Substitution of the degenerate smooth muscle (SM) alpha-actin CC(A/T-rich)6GG elements with c-fos serum response elements results in increased basal expression but relaxed SM cell specificity and reduced angiotensin II inducibility.

We have previously demonstrated that both CC(A/T-rich)6GG (CArG) elements A and B of the smooth muscle (SM) alpha-actin promoter are required for smooth muscle cell (SMC)-specific expression and angiotensin II (AII)-induced stimulation. Moreover, results provided evidence that AII responsiveness of SM alpha-actin was at least partially dependent on modulation of serum response factor (SRF) binding to the SM alpha-actin CArGs by the homeodomain containing protein, MHox. The goal of the present study was to investigate whether the degeneracy of the SM alpha-actin CArGs (both contain a Gua or Cyt substitution in their A/T-rich center) and their reduced SRF binding activity as compared with c-fos serum response element (SRE) is important for conferring cell type-specific expression and AII responsiveness. Transient transfection assays using SM alpha-actin reporter gene constructs in which the endogenous SM alpha-actin CArGs were replaced by c-fos SREs demonstrated the following: 1) relaxation of cell-specific expression, 2) a 50% reduction in AII responsiveness, and 3) reduced ability to be transactivated by MHox. In addition, we also showed that the position of the SM alpha-actin CArGs was important in that interchanging them abolished both basal and AII-induced activities. Taken together, these results suggest that the reduced SRF binding activities of the SM alpha-actin CArGs and CArG positional context contribute to SMC-specific expression of SM alpha-actin as well as maximal AII responsiveness.

The CArG 1 motif, characterized by the consensus sequence CC(A/T-rich) 6 GG, is found in the promoters of several immedi-ate-early response genes (1)(2)(3)(4)(5) including c-fos and has been shown to confer serum-and growth factor-induced transcriptional activation of these genes (reviewed in Ref. 6). CArG boxes are also present in the promoters of many skeletal and cardiac muscle-specific genes and are required for developmental and tissue-specific expression (Refs. 7-15, reviewed in Ref. 16). Although CArG elements bind the ubiquitously expressed transcription factor SRF, it is unlikely that SRF alone is sufficient to confer the functional diversity of CArG elements. A large body of evidence has accumulated suggesting that CArGdependent gene regulation is modulated by post-translational modification of SRF (17), interaction of SRF with SRF accessory proteins (reviewed in Refs. 6 and 18), and combinatorial interaction with other trans-factors in a promoter-specific fashion (15,19,20). For example, Sartorelli et al. (21) demonstrated that muscle-specific expression of cardiac ␣-actin required the presence of CArG boxes, binding sites for SP 1, and the musclespecific factor MyoD. Previous results from our laboratory also provided evidence for involvement of the SM ␣-actin CArGs in cell type-specific expression of SM ␣-actin in concert with other regulatory elements (20). Moreover, CArG elements have also been shown to play a role in regulation of virtually all SMC differentiation marker genes characterized to date including SM-22␣ (22,23), SM MHC (24,25), telokin (26), and h-caldesmon (27).
In contrast to the c-fos gene which contains a single high affinity binding site for SRF in its promoter, many musclespecific genes including skeletal, cardiac, SM ␣-actin, SM MHC, and SM-22␣ contain two or more CArG elements (7,20,24,28,29). Based on direct measurements of SRF binding (30,31) and/or predicted SRF binding affinity based on DNA sequence analysis (see Leung and Miyamoto (32) for criteria) many of these CArG elements bind SRF with relatively low affinity as compared with the c-fos SRE. For example, both SM ␣-actin CArGs A and B, the two distal skeletal ␣-actin SREs (31), human cardiac ␣-actin CArG 2, 3, and 4 (28), as well as the SM MHC CArG 2 (24) contain Gua or Cyt substitutions within their central A/T-rich region that reduces SRF binding activity (32). In addition, the most proximal cardiac ␣-actin CArG, although not containing a Gua or Cyt substitution, was shown to bind SRF less effectively than the c-fos SRE (30). The preceding observations raise the question as to why relatively low affinity CArG elements have been highly conserved during evolution, especially in the light of strong evidence indicating that increased transcriptional activity of a number of CArGdependent genes is associated with increased SRF binding activity (33)(34)(35)(36). One possible hypothesis is that weaker CArG elements might offer an additional level of control through mechanisms that influence SRF binding. In contrast, strong SRF binding CArGs appear to be regulated primarily at post-SRF-CArG binding steps through interaction with SRF accessory proteins whose activity is controlled by kinase/phosphatase regulatory systems (reviewed in Ref. 37).
