α7 Integrin Expression Is Negatively Regulated by δEF1 during Skeletal Myogenesis*

α7 integrin levels increase dramatically as myoblasts differentiate to myotubes. A negative regulatory element with putative sites for δEF1 is present in the α7 proximal promoter region. To define the role of δEF1 in regulating α7 integrin expression, we overexpressed δEF1 in C2C12 myoblasts. This resulted in a major down-regulation of α7 protein expression. Promoter assays revealed that C2C12 myoblasts transfected with δEF1 showed a decrease in activity of the 2.8-kb α7 promoter fragment, indicating regulation of α7 integrin at the transcriptional level. We have identified two E-box-like sites for δEF1 in the negative regulatory region. Mutation of these sites enhanced α7 promoter activity, indicating that these sites function in repression. MYOD, an activator of α7 integrin transcription, can compete with δEF1 for binding at these sites in gel shift assay. By using chromatin immunoprecipitation, we demonstrated a reciprocal binding of δEF1 and MYOD to this regulatory element depending on the stage of differentiation: δEF1 is preferentially bound in myoblasts to this region, whereas MYOD is bound in myotubes. The N-terminal region of δEF1 is necessary for α7 repression, and this region also binds the co-activator p300/CBP. Importantly, we found that the p300/CBP co-activator can overcome repression by δEF1, suggesting that δEF1 can titrate limiting amounts of this co-activator. These findings suggest that δEF1 has a role in suppressing integrin expression in myoblasts by displacing MYOD and competing for p300/CBP co-activator.

Integrins are heterodimeric molecules involved in cell-matrix binding and regulate diverse processes such as proliferation, survival, differentiation, and motility. Muscle myoblasts bind extracellular matrix proteins by using specific integrins, including the laminin-binding ␣7 integrin. Levels of ␣7 integrin increase over 50-fold during myodifferentiation. Tight regulation of ␣7 integrin ensures low levels of ␣7 integrin in myoblasts, which facilitates motility during embryogenesis and helps in muscle repair after injury in the adult (1)(2)(3)(4). In contrast, myotubes and myofibers have high levels of the ␣7 integrin essential for stable costameric and myotendinous junctions (5). It is important to define how expression of ␣7 is regulated, not only for its role in myoblast movement, anchorage, and myotendinous junction formation but also because loss of ␣7 integrin is associated with certain types of muscular dystrophy/myopathy (6).
It is known that muscle-specific gene expression is regulated positively by the master regulators of the MRF family (myoD, myf5, myogenin, and MRF4) and negatively by Twist (7), Id (8), I-mf (9), ␦EF1 (10), and several other transcription factors (11)(12)(13). Although MYOD is expressed in undifferentiated myoblasts at modest levels, cells do not enter the terminal differentiation program. This failure to differentiate also reflects modulation of MYOD activity that is negatively regulated by several mechanisms, including the transcription factors mentioned above, by subcellular localization (14), and by association with the chromatin-regulating enzymatic activity of histone acetylase (p300/CBP) 2 / deacetylase (15). Consequently, myoblasts remain in an undifferentiated state and are prevented from premature myodifferentiation. Modulation of ␣7 integrin expression may also be regulated by these mechanisms during skeletal myogenesis.
To understand the regulation of ␣7 integrin, a 2.8-kb promoter of the mouse ␣7 integrin promoter was cloned previously and shown to be positively regulated by myoD and myogenin but not by myf-5 or MRF4 (16). Recently, c-Myc, a repressor of myogenesis, was shown to negatively regulate ␣7 integrin promoter activity in myoblasts by directly binding to the Ϫ2.0to Ϫ2.6-kb region of the promoter (17). c-MYC levels are high in proliferating myoblasts but decrease during myotube formation enabling ␣7 integrin levels to increase during differentiation.
Deletion constructs of the ␣7 integrin promoter have shown that an additional negative regulatory region is present in the Ϫ401to Ϫ292-bp region of the mouse ␣7 integrin promoter (16). Compared with the negative element between Ϫ2.1 and Ϫ2.8 kb, the region between Ϫ401 and Ϫ292 bp produces the dominant repression on ␣7 integrin promoter activity. Although this region is not responsive to c-MYC, it contains potential binding sites for ␦EF1, a zinc finger transcription factor with a homeodomain between the N-and C-terminal zinc fingers. ␦EF1 was first identified as a repressor of ␦1-crystallin expression in lens (10). Because ␦EF1 binds E-boxes, the binding elements for basic helix-loop-helix activators such as MYOD, a role for this protein in myogenesis was postulated. Confirming this prediction, ␦EF1 was found to negatively regulate myogenesis, and overexpression of ␦EF1 inhibited MYOD-induced muscle differentiation in 10T1/2 cells and repressed MYOD-dependent activation of the muscle creatine kinase enhancer (10). Similar results were observed when the human homolog ZEB was overexpressed in mouse C2C12 myoblasts, where myotube formation was blocked and expression of myosin heavy chain and ␣4 integrin markers of differentiation were down-regulated (18,19). Most interestingly, although ␦EF1/ZEB is a repressor of muscle differentiation, its levels are not down-regulated during muscle differentiation (18). It is thought that increased levels of MYOD displace ␦EF1 from E-boxes and overcome ␦EF1-mediated repression.
