showed that both E1A gene products, the 289R and 243R E1A, were expressed at equal levels (Fig 1B). The E1A-positive clones expressed low levels of either 19K or 55K E1B by Western blot which was not detectable by IHC. Transfection with the E1B construct yielded 30 zeocin-resistant clones, 6 of which (designated B-clones) were positive for the 55K E1B protein as determined by IHC. As controls, we generated 7 stable transfectants with the empty vector expressing GFP (designated T clones).
Figure 1 E1A expression in AB clones. IHC for clone AB96 and for other AB clones demonstrated a strict nuclear localization of expressed E1A (A, left panel) in E1A-positive clones. The right panel is stained with a control IgG (200 fold magnification). Expression of E1A by AB clones was confirmed by Western blot (B), with HEK293 cells as positive control. Please note the equal expression of both the 289R- and the 243R-E1A proteins.
Effect of E1A expression on IL-8 and IL-6 secretion in response to LPS and TNF-α
We next evaluated the TNF-α- and LPS-induced IL-8 and IL-6 responses of the various clones. Hyperresponsive and hyporesponsive clones are defined respectively as clones with a significant (p < 0.05) 2.5-fold increased or 2.5-fold decreased (i.e. 0.4 times) IL-8 or IL-6 production relative to the mean of the control (T) clones, for all tested doses of a stimulus. Normoresponsive clones are those that stay within this 2.5-fold range. This definition is based on the observation of Hogg and coworkers, who showed at least a 2.5-fold increase in IL-8 production in response to 0.01 μg/ml LPS for an E1A-positive clone compared to an E1A-negative clone [5].
Of the 6 AB clones tested, 2 clones (AB38 and AB96) showed a hyperresponsive IL-8 production in response to both LPS and TNF-α, whereas the other 4 did not (Fig 2, Table 1). One B-clone (B1) was hyperresponsive to TNF-α but not to LPS, whereas two B-clones (B3 and B4) were hyporesponsive to both TNF-α and LPS. With respect to IL-6, none of the tested clones displayed a hyperresponsive IL-6 production in response to LPS or TNF-α (Fig 3, Table 1). In fact, one AB clone (AB157), and two B clones (B4 and B5) displayed hyporesponsive IL-6 production to LPS. All AB clones except AB38 showed hyporesponsive IL-6 production to TNF-α (Fig 3B, Table 1), and the same two B clones (B4 and B5) that had a hyporesponsive IL-6 production to LPS also displayed a hyporesponsive IL-6 production to TNF-α.
Figure 2 IL-8 production by various clones in response to exposure to LPS and TNF-α. Equal cell numbers from clones were exposed to a concentration range of LPS (see z-axis; 0–1 μg/ml, A) or TNF-α (0–5 ng/ml, B) for 8 hrs. IL-8 in culture supernatant was measured by ELISA. T represents the mean IL-8 production (in pg/ml; y-axis) of 7 T-clones. Individual B and AB clones as designated on the x-axis. Data are shown as the mean of two independent experiments (triplicate samples). Due to the representation as a 3D-matrix, no standard deviation can be shown. Asterisks indicates significant (* = p < 0.05, ** = p < 0.01, *** = p < 0.001) hyperresponsiveness, # indicates significant hyporesponsiveness (# = p < 0.05, ## = p < 0.01, ### = p < 0.001).
Table 1 Number of B- and AB-clones displaying hyper- or hyporesponsive IL-8 and IL-6 production to LPS or TNF-α. Clones transfected with E1A plus E1B are designated AB and clones transfected with E1B are designated B. Hyperresponsive and hyporesponsive clones are defined respectively as clones with a significant (p < 0.05) 2.5-fold increased or 2.5-fold decreased (i.e. 0.4 times) IL-8 or IL-6 production relative to the mean of the control (T) clones, for all tested doses of a stimulus.
Hyperresponsive Hyporesponsive
LPS TNF-α LPS TNF-α Total
IL-8 B 0 1 2 2 6
AB 2 2 0 0 6
IL-6 B 0 0 2 2 6
AB 0 0 1 3 4
Figure 3 IL-6 production in response to exposure to LPS and TNF-α. Equal cell numbers from clones were exposed to a concentration range of LPS (see z-axis; 0–1 μg/ml, A) or TNF-α (0–5 ng/ml, B) for 8 hrs. IL-6 in culture supernatant was measured by ELISA. T represents the mean IL-6 production (in pg/ml; y-axis) of 7 T-clones. Individual B and AB clones as designated on the x-axis. Statistics as described in figure 2.
Analysis of transcriptional activation by EMSA
Our previous studies showed that TNF-α- and LPS-induced transcription of IL-8 in NCI-H292 cells is dependent mainly on the transcription factor NFκB and to a lesser extent on AP-1 [11]. We compared activation of NFκB in both AB clones with a hyperresponsive IL-8 production to LPS (AB96 and AB38) to that of a normoresponsive AB-clone (AB66), a T-clone (T4) and a B-clone (B3) (Fig. 4A). We found no further upregulation of NFκB activation by E1A expression. Similar findings were observed upon stimulation with TNF-α. (Fig 4B).
Figure 4 Analysis of nuclear NFκB recruitment by EMSA. Equal cell numbers from clones, indicated on the x-axis, were stimulated with 0.1 μg/ml LPS (A) or 5 ng/ml TNF-α (B) for one hour, before nuclear extracts were prepared. Specific bands were identified by supershift using a p65 antibody for NFκB and cold oligo competition. Bands were quantified by phosphorimager and expressed relative to that of the stimulated T clone, which was set at one. Data represent mean ± SEM from 2 independent experiments.
We also determined nuclear recruitment of transcription factors AP-1 and C/EBP; the latter is involved in regulation of IL-6 gene transcription in NCI-H292 cells [10]. Again, we found no altered upregulation of AP-1 or C/EBP recruitment in E1A-expressing clones to either TNF-α or LPS (data not shown).
Analysis of IL-8 and IL-6 mRNA half-life
An increased half-life of IL-8 and IL-6 mRNA can be accompanied by an enhanced production of these cytokines [9,12]. Therefore, we tested whether the difference in responsiveness to TNF-α and LPS of the clones was paralleled by alterations in the half-life of IL-8- and IL-6 mRNA. One hyperresponsive AB clone (AB96) tended to have an increased half-life of IL-8 mRNA compared to the normoresponsive T clones (p= 0.06). However, the normoresponsive clone B6 had a similarly increased IL-8 mRNA half-life (Fig. 5A). Moreover, clone AB96, that displayed hyporesponsive IL-6 production, had a relative stable IL-6 mRNA (Fig. 5B). Similar heterogeneous results emerged when clones were stimulated with LPS (data not shown), and thus differences in responsiveness between the various clones did not parallel changes in the half-life of IL-8 and IL-6 mRNA.
Figure 5 IL-8 and IL-6 mRNA half-life after exposure to TNF-α. Equal cell numbers from clones, indicated on the x-axis, were stimulated with 5 ng/ml TNF-α for one hour before 5 μg/ml Actinomycin D (ActD) was added to block further transcription. At 0, 40 and 80 min after ActD addition, total RNA was extracted with TriZol. RNA was dot blotted and hybridized with 32P-labelled IL-8, IL-6 and GAPDH probes. Signals were quantified on a phosphorimager and IL-8 and IL-6 mRNA levels were normalized for variable loading using GAPDH mRNA levels. Half-life of the mRNA in the clones was calculated using linear regression. Data represent the mean ± SEM from 2 independent experiments (triplicate samples).
Responses of E1A- and E1B-expressing NCI-H292 subclones
Together, these data indicated that expression of biologically active E1A did not correlate with an enhanced IL-8 production, nor did E1A expression uniformly affect mechanisms regulating IL-8 production. For IL-6, there was a similar heterogeneity, but E1A expression appeared to inversely correlate with IL-6 production. To exclude that the observed heterogeneity in responses was due solely to an intrinsic heterogeneity of the mother cell line, we subcloned NCI-H292 cells and tested IL-6 responses to TNF-α of 15 subclones. We found up to a 15-fold difference in maximal IL-6 responses between clones (this response range is calculated by dividing the maximal response by the minimal response at 5 ng/ml TNF-α for a group of clones; data not shown), which is similar to that for the T-clones (11-fold difference), but less than that for B- and AB-clones, 30- and 40-fold difference, respectively. This suggests that the larger response range in AB- and B-clones is caused by the presence of E1A and E1B DNA, although part of the response range appears due to biological variation of the motherline and/or results from the procedure of subcloning. To further test the latter we cloned an earlier derived subclone and tested TNF-α- and LPS-induced responses from 11 derived sub-subclones. This time, we found a 4-fold difference between clones in their maximal IL-6 and also IL-8 responses, which indicates that there is some variation in the NCI-H292 motherline.
To confirm the effect of E1A and E1B expression on the response range, we transfected one of the clones derived from subcloning the NCI-H292 motherline, with E1A plus E1B. Ten E1A-positive clones differing in the level of E1A expression were obtained (Fig 6A) and tested for IL-8 and IL-6 responses to LPS and TNF-α (Fig 6B–E). Compared to E1A-negative clones, nine out of ten clones showed a significant hyporesponsive IL-6 production to LPS or TNF-α; the clone with the lowest E1A expression having a normoresponsive IL-6 production. Even though the absolute IL-6 production was much lower in E1A-positive clones, the range of IL-6 production was much larger in the E1A-positive clones than E1A-negative clones, showing difference of upto 130- and 3-fold, respectively.
Figure 6 Expression levels of E1A protein and IL-8 and IL-6 responses of transfected NCI-H292 subclones. A) Transfection of a NCI-H292 subclone yielded 10 E1A-positive clones. Expression level of E1A in the different clones was determined by Western blot. Exposed films were scanned and quantified. Signal is expressed in arbitrary units. Even though E1A shows as a single band in the printed figure, close examination of the electronic figure reveals both the 289R and the 243R E1A protein. Equal cell numbers from clones were exposed to a concentration series of LPS (0, 0.01, 0.1 and 1 μg/ml, 6B&D) or TNF-α (0, 0.5 and 5 ng/ml, 6C&E) for 8 hrs. IL-8 en IL-6 concentrations in culture supernatant were measured by ELISA. E1A-negative clones are represented by dashed curves, E1A-positive clones by solid curves. For clarity, the clone numbers at the right are ranked for IL-8 and IL-6 production, and hyporesponsive clones (p < 0.05) are marked with #. Data are from one experiment (triplicate samples) out of two independent experiments with similar results.
None of the clones displayed significant hyperresponsive IL-8 production to LPS, whereas 4 out of 10 E1A-positive clones showed hyporesponsive IL-8 production to LPS. Similarly, none of the clones displayed a hyperresponsive IL-8 response to TNF-α, and 3 out of 10 clones showed a hyporesponsive IL-8 production. In line with our previous results, the response range for IL-8 in the E1A-positive cells reached up to 48 for TNF-α and 23 for LPS, whereas the control clones displayed a response range of 3.
siRNA results
To provide ultimate proof of the role of E1A protein, we attempted to knock down E1A expression using siRNA. Five different siRNA duplexes directed to both 12S and 13S E1A mRNA, coding for the 243R and 289R E1A proteins [19], were generated and transfected into 4 different AB clones. As a positive control, a GAPDH siRNA was used. Transfection of GAPDH siRNA in a final concentration of 4 nM resulted in a specific downregulation of more than 90% of GAPDH mRNA, 24 h and 48 h after transfection, indicating that both siRNA synthesis and transfection protocols were functional. None of the siRNAs directed to E1A downregulated E1A mRNA at a concentration of 4 nM. Transfection of higher concentrations of E1A siRNA (up to 35 nM) resulted in aspecific silencing of both E1A and GAPDH mRNA.
We next used the shRNA DNA vector pSP-E1A (a kind gift from Dr. David Hacker, LBTC EPFL, Lausanne, Switzerland) that was shown to reduce E1A mRNA by 75% in HEK293 cells [16], but due to either a low transfection efficiency of DNA vectors in our clones (<30%) or incompatibility with the U6 promoter, we could not detect downregulation of E1A expression. To circumvent the latter, an siRNA using the same sequence was generated, which neither knocked down E1A mRNA expression in our clones.
Discussion
The present study aimed to investigate and extend the hypothesis that E1A expression in lung epithelial cells, contributes to increased IL-8 responses. We analyzed a panel of stable transfectants expressing E1A, but we found no uniform effect on IL-8 protein production, whereas the effect on IL-6 production was more uniform. If anything, clones expressing E1A displayed a much larger response range for IL-8 than control clones, suggesting that expression of E1A dysregulates IL-8 production, leading to either an increased or a suppressed IL-8 production. Similar findings were obtained for E1B, indicating that the effect is not specific for E1A.
These findings are in apparent contradiction with those reported by Hogg et al., who described an increase in IL-8 production in E1A-expressing ras-transformed alveolar-type II-like A549 cells as well as E1A/E1B-transformed primary bronchial epithelial cells [5,7]. In the present studies we have used lung epithelial NCI-H292 cells, which have been studied extensively in relation to IL-8 and IL-6 production [9-12] and showed regulation similar to that of other lung epithelial cell lines and primary bronchial epithelial cells. For example, effects of IL-17 on TNF-α induced IL-8 and IL-6 secretion were similar and regulated in the same way [20]. Most importantly, post-transcriptional regulation of IL-8 and IL-6 responses, which is a major mode of regulating these responses, are directed by various conditions to a similar extent in H292 cells and primary cells. As NCI-H292 cells are also susceptible to infection by many strains of adenoviruses [21], including serotype 5, we considered NCI-H292 cells a good model to study the effect of E1A on IL-8 and IL-6 production, although we cannot exclude that the apparent contradictory results are due to the use epithelial cells from different origin.
E1A protein expressed in our clones was biologically active as is evident from the reduced IL-6 production by E1A-expressing clones (Fig 6), in line with the reported inhibition of IL-6 gene transcription by the E1A protein [13]. Furthermore, immunohistochemical staining of E1A protein was restricted to the nucleus in all clones, indicative of nuclear recruitment which is essential for E1A protein to display its biological activity. Despite the presence of active E1A protein, and the relation between IL-6 responses and the expression levels of E1A, we found no correlation between the level of E1A protein expression and IL-8 production in the present clones. The E1A gene encodes two primary regulators of viral and cellular gene expression, the E1A-243R and E1A-289R proteins [19]. Distinct biological functions have been described for the 243R and 289R E1A products in mouse lung. Preferential expression of the 243R E1A was associated with cellular hyperplasia and low level of p53-mediated apoptosis, whereas preferential 289R E1A expression led to pro-apoptotic injury and acute pulmonary inflammation [22]. Our clones all expressed the 243R and 289R in an apparent 1:1 ratio as detected by Western blot, however the E1A/E1B transformed primary bronchial epithelial cells as described by Higashimoto et al. [7] preferentially expressed the 13S transcript coding for the 289R E1A protein and hence may explain the observed difference. As yet there is no data available on preferential expression of either transcript in lung tissue of COPD patients, or in the A549 in vitro model used by Hogg and coworkers. Other reasons for the observed differences may arise simply from the fact that we analyzed a larger panel of clones which allowed us to identify the different effects or from the cooperation of Ras with E1A in A549 cells.
Analysis of the transcriptional and post-transcriptional mechanisms of IL-8 and IL-6 production in the various E1A-expressing clones revealed marked heterogeneity. There was no increased nuclear recruitment of the transcription factors NFκB, AP-1 and C/EBP in E1A-positive cells. Also, there was no change observed in the electrophoretic mobility of the protein-DNA complexes that could indicate a change in the composition of the complex. In fact, clone AB38 showed constitutive expression of high levels of C/EPB activation (data not shown), and had a high basal IL-6 production, but was not hyperresponsive. The normoresponsive clone T4 showed the highest nuclear recruitment of NFκB, but there was no correlation between the recruitment of NFκB and the amount of IL-8 released upon LPS stimulation. With respect to the half-life of IL-8 mRNA, there was a similar heterogeneity. One of the hyperresponsive clones (AB96) had an increased stability of IL-8 mRNA, however, IL-8 mRNA in another hyperresponsive clone (AB38) had a normal half-life. Taken together, there was no uniform effect of E1A expression on the transcriptional and post-transcriptional regulation of IL-8 mRNA. In fact, for some transfectants it is unclear why they produce more IL-8 or less IL-6, as there is no apparent effect on transcription or mRNA degradation. An alternative explanation not addressed in the current study is that translational control may be affected in these transfectants.
Unexpectedly, some E1B transfectants showed similar effects on IL-8 and/or IL-6 production to TNF-α and LPS as E1A transfectants. This may suggest that the mere presence of adenoviral DNA or protein affects the regulation of IL-8 and IL-6 production. A possible explanation is that integration of vector DNA in the genome (E1A and E1B normally co-integrate), a poorly understood process, may affect the expression of a functional gene involved in the cascade controlling IL-8 or IL-6 production, leading to an altered cytokine production. The exact mechanism remains to be determined. An approach could be to generate a panel of inducible transfectants.
The effect of E1A and E1B on the IL-8 production is probably best described by increasing the response range. Subclones of the NCI-H292 mother line showed biological variation in IL-6 and IL-8 production, giving a response range of 15. This variation is comparable to that for NCI-H292 clones transfected with an empty vector, but markedly smaller than that for E1A plus E1B or E1B-transfected clones. Similarly, E1A-expressing clones generated from subclones of NCI-H292 showed an enhanced response range for IL-8, which was at large due to reduced minimal IL-8 production.
Ultimate proof for, or against, a role of the E1A protein in increasing IL-8 production in lung epithelial cells would be to knock down E1A expression using siRNA, allowed by the short half-life (30–120 min) of the E1A protein [23]. We generated 6 different E1A siRNAs, but none reduced E1A mRNA or protein. GAPDH, however, was readily downregulated by our GAPDH siRNA, validating the method we used. The published shRNA vector driven by the RNA polymerase-III U6 promotor, which reduced E1A expression by 75% in HEK293 cells [16], neither reduced E1A expression in our cells. The reason for this failure may be due to incompatibility of the U6 promoter with our cells. Furthermore, steric hindrance by proteins bound to E1A mRNA in NCI-H292 cells may have prevented the siRNA to bind. Hence, it remains elusive whether the dysregulated IL-8 production comes about by the actual viral proteins being expressed, or whether the incorporation of the DNA is sufficient.
The question arises whether there still is a role for E1A expression in the pathogenesis of COPD as proposed by Hogg et al. From theirs and an other study [24] it follows that E1A DNA and E1A protein is more abundantly expressed in epithelial cells from COPD patients, though the reason for increased E1A expression in COPD patients remains elusive. Based upon our findings it is doubtful that E1A expression per se contributes to the pathogenesis of COPD by means of an increased IL-8 production. We cannot exclude, however, that E1A DNA integration following adenoviral infection differs from that obtained with our transfection approach. In addition, our transfection approach allowed us to analyze adenoviral integrations that may be very rare events by adenoviral infection. Indeed we found large differences in transfection efficiency for the various constructs. This difference may be explained by the cellular effect of the gene that is expressed. As for transfection with E1A alone, there were no viable clones expressing E1A, which we assume was due to apoptosis. In a later study performed by Hogg et al it was also described that with primary bronchial cells, it is not possible to obtain clones expressing only E1A, and that E1B expression is needed together with E1A to generate stable clones. The transfection efficiency was much higher with the non-apoptotic E1B and empty vector. Second, the transfection efficiency may depend on the size of the insert in the vector. The mechanism by which a vector integrates remains largely elusive, but in order to integrate the vector needs to linearize, i.e. break open. The point where the vector breaks is random, and the larger the insert is compared to the rest of the vector, the larger the chance that it breaks in the gene of interest, leading to a reduction in efficiency of inserting the full gene. This may underlie the lower efficiency of the E1A/E1B clones, as this insert is of a similar size as the vector, and 2Kb larger than then E1B alone. The much smaller empty vector containing the Zeocyn resistance GFP tagged protein showed the highest transfection efficiency. Furthermore, incomplete insertion of the E1A/E1B sequence may also lead to expression of E1A without E1B, leading again to cell death. As we observed some clones transfected with E1A and E1B that were positive only for E1B, it is likely that this also occurs.
The previously unrecognized decreased IL-6 production by the E1A- and E1B-expressing clones, could contribute to increased inflammation. IL-6 is regarded a pro-inflammatory mediator, but has also been shown to exert many anti-inflammatory and immunosuppressive effects (see ref. [25] for review). Moreover, in several murine models for pulmonary inflammation IL-6 was shown to protect against lung damage [26-29]. The reduced IL-6 production from epithelial cells thus may play a role in the early pathogenesis of COPD, as it is hypothesized that latent adenoviral infection, accompanied by expression of E1A, is established in early childhood [2]. Even though in exhaled breath condensate of patients diagnosed with COPD high levels of IL-6 are found [30], this IL-6 may be derived from sources other than lung epithelial cells. Notably, the increased IL-6 levels are observed in patients already diagnosed with COPD, and thus do not exclude a role for reduced IL-6 production from epithelial cells in the early pathogenesis of COPD. Whether E1A expression affects the expression of other genes in these transfectants is unknown as yet.
Conclusion
Taken together, our study does not provide support for an unique role of Ad5 E1A protein in the enhanced IL-8 production. Expression of E1A and E1B, however, did affect the IL-8 response to LPS and TNF-α, as was evident by the increased response ranges, in particular due to a reduced IL-8 production of some clones. Both E1A and E1B reduced IL-6 production, which could play a role during the early pathogenesis of COPD. These results warrant further research into the impact of integrated genomic viral sequences on inflammatory responses.
Abbreviations
AP-1: Activator Protein-1
C/EBP: CAAT/Enhancer Binding Protein
IL: Interleukin
NFκB: Nuclear Factor κB
TNF: Tumor Necrosis Factor
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AB prepared DNA constructs, generated stable transfectants, performed all studies mentioned and drafted the manuscript. MS assisted with ELISAs and carried out dotblot hybridizations. HJ participated in the study design and coordination, and helped to draft the manuscript. RL conceived the study, participated in its design and coordination, and revised the draft.
Acknowledgements
We would like to thank Dr. Robert G. Vries (Department of Molecular Cell Biology, Leiden University Medical Center, Leiden, Netherlands) for providing the pAt-153-Xho vector and the antibodies against E1B, Dr. David Hacker (EPFL-SV-IGBB-LTC, Lausanne, Switzerland) for providing the pSP-E1A vector, and Dr. T.A. Out and professor René van Lier for critical reading of the manuscript. This study was supported financially by the Netherlands Asthma Foundation, grant 99.27.
==== Refs
Fletcher C Peto R The natural history of chronic airflow obstruction Br Med J 1977 1 1645 1648 871704
Matsuse T Hayashi S Kuwano K Keunecke H Jefferies WA Hogg JC Latent adenoviral infection in the pathogenesis of chronic airways obstruction Am Rev Respir Dis 1992 146 177 184 1626800
Elliott WM Hayashi S Hogg JC Immunodetection of adenoviral E1A proteins in human lung tissue Am J Respir Cell Mol Biol 1995 12 642 648 7766428
Graham FL Rowe DT McKinnon R Bacchetti S Ruben M Branton PE Transformation by human adenoviruses J Cell Physiol Suppl 1984 3 151 163 6611340 10.1002/jcp.1041210418
Keicho N Elliott WM Hogg JC Hayashi S Adenovirus E1A upregulates interleukin-8 expression induced by endotoxin in pulmonary epithelial cells Am J Physiol 1997 272 L1046 L1052 9227502
Keicho N Elliott WM Hogg JC Hayashi S Adenovirus E1A gene dysregulates ICAM-1 expression in transformed pulmonary epithelial cells Am J Respir Cell Mol Biol 1997 16 23 30 8998075
Higashimoto Y Elliott WM Behzad AR Sedgwick EG Takei T Hogg JC Hayashi S Inflammatory mediator mRNA expression by adenovirus E1A-transfected bronchial epithelial cells Am J Respir Crit Care Med 2002 166 200 207 12119233 10.1164/rccm.2111032
Vitalis TZ Kern I Croome A Behzad H Hayashi S Hogg JC The effect of latent adenovirus 5 infection on cigarette smoke-induced lung inflammation Eur Respir J 1998 11 664 669 9596119
van Wissen M Snoek M Smids B Jansen HM Lutter R IFN-gamma amplifies IL-6 and IL-8 responses by airway epithelial-like cells via indoleamine 2,3-dioxygenase J Immunol 2002 169 7039 7044 12471139
Roger T Out TA Jansen HM Lutter R Superinduction of interleukin-6 mRNA in lung epithelial H292 cells depends on transiently increased C/EBP activity and durable increased mRNA stability Biochim Biophys Acta 1998 1398 275 284 9655919
Roger T Out T Mukaida N Matsushima K Jansen H Lutter R Enhanced AP-1 and NF-kappaB activities and stability of interleukin 8 (IL-8) transcripts are implicated in IL-8 mRNA superinduction in lung epithelial H292 cells Biochem J 1998 330 ( Pt 1) 429 435 9461540
Lutter R Loman S Snoek M Roger T Out TA Jansen HM IL-6 protein production by airway epithelial(-like) cells disabled in IL-6 mRNA degradation Cytokine 2000 12 1275 1279 10930312 10.1006/cyto.1999.0728
Janaswami PM Kalvakolanu DV Zhang Y Sen GC Transcriptional repression of interleukin-6 gene by adenoviral E1A proteins J Biol Chem 1992 267 24886 24891 1332971
Bernards R Schrier PI Bos JL Van der Eb AJ Role of adenovirus types 5 and 12 early region 1b tumor antigens in oncogenic transformation Virology 1983 127 45 53 6858003 10.1016/0042-6822(83)90369-0
Zantema A Fransen JA Davis-Olivier A Ramaekers FC Vooijs GP DeLeys B Van der Eb AJ Localization of the E1B proteins of adenovirus 5 in transformed cells, as revealed by interaction with monoclonal antibodies Virology 1985 142 44 58 2932843
Hacker DL Bertschinger M Baldi L Wurm FM Reduction of adenovirus E1A mRNA by RNAi results in enhanced recombinant protein expression in transiently transfected HEK293 cells Gene 2004 341 227 234 15474305 10.1016/j.gene.2004.06.054
Lin HJ Eviner V Prendergast GC White E Activated H-ras rescues E1A-induced apoptosis and cooperates with E1A to overcome p53-dependent growth arrest Mol Cell Biol 1995 15 4536 4544 7623844
Rao L Debbas M Sabbatini P Hockenbery D Korsmeyer S White E The adenovirus E1A proteins induce apoptosis, which is inhibited by the E1B 19-kDa and Bcl-2 proteins Proc Natl Acad Sci U S A 1992 89 7742 7746 1457005
Perricaudet M Akusjarvi G Virtanen A Pettersson U Structure of two spliced mRNAs from the transforming region of human subgroup C adenoviruses Nature 1979 281 694 696 551290 10.1038/281694a0
van den Berg A Kuiper M Snoek M Timens W Postma DS Jansen HM Lutter R Interleukin-17 induces hyperresponsive interleukin-8 and interleukin-6 production to tumor necrosis factor-alpha in structural lung cells Am J Respir Cell Mol Biol 2005 33 97 104 15845864 10.1165/rcmb.2005-0022OC
Hierholzer JC Castells E Banks GG Bryan JA McEwen CT Sensitivity of NCI-H292 human lung mucoepidermoid cells for respiratory and other human viruses J Clin Microbiol 1993 31 1504 1510 8314992
Yang Y McKerlie C Borenstein SH Lu Z Schito M Chamberlain JW Buchwald M Transgenic expression in mouse lung reveals distinct biological roles for the adenovirus type 5 E1A 243- and 289-amino-acid proteins J Virol 2002 76 8910 8919 12163610 10.1128/JVI.76.17.8910-8919.2002
Spindler KR Berk AJ Rapid intracellular turnover of adenovirus 5 early region 1A proteins J Virol 1984 52 706 710 6333516
He B Zhao M Li X [Latent adenovirus infection in chronic obstructive pulmonary disease] Zhonghua Jie He He Hu Xi Za Zhi 2001 24 520 523 11758163
Tilg H Dinarello CA Mier JW IL-6 and APPs: anti-inflammatory and immunosuppressive mediators Immunol Today 1997 18 428 432 9293158 10.1016/S0167-5699(97)01103-1
Chen W Havell EA Gigliotti F Harmsen AG Interleukin-6 production in a murine model of Pneumocystis carinii pneumonia: relation to resistance and inflammatory response Infect Immun 1993 61 97 102 8418070
Bermudez LE Wu M Petrofsky M Young LS Interleukin-6 antagonizes tumor necrosis factor-mediated mycobacteriostatic and mycobactericidal activities in macrophages Infect Immun 1992 60 4245 4252 1328056
Denis M Interleukin-6 in mouse hypersensitivity pneumonitis: changes in lung free cells following depletion of endogenous IL-6 or direct administration of IL-6 J Leukoc Biol 1992 52 197 201 1506776
Ulich TR Yin S Guo K Yi ES Remick D del Castillo J Intratracheal injection of endotoxin and cytokines. II. Interleukin-6 and transforming growth factor beta inhibit acute inflammation Am J Pathol 1991 138 1097 1101 2024703
Bucchioni E Kharitonov SA Allegra L Barnes PJ High levels of interleukin-6 in the exhaled breath condensate of patients with COPD Respir Med 2003 97 1299 1302 14682411 10.1016/j.rmed.2003.07.008
|
16185356
|
PMC1261537
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Sep 26; 6(1):111
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-111
|
oa_comm
|
==== Front
World J Surg OncolWorld Journal of Surgical Oncology1477-7819BioMed Central London 1477-7819-3-621618802210.1186/1477-7819-3-62Case ReportPapillary cystic and solid tumour of the pancreas: Report of a case and literature review Kasem Abdul [email protected] Zainab [email protected] Joseph [email protected] Department of General Surgery, Princess Royal University Hospital,Farnborough Common, Orpington, BR6 8DN UK2 Department of Pathology, Princess Royal University Hospital,Farnborough Common, Orpington, BR6 8DN UK2005 27 9 2005 3 62 62 23 5 2005 27 9 2005 Copyright © 2005 Kasem et al; licensee BioMed Central Ltd.2005Kasem 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 papillary cystic and solid tumour of the pancreas (PCSTP) is a rare primary neoplasm of unknown pathogenesis typically found in young women. PCSTP is a low-grade malignant tumour, which is often asymptomatic but it may present with abdominal pain.
Case presentation
A 38 year old female patient who presented with one day history of epigastric pain was diagnosed as PCSTP. The patient was successfully treated with distal pancreatectomy.
Conclusion
It is important to differentiate this tumour from other pancreatic tumours because, unlike malignant pancreatic tumours, this neoplasm does not usually metastasise and is amenable to cure after complete surgical resection. However, the cell origin and the aetiology of this tumour are not clear and further studies are warranted in its pathogenesis.
==== Body
Background
The Papillary cystic and solid tumour of the pancreas (PCSTP) is an unusual low-malignant epithelial tumour, which mostly affects young females with a mean age 25 years [1-3]. Nearly 90–95% of the patients are female [2-4]. It has also been referred to as a solid-cystic epithelial tumour, solid-pseudopapillary tumour or papillary-cystic tumour [5]. It makes up 0.2–2.7% of all pancreatic cancers [6,7]. Since the original description by Frantz in 1959 the incidence of PCSTP has been increasing [8], although it may be that it is increasingly being diagnosed. Furthermore, some authors believe that the tumour is not rare but occasionally is misdiagnosed as carcinoma [1]. we report here a case andpresent the literature review.
Case presentation
A 38-year-old female presented with a one day history of acute epigastric pain. There was no nausea or vomiting. She had developed non-Insulin dependant diabetes 18 months previously. On examination she was overweight and there was epigastric tenderness, but no mass was palpable. All haematological and biochemical parameters were within normal limits.
Abdominal ultrasonography demonstrated a 5 × 3 cm partially cystic mass in distal pancreas with a possibility of a cystic neoplasm. A contrast enhanced computed tomography (CT) scan confirmed a 2 cm mass in the body of pancreas (Figure 1). Dynamic contrast enhanced magnetic resonance imaging (MRI) showed a 4 cm cystic lesion in the body of Pancreas suggestive of a mucinous cyst adenocarcinoma (Figure 2). No liver or abdominal metastases were detected on either the ultrasound or CT examination.
Figure 1 CT scan showing lesion in the pancreas.
Figure 2 MRI showing cystic and solid lesion in the pancreas.
The patient underwent staging laparoscopy which showed no metastases. Laparotomy was performed. An en-block distal pancreatectomy, including the pancreatic mass, and splenectomy was performed with full clearance of peri-pancreatic and coeliac nodes. The patient made an uneventful recovery.
