Nuclear factor of activated T cells and serum response factor cooperatively regulate the activity of an alpha-actin intronic enhancer.

Expression of alpha-actin in smooth muscle cells (SMCs) is regulated, in part, by an intronic serum response factor (SRF)-binding CArG element. We have identified a conserved nuclear factor of activated T cells (NFAT) binding site that overlaps this CArG box and tested the hypothesis that this site plays a previously unrecognized role in regulating alpha-actin expression. A reporter construct prepared using a 56-bp region of the mouse alpha-actin first intron containing SRF, NFAT, and AP-1 sites (SNAP) acted as an enhancer element in the context of a minimal thymidine kinase promoter. Basal reporter activity following expression in SMCs was robust and sensitive to the calcineurin-NFAT pathway inhibitors cyclosporin A and FK506. Mutating either the NFAT or SRF binding site essentially abolished reporter activity, suggesting that both NFAT and SRF binding are required. Basal activity in non-smooth muscle HEK293 cells was SRF-dependent but NFAT-independent and approximately 8-fold lower than that in SMCs. Activation of NFAT in HEK293 cells induced an approximately 4-fold increase in activity that was dependent on the integrity of both NFAT and SRF binding sites. NFATc3.SRF complex formation, demonstrated by co-immunoprecipitation, was facilitated by the presence of SNAP oligonucleotide. Inhibition of the calcineurin-NFAT pathway decreased alpha-actin expression in cultured SMCs, suggesting that the molecular interaction of NFAT and SRF at SNAP may be physiologically relevant. These data provide the first evidence that NFAT and SRF may interact to cooperatively regulate SMC-specific gene expression and support a role for NFAT in the phenotypic maintenance of smooth muscle.

Smooth muscle cells (SMCs), 1 unlike terminally differentiated cardiac and skeletal myocytes, are phenotypically dy-namic and maintain their differentiated phenotype through the regulated expression of a repertoire of smooth muscle-specific genes (1). This phenotypic flexibility is consistent with the physiological demands placed on vascular tissue and may underlie changes in smooth muscle structure and function that accompany pathological processes, such as those that occur in hypertension and atherosclerosis (2)(3)(4).
Serum response factor (SRF), a member of the MADS (MCM1, Agamous, Deficiens, SRF) family of transcription factors (5), is central to the regulation of a large subset of SMCspecific genes (1,6,7). Almost without exception, important regulatory regions of these SMC-specific genes contain CArG boxes (CC[A/T] 6 GG) that are indispensable for expression in smooth muscle (6). One motif that is common to many of these genes is a pair of CArG boxes, designated A and B, in the proximal upstream promoter region of the gene. The integrity of each of these elements, as well as conservation of their spacing and orientation, is required for efficient SMC expression (6). In addition to proximal CArG boxes, important CArG box-containing regulatory elements have been identified in intronic regions of SMC-specific genes, most notably within the first intron of genes for smooth muscle (SM) myosin heavy chain (MHC) and SM ␣-actin (6, 8 -13).
SRF transcriptional activity may be modulated by additional cofactors, some of which form complexes with SRF-CArG DNA, whereas others act to enhance SRF binding to CArG boxes without forming a detectable ternary complex (6,14,15). Although SRF activity is critically important in the regulation of SMC differentiation and SMC marker gene expression, CArG boxes alone are not sufficient to direct SMC-specific gene expression; additional cis-elements and trans-acting factors are required (reviewed in Ref. 6). Mechanisms that exist in parallel with the SRF-CArG pathway, as well as SRF-intersecting mechanisms, contribute to SMC phenotypic modulation and maintenance.
One of the more recent entries in the smooth muscle phenotypic maintenance derby is the calcium (Ca 2ϩ )-dependent transcription factor nuclear factor of activated T cells (NFAT), which regulates the expression of genes in a diverse array of immune and non-immune cells and is closely linked to developmental processes that involve modulation of cellular phenotypes (16 -24). NFAT also contributes to pathological processes, as exemplified by its role in the etiology of pathological, but not exercise-induced, cardiac hypertrophy (25)(26)(27).
NFAT activation is regulated primarily through control of its subcellular localization (16,28,29). In response to Ca 2ϩ -elevating stimuli, NFAT is dephosphorylated at multiple N-terminal phosphoserines by the Ca 2ϩ -dependent phosphatase, calcineurin, which promotes NFAT nuclear translocation and DNA binding competence (30,31). NFAT binds DNA with very low affinity in the absence of a cofactor, and in general, forma-tion of an NFAT⅐cofactor complex is required for significant NFAT-mediated transcriptional activity. The NFAT family consists of four members (NFAT1/c2, NFAT2/c1, NFAT3/c4, and NFAT4/c3) that share the property of Ca 2ϩ -dependent nuclear translocation and a fifth member, NFAT5, which is Ca 2ϩ -independent and shares limited homology with the other family members.
