Repression of interleukin-4 in T helper type 1 cells by Runx/Cbfβ binding to the Il4 silencer

We first examined expression of Runx proteins during differentiation of CD4⁺ Th cells. Purified CD4⁺CD25⁻CD62L⁺ naive T cells were stimulated with immobilized anti-CD3 and soluble anti-CD28 antibodies. 2 d after stimulation, Runx1 protein was substantially decreased (Fig. 1). After another 4 d of culture, although expression of Runx1 protein was restored and detected in both Th1 and Th2 cells, Runx3 protein was detected almost specifically in Th1 cells (Fig. 1). Thus, both Runx1 and Runx3 proteins are expressed in polarized Th1 cells. Expression of distal (P1) promoter–derived Runx1 or Runx3 transcript was well correlated with that of Runx1 or Runx3 protein, suggesting that activation of a distal promoter is important for regulated expression of Runx proteins. Considering the redundant function of Runx1 and Runx3 in Cd4 silencing in CD8⁺ T cells (11), it is also possible that these two transcription factors function redundantly in CD4⁺ T cells. Because association with the nonredundant Cbfβ protein is essential for the function of both Runx1 and Runx3, we analyzed the effect of loss of Cbfβ on Th cell differentiation (10, 16).

Because germline-null mutations of Cbfβ and Runx3 result in embryonic and neonatal lethality, respectively (16–18), we generated Loxp-flanked Cbfβ^fllox(Cbfβ^f) and Runx3^flox (Runx3^f) mutant alleles by gene targeting (Fig. S1). Mice harboring either Cbfβ^f or Runx3^f alleles were crossed with Lck-Cre or Cd4-Cre transgenic mice, to inactivate the targeted genes at CD4⁻CD8⁻ DN or CD4⁺CD8⁺ DP stages, respectively. Whereas inactivation of Runx1 at the DN stage resulted in a more than fivefold reduction in the number of total thymocytes (9), the reduction was only approximately twofold in Cbfβ^f/f: Lck mice, although development of mature thymocytes was severely impaired (Fig. 2 A and Fig. S2). In contrast, the number of mature thymocytes was only moderately reduced in Cbfβ^f/f: Cd4 mice (Fig. 2 A and Fig. S2). Although Cre-mediated recombination of the Cbfβ^f allele appeared to be very efficient in DP thymocytes by both Lck- and Cd4-Cre transgene, a significant amount of Cbfβ protein could be detected in those cells from the Cbfβ^f/f: Cd4 mice (Fig. 2 B). However, in the peripheral TCRαβ cells from Cbfβ^f/f: Cd4 mice, no Cbfβ protein was detected (Fig. 2 E), indicating that Cbfβ protein was gradually lost after inactivation of the gene. Similarly efficient inactivation of Runx3^flox allele by Cd4-Cre transgene in thymus resulted in a loss of Runx3^flox allele in peripheral T cells (Fig. 2, C and F), which is consistent with loss of Runx3 protein in CD8⁺ T cells from Runx3^f/f: Cd4 mice (Fig. 1).

In peripheral lymphoid tissues from Cbfβ^f/f: Cd4 mice, mature TCRαβ T cells consisted of two major subsets, CD4⁺CD8⁻ and CD4⁺CD8^int cells (Fig. 2 D). Perforin expression in CD4⁺CD8^int cells was comparable to that in wild-type CD8⁺ T cells (Fig. S3), which is consistent with the CD4⁺CD8^int phenotype resulting from the loss of Cd4 silencing in CD8⁺ cytotoxic-lineage cells in the absence of Cbfβ protein and Runx complexes (9).

It has been shown that outbred Runx3-deficient mice develop a transient inflammatory infiltrate in their lungs and elevated serum IgE (13, 14). In Cbfβ^f/f: Cd4 mice, serum IgA, IgG1, and IgE titers were significantly elevated (Fig. 3 A). Numerous lymphocytes and eosinophils were found to infiltrate bronchioles, perivascular space, and alveolar septae in the lung from all Cbfβ^f/f: Cd4 mice examined (Fig. 3 B). Mild infiltration of lymphoid cells in the bronchioles and perivascular space, but not in alveolar septae, was also observed in about one-third of Runx3^f/f: Cd4 mice (Fig. 4). Thus, T cell–specific loss of Cbfβ protein led to spontaneous development of asthma-related features, and a similar, but milder, disease developed in mice lacking Runx3 in T cells. Because such asthma-related findings are often correlated with enhanced Th2 responses, we next examined cytokine production and differentiation of CD4⁺ T cells in the absence of Cbfβ.

