STAT4 Expression in Human T Cells Is Regulated by DNA Methylation but Not by Promoter Polymorphism¹

The ability of committed Th1 and Th2 cells to function in altered cytokine environments is a central issue in autoimmune and immune-mediated diseases. Recent observations strongly suggest that the polarization of Th1 and Th2 differentiation is orchestrated by the specific and precise regulation of various cytokine genes (reviewed in Ref.1). Th1 cells produce one set of characteristic cytokines, most notably IL-2 and IFN-γ, whereas Th2 cells secrete a second set of cytokines, including IL-4, IL-5, IL-10, and IL-13 (reviewed in Ref.2). The stringent regulation of cytokine gene expression requires the cooperation of T cell-specific transcription factors, such as STAT4 and STAT6 (1), which mediate intracellular signaling triggered by the binding of cytokines to their receptors. STAT6, which is activated by IL-4 and IL-13, is primarily responsible for the transcriptional effects of these cytokines (3, 4, 5). Mice lacking STAT6 have defects in IL-4-mediated functions, such as B cell proliferation, Th2 cell development, and secretion of IgE, and therefore have a dominantly Th1 phenotype (6). In contrast, STAT4 induces the transcription of IFN-γ in response to the interaction of IL-12 with the IL-12R (7, 8). STAT4-deficient mice lack many IL-12-stimulated responses, including the induction of IFN-γ secretion and the differentiation of Th1 cells (9, 10). These mice are generally resistant to autoimmune diseases such as proteoglycan-induced arthritis, experimental autoimmune encephalomyelitis, and diabetes (reviewed in Ref.11). Interestingly, in a New Zealand mixed mouse model of lupus, mice deficient in STAT4 showed accelerated nephritis and increased mortality (12, 13). In an asthma model, STAT4^−/− mice showed significant decreases in airway hyperreactivity and eosinophil accumulation (14). Collectively, these findings strongly indicate that a deficiency of STAT4 expression is directly associated with impaired Th1 responses and associated immune diseases. However, the molecular mechanisms for transcriptional or posttranscriptional regulation of STAT4 expression have not yet been elucidated.

In this study, we identify and characterize the transcriptional regulatory elements in the promoter region of the human STAT4 gene. There was no evidence of STAT4 promoter region mutations in genomic DNA isolated from 91 patients who had asthma or rheumatoid arthritis (RA)³ (diseases implicated by the mouse models). Instead, our results revealed that STAT4 expression is regulated by DNA hypermethylation in human primary T cells and CD4⁺ T cells.

A 2.8-kb fragment containing the 5′ flanking region upstream of the first exon of STAT4 was amplified using genomic DNA isolated from Jurkat T cells. The amplified product was cloned into the pXP2 vector, which contains a promoterless luciferase gene. To identify the minimal promoter region, deletion constructs were progressively generated by PCR, and cloned into pXP2.

A SMART RACE cDNA amplification kit (BD Clontech) was used to identify the transcription start sites in the STAT4 promoter. Total RNA was isolated from Jurkat T cells, first-strand cDNA was synthesized, and 5′-RACE was performed using universal primer A and a STAT4-specific primer (5′-TGC CGC CGC TTC CAG TCT TGC AG-3′). The subsequent PCR step was performed using nested universal primer A and a nested STAT4-specific primer (5′-CCC TGG ACA GGC ATG TTG GCT GCA G-3′). Amplified PCR products were cloned into the pGEM-T-easy vector (Promega), and 20 clones were sequenced. Adenosine (at position +1) was a major transcription start site in 11 clones.

Jurkat T cells were transfected with promoter construct DNA (9 μg) by electroporation (Bio-Rad) and were harvested with reporter lysis buffer (Promega) 48 h later. All transfections were normalized against an internal control (pCMV-β-galactosidase). For immunoblot analysis, the cells were harvested by scraping, washed twice in cold PBS, and lysed in a buffer containing 50 mM HEPES (pH 7.2), 250 mM NaCl, 2 mM EDTA, 0.1% Nonidet P-40, 1 mM DTT, 1 μg/ml aprotinin, 1 μg/ml leupeptin, and 50 μg/ml PMSF. Equal amounts of protein from each sample, as determined by Bio-Rad assay, were subjected to SDS-PAGE, transferred to nitrocellulose membranes, and analyzed with anti-STAT4 (Santa Cruz Biotechnology) and anti-actin Abs (Sigma-Aldrich).

