Structural Organization and Promoter Analysis of the Bovine Cytochrome c Oxidase Subunit VIIc Gene A FUNCTIONAL ROLE FOR YY1*

Cytochrome c oxidase (COX) subunit VIIc is one of the nuclear encoded subunits of the 13-subunit holoenzyme that carries out the terminal step in the electron trans- port chain. We have isolated the gene for this subunit, previously shown to be ubiquitously expressed from a single copy gene in the genome, and show that 167 base pairs of DNA surrounding the transcriptional start site contain the minimal promoter of this gene. This basal promoter contains two YY1 sites and at least one site for NRF-2, which show binding to their cognate factors. Mutation of both YY1 sites eliminates most of the pro- moter activity. Mutation at the upstream YY1 site signif-icantly reduces the efficiency of transcript initiation at the major start site and thus plays the dominant role in COX7C regulation. COX7C is, thus, the second nuclear gene of COX that is regulated by YY1, suggesting that it is a third common factor, along with NRF-1 and NRF-2, to be associated with COX gene regulation. Mammalian cytochrome c oxidase (COX) 1 a 13-subunit protein complex located on the inner mitochondrial membrane that catalyzes the transfer of electrons from reduced cytochrome c to molecular oxygen and in the translocation of protons across the membrane. Proton translocation generates an electrochemical gradient that drives the synthesis of ATP. The complex consists of three primarily catalytic subunits encoded by mitochondrial DNA and 10 polypeptides encoded in the nucleus that are suggested to function in regulation and/or assembly (1–3). Three of the nuclear-encoded subunits (VIa, VIIa, and VIII) exist as a pair of isoforms (4–6) in most mammalian species: a liver (L) form that is present in most and a heart/muscle

Cytochrome c oxidase (COX) subunit VIIc is one of the nuclear encoded subunits of the 13-subunit holoenzyme that carries out the terminal step in the electron transport chain. We have isolated the gene for this subunit, previously shown to be ubiquitously expressed from a single copy gene in the genome, and show that 167 base pairs of DNA surrounding the transcriptional start site contain the minimal promoter of this gene. This basal promoter contains two YY1 sites and at least one site for NRF-2, which show binding to their cognate factors. Mutation of both YY1 sites eliminates most of the promoter activity. Mutation at the upstream YY1 site significantly reduces the efficiency of transcript initiation at the major start site and thus plays the dominant role in COX7C regulation. COX7C is, thus, the second nuclear gene of COX that is regulated by YY1, suggesting that it is a third common factor, along with NRF-1 and NRF-2, to be associated with COX gene regulation.
Mammalian cytochrome c oxidase (COX) 1 is a 13-subunit protein complex located on the inner mitochondrial membrane that catalyzes the transfer of electrons from reduced cytochrome c to molecular oxygen and participates in the translocation of protons across the membrane. Proton translocation generates an electrochemical gradient that drives the synthesis of ATP. The complex consists of three primarily catalytic subunits encoded by mitochondrial DNA and 10 polypeptides encoded in the nucleus that are suggested to function in regulation and/or assembly (1)(2)(3). Three of the nuclear-encoded subunits (VIa, VIIa, and VIII) exist as a pair of isoforms (4 -6) in most mammalian species: a liver (L) form that is present in most tissues and a heart/muscle (H) form specific for adult cardiac and skeletal muscle.
Several COX nuclear genes, including those of isoforms, have now been isolated and characterized and provide a means of assessing their role in function and regulation; in addition, defects in the expression of genes for COX nuclear subunits are likely to cause human disease. A detailed analysis of the basal promoters of three ubiquitously expressed COX genes thus far available (COX4, COX5B, and COX7AL) reveal that they have regulatory sites in common (7)(8)(9)(10)(11)(12). COX4, COX5B, and COX7AL contain multiple, functional binding sites for nuclear respiratory factor-2 (NRF-2 (Ref. 8); also called GABP, GAbinding protein). COX5B and COX7AL also harbor a functional site for NRF-1 (13,14), which binds to a number of housekeeping genes; a functional site has also been detected in COX6C, although the promoter of this gene has not been extensively analyzed (15). Binding sites for the ubiquitous factor Sp1 are found in the basal promoters of COX4, COX5B, and COX7AL. Finally, the multifunctional factor Yin-Yang-1 (YY1) has only been found in COX5B, where it acts by initiating basal transcription. None of the above elements, however, is found in the basal promoter of COX6AH, the only H isoform COX promoter so far characterized (16). The basal promoter of this gene is dependent on muscle-specific elements such as the E-box (MyoD binding site) and MEF-2 (myocyte-specific enhancerbinding factor-2) binding sites.
