Properties and Mode of Action of a Partially Purified Deoxyribonucleic Acid Polymerase from the Mitochondria of HeLa Cells*

of synthesis at sites of single strand scissions with covalent of nucleotides the template

Circular species of DNA have been identified as a genetic component in many systems, including the mitochondria of animal cells. Mitochondria isolated from several animal cells and tissues have been shown to incorporate deoxyribonucleotides into mitochondrial DNA in situ (l-8). Kalf and Ch'ih (9) and Meyer and Simpson (10) described the isolation and partial purification from rat liver mitochondria of a DNA polymerase with fractionation and reaction properties that differ from DNA polymerase activity associated with the cell nucleus.
The mitochondria of animal cells are thus a source of DNA polymerase(s) which operate on circular duplex DNA in viva.
Current understanding of in V&JO DNA replication at the level of enzyme action is largely speculative.
1n vitro studies of DNA polymerase activity from animal cell mitochondria were carried out with the object of extending our understanding of the process of circular DNA replication.
We have isolated and partially purified a DNA polymerase-containing fraction from sonic ex-tracts of HeLa cell mitochondria and have investigated the mechanism of DNA synthesis on circular DNA templates.

Materials
Nucleic Acids and Precursors-Unlabeled nucleotides were purchased from Sigma. 5-Bromodeoxycytidine 5'-triphosphate and 5-bromodeoxyuridine 5'-triphosphate were prepared as described under "Methods." [aH]dTTP was obtained from Schwarz BioResearch and New England Nuclear. Other 3Hlabeled nucleotides were obtained from Schwarz.
Labeled and unlabeled nucleotide solutions were mixed to give samples with lower specific activity and higher concentration.
The mixtures were analyzed by Dowex 1 chromatography (0.2 ml of approximately 1 mM nucleotide applied to 0.3 x 3.0 cm column; 60-ml linear gradient elution, from 10 mM Tris, pH 7.5, to 0.2 M HCI, 0.5 M LiCl) to determine the resultant specific activity and assess the purity of the nucleotide preparation. In each case, more than 85y0 of the applied radioactivity and optical density at the wave length of the nucleotide's maximum absorbance were recovered as the deoxynucleoside t&phosphate with constant specific activity across the peak.
Native calf thymus DNA was purchased from Sigma. T7 phage DNA was a gift of Dr. R. W. Hyman.
SV40 viral DNA and SV40 intracellular DNA were prepared in the manner of Rush et al. (11). HeLa cell closed circular mitochondrial DNA was extracted from mitochondria prepared in the manner described under "Methods," omitting, however, the two buoyant sucrose gradients.
32P-labeled SV40 circular DNA was a gift of Dr. R. Eason.
Further samples were prepared as needed according to Rush et al. (11).
[14C]Thymidine-labeled 4X174 single-stranded circular DNA was extracted from a phage preparation supplied by R. Benbow.
DNA preparations were treated with 20 mg per ml of SDS' at room temperature for 10 min prior to banding in a CsCl equilibrium density gradient. The gradients for the isolation of closed circular DNA contained ethidium bromide (12). The dye was removed with a Dowex 50 column (0.3 x 3.0 cm) in the presence of 4 M NaCl or by extraction with isoamyl alcohol. 3365 DNA samples were then dialyzed exhaustively against 10 mM Tris, 1 mM EDTA, pH 7.5. The integrity of the DNA samples was checked by analytical band velocity centrifugation.
Enzymes-Escherichia coli exonucleases I and III were stored frozen at 15,000 units per ml and 490,000 units per ml, respectively.
The preparations were gifts of D. Brutlag, Stanford University.
CsCl was purchased from Harshaw Chemical Company. Bovine serum albumin, fraction V (crystalline), was from Miles Laboratories. Ethidium bromide was donated by Boots Pure Drug Co. Ltd., Nottingham, England Methods Chromatography-Ion exchange resins were washed with 0.5 M NaOH, 0.5 RI HCl, 0.1 M EDTA, 4 M NaCl, and then washed repcatedly with glass-distilled water to pH equivalence. Washed resins were stored in 4 M NaCI. After packing, at least 20 bed volumes were eluted from columns before application of a sample. Bio-Gel P-150, 100 to 200 mesh, was prepared by swelling in 10 m&I Tris, 10 mM 2-mercaptoethanol, pH 7.5, at least 8 hours before packing.
One bed volume of elution buffer was passed through the column before application of the sample.
The bed volume of the gel filtration columns was 100 to 200 ml and sample volume was less than 5 ml.
Spectrophotometry-,bsorption spectra were recorded using a Gary model 14 spectrophotometer.
Optical density was measured as the difference in absorbance of the sample and blank solutions determined separately in the same quartz cell with an open reference beam path.
Protein concentration was assessed by absorbance at 280 nm and 260 nm using the method of Warburg and Christian (13). DNA concentration was measured at 260 nm using a value of 1.00 absorbance unit for a solution of 50 pg per ml of native duplex DNA.
Fluorescence measurements of ethidium bromide solutions were performed with quartz cells in a Hitachi Perkin-Elmer MPF-2A fluorescence spectrophotometer.
The excitation and emission wave lengths were 380 nm and 590 nm. Slit widths were varied according to the needs of each experiment.
A 300-to 400-nm bandpass filter was placed between the exit slit of the excitation monochromator and the sample cell. A 430~nm cutoff filter was placed between the sample cell and the entrance of the emission monochromator.
Measurements were recorded in the ratio mode to correct for fluctuations in lamp intensity.
