The Interaction Between Environmental Triggers and Epigenetics in Autoimmunity

Abstract

Systemic lupus erythematosus flares when genetically predisposed people encounter environmental agents that cause oxidative stress, such as infections and sunlight. How these modify the immune system to initiate flares is unclear. Drug induced lupus models demonstrate that CD4+ T cells epigenetically altered with DNA methylation inhibitors cause lupus in animal models, and similar T cells are found in patients with active lupus. How infections and sun exposure inhibit T cell DNA methylation is unclear. DNA methylation patterns are replicated each time a cell divides in a process that requires DNA methyltransferase one (Dnmt1), which is upregulated as cells enter mitosis, as well as the methyl donor S-adenosylmethionine, created from dietary sources. Reactive oxygen species that inhibit Dnmt1 upregulation, and a diet poor in methyl donors, combine to cause lupus in animal models. Similar changes are found in patients with active lupus, indicating a mechanism contributing to lupus flares.

Introduction

Autoimmune rheumatic diseases including systemic lupus erythematosus, rheumatoid arthritis, Sjogren’s Syndrome and progressive systemic sclerosis require both a genetic predisposition and an environmental exposure for disease onset and flares, and two or more of these diseases can develop in the same patient [1]. While significant progress has been made in identifying the predisposing genes, the nature of the environmental agents and the mechanisms by which they induce autoimmunity remain incompletely understood. However, reports that drugs such as procainamide and hydralazine can induce lupus-like autoimmunity in genetically predisposed people have led to studies of how these and other environmental agents may interact with the immune system to break tolerance and cause flares of lupus and possibly contribute to flares of the other autoimmune rheumatic diseases as well. Current evidence indicates that these lupus-inducing drugs as well as environmental agents that cause inflammation may alter the immune system by epigenetically modifying gene expression in immune cells, breaking tolerance and causing disease flares. In drug-induced lupus, drugs such as procainamide and hydralazine alter T cell gene expression to break tolerance such that the cells respond to self-antigens, causing lupus-like flares [2], while oxidative stress and transmethylation micronutrient deficiencies can contribute to flares of idiopathic lupus [3]. These agents modify T cell gene expression by altering epigenetic mechanisms, and evidence for similar epigenetically modified T cells are found in patients with active lupus as well as in patients with other active autoimmune rheumatic diseases. These studies are summarized below.

DNA methylation, T cells and lupus

DNA methylation refers to the methylation of dC bases in CpG pairs, and is a repressive modification. The first evidence that altered T cell DNA methylation might contribute to the development of lupus came from studies that examined the effects DNA methylation inhibitors on T cell function. These experiments demonstrated that inhibiting the replication of CD4+ T cell DNA methylation patterns during mitosis with 5-azacytidine (5-azaC), a DNA methyltransferase 1 (Dnmt1) inhibitor, activates expression of genes normally suppressed by DNA methylation, converting normal antigen specific CD4+ T cells into autoreactive cells that respond to self class II MHC determinants alone [4]. The significance of the autoreactivity was tested by treating murine CD4+ T cells with 5-azaC then injecting the treated cells into syngeneic mice. The recipients developed a lupus-like disease, suggesting that T cell DNA demethylation might contribute to the development of human lupus [5]. Subsequent studies demonstrated that procainamide and hydralazine are also T cell DNA methylation inhibitors, that murine CD4+ T cells treated with these drugs also cause a lupus-like disease in syngeneic mice [6], and that similar epigenetically altered T cells are found in patients with active lupus [7]. The lupus T cell DNA methylation defect was found to be caused by low Dnmt1 levels, due to a failure to upregulate the enzyme as T cells enter mitosis. The signaling defect was then traced to decreased PKCδ signaling [8], caused by oxidative damage (nitration) of the protein [9].

