ETS Transcription Factors Regulate the Expression of the Gene for the Human Mitochondrial ATP Synthase P-Subunit”

Elements responsible for the transcriptional activity of the human ATP synthase psubunit (ATPsynp) gene promoter have been studied through transient expres- sion in HepG2 hepatoma cells of a CAT gene connected with various 5”deletion mutants of the S‘-flanking re- gion. Promoter activity was mostly dependent upon a single CCAAT motif as well as a nearby Ets domain bind- ing region. This last region contains two sites that bind Ets-related proteins present in liver nuclear extracts as well as recombinant purified Ets-1 protein. The ATPsynp promoter was trans-activated by Ets-1 and Ets-2 expression vectors, and this effect was lost when the Ets binding region was deleted. The Ets binding re- gion of the ATPsynp promoter increased basal expression and conferred Ets-l- and Ets-2-dependent trans-ac- tivation to the herpes symplex thymidine kinase minimal promoter. A double-point mutation of the main Ets-binding site, which suppresses Ets binding, blocks Ets-dependent trans-activation. It is concluded that the gene for the mitochondrial ATPsynp is a target of transcriptional activation by members of the Ets family of transcription factors. It is suggested that Ets transcription factors may be involved in the enhanced

The mitochondrial respiratory chain and oxidative phosphorylation activities rely on enzyme complexes present in the mitochondrial inner membrane of eukaryotic cells. In mammals each of these enzymatic complexes contains different protein subunits encoded by either the mitochondrial or the nuclear genomes. The mitochondrial respiratory chaidoxidative phosphorylation system is present in most mammalian cells and constitutes the main source of ATP synthesis.
The expression of nuclear genes for proteins of the respiratory chain and oxidative phosphorylation complexes shows several particular characteristics: a certain degree of ubiquitous constitutive synthesis of each component is necessary to maintain regular mitochondrial function, the amount of each component must be modulated according to the cell type and physiological demands for ATP, and functional complexes can be * This research was supported in part by Grant PB-92-0865 from Direcci6n General de Investigacion Cientifica y Bcnica, Ministerio de Educaci6n y Ciencia Spain. 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 constructed only when the expression of nuclear genes is coordinated both between each other and with the expression of the mitochondrial genome. Changes in the transcription rates of the nuclear genes for subunits of the respiratory chain and oxidative phosphorylation complexes play an important role in the modulation of their expression (for reviews see Refs. 1,2).
The ATPsynp' is a major component of the catalytic site of the mitochondrial F,F,-ATP synthase complex, and it is encoded by the nuclear genome (3). Whereas many nuclear-encoded components of the respiratory chain and oxidative phosphorylation systems show multiple isoforms encoded by several different genes (4-7), the ATPsynp is encoded by a single copy gene which is ubiquitously expressed in mammalian cells (8). Therefore, the identification of the elements responsible for the transcriptional regulation of the ATPsynp gene is expected to provide substantial information on the molecular mechanisms responsible for the coordinated expression of components of the respiratory chain and the oxidative phosphorylation systems. The human ATPsynP gene has been isolated and cloned (8,9). The structure of the promoter shows particular features: it lacks a canonical TATA box, it contains four CCAAT boxes and a GC box sequence and, depending on the authors, a single (8) or two (9) transcription start sites. Several elements involved in the transcription activity of this gene have been reported. An enhancer in the far upstream region of the gene has been identified (10, 11) as well as a more proximal region (OXBOX-REBOX complex) that includes muscle-specific and non-tissuespecific regulatory elements (12). A GABPmRF-a-binding site has also been identified in an intron of the ATPsynp (13).
Here we report that an Ets domain-binding site is a major element for the ATPsynp promoter activity, and this gene is found to be transactivated by members of the Ets family of transcription factors. The Ets proteins resemble the v-ets oncogene in the "ets domain." This domain interacts specifically with purine-rich sequences of DNAin viral or cell gene promoters. In addition to the c-ets-1 protooncogene product, Ets-1, about 30 Ets proteins have been found in different species, including humans, and they are involved in the regulation of gene expression during a variety of biological processes including growth control, transformation, and developmental programs of many organisms (for review, see Ref. 14). Recent reports have shown that transcription of the nuclear genes for two subunits of the cytochrome oxidase complex also depends on Ets-binding sites in their promoter regions (13, 15). The present finding that ATPsynp is a target for transcriptional regulation by Ets proteins suggests that these transcription factors are involved in coordinating the expression of components of the respiratory chain and oxidative phosphorylation systems in the mammalian cell.

