In Vitro Polyoma DNA Synthesis: Characterization ... - Journal of Virology

Vol. 13, No. 1 Printed in U.S.A.

JOURNAL OF VIROLOGY, Jan. 1974, p. 125-139 Copyright 0 1974 American Society for Microbiology

In Vitro Polyoma DNA Synthesis: Characterization of a System from Infected 3T3 Cells TONY HUNTER AND BERTOLD FRANCKE The Armand Hammer Center for Cancer Biology, The Salk Institute, San Diego, California 92112

Received for publication 9 July 1973

A lysate from hypotonically swollen polyoma-infected BALB/3T3 cells incorporated labeled deoxynucleotide triphosphates into both viral and cellular DNAs. The incorporation was stimulated by the presence of ATP, deoxynucleotide triphosphates, thiols, and magnesium ions. Strong inhibition of incorporation was observed with thiol reagents and arabinosyl nucleotide triphosphates. The rate of in vitro synthesis increased with the temperature of incubation as expected. Incorporation into cellular DNA for up to 2 h was observed in lysates from virus-infected and serum-stimulated cells but not from resting cells. Synthesis in the system, therefore, appeared to reflect the physiological state of the cells before preparation of the lysate. Incorporation into viral DNA stopped far sooner than that into cellular DNA. During the initial phase of the in vitro incubation, incorporation occurred into viral replicative intermediates (RI). These RIs had identical properties to those isolated after in vivo pulse labeling and a substantial proportion of them was matured to form I DNA at later times in the incubation through all the stages known to occur in vivo. Density labeling of the in vitro product showed that practically all of the RIs pre-existing in the infected cell took part in the in vitro reaction. Analysis of DNA labeled in vitro in the presence of 5-bromodeoxyuridine triphosphate showed that synthesis occurred on Rls at all stages of replication and that the progeny strands were elongated by up to 80% of unit viral DNA length. Pre-existing RIs, pulse labeled in vivo, showed evidence of a pool at a late stage of replication which required elongation of their progeny strands by approximately 25% during conversion to form I molecules. From density-labeling experiments, we were also able to show that viral DNA synthesis in vitro was semiconservative. The major reason for cessation of viral DNA synthesis in vitro was the very limited ability of the lysate to initiate new rounds of viral DNA synthesis. The synthesis of both viral and cellular DNAs in eukaryotic cells has been extensively studied with the aid of cultured cell lines. For an understanding of the detailed mechanism and enzymology of DNA synthesis, however, there are obviously serious limitations to the use of whole cells. It is clear that, in the case of Escherichia coli, the development of highly active in vitro DNA-synthesizing systems has led to a fuller understanding of the detailed mechanism of DNA synthesis (23, 26, 33). Isolated nuclei from mammalian cells have been used by several groups of workers to study different aspects of eukaryotic DNA synthesis in vitro (12, 19). By using such systems it was possible to demonstrate the involvement of cytoplasmic factors in DNA synthesis (17) and also a correlation between the in vitro activity and the physiological state of the cells before the isolation of the nuclei (19). There are many 125

difficulties inherent in these studies, such as the poorly defined nature of the synthetic product and the problems of characterizing the highmolecular-weight DNA involved. For these reasons, it seemed best to use a simple mammalian DNA virus, known to replicate in the nucleus, as a tool for the study of eukaryotic DNA synthesis in vitro. Polyoma and SV40 viruses are ideal candidates containing well-characterized DNA molecules with replication cycles utilizing, for the most part, the enzymatic and structural machinery involved in host chromosomal DNA synthesis. The replication cycle of the small, doublestranded DNA viruses, polyoma and SV40, has been established in detail by in vivo studies on infected cells. From analysis of temperaturesensitive mutants, it is clear that, after the onset of viral DNA synthesis, continued viral replication depends on at least one known viral

126

HUNTER AND FRANCKE

function (11, 29). This function appears to be involved in the initiation of each single round of replication (a single round is defined as going from a closed-circular double-stranded molecule to two daughter molecules). For SV40, replication starts at a unique origin (24) and proceeds bi-directionally (1, 8). The synthesis of new strands occurs at least in part discontinuously (10, 20; B. Francke and T. Hunter, unpublished observations), whereas the parental template strands remain covalently closed (2, 15, 27). The existence of partially supercoiled replicating molecules demonstrates that the superhelical turns present in the parental form I are gradually removed as replication proceeds (18). How the template strands separate once the progeny strands have reached unit length is at present unknown. The existence of short-lived form II molecules containing the progeny strands in a uniquely nicked form, as kinetic precursors to closed form I molecules,

has been demonstrated (9). To complement the availability of a number of DNA polymerases from eukaryotic cells (5, 32), the development of an efficient in vitro system for the replication of these viral DNAs would be desirable to study a variety of still open questions regarding the detailed mechanism and enzymology of the process. In particular, a system that efficiently reinitiates new rounds of replication could serve as an in vitro complementation assay for the viral function required for this step by using viral mutants temperature sensitive in initiation (11). An in vitro system using purified nuclei from polyoma-infected 3T6 cells has been described (21, 34). This system is capable of a rather limited synthesis of viral DNA in a semiconservative fashion on pre-existing replicative intermediates. It has been used to demonstrate the involvement of RNA-linked DNA chains in viral DNA synthesis (20). This communication deals with the preparation and general properties of an in vitro system for polyoma DNA synthesis consisting of a concentrated lysate from hypotonically treated polyoma-infected 3T3 cells. This unfractionated lysate also continues synthesis in a semiconservative fashion on pre-existing replicative intermediates, but it is capable of completing most of them to mature form I DNA. We have examined the type of discontinuous synthesis carried out by the system, and this work will be published elsewhere. We have also demonstrated the involvement of RNA-linked DNA chains in the synthesis of viral DNA (14). Despite the efficiency of completion of preexisting replicative forms, the amount of initia-

J. VIROL.

tion occurring in the system is extremely limited. Since most, if not all, synthetic events during one round of replication of viral DNA are accomplished by cellular functions, results obtained with viral DNA synthesis are expected to be of major significance in understanding the mechanism and enzymology of mammalian chromosomal DNA replication. MATERIALS AND METHODS Virus stock. The temperature-sensitive mutant of polyoma, ts1260, was used in all experiments. The virus was propagated by passage at 32 C on primary baby mouse kidney cells at low multiplicity (less than 0.1 PFU/cell). When extensive cytopathic effects were observed, the culture was frozen and thawed three times, clarified by low-speed centrifugation, and stored at -20 C. Ts1260 is a late mutant of polyoma (W. Eckhart, unpublished observations) as defined by complementation tests (7). It behaves like wild-type virus with respect to viral DNA synthesis. Ts1260 has been used for these studies because it yields very high-titer stocks (109 to 2 x 10' PFU/ml) by the procedure described and gives consistently more active lysates than does wild-type virus as judged by the criteria in Results. Growth and infection of BALB/3T3 cells. BALB/3T3 cells (obtained from S. Aaronson) were transferred at low cell densities (about 105 cells per 9-cm dish) every 3 to 4 days in Dulbecco-modified Eagle medium containing 10% calf serum. Cells were infected when they reached 60 to 90% confluency (2 x 10 to 3 x 106 cells per 9-cm plate). The medium was removed, and the cells were washed with 5 ml of Tris-buffered saline. Virus (0.5 ml) diluted to 2 x 108 PFU/ml in Tris-buffered saline was added to each 9-cm dish and allowed to absorb at 37 C for 1 h with occasional shaking (the multiplicity of infection was approximately 20). Fresh medium containing 5% (for confluent cultures) or 10% (for subconfluent cultures) calf serum was added to the plates, which were then incubated at 32 C for 36 h. Two hours before the preparation of the lysate, the plates were shifted to 39 C. For reasons not understood, this resulted in an increase in the incorporation of ['HJTTP into the Hirt supernatant of up to 50%. The same was true for wild type-infected cell lysates. Preparation of the lysate for in vitro DNA synthesis. All procedures were carried out at 4 C. The medium was removed, and the infected cells were washed once with cold Tris-buffered saline. Hypotonic buffer (5 ml) was then added (20 mM N-2hydroxyethylpiperazine-N'-2'-ethane sulfonic acid [HEPES], pH 7.8, 1 mM MgCl,, 0.5 mM CaCl2, 1 mM DTT), and the cells were left to swell for 5 min. The hypotonic buffer was aspired, and the plates were left to drain till no more buffer could be removed. The swollen cells were scraped off the plate with a rubber policeman, yielding 0.1 to 0.15 ml of lysate per dish containing 2 x 108 to 3 x 10' cells. Isotonicity was restored by addition of 0.1 volume of 2.5 M sucrose,