In this study, we investigated the regulation of ␣7 integrin expression by ␦EF1. Multiple ␦EF1-like binding sites are present in the proximal promoter regions that negatively regulate ␣7 promoter activity through interaction with ␦EF1. Further characterization revealed that this regulatory element also binds MYOD. Inhibition of the integrin promoter by ␦EF1 is dependent on myodifferentiation status.

EXPERIMENTAL PROCEDURES
Cell Culture-C2C12 myosatellite cells were maintained in growth medium (Dulbecco's modified Eagle's medium (DMEM) supplemented with 20% fetal bovine serum). For differentiation of C2C12 myoblasts to myotubes, cells were grown to ϳ80% confluence in growth medium and subsequently differentiated for 8 days with DMEM supplemented with 2% horse serum (Invitrogen), by which time most of the cells had fused into myotubes (16). 10T1/2 cells were maintained in Earle's minimum medium with 10% FBS.
Plasmid Construction-The deletion constructs of the ␣7 integrin promoter p300, p400, p600, p1.2 kb, and the full-length p2.8-kb promoter have been described previously (16). We have, however, changed the numbering of the promoter region. In this paper, ϩ1 indicates the start site of translation. The p300 construct spans Ϫ292 to ϩ3 and p400 spans Ϫ401 to ϩ3. By using the QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, CA), the AGGTG at Ϫ378 bp in the p400 construct was mutated to AACTC. The resulting construct was designated p400mut1. Additionally, the GACCTG site at Ϫ333 bp was mutated to GCCCGA and resulted in p400mut2. The PAGE-purified primers (IDT, Coralville, IA) used to generate these mutant constructs were p400-MUT(Ϫ378) forward and p400MUT(Ϫ333) forward (TABLE ONE). The p300 construct was used to generate mutations at Ϫ123 and Ϫ58 bp by using the primers p300MUT(Ϫ123) forward and p300-MUT(Ϫ58) forward. The resulting construct was p300mut1 with a mutant Ϫ123-bp site and p300mut2 with mutant Ϫ123and Ϫ58-bp sites. The ␦EF1 expression vector was a kind gift from Prof. Yojiro Higashi (Osaka University, Japan). Deletion constructs of ␦EF1 were made by PCR amplification of the C-zinc finger region (Ϫ2426 to 3243 bp), the C-zinc finger plus the homeodomain (CZF-HD) from 1357 to 3243 bp, and the C-zinc finger, homeodomain, and N-zinc finger regions (CZF-HD-NZF) from 454 to 3243 bp. The primers used are listed in TABLE ONE. These constructs were cloned at the BamHI and KpnI sites of pcDNA3.1 A HisMyc(Ϫ). The vectors GAL4 Sp1 and the luciferase reporter vector GAL4Luc were the kind gifts from Dr. Reshma Taneja (Mount Sinai Medical School, New York). The expression vectors for MYOD, MEF2, CBP (CREB-binding protein), and the internal control vector for transfection efficiency, pRL␤-galactosidase, were the gifts from Dr. Rik Dernyck (University of California, San Francisco).
Electrophoretic Mobility Shift Assays-Nuclear extract was prepared from C2C12 myoblasts or myotubes grown in 10-cm plates according to the protocol of Schreiber et al. (20). The binding reaction (20 l) included binding buffer (13 mM Hepes, pH 7.9, 60 mM KCl, 0.13 mM EDTA, 2 mM dithiothreitol, and 10% glycerol), 50,000 cpm of radiolabeled probe, and 3-4 g of nuclear extract. The probes were endlabeled by using [␥-32 P]ATP and polynucleotide kinase. DNA-protein binding reactions were carried out for 30 min on ice. Reactions were loaded onto a 5% polyacrylamide gel in 0.5ϫ TBE (45 mM Tris, 45 mM boric acid, and 1 mM EDTA). Gels were dried and autoradiographed. For supershift assays, antibodies (1-2 l) were added overnight to binding reactions. Antibody to ␦EF1 was obtained from Dr. Yojiro Higashi (Osaka University, Japan). Anti-MYOD antibody was from Novocastra (Newcastle upon Tyne, UK).
Transient Transfections and Promoter Assays-C2C12 cells were plated at 150,000 cells/well in 6-well plates (Nunc, Denmark) and transfected 16 h later using Lipofectamine Plus (Invitrogen) as described in the manufacturer's protocol. Equal amounts of DNA were transfected in duplicate wells, and total DNA transfected per well ranged from 2 to 4 g. Transfection efficiency was normalized by co-transfection of 80 ng of pRL␤-galactosidase as internal control. After 48 h, cells were rinsed with phosphate-buffered saline and lysed with 300 l of lysis buffer supplied with the CAT enzyme-linked immunosorbent assay kit (Hoffmann-La Roche). For myotubes, 200,000 cells/well were plated in a 6-well plate and transfected as above. The day after transfection the medium was changed from DMEM with 10% FBS to DMEM with 2% FBS. Lysates were prepared after 96 h. Lysates from myoblasts or myotubes were collected, and either CAT or luciferase promoter assays were performed on a luminometer (Turner BioSystems, Sunnyvale, CA). CAT activity was determined from a standard curve using a standard CAT enzyme provided by the manufacturer and expressed as CAT activity relative to that of the empty vector, pCAT. Luciferase activities were determined using the luciferase assay system from Promega (Mad-

Primers used
The reverse primers used for site-directed mutagenesis as well as for gel shift are not shown. Complementary primers based on the sequence of the forward primers were designed and annealed to the forward primers before use. The bold face indicates mutated nucleotides. Underlined characters represent E-boxes.

