Cell-specific epigenome-wide DNA methylation in peripheral CD4(+) lymphocytes from patients with primary biliary cholangitis

Abstract

Background/aims

Primary biliary cholangitis (PBC) is a chronic cholestatic autoimmune liver disease, triggered by a complex interplay between genetic, environmental and epigenetic factors. We investigated the methylation profile of peripheral CD4(+) lymphocytes from PBC patients compared to healthy controls (HC) and autoimmune hepatitis (AIH) patients, to elucidate gene specific epigenetic modifications that contribute to PBC pathogenesis, as similar data are limited.

Methods

CD4(+) lymphocytes were isolated from 8 PBC treatment-naïve patients, 9 HC and 10 AIH patients at diagnosis by ROBOSEP platform. Whole genome methylation analysis was performed by 850k array of Illumina. Candidate genes’ transcriptional expression was quantified by RT-PCR.

Results

Comparison between PBC patients and HC, revealed 1016 differentially methylated positions (DMPs) on autosomal chromosomes and 1203 DMPs on X chromosome (>98 % hypermethylated), corresponding to 695 and 322 genes, respectively (p < 0.05). Hypermethylation mainly affected pathways of immune cells differentiation and signalling (CAMTA1, PRKARB, LNC01993, TLR9). Methylation analysis between PBC and AIH revealed >5000 DMPs (98 % hypermethylated in PBC) corresponding to >4000 genes. Pathway analysis retrieved an enrichment of “cytokine” and “interleukin” signalling (C1GTNF3, SMAD3). Analysis of differentially methylated regions showed enrichment on gene promoters and hypermethylation of IFN-regulated genes (IFNGR2, TICAM2) in PBC patients. Genes expression at the transcriptional level showed over-expression of IFNGR2 in PBC patients.

Conclusions

Hypermethylation characterizes most genes of peripheral CD4(+) lymphocytes in PBC. The epigenetic modifications mainly affect pathways of immunological responses. The significant number of X chromosome located DMPs, further supports the role of sex in PBC pathogenesis.

Graphical abstract

Highlights

• •Hypermethylation of the majority of genes of peripheral CD4(+) lymphocytes in PBC.
• •Enrichment of hypermethylated loci on ChrX supporting the role of gender in PBC.
• •Hypermethylation implicates pathways of immune response and signalling.
• •Distinct methylation patterns between AIH and PBC.

1. Introduction

Primary biliary cholangitis (PBC) is a chronic cholestatic autoimmune liver disease characterized by female predominance, antimitochondrial (AMA) and/or antinuclear (ANA) PBC-specific autoantibodies and a chronic non suppurative granulomatous cholangitis of small bile ducts on liver histology [[1], [2], [3]]. A complex interplay between genetic, environmental and epigenetic factors is implicated in disease pathogenesis [[4], [5], [6]].

Genetic predisposition of the disease has been mainly associated with specific HLA alleles: HLA-DRB1*08 is thought to confer to disease susceptibility, while HLADRB1*11 and HLADRB1*13 have been considered as protective alleles [4,[6], [7], [8]]. In addition, more than 40 non-HLA polymorphisms have been found to contribute to disease pathogenesis [9]. However, genetics are insufficient to explain the observed risk of disease occurrence, suggesting that epigenetic changes are also implicated.

Indeed, epigenome-wide association studies (EWAS) have showed altered methylation levels of several gene promoters (hypomethylation of CD40L and CXCR3; hypermethylation of UBE2A and FUNDC2) located at the X chromosome (ChrX), in peripheral mononuclear cells (PBMCs) and peripheral lymphocytes from PBC patients [[10], [11], [12]]. In addition, a study from our group highlighted that different transcriptional expression of DNA methyl-transferase-1 (DNΜΤ1) and ten-eleven-translocation-3 (TET3) enzymes, characterizes peripheral CD19(+) B-cells from PBC patients compared to healthy controls (HC) and autoimmune hepatitis (AIH) patients, although this expression was not associated with alterations in cells’ global methylation (5^mC)/hydroxymethylation (5^hmC) levels [13].

However, epigenetic studies in PBC are limited. Most of them investigated epigenetic alterations on mixed cell populations such as whole blood or PBMCs, so the results could be affected by the underlying heterogeneity in cell type composition, while the methylation analysis was frequently restricted to the genes of ChrX [[10], [11], [12], [13], [14]].

Therefore, in the present study we aimed to investigate the epigenetic profile of pure peripheral CD4(+) lymphocytes from PBC patients, compared to HC and AIH patients, in an attempt to determine the potential involvement of epigenetic modifications in PBC pathogenesis and detect the potential differences in the methylation profile between the two major autoimmune liver diseases, PBC and AIH.