VOL. 13, 1974

IN VITRO POLYOMA DNA SYNTHESIS

200 mM HEPES, pH 7.8, and 500 mM KCl. There were no detectable whole cells present in the lysate, which appeared by microscopy to consist of clumps of cytoplasm with many nuclei embedded in each clump. In vitro incubation for DNA synthesis. Including the concentrations of components in the hypotonic and isotonic buffers, the following final added concentrations were present in the in vitro incubation: 1 mM ATP, 100 pM CTP, GTP, UTP, dATP, dCTP, dGTP, and dTlP (the labeled deoxynucleotide triphosphate was present at a final concentration of 10 pM), 5 mM creatine phosphate, creatin phosphokinase, 1 mM dithiothreitol, 1 mM ethyleneglycol-bis-(B-aminoethyl ether)-N,N'-tetra-acetate, 40 mM HEPES, pH 7.8, 100 mM KCI, 5 mM MgCl,, 1 mM MnCl,, and 0.5 mM CaCl,. The lysate occupied nine-tenths of the incubation volume, with additions making up the other one-tenth. Variations on this incubation mixture will be cited for individual experiments. The reaction was terminated by a modified Hirt extraction (13). The incubation was diluted 5- to 20-fold with 20 mM HEPES, pH 7.8, and 10 mM EDTA, and sodium dodecyl sulfate was added to a final concentration of 2%. After 10 min at 32 C, the mixture was made 1 M in NaCl and left at 4 C for 4 h or longer. After centrifugation for 60 min at 10,000 rpm in a Sorvall RC2B centrifuge at 0 C, the pellet containing the majority of the cellular DNA was washed in 70% ethanol and 0.1 SSC and solubilized in 1 M NaOH at 70 C for 1 h. The radioactivity in the Hirt pellet and the Hirt supernatant was determined as described below. (Hirt supernatant and Hirt pellet refer to DNA extracted and fractionated by Hirt procedure [13]. DNA in the Hirt supernatant is of low molecular weight and primarily viral. DNA in the Hirt pellet is of high molecular weight and primarily cellular.) Purification of viral DNA. Since the Hirt supernatant fraction contained varying amounts of cellular DNA, viral DNA was routinely purified by sedimenting the Hirt supernatant through a 34-ml 5 to 20% linear sucrose gradient in 1 M NaCl, 10 mM Tris-chloride, pH 8.1, and 1 mM EDTA in a Beckman SW27 rotor for 12 h at 27,000 rpm at 5 C. Fractions (1.2 ml) were collected and sampled for determination of radioactivity. The fractions containing the viral DNA were pooled and concentrated by ethanol precipitation (2 volumes of ethanol, - 20 C, overnight, followed by centrifugation for 60 min at 10,000 rpm at 0 C). The pellet was resuspended in 0.2 M NaCl, 10 mM Tris-chloride, pH 8.1, and 5 mM EDTA and extracted with phenol saturated with the same buffer. The DNA was again precipitated with ethanol and stored at - 20 C for further analysis. Centrifugation techniques. (i) Alkaline sucrose gradient sedimentation: 3.6 ml of linear 5 to 20% sucrose gradients in 10 mM EDTA and 0.25 M NaOH were overlayered with approximately 200 pliters of sample and centrifuged for 100 min at 55,000 rpm at 5 C in a Beckman SW56 rotor. (ii) Equilibrium centrifugation in CsCl gradients containing ethidium bromide: the DNA sample in a total volume of 7.0 ml in 10 mM Tris-chloride, pH 8.1, 5 mM EDTA, and 0.2 M NaCl, containing 50 pg of native calf thymus DNA

127

and 500 pg of ethidium bromide, was mixed with 6.62 g of CsCl and centrifuged at 20 C in a Beckman 50 angle rotor for 24 h at 50,000 rpm followed by 48 h at 35,000 rpm. (iii) Equilibrium centrifugation at density-labeled DNA in CsCl: the DNA sample in a total volume of 2.0 ml in 10 mM Tris-chloride, pH 8.1, 5 mM EDTA, and 0.2 M NaCl, containing 20 pg of native calf thymus DNA, was mixed with 2.8 g of CsCl and centrifuged for 60 h at 35,000 rpm at 20 C in a Beckman SW56 rotor. (iv) Equilibrium centrifugation in alkaline Cs,SO4: the DNA sample in a total volume of 2.0 ml in 10 mM Tris-chloride, pH 8.1, 5 mM EDTA, and 0.2 M NaCl, containing 20 pg of native calf thymus DNA, was mixed with 1.2 g of Cs,SO,, made 0.1 M in NaOH, and centrifuged for 60 h at 20 C in a Beckman SW56 rotor. Form I DNA was nicked before centrifugation by boiling for 4 min in 10 mM Tris-chloride, pH 8.1, and 5 mM EDTA. Preparation of viral marker DNAs. Dishes (9 cm) of BALB/3T3 cells were infected as described with polyoma wild-type virus and incubated at 37 C. After 12 h, the cells were washed with Tris-buffered saline and overlayered with 5 ml of medium containing the desired labeled precursor. ['H]dThd (100 pCi/ml) and [I4C]dThd (0.05 pCi/ml) were used for 'H- and "4C-labeling. For "P-labeling, Dulbecco-modified Eagle medium lacking phosphate was used, supplemented with 2 x 10-5 M NaH2PO4, 5% dialyzed calf serum, and carrier-free "2P (100 uCi/ml). For density labeling, the following additions were made: cytidine (12 pg/ml), fluorodeoxyuridine (5 pg/ml), and bromodeoxyuridine (20 pg/ml). After 24 to 36 h, standard Hirt extractions were performed, and the supernatants were extracted with phenol. The viral DNA was purified by neutral sucrose gradient centrifugation followed by equilibrium centrifugation in ethidium bromide-CsCl (see above). The density of unsubstituted "2P-marker polyoma DNA in neutral CsCl equilibrium gradients was 1.703, and that of bromouracil-substituted DNA was 1.804. According to the arguments of Magnusson et al. (21), this corresponds to polyoma DNA in which 97% of the thymine residues have been replaced by bromouracil. Estimation of radioactivity. Samples containing labeled nucleotide triphosphates were precipitated with 5% trichloroacetic acid containing 0.1 M sodium pyrophosphate and filtered onto Whatman GF/C glass fiber filters, which were washed with 2 x 5 ml pyrophosphate containing trichloroacetic acid, 1 x 5 ml 5% trichloroacetic acid, and finally ethanol. The filters were dried and counted in a toluene scintillator containing 5 g of 2,5-diphenyloxazole in a Beckman LS200 scintillation counter. Samples not containi( labeled nucleotide triphosphates were precipitatg& with 5% trichloroacetic acid and then treated as above. Alkaline sucrose gradients, ethidium bromide-CsCl gradients, neutral CsCl gradients, and alkaline CsSO4 gradients were all collected dropwise directly onto Whatman GF/C glass fiber filters which were dried and thoroughly washed in 5% trichloroacetic acid and ethanol before being counted. 'H-counting efficiency was approximately 20%, "4C-efficiency was 50%; "2P-efficiency was 90%. "4C-overlap into the 'H-channel was approximately 16%, and "2P-overlap into the "4C-channel was 20%. These overlaps were