NFAT activity has been shown to contribute to the maintenance of the differentiated smooth muscle phenotype through regulation of ␣1 integrin and caldesmon expression (32) and has been implicated in regulating the expression of the SM MHC gene (33). More recently, NFAT has been proposed to regulate the expression of the K V 2.1 voltage-dependent potassium channel in SMCs and thereby regulate arterial smooth muscle excitability (34). A number of transcription factors that play a role in regulating SMC-specific gene expression are known to act as NFAT cofactors, including Egr-1, AP-1, YY1, the MADS family member MEF2, and members of the GATA family (Ref. 6 and the references therein). Using intact cerebral artery preparations, we have recently found that the NFATc3 isoform exhibits nitric oxide/protein kinase G-dependent nuclear localization in vascular smooth muscle cells in response to a normotensive pressure stimulus (35). The implication is that NFATc3 is constitutively nuclear and potentially active under normal physiological pressure and may thereby provide a link to the postulated role for protein kinase G in the phenotypic maintenance of differentiated smooth muscle (36). Mack and Owens (10) have demonstrated that in the context of a construct containing 5Ј CArG boxes A and B, exon 1, and intron 1, a double mutation in the ␣-actin intronic CArG box that disrupts SRF binding also disrupts normal SMC-specific expression, decreasing reporter activity in vitro and in vivo. By sequence analysis, we have identified a highly conserved NFAT binding site that overlaps this intronic SRF binding site. Intriguingly, the CArG box mutations that had been shown to disrupt SMC-specific expression also resulted in disruption of the overlapping NFAT site. The individual contributions of SRF and NFAT to enhancer activity have not been experimentally tested. Here we show that a 56-bp ␣-actin intronic element that contains conserved SRF, NFAT and AP-1 binding sites is capable of acting as an enhancer element, increasing the basal activity of a minimal promoter-reporter construct in both SMCs and HEK293 cells and conferring responsiveness to NFAT-activating stimuli in HEK293 cells. Mutations that independently disrupt either the NFAT site or the SRF binding site are each capable of essentially eliminating basal activity in SMCs and inducible activity in HEK293 cells, suggesting that both SRF and NFAT are required for enhancer activity under these conditions. In support of the idea that SRF and NFAT cooperate to mediate enhancer activity, we found that both SRF and NFAT bind simultaneously to the overlapping CArG box/NFAT site in this conserved intronic regulatory element under conditions that are associated with enhancer activity. Finally, in vitro expression of ␣-actin in SMCs was decreased by inhibiting NFAT activation, suggesting that the molecular interaction of NFAT and SRF at the intronic enhancer element may be physiologically relevant.

EXPERIMENTAL PROCEDURES
Cell Culture-The rat A7r5 aortic smooth muscle cell line and an HEK293 human embryonic kidney-derived cell line (BD Biosciences) were grown in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% fetal bovine serum and penicillin-streptomycin (10 units/ ml; Invitrogen). G418 (2 mg/ml) was included in cultures of this reverse tetracycline transactivator (rtTA)-expressing HEK293 derivative (used in experiments involving co-transfection of a Tet operator-EGFP-NFATc3 construct) to maintain stable rtTA expression. The cultures were maintained in a humidified incubator at 37°C and 5% CO 2 .
Construction of Plasmids-A 56-bp region of the ␣-actin first intron (corresponding to nucleotides 1035-1090 in GenBank TM accession number U63129) containing a previously identified SRF-binding CArG box, an overlapping NFAT binding motif, and an AP-1 binding site, designated SNAP (SRF/NFAT/AP-1), was used as a starting point in the design of reporter constructs (see Fig. 1). A double-stranded oligonucleotide covering this region was prepared by combining equimolar amounts of single-stranded sense and antisense oligonucleotides (IDX Technologies), heat denaturing (95°C, 5 min), and allowing them to slowly anneal at room temperature. SNAP oligonucleotides, synthesized to contain NheI sites at each end (61 bp, total length), were cloned into an NheI site present in the multi-cloning region of a luciferase reporter plasmid containing a minimal thymidine kinase (TK) promoter (pTAL; BD Biosciences). After transforming competent Escherichia coli (Invitrogen) and preparing DNA minipreps from isolated colonies (Qiagen), individual clones were screened by restriction analysis to determine the number of inserted copies and then sequenced to determine the orientation of each insert. Reporter constructs containing mutations in the NFAT (AGGTTT), SRF (AATAATTAGG), or AP-1 (TCATACA) binding sites were similarly prepared.
Full-length NFATc3 was prepared from a plasmid (pSH205A) containing an alternatively spliced murine NFATc3 cDNA (37) kindly provided by Dr. Gerald Crabtree. The N terminus of native NFATc3 was amplified from mouse smooth muscle total RNA by reverse transcription-PCR and ligated into an XbaI site defining the 5Ј end of the NFATc3 cDNA. A Tet-operated EGFP-NFATc3 fusion protein expression plasmid (pTetOP-EGFP-NFATc3-HGHpA) was prepared by cloning full-length NFATc3 into a pTetOP-EGFP-HGHpA plasmid constructed from pTetOP-HGHpA (containing a Tet operator and human growth hormone poly(A) tail) (38) and pEGFP-C3 (Clontech).
