Expression of the smooth muscle myosin heavy chain gene is regulated by a negative-acting GC-rich element located between two positive-acting serum response factor-binding elements.

To identify cis- and trans-acting factors that regulate smooth muscle-specific gene expression, we studied the smooth muscle myosin heavy chain gene, a rigorous marker of differentiated smooth muscle. A comparison of smooth muscle myosin heavy chain promoter sequences from multiple species revealed the presence of a highly conserved 227-base pair domain (nucleotides −1321 to −1095 in rat). Results of a deletion analysis of a 4.3-kilobase pair segment of the rat promoter (nucleotides −4220 to +88) demonstrated that this domain was necessary for maximal transcriptional activity in smooth muscle cells. Gel-shift analysis and site-directed mutagenesis demonstrated that one true CArG and another CArG-like element contained within this domain were both recognized by the serum response factor and were both required for the positive activity attributable to this domain. Additional studies demonstrated that mutation of a GC-rich sequence within the 227-base pair conserved domain resulted in a nearly 100% increase in transcriptional activity. Gel-shift analysis showed that this GC-rich repressor element was recognized by both Sp1 and Sp3. These data demonstrate that transcriptional control of the smooth muscle myosin heavy chain gene is highly complex, involving both negative and positive regulatory elements, including CArG sequences found in the promoters of multiple smooth muscle differentiation marker genes.

To identify cis-and trans-acting factors that regulate smooth muscle-specific gene expression, we studied the smooth muscle myosin heavy chain gene, a rigorous marker of differentiated smooth muscle. A comparison of smooth muscle myosin heavy chain promoter sequences from multiple species revealed the presence of a highly conserved 227-base pair domain (nucleotides ؊1321 to ؊1095 in rat). Results of a deletion analysis of a 4.3-kilobase pair segment of the rat promoter (nucleotides ؊4220 to ؉88) demonstrated that this domain was necessary for maximal transcriptional activity in smooth muscle cells. Gel-shift analysis and site-directed mutagenesis demonstrated that one true CArG and another CArG-like element contained within this domain were both recognized by the serum response factor and were both required for the positive activity attributable to this domain. Additional studies demonstrated that mutation of a GC-rich sequence within the 227-base pair conserved domain resulted in a nearly 100% increase in transcriptional activity. Gel-shift analysis showed that this GC-rich repressor element was recognized by both Sp1 and Sp3. These data demonstrate that transcriptional control of the smooth muscle myosin heavy chain gene is highly complex, involving both negative and positive regulatory elements, including CArG sequences found in the promoters of multiple smooth muscle differentiation marker genes.
Intimal migration and proliferation of vascular smooth muscle cells (SMCs) 1 are known to play an integral role in development of atherosclerotic disease (1,2), and numerous factors have been identified that have growth promoting and/or chemotactic activity for SMCs (2). An additional feature of SMCs within atherosclerotic lesions is that cells exhibit marked differences in morphology and protein expression patterns as compared with normal medial SMCs (3)(4)(5)(6), a process referred to as "phenotypic modulation" (7). This is characterized by decreased expression of proteins that are characteristic of differentiated SMCs, including the SM isoforms of contractile proteins, as well as altered growth regulatory properties, lipid metabolism, matrix production, and decreased contractility (reviewed in Ref. 8). Of particular significance, many of these phenotypic alterations in intimal SMCs cannot simply be viewed as a consequence of atherosclerotic disease, but rather are likely to play a major role in its development and/or progression.
Although the SMC exhibits a high degree of plasticity and, unlike skeletal and cardiac myocytes, does not undergo terminal differentiation, it is a very specialized cell, whose differentiated function is dependent upon the coordinate regulation of a large number of cell type-specific/selective products, including contractile proteins, receptors, signal transduction molecules, and ion channels (8). Contractile proteins that are highly abundant and specific to the SMC represent obvious candidates for studying molecular control of SMC differentiation. Consequently, it is not unexpected that the best studied SM-specific marker is smooth muscle ␣-actin, which first appears in the putative SMCs that envelope the dorsal aortae (9). Although it is transiently expressed during development in other mesodermally derived tissues (10) and in fibroblasts during wound repair (11), smooth muscle ␣-actin is normally expressed only in SMCs and SMC-like cells in the adult (12). A second well characterized SM-specific contractile protein is smooth muscle myosin heavy chain (SM-MHC). The SMC contains four isoforms of SM-MHC (SM-1A, SM-1B, SM-2A, SM-2B) (13)(14)(15), which are all derived from a single gene via alternative splicing (16). In situ hybridization and RNase protection analysis of SM-MHC gene expression during mouse embryogenesis revealed that SM-MHC mRNAs were present only in smooth muscle-containing tissues; no transcripts were detected in the developing brain, heart, or skeletal muscle (17). Immunohistochemical studies that discriminated between the SM-1 and SM-2 isoforms revealed that, in both rabbit and human aortae, SM-MHC protein was specific to the SMC and that SM-1 predominated early during embryonic development, with SM-2 appearing postnatally (18,19). These studies indicate that, similar to smooth muscle ␣-actin, SM-MHC represents a highly rigorous SMC marker and a good candidate gene for discerning transcriptional mechanisms important for maintenance of the differentiated SMC phenotype. the University of Virginia Academic Enhancement Program in Gene Transfer and Gene Therapy in the Cardiovascular System. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U83321.
