In Vitro Initiation of DNA Replication in Simian Virus 40 Chromosomes*

then fractionated into SV40 chro- matin (HSC and LSC, respectively) and nucleosol (HSN and LSN, respectively) by sedimentation. In uitro DNA replication assays were incubated for 50 min in the presence of [a-32P]dNTP, and the resulting [32P]DNA hybridized with ori (mSVO1 + mSV02) or with sequences 1800 bp from ori (mSV07 + mSV08). Results were nor- malized to a constant amount of SV40 [3H]DNA that had been prelabeled in uiuo. A value of 1.0 represented 12 pmol of dNMPs incomorated/h/350 na of DNA in a 50-ul reaction.

A soluble system has been developed that can initiate DNA replication de novo in simian virus 40 (SV40) chromatin isolated from virus-infected monkey cells as well as in circular plasmid DNA containing a functional SV40 origin of replication (ori). Initiation of DNA replication in SV40 chromatin required the soluble fraction from a high-salt nuclear extract of SV40infected cells, a low-salt cytosol fraction, polyethylene glycol, and a buffered salts solution containing all four standard deoxyribonucleoside triphosphates. Purified Replication began at ori and proceeded bidirectionally to generate replicating DNA intermediates in which the parental strands remained covalently closed, as observed in uiuo. Partial inhibition of DNA synthesis by aphidicolin resulted in accumulation of newly initiated replicating intermediates in this system, a phenomenon not observed under conditions that supported completion of replication only. However, conditions that were optimal for initiation of replication repressed conversion of late-replicating intermediates into circular DNA monomers. Most surprising was the observation that p-n-butylphenyl-dGTP, a potent and specific inhibitor of DNA polymerase-a, failed to inhibit replication of SV40 chromatin under conditions that completely inhibited replication of plasmid DNA containing the SV40 on' and either purified or endogenous DNA polymerase-a activity. In contrast, all of these DNA synthesis activities were inhibited equally by aphidicolin. Therefore, DNA replication in mammalian cells is carried out either by DNA polymerasea that bears a unique association with chromatin or by a different enzyme such as histone composition and nucleosome structure are essentially the same as those of its host (reviewed in Refs. 1 and 2). Initiation of SV40 DNA replication requires a genetically defined, cis-acting viral sequence identified as the origin of replication (ori), the 81,500-dalton virus-encoded large tumorantigen (T-ag),' and one or more factors from monkey or human cells that permit viral DNA replication. Several subcellular systems have been described that faithfully complete SV40 or polyoma virus DNA replication and chromatin assembly in viral chromosomes that initiated replication in vivo, although these systems initiate few, if any, new rounds of viral DNA replication in vitro (reviewed in Ref. 3). Using purified plasmid DNA containing the SV40 or polyoma virus ori, soluble extracts of permissive cells supplemented with the homologous viral T-ag have been shown to be capable of initiating bidirectional DNA replication at the SV40 or polyoma ori sequence (4)(5)(6)(7)(8). Since T-ag is the only virally encoded protein required for initiation of viral DNA replication, all other proteins and cofactors must be provided by the permissive cell host. So far, only DNA primase-DNA polymerase-a has been implicated strongly in the initiation process (7)(8)(9)(10)(11)(12). Current experimental strategies to identify the remaining components involved in initiation and the sequence of interactions that lead to initiation rely heavily on the application of specific enzyme inhibitors, antibodies, and in vitro complementation assays. Therefore, the utility of these systems in characterizing initiation of viral DNA replication will depend on how accurately they mimic the replication of viral chromosomes in uiuo.
Since SV40 and polyoma virus replicate in uiuo as circular chromosomes, we addressed the question of whether or not these endogenous viral chromosomes are capable of initiating replication in a soluble system, and, if so, whether the characteristics of viral chromatin replication would be the same as plasmid DNA replication. The original low-salt extract of SV40-infected CV-1 cells described by Su and DePamphilis (13,14) that continues replication of replicating viral chromosomes in vitro was modified to include a high-salt nuclear extract and polyethylene glycol. These modifications stimulated T-ag-dependent, ori-dependent initiation of bidirectional replication in viral chromosomes and in purified plasmid DNA. Surprisingly, p-nbutylphenyl-dGTP, a potent, highly specific inhibitor of DNA polymerase-a (15), inhibited replication of SV40 chromatin at concentrations 100-fold higher than required to inhibit DNA polymerase-a. These results are the first demonstration The abbreviations used are: T-ag, tumor antigen; Hepes, 4-(2-hydroxyethy1)-1-piperazineethanesulfonic acid; EGTA, [ethylenebis(oxyethylenenitri1o)ltetraacetic acid; PEG, polyethylene glycol; bp, base pairs; RI, replicating intermediates; RI*, late replicating intermediates; BuPdGTP, p-n-butylphenyl-dGTP. that chromosomes can initiate replication in vitro and suggest either that a more complex form of DNA polymerase-a is associated with replicating chromosomes or that a different enzyme, such as DNA polymerase-6, is the replicative DNA polymerase.