Transient Transfection and Luciferase Assay-HEK293 cells were transfected with TK-luciferase reporter constructs containing one, two, or three tandem copies of the SNAP element in the all-forward or all-reverse orientation using the FuGENE 6 transfection reagent (3:1 ratio of transfection reagent (l)/total plasmid DNA (g)) according to the manufacturer's protocol (Roche Applied Science). Luciferase activity in cell lysates was determined using the Luciferase Assay System according the manufacturer's protocol (Promega) and expressed relative to empty pTAL vector controls. Luciferase activity was normalized to protein concentration.
A7r5 cells were co-transfected with wild-type or mutant SNAP reporter constructs or pTAL empty vector and pRL-TK (Renilla expression plasmid; Promega) using Superfect transfection reagent (10 l reagent/2 g total DNA/well) according to the manufacturer's protocol (Qiagen). Luciferase and Renilla activity were simultaneously detected (Dual-Luciferase Reporter Assay System; Promega); Renilla activity was used to normalize luciferase activity.
In some experiments, exogenously expressed EGFP-NFATc3 was used to monitor NFAT subcellular localization and transfection efficiency. For experiments employing HEK293 cells, an HEK293 derivative stably expressing rtTA was co-transfected with TetOP-EGFP-NFATc3 and cultured in the presence of doxocyclin (2 g/ml) to induce EGFP-NFATc3 expression. A7r5 cells were also co-transfected with an rtTA expression plasmid, but were otherwise treated the same as HEK293 cells.
Whole-cell Extracts-After trypsinization, cells were washed three times in PBS, resuspended in lysis buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, and 1% Nonidet P-40) containing protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, 9 g/ml leupeptin, and 10 g/ml pepstatin), and incubated for 30 min on ice. Lysates were cleared by centrifugation at 14,000 rpm for 5 min at 4°C, and the supernatants were assayed for protein content using the Bradford assay with bovine serum albumin as the protein standard. Aliquots were stored at Ϫ80°C prior to use. Whole-cell extracts, containing a mixture of phosphorylated (primarily cytoplasmic) and dephosphorylated (primarily nuclear) NFAT, are a convenient source of material for DNA binding assays (below) because only the calcineurin-activated, dephosphorylated form exhibits significant DNA binding competence.
Avidin Biotin Conjugated DNA Binding (ABCD) Assay-Binding of SRF and/or NFAT to the SNAP element was determined using cellular extracts and biotin-conjugated nucleotide probes/streptavidincoated agarose beads in a modification of the DNA-protein binding assay developed previously by Wu and co-workers (39,40). For each binding assay, whole-cell extracts containing 500 g of protein were incubated with 400 ng of biotinylated oligonucleotide probes in binding buffer (20 mM Tris, pH 8.0, 1 mM EDTA, 100 mM NaCl, 0.5% Nonidet P-40, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl flu-oride, 5 g/ml aprotinin, 9 g/ml leupeptin, 10 g/ml pepstatin, and 10 mM sodium molybdate) for 30 min on ice. Streptavidin-conjugated agarose beads (Pierce) were added (25 l of a 50% slurry), and the suspensions were placed on a gyrating shaker platform at 4°C for 1-3 h. The beads were pelleted by centrifugation at 9000 ϫ g (4°C), washed three times in binding buffer, and eluted in 2ϫ SDS loading buffer for SDS-PAGE and Western analysis.
Co-immunoprecipitation-Whole-cell lysates from HEK293 cells (500 g total protein content/sample, treated as described in the text) were incubated with 500 ng of mouse anti-NFATc3-agarose conjugate (Santa Cruz Biotechnology) in DNA-protein binding buffer for 24 h at 4°C with constant rocking in the presence or absence of unlabeled SNAP oligonucleotide. The beads were pelleted by centrifugation (2000 ϫ g at 4°C for 5 min), and the pellet was washed three times in a buffer consisting of 10 mM Hepes, pH 7.5, 150 mM NaCl, and 0.1% Nonidet P-40. After the third wash, the pellet was resuspended in SDS-loading buffer and boiled for 5 min at 95°C. The proteins were separated by SDS-PAGE, transferred to nitrocellulose, and probed for SRF or NFATc3 by Western blotting. The absence of nonspecific binding to NFATc3-agarose beads was confirmed by stripping and reprobing for glyceraldehyde-3phosphate dehydrogenase (primary antibody from Chemicon International; 1:1000 dilution for 1 h at room temperature; data not shown).
Western Blot Analysis-Protein samples were separated by SDS-PAGE using 7.5% acrylamide gels (Bio-Rad) in standard Laemmli buffer and transferred to nitrocellulose membranes (Bio-Rad). After blocking for 1 h at room temperature in Tris-buffered saline containing 5% nonfat milk and 0.1% Tween 20, blots were exposed to primary antibodies (diluted in blocking solution) overnight at 4°C or for 1 h at room temperature. The primary antibodies used were mouse monoclonal NFATc3 (1:200; Santa Cruz Biotechnology) and rabbit polyclonal SRF (1:4000; Santa Cruz Biotechnology). After multiple washes in Tris-buffered saline, horseradish peroxidase-conjugated secondary antibody (from Santa Cruz Biotechnology (1:10,000) or Pierce (1:2000) in blocking solution) was applied for 1 h at room temperature. Bands were visualized using standard or high sensitivity chemiluminescence detection systems (Pierce Supersignal West Dura or Femto, respectively).