The T cell autoreactivity caused by these agents is due to demethylation and overexpression of ITGAL (CD11a), a subunit of the LFA-1 adhesion molecule which surrounds the T cell receptor-class II MHC complex to form the immunologic synapse. The synapse normally stabilizes the T cell receptor-class II MHC complex interaction and provides a stimulatory signal [10]. Treatment of CD4+ T cells with 5-azaC also causes aberrant overexpression of other genes normally suppressed by DNA methylation. These include the B cell costimulators TNFSF7 (CD70) and CD40LG (CD40L), the adhesion molecule ITGAL (CD11a), the cytotoxic molecule perforin (PRF1), the inflammatory cytokine interferon gamma (IFNG) [11] and the killer cell immunoglobulin-like receptor (KIR) genes, which are normally expressed clonally by NK cells but not expressed by normal T cells [12]. Together these changes convert normal, antigen specific CD4+ “helper” T cells into autoreactive, cytotoxic, pro-inflammatory cells that are capable of inducing a lupus-like disease [5].

The epigenetically altered sequences have been identified for some of these genes. The region −490 to −229 bp 5′ to the transcription start site of TNFSF7, the gene encoding CD70, is methylated in non-expressing cells but not in CD70 expressing cells, and methylation of this sequence in reporter constructs suppresses CD70 expression in transfected T cells [13]. Interestingly, CD40L is encoded on the X chromosome, one of which is inactivated by DNA methylation in female T cells, so inhibiting DNA methylation in female CD4+ T cells demethylates sequences on the inactive X chromosome, increasing expression of CD40L and other X-linked genes, while inhibiting DNA methylation in male CD4+ T cells does not [14]. Together, these changes in gene expression convert normal antigen specific human and mouse CD4+ T cells into autoreactive, cytotoxic, pro-inflammatory cells that are sufficient to cause a lupus-like autoimmunity in animal models [5].

As noted above, procainamide and hydralazine, two drugs that cause a lupus-like disease in genetically predisposed people [15], also inhibit T cell DNA methylation [16], and murine CD4+ T cells treated with these drugs also become autoreactive and cause lupus-like autoimmunity when transferred into syngeneic mice, similar to 5-azaC treated CD4+ murine T cells [6]. Procainamide was found to be a competitive inhibitor of Dnmt1 enzymatic activity [17], while hydralazine inhibits the signaling molecule PKCδ, decreasing T cell ERK pathway signaling and preventing Dnmt1 upregulation during mitosis [18].

Similarly, ultraviolet light causes oxidative stress and lupus flares, and CD4+ human T cells exposed to UV light also overexpress methylation sensitive genes and become autoreactive, similar to T cells treated with 5-azaC, procainamide and hydralazine [19]. The oxidative stress generated by agents such as UV light and infections inactivates PKCδ by nitrating the protein, which prevents DNMT1 upregulation as T cells enter mitosis similar to the effects of hydralazine [8]. Related studies found the same changes in DNA methylation and gene expression in CD4+ T cells from patients with active lupus that are found in human CD4+ cells treated with Dnmt1 inhibitors [15]. Other studies revealed that T cell Dnmt1 levels as well as overall DNA methylation levels decline with age [20], as noted by the NIEHS Workshop [21], and may contribute to late onset lupus. The importance of CD4+ T cells in lupus pathogenesis is further supported by reports that anti-CD4 antibodies treat lupus in mice [22], and that lupus goes into remission as CD4+ T cell numbers decline in AIDS patients, then flares when the AIDS is treated [23].

Together these studies indicate that these agents, which trigger lupus flares, are T cell DNA methylation inhibitors, that T cells treated with these agents are sufficient to cause lupus in mice, and that the same changes in CD4+ T cell DNA methylation and gene expression are found in patients with active lupus. This suggests that other environmental agents that trigger lupus flares may also inhibit T cell DNA methylation. The mechanism(s) by which these environmental agents inhibit DNA methylation though, were incompletely understood.

