Regulation of the Th1 Genomic Locus from Ifng through Tmevpg1 by T-bet

The cytokine IFN-γ plays a critical role in protecting vertebrate organisms from invading pathogens, including bacteria and viruses, as well as preventing tumor development (1–3). Naive CD4⁺ T cells do not produce IFN-γ efficiently in response to antigenic stimulation, but they differentiate into cells termed effector Th1 cells in the presence of antigenic stimulation and cytokine stimulation, such as IL-12, which are capable of producing high levels of IFN-γ in response to antigenic stimulation. This process is governed by transcription factors Stat1 and Stat4 that promote expression of the Th1 transcription factor T-bet. This combination of transcription factors promotes epigenetic remodeling at the Ifng locus (4–8). A general view is that this remodeling process allows for efficient transcription of Ifng mRNA in response to secondary antigenic stimulation (9–12).

Long noncoding RNAs (lncRNAs) are newly discovered species of regulatory RNAs (13, 14). Thousands of lncRNAs have been discovered, but the functions of only a few have been determined. In certain instances, genes encoding lncRNAs are positioned in the genome adjacent to the protein-coding genes that these lncRNAs regulate (15, 16). Examples are HOTAIR and HOTTIP, which regulate the locus of developmentally regulated HOX genes (17, 18). One mechanism by which these lncRNAs target the HOX protein-coding genes is by acting as scaffolds to recruit histone-modifying complexes to target gene loci to produce either repressing (HOTAIR) or activating (HOTTIP) epigenetic modifications. Thus, these lncRNAs appear to function by remodeling the epigenetic landscape rather than by directly repressing or activating transcription of protein-coding genes (19). In other instances, lncRNA genes are adjacent to protein-coding genes in the genome and are coexpressed with these adjacent protein-coding mRNAs, but they have regulatory functions that are distinct from regulating expression of the adjacent protein-coding gene. An example is lncRNA-Cox2 and Ptgs2 (gene that encodes Cox-2) (20). Other examples exist in which adjacent lncRNA-coding genes and protein-coding genes are coexpressed in response to different stimuli (13, 16, 21). Thus, it seems that coexpression of adjacent lncRNA-coding genes and protein-coding genes is a common property, but little is known about the underlying epigenetic and transcriptional mechanisms that produce this phenomenon.

The lncRNA transcript Tmevpg1, also named NeST and Ifng-AS1, which is encoded by a gene adjacent to Ifng in both human and mouse genomes, is induced in response to the Th1-differentiation program by mechanisms dependent upon Stat4 and T-bet (22–24). Tmevpg1 cooperates with T-bet to promote Ifng transcription. One mechanism by which Tmevpg1 functions is to bind to WDR5, a component of the histone H3 lysine 4 (H3K4) methyltransferase complex and enhance marking of the Ifng locus with H3K4 methylation. Importantly, studies in transgenic mice demonstrate that Tmevpg1 (NeST, Ifng-AS1) plays a critical role in establishing resistance to Salmonella enterica pathogenesis and preventing clearance of Theiler’s virus.

Although epigenetic remodeling and transcriptional regulation of lineage-specific protein-coding gene loci have been an area of significant investigation, less is known about the roles of epigenetic remodeling and transcriptional regulation of gene loci that encode lineage-specific lncRNAs. In this article, we show that T-bet binds to the Tmevpg1 gene promoter/proximal enhancer, as well as upstream distal enhancers, in developing and differentiated effector Th1 cells. T-bet also plays a critical role in epigenetic remodeling of these genomic regions. These genomic sites actively recruit inducible transcriptional regulators, such as NF-κB and Ets-1, which is largely dependent upon T-bet. Stimulus-dependent expression of Tmevpg1 is also dependent upon the activity of NF-κB. These findings expand our current understanding of the multiple mechanisms by which T-bet regulates Th1-lineage commitment and the underlying mechanisms that contribute to coexpression of adjacent lncRNA-coding genes and protein-coding genes.

