Differential Regulation of the Catalytic and Accessory Subunit Genes of Drosophila Mitochondrial DNA Polymerase*

The developmental pattern of expression of the genes encoding the catalytic (α) and accessory (β) subunits of mitochondrial DNA polymerase (pol γ) has been examined inDrosophila melanogaster. The steady-state level of pol γ-β mRNA increases during the first hours of development, reaching its maximum value at the start of mtDNA replication inDrosophila embryos. In contrast, the steady-state level of pol γ-α mRNA decreases as development proceeds and is low in stages of active mtDNA replication. This difference in mRNA abundance results at least in part from differences in the rates of mRNA synthesis. The pol γ genes are located in a compact cluster of five genes that contains three promoter regions (P1–P3). The P1 region directs divergent transcription of the pol γ-β gene and the adjacent rpII33 gene. P1 contains a DNA replication-related element (DRE) that is essential for pol γ-β promoter activity, but not for rpII33 promoter activity in Schneider's cells. A second divergent promoter region (P2) controls the expression of theorc5 and sop2 genes. The P2 region contains two DREs that are essential for orc5 promoter activity, but not for sop2 promoter activity. The expression of the pol γ-α gene is directed by P3, a weak promoter that does not contain DREs. Electrophoretic mobility shift experiments demonstrate that the DRE-binding factor (DREF) regulatory protein binds to the DREs in P1 and P2. DREF regulates the expression of several genes encoding key factors involved in nuclear DNA replication. Its role in controlling the expression of the pol γ-β and orc5genes establishes a common regulatory mechanism linking nuclear and mitochondrial DNA replication. Overall, our results suggest that the accessory subunit of mtDNA polymerase plays an important role in the control of mtDNA replication in Drosophila.

mitochondrial proteins, including those involved in mtDNA replication and transcription, are encoded in the nucleus, translated on cytoplasmic ribosomes, and targeted by a complex import mechanism to the mitochondrial compartment (1,2).
The molecular mechanisms involved in mitochondrial proliferation and maturation (i.e. mitochondrial biogenesis) during cell division, growth, and differentiation are poorly understood. During early embryonic stages, the mitochondria that are stored in the oocyte are segregated into various cellular territories; as development proceeds, mitochondrial proliferation begins. When differentiation ceases, the cells contain the correct number of mitochondria with the specific morphology and intracellular distribution to meet the highly variable energetic requirements of the various tissues of the organism (3,4). How the mitochondrial mass, the extent of mitochondrial differentiation, and the mtDNA copy number are controlled in a defined cell type is not known.
In addition to this increase in mitochondrial biogenesis in response to unidentified developmental cues, cells can also regulate the expression of the genes encoding mitochondrial proteins in response to a variety of physiological signals, including hormones, cyclic nucleotides, contractile activity, and thermogenesis (5)(6)(7)(8). Several DNA elements recognized by regulatory proteins involved in the transcriptional regulation of nuclearly encoded mitochondrial genes have been identified in mammals, including Mts, REBOX/OXBOX, and GRBOX elements (9 -11). In addition, two transcription factors involved in nucleo-mitochondrial communication, nuclear respiratory factors 1 and 2 (NRF1 1 and NRF2), have been characterized during the last few years by Scarpulla and co-workers (12)(13)(14). NRF1 activates the expression of several genes that encode functional components of the oxidative phosphorylation pathway and key elements of the transcriptional/replication machinery of mitochondria such as Tfam (also known as mitochondrial transcription factor A) and the RNase mitochondrial RNA processing (7). Binding sites for NRF1 have been also identified in several genes involved in cell proliferation and intermediary metabolism (12). Thus, NRF1 likely plays an important role in the control of mammalian mitochondrial biogenesis under a variety of conditions requiring high energy supply.
Although mtDNA replication is required for eukaryotic cell proliferation, the mechanisms that regulate this process are poorly understood. Recent studies have shown that the cy-toskeleton plays an active role in the mitochondrial dynamics of the cell (for a review, see Ref. 15) and that the replication of the mitochondrial genome occurs in the perinuclear space (16).
The key enzyme in mtDNA replication is DNA polymerase ␥ (pol ␥), which constitutes only ϳ1% of the total cellular DNA polymerase activity (17,18). The pol ␥ holoenzyme has been characterized extensively in Drosophila (19); it is a heterodimer of a 125-kDa ␣ subunit containing DNA polymerase and 3Ј-5Ј exonuclease catalytic activities (20) and a 35-kDa accessory subunit that increases the catalytic efficiency of the holoenzyme and is likely involved in primer recognition and in enhancing processivity (21,22). Both subunits have also been identified in other systems, and their structures are evolutionarily conserved. The catalytic subunit shares homology with the family A DNA polymerases (20,23,24), whereas the accessory subunit is related to aminoacyl-tRNA synthetases (22,25). In Drosophila, the nuclear genes and the corresponding cDNAs have been cloned (20,26,27) and map in the same genomic region (Adh region, chromosome II) within a tight cluster of five genes (28). Here we report on the transcriptional regulation of the pol ␥-␣ and pol ␥-␤ genes. Whereas our data indicate that the catalytic subunit of pol ␥ is expressed constitutively at a low level, the developmental pattern of expression of the pol ␥-␤ gene suggests that the accessory subunit plays an important role in the control of mtDNA replication. Moreover, the transcription factor DREF, which activates the expression of genes encoding components of the nuclear replication machinery (29), is critical for pol ␥-␤ promoter activity. This regulatory protein is also critical for the activity of the mitochondrial singlestranded DNA-binding protein (mtSSB) promoter (30), establishing firmly a molecular link between nuclear and mitochondrial DNA replication.