Deletional
Analysis of the 5'-Flanking Sequence of the ATPsynS Gene-Various portions of the ATPsynj3 promoter were placed upstream of the CAT reporter gene (Fig. 1) and transfected in the HepG2 human hepatoma cell-line. The construct containing the fragment from -786 to +89 yielded significant levels of CAT expression with respect to the promoterless CAT vector PBLCAT3. The levels of CAT activity driven by this fragment of the ATPsynp gene were within the range of those of other mammalian or viral promoters expressed in HepG2 cells (80 and 75% of the CAT activities driven by the -490/+73 rat phosphoenolpyruvate carboxykinase and the -105/+X HSV thymidine kinase promoters, respectively). All deletion constructs gave significant CAT expression with the exception of the -U/+89 construct, the CAT activity of which was indistinguishable from PBLCATB expression background. The fragment spanning nucleotide -786 to -381 appeared to contain weak negative control elements, since its deletion was associated with a small but significant (p 5 0.001) stimulation of CAT activity. Deletion of this fragment suppressed a GC box, putative Spl-binding site, at -3881-382 (9) which does not appear to have a positive functional role here. Deletion of the region spanning from -381 to -339 did not further affect CAT activity, whereas the suppression of the fragment between -339 and -246 resulted in a significant (p 5 0.001) reduction. Further deletion of the fragment between -246 and -220 caused a The DNA fragmrnt was end-lahelrd by using Klenow enzyme activity over the 5' extruding end resulting from thr XhnI digestion of the p381CAT plasmid (see "Experimental Procedures" for details). It was suhjected to DNase 1 fnotprinting using 0.5, 10, or 25 pg of nuclear protein extract from rat liver and electrophoresed using a 6Rr denaturing acrylnmide gel. The boxes on thr Irfl rnclose the rrgions protected from DSasr I digestion.
significant ( p 5 0.001) 4-fold decrease in CAT activity. Therefore, positive elements for the promoter expression are present in both regions. Deletion of -246 to -220 suppresses the upstream CCAAT sequence in the promoter, and the low CAT activity levels found in this construct indicates that this CCAAT box may be an essential element for ATPsynp promoter activity. The region between -220 to +89, which contains three additional CCAAT motifs, drove a CAT activity significantly above the background, but it was 4-fold lower than that found when the -2461-220 region was present.
Identification of Protein-hinding Sites in the ATPsynp Gene Promoter-Since the region between -339 to -220 was responsible for most of the ATPsynp promoter activity in HepG2 cells, DNase I protection assays were performed to identify binding sites for potential regulatory proteins using nuclear extracts from rat liver. Fig. 2 shows an example of this analysis. Two distinct areas of protection were detected. The first corresponded to -2481-230 and overlapped the upstream CCAAT sequence in the promoter. Nuclear proteins binding at that region showed a high affinity and/or high abundance in the rat liver extracts as amounts as low a s 1 pg of nuclear protein extracts yielded a marked footprint signal (not shown). The second area of protection, named E region, was the only one found within the remaining DNA region from -339 to -248 where positive acting elements for promoter activity had been detected. It was placed between -306 and -276, and the analysis of the corresponding DNA sequence indicated the presence of two sites with a close homology with the DNA binding sequence for the ETS family of transcription factors (see below).
The -3061-266 Region of the ATPsvnP Gene Promoter Rinds Ets-related Proteins-Two sequences (El and E2) homologous to the DNA-binding sites for Ets transcription factors were detected within the E region of the ATPsynP promoter. A third Ets-related sequence (E.3) was also detected in the 3' immediate vicinity although it did not correspond to a DNase I footprint protected site (Fig. 3A ). They are compared with the core Ets-binding consensus DNA sequence (2.5). and they show a full conservation of the A/C GGAA core motif characteristic of the Ets-binding sites.
Identification of the proteins binding the -3061-266 region as members of the Ets family was first achieved by using competition gel shift analysis. When two DNA fragments containing t h e E l sequence (-3061-293) or the E2 (+E3) sequences (-2921 -266) were incubated with protein extracts from rat liver nuclei, several retarded bands were observed in the gel-shift assay (Fig. 3R ). Oligonucleotides corresponding to DNA-binding sites for Ets-2 and GABPMRF2, two members of the Ets family known to be expressed in the liver (26, 27). were used for competition. These were the Ets-lEts-2-binding site (-2081 -192) in the stromelysin promoter (28) and the GABPINRF2binding site (+19/+35) of the rat cytochrome oxidase subunit IV promoter (13.29) (Fig. 3B, right ). An AT-rich, MEF-2 consensus oligonucleotide was used a s a negative control (30). Results indicated that the El site originated a main retarded band that was lost when an excess of either GARPINRF2 or Ets-l/Ets-2 oligonucleotide was added to the incubation mixture (see Fig.  3B, left ). However, lower amounts of the GABPMRF2 oligonucleotide were able to compete fully for binding when compared with the Ets-l/Ets-2 oligonucleotide action. This indicates that the El site of the ATPsynp promoter binds Ets proteins present in liver nuclei, and they have a somewhat higher affinity for GABPMRF2 sites than for Ets-l/Ets-2-binding sites. Conversely, only high amounts of the Ets-l/Ets-2 oligonucleotide caused a loss of shifted bands when the fragment containing the E2 and E3 sequences was analyzed (see Fig. 3R, center) thus indicating that the E2+E3 region has a weaker affinity for the Ets-related proteins present in liver nuclear extracts.
The presence of shifted bands that are not affected by Etsrelated competitors indicated that non-Ets proteins also interact with the -3061-266 region of the ATPsynp. Databank analysis of the DNA sequence within this region did not yield any consistent homology with previously identified DNA binding motif., with the exception of Ets. The GC-rich stretch between the El and E2 sites showed a single mismatch for full homology with a GC box putative Spl-binding site. It was tested in competition gel-shift assays using a consensus Spl-binding site oligonucleotide (31), and result.. were negative (not shown).