A number of mechanisms have been shown to increase SRF binding to CArG elements including post-translational modification of SRF (17), increased SRF protein expression (34), and interaction of SRF with homeodomain factors that modulate SRF binding or kinetics (38). For example, Croissant et al. (34) have shown that increased SRF protein levels appeared to be obligatory for increased skeletal ␣-actin expression during myoblast differentiation. In addition, we have previously provided evidence that TGF-␤ and AII-induced increases of SM ␣-actin expression in SMC were accompanied by increased SRF binding to CArG A and B (39,40). Whereas TGF-␤ increased SRF protein expression, AII treatment did not. Our work suggested that the AII effects on SM ␣-actin transcription were mediated, at least in part, by modulation of SRF binding to CArG B by the homeodomain containing protein MHox, which has been shown to increase SRF binding to CArG B in vitro (40). Walsh and co-workers (41) have carried out extensive and eloquent studies of the importance of both the CArG central A/T region and flanking sequences of the skeletal ␣-actin muscle regulatory element for muscle-specific expression. For example, the authors demonstrated that replacement of the muscle regulatory element by a c-fos SRE resulted in loss of musclespecific expression of skeletal ␣-actin. However, to our knowledge, no studies have specifically addressed the importance of weakly binding degenerate CArG elements per se in regulation of muscle-specific genes nor have any studies addressed the importance of such elements in control of SMCspecific and agonist-induced transcriptional regulation.
The aim of the present study was to determine whether sequence degeneracy of the SM ␣-actin CArG elements and their reduced SRF binding activity contributes to cell-specific SM ␣-actin expression as well as the ability of the gene to be regulated in response to AII or the mesodermally restricted homeodomain containing protein MHox.

Construction of Promoter-CAT Expression
Plasmids-The generation of various SM ␣-actin promoter CAT constructs, including the CArG A and B mutants, have been described previously (20). Additional CArG mutations within a 155-bp or 2.8-kb SM ␣-actin promoter CAT construct (pProm CAT) were generated using the Ex-site mutagenesis kit according to the manufacturer's instructions (Stratagene). The integrity and accuracy of the mutated constructs were determined by dideoxy sequencing (42).
All promoter-CAT plasmid DNAs used for transfections were prepared using an alkaline lysis procedure (43) followed by banding on two successive ethidium bromide cesium chloride gradients. Transfection results were not altered when independent plasmid preparations were tested.
Cell Culture, Transient Transfections, and Reporter Gene Assays-SMC from rat thoracic aorta and bovine endothelial cells (BAEC) were isolated and cultured as described previously (20,44). The culture conditions for the rat L6 skeletal myoblast were also described previously (20), and fusion into myotubes was induced by reducing fetal bovine serum (FBS) concentrations to 1% when cells reached confluency. AKR-2B mouse embryonic fibroblasts were a gift of Dr. Harold Moses (Vanderbilt University, Nashville, TN) and were cultured in McCoy's 5A medium (Life Technologies, Inc.) supplemented with 5% FBS, 0.68 mM L-glutamine (Sigma), 100 units/ml penicillin, and 100 g/ml streptomycin (Life Technologies, Inc.). SMC (passages 20 -30), L6 skeletal myoblasts, and AKR-2B fibroblasts were seeded for transient transfection assays into 6-well plates at a density of 1.5 ϫ 10 4 cells/cm 2 and BAEC at a density of 2 ϫ 10 4 cells/cm 2 . Transfection of the CAT reporter gene constructs (4 g of DNA per well) was performed in triplicate 30 h (in case of L6 myoblasts, 48 h) after plating using the transfection reagent DOTAP (Boehringer Mannheim) (6.7 l/g DNA) according to the manufacturer's recommendations. BAEC were trans-fected using the transfection reagent Transfectam (Promega) according to the manufacturer's instructions since transfection efficiency in these cells was lower with DOTAP. 2 No differences in transfection efficiencies were observed between Transfectam and DOTAP in other cell types. Cells were exposed to the DNA/DOTAP or DNA/Transfectam mixture for 5 h under serum-free conditions. After incubation, the transfection medium was replaced by serum containing medium, and the cells were harvested for the reporter assay 72 h later by scraping into ice-cold buffer A (15 mM Tris (pH 8.0), 60 mM KCl, 15 mM NaCl, 2 mM EDTA, 0.15 mM spermine tetrahydrochloride, 1 mM dithiothreitol) (45). Cell lysates were prepared by four freeze-thaws, followed by 10 min heat inactivation at 65°C; 95-l aliquots of each cell extract were assayed for CAT activity by enzymatic butyrylation of tritiated chloramphenicol (NEN Life Science Products) (46). CAT activities were normalized to that of a control promoterless construct set to one as described previously (20). This permits comparison of the activity of the wild-type p155 region versus that of the various mutants including disruption of the CArGs or substitution with the c-fos SRE. Normalization of CAT activity to that of SM ␣-actin constructs containing mutations of CArG A and B was not performed because CArG mutations have differential effects in SMC versus non-SMC, and such normalization would preclude comparison of the cell-specific functionality of the SM ␣-actin CArGs versus the c-fos SRE, a major aim of the present studies. Experiments were repeated two to three times, and relative CAT activity data were expressed as the mean Ϯ S.D. unless otherwise noted.
