Mutagenesis Advance Access originally published online on December 2, 2006
Mutagenesis 2007 22(1):5-13; doi:10.1093/mutage/gel052
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Determination of genetic toxicity and potential carcinogenicity in vitro—challenges post the Seventh Amendment to the European Cosmetics Directive
The Medical School, University of Wales Swansea Singleton Park, SA2 8PP, UK 1 Safety and Environmental Assurance Centre, Unilever Sharnbrook, MK44 1LQ, UK
Genetic toxicology and its role in the detection of carcinogens is currently undergoing a period of reappraisal. There is an increasing interest in developing alternatives to animal testing and the three R's of reduction, refinement and replacement are the basis for EU and national animal protection laws the Seventh Amendment to the EU Cosmetics Directive will ban the marketing of cosmetic/personal care products that contain ingredients that have been tested in animal models. Thus in vivo tests such as the bone marrow micronucleus test, which has a key role in current testing strategies for genotoxicity, will not be available for this class of products. The attrition rate for new, valuable and safe chemicals tested in an in vitro-only testing battery, using the in vitro tests currently established for genotoxicity screening, will greatly increase once this legislation is in place. In addition there has been an explosion of knowledge concerning the cellular and molecular events leading to carcinogenesis. This knowledge has not yet been fully factored into screening chemicals for properties that are not directly linked to mutation induction. Thus there is a pressing need for new, more accurate approaches to determine genotoxicity and carcinogenicity. However, a considerable challenge is presented for these new approaches to be universally accepted and new tests sufficiently validated by March 2009 when the animal testing and marketing bans associated with the Seventh Amendment are due to come into force. This commentary brings together ideas and approaches from several international workshops and meetings to consider these issues.
 |
Introduction |
|---|
 Top Introduction Current tests and protocols...New tests and approachesNew insights into cancer...ConclusionsReferences |
|---|
Genetic toxicology and its role in the detection of carcinogens, is currently undergoing a period of reappraisal. There is an increasing interest in developing alternatives to animal testing and the three R's of reduction, refinement and replacement are the basis due to a variety of reasons including impending regulatory initiatives, new assessments of the performance of the accepted battery of in vitro tests and new knowledge in cancer biology. There is an increasing interest in developing alternatives to animal testing and the three R's of reduction, refinement and replacement are the basis for EU and national animal protection laws. The Seventh Amendment to the EU Cosmetics Directive 76/768/EEC, [http://eur-lex.europa.eu/LexUriServ/site/en/oj/2003/l_066/l_06620030311en00260035.pdf] will ban the marketing of cosmetic/personal care products that contain ingredients that have been tested in animal models. Thus in vivo tests such as the bone marrow micronucleus test, which has a key role in current testing strategies for genotoxicity, will not be available for this class of compounds. On the other hand, the REACH initiative (Registration, Evaluation and Authorization of Chemicals) [http://ec.europa.eu/environment/chemicals/reach.htm], will involve the assessment of tens of thousands of chemicals in common usage and if current approaches are used, could involve the use of hundreds of thousands of animals in screening for genotoxicity alone, unless improved in vitro testing strategies become accepted by the Regulators. This comes at a time when the standard battery of in vitro genetic-toxicology tests is being considered more critically. These assays, necessary for regulatory submissions, measure gene mutations and damage to chromosomes and are used to predict the carcinogenic potential of the compounds under test. When positive results are obtained in the initial in vitro assays of the battery, further in vivo genotoxicity, and potentially rodent carcinogenicity studies, may be undertaken to validate or negate the in vitro results. However, a recent analysis of published results on nearly 1000 chemicals (1
) has highlighted the high sensitivity, but poor specificity of the current in vitro genetic-toxicity test battery when compared to the outcome of rodent carcinogenicity studies. When a standard two or three in vitro test battery was performed, at least 70% of the non-carcinogenic agents tested (177) were scored as mutagenic or genotoxic. These findings have been confirmed by a separate FDA analysis (2). Assays using mammalian cells (e.g. chromosome aberration test, mouse lymphoma L5178Y tk+/tk– test) have the poorest specificity. This is of particular concern where in vivo work to follow-up in vitro positives will be prevented in the future by the Seventh Amendment of the Cosmetics Directive. The attrition rate for new, valuable and safe chemicals tested in an in vitro-only testing battery, using the in vitro tests currently established for genotoxicity screening, will greatly increase, once this legislation is in place. There is also a small number of in vivo genotoxins that appear not to be easily detected in vitro that could be missed by current all in vitro test batteries (3). In addition there has been an explosion of knowledge concerning the cellular and molecular events leading to carcinogenesis. This knowledge has not yet been fully factored into screening chemicals for properties that are not directly linked to mutation induction but are important in carcinogenic potential. New approaches are required to enhance the detection of such chemicals. With this increased understanding of the carcinogenic process has come the realization that, in some cases, chemically induced cancer in rodents may not be predictive of carcinogenic potential in humans. (4). Thus there is a pressing need for new, more accurate approaches to determine genotoxicity and carcinogenicity. However, a considerable challenge is presented for these new approaches to be universally accepted and for new tests to be sufficiently validated by March 2009 when the animal testing and marketing bans associated with the Seventh Amendment are due to come into force.
This Commentary brings together ideas and potential approaches from several international workshops and meetings considering the above issues, including a workshop hosted by Unilever in September 2005, a European Centre for the Validation of Alternative Methods (ECVAM) sponsored workshop held in Ispra in April 2006 (5) and the UK Environmental Mutagen Society (UKEMS) meeting held at the University of Warwick, March 2006.
|
Current tests and protocols for genotoxicity screening in vitro |
|---|
|
TopIntroductionCurrent tests and protocols...New tests and approachesNew insights into cancer...ConclusionsReferences |
|---|
Although in vitro genotoxicity tests have been established for several decades and protocols have become increasingly refined, the poor specificity of current mammalian cell-based assays suggests that there are intrinsic problems in both the cell lines and protocols used, in terms of accurately discriminating genotoxic carcinogens from non-genotoxic non-carcinogens. Recent workshops have explored stability of test materials, karyotypic stability of cell lines, effects of high cytotoxicities and the appropriateness of the metabolic activation systems used. All have revealed potential for generation of ‘false positive’ results.