2. Patients and methods

2.1. Study samples and PBMCs preparation

Patients with well characterized PBC, attended the Department of Medicine and Research Laboratory of Internal Medicine, Expertise Center in Autoimmune Liver Diseases, General University Hospital of Larissa, Greece, were enrolled in the study. Our cohort included 8 treatment-naϊve PBC patients, 9 HC and 10 AIH patients at diagnosis before immunosuppressive treatment initiation as disease control group. The 8 PBC patients were age and sex-matched with 6 HC and 8 AIH patients, while the HC and AIH groups included 3 and 2 male individuals respectively, that were also age matched.

The diagnosis of PBC and AIH was based on the respective international established criteria [1,[15], [16], [17]]. ANA, smooth muscle antibodies (SMA) and AMA, were detected by indirect immunofluorescence on fresh frozen sections of in-house rodent kidney, liver and stomach tissue substrates as described previously [18,19]. In addition, AMA and anti-soluble liver antigen/liver pancreas antibodies were detected by commercially available ELISA kits (INOVA Diagnostics, San Diego, CA, USA) and immunoblotting using rat liver mitochondrial and cytosolic extracts [18,19]. No patient was cirrhotic at diagnosis. Demographic, clinical, laboratory and histological data of the patients and HC are summarized in Table 1.

Table 1.

Patient's characteristics.

Patients' characteristics	PBC (n = 8)	HC (n = 9)	AIH (n = 10)
Age (years)	58 (37–76)	49 (29–78)	53 (33–78)
Gender (male/female)	0/8	3/6	2/8
Disease duration (months)	41 (4–184)		13 (1–122)
AST (<40 IU/L)	37 (16–88)	20 (15–27)	107 (40–514)
ALT (<40IU/L)	49 (11–120)	18 (13–38)	230 (59–548)
Bilirubin (0.1–1 mg/dL)	0.62 (0.18–2.28)	0.51 (0.2–1.1)	1.3 (0.43–9.96)
γ-GT (<45U/L)	171 (12–297)	18 (5–43)	84 (20–273)
ALP (<120U/L x ULNa)	1.4 (0.5–2.25)a	0.6 (0.4–0.8)a	0.7 (0.3–1.1)a
IgG (<1650 mg/dL)	1400 (975–1622)	881 (658–1100)	2250 (1480–2920)
IgM (64–249 mg/dL)	246 (88–801)	106 (69–212)	166 (22–332)
ANA (positive/negative)	1/7	0/9	3/7
SMA (positive/negative)	0/8	0/9	10/10
AMA (positive/negative)	8/8	0/9	0/9
SLA/LP (positive/negative)	0/8	0/9	1/9
Simplified AIH score (>6 definite ΑIΗ)			7 (6–8)

Values are expressed as median (range). Abbreviations: AIH, autoimmune hepatitis; PBC, primary biliary cholangitis; HC, healthy controls; ANA, antinuclear antibodies, AMA, anti-mitochondrial antibodies; SMA, smooth muscle antibodies; anti-SLA/LP, antibodies against soluble liver antigen/liver pancreas.

^aValues are expressed as ratio to the upper limit of normal (ULN).

PBMCs were isolated by gradient centrifugation separation medium (Histopaque-1077, Sigma-Aldrich, St. Louis, MI, USA), and were then mixed with freezing medium (FBS with 10 % DMSO Sigma-Aldrich) and cryopreserved in liquid nitrogen.

All subjects agreed to participate by written consent at the time of initial evaluation. The ethical committee of the General University Hospital of Larissa approved the study protocol which conforms to the ethical guidelines of the Declarations of Helsinki, as reflected in a priori approval by the institution's human research committee (2258/21-3-2016).

2.2. CD4(+) T-lymphocytes isolation and flow cytometry analysis

Isolation and purification of CD4(+) T-lymphocytes were performed by ROBOSEP^TM-S platform (Stemcell Technologies, Vancouver, USA) and a FITC-anti-CD3/PE-anti-CD8 antibody (Biolegend Inc.), respectively, as we described in our previous work [13]. Flow cytometry analysis for the determination of the number and purity of cells was performed on Coulter FC-500 flow cytometer (Beckman-Coulter, Brea, CA, USA) (Supplementary Fig. 1). An average of 2-5x10⁶ number of CD4(+)-cells were isolated per sample and stored at −80⁰ C.

2.3. DNA/RNA extraction and quantification

Genomic DNA was extracted from CD4(+)-cells with QIAamp Blood mini purification kit (Qiagen, Hilden, Germany). Quantification was initially measured at 260 nm in a UV–VIS Nanodrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and DNA purity was determined by the ratio of 260–280 nm absorbance levels. Afterwards DNA quantity and purity was further confirmed using a highly sensitive fluorescent dye, PicoGreen. Approximately, 1–3 μg of genomic DNA were isolated from CD4(+)-cells.

