BATF-Interacting Proteins Dictate Specificity in Th Subset Activity Free

In response to the signals from APC, naive CD4⁺ T cells differentiate into distinct Th subsets, including Th1, Th2, Th17, regulatory T (Treg), and Th9 cells (1). Th cells are critical in the adaptive immune system by producing multiple cytokines and chemokines. In many subsets, a master transcription factor has been identified that defines the Th lineages, such as T-box transcription factor (T-bet) in Th1 cells or forkhead box P3 (Foxp3) in Tregs (2). However, emerging data show that the concept of master transcription factor is not sufficient to define the complexity of the Th cell phenotypes. The transcription factors shared by multiple Th subsets also play a critical role in Th cell differentiation. To further understand the underlying mechanism of Th cell differentiation, it is necessary to define how the transcription factors expressed in multiple lineages lead to the development of specific Th cells and the interplay among those transcription factors in different Th subsets.

IL-9 is the major cytokine produced by Th9 cells, a recently defined Th subset (3). IL-9– and IL-9–producing cells have been linked to autoimmunity, allergic disease, asthma, the immunity to parasites, and antitumor immunity (4, 5). The development of Th9 cells requires IL-4, TGF-β, and IL-2. Each of these cytokines activates specific signaling pathways and downstream transcription factors to promote IL-9 expression (6). The development of Th9 cells requires a balance of signaling pathways and a network of transcription factors including STAT6, basic leucine zipper transcription factor ATF-like (BATF), IFN regulatory factor 4 (IRF4), Smad proteins, PU.1, and STAT5 (7–12).

BATF belongs to the AP-1 transcription factor family that has a basic DNA-binding region and regularly spaced leucine residues termed a basic leucine zipper (bZIP). BATF requires heterodimer formation with other transcription factors such as Jun family members to regulate target gene expression, possibly because it lacks a transactivation domain (13). BATF is required for the differentiation of multiple Th subsets. In Th9 cells, BATF cooperates with IRF4 to promote IL-9 production in Th9 cells (11, 12). Moreover, BATF and IRF4 cooperate in the induction of IL-17 production in Th17 cells (14, 15). However, the mechanism through which BATF selectively regulates the lineage-specific cytokine production in various Th subsets is still not well defined.

Because BATF plays an essential role in driving the differentiation of Th9 and Th17 subsets, we questioned how BATF selectively promoted IL-9 production in Th9 cells, but not in Th17 cells. We found that there is preferential expression of BATF and its binding partners between Th9 and Th17 cells. The low expression of JunB and c-Jun in Th17 cells limits the capacity of BATF to induce IL-9 in Th17 cells. Broad complex/tramtrack/bric a brac and Cap‘n’collar homology 2 (Bach2) also regulates the activity of AP-1 transcription factors by competing for binding sites on target genes (16, 17). Surprisingly, we find that Bach2 amplified the function of BATF by binding to the Il9 locus and by regulating the expression of transcription factors that induce IL-9 expression. Thus, our study suggests that one factor dictating the specificity of BATF in Th subsets is the availability of BATF binding partners.

C57BL/6 mice were purchased from The Jackson Laboratory. Batf^ΔZ/ΔZ (C57BL/6), Irf4^fl/fl, and Bach2^fl/fl mice were previously described (11, 18, 19). All experiments were performed with the approval of the Indiana University Institutional Animal Care and Use Committee.

Naive CD4⁺CD62L⁺ T cells were isolated from the spleens and lymph nodes of the mice by using the magnetic separation following the supplier’s protocol (Miltenyi Biotec, Auburn, CA). Cells were cultured in complete RPMI 1640 on plates coated with anti-CD3 and soluble anti-CD28. Cell cultured under Th9 condition were supplemented with human TGF-β1 (2 ng/ml), murine IL-4 (20 ng/ml), human IL-2 (hIL-2; 50 U/ml), anti–IL-10R (10 mg/ml), and anti–IFN-γ (10 mg/ml). Th2 cells were cultured with hIL-2 (50 U/ml), murine IL-4 (20 ng/ml), and anti–IFN-γ (10 mg/ml). Th17 cells were cultured with murine IL-6 (100 ng/ml), human TGF-β1 (2 ng/ml), murine IL-1β (10 ng/ml), murine IL-23 (10 ng/ml), anti–IFN-γ (XMG; 20 mg/ml), anti–IL-4 (11B11; 10 mg/ml), and anti–IL-2 (10 mg/ml). After the initial 3 d culture, cells were removed from anti-CD3–coated plates and expanded with three volume of fresh media containing the same cytokine concentrations.