To further elucidate the mechanisms of COX regulation and expression, we have characterized the bovine gene encoding the 63-amino acid precursor of subunit VIIc (COX7C), a gene encoding a subunit without isoforms and thus expressed in all tissues. We present here the sequence of the COX7C gene and identify the transcription factors that regulate its expression. We show that its basal promoter is regulated by two YY1 sites, one of which is required for efficient initiation of transcription from the major start site. COX7C is the second COX nuclear gene in which YY1 plays a role in promoter function, suggesting that YY1 may be a third common factor, along with NRF-1 and NRF-2, involved in COX gene regulation. Cell Culture-HeLa (2-CCL), HeLa S3 (CCL-2.2), and C 2 C 12 mouse myoblasts (1772-CRL) were obtained from American Type Culture Collection. Media and supplements were obtained from Life Technologies, Inc. HeLa cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) containing penicillin (40 units/ml) and streptomycin (40 g/ml). HeLa S3 suspension cultures were grown in a similar medium containing 5% FBS. * This work was supported by National Institutes of Health Grant GM 48517. 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.

Materials-Ribonuclease
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U58655.
Primer Extension Assay-Primer extension was carried out as described previously (22), using total RNA isolated from adult bovine heart tissues and a 23-nt antisense primer corresponding to nt 183-161 of the first exon (Fig. 3). The primer-extended products were resolved on 6% polyacrylamide, 6 M urea gels and autoradiographed for 1-2 weeks at Ϫ70°C. A pBluescript KS(Ϫ) plasmid sequenced with the universal forward primer served as a DNA size marker.
Ribonuclease Protection Assay-A 395-bp SacII-HincII fragment comprising 121 bp of the first intron, at least 131 bp of the first exon, as deduced from the cDNA (19), and an additional contiguous region of 143 bp at the 5Ј end, was cloned at the SacII-HincII site of pBluescript KS(Ϫ) and used for nuclease protection with 40 g of total RNA as described previously (23).
CAT Constructs and Transfection Analysis-EcoRI, PstI, or SacII sites located upstream of the transcriptional start site and a HpaII site located within the non-coding portion of the first exon were used to isolate fragments that were cloned into pGKOCAT (24) to generate 1.8-kb EcoRI-7cCAT, 1.5-kb PstI-7cCAT, and 93-bp SacII-7cCAT. Recombinant plasmids were isolated, verified by sequencing, and purified through two CsCl gradients.
HeLa cells (8 ϫ 10 5 /ml) were grown, transfected, and the normalized CAT activity determined as described (9). For each CAT construct 3-4 dishes were used and each experiment repeated at least four times. C 2 C 12 myoblasts (2 ϫ 10 5 cells) and myotubes (5 ϫ 10 5 cells) were similarly transfected. The medium was replenished 17-20 h after transfection. Myoblasts were allowed to grow for another 48 h; to generate myotubes the medium was switched to 10% horse serum 24 h after replenishment and cells allowed to grow for an additional 48 h before harvesting, as described for HeLa cells.
Site-directed Mutagenesis of the YY1 Motifs in COX7C-The 167-bp basal promoter fragment was resynthesized from a set of 6 overlapping oligonucleotides spanning both strands as described (27). Appropriate mutant or wild-type (WT) complementary oligonucleotides were used to create the desired upstream (U), downstream (D), and upstream-downstream (UD) YY1 mutant, or WT constructs. Mutations were created by introducing the sequence AAAGGG for the YY1 core recognition motif NCCATN. For YY1-D, the introduced changes did not alter the overlapping NRF-2 site (Fig. 5). Briefly, the oligonucleotides were phosphorylated, annealed, and ligated; the 167-bp fragment recovered by PAGE was cloned into the SmaI site of pGKOCAT vector. After transformation into Escherichia coli DH5␣ cells, recombinants harboring the desired inserts were identified by restriction enzyme digestion and sequencing.
