A Novel Site in the Muscle Creatine Kinase Enhancer Is Required for Expression in Skeletal but Not Cardiac Muscle*

Expression of the muscle creatine kinase (MCK) gene in skeletal and heart muscle is controlled in part by a 5 tissue-specific enhancer. In order to identify new regu- latory elements, we designed mutations in a previously untested conserved portion of this enhancer. Transfec- tion analysis of these mutations delineated a new control element, named Trex (Transcriptional regulatory element x), which is required for full transcriptional activity of the MCK enhancer in skeletal but not cardiac muscle cells. Gel mobility shift assays demonstrate that myocyte, myoblast, and fibroblast nuclear extracts but not primary cardiomyocyte nuclear extracts contain a trans -acting factor that binds specifically to Trex. The Trex sequence is similar (7/8 bases) to the TEF-1 consen- sus DNA-binding site involved in regulating other muscle genes. To determine if TEF-1 interacts with Trex, selected TEF-1 binding sites such as GTIIc and M-CAT and two anti-TEF-1 antisera were used in gel shift as- says. These experiments strongly suggest that a factor distinct from TEF-1 binds specifically to Trex. Thus it appears that MCK transcription is regulated in skeletal muscles through a Trex-dependent pathway while Trex is not required for MCK expression in heart. This distinction could account partially for the difference in levels of muscle creatine kinase in these tissues.

Expression of the muscle creatine kinase (MCK) gene in skeletal and heart muscle is controlled in part by a 5 tissue-specific enhancer. In order to identify new regulatory elements, we designed mutations in a previously untested conserved portion of this enhancer. Transfection analysis of these mutations delineated a new control element, named Trex (Transcriptional regulatory element x), which is required for full transcriptional activity of the MCK enhancer in skeletal but not cardiac muscle cells. Gel mobility shift assays demonstrate that myocyte, myoblast, and fibroblast nuclear extracts but not primary cardiomyocyte nuclear extracts contain a trans-acting factor that binds specifically to Trex. The Trex sequence is similar (7/8 bases) to the TEF-1 consensus DNA-binding site involved in regulating other muscle genes. To determine if TEF-1 interacts with Trex, selected TEF-1 binding sites such as GTIIc and M-CAT and two anti-TEF-1 antisera were used in gel shift assays. These experiments strongly suggest that a factor distinct from TEF-1 binds specifically to Trex. Thus it appears that MCK transcription is regulated in skeletal muscles through a Trex-dependent pathway while Trex is not required for MCK expression in heart. This distinction could account partially for the difference in levels of muscle creatine kinase in these tissues.
The differentiation of embryonic mesodermal cells into skeletal or cardiac muscle cells requires the controlled expression of numerous muscle-specific genes (1)(2)(3)(4)(5). Because the muscle creatine kinase (MCK) 1 gene is expressed at different levels in mature skeletal and cardiac muscle cells (6), it is an instructive model for comparing the mechanisms that regulate expression of the same muscle-specific gene in both striated muscle tissues. The current study identifies a new regulatory element within the MCK enhancer, which exhibits such differential activity in skeletal and cardiac muscle cells.
Expression of the MCK gene in differentiated muscle cells is controlled by an array of cis-elements and trans-acting factors. Particular attention has been given to the analysis of a 206-bp upstream enhancer, which confers muscle-specific expression both in cultured cells and in transgenic mice (7)(8)(9). Mutational analysis, gel mobility shift studies, and footprinting assays have identified six transcriptional regulatory elements in the MCK enhancer (from 5Ј to 3Ј): the CArG, AP2, A/T-rich, left and right E boxes, and MEF-2 sites (6,10,11). Each of these MCK gene control elements has been well conserved during mammalian evolution (7,(12)(13)(14), and similar nucleotide motifs have been implicated in the transcriptional control of numerous other muscle-specific genes (15).
Mutating these six MCK enhancer control elements causes different relative effects in cultured cardiomyocytes and skeletal muscle cells (6). Mutation of the MCK right E box site decreases enhancer activity more dramatically in skeletal than in cardiac muscle cells, while mutations of the CArG site or the A/T-rich site seem more deleterious in cardiac than in skeletal muscle cells. In contrast to these differential effects, mutations of the MEF-2 site or the left E box site decrease enhancer activity to about the same extent in both cell types. Unlike the other MCK enhancer control elements, the AP2 site seems to repress transcription in cultured skeletal and cardiac muscle cells as its mutation leads to increased expression in both cell types.