On gross examination the pancreatic tumour was oval, 7.0 cm in diameter, and was surrounded by a fibrous pseudocapsule. Its cut surface showed solid and cystic spaces (Figure 3). On microscopy, the solid portion of the tumour revealed sheets of uniform polygonal cells as well as dyscohesive papillae arranged around fine fibrovascular cores. The cyst wall was composed of dense acellular fibrous tissue within which the tumour cells were arranged as cords and trabeculae set within a mucinous background (Figure 4).
Figure 3 Gross specimen of the lesion.
Figure 4 Photomicrograph of papillary cystic and solid tumour of the pancreas (haematoxylin and eosin ×40).
No invasion of the tumour into surrounding normal tissue was present. The tumour cells were strongly positive for vimentin and alpha-1 antitrypsin and showed faint focal positivity with MNF116 and chromogranin.
Discussion
The presenting features of PCSTP are relatively non-specific. In the case of large tumours, patients present with symptoms related to compression of adjacent structures. The tumours are either found incidentally or they generally cause mild abdominal symptoms such as abdominal discomfort or chronic and acute pain. Jaundice is rare, even in tumours that originate from the head of the pancreas and usually there is no associated functional endocrine syndrome [9-12]. Rarely, these tumours are found due to haemoperitoneum from rupture of the tumour [3,13]. Physical examination might show a palpable mass and epigastric tenderness [3].
PCSTP usually arises in the tail, the body, or occasionally in the head of the pancreas. The body and the tail of the pancreas are most frequently affected [3,4]. PCSTP may appear to be extra-pancreatic with only localised connection to the gland [9]. Development of the tumour in ectopic sites of pancreatic tissue has also been described [1].
Laboratory investigations provide little additional information [14]. Elevated serum tumour markers (CEA, CA19-9) have not been described with PCSTP [8,15].
Ultrasonographic findings are not characteristic and they may be suggestive of papillary tumour [16]. In PCSTP, ultrasonography reveals a sharply demarcated, well circumscribed, variably solid and cystic mass without any internal septation [3,17]. CT scaning presents a variable picture depending on the relationship of cystic-necrotic to solid components. Typically, on CT scan PCSTP tumours appear as sharply circumscribed, well-encapsulated, heterogeneous and hypodense lesions [4,12].
Koito et al [18], reported that endoscopic ultrasonography may provide an accurate diagnosis of pancreatic cystic tumours<2 cm. Additional studies are warranted in this area [19].
MRI offers good visualization of haemorrhagic areas [20]. On MRI, PCSTPs are sharply demarcated and have areas of high signal intensity corresponding to foci of haemorrhage [9].
The origin and histogenesis of this tumour is controversial [3,21] and little is known about it [22]. The fact that PCSTP tumours may express epithelial as well as mesenchymal markers and occasionally show exocrine and endocrine features suggesting an origin from a stem cell. However, the nature of the tumour cells is not obvious from their structure and behaviour. In addition, their strongly sex-linked occurrence is not in keeping with an origin from a stem cell. In one case, the tumour cells were highly reactive to progesterone antibody, while they were negative for estrogen. These findings suggest a ductal origin for PCSTP, and also suggested that the sex hormones play a role in its growth, but not its genesis [23].
The neoplasm is generally encapsulated and is well demarcated from the remaining pancreas[15]. The cut surface reveals lobulated, light brown solid areas admixed with zone of haemorrhage and necrosis as well as cystic spaces filled with necrotic debris[22].
PCSTP has mixed histological features including: solid monomorphous pattern with variable sclerosis, a pseudopapillary, trabecular and microcystic patterns [24]. Although criteria of malignancy have not been clearly established [15], unequivocal perineural invasion or angioinvasion, with or without deep invasion into the surrounding tissue, is taken to indicate malignant behaviour [11,22]. The most useful markers are alpha-1-antitrypsin, alpha-1-antichymotrypsin, neuron spesific enolase (NSE) and Vimentin [22,25].
One study found a definite increase in both oestrogen and progesterone receptors in the PCSTP tumour relative to both presently determined and previously published levels in normal pancreas. The very high levels of progesterone receptor detected in this tumour support the hypothesis that the PCSTP is oestrogen-responsive, since expression of progesterone receptor is induced by oestrogen, and thus constitutes an index of effective oestrogen action in a given tissue [26]. Oestrogen and progesterone receptors have been demonstrated by biochemical assays in four solid-pseudopapillary tumour [11,26]. Immunohistologically, some studies fail to detect nuclear oestrogen receptors [11,25] while progesterone receptors had been demonstrated in eight tumours [11].
Although a few case reports suggested that the preoperative diagnosis of PCSTP is possible by using fine needle aspiration (FNA), especially in clinically typical examples [9,20,22]. FNA was avoided in many cases because of the potential risk of tumour spillage and this may compromise surgical cure [1,27]. However, cytology obtained via FNA may not be useful to differentiate between pancreatoblastoma and PCST and if the tumour is operable, FNA is not necessary.
The differential diagnosis of PCSTP includes any cystic and/or solid pancreatic process such as: congenital pancreatic cysts, haemorrhagic pseudocyst, parasitic hydatid cyst, and other common cystic neoplasms of the pancreas, such as serous cystadenoma or cystadenocarcinoma, mucinous cystic neoplasms, cystic islet cell tumours, pancreatoblastoma, or mucinous duct ectasia [2,4,19]. Cases of PCSTP can mimic pancreatic cyst, occasionally mistakenly treated by cystogastrotomy [27].
Complete resection is the treatment of choice [14], and the standard therapy should involve complete removal of the tumour, the associated lymph nodes, the involved pancreas and any adjacent involved organs. Local invasion, recurrence, or limited metastases should not be considered contraindications to resection [1,27]. Surgery should be performed even if local infiltration is present, and in selected cases when complete excision is not possible, excision combined with postoperative radiotherapy to the residual mass has been used. [3]
Surgical management has been tailored to the slow-growing, non-invasive nature of this tumour. Depending on the location of the PCSTP, the surgical operation is chosen. With tumour involvement of the head of pancreas, a pylorus-preserving pancreaticoduodenectomy is recommended [27]. PCSTPs involving the neck or body of the pancreas were resected by central pancreatectomy and reimplantation of the pancreatic remnant into the stomach, with theoretical benefit of preserving pancreatic parenchyma and spleen [27]. When the tumour was located at the pancreatic tail, tail and body, or body of pancreas, distal pancreatectomy with splenectomy was employed in many cases [3]. Many authors recommended splenic conservation following distal pancreatectomy when possible [1,27].
Given the low grade malignancy and the excellent prognosis of PCSTP, conservative resection such as enucleation, evolution, lumpectomy, central pancreatectomy and partial resection of the head of pancreas have been suggested as safe and effective surgical procedures, especially in paediatric patients [3,5,16,28,29].
However, if at all possible, complete and radical excision should be the aim, as surgical curability is high and there is no clearly established role for radio-chemotherapy or embolisation in the treatment of PCSTP [20]. However, sporadic reports are present. Matusuda et al, reported a case of multiple hepatic metastases which responded to chemo-embolisation of the tumour [1], Fried et al, observed substantial shrinkage of an unresctable tumour after 6 weeks of radiotherapy [1].
PCSTP is a remarkably indolent neoplasm, and is regarded as a carcinoma of low malignant potential [9]. Most authors consider PCSTP as a benign or low-grade malignancy [20]. More than 95% of patients with PCSTP limited to the pancreas are cured by complete surgical excision. Clear resection margins are necessary to prevent local recurrence [9]. Surprisingly, even patients with metastatic disease have experienced long-term survival[9].
Conclusion
PCSTP is uncommon primary pancreatic neoplasm of unknown aetiology with low malignant potential generally occurring in young women. The sex and age distribution suggests that hormonal factors may be important in the pathogenesis of PCSTP. PCSTP should be considered in the differential diagnosis of any pancreatic mass, especially in young women. Unlike other pancreatic malignant tumours, this neoplasm is indolent and metastases are rare. The treatment of choice is complete surgical removal and the prognosis is excellent after complete surgical resection.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AK: Reviewed the literature and prepared the draft manuscript
ZA: Contributed the photomicrographs and pathological part of the manuscript
JE: Supervised the manuscript preparation and edited the manuscript
All authors read and approved the manuscript
Acknowledgements
Patients consent was obtained for publication of this case report.
==== Refs
Petrakis I Vrachassotakis N Kogerakis N Hatzidakis A Zoras O Chalkiakis G Solid pseudopapillary neoplasm of the pancreas: Report of a case after a 10-year follow-up and review of the literature Pancreatology 2001 1 123 128 12120190 10.1159/000055804
Buetow PC Buck JL Pantongrag-Brown L Beck KG Ros PR Adair CF Solid and papillary epithelial neoplasm of the pancreas: Imaging-pathologic correlation in 56 cases Radiology 1996 199 707 11 8637992
Lee WJ Park YT Choi JS Chi HS Kim BR Solid and papillary neoplasm of the pancreas Yonsei Med J 1996 37 131 141 8711936
Merkle EM Weber CH Siech M Kolokythas O Tomczak R Rieber A Brambs HJ Papillary cystic and solid tumour of pancreas Z Gastroenterol 1996 34 743 746 8956478
Minz S Sharma HP Kumar P Nirala KP Shrivastava SK Khandewal C Solid-cystic papillary tumour of pancreas Indian Journal of Pathol Microbiol 2001 44 463 464 12035367
Nakamura S Okayama Y Imai H Aoki S Kobayashi S Hattori T Shiraki S Goto K Sano H Ohara H Nomura T Joh T Yoshifumi Y Itoh M A Solid Cystic Tumour of the pancreas with ossification and possible malignancy, coexisting nonfusion of the pancreatic ducts J Clin Gastroenterol 2001 33 333 336 11588552 10.1097/00004836-200110000-00017
Nadler EP Novikov A Landzberg BR Pochapin MB Centeno B Fahey TJ Spigland N The use of endoscopic ultrasound in diagnosis of solid pseudopapillary tumours of pancreas in children J Pediatr Surg 2002 37 1370 1373 12194139 10.1053/jpsu.2002.35028
Frantz VK Tumour of the pancreas Atlas of tumour pathology, Section VII, Fascicles 27 and 28 1959 Washington, DC: Armed Forces Institute of Pathology 32 33
Klimstra DS Wenig BM Heffess CS Solid-pseudopapillary tumour of pancreas: A typically cystic carcinoma of low malignant potential Semin Diagn Pathol 2000 17 66 80 10721808
Bulligan MG Lucca E Risaliti A Terrosu G Intini S Donini A Anania G Sorrentino M Rocco M Solid papillary tumour of pancreas: A clinical case Minerva Chir 1996 51 983 988 9072729
Solcia E Capella C Kloppel G Tumors of the pancreas Atlas of tumor pathology, 3rd series, Fascicles 20 1997 Washington, DC: Armed Forces Institute of Pathology 120 144
Dong PR Lu DS Degregario F Fell SC Au A Kadell BM Solid and papillary neoplasm of the pancreas: Radiological-pathological study of five cases and review of the literature Clin Radiol 1996 51 702 705 8893639 10.1016/S0009-9260(96)80242-X
Colovic R Micev M Zogovic S Colovic N Stojkovic M Grubor N Solid and cystic-papillary tumour of the pancreas Srp Arh Celok Lek 2000 128 393 396 11337920
Yang YJ Chen JS Chen CJ Lin PW Chang KC Tzeng CC Papillary cystic tumour of the pancreas in children Scand J Gastroenterol 1996 31 1223 1227 8976016
Pezzolla F Lorusso D Caruso ML Demma I Solid pseudopapillary tumour of the pancreas. Consideration of two cases Anticancer Res 2002 22 1807 1812 12168873
Castillo Grau P Cerezo Lopez E Berges Magana MA Erdozain Sosa JC Gonzalez Sanz-Agero P Segura Cabral JM Muro Gonzalez J Ultrasonographic aspects of cystic papillary tumour of the pancreas. Apropos of 2 cases Gastroenterol Hepatol 1996 19 52 54 8616680
Lee DH Yi BH Lim JW Ko YT Sonographic findings of solid and papillary epithelial neoplasm of the pancreas J Ultrasound Med 2001 20 1229 1232 11758028
Koito K Namieno T Nagakawa T Shyonai T Hirokawa N Morita K Solitary cystic tumour of the pancreas: EUS-pathologic correlation Gastrointest Endosc 1997 45 268 276 9087833
Crawford BE 2nd Solid and papillary epithelial neoplasm of the pancreas, Diagnosis by cytology South Med J 1998 91 973 977 9786298
Schwartz DC Campos MA A woman with recurrent abdominal pain Am J Med Sci 2001 321 352 354 11370800 10.1097/00000441-200105000-00009
Buchino JJ Fine-needle aspiration of solid and papillary cystic tumour of the pancreas Pediatr pathol 1996 16 235 242 10.1080/107710496175714
Grant LD Lauwers GY Meloni AM Stone JF Betz JL Vogel S Unbalanced chromosomal translocation, der (17)t(13;17)(q14;p11) in a solid and cystic papillary epithelial neoplasm of the pancreas Am J Surg Pathol 1996 20 339 345 8772788 10.1097/00000478-199603000-00011
Remadi S Mac Gee W Doussis-Anagnostopoulou I Berger SD Ismail A Papillary-cystic tumour of the pancreas Diagn Cytopathol 1996 15 398 401 8989542 10.1002/(SICI)1097-0339(199612)15:5<398::AID-DC8>3.0.CO;2-8
Yoon DY Hines OJ Bilchik AJ Lewin K Cortina G Reber HA Solid and papillary epithelial neoplasms of the pancreas: Aggressive resection for cure Am Surg 2001 67 1195 1199 11768829
Miettinen M Partanen S Fraki O Kivilaakso E Papillary cystic tumour of the pancreas. An analysis of cellular differentiation by electron microscopy and immunohistochemistry Am J Surg Pathol 1987 11 855 865 3674283
Ladanyi M Mulay S Arseneau J Bettez P Estrogen and progesterone receptor determination in the papillary cystic neoplasm of the pancreas with immunohistochemical and ultrastructural observations Cancer 1987 60 1604 11 3304615
Panieri E Krige JE Bornman PC Graham SM Terblanche J Cruse JP Operative management of papillary cystic neoplasm of the pancreas J Am Coll Surg 1998 186 319 324 9510263 10.1016/S1072-7515(98)00015-5
Iacono C Bortolasi L Serio G Is there a place for central pancreatectomy in pancreatic surgery J Gastrointest Surg 1998 2 509 516 Discussion 516-5117 10457309 10.1016/S1091-255X(98)80050-4
Nakamura Y Egami K Maeda S Hosone M Onda M Solid and papillary tumour of the pancreas complicating agenesis of the dorsal pancreas J Hepatobiliary Pancreat Surg 2001 8 485 489 11702261 10.1007/s005340100014
|
16188022
|
PMC1261538
|
CC BY
|
2021-01-04 16:39:04
|
no
|
World J Surg Oncol. 2005 Sep 27; 3:62
|
utf-8
|
World J Surg Oncol
| 2,005 |
10.1186/1477-7819-3-62
|
oa_comm
|
==== Front
World J Surg OncolWorld Journal of Surgical Oncology1477-7819BioMed Central London 1477-7819-3-631618803010.1186/1477-7819-3-63ResearchQuality of Life determinants in women with breast cancer undergoing treatment with curative intent Pandey Manoj [email protected] Bejoy Cherian [email protected] Padmakumar [email protected] Kunnambath [email protected] Kuttan [email protected] Sankarannair [email protected] Beela S [email protected] Balakrishnan [email protected] Departments of Surgical Oncology, Regional Cancer Centre, Trivandrum, India2 Department of Radiation Oncology, Regional Cancer Centre, Trivandrum, India3 Department of Surgical Oncology, Jawaharlal Nehru Cancer hospital and Research Centre, Bhopal, India2005 27 9 2005 3 63 63 10 6 2005 27 9 2005 Copyright © 2005 Pandey et al; licensee BioMed Central Ltd.2005Pandey 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 diagnosis of breast cancer and its subsequent treatment has significant impact on the woman's physical functioning, mental health and her well-being, and thereby causes substantial disruption to quality of life (QOL). Factors like patient education, spousal support and employment status, financial stability etc., have been found to influence QOL in the breast cancer patient. The present study attempts to identify the determinants of QOL in a cohort of Indian breast cancer patients.
Patients and methods
Functional Assessment of Cancer Therapy-Breast (FACT-B) Version 4 Malayalam was used to assess quality of life in 502 breast cancer patients undergoing treatment with curative intent. The data on social, demographic, disease, treatment, and follow-up were collected from case records. Data was analysed using Analysis of Variance (ANOVA) and multinomial logistic regression.
Results
The mean age of the patients was 47.7 years with 44.6% of the women being pre-menopausal. The FACT-B mean score was 90.6 (Standard Deviation [SD] = 18.4). The mean scores of the subscales were – Physical well-being 19.6 (SD = 4.7), Social well-being 19.9 (SD = 5.3), Emotional well-being 14 (SD = 4.9), Functional well-being 13.0 (SD = 5.7), and the Breast subscale 23.8 (SD = 4.4). Younger women (<45 years), women having unmarried children, nodal and/or metastatic disease, and those currently undergoing active treatment showed significantly poorer QOL scores in the univariate analysis. However multivariate analysis indicated that the religion, stage, pain, spouse education, nodal status, and distance travelled to reach the treatment centre as indicative of patient QOL.
Conclusion
QOL derangements are common in breast cancer patients necessitating the provisions for patient access to psychosocial services. However, because of the huge patient load, a screening process to identify those meriting intervention over the general population would be a viable solution.
==== Body
Background
Breast cancer is the leading cause of cancer death among women around the world. In India it shows mix incidence pattern with breast cancer being second to cancer of the cervix in rural areas [1,2], however, in metropolitan cities like Mumbai, New Delhi and Trivandrum, the incidence of breast cancer has crossed that of cervix. The incidence of breast cancer in India ranges from 8.8/100,000 at Barshi to 28.6/100,000 at Mumbai [2]. In Trivandrum, the age-adjusted-rate (AAR) is 31.7/100,000 for the urban population and 16.5/100,000 for rural population [3].
The focus of breast cancer care, in addition to examining short-term treatment related quality of life (QOL) outcomes, has expanded to include acute treatment-related side effects and long-term factors that influence the quality as well as quantity of survival [4-7]. Considerable efforts are directed to reduce morbidity from treatment and rehabilitation. Scenario in India is little different. In absence of screening programmes, majority of the breast cancers are still diagnosed in locally advanced stage and achieving longer survival is still a priority. A few studies on QOL in the Indian context exist, factors like patient education, spousal support and employment status, financial stability, disease stage, etc., have been found to influence patient QOL [8,9].
QOL domains like levels of physical, social, and psychological well-being have been found to be comparable to those of women without the disease [10,11]. Initially, women with breast cancer, especially younger women, tend to suffer substantial disruption in their physical functioning, mental health and well-being [12,13]. Due to this wide variability in QOL [14,15] identification of factors that render women vulnerable to negative outcomes and poor QOL is essential [6]. This study aims at identifying the determinants of QOL of Indian women with breast cancer treated with curative intent, on a cross-sectional cohort of patients interviewed at a single cancer care centre.
Patients and Methods
The study sample consisted of 504 breast cancer patients who were undergoing or had undergone curative treatment at our centre. The tool was administered either at the beginning of the treatment or at follow-up after the treatment. The earlier validated local language version [8] of the Functional Assessment of Cancer Therapy-Breast, Version 4 (FACT-B) [16] was used. FACT-B is a 36 item self administered scale containing 4 general subscales i.e. physical, social/family well being, functional and emotional well being, the fifth subscale contain 9 items and is specific for breast cancer. Written consent was obtained from all the patients prior to administering the tool. The study was approved by the Institutional research board and the Ethics committee. The test was administered and scored in accordance with the instructions in the manual for the Version 4 of the Functional Assessment of Chronic Illness Therapy (FACIT) Measurement System [17]. Group comparisons were carried out by using one-way analysis of variance (ANOVA). Multivariate analysis was carried out using multiple logistic regression, the data was dichotomised using the median value and factors identified by literature search, and univariate analysis were entered into the model in single step (step method).
Results
Mean age of the patients was 47.6 years (SD = 11, range 20–80, median 47 years). Of the 502 patients almost equal number belonged to upper, middle and lower class (Table 1). Majority of the patients were Hindus (78%) resided within 150 km of the centre and most were married (75%). Other population characteristics are described in table 1.
Table 1 Frequency and percentage distribution of demographic characteristics of study population
Variable Grouping Codes Group N Percentage
Interviewer 1 BCT 197 39.2%
2 SR 305 60.8%
Social class 1 Low 167 33.3%
2 Middle 164 32.7%
3 High 171 34.1
Distance to centre (Km) 1 Local 150 29.9%
2 <150 241 48.0%
3 <150–250 61 12.2%
4 >250 49 9.8%
5 Don't know 1 0.2%
Religion 1 Hindu 323 64.3%
2 Muslim 71 14.1%
3 Christian 94 18.7%
9 Others/Don't know 13 2.6%
Marital status 1 Single 23 4.6%
2 Married 377 75.1%
3 Widow/Divorce 100 19.9%
Self education 1 Illiterate 23 4.6%
2 ≤5 96 19.1%
3 6–10 255 50.8%
4 11–12 55 11.0%
5 Graduate/tech 40 8.0%
6 Post graduate 29 5.8%
Spouse education 1 Illiterate 12 2.4%
2 ≤5 64 12.7%
3 6–10 198 39.4%
4 11–12 26 5.2%
5 Graduate/tech 44 8.8%
6 Post graduate 31 6.2%
7 Don't know 127 25.3%
Self Occupation 1 HW/Unemployed 380 75.7%
2 Employed 81 16.1%
3 Self/Business/Daily 31 6.2%
9 Don't know 10 2.0%
Spouse Occupation 1 HW/Unemployed 50 10%
2 Employed 135 26.9%
3 Self/Business/Daily 170 33.9%
9 Don't know 146 29.1%
Over 90% of the patients had been diagnosed prior to being referred to tertiary centre for treatment and 22% of them had underwent surgery in form of either modified radical mastectomy or breast conservation elsewhere (Table 2). Most of the patients had T2 disease (34.7%) followed by T3 (16%) and T4 (15%). Axillary nodes were present in 42% of the sample (Table 2).
Table 2 Frequency and percentage distribution of disease characteristics in patients
Variable Grouping Codes Group N Percentage
Symptoms 1 Lump 426 84.9%
2 Ulcer 3 0.6%
3 Discharge 28 5.6%
9 Don't know 44 8.8%
Pain 1 Yes 133 26.5%
2 No 334 66.5%
3 Don't know 35 7.0%
How diagnosed 1 Biopsy 212 42.2%
2 FNAC 239 47.6%
3 Mammogram only 4 0.8%
9 Don't know 46 9.2%
Previous Treatment 0 Nil 218 43.4%
1 Excision 84 16.7%
2 MRM 110 21.9%
3 BCT 2 0.4%
9 Don't know 88 17.6%
Tumour staging 1 T1 33 6.6%
2 T2 174 34.7%
3 T3 80 15.9%
4 T4 76 15.1%
9 TX 139 27.7%
Nodal involvement 0 N0 140 27.9%
1 N1 166 33.1%
2 N2 43 8.6%
3 N3 4 8.0%
9 NX 149 29.7%
FNAC: Fine needle aspiration cytology; BCT: Breast conservation treatment; MRM: Modified radical mastectomy.
The over all mean (±SD) quality of life score was 90.5 (±18.4) (median 87) ranging from 38–136.5. Mean score for various subscales were: physical well-being (GP) 19.8 ± 4.7; social family well-being (GS) 19.9 ± 5.3; Emotional well-being (GE) 14 ± 14.9 and functional well-being (GF) 13 ± 5.7. The mean scores for breast subscale was 23.07 ± 4.3 (median 24.8 range 10–34.7). The median and score range is detailed in table 3. The mean (±SD) subscale and scale scores for various variables are detailed in additional tables 1 and 2.
Table 3 Mean, standard deviation, median and range of the QOL scale, subscale scores and patient age.
Parameter Mean ± SD Median Range
Age 47.65 ± 11 47 20–80
General Physical well-being (GP) 19.8 ± 4.7 20 3–28
General Social well-being (GS) 19.9 ± 5.3 20.5 2.8–28
General Emotional well-being (GE) 14 ± 4.9 14 1–24
General Functional well-being (GF) 13 ± 5.7 12 2–28
Breast specific subscale (B) 23.7 ± 4.3 24.8 10–34.7
Total FACT-B score 90.5 ± 18.4 87 38–136.5
On univariate analysis, patient's education (p = 0.004), spouse occupation (0.01), number of children (p = 0.02), previous treatment (p = 0.02), nodal stage (p = 0.03), metastasis (p = 0.000) and composite stage (p = 0.000) of the disease were found to influence physical well-being (additional file 1).
The distance travelled to reach the treatment centre (p = 0.04), religion of the patient (p = 0.006) marital status (p = 0.002), education (p = 0.04) self (p = 0.02) and spouse occupation (0.04), method of diagnosis (p = 0.000), previous treatment (p = 000) and nodal status (p = 0.02) were found to significant influence emotional well-being.
Functional well-being was found to be influenced by religion (p = 0.000), patients education (p = 0.000), self (p = 0.000) and spouse occupation (p = 0.001), mode of diagnosis (p = 0.01), previous treatment (p = 0.02), and nodal status (p = 0.01). While distance travelled to the centre (p = 0.003) patients education, mode of diagnosis, previous treatment, presence of metastasis and composite stage significantly influenced breast specific subscale.
The overall quality of life was found be significantly affected by income (p = 0.03), Religion (p = 0.005), patients education (p = 0.000), self (p = 0.004) and spouse occupation (p = 0.000) presence of pain (p = 0.001), method of diagnosis (p = 0.000), previous treatment (p = 0.02), nodal stage (p = 0.01), presence of metastasis (p = 0.04) and composite stage (p = 0.005) (additional file 1).
Result of multiple logistic regression
Additional file 2 shows results of multivariate analysis. Distance travelled to the treatment centre and presence of nodal metastasis at initial presentation was found to significantly influence physical well-being. Social and family well-being was affected by religion. Emotional well-being was significantly influenced by religion and tumour stage at presentation. Functional well-being was influenced by religion, presence or absence of pain and tumour stage at presentation. While education of spouse was found to influence breast specific subscale, the overall quality of life was found to be significantly influenced by religion and tumour stage of the disease at presentation (Figure 1).
Figure 1 Factors influencing quality of life of women with breast cancer in India identified using logistic regression.
Discussion
In India, comprehensive cancer care is provided in the tertiary care centres and due to fewer numbers of such centres there are ever increasing patient load on each of them. Most patients present in locally advanced stage and achieving a good survival is still a priority. However, a few attempts have been made to comprehend and address the psychological and social needs of cancer patients [8,18,19].
The state of Kerala has a unique distinction of being 90% literate and having more females than males in the society [20]. However, as the state offers few employment opportunities, the per-capita income is low, and migration to other states and countries is high. It is also seen as a borrower's economy and is often termed as a consumer state. All these factors contribute to the state's high cost of living despite poor average earnings. Hence, developing a chronic illness or having a spouse with chronic illness like cancer would mean loss of that day's income, and extra expenditures. This reflects in the present study as well as the family income was found to significantly influence the overall quality of life.
Initial diagnosis has been shown to evoke a state of shock, fear and disbelief [21] thus creating not only a psychological crisis but an existential one as well [22]. Education has been found to significantly help one cope with these situations. In the present study too, the education was found to be a significant predictor of overall QOL in univariate analysis, however this significance was lost in the multivariate analysis. Spouse education was found to significantly influence social well-being in the univariate analysis, however in multivariate analysis it was found to significantly influence the breast specific subscale.
Culturally, Indian parents are substantially involved in their offspring's personal and social development, education, and more importantly their marriage, as majority of the marriages are arranged. Such marriages are stressful particularly for the parents of girls. Issues around dowry, sometimes described as a "social evil", play a significant determining role in marriage alliances. Adding the taboo of a parent with cancer affords even greater psychological pressure and financial burden on a family with unmarried children. This is more in patients with lower and middle income where the resources are meagre. The diagnosis of a cancer in the family also has its social stigma, which may influence the marriage prospects of the children. This is reflected in the results of the present study where number of unmarried children was found to significantly affect emotional well-being.
The present study has identified several factors that influence the QOL of the Indian breast cancer patient. Presence of pain has been identified to significantly influence physical well-being and overall QOL, stage of disease has been identified to influence functional well-being and breast specific subscale. In the univariate analysis, the distance travelled by a patient to the treatment centre significantly influenced the breast specific QOL and emotional well-being, however in the multivariate analysis it was found to significantly influence only the patient's physical well-being as expected. It was also interesting to note that though the univariate analysis did not indicate 'tumour stage' as an indicator of QOL in the breast cancer patient, the multivariate analysis showed its significant influence on emotional and functional well-being as well as on over all QOL score. In contrast to this, 'nodal involvement' was noted to influence the physical, emotional, functional well-being, and overall QOL score in the univariate analysis, but was found to significantly influence only the patients physical well-being in multivariate analysis. Several other variables that were found to have significant effect on quality of life and subscales in the univariate analysis turned out as insignificant in the multivariate analysis, viz. gender of the interviewer, and patient occupation etc.
The need for psychosocial intervention amongst cancer patients cannot be understated. The goals of planning a psychosocial intervention in the Indian breast cancer context would be to support the patient's ability to cope with the stress of treatment, helping them to tolerate short-term loss for long-term gain, and to assist in symptom management [21,24-26]. However, owing to increased patient burden, in-depth psychological intervention to each patient may not be feasible, and some sort of mechanism to cater to psychosocial problems need to be identified. Identification of the subset of women at risk is one such way forward, followed by targeted intervention that could be in form of patient education and counselling.
Competing interests
MP is the editor-in-chief of World Journal of Surgical Oncology, published by Open Access publishers Biomedcentral, which depends on Open Access model for substantial portion of its revenue.
Ethical approval
The study is approved by the institutional review board and the ethics committee.
Authors' contributions
MP: designed and coordinated the study, participated in statistical analysis, helped in preparing the draft manuscript and edited the final version for publication, beside contribution to patient management.
BCT: Participated in the study, data collection and statistical analysis and drafted the manuscript
PS: Participated in data collection and preparation of the manuscript
KR, KR, SP, BSM and BR contributed in patient management, study design and conduct and interpretation of results. They also contributed to the intellectual content of the manuscript.
All the authors read and approved the final version of the manuscript for publication
MP and KR are the guarantors of the manuscript
Supplementary Material
Additional file 1
QOL Breast 2005 showing results of univariate analysis.
Click here for file
Additional file 2
QOL breast 2005 showing results of multivariate analysis, multivariate odds ratios and p values.
Click here for file
Acknowledgement & Funding
This study is funded by a generous grant from the Indian Council of Medical Research (ICMR) New Delhi, India vide grant number 5/13/74/2000/NCD-III. The funding organisation had no role is study design. The study is independent from the funders.
