Demethylation of CD40LG on the Inactive X in T Cells from Women with Lupus¹

Systemic lupus erythematosus is a female predominant disease characterized by autoantibody formation and immune complex deposition. Genetic elements predispose to lupus, but incomplete concordance in identical twins, and reports that UV light and drugs like procainamide and hydralazine trigger a lupus-like disease, indicate an environmental contribution (1).

DNA methylation is an epigenetic modification that suppresses gene expression (2). CD4⁺ T cell DNA methylation is impaired in human lupus, due to decreases in DNA methyltransferase expression and activity (3, 4), causing genome-wide DNA hypomethylation and aberrant gene overexpression that contributes to autoantibody formation. The lupus-inducing drugs procainamide and hydralazine as well as UV light are DNA methylation inhibitors, and the same regulatory elements demethylate in T cells treated with these drugs as in CD4⁺ T cells from lupus patients, causing overexpression of autosomal genes that contribute to autoimmunity (5, 6, 7). Furthermore, CD4⁺ T cells treated with DNA methylation inhibitors including procainamide and hydralazine cause lupus-like autoantibodies and kidney disease in murine models (8). These observations suggest that DNA hypomethylation may be a mechanism contributing to the development of idiopathic and some forms of drug-induced human lupus.

Idiopathic lupus afflicts women nine times more often than men (9). The reason for the female predominance is unknown. DNA methylation also contributes to the silencing of one X chromosome in women (10). We hypothesized that autoantibody promoting genes on the inactive X may also demethylate in CD4⁺ T cells, causing greater expression in women than men with lupus. CD40LG is a B cell costimulatory molecule encoded on the X. CD40LG is overexpressed on lupus lymphocytes, contributing to overproduction of pathogenic autoantibodies (11, 12, 13, 14). We therefore characterized the role of DNA methylation in silencing CD40LG on the inactive X, and determined whether CD40LG on the inactive X demethylates and is overexpressed uniquely in women with lupus.

Healthy subjects were recruited by advertising. The average age of the men was 38 ± 13 years and 41 ± 14 for the women (mean ± SD). Lupus patients met criteria for lupus (15) and were recruited from the Michigan Lupus Cohort. Disease activity was quantitated using the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI)³ (16). The average age of the lupus women was 39 ± 11 and the men were 44 ± 10 (mean ± SD). Patient demographics and medications are shown in Table I. The protocols were reviewed and approved by the University of Michigan Institutional Review Board.

T cells were isolated, PHA stimulated and treated with 5-azacytidine (5-azaC) as previously described (7). For CD40L expression, the cells were then restimulated with 5 ng/ml PMA and 500 ng/ml ionomycin for an additional 6 h (17). Where indicated, the cells were fractionated by negative selection into CD4⁺ and CD8⁺ subsets using magnetic beads and protocols provided by the manufacturer (Miltenyi Biotec).

Cells were stained with anti-CD40L-FITC, anti-CD8-allophycocyanin, and anti-CD3-PE then analyzed by flow cytometry using published protocols (7). CD3⁺CD8⁺ cells are referred to as CD8⁺. Because CD4 expression diminishes following PMA and ionomycin stimulation (17), the CD3⁺CD8⁻ population is referred to as the CD4⁺ population.

CD40L, CD70, and β-actin transcripts were measured in magnetic bead purified CD4⁺ and CD8⁺ T cells using a Rotor-Gene (Corbett) thermocycler. CD70 and β-actin primers and amplification were as published (5, 7). CD40L primers were: forward: CACCCCCTGTTAACTGCCTA, reverse: CTGGATGTCTGCATCAGTGG, and FoxP3 primers were forward: CAAATGGTGTCTGCAAGTGG, reverse: CACAGATGAAGCCTTGGTCA.

T cell DNA was isolated and bisulfite treated using published protocols (18), then the promoter and enhancer were amplified using nested primers. Promoter primers were: round I, forward: (−448 to −407) GAAGAATTCAGTTGATGGGATATTAGTTATAAAATTAATTT, reverse: (+194 to +153) AAATCTAGACCCAATCATCTAAATAATAAAAAAAACAA. Round II: forward: (−402 to −366) TTTGAATTCATGTGTTTTTTTTTTTATATATTAGGTTTT, reverse: (+150 to +116) AATTCTAGAAAATTTTCATACTAATAAACTATCCAATAA. The enhancer was amplified in three fragments.