Transcriptional Analysis of YY1 Mutations in Transfected Cells-Total RNA from C 2 C 12 myoblasts, myotubes, and HeLa cells transfected with the WT basal promoter, as well as those harboring the various YY1 mutations, was subjected to a ribonuclease protection assay. Two antisense probes were generated by transcription from the T3 promoter using appropriately digested pBluescript plasmid containing the WT 167-bp basal promoter fragment (WT probe) or one with a mutant YY1-D site (D-probe) (Fig. 9B). The WT probe was used to detect transcripts arising from transfected WT and YY1-U mutant basal promoter CAT constructs and the D-probe was used to detect transcripts arising from the YY1-D and YY1-UD mutated constructs. Both the probes and the transcripts arising from the transfected plasmids include a common 62-bp vector sequence (Fig. 9B) that serves to differentiate these transcripts from any endogenous ones that may be protected by the probe. Both probes will protect all transcripts initiating from the ϩ1 site to the beginning of the CAT gene. Yeast tRNA and mock-transfected cells were used as a control for each probe. Individual transcripts were quantitated by densitometric scanning with a Molecular Dynamics PhosphorImager. The results from the ribonuclease protection assays were confirmed by primer-extension analysis (22) FIG. 1.
Gene organization of COX7C. A, restriction map and sequencing strategy of a 2.5-kb region harboring the COX7C gene. B, intron-exon organization of COX7C. Exons are stippled. Exons 1-3 are 185, 129, and 218 bp, and introns 1 and 2 are 838 and 915 bp, respectively. C, COX7C is located at a CpG island. Distribution of CpG (above) and GpC (below) dinucleotides spanning the 2.5-kb region is shown.

FIG. 2. Determination of the transcription start site of COX7C.
A, primer-extension analysis. Total bovine heart RNA (CO) and yeast tRNA (Y) were used. Products were resolved on 6% polyacrylamide-urea gels using a DNA sequencing ladder as a size marker. The largest primer-extended product (arrow) is 183 nt (two shorter products that occur within 10 bp of this product are not evident from this figure). B, ribonuclease protection assay. Total bovine heart RNA (CO) or yeast tRNA (Y) was protected with labeled in vitro synthesized sense (A) or antisense (AS) transcripts. Protected fragments were resolved on 6% polyacrylamide-urea gels and size compared with a DNA sequencing ladder. The largest protected fragment (arrow) is 180 nt.
using an antisense primer specific to the 5Ј-untranslated region of the CAT gene (5Ј-TAGCTCCTGAAAATCTCGCC) (data not shown).

Isolation and Organization of the COX7C Gene-The COX7C
genomic region from BGL-2J (see "Experimental Procedures") was completely sequenced on both strands (Fig. 1A). The first exon (185 nt) encodes the complete 16-amino acid presequence and the first 9 amino acids of the mature protein; it also contains a 5Ј-non-coding region of 110 nt. The second exon (129 bp) encodes the remaining 38 amino acids of the mature protein and 15 nt of the 3Ј-non-coding region, whereas the third encodes 218 nt of the non-coding region. All the intron/exon junctions follow the GT-AG rule (28); the first and second introns are 838 and 915 bp, respectively.
The gene is located at a CpG island (29), based on the distribution of CpG residues. Of the 52 CpGs that span the sequenced gene (Fig. 1C), 94% are found near the transcriptional start site and the first intron; by contrast, the distribution of GpC residues appears to be fairly uniform (Fig. 1C).