==== Refs
Parkin DM Whelan SL Ferlay J Raymond L Young J (eds) Cancer incidence in five continents 1997 VII International agency for research on cancer, Lyon, France. IARC Sci Pub No. 143
National Cancer Registry Programme Biennial Report (1988–89): An epidemilogical study 1992 Indian Council of Medical Research, New Delhi
Mathew A Vijayaprasad B Cancer incidence and mortality in Trivandrum (1998–1999), population based cancer registry 2002 Regional Cancer Centre, Trivandrum
Ferrell BR Grant M Funk B Otis-Green S Garcia N Quality of life in breast cancer – Part II: psychological and spiritual well-being Cancer Nursing 1998 21 1 9 9494225
Cella DF Bonomi AE Measuring QOL: 1995 update Oncology 1995 9 47 60 8608056
Stanton AL Krishnan L Collins CA Form or function? Part 1. Subjective cosmetic and functional correlates of quality of life in women treated with breast-conserving surgical procedures and radiotherapy Cancer 2001 91 2273 2281 11413515 10.1002/1097-0142(20010615)91:12<2273::AID-CNCR1258>3.0.CO;2-1
Krishnan L Stanton AL Collins CA Jewell WR Form or function? Part 2. Objective cosmetic and functional correlates of quality of life in women treated with breast-conserving surgical procedures and radiotherapy Cancer 2001 91 2282 2287 11413516 10.1002/1097-0142(20010615)91:12<2282::AID-CNCR1259>3.0.CO;2-0
Pandey M Thomas BC Ramdas K Eremenco S Nair MK Quality of life in breast cancer patients: validation of a FACT-B Malayalam version Qual Life Res 2002 11 87 90 12018741 10.1023/A:1015083713014
Pandey M Singh SP Behere PB Roy SK Singh S Shukla VK Quality of life in patients with early and advanced carcinoma of the breast Eur J Surg Oncol 2000 26 20 24 10718174 10.1053/ejso.1999.0734
Vinokur AD Threatt BA Caplan RD Qimmerman BL Physical and psychosocial functioning and adjustment to breast cancer: long-term follow-up of a screening population Cancer 1989 63 394 405 2910447
Ganz PA Rowland JH Desmond K Meyerowitz BE Wyatt GE Life after breast cancer: understanding women's health-related quality of life and sexual functioning J Clin Oncol 1998 16 501 514 9469334
Vinokur AD Threatt BA Vinokur-Kaplan D Satariano WA The process of recovery from breast cancer for younger and older patients: changes during the first year Cancer 1990 65 1242 1254 2302673
Levy SM Haynes LT Herberman RB Lee J McFeeley S Kirkwood J Mastectomy versus breast cinservation surgery: mental health effects at long-term follow-up Health Psychol 1992 11 349 354 1286653 10.1037/0278-6133.11.6.349
Derogatis LR Morrow GR Fetting J Penman D Piasetsky S Schmale AM Henrichs M Carnicke CL Jr The prevalence of psychiatric disorders among cancer patients JAMA 1983 249 751 757 6823028 10.1001/jama.249.6.751
Stanton AL Snider PR Coping with a breast cancer diagnosis: a prospective study Health Psychology 1993 12 16 23 8462494 10.1037/0278-6133.12.1.16
Brady MJ Cella DF Mo F Bonomi AE Tulsky DS Lloyd SR Deasy S Cobleigh M Shiomoto G Reliability and validity of the Functional Assessment of Cancer Therapy-Breast quality-of-life instrument J Clin Oncol 1997 15 974 986 9060536
FACIT Manual Cella D Manual of the Functional Assessment of Chronic Illness Therapy (FACIT) Measurement System Version 4 1997 CORE, Evanston Northwestern Healthcare, Evanston, IL, USA
Thomas BC Nandamohan V Thomas I Pandey M Psychological distress among cancer patients: Development of a distress inventory for cancer J Postgrad Med 2002 48 16 20 12082320
Pandey M Thomas BC Rehabilitation of cancer patients J Postgrad Med 2001 47 62 65 11590299
Census of India 2001 Census results
Holland JC Gooen-Piels J Bast RC, Kufe DW, Pollock RE, Weichselbaum RR, Holland JF, Frei E Principles of Psycho-Oncology Cancer Medicine 2000 5 BC Decker, Ontario
Thomas BC Pandey M Ramdas K Nair MK Psychological distress in cancer patients: Hypothesis of a distress model Eur J Cancer Prev 2002 11 179 185 11984137 10.1097/00008469-200204000-00011
Ganz PA Hirji K Sim MS Schag CA Fred C Polinsky ML Predicting psychosocial risk in patients with breast cancer Med Care 1993 31 419 431 8501990
Jacobsen PB Bovbjerg DH Schwartz MD Hudis CA Gilewski TA Norton L Conditioned emotional distress in women receiving chemotherapy for breast cancer J Consult Clin Psychol 1995 63 108 114 7896975 10.1037/0022-006X.63.1.108
van't Spijker A Trijsburg RW Duivenvoorden HJ Psychological sequelae of cancer diagnosis: a meta-analytical review of 58 studies after 1980 Psychosom Med 1997 59 280 293 9178339
Payne DK Sullivan MD Massie MJ Women's psychological reactions to breast cancer Semin Oncol 1996 23 89 97 8614852
|
16188030
|
PMC1261539
|
CC BY
|
2021-01-04 16:39:03
|
no
|
World J Surg Oncol. 2005 Sep 27; 3:63
|
utf-8
|
World J Surg Oncol
| 2,005 |
10.1186/1477-7819-3-63
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1624470510.1371/journal.pgen.001004405-PLGE-RA-0099R2plge-01-04-03Research ArticleBioinformatics - Computational BiologyEvolutionInfectious DiseasesGenetics/Comparative GenomicsDrosophilaVirusesPositive Selection of Iris, a Retroviral Envelope–Derived Host Gene in Drosophila melanogaster
A Retroviral
Envelope-Derived Host Gene
Malik Harmit S 1*Henikoff Steven 121 Basic Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, Washington, United States of America
2 Howard Hughes Medical Institute, Fred Hutchinson Cancer Research Center, Seattle, Washington, United States of America
Barsh Gregory EditorStanford University School of Medicine, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 21 10 2005 1 4 e4410 5 2005 1 9 2005 Copyright: © 2005 Malik and Henikoff.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.Eukaryotic genomes can usurp enzymatic functions encoded by mobile elements for their own use. A particularly interesting kind of acquisition involves the domestication of retroviral envelope genes, which confer infectious membrane-fusion ability to retroviruses. So far, these examples have been limited to vertebrate genomes, including primates where the domesticated envelope is under purifying selection to assist placental function. Here, we show that in Drosophila genomes, a previously unannotated gene (CG4715, renamed Iris) was domesticated from a novel, active Kanga lineage of insect retroviruses at least 25 million years ago, and has since been maintained as a host gene that is expressed in all adult tissues. Iris and the envelope genes from Kanga retroviruses are homologous to those found in insect baculoviruses and gypsy and roo insect retroviruses. Two separate envelope domestications from the Kanga and roo retroviruses have taken place, in fruit fly and mosquito genomes, respectively. Whereas retroviral envelopes are proteolytically cleaved into the ligand-interaction and membrane-fusion domains, Iris appears to lack this cleavage site. In the takahashii/suzukii species groups of Drosophila, we find that Iris has tandemly duplicated to give rise to two genes (Iris-A and Iris-B). Iris-B has significantly diverged from the Iris-A lineage, primarily because of the “invention” of an intron de novo in what was previously exonic sequence. Unlike domesticated retroviral envelope genes in mammals, we find that Iris has been subject to strong positive selection between Drosophila species. The rapid, adaptive evolution of Iris is sufficient to unambiguously distinguish the phylogenies of three closely related sibling species of Drosophila (D. simulans, D. sechellia, and D. mauritiana), a discriminative power previously described only for a putative “speciation gene.” Iris represents the first instance of a retroviral envelope–derived host gene outside vertebrates. It is also the first example of a retroviral envelope gene that has been found to be subject to positive selection following its domestication. The unusual selective pressures acting on Iris suggest that it is an active participant in an ongoing genetic conflict. We propose a model in which Iris has “switched sides,” having been recruited by host genomes to combat baculoviruses and retroviruses, which employ homologous envelope genes to mediate infection.
Synopsis
Mobile genetic elements have made homes within eukaryotic (host) genomes for hundreds of millions of years. These include retroviruses that integrate into host genomes as an essential step in their life cycle. While most such integration events are likely to be either deleterious or of little consequence to the host, on rare occasions host genomes can preserve and exploit capabilities of mobile elements for their own function. Especially intriguing are instances where host genomes have chosen to retain the envelope genes of retroviruses; the same envelope genes are responsible for conferring infectious ability to retroviruses. Primates and rodent genomes each have domesticated retroviral envelope genes (called “syncytin” genes) for important roles in placental function.
Now, Harmit Malik and colleagues show that a similar, ancient domestication event has taken place within the fruit fly Drosophila melanogaster. They identify a gene, Iris, which was acquired from an envelope gene of insect retroviruses, and has been maintained as a host gene for more than 25 million years. Unexpectedly, the authors find that Iris continues to evolve rapidly whereas previous studies have shown that mammalian syncytin genes do not. They suggest a model in which the Iris gene has “switched sides,” from its original role in causing infections to its current role in preventing them.
Citation:Malik HS, Henikoff S (2005) Positive selection of Iris, a retroviral envelope–derived host gene in Drosophila melanogaster. PLoS Genet 1(4): e44.
==== Body
Introduction
Despite the fact that mobile elements are generally detrimental to host fitness, there are several instances where eukaryotic genomes have harnessed the enzymatic machinery of transposable elements to perform a myriad of important functions. For instance, the reverse transcriptase activity of the telomerase enzyme, which protects the ends of linear chromosomes [1], is believed to be the ancient descendant of prokaryotic mobile genetic elements [2]. In several species of Drosophila, active Het-A and TART retroposons still carry out this important function [3,4]. The core enzymatic machinery used to carry out V(D)J recombination in the generation of antigen recognition diversity is encoded by the RAG1/RAG2 proteins, believed to be descended from a previously autonomous transposon [5,6]. Many human genes are derived from the integrase machinery of transposable elements [7–9], and although their function is still unknown, many of them appear to have conserved their enzymatic ability [10]. Host genomes can also employ mobile elements' genes for genome defense. In murine genomes, a domesticated retroviral gag gene, Fv1, can defend mouse cells against infections by exogenous retroviruses [11]. These represent examples of how host genomes can acquire and eventually exploit the enzymatic capabilities of mobile elements for host functions.
“Domestication” of retroviral envelope (env) genes is especially intriguing in this context. While the env gene usually confers infectious ability to retroviruses, the human endogenous retrovirus-W env gene now appears to play a critical role in placental morphogenesis in higher primate genomes [12]. This gene, called syncytin, is still present in the context of a human endogenous retrovirus-W provirus that entered the primate lineage about 35 million years ago [13], indicating that it is still at the early stages of “evolutionary domestication” in its transition from a retroviral env to a host gene [14,15]. Indeed, selection pressures on the rest of the retroviral sequence show early signs of decay, but the syncytin gene itself is under strong selective constraints and is conserved among all hominoids and Old World monkeys [14]. Thus, while the endogenous retrovirus itself has lost the service of its env gene, host genomes now exploit this gene's membrane-fusion ability to carry out the important process of trophoblast differentiation [12,16]. Recently, three other retrovirus env-derived host genes have been described. Syncytin-2 is a 35-million-year-old host gene also found in primate genomes, which is derived from human endogenous retrovirus-FRD and appears to be predominantly expressed in placenta [17]. Two separate retrovirus-derived fusogenic env genes, syncytin-A and syncytin-B, have been shown to be expressed in murine placental tissues [18]. These genes represent a remarkable case of convergent evolution where rodent and primate genomes have each acquired retroviral env genes for important roles in placental differentiation.
Most retroviruses appear to be derived from ancestral non-viral retrotransposons that lacked infectious ability [19,20]. Phylogenetic analysis suggests that the acquisition of env genes drove the evolutionarily important transition from a non-viral retrotransposable element to an infectious retrovirus on at least nine occasions [20,21]. Two of these instances led to the gypsy and roo retroviruses in Drosophila, which have both separately acquired homologous env genes from baculoviruses, double-stranded DNA viruses with large genomes [20,22]. Many baculoviruses employ this env gene for mediating infection [23]. In both retroviruses and baculoviruses, infectious ability requires a proteolytic cleavage to separate the envelope protein into the SU (receptor-binding component) and TM (brings membranes into close apposition and causes fusion) proteins. Just downstream of furin cleavage site is a hydrophobic fusion peptide that is also required for membrane fusion [24,25].
The release of the D. melanogaster genome sequence [26] provided a unique resource to help address the chronology of env acquisition by retroviruses. For instance, it gave a sequence snapshot of all proviral insertions in the D. melanogaster genome [27,28]. Compared to mammalian genomes, Drosophila genomes have a higher rate of DNA loss [29], thus proviral sequences are more likely to reflect recent insertion events or insertions that have been selectively retained. In our survey, we unexpectedly found that the D. melanogaster genome contains a host gene, CG4715 (renamed Iris in this paper), which is homologous to the env genes from baculoviruses and insect retroviruses (also identified in [22] [30]). We have now investigated the evolution of Iris in insect genomes, and found it to be conserved in most Drosophila species of the Sophophora subgenus. We can trace the acquisition of this env gene to a sister lineage of the roo insect retroviruses (named Kanga in this paper). Investigation of the selective constraints on Iris reveals that it has been subject to positive selection throughout its evolution in Drosophila. This unusual finding of positive selection on a domesticated retroviral env gene suggests that it is an active participant in an extant genetic conflict in its host genomes, possibly to combat against insect viruses that bear homologous env genes.
Results
CG4715 is a Viral Envelope–Related Host Gene in Drosophila
In order to investigate whether or not the D. melanogaster genome had domesticated any retroviral genes, we initiated searches of the databases by PSI-BLAST using the various encoded genes from the gypsy and roo insect retroviruses. We found a strong match to their env genes in a previously unannotated gene, CG4715, in the D. melanogaster genome [22]. The genomic regions surrounding CG4715 bear no discernible similarity to baculoviral or retroviral sequences, ruling out the possibility that CG4715 represents the evolutionary remnant of a recent retroviral-introduced provirus or a baculoviral insertion. Figure 1A schematizes the genomic contexts of the env homologs found in baculoviral, retroviral, and the D. melanogaster genomes. CG4715 bears many of the hallmarks of the gypsy and roo env genes, including the same architecture consisting of a signal peptide and a carboxyl-terminal hydrophobic peptide that is likely to be membrane-spanning (Figure 1B and 1C).
Figure 1
CG4715 Homologs
(A) Baculoviral and insect retroviral env genes shown in their respective genomic context. Baculoviruses, represented by Autographa californica nucleopolyhedrovirus (ACNV) and Lymantria dispar nucleopolyhedrosis virus (LDNV) are double-stranded DNA viruses whose genome size is close to 150 kilobases [72], while retroviruses, represented by roo and Gypsy, are close to 7 kilobases in length [73]. CG4715 is an open reading frame found in the same genomic context in many species of Drosophila. CG4715/Iris and its env homologs are shown in black (open reading frame direction shown by arrows) while neighboring genes are shown in gray. Note that the gypsy env is expressed through a spliced message. Kyte-Doolittle hydropathy plots of encoded protein products from CG4715 (B) and the roo env gene (C) are shown. The putative signal peptide (SP) and C-terminal, transmembrane hydrophobic peptide (Tm) are highlighted in bold, while the furin cleavage site in the roo envelope protein is indicated by an arrowhead.
We obtained CG4715 sequence from ongoing genome sequencing projects in several species of Drosophila using synteny (gene order) and TBLASTN searches. We screened for the presence of CG4715 in closely related species of the Sophophora subgenus of Drosophila using PCR and primers designed to flanking sequences (see Materials and Methods), and were able to confirm the presence of CG4715 in several additional species of Drosophila (Figure 2A and 2C). During our sequencing efforts, we uncovered two CG4715-related genes in tandem orientation in all species of the takahashii/suzukii subgroups. Figure 2C represents the phylogenetic analysis of CG4715 genes in the Sophophora subgenus of Drosophila (based on a partial alignment of their coding sequences), whose phylogenetic relationship is in good agreement with the widely accepted phylogeny of this genus [31,32] (schematized in Figure 2B). This indicates that this gene has been inherited strictly by vertical inheritance rather than by horizontal transfer, a conclusion that is supported by the fact that CG4715 is found in the same syntenic location in different species (Figure 2A). Of the two CG4715 genes found in the takahashii/suzukii groups, the 5′ gene (referred to as CG4715-A) represents the true ortholog, while the second (CG4715-B) represents a gene duplication whose phylogenetic position (Figure 2C) is incongruent with the expected species phylogeny (schematized in Figure 2B). This phylogenetic placement could result from altered selective constraints (and different evolutionary rates) that could lead to a phylogenetic artifact known as “long-branch attraction” [33]. While we cannot rule out an ancient origin of the B lineage, this would lead to the unparsimonious implication that this gene was subsequently lost in all species except those from the takahashii/suzukii species groups.
Figure 2 Phylogenetic Analysis of CG4715 Homologs
(A) CG4715 has been preserved in its syntenic location in Drosophila species. In species from the takahashii/suzukii species groups like D. lutescens, an additional paralog, CG4715-B (gray shading) is found in tandem orientation. D. ananassae has an additional transposon insertion in this syntenic location between CG4715 and CG4552, while the genomes of D. mojavensis and D. virilis lack CG4715 orthologs between CG4577 and CG4552. For D. ananassae and D. pseudoobscura, sequence was obtained from genome sequencing data (indicated with an asterisk) and confirmed by sequencing.
(B) An “expected” phylogeny of Drosophila species is shown, summarizing results from many genes [30,31].
(C) A neighbor-joining phylogeny of CG4715 orthologs based on C-terminal amino acid sequence is presented. (For some species, only the C-terminal sequence was obtained (indicated by a “p” for partial)). This phylogeny is largely in agreement with the accepted species phylogeny in (B), indicating that the gene has been inherited by strict vertical inheritance. Although there is a slight discordance in phylogenetic placement of the D. ananassae, D. eugracilis, and D. auraria, these branches have only a low bootstrap support. A second lineage of CG4715 paralogs, CG4715-B is evident (gray shading) in the takahashii/suzukii species groups.
In D. mojavensis and D. virilis, whose genome sequences are still incomplete, CG4715 is absent from its syntenic location, and we have not found true orthologs in other genomic locations. While it remains formally possible that the location of CG4715 is altered in these two species, it is more likely that CG4715 does not exist as a host gene in these species (BLAST searches did not reveal any orthologs). The latter possibility could be due to a subsequent loss event in D. mojavensis and D. virilis (both belong to the Drosophila subgenus, Figure 2B) or because CG4715 originated only after the separation of the Sophophora and Drosophila subgenera. Completion of ongoing sequencing projects in the D. willistoni, D. mojavensis, D. virilis, and D. grimshawi species will help distinguish among these possibilities.
CG4715 is the Domesticated Envelope Gene of the Kanga Insect Retroviruses
Where did CG4715 come from? The closest homologs to CG4715 in the available sequences of all Drosophila genomes are the env genes of a novel lineage of retroviruses, which appear in several species of the Sophophora subgenus (ongoing sequencing projects, see Materials and Methods). This lineage of retroviruses is most closely related to the roo lineage of BEL-like retroviruses, and we refer to it as Kanga [34–36]. In a detailed phylogenetic analysis of all CG4715-env related genes (Figure 3A), the CG4715 orthologs unambiguously branch together with the env genes of Kanga. We also investigated genome sequences from other insects for CG4715 homologs. Remarkably, the Anopheles gambiae genome also contains a homolog of CG4715 with the same architecture. Like the Drosophila CG4715 gene, the A. gambiae gene is not flanked by regions homologous to either retroviral or baculoviral sequences. Using the A. gambiae gene as a query, we were able to successfully retrieve its Aedes aegyptii ortholog as well. We can detect Kanga-roo-like retroviruses in the lepidopteran Bombyx mori (silkmoth) genome, but not in the Apis mellifera (honeybee) genome. Intriguingly, while the A. gambiae genome has multiple retrotransposons related to the Kanga-roo retroviruses, none of these is predicted to encode an env gene. The primary reason that the Kanga retroviruses have not been described so far appears to be their absence in the earliest sequenced insect genomes, including D. melanogaster and A. gambiae.
Figure 3
CG4715/Iris Relationships to Viral Envelope Genes
(A) The central domains of CG4715 and related viral env genes were aligned, and a neighbor-joining phylogenetic tree constructed. The tree separates the CG4715-env superfamily into four groups: the baculoviruses, the BEL clade retroviruses roo and Kanga, the gypsy-like retroviruses, and host genome borne CG4715 orthologs in Drosophila and mosquito genomes. While the tree overall does not provide high resolution to discern the order of divergence of each of the clades, there is very strong phylogenetic resolution (bootstrap support of key nodes shown) to unambiguously group CG4715 orthologs with the Kanga retrovirus lineage, indicating that this lineage of retroviruses is the likely source of the CG4715 lineage.
(B) Neighbor-joining phylogeny of selected representatives from the BEL clade of retrotransposons indicates that the Kanga retroviruses from Drosophila genomes form a monophyletic clade (the presumed ancestor of CG4715 is indicated by a yellow oval). Most retrotransposons in the BEL lineage do not possess an env gene (blue lettering) while many elements that do (red) have acquired non-homologous env genes acquired from a different viral source [19,20].
On a phylogenetic tree of all CG4715-env related homologs (Figure 3A), the two mosquito genes represent a distinct lineage from that of the Drosophila CG4715 orthologs and Kanga retroviruses. Parsimony criteria based on the phylogeny in Figure 3A strongly argues that the retroviral borne env gene represents the ancestral form. We can say with high confidence that the Drosophila CG4715 genes have derived from within the Kanga retroviral lineage (bootstrap support on relevant nodes is highlighted in Figure 3A). Thus, we conclude that there have been two separate domestications of insect retroviral env genes in fruit fly and mosquito genomes, respectively. The domestication event in the Sophophora genus of Drosophila led to CG4715, which now has been preserved as a host gene. It is present in all species tested, and appears to have been inherited strictly by vertical inheritance for at least 25 million years (the estimated time of separation of D. melanogaster and D. pseudoobscura [31]).
To better gauge the evolutionary origins of the newly identified Kanga retroviruses, we compared the majority of the pol sequence (PR-RT-RNH domains) of Kanga to other known insect retrotransposon lineages. These showed that Kanga retroviruses belong unambiguously to the BEL clade, which also includes the roo but not the gypsy retroviruses (Figure 3B). Previous analyses have shown that only a few lineages of the BEL elements also possess env genes (red in Figure 3B) and that the Caenorhabditis elegans and D. melanogaster retroviruses of this lineage have non-homologous env genes[19,20], indicating that the non-infectious retrotransposons (blue lineages) are likely to be the ancestral form.
Etymology
Based on our phylogenetic analyses (Figure 3A), it is clear that CG4715 orthologs represent a sister lineage to the env genes from Kanga retroviruses. In Greek mythology, the Titan Thaumas and Electra had two sets of offspring. The first were the winged monsters, the Harpies (which we liken to the insect retroviral and baculoviral env genes). The second was Iris, the goddess of the rainbow and the messenger of the god Zeus and his wife, Hera. Since CG4715 is clearly maintained as a host gene, we use the analogy to the benevolent sibling of the mythological Harpies to propose the name Iris to represent the CG4715 orthologs, since they are related to viral env genes but are presumably beneficial to the host genome, based on their conservation.
Iris Expression
Its strong conservation suggested that Iris might perform some important function in insects. To investigate this, we examined Iris expression in D. melanogaster. Using RT-PCR and Northern blots on pools of polyA RNA representing all life-stages of D. melanogaster, we determined that Iris is expressed primarily in adults in both males and females, with weak expression at the third instar larval stages (Figure 4A and 4B). By RT-PCR analysis on individually dissected tissues, we found Iris is transcribed in most adult tissues, with expression only slightly lower in ovaries and testes (Figure 4C). Our RT-PCR results are consistent with what was observed in a recent large-scale survey of Drosophila gene expression patterns in ovaries, testes, and the soma [37]. The expression pattern appears to suggest that Iris may have been domesticated for some role in adult flies, either within germline or somatic tissues, or both.
Figure 4
Iris Expression in D. melanogaster
Iris expression through various stages of development was assayed using (A) RT-PCR and (B) Northern blots. Both show that Iris is predominantly expressed in adult females and males. (C) RT-PCR analysis on individually dissected tissues from adult flies shows that Iris is expressed in somatic tissues but expression is slightly reduced in ovaries and testes. RT-PCR to Karyopherin alpha-3 (αKap3, a ubiquitous nuclear import factor- CG9423) is shown as a control for amounts of template RNA in the RT-PCR reaction, and to show that there is no detectable contamination from genomic DNA.
Conserved Features among Iris Orthologs and Paralogs
An amino acid alignment of all full-length Iris orthologs is shown in Figure 5. Several features are conserved, including a signal peptide (putative cleavage site shown by arrowheads) and a C-terminal hydrophobic peptide that presumably represents a membrane-spanning segment by analogy to the retroviral envelope proteins. In addition, several cysteine residues (highlighted in yellow) are variably conserved. Co-conservation of particular cysteine residues suggests that these cysteine residues participate in a disulfide bond either within the same molecule or across different molecules (“1–1” and “2–2”). Six cysteine residues (c1 through c6) are invariant; these are also highly conserved across all of the homologous retroviral env genes (Figure 3A). In general, the C-terminal domain of Iris is more conserved than the N-terminus among orthologs, and between Iris and retroviral envelope proteins. Some residues at the C-terminus, after the predicted membrane-spanning peptide, are also highly conserved (PLLEK amino acid residues). Based on bioinformatic predictions and by analogy to retroviral envelope proteins, this represents the cytoplasmic tail of Iris, and suggests that this may participate in either the physical anchoring of Iris at the cell membrane, or some downstream signal transduction.
Figure 5 Complete Alignment of Iris Proteins
An alignment of full-length Iris proteins from various Drosophila species is shown. All invariant residues are shown against a black background (except cysteines that are highlighted in yellow), while similar residues are highlighted in gray background. We did not include the Iris-B lineage here for ease of presentation (these are presented in Figure 6). Several features are conserved, including the signal peptide (predicted cleavage site indicated by arrowheads), C-terminal transmembrane domain (shown as a box), and several invariant cysteine residues (c1 through c6, highlighted in yellow) that are a characteristic feature of Iris and related envelope proteins. Other cysteine residue pairs (1–1 and 2–2, also highlighted in yellow) show co-conservation, i.e., loss of one results in loss of the other.
Figure 6
Iris Paralogs in the takashii/suzukii Species Groups
(A) An alignment of representative Iris-A and Iris-B proteins from the takahashii/suzukii species groups is shown. Iris-A and Iris-B proteins are highly similar to each other. Notable differences include pairs of cysteine residues that are conserved in the B lineage (indicated with “B”), but not in A. The B lineage also has a shorter cytoplasmic tail and is missing several residues (PLLEK amino acid residues) that are invariant in the A lineage. In addition, an internal segment of the Iris-A protein is lost from the Iris-B protein, by virtue of this genomic sequence becoming an intron (confirmed by RT-PCR).
(B and C) Hydropathy plots of representative Iris-A and Iris-B proteins show that the overall architecture of the two proteins is largely unaffected by the differences between the two lineages.
(D) A hypothetical model for the origin of the divergent Iris-B gene starts with the tandem gene duplication. A cryptic SA site is encountered by mutation, but this can be neutrally maintained. However, the simultaneous occurrence of an SD site activates the SA site and leads to a portion of the coding exon being spliced out from the mature message. If this is deleterious, the SD-SA combination is culled out by selection. However, in rare cases, like the Iris-B gene, this could lead to a novel functional gene that is favored by selection. Subsequently, the SD and SA sites are maintained by purifying selection.
One feature that is almost universally conserved among retroviral envelope proteins is a furin cleavage site followed by a hydrophobic peptide that represents the fusion peptide. Surprisingly, we found that the Iris protein in D. melanogaster lacks the central furin cleavage site and fusion peptide found in all env genes capable of mediating infection. We investigated when this cleavage site degenerated on the Iris-env phylogeny (Figure 3A). We employed a MAST search [38] using a position-specific scoring matrix constructed from a subset of retroviral homologs as previously described [20]. As a positive control, we used retroviral and baculoviral env genes where we knew that the furin cleavage site was conserved. For a negative control, we used homologous baculoviral genes where the furin cleavage site has been shown to have degenerated [39]. We queried three Iris proteins (from D. melanogaster, D. ananassae, and D. pseudoobscura), one domesticated mosquito gene (from A. gambiae), and the env gene from the Kanga retroviruses using this consensus. Using this strategy, we find that both Kanga retroviruses and the domesticated envelopes from mosquito genomes have a conserved furin cleavage and fusion peptide (E-value < 0.001), while this site is not conserved in any of the Iris proteins (E-value > 10). This suggests that the fruit fly and mosquito domestication events differ both chronologically and qualitatively, and that this cleavage site has been lost in the Iris lineage. This loss of cleavage is especially noteworthy since other architectural features, including several conserved pairs of cysteine residues (c1 through c6) presumed to be necessary for membrane fusion ability and post-cleavage interactions between the SU and TM domains, are still conserved [22] (Figure 5). This suggests that while these features are essential for membrane fusion, they may also serve another function.
A Second Iris Gene in the takahashii/suzukii Species Groups: A New Mode of Neofunctionalization?
All Drosophila species that we have investigated in the Sophophora subgenus (Figure 2B) possess an Iris ortholog in a syntenic location. Surprisingly, the takahashii/suzukii species groups have two genes that are found in tandem orientation (Figure 2A). We have shown that the first of these (Iris-A) represents the true ortholog while the second (Iris-B) is paralogous (Figure 2C). At first glance, the second gene (Iris-B) appears to be a pseudogene. All other Iris orthologs are intron-less genes. Based on this expectation, Iris-B (which is the same length as Iris-A) has frameshifts and nonsense codons. However, when we did RT-PCR on this gene in D. lutescens and D. prostipennis, we found that these genes had spliced out an intron of ~70 nucleotides. Removing this intron now recapitulates an open reading frame that is highly homologous to that of Iris-A. We found that the splice acceptor (SA) and splice donor (SD) sequences are invariant, and we conclude that all Iris-B genes possess a single intron. An amino acid sequence comparison of representative Iris-A and Iris-B genes is presented in Figure 6A. Once again, the C-terminal half of the gene is well conserved (including c1 through c6, highlighted in Figure 5), with more variation in the N-terminus. Hydropathy plots (Figure 6B and 6C) illustrate that despite the loss of exonic sequence, the overall architecture of the Iris-B proteins is largely unaltered. Some differences are apparent, however. For instance, Iris-B lacks the conserved residues at the C-terminus of Iris-A (PLLEK amino acid residues). Additionally, Iris-B has some conserved pairs of cysteine residues that are not found in Iris-A, suggesting that Iris-B now operates under altered selective constraints. This may be partly responsible for the “early branching” of the B lineage in the Iris phylogeny (Figure 2C).
Their maintenance since the evolutionary origin of the takahashii/suzukii groups suggests that the Iris-B genes are evolving under selective constraints. Following gene duplication, a duplicate gene can suffer three fates: non-functionalization (degeneration of function), neofunctionalization (new function), or subfunctionalization (assortment of ancestral functions). We cannot distinguish between the latter two possibilities. Nonetheless, the striking differences in Iris-B compel a hypothetical parsimonious model (Figure 6D) as to how the differences arose. Cryptic SA sites likely occur and are lost neutrally, but the simultaneous gain of an SD site leads to a selective decision. If the spliced product is non-functional, the SD-SA combination is lost. However, if there is a sufficient selective advantage for the truncated protein, then the SD-SA combination will sweep through the population and be maintained by purifying selection. Previously, there has been at least one well-documented instance of a previously intronic or intergenic sequence becoming exonic and a previously exonic sequence becoming a promoter (the Sdic gene in D. melanogaster [40]), but the de novo “invention” of an intron in what previously was exclusively exonic sequence appears to be a novel finding. The scenario we have presented in Figure 6D is simple but likely to be quite rare. However, unlike the much more frequent event of intron transposition, it is conceivable that intron invention may have contributed extensively to the gain of new protein functions.
Molecular Evolution of Iris in Drosophila Species
Most retroviral insertions into the host genome are either detrimental or selectively neutral. Therefore, upon insertion into host genomes, these proviral genes start decaying due to mutation. However, retroviral genes that are beneficial to the host genome can be domesticated; these genes can evolve either under purifying or positive selection. In the first scenario, the newly domesticated gene now carries out a housekeeping function, and selective pressures cull out deleterious mutations, including the majority of those that change the amino acid sequence. The mammalian domesticated syncytin gene falls into this category [12,14,15]. On the other hand, the host could also recruit a retroviral gene to protect itself from future rounds of infections, as murine genomes appear to have done with the domestication of a gag gene, Fv1 [11,41], or an env gene, Fv4 [42]. In either scenario, i.e., housekeeping or defense, the domesticated gene is likely to be well conserved because it confers a selective advantage, but the selective pressures are quite distinct and likely to discriminate among possibilities of function. For instance, in the latter host defense scenario, the newly domesticated gene might evolve rapidly at the amino acid level due to selective pressures to keep pace with rapidly evolving infectious agents, as is the case for Fv1 [41].
What selective constraints have shaped Iris evolution? Since Iris is a host gene related to retroviral env genes, we were interested in investigating the selective pressures under which it has evolved. We compared synonymous (dS) and non-synonymous (dN) nucleotide changes in five, non-overlapping pair-wise comparisons across the Drosophila phylogeny [43]. These results are presented in Figure 7 and highlight the variable nature of selective constraints, which have acted on Iris in the course of its evolution in Drosophila species. We find several windows where dN/dS significantly exceeds 1, but the location of these windows is variable from one pair-wise comparison to the next. In the case of the Iris paralogs in the takahashii species group, we find no evidence of positive selection in the Iris-A comparison (Figure 7D), but a significant window in the N-terminus of Iris-B (Figure 7E). This could simply reflect stochastic differences, but it is intriguing that the Iris-A comparison is the only one in our set that did not show any windows where dN/dS significantly exceeds 1.