Human naive CD4⁺ T cells were isolated from human PBMCs by using magnetic separation (Miltenyi Biotec). The cells were activated with a receptor cross-linking bead, Dynabeads Human T-Activator CD3/CD28 (Thermo Fisher Scientific) in complete media to generate Th9 cells (20 ng/ml hIL-4, 2 ng/ml human TGF-β1, 50 U/ml hIL-2, and 10 μg/ml anti–IFN-γ) and Th17 cells (2 ng/ml human TGF-β1, 20 ng/ml hIL-6, 10 ng/ml hIL-23, 10 ng/ml hIL-1β, 10 μg/ml anti–IL-4, and 10 μg/ml anti–IFN-γ). Cells were grown at 37°C under 5% CO₂ and were expanded after 3 d with original concentration of cytokines in fresh medium. Cells were harvested on days 5 and 7 for analysis.

Total RNA was isolated from cells cultured for 5 d using TRIzol (Life Technologies). RNA was reverse transcribed according to manufacturer’s directions (Quantabio, Beverly, MA). Quantitative RT-PCR (qRT-PCR) was performed with commercially available primers (Life Technologies) with a 7500 Fast-PCR machine (Life Technologies). Gene expression was normalized to housekeeping gene expression (β2-microglobulin). For cytokine staining, CD4⁺ T cells were stimulated with PMA (50 ng/ml; Sigma-Aldrich) and ionomycin (500 ng/ml; Sigma-Aldrich) for 2 h, followed by monensin (2 μM; BioLegend) for total 5 h at 37°C. After restimulation, cells were washed with FACS buffer (PBS with 0.5% BSA). Cells were then stained with a fixable viability dye (eBioscience) and surface markers (CD4, RM4-4, CD90.1, and OX-7; BioLegend) for 30 min at 4°C, followed by washing and fixation with 4% formaldehyde for 10 min at room temperature. Cells were then permeabilized with permeabilization buffer (eBioscience) and stained for cytokines (IL-4, 11B11, IL-9, RM9A4, IL-10, and JES5-16E3 [BioLegend] and IL-17A and eBio17B7 [eBioscience]). For transcription factor staining, after surface staining, cells were fixed with Fixation and Permeabilization Buffer (eBioscience) for 2 h at 4°C and then permeabilized with permeabilization buffer (eBioscience). Transcription factors (BATF [BD Biosciences], IRF4 [BioLegend], JunB [Santa Cruz], and Bach2 [Bioss]) were stained for 30 min at 4°C.

Bicistronic retroviral vectors containing BATF, IRF4, JunB, c-Jun, dominant-negative Jun (DN-Jun), and Bach2 were used to generate retrovirus as previously described (20). Briefly, after 1 d of culture, activated CD4 T cells were infected with the retrovirus containing the interested gene or their empty vectors as controls in the presence of polybrene (8 μg/ml; Sigma-Aldrich) by centrifugation at 2300 rpm for 90 min at 32°C. After centrifugation, the supernatant was replaced with Th cell–polarizing media. Cell were expanded on day 3 and analyzed on day 5.

Activated T cells were transfected on day 1 with 1 μM SMARTpool small interfering RNA (siRNA) targeted the specific mRNA or the nontarget control (Thermo Scientific Dharmacon) in the Accell siRNA Delivery Media with the Th cell condition Abs and cytokines described as above. Cells were harvested for qRT-PCR analysis on day 3 and cytokine production on days 4 and 5.

Whole-cell protein lysates were extracted from Th9 and Th17 cells on day 5. Cell lysates were precleared with protein A–agarose bead slurry (50%; MilliporeSigma) for 1 h at 4°C. Equal amount of protein was incubated with BATF Ab (Cell Signaling Technology) at 4°C overnight with rotation. Immunocomplexes were captured by protein A–agarose for 4 h at 4°C with rotation and eluted at 100°C for 5 min in Laemmli sample buffer. Immunoprecipitates were separated by 4–20% SDS-PAGE and immunoblotted with BATF, JunB, c-Jun, and Bach2 Abs (Cell Signaling Technology).