Materirrls and Standard Solutions
Restriction endonucleases were obtained from New England Biolabs, Beverly, MA, with the exception of EcoRII. Yeast tRNA and EcoRII were obtained from Bethesda Research Laboratories. Adenosine-5'-tetraphosphate-5'-adenosine (lithium salt) was obtained from Boehringer Mannheim. All radioactive isotopes were from Du Pont-New England Nuclear. p-n-Butylphenyl-dGTP was a gift from Dr. G. Wright (University of Massachusetts Medical School) (16), and HeLa cell DNA primase-DNA polymerase-a was a gift from Dr. E. Baril  Cell Lysate-Ten 15-cm plates of 80% confluent CV-1 cells were infected with SV40 wt800 as described previously (10). At 36-38 h of postinfection, the media was replaced with 1 ml of TS buffer, 10% calf serum (GIBCO), and 500 pCi of t3H]thymidine (80 Ci/mmol) and incubated for an additional 90 min to uniformly label SV40 DNA (4 X IO6 cpm 3H/pg of DNA). All subsequent steps were carried out at 0-4 "C. Cells were washed twice with T S buffer containing 250 mM sucrose to prevent premature lysis of infected cells and then once with low-salt buffer. Excess buffer was removed by aspiration, and the cells were scraped free with a rubber policeman, transferred to a Dounce homogenizer (Kontes, Vineland, NJ), and lysis completed with 3-4 strokes of pestle B. This cell lysate, containing swollen but intact nuclei, was either used directly for in uitro replication of endogenous SV40 viral chromatin (18) or fractionated further as described below.
Cytosol and Nucbar Extract-Nuclear extracts were prepared using a modification of the procedure of Su and DePamphilis (14). Cell lysate from 10 15-cm dishes was centrifuged at 1,200 X g for 5 min to remove nuclei, and the supernatant was centrifuged at 100,000 X g for 1 h in a Beckman Ti-50 rotor. This "cytosol" fraction consisted of about 2.5 ml containing 7-12 mg/ml protein but no detectable [3H] DNA and was stored at -70 "C. The 1,200 X g pellet, containing the nuclei, was resuspended in 1.3 ml of low-salt buffer and incubated on ice for 90 min with occasional agitation. Approximately 75% of the SV40 3H-chromatin was released from the nuclei. The nuclei were then sedimented at 8,000 X g for 10 min, and the resulting "low-salt nuclear extract" was stored at -70 "C. "High-salt nuclear extract" was prepared in the same way, except that nuclei were resuspended in low-salt buffer plus 500 mM potassium acetate.
Nucleosol and SV40 Chromatin-Nuclear extracts were fractionated by sedimentation in a Beckman SW60 Ti rotor at 300,000 X g for 1 h, and the "nucleosol" supernatant (6-10 mg/ml protein) was stored at -70 'C. was added when low salt nuclear extracts were used. The final reaction volume was 50 pl. Reaction mixtures were preincubated on ice for 15 min and then incubated at 30 "C for the times indicated. When nucleosol and SV40 chromatin were substituted for nuclear extracts, 8 pl of nucleosol and 2 pl of chromatin were added per 50 pl of reaction. I n uitro DNA replication conditions for unfractionated cell lysates were the same as described by DePamphilis et al. (18).
Replication of plasmid DNAs (pSVori, pSVoriA6) was carried out under the same conditions as described above for SV40 chromatin in the presence of nuclear extracts. Plasmid pSVori was made by substituting nucleotides 29-562 of plasmid pML-1, a pBR322 derivative in which the mammalian DNA replication poison sequence has been deleted (19), with the 213-bp HindIII-SphI segment of SV40 wt800 containing the origin of DNA replication (ori). Plasmid pSVoriA6 is pSVori but with a 6-bp deletion within ori centered at the BglI site (6). In some cases (Fig. 7), plasmid DNA was incubated in the presence of uninfected CV-1 extracts, supplemented with purified SV40 large T-antigen under otherwise equivalent conditions.
