Requirement of Two NFATc4 Transactivation Domains for CBP Potentiation*

Recruitment of the coactivator CREB-binding protein (CBP) to transcription factors is important for gene expression. Various regions of CBP such as the KIX and CH3 domains have been shown to interact with numerous transcription factors. The NFAT group of transcription factors is involved in multiple biological processes. NFATc4/NFAT3 has been proposed to play an important role in heart hypertrophy, adipocyte differentiation, and learning and memory. We demonstrate here that two transactivation domains, located at the NH 2 and COOH termini of NFATc4, are critical for interacting with CBP. Each transactivation domain interacts with a distinct region of the CBP protein (the KIX and CH3 domains). Binding of CBP potentiates NFATc4-mediated transcription activity. Both transactivation domains of NFATc4 are required for CBP function. Removal of either NFATc4 transactivation domain abolishes CBP potentiation. Conversely, mutation of the KIX or CH3 domain prevents CBP-mediated potentiation of NFATc4 transcription activation. These data demonstrate that formation of a functional NFATc4 (cid:1) CBP transcription complex requires interactions at two distinct sites. The transcription factor NFAT group was first characterized as an important regulator of interleukin-2 (IL-2) 1 gene expression (1). Other cytokines such as IL-4, IL-5, and tumor necrosis factor- (cid:1) (TNF

The transcription factor NFAT group was first characterized as an important regulator of interleukin-2 (IL-2) 1 gene expression (1). Other cytokines such as IL-4, IL-5, and tumor necrosis factor-␣ (TNF␣) have also been shown to be regulated by NFAT (2)(3)(4). Four members of the NFAT group (NFATc1/NFATc/ NFAT2, NFATc2/NFATp/NFAT1, NFATc3/NFAT4/NFATx, and NFATc4/NFAT3) have been identified (reviewed in Refs. 5 and 6). The wide tissue distribution of these NFAT isoforms suggests that NFAT may participate in multiple physiological processes. Within the NFAT group of proteins, NFATc4/NFAT3 is the member that is primarily expressed in nonimmune tissues (7). Recently, NFATc4 activities have been implicated in heart hypertrophy, adipocyte differentiation, and learning and memory (8 -10). Thus, elucidation of mechanisms of NFATc4 activation is critical for understanding these biological processes.
Activated nuclear NFAT interacts with other transcription factors to induce gene expression (6). Indeed, structural studies have demonstrated interactions between NFAT and the Fos⅐Jun (AP-1) complex (24,25). Formation of a ternary NFAT⅐Fos⅐Jun protein complex on DNA suggests that this complex may function as a composite transcription enhancer. The interaction of NFAT with GATA, myocyte enhancer factor-2, and peroxisome proliferator-activated receptor-␥ suggests that NFAT may often function at composite DNA elements (9, 26 -29). Further studies on NFAT transactivation will thus provide better understanding on NFAT-mediated gene expression.
The coactivator CBP was first identified as a binding partner for the transcription factor CREB and the adenoviral E1A protein (30 -32). CBP was subsequently found to interact with numerous transcription factors (reviewed in Refs. [33][34][35][36]. It is thought that the intrinsic histone acetyltransferase activity of CBP acetylates histone proteins and specific transcription factors to promote the remodeling of compact chromatin structures and hence modulate transcription activity. On the other hand, the adenoviral protein E1A inhibits transcription, in part, by sequestering CBP (37). Thus, stable interactions with transcription factors are important for CBP function.
The purpose of this study was to examine transactivation mediated by NFATc4. Since NFATc4 has been implicated in multiple biological processes, understanding the molecular mechanism of NFATc4-mediated transcription activation is an important goal. We report that binding of CBP potentiates NFATc4-mediated transcription activity. Two transactivation domains on the NFATc4 protein are identified. Each transactivation domain interacts with CBP on a distinct region. Both transactivation domains of NFATc4 are required for CBP potentiation of transcription activity. Removal of one of the trans-activation domains of NFATc4 abolishes CBP potentiation. Conversely, mutation of the KIX or CH3 domain prevents CBPmediated potentiation of NFATc4 transcription activity. These results indicate that NFATc4 functionally interacts with CBP at two distinct sites.