Purified naive T cells from Cbfβ^f/f, Cbfβ^f/f: Cd4, and Runx3^f/f: Cd4 mice were stimulated with anti-CD3 and -CD28 antibodies. After 48 h, the level of IL-4 secreted from Cbfβ-deficient cells was 10-fold higher than that from control cells (Fig. 4 A). After an additional 4 d of culture with IL-2, cells were analyzed by intracellular IL-4 and IFNγ staining. Consistent with the higher IL-4 production observed after 2 d, IL-4–producing cells, including IL4/IFNγ double producers, were differentiated efficiently from Cbfβ-deficient naive CD4⁺ T cells (Fig. 4 B). The differentiation of IFNγ-producing Th1 cells was also enhanced by Cbfβ-deficiency by yet uncharacterized mechanisms.

When Cbfβ-deficient naive CD4⁺ T cells were cultured under Th1 polarizing conditions, in the presence of IL-4 neutralizing antibody, cells producing both IL-4 and IFNγ were detected (Fig. 4 B). These IL-4/IFNγ double producers were, thus, differentiated independently of IL-4 signaling. In contrast, only IL-4–producing Th2 cells were differentiated under Th2-skewed conditions. These results indicate that expression of both IL-4 and IFNγ in the same cell is an outcome of IL-4 derepression in Th1 cells, rather than IFNγ derepression in Th2 cells. Previous studies showed that constitutive expression of Gata-3 induced derepression of the Il4 gene in Th1 cells (19, 20). However, we observed no difference in the level of induced T-bet or Gata-3 (Fig. 4 C), indicating that IL-4 derepression was not a consequence of dysregulation of these transcription factors.

Because derepression of the Il4 gene was induced in Th1 cells upon deletion of the HS IV Il4 silencer, which contains a putative Runx recognition motif (5′-ACCRCA-3′) (7), we next examined whether Runx complexes directly associate with the Il4 silencer by chromatin immunoprecipitation (ChIP) assays. The Il4 silencer region was efficiently amplified from DNA precipitated with anti-Cbfβ2 antibody, but not with control antibody, from both naive CD4⁺ T cells and Th1 cells (Fig. 5 A). In sharp contrast, anti-Cbfβ2 antibody failed to precipitate the Il4 silencer from Th2 cells, although the antibody precipitated control Tcrβ enhancer, which is known to be regulated by Runx complexes, from both Th1 and Th2 cells. Thus, binding of Runx complexes is well correlated with the specificity of Il4 silencer activity.

The level of IL-4 production and the severity of asthma-related symptoms were higher in Cbfβ^f/f: Cd4 mice than in the Runx3^f/f: Cd4 mice (Figs. 3 and Figs.4). This discrepancy suggests a compensatory function within Runx family members in the regulation of Il4 silencer activity. Indeed, Runx1 protein is expressed in naive CD4⁺ T cells and Th1 cells (Fig. 1). Therefore, we analyzed whether Runx1 protein binds to the Il4 ­silencer by using an anti-Runx1 antibody in ChIP assays. To eliminate possible cross-reactivity of the Runx1 antibody with Runx3 protein, we used Runx3-deficient cells as a control. The Il4 silencer region was precipitated from both control cells and Runx3-deficient cells by the anti-Runx1 antibody (Fig. 5 B). Thus, it is likely that Runx1 is involved in regulating Il4 silencer function, at least when Runx3 is not present.

Our results indicated that Runx complexes dissociate from the Il4 silencer in Th2 cells, despite its expression. This result suggests that there is either a mechanism that inhibits Runx binding to the Il4 silencer in Th2 cells or one that permits binding only in Th1 cells. Therefore, we examined the effect of enforced expression of Th2- and Th1-specific factors, Gata-3 and T-bet, on the binding of Runx complex to the Il4 silencer. Although Gata-3 expression in Th1 cells induced dissociation of Runx complexes from the Il4 silencer, T-bet expression in Th2 cells only induced Runx3 expression, but not Runx complex association with the Il4 silencer (Fig. 5, C and D). Furthermore, Gata-3 induced IL4 expression in polarized Th1 cells (Fig. S4), as previously reported (20). These results demonstrate that Gata-3 functions to inhibit binding of Runx complexes to the Il4 silencer.

Our results are consistent with the recent description of T-bet–dependent Runx3 expression in Th1 cells and Runx3 binding to the Il4 silencer (21). However, based on the expression patterns of Runx1 and Runx3 proteins, we propose that Runx1 is mainly involved in repressing the Il4 gene in naive CD4⁺ T cells. When Runx1 expression is reduced after encounter with antigen, newly expressed Runx3 plays a role in initiating Il4 repression in the early phase of Th1 differentiation. This would be followed by maintenance of Il4 repression by both Runx1 and Runx3 proteins. In contrast, during Th2 cell differentiation, Gata-3 induces dissociation of Runx complexes from the Il4 silencer by a yet uncharacterized mech­anism and induces IL4 expression.

