TWINKLE Has 5' -> 3' DNA helicase activity and is specifically stimulated by mitochondrial single-stranded DNA-binding protein.

Mutations in TWINKLE cause autosomal dominant progressive external ophthalmoplegia, a human disorder associated with multiple deletions in the mitochondrial DNA. TWINKLE displays primary sequence similarity to the phage T7 gene 4 primase-helicase, but no specific enzyme activity has been assigned to the protein. We have purified recombinant TWINKLE to near homogeneity and demonstrate here that TWINKLE is a DNA helicase with 5' to 3' directionality and distinct substrate requirements. The protein needs a stretch of 10 nucleotides of single-stranded DNA on the 5'-side of the duplex to unwind duplex DNA. In addition, helicase activity is not observed unless a short single-stranded 3'-tail is present. The helicase activity has an absolute requirement for hydrolysis of a nucleoside 5'-triphosphate, with UTP being the optimal substrate. DNA unwinding by TWINKLE is specifically stimulated by the mitochondrial single-stranded DNA-binding protein. Our enzymatic characterization strongly supports the notion that TWINKLE is the helicase at the mitochondrial DNA replication fork and provides evidence for a close relationship of the DNA replication machinery in bacteriophages and mammalian mitochondria.

The molecular mechanisms by which mtDNA is replicated in mammalian cells are of fundamental biological interest. Saccharomyces cerevisiae has served as a model system for studies of mammalian mtDNA replication, but there are significant differences between yeast and mammalian cells (1). Replication of the S. cerevisiae mtDNA is initiated from multiple sites of the ϳ86-kb genome, and the mtDNA molecules frequently undergo recombination. In contrast, the smaller mammalian mtDNAs (ϳ16 kb) initiate DNA replication from two specific origins of replication, oriH and oriL, and recombination is a rare or possibly even non-existent phenomenon (2).
Mammalian mtDNA contains two major promoters, the light and heavy strand promoters, which produce near genomic length transcripts that, after RNA processing, release individual mRNAs, tRNAs, and rRNAs. A separate transcription unit for the rRNA genes in mammalian mitochondria has also been reported (3). Transcription from light strand promoters is not only necessary for gene expression but also produces the RNA primers required for initiation of mtDNA replication at oriH (1,4). DNA synthesis from oriH is unidirectional and proceeds to displace the parental heavy strand. The nascent H strands frequently terminate 700 bp downstream of oriH, giving rise to 7 S DNA (D-loop strand). This termination event produces a characteristic triple-stranded structure, called the D-loop (5). The function of the D-loops is unknown, but they presumably play a role in regulating mtDNA replication.
The mitochondrial DNA polymerase ␥ is a heterodimer comprising catalytic (A) and accessory (B) subunits of 140 and 54 kDa, respectively. The accessory subunit, polymerase ␥ B, which is not present in yeast, has been characterized as a processivity factor for the polymerase (6,7). Polymerase ␥ B increases the affinity of the polymerase for DNA and promotes tighter nucleotide binding, increasing the polymerization rate (7,8). The processivity of DNA polymerase ␥ is specifically stimulated by the mitochondrial single-stranded DNA-binding protein, mtSSB 1 (9,10).
The TWINKLE gene was originally identified in a search for mutations associated with chromosome 10q24-linked autosomal dominant progressive external ophthalmoplegia (adPEO) (11), which is a human disorder with exercise intolerance, muscle weakness, peripheral neuropathy, deafness, ataxia, cataracts, and hypogonadism. Homology searches revealed a striking sequence similarity between TWINKLE and the bacteriophage T7 gene 4 protein, which contains both the DNA helicase and the primase activities needed at the bacteriophage replication fork. Primary sequence analysis revealed that TWINKLE contains sequence motives typically found in DNA helicases, whereas no obvious similarities could be identified with known primases.
Interestingly, adPEO is characterized by the presence of multiple mtDNA deletions, and the disorder has also been linked to mutations in DNA polymerase ␥ (12,13). Taken together, these data demonstrate a functional relationship between TWINKLE and the DNA polymerase and suggest that TWINKLE may be a DNA helicase active in mammalian mitochondrial DNA replication. A molecular characterization of TWINKLE would therefore not only be of interest for a molecular understanding of adPEO but would also reveal important insights into the molecular mechanisms of mtDNA replication.
We have purified TWINKLE in recombinant form and studied its enzymatic functions. We demonstrate here that TWINKLE is a 5Ј 3 3Ј DNA helicase, which is specifically stimulated by mtSSB. Our findings strongly support the notion that TWINKLE is involved in mammalian mtDNA replication and further demonstrate the remarkable structural and functional similarities between the DNA replication machinery in bacteriophages and mammalian mitochondria.