Immunofluorescence-HEK293 and A7r5 cells were grown in an 8-well Lab-Tek chamber slide (Nalge Nunc). Cells were fixed with 4% formaldehyde in PBS for 15 min, permeabilized in 0.1% Triton in PBS for 10 min, and then blocked in 3% donkey serum in PBS for 1 h. Primary antibody (rabbit polyclonal anti-NFATc3, 1:100; Santa Cruz Biotechnology) was prepared in 0.1% gelatin in PBS and applied overnight at 4°C. Secondary antibody (anti-rabbit Cy5, 1:500; Jackson Immunoresearch Laboratories) was prepared in 0.1% gelatin in PBS and applied for 1 h at room temperature. Nuclei were stained using SYTOX green (1:5000 in PBS).
Cells were examined at ϫ40 magnification using a Bio-Rad 1000 laser scanning confocal microscope. NFAT and nuclear staining were detected by sequentially monitoring the Cy5 and SYTOX fluorescence using excitation wavelengths of 650 and 488 nm and emission wavelengths of 670 and 520 nm, respectively. Specificity of immune staining was confirmed by the absence of fluorescence in cells incubated with secondary antibody alone. For scoring of NFAT-positive nuclei, multiple fields for each replicate of each experimental condition were imaged and counted by two independent observers under double-blind conditions with the aid of Metamorph software (Universal Imaging Corp). The software was programmed so that individual pixels within a given image would appear white if co-localization of the green nucleic acid stain and the Cy5-NFAT stain occurred. Thus, for quantification purposes, a cell was considered positive if co-localization (white) was distributed throughout the nucleus, whereas a cell was considered negative if no co-localization (green only) or only isolated co-localization was observed.
In-cell Western Analysis-SM ␣-actin expression in SMCs was quantified using an infrared-based in-cell Western assay system as described by the manufacturer (Li-Cor Biosciences). Briefly, confluent cultures of A7r5 cells were trypsinized and seeded into 96-well plates (70% confluence). After 48 h in 1% fetal bovine serum, cells were treated as described in the text, washed once with PBS, fixed with 4% formaldehyde in PBS for 15 min at room temperature, and blocked (Aqua Block with 0.1% Tween 20). Primary antibody (mouse monoclonal anti-SM ␣-actin (1:250; Biogenex) or mouse monoclonal anti-␤-actin (1:200; Sigma)) was applied overnight at 4°C, followed by secondary antibody (goat anti-mouse IRDye 800 (1:800 dilution in Aqua Block with 0.1% Tween 20; Rockland)) for 1 h. TOTO-3 (1:1000; Molecular Probes) was used to counterstain DNA. Infrared fluorescence was monitored using an Odyssey Imager (700 nm for TOTO-3 and 800 nm for IRDye 800). The intensity of both channels was set at 7.5, with a focus setting of 3 mm. Integrated intensity of SM ␣-actin or ␤-actin (housekeeping protein) stain was normalized to the integrated intensity of DNA stain. Because of the ease and accuracy with which multiple individual replicates can be independently analyzed, this approach is especially useful where quantitation is the goal.

A CArG Box-containing Region of the Mouse ␣-Actin First
Intron Acts as an NFAT-and SRF-dependent Enhancer Element-We have identified a putative NFAT binding site in the ␣-actin first intron that overlaps a functionally important SRFbinding CArG box (10). The sequence of this SRF-NFAT binding site-containing region is remarkably conserved among rodent, human, and chicken ␣-actin and includes a completely conserved AP-1 site at a position ϳ24 bp downstream of the NFAT site (Fig. 1). To test these putative binding sites for potential enhancer activity, an oligonucleotide encompassing this mouse ␣-actin first intronic region containing SRF, NFAT, and AP-1 binding sites (SNAP) was cloned into a luciferase reporter plasmid upstream of a minimal TK promoter. Constructs containing one, two, or three tandem copies of SNAP in either forward or reverse orientation were tested in HEK293 cells and SMCs for luciferase expression. The minimal TK promoter-luciferase construct containing three tandem copies of the insert in the reverse orientation showed the highest reporter activity of those constructs tested (data not shown) and was used in subsequent experiments. The wild-type reporter construct (SNAP-Luc) and variants containing mutated SRF (S mut NAP-Luc), NFAT (SN mut AP-Luc), or AP-1 (SNAP mut -Luc) binding sites are depicted schematically in Fig. 2A. Fig. 2, B and C, shows luciferase activity, normalized to the minimal TK promoter-containing (empty) pTAL vector, in SNAP-Luc-transfected SMCs and HEK293 cells, respectively. In both cell lines, the SNAP element promoted a significant increase in reporter activity under basal (unstimulated) conditions, demonstrating that this small region of the mouse ␣-actin first intron is capable of acting as a functional enhancer element. Inhibition of the calcineurin/NFAT pathway with FK506 (1 M) and CsA (1 M) significantly decreased reporter activity in SNAP-Luc-transfected SMCs (Fig. 2B), suggesting NFAT involvement.