These six enhancer regulatory elements are the target sites for an array of muscle and non-muscle-specific DNA binding factors. The left and right E boxes contain the consensus core sequence CANNTG, which is the target for skeletal musclespecific determination factors of the MyoD family of transcription factors (16). These two E boxes, which differ in their flanking regions, exhibit noticeably different transcriptional activities and protein binding properties (17,18). Although E boxes are involved in the cardiac regulation of MCK as well as a number of other cardiac muscle genes (19 -23), 2 MyoD family members are not detected in the heart at any time during development (4). The MEF-2 site and the A/T-rich site are both rich in adenine and thymidine (25,26). The MEF-2 site is the target of MEF2 proteins, which are also known as the Related to Serum Response Factor (RSRF) family (27,28), and BBF-1, a serum-inducible factor identified in cardiac but not in skeletal muscle (29). MEF2 proteins have been shown to cooperate in transactivation with members of the MyoD family of transcription factors (30 -33) and bind with lower affinity to the MCK A/T-rich site (34). The mouse MCK A/T-rich element is also a DNA-binding site for both the homeoprotein MHox and the ubiquitous Oct-1 factor (26,34). The CArG motifs have been shown to bind serum response factors (28).
The MCK enhancer does not contain all of the regulatory elements identified in other muscle genes. For example it does not contain functional Sp1 and CCAC sites found in regulatory regions of the ␣ actin (22) and myoglobin genes (35). The MCK enhancer also appears to contain no perfect match to the M-CAT sites, key elements in regulating the cardiac troponin T * This work was supported by National Institutes of Health Grants AR 18860 and 3P01-HL 03174 and by a grant from the Muscular Dystrophy Association (to S. D. H.). 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.
‡ Supported by National Institutes of Health Biotechnology Predoctoral Training Grant GM-08437-05.
M-CAT sites were of particular interest to the present study because a newly identified MCK enhancer control element called Trex (Transcriptional regulatory element x) bears close similarity to the consensus MCBF/TEF-1-binding sites in other genes. We show here that the MCK enhancer Trex site is required for full transcriptional activity of the enhancer in MM14 skeletal muscle cell cultures. However, mutations in the Trex site that greatly reduce MCK reporter gene expression in these cultures exhibit virtually no effect in heart cell cultures. Furthermore, despite its similarities to the consensus TEF-1binding site, we show that the Trex site is not a target for TEF-1, nor do previously identified TEF-1-binding sites (such as M-CAT or the SV40 enhancer GTIIc site) compete with the Trex sequence for specific Trex binding. A Trex-specific binding complex has been identified in MM14 skeletal myocyte, MM14 myoblast, NIH-3T3 fibroblast, and hybridoma cell nuclear protein extracts. However, no such complex was identified in nuclear protein extracts from neonatal cardiomyocytes. Trex may thus be partially responsible for the differential expression of MCK and potentially other genes in skeletal versus cardiac muscle cells.
Cell Cultures and Transfection Procedures-Mouse MM14 skeletal myoblasts were propagated to a density of 5 ϫ 10 5 /100-mm collagencoated tissue culture dish in proliferation medium (PL) (Ham's F-10C, 15% horse serum, and 2 ng/ml basic fibroblast growth factor) as described previously (45). Cells were transfected with 8 g of test plasmid plus 2 g of a reference pUCSV2pap plasmid via calcium-phosphate-DNA precipitation (6). Four hours later, the cells were glycerol-shocked, and myogenic differentiation was induced by switching to differentiation medium (F-10C with 0.75% horse serum, 6 g/ml insulin, and no basic fibroblast growth factor). For stable transfection, glycerol-shocked MM14 cells were selected in PL medium supplemented with 500 g/ml Geneticin (G418 sulfate) for 10 -15 days prior to inducing myogenic differentiation.
Skeletal and cardiac muscle cells were maintained in differentiation medium at least 24 h before harvesting. In both transient and stable cultures, at least 95% of the cells were differentiated as assessed by myosin immunostaining (47). The cells were analyzed either for chloramphenicol acetyltransferase (CAT) and placental alkaline phosphatase gene expression in the case of transient transfections (6) or for CAT and protein concentration (Bradford assay) in the case of stable transfections (48). For each MCKCAT construct tested, at least two plasmid preparations were used. The effect of each mutation was evaluated as the relative amount of CAT/placental alkaline phosphatase or CAT/ [protein] and was compared with the wild-type enhancer-CAT construct whose activity was set to 100. The results of independent transfections are shown as the average of at least three transient transfections of three plates each or three plates of stably transfected cells.