Stability and quality of test materials
Genetic-toxicology tests may be carried out by some laboratories for stability of test compounds used in the media, but stability testing is not universal. A variety of culture media in common usage have the potential to oxidize a wide variety of chemicals to produce hydrogen peroxide (6,7). As hydrogen peroxide is a clastogen, chromosome damage and the induction of small colonies in the mouse lymphoma assay will ensue if sufficient amounts of hydrogen peroxide are generated, regardless of the inherent genetic toxicity (or lack of it) of the compound under test. It is clearly prudent to check for hydrogen peroxide production in the chosen test medium with new test chemicals, before initiating genetic-toxicity testing (5). Test compound quality may also have contributed to the generation of ‘false’ positive results for some compounds reported in the literature, in that there may have been mutagenic and/or clastogenic contaminants in batches of an otherwise non-genotoxic material.
Cell line stability and quality
The rodent cell lines used around the world for genetic-toxicity testing have been cultured and sub-cultured for many decades. It is highly likely that laboratories have obtained their particular cell lines from disparate sources, as well as sub-culturing them in-house for many years. When karyotyping of the cell lines of interest has been carried out over time it is clear that the karyotype of each cell line is drifting over time, with the appearance of new rearrangements, chromosomal deletions, inversions etc. (8). It is unclear how much this contributes to high false positive rates but it is essential that the impact of genetic drift on the performance of genotoxicity assays is understood.
The question arises as to whether one cell type, currently in widespread use, is more likely to give more accurate results than others? The available evidence suggests that most variation comes from the differing protocols that have been used in the past, although the CHL cell line may be more susceptible to clastogenic damage than CHO cells (9) and primary human peripheral lymphocytes may give fewer false positive results than V79, L5178Y and TK6 cells in in vitro micronucleus tests (5).
However, all mammalian cell types, including primary human lymphocytes give high rates of positive results with compounds that are unlikely to be genotoxic carcinogens.
Many of the established cell lines are now known to have deficiencies in genes involved with aspects of DNA repair, cellular defence systems, cell cycle control and apoptosis, e.g. p53 (10,11), which may have an impact on response to chemicals. These systems should be fully characterized for each cell line.
Lymphocytes, although karyotypically stable, when used in chromosome aberration assays have been forced out of the resting phase by exposure to powerful mitogens, which again may have an effect on their response to chemicals and toxins. In addition, there will be donor-to-donor variation based on the genetics and life-style of the person donating, which again may alter responses to genotoxins.
Target levels of cytotoxicity for a valid test
In the formative years of genetic toxicology as a separate field, it was clear that with the models available, some carcinogens could only be detected as genotoxic in in vitro models using mammalian cells at high levels of cytotoxicity, e.g. 50% reduction in the mitotic index in chromosome aberration assays and up to 90% cytotoxicity in the mouse lymphoma assay. Thus to define a negative compound in these assays, guidelines require data from tests carried out at highly cytotoxic concentrations such as these. It is apparent that mechanisms other than frank genotoxicity, e.g. apoptosis, release of nucleases from lysosomes etc. can contribute to chromosome damage at such toxic concentrations (12,13). In addition, different measures of cytotoxicity can lead to the identification of markedly different concentrations that give 50% toxicity, leading to further confusion (14). A thorough comparison of the different measures of cytotoxicity is needed to identify what is appropriate.
Most importantly, for non-toxic, soluble compounds, guidelines require tests to be carried out at concentrations up to 10 mM in mammalian cell tests, as long as osmolality and pH is not changed. For bacterial assays the target concentration is 5 mg per plate. These high concentrations can be entirely inappropriate with regard to metabolic activation by enzymes in the rat liver extracts used in the tests (see below) and, in most cases, will represent concentrations well in excess of any physiologically-relevant concentration that might be representative of likely tissue exposure levels in vivo.
Metabolic activation
It is known that the cell lines in common usage in mammalian cell genotoxicity do not possess the full complement of enzymes relevant to the metabolic activation of many different classes of genotoxic carcinogens. What enzymes that are present have never been systematically identified and quantified in the different cell lines and thus their contribution to responses of these cells to specific genotoxins is unknown. The use of rat liver S9 fraction, obtained from animals treated with enzyme inducers such as phenobarbitone, was developed as a surrogate for mammalian metabolism in such in vitro tests. However, it has been recognized from the beginning that this activation system is not a replicate of normal animal or human metabolism. It has inflated levels of particular P450 cytochromes and reduced levels of others of relevance to metabolism by human cells in vivo. It is also deficient in Phase II metabolism (generally detoxification conjugating enzymes); these enzymes are essentially inactive in standard S9, as their cofactors are not added (5). These deficiencies are often overlooked, but may well increase sensitivity for the detection of some carcinogens whilst providing opportunities to activate otherwise non-carcinogenic molecules that may be rapidly detoxified and eliminated in vivo. Regarding the predictive value of in vitro genotoxicity assays for detecting human carcinogens, it would be advantageous to ensure that the metabolic activation of molecules within these tests is a close as possible to that which will occur in the human (i.e. human-specific rather than rodent-specific metabolism).
The pressure for in vitro-only test batteries has regenerated interest in metabolically competent and engineered cell lines containing genes expressing human P450 cytochromes (CYP). Among the human hepatic cell lines, HepG2 is derived from a human liver tumour and possesses some xenobiotic-metabolizing activities as compared with fibroblasts. The levels of mRNA for most CYPs in primary cultures of hepatocytes are 
400-fold that of HepG2 cells (15). HepG2 cells do however retain the expression of some enzymes relevant to metabolism, including some phase II enzymes. Specific CYPs are also inducible in these cells in the presence of particular compounds such as benzo[a]pyrene. S9 derived from HepG2 cells can also activate some compounds to form mutagens. However, there is a paucity of data at present from testing of compounds that are not expected to be genotoxins or genotoxic carcinogens so that specificity cannot be determined accurately.
An engineered human cell line of interest is the MCL-5 line, which contains cDNA's for four human CYPs plus epoxide hydrolase (16). This cell line is heterozygous for p53 but undergoes normal DNA repair and apoptosis responses after exposure to mutagens. It has a modal chromosome number of 46, although is not likely to be euploid. MCL-5 has been used successfully to detect a variety of compounds requiring metabolic activation (17), but as for HepG2 cells, there is a paucity of data on non-genotoxic/non-carcinogenic compounds so that the specificity of genotoxicity tests using this cell line is not known accurately.
There is also a need to consider organ-specific metabolism in in vitro tests, with reference to the proposed or likely route(s) of exposure for the consumer (e.g. oral, dermal, inhalation). In particular, the question of how is the chemical metabolized at the site of contact (e.g. skin, lung, gut) and systemically, should be addressed. At present, for many chemicals, standard in vivo tests (e.g. rodent bone marrow micronucleus test) do not routinely account for the proposed route of human exposure in the first instance.