CD4 T cells were cross-linked with 1% formaldehyde for 15 min at room temperature with rotation. The reaction was quenched with 0.125 M glycine for 5 min. Cells were lysed by adding the cells lysis buffer for 10 min on ice and nuclear lysis for 20 min on ice. Nuclear extracts were fragmented by sonication. After preclearing with Salmon Sperm DNA (Thermo Fisher Scientific), BSA, and protein A–agarose bead slurry (50%; MilliporeSigma), the supernatant was incubated with either rabbit polyclonal BATF, JunB, IRF4, Bach2 (Cell Signaling Technology), H3K4me3, H3K27ac (Abcam) Ab, or normal rabbit IgG (MilliporeSigma) at 4°C overnight with rotation. The immunocomplexes were incubated with protein A–agarose beads for 4 h at 4°C, washed, eluted, and reverse cross-linked at 65°C overnight. DNA was purified by phenol/chloroform extraction and resuspended in 200 μl of nuclease free water and analyzed by qRT-PCR as previously described. The supernatant from IgG samples was kept as input samples. After normalization to the input DNA, the amount of output DNA of each target protein was calculated by subtracting that of the IgG control. Chromatin immunoprecipitation (ChIP) primers were used as described (20).

All data analysis was done using Prism software version 7 (GraphPad Software) using a Student t test. Post hoc Tukey test was used for multiple comparisons. A p value ≤0.05 was considered statistically significant.

BATF cooperates with IRF4 in driving Th subset differentiation (13), and IL-9 production is diminished in BATF- and IRF4-deficient Th9 cultures (11, 12). Thus, IRF4 might be a key factor that affects the specificity of BATF in Th subsets. To test this, we measured BATF and IRF4 expression in Th9 and Th17 cells. Th9 cells showed greater expression of both BATF and IRF4 than Th17 cells (Fig. 1A). To examine whether the low expression of BATF or IRF4 limits IL-9 production in Th17 cells, Th9 and Th17 cells were transduced with BATF- or IRF4-expressing retrovirus separately (Fig. 1B). The transduction significantly increased the protein level of BATF or IRF4 in the cells (Fig. 1B). Consistent with previous studies (11, 15), BATF and IRF4 promoted IL-9 production in Th9 cells and IL-17 production in Th17 cells (Fig. 1C). In contrast, BATF or IRF4 transduction failed to enhance IL-9 production in Th17 cells or IL-17 production in Th9 cells (Fig. 1C). The inability to enhance IL-9 production in Th17 cultures might be due to a requirement for high amounts of both BATF and IRF4. Therefore, Th9 and Th17 cells were cotransduced with BATF and IRF4. Similar to what we observed with the single transduction of BATF or IRF4, cotransduction of BATF and IRF4 only promoted IL-9 production in Th9 cells and was still not capable of inducing IL-9 production in Th17 cells (Fig. 1D). BATF and/or IRF4 transduction also failed to increase IL-9 production in Th2 cells (Supplemental Fig. 1). Together, these data indicate that the low expression of BATF and/or IRF4 is not the sole limiting factor for IL-9 production in Th17 cells. Thus, other factors must contribute to the specificity of BATF in regulating IL-9 production.

To determine the binding partners of BATF in Th9 and Th17 cells, we performed coimmunoprecipitation and found several factors that interacted with BATF, including JunB and c-Jun (Supplemental Fig. 2). Heterodimers of BATF and Jun family members regulate cytokine production in Th2, Th17, and T follicular helper cells (18, 21–26). Yet, the function of Jun family members in Th9 cells has not been studied. Therefore, we hypothesized that Jun family members were crucial partners of BATF in Th9 cells. Based on our previous microarray data, the expression of Jun family members is higher in Th9 cells than in Th2 and Treg cells (11). In this study, we compared the expression of Jun family members in mouse Th9 and Th17 cells. Th9 showed greater expression of JunB protein (Fig. 2A). Both Junb and Jun had higher expression in Th9 cells than in Th17 cells, whereas Jund had similar expression in Th9 and Th17 cells (Fig. 2A). Consistent with the expression pattern in mouse T cells, human Th9 cells also showed higher expression of JUNB and JUN than in Th17 cells (Fig. 2B).

