Ca2+-dependent Gene Expression Mediated by MEF2 Transcription Factors*

Ca2+ induction of a subset of cellular and viral immediate-early activation genes in lymphocytes has been previously mapped to response elements recognized by the MEF2 family of transcription factors. Here, we demonstrate that Ca2+ activation of MEF2 response elements in T lymphocytes is mediated in synergy by two Ca2+/calmodulin-dependent enzymes, the phosphatase calcineurin, and the kinase type IV/Gr (CaMKIV/Gr), which promote transcription by the MEF2 family members MEF2A and MEF2D. Calcineurin up-regulates the activity of both factors by an NFAT-dependent mechanism, while CaMKIV/Gr selectively and independently activates MEF2D. These results identify MEF2 proteins as effectors of a pathway of gene induction in T lymphocytes which integrates diverse Ca2+ activation signals and may be broadly operative in several tissues.

Ca 2ϩ plays a central role in regulating gene transcription in lymphocytes at different stages of their development and differentiation. The rise in free intracellular Ca 2ϩ upon antigen receptor engagement has been implicated in the induction of a diverse group of gene products including transcriptional activators, growth factors, and effectors of activation-induced cell death. A major pathway mediating Ca 2ϩ signaling in lymphocytes involves the Ca 2ϩ /calmodulin-dependent phosphatase calcineurin and members of the calcineurin-regulated NFAT family of transcription factors (1). Calcineurin dephosphorylates NFAT proteins, a requisite step in their accessing the nucleus. NFAT proteins support transcription of a large number of Ca 2ϩ -responsive genes by binding to specific response elements in their promoters (2). A number of potent immunosuppressive agents such as cyclosporin A (CsA) 1 and FK506 act to inhibit calcineurin function and consequently abrogate NFATdependent gene transcription (3).
Ca 2ϩ responsiveness and CsA sensitivity of a number of promoters/enhancers have been mapped to another set of response elements distinct from those of NFAT and which are recognized by the MEF2 family of transcription factors (also known as serum-response factor-related proteins or RSRF).
MEF2 proteins have homologous DNA-binding domains located at their amino termini which are characterized by the presence of a MADS box motif. This motif defines a superfamily of DNA-binding proteins that, in addition to the MEF2 family, includes the serum response factor, the yeast regulatory proteins MCM1 and ARG80, and the plant homeotic gene products Agamous and Deficiens (4).
To date, four MEF2 proteins have been identified: MEF2A, -B, -C, and -D (5)(6)(7)(8)(9)(10)(11). They are encoded by distinct genes, and are expressed in several tissues including muscle, neurons, and lymphocytes. MEF2 proteins form homo-and heterodimers that constitutively bind to response elements bearing the consensus sequence CTA(A/T) 4 TAG (6). MEF2 sites are present in promoters/enhancers of a number of muscle-specific genes, where they promote gene expression during myogenic development (12). However, MEF2 sites are also present in promoters/ enhancers of a number of immediate-early activation genes, including the steroid orphan receptor Nur77 (13)(14)(15)(16), the Epstein-Barr virus lytic cycle switch gene BZLF1 (17), and the proto-oncogene c-jun (18). MEF2 proteins have been implicated in the induction of these genes by extracellular signals (16 -19), a role similar to that played by the MADS family member serum response factor in the induction of c-fos (20).
An important attribute of all three MEF2-regulated immediate-early activation genes is their induction by Ca 2ϩ in an immunosuppressant-sensitive fashion (16,21,22). Studies on the promoter/enhancers of both Nur77 (16) and BZLF1 (17) have mapped their Ca 2ϩ and immunosuppressant sensitivity to MEF2 response elements, but failed to demonstrate direct NFAT binding at these sites. These findings, which contrast with the regulation of other Ca 2ϩ and immunosuppressantsensitive promoters by either classical or B-related NFATbinding elements (2), raised the possibility that MEF2 activation by Ca 2ϩ is NFAT-independent (16).
Based on these observations, this study was undertaken to elucidate the mechanism by which Ca 2ϩ activates MEF2-dependent transcription in T lymphocytes, using Nur77 as a prototypic MEF2-regulated immediate-early activation gene. We demonstrate that Ca 2ϩ activation of MEF2-dependent transcription involves two independent pathways of Ca 2ϩ signaling: calcineurin, which promotes formation of MEF2/ NFATc2 transcriptional complexes, and CaMKIV/Gr, which differentially activates MEF2D. ⌬CaMKIVc, and constitutively active CaMKII-␥ B mutant (CaMKIIc). These were prepared and subcloned into pSG5 vector as described elsewhere (23)(24)(25). cDNA encoding a constitutively active mutant of the murine calcineurin ␣1 subunit (abbreviated CNM) and human NFAT family members NFATc1, NFATc2, and NFATc3 (corresponding to NFAT2a, NFAT1a, and NFAT4x according to the proposed nomenclature of Rao et al. (2)) were derived and subcloned into pBJ5 vector as described elsewhere (26,27). Dominant-negative human NFATc2 mutant lacking the amino-terminal transcriptional activation domain was derived and sublconed in pREP-4 expression vector as described (28).
A genomic clone encompassing rat NGFI-B/Nur77 promoter/enhancer extends from Ϫ1800 to ϩ119 bp relative to the transcription initiation site of the rat NGFI-B/Nur77 gene (15), and was subcloned into the BamHI site of the reporter plasmid p0Fluc to generate Ϫ1800Nur77Luc. Nur77 promoter/enhancer sequences extending from Ϫ319 to ϩ46 and from Ϫ252 to ϩ46 were derived from Ϫ1800Nur77Luc by polymerase chain reaction, subcloned into pGL2Luc basic vector (Promega) to generate Ϫ319Nur77Luc and Ϫ252Nur77Luc, respectively. The sequence spanning bp Ϫ252 to Ϫ98 of the Nur77 promoter was removed from Ϫ252Nur77Luc by digestion with the restriction enzymes BalI and SmaI, and the digested vector was reannealed by blunt-end ligation to generate Ϫ98Nur77Luc. A constitutively active luciferase reporter driven by cytomegalovirus immediate early promoter (CMVLuc) has been previously described (24).
Transfections and Reporter Assays-DO11. 10 and Jurkat cells were transfected in aliquots of 1 ϫ 10 7 and 1.5 ϫ 10 7 cells suspended in 0.4 ml of 10% fetal calf serum RPMI medium, respectively. The cells were transfected by electroporation at a setting of 250 V and 960 microfarads using a Bio-Rad Gene Pulsar apparatus. For reporter assays (except for GAL4 constructs), transfected DNA included 5 g of reporter plasmids, 10 g of kinase plasmids, or the corresponding empty vector pSG5, and 5 g of CNM and NFAT plasmids or the corresponding empty vector pBJ5. For GAL4-reporter assays, Jurkat cells were transfected with 250 ng of the respective GAL4 fusion construct, 5 g each of 5xGAL4-E1bLuc and pCMV-␤-gal, 1 g of CNM, 10 g of CaMKIVc or the respective empty vector, as indicated. For induction of Nur77 protein expression by ⌬CaMKIVc and CNM (Fig. 1B), DO11.10 cells were transfected with 30 g of the indicated kinase construct and 5 g of CNM. For studies on MEF2 association with NFAT, cells were transfected with 30 g each of NFATc2 and the respective GAL4 fusion constructs. Tansfected cells were maintained in culture for 18 -24 h, following which they were harvested and processed. For luciferase reporter assay, cells were resuspended in 100 l of 0.25 M Tris chloride, pH 7.8, and then lysed by freeze thawing. The lysates were cleared of debris then assayed for luciferase and ␤-galactosidase activities as described previously (31). Results represent means of 2 to 5 experiments Ϯ S.E.
