Atherosclerosis, which accounts for a large proportion of the morbidity and mortality associated with cardiovascular disease(1, 2) , is characterized by the abnormal growth and phenotypic modulation of vascular smooth muscle cells (SMCs)()(3) . Intimal SMCs within atherosclerotic lesions of man, as well as experimental animals, show diminished expression of a number of proteins that are characteristic of fully differentiated SMCs, including SM -actin and SM myosin heavy chain(4, 5) . As such, an understanding of the mechanisms that control SMC differentiation is likely to be important for understanding atherogenesis.

A key to understanding SMC differentiation is to identify factors that regulate the coordinate expression of cell-specific genes that distinguish one cell type from another. Studies of skeletal muscle development have lead to the characterization of a number of ``master control'' genes that regulate skeletal muscle differentiation or lineage determination, including myoD and related members of the helix-loop-helix family(6, 7) . These factors have been shown to activate the expression of genes characteristic of mature skeletal muscle cells by functioning as transcription factors that bind to a specific DNA-binding motif, referred to as an E-Box, that is found in the 5`-flanking regions of a number of skeletal muscle-specific genes (7) . More recently, Tontonoz et al.(8) identified a helix- loop-helix transcription factor that appears to regulate differentiation-specific gene expression in adipocytes. Thus, it is likely that the expression of at least some SM-specific genes and the molecular mechanisms involved in SMC differentiation rely on analogous transcriptional regulatory systems, although no SM-specific regulatory/transcription factors have as yet been characterized.

The principle function of mature, fully differentiated SMCs is contraction; therefore, genes encoding proteins involved in SM contraction, such as -actin, myosin heavy chain, myosin light chain, caldesmon, vinculin, calponin, and metavinculin, are candidates to study transcriptional regulatory systems in SMC differentiation. The SM isoforms of myosin light chain, caldesmon, vinculin, and metavinculin are produced by alternative splicing of gene transcripts expressed in a variety of cell types(9, 10, 11, 12, 13) . Thus, these genes are not likely to be suitable models to study SM-specific transcriptional regulatory systems. The SM myosin heavy chain gene shows a high degree of SM-specific expression(14) ; however, sequence information on this promoter has not been reported. Currently, one of the best genes to begin to characterize transcriptional cis-elements and trans-factors involved in control of SMC differentiation is SM -actin, which is the most abundant of the SM contractile proteins in mature vascular SMC(15) . Although SM -actin is transiently expressed in cardiac (16, 17) and skeletal (17) muscle during development and in myofibroblasts within tumors and healing wounds (18) , it is exclusively expressed in SMCs and SM-related cells in the normal adult animal(19, 20, 21) . Additionally, the expression of SM -actin is modulated in proliferating SMCs found within human atherosclerotic lesions(4, 22, 23) .

Studies of the chicken, mouse, and human SM -actin genes have been performed previously and have demonstrated the importance of cell context in the characterization of promoter elements that function in transcriptional regulation(24, 25, 26, 27) . Studies of the chicken SM -actin gene demonstrated that the first 122 bp of the promoter were sufficient to confer a moderate to high level of transcriptional activity in chicken SMCs, skeletal myoblasts, and fibroblasts and that the addition of a 29-bp promoter segment (to -151) restricted expression in fibroblasts and abolished expression in skeletal myoblasts. However, this same segment induced a doubling of activity in SMCs(24, 27) . Studies utilizing the mouse SM -actin promoter transfected into mouse AKR-2B embryonic fibroblasts and subconfluent mouse skeletal myoblast-like BC3H1 cells demonstrated that the region between -224 and -192 abolished expression of the first 191 bp of the promoter in both cell types. However, site-directed mutagenesis studies showed that the negative-acting cis-element within the -244 to -192 region that abolished expression in BC3H1 cells was different from the negative element in fibroblasts(25) . These studies, therefore, demonstrated that the regulatory elements responsible for transcriptional expression of the SM -actin gene are highly dependent upon the cell type in which the promoter is studied.

The SM -actin promoter contains a number of highly conserved putative regulatory elements. One such element is the CArG box, which has the general sequence motif CC(A/T-rich)GG. The 5`-flanking regions of both the skeletal and cardiac -actin genes contain conserved CArG box elements that have been shown to direct developmental and tissue-specific expression of these genes(28, 29, 30, 31) , and CArG elements have also been implicated in the regulation of the myosin light chain(32, 33) , muscle creatine kinase(34) , and dystrophin genes(35) . Additionally, a CArG box motif is the core component of the serum response element (SRE), a DNA element that is required for the transient transcriptional response of many nonmuscle-specific immediate-early response genes, such as c-fos and actin, upon serum or growth factor stimulation(36, 37) . Although previous studies have demonstrated that the SM actin CArG elements can function in serum-inducible expression of the gene in fibroblasts(38, 39) , the expression of the endogenous SM -actin gene is not serum-inducible in SMCs, even though -actin expression is serum-stimulated under the same conditions(40) . Indeed, although previous studies have shown that promoter regions containing the CArG boxes were transcriptionally active in SMCs(26, 27) , no direct evidence for a regulatory role of the SM -actin CArG elements in SMCs has been reported.

The goals of the current study were to identify cis-acting DNA elements within the rat SM -actin promoter that govern SM-specific transcriptional expression, to determine whether the CArG motifs play an important role in the regulation of SM -actin expression in SMCs, and to determine whether there is evidence for the binding of SM nuclear proteins to these regulatory sites. Although not a goal of the present study, the long term objective is to determine if these SM nuclear proteins act as transcription factors that not only regulate the transcription of SM -actin but also function as global regulators of other SM-specific genes; i.e. function as master regulatory factors. The results of these studies provide evidence that 1) the cell-specific expression of SM -actin is due to a complex combination of both negative and positive acting cis-elements found within the first 547 bp of the promoter; 2) the CArG box elements, along with additional 5`- or 3`-flanking regions, are important in the tissue-specific positive activation of SM -actin; and 3) the SM-specific trans-factors that bind the CArG boxes include serum response factor (SRF) or an SRF-like protein. These studies also support the utility of the SM -actin promoter in understanding the molecular regulation of SMC differentiation and also underscore the potential usefulness of this promoter in expressing bioactive molecules in a SMspecific manner.

