James K. Chipman

Development and application of the adverse outcome pathway framework for understanding and predicting chronic toxicity: I. Challenges and research needs in ecotoxicology

To elucidate the effects of chemicals on populations of different species in the environment, efficient testing and modeling approaches are needed that consider multiple stressors and allow reliable extrapolation of responses across species. An adverse outcome pathway (AOP) is a concept that provides a framework for organizing knowledge about the progression of toxicity events across scales of biological organization that lead to adverse outcomes relevant for risk assessment. In this paper, we focus on exploring how the AOP concept can be used to guide research aimed at improving both our understanding of chronic toxicity, including delayed toxicity as well as epigenetic and transgenerational effects of chemicals, and our ability to predict adverse outcomes. A better understanding of the influence of subtle toxicity on individual and population fitness would support a broader integration of sublethal endpoints into risk assessment frameworks. Detailed mechanistic knowledge would facilitate the development of alternative testing methods as well as help prioritize higher tier toxicity testing. We argue that targeted development of AOPs supports both of these aspects by promoting the elucidation of molecular mechanisms and their contribution to relevant toxicity outcomes across biological scales. We further discuss information requirements and challenges in application of AOPs for chemical- and site-specific risk assessment and for extrapolation across species. We provide recommendations for potential extension of the AOP framework to incorporate information on exposure, toxicokinetics and situation-specific ecological contexts, and discuss common interfaces that can be employed to couple AOPs with computational modeling approaches and with evolutionary life history theory. The extended AOP framework can serve as a venue for integration of knowledge derived from various sources, including empirical data as well as molecular, quantitative and evolutionary-based models describing species responses to toxicants. This will allow a more efficient application of AOP knowledge for quantitative chemical- and site-specific risk assessment as well as for extrapolation across species in the future.

DNA strand breakage in aquatic organisms and the potential value of the comet assay in environmental monitoring

This review considers the potential for DNA strand breaks, particularly as measured by the comet assay, to act as a biomarker of genetic toxicity in fish and other aquatic species. The background need for such biomarkers is introduced in relation to carcinogenicity, reproductive effects and other adverse effects of pollution. Sensitive measurements of DNA strand breakage can be achieved, e.g., by alkaline elution, alkaline unwinding or by single cell gel electrophoresis (comet) techniques. The DNA damage can be a reflection, not only of direct strand breakage, but also of alkali-labile sites and of repair enzyme-mediated breakage (i.e., is non-specific). A range of genotoxic chemicals (both with and without the requirement for metabolic activation) give positive effects in various cell types of vertebrate and invertebrate aquatic species, following in vitro and in vivo exposures under laboratory conditions. A limited number of analyses of organisms exposed to polluted waters or sediments in the field have implicated DNA strand breakage as a relatively sensitive, rapid and broad specificity indicator of genotoxic pollutant exposure. The comet assay deserves further exploitation to assess inter-individual and inter-cell variability in response to pollutants and naturally occurring genotoxic stimuli, and to assess the persistence of these effects.

A DNA expression array to detect toxic stress response in European flounder (Platichthys flesus).

As a first stage in developing a DNA array-based approach to investigating the effects of pollutants on an environmentally relevant European fish species, we have constructed a 160-gene custom microarray for European flounder. Degenerate primers were used to amplify 110 different fragments of stress-related and other genes from European flounder cDNA and genomic DNA. Additionally, 22 fragments were obtained by suppressive subtractive hybridisation (SSH). These fragments were cloned and sequenced, then, with additional control genes, used to create a cDNA microarray for flounder. After optimisation of the arraying process, hepatic mRNA was isolated from flounder caught in the polluted Tyne and relatively unpolluted Alde estuaries. Fluorescent cDNA probes were synthesised from the mRNA and used in dual-colour hybridisations to the microarray. A number of transcripts were differentially expressed between Tyne and Alde female flounder but these changes were not significant, due to high inter-individual variation. However, in comparisons between Tyne and Alde male flounder, 11 transcripts were found to significantly differ in expression (P<0.05). Seven transcripts were more highly expressed in the Tyne male fish (CYP1A, UDPGT, alpha-2HS-glycoprotein, dihydropyrimidine dehydrogenase, Cu/Zn SOD, aldehyde dehydrogenase and paraoxonase). Four transcripts (Elongation factor 1 (EF1), EF2, Int-6 and complement component C3) were found to be significantly less abundant in the Tyne male fish. Selected genes were assayed by real-time PCR, then normalised to alpha-tubulin. These assays confirmed the significance of the array results for CYP1A, UDPGT and EF1, but not for Cu/Zn SOD. This study provides a link between traditional single-gene biomarker studies and the emerging field of eco-toxicogenomics, demonstrating the utility of microarray studies on environmentally sampled, non-model organisms.

