Case study on the utility of hepatic global gene expression profiling in the risk assessment of the carcinogen furan

Abstract

Furan is a chemical hepatocarcinogen in mice and rats. Its previously postulated cancer mode of action (MOA) is chronic cytotoxicity followed by sustained regenerative proliferation; however, its molecular basis is unknown. To this end, we conducted toxicogenomic analysis of B3C6F1 mouse livers following three week exposures to non-carcinogenic (0, 1, 2mg/kgbw) or carcinogenic (4 and 8mg/kgbw) doses of furan. We saw enrichment for pathways responsible for cytotoxicity: stress-activated protein kinase (SAPK) and death receptor (DR5 and TNF-alpha) signaling, and proliferation: extracellular signal-regulated kinases (ERKs) and TNF-alpha. We also noted the involvement of NF-kappaB and c-Jun in response to furan, which are genes that are known to be required for liver regeneration. Furan metabolism by CYP2E1 produces cis-2-butene-1,4-dial (BDA), which is required for ensuing cytotoxicity and oxidative stress. NRF2 is a master regulator of gene expression during oxidative stress and we suggest that chronic NFR2 activity and chronic inflammation may represent critical transition events between the adaptive (regeneration) and adverse (cancer) outcomes. Another objective of this study was to demonstrate the applicability of toxicogenomics data in quantitative risk assessment. We modeled benchmark doses for our transcriptional data and previously published cancer data, and observed consistency between the two. Margin of exposure values for both transcriptional and cancer endpoints were also similar. In conclusion, using furan as a case study we have demonstrated the value of toxicogenomics data in elucidating dose-dependent MOA transitions and in quantitative risk assessment.

BMDExpress Data Viewer ‐ a visualization tool to analyze BMDExpress datasets

“…To illustrate the features and functionality of BMDExpress Data Viewer, BMDExpress output datasets were obtained from two case studies, i.e., (1) hepatic gene expression (Agilent) profiles of mice orally exposed to increasing doses of furan by oral gavage (Jackson et al ., 2014), and (2) thyroid gene expression (Affymetrix) profiles from rats that were orally exposed to increasing doses of the chemical MDMB (Thomas et al ., 2013b). …”

“…By applying the Functional Enrichment Analysis tool with Fisher's exact test P ≤ 0.05 (5%), and the minimum number of significant genes equal to three (in accordance with Jackson et al . [2014]), we obtained 24 pathways (Table 1). Fourteen (58%) of the identified pathways were also found in our previous IPA analysis (Jackson et al ., 2014).…”

“…[2014]), we obtained 24 pathways (Table 1). Fourteen (58%) of the identified pathways were also found in our previous IPA analysis (Jackson et al ., 2014). The mean and median values for the BMDL means for these pathways were 1.77 and 1.63 mg kg –1 day –1 , respectively.…”

“…Agilent DNA microarrays were used to examine hepatic transcriptional profiles for these mice. Detailed methods for the preparation of samples, microarray gene expression and normalization of the data are described in the publications (Jackson et al ., 2014; Webster et al ., 2014). In the second dataset (GSE45892), rats were exposed to increasing doses of one of six chemicals for 5, 14, 28 and 90 days.…”

“…In addition to modeling the dose–response of individual genes, BMDExpress also computes BMD values for biological categories (e.g., pathway BMD means, medians, 5th percentiles, etc.) using a feature called “Functional Classifications.” Global transcriptional BMD modeling using BMDExpress has been used to demonstrate that toxicogenomic BMDs fall within the range of BMDs derived from apical data, and has been useful for investigating the mode of action of several toxicants (Bhat et al ., 2013; Bourdon et al ., 2013; Jackson et al ., 2014; Moffat et al ., 2015). These studies support the notion that transcriptional BMDs can be used to derive points of departure for human health risk assessment, particularly in instances when apical data are not available (Chepelev et al ., 2015; Moffat et al ., 2015; Thomas et al ., 2012, 2013a, b).…”