BALB/cJ and DO11.10 mice were obtained from The Jackson Laboratory (Bar Harbor, ME). All mice were bred in the Vanderbilt University animal facilities. Research using mice complied with all relevant institutional and federal guidelines and policies. Human IFNG BAC–transgenic mice, which contain a 190-kb transgene that contains the IFNG and IL26 genes but not the TMEVPG1 gene, were described previously and were backcrossed for >12 generations to the C57BL/6 genetic background (25, 26).

Primary BALB/cJ or DO11.10 splenocytes (1 × 10⁶ cells/ml) were stimulated with immobilized anti-CD3 (2C11; American Type Culture Collection) or OVA_323–339 peptide Ag (10 μg/ml; InvivoGen), respectively, under neutral Th0 (10 μg/ml anti–IFN-γ and 10 μg/ml anti–IL-4), Th1 (10 ng/ml IL-12 and 10 μg/ml anti–IL-4, 11B11; American Type Culture Collection), or Th2 (10 ng/ml IL-4 and 10 μg/ml anti–IFN-γ, R4-642; American Type Culture Collection) polarizing conditions for 3 d, generating primary cultures. Effector cultures were generated by restimulation for an additional 48 h with immobilized anti-CD3. Jurkat T lymphocytes (American Type Culture Collection) were maintained in complete RPMI 1640 at a density of 5 × 10⁵ cells/ml. CD4⁺ T cells and NK cells were purified by negative magnetic selection (Miltenyi Biotec).

Total RNA was isolated with TRI Reagent (Ambion). cDNA was synthesized with the SuperScript III kit (Invitrogen). Tmevpg1 (Life Technologies TaqMan assay ID: Mm01161206_m1) and Ifng (TaqMan; Life Technologies) transcript levels were measured by RT-PCR and calculated relative to Gapdh (TaqMan; Life Technologies) by the δ-δ Ct method.

Chromatin immunoprecipitation (ChIP) procedures were followed as described (11, 26) using the following Abs: anti-H4Ac (06-866; Millipore), T-bet (4B10, sc-21749; Santa Cruz), Ets-1 (C-20, sc-350; Santa Cruz), NF-κB p65 (ab7970; Abcam), and anti-rabbit IgG (Sigma-Aldrich). Primers used to amplify DNA across the Tmevpg1 locus and conserved DNA sequences at DNase 1 hypersensitivity sites (HSs) (27) are listed in Supplemental Table I. Real-time PCR with SYBR Green was used to measure DNA after ChIP and DNA purification. Melt curves after amplification showed that primer pairs generated single products. Quantitative ChIP assays were performed in triplicate, and results from two or three independent experiments were averaged. Samples with a difference < 2.0 in mean Ct value for specific and nonspecific (isotype Ig control) immunoprecipitation (IP) were considered nonspecific. Standard curves from the input DNAs also were generated using the different primer pairs. The slopes of these curves were essentially identical. Therefore, we elected not to include input DNA results in our calculations because this would only add an additional source of error. Relative enrichment was determined using the following formula: 2^{[Ct (Ig control) − Ct (Ab-specific IP)]}, where Ct (Ig control) is the Ct obtained from the IgG isotype control ChIP, and Ct (Ab-specific IP) is the Ct obtained from the Ab-specific ChIP. For example, a mean Ct value of 31 for the isotype-control ChIPs and a mean Ct value of 26 for the Ab-specific ChIPs yielded a relative enrichment of 2⁵ or 32.

Human HS elements were amplified with the primers listed in Supplemental Table I from Jurkat T cell genomic DNA and cloned into the minimal promoter-luciferase or promoter-less luciferase construct (pGL4.24 and pGL4.10, respectively; Promega). The NFAT-IL4-Pr-luciferase construct was described previously (28). Briefly, the construct contains three copies of the distal NFAT binding site within the Il4 promoter, termed P1, linked to a minimal promoter and the firefly luciferase gene. Jurkat T cells were transfected with luciferase constructs by the DEAE transfection method (29) at 1 μg plasmid DNA/10⁶ cells. After overnight recovery, cells were stimulated with 50 ng/ml PMA and 1 μM ionomycin for 6 h before luciferase activity was measured with the luciferase activity system (Promega). Cells were treated with BAY 11-7085 NF-κB inhibitor (5 μM; Sigma-Aldrich) for 1 h (30). Promoter truncations were generated with the primers listed in Supplemental Table I and cloned into the pGL4.10 construct.

