DNA methylation profiling of the X chromosome reveals an aberrant demethylation on CXCR3 promoter in primary biliary cirrhosis

Although the etiology of primary biliary cirrhosis (PBC) remains enigmatic, there are several pieces of data supporting the thesis that a strong genetic predisposition and environmental factors interact to produce a selective loss of tolerance. The striking female predominance of PBC has suggested that this sex predisposition may be secondary to epigenetic alterations on the X chromosome. In the present study, we rigorously defined the X chromosome methylation profile of CD4, CD8, and CD14 cells from 30 PBC patients and 30 controls. Genomic DNA from sorted CD4, CD8, and CD14 subpopulations was isolated, sonicated, and immunoprecipitated for analysis of methylation. All products were hybridized to a custom-tiled four-plex array containing 27,728 CpG islands annotated by UCSC and 22,532 well-characterized RefSeq promoter regions. Furthermore, bisulfite sequencing was then used for validation on a subsequent group of independent samples from PBC patients and controls. Thence, expression levels of selected X-linked genes were evaluated by quantitative real-time PCR with cDNA samples from all subjects. We report herein that a total of 20, 15, and 19 distinct gene promoters reflected a significant difference in DNA methylation in CD4+ T, CD8+ T, and CD14+ cells in patients with PBC. Interestingly, there was hypermethylation of FUNDC2 in CD8+ T cells and a striking demethylation of CXCR3 in CD4+ T cells, which inversely correlated with CXCR3 expression levels in CD4+ T cells from early-stage PBC patients. Our data provides a set of genes with epigenetic alteration likely to be indicators of autoimmunity and emphasizes the role of CXCR3 in the natural history of PBC.


Background
Autoimmune diseases are multifactorial in origin. Genetic background confers susceptibility to disease onset but is not sufficient for disease development. Epigenetic mechanisms such as DNA methylation and histone modifications regulate gene expression levels and provide an alternative mechanism for modulating gene function and genetic changes. Importantly, DNA methylation plays an important role in T cell differentiation [1,2], and the potential role of epigenetics in environment and genetics interaction has been widely suggested in autoimmunity [3,4]. With these data in mind, we note that primary biliary cirrhosis (PBC) is characterized by the presence of high titers of circulating antimitochondrial antibodies (AMAs), innate immune system activation [5][6][7], and liver-infiltrating autoreactive CD4 and CD8 T lymphocytes [8][9][10], leading to the progressive destruction of small intrahepatic bile ducts [11,12]. Genetic polymorphisms account for individual susceptibility to PBC in a minority of cases, as supported by genome-wide association (GWAS) data [13][14][15][16][17], while environmental factors may act through epigenetic changes, as demonstrated for specific genes [18][19][20]. It has long been known that there is a striking female sex bias in PBC [21], and recent data has suggested involvement of sex chromosomes [22,23]. For example, an enhanced X monosomy rate in peripheral lymphocytes has been demonstrated in women with PBC [24], despite random X chromosome inactivation [25], as well as an increased Y loss in men with PBC [26].
We thus hypothesized that there is an X chromosome epigenetic component in PBC. Herein we performed an extensive X chromosome methylation analysis in CD4, CD8, and CD14 cells from PBC patients and controls, coupled with gene expression profile from the same cells. The frequency of CXCR3-expressing cells in peripheral blood is known to be higher in PBC patients, and CXCR3positive cells are found in the portal areas of diseased livers, primarily on CD4+ T cells [27]. Moreover, we report that the CXCR3 promoter is demethylated in CD4+ T cells from PBC patients and the expression level of CXCR3 in CD4+ T cells is significantly increased in comparison to controls and inversely correlated with the overall methylation levels of the CXCR3 promoter, indicating that promoter demethylation status contributed to CXCR3 overexpression in CD4+ T cells. Hence, we submit that there is an epigenetic mechanism that underlies CXCR3 expression in PBC. Our findings not only identify potentially relevant DNA methylation markers for the clinical characterization of PBC patients but also support the thesis that epigenetic changes may be critical for the clinical manifestations of PBC.

