Epigenetics and methylation in the rheumatic diseases

Seminars in Arthritis and Rheumatism ] (2013) ]]]–]]]

Contents lists available at ScienceDirect

Seminars in Arthritis and Rheumatism journal homepage: www.elsevier.com/locate/semarthrit

Epigenetics and methylation in the rheumatic diseases Flore Zufferey, MD, Frances M.K. Williams, BSc, MBBS, PHD, FRCP(E), Tim D. Spector, MBBS, MSc, FRCP, MDn Department of Twin Research & Genetic Epidemiology, King's College London, London, United Kingdom

a r t i c l e in fo

Keywords: Epigenetics Methylation EWAS Autoimmune diseases Rheumatic diseases

a b s t r a c t Objectives: Rheumatic diseases encompass a wide range of conditions of poorly characterized etiopathology, many having both genetic and environmental susceptibility factors. Epigenetic studies are providing new insights into disease pathogenesis. Recent rheumatology literature related to DNA methylation studies—both epigenome-wide and candidate gene—are discussed, as well as methodological issues. Method: A PubMed search for articles published until April 2013 was conducted using the following keywords: (“methylation” OR “epigenetics”) AND (“rheumatoid arthritis” OR “lupus” OR “autoimmune disease” OR “osteoporosis” OR “osteoarthritis” OR “musculoskeletal disorder”) and EWAS. The reference lists of identified articles were searched for further articles. Results: Several genome-wide methylation studies have been reported recently, mostly in autoimmune rheumatic diseases. Overall, these studies have identified methylation signatures in disease, clustering of subgroups as well as new and known epigenetic associations. Methodological issues, small sample sizes and reduced coverage of methylation assays render many results preliminary. Conclusions: There have been a number of epigenetic advances in rheumatic diseases recently. The new technologies and emerging field of epigenome-wide association will provide novel perspectives in disease etiology, diagnosis, classification, and therapy. & 2013 Elsevier Inc. All rights reserved.

Introduction The field of epigenetics in medicine has grown exponentially over the last few years, benefiting from the advanced application of high-throughput technologies. Epigenetics has however been harder to define with any consensus [1,2], not least because of the wide variety of regulatory mechanisms including histone modifications, chromatin folding, DNA methylation (DNAm), and miRNA expression (micro-RNAs). Among these, DNA methylation is currently the most well studied because of the availability of suitable technology. DNA methylation forms the subject of this review—for other epigenetic mechanisms in the field, see reviews [3–6]. Rheumatic disease encompasses a wide range of different conditions, from autoimmune, inflammatory diseases to degenerative and mechanical conditions. Disease classification relies mainly on the evolving clinical, radiologic, and biological composite criteria because the etiopathology of few such diseases has been elucidated fully. Although twin and family studies have established genetic risk factors for most, if not all the conditions,

Dr. Zufferey is recipient of a fellowship from the Swiss Society of Rheumatology and SICPA foundation, Switzerland. Tim Spector is recipient of an ERC Senior Researcher award. n Corresponding author. E-mail address: [email protected] (Tim D. Spector). 0049-0172/$ - see front matter & 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.semarthrit.2013.10.003

genetic variants identified so far explain a small proportion of the total phenotypic variance. Epigenetics has great promise, therefore, in explaining and redefining diseases in the field as well as accounting for the role of some of the environmental factors. In this review, we aim to provide clinicians with an overview of the current methodological concepts, challenges, and issues in methylation analysis. We present results of methylation studies in the context of rheumatic diseases and discuss future perspectives.

Methylation architecture and distribution The genome sequence is composed of four nucleotides—A (adenosine), T (thymine), C (cytosine), and G (guanine). DNA methylation (DNAm) refers to the addition of a methyl group to cytosine at the carbon 5 position (5-methylcytosine, 5mC). Other types of methylation marks exist such as hydoxymethylcytosine (5hmC), which has been identified as a modification of developmental importance at least in early life [7]. The global distribution of cytosine modifications across the genome is referred to as the DNA methylome. In humans, methylation occurs mainly when cytosine is followed by a guanine on the same strand of DNA linked by a phosphate group (called CpGs or CpG dinucleotides), although several other types of methylation, e.g., at CpHpG (H ¼ A, T, or C),

2

F. Zufferey et al. / Seminars in Arthritis and Rheumatism ] (2013) ]]]–]]]

exist as well [8]. Overall distribution of methylation in the genome is bimodal; the majority of CpGs across the genome are methylated (70–80%), whereas a high density of CpGs enriched in the promoter region of the genes (CpGs islands) are generally unmethylated [9]. It is therefore challenging to summarize these changes over the genome.

Methylation properties and functions in human biology and disease Methylation plays a key role in physiologic conditions, and alterations in methylation regulation have been identified in pathological processes [10,11]. Current knowledge indicates that the function of methylation and its relationship to gene expression varies according to the position in the genome as well as in regions with poor or rich CpG content [12,13]. High levels of 5mC in CpGrich promoter regions are usually strongly associated with transcriptional repression [14], whereas gene body methylation may not be and shows a more context-dependant relationship [15]. DMAm does not act alone on gene regulation but is closely interconnected with other genetic and epigenetic mechanisms (e. g., transcription factors and histone-modifying enzymes), acting together on chromatin structure to determine the state of gene accessibility. Methylation is thought to reinforce the maintenance of closed chromatin through subsequent cell divisions and mediate long-lasting changes in gene expression [13]. The aforementioned regulatory mechanisms rely on two central properties of DNAm: stability (mitotic inheritance) and plasticity. Plasticity refers to the constant adaptation of the methylome to specific cell regulatory processes in response to internal and external stimuli. Physiological variations in the methylome Gametogenesis and pre-implantation zygotic phases are both associated with extensive methylation resetting and reprogramming [16]. Later, in post-zygotic somatic development, most blocks of genomic methylation patterns are static across tissues and throughout life, changing only in local context as specific cellular processes are activated or repressed [13]. These include, for example, mechanisms sustaining cellular fate and differentiation, responsible for specific methylome signatures across different cell types (cell-specific methylome) [17]. Aging—which plays a role in many rheumatic diseases—seems also to be associated with global reproducible methylation changes with a tendency towards hypermethylation [18]. Epigenetic epidemiology and human diseases Parental contributions of imprinted genes are currently the clearest examples of how human disease may result from epigenetic deregulation [19]. Examples of monogenic epigenetic diseases are congenital syndromes such as Beckwith–Wiedemann or Prader–Willi/Angelman syndromes. These are recognizable syndromes associated with congenital abnormalities and/or developmental delay. These syndromes are due to the consequence of the loss of normal imprinted gene regulation in specific subdomains due to different underlying genetic mechanisms (microdeletion and uniparental disomy) [19]. In common diseases, it has been suggested more recently that each individual has many “epigenomes” and that these may play a central role in disease pathogenenesis [19,20]. Although there is some evidence of epigenetic change influenced by the environment [21,22], such as smoking in humans [23], it remains largely unknown as to precisely how the environment triggers alterations in the epigenome leading to disease susceptibility [24]. There is

Table 1 Disease concordance rate within MZ twin pairs and heritability estimates—the proportion of phenotypic variance attributed to additive genetic factors—in several rheumatic diseases based on twin studies Disease/trait

MZ concordance rate (%)

Heritability estimates (%)

References

RA SLE OA Osteoporotic fractures Hyperuricemia Gout Psoriasis Psoriatic arthritis Ankylosing spondylitis