2.4. EWAS

According to our previous report [13], 500 ng of DNA from CD4(+) lymphocytes from patients with PBC, AIH and HC were bisulfite-converted (Zymo Research, Irvine, California, USA), and DNA methylation was assessed by hybridising bisulfite-converted DNA to the Human Methylation EPIC array BeadChip (Diagenode SA, Belgium), which permits the interrogation of over 850000 methylation sites throughout the genome at single-nucleotide resolution. These investigations were done by using NXT-DX Company (Gent, Belgium) according to the instructions of the manufacturer. Data regarding methylation were provided as β values: β = M/(M + U), where M is the fluorescent signal of methylation and U the respective signal of unmethylated probe. The β values ranged from 0 to 1 (no methylation to 100 % methylation, respectively). Bioconductor R package, Chip Analysis Methylation Pipeline (ChAMP) was used to assess quality control, according to which none of the samples had a proportion of failed probes >0.1. No significant source of variation was identified after normalization.

2.5. Reverse transcription polymerase chain reaction (RT-PCR) analysis

To provide additional support of the EWAS findings, we used RT-PCR in the same patient cohort to evaluate expression of certain candidate genes that emerged from the DNA methylation. C1q and Tumor Necrosis Factor Related Protein 3 (C1QTNF3), Toll-like receptor (TLR) 9 and interferon gamma receptor 2 (IFNGR2) were selected for mRNA quantification. One μg of total RNA from CD4(+) lymphocytes was reversed transcribed using random hexamer primers and 10 Units of Transcriptor Reverse Transcriptase according to the Transcriptor First Strand cDNA synthesis kit (Roche Diagnostics, Basel, Switzerland) in a 20 μL reaction for 10 min at 25 °C, 60 min at 50 °C and 5 min at 85 °C. Quantitative PCR was carried out using Quantinova SYBR Green PCR Kit (Qiagen GmbH, Hilden, Germany) in a total volume of 15 μL containing 0.7 μM of LNA-PCR specific forward and reverse primers. Amplification and detection were performed in a Lightcycler^R 480 Instrument (Roche Life Sciences, Bavaria, Germany). For the quantification of the expression of human C1QTNF3, TLR9, IFNGR2 and GAPDH genes, commercially LNA PCR Assays were used according to the manufacturers recommendations: HS_C1QTNF3_1539589, HS_TLR9_2475290, HS_IFNGR2_1318476, HS_GAPDH_2475369 (Qiagen). The mRNA of human GAPDH was chosen as internal control and relative quantification on gene expression was based on the comparative CT method [2^{−(ΔCT target−ΔCT calibrator)} or 2^−ΔΔCT].

2.6. Statistical analysis

The SPSS 24 and Graph Pad-Prism 8.0 software were used for the analysis. Results were expressed as mean ± standard deviation (SD) or median (range) where appropriate. Kruskall-Wallis and Mann-Whitney U test were applied for the detection of differences between independent samples. Statistical analysis of the EWAS was performed with the Bioconductor R package ChAMP [20]. After normalization, identification of differentially methylated positions (DMPs), applying to differentially methylated cytosine-phosphate-guanine dinucleotides (CpGs), between groups was performed using Benjamini-Hochberg adjusted p value < 0.05. Differentially methylated regions (applying to regions of the genome mostly enriched by DMPs) (DMRs) between groups, were identified with BumpHunter method and adjusted p value < 0.05 [21]. Gometh method was used for gene set enrichment analysis, considering the genes with the largest number of differentially methylated CpGs [22,23]. Finally, in order to identify common molecular functional properties between enriched genes, we used the Panther Geneontology Unifying Biology Classification System 16.0 (http://www.pantherdb.org/) which classifies genes according to the function of the protein by itself or with directly interacting proteins at a biochemical level and creates gene pathways specifying the relationships between the interacting molecules. Gene pathways were built using the Fisher's exact test.

3. Results

3.1. EWAS in CD4(+) lymphocytes from PBC patients compared to HC

Comparison between PBC patients and HC, revealed 1016 DMPs (99.8 % hypermethylated in PBC patients) corresponding to 695 unique and annotated genes in autosomal chromosomes (p < 0.05) (Fig. 1A). The majority of DMPs were present in genomic areas far from CpG islands and within open sea regions and gene bodies as shown in Fig. 1B and C.

Fig. 1.

