Uptake and effects of dioxin like compounds in fish : in vitro in vivo and cross species extrapolation using physiologically based toxicokinetic models

The past decades have witnessed an unparalleled interest of the public and of the scientific community in dioxins and dioxin-like compounds (DLCs) such as polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs), as well as dioxin-like polychlorinated biphenyls (dl-PCBs). DLCs belong to the group of persistent organic pollutants, which are characterised by low environmental degradation rates and environmental half-lives ranging from months to decades. In the water column, the strong adsorption of DLCs to suspended solids renders the sediment a long-term sink for, and potential secondary source of, pollution. Although DLC emissions have decreased substantially because of international efforts, they are still re-distributed in the environment by poorly understood processes. DLCs biomagnify in higher trophic levels of the aquatic food web due to their lipophilic character and can thus still pose a considerable risk to humans, e.g. via consumption of contaminated fish. Whether sediment-bound DLCs are also available to fish through non-dietary routes, e.g. during the re-suspension of sediments, is a contentious issue. Consequently, no sediment-based environmental quality standards (EQSs) for DLCs exist, despite their legal requirement by European directives such as the water framework directive (WFD) and the environmental quality standards directive (EQSD). DLCs are known to cause a plethora of acute and chronic toxic effects in humans and wildlife, including neuro-, immuno-, and hepatotoxic effects; they can also affect reproduction and increase the risk for specific cancers. All DLCs share planar structural properties and thus show high binding affinities for the cytosolic aryl hydrocarbon receptor (AhR). Many of the toxic effects associated with DLCs are believed to be mediated by AhR activation.Conventional chemical analyses of DLCs using high resolution gas chromatography and high resolution mass spectrometry are costly and require highly specialised personnel. It is evident that higher throughput methods are urgently required. Bioanalytical in vitro screening assays with permanent cell lines are one potential solution. Such bioassays are rapid and inexpensive and the resulting biological 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) equivalents (TEQs) provide an integrative measure of DLC contamination. On the other hand, many compounds with different physicochemical properties are subsumed under just one value. Consequently, it is difficult to characterise the distribution of TEQs between environmental compartments and within biota, and the predictive value of such assays for the in vivo effects of DLCs remains unclear. These aspects currently limit the use of bioassays in the assessment of contaminated sediment toxicity. The DioRAMA project, performed in collaboration between RWTH Aachen University and the German Federal Institute of Hydrology, aimed to overcome these limitations. The research presented in this thesis was conducted in close collaboration with the DioRAMA project. With the support of sophisticated computational methods, it aimed to develop a mechanistic understanding of: (a) how in vitro bioassay data related to the associated effects in vivo; (b) the toxicokinetic differences between fish species; (c) the differences and commonalities between the transcriptional response of fish to separate classes of DLCs; and (d) how the concentrations of DLCs in sediments relate to their availability to, and effects on, fish during sediment re-suspension.In the first study, a physiologically based toxicokinetic (PBTK) model for rainbow trout (Oncorhynchus mykiss) was used to extrapolate the results of in vitro bioassays to the in vivo level. In such models, different organs and tissues (i.e. liver, richly perfused tissues, poorly perfused tissues, kidney, and adipose fat) are explicitly represented. Each of the compartments is characterised by its volume, its total lipid and water contents, and blood flow to the compartment. PBTK models are capable of predicting a chemical’s internal concentration in the whole fish and in various tissues at any time during exposure; this facilitates analysis of the link between the biochemical effects and the chemical concentration at the site of toxic action, e.g. the liver. The most important biomarker for DLC exposure is probably hepatic 7-ethoxyresorufin-O-deethylase (EROD) activity. Previously published experimental in vivo median effect concentrations (EC50s) for EROD induction (based on aqueous concentrations) did not correlate with the in vitro EC50s. Recalculation of aqueous EC50s using the internal concentration in the liver calculated by the PBTK model resulted in an excellent correlation with the in vitro EC50s. This study confirmed that in vitro bioassays can be predictive of the in vivo effects if a meaningful measure of concentration is employed. Consequently, it is vital that PBTK models are available for a number of different fish species. It was the aim of the second study to apply the PBTK model to two new fish species: arguably the most important model species, zebrafish (Danio rerio), and the ecologically relevant non-model species, roach (Rutilus rutilus). These new models were combined with three existing ones. The resulting multi-species model allowed for cross-species extrapolation of the bioaccumulative potential of neutral organic compounds, including DLCs. Predictions were accurate for most compounds and the model can be used efficiently for cross-species extrapolation of a chemical’s potential for bioaccumulation. To demonstrate the applicability of the approach to fish species with a less common physiology (high lipid content, low accumulation rate), a PBTK model for the European eel (Anguilla anguilla) was developed in the third study. This facultative catadromous fish species has a complex life cycle. Its alarming population status might be – among other causes – the result of anthropogenic pollutants.Another aspect of the on-going discussion about the grouping of several DLCs under one TEQ is the assumption that one DLC can be substituted for another by taking their relative potency, i.e. their toxic equivalency factor (TEF), into account. While this clearly provides a convenient approach to simplifying the assessment of complex mixtures of compounds, it does not imply that different classes of DLCs have identical effects in fish. Approaching this common misconception was the main objective of the fourth study. Exposure of the aforementioned non-model fish species (roach) to three DLCs (TCDD, dl-PCB 156 and the dioxin-like polycyclic aromatic hydrocarbon, benzo[k]fluoranthene) led to differences in the transcriptome-wide responses determined by RNA sequencing. Interestingly, PCB 156 caused no substance-specific transcriptional changes, while TCDD and benzo[k]fluoranthene produced marked substance-specific effects.The two final studies described in this thesis investigated whether PBTK models could be used in combination with desorption models to predict the uptake of chemicals from sediments. Furthermore, the time- and concentration-dependent uptake of DLCs in rainbow trout and the associated effects were investigated. Semi-static 90-d exposure experiments were conducted using suspensions generated from serial dilution of three sediments with different contamination levels, collected from the rivers Rhine and Elbe. Particle-bound DLCs were readily bioavailable during re-suspension. Their uptake and the associated effects in fish (EROD activity, micronuclei and other nuclear aberrations, histopathological and gross pathological lesions) were proportional to their concentration in the sediment, thus clearly demonstrating that re-suspended sediments can provide a source of DLCs capable of accumulating in organisms and potentially producing adverse effects. The PBTK model predicted the measured concentrations of dl-PCBs in muscle tissue accurately.The present thesis clearly demonstrates that PBTK models provide powerful tools for diverse research questions in the context of DLC risk assessment. Three new PBTK models for roach, zebrafish and European eel were developed and applied to published data. For the first time, PBTK models were used for: (a) in vitro-in vivo extrapolation of mechanism-specific effects in fish; (b) cross-species extrapolation of the bioconcentration potential of chemicals; and (c) predicting the uptake of chemicals present in sediments by fish. Furthermore, it was demonstrated experimentally that sediment-bound DLCs are bioavailable when sediments are re-suspended. The sophisticated modelling approaches and experimental datasets presented within this thesis provide a powerful framework for the derivation of EQSs for DLCs in sediments and thus represent an important contribution to the achievement of the WFD and EQSD goals