The role of epigenetic variation in the pathogenesis of systemic lupus erythematosus

The focus of the present review is on the extent to which epigenetic alterations influence the development of systemic lupus erythematosus. Lupus is a systemic autoimmune disease characterized by the production of autoantibodies directed at nuclear self-antigens. A DNA methylation defect in CD4+ T cells has long been observed in idiopathic and drug-induced lupus. Recent studies utilizing high-throughput technologies have further characterized the nature of the DNA methylation defect in lupus CD4+ T cells. Emerging evidence in the literature is revealing an increasingly interconnected network of epigenetic dysregulation in lupus. Recent reports describe variable expression of a number of regulatory microRNAs in lupus CD4+ T cells, some of which govern the expression of DNA methyltransferase 1. While studies to date have revealed a significant role for epigenetic defects in the pathogenesis of lupus, the causal nature of epigenetic variation in lupus remains elusive. Future longitudinal epigenetic studies in lupus are therefore needed.

and organ systems of the human body. In eff ect, epigenetic modifi cation facilitates the diff erentiation of diverse cell types from static genomic DNA. Th e possibility for epigenetic variation underlying the disease process of lupus is intriguing as patients experience periods of calm punctuated by disease fl ares. Because epigenetic states vary with time and between cell types, it is in the context of disease activity that epigenetic variation in lupus is likely to stand in sharpest relief against the epigenome of normal tissues.

Epigenetic regulation in normal lymphocytes
Th e human immune system is composed of a number of cell types that serve specifi c functions in establishing and maintaining cell-mediated and humoral immunity. Th e diversity of lymphocyte populations that comprise the human immune system is established and maintained through epigenetic modifi cation. Distinct and highly conserved epigenetic profi les of DNA methylation are observed in normal B and T lymphocytes [10,11]. While DNA methylation patterns are broadly conserved within cell types, widespread diff erences are observed between lymphocyte subtypes. DNMT1 function and DNA methy lation patterns are essential in ensuring T cell diff er entiation and function [12]. DNA methylation and histone modifi cations reinforce repressive and permissive chromatin states at key regulatory loci in CD4+ T cells lineages [13,14].
Demethylation and changes in chromatin architecture at transcription factor and cytokine genes accompany eff ector cell diff erentiation in response to cytokine signaling among lymphocyte populations. Chromatin remodel ing of the IL17 locus is observed upon diff erentiation of Th 17 cells [15]. Likewise, DNA demethylation and histone acetylation of the IFNG (IFN-γ) and the IL4-IL5-IL13 genetic loci, upon Th 1 and Th 2 diff erentiation, respectively, is mediated by T cell lineage-specifi c transcription factors following T-cell receptor signaling [16]. Indeed, Th 2-specifi c chromatin remodeling and demethyla tion takes place in a locus control region that fl anks a number of Th 2 cytokines [17]. Further, these epigenetic changes are accompanied by changes in long-range interactions and chromatin architecture at cytokine gene loci under Th 1 and Th 2 polarizing conditions [18].

Drug-induced lupus points to epigenetic dysregulation in autoimmunity
While approximately 90% of lupus cases are idiopathic, prolonged exposure to certain medications is known to induce lupus. It was eventually discovered that procainamide and hydralazine, medications that can cause lupus, inhibit DNA methylation in T cells [19]. It was further shown that in vitro treatment of Th 2 cells with procainamide or 5-azacytidine (a DNA methylation inhibitor) leads to their autoreactivity, and upon adoptive transfer these cells are capable of inducing autoimmunity in mice [20].
A better understanding of the mechanisms through which drug-induced lupus occurs has provided insight into the epigenetic mechanisms and pathways involved in idiopathic lupus. Global hypomethylation of genomic DNA is observed in CD4+ T cells of lupus patients [21]. Th is decline is attributable to a reduction in methyltransferase activity. Procainamide acts as a competitive inhibitor of DNMT1 [22]. Hydralazine reduces DNA methylation via inhibiting extracellular signal-regulated kinase (ERK) pathway signaling [23], which regulates the expression of DNMT1 [24,25]. Th e ERK pathway further ensures proper immune cell development and function [26].