The DNA products of the replication reaction were purified as described previously (6,11). When replication was carried out in crude cell lysates or isolated nuclei, viral DNA was isolated by the method of Hirt (20).

DNA Hybridization Assay
Hybridization assays were done using circular, single-stranded M13mp7 DNAs into which regions of SV40 wt800 have been cloned (mSV DNAs). mSVOl and mSV02 contain complementary strands of the 311-bp SV40 BstNI G fragment containing the ori region. mSV07 and mSV08 contain complementary strands of the 364-bp A d -P s t I fragment 1626-1988 bp to the late gene side of ori (10). Ten pg of either of these DNAs, or M13mp7 DNA alone, were denatured and adsorbed onto nitrocellulose discs (13 mmZ) using a "Dot-Blot" filtration manifold (Minifold SRC-97; Schleicher & Schuell) as described by Wirak et al. (21). Seventy-five % of the DNA was retained by the nitrocellulose membrane, and adsorbtion was proportional to the amount of DNA added within a range of 1-15 pg of added DNA.
Individual discs (13 mm') were cut from the membrane and incubated in a prehybridization solution as previously described (21), but using a 5 X concentration of Denhardt's solution and 0.1 mg/ml salmon sperm DNA. SV40 [32P]DNA from the in uitro replication reactions was digested with either BstNI endonuclease (ori) or AccI+PstI endonucleases (control), denatured at 100 "C, and chilled on ice. 5-50 ng of SV40 DNA (22000 cpm 3H) was added per disc; equal aliquots were also incubated with discs containing mp7 DNA alone. Hybridization reactions (115 pl) were carried out overnight at 68 'C in sealed microtiter dishes (3696; Costar, Cambridge, MA) under prehybridization conditions but substituting 0.3 mg/ml tRNA for salmon sperm DNA. Following hybridization the discs were washed twice in 0.6 M Tris-HC1 (pH 7.5), 3.0 M NaCl, 10 mM EDTA at room temperature, once in 3 x standard saline citrate (SSC; Ref. 26), 2.5% sodium dodecyl sulfate at 68 "C, and once in 2 X SSC, 2.5% sodium dodecyl sulfate at 68 "C. The discs were dried, and bound radioactivity was determined by liquid scintillation. The efficiency of the assay was determined by hybridizing aliquots of between 10 to 250 ng of 5'-32P-end-labeled (125 cpm/ng; Ref. 26) BstNI restriction fragments of SV40 Form I DNA to discs containing mSVOl + mSV02 DNAs. The efficiency of annealing was 80%, and reproducibility between duplicate samples was within 4% (standard deviation/mean). Less than 6% of the [5'-32P]DNA bound by mSVOl + mSV02 was retained by M13mp7 alone, and the amount of [5'-3'P] DNA bound was proportional to its concentration throughout the range tested (10-250 ng). Hybridization with mSVOl + mSV02 DNAs specifically removed the ori-containing BstNI G fragment as shown by analysis of the [5'-32P]DNA before and after hybridization by electrophoresis in 1.2% agarose gels.

Electrophoresis
Purified DNA products were analyzed by electrophoresis in 0.6% agarose in Tris.borate, EDTA buffer (TBE, Ref. 22). Where indicated, DNA products were digested by restriction endonucleases under conditions suggested by Maniatis et al. (22) and fractionated by electrophoresis in 6% polyacrylamide in TBE. Radioactive DNA was visualized by fluorography using a Cronex (Du Pont). Relative amounts of DNA forms in the gel were estimated by densitometry (6).