Primer name Sequence
ison, WI) and expressed as arbitrary units. ␤-Galactosidase activity was determined with a kit from Tropix (Applied Biosystems, Foster City, CA). All transfection experiments were done in duplicate and repeated at least three times. Results are expressed as mean Ϯ S.D.
Western Blotting-Whole cell lysates were prepared from C2C12 myoblasts or myotubes. In brief, cells were rinsed with several changes of cold cell rinse buffer (40 mM Tris-HCl, pH 7.4, 1 mM EDTA, 0.15 M NaCl, 0.1 M phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml pepstatin). Cells were scraped into a cell rinse buffer using a rubber policeman and pelleted. The pellet was resuspended in chilled cell resuspension buffer (40 mM Hepes, pH 7.9, 0.4 M KCl, 1 mM dithiothreitol, 10% glycerol). Protease inhibitors (0.1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml pepstatin) were added to the buffer just before use. The cells were subjected to three rounds of freezing and thawing. Debris was removed by spinning at 13,000 rpm at 4°C. The supernatant was collected, frozen in aliquots, and stored at Ϫ70°C. Protein levels were quantified by the BCA protein assay (Pierce). Equal amounts (20 -50 g) of total proteins from each sample were loaded onto FIGURE 1. ␣7 integrin promoter analysis. A, promoter activity of deletion constructs of mouse ␣7 integrin promoter. C2C12 myoblasts or 10T1/2 cells were transiently transfected with deletion constructs of mouse ␣7 integrin promoter. The deletion constructs used were p300 (Ϫ292 to ϩ3 bp), p400 (Ϫ401 to ϩ3 bp), p600 (Ϫ617 to ϩ3 bp), p1.2 (Ϫ1160 to ϩ3 bp), and p2.8 (Ϫ2809 to ϩ3 bp). Cell lysates were prepared 48 h after transfection, and CAT promoter activity was measured and expressed as activity relative to that of empty vector pCAT. Transfection efficiency was normalized by using the ␤-galactosidase vector as an internal control. Negative regulatory elements were present between Ϫ401 and Ϫ292 bp and between Ϫ2809 and Ϫ1160 bp. The control 10T1/2 cells did not express ␣7 integin and showed negligible ␣7 promoter activity. Data represent the mean of three separate experiments with error bars indicating S.D. B, promoter sequence analysis of the mouse ␣7 integrin gene. The proximal promoter region of the ␣7 integrin gene is shown. Potential E-box-like sites for ␦EF1(a/g/T/CACCT) and MYOD (CANNTG) were identified at Ϫ378, Ϫ333, Ϫ123, and Ϫ58 bp (boxed region) upstream of the translation initiation site in the mouse ␣7 promoter. Several Sp1 sites as well as an AP1-binding site were present as reported previously. An initiator-like sequence at Ϫ174 bp and the translation start site at ϩ1 bp are also indicated.
an 8% SDS-polyacrylamide gel. Proteins were separated by SDS-PAGE, transferred onto polyvinylidene difluoride transfer membranes (Millipore, Billerica, MA), blotted with specific antibodies, and detected with an ECL Western blotting kit (Amersham Biosciences) according to the protocol provided by the manufacturer. Antibody to mouse ␦EF1 was a gift from Dr. Yojiro Higashi (Osaka University, Japan). To detect ␣7 integrin, polyclonal 1211 antibody against mouse ␣7 integrin was used (21). Anti-MYOD antibody (sc760) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA) and tubulin antibody from NeoMarkers (Fremont, CA).
RT-PCR-C2C12 myoblasts were transiently transfected with either empty vector or an equal amount of ␦EF1 expression vector using Lipofectamine Plus as mentioned above. Total RNA was prepared after 48 h using the RNeasy kit from Qiagen (Valencia, CA). Equal amounts of control or ␦EF1-transfected RNA were reverse-transcribed to cDNA using Stratascript RT enzyme (Stratagene, La Jolla, CA). PCR was carried out as described previously (17) by using primers against ␣7 integrin (Ϫ378 F and ␣7 R) and GAPDH (mGAPDH F and mGAPDH R) listed in TABLE ONE. PCR products were analyzed on a 2% agarose gel.
Chromatin Immunoprecipitation (ChIP)-C2C12 myoblasts and myotubes were used for ChIP assay. A kit supplied by Upstate (Charlottesville, VA) was used following the manufacturer's protocol. C2C12 myoblasts and myotubes were grown in 100-mm plates as described above. Proteins bound to DNA were fixed by the addition of formaldehyde (final 1%) to the medium. Cells were washed with cold phosphatebuffered saline and lysed with SDS-Lysis buffer supplied with the kit. The DNA in the lysate (from 1 ϫ 10 6 cells) was sheared to between 500 and 1000 bp by sonication using a Vibracell sonicator (Sonics Materials, Inc., Newtown, CT) set to 25% output (four 10-s pulses with cooling between pulses). The lysate was further processed as per the manufacturer's protocol. Immunoprecipitation was carried out by incubation overnight with antibodies against ␦EF1 (from Dr. Yohiro Higashi, Osaka University, Japan), MYOD (Santa Cruz Biotechnology), or control IgG antibody. DNA-antibody complexes were collected and processed as per the kit protocol. Eluted DNA was resuspended in 30 l of TE buffer (10 mM Tris, pH 8.0, 1 mM EDTA). For PCR, primers that span the ␦EF1 sites at Ϫ378 and Ϫ333 bp were used (Ϫ378 F and ␣7 R), see TABLE ONE. Thirty five cycles of PCR were performed, and the amplified products were run on a 2% agarose gel.

