Ecotoxicogenomics: linkages between exposure and effects in assessing risks of aquatic contaminants to fish

Reproductive Toxicology 19 (2005) 321–326

Review

Ecotoxicogenomics: linkages between exposure and effects in assessing risks of aquatic contaminants to fish Ann L. Miracle a,∗ , Gerald T. Ankley b a

U.S. Environmental Protection Agency, Ecological Exposure Research Division, Cincinnati, OH, USA b U.S. Environmental Protection Agency, Mid-Continent Ecology Division, Duluth, MN, USA Received 22 March 2004; received in revised form 8 June 2004; accepted 14 June 2004 Available online 17 July 2004

Abstract Understanding the biological effects of exposures to chemicals in the environment relies on classical methods and emerging technologies in the areas of genomics, proteomics, and metabonomics. Linkages between the historical and newer toxicological tools are currently being developed in order to predict and assess risk. Being able to classify chemicals and other stressors based on effects they have at the molecular, tissue, and organismal levels helps define a systems biology approach to development of streamlined, cost-effective, and comprehensive testing approaches for evaluating environmental hazards. The challenges of the individual technologies and the combinations of tools for ecotoxicogenomics are discussed in application to aquatic toxicology with a particular emphasis on fish testing. Published by Elsevier Inc. Keywords: Aquatic toxicology; Genomics; Exposure; Risk assessment

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Historical and current perspectives in fish toxicology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Omics technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Establishing linkages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Diagnostic and predictive risk assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

321 322 323 323 324 325

Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

325

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

325

1. 2. 3. 4. 5. 6.

1. Introduction Recently developed and emerging genomics technologies have significant potential implications for human and ecological risk assessment issues. Data generated from genomics, transcriptomics, proteomics, metabonomics, population genetics, histopathology, and ecology can be linked together through bioinformatics to generate a landscape of ∗ Corresponding author. Tel.: +1 513 569 7289; fax: +1 513 569 7609. E-mail address: [email protected] (A.L. Miracle).

0890-6238/$ – see front matter. Published by Elsevier Inc. doi:10.1016/j.reprotox.2004.06.007

events occurring within a given organism, or collection of organisms from source of stressor(s) through exposure and ultimately, to outcomes. Current ecological exposure and risk assessment models are not sufficient to examine a dose to effect continuum without extrapolation. Recognition of linkages connecting the data produced by these various tools serves to expand the scope of current technologies. Building exposure-to-outcome databases will become important in collecting diverse ecotoxicogenomic data sets

322

A.L. Miracle, G.T. Ankley / Reproductive Toxicology 19 (2005) 321–326

and defining systems biology for toxicology, which is ultimately required to effectively reduce uncertainties in comprehensive risk assessment for the environment and human health. The issues and challenges in aquatic ecotoxicology focused on fish will be used as an example of the need for linkages between classic and emerging toxicological disciplines.

2. Historical and current perspectives in fish toxicology A comprehensive review of the toxicology of fishes is beyond the scope of this paper; for the interested reader, authoritative references on the topic are available [1]. To help provide context for this article, however, we provide a brief overview of the nature of fish tests, including species and endpoints that commonly have been used for regulatory/monitoring programs. With this background, we then discuss how traditional tests with fish would/could be enhanced through the application of emerging computational toxicology approaches. Fish are, by far, the most numerous class of vertebrate animals in terms of number of species. They occupy virtually every aquatic habitat available and, in doing so, have evolved a rich diversity of life-history strategies which make it virtually impossible to identify any one species as a generalized model. A wide variety of both marine and freshwater fish species have been used in short-term lethality assays. However, partial- and full-life-cycle toxicity testing has focused on a much smaller number of species, usually selected for ease of culture and handling in a lab setting, ecological relevance (i.e., representative of other species in habitats of concern) and, occasionally, economic importance. Although there are some exceptions, most fish used for partial- and full-life-cycle tests have been freshwater rather than marine species. In the US and Europe, salmonids such as the rainbow trout (Oncorhynchus mykiss) have received a fair amount of historical attention with respect both to toxicology and physiology research. Many salmonids can be cultured under controlled conditions and they clearly are ecologically and economically relevant. However, the comparatively long life-cycle (1–2 years) and large size of most salmonids are not conducive to routine partial- or full-life-cycle testing. In other parts of the world (e.g., Japan), larger cyprinid species such as the common carp (Cyprinus carpio) and goldfish (Carassius auratus) also have commonly been used for toxicology and physiology research; however, these species have the same general draw-backs as salmonids in terms of body size and life-cycle duration. Because of the logistic constraints associated with the longer lived, large fish species, much of the current toxicology research with fish utilizes small “aquarium” species; three small freshwater fishes which are particularly widely used are the fathead minnow (Pimephales promelas), the Japanese medaka (Oryzias latipes) and the zebrafish (Danio rerio). There is some level of regional preference for use of the three species for regu-

