Loss of liver-specific and sexually dimorphic gene expression by aryl hydrocarbon receptor activation in C57BL/6 mice

The aryl hydrocarbon receptor (AhR) is a highly conserved transcription factor that mediates a broad spectrum of species-, strain-, sex-, age-, tissue-, and cell-specific responses elicited by structurally diverse ligands including 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). Dose-dependent effects on liver-specific and sexually dimorphic gene expression were examined in male and female mice gavaged with TCDD every 4 days for 28 or 92 days. RNA-seq data revealed the coordinated repression of 181 genes predominately expressed in the liver including albumin (3.7-fold), α-fibrinogen (14.5-fold), and β-fibrinogen (17.4-fold) in males with corresponding AhR enrichment at 2 hr. Liver-specific genes exhibiting sexually dimorphic expression also demonstrated diminished divergence between sexes. For example, male-biased Gstp1 was repressed 3.0-fold in males and induced 4.5-fold in females, which were confirmed at the protein level. Disrupted regulation is consistent with impaired GHR-JAK2-STAT5 signaling and inhibition of female specific CUX2-mediated transcription as well as the repression of other key transcriptional regulators including Ghr, Stat5b, Bcl6, Hnf4a, Hnf6, Foxa1/2/3, and Zhx2. Attenuated liver-specific and sexually dimorphic gene expression was concurrent with the induction of fetal genes such as alpha-fetoprotein. The results suggest AhR activation causes the loss of liver-specific and sexually dimorphic gene expression producing a functionally “de-differentiated” hepatic phenotype.


Introduction
2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) is a persistent environmental contaminant that bioaccumulates and elicits a broad spectrum of biochemical and toxic effects. These effects are largely mediated by the aryl hydrocarbon receptor (AhR), a ligand-activated transcription factor [1]. Following ligand binding, cytosolic AhR dissociates from its chaperone proteins (e.g., HSP90, XAP2, and P23) and translocates to the nucleus where it dimerizes with the aryl hydrocarbon receptor nuclear translocator (ARNT). This complex interacts with dioxin response elements (5'-GCGTG-3') to elicit differential gene expression. AhR-mediated gene expression been observed. On PND 28 and every following 4th day (d), animals were orally gavaged with 0.1 mL sesame oil or 0.01, 0.03, 0.1, 0.3, 1, 3, 10 and 30 μg/kg TCDD (Dow Chemical Company, Midland, MI) for a total of 28 or 92d (females only). Male mice exhibited greater sensitivity to TCDD and therefore were not treated for 92 days to avoid overt toxicity and death. Four days following the final dose mice were sacrificed by cervical dislocation. Liver samples were collected and immediately frozen in liquid nitrogen. Doses used compensate for the short study duration compared to lifelong cumulative human exposure from diverse AhR ligands, their bioaccumulative nature, and differences in metabolism and half-life. Furthermore, doses result in hepatic tissue levels that span human background serum concentrations reported in the United States, Germany, Spain, and the United Kingdom as well as serum levels in Viktor Yushchenko 4-39 months following intentional poisoning [30]. All procedures were approved by the All-University Committee on Animal Use and Care.
Transcriptomic, AhR ChIP-seq, and putative dioxin response element data Transcriptomic, AhR ChIP-seq, and computationally identified putative dioxin response element (pDRE) data were previously published [7,8,28]. Genes were considered differentially expressed when |fold-change| ! 1.5 and posterior probability values (P1(t)) ! 0.8 determined using an empirical Bayes approach [28]. Significant AhR ChIP-seq binding used a FDR 0.05. pDREs were considered functional when the matrix similarity score (MSS) ! 0.85. Sequencing data is available in the Gene Expression Omnibus (GEO; GSE62902, GSE81990, GSE87519, GSE97634, GSE97636). Sexually dimorphic genes were identified using RNA-seq datasets from male and female mice from vehicle treatment groups from the studies described above (GSE62902 and GSE87519) to account for age and other experimental factors. Following read mapping to the mouse reference genome (GRCm38) and gene count determinations [28], counts were transformed through variance stabilizing transformation (VST) using DESeq in R [31]. Data was normalized using a semi-parametric approach in SAS v9.3. An empirical Bayes approach was used to calculate posterior probabilities (P1(t)) values. Genes were considered sexually dimorphic if there was a |fold-change| ! 2.0 and P1(t) ! 0.8 difference between male and female transcriptomic data sets (S1 Table). Identification of over-represented transcription factor binding sites for ChIP-seq peaks was performed using Pscan-ChIP [32].

