Isolation and characterization of a DNA primase from human mitochondria.

A family of enzymatic activities isolated from human mitochondria is capable of initiating DNA replication on single-stranded templates. The principal enzymes include at least a primase and DNA polymerase gamma and require that rNTPs as well as dNTPs be present in the reaction mixture. Poly(dC) and poly(dT), as well as M13 phage DNA, are excellent templates for the primase activity. A single-stranded DNA containing the cloned origin of mitochondrial light-strand synthesis can be a more efficient template than M13 phage DNA alone. Primase and DNA polymerase activities were separated from each other by sedimentation in a glycerol density gradient. Using M13 phage DNA as template, these mitochondrial enzymes synthesize RNA primers that are 9 to 12 nucleotides in size and are covalently linked to nascent DNA. The formation of primers appears to be the rate-limiting step in the replication process. Replication of M13 DNA is sensitive to N-ethylmaleimide and dideoxynucleoside triphosphates, but insensitive to rifampicin, alpha-amanitin, and aphidicolin.


Tai Wai WongS and David A. Clayton
From the Department of Pathology, Stanford University School of Medicine, Stanford, California 94305 A family of enzymatic activities isolated from human mitochondria is capable of initiating DNA replication on single-stranded templates. The principal enzymes include at least a primase and DNA polymerase y and require that rNTPs as well as dNTPs be present in the reaction mixture. Poly(dC) and poly(dT), as well as M13 phage DNA, are excellent templates for the primase activity. A single-stranded DNA containing the cloned origin of mitochondrial light-strand synthesis can be a more efficient template than M 1 3 phage DNA alone. Primase and PNA polymerase activities were separated from each other by sedimentation in a glycerol density gradient. Using M 1 3 phage DNA as template, these mitochondrial enzymes synthesize RNA primers that are 9 to 12 nucleotides in size and are covalently linked to nascent DNA. The formation of primers appears to be the rate-limiting step in the replication process. Replication of M 1 3 DNA is sensitive to N-ethylmaleimide and dideoxynucleoside triphosphates, but insensitive to rifampicin, a-amanitin, and aphidkolin.
The mitochondrial genome of animal cells represents a versatile subject for studying eukaryotic DNA replication and gene expression. In mammals, the mitochondrial genome is a closed circular DNA of about 16.5 kilobases. This relatively small size and a compact gene organization render it more amenable to experimental manipulation than the complex chromosomes of the nucleus (1). The nucleotide sequences of three mammalian mtDNAs have been determined (2-4). It has become apparent from this sequence information and from biochemical data that most, if not all, of the enzymes that are involved with transcription and replication of mtDNA must be imported from the cytosol. Hence an understanding of the biosynthesis of mitochondrial nucleic acids and its regulation should provide useful insights into the mechanism of nuclear-cytoplasmic interactions.
Previous work in this laboratory has resulted in the isolation of a transcription activity from human mitochondria (5). The ability to reproduce the transcription process in uitro has allowed the identification of the major promoters for transcription of human mtDNA (6). We have also undertaken an effort to study in vitro replication of mtDNA. Our first goal is to isolate mitochondrial enzymes that are capable of initiating DNA replication. Here we report the identification and * This investigation was supported by Grant GM-33088-15 from the National Institute of General Medical Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Postdoctoral Fellow of The Helen Hay Whitney Foundation.
characterization of a fraction of mitochondrial proteins, isolated from human tissue culture cells, that can initiate replication on single-stranded DNA templates.

EXPERIMENTAL PROCEDURES
Reagents-Nucleoside triphosphates, poly(dA), poly(dC), poly-(dG), poly(dT), and polyA.oligo(dT) were purchased from P-L Biochemicals. c~-~'P-labeled dATP, dCTP, dGTP, GTP, and TTP were M13 DNA Templates-The single-stranded DNA template, M13KBL0"H,' contains a 226-nucleotide HincII fragment (nucleotides 5694-5919) of human mtDNA and was cloned in M13mp7 by Maureen J. Bibb of this laboratory. The insert contains H-strand DNA sequence of the origin of L-strand replication, flanked by tRNA genes. Phage particles were prepared from 500-ml cultures of infected E. coli JMlOl cells and were purified by two cycles of centrifugation in CsCl density gradients as described (7). Purified phage preparations were extracted with aqueous phenol and phage DNA was precipitated with ethanol.