Figure 7 Sliding Window dN/dS Analyses of Different Drosophila Iris Genes
We have chosen non-overlapping sets of the Drosophila species to do a pair-wise analysis of dN compared to dS. We present a sliding window analysis (window size 150 base pairs, slide of 50 base pairs) of dS and the dN/dS ratio (y-axis) plotted against nucleotide position (x-axis). Under neutrality, a dN/dS ratio of 1 is expected (dashed lines). We present a comparison of (A) D. melanogaster versus D. simulans, (B) D. yakuba versus D. teissieri, (C) D. erecta versus D. orena, (D) D. paralutea A versus D. lutescens A, and (E) D. paralutea B versus D. prostipennis B. In all these comparisons except (D), at least one window where dN/dS significantly exceeds 1 is seen (indicated by asterisks; significance tested by simulations in the K-estimator program [43]).
We also performed a maximum likelihood based analysis of selective pressures acting on Iris using the PAML and random effects likelihood (REL) and fixed effects likelihood (FEL) programs [44,45]. We chose only a closely related set of full-length Iris orthologs (12 total up to D. eugracilis) for this purpose, to minimize the number of gapped positions in the alignment. We excluded all positions with gaps to avoid any ambiguities in alignments. Notably, these gapped regions had the maximum variability in sequence. Results from these analyses are shown in Figure 8A and Table 1. A whole-gene assignment of dN/dS ratios to the different branches of the Iris phylogeny is shown in Figure 8A. Only three branches were shown to have a dN/dS greater than 1. This is not surprising because domains subject to purifying selection (where dN/dS is less than 1) can mask the signal of windows of positive selection such that the overall dN/dS in the gene does not exceed 1. In spite of this, we found that the lineage leading up to the sibling species D. mauritiana, D. simulans, and D. sechellia had a dN/dS ratio of 1.82. Using PAML comparisons in which this branch was fixed at dN/dS = 1 versus dN/dS = 1.82, we found weak evidence that positive selection occurred on this branch (highlighted in Figure 8A; 2ΔlnL = 3.1 and 1 degree of freedom, p < 0.08) despite the fact that the whole gene was being analyzed.
Figure 8 PAML Analyses of Iris Evolution
(A) A free-ratio model for Iris evolution in Drosophila is presented with numbers above branches indicating (whole gene) dN/dS ratios estimated for each individual branch. Only the lineage leading to the sibling species D. mauritiana, D. sechellia, and D. simulans (thick line) has a dN/dS ratio that appears to be greater than 1. When this value of dN/dS = 1.82 was compared against the neutral expectation of 1, the higher value fit the data marginally better (p < 0.08).
(B) Individual residues highlighted by PAML analyses as having being subject to recurrent positive selection are shown by inverted triangles. Also schematized are the signal peptide cleavage site (arrowheads) and C-terminal hydrophobic peptide (box). Dark, dashed lines indicate the ten cysteine residues (1–1, 2–2, c1 through c6) highlighted in Figure 5. We note that most of the residues identified at high confidence appear to cluster around the 2–2 pair of cysteine residues, suggesting a functional interaction surface here [46].
Table 1 PAML and REL Analyses of Iris in Drosophila [44]
A whole gene dN/dS ratio comparison can fail to identify specific domains or residues subject to positive selection. We investigated this latter possibility on the multiple alignment of Iris from 12 Drosophila species using a comparison of NSsites model M7 (a beta distribution with no positive selection) and model M8 (a beta distribution with positive selection permitted). We find that model M8, which allows one class of codons to have allowed under positive selection, fits the data significantly better (Table 1, p < 0.002). Thus, we conclude that Iris has been subject to positive selection through this period of Drosophila evolution. This analysis also highlights a few residues as being repeatedly subjected to positive selection (posterior probability > 0.95 in Table 1). There is remarkable congruence between these results and those obtained from a similar REL analysis and the more conservative FEL analysis (Table 1). Of the nine residues that were identified by the PAML analysis over the entire protein (~ 500 residues compared), six are clustered within 15 amino acids around the 2–2 pair of cysteine residues (Figures 5 and 8B). We have previously tested “patches” of positive selection similarly identified by PAML analyses in the retroviral defense gene TRIM5α and have shown that they represent interaction interfaces between host and viral proteins [46]. These analyses suggest that the 2–2 pair of cysteine residues may encode a similar interaction interface.
To investigate the effects of positive selection on standing genetic variation, we sequenced Iris from a variety of strains of D. melanogaster (14 strains) and D. simulans (17 strains) to carry out population genetic analyses. Using the McDonald-Kreitman test, we first compared fixed interspecies differences to intraspecific polymorphisms at replacement and synonymous sites [47]. Fixed Rf:Sf changes between the two species are 77:25, while the polymorphic Rp:Sp ratio is 90:36. These values are not significantly different from each other (p ~ 0.5). Polarizing changes to just the D. melanogaster lineage (40:17 versus 44:17) or just the D. simulans lineage (49:16 versus 46:21) also did not reject the null expectation. One potential source of discordance between the dN/dS and the McDonald-Kreitman test results could be strong selective pressures acting on the intraspecific polymorphisms, compared to interspecific divergence. This could suggest, for instance, that the bulk of the dN/dS signal observed in Figure 7A was in fact due to intraspecific polymorphisms. However, we confirmed that this was not the case by reconstructing the hypothetical ancestor of all D. melanogaster and all D. simulans strains and performing a pair-wise dN/dS comparison, which is practically identical to Figure 7A (unpublished data).
Iris and the Phylogeny of D. simulans Sibling Species
Positive selection may have had a strong impact on Iris evolution even in closely related species, due to species-specific infections by mobile elements. Horizontal transfers of DNA-mediated transposons and LTR-retrotransposons [28,48–50] can lead to species-specificity of transposon propagation. These selective pressures could be predicted to lead to the rapid fixation of Iris polymorphisms in a species-specific manner, which might subsequently resist introgression of alleles from other species because of constant selective pressures. We tested these possibilities by comparing Iris sequences from several strains of D. simulans, D. mauritiana, and D. sechellia since these species appear to have the most striking signature of positive selection (Figure 8A). These three species are believed to have separated less than 500,000 years ago [51]. In our phylogenetic analysis (Figure 9A), Iris sequences from each species form their own exclusive clade to a high degree of statistical support, in large part due to six nucleotide differences that are unambiguously diagnostic for branching order within these three sibling species.
Figure 9
Iris Phylogeny in Closely Related Species
(A) Phylogenetic analysis of Iris coding regions from different strains of D. melanogaster, D. simulans, D. sechellia, and D. mauritiana, the latter three species believed to have diverged less than half a million years ago [51]. Based on distance, parsimony or likelihood methods (bootstrap values indicated in ovals), the phylogeny clearly separates the three species. This is largely due to six sites that are “unambiguous” as far as phylogenetic information is concerned, indicated with “!.” An unambiguous site is defined as one in which the same derived nucleotide is found fixed in two of the three species (e.g., D. simulans and D. sechellia), whereas the third species (e.g., D. mauritiana) is fixed for the ancestral nucleotide, corresponding to the out-group, D. melanogaster.
(B) Iris is only the second known gene to inform about the phylogeny of the three sibling species D. simulans, D. sechellia, and D. mauritiana with statistical significance. In the Iris phylogeny, D. mauritiana branched earliest while previously, D. sechellia was found to branch earliest. This suggests that speciation events' chronology among these three species is more complicated than suggested previously [52].
The ability to phylogenetically separate these three species has only been seen previously for the Odysseus homeobox (OdsH) gene [52] that has been proposed to play a role in hybrid sterility [53]. The difficulty in resolving these relative recent speciation events is likely to result from the persistence and possibly introgression of ancestral alleles following speciation [51,52]. Indeed, since only speciation genes would be able to resist the effects of introgressed alleles from other species, it has been previously suggested that only these would have the required resolution to trace the exact chronology of reproductive isolation among recently diverged species. Based on the OdsH gene, the case has been made for allopatric speciation among the sibling species D. simulans, D. mauritiana, and D. sechellia, with D. sechellia branching first [52].
Our results call into question the generality of these previous conclusions. While Iris also resolves the phylogeny to almost the same degree of certainty, the chronology of events traced by Iris are different from those traced by OdsH. Thus, in the case of OdsH, six “unambiguous” sites indicated that D. sechellia was the out-group, while one site indicated that D. simulans was the out-group [52]. In the case of Iris, five sites (all in the N-terminus) indicate that D. mauritiana was the out-group species while one (the most C-terminal) indicates that D. sechellia was the out-group. We suggest that it is likely that all these phylogenetic reconstructions simply reflect the fact that a recent episode of positive selection affected only two out of three species, rather than the true chronology of reproductive isolation. Notably, OdsH is under strong positive selection between D. mauritiana and D. simulans, and its phylogeny groups these two species [52,53]. Similarly, there appears to be clear evidence that Iris is significantly diverged because of a species-specific selective pressure.
An important caveat is that both these genes reside on different chromosomes: OdsH on the X and Iris on 2L. Divergent selective regimes could have led to independent, species-specific chromosomal “speciation” events, although it is difficult to imagine how this could have been achieved in strict allopatry if they occurred simultaneously. Alternatively, the OdsH and Iris phylogenies could reflect temporal differences, with positive selection acting on OdsH at the speciation bottleneck that occurred in allopatry, while a different episode of positive selection acted on Iris subsequently. Interestingly, we find that Iris also separates the Zimbabwe strains from the cosmopolitan strains of D. melanogaster (Figure 9A phylogeny; unpublished data), consistent with known premating isolation between these populations [54].
Discussion
The evolutionary origin of viruses has long fascinated evolutionary biologists. Are they remnants of an ancestral lifestyle, or more recent escapees from traditional genomes [55]? The env genes of retroviruses are an important key to unlocking this conundrum; as first suggested by Howard Temin [56], their acquisition is the single event that allows previously genome-bound retrotransposons to adopt an infectious lifestyle. The genes that confer this ability appear to have been very desirable for eukaryotic genomes. In particular, the syncytin genes have been acquired in two mammalian lineages, while Iris-like genes have been acquired in two insect lineages. However, there are significant differences between the syncytin and Iris domestication events. First, the syncytin genes show a signature of purifying selection in primates, consistent with their domesticated role in placental function [15]. Iris, on the other hand, appears to be an active participant in an ongoing genetic conflict as evidenced by the signature of positive selection. Second, the syncytin gene has retained the same architecture of the ancestral retroviral env gene including the SU/TM furin cleavage site, since it still carries out the ancestral membrane fusion function [12,14]. Iris, on the other hand, has degenerated this cleavage site, suggesting that Iris's current function does not require membrane fusion. Third, while syncytin clearly derived from an endogenous retrovirus, the donor retroviruses appear to be extinct, especially in the human genome. However, the Kanga retroviruses appear to be active, which may greatly aid studies on this interesting domestication of a retroviral env in an organism with more facile genetics.
Are the selective pressures on Iris unique? We know of two other cases of proviral env genes domesticated for host defense: Fv4 and Rmcf. Neither has been investigated for selective constraint. However, in the case of both Fv4 and Rmcf [42,57], the mode of defense is by the domesticated env gene blocking the receptor required for retrovirus entry [58,59]. Under this scenario, unless the receptor is subject to positive selection, the domesticated gene does not have a “moving target” and is not expected to be subject to positive selection. Indeed, the defense function of Fv4 and Rmcf may involve the stable co-evolution of the receptor and the domesticated ligand. Iris, on the other hand, is subject to positive selection, suggesting that its mode of action is likely to be directly at a protein–protein interaction surface with its antagonist [46]. Thus, we predict that Iris action is likely to be distinct from the receptor blockade mechanism.
What genetic conflict could Iris be subject to? Previously, there has been one case of positive selection of a viral gene that was recruited as an inhibitor of subsequent infections. The Fv1 restriction factor that guards against murine retroviral infections is a “domesticated” gag gene from a lineage of retroviruses [11] that has been proposed to be subject to positive selection in murine genomes [41]. Based on our finding of positive selection, and the precedent of the Fv1 gene, we propose that Iris has been recruited as a host gene specifically to defend adults against recurrent invasions by retroviruses and baculoviruses, which share a homologous env. Two hypothetical scenarios by which this defense could be achieved are schematized in Figure 10.
Figure 10 Two Hypothetical Models to Explain Positive Selection of Iris
(A) Under the first model, Iris has been domesticated for a role other than host defense. As part of this housekeeping function, Iris proteins reside on the cell surface, where they can be recognized as receptors by viral envelope proteins. Variants of Iris that cannot be recognized by the viral envelopes have a selective advantage.
(B) A second model considers the possibility that Iris can act as a dominant negative agent that counteracts retroviral envelope trimers (red) from mediating infection. In this scenario, viruses encode for envelope trimers that can be cleaved into the SU ligand interaction and TM membrane fusion domains. In the absence of Iris, or if Iris lacks the specificity to bind the envelope trimers, the viral envelopes can mediate infection of the target cell. However, if the Iris protein (blue) can bind the viral envelopes and arrest the membrane fusion step, then the host defends against the viral infection. In this scenario, Iris directly acts as a host defense protein. Note that in both scenarios, Iris is predicted to be subject to positive selection (to decrease virus binding in the first model, and to increase virus binding in the second).
In the first model (Figure 10A), Iris is present on cell surfaces as part of a housekeeping function, as is the case for syncytin. But by virtue of its homology, it continues to act as a receptor for retroviruses. Under this scenario, the positive selection on Iris would cause it to avoid interacting with retroviral envelopes. Whether Iris has a housekeeping function can be directly tested with flies carrying mutations in the Iris gene. Under the second model (Figure 10B) Iris serves as a dominant negative inhibitor of retroviral trafficking. Since the Iris-encoded protein is expected to largely share the same architecture as the retroviral envelope proteins, it is expected to form multimers with the retroviral encoded envelope proteins. However, if the protein encoded by Iris is not cleaved, these may form multimers with retroviral envelopes that are incapable of mediating infection. In this scenario, the positive selection of Iris would act to improve recognition of retroviral envelope proteins to trap them in defective multimers (Figure 10B), while the latter would evolve away from this inhibitory interaction. We favor this second model because it provides a rational hypothesis for why the furin cleavage site has not been conserved. Under this model, we expect that Iris could defend against either horizontal transfers or germline transposition events. Germline tissues (ovaries and testes) are primarily where genome-bound retroelements need to transpose in order to increase their copy number within the genome. Gypsy-like retroviruses appear to infect the female oocyte [60], and recent studies indicate that this infection does not require the retroviral env genes [61–63]. However, these retroviruses have also been shown to be able to horizontally transfer to new hosts within the same species [64] and possibly to new species [65], and this activity depends on retroviral env activity.
Both models presented in Figure 10 are predicted to result in positive selection on Iris; genes subject to constantly antagonistic interactions (the “Red Queen” hypothesis [66]) are frequently subject to positive selection affecting the protein–protein interaction interface [46]. Our results raise the possibility that a number of retrovirus-derived “fossils” that can be found in many genomes, including our own [9], may represent new and old recruits in an ongoing battle for evolutionary supremacy. Such recruitments are easier to identify in genomes like Drosophila, where genes that are not under selection are quickly abraded [29], rather than in mammalian genomes, where pseudogenes can survive for tens of millions of years. In both cases, only detailed investigations of function or selective constraint can ascertain whether a retroviral remnant has been functionally retained, or is simply a paleontological relic of a past infection.
Materials and Methods
Drosophila strains.
Drosophila strains used in this study were obtained from the Drosophila Species Stock Center (Tucson, Arizona, United States), except for the Zimbabwe strains of D. melanogaster that were a gift from Y. Chen and W. Stephan.
PCR.
PCR was used to amplify the Iris coding region from Drosophila species using degenerate primers designed to Iris, CG4715degF: 5′- CTGGTGGACACCGAAACACCNTACYTNGG-3′, and to a conserved gene found downstream and in opposite orientation to Iris (CG4552)-primer CG4552degF: 5′- GCGACCTCATCACGTTYAARTAYGG-3′ (Figure 1A). This pair of primers enabled the amplification of the 3′ end of the Iris coding region and the design of species-specific primers. For all strains from D. melanogaster, and sibling species D. simulans, D. mauritiana, and D. sechellia, we employed specific primers CG4715eATG: 5′- AACGATCACCTCTACAAGCGAAAGATG-3′, and CG4715R2: 5′- GAAGACTGGTTCCGTATGGCCGC-3′ in forward and reverse orientations, respectively, to get the complete coding Iris sequence. In the case of the other Drosophila species, we employed a forward primer 500 base pairs upstream of CG715: 5′- CACTTCGACTGTTCTGAATGAACTGACG-3′ to obtain nearly the entire coding region, in conjunction with primers designed specifically to the 3′ end of the particular Iris gene from that species. Specific primers to D. pseudoobscura and D. ananassae were made based on the draft sequences of the genome from the two species. The A. gambiae sequence was directly obtained from the Anopheles genome sequence, while the A. aegypti sequence was reconstructed using synteny to Anopheles, from the database of trace sequences. Sequences of the Kanga and roo retroviruses were obtained from the ongoing genome sequencing efforts in 12 Drosophila species. Most products were directly sequenced using ABI Big-Dye sequencing. In cases where PCR products were too weak to be directly sequenced, they were cloned using TopoTA cloning kits (Invitrogen, Carlsbad, California, United States) and then sequenced using vector-specific primers: at least four separate clones were sequenced for each PCR product. All sequences obtained or annotated in this study have been deposited in GenBank (http://www.ncbi.nlm.nih.gov/Genbank).
RT-PCR.
RT-PCR analysis was carried out on pools of polyA D. melanogaster RNA that were a gift from S. Parkhurst (Fred Hutchinson Cancer Research Center) using primers CG4715eATG and CG4715R2 (described above) using the SuperScript One Step RT-PCR System from Invitrogen.
Northern analysis.
Northern analysis was also carried out using a blot containing the same pools of polyA RNA, using D. melanogaster Iris gene PCR fragment (CG4715eATG and CG4715R2 primers) as probe. For the tissue-specific RT-PCR analysis, individual flies were dissected for ovaries, testes, carcasses, and heads. RNA was isolated using the Qiagen RNeasy Kit (Valencia, California, United States) and treated with the DNase-Free kit from Ambion (Austin, Texas, United States) to remove trace amounts of DNA. RNA amounts were measured using a spectrophotometer. Roughly equal amounts of RNA were used as template in the individual RT-PCR reactions. As a loading control, and to rule out genomic DNA contamination, a separate RT-PCR was carried out to the Karyopherin α3 gene using primers 5′- CGTTGAGCTGAGGAAGAACAAGCG-3′ and 5′-GTGGCTGCACGACTCCGTGC-3′, which span an intron, allowing cDNA to be distinguished from genomic DNA. For the Iris-B genes from D. prostipennis and D. lutescens, RNA was isolated from pooled adult male and female flies. RT-PCR was used to validate the intron positions.
Bioinformatic analyses.
We used PSI-BLAST analyses to obtain all homologous sequences to Iris (CG4715) using Iris, gypsy env, and Autographa californica nucleopolyhedrovirus orf23 genes as search seeds, allowing the search up to three iterations. The various homologous sequences obtained by PSI-BLAST and our PCR results were aligned using CLUSTALX [67], eliminating all domains that were not unambiguously aligned in order to get a conservative alignment. Alignments were presented using the MacBoxshade program (written by M. Baron). We then used this alignment to obtain phylogenetic trees using the PAUP* suite of programs [68], employing both neighbor-joining, maximum likelihood, and maximum parsimony (heuristic) searches, followed by bootstrap analyses. The Kanga retroviral sequences used in the analysis presented in Figure 3 represent best match hits (using Iris as a query) in the individual genomes. Each hit to a retroviral env gene was used to analyze the genomic region containing the retrovirus for additional open reading frames, including the gag- and pol-like genes (used in Figure 3B). We used the SignalP program (version 3.0) to identify putative signal peptide cleavage sites [69].
Population genetic analyses.
Population genetic analyses were carried out using the DNASP program [70]. We used the program to carry out various tests for positive selection, including the McDonald-Kreitman test [47]. dN/dS ratios were computed in a sliding window using the Kestimator package [43]. Given calculated transition: transversion ratios and G+C content at third positions of codons, 1,000 trials of simulating dN equal to dS were generated. Significant deviations from neutrality (dN/dS ~1) were evaluated by comparing the range of simulated dN values to actual dN [43].
Maximum likelihood analyses.
Maximum likelihood analyses were performed with Codeml in the PAML software package [44]. Global ω ratios for the tree (Figure 8A) were calculated by a free-ratio model, which allows ω to vary along different branches. To detect selection, multiple alignments were fitted to either the F3 ×4 or F61 models of codon frequencies. Log-likelihood ratios of the data were compared using different site-specific (NSsites) models: M7 (fit to a beta distribution, ω > 1 disallowed) to M8 (similar to model 7 but ω > 1 allowed). The likelihood ratio test is performed by taking the negative of twice the log-likelihood difference between the two models and comparing this to the χ2 distribution with degrees of freedom equal to the difference in the number of parameters between the models. In all cases, permitting sites to evolve under positive selection gave a much better fit to the data (Table 1). These analyses also identified certain amino acid residues with high posterior probabilities (greater than 0.95) of having evolved under positive selection under the naïve empirical Bayes (NEB) model (Table 1 and Figure 8B). A more conservative Bayes empirical Bayes (BEB) evaluation of whether codons had evolved under positive selection was also carried out. REL and FEL analyses were carried out using the online server at http://www.datamonkey.org [45,71].
Supporting Information
Accession Numbers
The GenBank (http://www.ncbi.nlm.nih.gov/Genbank) accession numbers in this paper are: Iris-A sequences (DQ 177366–DQ177418) and Iris-B sequences (DQ 185599–DQ 185602).
The FlyBase (http://flybase.bio.indiana.edu) accession numbers in this paper are: CG4715 orthologs (FBgn0031305) and Anopheles gambiae homolog of CG4715 (XP_314732).
We thank Susan Parkhurst and Miriam Rosenberg for the RNA pools and Northern blots, and Danielle Vermaak for help with the RNA isolation, RT-PCR analysis, helpful discussions, and constructive criticism. We also thank George Rohrmann for helpful comments, encouragement, and advice throughout this project. We thank Josh Bayes, Michael Emerman, Scott Goeke, Julie Kerns, Katie Peichel, Sara Sawyer, Danielle Vermaak and especially one anonymous reviewer for their helpful suggestions on the manuscript. This work was initially supported by a postdoctoral fellowship from the Helen Hay Whitney Foundation to HSM and funds from the Howard Hughes Medical Institute to SH. HSM is currently supported by startup funds from the Fred Hutchinson Cancer Research Center and by a Searle Scholar Award from the Kinship Foundation. HSM is an Alfred P. Sloan Fellow in Computational and Evolutionary Molecular Biology.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. HSM conceived and designed the experiments, performed the experiments, and analyzed the data. HSM and SH wrote the paper.
Abbreviations
BEBBayes empirical Bayes
dNnumber of replacement changes per site
dSnumber of synonymous changes per site
env
envelope
FELfixed effects likelihood
NEBnaïve empirical Bayes
RELrandom effects likelihood
SAsplice acceptor
SDsplice donor
==== Refs
References
Nakamura TM Morin GB Chapman KB Weinrich SL Andrews WH 1997 Telomerase catalytic subunit homologs from fission yeast and human Science 277 955 959 9252327
Eickbush TH 1997 Telomerase and retrotransposons: Which came first? Science 277 911 912 9281073
Pardue ML DeBaryshe PG 2003 Retrotransposons provide an evolutionarily robust non-telomerase mechanism to maintain telomeres Annu Rev Genet 37 485 511 14616071
Levis RW Ganesan R Houtchens K Tolar LA Sheen FM 1993 Transposons in place of telomeric repeats at a Drosophila telomere Cell 75 1083 1093 8261510
Agrawal A Eastman QM Schatz DG 1998 Transposition mediated by RAG1 and RAG2 and its implications for the evolution of the immune system Nature 394 744 751 9723614
Melek M Gellert M van Gent DC 1998 Rejoining of DNA by the RAG1 and RAG2 proteins Science 280 301 303 9535663
Smit AF Riggs AD 1996 Tiggers and DNA transposon fossils in the human genome Proc Natl Acad Sci U S A 93 1443 1448 8643651
Smit AF 1999 Interspersed repeats and other mementos of transposable elements in mammalian genomes Curr Opin Genet Dev 9 657 663 10607616
Lander ES Linton LM Birren B Nusbaum C Zody MC 2001 Initial sequencing and analysis of the human genome Nature 409 860 921 11237011
Volff J Korting C Schartl M 2001 Ty3/Gypsy retrotransposon fossils in mammalian genomes: Did they evolve into new cellular functions? Mol Biol Evol 18 266 270 11158386
Best S Le Tissier P Towers G Stoye JP 1996 Positional cloning of the mouse retrovirus restriction gene Fv1
Nature 382 826 829 8752279
Mi S Lee X Li X Veldman GM Finnerty H 2000 Syncytin is a captive retroviral envelope protein involved in human placental morphogenesis Nature 403 785 789 10693809
Kim HS Takenaka O Crow TJ 1999 Isolation and phylogeny of endogenous retrovirus sequences belonging to the HERV-W family in primates J Gen Virol 80 2613 2619 10573154
Mallet F Bouton O Prudhomme S Cheynet V Oriol G 2004 The endogenous retroviral locus ERVWE1 is a bona fide gene involved in hominoid placental physiology Proc Natl Acad Sci U S A 101 1731 1736 14757826
Bonnaud B Bouton O Oriol G Cheynet V Duret L 2004 Evidence of selection on the domesticated ERVWE1 env retroviral element involved in placentation Mol Biol Evol 21 1895 1901 15254254
Frendo JL Olivier D Cheynet V Blond JL Bouton O 2003 Direct involvement of HERV-W Env glycoprotein in human trophoblast cell fusion and differentiation Mol Cell Biol 23 3566 3574 12724415
Blaise S de Parseval N Benit L Heidmann T 2003 Genomewide screening for fusogenic human endogenous retrovirus envelopes identifies syncytin 2, a gene conserved on primate evolution Proc Natl Acad Sci U S A 100 13013 13018 14557543
Dupressoir A Marceau G Vernochet C Benit L Kanellopoulos C 2005
Syncytin-A and syncytin-B, two fusogenic placenta-specific murine envelope genes of retroviral origin conserved in Muridae
Proc Natl Acad Sci U S A 102 725 730 15644441
Frame IG Cutfield JF Poulter RT 2001 New BEL-like LTR-retrotransposons in Fugu rubripes, Caenorhabditis elegans, and Drosophila melanogaster
Gene 263 219 230 11223261
Malik HS Henikoff S Eickbush TH 2000 Poised for contagion: Evolutionary origins of the infectious abilities of invertebrate retroviruses Genome Res 10 1307 1318 10984449
Eickbush TH Malik HS 2002 Origins and evolution of retrotransposons Craig NL Craigie R Gellert M Lambowitz AM Mobile DNA II Washington, D.C. ASM Press 1111 1146
Rohrmann GF Karplus PA 2001 Relatedness of baculovirus and gypsy retrotransposon envelope proteins BMC Evol Biol 1 1 11244578
Pearson MN Groten C Rohrmann GF 2000 Identification of the lymantria dispar nucleopolyhedrovirus envelope fusion protein provides evidence for a phylogenetic division of the Baculoviridae
J Virol 74 6126 6131 10846096
Pearson MN Rohrmann GF 2004 Conservation of a proteinase cleavage site between an insect retrovirus (gypsy) Env protein and a baculovirus envelope fusion protein Virology 322 61 68 15063117
Pearson MN Russell RL Rohrmann GF 2002 Functional analysis of a conserved region of the baculovirus envelope fusion protein, LD130 Virology 304 81 88 12490405
Adams MD Celniker SE Holt RA Evans CA Gocayne JD 2000 The genome sequence of Drosophila melanogaster
Science 287 2185 2195 10731132
Kaminker JS Bergman CM Kronmiller B Carlson J Svirskas R 2002 The transposable elements of the Drosophila melanogaster euchromatin: A genomics perspective Genome Biol 3 DOI: 10.1186/gb-2002–3–12-research0084
Bowen NJ McDonald JF 2001
Drosophila euchromatic LTR retrotransposons are much younger than the host species in which they reside Genome Res 11 1527 1540 11544196
Petrov DA Lozovskaya ER Hartl DL 1996 High intrinsic rate of DNA loss in Drosophila
Nature 384 346 349 8934517
Lung O Blissard GW 2005 A cellular Drosophila melanogaster protein with similarity to baculovirus F envelope fusion proteins J Virol 79 7979 7989 15956544
Russo CA Takezaki N Nei M 1995 Molecular phylogeny and divergence times of Drosophilia species Mol Biol Evol 12 391 404 7739381
Yang Y Zhang YP Qian YH Zeng QT 2004 Phylogenetic relationships of Drosophila melanogaster species group deduced from spacer regions of histone gene H2A-H2B
Mol Phylogenet Evol 30 336 343 14715225
Felsenstein J 1978 Cases in which parsimony or compatibility methods will be positively misleading Syst Zool 27 401 410
Milne AA Shepard EH 2001 The complete tales and poems of Winnie-The-Pooh New York Dutton Books 557 p.