RNA Extraction and Northern Analysis
Total RNAs from staged embryos; first-, second-, and third-instar larvae; early and late pupae; and adults of Drosophila melanogaster Oregon R were extracted using the Trizol kit (Life Technologies, Inc.) according to the manufacturer's directions. For Northern analysis, 30 g of total RNA was electrophoresed on 1.8 M formaldehyde and 1.2% agarose gels, blotted on a Zeta Probe membrane (Bio-Rad), and probed in ZAP buffer (0.25 M phosphate buffer, pH 7.2., 7% SDS) at 65°C using various [␣-32 P]dCTP-labeled cDNA clones as probes. Filters were washed in 0.1% SDS and 0.1ϫ SSC at 65°C and autoradiographed with intensifying screens at Ϫ70°C.

Nuclear Run-on Transcription Analysis
To obtain nuclei from Drosophila adults, flies (100 in a typical experiment) were homogenized in 1 ml of buffer containing 10 mM KCl, 15 mM HEPES (pH 7.6), 5 mM MgCl 2 , 0.35 M sucrose, 0.5 mM EGTA, 0.1 mM EDTA, 1 mM dithiothreitol, and 10 mM phenylmethylsulfonyl fluoride. The homogenate was filtered through Miracloth (Calbiochem) and centrifuged at 400 ϫ g for 5 min. The supernatant was centrifuged at 700 ϫ g for 10 min, and the nuclear pellet was resuspended in buffer containing 50 mM Tris-HCl (pH 8.3), 5 mM MgCl 2 , 0.1 mM EDTA, and 40% glycerol. Embryos were dechorionated by sodium hypochloride treatment and washed extensively before homogenization. Transcriptional run-on reactions were performed essentially as described by Linial et al. (31) using 150 Ci of [␣-32 P]UTP and ϳ10 7 nuclei. Radiolabeled RNA was purified by phenol/chloroform extraction followed by precipitation with isopropyl alcohol and hybridized in ZAP buffer at 65°C to 5 g of 18 S rDNA and pol ␥-␣ and pol ␥-␤ cDNAs that were immobilized on nylon filters using the Manifold II system (Schleicher & Schü ll) and denatured by NaOH treatment. Filters were washed in 0.1% SDS and 02 ϫ SSC at 65°C and autoradiographed with intensifying screens at Ϫ70°C.

Promoter Constructs
DNA fragments containing the promoter regions (P1-P3) of the five genes were amplified by PCR using several sets of oligonucleotide primers ( Table I). The resulting parental DNA fragments were used as substrates for generation of deletion constructs. After digestion with restriction endonucleases, DNA fragments were excised from agarose gels and inserted into the pxp2 vector, which contains the luciferase gene as reporter. The nucleotide sequences of the parental DNA fragments and each of the promoter constructs were confirmed by DNA sequence analysis.
pol ␥-␣ Promoter (P3)-Deletions of the parental DNA fragment were generated as follows. EcoRV/XhoI and XmnI/XhoI fragments were cloned into the pxp2 vector that was digested with SmaI/XhoI to generate constructs ␣Ϫ1710/ϩ113 and ␣Ϫ1197/ϩ113 (where ϩ1 corresponds to the transcriptional start site). The XmnI/XhoI fragment was also digested with DraI, and the XmnI/DraI fragment was cloned into pxp2 that was digested with SmaI to generate construct ␣Ϫ1197/Ϫ183. To obtain construct ␣Ϫ407/ϩ113, a 520-bp BglII/XhoI fragment was cloned into pxp2 that was digested with BglII/XhoI. To mutate the putative NRF1-like binding site in P3, we used the oligonucleotide 5Ј-GCCAGCCTAGGTATTTTTGccCcCTccCcCGGCGG-3Ј (from positions ϩ42 to ϩ8 in the template strand for transcription); the lowercase letters represent the nucleotides that were changed from the original sequence (5Ј-GCCAGCCTAGGTATTTTTGTGCGCTTGCGCGGCGG-3Ј), and those that are underlined represent the mutated NRF1 site. After several steps of subcloning, PCR, and additional subcloning, construct ␣Ϫ407/ϩ113mut was obtained. This plasmid is identical to its parental construct, ␣Ϫ407/ϩ113, except for the six nucleotide changes within the NRF1-like site.
rpII33/pol ␥-␤ Promoter (P1)-Deletions in the parental PCR fragment were generated as follows. HindIII/XhoI and PvuII/XhoI fragments were cloned into the pxp2 vector that was digested with HindIII/ XhoI and SmaI/XhoI, respectively, to generate constructs ␤Ϫ1806/ϩ467 and ␤Ϫ1319/ϩ467. To obtain constructs ␤Ϫ569/ϩ467 and ␤Ϫ212/ϩ467, the parental DNA fragment was digested with PstI and then with ClaI, rendered blunt-ended with T4 DNA polymerase, and cut with XhoI. Purified fragments were then ligated into the pxp2 vector that was digested with SmaI/XhoI. A 797-bp DNA fragment from positions Ϫ326 to ϩ473 was also amplified by PCR using two oligonucleotides as primers: the same reverse primer described above (Table I) and a new forward primer (5Ј-GCACTGCAGTTTTGTTTTGATGAA-3Ј) containing a PstI site at position Ϫ318. The amplified product was digested with PstI and XhoI and cloned into the pBluescript vector that was digested with the same restriction enzymes, generating pBS-PI. The construct was then digested with ClaI; end-filled with Klenow DNA polymerase; and religated to obtain pBS-PImut, in which the DREF-binding site is changed from TATCGATT to TATCgcGATT. The BamHI/XhoI fragments of pBS-PI and pBS-PImut were cloned into pxp2 that was digested with BamHI/XhoI to generate the ␤Ϫ318/ϩ467 and ␤Ϫ318/ ϩ467mut constructs, respectively.