EWAS between PBC and HC. A) Heatmap of significant CpGs corresponding to 1016 DMPs detected between PBC (n = 8) and HC (n = 9) (P1-8: PBC, C1-9: healthy controls). B) Genomic distribution of significant DMPs. DMPs are enriched in open sea regions and shores. C) Most DMPs are located on gene bodies and IGR. D) Methylation status of 7 DMPs comprising SGCE gene. SGCE gene is hypermethylated in PBC compared to HC and all DMPs are located on TSS1500 and within the CpG island. Abbreviations: PBC, primary biliary cholangitis; HC, healthy controls; CpGs, cytosine-phosphate-guanine dinucleotides; DMPs, differentially methylated positions; SGCE, sarcoglycan epsilon; TSS, transcription start sites; IGR, intergenic regions; 5′UTR, 5′-untranslated region; 3′UTR, 3′-untranslated region; Hyper, hypermethylated; Hypo, hypomethylated.

Among the top 50 genes with the greatest number of DMPs (≥2 DMPs per gene) (Supplementary Fig. 2), the most enriched one was SGCE (sarcoglycan epsilon) gene (Fig. 1D), a transmembrane protein that is expressed also in the liver and links the actin cytoskeleton to the extracellular matrix [24]. A significant number of hypermethylated DMPs in PBC patients located on genes of innate and adaptive immune response pathways, as shown in Table 2. Given that more prominent methylation differences, may reflect more meaningful biological alterations, we also searched for genes with a Δβ cut-off ≥0.1 (Supplementary File 2), and identified that 14/50 (28 %) of the mostly enriched genes presented at least one DMP with Δβ ≥0.1, with the greatest methylation difference (Δβ = 0.167) observed in arachidonic acid 12-lipoxygenase (ALOX12) gene.

Table 2.

Top differentially methylated genes between PBC and HC implicated in innate and adaptive immune response pathways.

Gene	Chr	Description	DMPs	Δβ (PBC vs HC)	p	Pathways
ALOX12	1	Arachidonate 12-Lipoxygenase	cg03760483	0.169	0.04	Eicosanoid Synthesis
			cg03404566	0.161	0.04
			cg05657792	0.145	0.02
C1QTNF3	5	Complement C1q Tumor Necrosis Factor-Related Protein 3	cg24699745	0.132	0.04	Host defence, inflammation, and glucose metabolism
			cg17617843	0.080	0.02
			cg18752880	0.050	0.02
CAMTA1	1	Calmodulin Binding Transcription Activator 1	cg24865247	0.063	0.03	T-lymphocytes differentiation
			cg05685045	0.055	0.04
			cg06129585	0.054	0.02
CD19	16	B-Lymphocyte Antigen CD19	cg24900963	0.091	0.03	Downstream signalling events of B Cell Receptor
			cg07322144	0.072	0.04
LINC01993	17	Long Intergenic Non-Protein Coding RNA 1993	cg13074526	0.122	0.04	Development of diverse immune cells and mechanisms of immune cell activation
			cg01864395	0.086	0.03
			cg03067296	0.080	0.04
PRKAR1B	7	Protein Kinase CAMP-Dependent Type I Regulatory Subunit Beta	cg10251229	0.107	0.03	T lymphocytes transport
			cg04952627	0.075	0.03
			cg05725489	0.056	0.04
RPTOR	17	Regulatory Associated Protein of mTOR Complex 1	cg03641032	0.066	0.04	Control of the target of rapamycin, regulation of cell growth, TH17 cell development
			cg15946337	0.065	0.04
			cg06053702	0.051	0.04
			cg14596352	0.045	0.04
SQSTM1	5	Ubiquitin-binding protein p62	cg15126957	0.076	0.02	Autophagy receptor required for selective macro-autophagy, regulation by FXR
			cg01325271	0.059	0.02
TLR9	3	Toll like receptor, CD289	cg16302310	0.072	0.03	Immune response against pathogens through recognition of molecular patterns specific to microorganisms
			cg17716965	0.051	0.04

Chr, chromosome; DMP, differentially methylated positions; Δβ, methylation difference PBC; primary biliary cholangitis; HC, healthy controls; FXR, farnesoid X receptor.

Separate analysis was conducted for genes located on ChrX. We identified 1203 DMPs corresponding to 322 unique and annotated genes (p < 0.05) (98.5 % hypermethylated in PBC patients compared to HC). Analyses of the genes with the greatest number of DMPs (≥5 DMPs per gene, n = 40) (Supplementary Table 1) revealed hypermethylation of TSC22 domain family member 3 (TSC22D3), Dual-Specificity Phosphatase 9 (DUSP9), as well as, of two zinc-finger proteins ZBTB33 and ZNF75D. A 63 % (25/40) accounted for at least one DMP with Δβ ≥0.1, with the most prominent difference observed for ARHGEF6 (Δβ = 0.243), involved in CD4(+) T cell migration and development (Supplementary File 2). To further assess whether the ChrX DMPs identified might be influenced by sex-related epigenetic regulation, we investigated whether they are known to escape X chromosome inactivation (XCI). Using published XCI escape maps from lymphoblastoid and immune cell–based datasets [[25], [26], [27]], we found that 20 of the 40 genes (50 %) are located in regions that commonly escape XCI across tissues (Supplementary Table 1). Notably, ARHGEF6, ZBTB33 and ZNF75D genes, all of which were found to be highly hypermethylated in PBC, were also found to escape the inactivation process.