128

HUNTER AND FRANCKE

calculated and subtracted together with the background in all cases. There was less than 1% overlap of 32p into the 3H-channel which was not corrected for. Materials. Materials used were as follows: [methyl'H]thymidine (53 Ci/mmol), [2-"4C]thymidine (52 mCi/mmol), and carrier-free [32Piphosphoric acid from New England Nuclear Corp.; [methyl-'HJTTP (13.4 Ci/mmol), [8-_H]dATP (16.8 Ci/mmol), [53H]dCTP (22.6 Ci/mmol), and mitomycin C from Schwarz/Mann; [8-'H]dGTP (9.8 Ci/mmol) from Amersham Searle; dithiothreitol, hydroxyurea, and N-ethylmaleimide from Calbiochem; and 9-p-Darabinofuranosyl triphosphate (araATP) and 1-0-Darabinofuranosyl triphosphate (ara CTP) from Terramarine Bioresearch, La Jolla, Calif. a-'2P-labeled deoxynucleotide triphosphates were synthesized according to the method of C. K. Biebricher (personal communication). 5-Bromodeoxyuridine triphosphate (BrdUTTP) was synthesized according to a modification of the method of Chamberlin and Berg (4). The rifamycin analogue AF05 was kindly donated by Luigi Silvestri, Lepetit, Milan. RESULTS

Optimization of in vitro DNA synthesis. The in vitro system consisted of a concentrated cell lysate which was obtained from BALB/3T3 cells infected with ts1260 (ts1260 is a temperature-sensitive mutant of polyoma in a late class which produces normal amounts of viral DNA at the nonpermissive temperature) as described in Materials and Methods. Results obtained with lysates from wild-type infected cells were essentially identical to those obtained for ts1260 (data not shown). After in vitro synthesis had occurred, DNA was extracted from the lysates by a modification of the method of Hirt (13). The majority of the DNA found in the Hirt supernatant was determined to be viral in origin by sucrose gradient analysis (see Fig. 2a-d). Therefore, incorporation of [3H]ITTP into acidinsoluble material in the Hirt supernatant was used to optimize the system for viral DNA synthesis. The system showed a broad optimum for added Mg2+ ions between 3 and 6 mM final concentration, whereas Mn2+ ions at a final concentration of 1 mM were found to stimulate by up to 10% at optimal Mg2+ concentrations. Concentrations of Mg2+ above 7 mM were inhibitory. Routinely, the final concentrations of MgCl2 and MnCl2 in the incubations were 5 and 1 mM, respectively. In the presence of an excess of EDTA over Mg2+, no viral DNA synthesis was observed (Table 1). A broad optimum between 60 and 90 mM was observed for the monovalent cation. Routinely, a final added concentration of 75 mM KCI was used. The presence of Ca2+ ions during the preparation of the lysate led to more prolonged incorporation, giving a stimulation of up to 30%. The

J. VIROL.

absence of Ca2+ ions during the preparation resulted in clumping of the lysate, which was then extremely difficult to handle quantitatively. A slight molar excess of EGTA was added to the in vitro incubation to chelate Ca2+, and this slightly prolonged incorporation (Table

1).

In vitro DNA synthesis was extremely sensitive to thiol reagents (e.g., p-chloromercuribenzoic acid [PCMB ] and N-ethylmaleimide [NEM 1) (Table 1). For this reason, the lysate was maintained in 1 mM dithiothreitol at all times. In vitro DNA synthesis was not totally dependent upon added deoxynucleotide triphosphates, because there are sizeable pools of all four deoxynucleotide triphosphates present in the lysate (see below). To obtain maximal incorporation of [3H]TTP, the minimal added concentrations of deoxynucleotide triphosphates reTABLE 1. Properties of the system Incubation conditions

Control

Omit ATP .............................. Omit ATP, Creatine phosphate, and creatine phosphokinase ........ ........ Omit EGTA ............................. Omit Ca2+ in preparation of the lysate ..... + 10 mM EDTA ......................... + Mitomycin C (130 ,g/ml) ........ ...... + alpha-amanitin (33 Ag/ml) ....... ...... + Rifamycin AF05 (100 ;g/ml) ....... .... + Cycloheximide (50 jg/ml) ........ ...... + Pancreatic DNase (50 ig/ml) ....... .... + PCMB (5 mM) ........................ + NEM (1 mM) .........................

80

+NEM(2mM) ......................... + NEM (5 mM) ......................... + Hydroxyurea (10 mM) ......... ........ + araCTP (10 LM) .............. ......... + araCTP (100,uM) ............. ......... + araATP (10MM) .............. .........

+araATP(100L M) ......................

31 96 60 2 50 99 72 99 85 1 70 3 1 78 25 5 23 6

aIn each case, incubations were set up with 50

jliters of lysate from ts1260-infected 3T3 cells. Incu-

bations were for 60 min at 32 C. Since not all of the incubations were done at the same time, the results are expressed as the percentage of the control incorporation of [3H]TTP into acid-precipitable radioactivity in the Hirt supernatant. At the concentrations used, alpha-amanitin (greater than 90% inhibition in isolated nuclei) and rifamycin AF05 (30% inhibition in whole cells) were shown to inhibit RNA synthesis. At 10 mM, hydroxyurea inhibited DNA synthesis in whole cells more than 95% after 15 min. Cycloheximide at 50 gg/ml inhibited protein synthesis in whole cells by more than 99%, and it also inhibited [1"CJamino acid incorporation in the lysate to the same extent. The pancreatic DNase was shown to be active against purified labeled viral DNA.

VOL. 13, 1974

129


quired were between 10 and 20 iuM. Concentrations of deoxynucleotide triphosphates up to 500 .uM could be included in the incubation without inhibiting incorporation. The standard reaction conditions contained the labeling deoxynucleotide triphosphate at 10 uM and the other three at 100 MM. By using the four labeled deoxynucleotide triphosphates separately, we could show that all four were incorporated into viral DNA as judged by sucrose gradient analysis. Almost total inhibition of viral DNA synthesis was observed with either of the two arabinosyl nucleotide analogues araCTP or araATP (Table 1). These effects could be reversed by the addition of dCTP or dATP, respectively. Therefore, the effective omission of these two deoxynucleotide triphosphates is sufficient to block DNA synthesis, which indicates that the in vitro DNA synthesis is truly dependent upon the presence of all four deoxynucleotide triphosphates. The requirement for ATP was not striking when an energy-generating system was included in the incubation (Table 1). The omission of both ATP and the energy-generating system led to a marked diminution of incorporation (Table 1). The lack of a strict dependence on ATP is probably attributable to the presence of a sizeable pool of adenine nucleotides in the lysate. In dialyzed lysates, long-term synthesis was considerably more dependent upon ATP and also showed a slight dependence upon the ribonucleotide triphosphates other than ATP. For this reason, they were included in the standard incubation. The main criteria for obtaining an active in vitro system with respect to both maximal [3H]T'IP incorporation and maximal conversion to form I DNA were (i) multiplicities of infection greater than 20; (ii) isolation of the system during the log phase of increase in viral DNA synthesis, before noticeable cytopathic effects occurred; and (iii) the use of cells which showed good density-dependent inhibition. Kinetics of in vitro DNA synthesis. A typical time course of incorporation of [3H]T'TP into viral DNA is shown in Fig. la. Synthesis. continued for up to 50 min at 32 C. The duration of incorporation at a linear rate varied from 10 to 30 min in different experiments. No appreciable acid-precipitable radioactivity was found in the Hirt supernatant when lysates from resting or mock-infected cells were used (Fig. la). In Fig. lb, the incorporation into the Hirt pellet is shown. Resting cell lysates showed no incorporation of [3H]TTP. From this it is clear that the preparation of the lysate does not itself trigger any form of DNA synthesis. Mock-

infected cell lysates incorporated [3H]TTP into cellular DNA linearly for up to 60 min. This synthesis reflects the burst of cellular DNA synthesis initiated by the serum change given at the time of mock infection. The virus-infected cell lysates showed an even greater incorporation of [3H]'TTP into cellular DNA. This synthesis was the result of the rather synchronous wave of cellular DNA synthesis induced by viral infection. Linear synthesis of cellular DNA was often maintained for up to 90 min at 32 C. For this reason, infected cell lysates would seem to offer a good system for the study of cellular DNA synthesis in vitro. The greater inherent stability of cellular DNA synthesis to incubation in vitro suggests that continued viral DNA synthesis is dependent upon a process which is not required or is required to a lesser extent for

10

0' x

x

0.

a.

z

z

A.