DNA methylation
Initially, X chromosome methylated DNA immunoprecipitation (MeDIP) arrays were performed on cell subsets from 10 PBC patients and 10 controls (Fig. 1). We identified 8 hypomethylated genes and 12 hypermethylated genes in CD4+ cells, 3 hypomethylated genes and 12 hypermethylated genes in CD8+ cells, and 9 hypomethylated genes and 10 hypermethylated genes in CD14+ cells in PBC compared with normal controls (Fig. 2, Table 1). Gene ontology (GO) analysis was conducted to identify the cell component, molecular function, and biological processes of these genes. Bisulfite sequencing was thence used to validate the results of the MeDIP-ChIP. Based on the GO analysis, four genes were chosen for validation in CD4+, CD8+, and CD14+ cells from 20 additional PBC patients and 20 additional normal controls. Importantly, CXCR3 was demonstrated to be significantly hypomethylated in CD4+ T cells from PBC patients (Fig. 3), whereas UBE2A and FUNDC2 were hypermethylated in CD4+ T cells and CD8+ T cells, respectively, in PBC patients (Figs. 4 and 5). IL1RAPL2 was hypermethylated in CD14+ cells in PBC (Fig. 6).
Importantly, CXCR3 was significantly overexpressed in CD4+ T cells in PBC (p = 0.0055), whereas FUNDC2 was significantly less expressed in CD8+ T cells from PBC than controls (p = 0.0134). These results are concordant with the methylation data. IL1RAPL2 was not expressed in any cell subset in PBC or controls, and no significant differences were identified for UBE2A in the three subsets investigated (data not shown). Finally, a  significant negative correlation between CXCR3 expression and overall methylation levels in CD4+ T cells was found in early-stage PBC patients (p < 0.05) (Fig. 8a). Of note, there was no difference observed based on UDCA treatment (data not shown). In addition, our data demonstrated that CXCR3 expression levels in CD4+ T cells were significantly associated with duration of disease (p < 0.01) ( Fig. 8b) but not associated with age (data not shown).

CD4 T cell phenotype
Finally, we quantified CD4 T cell populations expressing CXCR3 in both PBC (n = 20) and controls (n = 10) by flow cytometry. Based on the expression data, patients with PBC had an increased number in the periphery of CD4 T cells expressing CXCR3. In addition, both naïve and memory CD4 T cells from PBC expressed a higher frequency of CXCR3 cells in the periphery than controls. However, only activated CD4+CD45RO+ cells expressed significantly higher levels of CXCR3 (41.82 % ± 4.37 versus 27.57 % ± 2.61, p = 0.0094) (Fig. 9). Please note we included both early (n = 10) and advanced (n = 10) stage patients with PBC (data not shown).