Next, we investigated if a specific Jun family member was required for IL-9 production in Th9 cells. To approach this, Th9 cells were transfected with JunB-, c-Jun– or JunD-specific siRNA (Fig. 2C, Supplemental Fig. 3). The knockdown efficiency was confirmed 48 h after the transfection (Fig. 2C, Supplemental Fig. 3). Knocking down Junb or Jun did not interfere with the expression of other Jun family member expression (Fig. 2C). IL-9 expression was diminished when the expression of Junb or Jun was decreased (Fig. 2C). However, decreased Jund expression did not affect IL-9 expression (Supplemental Fig. 3). The reduced expression of Junb or Jun did not affect Batf or Irf4 expression in Th9 cells. These results indicate that JunB and c-Jun promote IL-9 production in Th9 cells.

The Il9 gene has been characterized by three conserved noncoding sequences (CNS): the promoter (CNS-1), a region 6 kb upstream of the transcriptional start site (CNS-0), and a region 5.4 kb downstream of the transcriptional start (CNS-2) (6). Our group recently described an enhancer region, 25 kb upstream of the Il9 gene, that promoted Il9 gene expression (Fig. 2D) (20). In our previous study, we found that BATF bound to the Il9 promoter and enhancer regions in Th9 cells (11, 20). As a major binding partner of BATF, it is possible that Jun family members directly bound to the Il9 gene and interacted with BATF at the gene locus. As expected, JunB strongly bound at the Il9 promoter in Th9 cells, but not in Th17 cells (Fig. 2D). Binding in Th9 cells was predominantly in the promoter region. The binding of JunB at the Il9 promoter is decreased in both BATF-deficient and IRF4-deficient Th9 cells (Fig. 2D). This result suggests that IRF4 stabilizes a BATF/JunB heterodimer at the Il9 gene in Th9 cells.

To investigate if Jun family members are required by BATF to promote IL-9 production, we used a DN-Jun retroviral vector (27). The N-terminal activation domain of Jun is truncated, so it dominantly forms a DNA-binding complex with BATF predicted to inhibit BATF function. DN-Jun transduction significantly inhibited IL-9 expression in Th9 cells, indicating that the transactivation domain of Jun family members is indispensable for IL-9 production in Th9 cells (Fig. 2E). To further confirm the transactivation domain of Jun family members is required by BATF in Th9 cells, we cotransduced BATF and DN-Jun into the Th9 cells. Similar to the result in Fig. 1C, ectopic transduction of BATF increased IL-9 production (Fig. 2F). However, the BATF and DN-Jun cotransduced cells expressed significantly less IL-9 than cells transduced with BATF alone (Fig. 2F). These data indicate that DN-Jun inhibits the function of BATF in Th9 cells and further supports that the transactivation domain of JunB and c-Jun family members is required by BATF to promote IL-9 production in Th9 cells.

Jun family members are required for IL-17 production in Th17 cells, although the structural requirements are not well defined (22, 23, 28, 29). To test the requirement for the Jun family member transactivation domain in IL-17 regulation, Th17 cells were transduced with DN-Jun expressing retrovirus. Surprisingly, IL-17 production was unchanged, suggesting that the Jun transactivation domain is not required for IL-17 production (Fig. 3A). Compared with Th9 cells, Th17 cells showed lower expression of JunB and c-Jun (Fig. 2A). Therefore, the low expression of JunB and c-Jun might restrict the capacity of BATF to induce IL-9 production in Th17 cells. However, the transduction of JunB or c-Jun failed to increase IL-9 production in Th17 cells (Fig. 3B). This might be due to a requirement in Th17 cells for increased expression of JunB/c-Jun coupled with increased BATF. To test this hypothesis, we cotransduced BATF with JunB or c-Jun into Th17 cells (Fig. 3C). Compared with control vector–transduced cells, cotransduction of BATF and JunB or c-Jun significantly increased IL-9 production, albeit at percentages much less than those observed in Th9 cultures (Fig. 3C). In all, these results demonstrate that the low expression of JunB and c-Jun are limiting factors in the ability of BATF to promote IL-9 production. Moreover, JunB and c-Jun functionally cooperate with BATF in promoting IL-9 production, and the requirement for the Jun family transactivation domain distinguishes the contribution of family members at specific cytokine loci.