U

MINUTES

FIG. 1. Time courses and temperature dependence of incorporation of [3H]TTP into viral and cellular DNA in vitro. Three types of lysate were prepared as described in Materials and Methods from the following sets of cells: (i) 3T3 cells infected with ts1260 for 36 h at 32 C followed by 2 h at 39 C; (ii) 3T3 cells (60 to 80%o confluent) stimulated with fresh medium containing 10%o calf serum for 24 h at 37 C; (iii) 3T3 cells allowed to come to rest for 2 days at 37 C after reaching confluency. Incubations (50 gliters) were set up with ['H]TTP (10 AsM, 20 pCi/ml) and incubated at the temperatures shown for the times indicated. Acid-precipitable radioactivity incorporated into the Hirt supernatant and the Hirt pellet was analyzed separately. (a) Hirt supernatant. (b) Hirt pellet. 0, Infected cells; 0, serum-stimulated cells. 0, Resting cells. Panels (c) and (d) are data derived exclusively from the infected cell lysate. (c) Hirt supernatant. (d) Hirt pellet. O, 20 C; 0,28 C. 0,32 C; 039 C.

130

HUNTER AND FRANCKE

cellular DNA synthesis. The ratio of viral to cellular DNA synthesis increased with the time after infection and multiplicity of infection and decreased with the passage level of the cells. The temperature dependence of in vitro DNA synthesis is shown in Fig. lc and d. At 39 C the final level of synthesis was lower than that at 32 C. Indeed, at late times at 39 C there was often a decrease in net incorporation which may have resulted from nucleolytic activity but was not investigated further. At 20 C, maturation to form I viral DNA was very inefficient. The initial rates of synthesis of both viral and cellular DNAs showed the expected temperature dependence. Arrhenius plots derived from the initial rates of synthesis are reasonably linear (not shown). The slopes of the plots are the same for viral and cellular DNA synthesis, suggesting a common mechanism for DNA chain elongation. The activation energy for both processes was about -20 Cal. Determination of the deoxynucleotide triphosphate pools in the lysate. The method used to prepare the lysate involved no step in which the low-molecular-weight compounds are removed. Thus, the lysate is likely to contain most of the pools of deoxynucleotides and ribonucleotides pre-existing in the cells at the time of isolation of the lysate. Without knowledge of the final specific activity of the labeled deoxynucleotide triphosphates in the incubation, measurement of the absolute amount of DNA synthesized in vitro is impossible. For this reason, we determined the pool sizes for all four deoxynucleotide triphosphates in two different ways, both involving isotope dilution (Table 2). There were significant differences in the levels of the deoxynucleotide triphosphates. Dialysis of the lysate did not dramatically lower the pools of deoxynucleotide triphosphates (Table 2). The pool sizes determined here are in reasonable agreement with those determined for growing 3T3 cells (31), although the pools may have been expanded to a slightly greater extent than in growing cells due to the synchrony of DNA synthesis (28). By using the value obtained for the TTP pool, the final level of [3H]TTP incorporated into viral DNA in Fig. la was about 25 pmol per 50-pliter reaction, corresponding to 1 nmol of [3H]TTP incorporated into viral DNA per mg of total cellular DNA. Inhibitors of in vitro DNA synthesis. Listed in Table 1 are the effects of a number of compounds on long-term in vitro labeling of DNA. Synthesis was particularly sensitive to thiol reagents. Almost complete inhibition was

J. VIROL.

TABLE 2. Determination of the pool sizes of deoxynucleotide triphosphate in the lysatea Deoxynucleotide triphosphate

dATP ........ dCTP ........ dGTP ........ dTTP ........

Lysate (MM) Expt 1

Expt 2

9 33 3 10

7 22 3 10

Dialyzed lysate (jsM) Expt 2

5 19 3 10

a In experiment 1, 50-gliter incubations of lysates from ts1260-infected cells were set up in duplicate with the appropriate deoxynucleotide triphosphate missing. For each deoxynucleotide, incubations of 10 min at 32 C were made with final concentrations of 10, 20, and 40 uM of added labeled triphosphate. The specific activity of the added [3H]deoxynucleotide triphosphates was approximately 3 Ci/mmol in each case. Incorporation into DNA is proportional to y/x + y, where y is the final concentration of added precursor and x is the pool size of the precursor. 1/incorporation was plotted against 1/added concentration of precursor, and the intercept on the abscissa was used to determine the pool size. In experiment 2, the same incubations were set up except that the [3Hjdeoxynucleotide triphosphate was present in all incubations at 10 uCi/ml. Unlabeled deoxynucleotide triphosphates were added to give final concentrations of 0, 5, 10, 20, 40, and 60 uM. In this case, incorporation into DNA is proportional to 1/x + y. 1/incorporation was plotted against y, and the intercept on the abscissa was used to determine the pool size. Because the pool sizes for dGTP were so small, there was considerable error in determining those values. Dialysis was against isotonic buffer for 3 h at 0 C.

is interesting to note that only one of the two DNA polymerases known in eukaryotic cells is inhibited by thiol reagents (32). Pancreatic DNase, at the concentration used, had little effect on the in vitro synthesis. At the same time, the radioactivity in long-term, prelabeled, viral DNA was found to be resistant to the presence of DNase in the incubation. Whether this was due to protection of the viral DNA or to an inhibitor of pancreatic DNase in the lysate is not known. Compounds known to inhibit mammalian RNA polymerases by interaction with the enzyme rather than the substrate had little (AF05) or no (alpha-amanitin) effect on DNA polymerization at the concentrations used. Cycloheximide also showed no inhibition, although in vivo long-term viral DNA synthesis is sensitive to cycloheximide (3). 1-#t-DArabinofuranosylcytosine is a specific and potent inhibitor of DNA synthesis in vivo (6). AraCTP proved to be very inhibitory in vitro even at concentrations lower than the endogenous dCTP pool. Likewise, araATP was found seen with concentrations of NEM and PCMB in excess of the 1 mM DTT present in the lysate. It to be a good inhibitor. Both of these compounds

VOL. 13, 1974


have the advantage that their action can be reversed in the presence of a large excess of the corresponding deoxynucleotide triphosphate. We are continuing to examine the mechanism of inhibition by arabinosylnucleotides. Mitomycin C, a good inhibitor of DNA synthesis in vivo (30), was not extremely effective in vitro. This may reflect its mode of action in cross-linking DNA (30). Hydroxyurea, which is known to block DNA synthesis in vivo by virtue of its inhibition of the enzyme ribonucleotide diphosphate reductase (16), inhibited only slightly in vitro. Presumably, the added deoxynucleotide triphosphates can act directly as precursors for DNA synthesis in vitro without a requirement for ribonucleoside diphosphate reductase. Stability of the DNA synthesis capacity of the lysate. The DNA-synthesizing capacity of the lysate was diminished by less than 10% by storage at 0 C for up to 4 h. It also proved possible to dialyze the lysate against an isotonic buffer for up to 4 h without much loss of synthetic activity, although this was difficult to determine accurately because of the slight variation in the size of the precursor pools upon dialysis. Although, as shown above, dialysis did not reduce the pools of deoxynucleotide triphosphates to any considerable extent, we have observed that, even in the presence of an energy-generating system, added ATP (0.1 mM) was required to obtain maximal DNA synthesis in dialyzed lysates, which was not the case with undialyzed lysates. Preincubation of the lysate at 39 C for 15 min, while DNA synthesis was blocked with EDTA or araCTP, resulted in greater than 90% inhibition of viral DNA synthesis after release of the block. A 50% inhibition was observed with a 5-min preincubation. Inactivation was also seen with preincubation at 32 C. The sensitivity of cellular DNA synthesis to preincubation (halftime for inactivation was 15 min at 39 C) appeared to be less than that of viral DNA synthesis. Characterization of the in vitro product. Incorporation of labeled deoxynucleotide triphosphates into acid-precipitable radioactivity in the Hirt supernatant was used as a measure of virus-specific DNA synthesis to optimize the system. However, the Hirt supernatant was routinely analyzed by sedimentation through a preparative neutral sucrose gradient (see Materials and Methods), which at the same time served to purify viral DNA from incompletely precipitated cellular DNA. Such cellular DNA fragments usually pelleted to the bottom of the tube and did not appear in the gradient. Only viral DNA purified by this procedure was used