These factors all lead to the need for new approaches to incorporating relevant metabolic activation systems in genetic-toxicology tests (see the Section on New tests and approaches below).
Computer assisted structure activity relationships (CSAR)
The deficiencies and opportunities provided by CSAR for prediction of toxicity have recently been reviewed (18). The most widely used systems for predicting genotoxicity and carcinogenicity are MCASE (Multiple Computer Automated Structure Evaluation) and DEREK (Deductive Estimation of Risk from Existing Knowledge). These are normally used in conjunction with other information e.g. companies own heritage data, or systems are used in conjunction with each other. The current success of these programs in terms of uptake and accuracy is moderate, but growing, as they are becoming an essential part the strategies to identify potential toxins of many pharmaceutical companies, and other chemically based companies, influencing compound selection and timing of studies.
However, individual systems have not been without problems. These include: unsupported IT platforms; restricted chemical space; poor standardization of key data; black box predictions without justification; no accessibility to underlying data; synergism or antagonism of substructures not covered; variable quality of key data; inability of users to input new data; inability to predict non-positive structures.
Although many companies are using CSAR, there is no agreed standard approach although guidelines do exist for how Qualitative Structure Activity Relationship (QSAR) approaches may be validated for use in a regulatory setting (OECD Environment Health and Safety Publications, Series on Testing and Assessment, Number 49, Paris, November, 2004, www.oecd.org/ehs/). It is agreed that input from experts in chemistry can add value in interpretation of the output from these systems.
There are further developments and enhancements of existing systems in progress.
Developments of MCASE include MCASE-ES, co-developed with the US Food and Drug Administration (FDA).
MCASE-ES has increased coverage including more proprietary data, data covering stereochemistry, standardized data on compound acceptability, evaluation of alerts, MCASE modulators etc. This version has increased reliability of predictions including fewer false positives, but users cannot see the data on which the predictions are based.
The MCASE databases are now commercially available including the FDA database, which is available through Leadscope (www.leadscope.com).
MCASE has been updated and a version for Windows based PC's (MC4PC) and a web based version are available including a metabolism predictor (META).
DEREK is the most widely used predictive software and uses a rule-based expert system (Windows and web based versions available from Lhasa Ltd, www.lhasalimited.org).
Lhasa Ltd provides regular new rule updates based on user input and experience so the utility and chemical space covered by the system is increasing with time. Improvements include links to METEOR, a metabolism predictor and VITIC, a structure-searchable database capable of batch processing thousands of structures There are plans to include linking DEREK to various physicochemical calculators and also to release further modules to allow prediction of clastogenic potential of chemical structures in addition to Ames test prediction.
Leadscope and Lhasa Ltd are jointly supporting ToxML (www.leadscope.com/toxml.php), a public domain communication standard for chemical and toxicological data exchange, which will ease the ability of companies to compare regulatory data with their own proprietary data.
There is a need to be able to predict mechanisms of genotoxicity from chemical structure and related data e.g. thresholded mechanisms versus non-thresholded mechanisms. Most systems are geared towards predicting positive responses, rather than negative responses, but both have value. There is a huge increase in good quality data being generated, providing an increasing resource for data mining and determining predictive rules in a wider area of chemical space. The ability to predict the likelihood of false positive results with a new chemical would be a boon, but at present most of these programmes do not predict the outcome of mammalian mutagenicity tests and thus will remain as priority setting systems for biological testing in the immediate future.
|
New tests and approaches |
|---|
|
TopIntroductionCurrent tests and protocols...New tests and approachesNew insights into cancer...ConclusionsReferences |
|---|
It is possible to define an ideal cell line for use in genotoxicity testing. The chosen cell line would have a normal karyotype; would be karyotypically stable; would have a full complement of human phase I and phase II metabolic enzymes (representative of all human organs); its inherent metabolic activation and detoxification enzymes would be known and quantified as would its cellular defence systems and DNA repair capacity. The cells would be robust to culture and be readily available from a point source; would have a short cell cycle time and would not have over-active transporter systems so that molecules under test are not rapidly transported out of the cell. The cellular target(s) for determining genotoxicity would be responsive to a wide range of genotoxins acting via the full panoply of relevant genetic changes. At present no such cells exist! However, there are some systems under development and consideration that meet some of the requirements above.
The GreenScreen HC assay
This assay uses human lymphoblastoid TK6 cells transfected with the GADD45
genotoxic stress-specific response gene linked to a green fluorescent protein reporter (19). In the initial validation study of the assay, which was carried out in microtitre trays, 31 out of 34 of the expected genotoxins gave a positive response, whilst none of the 36 non-genotoxins were positive. This group included 11 cytotoxic non-carcinogens that had given positive responses in chromosome aberration in vitro studies, when tested at high levels of cytotoxicity. Only direct-acting chemicals (not requiring metabolic activation) were included in the initial trial as S9 is fluorescent and absorbing at the relevant wavelengths, and can interfere with the standard protocol. However, a flow cytometric version of the assay has been developed which has allowed the detection of several compounds that require metabolic activation. New versions of the assay using metabolically competent cells are being considered. Multi-centre trials are under way to determine the robustness of this assay.
Cellular defences against reactive oxygen species
The cellular defence capability against reactive oxygen species (ROS) of the cells used in genotoxicity testing is not completely defined. There is evidence that such defence mechanisms can vary markedly in particular cells within the body and also in cell lines derived from them. These differences have been shown to significantly change response to DNA-damaging agents such as UV (20). It appears that cells in culture can have lower than normal defence against ROS, which may enhance responses to genotoxins. A human-derived immortalized keratinocyte cell line HaCaT has been successfully transfected with a vector containing the SOD-1 gene at Unilever's UK laboratories (21). This gene codes for a cytosolic Cu/Zn superoxide dismutase and other antioxidant defence genes are also being explored. Trials are underway to determine the effects enhanced SOD-1 expression has on the response of this cell line to pro-oxidant genotoxins. This approach may have promise in the risk assessment of pro-oxidant chemicals but finding the right ‘balance’ of expression to reflect in vivo responses in an in vitro system will be difficult.
Engineered cells and cell lines with metabolic competence
Coecke et al. (22) have produced a useful review of the issues and possibilities for test developments incorporating metabolism systems in in vitro testing in all areas of toxicology; this is a report from ECVAM workshop 54.
Lines of V79 cells have been genetically engineered to express various human P450's. Several chemicals that are known to be in vivo genotoxic, DNA-reactive mutagenic carcinogens could be detected at very much lower concentrations than when S9 is used as the activation system (23). This is highly relevant to overcoming the current need for very high concentrations in the conventional assays, which may lead to spurious results in some cases. It would be helpful to extend the testing database of these engineered lines to see if specificity is affected with non-genotoxic carcinogens.