RESULTS
Polyethylene Glycol Stimulates SV40 DNA Synthesis in Vitro-Su and DePamphilis (13,14) showed that a low-salt extract of nuclei from SV40-infected CV-1 cells, free of cellular chromatin, allowed the endogenous SV40-replicating chromosomes to complete replication in uitro when supplemented with cytosol from either infected or uninfected CV-1 cells. However, this system was never observed to initiate DNA replication at ori. In an effort to correct this problem, a high-salt nuclear extract was substituted for the low-salt nuclear extract in the hope that it would contain higher concentrations of initiation factors such as large T-ag. Although this change stimulated total SV40 DNA synthesis about 50%, analysis of newly replicated DNA by electrophoresis in agarose gels revealed the same distribution of DNA forms originally described by Su and DePamphilis (13,14). Nascent DNA first appeared in replicating intermediates (RI) at all stages of replication, and then, with time, about half the nascent DNA accumulated as late replicating intermediates 90% completed (RI*) and about half as circular DNA monomers that were either covalently closed and superhelical (Form I) or topologically relaxed (Form 11; Fig. 1, lanes a-e). The electrophoretic mobility of various forms of SV40 DNA has been described previously (25).
Remarkably, addition of 5% PEG to this high-salt nuclear extract stimulated SV40 DNA synthesis at least 4-fold and dramatically increased the amount of nascent DNA that appeared initially in newly initiated RI (Fig. 1, lune g). This RI migrated between Form I and I1 DNA during gel electrophoresis, and its nascent (radiolabeled) DNA was localized in the ori region (11,25). The amount of radiolabel in early RI was more pronounced after 30 min (lane g) of incubation than after 10 min (lane f ) , suggesting that initiation of replication was delayed for at least 10 min. The difference in the DNA products produced in the presence and absence of PEG was most apparent when equal amounts of reaction mixture that had been incubated for the same time were compared directly (lanes k and I).
Addition of PEG also resulted in accumulation of replicating intermediates about 90% completed (RI*) at the expense of circular DNA monomer production ( Fig. 1, lanes i and j ) , indicating that PEG interfered with termination of DNA replication. To confirm that RI was still converted into Form I and I1 in the presence of PEG, albeit less efficiently, a reaction mixture was incubated in the presence of [o-~*P] dNTPs for 30 min ("pulse"), 500 p~ unlabeled dNTPs were added and the incubation continued ("chase"). Radiolabel was no longer incorporated during the chase period, although prelabeled [3H]RI was converted into Form I and I1 (data not shown). After the 30 min pulse (0-min chase), a broad distribution of RI at various stages in its replication was apparent with some accumulation of RI* (Fig. 1, lane m). After a 30min chase, early RI had completely disappeared, and several forms of middle-to-late RI were present (lane n), and after a 60-min chase (lane 0) RI was converted into Form I and 11. In fact, the distribution of 32P-radiolabel among the various forms of SV40 DNA after a 60-min chase (lane 0) was equivalent to that observed after continuously labeling nascent DNA for 90 min (lunej). Thus, it appears that PEG stimulated initiation of SV40 DNA replication within the first 30 min but inhibited subsequent conversion of RI* into circular DNA monomers.
Polyethylene Glycol Stimulates DNA Synthesis Specifically in the Ori Region-To determine whether or not PEG stimulated initiation of new rounds of DNA replication, newly synthesized SV40 [32P]DNA was annealed to unique segments of the SV40 genome that had been cloned into single-stranded M13 virion DNA. mSVOl and mSV02 contain complementary strands of a 311-bp SV40 DNA restriction fragment that includes the 64-bp ori-core sequence as well as both 45-bp flanking ori-auxiliary sequences. Thus, hybridization of nascent DNA to mSVOl + mSV02 measured the amount of DNA synthesis in ori. mSV07 and mSV08 contain the complementary strands of a 364-bp SV40 DNA restriction fragment centered about 1800 bp to the late gene side of ori. Hybridization of nascent DNA to mSV07 + mSV08 measured the amount of DNA synthesis at replication forks far removed from ori.
A high-salt nuclear extract supplemented with cytosol and 4-896 PEG stimulated DNA synthesis at least 7-fold in the ori region with only minor stimulation of DNA synthesis in late RI (control, Fig. 2). PEG preparations from 6,000 to 20,000 daltons in size were tested at concentrations of 1 and 5% and found to have the same amount of activity. Therefore, 14,000-dalton PEG was routinely used because its lower viscosity made it easier to handle. PEG did not stimulate initiation of SV40 chromatin replication in isolated nuclei from infected CV-1 cells supplemented with cytosol, a system that faithfully continues replication in SV40 RI (22). In fact, DNA synthesis in isolated nuclei was inhibited in proportion to PEG concentration. Polyvinyl alcohol also stimulated DNA synthesis in high-salt nuclear extracts at concentrations similar to those used with PEG. However, polyvinyl alcohol was more viscous than PEG, and some lots were inhibitory.
