ATPF1 binding site, a positive cis-acting regulatory element of the mammalian ATP synthase alpha-subunit gene.

An analysis of the promoter of a bovine and human nuclear-encoded mitochondrial ATP synthase alpha-subunit gene (ATPA) revealed the presence of a positive control element. DNase I footprinting and electrophoretic mobility shift assays demonstrated that this cis-acting regulatory element contains a binding site for a protein present in human HeLa nuclei, termed ATP factor 1 (ATPF1). The ATPF1 binding site contains the sequence, CANNTG, a sequence identical to the recognition site for a family of transcription factors containing a basic region adjacent to a helix-loop-helix domain. Site-directed point mutations of the basic helix-loop-helix binding site demonstrated the critical role of this element for both binding of ATPF1 and also for transcriptional activation of the ATPA gene.

is a complex process requiring the coordinate expression of both nuclear and mitochondrial genes. For example, two of the subunits of the ATP synthase are encoded by the mitochondrial genome in animal cells, whereas the remaining subunits of this enzyme complex are encoded by nuclear genes and imported into mitochondria (for review see Ref. 1). The ATP synthase complex offers a model system to study the molecular mechanism(s) of nuclear and mitochondrial gene interactions.
A regulatory system must exist to coordinate the expression of the nuclear and mitochondrial genes that encode proteins of the oxidative phosphorylation system to meet varying energy demands. It has been found in the yeast Saccharomyces cereuisiae that transcriptional control of nuclear genes encoding proteins of the oxidative phosphorylation system is mediated through specific cis-and trans-acting elements in response to heme metabolism, oxygen, and catabolite repression (see Refs. nuclear genes that encode subunits of the mammalian mitochondrial ATP synthase complex. Recently, we reported the isolation and characterization of a nuclear gene (ATPA) that encodes a bovine (4) and human (5) mitochondrial ATP synthase a-subunit protein.
We have determined that the 5"flanking region of each of these genes acts as a promoter element (4, 5). In this study, we have analyzed the cis-acting sequences which regulate the expression of these genes. Using a deletion analysis, we have identified a positive cis-acting regulatory element, that is required for basal expression of the ATPA gene. We have determined that a trans-acting factor present in human HeLa nuclei, termed ATP factor 1 DNA Sequence Analysis-DNA sequences of alkaline-denatured double-stranded plasmid DNks (7, 8 ) were determined by the dideoxy chain termination method (9) using Sequenase kits (U. S. Biochemical Corp.) and the method recommended by the manufacturer.
Polymerase Chain Reaction Amplification-Polymerase chain reactions (PCR) were carried out in a total volume of 100 pl using 100 ng of each primer, 100 ng of u20B DNA (4), 200 p~ of each dNTP, 7bq DNA polymerase buffer (Promega, Madison, WI), and Taq DNA polymerase (2.5 units, Promega). Samples were heated at 95 "C for 3 min to denature the template DNA and subjected to 30 cycles of amplification of 94 "C for 1 min, 55 "C for 1 min, and 72 "C for 1 min. Mutations were introduced into the ATPA gene promoter using oligonucleotides containing specific mismatches and PCR reactions.
Plasmid Construction-To facilitate cloning of the 5"flanking region of the ATPA gene into the reporter plasmid pCAT-Basic (Promega), primers were synthesized containingHindII1 andXbaI restriction sites.
After PCR amplification, the products were digested with HindIII and XbaI (Life Technologies, Inc.) and ligated to pCAT-Basic DNA that had been digested with HindIII and XbaI.
'Ifansfections and Enzyme Assays-Human HeLa cells were used in the transfection experiments. Cells were maintained in Dulbecco's modified Eagle's medium (Irvine Scientific, Irvine, CA) supplemented with 5% calf serum. For DNA transfections, 20 pg of cesium chloridepurified PATPA-CAT DNA and 5 pg of pCMV-P-galactosidase DNA (10) were co-precipitated with calcium phosphate (8). After 2 days, the cells were harvested and the lysates were assayed for P-galactosidase (7) and CAT (11) activities. Prior to assaying for CAT activity, cell extracts were heated at 65 "C for 15 min. CAT activity was assayed using L3H1acetyl-CoA (Amersham Corp.) as a substrate. Promoter activity values represent the average of at least three separate transfections of three plates each.