There is renewed interest in cell lines such as the human liver derived HepG2 that retain an element of metabolic competency, which can detect some genotoxic carcinogens not detected by bacterial mutagenicity tests (24). Several transfected lines of HepG2 have been constructed which express increased levels of phase I enzymes (such as CYP1A1, CYP1A2, CYP2E1 etc.); furthermore, cell lines are available which express human glutathione-S-transferases. HepG2 cells have also been used successfully as a source of S9. As for most of these possible options, there is little published information on the accuracy of tests using this cell line to find non-genotoxic carcinogens and non-carcinogens negative. There is some information that high concentrations of ascorbic acid and beta-carotene are positive in tests using these cells, which demonstrates that there may be problems with specificity.
Another promising cell line in this class is the Hepa RG, which is a human hepatocellular carcinoma derived line that retains good activity of many cytochromes and phase II enzymes (25). Several P450 cytochromes have been shown to be inducible in this cell line and it looks to be a useful surrogate for human hepatocytes. This cell line has been patented by the Institut National de la Santé et la Recherche Médicale (INSERM). More data are required using Hepa RG cells in genotoxicity testing so that their utility can be assessed.
The proliferating AR42J-B13 pancreatic cell line is known to respond to glucocorticoid treatment by producing foci of cells (B13-H) that express liver-specific or liver-enriched functional cytochrome P450 proteins (and phase II enzymes) when stimulated to trans-differentiate into hepatocytes by glucocorticoid treatment. These data suggest that this cell line has an unusual ability to trans-differentiate into functional hepatocytes and that it could be possible to generate a limitless supply of functional hepatocyte-like cells in vitro (26,27). Such cells again could be a source of metabolic activation for use in genotoxicity testing. If sufficient levels of metabolic activation could be achieved from the use of a cell line in in vitro genotoxicity assays, this may allow more flexibility in the duration of treatment that could be employed with these metabolizing systems. At present, where rat liver S9 is used as the standard metabolic activation system, the treatment time in mammalian cells is generally limited to a few hours owing to cytotoxicity of the S9 preparation itself.
Use of ‘omic’ technologies
Newer technologies such as transcriptomics, proteomics and metabolomics probably have the greatest potential to provide accurate assessments of genotoxic potential of chemicals in the longer term. They provide the opportunity to gain insight into genotoxic mechanisms of carcinogenicity beyond the initial DNA-damaging events, provide new markers in vitro and in vivo and possibly increase the accuracy of discriminating carcinogens from non-carcinogens. However, at present the sheer complexity of responses to genotoxic stresses provides major difficulties in formulating a clear way forward. In addition ‘omic’ sciences face technical limitations, managing the sheer volume of data produced, database challenges, and high costs. Intelligent analysis of this data requires sophisticated software to perform a range of pattern recognition techniques, including hierarchical cluster analysis, principal components analysis, partial least squares and neural networks. This in turn poses a major challenge to the scientists interrogating these analyses to formulate and grasp the biological implications that can be determined from the data. Therefore progress in these areas is likely to be slower and more tedious than promised by the initial hype. It must also be remembered whenever considering the use of new measurement technologies such as transcriptomics, proteomics and metabolomics that the issues mentioned earlier regarding suitability and characterization of the test system (e.g. metabolic capacity, cell line stability, physiologically-relevant cellular defences) are equally as important for these measurements as they are for traditional genetic-toxicology measurements (e.g. gene mutation, chromosome damage etc.).
Transcriptomics
The possible role of transcriptomics in genotoxicology has been reviewed by Aubrecht and Caba (28).
The yeast Saccharomyces cerevisiae has been used as a model to explore global gene expression as it is accepted that there is a high degree of conservation of key cellular processes between yeast and mammalian cells. Studies have revealed the enormous complexity of the so-called environmental stress response (ESR), which involves expression of up to 15% of the genes in the yeast genome. ESR consists of down regulation of protein synthesis genes together with up regulation of genes involved in lipid, carbohydrate and DNA metabolism (29). Exposure to the alkylating agent methylmethanesulphonate induced transcriptional changes in 20–30% of the yeast genome, only 8–10% of these belong to the stress category (30). Agent-specific profiles have been obtained, but in one study, only 21 transcripts responded to all tested compounds in a similar fashion, none of which belong to the DNA repair/damage category (31).
A comprehensive study using p53 proficient human lymphoblastoid TK6 cells and an isogenic p53 deficient cell line demonstrated an important role for p53 dependent pathways for DNA-damaging agents, whilst no such response was obtained for non-carcinogenic compounds at cytotoxic concentrations (32).
There have been several collaborative programmes exploring the use of transcriptomics in toxicology and genotoxicology, including a research programme organized by the Health and Environmental Sciences Institute (HESI) Genomics Committee. This included measurements of gene expression changes in mouse lymphoma (p53 deficient) cells exposed to a series of genotoxins acting through a variety of different mechanisms. The conclusions of this work demonstrated that traditional methods, including measurement of DNA adducts, were more sensitive, in that changes with genotoxic carcinogens were seen at lower concentrations with traditional methods, than gene expression changes in detecting the chosen genotoxins. There were few gene expression changes at low doses and most changes seen were less than three fold e.g. the NF-Kappa Beta pathway involved in cell survival. The exception was the GADD45 gene family where changes up to 15-fold were seen (see GreenScreen HC assay above). It is unlikely therefore that transcriptomic approaches will form the basis of new assays for detecting genotoxic carcinogens. One area where there may be a potential advantage in the use of transcriptomic technology however is to distinguish mechanisms of genotoxicity, e.g. those that have a threshold from those that act without a threshold (33). This conclusion has been reached by other similar studies (28). With this complex background, it may be naive to expect measurement of one endpoint or changes in one gene to give a true picture of response to genotoxins and cytotoxic non-genotoxins. The current development and use of bioinformatics tools will assist in this process; enabling complex overviewing of responses with mechanistic understanding. The use of transcriptomics will also become more powerful as databases of expression profiles become populated with larger numbers of studies of genotoxins acting by a variety of mechanisms, as well as studies of non-genotoxins. Choice of cell type following the considerations above, will be vital in maintaining and extending the utility of such databases. One new initiative in this area is the multi-centre ‘Carcinogenomics’ collaboration, co-ordinated by the University of Maastricht.