Gene set enrichment and neighborhood gene expression regulation analyses
Gene set enrichment analysis (GSEA) [33] of sexually dimorphic and liver-specific genes was performed using a pre-ranked gene list based on the magnitude of the fold-change (largest induction to most repressed). The liver-specific gene set was identified using published microarray data representing 96 different tissues and cell types including adult male liver [34]. Tissue specificity was determined based on the difference of the gene in the tissue of interest and subtracting the highest signal in all other tissues (S1 Fig). Genes were considered liver-specific when the difference in microarray signal was ! 5,000 (S2 Table). Genomic neighborhoods consisting of several nearby genes exhibiting similar differential expression, indicating putative shared regulation, was determined by calculating a running sum of log 2 fold-changes. Firstly, the differentially expressed gene (|fold-change| ! 1.5) with smallest transcription start site (TSS) genomic coordinate was identified and that fold-change (in log 2 space) was set as the initial score. The next differentially expressed gene based on TSS genomic coordinate was then found and that value (in log 2 space) was added to the score. When a gene TSS was identified but did not exhibit a change in expression, the existing score was recorded. This process was continued across all genes in the chromosome, then all chromosomes. Scores were plotted based on the TSS coordinate and assessed for neighborhoods identified as dramatic increases (induction) or decreases (repression) revealing genome regions of closely located genes that may be co-regulated.

Histopathology
Paraffin embedded livers were sectioned at 4 μm, placed on slides coated with 2% 3-Aminopropyltriethoxysilane, and dried at 56˚C overnight. Slides were deparaffinized in Xylene and hydrated through descending grades of ethyl alcohol to distilled water then placed in Tris Buffered Saline pH 7.4 (Scytek Labs, Logan, UT) for 5 minutes (min). Endogenous peroxidase was blocked using 3% Hydrogen Peroxide/Methanol bath for 30 min followed by running tap and distilled water rinses. Standard micro-polymer complex staining steps were performed at room temperature on the IntelliPath™ Flex Autostainer. All staining steps are followed by TBS Autowash buffer (Biocare Medical, Concord, CA) rinses. After blocking for non-specific protein with Background Punisher (Biocare) for 10 min, sections were incubated with Polyclonal Rabbit anti-AFP (Protein Tech, 14550-1-AP; 1:100) in normal antibody diluent (NAD-Scytek) for 60 min. Following 30 min in ProMark Rabbit anti-Rodent™ Micro-Polymer (Biocare), reaction development was performed with Romulin AEC™ (Biocare) for 5 minutes and counterstained with CAT Hematoxylin (Biocare). Histological processing and staining was performed at the Michigan State University Investigative Histopathology Lab (https:// humanpathology.natsci.msu.edu/).

TCDD-mediated repression of liver-specific gene expression
To further explore the effect of TCDD on hepatic gene expression, 181 genes primarily expressed in the liver were identified using 96 GeneAtlas MOE430 datasets representing tissue/ cell types from naïve male mice [34]. These 181 genes are subsequently referred to as "liverspecific" based on a microarray signal difference of !5,000 units when comparing constitutive expression in liver to other non-hepatic tissues/cell types. Signal difference was used instead of fold change to capture only highly expressed genes. Our liver-specific gene set included previously identified liver-specific genes including albumin (Alb), methionine adenosyltransferase (Mat1a), fatty acid binding protein 1 (Fabp1), and fetuin-A (Ahsg). GSEA revealed repression by TCDD in both male and female mice at 28 and 92d suggesting an overall loss of liver-specific expression in both sexes (Fig 1A-1C). Repressed genes included Alb, Mat1a, glutamate dehydrogenase 1 (Glud1), and hydroxyacid oxidase 1 (Hao1) in both sexes. Hepatic HAO1 The 181 liver-specific genes were identified using published microarray datasets representing 96 tissues/cell types [34]. Identification of liver-specific genes is described in materials and methods. TCDD elicited gene expression changes were ranked from most induced (left-red) to most repressed (right). Vertical black line represents identified liver-specific genes. The top panel (green line) represents a running-sum statistic (enrichment score) based on the lower panel, increasing when a gene is a member of the liver-specific gene set and decreasing when it is not. Enrichment scores increased most dramatically on the right indicating most of the liver-specific genes were repressed by TCDD. (D) Heat map of liver-specific gene expression changes elicited by TCDD. (E) Heatmap of TCDD-elicited repression of hepatokines. For heatmaps (D and E) blue indicates repression while red represents induction. The presence of pDREs (MSS ! 0.856) and hepatic AhR enrichment peaks (FDR 0.05) at 2h are shown as green boxes. Read count represents the maximum raw number of aligned reads to each transcript where yellow represents a lower level of expression ( 500 reads) and pink represents a higher level of expression (! 10,000). protein levels were also repressed in males while hepatic ALB levels increased contrary to its transcriptional repression (Fig 2). Liver-specific genes demonstrating increased expression in males and females included Cyp1a2 and Ugt2b35, members of the AhR gene battery [35].