Meyerowitz EM Hogness DS 1982 Molecular organization of a Drosophila puff site that responds to ecdysone Cell 28 165 176 6279311
Scherer G Tschudi C Perera J Delius H Pirrotta V 1982
B104, a new dispersed repeated gene family in Drosophila melanogaster and its analogies with retroviruses J Mol Biol 157 435 451 6181263
Parisi M Nuttall R Edwards P Minor J Naiman D 2004 A survey of ovary-, testis-, and soma-biased gene expression in Drosophila melanogaster adults Genome Biol 5 R40 15186491
Bailey TL Gribskov M 1998 Methods and statistics for combining motif match scores J Comput Biol 5 211 221 9672829
Pearson MN Rohrmann GF 2002 Transfer, incorporation, and substitution of envelope fusion proteins among members of the Baculoviridae, Orthomyxoviridae, and Metaviridae (insect retrovirus) families J Virol 76 5301 5304 11991958
Nurminsky DI Nurminskaya MV De Aguiar D Hartl DL 1998 Selective sweep of a newly evolved sperm-specific gene in Drosophila
Nature 396 572 575 9859991
Qi CF Bonhomme F Buckler-White A Buckler C Orth A 1998 Molecular phylogeny of Fv1
Mamm Genome 9 1049 1055 9880676
Ikeda H Laigret F Martin MA Repaske R 1985 Characterization of a molecularly cloned retroviral sequence associated with Fv-4 resistance J Virol 55 768 777 2991595
Comeron JM 1999 K-Estimator: Calculation of the number of nucleotide substitutions per site and the confidence intervals Bioinformatics 15 763 764 10498777
Yang Z Wong WS Nielsen R 2005 Bayes empirical bayes inference of amino acid sites under positive selection Mol Biol Evol 22 1107 1118 15689528
Pond SL Frost SD 2005 Datamonkey: Rapid detection of selective pressure on individual sites of codon alignments Bioinformatics 21 2531 2533 15713735
Sawyer SL Wu LI Emerman M Malik HS 2005 Positive selection of primate TRIM5α identifies a critical species-specific retroviral restriction domain Proc Natl Acad Sci U S A 102 2832 2837 15689398
McDonald JH Kreitman M 1991 Adaptive protein evolution at the Adh locus in Drosophila
Nature 351 652 654 1904993
Jordan IK Matyunina LV McDonald JF 1999 Evidence for the recent horizontal transfer of long terminal repeat retrotransposon Proc Natl Acad Sci U S A 96 12621 12625 10535972
Daniels SB Peterson KR Strausbaugh LD Kidwell MG Chovnick A 1990 Evidence for horizontal transmission of the P transposable element between Drosophila species Genetics 124 339 355 2155157
Sanchez-Gracia A Maside X Charlesworth B 2005 High rate of horizontal transfer of transposable elements in Drosophila
Trends Genet 21 200 203 15797612
Kliman RM Andolfatto P Coyne JA Depaulis F Kreitman M 2000 The population genetics of the origin and divergence of the Drosophila simulans complex species Genetics 156 1913 1931 11102384
Ting CT Tsaur SC Wu CI 2000 The phylogeny of closely related species as revealed by the genealogy of a speciation gene, Odysseus
Proc Natl Acad Sci U S A 97 5313 5316 10779562
Ting CT Tsaur SC Wu ML Wu CI 1998 A rapidly evolving homeobox at the site of a hybrid sterility gene Science 282 1501 1504 9822383
Wu CI Hollocher H Begun DJ Aquadro CF Xu Y 1995 Sexual isolation in Drosophila melanogaster: A possible case of incipient speciation Proc Natl Acad Sci U S A 92 2519 2523 7708677
Campbell A 2001 The origins and evolution of viruses Trends Microbiol 9 61
Temin HM 1980 Origin of retroviruses from cellular moveable genetic elements Cell 21 599 600 6254661
Jung YT Lyu MS Buckler-White A Kozak CA 2002 Characterization of a polytropic murine leukemia virus proviral sequence associated with the virus resistance gene Rmcf of DBA/2 mice J Virol 76 8218 8224 12134027
Ikeda H Sugimura H 1989
Fv-4 resistance gene: A truncated endogenous murine leukemia virus with ecotropic interference properties J Virol 63 5405 5412 2555565
Goff SP 2004 Retrovirus restriction factors Mol Cell 16 849 859 15610729
Song SU Kurkulos M Boeke JD Corces VG 1997 Infection of the germ line by retroviral particles produced in the follicle cells: A possible mechanism for the mobilization of the gypsy retroelement of Drosophila Development 124 2789 2798 9226450
Leblanc P Desset S Giorgi F Taddei AR Fausto AM 2000 Life cycle of an endogenous retrovirus, ZAM, in Drosophila melanogaster
J Virol 74 10658 10669 11044110
Chalvet F Teysset L Terzian C Prud''homme N Santamaria P 1999 Proviral amplification of the Gypsy endogenous retrovirus of Drosophila melanogaster involves env -independent invasion of the female germline Embo J 18 2659 2669 10228177
Pelisson A Mejlumian L Robert V Terzian C Bucheton A 2002
Drosophila germline invasion by the endogenous retrovirus gypsy: Involvement of the viral env gene Insect Biochem Mol Biol 32 1249 1256 12225916
Song SU Gerasimova T Kurkulos M Boeke JD Corces VG 1994 An env-like protein encoded by a Drosophila retroelement: Evidence that gypsy is an infectious retrovirus Genes Dev 8 2046 2057 7958877
Syomin BV Fedorova LI Surkov SA Ilyin YV 2001 The endogenous Drosophila melanogaster retrovirus gypsy can propagate in Drosophila hydei cells Mol Gen Genet 264 588 594 11212913
Van Valen L 1973 A new evolutionary law Evolutionary Theory 1 1 30
Thompson JD Gibson TJ Plewniak F Jeanmougin F Higgins DG 1997 The CLUSTAL_X windows interface: Flexible strategies for multiple sequence alignment aided by quality analysis tools Nucleic Acids Res 25 4876 4882 9396791
Swofford DL 2000 PAUP*: Phylogenetic Analysis Using Parsimony (*and Other Methods), 4th edition [computer program] Sunderland (Massachusetts) Sinauer
Bendtsen JD Nielsen H von Heijne G Brunak S 2004 Improved prediction of signal peptides: SignalP 3.0 J Mol Biol 340 783 795 15223320
Rozas J Sanchez-DelBarrio JC Messeguer X Rozas R 2003 DnaSP, DNA polymorphism analyses by the coalescent and other methods Bioinformatics 19 2496 2497 14668244
Kosakovsky Pond SL Frost SD 2005 Not so different after all: A comparison of methods for detecting amino acid sites under selection Mol Biol Evol 22 1208 1222 15703242
Kuzio J Pearson MN Harwood SH Funk CJ Evans JT 1999 Sequence and analysis of the genome of a baculovirus pathogenic for Lymantria dispar
Virology 253 17 34 9887315
Marlor RL Parkhurst SM Corces VG 1986 The Drosophila melanogaster gypsy transposable element encodes putative gene products homologous to retroviral proteins Mol Cell Biol 6 1129 1134 3023871
|
16244705
|
PMC1262188
|
CC BY
|
2021-01-05 08:00:57
|
no
|
PLoS Genet. 2005 Oct 21; 1(4):e44
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010044
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1624470610.1371/journal.pgen.001004705-PLGE-RA-0043R2plge-01-04-02Research ArticleGenetics/Comparative GenomicsGenetics/Genetics of DiseaseMammalsEvolutionary Comparison Provides Evidence for Pathogenicity of RMRP Mutations RMRP Mutations in Cartilage-Hair HypoplasiaBonafé Luisa 1*Dermitzakis Emmanouil T 23Unger Sheila 45Greenberg Cheryl R 6Campos-Xavier Belinda A 1Zankl Andreas 1Ucla Catherine 2Antonarakis Stylianos E 2Superti-Furga Andrea 14Reymond Alexandre 271 Division of Molecular Pediatrics, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland
2 Department of Genetic Medicine and Development, University of Geneva Medical School and University Hospitals of Geneva, Geneva, Switzerland
3 The Wellcome Trust Sanger Institute, Cambridge, United Kingdom
4 Center for Pediatrics and Adolescent Medicine, Freiburg University Hospital, Freiburg, Germany
5 Division of Clinical and Metabolic Genetics, Hospital for Sick Children, Toronto, Ontario, Canada
6 Metabolic Service, Section of Genetics and Metabolism, Health Sciences Centre, Winnipeg, Manitoba, Canada
7 Center for Integrative Genomics, University of Lausanne, Lausanne, Switzerland
Valle David EditorJohns Hopkins Institute, United States of America*To whom correspondence should be addressed. E-mail: [email protected] 2005 21 10 2005 1 4 e4715 3 2005 7 9 2005 Copyright: © 2005 Bonafé 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.Cartilage-hair hypoplasia (CHH) is a pleiotropic disease caused by recessive mutations in the RMRP gene that result in a wide spectrum of manifestations including short stature, sparse hair, metaphyseal dysplasia, anemia, immune deficiency, and increased incidence of cancer. Molecular diagnosis of CHH has implications for management, prognosis, follow-up, and genetic counseling of affected patients and their families. We report 20 novel mutations in 36 patients with CHH and describe the associated phenotypic spectrum. Given the high mutational heterogeneity (62 mutations reported to date), the high frequency of variations in the region (eight single nucleotide polymorphisms in and around RMRP), and the fact that RMRP is not translated into protein, prediction of mutation pathogenicity is difficult. We addressed this issue by a comparative genomic approach and aligned the genomic sequences of RMRP gene in the entire class of mammals. We found that putative pathogenic mutations are located in highly conserved nucleotides, whereas polymorphisms are located in non-conserved positions. We conclude that the abundance of variations in this small gene is remarkable and at odds with its high conservation through species; it is unclear whether these variations are caused by a high local mutation rate, a failure of repair mechanisms, or a relaxed selective pressure. The marked diversity of mutations in RMRP and the low homozygosity rate in our patient population indicate that CHH is more common than previously estimated, but may go unrecognized because of its variable clinical presentation. Thus, RMRP molecular testing may be indicated in individuals with isolated metaphyseal dysplasia, anemia, or immune dysregulation.
Synopsis
Cartilage-hair hypoplasia is a genetic condition named after two of its most conspicuous features, short bones and sparse hair, but it affects blood-forming tissues, immune system, and intestine. It is caused by sequence mutations in RMRP, a small gene that codes for a structural RNA component of an RNAse complex whose biological functions have been elusive so far. The small RMRP gene carries a surprisingly high number of sequence variations, and because its transcript is not translated into protein and its function in the cell is still unclear, distinction between harmless variants and disease-causing mutations (more than 60 have been found so far by the authors and others) is difficult. The authors have sequenced the RMRP gene in several species covering the whole class of mammals and found that the gene is remarkably conserved between species. Interestingly, mutations occurring in conserved (probably functionally important) regions of the gene appear to be disease-producing, whereas those occurring in regions where evolution is more relaxed seem to be harmless variants. These results will help in counseling affected individuals and their families, and may lead to the discovery of the real function of this mysterious gene.
Citation:Bonafé L, Dermitzakis ET, Unger S, Greenberg CR, Campos-Xavier BA, et al. (2005) Evolutionary comparison provides evidence for pathogenicity of RMRP mutations. PLoS Genet 1(4): e47.
==== Body
Introduction
Cartilage-hair hypoplasia (CHH; MIM#250250) is a pleiotropic disease involving different organs and tissues. Its most frequent clinical manifestations are short-limb dwarfism with metaphyseal dysplasia, joint laxity, fine and sparse hair, anemia, and immune deficiency [1–4], but Hirschsprung disease, impaired spermatogenesis, and malignancies (particularly leukemias, lymphomas, and skin cancer) also occur at increased frequency [5–8].
Mutations in RMRP, a gene encoding an RNA component of the mitochondrial RNA processing ribonuclease (RNase MRP) [9], were found to be responsible for CHH [10,11]. Among the known functions of RNase MRP are pre-rRNA processing in the nucleolus and cleavage of RNA primers necessary for DNA replication in the mitochondrion; but it is likely that the RNA molecule coded by RMRP has other yet-unidentified functions [12–18]. Forty-two different mutations have been reported so far in 91 Finnish, 47 non-Finnish Caucasian, and two Japanese families [11,19–22]. Given the clinical and radiographic variability of CHH, molecular diagnosis is important both for appropriate counseling and for allowing relevant preventive measures. However, because the RMRP RNA is not translated into a protein, it is difficult to evaluate the pathogenicity of a sequence variant. Moreover, the small RMRP gene (coding region of 265 base pairs [bp]) has two remarkable features: a very high density of single nucleotide polymorphisms (SNPs) [19,20,22] and an extensive series of allelic, putatively pathogenic variants. For this reason, molecular diagnosis and genetic counseling are particularly difficult.
We report here yet another series of 20 novel RMRP mutations in CHH; more importantly, we show that RMRP pathogenic mutations, but not SNPs, occur in nucleotides conserved in the entire mammalian class. The implications of these findings in connection with the wide clinical spectrum of CHH are discussed.
Results
Mutations and Polymorphisms in CHH Patients
We performed mutation analysis of the RMRP gene in 36 patients with suspected CHH. Clinical data are summarized in Table 1. Twelve patients had only skeletal manifestations of the disease during infancy and childhood, 11 patients had also severe immune deficiency, whereas 13 patients had an intermediate phenotype including skeletal dysplasia and at least one extra-skeletal manifestation: recurrent infections and/or subclinical immune deficiency, anemia and/or leucopenia, autoimmune manifestations, hair hypoplasia (defined as thin hair, often lightly colored, and rarely needing a haircut), or Hirschsprung disease.
Table 1 Clinical and Molecular Data of the 36 Studied CHH Patients
Thirty probands were found to be compound heterozygotes for putative pathogenic variants (Table 1), whereas the six remaining probands were true homozygotes for a known pathogenic mutation (see below). We identified 20 previously unreported, putatively pathogenic mutations; five of them are located in the promoter region and 15 in the transcribed region. Table 2 details all RMRP mutations reported to date and their localization, as well as the number of reported affected families.
Table 2
RMRP Variants: 20 New Putative Pathogenic Mutations and Review of Reported Mutations
Figure 1
RMRP Gene Product
The RNA molecule of the human RNase MRP complex is shown according to the model of Welting et al. [29]. Nucleotides affected by mutations (single nucleotide changes) are in red; polymorphisms and rare variants are in blue. Insertion/duplication/deletions are indicated as red (mutations) or blue (putative polymorphisms) arrows. Regions of base-pairing are indicated as gray boxes. The P3 domain (nucleotides 22–66) is shaded.
We identified five new insertions/duplications of 8 to 32 bp localized in the last 25 nucleotides of the promoter region 5′ to the transcription initiation site (Tables 1 and 2). All duplications/insertions in the promoter region result in an increased distance between the TATA box and the transcription initiation site. This type of mutation was always found in compound heterozygosity with a single nucleotide mutation in the transcribed region and never homozygous or in compound heterozygosity with another promoter duplication/insertion.
Among all mutations located within the transcribed region, six are duplication/insertions of 1–17 nucleotides, one is a 2-bp deletion, and the remaining variants are single nucleotide changes (Table 2). Mutations in the transcribed region are located throughout the RNA molecule (Figure 1; Table 2), with no preference in any of the different structural domains. Among the 39 single nucleotide mutations in the transcribed region, 12 are located at nucleotides not involved in base-pairing in the secondary structure of the RNA molecule (Figure 1; Table 2) [18,23,24]. None of the 20 new putative pathogenic mutations reported here were found in 100 non-Finnish Caucasian controls.
The different parental origin of mutations in compound heterozygote patients was confirmed in 23 out of 24 families for which parental DNA was available. In family 27 (Table 1), mutation g.248C>T was carried by the father, whereas the g.127G>A change was not present in maternal DNA or in that of two unaffected sons. Microsatellite marker analysis of the region around RMRP in this family (Figure S1) showed that the same maternal allele is carried by the affected child and by one of the healthy siblings, suggesting that g.127G>A is a de novo mutation in this patient, occurring on the maternal allele. However, maternal germinal mosaicism could not be excluded.
Unlike in the Finnish/Amish population, only five patients (24, 25, 31, 32, and 36 in Table 1) were homozygous for the g.70A>G CHH mutation. One patient (34 in Table 1) was homozygous for another known pathogenic mutation (g.261C>T). No parental consanguinity was reported in our sample.
We have previously identified seven frequent SNPs within and around RMRP and four rare variants (Table 3) [19]. In a Japanese control population (130 alleles), four other rare sequence changes (g.36T>G, g.55–56insC, g.162C>T, and g.172C>T) have been reported [22]. These variants were found in controls only in heterozygosity and therefore their neutrality is not proven. One more polymorphism has been identified in the present study: g.127G>C. This variant was found in 1% of non-Finnish Caucasian controls (two chromosomes out of 200) and observed in one control individual in proven compound heterozygosity with a known pathogenic mutation (g.70A>G) (trans phase confirmed by parental DNA testing).
Table 3
RMRP Polymorphisms and Other Rare Variants
Sequence analysis of 100 controls did not detect any putative pathogenic mutation (newly identified and previously described).
Comparative Genomic Analysis of RMRP
To address the significance of sequence changes identified in humans, we compared the human RMRP gene against a set of orthologous sequences from mammalian species covering Classes I, II, III, and IV of the Eutherians [25], the Metatherians, and the Monotremes. We successfully PCR amplified and sequenced the RMRP transcript regions from genomic DNA of green monkey (Cercopithecus aethiops), ring-tailed lemur (Lemur catta), brush-tailed porcupine (Atherurus africanus), rabbit (Oryctolagus cuniculus), pig (Sus scrofa), cat (Felis catus), white-toothed shrew (Crocidura russula), nine-banded armadillo (Dasypus novemcinctus), African elephant (Loxodonta africana), and tammar wallaby (Macropus eugenii) and aligned them to human and mouse RMRP (Figure 2). The promoter region of the gene was amplified from white-toothed shrew (Cr. russula), brush-tailed porcupine (A. africanus), rabbit (Ory. cuniculus), green monkey (Ce. aethiops), ring-tailed lemur (Le. catta), cow (Bos taurus), and African elephant (Lo. africana) (Figure 3).
Figure 2 Multiple Sequence Alignment of the RMRP Transcribed Region
The sequence alignment of 12 mammalian species (11 placental and 1 marsupial) of the RMRP transcribed region (from nucleotide g.28 to g.247 of the human sequence) is shown. Putative pathogenic mutations are indicated as red boxes (single nucleotide changes) or red arrows (insertion/duplications/deletions); polymorphisms and rare variants are indicated as blue boxes (single nucleotide changes) or blue arrows (insertion/duplications/deletions). Conservation of nucleotides was analyzed for the single nucleotide substitutions (putative mutations or polymorphisms) included in the alignment interval (from g.28 to g.247). Positions were considered as conserved if 11 out of 12 species had the same nucleotide.
Figure 3 Multiple Sequence Alignment of the RMRP Promoter Region
The sequence alignment of the RMRP promoter region of nine mammalian species (from nucleotide g.−80 to the transcription initiation site of the human sequence) is shown. Polymorphisms and rare variants are indicated as blue boxes (single nucleotide changes); red arrows indicate pathogenic insertions, and red lines indicate pathogenic duplications and triplications. Conservation of nucleotides was analyzed for the single nucleotide substitutions (putative polymorphisms) included in the alignment interval (from g.−80 to g.−1). Positions were considered as conserved if eight out of nine species had the same nucleotide.
The multiple species alignment of the RMRP transcribed region showed that 23 putative pathogenic mutations out of 29 are located in strongly conserved nucleotides (same nucleotide in at least 11 out of 12 species) (Figure 2). In contrast, the three polymorphisms located within the coding region (positions g.127, g.156, and g.177) are located in non-conserved nucleotides. Similarly, multiple sequence alignment of the promoter region from −80 to the transcription initiation site showed that only one out of four polymorphisms in this region is located in a conserved position (Figure 3). Three out of four examined rare variants also affect non-conserved nucleotides (g.−24, g.36, and g.162) (Figures 2 and 3).
Discussion
RMRP Structure and Mutations
The RMRP gene encodes a non-translated RNA that forms a cage-shaped structure in the core of the ribonuclease enzymatic complex. Although initially isolated in the mitochondrion, the enzyme is predominantly localized in the nucleolus [26]. Different protein subunits bind to specific domains of the RNA secondary structure [18,24,27–29]. The most recent model of protein–RNA interactions of RNase MRP [29] shows that some regions of the RNA molecule are critical for protein binding (Figure 1). The P3 domain (nucleotides 22–66) is involved in direct binding of several protein subunits (hPop1, hPop4, Rpp20, and Rpp25) [24,27,29]. This region contains also the nucleolar localization signal region (nucleotides 23–62), previously reported as mutation free [20] except for one insertion (g.57insTTCCGCCT), which did not change the signal sequence. We observed three novel mutations (g.35C>T, g.40G>A, and g.45_53dupTGTTCCTCC) in the nucleolar localization signal region. In the P3 domain mutation, g.63C>T is recurrent and was observed in five unrelated families in this study: four of them are of Dutch origin and two of them share the same [g.63C>T] + [g.70A>G] genotype (Table 1); furthermore, in all four patients, g.63C>T segregates with the Finnish haplotype (data not shown), suggesting a possible common origin of the mutation.
Nucleotide 70 maps just after the end of the P3 domain in a conserved region, which can form, in yeast, a 9-bp duplex with 5.8S rRNA [17], suggesting a possible role in small rRNA processing. The Finnish/Amish founder mutation, g.70A>G, was found in 18 out of 72 chromosomes in our series of patients. This frequency (26%) is far lower than the 48% reported for non-Finnish populations [20]. It has been shown that this mutation results from enrichment in an isolate population and is thus not a mutational hotspot [30].
Another recurrent mutation in our series is g.4C >T, identified in five of the studied families; it is not associated with a specific haplotype (data not shown), suggesting a multiple origin of the mutation. The other mutations are located throughout the RNA molecule, and no clear genotype-phenotype correlation could be recognized. Phenotypes were different in patients with the same genotype, as previously reported for g.70A>G homozygote patients in several Finnish studies [5,11].
Compound heterozygosity for a duplication/insertion in the promoter region and a single nucleotide change in the transcribed region occurs frequently: ten out of 36 families in our sample, and seven out of 44 families in previous reports [20]. Duplication/insertions in the promoter region have never been reported in the homozygous state or in compound heterozygosity with another mutation located within the promoter region, either in our series or in previous works [11,19,20,22]. Because RMRP is transcribed by RNA polymerase III [18], the distance between the TATA box and the transcription initiation site is critical for efficient transcription [31]. It has been found that duplications/insertions in this region cause reduced quantity of transcript [11]; the fact that no patient has been found to be homozygous or compound heterozygous for such mutations may indicate that the mutations are phenotypically lethal. Thus, it appears that having drastically reduced amounts of RMRP RNA, albeit of normal structure, is more deleterious than having normal amounts of RMRP RNA containing subtle structural changes.
Both the abundance of SNPs and the number and diversity of pathogenic mutations in RMRP are remarkable and at odds with the high conservation of the RMRP gene throughout species. It is unclear at present whether the RMRP gene is affected by a particularly high incidence of point mutations, or whether there is reduced efficiency of DNA repairing mechanisms specifically for the RMRP gene.
In the background of this mutational heterogeneity, we observed in our patient population a rarity of parental consanguinity and a low incidence of homozygosity for any given mutation. The known inverse relationship between parental consanguinity rate and frequency of a recessive disease suggests that CHH is relatively common among recessive diseases. Precise data on the incidence of CHH outside Finland are lacking, and direct ascertainment is difficult because of the variable clinical expression.
Comparative Genomics
In molecular diagnosis of CHH, the definition of criteria of pathogenicity for any given mutation is a key issue. Because RMRP is not translated into protein, any prediction of the consequences of a nucleotide change is difficult. Correct segregation of mutations in the families and absence of mutations in controls are two commonly used criteria. Other suggested criteria are evolutionary conservation of nucleotides involved in CHH-associated mutations and involvement in base-pairing in the secondary structure model [19]. Several mutations that co-segregate with the disease in multiple unrelated families with CHH do not map in base-pairing positions (Figure 1, Table 2). The best example is g.70A>G, the most frequent pathogenic allele worldwide, but 12 other putatively pathogenic mutations are also located in non-paired nucleotides of the RNA molecule. Thus, this criterion appears to be of limited usefulness in defining pathogenicity. Conservation through species, instead, is strongly associated with putative pathogenic mutations. Previously reported multiple sequence alignments [23,32] took into account organisms that were phylogenetically quite distant, and did not examine the promoter region of the gene. We aligned the promoter and transcribed region of RMRP from multiple placental and one marsupial mammal. Our analysis shows that putative pathogenic CHH-associated mutations involve strongly conserved nucleotides (23 out of 29). Furthermore, positions corresponding to SNPs are mainly non-conserved, both in the promoter and in the transcribed region (nine out of 11, including rare variants).
We noticed two different changes at position 127, the mutation g.127G>A, associated with CHH in two different families, and the polymorphism g.127G>C, found in 1% of non-Finnish Caucasian controls. This nucleotide is only partially conserved through species, being 127C in Xenopus and 127A in yeast [23,32]. In our alignment, armadillo has 127A whereas all the other placentals and wallaby have 127G. Interestingly, we could document a de novo g.127G>A transition in family 27 (Figure S1), indicating a possible mutational hot spot at this position. Similarly at nucleotide position 182, there are three different RMRP mutations in CHH patients, g.182G>T, g.182G>C, and g.182G>A; the latter was also observed de novo in a Japanese family [22].
Among all RMRP variants found in controls (Table 3), eight single nucleotide changes are certainly neutral polymorphisms; the other nine variants were found in controls by us and others only in heterozygosity and are therefore not certainly neutral. Interestingly, the majority of rare variants affect non-conserved nucleotides (Figures 2 and 3), suggesting that they represent rare polymorphisms rather than pathogenic mutations. In conclusion, our observations further validate the role of phylogenetic footprinting in assessing pathogenicity.
CHH Phenotypic Spectrum
CHH is both pleiotropic, with potential involvement of different organs, and variable in its clinical severity. Skeletal manifestations are characterized by metaphyseal dysplasia of variable degree; extra-skeletal manifestations are sparse hair, anemia, leucopenia, immune deficiency, Hirschsprung disease, and other possible organ involvement (endocrine, autoimmune, and digestive) including malignancies. All patients in our sample were ascertained because of short stature and some degree of skeletal changes. We reviewed personally all available radiographs and selected for those patients who had the typical pattern of phalangeal and metaphyseal striations, the degree of which can be variable. We then separated patients in our sample into three arbitrary groups: approximately 33% (12 out of 36) with short stature and skeletal manifestations only; approximately 30% (11 out of 36) with short stature and severe immune deficiency and/or anemia; and approximately 36% (13 out of 36) with skeletal dysplasia and at least one extra-skeletal manifestation, whose immunological and hematological abnormalities were subclinical or caused only mild clinical symptoms. The severity of skeletal manifestations did not correlate with the severity of extra-skeletal disturbances. Indeed, two patients with severe immune deficiency had short stature but only very mild changes on skeletal survey (Table 1). Thus, the spectrum of CHH phenotypes is wide, ranging from isolated metaphyseal dysplasia to severe immune deficiency associated with short stature of variable degree. If at one end of the spectrum there are several patients with only skeletal manifestations of the disorder (skeletal variant) [19,33], we might also expect, at the opposite end of the spectrum, CHH patients with anemia/leucopenia/immune deficiency and normal stature. Although all patients we had studied were short at the time of diagnosis, not all of them were short at birth, indicating that prenatal growth can be normal. Normal birth length was documented in four patients who later developed clear skeletal signs of the disease and no immunologic abnormalities.
The presence of hair hypoplasia is often associated with other extra-skeletal manifestations: Seven patients out of nine with this characteristic had also immunological anomalies and/or Hirschsprung disease. The absence of hair changes and of hematological/immunological involvement should not exclude CHH from the differential diagnosis of a short child with metaphyseal dysplasia. Short fingers and cone-shaped phalangeal epiphyses are a very sensitive marker for CHH; our data indicate that molecular testing for RMRP mutations should be considered in all unclear metaphyseal dysplasias. Long-term follow-up will clarify whether patients with RMRP mutations, but only skeletal signs at diagnosis, are at increased risk of malignancy like those with immune deficiency. Indeed, patient 36 (Table 1) had short stature with skeletal dysplasia, but was otherwise healthy until developing non-Hodgkin lymphoma in the fourth decade of life.
Hirschsprung disease was diagnosed in four patients out of 36 (11%), a frequency similar to that reported for Finnish patients (9%) [7]. As previously suggested [7], Hirschsprung disease is associated with severely affected patients (Table 1).
Three patients in our series were reported to have growth hormone (GH) deficiency. Previous studies showed that low IGF-1 levels correlate, in CHH patients, with reduced erythropoiesis [34], but true GH deficiency, defined as insufficient response to GHRH (growth hormone-releasing hormone) stimulation, was never demonstrated [35]; moreover, GH treatment has proven to be of little value in CHH patients [36], as well as in other skeletal dysplasias [37,38]. Because the biochemical demonstration of GH deficiency is difficult and dysregulation of the GH-IGF axis has previously been observed in CHH patients, we would consider these data with caution. In addition, other subclinical endocrine abnormalities seem to occur in CHH patients: Abnormal thyroid function was found in three patients in this study.
The association of hypobetalipoproteinemia and CHH (patient 10 in Table 1) has not been previously described. As clinical symptoms of hypobetalipoproteinemia are often not evident in childhood, it is possible that this abnormality is not frequently tested in CHH patients. However, it is unclear whether hypobetalipoproteinemia is primary in this patient, or secondary to subclinical malabsorption. Some clinical manifestations in this CHH family overlap with Shwachman-Diamond syndrome (SDS; MIM#260400) [39], a disease characterized by pancreatic exocrine insufficiency, hematological dysfunction, and skeletal abnormalities, which is in the differential diagnosis of CHH. Interestingly, the gene causing SDS, SDBS, plays a role in RNA metabolism [40], suggesting a possible pathogenetic mechanism common to CHH and SDS. Intestinal malabsorption has been described in CHH [41–43], also in association with celiac disease [44]. Although some instances may be explained by opportunistic bowel infections, it is possible that enteropathy is another rare manifestation of CHH pleiotropism. However this association in single cases may also be caused by chance alone.
Autoimmune disorders were present in four patients in our series. Previous reports suggested an autoimmune mechanism underlying the anemia [45,46] and neutropenia [47] in some CHH patients. There is no evidence of autoimmune anemia in most CHH patients, but it is possible that immune dysregulation favors the development of autoimmunity against other organs and tissues in CHH.
Materials and Methods
Sample population.
A total of 59 patients referred to our center for short stature and skeletal changes were included in this study; of these, 40 patients showed the typical pattern of phalangeal epiphyseal changes, metaphyseal cupping and striations of CHH (as judged by two or more of the radiographic experts among us: LB, SU, AZ, and ASF), albeit to a variable degree; and 19 had radiological changes that were considered not typical but still compatible with CHH. All 59 patients were tested for RMRP mutations by sequencing of the entire RMRP coding sequence as well as promoter region. Twenty-three patients tested negative (all 19 patients in the non-typical group, plus four patients considered to have radiographic changes typical for CHH), and 36 patients had mutations in RMRP. The patient material was referred to us for non-commercial diagnostic help and accepted under the research terms specified in the European Skeletal Dysplasia Network program (http://www.esdn.org/) (for European patients) or analogously, for non-European subjects. Appropriate informed consent was obtained by the physicians in charge. The clinical data and the radiographic features of each patient were evaluated before molecular testing. The clinical data of the 36 unrelated patients with CHH are summarized in Table 1.
Parental DNA was studied in 24 families. For the remaining 12 patients (patients 2, 7, 8 [adopted], 11, 21, 24 [homozygote], 25 [homozygote], 26, 31 [homozygote], 32 [homozygote], 35, and 36 [homozygote]), parental DNA was not available.
Molecular analysis of RMRP gene.
Genomic DNA was extracted from blood leukocytes according to standard methods; in one patient (patient 5) genomic DNA was extracted from fibroblasts because bone marrow transplantation had been performed before the study.
Genomic DNA of 100 non-Finnish Caucasian control individuals (German, Italian, British, French, Swiss, Australian, and Canadian) has been studied to test for the presence or absence of all variants, including new putative pathogenic mutations, previously reported mutations, and putative polymorphisms.
The coding and promoter regions of the RMRP gene were amplified from genomic DNA by PCR and analyzed by fluorescent bidirectional direct sequencing as previously described [19]. Each variant was confirmed by reamplification and resequencing. Trans phase of mutations in 24 compound heterozygote patients was verified by confirming the presence of mutations in parental DNA .
Microsatellite marker analysis of the region around RMRP in family 27 was performed with the ABI Prism Linkage Mapping Set version 2.5 kit and protocol, in an ABI Prism 3100Avant automatic sequencer (Applied Biosystems, Foster City, California, United States).
Comparative genomics.
Tissues or genomic DNA from the following species were obtained from different sources (for details see Table S3 of Dermitzakis et al. [48]): S. scrofa, F. catus, My. myotis, Cr. russula, A. africanus, Ory. cuniculus, Ce. aethiops, Le. catta, D. novemcinctus, Lo. africana, Ma. eugenii, Orn. anatinus, B. taurus. Pairs of oligonucleotides were designed on human DNA sequences in highly conserved regions between human and mouse to amplify either the promoter (5′-GCCACCAACTTTCTCACC-3′ and 5′-GGGACTTTCCCCTAGGC-3′) or the transcribed region (5′-GGCCTGTATCCTAGGCTACA-3′ and 5′-AGCCGCGCTGAGAATGAG-3′) of the RMRP gene.
Conservation of single nucleotide substitutions was analyzed on the multiple sequence alignments of the transcribed region (Figure 2) and the promoter region from g.–80 to g.−1 (Figure 3), based on the number of species harboring the same nucleotide. We considered as conserved the positions where at least 11 out of 12 species (in the transcribed region) or eight out of nine species (in the promoter region) had the same nucleotide.
Supporting Information
Figure S1 Microsatellite Marker Analysis of the Region around RMRP in Family 27
The same maternal haplotype is carried by the affected child and by one of the healthy siblings. Neither the mother nor the unaffected sibling carry g.127G>A, suggesting that this mutation occurred de novo in the affected individual (see also Table 1).
(378 KB PDF)
Click here for additional data file.
Accession Numbers
The GenBank (http://www.ncbi.nlm.nih.gov/Genbank) accession numbers for the PCR-amplified RMRP sequences are CC935687, CC935890, CC935995, CC936164, CC936319, CC936439, CC936561, CC936652, and CC936711; the GenBank accession number for the RMRP gene is M29916.
We thank all physicians providing patients' samples and data: Prof. Maja DiRocco, Istituto Gaslini, Genova (Italy); Prof. U. Ramenghi, University of Turin (Italy); PD Dr. C. Aebi, University Children's Hospital, Bern (Switzerland); Prof. M. LeMerrer, Hôpital Enfants Malades, Paris (France); Prof. G. Camera, Ospedale Galliera, Genova (Italy); Prof. J. Kunze and Dr. L. Neumann, Institute for Human Genetics Charité, Berlin (Germany); Dr. K. Devriendt, University of Leuven (Belgium); Prof. B. Zabel, University Children's Hospital, Mainz (Germany); Dr. L. Kobrynski, Emory University, Atlanta, Georgia (United States); Dr. B. Prager, Dresden (Germany); Dr. I. van der Burgt, Universitair Medisch Centrum Nijmegen (The Netherlands); Dr. I. Stolte-Dijkstra, University Hospital Groningen (The Netherlands); Dr. E. Kirk, Children's Hospital, Sydney (Australia); Dr. E.C. Prott, University of Essen (Germany); Dr. G. Mortier, University of Ghent (Belgium); Dr. W. Reardon, National Centre for Medical Genetics, Dublin (Ireland); Dr. A. Fryer, Countess of Chester Hospital, Chester (United Kingdom); Prof. J. Battin, Medical Genetics, Bordeaux (France); Dr. C. Adam, Trinidad (West Indies); Dr. C. Altuzarra, Unité Marfan CHUB, Besançon (France); Dr. L. A. Demmer, Tufts University of Medicine, Boston, Massachusetts (United States).