latory testing, and each has somewhat different advantages and disadvantages. However, all share several similar positive attributes, including: (1) small body size (minimizing space needed for their maintenance/testing), (2) existence of standard, validated techniques for culture in the lab, (3) relatively rapid life-cycles, and (4) a large amount of existing information with respect to basic biological/toxicological attributes [2]. Testing with fish has evolved over the last 40 years to become a critical tool for determining the toxicity of single chemicals and complex mixtures in environmental risk assessments. Data from fish toxicity tests currently are used for a variety of regulatory programs in the US, including pesticide registration, Superfund site assessments, ambient water quality criteria, and screening/testing classes of chemicals of specific concern, such as endocrine disrupters. Early work with fish focused almost exclusively on short-term (generally 96 h) lethality assays with animals at different life-stages [3,4]. As recognition of the importance of long-term exposures and sub-lethal effects became more apparent, full-life-cycle testing became more common [3]. Life-cycle tests, however, are extremely resource intensive, and take months to complete, even in small fish models [2]. Thus, although life-cycle assays are comprehensive in terms of endpoint coverage they are not practical for routine testing. To address this, a variety of shorter term (weeks to months), partial-life-cycle tests have been developed for use in regulatory programs. The most commonly used partial-life-cycle tests emphasize different aspects of early development, as this seems to be a very sensitive life stage, at least to some classes of contaminants [3]. Early developmental tests typically start with newly fertilized eggs, and proceed through yolk-sac absorption (onset of active feeding), occasionally to juvenile life-stages. More recently, however, there has been a realization that life-stages other than early development, in particular active reproduction, can be very sensitive to chemicals with some modes/mechanisms of action (MOAs). Chemicals of concern here include those which disrupt reproductive endocrine function through alterations in the hypothalamic–pituitary–gonadal (HPG) axis. To address this, different variations of partial-life-cycle tests with reproductively active adult fish also have been developed and are starting to be more broadly used [2]. Most ecological risk assessments with toxic chemicals are focused upon possible adverse impacts at the population level [5,6]. Hence, for virtually all regulatory applications, endpoints most commonly assessed in toxicity tests with fish are those which can be directly related to the status of populations: survival, growth (generally during early life-stages), and measures of reproductive success (e.g., fecundity, fertility, hatch). Occasionally morphological alterations in exposed animals are assessed, but these are not typically used for quantitative assessments of risk. Although useful for predicting possible adverse effects of chemicals at the population level, an exclusive focus on these types of

A.L. Miracle, G.T. Ankley / Reproductive Toxicology 19 (2005) 321–326

“apical” endpoints can result in shortcomings in both diagnostic and predictive risk assessments [7].