Discussion
As the case for most autoimmune diseases, PBC results from the contribution of genetic and environmental factors, reflected by incomplete concordance in monozygotic twins [28]. Moreover, our group has suggested a significant role for X chromosome defects in PBC, based on the observation that women with PBC have a significantly enhanced monosomy X frequency in peripheral white blood cells compared to age-and disease stage-matched women with chronic viral hepatitis and to age-matched healthy women [24]. Epigenetic mechanisms act at the interface of genetic and environmental influences on human phenotype and disease risk by altering gene expression levels without altering DNA sequence or chromosome structure. In the case of autoimmune diseases, epigenetic mechanisms have been implicated in the pathogenesis of systemic lupus erythematosus [29,30] and type I diabetes [31]. Moreover, we had previously explored the X chromosome epigenetic component in PBC [18].
In this study, we investigated for the first time the DNA methylation status of the X chromosome in CD4+, CD8+, and CD14+ cells isolated from PBC patients and well-matched controls. We herein demonstrate that the CXCR3 promoter is demethylated in CD4+ T cells from PBC patients compared to controls, which is supported by a significantly higher expression of CXCR3 in the same cell subtype of PBC patients. CXCR3 gene encodes a G protein-coupled receptor with selectivity for three chemokines: CXCL9 or MIG (monokine induced by IFNγ), CXCL10 or IP10 (IFNγ-inducible 10 kDa protein), and CXCL11 or I-TAC (IFN-inducible T cell αchemoattractant). CXCR3 plays an important role in regulating leukocyte trafficking; indeed binding of chemokines induces cellular responses that are involved in leukocyte traffic, most notably integrin activation, cytoskeletal changes, and chemotactic migration. CXCR3 is expressed primarily on activated T lymphocytes and NK cells and some epithelial cells and endothelial cells. It had been implicated that CXCR3 is highly expressed on pro-inflammatory human CD4+Th17 cells [35], "pathogenic" human Th1/17 cells [36], and myelin oligodendrocyte glycoprotein-specific CD4+ T cells in experimental autoimmune encephalomyelitis mice in combination with other biomarkers [37]. Increased CXCR3 in portal tracts has also been reported in PBC patients [27]. Additionally, levels of CXCR3 ligands are markedly increased in inflammatory liver diseases [38,39], including PBC [27]. Further, knockout of CXCR3 delays PBC progression in an animal model [40]. Most importantly, our data indicates that CXCR3 expression in CD4+ T cells inversely correlates with methylation levels and positively correlates with duration of disease. The accumulated data suggest a critical role of the CXCR3/chemokine system in the development of autoimmune disorders. In addition to its role in lymphocyte recruitment to tissues, CXCR3 is also implicated in the development of Th1 cells within lymph nodes [41]; thus, hypomethylation of CXCR3 and the consequent increased expression on CD4 T cells could promote disease progression through enhanced Th1 differentiation and increased recruitment. X chromosome demethylation has been shown to result in increased CXCR3 expression in lupus, suggesting this could be a common mechanism that contributes to the female susceptibility to autoimmune diseases [4]. Our findings are in agreement with previous studies; nevertheless, our findings strongly suggest an epigenetic (See figure on previous page.) Fig. 2 Differential methylation between PBC-affected subjects (n = 10) and healthy controls (n = 10) in CD4+, CD8+, and CD14+ cells represented in a heat map. The genomic locations of genes are shown on the left of the X chromosome ideogram. On the right, different methylated regions (DMR) of genes are identified in at least 6/10 samples in one group and 3/10 samples in the other. The predicted DMR along with the names of the genes in each cell examined is depicted as a two-dimensional heat map, respectively. Black blocks represent zero peaks, and yellow blocks symbolize different peak numbers corresponding to the scale bar at the bottom   There are significant limitations in our study that should be discussed. First, of the genes studied herein, we note there were no significant differences identified for UBE2A gene expression, although UBE2A has been demonstrated to be hypermethylated in CD4+ T cells. This suggests that UBE2A expression is regulated by other epigenetic mechanisms rather than by methylation. Second, the pathogenetic mechanisms that underlie CXCR3 demethylation remain to be defined. Third, our study did not include men with PBC. This is of particular interest given the relevant function of the X chromosome. Fourth, the CD4 bias is interesting and it will be important to determine if this bias is found on other T cell subpopulations, particularly Th17, Tregs, or Th1 cells, all of which require CXCR3 to enter the liver. Finally, further studies will also need to focus on the differential expression of central versus effector memory cells. Our data highlight, however, the critical importance of epigenetics in PBC and likely for other autoimmune diseases.

Conclusions
CXCR3 promoter demethylation in CD4+ T cells from PBC patients, supported by a significantly higher expression of CXCR3 in the same cell subtype, emphasizes the well-known role of CXCR3 in the pathogenesis of PBC. Moreover, CXCR3 expression levels in CD4+ T cells are significantly associated with duration of disease. Whether CXCR3 may contribute to the female susceptibility to PBC warrants further investigation.

Subjects
Fresh heparinized peripheral blood samples were obtained from 30 patients diagnosed with PBC and 30 unaffected age-and sex-matched controls [11]. All patients (Table 2) were female, and 90 % had readily detectable AMA; the diagnosis was made based on internationally accepted criteria [11]. The mean age was 61 (43-78) years old (range years), and 100 % of them were taking ursodiol. Four percent of the PBC patients included in this study were histologically characterized as cirrhotics. Serum liver function tests were performed utilizing routine laboratory methods. Subjects were excluded from the study if they had malignancies or were using immunosuppressive drugs. Patients and controls were matched for age and sex. After approval from appropriate institutional review board, all subjects provided written informed consent prior to enrollment in the study. The study design is illustrated in Fig. 1.

Methylated DNA immunoprecipitation and methylation microarrays
DNA samples from PBC patients (n = 10) and controls (n = 10) were sonicated to generate fragments between 200 and 1000 base pairs (bp), and then immunoprecipitated with a murine monoclonal antibody that specifically recognizes 5-methylcytidine (Diagenode, Liège, Belgium). The immunoprecipitated DNA fragments were recovered using anti-mouse IgG magnetic beads. The MeDIPenriched DNA was amplified by PCR using universal primers and a limited number of cycles (GenomePlex® Complete Whole Genome Amplification kit). The amplified DNA samples were then purified with QIAquick PCR purification kit (Qiagen, Valencia, CA, USA). One microgram of immunoprecipitated and reference DNA was tagged respectively with Cyanine-5 (Cy5) and Cyanine-3 (Cy3) labeled random 9-mers and then hybridized by the NimbleGen Array Hybridization Kit (Roche NimbleGen, Madison, WI, USA). The labeled DNA was then purified by isopropanol/ethanol precipitation. For array hybridization, Roche NimbleGen's Human Promoter plus CpG Island array was used, a 3 × 720k format array design containing 27,728 CpG islands and all wellcharacterized promoter regions (from about −2440 to +610 bp of the TSSs) totally covered by~720,000 probes.