Preparation of Nuclear Extracts-Crude nuclear extracts were prepared essentially as described by Dignam et al. (49) and Buskin and Hauschka (10) from MM14 myoblasts and myocytes, NIH-3T3 fibroblasts, F5D hybridoma cells, rat primary cardiomyocytes, and cardiac fibroblasts (6). The percentage of differentiated cells was determined by immunostaining a parallel culture for myosin heavy chain. To minimize the effect of contaminating myocytes in myoblast nuclear extracts, only cultures with typically less than 5% myosin heavy chain-positive cells were used.
Gel Mobility Shift Assays-5Ј-32 P-labeled DNA oligomer probes were purified on a 12% polyacrylamide gel before use. The binding reactions were carried out in 10-l volumes. 1.5-2 g of nuclear protein extracts were mixed with 10 g of bovine serum albumin in 10 mM Tris-Cl, pH 7.4, 0.1 M NaCl, 0.1 mM EGTA, 0.5 mM dithiothreitol, 0.3 mM MgCl 2 , 0.5 mM phenylmethylsulfonyl fluoride, and 4% glycerol and incubated with or without unlabeled oligonucleotide competitor at room temperature for 15 min prior to addition of 0.15 ng of oligonucleotide probe plus 360 ng of poly(dI-dC) as nonspecific competitor. The incubation was then continued for 10 min at room temperature prior to loading the mixture onto the gel. For antibody supershift experiments, 1 l of TEF-1 antiserum replaced the competitors. Two TEF-1 specific antisera (50), denoted as N and C, were kindly provided by Drs. Irvin Davidson and Pierre Chambon (Institut de Chimie Biologique, INSERM-U.184, Strasbourg, France). N and C were prepared against an amino-terminal peptide (amino acids 16 -25)

MCK Enhancer
Sequence Alignment-Nucleotide sequence comparisons of the MCK gene 5Ј enhancers from mouse (7), rat (12), human (13), and rabbit (14) identify regions that appear to have been conserved during mammalian evolution and which may therefore be important for enhancer function. Fig. 1A shows the alignment of 65 bp, between the mouse MCK enhancer residues Ϫ1241 and Ϫ1176. Within these 65 bp, 74% are conserved across the four species. While the mouse and rat sequences are the most similar, even the mouse and human sequences are 88% identical. In comparison only 49% of the entire 206-bp enhancer is identical between the same four species. Interestingly, nucleotide alignment of the four mammalian MCK enhancers identifies an additional highly conserved region located between the AP2 and A/T-rich sites (Fig.  1A). Due to its high degree of evolutionary conservation and to its sequence similarity to the M-CAT control elements of other muscle genes, this region (subsequently named "Trex" for Transcriptional regulatory element x) was analyzed for its transcriptional activity.

Effects of Trex Region Mutations on MCK Reporter Gene Expression in Skeletal and Cardiac Muscle Cell Cultures-To
investigate the possibility that an additional regulatory site exists between the AP2 and A/T-rich sites, a 10-nucleotide mutation (M1) was constructed to test this region (Fig. 1B). The mutation was carried in constructs shown in Fig. 1C and tested by transient and stable transfection assays of MM14 skeletal muscle cells and by transient transfections of rat primary cardiomyocytes (Fig. 2).
In the context of the full-length promoter 1256MCKCAT, mutation M1 decreased enhancer activity 2-4-fold in both transiently and stably transfected MM14 myocyte cultures ( Fig. 2A). In contrast, the 1256MCKCAT enhancer construct bearing the M1 mutation was expressed at a similar level to the wild-type enhancer in rat primary cardiomyocyte cultures ( Fig.  2A). These results suggest that mutation M1 disrupts a positive regulatory site that is required for full transcriptional activity of the enhancer in skeletal myocytes but not in cardiomyocytes. A more dramatic effect of mutation M1 was observed in constructs containing the enhancer, in either its normal (ϩ) or reversed (Ϫ) orientation, fused to a truncated promoter ((ϩenh)80MCKCAT and (Ϫenh)80MCKCAT). In these contexts the M1 mutation caused nearly a 10-fold decrease in enhancer activity in both transient (Fig. 2B) and stable transfection assays of MM14 myocytes (data not shown). The greater relative effects of Trex site mutations when tested in (ϩenh)80MCKCAT versus 1256MCKCAT contexts may be due to compensatory effects of sequences in the Ϫ1050to Ϫ81-bp region contained in the later construct; however, no Trex site has been identified in this region.
