Vascular smooth muscle alpha-actin gene transcription during myofibroblast differentiation requires Sp1/3 protein binding proximal to the MCAT enhancer.

The conversion of stromal fibroblasts into contractile myofibroblasts is an essential feature of the wound-healing response that is mediated by transforming growth factor beta1 (TGF-beta1) and accompanied by transient activation of the vascular smooth muscle alpha-actin (SmalphaA) gene. Multiple positive-regulatory elements were identified as essential mediators of basal SmalphaA enhancer activity in mouse AKR-2B stromal fibroblasts. Three of these elements bind transcriptional activating proteins of known identity in fibroblasts. A fourth site, shown previously to be susceptible to single-strand modifying agents in myofibroblasts, was additionally required for enhancer response to TGF-beta1. However, TGF-beta1 activation was not accompanied by a stoichiometric increase in protein binding to any known positive element in the SmalphaA enhancer. By using oligonucleotide affinity isolation, DNA-binding site competition, gel mobility shift assays, and protein overexpression in SL2 and COS7 cells, we demonstrate that the transcription factors Sp1 and Sp3 can stimulate SmalphaA enhancer activity. One of the sites that bind Sp1/3 corresponds to the region of the SmalphaA enhancer required for TGF-beta1 amplification. Additionally, the TGF-beta1 receptor-regulated Smad proteins, in particular Smad3, are rate-limiting for SmalphaA enhancer activation. Whereas Smad proteins collaborate with Sp1 in activating several stromal cell-associated promoters, they appear to operate independently from the Sp1/3 proteins in activating the SmalphaA enhancer. The identification of Sp and Smad proteins as essential, independent activators of the SmalphaA enhancer provides new insight into the poorly understood process of myofibroblast differentiation.

Although the molecular control of myofibroblast differentiation is largely unknown, several studies have shown that transforming growth factor ␤1 (TGF-␤1) may be particularly important in their recruitment to sites of tissue inflammatory damage. TGF-␤1 is a potent Sm␣A gene transcriptional activator in both granulation tissue and isolated fibroblasts (2,5,7,15). TGF-␤1 also induces expression of Sm␣A in cultured rat aortic smooth muscle cells, bovine aortic endothelial cells, and rat fibroblasts, where specific transcriptional control elements contained within the rat Sm␣A gene promoter were shown to mediate TGF-␤1-dependent activation (16,17). However, the sequence context and position of DNA-regulatory elements identified in studies on the rat Sm␣A promoter and aortic smooth muscle cells differed from those in the mouse Sm␣A enhancer that undergoes TGF-␤1-dependent chromatin conformational changes in AKR-2B stromal fibroblasts (15). TGF-␤1 elicits both tissue-specific and cell culture context-specific responses (18,19), yet studies performed to date have not resolved whether sequence elements required for activation from TGF-␤1-dependent promoters are distinctly different or evolutionarily conserved.
Previous studies from our laboratories have indicated that the mouse Sm␣A gene is regulated by an array of both positive and negative cis-acting elements that behaved differently in fibroblasts compared with muscle cells (20 -27). These studies resulted in the identification of a minimal enhancer element that was constitutively active in fibroblasts, immature myoblasts, and cultured aortic smooth muscle cells, but not in quiescent, differentiated skeletal muscle myocytes. The aim of the present work was to determine whether the mouse Sm␣A enhancer also conveyed the TGF-␤1 response in fibroblasts and whether sequence elements and transcription factors previously identified as responsible for TGF-␤1 inducibility in rat aortic smooth muscle cells also directed this effect in mouse fibroblasts. Our results indicate that TGF-␤1 amplifies basal expression of the mouse Sm␣A enhancer by concerted action at five positive regulatory sites. Interestingly, stoichiometric changes in nuclear protein binding activity at these specific sites were not observed in TGF-␤1-activated mouse fibroblasts. Whereas three of these sites bind activating proteins of known identity, proteins interacting with the previously identified TGF-␤1 control element (TCE) plus a newly identified TGF-␤1 hypersensitivity region (THR) in fibroblasts were determined to be the Sp1 and Sp3 transcriptional regulatory proteins. Our data indicate that both Sp1 and Sp3 can mediate basal expression from both the TCE and THR but convey TGF-␤1-inducible expression only at the THR in the mouse Sm␣A enhancer. A rate-limiting role in the activation of the mouse Sm␣A enhancer is indicated for the TGF-␤1 receptor-regulated Smad proteins previously shown to govern transcription of TGF-␤1dependent genes associated with wound healing and extracellular matrix remodeling in fibroblasts.