Bach2 represses the production of multiple cytokines in effector CD4 T cells and maintains the stability of Tregs (30). Additionally, Bach2 restrains the ability of BATF to promote IL-4 production in Th2 cells (18). BATF immunoprecipitation showed that Bach2 coimmunoprecipitated with BATF (Supplemental Fig. 2), which suggested that Bach2 might also be a regulator of BATF activity in IL-9 production. Bach2 is expressed in both mouse and human Th9 cells and Th17 cells, with greater expression in Th17 cells (Fig. 4A, 4B). To test the binding of Bach2 to the Il9 gene, ChIP assay was performed in Th9 and Th17 cells. To our surprise, the binding of Bach2 at the Il9 promoter in Th9 cells was much higher than in Th17 cells, even though Bach2 showed greater expression in Th17 cells (Fig. 4C). The binding of Bach2 on the Il9 promoter region was modestly decreased in BATF-deficient Th9 cells but was undetectable in IRF4-deficient Th9 cells (Fig. 4C). These results suggested that Bach2 binding at the Il9 promoter was dependent on IRF4.

Because Th9 cells demonstrated lower Bach2 expression than Th17 cells, we wanted to determine if ectopic expression of Bach2 altered IL-9 production in Th9 cells. To test this, we introduced Bach2 in Th9 cells by retroviral transduction. Interestingly, ectopic Bach2 increased the production of IL-9 (Fig. 4D). Furthermore, Bach2 transduction promoted the expression of BATF (Fig. 4D).

We previously reported that IL-10 was a repressor of IL-9 production in Th9 cells (31), and Bach2 can repress IL-10 (32). Even though we compensate for the repressive effect of IL-10 by incubating Th9 cultures in the presence of anti–IL-10R, we wanted to determine if decreased endogenous IL-10 production could explain the result through an indirect mechanism. However, Bach2 induced IL-10 production in a small percentage of cells (Fig. 4D), suggesting that limiting IL-10 production is not an intermediate step in the ability of Bach2 to regulate IL-9.

To further assess the requirement of Bach2 for IL-9 production, Bach2 expression was inhibited by Bach2-specific siRNA. The knockdown of Bach2 significantly diminished IL-9 production and other IL-9–inducing transcription factors, including Batf, Irf4, Junb, and Jun, whereas lower Bach2 did not significantly change Il10 expression (Fig. 4E). Similarly, Bach2-deficient Th9 cells showed decreased IL-9 production (Fig. 4F). However, Bach2 siRNA did not diminish the low amount of IL-9 seen early in Th2 differentiation (Supplemental Fig. 4A) (12). Furthermore, the knockdown of Bach2 in Th9 cells results in decreased H3K4me3 at the promoter and decreased H3K27ac at the Il9 CNS-25 (Fig. 4G). Taken together, these data suggest that Bach2 promotes IL-9 production by directly altering chromatin modifications at the Il9 gene locus and/or indirectly by enhancing IL-9–inducing transcription factor expression in Th9 cells.

To then determine if Bach2 cooperates with BATF in Th9 cells, we cotransduced BATF/internal ribosomal entry site (IRES)/enhanced GFP with BACH2/IRES/Thy1.1 retroviruses into Th9 cells and analyzed IL-9 production on day 5 (Fig. 4H). By gating on GFP⁺, Thy1.1⁺, and GFP⁺Thy1.1⁺ populations, we distinguished cells transduced with BATF alone, Bach2 alone, or both BATF and Bach2, respectively (Fig. 4H). In agreement with the previous results, BATF- or Bach2-transduced cells showed increased IL-9 expression. Importantly, cells transduced with both BATF and Bach2 and showed significantly increased IL-9 production compared with IL-9 production in BATF- and Bach2-single-transduced cells, implying that Bach2 augments the ability of BATF to promote IL-9 production (Fig. 4H, Supplemental Fig. 4E). Thus, Bach2 cooperates with BATF in promoting IL-9 production in Th9 cells.

Because Bach2 cooperated with BATF in regulating IL-9 production in Th9 cells, we speculated that Bach2 might also regulate IL-9 production in Th17 cells. To define Bach2 function in Th17 cells, a Bach2-expressing retrovirus was transduced in Th17 cells. Transduction of Bach2 inhibited IL-17 production but did not enhance IL-9 production in Th17 cells (Fig. 5A). Because Bach2 cooperated with BATF in inducing IL-9 production in Th9 cells, we hypothesized that high expression of both BATF and Bach2 is required for IL-9 production in Th17 cells. To test this hypothesis, we cotransduced BATF/GFP and Bach2/Thy1.1 into Th17 cells (Fig. 5B). Compared with the cells transduced with the control vectors, cotransduction of BATF and Bach2 significantly increased the IL-9 production to over 20% in Th17 cells, near the amount of IL-9 production in Th9 cells (Fig. 5C). Taken together, these data indicate that BATF and Bach2 can cooperate to induce IL-9 production in Th17 cells (Fig. 6).