EXPERIMENTAL PROCEDURES
Reagents-The NFAT-luciferase reporter plasmid (7) and the expression vectors for calcineurin (38), NFATc4 (7), dominant-negative NFAT (39), and CBP (31,32,40,41) have been described. NFATc4 COOHterminal truncation mutants were constructed using restriction enzyme digestion to remove the indicated fragments. The internal deletion mutant of NFATc4 (NFATc4-(⌬130 -237)) was constructed by removing the Eco47III-BglI fragment. Gal4-NFATc4 plasmids were constructed by polymerase chain reaction. The TNF␣-luciferase reporter plasmid was constructed by removal of the BamHI-HindIII fragment from the TNF␣-ProCAT plasmid (42) and insertion into the pGL3-Basic luciferase reporter plasmid (Promega). Mutational removal of the NFATbinding element (Ϫ93 to Ϫ82) on the TNF␣ promoter reporter plasmid was performed by polymerase chain reaction. All mutants were sequenced with an Applied Biosystems machine. Bacterially expressed CBP was purified by glutathione affinity chromatography as described previously (14). Monoclonal antibody M2 was obtained from Sigma. Rabbit polyclonal antibody against the Gal4 DNA-binding domain was obtained from Upstate Biotechnologies, Inc.
Luciferase Assays-An NFAT expression vector (0.1 g) was cotransfected with an NFAT-luciferase reporter plasmid (0.3 g) and the control plasmid pRSV-␤-galactosidase (0.2 g) into BHK fibroblasts and human embryonic kidney 293T cells. The CBP expression vector plasmid (0.4 g) was cotransfected as indicated. The Gal4 and Gal4-NFATc4 expression vectors (0.1 g) were cotransfected with a luciferase reporter plasmid (0.3 g) containing five Gal4-binding sites located upstream of the E1B gene promoter. Luciferase and ␤-galactosidase activities were measured 48 h after transfection. Cells were stimulated with ionomycin (2 M) plus PMA (100 nM) as indicated. The data are presented as relative luciferase activity, calculated as the ratio of luciferase activity to ␤-galactosidase activity (means Ϯ S.E., n ϭ 4).

CBP Interacts with NFATc4 and Potentiates Transcription
Activity-CBP acts as a transcription coactivator in vivo. Multiple regions of CBP interact with transcription factors (Fig.  1A). The receptor-interacting domain located at the NH 2 -terminal end of CBP binds members of the nuclear receptor group of proteins (e.g. retinoic acid receptor) (44). The KIX domain (the region for interacting with protein kinase A-phosphorylated CREB) of CBP binds a wide variety of transcription factors, including CREB, c-Jun, and activating transcription factor-4 (31)(32)(33)45). The CH3 (cysteine-and histidine-rich 3) domain of CBP also interacts with multiple factors (e.g. E1A, c-Fos, and GATA) (30, 34, 46 -50). To test whether the transcription factor NFATc4 interacts with CBP, we performed in vitro binding assays using recombinant CBP proteins (Fig. 1B). Purified GST-CBP proteins were immobilized on glutathione-Sepharose. Control experiments demonstrated that NFATc4 did not bind to immobilized GST. The COOH-terminal region of CBP (CBP-(1892-2441)) also did not bind NFATc4. However, binding of NFATc4 to the KIX domain (CBP-(451-682)) and the CH3 domain (CBP-(1680 -1891)) of CBP was detected. Together, these data indicate that NFATc4 interacts with two regions on CBP.
Recruitment of CBP has been demonstrated to potentiate transcription activity. We therefore examined whether CBP regulates NFATc4 transcription (Fig. 1C). NFATc4 was cotransfected with an NFAT-luciferase reporter plasmid into BHK cells. Expression of NFATc4 increased luciferase reporter gene activity (data not shown). Coexpression of CBP further increased NFATc4 transcription activity. These data indicate that CBP interacts with NFATc4 and potentiates transcription activity.