Although GOmeth did not reveal any enriched pathway, Panther could categorize 763 genes in 8 groups according to molecular functions (Supplementary Fig. 2). Interestingly, among implicated pathways, “Toll receptor signalling” and “T cell activation” pathways were included, comprising of 10 genes (Supplementary Table 2). The main functional annotations over-represented were those classified as “kinase activity” and “protein binding” (Fold enrichment 1.75, p < 0.001; Fold enrichment 1.15, p < 0.001 respectively).

Further analysis of extended segments of the genome revealed 331 DMRs, 76 % located on gene promoters and transcription start sites (TSS). Among the key immunity related genes identified were CD24 (hypermethylated, p = 0.002), CD96 (hypomethylated, p = 0.01), chemokine receptor like 2 (CCRL2; hypomethylated, p = 0.007), and two interferon-regulated genes, interferon gamma receptor 2 (IFNGR2; hypermethylated, p = 0.01) and TLR adaptor molecule 2 (TICAM2; hypermethylated, p = 0.011).

3.2. EWAS in CD4(+) lymphocytes from PBC patients compared to AIH patients

To investigate the differences in the methylation status between the two major autoimmune liver diseases we conducted a comparison of whole genome methylation analysis between PBC (n = 8) and AIH patients (n = 10). Analysis at the CpG level revealed 5766 DMPs corresponding to 4018 unique and annotated genes, 98 % hypermethylated in PBC (Fig. 2A). Association analysis of ChrX located DMPs revealed 73 DMPs corresponding to 51 genes, while functional genomic distribution retrieved enrichment in open sea regions, present at more than 4 Kb of CpG island and within gene bodies and intergenic regions (IGR) (Fig. 2B and C).

Fig. 2.

EWAS between PBC and AIH

A) Heatmap of significant CpGs corresponding to 5000 DMPs detected between PBC (n = 8) and AIH (n = 10) (P1-8: PBC, AIH1-10: autoimmune hepatitis). B) Genomic distribution of significant DMPs at CpG island: DMPs are enriched in open sea regions and shores. C) Genomic distribution of significant DMPs at gene: Most DMPs are located on gene bodies and IGR. Abbreviations: PBC, primary biliary cholangitis; AIH, autoimmune hepatitis; CpGs, cytosine-phosphate-guanine dinucleotides; DMPs, differentially methylated positions; TSS, transcription start sites; IGR, intergenic regions; 5′UTR, 5′-untranslated region; 3′-UTR, 3′untranslated region; Hyper, hypermethylated; Hypo, hypomethylated.

Among the top 70 genes mostly enriched by significant DMPs (≥3) (Supplementary Fig. 3), 49/70 (70 %) presented at least one DMP with a Δβ ≥1 (Supplementary File 2). The most pronounced difference in methylation was detected for AT-Rich Interaction Domain 5B (ARID5B) gene, a transcriptional co-activator. Top differentially methylated genes between PBC and AIH patients were implicated in inflammatory and immune response pathways (Table 3). Of note, actinin alpha 4 (ACTN4) gene was found among the most hypermethylated (Δβ ≥0.1) in PBC compared to AIH patients.

Table 3.

Top differentially methylated genes between PBC and AIH patients implicated in immune response pathways.

