The role of an E box binding basic helix loop helix protein in the cardiac muscle-specific expression of the rat cytochrome oxidase subunit VIII gene.

We have characterized the rat gene for muscle-specific cytochrome oxidase VIII (COX VIII(H)) and mapped the distal promoter region responsible for transcription activation in C2C12 skeletal myocytes and H9C2 cardiomyocytes. In both cell types, the promoter elements responding to the induced differentiation of myocytes map to two E boxes, designated as E1 and E2 boxes with a core sequence of CAGCTG. Gel mobility shift analysis showed that both E1 and E2 box motifs form complexes with nuclear extracts from H9C2 cardiomyocytes that were supershifted with monoclonal antibody to E2A but not with antibody to myo-D. Extracts from induced and uninduced H9C2 cardiomyocytes yielded different gel mobility patterns and also different E2A antibody supershifts suggesting a difference in the DNA-bound protein complexes cross-reacting with the E2A antibody. Transcriptional activity of the promoter construct containing intact E boxes was inhibited by coexpression with Id in differentiated H9C2 cardiomyocytes. Our results show the involvement of an E box binding basic helix loop helix protein in the cardiac muscle-specific regulation of the COX VIII(H) promoter.

We have characterized the rat gene for muscle-specific cytochrome oxidase VIII (COX VIII(H)) and mapped the distal promoter region responsible for transcription activation in C2C12 skeletal myocytes and H9C2 cardiomyocytes. In both cell types, the promoter elements responding to the induced differentiation of myocytes map to two E boxes, designated as E1 and E2 boxes with a core sequence of CAGCTG. Gel mobility shift analysis showed that both E1 and E2 box motifs form complexes with nuclear extracts from H9C2 cardiomyocytes that were supershifted with monoclonal antibody to E2A but not with antibody to myo-D. Extracts from induced and uninduced H9C2 cardiomyocytes yielded different gel mobility patterns and also different E2A antibody supershifts suggesting a difference in the DNA-bound protein complexes cross-reacting with the E2A antibody. Transcriptional activity of the promoter construct containing intact E boxes was inhibited by coexpression with Id in differentiated H9C2 cardiomyocytes. Our results show the involvement of an E box binding basic helix loop helix protein in the cardiac muscle-specific regulation of the COX VIII(H) promoter.
Cytochrome c oxidase (COX) 1 is the terminal enzyme of the mitochondrial electron transport chain, which catalyzes the transfer of electrons from reduced cytochrome c to molecular oxygen and concomitantly generates the transmembrane H ϩ gradient by transferring protons outside the mitochondrial inner membrane (1). The mammalian COX complex contains 13 different subunits, of which the 3 largest catalytic subunits are mitochondrial encoded and the remaining 10 smaller subunits are encoded by nuclear genes (1). The structural organization of a number of nuclear genes from different mammalian sources (2)(3)(4)(5)(6)(7)(8) and the promoter elements of the ubiquitously expressed COX IV and Vb genes have been extensively investigated (9 -12). Both the COX IV and Vb genes contain functionally important GABP factor (13) (variably referred to as NRF2 (12) or E4TF1 (14)) binding sites at or around the transcription initiation sites as well as at upstream promoter regions (3,9,10,15). The COX Vb promoter also contains multiple YY-1 binding sites that are important for both transcription initiation and tissue-dependent transcription suppression (9,16). Additionally, both of the promoters contain functionally important NRF1 factor binding sites (17). Recently, bovine genes for the muscle-specific isologs VIa(H) and VIIa(H) have been sequenced (4,5,18). However, the nature of the promoter elements controlling cardiac and skeletal muscle-specific expression of these genes has not yet been investigated.
Transcription regulation of a large majority of skeletal muscle-specific genes is dependent on E box binding bHLH transcription factors (19 -22), although factors binding to immediately flanking CArG or A/T-rich MEF2 motifs in conjunction with the E box binding protein appear to be important for the expression of some muscle-specific promoters (23,24). Myo-D, myogenin, MRF4, and myf-5 are members of the skeletal muscle-specific class II bHLH proteins that are important determinants of the skeletal muscle lineage (22,25). In particular, heterodimerization between the class I type ubiquitous E2A family proteins with cell-and tissue-specific class II proteins plays a pivotal role in the transactivation of skeletal musclespecific genes. In contrast, binding of E2A homodimers to E box motifs has been reported to play a key role in the B cell-specific expression of the immunoglobulin heavy chain gene (26,27). The expression of muscle-specific actin, MCK, troponin C, and other genes in cardiac cells has been shown to depend upon serum response factor binding to CArG box sequences (24,28), factor MEF2 binding to A/T-rich sequence motifs (29,30), or GATA-4 factor binding to its cognate sequence motifs (31). Immunohistochemical (32) and gel mobility shift (24) studies initially suggested the presence of bHLH proteins in developing cardiac muscle cells. More recently three different groups (33)(34)(35) have independently characterized bHLH factors dHAND and eHAND (variably termed Th1, Th2 or HXT, HED), which may have roles in the cardiac muscle development. These factors are detected only at certain stages during chick and mouse embryogenesis, and currently their role in the regulation of muscle-specific genes in differentiated cardiac muscle cells remains unclear. In the present paper we have characterized the rat gene for the muscle-specific COX VIII(H) and found that a tandemly duplicated consensus E box is necessary for transcription activation of the promoter in both skeletal and cardiac muscle cell lines. We also show that a bHLH E box binding superfamily factor is involved in transcriptional activation of the promoter in differentiated H9C2 cardiomyocytes.

