Gfi1-mediated Stabilization of GATA3 Protein Is Required for Th2 Cell Differentiation*

The differentiation of naive CD4 T cells into Th2 cells requires the T cell receptor-mediated activation of the ERK MAPK cascade. Little is known, however, in regard to how the ERK MAPK cascade regulates Th2 cell differentiation. We herein identified Gfi1 (growth factor independent-1) as a downstream target of the ERK MAPK cascade for Th2 cell differentiation. In the absence of Gfi1, interleukin-5 production and the change of histone modification at the interleukin-5 gene locus were severely impaired. Furthermore, the interferon γ gene showed a striking activation in the Gfi1–/– Th2 cells. An enhanced ubiquitin/proteasome-dependent degradation of GATA3 protein was observed in Gfi1–/– Th2 cells, and the overexpression of GATA3 eliminated the defect of Th2 cell function in Gfi1-deficient Th2 cells. These data suggest that the T cell receptor-mediated induction of Gfi1 controls Th2 cell differentiation through the regulation of GATA3 protein stability.

It has been established that effector T helper (Th) 2 cells can be classified into at least four subsets, namely Th1, Th2, Th17, and T regulatory cells. Th1 cells produce a large amount of IFN␥ and control cell-mediated immunity against intracellular pathogens, whereas Th2 cells produce IL-4, IL-5, and IL-13, and they are also involved in humoral immunity and allergic reactions (1)(2)(3). A recently identified new subset, Th17 cells, is thought to be involved in various inflammatory diseases. T regulatory cells are known to suppress various immune responses including those of autoimmune diseases (4). The direction of T helper cell differentiation depends on the types of cytokine in the environmental milieu (5). IL-12 and IFN␥ induce Th1 cell differentiation, whereas IL-4 and IL-2 induce Th2 cell differentiation (1,6,7). The combination of IL-6 and transforming growth factor-␤ is required for murine Th17 cell differentiation (8 -11), and the transforming growth factor-␤-dependent generation of T regulatory cells has also been reported (12,13).
In addition to the cytokines mentioned above, the activation of TCR-mediated signaling is also indispensable for Th cell differentiation. We previously reported that Th2 cell differentiation is highly dependent on the extent of TCR-mediated activation of the p56lck, calcineurin, and the Ras-ERK MAPK signaling cascade (14 -16). In particular, the inhibition of the activation of the Ras-ERK MAPK cascade caused a shift from Th2 to Th1 cell differentiation, thus suggesting that the direction of Th1/Th2 cell differentiation is controlled by the TCRmediated activation of the Ras-ERK MAPK cascade (15,17). On the other hand, Th1 cell development appeared to be regulated by another MAPK, c-Jun N-terminal kinase (JNK) (18,19).
Several transcription factors that govern Th2 cell differentiation have been reported. Among them, GATA3 appears to be a key factor for Th2 cell differentiation (20,21). The up-regulation of the GATA3 mRNA expression is selectively induced in developing Th2 cells by IL-4-mediated STAT6 activation (22,23). In addition to the transcriptional regulation, the expression of GATA3 is also regulated by a post-transcriptional mechanism. We recently reported that the Ras-ERK MAPK cascade controls GATA3 stability through the ubiquitin-proteasomedependent pathway (24).
To identify the candidate genes that play an important role in Th2 cell differentiation and are induced by the activation of the Ras-ERK MAPK cascade, a microarray analysis was performed. We found that the expression of Gfi1 (growth factor independent-1) is dramatically up-regulated by TCR stimulation in an ERK MAPK/calcineurin activation-dependent manner. Originally, Gfi1 has been reported to play an important role in promoting cell proliferation and in preventing apoptosis (25)(26)(27). In addition, Gfi1 is involved in the self-renewal of hematopoietic stem cells (28,29). Recently, the IL-4/STAT6-mediated induction of the Gfi1 expression in developing Th2 cells has been reported (30). Our results in this paper demonstrate that Gfi1 plays an important role in the stable expression of GATA3 protein and the subsequent differentiation of Th2 cells.