Centr+gation-Preparative ultracentrifugation studies were performed in Beckman model L or L2-65B centrifuges with SW 50L or SW 50.1 rotors.
The CsCl self-generating density gradient system was used for band sedimentation analysis. Samples (less than 0.2 ml) were layered through Bayol onto at least 3.0 ml of 1.4 g per ml CsCl solution prior to centrifugation at 35,000 rpm. Gradients for alkaline sedimentation velocity analysis contained 0.1 M KOH.
Buoyant equilibrium CsCl density gradients were run at 35,000 rpm, 20 to 25", for at least 36 hours. Initial solution density was 1.70 g per ml for neutral CsCl gradients, 1.74 g per ml for alkaline CsCl gradients, and 1.56 g per ml for neutral CsCl gradients containing ethidium bromide (250 pg per ml). Alkaline buoyant gradients contained 50 mM K3P04 from a 0.25 M solution of KaHPOd adjusted to pH 12.5 by addition of KOH.
Cellulose nitrate and polyallomer tubes were used for centrifugation at neutral and alkaline pH, respectively. Velocity and buoyant gradients were punctured at the bottom of the tubes and lo-drop (about 100 ~1) fractions were collected for analysis.
Liquid Xcintillation Counting-Samples were applied to Whatman No. 3MM filter papers.
The volume added was 5 ml for counting 3H, 10 ml for 3H and/or 14C, and 15 ml for 3H and/or 32P. Counting efficiency and channel spillover in the Packard Tri-Carb liquid scintillation spectrometer were determined with standards applied to paper filters.
Double-label experiments were analyzed using the 3H channel exclusion method.
Electron JJicroscopy-Specimens were prepared using the formamide modification of the Kleinschmidt procedure (14). The DNA samples were first dialyzed against 50% formamide, 100 mM Tris, 10 mM EDTA (pH 8.5). Cytochrome c was added to 50 pg per ml prior to casting on a hypophase containing 17% formamide, IO mM Tris, 1 mM EDTA (pH 8.5). Grids were shadowed on a rotating platform 8 degrees below and 5 cm from the point of evaporation of a Pt-Pd wire.
Electron micrographs were recorded on 35-mm film using the Philips 300 electron microscope .2 Preparation of Brominated Nucleolides-5-Bromodeoxyuridine 5'-triphosphate was prepared as described by Bessman et al. (15), by bromination of dCTP in formamide and subsequent deamination of the 5-bromodeoxycytidine 5'.triphosphate with nitrous acid. The deoxynucleoside triphosphates were isolated from reaction mixtures by adding BaBrn to 0.25 M and precipitating with ethanol.
The precipitates were washed in ethanol, dried in an airstrcam, and resuspended in 10 mM Tris (pH 7.5). The suspended nucleotides were metathesized to soluble potassium salts with Dowex 50. Ultraviolet absorption spectra and Dowex 1 chromatography were used to check the quality of the preparations.
Tissue Culture-HeLa cells were maintained in suspension culture using the Dulbecco modification of Eagle's phosphate medium supplemented with 5% calf serum. Spinners were inoculated at 3 x lo4 cells per ml, 72 hours prior to the harvest of cells for the preparation of mitochondria. BSC-1 cells were maintained on plastic dishes and SV40 infections were carried out as described by Rush et al. (11).
Preparation of HeLa Cell illitochondria-At the time of harvest the concentration of cells was 3 to 4 X lo5 cells per ml. Cells were recovered from suspension by centrifugation in batches for 3 min each at 1400 x g in the International PR-6, and in the latter part of the work at 3000 X g in the Sorvall RC2 using the Szent-Gyorgyi and Blum continuous flow system. A preparation of 20-liter suspension can be processed by continuous flow in less than 1 hour with a normal yield of 30 to 40 ml of packed cells. Cell pellets were resuspended in 150 ml of Buffer A (137 mu NaCl, 5.8 mM KCl, 0.7 mM NatHPO+ 25 mM Tris, pH 7.5). Examination in the phase contrast microscope did not reveal fragmentation of the cells at the higher fields used in the continuous flow harvesting operation.
Operations following cell harvest were performed at 4". Cells were pelleted in four 50-ml conical centrifuge tubes for 5 min at 900 x g in the International PR-6. After removing the supernatant, Buffer B (10 mM 'Iris, 10 mM NaCl, 1.5 m MgClz, pH 7.5) was added to 30 ml in each tube and the cells were gently resuspended with a wide bore pipette.
Cells were allowed to swell in this hypotonic buffer for about 30 min. Each batch was homogenized using two strokes of the 40-ml glass Dounce homogenizer (Kontes) with tight-fitting pestle. One-third volume of 1.1 M sucrose, 10 rnM Tris, 1 m&l EDTA (pH 7.5) was added to the homogenate followed by an additional stroke of the homogenizer for mixing.
Phase contrast microscopy confirmed greater than 90% disruption.
The free nuclei appeared intact. Cells, nuclei, and large debris were removed by centrifugation three times for 5 min at 2000 x g, discarding the pellets each time. The resulting supernatant was layered onto 1.5 M sucrose, 10 rnM Tris, 1 mM EDTA (pH 7.5) in six tubes containing 25 ml and 10 ml of the supernatant and 1.5 M sucrose stages, respectively. Centrifugation followed for 30 min in the SW27 rotor at 25,000 rpm. Mitochondria were recovered from the interface and resuspended in 90 ml of Buffer C (210 rn_RI mannitol, 70 mM sucrose, 10 mivr Tris, 1 rnM EDTA, pH 7.5). The mitochondrial sample in Buffer C was layered onto six preformed, two-step gradients containing 10 ml of 1.1 hi sucrose, 10 m&r 'l'ris, I m&r EDTA, 1X 7.5, over 10 ml of 1.5 M sucrose, 10 m&t 'I'ris, I miw EDT,4, pI1 7.5. Cent)rifugation followed for 30 min at 25,000 rpm as before.