The T cell DNA methylation defect in idiopathic lupus was first traced to low Dnmt1 levels [15], and subsequent studies revealed that impaired T cell ERK pathway signaling contributes to the low Dnmt1 levels by preventing Dnmt1 upregulation during mitosis [24]. As noted above, hydralazine similarly decreases T cell Dnmt1 levels by inhibiting ERK pathway signaling [18]. The relevance of decreased T cell ERK pathway signaling to lupus pathogenesis was confirmed using transgenic mice in which expression of a dominant negative MEK (dnMEK) is induced selectively in T cells by adding doxycycline (dox) to their drinking water. Giving dox to these mice activates expression of the dnMEK transgene in T cells, which decreases ERK pathway signaling, decreases T cell Dnmt1 levels, demethylates T cell DNA, and increases expression of genes normally suppressed by DNA methylation, similar to those activated by DNA methylation inhibitors in human T cells [25]. These studies also demonstrated that dox induced anti-DNA antibodies but not kidney disease in the double transgenic mice on a C57BL6 background, a strain not genetically predisposed to lupus [25]. However, giving dox to double transgenic C57BL/6 mice crossed with SJL mice, which have lupus susceptibility genes [26], caused higher titers of anti-DNA antibodies, and backcrossing them a second time with SJL mice further increased anti-DNA antibody levels. The C57BL/6xSJL mice receiving dox also developed an immune complex glomerulonephritis while the C57BL/6 mice did not [27]. These results are similar to the gene/environment interactions required for human lupus to flare.

Subsequent studies traced the ERK pathway signaling defect to PKCδ inactivation (PKCδ→ras→raf→MEK→ERK) in T cells from patients with active lupus similar to hydralazine treated T cells [8]. The importance of PKCδ inactivation in lupus flares was confirmed by transfecting T cells with a dominant negative PKCδ (dnPKCδ). The dnPKCδ decreased Dnmt1 levels, causing demethylation and overexpression of methylation sensitive T cells genes [8]. The importance of T cell PKCδ inactivation in lupus pathogenesis was further confirmed by creating a second double transgeneic mouse strain in which dox induces expression of a dnPKCδ rather than a dnMEK in T cells. These mice also developed lupus-like autoimmunity on a C57BL6 X SJL background [28]. Importantly, PKCδ null mutations cause a lupus-like disease in both humans [29] and mice [30]. The mechanism inactivating PKCδ in lupus T cells though, was unclear.

Oxidative stress, protein nitration and lupus

As noted by the NIEHS, environmental agents that cause oxidative stress can trigger lupus flares [21], and the inflammation caused by the flares can further increase oxidative stress, amplifying the effects of the initial exposure. The reactive oxygen species generated by this process cause oxidative modifications of serum proteins in patients with active lupus, including the formation of carbonyls, methionine sulfoxide and 3-nitrotyrosine (3-nitro-Tyr). The reactive oxygen species also decrease thiol and methionine levels in serum proteins from patients with active lupus. Levels of these modifications correlate directly with lupus disease activity [31].

The protein nitration caused by inflammation plays a critical role in lupus flares. Protein nitration is caused by superoxide (O₂⁻), generated during inflammatory responses and lupus flares by mechanisms including increased T cell mitochondrial hyperpolarization and oxidative metabolism [32]. Activated macrophages also generate reactive oxygen species [31]. Superoxide then reacts with nitric oxide (NO), an intracellular signaling molecule [33], to produce peroxynitrite (ONOO⁻), a highly reactive molecule that nitrates protein Tyr residues as well as lipids and DNA [34]. Importantly, serum nitrate and nitrite levels are elevated in lupus patients, and the levels correlate directly with disease activity [34]. Lupus patients with proliferative nephritis have even higher serum protein nitration levels, and the elevated levels correlate with the accumulation of renal damage and a lack of response to therapy [34]. The nitration may also be increased by inducible nitric oxide synthase (iNOS) stimulation by IFN-α [35] and IFN-γ [36], which are also increased in lupus flares [37]. Notably, estrogen also stimulates NO production by splenocytes treated with T cell mitogens [36], which may contribute to lupus flares associated with the menstrual cycle and pregnancy [38]. Evidence for oxidative stress in lupus flares is further supported by a study that used liquid chromatography/mass spectrometry (LC/MS) and gas chromatography/mass spectrometry (GC/MS) platforms to identify metabolic disturbances in patients with modestly active SLE as defined by a mean Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) [39] score of 5. The lupus metabolome was characterized by profound increases in lipid peroxidation and deficiencies in the anti-oxidant glutathione as well as deficiencies in methyl group donors including cysteine, methionine and choline, all indicative of ongoing inflammation, oxidative stress and transmethylation reaction impairment [40].