Collectively, our results demonstrate that loss of Runx complex function in CD4⁺ Th cells leads to spontaneous development of asthma-related symptoms caused by enhanced Th2 responses that are caused, at least in part, by failure of Il4 silencing. It is not documented whether Il4 silencer-deficient mice spontaneously develop similar symptoms, although impaired Th1-mediated immunity upon Leishmania major infection of these mice was reported (7). It is possible that, in addition to loss of Il4 silencing, additional mechanisms caused by loss of Runx complex function in T cells facilitate disease development. Alternatively, because an asthma-like phenotype was also attributed to loss of Runx3 function in dendritic cells (13, 14), we must consider the potential involvement of cells other than T cells in disease development. Because the human RUNX3 locus is closely linked to one of the asthma-susceptibility loci (15), dysfunction of Runx complexes may be involved in the human disease. A better understanding of Runx complex function during Th cell differentiation should provide important additional insights into the pathogenesis of allergic diseases.

The targeting vectors for generating Cbfβ^flox and Runx3^flox alleles were constructed in the pL2-Neo plasmid (8), with genomic fragments containing the loxP-flanked coding exon and neo^R gene inserted within the short 5′ side homology region (Fig. S1). The vector for the Cbfβ genomic fragment was obtained from S.-C. Bae (Chungbuk University, Cheongju, South Korea). Transfection into E14 ES cells was performed as previously reported (9). The Lck-Cre and Cd4-Cre transgenic mice were provided by C. Wilson (University of Washington, Seattle, WA). Mouse colonies were maintained in an animal facility in the Research Center for Allergy and Immunology RIKEN Institute, and experiments were performed according to the institutional guidelines for animal care.

Anti-Runx3 antibody was provided by Y. Ito (Institute of Molecular and Cell Biology, Singapore) (22). Anti-Runx1 and -Cbfβ2 antibodies were generated by immunizing rabbits with peptides corresponding to the N terminus of the distal promoter–derived Runx1 protein and to the C-terminal end of Cbfβ2, respectively. Anti-Gata3 (HG3-31) and −T-bet (4B10) antibodies were purchased from Santa Cruz Biotechnologies, and all monoclonal antibodies used for staining cells were obtained from BD Biosciences.

Naive CD4⁺CD25⁻CD62L⁺ T cells were sorted by flow cytometry. Differentiation of CD4⁺ T cells was induced as previously described (12). In brief, for inducing Th1 or Th2 differentiation, the naive cells stimulated with 2 μg/ml of immobilized anti-CD3 and 2 μg/ml of soluble anti-CD28 antibody were cultured in the presence of 5 ng/ml IL-12 and 1 μg/ml anti–IL-4 antibody or 10 ng/ml IL-4 and 1 μg/ml anti-IFNγ antibody and 1 μg/ml anti–IL-12 antibody, respectively, during the first 2 d. Cells were then maintained in the medium supplemented with 20 U/ml rIL-2 for an additional 4 d before staining for intracellular cytokines (12).

Whole-cell lysates were resolved by SDS-PAGE and transferred to Hybond-P membranes (GE Healthcare). The membranes were probed with an appropriate primary antibody, and immunocomplexes were detected using ECL reagents (GE Healthcare). Cytokine and serum immunoglobulin levels were assessed by ELISA using Quantikine (R&D Systems) and Mouse Ig ELISA Quantitation kits (Bethyl), respectively.

ChIP assays were performed according to protocols provided for the ChIP Assay kit (Millipore). Chromatin DNA was fragmented by sonication to a mean length of 500 bp, and was immunoprecipitated with control, anti-Cbfβ2, or -Runx1 antibody. The precipitated DNA was subjected to PCR amplification. The primer sequences used in ChIP assay and RT-PCR are described in Fig. S5.

The pMigRI–T-bet and the pMX–Gata-3 vectors were provided by S.L. Reiner (University of Pennsylvania, Philadelphia, PA) and M. Kubo (RIKEN Research Center for Allergy and Immunology, Yokohama, ­Japan), respectively. Naive T cells were stimulated in Th1 or Th2 conditions for 3 d, infected with retroviruses, and sorted 2 d after infection for analyses.

Fig. S1 shows the targeting strategy used to generate Cbfβ^flox and Runx3^flox alleles. Fig. S2 shows the decreased number of total and mature thymocytes in Cbfβ^f/f: Lck and Cbfβ^f/f: Cd4 mice. Fig. S3 shows the expression of perforin in peripheral T cells from Cbfβ^f/f: Cd4 mice. Fig. S4 shows the induction of IL4 expression in Th1 cells by Gata-3 transduction. Fig. S5 provides primer sequences used in this study.

We thank Chieko Tezuka for the genotyping of mice. We are grateful to Dr. Suk-Chul Bae for providing the vector containing the Cbfβ genomic fragment, and Dr. Yoshiaki Ito and Dr. Kosei Ito for providing anti-Runx3 antibody. We also thank Dr. Takeshi Egawa for his critical reading of the manuscript.

This work was supported by grants from Precursory Research for Embryonic Sciences and Technology, Japan Science and Technology Agency (I. Taniuchi), and the Sandler Program for Asthma Research (D.R. Littman).

The authors have no conflicting financial interests.