Bisulfite sequencing
The CpG island DNA methylation status was determined by sequencing bisulfite-modified genomic DNA [43] from PBC patients (n = 20) and controls (n = 20). One microgram of genomic DNA was bisulfite-converted using the EpiTect Bisulfite Kit (Qiagen, Valencia, CA, USA). For  Table S1.

Reverse transcription and quantitative polymerase chain reaction
Total RNA was extracted from all isolated samples (PBC n = 30, controls n = 30) using TRI Reagent (Sigma-Aldrich, St. Louis, MO, USA). cDNA was synthesized from all different cell types using High Capacity cDNA Reverse Transcription Kits (Life Technologies-Applied Biosystems, Carlsbad, CA, USA). One hundred nanograms of cDNA in a total volume of 20 μl was amplified for 40 cycles on an Applied Biosystems 7900 HT Sequence Detection System, using a TaqMan Gene Expression Assay specific for CXCR3, UBE2A, FUNDC2, and IL1RAPL2 (Cat. #4331182, #4351372, Life Technologies-Applied Biosystems, Carlsbad, CA, USA). TaqMan® Gene Expression Assays consist of a pair of unlabeled PCR primers and a TaqMan® probe with a FAM™ dye label on the 5′ end and minor groove binder (MGB) nonfluorescent quencher (NFQ) on the 3′ end. RNA from samples of interest is reverse transcribed into cDNA, and the synthesized cDNA serves as the template for real-time PCR. ROX™ dye is used as a passive reference to normalize real-time PCR reactions. ROX™ dye provides an internal reference to which the FAM™ reporter dye signal can be normalized, which is necessary to correct for fluorescent fluctuations due to changes in concentration, volume, or light source intensity. Normalization of the reporter dye signal results in increased data precision.  All reactions were run in duplicate. The relative mRNA expression level of each gene was determined by normalizing its mRNA level to the internal control glyceraldehyde 3-phosphate dehydrogenase (GAPDH).

Flow cytometry
Phenotypic analysis of CD4 T cells was assessed using fluorochrome-conjugated monoclonal antibodies against cell surface markers on a FACSCantoII (BD Biosciences, San Diego, CA, USA) using FACSDiva™ software (BD Biosciences, San Diego, CA, USA). Monoclonal antibodies against human CD3, CD4, CD45RA, CD45RO, and CD183 (CXCR3) were purchased from BioLegend (San Diego, CA, USA). Frozen cells were thawed and prepared for flow cytometry as described [44]. Cells were stained for 15 min at room temperature with Aqua viability dye (Life Technologies). Optimal concentrations of the mAbs were used throughout, and all assays included positive and negative controls.

Statistical methods
NimbleScan software v2.5 (Roche NimbleGen, Madison, WI, USA) was utilized for DNA methylation data analysis using a threshold p value of 0.05 equivalent to 1.31 based on the Gaussian distribution of data. Second, exclusive elements corresponding to specific microarray probes were identified using a one-sided Kolmogorov-Smirnov (KS) test in affected and healthy subjects; peaks found only in either group were selected for further analysis. Peaks within 500 bp of each other are merged. If several adjacent probes rise significantly above a set threshold, the region is assigned to an enrichment peak (EP). NimbleScan detects peaks by searching for at least two probes above a p value minimum cutoff (−log10) of 2. Third, elements of interest were inserted into the UCSC Genome Browser to identify corresponding genes. The filter condition of different methylated regions was set to be no less than six EPs in one group and not more than three EPs in the other group. Based on GO analysis and potential involvement in the pathogenesis of PBC, four genes were selected for validation using isolated CD4+, CD8+, and CD14+ cells from 20 additional patients with PBC and 20 additional normal controls. For RT-qPCR, the data were analyzed with Delta Delta Ct using global normalization. Statistical differences between groups were identified using an unpaired t test taking into account transcripts with p value <0.05.

Additional file
Additional file 1: Supplementary table.