SMC used for transfection experiments involving AII stimulation were plated at a density of 3 ϫ 10 3 /cm 2 , grown to confluency in 10% serum containing medium, and then growth-arrested for 4 days in serum-free medium (SFM) (47) prior to stimulation with AII (Peninsula Laboratories, 10 Ϫ6 M) or SFM. Cells used for these experiments were between the 6th and the 12th passage. SMC that are growth-arrested in this fashion express multiple SMC differentiation marker proteins including SM ␣-actin, SM MHC, h-caldesmon, h 1 calponin, SM tropomyosin, and SM myosin light chain (MLC 20 ) (48 -50). 3 Confluent, growtharrested SMC were then transiently transfected (in triplicate in 6-well plates) with 5 g of DNA using the transfection reagent DOTAP (Boehringer Mannheim) according to the manufacturer's recommendations. After an incubation period of 12-14 h, the medium was replaced with fresh serum-free medium, and AII (10 Ϫ6 M) or vehicle were added. Cells were harvested 72 h later and processed for the reporter assay as described above.
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assays (EMSA)-Crude nuclear extracts from SMC were prepared by the method of Dignam et al. (51). SMC were either grown to confluency in 10% FBS or growth-arrested in SFM for 4 days when treated with AII. SMC were then exposed to AII (10 Ϫ6 M) or SFM for 4 h. Protein concentrations were measured by the Bradford assay (Bio-Rad). Probes for EMSA were obtained by end-labeling 20 M single-stranded oligonucleotides with 150 Ci of [␥-32 P]ATP (6000 Ci/mmol) and T4 polynucleotide kinase. Labeled single-stranded oligonucleotides were annealed, and unincorporated nucleotides were removed using Nuc Trap Push columns (Stratagene) as recommended by the manufacturer. The specific activity for all probes used was 1.0 -1.1 Ci/pmol. The following nucleotides, purchased commercially (Operon Technologies, Inc.), were used as a probe (only sense strand shown): CArG B, 5Ј GAGGTC-CCTATATGGTTGTG 3Ј; CArG A, 5Ј TTGCTCCTTGTTTGGGAAGC 3Ј; c-fos SRE, 5Ј GATGTCCATATTAGGACATC 3Ј.
EMSA were performed with 20-l binding reactions that contained ϳ50 pg of 32 P-labeled annealed oligonucleotides, nuclear extracts (3 or 5 g in Dignam buffer D), human recombinant SRF (1 or 2 l), 100 mM KCl, 5 mM HEPES (pH 7.9), 1 mM EDTA, 35 mM Tris (pH 7.5), 1.125 mM dithiothreitol, 10% glycerol, and 0.125 g of poly(dI-dC) as a nonspecific competitor. Specific antibodies against SRF (Santa Cruz, 2 g/reaction) were added when indicated. The binding reaction was incubated for 20 min at room temperature before radiolabeled probe was added, followed by another 20 min at room temperature incubation. All binding reactions were loaded on a 4.5% polyacrylamide gel and electrophoresed at 170 V in 0.5ϫ TBE. The gels were dried and subjected to autoradiography at Ϫ70°C.
In Vitro Synthesis of SRF-In vitro synthesis of SRF was performed using a TNT®-coupled reticulocyte lysate translation system (Promega) with the human SRF cDNA clone p T7 AE ATG (52) as a template.

The SM ␣-Actin CArG Elements A And B Showed Reduced SRF Binding as Compared with the c-fos SRE-There is extensive evidence showing that the internal A/T-rich center of
CArG elements affects SRF binding affinities (24,31,32,53). Leung and Miyamoto (32) demonstrated that substitution of A/Ts in the core of the c-fos SRE by guanidines or cytosines resulted in marked reduction of SRF binding affinity, especially when the substitution was made in the middle of the A/T-rich center. In contrast to the c-fos SRE, the A/T-rich center of CArG B of the SM ␣-actin promoter contains a cytosine substitution (see Fig. 1), and CArG A contains a guanidine substitution in the middle of the A/T-rich center (see Fig. 1). Thus, we would predict that binding activity of the SM ␣-actin CArGs for SRF would be reduced as compared with the c-fos SRE. To test this directly, recombinant, in vitro translated SRF was incubated at different concentrations with radiolabeled CArG A, B, and c-fos SRE oligonucleotides. EMSA demonstrated that SRF binding activity was lowest with the CArG A probe, intermediate with the CArG B probe, and highest with the c-fos SRE probe ( Fig. 2A, lanes 1-6). To determine whether the pattern of SRF binding activities to the different CArG elements was similar with SMC nuclear extracts as compared with those observed with recombinant SRF, we performed gel shift assays using nuclear extracts derived from SMC grown in 10% serum. Similar to the results obtained with recombinant SRF, SRF binding activity derived from SMC was lowest with CArG A, intermediate with CArG B, and highest with the c-fos SRE (Fig. 2B, lanes 1-3). These identical binding patterns suggest that SRF obtained from the SMC extract was not modified in such a manner as to significantly affect binding activity to each of these three probes and that SRF binding activity was not altered by any additional SMC proteins present in the extract, at least under the conditions of our gel shift assays.