The promoters of the rat, mouse, and rabbit SM-MHC genes have been recently cloned and partially characterized. A reporter construct containing 2266 bp of the 5Ј-flanking region of the rabbit SM-MHC gene was shown to be highly active in cultured rat aortic SMCs and only minimally active in other cell types (20). In a separate study on the rabbit SM-MHC promoter, Kallmeier et al. (21) identified a 107-bp fragment (nucleotides Ϫ1332 to Ϫ1225) that enhanced transcription in what appeared to be a highly SMC-specific manner. Analysis of the rat SM-MHC promoter in multiple cell types indicated that a 68-bp domain (nucleotides Ϫ1249 to Ϫ1317) was important for restriction of expression to the SMC (22). Constructs longer than nucleotide Ϫ1317 were found to be significantly active only in the SMC. Another recent report on mouse SM-MHC showed that, similar to rat and rabbit, 1500 bp of the mouse SM-MHC promoter was strongly active only in cultured SMCs (23). In all of the above studies, several potential cis-elements, including multiple E boxes, GC boxes, and CArG or CArG-like boxes, were identified based on sequence similarities to known cis-elements. Some of these potential elements were also present in regions with functional activity. However, only in the mouse promoter study by Watanabe et al. (23) were any potential cis-elements specifically mutated and tested for function. They demonstrated by site-directed mutagenesis that two CCTCCC boxes located proximal to the TATA box were bound by Sp1 and functioned as positive-acting cis-elements.
In this study, we utilized ϳ4.2 kilobase pairs of the rat SM-MHC gene 5Ј-flanking region and created a series of mutated CAT fusion constructs to identify functionally important cis-elements. We specifically targeted those sequences that exhibited homology to known cis-elements and that were also conserved between species. A sequence alignment of the rat, mouse, and rabbit SM-MHC promoters revealed a highly conserved 227-bp domain that spanned nucleotides Ϫ1321 to Ϫ1095 in rat. Deletion analysis showed that this domain was required for maximal promoter activity in SMCs. Site-directed mutagenesis was utilized to further demonstrate that two of the three CArG or CArG-like boxes contained within the 227-bp domain functioned as positive-acting cis-elements recognized by the serum response factor (SRF) or an SRF-like protein. A GC-rich sequence located within this domain was determined to function as a negative-acting cis-element recognized by members of the Sp1 family. The fact that CArG motifs are required for expression of both the SM-MHC gene (this study) and the SM ␣-actin gene (24) suggests that this motif may be important for the coordinate regulation of SM-specific genes during SMC differentiation.

Construction of Rat SM-MHC Promoter-CAT Expression Plasmids-
The cloning, determination of the ϩ1 start site, and nucleotide sequence (nucleotides Ϫ1699 to ϩ121) of the rat SM-MHC promoter have been previously reported (22). In this study, we ligated a 4308-bp BglII fragment, spanning nucleotides Ϫ4220 to ϩ88, into a partially filled-in SalI site of the pCAT-Basic reporter vector (Promega). This clone (pCATϪ4220) was sequenced by the dideoxy method of Sanger et al. (25) using either an automated sequencer (Applied Biosystems Inc.) or a Sequenase kit (U. S. Biochemical Corp.). The sequence from nucleotides Ϫ4220 to Ϫ2400 was generated from analysis of only a single strand, whereas both strands were sequenced from nucleotides Ϫ2400 to ϩ88. Serial deletion constructs were generated from the pCATϪ4220 clone using exonuclease III and the protocol provided by the manufacturer (Stratagene). Deletion clones that targeted specific sites were generated using Taq polymerase and a thermal cycler (Perkin-Elmer). Oligonucleotide primers were purchased from a commercial source (Operon Technologies, Inc.) and contained HindIII and XbaI linkers such that PCR products could be directionally cloned into the pCAT-Basic vector (Promega). Site-directed mutagenesis of the pCATϪ1346 construct was performed using the Ex-site mutagenesis kit according to the manufacturer's instructions (Stratagene). The orientation and in-tegrity of the mutated constructs were determined by dideoxy sequencing (25).