Adaptive differences in gene expression in European flounder (Platichthys flesus)

Population structure was previously believed to be very limited or absent in classical marine fishes, but recently, evidence of weakly differentiated local populations has been accumulating using noncoding microsatellite markers. However, the evolutionary significance of such minute genetic differences remains unknown. Therefore, in order to elucidate the relationship between genetic markers and adaptive divergence among populations of marine fishes, we combined cDNA microarray and microsatellite analysis in European flounders (Platichthys flesus). We demonstrate that despite extremely low levels of neutral genetic divergence, a high number of genes were significantly differentially expressed between North Sea and Baltic Sea flounders maintained in a long‐term reciprocal transplantation experiment mimicking natural salinities. Several of the differentially regulated genes could be directly linked to fitness traits. These findings demonstrate that flounders, despite little neutral genetic divergence between populations, are differently adapted to local environmental conditions and imply that adaptation in gene expression could be common in other marine organisms with similar low levels of population subdivision.

Identifying health impacts of exposure to copper using transcriptomics and metabolomics in a fish model.

Copper (Cu) is a micronutrient essential for the biochemical functioning of numerous processes in vertebrates but is also often present in the aquatic environment at concentrations able to cause adverse health effects in aquatic organisms. This study investigated the signaling pathways mediating the effects of exposure to Cu using a toxicogenomic approach in a fish model, the stickleback ( Gasterosteus aculeatus ). Freshwater-acclimated male fish were exposed via the water to Cu, including at environmentally relevant concentrations (3.2-128 microg of Cu/L for 4 days), and the biological responses explored through analyses of the hepatic transcriptome and metabolome and phenotypic end points, including assessment of DNA damage in blood cells. The Cu exposures resulted in DNA strand breaks in blood cells at all exposure concentrations and alterations in hepatic gene expression and metabolite concentrations in a concentration-dependent manner (from 10 microg of Cu/L). Genes associated with the cholesterol biosynthesis pathway were significantly over-represented and consistently down-regulated (at 128 microg of Cu/L), similar to that occurring in a mouse model for Wilsons disease. Additionally, inductions in metallothionein and catalase were also observed. The concentrations of NAD(+) and lactate increased significantly with the Cu exposure, consistent with a shift toward anaerobic metabolism, and these aligned closely with changes observed in gene expression. The pathways of Cu toxicity identified in our study support the conserved mechanisms of Cu toxicity from lower vertebrates to mammals, provide novel insights into the deleterious effects of Cu in fish, and further demonstrate the utility of fish as environmental sentinels for chemical impacts on both environmental and human health.

Non-genotoxic carcinogens: early effects on gap junctions, cell proliferation and apoptosis in the rat

