ToxSci Advance Access originally published online on September 8, 2006
Toxicological Sciences 2007 95(1):5-12; doi:10.1093/toxsci/kfl103
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Published by Oxford University Press 2006.
The ToxCast Program for Prioritizing Toxicity Testing of Environmental Chemicals
National Center for Computational Toxicology (D343-03), Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711
1 To whom correspondence should be addressed. Fax: (919) 541-1194. E-mail: dix.david{at}epa.gov.
Received May 24, 2006; accepted August 30, 2006
 |
ABSTRACT |
|---|
The U.S. Environmental Protection Agency (EPA) is developing methods for utilizing computational chemistry, high-throughput screening (HTS), and various toxicogenomic technologies to predict potential for toxicity and prioritize limited testing resources toward chemicals that likely represent the greatest hazard to human health and the environment. This chemical prioritization research program, entitled "ToxCast," is being initiated with the purpose of developing the ability to forecast toxicity based on bioactivity profiling. The proof-of-concept phase of ToxCast will focus upon chemicals with an existing, rich toxicological database in order to provide an interpretive context for the ToxCast data. This set of several hundred reference chemicals will represent numerous structural classes and phenotypic outcomes, including tumorigens, developmental and reproductive toxicants, neurotoxicants, and immunotoxicants. The ToxCast program will evaluate chemical properties and bioactivity profiles across a broad spectrum of data domains: physical-chemical, predicted biological activities based on existing structure-activity models, biochemical properties based on HTS assays, cell-based phenotypic assays, and genomic and metabolomic analyses of cells. These data will be generated through a series of external contracts, along with collaborations across EPA, with the National Toxicology Program, and with the National Institutes of Health Chemical Genomics Center. The resulting multidimensional data set provides an informatics challenge requiring appropriate computational methods for integrating various chemical, biological, and toxicological data into profiles and models predicting toxicity.
Key Words: high-throughput screening; toxicogenomics; chemoinformatics; bioinformatics.
|
INTRODUCTION |
|---|
Across several U.S. Environmental Protection Agency (EPA) programs, there is a clear need to develop methods for evaluating large numbers of environmental chemicals for potential toxicity and to use the resulting information to prioritize the use of testing resources toward those chemicals and endpoints that present the greatest likelihood of risk to human health and the environment. This need can be addressed through the experience of the pharmaceutical industry in the use of state of the art, high-throughput screening (HTS), toxicogenomics, and computational chemistry tools for the discovery of new drugs (Table 1), with appropriate adjustments to the needs of environmental toxicology. Thus, a research program entitled "ToxCast" has been initiated within EPA to develop an ability to forecast toxicity based on bioactivity profiling. Ultimately, ToxCast's purpose is to develop methods of prioritizing chemicals for further screening and testing to assist EPA programs in the management and regulation of environmental contaminants.
|
Over the past decade, HTS has developed into a primary tool for drug discovery based upon bioactivity screening of the drugable proteome (Fliri et al., 2005b
; Janzen and Hodge, 2006). On a more limited scale, HTS has also been adapted to agrochemical discovery for the analysis of target species and model organisms (Smith et al., 2005; Tietjen et al., 2005). Recently, HTS applications to toxicology have been expanding as a useful complement to traditional toxicology (Bhogal et al., 2005; Fliri et al., 2005a; Kikkawa et al., 2006). In the federal sector, the National Institutes of Health Chemical Genomics Center (NCGC) has been established (http://www.ncgc.nih.gov/). The NCGC is using industrial-scale HTS technologies to collect data that is useful for developing small-molecule chemical probes for basic biological research (Austin et al., 2004). Traditional toxicology testing involves screening compounds through in vivo and in vitro tests focused on defined endpoints (e.g., neurotoxicity, developmental toxicity) or mechanisms of action (e.g., mutagenicity, cytotoxicity, regenerative hyperplasia). However, EPA is confronted with a large number of compounds to evaluate and faced with the difficulty of prioritizing scarce resources. Thus, environmental toxicology is challenged by (1) too many compounds to evaluate through endpoint-based in vivo testing and (2) inadequate models or knowledge of mechanism for many types of toxicity to design suitable in vitro testing. These challenges are also faced by other organizations including the U.S. Food and Drug Administration, European Union member countries, the Organization for Economic Cooperation and Development, and the regulated community (i.e., the pharmaceutical, agrochemical, and consumer products industries). There is an important need to distinguish between compounds that present little or no concern from those with the greatest likelihood of causing an adverse effect in the target species. High-throughput, high-content, and toxicogenomic screening methods applied to predictive toxicology provide opportunities for addressing these challenges.