Cell Culture, Transient Transfections, and Reporter Gene Assays-SMCs from rat thoracic aorta were isolated and cultured as described previously (26). Rat L6 skeletal myoblasts were cultured as described previously (24) and fused into myotubes using the protocol of Yaffe (27). SMCs (passages 10 -22) and L6 myoblasts were seeded for transient transfection assays into 6-well plates at a density of 2 ϫ 10 4 . These densities were chosen so that the cells would be 70 -80% confluent at the time of transfection (24 h after plating). Transfections of the CAT reporter gene constructs, subsequent growth conditions of the SMCs and L6 myoblasts, and preparation of the cell extracts for measurement of CAT activity were all performed as described previously (24), with only slight modification, namely, the amount of plasmid DNA added per well was 4 g, and N- [1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate (7.5 l/g of DNA; Boehringer Mannheim) was utilized as the transfection reagent. All CAT activity values were normalized to the protein concentration of each cell lysate as measured by the Bradford assay (28). Early transfection experiments included a ␤-galactosidase plasmid as a cotransfection partner for measurement of transfection efficiency. However, the ␤-galactosidase activity measurements did not result in qualitative changes in the data, nor did they affect experimental variability. Since the cotransfections conferred no advantage in the reduction of experimental error and had the potential of competing for limited trans-acting factors that regulate SM-MHC transcription, the ␤-galactosidase cotransfections were discontinued. In each experiment, the promoterless pCAT construct was also transfected into triplicate wells to serve as the base-line indicator of CAT activity, and the activity of each promoter construct is expressed relative to the promoterless construct set to 1. Additionally, an SV40 promoter-CAT construct with enhancer (Promega) served as a positive control of transfection and CAT activity. All SMC CAT activity values represent at least three independent experiments, with each construct tested in triplicate per experiment. L6 myoblast and myotube CAT activity values represent two independent experiments. Relative CAT activity data are expressed as the means Ϯ S.D. computed from the results obtained from each set of transfection experiments. One-way analysis of variance followed by the Newman-Keuls's multiple range test were used for data analysis. Values of p Ͻ 0.05 were considered statistically significant.
Preparation of SMC Nuclear Extracts and Electrophoretic Mobility Shift Assays (EMSAs)-Nuclear extracts were prepared from rat aortic SMCs grown to confluency under the same media conditions utilized for the transfection assays. The actual preparative method used was essentially that of Dignam et al. (29); protease inhibitors were added as described previously (24).
The Sp1, Sp3, SRF, and CTF/NF1 antibodies used for EMSA supershift experiments were all purchased commercially (Santa Cruz). EMSA binding reactions were set up as described above and incubated for 20 min; 2 l of the antibody was added to the mixture; and the reaction was incubated for another 15 min at room temperature and then loaded onto the gel for electrophoresis.
DNase I Footprinting-Binding conditions were as described for EMSAs, except that 40 g of SMC nuclear extract was used per reaction. DNase I digestions and electrophoresis of the samples were performed as described previously (30). Location of the footprinted regions was determined by including on the gel a dideoxy sequence marker ladder, and a Maxam-Gilbert G-specific ladder of the labeled fragment was also generated to verify the integrity of the probe (data not shown). The plasmid construct pCATϪ1346 was utilized as template DNA for PCR-generated footprint probes. The oligonucleotides utilized in the PCRs were 5Ј-ccataacgctccatgggct-3Ј (nucleotides Ϫ1346 to Ϫ1328) and 5Ј-gcatggagagtgggaggga-3Ј (nucleotides Ϫ1089 to Ϫ1071). The Ϫ1346/ Ϫ1328 oligonucleotide primer used in the PCRs was labeled at the 5Ј-end with T4 polynucleotide kinase (Life Technologies, Inc.) and [␥-32 P]ATP (DuPont NEN) to generate an asymmetrically labeled double-stranded probe. Probes were purified on a 5% nondenaturing polyacrylamide gel.