MATERIALS AND METHODS
Cell Culture and Transfection Methods-AKR-2B embryonic fibroblasts were maintained in McCoy's 5A medium (Biowhittaker, Walkersville, MD) supplemented with 5% heat-inactivated fetal bovine serum (Invitrogen) and penicillin-streptomycin (20,26,27). Nonhuman primate COS7 kidney fibroblasts were maintained in Dulbecco's modified Eagle's medium (4.5 g/liter D-glucose) supplemented with penicillin-streptomycin and 10% FBS. Primary cultures of mouse embryonic fibroblasts derived from wild-type and Smad3 knockout mice were kindly provided by Dr. X.-F. Wang (Duke University, Durham, NC) and maintained in Dulbecco's modified Eagle's medium containing 10% FBS and penicillin-streptomycin. All fibroblast preparations were rendered quiescent by a 48-h exposure to HEPES-buffered Dulbecco's modified Eagle's medium (1.0 g/liter D-glucose), 0.5% FBS, and penicillin-streptomycin. Recombinant human TGF-␤1 (2-5 ng/ml, final concentration; R&D Systems, Minneapolis, MN) or an equivalent volume of vehicle (1 mg/ml bovine serum albumin, 4 mM HCl) was added to fibroblast cultures for varying periods before preparation of cell extracts for either reporter gene determinations or electrophoretic mobility shift assays. Mithramycin A (0.1-0.8 M, final concentration; Sigma) was used in some experiments as specified in the figure legends to specifically block Sp1 binding at GC-rich motifs within the Sm␣A enhancer (30). Plasmids used for transfections were purified using Qiagen TM preparative resin and a protocol provided by the manufacturer (Qiagen, Chatsworth, CA). In studies using AKR-2B and mouse embryonic fibroblasts, Sm␣A enhancer:CAT reporter gene fusion plasmids were combined with 5 g of either pSV-␤-galactosidase (Promega, Madison, WI) or pXGH5 (28) reporter gene constructs to normalize CAT expression for variation in transfection efficiency. pXGH5 encodes secreted human growth hormone under control of the constitutive metallothionein promoter. CAT, ␤-galactosidase, and human growth hormone assays were performed in triplicate using commercial enzyme-linked immunosorbent assay kits (Roche Molecular Biochemicals) in accordance with the manufacturer's instructions. For Sp and Smad protein overexpression studies using high transfection efficiency SL2 and COS7 cells (see below), Sm␣A:CAT fusion gene output was normalized to total protein rather than output from a secondary ␤-galactosidase or human growth hormone reporter gene. This was done because direct trans-regulation of the pSV-␤-galactosidase and pXGH5 promoters by the overexpressed proteins could compromise their role in monitoring transfection efficiency. For transfection, cells at 50% confluence in 6-well plates first were washed with serum-and antibiotic-free medium. Optimized mixtures of Sm␣A promoter:reporter fusion plasmids (pC3VSMP4 and pC3VSMP3) and plasmids encoding various transcriptional regulatory proteins (see below) were combined (total plasmid payloads were between 0.5 and 2.5 g) along with 15 l of LipofectAMINE TM reagent (Invitrogen) and incubated for 30 min at room temperature before applying them to cell monolayer for a 5-h period. After washing out unincorporated DNA, the cells were maintained in complete medium for an additional 36 -48 h before harvesting for reporter gene assays as described above. Transfections were routinely performed in triplicate, and each experiment was repeated three to five times. Mean values for normalized CAT activity (ϮS.E.) were evaluated by analysis of variance with statistical significance set at p Յ 0.05.
Schneider Drosophila 2 (SL2; kindly provided by Dr. M. Seeger; Ohio State University) cells were grown in Schneider's medium (Invitrogen) supplemented with penicillin, streptomycin, and 10% FBS. For transfection, replicate preparations of SL2 cells were plated at 50% confluence 1 day before DNA delivery and washed briefly with 2 ml of serum-free Schneider's medium just before transfection. For each transfection, 2 g of Sm␣A promoter:reporter fusion plasmid (pC3VSMP3 or pC3VSMP4) plus 4 g of Sp insect expression plasmid (empty pPAC vector, pPACSp1, and pPACSp3 driven by the Drosophila actin promoter and kindly provided by Dr. G. Suske; University of Marburg) were combined with 9 l of Cellfectin TM reagent (Invitrogen), incubated for 15 min at room temperature, diluted with serum-free medium, and then distributed to SL2 cells. After 20 h, the DNA-containing medium was removed, and the cells were washed briefly and incubated with complete growth medium for 48 h before harvesting extracts for CAT assays.