Xtochondria were recovered at the interface between the 1.1 and 1.5 in sucrose solutions, resuspended in Buffer C, and pelleted for 15 min at 12,000 rpm in the Sorvall KCS-1~ with an SS-34 rotor. The mitochondrial pellet was resuspended in 2y0 SI)S, 0.1 of ED'I'A, pI1 7.5, for estraction of mitochondrial DNA, or 10 rnnl Tris, 10 xnM 2-mercaptoethanol, pH 7.5, for extraction of soluble mitochondrial proteins by sonication. Enzyme Assays-DNA polymcrasc was assayed using a modification of t#he procedure described by Bollum (16) based on the incorporation of labeled deoxynucleoside triphosphates into acid-precipitable material retained on paper filters through a batch washing procedure.
B typical reaction mixture contained 10 mM Tris (pH 7.5), 3.3 mM 2-mercaptoethanol, 33 mM NaCl, 3.3 rnnf MgC12, 0.33 mM EDTA, 170 pM dGTP, 155 pM dCTP, 310 ~.lhl dTTP, 84 PM [3H]dhl'P (specific activity, 250 Ci per mole), 20 pg per ml of native calf thymus DNA, and enzyme protein in the range of 50 to 200 pg per ml. Prior to incubation, materials were kept at 4". After zero, 30, and 60 min at 37", 50-~1 aliquots were withdrawn from the incubation mixture, with the same micropipette, onto two numbered paper filters, one of which was immediately placed in a beaker containing 500 ml of cold 5% trichloroacetic acid, 1 y0 sodium pyrophosphate. Thirty minutes after t)he addition of the last filter to the bath, the filters were washed with cold 5y0 trichloroacetic acid three times for 20 min and then with 95% ethanol, followed by ethyl ether and drying.
Incorporation of [3H]dATP was assessed in two ways. The 3H counts per min retained on the washed filter divided by the specific activity of the [3H]dATP was the amount of dcosyadenylate (nanomoles) in the 50.~1 aliquot rendered insoluble by action of t,he enzyme system. The ratio of 3H counts per min on the washed and unwashed filters corresponds to the fraction of deoxyadenylate in the reaction which is acidprecipitable.
This fraction and the initial concentration of dATP were used to calculate the total amount of deoxyadenylate polymerized by the enzyme system. The two methods gave the same result.
Linear incorporation kinetics (Figs. 4 and 5) justified the calculation of the rate of incorporation from a single determination of the amount of nucleotide rendered acidprecipitable.
One unit of DNA polymerase activity is defined as the conversion of 1 nmoIe of labeled nucleotide into acid-precipitable material after 60 min at 37". Specific activity of enzyme preparations is espressed as units per mg of protein.
Levels of deoxyribonuclease activity were so low that the Kunitz assay (17) and attempts to demonstrate the solubilization of radioactive-labeled, acid-precipitable DNA samples by the action of the enzyme system were not satisfactory.
A more sensitive endonuclease assay developed by Paoletti el al. (18) and based on the conversion of closed circular DNA to nicked circular DNA (or to linear DNA) was used. On binding to DNA, the fluorescence of ethidium bromide is enhanced.
Endonuclease action relieves a restriction on the amount of dye which can bind to closed circular DNA.
The increase in fluorescence of an ethidium bromide solution containing samples of initially closed circular DNA can be related to the extent of endonuclease action. Equal aliquots were taken at 5-min intervals from a reaction mixture containing initially closed circular SV40 DNA, and diluted 25-fold in 1 IIIM EDTA, 1 pg per ml of ethidium bromide, pH 7.5. The fluorescence enhancement, E, is the measured increase in the fluorescence of the ethidium bromide solution after adding the sample. The fraction of closed circular molecules surviving in a given sample was calculated from the observed fluorescence enhancement using the linear relation En -E Fraction SV40-I = ~ Err -EI where El and Brr are the fluorescence enhancements of solutions of SV40-I and SV40-II, respectively, at the same concentrations as the DNA in the sample used to measure E.

Extraction
and Partial Purijkation of Xitcchondrial DNA Polymeruse-Freshly prepared HeLa cell mitochondria were suspended in 3 to 5 ml of 10 mM Tris, 10 rnM 2-mercaptoethanol (pH 7.5) and subjected to vigorous sonication.
The tip of the horn of the Model S-125 Branson sonifier was immersed directly into the suspension during three 20-s bursts at level six. The 12.ml conical test tube was chilled in ice after each burst.
The turbid suspension clarified.
Following sonication, addition of NaCl to 1 M and extraction for up to 12 hours did not increase the yield of protein or DNA polgmerase activity.
Addition of salts prior to sonication greatly interfered with the extraction.
The sonicate was then centrifuged in polycarbonate tubes at 40,000 rpm for 1 hour in the type 65 fixed angle rotor.
Most of the clear, pale yellow supernatant was carefully withdrawn from the top of the tube with a Pasteur pipette.
The last 0.5 to 1.0 ml of fluid gave rise to visible schlieren effects when drawn into the pipette.
This material and the clear gelatinous pellet were routinely discarded.