RESULTS
Identification of a 227-bp Sequence Domain Conserved between the SM-MHC Promoters of Rat, Mouse, and Rabbit-The rat SM-MHC promoter has been recently cloned and sequenced from bp Ϫ1699 to ϩ121 (22). To assist in the identification of potential cis-elements, we sequenced an additional 2500 bp of the rat SM-MHC promoter (up to nucleotide Ϫ4220) and compared the rat promoter sequence with the published sequences of rabbit (Ϫ2266 bp) (20) and mouse (Ϫ1500 bp) (23). We identified only two domains that shared obvious sequence similarities among all three species (Fig. 1A). The distal (relative to the ϩ1 start site) conserved domain was located in the rat promoter between nucleotides Ϫ1321 and Ϫ1095 (Fig. 1B). Within this 227-bp domain, we identified five known cis-elements, all of which were totally conserved between the three species. These elements included two CArG-like boxes and one true CArG box (hereafter referred to as CArG boxes 1, 2, and 3, in order of proximal to distal), a GC-rich element, and a CTF/ NF1 site. These three elements represent potential binding sites for SRF (31), Sp1 (32), and CTF/NF1 (33), respectively. It should be noted that the centrally located nucleotides of the 13-bp CTF/NF1 site (TGGC(N) 5 GCCA) contain sequence differences between the three species; however, these residues are not considered to be important for factor binding (34). The second conserved domain encompassed the TATA box and extended to nucleotide Ϫ128 in rat. Contained within this domain were two conserved CCTCCC elements that were previously shown in transient transfection studies to function as positiveacting cis-elements in the mouse SM-MHC promoter (23).

Deletion Analysis of the Rat SM-MHC Promoter Revealed That the Conserved 227-bp Domain Is Required for Maximal
Activity-To identify functional elements contained within the rat SM-MHC promoter, a 4308-bp BglII fragment (bp Ϫ4220 to ϩ88) was cloned into a pCAT reporter vector. A series of deletion constructs were created by treating the pCATϪ4220 construct with exonuclease III and, when targeting specific regions of interest, by PCR (see "Materials and Methods"). Transient transfection of these constructs into cultured rat aortic SMCs revealed that pCATϪ1346, which contained both of the conserved domains, was the most active construct, with an activity level ϳ48-fold over promoterless pCAT (Fig. 2). The addition of more promoter 5Ј-flanking DNA reduced transcriptional activity to 29-fold (pCATϪ3443) and 24-fold (pCATϪ4220) over promoterless pCAT. This observed decrease in activity suggests that negative regulatory elements exist upstream of nucleotide Ϫ1346. The pCATϪ1182 construct, which is missing CArG boxes 3 and 2 and the GC-rich element, yielded an activity 34-fold over promoterless pCAT. When the entire 227-bp conserved domain was deleted (pCATϪ1102), transcriptional activity further decreased to ϳ17-fold over promoterless pCAT. The activity of the pCATϪ562 construct was found to be approximately equal (14-fold over promoterless pCAT) to that of the pCATϪ1102 construct. The notable difference in activity between pCATϪ1346 and pCATϪ1102 suggests that a positive-acting element(s) resides within the 227-bp conserved domain. The incremental gain of activity noted for the pCATϪ1182 and pCATϪ1346 constructs indicates that more than one positive-acting cis-element is present within this region.
Transcriptional activity of the SM-MHC promoter has been measured in several other non-SMC types, including L8 skeletal myoblasts (22), NIH3T3 mouse fibroblasts (23), and C 2 C 12 mouse myoblasts (20). For comparative purposes, the deletion constructs were tested for activity in L6 myoblasts and myotubes (Fig. 3). Transient transfection analysis of L6 myotubes revealed that all constructs were only minimally active relative to SMCs, ranging from 8 to 14-fold activity over promoterless pCAT. Similar results were seen in L6 myoblasts, except that the range in activity (4 -7-fold over promoterless pCAT) was about half that of myotubes. In contrast to the SMC data, the addition of the 227-bp domain failed to enhance transcription in either L6 myoblasts or myotubes, suggesting that the activity of this domain is regulated in a cell type-specific manner.