Mammalian Protein Overexpression Plasmids and Construction of VSM ␣-Actin Promoter Mutations-Recombinant forms of human Sp1 and Sp3 (kindly provided by Dr. J. Horowitz; North Carolina State University) and the human Smad proteins (kindly provided by Drs. L. Choy and R. Derynck; UCSF) were subcloned into mammalian expression plasmids under control of the cytomegalovirus promoter. Transversion mutants of the VSM ␣-actin promoter:CAT reporter fusion constructs pC3VSMP3 and pC3VSMP4 (abbreviated henceforth in this work as VSMP3 or SMP3 and VSMP4 or SMP4) were constructed using PCR amplification or site-directed mutagenesis using a commercial mutagenesis kit (Altered Sites TM ; Promega). Custom oligonucleotide primers harboring transversion mutations of nucleotides comprising regulatory elements in VSMP3 and VSMP4 were purchased from Integrated DNA Technologies, Inc. (Coralville, IA). Primers were annealed to 0.1-1 g of template DNA (VSMP4) and amplified for 30 -35 cycles in a PerkinElmer Life Sciences DNA Thermacycler TM using PCR kit (PerkinElmer Life Sciences) reagents and reaction times (94°C ϫ 1 min; 50°C ϫ 2 min; 72°C ϫ 3 min). PCR products were purified from 1-2% agarose gel slices using Spin-X TM columns (Costar, Cambridge, MA). The promoter construct harboring the MCAT mutation (MCATmut) was created by amplifying the VSMP4 insert with a forward primer containing a 5Ј SalI restriction site and substituting CTTCCGT for the native MCAT site between Ϫ182 and Ϫ176. The promoter construct harboring the THR mutation (THR-mut) was created by amplifying the VSMP4 insert with a forward primer containing a 5Ј SalI restriction site that substitutes CTTA for AGGC between Ϫ160 and Ϫ157. Mutation of the putative TGF-␤1 control element (TCE-mut) was made using a primer that substituted TT for GG at positions Ϫ50 and Ϫ49 in VSMP4. Promoters harboring the downstream element mutation (DE-mut) were created by amplification with a reverse primer containing a 3Ј BamH1 restriction site and substituting AATTTCA-CAACAA for CCGGGACACCACC between Ϫ2 and ϩ11 in both VSMP3 and VSMP4 context. Promoter constructs harboring StuI recognition site substitution mutations in place of CARG-like elements A, B, or D (A-mut, B-mut, and D-mut, respectively) were prepared as described previously (26). All sequence modifications were confirmed by doublestranded dideoxy sequencing.
Preparation of DNA-binding Protein Extracts and Electrophoretic Mobility Shift Assays (EMSAs)-EMSA reactions typically contained 5-10 g of nuclear protein extract, 1 g of poly(dI-dC), 10 mM Tris, pH 7.5, 50 mM NaCl, 0.5 mM dithiothreitol, 0.5 mM EDTA, 0.12 mM phenylmethylsulfonyl fluoride, 4% glycerol, and 20,000 cpm (5-50 fmol) of 32 P-labeled probe in a 10-l reaction volume. EMSA probes were constructed in native copy number context with the exception of the CARG D oligonucleotide, which contained a dimerized version of the native motif to improve binding affinity. Mutated versions of each EMSA probe contained nucleotide substitutions identical to those contained in the reporter gene counterparts described in the previous section. Oligonucleotide probes were labeled with Klenow enzyme (Invitrogen) and [␥-32 P]ATP (ICN Biomedical, Costa Mesa, CA), purified by electrophoresis on 8% polyacrylamide gels, eluted, and ethanol-precipitated before use. EMSA reaction mixtures were incubated for 30 min at room temperature before electrophoretic analysis on 5% nondenaturing polyacrylamide gels in 0.5ϫ TBE buffer (0.045 M Tris borate and 1 mM EDTA). For competition EMSAs, an excess of unlabeled oligonucleotide competitor was added to the reaction before adding protein extract or labeled probe. Antibody supershift EMSAs using commercial Sp1 and Sp3 rabbit polyclonal antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) were performed as described above but included a 20 min preincubation with 2 l of anti-Sp antibody before the addition of probe. Purified mouse IgG was used as a negative control in antibody supershift experiments.
Oligonucleotide Affinity Capture, Immunoblot, and Northern Blot Analysis-Reaction mixtures containing nuclear extract (100 g of pro-tein) and biotinylated oligonucleotides (100 pmol; Integrated DNA Technologies) corresponding to the Sp1 consensus site, native TCE site, or mutated TCE site (TCE-mut) were incubated under conditions that mimicked an EMSA reaction. Sp1/Sp3-biotin-DNA complexes were captured on streptavidin-immobilized paramagnetic particles (Promega; 0.6 ml/reaction, 30-min incubation). After washing four times with buffer containing 25 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 100 mM NaCl, bound protein was eluted using 2ϫ protein denaturation buffer and analyzed by Western blot (22). Eluted protein was processed by SDS-PAGE using 10% polyacrylamide gels and electrophoretically transferred to nitrocellulose membranes (Schleicher & Schü ll). After overnight blocking at 4°C in TBS (25 mM Tris-HCl, pH 7.5, 150 mM NaCl) containing 3% (w/v) nonfat dry milk, blots were incubated with anti-Sp1 or anti-Sp3 rabbit polyclonal antibodies for a 60 -90-min incubation at ambient temperature with gentle mixing. Goat anti-rabbit IgG-HRP (horseradish peroxidase) (Vector Laboratories, Burlingame, CA) diluted 1:2000 in TBST (TBS containing 0.05% Tween 20) was then applied for 30 -45 min. After washing in TBST, the blot was incubated with Vectastain TM reagent for 30 min. Blots were washed four times over a 30-min period and processed for chemiluminescence development for 1 min (ECL TM , reagents; PerkinElmer Life Sciences), and the immune complexes were visualized on x-ray film (X-Omat; Eastman Kodak) after a 5-10-s exposure. Both antibodies recognized multiple Sp1 and Sp3 isoforms. Northern blot analysis was performed using a Sm␣A-specific 3Ј(untranslated region) cDNA probe as described previously (15,53).