The BamHI/XhoI fragments of pBS-PI and pBS-PImut were also cloned into pxp2 that was digested with BamHI/XhoI to generate constructs rpIIϪ715/ϩ71 and rpIIϪ715/ϩ71mut, with the inverse orientation of the DNA fragments. These constructs are numbered with respect to the transcriptional start site (ϩ1) of the rpII33 gene.
orc5/sop2 Promoter (P2)-Deletions of P2 were generated by digestion of the parental PCR fragment as follows. The HindIII/XhoI and SalI/XhoI fragments were cloned into pxp2 that was digested with the same enzymes to generate constructs sopϪ1616/ϩ307 and sopϪ604/ ϩ307. (ϩ1 represents the transcriptional start site in the sop2 gene that maps at a position 326 nucleotides upstream of the translational start

AACGTGCGTAACAGCACTGG TCCTCGAGTTAATAAAGGAAACAT
codon. This ϩ1 position is shared in the complementary strand by the orc5 gene.) A 627-bp DNA fragment from positions Ϫ326 to ϩ473 was amplified by PCR using two oligonucleotides as primers: the same reverse primer described above (Table I) and a forward primer (5Ј-CCAGAGCTCTCGTCTGCTTT-3Ј) containing a PvuII site at position Ϫ306. The PvuII/XhoI fragment was cloned into the pBluescript vector that was digested with EcoRV/XhoI to obtain construct pBS-PII. The construct was then digested with ClaI; end-filled with Klenow DNA polymerase; and religated to obtain construct pBS-PIImut, in which the region containing the two DREF sites is modified by a small deletion, changing the sequence TATCGATataacgaaaTAACGATAgtattgagggccatcgatt, where the ClaI sites are underlined and the DREF sites are in uppercase letters. The BamHI/XhoI fragments of pBS-PII and pBS-PIImut were cloned into pxp2 that was digested with BamHI/XhoI to obtain constructs sopϪ303/ϩ307 and sopϪ303/ϩ307mut. A 2230-bp DNA fragment from positions Ϫ2080 to ϩ351 (where ϩ1 corresponds to the transcriptional start site of the orc5 gene) was amplified by PCR using the two oligonucleotide primers 5Ј-CCAGAGCTCTCGTCTGCTT-T-3Ј (forward) and 5Ј-GGCCCTCGAGGGTGGCTAAATGCG-3Ј (reverse), in which a PvuII site was introduced at a position corresponding to ϩ343. The SalI/PvuII fragment was cloned into pxp2 that was digested with SalI/SmaI to generate construct orcϪ1405/ϩ346. The SalI/ KpnI fragment was cloned into pxp2 that was digested with SalI/KpnI to generate construct orcϪ1405/ϩ117. Finally, the BamHI/EcoRV fragments of pBS-PII and pBS-PIImut were cloned into pxp2 that was digested with SmaII/BglII to obtain the orcϪ170/ϩ346 and orcϪ170/ ϩ346mut constructs, respectively. DREF and Erect Wing Constructs-For cotransfection experiments, we cloned the complete Erect Wing (EWG) and DREF cDNAs in the vector pMK26 (32), which allows strong expression under the control of the actin 5C promoter. A BamHI/XhoI fragment containing the complete DREF cDNA (30) was end-filled with Klenow DNA polymerase and subcloned into the pBluescript vector that was digested with EcoRV to obtain pBS-DREF. For EWG plasmid construction, the pET3dEWG6His clone (a kind gift of Dr. Richard Scarpulla) was first digested with XbaI/BamHI, end-filled with Klenow DNA polymerase, and then inserted into the pBluescript vector that was digested with EcoRV to obtain pBS-EWG. Finally, the HindIII/PstI fragments obtained from pBS-DREF and pBS-EWG were cloned into the pMK26 vector that was digested with HindIII/PstI, generating constructs pMK-DREF and pMK-EWG, respectively.

Cell Transfection and Reporter Activity
Transfection in Schneider's S2 cells was carried out according to Soeller et al. (33) with some modifications. Streptomycin (100 g/ml) and penicillin (100 IU/ml) were added to the medium, and cells were transfected with 10 g of the vector pSV␤gal (Promega) and 5 g of the corresponding pxp construct. After transfection, cells were incubated for 24 h at 25°C, washed twice with phosphate-buffered saline, resuspended in 5 ml of fresh medium, and incubated for 60 -80 h at 25°C. To prepare extracts, cells were harvested by centrifugation; washed with phosphate-buffered saline and then with fresh buffer containing 40 mM Tris-HCl (pH 7.5), 1 mM EDTA, and 150 mM NaCl; resuspended in 100 l of 0.1 M Tris-HCl (pH 7.5); and subjected to five cycles of freezing for 60 s at Ϫ70°C and heating for 60 s at 37°C. Cell debris was removed by centrifugation at 13,000 ϫ g for 5 min. Promoter activities were calculated by normalizing luciferase activity (pxp constructs) to ␤-galactosidase activity (pSV␤gal). Luciferase activity was determined using the luciferase assay system (Promega) according to the manufacturer's recommendations, and ␤-galactosidase activity was measured as described previously (34).

