Genome-wide DNA methylation pattern in systemic sclerosis microvascular endothelial cells: Identification of epigenetically affected key genes and pathways

Abstract

Background:

The etiology of systemic sclerosis is not clear, but there is evidence suggesting a critical role for epigenetic alterations in disease pathogenesis and clinical expression. We sought, in this study, to characterize the genome-wide DNA methylation signature in systemic sclerosis microvascular endothelial cells.

Methods:

We performed a genome-wide DNA methylation study in microvascular endothelial cells derived from seven diffuse cutaneous systemic sclerosis patients compared to seven age-, sex-, and ethnicity-matched healthy controls. We paired matched samples on Illumina HumanMethylation450 (three diffuse cutaneous systemic sclerosis microvascular endothelial cells and three controls), and reproduced the results in an independent set of matched patient and controls using Illumina Infinium MethylationEPIC (four diffuse cutaneous systemic sclerosis patients and four controls) to identify differentially methylated genes.

Results:

We identified 71,353 differentially methylated CpG sites in systemic sclerosis microvascular endothelial cells using Infinium MethylationEPIC microarray in the first group (0.081% of representative probes) and 33,170 CpG sites in the second group using HumanMethylation450 microarray (0.073% of representative probes) in diffuse cutaneous systemic sclerosis microvascular endothelial cells. Among the two groups of subjects, we identified differential methylation of 2455 CpG sites, representing 1301 genes. Most of the differentially methylated CpG sites were hypermethylated (1625 CpG), corresponding to 910 genes. Common hypermethylated genes in systemic sclerosis microvascular endothelial cells include NOS1, DNMT3A, DNMT3B, HDAC4, and ANGPT2. We also identified hypomethylation of IL17RA, CTNNA3, ICAM2, and SDK1 in systemic sclerosis microvascular endothelial cells. Furthermore, we demonstrate significant inverse correlation between DNA methylation status and gene expression in the majority of genes evaluated. Gene ontology analysis of hypermethylated genes demonstrated enrichment of genes involved in angiogenesis (p = 0.0006). Pathway analysis of hypomethylated genes includes genes involved in vascular smooth muscle contraction (p = 0.014) and adherens junctions (p = 0.013).

Conclusion:

Our data suggest the presence of significant genome-wide DNA methylation aberrancies in systemic sclerosis microvascular endothelial cells, and identify novel affected genes and pathways in systemic sclerosis microvascular endothelial cells.

Introduction

Systemic sclerosis (SSc) is an autoimmune connective tissue disorder that is characterized by microvascular injury, excessive tissue fibrosis, and distinctive visceral changes that can affect the lungs, heart, kidneys, gastrointestinal tract, and other organs. ¹ The exact etiology of SSc is still unknown. The microvascular endothelial cells (MVEC) exist as continuous monolayer of cells connected closely with basal lamina and they are involved in important functional tasks such as regulation of coagulation and fibrinolysis, permeability, vasoreactivity, cellular metabolism, and nutrition. ² Also, the MVEC have other functions that include fluid filtration, blood vessel tone, hemostasis, neutrophil recruitment, and hormone trafficking. ³ Dysregulation of MVEC function within the vascular wall plays an important role in vascular tone control and vascular remodeling associated with the fibro-proliferative vasculopathy observed in SSc. In fact, MVEC injury is proposed as a crucial initiating event that leads to vascular remodeling in SSc with intimal proliferation of arterioles and capillary breakdown and finally, blood vessel occlusion. ⁴ Therefore, studying MVEC dysfunction in SSc is important in identifying endothelial biological pathways, which lay the ground work for understanding the mechanism/(s) of the disease that may lead to devising new and improved therapies for SSc. ⁵

Epigenetic alterations of genomic DNA play a critical role in many important human diseases, and in SSc. ⁶ DNA methylation is considered the core epigenetic mechanism that regulates gene expression by altering transcriptional accessibility of regulatory regions within gene sequences. There is a growing evidence of gene-specific epigenetic alterations in SSc.^7,8 We have identified epigenetic aberrancies in SSc fibroblasts at the genome-wide level that contribute to the abnormal activated phenotype of SSc fibroblasts. ⁹

In this study, we performed a genome-wide DNA methylation study in MVEC from patients with diffuse cutaneous systemic sclerosis (dcSSc) compared to control-MVEC. Our aim was to gain an unbiased understanding of the pathogenesis of SSc through identifying the differentially methylated genes in dcSSc–MVEC and to characterize the biological pathways enriched by these genes. We identified aberrant methylation in key genes and pathways between dcSSc and control-MVEC that are pertinent to the pathogenesis of SSc. Moreover, we demonstrate that these aberrancies have functional implications in terms of gene expression represented by inverse correlation between gene-specific DNA methylation levels and gene expression in the majority of genes that we evaluated in this study.