131

when further analysis of the viral DNA was to be made. Analysis of viral DNA synthesized during a standard in vitro reaction is presented in Fig. 2. To ascertain the stability of the pool of form I DNA present in the infected cells during the in vitro reaction, the cultures were long termlabeled in vivo with [14C]dThd, which at the same time provided an internal marker for subsequent gradient centrifugation. To follow the fate of the viral DNA molecules that were in the process of replicating at the time of isolation of the lysate, a pulse label with [3H]dThd was administered immediately before preparation of the system. The in vitro reaction was carried out at 32 C in the presence of [a-32P]dCTP to label the in vitro product. The viral DNA was extracted from samples taken at 0, 5, 15, and 60 min and sedimented through neutral sucrose gradients, separating form II (16S), form I (20S), and replicative intermediate (RI) (heterogeneous around 25S) (Fig. 2a-d). The total viral DNA (16 to 25S) was recovered from each gradient and analyzed in two other ways: alkaline sucrose gradient centrifugation, separating the rapidly sedimenting form I (53S) from single-stranded DNA generated by denaturation from form II and RI (Fig. 2e-h), and equilibrium centrifugation in CsCl gradients containing ethidium bromide, separating supercoiled form I DNA from the relaxed form II DNA (Fig. 2i-l). In such gradients RI, at least at early stages of replication, is found at densities intermediate between forms I and II. The results shown in Fig. 2 can be summarized as follows. The pool of viral DNA (as represented by the long-term "4C-prelabel) existed almost exclusively as form I molecules in the infected cells. This pool was stable upon incubation in the lysate for at least 1 h and did not participate detectably in the in vitro reaction. This was borne out by experiments in which the pool of form I was labeled with a 4-h pulse of [3H]dThd and a 30-min chase, followed by incubation in vitro with BrdUTP. No 3Hmaterial with density greater than that of marker DNA was found, although when the density-labeling step was performed in vivo, 3H-material with an altered density could be detected within 30 min (T. Hunter and B. Francke, unpublished observations). Replicating DNA (as labeled by the 3H-pulse) exhibited the known features of viral RI: it sedimented faster than form I at neutral pH; it yielded a broad peak of single strands at alkaline pH; and it had a high proportion of material at intermediate density in ethidium bromide-CsCl gradients (early RI). Soon after the beginning of the

a

I

l

N

0 xX

0C 0

N I

0

x

aW

it)

X

x N_

If)

FRACTION NUMBER

FIG. 2. Analysis of in vitro-labeled viral DNA. Four 9-cm plates of 3T3 cells infected with tsl260 were labeled with ["4C]thymidine (0.05 pCi/ml) at 20 h postinfection at 32 C. At 36 h postinfection, the plates were shifted to 39 C for 2 h. At this time, the [14Clthymidine medium was removed and replaced with 2.5 ml of fresh medium containing ['H]dThd (200 ;Ci/ml). Labeling continued for 2.5 min before preparation of the lysate. Four 100-j.liter incubations were set up with [a-_2P]dCTP (10 puM, 40 pCi/ml) and incubated for 0, 5, 15, and 60 min at 32 C. The Hirt supernatants from the four incubations were centrifuged through neutral sucrose gradients, and from each fraction 0.1-ml samples were taken to determine the radioactivity (Materials and Methods). The areas corresponding to viral DNA were pooled and extracted with phenol, and samples were subjected to alkaline sucrose gradient sedimentation and ethidium bromide-CsCI centrifugation as described in Materials and Methods. For the sucrose gradients, sedimentation was from right to left. Density increases in the ethidium bromide-CsCI gradients towards the left. Symbols: 0, 'H-radioactivity; 0, "4C-radioactivity; *, "2P-radioactivity. (a)-(d), Neutral sucrose gradients; (e)-(h), alkaline sucrose gradients; (i)(l), ethidium bromide-CsCI gradients. 132

VOL. 13, 1974

133

IN VITRO POL YOMA DNA SYNTHESIS

in vitro incubation, the density of the pulselabeled RI shifted to that of fully relaxed DNA (late RI) and later to that of form I, until after 60 min approximately 70% of the 'H-prelabel was in form I as judged by the ethidium bromide-CsCl and alkaline sucrose gradients. The in vitro label ([a-_2P]dCTP) first appeared in 25S material and, as is evident from Fig. 2, was incorporated into Rls at all stages of replication. At later times there was less incorporation into 25S material, indicating that preexisting RIs matured more efficiently than new ones were generated. After 60 min, approximately 55% of the radioactivity from the in vitro label was in form I DNA as judged by the ethidium bromide-CsCl and alkaline suctose gradients. In different experiments, values for the percentage of the in vitro label in form I DNA ranged from 50 to 80%. The percentage of the prelabeled RIs converted to form I was always higher than the percentage of the in vitro label found as form I. Occasionally a small fraction (usually less than 20%) of the in vitrolabeled DNA isolated as form I on ethidium bromide-CsCl was alkali labile and sedimented as form II on alkaline sucrose gradients. The reason for this sensitivity is not known. For the characterization of the in vitro system, these results mean that (i) the incorporation of labeled precursors into the Hirt supernatant is indeed into viral DNA; (ii) the system utilizes pre-existing RI molecules; (iii) the stages of replication demonstrated by the methods used here are indistinguishable from the known stages of viral DNA replication in vivo; (iv) the maturation to form I of pre-existing RIs is efficient (60% or more); and (v) the generation of new Rls is the process that is most sensitive to the preparation and incubation of the lysate. Analysis of the density of in vitro products labeled with BrdUTP. In the following set of experiments, BrdUTP was substituted for T'IP in the reaction mixture, and therefore the effect of BrdUTP on the overall synthesis of viral DNA was first evaluated. Incorporation of [a32P]dCTP was unaffected by the presence of 100 or 200 uM BrdUTP over time periods up to 60 min, whereas the incorporation of [3HJTTP was reduced to 8% (100 gM BrdUTP) or 3% (200 1sM BrdUTP) even at short reaction times. Thus, BrdUTP can substitute efficiently for TTP without affecting the overall synthesizing capacity of the lysate. Analysis of the density of viral DNA pulseprelabeled with [3HJdThd in vivo and labeled with [a-32P]dCTP in the presence of 150 gM BrdUTP in vitro is shown in Fig. 3. Viral DNA

extracted before and at 15 and 60 min after the start of the in vitro reaction was purified and centrifuged to equilibrium in CsCl together with 14C fully light- and "sP fully bromouracil-substituted viral DNA markers (see Materials and Methods for preparation and properties of marker DNAs). The pulse prelabel was short enough to label mostly RI (90% 25S, 10% form I, gradients not shown here). Its density profile matches that of light marker DNA (Fig. 3a). Thus, the single-stranded regions known to be present in RI do not increase its density noticeably. During the in vitro incubation, the density (a)

HL

H

LL

3

0*~~~~~~~~~ 2

(b)

I Z 0

49

l

5

6z 0

3

4

2

UA.

IL

a"I o I

2 x

x

a.

II

D-

'5

U

aX

Cc) 10"

60' 4

8

3

6

21

4

2

I

I

10

20

I

W0 40 FRACTION NUMBER

50 60 _p

FIG. 3. CsCI gradient analysis of viral DNA pulse labeled in vivo with ['HJdThd and in vitro in the presence of BrdUTP and [a-"P]dCTP. Four 9-cm plates of 3T3 cells infected with ts1260 for 36 h at 32 C and 2 h at 39 C were pulse labeled with ['HJdThd as described in Fig. 2. Three 150-uliter incubations were set up with 150 jum BrdUTP instead of TTP and [a-_"P]dCTP (10 ;M, 40 uCi/ml) and incubated for 0, 15, and 60 min at 32 C. Viral DNA was purified as described and centrifuged to equilibrium in CsCI (Materials and Methods). The vertical lines indicate the positions of 14C fully light DNA and "P fully

bromouracil-substituted DNA markers. Density infrom right to left. Symbols: 0, 'H-radioactivity; 0, "2P-radioactivity. (a) 0-min incubation; (b) 15-min incubation; (c) 60-min incubation. creases

