Aberrant DNA methylation and expression of ERAP1 gene in ankylosing spondylitis

,


Summary
Objective : Endoplasmic reticulum aminopeptidase 1 (ERAP1) is known to participate in the pathogenesis of ankylosing spondylitis (AS) cooperated with HLA-B27.This study aimed to evaluate the relationship between promoter methylation and mRNA levels of ERAP1 and AS.
Methods : The DNA methylation level of 100 AS patients and 100 health controls (HCs) were tested using targeted bisulfite sequencing assay.Besides, the mRNA level of 20 AS patients and HCs was measured used quantitative real-time reverse transcription-polymerase chain reaction to verify the results of DNA methylation.
Results : The methylation levels of two CpG islands containing 31 loci in ERAP1 promoter were measured.ERAP1 1 (P < 0.001) and ERAP1 2 (P < 0.001) islands were significantly hyperrmethylated in AS patients compared with healthy controls.Correspondingly, the mRNA level was significantly lower in AS patients.The ROC curve analysis reported the sensitivity, specificity and area under curve were 0.717, 0.737 and 0.779 of differential methylated CpG loci of ERAP1 for AS diagnosis.Besides, we also found that the methylation level was associated with the family history, non-steroidal anti-inflammatory drugs use, X-ray classification and clinical manifestations.

Introduction
Ankylosing spondylitis (AS) is a common chronic rheumatic arthritis, characterized with progressive bone proliferation of axial skeleton and sacroiliac joints.The exactly pathogenesis of AS is still obscure now, but studies indicated that gene and environmental interaction plays roles in the development and progression of AS (1,2).Twins and family based studies estimated the heritability of AS is about 90% (2)(3)(4).The human leukocyte antigen (HLA)-B27, which is encoded in class 1 major histocompatibility complex region, is the strongest risk factor of AS (4).Recent genome-wide association and case-control studies indicated that endoplasmic reticulum aminopeptidase 1 (ERAP1 ) is significant associated with AS through cooperating with HLA-B27 (5).The total of 114 loci are established to associate with AS; however these genetic variants are reported to only accounting for about 30% of genetic risk (6).Recent studies indicated that epigenetics may partly account for the inter-individual variance of heterogeneity.
DNA methylation as a most common reported epigenetic modification plays pivotal roles in various life courses, as growth and differentiation, through programmed gene expression regulation in the genome.DNA methylation is the addition of a methyl group to 5' position of a cytosine DNA base in the middle of cytosine-guanine dinucleotide (CpG) (7).The abnormal DNA methylation in the gene promoter is generally associated with transcriptional silencing and linked to ranges of diseases (8,9).Recently, increasing number of epigenome-wide association studies (EWAS) indicated that DNA methylation plays pivotal roles in the mechanism of rheumatic diseases as systematic lupus erythematosus, rheumatoid arthritis, and AS (10)(11)(12)(13)(14).One EWAS found 1915 altered DNA methylation loci of AS.Besides, candidate targeted gene methylation studies also reported differential methylation loci of AS patients.Methylation of SOCS-1 gene was detected in serum of HLA-B27 positive AS patients but not B27 positive controls, and significantly associated with higher serum cytokines and severity of clinical manifestations of AS patients (15).Hypermethylation and decreased expression of DNMT1 and BCL11 B genes were both reported to associate with AS (16,17).Nevertheless, study focus on DNA methylation and AS is still scarce and urgent.
ERAP1 is a polymorphic aminopeptidase within the endoplasmic reticulum, known as "molecular ruler" to trim peptides to nine amino acids in length for binding to HLA class I molecules on antigen-presenting cells for subsequent interaction with CD8 + T cells (18,19).Recent single nucleotide polymorphism studies of our team and other scholars also proved that ERAP1 plays pivotal roles in the pathogenesis of AS through cooperating with HLA-B27 (5,14).Taken into account of these factors, we designed a two stage casecontrolled study to evaluate the promoter methylation and transcriptional profile ofERAP1 gene, respectively, in peripheral blood mononuclear cells of AS patients and healthy controls (HCs).

Study populations
A total of 100 AS patients and 100 age and gender matched HCs were enrolled in the DNA methylation examination stage, and 20 patients and 20 controls were recruited in the mRNA expression verification stage.All patients and HCs were recruited from the Department of Rheumatology at the First Affliation Hospital of Anhui Medical University.Diagnosis of AS was made by qualified rheumatologists according to the modified New York criteria (20).Blood donors with no history of rheumatic diseases or other chronic diseases were included as controls.DNA and mRNA samples were extracted from the 5 ml peripheral venous blood of all participants.Besides, all patients have filled out a questionnaire about the general geographic and clinical characteristics.Detailed clinical indicators as medication use, HLA-B27, C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) and Bath Ankylosing Spondylitis Functional Index (BASFI) were reported.This study was approved by the ethics committee of Anhui Medical University with the serial number of 20170225.All patients and controls provided their written informed content.