RESULTS
␦EF1 Is a Negative Regulator of ␣7 Integrin Expression-Previous studies have shown that ␣7 integrin remains at basal levels in myoblasts, but mRNA and protein expression are strongly elevated following myodifferentiation (17,22). To gain an understanding of the promoter regulatory regions that control ␣7 integrin levels during differentiation, several deletion constructs were analyzed for promoter activity in C2C12 myoblasts. As shown in Fig. 1A, the region from Ϫ401 to Ϫ292 bp displays significant negative regulation (compare p400 and p300). Two E-box sites are present in this region (Fig. 1B); however, we did not identify binding sites for any other known potential repressors. E-box sites are known to be involved in muscle gene activation through helixloop-helix proteins of the MRF family (MYOD, myf-5, myogenin, and MRF). A search for E-box-binding myogenic repressors identified the ␦EF1 transcription factor. A consensus ␦EF1-binding site (CACC(T/ A)GGTG) is present at Ϫ378 bp upstream of the translation start site (Fig. 1B), and a second potential binding site GACCTG is also present at Ϫ333 bp.
To test if ␦EF1 has a role in the regulation of ␣7 integrin, we transfected C2C12 mouse myoblasts with an expression vector for ␦EF1. Myoblasts transfected with the ␦EF1 expression vector showed a strik-ing decrease in ␣7 integrin expression at the protein level by Western blot compared with cells transfected with a control vector ( Fig. 2A). Tubulin levels indicate equal protein loading. Most interestingly, MYOD levels were unchanged following ␦EF1 overexpression, indicating that down-regulation in ␣7 integrin levels is not an indirect consequence of lower MYOD levels. To examine whether ␦EF1-induced loss of ␣7 expression resulted from transcriptional down-regulation, we transfected C2C12 myoblasts with a full-length 2.8-kb mouse ␣7-CAT promoter construct and performed promoter activity assays. Cells were co-transfected with either an empty expression vector or with increasing concentrations of ␦EF1 expression vector. As seen in Fig. 2B, in the FIGURE 2. ␦EF1 down-regulates ␣7 integrin expression. A, effect of ␦EF1 on ␣7 integrin protein expression. C2C12 myoblasts were transiently transfected with either 4 g of empty vector or the expression vector for ␦EF1. Cell lysates were analyzed by Western blotting for expression levels of ␣7 integrin, MYOD, and ␦EF1. Following overexpression of ␦EF1, ␣7 integrin expression was significantly decreased, whereas MYOD levels did not change. Tubulin levels indicate protein loading. B, overexpression of ␦EF1 represses ␣7 integrin promoter activity. C2C12 cells were transiently transfected with the fulllength promoter p2.8 and co-transfected with either empty pcDNA vector or with increasing amounts of ␦EF1 expression vector. CAT activity was determined as in Fig. 1. A dose-dependent decrease in promoter activity was induced with increasing amounts of ␦EF1 vector. The data represent the mean Ϯ S.D. of three independent experiments. C, effect of ␦EF1 on ␣7 integrin mRNA levels. C2C12 cells were transiently transfected with either empty vector or ␦EF1 vector. RNA was prepared after 48 h, and RT-PCR for ␣7 integrin and GAPDH was performed as described under "Experimental Procedures." ␦EF1 induced a decrease in ␣7 integrin mRNA. A, mutation of putative ␦EF1-binding sites at Ϫ378 and Ϫ333 bp on the ␣7 promoter resulted in release of repression. C2C12 cells were transiently transfected with either the wild-type p400 construct or with p400mut1 (mutant Ϫ378-bp site) or p400mut2 (mutant Ϫ378and Ϫ333-bp sites). CAT activity was determined and expressed as CAT activity relative to that of empty vector pCAT. Mutation of the E-box sites at Ϫ378 and Ϫ333 bp resulted in release of repression. To test if repression was DNA-binding dependent, C2C12 myoblasts were transfected with wild-type p400 or p400mut2 and co-transfected with either control vector or expression vector for ␦EF1. Overexpression of ␦EF1 resulted in repression of the wild-type p400 as well as the mutant p400 construct, indicating the possibility of repression through non-DNA binding mechanisms or the presence of additional ␦EF1-binding sites. B, mutation of putative ␦EF1-binding sites at Ϫ123 and Ϫ58 bp resulted in minimal release of repression. C2C12 cells were transiently transfected with the wild-type p300 construct or p300mut2 (mutant E-box sites at Ϫ123 and Ϫ58 bp). CAT activity was determined as above. Mutation of the Ϫ123and Ϫ58-bp sites resulted in a slight increase in promoter activity. To determine whether ␦EF1 could repress the wild-type p300 and mutant p300 constructs, C2C12 cells were transfected with the wild-type or p300mut2 construct and co-transfected with either control vector or expression vector for ␦EF1. Lysates were prepared, and CAT activity was determined as above. Overexpression of ␦EF1 resulted in a slightly greater repression of the wild-type p300 promoter as compared with the p300mut2 construct. Repression of both constructs indicated that ␦EF1 may bind the wild-type and mutant sites when overexpressed. C, the E-box-like sites at Ϫ123and Ϫ58-bp function in activation. C2C12 myoblasts were transiently transfected with the wild-type p300mut1 or p300mut2 promoter fragments and co-transfected with either the control vector or MYOD expression vector and processed as above. MYOD activated the wild-type promoter but mutation of the Ϫ123and Ϫ58-bp sites prevented the MYOD-induced activation of the p300mut2 construct. D, transactivation by Sp1 was not blocked by ␦EF1. C2C12 cells were transiently transfected with the GAL4 reporter construct GAL4Luc and co-transfected with expression vectors for fusion proteins GAL4Sp1 or GAL4Sp1 and ␦EF1. Luciferase activity was measured as described under "Experimental Procedures." Transfection efficiency was normalized by using the ␤-galactosidase vector as an internal control. Results reflect three independent experiments expressed as the means Ϯ S.D. The results indicate that ␦EF1 does not modify transactivation of promoter activity by Sp1. presence of increasing concentrations of ␦EF1, ␣7 promoter activity decreased progressively and plateaued with maximal inhibition near 70% with 3-4 g of ␦EF1, indicating that ␦EF1 decreases ␣7 integrin expression through a transcriptional mechanism. In order to confirm repression at a transcriptional level, we also performed semi-quantitative RT-PCR for ␣7 integrin in C2C12 cells transfected with either empty vector or ␦EF1 expression vector. As seen in Fig. 2C, ␣7 integrin mRNA levels decrease when ␦EF1 is overexpressed, confirming that ␦EF1 exerts its effect at a transcriptional level.