Reduced ERK pathway signaling in lupus CD4+ T cells
An ERK pathway signaling defect has been observed in lupus CD4+ T cells [27,28] and has been attributed to originate from decreased signaling through protein kinase C delta in lupus CD4+ T cells [29]. Decreased protein kinase C signaling may further account for defective T cell activation in lupus [30]. Additionally, mice lacking the protein kinase C delta gene display germinal center formation in the absence of stimulation and develop autoimmunity [31].
A reduction of ERK pathway signaling leading to reduced expression of DNMT1 has been shown to lead to T cell autoreactivity. Mice that express a dominant-negative form of the ERK pathway component MEK show reduced ERK pathway signaling and reduced expression of DNMT1. Th is leads to overexpression of methylationsensitive genes, anti-double stranded DNA antibody produc tion, and activation of IFN-regulated signaling pathways similar to lupus patients [32].

Sex-chromosome complement and skewed X chromosome inactivation in lupus
Lupus is a sexually dimorphic autoimmune disease with a 9:1 female bias, a fact that makes sex chromosome complement and skewed X chromosome inactivation interesting candidates for investigating the origins of the disease. Under normal circumstances, females have only one X chromosome from which genes are actively transcribed, while the additional X chromosome remains transcriptionally inactive through epigenetic repression. As previously stated, widespread demethylation of genomic DNA is observed in lupus CD4+ T cells, and this, it is thought, might lead to the reactivation of methylation-sensitive genes on the normally inactive X chromosome. An example of the reactivation of a gene on the inactive X chromosome in lupus CD4+ T cells is that of CD40 ligand (CD40LG). CD40LG is a B-cell costimulatory molecule that, upon binding with CD40, promotes B cell activation, plasma cell diff erentiation, immunoglobulin production and antibody class switching [33,34]. Overexpression of CD40LG is observed in lupus CD4+ and CD8+ T cells as well as lupus B cells [35]. CD4+ T cells from women but not men overexpress CD40LG when treated with 5-azacytidine, while the overexpression of autosomal methylation-sensitive genes is comparable between the sexes [36]. CD40LG-transfected normal human T cells lead to autologous B cell activation and plasma cell diff erentiation in co-culture assays [37]. Further, T cells from women but not men increase B cell IgG production when treated with DNA methylation inhibitors, which can be inhibited by anti-CD40LG antibody [37].
Th ere is evidence for skewed X chromosome inactivation in a number of other autoimmune diseases apart from lupus, including autoimmune thyroiditis and scleroderma, many of which also have a strong female bias [38]. Further evidence for the role of sex chromosome complement in the pathogenesis of lupus comes from Klinefelter's syndrome (47, XXY), a condition in which phenotypic male individuals possess an additional X chromosome. It was found that lupus is approximately 14 times more frequent among men with Klinefelter's syndrome than it is among (46, XY) men, a rate similar to that observed among women [39].