RESULTS
Basal transcription of the native Sm␣A gene in mouse AKR-2B stromal fibroblasts was minimally enhanced 6-fold up to as much as 20-fold in higher cell density preparations within 6 h of treatment with TGF-␤1 (Fig. 1a). To localize sequence elements responsible for TGF-␤1 activation, we performed deletion analysis of a 3.6-kb segment of the mouse Sm␣A promoter (VSMP8) that exhibits correct developmental regulation in transgenic mice. At the low cell density required for efficient transfection, VSMP8 transcription was increased ϳ4-fold by TGF-␤1, indicating that elements contained within the 5Јflanking/first intronic region mediated growth factor responsiveness ( Fig. 1, b and c). A subfragment of the 5Ј-flanking region, located between Ϫ191 and ϩ46 relative to the start of transcription (Fig. 1b), contains a potent core enhancer element previously shown to contain a MCAT motif required for highlevel constitutive expression and serum inducibility in stromal fibroblasts (21,26), aortic smooth muscle cells (25,29), and undifferentiated striated muscle myoblasts (21). The VSMP4 construct containing this enhancer was induced about 6-fold by TGF-␤1. Further 5Ј truncation by 41 bp (VSMP5) or 101 bp (VSMP6) completely eliminated basal and TGF-␤1-inducible transcriptional activity, indicating that the region of the mouse Sm␣A promoter between Ϫ191 and Ϫ150 was minimally required to mediate the TGF-␤1 response in fibroblasts (Fig. 1, b  and c).
Site-specific mutations were created within VSMP4 context to evaluate the relative importance of the known positive elements in mediating TGF-␤1 inducibility in fibroblasts. Altered sites included an inverted MCAT motif (AGGAATG, between Ϫ182 and Ϫ176) and CARG-like elements B and A (CCCTATATGG, between Ϫ120 and Ϫ111, and CCTTGTT-TGG, between Ϫ70 and Ϫ61) that bind the TEF1 and serum response factor transcriptional activating proteins, respectively. Both sites were required for basal expression of VSMP4 in mouse fibroblasts (24,26), neonatal rat aortic A7r5 smooth muscle cells (29), and undifferentiated mouse BC3H1 myoblasts (21). Also included in the analysis was a variant containing a 4-bp transversion (AGGC to CTTA) within a recently described 20-bp subdomain of VSMP4 that exhibited differential reactivity to DNA-modifying agents in the presence and absence of TGF-␤1 (15). This region, located between Ϫ170 and Ϫ150, is referred to as the THR. A construct containing a mutation within a putative TCE located between Ϫ53 and Ϫ43, previously identified in the rat Sm␣A promoter (16), was also included in the analysis. Fig. 2 shows that all mutations exhibited significantly reduced basal expression relative to native VSMP4. However, all VSMP4 mutants retained TGF-␤1 inducibility, except for the construct harboring a THR mutation. Whereas fold induction in the presence of TGF-␤1 (relative to that exhibited by native VSMP4) was highest for the construct lacking a functional MCAT element, mutations in the TCE, CARG A, or CARG B motifs all showed approximately the same fold increase. The slight induction exhibited by the THR mutation was not statistically significant (Fig. 2). Thus, with the exception of the THR, mutation of all other activating elements (MCAT, CARG B, CARG A, and TCE) did not fully eliminate the mouse Sm␣A enhancer responsiveness to TGF-␤1 in transfected fibroblasts.
EMSAs were performed using extracts prepared from TGF-␤1-treated fibroblasts to examine nuclear protein binding to Multiple trials revealed levels of induction as high as 20-fold in TGF-␤1-treated fibroblasts relative to quiescent cells. b, the mouse Sm␣A promoter (SMP8), previously shown to direct smooth muscle-specific expression in several adult transgenic mouse models. Multiple positive regulatory elements are located in the proximal 210-bp portion of SMP8, and deletion mutants (SMP4, SMP5, and SMP6) harbor different combinations of these elements. Normalized activity for each SMP construct in the presence (f) and absence (o) of 5 ng/ml TGF-␤1 is shown in c. Statistically significant basal expression and TGF-␤1 responsiveness (p Ͻ 0.05) in stromal fibroblasts was observed only for SMP4 and SMP8, which both contain the MCAT enhancer located between Ϫ191 and Ϫ150 of the 5Ј-flanking region. All transfections were performed in triplicate, and each experiment was repeated five times. the Sm␣A enhancer. No net increase in binding activity to the MCAT, CARG B, CARG A, and TCE probes was observed after treatment with TGF-␤1 (data not shown). Protein binding specificity was confirmed by competition with unlabeled excess DNA (see below). In the mouse Sm␣A enhancer, CARG B was previously shown to bind serum response factor (26,29), whereas the MCAT element binds TEF1 (20,26,27). Whereas the identity of the mouse fibroblast TCE-binding protein was not known, an identical binding activity was noted when extracts were assayed using a probe containing the rat Sm␣A promoter TCE element (data not shown). Because the mouse TCE (TGGGAGGGG) shares homology with GC-rich Sp1/3 transcription factor consensus binding sites (GGGGCGGGG or GGTGTGGGG), we compared Sp1/3 consensus and mouse Sm␣A TCE probes in both native and mutant context for the ability to bind fibroblast nuclear proteins in oligonucleotide affinity capture assays (Fig. 3). Western blot analysis of eluted proteins revealed that authentic Sp1 and Sp3 proteins bound to both probes equally well but not to the probe harboring a TCE mutation.
