To Örebro University

Environmental pollutants often induce morphological alterations in developing organisms, yet assessments are commonly subjective, limiting reproducibility and sensitivity. We developed and validated a semi-automated brightfield high-content imaging (HCI) pipeline to quantitatively detect morphological changes in zebrafish embryos. Using FishInspector software, we adapted image analysis for microscopy systems without automated embryo positioning, extending applicability across standard laboratory setups.To validate the approach, zebrafish embryos were exposed for 96 hours to two previously characterized pollutant mixtures (PFOS + PCB126; PFOS + B[a]P + arsenate) known to cause developmental effects. The pipeline sensitively quantified phenotypes including reduced swim bladder and shortened body length. These endpoints reflect developmental delay, highlighting the method's ability to capture mechanistically relevant effects. Such changes may reduce physiological performance and behavior, ultimately impacting fish populations.While earlier subjective scoring identified some similar alterations, our findings underscore the advantages of quantitative, semi-automated morphology assessment. The method improves reproducibility, enables standardized comparisons across studies, and increases sensitivity to detecting subtle morphological effects. By integrating brightfield imaging with semi-automated analysis, this approach broadens the toxicological toolbox for developmental hazard assessment and mixture toxicity research.

Polycyclic aromatic compounds (PACs) are prevalent in the environment, typically found in complex mixtures and high concentrations. Our understanding of the effects of PACs, excluding the 16 priority polycyclic aromatic hydrocarbons (16 PAHs), remains limited. Zebrafish embryos and in vitro bioassays were utilized to investigate the embryotoxic, behavioral, and molecular effects of a soil sample from a former gasworks site in Sweden. Additionally, targeted chemical analysis was conducted to analyze 87 PACs in the soil, fish, water, and plate material. CALUX® assays were used to assess the activation of aryl hydrocarbon and estrogen receptors, as well as the inhibition of the androgen receptor. Larval behavior was measured by analyzing activity during light and darkness and in response to mechanical stimulation. Furthermore, qPCR analyses were performed on a subset of 36 genes associated with specific adverse outcomes, and the total lipid content in the larvae was measured. Exposure to the sample resulted in embryotoxic effects (LC50 = 0.480 mg dry matter soil/mL water). The mixture also induced hyperactivity in darkness and hypoactivity in light and in response to the mechanical stimulus. qPCR analysis revealed differential regulation of 15 genes, including downregulation of opn1sw1 (eye pigmentation) and upregulation of fpgs (heart failure). The sample caused significant responses in three bioassays (ERα-, DR-, and PAH-CALUX), and the exposed larvae exhibited elevated lipid levels. Chemical analysis identified benzo[a]pyrene as the predominant compound in the soil and approximately half of the total PAC concentration was attributed to the 16 PAHs. This study highlights the value of combining in vitro and in vivo methods with chemical analysis to assess toxic mechanisms at specific targets and to elucidate the possible interactions between various pathways in an organism. It also enhances our understanding of the risks associated with environmental mixtures of PACs and their distribution during toxicity testing.

Current environmental risk assessment of polluted sites primarily relies on single compound evaluation. However, in the environment, organisms are often exposed to complex mixtures of pollutants. To further develop risk assessment of polluted sites and evaluate the risks that mixtures pose to humans and wildlife, a mechanistic understanding of mixture toxicity is needed.

The overall aim of this thesis was to increase our knowledge of the toxic effects caused by chemical mixtures and to develop new approaches to investigate the molecular mechanisms underlying such effects. To reach this aim, a comprehensive set of methods was applied, considering molecular and phenotypical alterations as well as chemical analyses.

The investigations revealed that the acute toxicity caused by mixtures of the pollutants B[a]P, PFOS, PCB126, and Arsenate is mainly predictable by concentration addition. The results also showed some specific sublethal effects of the various mixtures that were not observed for the single components. In addition, each mixture caused very specific patterns of behavioral alterations, gene expression changes, altered lipid content, and altered organ growth. A complex environmental mixture from soil contaminated with PACs caused for instance behavioral alterations in zebrafish, in addition to dysfunction of genes that are critical for eye development.

In summary, this thesis contributes to an increased understanding of the mechanistic pathways underlying the mixture toxicity of selected pollutants and environmental samples. In addition, it provides insights for the development of new approaches that may be included in risk assessment, such as image analysis and effect-directed analysis.

Risk assessment of chemicals is still primarily focusing on single compound evaluation, even if environmental contamination consists of a mixture of pollutants. The concentration addition (CA) and independent action (IA) models have been developed to predict mixture toxicity. Both models assume no interaction between the components, resulting in an additive mixture effect. In the present study, the embryo toxicity test (OECD TG no. 236) with zebrafish embryos (Danio rerio) was performed to investigate whether the toxicity caused by binary, ternary, and quaternary mixtures of organic (Benzo[a]pyrene, perfluorooctanesulfonate, and 3,3´,4,4´,5-pentachlorobiphenyl 126) and inorganic (arsenate) pollutants can be predicted by CA and IA. The acute toxicity and sub-lethal alterations such as lack of blood circulation were investigated. The models estimated the mixture toxicity well and most of the mixtures were additive. However, the binary mixture of PFOS and PCB126 caused a synergistic effect, with almost a ten-fold difference between the observed and predicted LC₅₀-value. For most of the mixtures, the CA model was better in predicting the mixture toxicity than the IA model, which was not expected due to the chemicals' different modes of action. In addition, some of the mixtures caused sub-lethal effects not observed in the single compound toxicity tests. The mixture of PFOS and BaP caused a division of the yolk and imbalance was caused by the combination of PFOS and As and the ternary mixture of PFOS, As, and BaP. Interestingly, PFOS was part of all three mixtures causing the mixture specific sub-lethal effects. In conclusion, the present study shows that CA and IA are mostly resulting in good estimations of the risks that mixtures with few components are posing. However, for a more reliable assessment and a better understanding of mixture toxicity, further investigations are required to study the underlying mechanisms.

The numbers of potential neurotoxicants in the environment are raising and pose a great risk for humans and the environment. Currently neurotoxicity assessment is mostly performed to predict and prevent harm to human populations. Despite all the efforts invested in the last years in developing novel in vitro or in silico test systems, in vivo tests with rodents are still the only accepted test for neurotoxicity risk assessment in Europe. Despite an increasing number of reports of species showing altered behaviour, neurotoxicity assessment for species in the environment is not required and therefore mostly not performed. Considering the increasing numbers of environmental contaminants with potential neurotoxic potential, eco-neurotoxicity should be also considered in risk assessment. In order to do so novel test systems are needed that can cope with species differences within ecosystems. In the field, online-biomonitoring systems using behavioural information could be used to detect neurotoxic effects and effect-directed analyses could be applied to identify the neurotoxicants causing the effect. Additionally, toxic pressure calculations in combination with mixture modelling could use environmental chemical monitoring data to predict adverse effects and prioritize pollutants for laboratory testing. Cheminformatics based on computational toxicological data from in vitro and in vivo studies could help to identify potential neurotoxicants. An array of in vitro assays covering different modes of action could be applied to screen compounds for neurotoxicity. The selection of in vitro assays could be guided by AOPs relevant for eco-neurotoxicity. In order to be able to perform risk assessment for eco-neurotoxicity, methods need to focus on the most sensitive species in an ecosystem. A test battery using species from different trophic levels might be the best approach. To implement eco-neurotoxicity assessment into European risk assessment, cheminformatics and in vitro screening tests could be used as first approach to identify eco-neurotoxic pollutants. In a second step, a small species test battery could be applied to assess the risks of ecosystems.