Mutation of Putative ␦EF1 Sites Results in Release of Repression-In order to determine whether the two potential ␦EF1-binding sites at Ϫ378 and Ϫ333 bp function in negative regulation, we used site-directed mutagenesis to generate mutant binding sites and then performed CAT promoter activity assays. We used the p400(Ϫ401 to ϩ3) construct to generate p400mut1 having a mutant Ϫ378-bp E-box as well as a construct p400mut2 with mutant E-box sites at Ϫ378 and Ϫ333 bp. C2C12 cells were transfected with wild-type or mutant constructs, and promoter activity was assayed. Mutation of the Ϫ378-bp site resulted in release of repression (activity increased over 2-fold), verifying the site's function as a negative regulatory element (Fig. 3A). Mutation of the site at Ϫ333 bp resulted in only a small additional increase in promoter activity, indicating that this site may have a minor role in suppression of ␣7 integrin expression. These results indicate that endogenous ␦EF1 may be repressing the ␣7 promoter by binding at the Ϫ378-bp site and that the mechanism of repression by ␦EF1 depends on its targeting these sequences.
We were interested in determining whether DNA binding by ␦EF1 is the only mechanism of repression. To test this, C2C12 cells were cotransfected with ␦EF1 and either wild-type or mutant p400 constructs. If ␦EF1 repression requires DNA binding, then mutation of the ␦EF1binding sites should prevent repression. As seen in Fig. 3A, ␦EF1 was able to repress the wild-type p400 as well as the mutant p400mut2 construct, indicating that non-DNA binding mechanisms may be operative. It is also possible that additional unidentified ␦EF1-binding sites may be present in the p400 construct. We identified an AGGTG site located at Ϫ123 bp and a GACCTG site at Ϫ58 bp upstream of the ATG translation start site, respectively. We examined if these two sites are involved in repression by ␦EF1 by using a shorter promoter fragment, p300 (Ϫ292 to ϩ3 bp), that has the Ϫ123and Ϫ58-bp sites but lacks the two ␦EF1 sites at Ϫ378 and Ϫ333 bp. The site at Ϫ123 bp was mutated to give the construct p300mut1. A second construct p300mut2 contained mutations at Ϫ123 and Ϫ58 bp. C2C12 cells were transfected with the wild-type p300 or the mutant p300mut2 promoter constructs, and promoter activity was assayed. As seen in Fig. 3B, mutation of these two sites in the p300 fragment resulted in only a slight increase in promoter activity, indicating that these sites may not be important in negative regulation. We tested whether DNA binding is the exclusive mechanism of repression by transfecting C2C12 cells with ␦EF1 and wild-type p300 or mutant constructs. ␦EF1 could repress the wild-type p300 (activity decreased to ϳ50%) as well as the mutant p300mut2 construct (activity decreased to ϳ69%) (Fig. 3B). Because we did not identify any additional ␦EF1-binding sites in the p300 construct, we tested the possibility that, under conditions of overexpression, ␦EF1 is able to bind the mutant E-box. We found by electrophoretic mobility shift assay (EMSA) that the mutant site does compete for binding ␦EF1, although with lower affinity than wild-type ␦EF1 (data not shown), and may explain the ability of ␦EF1 to repress the mutant p400 and p300 constructs. ␦EF1 may repress the p300 promoter through non-DNA binding mechanisms, but further work is needed to test this possibility.
Because the two E-box sites in the p300 construct did not appear to play a major role in repression, we next assessed whether these two sites might activate the ␣7 integrin through E-box binding by MYOD. C2C12 myoblasts were transfected with myoD and the wild-type p300 construct or mutant p300mut1 and p300mut2 constructs. We found that MYOD could activate the wild-type p300 construct (Fig. 3C); however, mutation of the Ϫ123-bp site (p300mut1) resulted in decreased activation by MYOD. Further mutation of both sites led to a complete loss of activation by MYOD (p300mut2). These results suggest that ␦EF1 represses the ␣7 promoter through the Ϫ378-bp site and that MYOD activates the promoter through E-box sites at Ϫ123 and Ϫ58 bp immediately upstream of the translation start site. MYOD is known to activate the full-length p2.8 ␣7 promoter (16,17), and the current experiments have identified additional MYOD-binding sites in the ␣7 promoter.