MATERIALS AND METHODS
Isolation and Sequencing of the COX VIII(H) Gene-A rat genomic library (Stratagene Inc.) in Lambda Dash vector was screened using 32 P-labeled rat COX VIII(H) cDNA probe (36) by the colony hybridization method. DNA from positive clones was isolated and subjected to restriction analysis by digestion with SalI, NcoI, BamHI, and AvaI enzymes individually or in combination. Restriction fragments hybridizing with the COX VIII(H) cDNA probe were subcloned into pGEM 5z or 7z plasmids (Promega Biotech Corp.) and sequenced in both directions using the Sequenase sequencing kit (U.S. Biochemical Corp.) by the dideoxy chain termination method of Sanger et al. (37). S1 Nuclease Protection and Primer Extension Analyses-S1 nuclease protection and primer extension analyses were carried out to determine the transcription start site of the rat COX VIII(H) gene. A 19-mer antisense primer (5ЈTTGGAGCAGCCGCAGGATA3Ј) complementary to nucleotides 85-103 of the cDNA sequence was 5Ј-end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase to a specific activity of 5 ϫ 10 5 cpm/ng and used for generating the S1 probe essentially as described before (2). About 3 ϫ 10 5 cpm of the 194-nucleotide-long single-stranded probe containing the first exon and the 5Ј-flanking sequence was annealed with 100 g each of total RNA from rat liver, heart, skeletal muscle, or Escherichia coli tRNA and subjected to S1 digestion using 150 -200 Vogt units of S1 nuclease at 37°C for 30 min as described (38). The same 5Ј-end-labeled primer (1 ϫ 10 6 cpm) was used for the avian myeloblastosis virus reverse transcriptase-based primer extension analysis as described (2). The primer extension and the S1 nucleaseresistant products were resolved on a 6% polyacrylamide, 8 M urea sequencing gel.
Reporter Plasmid Construction and Mutagenesis-The parent promoter plasmid, Ϫ994CAT was constructed by cloning a PCR-amplified 1044-base pair DNA fragment consisting of the Ϫ994/ϩ50 sequence of the rat COX VIII(H) gene in SalI and blunt-ended XbaI sites of the promoter-less, enhancer-less CAT reporter plasmid, pCAT basic (Promega Biotech Corp). The Ϫ994CAT DNA was digested with HindIII/ SacI, HindIII/ApaI, or PstI alone followed by blunt ending and recircularization of the vector to generate the 5Ј deletion clones Ϫ579CAT, Ϫ158CAT, and Ϫ44CAT, respectively. Point mutations were targeted to the two E boxes contained within the Ϫ158CAT DNA by overlap extension PCR. Specific base substitutions at the E1 and E2 boxes are as shown for the E1mut and E2mut oligonucleotides in Table I.
Transient Transfection in Myogenic Cell Lines-Conditions for the growth of COS cells and transfection were described previously (9,11). C2C12 and H9C2 myocytes in logarithmic growth were seeded at an initial plating density of 5 ϫ 10 5 cells/10-cm dish in a medium supplemented with 10% fetal bovine serum and 50 g/ml gentamycin or 100 g/ml penicillin/streptomycin. After an overnight recovery period, one group of cells was transfected and harvested 2 days later as subconfluent myocytes (50 -60% confluent). A companion group of transfected cells was grown for 2 additional days to 80 -100% confluency and induced by supplementing the medium with 2% heat-inactivated fetal bovine serum. Cells were harvested after 2 days when 70 -80% of the cells had fused to form multinucleated myotubes. In all cases cells were transfected in replicate plates with CsCl-purified CAT reporter plasmid DNAs (5-10 g/10-cm dish) using the calcium phosphate coprecipitation method (39). Cell extracts were assayed for CAT activities using [ 14 C]chloramphenicol (40).
Cotransfections with E2-5, Id1, and myo-D expression constructs (generously provided by Dr. Tom Kadesch) and dHAND and eHAND expression plasmids (generously provided by Dr. Eric Olson) were carried out in C2C12 and H9C2 cell lines. In each experiment 8 g of reporter DNA alone or in combination with 3-30 g of expression plasmids were used. Cytomegalovirus vector DNA was added to make up the final DNA concentration in both control and experimental dishes to the same level.