To determine the relative contribution of SRF, NFAT, and/or AP-1 sites to the enhanced reporter activity, HEK293 cells and SMCs were transfected with SNAP-Luc constructs containing mutations at each of these sites, designed such that mutations in one binding element were independent of each of the other binding elements. Mutating either the NFAT site (SN mut AP-Luc) or the SRF binding site (S mut NAP-Luc) almost completely abolished basal enhancer activity in SMCs, suggesting that both NFAT and SRF binding sites were required for this activity (Fig. 2B). Mutating the AP-1 binding site (SNAP mut -Luc) partially decreased basal activity in SMCs (Fig. 2B), suggesting that full activity may require formation of an SRF⅐NFAT⅐AP-1 complex. In contrast, only SRF binding appears to be required for basal activity in HEK293 cells because a mutation in the  A and B, exon 1, and intron 1. The intronic SNAP element used in these studies is composed of an SRF binding element (CArG box), an overlapping NFAT binding site, and an AP-1 binding site shown in red, blue, and green, respectively. The CArG box and NFAT site are identical in mouse and rat; nucleotides in capital letters are completely conserved among chicken, mouse, and human.
NFAT/AP-1 Pathway Stimulation Increases SNAP Enhancer Activity in HEK293 Cells-Concurrent treatment with the Ca 2ϩ ionophore, ionomycin (Io), and phorbol 12-myristate 13acetate acetate (PMA) is a stimulus paradigm that is commonly used to simultaneously activate NFAT and NFAT cofactor pathways (41)(42)(43). Dephosphorylation by the Ca 2ϩ -dependent phosphatase, calcineurin, serves to promote both NFAT nuclear accumulation and DNA binding (30). Stimulation of SNAP-Luc-transfected HEK293 cells with Io (1 M) and PMA (100 nM) induced an ϳ4-fold increase in luciferase activity over unstimulated conditions (Fig. 3). This inducible luciferase activity was significantly inhibited by pre-treatment with the calcineurin inhibitors CsA and FK506, suggesting that dephosphorylation-dependent translocation of NFAT from the cytosol to the nucleus contributes to the observed increase in activity. The Io/PMA-induced increase in reporter activity, like the basal enhancer activity in SMCs, was dependent on the integrity of both SRF and NFAT binding sites, as indicated by the virtual absence of inducible activity in HEK293 cells transfected with reporter constructs containing mutations in the NFAT (SN mut AP-Luc) or SRF (S mut NAP-Luc) sites. Similar to results obtained under basal conditions in SMCs, a mutation in the AP-1 site (SNAP mut -Luc) partially but significantly reduced the response to Io/PMA stimulation in HEK293 cells. Interestingly, the high basal activity of the SNAP-Luc construct in SMCs (ϳ8-fold higher than that in similarly transfected HEK293 cells) cannot be further enhanced by treatment with Io/PMA (data not shown; see "Discussion").
NFATc3 and SRF Simultaneously Bind to the SNAP Enhancer Element-Functional effects of individual site mutations on SNAP enhancer activity strongly suggest that NFAT and SRF bind simultaneously at overlapping SRF and NFAT sites within the ␣-actin intronic regulatory element under study. To determine whether this was the case, we performed ABCD assays to pull down SNAP-associated proteins from cellular extracts, followed by Western analysis. Fig. 4A shows a representative ABCD assay employing whole-cell lysates from non-stimulated and Io/PMA-stimulated HEK293 cells. Consistent with the functional data, only SRF was bound to the oligonucleotide probe under non-stimulated conditions (Fig. 4A,  lane 1). Under stimulated conditions, both SRF and NFATc3 were present in the DNA⅐protein complex (Fig. 4A, lane 2). Thus, the overlapping CArG box/NFAT site in the ␣-actin first intron can bind both NFATc3 and SRF in response to an NFATactivating stimulus.
In competition assays employing either HEK293 or SMC whole-cell lysates, SRF binding to biotinylated SNAP was abrogated in the presence of an excess of unlabeled wild-type SNAP oligonucleotides. Unlabeled probes containing a scrambled sequence or a mutation in the SRF binding site (S mut NAP), however, failed to compete for SRF binding, demonstrating that the observed binding is sequence-specific (Fig. 4, B and C). To rule out the possibility that mutations in the adjacent NFAT or AP-1 sites might disrupt SRF binding, even though they do not disrupt the CArG site itself, we performed competition binding assays using unlabeled SNAP containing mutated NFAT (SN mut AP) or AP-1 (SNAP mut ) binding sites. In each case, an excess of unlabeled oligonucleotides was able to prevent SRF binding to the labeled probe, indicating that an intact CArG box alone is sufficient to bind SRF.

FIG. 2. A CArG box-containing region of the mouse ␣-actin first intron acts as an NFAT-and SRF-dependent enhancer under basal conditions.