Fractionation of Human Mitochondrial Proteins-Human epidermal carcinoma KB cells were grown in suspension cultures in Eagle's minimal essential medium supplemented with 10% calf serum. Mitochondria were prepared and were purified by centrifugation in sucrose density gradients as described (8). Unless otherwise specified, the following procedures were carried out at 4 "C. Mitochondria were resuspended in buffer A (10% glycerol, 20 mM Tris. HCl, pH 8.0, 0.2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 14 mM 2mercaptoethanol) to about 0.3 g of mitochondria/ml of buffer. Triton X-100 and NaCl were added to 1% (v/v) and 0.1 M, respectively. The suspension was incubated on ice for 15 min and mixed every 5 min by vigorous vortexing. It was then centrifuged in a Beckman 75 Ti rotor at 45,000 rpm for 1 h. The supernatant was recovered and diluted with one-half volume of buffer A to yield fraction I. Fraction I was applied to a 1.0 X 20-cm column of DE52 equilibrated in buffer A containing 70 mM NaC1. The column was washed with the same buffer at 20 ml/h until A z M of effluent was less than 0.1 The column was then washed with buffer A containing 0.3 M NaCl. Fractions that contained protein, as determined by AM, were pooled to yield fraction 11. Fraction I1 was immediately applied to a 1.5 X 10-cm column of phosphocellulose that had been equilibrated in buffer A containing 0.3 M NaCl. The column was washed with 2 column volumes of the same buffer and then with buffer A containing 0.6 M NaCl. Fractions that contained more than 0.05 mg of protein/ml were pooled and dialyzed against 1 liter of buffer that contained 50% glycerol, 20 mM Tris'HCl, pH 7.5, 0.2 mM EDTA, and 14 mM 2-mercaptoethanol to yield fraction 111. Fraction 111 was stored at -20 "C.
Alternatively, fraction I1 was subjected to ammonium sulfate fractionation. Solid ammonium sulfate was added to the fraction over a The abbreviations used are: L.,"H, heavy-strand template DNA sequence of light-strand replication origin; dh'TP, deoxyribonucleoside triphosphate; rNTP, ribonucleoside triphosphate, ddTTP, dideoxythymidine triphosphate; H and L strand, heavy and light strand, respectively; OL, origin of L-strand replication.

~i~o e~~r~~
DNA Primase 11531 20-min period to reach 45% saturation. The suspension was stirred gently for another 30 min and was centrifuged in a Beckman 75 Ti rotor at 25,000 rpm for 20 min. The pellet was resuspended in buffer B (10% glycerol, 20 mM Tris.HC1, pH 7.5, 0.2 mM EDTA, and 14 mM 2-mercaptoethanol) to a final concentration of 10 to 20 mg of proteinfml. The material was then dialyzed against two changes (1 liter each) of buffer B. Insoluble materials were removed by centrifugation and the supernatant (referred to as ammonium sulfate fraction in the text) was stored at -70 "C. Protein concentrations were determined as described by Bradford (9). DNA Replication Assay-DNA synthesis was assayed by measuring incorporation of 32P-labeled TMP or dAMP into acid-insoluble nucleic acids. Each assay was carried out in a total volume of 25 p1 and contained 20 mM Tris.HC1, pH 7.5, 10 mM MgC12, BSA at 0.1 mg/ ml, 1 mM 2 -m e r c a p t~t h~o l , 2 mM ATP, 200 pM each of CTP, GTP, and UTP, 10 p~ of [LX-~*P]~ATP or [cY-~'P]TTP (specific activity of 5,000 to 20,000 cpm/pmol) and the other three dNTPs at 100 p~ each, DNA template was either M13 phage DNA at 20 pg/ml or poly(dT) at 10 pg/ml. In assays of poly(dT) replication, ATP and dATP were the only nucleoside triphosphates in the reaction mixtures. The reaction was initiated by adding m i~h o n d r i a l fractions and was terminated by adding EDTA to 50 mM and 30 pg of sonicated calf thymus DNA. Incorporation was assayed by precipitation of reaction products with 1 N HCl containing 1% (w/v) sodium pymphosphate. Acid-insoluble materials were collected on Whatman GF/ A filters, which were then washed, dried, and counted in a toluenebased scintillation fluid. One unit of DNA primase activity is defined as the amount that will catalyze polymerization of 1 pmol of dAMP on poly(dT) template at 37 "C in 1 h.
