Het gebruik van iPSC-afgeleide hepatocytachtige cellen bij hepatotoxiciteitstesten

Abstract Due to rapid advances in the pharmaceutical, cosmetics, nanomaterial, and other industries, humans are constantly exposed to a wide variety of chemicals at all times. The assessment of risks posed by these chemicals to human health is thus necessary. Pharmaceuticals, herbal products, and cosmetics in particular have greater potential to cause adverse effects due to the increased exposure concentrations. Most of these chemicals have been tested in whole animal models in the past decades, and the majority of the available exposure, metabolism, and toxicity information are based on these studies. However, the differences in phase I and II metabolising enzymes and transporters in non-human models negatively impact risk assessment accuracy and extrapolation to chemical responses in humans. This is also a major reason for the high cost and poor efficiency of drug development, resulting in black box warnings on many commonly prescribed drugs, and unwanted side effects from herbal products and cosmetics. In addition, it is extremely time consuming and prohibitively expensive to perform risk assessment studies on animals. The ethical concerns of using animals for safety evaluation have led to the directive 2010/63/EU by the European Union, for the use of Reduction, Refinement and Replacement (3Rs) principles. Already, sale of cosmetic products that use animals for risk assessment studies has been banned in Europe since 2009 (EC 1223/2009). Moreover, the United States Environmental Protection Agency issued a directive in 2019, outlining their plans to eliminate animal testing from regulatory requirements for pesticides and industrial chemicals by 2035 . Thus, initiatives are needed for the development of animal-free chemical safety evaluation models that are predictive for human toxicological responses and can be put into use by industry and regulatory bodies. As an alternative, one can use human in vitro models, which are potentially better models due to improved predictive abilities, and offer the possibility of high-throughput studies and multiplexing. As the liver performs most of the metabolism and biotransformation of xenobiotic substances in the body, there is a particular need for good in vitro liver, especially hepatocyte, models. Current in vitro models of hepatotoxicity include primary human hepatocytes (PHHs) or stable transformed cell lines of hepatic origin, such as the HepG2 and HepaRG models. Although primary human hepatocytes are the gold standard for in vitro cellular risk assessment, donor-specific polymorphisms can lead to differences in drug metabolism and clearance rates, and this can cause batch-to-batch differences. Moreover, PHHs rapidly lose mature hepatic functions in two-dimensional (2D) culture, rendering them unsuitable for long-term, but also shorter term, exposure studies. Many of the transformed cell line models do not have the drug/chemical biotransformation capacity comparable to that of freshly isolated PHHs. In addition, mutations in the cell lines are amplified upon repeated passaging, and may affect their susceptibility to certain stressors. After the discovery by Yamanaka and colleagues in 2006 that nucleated somatic cells can be reprogrammed to induced pluripotent stem cells (iPSCs), an inexhaustible human cell population can now be created that theoretically can differentiate into any human cell type. Several research groups have improved upon the initial reprogramming protocol and made the generation of iPSCs from adult somatic cells faster and more efficient. iPSC-derived cells are being studied for their potential use as predictive and mechanistic risk assessment models due to their advantages over other models. These are discussed in further detail in section 1.4. As part of the in3 project (Marie Sklodowska-Curie Action-Innovative Training Network 2017-2020 grant no. 721975, www.estiv.org/in3), iPSCs obtained from StemBANCC were differentiated to different cells representing lung, kidney, liver, and brain, for toxicity risk assessment studies, and the work on hepatocytic progeny is presented here. The Verfaillie lab recently described that overexpression of three transcription factors (TFs; termed 3x) combined with the use of a metabolically improved medium (consisting of high concentrations of amino acids and glycine, termed AAGly) generates PSC-derived hepatocyte-like cells (HLCs) that are transcriptionally more mature and display improved cytochrome functional activity. These cells can also be maintained in culture without loss of function for at least 50 days. The development of chemical-induced toxicity is a highly complex process and it is dependent on a number of factors such as chemical structure, concentration in tissue, biological activity of the chemical and/or its metabolite(s), and the activity of cellular defence mechanisms. Until recently, the study of toxicology used mostly cell viability, and biochemical and/or histopathological changes in cells upon treatment, for the classification of toxic and non-toxic chemicals. However, insights into mechanisms of toxicity may improve predictions of toxicity risk at lower doses and shorter exposure times. Transcriptomics studies enable the measurement of thousands of transcripts simultaneously and hence provide the means of understanding toxicity at a pathway-level, insights that could also be used to predict toxicity caused by structurally similar chemicals. A limited number of studies have been published that address mechanistic toxicity using transcriptomics methods and/or highcontent imaging. In this study, we used day 40 3x-genome and AAGly-metabolically engineered iPSC-HLC progeny to evaluate their potential as a model for traditional concentration-response toxicity studies and their ability to identify mechanisms of adverse effects caused by the chemicals. The 3x-AAGly-HLCs correctly categorised Amiodarone, Paraquat, Diclofenac and Clozapine as acutely cytotoxic compounds. Transcriptomics data (using the TempO-Seq method) on cells treated with Amiodarone, Paraquat, Clozapine, Diclofenac, and Ibuprofen at or below concentrations inducing 10% loss of cell viability (IC10), was used to gain mechanistic insights in how these chemicals cause hepatotoxicity. The model could identify cellular stress pathways that were previously linked to mechanisms of toxicity induced by these chemicals, and could identify deregulation of specific genes known to be implicated in these processes. We also examined if genes involved in the endoplasmic reticulum (ER) stress and unfolded protein response (UPR) pathways were induced increasingly over time upon chemical treatment, and found that several genes in the PERK-EIF2A-ATF4 and IRE1 cascade were differentially expressed in response to a gradient of chemical treatment concentrations and at both early and later time-points of exposure. Finally, we compared our transcriptome dataset with publicly available datasets of other in vitro liver toxicity models, to benchmark the sensitivity of 3x-AAGly-HLCs to diclofenac. Using transcriptomics data of primary human hepatocytes treated with diclofenac for 24 hours as a benchmark, diclofenac-treated HLCs were compared to other hepatocyte models from literature. Our HLCs identified more differentially expressed genes than treated rat primary hepatocytes, hepatic cell lines and ESC-HLCs, and at lower concentrations of diclofenac treatment. In conclusion, the SBAD2-3x derived HLCs differentiated using the method and medium optimised previously by the Verfaillie laboratory, hold promise for both traditional cytotoxicity-based studies, as well as predictive and mechanistic studies using transcriptomics for the purpose of chemical risk assessment.