Immunoprecipitation and Immunoblotting-Immunoprecipitation and immunoblotting were done as described previously (32). Briefly, transfected cells (1.5 ϫ 10 7 /aliquot) were lysed in an Nonidet P-40 detergent buffer and nuclei and insoluble material were removed by centrifugation. Supernatants were precleared for 1 h with 25 l of packed protein G-Sepharose (Roche Molecular Biochemicals). Immunoprecipitates were then derived by incubating the precleared lysates for 2 h with 25 l of protein G-Sepharose and 5 g of the indicated antibody (described below). The beads were washed 3 times in lysis buffer, and bound proteins were resolved by SDS-PAGE using 7.5% acrylamide gels, transferred to nitrocellulose membranes, and subjected to immunoblotting.
For detection of proteins in Nonidet P-40-detergent lysates, samples containing 100 g of protein were resolved by SDS-PAGE, then electroblotted onto nitrocellulose membranes. Immunoblotting was carried . Cell aliquots were then either treated with vehicle alone, with ionomycin (1 M), or with ionomycin ϩ PMA (1 M and 25 ng/ml, respectively). The cultures were maintained for an additional 2 h following which nuclear extracts were prepared and examined for Nur77 expression by immunoblotting. Nur77 is expressed as protein species ranging in size from 70 to 90 kDa (indicated by bracket). The higher molecular weight species result from activation-induced protein phosphorylation (16,54,55). B, CaMKIV/Gr and calcineurin synergistically induce Nur77 protein expression. DO11.10 cells were transiently transfected with either control vectors (Vector), individual constructs encoding constitutively active mutants of CaMKIV/Gr (⌬CaMKIVc), CaMKII (CaMKII(c)), or CNM, or with CNM together with ⌬CaMKIVc or CaMKII(c). The cells were maintained in culture for 18 h following which nuclear extracts were derived and probed for Nur77 expression by immunoblotting. C, expression of recombinant kinase molecules in transfected DO11.10 cells. Expression was detected by employing a murine monoclonal antibody against the FLAG epitope, which was engineered into the kinase expression vectors. out with one of the following antibody preparations: the M2 murine monoclonal anti-FLAG epitope antibody (Eastman Kodak Co.), murine anti-GAL4-binding domain (GAL4-BD) monoclonal antibody, or a polyclonal rabbit anti-NFAT antiserum (RK5C1 and K-18-R; both from Santa Cruz Biotechnology, Inc.). Nur77, which localizes to the nucleus, was detected in nuclear extracts (24) using murine anti-rat Nur77 monoclonal antibody (clone 2E1, kind gift of Dr. Jeff Milbrandt, Washington University, St. Louis, MO). Following incubation with the indicated primary antibody, the filters were washed and then incubated with the respective horseradish peroxidase-conjugated secondary reagent (sheep anti-mouse or donkey anti-rabbit IgG antibodies (Amersham Pharmacia Biotech)). The blots were then washed and subsequently developed using an enhanced chemiluminescence system for peroxidase-based detection (Amersham Pharmacia Biotech).
Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared as described previously (24). For electrophoretic mobility shift assays, 5 g of nuclear extracts were incubated together with 0.1 ng of the indicated 32 P-labeled double-stranded oligonucleotide probe in a 20-l reaction mixture containing 10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM dithioerythritol, 0.5 mM EDTA, 5% glycerol, and 1 g/ml poly(dI-dC). Radiolabeled oligonucleotide probes included the aforementioned wtMEF2(Ϫ312) and wtMEF2(Ϫ278) (spanning bp Ϫ312 to Ϫ291 and Ϫ278 to Ϫ257 of the rat NGFI-B/Nur77 promoter), and the distal NFAT-binding site at position Ϫ295 of the murine IL-2 promoter/enhancer: 5Ј-GATCGCCCAAAGAGGAAAATTTGTTTCATACAG-3Ј (33). Competition assays were carried out by including 100-fold molar excess of unlabeled oligonucleotide in the reaction mixture. For supershifts assays, 1 l of the relevant antiserum was added to the reaction mixture. Polyclonal rabbit anti-MEF2A, -B, or -D antisera were derived as described elsewhere (18). Polyclonal rabbit anti-MEF2C antiserum was a kind gift of Dr. J. Han (Scripps Research Institute, La Jolla). Polyclonal rabbit anti-NFAT antiserum (67.1) was a kind gift of Dr. Anjana Rao (Center for Blood Research, Boston, MA). Antisera against Ets family members (Ets-1/2, Elk-1, ERM, Elf-1, SAP-1, and PEA3) were all obtained from Santa Cruz Biotechnology, Inc. Incubations were carried out for 30 min at 23°C, following which the reaction mixtures were resolved on a pre-run 5% nondenaturing polyacrylamide gel and the protein-DNA complexes were visualized by autoradiography.
DNase I Footprinting-DNA templates were prepared by digesting the Ϫ316/Ϫ252Nur77Luc construct with either HindIII (sense strand labeling) or BglII (antisense strand labeling), filling up with [ 32 P]dNTP and Klenow, cutting with BglII (sense strand) or HindIII (antisense strand), then gel purifying by PAGE followed by overnight elution and ethanol precipitation. Approximately 75,000 cpm of probe (1-2 ng) were incubated with 5 g of the indicated nuclear extracts for 30 min at room temperature in a 50-l volume containing 20 mM Tris, pH 7.5, 50 mM NaCl, 1 mM dithioerythritol, 0.5 mM EDTA, 7.5% glycerol, 2% polyvinyl alcohol, and 1 g/ml poly(dI-dC). Control reactions were run in parallel by omitting the addition of nuclear extracts. DNA digestion was initiated by adding 100 l of DNase I solution in 10 mM MgCl 2 , and was continued for 1 min then stopped by adding 150 l of a stop solution containing 8 M urea, 0.5% SDS, and 5 mM EDTA. The amount of DNase I used (0.2 unit/reaction) was derived in separate titration experiments. Following digestion, samples were subjected to phenol/chloroform extraction, ethanol precipitated, and resuspended in loading buffer. Equivalent counts from each samples were denatured at 90°C for 3 min were transfected with either empty pOLuc reporter plasmid or Ϫ1800Nur77Luc together with one or combinations of the following constructs, as indicated: control vectors (Vector), constitutively active mutants of CaMKIV/Gr (⌬CaMKIVc), CaMKII (CaMKII(c)), or calcineurin (CNM), and/or individual NFAT proteins (NFATc1, -c2, or -c3). The cells were harvested 18 h later and assayed for luciferase activity. C, expression of NFAT proteins in transfected cells. Cytosolic and nuclear extracts were derived from cells transfected with empty pBJ5 vector (Vector), NFATc1, -c2, or -c3 were derived and resolved by SDS-PAGE, transferred to nitrocellulose filter, then immunoblotted with an anti-NFAT antiserum which recognizes all three proteins. NFATc1 and -c2 migrated as 110 -120 kDa proteins, while NFATc3 migrated as 180-kDa protein, as reported (56). Equal protein amounts (50 g) were loaded in each lane. However, the total pool of cytosolic proteins recovered was on average 5-fold more abundant than its nuclear counterpart. then loaded onto a 10% polyacrylamide-urea gel, and run at 1500 V for 2 h. Maxam-Gilbert A/G ladders were run alongside the DNase Itreated samples. Following electrophoresis, the gels were dried and subjected to autoradiography.