Patients with asthma (n = 41; age 50 ± 12 years) or RA (n = 50; age 55 ± 11 years) were identified at Seoul National University and the Catholic University of Korea (Seoul, Korea), respectively. The study protocols were approved by the institutional review boards of both hospitals, and informed consent was obtained. The diagnosis of asthma was made when a subject with symptoms of dyspnea or wheezing showed reversible airway obstruction based on the National Institutes of Health guidelines (15), and asthma severity was determined according to lung function and the medication use index needed to obtain control, as previously described (16, 17). The diagnosis of RA was made according to the 1987 revised criteria of the American College of Rheumatology (formerly the American Rheumatism Association) (18).

Peripheral blood samples were taken from healthy donors and PBMC were isolated by Ficoll-Hypaque (Amersham Pharmacia Biotech) density gradient centrifugation. Cells were purified by negative selection using MACS. The negatively selected CD4⁺ T cells were sequentially incubated with a biotin-Ab mixture and anti-biotin conjugated microbeads, followed by depletion of negative cells using LS+ MACS columns (Miltenyi Biotec). The purity of the selected cell populations was 96–98%. Human CD4⁺ T cells were stimulated with anti-CD3 and anti-CD28 mAbs (BD Pharmingen), incubated for 24 h, stained by PE-conjugated anti-CD4 and FITC-conjugated anti-CD69 for 30 min at 4°C, and then analyzed by FACS.

Total RNA was extracted from Jurkat T and CD4⁺ T cells using the RNeasy RNA extraction kit (Qiagen), and 1 μg each of RNA sample was reverse-transcribed (First Strand cDNA kit; Roche). The first-strand cDNA (5 ng) was subjected to 40 cycles of real-time PCR amplification in a LightCycler (Roche) using primers 5′-CAC CTG CCA CAT TGA GTC AAC TA-3′ and 5′-TAA GAC CAC GAC CAA CGT ACG A-3′. The reference gene, GAPDH, was amplified with intron-spanning primers and used for normalization of the cDNA concentrations.

EMSA was performed with oligonucleotides designed to span the AP1/CREB, SP1, OCT1/NF-κB, CREB, NF-κB, AP1, and p53 elements of the STAT4 promoter, as follows: AP1/CREB (nt −56 to −33) 5′-ACT CGA ACA CTG ACG CAC AGG AAA-3′; SP1 (nt −32 to −9) 5′-GCC TCA AGT GGG AGG AGA AAT GCA-3′; OCT1/NF-κB (nt −20 to +5) 5′-AGG AGA AAT GCA AAT CCC CTA CTG A-3′; CREB (nt −3 to +25) 5′-CCT ACT GAT GAT GGC GTC AGC GGC TTT C-3′; NF-κB (nt +25 to +52) 5′-CTC CTA GGG ACT GTG AGG GGC GCT TCT G-3′; AP1 (nt +43 to +67) 5′-GGC GCT TCT GAC TTT GGA CTT GAG C-3′; and p53 (nt +138 to +163) 5′-AGG AGG ACA TGC TTA TTA TGC AGG AT-3′. Jurkat T cell nuclear extracts (10 μg) were incubated for 30 min at room temperature in final volumes of 20 μl containing 2 μg of poly(dI-dC), 10 mM HEPES (pH 7.9), 60 mM KCl, 1 mM EDTA, 1 mM DTT, 7% glycerol, and 0.0175 pmol [γ-³²P]dATP-labeled DNA probe (purified on a G25 column; Roche). Samples were loaded on a 6% polyacrylamide nondenaturing gel and electrophoresed at 180 V, and the dried gels were exposed to film on an intensifying screen for 12 h at −70°C. For competition assays, the previously known consensus sequences for AP1, CREB, NFAT, NF-κB, OCT1, p53, and SP1 (Santa Cruz Biotechnology) were used. Before the addition of the labeled probe, nuclear extracts were incubated with 50-fold molar excesses of unlabeled consensus and mutant oligonucleotides of each transcription factor as competitors (see Fig. 2 D).