Earlier experiments in which the lower phase was included in the crude extract yielded final preparations of DNA polymerase with similar levels of activity, but with significantly lower ratios of Azg,,:A 260, suggesting greater nucleic acid contamination.
The crude extract, designated Fraction CE, was then applied to a Bio-Gel P-150 column for gel filtration, eluting with 10 mM Tris, 10 mM 2-mercaptoethanol (pH 7.5). Elution of the sample was monitored by absorbance at 280 nm. DNA polymerase can be located by assay of the eluted fractions, but the protein is dilute at this stage and the levels of incorporation of labeled nucleotides are low. The procedure now employed is to postpone the assay until each of the fractions in which the enzyme is routinely recovered has been individually subjected to the fractional ammonium sulfate precipitation and concentration step described next. Fig. 1 shows a typical elution profile from the Bio-Gel P-150 column and the location of DNA polymerase activit,y in the fractions following elution of the void volume.

Fraction Number
FIG. 1. Gel filtration of supernatant fraction following centrifugation of mitochondrial sonicate. The sample (4 ml) was applied to a Bio-Gel P-150 column having a bed volume of 170 ml. The column was pre-equilibrated with 10 mM Tris, 10 mM 2-mercaptoethanol, pH 7.5, the buffer used for elution. All column operations were performed at ambient room temperature.
Aliquots from Fractions 10 through 17 were assayed for DNA polymerase activity (O-O) as described under "Methods," except that the radioactive label was 6 PM [3H]dTTP (7 Ci per mmole) and dATP was present at 175 PM. The DNA polymerase activity is represented by the radioactivity rendered acid-precipitable after incubation for 30 min at 37".
Selected column fractions were raised to 35% saturated ammonium sulfate by the addition of 0.54 ml of saturated ammonium sulfate per ml of sample. After 30 min at 4' the precipitated material was removed by centrifugation at 40,000 rpm, 30 min, in the type 65 rotor. The supernatant was raised to 50% saturated ammonium sulfate by the addition of 0.30 ml saturated ammonium sulfate per ml of sample, chilled 30 min at 4", and the precipitate collected as before. The precipitated material was carefully resuspended in 0.5 ml of 10 mM Tris, 10 mM 2-mercaptoethanol, 10 mu MgClz, 100 mu NaCl, pH 7.5.
A typical preparation from 20 liters of suspension culture yielded 40 ml of packed cells, 1 ml of wet, purified mitochondria, about 10 mg of protein in the crude extract, and 0.5 mg of protein in Fraction AS representing the pooled fractions containing DNA polymerase activity. The expected yield of mitochondrial DNA polymerase was calculated on the basis of the ad hoc assumption of 1 to 10 molecules of the enzyme per molecule of mitochondrial DNA in HeLa cells. The yield of mitochondrial DNA from HeLa cells is about 2 kg or 2 x lo-la mole per ml packed cells. If the molecular weight of the DNA polymerase in Fraction AS is 1.5 x lo5 g per mole (see Fig. l), we would expect about 1 to 10 pg of pure DNA polymerase in the typical preparation of Fraction AS which leads to 500 pg of total protein derived from 40 ml of packed cells.

Attempts
to purify the enzyme further by DEAE-cellulose chromatography were unsuccessful. None of the DNA polymerase activity applied to the columns was recovered in individual eluted fractions (step-gradient elution), or after pooling and concentration of the total eluted protein.
The time required for a preparation of Fraction AS is about 14 hours. The DNA polymerase activity in Fraction AS decays to about 50% in 1 week at 4". Reactions described in the text were performed within 48 hours of a preparation of Fraction AS.
An increase in the total DNA polymerase activity was observed at each step in the partial purification, presumably due to the remnval of materials inhibiting DNA synthesis. Attempts to demonstrate DNA polymerase activity in discarded fractions of the sonicate were unsuccessful. Reaction Requirements-The requirements of Fractions CE and AS for DNA synthesis were assessed in a series of reactions in which certain components were omitted or replaced (Table I). Both fractions have an absolute requirement for Mg++. Fraction AS has an absolute requirement for added DNA template and the four deoxynucleoside triphosphates. In this experiment the concentration of [*H]dTTP was 4.6 pM. The activity of Fraction AS was found to be lo-to 20-fold greater when the concentration of each of the four deoxyribonudeotides in the reaction mixture was in the range of 75 to 150 pM (see Tables  II and III). These requirements also hold for Fraction CE but appear to be less stringent, perhaps because of DNA and nucleotides present in the crude extract. No radioactive nucleotides were rendered acid-precipitable in reactions lacking protein.
ATP was included at 10 times the concentration of the deoxy- The ordinate refers to incorporation of the radioactivity labeled nucleotide into acid-precipitable product.
centrations of NaCl and template DNA were below optimum levels for DNA synthesis.
Salt ~:$ct.s-rTl~e capacity of Fraction AS for DNA synthesis with native calf thymus DNA template was studied as a function of t,he concentration of MgCl~ and NaCl present in the reaction mixture.
The results shown in Fig. 3 indicate that 30 rnM NaCl and 3 mIlr MgC12 provided maximum activity.
Unless otherwise indicated, these concentrations were used in reactions described subsequently.
Concentrations of NaCl higher than 70 mM were found to inhibit DNA synthesis, in distinct contrast to the optimum concentration of NaCl, 150 mM, found by Meyer and Simpson for the rat liver mitochondrial DNA polymerase (10). The replacement of NaCl and MgC12 with other salts was studied and the results are presented in Table II. The monovalent cations, added at 32 mM, all enhance the activity of Fraction AS, but to different extents.