Interestingly, DNA-protein complexes with electrophoretic mobilities identical to the Sp1/Sp3-TCE complex were also observed when probes corresponding to other partially or substantially GC-rich regions of the mouse Sm␣A enhancer were used in gel shift analysis (Fig. 4). These additional probes (depicted in Fig. 1b) encompassed the following regions in the mouse Sm␣A core promoter: 1) a 13-bp DE located between Ϫ2 and ϩ11 in proximity to the transcription start site, 2) an 11-bp portion of the THR described above, located midway between the MCAT and CARG B elements (15), and 3) a 10-bp element previously referred to as CARG D located upstream to the MCAT motif and shown to be involved in transcriptional repression in undifferentiated myoblasts but functionally irrelevant in fibroblasts (21). Although only the TCE sequence strictly resembled an authentic Sp1/3 binding site, each of the four mouse Sm␣A enhancer elements in native, but not mutant, context specifically competed for protein complexes that bound to a Sp1/3 consensus oligonucleotide probe (Fig. 4). The authentic Sp1/3 consensus element was the most effective competitor, followed by the TCE, DE, THR, and CARG D sites. For all probes, mobility supershift analysis using polyclonal antibodies revealed that the slowest-migrating DNA-protein complexes were specifically eliminated/supershifted by the Sp1specific antibody, whereas the two more rapidly migrating complexes were eliminated/supershifted by the Sp3-specific antibody (Fig. 5). Taken together, the results demonstrated that both Sp1 and Sp3 were capable of binding at least four sites within or proximal to the mouse Sm␣A enhancer.
The functional effect of Sp1 and Sp3 overexpression on mouse Sm␣A enhancer activity was examined in both Drosophila SL2 and nonhuman primate COS7 cells. Whereas VSMP4 showed little baseline activity in SL2 cells, which naturally lack authentic Sp1/3 proteins, expression of either Sp1 or Sp3 stimulated VSMP4 transcription to a different extent, with co-expression of both Sp1 and Sp3 showing an intermediate level of activation (Fig. 6). For both SL2 and COS7 cells, Sp3 was a less efficient activator of VSMP4 transcription. Western blot analysis of transfected COS7 cells revealed that reduced efficiency of Sp3 enhancer activation was not due to reduced Sp3 protein synthesis relative to cells transfected with identical mass amounts of Sp1 expression plasmid (data not shown).
Whereas the THR and TCE Sp1/3 binding sites were both required for basal activity of the mouse Sm␣A promoter in fibroblasts (Fig. 2), the functional significance of the newly identified DE and CARG D Sp1 binding sites was unknown and therefore examined in cell transfections assays using reporter gene constructs containing site-specific mutations (Fig. 7). The DE site was fully contained within the 191-bp VSMP4 promoter, but the CARG D binding site was farther upstream and thus mutated in the context of a slightly larger 224-bp Sm␣A promoter construct (VSMP3) that is normally repressed in fibroblasts, undifferentiated myoblasts, and proliferating cultured aortic smooth muscle cells (29,21). The construct containing a THR mutation served as a negative control in this experiment and, as before (refer to Fig. 2), was inactive when compared with the native VSMP4 promoter (Fig. 7). In contrast, disruption of the DE Sp1/3 binding site had no significant effect on VSMP4 activity. Moreover, mutation of the CARG D Sp1/3 binding site in context of the repressed VSMP3 construct had no de-repression/activating effect in fibroblasts when present alone or in combination with a DE mutation. Taken together with the EMSA results, the data suggested that Sp1/3 proteins function as transcriptional activators in fibroblasts, most likely through their essential interaction with the TCE and THR sites.