Transcriptional Activity of SM ␣-Actin Varied as a Function of SRF Binding Activities of Its CArG Motifs-Previous studies including our own (33-36, 39, 40) have shown a correlation between increased SRF binding activity and transcriptional activity of a number of CArG-dependent genes. Thus, replacement of the SM ␣-actin CArGs by stronger or weaker SRF binding sites should alter SM ␣-actin transcription correspond-ingly. To test this, we generated various combinations of "strong CArGs" (c-fos SRE), "intermediate CArGs" (CArG B), and "weak CArGs" (CArG A) within a 155-bp SM ␣-actin promoter context by changing the internal CArG sequences as depicted in Fig. 1. Since the focus of the present study was to investigate the effects of SRF binding activities on transcriptional regulation of SM ␣-actin, we did not change the flanking sequences. Results of transient transfection assays, carried out in SMC growing in 10% serum, showed that replacement of either CArG A or B, or both with a c-fos SRE, resulted in increased SM ␣-actin transcription as compared with a wildtype p155 CAT construct (2-3.5-fold) with maximal increases seen with the SRE-SRE construct (Fig. 3). Due to the high activity of c-fos SRE-containing constructs, special care was taken to ensure that the CAT assay was performed in the linear range of the assay. In contrast, when two CArG A elements were present, almost all activity was lost. These results indicate that SM ␣-actin transcription varies as a function of the relative SRF binding activities of its CArG elements.
Positioning of CArG Elements Was Critical for SM ␣-Actin Transcriptional Activity-Previous studies study by Lee et al. (54) provided evidence that the three skeletal ␣-actin CArGs are bound by SRF in a specific order as determined by their relative SRF binding affinities and that the position of the individual CArGs to each other plays an important role in formation of transcriptionally active complexes. To test whether CArG positioning was also critical for activation of the SM ␣-actin promoter, we tested three sets of paired constructs in which the SRF binding elements (c-fos SRE, CArGs A, and B) were replaced at position 1 (proximal) and position 2 (distal) in reversed combinations (see Fig. 1). As shown in Fig. 4, differences in activities between these constructs clearly demonstrate that positions 1 and 2 are not functional equivalents of each other. In all instances, when the lesser affinity CArG was located at position 2 activity was decreased. Particularly striking was the significant decrease in activity when CArG A and B were switched (pBL 155 CAT A-B). This loss was not due to a negative effect of placing CArG B at position 1 since pBL 155 CAT A-A also had essentially the same low activity (see Fig. 3). No significant changes in transcriptional activity were observed when the positions of the two CArG elements with relatively high SRF binding capacities (CArG B and c-fos SRE) were interchanged. These results demonstrate that positioning of the CArG elements is critical in the regulation of the SM ␣-actin gene and that maximal levels of activity appeared to require the presence of an intermediate or high affinity SRF binding motif at position 2.
Polyclonal SRF antibodies (Santa Cruz) raised against the COOH terminus of human SRF was added at a concentration of 2 g/reaction (lanes 7 and 9). Unprogrammed lysate (UP) was incubated with a CArG B probe (lane 10). B, nuclear extracts (5 g) from rat SMC growing in 10% FBS were incubated with radiolabeled CArG A, B, and c-fos SRE oligonucleotides (lanes 1-3). An SRF antibody (Santa Cruz) was added to the binding reaction in lanes 4 -6.  1 and 2), CArG B (lanes 3 and 4), and c-fos SRE (lanes 5 and 6) double-stranded oligonucleotides were incubated with 1 or 2 l of in vitro translated human SRF (rec.SRF) for 20 min at room temperature.