MATERIALS AND METHODS

Construction of Promoter-CAT Expression Plasmids

Site-directed CArG mutations were generated using the Altered Sites in vitro mutagenesis system (Promega) according to the manufacturer's recommendations. The mutated promoter fragments were then PCR amplified, subcloned into pCAT-Basic, and sequence verified as described above.

All promoter-CAT plasmid DNAs used for transient transfections were prepared using an alkaline lysis procedure (42) followed by banding on two successive ethidium bromide-cesium chloride gradients. Each plasmid preparation was examined by electrophoresis on 1% agarose gels, and preparations were judged acceptable if >50% of the DNA was supercoiled and if the relative amount of supercoiled to nicked plasmid DNA was approximately the same for all constructs used in the same set of transfections. Each promoter construct is designated by a ``p'' followed by a number that indicates the 5`-most promoter sequence relative to the transcription start site. The 3` end of each promoter fragment ends at position +20 within the first exon.

Cell Culture, Transient Transfections, and Reporter Gene Assays

Cells were seeded for transient transfection assays into 6-well plates (Corning Glass; Corning, NY) at a density of 2 10 cells/cm for SMCs and L6 skeletal myoblasts, 3 10 cells/cm for BAECs, and 4 10 cells/cm for RAECs. These densities were chosen so that the cells would be at 60-80% confluency at the time of transfection (30 h after plating). Transient transfection experiments consisted of transfecting each promoter-CAT plasmid in triplicate using Transfectam reagent according to the manufacturer's recommendations (Promega). Briefly, for each well to be transfected, 5 µg of a plasmid construct (in 10 µl of 10 mM Tris (pH 8.0), 1 mM EDTA) was incubated for 10 min in a tissue culture-sterile microcentrifuge tube with 10 µl of Transfectam reagent and 380 µl of transfection media. During the incubation, each well was washed twice with 3 ml of transfection medium to remove residual serum, and after the second media wash was removed, 250 µl of fresh transfection medium was added to each well. After incubation, 400 µl of DNA/Transfectam/media solutions were added to each well. The transfection media consisted of the following: a 1:1 formulation of Dulbecco's modified Eagle's medium and Ham's F-12 medium (Life Technologies, Inc.) for SMCs, Dulbecco's modified Eagle's medium for L6 skeletal myoblasts and myotubes, MCDB 105 for RAECs, and Waymouth's basal medium (Life Technologies, Inc.) for BAECs. Each of the transfection mediums also contained 0.68 mML-glutamine, 100 units/ml penicillin, and 100 µg/ml streptomycin. At 14 h post-transfection, the cells were changed to the cell-specific culture medium described above. In the case of skeletal myotubes, culture/differentiation medium was the same as for the skeletal myoblasts except that only 1% FBS was included. After 48 h in culture medium, the cells had reached confluency and were harvested by scraping in harvesting buffer (30 mM Tris, 6 mM EDTA, 150 mM NaCl (pH 7.5)), pelleted, resuspended in 100 µl of cold 100 mM Tris (pH 7.5), and stored at -70 °C.

Cell extracts were prepared by four cycles of freezing in a dry ice/ethanol bath, thawing, and vigorous vortexing followed by centrifugation to remove cellular debris. A 60-µl aliquot of each cell extract was heat-inactivated, and 55 µl of each heat-inactivated extract was transferred to a fresh microfuge tube and was assayed for CAT activity by enzymatic butyrylation of tritiated chloramphenicol (DuPont NEN) as described previously(49) . All CAT activity values were normalized to the protein concentration of each cell lysate as measured by the Bradford assay (50) using Bio-Rad's protein assay reagent (Bio-Rad). Early transfection experiments utilized a -galactosidase plasmid as a co-transfection partner to measure transfection efficiency. However, the -galactosidase activity measurements did not result in qualitative changes in the data nor did they affect experimental variability. Since the co-transfections conferred no advantage in the reduction of experimental error and had the potential of competing for limited trans-factors that regulate SM -actin transcription, the -galactosidase co-transfections were discontinued. In each experiment, the promoterless pCAT-Basic 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 one. Additionally, a SV40 enhancer/promoter-CAT construct (Promega) served as a positive control of transfection and CAT activity. All CAT activity values represent at least three independent transfection experiments with each construct tested in triplicate per experiment. Relative CAT activity data are expressed as the means ± standard error computed from the results obtained from each set of transfection experiments. One-way analysis of variance followed by the Newman-Keuls' multiple range test were used for data analysis. Values of p < 0.05 were considered statistically significant.

Preparation of Nuclear Extracts, in Vitro Synthesis of SRF, Electrophoretic Mobility Shift Assays (EMSA), and Cross-linking Assays

The 95-bp promoter segments were generated by PCR amplification using wild-type or CArG box-mutated promoter-CAT plasmids as the templates and PCR primers containing EcoRI restriction sites. After the reactions were chloroform extracted, the PCR products were EtOH precipitated, and EcoRI restriction endonuclease (Promega) was digested and gel-purified in a 1% agarose gel using the Magic PCR prep DNA purification system (Promega). The EcoRI 5`-overhangs were filled in using Klenow enzyme (Boehringer-Mannheim), [-P]dATP (6000 Ci/mmol) (DuPont NEN), and unlabeled dCTP, dGTP, and dTTP (Promega). The unincorporated nucleotides were removed using NucTrap Push columns (Stratagene) as recommended by the manufacturer. After fill-in labeling, the 95-bp fragments included nucleotide positions -137 to -43.

CArG oligonucleotides for EMSA contained 5 bp 5` and 3` of each CArG element (20 bp total). Mutant CArG oligos involved changing both the flanking CCs and GGs of the CArG motif (CC[AT]GG) to AAs. Both oligonucleotides of a double-stranded duplex were end-labeled separately with T4 polynucleotide kinase (Promega) and [-P]ATP (6000 Ci/mmol) (DuPont NEN) and annealed together. Unincorporated nucleotides were removed as described above.