The replacement of Mg++ by Ca++, Sr++, or Ba++ resulted in a loss of activity. The results presented in Table III show that DNA synthesis in reactions containing n'lgClz is inhibited by the addition of CaC12 or SrBrz but not BaC12.
lncorporalion Ziinelics-The incorporation of [3H]dTTP into DNA by Fractions CE and AS was studied with native calf thymus DNA template on prolonged incubation at 37". Aliquots were withdrawn from reaction mixtures at selected times to determine the extent of incorporation. Fig. 4 shows the incorporation of t3H]dTTP by Fraction AS to be linear for at least 235 hours. The initial activity of Fraction CE was lower and the activity reached a plateau.
Other experiments of this nature have shown that nucleotide incorporation by Fraction CE reached a maximum level followed by a gradual loss of acidprecipitable [3H]thymidine. This is presumably due to degradative activities in Fraction CE which are greatly reduced in the purification leading to Fraction AS. Circular DNA templates for DNA synthesis are an important aspect of this report.
SV40 DNA was used to evaluate the dependence of incorporation kinetics on template DNA concentration.
Linear incorporation kinetics over 60 min with Fraction AS was observed with different concentrations of SV40 DNA template (Fig. 5). 4 double reciprocal plot of the rate of DNA synthesis at different template concentrations was linear and gave values of 2.6 nmoles of deoxyadenylate per hour and 8 pg per ml for the maximum rate of incorporation and the concentration of template DNA required for half the maximum rate, respectively.
The reaction mixtures contained 96 pg per ml of Fraction AS protein.
At high DNA template concentrations, the specific activity approached 27 units per mg of protein.
Assuming the product contains 29.5% deosyadenylate, in accord with the base composition of the SV40 template (19), the ratio of product DNA to initial template DNA can be calculated and used as an indication of the extent of synthesis.
In the above experiment (Fig. 5) we find that after 60 min at 37" the product to template ratio is 0.11, 0.26, and 0.48 for 40, 10, and 2.5 pg per ml of template, respectively. The reaction in which DNA synthesis was to proceed contained [3H]dATP, 46,000 cpm per pg of product DNA assumed to contain 29.5% deoxyadenylate.
The incorporation of [3H]dATP into product DNA and the fractional loss of acid-precipitable template DNA were determined in the course of 60 min at 37" (Table IV).
The template was quantitatively recoverable throughout the period of incubation. The extent of product DNA synthesis after 1 hour was approximately 10 times greater than the error associated with the template recovery analysis.
The endonucleolytic conversion of closed circular SV40 DNA to nicked circular or linear DNA, or both, was followed by the fluorometric assay (18). First order survival kinetics of SV40-I was observed during incubation with Fraction AS protein.
In 18 the experiment presented in Fig. 6, the half-life of SV40-I was 7 min. If the endonucleolytic process is considered to be a collection of randomly placed single-or double-strand scissions, or both, with all phosphodiester bonds equally susceptible, then the average number of breaks per molecule per min may be calculated by dividing In 2 by the measured half-life of closed circular DNA. For the experiment shown in Fig. 6, the rate was found to be 0.10 break per molecule per min for at least 20 min. We extrapolate that there would be an average of six breaks, single-or doublestrand scissions, per molecule after 1 hour.
No significant difference in the half-life of closed circular XV40 DNA at concentrations of 33 and 66 pg per ml was found, a result which suggests a linear dependence of nuclease activity on DNA concentration. In this system SV40 DNA sustains an average number of breaks per molecule per min which is independent of DNA concentration up to at least 66 pg per ml.
Two lines of evidence show that the endonuclease action is predominantly a single-strand scission process. A mixture of 32P-labeled SV40-I and SV40-II was incubated at 37" in two reactions, with and without Fraction AS. Sedimentation velocity analysis revealed that the DNA remained circular through both incubations, although the SV40-I was converted to SV40-II in the presence of Fraction AS. As seen in Fig. 7 the DNA incubated with Fraction AS sedimented as a sharp band corresponding to the position of SV40-II. Significant doublestrand scission activity would have generated linear fragments. The sedimentation profile, although still peaked at 14 to 16 S, would have been skewed back to the meniscus.
Electron microscope examination of a sample of initially closed circular SV40 DNA, incubated for 60 min with Fraction AS, revealed only 5% linear molecules (500 molecules counted).
One might argue that the linear molecules refiect fragmentation of the more extensively nicked molecules in the sample due to proximity of single-strand scissions on opposite strands.
Assuming random disposition of the breaks in the population of initially closed circular molecules, the frequency of molecules with a specified number of prior to centrifugation. The buoyant density of the SV40 [3LP]I)NA under the equilibrium conditions employed is known and the density gradient in the region of the marker DNA was calculated with data given by Vinograd and Hearst (20). The magnitude of the buoyant density differences between the DNA distributions was estimated from differences in the centers of gravity of the radioactive marker and product DNA distributions.
At the end of incubations the reaction mixtures were chilled was diluted to about 3 ml with 10 mM Tris (pH 7.5), and ethidium bromide was added to 250 pg per ml. Solid CsCl was added to raise the density to 1.55 g per ml. Centrifugation, fractionation, and analysis of radioactivity distributions were the same as described in the legend for Fig. 8. The more dense band in the 32P distribution corresponds to the closed circular SV40 DNA.
to 4" and EDTA was added to at least 10 mM. The samples were t'hen treated with SDS (20 mg per ml) for 10 min at room temperature, followed by addition of NaCl or CsCl to 1 M, chilling to 4", and removal of the precipitated protein and SDS. The extent of DNA synthesis was determined from the measured incorporation of 311-labeled nucleotide.