The -3061-266 Region of the ATPsynp Promoter Contains Binding Sites for Ets-I-For
further evaluation of the ability of sequences from -306 to -266 in the ATPsynp gene to bind Ets-related proteins, the two fragments -3061-29.3 and -292f -266 were subjected to gel-shift assays using recombinant purified Ets-1 protein. As shown in Fig. 4, incubation of either fragment with Ets-1 protein resulted in a single retarded band. Binding was specific, as an excess of cold Ets-l/Ets-2 oligonucleotide suppressed the appearance of the retarded band, whereas the MEF-2 oligonucleotide did not. When equimolar

-266 c t a g C M C A~C T C O O C C C C T l l C C T~C G T A G l l c C l C T~ CTTGTCCTTOAOCCOOOCAMCOATTTGCATCM~AGACTgatc
ee-E l E2 E3

AIC C 0 A A GIC E t s core consensus A O G A A C E l A G G A A A E2 A G G A A C E3
(he two end-labelrd fragmrnts rrsulting from the IInrIII digestion of the -3OW -2fi6 doublr-stranded oligonucleotide dr- amounts of labeled -3061-293 ( E l ) or -29Z-266 (E2+E3) fragments were incubated with equal amounts of Ets-1, the percentage of radioactivity in the retarded band was always lower in the E2+E3 fragment than in the El. These results confirm the presence of Ets-binding sites in both fragments, but also the much lower affinity of the E2+E3 fragment than the E l in binding Ets proteins. On the other hand, the mobility of the retarded bands caused by recombinant purified Ets-1 was always lower than the Ets-related bands obtained using liver nuclear extracts (not shown).