J. VIROL. HUNTER AND FRANCKE 134 of the prelabeled DNA increased until at 60 BrdUTP was fractionated into form I and 25S min, when practically all of the Rls had been DNAs (gradient not shown), which in turn were completed, it was spread between fully light centrifuged to equilibrium in CsCl (Fig. 4a and and hybrid DNA (Fig. 3b and c). The degree of b). The average density shift of form I DNA was the density shift for any molecule reflects the about 66% of the way towards fully hybrid amount of bromouracil-substituted DNA added density (Fig. 4a), in agreement with the data in to that molecule in vitro. The peak of 3H-prela- Fig. 3c). The 25S material showed a slightly bel at 60 min of incubation was found to be greater average density shift than did form I shifted by about one-fourth of the density (Fig. 4b). In another experiment carried out in difference between light and fully hybrid mole- the same manner, the density shift of the 25S cules, indicating that on average the progeny material was clearly greater than that of form I strands in the prelabeled RIs were 75% complete (Fig. 5a and b). Since incomplete RIs contain a at the time the in vitro reaction was begun. This molar excess of parental DNA over progeny is in general agreement with the predominance DNA, the average density of bromouracil-subof late Rls found in vivo (22). Approximately stituted RIs should have been less than that of 85% of the 3H-prelabel is found shifted to higher densities to some degree (Fig. 3c). Since 10%/, of HH HL LL (a) I the prelabeled DNA was found as form I, this fraction of the label would not be expected to change density during the in vitro synthesis in A the presence of BrdUTP. Therefore, the value of I^, 85% of the 3H-material showing a density shift indicates that practically all of the pre-existing d 25s MIs participate in the in vitro reaction. This is in agreement with the observation that a high proportion of the pre-existing RIs are matured to form I in vitro. The 32P-radioactivity repreII__ _ _ senting the bromouracil-substituted DNA X -0 -40 added in vitro showed a maximum in density It IC) 10 shift which was three-fourths of the way towards fully hybrid DNA after 60 min of incubation (Fig. 3c). This illustrates that early Rls containing little 3H-prelabel required more DNA synthesis for completion than the pool of 30 35 I() 25S / 25 late RIs, and were therefore found at higher Ji densities. No radioactivity was seen at densities greater than hybrid density, suggesting that the in vitro synthesis was semiconservative. This was confirmed by banding in vitro-labeled 30 35 25 20 FRACTION NUMBER DNA, synthesized in the presence of BrdUTP, -i

I

z

x

I

-

v

z

before and after sonication in neutral CsCl and alkaline Cs,SO gradients. The in vitro label showed the expected shifts (21) to the position of hybrid density in the neutral CsCl gradient and to the density of fully substituted DNA in the alkaline CS2SO gradient after sonication. Any molecules labeled in vitro with fully hybrid density must have had approximately 50% of their DNA synthesized in vitro. Such molecules were either a result of in vitro initiation events or were derived from RIs that had been initiated in vivo just before preparation of the lysate. Less than 10% of the in vitro label was found at fully hybrid density (Fig. 3c), indicating that initiation is a rare event in vitro. Further analysis of density-labeled progeny strands is shown in Fig. 4. In this case, viral DNA labeled with [3H]dATP in the presence of

FMG. 4. Analysis of the density of viral DNA synthesized in vitro in the presence of BrdUTP and ['H]dATP. Lysate (0.3 ml) from ts1260-infected 3T3 cells was incubated in the presence of 150 uM BrdUTP and ['H]dATP (20 MM, 75 MCi/ml) for 90 min at 32 C. The Hirt supernatant was centrifuged through a neutral sucrose gradient (Materials and Methods), and the viral DNA was separated into 20S and 25S fractions. These two fractions were then centrifuged to equilibrium in CsCI (Materials and Methods). Two more samples were centrifuged to equilibrium in alkaline CsSO4 (Materials and Methods). The form I fraction was nicked by boiling for 4 min before this centrifugation. The vertical lines indicate the positions of 32P fully bromouracil-substituted and "2P fully light marker DNAs. (a) Form I neutral CsCI gradient; (b) 25S DNA neutral CsCI gradient; (c) Form I alkaline Cs,SO4 gradient; (d) 25S DNA alkaline Cs2S04 gradient.

VOL. 13, 1974 12


135

the

(a) I

HH

HL

LL

same length of time, through alkaline sucrose gradients, excluding breakage as a reason

for the relatively high density. We can analyze the data in Fig. 4c and d as : follows. If (i) no initiation occurs in vitro, (ii) L ,'- 4 the pre-existing RIs are randomly distributed throughout the replication cycle, and (iii) all of x the pre-existing RIs are completed to form I in 0.3 vitro, then a plot like that in Fig. 4c would show RI a straight line passing through the abscissa at the position of fully light density and with a _ ~ ~~~~~~~ slope such that the radioactivity at 50% intermediate density was half that at fully substituted density. Since very little initiation seems to occur in vitro, deviations from such a line 10 20 30 40 50 must mean that either the pool of pre-existing FRACTION NUMBER _p RIs was not randomly distributed throughout FIG. 5. Analysis of the density of form I and 25S the replication cycle or not all of the pre-existviral DNA synthesized in vitro in the presence of ing RIs were completed in vitro. It is clear that BrdUTP and ['HldA TP. Lysate (0.5 ml) from ts1260- not all of the pre-existing RIs are matured to infected 3T3 cells was incubated in the presence of 150 pMBrdUTP and [1HJdA TP (10 pM, 40 ,Ci/ml) for 90 form I in vitro. However, most of both of the min at 32 C. The Hirt supernatant was centrifuged prelabeled DNAs and the in vitro-labeled DNA through a neutral sucrose gradient (Materials and is near unit viral length at the end of the Methods), and the viral DNA was separated into 20S incubation (data not shown), so that a summaand 25S fractions. These two fractions were then tion of the data in Fig. 4c and d gives a rough centrifuged to equilibrium in CsCl (Materials and indication of what the distribution of densities Methods). The vertical lines indicate the positions of would look like if all of the pre-existing Rls were 82P fully bromouracil-substituted and 2P fUlly light completed in vitro. Deviations from the straight marker viral DNAs. line can therefore be taken to reflect the distrimature viral DNA with progeny strands substi- bution of the pre-existing RIs within the replicatuted to the same extent. Therefore, the fact tion cycle. The shoulder seen towards the light that the density shift seen with 25S molecules side in Fig. 4c,presumably results from the pool was as great or greater than that for form I of late RIs known to exist in whole cells (22). molecules suggests that the residual RIs left at Otherwise, the distribution of radioactivity sugthe end of the incubation were, before the gests that synthesis occurred on RIs rather preparation of the lysate, at an earlier stage of randomly distributed throughout the replicareplication than those RIs matured to form I in tion cycle. Progeny strands had been elongated vitro. The two preparations of DNA from the by up to 90% of the unit viral DNA length, the experiment in Fig. 4 were also centrifuged to average elongation being about 50%. It is interequilibrium in alkaline Cs2SO4 gradients to esting to note that in the case of both form I and examine the density of the progeny strands. In 25S materials, there was a maximal density order to allow complete denaturation of form I shift of approximately 90%. This may mean DNA, this was first nicked by boiling so that that the initiation event involves the synthesis there were no supercoiled molecules left and the of a small amount of DNA as a priming event average size of the labeled DNA was 12S as and that it is this which does not occur in vitro. determined by sedimentation analysis. The Since the residual, labeled 25S material condensity distributions obtained (Fig. 4c and d) tains nearly unit viral length progeny strands of show the profiles expected from the densities of a high average density shift, it seems likely that the native DNAs (Fig. 4a and b). For form I such molecules are not derived by initiation DNA, there was a spread of material between events, but are rather derived from early Rls fully light DNA and fully bromouracil-sub- which for some reason have been unable to stituted DNA, with the peak toward the fully mature. Although it is clear that the incorporasubstituted side. For 25S DNA, there was a tion of labeled precursors in vitro represent more pronounced peak near the fully substisynthesis on RIs that were randomly distributed tuted marker. The integrity of the bromouracil- throughout the replication cycle, the increase in labeled single strands, after the alkaline Cs,SO4 incorporation between 15 and 60 min (70% in gradients, was ascertained by sedimenting con- the experiment in Fig. 3) probably reflects trol samples, kept under identical conditions for synthesis on RIs at relatively early stages of