The underlying hypothesis for ToxCast is that toxicological response is driven by interactions between chemicals and biomolecular targets. In most cases, these targets are part of the cellular proteome (e.g., receptors, ion channels, kinases). However, for most environmental chemicals the protein targets and biological effects underlying potential adverse effects have yet to be defined or characterized. Because suitable assays to query these have remained elusive, a more global approach of bioactivity profiling is a critical goal in environmental toxicology. This goal is embodied in the ToxCast program, which will focus on a multiple target matrix approach rather than a single target, directed vector approach. The matrix contains an expanded number of potential targets whose chemical interactions may be characterized by in silico models, biochemical assays, cell-based in vitro assays, and nonmammalian animal models.
|
ENABLING HTS AND TOXICOGENOMICS TECHNOLOGIES |
|---|
Modern computational chemistry and molecular and cellular biology tools allow researchers to characterize a broad spectrum of physical and biological properties for large numbers of chemicals (Bredel and Jacoby, 2004
; Table 1). Genomics, transcriptomics, proteomics, and metabolomics technologies are components of this modern molecular biology toolkit. However, though omics technologies produce large amounts of data per sample, they are not truly high throughput, and the per chemical cost can be significant. Thus, the primary driver transforming drug discovery has been HTS technologies (Macarron, 2006). HTS is comprised of assays in miniaturized format that can be either target or phenotype based. Target-based assays usually measure either binding or function of proteins cell free or in engineered cells. Phenotype-based assays monitor more complex endpoints in cells or whole organisms. These assays utilize small quantities of reagents and test chemicals and can be quite cost and time effective for analyzing larger numbers of chemicals.
The ability to generate broad-based bioactivity profiles for large libraries of compounds in coordinated portfolios of biochemical and cellular assays has become the norm in the pharmaceutical sciences for drug discovery. As bioactivity profiles for compound libraries have grown, the potential of these profiles for identifying off-target mechanisms and potential liabilities has begun to emerge (Bhogal et al., 2005; Fliri et al., 2005b,c; Klekota et al., 2006; Melnick et al., 2006). HTS technology optimized for drug discovery is now being refocused to applications in toxicological screening. It is important to appreciate, however, the significant and substantial differences between the application of HTS to pharmaceutical research versus environmental toxicology (Table 2). The chemical space and numbers, the targeting and potency, and most importantly the intolerance for false negatives are all key differences that will impact assay selection and study design for ToxCast. The aim of drug discovery HTS is to find a small number of active compounds amenable to subsequent optimization for drug development, and in this pursuit, false negatives are generally not a major concern. HTS for toxicology must determine the activity of all compounds tested, and false negatives are of greater concern from a public safety standpoint.
|
|
ENVIRONMENTAL CHEMICALS RELEVANT TO TOXCAST |
|---|
 TOPABSTRACTINTRODUCTIONENABLING HTS AND TOXICOGENOMICS... ENVIRONMENTAL CHEMICALS RELEVANT... DESIGN OF THE TOXCAST...SELECTION OF PROOF-OF-CONCEPT...INTEGRATING CHEMICAL AND...CONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
There are potentially 10,000 or more environmental chemicals from several EPA programs in need of prioritization for further testing. Antimicrobials, pesticidal inerts, high production volume (HPV: > 1 million lbs/year) chemicals, inventory update rule (> 10,000 lbs/year, < 1 million lbs/year) chemicals, and drinking water contaminant candidate list chemicals (Fig. 1) generally have limited toxicological data available for hazard and risk assessments. As ToxCast moves beyond initial proof-of-concept, thousands of environmental chemicals from various EPA domains can be considered for the ToxCast program. Looking beyond U.S. borders, there may be utility for a program like ToxCast in Europe's Registration, Evaluation, and Authorization of Chemicals (REACH) program. In 2003, the European Commission adopted the REACH proposal as a new regulatory framework for chemicals manufactured or imported at more than 1 ton per year. After final adoption of the REACH legislation, which is expected by the end of 2006, REACH legislation is likely to be in force by mid 2007 (http://ec.europa.eu/environment/chemicals/reach/reach_intro.htm).