Proteomics and transformation assays
There appears to be few publications on the use of proteomics to study response to genotoxins and carcinogens. However, proteomics has been used extensively as an aid to cancer detection and risk assessment. Laser Capture Microdissection (LCM), coupled with downstream proteomics applications, such as 2D polyacrylamide gel electrophoresis and SELDI (surface enhanced laser desorption ionization) separation followed by mass spectrometry (MS) analysis, can facilitate the characterization and identification of protein expression changes that track normal and cancer phenotypes. Identification of such proteins could enable the development of protein chips allowing changes relevant to genotoxicity and carcinogenicity to be measured. For example, in a study of the tumour promoter TPA in cells from a poorly differentiated squamous cell carcinoma NPC cell line, CNE2, was found to up-regulate the expression of triosephosphate isomerase and 14-3-3 protein sigma and down-regulate the expression of reticulocalbin 1 precursor, nucleophosmin, mitochondrial matrix protein p1 precursor, and stathmin. These changes suggested that TPA induced CNE2 cells to antiproliferation and to apoptosis, which was confirmed by subsequent apoptosis detection (34).
Another area where proteomics may add value is in providing objective biomarkers for in vitro transformation assays. Although the SHE transformation assay appears to be a reasonable predictor of rodent carcinogenesis (genotoxic and non-genotoxic) as shown by the results of the ILSI trial (35) the ‘transformed’ phenotype is uncharacterized at the cellular and molecular level, and relies on somewhat subjective morphological criteria. Unilever are supporting a project with Invitrogen-Bioreliance to define proteomic biomarkers of transformation using SILAC (Stable Isotope Labelling with Amino acids in Cell culture), which is a mass spectrometric, quantitative proteomics technique. Other novel methodologies such as Infrared (IR) microspectroscopy may also have promise in such transformation biomarker identification, as has been the case in identification of neoplastic cells in tissue sections (36) and exfoliative cytology (37).
It is believed that only a fraction of the cells within a SHE cell population are capable of morphological transformation; therefore a better understanding of the events leading to morphological cell transformation may allow the identification of this subpopulation and thereby increase the statistical power of the assay.
Ideally a human cell-based system would be preferable to rodent based transformation assays, however, the difficulties associated with the extremely low transformation rate of human cells and the use of suitable populations of cells (e.g. human stem cells) would need to be overcome. Research in this area should be encouraged.
Metabolomics
Metabolomics, the study of global, small molecular weight metabolite profiles in cells, tissues, and organisms, may allow the metabolic consequences of gene expression and protein activity changes to be understood. It is known that cancer cells can produce a distinct pattern of metabolites, which can give an indication of prognosis (38). The application of metabolomics in this area may give new insights into intracellular signalling pathways and regulatory networks. Unilever, in conjunction with Cambridge University, are carrying out a metabolomics pilot study of intracellular metabolites in yeast strains using GC-MS and NMR. Genotoxic and non-genotoxic chemicals are being investigated to gain further insights into changes in cellular biochemistry in response to carcinogens. As for most of the ‘omic approaches described above, the outcome of such studies are not predictable at this stage and are unlikely to provide the required alternatives to current approaches in time for implementation of the Seventh Amendment in March 2009.
Organ models
Work using more complex culture models that possess characteristics more relevant to the intact organ could be used to put positive results from in vitro cell lines into context. Of particular interest to the cosmetic industry are 3D skin models such as EpiDermTM and EpiskinTM; 3D human skin models comprised of a reconstructed epidermis and a functional stratum corneum. These models have been developed primarily for measuring the potential of chemicals to induce skin corrosivity, irritation and sensitization. However, a task force initiated by the European Commission and led by ECVAM recommended the use of genotoxicity assays using skin models in lieu of animal testing (39). Taking on this recommendation, the European Cosmetic, Toiletry and Perfumery Association (COLIPA) Genotoxicity Task Force is developing a joint-industry testing programme to help validate this approach. This aims to evaluate the models using genotoxicity measured via the comet and micronucleus endpoints. The reproducibility of the models and assays will be assessed between laboratories and the ability of the tests to correctly identify carcinogens/genotoxins determined. Furthermore, an extensive assessment of the metabolic capability of the 3D models will be conducted, comparing the activities to that found in normal human skin. There is a report of the development of such a micronucleus assay using isolated cells from the EpiDermTM model (40). At least four, direct-acting genotoxins have been shown to induce micronuclei in this model, whilst four rodent skin non-carcinogens were negative. The cells isolated have been shown to possess metabolic capability similar to that found in human skin cells and further experiments are planned to test genotoxins requiring metabolic activation (5). A pilot study using the Comet assay in isolated keratinocytes EpiskinTM has also been reported. The induction of DNA damage by UV-A light is enhanced by the fluoroquinolone lomefloxacin and 4-nitroquinoline-N-oxide induces DNA damage in this model (41). Other 3D skin models are also being produced by industry (e.g. Henkel/Phenion) and Academia (e.g. University of Leiden). Models of other organs have also been developed including three-dimensional, multicellular spheroids of liver cells (42), bone marrow cell cultures (43) etc., that could be used to develop new models for genotoxicity testing.
Full characterization of such tissue models would be essential prior to their inclusion in any decision-making strategy. Large collaborative programs are required to determine if these models can play a useful role in all in vitro test batteries. Such programs would need to consider carefully where these models would fit in a testing battery, e.g. if it is proposed that these models would only be used in a second tier once in vitro positive results have been obtained in standard genotoxicity assays this would need to be taken into account in the choice of chemicals used for any validation studies. It would also be essential to understand the reason for negative results in these tissue models if positive results had previously been obtained in in vitro cell-based genotoxicity assays. This would probably rely on a robust demonstration that the tissue models were closer to the in vivo situation than existing tests as well as a clear demonstration that documented non-carcinogens that are known to give positive results in standard genotoxicity assays give negative results in such tissue models. This would therefore be a long-term aspect of any research and would need to involve clinical research groups, tissue engineering groups and the exchange of information at relevant scientific meetings.
|
New insights into cancer biology and opportunities for new screens |
|---|
|
TopIntroductionCurrent tests and protocols...New tests and approachesNew insights into cancer...ConclusionsReferences |
|---|
Epigenetics
The term ‘epigenetics’ defines all heritable changes in gene expression and chromatin organization that are not coded in the DNA sequence. Epigenetic inheritance including DNA methylation and histone protein modifications are essential mechanisms that allow the stable propagation of gene activity states from one generation of cells to the next. Dysregulated epigenetic mechanisms are implicated in cancer development, although the precise contribution they make is not completely known.