EMSA were performed with 20 µl of binding reaction that contained 0.24 ng of P-labeled 95-bp fragment or 30 pg of P-labeled annealed oligonucleotides, 5 or 10 µg of crude nuclear extract (in 5 µl of Dignam buffer D(51) ), or 2.5 µl of in vitro synthesized SRF, 35 mM Tris (pH 7.5), 5 mM HEPES, 200 mM KCl, 1.125 mM dithiothreitol, 1.05 mM EDTA, 0.25 µg of poly(dI-dC), 10% glycerol, and competitor oligonucleotides where indicated. All binding reactions were incubated for 30 min at room temperature. When affinity-purified anti-SRF antibodies (generous gifts from Ravi P. Misra and Michael E. Greenberg; Harvard Medical School; Boston, MA) were included in the EMSA, they were also incubated in binding reactions for 30 min as described above except without the radiolabeled DNA. The radiolabeled DNA was subsequently added and incubated for an additional 30 min. All binding reactions were then loaded on 5% polyacrylamide gels (30:1 acrylamide/bis-acrylamide (Bio-Rad)), which had been prerun at 170 V for 30 min and electrophoresed at 170 V in 0.5 TBE (45 mM Tris borate, 1 mM EDTA). The gels were then dried and exposed to Kodak X-Omat LS film at -70 °C in the presence of an intensifying screen. The binding reactions for UV cross-linking experiments were performed in a similar manner except that 7.5 µg of SMC nuclear extract or 4 µl of in vitro synthesized SRF were used per 20-µl reaction. Following electrophoresis in 5% polyacrylamide gels, the gels were UV irradiated at a wavelength of 254 nm and a distance of 6 cm for 30 min at 4 °C and exposed to film at 4 °C overnight. The bands corresponding to the protein-DNA complexes of interest were excised from the gels, placed in 50 µl of sample buffer (50 mM Tris (pH 6.8), 1% SDS, 25 mM dithiothreitol, 10% glycerol, and 0.001% bromphenol blue tracking dye), and electrophoresed on one-dimensional SDS-polyacrylamide gels as described previously(15) . The gels were then dried and autoradiographed as described above.

RESULTS

Negative-acting cis-Elements Are Required for Cell-type Specific Expression of Smooth Muscle -Actin

Figure 1: A, structures of the rat SM -actin promoter-CAT reporter gene constructs. Promoter sequences were PCR amplified and ligated into a plasmid containing a promoterless CAT gene. The 5`-most nucleotide position is indicated above the diagram of each chimera. The relative positions of putative cis-acting regulatory regions are also indicated. B, expression of SM -actin promoter-CAT constructs in adult rat aortic smooth muscle cells. Subconfluent cultures were transiently transfected with the plasmid constructs pictured in A. CAT activity was expressed relative to the base-line CAT activity of a promoterless-CAT construct set to one. An SV40 enhancer/promoter-CAT construct served as a positive control for transfection and CAT activity. The data represent six independent experiments, each experiment performed with triplicate samples per plasmid construct.

To identify cis-elements responsible for cell-type specific regulation of SM -actin, the truncated promoter-CAT constructs were also transfected into non-SMC lines. These cell lines included aortic endothelial cells, which do not express endogenous SM -actin based on both Northern analysis as well as evaluation of SM -actin protein expression by two-dimensional gel electrophoresis and immunostaining with SM actin-specific antibody (data not shown), and L6 skeletal myoblasts, which also do not express SM -actin but which stimulate endogenous SM -actin expression when induced to differentiate into myotubes ((53) and data not shown). In skeletal myoblasts, the shorter promoter constructs (p125CAT to p371CAT) only modestly activated the transcription of the CAT reporter gene (up to 6-fold above promoterless-CAT), while the longer promoter constructs (p547CAT to pPromCAT) failed to elicit any increase in CAT expression over that of the promoterless-CAT construct (Fig. 2A). These data suggest that the promoter contains an element(s) within the region between -547 and -371 that represses transcriptional expression of SM -actin in skeletal myoblasts. This contrasts with results seen in SMCs in which there was a prominent CAT activity produced by the shorter promoter constructs as well as by p547CAT to pPromCAT.

Figure 2: Expression of SM -actin promoter-CAT constructs in rat L6 skeletal myoblasts and myotubes (A) and rat and bovine aortic endothelial cells (B). Subconfluent cultures were transiently transfected with the plasmid constructs pictured in Fig. 1A. CAT activity was expressed relative to the base-line CAT activity of a promoterless-CAT construct set to one. An SV40 enhancer/promoter-CAT construct served as a positive control for transfection and CAT activity. The skeletal myoblast, myotube, and RAEC data represent three independent experiments, and the BAEC data represent five independent experiments, each experiment performed with triplicate samples/plasmid construct.

When the truncated promoter-CAT constructs were transiently transfected into skeletal myoblasts that were then induced to differentiate into myotubes (Fig. 2A), the results again differed from that seen in SMCs in that the shortest promoter constructs (p125CAT to p208CAT) moderately activated transcription (up to 9-fold), similar to the level of activity seen in skeletal myoblasts. However, upon addition of the region between -271 and -208 (p271CAT), which contains two putative E-Box motifs, CAT expression increased 18-fold compared with promoterless-CAT in skeletal myotubes but had no affect in skeletal myoblasts or SMCs. Additionally, the longest promoter-CAT constructs retained transcriptional activity (up to 20-fold above promoterless-CAT) in skeletal myotubes, similar to that seen in SMCs. The differential expression of the full-length promoter (pPromCAT) between SMCs, skeletal myoblasts, and skeletal myotubes correlated with endogenous SM -actin mRNA levels in these cells in that SMCs and skeletal myotubes expressed a high level of SM -actin mRNA, as determined by Northern analysis, while skeletal myoblasts did not (data not shown).

Cell-type specific expression of the truncated promoter-CAT constructs was further tested in RAECs, another cell line that does not express its endogenous SM -actin gene. Surprisingly, p125CAT exhibited high CAT activity (up to 18-fold above promoterless-CAT). This was subject to negative regulation with addition of upstream promoter sequences (Fig. 2B). However, in contrast to the activity observed in SMCs, the longer promoter-CAT constructs (p547CAT to pPromCAT) had no activity above the promoterless-CAT level in RAECs. Studies were repeated in BAECs, which also did not express their endogenous SM -actin gene at the mRNA or protein level (data not shown). As shown in Fig. 2B, the general pattern of promoter activities in BAECs was the same as seen in RAECs. The similarity between RAECs and BAECs suggests that the mechanisms and trans-factors that regulate SM -actin expression in RAECs and BAECs are likely to be the same. Since RAECs were more difficult to transfect and more costly to grow in culture than BAECs, subsequent studies requiring endothelial cells utilized BAECs.