|
For the ToxCast proof-of-concept, conventional chemical pesticide actives are an ideal set of compounds for a number of reasons. Currently, registered pesticide actives are relatively modest in number (about 800), yet these actives represent a fairly diverse set of structural classes (Table 3). Furthermore, these chemicals were all designed to have biological activity targeted against a pest species. This biological activity promises to provide a diverse range of positive results in the biochemical and cellular assays of ToxCast. Most importantly for the purposes of ToxCast, the pesticide actives have a wealth of uniform toxicological test data to inform hazard characterization. This existing information, and EPA's evaluation and interpretation of these data in current risk assessments, will provide the critical and necessary context for interpreting ToxCast data.
|
|
DESIGN OF THE TOXCAST RESEARCH PROGRAM |
|---|
ToxCast is designed to populate multiple data domains of increasing biological relevance and experimental cost, from in silico to in vitro, and perhaps even in vivo with nonmammalian model organisms (Fig. 2). Associations between data domains and across chemicals can be made in order to generate bioactivity fingerprints and to group or bin chemicals. It is these larger patterns gleaned from bioactivity profiling across a broad range of assays that can be associated with either chemical structure (Fliri et al., 2005b
,c; Melnick et al., 2006), or with known toxicity of reference chemicals in a proof-of-concept study. It is from these associations or correlations between chemical structure, bioactivity profile, and toxicity outcome that the predictive power of ToxCast will be derived. The chemical and biological diversity in ToxCast will afford an opportunity to establish qualitative connections and quantifiable linkages between chemical structure, biological activity, and known or predicted toxicity.
|
In the course of identifying screening targets (Table 4) or assays suitable for ToxCast (Table 5), two key considerations are the technical and economic feasibility of pursuing that target or assay for thousands of chemicals. Rather than just the drugable proteome, ToxCast sets out to survey a broad spectrum of genes, proteins, and metabolites that comprise the cellular "interactome." Pathway-based analyses may also identify effects on higher level signaling, in addition to discrete targets within the cellular interactome. These pathways could serve as a good middle ground between biochemical or other target-focused assays and more phenomenological, phenotypic, or high-content assays. Thus, the range of potential targets and assays is very broad, and increasing biological relevance will have to be balanced against increasing cost for various data domains (Fig. 2). Two abiding requirements for ToxCast assays will be the ability to minimize false negatives relating to hazards to human health and the current availability of these assays from reliable sources.
|
|
The majority of ToxCast data will come from a diverse series of assay types that collectively evaluate a broad spectrum of bioactivities (Fig. 3). Like prior examples in the literature (Fliri et al., 2005c
; Janzen and Hodge, 2006; Melnick et al., 2006), ToxCast data will include numerous HTS assays delineating biological effects. Eventually, ToxCast is designed to research thousands of chemicals, requiring a managed library of environmental chemicals and sophisticated chemical and biological informatics to identify meaningful data associations. HTS biochemical assays will be supplemented by cellular assays for more complex biological effects and toxicities (Table 4, Fig. 3). The potential of high-content screening (Borchert et al., 2005; O'Brien et al., 2006) and microelectronic monitoring (Xing et al., 2005) of cells for detection of specific toxicities will be researched. In addition, the use of the nematode Caenorhabditis elegans (in collaboration with the National Institute of Environmental Health Sciences [NIEHS]; Schwartz et al., 2004), and the zebrafish Danio rerio will be explored as models of mammalian toxicity. Biological samples from these various in vitro and in vivo assays will also be utilized for supplemental genomics and metabolomics.