We thank C. Rossier and C. Chiesa for technical assistance. We are grateful to M. Casellini, F. M. Catzeflis, D. L. Dittmann, J. A. Marshall Graves, M. Ruedi, F. Sheldon, P. Vogel, C. Wenker, and M. Westerman for DNA samples. This project was supported by the Swiss National Science Foundation, the Telethon Action Suisse, the Désirée and Niels Yde Foundation, the Jérôme Lejeune Foundation, and the European Skeletal Dysplasia Network (http://www.esdn.org/). Tissues were provided by: “Institut des Sciences de l'Evolution,” Montpellier University 2; the Louisiana Museum of Natural History, Louisiana State University; the Comparative Genomics Research Group, Australian National University; the Department of Genetics, La Trobe University; the Basel Zoo; the Museum of Natural History of Geneva; and the Institute of Ecology, University of Lausanne.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. LB, AS, and AR conceived and designed the experiments. LB, ETD, CU , and AR performed the experiments. LB, ETD, SU, BACX, and AR analyzed the data. CRG and AZ contributed reagents/materials/analysis tools. LB, SEA, ASF, and AR wrote the paper.
Abbreviations
bpbase pair(s)
CHHcartilage-hair hypoplasia
GHgrowth hormone
SDSSchwachman-Diamond syndrome
SNPsingle nucleotide polymorphism
==== Refs
References
McKusick VA Eldridge R Hostetler JA Ruangwit U Egeland JA 1965 Dwarfism in the Amish. II. Cartilage-hair hypoplasia Bull Johns Hopkins Hosp 116 285 326 14284412
Trojak JE Polmar SH Winkelstein J Hsu S Francomano CA 1981 Immunologic studies of cartilage-hair hypoplasia in the Amish Johns Hopkins Med J 148 157 164 6970848
Makitie O 1992 Cartilage-hair hypoplasia in Finland: Epidemiological and genetic aspects of 107 patients J Med Genet 29 652 655 1404295
Makitie O Rajantie J Kaitila I 1992 Anaemia and macrocytosis—Unrecognized features in cartilage-hair hypoplasia Acta Paediatr 81 1026 1029 1290847
Makitie O Kaitila I 1993 Cartilage-hair hypoplasia—clinical manifestations in 108 Finnish patients Eur J Pediatr 152 211 217 8444246
Makitie O Pukkala E Teppo L Kaitila I 1999 Increased incidence of cancer in patients with cartilage-hair hypoplasia J Pediatr 134 315 318 10064668
Makitie O Kaitila I Rintala R 2001 Hirschsprung disease associated with severe cartilage-hair hypoplasia J Pediatr 138 929 931 11391344
Makitie OM Tapanainen PJ Dunkel L Siimes MA 2001 Impaired spermatogenesis: An unrecognized feature of cartilage-hair hypoplasia Ann Med 33 201 205 11370774
Topper JN Clayton DA 1990 Characterization of human MRP/Th RNA and its nuclear gene: Full length MRP/Th RNA is an active endoribonuclease when assembled as an RNP Nucleic Acids Res 18 793 799 1690392
Sulisalo T Sistonen P Hastbacka J Wadelius C Makitie O 1993 Cartilage-hair hypoplasia gene assigned to chromosome 9 by linkage analysis Nat Genet 3 338 341 7981754
Ridanpaa M van Eenennaam H Pelin K Chadwick R Johnson C 2001 Mutations in the RNA component of RNase MRP cause a pleiotropic human disease, cartilage-hair hypoplasia Cell 104 195 203 11207361
Tollervey D Kiss T 1997 Function and synthesis of small nucleolar RNAs Curr Opin Cell Biol 9 337 342 9159079
Cai T Reilly TR Cerio M Schmitt ME 1999 Mutagenesis of SNM1, which encodes a protein component of the yeast RNase MRP, reveals a role for this ribonucleoprotein endoribonuclease in plasmid segregation Mol Cell Biol 19 7857 7869 10523674
Chang DD Clayton DA 1987 A mammalian mitochondrial RNA processing activity contains nucleus-encoded RNA Science 235 1178 1184 2434997
Chang DD Clayton DA 1987 A novel endoribonuclease cleaves at a priming site of mouse mitochondrial DNA replication EMBO J 6 409 417 3582365
Chu S Archer RH Zengel JM Lindahl L 1994 The RNA of RNase MRP is required for normal processing of ribosomal RNA Proc Natl Acad Sci U S A 91 659 663 8290578
Clayton DA 1994 A nuclear function for RNase MRP Proc Natl Acad Sci U S A 91 4615 4617 8197106
van Eenennaam H Jarrous N van Venrooij WJ Pruijn GJ 2000 Architecture and function of the human endonucleases RNase P and RNase MRP IUBMB Life 49 265 272 10995027
Bonafe L Schmitt K Eich G Giedion A Superti-Furga A 2002 RMRP gene sequence analysis confirms a cartilage-hair hypoplasia variant with only skeletal manifestations and reveals a high density of single-nucleotide polymorphisms Clin Genet 61 146 151 11940090
Ridanpaa M Sistonen P Rockas S Rimoin DL Makitie O 2002 Worldwide mutation spectrum in cartilage-hair hypoplasia: Ancient founder origin of the major70A→G mutation of the untranslated RMRP Eur J Hum Genet 10 439 447 12107819
Kuijpers TW Ridanpaa M Peters M de Boer I Vossen JM 2003 Short-limbed dwarfism with bowing, combined immune deficiency, and late onset aplastic anaemia caused by novel mutations in the RMPR gene J Med Genet 40 761 766 14569125
Nakashima E Mabuchi A Kashimada K Onishi T Zhang J 2003 RMRP mutations in Japanese patients with cartilage-hair hypoplasia Am J Med Genet 123A 253 256 14608646
Schmitt K Bennett JL Dairaghi DJ Clayton DA 1993 Secondary structure of RNase MRP RNA as predicted by phylogenetic comparison FASEB J 7 208 213 7678563
van Eenennaam H Vogelzangs JHP Lugtenberg D Van Den Hoogen FHJ van Venrooij WJ 2002 Identity of the RNase MRP- and RNase P-associated Th/To autoantigen Arthritis Rheum 46 3266 3272 12483731
Murphy WJ Eizirik E O'Brien SJ Madsen O Scally M 2001 Resolution of the early placental mammal radiation using Bayesian phylogenetics Science 294 2348 2351 11743200
Li K Smagula CS Parsons WJ Richardson JA Gonzalez M 1994 Subcellular partitioning of MRP RNA assessed by ultrastructural and biochemical analysis J Cell Biol 124 871 882 7510714
Pluk H van Eenennaam H Rutjes SA Pruijn GJ van Venrooij WJ 1999 RNA-protein interactions in the human RNase MRP ribonucleoprotein complex RNA 5 512 524 10199568
van Eenennaam H van der Heijden A Janssen RJ van Venrooij WJ Pruijn G 2001 Basic domains target protein subunits of the RNase MRP complex to the nucleolus independently of complex association Mol Biol Cell 12 3680 3689 11694598
Welting TJ van Venrooij WJ Pruijn GJ 2004 Mutual interactions between subunits of the human RNase MRP ribonucleoprotein complex Nucleic Acids Res 32 2138 2146 15096576
Ridanpaa M Jain P McKusick VA Francomano CA Kaitila I 2003 The major mutation in the RMRP gene causing CHH among the Amish is the same as that found in most Finnish cases Am J Med Genet C Semin Med Genet 121 81 83
Murphy S Di Liegro C Melli M 1987 The in vitro transcription of the 7SK RNA gene by RNA polymerase III is dependent only on the presence of an upstream promoter Cell 51 81 87 3652210
Sbisa E Pesole G Tullo A Saccone C 1996 The evolution of the RNase P- and RNase MRP-associated RNAs: Phylogenetic analysis and nucleotide substitution rate J Mol Evol 43 46 57 8660429
Ridanpaa M Ward LM Rockas S Sarkioja M Makela H 2003 Genetic changes in the RNA components of RNase MRP and RNase P in Schmid metaphyseal chondrodysplasia J Med Genet 40 741 746 14569119
Makitie O Juvonen E Dunkel L Kaitila I Siimes MA 2000 Anemia in children with cartilage-hair hypoplasia is related to body growth and to the insulin-like growth factor system J Clin Endocrinol Metab 85 563 568 10690856
Halle MA Collipp PJ Roginsky M 1970 Cartilage-hair hypoplasia in childhood N Y State J Med 70 2705 2708 5273153
Bocca G Weemaes CM van der Burgt I Otten BJ 2004 Growth hormone treatment in cartilage-hair hypoplasia: Effects on growth and the immune system J Pediatr Endocrinol Metab 17 47 54 14960021
Hagenas L Hertel T Skeletal dysplasia, growth hormone treatment and body proportion: Comparison with other syndromic and non-syndromic short children Horm Res 60 65 70 14671400
Kanazawa H Tanaka H Inoue M Yamanaka Y Namba N 2003 Efficacy of growth hormone therapy for patients with skeletal dysplasia J Bone Miner Metab 21 307 310 12928832
Makitie O Ellis L Durie PR Morrison JA Sochett EB 2004 Skeletal phenotype in patients with Shwachman-Diamond syndrome and mutations in SBDS Clin Genet 65 101 112 14984468
Boocock GRB Morrison JA Popovic M Richards N Ellis L 2003 Mutations in SBDS are associated with Shwachman-Diamond syndrome Nat Genet 33 97 101 12496757
van der Burgt I Haraldsson A Oosterwijk JC van Essen AJ Weemaes CM 1991 Cartilage hair hypoplasia, metaphyseal chondrodysplasia type McKusick: Description of seven patients and review of the literature Am J Med Genet 41 371 380 1789294
Lux SE Johnston RBJ August CS Say B Penchaszadeh VB 1970 Chronic neutropenia and abnormal cellular immunity in cartilage-hair hypoplasia N Engl J Med 282 231 236 4188537
Wilson WG Aylsworth AS Folds JD Whisnant JK 1978 Cartilage-hair hypoplasia (metaphyseal chondrodysplasia, type McKusick) with combined immune deficiency: Variable expression and development of immunologic functions in sibs Birth Defects Orig Artic Ser 14 117 129 728552
Burgert EOJ Dower JC Tauxe WN 1965 A new syndrome—aregenerative anemia, malabsorption (celiac), dyschondroplasia and hyperphosphatemia [abstract] J Pediatr 67 711 712
Ashby GH Evans DI 1986 Cartilage hair hypoplasia with thrombocytopenic purpura, autoimmune haemolytic anaemia and cell-mediated immunodeficiency J R Soc Med 79 113 114 3950893
Berthet F Siegrist CA Ozsahin H Tuchschmid P Eich G 1996 Bone marrow transplantation in cartilage-hair hypoplasia: Correction of the immunodeficiency but not of the chondrodysplasia Eur J Pediatr 155 286 290 8777921
Rubie H Graber D Fischer A Tauber MT Maroteaux P 1989 [Hypoplasia of cartilage and hair with combined immune deficiency] Ann Pediatr 36 390 392
Dermitzakis ET Reymond A Scamuffa N Ucla C Kirkness E 2003 Evolutionary discrimination of mammalian conserved non-genic sequences (CNGs) Science 302 1033 1035 14526086
Thiel CT Horn D Zabel B Ekici AB Salinas K 2005 Severely incapacitating mutations in patients with extreme short stature identify FNA-processing endoribonuclease RMRP as an essential cell growth regulator Am J Hum Genet 77 795 806 16252239
|
16244706
|
PMC1262189
|
CC BY
|
2021-01-05 08:00:24
|
no
|
PLoS Genet. 2005 Oct 21; 1(4):e47
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010047
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1624470710.1371/journal.pgen.001005105-PLGE-RA-0201R1plge-01-04-04Research ArticleBioinformatics - Computational BiologyEvolutionVirologyVirusesThe Evolutionary Value of Recombination Is Constrained by Genome Modularity Genome Modularity and RecombinationMartin Darren P 1*van der Walt Eric 2Posada David 3Rybicki Edward P 21 Institute of Infectious Diseases and Molecular Medicine, University of Cape Town, Cape Town, South Africa
2 Department of Molecular and Cell Biology, University of Cape Town, Cape Town, South Africa
3 Department of Biochemistry, Genetics and Immunology, University of Vigo, Vigo, Spain
Gibson Greg EditorNorth Carolina State University, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 21 10 2005 22 9 2005 1 4 e5116 8 2005 22 9 2005 Copyright: © 2005 Martin 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.Genetic recombination is a fundamental evolutionary mechanism promoting biological adaptation. Using engineered recombinants of the small single-stranded DNA plant virus, Maize streak virus (MSV), we experimentally demonstrate that fragments of genetic material only function optimally if they reside within genomes similar to those in which they evolved. The degree of similarity necessary for optimal functionality is correlated with the complexity of intragenomic interaction networks within which genome fragments must function. There is a striking correlation between our experimental results and the types of MSV recombinants that are detectable in nature, indicating that obligatory maintenance of intragenome interaction networks strongly constrains the evolutionary value of recombination for this virus and probably for genomes in general.
Synopsis
Genetic exchange between organisms, called recombination, occurs in all biological kingdoms and is also common in viruses in which it may threaten the long-term control of important human pathogens such as HIV and influenza. Although recombination can produce advantageous gene combinations, bioinformatic analyses of bacterial genomes have suggested that recombination is not well tolerated when it involves exchanges of genes that interact with a lot of other genes. Using laboratory-constructed recombinants of a small plant virus called MSV, Martin and co-workers provide the first direct experimental evidence that the evolutionary value of exchanging a genome fragment is constrained by the number of ways in which the fragment interacts with the rest of the genome. They note that fitness losses suffered by artificial MSV recombinants increase with decreasing parental relatedness. Furthermore, these losses accurately anticipate the patterns of genetic exchange detectable in natural MSV recombinants, suggesting that they accurately reflect the impact of deleterious selection on natural isolates of the virus.
Citation:Martin DP, van der Walt E, Posada D, Rybicki EP (2005) The evolutionary value of recombination is constrained by genome modularity. PLoS Genet 1(4): e51.
==== Body
Introduction
Genetic recombination may predate the evolution of cellular life [1] and is the basis of ubiquitous biological processes such as DNA repair and sexual reproduction. The combinatorial nature of recombination can provide organisms with vastly more evolutionary options than are available through mutation alone [2–4]. However, kingdom-wide analyses of bacterial recombination [5] and DNA-shuffling studies [6,7] have indicated that the evolutionary value of recombination can vary depending on both the genes and the sub-gene modules transferred. In bacteria, the complexity hypothesis has been proposed to explain an imbalance in detectable informational and operational gene transfers between species [5]. Similarly, the schema hypothesis has been proposed to explain patterns of sequence mosaics observed in DNA-shuffling experiments [7]. Although the complexity hypothesis concerns genes within the context of genomes, the schema hypothesis concerns amino acids within the context of proteins. Both hypotheses are conceptually related and propose that the functionality of sequence fragments in foreign genetic backgrounds is inversely correlated with the complexity of interaction networks within which they must function.
Here we provide experimental support for these hypotheses using the small single stranded plant DNA virus, Maize streak virus (MSV; Geminiviridae, Mastrevirus) as a model organism to investigate the effect of genomic recombination on viral fitness.
Results
Recombinant Fitness
The MSV genome is approximately 2,690 nucleotides long and contains only three genes and two intergenic regions. We constructed 18 paired reciprocal recombinants (36 genomes in total) from four pairs of MSV isolates sharing genome-wide nucleotide sequence identities of 98%, 95%, 89%, and 78% [8]. Recombinant viruses were constructed in which the three genes and two intergenic regions of MSV were reciprocally exchanged between the four pairs of viruses (Figure 1).
Figure 1 Laboratory Constructed MSV Recombinants and Their Relative Fitness (as Measured by ICLAs) in Maize
Five genome regions corresponding to the three MSV genes (MP, CP, and Rep) and two intergenic regions (LIR and SIR) were reciprocally exchanged between four pairs of MSV isolates (MSV-Mat/MSV-Kom, MSV-Mat/MSV-R2, MSV-Mat/MSV-VW, and MSV-Kom/MSV-Set). Genome-wide sequence identities are indicated. SD, standard deviation.
As a correlate of viral fitness, we determined the induced chlorotic leaf areas (ICLA) of parental and recombinant viruses in infected maize plants [8]. The relationship between symptom severity and fitness is complex for most pathogens. However, MSV populates mesophyll cells within precisely defined chlorotic lesions of infected maize leaves [9] and the chlorotic surface area of an infected leaf is positively correlated with the total amount of viral DNA within the leaf [10,11]. The correlation between MSV pathogenicity and fitness is also evident in the greater geographical distribution and incidence of more pathogenic MSV genotypes relative to less pathogenic genotypes [12].
For each pair of reciprocal recombinants, we defined the recombination tolerance index (Ti) as the average ICLA of the recombinant pair divided by the average ICLA of their parental viruses. If reciprocally exchanged sequence fragments continue to function as well within recombinant genomes as they did in their original genomic backgrounds, we expect that the average ICLA of reciprocal pairs should be identical to that of their parental pairs—i.e., we would expect a Ti = 1.0. Conversely a drop in Ti below 1.0 would indicate that reciprocally exchanged sequences might not function as well in their new genomic backgrounds as they did in their original backgrounds.
In 17 out of 18 recombination experiments, the average ICLA of reciprocal recombinant pairs was lower than that of their parental viruses (i.e., Ti < 1). Values of Ti generally decreased with increasing divergence of exchanged sequences, with a distinct rate of decrease for each genome region exchanged (Figure 2). Given equal degrees of divergence, it seems that the short intergenic region (SIR) and movement protein gene (MP) function better in foreign genetic backgrounds than do the replication-associated protein gene (Rep), the coat protein gene (CP), or the long intergenic region (LIR). In other words, the SIR and MP appear more modular than the other regions of the genome.
Figure 2 Tolerance of Recombination in MSV Differs According to the Genome Region Involved and the Degree of Divergence between Exchanged Sequences
Each plotted point represents a Ti value calculated as the average fitness of a pair of recombinant viruses with reciprocally exchanged MP genes (cyan circles), CP genes (orange diamonds), Rep genes (blue inverted triangles), SIR (red triangles), or LIR (green squares) divided by the average fitness of their parental viruses. Error bars represent the standard deviations of Ti values. Curved lines represent quadratic regressions of Ti values against parental SIR, MP, LIR, CP, and Rep nucleotide sequences.
Genome Modularity
To better understand these differences in modularity, we examined the network of known direct protein–protein and protein–DNA interactions that occur during an MSV infection of maize (Figure 3) [13–22]. Whereas every other genome component or its expression product participates in multiple specific protein–DNA and/or protein–protein interactions with other virus components, the SIR apparently interacts only with host transcription and DNA replication factors [16]. The only known specific interaction of the MP with another virus component is a protein–protein interaction with the CP gene [18].
Figure 3 The Network of Known Protein–Protein and Protein–DNA Interactions during a MSV Infection of Maize
Rep/RepA indicates the two alternative expression products of the replication-associated protein gene. Solid lines represent specific protein–protein interactions [14,16–18,21,22], dotted lines represent specific protein–DNA interactions [14,16,19–21], and dashed lines indicate CP-DNA interactions of unknown specificity [18–20,22]. For protein–DNA interactions, arrows point from the protein component to the DNA component of the interactions. Rep interacts with the LIR at three distinct sites [14]. CP and Rep form oligomers (solid circular arrows) [14,22]. Although CP must interact with the rest of the genome (including its own gene) during encapsidation, the sequence specificity of these interactions is unknown.
To determine whether the known network of interactions occurring during an MSV infection is anticipated by our Ti data, it was necessary to first extract a relative modularity score for each of the MSV genome regions from the Ti plots. To do this we fitted quadratic equations to the plots (see Figure 2) to estimate similarly tolerable degrees of recombination-induced diversification (RID) in the five genomic regions. Our aim in fitting a line to the Ti plots was to objectively estimate degrees of RID tolerated in the different genome regions at particular Ti values. For example, fitting quadratics to the plots and picking a Ti value of 0.9 (reciprocal recombinants have an average ICLA 90% that of their parents) the corresponding estimates of tolerable recombination induced diversification in the SIR, MP, CP, LIR, and Rep are 15.3%, 8.3%, 2.7%, 3.3%, and 3.8% respectively. To avoid any biases due to “cherry picking” the Ti values we used to compare the different genome regions, we examined estimated recombination tolerances over the range of Ti values between 0.99 and 0.7. There is good correlation between the number of known MSV intergenome component protein–protein and protein–DNA interactions for different genome regions (SIR = 1, MP = 2, CP = 5, LIR = 4, and Rep = 4) and recombination tolerance estimated for these same regions over the entire Ti range between 0.94 and 0.7 (Pearson's R
2 > 0.87, p < 0.05, Spearman's rho corrected for ties = −0.975, p = 0.051).
Analysis of Natural Recombinants
To investigate whether experimentally determined Ti values provide any insights into real processes that influence the survival of recombinants in nature, we examined all available Mastrevirus sequences in GenBank for evidence of recombination. In each of the five genomic regions (LIR, MP, CP, SIR, and Rep), we identified evidence for unique recombination events involving MSV isolates or the closely related African streak viruses. We only retained evidence of recombination events detectable in genomes with proven viability (as determined by the existence of infectious clones of these genomes). For each recombination event involving two identifiable parental sequences and comprising two easily identifiable breakpoints both unambiguously within one of the five defined genomic regions, we used the identified parental sequences to infer the number of nucleotide differences between the transferred sequence and the sequence it replaced (Table 1). The greatest number of nucleotide changes observed in a single recombination event in each of the five genomic regions provides a gross estimate of the maximum parental divergence tolerated in nature. The set of values thus determined (SIR = 15.0%, MP = 7.7%, CP = 3.2%, LIR = 5.4%, and Rep = 3.9%) was significantly correlated with equivalent sets of values derived from the quadratic regressions shown in Figure 2 (Figure 4; Pearson's R
2 > 0.96, p < 0.01, Spearman's rho = 0.9, p = 0.037 over the Ti range 0.95–0.7).
Table 1 Evidence of Recombination within Five Defined Regions of Publicly Available African Streak Virus Genome Sequences
Figure 4 The Experimentally Determined Relative Recombination Tolerance of Five MSV Genomic Regions Is Highly Correlated with Values Inferred from Nature
We used the quadratic regressions presented in Figure 2 to derive experimental estimates of similarly tolerable degrees of RID in the five genomic regions over a range of Ti values between 0.99 and 0.70 (290 values at 0.01 intervals). Using a range of Ti values avoids any biases that might occur due to inadvertently choosing particularly poor/favourable Ti values for estimating similarly tolerable degrees of RID from the experimental data. For example, the set of similarly tolerable degrees of RID calculated when Ti = 0.9 is 15.3%, 8.3%, 2.7%, 3.3%, and 3.8% for the SIR, MP gene, CP gene, LIR, and Rep gene, respectively. Each of the 290 sets of values thus determined was linearly regressed against the set of values for the maximum tolerable RID inferred for the same five regions from an examination of natural recombinants (15.0%, 7.7%, 3.2%, 5.4%, and 3.9%, for SIR, MP, CP, LIR, and Rep, respectively). R
2 values determined for these 290 regressions are plotted against their corresponding Ti values (solid line). The correlation is significant (broken line = R
2 value corresponding to a p-value < 0.01) over the Ti range 0.95–0.7 (Pearson's R
2 > 0.96, p < 0.01; Spearman's rho = 0.9, p = 0.037).
Discussion
By demonstrating a negative correlation between the relative modularity of defined genomic regions and the complexity of interactions in which they are involved we have provided experimental support for the complexity hypothesis [5]. This hypothesis was proposed to explain the disparity in informational (those involved in transcription, translation, and related processes) and operational (those involved in housekeeping) gene transfer rates in bacteria. It states that because informational genes are generally involved in more complex interactions than operational genes, they are less likely to continue functioning well after horizontal transfer.
The progressive decrease in tolerance of recombination with increasing divergence of exchanged sequences observed in Figure 2 has strong parallels with parental sequence imbalances observed in “family shuffling” variants of DNA-shuffling experiments [2]. The functional genes produced by shuffling three or more distinct sequences (i.e., 60%–85% identical) are usually derived predominantly from either one sequence or combinations of the most similar sequences [2,6,23,24]. The schema hypothesis proposes that these imbalances are due to the probability of recombinant protein-fold disruption increasing with increasing divergence of parental sequences [6,7]. We shuffled entire intergenic regions and genes, and therefore the effect we observed cannot be explained directly in terms of the schema hypothesis. We have, however, provided empirical evidence for a whole-genome analogue of this hypothesis: The probability of the normal network of intragenome interactions being disrupted by recombination increases with increasing divergence of the exchanged fragments.
The successful inheritance of genomic fragments through recombination is expected to depend on the maintenance of important intragenome interactions. After all, the exchange of a genome fragment could be seen as a simultaneous introduction of multiple mutations. Negative or purifying selection should remove those recombinants that break the epistatic interactions that define the architecture of a particular genome, whereas genetic drift might permit the survival and spread of “neutral” recombinants. Alternatively, positive selection should favour the spread of rare recombinants with improved genomic interactions. The genomic interactions in the (natural) parental viruses used in these experiments have most likely been optimised through selection over long evolutionary periods. None of the recombinants generated from these viruses was more fit than the fitter of its parents, which is expected if negative selection is the dominant force that now maintains the integrity of these genomes.
The relative degree of modularity that we demonstrated experimentally for each genome component appears to be reflected in the recombination events detected within the same regions in natural viruses represented in GenBank. This correlation is surprising because the natural recombinants—unlike the recombinants we constructed in the laboratory—involve exchanges of fragments of genes or intergenic regions. Such exchanges may disrupt intraprotein or intra-intergenic region interactions as well as interaction networks amongst whole genes and intergenic regions. Survival of the natural viruses with detectable recombination events in coding regions presumably depended on their inheritance of sequences that did not overly disrupt either intraprotein or intergene/intergenic region interaction networks; survival of natural intergenic region recombinants and those we generated in the laboratory would have been subject only to the latter constraint. The correlation between our experimental results and the inferred natural recombinants may indicate that maintenance of intergenome component interactions is the principal determinant of recombination tolerance (at least for MSV and closely related viruses). Alternatively, a requirement for the preservation of both intergenome component, and intragene, interaction networks has a net effect that is difficult to distinguish from either constraint operating alone.
We have provided experimental support for the complexity hypothesis by demonstrating a relationship between the relative modularity of defined genomic regions and the complexity of interactions in which they are involved. The striking correlation between our experimental results and the types of recombination observed in nature lends credence to the notion that these detectable modularity differences are evolutionarily relevant. Our results also suggest that the degree of similarity between an inherited sequence and the sequence it replaces is an important additional determinant of recombinant fitness. Whereas recombination can substantially increase the evolutionary options of an organism, the obligatory maintenance of co-evolved interaction networks may severely restrict its evolutionary value.
Materials and Methods
Recombinant construction.
We have previously described the construction and symptom analysis of 15 reciprocal recombinant pairs for the MSV isolates MSV-Mat, MSV-Kom, MSV-R2, and MSV-VW [8]. PCR mutagenesis was used to introduce NcoI restriction sites immediately upstream of the CP gene start codons of the MSV isolates MSV-Set and MSV-Kom to obtain SNco and KNco, respectively. Reciprocal MP gene recombinants were obtained by exchanging BamHI-NcoI restriction fragments containing the entire MP between MSV-SNco and MSV-KNco. Reciprocal CP recombinants were obtained by exchanging NcoI-NcoI restriction fragments containing the entire CP between MSV-SNco and MSV-KNco. Agroinfectious partially dimeric clones of recombinant viruses were constructed as previously described [10].
Fitness testing.
The fitness of recombinant and parental viruses as indicated by ICLA was determined in maize by agroinoculation of 3-d-old cv. Jubilee seedlings with image analysis quantification of ensuing disease symptoms [8]. An ICLA score for each virus was determined by averaging the percentage chlorotic areas determined on leaves two through six for between three and 14 symptomatic plants in each of three to seven replicated agroinoculation experiments.
Recombination detection.
All available Mastrevirus sequences were obtained from GenBank and aligned using POA [25] with gap open and gap extension penalties of 12 and 6, respectively. Identification of potential recombinant and parental sequences, and localisation of possible recombination breakpoints was carried out using the RDP [26], Geneconv [27], RecScan [28], Maximum Chi Square [29], Chimaera [30], and SisterScan [31] methods as implemented in RDP2 [30]. The analysis was performed with default settings for the detection methods, a Bonferroni corrected p-value cutoff of 0.05, and a requirement that any potential event be detectable by two or more methods.
Supporting Information
Accession Numbers
The GenBank (http://www.ncbi.nlm.nih.gov/Genbank) accession numbers for the genome sequences are CSMV (NC001466), DSV (NC001478), MiSV (NC003379), MSV-Ama (AF329878), MSV-Gat (AF329879), MSV-Jam (AF329887), MSV-K (X01089), MSV-KA (AF329885), MSV-Km (AF395891), MSV-Kom (AF003952), MSV-MakD (AF329884), MSV-Nig (X01633), MSV-Pat (AF329888), MSV-Raw (AF329889), MSV-Reu (X94330), MSV-SA (NC001346), MSV-Sag (AF329880), MSV-Set (AF007881), MSV-Tas (AF239962), MSV-VM (AF239961), MSV-VW (AF239960), PanSV-Kar (NC001647), PanSV-Ken (X60168), SSEV-Ben (AF039529), SSRV (NC004755), and SSV-N (NC003744).
We would like to thank the South African National Research Foundation and National Bioinformatics Network for funding this work. DP is supported by the Ramón y Cajal Programme of the Spanish Government, and funded by grants R01-GM55276 from the United States National Institutes of Health, and BFU2004–02700 from the Spanish Education and Science Ministry (MEC).
Competing interests. The authors have declared that no competing interests exist.
Author contributions. DPM, EVDW, and EPR conceived and designed the experiments. DPM, EVDW, and EPR performed the experiments. DPM, EVDW, DP, and EPR analyzed the data. DPM contributed reagents/materials/analysis tools. DPM, EVDW, DP, and EPR wrote the paper.
A previous version of this article appeared as an Early Online Release on September 22, 2005 (DOI: 10.1371/journal.pgen.0010051.eor).