The transfection results in endothelial cells demonstrated that there are negative-acting cis-elements within the promoter regions -547 to -371, -208 to -271, and -125 to -155 that repress expression of SM -actin in these cells. These results are similar to those obtained in skeletal myoblasts in which the larger promoter constructs were not transcriptionally active. In contrast, in SMCs and skeletal myotubes, which transcribe their endogenous SM -actin gene, there appeared to be differential regulation of the gene since the first 125 bp of the promoter were sufficient to express a very high level of transcriptional activity in SMCs but not in skeletal myotubes. Taken together, these data demonstrate that the first 547 bp of the promoter are sufficient to confer tissue-specific expression of the SM -actin gene in these cultured cell types.

The CArG Boxes Function in the Cell-specific Activation of SM -Actin Transcription

Figure 3: Smooth muscle -actin promoter sequence. This figure compares the 5`-flanking regions of rat (27) , mouse(25) , human(84) , and chicken (85) SM -actin genes. Nucleotides conserved with the rat sequence are indicated by dashedlines. Spaces indicate gaps introduced in the sequence to optimize the alignment. Numbers in the right margin refer to the position of the last nucleotide in each line relative to the transcriptional start site. Putative regulatory sequences that are conserved between at least two of the depicted species are boxed. This figure was previously published by Blank et al.(27) except for the nucleotide corrections in the rat sequence in the region between -161 and -159.

Figure 4: A, CArG mutations introduced into the rat SM -actin promoter-CAT reporter constructs. Site-directed mutations were introduced using Altered Sites in vitro mutagenesis system. The relative positions of mutated CArG box elements are also indicated. B, effects of CArG mutations in SM -actin p125CAT constructs transfected into rat aortic SMCs and endothelial cells. Subconfluent cultures were transiently transfected with CArG mutant constructs pictured in Fig. 4A. CAT activity was expressed relative to the base-line CAT activity of a promoterless-CAT construct set to one. The SMC data represent four independent experiments, and the EC represent three independent experiments, each experiment performed with triplicate samples/plasmid construct. Similar results were obtained with mutation of the CArG motifs in the p271CAT construct (data not shown).

To determine whether mutation of CArG-A or CArG-B would also abolish the transcriptional activity induced by the full-length promoter, the same site-specific CArG mutations were created in the pPromCAT construct. This resulted in approximately a 50% reduction in the CAT activity in SMCs (Fig. 5), demonstrating that while both CArG-A and CArG-B were essential for p125CAT to be transcriptionally active in SMCs, there is likely an additional upstream cis-element(s) required for full activity of the full-length promoter. This additional positive-acting cis-element(s) may be upstream of -699 of the promoter since activity of pPromCAT had significantly greater activity (p < 0.05) than p699CAT (Fig. 1B). Consistent with this, in studies with the human SM -actin gene transiently transfected into rat SMCs, Nakano et al.(26) found suggestive evidence for a positive-acting upstream regulatory element in the region between -891 and -597 of the human promoter.

Figure 5: Effects of CArG mutations in SM -actin pPromCAT constructs transfected into SMCs and skeletal myotubes. Subconfluent cultures were transiently transfected with pPromCAT or pPromCAT constructs mutated in CArG-A, CArG-B, or both. CAT activity was expressed relative to the base-line CAT activity of a promoterless-CAT construct set to one. The data for each cell line represents three independent experiments, each experiment performed with triplicate samples/plasmid construct.

The pPromCAT/CArG mutant constructs were also studied in skeletal myotubes, which express their endogenous SM actin gene. Results demonstrated that mutation of either CArG box reduced transcriptional activity of pPromCAT in myotubes by approximately 50% (Fig. 5). Therefore, the CArG boxes appear to be involved in the transcription of SM -actin in both SMCs and skeletal myotubes. However, the mode of regulation via the CArG box elements is different in the two cell-types. Whereas the first 125 bp of the promoter were sufficient for a high level of transcription in SMCs (Fig. 1B) and the activity was completely dependent upon the CArG boxes in SMCs (Fig. 4B), the same 125-bp promoter segment induced only modest activity in skeletal myotubes (Fig. 2A). This implies that transcriptional activation via the CArG elements requires an interaction with sequence elements upstream of -125 in skeletal myotubes but not in SMCs. The upstream element(s) required for expression in skeletal myotubes may reside between -271 and -208, which contains two putative E-Boxes, since addition of this region was necessary for high levels of transcriptional activation (Fig. 2A). Therefore, although the CArG box elements functioned in the transcriptional regulation of SM actin in both SMCs and skeletal myotubes, there was a qualitative difference in the regulation of transcription via the CArG box elements in these cell-types, and this difference is likely to be due to a complex array of cell-specific cis-elements and trans-factors.

CArG Boxes Interact with SM Nuclear Factors

Figure 6: A, structure of rat SM -actin promoter segments used in electrophoretic mobility shift assays. Oligonucleotide primers with engineered EcoRI restriction sites were used to PCR amplify the 95-bp promoter sequences for subsequent restriction and Klenow enzyme fill-in with [-P]dATP. The 20-bp CArG box oligonucleotides were end-labeled with [-P]ATP and T4 polynucleotide kinase and annealed to form double-stranded duplexes. The 5`- and 3`-most nucleotide positions are indicated above the diagram of each promoter segment. The relative positions of mutated CArG box elements are also indicated. B, binding of SMC nuclear proteins to the wild-type 95-bp SM -actin promoter segment. The radiolabeled 5`-flanking DNA segment from -137 to -43 (*WT), which contained wild-type CArG-A and CArG-B sequences, was incubated with 5 µg of SMC nuclear extract. Competition reactions were performed with 20-bp double-stranded oligonucleotide duplexes of the CArG-A element (wtA), CArG-B element (wtB), or their mutations (mA and mB, respectively). Competitor oligonucleotides were added at a 50-400-fold molar excess relative to the radiolabeled DNA. The positions of the three principle nucleoprotein/DNA complexes (labeled 1-3) are indicated by the arrows.

Incubation of the wild-type 95-bp fragment (*WT) with SMC NE resulted in the formation of three shift bands as well as a fourth band below band 3 with some SMC NE preparations (Fig. 6B, lane1). Addition of unlabeled 20-bp double-stranded oligonucleotide duplexes homologous to CArG-A (wtA) or CArG-B (wtB) sequences competed for binding of nuclear factors and inhibited the formation of bands 1 and 2 (Fig. 6B, lanes2-6 and 9-13, respectively). In contrast, mutant double-stranded CArG-A and CArG-B oligonucleotide duplexes (mA and mB, respectively) failed to compete (lanes7-8 and 14-15). This demonstrates that SMC nuclear proteins are binding the CArG-A and CArG-B elements leading to shift bands 1 and 2. This contrasts with band 3 (and 4) that appeared to represent binding to other areas of the 95-bp segment since these bands were not competed for by CArG box oligonucleotides.