Assay of DNA P a l y~e r~e y Activity-Each reaction mixture was in a total volume of 25 pl and contained 20 mM Tris.HC1, pH 7.5, 0.5 mM MnC12, BSA at 0.1 mg/ml, 50 mM sodium phosphate, 0.1 M NaCl, polyA'oligo(dT) at 25 pg/ml, 10 p~ [ C~-~~P J T T P (8,000 to 15,000 cpm/pmol), 1 mM 2-mercaptoethanol and varying amounts of mitochondrial enzymes. Reaction mixtures were incubated at 37 "C and incorporation was assayed by acid p r~i p i t a t~o n a s described above. One unit of activity is defined as the amount that will result in incorporation of 1 pmol of TMP into acid-insoluble materials at 37 "C in 1 h.
Assay of RNA Primer Synthesis-The reaction was carried out as described for assaying M13 DNA replication except that [(u-~*P]GTP ( 1 5 0 ,~ cpm/pmol) was used at 4 p~ as the only radiolabeled substrate. The reaction was terminated by extraction of the mixture with aqueous phenol. Ammonium acetate was added to 2 M and nucleic acids were precipitated with ethanol. Reaction products were resuspended in 5 pl of sample buffer (80% formamide, 50 mM Tris . borate, pH 8.3, 0.02% xylene cyanol, 0.02% bromphenol blue, and 1 m M EDTA) and were incubated at 90 "C for 3 min. They were than separated by electrophoresis in a 20% polyacrylamide gel containing 7 M urea as described (10). An alkaline hydrolysate of [(~-~~P]poly (A) was used as size markers.
DNase I digestions of synthetic products were carried out in 25 pl that contained 20 mM Tris. HCl, pH 7.5,lO mM MgC12, 1 mM CaC12, and RNase-free DNase I at 0.8 mg/ml. The reaction mixtures were incubated at 37 "C for 1 h and were then extracted with aqueous phenol. Nucleic acids were recovered by precipitation with ethanol.
Glycerol Density Gradient Sedimentation-4.5 pg of fraction 111 was adjusted by dilution to a final concentration of 10% glycerol, 20 mM Tris.HC1, pH 7.5, 0.2 mM EDTA, 0.1 M NaCl, and 14 mM 2mercaptoethanol and was layered onto a 4-ml linear gradient of 10 to 30% glycerol (v/v) in the same buffer. Centrifugation was in a Beckman SW60 Ti rotor at 45,000 rpm at 4 "C for 12 h. Fractions of 0.18 ml were collected from bottom of the centrifuge tube and were assayed for primase and DNA polymerase y activities as described above. In primase assays, the reaction mixtures contained also 0.4 unit of E. coli DNA polymerase I large fragment.

DNA Replication Enzymes in Human
Mitochondria-Our attempts to study in vitro replication of mtDNA began with the isolation of mitochondrial enzymes that are capable of priming and elongating nascent DNA molecules. In all of the eukaryotic and prokaryotic DNA replication systems that have been examined, RNA priming is the most common mechanism of initiating DNA synthesis (11). That a similar mechanism may be operative in mammalian mitochondria has been suggested by previous experimental data 02-14). For these reasons, we have chosen as criteria for DNA replication enzymes with the ability to catalyze the synthesis of RNA primers on a DNA template, and also to elongate these primers into daughter DNA strands. One of the DNA templates we have used is a 226-nucleotide fragment containing H-strand sequence of human mtDNA where L-strand replication initiates in uiuo (15). The fragment was cloned in bacteriophage M13mp7 and phage DNA (M13KBL0fiH) was used in assaying mitochondrial enzymatic activities. The use of single-stranded DNA was designed to mimic the physical state of the origin of L-strand replication (OL) when replication begins at that origin (16).