Identification of the CBP-binding Sites on NFATc4 -To delineate the CBP-binding sites on NFATc4, we examined the interaction between various NFATc4 deletion mutants ( Fig detected by immunoblot analysis (Fig. 2B). Progressive COOHterminal truncations of NFATc4 reduced binding to CBP-(451-682) (Fig. 2C). Similarly, reduced binding to CBP-(1680 -1891) was also detected in these NFATc4 deletion mutants (Fig. 2D). Deletion to residue 160 of NFATc4 completely abolished CBP binding. These data suggest that CBP interacts with multiple regions of NFATc4.
Two Transactivation Domains Are Located on NFATc4 -CBP frequently interacts with the activation domains of transcription factors. Both the NH 2 -and COOH-terminal regions of NFATc1 and NFATc2 have been shown to mediate transcription activity (17,51). We tested whether the NH 2 -and COOHterminal regions of NFATc4 supported gene transcription. The NH 2 -terminal region (residues 1-217, Gal4-NFATc4-(N)) and the COOH-terminal region (residues 771-902, Gal4-NFATc4-(C)) of NFATc4 were fused in frame with the DNA-binding domain of the Gal4 transcription factor. A luciferase reporter plasmid containing five copies of the Gal4-binding site was cotransfected with Gal4-NFATc4 expression vectors into BHK cells. Control experiments indicated that expression of the Gal4 DNA-binding domain alone supported minimal transcription activity (Fig. 3). However, expression of Gal4-NFATc4-(N) and Gal4-NFATc4-(C) markedly increased transcription activity (Fig. 3). These data indicate that there are two transactivation domains located on the NFATc4 protein.
Both NFATc4 Transactivation Domains Are Required to Mediate CBP Potentiation-Binding of NFATc4 to two different regions of CBP suggested that both transactivation domains might be required for NFATc4 function in the nucleus. To test NFATc4 nuclear function, we first examined DNA-binding activity of the NFATc4 proteins. Nuclear extracts were prepared from COS cells transfected with NFATc4 expression vectors. Oligonucleotides encoding the NFAT-binding site from the IL-2 promoter were used as a probe in DNA gel mobility shift assays (Fig. 6A). Wild-type NFATc4 formed a complex with the IL-2 NFAT DNA oligonucleotides (Fig. 6A). Similar NFATc4⅐DNA complexes were detected in NFATc4 deletion mutants (Fig.  6A). These data indicate that removal of the transactivation domain does not affect DNA binding of NFATc4.
NFATc4 is implicated in the transcriptional regulation of the expression of several cytokine genes, including IL-2, IL-4, and TNF␣. Previous studies indicated that there are two NFATbinding elements present in the TNF␣ promoter (4). Binding of NFAT to both elements is required for TNF␣ induction. We examined TNF␣ promoter activity mediated by NFATc4 proteins in the presence and absence of coexpressed CBP. Coexpression of CBP and NFATc4 increased TNF␣ promoter activity (Fig. 6C). However, removal of either the NH 2 -terminal (NFATc4-(⌬130 -237)) or COOH-terminal (NFATc4-(1-771)) transactivation domain of NFATc4 abolished CBP potentiation (Fig. 6C). These data further support that both transactivation domains of NFATc4 are required for CBP function.
Since there are two NFAT-binding sites present on the TNF␣ promoter, we tested whether coexpression of the two CBP binding-defective NFATc4 mutants (NFATc4-(⌬130 -237) and NFATc4-(1-771)) would reconstitute CBP potentiation. Various combinations of NFATc4-(⌬130 -237) and NFATc4-(1-771) expression vectors were cotransfected with the CBP expression plasmid into BHK cells. However, potentiation of NFATc4 activity by CBP was not detected under any tested condition (data not shown). Together, these data suggest that NFATc4/ CBP interaction mediated by two distinct binding sites is required for transcription activation.