Preparation of Nuclear Extracts and DNA-Protein Binding by Gel Mobility Shift-Nuclear extracts from induced and uninduced myocytes and other cells were prepared as described by Dignam et al. (41). Miniprep nuclear extract from induced H9C2 myocytes was prepared by the procedure of Schreiber et al. (42), and rat heart nuclear extract was prepared according to Gorski et al. (43). DNA-protein binding was assayed by gel mobility shift essentially as described previously (44) using 0.1 to 0.2 ng (ϳ20,000 cpm) of double-stranded DNA probes, end-labeled with 32 P. Wild type and mutant oligonucleotides used as probes or competitors are listed in Table I. Binding reactions (20 l final volumes) were carried out on ice for 25 min using 1-4 l (5-20 g of protein) of nuclear extracts or 0.1-0.3 g of purified protein and 1 g of poly[d(I⅐C)] to minimize nonspecific binding, as described previously (44). The DNA-bound complexes were resolved by electrophoresis through 4% polyacrylamide gels using 0.25 ϫ TBE (1 ϫ TBE ϭ 89 mM Tris base, 89 mM Boric acid, and 2 mM EDTA) as the running buffer. Competition with 100-fold molar excess (20 ng) of specific or nonspecific competitor DNAs was carried out as described (10,11). For antibody supershift assays, 1-2 g of either preimmune IgG or myo-D and E2A-specific antibodies (Santa Cruz Biotech Corp.) were added to the reaction mixture and incubated for 20 min at room temperature, followed by the addition and incubation with the labeled probe as described above.

FIG. 1. Nucleotide sequence of the rat COX VIII(H) gene.
The intron sequence is presented in lowercase letters, and the intron size is indicated. The transcription start site at ϩ1 was based on the S1, and primer extension analyses are presented in Fig. 2. Various protein binding motifs identified by a computer search have been presented in bold letters and are double underlined. Translation start (at ϩ68 position) and termination sites are marked with bold letters. A putative polyadenylation signal is marked with a single underline.

Organization of the Rat COX VIII(H) Gene-
The nucleotide sequence of the rat COX VIII(H) gene and about 1.0 kb of the 5Ј-flanking region are presented in Fig. 1. Two NcoI fragments (4 and 1.2 kb) of the genomic clone that hybridized with the cDNA probe were sequenced and found to contain the entire COX VIII protein coding region and over 2 kb of the 5Ј-flanking sequence. The gene contains two exons of 178 and 151 nucleotides. The first exon contains the 67-base pair long 5Ј-untranslated region, sequences for the 24-amino acid long cleavable mitochondrial targeting sequence, and sequences for amino acids 1-13 of the mature protein. The second exon contains sequences for amino acids 14 -46 of the mature protein and the entire 3Ј-untranslated region, including a putative polyadenylation signal. A single intron of ϳ1.2 kb follows the 5ЈGT . . . AG3Ј consensus rule at the exon-intron junctions.
Identification of the Transcription Start Site-A combination of primer extension analysis and S1 nuclease protection was carried out to determine the transcription start site of the COX VIII(H) gene. As shown in Fig. 2A, primer extension using both the rat skeletal muscle and heart RNAs yielded a major extension product of 103 bases, which was not observed in extension products with E. coli tRNA. A major S1-resistant fragment of identical size was obtained with the rat skeletal muscle and heart RNAs (see Fig. 2A), but no significant protection was seen with E. coli tRNA or rat liver RNA. These results, along with the Northern blot hybridization presented in Fig. 2B, show that the 0.37-kb COX VIII(H) mRNA is expressed in skeletal muscle and heart tissues but not in the liver, kidney, brain, and lung tissues at significant levels. Thus, the major mRNA start site in both skeletal muscle and heart corresponds to an "A" residue 67 nucleotides upstream of the translation initiation site. The Northern blot in Fig. 2C shows that induced C2C12 as well as H9C2 myocytes, but not the uninduced myocytes, contain significant COX VIII(H) mRNA suggesting the induction of mRNA expression during myogenic differentiation.
Structural Features of the Promoter and Its Expression in Myogenic Cell Lines-A computer search for transcription promoter elements (see Fig. 1) showed the presence of a prominent TATA box 22 nucleotides upstream of the transcription start site and an Sp1 site at position Ϫ42 to Ϫ33. The 5Ј-flanking sequence also contains three E boxes with CANNTG consensus sequences at positions Ϫ464 to Ϫ459, Ϫ90 to Ϫ85, and Ϫ52 to Ϫ47 that have been marked as E box, E1 box, and E2 box, respectively (see Fig. 1). In addition, the 5Ј-flanking region contains one or more GRE, XRE, GATA-1, GCN-4, PEA-3, AP1, and AP2 consensus motifs and also three imperfect CArG sites (⌿CArG) as indicated in Fig. 1.