Electrophoretic Mobility Shift Assays
Electrophoretic mobility shift assays (EMSAs) were carried out using protein prepared from two sources: the in vitro TNT transcription/ translation system (Promega) for DREF binding assays and Schneider's cell nuclear extracts for DREF and NRF1 binding assays. The binding reactions were carried out using 2 l of in vitro transcription and translation product or 5 g of nuclear protein and 50,000 cpm of the DNA probe in binding buffer containing 20% glycerol, 1 mM dithiothreitol, 20 mM HEPES (pH 9.0), 5 mM MgCl 2 , 0.2 mM EDTA, and 200 mM KCl. After incubation for 30 min at 4°C, reaction products were electrophoresed on 4.5% polyacrylamide gels (for PCR probes) or 6% polyacrylamide gels (for oligonucleotide probes) containing 0.5ϫ Tris borate/EDTA. DNA probes were labeled using [␥-32 P]ATP and T4 polynucleotide kinase. For the P1 probes, two DNA fragments containing the DRE were used as probes: a double-stranded 30-mer oligonucleotide from positions Ϫ225 to Ϫ204 and a 173-bp fragment from positions Ϫ325 to Ϫ153 that was obtained by PCR amplification (where ϩ1 corresponds to the transcriptional start site for pol ␥-␤). For the P2 probe, a 120-bp DNA fragment from positions Ϫ132 to Ϫ13 containing the DREs was amplified by PCR (where ϩ1 corresponds to the transcriptional start site for sop2). In DREF EMSA experiments, a 100-fold molar excess of various oligonucleotides was used as competitor, including unlabeled DNA probes and oligonucleotides from the mtSSB promoter containing wild-type or mutated DRE sites (30). For the P3 probe, a double-stranded 35-mer oligonucleotide from positions ϩ42 to ϩ8 containing the NRF1-like site and its mutated version (described above) was used. DREF antiserum was prepared as described (30). Antiserum against EWG was kindly provided by Dr. Kalpana White. In supershift assays, 2 l of a 1:100 dilution of the relevant antiserum was used.

RESULTS
Differential Expression of the Genes Encoding the pol ␥ Catalytic and Accessory Subunits during D. melanogaster Development-The genes encoding the two subunits of the mitochondrial DNA polymerase map within a cluster of genes, the pol ␥ cluster, that extends over a region of 10.5 kilobases in the alcohol dehydrogenase region (34D-E) of the second chromosome of D. melanogaster (27,28). The cluster contains five genes that encode proteins involved in essential cellular processes, including the two pol ␥ subunits (pol ␥-␣ and pol ␥-␤), one subunit of RNA polymerase II (RpII33), one subunit of the origin recognition complex (ORC5), and one subunit of the actin-related protein 2/3 complex (SOP2/ARC41). The pol ␥ cluster is extremely compact: the genes are tightly packed with very little intergenic space, with some overlap in their 5Ј-and 3Ј-ends (Fig. 1). The cluster contains three promoter regions (P1-P3), and two of them direct bidirectional transcription (28). This tight organization suggests that the structure of the pol ␥ cluster plays an important role in regulating the expression of its genes. This is a particularly important consideration given that two of the genes encode the subunits of the key enzyme in the mitochondrial DNA replication process (pol ␥-␣ and pol ␥-␤), representing a rare example in eukaryotes of genes encoding functionally related proteins that are linked in the genome. Moreover, a third gene of the pol ␥ cluster, orc5, encodes an essential component of the nuclear DNA replication machinery, suggesting the presence of potential common regulatory mechanisms for nuclear and mitochondrial DNA replication.
As a first step to study the factors involved in regulating the expression of the pol ␥ genes, we determined (by Northern analysis) their patterns of expression during Drosophila development and compared them with the expression profiles of the other genes encoded in the pol ␥ cluster. The level of pol ␥-␣ mRNA was relatively high in eggs and decreased rapidly during embryonic development ( Fig. 2A). This result is perhaps surprising because mtDNA replication started ϳ10 h after egg laying (AEL); and at this developmental stage, the level of pol ␥-␣ mRNA has declined substantially. During the larval stages, a period characterized by increases in body weight and cell ploidy, the steady-state level of pol ␥-␣ mRNA remained very low, whereas that in adults was comparatively higher. The spatiotemporal pattern of expression of pol ␥-␣ mRNA, examined by whole-mount RNA in situ hybridization, is in good agreement with the Northern results. pol ␥-␣ mRNA was distributed homogeneously in early blastoderm, and the intensity of the signal decreased through gastrulation and was very faint or undetectable in older embryos (data not shown).
pol ␥-␤ mRNA showed a different developmental profile ( Fig.  2A). The level present in eggs was relatively low, but its steadystate level increased in early embryonic stages, reaching its maximum between 6 and 12 h AEL. In late embryos, its steadystate level decreased; and in first-instar larvae, there was a moderate increase. These results suggest that the expression of the pol ␥-␤ subunit is regulated precisely during Drosophila development. Moreover, the steady-state level of pol ␥-␤ reached its maximum just before the start of mtDNA synthesis that occurred 10 -12 h AEL.
To determine whether the differences observed in the pattern of expression of pol ␥-␣ and pol ␥-␤ mRNAs are due to different rates of transcription, we performed a series of run-on transcription experiments using nuclei extracted from 8 -10-h AEL embryos and adults. A higher rate of mRNA synthesis was detected in the pol ␥-␤ gene in both embryos and adults (Fig. 3). Although we have not evaluated possible differences in mRNA stability, the different rates of mRNA synthesis correspond well with the mRNA steady-state levels detected by Northern analysis (Fig. 2A), documenting a differential regulation of the pol ␥-␣ and pol ␥-␤ genes at the transcriptional level.
Unfortunately, we were unable to detect pol ␥-␤ mRNA by in situ hybridization despite systematic attempts using a variety of experimental conditions (data not shown). A lack of sensitivity of the technique is an unlikely explanation because Northern analysis showed that the steady-state level of pol ␥-␤ mRNA was similar or higher than that of pol ␥-␣ mRNA. We have also used in parallel several control probes that detect other unrelated mRNAs, obtaining in each case the expected pattern. Although the reason for the lack of in situ detection of pol ␥-␤ mRNA is presently unknown, our data suggest the possibility of some type of sequestration of pol ␥-␤ mRNA that does not occur with pol ␥-␣ mRNA.