Genomic DNA was restriction digested with BstEII and EcoRI, denatured with sodium hydroxide (0.3 N) at 39°C for 30 min, and then treated with hydroquinone (5.5 mM) and sodium bisulfite (pH 5, 2.5 M). The samples were incubated in a Thermocycler at 55°C for a total of 16 h and punctuated every 3 h by a 5 min denaturation at 95°C. The resulting converted DNAs were desalted using a Wizard DNA clean-up kit (Promega), denatured with sodium hydroxide (0.3 N) at 37°C for 15 min, and purified by ethanol precipitation. The bisulfite-modified DNAs were PCR amplified using primers that targeted two regions as follows: region one, 5′-AGG GGT GG TTT ATT AGG TTG AGT GGA GAG GGA GAG A-3′ and 5′-TAA ACA TAT CCT CCT TAC AAA AAA CAA CC-3′ for the first amplification, and 5′-AGG GAG AGA GGA AGT TGA AGA ATT GG-3′ for the nested PCR amplification; region two, 5′-GAG TAT TGT TTG GGA TTT GTG TTG AGA GAG-3′ and 5′-CCT TTA AAC ATC CTT CAA CAA ATT TCT TTA C-3′ for the first amplification, and 5′-TTT GAG GAG GTT TAT TAT AGA AGG AAG-3′ and 5′-ACT TCT AAA AAC TCA ATA CCT ATC TAT CC-3′ for the nested PCR. The products from the nested PCR amplifications were cloned into the pGEM-T vector (Promega), and 10 clones of each were sequenced using an M13 reverse primer.

To investigate the transcriptional or posttranscriptional mechanism responsible for regulating STAT4 expression, we first identified the STAT4 promoter sequences via a BLAST search of the human genome against the cDNA sequence of the STAT4 gene. Using genomic DNA from Jurkat T cells as a template, we amplified a 2.8-kb fragment containing the STAT4 5′ flanking region, upstream of the translation initiation start site (ATG). To determine whether this fragment (nt −2225 to +605) (Fig. 1,A) conferred the expected promoter activity, we fused the fragment to the luciferase cDNA and transiently transfected the resulting reporter construct into Jurkat T cells. The luciferase activity in these cells was ∼150-fold higher than that in cells transfected with the promoterless luciferase backbone vector pXP2. To define the minimal sequences required for the transcriptional initiation of STAT4 expression, we generated promoter mutants by progressive deletion of nucleotides from the 3′ and/or 5′ ends and placed these fragments in the luciferase reporter construct. Our results revealed that the minimal STAT4 promoter region consisted of the region from nt −48 to +172 (GenBank Accession no. DQ003482) (Fig. 1, B and C). We then identified potential cis-acting regulatory elements within the optimal STAT4 promoter region, using the MatInspector version 2.2 and MATCH version 1.0 computer programs. The identified elements included multiple putative transcription factor binding sites, such as those for AP1, CREB, NFAT, NF-κB, OCT1, p53, and SP1 (Fig. 1,C). The transcription start site was identified by 5′-RACE analysis of the STAT4 mRNA (data not shown); analysis of the region surrounding this transcriptional start site failed to reveal a TATA box. Interestingly, the comparison of STAT4 promoter gene between human and mouse sequences showed high conservation in their putative cis-acting binding elements (Fig. 1 D).