The efficiency of the promoter and requirements for maximal transcriptional activity in skeletal C2C12 and cardiac H9C2 myocytes as well as in COS cells were tested by transient transfection with intact and 5Ј deletion promoter CAT constructs. As shown in Fig. 3, the intact promoter construct, FIG. 2. Localization of the major transcription start site and muscle-specific expression of the COX VIII(H) gene. A, the transcription start site was mapped by using a combination of S1 nuclease protection and primer extension analysis. A 19-mer primer complementary to cDNA sequence 85-103 was used to generate the single-stranded S1 probe, the reverse transcriptase based primer extension, and also for generating the sequence ladder. The details of S1 analyses and primer extension are described under "Materials and Methods." In the gel pattern for S1 protection, duplicate lanes with heart and skeletal muscle RNAs represent S1 reactions with 150 and 200 Vogt units of the enzyme. The major transcription start site is indicated by an arrow. B represents Northern blot analysis of rat heart, skeletal muscle, liver, kidney, lung, and brain RNAs. C represents RNA from induced and uninduced C2C12 and H9C2 myocytes. In each case 25 g of RNA was hybridized with 32 P-labeled COX VIII(H) cDNA probe (13). The RNA size was determined by comparing with the migration of size markers. The loading level was assessed by hybridizing the same blot with labeled 18 S rRNA probe.
Ϫ994CAT, and also 5Ј deletion constructs Ϫ579CAT and Ϫ158CAT yielded nearly comparable CAT activities of 74 -104% in induced C2C12 skeletal myocytes and 92-106% in induced H9C2 cardiomyocytes, respectively. The deletion clone Ϫ44CAT, which contains the basal promoter elements, Sp1 and the TATA box, on the other hand, yielded reduced CAT activities of 18 -24% of the full-length construct in both of the induced myogenic cell lines. In contrast, all the four constructs gave comparable activities in COS cells. These results show that the Ϫ44 to Ϫ158 region of the promoter functions as a cell-specific transcription activator in both cardiac and skeletal muscle cell lines and that it failed to enhance the transcription activity of the basal promoter in nonmuscle COS cells.
Results of transient transfections presented in Fig. 3 also show that the Ϫ158CAT and Ϫ44CAT constructs gave relatively low CAT activities in the range of 7.2-8.2% in uninduced C2C12 myoblasts. These results demonstrate that the activity of the Ϫ158CAT construct is induced 12-14-fold during skeletal myogenesis, whereas the activity of the basal promoter construct Ϫ44CAT is induced about 2.5-fold. Similarly, the activities of the Ϫ994CAT, Ϫ579CAT, and Ϫ158CAT constructs were induced 14 -15-fold during induced myogenesis in H9C2 cardiac myocytes, whereas the activity of the basal promoter construct, Ϫ44CAT, was induced by 4 -5-fold. These results not only demonstrate that the COX VIII(H) promoter region being analyzed responds to cardiac muscle-specific differentiation signals, but also show that the promoter region responsible for most of the induction is contained within sequences Ϫ158 to Ϫ44. The reason for the 2.5-5-fold higher activity of the basal promoter construct during induced myogenesis in both cell types remains unknown, although it may be due to a higher level of general transcription factors in induced myocytes. Alternatively, a minor element that responds to differentiation signals could be present within the Ϫ44CAT construct.
Role of the E1 and E2 Boxes in Muscle-specific Transcription Activation-Although a majority of muscle-specific genes contain consensus E box motifs, a functional role for the E box motif in cardiac-specific gene expression has been implied only in a limited number of cases (24). Results presented in Fig. 4A show that promoter constructs with single E box mutations (E1mut and E2mut) yielded only 25-30% activity, which is similar to the activity of the double mutant E1/E2mut as well as the basal promoter Ϫ44CAT construct. Nearly similar effects of the E box mutations on transcriptional activity were observed in C2C12 skeletal myocytes. Furthermore, although results are not presented, the Ϫ158CAT construct when coexpressed with 1 g of Myo-D cDNA in induced C2C12 myocytes yielded about 7-fold higher activity. In contrast, single E box mutants E1mut or E2mut yielded only a marginal increase in activity when co-expressed with Myo-D cDNA, and the level of expression of the double mutant E1/E2mut was not affected by Myo-D co-expression. Results in Fig. 4B show that both single and double E box mutants yielded 3-4-fold transcriptional stimulation during induced myogenesis of H9C2 myocytes. This level of stimulation is similar to that observed with the basal promoter Ϫ44CAT suggesting that these mutants show reduced ability to respond to the myogenic differentiation signals. The mutational results presented in Fig. 4 thus suggest that the same consensus E1 and E2 box motifs are important in the cardiac as well as skeletal muscle-specific transcription activation of the rat COX VIII(H) gene and also suggest a functional cooperativity between the two E boxes.