DNase I Footprinting Assays-Fragments were prepared from the plasmid pATPA-9ICAT by digestion with either XbaI (to label the coding strand) or HindIII (to label the noncoding strand) and 3'-end-labeled with [&"P]dNTPs using Klenow enzyme. Binding reactions contained approximately 10 fmol of labeled fragment in a 100-pl reaction volume containing 25 m~ Tris, pH 7.9, 6.25 rn MgCIZ, 1 m~ EDTA, 1 rn dithiothreitol, 50 m~ KC1, 0.1 pg of poly(d1-dC), plus approximately 30 pg of HeLa nuclear extract (12). After incubation at room temperature for 15 min, 50 pl of 10 m~ MgCI,, 10 m~ CaCI, was added and incubated at room temperature for 1 min. DNase I (Promega) was then added (0.10 unit for reactions without extract and 1.5 units for reactions with extract) and incubated for an additional 1 min at room temperature. Reactions were then terminated, extracted with phenol, and precipitated with ethanol. Samples were then separated by electrophoresis on denaturing polyacrylamide-urea gels.
Electrophoretic Mobility Shifi DNA-Protein Binding Assays-A double-stranded oligonucleotide encompassing the +23 to +50-bp region of the bovine ATPA gene was synthesized and used as a probe for electrophoretic mobility shift protein binding assays. The probe was labeled at its 3'-end using the Klenow fragment of DNA polymerase I and the appropriate [32PldNTP. Approximately 0.1 ng of probe (5,000-10,000 cpm) was incubated with 1 pg of HeLa cell nuclear extract as described (8). Complexes were resolved on 4% polyacrylamide gels using 1 x Tris-borate-EDTA (7) as the running buffer.

RESULTS
Mapping the Basal Promoter Activity of the ATPA Gene-We have shown previously that a 700-bp fragment from the upstream region of the bovine ATPA gene (from -538 to +161 bp relative to the most 5'-transcription start site; Ref. 4) can efflciently drive expression of a reporter CAT gene in HeLa cells. Furthermore, we have demonstrated that this fragment acts in an orientation-dependent manner (4). This plasmid was used as the starting point for mapping cis-acting DNA regulatory elements required for controlling expression of the ATPA gene.
Potential cis-acting elements were initially identified using a series of 5"deletion mutants. These deletion mutants were constructed using either naturally occurring restriction endonuclease sites or synthetic oligonucleotides and PCR reactions. Each of these deletion mutants was transfected into human HeLa cells, and the levels of the reporter CAT enzyme were determined. The results of these transfection experiments indicated that deletion of the 5"flanking sequences of the ATPA gene to within 9 bp of the most 5'4ranscription start site had essentially no effect on promoter activity (Fig. 1). Furthermore, deletion of sequences to +25 bp downstream of this start site also had little effect on promoter activity (Fig. 1). However, further deletion to +49 bp dramatically reduced promoter activity and deletion to +71 bp essentially reduced promoter activity to background levels (Fig. 1). These results indicate the importance of the +25 to +70-bp region of the ATPA gene for basal promoter function. Ref. 4) was amplified by PCR and cloned upstream of the Escherichia coli CAT gene in the vector, pCAT-Basic. The generalized fusion gene structure is shown at the top. Major sites of transcription initiation of the ATPA gene are indicated by arrows (4). A series of 5"deletions in the ATPA gene were constructed using either restriction endonuclease sites or oligonucleotides primers and PCR reactions. The position in base pairs of each of the deletions is indicated above the lines. Each pATPA-CAT construct was transfected into human HeLa cells, together with a P-galactosidase expression plasmid. The CAT activities in the cell extracts were assayed and normalized relative to the P-galactosidase activities. The activity obtained with the pATPA-538ICAT plasmid was considered 100% activity.