EXPERIMENTAL PROCEDURES
Mice-C57BL/6 mice and BALB/c mice were purchased from Clea. The Gfi1-deficient mice were generously provided by Dr. Stuart Orkin (Harvard Medical School, Boston, MA) (31). Gfi1-deficient mice were backcrossed more than 10 times with C57BL/6. All of the mice were maintained under specific pathogen-free conditions and then were used at 4 -8 weeks of age. All of the experiments using mice received approval from the Chiba University Administrative Panel for Animal Care. All of the animal care was conducted in accordance with the guidelines of Chiba University (Chiba, Japan).
Expression Plasmids and Gene Transfer-The expression plasmids were transfected into 293Tcells using FuGENE reagent (Roche Applied Science) according to the manufacturer's protocol. Retrovirus vector, pMXs-IRES-hNGFR (human nerve growth factor receptor p75) was generated from the pMXs-IRES-GFP plasmid by replacing the EGFP with the cytoplasmic region-deleted hNGFR cDNA. The method for the generation of virus supernatant and the infection into developing Th2 cells was described previously (33,34). Infected cells were detected by staining with anti-human NGFR mAb (C40-1457; BD Bioscience).
Chromatin Immunoprecipitation Assay-A ChIP assay was performed as previously described (35). Anti-acetylhistone H3-K9/14 antibody was purchased from Upstate Biotechnology, and anti-trimethylhistone H3-K4 antibody (ab8580) was from Abcam Co. (Cambridge, UK). The specific primers and TaqMan probes used in this experiments are described in the supplemental material or described previously (25,35,36). In the case of the semi-quantitative PCR analysis, the PCR products were resolved in an agarose gel and visualized by ethidium bromide. The images were recorded and quantified using an ATTO L&S analyzer (ATTO, Tokyo, Japan).
Quantitative RT-PCR-Total RNA was isolated using TRIzol reagent (Invitrogen). Reverse transcription was done using Superscript II (Invitrogen). For quantitative PCR, a TaqMan universal PCR Master Mix was used for all reactions (Applied Biosystems), and the ABI Prism 7000 sequence detection system was used (Applied Biosystems). The primers and TaqMan probes for the detection of mouse GATA3, Gfi1, IL-4, IL-5, IL-13, IFN␥, hypoxanthine phosphoribosyltransferase (HPRT), and human 18 S rRNA were purchased from Applied Biosystems. The expression of mRNA was normalized using the HPRT or human 18 S rRNA signal.

TCR-mediated Activation of the ERK MAPK and Calcineurin
Pathways Induces Gfi1 mRNA Expression-We previously reported that the TCR-mediated activation of the Ras-ERK MAPK cascade is required for Th2 cell differentiation (15,17). To identify candidate genes, which are induced as downstream targets of the Ras-ERK MAPK and regulate Th2 cell differentiation, a DNA microarray analysis was performed. Naïve CD4 T cells were stimulated with immobilized anti-TCR mAb under Th2 conditions (IL-2, IL-4, and anti-IFN␥) for 48 h in the absence or presence of a specific inhibitor of MEK1/2, U0126. Next, total RNA was prepared and subjected to a DNA microarray analysis. We focused on the nuclear factors, and the potentially interesting 27 genes were selected (Table 1). Among the candidate genes, Gfi1 mRNA was rapidly induced after TCR stimulation in CD4 T cells, cultured under both Th1 and Th2 conditions (supplemental Fig. S1A). The STAT6-dependent induction of Gfi1 mRNA has been previously reported (30). As shown in supplemental Fig. S1B, the induction (2 h of stimulation) of Gfi1 mRNA was not dependent on STAT6; however, the reduction was more severe in STAT6-deficient cells cul-tured under Th2 conditions for 5 days. In contrast, the induction of GATA3 mRNA expression was completely dependent on the STAT6 activation (supplemental Fig. S1B, right panel). We confirmed the ERK MAPK-dependent induction of Gfi1 mRNA in developing Th2 cells (Fig. 1A). Splenic CD4 T cells were stimulated under Th2 conditions in the absence or presence of the indicated inhibitors for 2 h, and the expression level of Gfi1 mRNA was determined by quantitative RT-PCR. As expected, the induction of Gfi1 mRNA expression was completely blocked by the treatment with MEK inhibitor (U0126). A calcineurin inhibitor, FK506, also exhibited an inhibitory effect on Gfi1 induction. In contrast, a specific inhibitor for the p38 MAPK (SB203580) and for the JNK had no effect on the expression of Gfi1 mRNA. These results indicated that the induction of Gfi1 mRNA at the early phase of T cell activation was dependent on the ERK MAPK and the calcineurin signaling pathways.