To determine if Trex is involved in silencing MCK gene expression in cells that do not express MCK, NIH-3T3 fibroblasts and MM14 myoblasts were transfected with 1256MCK-CAT constructs. As expected, the wild-type MCK enhancer construct was expressed at a low background level (data not shown), consistent with the lack of endogenous MCK expression in these cell types. Mutation M1 did not result in elevated levels of reporter gene expression, indicating that Trex is not responsible for silencing the MCK gene in fibroblasts and myoblasts.
To further delineate the Trex site the sequence between the AP2 and A/T-rich sites was examined via two smaller mutations, M2 and M3 (Fig. 1B). Mutation of bases in the 5Ј-region of the M1 mutation (mut M3) had no effect on expression of the (enh)80MCKCAT reporter gene when tested in either the pos-itive or negative enhancer orientation (Fig. 2B). These results indicate that none of the residues modified by M3 are critical parts of Trex. In contrast, disruption of bases within the more 3Ј-sequence by mutation M2 impaired expression severely, reducing the enhancer function about 8-fold (Fig. 2B). Since the nucleotide substitutions made in mutation M2 differ markedly from those made in mutation M1, it seems very unlikely that both mutations would create negative regulatory sites with similar repressor activity. The most straightforward interpretation is that the M1 and M2 mutations both disrupt the same positive control element.
Transfections of MM14 myocytes or 3T3 fibroblasts with either Ϫ80MCKCAT or a multimerized Trex sequence ligated to the basal MCK promoter, (Trex) 4 ATCAT, result in background levels of CAT gene expression (data not shown), indicating that four copies of the Trex site are not sufficient to promote transcription. Therefore, the Trex site appears to require additional regulatory elements in order to confer transcriptional activity to skeletal muscle genes.
The Trex Site Exhibits Specific Nuclear Protein Binding-DNA-protein binding experiments were used to characterize trans-acting factors that interact with the MCK Trex site. A 22-bp Trex oligonucleotide was bound in gel mobility shift assays by nuclear proteins from MM14 skeletal myocytes, myoblasts, and NIH-3T3 fibroblasts (Fig. 3, lanes 2, 5, and 8) as well as by nuclear extracts from cardiac fibroblasts and F5D hybridoma cells (data not shown). The binding of nuclear proteins was sequence-specific as assessed by competition with excess unlabeled oligonucleotides bearing the wild-type Trex sequence (Fig. 3, lanes 3, 6, and 9) and the lack of competition by excess oligonucleotides bearing the functionally inactive M1 sequence (Fig. 3, lanes 4, 7, and 10). A similar competition pattern was observed when the entire 206-bp wild-type or the mutant M1 enhancer was used as competitor (data not shown). Further titration of the myocyte Trex binding complex with increasing amounts of unlabeled competitors confirmed its sequence specificity (Fig. 4B, see below). Since the Trex binding complex migrated to approximately the same gel location irrespective of whether the nuclear extracts were derived from MM14 myocytes, MM14 myoblasts, NIH-3T3 fibroblasts, rat primary cardiac fibroblasts, or mouse hybridoma cells, it seems likely that the same Trex-binding protein, termed TrexBF (Trex Binding Factor), is present in these diverse cell types.
Characterization of the Trex Site-The Trex site shows a high degree of sequence similarity, except for the two underlined mismatches, with the SphI control element (GAGATG-CATGCTT) located in the SV40 enhancer A domain (52). The SphI motif binds TEF-1, a ubiquitous factor that also interacts with SphII and GTIIc motifs in the SV40 enhancer (50), and with the muscle-specific M-CAT motifs in the cardiac troponin T promoter (42). Farrance and colleagues (42) have shown that SphI, GTIIc, and M-CAT motifs are functionally interchangeable and have proposed a consensus TEF-1-binding site (A/ G)CATNC(C/T)(T/A) that differs from the Trex site by the single underlined base pair. Several complementary approaches were thus used to determine whether TEF-1 is indeed the Trex binding factor.