Together, these studies demonstrate ongoing generation of reactive oxygen species that cause protein nitration and other oxidative modifications of serum proteins, as well as proteins in T lymphocytes and perhaps other cells, sustaining lupus flares. Interestingly, mycophenolate mofetil, used to treat lupus flares [41], inhibits inducible nitric oxide synthase (iNOS) mediated NO production, possibly decreasing protein nitration in lupus patients in addition to causing apoptosis of activated T cells [41].

Oxidative stress, DNA demethylation and lupus

The observations that PKCδ catalytic activity is decreased in T cells from patients with active lupus [8], and that serum proteins as well as T cell proteins are nitrated in patients with active lupus [31], suggested that lupus T cell PKCδ may be inactivated by nitration. CD4+ T cells from patients with active lupus were therefore treated with phorbol myristate acetate (PMA), which stimulates autophosphorylation of PKC isoenzymes [42]. The cells were then lysed, proteins fractionated by SDS-PAGE then phospho-PKCδ levels were compared by immunoblotting. PMA stimulated greater amounts of phospho-PKCδ in CD4+ T cells from controls than lupus patients (p<0.001) [8], suggesting that environmentally-induced oxidative stress may inactivate T cell PKCδ through protein nitration to decrease ERK pathway signaling, decrease Dnmt1 levels, and demethylate DNA in lupus patients.

The relationship between oxidative stress and T cell DNA demethylation was tested by culturing human CD4+ T cells with H₂O₂ or ONOO⁻, stimulating the cells with PMA, then comparing PKCδ activation, ERK pathway signaling, Dnmt1 levels, and demethylation and overexpression of the TNFSF7 (CD70) and KIR2DL4 genes in untreated and treated cells. Hydralazine, which inhibits PKCδ [8], was included as a positive control. H₂O₂ and ONOO⁻ inhibit PMA stimulated PKCδ autophosphorylation in CD4+ T cells. PMA-stimulated Raf, MEK and ERK phosphorylation were similarly decreased by H₂O₂ and ONOO⁻, and Dnmt1 mRNA levels were decreased by ~80% in cells treated with by H₂O₂ (p=0.002) and ONOO⁻ (p=0.001) [43]. Importantly, the effects of H₂O₂ and ONOO⁻ were completely inhibited by the anti-oxidant N-acetylcysteine (p<0.04), confirming that H₂O₂ and ONOO⁻ inhibit PKCδ by mechanisms involving reactive oxygen species [43]. Similarly, N-acetylcysteine as well as other anti-oxidants decrease flare severity in lupus patients [44]. Proliferating CD4+ human T cells treated with H₂O₂ or ONOO⁻ also aberrantly overexpress the methylation-sensitive genes KIR and CD70 in a dose dependent fashion [13, 45], and that ONOO⁻ is more potent than H₂O₂ [43]. No effects on CD4 expression were seen, similar to previous reports using 5-azaC [4]. Controls included demonstrating that both H₂O₂ and ONOO⁻ demethylate the same regulatory regions in the KIR2DL4 and TNFSF7 (CD70) genes (not shown) [43]. However, while the same sequences demethylate with both agents, ONOO⁻ was again more potent than H₂O₂. This may be due to different mechanisms by which these agents inhibit T cell DNA methylation. PKCδ is inactivated by nitration in T cells from active lupus, thereby decreasing ERK pathway signaling and decreasing Dnmt1 levels [8]. ONOO⁻ directly nitrates proteins, while H₂O₂ must combine with NO to form ONOO⁻ and nitrate proteins and H₂O₂ has additional effects and toxicities such as the oxidation of cysteine thiols [46]. Together, these studies suggest that reactive oxygen species inactivate T cell PKCδ by nitration, and that PKCδ inactivation results in decreased T cell ERK pathway signaling, decreased Dnmt1 levels, and demethylation of regulatory elements and overexpression of human and mouse T cell genes normally silenced by DNA methylation, promoting lupus flares.