We next assessed the histone modification status of the Th2 cytokine gene loci and the IFN␥ gene locus in Gfi1 Ϫ/Ϫ developing Th2 cells by ChIP assay. As expected, the trimethylation level of H3-K4 at the IL-5 promoter locus was severely impaired in Gfi1 Ϫ/Ϫ Th2 cells, whereas the methylation at the IFN␥ promoter was enhanced (Fig. 1E). The methylation status of histone H3-K4 at the IL-4 promoter, IL-13 promoter, and RAD50 promoter in Gfi1 Ϫ/Ϫ Th2 cells was equivalent to those in the Gfi1 ϩ/ϩ Th2 cells (Fig. 1E). These results suggest that Gfi1 plays an important role in the acquisition of IL-5 production ability and for the suppression of the IFN␥ gene in developing Th2 cells.
Gfi1 Is Required for the Establishment of an Active Chromatin Status at the IL-5 Gene Locus-We previously reported the induction of Th2 cell-specific hyperacetylation at the intergenic region between the IL-5 and the RAD50 gene during Th2 cell differentiation, as well as the IL-5 gene locus (37). Therefore, we analyzed the histone modification status around the

Mybl2
Myeloblastosis oncogene-like 2 Ϫ7.5 Gfi1 Growth factor-independent 1 Interferon regulatory factor 4 Ϫ14.9 Egr2 Early growth response 2 Ϫ6.5 PBX1 Pre-B cell leukemia transcription factor 1 Ϫ6.5 Id3 Inhibitor of DNA binding 3 6.5 IRF7 Interferon regulatory factor 7 4.3 Klf2 Kruppel-like factor 2 39.4 Klf3 Kruppel-like factor 3 13.9 Klf7 Kruppel-like factor 7 4.3 Foxp1 Forkhead box P1 4.0 Foxp3 Forkhead IL-5 gene locus in Gfi1 Ϫ/Ϫ Th2 cells more precisely. A schematic representation of the IL-5 gene locus and the location of the designed primer pairs is indicated in Fig. 2A. The actual band patterns of each ChIP assay (Fig. 2B) and the relative band intensities (Me3 H3-K4/Input DNA or Ac H3-K9/14/Input DNA) of the eight primer pairs (Fig. 2C) were depicted. As previously reported, a long range Th2-specific histone modification was observed from the upstream region of the RAD50 promoter (corresponding to primer 1) through the end of the IL-5 intron 2 (primer 8) in Gfi1 ϩ/ϩ Th2 cells. Nevertheless, the H3-K4 methylation and H3-K9/14 acetylation status from the primer 2 to the primer 8 regions were lower in Gfi1 Ϫ/Ϫ Th2 cells, and the levels were similar to those in Th1 condition cells (Fig. 2C). The reduction of histone modifications of the IL-5 gene locus was confirmed by a ChIP assay with quantitative PCR (Fig. 2D). Neither of the modifications of the RAD50 promoter was affected (Fig. 2D). These results suggest that Gfi1 is required for the formation of active chromatin at the IL-5 gene locus. Gfi1 Is Required for the Suppression of the IFN␥ Gene Locus Activation in Developing Th2 Cells-To assess the status of histone modifications at the IFN␥ locus in Gfi1 Ϫ/Ϫ Th2 cells more precisely, we designed a series of primers of the IFN␥ locus, including the promoter, regulatory elements, and DNase I-hypersensitive sites (Fig. 3A) and performed a ChIP assay. The actual band patterns of each ChIP assay (Fig. 3B) and the relative band intensities (Me3 H3-K4/Input DNA or Ac H3-K9/14/ Input DNA) of the 14 selected primer pairs (Fig. 3C) are shown. Th1-specific hypermethylation of H3-K4 at the IFN␥ locus was similarly observed both in Gfi1 ϩ/ϩ and in Gfi1 Ϫ/Ϫ Th1 cells (Fig. 3C, upper panel). H3-K9/14 acetylation in the Gfi1 Ϫ/Ϫ Th1 cells was moderately increased (Fig. 3C, lower panel). In Gfi1 ϩ/ϩ Th2 cells, the histone methylation and the acetylation at the IFN␥ gene locus was suppressed, and those levels were significantly lower in comparison with Th1 condition cells (Fig.  3, B and C). However, in Gfi1 Ϫ/Ϫ Th2 cells, high level H3-K4 methylation and H3-K9/14 acetylation were observed at the IFN␥ gene locus (Fig. 3, B and C). A substantial increase in the H3-K4 methylation and H3-K9/14 acetylation at the IFN␥ gene locus in Gfi1 Ϫ/Ϫ Th2 cells was confirmed by a ChIP assay with quantitative PCR (Fig. 3D). These results suggest that Gfi1 plays a functional role in the IFN␥ gene suppression of developing Th2 cells.