Statistical significance was determined by a Student t test using Prism software. A p value < 0.05 was considered statistically significant.

Th1 polarization promotes the expression of IFN-γ while suppressing Th2-specific genes, a process that is orchestrated, in part, by transcription factor recruitment, such as with T-bet, to noncoding regulatory elements (10). T-bet is known to associate with the Ifng promoter and numerous sites along a 140-kb intergenic distance between Ifng and Tmevpg1. Further, T-bet is required for the formation, spreading, and maintenance of certain histone modifications across this region (10). Because Tmevpg1 expression is also dependent upon T-bet, we investigated T-bet association with the Tmevpg1 locus. Splenic cultures were polarized under Th1 conditions for 3 d to generate primary Th1 cultures and for 5 d, with an additional 48 h of stimulation with immobilized anti-CD3, to generate effector Th1 cultures. T-bet binding was assessed from +3.0 kb upstream to −160 bp downstream of the Tmevpg1 transcriptional start site (TSS) by ChIP. T-bet associated with chromatin mapping to the Tmevpg1 locus in primary Th1 cells, as well as in effector Th1 cells (Fig. 1). T-bet was similarly found to associate with the Ifng promoter in both populations; however, it was markedly enriched in effector Th1 cultures. For comparison, a conserved DNA element located 40 kb upstream of Ifng that previously was determined not to recruit T-bet (10) also did not recruit T-bet in our experiments, thus serving as a negative control. Moreover, the majority of T-bet association in the primary Th1 cultures was within ± 100 bp of the Tmevpg1 TSS, whereas T-bet association was found to spread across the entire 3-kb region in effector Th1 cells. Further, T-bet did not associate with the Tmevpg1 locus under Th2 culture conditions. Thus, T-bet bound to the Tmevpg1 promoter in primary Th1 cells and spread across up to +3 kb of the Tmevpg1 region as cells further differentiated into effector Th1 cells.

Given the above results, we next examined T-bet–dependent H4Ac patterns around the Tmevpg1 TSS. Spleen cell cultures from DO11.10 wild-type or DO11.10.Tbx21^−/− mice were polarized under Th1 conditions for 3 d to generate primary Th1 cultures and for 5 d, with 48 h of additional stimulation with immobilized anti-CD3, to generate effector Th1 cultures. H4Ac histone modifications were assessed by ChIP spanning +2.3 kb upstream to −165 bp downstream of the Tmevpg1 TSS. In primary Th1 cultures, H4Ac modifications were enriched at genomic regions surrounding the Tmevpg1 TSS, and these modifications required the presence of T-bet (Fig. 2A). H4Ac modifications also were enriched in effector Th1 cells, and these also were dependent upon T-bet (Fig. 2B). Further, the level of H4Ac modifications was higher in effector Th1 cultures than in primary Th1 cultures, and H4Ac modifications appeared to spread across this ∼3-kb genomic region. Histone H4Ac modifications at these sites in primary and effector Th2 cells were essentially undetectable and were equivalent to isotype controls. These findings indicate that T-bet associated with sites along the Tmevpg1 TSS and was necessary for seeding and spreading of H4Ac marks across this genomic region. Alternatively, T-bet may play a critical, but indirect, role in stimulating the formation of these histone modifications by promoting Th1 differentiation through mechanisms that are incompletely defined.