|
While much of the ToxCast data are likely to come from HTS enzyme and receptor assays, an important complement to these data will be derived from assays using complex formats of human, nonhuman primate, or rodent cells for detecting biotransformation and complex toxicities. These are capable of detecting secondary effects (e.g., altered membrane permeability) resulting from chemically induced perturbations of the interactome. For example, in vitro primary hepatocyte models of the liver are commonly used to screen for metabolism and toxicity of xenobiotics, but primary hepatocytes rapidly lose liver-specific functions under standard cell culture conditions. Advances in tissue engineering and in silico modeling are enabling development of novel engineered approaches (Allen et al., 2005
; Sivaraman et al., 2005) that could improve chemical hazard testing by recreating the three-dimensional microscale of the liver. Such tissue engineering raises new possibilities for the study of complex toxicological processes in vitro (Griffith and Swartz, 2006), and the convergence of HTS and toxicogenomic data with systems biology is creating opportunities for developing bioengineered and computational models that more realistically replicate hepatic architecture and function. Toward this end, ToxCast could provide critical data for defining processes such as nuclear receptor–mediated regulation of xenobiotic-metabolizing enzymes, useful to systems biology models of toxicity. It is through systems biology that the issue of metabolism and biotransformation may be best addressed within ToxCast. A recent workshop organized by the European Center for the Validation of Alternative Methods (Coecke et al., 2006) emphasized the need to account for biotransformation with appropriate methods and to consider how such information can be incorporated into computer models for hazard identification.
Toxicogenomic assays, specifically the highly parallel profiling of gene expression and cellular metabolites in ToxCast biological samples can be an important adjunct or alternative to biochemical HTS profiling. For example, nuclear receptor binding and activity could be assessed by monitoring expression of suites of genes that are the transcriptional targets for specific nuclear receptors of interest. The appropriate target genes can be identified by a complementary suite of positive internal control ligands (e.g., testosterone for the androgen receptor, rifampicin for the human pregnane X receptor) utilized in ToxCast cellular assays. Receptor activities could then be assessed based on expression of receptor-modulated genes and utilized as an efficient toxicogenomics in vitro assay for characterization of environmental chemicals (Yang et al., 2006a).
|
SELECTION OF PROOF-OF-CONCEPT CHEMICALS |
|---|
The essential first step for the ToxCast program is to conduct a demonstration phase using reference chemicals that have an existing, rich toxicological database (i.e., registered chemical pesticide actives). Several hundred reference chemicals representative of differing structural classes and phenotypic outcomes (e.g., carcinogens, reproductive toxicants, neurotoxicants) will need to be evaluated in ToxCast's wide net of assays and endpoints for this proof-of-concept. As the program matures, the assays and endpoints may be narrowed or modified based on predictive value, derived from associations between various data domains and the known toxicological properties of the reference chemicals. From this proof-of-concept, a broader strategy for identifying toxicity potential, minimizing false negatives, and prioritizing subsequent testing can be developed for larger number of environmental chemicals having limited toxicological data. This proof-of-concept will be especially important because of the challenges of ToxCast, as compared to conventional drug discovery, attributable in part to the diversity of environmental chemicals and issues relating to solubility, volatility, or confounding cytotoxicity.