To determine which cis-acting elements in the STAT4 promoter are important for transcriptional activity, we generated eight deletion mutants of the STAT4 promoter (Fig. 2,A). We found that disruptions of the regions −47 to −40 (ΔI), −23 to −17 (ΔIII), −14 to −2 (ΔIV), and +10 to +17 (ΔV), containing the CREBs, SP1, and OCT1/NF-κB motifs, respectively, significantly reduced STAT4 promoter activity (Fig. 2,B). However, deletions of the other transcriptional elements had no significant effect on transcriptional activity. We then used a gel-shift competition assay to assess whether these cis-acting elements physically interacted with the transcription factors. Double-stranded oligonucleotides containing putative factor binding sites for CREB (I and V), SP1 (III), and OCT1/NF-κB (IV) were radiolabeled. As shown in Fig. 2, C and D, EMSA analysis of Jurkat T nuclear extracts was performed by competition with unlabeled oligonucleotides (cold, Fig. 2,D, lanes 2 and 6), the transcription factor binding consensus sequences encoding CREB, NF-κB, OCT1, or SP1 (consensus, Fig. 2,D, lanes 3, 7, 10, 13, and 16), or their mutant sequences as a negative control (mutant, Fig. 2,D, lanes 4, 8, 11, 14, and 17). Consistent with the results of our reporter gene assay, competition with the CREB, SP1, and OCT1 oligonucleotides dramatically eliminated specific binding in Jurkat T cell nuclear extracts, whereas the corresponding mutated oligonucleotides (negative controls) did not. In contrast, the DNA-protein interactions were not altered by competition with oligonucleotides corresponding to NF-κB (Fig. 2 D), NFAT, AP1, or p53 (data not shown). These data suggest that multiple factors including CREB, SP1, and OCT1 cooperate for the maximal activation of STAT4 transcription.

To determine whether polymorphic base changes in the STAT4 promoter region affect STAT4 transcription, we sequenced this region in genomic DNA isolated from 41 patients with asthma and from 50 patients with RA. We did not identify any polymorphic changes in the defined STAT4 essential promoter region (data not shown), indicating that promoter polymorphisms may not be involved in the modulation of STAT4 promoter activity. We did, however, identify a novel allelic variation (−149A/G) in noncoding exon 1 of the STAT4 gene, which is in the 5′ flanking region of the essential promoter. Direct sequencing analysis revealed that the heterozygous (G/A) genotype was present in ∼30% of the RA patients and ∼17% of the asthma patients, whereas the mutant homozygous allele (A/A) was present in 6 and 3% of these patients, respectively (Fig. 3,A). As this noncore promoter region allelic variation appeared frequently in patients with RA and asthma, we performed luciferase reporter gene assays using STAT4 promoter constructs (−288 to +605) containing either the wild type (G/G) or mutant homozygous (A/A) alleles to examine possible effects on promoter activity. We observed no significant difference in transcriptional activity between the alleles (Fig. 3 B), suggesting that this promoter polymorphism is not involved in the regulation of STAT4 expression.

In addition to genetic lesions, changes in promoter hypermethylation patterns have been shown to regulate gene expression. The promoter regions of Th1-cytokines, such as IFN-γ, are hypomethylated in murine Th1 clones, but hypermethylated in murine Th2 clones (19, 20). In addition, in vitro methylation of the IFN-γ promoter was found to markedly reduce the binding of specific transcription factors to the promoter (21, 22, 23, 24). Together, these results indicate that promoter methylation is important in regulating the expression of T cell-specific transcription factors and cytokine genes, such as IFN-γ and IL-4, during the development of Th populations. To determine whether DNA methylation is involved in transcriptional regulation of STAT4 expression, we isolated human peripheral blood T cells from healthy donors, stimulated the cells with anti-CD3 and anti-CD28 Abs, treated the cells with an irreversible DNA methyltransferase inhibitor 5-aza-2′-deoxycytidine (5-aza-2-DC), and used real-time RT-PCR to measure STAT4 transcript levels. Our results revealed that the mixed populations of primary T cells showed dramatic increases in the mRNA levels of both STAT4 and IFN-γ following treatment with the demethylating drug (Fig. 4 A).