fragment of the ATPA gene promoter from -9 to +135 bp was end-labeled and used as a probe in these assays. The results of these experiments revealed that two regions of the ATPA gene promoter were protected from nuclease digestion by the HeLa extracts. One region extended from nucleotide +26 to +49 and coincided with one of the regions identified by the 5'-deletion analysis to be critical for basal promoter activity (Fig. 2). A second (and weaker) protected region of the ATPA gene promoter extended from nucleotide +72 to +81 (Fig. 2).
Detection of DNA-Protein Interactions Using Electrophoretic Mobility Shift Assays-The 5"deletion analysis and the DNase I footprinting assays identified a region of the ATPA gene within the sequence from +25 to +49 bp that could serve as a potential regulatory site involved in DNA-protein interactions.
To further examine the ability of these sequences to interact with trans-acting factors, complementary oligonucleotides that encompass this region were tested in electrophoretic mobility shift assays together with nuclear extracts prepared from HeLa cells. When the products of these binding reactions were separated by electrophoresis in low ionic strength polyacrylamide gels, several complexes were detected (Fig. 3). All of these complexes were competed by excess oligonucleotide (Fig. 3). No competition was detected using nonspecific oligonucleotides (data not shown).
A search for consensus sequences for binding sites of known transcription factors (13) within this region of the ATPA gene revealed two potential binding sites for previously characterized factors. One of these sequences, ACATCCGG, matches the complement of the consensus binding sequence, C C G G M TGT/C, for the ets-1 family of transcription factors (14). The other sequence, CACGTG, is the consensus recognition sequence for the bHLH-containing family of DNA-binding proteins (6). To determine if either of these sequences contribute to the ATPA-protein complexes observed in the electrophoretic mobility shift assays, oligonucleotides corresponding to the ETS and bHLH binding sites were tested as competitors in mobility shift assays. The results of these experiments revealed that the ETS oligonucleotide competed only weakly for formation of the ATPA-protein complexes (Fig. 3). In contrast, an oligonucleotide containing the bHLH binding site effectively competed for formation of most of the ATPA.HeLa complexes (Fig. 3).
USF is a member of the bHLH family of DNA-binding proteins (15). USF was originally identified as a a cellular factor that binds to an upstream stimulatory element in the adenovirus major late promoter (16, 17) and has also been found to regulate the transcription of several cellular genes (18-21). To determine if the protein complexed to the ATPA regulatory sequence was the same as USF, we tested the effect of antiserum to USF (15) on the mobility of the ATPA-protein complexes using electrophoretic mobility shift assays. The results of these experiments revealed that there was essentially no difference between the mobility of the ATPA-HeLa complexes that were reacted with anti-USF antiserum compared with those that were treated with a preimmune serum, suggesting that the nuclear factor bound to the ATPA regulatory element is distinct from USF (Fig. 3).
To identify specific residues within this cis-acting regulatory sequence that are involved in the DNA-protein interactions, several complementary oligonucleotides were synthesized containing point mutations within this sequence (Fig. 4). These oligonucleotides were tested as competitors in electrophoretic mobility shift experiments using the wild-type ATPA regulatory site as a probe. The results of these experiments revealed that oligonucleotide mut 1, which contains a 2-bp substitution in the ETS domain binding site, effectively competed for binding of HeLa nuclear extracts to the wild-type ATPA probe (Fig. 4). Oligonucleotide mut 2, which has a 3-bp substitution in the sequence adjacent to the core bHLH site, competed partially for formation of the ATPA-HeLa complexes. In contrast, oligonucleotide mut 3, which contains a mutation altering the CA 1 2 3 4 5 6 7 FIG. 3. Characterization of ATPA-HeLa complexes using electrophoretic mobility shift assays. A double-stranded oligonucleotide encompassing the +23 to +50-bp region of the ATPA gene was synthesized, labeled with I:'2PldATP, and used as a probe in electrophoretic mobility shift assays. Binding was performed in the absence (lane 1 ) or the presence (lanes 2-7) of approximately 1 pg of HeLa nuclear extract. Complexes were formed in the absence (none) or the presence of competitor oligonucleotides or antiserum, as indicated above the lanes. A 100-fold molar excess of unlabeled competitor oligonucleotide was used.