9a

u

I

136

HUNTER AND FRANCKE

J. VIROL.

replication, since by 15 min more than 40% of tem. The exact nature of the requirement for the label in the pre-existing pool of late RIs ATP in DNA replication, even in the more (75% completed) had already matured to form I. purified in vitro system from E. coli (25), is not well understood. In addition to deoxynucleotide DISCUSSION triphosphates, the lysate utilized externally Although there have been a number of studies added deoxynucleosides and their mono- and of eukaryotic DNA replication in vitro using diphosphates, reflecting the high phosphorylatnuclei isolated from cells in the process of ing capacity of the lysate and indicating one making DNA (12, 19), the poorly defined and possible role for ATP. The incorporation of high-molecular-weight nature of the chromo- nucleosides, however, was effectively competed somal DNA makes characterization of the syn- by the respective triphosphates, indicating thetic product difficult. For this reason, we that, in all likelihood, the true precursors for chose to study the in vitro replication of the DNA polymerization are deoxynucleotide triwell-defined low-molecular-weight DNA of pol- phosphates. The lack of inhibition by comyoma virus as a model system. Since it is likely pounds known to have indirect effects on DNA that most of the processes within the viral synthesis in vivo (cycloheximide and hydroxyreplication cycle utilize the enzymatic and urea) and the well-defined inhibition by structural machinery involved in cellular DNA araCTP and araATP (acting at the level of desynthesis, such a study should afford an under- oxynucleotide triphosphates) further substanstanding not only of viral DNA replication but tiate this point. Thus, under the conditions described, the system can be used to study in vitro of chromosomal DNA replication as well. Our approach in developing an in vitro DNA DNA polymerization, irrespective of other procreplication system was to aim for maximal esses carried out by the lysate. We have used DNA-synthesizing activity while retaining as the system to study the detailed mechanism of many of the in vivo features of DNA replication discontinuous chain growth of polyoma DNA as possible. A concentrated lysate of hypotoni- (B. Francke and T. Hunter, unpublished obsercally treated infected cells, as described in this vations) and to show the covalent association of paper, was found to meet these requirements RNA with the small, discontinuously growing best. After the lysate was prepared, no further DNA fragments. Neither of these studies would fractionation or purification steps were carried have been possible in vivo. The analysis of the product synthesized in out. Thus, it is reasonable to suppose that the lysate contains all of the macromolecules in- vitro has been discussed in detail in Results. In volved, in vivo, in DNA synthesis. In addition, general, for the duration of in vitro activity, most of the low-molecular-weight compounds in incorporation into cellular and viral DNAs rethe cell would be expected to be present, and flected the pattern found in vivo. In vitro this is bome out by the lack of absolute depend- labeling of cellular DNA was only observed ence for DNA synthesis on added nucleotide when the cells had been stimulated to synthetriphosphates and the sizeable pools of deox- size DNA in vivo either by virus infection or by ynucleotide triphosphates found in the lysate serum stimulation. Resting cells showed no (Table 2). Not surprisingly, such a crude system DNA synthesis in vitro. We conclude that the in is also able to carry out some RNA and protein vitro labeling of cell DNA constitutes the consynthesis as demonstrated by the incorporation tinuation of the regular in vivo process rather of the appropriate precursors. For this work, we than being synthesis, for instance of a repair chose to use the ts1260 temperature-sensitive type, induced by the preparation of the lysate. mutant of polyoma, which is a mutant in a late Cellular DNA synthesis proceeded in vitro for complementation class (7; W. Eckhart, unpub- rather long time periods at a linear rate, further lished observations). Ts1260 has a DNA replica- emphasizing the potentiality of the system for a tion cycle identical to that of wild-type and study of chromosomal DNA replication. With respect to polyoma DNA synthesis, produces normal amounts of viral DNA at the nonpermissive temperature. We have checked, during the initial phase of the incubation, however, that all of the events in viral DNA labeled deoxynucleotide triphosphates were inreplication observed in vitro with ts1260- corporated mainly into viral RI. The properties infected lysates are also found with wild type- of RI labeled in vitro were indistinguishable from those of RI labeled in vivo. It sedimented infected lysates. When optimized, viral DNA synthesis in the at approximately 25S in neutral sucrose gradilysate depended most stringently on the pres- ents, had an intermediate density in ethidium ence of magnesium ions, a thiol, and ATP, or bromide-CsCl gradients, showed a broad distrimore correctly upon an energy-generating sys- bution of progeny chain lengths in alkaline

VOL. 13, 1974


sucrose gradients (all Fig. 2), and was retained on BND cellulose columns in 1 M NaCl (B. Francke and T. Hunter, unpublished observations). The processing of this viral RI in vitro followed the steps known in vivo. We showed that DNA chain growth occurs at least partly discontinuously on RIs at all stages of replication (B. Francke and T. Hunter, unpublished observations). The completed daughter molecules were found initially as form II molecules in which the progeny strands were nicked (B. Francke and T. Hunter, unpublished results), as has been shown in vivo for SV40 (9). Maturation of a high percentage of the labeled RI to form I occurred at later stages of the incubation. The maturation process appeared to be temperature dependent, being rather inefficient at lower temperatures. We have not determined whether conditions optimal for DNA synthesis are necessarily optimal for maturation to form I. By in vivo prelabeling of viral RIs, we were able to show that practically all pre-existing RIs participated in the in vitro reaction and that a high proportion of them were matured to form I. This was confirmed by the shift in density obtained with prelabeled RIs when synthesis was allowed to occur in the presence of BrdUTP in vitro. The degree of the density shift obtained (Fig. 3c) suggests that most of the prelabeled molecules were at rather a late stage of replication at the time of preparation of the lysate. Magnusson et al. (21) have demonstrated in a similar fashion that all of the prelabeled RIs participate in the in vitro reaction occurring in nuclei derived from polyoma-infected 3T6 cells. We have confirmed that synthesis on viral Rls was semi-conservative, as has already been shown for the replication of polyoma DNA in 3T6 nuclei (21). The pool of prelabeled form I DNA present in the cells at the time of isolation of the lysate was completely stable during the in vitro incubation (see Fig. 2). There was, however, a rapid loss of synthesizing activity upon preincubation of the lysate in the absence of DNA synthesis, as noted by Magnusson et al. (21) for polyoma DNA synthesis in 3T6 nuclei. It is not clear whether the sensitivity of viral DNA synthesis is related to the inability of the lysate to initiate new rounds of viral DNA synthesis, although the sensitivity of cellular DNA synthesis to preincubation suggests that some factor(s) common to both types of synthesis is inactivated by preincubation. The rate of in vitro synthesis increased with the temperature of incubation as expected. The absolute rate of DNA synthesis in the lysate, determined by using the methods described by Winnacker et al. (34), was found to

137

be between 30 and 50% of the in vivo rate. Incorporation into viral DNA generally leveled off sooner than that into cellular DNA. From the small size of the viral DNA, this is what one would expect if reinitiation of new rounds of DNA synthesis was the process most sensitive to the in vitro conditions. However, it was possible that the incorporation observed at later times in the incubation when most of the pre-existing RIs appeared to have been completed was a result of initiation events. Molecules initiated and completed in vitro in the presence of BrdUTP would be expected to have fully hybrid density. Molecules initiated but not completed would be expected to have fully dense progeny strands. Analysis of the product synthesized in the presence of BrdUTP (Fig. 4) showed that rather a small fraction of the mature viral DNA was of fully hybrid density and also that only a small proportion of the progeny strands of incomplete viral RIs were fully dense. These results indicate that initiation is indeed the limiting factor in continued synthesis in vitro. The density distribution of in vitro-labeled DNA (Fig. 4c and d) indicates that in vitro synthesis occurs on RIs rather randomly distributed throughout the replication cycle, although there is some evidence for a pool of late RIs awaiting completion. The progeny strands synthesized in vitro showed a maximal density shift of 90% of the way to full substitution, which implies that the maximal elongation of progeny strands was up to 90% of unit viral DNA length. It seems that there are few RI molecules with less than 10% of their DNA copied into progeny strands at the start of the incubation, upon which in vitro synthesis occurred. It is possible that a "priming step" is necessary for viral form I molecules to enter the replication cycle. Such molecules might have up to 10% of their DNA synthesized during the initiation priming event. Such a step could be similar to that known to occur during mitochondrial DNA synthesis. Since the initiation step is most likely the only one in viral DNA replication controlled by a viral gene (11, 29), it would be desirable to study it in vitro. We are presently testing modified ways to prepare the lysate in order to obtain initiation in vitro. It is interesting to compare our system with that for in vitro polyoma DNA synthesis recently reported by Winnacker et al. (34) and Magnusson et al. (21) while the present work was in progress. Their system consists of nuclei isolated from polyoma-infected 3T6 cells. Synthesis also occurred semi-conservatively, mainly into pre-existing RIs, but the incorporation ceased after relatively shorter times, and