In confirming studies CD4+ T cells from female SJL mice were stimulated with concanavalin A, treated with 5-azaC, H₂O₂ or ONOO⁻, then injected into syngeneic recipients [47]. The mice receiving the treated cells developed anti-DNA antibodies by 12 weeks, and again ONOO⁻ is the most potent. Mice receiving 5-azaC, H₂O₂ and ONOO⁻ treated T cells, but not untreated T cells, also developed an immune complex glomerulonephritis and proteinuria (5-azaC p=0.01, ONOO⁻ p=0.01, H₂O₂ p=0.038) [48]. The observation that 5-azaC, ONOO⁻ and H₂O₂ all stimulate autoantibody formation and cause an immune complex glomerulonephritis and proteinuria support the association between environmentally induced T cell DNA demethylation and lupus flares. Other studies demonstrated that CD4+ T cells treated with DNA methyltransferase inhibitors (5-azacytidine and procainamide) or ERK pathway inhibitors also became autoreactive and stimulated antibody overexpression by autologous B cells [49]. Together these results demonstrate that CD4+ T cells experimentally demethylated with Dnmt inhibitors or with oxidizing agents are sufficient to cause a lupus-like disease in genetically predisposed mice, similar to studies using 5-azaC, procainamide or hydralazine treated T cells [47].

Diet, DNA methylation and lupus

Dietary micronutrients such as folate and methionine also play a critical role in transmethylation reactions including DNA methylation. DNA methylation patterns are established during development then are copied during mitosis by the maintenance DNA methyltransferase Dnmt1. Dnmt1 binds the replication fork and “reads” CG pairs. Where dC in the parent strand is methylated, Dnmt1 catalyzes transfer of the methyl group from S-adenosylmethionine (SAM), the methyl donor, to the corresponding dC in the daughter strand, creating deoxymethylcytosine (d^mC) and S-adenosyhomocysteine (SAH), an inhibitor of transmethylation reactions [50] i.e. dC + SAM → d^mC + SAH. Like other intracellular transmethylation reactions, the reaction velocity (V) is directly related to intracellular SAM pools and Dnmt1 levels, and is inhibited by SAH. This may be written as V=k[SAM][Dnmt1]/]SAH], where k is a constant to correct for units [51]. This equation implies that when Dnmt1 levels are low, as they are in T cells from patients with active lupus [52], SAM levels will need to increase and/or SAH levels decrease, in order to maintain efficient copying of DNA methylation patterns in dividing cells as Dnmt1 progresses along the DNA strands. Importantly, as noted above, T cell Dnmt1 levels are low in patients with active lupus due to oxidative PKCδ inactivation, and others have reported that lupus patients have high homocysteine (Hcy) levels [53], suggesting synergistic effects.

