PFAS) in Primary Human Liver Spheroids to Inform Read-Across
Total Page:16
File Type:pdf, Size:1020Kb
bioRxiv preprint doi: https://doi.org/10.1101/2020.10.15.341362; this version posted October 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. High-throughput transcriptomic evaluation of per- and polyfluoroalkyl substances (PFAS) in primary human liver spheroids to inform read-across A. Rowan-Carroll1, A. Reardon1, K. Leingartner1, R. Gagné1 , A. Williams1, B. Kuo 1, J. Bourdon-Lacombe2, I. Moffat2, R.Carrier2, A. Nong1, L. Lorusso3, S.S. Ferguson4, E. Atlas1 *, C. Yauk5,1* 1Environmental Health Science and Research Bureau, Healthy Environments and Consumer Safety Branch (HECSB), Health Canada, 2 Water and Air Quality Bureau, HECSB, Health Canada, 3 Chemicals and Environmental Health Management Bureau, HECSB, Health Canada, 4 US National Institute of Environmental Health Sciences, 5Dept. of Biology, University of Ottawa. For submission to Toxicological Sciences * To Whom correspondence should be addressed: [email protected], [email protected] and [email protected] Keywords: TempO-Seq, PFAS, Liver spheroids, benchmark concentration, new approach methodology, bioactivity exposure ratio The following secure token has been created to allow review of record GSE144775 while it remains in private status: wlqxagaadnmjxmn. 1 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.15.341362; this version posted October 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Abstract (250word limit) Per- and poly-fluoroalkyl substances (PFAS) are widely found in the environment because of their extensive use and persistence. Although a few PFAS are well studied, most lack toxicity data to inform human health hazard and risk assessment. This study focussed on four model PFAS: perfluorooctanoic acid (PFOA; 8 carbon), perfluorobutane sulfonate (PFBS; 4 carbon), perfluorooctane sulfonate (PFOS; 8 carbon), and perfluorodecane sulfonate (PFDS; 10 carbon). Human primary liver cell spheroids (i.e., pooled-donor) were exposed to 10 concentrations of PFAS over four time-points. The approach aimed to: (1) identify the extent to which the PFAS modulated gene expression; (2) identify similarities in biological responses; (3) compare PFAS potency through benchmark concentration (BMC) analysis; and (4) derive bioactivity exposure ratios (BERs: ratio of concentration at which biological response occurs converted to administered equivalent dose relative to human daily exposure). All PFAS induced transcriptional changes of cholesterol biosynthesis and lipid metabolism, and appeared to activate PPARα. PFOS exhibited the most transcriptional perturbations and had a highly similar gene expression profile to PFDS. PFBS induced the least transcriptional changes and had the highest BMCs. The data indicate that these four chemicals may have common molecular targets and toxicities, but that PFOS and PFDS are the most similar. BERs derived for PFOA and PFOS had relatively low margins; the transcriptomic BER was slightly more conservative than BERs derived from rodent apical endpoints used as points of departure in risk assessment. The data provide a baseline on which to compare the toxicity of other PFAS using this testing strategy. 2 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.15.341362; this version posted October 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Introduction (750word limit) Per- and poly-fluoroalkyl substances (PFAS) are a class of chemicals ubiquitously found in the environment. Their unique hydrophobic and lipophobic properties has led to their use in many consumer products and their persistence, and bioaccumulation potential have led to their detection in wildlife and humans worldwide (Armitage et al., 2006; Wania, 2007; Schenker et al., 2008; Stemmler and Lammel, 2010; Armitage et al., 2009a, 2009b). Although perfluorooctane sulfonate (PFOS) and perfluorooctanoate (PFOA) have been retired from North American production, they continue to be found in drinking water and contaminated sites within Canada (Paul and Auer, 2000; Canada. Health Canada., 2017; Shi et al., 2015; Z. Wang et al., 2014; Wang et al., 2016; Liu et al., 2017). Moreover, there are thousands of PFAS that remain in production globally, with little toxicological data available for the vast majority of them (Linda S. Birnbaum, 2018). The adverse health effects of PFOS and PFOA in humans are well documented, including increased cholesterol levels (Eriksen et al., 2013; Geiger et al., 2014), decreased human