As a further test of CArG box interaction with SMC nuclear proteins, the 95-bp DNA segment containing mutations of CArG-A and/or CArG-B were used in EMSA. Mutation of CArG-A preferentially decreased formation of shift band 1 relative to shift band 2, and competition with double-stranded CArG-A and CArG-B oligonucleotides inhibited formation of shift band 1 but not band 2 (Fig. 7A). In a similar manner, mutation of CArG-B by itself or of CArG-A and CArG-B together nearly abolished shift band 1 formation, and competition with double-stranded CArG-A and CArG-B oligonucleotides had no additional affect on any of the shift bands (Fig. 7B). For reasons that are not clear, CArG oligonucleotide competitors did not compete for formation of band 2 when using the 95-bp DNA segments containing mutations of CArG-A or -B alone or together, although they did compete for this band using the wild-type 95-bp DNA segment (Fig. 6B).

Figure 7: Binding of SMC nuclear proteins to the 95-bp SM -actin promoter segment containing mutations of CArG-A (A) or mutations of CArG-B or CArG-AB (B). Radiolabeled 5`-flanking DNA segment from -137 to -43 that contained a mutated CArG-A (*mA), a mutated CArG-B (*mB), or both (*mAB) was incubated with 5 µg of SMC nuclear extract. The wild-type 95-bp segment (*WT) was also included in these experiment as a comparison (lanes1). Competition reactions were performed with 20-bp double-stranded oligonucleotide duplexes of the CArG-A element (wtA), CArG-B element (wtB), or their mutations (mA and mB, respectively). Competitor oligonucleotides were added at a 50-400-fold molar excess relative to the radiolabeled DNA. The positions of the three principle nucleoprotein/DNA complexes (labeled 1-3) are indicated by the arrows.

EMSA experiments using P-labeled 20-bp double-stranded CArG-A and CArG-B oligonucleotide duplexes (Fig. 6A) were also performed to determine whether these short CArG fragments are sufficient to bind nuclear factors and whether the factors that bind the different CArG boxes are equivalent. Incubation of the wild-type CArG-B oligonucleotide duplex (*wtB) with SMC NE produced two slowly migrating shift bands, labeled A and B, that were both specifically competed away with increasing amounts of CArG-A and CArG-B competitor oligonucleotides (Fig. 8, lanes2-4 and 7-9; also see Fig. 10A, lane3) but not with mutant competitors (Fig. 8, lanes5-6 and 10-11). Shift bands A and B were also formed with the wild-type CArG-A oligonucleotide duplex (*wtA) (see Fig. 10A, lane1), although their intensities were less. These were also inhibited by CArG oligonucleotide competitors (data not shown). In contrast, shift bands A and B did not form when the mutated CArG-A and CArG-B oligonucleotide duplexes (*mA and *mB) were used (see Fig. 10A, lanes2 and 4), further demonstrating the CArG specificity of bands A and B. Reaction of the CArG oligonucleotides with SMC NE also resulted in the formation of two faster migrating shift bands/complexes (labeled C and D). However, these shift bands/complexes appeared to represent nonspecific binding since mutation of CArG-A or CArG-B (*mA and *mB) did not abolish formation of these shift bands (see Fig. 10A) and competitions with unlabeled CArG oligonucleotides did not effectively decrease their formation (Fig. 8). Thus, these results confirm that SMC nuclear proteins can bind specifically to both CArG box elements, consistent with EMSA using the 95-bp segments. Additionally, the results shown in Fig. 6B and Fig. 8also revealed that lower concentrations of CArG-B competitor were needed to inhibit the formation of CArG-specific shift bands, suggesting that CArG-B has a greater affinity for the CArG binding factor(s) than does CArG-A. The differences in shift band intensities between *wtA and *wtB were not due to a difference in the specific activities of those oligonucleotides since the difference in shift band intensities were observed even when the CArG-A oligonucleotide duplex had a higher specific activity.

Figure 8: Binding of SMC nuclear proteins to the wild-type CArG-B. Radiolabeled wild-type 20-bp double-stranded CArG-B (*wtB) oligonucleotide duplex was incubated with 5 µg of SMC nuclear extract. Competition reactions were performed with unlabeled 20-bp double-stranded oligonucleotide duplexes of the CArG-A element (wtA), CArG-B element (wtB), or their mutations (mA and mB, respectively). Competitor oligonucleotides were added at a 25-400-fold molar excess relative to the radiolabeled DNA. The positions of four nucleoprotein/DNA complexes (labeled A-D) are indicated by the arrows.

Figure 10: A, binding of SMC and EC nuclear proteins to the wild-type and mutated 20-bp CArG-A and CArG-B elements and to the 22-bp SRE CArG element. Radiolabeled wild-type and mutated 20-bp double-stranded CArG-A (*wtA and *mA, respectively), CArG-B (*wtB and *mB, respectively), and the 22-bp SRE CArG (*SRE) oligonucleotide duplexes were incubated with 10 µg (for CArG-A duplexes) or 5 µg (for CArG-B and SRE duplexes) of SMC or EC nuclear extracts. The positions of four nucleoprotein-DNA complexes (labeled A-D) are indicated by the arrows. B, binding of nuclear proteins from skeletal myoblast and myotube cell lines to the wild-type 95-bp SM -actin promoter segment. The radiolabeled wild-type 5`-flanking DNA segment from -137 to -43 (*WT) was incubated with 5 µg of rat SMC NE (lane1), rat L6 skeletal myoblast NE (lane2), or rat L6 skeletal myotube NE (lane3). The arrows labeled 1 and 2 indicate the CArG-specific SMC nucleoprotein complexes. The unlabeled arrows identify the positions of the nucleoprotein complexes formed with L6 skeletal cell line nuclear extracts. C, comparison of SMC nuclear proteins and in vitro synthesized serum response factor bound to the wild-type 95-bp SM -actin promoter segment and the wild-type 20-bp CArG-B element. Lane1, radiolabeled wild-type 95-bp promoter segment (*WT) reacted with 5 µg of SMC nuclear extract; lane2, radiolabeled wild-type CArG-B oligonucleotide duplex (*wtB) with 5 µg of SMC nuclear extract; lane3, *WT incubated with 2.5 µl of SRF in vitro synthesized with rabbit reticulocyte lysate; lane4, *WTincubated with 2.5 µl of unprogrammed (UP) rabbit reticulocyte lysate; lane5, *wtB with 2.5 µl of SRF lysate; and lane6, same as lane1. This gel was run twice as long as the previous gels in order to optimize comparisons of shift band positions. The three principle SMC nucleoprotein complexes that form with *WT are indicated by the arrows labeled 1-3 (compare Fig. 6B). Three of the principle SMC nucleoprotein complexes that form with *wtB are indicated by the arrows labeled A-C (compare Fig. 8B).