Some Th subsets express unique transcription factors, such as T-bet in Th1 cells. However, multiple Th subsets share common transcription factors (2, 6). The mechanism that contributes to the specificity of the shared transcription factors in various Th subsets is still not clear. In this study, we find the binding partners affect the specificity of BATF in inducing IL-9 production in Th9 and Th17 cells. We demonstrate JunB and c-Jun are essential for the ability of BATF to promote IL-9 expression in Th9 cells. It was interesting to find that instead of functioning as a transcriptional repressor, Bach2 positively regulates IL-9 production. JunB/c-Jun and Bach2 facilitate the capacity of BATF to promote IL-9 production in Th9 cells.

IRF4 has been shown to interact with BATF-Jun and bind to AP-1–IRF composite elements on target genes in Th2 and Th17 cells (13, 22, 33). Previously, we demonstrated that BATF cooperated with IRF4 for IL-9 production in Th9 cells. The binding of BATF or IRF4 was diminished in the reciprocal gene-deficient Th9 cells (11). Initially, we thought IRF4 might limit the ability of BATF to promote IL-9 production in Th17 cells. Surprisingly, transduction of BATF and/or IRF4 was not sufficient to induce IL-9 production in Th17 cells, which indicated additional factors interact with BATF alone or BATF and IRF4 to regulate IL-9 production. Importantly, the binding of BATF-containing complexes with JunB or Bach2 are dependent on IRF4 in Th9 cells. As IRF4 also seems to integrate TGF-β signals (34), it is clearly a critical factor in organizing the IL-9-inducing enhanceosome.

Our results do not diminish the importance of distinct Jun family members in other subsets. Both JunB and JunD interact with BATF to regulate IL-4 and IL-13 production in Th2 cells (18, 21). Jun family members also facilitate the DNA-binding activity of BATF and IRF4 on target genes in Th17 cells (28). In this report, we demonstrated that JunB and c-Jun functionally cooperated with BATF in promoting IL-9 production. The specificity of BATF in Th9 and Th17 cells is also distinguished by the requirement for the transactivation domain of Jun family members, with results suggesting that the activation domain of Jun family partners can compensate for the absence of the Jun activation domain. In contrast, JunD knockdown did not impair IL-9 production. It was striking to find that JunB or c-Jun knockdown did not affect the expression of other Jun family members in Th9 cells (Fig. 2), which is in contrast with the ability of JunD to functionally compensate for the loss of JunB in Th17 cells (29). This is indicative of JunB and c-Jun having nonredundant function in promoting IL-9 production in Th9 cells. Whether this is entirely through interactions with BATF is not clear. JunB also interacts with Fos-related antigen 1 (FRA1) to directly regulate IL-17 production in Th17 cells (35), whereas the c-Jun/c-Fos complex inhibits IL-17 production in Th17 cells (36). Together, these data suggest that Jun family members have distinct functions among the Th subsets and that the balance of function of Fos/Jun and BATF/Jun family members likely dictates gene expression patterns.

Although Bach2 represses IL-4, IL-13, and IFN-γ (16, 30), we observe that Bach2 promotes IL-9 production, suggesting its transcription regulatory function is gene dependent. Bach2 induces cytokines in other contexts, including IL-2 (37) (Supplemental Fig. 4B). Bach2 also has unique effects on the Il10 gene in Th9 cells. Bach2 transduction promotes IL-10 production in Th9 cells, which is different from the inhibitory effect of Bach2 on IL-10 in naive CD4 T cells and induced Tregs (32, 38). However, Bach2 knockdown did not affect Il10 expression in Th9 cells (Fig. 4), suggesting that retroviral expression of Bach2 might have nonphysiological effects on the Il10 gene. The mechanism of switching between activating and repressing Bach2 functions is still not clear. Previous studies have shown that Bach2 antagonizes Bcl6 and Blimp1 in the development of B cells (39–41). Both Bcl6 and Blimp1 have repressive effects on IL-9 production (42, 43). Thus, it is possible that Bach2 could promote IL-9 production by inhibiting the repressive effectors of Blimp1 and Bcl6. Bach2 also regulates the expression of IL-9–inducing transcription factors in Th9 cells (Fig. 4), and some of the regulatory effects of Bach2 could be indirect. BATF also cooperates with c-Maf in promoting IL-4 production in T follicular helper cells (24). Bach2 forms a heterodimer with Maf family members and binds to Maf-recognition elements on target gene locus (16). We previously reported that ectopic c-Maf expression inhibited IL-9 production in Th9 cells (9). Bach2 might compete with c-Maf in repression of the Il9 locus. Bach2 might form an active complex with BATF that inhibits the interaction between BATF and c-Maf, a potential repressive complex for IL-9 production. Modulating c-Maf activity might also explain why Bach2 shows opposite effects in Th2 and Th9 cells. Further investigations are expected to reveal the functions of the complex and context-dependent interactions among these transcription factors.