The primers were as follows.

Round I: forward: (+11,311 to +11,348) GGAGAATTCTTATGGAGAATTATTTATTATATATTTT, reverse: (+11,868 to +11,901): TAATCTAGAAAATTAAAAAAAAAAACTAAAAAAAAAAAA. Round II: forward: (+11,355 to +11,395) TGTGAATTCGTAATAAGTGAATTATAGGTTATATGAAATT, reverse: (+11,747 to +11,786) AAATCTAGAAAAAAAAAATTAAAAAAAAATAAATAACTA.

Round I: forward: (+11,466 to +11,506): TTTGAATTCAAGATAAGGTTATTATGTATAGGTTGAATTT, reverse: (+12,311 to +12,271): ATTTCTAGAAAAACAAAATTTATAAAAACATCACAAAAA. Round II: forward: (+11,498 to +11,536): TTTGAATTCGAGTAAATAGTAGATAATTTGTTAAGTTT, reverse: (+12,259 to +12,221): CCCTCTAGATACTTCCTCTTTCCCCAACCTAACTAAAAA.

Round I: forward: (+12,239 to +12,279) TATGAATTCGGTAGTATATTTTTTGGAAGTTTGTGATAGT, reverse: (+12,505 to +12,545): TCTTCTAGAAATAAAACCACCTAAACAACTATTAATAAA. Round II: forward: (+12,286 to +12,327), TTAGAATTCTTATATTATAAAATATTTTGGTGGTAGTATAT, reverse: (+12,470 to +12,510): TAATCTAGACCACTACTCCTTATTTAAAAACCCTCTCCAA. The amplified fragments were ligated into pBS⁺ (Stratagene), cloned, and 10 fragments sequenced (University of Michigan DNA Sequencing Core) for each amplified region from each subject.

The CD40LG upstream enhancer and promoter (−1,227 to +60) and downstream enhancer (+11,371 to +12,224) were amplified by PCR and cloned into pGL3-Basic or pGL3-Promoter (Promega) as appropriate. An SpeI site was engineered into the promoter using the QuikChange site-directed mutagenesis kit (Stratagene) as described (5). The regions of interest were excised with the appropriate restriction endonucleases, gel purified, methylated with SssI and S-adenosylmethionine, ligated back into the appropriate construct, and transfected into Jurkat cells using previously described protocols (18). Controls included β-galactosidase transfection as well as mock methylated constructs similarly generated but omitting the SssI.

MS-PCR primers were designed to hybridize with methylated (forward TGTTAAGTTTAGTTTTGTTTTTTTGC, reverse TCGATTCTCCACTATATTAAACGTT) or unmethylated (forward TTAAGTTTAGTTTTGTTTTTTTGTG, reverse TTCAATTCTCCACTATATTAAACATT) CD40LG enhancer sequences (bp 11,526–11,552 and 11,711–11,686 for methylated, and bp 11,528–11,554 and 11,712–11,686 for unmethylated). The loading control consisted of primers specific for a region lacking CG pairs (forward TTGGAGGGGGTTAGGTTTTAG, reverse AAAAAAACAAAAAAATTAAAAAAAA). Results were standardized to the loading control and expressed as the methylation index (methylated/(methylated + unmethylated)).

Statistical analysis was performed using the t test or ANOVA and post-hoc testing with Fisher’s least significant difference test. Unless otherwise noted, results are presented as the mean ± SEM.

CD40LG is regulated by a core promoter and upstream and downstream enhancers (19). Fig. 1,A shows the transcription start site, promoter, three NF-AT sites, and all CG pairs. There are 10 CG pairs in the promoter and surrounding the start site and 3 in the upstream enhancer. Fig. 1 B similarly shows the downstream enhancer with 11 CG pairs and an NF-κB site.

CD40LG promoter and downstream enhancer methylation patterns were compared in CD4⁺ T cells from healthy men and women. Fig. 2,A shows representative methylation patterns of the eight CG pairs surrounding the start site in 10 cloned fragments from a woman. Half the fragments are unmethylated, while the other half are largely methylated, consistent with one active and one inactive gene. In contrast, all cloned fragments from a representative male subject are largely unmethylated (Fig. 2,B), consistent with their one active gene. The average methylation of each CG pair in the CD40LG promoter of CD4⁺ T cells from five women is shown in Fig. 2,C, and of three men in Fig. 2,D. Approximately half the pairs are methylated in women, while almost none are in men. Fig. 2,E shows a similar analysis of the CD40LG enhancer in T cells from the same women; Fig. 2 F shows the men. Again, ∼half the CGs in the female enhancer are methylated and half are not. In men, the proximal enhancer is somewhat more methylated than the promoter although overall methylation is still less than in women.