A, reporter constructs containing three tandem copies (depicted in forward orientation) of the SM ␣-actin first intronic enhancer element (SNAP) used in transfection experiments: wild-type (SNAP-Luc), NFAT-mutated (SN mut AP-Luc), SRF-mutated (S mut NAP), or AP-1mutated (SNAP mut ). B, A7r5 SMCs were transfected with the SNAP-Luc construct or its mutants, as described under "Experimental Procedures." Activity is expressed relative to the empty pTAL vector, which contains a minimal thymidine kinase promoter-luciferase expression cassette (TK-Luc). A subgroup of cells transfected with SNAP-Luc (wild-type) was pre-treated with 1 M FK506 and 1 M CsA. *, p Ͻ 0.001 versus all; n ϭ 6 -9. C, HEK293 cells were transfected with the SNAP-Luc wild-type and mutant constructs, as described. Luciferase activity was normalized to reporter activity of cells transfected with empty vector. *, p Ͻ 0.01 versus all; n ϭ 6.
These experiments do not rule out the formal possibility that NFATc3 and SRF are binding individually to separate populations of probe molecules rather than binding simultaneously to the same probe. To address this, we immunoprecipitated NFATc3 in the presence or absence of unlabeled SNAP oligonucleotide and probed for associated SRF by Western analysis (Fig. 5). Under standard detection conditions (Fig. 5A), a strong SRF band is evident in the presence of the SNAP probe (lane 1), whereas no SRF band is evident in the absence of SNAP (scrambled oligonucleotide, lane 2; no oligonucleotide, lane 3). By increasing the gain (i.e. using a higher sensitivity chemiluminescence detection system), we are able to show that SRF is present in NFATc3 immunoprecipitates in the absence of SNAP, albeit to a lesser extent (Fig. 5B). These results indicate that SNAP acts as a sequence-specific scaffold to facilitate NFATc3 and SRF interactions, but they also suggest that NFATc3 and SRF may interact in solution.
NFATc3 Exhibits Constitutive Nuclear Localization in A7r5 Cells but Not HEK293 Cells-In general, NFAT is localized to the cytosol of unstimulated cells. However, in the A7r5 cells used in this study, we found that in an overwhelming majority of cells (95.33 Ϯ 2.53%, n ϭ 3 images (81 cells)), endogenous NFATc3 is constitutively localized to the nucleus under basal conditions (Fig. 6A, right panel). Inhibition of calcineurin activity by treatment with CsA and FK506 (15 min) induced a substantial redistribution of nuclear NFATc3 to the cytoplasm (NFATc3-positive nuclei ϭ 27.40 Ϯ 7.40%, n ϭ 3 images (43 cells); p Ͻ 0.001 versus basal). Overexpressed EGFP-NFATc3 also exhibits nuclear localization in SMCs in the absence of an NFAT-activating stimulus (Fig. 6B, right panel). This distribution is strikingly different from the primarily cytosolic and perinuclear localization of both endogenous (Fig. 6A, left panel) and exogenously expressed (Fig. 6B, left panel) NFATc3 in HEK293 cells. These results are consistent with both the elevated basal activity of SNAP-Luc in SMCs and the requirement for an NFAT-activating stimulus for maximal reporter activity in HEK293 cells.

Calcineurin/NFAT Pathway Inhibition Reduces ␣-Actin Expression in Smooth Muscle Cells-The fact that NFAT and SRF
can simultaneously bind to a CArG/NFAT site corresponding to a region of the ␣-actin intron known to be important for regulating SMC-specific ␣-actin expression (8,10) suggests that NFAT cooperates with SRF in the regulation of ␣-actin expression. To confirm that the calcineurin/NFAT pathway is involved in regulating the expression of SM ␣-actin, we treated confluent, serum-deprived A7r5 SMCs with the calcineurin inhibitors FK506 and CsA, and we measured changes in SM ␣-actin protein expression by in-cell Western analysis (see "Ex-  1 and 2) or absence (lanes 3 and 4) of biotin-labeled SNAP (Biotin-SNAPwt) probe. Streptavidin-agarose-conjugated beads were added and incubated for 2 h. Pellet eluted was separated by SDS-PAGE and blotted for NFATc3 or SRF. Lanes 5 and 6 contain 30 g of whole-cell extract. B, whole-cell lysates (500 g of protein) from Io/PMA-treated HEK293 cells were pre-incubated for 30 min with non-labeled probe containing wild-type (SNAPwt), NFAT-mutated (SN mut AP), SRF-mutated (S mut NAP), AP-1mutated (SNAP mut ), or scrambled sequences (lanes 4 -8). Lane 9 contains 30 g of whole-cell extract. C, A7r5 whole-cell lysates (500 g of protein) were pre-incubated for 30 min with non-labeled probe containing wildtype, mutated, or scrambled sequences (lanes 3-7). Lane 8 corresponds to 30 g of whole-cell extract. Representative experiments of n ϭ 3. perimental Procedures"). Results are expressed relative to ␤-actin, and expression of both proteins was normalized to DNA content. Fig. 7 shows that inhibition of NFAT activation significantly decreased SM ␣-actin expression, reducing protein levels by nearly 50% compared with untreated controls. These results suggest that the NFAT-dependent regulation of SNAP enhancer activity, described above, is relevant to the regulation of ␣-actin expression. DISCUSSION We have found that a 56-bp region of the mouse SM ␣-actin first intron that contains a highly conserved