Targeted bisulfite sequencing assay
CpG islands in the proximal promoter of ERAP1 gene were examined from 2k bp upstream of transcript start site to 1k bp downstream of the first exon satisfying the following criteria: a) observed to expected ratio of CpG dinucleotide > 0.60; b) percentage of cytosine and guanine > 50%; c) length >200bp.DNA sequences of the CpG islands in ERAP1 promoter region were determined by an online database of the University of California, Santa Cruz (http://www.genome.ucsc.edu),and the primers sequences for ERAP1 methylation were designed by the EpiDesigner online software (http:// www.epidesigner.com)accordingly.And 31 methylation sites of two CpG islands (ERAP1 1 and ERAP1 2) were analyzed in our study.The primers sequence of ERAP1 1 and ERAP1 2 methylation were listed as following: ERAP1 1, forward: 5'-AGGGTTAGGGGTATGTAGGAAAG-3', reverse: 5'-CCTTCCTCCTCTACAACATCTCC-3'; ERAP1 2, forward: 5'-GTTTTGGGGTYGTTTTTATTTTTG-3', reverse: 5'-TTACCCTTTCCCCAACTCC -3'.
Genomic DNA was firstly extracted from peripheral venous blood of participants using DNeasy Blood Tissue Kit QIAGEN Kit (QIAGEN, Germany) in line with manufacturer's protocol, and quantified and qualified through NanoDrop TM 2000 Spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA.).Genomic DNA (400 ng) was bisulfite converted using EZ DNA Methylatio TM -GOLD Kit (Zymo Research, Irvine, USA).Multiplex polymerase chain reaction (PCR) was performed with above primers combination, and PCR amplicons were separated and purified through agarose electrophoresis and QIAquick Gel Extraction kit (QIAGEN, Germany).Corresponding libraries were sequenced on Illumina NextSeq platform according to manufacturer's protocols.

Quantitative real-time PCR assay
Peripheral blood mononuclear cells were isolated from peripheral blood using density gradient centrifugation.Then, total RNA were extracted and purified from peripheral blood mononuclear cells by TRIzol TM LS reagent.Total RNA was reverse translated as complementary DNA after the degradation of mixed genomic DNA using PrimeScript TM RT reagent Kit with gDNA Eraser (Takara Bio Inc., Japan).SYBR Green kit (Takara Bio Inc., Japan) was used in quantitative real-time PCR process based on QuantStudio TM 7 Flex (Life Technologies, USA).The expression data of ERAP1 was normalized to internal reference β-actin, and the relative expression level of ERAP1 was calculated by comparative 2 -ΔΔ῝τ method.Primers sequence of ERAP1 and β-actin were listed as following: β-actin, forward: 5'-CTCCATCCTGGCCTCGCTGT-, reverse: 5'-GCTGTCACCTTCACCGTTCC-; ERAP1, forward: 5'-TTTGAACTTGGCTCATCTTCC-, reverse: 5'-AATTGTCTGTTGGACACAACG .

Statistical analysis
The distribution of the included variables of our study was tested via Shapiro-Wilk method.Normal distribution data was presented as mean ± standard variance, otherwise median and (interquartile range) was presented.Correspondingly, student's t test or Mann-Whitney U test were used to test the between group difference of the mean or mean rank, respectively.Spearman's rank correlation efficiency was test to evaluate the association between ERAP1 methylation level and other continuous or hierarchical variables.Univariate logistic regression analysis was used to calculate the odds ratio (OR).Odds ratio (OR) with 95% confidence interval (CI) and forest plot were used to evaluated association of individual methylation site and AS visualized.Multiple logistic regression analysis was used to establish a regression model.Receiver operating characteristic (ROC) curve and area under curve (AUC) were applied to assess the predictive value of the ERAP1 methylation as a biomarker of AS.All these data analysis diagram plotting were accomplished by SPSS 23.0 software (SPSS Inc., Chicago, IL, USA), GraphPad Prism 7.00 (GraphPad Software Inc., CA, USA) and R software.A two-tailed P < 0.05 was taken as statistically significant, and multiple comparisons were adjusted through Bonferroni method.