Methods and materials

MVEC isolation

We obtained punch skin biopsies from seven patients with dcSSc and seven age-, sex-, and ethnicity-matched controls. Table 1 summarizes the clinical characteristics of subjects who participated in this study. The average age of patients with SSc was 55.3 years and for controls was 54.7 years (p = 0.8). The average duration of SSc was 5.9 years from the onset of the first non-Raynaud’s phenomena manifestation. We obtained 4 mm punch skin biopsies under strict aseptic technique from involved skin of the forearms of patients with dSSc and a similar site from control subjects. The study was approved by the institutional review board of the University of Toledo Medical Center. All study participants signed an informed consent to participate in this study. After obtaining punch skin biopsies, we dissected and removed the epidermis, and the dermis layer was cut into small pieces and incubated for 1 h at 37°C in serum-free medium (EGM-2 MV, Lonza) containing trypsin at a final concentration of 0.5% as described. ¹⁰ At the end of the incubation period, a 10 µL of 10 mg/mL of DNase1 (Millipore) was added to the mixture to a final concentration of 0.2 mg/mL for additional 15 min at 37°C. Fetal bovine serum was then added to a final concentration of 10% and the cells were spun down at 18°C for 5 min at 1250 r/min. Pellets were suspended in complete growth medium (SMG-2 MV, Lonza) and grown in a six-well dish for 10–14 days, then harvested and grown in T75 flasks. We isolated vascular endothelial cells by using magnetic CD31 MicroBead kit following the company protocol (Miltenyi Biotec). The purity of population was determined by flow-cytometry using anti CD-31-PE antibodies (Miltenyi Biotec). We re-plated the purified endothelial cells and harvested the cells upon confluence for subsequent studies. The cells were used at passages three to six in all studies.

Table 1.

Demographic characteristics of the study participants.

	Controls			Patients
	ID	Age (years)	Sex	ID	Age (years)	Sex	Class	Duration	mRSS	Manifestations	Serologies
HumanMethylation450	C1	57	F	P1	56	F	dcSSc	5	37	ILD, PAH, joint, GERD	SCL70
	C2	59	F	P2	61	F	dcSSc	5	17	Calcinosis	ANA
	C3	57	F	P3	62	F	dcSSc	6	14	SRC, joint, GERD, PAH, digital ulcers, and calcinosis	RNA polymerase III
MethylationEPIC microarray	C4	60	F	P4	58	F	dcSSc	7	15	Joint, GERD	ANA
	C5	52	F	P5	52	F	dcSSc	1	21	Joint, GERD	RNA polymerase III
	C6	51	F	P6	49	F	dcSSc	10	26	Digital ulcers	SCL70
	C7	47	F	P7	49	F	dcSSc	7	14	GERD, ILD	SCL70
	Average	54.7			55.3			5.9	20.6
	p-value	0.8

mRSS: modified Rodnan skin score; dcSSc: diffuse cutaneous systemic sclerosis; ILD: interstitial lung disease; PAH: pulmonary arterial hypertension; GERD: gastroesophageal reflux disease; ANA: antinuclear antibody; SRC: scleroderma renal crisis.

Mean age is 49.3 years for patients with dSSc and 49.1 years for their controls.

p-value = 0.99 and 0.88, respectively.

Global DNA methylation

DNA from control and SSc-MVEC was isolated using QIAGEN kit and the concentration was determined by using Nano drop machine. For global methylation, we used 50 ng from each DNA sample and we followed the company protocols for MethylFlash^TM Global DNA Methylation assay kit (E_TEPIGENTEK).

Genome-wide DNA methylation studies

We performed genome-wide DNA methylation study in MVEC derived from seven dcSSc patients compared to seven age-, sex-, and ethnicity-matched healthy control MVEC. Briefly, DNA was first purified using QIAGEN kit and the concentration was determined by using Nano drop machine; then we used bisulfite conversion, which resulted in conversion of unmethylated cytosines to uracil, whereas methylated cytosines do not get converted.