MATERIALS AND METHODS
Recombinant Proteins-Spodoptera frugiperda (Sf9) cells were maintained and propagated in suspension in SFM 900 medium (Invitrogen), containing 5% fetal calf serum, at 27°C. DNA fragments encoding TWINKLE and mtSSB were PCR-amplified from human cDNAs and cloned into the pBacPAK9 vector (Clontech). The TWINKLE expression construct encoded the mitochondrial form of the protein lacking the import signal (1-42 amino acids). A His 6 tag was introduced at the N terminus. The construct for mtSSB also encoded the mitochondrial form of the protein without the import signal (1-16 amino acids) but had no affinity tag. Autographa californica nuclear polyhedrosis viruses recombinant for the individual expression constructs were prepared as described in the BacPAK TM manual (Clontech). Protein expression was performed by growing 400 ml of Sf9 cells to a density of 2 ϫ 10 6 cells/ml in suspension. The cells were infected with 10 plaque-forming units/cell of recombinant baculovirus and harvested 72 h after infection. Infected cells were frozen in liquid nitrogen and thawed at 4°C in 20 ml of lysis buffer containing 25 mM Tris-HCl, pH 8.0, 10 mM ␤-mercaptoethanol, and 1ϫ protease inhibitors (for all purifications, a 100ϫ stock of protease inhibitors contained 100 mM phenylmethlsulfonyl fluoride, 200 mM pepstatin A, 60 mM leupeptin, and 200 mM benzamidine in 100% ethanol). The cells were incubated on ice for 20 min, transferred to a Dounce homogenizer, and disrupted using 20 strokes of a tight-fitting pestle. Next, NaCl was added to a final concentration of 1.0 M, and the homogenate was swirled gently for 45 min at 4°C. The extract was cleared by centrifugation at 36,000 rpm for 30 min at 4°C using a Beckman TLA 100.3 rotor.
Protein Purification-TWINKLE was purified by diluting the protein extracts with an equal volume of buffer A (50 mM Tris-HCl, pH 8.0, 0.6 M NaCl, 20% glycerol, 10 mM ␤-mercaptoethanol, and 1ϫ protease inhibitors) containing 20 mM imidazole. The extract was then added to 1 ml of Ni 2ϩ -agarose Superflow beads (Qiagen) and incubated for 1 h at ϩ4°C. Ni 2ϩ -agarose beads were collected by centrifugation (JA-17, 2,500 rpm, 10 min, ϩ4°C), washed once with 15 ml of buffer A containing 40 mM imidazole, again collected by centrifugation, and finally loaded into a column. The column was washed with 10 column volumes of buffer A containing 40 mM imidazole and eluted with 15 column volumes of buffer A containing 250 mM imidazole. The recombinant protein was identified by SDS-PAGE and Coomassie Brilliant Blue staining, and the peak fractions were pooled and diluted with an equal volume of buffer B (10 mM K⅐PO 4 , pH 7.2, 10% glycerol, 1 mM dithiothreitol, 100 mM NaCl, and 1ϫ protease inhibitors). TWINKLE was further purified on a 2-ml hydroxyapatite column (Bio-Rad) equilibrated in buffer B. The column was washed with 3 volumes of buffer B, and the proteins were eluted with a linear gradient (20 ml) of buffer B to buffer B containing 400 mM K⅐PO 4 . The TWINKLE protein eluted at a concentration of ϳ300 mM K⅐PO 4 . The peak fractions were diluted with 2 volumes of buffer C (20 mM Tris-HCl, pH 7.7, 10% glycerol, 1 mM dithiothreitol, and 0.5 mM EDTA, pH 8.0) and loaded on a 1-ml HiTrap SP column (Amersham Biosciences), which had been equilibrated with buffer C containing 150 mM NaCl. The TWINKLE protein eluted in the flow-through. The flow-through was collected and loaded on a 1-ml HiTrap heparin-Sepharose column (Amersham Biosciences), which had been equilibrated with buffer C with 150 mM NaCl. The column was washed with 3 volumes of buffer C with 0.3 M NaCl. The peak fractions were collected and eluted with a linear gradient (6 ml) of buffer C with 0.3-1 M NaCl. The TWINKLE protein eluted ϳ700 mM NaCl. The peak fractions were collected and dialyzed against buffer C with 0.2 M NaCl, frozen in liquid nitrogen, and finally stored at Ϫ80°C. The yield of TWINKLE protein was ϳ2 mg from a 400-ml culture. The purity of the protein was at least 95% as estimated by SDS-polyacrylamide gel electrophoresis and Coomassie Brilliant Blue staining (Fig. 1A).