The KIX and CH3 Domains of CBP Are Both Required to Mediate Potentiation of NFATc4 Transcription Activity-Previous studies demonstrated that specific point mutations (K606E and E655K) in the CBP KIX domain reduce transcription activity mediated by CREB (41). Similarly, removal of the CH3 domain of CBP (CBP⌬CH3) decreases transcription activity mediated by the STAT1 transcription factor (40). However, the CBP⌬CH3 mutant mediates transcription activity of the retinoic acid receptor as effectively as wild-type CBP. These results indicate a differential use of CBP domains for transactivation by various nuclear factors. Since NFATc4 interacts with CBP at two different sites, we examined NFATc4 transcription activity in the presence of CBP mutants (CBP-K606E and CBP⌬CH3) (Fig. 7A). Coexpression of wild-type CBP and NFATc4 potentiated TNF␣ promoter activity (Fig. 7B). However, a mutation in the KIX domain (CBP-K606E) or deletion of the CH3 domain (CBP⌬CH3) of CBP abolished transcription potentiation (Fig. 7B). These data indicate that both the KIX and CH3 domains are required for NFATc4 transactivation.
Replacement of Leu 607 with Phe in the KIX domain of CBP (CBP-L607F) increases CREB-dependent transcription by promoting interaction between CREB and CBP (41). Increased KIX domain binding may thus relieve the requirement of two NFATc4-binding sites for CBP potentiation. We tested whether the CBP-L607F mutant would rescue transcription potentiation mediated by the NFATc4-(⌬130 -237) mutant, which binds to the KIX domain as effectively as wild-type NFATc4, but whose interaction with the CH3 domain of CBP is abolished. Expression of CBP-L607F (although to a lesser extent) potentiated TNF␣ promoter activity mediated by wild-type NFATc4 (Fig. 7C). Removal of the CH3 domain-binding region of NFATc4 (NFATc4-(⌬130 -237)) abolished CBP potentiation. Expression of CBP-L607F did not further increase TNF␣ promoter activity mediated by the NFATc4-(⌬130 -237) mutant (Fig. 7C). These data indicate that two interacting sites are required for NFATc4/CBP transcription potentiation.
Inhibition of NFAT Reduces CBP-mediated Potentiation of TNF␣ Promoter Activity-Previous studies indicated that treatment with the immunosuppressive drug cyclosporin A (CsA) reduces TNF␣ gene transcription (52)(53)(54)(55)(56). Molecular analysis indicates that CsA reduces cytokine gene expression by inhibiting calcineurin phosphatase and thereby blocks NFAT activation (15,16). Recently, TNF␣ gene transcription has been shown to be dependent on the dosage of CBP (57). We tested whether inhibition of NFAT reduced CBP potentiation on the TNF␣ promoter. Treatment with CsA reduced TNF␣ promoter activity (Fig. 8A). Coexpression of CBP potentiated TNF␣ promoter activity. Administration of CsA also reduced CBP-mediated potentiation of TNF␣ promoter activity (Fig.  8A). A similar inhibition of CBP potentiation of the TNF␣ promoter was observed with coexpression of the dominantnegative NFAT inhibitor (39) (Fig. 8A). These data indicate that NFAT plays an important role in CBP-mediated TNF␣ gene transcription.
Multiple transcription factor-binding sites have been characterized in the TNF␣ promoter. These transcription factors are likely involved in the assembly of the transcription activation complex. To examine the role of NFAT in the TNF␣ promoter, we constructed a minimum promoter reporter plasmid (Ϫ206 to Ϫ1) containing an NFAT-binding site. This TNF␣

NFATc4 Interacts with CBP on Two Distinct Regions
promoter reporter plasmid remained responsive to ionomycin plus PMA stimulation and CBP potentiation (Fig. 8B). However, mutational removal of the NFAT-binding site in this TNF␣ promoter abolished CBP potentiation (Fig. 8B). These data further indicate an important role of NFAT in CBP potentiation of the regulation of TNF␣ gene transcription. FIG. 6. Both NFATc4 transactivation domains are required to mediate CBP potentiation. A, NFATc4 proteins were expressed in COS cells, and nuclear extracts were prepared for gel mobility shift assays. Oligonucleotides encoding the NFAT-binding site from the IL-2 gene promoter were labeled with [␥-32 P]ATP and used as probes. NFATc4⅐DNA complexes were visualized by autoradiography. B, fulllength NFATc4 and NFATc4 mutants were transfected with an NFATluciferase reporter plasmid without (Control) and with a CBP expression vector. Transfection efficiency was monitored by measurement of ␤-galactosidase activity. The cells were stimulated with ionomycin (2 M) plus PMA (100 nM) for 16 h before harvest. C, full-length NFATc4 and NFATc4 mutants were transfected with a TNF␣-luciferase reporter gene plasmid without (Control) and with CBP. The cells were stimulated with ionomycin (2 M) plus PMA (100 nM) for 16 h before harvest. Transfection efficiency was monitored by measurement of ␤-galactosidase activity.