We divided samples from patients and controls into two groups of matched SSc subjects and controls. Cytosine methylation was quantified across the genome using Illumina HumanMethylation450 (three dcSSc-MVEC and three controls), which covers 99% of RefSeq genes, with an average of 17 CpG sites per gene across the promoter region, 5′ untranslated regions (5′-UTR), first exon, gene body, and 3′-UTR. It also covers 96% of CpG islands. Non-CpG-methylated sites recently identified in human stem cells are also covered as well as microRNA promoter regions. Also, we used second microarray to reproduce the results in an independent set of patient and controls to identify differentially methylated genes. Specifically, we used the most recent Illumina infinium MethylationEPIC (four dcSSc patients and four controls) microarray (Illumina, San Diego, California, USA), which provide more extensive coverage for around 850,000 CpG sites across the genome. The average level of DNA methylation (β) on each CpG site was compared between patients and controls. To identify differentially methylated CpG sites between cases and controls, we used three data filtering criteria: (1) CpG site with an average difference in DNA methylation level of at least 1.2-fold; (2) differential methylation score of ⩾ 22 (p ⩽ 0.01) after adjusting for multiple testing using a false discovery rate (FDR) of 5%, and (3) exclusion of CpG sites assessed by probes with a genetic variant located within 10 bp of the 3′ end of the probe, as described before. ⁹ We included common differentially methylated CpG sites that were identified by both microarrays in final analysis.

To systematically highlight the most over-represented biological terms, we performed gene ontology (GO) network and pathway analysis using the Database for Annotation, Visualization and Integrated Discovery (DAVID) V. ¹¹ In pathway and GO analysis, we used Expression Analysis Systematic Explorer (EASE) Score threshold of < 0.1 for the detection of gene enrichment analysis (EASE score represents a modified Fisher’s exact p-value, which is considered a measure to examine the significance of gene-term enrichment), in addition to fold enrichment of 1.5 and FDR for correction of multiple testing < 10%. ¹²

Gene expression analysis

Quantitative real-time polymerase chain reaction (qRT-PCR) was performed to assess the correlation between DNA methylation and gene expression levels in selected set of genes that were differentially methylated in SSc-MVEC. As described before, ¹³ we isolated RNA from SSc MVEC (n = 3) and controls (n = 3) using TRIzol reagent following the company protocol (Invitrogen, Carlsbad, CA). Extra purification step of RNA was performed using column clean-up (QIAGEN, Valencia, CA). First-strand complementary DNA (cDNA) from RNA was prepared using iScript cDNA synthesis kit following the Bio-Rad protocol (Bio-Rad, Hercules, CA), and cDNA was synthesized from 1 µg total RNA. Gene-specific primers for CTNNA1 (5′-TCCTACGTCGCCTCTACCAA and 3′-CTGGACTGGTTCAGTCTGCC), IL17RA (5′-TACATCCGCTCCAGGGAGAA and 3′-AGGCAAGGTCTGAGAGTGGA), IL17REL (5′-CATCAGGACAGGACTCTGCC and 3′-GACATTCCAGAGACCCCTGC), DNMT3A (5′-CCCAGGCAGCCATTAAGGAA and 3′-TCCACCTGAATGCCCAAGTC), PDGFA (5′-CAGGACAGTGCGACGGTATT and 3′-GGCACACCAACAACACAGAC), ANGPT-2 (5′-TGCACAATGGTCTCACGTTC and 3′-ACTTGCAGGTGCTATGGTCT), DNMT3B (5′-TGCACTACACAGACGTGTCC and 3′-CCCACTGAGCAAGAGGGAAC), EDNRA (5′-CCACAACACAGACCGGAGCA and 3′-TCATTCTCCCCCAGCTCCCA), NOS1 (5′-TGCCGCCACTGTTTTCATTG and 3′-TGAGAGCTCAGAGGTAGGGG), ICAM-1 (5′-ACAACAGGCCCAAAAAGGGA and 3′-ATCAGATGCGTGGCCTAGTG), FN1 (5′-ACCGTGGGCAACTCTGTCAA and 3′-AGGAGCAAATGGCACCGAGA), and GAPDH (5′-CCCTGCCTCTACTGGCGCTG and 3′-GGCCATGAGGTCCACCACCC) genes for qRT-PCR were designed using Primer Blast software (NCBI) and were synthesized commercially (IDT, Coralville, Iowa). All cDNA samples were processed in parallel, and cycle threshold (Ct) values obtained from three replicate runs using power SYBR Green PCR master mix and Applied Biosystems machine. Briefly, the amplification conditions were: cycle 1, 95 °C for 3 min; cycle 2 (40 cycles), step 1 denaturation at 95 °C for 30 s, step 2 annealing at 58 °C for 1 min, step 3 elongation at 72 °C for 1 min; cycle 3, final elongation at 72 °C for 5 min; and cycle 4, and hold at 4 °C. PCR product specificity was evaluated by analysis of melt curves and agarose gel electrophoresis.