We then determined whether 5-azaC, an irreversible DNA methyltransferase inhibitor, increases CD40LG expression more in women than men. PBMC from five to six men or women were stimulated with PHA, treated with graded amounts of 5-azaC, and T cell CD40L expression was compared between the men and women using flow cytometry. Fig. 3,A shows the effect on percent CD4⁺CD40L⁺ cells. No significant differences were found between men and women with or without 5-azaC treatment, although there was a small increase in CD4⁺CD40L⁺ cells in women. Fig. 3,B shows the effect of 5-azaC on CD40L mean fluorescence intensity (MFI). There was a 27% increase in CD40L levels on male CD4⁺ cells. In contrast, CD40L levels approximately doubled (216%) on 5 μM 5-azaC-treated female CD4⁺ cells. Higher concentrations gave no further increase (data not shown). No difference was seen in CD40L MFI on untreated CD4⁺ T cells from men and women (female/male 0.86 ± 0.09, n = 7 overall). Fig. 3,C shows the effect on percent CD8⁺CD40L⁺ cells. Relatively few CD8⁺ T cells express CD40L, and 5-azaC increased CD8⁺CD40L⁺ cell numbers equally in men and women. No significant difference in CD40L expression levels was seen in untreated CD8⁺ cells from men and women (female/male 0.90 ± 0.08, n = 7), or on 5-azaC-treated CD8⁺ cells from men and women (Fig. 3 D).

These results were confirmed at the mRNA level in four men and four women. 5-azaC approximately doubled CD40L mRNA in CD4⁺ cells from the women but not the men (fold increase 2.32 ± 0.19 vs 1.43 ± 0.23, women vs men, p = 0.02). Others reported that ∼1.8-fold increases in CD40L cause a significant increase in high-affinity IgG autoantibodies (14), suggesting that the increase in CD40L expression on CD4⁺ T cells from women is functionally significant. In contrast, and similar to results seen at the protein level, CD40L mRNA increased equally in male and female CD8⁺ cells (fold increase 2.52 ± 0.85 vs 2.38 ± 0.85, women vs men). As a control, we compared 5-azaC effects on an autosomal gene, TNFSF7 (CD70) in CD4⁺ T cells from the same men and women. We previously reported that 5-azaC demethylates the T cell TNFSF7 promoter and increases CD70 expression in women and men (7). In contrast to CD40LG, no significant difference was seen in the TNFSF7 increase caused by 5-azaC in CD4⁺ cells from women and men (CD4 fold increase 2.9 ± 0.9 vs 3.3 ± 0.6, women vs men).

We compared 5-azaC effects on CD40LG promoter and enhancer methylation in T cells from men and women. Fig. 4,A shows the promoter methylation pattern in untreated CD4⁺ T cells from four women, and Fig. 4,B shows the pattern in their 5-azaC-treated cells. 5-azaC causes a decrease in methylation across the region. Fig. 4, C and D, similarly compare enhancer methylation in the same cells. Again, 5-azaC decreases deoxymethyl cytosine content. The degree of CD40L overexpression in women was inversely proportional to enhancer methylation as measured by MS-PCR (R² = 0.5, p = 0.05). Others have reported that the effects of 5-azaC on cellular phenotype and DNA methylation are dose dependent over this concentration range (20). The promoter was nearly completely unmethylated in untreated and 5-azaC-treated CD4⁺ T cells from men (3 ± 2% vs 2 ± 2% methylated, n = 3), and 5-azaC had a small but insignificant effect on enhancer methylation (2 ± 1%). In contrast to CD4⁺ cells, 5-azaC demethylated the promoter equally in CD8⁺ T cells from three men and three women (69 ± 6% vs 46 ± 12% methylated, untreated vs treated, p = 0.05 by paired t test).