Enhanced Ubiquitin/Proteasome-dependent Degradation of GATA3 Protein in Gfi1 Ϫ/Ϫ Developing Th2 Cells-GATA3 has been reported to be a critical transcriptional factor for the suppression of IFN␥ production (38 -40) and the activation of IL-5 transcription (41)(42)(43). Therefore, we checked the expression of GATA3 in the Gfi1 Ϫ/Ϫ Th2 cells. As shown in Fig. 4A, the protein expression level of GATA3 in Gfi1 Ϫ/Ϫ Th2 cells was  OCTOBER 17, 2008 • VOLUME 283 • NUMBER 42 substantially decreased, and its level was similar to that in developing Th1 cells. The expression of c-Maf, NFAT1, JunB, and T-bet proteins in Gfi1 Ϫ/Ϫ Th2 cells was equivalent to that of Gfi1 ϩ/ϩ Th2 cells (Fig. 4A). Interestingly, the quantitative RT-PCR revealed that GATA3 mRNA was not decreased in the Gfi1 Ϫ/Ϫ Th2 cells (Fig. 4B). We previously reported that GATA3 is rapidly degraded through the 26 S proteasome-dependent pathway (24). Consequently, we wanted to determine whether the 26 S proteasome pathway is involved in the reduction of GATA3 protein in Gfi1 Ϫ/Ϫ Th2 cells. The Gfi1 Ϫ/Ϫ Th2 cells were treated with a 26 S proteasome inhibitor, MG132, for 2 h and were subjected to Western blotting. The expression of GATA3 protein in the Gfi1 Ϫ/Ϫ Th2 cells was strikingly increased by the treatment with MG132, whereas GATA3 protein in Gfi1 ϩ/ϩ Th2 cells showed a moderate increase (Fig. 4C). Enhanced protein expression for GATA3 was not elicited in developing Th1 cells by MG132 treatment (Fig. 4C, right panel). Collectively, these results imply that GATA3 protein rapidly degraded in Gfi1 Ϫ/Ϫ Th2 cells via the 26 S proteasome-dependent pathway. Accordingly, we assessed the levels of the ubiquitinated-form of GATA3 protein in Gfi1 Ϫ/Ϫ Th2 cells in the presence of MG132. The ubiquitination of GATA3 protein was dramatically increased in Gfi1 Ϫ/Ϫ Th2 cells, whereas the total protein amount of GATA3 slightly decreased (Fig. 4D).

Critical Role of Gfi1 in GATA3 Protein Stabilization
To confirm the effect of Gfi1 on the stabilization of GATA3 protein, Gfi1 was introduced into Gfi1 Ϫ/Ϫ developing Th2 cells.  Naive CD4 T cells from Gfi1 Ϫ/Ϫ mice were stimulated with immobilized anti-TCR mAb under the Th2 conditions for 2 days, and then the cells were infected with a retrovirus vector containing Gfi1 cDNA. Three days after infection, the protein and mRNA expression levels of GATA3 were assessed. A significant increase in the level of GATA3 protein was detected in the Gfi1-introducing Gfi1 Ϫ/Ϫ Th2 cells in comparison with the control mock-infected Gfi1 Ϫ/Ϫ cells (Fig. 4E, upper panel). The GATA3 mRNA expression was not affected by the introduction of Gfi1 (Fig. 4E, lower panel). The enhancement of IL-5 production and the reduction of IFN␥ production from Gfi1-introducing Gfi1 Ϫ/Ϫ Th2 cells were confirmed by intracellular staining (supplemental Fig. S4A) and quantitative RT-PCR (supplemental Fig. S4B). These results suggest that Gfi1 stabilizes GATA3 protein through the inhibition of the ubiquitin/proteasomemediated degradation of GATA3 protein.

