A human induced pluripotent stem cellbased model to test for embryotoxicity of chemicals and mixtures

Humans are continuously exposed to numerous exogenous chemicals from food, drugs, care products and our environment. For pregnant women in particular, this chemical exposure can be problematic as it can potentially harm the developing fetus. To ensure more descriptive, reliable and ethically sound toxicological screening of a large number of chemicals and drugs, it is paramount to develop and implement new approach methodologies (NAMs). The development of NAMs based on human cells for the assessment of embryotoxic effects is crucial to generate human relevant data.

The aim of this work was to contribute to the progress of the field of NAMs, specifically within the research area of human stem cell-based in vitro assays for developmental and/or embryotoxicity testing based on cardiomyocyte differentiation. The thesis is based on work performed to build upon the predictive capabilities of the PluriLum assay.

While several other developmental toxicity assays exist measuring the impact of chemicals on cardiomyocyte differentiation, these assays may present several disadvantages, including, for example: making use of ethically challenged human embryonic stem cells, not being representative of the in vivo scenario when conducted in 2D cultures; having long, and consequently expensive and laborious, protocols; or relying on subjective endpoints, such as visual detection of cardiomyocyte contractility. The PluriLum assay is currently the only available embryotoxicity 3D reporter gene assay based on cardiomyocyte differentiation of human induced pluripotent stem cells (hiPSCs), meaning that the model does not entail any relevant ethical issues, the assay is robust with an objective readout, and it relies on 3D embryoid bodies that mimic features of the very early blastocyst, as formed a few days after fertilization.

First, we investigated the applicability domain of the PluriLum assay by testing a panel of 10 chemicals in the assay. Our findings showed that the assay correctly classified two known nonembryotoxic chemicals – ascorbic and folic acid –, and two out of three well-established developmental toxicants – 5-fluorouracil and all-trans retinoic acid –, while failing to detect toxic effects from the teratogen valproic acid. The assay was also able to detect embryotoxicity of three per- and polyfluoroalkyl substances (PFAS), and the two fungicides, epoxiconazole and prochloraz. In addition, we demonstrated that the reporter-gene luminescence readout is, overall, more sensitive than the visual inspection of contractility.

Secondly, we optimized the assay protocol by making it more reliable as well as shortening the protocol. Based on our findings, we introduced a minimum threshold for the diameter of the forming embryoid bodies of 500 μm, resulting in more robust cardiomyocyte differential outcomes. Additionally, we were able to shorten the protocol length by 24 h, from a total of 8 to 7 days, without sacrificing, or even improving, the sensitivity of the assay.

Third, we investigated how the assay performed as a tool for assessments of “real-life” complex chemical mixtures extracted from water, fish, milk and serum, as part of the EU H2020 PANORAMIX project. Here we found that the assay is able to decern between the toxicity of the different matrices, contributing to the understanding of the toxicity arising from human exposure to chemicals from diverse sources.

Lastly, we used the PluriLum assay to screen the embryotoxicity of four individual legacy PFAS, and to assess the potential embryotoxicity of component-based mixtures of PFAS. We found that the assay was able to determine varying relative potency of the individual PFAS, and that effects of component-based mixtures, either equipotent or unbalanced, follow the principle of concentration addition for mixture effects. Importantly, by testing mixtures of PFAS prepared based on average European exposure levels, we showed a wide safety margin of exposure of adults and children to these substances in regard to potential embryotoxic effects. Additionally, we tested if the PluriLum assay could be used as a tool for revealing mechanistic insights into the toxicological effects of these substances. For this purpose, we conducted mRNA sequence analysis on cells exposed to the single PFAS, and we found some commonalities regarding impacted functional pathways/targets, such as ion channel activity and extracellular matrix components.

In conclusion, the work conducted here demonstrates that the PluriLum assay has the potential to become a tool for the assessment of embryotoxic effects of both single chemicals and mixtures. We envision the inclusion of the assay in a battery of complementary in vitro assays for embryotoxicity testing or as part of a Defined Approach for the assessment of embryotoxicity/developmental toxicity.