fertility (Bach et al., 2016), and reduced immune response (National Toxicology Program NTP NTP Monograph Immunotoxicity Associated with Exposure to Perfluorooctanoic Acid or Perfluorooctane Sulfonate, 2016; Kielsen et al., 2016; Health Canada, 2018b). Recently the US Environmental Protection Agency has recognized the need for, and developed an action plan to address concerns of PFAS in drinking water, reduction of exposure, safety in commerce and support for communications (United States Environmental Protection Agency, 2020). The effects of several PFAS have been extensively studied in animal models. For example, wild-type and PPARα-null mice treated with perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), or perfluorohexane sulfonate (PFHxS) exhibit steatosis, which is characterised by an accumulation of lipids in hepatocytes (Das et al., 2017). The accumulation of fatty acids ultimately leads to liver disease and liver failure (Das et al., 2017). Studies conducted in rodents have also shown that PFOS and PFOA exposure can lead to cancer (Health Canada, 2018b; Filgo et al., 2015; Biegel et al., 2001; Butenhoff et al., 2012). Conversely, a critical review of the human literature found 3 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.15.341362; this version posted October 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. no causal association between cancer in humans and exposure to PFOA and PFOS (Chang et al., 2014). As such, the liver has been identified as the critical epidemiological endpoint in Health Canada’s existing regulatory guidelines on PFAS (Health Canada, 2018b). Acquiring information on data-poor substances for risk assessment has been a challenge for regulatory agencies worldwide given the cost and time required to conduct traditional toxicological research. To accelerate the pace of chemical risk assessments, global efforts have focused on increasing the use of New Approach Methodologies (NAMS) (Kavlock et al., 2018; Krewski et al., 2020). In silico predictions, high throughput screening, novel in vitro models, in vitro to in vivo extrapolation (IVIVE), and read-across are some of the NAMs with increasing use in risk assessment for chemicals with limited data. High-throughput transcriptomics, when paired with physiologically- relevant model systems, is a particularly powerful methodology to obtain information about the broad scope of biological perturbations that result from chemical exposures. New technologies now enable gene expression analyses directly from cell lysates, which significantly increases the ease and speed with which transcriptomic data can be produced (Trejo et al., 2019). To address the need for efficient assessment of data-poor PFAS, we sought to develop an approach to broadly screen for PFAS-induced biological perturbations through global gene expression profiling of human primary cell liver spheroids. A main objective was to establish baseline transcriptomic profiles and an analytical pipeline for future evaluations of data-poor PFAS. Liver spheroids were exposed to nine concentrations of PFAS and sampled at four time points. To establish our approach, we focused on four prototype PFAS: PFOS (8 carbon), PFOA (8 carbon), perfluorodecane sulfonate (PFDS: long chain, 10 carbon), and perfluorobutane sulfonate (PFBS: short chain, 4 carbon). Microscopy was used to evaluate phenotypic changes in the spheroids, and the lactose dehydrogenase activity assay was used to assess cytotoxicity over time. The TempO-Seq platform using the NIEHS S1500 gene set (https://federalregister.gov/a/2015-08529) was applied for transcriptomic analysis. PFAS were compared based on the induction of differentially expressed genes (DEGs), 4 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.15.341362; this version posted October 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. hierarchical clustering and principle component analysis (PCA), pathway and upstream regulator enrichment, and transcriptomic benchmark concentrations (BMCs). Bioactivity- exposure ratios (BERs) for PFOS and PFOA were calculated by comparing human administered equivalent dose (AED: the daily dose required to achieve the in vitro concentration of interest) to human exposure levels. A small BER suggests that biological perturbations occur at concentration that approach external human exposure levels. This initial research project was used to inform experimental design and analytical approach for analysis of a second larger group of data-poor PFAS (Reardon et al., 2020