The importance of E box binding bHLH proteins in the cardiac muscle-specific activation of the COXVIII(H) promoter was investigated by overexpression of Id (Id1 cDNA) in induced H9C2 cardiomyocytes. It is known that Id inhibits bHLH factordependent transcription activation in skeletal myocytes (45) and B cells (41,46) by heterodimerization with members of the E2A family of factors. Transient transfection results presented in Table II show that co-expression with increasing amounts of Id expression cDNA (20 and 30 g) caused a 2.5-5-fold reduction in activity of the Ϫ158CAT construct in H9C2 myocytes. However, co-expression with Id had no significant inhibitory effect on the activity of the basal promoter Ϫ44CAT as well as the COX IV promoter Bcd142 (10, 11), both of which lack E box motifs and fail to respond to myogenic signals. Furthermore, co-expression with 20 g of Id cDNA caused a 2.5-fold reduction in activity of the muscle-specific MCK110 enhancer construct (47), suggesting the specificity of the Id-mediated effects under the transfection conditions used. These co-expression results suggest the involvement of a member of the bHLH family of transcription factors in cardiac muscle-specific activation of the COXVIII(H) promoter.
Patterns of E Box Binding Proteins in Induced H9C2 Cardiac Myocytes-In an attempt to correlate the mutational data with tissue-specific protein binding, gel mobility shift assays were carried out with the E1 and E2 box synthetic DNA probes using nuclear extracts from induced C2C12 and H9C2 myocytes. Although both E box probes yielded some similarly migrating complexes with the two myogenic cell extracts, we were not certain if these comigrating complexes contained the same proteins. For this reason, complexes formed with the C2C12 and H9C2 extracts have been labeled differently, the former with roman numerals beginning with letter C and the latter with arabic numerals starting with letter H. The gel mobility shift pattern in Fig. 5A shows that the E2 DNA probe formed two major complexes CII and CIII and very minor complexes CIV and CV with the C2C12 extract that were competed by excess unlabeled E1 and E2 DNAs but not by mutated E1 or E2 DNAs. An expanded picture of this region of gel pattern is presented on the left-hand side to allow a better visualization of these complexes. Some batches of C2C12 nuclear extract also yielded a minor complex that migrated slower than complex CII. However, this complex was readily competed by E1mut

FIG. 3. Transcription activities of the intact and 5 deletion CAT constructs in COS cells and induced myocytes. The
Ϫ994CAT construct and its 5Ј progressive deletions were generated and transfected in COS cells and in induced (I) and uninduced (U) C2C12 skeletal, as well as H9C2 cardiac myocytes as described under "Materials and Methods." The locations of various factor binding motifs including the E1 box, E2 box, TATA box, and the transcription initiation site are indicated. The 5Ј ends of the progressive deletion clones are indicated with upwardly directed arrows. The CAT activities represent averages of three to four independent transfections. DNA (see Fig. 5A) suggesting that it is not E box-specific. The gel shift also yielded a number of fast migrating complexes that were not competed with the wild type or mutant DNAs suggesting nonspecific binding. The H9C2 extract, on the other hand, yielded a major complex H1 and two less intense complexes H4 and H5 with the E2 DNA probe (see Fig. 5A). All three complexes were efficiently competed by E1 and E2 wild type DNAs, whereas only complex H5, but not H1 and H4, was competed by mutated DNAs. These results suggest that complex H5 may not represent an E box-specific bound complex. Similarly, a number of faster migrating complexes that were not competed by excess E1 or E2 DNAs were observed suggesting that they may represent nonspecific binding. Because of these reasons and also because of the variable nature of these fast migrating complexes with both of the myogenic cell extracts in different gel shift assays, complexes migrating faster than CIII and H4 were not considered for further analysis.
A relatively more complex gel mobility shift pattern was obtained with the E1 DNA probe (see Fig. 5B). The C2C12 extract yielded prominent complexes CII and CIII in addition to minor complexes CIV to CVI. All of these complexes were competed by the wild type E1 DNA as well as E4 and E5 DNAs which are E box sequences found on the immunoglobulin heavy chain enhancer (48). However, these complexes were not competed by mutant E1 DNA. The H9C2 extract, on the other hand, yielded prominent complexes H1, H3, and H6. All of the complexes were efficiently competed by the wild type E1 DNA, as well as E4 and E5 DNAs. Additionally, only the H6 complex but not complexes H1 to H5 were competed by E1mut DNA. Since the relative levels of CIV to CVI and H4 to H6 varied in different experiments, they were not further analyzed in this study. The gel shift data presented in Fig. 5 show that induced H9C2 cardiac myocytes and C2C12 skeletal myocytes contain proteins binding to the E1 and E2 box motifs.