Protein extraction and Western blots

Cytoplasmic and nuclear proteins were extracted as previously described ¹⁰ and protein concentration was determined by Bradford procedure using Bio-Rad reagent (Bio-Rad, Hercules, CA). Approximately 20 µg of total cytoplasmic and/or nuclear proteins were loaded in each lane of sodium dodecyl sulfate (SDS) gels. Proteins were transferred to polyvinylidene difluoride (PDV) membrane and probed with various antibodies to test protein expression.

Matrix metallopeptidase 9 protein activity and Zymography gels

SSc-MVEC and control-MVEC were grown until they reached 80% to 90% confluence in normal medium (EGM-2 MV, Lonza); then cells were incubated with serum free EGM-2 MV medium for another 48 h. We collected the supernatant media and concentrated it with Amicon Ultra centrifugal filter unite (Millipore) with 10 kDa molecular weight cut. Protein concentration was determined for the concentrated media and 5 µg of medium secreted protein was applied to 8% zymography gel using gelatin type B as substrate following Abcam company gelatin zymography protocol.

MVEC tube formation for angiogenesis

SSc-MVEC and control-MVEC were seeded with population of 1 × 105 in 48-well plates, coated prior to seeding with 150 µL of 8.8 mg/mL of Matrigel^® Matrix (Millipore). Tube formation was observed by microscopy and pictures were taken once the tube was formed (approximately after 4–6 h). The cells were treated with phalloidin to observe the actin cytoskeleton organization.

Results

Global hypomethylation of SSc-MVEC compared to controls

To evaluate differences in DNA methylation levels between MVEC from patients with SSc compared to controls, we evaluated Global DNA methylation levels between the two groups using MethylFlash^TM Global DNA Methylation assay kit. We demonstrate significant global hypomethylation of SSc-MVEC compared to control (p = 0.02) (Supplementary Figure 1). Although there was global hypomethylation of MVEC genome in SSc, there were more differentially hypermethylated sites in SSc-MVEC as discussed below.

Differentially methylated specific CpG sites and genes

We evaluated DNA methylation changes in MVEC in two independent sets of dSSc patients and age-, sex-, and ethnicity-matched controls. Of the 850,000 CpG sites that were evaluated at a genome-wide level, we identified a total of 71,353 differentially methylated CpG sites in SSc-MVEC compared with healthy controls using Infinium MethylationEPIC microarray in the first group (0.081% of representative probes), and 33,170 CpG sites were differentially methylated in the second group using HumanMethylation450 microarray, which covers 485,000 CpG sites of the genome (0.073% of representative probes; Table 2). The difference in DNA methylation changes between the two groups is partly related to the number of CpG sites represented on these two platforms as described above.

Table 2.

Differentially methylated genes and CpG sites.

		Increased methylation	Decreased methylation	Total
Group 1 (EPIC)	CpG sites	34,961	36,392	71,353
	Genes	10, 957	11,277	15,777
Group 2 (450)	CpG sites	20,568	12,602	33,170
	Genes	6844	4841	9594
Common between Group 1 + 2	CpG sites	1625	830	2455
	Genes	910	485	1301

Between the two groups of subjects, we identified differential methylation of 2455 CpG sites, representing 1301 genes that were common between the two groups. Most of the differentially methylated CpG sites were hypermethylated (1625 CpG), corresponding to 910 genes. We provided a complete list of common hypermethylated genes in the two groups in (Supplementary Table 1). There were 830 hypomethylated CpG sites, representing 485 genes, as demonstrated in (Supplementary Table 2). The average fold change and differential score for differentially methylated CpG sites in SSc and controls were 0, 41, −36.82 and 1, 71, 67.83 for hypomethylated and hypermethylated CpG sites, respectively (Table 3).