Gel mobility shift assays showed that the specific complex formed by Trex and nuclear proteins from MM14 myocytes was not abolished by competition with unlabeled GTIIc site or M-CAT site oligonucleotides (Fig. 4A, lanes 2-4 and lanes 5-7). Similarly, this complex persisted upon addition of a 22-mer oligonucleotide (Fig. 1B, Trex G-C) containing a 1-base pair mutation that converts the Trex site into a sequence equivalent to the TEF-1 consensus sequence (Fig. 4B, lanes 17-19). Crosscompetition studies showed that binding to GTIIc, TrexG-C, or to M-CAT probes was not altered by addition of unlabeled Trex oligonucleotide (data not shown). Since binding to Trex is not altered by increasing amounts of known TEF-1-binding sites, it appears that TrexBF is not TEF-1.
Further evidence that the TrexBF differs from TEF-1 was obtained by immunological studies using anti-TEF-1 sera that recognize both human and mouse TEF-1 (40). Addition of either N-terminal or C-terminal anti-TEF-1 polyclonal antiserum to MM14 nuclear extracts produced no change in migration of the TrexBF complex (Fig. 5, lanes 6 and 7). However, both antisera caused a supershift when the same skeletal muscle extracts were combined with GTIIc oligonucleotides (Fig. 5, lanes 2 and  3). As a control, no supershift was observed upon addition of nonimmune serum (Fig. 5, lanes 1 and 5). This result indicates that MM14 nuclear extracts contain TEF-1 or a TEF-1-related protein that can participate as part of a GTIIc binding complex. However, the DNA binding activity defined as TrexBF is not TEF-1.
Taken together these results indicate that both TEF-1 and a previously unrecognized factor, TrexBF, are present in MM14 nuclear extracts. However, despite the close sequence similarity between the MCK Trex site and the TEF-1-binding sites, TrexBF is not TEF-1.
The Trex Binding Factor Is Not Present in Cardiomyocyte Nuclear Extracts-Results from transfection experiments showed that the Trex region is not necessary for cultured cardiomyocytes to express 1256MCKCAT; yet, TrexBF could be  -7), or NIH-3T3 fibroblasts (lanes 8 -10) and analyzed via gel mobility shift assay. Lane 1 shows the mobility of the free probe, indicated by an arrow at the bottom of the gel. A slowly migrating DNA-protein complex of the same apparent mobility was formed with the Trex probe and nuclear proteins from each cell type. The Trex binding complex, indicated by an arrow, is abolished by addition of a 100-fold molar excess unlabeled Trex probe (lanes 3, 6, and 9) but not by addition of an equivalent amount of a similar 22-bp oligonucleotide carrying the mutation M1 (lanes 4, 7, and 10, respectively). present in cardiomyocytes. To test this hypothesis, cardiomyocyte nuclear extracts were used in gel mobility shift assays. No shifted bands were observed when cardiomyocyte nuclear protein extracts were added to the labeled Trex probe (Fig. 6, lanes   1-3), suggesting that no cardiomyocyte nuclear-protein recognizes and binds to the Trex sequence. To ascertain the quality of the cardiac nuclear extracts, the presence of other specific binding factors in these extracts was investigated using a TrexG-C probe (Fig. 6, lanes 4 -7) and the previously described MCK CArG site (data not shown) (6). Two DNA-protein complexes were formed between cardiomyocyte nuclear proteins and the TrexG-C-labeled probe (Fig. 6, lane 4). Both complexes were competed by the addition of a 100-fold molar excess unlabeled Trex G-C sequence (Fig. 6, lane 5) or GTIIc sequence (Fig. 6, lane 6). Conversely, neither complex was abolished by addition of unlabeled wild-type Trex sequence (Fig. 6, lane 7). These results suggest that cardiomyocyte nuclear extracts do not contain any TrexBF and confirm that they contain intact binding proteins for the TEF-1 consensus site (40,42).
To examine the possibility that cardiomyocyte extracts contain an inhibitor that prevents binding to Trex, up to a 3-fold excess of cardiomyocyte nuclear extract was added to a skeletal extract and tested in gel mobility shift assays. In no case was there a specific reduction of skeletal muscle extract binding to Trex (data not shown), indicating that no inhibitor of Trex binding exists in these extracts.