TGF-␤1 appeared to amplify Sm␣A enhancer output in fibroblasts in the absence of a net increase in Sp1/3 binding to the THR, suggesting that another regulatory protein may be ratelimiting for this response. In this regard, Sp1/3 has been reported by several investigators to mediate transcriptional response to TGF-␤1 through both direct and indirect association with members of the Smad family of TGF-␤1 receptor-regulated proteins (31)(32)(33)(34). Western blot analysis revealed that Smad2 and Smad3 co-activating proteins were rapidly translocated to the nuclear compartment of fibroblasts within 30 min of exposure to TGF-␤1 (Fig. 8). Suggestive of Smad dependence, VSMP4 was induced by TGF-␤1 in transfected normal mouse embryonic fibroblasts but failed to respond in transfected embryonic fibroblasts derived from Smad3-deficient mice (Fig. 8). Either Smad2 or Smad3 was capable of transactivating VSMP4 in transfected COS7 cells, but only when an expression plasmid encoding the collaborating Smad4 protein was included (Fig. 9). Interestingly, co-expression of Smad2/3/4 and Sp1/3 proteins in COS7 cells did not increase VSMP4 output beyond that observed when Smad or Sp1/3 proteins were overexpressed individually. These observations suggest that either Sp1/3 or Smad proteins may be equally capable of activating the Sm␣A enhancer via a mechanism not based on transcriptional synergy. This idea was further supported by experiments using the Sp1-selective transcriptional inhibitor, mithramycin A. Mithramycin A is an intercalating agent that preferentially binds GC-rich sequences in both native and synthetic oligonucleotide contexts (30). Basal expression of Sm␣A mRNA was reduced in mithramycin A-treated fibroblasts, yet in three independent Northern blot studies, typical levels of TGF-␤1 inducibility were still fully preserved (compare LSϩM and TGFϩM in Fig. 10 showing 6-fold induction). Allowing for more rigorous statistical evaluation, studies using COS7 cells transfected with the Sm␣A enhancer and Smad protein expression plasmids revealed that enhancer output in the presence of Smad2/3/4 was reduced by about 55% (p Ͻ 0.01) in the presence of mithramycin A (Fig. 11, compare SMP4/Smads and SMP4/ Smads/MA). However, this residual expression still represented a nearly 17-fold increase (p Ͻ 0.006) over baseline expression in cell lacking exogenous Smad proteins (Fig. 11, compare SMP4/MA and SMP4/Smads/MA). Taken together, the results were consistent with a model for Sm␣A enhancer activation that includes both Sp1-dependent (mithramycin Asensitive) and Sp1-independent/Smad-mediated (mithramycin A-resistant) components. DISCUSSION TGF-␤1 transiently induces expression of Sm␣A in stromal fibroblasts and therefore may represent an important mediator of myofibroblast differentiation (15) and short-term contractile function during the wound-healing process (35). Tightly regulated activation of the Sm␣A gene and consequent assembly of smooth muscle actin stress fibers are likely to be crucial features of myofibroblast differentiation that contribute to the generation of contractile force, rapid wound closure, and effi- cient tissue healing. Tissue fibrosis and scar formation may result if myofibroblast differentiation is poorly controlled as in chronic inflammatory diseases such as liver cirrhosis, scleroderma, cardiac allograft dysfunction, and pulmonary fibrosis (2,3,14,36). In this regard, we have identified transcriptional regulatory elements and cognate repressor protein complexes that may be important not only in preventing unscheduled activation of this gene in quiescent stromal fibroblasts but also in limiting the magnitude and duration of Sm␣A transcription in injury-activated myofibroblasts (15,20,22,27,29).
Deletion mutation analysis of the smooth muscle-specific VSMP8 promoter construct revealed that an amplified transcriptional response to TGF-␤1 in AKR-2B fibroblasts minimally required sequence elements located within the first 191 bp of the 5Ј-flanking region generally referred to as the MCAT enhancer. Basal activity required multiple positive elements within the MCAT enhancer that were also required for transcriptional amplification after exposure of stromal fibroblasts to TGF-␤1. A recent analysis of native mouse fibroblast chromatin encompassing the Sm␣A enhancer region (15) indicated However, Smads appear to act independently from the Sp1 and Sp3 proteins (right panel) because combined overexpression of all three Smads plus either the Sp1 or Sp3 proteins (SMP4/Sp1/Smads and SMP4/Sp3/Smads) did not activate SMP4 to levels greater than that observed with individual Sp proteins (SMP4/Sp1 and SMP4/Sp3) or Smad2/3/4 (SMP4/Smads) alone. pCMV, empty expression vector; Smads, combination of the Smad2/3/4 expression plasmids. Transfections were performed in triplicate, and each experiment was repeated five times.

FIG. 10. Effect of mithramycin A on actin mRNA expression in fibroblasts.
Mithramycin A specifically intercalates at GC-rich Sp1/3 DNA-binding sites and had a marked inhibitory effect on basal expression of Sm␣A mRNA, but not TGF-␤1 inducibility per se. There were no discernable effects of mithramycin A on the level of nonmuscle ␤,␥-actin or glyceraldehyde-3-phosphate dehydrogenase mRNA expression. LS, cells receiving 0.5% FBS alone; TGF, cells receiving 0.5% FBS and then exposed to 5 ng/ml TGF-␤1 for 12 h; M, cells receiving a 6-h dose of mithramycin A (0.8 M) in 0.5% FBS alone or before the 12-h TGF-␤1 treatment. The Northern blot depicted is representative of three independent experiments that essentially showed the same trends.