Characteristics of participants
In the candidate targeted gene methylation detection stage, of the enrolled 100 patients and 100 HCs, one patient and one HC failed in bisulfite conversion assay.Detailed demographic and clinical characteristics were listed in Table 1.There were no significant difference in gender (χ 2 = 1.172,P = 0.279) and mean age (t = 0.564, P = 0.574) existed between patients and HCs.In the second stage, 20 patients and 20 HCs were tested the mRNA level ofERAP1 gene.Similarly, no any significant difference in mean age (t = 0.102, P = 0.919) and gender (χ 2 = 0.125,P = 0.723) were found.Data medication use and unilateral X-ray classification were only retrieved from a proportion of AS patients.

DNA methylation of ERAP1 gene
In total, two CpG islands containing 31 loci were tested the proportion of methylated cytokine accounting for the total cytokine.Among the 15 loci of ERAP1  1A.Among them, ERAP1 2 28 located at chr5: 96143595 was reported to have highest effect size with OR = 5.390 (95% CI: 2.048 to 14.184) of as a percent of total cytokine methylated.The average methylation levels of ERAP1 1 (Z = -4.831,P < 0.001) and ERAP1 2 (Z = -6.140,P< 0.001) islands were also calculated to associated with AS (Figure 1B & 1C).ROC curve analysis on the average methylation levels of the two CpG islands also indicated their potential as biomarkers of AS.The AUC of ROC curve for ERAP1 1 island was 0.669 (95% CI: 0.626 to 0.707, P < 0.001), with a sensitivity of 0.818 and a specificity of 0.531 for the best cutoff point 0.671 (Figure 1D, Table 3).Correspondingly, the AUC of ERAP1 2 was 0.750 (95% CI: 0.683 to 0.818,P < 0.001), and the best cutoff point was 0.681, with a sensitivity of 0.707 and specificity of 0.684 (Figure 1D, Table 3).We also established a regression model included the above mentioned 13 significant CpG loci detailed in Table S1, with each sample obtained a probability value.ROC curve analysis on the probability value also reported that the AUC was 0.779 (95% CI: 0.714 to 0.844, P < 0.001), and best cutoff point was 0.500, with a sensitivity of 0.717 and specificity of 0.737 (Figure 1D, Table 3).

Subgroup and correlation analyses and of ERAP1 methylation
ERAP1 was known to play roles cooperated with HLA-B27 in AS.Nevertheless, we found that the methylation levels of ERAP1 1 and ERAP1 2 islands of HLA-B27 positive patients were not significantly different from HLA-B27 negative patients (Z = -0.180,P = 0.857).The known environmental factors associated with DNA methylation of smoking (Z = -0.268,P = 0.789) and alcohol use (Z = -0.969,P = 0.332) were also reported not associated with the methylation level in AS patients detailed in Table 4.Among the demographic and clinical factors, we found that patients with family history have lower methylation level of ERAP1 1 island (Z = -2.258,P = 0.024).Non-steroidal anti-inflammatory drugs (NSAIDs) use was significantly associated with higher methylation level of ERA1 2 island (Z = -2.113,P = 0.035), but not any medication use history (Z = -0.605,P = 0.545).As to the variates about disease activity and function, we also found that the methylation level of ERAP1 1 island was significant associated with global back pain (r s = 0.502, P = 0.002) of AS patiens.X-ray classification of sacroiliac joint (r s = 0.548, P = 0.018) was associated with the methylation level of ERAP1 2 island (Table 5).

Expression level and correlation analysis of ERAP1 mRNA
In order to verify the differential methylation level of ERAP1 , we also measured the mRNA expression level in 20 AS patients and 20 HCs.The expression level of ERAP1 was significant decreased in AS patients compared with HCs ((Z = -4.048,P < 0.001, Figure 1e).In the correlation analysis, we also found that body mass index (r s = -0.659,P = 0.002), chest expansion (r s = 0.697, P = 0.001) and Schober test score (r s = 0.537, P = 0.018) were associated with ERAP1 level.