The rpII33 gene is divergently transcribed from the same promoter as the pol ␥-␤ gene, yet its developmental pattern of expression is entirely different (Fig. 2B). rpII33 mRNA was accumulated in eggs, and its level was maintained fairly con-FIG. 2. Developmental pattern of expression of the five genes encoded in the Drosophila pol ␥ cluster. Total RNAs from 0 -18-h embryos, larvae (first-, second-, and third-instar (L1-L3, respectively)), pupae (P1-P3), and adults (A) of D. melanogaster were analyzed by Northern blotting using the corresponding cDNAs as probes. RNAs were fractionated on formaldehyde-containing 1.2% agarose gels, transferred to nylon membrane, and hybridized with the corresponding radiolabeled probes, and the blots were autoradiographed. Levels of RNA loaded were evaluated by ethidium bromide (EthBr) staining. The expression patterns of the pol ␥ genes are shown in A, and those of rpII33, sop2, and orc5 are shown in B. stant during early embryogenesis up to 6 h AEL. From 9 h AEL onward, its steady-state level decreased and was very low during the larval and pupal periods. Adults also contained a relatively low amount of the rpII33 transcript. A similar result was observed in the case of the orc5/sop2 genes, and the pattern of expression was also different for each gene (Fig. 2B). sop2 mRNA was present at relatively high levels in all developmental stages, whereas the steady-state level of orc5 mRNA was very high in eggs and decreased sharply as development proceeded. In larvae and pupae, its concentration was low and increased again in adults. This pattern is generally similar to that for pol ␥-␣. Our results indicate clearly that the pairs of genes transcribed from the bidirectional promoters P1 and P2 are differentially regulated during development.
Analysis of the rpII33/pol ␥-␤ Promoter Region (P1)-The developmental pattern of expression of the pol ␥-␤ gene suggests that the accessory subunit of the pol ␥ enzyme may play an important role in regulating mtDNA replication. The pol ␥-␤ gene is divergently transcribed from the rpII33 gene, and their 5Ј-ends are separated by only 245 bp (Fig. 1). To identify and characterize transcriptional regulatory elements in the rpII33/ pol ␥-␤ intergenic promoter region (P1), we cloned defined DNA fragments in the pxp2 vector and determined their promoter activities in transient transfection experiments in Schneider's S2 cells, measuring luciferase activity in cell extracts (see "Experimental Procedures").
A construct containing the 5Ј-UTR and 1.8 kilobases of the proximal 5Ј-upstream region of the pol ␥-␤ gene (␤Ϫ1806/ϩ467, where ϩ1 corresponds to the pol ␥-␤ transcriptional start site), which includes the complete adjacent rpII33 gene, directed a substantial level of luciferase activity in Schneider's cells, Ͼ1000-fold as compared with the pxp2 vector (Fig. 4). A similar level of activity was maintained in shorter constructs, including ␤Ϫ318/ϩ467, which contains the rpII33/pol ␥-␤ intergenic region and the 5Ј-UTRs of both genes. Computer analysis of the DNA sequence spanning positions Ϫ318 to ϩ467 revealed the presence of a DRE motif (TATCGATT) located at position Ϫ216 within the rpII33/pol ␥-␤ intergenic region, close (ϳ30 bp) to the transcriptional initiation site of the rpII33 gene. The DRE is recognized by DREF, a transcription factor critical in the regulated expression of several genes encoding proteins involved in nuclear DNA replication (29). We have also demonstrated recently that it is essential for the activity of the D. melanogaster mtSSB promoter (30). Mutation of the DRE-binding site reduced the activity of construct ␤Ϫ318/ϩ467mut by 80% (p Ͻ 0.05), suggesting strongly that DREF is critical to The DRE sequence is represented as an ellipse. DNA fragments cloned upstream of the luciferase reporter gene and used in the analysis of the pol ␥-␤ promoter activity are shown above the schematic, and those used in the analysis of the rpII33 promoter activity are shown below. Plasmid designations indicate the specific DNA sequences present in the constructs, with numbers relative to the position of the transcriptional start site. The internal X in the DRE sequence indicates that the site is mutated. The relative promoter activities of the constructs, as measured in the luciferase assay, are indicated to the right. The luciferase activity of the pxp2 vector was defined as 1.0 and was used as the standard for comparison. Luciferase activity was normalized to ␤-galactosidase activity for each construct. Values are means Ϯ S.D. of at least three independent experiments. Statistical analysis was carried out using the Wilcoxon signed-rank test. * indicates p Ͻ 0.05. achieve full pol ␥-␤ promoter activity. Furthermore, construct ␤Ϫ212/ϩ417, with the DRE site eliminated, abolished promoter activity in Schneider's cells.
To determine whether the DRE sequence is also important for the activity of the rpII33 promoter, we determined the luciferase activity of the ␤Ϫ318/ϩ467 fragment cloned in the pxp2 vector in the opposite orientation (rpIIϪ715/ϩ71). An 800-fold induction in luciferase activity was detected, indicating that the rpII33/pol ␥-␤ intergenic region divergently directs the transcription of the two genes. Importantly, when the assay was carried out with a construct containing the same DNA fragment with the DRE site mutated (rpIIϪ715/ϩ71mut), no significant change in luciferase activity was observed. This result demonstrates that the DRE site located in the P1 region is essential only for the promoter activity of the pol ␥-␤ gene.
Analysis of the orc5/sop2 Promoter Region (P2)-Two of the genes in the pol ␥ cluster, orc5 and sop2, overlap in their 5Ј-ends (Fig. 1). We found that a DNA fragment encompassing the 5Ј-upstream region of the two genes contains potent bidirectional promoter activity. Construct sopϪ1616/ϩ307, which contains the sop2 5Ј-UTR and most of the orc5 gene, directed a high level of luciferase activity, Ͼ9000-fold as compared with the pxp2 vector (Fig. 5). A high level of activity was maintained in the sopϪ303/ϩ307 construct, which contains the complete 5Ј-UTRs of the two overlapping genes. The 5Ј-UTR of the orc5 gene contains two DRE sites. To examine the functional rele-vance of the DRE sites in sop2 promoter activity, we mutated both sites in construct sopϪ303/ϩ307mut. We observed no change in the level of luciferase activity, indicating that the DRE sites do not play an essential role in the sop2 promoter activity in Schneider's cells.