We also treated negatively selected CD4⁺ T cells with anti-CD3 and anti-CD28 Abs followed by 5-aza-2-DC. Twenty-four hours later, the stimulated CD4⁺ T cells were identified by FACS analysis using PE-conjugated anti-CD4 and FITC-conjugated CD69 Abs (Fig. 4,B), and the levels of STAT4 mRNA in the selected CD4⁺ T cells were measured by quantitative RT-PCR and expressed relative to GAPDH mRNA expression (Fig. 4,C). As expected, treatment with 5-aza-2-DC resulted in increased STAT4 promoter activity (data not shown). In addition, the CD4⁺ T cells expressed approximately four times more STAT4 mRNA than did untreated CD4⁺ T cells (Fig. 4,C). Finally, the immunoblotting revealed that demethylation dramatically increased the protein levels of STAT4 but not actin (Fig. 4 D). Collectively, these findings indicate that DNA methylation affects the mRNA and protein expression levels of STAT4.

The STAT4 gene promoter and its flanking regions contain ∼30 methylatable CG pairs (Fig. 5,A). To examine the methylation patterns of these potential substrate sites, DNA isolated from Jurkat T cells was treated with sodium bisulfite, two PCR fragments (Fig. 5,A, PCR no. 1 and PCR no. 2) were cloned and 10 independent clones per PCR fragment were sequenced. We found that the CG pairs in the 3′ flanking region of the proximal STAT4 promoter (nt +521 to +647) were highly methylated and in the 5′ flanking region (nt −291 to −126) contained a methylated CG pair and four singly methylated deoxycytidine (dC) bases. In contrast, the core region of the STAT4 promoter (nt −48 to +172) showed no significant pattern of CG methylation (Fig. 5 B).

Previous studies have reported that DNA methylation proximal to the promoter may indirectly repress or silence gene expression by increasing chromatin compaction, thus rendering the promoter/enhancer regions less accessible to transcription factors (22). We were unable to identify methylated CpG sites in the essential STAT4 promoter region, indicating that the repression or silencing of STAT4 transcription may be indirectly mediated by DNA methylation. To investigate whether methylation of the four dC bases and the CpG sites in the 3′ and 5′ flanking regions of the STAT4 promoter affects STAT4 transcription, we generated additional STAT4 mutant promoter constructs containing 5′ flanking (nt −305 to +172) and 3′ flanking (nt −48 to +682) regions, and transfected these constructs into Jurkat T cells for reporter gene assays. As shown in Fig. 5,C, deletion of the 3′ methylatable dC bases and single CG pair did not alter the transcriptional activity of the STAT4 promoter (compare nt −305/+172 to −48/+172), whereas deletion of the 5′ CG pairs dramatically increased the transcriptional activity driven by the STAT4 essential promoter (compare nt −48/+682 to −48/+172), indicating that the 5′ methylation sites in the proximal STAT4 promoter region are involved in the regulation of transcriptional activity. Next, we assessed whether the STAT4 promoter is less methylated following 5-aza-2-DC treatment, which enhances STAT4 protein levels by demethylation. DNA isolated from Jurkat T cells cultured in the absence or presence of 5-aza-2-DC was treated with sodium bisulfite, cloned and sequenced. As shown in Fig. 5 D, most of the methylated residues in the proximal STAT4 promoter sequences were demethylated by 5-aza-2-DC treatment. Collectively, these findings indicate that the methylation status of the STAT4 promoter region directly affects STAT4 expression.

Initial RT-PCR experiments comparing STAT4 mRNA levels revealed that Jurkat T cells expressed at least 15 times more STAT4 mRNA than did non-T cells such as HeLa cells (Fig. 6,A). Interestingly, this is consistent with differences in the methylation patterns between HeLa and Jurkat T cells; the DNA methylation sequencing experiments showed that the 3′ and 5′ flanking regions of the STAT4 promoter were significantly more hypermethylated in HeLa cells than in Jurkat T cells (Fig. 6,B). Similar results were obtained in other non-T cell lines, including BEAS2B (lung epithelial cells) and THP1 (monocytic leukemia cells) (Fig. 6 C). Immunoblotting using an anti-STAT4 Ab revealed that in the absence of the demethylating drug, STAT4 levels were very low in all tested cell lines, but that the basal STAT4 protein levels were higher in Jurkat T cells than in the other tested cells. Similarly, whereas 5-aza-2-DC treatment significantly enhanced STAT4 expression in all tested cell lines, the degree of enhancement was much more pronounced in Jurkat T cells compared with the non-T cells. These findings may indicate that demethylation of the STAT4 gene is relatively inefficient in non-T cells, or that T cells have an active mechanism for this demethylation. Future work will be required to assess why treatment with the demethylating agent has a more pronounced effect in T cells vs non-T cells, but the present findings strongly suggest that STAT4 expression in human T cells is regulated by selective DNA methylation.