Induction of liver progenitor cell marker alpha-fetoprotein
The 1.6-and 15.0-fold induction of alpha-fetoprotein (Afp; 30 μg/kg) as well as 7.4-and 7.8-fold induction of TNF receptor superfamily 19 (Tnfrsf19; 30 μg/kg) in female and male mice, respectively, contrasts the overall repression of adult liver-specific gene expression. AFP is typically expressed during development in hepatic progenitor cells or hepatocellular carcinoma cells. Despite the lack of histological evidence of tumorigenesis, liver sections were evaluated for AFP expression. Increased cytosolic and intranuclear AFP staining was observed in male liver sections at 30 μg/kg TCDD including within hypertrophic differentiated hepatocytes showing lipid vacuolization, but not in female sections (Fig 3). The apparent difference in scale of histological images reflect the increased sensitivity of male mice to TCDD (Fig 3D). Expression of other albuminoids which play important roles in systemic transport such as albumin (Alb), afamin (Afm), and vitamin-D binding protein (Gc) were repressed 3.7-, 7.8-, and 4.3-fold, respectively, in male mice at 30 μg/kg. Repression of Alb, Afm and Gc was also observed in females following 92d of exposure, albeit to a lesser extent. Notably, Alb, Afp, Afm and Gc are sequentially co-localized on the same chromosome and arranged in the same transcriptional orientation in all mammals [38]. AhR enrichment was observed at 3 sites without pDREs~13 -30kb upstream of Alb in both male and female mice 2 h after TCDD treatment suggesting putative AhR-dependent co-regulation (Fig 4). The transcriptional repressor, zincfinger and homeoboxes 2 (Zhx2), is also known to inhibit the fetal hepatic expression of Afp, H19, and Gpc3 [20]. Accordingly, 30 μg/kg TCDD repressed Zhx2 1.6-fold while fetal markers Afp, H19 and Gpc3 were induced 15.0-, 2.9-and 2.5-fold, respectively.

AhR enrichment at liver-specific genes
AhR enrichment was observed within 136 of the 181 liver-specific genes (79 in both sexes, 56 in males only, and 1 in females only; Fig 5A). pDREs were present in the majority of these binding sites (26 of 56 male sites, 57 of 79 male and female sites, and 0 of 1 female only sites), suggesting direct AhR-DRE-dependent regulation. Over-represented transcription factor motif analysis within AhR enriched regions identified binding sites for HNF4α, COUP-TFI and III, retinoic acid receptor (RXRA, RARG), CCAAT/Enhancer binding proteins (CEBPB, D, E, and G), and vitamin D receptor (VDR) (Fig 5B). In male mice, Cebpd, Rarg, and Vdr were induced 2.0-, 2.8-, and 2.1-fold, respectively, while Cebpe was repressed 4.1-fold at 30 μg/   at 30 μg/kg) [39]. HNF6 and HNF3 repression is consistent with the 8.1-fold repression of transthyretin (Ttr), a male target gene. Smaller changes were observed in females after 92d but not at 28d. These results are consistent with proposed AhR-COUP-TFII-mediated repression [2].