Aberrant methylation and gene expression in lupus CD4+ T cells
Global reduction of DNA methylation is a hallmark of CD4+ T cells in lupus ( Figure 1). Several autosomal genes are known to be hypomethylated in lupus CD4+ T cells. Many of these genes encode proteins involved in processes related to immune function and infl ammation. It is thought that the overexpression of methylation-sensitive genes leads to lupus T cell autoreactivity and subsequent induction of autoreactive B cell immunoglobulin production.
Lymphocyte function-associated antigen 1 (LFA-1) is a multi-unit, integrin family protein involved in lymphocyte adhesion that is composed of the products of two genes, CD11a and CD18. DNA methylation and local chromatin architecture regulate the tissue-specifi c expres sion of CD11a [40]. Promoter hypomethy lation of the ITGAL (CD11a) gene is observed in lupus CD4+ T cells [41]. Th is leads to overexpression of LFA-1 among an autoreactive T-cell subset. Further, ectopic expression of CD18 recapitulates the eff ects of treatment with any of the multiple DNA methylation inhibitors in inducing Tcell autoreactivity [42]. LFA-1 overexpression is a driver of autoreactivity directly linked to reduced levels of DNA methylation. CD70 (TNFSF7) is a membrane bound, B-cell costimulatory protein involved in regulating B-cell immunoglobulin production. CD70 is overexpressed in lupus CD4+ T cells and T cells treated with DNA methylation inhibitors, and this overexpression stimulates B cell IgG production in vitro [43]. Indeed, the promoter sequence of CD70 that demethylates upon treating normal CD4+ T cells with DNA methylation inhibitors is similarly demethylated in lupus CD4+ T cells [44]. Reduced methylation of a CpG island in the CD70 promoter was also observed to correspond with higher CD70 expression and reduced DNMT1 mRNA levels in MRL/lpr mice at 16 weeks when autoimmunity is established [45]. Over expression of CD70 in response to reduced promoter methylation is also observed in patients with subacute cutaneous lupus erythematosus. Increased expression was observed at both the mRNA and protein level by fl ow cytometry, and although the overall proportion of CD70-expressing cells was unchanged, the mean fl orescent intensity was greater among subacute cutaneous lupus erythematosus patient samples [46].
Many transcription factors are known to infl uence local chromatin architecture through specifi c interaction with DNA and histone methyltransferases. Decreased levels of RFX1 at CD11a and CD70 promoters leads to decreased methylation and increased expression of both of these genes in lupus CD4+ T cells [47]. Moreover, this change in DNA methylation and expression is reinforced by concomitant reductions in histone 3 lysine 9 trimethylation (H3K9me3) at these promoters in a RFX1-SUV39H1 interaction-dependent manner [48].
A role for the Gadd45A (growth arrest and DNAdamage-inducible 45 alpha) gene in autoimmunity has long been suggested. Initial knock-out studies of Gadd45a reported by Salvador and colleagues revealed a protective role for the gene product, as knockout Gadd45A-/-mice grew to develop a lupus-like systemic autoimmune disease [49]. More recently, transfection of HEK293 cells with a Gadd45A overexpression vector was shown to result in both site-specifi c and global DNA demethylation. Further, increased expression of Gadd45A following exposure to ultraviolet light was shown to lead to reduced genomic DNA methylation, a trend partially counteracted by knockdown of Gadd45A [50]. Th ese fi ndings were recapitulated in recent work in CD4+ T cells in which overexpression of Gadd45A and exposure to UV-B light radiation led to increased expression and promoter demethylation of CD70 and CD11a [51].
Perforin is a protein encoded by the PRF1 gene, expressed primarily by CD8+ T lymphocytes and natural killer cells. Perforin acts to promote target cell lysis by promoting the formation of pores in the target cell membrane. Initially, overexpression of perforin in T cells was observed in response to treatment with DNA methylation inhibitors. Promoter demethylation and perforin overexpression were accompanied by changes in chromatin architecture as evidenced by the gain of a DNAse hypersensitive site upstream of the PRF1 transcription start site [52]. Further, this trend of PRF1 promoter demethylation leading to increased perforin expression was then confi rmed in lupus CD4+ T cells, where increased perforin was observed via fl ow cyto metry in CD4+ T cells of lupus patients with active disease [53].
Killer immunoglobulin-like receptors (KIRs) comprise a diverse set of polymorphic genes involved in HLA type I stimulatory and inhibitory signals in natural killer cells and CD4+ CD28-T cells [54]. Expression of KIR genes is regulated by methylation of their promoter regions, and There is overlap between the DNA methylation defect in lupus and other epigenetic mechanisms. Overexpression of miR-148a and miR-126 aff ect DNA methylation levels by directly targeting DNMT1 transcripts. miR-21 overexpression also aff ects DNA methylation; however, this microRNA indirectly regulates DNMT1 expression through ERK pathway signaling. Variable expression of other individual regulatory microRNAs also contributes to the disease process. Overexpression of miR-17-92 in murine T lymphocytes is observed and may aff ect lymphocyte stability. miRNA-155 is overexpressed in CD4+ T cells and contributes to reduced stability of regulatory T cell populations. miR-31 negatively regulates the expression of Foxp3 and has been observed to be overexpressed in murine splenocytes. Underexpression of miR-146 leads to increased type I IFN signaling in lupus CD4+ T cells. Reduced expression of miR-125a leads to increased expression of the infl ammatory chemokine RANTES by reduced targeting of KLF13 transcripts. Overexpression of miR-101 has been reported in murine T lymphocytes. miR-101 has also been reported to negatively regulate EZH2, a histone methyltransferase involved in epigenetic repression. Aberrant patterns of histone modifi cation are also observed in lupus CD4+ T cells. Reduced levels of histone 3 lysine 9 trimethylation (H3K9me3) are observed as well as reduced p300 histone acetyltransferase activity. Further, EZH2 levels are reduced in lupus CD4+ T cells. Reduced levels of the transcription factor RFX1 lead to loss of local epigenetic repression characterized by reduced levels of DNA methylation and H3K9me3 via the histone methyltransferase SUV39H1. treatment of T cells with 5-azacytidine leads to activation of a number KIR genes [55]. Interestingly, overexpression of KIR genes is also observed in lupus CD4+ T cells, and this overexpression is seen to correspond to disease activity [56].