SRF or an SRF-like Factor in SMC Nuclear Extracts Binds the CArG Box Elements and May Be Responsible for Transactivation of SM -Actin

Initial studies utilized an affinity-purified polyclonal antibody raised against the amino terminus of SRF (referred to as R1119, a generous gift from Ravi P. Misra and Michael E. Greenberg; Harvard Medical School; Boston, MA). This antibody was previously shown to react with SRF from both murine and human cell lines, as well as with in vitro synthesized human SRF(64) . In EMSA, the antibody reacted with the SMC nuclear factors that form CArG-specific shift bands 1 and 2 with *WT and resulted in at least one supershifted band of slower electrophoretic mobility (Fig. 9A). The anti-SRF antibody did not, however, react with nuclear factors in non-CArG-specific shift band 3. In a similar manner, the migration of CArG-specific shift bands A and B formed with *wtB was also altered when reacted with the same antibody, while bands C and D were unaffected. In contrast, addition of pre-immune serum or non-SRF antibodies did not alter shift patterns (data not shown). A second affinity purified antibody against the amino terminus of SRF (R1122) and immune serum raised against SRF's carboxyl terminus (R1443; also gifts from R.P. Misra and M. E. Greenberg) yielded similar results to those with SRF R1119 antibody (data not shown). Furthermore, UV cross-linking assays demonstrated that the size of the SM CArG box-binding protein binding to the CArG-B element was approximately 67 kDa (data not shown), similar to the known size of SRF(52) . This strongly suggests that SRF, or a protein related to SRF in size and antigenicity, is at least one of the SMC nuclear factors binding the SM -actin CArG elements.

Figure 9: A, binding of the wild-type 95-bp SM -actin promoter segment and the wild-type 20-bp CArG-B element to SMC nuclear proteins that had been reacted with an antibody raised against the amino terminus of serum response factor. The leftside of the figure shows the results with wild-type 5`-flanking DNA segment from -137 to -43 (*WT). The arrows labeled 1-3 indicate the positions of the three principle nucleoprotein/DNA complexes that formed when *WT was reacted with SMC nuclear extract by itself (lane1). The rightside of the figure shows the results with the wild-type CArG-B oligonucleotide duplex (*wtB). The arrows labeled A and B indicate the positions of the two CArG-specific nucleoprotein/DNA complexes that formed when *wtB was reacted with SMC nuclear extract by itself (lane7). The radiolabeled DNAs were incubated with 5 µg of SMC nuclear extract and various dilutions of affinity-purified polyclonal antibodies (referred to as R1119) raised against SRF. The top, unlabeled arrows identify the supershifted nucleoprotein-DNA complexes that formed upon reaction with anti-SRF antibody. B, binding of in vitro synthesized serum response factor to the wild-type 20-bp CArG-B element. Radiolabeled wild-type CArG-B oligonucleotide duplex (*wtB) was incubated with 2.5 µl of SRF in vitro synthesized with rabbit reticulocyte lysate. Competition reactions were performed with unlabeled 20-bp double-stranded oligonucleotide duplexes of the CArG-A element (wtA), CArG-B element (wtB), or their mutations (mA and mB, respectively). Competitor oligonucleotides were added at a 100-400-fold molar excess relative to the radiolabeled DNA. *wtB incubated with 5 µg of SMC nuclear extract was also included in this experiment as a comparison (lane1), and the positions of the two CArG-specific SMC nucleoprotein/DNA complexes (labeled A and B) are indicated by the arrows. Other comparison lanes included *wtBincubated with 2.5 µl of unprogrammed (UP) rabbit reticulocyte lysate (lane13), mutated CArG-B oligonucleotide duplex (*mB) incubated with 2.5 µl of SRF lysate (lane14), and wild-type CArG-A oligonucleotide duplex (*mA) incubated with 2.5 µl of SRF lysate (lane15).

To further investigate the possible involvement of SRF in binding the SM -actin CArG box elements, the binding characteristics of in vitro synthesized SRF were compared with the SMC nuclear proteins. As shown in Fig. 9B, reacting *wtB with in vitro synthesized SRF resulted in shift bands with similar electrophoretic mobilities and CArG box-specific competition properties as SMC NE bands A, B, and C (compare with Fig. 8and 10A). As with CArG-specific shift bands A and B with the SMC NE, the two slowest migrating SRF shift bands also appeared to be CArG box-specific for the following reasons. First, competition reactions with the wild-type CArG-B oligonucleotides inhibited the formation of the two top bands (Fig. 9B, lanes8-10), while the mutant CArG-B oligonucleotides did not (lanes11 and 12). Second, unprogrammed lysate, which was reacted with *wtB (lane13), and SRF lysate, which was reacted with the mutated CArG-B oligonucleotides (*mB) (lane14), did not result in formation of shift band B and formed a greatly diminished shift band A. These results provide additional evidence that SRF is the factor in SMC NE that contributes to shift bands A and B. In contrast, the wild-type CArG-A oligonucleotides were not effective competitors (lanes3-7) nor did *wtA efficiently bind SRF lysate to form bands A or B (lane15). This is also consistent with earlier results demonstrating that CArG-B had a greater affinity for the SMC NE CArG-binding factor(s) than did CArG-A (Fig. 6B, 8, and 10A).

Fig. 10A also demonstrates the results of reacting SMC nuclear extract with a radiolabeled 22-bp double-stranded oligonucleotide duplex conforming to SRF's optimum CArG box binding site (*SRE; lane5)(65) . This resulted in shift bands similar to those formed with labeled SM -actin CArG-B (*wtB; lane3), providing additional evidence that the SM -actin CArG box elements are binding SRF, or SRF-like factor, in SMC nuclear extract.