To examine if a non-DNA binding mechanism of repression might also be operative, we investigated whether ␦EF1 interacts with other transcription factors that bind the ␣7 promoter within the Ϫ401to ϩ3-bp region. We used a GAL4 reporter assay to test the effect of ␦EF1 on transactivation by Sp1, as several Sp1 sites are present in this region. As seen in Fig. 3D, ␦EF1 had no effect on transactivation by Sp1. This result indicates that ␦EF1 does not repress through an effect on Sp1. It is possible that ␦EF1 may target other transcription factors, such as AP1, which we have not yet tested.
Identification of Potential ␦EF1-binding Sites; Repression Correlates with ␦EF1 Binding-Functional studies on the ␣7 promoter identified the Ϫ378-bp site as responsible for repression and the Ϫ123and Ϫ58-bp sites as involved in activation. We wanted to analyze the binding of ␦EF1 and MYOD to the various E-box-like sites in this region. The consensus binding site for ␦EF1 is a/g/T/CACCT or the palindromic AGGTG/A/c/t; the lowercase represents a lower frequency of occurrence of the nucleotide in the ␦EF1 consensus binding site (23). The AAGGTG site at Ϫ378 bp is a consensus ␦EF1-binding site (23). The GACCTG site at Ϫ333 bp may also bind ␦EF1. In order to check for ␦EF1 binding at the Ϫ378and Ϫ333-bp sites, we used EMSA and supershift assay using C2C12 nuclear extracts. In these assays, ␦EF1 was found to bind to the Ϫ378-bp site, where a single band was shifted by the ␦EF1 but not to the Ϫ333-bp site (Fig. 4A, left panel). Similarly, EMSA also indicates that MYOD can bind the E-boxes at Ϫ378 and Ϫ333 bp. MYOD may thus compete with ␦EF1 for binding and displace ␦EF1 from the Ϫ378-bp site when MYOD levels are high, for example, in differentiating myotubes. Additionally, the results on binding correlate with the results on mutation of the Ϫ378-bp site which resulted in release of repression (Fig. 3A). Two additional sites that may bind ␦EF1 are present at Ϫ123 and Ϫ58 bp. We tested ␦EF1 and MYOD binding to MYOD and MEF2 could overcome repression of ␣7 promoter by ␦EF1, indicating a reciprocal activity with ␦EF1. Similarly, C2C12 cells were transfected with the p400 or the full-length p2.8 ␣7 promoter and co-transfected with ␦EF1 following induction of differentiation as described under "Experimental Procedures." Cell lysates were prepared from myotubes, and promoter activity was determined. The results show that ␦EF1 failed to repress the ␣7 promoter under physiological conditions of myoblast differentiation. C, ␦EF1 expression during myoblast differentiation to myotube. Equal amounts of whole cell lysates prepared from C2C12 myoblasts or differentiated myotubes were separated by SDS-PAGE and blotted with anti-␦EF1 antibody. The results indicate that ␦EF1 levels did not change during myoblast differentiation. D, ␦EF1 binding using ChIP analysis. In vivo binding of ␦EF1 and MYOD to the ␣7 integrin promoter is shown. DNA-protein cross-linking was carried out as detailed under "Experimental Procedures." Antibodies to ␦EF1 and MYOD were used to immunoprecipitate promoter-bound factors. Input is DNA not processed for immunoprecipitation. Negative control represents a sample that was immunoprecipitated by using normal IgG. PCR was carried out using primers (Ϫ378F and ␣7R) spanning the p400 region. The results indicate that ␦EF1 but not MYOD preferentially binds the ␣7 promoter in C2C12 myoblasts, whereas MYOD but not ␦EF1 binds to the ␣7 promoter after myoblast differentiation. the putative binding site at Ϫ123 bp (GAGGTG) and the Ϫ58-bp site (GACCTG). We found that ␦EF1 appears to bind the Ϫ58-bp site but not the Ϫ123-bp site, whereas MYOD binds both sites (Fig. 4A, center  panel). We reported earlier that mutation of the Ϫ123and Ϫ58-bp sites indicated that these sites may be more relevant in activation by MYOD, rather than repression by ␦EF1 (Fig. 3C). These results correlate with the EMSA results that indicate that MYOD, rather than ␦EF1, binds at these sites.
Role of ␦EF1 during Myodifferentiation-The above results indicate that ␦EF1 and MYOD may compete for binding the Ϫ378-bp site in the ␣7 promoter. We tested whether MYOD and ␦EF1 have reciprocal effect on ␣7 promoter activity. We transfected C2C12 cells with the full-length p2.8 and p400 fragment of the ␣7 promoter. MYOD and MEF2 overcame repression by ␦EF1, indicating that ␦EF1 and MYOD have reciprocal effects on ␣7 promoter activity (Fig. 4B). We further studied this activity in the physiological context of myoblast-to-myotube differentiation by assessing repression of ␣7 promoter activity by ␦EF1 in differentiating C2C12 cells. We found that ␦EF1 failed to repress the p400 or p2.8 promoter constructs on myodifferentiation (Fig. 4B). This may be due to loss of ␦EF1 binding to the Ϫ378-bp site during myodifferentiation. We also used EMSA analysis of nuclear extracts prepared from differentiating myotubes to assess binding of ␦EF1 to the Ϫ378-bp site. ␦EF1 did not bind to this site after myogenic differentiation as evidenced by the absence of band shift by anti-␦EF1 antibody; however, MYOD retained its binding to this site as shown by band shift with anti-MYOD antibody (Fig. 4A, right panel).