DNA methylation in lupus B cells
Several genes are observed to be diff erentially regulated in lupus B cells. Surface expression of CD5, which is important in maintaining B cell anergy [57], is known to be reduced in lupus B cells. IL-6 has been observed to alter CD5 expression levels in lupus B cells by inhibiting DNA methylation [58]. Increased exposure to IL-6 leads to ERK pathway inhibition, decreased DNMT1 levels, and increased expression of CD5-1EB, the cytoplasmic form of CD5, which in turn limits the expression of CD5 on the cell surface. Th is trend was observed in lupus B cells and confi rmed among B cells from healthy controls exposed to increased levels of IL-6 or treated with DNA methylation inhibitors [58].

Widespread DNA methylation diff erences between lupus patients and healthy controls
Studies in disease-discordant monozygotic twins are useful when investigating epigenetic aspects of complex human disease. Monozygotic twin studies enable the control of many of the confounding infl uences of genetic variation, although there is evidence for copy number variation among monozygotic twins [59], while focusing on disease-relevant, inter-individual epigenetic variation. Monozygotic twins display signifi cant life course variation in epigenetic modifi cations, and this variation, it is thought, may drive disease pathogenesis [60]. A recent twin study in lupus revealed signifi cant epigenetic variation in white blood cells between disease-discordant monozygotic twins [61]. Th e study reported and validated a number of genes found to possess promoter methylation diff erences. Specifi cally, 49 genes were found to be hypomethylated in lupus patients, highlighting ontologies relevant to autoimmunity. In addition, variable methylation of ribsomal genes was reported whereby 18S and 28S ribosomal subunit genes are hypomethylated in lupus individuals of discordant pairs and hypomethylated in both individuals of concordant lupus twin pairs examined. A recent case-control methylation study utilizing highthroughput methylation arrays identifi ed a number of diff erentially methylated genes and pathways in lupus CD4+ T cells. Th e methylation status of over 27,000 CpG sites in promoter regions of nearly 15,000 genes was collected and analyzed for 12 female lupus cases and 12 controls. In the study, Jeff ries and colleagues report hypo methylation of 236 CpG sites and hypermethylation of another 105 (representing 232 and 104 genes, respectively). Among hypomethylated genes, ontology analysis highlighted genes involved in connective tissue development, including ADAMTS1, ALX4, CD9, ESRRA, FGF8, HOXA13, HOXD11, MMP9, MSX1, PDGFRA, and SOX5 [62]. Hypermethylation of genes involved in folate biosynthesis, a pathway related to maintenance of DNA methy lation, was observed [62]. Interestingly, RUNX3, which belongs to a transcription factor family involved in mediating cell proliferation signals [63], was found to be hypermethylated in lupus CD4+ T cells. Protein interaction network analysis also revealed diff erential methylation of several apoptosis-associated genes, and genes involved in cell growth, tissue development, and cell division. Th is study also provided data to support a potential role for some specifi c DNA methylation changes as novel biomarkers for lupus disease activity, with the caveat that these need to be fi rst validated and then examined in the setting of a clinical trial before being considered of any clinical benefi t ( Figure 2).