A Novel Regulatory Element, Trex, Is Required for Full Transcriptional Activity of the MCK Enhancer in Skeletal
Myocytes-This study has identified a novel cis-acting regulatory element, Trex, within the MCK enhancer. Transfection analyses in MM14 myocytes indicate that the Trex site functions as a positive control element. The relative loss in skeletal muscle transcriptional activity due to Trex mutation M1 in 1256MCK-CAT is comparable with the loss of activity resulting from mutations in the MEF2 site, the left E box, or the CArG MCK enhancer elements (6). The same Trex mutation neither decreased nor increased the low level of reporter gene expression in proliferating myoblasts or NIH-3T3 fibroblasts. This sug-  1 and 9) is competed away by addition of an excess of unlabeled Trex oligonucleotide (lanes 10 -13) but not by unlabeled oligonucleotides carrying the mutation M1 (lanes 14 -16), the single base pair mutation TrexG-C (lanes [17][18][19], the closely related GTIIc oligonucleotide (lanes 2-4), or the M-CAT oligonucleotide (lanes 5-7).  2 and 3). In contrast, no supershift is observed with the Trex probe (lanes 6 and 7). Control assays with non-immune sera (P) are shown in lanes 1 and 5, respectively.
gests that the Trex site and associated factors do not act as an inhibitor of MCK gene expression in myoblasts or non-muscle cells. When examined in the (enh)80MCKCAT configurations, mutation M1 was even more deleterious than mutations in most of the other enhancer elements and nearly as deleterious as mutations of the right E box. The relative effects of the smaller overlapping mutations M2 and M3 suggest that the crucial Trex bases are located within the central and 3Ј-portion of the conserved Trex region.
Trex Is Inactive in Cardiomyocytes-Results from transient transfection assays of primary rat cardiomyocytes indicate that Trex does not function as a positive control element in cardiac muscle cells since the M1 mutation has no significant effect on reporter gene expression ( Fig. 2A). This finding is consistent with the observation that TrexBF is undetectable in cardiomyocyte nuclear protein extracts (Fig. 6), most likely due to its absence in these cells. However, an alternative explanation is that TrexBF is present in cardiomyocyte extracts but posttranscriptionally modified and thereby unable to bind DNA. Behavior of this type has been reported for MyoD, which is present in skeletal muscle myoblasts (53) and yet unable to bind control elements such as the MCK enhancer right E box (MEF-1) site (10). This inhibition appears to be due to members of the Id family, which inhibit differentiation by sequestering E proteins (E12/E47), thereby preventing MyoD family members from forming DNA-binding heterodimers (54,55). The possibility that cardiomyocyte extracts might contain an inhibitor that prevents TrexBF binding was tested by adding increasing amounts of cardiac muscle nuclear extracts to skeletal muscle extracts in gel mobility shift assays. No inhibition of skeletal muscle TrexBF binding was detected using this strategy. Taken together, our data are consistent with the hypothesis that TrexBF is not present in cardiomyocytes and with the corollary that the MCK Trex site is functionally silent in these cells.
TrexBF Is Present in Cultured Skeletal Muscle and in Nonmuscle Cells-Trex binding complexes of similar mobility are observed in skeletal myocyte, myoblast, and fibroblast nuclear extracts (Fig. 3). Although the Trex site does not appear to be active on its own, the presence of TrexBF among proliferating and differentiated skeletal muscle cells as well as among several non-muscle cell types suggests that TrexBF may also be involved in regulating non-muscle genes. Why Trex participation in such control mechanisms should be precluded in cardiomyocytes is unclear. However, the requirement for Trex seems to correlate with the higher levels of MCK gene expression in skeletal versus cardiac muscle cells (6).
Although TrexBF is present in various cell extracts, its activity has not been defined. TrexBF's activity could be regulated in these cell types through several mechanisms, for example by post-translational modification (e.g. phosphorylation) or by interaction with ubiquitous or cell-type specific accessory factors. Considering the fact that TrexBF activity is found in both replicating and postmitotic MM14 cells, mitogen-regulated phosphorylation is probably not involved in TrexBF activity. The alternative mechanism, involving protein-protein interactions, has been proposed for regulation of the cardiac troponin T gene. In this gene the ubiquitous TEF-1 factor binds to M-CAT control elements (42), but its full transcriptional activity requires intermediary factors, termed TIFs (50). Understanding the mechanisms involved in regulating TrexBF function awaits its cloning and characterization.