Mitochondria were purified from KB cells and lysed with Triton X-100. Using the crude lysate, we were not able to demonstrate replication activity on any DNA template. However, an ammonium sulfate fraction of mitochondrial proteins was able to stimulate DNA synthesis in the presence of M13KBL,,H DNA, dNTPs, rNTPs, and Mg2+ (Fig. 1). DNA synthesis was dependent on the presence of rNTPs in the reaction mixture, although a reduced level of DNA synthesis also occurred in the absence of rNTPs. Also, the rNTPdependent DNA synthesis on M13KBL,,iH was more efficient than that obtained with M13mp7 DNA maximal synthesis with M13KBL,iH was almost twice that with vector DNA alone. At present it is not possible to evaluate the significance of this apparent difference in the extent of DNA synthesis because of the crude nature of the enzyme preparation. Examination of the reaction mixtures by agarose gel electrophoresis revealed that greater than 90% of the template DNA was degraded in the course of the incubation (data not shown). A nuclease activity was present among the mitochondrial proteins and may account for the generation of 3'-hydroxyl ends that results in significant levels of rNTP-independent DNA synthesis with either template. The presence of nuclease in the ammonium sulfate fraction also makes it impossible to determine the nature of the OL-specific DNA synthesis using this particular enzyme fraction. We therefore proceeded to fractionate further the DNA replication activities.
The high-speed supernatant of the mitochondrial lysate was subjected to successive chromatography on DEAE and phosphocellulose columns. Fractions obtained throughout the purification were assayed for DNA polymerase y and primase activity. The latter was assayed using the synthetic template poly(dT). The two enzyme activities were adsorbed on both columns and were recovered by .elution at higher ionic strength. The final preparation, fraction 111, was completely free of any DNase activity. The 3-fold enhancement in recovery of primase activity following chromatography on DEAEcellulose is most likely due to removal of nuclease or other inhibiting activities (Table I). The simple chromatographic procedures resulted in an 85-fold and 760-fold purification of DNA polymerase and primase, respectively. When stored at -20 "C, both activities were stable for at least 2 months.
Separation of Mitochondrial Primase and DNA Polymerase-The chrom&ographic fractionation of mitochondrial enzymes was designed to allow maximal recovery of activities that may be involved in DNA replication, such as DNA polymerase and primase, and at the same time permit removal of all nuclease or other inhibitory activities. During purification by this scheme, we observed co-fractionation of DNA polymerase and primase activities. A tight association between cytosolic DNA primase activity and the major nuclear DNA polymerase activity has been observed previously in the purification of other eukaryotic DNA primases (17)(18)(19)(20)(21)(22). In order to determine whether or not the mitochondrial primase and DNA polymerase activities are tightly associated, we subjected fraction I11 to analytical centrifugation in glycerol density gradients. As shown in Fig. 2, mitochondrial primase activity appeared to be a very fast sedimenting entity, with an apparent sedimentation coefficient considerably greater than 11.3 S . On the other hand, DNA polymerase activity had an apparent sedimentation coefficient between 7.6 S and 11.3 S . While the bulk of the two enzymatic activities were well resolved from each other, there was a fraction (10 to 20%) of each activity that co-sedimented with the other activity. DNA polymerase y has been purified from chick embryos and shown to be an oligomeric enzyme with subunits of 47,000 daltons (23). The sedimentation rate of the peak of DNA polymerase activity shown in Fig. 2 is therefore consistent with the physical property of DNA polymerase y. Mitochondrial RNA polymerase has also been identified in KB cells and was found to have a sedimenation coefficient of about 8 S ( 5 ) . Thus it appears that the mitochondrial enzyme responsible for priming DNA replication on single-stranded templates is different from the major mtRNA polymerase.