The possibility that MYOD may displace ␦EF1 from its binding site during myodifferentiation was further explored. Although it is well known that MYOD levels rise during myodifferentiation, the relative protein levels of ␦EF1 appear to be constant in C2C12 myoblasts and myotubes (Fig. 4C). We next used chromatin immunoprecipitation to define further the relative interaction of ␦EF1 and MYOD with the proximal promoter region in vivo. ChIP assays were performed using anti-␦EF1 and anti-MYOD antibodies to immunoprecipitate ␣7 promoter fragments bound by the two transcription factors. Antibody to ␦EF1 immunoprecipitated the ␣7 promoter fragment from myoblast (primers amplified a region between Ϫ378 and ϩ3 bp) as identified by PCR (Fig. 4D). In contrast, in myoblasts, anti-MYOD failed to yield any detectable product in the Ϫ378to ϩ3-bp region of the ␣7 promoters. The situation was reversed after myoblast differentiation to myotubes, where antibody to MYOD effectively immunoprecipitated the region between Ϫ378 and ϩ3 bp, but anti-␦EF1 antibody failed to recover ␣7 promoter fragments in this region. This indicates that the mechanism by which ␣7 integrin levels increase during differentiation may be due to the reciprocal loss of ␦EF1 binding and an increase in MYOD binding to the ␣7 promoter in the negative regulatory region (Ϫ401 to ϩ3 bp).
Mechanism of Repression by ␦EF1-ChIP assays indicated that ␦EF1 binds to the endogenous gene, suggesting a direct DNA-binding mechanism of repression and possibly a displacement of ␦EF1 by MYOD during muscle differentiation. We further investigated the domains of ␦EF1 required for repression of the ␣7 promoter by using the following three vectors expressing truncated forms of ␦EF1: construct CZF expressed the C-zinc finger region, construct CZF-HD expressed C-zinc fingers/homeodomain region, and construct CZF-HD-NZF expressed C-zinc fingers/homeodomain/N-zinc fingers. Only the fulllength ␦EF1 protein was able to fully repress the ␣7 promoter (Fig. 5A). Vectors expressing CZF, CZF-HD, or CZF-HD-NZF were not able to efficiently repress the ␣7 promoter. This indicates that a crucial repression domain lies upstream of the N-zinc fingers in the region called the NR (negative region) domain. Most interestingly, the NR-containing region is known to bind p300/CBP (CREB-binding protein) co-activator proteins (24). Subsequently, we tested whether overexpression of CBP could reverse repression because of ␦EF1. In promoter assays, we found that overexpression of MYOD could lead to increase activity, but this increase was lost in the presence of higher levels of ␦EF1 (Fig. 5B). Furthermore, CBP could reverse the repression by ␦EF1, indicating that ␦EF1 can compete for limited amounts of CBP. Similar results were obtained with full-length p2.8 promoter fragment (results not shown). The ability of CBP to overcome repression by ␦EF1 is comparable with that of MYOD (Fig. 5B). FIGURE 5. Identification of ␦EF1 domains important for repression. A, the N-terminal region of ␦EF1 is needed for repression of ␣7 integrin promoter activity, indicating an active mechanism of repression. C2C12 myoblasts were transiently transfected with fulllength ␣7 integrin p2.8 promoter or with control vector, pCAT. Cells transfected with p2.8 were co-transfected with either full-length ␦EF1 vector or with a deletion construct CZF (C-zinc finger region), CZF-HD (C-zinc finger and homeodomain region), or CZF-HD-NZF (C-zinc finger and homeodomain and N-zinc finger region). Cell lysates were prepared 48 h after transfection, and CAT promoter activity was determined as described under "Experimental Procedures." Transfection efficiency was normalized using the ␤-galactosidase vector as an internal control. Results of three independent experiments are plotted as the means Ϯ S.D. The results indicate that only full-length ␦EF1 repressed the ␣7 promoter. B, CBP relieves repression by ␦EF1. C2C12 cells were transiently transfected with the p400 fragment of the ␣7 integrin promoter construct or a control pCAT construct. Cells transfected with the p400 construct were co-transfected with either empty vector or with expression vectors for ␦EF1, ␦EF1, and MYOD or ␦EF1 and CBP. CAT activity was measured as described under "Experimental Procedures." The results indicate that CBP could overcome repression by ␦EF1.

DISCUSSION
Regulation of the ␣7 integrin during myodifferentiation is complex. Controlled expression of this integrin is important because it contributes to mechanical stability during muscle contraction, and its loss leads to congenital myopathy (6,25). Importantly, ␣7 integrin expression in myoblasts remains low, facilitating cell motility needed during tissue remodeling and alignment of cells prior to fusion and myotube formation. Induction of differentiation to myotubes generates more than a 20-fold increase in ␣7 mRNA levels and a parallel increase in promoter activity (17,22). Our previous studies have shown that the ␣7 promoter is tightly regulated, and gene expression is intricately modulated by MYOD family members as well as c-MYC (17). A negative regulatory element has been identified in the proximal promoter that contains E-box-like sequences that may be potential binding sites for ␦EF1. The current results indicate a role for ␦EF1 in suppression of ␣7 promoter activity through targeting this regulatory region, whereas during myotube differentiation this inhibitory activity is attenuated by increased levels of competing MYOD.