Characterization of E Box Binding Complexes by Antibody Supershift-The nature of the E box binding complexes with nuclear extracts from induced C2C12 skeletal myocytes and H9C2 cardiomyocytes were compared by antibody supershift assay using polyclonal antibody against myo-D and a monoclonal antibody against E2A, termed Yae. The Yae antibody was previously shown to interact with myo-D⅐E2A heterodimer as well as E2A homodimer complexes, causing supershifts (20,27). The Myo-D antibody is known to interact only with complexes containing myo-D⅐E2A heterodimers (20). The gel mobility shift with the E1 DNA probe (see Fig. 6A) shows that Myo-D antibody efficiently supershifted the C2C12 nuclear protein-derived complex (SSCC), while it failed to show any effect on bound complexes with the H9C2 nuclear proteins. Based on the relative band intensities in lanes with added Myo-D antibody, CII appears to be preferentially supershifted to form SSCC. The above results therefore suggest that the major complexes H1 and H3 obtained with extract from induced H9C2 myocytes lack Myo-D and support the widely prevailing view on the lack of significant Myo-D gene expression in cardiac tissue. On the contrary, the Yae antibody produced a supershifted complex SSHC with the H9C2 cardiomyocyte extract (Fig. 6B), although it is not clear if the H1 and/or H3 complexes are supershifted. Similarly with the C2C12 extract, the Yae antibody efficiently supershifted the same slow migrating CII complex as was seen with Myo-D antibody in a concentration-dependent manner to form an cated at the bottom. B, the extent of induction of transcription activity of the wild type and mutant promoter constructs during induced myogenesis was determined by comparing the transcription activities in uninduced and induced H9C2 myocytes. The average and standard deviations were calculated from three to five transfections. ؊158CAT promoter in C2C12 and H9C2 myocytes. A, the wild type Ϫ158CAT construct and mutations targeted to single E boxes (E1MUT and E2MUT), as well as both E boxes (E1/E2MUT), were transfected in C2C12 myocytes and also H9C2 myocytes, and cell extracts from fully induced myocytes were assayed for CAT activities as described under "Materials and Methods." The putative protein binding sites of each of the promoter construct and mutational sites are indi-SSCC, whereas preimmune IgG had no effect. The SSCC complex comigrated with a slow migrating complex as seen with some batches of C2C12 extract described in Fig. 5. The supershifted complex with H9C2 extract, SSHC, however, migrated differently from the SSCC, further confirming that the Yae antibody-reactive complexes from the two myogenic extracts are different. Furthermore, similar antibody supershift patterns were observed with the C2C12 and H9C2 nuclear protein bound complexes with the E2 DNA probe (Fig. 6C).

FIG. 4. Effects of E box mutations on the transcription activities of the
The patterns of E box binding proteins in uninduced and induced H9C2 extracts were compared to see if the characteristic Yae antibody interacting protein was induced during cardiac myogenesis. The binding patterns with nuclear extracts from induced and uninduced H9C2 myocytes is presented in Fig. 7A. Extract from uninduced myocytes yielded complexes comigrating with complexes H1 and H3 obtained with extract from induced myocytes, in addition to an intermediary complex I. All of these complexes were efficiently competed with excess unlabeled E1 DNA. The complex comigrating with complex H1 was also competed by E1mut DNA suggesting that it is either a nonspecific complex or not E box-specific. The Yae antibody caused a supershift of the intermediary complex (complex I) obtained with the uninduced H9C2 extract (Fig. 7B). Furthermore, the antibody supershifted complex with uninduced H9C2 extract (marked as SS) migrated differently from the SSHC obtained with the nuclear extract from induced myocytes. These results show that although both the induced and uninduced H9C2 myocytes contain E box binding proteins, the Yae antibody-reactive complexes in these two extracts are different, suggesting the induction or activation of a specific protein during cardiomyocyte differentiation.
Nature of the Yae Antibody-reactive Protein in Induced H9C2 Cardiac Myocytes-It was recently shown that E2A homodimer binding to DNA is critical for the IgH gene expression and possibly in differentiation of the B cell lineage (27). We therefore rationalized that the Yae antibody supershifting complex obtained with the induced H9C2 nuclear extract (SSHC) may represent either an E2A homodimer or a heterodimer with a cardiac cell-specific bHLH factor. To test these possibilities H9C2 myocytes were transfected with the E2-5 cDNA that encodes a full-length E2A protein. Mini-cell extracts from E2A overexpressing and control cells were compared for binding to the E1 DNA probe. The gel mobility shift pattern presented in Fig. 8 shows that the nuclear extract from control H9C2 cells yielded two complexes consistent with complexes H1 and H3 described previously. The extract from E2A overexpressing cells, however, yielded an additional band that migrated faster than complex H1 (see Fig. 8). The levels of both H1 and H3 complexes remained relatively unchanged in H9C2 cells trans- FIG. 5. Characterization of E1 and E2 box binding proteins by gel mobility shift analysis. 32 P-Labeled, gel-purified E2 box (A) and E1 box (B) DNA probes were incubated with nuclear extracts from induced C2C12 or H9C2 myocytes, and the bound complexes were resolved on 4% polyacrylamide gels as described under "Materials and Methods." An enlarged picture of complexes CI to CV from the control lane in A is presented at the left-hand side. Competitions were carried out using 100 M excess (20 ng) of unlabeled wild type or mutant DNAs as indicated. H9C2 myocytes were transfected with 8 g each of the Ϫ158CAT, Ϫ44CAT, MCK110, or 5 g of Bcd142 promotor construct alone or co-transfected with 20 or 30 g of ID expression construct as described under the "Materials and Methods." Extracts from induced myocytes were assayed for relative CAT activities. The values represent averages of 3-6 transfections.