In the presence of PEG, newly synthesized ori region DNA accumulated relative to nascent DNA in RI at later stages of replication (Fig. 3A), consistent with initiation of SV40 DNA  Fig. 1 (lane h)   -PEG) of polyethylene glycol (see Fig. 1). Aliquots of samples shown in Fig. 1 (lanes a-j) were assayed by hybridization to mSVO1 + mSV02 (closed symbols, ori) or mSV07 + mSV08 (open symbols, control) DNA. B, SV40 chromatin was incubated in the absence of radiolabeled nucleotides for the times indicated before adding 15 pCi of [ C Y -~~P I~C T P and [cY-~'P]~TTP and continuing the incubation for additional 20-min intervals. The concentration of dCTP and dTTP was 10 p~ each. Aliquots were then hybridized to SV40 ori or control DNA as described in Fig. 2. replication in vitro. The low level of DNA synthesis in the ori region observed in the absence of PEG may represent either initiation of DNA replication due to the presence of high-salt nuclear extract, or early RI that were initiated in uiuo, but whose replication forks had not yet progressed outside of the 311-bp region containing ori. PEG stimulation of DNA synthesis at ori did not begin until 10-20 min of incubation, consistent with the appearance of early 32P-RI (Fig. 1). Stimulation of DNA synthesis at regions distant from ori (control, Fig. 3A) did not occur until 30-50 min of incubation as the newly initiated replication forks progressed around the genome. In the absence of PEG, DNA synthesis was more pronounced at replication forks in late RI (control) than at replication forks in ori. This was consistent with DNA synthesis occurring in pre-existing RI that were at all stages of replication with a bias toward late RI, consistent with previous observations that the in vivo pool of SV40 RI contains 2-to %fold more late RI than early RI (26,27).
To determine whether or not new initiation events continued to occur in vitro, high-salt nuclear extracts supplemented with PEG were pulse-labeled with [a!-32P]dNTPs for 20-min intervals from 0 to 80 min after beginning the incubation at 30 "C (Fig. 3B). These data confirmed that the highest rate of ori synthesis occurred from 20 to 40 min after incubation began and that initiation continued to occur at decreasing rates as incubation continued. Similarly, the rate of DNA synthesis in late RI increased continuously as the in vitro reaction progressed, consistent with the appearance of RI* late during the reaction (Fig. 1). The observations made by hybridization of nascent DNA to unique segments of the SV40 genome were confirmed by measuring the relative amounts of DNA synthesis per base pair throughout the SV40 genome. Using the same in vitro conditions described above, [32P]DNA products from various times of incubation were digested with BstNI restriction endonuclease into 15 fragments that varied in length from 54 to 993 bp. In the presence of PEG, at least 65% of the newly synthesized DNA was in the ori region, even after 20 min of incubation (Fig. 4, H S N E + PEG). After 45 min, ["'PIDNA was distributed over 2000 bp of the genome centered around ori, and after 95 min, [32P]DNA was spread through the entire genome, coincident with the large accumulation of RI* and appearance of Form I DNA (Fig. 1, lanes f-j).
In contrast, a lysate of SV40-infected CV-1 cells capable only of continuing replication of RI that were initiated in uiuo showed the opposite effect (Fig. 4, Cell Lysate -PEG). Newly synthesized DNA first appeared throughout the SV40 genome. After 1 h of incubation, newly synthesized DNA had accumulated as an increasing gradient beginning at ori and ending at the termination region. Thus, RI that had initiated replication in vivo were able to complete replication in this cell lysate. To confirm that RI had been converted into Form I DNA, SV4O-infected CV-1 cells were incubated with [3H] thymidine for 4 min to label RI molecules predominantly, and a cell lysate was prepared and incubated in the absence of [a-32P]dNTPs, and the DNA was fractionated by equilibriumcentrifugation in CsC1-ethidium bromide (18). After 5 min, 12% of the viral DNA was Form I, and at 60 min 45% was Form I, in agreement with the 50% conversion originally reported by DePamphilis et al. (18).