The mitochondrial SSB was purified by dialyzing the clarified Sf9 lysate against buffer C with 0.1 M NaCl. The lysate was loaded onto a 5-ml CM Superose column (Amersham Biosciences), which had been equilibrated with buffer C with 0.1 M NaCl. MtSSB eluted in the flow-through was collected and loaded onto a 5-ml HiTrap heparin-Sepharose column (Amersham Biosciences), which had been equilibrated with buffer C (0.1 M NaCl). MtSSB was eluted with a linear gradient (50 ml) of buffer C (0.1-1 M NaCl), and the peak fractions of mtSSB were found at ϳ300 mM NaCl. The mtSSB fractions were collected and loaded on a 1-ml hydroxyapatite column, which had been equilibrated with buffer B (10 mM K⅐PO4). The column was washed by 3 volumes of equilibration buffer, and mtSSB was eluted with a linear gradient (10 ml) of buffer B (10 -400 mM K⅐PO 4 ). MtSSB, eluted at ϳ100 mM K⅐PO 4 , and the peak fractions were collected and dialyzed against buffer C (0.2 M NaCl), frozen in liquid nitrogen, and stored at Ϫ80°C.
The yield of mtSSB was ϳ20 mg from a 400-ml starting culture, and the purity was at least 95% as judged by SDS-PAGE and Coomassie Brilliant Blue staining (see Fig. 5A).
ATPase Assay-ATPase assays were performed in 20 l of 20 mM Tris⅐HCl (pH 7.8), 10 mM NaCl, 1.5 mM MgCl 2 , 17.4% glycerol, 0.3 mg/ml bovine serum albumin, 0.7 mM ATP, 250 -300 nCi of [␥ 32 P]ATP (Amersham Biosciences), and 10 g/ml activated calf thymus DNA. TWINKLE was added as indicated in the figure legends. Incubation was for 60 min at 37°C. The reaction was stopped by the addition of 400 l of a suspension of Norit A (12% in 0.1 M HCl, 10 mM K⅐PO 4 ). The mixture was briefly vortexed and then centrifuged for 3 min at 7,000 ϫ g. Two hundred l of the supernatant was mixed in 3 ml of scintillation mixture (Ready Safe; Beckman-Coulter), and the radioactivity was measured in a liquid scintillation counter. The enzyme-dependent release of phosphate was calculated by subtracting the release of phosphate in samples without enzyme.
Helicase Assay-The reaction mixture (15 l) contained 15 fmol of DNA substrate (DNA concentrations in this report are expressed in moles of molecules), 20 mM Tris-HCl (pH 7.6), 10% glycerol, 10 mM dithiothreitol, 4.5 mM MgCl 2 , 3 mM ATP, 100 g/ml bovine serum albumin, 40 mM NaCl, and the indicated amounts of TWINKLE, mtSSB, and Escherichia coli SSB. The reactions were incubated at 32°C for the times indicated and stopped by the addition of 2 l of stop solution (90 mM EDTA (pH 8.0), 6% SDS, 30% glycerol, 0.25% bromphenol, 0.25% xylene cyanol). We did not observe any significant levels of spontaneous reannealing of unwound DNA at the assay conditions used. The products were separated by electrophoresis through a 15% non-denaturing polyacrylamide gel, which was dried onto DE81 (Whatman) and autoradiographed overnight at Ϫ80°C with an intensifying screen. Intensities of the bands were quantified by densitometry using the program NIH Image (rsb.info.nih.gov/nih-image/).

RESULTS
The finding that TWINKLE mutations are associated with mtDNA deletions suggested that the protein might be involved in mtDNA replication, and we therefore characterized the enzymatic activities of the TWINKLE in vitro. We generated a recombinant baculovirus encoding the human TWINKLE gene to obtain sufficient quantities of the protein for studies of its associated biochemical activities. TWINKLE was expressed in insect cells and purified over Ni 2ϩ -agarose, hydroxyapatite, SP-Sepharose, and heparin-Sepharose to near homogeneity (Fig. 1A). Recombinant TWINKLE migrated as a doublet with an apparent molecular mass of about 70 kDa during SDSpolyacrylamide gel electrophoresis, corresponding to the predicted molecular mass of 72 kDa. The weaker, lower band is a shorter form of the protein, lacking 15 amino acids at the very C terminus as demonstrated by mass analysis with MALDI-TOF mass spectrometry (data not shown). The truncation is probably due to translational pausing since the relative levels of the two forms are unaffected by protease inhibitors. A strong contaminating exonuclease activity co-purified with TWINKLE over the first two columns, but it was lost at the SP-Sepharose step. Heparin column-purified TWINKLE co-migrated with a strong ATPase activity (Fig. 1B), supporting primary sequence analysis predictions that TWINKLE is a Walker-type ATPase.