Patterns of DNA methylation and chromatin structure are profoundly altered in neoplasia and include genome-wide losses of, and regional gains in, DNA methylation. Recent knowledge of how chromatin organization modulates gene transcription has further highlighted the importance of epigenetic mechanisms in the initiation and progression of human cancer. These epigenetic changes, in particular, aberrant promoter hypermethylation that is associated with inappropriate gene silencing, affect virtually every step in tumour progression (44).
As these epigenetic changes result in changes in transcription, it is possible that transcriptomic methods will provide insight into the consequences of epigenetic changes (see above). Screens measuring compound induced over- expression of specific genes, such as cytosine-DNA methyltransferases and /or screens measuring hypermethylation of CpG islands in the promoters of a small panel of target genes by methylation-specific PCR (45) may be useful in detecting compounds acting via such mechanisms.
Cellular hubs
Formation of many tumours may require the following steps: (i) loss of the response to antigrowth signals (e.g. inactivation of Rb and p53 cellular pathways); (ii) activation of growth promoting pathways (e.g. Ras activation); (iii) evasion of apoptosis; (iv) telomerase activation or alternative mechanisms of cellular immortalization; (v) angiogenic activity; and (vi) the ability to invade surrounding tissues and to metastasize (46). Other events such as inactivation of phosphatase 2A that causes changes in the phosphorylation and activity of several cellular proteins (47) may also be important. A recent review in Scientific American (48) highlights the importance of stem cells as the real source of malignancies. Following an initiation event, such a cell becomes irreversibly blocked from terminal differentiation, possibly due to changes in cell-to-cell communication. In the promotion phase there is a potentially reversible clonal expansion of the initiated cell by a combination of growth stimulation and inhibition of apoptosis. These expanded cells in a progression phase accrue mutations and epigenetic changes to become growth stimulus independent and resistant to growth inhibition. At this point the cells have the potential for invasive and metastatic phenotypes. Assays that take into account the unique nature of stem cells and the steps above must be a future goal in the field of carcinogen detection and risk assessment (49,50). Each one of these steps in carcinogenesis may be controlled by a cellular ‘hub’ involving complex interactions of different cellular functions. The future development of novel in vitro assays for the assessment of potential chemical carcinogenicity may need to embrace approaches that can overview the ‘systems biology’ of these events in order to effectively target the key ‘hubs’ and their molecular interplay.
An interesting approach in this respect has been to use a set of yeast strains, each one with a different gene deleted, to generate a profile of biochemical pathways and biological networks that are responsible for maintaining cellular viability in response to damage to macromolecules (51). These strains have been evaluated following exposure to a small number of genotoxins, the data collected and analysed by a programme called Cytoscape (52), to develop a yeast ‘interactome’ profile for each agent; thus assessing key interactive and cross-talking hubs. An equally fascinating and emerging approach, enabling a bridging between in vitro systems and in vivo human cancer biology, comes from work that combines computational and experimental systems biology. New language tools, such as that developed by Gene Network Sciences (53), create models of the interconnected signal transduction pathways and the gene expression networks (including information on receptor activation, mitogenic signalling, initiation of cell cycle and passage of checkpoints and apoptosis) that control human cell proliferation. Such tools may become part of the evaluation of complex data generated by new in vitro batteries in the future assessment of chemical carcinogenic potential.
|
Conclusions |
|---|
|
TopIntroductionCurrent tests and protocols...New tests and approachesNew insights into cancer...ConclusionsReferences |
|---|
The currently established in vitro tests for determining genotoxicity have good sensitivity but poor specificity for predicting rodent carcinogens. Thus all batteries composed only of in vitro tests, i.e. excluding animal tests, as required by the Seventh Amendment to the EU Cosmetic Directive, would result in a large proportion of tested chemicals without carcinogenic potential being discarded unnecessarily. It appears that there is no one factor that is responsible for this poor performance but various possibilities exist for improving the specificity of established tests. These include reduction of inadvertent generation of reactive oxygen species in culture; better characterization and control of the karyotype of the cells used; improvements in metabolic activation systems; reduction in the levels of cytotoxicity required to define negative results; use of improved predictive software etc. New models and approaches are at various stages of development including a mammalian cell model looking at induction of Gadd 45
; novel cell lines and engineered cell lines with metabolic competence or boosted cellular defences against reactive oxygen species; use of the ‘omic technologies; organ models particularly of the skin that could all make a valuable contribution to better prediction of genotoxicity and potential carcinogenicity. Models that measure compound induced epigenetic changes are available and may yield useful data for the prediction of carcinogenic potential. The ability to investigate the relevance of positive in vitro genotoxicity results for prediction of carcinogenicity in humans without the use of animals represents a significant scientific and technical challenge. As described above a number of factors must be taken into consideration and it is unlikely that a standardized experimental approach will be appropriate in every case. A range of accepted, robust tools to investigate in vitro positives should be available to allow appropriate experiments to be conducted on a case-by-case basis depending on the nature of the initial observation in the standard in vitro tests. A number of promising approaches have been described, however, in every case, even if major international collaborative trials are instigated in the near future, to change the current situation by March 2009 when the Seventh Amendment takes effect will be a significant challenge. However, there is real promise that in the longer term some of the approaches described in this Commentary will make a major impact on screening chemicals for genotoxicity and potential carcinogenicity in the absence of data from whole animals.
|
Acknowledgments |
|---|
Part of Unilever's ongoing effort to develop novel ways of delivering consumer safety.
Conflict of interest: I have a financial arrangement with the company Gentronix which is developing the GreenScreen HC Assay, this is not currently a commercial product. The reference to the GreenScreen assay is factual and no direct opinion on the utility of the assay is stated.
|
Notes |
|---|
*To whom correspondence should be addressed. Tel/Fax: +77 423 37093; Email: david{at}tweats.fslife.co.uk
|
References |
|---|
|
TopIntroductionCurrent tests and protocols...New tests and approachesNew insights into cancer...ConclusionsReferences |
|---|
-
1. Kirkland D., Aardema M., Henderson L., Müller L. (2005) Evaluation of the ability of a battery of 3 in vitro genotoxicity tests to discriminate rodent carcinogens and non-carcinogens. I. Sensitivity, specificity and relative predictivity. Mutat. Res. 584:1–256.[ISI][Medline]
2. Matthews E.J., Kruhlak N.L., Cimino M.C., Benz R.D., Contrera J.F. (2006) An analysis of genetic toxicity, reproductive and developmental toxicity, and carcinogenicity data: I. Identification of carcinogens using surrogate endpoints. Regul. Toxicol. Pharmacol. 44:83–96.[CrossRef][ISI][Medline]
3. Tweats D.J., Blakey D., Heflich R.H., Jacobs A., Jacobsen S.D., Morita T., Nohmi T., O'Donovan M.R., Sasaki Y.F., Sofuni T., Tice R. (2006) Report of the IWGT working group on strategies and interpretation of regulatory in vivo tests. II. Identification of ‘in vivo only’ positive compounds in the bone marrow micronucleus test. Mutat. Res. (in press).