3. Omics technologies The power of “omics” technologies is that they offer a dynamic picture of biological systems. Genomics defined here represents the transcription and repression of genes as a process that varies with metabolism and immediate environmental conditions. It is necessary to have the capability of examining scores of genes under a variety of exposure conditions in order to assess those which may be regulated by a specific stressor or exposure to provide a clearer picture of the acute and long term effects that an exposure exerts on gene expression. Information regarding transcriptional patterns that are indicative of expression changes for a specific exposure can then be discerned based on the patterns of transcriptional change. The use of DNA microarrays provides a high throughput diagnostic tool to screen the many variables required to effectively examine gene-expression patterns. Microarrays, either as cDNAs or oligonucleotides, permit thousands of genes to be screened in a single experiment to establish differential gene expression in stressor-treated versus control cell and tissue populations [8]. Microarrays have been used to study issues in pathology, pharmacology, oncology, cell biology, and recently, toxicology. With the advent of whole genome information readily available for select organisms, and the promise of additional, ecologically relevant organism genomic information, microarrays can be generated to sample a significant portion of the expressed genome. Microarray technology and more recently, protein characterization have revolutionized the ability to discern mechanistic pathways involved in disease processes, development, and toxicant action. For example, many types of human cancers can now be more accurately diagnosed as to subtype and prognosis based on correlation of molecular markers with histopathological data [9]. The combination of toxicology with proteomics and genomics has produced the field of toxicogenomics and is a widely investigated field in drug development [10]. To date, very little has been done to apply genomics technologies to ecological risk assessment, including aquatic species such as fish. This paucity is largely due to the lack of genome information and genomics tools for the classic environmental model species. However, recent genome sequencing efforts for several fish models such as zebrafish [11] and medaka [12] make these organisms potential candidates for large scale efforts to incorporate genomics technologies in understanding the mechanistic toxicity pathways for environmental stressors in aquatic settings. In addition, several smaller, “brute force” efforts are underway to develop genomic tools for other fish species more commonly used for environmental assessments of aquatic toxicity such as the fathead minnow [13], the European flounder (Platichthys flesus) [14], sheepshead minnow (Cyprinodon variega-

323

tus) [15], and large mouth bass (Micropterus salmoides) [16]. Proteomics technologies address the changes in protein expression between two distinct samples. Multiple techniques are employed to assess differential protein expression including various types of mass spectrometry, 2-dimensional (2D) gel electrophoresis, and more recently, protein arrays. While none of these individual techniques provides holistic information regarding expression, post-translational modifications, and identity of a protein, these approaches can be combined with microarray analyses to provide a clearer assessment of the impacts of a stressor on proteins within a sample, and to support and validate microarray data. Although proteomics is relatively new when compared with genomics, the use of proteomics is gaining importance in industry for biomarker and drug discovery [17,18] and as an application in clinical diagnostics by monitoring patterns of protein changes [19,20]. For ecotoxicogenomics, the field of proteomics is just starting to produce results. For example, Shrader et al. [21] recently published a proteomics study in zebrafish exposed to endocrine disrupting chemicals and describe two distinct patterns of protein expression on 2D gels that discriminated exposure to two separate chemicals with differing modes of action. More recent than proteomics is the advent of using metabonomics/metabolomics, which refers to the metabolic profile of a cell or the end products of the regulatory processes of a cell and hence cellular function. The measurement of metabolites specifically in biofluids is referred to a metabonomics, and can provide information regarding organ-specific toxicity profiles, identify biomarkers, and monitor onset and progress of toxic events [22]. The use of nuclear magnetic resonance (NMR) imaging and/or gas and liquid chromatography coupled to mass spectrometry provides identification and quantitation of metabolites which can be regarded as the end response of a biological system to an environmental change. Metabonomics provide a better assessment of a tissue response and leads to understanding the cellular metabolome. Metabonomic research would be very well suited for a systems biology or functional genomics analysis especially in combination with genomics and proteomics [23–25].

4. Establishing linkages Profiling of transcripts, proteins, and metabolites can help discriminate classes of toxicants and be helpful in understanding modes of action (MOAs) and disease [26]. However, the identification and importance of the endpoints of genomics and proteomics tools rely on the informatics employed to give the data biological meaning via statistical evaluations. Bioinformatics is widely recognized as the limiting step when it comes to interpretation of the wealth of data generated using microarrays and proteomics [27].