Derivation of Recombinant Histidine-tagged MEF2D 87-514 and in Vitro Phosphorylation Assays-A cDNA fragment spanning oligonucleotides 492-1778 of MEF2D (8) was derived by polymerase chain reaction and subcloned into the histidine-tag (His) expression plasmid pET28c (Novagen) to generate His-MEF2D 87-514 expression vector. The vector was used to transform Escherichia coli BL21(DE3), and recombinant His-MEF2D 87-514 was purified from bacterial lysates by nickel chelation chromatography (Novagen) following the manufacturer instructions. To derive CaMKIV/Gr, BJAB human Burkitt lymphoma cells, which are lacking in endogenous CaMKIV/Gr, were transfected with empty vector alone or with constructs encoding either wild type human CaMKIV/Gr or catalytically inactive CaMKIV/Gr mutant (CaMKIV/GrK75E), as described elsewhere (32). Following overnight incubation, transfected cells were resuspended in 1-ml aliquots at 10 7 cells/ml in culture medium and were either left untreated or treated with goat anti-human IgM (Southern Biotechnology) at 50 g/ml for 30 s at 37°C. The cells were then collected by rapid centrifugation and lysed, and lysates were subjected to immunoprecipitation with rabbit anti-human CaMKIV/Gr antiserum and protein G-agarose beads (35). In vitro phosphorylation reaction were carried out by incubating the agarose beads with 5 g of His-MEF2D 87-514 in a reaction mixture containing 25 mM HEPES buffer, pH 7.5, 50 M ATP, 10 mM MgCl 2 , 5 mM CaCl 2 , 600 nM calmodulin, and 10 Ci of [␥-32 P]ATP (Amersham Pharmacia Biotech). In vitro phosphorylation reaction was carried out for 30 min at 30 o C, following which the reaction mixture was resolved by SDS-PAGE on a 10% acrylamide gel then transferred to polyvinylidene difluoride membranes (Millipore). The membranes were subjected to autoradiography as well as immunoblotting with anti-CaMKIV/Gr antiserum. Radiolabeled His-MEF2 87-514 band was excised, and onedimensional phosphoamino acid analysis was carried out as described elsewhere (35).

Ca 2ϩ Induction of Nur77 Expression in T Cells Is Mediated by CaMKIV/Gr and Calcineurin
Nur77 protein expression is induced upon stimulation with Ca 2ϩ -mobilizing ionophores, either alone or in combination with phorbol esters (16). Previously, it has been shown that Nur77 expression is inhibited by the immunosuppressant agent CsA, which targets calcineurin (3), and in the case of the rat pheochromocytoma cell line PC12, by an inhibitor of multifunctional CaMK (36). To examine the contribution of multifunctional CaMKs and calcineurin to the induction by Ca 2ϩ of Nur77 expression in T cells, we tested the capacity of CsA and KN93 (Calbiochem), a potent pharmacological inhibitor of multifunctional CaMK including CaMKII and CaMKIV/Gr (37), to inhibit the induction of Nur77 protein expression in the murine T cell hybridoma cell line DO11.10. Fig. 1A demonstrates that KN93 achieved almost complete inhibition of Nur77 protein expression induced upon treatment of cells with Ca 2ϩ ionophore or with ionophore and PMA. This inhibition was similar in magnitude to that achieved with CsA, indicating that Ca 2ϩ induction of Nur77 expression in DO11.10 cells is equally dependent on the contribution of both CaMKs and calcineurin.
To determine the role of CaMKs and calcineurin in the induction of Nur77 by Ca 2ϩ , we examined Nur77 expression in DO11.10 cells transfected with constitutively active mutants of calcineurin (CNM) and/or CaMKs, including the ␥ B isoform of CaMKII (CaMKIIc) and CaMKIV/Gr (⌬CaMKIVc). These mutants maintain the specificities of the parent enzymes (24,38); and allow assessment of the isolated contribution of the respective enzyme to Nur77 induction. Fig. 1B demonstrates that transient co-transfection of D011.10 cell with CNM and ⌬CaMKIVc resulted in induction of Nur77 expression, while transfection with either construct alone resulted in modest effects. In contrast, the constitutively active CaMKIIc failed on its own to up-regulate Nur77 expression and was modestly effective when combined with CNM. This failure was not due to lack of CaMKIIc expression as evidenced by immunoblotting with an anti-FLAG epitope antibody (Fig. 1C). These results indicated that the capacity to up-regulate Nur77 expression is an attribute specific to CaMKIV/Gr. Ca 2ϩ signals induce Nur77 protein expression by activating transcription from the Nur77 promoter/enhancer (16). The capacity of CaMKIV/Gr and calcineurin to activate transcription from the Nur77 promoter/enhancer was investigated using constitutively active enzyme mutants. Accordingly, the constitutively active mutants ⌬CaMKIVc and CNM were transfected, alone or in combination, into DO11.10 cells together with the reporter construct Ϫ1800Nur77Luc, which contains a luciferase gene driven by a DNA fragment spanning bp Ϫ1800 to ϩ119 of the rat Nur77 gene. This fragment retains the Ca 2ϩ and mitogen inducibility of the native promoter/enhancer while exhibiting low basal activity. Fig. 2A demonstrates that on its own, ⌬CaMKIVc induced a 10-fold increase in luciferase re- Following overnight culture, transfected cells were divided into aliquots which were either left unstimulated, treated with ionomycin at 1 M, or treated with combination of ionomycin and PMA (25 ng/ml). Stimulation was carried out for 5 h, following which the cells were harvested and assayed for luciferase activity. Inset, panel A, expression of CaMKIV/GrK75E in transiently transfected DO11.10 cells. Expression was detected by immunoblotting using an anti-FLAG epitope antibody. porter expression whereas CNM had minimal effects. However, ⌬CaMKIVc and CNM synergized in activating Ϫ1800Nur77Luc expression up to 100-fold above baseline. In contrast, the two enzymes, alone or in combination, had negligible effect on the activity of a reporter construct lacking the Ϫ1.8-kilobase Nur77 promoter sequence. The results obtained in the DO11.10 cell line were reproduced in the human leukemia T cell line Jurkat (Fig. 2B), thus confirming the capacity of CaMKIV/Gr and calcineurin to activate the Nur77 promoter/ enhancer in different T cell lines of different species.