Identification of DNase 1 HSs is one method to identify potential functional genomic enhancer elements. Genome-wide analysis of DNase 1 HSs in human Th1 and Th2 cells has been performed and is available on the University of California Santa Cruz (UCSC) Genome Browser (31, 32). Inspection of these tracks within the genomic region between Ifng and Tmevpg1 (Fig. 3A, adapted from the UCSC Genome Browser) shows the presence of five Th1-specific DNase 1 HSs, indicating that these areas represent accessible, open chromatin. Using an unbiased deletion strategy to identify noncoding regions conferring Th1-specific Ifng expression, we examined the requirement of two large deletions of the 190-kb BAC transgene mapping to the region between IFNG and TMEVPG1. This BAC transgene contains the complete IFNG gene but does not contain the TMEVPG1 gene. CD4⁺ T cells were purified and cultured under neutral (Th0), Th1, and Th2 conditions for 5 d to generate primary Th1 cells or were restimulated for 48 h with immobilized anti-CD3 to generate effector Th1 cells. Compared with the full BAC transgene, deletion 1 and deletion 2 cultures exhibited equivalent levels of IFNG message relative to Gapdh in Th0, Th1, and Th2 primary and effector cultures (Fig. 3B, 3C), coinciding with levels of IFN-γ in culture supernatants (25). These results support the notion that this intergenic region is dispensable for IFNG expression. Therefore, if these HSs possess transcriptional enhancer activity, it may be that their function is to enhance TMEVPG1 transcription rather than IFNG transcription.

To explore this hypothesis further, we assessed the enhancer activity of each HS. Each HS, numbered one through five, was cloned into the pGL4.24 vector construct containing a minimal promoter followed by the luciferase gene. The constructs were transfected into Jurkat T lymphocytes at 1 μg/10⁶ cells. After overnight recovery, cells were stimulated with 50 ng/ml PMA and 1 μM ionomycin for 6 h, lysed, and assayed for luciferase activity. Transfection with HS1-, HS2-, HS3-, or HS4-pGL4.24 constructs resulted in a significant increase in luciferase activity relative to vector alone, indicating that these DNA elements possessed enhancer activity (Fig. 4A). Similar levels of enhancer activity were observed in HS1 and HS3, whereas HS4 exhibited markedly greater enhancer activity relative to the other HS elements. In contrast, HS5 exhibited a 10-fold decrease in luciferase activity relative to vector alone (Fig. 4B). These findings indicated that HS1, HS2, HS3, and HS4 are functional enhancer elements, whereas HS5 appeared to have suppressive function.

In addition to having enhancer activity, the HS1 element aligns with the TMEVPG1 TSS; thus, we evaluated HS1 for promoter activity. To do so, HS1 was cloned into the pGL4.10 promoterless luciferase construct and transfected into Jurkat T lymphocytes at 1 μg/million cells. After overnight recovery, cells were stimulated with 50 ng/ml PMA and 1 μM ionomycin for 6 h and assayed for luciferase activity. A 3-fold increase in luciferase activity was observed relative to the vector control, supporting the fact that, in addition to enhancer activity, HS1 contains the Tmevpg1 promoter (Fig. 4C).

The HS1 element contains two distinct peaks of Th1-specific DNase I hypersensitivity, either of which could contribute to enhancer, as well as promoter, activity. Therefore, we used a traditional promoter analysis truncation approach to identify regions of HS1 that contributed to the observed promoter activity. The HS1 region was divided into four segments, cloned into the pGL4.10 vector construct, and assayed for promoter activity relative to the full HS1 construct alone (Fig. 4D). The greatest promoter activity was found at the more distal regions of HS1, −281 to −11 bp and +274 to +518 bp, whereas constructs with intervening sequences possessed lower promoter activity (compare activity of the +274 to +518-bp construct with activity of the +7 to +518-bp construct), indicating the presence of both enhancer activity and repressor activity in HS1.