The interaction between Dnmt1 levels, Hcy and folic acid on methylation sensitive T cell gene expression was confirmed in an aging study. CD4+ T cell Dnmt1 levels [51] and total d^mC content [54] normally decline with age, causing overexpression of the same genes as those overexpressed by T cells from lupus patients [51]. PBMC from healthy people ages 21–75 were stimulated with PHA then cultured in custom media containing folic acid at normal (40 nM) or low (10 nM) serum concentrations, with or without 15 μM Hcy (normal 5–15 uM). KIR expression was then measured on T cells. KIR expression increases as Dnmt1 levels decline with age, and that borderline high (15 μM) Hcy levels further increases KIR expression in T cells from older people when folate levels are low [51]. In this study aging also sensitized CD4+ T cells from the older people to low Met concentrations (not shown) [51]. Dnmt1 siRNA “knockdowns” also sensitize CD4+ T cell gene expression to low Met levels. Similar results were seen with low folate levels in the same experiments [51]. Others have also reported lifestage-specific effects of diet on the epigenome (reviewed in [55]), including changing the coat color of mice during embryonic development [56].

The interactions between low transmethylation micronutrient concentrations and gene expression in lupus were then tested by culturing PHA stimulated CD4+ T cells from lupus patients and healthy controls in custom RPMI 1640 media with the usual media Met concentration (101 μM), with normal serum Met levels (30 μM) or with low Met levels (5 μM). Perforin expression was measured 3 days later by flow cytometry. T cells from lupus patients, with low Dnmt1 levels [57], express higher levels of perforin than controls when cultured in media with low Met levels (p=0.008). The pathologic consequences of the interactions between diet, Dnmt1 levels and lupus were then tested in a double transgenic mouse model where Dnmt1 is decreased selectively in T cells by adding dox to their drinking water. Gene expression in mice with normal T cell Dnmt1 levels was not significantly affected by a diet low in transmethylation micronutrients. However, when Dnmt1 levels were decreased by adding dox to their drinking water, anti-DNA antibody titers were increased by a diet poor in the micronutrients, while anti-DNA titers decreased in mice receiving a diet enriched in the micronutrients. The same study found that the methyl-poor diet increased the severity of glomerulonephritis and hematuria in these mice, while the enriched diet decreased glomerulonephritis severity and hematuria [58]. Others have reported that dietary interventions can prevent or even reverse pathologic epigenetic changes in other chronic inflammatory disorders such as allergy, asthma, obesity, type 2 diabetes, cardiovascular disease and cancer (reviewed in [55]).

More recently transmethylation micronutrient levels were measured in serum from 28–35 patients with mildly active lupus (mean SLEDAI 3). Significant decreases in Zn, B6 and methionine levels were found, as well as the increased Hcy levels noted by others [53] (manuscript submitted). Deficiencies in transmethylation micronutrients including cysteine, methionine and choline were also noted by Mohan et al in patients with somewhat more active lupus (mean SLEDAI 5) [40]. Together, these studies suggest that low Dnmt1 levels, caused by oxidative inactivation of PKCδ and perhaps other molecules during inflammatory responses, together with a diet poor in transmethylation micronutrients and elevated Hcy levels, may combine to promote T cell DNA demethylation, triggering lupus flares in genetically predisposed people. These studies also suggest that anti-oxidants together with transmethylation micronutrient supplementation may help prevent lupus flares and possibly decrease flare severity. Figure 1 summarizes how O₂⁻ generated by inflammatory responses may interact with estrogen-enhanced NO to form ONOO⁻, which inactivates T cell PKCδ and decreases Dnmt1 levels, and how the low Dnmt1 levels interact with a methyl-poor diet and increased Hcy levels to promote T cell DNA demethylation and lupus flares in genetically predisposed people. Figure 1 also shows that the lupus-inducing drug hydralazine decreases T cell Dnmt1 levels by inhibiting PKCδ and ERK pathway signaling [8] while procainamide, another lupus-inducing drug, is a competitive inhibitor of Dnmt1 activity [17, 59], and that T cell Dnmt1 levels decline with age [51], which may contribute to lupus onset in older, genetically predisposed people [60].

Figure 1.

Mechanism by which inflammation, estrogen, diet, and aging may contribute to T cell DNA demethylation and lupus flares in genetically predisposed individuals.