We found that forced expression of ␦EF1 in C2C12 myoblasts induced a major reduction in ␣7 integrin levels, suggesting a role for ␦EF1 as a selective negative regulator of ␣7 integrin gene transcription. MYOD levels were unchanged following ␦EF1 overexpression, indicating that ␦EF1 does not repress the ␣7 integrin gene by modulating MYOD levels. ␦EF1 may control ␣7 promoter activity by competing with MYOD for binding sites. We found by gel shift that ␦EF1 and MYOD compete for binding the ␣7 promoter in the negative regulatory region and that ␦EF1 and MYOD have opposite effects on ␣7 promoter activity. We tested for differential binding of ␦EF1 and MYOD during myogenic differentiation by ChIP analysis. We found that in myoblasts ␦EF1 preferentially binds the ␣7 promoter. However, in myotubes binding efficiency to the promoter shifts to MYOD, thereby releasing the repression and enhancing ␣7 gene transcription during differentiation.
However, simple displacement of MYOD by ␦EF1 does not seem to be the mechanism by which ␦EF1 represses the ␣7 promoter. MYOD binds E-boxes with the consensus sequence CANNTG, whereas ␦EF1 binds the E-box-like sequence a/g/T/CACCT. Because ␦EF1 binds only a subset of E-boxes, it is therefore not clear how it represses myogenesis. We found that only the full-length ␦EF1 construct could repress ␣7 integrin expression. We tested deletion constructs having the C-terminal zinc fingers (CZF), the C-terminal zinc fingers plus the homeodomain CZF-HD, or the C-terminal zinc fingers homeodomain and N-terminal zinc fingers CZF-HD-NZF. All of these constructs failed to significantly repress ␣7 integrin expression, indicating that the N-terminal domain of ␦EF1 is needed for repression. Sekido et al. (23) also reported that an NR (negative region) domain upstream of the N-terminal zinc fingers is needed for repression of the ␦-crystallin enhancer. However, ZEB, the human homolog of ␦EF1, represses the ␣4 integrin promoter in muscle through the region between the N-terminal zinc fingers and homeodomain. Neither the N-terminal nor the C-terminal zinc fingers were needed for repression of the ␣4 integrin (18). The same authors reported that the C-terminal domain of ZEB repressed muscle genes by targeting the trans-activation domain of MEF2 (19). This shows that ␦EF1 and ZEB are homologs that differ significantly in the domains needed for muscle-gene repression. However, both ZEB and ␦EF1 target MEF2, a gene essential for muscle gene expression.
In our study, the co-activator CBP could overcome repression by ␦EF1. It is known that ␦EF1 binds p300/CBP through a region upstream of the N-zinc fingers (24); therefore, our results suggest that ␦EF1 binds p300/CBP and may compete with MYOD for limited amounts of p300/ CBP. p300/CBP has histone acetylase activity that opens up the chromatin to allow transcription. MYOD and MEF2 activate muscle genes by binding p300 and using the histone acetylase activity of p300/CBP to allow transcription (26,27). Unfortunately, the exact position where p300/CBP binds ␦EF1 is not known, and it is currently not possible to test whether mutation of the CBP-binding sites of ␦EF1 might cause ␦EF1 to lose the ability to repress the ␣7 integrin gene. Postigo and Dean (19) also found that forced expression of p300 overcame the repression of c-myb and ets sites by ZEB in hematopoietic cells as well as the repression of MEF2 and NFB. However, in their study, region 1 between the N-zinc fingers and the homeodomain was implicated as targeting p300 activity. Because ␦EF1/ZEB binds p300 at the extreme N terminus upstream of the N-zinc fingers, it is not clear how region 1 may affect p300 activity. It seems that multiple domains in ZEB target the co-activator p300/CBP.
We reported earlier that c-Myc down-regulates ␣7 integrin expression (17). Normal levels of p300/CBP are needed to keep c-MYC repressed, and depletion of p300/CBP is known to result in raised c-MYC levels and inappropriate entry into S-phase (28). It is possible that titration of limited amounts of p300/CBP by ␦EF1 results in raised c-MYC levels and that c-MYC and ␦EF1/ZEB thereby cooperate to regulate muscle gene expression. It will be interesting in future studies to examine if ␦EF1 and c-MYC synergize to down-regulate ␣7 integrin expression in myoblasts.
In summary, this study shows that ␦EF1 controls ␣7 integrin expression in myoblasts by competing with MYOD for binding to the negative regulatory region of the ␣7 integrin promoter. During skeletal myogenesis increased levels of MYOD displace ␦EF1 from the promoter inducing elevated ␣7 integrin levels. The mechanism by which ␦EF1 represses the ␣7 integrin promoter is more complex than simply passive displacement of MYOD. Rather, it is likely that the ability of ␦EF1 to compete for limited amounts of the co-activator p300/CBP is responsible for repression of ␣7 integrin in myoblasts. Following differentiation into myotubes, sustained high levels of MYOD continue to override ␦EF1 repression, resulting in persistent elevated ␣7 integrin expression.