fected with increasing amounts of E2-5 cDNA up to 20 g. Furthermore, coexpression with E2-5 cDNA did not affect the in vivo transcription activity of the Ϫ158CAT construct in induced H9C2 cardiac myocytes (results not presented). These results suggest that the E box binding complex from induced cardiomyocytes, supershifted by the Yae antibody, might not consist of an E2A homodimer, although it may contain E2A as part of a heteromeric complex. DISCUSSION Nuclear genes encoding mitochondrial proteins contain one or more common sequence motifs in their 5Ј proximal promoter regions (49,50). The most common sequence motifs include the NRF1 (17), GABP (variably termed NRF2 and E4TF1 (10 -12)), and OXBOX (51). Although NRF1 and GABP (NRF2, E4TF1) are not unique to genes encoding mitochondrial proteins (14,(52)(53)(54), it is believed that these and other common transcription factor binding motifs might be involved in the coordinate regulation of nuclear genes coding for various mitochondrial proteins. There is evidence suggesting a global regulation of mitochondrial destined genes in yeast by members of the HAP1/HAP2/HAP3 family of transcription factors (55,56). Results presented in this paper indicate the absence of functionally important NRF1, GABP (NRF2, E4TF1), or OXBOX sequence motifs in the 5Ј promoter region of the rat COX VIII(H) gene, suggesting that a different set of regulatory controls may exist for the expression of tissue-specific mitochondrial genes. Additionally, our results demonstrate that the rat COX VIII(H) gene is predominantly expressed in skeletal muscle and heart tissues with no detectable expression in nonmuscle tissues. Interestingly, the same two E box sequence motifs closest to the transcription start site are important for both cardiac and skeletal muscle-specific regulation of the gene.
Extensive studies in the skeletal muscle system have provided compelling evidence that the expression of one or more bHLH factors like Myo-D, myogenin, Myf-5, and MRF-4 is sufficient to induce myogenesis and that these master regulatory genes play a central role in skeletal muscle cell determination and differentiation (22,25,57,58). It is well known that Myo-D and other myogenic determinants are expressed only in differentiated skeletal muscle cells and activate only the myogenic genes (19,21). Although a number of nonmuscle enhancers such as the IgH enhancer contain E boxes, such nonmuscle promoters are not activated in muscle cells due to the presence of negative acting elements that inhibit Myo-D-dependent transcription activation (47). In view of this, the 12-14-fold increase in the transcription activity of the Ϫ158CAT construct in induced C2C12 skeletal myocytes and also a 7-8-fold stimulation by co-expression with the Myo-D cDNA suggest that the two E boxes (E1 and E2 boxes) of the rat COX VIII(H) promoter characterized in this study function cooperatively as skeletal muscle-specific promoter elements. The results of mutations targeted to single E boxes also suggest some degree of cooperativity between the two protein-bound complexes for transcriptional activity. Such cooperative interactions between Myo-D and related heterodimeric bHLH factors binding to two tandemly spaced E box sites have been observed previously (59 -64).
Differentiation of cardiac myocytes requires coordinated expression of a set of tissue-specific genes including those coding for the contractile proteins, cell-and lineage-specific enzymes, ing of E2 DNA-bound complexes using antibodies to myo-D as well as E2A. In all cases ϩ represents 1 l, and ϩϩ represents 2 l of antibody or preimmune serum. SSCC represents supershifted C2C12 extractderived complex, and SSHC represents supershifted H9C2 extractderived complex.
FIG. 6. Antibody supershift patterns of protein-bound complexes with extracts from C2C12 and H9C2 myocytes. A represents the supershifting of E1 box DNA-bound complexes with polyclonal antibody to myo-D. B represents supershifting of E1 DNA-bound complexes with monoclonal antibody (Yae) to E2A. C represents supershift-and receptors. Transcriptional activation of these genes during cardiac muscle differentiation has been shown to involve multiple pathways, involving different enhancer binding proteins, including the recently identified bHLH factors (33)(34)(35). For example, the expression of the cardiac troponin T gene requires distinctly different sequence motifs than those required for its expression in skeletal myocytes (65). A minimal promoter region containing two CAAT-like motifs, designated as M-CAT, was sufficient for full expression in skeletal muscle cells. An additional ϳ75-nucleotide sequence upstream of the two M-CAT motifs was required for its expression in cardiac myocytes (66). Activation of the cardiac muscle-specific troponin C enhancer, on the other hand, requires the binding of zinc finger transcription factor GATA-4 (31). Additionally, many musclespecific genes contain CArG boxes (also referred to as SRE) that are important in the cardiac muscle-specific expression of a number of genes (24,28,67). Additionally, A/T-rich motifs binding to the myocyte-specific enhancer factor MEF-2 have been implicated in the cardiac-specific expression of a number of genes (29,46).