12–15 24–56 31–50 13–19

65 69 E50 27

[31] [76,77] [78–80] [81]

53 42 4–32 10 40–75

49 0a 66 – 94

[74] [74] [82] [83] [84,85]

RA ¼ rheumatoid arthritis; SLE ¼ systemic lupus erythematosus; OA ¼ osteoarthritis. a

95% Confidence interval ¼ 0–61.8%.

also accumulating evidence for genetic influence on DNAm [25–30], which may mediate disease susceptibility. Disease modeling using concordance rate between twin pairs or family members helps tease apart the respective contributions of genetics and environment in disease variance. Heritability is an estimate of how much of the variance of a trait may be attributed to genetic factors. Despite known heritable influence in the majority of rheumatic diseases (Table 1), concordance rates between identical (MZ) twin pairs are frequently surprisingly low (e.g., 12–15% in rheumatoid arthritis) [31]. This suggests that a variety of genetic and non-genetic factors as well as their complex interactions contribute to disease susceptibility. Among these, epigenetics has recently emerged as a potential explanatory element as signals vary within MZ pairs [32]. This hypothesis is for example sustained for many rheumatic diseases by the observation of gender bias, geographic differential distribution, remission and relapse as well as age-related prevalence but also decline of clinical symptoms with age, all of which are considered good epidemiological, clinical, and etiopathogenic indicators of epigenetic properties [20]. EWAS Epigenome-wide association studies (EWAS), which is related to the better known genome-wide association study (GWAS), aim to identify epigenetic marks associated with disease by performing a hypothesis-free testing across the whole genome [33]. Before discussing the application of EWAS in rheumatic diseases, it is important to consider the current methodological aspects to allow interpretation of the results obtained so far [34]. (Fig.) depicts a flowchart on how EWAS are carried out and the following paragraphs discuss the different points in more details.

Research question Results of epigenetic studies provide insight into disease mechanisms or potential biomarkers. Indeed, although challenging to detect [35], in contrast to genetic polymorphisms, plastic DNAm properties may help identify much-needed biomarkers for disease prognosis, monitoring progression, or predicting response to treatment with potential direct translation into the clinics. Comparison of patients with positive specific autoimmune antibodies but free of clinical symptoms versus clinically affected or in responders versus non-responders for a specific medication may help identify such biomarkers.

F. Zufferey et al. / Seminars in Arthritis and Rheumatism ] (2013) ]]]–]]]

3

Research queson Idenficaon of differently methylated regions (DMRs) associated with: e.g. -disease pathogenesis? -disease progression? - therapeuc biomarkers?

Study design -Populaon based: case-control -Family based: MZ discordant twins, parent–offspring, transgeneraonal

Populaon selecon -sample size -case definion -gender, age matching

Sample collecon -selecon of one or more relevant ssue/cell populaon for DNA extracon -longitudinal versus cross-seconal sampling -assessment of potenal confounders (e.g.: smoking, drugs)

Methylaon assessment -array-based plaorms, enrichment-based methods, bisulfite sequencing

Stascal analyses -DMRs idenficaon -significance threshold seng (correcon for mulple tesng) -correcon for confounders

Validaon -confirmaon of methylaon status by different methods

Replicaon -replicaon of significant results in at least one independent populaon

Integrave analysis with other omics and experimental data -gene expression -genotyping -histone modificaons, miRNAs -funconal experiments Fig. EWAS flowchart.

Study design The two most frequently used population-based study designs in EWAS are case–control and MZ discordant twin pairs [33]. The former comprises two groups of unrelated individuals—generally a healthy control group and a case group affected by a disease of interest drawn from the same population. Ideally cases and controls should be matched for age, sex, ethnicity, and other potential confounders [23]. The discordant MZ twin design uses a different approach. Methylation patterns between MZ twin pairs discordant for a disease are compared within each pair. In this situation, there is internal control of genetic and age/cohort variation. In addition, many other factors such as upbringing and social factors are closely matched [36]. Population selection: power and sample size Power calculation in EWAS is a matter of on-going research and debate. It remains somewhat speculative as the real effect size of methylation differences is not completely known [36]. However, most findings so far have shown that effect size differences are small, at about 5%. Based on the very scarce current literature, around 15–20 MZ discordant twin pairs and several hundreds of cases and controls would be necessary to reach sufficient power [33,37] under specific experimental and study conditions. The discovery sample would also then need replication in a larger

independent replication sample. However epigenome-wide calculations are not possible with any degree of accuracy as many known and unknown variables influence power calculation in EWAS that vary across studies depending on factors such as the study design, the coverage of the methylation assay and its variability, confounders, and the trait under consideration [36]. Sample collection Choice of biological sample The choice of material sample for DNA extraction will depend on the trait of interest and should be of developmental or functional relevance. More than one cell or tissue of interest may be relevant, for example, in RA, where both fibroblast-like synoviocytes (FLS) and T cells have been shown to be involved in the pathogenesis [38]. As methylation is determined across different cells from a sample tissue or organ and represents an average measurement across the sampled DNA, cellular heterogeneity may confound the results. This is particularly pertinent for venous whole blood, which although highly accessible, shows great cellular heterogeneity [39]. Moreover, heterogeneity of cell composition may be responsible for systematic bias between cases and controls. Neutropenia which may be drug induced or autoimmune-related as in Felty's syndrome may be such an example. Saliva samples for DNA have proved problematic as they contain heterogeneous cell composition, like blood, but which can

4

F. Zufferey et al. / Seminars in Arthritis and Rheumatism ] (2013) ]]]–]]]

Confounders Any lifestyle effect such as smoking or alcohol or pathophysiology of the disease and/or any treatment prior to sample collection could itself generate epigenetic change, which may be of interest or a confounder [23]. There are also recognized technical sources of variation that should be taken into account in both sample collection and assay randomization (especially batch effects) [33]. The use of different array platforms and/or at a different date may account for the observed variability in DNAm between cases and controls and give a spurious association, which is called a batch effect.

dataset (global hypomethylation versus hypermethylation), (ii) the detection of differently methylated regions (DMRs) after statistical comparisons between groups (e.g., cases versus controls). DMRs may be defined according to statistical significance after comparisons and relative and absolute differences or ratios in DNAm between groups and controls. After normalization, absolute DNAm levels calculated from bisulfite-sequencing data are generally expressed as beta values (β). β Values conceptually stem for the observation that the distribution of DNAm levels across the genome resembles a β distribution with a finite scale of 0–1 (where 0 ¼ absence of methylation and 1 ¼ complete methylation). As for GWAS, multiple testing must be taken into account in EWAS. The principle is adjusted statistically for the number of tests performed according to the coverage of the methylation assay to lower the number of false-positive signals due to multiple comparisons and obtain an adjusted p-value. There is however no clear consensus for epigenome-wide significance level as there is for GWAS yet [33], and approaches remain conservative. A DMR may be situated in the promoter region or in a gene body, but also in inter-genic regions depending on the coverage and type of methylation assay. From the list of DMRs, mapping the corresponding gene and determining its level of methylation (one gene may have many DMRs, possibly in opposite directions) is a further challenge. Usually, bioinformatic tools such as gene pathway analyses help identify biologically meaningful trends in the lists of DMRs.