Replacement of the SM ␣-Actin CArGs by c-fos SREs Resulted
in Relaxation of Cell Specificity-Previously, we have demonstrated that the SM ␣-actin CArG elements A and B contribute to cell-specific regulation of this gene (20) in that mutation of either CArG A or B completely abolished activity of a p125 bp SM ␣-actin promoter CAT construct in SMC but not in BAEC. These results suggested that factors other than SRF were required for the high transcriptional activity of the p125 CAT construct in BAEC. Results of our earlier studies (20) also demonstrated that the region of the SM ␣-actin promoter upstream from Ϫ125 bp to Ϫ2.8 kb contained negative regulatory elements that completely suppressed the transcriptional activity of the Ϫ125-bp region in BAEC. To test whether CArG elements with strong SRF affinities like the c-fos SREs would alter the activity of the SM ␣-actin promoter in BAEC, we transfected a pBL 155 CAT wild-type (Wt) construct and a pBL 155 CAT construct containing two c-fos SREs (pBL 155 CAT SRE-SRE) into BAEC (Fig. 5). For comparison, constructs were also transfected in parallel into SMC. Transfection data demonstrated that activity of the pBL 155 CAT Wt construct was ϳ6-fold higher in SMC as compared with BAEC. Consistent with our earlier studies, mutations of the CArG elements A and B alone or in combination completely abolished CAT activities in SMC but had little effect in BAEC. However, replacement of the SM ␣-actin CArGs with c-fos SREs (i.e. pBL 155 CAT SRE-SRE) resulted in a marked increase in CAT activity in BAEC (ϳ600-fold over promoterless controls). This activity was CArG-dependent, since mutation of the SRE at either position 1 or 2 reduced CAT activity in BAEC to that of the pBL 155 CAT Wt construct. Importantly, single SRE mutations reduced but did not abolish CAT activities in SMC. This is in marked contrast to observations with wild-type promoter constructs where mutation of CArG A or B alone completely abolished activity in SMC. Taken together, these results suggest the following: 1) replacement of the relatively weak SRF binding sites of the SM ␣-actin promoter with strong CArG elements resulted in high level SM ␣-actin expression in BAEC; 2) the gain in transcriptional activity obtained with the pBL 155 CAT SRE-SRE construct in BAEC was mediated through a CArGdependent mechanism whereas CArG-independent factors regulated the activity of the pBL 155 CAT Wt construct; and 3) the presence of a c-fos SRE at either position 1 or 2 reduced the dependence of transcriptional activity on a second CArG element in SMC.
Since we previously have shown that upstream sequences in the SM ␣-actin promoter selectively repressed transcriptional activity of SM ␣-actin in BAEC, we also tested whether c-fos SRE substitutions in the context of a 2.8-kb SM ␣-actin promoter (designated pProm CAT SRE-SRE) could overcome the effects of these negative acting elements. Transfection results demonstrated that the activity of the pProm CAT SRE-SRE construct in BAEC was increased as compared with the wildtype (pProm CAT) construct (Fig. 6). However, activity of pProm CAT SRE-SRE in BAEC was much less than the activity of the same construct in SMC. Taken together, these results indicate that strong CArG elements could only partially overcome negative regulatory elements between Ϫ125 bp to Ϫ2.8 kb that suppress activity in BAEC.
To test further the importance of the SM ␣-actin CArGs for SMC-specific regulation of this gene, we tested the activity of SRE containing SM ␣-actin constructs in L6 skeletal myotubes. Skeletal myotubes express SM ␣-actin transiently during development (55), but expression is differentially regulated as compared with SMC (20). For example, high transcriptional activity in skeletal myotubes was shown to be dependent on the upstream region from Ϫ125 bp to Ϫ271 bp, whereas little or no activity was seen with the p125 CAT construct (20). Moreover, we have shown that this gene is a target of the skeletal musclespecific HLHs whose effects are mediated by two E boxes located at Ϫ214 bp and Ϫ254 bp (56). Transfection results demonstrated that the pBL 155 CAT SRE-SRE construct had markedly higher activity in skeletal myotubes as compared with the wild-type construct (Fig. 7A). Indeed, activities exceeded that of the pProm CAT construct indicating that the presence of two strong CArG elements supplants the normal requirement for E boxes for expression of SM ␣-actin in skeletal myotubes.
Cell specificity of the SM ␣-actin CArGs was also tested in AKR-2B fibroblasts that do not express their endogenous SM ␣-actin gene under normal circumstances, although they express mouse SM ␣-actin promoter/reporter constructs containing disruption or deletion of the promoter region from Ϫ191 bp to Ϫ221 bp which acts as a repressor in these cells (57). Results of our studies demonstrated significant activity of the p155 CAT construct in AKR-2B cells that was increased to extremely high levels by replacement of the CArGs with c-fos SREs (i.e. p155 CAT SRE-SRE was ϳ600-fold over control) (Fig. 7B). However in marked contrast to observations in SMC, inclusion of the upstream region from Ϫ155 bp to Ϫ2.8 kb completely abolished activity of the wild-type p155 CAT construct in AKR-2B cells, and also greatly reduced the activity of the pProm CAT SRE-SRE construct. These data indicate that strong CArGs cannot counteract the potent repressor activity associated with the region upstream of Ϫ155 bp in AKR-2B fibroblasts, whereas they can override negative regulatory elements upstream from Ϫ155 bp in SMC.