Table 3.

Average fold change and mean differential score of the differentially methylated CpG sites.

	Mean fold change	Mean Differential score
Hypomethylated CpG	0.41	−36.82
Hypermethylated CpG	1.71	67.83

Differential methylation of genes involved in pathogenesis of SSc

Next, we screened differentially methylated genes based on known effect in SSc. We identified hypermethylated genes (location designated in brackets) in dSSc MVEC include NOS1 (body of the gene), which encodes for nitric oxide synthase-1, DNMT3A (Transcription Start site (TSS) 1500), DNMT3B (TSS1500, TSS 1500, TSS 1500), HDAC4 (body, body, body), and ANGPT2 (5’UTR; 1stExon; 5’UTR; 1stExon; 5’UTR; 1stExon; body and body; body; body; body). We also identified hypomethylation of IL17RA (3’UTR), IL17REL (TSS200), CTNNA1 (body; TSS1500; body), ICAM2 (TSS1500; 5’UTR; 5’UTR; 5’UTR; 5’UTR; and TSS1500; 5’UTR; 5’UTR; 5’UTR; 5’UTR) and SDK1 (body) in dSSc-MVEC.

Gene ontology and pathway analysis

Gene ontology analysis of hypermethylated genes demonstrated enrichment of genes involved in homophilic cell adhesion via plasma membrane adhesion molecules (p = 2.10E⁻⁰⁷) and angiogenesis (p = 0.0006). Pathway analysis of hypomethylated genes includes genes involved in Wnt signaling pathway (p = 0.001), vascular smooth muscle contraction (p = 0.014), and adherens junctions (p = 0.013; Table 4).

Table 4.

Gene ontology and pathway analysis of differentially methylated genes.

Gene ontology of hypermethylated genes
GO term	p-value
Homophilic cell adhesion via plasma membrane adhesion molecules	2.10E−07
Osteoblast development	7.63E−04
Actin cytoskeleton organization	8.62E−04
Embryonic skeletal system morphogenesis	0.0017
Angiogenesis	0.0067
KEGG pathways of hypomethylated genes
Wnt signaling pathway	0.0012
Pathways in cancer	0.00431
Adherens junction	0.01180
cGMP–PKG signaling pathway	0.01386
Vascular smooth muscle contraction	0.01464

GO: gene ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes; cGMP: cyclic guanosine 3′,5′-monophosphate; PKG: protein kinase G.

Correlation between methylation level and gene expression of selected genes

We evaluated gene expression levels of selected genes based on their potential biological significance in pathogenesis of SSc, or potential rule in epigenetic modification of gene expression. We demonstrated significant inverse correlation between DNA methylation status and gene expression in the majority of genes evaluated. We used qRT-PCR to evaluate expression levels of selected differentially methylated genes in SSc-MVEC compared to controls. In the hypermethylated genes (Figure 1(a)), we examined the gene expression of ANGPT-2, PDGFA, DNMT3A, DNMT3B, EDNRA, and NOS1 and detected significantly decreased gene expression levels (p-values were 0.0075, 0.002, 0.011, 0.003, 0.019, and 0.0044, respectively). Furthermore, we examined the protein expression of a selected gene based on its potential biological significance, ANGPT-2, using Western blot analysis and we demonstrate significant decrease in ANGPT-2 protein expression (p = 0.014) in dSSc-MVEC compared to control cells (Figure 2). While in hypomethylated genes (Figure 1(b)), we examined the expression of CTNN1, IL17RA, IL17REL, ICAM1, and FN1 genes and we demonstrated significant increase in the gene expression of IL17RA, IL17REL, ICAM1, and FN1 (p-values 0.037, 0.042, 0.046, and 0.007, respectively). Although we did not find significant change in CTNN1 gene expression levels in SSc-MVEC compared to controls, we demonstrated significant increase of β-catenin protein in the nucleus of SSc-MVEC (p = 0.01; Figure 3). There are many factors that affect the translocation of β-catenin to the nucleus and among these factors is the increase in protease activity of matrix metallopeptidase 9 (MMP-9) protein. As shown in Figure 4, by using the zymography techniques and the gelatin type B as specific substrate for MMP-9 and MMP-2 proteins, we are able to demonstrate increased activity of MMP-9 in dSSc-MVEC in comparison with controls (p = 0.04).

Figure 1.