The following oligonucleotides were used as competitors. The ATPA oligonucleotide is the +23 to +50-bp region of the bovine ATPA gene. The ETS oligonucleotide is the +13 to +36-bp region of the rat cytochrome c oxidase subunit IV gene which has been shown to contain a functional ETS binding site (33)(34)(35). The bHLH oligonucleotide contains the USF binding site of the adenovirus major late promoter (16, 17). The anti-USF antiserum (1:lOO dilution) was a polyclonal antiserum raised against the 43-kDa USF (15). residues of the core bHLH site (CANNTG) together with the 5"flanking T residue, competed only weakly for formation of the ATPA-protein complexes (Fig. 4). However, oligonucleotides mut 4 and mut 5, which contain base pair substitutions changing the NNTG residues of the bHLH binding site, effectively competed with the wild-type ATPA regulatory element for formation of the DNA-protein complexes (Fig. 4). These data demonstrate the importance of the CA residues of the bHLH binding site for formation of the ATPA-HeLa complexes.
The nucleotide sequence of this cis-acting regulatory element identified in the bovine ATPA gene is highly homologous to the corresponding region of the human ATPA gene (Fig. 5;Ref. 5). It is therefore likely that the corresponding region of the human gene also acts as a regulatory element controlling expression of this gene. In support of this idea is the finding that oligonucleotides containing this region of the human ATPA gene effectively competed for binding of HeLa nuclear extracts to the bovine regulatory element (data not shown).
D-ansfection of HeLa Cells with Mutated ATPA-CAT Constructs-The electrophoretic mobility shift and DNase I protection assays indicated that a factorb) in HeLa nuclear extracts could bind to the ATPA promoter in the region from +25 and +49 bp. Furthermore, site-directed mutations in this region identified several nucleotides which appear to be critical for this binding. These in vitro assays do not, however, address the functional consequences of these DNA-protein interactions. In order to assay the effects of mutations in the ATPA cis-acting regulatory element on basal transcription, each of the mutant constructs, pATPA mut 1, mut 2, mut 3, mut 4, and mut 5-CAT, was transfected into HeLa cells in parallel with the corresponding unmutated PATPA-CAT construct. The results of these experiments revealed that the pATPA mut 1-CAT construct had slightly reduced transcriptional activities when compared with the wild-type construct ( Table I) mid showed a reduction in transcriptional activity to levels approximately 36% of wild-type (Table I). The pATPA mut 3-CAT construct showed an even more dramatic reduction in transcriptional activity to levels approximately 11% of wildtype, whereas the pATPA mut 4-CAT and pATPA mut 5-CAT constructs had activities that were comparable or even greater than those of the wild-type PATPA-CAT plasmid ( Table I).
These results indicate that residues within this cis-acting regulatory element of the ATPA gene promoter are essential for forming a DNA-protein complex required for expression of this gene.

DISCUSSION
In this paper we define a cis-acting regulatory sequence that is required for basal expression of a bovine and human nuclearencoded mitochondrial ATP synthase 0-subunit gene (ATPA). Using electrophoretic mobility shift and DNase I footprinting assays, we have determined that a trans-acting nuclear factor(s) present in human HeLa nuclei, termed ATPF1, binds to this regulatory sequence.