138

HUNTER AND FRANCKE

only a small fraction of the in vitro product consisted of mature form I. From density shift experiments, Magnusson et al. (21) reported that the percentage elongation of the progeny strands is considerably shorter than for the lysate. Some of the differences could be accounted for by a requirement for cytoplasmic factors, removed by purification of the nuclei. In our hands, the homogenization and purification steps involved in preparing the nuclei reduced in vitro synthesizing activity markedly, both with respect to total incorporation of (3H]TTP and also efficiency of maturation to form I. This suggests that the decreased activity might also be caused by mechanical damage. It is also possible that some of the differences between the two systems are due to the different cell lines used. It is not clear which of the DNA polymerases known in eukaryotic cells (5, 32) are involved in DNA replication. Since it has been shown that viral DNA synthesis occurs, at least in part, via a discontinuous mechanism (10, 20, 14), the possibility has been raised that more than one DNA polymerase may be involved in chain growth (20). We are presently examining the change in pattern of discontinuous DNA synthesis with arabinosyl nucleotides and thiol reagents, inhibitors which have differential effects on the two known DNA polymerases, in the hope of defining the roles of the different polymerases. We are also using ATP analogues to investigate the role of ATP in DNA synthesis. ACKNOWLEDGMENTS We thank Helen Hesser and Mary Anne Hutchinson for their technical assistance, and Walter Eckhart, in whose laboratory this work was performed, for his interest, advice, and support. This research was supported by Public Health Service grant no. CA 13884 from the National Cancer Institute and contract no. 67-1147 of the Virus Cancer Program of the National Cancer Institute. B.F. is a recipient of a fellowship from the Deutsche Forschungsgemeinschaft. LITERATURE CITED 1. Borgaux, P., and D. Borgaux-Ramoisy. 1971. A symmetrical model for polyoma virus DNA replication. J. Mol. Biol. 62:513-524. 2. Borgaux, P., and D. Borgaux-Ramoisy. 1972. Unwinding of replicating polyoma virus DNA. J. Mol. Biol. 70:399-413. 3. Branton, P. E., W. P. Cheevers, and R. Sheinin. 1970. The effect of cycloheximide on DNA synthesis in cells productively infected with polyoma virus. Virology 42:979-992. 4. Chamberlin, M., and P. Berg. 1964. Mechanism of RNA polymerase action: formation of DNA-RNA hybrids with single-stranded templates. J. Mol. Biol. 8:297-313. 5. Chang, L. M. S., and F. J. Bollum. 1971. Low molecular weight DNApolymerase in mammalian cells. J. Biol. Chem. 246:5835-5837.

J. VIROL.

6. Cohen, S. S. 1966. Introduction to the biochemistry of D-arabinosyl nucleosides. Progr. Nucl. Acid Res. Mol. Biol. 5:1-88. 7. Eckhart, W. 1969. Complementation and transformation by temperature sensitive mutants of polyoma virus. Virology 38:120-125. 8. Fareed, G. C., C. F. Garon, and N. P. Salzman. 1972. Origin and direction of SV40 DNA replication. J. Virol. 10:484-491. 9. Fareed, G. C., M. L. McKerlie, and N. P. Salzman. 1972. Characterization of simian virus 40 DNA component II during viral DNA replication. J. Mol. Biol. 74:95-111. 10. Fareed, G. C., and N. P. Salzman. 1972. Intermediate in SV40 chain growth. Nature N. Biol. 238:277-279. 11. Francke, B., and W. Eckhart. 1973. Polyoma gene function required for viral DNA synthesis. Virology, in press. 12. Friedman, D. L., and G. C. Mueller. 1968. A nuclear system for DNA replication from synchronized Hela cells. Biochim. Biophys. Acta 161:455-468. 13. Hirt, B. 1967. Selective extraction of polyoma DNA from infected mouse cell cultures. J. Mol. Biol. 26:365-369. 14. Hunter, T., and B. Francke. 1973. In vitro polyoma DNA synthesis: involvement of RNA in discontinuous chain growth. J. Mol. Biol., in press. 15. Jaenisch, R., A. Mayer, and A. Levine. 1971. Replicating SV40 molecules containing closed circular template DNA strands. Nature N. Biol. 233:72-76. 16. Krakoff, I. H., N. C. Brown, and P. Reichard. 1968. Inhibition of ribonucleotide diphosphate reductase by hydroxyurea. Cancer Res. 28:1559-1565. 17. Kumar, K. V., and D. L. Friedman. 1972. Initiation of DNA synthesis in HeLa cell free system. Nature N. Biol. 239:74-76. 18. Levine, A. J., H. S. Kang, and F. E. Billheimer. 1970. DNA replication in SV40 infected cells. I. Analysis of replicating SV40 DNA. J. Mol. Biol. 50:549-568. 19. Lynch, W. E., R. F. Brown, T. Umeda, S. G. Langreth, and J. Lieberman. 1970. Synthesis of DNA by isolated liver nuclei. J. Biol. Chem. 245:3911-3916. 20. Magnusson, G., V. Pigiet, E. L. Winnacker, R. Abrams, and P. Reichard. 1973. RNA-linked short DNA fragments during polyoma replication. Proc. Nat. Acad. Sci. U.S.A. 70:412-415. 21. Magnusson, G., E. L. Winnacker, R. Eliasson, and P. Reichard. 1972. Replication of polyoma DNA in isolated nuclei. II. Evidence for semi-conservative replication. J. Mol. Biol. 72:539-552. 22. Mayer, A., and A. J. Levine. 1972. DNA replication in SV40 infected cells. VIII. The distribution of replicating molecules at different stages of replication in SV40 infected cells. Virology 50:328-338. 23. Moses, R. E., and C. C. Richardson. 1970. Replication and repair of DNA in cells of E. coli treated with toluene. Proc. Nat. Acad. Sci. U.S.A. 67:674-679. 24. Nathans, D., and K. Danna. 1972. Specific origin in SV40 DNA replication. Nature N. Biol. 236:200-202. 25. Pisetsky, D., I. Berkower, R. Wickner, and J. Hurwitz. 1972. Role of ATP in DNA synthesis in E. coli. J. Mol. Biol. 71:557-571. 26. Schaller, H., B. Otto, V. Niisslein, J. Huf, R. Hermann, and F. Bonhoeffer. 1972. Deoxyribonucleic acid replication in vitro. J. Mol. Biol. 63:183-200. 27. Sebring, E. D., T. J. Kelly, M. M. Thoren, and N. P. Salzman. 1971. Structures of replicating simian virus 40 DNA molecules. J. Virol. 8:478-490. 28. Skoog, L. K., B. A. Nordenskjold, and K. G. Bjursell. 1973. Deoxyribonucleoside triphosphate pools in synchronized hamster cells. Eur. J. Biochem. 33:428-432. 29. Tegtmeyer, P. 1972. Simian virus 40 DNA synthesis: the viral replicon. J. Virol. 10:591-598. 30. Waring, M. J. 1968. Drugs which affect the structure and

VOL. 13, 1974


function of DNA. Nature (London) 219:1320-1325. 31. Weber, M. J., and G. Edlin. 1971. Phosphate transport, nucleotide pools and ribonucleic acid synthesis in growing and in density inhibited 3T3 cells. J. Biol. Chem. 246:1823-1833. 32. Weissbach, A., A. Schlabach, B. Fridlender, and A. Bolden. 1971. DNA polymerases from human cells. Nature N. Biol. 231:167-170.

139

33. Wickner, W., D. Brutlag, R. Schekman, and A. Kornberg. 1972. RNA synthesis initiates in vitro conversion of M13 DNA to its replicative form. Proc. Nat. Acad. Sci. U.S.A. 69:965-969. 34. Winnacker, E. L., G. Magnusson, and P. Reichard. 1972. Replication of polyoma DNA in isolated nuclei. I. Characterization of the system from mouse fibroblast 3T6 cells. J. Mol. Biol. 72:523-537.