Correlation of gene expression and methylation status. To evaluate the functional significance of methylation status on gene expression, we performed RT-qPCR for (a) six selected hypermethylated and (b) five selected hypomethylated genes that were differentially methylated in SSC-MVEC. We used three samples of dcSSc MVEC and three controls for these studies. Overall, most of the hypomethylated genes were overexpressed, and most hypermethylated genes were under-expressed in SSC-MVEC. The columns in the graph show the average individual gene expression level in SSc-MVEC in reference to control-MVEC; error bars represent the standard deviation. T-test was used to compare the mean of individual gene expression level in each SSc subset, as well as all SSc samples.

*p-value ⩽ 0.05.

Figure 2.

We performed Western blot to evaluate expression of ANGPT-2 using three SSc-MVEC and healthy controls MVEC. There was significant decrease in ANGPT-2, consistent with the observed increased methylation of ANGPT-2.

Figure 3.

The graph shows nuclear β-Catenin expression in SSc-MVEC and control MVEC from three patients and healthy controls.

Figure 4.

Increased activity of MMP-9 in SSc-MVEC compared to control MVEC as demonstrated by Zymogram. The columns show the average zymographic activity of three SSc-MVEC and three control-MVEC (p = 0.04).

We identified several hypermethylated genes in dSSc-MVEC that are relevant to angiogenesis (specifically, we identified the following genes by bioinformatics assessment of gene ontology: ANGPT1, ANGPT2, ANXA3, ALOX12, YP1B1, CD34, EPHA1, FN1, LRP5, SRPK2, SOX18, GPI, ENDRA, HAND2, HSPG2, HDAC7, MMRN2, NOS1, NRXN3, NRCAM, PITX2, PARVA, PDE3B, PLCD1, PRKCA, PPP1R16B, RNF213, SEMA5A, SLC12A6, TCF21, TGFB2, TGFBR2, and UNC5B). We used tube formation technique to investigate the angiogenesis potential of MVEC. Tube formation is the first step in angiogenesis, and it is well established that there is defect in tube formation in SSc. We confirmed that dSSc-MVEC were defective in tube formation in comparison to control cells. These results demonstrated that there is a major epigenetic modification in both the phenotype and gene expression in dSSc-MVEC.

Discussion

This is the first unbiased genome-wide DNA methylation study in MVEC from patients with dcSSc compared to healthy controls. The important findings of this study reveal differential methylation of genes involved in pathogenesis of SSc, associated with differential expression of some of the genes that we examined. We show that several groups of genes are hypermethylated in dSSc-MVEC in comparison with controls. These hypermethylated groups of genes represent embryonic skeletal system morphogenesis (p = 0.001), cell adhesion via plasma membrane adhesion molecules (p = 2.10E⁰⁷), osteoblast development (p = 7.63E⁰⁴), actin cytoskeleton organization (p = 8.62E⁰⁴), and angiogenesis (p = 0.0067).

Angiogenesis is a well-programmed cascade of events and a multi-step process that requires involvement of several factors to stimulate and enhance the formation of new vessels. ¹⁴ Among these factors are ANGPT-2 and PDGFA proteins. In SSc, the angiogenesis is dysregulated and lacks appropriate vascular recovery. ANGPT-2 is a key regulator of angiogenesis that exerts context-dependent effects on endothelial cells. ¹⁵ ANGPT-2 binds the endothelial-specific receptor tyrosine kinase 2 (TIE2) ¹⁵ and acts as a negative regulator of ANGPT-1/TIE2 signaling during angiogenesis, thereby controlling the responsiveness of endothelial cells to exogenous cytokines. Nevertheless, ANGPT-2 has been shown to act as an anti-apoptotic protective factor for stressed endothelial cells. ¹⁶ ANGPT-2 ablation in 129/J background results in mice that are born normal but most (95–99%) die within 2 weeks with severe chylous ascites, peripheral lymphedema, and ANGPT-2^−/− mesenteric lymphatic vessels that have poor smooth muscle coverage. ¹⁷ ANGPT-2 is mainly secreted by endothelial cells at sites of active vascular remodeling and functions in an autocrine manner. ¹⁸ Here, we demonstrate hypermethylation and decreased expression of ANGPT2. This finding demonstrates a role of DNA methylation machinery in angiogenesis.