Enforced Expression of Gfi1 Stabilizes GATA3 Protein in 293
T Cells-To elucidate the direct stabilization of GATA3 protein by Gfi1, Myc-tagged human GATA3 (hGATA3) and FLAGtagged Gfi1 or FLAG-tagged control vector was cotransfected into 293T cells, and then the protein expression level of GATA3 was determined. As shown in Fig. 5A, a Gfi1 dose-dependent expression of GATA3 protein (Fig. 5A, upper panel) was detected without affecting the GATA3 mRNA level (Fig. 5A,  lower panel). The stabilization of GATA3 protein by Gfi1 was confirmed using hGATA3-IRES-EGFP vector, which encodes GATA3 and EGFP on the same mRNA. As expected, GATA3 protein substantially increased by the expression of Gfi1 without affecting of the green fluorescent protein level (Fig. 5B). Consequently, we studied the effect of Gfi1 expression on the degradation of GATA3 protein by a plus-chase analysis. 293T cells were transfected with GATA3 plus Gfi1 or control vector, and the amount of GATA3 protein was assessed at the indicated time points. The degradation of 35 S-labeled nascent GATA3 protein in Gfi1-expressed 293T cells was slower in comparison with mock vector-introduced control cells (Fig.  5C). After a 2-h chase, 70% of the labeled GATA3 remained in the Gfi1-expressing cells, whereas the level decreased at 30% in the control cells. We next examined whether the enforced expression of Gfi1 can suppress the ubiquitination of GATA3. 293T cells were introduced with Myc-tagged hGATA3 plus FLAG-tagged Gfi1 or control FLAG vector, and the ubiquitination of GATA3 was assessed by immunoblotting with antiubiquitin mAb following immunoprecipitation with anti-GATA3 mAb. In the Gfi1 introduced 293T cells, the ubiquitination of GATA3 dramatically reduced (Fig. 5D, left  panel), whereas the expression levels of GATA3 protein were increased (Fig. 5D, right panel). These results suggest that Gfi1 controls the protein expression of GATA3 protein through regulation of the ubiquitin/proteasome-dependent GATA3 protein.
The SNAG Domain of Gfi1 Is Required for the Stabilization of GATA3 Protein-In addition to the C2H2 type zinc finger domain, Gfi1 contains a unique region called the Snail/Gfi1 (SNAG) domain. The importance of the SNAG domain in the Gfi1-mediated transcriptional repression has been reported (26). Gfi1 can form a repressor complex, which contains the CoREST corepressor, the histone demethylase LSD1, and histone deacetylases via the SNAG domain (44). The proline residue at position 2 in the SNAG domain is crucial for the Gfi1mediated formation of the repressor complex. To assess the role of the SNAG domain for the Gfi1-mediated stabilization of GATA3 protein, a SNAG domain deletion mutant (dSNAG) and a proline substitution mutant (P2A) were generated (Fig. 6A). FLAGtagged human GATA3 (hGATA3) and Myc-tagged WT or mutant Gfi1 was cotransfected into 293T cells, and then the protein expression level of GATA3 was determined. As shown in Fig. 6B, the dSNAG mutant failed to stabilize GATA3 protein (upper panel, fifth lane), whereas the P2A mutant succeeded to stabilize (upper panel, fourth lane) without affecting the GATA3 mRNA level (lower panel). WT and mutants Gfi1 proteins expressed at a similar level. Next, we studied the effect of Gfi1 mutants on GATA3 ubiquitination. As expected, the inhibitory function for GATA3 ubiquitination was dramatically reduced in the dSNAG mutant (lane 5), whereas the P2A mutant suppressed GATA3 ubiquitination equivalently (lane 4) in comparison with WT Gfi1 (lane 3) (Fig. 6C,  upper panel). These results suggest that Gfi1 controls GATA3 protein stability through its SNAG domain. However, the repressor activity of Gfi1 is not essential for GATA3 stabilization.