Unlike CaMKIV/Gr, CaMKIIc failed on its own to activate transcription from Ϫ1800Nur77Luc reporter construct. When transfected with CNM it either had no effect (DO.11.10; Fig.  2A) or exhibited a modest, 2-fold up-regulation of CNM activation (Jurkat cells; Fig. 2B). These results are in agreement with the failure of CaMKIIc to effectively induce Nur77 protein expression in DO11.10 cells.
To determine the capacity of NFAT proteins to substitute for calcineurin in activating Nur77 promoter/enhancer, we examined the transactivation by individual NFAT proteins of Ϫ1800Nur77Luc. To that end, DO11.10 or Jurkat cells were transfected with individual NFAT constructs encoding NFATc1, -c2, or -c3, alone or in combination with ⌬CaMKIVc, together with Ϫ1800Nur77Luc then examined for reporter gene activation. Fig. 2 demonstrates that NFATc2, but not NFATc1 or NFATc3, acted in synergy with ⌬CaMKIVc to induce Ϫ1800Nur77Luc to levels comparable to those achieved with CNM in both DO11.10 and Jurkat cells (Fig. 2, A and B, respectively). Further treatment with Ca 2ϩ ionophore did not enhance Nur77 reporter gene activation by ⌬CaMKIVc and individual recombinant NFAT proteins (data not shown), in agreement with previous observations on the capacity of transfected NFAT proteins to bypass the requirement for Ca 2ϩ mobilization when mediating gene induction (27). This probably reflects constitutive localization of transfected NFAT molecules to both nuclear and cytosolic compartments (Fig. 2C), which obviates the need for an additional Ca 2ϩ mobilizing signal normally required to translocate cytosol-bound NFAT proteins to the nucleus. Overall, these results indicate that NFATc2 can fully substitute for calcineurin in activating Nur77 transcription.
To further verify the role of CaMKIV/Gr in Nur77 gene transcription, we examined the capacity of a dominant-negative, kinase-deficient CaMKIV/Gr mutant, CaMKIV/GrK75E, to antagonize the activation by Ca 2ϩ of the reporter construct Ϫ1800Nur77Luc. Accordingly, DO11.10 cells were transiently transfected with Ϫ1800Nur77Luc together with increasing concentrations of CaMKIV/GrK75E. Transfected cells were treated with Ca 2ϩ ionophore either alone or in combination with PMA, then examined for reporter gene transcription. 3A demonstrates that CaMKIV/GrK75E completely inhibited the activation by Ca 2ϩ ionophore of reporter gene expression; it also inhibited by about 85% reporter induction by combined PMA and ionomycin treatment. The extent of reporter inhibition correlated with the level of CaMKIV/GrK75E expression, as detected by immunoblotting with anti-FLAG epitope antibody (Fig. 3A, inset). The specificity of inhibition by CaMKIV/ GrK75E of Ϫ1800Nur77Luc was confirmed by failure of the kinase mutant to inhibit a constitutively active reporter gene driven by CMV immediate-early promoter (Fig. 3B). These results are consistent with a critical role for CaMKIV/Gr in promoting Nur77 promoter/enhancer activation by Ca 2ϩ .
Mapping of CaMKIV/Gr and Calcineurin Activation to MEF2 Response Elements-It has been previously demonstrated that Ca 2ϩ induction of murine Nur77 transcription was dependent on two MEF2 elements in the Nur77 promoter/ enhancer (16). These elements are conserved in the rat Nur77 promoter/enhancer; their core consensus sequences span bp Ϫ306 to Ϫ297 and bp Ϫ272 to Ϫ263, respectively. To determine the contribution of these elements to CaMKIV/Gr and calcineurin activation, we generated two deletion mutants which flank the equivalent rodent elements: one extending from Ϫ319 to ϩ46 and the other from Ϫ252 to ϩ46 of the rat Nur77 gene. These were cloned upstream of a luciferase reporter gene, and were examined for their capacity to mediate reporter gene transcription in response to CaMKIV/Gr and calcineurin. Fig.  4A demonstrates that the Ϫ319Nur77Luc construct was potently activated by ⌬CaMKIVc, and that this activation was further up-regulated by CNM. In contrast, the Ϫ252Nur77luc reporter was resistant to ⌬CaMKIVc activation, and was activated by ⌬CaMKIVc ϩ CNM at levels approximately 20% that of Ϫ319Nur77Luc. The residual Ca 2ϩ responsiveness of the Ϫ252Nur77Luc reporter was largely localized to a region spanning bp Ϫ252 to Ϫ98 which harbors AP-1 responsive elements. This was evidenced by the minimal Ca 2ϩ responsiveness of a reporter construct driven by bp Ϫ98 to ϩ46 of the Nur77 promoter (Fig. 4A). Overall, these results identified the Ϫ319 to Ϫ252 sequence as a major Ca 2ϩ and ⌬CaMKIVc ϩ CNMresponsive site within the Nur77 promoter/enhancer.
To determine the identity of elements mediating Ca 2ϩ responsiveness within the Ϫ319 to Ϫ252 sequence, we performed DNase I footprinting analysis with using an oligonucleotide were transiently co-transfected with the indicated effector constructs and one of the following ␤GLuc reporters driven by individual wild type or mutant MEF2 elements: wtMEF2(Ϫ312)␤GLuc, driven by an oligonucleotide (Ϫ312/Ϫ291) spanning the Nur77 distal MEF2 site; wtMEF2(Ϫ278)␤GLuc, driven by an oligonucleotide (Ϫ278/Ϫ257) spanning the Nur77 proximal MEF2 site; mutMEF2(Ϫ312)␤GLuc and mutMEF2(Ϫ278)␤GLuc carry corresponding oligonucleotides in which the respective MEF2 sites were inactivated. D, a putative NFAT-binding sequence contributes to enhanced Ca 2ϩ responsiveness of wtMEF2(Ϫ278). Jurkat cells were transiently co-transfected with the indicated effector constructs together with ␤GLuc reporters driven either by an individual wtMEF2(Ϫ278) element, by a mutant element (mutETS) in which the 5Ј GGA sequence was changed to GAG, or by a mutant element (mutNFAT) in which the 5Ј terminal GG dinucleotide sequence was changed to TT. The cells were subsequently harvested and examined for reporter activity.
overlapping this sequence (Ϫ316 to Ϫ252) and nuclear extracts from mock stimulated (vehicle) and PMA ϩ ionomycin-stimulated DO11.10 cells (Fig. 4B). Results, summarized in Fig. 4C, revealed a pattern of protection involving consensus residues of the two MEF2 elements within the Ϫ319 to Ϫ252 sequence: in both sense (coding) and antisense strands as well as some flanking residues (Fig. 4, B and C). Protection was observed with extracts of both unstimulated and stimulated cells, although it was frequently accentuated when using activated cell extracts. These results were consistent with the proposition that the two MEF2 elements within the Ϫ316 to Ϫ252 segment contribute to Ca 2ϩ responsiveness.