AII Increased SRF Binding to the SM ␣-Actin CArGs as Well as to the c-fos SRE-The preceding studies provide clear evidence that the relatively weak SRF binding sites of the SM ␣-actin promoter are important for cell type-specific expression of SM ␣-actin. Results of recent studies from our laboratory demonstrated that AII inducibility of SM ␣-actin was dependent on both CArG boxes A and B and partially dependent on a MHox binding site (ATTA) situated 5Ј to CArG B (40). Moreover, AII-induced stimulation of SM ␣-actin was associated with markedly increased SRF binding to both CArG elements, and MHox was shown to enhance SRF binding to CArG B and overexpression of MHox transactivated SM ␣-actin expression. These results suggest that enhancement of SRF binding to lesser affinity SRF binding sites may represent an important mechanism to maximally up-regulate SM ␣-actin expression. If so, then replacement of low SRF binding sites by high SRF binding sites should result in higher constitutive activity and reduced responsiveness to AII.
To test this hypothesis, we first addressed whether SRF binding activity was increased in nuclear extracts from SMC treated with AII as compared with SFM vehicle and whether similar differences in binding activity of probes for CArG A, CArG B, and the c-fos SRE were seen with SMC extracts as observed with recombinant SRF (Fig. 2). This is important to rule out possible post-translational modifications of SRF that might differentially affect binding to these probes. Consistent with our previous findings, results showed that AII treatment was associated with increased SRF binding activity, with binding activity being greatest with the c-fos SRE, intermediate with CArG B, and lowest with CArG A in a manner similar to that seen with recombinant SRF (Fig. 8). Interestingly, the fact that AII was capable of increasing SRF binding to the c-fos SRE suggests that some degree of AII-mediated stimulation of constructs containing c-fos SREs might be expected.

FIG. 8. Gel shift analysis of the effects of AII treatment of SMC on SRF binding to CArG A and B and c-fos SRE oligonucleotides.
Radiolabeled 20-bp CArG A (lanes 1 and 2), CArG B (lanes 3 and 4), and c-fos SREs (lanes 5 and 6) double-stranded oligonucleotides were incubated with nuclear extracts (3 g) from rat SMC treated with AII (10 Ϫ6 M) or SFM. Polyclonal SRF antibodies (Santa Cruz) were added to the binding reaction in the absence of radiolabeled probe and incubated for 20 min at room temperature. ness of SM ␣-actin, we performed transfection studies using various constructs described earlier (see Figs. 1-3). It should be noted that transfections were carried out under serum-free conditions in confluent, growth-arrested SMC, since we (44,47) and others (58) have previously demonstrated that such conditions are necessary for AII-induced hypertrophic responses. However, transfection efficiencies and corresponding reporter activities are lower under such conditions as compared with SMC growing exponentially in serum. For this reason, and due to other undefined changes due to serum withdrawal, results of these AII experiments cannot be directly compared with results of our earlier experiments (Figs. 1-7) which were carried out in growing, subconfluent cells in serum-containing media. Results demonstrated that replacement of SM ␣-actin CArGs by either one or two c-fos SREs resulted in an increase in basal as well as AII-stimulated activity of each of the c-fos SRE-substituted pBL 155 CAT constructs (Fig. 9A). However, as depicted in Fig.  9B, the extent of AII inducibility of these constructs (pBL 155 CAT B-SRE, pBL 155 CAT SRE-SRE, and pBL 155 CAT SRE-A) was reduced approximately by half as compared with the pBL 155 CAT wild-type construct. AII inducibility was also found to be greatly reduced (pBL 155 CAT A-SRE) or completely abolished (pBL 155 CAT A-B or A-A) when a weak CArG was placed at position 2. Taken together, these results indicate that AII inducibility of SM ␣-actin is reduced when the endogenous SM ␣-actin CArGs are replaced by high affinity c-fos SREs. In addition, results suggest that AII inducibility requires the presence of at least an intermediate SRF binding site at position 2.