Reports of cardiac muscle-specific transcription regulation involving the bHLH family of transcription factors have been relatively rare. Studies with the human cardiac ␣-actin gene suggested that the CArG box motif was essential for enhancer activity, although an E box motif located immediately downstream was also important for maximal activity in cardiac myocytes (24,67), suggesting a functional cooperativity between proteins binding to the two sequence motifs. In the present study, a minimal promoter region containing the two proximal E boxes and a GRE (sequences Ϫ158 to Ϫ44) exhibited cardiac muscle-specific transcription activation in induced H9C2 myocytes (Fig. 3). Our results also show that mutations in a single E box caused a reduction in activity comparable with the activity of the basal promoter and a reduction in its ability to respond to cardiac muscle-specific differentiation signals. Further support for the myogenic specificity of the E box containing distal promoter comes from the observation that the putative enhancer sequence is unable to stimulate transcription in nonmuscle cells such as COS cells (Fig. 3). To our knowledge, the COX VIII(H) gene is the first gene to be reported where duplicated E boxes act as tissue-specific transcription activators in both skeletal-and cardiac-specific expression. Although the 5Ј upstream region of the promoter contains three imperfect CArG-like sequence motifs (⌿CArG), results of progressive deletion studies suggested no apparent functional significance for this region in both the cardiac and skeletal muscle-specific activation of transcription. This is in contrast to a possible functional cooperativity between SRF and E box binding proteins in the tissue-specific expression of a number of genes (23,24,67).
Gel shift analysis and Yae antibody supershift patterns presented in this study (Figs. 5 and 6) confirm and extend earlier observations (24, 32) on the possible occurrence of an E box binding bHLH protein in differentiated cardiomyocytes. Direct support for the functional significance of the bHLH protein in FIG. 7. Comparison of E box binding proteins from uninduced and induced H9C2 myocytes. A and B represent gel shift patterns with 32 P-labeled E1 DNA probe using nuclear extracts from induced (I) and uninduced (U) H9C2 myocytes. A, 20 ng of unlabeled wild type or mutated E1 box DNA were used for competition as indicated at the top of lanes. B, the E2A-specific Yae monoclonal antibody was used to compare the antibody supershift patterns of bound complexes with induced and uninduced H9C2 extracts. SS denotes supershifted complex with extract from uninduced H9C2 myocytes. SSHC represents supershifted complex with extract from induced H9C2 myocytes.
FIG. 8. Effects of E2A overexpression on complex formation with H9C2 extract. Mini-cell extracts from control H9C2 cells, cells mock-transfected with 10 g of vector DNA, or transfected with 5, 10, or 20 g of E2-5 CMV DNA were used in binding assays with E1 DNA probe. In all cases 3 l of nuclear extracts from induced myocytes were used for binding. 1 l of full-length E47 protein expressed in E. coli (1-2 g) was used for comparison. An intermediary complex migrating distinctly from complex I in extracts from E2A overexpressing cells is shown as E2A.
cardiac muscle-specific transcription activation of the COX VIII(H) promoter comes from the observation that Id overexpression specifically inhibited the transcriptional activity in induced H9C2 myocytes (Table II). Our results also show a 12-14-fold difference in the in vivo transcription activities of the Ϫ158CAT construct in uninduced versus induced H9C2 myocytes (Fig. 3). Additionally, results of gel mobility analysis with extracts from uninduced and induced H9C2 myocytes (Fig. 7) show marked differences in the migration of Yae antibody supershifted complexes. These results provide evidence that the E box binding protein necessary for cardiac musclespecific activation of the COX VIII(H) promoter is either induced or activated during myogenesis. Recent studies using the yeast two-hybrid system (34,35) demonstrated the expression of two bHLH proteins, dHAND and eHAND (Th1, Th2) during heart muscle early embryonic stages, that are not detected in the adult heart (34). In support of these observations, coexpression with both dHAND and eHAND did not affect the transcription activity of the Ϫ158CAT construct in differentiated H9C2 cells (results are not presented). Although, our results suggest that the antibody supershifting complex may be different from an E2A homodimer, this may contain an E2A-like protein in association with a cell-specific bHLH protein. UV cross-linking and Western blot analyses (results not shown) indeed show the presence of E2A antibody-reactive proteins in nuclear extracts from induced H9C2 myocytes and adult rat heart.