Working from EPA databases, 826 conventional chemical pesticide actives that are currently registered or undergoing registration were identified. Of these 826, at least 270 are food-use pesticides that have the most extensive testing requirements. Table 3 presents EPA's Office of Pesticide Programs (OPP) use categories and chemical classes for the majority of these pesticides. Table 5 lists the general selection criteria that were used for ranking chemical pesticide actives as candidates for the ToxCast proof-of-concept. Structural annotation was added to these pesticide actives, and further chemoinformatic analysis was conducted using LeadScope Enterprise (http://www.leadscope.com; Table 6). These 785 chemicals were characterized into 101 structural classes, with 28 of these classes being singletons. For proof-of-concept, the chemicals were prioritized based on several criteria. High priority was generally given to those chemicals in common with the 1408 chemicals that the National Toxicology Program (NTP) has provided NCGC in early 2006 for HTS. Compatibility with standard HTS assays was also considered; thus, low priority was given to inorganics, organometallics, high ALogP (octanol/water partitioning), and molecular weights < 150. The 328 prioritized chemicals were secondarily ranked in descending order of representation in other toxicological databases annotated in the EPA DSSTox Structure Data File collection (http://www.epa.gov/ncct/dsstox/index.html), or in other EPA programs (e.g., industrial HPV chemicals) that correlate in some fashion to ToxCast. A small minority of inorganics and organometallics are included in this set of 328 chemicals because of their relevance to other toxicological programs. The remaining chemicals included an additional 219 chemical pesticides that might be suited to HTS.
|
|
INTEGRATING CHEMICAL AND BIOLOGICAL DATA TO FORECAST TOXICITY |
|---|
Within ToxCast, data will be generated on an environmental chemical library using numerous types of assays evaluating a broad spectrum of bioactivities (Fig. 3). These data will need to be relationally linked within the ToxCast database to other physical-chemical, toxicological, and in silico information, and a structured strategy developed to predict toxicity based on this entire data set. This structured strategy will be forged upon the known toxicities of the proof-of-concept chemical pesticides.
We are currently in the process of collecting toxicological data on pesticides and working with the OPP on how to accurately and precisely capture this information into a relational database. The OPP evaluates submitted toxicological studies in a standardized review process, which is captured in a Data Evaluation Record (DER). Information is being culled from DERs on endpoints, dose-response, and critical effects in mammalian test species for approximately 400 chemical pesticides. The DERs being used are primarily from neurotoxicity, developmental, reproductive, subchronic, chronic, and cancer guideline toxicology tests. The OPP conventional toxicology for the proof-of-concept pesticides will complement the chemoinformatic, HTS, and toxicogenomic information in the ToxCast database, allowing us to develop and validate ToxCast's predictive power. In addition, toxicological data from other EPA Programs (e.g., HPV Challenge) and the NTP will also be helpful in developing ToxCast. Throughout the course of methods and data development, our goal is to keep ToxCast a public and transparent enterprise.
Another ongoing informatics effort is aimed at generating, collating, reviewing, and organizing unambiguous definitions of chemical identity and structure for the various environmental chemical domains relevant to ToxCast and EPA. To accomplish this, we are building on the DSSTox project. This will also aid in the identification of other potentially useful sources of data relative to the ToxCast candidate chemical list, as well as help identify structurally similar chemicals for which toxicity or bioassay data might be available.
Figure 4 presents a flowchart for applying ToxCast data and predictions to the process of prioritizing chemicals. Hazard prediction represents both the primary goal and the key bioinformatics and chemoinformatics challenge of this approach, and the value of such an approach is self-evident so long as false negatives are minimized. Over the past several years, a number of studies have been published presenting alternative, in vitro, and in some cases HTS methods for integrated testing of chemicals for bioactivity and associations with toxicity or side effects. One example relevant to ToxCast was an integrated, tiered approach using computational and experimental in vitro data for hazard assessment, although limited to only 10 environmental chemicals (Gubbels-van Hal et al., 2005). The hazard assessment for these 10 substances was performed on the basis of available nonanimal data, quantitative structure activity relationship, physiologically based pharmacokinetic modeling, and additional new in vitro testing. Based on these data, predictions of various toxicities were made and then compared with prior in vivo testing to demonstrate at least a partial success. However, the limited number of chemicals included in the study of Gubbels-van Hal et al. did not allow conclusions to be drawn for the thousands of chemicals subject to REACH. It is apparent that methods compatible with larger numbers of chemicals, which do not lead to substantially higher costs for industry, need to be developed. We suggest that HTS technologies, larger chemical libraries, and expanded data analysis techniques may accomplish these broader goals within ToxCast.
|
|
CONCLUSIONS |
|---|
The strategy of ToxCast encompasses a diverse range of data types. No single assay or endpoint will have a large impact on interpretation of the fingerprint or bioactivity profile. It will be the overall pattern across many assays and data types that will be the predictor of toxicity used for prioritizing chemicals. This will be the main goal of ToxCast, taking advantage of HTS and toxicogenomic technologies for bioactivity profiling of environmental chemicals related in structure or mechanism of action. Although the primary purpose is not to identify mechanisms of action of environmental toxicants per se, this might be a future benefit of the program. The availability of a biologically and chemically based system to categorize chemicals of like properties and activities will provide EPA Program Offices with a valuable tool that heretofore has been seriously lacking.