324

A.L. Miracle, G.T. Ankley / Reproductive Toxicology 19 (2005) 321–326

Given an increasing need to connect various outputs and endpoints from toxicological studies, genomics, proteomics, and phenotype or pathology, the computational challenges required to link these data together will require collaborative efforts across government, academia, and industry. Efforts to make these associations are already evident with database developments like those described by Kriete et al. [28] to combine histomorphometric analyses with gene expression. The extensible morphometric relational gene-expression analysis (EMeRGE) links information regarding regulation of genes with possible downstream effects. Perhaps a more complementary database, the Chemical Effects in Biological Systems (CEBS), incorporates toxicology data from genomics, proteomics, metabolomics and conventional toxicology and is being developed by the National Center for Toxicogenomics, National Institute of Environmental Health Sciences [29]. Although the aquatic models mentioned in the manuscript are not currently represented in the CEBS database, there is considerable interest in using CEBS for cross-species expression profiling that ranges from microbial to mammalian organisms. With the recent advent of metabolomics/metabonomics technologies, there are also efforts being made through the Consortium for Metabonomic Toxicology (COMET) project to generate databases that help to predict toxic effects of xenobiotics [30]. Development of these various databases encourages the disparate “omics” disciplines to tie together the various components to gain an understanding of the changes in the biological systems in contact with a stressor, and to assess the ecological risk of these changes.

5. Diagnostic and predictive risk assessment Diagnostic, or retrospective, assessments usually deal with observed alterations in populations of animals or, occasionally, impacts on individuals that could be due to any of a number of causative agents, chemical or otherwise. Further, in these situations there are invariably multiple potential stressors, only a subset of which (perhaps one) have any role in producing the adverse outcome of concern. Examination of apical endpoints lend little or no insight into causative modes of action (MOAs). Remedial action in this

type of situation is impossible in the absence of mechanistic data from lower levels of biological organization (Fig. 1). Even in situations where there are, perhaps significant, resources available for testing, knowledge of possible MOAs could guide test and endpoint selection in predictive assessments. For example, if short-term gene-expression assays with fish indicate a reasonable likelihood that a pesticide would affect some aspect of the HPG axis, subsequent fish assays could be focused on aspects of sexual development and active reproduction as opposed, for example, to survival and growth during early development. In addition to helping to prioritize chemicals for testing and aid in test selection, MOA information afforded by knowledge of effects data at lower levels of biological organization could help address a number of other uncertainties associated with predictive assessments. For example, risk assessments for mixtures currently are problematic and feasible only for groups of chemicals (e.g., Ah receptor agonists, organophosphate pesticides) for which extensive research has established/documented a common MOA. If reliable models/techniques were available to identify chemicals with similar MOA, consideration of potential mixture toxicity could be made much earlier in tiered risk assessments. Species extrapolation also is a challenge for ecological risk assessments. An understanding of MOA resulting from collection of data from lower levels of biological organization would help reduce uncertainty associated with extrapolation to fish, or even other vertebrates [7]. As is true for diagnostic assessments, predictive (prospective) risk assessments based on data from fish toxicity tests are somewhat limited by reliance solely on apical endpoints, albeit for different reasons. An increasing emphasis on collection/use of information from lower levels of biological organization would result in more efficient and effective evaluations of greater numbers of chemicals by regulatory bodies such as U.S. Environmental Protection Agency. Specifically, due to resource limitations, comparatively few chemicals undergo fish testing as part of processing/registration. This results in extensive data for a few chemicals (e.g., pesticides) and little (or no) empirical data for thousands of other existing/new chemicals. If it were possible to use validated techniques based on gene expression (transcriptomics), protein expression, or metabolite profiles to assign chemicals to MOA classes,

Fig. 1. Conceptual linkages across biological levels of organization in ecological risk assessments.