TWINKLE Is a DNA Helicase-To investigate whether TWINKLE is an active DNA helicase in vitro, we annealed a 32 P-labeled 60-nt oligonucleotide to the complementary region of M13mp18 single-stranded DNA to form a helicase substrate with a 20-bp double-stranded region and a 40-nt 5Ј-singlestranded tail. Examination of the purified protein showed that FIG. 1. ATPase and DNA helicase activities follows the TWINKLE protein peak from the heparin column. In A, TWINKLE peak fractions (10 l) from the heparin column were separated by SDS-PAGE (15%) and revealed with Coomassie Brilliant Blue staining. The protein concentration in the peak (fraction 26) was 0.25 mg/ml. M, molecular size marker. In B, the ATPase activity of the protein fractions (1 l) in panel A was analyzed in the presence of activated calf thymus DNA. C, a helicase assay over the peak fraction from the heparin column. One l from each fraction was added to the reaction mixture as described under "Materials and Methods" and incubated for 30 min. Lane 1, substrate heated to 100°C before loading; Lane 2, untreated substrate; S, double-stranded substrate; P, singlestranded product. it indeed possessed a strong DNA unwinding activity, which coincided perfectly with the peak of TWINKLE protein eluting from heparin-Sepharose (Fig. 1C). The helicase activity was not found when mock-infected insect cell extracts were purified in a similar way (data not shown). We next analyzed DNA unwinding as a function of TWINKLE protein concentration and found that addition of increasing concentrations of protein to our helicase substrate revealed a linear increase of displaced oligonucleotide (Fig. 2, A and B). Extensive DNA unwinding (Ͼ25%) was observed at protein concentrations above 30 nM of the protein, demonstrating that TWINKLE is a potent helicase in vitro. A time course experiment (Fig. 2C) demonstrated that TWINKLE initiates unwinding of the substrate without any apparent lag phase. The unwinding reaction is dependent on NTP hydrolysis and was inhibited at low concentrations of the non-hydrolyzable ATP analogue, ATP␥S (Fig. 3A), TWINKLE was next incubated with a variety of different nucleoside 5Јtriphosphates to analyze their ability to act as co-effectors for the helicase activity (Fig. 3B). UTP efficiently supported DNA unwinding, and to a lesser extent, so did ATP, GTP, and dTTP. CTP was a very poor co-effector. TWINKLE Has 5Ј 3 3Ј Helicase Directionality-We next investigated the efficiency by which TWINKLE could unwind different DNA substrates. To this end, we made a series of constructs with a 60-nt-long oligonucleotide annealed to complementary oligonucleotides to form helicase substrates with a 20-bp double-stranded region, a 40-nt single-stranded 5Ј-tail, and 3Ј-tails of varying lengths (0, 10, 15 nt). No unwinding was observed with substrate lacking a 3Ј-tail, low levels of unwinding were observed with the 10-nt 3Ј-tail, whereas the 15-nt 3Ј-tail template was efficiently unwound (Fig. 3C). Similar experiments were also performed with a 40-nt 3Ј-tail substrate and varying lengths of the 5Ј-tail (0, 10, 15 nt). We found that unwinding was dependent on a free 5Ј-tail as well. No unwinding was observed on the template lacking a 5Ј-tail. TWINKLE displayed a moderate activity on the 10-nt 5Ј-tail substrate and efficiently unwound the 15-nt 5Ј-tailed template (data not shown).
TWINKLE needs a fork-like structure with both a 5Ј-and a 3Ј-single-stranded stretch of DNA to efficiently initiate DNA unwinding, similar to what had been shown previously for the T7 gp4 helicase (15). We could not use common DNA substrates for a directionality assay, due to this specific substrate requirement. Specific DNA substrates have been developed to circumvent this problem for the T7 gp4 protein. We used a directionality assay with substrates containing a 20-bp double-stranded region with one single-stranded and one double-stranded tail (Fig. 4, A and B) (16). TWINKLE requires a single-stranded stretch of DNA to initiate unwinding, and we could therefore determine the directionality of the enzyme by introducing a single-stranded 5Ј-or 3Ј-tail. We found that TWINKLE could only unwind the substrate with a 5Ј-single-stranded tail, thus demonstrating that TWINKLE unwinds DNA in the 5Ј to 3Ј direction, as does the T7 gp4 helicase.