Newly Initiated RZ Accumulate in the Presence of Aphidi-
Colin-Partial inhibition of DNA synthesis in vitro by aphidicolin, a specific inhibitor of DNA polymerase-a! and -6 (16,28), also revealed that new rounds of SV40 chromatin replication occurred i n vitro in the presence of PEG. At concentrations greater than 20 p~, SV40 chromatin replication was inhibited 90%. However, in the presence of PEG, 5-15 p~ aphidicolin resulted in a dramatic accumulation of newly replicated DNA that migrated between Form I and Form I11 during electrophoresis in agarose gels (Fig. 5 , lanes c and d).
The mobility of these molecules corresponded to early RI, and digestion of these molecules with various restriction endonucleases revealed that all of the newly synthesized DNA was (HSNE + PEG) as described in Fig. 1 (lanes f-j) for 20 min (solid area), 45 min (medium-shaded area), or 95 min (lightly shaded area).
Cell lysates containing SV40 chromatin (Cell Lysate -PEG), "Experimental Procedures," were incubated under the same reaction conditions as high-salt nuclear extracts except that PEG was not added and incubation times were for 5 min (solid area) or 60 min (lightly shaded area).
[32P]DNA was purified, digested with BstNI restriction endonuclease, and the products fractionated by electrophoresis in a 6% polyacrylamide gel.
[32P]DNA fragments were identified by autoradiography, and the relative amounts of DNA synthesis quantitated by densitometry of the x-ray film. Several film exposures per gel were analyzed to ensure a linear range in film response. The area under each peak was divided by the number of base pairs (bp) in that fragment to compensate for differences in fragment length, and the relative amount of DNA synthesis/base pair in each fragment was normalized relative to the ori-containing fragment. The genomic locations of BstNI cleavage sites in SV40 wt800 are indicated on the abscissa. Fragments containing ori and the termination region (ter) are indicated.
localized within a 600-700-bp region centered at ori (11). Furthermore, hybridization of the [3ZP]DNA in Fig. 5, lane d, to ori and control sequences confirmed the amplification of ori-specific DNA (data not shown). No accumulation of early RI was observed when PEG was omitted from the in vitro reaction (Fig. 5, lanes a and b). T-Antigen is Required for Initiation of SV40 Chromosome Replication in Vitro-Addition of PAb419, a monoclonal antibody directed against SV40 T-ag (24), to the in vitro system resulted in a 90% reduction in DNA synthesis and a marked decrease in the fraction of nascent DNA in RI (Fig. 5, lanes e and f). The remaining prominent [32P]DNA species was RI*, consistent with a requirement for T-ag in the initiation of DNA replication, but not in the elongation of DNA. In order to examine the initiation of DNA replication specifically, sufficient aphidicolin was added to the in vitro system to inhibit DNA synthesis by 50% and thereby cause an accumulation of early RI (11). This accumulation of early RI molecules was specifically inhibited by addition of increasing amounts of PAb419, while the small amount of RI* remains unaffected (Fig. 5, lanes g-1). Addition of either control IgG (Fig. 5, lanes m-r) or nonimmune media from the parental NS-1 myeloma cell line (data not shown) had no effect on SV40 chromatin replication. PAb419 immunoprecipitated Tag and reacted with T-ag in an enzyme-linked immunoadsorption assay, whereas control mouse IgG did not (data not shown).
Initiation of SV40 chromatin replication in vitro required a high-salt nuclear extract of SV40-infected CV-1 cells, but addition of 200 ng or more of immunopurified SV40 T-ag increased nucleotide incorporation 2-fold (Table I). Since addition of T-ag also doubled the amount of newly synthesized DNA that accumulated as early RI in the presence of aphidicolin, as well as increased labeling of BstNI fragment G (data not shown), addition of purified T-ag further stimulated initiation of SV40 chromatin replication in vitro. Apparently, high-salt nuclear extracts are already rich in T-ag. This was demonstrated by the fact that low-salt nuclear extracts supplemented with PEG did not initiate SV40 chromatin replication unless immunopurified T-ag was added (Fig. 3 in Ref. 11). Taken together, these data demonstrate that T-ag is required for initiation of SV40 chromatin replication in uitro.