A.L. Miracle, G.T. Ankley / Reproductive Toxicology 19 (2005) 321–326

the regulatory process could consider, in a timely and inexpensive fashion, the potential toxicity of any chemical for which there may be concern. The omics technologies would not be the risk assessors’ tools in and of themselves, but would provide a “landscape” of probable effects which would be linked with phenotypic endpoints indicative of chemical effects. Those chemicals which are assigned to MOA classes of particular concern then could be flagged and prioritized in terms of targeting the limited resources available for longer term partial- and full-life-cycle fish tests to validate the predicted outcomes.

6. Conclusions So, given the multitude of potential advantages that knowledge of MOA brings to both diagnostic and predictive risk assessments, why is this information not more broadly sought or used? Fig. 1 illustrates the answer to this question. Specifically, an understanding and/or identification of MOA can be achieved through collecting data at lower biological levels of organization, usually at the molecular and cellular levels. But, the significance of alterations at these levels of organization generally is uncertain in terms of ultimate adverse outcomes at the individual and population levels. To address this uncertainty requires an understanding of linkages between events occurring at different levels of organization. This can be achieved through research focused on the delineation of “toxicity pathways”, defined here as the cascade of events that occur from a molecular “initiating” event (e.g., receptor interactions, enzyme induction or inhibition, DNA binding, etc.) through responses at the cellular and tissue levels to culmination in an adverse outcome in the animal. Only through systematic efforts to generate knowledge of this type can mechanistic information achieve its full potential utility in diagnostic and predictive assessments of the risk of toxic chemicals in tests with fish or any other model species used for ecological risk assessment.

Acknowledgments This document has been reviewed in accordance with U.S. Environmental Protection Agency policy and approved for publication. Approval does not signify that the contents necessarily reflect the views or policies of the Agency nor does mention of trade names or commercial products constitute endorsement or recommendation for use. References [1] Di Giulio R, Hinton DE, editors. Toxicology of fishes. London, UK: Taylor and Francis, in press. [2] Ankley GT, Johnson RD. Small fish models for identifying and assessing the effects of endocrine-disrupting chemicals. ILAR J, in press.