DISCUSSION
NFAT/CBP Interactions-In this study, we demonstrate that NFATc4 interacts with CBP on two distinct regions. The presence of two interacting sites may allow an efficient recruitment of CBP to the NFAT transcription complex. Formation of a multicomponent protein complex is determined, in part, by the on-rate of the individual component and the off-rate of the complex. The presence of two interacting sites will increase the opportunities for two proteins to associate and hence promote NFATc4-mediated gene transcription. In addition, NFAT is a cytosolic protein in resting cells (5,6). Upon stimulation, NFAT translocates from the cytosol to the nucleus to mediate gene expression. Since the CBP protein is proposed to be limited in quantity inside the nucleus (37), the presence of two binding sites may allow better competition for CBP once NFAT is resident in the nucleus. Furthermore, the presence of two interacting sites may allow a longer duration for transcription mediated by the NFATc4⅐CBP complex by retaining CBP in the vicinity of NFAT downstream target promoters. Together, the presence of two CBP-interacting sites may promote immediate prompt downstream actions, in response to extracellular stimuli, to recruit CBP and to increase expression of NFAT-mediated genes.
Inactive NFAT is proposed to adopt a compact structure in the cytosol (6). Masking of nuclear localization sequences is achieved by both intra-and intermolecular interactions (14,20,22,58). Phosphorylation has been demonstrated to play an important role in these interactions. Upon stimulation, dephosphorylation mediated by the calcineurin phosphatase unmasks nuclear localization sequences and facilitates translocation of NFAT from the cytosol to the nucleus. Mechanisms for maintaining NFAT proteins in a phosphorylated compact structure in the cytoplasm or in a dephosphorylated active state inside the nucleus remain to be determined. Our results indicate that CBP interacts with NFAT at two different sites. These sites are located in the NH 2 -and COOH-terminal regions of NFAT. Formation of an NFATc4⅐CBP complex may promote and stabilize an open structure of nuclear NFAT. The NFATc4⅐CBP complex may also prevent binding of protein kinases and subsequent phosphorylation of NFATc4 in the nucleus. Alternatively, the NFATc4⅐CBP complex may interfere with binding of nuclear factors that facilitate export (e.g. export receptor CRM1). Thus, CBP may sequester activated NFAT in the nucleus to mediate gene transcription.
Recent studies indicate that transcription of the TNF␣ gene is dependent on the dosage of CBP (57). Transcriptional induction of TNF␣ in response to T cell receptor stimulation requires both copies of the CBP gene. However, one copy of CBP is sufficient to induce TNF␣ gene expression in response to viral infection. In addition, different transcription enhancer complexes are identified under viral and T cell receptor-mediated stimulation (59). These observations suggest that stimulation by viral infection, but not by T cell receptor stimulation, may induce additional regulation to recruit CBP to potentiate transcription. Our results indicate that CBP potentiation of TNF␣ gene transcription is inhibited by treatment with CsA and coexpression of dominant-negative NFAT. In addition, our results suggest that the presence of two transactivation domains on NFAT may favor competition for CBP. Since both transactivation domains are required for potentiation, the requirement for two CBP-binding sites may be relieved in cells with a low dosage of CBP. Furthermore, specific post-translational modifications of the interacting domains between NFAT and CBP may stabilize NFAT/CBP interactions and hence potentiate TNF␣ gene expression. Thus, in low levels of CBP proteins, viral infection may modulate NFAT⅐CBP complex formation to induce TNF␣ gene expression.