Our studies have focused on in vitro–derived Th9 and Th17 cultures. The question remains as to what effect Bach2 has during in vivo Th9 development. Previous reports on Bach2 function demonstrated amplified type 2 inflammation (18). Indeed, in that model, there is detectable Il9 expression above the naive background levels (Supplemental Fig. 4C). Coupled with our in vitro results, these data suggest that Bach2 is an efficiency factor for IL-9 production, and not a critical factor for expression. In agreement with this idea, Th9 cultures with either knockdown or Cre-mediated deletion of Bach2 resulted in only a partial decrease in IL-9 production. The ability of Bach2 to control Th2 cytokines (18), even in Th9 cultures (Supplemental Fig. 4D), obscures our ability to distinguish the effects of Bach2 on Il9. It is possible that a model of Th9 function that is independent of type 2 inflammation would allow isolation of the effects of Bach2 on IL-9 production in vivo, and that will be an important experiment for the future.

BATF binds to the Il9 promoter and CNS-25 enhancer regions in Th9 cells (11, 20). Results from the ChIP assay shows both JunB and Bach2 bind preferentially to the Il9 promoter region. Binding of all three factors is diminished in Th17 cells, compared with Th9 cells. Despite the Il9 CNS-25 enhancer being required for optimal IL-9 production and BATF binding to this region (20), binding of Bach2 and JunB was not significantly detected in this region. Although we have shown IRF4 binds to the CNS-25 region in Th9 cells, it is possible that BATF cooperates with other partners at the enhancer region to activate Il9 gene expression. The ability of BATF to integrate site-specific complex formation might be a key feature of its ability to regulate Il9.

Although Il9 transcription relies on many transcription factors, it is important to note that overexpression of any one bZIP factor is not sufficient to transdifferentiate Th17 cells into IL-9 producers. Transduction of BATF, Bach2, or Jun family members was not sufficient to induce IL-9 in Th17 cells (Fig. 6). The combination of BATF and Jun family members was sufficient to induce modest IL-9 in Th17, whereas BATF and Bach2 were more potent at inducing IL-9. However, even the combination of BATF and Bach2 did not induce IL-9 to the amounts observed in Th9 cultures (Fig. 6). This highlights that although ectopic transcription factor expression likely shifts the equilibrium from a closed chromatin state in Th17 cells to a more open configuration similar to Th9 cells by recruiting additional chromatin modifying enzymes to the target loci, that shift might not be sufficient to recapitulate the chromatin signature at the target gene compared with a cell cultured under optimal cytokine-inducing conditions. The chromatin structure at the Il9 locus in Th17 cells is likely another limiting factor in activating expression. Thus, in activating IL-9 production, BATF and Bach2 are not rewriting the network required for Th9 differentiation and not recreating the optimal Th9 phenotype.

These studies highlight multiple mechanisms at the core of transcription factor specificity in Th subsets. First, there may be differences in the amount of expression. BATF and Jun family members are expressed in all Th subsets, but in varying amounts (11). Second, the requirement for interacting proteins may differ among subsets. BATF appears to require IRF4 for gene induction in multiple subsets (11, 13, 22), but the requirement for specific Jun family members, and Jun functional domains, might be distinct among subsets. Third, there are likely temporal differences in expression among BATF-interacting proteins as well as time-dependent, and perhaps subset-specific, differences in the signals that regulate heterodimer formation. Finally, bZIP-interacting proteins might have distinct functions in each subset. Bach2 represses IL-17 production, but induces IL-9 production. Each of these mechanisms likely contributes to the specificity of transcription factor function.

We thank Drs. Baohua Zhou and Alexander Dent for review of this manuscript.

The authors have no financial conflicts of interest.