Multiple ments were performed using nuclear extracts from cultured rat aortic SMCs. Using a probe that extended from nucleotides Ϫ1346 to Ϫ1071, we identified three domains of protein binding (Fig. 4). These same three domains were also identified in a DNase I analysis of the antisense strand (data not shown). A summation of the footprint data shows that footprint 3 extended from nucleotides Ϫ1234 to Ϫ1206 and encompassed CArG box 2 and an adjacent GC-rich element. Footprint 2 extended from nucleotides Ϫ1146 to Ϫ1121 and encompassed the CTF/NF1 site. Footprint 1 extended from nucleotides Ϫ1115 to Ϫ1098 and encompassed CArG box 1. The identification of multiple sites of protein-DNA interaction is consistent with the promoter deletion data that indicated the presence of multiple regulatory elements within this region.
CArG Boxes 1 and 2 Are Positive-acting cis-Elements, and the GC-rich Sequence Functions as a Negative-acting cis-Ele-ment-To determine if the sites of protein-DNA interaction described in Fig. 4 represented functional elements, we mutated each of the four known cis-elements contained within footprints 1-3. CArG boxes 1 and 2, the GC-rich element, and the CTF/NF1 site were all mutated within the context of the most active construct (pCATϪ1346) and tested for functional activity. Although CArG box 3 did not display any protein binding activity, we mutated this element anyhow due to its similarity to other SRF recognition sites and sequence conservation among all three species. Transient transfection of the mutated constructs into rat aortic SMCs revealed that the SM-MHC promoter was regulated by both positive-and negative-acting cis-elements. As shown in Fig. 5A, CArG box 3 was totally deleted, and CArG boxes 1 and 2 were mutated at the conserved 5Ј-CC doublet. The presence of these two cytosine residues has been previously shown to be necessary for SRF binding and functional activity (24). Transfection analysis of the three CArG boxes showed that deletion of CArG box 3 had virtually no effect on functional activity relative to wild-type pCATϪ1346 (Fig. 5B). However, mutations of CArG boxes 1 and 2 resulted in an ϳ50% decrease in transcriptional activity (20-and 23-fold over promoterless pCAT, respectively) when compared with wild-type pCATϪ1346 (48-fold over promoterless pCAT). The reduced activity of pCAT-CArG1Mut and pCAT-CArG2Mut suggests that a major portion of the positive activity of the 227-bp conserved domain (the difference in activity between pCATϪ1102 and pCATϪ1346) is dependent on the presence of CArG boxes 1 and 2.
The sequence identified as the GC-rich element contains a core Sp1 site (CCCGCCC); thus, the central nucleotides of this sequence were specifically mutated such that any putative Sp1 binding would be abrogated (32). Alteration of this element resulted in a marked increase in transcriptional activity (Fig.  5B). The actual level of activity (94-fold over promoterless pCAT) was approximately double that of wild-type pCATϪ1346 and nearly equal to that of the powerful SV40 viral promoter (pSV40-CAT). This result suggests that the GC-rich element located at nucleotide Ϫ1215 functions as a negative-acting cis-element. The CTF/NF1 site in pCAT-NF1Mut was completely deleted due to previous reports indicating that this factor is capable of binding to either half of its palindromic recognition sequence (TGGC(N) 5  pCAT-NF1Mut was determined to be slightly higher than that of pCATϪ1346 (52-versus 47-fold over promoterless pCAT). However, these differences were not statistically significant, and further analyses of this element were not performed.
The five pCATϪ1346 mutated constructs were transfected into L6 myoblasts and myotubes to test whether any of the sequence mutations affected activity in different cell types. Relative to wild-type pCATϪ1346, none of the mutated ciselements resulted in any significant gain or loss of transcriptional activity (Fig. 5C). Of particular interest, the GC-rich mutant (pCAT-GC-richMut), which was almost twice as active as wild-type pCATϪ1346 in SMCs, was actually slightly less active than the wild type in L6 myoblasts or myotubes. These data suggest that the positive and negative elements contained within the 227-bp conserved domain function in a cell typespecific manner. Fig. 1 functioned as SRF recognition sites. Individual oligonucleotides containing CArG boxes 1-3, as well as 5 bp of 5Ј-and 3Ј-flanking sequence, were labeled, incubated with 10 g of SMC nuclear extracts, and analyzed by EMSA (Fig. 6A). Duplicate reactions that included an SRF antibody were also electrophoresed on the same gel (Fig. 6A, right panel). For comparison, we included a similar sized oligonucleotide that contained the well characterized SRF-binding site present in the c-fos promoter, the SRE (31). Similar to a previous study on SRF binding to the SM ␣-actin CArG boxes (24), we identified two closely migrating protein-DNA complexes with the CArG1 probe. The mobilities of these two bands were supershifted upon SRF antibody addition. Two bands were also identified with the SRE probe that comigrated alongside the CArG1⅐SRF complexes and supershifted with SRF antibody addition. These data demonstrate that the protein-CArG1 complexes contain SRF or an SRF-like factor. No protein-DNA complexes whose mobilities supershifted upon addition of SRF antibody were detected with either the CArG box 3 or box 2 probes.