Methylation assessment

Validation and replication

Methylation may be assessed at specific candidate gene loci or genome-wide. A wide range of methods have been developed to explore methylation changes across the whole genome in a hypothesis-free manner (reviewed in Refs. [41–44]). We will pinpoint some basic concepts to understand this rapidly evolving topic. There are two types of platforms in genome-wide methylation analysis: (1) array-based (bisulfite microarrays) and (2) sequencingbased methods (bisulfite sequencing and enrichment-based methods) [42]. Bisulfite sequencing is considered the gold standard, it offers resolution to a single-base and is generally accurate and reproducible, but remains expensive. Enrichment-based methods include for example the use of DNA-methylation-specific antibodies as in methylated DNA immunoprecipitation coupled with highthroughput sequencing (MeDIP–seq) to enrich for a fraction of highly methylated (or unmethylated) DNA fragments. It is then followed by quantification of the enrichment of specific fragments by next-generation sequencing. Bisulfite microarrays are 10-fold cheaper currently and measure DNAm at a preselected fraction of Cytosines throughout the genome, each of which is represented by dedicated probes on the microarray. Resolution is high (at a single CpG), but coverage is limited and may not reflect the most interesting areas because methylation is only assessed at the CpGs on the probes. Illumina Infinium assays are the most widely used method for methylation. The latest version covers around 450,000 CpGs sites, which corresponds to only about 1.5% of CpGs in the human genome; it focuses on promoter regions and includes only few CpG-dense regions. More recently, it has been shown that the array only contains less than 9% of the important variable CpG regions [45]. Currently most of the studies published in the rheumatic disease literature have used bisulfite microarray methods.

Reliability of the data and their significance should be confirmed by validation and replication of the results in another independent sample.

Statistical analyses

Major methodological issues render the results published so far preliminary, and interpretation, like in early candidate gene days, should be cautious. Many studies are underpowered, lack replication data, and have low methlylation coverage. Indeed, all EWAS

vary within the same day, adding considerably to the noise in the analysis. Temporality–causality Methylation changes found associated with a condition may take place prior to the phenotype but not necessarily be causally linked. In addition, methylation change may occur as a consequence of disease and not necessarily be the triggering step in the disease susceptibility: conclusions on causality cannot be drawn from retrospective data. Ideally, longitudinal cohorts or utilization of Guthrie cards (stored blood for metabolic screening in newborns) [40] will allow the sequence of change to be determined and point to underlying causality. Although methylation may be the link between early life changes and adult disease, no methylation study for any disease has yet proven causality in humans.

Assessment of methylation sites is generally treated as a binary variable: (i) the assessment of global trends in DNAm in the

Integrative analysis with other omics and experimental data Comparison of DNAm results with gene expression analysis or other epigenetic marks (e.g., histone modifications, miRNAs, and open chromatin state) would provide a useful overall picture of the epigenetic regulatory landscape. Moreover, as methylation status has a genetic component, the addition of genotyping or DNA sequencing data in case–control studies allows the exploration of the relationship between sequence variation and methylation regulation—called Methylation QTLs (MeQTLs).

Methylation changes in genes and across the genome in rheumatic diseases Methylation changes in specific genes have been identified in many rheumatic diseases through candidate gene methylation studies and, more recently, EWAS. In the present review, all published EWAS in rheumatic diseases until April 2013 have been included, along with the most significant candidate gene studies for key diseases. Recent literature on EWAS and rheumatic diseases has mainly focused on autoimmune diseases and includes studies on rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), and dermatomyositis (DM). In addition, there are recent studies of osteoporosis (OP) and osteoarthritis (OA) (Table 2). General overview

Table 2 EWAS in the rheumatic diseases Disease Study design

Phenotype

Sample size

Replicationa Cell/tissue

Methylation assay

Methylation profile

Statistical criteria for DMRs definition

Gene expression platformb

Main results

References

–

No significant changes

[50]

qRT-PCR

[57] Hypomethylation in genes relevant to RA, such as CHI3L1, CASP1, STAT3, MAP3K5, MEFV, and WISP3 and hypermethy;lation in TGFBR2 and FOXO1.

qRT-PCR

Altered DNAm in genes such as IL6R, CAPN8, DPP4, and HOX.

–

[59] Two clusters within the MHC region that may, through differential methylation, mediate the risk of RA after correction for cellular heterogeneity. Partial replication in monocyte cell fractions in an independent cohort of 12 cases and 12 controls. Hypomethylation in immune-related genes such as [50] IFNGR2, MMP14, LCN2, CSF3R, PECAM1, CD9, and AIM2.

Genome- DMRs DMGs wide RA

5

–

WBCs

Illumina GoldenGate

¼

CC

RA versus OA

6–5

–

FLS

Illumina 450 K

¼

CC

RA versus OA

6–6

–

FLS

Illumina 450 K

¼

CC

ACPA þ RA versus controls

354–337 12–12

WBCs

Illumina 450 K

NP

MZ pairs SLE

5

–

WBCs

Illumina GoldenGate

↓

CC

SLE versus controls

12–12

–

CD4 þ T

Illumina 27 K

NP

CC

SLE versus controls SLE versus controls

12–12

–

WBCs

↓

18

18

Naïve CD4 þ T

Illumina 27 K Illumina 450 K

CC

0 p o 0.05 (adjusted for multiple testing) and min mean methylation change of 10% 1859 1206 p o 0.05 (adjusted for multiple testing) and average difference in methylation level 40.1 2571 1240 p o 0.05 and absolute value of the median difference between b-values 4 0.1 51,476 18,227 p o 0.05 (adjusted for multiple testing) 0

49

341

qRT-PCR 49 p o 0.05 (adjusted for multiple testing) and min mean methylation change of 10% 336 p o 0.05 and at least – 1.2-fold difference in methylation level

2165

2165

NP

86

47

DM

MZ pairs DM

5

–

WBCs

Illumina GoldenGate

¼

0

0

OA

CC

OA versus controls

25–20

–

Articular cartilage

Illumina 27 K

↓

91

91

OP

CC

Hip fracture versus hip OA

27–26

–

Trabecular bone

Illumina 27 K

NP

241

228

[58]

[51] Hypomethylation of immune-related genes such as CD9, MMP-9, PDGFRA, and BST2. Hypermethylation in RUNX3 involved in T cell maturation and genes involved in folate biosynthesis. o0.05 – Hypomethylation of the IL10 and ILR2 genes in SLE [86] patients Hypomethylation in interferon-regulated genes in HumanHT-12 r0.001 (adjusted for [52] naïve CD4 þ T cells in SLE, including IFIT1, IFIT3, MX1, v4 Expression multiple testing) and at least 1.2-fold difference in BeadChip array STAT1, IFI44L, USP18, TRIM22 and BST2 (Illumina) methylation level – No significant changes [50] o0.05 (adjusted for multiple testing) and min mean methylation change of 10% o0.05 (adjusted for – DMRs mainly related to the inflammatory/defense [66] multiple testing) response (demethylated in OA). RUNX1 and MSX1 were respectively the 2 most hypo and hypermethylated genes in OA. o0.05 (adjusted for – Majority of hypomethylated genes in OP compared to [68] multiple testing) OA. No classic gene bone candidate but enrichment for GWAS bone trait-associated genes and the homeobox group of genes.