STAT4 appears to be a major player in IL-12-induced responses and Th1 cell differentiation (reviewed in Ref.25). Studies have shown that STAT4^−/− mice have defective Th1 cell proliferation, IFN-γ production, and NK cell activity in response to IL-12 stimulation (9, 26). Expression of IL-12Rβ2 is reduced in STAT4^−/− mice, and transgenic expression of IL-12Rβ2 in the absence of STAT4 is not sufficient to restore Th1 differentiation in these mice (27, 28). STAT4 is activated by phosphorylation mediated by the IL-12 signaling pathway (reviewed in Ref.29), strongly suggesting that STAT4 mediates the Th1 immune response through IL-12 signaling.

STAT4 binds to the first intron of the IFN-γ locus (30), and the formation of a complex between activated STAT4 and the AP1 transcription factor has been found to up-regulate IFN-γ transcriptional activity (31). Thus, STAT4 expression is critical for induction of IFN-γ gene transcription, which pushes Th cells toward the Th1 phenotype. Despite its importance, however, the molecular mechanism regulating STAT4 mRNA expression has not been previously determined. In this study, we show that a 220-bp fragment upstream of the STAT4 coding sequence confers potent STAT4 transcriptional activity, and that disruption of transcriptional elements covering the CREB, SP1, and OCT1 binding sites decreases transcriptional activity. In addition, our results revealed that changes in STAT4 mRNA and protein expression are associated with DNA methylation changes in human PBL and CD4⁺ T cells, providing a novel mechanism by which STAT4 expression is regulated.

STAT4 expression is very low in resting T cells and is up-regulated by stimulation of the TCR in response to IL-12 (7, 11). Unlike IL-12Rβ2, STAT4 is expressed in both Th1 and Th2 cells, but to a greater degree in Th1 cells. Interestingly, the promoter region of the IFN-γ gene is hypomethylated in established Th1 clones, but hypermethylated in Th2 clones (19, 20), providing molecular insight into the selective expression of IFN-γ in Th1 cells. Initial experiments using RT-PCR to quantitate STAT4 mRNA levels revealed that Jurkat T cells constitutively expressed STAT4, but HeLa cells did not. The STAT4 promoter region is highly hypermethylated in HeLa cells compared with Jurkat T cells (Fig. 4), making it likely that the transcriptional regulatory elements of the STAT4 gene are highly methylated in Th2 cells, providing a mechanistic explanation for previous experimental observations.

Mutations in the IL-12Rβ2 gene have been reported in atopic patients, and these mutations were associated with reduced STAT4 phosphorylation and IFN-γ production in response to IL-12 stimulation (32, 33). A previous work failed to find any association between STAT4 mutations/polymorphisms and immune diseases such as allergy and asthma (34). Similarly, we found no evidence of polymorphisms in the STAT4 promoter in patients who have asthma or RA, except for a single silent polymorphism in the proximal regulatory elements of the core promoter. Together, these findings strongly indicate that epigenetic control, such as DNA methylation, may be the major regulator of STAT4 expression. Our results revealed that demethylation of the STAT4 promoter region, either by treatment with a methyltransferase inhibitor or by truncation of methylatable CG pairs, markedly enhanced STAT4 transcription and protein expression.

In summary, we show that STAT4 expression in human T cells is associated with DNA methylation status but not with promoter polymorphisms, providing novel insights into the molecular mechanism responsible for regulating STAT4 transcription in T cells.

The authors have no financial conflict of interest.