The requisite role of MEF2 elements in mediating transcriptional activation by CaMKIV/Gr and calcineurin was verified using a luciferase reporter gene driven by a minimal ␤-globin promoter (␤GLuc) and synthetic oligonucleotides spanning both MEF2 sequences or each individually. First, it could be demonstrated that an oligonucleotide spanning bp Ϫ316 to Ϫ252 of rodent Nur77 promoter/enhancer conferred sensitivity to induction by CaMKIV/Gr and calcineurin (Fig. 5A). In contrast, introduction of inactivating mutations at both MEF2 elements abolished reporter induction by the two enzymes (Fig. 5A).
Next, we examined the activation by CaMKIV/Gr and calcineurin of transcription from individual MEF2 sites. Oligonucleotides extending 6 bp 5Ј and 3Ј of the consensus sequences of the distal and proximal Nur77 MEF2 elements (Ϫ312 to Ϫ291 and Ϫ278 to Ϫ257, respectively) were placed upstream of the minimal ␤-globin promoter in ␤GLuc reporters and examined for their capacity to confer sensitivity to CaMKIV/Gr and cal-cineurin. Reporter constructs carrying mutant MEF2 sequences were used as controls. Fig. 5 demonstrates that reporter genes driven by either the proximal or distal Nur77 MEF2 sites (wtMEF2(Ϫ278)␤GLuc and wtMEF2(Ϫ312)␤GLuc, respectively), but not their mutant counterparts, were potently activated by ⌬CaMKIVc and CNM both in DO11.10 cells and Jurkat cells (Fig. 5, B and C).
The oligonucleotide spanning the proximal Nur77 MEF2 site (MEF2(Ϫ278)) was about 2-fold more effective in driving reporter gene transcription as compared with its counterpart spanning the more distal Nur77 MEF2 site (MEF2(Ϫ312)). A major difference between the two oligonucleotides is the presence of a sequence (GGAAAA) at the 5Ј end of the MEF2(Ϫ278) but not MEF2(Ϫ312) oligonucleotide. This sequence includes core residues found in binding sites of both NFAT (GGAAAA) and Ets (GGAA) families of transcription factors (2,39). To determine the role of this sequence in the differential transcriptional activity of the two oligonucleotides, we tested reporter gene activation by mutant MEF2(Ϫ278) oligonucleotides in which the putative Ets and NFAT core sequences are differentially disrupted. One oligonucleotide (mutETS) had the GGAA sequence at the 5Ј end of the wtMEF2(Ϫ278) changed to GAGA, which results in the disruption of canonical Ets-binding sites and abrogation of Ets-dependent activation. Another oligonucleotide, mutNFAT, had the putative NFAT-binding sequence disrupted by changing the GG dinucleotide at the 5Ј end of the wtMEF2(Ϫ278) into TT (2). The results, shown in Fig.  5D, demonstrate that mutETS was activated by ⌬CaMKIVc ϩ CNM in a manner indistinguishable from that of wt-MEF2(Ϫ278). In contrast, activation of mutNFAT by ⌬CaMKIVc ϩ CNM was reduced by about 50% compared with wtMEF2(Ϫ278), and closely matched activation levels of MEF2(Ϫ312) (compare Fig. 5, panels C and D). These observations indicate that the putative NFAT-binding sequence promotes Ca 2ϩ responsiveness of Nur77MEF2(Ϫ278).
To establish the identity of proteins bound to the MEF2 sites of the Nur77 promoter/enhancer, we performed electrophoretic mobility supershift assays. Fig. 6 demonstrates that a specific protein-DNA complex was detected upon incubation of nuclear extracts derived from unstimulated DO11.10 cells with radiolabeled oligonucleotide corresponding to bp Ϫ278 to Ϫ257 of the Nur77 promoter/enhancer (wtMEF2(Ϫ278)), which spans the proximal MEF2 site (Fig. 6, lane 2). The same complex was detected in nuclear extracts of stimulated cells (lane 3), and its specificity was established by its inhibition upon the addition of excess unlabeled wild type but not mutant MEF2(Ϫ278) oligonucleotide (lanes 4 and 5, respectively). The MEF2 complex was not inhibited by excess unlabeled NFAT oligonucleotide (lane 6), a treatment which was effective in inhibiting the binding of specific protein complex to radiolabeled NFAT oligonucleotide probe (Fig. 6B, lanes 1 and 2).
We next examined the binding of MEF2 and NFAT proteins to the wtMEF2(Ϫ278) probe by supershift assays using specific antisera. Fig. 6 demonstrates that the MEF2 complex (arrow) was completely supershifted by specific antisera directed against MEF2A (lane 8) and MEF2D (lane 11) but not MEF2B (lane 9) or MEF2C (lane 10). In contrast, neither complex was affected upon the addition of a control rabbit antiserum (lane 7) or an anti-NFAT antiserum (lane 12). The specificity of the anti-NFAT antiserum was demonstrated by its capacity to completely supershift specific protein complexes bound to a radiolabeled NFAT oligonucleotide probe (Fig. 6B, lane 3). Identical results were obtained when using a radiolabeled oligonucleotide probe corresponding to bp Ϫ312 to Ϫ291 of the Nur77 promoter/enhancer (wtMEF2(Ϫ312)), which spans the Nur77 distal MEF2 site (data not shown). The presence in MEF2-DNA complexes of ETS family proteins was also examined by supershift assays using antisera specific for the following ETS members: Ets-1/2, Elk-1, Elf-1, SAP-1, ERM, and PEA3. None of the antibodies inhibited or supershifted MEF2-DNA complexes. Overall, these results indicated that the Nur77 MEF2 elements bound heterodimers of MEF2A and MEF2D, but failed to demonstrate binding of NFAT proteins.
Ca 2ϩ Signalling Pathways Target MEF2 Proteins-To determine whether MEF2 proteins are the direct target of CaMKIV/Gr and calcineurin, we employed constructs encoding hybrid proteins composed of full-length MEF2A or MEF2D fused at their amino termini with the DNA-binding domain of the yeast transcription factor GAL4. Jurkat cells were transiently co-transfected with either GAL4-MEF2A or GAL4-MEF2D together with a luciferase reporter gene harboring 5 GAL4-binding sites upstream of a minimal promoter (pG5ElbLuc), and a ␤-galactosidase reporter gene driven by the CMV immediate early promoter which served as an internal control. Ca 2ϩ responsiveness of the respective GAL4-MEF2 fusion protein was determined monitoring the activity of the luciferase reporter, normalized for ␤-galactosidase activity, upon co-transfection of ⌬CaMKIVc, CNM, or both. Fig. 7A demonstrates that CNM activated transcription by GAL4-MEF2A by about 5-fold while CaMKIVc had no effect. Cotransfection of CaMKIVc had marginal effects on activation compared with CNM alone. In contrast, co-transfection of individual CaMKIVc and CNM constructs resulted in a 5-and 17-fold increase, respectively, in the transcriptional activity of GAL4-MEF2D. Significantly, co-transfection of CaMKIVc and CNM resulted in synergistic activation of GAL4-MEF2D, reaching in excess of 200-fold above basal activity. These results indicated that MEF2 proteins are directly targeted by Ca 2ϩ signaling pathways. They also indicated that MEF2A and MEF2D are differentially regulated by Ca 2ϩ signaling pathways, with CNM activating both proteins while ⌬CaMKIVc selectively activating MEF2D.