F. Zufferey et al. / Seminars in Arthritis and Rheumatism ] (2013) ]]]–]]]

SLE

MZ pairs RA

MZ pairs stands for MZ discordant twin pairs. Abbreviations: CC ¼ case–control; DMRs ¼ differently methylated regions; DMGs ¼ differently methylated genes; SLE ¼ systemic lupus erythematosus; RA ¼ rheumatoid arthritis; DM ¼ dermatomyositis; OA ¼ osteoarthritis; OP ¼ osteoporosis; WBCs ¼ white blood cells; FLS ¼ fibroblast-like cells; NP ¼ not provided; qRT-PCR ¼ quantitative real-time PCR. Symbols: Similar global methylation level (¼ ), global hypomethylation (↓). a b

In an independent population. Only gene expression analyses performed on the same population as for DMRs discovery were considered. 5

6

F. Zufferey et al. / Seminars in Arthritis and Rheumatism ] (2013) ]]]–]]]

in rheumatic diseases published so far, have been generated using methylation arrays of variable coverage. In addition, methods and statistical analyses—notably criteria to define a probe as a DMR— differ greatly between studies rendering meaningful crosscomparison a challenge. Correction for multiple testing or potential confounders has not been taken into account in several studies. In addition, in some instances, the control group was not always composed of healthy individuals, but those with different diseases (e.g., RA versus OA or OA versus OP), making generalizability difficult. Only a few studies have integrated other experiments like gene expression or other omics experiments, which will be necessary to interpret the results in a biologically meaningful manner. Overall, these studies provide preliminary evidence for specific methylation changes in rheumatic disease and suggest methylation change in novel candidate genes and clustering of disease subgroups according to methylation profile. Average differences in methylation levels (β values) between the groups compared were generally modest (o 15%) across all studies, suggesting relatively small effect sizes. When examined, expression data showed an overall inverse—but variable and inconsistent—relationship between DNAm and particular gene expression levels, highlighting again the complexity of regulation of gene expression.

Systemic lupus erythematosus First evidence of altered DNAm status in SLE came from a study that found that T cells from patients with active lupus have a decreased global DNAm level and that 5-azacytidine (5-azaC), an inhibitor of DNA methylation, could induce auto-reactivity in cloned CD4 þ T cells and autoimmune syndrome [46,47]. Moreover, classic methylation-sensitive autoimmunity-related genes have been identified in CD4 þ T cells in SLE, such as ITGAL (CD11A), PRF1 (perforin), TNFSF7 (CD70), and CD40LG (reviewed in Ref. [48]). EWASs published so far suggest a tendency towards hypomethylation in SLE compared to healthy controls and support the central role of type I interferon pathway in SLE pathogenesis [49] as interferon-related genes have been identified as differently methylated across studies. However, major limitations include small sample size, low coverage of the epigenome, and scarce clinical data with heterogeneous case population. The first EWAS study using the MZ discordant twins design was published in 2010 and included 15 pairs of MZ autoimmune discordant twins (five with RA, five with SLE, and five with DM) [50]. Using a low-coverage methylation array representing 807 gene promoters, the authors identified consistent methylation changes in SLE twins relative to unaffected co-twins in certain genes influencing immune function and a global decrease in DNAm levels. One year later, the methylation profile of CD4 þ T cells of 12 unrelated lupus patients and 12 normal controls partially matched for ethnicity was compared [51]. The authors identified 341 differently methylated (DM) CG between cases and controls—the majority were hypomethylated in SLE, notably in known connective tissue-related genes (e.g., MMP-9 and PDGFRa). Of note, PDGFRa autoantibodies are present in less than half of lupus patients and contribute to the development of autoimmune hemolytic anemia. Using a more powerful assay and two independent cohorts (one for discovery (n ¼ 18) from the same institution as [51] and one for replication (n ¼ 18)), Coit et al. [52] identified and replicated 86 DMRs between naïve T cells from patients and controls (36 pairs) in 47 genes (mainly hypomethylated). A majority of them were regulated by type I interferon. However, gene expression analysis assessed from microarray in a subset of study participants did not reveal different expression levels of

these DMGs in naïve CD4, raising the question of their biological significance. Lin et al. identified global hypomethylation in SLE as compared to controls in a small population of 12 active SLE cases and 12 controls. No correction for multiple testing was apparently applied, and limited clinical and demographic data were provided. Among the 2165 genes with significant methylation changes, the authors focused on IL1R2 and IL10 (the latter was also hypomethylated in Ref. [50]) as biologically relevant candidates for further confirmation in 66 SLE and 102 healthy controls. Both genes were hypomethylated in SLE relative to healthy controls but as well in RA patients (n ¼ 12). This finding corroborates recent evidence for overlapping methylome signatures between different autoimmune diseases similar to that found for single nucleotide polymorphisms in GWAS [53].

Rheumatoid arthritis The investigation of DNAm in RA began with relatively small candidate gene studies showing altered methylation status of single gene promoters (generally at a single CpG site), including DR3, IL6, IL10, IL1R2, and CXCL12 identified in peripheral blood mononuclear cells and FLSs (reviewed in Ref. [54]). In addition, evidence for epigenetic involvement in RA with respect to gender bias is supported by one study, which shows significantly increased skewed X-chromosome inactivation pattern in female RA patients (n ¼ 106) compared to controls (n ¼ 257) [55]. Also, demethylation of the promoter of the CD40L gene on X chromosome was observed in CD4 þ T cells from female (n ¼ 11) but not male (n ¼ 9) RA patients [56]. Several EWAS in RA have been published recently [50,57–59]. They suggest the presence of specific methylation changes in RA, and interesting regions such as the IL6R gene and the MCH region which have been identified as differently methylated. However, cross-comparisons are difficult because of the use of different tissue (whole blood versus FLSs) and different “control” groups. Moreover, due to the known clinical heterogeneity of RA, only one study included only ACPA-positive patients, the others included heterogeneous and treated individuals, which may have biased the results. In the twin discordant study from Javierre et al. [50], there were no significant changes in methylation profile within the five discordant RA twin pairs. More recently, two case–control studies compared methylation profiles derived from a small number of cultured FLSs in patients with RA or OA (n ¼ 6 RA versus n ¼ 5 OA in the first study and n ¼ 5 RA versus n ¼ 6 OA in the second one) [57,58]. The authors provided limited clinical and demographic data, and statistical analyses were not adjusted for age, medication, or batch effects. Having said that, both research groups identified an RA “methylome signature” with partially overlapping genes between both sets, including genes like MMP20, RASGRF2, ADAMTS2, EGF, and TIMP2. Hypomethylated loci were identified by the first group in key genes relevant to RA. In a study by Nakano et al. [57], gene expression determined by PCR for 7 genes picked up from the DMRs list on the same samples of patients (6 OA and 13 RA FLS lines) showed overall greater expression in RA in comparison with OA, but the relationship for individual genes was variable. de la Rica and co-authors reanalyzed available RA FLSs and OA FSs array expression data from the Gene Expression Omnibus (GEO) database and compared the results with their DNAm data. Two hundred eight annotated CpGs displayed an inverse correlation between expression and methylation levels. Then, they investigated the expression status of 10 genes displaying a change in DNAm in the set of samples used in this study.