FIG. 11. The effect of mithramycin A on Smad protein-dependent Sm␣A enhancer activity in COS7 cells. COS7 cells were transfected with either SMP4 alone (SMP4) or SMP4 plus Smad2/3/4 (SMP4/Smads) and subsequently exposed to mithramycin A (0.8 M) for 24 h before cell extraction and CAT assays. The data are presented as the normalized means (units of CAT activity/g protein) Ϯ S.E. Only 55% of the total Smad-dependent enhancer output was inhibited by mithramycin A (compare SMP4/Smads and SMP4/Smads/MA, p Ͻ 0.006). Transfections were performed in triplicate, and each trial was repeated five times. that TGF-␤1 markedly affected chemical reactivity of nucleotides located between Ϫ170 and Ϫ150 that fully encompassed the THR identified by functional cell transfection studies reported in this current work. Sm␣A promoter constructs harboring mutations in five sequence elements (MCAT, THR, CARG A, CARG B, and TCE) all exhibited impaired basal transcriptional activity in mouse fibroblasts, yet only the THR was required for an amplified response to TGF-␤1. Mechanisms governing TGF-␤1 inducibility of the 191-bp mouse Sm␣A enhancer in fibroblasts apparently differ from those in rat aortic smooth muscle cells, bovine aortic endothelial cells, and rat fibroblasts, where elements required for basal expression of a 125-bp segment of the rat Sm␣A promoter were shown to be necessary and sufficient for TGF-␤1 inducibility (16,17). Moreover, those investigators noted increases in the amount and/or DNA binding activity of nuclear protein factors that bound to transcriptional activating elements in the rat Sm␣A core enhancer (CARG A, CARG B, and TCE) after exposure of quiescent rat aortic smooth muscle cells to TGF-␤1. However, we noted that although TGF-␤1-amplified output from mutant Sm␣A enhancers in mouse AKR-2B fibroblasts was markedly diminished, each mutation nonetheless retained significant inducibility. These observations thus preclude strict classification of the mouse Sm␣A CARG A, CARG B, and TCE sites as TGF-␤1 response elements, although each was required for maximum enhancer activity in mouse fibroblasts. This interpretation is more than a semantic argument because transcriptional activity from each mutant enhancer was quite substantial in TGF-␤1-treated fibroblasts and comparable with the basal activity elicited from the native VSMP4 enhancer in the absence of growth factor. The observation that net protein binding to known Sm␣A enhancer elements was not affected by TGF-␤1 further suggests that none of the known enhancerbinding proteins are rate-limiting. Rather TGF-␤1 amplification of the mouse VSM ␣-actin promoter in fibroblasts appears to require participation of all sequence elements that normally contribute to basal transcriptional activity. TGF-␤1 may stabilize higher order protein complexes between these multiple Sm␣A enhancer sites or permit unique protein-DNA conformational arrangements that do not depend on increased stoichiometric binding of any one individual class of enhancer-binding protein.
Differences in TGF-␤1 responsiveness between the mouse and rat Sm␣A enhancers may be further explained by the existence of additional positive regulatory elements in the mouse promoter located between Ϫ191 and Ϫ125 (THR and MCAT) that, as shown here and elsewhere, are required for high level constitutive expression of the mouse Sm␣A core promoter in both muscle and nonmuscle cells from the mouse (21) but appear to be dispensable in the rat gene (16,17). Shorter mouse Sm␣A promoter constructs consisting of sequence elements between Ϫ143 and Ϫ1 were inactive in mouse muscle cells and fibroblasts as well as primary rabbit (25) and A7r5 neonatal rat (29) aortic smooth muscle cells. Positive regulatory elements between Ϫ191 and Ϫ143 may facilitate the formation of more complex DNA structures or assembly of novel DNA-protein complexes that could diminish the importance of any one site while enhancing cooperative interactions that maximize the transcriptional response to TGF-␤1. Currently, we are analyzing dynamic structural changes within the closely positioned transcriptional silencing and activating sequences located in the Sm␣A enhancer as well as evaluating the higher order composition of enhancer-protein complexes that form in this region.
Basal and TGF-␤1-inducible promoter activity necessarily relies on transcription factor availability within the cell nu-cleus. Mouse AKR-2B fibroblasts do not appear to contain novel, potentially rate-limiting transcription factors that were reported to bind the TCE site in rat aortic smooth muscle cells (16,17). However, given the high homology of the GC-rich TCE motif to binding sites for the ubiquitous Sp1/Sp3 DNA-binding proteins, we examined whether these proteins could contribute to basal Sm␣A transcriptional activity in mouse fibroblasts, even though they do not appear to mediate promoter activity in the rat. The Sp1/Sp3 proteins are well-characterized members of a multigene family (37)(38)(39) that bind to widely distributed GC-rich promoter elements such as the GC box (GGGGCGG-GG) and the related GT/CACCC box (GGTGTGGGG). We analyzed protein factors that bound to four GC semi-rich elements contained within truncated portions of the VSM ␣-actin promoter that were either constitutively active (VSMP4) or fully repressed (VSMP3) in mouse fibroblasts. Whereas all four sites bound Sp1/Sp3, mutation of only two of these sites (THR and TCE) had functional consequences in transfection assays, and only mutations in the THR significantly affected TGF-␤1 inducibility of the Sm␣A enhancer. Whereas Sp1 has been shown to function exclusively as a transcriptional activator, Sp3 can either activate or repress transcription, depending on the promoter context (40 -45). In general, Sp3 behaves as an activator when targeted to a single binding site within a promoter but acts as a repressor when bound to multiple sites within a promoter (46), although exceptions to this scheme have been reported (45). One such exception may pertain to stromal fibroblasts because we showed that either Sp1 or Sp3 alone can activate Sm␣A enhancer constructs containing multiple Sp1/Sp3 binding sites when overexpressed in Sp factordeficient Drosophila SL2 cells. When expressed in combination using equimolar concentrations of plasmid DNA, activation was intermediate to that observed in the presence of equivalent amounts of Sp1 or Sp3 alone, implying that their effects are simply additive rather than antagonistic or synergistic. However, our inability to detect a change in the level of Sp1/Sp3 binding, or any other activator for that matter, at any one site in the mouse Sm␣A enhancer after exposure of fibroblasts to TGF-␤1 leads us to speculate that growth factor activation of this gene may require more than individual protein loading events at functionally insulated cis-acting elements. For example, Li et al. (43) reported that phosphorylation of the B activation domains of Sp1 rather than overt changes in Sp1 binding activity to DNA may be an important molecular feature of TGF-␤1-activated transcription.