In late 2005, EPA organized the Chemical Prioritization Community of Practice (CPCP) to provide a forum for discussing the utility of computational chemistry, HTS, and various toxicogenomic technologies for chemical prioritization and Agency use. The CPCP has brought together experts and interested parties to discuss chemical prioritization research. This has afforded various groups the opportunity to consider the ToxCast concept, from EPA Program Offices, to external stakeholders such as the American Chemistry Council, the Center for Alternatives to Animal Testing, CropLife America and Environmental Defense. In addition, the CPCP has been helpful in building partnerships and communicating with the NTP, the NIEHS, and the NCGC.
Many hurdles remain to be cleared by ToxCast as it transits from concept to proof-of-concept and ultimately to a useful prioritization tool, including (1) accessing a chemical library providing coverage of sufficient chemical space, (2) identifying an upper limit on the per chemical cost of obtaining screening level data, (3) selecting assays within available resources that produce predictive bioactivity profiles, (4) evaluating the impact of metabolism on the efficiency and accuracy of assays, (5) developing a bioinformatic approach to mining ToxCast data and identifying predictive signatures, and (6) carrying out a prospective prioritization for chemicals currently entering a traditional testing process, in such a way that minimizes false negatives. These hurdles will be the focus of the ToxCast program over then next few years.
|
SUPPLEMENTARY DATA |
|---|
Supplementary data are available online at http://toxsci.oxfordjournals.org/.
|
NOTES |
|---|
Disclaimer: This work was reviewed by EPA and approved for publication but does not necessarily reflect official Agency policy.
|
ACKNOWLEDGMENTS |
|---|
We thank the many EPA, NTP, and NCGC colleagues who have supported our initial and ongoing efforts to develop ToxCast. Particular thanks are due to Maritja Wolf (Lockheed Martin, contractor to the EPA) for chemoinformatics assistance; Raymond Tice, Cynthia Smith, and Kristine Witt for coordination with the NTP HTS program; Chris Austin and Jim Inglese for coordination with the NCGC; Tina Levine, Elizabeth Mendez, Elissa Reaves, and Jess Rowland, for assistance with EPA/OPP toxicological data; and to Vicki Dellarco (EPA/OPP) and Phil Sayre (EPA/OPPT) for guidance and helpful comments in the review of this article.
|
REFERENCES |
|---|
Allen JW, Khetani SR, Bhatia SN. (2005) In vitro zonation and toxicity in a hepatocyte bioreactor. Toxicol. Sci. 84:110–119.
Austin CP, Brady LS, Insel TR, Collins FS. (2004) NIH Molecular Libraries Initiative. Science 306:1138–1139.
Berg EL, Kunkel EJ, Hytopoulos E, Plavec I. (2006) Characterization of compound mechanisms and secondary activities by BioMAP analysis. J. Pharmacol. Toxicol. Methods 53:67–74.[CrossRef][Medline]
Bhogal N, Grindon C, Combes R, Balls M. (2005) Toxicity testing: Creating a revolution based on new technologies. Trends Biotechnol. 23:299–307.[CrossRef][Web of Science][Medline]
Borchert KM, Galvin RJ, Frolik CA, Hale LV, Halladay DL, Gonyier RJ, Trask OJ, Nickischer DR, Houck KA. (2005) High-content screening assay for activators of the Wnt/Fzd pathway in primary human cells. Assay Drug Dev. Technol. 3:133–141.[CrossRef][Web of Science][Medline]
Bredel M and Jacoby E. (2004) Chemogenomics: An emerging strategy for rapid target and drug discovery. Nat. Rev. Genet. 5:262–275.[CrossRef][Web of Science][Medline]
Coecke S, Ahr H, Blaauboer BJ, Bremer S, Casati S, Castell J, Combes R, Corvi R, Crespi CL, Cunningham ML, 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]
Ekins S, Berbaum J, Harrison RK. (2003) Generation and validation of rapid computational filters for cyp2d6 and cyp3a4. Drug Metab. Dispos. 31:1077–1080.