To evaluate the orc5 promoter, we assayed the activity of the orcϪ170/ϩ346 construct and observed a high increase in luciferase activity, Ͼ8000-fold as compared with the pxp2 vector. The same region containing mutated DRE sites (orcϪ170/ ϩ346mut) directed significantly reduced activity, only 30% of that detected in the orcϪ170/ϩ346 construct (p Ͻ 0.05). Thus, similar to the results obtained in the bidirectional rpII33/pol ␥-␤ promoter region, the DRE sites located in the 5Ј-UTR of the orc5 gene are essential for the activity of one promoter (orc5), but have no effect on the activity of the other (sop2). More importantly, these data establish a common molecular mechanism involved in the regulated expression of two genes in the pol ␥ complex that are involved in nuclear and mitochondrial DNA replication, orc5 and pol ␥-␤, respectively.
DREF Binds to the pol ␥-␤ and orc5 Promoter Regions-To characterize further the DRE sequences in P1 and P2, we carried out electrophoretic mobility shift assays using probes covering the P1 and P2 regions that contain the DRE sites. We used either DREF produced by in vitro transcription/translation or nuclear extracts prepared from cultured Schneider's cells as the protein source. When a radiolabeled 173-bp frag- FIG. 5. Functional analysis of the P2 region. A schematic representation of the divergent P2 region and the adjacent orc5 and sop2 genes is shown in the middle. DNA fragments cloned upstream of the luciferase reporter gene and used in the analysis of the sop2 promoter activity are shown above the schematic, and those used in the analysis of the orc5 promoter activity are shown below. The relative promoter activities of the constructs, as measured in the luciferase assay, are indicated to the right. The luciferase activity of the pxp2 vector was defined as 1.0 and was used as the standard for comparison. Luciferase activity was normalized to ␤-galactosidase activity for each construct. Values are means Ϯ S.D. of at least three independent experiments. Statistical analysis was carried out using the Wilcoxon signed-rank test. * indicates p Ͻ 0.05. ment containing the single DRE site present in the P1 region (P1-DRE) was incubated with recombinant DREF protein, a single retarded protein-DNA complex was detected, which was eliminated when a 100-fold molar excess of unlabeled probe was included in the binding reaction (Fig. 6A). Inclusion of rabbit anti-DREF serum produced a clear supershift, demonstrating that DREF interacts with the DRE site in the pol ␥-␤ promoter region. The supershift was also eliminated by inclusion of a 100-fold molar excess of unlabeled probe. When a similar experiment was carried out in which a 120-bp fragment containing the two DRE sites present in the P2 region (P2-DRE) was used as probe, similar results were obtained (Fig.  6B). In this case, two retarded protein-DNA complexes were apparent, and both were supershifted by the anti-DREF antiserum. As a control, we included preimmune serum in the binding reaction, and no supershift was produced. The same EMSA patterns were observed with both promoter fragments when the source of DREF was a nuclear extract prepared from Schneider's cells (Fig. 6, A and B).
Additional EMSA experiments were carried out with Schneider's cell nuclear extracts and functional DRE sequences that were characterized previously in the Drosophila mtSSB promoter as competitors (30). Here the probe for P1 was a doublestranded 30-mer oligonucleotide spanning positions Ϫ225 to Ϫ204 (Fig. 7A), and that for P2 was as described above (Fig.  7B). Both probes gave rise to shifted bands, and inclusion of anti-DREF antiserum in the binding reaction produced supershifts not observed upon inclusion of preimmune serum. When competition of DREF binding to P1 and the corresponding anti-DREF antiserum-mediated supershift was carried out using a 100-fold molar excess of unlabeled P1, P2, or ssb (a double-stranded oligonucleotide containing the two functional DRE sites in the mtSSB promoter (30)) DNA, the retarded bands and supershifts were abolished. However, if the competition was performed with ssb Ϫ/Ϫ DNA (a double-stranded oligonucleotide containing mutated DRE sites from the ssb promoter), neither the retarded bands nor the supershifts were eliminated, providing further support for the binding of DREF to P1. Similar results were obtained with DREF binding to P2. A 100-fold molar excess of P2 or ssb DNA eliminated the retarded bands and supershifts, but competition with ssb Ϫ/Ϫ DNA had no effect. Notably, however, only partial competition of DREF binding to the P2 region was observed with P1 DNA, likely due to the fact that P1 contains a single DRE site, whereas P2 contains two. Taken together, the EMSA experiments demonstrate that the DRE sites present in the pol ␥-␤ and orc5 promoters are bound specifically by DREF.
Analysis of the pol ␥-␣ Promoter Region (P3)-A construct (␣Ϫ1710/ϩ113) containing the complete coding sequence of the sop2 gene, the ϳ140-bp sop2/pol ␥-␣ intergenic region, and 119 bp of the 5Ј-UTR of the pol ␥-␣ gene directed 300-fold higher luciferase activity than the pxp2 vector (Fig. 8). Constructs containing shorter DNA fragments maintained a similar level of activity, including construct ␣Ϫ407/ϩ113, which contains only the sop2/pol ␥-␣ intergenic region and the pol ␥-␣ 5Ј-UTR. Luciferase activity was abolished in construct ␣Ϫ1197-183, which has the transcriptional start site eliminated. These data indicate that the proximal 5Ј-upstream region of the pol ␥-␣ gene has weak but significant promoter activity in Schneider's cells.