4. Pastoor T. and Stevens J. (2005) Historical perspective of the cancer bioassay. Scand. J. Work Environ. Health 31:119–122.
5. Kirkland D., Pfuhler S., Tweats D, et al. (2006) How to reduce false positive results when undertaking in vitro genotoxicity testing and thus avoid unnecessary follow-up animal tests: report of an ECVAM Workshop. Mutat. Res. (in press).
6. Halliwell B. (2003) Oxidative stress in cell culture: an under-appreciated problem? FEBS Lett. 540:3–6.[CrossRef][ISI][Medline]
7. Wee L.M., Long L.H., Whitemen M., Halliwell B. (2003) Factors affecting the ascorbate- and phenolic-dependent generation of hydrogen peroxide in Dulbecco's modified Eagles medium. Free Radic. Res. 37:1123–1130.[CrossRef][ISI][Medline]
8. Tweats D.J. (1992) Cell culture and mutagenicity assessment. Analyt. Proc. 29:476–481.
9. Sofuni T., Matsuoka A., Sawada M., Ishidate M. Jr, Zeiger E., Shelby M.D. (1990) A comparison of chromosome aberration induction by 25 compounds tested by two Chinese hamster cell (CHL and CHO) systems in culture. Mutat. Res. 241:175–213.[CrossRef][ISI][Medline]
10. Oka H., Ikeda K., Yoshimura H., Ohuchida A., Honma M. (2006) Relationship between p53 status and 5-fluorouracil sensitivity in 3 cell lines. Mutat. Res. 606:52–60.[ISI][Medline]
11. Chaung W., Mi L.-J., Boorstein R.J. (1997) The p53 status of Chinese hamster V79 cells frequently used for studies of DNA damage and DNA repair. Nucleic Acid Res. 25:992–994.
12. Kirkland D. (1992) Chromosomal aberration tests in vitro: problems with protocol design and interpretation of results. Mutagenesis 7:95–106.
13. Scott D., Galloway S.M., Marshall R. Jr, Ishidate M. Jr, Brusick D., Ashby J., Myhr B.C. (1991) Genotoxicity under extreme culture conditions. A report from ICPEMC Task group 9. Mutat. Res. 257:147–204.[CrossRef][ISI][Medline]
14. Greenwood S.K., Hill R.B., Sun J.T., Armstrong M.J., Johnson T.E., Gara J.P., Galloway S.M. (2004) Population doubling: a simple and more accurate estimation of cell growth suppression in the in vitro assay for chromosomal aberrations that reduces irrelevant positive results. Environ. Mol. Mutagen. 43:36–44.[CrossRef][ISI][Medline]
15. Rodriguez-Antona C., Donato M.T., Boobis A., Edwards R.J., Watts P.S., Castell J.V., Gomez-Lechon M.J. (2002) Cytochrome P450 expression in human hepatocytes and hepatoma cell lines: molecular mechanisms that determine lower expression in cultured cells. Xenobiotica 32:505–520.[CrossRef][ISI][Medline]
16. Crespi C.L., Gonzalez F.J., Steimel D.T., Turner T.R., Gelboin H.V., Penman B.W., Langenbach R. (1991) A metabolically competent human cell line expressing 5 cDNAs encoding procarcinogen-activating enzymes: Application to mutagenicity testing. Chem. Res. Toxicol. 4:566–572.[CrossRef][ISI][Medline]
17. Doherty A.T., Ellard S., Parry E.M., Parry J.M. (1996) An investigation into the activation and deactivation of chlorinated hydrocarbons to genotoxins in metabolically competent human cells. Mutagenesis 11:247–274.
18. Johnson D.E. and Rodgers A.D. (2006) Computational toxicology: heading toward more relevance in drug discovery and development. Curr. Opin. Drug Discov. Dev. 9:29–37.[ISI][Medline]
19. Hastwell P.W., Chai L.L., Roberts K.J., Webster T.W., Harvey J.S., Rees R.W., Walmsley R.M. (2006) High-specificity and high-sensitivity genotoxicity assessment in a human cell line: validation of the GreenScreen HC GADD45a-GFP genotoxicity assay. Mutat. Res. 607:160–175.[ISI][Medline]
20. Leccia M.T., Richard M.J., Joanny-Crisci F., Beani J.C. (1998) UV-A1 cytotoxicity and antioxidant defence in keratinocytes and fibroblasts. Eur. J. Dermatol. 8:478–482.[ISI][Medline]
21. Maskell S., Malcomber S., Windebank S., Carmichael P., Scott A. (2006) Development of a human keratinocyte cell line that is up-regulated for the expression of superoxide dismutase 1 (SOD1). Toxicology (in press).
22. Coecke S., Ahr H., Blaauboer B.J., et al. (2006) Metabolism: a bottleneck in in vitro toxicological test development. The report and recommendations of ECVAM workshop 54. Altern Lab Anim. 34:49–84.[Medline]
23. Glatt H., Schneider H., Liu Y. (2005) V79-hCYP2E1-hSULT1A1, a cell line for the sensitive detection of genotoxic effects induced by carbohydrate pyrolysis products and other food-borne chemicals. Mutat. Res. 580:41–52.[ISI][Medline]
24. Knasmuller S., Mersch-Sundermann V., Kevekordes S., Darroudi F., Huber W.W., Hoelzl C., Bichler J., Majer B.J. (2004) Use of human-derived liver cell lines for the detection of environmental and dietary genotoxicants; current state of knowledge. Toxicology 198:315–328.[CrossRef][ISI][Medline]
25. Aninat C., Piton A., Glaise D., Le Charpentier T., Langouet S., Morel F., Guguen-Guillouzo C., Guillouzo A. (2006) Expression of cytochromes P450, conjugating enzymes and nuclear receptors in human hepatoma HepaRG cells. Drug Metab. Dispos. 34:75–83.