We inspected the HS1 DNA sequence for potential transcription factor binding sites and identified both Ets-1 and NF-κB binding motifs. Ets-1 is a transcription factor predicted to function with T-bet in the Th1-polarization pathway (33, 34). These results indicate that TMEVPG1 is surrounded by functional enhancer sequences, whereas HS1 also possesses promoter activity. Although IFN-γ production by primary and effector Th1 cells is dependent upon T-bet, it is known that T-bet is dispensable for Ifng expression by memory cells (35). Inducible transcription factors, such as NF-κB, in particular, contribute to rapid IFN-γ responses by effector memory Th1 cells (36). As such, we examined NF-κB and Ets-1 recruitment to the Tmevpg1 locus. Spleen cell cultures from DO11.10 wild-type mice were polarized under Th1 or Th2 conditions for 3 d for primary cultures and for 5 d with 48 h of additional stimulation with immobilized anti-CD3 to generate effector cultures. NF-κB and Ets-1 recruitment to Tmevpg1 was assessed by ChIP assay spanning +2.3 kb upstream to −165 bp downstream of the Tmevpg1 TSS. NF-κB recruitment to Tmevpg1 was relatively low across the locus, particularly in primary Th1 cultures (Fig. 5A). However, a distinct peak in NF-κB binding was seen at −70 bp downstream of the Tmevpg1 TSS in effector Th1 cultures. Conversely, in these same samples, Ets-1 was recruited at multiple HS1 sites in both primary and effector Th1 cultures (Fig. 5B). These same transcription factors were not recruited to HS1 in primary or effector Th2 cultures. ChIP determinations at the B2m promoter (37) and the Cd3e promoter (38) for NF-κB and Ets-1, respectively, served as positive signals not dependent upon Th1/Th2 differentiation. Thus, both NF-κB and Ets-1, along with T-bet, were recruited to specific Tmevpg1 HS1 sites in primary and effector Th1 cells but not effector Th2 cells.

Because both NF-κB and Ets-1 associated with Tmevpg1 genomic promoter/enhancers, we examined recruitment of these transcription factors to murine HS (mHS) enhancer elements that were identified by sequence conservation to the above human HS elements (Fig. 3, the HS2 element is not conserved between humans and mice). To address this question, spleen cell cultures from DO11.10 Tbx21^+/+ or DO11.10.Tbx21^−/− mice were polarized under Th1 or Th2 conditions for 3 d for primary cultures and for 5 d with 48 h of additional stimulation with immobilized anti-CD3 to generate effector Th1 or Th2 cultures. We found that NF-κB was enriched at these conserved mHS regions in primary and effector Th1 cultures, but not Th2 cultures, and recruitment was largely T-bet dependent (Fig. 6A, 6B). The magnitude of recruitment to mHS3 and mHS4 was greater than recruitment to the Ifng promoter. Ets-1 also was recruited to each of the mHS enhancer sequences in primary and effector Th1 cultures, but not Th2 cultures, and it was found to be largely, but not absolutely, dependent upon the presence of T-bet (Fig. 6C, 6D), similar to the pattern observed for NF-κB enrichment. Lastly, Ets-1 associated with the Ifng promoter in a T-bet–dependent manner in effector cultures. These results support that T-bet is required, in part, for recruitment of the inducible transcription factors Ets-1 and NF-κB to these intergenic enhancers.

Additionally, we assessed pharmacological inhibition of NF-κB activity by treatment with the extensively used IκBα inhibitor BAY 11-7085. HS1-, HS2-, HS3-, HS4-, and HS5-pGL4.24 constructs were transfected into Jurkat T lymphocytes at 1 μg/10⁶ cells. After overnight recovery, cells were incubated in the presence of the inhibitor for 1 h, followed by stimulation with 50 ng/ml PMA and 1 μM ionomycin for 6 h before determining luciferase activity. Relative to control cultures, cultures treated with the NF-κB inhibitor exhibited a marked decrease in the activity of each enhancer, HS1–HS4, to varying degrees (Fig. 7A). In contrast, BAY 11-7085 did not inhibit transcriptional activity directed by a known NFAT response element from the Il4 promoter (NFAT-Il4-Pr). Further, treatment of effector Th1 cultures with BAY 11-7085 before restimulation impaired Tmevpg1 expression, as well as Ifng expression (Fig. 7B, 7C). In contrast, BAY 11-7085 did not inhibit expression of the stimulus-induced Fos gene (Fig. 7D). These results support an important role for NF-κB activation and recruitment to the Tmevpg1 proximal promoter/enhancer region, as well as HS1–HS5 elements, to achieve stimulus-dependent transcriptional enhancer activity and induction of Tmevpg1 expression.