SRF/NFAT/AP-1 (SNAP) composite binding site sequence is capable of acting as an enhancer element, increasing reporter activity in the context of a minimal TK promoter. These results are consistent with a previous report showing that a larger (180-bp) segment of the corresponding region of the human ␣-actin first intron, cloned upstream or downstream of a minimal SV40 promoter, exhibits enhancer activity (13) and further support more recent results showing that the corresponding intronic region of the rat ␣-actin gene is required for robust in vitro and in vivo reporter expression in the context of a construct that includes 5Ј-regulatory elements and the first exon (10). More importantly, our results indicate that the previously unrecognized NFAT site and the overlapping SRF-binding CArG element are both absolutely required for the elevated basal activity in SMCs and induced activity in HEK293 cells. The SNAP region simultaneously binds both SRF and NFAT in HEK293 cells under conditions in which activity is induced but binds only SRF in the absence of stimulation, providing the first evidence that NFAT and SRF may interact to cooperatively regulate gene expression. Our results also show that inhibition of the calcineurin/NFAT pathway decreases ␣-actin expression in SMCs, consistent with a role for NFAT in the regulation of ␣-actin transcription.
In HEK293 cells transfected with the SNAP-Luc construct, reporter expression in the absence of stimulation was signifi- cantly increased over that in cells transfected with a control luciferase reporter plasmid containing only a minimal TK promoter. The basal level of expression in HEK293 cells, however, was ϳ8-fold lower than that in SNAP-Luc-transfected SMCs, consistent with a potential SMC-selective role for this element. Basal expression in HEK293 cells was sensitive to disruption of the CArG box but unaffected by mutations in the NFAT site or the AP-1 site. This is in sharp contrast to Io/PMA-stimulated HEK293 cells (or unstimulated SMCs), in which mutations in NFAT or SRF sites were equally effective in abrogating reporter activity and a mutation to the AP-1 site partially inhibited activity. Thus, in non-smooth muscle HEK293 cells, SRF alone appears capable of mediating a low basal level of transcription from the SNAP enhancer element, whereas the induced activity in these cells and the much higher basal activity in SMCs require both NFAT and SRF.
Interestingly, in SMCs, stimulation of the calcineurin/NFAT pathway with Io/PMA failed to increase the reporter activity of the SNAP-Luc construct beyond the already elevated basal level. The absence of inducible activity in SNAP-Luc-transfected SMCs, as well as the profound difference in basal enhancer activity between SMCs and HEK293 cells, may be partially accounted for by differences in the subcellular distribution of the NFATc3 isoform between these two cell types. Although NFAT is normally presumed to localize to the cytosol of unstimulated cells, we found that a preponderance of endogenous as well as exogenously expressed NFATc3 is constitutively localized to the nucleus of A7r5 cells. In HEK293 cells, both endogenous and exogenously expressed NFATc3 is localized to the cytosol under basal conditions. Thus, a virtual absence of nuclear NFATc3 in unstimulated HEK293 cells is associated with low level, NFAT-independent SNAP enhancer activity. Stimulation of HEK293 cells with Io/PMA increases nuclear NFATc3 (data not shown) and SNAP enhancer activity. In SMCs, in which NFATc3 is constitutively nuclear, SNAP activity is high, NFAT-dependent, and unresponsive to Io/PMA treatment. Although these results do not necessarily indicate that SNAP preferentially binds the NFATc3 isoform, they are consistent with the interpretation that sufficient nuclear NFAT in the form of NFATc3 is present in SMCs to drive maximal enhancer activity under basal conditions. The constitutively nuclear NFATc3 observed in SMCs is redistributed to the cytosol following inhibition of calcineurin with FK506 and CsA (Fig. 6C). This result suggests that NFAT import mechanisms may be constitutively elevated in these cultured SMCs, an observation that could be explained by elevated calcineurin activity. Alternatively, mechanisms that act to promote nuclear export may be less active under basal conditions. Recent work from our laboratory has shown that regulation of NFATc3 nuclear export may be central to the activation state of NFAT in smooth muscle (reviewed in Ref. 44). In unpressurized arteries, basal NFATc3 nuclear export is high as a result of elevated JNK2-mediated export; stimuli that induce a net nuclear accumulation of NFATc3 are those that not only elevate intracellular Ca 2ϩ but also result in a decrease in JNK2-mediated export (45). In response to a normotensive pressure stimulus, JNK2 activity in native smooth muscle is reduced via a nitric oxide/protein kinase G-dependent pathway, and NFATc3 is predominantly localized to the nucleus (35). In this respect, A7r5 may exist in a state similar to that of pressurized arteries, with reduced JNK2-mediated export activity and constitutively nuclear NFATc3.