To determine whether MEF2D activation by CaMKIV/Gr and calcineurin involved independent or redundant mechanisms, we examined the susceptibility of the respective activation pathway to inhibition by CsA. Fig. 7B demonstrates that CsA did not inhibit ⌬CaMKIVc-induced activation of MEF2D. However, it completely reversed activation by CNM and upregulation by CNM of ⌬CaMKIVc activation. These results indicated that CaMKIV/Gr activates MEF2D by a mechanism independent of that of calcineurin.
Calcineurin Activation of MEF2 Factors Is NFAT-dependent-Despite the failure to detect NFAT binding to MEF2 elements by electrophoretic mobility shift assay, other evidence indicates that MEF2 activation by calcineurin is NFAT-dependent. First, and as shown in Fig. 8A, a dominant- by calcineurin of GAL4-MEF2D. Dominant-negative NFAT also inhibited by a similar magnitude the capacity of calcineurin to up-regulate GAL4-MEF2D activation by CaMKIV/ Gr, but had minimal effect on CaMKIV/Gr activation alone (Ͻ10%) (Fig. 8A). Dominant-negative NFAT was similarly effective in inhibiting CNM activation of GAL4-MEF2A (data not shown). The efficacy of dominant-negative NFAT in inhibiting GAL4-MEF2 activation by calcineurin and ⌬CaMKIVc closely approximated that observed using a reporter gene driven by NFAT response elements (Fig. 8B). The specificity of the dominant-negative NFAT mutant was demonstrated by its failure to affect the constitutive activity of a luciferase reporter gene driven by CMV immediate early promoter (Fig. 8, legend).
In reciprocal experiments, it could be demonstrated that the transcription factor NFATc2 activates MEF2 factors in a manner similar to that of calcineurin. Fig. 8C demonstrates that NFATc2 but not NFATc1 up-regulated GAL4-MEF2A activity by more than 20-fold. ⌬CaMKIVc had no effect on its own (similar to results shown in Fig. 7A), and it failed to up-regulate GAL4-MEF2A activation by NFATc2. NFATc2 but not NFATc1 also activated GAL4-MEF2D, as did ⌬CaMKIVc. However, and unlike the case of GAL4-MEF2A, NFATc2 was potently synergistic with ⌬CaMKIVc in activating GAL4-MEF2D, similar to what observed with CNM (Fig. 7A).
The capacity of NFATc2 to activate MEF2-dependent transcription was also established in studies using luciferase reporter genes driven by MEF2 response elements. Fig. 8D demonstrates that NFATc2, but not NFATc1 or NFATc3, synergized with ⌬CaMKIVc to activate transcription of a ␤GLuc reporter driven by the proximal Nur77 MEF2 site (wtMEF2(Ϫ278)␤GLuc) to levels equivalent to those achieved by ⌬/CNM CaMKIVc and calcineurin. In contrast, NFATc1 and NFATc3 exhibited only modest effects. Similar results were found for the Nur77 distal MEF2 site (data not shown). This is consistent with the results obtained with the full-length Nur77 promoter/enhancer sequence, the transcription of which was FIG. 8. MEF2 activation by calcineurin proceeds by an NFATc2-dependent mechanism. A, dominant-negative (DN) NFAT mutant inhibits MEF2D activation by calcineurin. Jurkat cells were co-transfected with the reporter plasmid pG5ElbLuc, the ␤-galactosidase expression vector pCMV-␤-gal, GAL4-MEF2D expression vector, the constitutive active mutants CNM, ⌬CaMKIVc, or their respective empty vector, as indicated, together with 10 g of either empty pREP vector (Vector) or a pREP plasmid encoding dominant-negative NFATc2 mutant. Following overnight culture, cellular extracts were derived and assayed for luciferase activity, which were normalized for ␤-galactosidase activity. The specificity of dominant-negative NFAT was established by its failure to affect the activity of a luciferase reporter driven by CMV immediate early promoter (vector: 757 Ϯ 20 light units/g of protein; dominant-negative NFAT 707 Ϯ 91 light units/g of protein). Dominant-negative NFAT similarly did not affect the basal activity of pCMV-␤-gal vector (data not shown). B, inhibition of NFAT-driven reporter gene activation by dominant-negative NFAT. Jurkat cells were transiently transfected with a luciferase reporter plasmid driven by NFAT response elements (NFAT-luciferase) then, following overnight incubation, stimulated for 5 h with PMA and ionomycin and examined for reporter gene activation. C, activation of MEF2-dependent transcription by NFATc2. Jurkat cells were co-transfected with the reporter plasmid pG5ElbLuc, the ␤-galactosidase expression vector pCMV-␤-gal, GAL4-MEF2A, or -MEF2D expression vectors, and NFATc1, NFATc2, ⌬CaMKIVc, or their respective empty vector, as indicated. D, Jurkat cells were co-transfected with a ␤GLuc reporter construct driven by one copy of the Nur77 proximal MEF2 site (wtMEF2(Ϫ278)␤GLuc), together with the indicated effector constructs. Transfected cells were subsequently harvested and assayed for luciferase activity. also supported by NFATc2 (Fig. 2).
Co-precipitation of Recombinant NFATc2 and MEF2 Proteins from Cellular Lysates-The identification of MEF2 proteins as direct targets of CNM, coupled with the demonstration that CNM activates MEF2-dependent transcription in an NFATdependent manner, prompted an investigation of physical association of NFAT and MEF2 proteins. To that end, NFAT and GAL4-MEF2 fusion proteins were examined for their co-immunoprecipitation upon their co-expression in Jurkat cells. Fig.  9A (upper panel) demonstrates that NFATc2 co-precipitated with GAL4-MEF2A and GAL4-MEF2D fusion proteins upon their immunoprecipitation with a monoclonal antibody directed against GAL4-BD. NFATc2 was also co-precipitated albeit with reduced levels with GAL4-MEF2D mutant (GAL4-MEF2DCЈ) lacking the NH 2 -terminal 93 amino acids spanning MADS box domain and the adjacent MEF2 domain. This indicates that the two domains are permissive but not essential for NFAT binding to MEF2D. In contrast, NFATc2 did not coprecipitate with a GAL4-CREB fusion protein despite the successful immunoprecipitation of this protein with an anti-GAL4 DNA-binding domain (GAL4-BD) antibody (Fig. 9A, middle  panel). Given that equal amounts of NFATc2 were expressed in lysates of the respectively transfected cells (Fig. 9A, lower  panel), these results support the specificity of NFATc2-MEF2 co-precipitation results.