325

[3] McKim JM. Evaluation of tests with early life stages of fish for predicting long term toxicity. J Fish Res Board Can 1977;34:1148– 54. [4] Mount DI. Present approaches to toxicity testing—a perspective. In: Mayer FL, Hamelink JL, editors. Aquatic toxicology and hazard evaluation, STP 634. Philadelphia, PA: American Society for Testing and Materials; 1977. p. 5–17. [5] Suter II GW, Rosen AE, Linuer E, Parkhurst DF. Endpoints for responses of fish to chronic toxic exposures. Environ Toxicol Chem 1987;6:793–809. [6] Kavlock RJ, Ankley GT. A perspective on the risk assessment process for endocrine-disruptive effects on wildlife and human health. Risk Anal 1996;16:731–9. [7] Ankley GT, Johnson RD, Detenbeck NE, Bradbury SP, Toth G, Folmar LC. Development of a research strategy for assessing the ecological risk of endocrine disruptors. Rev Toxicol 1997;1:71–106. [8] Nuwaysir EF, Bittner M, Trent J, Barrett JC, Afshari CA. Microarrays and toxicology: the advent of toxicogenomics. Mol Carcinog 1999;24:153–9. [9] Luo J, Issacs WB, Trent JM, Duggan DJ. Looking beyond morphology: cancer gene expression profiling using DNA microarrays. Cancer Invest 2003;21:937–49. [10] Guerreiro N, Staedtler F, Grenet O, Kehren J, Chibout SD. Toxicogenomics in drug development. Toxicol Pathol 2003;31:471–9. [11] Rasooly RS, Henken D, Freeman N, Tompkins L, Badman D, Briggs J, et al. Genetic and genomic tools for zebrafish research: the NIH zebrafish initiative. Dev Dyn 2003;228:490–6. [12] Henrich T, Ramialison M, Quiring R, Wittbrodt B, Furutani-Seiki M, Wittbrodt J, et al. MEPD: a medaka gene expression pattern database. Nucleic Acids Res 2003;31(1):72–4. [13] Miracle AL, Toth G, Lattier DL. The path from molecular indicators of exposure to describing dynamic biological systems in an aquatic organism: microarrays and the fathead minnow. Ecotoxicology 2003;12:457–62. [14] Williams TD, Gensberg K, Minchin SD, Chipman JK. A DNA expression array to detect toxic stress response in European flounder (Platichthys flesus). Aquat Toxicol 2003;65(2):141–57. [15] Larkin P, Folmar LC, Hemmer MJ, Poston AJ, Denslow ND. Expression profiling of estrogenic compounds using a sheepshead minnow cDNA macroarray. EHP Toxicogenomics 2003;111:29– 36. [16] Larkin P, Sabo-Attwood T, Kelso J, Denslow ND. Analysis of gene expression profiles in largemouth bass exposed to 17-beta-estradiol and to anthropogenic contaminants that behave as estrogens. Ecotoxicology 2003;12:463–8. [17] Celis JE, Gromov P, Cabezon T, Moreira JM, Ambartsumian N, Sandelin, et al. Proteomic characterization of the interstitial fluid perfusing the breast tumor microenvironment: a novel resource for biomarker and therapeutic target discovery. Mol Cell Proteomics 2004 [Epub]. [18] Zolg JW, Langen H. How industry is approaching the search for new diagnostic markers and biomarkers. Mol Cell Proteomics 2004 [Epub]. [19] Clarke W, Zhang Z, Chan DW. The application of clinical proteomics to cancer and other diseases. Clin Chem Lab Med 2003;41:1562– 70. [20] Petricoin EE, Paweletz CP, Liotta LA. Clinical applications of proteomics: proteomic pattern diagnostics. J Mammary Gland Biol Neoplasia 2002;7:433–40. [21] Shrader EA, Henry TR, Greeley MS, Bradley BP. Proteomics in zebrafish exposed to endocrine disrupting chemicals. Ecotoxicology 2003;12:485–8. [22] Reo NV. NMR-based metabolomics. Drug Chem Toxicol 2002;25:375–82. [23] Forster J, Gombert AK, Nielsen J. A functional genomics approach using metabolomics and in silico pathway analysis. Biotechnol Bioeng 2002;79:703–12.

326

A.L. Miracle, G.T. Ankley / Reproductive Toxicology 19 (2005) 321–326

[24] Viant MR, Rosenblum ES, Tjeerdema RS. NMR-based metabolomics: a powerful approach for characterizing the effects of environmental stressors on organism health. Environ Sci Technol 2003;37:4982–9. [25] Weckwerth W. Metabolomics in systems biology. Annu Rev Plant Biol 2003;54:669–89. [26] Amin RP, Hamadeh HK, Bushel PR, Bennett L, Afshari CA, Paules RS. Genomic interrogation of mechanism(s) underlying cellular responses to toxicants. Toxicology 2002;181/182:555– 63. [27] Yu U, Lee SH, Kim YJ, Kim S. Bioinformatics in the post-genome era. J Biochem Mol Biol 2004;37:75–82.

[28] Kriete A, Anderson MK, Love B, Freund J, Caffrey JJ, Young MB, et al. Combined histomorphometric and gene-expression profiling applied to toxicology. Genome Biol 2003;4:R32. [29] Waters M, Boorman G, Bushel P, Cunningham M, Irwin R, Merrick A, et al. Systems toxicology and the Chemical Effects in Biological Systems (CEBS) knowledge database. EHP Toxicogenomics 2003;111:811–24. [30] Lindon JC, Nicholson JK, Holmes E, Antti H, Bollard ME, Keun H, et al. Contemporary issues in toxicology: the role of metabonomics in toxicology and its evaluation by the COMET project. Toxicol Appl Pharmacol 2003;187:137– 46.