Platelet-derived growth factor A (PDGFA) is produced by endothelial cells ¹⁹ and it plays a significant role in blood vessel formation by recruiting pericytes, which stabilize the new blood vessels.^19,20 PDGFA plays an essential role in the regulation of embryonic development, cell proliferation, cell migration, survival, and chemotaxis. ²¹ Also, it is required for normal lung alveolar septum formation during embryogenesis and normal development of the gastrointestinal tract. ²² In this study, we demonstrate hypermethylation and decreased expression of PDGFA, which could be contributing to MVEC dysfunction in SSc.

We were also able to link the deficiency of the angiogenesis process in SSc disease to the impairment of angiogenesis factors, which are significantly affected by methylation machinery and subsequently gene expression and protein translation. We show that NOS1 and EDNRA genes were hypermethylated and their gene expression levels were significantly decreased in dSSc-MVEC. Nitric oxide synthases (NOSs) are a family of three isoforms of NOS-1, NOS-2, and NOS-3. ²³ NOSs catalyze the production of nitric oxide (NO) from L-arginine.^24,25 NO has a dual function and it can be beneficial (as a vasodilator)^26–28 and harmful (as a cytotoxic effector).^29,30 In this report, we demonstrate increased methylation levels of NOS1 gene associated with reciprocal decrease and the transcription of NOS1, which may contribute to deficient NO production by SSc-MVEC.

Other genes that were hypermethylated in SSc-MVEC include the Histone deacetylase 4 (HDAC4) and DNA methyltransferase enzymes (DNMT3A and DNMT3B). The acetylation and deacetylation of histones play an important role in the regulation of gene transcriptions.^31,32 This observation highlights an interplay between DNA methylation and histone code epigenetic machineries that might be contributing to MVEC dysfunction.

It was interesting to find that Guanylyl cyclase gene to be activated by hypomethylation in SSc-MVEC. Guanylyl cyclase catalyzes the conversion of guanosine 5′-triphosphate (GTP) to cyclic guanosine 3′,5′-monophosphate (cGMP). Once cGMP is produced, it can have a number of effects in cells, and many of these effects are mediated through the activation of protein kinase G (PKG). PKG has been shown to catalyze the phosphorylation of a number of physiologically relevant proteins. Previous studies have shown that there is a direct relation between cGMP and Wnt/β-catenin signaling pathway.^33,34 When the intracellular cGMP level increases, it causes inhibition of GSK-3. ³⁵ GSK3 has been implicated in the regulation of many biological processes in cells and, in particular, emphasis in its role in the phosphorylation and modulation of the transcriptional activity in the context of the “canonical” Wnt/β-catenin pathway. There is significant evidence for a crucial role of canonical Wnt/β-catenin signaling pathway in cell development, proliferation, differentiation, and migration. ³⁶ Wnt/β-catenin regulates the expression of several extracellular matrix (ECM) components including fibronectin ³⁷ and degrading MMPs such as MMP-3 and MMP-9.^38,39 The Wnt/β-catenin signaling pathway has been established as a critical pathway in the activation of fibroblasts in wound healing and maintenance of fibrogenesis in SSc.^40–42 Apart from the role of cGMP in the activation of Wnt/β-catenin pathway, we also demonstrated that Wnt gene in SSc-MVEC is activated by hypomethylation. There are three well-characterized Wnt signaling pathways when activated and they are the canonical Wnt pathway, the noncanonical planar cell polarity pathway, and the noncanonical Wnt/calcium pathway. All three pathways are activated by binding a Wnt-protein ligand to a Frizzled family receptor, which passes the biological signal to the disheveled protein inside the cell. β-catenin is the end-product of the canonical Wnt pathway. β-catenin is significantly activated as shown by its translocation to the nucleus in SSc-MVEC in comparison to control cells. Our observation supports a role of Wnt/β-catenin in SSc pathogenesis and provides evidence for epigenetic alterations of this pathway in SSc-MVEC.

Cell-to-cell junctions are important regulators of endothelial responses both in quiescent and angiogenic vessels. Endothelial cells express tight and adherens junctional structures, which are formed by transmembrane adhesive proteins that bind to identical proteins on an adjacent cell and start a sequence of signaling events. ⁴³ CTNNA1 encodes for an α-catenin, a protein involved in cell adhesion by anchoring the β-catenin-Ecadherin complex to the actin cytoskeleton. Loss of CTNNA1 gene causes global loss of cell adhesion in E-cadherin-expressing human breast carcinoma cells, ⁴⁴ which demonstrate the importance of α-catenin in maintaining the integrity of adherens junctions. In our study, we found hypomethylation of CTNNA1 gene but we did not find any changes in the gene expression level in comparison to control.