TrexBF Is Not TEF-1-The MCK enhancer Trex site exhibits sequence similarity with the consensus MCBF/TEF-1-binding site defined by the alignment of muscle gene M-CAT sites with TEF-1 binding motifs (42). In addition to their role in cardiac troponin T gene regulation (56), M-CAT motifs are also involved in regulating the ␣-(21, 37) and ␤-myosin heavy chain genes (39,40,57,58), as well as skeletal ␣-actin (41). The possibility that the MCK enhancer Trex site could be M-CAT or another TEF-1-binding site was examined. Gel mobility shift assays using M-CAT and other TEF-1-binding sites as competitors as well as supershift experiments using two anti-TEF-1 antisera confirmed that TrexBF is not TEF-1. Interestingly, the amino-terminal-specific antiserum to TEF-1 slightly diminishes formation of the TrexBF complex (Fig. 5, lane 6). Because the DNA-binding domain of TEF-1 (TEA domain) is located in its N-terminal region, it is possible that antiserum N recognizes a TEA-like domain in TrexBF.
Five TEF-1 isoforms have been characterized (59). Although we cannot rule out the possibility that TrexBF is a novel TEF-1 isoform, this seems unlikely since neither TEF-1 antiserum supershifted the Trex-TrexBF complex (Fig. 5). In addition, a single base pair mutation that converts the MCK enhancer Trex site to a consensus TEF-1-binding site does not compete for Trex binding (Fig. 4B). This suggests that association of TrexBF with the Trex site is very sequence-specific. Since a single amino acid substitution in a helix-loop-helix protein can change the protein specificity for bases flanking the core E box motif (60), it is possible that TrexBF and TEF-1 may have very similar structures but that slight changes are sufficient to alter their binding specificity. An additional distinction between TEF-1 and TrexBF is that TEF-1 is enriched in skeletal mus-FIG. 6. Trex-specific binding activity is undetectable in cardiac myocyte nuclear extracts. Nuclear extracts from primary rat cardiomyocytes were mixed with 5Ј-labeled Trex oligonucleotide (lanes 1-3) or TrexG-C, a single base pair mutant Trex oligonucleotide that conforms to the TEF-1 consensus binding sequence (lanes 4 -7). No nuclear proteins from the cardiomyocyte extract form a shifted complex with the Trex probe (lanes 1-3). In contrast, two DNA-protein complexes formed between cardiac proteins and the TrexG-C labeled probe (lane 4). Both complexes are competed by addition of a 100-fold molar excess unlabeled Trex G-C sequence (lane 5), as well as by excess GTIIc sequence (lane 6). Neither complex is abolished by addition of excess unlabeled wild-type Trex sequence (lane 7). cles as well as in heart tissue (40,59), whereas TrexBF is undetected in heart. Moreover, knocking out the TEF-1 gene affects cardiac muscle but not skeletal muscle development (61). When considered as a whole, these observations strongly support the hypothesis that TrexBF is not TEF-1.
The activation and steady state expression of the same genes in both skeletal and cardiac muscles could imply transcription factors that are common to both striated muscle types. Indeed since combinatorial interactions of multiple cis-acting elements underlie the regulation of muscle-specific genes, the relative ratios of common factors could be responsible for the quantitative differences in a gene's expression level in each type of muscle (35,38). However, in many cases heart and skeletal muscle cells use alternative transcriptional strategies to regulate tissue-specific expression of the same gene (3,62). For instance, distinct enhancer fragments control expression of the cardiac troponin C, cardiac troponin T, myosin light chain 2, and ␤-myosin heavy chain genes in cardiac versus skeletal muscle cells (24,56,63,64). Here we present evidence that a novel MCK enhancer regulatory element, Trex, is required for skeletal muscle but not for cardiac muscle expression of the MCK gene. Thus it appears that MCK is expressed in skeletal muscle through a Trex-dependent pathway and in heart through a Trex-independent pathway. It will be instructive to determine both how TrexBF is activated in skeletal muscles and how this factor is restricted in cardiomyocytes. It will also be important to extend our observations to transgenic mice as well as to evaluate the role of Trex in other muscle genes.