F. Zufferey et al. / Seminars in Arthritis and Rheumatism ] (2013) ]]]–]]]

Results showed inconsistent relationship between methylation level and gene expression. Finally, Liu et al. [59] analyzed a much larger cohort of RA patients (n ¼ 354 and 337 controls) from a population-based cohort. As a major improvement on other studies, cases comprised only ACPA-positive RA patients, free of treatment and the results were replicated in a second independent sample. Of the 51,476 identified DMRs, the authors identified genotype (single nucleotide polymorphisms)-dependant DMRs to help identify methylation that was inferred as a likely result of disease. Of note, the potential causal relationship was not based on observational data (or Mendelian Randomization methods) but was determined statistically using a series of cross-sectional conditional correlation analyses based on the causal inference literature. The authors identified two clusters within the major histocompatibility complex region, whose differential methylation potentially mediates genetic risk for rheumatoid arthritis. As the analyses were performed on whole blood-derived DNA samples cells, a correction was applied for cellular heterogeneity by estimating and adjusting for cell-type proportions, as well as for age, gender, and smoking status. The fact that DMRs were found in the HLA region suggests that the results are likely to be real but the study was powered to find major effects only. The authors partially replicated their findings on additional 12 RA patients and 12 controls on the monocyte fraction of peripheral blood cells.

Osteoarthritis (OA) The role of DNAm in OA has in general focused on methylation status and gene expression at the promoters of candidate genes known to be important for cartilage maintenance (reviewed in Ref. [60]). These in vitro studies have included the promoters of a number of metalloproteinases at a single CpG (e.g., MMP13 and ADAMTS4 in end-stage OA chondrocytes in one study on 16 patients with OA and 10 controls) [61]. The expression of MMP13 was further shown to be modulated by epigenetic change in the promoter regions of the leptin gene in a candidate gene study on osteoarthritic articular cartilage from 15 participants [62]. It was also postulated that long-term inflammatory cytokines might induce overexpression of IL1B through the methylation of its promoter at a specific CpG based on an experiment on human cultured chondrocytes. A single nucleotide polymorphism in GDF5 was postulated to mediate GDF5 differential expression in OA through DNAm [63] and SOD2 gene—related to the oxidative stress process—may also be regulated by DNAm of several CpG sites within its promoter in OA chondrocytes [64]. No changes were seen in total DNA methylation content in OA chondrocytes relative to control by chromatography [65]. Only one unreplicated small EWAS on knee OA (25 cases versus 20 controls, articular cartilage samples) has been published. The authors found global hypomethylation in OA and identified 91 differentially methylated (DM) probes between cases and controls [66]. Correction for age but not for cellular heterogeneity was applied. The authors identified a small sub-group of cases (7 out of 23) with a particular methylation pattern (1357/23,837 probes differentiated this group from the others using a statistics tool for unsupervised class discovery). This sub-group was enriched for genes associated with inflammatory response (hypomethylated) and extracellular matrix constituents (hypermethylated). Although potentially interesting, this finding is based on a small sample and clinical details were not discussed, making these results still preliminary. Osteoporosis The contribution of epigenetics and methylation to bone mass and homeostasis is poorly understood, limited mainly to genes of

7

importance in osteoblast and osteoclast differentiation from small candidate gene studies (reviewed in Ref. [67]). Genome-wide methylation profiles from pieces of trabecular bone of the femoral head from women (27 cases, mean age 80 7 3 years) undergoing hip surgery for fracture (arguably considered as a proxy for OP) were recently compared to those of patients with hip OA (26 patients, mean age 75 7 6 years) [68]. Of the 228 DMGs (the majority was less methylated in OP than in OA), there was enrichment for GWAS bone trait-associated genes and in genes participating in skeletal development, particularly in the homeobox superfamily of transcription factors. Of note, the leptin gene showed significant methylation difference between OP and OA in this study, although there were no other DMRs in classic bone gene candidates. This unique study without replication and integrative experiments is too preliminary to draw any firm conclusions.

Pain sensitivity Chronic pain, even if not a disease per se, is a major symptom of most rheumatic diseases and a key therapeutic aim. The concept of central sensitivity syndrome may underlie conditions like fibromyalgia and overlapping conditions such as chronic fatigue syndrome [69]. There is increasing evidence for epigenetic mechanisms in both the development and maintenance of pain states [70]. Heat pain suprathreshold response was determined among 2500 twins from the TwinsUK cohort, as these measures have been shown to reflect propensity to develop chronic pain. Using genome-wide MeDIP-sequencing on whole blood from 25 discordant MZ twin pairs (discovery cohort) for objective measures of heat pain sensitivity and 50 unrelated samples, Bell et al. [71] identified nine DMRs in eight unique regions associated with high or low pain sensitivity. MZ twin and unrelated sample EWAS were performed separately followed by a meta-analysis of combined results. The strongest association signal was in the promoter of the ion channel gene TRPA1, a known pain gene whose product is expressed in peripheral nociceptors [72]. Of interest, most of the top DMRs were not covered on the Illumina HumanMethylation450 BeadChip (used for validation), stressing the importance of increased methylation coverage in EWAS. Variability in WBC sub-type (lymphocytes, neutrophils, basophils, and eosinophils) did not show a major effect on the most significant DMRs in this study. Longitudinal data were obtained for 33 individuals from the MZ twin sample (2–3 years apart) and patterns at nine DMRs showed relatively stable methylation over time. This study illustrates the efficiency of adequate sample size and in-depth methylation coverage for gene discovery in EWAS. It suggests epigenetic influence on sensitivity to pain in normal individuals, leading to potential future interesting development in pathological conditions of chronic pain states such as chronic back pain or fibromyalgia.

Future perspectives Other unexplored rheumatic diseases may represent good candidates for epigenetic investigation. For instance, gout has a high and increasing prevalence (E 3.9%) [73] and cohorts with extended genomic data already exist. Interestingly, in contrast to high genetic contribution to plasma urate level, gout has a relatively low estimated heritability with an important component of non-shared environment in phenotypic variance, suggesting the environmental influence is strong and possibly epigenetically mediated [74]. Further, the application of alternative study designs might be suitable for some diseases. For instance, comparison of affected

8

F. Zufferey et al. / Seminars in Arthritis and Rheumatism ] (2013) ]]]–]]]

versus unaffected cells within a tissue from the same individual may provide insight into acquired epigenetic modification in a similar approach to cancer. This may be applicable, for example, to Paget's disease or from different joints in localized forms of inflammatory arthritis. In addition, EWAS, if sufficiently powered with large sample sizes or twin designs with epigenome-wide coverage, have great potential for the identification of markers for disease onset, progression, or response to treatment for diseases such as RA or SLE. Collecting samples at diagnosis before medication starts and then longitudinally comparing the methylation profile between groups of responders versus non-responders may show methylation differences induced by treatment response or previous differences preluding drug resistance (e.g., in RA patients treated with specific DMARDs or biological drugs) [75].