In reciprocal experiments, it could be demonstrated that both MEF2A and MEF2D specifically co-precipitate with NFATc2 upon immunoprecipitation of the latter with a monoclonal antibody directed against a FLAG epitope which was engineered at NFATc2 amino terminus (Fig. 9B, upper panel). GAL4-MEF2D exhibited enhanced association with NFATc2 as compared with GAL4-MEF2A, and was also selectively recovered in NFATc2 immunoprecipitates from cells co-transfected with both GAL4-MEF2A and GAL4-MEF2D. Comparison with total GAL4-MEF2 protein content in cellular lysates revealed that up to 5% of total GAL4-MEF2D and lesser amounts of GAL4-MEF2A co-precipitated with NFATc2. This was not related to differential retrieval of NFATc2 as equal amounts of this protein were present in all immunoprecipitates (Fig. 9B,  middle panel). Similarly, the enhanced association of GAL4-MEF2D with NFATc2 was not due to the abundance of the respective GAL4-MEF2 protein in cell lystates as MEF2A was expressed at higher levels in transfected cells as compared with MEF2D ( Fig. 9, B, lower panel, see also A, middle panel). Differential retrieval of GAL4-MEF2D in NFATc2 immunoprecipitates may relate to a detrimental influence of the anti-FLAG epitope antibody on NFATc2/MEF2A interaction. More likely, it reflects preferential co-precipitation of NFATc2 with GAL4-MEF2D as compared with GAL4-MEF2A. In that context, the relatively equal amount of NFATc2 found in GAL4-  4). NFATc2 immunoprecipitates were derived with monoclonal anti-FLAG epitope antibody, then subjected to sequential immunoblotting first with monoclonal anti-GAL4-BD monoclonal antibody (upper panel) followed by anti-FLAG epitope monoclonal antibody (middle panel). Cellular lysates were also examined for GAL4-MEF2 content using GAL4-BD monoclonal antibody (lower panel). Fig. 9A can be explained by the severalfold overexpression in transfected cells of GAL4-MEF2A as compared with GAL4-MEF2D (Fig. 9A, middle  panel).

MEF2A and -D immunoprecipitates in
Co-precipitation of recombinant GAL4-MEF2 and NFATc2 proteins proceeded in the absence of a Ca 2ϩ mobilizing signal. This is most likely due to co-expression in the same (nuclear) compartment of a significant portion of recombinant NFATc2 (Fig. 2C) together with the bulk of GAL4-MEF2 proteins (data not shown). A significant portion of recombinant NFATc1 also co-localizes with GAL4-MEF2D in the nucleus. However, and in contrast to NFATc2, NFATc1 could not be detected in immunoprecipitates of GAL4-MEF2 proteins (data not shown). This is in agreement with the results presented in Fig. 8 showing that NFATc2 but not NFATc1 supports MEF2-dependent transcription.
Formation of MEF2-NFATc2-DNA Ternary Complexes-Given the capacity of recombinant MEF2 and NFAT proteins expressed in cells to co-precipitate, the failure to detect MEF2-NFAT-DNA complexes by gel shift assays may stem from the dissociation of such complexes upon gel electrophoresis. To detect such complexes, we examined the capacity of a biotinylated, double stranded wtMEF2(Ϫ278) oligonucleotide to coprecipitate endogenous MEF2 proteins and recombinant NFATc2 upon expression of the latter in DO11.10 nuclei. To that end, biotinylated wtMEF2(Ϫ278) oligonucleotide was incubated with nuclear extracts of NFATc2-transfected DO11.10 cells. Bound complexes were retrieved on streptavidin-agarose beads, washed, and resolved on SDS-PAGE, and examined for the presence of recombinant NFATc2 and endogenous MEF2D proteins by immunoblotting with the respective specific antibody. Fig. 10 (panels A and B, lane 2) demonstrates that both recombinant NFATc2 and endogenous MEF2D co-precipitated with biotinylated wtMEF2(Ϫ278) oligonucleotide. The impact of disrupting core MEF2-and NFAT-binding residues of wt-MEF2(Ϫ278) on the precipitation of the respective protein was also examined. Precipitation with biotinylated mut-MEF2(Ϫ278), in which core MEF2-binding residues are mu-tated, resulted in the retrieval of very low levels of endogenous MEF2D (Fig. 10B, lane 3). Importantly, NFAT co-precipitation with the same oligonucleotide was also greatly decreased, consistent with a requirement for functional MEF2-binding sites for optimal NFAT binding (Fig. 10A, lane 3). Precipitation with biotinylated mutNFAT(Ϫ278), in which core NFAT-binding residues are mutated, results in the retrieval of very low levels of NFATc2 (Fig. 10A, lane 4). In contrast, the binding of endogenous MEF2D to mutNFAT(Ϫ278) was not impaired relative to wtMEF2(Ϫ278) (Fig. 10B, lane 4). Finally, precipitation with an oligonucleotide in which both MEF2 and NFAT core residues of wtMEF2(Ϫ278) have been mutated (dbmut(Ϫ278)) resulted in the retrieval of very low amounts of both proteins (Fig. 10, A and B, lane 5).
Comparison of different NFAT proteins revealed that NFATc2 bound preferentially to biotinylated Nur77 MEF2(Ϫ278) as compared with NFATc1; also, ternary complex formation proceeded independent of ⌬CaMKIVc co-expression (data not shown). Overall, these results demonstrate that NFAT and MEF2 proteins do form ternary complexes together with wtMEF2(Ϫ278) oligonucleotide, and that optimal incorporation of NFAT proteins into these complexes is dependent on both NFAT and MEF2 contact residues of wtMEF2(Ϫ278).