The product DNA forms a band at the same position in a pH 7.5 CsCl density gradient as the SV40 marker [32P]DNA (Fig. 8). In a CsCl density gradient containing ethidium bromide (Fig. 9)) closed circular and nicked circular SV40 DNA separate and the product DNA is found in one band, coincident with nicked circular marker DNA.
Product DNA samples were prepared using [W]dATP, 5bromodeoxyuridine 5'.triphosphate, and different concentrations of SV40 DNA template.
At higher concentrations of template the total amount of product synthesized by Fraction AS is greater, but the amount of product relative to the template is lower.
In the CsCl density gradients shown in Fig. 10 product was synthesized-in reactions containing -different concentrations of unlabeled. thvmidine-containing SV40 DNA template in order to obtain samples with different extents of synthesis relative to template.
After 60 min incubation at 37" the reactions were stopped by addition of ISDTA to 50 mM. The samples were then treated with SDS as described in the legend for Fig. 8. After dilution with 10 rnhf Tris (pH 7.5) solid CsCl was added to raise the density to 1.72 g per ml. About 1400 cpm of SV40 [32P]DNA were added to each sample. Centrifugation, fractionation, and analysis of radioactivity distributions were performed as described in the legend for Fig. S. The amount of product synthesized in each of the reactions was evaluated from the total acid-precipitable 3H radioactivity recovered in the density gradient and the specific activity of the [3H]dATP as described under "Methods " The extents of synthesis (product/ (product + template)) w&e 0.09 (lop), 0.21 (midcUe), and 0.25 (bottom). product DNA is shifted to higher densities as the fraction of product in the sample increases. So peak appears at the position expected (19, 21) for hybrid or fully 5-bromodeoxyuridinclabeled SV40 DNA, Fractions 18 to 20 and 10 to 12, respectively. The product DNA is apparently associated with template J)N,4 molecules, Fractions 8 to Il.
The measured density shifts are linearly correlated with the extent of synthesis, defined as the ratio of product I)NA to the sum of product and template I)NA (Fig. 11). The point representing no density shift is taken from an experiment in which the extent of synthesis was 0.10, but the reaction contained dTTP instead of 5bromodeoxyuridine 5'triphosphate.
The slope corresponds to 125 mg per ml density shift for unit extent of synthesis.
The primary sources of error in this experiment were the estimation of the density gradient and the extent of synthesis.
Combined, these amount to probably less than a 10% error in the value of the slope. The magnitude of the density shift relative to extent of synt.hesis will be discussed later.
Extent of Synthesis FIG. 11. Linear relation between extent of synthesis with 5-bromodeoxyuridine 5'-triphosphate and the buoyant density shift relative to thymidine-containing SV40 DNA. Buoyant density differences between the 3H-labeled product DNA samples and SV40 marker [32P]DNA were evaluated from the data presented in Fig. 10. The centers of gravity of the radioactivity distributions were determined in terms of fraction number. The density differences were then calculated from the separation of the 3H and 3zP band centers and the calculated density gradient in the region of the SV40 marker [32P]DNA (buoyant density of SV40 DNA, 1.694 g per ml; density gradient at 35,000 rpm, SW 501, rotor, 0.090 g per cm4 (20) ; radial displacement per fraction, 0.082 cm; density gradient, 7.4 mg per ml per fraction).
The extents of synthesis were given in Fig. 10. The point representing 0.00 extent of synthesis is taken from the experiment described in Fig.  8 in which dTTP was used in place of 5-bromodeoxyuridine 5'triphosphate.
In that case the 3H-labcled product DNA and 32P-labeled marker DNA distributions were coincident. The slope of the relation is 125 mg per ml density shift per unit extent of synthesis with the brominated nucleotidc.
The nature of the association of product and template USA was examined ill alkaline CsCl density gradients (Fig. 12). Under these conditions (pH 12.5) strand separation occurs for all but closed circular I>KA.
Closed circular SV40 [321']I>SA was added as a marker prior to banding in alkaline CsCl densit gradients.
The fraction oi' product 7)n'A in the IWO samples was 10%. In the case of thymidiue incorporation the % labeled ljroduct ~-as found to be 20 mg per ml liglltcr than the 32P-labeled marker.
With incorporation of S-bromodeoxyuridine, however, the separation of the product and marker distributions was only 1.2 mg per ml. The product distribution was skcmed to higher densities in this cast, but very little product was found in the region expected for fully 5-bromodeoxyuridinelabeled single-stranded I>KA.
The very small light shoulder in the 32P-labelecl marker DNA distributions indicates the limited extent of alkaline hydrolysis during centrifugation.
The result of this experiment suggests that most of the product I;KA is covalently bound to the template I)KA strands.
It was observed that omission of the SI?S treatment after incubation with the normal complement of four nucleotides led to the appearance of product DNA at slightly lower density than the malkcr DKA, suggesting that some protein remains bound to the IlKA in untreated samples at. high CsCl concentrations.
This effect was not observed when untreated samples were banded in gradients with ethidinm bromide or adjusted to alkaline pII. 13. Rand sedimentation analysis of 3H-labeled product DNA (O--O) in a self-generating CsCl density gradient (pH 7.5).
Product DNA was synthesized as described in the legend for Fig. 12 (lop).