Non-genotoxic carcinogens are thought to induce tumour formation by disturbing the balance between cell growth and cell death. Gap junctions (GJ) contribute to the maintenance of tissue homeostasis by allowing the intercellular exchange of growth regulatory signals and potential inhibition of GJ intercellular communication through loss of connexin (Cx) plaques has been shown to be involved in the cancer process. We have investigated the time- and dose-dependent effects of the non-genotoxic hepatocarcinogens Wy-14,643, 2,3,7,8-tetrachlorodibenzo-p-dioxin, methapyrilene and hexachlorobenzene and the male rat kidney carcinogens chloroform, p-dichlorobenzene and d-limonene on gap junction plaque expression in relation to proliferation and apoptosis. With the exception of limonene, all non-genotoxic carcinogens significantly reduced the expression of GJ plaques containing Cx32 in their respective target tissue. No dose-dependent, significant effects were seen in non-target organs. Although alteration of Cx32 expression did not appear to correlate with induction of cell proliferation, out data suggest that the interaction of both processes-interference of GJ coupled with a proliferative stimulus (at the carcinogenic dose)-may be important in non-genotoxic carcinogenesis and provide a potential alert for non-genotoxic carcinogens in short-term toxicity tests.

Hepatic transcriptomic profiles of European flounder (Platichthys flesus) from field sites and computational approaches to predict site from stress gene responses following exposure to model toxicants.

Genomic technologies offer opportunities to gain a more global assessment of the health status of an organism through an understanding of the functional pathways that are responding to pollutant exposure. We have developed a 13,000 clone cDNA toxicogenomics microarray for Platichthys flesus, the European flounder (EU-GENIPOL Project). We aimed to distinguish the origins of flounder taken from six sampling sites of different pollution status in Northern Europe according to their hepatic gene expression profile using bioinformatic approaches. To determine which gene expression differences may relate to pollutant impact, we have completed complementary laboratory exposures of flounder to selected toxicants and determined the associated gene expression profiles. Using multivariate variable selection coupled with a statistical modelling procedure (GALGO) we can predict geographical site but the accuracy is limited to specific sites. The search space for a combination of genes that effectively predicts class membership is very large, however, by combining the signatures derived from acute laboratory exposure to individual chemicals to limit the search space, a very accurate model for classification of all the different environmental sites was achieved. The final model utilised the expression profiles of 16 clones and validation with a qPCR array comprising these genes correctly assigned the site of origin for fish obtained from three of the sites in an independent sampling. These data would imply that the gene expression fingerprints obtained with these arrays are primarily attributable to variations in chemical pollutant responses at the different sites, indicating their potential utility in environmental impact assessment.

Development and validation of alternative metabolic systems for mutagenicity testing in short-term assays

We present here the results obtained within the framework of an EU funded project aimed to develop and validate alternative metabolic activating systems to be used in short-term mutagenicity assays, in order to reduce the use of laboratory animals for toxicology testing. The activating systems studied were established cell lines (Hep G2, CHEL), genetically engineered V79 cell lines expressing specific rat cytochromes P450, erythrocyte-derived systems, CYP-mimetic chemical systems and plant homogenates. The metabolically competent cell lines were used as indicator cells for genotoxic effects as well as for the preparation of external activating systems using other indicator cells. The following endpoints were used: micronuclei, chromosomal aberrations and sister chromatid exchanges, mutations at the hprt locus, gene mutations in bacteria (Ames test), unscheduled DNA synthesis and DNA breaks detected in the comet assay. All metabolic systems employed activated some promutagens. With some of them, promutagens belonging to many different classes of chemicals were activated to genotoxicants, including carcinogens negative in liver S9-mediated assays. In other cases, the use of the new activating systems allowed the detection of mutagens at much lower substrate concentrations than in liver S9-mediated assays. Therefore, the alternative metabolizing systems, which do not require the use of laboratory animals, have a substantial potential in in vitro toxicology, in the basic genotoxicity testing as well as in the elucidation of activation mechanisms. However, since the data basis is much smaller for the new systems than for the activating systems produced from subcellular liver preparations, the overlapping use of both systems is recommended for the present and near future. For example, liver S9 preparations may be used with some indicator systems (e.g., bacterial mutagenicity), and metabolically competent mammalian cell lines may be used with other indicator systems (e.g., a cytogenetic endpoint) in a battery of basic tests.