Introduction of GATA3 Compensates Gfi1 Ϫ/Ϫ Th2 Cell Phenotypes-Finally, we examined whether the introduction of GATA3 can compensate the phenotypes of Gfi1 Ϫ/Ϫ Th2 cells. The GATA3 was introduced into Gfi1-deficient developing Th2 cells, and the IFN␥/IL-5 production profile was determined by intracellular staining. The number of IL-5-producing cells (the sum of the upper left and upper right areas) in Gfi1 Ϫ/Ϫ Th2 cells was substantially recovered by the introduction of GATA3 (2.9% versus 10.7%) (Fig. 7A). Moreover, the generation of IFN␥-producing cells (the sum of the lower right and upper right areas) was inhibited (25.4% versus 8.9%) (Fig. 7A). The compensatory effects of GATA3 on the expression of IL-5 and IFN␥ in Gfi1 Ϫ/Ϫ Th2 cells were confirmed by quantitative RT-PCR (Fig. 7B). Finally, we assessed the histone modification at the IL-5 and IFN␥ gene locus in GATA3-introduced Gfi1 Ϫ/Ϫ Th2 cells. The trimethylation of H3-K4 at the IL-5 promoter substantially increased in Gfi1 Ϫ/Ϫ Th2 cells by the introduction of GATA3 (Fig. 7C, left panel), whereas the level at the IFN␥ promoter was suppressed (Fig. 7C, middle panel). The methylation at the IL-4 promoter was not affected by the introduction of GATA3 (Fig. 7C, right panel).

DISCUSSION
In this report, we demonstrate that Gfi1 plays an important role in the regulation of IL-5 and IFN␥ production in Th2 cells. Gfi1 appears to control these Th2 functions through the regu- lation of GATA3 protein stability via the repression of ubiquitin/proteasome-dependent GATA3 degradation. Therefore, the cooperation of Gfi1 with GATA3 is required for the proper Th2 cell differentiation.
We previously reported that the activation of the Ras-ERK MAPK cascade inhibits the ubiquitin/proteasome-dependent degradation of GATA3 (24). In the present study, we found that Gfi1 is rapidly induced after TCR-mediated activation in an ERK MAPK/calcineurin-dependent manner (Fig. 1A) and suppressed the degradation of GATA3 protein (Figs. 4 -6). These results raise the possibility that ERK MAPK regulates GATA3 stability through the induction of Gfi1 expression. Gfi1 has been reported to be associated with PIAS3, a small ubiquitinlike modifier (SUMO) ligase, and inhibits the binding with STAT3 (45). STAT3 could escape from PIAS-mediated sumoylation and increase transcriptional activity. We previously reported that Mdm2 binds to GATA3 and acts as an ubiquitin ligase for GATA3 (24). As in the case of PIAS3, it is possible that Gfi1 binds to Mdm2 and sequesters its E3 ligase activity for GATA3. In fact, the association of Gfi1 with Mdm2 was detected in the transfected 293T cells and Th2 cells. 3 Furthermore, the ubiquitination of Gfi1 protein has also been reported (46,47). Although further studies are required to elucidate the precise mechanism of Gfi1-mediated inhibition in the GATA3 ubiquitination, it is likely that Gfi1 regulates GATA3 protein stability through the repression of ubiquitination.
Gfi1 acts as a transcriptional repressor through a unique N-terminal SNAG domain (26). The proline residue at position 2 was crucial for repressor activity of Gfi1, because the substitution of proline to alanine (Gfi1 P2A mutant) abolished the transcriptional repressor ability. Although the SNAG domain of Gfi1 was required for the stabilization of GATA3 protein, Gfi1 P2A mutant could suppress GATA3 ubiquitination and stabilize GATA3 protein (Fig. 6, B and C). These results demonstrated that the repressor activity of Gfi1 was not essential for GATA3 stabilization and that the SNAG domain might has a distinct function from transcriptional repression.
In this report, we mostly analyzed the Gfi1 effects on the GATA3 stabilization using 293 T cells. Because Gfi1-mediated stabilization of GATA3 was observed in 293T cells as well as Th2 cells , it is likely that Gfi1 stabilized GATA3 protein in Th2 cells through the similar machinery that would operate in 293T cells. However, GATA3 and Gfi1 were not expressed endogenously in 293T cells. The identification of a Gfi1 target ubiquitin E3 ligase may help us to better understand the molecular mechanism controlling GATA3 degradation.