Fraction ZZI Catalyzes RNA-primed DNA Synthesis-Fraction I11 was characterized for its ability to support DNA replication on biological and artificial DNA templates. With single-stranded M13 DNA, fraction I11 catalyzed DNA synthesis that was dependent on the presence of rNTPs in the reaction mixture (Fig. 3A). The activity increased sigmoidally with increasing enzyme concentration up to about 35 pglml. Higher concentrations of enzyme resulted in a drastic decrease in DNA synthesis. A small but significant amount of DNA synthesis occurred also in the absence of rNTPs. Analysis of these latter reaction products by agarose gel electrophoresis showed that they co-migrated with linear single-stranded DNA. They are therefore most likely products of elongation of 3'-hydroxyl ends of fold-back structures present in nicked phage DNA. Unlike the ammonium sulfate fraction, fraction 111 did not stimulate mitochondria-specific DNA synthesis when M13KBLmiH DNA was used as template (data not shown). This apparent discordance may be because the fractionation on phosphocellulose resulted in a reduction in concentrations of factors and enzymes that are required for interaction with OL (see "Discussion").
With poly(dT) as template, fraction I11 was also able to support the synthesis of poly(dA) (Fig. 3B). The activity was absolutely dependent on the presence of ATP and no DNA synthesis was detected in its absence. The extent of DNA synthesis could be enhanced 7-to 10-fold by the addition of 0.4 unit of E. coli DNA polymerase I large fragment (data not shown). We also compared different homopolymers for their ability to serve as templates for fraction 111. As shown in Table 11, poly(dC) appears to be the most efficient template for DNA replication by the mitochondrial primase. The rate of DNA synthesis with poly(dT) was only 25% of that obtained with poly(dC). In contrast, both poly(dA) andpoly(dG) were relatively ineffective templates for the primase activity.
Supplementation of the latter two reactions with ATP and GTP did not result in any increase in the extent of DNA synthesis. At present the basis for this rather specific template preference is not understood.
The replication of M13 DNA using fraction I11 was assayed as a function of time of incubation and was found to exhibit an initial lag of about 15 min before achieving its maximal rate (Fig. 4, closed circles). The initial lag varied from 15 to 17 min, after which the reaction remained linear for more than 60 min. The lag time was not due to temperature reequilibration because all components had been preincubated separately at 37 "C before the reaction was initiated. However, a 15-min preincubation of enzyme, DNA, M$+, and rNTPs together was able to reduce the lag time to less than 5 min (Fig. 4, squares). All four rNTPs had to be present in the preincubation in order for the shift in time course to be observed. In contrast, rNTP-independent DNA synthesis achieved a linear rate as soon as the reaction was started. These results suggest that the synthesis of primer DNA is a relatively slow process compared to DNA chain elongation. Primer formation may therefore be a rate-limiting step in single-stranded DNA replication involving the mitochondrial primase.
Discrete Oligoribonucleotide Primers Are Synthesized by Mitochondrial Fraction IZZ Enzymes-Data presented above suggest that fraction I11 is capable of catalyzing the synthesis of RNA primers that are subsequently elongated by DNA polymerase to form nascent DNA. In order to determine the nature of these RNA primers, we carried out a replication assay with fraction I11 and M13 DNA. The reaction mixture contained also radiolabeled GTP and the other three rNTPs, but no added dNTPs. Upon fractionation in a denaturing poiyacrylamide gel, the radiolabeled products were found to be mostly oligonucleotides (Fig. 5, lane 1). 'I'he sizes of these oligonucleotides exhibited a modal distribution ranging from Protein m., pg/ml

TABLE I1
Template specificity of mitochondrial primase DNA replication assays were carried out as described under "Experimental Procedures." Each reaction mixture contains DNA at 12 pg/ml, fraction 111 at 18 pg/ml, rNTP at 0.8 mM, and [w3'P]dNTP at 10 p~ (10, 9 to 12, 15 to 25, and 29 to 32 nucleotides. Present in lower abundance were also some larger species, greater than 600 nucleotides in size, that remained at the origin of the gel. If the reaction products were digested with RNase-free DNase I, the species at the origin were not present. At the same time, there was a proportionate increase in abundance of the smaller oligoribonucleotides (Fig. 5, lane 2). Similar results were obtained if all four dNTPs were also present in the reaction mixture. However, the major radiolabeled products seemed to be of high molecular weight and there was a notable absence of oligoribonucleotides of 9 to 12 nucleotides (Fig. 5, lane 3). Upon digestion of reaction products with DNase I, there was a complete conversion of the high molecular weight radiolabeled products to oligonucleotides of 9 to 12 nucleotides (lane  I and 2 ) or presence (lanes 3 and 4 ) of added dNTPs. Reaction products were analyzed directly (lanes I and 3) or after digestion with DNase I (lanes 2 and 4). They were fractionated by electrophoresis in a 20% polyacrylamide gel that contained 7 M urea. Numbers on the left side indicate sizes of markers in nucleotides.