SDS treatment, sedimentation, fractionation, and radioactivity analysis were performed as described in the legend for Fig. 7 still dependent on added tcmplatc DNA and may result from a noncovalent initiation process and subsequent dissociation from the template.
The amount of this material apparently depends in an unknown may on the preparation of Fraction AS. With some preparations, the slow band was not detected. When present, the relative amount increased to as much as 30% of the product DNA.
The relative amount appeared to increase when nucleotide or template concentrations were limiting. Sedimentation studies were also performed in alkaline selfgenerating CsCl density gradients.
The %labelcd product of DNA synthesis with SV40-I template sedimented with a predominant peak in the region expected for full-length linear single strands of SV40 1)Nh (Fig. 14). The distribution of product DKA was skewed back to the meniscus.
This sedimentation behavior is consistent with an average of two to three single-st,rand scissions in the circular template strands and with the covalent attachment of product DNA to template DNA strands.
Susceptibility of Product and Template to Adi on oj Escherichia coli konuclease I and II/---The specific reactions of IL?. coli csonucleases I and III make them suitable for the investigation of the structure of molecules containing both template and product DXA.
Exonuclcase I (20 units per ml) was added to one, esonuclease III (650 units lxx ml) to the second, while the third remained unaltered.
The esonuclease levels were in excess of those required to completely digest the l>NA in the csperiments, if susceptible.
The course of the reaction in the three cases was determined by removing aliquots for analysis of acid-yrecipitable product and template DNA as a function of time (Fig. 15) the presence of esonuclease I, but at a somewhat slower rate. Both template and product DNA were degraded by exonuclease III.
The latter result indicates that product, as well as template, DNA are in duplex regions.
The site(s) of action of the exonuclease III might be either the single-strand scissions produced by endonucleolytic activity in Fraction AS or the 3'-OH end of a growing product DNA strand.
The product DNA isolated and purified from standard incorporation reaction was also treated with exonuclease I. The DNA was isolated in CsCl-ethidium bromide density gradients following treatment with SDS (20 mg per ml) as previously described.
After removing the dye, the DNA containing 10% product was dialyzed against 10 mM Tris, 1 mM EDTA (pH 7.5). Exonuclease I was diluted to 300 units per ml in 10 my Tris, 0.18 M (NH&SO+ 0.1% bovine serum albumin (pH 7.5). Ten microliters of 1 ht MgCIZ and 50 ~1 of either the exonuclease I solution or its dilution buffer were added to 0.5 ml of the DNA sample.
After 30 min at 37", the reactions were stopped by chilling and adding 100 ~1 of 0.1 M EDTA (pH 7.5). Preparative velocity gradients showed no change in the sedimentation behavior of the product due to the action of exonuclease I. In another experiment the DNA was dialyzed against 50 mM NaCl, 10 mM Tris, 1 mM EDTA (pH 7.5). The sample was denatured by heating at 90" for 5 min, then quickly chilled to 0". Exonuclease I was diluted to 60 units per ml in 10 mM Tris, 100 mM MgCle, 0.05% bovine serum albumin (pH 7.5). On incubation at 37" of a solution containing equal volumes of the denatured DNA and exonuclease I solutions, more than 95% of the 3Hlabeled product was solubilized in 5 min. The product DNA is, therefore, rendered susceptible to the action of exonuclease I only after denaturation.
Electron il4icroscopy of Template-Product DNA-Basic protein film electron microscopy was used to visualize molecules containing regions of product and template DNA.
The formamide system (14)  have been due to persistcncc of contaminating protein in the samples, despite treatment with SDS and banding in CsCl density gradients.
The formamide may have dissociated protein-DNA complexes.
In addition, the formamide technique allows discrimination between single-and double-stranded wgiotis ill I )NX molccnlcs. Single-st'randed l>?lr~~ appears wit'h somewhat wtluwd cont,rast from background and a more kinked cwltour th:ul the: thicker and smoother duplex I)SA regions. ,l'h(l (~lwtrotl micrographs presented in Fig. 16 show rcpresenta-tiw molecules in a sample containing 187; product DSA in a l-hour incubation of circular SV40 template with Fraction AS. The most apparent structures are circular duplex, SV40-length molecules with sin&stranded branches. About half of the circular duplex molecules have au easily visualized singlestranded branch.
Molecules with more than one branch wcrc not uncommon.
Most of the branches observed were of the size seen in the molecules of Fig. 16A, but occasionally much larger branches were found as in the molcculcs of Fig. 16, B and C. The foregoing results, togcthcr with the conclusions drawn from the cent,rifugation and exonuclease susceptibility stud&:, indicate the displacement of a template I>lY,4 strand in the course of product I)XA synthesis.
'J'hc length of the displaced template DSA sllould be the samt as the associated region of product DXA, since no dcgradatiou of template rould be demonstrated in the course of product USA synthesis.
The more frequent smaller branches wcrc about 27; by mass of an ST'40 molecule. The longer branches shown in Fig. 16, B and C, wcrc 205; and 107; of SV40 molecular weight, respectively.
:1n 185; rstent of sytlthesls would lead to an 18 o/o increase in the average molecular weight of the SV40 molcculcs in the sample.
The frequcnc> of molecules with long branches was too small to iigiiificantly affect the average molecular weight.