MHox-mediated Transactivation of SM ␣-Actin Was Markedly Reduced by Replacement of the SM ␣-Actin CArGs by c-fos
SREs-Previous studies in our laboratory demonstrated that an MHox binding site at Ϫ145 bp of the SM ␣-actin promoter was required for maximal AII responsiveness of SM ␣-actin and that overexpression of MHox transactivated SM ␣-actin expression 3-4-fold (40). To gain insight into the mechanisms contributing to reduced AII inducibility in the presence of c-fos SREs, we tested whether MHox-mediated transactivation of SM ␣-actin was altered when SM ␣-actin CArGs were replaced by c-fos SREs. An MHox expression vector was co-transfected along with a p155 CAT construct containing either the SM ␣-actin wild-type CArGs or c-fos SRE substitutions, whereas controls were co-transfected with the empty expression vector. Results showed that MHox-mediated transactivation of the wild-type p155 CAT was ϳ400% and only ϳ40% in the presence of c-fos SREs (Fig. 10). No transactivation was observed with a construct carrying two weak CArGs (CArG A). Consistent with these results, we have previously shown that MHox fails to enhance SRF binding to CArG A (40) and Grueneberg et al. (53) showed that Phox1, the human homologue of MHox, failed to impart serum responsiveness to weak SRF binding sites. In summary, these results suggest that the ability of MHox to transactivate transcription is markedly influenced by the SRF binding activities of CArG elements. DISCUSSION The aim of the present study was to determine whether the reduced SRF binding activities of CArG A (Ϫ62) and B (Ϫ112) within the SM ␣-actin promoter contribute to cell type-specific expression of SM ␣-actin and responsiveness to AII, a contract- ile agonist shown to mediate hypertrophic growth responses in SMC. We demonstrated that replacement of CArG A and CArG B by the strong SRF binding site c-fos SRE resulted in increased basal activity. However, it also resulted in relaxed cell-specific expression, reduced inducibility by AII, and markedly reduced transactivation by MHox. Moreover, we provided evidence that the position of the CArG elements was also important for basal and AII-mediated stimulation of SM ␣-actin expression. These results suggest that CArG elements with reduced SRF binding activities contribute to cell type-specific expression of SM ␣-actin and are required for maximal AII responsiveness.
CArG elements are present in the promoters of many genes that are independently regulated including immediate-early genes (Refs. 1-5, reviewed in Ref. 6) and muscle-specific genes (Refs. 7-15, reviewed in Ref. 16). Gene-specific differences are due at least in part to sequence variations within the internal A/T-rich center of the CArGs as well as by changes in flanking sequences that provide additional binding sites for factors that interact with SRF and modulate its function (reviewed in Refs. 6,37,38,40,53,54,59). For example, Walsh and co-workers (41,60) demonstrated that CArG elements are not functionally interchangeable. Replacement of the most proximal CArG box of the skeletal (SK) ␣-actin promoter (muscle regulatory element) by a c-fos SRE led to constitutive expression of SK ␣-actin in non-muscle cells. Consistent with the observations by Walsh et al. (41,60), we also found that replacement of the SM ␣-actin CArGs by c-fos SREs was associated with relaxed cell type-specific expression of SM ␣-actin. However, our studies also revealed a fundamental difference between cells that express their endogenous SM ␣-actin gene (SMC and skeletal myotubes) and those that do not (e.g. BAEC and AKR-2B fibroblasts) with respect to the effects of c-fos SRE substitutions when examined in a short (155 bp) versus a longer (2.8 kb) SM ␣-actin promoter context. In SMC, strong CArGs completely overcame the effects of negative acting elements located between Ϫ155 bp and Ϫ2.8 kb. However, whereas the presence of strong CArGs within a Ϫ155-bp context resulted in high transcriptional activity in BAEC and AKR-2B, strong CArG elements were only modestly effective in overcoming the repressor effects of negative regulatory elements upstream of Ϫ155 bp in the longer promoter context. In L6 skeletal myotubes, expression of wild-type SM ␣-actin promoter constructs was dependent on the combinatorial interaction of the CArG boxes and E boxes (20). However, substitution of the SM ␣-actin CArGs with c-fos SREs resulted in high transcriptional activity of the p155 CAT construct that lacks the E box elements that are normally required for expression in this cell type. Taken together, these results indicate that cell type-specific expression is not only dependent on the reduced SRF binding activity of the degenerate SM ␣-actin CArGs but also on powerful cell type-specific repressor elements that limit expression even in the presence of c-fos CArG elements that bind SRF with very high affinity.
Based on genomic footprint analysis showing that the c-fos SRE is constitutively occupied, many models of CArG-dependent regulation of c-fos have assumed that SRF-CArG interaction was not rate-limiting/or regulated (61). Rather, serum and growth factor responsiveness of the c-fos gene was shown to be mediated at least in part by the interaction of SRF with members of the ets domain containing transcription factors including ELK-1 and SAP-1 (reviewed in Refs. 18, and 37) that formed ternary complexes with SRF as well as non-ternary complex-dependent pathways that signal via Rho family GTPases (62). A growing body of recent evidence, however, suggests that CArG-dependent regulation of the c-fos as well as other cell type-specific genes is also regulated at the level of SRF-CArG interactions (33-36, 39, 40). This includes post-translational modifications of SRF that modify its binding (17,63), alterations in the level of SRF expression (34,39), and interaction of SRF with factors including YY1 (54,64,65) and homeodomain factors (38,40,53). Studies by Grueneberg et al. (53) have shown that Phox1, the human counterpart of MHox, enhanced SRF binding to the c-fos SRE and that overexpression of Phox1 transactivated a test construct consisting of a single copy of the c-fos SRE coupled to a minimal c-fos promoter. The present studies are consistent with those of Grueneberg et al. (53) and for the first time identify a gene and a cell context in which regulation of SRF binding to highly conserved degenerate CArG motifs by a homeodomain factor are critical for both SMC-specific and agonist inducibility. Our results in this and previous studies (40) suggest that the presence of relatively weak SRF binding CArG motifs within the SM ␣-actin promoter permits a greater extent of regulation of this gene than would be possible with strong CArGs that are likely to constitutively bind SRF even under basal conditions. Several lines of evidence support this hypothesis. First, AII responsiveness was reduced by half in the presence of c-fos SREs. Second, overexpression of MHox transactivated the wild-type promoter by ϳ400%, whereas transactivation of a construct containing two c-fos SREs was only ϳ40% (Fig. 10). Third, the presence of a homeodomain binding site at Ϫ145 was required for maximal MHox-induced transactivation in that effects were reduced by 50% with a p125 CAT construct missing the MHox binding site, or by mutation of the homeodomain site within a p155 CAT context (40). However, when the SM ␣-actin CArGs were substituted by c-fos SREs within the 155or 125-bp context, no differences in transcriptional activities were observed 2 indicating that transcriptional activity of SM ␣-actin is less dependent on MHox in the presence of strong SRF binding sites.