Conclusion Epigenetics is a major part of the future of rheumatology. The increasing number of published epigenome-wide methylation studies brings new challenges of interpretation to the medical and scientific community. Current data from genome-wide methylation analyses suggest specific methylation changes in several rheumatic diseases compared to healthy controls or control groups. These studies have allowed the identification of differentially methylated genes previously implicated in the disease pathology but have also highlighted new gene associations with individual diseases. Improved methodology and generation of methylation data using more powerful techniques with coverage of the methylome are likely to follow and will shed new light on disease pathogenesis. Larger sample sizes, replication, and biologic validation will be necessary to integrate these data and allow their potential translation into the clinical arena. Longitudinal data will be advantageous in teasing apart cause and effect. We can however expect advances in epigenetics to aid disease diagnostics, classification, and therapy in the future. References [1] Bird A. Perceptions of epigenetics. Nature 2007;447:396–8 [PubMed PMID: 17522671]. [2] Berger SL, Kouzarides T, Shiekhattar R, Shilatifard A. An operational definition of epigenetics. Genes Dev 2009;23:781–3 [PubMed PMID: 19339683]. [3] Lu Q. The critical importance of epigenetics in autoimmunity. J Autoimmun 2013;41:1–5 [PubMed PMID: 23375849]. [4] Baxter D, McInnes IB, Kurowska-Stolarska M. Novel regulatory mechanisms in inflammatory arthritis: a role for microRNA. Immunol Cell Biol 2012; 90:288–92 [PubMed PMID: 22249200]. [5] Oppermann U. Why is epigenetics important in understanding the pathogenesis of inflammatory musculoskeletal diseases? Arthritis Res Ther 2013;15:209 [PubMed PMID: 23566317]. [6] Ballestar E. Epigenetic alterations in autoimmune rheumatic diseases. Nat Rev Rheumatol 2011;7:263–71 [PubMed PMID: 21343899]. [7] Khare T, Pai S, Koncevicius K, Pal M, Kriukiene E, Liutkeviciute Z, et al. 5-hmC in the brain is abundant in synaptic genes and shows differences at the exon– intron boundary. Nat Struct Mol Biology 2012;19:1037–43 [PubMed PMID: 22961382. Pubmed Central PMCID: 3465469]. [8] Lister R, Pelizzola M, Dowen RH, Hawkins RD, Hon G, Tonti–Filippini J, et al. Human DNA methylomes at base resolution show widespread epigenomic differences. Nature 2009;462:315–22 [PubMed PMID: 19829295. Pubmed Central PMCID: 2857523]. [9] Irizarry RA, Ladd-Acosta C, Wen B, Wu Z, Montano C, Onyango P, et al. The human colon cancer methylome shows similar hypo- and hypermethylation at conserved tissue-specific CpG island shores. Nat Genet 2009;41:178–86 [PubMed PMID: 19151715. Pubmed Central PMCID: 2729128]. [10] Kilpinen H, Dermitzakis ET. Genetic and epigenetic contribution to complex traits. Hum Mol Genet 2012;21:R24–8 [PubMed PMID: 22976472]. [11] Portela A, Esteller M. Epigenetic modifications and human disease. Nat Biotechnol 2010;28:1057–68 [PubMed PMID: 20944598]. [12] Jones PA. Functions of DNA methylation: islands, start sites, gene bodies and beyond. Nat Rev Genet 2012;13:484–92 [PubMed PMID: 22641018]. [13] Smith ZD, Meissner A. DNA methylation: roles in mammalian development. Nat Rev Genet 2013;14:204–20 [PubMed PMID: 23400093].

[14] Deaton AM, Bird A. CpG islands and the regulation of transcription. Genes Dev 2011;25:1010–22 [PubMed PMID: 21576262. Pubmed Central PMCID: 3093116]. [15] Lorincz MC, Dickerson DR, Schmitt M, Groudine M. Intragenic DNA methylation alters chromatin structure and elongation efficiency in mammalian cells. Nat Struct Mol Biol 2004;11:1068–75 [PubMed PMID: 15467727]. [16] Smallwood SA, Tomizawa S, Krueger F, Ruf N, Carli N, Segonds-Pichon A, et al. Dynamic CpG island methylation landscape in oocytes and preimplantation embryos. Nat Genet 2011;43:811–4 [PubMed PMID: 21706000. Pubmed Central PMCID: 3146050]. [17] Varley KE, Gertz J, Bowling KM, Parker SL, Reddy TE, Pauli-Behn F, et al. Dynamic DNA methylation across diverse human cell lines and tissues. Genome Res 2013;23:555–67 [PubMed PMID: 23325432. Pubmed Central PMCID: 3589544]. [18] Bell JT, Tsai PC, Yang TP, Pidsley R, Nisbet J, Glass D, et al. Epigenome-wide scans identify differentially methylated regions for age and age-related phenotypes in a healthy ageing population. PLoS Genet 2012;8:e1002629 [PubMed PMID: 22532803. Pubmed Central PMCID: 3330116]. [19] Feinberg AP. Phenotypic plasticity and the epigenetics of human disease. Nature 2007;447:433–40 [PubMed PMID: 17522677]. [20] Petronis A. Epigenetics as a unifying principle in the aetiology of complex traits and diseases. Nature 2010;465:721–7 [PubMed PMID: 20535201]. [21] Heijmans BT, Tobi EW, Stein AD, Putter H, Blauw GJ, Susser ES, et al. Persistent epigenetic differences associated with prenatal exposure to famine in humans. Proc Natl Acad Sci U S A 2008;105:17046–9 [PubMed PMID: 18955703. Pubmed Central PMCID: 2579375]. [22] Katari S, Turan N, Bibikova M, Erinle O, Chalian R, Foster M, et al. DNA methylation and gene expression differences in children conceived in vitro or in vivo. Hum Mol Genet 2009;18:3769–78 [PubMed PMID: 19605411. Pubmed Central PMCID: 2748887]. [23] Zeilinger S, Kuhnel B, Klopp N, Baurecht H, Kleinschmidt A, Gieger C, et al. Tobacco smoking leads to extensive genome-wide changes in DNA methylation. PLoS One 2013;8:e63812 [PubMed PMID: 23691101. Pubmed Central PMCID: 3656907]. [24] Feil R, Fraga MF. Epigenetics and the environment: emerging patterns and implications. Nat Rev Genet 2011;13:97–109 [PubMed PMID: 22215131]. [25] Gordon L, Joo JE, Powell JE, Ollikainen M, Novakovic B, Li X, et al. Neonatal DNA methylation profile in human twins is specified by a complex interplay between intrauterine environmental and genetic factors, subject to tissuespecific influence. Genome Res 2012;22:1395–406 [PubMed PMID: 22800725. Pubmed Central PMCID: 3409253]. [26] Kerkel K, Spadola A, Yuan E, Kosek J, Jiang L, Hod E, et al. Genomic surveys by methylation-sensitive SNP analysis identify sequence-dependent allele-specific DNA methylation. Nat Genet 2008;40:904–8 [PubMed PMID: 18568024]. [27] Bjornsson HT, Sigurdsson MI, Fallin MD, Irizarry RA, Aspelund T, Cui H, et al. Intra-individual change over time in DNA methylation with familial clustering. J Am Med Assoc 2008;299:2877–83 [PubMed PMID: 18577732. Pubmed Central PMCID: 2581898]. [28] Zhang D, Cheng L, Badner JA, Chen C, Chen Q, Luo W, et al. Genetic control of individual differences in gene-specific methylation in human brain. Am J Hum Genet 2010;86:411–9 [PubMed PMID: 20215007. Pubmed Central PMCID: 2833385]. [29] Bell JT, Pai AA, Pickrell JK, Gaffney DJ, Pique-Regi R, Degner JF, et al. DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome Biol 2011;12:R10 [PubMed PMID: 21251332. Pubmed Central PMCID: 3091299]. [30] Tsai PC, Spector TD, Bell JT. Using epigenome-wide association scans of DNA methylation in age-related complex human traits. Epigenomics 2012; 4:511–26 [PubMed PMID: 23130833]. [31] MacGregor AJ, Snieder H, Rigby AS, Koskenvuo M, Kaprio J, Aho K, et al. Characterizing the quantitative genetic contribution to rheumatoid arthritis using data from twins. Arthritis Rheum 2000;43:30–7 [PubMed PMID: 10643697]. [32] Bell JT, Spector TD. A twin approach to unraveling epigenetics. Trends Genet 2011;27:116–25 [PubMed PMID: 21257220. Pubmed Central PMCID: 3063335]. [33] Rakyan VK, Down TA, Balding DJ, Beck S. Epigenome-wide association studies for common human diseases. Nat Rev Genet 2011;12:529–41 [PubMed PMID: 21747404. Pubmed Central PMCID: 3508712]. [34] Heijmans BT, Mill J. Commentary: the seven plagues of epigenetic epidemiology. Int J Epidemiol 2012;41:74–8 [PubMed PMID: 22269254. Pubmed Central PMCID: 3304528]. [35] Cortessis VK, Thomas DC, Levine AJ, Breton CV, Mack TM, Siegmund KD, et al. Environmental epigenetics: prospects for studying epigenetic mediation of exposure-response relationships. Hum Genet 2012;131:1565–89 [PubMed PMID: 22740325. Pubmed Central PMCID: 3432200]. [36] Bell JT, Spector TD. DNA methylation studies using twins: what are they telling us? Genome Biol 2012;13:172 [PubMed PMID: 23078798. Pubmed Central PMCID: 3491399]. [37] Kaminsky Z, Petronis A, Wang SC, Levine B, Ghaffar O, Floden D, et al. Epigenetics of personality traits: an illustrative study of identical twins discordant for risk-taking behavior. Twin Res Hum Genet 2008;11:1–11 [PubMed PMID: 18251670]. [38] Neumann E, Lefevre S, Zimmermann B, Gay S, Muller-Ladner U. Rheumatoid arthritis progression mediated by activated synovial fibroblasts. Trends Mol Med 2010;16:458–68 [PubMed PMID: 20739221]. [39] Adalsteinsson BT, Gudnason H, Aspelund T, Harris TB, Launer LJ, Eiriksdottir G, et al. Heterogeneity in white blood cells has potential to confound DNA