Gene	Chr	Description	DMPs	Δβ (PBC vs AIH)	p	Pathways
ADAP1	7	ArfGAP With Dual PH Domains 1	cg01075915	0.05	0.04	B Cell Receptor Signalling
			cg02381321	0.04	0.03
			cg13134104	0.03	0.03
ADGRG3	16	Adhesion G Protein-Coupled Receptor G3	cg23698652	0.09	0.04	Regulator of migration of lymphatic endothelial cells in vitro via the small GTPases RhoA and CDC42
			cg13697988	0.07	0.04
			cg24979119	0.06	0.04
C1QTNF3	5	C1q And TNF Related 3	cg24699745	0.128	0.04	Growth factor that inhibits production of inflammatory cytokines from human adipocytes, monocytes and fibroblasts
			cg23051392	0.063	0.04
			cg17617843	0.064	0.04
			cg18752880	0.041	0.04
			cg00760266	0.012	0.03
ITGAE	17	Integrin Subunit Alpha E	cg17074014	0.135	<0.001	Mediator of cell adhesion of intra-epithelial T-lymphocytes to epithelial cell monolayers-ERK signalling
			cg24738036	0.152	<0.001
			cg23770271	0.128	<0.001
			cg09552070	0.118	<0.001
			cg07323451	0.062	<0.001
MSI2	16	Musashi RNA Binding Protein 2	cg14884512	0.140	0.04	RNA binding protein that regulates the expression of target mRNAs at the translation level
			cg21668652	0.093	0.04
			cg11872669	0.096	0.04
			cg03984067	0.057	0.04
			cg20318252	0.022	0.04
			cg06247967	0.021	0.04
			cg19480933	0.023	0.04
			cg19098004	0.029	0.04
MLST8	16	MTOR Associated Protein, LST8 Homolog	cg01500402	0.082	0.04	mTOR signalling and downstream signalling events of B Cell Receptor
			cg24464795	0.063	0.04
			cg15487687	0.064	0.04
			cg03910196	0.052	0.04
			cg08737236	0.031	0.03
SMAD3	15	SMAD Family Member 3	cg05438378	0.136	0.04	Differentiation of CD4(+) T cells in ΤΗ17 και FOXP3+ lymphocytes
			cg21106065	0.129	0.04
			cg04133879	0.125	0.04
			cg10711190	0.043	0.02
SH3BP5	3	SH3 Domain Binding Protein 5	cg19069769	0.140	0.04	Negative regulatory role in BTK-related cytoplasmic signalling in B-cells
			cg01597905	0.136	0.04
			cg03495084	0.097	0.04
			cg07480608	0.189	0.04
TRIM26	6	Tripartite Motif Containing 26	cg14799927	0.093	<0.01	Innate immune system and interferon gamma signalling
			cg19714347	0.072	<0.01
			cg19321263	0.013	<0.01
			cg07011235	0.014	<0.01

Chr, chromosome; DMPs differentially methylated positions; cytosine-phosphate-guanine dinucleotide; Δβ, methylation difference; PBC, primary biliary cholangitis; AIH, autoimmune hepatitis; GTP, guanosine triphosphate; RhoA, ras homolog family member A; CDC42, cell division control protein 42 homolog; mTOR, mammalian target of rapamycin; ERK, extracellular signal-regulated kinases; BTK, Bruton's tyrosine kinase.

Although GOmeth did not identify any enriched pathway, Panther could categorize 20 pathways comprising of 2776 identified genes. Among the enriched pathways, “Interleukin signalling” and “Inflammation mediated by chemokine and cytokine signalling” were identified, comprising of 20 and 55 genes respectively (Supplementary Table 3).

In addition, we further analysed the 68 DMRs detected between PBC and AIH patients. Interestingly, 52 DMRs were hypermethylated in PBC compared to AIH, while 72 % of them were around gene promoters and TSS. Among the corresponding genes, C-Type Lectin Superfamily 2 Member D (CLEC2D; (hypomethylated, p = 0.002), Interferon Regulatory Factor 6 (IRF6; hypermethylated, p = 0.01), Lymphocyte Transmembrane Adaptor 1 (LAX1; hypermethylated, p = 0.01), S100 Calcium-Binding Protein P (S100P; hypermethylated, p = 0.03) and Zeta-Chain Associated Protein Kinase, 70kD (ZAP70; hypomethylated, p = 0.01) were retrieved.

3.3. C1QTNF3, TLR9 and IFNGR2 transcriptional expression between groups

In order to explore the effects of methylation alterations on genes transcriptional expression we proceeded to mRNA quantification of selected genes C1QTNF3, TLR9 and IFNGR2, which were found among the top hypermethylated genes in PBC, but also among the genes included in the mostly enriched pathways in PBC patients. Our results showed higher expression of INFGR2 in PBC patients compared to HC (8.312 vs.1.122, p < 0.001) and AIH patients (8.312 vs. 2.800, p < 0.05), while a trend of higher expression of C1QTNF3 was observed in PBC patients compared to HC (3.950 vs.2.266, p = 0.069). No difference in TLR9 expression at the transcriptional level was observed between groups (Fig. 3).

Fig. 3.

C1QTNF3, TLR9 and IFNGR2 transcriptional expression between groups. The levels of mRNA for IFNGR2, C1QTNF3, and TLR9 in PBC (n = 8), HC (n = 9) and AIH (n = 10) are presented in a logarithmic scale. Abbreviations: IFNGR2, interferon gamma receptor 2; C1QTNF3, C1q and tumor necrosis factor related protein 3; TLR9, Toll-like receptor 9; PBC, primary biliary cholangitis; AIH, autoimmune hepatitis; HC, healthy controls.