Hepatic transcriptomic and metabolomic responses in the Stickleback (Gasterosteus aculeatus) exposed to ethinyl-estradiol

An established three-spined stickleback (Gasterosteus aculeatus) cDNA array was expanded to 14,496 probes with the addition of hepatic clones derived from subtractive and normalized libraries from control males and males exposed to model toxicants. Microarrays and one-dimensional (1)H nuclear magnetic resonance (NMR) spectroscopy, together with individual protein and gene biomarkers were employed to investigate the hepatic responses of the stickleback to ethinyl-estradiol (EE(2)) exposure. Male fish were exposed via the water to EE(2), including environmentally relevant concentrations (0.1-100ng/l) for 4 days, and hepatic transcript and metabolite profiles, kidney spiggin protein and serum vitellogenin concentrations were determined in comparison to controls. EE(2) exposure did not significantly affect spiggin concentration but significantly induced serum vitellogenin protein at the threshold concentration of 32ng/l. (1)H NMR coupled with robust univariate testing revealed only limited changes, but these did support the predicted modulation of the amino acid profile by transcriptomics. Transcriptional induction was found for hepatic vitellogenins and choriogenins as expected, together with a range of other EE(2)-responsive genes. Choriogenins showed the more sensitive responses with statistically significant induction at 10ng/l. Real-time polymerase chain reaction (PCR) confirmed transcriptional induction of these genes. Phosvitinless vitellogenin C transcripts were highly expressed and represent a major form of the egg yolk precursors, and this is in contrast to other fish species where it is a minor component of vitellogenic transcripts. Differences in inducibility between the vitellogenins and choriogenins appear to be in accordance with the sequential formation of chorion and yolk during oogenesis in fish.

Interindividual Variability in Response to Sodium Dichromate–Induced Oxidative DNA Damage: Role of the Ser326Cys Polymorphism in the DNA-Repair Protein of 8-Oxo-7,8-Dihydro-2′-Deoxyguanosine DNA Glycosylase 1

Although the genotoxic mechanism(s) of hexavalent chromium (CrVI) carcinogenicity remain to be fully elucidated, intracellular reduction of CrVI and concomitant generation of reactive intermediates including reactive oxygen species and subsequent oxidative damage to DNA is believed to contribute to the process of carcinogenesis. In the current study, substantial interindividual variation (7.19-25.84% and 8.79-34.72% tail DNA as assessed by conventional and FPG-modified comet assay, respectively) in levels of DNA strand breaks after in vitro treatment of WBC with sodium dichromate (100 μmol/L, 1 hour) was shown within a group of healthy adult volunteers (n = 72) as assessed by both comet and formamidopyrimidine glycosylase–modified comet assays. No statistically significant correlation between glutathione S-transferases M1 or T1, NADPH quinone oxidoreductase 1 (codon 187) and X-ray repair cross complementation factor 1 (codon 194) genotypes and individual levels of DNA damage were observed. However, individuals homozygous for the Cys326 8-oxo 7,8-dihydro-2′-deoxyguanosine glycosylase 1 (OGG1) polymorphism had a statistically significant elevation of formamidopyrimidine glycosylase–dependent oxidative DNA damage after treatment with sodium dichromate when compared with either Ser326/Ser326 or Ser326/Cys326 individuals (P = 0.008 and P = 0.003, respectively). In contrast, no effect of OGG1 genotype on background levels of oxidative DNA damage was observed. When individuals were divided on the basis of OGG1 genotype, Cys326/Cys326 individuals had a statistically significant (P < 0.05, one-way ANOVA followed by Tukey test) higher ratio of oxidative DNA damage to plasma antioxidant capacity than either Ser326/Ser326 or Ser326/Cys326 individuals. The results of this study suggest that the Cys326/Cys326 OGG1 genotype may represent a phenotype that is deficient in the repair of 8-oxo-7,8-dihydro-2′-deoxyguanosine, but only under conditions of cellular oxidative stress. We hypothesize that this may be due to oxidation of the Cys326 residue. In conclusion, the homozygous Cys326 genotype may represent a biomarker of individual susceptibility of lung cancer risk in individuals that are occupationally exposed to CrVI.