Fliri AF, Loging WT, Thadeio PF, Volkmann RA. (2005a) Analysis of drug-induced effect patterns to link structure and side effects of medicines. Nat. Chem. Biol. 1:389–397.[CrossRef][Web of Science][Medline]
Fliri AF, Loging WT, Thadeio PF, Volkmann RA. (2005b) Biological spectra analysis: Linking biological activity profiles to molecular structure. Proc. Natl. Acad. Sci. USA. 102:261–266.
Fliri AF, Loging WT, Thadeio PF, Volkmann RA. (2005c) Biospectra analysis: Model proteome characterizations for linking molecular structures and biological response. J. Med. Chem. 48:6918–6925.[CrossRef][Web of Science][Medline]
Griffith LG and Swartz MA. (2006) Capturing complex 3D tissue physiology in vitro. Nat. Rev. Mol. Cell Biol. 7:211–224.[CrossRef][Web of Science][Medline]
Gubbels-van Hal WM, Blaauboer BJ, Barentsen HM, Hoitink MA, Meerts IA, van der Hoeven JC. (2005) An alternative approach for the safety evaluation of new and existing chemicals, an exercise in integrated testing. Regul. Toxicol. Pharmacol. 42:284–295.[CrossRef][Web of Science][Medline]
Janzen WP and Hodge CN. (2006) A chemogenomic approach to discovering target-selective drugs. Chem. Biol. Drug Des. 67:85–86.[CrossRef][Web of Science][Medline]
Kikkawa R, Fujikawa M, Yamamoto T, Hamada Y, Yamada H, Horii I. (2006) In vivo hepatotoxicity study of rats in comparison with in vitro hepatotoxicity screening system. J. Toxicol. Sci. 31:23–34.[CrossRef][Medline]
Klekota J, Brauner E, Roth FP, Schreiber SL. (2006) Using high-throughput screening data to discriminate compounds with single-target effects from those with side effects. J. Chem. Inf. Model 46:1549–1562.[CrossRef][Web of Science][Medline]
Macarron R. (2006) Critical review of the role of HTS in drug discovery. Drug Discov. Today 11:277–279.[CrossRef][Web of Science][Medline]
Melnick JS, Janes J, Kim S, Chang JY, Sipes DG, Gunderson D, Jarnes L, Matzen JT, Garcia ME, Hood TL, et al. (2006) An efficient rapid system for profiling the cellular activities of molecular libraries. Proc. Natl. Acad. Sci. USA. 103:3153–3158.
O'Brien SE and de Groot MJ. (2005) Greater than the sum of its parts: Combining models for useful ADMET prediction. J. Med. Chem. 48:1287–1291.[CrossRef][Web of Science][Medline]
O'Brien PJ, Irwin W, Diaz D, Howard-Cofield E, Krejsa CM, Slaughter MR, Gao B, Kaludercic N, Angeline A, Bernardi P, et al. (2006) High concordance of drug-induced human hepatotoxicity with in vitro cytotoxicity measured in a novel cell-based model using high content screening. Arch. Toxicol 80:580–604.[CrossRef][Web of Science][Medline]
Poroikov VV, Filimonov DA, Ihlenfeldt WD, Gloriozova TA, Lagunin AA, Borodina YV, Stepanchikova AV, Nicklaus MC. (2003) PASS biological activity spectrum predictions in the enhanced open NCI database browser. J. Chem. Inf. Comput. Sci. 43:228–236.[CrossRef][Web of Science][Medline]
Richard AM, Gold LS, Nicklaus MC. (2006) Chemical structure indexing of toxicity data on the internet: moving towards a flat world. Curr. Opin. Drug Discov. Dev 9:314–325.[Web of Science][Medline]
Scherf U, Ross DT, Waltham M, Smith LH, Lee JK, Tanabe L, Kohn KW, Reinhold WC, Myers TG, Andrews DT, et al. (2000) A gene expression database for the molecular pharmacology of cancer. Nat. Genet. 24:236–244.[CrossRef][Web of Science][Medline]
Schwartz DA, Freedman JH, Linney EA. (2004) Environmental genomics: A key to understanding biology, pathophysiology and disease. Hum. Mol. Genet. 13:R217–224.