CArG Boxes 1 and 2 Function as Recognition Sites for SRF or an SRF-like Protein-EMSA and antibody supershift analysis were utilized to determine if the three CArG boxes identified in
Transfection data of pCAT-CArG3Mut indicated that CArG box 3 was nonfunctional; as such, we did not further pursue experimentation on protein binding to this site. However, we did perform further EMSAs on CArG box 2. By increasing the amount of SMC nuclear extract from 10 to 30 g in the gel-shift reaction, we were able to detect a band that comigrated with the SRE⅐SRF complex and also supershifted upon SRF antibody addition (Fig. 6B). We have previously demonstrated that mutation of the 5Ј-CC doublet of the CArG box completely abrogated SRF binding (24). However, given the amount of extract needed for detection of the CArG2⅐SRF (or SRF-like) complex, we wanted to verify that, under these conditions, SRF did not bind to the mutated CArG2 oligonucleotide. A CArG2Mut oligonucleotide that duplicated mutations created in the pCAT-CArG2Mut construct was synthesized and tested for SRF binding. As shown in Fig. 6C (right lane), the CArG2Mut probe failed to generate a band that comigrated with the SRE⅐SRF complex, even when 30 g of SMC extract was used. In conjunction with the transfection data, these protein binding studies suggest that SRF interaction with CArG boxes 1 and 2 functions to activate transcription of the SM-MHC gene. Although the binding affinities for the three CArG boxes were not determined, the data presented in Fig. 6 suggest that the binding affinity of SRF, or an SRF-like protein, is much greater for CArG1 than for CArG2.
The Negative-acting GC-rich Element Functions as a Sp1/ Sp3 Recognition Site-Due to the considerable sequence similarities between the negative-acting GC-rich element located within the 227-bp conserved domain and the published Sp1binding site, we wanted to determine if this element functioned as a Sp1 recognition site. Using a labeled oligonucleotide that encompassed the SM-MHC GC-rich element, EMSAs were performed in the presence of different unlabeled competitors (Fig.  7). The GC-rich probe formed three predominant protein-DNA complexes (bands 1-3) in the presence of SMC nuclear extract. Bands 1 and 2 exhibited very similar mobilities and migrated through the gel slower than did band 3. This particular migration pattern of protein-DNA complexes was similar to the pattern previously described for Sp1 and Sp3 binding; using antibody supershift analysis and a Sp1 probe, Dennig et al. (36) FIG. 4. DNase I analysis of protein-DNA interaction within the 227-bp conserved sequence domain. A 276-bp DNA fragment spanning nucleotides Ϫ1346 to Ϫ1071 of the rat SM-MHC promoter was generated by PCR such that the sense strand was labeled. The gelpurified fragment was incubated with 40 g of rat SMC nuclear extract (ϩProtein lanes) and treated with DNase I. Control reactions (Control lanes) were performed in a similar manner, omitting protein. Following digestion with DNase I, products were resolved on a 6% denaturing polyacrylamide gel. To the right is a sequence summary of the three domains labeled as footprints 1-3 (thick bars). demonstrated that band 1 contained Sp1 protein and that bands 2 and 3 represented Sp3-probe complexes. It should be noted that with longer electrophoresis times, we were able to better resolve bands 1 and 2 (data not shown). Further analysis of Fig. 7 shows that the consensus Sp1 oligonucleotide competed very efficiently for factor binding, even more so than did the wild-type GC-rich competitor. The fact that all bands disappeared equally with addition of the Sp1 competitor oligonucleotide is consistent with Sp1 and Sp3 binding since both factors bind to the same element with equal affinities (37). Nucleotide alterations in the GC-richMut oligonucleotide were created to duplicate mutations made in the pCAT-GC-richMut construct. No competition was evident for the GC-richMut oligonucleotide even at the highest concentration utilized (250 M excess). Antibody supershift analysis were employed to determine if the bands that were competed away with the Sp1 consensus element truly represented Sp1-and Sp3-containing complexes (Fig. 8). Similar to the results found in the aforementioned study, the addition of Sp1 antibody resulted in the disappearance of band 1 and the formation of supershifted complexes. Likewise, the addition of Sp3 antibody resulted in the disappearance of bands 2 and 3 and the formation of supershifted complexes. The data from Figs. 7 and 8 demonstrate that the GC-rich element located within the 227-bp conserved domain functions as a recognition element for both Sp1 and Sp3. Furthermore, the lack of competition with the GC-richMut oligonucleotide suggests that any potential Sp1/Sp3 binding at this site would have been abolished in the highly active pCAT-GC-richMut construct. DISCUSSION The goal of this study was to identify cis-elements and transacting factors important for transcriptional regulation of the SM-MHC gene. By comparing the 5Ј-flanking sequences of the rat, mouse, and rabbit SM-MHC genes, we identified two domains that were conserved among all three species. The most proximal domain encompassed the TATA box and contained two conserved CCTCCC elements. This region was previously studied in mouse, and the CCTCCC elements were shown to be necessary for the basal promoter activity ascribed to this domain (23). Therefore, we focused our studies on the distally conserved domain located between nucleotides Ϫ1095 and Ϫ1321 in rat. From deletion analysis, we determined that inclusion of this 227-bp domain was necessary for maximal expression of the SM-MHC gene in cultured rat aortic SMCs. We further demonstrated that several of the known cis-elements present within this domain were important for SM-MHC expression, both as positive and negative regulators.