The core of this cis-acting regulatory site contains the sequence, CACGTG, which matches the consensus binding site (CANNTG) for a large family of DNA-binding proteins that have a common structural feature termed the bHLH domain. This domain consists of a stretch of basic amino acids followed  I I I I I I I I I I I I I I I I I I I I I l l FIG. 5. Comparison of the bovine ATPA cis-acting regulatory element with the corresponding region of the human ATPA gene. The cis-acting regulatory element identified in the bovine ATPA gene (+23 to +49 bp) is aligned with the corresponding region (-57 to -31 bp) of the human ATPA gene (5). Bases which are identical between the two sequences are indicated by a vertical line.

TABLE I
The effect of mutations in the ATPA gene on basal promoter activity The ATPA gene from +23 to +135 bp was cloned into the plasmid pCAT-Basic to generate pATPA wild-type-CAT. Mutations were introduced into this sequence using oligonucleotide primers and PCR reactions. The bases which were mutated are the same as those shown in the legend to Fig. 4. Wild-type and mutant pATPA-CAT constructs were transfected into HeLa cells. The CAT activities of the transfected cells were assayed and normalized relative to the 0-galactosidase activities. The normalized CAT values represent the average of a t least three independent determinations. The activity obtained with the pATPA wild-type-CAT plasmid was considered 100% activity.

Constructs
Relative CAT activity pCAT-Basic 0 pATPA wild-type-CAT 100 pATPA mut I-CAT 74 pATPA mut 2-CAT 36 pATPA mut 3-CAT 11 pATPA mut 4-CAT 161 pATPA mut 5-CAT 94 by a conserved sequence of amino acids presumed to form two 0-helices interrupted by a loop. Members of this family include the transcription factor, USF, that binds to an upstream sequence of the adenovirus major late promoter (15, 16); transcription factors, TFE3 (22), TFEB (23), and E12, E47 (24), which bind to specific DNA sequences found in immunoglobulin enhancers; proteins such as MyoD (25), myf-5 (26), myogenin (27), achaete-scute (28), and daughterless (29), which play important roles in cell determination; and the Myc family of oncoproteins (see Ref. 301, together with their binding partners, such as M d y n (31). It is likely that the trans-acting protein factor, ATPF1, required for basal expression of the ATPA gene is a member of this bHLH family of DNA-binding proteins. Indeed, using a mutational analysis we have determined that binding of ATPFl to specific residues within the bHLH sequence is critical for transcriptional activation of the ATPA gene.
One member of the bHLH family, transcription factor USF, has been found to play an important role in the regulation of several cellular genes, including genes for mouse metallothionein I (18), rat fibrinogen (19), rat human growth hormone (20), Xenopus transcription factor TFIIIA (32), and duck histone H5 (21). To determine ifATPFl was the same as USF, we tested the effect of anti-USF antiserum (15) on the mobility of the ATPA.HeLa protein complexes. The results of these experiments suggested that ATPFl is distinct from USF (Fig. 3).