3.4. Comparison with published data

Finally, differentially methylated genes and DMPs obtained from the analysis of ChrX in PBC patients compared to HC, were compared with reported data from PBMCs and peripheral CD4(+) cells [10,11]. As presented in Fig. 4, a significant overlap of differentially methylated genes was observed between different studies. More precisely, 26/30 genes were found hypermethylated in all studies, while 4/30 were found hypomethylated in one of the previous studies and hypermethylated in our study (Supplementary Table 4). Of note, isocitrate dehydrogene NAD(+) gene (IDH3G), which plays a role in the allosteric regulation of the enzyme catalysing the decarboxylation of isocitrate into alpha-ketoglutarate, was found differentially methylated in all studies.

Fig. 4.

Overlap of differentially methylated genes of ChrX with previous studies. Twenty genes were found hypomethylated in the present study and the study of Selmi et al., [11] while 10 genes were found differentially methylated in the present study and the study of Lleo et al. [12]. One gene, isocitrate dehydrogene NAD(+) gene (IDH3G), was found differentially methylated in all studies.

4. Discussion

In the present study we performed a whole genome methylation analysis of peripheral CD4(+) lymphocytes from PBC patients compared to HC and patients with AIH. Our results confirm the hypermethylation of most genes of peripheral CD4(+) lymphocytes in PBC and demonstrate for the first time that PBC and AIH are characterised by different methylation profiles. Differences in methylation affect pathways of CD4(+) T-cells immune response, differentiation and signalling and suggest an epigenetic contribution to their pivotal role in PBC pathogenesis. In addition, our study further supports the notion that the female predominance in PBC is driven in part by immune genes differentially methylated on ChrX.

The hypermethylation of several genes of peripheral immune cells in PBC has been described in previous studies [10,11]. Our results confirm these findings, not only by highlighting a hypermethylation of the vast majority of genes in peripheral CD4(+) lymphocytes from PBC patients compared to HC, but also by demonstrating a significant overlap between the hypermethylated genes annotated by different studies. The discrepancies between ours and previous results, regarding the different methylation pattern of ATP6AP2, RAP2C, HTATSF1 and IDH3G genes, (hypomethylated in previous studies), could be attributed to the different methods used for the methylation analysis, the analysis of different cell populations and the inclusion of treatment-naïve patients versus both untreated and treated patients in previous studies.

Although the mechanism is not clear, CD4 (+) T-cells seem to play a pivotal role in PBC pathogenesis [28]. PRKARB gene, encoding a principal signal transduction system in T-cells [29], and LINC01993 gene, codifying a long non-coding RNA that coordinates the differentiation and activation of distinct T-cell subsets [30], were found hypermethylated in PBC patients. Accordingly, we detected hypermethylation of CAMTA1 gene, a calcium dependent transcriptional factor participating in T-cell differentiation, whose methylation pattern confers to the characterization of different Treg phenotypes [31]. In addition, gene set enrichment analysis highlighted the “TLR signalling” and the “T-cell activation” among the mostly enriched hypermethylated pathways. All together these findings support an epigenetic contribution in the regulation of T-cell responses in PBC, that precises further investigation.

C1QTNF3, TLR9, and IFNGR2 were selected for transcriptional analysis based on a combination of statistical significance, pathway enrichment, and known relevance to PBC pathogenesis. C1QTNF3 was differentially methylated across all three groups (Δβ = 0.132 for PBC vs. HC, Δβ = 0.128 for PBC vs. AIH), suggesting a robust disease-specific signal. TLR9, despite showing lower Δβ values (<0.1), was included due to its key role in the IgM production in PBC and its presence in highly enriched pathways (Toll Receptor signalling) [32]. IFNGR2 was among the most significantly hypermethylated DMRs in PBC vs. HC (p = 0.01) and was selected because of its direct involvement in the “interferon signature” [33], a mechanism strongly implicated in PBC and also highlighted in previous epigenetic studies [11]. However, only IFNGR2 mRNA expression was higher in PBC compared to HC and AIH patients. This indicates that the effect of DNA methylation on gene's transcription is not completely predictable and is affected by other epigenetic mechanisms or post-transcriptional modifications. Increased IFNGR2 mRNA expression is in accordance with the “interferon signature” in PBC pathogenesis. This is further supported by the fact that among annotated genes of detected DMRs we could retrieve two IFN-regulated genes, IFNGR2 and TICAM2, and one IFN-regulatory element (IRF6). Recently, an EWAS study in the whole blood of PBC patients [14] revealed a hypomethylation of several CpGs associated with an enrichment in T-cell activation and IFN-signalling. This indicates that beyond discrepancies in the methylation status and the variability of annotated genes among different studies the associated pathways remain the same.