The TWINKLE Helicase Is Stimulated by mtSSB-We determined whether human mtSSB had any stimulatory effect on the TWINKLE helicase activity by using the same substrate as in Fig. 1C. We expressed recombinant mtSSB in insect cells and purified the protein to near homogeneity (Fig. 5A). We found that mtSSB had a strong stimulatory effect on the unwinding activity of the TWINKLE protein (Fig. 5B). The stimulation by mtSSB was specific because no such effect was observed with the E. coli SSB. The assays were performed as described under "Materials and Methods." Increasing amounts of TWINKLE protein was added to 20 fmol of template and incubated for 40 min. As shown in A, TWINKLE cannot unwind the double-stranded template with a 3Ј-single-stranded tail. S, double-stranded substrate; P, single-stranded product. As shown in B, TWINKLE can efficiently unwind a double-stranded template containing a 5Ј-single-stranded tail. Lane 1, untreated substrate; lane 2, substrate heated to 100°C before loading.

DISCUSSION
The mechanisms of mammalian mtDNA replication have not yet been fully defined (1). We have initiated a project aimed at reconstituting mtDNA replication in vitro, which we hope will generate new insights into this fundamental cellular process. According to the generally accepted model, mammalian mtDNA replication is continuous on both strands and takes place in a strand-asymmetric mode. DNA synthesis is initiated from two different sites, one for each strand (17). Activation of oriL and the heavy strand origin oriH are physically and temporally distinct. DNA synthesis first commences at oriH, which is localized in the non-coding region of mtDNA. After leading strand synthesis has reached two-thirds of the genome, it comes to oriL, which is activated, and the DNA synthesis then initiates in the opposite direction. Recently, however, this strand-asymmetric model for mtDNA replication has been challenged by two-dimensional gel electrophoresis analysis demonstrating the presence of conventional duplex mtDNA replication intermediates, indicative of coupled leading and lagging-strand DNA synthesis (18,19). A detailed biochemical analysis of these processes in vitro may help to clarify the molecular mechanisms of mtDNA replication We demonstrate here that the mammalian TWINKLE protein displays the classical features of a DNA helicase: it catalyzes the ATP-dependent unwinding of a DNA duplex with a distinct polarity (5Ј 3 3Ј). The protein requires specific substrates with a single-stranded 5Ј-DNA loading site and a short 3Ј-tail to initiate unwinding. The preferred substrate thus resembles the conformation of a DNA replication fork, a structure with which the TWINKLE protein would be expected to interact. The substrate requirement is also similar to what has been described previously for the T7 gp4 protein (15) and other hexameric helicases, such as DnaB (20). The ability of the TWINKLE protein to utilize various nucleoside 5Ј-triphosphates as co-effectors for helicase activity is interesting. ATP efficiently supports TWINKLE-mediated DNA unwinding, but UTP is clearly a much more potent cofactor. The physiological relevance of this observation remains to be established.
MtSSB has a stimulatory effect on the rate of DNA unwinding, and this effect is specific as the E. coli SSB cannot substitute for mtSSB. The observed specificity may be due to a direct interaction between mtSSB and the TWINKLE protein. Physical interactions between replicative helicases and their endogenous single-stranded DNA-binding proteins have been demonstrated in other systems, e.g. the herpes simplex virus type 1 helicase-primase complex is specifically stimulated by the viral SSB, ICP8 (21,22). The specific stimulatory effect by mtSSB therefore supports the notion that TWINKLE is the replicative DNA helicase in mammalian mitochondria.
The identification of TWINKLE as a DNA helicase means that only one essential function, a primase, is missing from the mitochondrial replisome. A primase activity is constantly required at the replication fork for coordinated leading and lagging DNA strand synthesis, whereas a strand-asynchronous mode of replication would only need primase activity for initiation of lagging strand DNA synthesis at oriL. The T7 gp4 protein contains a primase activity, which synthesizes primers at the bacteriophage DNA replication fork (23). Primary sequence comparisons suggest that TWINKLE has lost its primase activity, but such an activity cannot be ruled out without careful biochemical analysis (11). Wong and Clayton (24,25) previously reported a primase activity specifically acting at oriL in mammalian mitochondria. The oriL region forms a conserved hairpin structure, which may function as an attenuation site for the TWINKLE helicase and the mitochondrial replisome. The paused replisome may then recruit a mitochondrial primase, which can synthesize the primer needed for initiation of light strand DNA synthesis. Alternatively, the hairpin structure could function as signal to activate an atypical primase activity in the TWINKLE protein itself. Future efforts will be directed toward investigating these intriguing possibilities for additional functions of TWINKLE.