The transcriptional start site of the COX7C gene was determined by primer-extension and nuclease protection analyses. A 23-nt antisense primer that anneals to the first exon, beginning 2 nt 5Ј (nt 183-161; Fig. 3) of the first intron, was used to reverse transcribe total bovine RNA. A major primer-extended product of 183 nt was obtained, suggesting that the first exon is 185 nt long. This product is clustered along with at least two minor products that all lie within 8 -10 nt of each other (but not clearly evident in Fig. 2A). This result was consistent with a ribonuclease protection assay. In the latter, a cloned 395-bp fragment, extending from the SacII site (nt Ϫ93) to the HincII site (nt ϩ306), containing at least 131 bp of the first exon as deduced from the cDNA, and an additional 143 bp of 5Ј-upstream sequence, was radioactively transcribed to yield sense and antisense transcripts and each was hybridized to total bovine RNA. Only the antisense transcript probe yielded protected fragments (Fig. 2B, lane CO-AS), the largest of which was 180 nt. At least three protected bands (including the 180-nt band) were seen, consistent with the primer extension data. The major band (arrow) observed by primer-extension ( Fig. 2A) was used to assign nt ϩ1. These results indicate that the gene contains heterogeneous start sites. The end of the third exon was determined by comparison with a poly(A) tailcontaining processed pseudogene (BCOX7c-1; Ref. 19) since the available bovine cDNAs were truncated in this region. The gene has two potential poly(A) addition signals, one matching the consensus AATAAA and another, AATTAA, located 132 nt and 32 nt upstream of the poly(A) tail, respectively (Fig. 3).
Analysis for the presence of known regulatory sites revealed two putative YY1 motifs (30,31) between Ϫ17 and ϩ69, one immediately flanking the transcriptional start site (referred to as the YY1-upstream site) and the other in the 5Ј-untranslated region, 55 bp downstream of the transcriptional start site (YY1downstream site) (Figs. 3 and 5). The two sites match the reported consensus (Fig. 5) (32, 33). The presence of YY1 motifs is especially interesting since a YY1 site has also been found in COX5B, where it acts as an initiator (10). Another notable motif present is for NRF-2 ((A/C)GGAA): two sites upstream of the translational start site (at ϩ69 to ϩ65 and Ϫ158 to Ϫ154; Fig. 3) and four sites in the first intron, including a pair of tandem motifs spaced 4 bp apart. Six more motifs can be found if the above consensus is extended to include a GGAT core: two in the upstream region (at Ϫ111 to Ϫ115 and Ϫ60 to Ϫ64), one at the start site (ϩ6 to ϩ2), and three in the first intron. There are no canonical TATA or CCAAT boxes or binding sites for NRF-1. There are also no GC-boxes (Sp1 binding sites) with the core sequence GGGCGG (34), but several that show partial homology are present. A perfect match to the Sp1 consensus ((G/T)(G/A)GGC(G/T)(G/A)(G/A)(G/T); Ref. 35) is present between ϩ579 and ϩ587 in the first intron and two more, with a single mismatch (8/9), are located between ϩ73 and ϩ65 in the 5Ј-untranslated region and ϩ691 and ϩ683 in the first intron.
Promoter Analysis of Flanking DNA-Three CAT constructs were made in pGKOCAT vector with 1.8 kb, 1.5 kb, and 93 bp of flanking DNA (Fig. 4). These were transfected into HeLa and C 2 C 12 cells, and their normalized CAT activity was compared with the largest CAT construct, EcoRI-7c CAT. In HeLa cells, significant activity (117 Ϯ 26%) is retained with the smallest CAT construct used: SacII-7cCAT, which contains only 93 bp of 5Ј-flanking DNA and 74 bp of the 5Ј-untranslated region (Fig.  5). Hence, the putative basal promoter must reside within the 167-bp fragment defined by SacII and HpaII ends. This fragment contains two putative sites for YY1 and at least a single site for NRF-2. Similarly, in C 2 C 12 cells, the SacII-7cCAT construct is as active as the largest construct both in myotubes (113 Ϯ 8%) and myoblasts (138 Ϯ 17%) (Fig. 4).