The cis-acting region of the ATPA gene found to be important for basal promoter activity also contains a potential binding site for transcription factors with an ETS domain (see Ref. 14). The results of both our in vitro ( Fig. 3 and 4) and in vivo ( Table  I) experiments suggest that binding of an ETS domain factor to the ATPA regulatory element is not essential for basal transcription of the ATPA gene. This contrasts to the rat and mouse nuclear genes that encode cytochrome c oxidase subunit rV, whose basal promoters consist in a large part of two tandemly linked ETS domain binding sites (33)(34)(35) (see also below). It is possible that an ETS domain protein acts in concert with ATPFl to modulate expression of the ATPA gene. ETS proteins have been found to cooperate with other transcription factors in the transcriptional activation of a number of genes (see Refs. It is becoming evident that not all genes encoding subunits of the oxidative phosphorylation system in mammalian cells are regulated by the same cis-and trans-acting factors. Several distinct cis-acting elements have now been identified which are important for the transcriptional initiation of some mammalian nuclear-encoded oxidative phosphorylation genes. For example, the 5"flanking region of the rat and human somatic cytochrome c genes contain a binding site for a trans-acting factor, termed 40). A NRF-1 site is also found in the genes that encode rat cytochrome c oxidase subunit VIc-2, human ubiquinone-binding protein, mouse and human MRP RNA, and bovine ATP synthase y-subunit (40). Similarly, transcription of the rat and mouse cytochrome c oxidase IV genes requires binding of a trans-acting factor, NRFS, to a cis-acting element (33). The core of this element has the sequence, GGAAG, which is the consensus binding site for proteins with an ETS domain (14). NRF-2 has been shown recently to be similar or identical to the ETS domain protein, GABPa (34,351. Other studies have shown that transcription of the mouse cytochrome c oxidase Vb gene involves a cis-acting regulatory sequence to which a factor that is similar or identical to NF-E1 (also called 6, YY1, FACT11 binds (41). This NF-E1 sequence is adjacent to two ETS binding sites which may also play a role in the transcriptional initiation of the COXVb gene (33,41). In addition, several cis-acting sequences, termed Mtl, Mt3, and Mt4, have been found in the 5'-flanking region of several human oxidative phosphorylation genes, including ATP synthase P-subunit, cytochrome cl, and the ubiquinone-binding protein (42). Similarly, two cis-acting elements, termed the OXBOX site and the REBOX site, have been identified which are important for transcriptional activation of the human ATP synthase P-subunit gene and the adenine nucleotide translocator 1 gene (43,44). The OXBOX sequence binds a muscle-specific protein and might play a role in the tissue-specific regulation of these genes. The REBOX element binds a factor which appears to be ubiquitous, the binding of which can be modulated by the RE-DOX potential and the pH of the medium (44). Last, a sequence that acts as a n "enhancer" element is present in the upstream region of the human gene encoding ATP synthase P-subunit, as well as the genes encoding human cytochrome c1 and pyruvate dehydrogenase subunit E l a (45). Our data adds another cisacting element and a trans-acting factor (ATPF1) to the growing list of factors important for the transcriptional regulation of mammalian oxidative phosphorylation genes. ATPFl appears to be distinct from the previously described factors. For example, there is no similarity between the recognition sequence forATPFl and those for NRF-1, NRF-2, or NF-E1 or to the Mtl, Mt3, Mt4, OXBOX, and REBOX elements. In addition, there was essentially no competition by oligonucleotides containing either an NRF-1 site (data not shown), a NRF-2 site (Fig. 3), or an OXBOX/REBOX site (data not shown) for the binding of HeLa nuclear extracts to the ATPF1-responsive element when tested using electrophoretic mobility shift assays.
It is intriguing that the regulatory sequence found in this work to be required for basal transcription of the mammalian ATP synthase a-subunit gene is homologous to two sequences that have been shown to be important for transcriptional activation of several nuclear-encoded mitochondrial genes in the yeast S. cerevisiae (Fig. 6). One of these sequences has the consensus RTCRYNNNNACG and has been found to bind a protein factor calledABF1 (also termed SBF-B, TAF, SUF, GFI, and BAF'1) (see Ref. 46). Binding of ABFl is required for the transcriptional activation of the genes that encode cytochrome c oxidase subunit VI (47,481 and subunit VI11 of the ubiquinolcytochrome c oxidoreductase complex (49) in response to carbon [36][37][38]. source. The second sequence has the consensus RTCACRTG and binds a protein factor called CPFl (also termed CBP1, CB1, and GFII) (46). Binding of CPFl is important for regulating the expression of the gene that encodes subunit VI11 of the ubiquinol-cytochrome c oxidase complex (49). CPFl is a member of the bHLH family of transcription factors (50). ABFl and CPFl also bind to a number of genes that encode proteins involved in processes important for cell growth and to elements important for cell division and it has been postulated that ABFl and CPFl are involved in coupling the rate of mitochondrial biosynthesis to cellular growth (46). It is possible that some transcription factors which control the expression of nuclear genes encoding mitochondrial proteins are similar between both lower and higher eukaryotic organisms.