Results of the present studies also provide evidence that the positional context of the SM ␣-actin CArGs was critical for transcriptional activity in that switching of CArG A and B resulted in almost complete loss of both basal and AII-induced activity. Whereas results of our previous studies (20) have clearly shown that both CArG A (position 1) and CArG B (position 2) are required for transcriptional activity in SMC, there appears to be a requirement for the stronger of the two CArGs be located in the more distal position (Figs. 3 and 4). This may be important because of proximity to the homeodomain binding site or alternatively may relate to structural requirements for formation of a higher order transcription initiation complex (66). Of interest, however, our data show that the SRF binding activity of CArG A appears to be too low to function effectively at position 2, although it is essential for activity in its normal context (20). Consistent with these observations, Grueneberg et al. (53) has shown that Phox1, the human homologue of MHox, failed to impart serum responsiveness to very poor SRF-binding sites, and we found that MHox did not transactivate a construct containing two CArG A elements. In addition, MHox did not increase SRF binding to CArG A in gel shift assays using recombinant proteins (40). Together these results suggest that binding of SRF to CArG B, promoted by MHox, may be a crucial initial step in transcriptional activation. Subsequently, SRF-CArG B interaction may induce DNA bending (67), allowing the weaker SRF-binding site CArG A to be occupied, thereby forming an energetically favorable multiprotein-DNA complex. Consistent with this model, Lee et al. (54) demonstrated that the two high affinity proximal and distal skeletal ␣-actin SREs are first bound cooperatively by SRF with concurrent DNA bending, which then facilitates SRF interaction with the weak central site CArG. Taken together, these data suggest that cooperative interaction of strong and weak CArGs contribute to CArG-dependent regulation of multiple muscle-specific genes including SM ␣-actin.
Although our studies focused on a single SMC differentiation marker, SM ␣-actin, the observation that SRF binding is modulated by homeodomain proteins may represent a general regulatory paradigm that may contribute to control of other CArGdependent muscle genes. Consistent with this, Chen and Schwartz (59), have demonstrated that the homeodomain protein Nkx-2.5, in concert with SRF, was required for expression of cardiac ␣-actin in nonmyogenic fibroblasts suggesting a role for Nkx-2.5 in conferring cell type-specific expression of cardiac ␣-actin. In addition, it is interesting to speculate that the interaction of homeodomain proteins and SRF may also contribute to the coordinate expression of multiple CArG-dependent genes that are characteristic of mature differentiated smooth muscle, including SM MHC and SM-22␣. However, there is also evidence that CArG-dependent expression of these SMC genes is regulated in a gene-specific manner. For example, both CArG boxes A and B were required for high level expression of SM ␣-actin (20). In contrast, a single proximal CArG element with relatively high SRF binding activity (predicted based on the lack of Gua or Cyt substitution in the A/T-rich CArG center) was sufficient for high level expression of SM MHC and SM-22␣ in SMC (22,24) suggesting that a single strong CArG might be sufficient to drive significant transcription of these genes. Consistent with this, our results demonstrated that SM ␣-actin promoter constructs containing a single c-fos SRE were sufficient to drive transcription of SM ␣-actin (Fig. 5), whereas a single wild-type SM ␣-actin CArG failed to do so. Finally, it is critical to emphasize that MHoxdependent regulation of SRF binding to the SM ␣-actin CArGs alone is unlikely to be sufficient to control cell type-specific expression of SM ␣-actin, since MHox expression is clearly not restricted exclusively to SMC, although it does show mesodermally restricted activity (68). Rather, cell type-specific expression of SM ␣-actin appears to depend on the combinatorial interaction of multiple cis-elements and trans-factors in a manner analogous to many cardiac specific genes (reviewed in Ref. 16).