Binding of Transcription Factors to the Core Promoter-We used EMSA to examine the ability of YY1 to bind to the basal promoter both with HeLa extracts (data not shown) and purified YY1 (Fig. 6). Increasing amounts of YY1 (0.1-600 ng) produced a distinct shift (Complex A) at 10 ng that progressively increased in intensity at 100 and 600 ng. A second shift (Complex B) of slower mobility was observed at 600 ng, indicating the probable presence of two YY1 binding sites in the basal promoter fragment, consistent with the YY1-U and -D sites identified within this fragment. We speculate that the faster migrating band represents the DNA fragment with one YY1 site bound and the slower one has both sites bound. To confirm binding of YY1 to each of these sites, we annealed two FIG. 4. Characterization of the basal promoter of COX7C. Up to 1.8 kb of the upstream region of COX7C was analyzed for promoter activity by transfection in HeLa, C 2 C 12 myoblasts, and myotubes. Constructs harboring 1.8 kb (EcoRI-7cCAT), 1.5 kb (PstI-7cCAT), and 93 bp (SacII-7cCAT) of upstream region were made in pGKOCAT. CAT activity normalized with ␤-galactosidase is expressed relative to EcoRI-7cCAT (100%).  pairs of complementary oligonucleotides for EMSA, each harboring one of these sites. The duplex oligonucleotide ds7c-6 contains the YY1-D site, and ds7c-3.5 contains the YY1-U site (Fig. 5). YY1 binds specifically to YY1-D (Fig. 7A), shown by competition with 100-and 1000-fold molar excesses of the specific (ds7c-6, lanes 3 and 4) but not nonspecific (lanes 5 and 6) DNA. Similarly, YY1 was able to bind specifically with YY1-U (ds7c-3.5) (Fig. 7B). These results confirm that the basal promoter fragment (Ϫ93 to ϩ74) has two functional binding sites for YY1 and that these are the U and D sites. YY1-D overlaps an NRF-2 motif (Fig. 5). We examined whether NRF-2 could bind specifically to this complex regulatory site and found that it does (Fig. 7C), since the specific DNA (ds7c-6, lanes 3 and 4) but not a nonspecific DNA (ds7c-3, lanes 5 and 6) can effectively compete for this binding.
Mutational Analysis of YY1 Sites-Mutations were introduced at YY1-U, -D, and at both sites (UD) in the 167-bp basal promoter fragment. The effects of these mutations on CAT activity were assessed in both undifferentiated and differentiated C 2 C 12 cells and in HeLa cells (Fig. 8). The mutations have similar effects in all three cell types. YY1-U appears to be the relatively more important site, since a mutation decreases activity by 68% in HeLa cells and 58% in myotubes, whereas a mutation in YY1-D decreases activity by 46% and 39%, respectively. However, when both sites are mutated, the CAT activity drops further (by a total of 79% and 71%, respectively). Myoblasts show a similar pattern, but at more modest levels. These decreases in promoter activity are significant in the three cell types tested, as determined by a standardized t test (36).
Transcriptional Analysis of the YY1 Mutations in Transfected Cells-The major transcriptional start site determined for the wild-type transfected promoter (Ϫ93 to ϩ 74) in cultured cells is consistent with that obtained from mRNA isolated from tissue (data not shown). This suggests that the transfected CAT gene is driven by the COX7C basal promoter and not by cryptic sequences. In addition to the major start site, as already noted, a cluster of heterogeneous transcripts beginning immediately downstream of this site is seen. To examine if these initiation sites are affected by the two YY1 mutations, we transfected HeLa cells with the WT and mutant YY1-U, -D, and -UD basal promoter constructs. Total RNAs from these cells were subjected to ribonuclease protection assays with probes that would detect the major transcript arising at the ϩ1 site as well as the heterogeneous cluster of minor transcripts initiating downstream. In Fig. 9A, the major start site (ϩ1) (larger arrow) and initiation sites for the three minor transcripts (smaller arrows) in the WT-transfected promoter are shown. The major transcript constitutes 47% of these transcripts by densitometric analysis. This pattern is retained in the D mutant, although the overall amount of transcripts is reduced. Significantly, the major transcript in WT cells is reduced in both the U and UD mutants (16 and 20%, respectively); one of the downstream transcripts now constitutes the largest fraction (32-33%, respectively). Overall, there is a net decrease in total transcript levels in the U, D, and UD mutants. Some of the decrease in the D lane is due to the lesser amount of total RNA used (Fig. 9A, legend). These observations hold true for both myoblasts and myotubes (data not shown). Thus, the binding of YY1 to the U site appears to correlate with initiating transcription from the ϩ1 site.