Evidence for SM-Specific CArG Box-binding Proteins/Complexes

To further characterize whether the same CArG-specific nuclear factors bind the CArG-B oligonucleotide duplex and the 95-bp DNA segment, the electrophoretic mobilities of the CArG-specific shift bands formed with the two DNA segments were compared. In order to optimize comparisons between bands, these gels were run twice as long as previous shifts (see ``Materials and Methods''). Results demonstrated that reaction of the *WT probe with SMC NE resulted in a band with a slower electrophoretic mobility (Fig. 10C, lanes1 and 6, band1) than was evident with any of the other combinations tested. This included *wtB probe + SMC NE (Fig. 10C, lane2), *WT probe + in vitro synthesized SRF (lane3), or *wtB + in vitro synthesized SRF (lane5). The difference in mobility of band 1 was not due to a difference in DNA fragment size since we found no difference in shift bands between *WT and *wtB when reacted with in vitro synthesized SRF. Results suggest either 1) that the wild-type 95-bp promoter fragment is able to interact with an SRF-like protein that is larger than SRF, while the 20-bp double-stranded CArG oligonucleotides contain only enough sequence information to bind SRF or 2) that the complexes formed with the 95-bp fragment and the 20-bp double-stranded oligonucleotides both contain SRF but that the 95-bp promoter fragment also contains DNA sequence information required to bind an SRF accessory protein(s), and the binding of the accessory protein(s) results in the decreased electrophoretic mobility.

DISCUSSION

The goals of the present study were to identify cis-elements and trans-factors that govern transcriptional expression of SM -actin in vascular SMCs. Results demonstrated that tissue-specific expression was dependent upon both negative- and positive-acting cis-elements, which include the CArG box elements, and upon SM trans-factors or complexes of factors that bound over the CArG boxes. Suppression of SM -actin promoter-CAT reporter gene constructs in cells that do not express their endogenous SM -actin gene was dependent upon negative-acting cis-elements that repressed the activity of shorter promoter constructs. For example, in skeletal myoblasts, addition of the promoter region between -547 and -371 abolished the transcriptional activity of the shorter constructs. Similarly, in endothelial cells, negative-acting cis-elements within three regions of the promoter, including the -547 and -371 region, abolished the high transcriptional activity of the p125CAT construct. Previous reports with the mouse SM -actin promoter also described a negatively acting cis-element, whose 5`-boundary was at -195, that was responsible for suppressing the transcriptional activity of a shorter promoter-reporter gene construct in fibroblasts and subconfluent (undifferentiated) BC3H1 skeletal-like myoblasts, which do not express their endogenous SM -actin gene(66) . Additionally, a 29-bp region (-151 to -123) of the chicken SM -actin promoter significantly reduced expression of the p122CAT ``core'' promoter in chicken fibroblasts and abolished expression in chicken skeletal myoblasts(24) . As of yet, the human SM -actin 5`-flanking region has not been studied sufficiently to determine whether it also contains a promoter region that turns off gene transcription. However, CArG box-binding factor-A has been shown to suppress transcription of the human SM -actin gene in mouse C2 skeletal myoblasts. The expression of CArG box-binding factor-A was also shown to decrease as the myoblasts were induced to form myotubes (67) , consistent with the induction of SM -actin during differentiation in this cell line(53) .

High transcriptional activity of SM -actin in SMCs required both CArG boxes and one or more additional upstream DNA elements. The CArG elements were also shown to bind SMC nuclear factors and to function in the complex regulation of tissue-specific expression. While mutation of either CArG by itself in the context of the 125-bp promoter abolished transcriptional expression in SMCs, it had no effect in endothelial cells. Only when both CArG boxes were mutated together was there any reduction in transcription, but this double CArG-mutated promoter still retained at least half of the activity seen with the wild-type sequence. In contrast to the high activity of the rat 125-bp promoter, previous studies in our laboratory utilizing the chicken 122-bp SM -actin promoter, which contains CArG boxes that are 100% conserved with the rat promoter but with extensive divergence outside the CArG elements (Fig. 3), was not active in endothelial cells(27) . Taken together, these findings suggest that the transcriptional activity of p125CAT in endothelial cells is elicted by nonconserved sequences outside the CArG motifs. Although the SM -actin CArG boxes may have some function in endothelial cells, there is clearly a qualitative difference in transcriptional activation via the CArG box elements in SMCs versus endothelial cells.

Skeletal muscle cells also showed a qualitative difference in the transcriptional activation of SM -actin via the CArG box elements. Although mature skeletal muscle cells do not express SM -actin in vivo, the gene is transiently expressed in immature skeletal cells during development in vivo(17) and constitutively expressed in cultured skeletal muscle cells upon differentiation into myotubes(53) . Mutating either CArG-A or CArG-B in the context of the full-length promoter resulted in about a 50% reduction of CAT activity in both SMCs and skeletal myotubes. However, the CArG elements were sufficient to induce a high level of transcriptional activity in the context of the p125CAT construct in SMCs but not in skeletal myotubes. In contrast to SMCs, activity of the CArG boxes in skeletal myotubes was dependent upon inclusion of the promoter region between -271 and -208. Interestingly, this region contains two E-Boxes at -252 and -214. E-Box elements bind to myoD and other members of the helix-loop-helix family and have been shown to be required for CArG-dependent transcriptional expression of a number of muscle-specific genes in skeletal muscle cells, including cardiac -actin(68, 69, 70) . The hypothesis that the differential regulation of SM -actin in SMCs versus skeletal myotubes involves the CArG elements is also supported by results of EMSA demonstrating that the SMC and skeletal myotube shift bands are distinct in the context of a 95-bp promoter segment containing two highly conserved CArG elements but not in the context of the CArG elements alone. Therefore, CArG boxes may function in the positive transcriptional activation of SM -actin within a variety of cultured cell lines that express their endogenous gene, but the mechanisms of differential transcriptional regulation are likely to involve cell-specific CArG box-binding proteins and/or protein complexes that involve more than one DNA element.