It was previously reported that Gfi1 is required for IL-2-dependent expansion of GATA3 high Th2 cells (48). In our experiment, the mRNA expression level of GATA3 was not changed in expanded Gfi1 Ϫ/Ϫ developing Th2 cells (Fig. 4B). Retrovirusmediated introduction of Gfi1 into Gfi1 Ϫ/Ϫ developing Th2 cells was sufficient to induce GATA3 protein expression (Fig.  4E). Furthermore, the IL-2-dependent expression of GATA3 protein was impaired in Gfi1 Ϫ/Ϫ effector Th2 cells. 3 Therefore, the reduced GATA3 high cell number in Gfi1 Ϫ/Ϫ Th2 cells might be also due to the enhanced degradation of GATA3 protein.
Enhanced production of IFN␥ was observed in Gfi1 Ϫ/Ϫ cells cultured under the Th2 condition and under the Th1 condition with low concentration of IL-12 (supplemental Fig. S3). Furthermore, the levels of histone H3-K9/14 acetylation and of H3-K4 methylation at the IFN␥ gene locus was dramatically increased in the Gfi1 Ϫ/Ϫ Th2 cells (Fig. 3). For the repression of the IFN␥ gene activation including the suppression of H3-K4 methylation and H3-K9/14 acetylation, the expression of GATA3 is required (34, 38 -40). In addition, the expression of the T-box type transcription factor, T-bet promotes IFN␥ production in Th1 cells (50). Although the level of T-bet was increased slightly in the Gfi1 Ϫ/Ϫ Th2 cells, the level was quite low in comparison with that in the Th1 cells (Fig. 4A). Therefore, the de-repression status of the IFN␥ gene locus in Gfi1 Ϫ/Ϫ Th2 cells appeared to be due to the decreased level of GATA3 protein.
Gfi1 contains six C2H2 zinc fingers and is able to bind to specific DNA target sequence (51). The consensus recognition sequence, determined by selection for Gfi1 binding from a random DNA sequence library, is TAAA(T/G)CAC(A/ T)GCA. We identified a potential Gfi1-binding site within the promoter region of the IL-13 gene, whereas there are no potential binding sites for Gfi1 in other regulatory regions of the Th2 cytokine gene locus (IL-4 promoter, IL-5 promoter, CGRE, CNS1, and V A enhancer) and the IFN␥ gene locus (IFN␥ promoter, CNS1, and CNS2). We also performed a ChIP assay and confirmed the Gfi1 binding within the Th2 cytokine gene locus and the IFN␥ gene locus. As expected, the binding of Gfi1 was only detected at the IL-13 promoter. 3 Therefore, it is unlikely that Gfi1 directly regulates IL-5 and IFN␥ production in Th2 cells.
Although the expression of Gfi1 mRNA was equivalently induced in CD4 T cells cultured under both Th1 and Th2 conditions, its expression was preferentially maintained in developing Th2 cells. 3 STAT6-deficient CD4 T cells cultured under Th2 conditions also failed to maintain Gfi1 mRNA expression (supplemental Fig. S1B). These results are consistent with previous data, in which Gfi1 expression is Th2 cell-specific and STAT6-dependent (30). Th2 cell-specific maintenance of Gfi1 expression may contribute to the Th2 cell-specific stabilization of GATA3 protein. In addition, our preliminary results indicate that the introduction of GATA3 can induce Gfi1 mRNA expression in developing Th1 cells. 3 These data raise the possibility that GATA3 and Gfi1 make a positive feedback loop for stable expression of both genes and control Th2 cell differentiation. In summary, the results of this study indicate that Gfi1 is a downstream target of the ERK MAPK cascade, and it plays an important role in the regulation of the GATA3 protein expression and Th2 cell functions.  Fig. 4E and restimulated by anti-TCR mAb for 2 h. The expression of each cytokine mRNA was assessed by quantitative RT-PCR. C, the levels of trimethylation of H3-K4 were determined by a ChIP assay with a quantitative PCR as described in Fig. 1E. Three independent experiments were performed with similar results.