DISCUSSION
The minimal promoter of COX7C can be localized to a sequence that contains 93 bp of 5Ј-flanking region and 74 bp of the 5Ј-untranslated region. This 167-bp region contains two sites for YY1 and at least one site for NRF-2. All three sites show binding to their cognate factors (Fig. 10). In addition to COX7C and COX5B (10), the nuclear encoded ␣-subunit of the mitochondrial F 0 F 1 ATP synthase is also regulated by YY1 (37). Thus, three subunits of the respiratory chain are regulated by this factor, a multifunctional zinc-finger protein, variously referred to as NF-E1 (31), ␦ (38), UCRBP (39), CF-1 (40), F-ACT 1 (41), and NMP-1 (42), that can activate, repress, or initiate transcription. The presence of two or more binding sites for YY1 and their location 3Ј to the transcriptional start site have been frequently observed (32,(42)(43)(44)(45).
COX7C expression appears to be more critically dependent FIG. 7. A, binding of YY1 to the D site. ds7c-6, a duplex oligonucleotide containing the YY1-D site, was used in EMSA. Lane 1, free probe; lane 2, incubation with YY1 alone; lanes 3 and 4, competition with 100-and 1000-fold molar excesses of specific competitor, respectively; lanes 5 and 6, competition with 100-and 1000-fold molar excesses of nonspecific competitor. B, binding of YY1 to the U site. ds7c-3.5, containing the YY1-U site, was used in EMSA and the experiment performed as in A. ds7c-6 is a nonspecific competitor. C, binding of NRF-2 to the basal promoter. The basal promoter contains an NRF-2 site overlapping YY1-D. ds7c-6, containing the NRF-2 site, was used in EMSA, and ds7c-3 was used as nonspecific DNA. Lane 2 represents incubation with NRF-2 ␣ ϩ ␤ 1 subunits; lanes 3 and 4 represent competition with 100-and 1000-fold molar excesses of specific competitor; lanes 5 and 6 represent similar competition with a nonspecific competitor.
FIG. 8. Site-directed mutagenesis of the YY1-motifs and their effect on COX7C promoter activity. Mutations (X) were introduced at the YY1-U, -D, and -UD motifs of the basal promoter (Ϫ93 to ϩ74) as described under "Experimental Procedures." The three mutant constructs and a WT construct were transfected into HeLa, C 2 C 12 myoblasts, and myotubes. Promoter activity was determined by expressing normalized CAT activity as a percent of the WT value. The transcriptional start site at ϩ1 is indicated. on the binding of YY1 to the U site than it is to the D site. Mutating both sites together further reduces CAT activity, suggesting that the binding of YY1 to both sites is required for maximal promoter activity; furthermore, consideration of the activity associated with each mutant site compared with the double mutation suggests that the sites act independently. The titration of the basal promoter fragment with YY1 (Fig. 6) shows few molecules with two bound sites at lower YY1 concentrations, suggesting that one of the sites has a tighter association constant than the other; we presume it is the U site, based on the argument that bound YY1 must be present there to give the WT initiation pattern (Fig. 9A). The mutational data also suggests that YY1 binds to the U and D sites as an activator. The observation that mutating the YY1 sites affects promoter activity similarly in the three cell types tested is not surprising, since COX7C is ubiquitously expressed in all tissues, is not known to be differentially regulated, and fits with the general observation that ubiquitously expressed genes are activated by YY1 whereas highly regulated genes are repressed by it (32).
In transfected cells, transcription initiates at a cluster that includes the major start site at ϩ1, consistent with that seen in tissues. Binding of YY1 to the U site appears to control the efficiency of initiation from the major start site, since a mutation at YY1-U decreases dramatically the synthesis of this transcript. A YY1-D mutation appears to have little or no effect on the relative proportion of WT transcripts but decreases their net amount. It is less clear how the interaction of YY1 to the D site affects promoter activity (but see below). We also note that the YY1 double mutation does not completely abolish CAT promoter activity. At least 20 -45% of the CAT activity is still retained in these mutants, suggesting that the NRF-2 motif in the basal promoter could also contribute to promoter activity.
The close proximity of YY1-U to the transcriptional start site indicates that this binding site could function as an initiator, as seen for COX5B. YY1-U may therefore be involved in the formation or stability of the preinitiation complex, perhaps by interaction with the basal transcription machinery. The ability of YY1 to recruit Pol II to the initiation site by interaction with TFIIB, or TBP, has been demonstrated (46). Consistent with this view, we see that mutating the U site nearly abolishes initiation at the major start site, which accounts for nearly half of the transcripts arising from the WT promoter.