SV40 Ori is Required for Initiation of DNA Replication in
Vitro-To determine whether or not the conditions that promoted T-ag-dependent DNA synthesis at ori in SV40 chromatin depended on a functional ori sequence, plasmid DNA containing either a functional (pSVori), or nonfunctional (pSVoriA6) SV40 ori sequence was incubated under a variety of in vitro conditions. Replication was measured by converting all DNA products to linear molecules through cleavage at the single EcoRI site, and then digesting with DpnI (6). Plasmid DNA propagated in dam + Escherichia coli is methylated at DpnI cleavage sites which allows them to be cut by DpnI. DNA that undergoes one or more rounds of replication in mammalian cells, which lack this methylase, becomes resistant to cleavage by DpnI. Thus, the amount of plasmid length Form I11 [32P]DNA is proportional to the fraction of DNA that replicated in vitro.
Addition of PEG to a low-salt nuclear extract of SV40infected CV-1 cells stimulated plasmid DNA synthesis 2.2fold to a final rate of 3.3 pmol of dNMPs/h/250 ng of plasmid DNA (Fig. 6, lanes a and 6 ) . This stimulation was dependent upon SV40 ori; pSVoriA6 did not replicate ( l a n e c). Addition of high-salt nucleosol plus PEG stimulated ori-dependent plasmid DNA replication at a rate of 5 pmol of dNMPs/h/ 250 ng of plasmid DNA (lanes e-g). Replication of endogenous 100% DNA synthesis was 6.7 pmol of dNMP incorporated/h/350 ng of SV40 DNA in a 50-p1 reaction at 30 "C. SV40 T-ag (0.1 pg/pl) was purified by immunoaffinity chromatography (23) using PAb419 (24).  Fig. 6 was greatly increased by high-salt nucleosol and PEG, consistent with the experiments described above. These results demonstrated that in uitro conditions necessary for initiation of replication in SV40 chromatin utilized a functional SV40 ori.
To compare ori-dependent plasmid DNA replication under the conditions described above with results reported by others (4)(5)(6)(7)(8), pSVori replication was carried out in the absence of endogenous SV40 chromatin by preparing extracts from uninfected CV-1 cells and supplementing them with purified Tag. pSVori DNA replication exhibited a 10-to 20-min delay (Fig. 7, lanes a-c), as previously reported (29, (lanes p , r,   t, u, and x). Since this Form I11 [32P]DNA was resistant to cleavage by DpnI, it had been replicated in uitro.
Aliquots of pSVori DNA were digested with MboI to ascertain the amount of reinitiation that occurred in the same  (lanes a-i) or else first digested with EcoRI, and DpnI or MboI restriction endonucleases (lanes j-y). Equal amounts of [32P]DNA were applied to lanes a-i, but equal volumes of reaction were applied to lanes j-y. DNA synthesis was linear for at least 3 h with an average rate of incorporation of 12 pmol of dNMPs/h/l30 ng of DNA. Z, ZZ-ZZZ designate the migration positions of pSVori Form Parameters that Affect Initiation of SV40 Chromatin Replication in Vitro-Parameters that affected efficiency of SV40 chromatin replication under conditions that promoted initiation of DNA synthesis a t ori are summarized in Table I. DNA   synthesis depended on the presence  of cytosol, nucleosol, PEG, and SV40 chromatin. Although the rate of DNA synthesis per microgram of DNA decreased with increasing chromatin concentration, 350 ng of chromatin DNA was used routinely because the total amount of DNA synthesis per assay was greatest. The optimal M$+ concentration was 5 mM, which represented about 1 mM uncomplexed M$+. Little effect of [M$+] was observed on elongation or replication of sequences 1800 bp distal from ori (i.e. mSV07 + mSV08). The optimal K+ concentration was 90 mM; initiation was more sensitive than elongation to ionic strength. Acetate was used to adjust the salt concentration since C1inhibits DNA polymerase-a (31).
Adenosine-5'-tetraphosphate-5'-adenosine (Ap4A) binds to DNA polymerase-a and can serve as a primer for templatedependent DNA synthesis (32) by a complex form of this enzyme (33). Furthermore, the intracellular concentration of Ap4A rises from 0.1 to 1 ~L M when quiescent cells are stimulated to divide, and Ap4A levels rise as high as 10 p~ in cells arrested in S-phase (34). However, Ap4A neither increased nor decreased the level of SV40 chromatin replication in uitro.
a-Amanitin concentrations sufficient to inhibit RNA polymerase I1 and I11 had no effect on SV40 chromatin replication in uitro. Similarly, omission of CTP, GTP, and UTP did not affect viral chromatin replication. These data suggest that transcription is not required to initiate SV40 DNA replication.