Variable regulation of microRNAs in lupus
MicroRNAs represent a post-transcriptional regulatory mechanism whereby single-stranded RNAs 20 to 24 bp in length homologously bind to mRNA transcripts, eff ectively blocking their subsequent translation [64]. MicroRNA regulatory networks work to fi ne-tune immune function. Further, diff erential regulation of microRNAs appears to play a role in the disease process of lupus.
MiR-146, which plays a role in negatively regulating innate immune responses [65], has been observed to be intrinsically underexpressed in lupus CD4+ T cells. Reduced expression of miR-146 leads to induction of type I interferon signaling, and miR-146 expression levels negatively correlate with disease activity. Interestingly, IFN signatures of lupus patient peripheral blood mononuclear cells were alleviated upon transfection with a miR-146 expression vector [66]. Zhao and colleagues observed targeting of the 3' UTR of the KLF13 gene by miR-125a in a dose-dependent manner. Underexpression of miR-125a was also reported in lupus CD4+ T cells leading to increased levels of the infl ammatory chemokine RANTES [67]. Normalization of miR-125a expression in CD4+ T cells of lupus patients resulted in reduced RANTES levels.
Investigation of microRNA patterns in three models of murine lupus reveals common microRNA pathways. Th e study investigated common defects in regulatory microRNAs among splenic T and B lymphocytes in B6/ lpr, MRL/lpr, and NZB/W mice when compared with their respective age-matched control mice. Dai and colleagues report a number of regulatory microRNAs, including miR-155, miR-150, miR-182-96-183, miR-31, miR-127, and miR-379, to be signifi cantly increased in splenic B and T cells of MRL/lpr mice. Signifi cantly, upregulation of miR-146, miR101a, and miR-17-92 in MRL/lpr mice was observed exclusively among T lymphocytes [68]. Mice with increased miR-17-92 expres sion display uncontrolled expansion and auto immunity characterized by reductions in the pro apoptotic protein Bim and the tumor suppressor PTEN (phosphatase and tensin homolog) [69]. Interestingly, genomic loss of miR-101 has been associated with over expression of the repressive histone methyl trans ferase EZH2 in human prostate cancer cells [70]. EZH2, the histone methyl trans ferase that mediates H3K27me3, functionally asso ciates with DNMT1 [71]. Overexpres sion of miR-101 might contribute to the loss of repression observed in lupus CD4+ T cells. Moreover, down regulation of EZH2 in CD4+ T cells has been reported in lupus patients with active disease [72].
A systematic, microarray-based investigation of micro-RNA expression in Epstein-Barr virus-transformed B cell lines of lupus patients with nephritis and controls in  [61] performed a genome-wide methylation scan investigating epigenetic diff erences in fi ve pairs of monozygotic twins discordant for lupus. The study investigated these diff erences in white blood cells (WBCs) of patients and their related controls. The study reported variable methylation of 49 of the 807 genes investigated. Pathway analysis of these genes reveals putative functions in immune response, cell activation, and cell proliferation. The study further identifi es hypomethylation and overexpression of ribosomal genes, which may correspond to increased ribosomal-autoantibody formation. (b) Jeff ries and colleagues [62] investigated the methylation status of over 27,000 individual CpG dinucleotides located in promoter regions of nearly 15,000 genes. Twelve lupus cases of varying disease activity and healthy control individuals were included in the study. The study reports hypomethylation of CpG dinucleotides in 232 genes and the hypermethylation of 104 genes. Pathway analysis reveals genes involved in folate biosynthesis, a pathway related to maintenance of DNA methylation. Further, modifi cations of individual genes observed to be dysregulated included the hypomethylation of CD9, encoding a known activator of T cell signaling, and hypermethylation of RUNX3, which encodes a transcription factor that mediates profi leration signals in lymphocytes. In both of the diagrams on the left side of each panel, empty red circles represent unmethylated CpG dinucleotides, whereas solid blue circles represent methylated CpG sites. In the lower panel, the diff erences in color of the gene body indicate the presence of genetic heterogeneity. Caucasian and African-American populations observed reproducible overexpression of a number of microRNAs across ethnicities [73]. Th is study also included a discordant monozygotic twin pair in which the broader microRNA expression patterns were partially conserved.
More recently, analysis of microRNA expression profi les in lupus peripheral blood mononuclear cells compared to controls revealed the diff erential expression of 27 microRNAs out of 365 analyzed [74]. Importantly, the levels of miR-21, miR-25, miR-106b and miR-148b correlated with disease activity in lupus patients, while miR-196a and miR-379 expression inversely correlated with disease activity, suggesting a potential role for microRNA profi ling as disease biomarker in lupus [74].
A recent comprehensive review on the role of micro-RNA in the pathogenesis of lupus and other autoimmune disease has been recently published [75].