Sivaraman A, Leach JK, Townsend S, Iida T, Hogan BJ, Stolz DB, Fry R, Samson LD, Tannenbaum SR, Griffith LG. (2005) A microscale in vitro physiological model of the liver: Predictive screens for drug metabolism and enzyme induction. Curr. Drug Metab. 6:569–591.[Medline]
Smith SC, Delaney JS, Robinson MP, Rice MJ. (2005) Targeting inputs and optimising HTS for agrochemical discovery. Comb. Chem. High Throughput Screen. 8:577–587.[CrossRef][Web of Science][Medline]
Tietjen K, Drewes M, Stenzel K. (2005) High throughput screening in agrochemical research. Comb. Chem. High Throughput Screen. 8:589–594.[CrossRef][Web of Science][Medline]
Walum E, Hedander J, Garberg P. (2005) Research perspectives for pre-screening alternatives to animal experimentation—On the relevance of cytotoxicity measurements, barrier passage determinations and high throughput screening in vitro to select potentially hazardous compounds in large sets of chemicals. Toxicol. Appl. Pharmacol. 207:393–397.[CrossRef][Medline]
Xing JZ, Zhu L, Jackson JA, Gabos S, Sun XJ, Wang XB, Xu X. (2005) Dynamic monitoring of cytotoxicity on microelectronic sensors. Chem. Res. Toxicol. 18:154–161.[CrossRef][Web of Science][Medline]
Yang Y, Abel SJ, Ciurlionis R, Waring JF. (2006a) Development of a toxicogenomics in vitro assay for the efficient characterization of compounds. Pharmacogenomics 7:177–186.[CrossRef][Web of Science][Medline]
Yang C, Richard AM, Cross KP. (2006b) The art of data mining the minefields of toxicity databases to link chemistry to biology. Curr. Comput. Aided Drug Des. 2:1–19.[CrossRef]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. T. Martin, E. Mendez, D. G. Corum, R. S. Judson, R. J. Kavlock, D. M. Rotroff, and D. J. Dix Profiling the Reproductive Toxicity of Chemicals from Multigeneration Studies in the Toxicity Reference Database Toxicol. Sci., July 1, 2009; 110(1): 181 - 190. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. R. Williams-Devane, M. A. Wolf, and A. M. Richard Toward a Public Toxicogenomics Capability for Supporting Predictive Toxicology: Survey of Current Resources and Chemical Indexing of Experiments in GEO and ArrayExpress Toxicol. Sci., June 1, 2009; 109(2): 358 - 371. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. E. Andersen and D. Krewski Toxicity Testing in the 21st Century: Bringing the Vision to Life Toxicol. Sci., February 1, 2009; 107(2): 324 - 330. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. W. Edwards and R. J. Preston Systems Biology and Mode of Action Based Risk Assessment Toxicol. Sci., December 1, 2008; 106(2): 312 - 318. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. C. K. Leung, P. L. Williams, A. Benedetto, C. Au, K. J. Helmcke, M. Aschner, and J. N. Meyer Caenorhabditis elegans: An Emerging Model in Biomedical and Environmental Toxicology Toxicol. Sci., November 1, 2008; 106(1): 5 - 28. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. D. Burgoon and T. R. Zacharewski Automated Quantitative Dose-Response Modeling and Point of Departure Determination for Large Toxicogenomic and High-Throughput Screening Data Sets Toxicol. Sci., August 1, 2008; 104(2): 412 - 418. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||