Modeling the mouse lymphoma forward mutational assay: the Gene-Tox program database

MultiCASE Platform for In Silico Toxicology

Predictive and computational toxicology, a highly scientific and research-based field, is rapidly progressing with wider acceptance by regulatory agencies around the world. Almost every aspect of the field has seen fundamental changes during the last decade due to the availability of more data, usage, and acceptance of a variety of predictive tools and an increase in the overall awareness. Also, the influence from the recent explosive developments in the field of artificial intelligence has been significant. However, the need for sophisticated, easy to use and well-maintained software platforms for in silico toxicological assessments remains very high. The MultiCASE suite of software is one such platform that consists of an integrated collection of software programs, tools, and databases. While providing easy-to-use and highly useful tools that are relevant at present, it has always remained at the forefront of research and development by inventing new technologies and discovering new insights in the area of QSAR, artificial intelligence, and machine learning. This chapter gives the background, an overview of the software and databases involved, and a brief description of the usage methodology with the aid of examples.

QSAR screening of 70,983 REACH substances for genotoxic carcinogenicity, mutagenicity and developmental toxicity in the ChemScreen project

The ChemScreen project aimed to develop a screening system for reproductive toxicity based on alternative methods. QSARs can, if adequate, contribute to the evaluation of chemical substances under REACH and may in some cases be applied instead of experimental testing to fill data gaps for information requirements. As no testing for reproductive effects should be performed in REACH on known genotoxic carcinogens or germ cell mutagens with appropriate risk management measures implemented, a QSAR pre-screen for 70,983 REACH substances was performed. Sixteen models and three decision algorithms were used to reach overall predictions of substances with potential effects with the following result: 6.5% genotoxic carcinogens, 16.3% mutagens, 11.5% developmental toxicants. These results are similar to findings in earlier QSAR and experimental studies of chemical inventories, and illustrate how QSAR predictions may be used to identify potential genotoxic carcinogens, mutagens and developmental toxicants by high-throughput virtual screening.

Dietary flavonoids bind to mono-ubiquitinated annexin A1 in nuclei, and inhibit chemical induced mutagenesis

In order to investigate the mechanisms of anti-mutagenic action by dietary flavonoids, we investigated if they inhibit mutation of the thymidine kinase (tk) gene in L5178Ytk(±) lymphoma cells. Silibinin, quercetin and genistein suppressed mutation of the tk gene induced in L5178Ytk(±) lymphoma cells by methyl methanesulfonate (MMS) and As(3+). Flavone and flavonol were less effective. To establish that mutation of the tk gene in L5178Ytk(±) lymphoma cells by MMS and As(3+) is mediated through mono-ubiquitinated annexin A1, L5178Ytk(±) lymphoma cells were treated with annexin A1 anti-sense oligonucleotide. The treatment reduced mRNA as well as protein levels of annexin A1, and suppressed mutation of the tk gene. Nuclear extracts from L5178Ytk(±) lymphoma cells catalyzed translesion DNA synthesis with an oligonucleotide template containing 8-oxo-guanosine in an annexin A1 dependent manner. This translesion DNA synthesis was inhibited by the anti-mutagenic flavonoids, silibinin, quercetin and genistein, in a concentration dependent manner, but only slightly by flavone and flavonol. Because these observations implicate involvement of annexin A1 in mutagenesis, we examined if flavonoids suppress nuclear annexin A1 helicase activity. Silibinin, quercetin and genistein inhibited ssDNA binding, DNA chain annealing and DNA unwinding activities of purified nuclear mono-ubiquitinated annexin A1. Flavone and flavonol were ineffective. The apparent direct binding of anti-mutagenic flavonoids to the annexin A1 molecule was supported by fluorescence quenching. Taken together, these findings illustrate that nuclear annexin A1 may be a novel and productive target protein of prevention for DNA damage induced gene mutation, ultimately conferring cancer chemoprevention.

The blood-based glycophorin A (GPA) human in vivo somatic mutation assay

The glycophorin A assay concurrently detects and quantifies erythrocytes with allele-loss phenotypes at the autosomal locus responsible for the polymorphic MN blood group. It uses a pair of allele-specific monoclonal antibodies and flow cytometry to efficiently analyze a standard population of five million cells. Two distinct variant phenotypes are detected: simple allele loss and allele loss followed by reduplication of the remaining allele; both are consistent with the mechanisms underlying "loss of heterozygosity" at tumor-suppressor genes. The assay is an intermediate biomarker of biological effect in the somatic mutational model of human cancer and has been applied to populations with a known or suspected genotoxic exposure, to patients with hereditary syndromes causing predisposition to cancer (where the assay has been applied diagnostically), and to patients manifesting cancer as a disease endpoint.