SMCs are known to express higher levels of SRF than other cell types (46,47), so it is conceivable that the difference between the levels of basal and induced enhancer activity between SMCs and HEK293 cells may also reflect limiting SRF levels in HEK293 cells. Although HEK293 cells do contain sufficient SRF to drive low basal levels of enhancer activity and to detect in a complex with the SNAP element, it may be that higher levels of SRF are required for efficient recruitment of NFAT and maximal enhancer activity.
Our reporter assay data showing that basal SNAP enhancer activity in SMCs and induced activity in HEK293 cells are dependent on both intact SRF and NFAT sites suggest a cooperative mechanism of action between SRF and NFAT. This idea is supported by the binding of both NFATc3 and SRF to the SNAP element under conditions in which the NFAT pathway is activated in HEK293 cells. The presence of SRF in NFATc3 immunoprecipitates provides compelling evidence that SRF and NFATc3 interact. The fact that associated SRF is increased in the presence of SNAP is most consistent with the interpretation that simultaneous binding of SRF and NFATc3 to their respective sites facilitates NFATc3/SRF interactions. SRF binding alone is insufficient to promote enhancer activity under NFAT stimulated conditions because a mutation at the NFAT binding site does not substantially inhibit SRF binding to the overlapping CArG box but almost completely abrogates induced reporter activity.
SRF, a member of the MADS family of transcription factors (5), has been shown to bind to the REL homology domain of nuclear factor-B (48) to enhance nuclear factor-B transcriptional activity. NFAT, which also contains a REL homology domain, has been previously shown to functionally interact with the MADS family member Mef2 (21). This is the first report, however, to suggest that NFAT and SRF may interact with one another to function as transcriptional co-activators. The SRF and NFAT domains that may be responsible for mediating protein-protein interactions in the putative SRF⅐NFAT complex, as well as the role that the individual NFAT and/or SRF transactivation domains may play in mediating interactions with the transcriptional machinery, are currently unknown. Our data also suggest that AP-1, which is a common NFAT cofactor (43) that has also been implicated in SRF signaling (49), contributes to NFAT/SRF enhancer efficacy, although the molecular details of this involvement remain to be defined.
SRF has been previously shown to interact with other transcription factors, such as ETS domain-containing transcription factors (e.g. Elk-1), the homeodomain proteins Barx1b and Nkx3.2, and the zinc finger protein GATA6 (6). In smooth muscle, the transcription factor myocardin has been shown to regulate SRF-dependent transcription of a Ϫ2.6/ϩ2.8-kb ␣-actin promoter/intronic construct through interactions that exhibit a combinatorial dependence on CArG elements, primarily involving CArG B and CArG A boxes (50). The apparent preferential interaction of myocardin with proximal promoter CArG boxes suggests that NFAT action at the intronic SNAP element may operate in parallel with myocardin activity, with both pathways required for full transcriptional activity. Considered in the context of the demonstrated Ca 2ϩ dependence of smooth muscle differentiation marker gene expression (51), it is possible that the NFAT pathway may contribute to the Ca 2ϩ sensitivity component of SMC gene expression, serving to integrate Ca 2ϩ signals at regulatory elements of smooth musclespecific genes, including ␣-actin.
In addition to the molecular evidence supporting a role for NFAT in the regulation of ␣-actin expression, this study shows that inhibition of the calcineurin/NFAT pathway in cultured rat SMCs decreases ␣-actin protein expression. This result is in accord with a recent report showing that blockade of calcineurin/NFAT in primary cultures of human vascular smooth muscle is associated with a reduction in both ␣-actin and SM MHC protein expression (33). Another recent study has also implicated NFAT in the regulation of ␣1 integrin and caldesmon expression in cultured visceral SMCs, showing that calcineurin activity is required for the expression of these smooth muscle markers and maintenance of the contractile smooth muscle phenotype (32). The proposed interactions between NFAT and SRF at the ␣-actin intronic composite binding site may provide a molecular framework to help explain the apparent NFAT dependence of ␣-actin expression in SMCs, but it appears unlikely that this interaction can be invoked as a general mechanism to account for the NFAT dependence of other smooth muscle-specific genes. For example, although the regulation of SM MHC expression in smooth muscle exhibits similarities to that of ␣-actin regulation in that dual 5Ј CArG boxes and an intronic CArG box are all required for smooth muscle-specific expression, neither the 5Ј nor intronic CArG box is associated with NFAT binding sites. Instead, NFAT appears to regulate SM MHC expression through an association with GATA at unidentified sites in the 5Ј-flanking sequence of the SM MHC gene (33). The single 5Ј-upstream CArG box of ␣1 integrin that is required for expression in differentiated SMCs (52) likewise lacks an associated NFAT site. In addition to the conserved intronic NFAT site characterized here, the first intron of ␣-actin is rich in potential NFAT binding sites, many of which are phylogenetically conserved and might be expected to contribute to NFAT-sensitive expression. To the extent that NFAT is involved in regulating the expression of SMC-specific genes, it is likely that it does so through multiple mechanisms involving interactions with different cofactors.