Phenobarbital mediates an epigenetic switch at the constitutive androstane receptor (CAR) target gene Cyp2b10 in the liver of B6C3F1 mice

Abstract

Evidence suggests that epigenetic perturbations are involved in the adverse effects associated with some drugs and toxicants, including certain classes of non-genotoxic carcinogens. Such epigenetic changes (altered DNA methylation and covalent histone modifications) may take place at the earliest stages of carcinogenesis and their identification holds great promise for biomedical research. Here, we evaluate the sensitivity and specificity of genome-wide epigenomic and transcriptomic profiling in phenobarbital (PB)-treated B6C3F1 mice, a well-characterized rodent model of non-genotoxic liver carcinogenesis. Methylated DNA Immunoprecipitation (MeDIP)-coupled microarray profiling of 17,967 promoter regions and 4,566 intergenic CpG islands was combined with genome-wide mRNA expression profiling to identify liver tissue-specific PB-mediated DNA methylation and transcriptional alterations. Only a limited number of significant anti-correlations were observed between PB-induced transcriptional and promoter-based DNA methylation perturbations. However, the constitutive androstane receptor (CAR) target gene Cyp2b10 was found to be concomitantly hypomethylated and transcriptionally activated in a liver tissue-specific manner following PB treatment. Furthermore, analysis of active and repressive histone modifications using chromatin immunoprecipitation revealed a strong PB-mediated epigenetic switch at the Cyp2b10 promoter. Our data reveal that PB-induced transcriptional perturbations are not generally associated with broad changes in the DNA methylation status at proximal promoters and suggest that the drug-inducible CAR pathway regulates an epigenetic switch from repressive to active chromatin at the target gene Cyp2b10. This study demonstrates the utility of integrated epigenomic and transcriptomic profiling for elucidating early mechanisms and biomarkers of non-genotoxic carcinogenesis.

Figure 1. Overview of study design and data integration.

(A) Experimental strategy for the identification and integration of PB-induced expression and epigenetic perturbations in target (liver) and non-target (kidney) tissues. RNA, DNA and chromatin was extracted from liver and kidney samples of control and PB-treated (4-week, 0.05% in drinking water) B6C3F1 male mice (n = 10 per group) and analyzed through the different profiling methodologies and platforms indicated. Abbreviations: liquid chromatography-mass spectrometry (LC-MS), Methylated DNA immunoprecipitation (MeDIP), Chromatin immunoprecipitation (ChIP), quantitative real-time PCR (qPCR). (B) Summary of bioinformatic data integration strategy. For each annotated gene present on gene expression and promoter arrays, the expression and DNA methylation values were mapped to the genome and correlated to examine the functional links between expression and methylation levels at individual loci upon phenobarbital treatment. For loci of interest the abundance of selected chromatin marks were quantified. Coverage of the promoter array (1.8 kb per promoter: 500 bp downstream and 1300 bp upstream of the transcriptional start site (TSS)) is shown. For the methylation analysis a window of 100 bp downstream and 800 bp upstream of the TSS was used. The figure shows an exemplary gene that upon treatment loses DNA methylation and gets expressed.

Figure 2. MeDIP-promoter array profiling of the methylome in the liver of control and PB treated B6C3F1 mice.

(A) An antibody directed against 5-methyl-cytosine (5mC) was used for immunoprecipitation of methylated DNA. Control sequences that are highly methylated (IAP, H19 ICR) or lack CpGs (CSa) were selected as controls for the MeDIP experiment prior to and following whole genome amplification (WGA). The relative enrichment in the bound over input fractions for 10 individual biological replicates was measured by qPCR. (B) Methylation comparison between liver of control and PB-treated mice (average log₂ (IP/total) for replicates) in all 23,428 Nimblegen probe sets. The colors indicate the CpG island class for those probe sets where the log₂ methylation ratio of PB-treated vs. control (difference in M-value) is significant (p≤0.01 and an absolute log₂ fold change of ≥0.2, 28 probe sets), non-differentially methylated regions are indicated in grey. A circle highlights Cyp2b10.

Figure 3. Cyp2b10 is selectively DNA demethylated and over expressed in the liver of PB-treated B6C3F1 mice.

(A) RT-qPCR analysis of Cy2b10 expression in the liver and the kidney of control and PB-treated animals. Fold changes are indicated relative to 18S RNA expression levels. (B) MeDIP-qPCR validation of Nimblegen data. Positive (H19 ICR, IAP), negative (CSa, Intergenic3, Hprt, Gapdh) and selected regions identified on the promoter array were assessed by qPCR from WGA amplified MeDIP and input DNA samples prepared from the liver of 6 control and PB-treated mice. qPCR identified selective demethylation at Cyp2b10 TSS, both in the promoter and first intron (location of PCR amplicons is shown in Figure 3C). Relative enrichment (IP/Input) for DNA methylation of 6 individual biological replicates is shown. (C) Bisulfite sequencing at Cyp2b10 first intron. Sequenced area is shown by the two arrows in the schematic gene map. Each line represents the sequence of a single clone. CpGs are shown as white (unmethylated) or black (methylated) circles. The values above summarizes the overall methylation level of this region (percentage of methylated CpG in all sequenced clones) (D) Quantitative DNA methylation analysis by pyrosequencing of two Cyp2b10 promoter CpG sites (CpG1: -914 and CpG2: -886 indicated in (C)) in the liver and kidney of control and PB-treated animals. Standard deviation was calculated from 10 biological replicates. Primers/genomic regions used for bisulfite sequencing are available in Figure S9.

Figure 4. Native chromatin immunoprecipitation identifies a PB-induced epigenetic switch at Cyp2b10 transcriptional start site (TSS).

(A) Following nuclei purification and micrococcal nuclease treatment, chromatin was immunoprecipitated with antibodies directed against the active marks H3K4me2 and H3K9ac or the repressive mark H3K27me3. IgG was used as a negative IP control in these experiments. The relative enrichment for different marks was analyzed around Cyp2b10 TSS. Ubiquitously expressed, active (Gapdh, beta-actin) and pluripotency associated, repressed (Hoxa9, Oct4) genes were used to control the selectivity of antibodies. Standard deviation was calculated from 3 (H3K9ac) and 5 (IgG, H3K4me2 and H3K27me3) biological replicates. (B) Model representing the epigenetic switch at the Cyp2b10 TSS upon 4 weeks of PB treatment. Blue circles represent methylated cytosines.