An analysis of genetic toxicity, reproductive and developmental toxicity, and carcinogenicity data: II. Identification of genotoxicants, reprotoxicants, and carcinogens using in silico methods

This study examined a novel method to identify carcinogens that employed expanded data sets composed of in silico data pooled with actual experimental genetic toxicity (genetox) and reproductive and developmental toxicity (reprotox) data. We constructed 21 modules using the MC4PC program including 13 of 14 (11 genetox and 3 reprotox) tests that we found correlated with results of rodent carcinogenicity bioassays (rcbioassays) [Matthews, E.J., Kruhlak, N.L., Cimino, M.C., Benz, R.D., Contrera, J.F., 2005b. An analysis of genetic toxicity, reproductive and developmental toxicity, and carcinogenicity data: I. Identification of carcinogens using surrogate endpoints. Regul. Toxicol. Pharmacol.]. Each of the 21 modules was evaluated by cross-validation experiments and those with high specificity (SP) and positive predictivity (PPV) were used to predict activities of the 1442 chemicals tested for carcinogenicity for which actual genetox or reprotox data were missing. The expanded data sets had approximately 70% in silico data pooled with approximately 30% experimental data. Based upon SP and PPV, the expanded data sets showed good correlation with carcinogenicity testing results and had correlation indicator (CI, the average of SP and PPV) values of 75.5-88.7%. Conversely, expanded data sets for 9 non-correlated test endpoints were shown not to correlate with carcinogenicity results (CI values <75%). Results also showed that when Salmonella mutagenic carcinogens were removed from the 12 correlated, expanded data sets, only 7 endpoints showed added value by detecting significantly more additional carcinogens than non-carcinogens.

Lack of predictivity of the rat lethality (LD50) test for ecological and human health effects

The relationship between acute toxicity in rats (LD50 values) and indicators of potential health hazards in humans was investigated, based on a chemical population-based paradigm (i.e. the "chemical diversity approach"). These structure-activity relationship-based analyses indicate that high toxicity in rats (i.e. a low LD50 value) is not a good predictor of health effects in humans. In fact, it was found that high acute toxicity to minnows, as well as toxicity to cultured cells, showed significantly greater associations with the potential for health effects than rat LD50 values.

SAR modeling of genotoxic phenomena: the consequence on predictive performance of deviation from a unity ratio of genotoxicants/non-genotoxicants

SAR approaches to the study of genotoxic phenomena are finding increased applications. However, the data being modeled are frequently not considered optimal due to the small size of the dataset and an uneven distribution of genotoxicants and non-genotoxicants in the dataset. The effects of such imbalances on the performance of one SAR approach were investigated with respect to the modeling of the induction of unscheduled DNA synthesis in rat hepotocytes and of sister chromatid exchanges and chromosomal aberrations in cultural CHO cells. The analyses indicate that if genotoxicants exceed non-genotoxicants, the performance of the SAR model can be improved if the dataset is supplemented with physiological chemicals which are assumed to be non-genotoxicants. On the other hand, if non-genotoxicants exceed genotoxicants, it was found that the predictive performance of the resulting SAR model is not improved by removal of genotoxicants from the dataset to achieve a ratio of genotoxicants/genotoxicants of unity. Overall, the present analyses did not result in the development of SAR models of greatly increased predictivity. Conceivably, for the particular datasets and SAR paradigm the limit of predictivity has been reached. The possibility of investigating the use of a "battery" of SAR paradigms should be considered.

Identification of structural components associated with cytostatic activity in MCF-7 but not in MDA-MB-231 cells

The National Cancer Institute's Developmental Therapeutics Program maintains the screening results obtained in 60 standardized cancer cell lines and contained 37,836 compounds for this study. This dataset has shown to be an outstanding resource for the development of structure-activity relationship (SAR) models describing anticancer activity. We report here a novel SAR modeling approach based on a subtractive protocol to develop models that describe cell type-specific molecular descriptors of cytotoxicity. The goal of this approach is to separate features associated with antiproliferative activity to many cell lines from those that effect only a specific cell type. To assess this approach, we developed SAR models for cytostatic activity against the human breast cancer cell lines MCF-7 and MDA-MB-231 and one differential activity model for compounds that were potent cytostatic agents in MCF-7 cells but relatively inactive against MDA-MB-231 cells. The models were between 72 and 84% accurate when challenged with compounds not in the learning sets. Structural features associated with the differential activity model highlighted how the use of this approach can selectively identify chemical moieties associated with potent cytostatic action to MCF-7 but not to MDA-MB-231 cells. We surmise that outgrowth of this method can facilitate the development of SAR models with sufficient resolution and clarity to identify chemical moieties associated with antiproliferative activity to selective individual cancer types while being innocuous to other cell types.