26. Marek C.J., Cameron G.A., Elrick L.J., Hawksworth G.M., Wright M.C. (2003) Generation of hepatocytes expressing functional cytochromes P450 from a pancreatic progenitor cell line in vitro. Biochem. J. 370:763–769.[CrossRef][ISI][Medline]
27. Burke Z.D., Shen C.N., Ralphs K.L., Tosh D. (2006) Characterization of liver function in transdifferentiated hepatocytes. J. Cell. Physiol. 206:147–159.[CrossRef][ISI][Medline]
28. Aubrecht J. and Caba E. (2005) Gene expression profile analysis: an emerging approach to investigate mechanisms of genotoxicity. Pharmacogenomics 6:419–428.[CrossRef][ISI][Medline]
29. Gasch A.P., Spellman P.T., Kao C.M., Carmel-Harel O., Eisen M.B., Storz G., Botstein D., Brown P.O. (2000) Genomic expression programs in the response of yeast cells to environmental changes. Mol. Biol. Cell 11:4241–4257.
30. Jelinsky S.A. and Samson L.D. (1999) Global response of Saccharomyces cerevisiae to an alkylating agent. Proc. Natl Acad. Sci. USA 96:1486–1491.
31. Jelinsky S.A., Estep P., Church G.M., Samson L.A. (2000) Regulatory networks revealed by transcriptional profiling of damaged Saccharomyces cerevisiae cells: Rpn4 links base excision repair with proteasomes. Mol. Cell. Biol. 20:8157–8167.
32. Amundson S.A., Do K.T., Vinikoor L., Koch-Paiz C.A., Bittner M.L., Trent J.M., Meltzer P., Fornace A.J. Jr. (2005) Stress-specific signatures: expression profiling of p53 wild-type and -null human cells. Oncogene. 24:4572–4579.[CrossRef][ISI][Medline]
33. Newton R.K., Aardema M., Aubrecht J. (2004) The utility of DNA microarrays for characterizing genotoxicity. Environ. Health Perspect. 112:420–422.[ISI][Medline]
34. Jiang P., Gan M., Huang H., Shen X., Wang S., Yao K. (2005) Proteome analysis of antiproliferative mechanism of 12-O-tetradecanoylphorbol 13-acetate on cultured nasopharyngeal carcinoma CNE2 cells. J. Proteome Res. 4:599–605.[CrossRef][ISI][Medline]
35. Mauthe M.J., Gibson D.P., Bunch R.T., Custer L. (2001) The Syrian hamster embryo (SHE) cell transformation assay; a review of the methods and results. Toxicol. Pathol. 29:Suppl., 138–146.
36. German M.J., Hammiche A., Ragavan N., et al. (2006) Infrared spectroscopy with multivariate analysis potentially facilitates the segregation of different types of prostate cell. Biophys. J. 90:1–13.
37. Walsh M.J., German M.J., Singh M., Pollock H.M., Hammiche A, Kyrgiou M., Stringfellow H.F., Paraskevadis E., Martin-Hirsch P.L., Martin F.L. (2006) IR microspectroscopy: potential applications in cervical cancer screening. Cancer Lett. [Epub ahead of print].
38. Ippolito J.E., Xu J., Jain S., Moulder K., Mennerick S., Crowley J.R., Townsend R.R., Gordon J.I. (2005) An integrated functional genomics and metabolomics approach for defining poor prognosis in human neuroendocrine cancers. Proc. Natl Acad. Sci. USA 102:9901–9906.
39. Maurici D., Aardema M., Corvi R., et al. (2005) Genotoxicity and mutagenicity. Altern. Lab. Anim. 33:Suppl. 1, 117–130.
40. Curren R.D., Mun G.C., Gibson D.P., Aardema M.J. (2006) Development of a method for assessing micronucleus induction in a 3D human skin model (EpiDermTM). Mutat. Res. 607: pp. 192–204.[ISI][Medline]
41. Flamand N., Marrot L., Belaidi J.P., Bourouf L., Dourille E., Feltes M., Meunier J.R. (2006) Development of genotoxicity test procedures with Episkin®, a reconstructed human skin model: towards new tools for in vitro risk assessment of dermally applied compounds? Mutat. Res. 606:39–51.[ISI][Medline]
42. Xu J., Ma M., Purcell W.M. (2003) Biochemical and functional changes of rat liver spheroids during spheroid formation and maintenance in culture: II. nitric oxide synthesis and related changes. J. Cell. Biochem. 90:1176–1185.[CrossRef][ISI][Medline]
43. Dal Negro G., Vandin L., Bonato M., Sciuscio D. (2006) Toward refinement of the colony-forming unit-granulocyte/macrophage clonogenic assay: inclusion of a metabolic system. Toxicol. In Vitro 20:743–749.[CrossRef][ISI][Medline]
44. Jones P.A. and Baylin P.A. (2002) The fundamental role of epigenetic events in cancer. Nat. Rev. Genet. 3:415–428.[ISI][Medline]
45. Vuillemenot B.R., Hutt J.A., Belinsky S.A. (2006) Gene promoter hypermethylation in mouse lung tumours. Mol.Cancer Res. 4:267–273.
46. Hanahan D. and Weinberg R.A. (2000) The hallmarks of cancer. Cell 100:57–70.[CrossRef][ISI][Medline]
47. Bocchetta M. and Carbone M. (2004) Epidemiology and molecular pathology at crossroads to establish causation: molecular mechanisms of malignant transformation. Oncogene 23:6484–6491.[CrossRef][ISI][Medline]
48. Clarke M.F. and Becker M.F. (2006) Stem cells:the real culprits in cancer? Sci. Amer. 295:52–59.[ISI][Medline]
49. Trosko J.E., Chang C.C., Upham B.L., Tai Mh H. (2004) Ignored hallmarks of carcinogenesis: stem cells and cell-cell communication. Ann. NY Acad. Sci. 1028:192–201.
50. Trosko J.E. and Upham B.L. (2005) The emperor wears no clothes in the field of carcinogen risk assessment. Mutagenesis 20:81–97.
51. Begley T.J., Rosenbach A.S., Ideker T., Samson L.D. (2002) Damage recovery pathways in Saccaromyces cerevisiae revealed by genomic phenotyping and interactome mapping. Mol. Cancer Res. 1:103–112.
52. Ideker T., Ozier O., Schwikowski B., Siegel A. (2002) Discovering regulatory and signalling circuits in molecular interaction networks. Bioinformatics 18:233–240.
53. Christopher R., Dhiman A., Fox J., Gendelmann R., Haberitcher T., Kagle D., Spizz G., Khalil I.G., Hill C. (2004) Data-driven computer simulation of human cancer cell. Ann. NY Acad. Sci. 120:132–153.
Received on August 24, 2006; revised on October 9, 2006; accepted on October 9, 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
| |||||||||||||||||||||||||||||||||||||||||||||