The proximal 5Ј-upstream region of the pol ␥-␣ gene does not contain DNA sequence motifs with significant homology to the DRE site. However, its 5Ј-UTR contains a proximal binding motif potentially recognized by the transcription factor NRF1. To evaluate its functional relevance, we mutated the NRF1-like site. We observed a moderate but significant decrease (50%; p Ͻ 0.01) in the promoter activity of construct ␣Ϫ407/ϩ113mut compared with the native construct ␣Ϫ407/ϩ113, suggesting that the NRF1-like site might be functional in Schneider's cells.
NRF1 likely plays a key role in nucleo-mitochondrial communications in mammals (12). Interestingly, its DNA-binding motif is conserved in two regulatory proteins identified in Drosophila (Erect Wing) and sea urchin (P3A2). To study the binding of putative regulatory proteins to the NRF1-like site, we pursued EMSA experiments using, as probes, a doublestranded 35-mer oligonucleotide spanning positions ϩ42 to ϩ8 from the P3 region that contains the NRF1-like site (NRF1 ϩ ) and a similar oligonucleotide containing a mutated NRF1-like site (NRF1 Ϫ ). When Schneider's cell nuclear extracts were incubated with the NRF1 ϩ probe, several retarded bands were observed that were eliminated specifically by a 250-fold molar excess of unlabeled NRF1 ϩ oligonucleotide as competitor (Fig.  9, left panel). Inclusion of anti-EWG antiserum in the binding reaction did not produce a supershift, suggesting that the EWG protein is not present in the retarded complexes. Competition with a 250-fold molar excess of the NRF1 Ϫ oligonucleotide eliminated the retarded fragments, indicating that the protein bound to the probe likely interacts with sequence elements other than the NRF1-like site. Furthermore, when the NRF1 Ϫ oligonucleotide was used as probe, several retarded bands were clearly visible, including an extra band not observed in the EMSAs with NRF1 ϩ (Fig. 9, right panel). Whereas an excess of the NRF1 Ϫ probe competed completely, only the shared retarded complexes were eliminated by a 250-fold excess of the NRF1 ϩ oligonucleotide. These results suggest that there is protein binding to NRF1 Ϫ that is mediated specifically through the mutant sequence, possibly explaining the inhibition of promoter activity detected in transient transfection experiments with the mutant NRF1-like construct (Fig. 8). This protein is not EWG because no supershift was observed in the EMSAs in the presence of anti-EWG antiserum. Overall, the data suggest that neither the EWG protein nor the native NRF1-like site per se is likely involved in the control of pol ␥-␣ promoter activity. Consistent with this interpretation, overexpression of EWG in Schneider's cells had no effect on the activity of the pol ␥-␣ constructs examined in transient transfection assays (data not shown). DISCUSSION Eggs are highly enriched in mitochondria that are deposited by the mother during oogenesis. As development proceeds, important changes in the morphological and functional maturation of mitochondria occur (3) in a process that requires a careful orchestration of nucleo-mitochondrial interactions. Evidence from studies in different systems including mammals shows that the mitochondria stored in the egg are sufficient for the first stages of embryonic development (35). Later on, mitochondria are distributed throughout the various embryonic territories, which exhibit specific patterns of organelle differentiation (36). In Drosophila, the developmental pattern of expression of several genes encoding mitochondrial proteins, including the nuclearly encoded ␣and ␤-H ϩ -ATP synthase subunits and the mitochondrially encoded ATPase-6 and -8 subunits, has been analyzed (37,38). In each of these cases, gene expression is activated coordinately 6 -12 h AEL, suggesting that during this period, the maternal contribution becomes limiting, and there is an activation of mitochondrial biogenesis in the embryo. Accordingly, it has been shown that mtDNA replication and likely mitochondrial proliferation start around this time of development (39).
In this work, we have evaluated the expression of the genes encoded in the pol ␥ cluster during Drosophila development and, in particular, those encoding the two subunits of the pol ␥ holoenzyme. Both pol ␥-␣ and pol ␥-␤ mRNAs are stored in the egg, probably as a result of the high level of mtDNA replication occurring during oogenesis. The steady-state level of pol ␥-␣ mRNA declines rapidly during development, reaching a minimum level 10 h AEL and onward. In contrast, the pol ␥-␤ mRNA level increases during the first hours of development and reaches its maximum level 6 -10 h AEL. Nuclear run-on transcription analyses indicate that the expression of the genes encoding the two pol ␥ subunits is regulated differentially at the transcriptional level during Drosophila development. Notably, only the pol ␥-␤ mRNA level is high in the period of active mtDNA replication. This might seem surprising because the pol ␥-␣ subunit contains the catalytic activities of the enzyme. However, the lack of correlation between the pol ␥-␣ mRNA level and mtDNA replication during Drosophila development is consistent with several results obtained previously in mammals. First, the pol ␥-␣ mRNA steady-state levels in different cell types do not correlate with their mtDNA content (40). Second, cells without mtDNA ( 0 ) contain similar levels of pol ␥-␣ mRNA and protein compared with wild-type cells (41), rendering it unlikely that pol ␥-␣ serves a key role in the control of mtDNA replication. Although the steady-state level FIG. 8. Functional analysis of the P3 region. A schematic representation of the P3 region, the adjacent sop2 gene, and the 5Ј-region of the pol ␥-␣ gene is shown. The NRF1-like sequence element is shown as an ellipse. DNA fragments cloned upstream of the luciferase reporter gene and used in the analysis of the pol ␥-␣ promoter activity are shown below. The relative promoter activities of the constructs, as measured in the luciferase assay, are indicated to the right. The luciferase activity of the pxp2 vector was defined as 1.0 and was used as the standard for comparison. Luciferase activity was normalized to ␤-galactosidase activity for each construct. Values are means Ϯ S.D. of at least three independent experiments. Statistical analysis was carried out using the Wilcoxon signed-rank test. * indicates p Ͻ 0.01.