Due to its strong female predominance, the role of ChrX in PBC occurrence was always speculated [[34], [35], [36]], but it was not until recently that the ChrX contribution to PBC genetics was revealed [37]. In the present study, we identified more than 300 hypermethylated genes on ChrX in PBC patients. While it is not possible to entirely exclude the influence of XCI on these findings, we performed a targeted analysis to address this issue. Among the 40 ChrX genes with the highest number of DMPs (≥5), 50 % are known to escape XCI, based on curated datasets derived from immune and lymphoid cells [[25], [26], [27]]. This includes key genes such as ARHGEF6, DUSP9, ELF4 and ZBTB33 that preferentially escape inactivation in immune cells. These findings suggest that many of the observed methylation changes may arise from loci that are epigenetically active on both X chromosomes in females and strengthen the biological plausibility of our findings and arguing against sex bias alone driving these results. Furthermore, differential methylation of ChrX genes was observed exclusively between PBC and HC, but not between PBC and AIH, despite both control groups included male individuals. This provides additional support that the ChrX signature reflects PBC-specific epigenetic alterations rather than general sex-related effects. Our results align with recent epigenome-wide studies that also demonstrated a significant enrichment of disease-associated signals among ChrX genes in PBC [6,38,39]. Taken together, these findings highlight a potential disease-specific, epigenetically mediated dysregulation of ChrX that may contribute to the striking female predominance observed in PBC.

One of the main annotations of our study was the detection of a methylation shift between PBC and AIH. The different methylation pattern was more pronounced in the “cytokine” and “interleukin” signalling pathways including SMAD3 and C1QTNF3 genes. SMAD3 acts as a transcriptional factor that induces differentiation of CD4(+)-cells in Th17 and FOXP3(+) cells, maintaining a balance between pro-inflammatory and immunoregulatory mechanisms [40], while C1QTNF3 has been shown to suppress Th17 cells differentiation in animal models [41]. In PBC an imbalance between Th17/Tregs in favour of an up-regulation of Th17 is described [42], while in AIH Th17 cells are key effector T-cells that regulate its pathogenesis via induction of hepatic IL-6 expression [43]. According to our findings, an epigenetic regulation may lie behind these mechanisms and could be the target for future studies. Another key gene that was found hypermethylated in PBC compared to AIH was ACTN4 which encodes one of the four isoforms of α-actinin, one of the main autoantigens in about 40 % of AIH patients, and a marker of disease activity and treatment response [[44], [45], [46]]. This suggests that DNA methylation may contribute to shaping disease-specific autoantigen profiles by modulating the expression of immune-relevant targets,

The main limitations of our study are the small cohort of patients and the analysis of epigenetic modifications in peripheral rather than in liver infiltrating lymphocytes. However, the analysis of treatment-naïve patients and of pure CD4(+)-cells confers reliability to our results. Secondly, our analysis is restricted to the investigation of alterations only in gene methylation, which cannot predict a clear transcriptional effect on the implicated gene, especially if we consider that methylation differences were not always found in gene promoters [47] and that DMPs with low absolute methylation (ΔΒ<0,1) difference, have less commonly functional implications. However, our results at the transcriptional level indicate that the hypermethylation of CD4(+) genes in PBC and the methylation shift between PBC and AIH, do have a transcriptional impact on the implicated immune response pathways and may contribute to disease pathogenesis.

In conclusion, our results showed an aberrant gene DNA hypermethylation in peripheral CD4(+) lymphocytes from PBC patients and highlight for the first time a distinct methylation profile between PBC and AIH, that involves pathways of immune response and signalling. This could reflect the differences in the pathogenesis of these two major autoimmune liver diseases. These preliminary results need further confirmation in order to clarify which is the role of epigenetic marks in the pathogenesis of PBC and AIH, how do these marks affect genes expression at the transcriptional, as well as, at the protein level and how these pathways could represent a new target for epigenetic based therapeutic interventions.

CRediT authorship contribution statement

Pinelopi Arvaniti: Writing – review & editing, Writing – original draft, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation. Kalliopi Zachou: Writing – review & editing, Writing – original draft, Supervision, Formal analysis, Conceptualization. Aggeliki Lyberopoulou: Writing – review & editing, Methodology, Investigation, Formal analysis, Data curation. Eirini Sevdali: Writing – review & editing, Methodology, Investigation, Formal analysis, Data curation. Stella Gabeta: Writing – review & editing, Methodology, Investigation, Formal analysis, Data curation. Matthaios Speletas: Writing – review & editing, Validation, Supervision, Methodology, Conceptualization. Yves Renaudineau: Writing – review & editing, Validation, Supervision, Methodology, Conceptualization. George N. Dalekos: Writing – review & editing, Writing – original draft, Supervision, Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

figs1.

figs2.

figs3.

figs4.

Data availability

Data will be made available on request.