Crosstalk between microRNAs and DNA methylation in lupus
Targeting of DNA methyltransferases by regulatory micro-RNAs is observed in both development and disease. MicroRNAs appear to play a critical role in regu lation of de novo methylation in mouse embryonic stem cells [76]. Targeting of DNMT3a and DNMT3b by miR-29 family members is observed in human lung cancer [77]. Two recent studies demonstrate down regulation of DNMT1 as a result of variable expression of regulatory microRNAs in lupus.
Pan and colleagues [78] demonstrated a connection between aberrant regulation of certain microRNAs and reduced expression of DNMT1. A genome-wide micro-RNA scan in both lupus patients and MRL/lpr lupusprone mice revealed increased expression of both miR-21 and miR-148a. miR-21 targets and inhibits RSGRP1, a signaling molecule involved in the ERK signaling pathway, through which it is able to indirectly downregulate DNMT1 expression. MiR-148a, however, directly targets the DNMT1 transcript.
MiR-126 is among a number of microRNAs observed to be variably regulated in lupus CD4+ T cells. Zhao and colleagues [79] report overexpression of miR-126 in lupus CD4+ T cells, which correlates with decreased DNMT1 expression. Th ey further demonstrate that miR-126 targets a 3' UTR region in DNMT1 and reduces DNMT1 protein expression. Furthermore, ectopic overexpression of miR-126 in CD4+ T cells from healthy donors led to hypomethylation and overexpression of the methylation sensitive genes CD11a and CD70 and induction of T cell autoreactivity [79].

Variable histone protein modifi cation in lupus
Covalent modifi cation of histone protein represents another layer of epigenetic regulation. Acetylation of histone tails leads to an open conformation of chromatin that is permissive to gene expression. Th e acetyl transferase activity of p300 in B cells is crucial in preventing autoimmunity. Mice with a B cell-specifi c knock-in mutant version of p300 that lacks only histone acetyltransferase activity develop a lupus-like autoimmune disease characterized by splenomegaly, nephritis, and vasculitis [80]. Reduced global levels of H3K9me3 and histone 3 acetylation have been reported in lupus CD4+ T cells, and diff erences in the activity of histone acetyltrans ferases have been observed in lupus CD4+ T cells [72].
Sullivan and colleagues [81] initially reported variability in histone 4 acetylation at the tumor necrosis factor alpha (TNF-α) locus in lupus monocytes. Th is same group further studied histone 4 acetylation at the genome level in lupus monocytes where they found hyperacetylation and overexpression of several genes with modest overlap with the acetylation and expression patterns of IFN-α treated cells [82]. Th ese fi ndings point to widespread changes in chromatin architecture in lupus monocytes.