IL-17RE complex may play a role in maintaining mucosal barrier integrity. ⁴⁵ In our study, we demonstrate hypomethylation and increased expression of IL-17RA and IL-17RE in SSc-MVEC compared to controls, which may suggest a role of this inflammatory pathway in activation of SSc-MVEC, which is providing important new direction toward understanding the pathogenesis of SSc vascular disease. Endothelial cell injury appears to be an initial event in the formation of vasculopathy. ⁴⁶ Immune factors mediate endothelial cell inflammation, which results in leukocyte recruitment, adherence to injured endothelial cells, and to leukocyte production of inflammatory mediators, such as IL-17A, to exacerbate the inflammatory reaction and aggravate blood vessel destruction and tissue hypoxia. IL-17A has been reported to mediate the recruitment, activation, and migration of neutrophils,^47,48 and induces the expression of adhesion molecules and chemokines.

In this report, we demonstrated hypomethylation and enhanced gene expression of the cell adhesion molecules in SSc-MVEC. Cell adhesion molecules are cell-surface proteins that account for cell-to-cell and/or cell-to- ECM interactions. Cell adhesion molecules are shown to play critical roles in the development and maintenance of cell adhesive mechanisms between many different types of cells. Several studies have compared the expression levels of adhesion molecules in SSc patients and healthy controls in serum^49–51 and in tissues.^50,52,53 Increased expression of the adhesion molecules in SSc may be an important factor for the process of immunologically induced fibrosis in the disease. We demonstrated that ICAM1 gene is hypomethylated in association with increased gene expression level in the SSc-MVEC. ICAM1 is an endothelial- and leukocyte-associated transmembrane protein and long known for its importance in stabilizing cell–cell interactions and facilitating leukocyte endothelial transmigration. More recently, ICAM-1 has been characterized as a site for the cellular entry of human rhinovirus. ⁵⁴ Because of these associations with immune responses, it has been hypothesized that ICAM-1 could function in signal transduction. ICAM-1 ligation produces pro-inflammatory effects such as inflammatory leukocyte recruitment by signaling through cascades involving a number of kinases, including kinase p56lyn, and in the regulation of vascular permeability.^55,56 A previous study demonstrated that there is a direct relation between IL-17 and ICAM1 expression. ⁵⁷ Gene ontology analysis of hypermethylated genes demonstrated enrichment of genes involved in homophilic cell adhesion via plasma membrane adhesion molecules and angiogenesis. Pathway analysis of hypomethylated genes includes genes involved in Wnt signaling pathway, vascular smooth muscle contraction, and adherens junctions (Table 1).

Our study certainly has limitations that include relatively small sample size that may lead to false discovery. To minimize that effect, we performed more stringent analysis where each differentially methylated site has to be reproduced in second cohort to qualify for further analysis. Further studies using larger sample size will be useful to confirm these findings. We used two platforms to evaluate DNA methylation, Illumina Infinium MethylationEPIC microarray and HumanMethylation450K microarray. To avoid patch effect, we have maintained age and gender match when we placed the samples on either platform, as SSc patient and control samples were paired on each chip. We performed limited gene expression analysis on genes of interest, which was mostly supportive of an inverse relationship between methylation and gene expression in these genes. Finally, our cohort represents largely patients with relatively late dcSSc, as the majority of patients are 5–10 years after the onset of dcSSc. The effect of disease duration on DNA methylation aberrancies is not established in SSc. Longitudinal studies from the same subjects and controls are warranted in the future to understand changes in the epigenome.

Conclusion

This is, to our knowledge, the first unbiased study to determine the methylation landscape of SSc-MVEC at the genome-wide level. This study highlights significant aberrancies in DNA methylation patterns between SSc-MVEC and controls, with aberrancies involving methylation patterns of specific genes and pathways that may be contributing to the abnormal phenotype of SSc-MVEC. The data we present in these studies highlight an inverse relationship between DNA methylation levels and gene expression level in the majority of the genes we selected for further evaluation. This suggests a functional role of DNA methylation in gene expression. Among the hypermethylated genes that were involved are genes related to angiogenesis. Our study emphasizes the potential role of DNA methylation changes in the pathogenesis of SSc.