The effect of mutating the D site is less clear. YY1-D overlaps an NRF-2 site. The core motif for YY1 (CCAT, top strand) and NRF-2 (GGAA, bottom strand) are spaced only 3 nt apart. However, the four zinc fingers of YY1 are presumed to interact with 12 nt of the bottom strand, with the core motif positioned at the center (33). This indicates that the binding of one factor would sterically hinder the binding of the other. As a repressor, YY1 is known to overlap the DNA binding motifs of a wide range of activator proteins by precluding their binding. For instance, overlapping binding sites have been observed for YY1 and NF-B in the serum amyloid A1 gene promoter (47), YY1 and serum response factor in the skeletal ␣-actin gene promoter (41), YY1 and MGF in the ␤-casein promoter (48) and YY1 and GATA-1 in the ⑀-globin gene promoter (49). In COX7C, the significance of the YY1/NRF-2 overlap is not clear, especially since NRF-2 is also a general activator of COX gene expression (8). NRF-2, a member of the ets family of proteins that regulate genes involved in development, growth control, and cell transformation, has been found in the promoters of all ubiquitously expressed COX nuclear genes so far characterized: COX4 (7,11,12), COX5B (10), COX7AL (9), and COX7C. It is conceivable that the binding of these factors to the overlapping YY1/NRF-2 motif in COX7C is dictated by the steadystate levels of YY1 and NRF-2 in different tissues. YY1 levels are known to be decreased during the onset of the myogenic program (41). Alternatively, YY1 binding at the D site may serve to alter the topology of the promoter by DNA bending (44), which would facilitate protein-protein interactions or enhance protein binding to DNA. In the c-fos serum response element, where the serum response factor and YY1 binding sites overlap, it has been shown that both proteins can cooccupy this element, apparently facilitated by a YY1-induced bending of the serum response element DNA to enhance binding of serum response factor (50). This possibility can be extended to the YY1/NRF-2 motif of the COX7C promoter, and may explain why a mutation at the YY1-D site decreases CAT activity. If a YY1-induced bend facilitates enhanced binding FIG. 9. Transcriptional analysis of YYI mutations in transfected cells. Top panel, constructs harboring the WT basal promoter and the various YY1 mutations were transfected into HeLa cells. In vitro labeled antisense probes were used to protect RNA derived from the transfected plasmids (shown above each lane). Twenty g of total RNA were used, except for D (16 g). The bold arrow depicts the transcript arising from the major start site, and the small arrows represent the minor transcripts of the cluster. The major transcript in the WT and D lanes is nearly abolished in the U and UD lanes. Sequencing marker lanes are shown at left. Bottom panel, the four constructs used for transfection (thick lines) and the two probes used for protection (thin lines) are depicted. Bold regions represent the basal promoter region (Ϫ93 to ϩ74), and the open boxes within, the mutated YY1 sites; the blank regions represent vector sequences of pGKOCAT. The 62-bp vector sequence serves to identify transcripts arising from transfected DNA. The probes fully protect all transcripts initiating downstream of the WT transcription start site (B, arrow) but not those initiating upstream. Similar results were also obtained for myoblasts and myotubes. and NRF-2 (shaded circle) to the basal promoter is depicted. Below each factor, the binding motifs for YY1-U (Ϫ17 to Ϫ3) and for YY1-D/NRF-2 (ϩ59 to ϩ74) are shown, with the core sequence depicted in bold. The transcriptional and translational start sites are indicated by arrows.
of NRF-2 at the D site, it follows that a mutation would decrease the kinetics of NRF-2 binding, thus explaining the modest decrease in CAT activity.
COX7C (YY1/NRF-2), COX5B (YY1/NRF-1/NRF-2), COX7AL (NRF-1/NRF-2), and COX4 (NRF-2) are four ubiquitously expressed genes whose promoters have been analyzed in some detail. At least three factors appear to contribute to their expression. It is not clear why some of these are NRF-1-dependent and others YY1-dependent. NRF-2 appears to be a common factor in these genes and its interaction, if any, with YY1 and NRF-1 is not known. It is conceivable that some of these factors may sense the redox state of the cell and coordinate their interactions with other factors. A clearer picture will emerge only when the promoters of additional COX genes are characterized.