Methylation effects on promoter and enhancer function were tested by cassette methylation (Fig. 5). Fragments containing the CD40LG upstream enhancer and promoter or downstream enhancer were cloned into the appropriate reporter constructs as described in Materials and Methods, then the upstream enhancer (bp −1,227 to −350), promoter (−350 to +60) or downstream enhancer (+11,371 to +12,224) were excised, methylated in vitro with SssI and S-adenosylmethionine, purified, ligated back into the constructs, and transfected into Jurkat cells. Methylation of the 5′ enhancer had a small suppressive effect, while methylation of the five CG pairs (−350 to +60) just upstream of the transcription start site had a somewhat greater effect. Methylation of the downstream enhancer also suppressed enhancer function to approximately the same degree as the promoter. This indicates that CD40LG promoter and enhancer function is methylation sensitive.

We then determined whether CD40LG demethylates in T cells from women with lupus. Fig. 6, A–E, compare CD40LG promoter and enhancer methylation in CD4⁺ cells from women with lupus and controls. Fig. 6,A shows the promoter methylation of five female controls, and Fig. 6,B the methylation of the same region in CD4⁺ cells from five women with mildly (SLEDAI = 4) active lupus. There is modest demethylation of the five CG pairs just 5′ to the start site. Fig. 6,C similarly shows the methylation pattern in CD4⁺ cells from three women with more active lupus (SLEDAI = 8.7). Methylation of the eight CG pairs surrounding the start site is less than controls and patients with less active lupus. Fig. 6, D and E, compare enhancer methylation patterns in CD4⁺ cells from the patients and controls. Overall methylation is diminished in the lupus patients. There was no significant effect of disease activity on enhancer methylation (data not shown).

Together, these results suggest that promoter and enhancer demethylation may cause CD40LG overexpression on CD4⁺ T cells from women but not men with lupus. CD40LG expression was therefore compared on CD4⁺ cells from men and women with lupus. Promoter demethylation is proportional to disease activity, so the men and women were matched for SLEDAI scores (4.33 ± 0.61 vs 3.86 ± 0.77, mean ± SEM, male vs female). Each subject was paired with a healthy age- and sex-matched control, and CD40L levels on CD4⁺ cells measured as in Fig. 3. Fig. 6 F compares CD40L levels in the men and women with lupus, standardized to the healthy controls. CD4⁺ cells from women but not men with lupus overexpress CD40L.

We confirmed and extended these results by comparing CD40LG and TNFSF7 transcripts in three healthy men and women and three men and women with lupus (SLEDAI 4.7 ± 1.8 and 4.3 ± 1.2, respectively). Fig. 7 A shows that CD4⁺ cells from lupus women have higher CD40LG levels than the other three groups (p = 0.023 by ANOVA). The relationship between CD40LG enhancer methylation status, measured by MS-PCR, and transcript levels was assessed in the control and lupus women. Similar to the 5-azaC-treated cells, there was an inverse relationship between the enhancer methylation index and CD40LG mRNA (R² = 0.866, p = 0.007).

Fig. 7 B compares TNFSF7 transcripts in CD4⁺ cells from the same subjects. Similar to experimentally demethylated cells, TNFSF7 mRNA was not significantly different in T cells from healthy men and women, and increased equally in CD4⁺ from both men and women with lupus (p = 0.01 lupus vs controls).

Additional studies compared expression of FOXP3, also encoded on the X chromosome, in CD4⁺ and CD8⁺ T cells from the same lupus patients and controls. No significant differences in FOXP3 transcripts were seen between male and female CD4⁺ lupus T cells (data not shown). However, CD8⁺ cells from lupus women had higher FoxP3 levels than lupus men (0.42 ± 0.07 vs 0.18 ± 0.003, mean ± SEM, p = 0.03).

These studies demonstrate differential CD40LG methylation in CD4⁺ T cells from healthy men and women. CD40LG is encoded on the X chromosome, so women have two copies and men have one. Bisulfite sequencing of DNA from CD4⁺ T cells revealed that women have one methylated and one unmethylated CD40LG gene, while the one gene in men is largely unmethylated. Because one X chromosome is inactivated in women by processes including DNA methylation (10), the methylated gene is from the inactive X. The unmethylated gene in men is consistent with reports that CD40L is expressed on all CD4⁺ T cell subsets, including naive, memory, Th0, Th1, Th2, and clones (21), indicating both that the necessary transcription factors are present in all CD4⁺ T cells, and that there is no subset-specific suppressive methylation.