F. Zufferey et al. / Seminars in Arthritis and Rheumatism ] (2013) ]]]–]]]

[40]

[41] [42] [43] [44] [45]

[46]

[47]

[48]

[49]

[50]

[51]

[52]

[53]

[54] [55]

[56]

[57]

[58]

[59]

[60]

[61]

[62]

methylation measurements. PloS One 2012;7:e46705 [PubMed PMID: 23071618. Pubmed Central PMCID: 3465258]. Beyan H, Down TA, Ramagopalan SV, Uvebrant K, Nilsson A, Holland ML, et al. Guthrie card methylomics identifies temporally stable epialleles that are present at birth in humans. Genome Res 2012;22:2138–45 [PubMed PMID: 22919074. Pubmed Central PMCID: 3483543]. Bock C. Analysing and interpreting DNA methylation data. Nat Rev Genet 2012;13:705–19 [PubMed PMID: 22986265]. Laird PW. Principles and challenges of genomewide DNA methylation analysis. Nat Rev Genet 2010;11:191–203 [PubMed PMID: 20125086]. Beck S, Rakyan VK. The methylome: approaches for global DNA methylation profiling. Trends Genet 2008;24:231–7 [PubMed PMID: 18325624]. Heyn H, Esteller M. DNA methylation profiling in the clinic: applications and challenges. Nat Rev Genet 2012;13:679–92 [PubMed PMID: 22945394]. Ziller MJ, Gu H, Muller F, Donaghey J, Tsai LT, Kohlbacher O, et al. Charting a dynamic DNA methylation landscape of the human genome. Nature 2013;500:477–81 [PubMed PMID: 23925113]. Richardson B, Scheinbart L, Strahler J, Gross L, Hanash S, Johnson M. Evidence for impaired T cell DNA methylation in systemic lupus erythematosus and rheumatoid arthritis. Arthritis Rheum 1990;33:1665–73 [PubMed PMID: 2242063]. Richardson BC, Strahler JR, Pivirotto TS, Quddus J, Bayliss GE, Gross LA, et al. Phenotypic and functional similarities between 5-azacytidine-treated T cells and a T cell subset in patients with active systemic lupus erythematosus. Arthritis Rheum 1992;35:647–62 [PubMed PMID: 1376122]. Zhang Y, Zhao M, Sawalha AH, Richardson B, Lu Q. Impaired DNA methylation and its mechanisms in CD4( þ )T cells of systemic lupus erythematosus. J Autoimmun 2013;41:92–9 [PubMed PMID: 23340289]. Bronson PG, Chaivorapol C, Ortmann W, Behrens TW, Graham RR. The genetics of type I interferon in systemic lupus erythematosus. Curr Opin Immunol 2012;24:530–7 [PubMed PMID: 22889593]. Javierre BM, Fernandez AF, Richter J, Al-Shahrour F, Martin-Subero JI, Rodriguez-Ubreva J, et al. Changes in the pattern of DNA methylation associate with twin discordance in systemic lupus erythematosus. Genome Res 2010;20:170–9 [PubMed PMID: 20028698. Pubmed Central PMCID: 2813473]. Jeffries MA, Dozmorov M, Tang Y, Merrill JT, Wren JD, Sawalha AH. Genome wide DNA methylation patterns in CD4 þ T cells from patients with systemic lupus erythematosus. Epigenetics 2011;6:593–601 [PubMed PMID: 21436623. Pubmed Central PMCID: 3121972]. Coit P, Jeffries M, Altorok N, Dozmorov MG, Koelsch KA, Wren JD, et al. Genomewide DNA methylation study suggests epigenetic accessibility and transcriptional poising of interferon-regulated genes in naive CD4 þ T cells from lupus patients. J Autoimmun 2013;43:78–84 [PubMed PMID: 23623029]. Tost TTN, Boudaoud S, Micelli C, Mariette X. High-throughput DNA methylation analysis of cell sorted blood cell populations reveals widespread epigenetic deregulation in Sjogren's syndrome. EULAR abstract; 2013. Klein K, Gay S. Epigenetic modifications in rheumatoid arthritis: a review. Curr Opin Pharmacol 2013;2:420–5 [PubMed PMID: 23384968]. Chabchoub G, Uz E, Maalej A, Mustafa CA, Rebai A, Mnif M, et al. Analysis of skewed X-chromosome inactivation in females with rheumatoid arthritis and autoimmune thyroid diseases. Arthritis Res Ther 2009;11:R106 [PubMed PMID: 19589151. Pubmed Central PMCID: 2745787]. Liao J, Liang G, Xie S, Zhao H, Zuo X, Li F, et al. CD40L demethylation in CD4( þ ) T cells from women with rheumatoid arthritis. Clin Immunol 2012;145: 13–8 [PubMed PMID: 22889643]. Nakano K, Whitaker JW, Boyle DL, Wang W, Firestein GS. DNA methylome signature in rheumatoid arthritis. Ann Rheum Dis 2013;72:110–7 [PubMed PMID: 22736089. Pubmed Central PMCID: 3549371]. de la Rica L, Urquiza JM, Gomez-Cabrero D, Islam AB, Lopez-Bigas N, Tegner J, et al. Identification of novel markers in rheumatoid arthritis through integrated analysis of DNA methylation and microRNA expression. J Autoimmun 2013;41:6–16 [PubMed PMID: 23306098]. Liu Y, Aryee MJ, Padyukov L, Fallin MD, Hesselberg E, Runarsson A, et al. Epigenome-wide association data implicate DNA methylation as an intermediary of genetic risk in rheumatoid arthritis. Nat Biotechnol 2013;31:142–7 [PubMed PMID: 23334450. Pubmed Central PMCID: 3598632]. Barter MJ, Bui C, Young DA. Epigenetic mechanisms in cartilage and osteoarthritis: DNA methylation, histone modifications and microRNAs. Osteoarthritis Cartilage 2012;20:339–49 [PubMed PMID: 22281264]. Roach HI, Yamada N, Cheung KS, Tilley S, Clarke NM, Oreffo RO, et al. Association between the abnormal expression of matrix-degrading enzymes by human osteoarthritic chondrocytes and demethylation of specific CpG sites in the promoter regions. Arthritis Rheum 2005;52:3110–24 [PubMed PMID: 16200590]. Iliopoulos D, Malizos KN, Tsezou A. Epigenetic regulation of leptin affects MMP-13 expression in osteoarthritic chondrocytes: possible molecular target for osteoarthritis therapeutic intervention. Ann Rheum Dis 2007;66:1616–21 [PubMed PMID: 17502362. Pubmed Central PMCID: 2095321].