) .
These data indicate that fraction I11 synthesized a group of small RNAs, mostly between 9 and 12 nucleotides in size, that were covalently linked to nascent DNA. The existence of DNase-sensitive products in reaction mixtures that contained no added dNTPs is most likely due to the presence of contaminating dNTPs in commercial preparations of rNTPs. Similar phenomena have previously been documented by other investigators (17, 18, 22).
Characteristics of DNA Replication Activities in Fraction ZZZ-The rNTP requirements of M13 DNA replication catalyzed by fraction 111 were examined and the results are shown in Table 111. While all four rNTPs are required for maximal DNA synthesis, ATP seems to be the most critical component; omission of ATP from the reaction mixture drastically reduces DNA replication. On the other hand, omission of any one of the other three rNTPs, such as GTP, resulted in only about a 50% reduction in DNA synthesis. We also examined the effects of a number of reagents on the replication reaction. The addition of rifampicin or a-amanitin to the reaction mixture did not have any deleterious effect on the replication of M13 DNA. These reagents are known to inhibit prokaryotic and eukaryotic nuclear RNA polymerases, respectively. Also, the reaction seemed to be insensitive to aphidicolin a t concentrations that inhibit DNA polymerase CY (26). Drastic inhibition of DNA replication was achieved with the use of

DISCUSSION
In this report we have described the identification and characterization of human mitochondrial enzymes that are capable of initiating DNA replication in uitro. These enzymes, referred to as fraction 111, support replication of singlestranded DNA templates such as M13 phage DNA and poly(dT). The replication process is absolutely dependent on the presence of rNTPs in the reaction mixture. We have presented evidence that the rNTPs are used to synthesize RNA primers of 9 to 12 nucleotides. These primers are elongated by an activity inferred to be DNA polymerase y on the basis of its sensitivity to several diagnostic reagents.
The assays that have been described in this report were not able to detect any specific initiation a t O L using fraction I11 and M13KBL0"H DNA. However, we have been able to achieve that goal using a more sensitive assay.* Under the assay conditions employed here, fraction I11 is not able to support replication of duplex DNA.3 The use of single-stranded DNA templates has significant biological relevance because the region of mtDNA where L-strand replication initiates is exposed as a single-stranded region at the onset of L-strand replication (16).
Properties of the primase activity in fraction I11 closely resemble those of other eukaryotic primase activities (17-22, 24,25). In all but one of these other cases, the primase activity was found to be tightly associated with the major nuclear DNA polymerase activity. This report represents the first description of a primase activity that is present in mitochondria and is presumably involved with replication of the mitochondrial genome. At present there is no indication as to whether the mitochondrial primase may be physically identical to the primase associated with DNA polymerase CY in KB cell cytosol (24). We have provided evidence that even though the bulk of DNA polymerase y and primase activities can be separated from each other, there exists a potential for the two T. W. Wong and D. A. Clayton, manuscript submitted. T. W. Wong, unpublished results. activities to participate in the formation of a multi-enzyme complex. The isolation of the plication enzymes should make it feasible to study the enzymatic interactions that are crucial to mtDNA replication and also to delineate the molecular mechanism of the replication process.