The altered epigenetic state of regulatory T cell populations in murine lupus
Foxp3+ CD25high regulatory T cells broadly serve to suppress autoimmunity. Expression of the transcription factor Foxp3 regulates the diff erentiation of the regulatory T cell lineage [83]. Mice that lack Foxp3 display a paucity of regulatory T cells and subsequently develop fatal autoimmunity [84]. Epigenetic modifi cations regulate the emergence and durability of regulatory T cells. Demethylation of the Foxp3 promoter accompanies the emergence of the regulatory T cell lineage in both mice and humans [85,86]. Foxp3 expression is also governed post-transcriptionally by microRNAs. Specifi cally, miR-31 is a negative regulator of Foxp3 expression, while miR-21 positively regulates Foxp3 [87].
Regulatory T cells of lupus patients with active disease are reduced in quantity commensurate to disease severity and display decreased suppressive function [88,89]. Divekar and colleagues [90] have recently reported a regulatory T cell defect of murine lupus whereby regulatory T cells exhibit reduced suppressive potential in MRL/lpr mice that is further characterized by decreased dicer expression and increased miR-155 expression. Interestingly, miR-155 plays a critical role in the regulation of germinal center formation and is critical in maintaining proper immune function [91,92]. Divekar and colleagues suggest that increased expression of miR-155 may account for the regulatory T cell defect observed in MRL/lpr mice. Th ese fi ndings further coincide with previous reports in which dicer defi ciency among regulatory T cells led to diminished regulatory T cell populations and autoimmune pathology [93,94].

Epigenetic variation in lupus: cause or consequence?
Th is review off ers clear evidence of the involvement of epigenetic variation in lupus. Th e studies mentioned herein describe widespread lupus-associated epigenetic variation, yet association and causation are not interchangeable terms. Many questions regarding the causal nature of epigenetic variation in lupus remain to be answered. At what stage of disease are epigenetic manifestations fi rst present? In what cell type(s) do epigenetic perturbations fi rst appear in lupus? Does epigenetic variation merely refl ect prior canalization of disease, or are these modifi cations ultimately causal? Is diseaseassociated epigenetic variation a proximate or an ultimate cause of disease?
Th e causal nature of epigenetic variation in lupus is a complicated question. In our view, large longitudinal cohort studies that will allow the epigenome to be examined before and after disease will be crucial to answer this question. Of course, these types of studies will be complicated to design, and samples from specifi c cell types will have to be maintained and subsequently examined. Th e evidence that inhibiting T cell ERK pathway signaling using a mouse model results in decreased DNMT1 expression, overexpression of methylation sensitive genes similar to lupus patients, the development of anti-double stranded DNA antibody, and an interferon signature-like gene expression profi le represents the strongest argument for a causal role of DNA methylation in lupus to date [32]. Recent studies have shown that epigenetic variation in lupus occurs along multiple axes. Unraveling the causal web of lupus epigenetics has become an increasingly complicated task as recent studies have revealed a complex, interconnected network of epigenetic regulation that leads to altered gene expression in lupus. Th ese studies pose new questions about the nature of complexity of regulatory events involved in lupus pathogenesis. Future epigenetic studies
investigating the origins of complex human disease will therefore need to incorporate multiple-level systems modeling and complexity science to uncover disease etiology in complex and dynamic biologic systems. Determining the environmental-epigenetic interactions that contribute to lupus is important in understanding the ultimate cause of lupus. Th ere is much reference made to environmental factors in the lupus epigenetics literature, yet these typically refer to a few examples or are invoked as an unknown explanatory variable. With a few notable exceptions, the specifi c mechanisms through which environmental triggers operate remain largely uncharacterized to date. Further characterization of the putative environmental-epigenetic tipping point between active and inactive disease is therefore needed.

Conclusion
Studies to date have revealed a substantial epigenetic component of disease in lupus, particularly in T cells (Table 1, Figure 1), yet ultimately the causal nature of disease associated epigenetic variation remains ambiguous. Although the etiology of lupus is indeterminate, some generalizations about the origins of lupus can be made: fi rst, there is a signifi cant genetic component to the disease; second, known genetic variation does not comprehensively describe the origin, development, and even the heritability of the disease; third, epigenetic variation correlates with variable expression of a number of genes in pathways involved in disease pathogenesis; and fourth, lupus is the result of complex interactions between genetic susceptibility, environmental exposure, and epimutation. Th e path forward in understanding lupus is not merely in the clear delineation of the various components of disease but in understanding emergent phenotypes that result from their interaction over time.