FIG. 9. Regulatory protein binding in the P3 region. Protein binding to the P3 region was examined by EMSAs. The probes were a double-stranded 35-mer oligonucleotide containing the NRF1-like site (NRF1 ϩ ) and a derivative in which the NRF1-like site was mutated (NRF1 Ϫ ). Arrowheads indicate the positions of shared retarded bands, and the arrow indicates the position of the retarded band produced only with the NRF1 Ϫ probe. The competitors used were a 250-fold molar excess of the unlabeled NRF1 Ϫ (1) and NRF1 ϩ (2) probes as indicated. Ab, antibody. of pol ␥-␤ mRNA has not been determined in mammals, its developmental profile in Drosophila suggests that the pol ␥-␤ subunit participates in the reactivation of mtDNA synthesis during development. Interestingly, the pol ␥ accessory subunit shares conserved protein domains with aminoacyl-tRNA synthetases (22,25) and has been suggested to participate in primer recognition (22) and demonstrated to increase the template-primer binding affinity, catalytic efficiency, and processivity of the holoenzyme (21,25,(42)(43)(44). Together, the transcriptional regulation shown here and the biochemical data regarding accessory subunit function in several animal systems including Drosophila support the conclusion that the accessory subunit of pol ␥ plays a key role in the control of mtDNA replication.
We have analyzed the promoter region of the pol ␥-␤ gene and shown that its expression is controlled by DREF, a regulatory protein essential for the transcriptional regulation of a set of genes encoding proteins involved in nuclear DNA replication and cell cycle control in Drosophila. These include proliferating cell nuclear antigen (which is an auxiliary protein for nuclear replicative DNA polymerase ␦), the 180-and 73-kDa subunits of nuclear initiator DNA polymerase ␣, cyclin A, E2F, and the proto-oncogene draf (45)(46)(47)(48)(49). Transient transfection assays and EMSA experiments demonstrate that the DRE site present in the pol ␥-␤ promoter is functional in Schneider's cells and that DREF binds to this site. Furthermore, DREF is expressed throughout Drosophila development, reaching its maximum level 2-8 h AEL (29), suggesting that it may contribute significantly to the expression of the pol ␥-␤ gene in vivo. During the cell cycle, nuclear DNA synthesis is extremely well controlled through complex regulatory mechanisms that have been characterized extensively during the last few years (50). In contrast, although mtDNA content, maintenance, and partitioning are also strictly controlled during cell proliferation, very little is known about the molecular mechanisms involved in the regulation of these processes. With regard to the control of nuclear and mitochondrial DNA replication, it is notable that mutations in the RPN11/MPR1 gene of Saccharomyces cerevisiae, which encodes a subunit of the proteasome, produce a cell cycle arrest phenotype with over-replication of the nuclear and mitochondrial DNA genomes coupled with severe alterations in mitochondrial morphology (51). The results presented in this report, in addition to our recent finding that the expression of the mtSSB gene is also controlled by the DRE/DREF system (30), establish in molecular terms a link between mitochondrial and nuclear DNA replication.
Interestingly, the pol ␥-␤ gene is transcribed from a promoter region that also directs the divergent transcription of the rpII33 gene, yet the DRE site is important only for the expression of the pol ␥-␤ gene. Remarkably, a second promoter region in the pol ␥ complex that directs the divergent transcription of two genes that overlap in their 5Ј-ends, orc5 and sop2, contains two copies of the DRE motif. When both DRE sites are mutated, there is a substantial reduction in orc5 promoter activity, consistent with the essential role of the origin recognition complex in nuclear DNA replication. In contrast, the same mutation has no effect on sop2 promoter activity. Although we do not know at present the physiological significance of the juxtaposition of the pol ␥-␤ and orc5 genes, it is clear that the DRE/DREF system provides an elegant mechanism to orchestrate nuclear and mitochondrial DNA replication during cell proliferation.
We have also delimited the promoter region of the gene encoding the pol ␥ catalytic subunit. This region does not contain DRE sites, suggesting that the expression of the pol ␥-␣ gene may not be cell proliferation-dependent, consistent with its developmental pattern of expression. However, in its 5Ј-UTR, there is a DNA element potentially recognized by NRF1, a transcription factor that likely plays a key role in nucleomitochondrial communication in mammals (7,8). Mutation of this site significantly reduces the activity of the pol ␥-␣ promoter in Schneider's cells, suggesting that the NRF1-like element could be involved in the transcriptional regulation of the gene. NRF1 belongs to a family of transcriptional regulatory proteins with a DNA-binding domain conserved in two transcription factors, P3A2 and EWG, which play an essential role in the development of the sea urchin and Drosophila, respectively (52,53). Although the erect wing gene is expressed in all cell types, it is essential only in neurons and myoblasts (54); and thus, erect wing mutants exhibit neural and muscle defects. However, EMSA experiments do not support the involvement of the Erect Wing protein in the control of pol ␥-␣ gene expression. Nonetheless, because our transient transfection data demonstrate that the proximal 5Ј-upstream region of the gene is critical for promoter activity, the identification and characterization of protein factors that bind to this region and that regulate the expression of the gene remain an avenue for future studies.
In summary, we have shown that the expression of the catalytic and accessory subunit genes of the Drosophila mitochondrial DNA polymerase is regulated by different mechanisms. In particular, the expression of the pol ␥-␤ gene is regulated by the transcription factor DREF, thereby linking mitochondrial and nuclear DNA replication. Parallel studies on the biochemical roles of the accessory subunit and the developmental control of its synthesis should increase our understanding of the regulation of mtDNA replication under both physiological and pathological conditions.