In this study, we provide evidence to extend the breadth of T-bet positive regulation of the Th1 locus from Ifng over 170 kb upstream to the gene encoding the Tmevpg1 lncRNA. In previous studies, we found that Tmevpg1 expression is dependent upon the master Th1 transcription factor T-bet, but this effect was poorly understood (23). To summarize our results, we find that T-bet associates with the Tmevpg1 locus, promoting H4Ac marks in primary and effector Th1 cells but not Th2 cells. T-bet is required for the formation and maintenance of H4Ac marks downstream of Ifng; however, this region is not required for IFNG expression (10, 25). The genomic region between Tmevpg1 and Ifng contains a number of Th1-specific DNase 1 HS sites to which the inducible transcription factors NF-κB and Ets-1 are recruited in a T-bet–dependent fashion. Four of these sites possess strong transcriptional enhancer activity, whereas the fifth possesses transcriptional repressor activity. Further, pharmacological inhibition of NF-κB reduces enhancer activity of the HS1–HS4 elements, supporting that enhancement from these sites is mediated, in part, by NF-κB. Taken together, our data support a mechanism by which Tmevpg1 is transcriptionally regulated by T-bet via epigenetic mechanisms, enabling recruitment of inducible transcription factors to both the Tmevpg1 promoter and gene body, as well as distal transcriptional enhancers and repressors (Fig. 8). Our findings not only provide a detailed description of lncRNA gene regulation as part of a developmental program, but they also contribute to the accepted mechanism of regulation of Ifng during Th1 lineage commitment.

Our model is generally consistent with recent studies examining the requirement of Tmevpg1 for expression of Ifng in response to infection with Salmonella (24). Tmevpg1 expression in CD8⁺ T cells promotes rapid IFN-γ expression in response to Salmonella infection through formation of covalent H3K4me3 marks at the Ifng locus, correlating with active transcription and rapid expression of IFN-γ by CD8⁺ T cells (24). Moreover, in an in vitro assay, Tmevpg1 physically associates with the WDR5 component of the MLL/Set1 histone-modifying complex responsible for producing the H3K4me3 mark (24). These findings are generally supported by studies examining functions of other lncRNAs. Although repressive lncRNAs mediate epigenetic regulation of genes through association with PRC2-containing histone complexes, lncRNAs imposing positive regulation on target genes are commonly found to associate with the MLL/Set1 complex and their associated proteins.

These results, combined with previous published studies, support a model whereby Stat4 and T-bet play integral roles in epigenetic remodeling of the Ifng locus spanning >300 kb. In addition to the activity of Ifng and Tmevpg1 proximal promoter/enhancers, distal enhancers seem to be divided into those used to drive Ifng expression at discrete stages of differentiation from initial Th1 polarization to responses by memory T cells and those used to drive Tmevpg1 expression. Transcription factors, such as NF-κB and Ets-1, which respond to TCR stimulation, are critical for Tmevpg1 expression and Ifng expression at later stages of differentiation. Results also support the notion that Tmevpg1 recruits the MLL/SET1/WDR complex to Ifng to establish a network of H3K4 methylation marks. Finally, three principal CTCF binding sites, one at the 3′ end of the Tmevpg1 gene, one within Ifng, and one upstream of Ifng, cooperate to loop this genomic region together (39, 40). It was suggested that these CTCF binding sites serve an insulator function to protect Ifng from the silencing effects of neighboring heterochromatin. We suggest that our results are more consistent with a model whereby CTCF sites are actually used to bring the Tmevpg1 gene in close proximity to the Ifng gene to facilitate localization of Tmevpg1 RNA transcripts to the Ifng genomic region so that Tmevpg1 RNA transcripts can carry out their function of enhancing Ifng transcription.

We thank Mark Boothby and Christopher Williams (Department of Pathology, Microbiology and Immunology, Vanderbilt University School of Medicine) for providing the Tbx21-knockout mice. We also thank Scott Collier for helpful discussions and comments.

The authors have no financial conflicts of interest