Our results indicate that NFATc4 interacts with CBP at two distinct sites. Since there are two NFAT-binding sites on the TNF␣ gene promoter, it is possible that one CBP molecule may interact with two NFATc4 proteins to mediate transactivation. In this study, we were not able to determine the stoichiometry of NFATc4 to CBP. However, expression of two CBP bindingdefective NFATc4 mutants (NFATc4-(1-771) and NFATc4-(⌬130 -237)) did not recapitulate transcription potentiation by the wild-type NFATc4 protein. Thus, formation of a stable NFATc4⅐CBP complex may require interactions with both transactivation domains. It is likely that one molecule of NFATc4 interacts with one molecule of CBP through two interacting sites. The presence of two NFAT-binding sites on the TNF␣ gene promoter may facilitate recruitment of CBP and promote formation of a secure NFAT⅐CBP complex, which then regulates gene expression.
NFAT Regulation-The NFAT signaling pathway is regulated at multiple levels (5,6). These include subcellular distribution, DNA binding, and transcription activation. Recently, phosphorylation has been shown to regulate subcellular distribution of NFAT (18 -23). Multiple NFAT kinases have been identified to phosphorylate NFAT. Phosphorylation of NFAT proteins opposes calcineurin-mediated nuclear localization. Once NFAT is translocated to the nucleus, nuclear NFAT pre- sumably interacts with DNA and mediates gene transcription. However, the regulatory mechanism for nuclear NFAT remains to be determined. The requirement of two intact binding sites for CBP potentiation suggests an additional mechanism that regulates NFATc4-mediated gene expression. Disturbing binding to one of the transactivation domains of NFATc4 abolishes CBP potentiation, although CBP may remain tethered to the transcription complex by the other CBP-interacting region. This mechanism may provide an inhibitory signal to suppress NFATc4-mediated gene transcription. On the other hand, the presence of two interacting sites may allow effective competition for CBP once NFAT is activated and translocated to the nucleus. The NFATc4/CBP interactions may be further regulated by post-translational modifications such as phosphorylation. Binding to other nuclear factors may also affect NFATc4/ CBP interactions. Alternatively spliced NFAT isoforms that encode the interacting sites or not may also modulate CBP potentiation.
The NFAT family includes four distinct genes, which all contain an NFAT homology domain at their NH 2 termini. Molecular cloning of cDNAs revealed alternatively spliced NFAT mRNAs that encode at least 10 different NFAT polypeptides (6). Several of these alternatively spliced mRNAs have various COOH termini and deletions within the transactivation domain. These COOH-terminal truncated NFAT spliced isoforms have an intact Rel homology domain for DNA binding. However, gene transcription mediated by these COOH-terminal truncated NFAT spliced isoforms may be attenuated because of the lack of transcription potentiation by CBP. Thus, a differential transcription activity would be achieved by various NFAT spliced isoforms to optimally regulate downstream target gene expression.
The NFAT group of proteins is characterized by the presence of an NH 2 -terminal NFAT homology domain for regulation and a Rel homology domain for DNA binding (reviewed in Refs. 5 and 6). Multiple proteins have been found to interact with the NFAT homology domains. For example, the phosphatase calcineurin has been shown to bind to the PXIXIT motif in the NFAT homology domain, which is conserved among all four NFAT members (12)(13)(14). On the other hand, glycogen-synthase kinase-3, casein kinase-1␣, and c-Jun NH 2 -terminal kinase have been shown to interact with and phosphorylate Ser residues in the NFAT homology domain (14,18,19,22). The signaling protein 14-3-3 has also been demonstrated to bind the NFAT homology domain (20). Phosphorylation of Ser 272 and Ser 289 of NFATc4 by protein kinase A promotes NFAT/14-3-3 interactions. We demonstrate here that NFATc4 binds to the transcription coactivator CBP. One of the binding regions (between residues 130 and 217 of NFATc4) is located in the NFAT homology domain. Previous studies indicated that the NH 2terminal regions of NFATc1 and NFATc2, which encompass the NFAT homology domain, also interact with CBP (17,51). Since the NFAT homology domain is highly conserved among the NFAT family members, it is conceivable that similar residues are responsible for CBP interaction. Further studies may allow formulation of a consensus binding motif for the interaction between CBP and NFAT. Such motifs may provide a therapeutic opportunity for the regulation of NFAT activation and subsequent cytokine production.