Abbreviations
ICLAinduced chlorotic leaf area
LIRlong interegenic region
MSVMaize streak virus
RIDrecombination-induced diversification
SIRshort intergenic region
Titolerance index
==== Refs
References
Lehman NA 2003 A case for the extreme antiquity of recombination J Mol Evol 56 770 777 12911039
Crameri A Raillard SA Bermudez E Stemmer WP 1998 DNA shuffling of a family of genes from diverse species accelerates directed evolution Nature 391 288 291 9440693
Drummond DA Silberg JJ Meyer MM Wilke CO Arnold FH 2005 On the conservative nature of intragenic recombination Proc Natl Acad Sci U S A 102 5380 5385 15809422
Stemmer WPC 1994 Rapid evolution of a protein in vitro by DNA shuffling Nature 370 389 391 8047147
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
Meyer MM Silberg JJ Voigt CA Endelman JB Mayo SL 2003 Library analysis of SCHEMA-guided protein recombination Prot Sci 12 1686 1693
Voigt CA Martinez C Wang ZG Mayo SL Arnold FH 2002 Protein building blocks preserved by recombination Nat Struct Biol 9 553 558 12042875
Martin DP Rybicki EP 2002 Investigation of Maize streak virus pathogenicity determinants using chimaeric genomes Virology 300 180 188 12350349
Lucy AP Boulton MI Davies JW Maule AJ 1996 Tissue specificity of Zea mays infection by maize streak virus Mol Plant Microbe Interact 9 22 31
Schnippenkoetter WH Martin DP Willment JA Rybicki EP 2001 Forced recombination between distinct strains of Maize streak virus J Gen Virol 82 3081 3090 11714986
Shepherd DN Martin DP McGivern DR Boulton MI Thomson JA 2005 A three-nucleotide mutation altering the Maize streak virus Rep pRBR-interaction motif reduces symptom severity in maize and partially reverts at high frequency without restoring pRBR-Rep binding J Gen Virol 86 803 813 15722543
Martin DP Willment JA Billharz R Velders R Odhiambo B 2001 Sequence diversity and virulence in Zea mays of Maize streak virus isolates Virology 288 247 255 11601896
Boulton MI 2002 Functions and interactions of mastrevirus gene products Phys Mol Plant Path 5 243 255
Castellano MM Sanz-Burgos AP Gutierrez C 1999 Initiation of DNA replication in a eukaryotic rolling-circle replicon: Identification of multiple DNA-protein complexes at the geminivirus origin J Mol Biol 290 639 652 10395820
Dickinson VJ Halder J Woolston CJ 1996 The product of maize streak virus ORF V1 is associated with secondary plasmodesmata and is first detected with the onset of viral lesions Virology 220 51 59 8659128
Donson J Morris-Krsinich BA Mullineaux PM Boulton MI Davies JW 1984 A putative primer for second-strand DNA synthesis of maize streak virus is virion-associated EMBO J 3 3069 3073 6526010
Horvath GV Pettko-Szandtner A Nikovics K Bilgin M Boulton M 1998 Prediction of functional regions of the maize streak virus replication-associated proteins by protein-protein interaction analysis Plant Mol Biol 38 699 712 9862488
Liu H Boulton MI Oparka KJ Davies JW 2001 Interaction of the movement and coat proteins of Maize streak virus: Implications for the transport of viral DNA J Gen Virol 82 35 44 11125156
Liu H Boulton MI Thomas CL Prior DA Oparka KJ 1999 Maize streak virus coat protein is karyophyllic and facilitates nuclear transport of viral DNA Mol Plant Microbe Interact 12 894 900 10517029
Liu H Boulton MI Davies JW 1997 Maize streak virus coat protein binds single- and double-stranded DNA in vitro J Gen Virol 78 1265 1270 9191917
Nikovics K Simidjieva J Peres A Ayaydin F Pasternak T 2001 Cell-cycle, phase-specific activation of Maize streak virus promoters Mol Plant Microbe Interact 14 609 617 11332725
Zhang W Olson NH Baker TS Faulkner L Agbandje-McKenna M 2001 Structure of the Maize streak virus geminate particle Virology 279 471 477 11162803
Aharoni A Gaidukov L Yagur S Toker L Silman I 2004 Directed evolution of mammalian paraoxonases PON1 and PON3 for bacterial expression and catalytic specialization Proc Natl Acad Sci U S A 101 482 487 14695884
Joern JM Meinhold P Arnold FH 2002 Analysis of shuffled gene libraries J Mol Biol 316 643 656 11866523
Lee C Grasso C Sharlow MF 2002 Multiple sequence alignment using partial order graphs Bioinformatics 18 452 464 11934745
Martin D Rybicki E 2000 RDP: Detection of recombination amongst aligned sequences Bioinformatics 16 562 563 10980155
Padidam M Sawyer S Fauquet CM 1999 Possible emergence of new geminiviruses by frequent recombination Virology 265 218 225 10600594
Martin DP Posada D Crandall KA Williamson CA 2005 A modified bootscan algorithm for automated identification of recombinant sequences and recombination breakpoints AIDS Hum Retro 21 98 102
Smith JM 1992 Analyzing the mosaic structure of genes J Mol Evol 34 126 129 1556748
Martin DP Williamson C Posada D 2005 RDP2: Recombination detection and analysis from sequence alignments Bioinformatics 21 260 262 15377507
Gibbs MJ Armstrong JS Gibbs AJ 2000 Sister-scanning: a Monte Carlo procedure for assessing signals in recombinant sequences Bioinformatics 16 573 582 11038328
|
16244707
|
PMC1262190
|
CC BY
|
2021-01-05 08:00:24
|
no
|
PLoS Genet. 2005 Oct 21; 1(4):e51
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010051
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623199310.1371/journal.pmed.0020287PerspectivesBioethicsInfectious DiseasesEpidemiology/Public HealthHealth PolicyHIV/AIDSSexual HealthUrologyEthicsHIV Infection/AIDSMedicine in Developing CountriesInfectious DiseasesSexually transmitted infections - other than HIV/AIDSThe First Randomised Trial of Male Circumcision for Preventing HIV: What Were the Ethical Issues? PerspectivesCleaton-Jones Peter Peter Cleaton-Jones is Chair of the Human Research Ethics Committee (Medical), University of the Witwatersrand, Johannesburg, South Africa. E-mail: [email protected]
Competing Interests: The author has declared that no competing interests exist.
11 2005 25 10 2005 2 11 e287Copyright: © 2005 Peter Cleaton-Jones.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.
Randomized, Controlled Intervention Trial of Male Circumcision for Reduction of HIV Infection Risk: The ANRS 1265 Trial
Does Male Circumcision Prevent HIV Infection?
A Landmark Paper in HIV Research?
Cleaton-Jones, chair of the ethics committee that approved the trial of circumcision for preventing HIV, shares with us the discussions that the committee had ahead of granting approval.
==== Body
In the November issue of PLoS Medicine, Auvert and colleagues report the first randomised controlled trial of circumcision for preventing HIV infection [1]. In the study, 3,274 uncircumcised men, aged 18–24 years, were randomised to a control or intervention group, with follow-up at three, 12, and 21 months. Circumcision was offered to the intervention group immediately after randomisation and to the control group at the end of the follow-up (only men who wished to be circumcised were eligible for the trial). There were 20 HIV infections in the intervention group and 49 in the control group, corresponding to a relative risk of 0.40 (95% confidence interval, 0.24–0.68; p < 0.001). The relative risk was unchanged when controlled for behavioural factors, including sexual behaviour, condom use, and health-seeking behaviour, suggesting that male circumcision provides a degree of protection against acquiring HIV infection.
Prior to conducting the study, the study proposal (protocol study number M020104) was reviewed in January 2002 by the Human Research Ethics Committee (Medical), which is the institutional review board (IRB) of the University of the Witwatersrand in Johannesburg, South Africa, and which I chair. The committee decided to approve the study, and in this article I discuss how we came to our decision.
Background to the Committee
The committee was formed in October 1966 soon after the seminal paper by Beecher on ethics and clinical research was published [2]. It has functioned continuously, ever since, according to local [3] and international [4] research ethics guidelines. The committee has United States federal-wide accreditation (FWA 0000715), and is one of the few IRBs outside the US that has had a site quality assurance visit by a team from the US Office of Human Research Protections (the visit was in 2002, headed by Deputy Director Melody Lin).
At the time when we reviewed Auvert and colleagues' protocol, there were 27 members, 12 women and 15 men, from diverse ethnic backgrounds. Twenty-four members were from the University of the Witwatersrand, of whom 16 had medical or scientific expertise and eight had nonmedical backgrounds in law, social work, ethics, and psychology. There were three independent members (with backgrounds in education, nursing, and religion). Seven members of the committee had, or were obtaining, postgraduate qualifications in applied ethics up to doctoral level. Over half the members had at least ten years of experience—the maximum was 28 years—and three members served, or had served, on the IRBs of the Medical Research Council and Human Sciences Research Council in South Africa. All but three were born in South Africa.
Importance of the Study
There were four reasons why the committee believed that the proposed study had local importance, and that it should be conducted urgently. First, the proposal dealt with a serious infection with a seroprevalence in 2002 of 26.5% among pregnant women, according to a national government survey [5].
Second, in 2002, government policy concerning HIV was to provide condoms, safe-sex counselling, and voluntary HIV testing after counselling—but no antiretroviral treatment to those infected. Such treatment was only available from government health centres, starting in 2004 [6]. In other words, at the time when we considered the study protocol, the public-sector standard of care did not include antiretroviral medication.
Third, anecdotal evidence at the time when we considered this proposal suggested that circumcision might be protective against heterosexual acquisition of HIV in men, but firm evidence was lacking. A year later, a meta-analysis of published studies concluded that while epidemiological evidence was supportive, the outcome of randomised controlled trials would be important in determining the value of circumcision [7].
Finally, in many African cultures in South Africa, initiation into manhood is accompanied by circumcision performed by a “traditional surgeon”. In recent times, because of high mortality from haemorrhage, infections, and dehydration [8], many African men now opt for circumcision in adult life by a medical practitioner.
Ethical Considerations
We were, therefore, convinced that the study was important, and we then went on to consider whether it would be ethical. There were four guiding principles in our ethical deliberations.
Autonomy
Autonomy was respected through a written informed-consent process, with verbal translation into vernacular if needed. This process also fulfilled a local legal requirement, since informed consent to participate in research is entrenched in the Bill of Rights of the South African Constitution [9]. No false promises were made that might have influenced a person's decision to participate.
Beneficence
All participants would receive their desired circumcision. They would also have repeated medical examinations and counselling at each follow-up visit. Such counselling involved education about safe-sex practices and the benefits of being tested for HIV. Participants who wished to know their HIV status were referred for an HIV test.
One criticism that has been levelled at studies of HIV in South Africa is that not all participants are routinely informed of their HIV status. Our committee understands the stigmatisation faced by people in South Africa who are HIV positive [10], and so we accepted the right of individuals in the study to choose whether or not to be tested, once they had been counselled.
If during the study a participant became infected with syphilis, or any other sexually transmitted disease for which treatment was provided at government health centres, the participant would be referred to the nearby health centre to receive treatment.
Beneficence was also allowed for through scheduled interim analyses, with results provided to a three-person data safety and monitoring board in order to stop the study as soon as statistically significant results had been found (see below).
Non-maleficence
The study's proposed participants were to be only those who wished to be circumcised. All participants would receive their desired procedure, half immediately and half at the end of the study, so there would be no withholding of the operation. A potential source of disadvantage would be if there was a selection process before enrolment, for example, testing for HIV and accepting only people who were HIV negative into the study. The researchers had anticipated this, and all who wished to enrol were accepted—adjustment for HIV status at enrolment was through a large sample size and statistical analysis.
A second concern of the researchers was that the results of the tests for HIV and herpes simplex virus should not be known by the researchers in order to keep them “blinded” and, hence, minimise bias. Because voluntary testing for these viruses was offered at each follow-up, the researchers' concern was accepted by the committee.
No treatment would be provided for HIV or herpes simplex virus infection, according to government policy. The Declaration of Helsinki recommends provision of treatment for a disease being studied, and so our committee debated at length the issue of whether the researchers had an ethical duty to provide treatment. We concluded, as put forward by Benatar [11], that a moral judgment had to be made considering local conditions, the short period of the study (which was less than the 5–7 years considered appropriate at the time to reach a clinical need for antiretroviral treatment), and the finances involved. In 2002, the South African government was not offering treatment for these two infections, and so we agreed that it would be ethical for no treatment to be provided to the study's participants.
Justice
Justice was adhered to by ensuring that the potential participants were recruited from disadvantaged groups who might not have had ready access to medical circumcision.
Interim Analyses
The first interim analysis showed no statistically significant effect. The second, at 12 months, did show such an effect, and so the data safety and monitoring board recommended that the study be stopped—and the committee agreed. At this stage, the researchers amended the study to allow HIV, herpes simplex virus, and syphilis results recorded in the study to be given to those participants, after counselling and informed consent, who had not previously wished to know.
By then the government policy had changed to providing antiretroviral therapy. The researchers undertook to do their best to provide HIV treatment for up to five years to those participants for whom such treatment was clinically indicated, and the committee approved this course of action.
Conclusion
This randomised trial presented a challenge to the IRB to combine general principles of research ethics with local conditions to permit a very important study to be done. Functioning, as the IRB does, in a developing country environment requires knowledge of local cultures, resources, and services. This is not always understood nor accepted by IRBs and researchers who operate in resource-rich environments. The fact that circumcision is so clearly protective should benefit many men, particularly those in countries with a high HIV prevalence.
The author is most grateful to P. A. Cooper of the Human Research Ethics Committee (Medical) for his critical comments.
Citation: Cleaton-Jones P (2005) The fi rst randomised trial of male circumcision for preventing HIV: What were the ethical issues? PLoS Med 2(11): e287.
Abbreviation
IRBinstitutional review board
==== Refs
References
Auvert B Taljaard D Lagarde E Sobngwi-Tambekou J Randomized, controlled intervention trial of male circumcision for reduction of HIV infection risk: The ANRS 1265 trial PLoS Med 2005 2 e298 10.1371/journal.pmed.0020298 16231970
Beecher HK Ethics and clinical research New Engl J Med 1966 274 1354 1360 5327352
Medical Research Council of South Africa Guidelines on ethics for medical research: General principles, revised ed 2002 Tygerberg (South Africa) Medical Research Council of South Africa Available: http://www.sahealthinfo.org/ethics/ethicsbook1.pdf . Accessed 30 September 2005
United States National Archives and Records Administration Code of federal regulations, title 45—Public welfare: Part 46—Protection of human subjects 2001 Bethesda Department of Health and Human Services, General Administration, National Institutes of Health Available: http://www.access.gpo.gov/nara/cfr/waisidx_01/45cfr46_01.html . Accessed 26 July 2005
Department of Health, Republic of South Africa National HIV and syphilis antenatal sero-prevalence survey in South Africa 2004 2004 Available: http://www.doh.gov.za/docs/reports/2004/hiv-syphilis.pdf . Accessed 28 September 2005
Department of Health, Republic of South Africa National antiretroviral treatment guidelines, first edition 2004 Pretoria Department of Health, Republic of South Africa Available: http://www.doh.gov.za/docs/factsheets/guidelines/artguidelines04/index.html . Accessed 28 September 2005
Siegfried N Muller M Volmink J Deeks J Egger M Male circumcision for prevention of heterosexual acquisition of HIV in men Cochrane Database Syst Rev 2003 2003 CD003362
Apps P S. Africa acts on illegal circumcision after deaths 2005 June 28 London Reuters Foundation AlertNet
Constitutional Court of South Africa Constitution of the Republic of South Africa, Act 108 of 1996 1996 Johannesburg Constitutional Court of South Africa Available: http://www.concourt.gov.za/site/theconstitution/english.pdf . Accessed 26 July 2005
Nicodemus A Africa still stigmatises HIV-positive people: Many HIV-positive people in South Africa fear that revealing their positive status could mean isolation from their communities or even murder 1999 May 7 Johannesburg Mail and Guardian Available: http://www.aegis.com/news/dmg/1999/MG990504.html . Accessed 26 July 2005
Benatar SR Imperialism, research ethics and global health J Med Ethics 1998 24 221 222 9752622
|
16231993
|
PMC1262554
|
CC BY
|
2021-01-05 10:38:16
|
no
|
PLoS Med. 2005 Nov 25; 2(11):e287
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020287
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623199410.1371/journal.pmed.0020293EditorialInfectious DiseasesEpidemiology/Public HealthHealth PolicyHIV/AIDSSexual HealthUrologyInfectious DiseasesHIV Infection/AIDSClinical trialsMedicine in Developing CountriesSexually transmitted infections - other than HIV/AIDSA Landmark Paper in HIV Research? EditorialE-mail: [email protected]
11 2005 25 10 2005 2 11 e293Copyright: © 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.
Randomized, Controlled Intervention Trial of Male Circumcision for Reduction of HIV Infection Risk: The ANRS 1265 Trial
Does Male Circumcision Prevent HIV Infection?
The First Randomised Trial of Male Circumcision for Preventing HIV: What Were the Ethical Issues?
==== Body
The clinical trial published in PLoS Medicine on whether male circumcision can prevent HIV (DOI: 10.1371/journal.pmed.0020298) is a “landmark paper” in HIV research. So said the first reviewer we sent this paper to. It is the first randomized trial to be completed of circumcision for HIV prevention, a topic that has generated a huge amount of epidemiological research and argument since the first observation that the incidence of HIV was lower in men who were circumcised. The trial was stopped at an interim analysis at the request of the trial's data safety monitoring board. At a mean of 18.1 months' follow-up there were 20 infections in the intervention (circumcised) group versus 49 in the control group, representing a protection of 60% (95% confidence interval 32%–76%).
In many ways this trial is a paradigm for the whole of HIV research; it encompasses many of the practical, social, moral, and scientific difficulties of doing such research.
The first, practical, problem of doing a trial that can test the effect of circumcision on preventing HIV infection is recruiting sufficient numbers of individuals who are prepared to be circumcised, in an area where the event rate (new infections) is likely to be sufficiently high that the trial can be done in a reasonable timeframe. It seems obvious, but perhaps not politically correct, to say that such a trial could not be done in New York or London or Paris. The trial we publish was done in South Africa, where the event rate is high and many African men now opt for circumcision in adult life by a medical practitioner.
But this circumstance leads to the second, social, problem: the ethical issues involved in doing such a trial and of applying Western standards to a non-Western setting. People in developing countries do not want to be used by Western researchers to do trials that might not be allowed in their own countries. Trials should only be done that address crucial local health problems and in a way that is appropriate for each country, and potential participants should be included in discussions about what is acceptable to them. This trial fulfills these criteria; however, HIV-positive men were not excluded from participating nor were they told of their HIV status during the trial. The authors' reasoning for not informing participants of their HIV status as a routine part of the trial (as would be likely if a similar trial could be done in a Western setting), and which was accepted by the two ethics boards that reviewed this trial, was as follows. In a country where there is stigma attached to being HIV positive, automatic exclusion of HIV-positive individuals would have potentially exposed them to discrimination. In addition, there may be benefits for HIV-positive individuals in being circumcised, including protection against other sexually transmitted diseases and against re-infection by other strains of HIV. At each visit all participants were offered advice on HIV and other sexually transmitted diseases and strongly encouraged to seek voluntary counseling and testing at a center close by where HIV status was disclosed to the patient if asked for. Although arguments rage over whether HIV status should be private or public knowledge, at the moment individuals cannot be forced into testing or indeed into knowledge of their status. Moreover, even if participants knew of their status, they would not have had access to antiretroviral drugs, as these were not available in South Africa when the trial was done. These ethical issues are discussed in two accompanying Perspectives: one by Peter Cleaton-Jones, chair of the South African ethics committee that approved the trial (DOI: 10.1371/journal.pmed.0020287), and another by Nandi Siegfried (DOI: 10.1371/journal.pmed.0020393).
The final problem that then arises is for editors: if you have any ethical questions about a study, should you publish it? This surely is what peer review and the editorial process is for. During the process of peer review here, ethical questions were raised and so also were scientific concerns. These concerns included the following: that the randomization used was unusual (but appropriate for this community); that participants were paid; that there were substantial numbers of dropouts; that the trial was stopped early (at the request of the data monitoring committee but, some might argue, too early for conclusive results); and that circumcision did appear to change the sexual behavior of participants—but in a way likely to make them more prone to infection, i.e., making the findings more robust. In considering papers for publication editors must weigh all these issues and reviewers' comments. We took particular note that this trial was approved by two experienced ethics boards.
The six reviewers and the academic editor who saw this paper were unanimous on one point: that this trial must be published, quickly. Ultimately, if these results are correct, then this is a study that offers hope. Clearly, further randomized studies will be needed to confirm the results (one in Kenya is scheduled for completion in 2007), but to not put this paper in the public domain quickly could be considered unethical in its own right.
Citation: The PLoS Medicine Editors (2005) A landmark paper in HIV research? PLoS Med 2(11): e293.
|
16231994
|
PMC1262555
|
CC BY
|
2021-01-05 10:39:16
|
no
|
PLoS Med. 2005 Nov 25; 2(11):e293
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020293
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1623197010.1371/journal.pmed.0020298Research ArticleInfectious DiseasesEpidemiology/Public HealthHealth PolicyHIV/AIDSSexual HealthUrologyInfectious DiseasesHIV Infection/AIDSSexually transmitted infections - other than HIV/AIDSMedicine in Developing CountriesRandomized, Controlled Intervention Trial of Male Circumcision for Reduction of HIV Infection Risk: The ANRS 1265 Trial Male Circumcision TrialAuvert Bertran
1
2
3
4
*Taljaard Dirk
5
Lagarde Emmanuel
2
4
Sobngwi-Tambekou Joëlle
2
Sitta Rémi
2
4
Puren Adrian
6
1Hôpital Ambroise-Paré, Assitance Publique—Hôpitaux de Paris, Boulogne, France,2INSERM U 687, Saint-Maurice, France,3University Versailles Saint-Quentin, Versailles, France,4IFR 69, Villejuif, France,5Progressus, Johannesburg, South Africa,6National Institute for Communicable Disease, Johannesburg, South AfricaDeeks Steven Academic EditorSan Francisco General Hospital, San Francisco, CaliforniaUnited 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: BA designed the study with DT, EL, and AP. DT and AP were responsible for operational aspects, including laboratory and field work and in-country administration of the study. BA monitored the study with input from EL and wrote the paper with input from all authors. BA analyzed the data with RS, with inputs from JST. RS conducted the interim analysis.
11 2005 25 10 2005 2 11 e29829 6 2005 26 9 2005 Copyright: © 2005 Auvert 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.
First Trial of Male Circumcision against HIV
The First Randomised Trial of Male Circumcision for Preventing HIV: What Were the Ethical Issues?
Does Male Circumcision Prevent HIV Infection?
A Landmark Paper in HIV Research?
Background
Observational studies suggest that male circumcision may provide protection against HIV-1 infection. A randomized, controlled intervention trial was conducted in a general population of South Africa to test this hypothesis.
Methods and Findings
A total of 3,274 uncircumcised men, aged 18–24 y, were randomized to a control or an intervention group with follow-up visits at months 3, 12, and 21. Male circumcision was offered to the intervention group immediately after randomization and to the control group at the end of the follow-up. The grouped censored data were analyzed in intention-to-treat, univariate and multivariate, analyses, using piecewise exponential, proportional hazards models. Rate ratios (RR) of HIV incidence were determined with 95% CI. Protection against HIV infection was calculated as 1 − RR. The trial was stopped at the interim analysis, and the mean (interquartile range) follow-up was 18.1 mo (13.0–21.0) when the data were analyzed. There were 20 HIV infections (incidence rate = 0.85 per 100 person-years) in the intervention group and 49 (2.1 per 100 person-years) in the control group, corresponding to an RR of 0.40 (95% CI: 0.24%–0.68%; p < 0.001). This RR corresponds to a protection of 60% (95% CI: 32%–76%). When controlling for behavioural factors, including sexual behaviour that increased slightly in the intervention group, condom use, and health-seeking behaviour, the protection was of 61% (95% CI: 34%–77%).
Conclusion
Male circumcision provides a degree of protection against acquiring HIV infection, equivalent to what a vaccine of high efficacy would have achieved. Male circumcision may provide an important way of reducing the spread of HIV infection in sub-Saharan Africa. (Preliminary and partial results were presented at the International AIDS Society 2005 Conference, on 26 July 2005, in Rio de Janeiro, Brazil.)
The first trial of male circumcision for reducing the risk of HIV finds significantly lower new cases in the treatment group.
==== Body
Introduction
Male circumcision (MC) is associated with various cultural factors, including religious sacrifice, rites of passage into adulthood, and the promotion of hygiene. The earliest documentary evidence for circumcision is from Egypt. Tomb artwork from the Sixth Dynasty (2345–2181 B.C.) shows circumcised men, and one relief from this period shows the rite being performed on a standing adult male. Genesis (17:11) places the origin of the rite among the Jews in the age of Abraham, who lived around 2000 B.C.
Presently, MC practices in Africa are varied. Whereas men in Muslim countries are circumcised, as in North Africa or a large part of West Africa, in other societies the prevalence of MC depends on other cultural factors, such as changes that occurred under colonization. In countries such as Cameroon and the Democratic Republic of Congo, which are predominately non-Muslim, most men are circumcised [1–3]. In Kenya, where only a minority of men are Muslims, men in all tribes except the Luo practice MC [4].
The first paper suggesting a protective effect of MC against HIV infection was published in 1986 [5]. Since then, many observational studies have been published, some of which have observed that most men living in East and southern Africa, the regions with the highest prevalence of HIV, are not circumcised [1–3]. A majority of these observational studies are cross-sectional, and a minority are prospective [6–11]. A systematic review and meta-analysis found that in sub-Saharan Africa MC is associated with a significantly reduced risk of HIV infection among men, with an adjusted relative risk of 0.42 (95% CI: 0.34%–0.54%) [12].
All of these studies were based on observational data, and, in the absence of experimental studies, a causal relationship between MC and protection against HIV infection could not be determined [13]. Direct experimental evidence is needed to establish this relationship and, should a protective effect of MC be proven, to convince public health policy makers of the role of MC in reducing the spread of HIV [7,13,14].
The primary objective of this study was to determine the impact of MC on the acquisition of HIV by young men through a randomized, controlled, blindly evaluated intervention trial. The secondary objective was to assess the role of behavioural factors known to be associated with HIV serostatus in explaining the possible impact. This study was conducted in the Gauteng province of South Africa, where HIV prevalence among pregnant women was estimated to be 29.6% in 2003 [15]. According to an earlier study in the research site area, 59% (95% CI: 55%–63%) of uncircumcised men said that they would be circumcised if it reduced their chance of acquiring HIV and STDs [16].
Methods
General Presentation
A randomized, controlled, blindly evaluated intervention trial was carried out in Orange Farm and surrounding areas, a semi-urban region close to the city of Johannesburg. The recruitment of participants took place in the general population from July 2002 to February 2004. Information about the trial was disseminated in the community through meetings during the recruitment period. Precise oral and written information was delivered at the investigation centre to potential participants during a pre-screen visit. Participants were then informed that the impact of MC on the acquisition of sexually transmitted infections (STIs), including HIV, is not known. A minimum of 3 d after the pre-screen visit, potential participants were screened for eligibility. Potential participants with genital ulcerations were temporarily excluded until successful treatment. The inclusion and exclusion criteria are listed in Table 1. The participants received a total of 300 South African Rand as compensation (1 South African Rand ~ 0.12 Euro). The protocol, the consent form, and the participant information sheet are provided as Text S1–S3.
Table 1 Inclusion and Exclusion Criteria
Randomization
At the end of the screen visit, following screening and written consent, participants were divided into two groups, using sealed envelopes. Participants requested to participate actively in the random assignment. Consequently, each participant was invited by the manager of the centre to choose an envelope containing the group name from a basket of ten envelopes. After each randomization, a new envelope was added to the basket. This added envelope was taken sequentially from a set of envelopes pre-prepared in such a way that each set of envelopes contained five for the “Control” and five for the “Intervention” arm. Participants of the intervention group were offered to be circumcised within a week. Participants of the control group were asked to wait until the end of the trial before being offered to be circumcised.
Follow-Up and Data Collection
After the screen visit, which took place at month 1 (M1), the three follow-up visits took place at the end of M3, M12, and M21. The M3 visit was designed to study the possible impact of surgery on HIV acquisition as a result of sexual activity during the healing phase following circumcision or contamination during surgery. These three follow-up visits defined three sequential periods, M1–M3, M4–M12, and M13–M21, with expected durations of 3, 9, and 9 mo, respectively. The duration of these periods was measured in days from the dates of the visits, the day after the end of a period being the beginning of the next period.
A participant lost to follow-up was defined as a participant who had not completed a planned visit in the 2 mo following the planned date of this visit and who did not complete any further visit. A missing visit was defined as a visit not completed prior to a completed visit.
At each of the four visits, each participant was invited to answer a face-to-face questionnaire, to provide a blood sample, and to have a genital examination and an individual counselling session. The questionnaire allowed for collection of data on background characteristics and reported sexual behaviour. The last section of the questionnaire allowed for the description of all sexual partnerships over the previous 3 mo for the M3 visit and over the previous 12 mo for all other visits. This section allowed each participant to describe the number of sexual contacts, the date of first and last sexual contact, the frequency of condom use (never, sometimes, always), and the type of partnership (spousal or non-spousal), a spousal partner being defined as a sexual partner with whom the respondent is married or living as married. Characteristics of sexual behaviour during the 9-mo periods M4–M12 and M13–M21 were determined from this section, using the dates of first and last sexual contact of each sexual partner. The genital examination was performed by a trained nurse who recorded the circumcision status and took a blood sample from each participant. Blood samples were tested for syphilis and HIV-1.
The counselling session (15–20 min) was delivered by a certified counsellor and focused on information about STIs in general and HIV in particular and on how to prevent the risk of infection. During this session, participants were encouraged to attend voluntary counselling and testing in a public clinic located 200 m away from the investigation centre or in a voluntary counselling and testing (VCT) centre funded by the project and located in the same building as the investigation centre. Condoms were provided in the waiting room of the investigation centre and were also provided by the counsellor. Participants who had symptoms of STIs, as assessed by the nurse during the genital examination, or who tested positive for syphilis were treated at the local clinic or by doctors working for the project. A specific programme for prevention of opportunistic infections and delivery of antiretroviral treatment, if required, was put in place at the VCT centre to assist participants who attended VCT and who tested positive for HIV. The arrangement will remain in place until the public sector programme becomes operational in the area.
The standard of care in South Africa at the beginning of the trial in July 2002 included VCT but not access to antiretroviral therapy. With the formal introduction of access to antiretroviral therapy in 2004, there were increased efforts to encourage participants to attend VCT and referrals to appropriate facilities were instituted. In this context, it was decided to include participants independent of VCT attendance. Consideration for making HIV testing compulsory for participation in the trial or recruiting only those who tested HIV-negative would certainly lead to stigmatization, and the investigators considered that the whole concept of VCT was that it should be voluntary. They considered it unethical to inform participants of their HIV status without their permission, even if they thought that participants should be aware of their HIV status. They also considered it unethical to deter from participating in the study potentially at-risk men who did not want to know their HIV status. Indeed, HIV-positive participants would benefit from the trial: (a) by receiving counselling at each visit, (b) by undergoing clinical examination and syphilis testing, and (c) by having a medicalized circumcision that could possibly protect them or their sexual partners against other STIs or even against re-infection by HIV.
Male Circumcision
The median (interquartile range [IQR]) duration between randomization and MC was one day (0–2). The circumcisions were performed by three local general practitioners in their surgical offices. The general practitioners were experienced in MC practices. The cost of each circumcision was 300 South African Rand and was paid for by the project. The procedure was standardized and used the forceps-guided method, as is widely practiced in South Africa, and was reviewed by the Department of Urology, University of the Witwatersrand Medical School, South Africa.
Quality of the Data, Blinding, Confidentiality, and Data Management
To ensure confidentiality, participants' files were kept in a locked room at the centre and each participant received a number that was used to identify all documents related to that person. To ensure blinding of study personnel, the randomization group information was not available to the personnel in charge of counselling or collecting information in the centre during the participants' visits.
Questionnaires were checked at the end of each interview. Participants failing to turn up for any follow-up visit were visited at home by trial staff, who encouraged them to come for the follow-up visits or ascertained the reasons for dropping out.
Laboratory results were stored in a database that was independent of the one used to store the information related to each participant. During the study, no HIV results were available to the investigation centre or to the investigators, apart from the statistician in charge of the interim analysis.
Laboratory results and data collected from questionnaires were entered twice in a database (Microsoft Access, Redmond, Washington, United States) by different people. The two entries were compared, and discrepancies were corrected. The data were then re-checked for inconsistencies using the source documents. After the data had been cleaned, they were imported into the statistical package SPSS for Windows version 8 (SPSS, Chicago, Illinois, United States) and R (version 2.0.1) for analysis [17].
Laboratory Procedures
Following the interview, a trained nurse collected whole blood samples in the investigation centre. One EDTA blood tube of 10 ml of venous blood was taken and immediately centrifuged at 400 g for 10 min, and five aliquots of plasma were frozen at −20 °C. The samples were identified only by the participant number and transported each week to the laboratory, where they were stored at −70 °C and tested.
An ELISA screen (Genscreen HIV1/2 version 2, Bio-Rad, France) and two ELISA confirmatory tests (Wellcozyme HIV recombinant, Abbott Murex, Dartford, United Kingdom, and Vironostika HIV Uni-Formm II plus O, bioMérieux, Boxtel, Netherlands) were used to test plasma for HIV-1 infection. Samples that were positive on all three ELISAs were regarded as “positive” and all others as “negative” [18].
Ethics
The research protocol was reviewed and approved by the University of Witwatersrand Human Research Ethics Committee (Medical) on 22 February 2002 (protocol study no. M020104). The trial was also approved by the Scientific Commission of the French National Agency for AIDS Research (ANRS; protocol study no. 1265; 2002, decision No. 50) and obtained authorization from the City of Johannesburg, Region 11, on 25 February 2002. A Data and Safety Monitoring Board was responsible for analyzing adverse events and for deciding on the results of the interim analysis.
Adverse Events
Adverse events (AEs) were documented and analyzed for all participants, including those who were HIV-positive at randomization. These AEs related to surgery, and that occurred in the first month post-surgery, were reported by the practitioners using a specific form. In addition, at each visit to the centre the nurse completed a questionnaire after the genital examination to record adverse events. During home visits for missing participants, any deaths were recorded.
Sample Size and Interim Analysis
The total sample size was initially calculated to be 2,580 HIV-negative participants in order to obtain a power of 80% to detect a 50% reduction in the proportion of HIV infection between the groups at a 5% significance level, assuming an HIV incidence of 2.2 per 100 person-years (py) in the control group. This number, calculated using Fisher's exact test, was increased to 3,035 to account for 15% of participants lost to follow-up. An interim analysis was planned for when all the M12 visits had been completed, and this was conducted blind with the database obtained on 29 November 2004. At the time of the interim analysis, the total follow-up included an estimated 63% of the total number of py that would have been collected at the end of the study, leading to a threshold value of 0.0095, as determined by the Lan-DeMets alpha-spending function method [19].