The presence of this 227-bp domain failed to enhance transcription in either rat L6 myoblasts or myotubes. This result suggests that the positive effects attributable to this region are potentiated in a cell type-specific manner. This observation is in agreement with an earlier SM-MHC promoter mapping study by Kallmeier et al. (21). They demonstrated that a rabbit SM-MHC construct containing 1332 bp of 5Ј-flanking DNA (including the 227-bp conserved region) was highly active in primary cultures of rabbit aortic SMCs (6-fold over the basal activity level of the Ϫ112 promoter construct). This Ϫ1332 construct was only minimally active (basal level activity) in five other non-SMC types. However, a complete review of the literature indicates that there exist several discrepancies with respect to certain SM-MHC deletion constructs and their activity levels in both SMC and non-SMC types. Katoh et al. (20) analyzed the rabbit promoter in primary cultures of rat aortic SMCs and found that Ϫ509 and Ϫ1392 deletion constructs were both at ϳ25% relative CAT activity when compared with the Ϫ2266 construct (100% CAT activity), their most active construct. All of the constructs tested were only minimally active in other non-SMCs. In contrast, White and Low (22) analyzed the rat SM-MHC promoter in primary cultures of rat aortic SMCs and reported that a Ϫ1249 deletion construct was the most active construct, exhibiting twice the level of activity of the next most active construct, Ϫ602 (26-versus 12-fold activity over promoterless plasmid). More important, they found that with the addition of 68 bp of more 5Ј-flanking sequence (the Ϫ1327 construct), transcriptional activity decreased to one-half that of the Ϫ602 construct. A significant decrease in activity was also noted when the Ϫ1249 and Ϫ1327 constructs were compared in rat L8 myoblasts. The noted high level of activity (150-fold over promoterless plasmid) of the Ϫ1249 construct in L8 myoblasts and the subsequent drop in activity (barely detectable levels) with the addition of 68 bp of 5Ј-flanking sequence led the authors to conclude that this domain was likely to contain a repressor element(s) necessary for cell type-specific expression. In another study, Watanabe et al. (23) reported that a mouse Ϫ188 deletion SM-MHC promoter construct, when tested in primary cultures of rabbit aortic SMCs, was the most active construct. The addition of up to 3 kilobase pairs of more 5Ј-flanking sequence only led to a reduction in transcriptional activity. No increase in activity was observed with inclusion of CArG boxes 1 and 2.