Ets-I and Ets-2 Expression Duns-actiuates the ATPsyn P-Subunit Gene Promoter Due to Elements Present at Position
-339/-247"The P786CAT plasmid was co-transfected with expression plasmids containing the cDNAs of either c-Ets-1, h-Ets-2, or the distantly related member of the Ets family Pu-1. A deletion mutant form of the c-Et..-1 expression vector unable for DNA binding was used as negative control (16). As shown in Fig. 5, both Ets-1 and Ets-2 trans-activated the ATPsynP gene promoter whereas the Pu-1 expression vector had no effect. Deletion mutants P381CAT and P339CAT were also trans-activated severalfold over the negative control by c-Ets-1 and h-Ets-2 expression vectors. The 8246 CAT and P l l CAT constructs were unresponsive to Ets (Fig. 5). These results indicate that the ATPsynP promoter is a target for trans-activation by Ets-related transcription factors, and that cis-acting elements responsible for trans-activation occur in the -3391-247 region, where Ets-binding sites had been detected. The two rrsulting lahelrd fragmrnts -306/-293 and -292/-266 were incuhnted with 0.5 pg of highly purifird bacterially expressrd murine Ets-1 protrin and rlectrophoresed on 5% acrylamide gel. The Et--1lEts-2 oligonucleotide competitor and the MEF-2 negative control (see Fig. 3) wrre added at the indicated molar rxcess. The ahsence of competitnr is drpictrd as -. activation-When the construct transfected to HepG2 cells contained a version of the -3061-266 region in which the AGGAA motif at -3021-297 (El site) had been mutated to AttAA, the capacity to be trans-activated by Ets-1 expression was lost (see Fig. 6). This is consistent with the previously reported requirement for an intact GG pair in Ets-responsive elements (23). However, the presence of the double point mutation did not block the enhanced basal expression of the construct with respect to TK promoter alone, previously observed for the wildtype fragment. Band shift analysis of a labeled probe corresponding to the mutated -3061-293 region using recombinant Ets-1 protein indicated that the double point mutation suppressed the ability of the E l site t o bind Ets-1 (not shown).

Y .
These results indicate that the main Ets-binding site in the ATPsynp promoter, site E l , is necessary for Ets-dependent trans-activation of the promoter.

DISCUSSION
In this study it is demonstrated that the main basal promoter activity of the ATPsynp gene in hepatic cells depends on two cis-acting regions: a CCAAT box site, analogous to that present in many eukaryotic gene promoters, and a nearby upstream region containing Ets domain-binding sites. The ATPsynp gene promoter is identified as a target for Ets-1 and Ets-2 trans-activation due to the presence of an Ets domain binding region.
The Ets-responsive region of the ATPsynp gene promoter presents a complex structure as it contains at least two Ets- binding sites (El and E2) in opposing orientation, each one showing different affinities for Ets proteins. Ets-responsive elements composed of different arrangements of more than a single Ets-binding site have been previously identified (23, 28, 29, 32, 33). The opposite alignment of the E l and E2 sites is similar to the palindromic alignments of individual Ets-binding sites in the Ets-l/Ets-2-responsive elements of the stromelysin (28) or the GATA-1 genes (34). The present observation that either Ets-1 or Ets-2 transactivate the ATPsynp promoter is also in agreement with previous reports showing that these two closely similar members of the Ets family activate mammalian or viral promoters through common Ets-1/Ets-2-responsive elements (28, 35). The lack of effect of Pu-l, the most divergent member of the Ets family (171, is also consistent with its distinct binding specificity and lack of cross-transactivation with other Ets-responsive elements (36). In this context, the finding that the Ets proteins in liver nuclei which bind the Ets-responsive elements in the ATPsynp gene have high affinities for both Ets-l/Ets-2 and GABPLNRF2 sites is not surprising. Crossbinding is a characteristic feature of closely related members of the Ets family. Ets-2, which is substantially expressed in the liver, binds the same DNA sequences that can bind Ets-1 (37, 38). In addition, the Ets-lEts-2-binding sites of the MSV and the polyoma PEA3 promoters are able to bind GABPINRF2, another Ets protein expressed in the liver (15, 29). Therefore, the combination of the functional and binding studies shown here does not necessarily point to a single Ets-related protein as the unique mediator of Ets-dependent ATPsynP promoter activity in liver cells. The higher affinity for GABP/NRF2 suggests that this member of the Ets family might be the main Ets protein in the liver interacting with the E l region of the ATPsynp gene. However, the relevance of the binding with other members of the Ets family such as Ets-2 and the possibility of a complex cross-talk of binding and trans-activation activities from different members of the Ets family cannot be ruled out.
The presence, within the Ets-responsive region of the ATPspP promoter, of DNA elements which bind proteins other than Ets, indicated by the presence of shifted bands unaffected by Ets competitors, is remarkable. Their functional relevance