9

[63] Reynard LN, Bui C, Canty-Laird EG, Young DA, Loughlin J. Expression of the osteoarthritis-associated gene GDF5 is modulated epigenetically by DNA methylation. Hum Mol Genet 2011;20:3450–60 [PubMed PMID: 21642387]. [64] Scott JL, Gabrielides C, Davidson RK, Swingler TE, Clark IM, Wallis GA, et al. Superoxide dismutase downregulation in osteoarthritis progression and endstage disease. Ann Rheum Dis 2010;69:1502–10 [PubMed PMID: 20511611]. [65] Sesselmann S, Soder S, Voigt R, Haag J, Grogan SP, Aigner T. DNA methylation is not responsible for p21WAF1/CIP1 down-regulation in osteoarthritic chondrocytes. Osteoarthritis Cartilage 2009;17:507–12 [PubMed PMID: 18954998]. [66] Fernandez-Tajes J, Soto-Hermida A, Vazquez-Mosquera ME, Cortes-Pereira E, Mosquera A, Fernandez-Moreno M, et al. Genome-wide DNA methylation analysis of articular chondrocytes reveals a cluster of osteoarthritic patients. Ann Rheum Diseases 2013:16 [PubMed PMID: 23505229]. [67] Delgado-Calle J, Garmilla P, Riancho JA. Do epigenetic marks govern bone mass and homoeostasis? Curr Genomics 2012;13:252–63 [PubMed PMID: 23115526. Pubmed Central PMCID: 3382279]. [68] Delgado-Calle J, Fernandez AF, Sainz J, Zarrabeitia MT, Sanudo C, GarciaRenedo R, et al. Genome-wide profiling of bone reveals differentially methylated regions in osteoporosis and osteoarthritis. Arthritis Rheum 2013;65: 197–205 [PubMed PMID: 23124911]. [69] Yunus MB. Fibromyalgia and overlapping disorders: the unifying concept of central sensitivity syndromes. Semin Arthritis Rheum 2007;36:339–56 [PubMed PMID: 17350675]. [70] Denk F, McMahon SB. Chronic pain: emerging evidence for the involvement of epigenetics. Neuron 2012;73:435–44 [PubMed PMID: 22325197]. [71] Bell JT, Loomis AK, Butcher LM, Gao F, Zhang B, Hyde CL, et al. Differential methylation of the TRPA1 promoter in pain sensitivity. Nat Commun [in press]. [72] Fajardo O, Meseguer V, Belmonte C, Viana F. TRPA1 channels mediate cold temperature sensing in mammalian vagal sensory neurons: pharmacological and genetic evidence. J Neurosci 2008;28:7863–75 [PubMed PMID: 18667618]. [73] Zhu Y, Pandya BJ, Choi HK. Prevalence of gout and hyperuricemia in the US general population: the National Health and Nutrition Examination Survey 2007–2008. Arthritis Rheum 2011;63:3136–41 [PubMed PMID: 21800283]. [74] Krishnan E, Lessov-Schlaggar CN, Krasnow RE, Swan GE. Nature versus nurture in gout: a twin study. Am J Med 2012;125:499–504 [PubMed PMID: 22365026]. [75] Webser APD, Eyre S, Flynn E, Martin P, Wilson G, Morgan AW, et al. Differential methylation related to response to etanercept in patients with rheumatoid arthritis. EULAR abstract; 2013. [76] Block SR, Winfield JB, Lockshin MD, D'Angelo WA, Christian CL. Studies of twins with systemic lupus erythematosus. a review of the literature and presentation of 12 additional sets. Am J Med 1975;59:533–52 [PubMed PMID: 1101680]. [77] Deapen D, Escalante A, Weinrib L, Horwitz D, Bachman B, Roy-Burman P, et al. A revised estimate of twin concordance in systemic lupus erythematosus. Arthritis Rheum 1992;35:311–8 [PubMed PMID: 1536669]. [78] MacGregor AJ, Antoniades L, Matson M, Andrew T, Spector TD. The genetic contribution to radiographic hip osteoarthritis in women: results of a classic twin study. Arthritis Rheum 2000;43:2410–6 [PubMed PMID: 11083262]. [79] Kirk KM, Bellamy N, O'Gorman LE, Kuhnert PM, Klestov A, Muirden K, et al. The validity and heritability of self-report osteoarthritis in an Australian older twin sample. Twin Res 2002;5:98–106 [PubMed PMID: 11931687]. [80] Page WF, Hoaglund FT, Steinbach LS, Heath AC. Primary osteoarthritis of the hip in monozygotic and dizygotic male twins. Twin Res 2003;6:147–51 [PubMed PMID: 12724001]. [81] Michaelsson K, Melhus H, Ferm H, Ahlbom A, Pedersen NL. Genetic liability to fractures in the elderly. Arch Intern Med 2005;165:1825–30 [PubMed PMID: 16157825]. [82] Grjibovski AM, Olsen AO, Magnus P, Harris JR. Psoriasis in Norwegian twins: contribution of genetic and environmental effects. J Eur Acad Dermatol Venereol 2007;21:1337–43 [PubMed PMID: 17958839]. [83] Pedersen OB, Svendsen AJ, Ejstrup L, Skytthe A, Junker P. On the heritability of psoriatic arthritis. Disease concordance among monozygotic and dizygotic twins. Ann Rheum Dis 2008;67:1417–21 [PubMed PMID: 18218666]. [84] Brown MA, Kennedy LG, MacGregor AJ, Darke C, Duncan E, Shatford JL, et al. Susceptibility to ankylosing spondylitis in twins: the role of genes, HLA, and the environment. Arthritis Rheum 1997;40:1823–8 [PubMed PMID: 9336417]. [85] Pedersen OB, Svendsen AJ, Ejstrup L, Skytthe A, Harris JR, Junker P. Ankylosing spondylitis in Danish and Norwegian twins: occurrence and the relative importance of genetic vs environmental effectors in disease causation. Scand J Rheumatol 2008;37:120–6 [PubMed PMID: 18415769]. [86] Lin SY, Hsieh SC, Lin YC, Lee CN, Tsai MH, Lai LC, et al. A whole genome methylation analysis of systemic lupus erythematosus: hypomethylation of the IL10 and IL1R2 promoters is associated with disease activity. Genes Immun 2012;13:214–20 [PubMed PMID: 22048455].