Discussion
TCDD induces a broad spectrum of hepatic pathologies from hepatic fat accumulation to HCC in a species-, strain-, sex-, age-, tissue-and cell-specific manner [4][5][6][7][8][9][10]26]. Much remains to be understood about the role of sex in liver disease and toxicity [14]. Consequently, we investigated the effect of TCDD on liver-specific and sexually dimorphic gene expression using hepatic RNA-seq datasets from male and female mice [7,8,28]. Overall, gene expression changes elicited by TCDD indicate a loss of liver-and sex-specific gene expression producing a sexually ambiguous and functionally de-differentiation liver.
The dose-dependent repression of liver-specific gene expression suggests diminution of the functional hepatic phenotype. Furthermore, AFP induction suggests the TCDD treated liver is retreating to a less differentiated transcriptome consistent with increases in hepatic and serum AFP levels typically associated with HCC or liver regeneration [44]. AFP induction in differentiated hepatocytes marked by vacuolization suggests TCDD caused a loss in hepatocyte differentiation as opposed to progenitor cell proliferation in response to liver regeneration needs. Moreover, AFP staining in liver cancer is only reported to be cytoplasmic [44] while both intranuclear and cytoplasmic staining was observed with TCDD in males at 30 μg/kg, the significance of which is unclear. Paradoxically, Alb, which is typically reciprocally regulated compared to Afp, was repressed as expected, yet hepatic protein levels were increased. Interestingly, plasma ALB levels are lower in TCDD exposed waste incineration workers [45] and PCB exposed transformer repair workers [46].
Further evidence of diminished liver function includes the repression of albuminoids, hemoglobin subunits, major urinary proteins, and hepatokines [36,47]. In addition to participating in the transport of vitamins, lipids, amino acids, steroids and metal ions, albuminoids also bind heme reducing lipoprotein oxidation and oxidative stress [7,47]. Alb repression along with hemoglobin Hba and Hbb subunits is also consistent with TCDD elicited decreases in serum proteins [5]. Similarly, fetuin-A (Ashg) and leukocyte derived chemotaxin 2 (Lect2) repression enhances insulin signaling [36] and may support improved glucose tolerance in the absence of altered insulin levels in TCDD treated animals [27]. The physiological significance of Mup repression is unclear as hepatic secretion is largely male specific and implicated in lipid and pheromone binding. Although Mups also regulate glucose and lipid metabolism by repressing gluconeogenic and lipogenic genes [48], they are not expressed in humans.
There was also concomitant loss of sexually dimorphic gene expression consistent with reports of male liver "feminization" and divergent regulation of sexually dimorphic genes [12,13,26]. Our comparative study indicates loss of sexually dimorphic expression as opposed to feminization or masculinization. This is likely due to the combined loss of both male gene repression and female gene induction in females, and vice-versa in males. Notably, the number of differentially expressed genes in TCDD treated males (9,314) was much larger than in females (3,766) [7,12,13]. This large difference in the number of dysregulated genes between sexes was also reported for HNF4α null mice which resulted in~1,000 fewer DEGs in female mice [43]. HNF4α regulates liver-specific and GHR-JAK2-STAT5-mediated sexually dimorphic gene expression, the latter of which is impaired by AhR activation through repression of Ghr and Jak2 [17,19,22,26]. Moreover, mouse models lacking hepatic GHR, JAK2, or STAT5 have increased liver damage and develop features of NAFLD [49] suggesting GHR-JAK2--STAT5 impairment contributes to TCDD-elicited NAFLD. The anti-estrogenic effects of AhR agonists in mediating these changes remains unclear [50,51].
The repression of the negative regulator ZHX2 may also be involved in the loss of liver-specific and sexually dimorphic gene expression [20,52,53]. ZHX2 inhibits fetal gene expression in adult liver, namely Afp, H19, and Gpc3, as well as several sex-specific cytochrome P450s [20,21,53,54], all of which were induced by TCDD. Similarly, ZHX2 is required for high expression of Mups [42] most of which were repressed by TCDD. Overall, ZHX2 repression is consistent with the loss of tissue-and sex-specific gene expression and the induction of fetal genes commonly associated with HCC.
Collectively, our studies suggest that TCDD elicits a loss of liver identity by diminishing both tissue-specific and sexually dimorphic gene expression. While the exact mechanism remains to be elucidated, impairment of GHR-JAK2-STAT5 signaling, dysregulation of HNFmediated transcription, and Zhx2 repression are implicated. These effects may also contribute to the development of NAFLD pathologies and other AhR-mediated toxicities. Indeed, the repression of HNF6 (Onecut1) which plays a regulatory role in sexually dimorphic and liverspecific gene expression, also regulates hepatic lipid metabolism via induction of the transcriptional repressor REV-ERBα (Nr1d1) [55] which was induced by TCDD [7]. In summary, persistent AhR activation promoted the loss of unique hepatic functions and sexual dimorphism that may factor in TCDD-elicited NAFLD and HCC.
Supporting information S1 Table. Gene expression changes of sexually dimorphic genes. (XLSX) S2 Table. Gene expression changes in TCDD treated mice for 181 liver-specific genes. (XLSX) S1 Fig. Generation of liver-specific gene set. Microarray datasets for basal gene expression for 96 different male mouse tissues/cell types were obtained from Lattin et al. [34]. The difference between the microarray signal in the liver and the maximum fluorescent signal on a per gene basis in all other tissues/cells was calculated (ΔSignal). A gene was considered liver-specific when ΔSignal ! 5,000 units.