Discussion
Our study first proved that the relationship betweenERAP1 and AS from the epigenetic aspect.The result proved that hypermethylation ofERAP1 promoter in the peripheral blood was associated with AS.The function and disease activity index and NSAIDs were also significantly associated with the methylation level.Correspondingly, the mRNA expression level was also proved to significantly decrease in AS patients.
Unlike the DNA text, the sequence of nucleotides, containing the genetic information, the annotation system of chemical modification was exited for instructing how and when to read the text in mammals.DNA methylation plays sophisticated roles in annotating genetic information.The existence of DNA methylation in gene regulatory regions, as promoter and enhancer, would recruit a group of factors programmed generating a closed chromatin structure and consequently repressing the gene expression.Our study reported that the DNA methylation of ERAP1 was associated with AS, consistent with results of previous studies thatEARP1 plays pivotal roles in the pathogenesis of AS.In the verification stage, we also found that the mRNA expression level of AS patients were significantly deceased, which is consistent with the result that ERAP1 gene was hypermethylated in AS patients.
Different form the sequence of DNA, previous study has proved that DNA methylation patterns derived from the gametes will erased before embryo implantation and the new methylation profile will established in each mammal (21,22).Intriguingly, we found something unusual that ERAP1 1 in AS patients with family history were hypomethylated.One reason may be that patients with family history inherited higher risk of AS, and the pathogenesis of AS propositus was more like to the results of exposure of environmental risk factors, which made them have higher methylation level than patients with higher inherited risk of AS.Even though ERAP1 was established to cooperate with HLA-B27 in AS, the results indicated that HLA-B27 status may not associated with the methylation level of ERAP1 in AS patients.Consistently, previous studies also indicated that family history was associated with methylation level of diseases related genes (23,24).Our study also suggested that NSAIDs was associated with the methylation of ERAP1 1, and the tumor necrosis factor inhibitor and sulfasalazine were irrelevant with the methylation level.Similarly, previous study also proved that celecoxib could reverse the DNA hypomethylation status in rat colon tumors.This study also indicated that preventive efficacy of various agents may be the results of their effect on reverse DNA hypomethylation in some extent (25).Our results also indicated that the efficacy of NSAIDs on repression the progression of ossification may be the results of epigenetic regulation of corresponding genes.However, the nature of cross-sectional design our study make us only can provide association result.The result should be verified by prospective study, and further study about the underlying mechanism was also helpful.
In the correlation analysis, we also found that the methylation level of ERAP1 2 was positively associated with the X-ray classification of sacroiliac joint.Besides, we also found that the mRNA level ofERAP1 was positively associated the thoracic and lumbar mobility.It seemed that ERAP1 was more associated with the long-standing joint ossification.And the higher methylation and corresponding lower mRNA levels of ERAP1 was associated with severer ossification.These results were consistent with the conclusion that ERAP1 were associated with AS.Besides, we also found something interesting that body mass index was positively associated with the mRNA level ofERAP1 .Relevant study was still devoid, and the relationship should be verified in further with larger sample size.The ROC curve analysis indicated that the ERAP1 methylation level could serve as biomarkers, by either CpG island or entire promoter, to distinguish AS patients from HCs.
Our study has several strengths.To best of our known, this study was the first research reported the relationship between ERAP1 methylation status and AS and verified the mRNA expression level ofERAP1 .Besides, we have also evaluated the relationship between various environmental factors, clinical manifestations or medication use and the methylation level.Considering the less invasiveness and higher diagnostic efficiency, methylation test was valuable in clinical setting for AS diagnosis.Recent study indicated that drugs as celecoxib was could prevent diseases as tumor (25), which suggested the treatment potential of agents specially regulated the methylation factors ofERAP1 .The finding that NSAIDs use was positive associated the methylation level of ERAP1 1 also indicated the treatment efficacy of NSAIDs on AS was via altering the methylation level of corresponding genes in some extent.
These are also some limitation should be considered.First, subjects for the testing of DNA methylation and mRNA were two separate groups, so the relationship analysis between DNA methylation and mRNA was not applicable.Second, the methylation profiles were different across tissues and cell subtypes.ERAP1 methylation level of peripheral blood mononuclear cells can not present the methylation of specific cell.In the end, the single center case-control study design limited the generalizability of results, and further lager scale prospective study and animal model research are necessary.

Conclusions
In summary, our study demonstrated that the ERAP1 is significantly hypermethylated in AS patients, and the result is also verified by the lower mRNA level of AS patients.Our findings suggested that aberrant methylation of ERAP1 promoter may take part in the pathogenesis of AS and can be considered as diagnostic tool and therapeutic target of AS.The finding that NSAIDs use could alter the methylation level indicated that the efficacy of NSAIDs on AS could be through altering the methylation level of corresponding genes in some extent.

Figure 1 .
Figure 1.Results of the association between methylation and mRNA levels of ERAP1 and AS.A) The odds ratio (ORs) and 95% confidence intervals (CIs) of the methylation 31 CpG loci in ERAP1 1 and ERAP1 -2 islands, and six loci in ERAP1 2 island were significantly associated with increased risk of AS.B) The methylation level of ERAP1 1 island was significantly increased in AS patients.C) The methylation level of ERAP1 2 island was significantly increased in AS patients.D) The ROC curves of the methylation levels of ERAP1 1 island, ERAP1 2 island and the combination of all the significantly differential methylated CpG loci.E) The ERAP1 relative expression level was significantly decreased in AS patiens.

Table 2
Detailed methylation status of the 31 CpG sites of AS patients and HCs