Phosphorylation of MEF2D by CaMKIV/Gr-To examine whether MEF2D is a direct target of CaMKIV/Gr phosphorylation, in vitro kinase assays were carried out using recombinant, His-Tagged protein spanning amino acids 87-514 of MEF2D (His-MEF2D 87-514 ). In these assays, recombinant wild type and catalytically inactive CaMKIV/Gr proteins were expressed in BJAB human Burkitt lymphoma cells, which lack endogenous CaMKIV/Gr. Transfected cells were treated with goat anti-human IgM antibodies to trigger Ca 2ϩ mobilization and kinase activation (32), following which recombinant kinase proteins were derived by immunoprecipitation and tested for their capacity to phosphorylate recombinant His-MEF2D 87-514 in vitro. The reaction mixture was resolved by SDS-PAGE and the status of MEF2D phosphorylation was determined by autoradiography. Fig. 11A demonstrates that recombinant wild type CaMKIV/Gr derived from anti-IgM-treated BJAB cells intensely phosphorylated His-MEF2D 87-514 (lane 5), whereas a catalytically inactive kinase mutant (CaMKIV/GrK75E) which has been derived under identical conditions failed to do so (lane 3). No phosphorylation was observed from immunoprecipitates of mock transfected cells (lane 1), or from cells transfected with either wild type or mutant kinase species but which have not been treated with anti-IgM (lanes 2 and 4, respectively). In these assays, equal amounts of wild type and mutant kinase proteins were present as revealed by immunblotting with an anti-kinase antiserum (Fig. 11B). Phosphoamino acid analysis revealed that CaMKIV/Gr-phosphorylated His-MEF2D 87-514 primarily on serine residues (Fig. 11C). These results suggest that MEF2D is amenable to CaMKIV/Gr phosphorylation. The functional significance of MEF2D phosphorylation remains under investigation. DISCUSSION Ca 2ϩ and immunosuppressant-sensitive transcriptional activation has regularly been mapped to NFAT response elements, either classical or B-like (2). Unexpectedly, this sensitivity mapped in some genes to response elements which recognize the MEF2 family of transcriptional factors but which did not appear to bind NFAT. In this report, we have established that Ca 2ϩ induction of MEF2-dependent transcription in lymphocytes is the result of a novel, combinatorial activation by two pathways of Ca 2ϩ signaling: CaMKIV/Gr and calcineurin. The two enzymes synergized in activating transcription from isolated MEF2 response elements derived from the Nur77 promoter/enhancer. Furthermore, products of genes harboring MEF2 elements, including Nur77 (this report) and BZLF1 (23), are potently induced by the synergistic action of CaMKIV/Gr and calcineurin. These results establish MEF2 elements as mediators of Ca 2ϩ -and immunosuppressant-sensitive transcription of a distinct subset of early activation genes.
Induction of MEF2-dependent transcription by calcineurin proceeded by an NFAT-dependent mechanism, most likely involving NFATc2. This was supported by several lines of evidence, including the capacity of NFATc2 but not NFATc1 or NFATc3 to substitute for calcineurin in supporting MEF2-dependent transcription, inhibition of Ca 2ϩ -activated, MEF2-dependent transcription by a dominant-negative NFATc2 mutant, co-precipitation of recombinant NFATc2 and MEF2 proteins upon their co-expression in cells, and the demonstration of ternary complexes composed of NFATc2 and MEF2 proteins and the Nur77 proximal MEF2 response element (MEF2(Ϫ278)). NFAT-MEF2 protein complexes formed independent of binding of either component to its respective response element (Fig. 9), and NFATc2 activated transcription by GAL4-MEF2 fusion proteins from GAL4 response elements which otherwise do not recognize NFAT (Fig. 8). Also, a MEF2 response element mutant (mutNFAT(Ϫ278)) which lacked core NFAT contact residues retained some responsiveness to calcineurin (Fig. 5D). However, the presence of adjacent NFAT contact residues, such as found at the 5Ј end of wtMEF2(Ϫ278), greatly enhanced both Ca 2ϩ responsiveness of and NFAT association with MEF2 response elements (Figs. 5D and 10). Taken together, evidence suggests that NFATc2 activation of MEF2-dependent transcription may proceed in the absence of demonstrable NFAT binding to DNA, but that such activation is most effective at composite NFAT/MEF2 elements such as wtMEF2(Ϫ278) which provide a nexus for NFAT interaction with DNA. The dual requirement for NFAT to interact with both MEF2 and DNA for optimal activation is evidenced by the observation that mutations at wtMEF(Ϫ278) which selectively disrupt MEF2 binding also result in greatly decreased NFAT association (Fig. 10). These results are also consistent with the recent demonstration of synergy between juxtaposed MEF2 and NFAT response elements in promoting Ca 2ϩ -dependent transcription (40).
Additional mechanisms may also contribute to MEF2/NFAT interaction, including accessory protein(s) such as transcriptional co-activator p300 which has been previously demonstrated to bind both MEF2 and NFAT proteins (41,42), and which may serve to stabilize NFAT-MEF2 complexes. Of note, MEF2 proteins have previously been described to associate with other transcription factors including MyoD and other myogenic and neurogenic basic helix loop helix-type transcription factors (43)(44)(45)(46). However, whereas previous reports have implicated the MADS box and MEF2 domains in pairing of MEF2 proteins with MyoD and other factors, the residual binding of NFATc2 to a truncated MEF2D mutant lacking the two domains ( Fig. 9) suggests that their role in NFAT binding, while salutary, is nevertheless dispensable.
While calcineurin activated both MEF2A and -D, CaMKIV/Gr selectively activated the latter by an independent mechanism. Previous studies on the regulation of MEF2-dependent transcription by phosphorylation have focused on the stress-activated p38 MAP kinase, which activates the MEF2 family members MEF2A and -C, but not MEF2D, by direct phosphorylation (47,48). However, activation by CaMKIV/Gr of MEF2D, but not MEF2A, -dependent transcription argues against involvement of p38 MAP kinase downstream of CaMKIV/Gr. The demonstration that CaMKIV/Gr mediates intense in vitro phosphorylation of recombinant MEF2D on serine residues suggests direct phosphorylation as a potential mechanism of MEF2D activation by CaMKIV/Gr (Fig. 11). Differential phosphorylation of MEF2 proteins, if confirmed by currently ongoing studies, may account for the selective activation by CaMKIV/Gr of MEF2D but not MEF2A. Activation by CaMKIV/Gr may also involve the transcriptional co-activator p300/CBP, which has been implicated in gene induction by both  32 P Radiolabeled His-MEF2D 87-514 was subjected to acid hydrolysis followed by one-dimensional phosphoamino acid analysis. P i denotes inorganic phosphate, and pS, pT, and pY the position of phosphoserine, phosphothreonine, and phosphotyrosine residues, respectively.
The integration by MEF2 elements of activation signals of two distinct Ca 2ϩ signaling pathways is a distinguishing regulatory feature which may serve to enhance the sensitivity of promoters/enhancers to induction by Ca 2ϩ even in the absence of other co-signals. Such a pathway may be particularly relevant to lymphocyte selection in the thymus, where Ca 2ϩ signaling is crucial to processes of lymphocyte selection. In this regard, it is interesting to note that CaMKIV/Gr achieves its highest levels in double positive thymocytes, an expression profile which may contribute to negative selection of these cells by Nur77 and its related family member Nor1 (50 -52).
Ca 2ϩ regulation of MEF2 activity may also be important in non-lymphoid cells such as neurons and myocytes, where MEF2 proteins have been implicated in regulating the expression of a number of genes (12). The recent description of a role for the NFAT family member NFAT3 in cardiac development and hypertrophy, and the association of NFAT3 with the cardiac transcriptional regulator GATA4 points to a broad partnership between NFAT and muscloskeletal transcription factors, including MEF2 proteins (53). The role of Ca 2ϩ signaling in regulating MEF2-dependent gene expression in non-lymphoid tissues remains the subject of ongoing investigation.