The present studies also demonstrate that demethylating female CD4⁺ cells with 5-azaC doubles CD40L expression, while similarly treated male T cells have a much smaller increase. The female-specific increase is associated with demethylation of the core promoter and downstream enhancer, and methylation of these regions in reporter constructs suppresses function, indicating that the methylation changes are transcriptionally relevant. PHA stimulation does not contribute to the demethylation, because CD40LG promoter and enhancer methylation patterns were essentially identical in unstimulated and PHA-stimulated CD4⁺ T cells from healthy women. This was also seen in similar studies on the methylation of the ITGAL, PRF1, and TNFSF7 genes in unstimulated and PHA-stimulated T cells (5, 6, 22). Thus, 5-azaC induced demethylation of CD40LG on the inactive X contributes to the doubling of expression in women, while CD40LG expression increases only slightly in men because their single gene is largely unmethylated. Additional mechanisms contribute to X chromosome inactivation including coating with Xist RNA, histone modifications, and histone variant macro-H2A recruitment (10). That 5-azaC is sufficient to reactivate CD40LG suggests that DNA methylation is a primary mechanism suppressing its expression. 5-azaC also demethylated the CD40LG promoter and induced expression equally in CD8⁺ cells from men and women. This suggests that methylation is a mechanism suppressing CD40LG expression in the CD8⁺ subset.

The CD40LG regulatory sequences demethylated by 5-azaC also demethylate in CD4⁺ T cells from women with lupus, correlating with overexpression in women but not men with lupus. Others have excluded T cell activation as a mechanism for the overexpression in lupus (19), and subset-specific effects are unlikely given the nearly complete demethylation observed in the women with more active lupus and lack of subset-specific differences in expression reported by others (21). The overexpression is unlikely to be due to medications, because the drugs commonly used to treat lupus do not inhibit DNA methylation (4, 23). T cell DNA also demethylates with age (24), but in this study the subjects were all within the same age range (Table I and Materials and Methods), and no effects of age were noted in individual subjects. We reported similar demethylation and overexpression of the ITGAL, PRF1, and TNFSF7 genes in 5-azaC-treated and lupus CD4⁺ cells (5, 6, 7), suggesting that demethylation occurs throughout the genome. Our present observation that TNFSF7 expression increases in 5-azaC-treated and lupus T cells from men and women confirms this. The mechanism causing DNA demethylation in idiopathic and hydralazine-induced lupus is decreased ERK pathway signaling, resulting in failure to up-regulate the maintenance DNA methyltransferase Dnmt1 during mitosis and consequently failure to replicate methylation patterns in the newly synthesized DNA (3, 25). Our present observation that the same CD40LG sequences demethylate in both 5-azaC-treated and lupus T cells supports our earlier report that proliferating T cells with treated DNA methyltransferase inhibitors, or ERK pathway inhibitors, demethylates the same DNA sequences (6).

T cells overexpressing ITGAL, PRF1, and TNFSF7 become autoreactive, killing autologous macrophages and overstimulating B cells (8). Macrophage killing releases apoptotic material and impairs its clearance, leading to generation of anti-chromatin Abs and accelerating lupus (26). CD40LG overexpression could amplify the anti-chromatin response more in women than men, increasing lupus susceptibility. The present observations also support the hypothesis that abnormal T cell DNA methylation contributes to lupus pathogenesis. There are >1000 genes on the X chromosome, some with immune function (27). CD40LG demethylation and overexpression raises the possibility that other X chromosome genes may also demethylate in lupus, amplifying autoimmune processes in women. This is supported by our observation that women with lupus also overexpress FOXP3 transcripts in CD8⁺ T cells relative to men. Others have reported a decrease in FoxP3 transcripts in CD4⁺ lupus T cells (28), but the functional significance of the gender-specific difference in the CD8⁺ subset is unknown.

Multiple mechanisms may contribute to female-biased diseases. DNA generally demethylates with age (24), so some female predominant diseases of aging could involve X chromosome reactivation. A recent report indicates that ∼15% of genes on the inactive X escape inactivation to some degree in fibroblasts (29), suggesting another mechanism predisposing women to diseases. Both X chromosomes are active during early development, then one is randomly inactivated, and skewed X chromosome inactivation could also be a mechanism predisposing women to disease (30). Finally, estrogen likely contributes to the propensity of women to develop lupus, but does not completely explain the female predominance (9). The present work is the first to add X chromosome demethylation to these mechanisms.

We thank Cindy Bourke for her expert secretarial assistance and Dr. Zhaohui Qin for his statistical advice.

The authors have no financial conflict of interest.