Statistical Analysis
While participants with a HIV-positive test at M1 were followed in the same way as the other participants, they were excluded from the statistical analysis. HIV status was considered as censored data with time being continuous, observed in a grouped form (at the end of each period), with non-uniform duration of periods. These data were modelled using a piecewise exponential, proportional hazards model in which the baseline hazard is constant in each period. This theoretical model allows the precise duration between each visit and time-dependent covariate to be taken into account. It was implemented by running a Poisson log-linear model on a dataset composed of lines corresponding to the periods M1–M3, M4–M12, and M13–M21, in which the participant stayed HIV-negative or became HIV-positive [20–22]. Consequently, in this dataset, each individual was represented by a maximum of three lines. This type of model gives an incidence rate and incidence rate ratio (RR) of HIV infection among men of the intervention group in comparison with men of the control group. The protection against HIV infection was calculated as 1 − RR.
At the interim analysis, the RR was 0.37 in the intervention group, as compared with the control group, with a p value of 0.00073, below the threshold value. The Data and Safety Monitoring Board advised the investigators to interrupt the trial and offer circumcision to the control group, who were then asked to come to the investigation centre, where MC was advised and proposed. The database corresponding to planned visits up to 30 April 2005 was then analyzed, and the results are presented in this paper. Because the study was interrupted, some participants did not have a full follow-up on that date, and their visits that were not yet completed are described as “planned” in this article.
Adjusted rates and RRs were obtained by taking into account covariates that were calculated for each period when they were time-dependent. Three nested models were developed. The model-1 included the period number, which was included as categorical variables, with the logarithm of the duration of exposure in each period in days as an offset. In the model-2, the calendar period of recruiting and background characteristics of the participants were added. In the model-3, behavioural time-dependent covariates, characterizing the behaviour of participants during each period, were also added.
The background characteristics of the participants considered were age (less than or equal to 21 y, more than 21 y), religion (Catholic or Protestant, African traditional, other), ethnic group (Zulu, Sotho, other), and alcohol consumption in the past month. The five reported sexual behaviour covariates considered were, for each period of follow-up, being at-risk behaviour (defined as having at least one sexual contact unprotected by condom), having a spousal partner, the number of non-spousal sexual partners, the number of sexual contacts, having at least one relationship with only one sexual contact. In addition, health-seeking behaviour was characterized by at least one visit to a clinic for a genital problem during the 12-mo period prior to a visit to the centre.
Additional analyses were also performed. (a) The impact of the intervention was assessed among those having completed their M21 visit. (b) The impact of the intervention on participants who were 1 mo or more late to at least one follow-up visit or missed one follow-up visit was compared with the impact of the intervention on other participants by testing the corresponding interaction term between this factor and the randomization group. (c) To analyze the impact of the 6-wk period of abstinence, the analysis was repeated with the duration of the period M1–M3 reduced by 42 d in the intervention group. Forty-two days was the median (IQR = 28–56) interval between MC and first sexual contact reported by sexually experienced participants of the intervention group. (d) The effects of MC across the ethnic groups were studied by assessing this impact among the two major ethnic groups of this study (Zulus and Sothos) and by testing the corresponding interaction term. (e) Finally, while all analyses were performed in intention-to-treat, a per-protocol analysis was performed using the circumcision status observed at each visit.
Six comparisons of the behavioural factors for each of the periods M4–M12 and M13–M21 were performed. Independence of behavioural categorical data between the randomization groups was tested using Fisher's exact test, and the Kruskal-Wallis test was used for quantitative behavioural variables. Assuming that these comparisons were independent, and to keep the overall risk of type I error equal to 0.05, the level of significance was set as 1.00 − 0.95 1/6 = 0.0085.
Results
Table 2 gives the baseline characteristics for the HIV-negative participants. The median age (IQR) was 21.0 y (19.6–22.5). Most of the participants had completed the primary level of education. Very few were married or living as married, and about half were at-risk behaviour. Figure 1 shows the trial flowchart. A total of 3,274 men participated in the trial. There were 146 (prevalence 4.5%) HIV-positive participants at randomization. The difference in size between the intervention and control group was 34 (1,620 versus 1,654).
Figure 1 Trial Profile
This figure describes the state of the trial corresponding to planned visits up to 30 April 2005. HIV-positive and HIV-negative participants were randomized. All were followed, but only participants HIV-negative at randomization were analyzed and are represented in the three follow-up visits of the figure. After randomization, the participants could attend the 3-mo visit, miss it, or be excluded from follow-up (death or loss to follow-up). The non-excluded participants who attended the 3-mo visit could then attend the 12-mo visit, miss it, or be excluded (death or loss to follow-up). The non-excluded participants of the 12-mo visit could then attend the 21-mo visit, be excluded (death or loss to follow-up) or were planning to attend the 21-mo visit but had not yet done so, because of the interruption of the trial.
*, did not come for the scheduled visit (refused, withdrew, moved away or died); **, no blood sample
Table 2 Baseline Characteristics of HIV-Negative Men Enrolled in the Trial
Among the 3,128 HIV-negative participants at randomization, the visits at M3, M12, and M21 took place at (median; IQR) 3.0 (3.0–3.2), 12.0 (11.9–12.1), and 20.9 mo (20.9–21.2) after randomization, respectively. The mean (IQR) follow-up was 18.1 mo (13.0–21.0).
The fraction of participants lost to follow-up was 8.0 % (251/3128), with 6.5% (100/1546) in the intervention group and 9.5% (151/1582) in the control groups (p = 0.0016, Fisher's exact test). Among the participants lost to follow-up at the visit M12 or M21, none (0/124) were HIV-positive at their previous completed visit.
During the study, 20 and 49 participants acquired HIV infection in the intervention and control groups, respectively, corresponding to incidence rates (95% CI) of 0.85 per 100 py (0.55–1.32) and 2.1 per 100 py (1.6–2.8) in the intervention and control groups, respectively. Using model-1, the RR of HIV infection for the intervention group in comparison with the control group was 0.40 (0.24–0.68), p = 0.00059 (Table 3). This RR corresponds to a protection of 60% (32–76) against HIV infection. This result is equivalent to saying that during the period M1–M21 the intervention prevented six out of ten potential infections.
Table 3 Characteristics of the Follow-Up Period
When considering only those participants who completed their M21 visit, the RR was 0.38 (0.22–0.67), p < 0.001. In comparison with the others, those who were 1 mo or more late to at least one follow-up visit or missed one follow-up visit (1178/3128; 37.7%) had the same risk of HIV infection (RR = 1.06; 0.65–1.73; p = 0.82) and were not differently protected by MC (p = 0.69). When reducing the M1–M3 period by 42 d in the intervention group, the RR was RR = 0.43 (0.26–0.73), p = 0.0016, a value close to the RR obtained in the intention-to-treat analysis. This indicates that the 6-wk period of abstinence plays a minor role in explaining the effect of the intervention during the period M1–M21. Among the two major ethnic groups of the participants, Zulus (n = 1,109) and Sothos (n = 1,506), the RR was 0.60 (0.25–1.41), p = 0.24, and 0.42 (0.20–0.88), p = 0.022, respectively. These two RRs were not significantly different (p = 0.55). The per-protocol analysis gave RR = 0.24 (0.14–0.44), p < 0.001, a value lower to the RR obtained in the intention-to-treat analysis. The difference of the results given by the two analyses is at least partly explained by the cross-overs. In the intervention group, 6.5% (93/1432) were not circumcised at M3, and in the control group, 10.3% (114/1105) were circumcised at M21 (Figure 1).
For the periods M1–M3, M4–M12, and M13–M21, the number of HIV infections was two, seven, and 11 in the intervention group and nine, 15, and 25 in the control group. The RR for each of these periods is given in Table 3. In the period M1–M3, there was an RR of 0.23 close to the significance level, which was slightly higher when taking into account the 42 d of abstinence (RR = 0.37; 0.08–1.72; p = 0.21).
Using model-2, an RR was found similar to that obtained with model-1: 0.38 (0.23–0.65), p < 0.001. This result is attributable to the randomization process, which distributed the characteristics equally between the intervention and control groups.
Of the five reported sexual behavioural factors, all were higher in the intervention group than in the control group during the period M4–M12, and four out of five were higher during the period M13–M21. Only the mean number of sexual contacts showed statistically significant differences during the period M4–M12 (5.9 versus 5.0, p < 0.001) and during the period M13–M21 (7.5 versus 6.4, p = 0.0015). The proportion of participants attending a clinic for a genital problem in the 12 mo prior to M12 was lower in the intervention group than in the control group (4.7% versus 7.2%, p = 0.0067).
Using model-3, the RR, adjusted on behavioural characteristics, reported by participants during the follow-up is similar to the RR obtained with model-1 (Table 4). This last result indicates that the protective effect of the intervention is not attributable to the change of reported behaviour associated with the intervention and shows that adjustment for potential confounders has little effect on the association of MC and HIV incidence.
Table 4 Multivariate RRs of HIV Incidence
Figure 2 shows the fitted infection-free probability as a function of time and of randomization. Table 5 describes the 60 (3.8%) AEs that were reported during surgery or in the first month following surgery among 1,568 MCs performed in the intervention group, HIV-positive at randomization included. The proportion of AE was higher among those who were HIV-positive at randomization, and the difference is close to significance (p = 0.056, Fisher's exact test). At M3, 98.5% of those who were circumcised (HIV-positive at randomization included), were “very satisfied” with the result of the circumcision. Adverse events recorded at the end of the follow-up (M21) are described in Table 6.
Figure 2 Infection-Free Probability As a Function of Time and of Randomization
This figure represents the infection-free probability using a piecewise exponential distribution with boundaries at M3, M12, and M21 obtained with a Poisson log-linear model (see text). Each segment of exponential has been fitted to the data in each period for each randomization group. The 95% confidence intervals have been represented in the middle of each period. x/y is the number of HIV infections observed in each period (x) and the number of persons at the beginning of the period (y).
Table 5 Adverse Events during Surgery or in the First Month following Surgery among Those Having Been Randomized in the Intervention Group, as a Function of HIV Status at Randomization
Table 6 Adverse Events at the End of the Follow-Up (M21) among Those Having Been Randomized in the Intervention Group, As a Function of HIV Status at Randomization
Home visits for late participants revealed 16 deaths among participants (HIV-positive at randomization included), of whom six had been circumcised, but examination of death certificates, reports from doctors who carried out the MC, interviews with relatives, and timing of these deaths revealed no deaths related to MC. The mortality rate from the South African Census 2001 data [23] in the age groups 15–19 and 20–24 for the black population of the Gauteng province was 2.4 and 3.9 per 1,000 per year. These figures lead to an estimate of 3.5 per 1,000 per year at the mean age (21.0 y) of our participants. In turn, this value leads to an estimated number of deaths of 15.7 using the mean follow-up, which is close to the number of deaths observed in our trial.
Discussion
This study provides the first experimental evidence of the efficacy of MC in protecting men against HIV infection. It was conducted in a general population, and it is the first randomized control trial testing the impact on health of MC. The demonstration in this study of a causal association between HIV infection and MC is consistent with protection suggested by meta-analyses of observational studies [12] but with a higher protective effect. This difference can be explained, at least partly, by the effect of bias and confounding factors associated with cross-sectional studies. High values ranging from 0.12 to 0.29 of protective effect of MC have been reported in prospective studies conducted in high-risk groups [6,8–11]. Our study is also the first experimental study demonstrating that surgery can be used to prevent an infectious disease. In addition, this finding is an a posteriori proof of the use of MC to improve hygiene in the common meaning of not being infected.
This study has some limitations. It was conducted in one area in sub-Saharan Africa and, therefore, may not be generalizable to other places. Nevertheless, because of the similar route of transmission of HIV in sub-Saharan Africa and because observational studies from various areas of sub-Saharan Africa have shown an association between HIV status and MC [12], the result of this trial is applicable to all of sub-Saharan Africa with some degree of confidence.
Even though some participants were lost during the follow-up, and the loss to follow-up rate was greater than the event rate, the impact of missing participants on the overall results of this study is likely to be small not only because the loss to follow-up was small for a cohort study conducted in a general population, but also because those who were late for at least one follow-up visit were protected by MC just as the other participants. The reason for this loss to follow-up was a result of participants moving from the area or being unreachable, and not a result of HIV infection.
Because the Data and Safety Monitoring Board recommended to stop the trial after the intermediate analysis, it was not possible to follow all the participants as initially planned, and, as a consequence, only those participants recruited at the beginning had a full follow-up. This potential bias was taken into account by adjusting the analysis for the recruitment period; such an adjustment cannot fully account for the confounding effect associated with partial follow-up. When restricting the analysis to those participants who had a full follow-up, the intervention had an effect that was similar in size and significance, suggesting that this potential bias had a negligible impact.
A specific survey was implemented after the end of the recruiting period in order to assess the satisfaction of the results of the randomization. Of the participants, 65.3% said they were happy. However, the results also showed that a limited number of participants (7.5%), strongly unhappy with their group of randomization, were allocated and recorded in the other group. They were analyzed in their randomization group in the intention-to-treat analysis. The findings were confirmed by the person in charge of randomization. This factor contributed to increase the cross-over, which remained low, and to dilute the measure of the effect of the intervention, which remained high.
Another limitation concerns the timescale of this study. Participants were followed up for a short period of time, and, therefore, this study did not explore the long-term protective effect of MC.
The protective effect of MC on HIV infection was unchanged when controlling for sexual behaviour, including condom use, which was taken into account when defining those at-risk behaviour, the period of abstinence in the intervention group following MC, and heath-seeking behaviour, which was considered because treatment of STIs can have an effect on HIV acquisition [24]. This shows that these factors play a minor role in explaining the protective effect of MC on HIV infection. The reasons for this protective effect of MC on HIV acquisition have to be found elsewhere, and several direct or indirect factors may explain this [25]. Direct factors may be keratinization of the glans when not protected by the foreskin, short drying after sexual contact, reducing the life expectancy of HIV on the penis after sexual contact with an HIV-positive partner, reduction of the total surface of the skin of the penis, and reduction of target cells, which are numerous on the foreskin [26]. Indirect factors may be a reduction in acquisition of other STIs, which in turn will reduce the acquisition of HIV. Our study does not allow for identification of the mechanism(s) of the protective effect of MC on HIV acquisition.
The first and obvious consequence of this study is that MC should be recognized as an important means to reduce the risk of males becoming infected by HIV. As shown by our study, MC is useful and feasible even among sexually experienced men living in an area with high HIV prevalence. Indeed, in our study the intervention delivered by local general practitioners resulted in a limited and reasonable number of adverse events and did not lead to an increase in deaths. In addition to the protective role in men, MC will indirectly protect women and, therefore, children from HIV infection because if men are less susceptible to HIV acquisition, women will be less exposed. Moreover, MC may also be protective against male-to-female HIV transmission, but this will require further investigation [7]. The role that women can play in promoting MC is potentially important. If women are aware of the protective effect of MC, this awareness could, in turn, have an impact on the prevalence of MC by encouraging males to become circumcised.
It was found that the protective effect of MC is high. MC provides a degree of protection against acquiring HIV infection equivalent to what a vaccine of high efficacy would have achieved. Consequently, the authors think that MC should be regarded as an important public health intervention for preventing the spread of HIV. MC could be incorporated rapidly into the national plans of countries where most males are not circumcised and where the spread of HIV is mainly heterosexual. This is even more important at a time when no vaccine or microbicides are currently available and when delivering antiretroviral treatments under WHO guidelines will have only a small impact on the spread of HIV [27]. In addition, MC is an inexpensive means of prevention, performed only once, and men can be circumcised over a wide age range, from childhood to adulthood.
The potential impact of prevention programmes based on MC is difficult to assess at population level and requires modelling. From the results of this study and of the meta-analysis quoted above, it can be predicted that widespread MC could lead to a strong reduction of the spread of HIV. The availability of a simple and ancient practice with a high potential effect on the spread of HIV is remarkable and should encourage decision makers to take MC into consideration as policy. Because most of southern and East Africa is concerned, the number of HIV infections that could be avoided by the widespread implementation of MC is high.
There are potential risks in promoting MC as way of reducing the risk of HIV infection. MC can be performed under poor hygienic conditions, leading to not only infection, bleeding, and permanent injury, but also HIV infection from non-sterilized instruments, and possible death if appropriate treatment of sequelae is not provided. In the healing period, sexually active men are likely to be at a higher risk of HIV infection, and this risk should not be underestimated. MC does not provide full protection and, if perceived as full protection, could lead to reduction of protection of men who, for example, decrease their condom use or otherwise engage in riskier behaviour. It was found that the intervention group had significantly more sexual contacts. While the protective effect of circumcision remained despite this increased risk, this should be a concern when considering implementation of circumcision as a means of preventing HIV infection. Finally, there is the danger of confusing MC with female circumcision, and that promotion of MC could be used by defenders of female circumcision to defend this practice.
Acceptability studies of the use of MC as a prevention measure against the spread of HIV have been conducted in South Africa [16,28], Kenya [29,30], Zimbabwe [31], and Botswana [32]. These studies, in which most of the uncircumcised African men expressed interest in becoming circumcised if performed safely and affordably, highlighted the potential of MC as a population-level intervention to reduce HIV spread. MC is a not a universal cultural practice, and cultural practices can be barriers in policy considerations. However, there are examples showing that the prevalence of MC can be changed. For example, in South Korea 50 years ago, almost no men were circumcised; today some 85% of Korean men 16–29 y old are circumcised [33].
The experimental demonstration of the protective effect of MC on the acquisition of HIV emphasizes the role of MC in explaining the heterogeneity of HIV prevalence in sub-Saharan Africa. From a multi-site study conducted in four African countries, MC, together with sexual behaviour, has been posited as an important factor in the heterogeneity of HIV prevalence in sub-Saharan Africa [34]. This role is confirmed and reinforced by the findings of the present study.
Supporting Information
Text S1 Effect of Medicalized Male Circumcision on the Incidence of HIV, Herpes Simplex Virus 2, and Genital Ulcerations
(204 KB PDF).
Click here for additional data file.
Text S2 Consent Form
(29 KB PDF).
Click here for additional data file.
Text S3 Participant Information Sheet
(45 KB PDF).
Click here for additional data file.
Patient Summary
Background
HIV/AIDS is one of the greatest threats to health worldwide. More than 3 million people died of AIDS last year, and about 5 million others became infected with HIV, bringing the total number of people living with the infection to nearly 40 million. The situation is particularly severe in Africa, which has 10% of the world's population but two-thirds of the world's people with HIV. In many African tribal groups, men are circumcised, usually in late childhood or early adolescence, and this is an important part of their cultural identity. In other African ethnic groups, men are not circumcised. By the late 1980s, researchers noticed that HIV infection rates were lower in those tribes where men were circumcised. But it was not clear whether it was circumcision itself or some other difference in behaviour between the circumcised and uncircumcised groups that gave some protection to the circumcised men against getting HIV.
What Did The Researchers Do?
The researchers wanted to find out whether circumcising men could reduce their chance of becoming infected by HIV. They offered young, sexually active, heterosexual, uncircumcised men in Johannesburg, South Africa, the chance to have the operation. They explained that half of those who came forward would be circumcised right away (the “treatment group”) and the other half would be circumcised 21 months later (the “control group”). Some 3,000 men joined the study. The group that each man was put into was decided at random. The plan was that all the men would visit the research clinic four times during this 21-month period, and that they would be tested for HIV each time. However, after 14 months, the number of new infections in the control group (49) was so much greater than the number in the treatment group (20) that it was considered unethical to continue the study. (The men in the control group were told they could be circumcised without any further delay.)
What Do These Findings Mean?
Infections were 60% fewer in the treatment group, which seems to indicate that circumcised men are much less likely to become infected with HIV when having sex with infected women. In communities where HIV is common, circumcision may prove to be a valuable tool for reducing men's risk of getting infected. However, as with most studies, criticisms could be made of some aspects of the methods used, and more research is needed before we can be sure. We must also remember that circumcised men can still become infected, even though the risk might be lower. They should still take other steps to prevent themselves from getting HIV.
Where Can I Get More Information Online?
The United Nations health agencies, including the WHO and UNAIDS, issued a statement when this research was first presented at a meeting in Brazil in July 2005:
http://www.who.int/mediacentre/news/releases/2005/pr32/en/
UNAIDS (http://www.unaids.org) has information about the state of the HIV/AIDS epidemic and prevention strategies. It produces an annual report and has documents on a wide range of topics. The Q&A documents are particularly useful:
http://www.unaids.org/EN/resources/questions_answers.asp#II
Many organizations provide information on AIDS prevention—for example, the Terrence Higgins Trust:
http://www.tht.org.uk
AEGIS is the world's largest searchable database on HIV and AIDS:
http://www.aegis.com
The authors thank all those who agreed to take part in this study, to answer the questions put to them, and to provide blood samples. The authors would like to thank Reathe Rain-Taljaard for her management support and assistance in this project, as well as Gaph Sipho Phatedi for his management of the recruitment process. The authors would like to thank the general practitioners who have performed the MCs for this study (Dr. Bhekuyise Gwala, Dr. George Shilaluke, and Dr. Dumiso Zulu), and Dr. Sergio Carmona for monitoring the MCs. The authors would like to thank Goliath Gumede for the clinical investigation and Zodwa Nkosi for interviewing all the respondents. They would also like to thank Bongiwe Klaas for the data capture, Mabel Hunter and the recruitment staff and all the assistants (Cynthia Dlamini, Sidwell Dumisi, Benjamin Masitenyane, Robert Matodzi, Tsietsi Mbuso, Anthony Motha, Sibongiseni Mpetsheni, Jabulani Nhlapo, Joseph Ntsele, Male Chakela, Audrey Tshabalala, Donald Mashamba, and Nkululeko Nhlapo) for their cooperation and support. Ewalde Cutler, Lesley Short, Moses Mashiloane, Beulah Miller, Beverley Singh, Sarah Hloma, and the HIV serology laboratory of the National Institute for Communicable Diseases, Johannesburg, South Africa, provided excellent technical assistance in regard to the laboratory testing and administration. The authors would like to thank Brian Williams, Philippe Aegerter, Phuong Pham, and Jean-Christophe Thalabard for their useful comments on an earlier draft of this manuscript.
The study was funded by ANRS, Paris, France; the National Institute for Communicable Diseases, Johannesburg, South Africa; and the Institut National de la Santé et de la Recherche Médicale, Paris, France. JST received support from SIDACTION, Paris, France. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Data Safety Monitoring Board: Peter Cleaton-Jones, Mohamed Haffejee (University of Witwatersrand, South Africa), and Jonathan Levin (MRC, South Africa)
This trial has been registered in http://www.clinicaltrials.gov under the number NCT00122525.
Citation: Auvert B, Taljaard D, Lagarde E, Sobngwi-Tambekou J, Sitta R, et al. (2005) Randomized, controlled intervention trial of male circumcision for reduction of HIV infection risk: The ANRS 1265 trial. PLoS Med 2(11): e298.
Abbreviations
AEadverse event
IQRinterquartile range
M[number]month [number]
MCmale circumcision
pyperson-year
RRrate ratio
STIsexually transmitted infection
VCTvoluntary counselling and testing
==== Refs
References
Bongaarts J Reining P Way P Conant F The relationship between male circumcision and HIV infection in African populations Aids 1989 3 373 377 2502151
Caldwell JC Caldwell P The African AIDS epidemic Sci Am 1996 274 62 63 66 68 8607017
Moses S Bradley JE Nagelkerke NJ Ronald AR Ndinya-Achola JO Geographical patterns of male circumcision practices in Africa: Association with HIV seroprevalence Int J Epidemiol 1990 19 693 697 2262266
Auvert B Buve A Lagarde E Kahindo M Chege J Male circumcision and HIV infection in four cities in sub-Saharan Africa Aids 2001 15 S31 40 11686463
Fink AJ A possible explanation for heterosexual male infection with AIDS N Engl J Med 1986 315 1167
Lavreys L Rakwar JP Thompson ML Jackson DJ Mandaliya K Effect of circumcision on incidence of human immunodeficiency virus type 1 and other sexually transmitted diseases: A prospective cohort study of trucking company employees in Kenya J Infect Dis 1999 180 330 336 10395846
Gray RH Kiwanuka N Quinn TC Sewankambo NK Serwadda D Male circumcision and HIV acquisition and transmission: Cohort studies in Rakai, Uganda. Rakai Project Team Aids 2000 14 2371 2381 11089626
Reynolds SJ Shepherd ME Risbud AR Gangakhedkar RR Brookmeyer RS Male circumcision and risk of HIV-1 and other sexually transmitted infections in India Lancet 2004 363 1039 1040 15051285
Cameron DW Simonsen JN D'Costa LJ Ronald AR Maitha GM Female to male transmission of human immunodeficiency virus type 1: Risk factors for seroconversion in men Lancet 1989 2 403 407 2569597
Telzak EE Chiasson MA Bevier PJ Stoneburner RL Castro KG HIV-1 seroconversion in patients with and without genital ulcer disease. A prospective study Ann Intern Med 1993 119 1181 1186 8239249
Mehendale SM Shepherd ME Divekar AD Gangakhedkar RR Kamble SS Evidence for high prevalence and rapid transmission of HIV among individuals attending STD clinics in Pune, India Indian J Med Res 1996 104 327 335 8996932
Weiss HA Quigley MA Hayes RJ Male circumcision and risk of HIV infection in sub-Saharan Africa: A systematic review and meta-analysis Aids 2000 14 2361 2370 11089625
Siegfried N Muller M Deeks J Volmink J Egger M HIV and male circumcision—A systematic review with assessment of the quality of studies Lancet Infect Dis 2005 5 165 173 15766651
Halperin DT Bailey RC Male circumcision and HIV infection: 10 years and counting Lancet 1999 354 1813 1815 10577659
Department of Health National HIV and syphilis antenatal sero-prevalence survey in South Africa 2003 2003 Pretoria (South Africa) Department of Health 18
Lagarde E Dirk T Puren A Reathe RT Bertran A Acceptability of male circumcision as a tool for preventing HIV infection in a highly infected community in South Africa Aids 2003 17 89 95 12478073
RDC Team R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing 2004 Available: http://www.R-project.org . Accessed 28 September 2005
Martin DJ Blackburn NK O'Connell KF Brant ET Goetsch EA Evaluation of the World Health Organisation antibody-testing strategy for the individual patient diagnosis of HIV infection (strategy III) S Afr Med J 1995 85 877 880 8545748
Lan KKG DeMets DL Discrete sequential boundaries for clinical trials Biometrika 1983 70 659 663
Frome EL The analysis of rates using Poisson regression models Biometrics 1983 39 665 674 6652201
Berry G The analysis of mortality by the subject-years method Biometrics 1983 39 173 184 6871346
Holford TR The analysis of rates and of survivorship using log-linear models Biometrics 1980 36 299 305 7407317
Dorrington R Moultrie TA Timaeus IM Estimation of mortality using the South African Census 2001 data 2004 University of Cape Town Center for Actuarial Research 88
Grosskurth H Gray R Hayes R Mabey D Wawer M Control of sexually transmitted diseases for HIV-1 prevention: Understanding the implications of the Mwanza and Rakai trials Lancet 2000 355 1981 1987 10859054
Szabo R Short RV How does male circumcision protect against HIV infection? BMJ 2000 320 1592 1594 10845974
Patterson BK Landay A Siegel JN Flener Z Pessis D Susceptibility to human immunodeficiency virus-1 infection of human foreskin and cervical tissue grown in explant culture Am J Pathol 2002 161 867 873 12213715
Auvert B Males S Puren A Taljaard D Carael M Can highly active antiretroviral therapy reduce the spread of HIV? A study in a township of South Africa J Acquir Immune Defic Syndr 2004 36 613 621 15097305
Scott BE Weiss HA Viljoen JI The acceptability of male circumcision as an HIV intervention among a rural Zulu population, KwaZulu-Natal, South Africa AIDS Care 2005 17 304 313 15832878
Bailey RC Muga R Poulussen R Abicht H The acceptability of male circumcision to reduce HIV infections in Nyanza Province, Kenya AIDS Care 2002 14 27 40 11798403
Mattson CL Bailey RC Muga R Poulussen R Onyango T Acceptability of male circumcision and predictors of circumcision preference among men and women in Nyanza Province, Kenya AIDS Care 2005 17 182 194 15763713
Halperin DT Fritz K McFarland W Woelk G Acceptability of adult male circumcision for sexually transmitted disease and HIV prevention in Zimbabwe Sex Transm Dis 2005 32 238 239 15788922
Kebaabetswe P Lockman S Mogwe S Mandevu R Thior I Male circumcision: An acceptable strategy for HIV prevention in Botswana Sex Transm Infect 2003 79 214 219 12794204
Kim DS Lee JY Pang MG Male circumcision: A South Korean perspective BJU Int 1999 83 28 33 10349412
Auvert B Buve A Ferry B Carael M Morison L Ecological and individual level analysis of risk factors for HIV infection in four urban populations in sub-Saharan Africa with different levels of HIV infection Aids 2001 15 S15 30 11686462
|
16231970
|
PMC1262556
|
CC BY
|
2021-01-05 10:39:16
|
no
|
PLoS Med. 2005 Nov 25; 2(11):e298
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020298
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 10.1371/journal.pmed.0020391SynopsisInfectious DiseasesEpidemiology/Public HealthHealth PolicyHIV/AIDSSexual HealthUrologyHIV Infection/AIDSInfectious DiseasesMedicine in Developing CountriesSexually transmitted infections - other than HIV/AIDSFirst Trial of Male Circumcision against HIV Synopsis11 2005 25 10 2005 2 11 e391Copyright: © 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.
Randomized, Controlled Intervention Trial of Male Circumcision for Reduction of HIV Infection Risk: The ANRS 1265 Trial
==== Body
Over 3 million people died of AIDS last year and about 5 million others became infected with HIV, bringing the total number of people living with the infection to nearly 40 million. The continuing rise in the number of new cases makes it a priority to investigate all possible measures that might reduce the risk of infection, particularly—but by no means only—in Africa, which has 10% of the world's population but two-thirds of the world's people with HIV.
In many African tribal groups, men are circumcised, usually in late childhood or as teenagers, and this is an important part of their cultural identity. In other African ethnic groups, men are not circumcised. From observational studies dating back to the 1980s, it has become clear that HIV infection rates are greater in those groups where men were not circumcised. It has, however, remained a matter of speculation as to whether it is circumcision itself or some other difference in behavior that has a protective effect. What has been needed to settle this question is a randomized controlled trial (RCT) of the use of circumcision as a preventive intervention.
Auvert et al. have completed the first such trial in the Orange Farm area, a semiurban region close to Johannesburg, South Africa. They offered young, heterosexual uncircumcised men the chance to have the operation, explaining that half of those who came forward would be circumcised straightaway and the others (the control group) 21 months later. Some 3,000 men joined the study. It was planned that all the men would visit the research clinic four times during this 21-month period, and that they would be tested for HIV each time. They were instructed not to have sex for six weeks after the operation, and asked at each clinic visit to provide detailed information about their sexual activity. However, after 17 months, the number of new infections in the control group (49) was so much greater than the number in the treatment group (20) that it was considered unethical to continue the study, which was thus terminated early. Since infections were 60% lower in the treatment group, we now have strong evidence favoring the use of circumcision as part of prevention efforts.
Preliminary results from the Orange Farm study were presented in July at the 3rd International AIDS Society Conference in Rio de Janeiro, where there was considerable interest in the findings, although some cautionary notes were also sounded. Some observers have expressed concerns that the researchers did not tell participants of their HIV status, although they gave them every encouragement to attend a counseling and testing clinic in order to find this out. These and other issues are discussed in our accompanying Editorial (DOI: 10.1371/journal.pmed.0020293) and in two Perspectives, one by Peter Cleaton-Jones (DOI: 10.1371/journal.pmed.0020287), chair of the ethics committee that approved the study, and the second by Nandi Siegfried (DOI: 10.1371/journal.pmed.0020393), lead author of a Cochrane systematic review of male circumcision for prevention of heterosexual acquisition of HIV in men.
The authors have called for the promotion of male circumcision as part of AIDS prevention efforts in Orange Farm and in other parts of Africa. However, others will take a more cautious view, believing that the positive results of this one study must be confirmed by further studies. Such trials are currently under way in Kenya and Uganda. We must also remember that adult circumcision carries risks, especially if performed by medical personnel or traditional healers without proper training. A further concern is that circumcised men, considering themselves to be “protected,” might be more likely to engage in unsafe sex. Research is also needed to find out whether male circumcision has a preventive effect only on female-to-male transmission, or whether it may also reduce male-to-female transmission or male-to-male transmission. In addition, it will be important to determine the mechanism by which circumcision exerts its apparent protective effect.
Many questions do therefore remain, but in the words of one of the study's peer reviewers, this first RCT may come to be regarded as “a landmark paper” in HIV prevention.
|
0
|
PMC1262557
|
CC0
|
2021-01-05 10:39:19
|
no
|
PLoS Med. 2005 Nov 25; 2(11):e391
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020391
|
oa_comm
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.