A data mining approach for the elucidation of the action of putative etiological agents: application to the non-genotoxic carcinogenicity of genistein

A procedure designated "the virtual similarity index" (VSI) is described to determine the probability that two or more toxicants are related mechanistically. The approach is structure-activity relationship (SAR) based and generates the virtual toxicological profiles of the chemicals under investigation. It also determines the similarities between them. That commonality is compared to the frequency with which it is found among a population of 10,000 chemicals representing the "universe of chemicals". The similarities between the candidate chemicals and chemicals known to act by other recognized mechanisms are also determined. If the similarities between the candidate chemicals are significantly greater than for the non-related ones, the chemicals are assumed to act by a common mechanism. In that context, the putative non-genotoxic mechanism responsible for the carcinogenicity of genistein (GEN) and its relationship to the action of diethylstilbestrol is examined.

A paradigm for determining the relevance of short-term assays: application to oxidative mutagenesis

A simple substructure-based approach was developed to determine whether a short-term assay under development is related mechanistically to the endpoint it seeks to predict. Thus, substructures associated with mutagenicity in Salmonella are also present in carcinogens and agents active in other mutagenicity and genotoxicity assays. When applied to test results obtained with an Escherichia coli strain designed to identify oxidative mutagens, there was no significant association with either carcinogens or mutagens and genotoxicants detected by other systems. There was, however, a significant association between alerts for oxidative mutagenesis and chemicals capable of inducing allergic contact dermatitis (ACD) in humans.

Allergic contact dermatitis and its relationship to carcinogenicity

The relationship between allergic contact dermatitis (ACD) and carcinogenicity was investigated using a recently developed and validated simulation approach. The analyses indicated that while there are electrophilic and non-electrophilic components to ACD, these were not identical to those operating in carcinogenicity. Accordingly, with respect to carcinogenicity prediction, the results of ACD do not improve the results based upon mutagenicity testing alone, the latter being a surrogate for potential electrophilicity.

Chemical categories for health hazard identification: a feasibility study

The use of chemical categories has been suggested in order to lower the number of chemicals tested in the High Production Volume (HPV) Chemical Challenge Program. In this investigation we examined the reliability of using organic chemical categories to classify chemicals as either toxic or nontoxic for individual toxicological effects as well as for panels of such endpoints. The analyses indicate that chemical categories are unable to consistently identify groups of chemicals with similar toxic responses either for a multiplicity of endpoints or for single effects. Our analyses suggest that if chemical categories are to be used to identify health hazards, that computer-based SAR approaches appear to be superior to arbitrary chemical categories for predicting specific toxicological effects but they are not, at this time, useful for defining the overall toxicity.

SAR modeling of genotoxic phenomena: the effect of supplementation with physiological chemicals

Structure-activity relationship (SAR) modeling of toxicological phenomena is optimal when the ratio of toxicants to non-toxicants included in the model is unity. Frequently, however, the experimental data available are enriched with toxicants, this appears to be especially true for genotoxicity data sets. It is demonstrated herein, using a Salmonella mutagenicity data set, that when there is a paucity of non-toxicants, the learning set may be augmented with physiological chemicals on the assumption that they are non-genotoxic.

Computational toxicology and the generation of mechanistic hypotheses: gamma-butyrolactone

In this study, we use SAR approaches in an attempt to elucidate the action of gamma-butyrolactone (GBL), an illicit drug and a dietary supplement, that can cause coma and deaths in humans while exhibiting low systemic toxicity towards rodents. The lack of systemic toxicity of GBL and of its metabolite(s) was also predicted by validated SAR models. In fact using diverse SAR models, the only significant SAR prediction was that GBL had the potential for inhibiting human cytochrome P4502D6 (CYP2D6). However, inhibition of that isozyme is not necessarily associated with toxicity. It is suggested that GBL users also abuse other substances. When GBL inhibits CYP2D6 this may prevent the CYP2D6-mediated detoxification of other toxicants simultaneously consumed by the GBL user.