Rate-limiting co-regulatory factors may enhance dynamic interactions between individual transcriptional activating proteins that bind to closely positioned sites within the Sm␣A enhancer. Relevant to control of the Sm␣A enhancer response to TGF-␤1 are members of the Smad family of co-regulatory proteins (47)(48)(49)(50), which are known to modulate transcription through direct or indirect interaction with Sp1 (31)(32)(33)(34). Results presented here show that both Smad2 and Smad3 rapidly accumulated in the fibroblast nucleus within 30 min after exposure to TGF-␤1. Consistent with a possible rate-limiting role, TGF-␤1 treatment failed to up-regulate the Sm␣A enhancer in transfected embryonic fibroblasts prepared from Smad3 knockout mice, and combined overexpression of all three Smad proteins in COS7 cells substantially activated the Sm␣A enhancer. However, unlike the results reported for the p15 Ink4B (32) and ␣2(I)collagen promoters (33,34), co-expression experiments that included all the Smad proteins plus either Sp1 or Sp3 showed no further Sm␣A enhancer transactivation than that obtained when either Smads or Sp proteins were expressed alone. One idea emerging from these results and currently under investigation is that Smad proteins di-rectly activate the Sm␣A enhancer by a mithramycin-resistant mechanism that does not require physical interaction of Sp1 or Sp3. In this regard, all activation sites in the Sm␣A enhancer may be physically occupied by latent transcriptional regulatory proteins/repressors awaiting coordinate activation/neutralization by TGF-␤1 during a wound-healing event. Within the vicinity of the MCAT and THR motifs, the single-strand-specific DNA-binding proteins MSY1 and Pur ␣/␤ are believed to block access of the TEF1 activating protein to the VSM ␣-actin enhancer via a conformational switch that disrupts duplex DNA structure. In vivo evidence for this switch (15) showed that TGF-␤1 promoted accumulation of both hyper-reactive and protected DNA, particularly in regions encompassing both the TEF1 and Sp1/Sp3 binding sites (the MCAT and THR sites, respectively). TGF-␤1 acting through the rate-limiting Smad pathway does not appear to enhance binding of positive regulatory proteins to probes spanning this conformational switch. However, it remains a distinct possibility that changes in chromatin conformation are associated not with a gain in activating factor binding at these sites but rather with depletion of the single-strand-specific transcriptional repressors Pur ␣/␤ and MSY1 from the enhancer. Relief of Sm␣A transcriptional repression through net loss of Pur and MSY1 repressor proteins from their corresponding single-stranded enhancer binding sites could be achieved via higher order interactions between activator proteins that permanently reside at the MCAT, THR, CARG, and/or TCE sites. The Smad proteins, which may possess intrinsic kinase activity, could facilitate these hypothetical protein-protein or protein-DNA interactions within the enhancer (32). In this regard, there are two Smad protein-binding CAGA motifs embedded in the TGF-␤1 hyper-reactivity region of the enhancer between Ϫ170 and Ϫ150. Identification of higher order complexes containing multiple transcriptional activators such as TEF1, Sp1/3, serum response factor, and Smad proteins may not be captured by the relatively short oligonucleotide probes used in our analysis and will require further analysis using longer DNA probes, antibody supershift EMSAs, competition EMSAs, and protein immunoprecipitations. Recently published studies from our group have shown that TEF1-serum response factor complexes can in fact be observed in extracts prepared from AKR-2B fibroblasts and aortic smooth muscle cells, lending support to the notion of cooperative interaction between multiple enhancer-activating proteins (29). Moreover, Sp proteins have previously been shown to interact with and affect the functional properties of both Pur ␣/␤ and MSY1 (51,52). In myofibroblasts, coordinated activation of the Sm␣A gene from multiple protein binding sites in the enhancer may provide for rapid accumulation of Sm␣A monomers that are uniquely suited for the formation of rigid stress fibers and generation of tensile forces required for efficient repair of tissue injuries (6,35).
In summary, we have shown that at least five cis-acting elements along with their cognate DNA-binding proteins together are required for full basal and TGF-␤1-inducible expression of the Sm␣A enhancer in mouse stromal fibroblasts. Two of these sites, the THR and TCE, are shown here for the first time to bind the transcription factors Sp1 and Sp3. TGF-␤1 induction of the enhancer is also Smad protein-dependent, which could have implications regarding the structural arrangement and cooperativity of multiple regulatory proteins that bind near the THR site required for TGF-␤1 amplification. Future studies of Sm␣A gene expression in mouse fibroblasts will focus on identifying individual components of the transcriptional activation complex that drives enhanced expression in the presence of TGF-␤1. These studies, in turn, may provide useful insight into the molecular mechanisms that control myofibroblast function in both normal and pathological settings.