Early microRNA indicators of PPARα pathway activation in the liver

Abstract

MicroRNAs (miRNAs) are short non-coding RNA species that play key roles in post-transcriptional regulation of gene expression. MiRNAs also serve as a promising source of early biomarkers for different environmental exposures and health effects, although there is limited information linking miRNA changes to specific target pathways. In this study, we measured liver miRNAs in male B6C3F1 mice exposed to a known chemical activator of the peroxisome proliferator-activated receptor alpha (PPARα) pathway, di(2-ethylhexyl) phthalate (DEHP), for 7 and 28 days at concentrations of 0, 750, 1500, 3000, or 6000 ppm in feed. At the highest dose tested, DEHP altered 61 miRNAs after 7 days and 171 miRNAs after 28 days of exposure, with 48 overlapping miRNAs between timepoints. Analysis of these 48 common miRNAs indicated enrichment in PPARα-related targets and other pathways related to liver injury and cancer. Four of the 10 miRNAs exhibiting a clear dose trend were linked to the PPARα pathway: mmu-miRs-125a-5p, -182-5p, -20a-5p, and -378a-3p. mmu-miRs-182-5p and -378a-3p were subsequently measured using digital drop PCR across a dose range for DEHP and two related phthalates with weaker PPARα activity, di-n-octyl phthalate and n-butyl benzyl phthalate, following 7-day exposures. Analysis of mmu-miRs-182-5p and -378a-3p by transcriptional benchmark dose analysis correctly identified DEHP as having the greatest potency. However, benchmark dose estimates for DEHP based on these miRNAs (average 163; range 126-202 mg/kg-day) were higher on average than values for PPARα target genes (average 74; range 29-183 mg/kg-day). These findings identify putative miRNA biomarkers of PPARα pathway activity and suggest that early miRNA changes may be used to stratify chemical potency.

Keywords: AIC, Akaike Information Criterion; ALT, alanine aminotransferase; AOP, adverse outcome pathway; AST, aspartate aminotransferase; Acox1, acyl-Coenzyme A oxidase 1; Adverse outcome pathway (AOP); AhR, aryl hydrocarbon receptor; BBP, n-butyl benzyl phthalate; BMD, benchmark dose; BMDA, apical-based benchmark dose; BMDL, BMD lower confidence interval; BMDT, transcriptional-based benchmark dose; BMR, benchmark response; BROD, benzyloxyresorufin O-debenzylation; Benchmark dose (BMD); Biomarkers; CAR, constitutive androstane receptor; DEGs, differentially expressed genes; DEHP, di (2-thylhexyl) phthalate; DEmiRs, differentially expressed miRNAs; DNOP, di-n-octyl phthalate; EPA, U.S. Environmental Protection Agency; EROD, ethoxyresorufin O-dealkylation; GEO, Gene Expression Omnibus; HCA, hepatocellular adenoma; HCC, hepatocellular carcinoma; Hepatocellular carcinoma; IPA, Ingenuity Pathway Analysis; Liver toxicity; MOA, mode of action; MicroRNAs; Mode of action (MOA); Nrf2, nuclear receptor erythroid 2-like 2; POD, point-of-departure; PPARα, peroxisome proliferator-activated receptor alpha; PROD, pentoxyresorufin O-depentylation; PXR, pregnane X receptor; Peroxisome proliferator-activated receptor alpha (PPARα); Phthalate; SDH, sorbitol dehydrogenase; TMM, trimmed mean of M-values; ddPCR, droplet digital polymerase chain reaction; mRNA, messenger RNA; miRNAs, microRNAs; mtDNA, mitochondrial; rRNA, ribosomal RNA; smallRNA-seq, small RNA sequencing; tRNA, transfer RNA.

Fig. 1

Global analysis of mouse liver miRNA alterations after short-term DEHP exposure. Male B6C3F1 mice were exposed to DEHP for 7 or 28 days through diet and liver miRNA alterations were measured using small RNA-seq. Significant differences were observed at both timepoints, with more robust miRNA alterations occurring at the later timepoint. (A) Average linkage clustering using normalized miRNA counts indicated clear distinct groupings of high (6000 ppm) treated and control mice after 28 days of exposure. (B) Principal component analysis (PCA) similarly indicated these differences at 28 days, but the groups were less distinct at the earlier time with the highest dose tested. (C) Forty-eight differentially expressed miRNAs (DEmiRs) were in common after 7 and 28 days of high DEHP exposure.

Fig. 2

Ingenuity Pathway Analysis (IPA) of mRNA linked to altered miRNA at 7 and 28 days following 6000 ppm DEHP exposure. Differentially expressed genes (DEGs) were linked to DEmiRs using the miRNA target prediction algorithms in IPA. These predicted miRNA target genes were analyzed for signaling pathway enrichment using IPA: (A) toxicological pathway lists and (B) upstream regulator analysis. For the toxicity analysis, each dot represents a different pathway that was significantly enriched by this DEG list (-log(p-value >1.3) and related to the general category listed along the x-axis. These pathways are fully described in Supplemental File 4, under “miRNA.linked DEGSs, tox function” and “miRNA.linked DEGs, upstream.reg” tabs.

Fig. 3

BMD analysis of mmu-miR-182-5p and mmu-miR-378a-3p using 7-day DEHP exposure data. (A) Small RNA-seq data for the top two miRNA candidates were fit to the best response model, with and without top dose, using benchmark dose (BMD) analysis. (B) Modeling was repeated for the same miRNA candidates using ddPCR data, with and without top dose, using the best fit model. Doses were converted to mg dose per mouse weight (kg) each day. Both BMD and benchmark dose for lower confidence interval (BMD_L) are noted in each graph. See Materials and Methods for full procedure description.

Fig. 4

BMD analyses for miRNA, gene expression, apical, and 2-year tumor data based on DEHP exposure in mice. Benchmark dose (BMD; closed shapes) and BMD lower confidence values (BMDL; open shapes) were calculated for select miRNA candidates and summarized for mRNA, apical measurements, and 2-year hepatocellular carcinoma (HCC) ± hepatocellular adenoma (HCA) based on previous calculations [8,32,39]. Where available, sequencing (circles), PCR (diamonds), and microarray (triangles) data were used to calculate transcriptomic (_T; here, both miRNA and mRNA) values. Associated pathways or transcriptional factors associated with these transcripts are marked below the gene/miRNA names. Functional apical (_A) BMD values, including measurements of CYP activity (PROD, BROD, EROD), liver enzymes (ALT and AST); small cell proliferation (Ki67), hepatocyte cytoplasmic alterations, and liver weight were derived from Lake et al. (2016) and denoted as squares. Exposure thresholds for HCC and HCC + HCA are based on two-year data, calculated in Wood et al. (2014), and shown for reference as asterisks/solid line (BMD) or pluses/broken lines (BMDL). Special notations for BMD values: ^•= BMD greater than top dose tested; UF = the response was dose-related but was an Unacceptable Fit (global goodness of fit p-value <0.1); NS = not significance compared to control values for any dose tested; NA = measurement not available.

Fig. 5

Serum measurements of mmu-miR-182-5p and mmu-miR-378a-3p. Digital drop PCR (ddPCR) was performed to measure miRNA candidates in serum isolated from DEHP treated and control mice from (A) 7-day and (B) 28-day exposures. Box plots of data shown. * = significant based on Mann-Whitney U test (p-value <0.05). n = 6-8.