To our knowledge, this is the first report that provides direct evidence that the CArG-box binding activity in SMCs includes SRF or an SRF-like factor. The identification of SRF as one of the putative SMC transcription factors was demonstrated by 1) supershifting the SMC CArG-specific shift bands with anti-SRF antibodies, 2) determining that the SMC CArG box-binding factor was similar in size to SRF, 3) establishing that the shift bands formed with in vitro synthesized SRF had the same mobilities and CArG-specific competition characteristics as the shift bands formed with SMC nucleoproteins whether the proteins were reacted with the oligonucleotide duplexes of CArG-A, CArG-B, or SRF's optimum binding site (SRE), and 4) ascertaining that the relative binding affinity of in vitro synthesized SRF for the CArG-A and CArG-B oligonucleotide duplexes was the same as the relative binding affinity of SMC nuclear factors for the same oligonucleotides. Although these results do not prove that SRF is involved in the transcriptional activation of SM -actin in SMCs, we also demonstrated that CArG box mutations that abolished transcription also inhibited in vitro binding of both SRF and the SMC CArG box-binding factor. Results are similar to those obtained in previous studies of the human (62) and Xenopus(54) cardiac -actin genes that demonstrated that SRF was likely to be the factor that regulates these genes.

This report provides evidence that SRF functions as a part of the tissue-specific transcriptional machinery of SM -actin. However, SRF was originally identified as the factor that binds to the SRE of the c-fos proto-oncogene(52, 55) , a member of a family of genes termed immediate-early response genes, many of which are thought to be of cardinal importance in governing cell growth and differentiation in nonmuscle (71, 72, 73) as well as in muscle cells(74, 75) . At the core of the c-fos SRE is a CArG motif, which has been shown to be the primary element required for SRF binding(55) . CArG box motifs have also been found in the SREs of a variety of other immediate-early response genes, and there is evidence for involvement of SRF in their regulation as well(36) . Although previous studies in our laboratory have shown that the transcription of nonmuscle -actin is rapidly and transiently stimulated in SMCs when exposed to serum-containing medium, the transcription of the endogenous SM -actin in SMCs was not altered under the same conditions(40) . Similar results have been reported for SM -actin in fibroblast cells(38) , as well as for the skeletal and cardiac -actins(62, 76) . Additional studies in our laboratory demonstrated that transcription of the truncated SM actin promoter-CAT constructs in SMCs was not affected by serum stimulation. ()Therefore, very similar SRE/CArG promoter elements appear to be involved in both the nontissue-specific transcription of a variety of nonmuscle-specific genes inducible by growth factors as well as in the constitutive expression of a number of muscle-specific genes in a tissuespecific manner. Several models, which are not mutually exclusive, could be postulated to explain this phenomenon.

First, we cannot rule out the possibility that the CArG box-binding factor that activates SM -actin in SMCs is similar but not identical to SRF. The SMC factor might be completely distinct from SRF or only one monomer subunit of the SRF dimer need be different to form a tissue-specific unit. Similar to helix-loop-helix factors(77) , dimerization among heterologous SRF-like subunits could produce a variety of heterodimers from a limited number of protein monomers, thereby expanding the regulatory potential. In this report, we demonstrated a difference in mobility between SMC nucleoprotein complexes and in vitro synthesized SRF complexes formed with the 95-bp promoter fragment, while the same complexes formed with the 20-bp CArG-containing oligonucleotides did not show this difference (Fig. 10C). This is consistent with a SMC nuclear factor distinct from SRF interacting with the SM -actin CArG elements, but only in the context of sequences contained within the 95-bp promoter fragment. This model is also consistent with studies by Walsh and co-workers(57) , which demonstrated that the chicken skeletal -actin CArG elements bound a chicken skeletal muscle nuclear factor distinct from SRF in EMSA. Alternatively, if SRF is the factor that activates SM -actin expression, a second model might involve tissue- or stimulus-specific, post-translational modifications of SRF that would dictate its activation properties. SRF is a phosphoprotein(62, 78) , and it is possible that distinct signaling pathways function to activate SRF under various conditions or in different cell-types. Alternatively, variations at the gene level, such as methylation patterns or permissive versus nonpermissive chromatin configurations, might only allow SRF to form an active complex over muscle CArG elements under basal conditions, while the interaction of SRF over immediate-early gene CArG elements might require a growth stimulus. A third possibility to explain SRF's involvement in both muscle specificity and serum inducibility is that SRF might bind in conjunction with a diverse group of accessory factors that are themselves responsible for the tissue- and stimulus-specific activities. The data in Fig. 10C are consistent with this model in that the unique slower mobility shift band seen with the 95-bp fragment reacted with SMC nuclear proteins may represent the binding of an accessory factor that is present only in SMC nuclear extract and whose binding is dependent on CArG flanking sequences not found in the 20-bp CArG oligonucleotides. Indeed, studies of c-fos regulation have shown that SRF binds an accessory protein to form a ternary complex over the SRE(79) , and it has been suggested that the accessory factor functions to integrate different signal transduction pathways at the SRE(79, 80) . In the case of the SRF accessory proteins p62/TCF, Elk-1, and SAP-1, a motif known as the Ets domain-binding element is located on the 5` side of the CArG box in the c-fos SRE and is required for binding these accessory proteins(79, 81, 82) . SM -actin CArG elements are not flanked by Ets domain-binding motifs; however, a yet uncharacterized DNA element may be located in the evolutionarily conserved regions that flank the 5` or 3` sides of CArG-A and/or CArG-B. The presence of CArG accessory factors specific to SMCs would also explain the difference in electrophoretic mobilities between shift bands formed with SMC versus skeletal myoblast/myotube NEs (Fig. 10B) and explain the differential expression of the promoter-CAT constructs in the various cell lines.

Consistent with all of the above models of SRF activity is the hypothesis that either the internal CArG nucleotides, the sequences that immediately flank the CArG elements, and/or cis-elements some distance from the CArG boxes are essential in determining the nature of the complexes that bind the CArG elements. This is true for helix-loop-helix proteins, which have distinct binding site preferences that depend on variable nucleotides within and around the E-box motif. Although the internal nucleotides are extremely variable from one CArG to the next, the internal sequence of any particular CArG element is usually highly conserved, and there is evidence that modification of the A/T core has a pronounced affect on transcriptional activity(83) . Additionally, the two 10-bp CArG elements in the SM -actin gene are each located within larger 16-bp domains that are 100% conserved between all species in which the promoter has been cloned, and a variable amount of flanking sequence is conserved for other gene CArG elements as well. Observations in the present studies suggest that the flanking and/or internal CArG sequences may also be important. First, as noted above, the 95-bp fragment bound a larger protein or complex of proteins in SMC nuclear e