The molwulcs with short branches could account for an illcrease in the average molecular weight of only 1 to 2c/c. In another sample of DNA containing 6yG product, the frequency of circular duplex molecules with small sin&-stranded branches was much lower, about 2 to 37;;;, which could account for only 1 c/; of the expected increase in average molecular weight of the SV40 DS,4. 7'1~ rather low frequency of bran&d molecules and the small size of most of the branches suggest that product LDIYA synthesis al,qo gives rise to short displaced tcmplatc branches, too small to bc +ualized. Such a result' is compatible with initiation of DNA .qynthesis at many of the single-strand wissiorls introduced during the incubation with Fraction AS. l'wvious experiments indicated that approximately 10 &&-stranded srissions per molwule occurred during a 60.mill incubatioll.
Furthermore, the soliiblc nuclear DS;i polymerase of II&a cells has I'rom three to fix,<> time5 greater activity with heat-dcnaturod than with uatiw USI trmplates (24), in contrast to tile native template 1)refcwllw or tllr> l)SA polymerax activity in Fraction ;\S. ll'prlls et nl. (24) ~c'rc also able to recover the nuclear enzyme from DEAE-cellulose columns, whereas all of the activity in Fraction AS was lost during DEAE -cellulose chromatography.
We therefore regard the DNA polymerase activity in Fraction AS to be of mitochondrial origin.
The preparations of rat liver mitochondrial DNA polymerase by Meyer and Simpson (10) and Kalf and Ch'ih (9) differed somewhat in their properties.
These differences appear to be due to t,he different levels of purification. Meyer and Simpson (25) noted that sonication yielded higher activity of DNA polymerase than their preferred method of grinding frozen mitochondria with alumina.
However, they were unable to recover the former activity from ion exchange columns successfully employed in the purification of the latter activity.
Our attempts to utilize ion exchange columns in purification of sonic extracted DNA polymerase activity from HeLa cell mitochondria were likewise unsuccessful.
It may very well be, as discussed in the following paper (21), that mitochondria contain two DNA polymerases and that the activity we have described here is functionally distinct from that described by the above investigators.
Under conditions optimized with respect to salt, Mg++, and nucleotide concentrations, Fraction AS can catalyze extensive synthesis of DNA without degradation of template. Buoyant equilibrium and sedimentation velocity studies led to the conclusion that most of the product was synthesized in a covalent extension of template DNA strands.
Initiation of DNA synthesis in this system is thus template-primed.
The initiation sites appear to be the single-strand scissions introduced by the limited endonucleolytic activity in Fraction AS. Base sequence fidelity is suggested by the requirement of all four deoxynucleotides or appropriate analogues for DNA synthesis.
To maintain fidelity of base sequence without degrading template, the template strand ahead of the point of polymerization must be displaced to allow base pairing interaction of added nucleotides with the opposing region of the complementary template strand. Electron microscopy provided evidence of such single-stranded regions in circular duplex molecules after DNA synthesis.
The assignment of these single-stranded branches as displaced regions of iemplate DNA is supported by the results of studies of the sensitivity of template and product DNA in these molecules to the action of E. coli exonucleases I and III.
The results characterize a mode of action consistent with the "rolling circle" model of Gilbert and Dressler (26). The model in Fig. 17a has a single-stranded tail with a length of about 20% of that of the duplex circle. After denaturation the product was rendered susceptible to digestion by exonuclease I. The product, therefore, does not contain short self-complementary regions at the growing point, such as in Fig. 17b. After an incubation long enough to achieve 10 v0 synthesis, several nicks would have been introduced in most molecules. Electron microscope examination showed that the lengths and frequency of observed single-stranded branches were too small to account for localization of product in one or perhaps two sites per molecule.
The formation of a pinwheel structure (Fig. 17~) accounts for all of the results in this work.
The buoyant density of a template-product complex representing the limit of unit extent of synthesis (product/(product + template)), with 5-bromodeoxyuridine 5'-triphosphate instead of dTTP, on an SV40 DNA template is the same as the buoyant density of 5-bromodeoxyuridine-labeled single-stranded SV40 DSA.
The density difference between such a species and thymidine-containing duplex SV40 DNA was calculated using the base composition of SV40 DNA, 41 mole per cent dG + dC (19), the 0.20 g per ml density difference between the synthetic polymers poly(dA-dT) and poly(dA-dBrU) (27), and the density difference between duplex and single-stranded DNA (28). We estimate the error associated with these values to be at least &I mole per cent dG + dC, ho.01 g per ml, and +1 mg per ml, respectively.
The density shift expected for the mode of DNA synthesis described is 126 to 142 mg per ml for unit extent of synthesis.
The value determined from the slope in Fig. 11 was 125 mg per ml.
Annealing of complementary displaced strands in the samples under investigation would have maximally lowered the slope by 16 mg per ml. The slope would represent an overestimate of the density shift per unit extent of synthesis if there were large variations in the amount of product DNA among the population of template molecules, leading to different mass distributions of product and template in the density gradients.
The assumption of coincident product and template distributions is supported by the electron microscopy studies which suggest that product DNA was localized in several small regions on most of the template molecules. Molecules with long strctchcs of product DNA synthesis were rare.
A similar mechanism of template priming of DNA synthesis and subsequent template strand displacement was observed in recent studies with the E. coli DNA polymerase I and nicked circular duplex PM2 DNA template (29). Two important distinctions in mode of action should be noted. The HeLa mitochondrial system (Fraction AS) does not show the 5'exonucleolytic activity of the E. coli DNA polymcrasc and does not appear to hairpin in the course of DNA synthesis.
Further studies with Fraction AS involved the USC of the natural template for the system, HeLa mitochondrial DNA. In addition to evidence for base sequence fidelity, wc have also observed a preference for the initiation of DNA synthesis on the