It is difficult to reconcile the discrepancies in the data presented in the above studies, not only with respect to our data, but also with respect to each other. Particularly perplexing is the result in mouse, where maximal activity in SMCs was achieved with a Ϫ188 construct. One possible explanation for many of the observed differences between these studies may be differences in the phenotypic state of the SMC in which the activity of the SM-MHC promoter was assessed. It is well documented that SMCs undergo rapid phenotypic modulation when placed in culture (38). In particular, there is an immediate and significant decrease in expression of the SM-MHC gene (39). However, we have demonstrated that rat aortic SMCs used in our studies re-express the SM-MHC gene upon reaching confluency (39). Western blot analysis of the SMCs utilized in this study showed that, although there was a marked de-FIG. 6. Binding of SM nuclear proteins to the three SM-MHC CArG motifs. A, radiolabeled 20-bp oligonucleotides encompassing CArG box 1, 2, or 3 or the SRE were incubated with 10 g of SMC nuclear extract only (lanes 1-4)  crease in SM-2 protein, the SM-1 isoform was present in confluent cells up to passage 22, the last passage analyzed. 2 Moreover, we have shown by protein electrophoretic analysis, Northern analysis, and nuclear run-on analysis (39, 40) 3 that the endogenous SM-MHC gene is actively transcribed in our SMC cultures under the exact transfection conditions used in the present study. In all but one of the studies cited above (22), it was not made clear if the endogenous SM-MHC gene was determined to be active at the time when the SM-MHC promoter-reporter gene fusion constructs were analyzed. Based on our observations, it is possible that differences in confluency at the time of transfection and harvest and differences related to culture methodology could account for the noted differences in expression. Additionally, inherent phenotypic differences between cultured rat and rabbit SMCs may represent another possible explanation. Despite several attempts, we have been unable to culture rabbit aortic SMCs such that they continue to express their endogenous SM-MHC gene, even at early passages. Potential effects due to phenotypic differences become even more plausible in view of the observation of White and Low (22) that certain SM-MHC deletion constructs were differentially expressed in SMCs isolated from either rat aortic or tracheal tissues. Finally, sequence differences (resulting from either construct design or species divergence) in the deletion constructs being compared could certainly explain some of the differences in expression. However, it is doubtful that this explanation could account for the mouse data, where the addition of CArG boxes 1 and 2 or any other 5Ј-flanking sequence failed to enhance transcription above the level of the Ϫ188 construct. Clearly, discerning the many factors likely to be involved in the expression of the SM-MHC gene in cultured SMCs will require further careful study in which each of the potential mitigating factors is analyzed.
This study demonstrated through site-directed mutagenesis that SM-MHC CArG boxes 1 and 2 functioned as positiveacting cis-elements. We further demonstrated that both CArGs, in the presence of aortic SMC nuclear extract, formed protein-DNA complexes that contained a factor antigenetically related to SRF. These data provide evidence that CArG-SRF interaction within the vascular SMC contributes significantly to the expression of the SM-MHC gene. Our laboratory previously demonstrated by site-directed mutagenesis that two CArG boxes located at nucleotides Ϫ112 and Ϫ62 in the rat SM ␣-actin promoter were absolutely required for transcriptional activity in cultured SMCs (24). Li et al. (41) further demonstrated that deletion of a region containing two CArG boxes in the mouse SM22␣ gene (another highly specific SMC marker) resulted in a significant drop in transcriptional activity in cultured SMCs. Thus, similar to several cardiac-and skeletal muscle-specific genes (42)(43)(44), CArG-SRF interaction appears to serve as a common transcriptional activation mechanism for multiple SM-specific genes.
It was somewhat surprising that mutation of the highly conserved CTF/NF1 site present within the 227-bp domain did not result in any significant alteration in activity. However, several studies indicated that, at least in some circumstances, CTF/NF1 functions to regulate transcription via a mechanism FIG. 7. Binding of SM nuclear proteins to the GC-rich repressor element. A radiolabeled oligonucleotide encompassing the GC-rich element was incubated with 10 g of SMC nuclear extract and subjected to EMSA in the presence of different unlabeled competitors. Competitor oligonucleotides (wild-type GC-rich, GC-richMut, and Sp1 consensus) were added at 10 -250 M excesses relative to the radiolabeled DNA. The positions of the three protein-DNA complexes (labeled 1-3) are indicated by arrows. that is likely to involve higher order chromatin structures (45,46). Such an activity would likely be missed in a transient transfection reporter gene analysis where "naked" plasmid DNA is utilized (35).
Another important regulatory element identified in this study is the GC-rich repressor that was recognized by both Sp1 and Sp3. Multiple mechanisms can be envisioned whereby Sp1 and/or Sp3 binding could inhibit transcription. The fact that the GC-rich element is positioned adjacent to CArG box 2 suggests that it may function as a negative regulator by preventing SRF binding to the CArG2 element in a manner similar to YY1-SRF interaction in the cardiac ␣-actin gene (47). Also, in contrast to Sp1, a number of recent promoter studies indicate that binding of Sp3 appears to negatively regulate transcription (48 -50). It will be of interest in the future to investigate how the elements identified in this study, as well as their potential binding factors, are involved in the altered expression of the SM-MHC gene in phenotypically modified SMCs found in atherosclerotic lesions in man (51).