To determine whether the stimulatory factor associated with high-salt nuclear extracts was associated with the chromatin or with the nucleosol component, nuclear extracts were prepared under either high-salt or low-salt conditions, SV40 chromatin was separated from nucleosol, and individual components were recombined (Table 11). The amount of DNA synthesis originally observed in unfractionated nuclear extracts was reconstituted when the same two fractions were recombined. Total DNA synthesis was 5-fold better in the high-salt nuclear extract than in the low-salt nuclear extract, and only the high-salt nuclear extract specifically stimulated DNA synthesis at ori. These properties were due solely to one or more factors found in high-salt nucleosol; high-salt nucleosol combined with SV40 chromatin from either high-salt or low-salt nuclear extracts stimulated initiation of SV40 chromatin replication whereas low-salt extracts were inactive regardless of which SV40 chromatin fraction was used as substrate. In all instances, recovery of initiation activity was confirmed through analysis of [32P]DNA fractionated by electrophoresis in agarose gels.
High-salt nuclear extracts contain at least one initiation factor in addition to T-ag that is required for initiation of SV40 chromatin replication. This was demonstrated by addition of saturating amounts of purified T-ag to a low-salt nuclear extract (100 ng of T-ag/50 p1 of reaction). These  Table I) were determined from the least squares fit through each data set. The two lines shown in panel B are for DNA polymerase-a and SV40 chromatin; pSVori was essentially the same as SV40 chromatin. Aphidicolin (6 mM) was prepared in dimethyl sulfoxide (11) and BuPdGTP (20 mM) in 0.1 M Hepes (pH 7.8) and 0.1 mM EDTA. Dilution of dimethyl sulfoxide alone into the assay mixture at the levels used to determine the effect of aphidicolin had no effect on DNA synthesis.

TABLE I1
Initiation of SV40 chromatin replication in a reconstituted system High-salt (HS) and low-salt (LS) nuclear extracts were prepared from SV40-infected CV-1 cells as described under "Experimental Procedures." Nuclear extracts were then fractionated into SV40 chromatin (HSC and LSC, respectively) and nucleosol (HSN and LSN, respectively) by sedimentation. conditions never yielded more than 56% of the maximum replication rate observed with high-salt nuclear extracts saturated with T-ag (Tables I and 11); additional T-ag did not further increase the rates of viral DNA synthesis. However, addition of high-salt nucleosol to the low-salt nucleosol supplemented with a saturating amount of T-ag increased the

SV40 Chromatin
Replication in Vitro rate of DNA synthesis a maximum of 2-fold (Table 11). This was equivalent to the maximum levels obtained with highsalt nuclear extracts saturated with T-ag (Table I). Therefore, one or more soluble factors, in addition to T-ag, are eluted from nuclei by extraction with high-salt buffer.
Is DNA Polymerase4 Involved in SV40 Chromatin Replication?-DNA polymerase-a and DNA polymerase4 exhibit similar sensitivities to aphidicolin, but a-polymerase is 1000 times more sensitive than &polymerase to inhibition by p-nbutylphenyl-dGTP (BuPdGTP; 15). Therefore, BuPdGTP should be useful in identifying which DNA polymerase is required for DNA replication. The sensitivity of DNA polymerase-a to this BuPdGTP was confirmed using a-polymerase purified from either CV-1 (38) or HeLa cells (17) on DNase I-activated DNA. Both enzymes were inhibited 50% by 0.1 p~ BuPdGTP in a standard a-polymerase assay (130 p~ dGTP; 39). However, inhibition of SV40 chromatin replication in high-salt nuclear extracts containing PEG required a 1000-fold higher concentration of BuPdGTP (114 p~) to inhibit 50% of the total SV40 DNA synthesis (Fig. SA). Analysis of the DNA synthesized under these conditions by agarose gel electrophoresis did not reveal an accumulation of SV40 early 32P-RI as previously observed with aphidicolin (data not shown). When the same activated DNA substrate was added either to a high-salt nuclear extract or high-salt nucleosol, 50% of the DNA synthesis was inhibited by 4 p~ BuPdGTP, demonstrating that the unexpectedly high concentration of BuPdGTP required to inhibit SV40 chromatin replication was not due to lability of the BuPdGTP in the presence of the cellular extracts. Furthermore, purified apolymerase was inhibited 50% by 1.4 p~ BuPdGTP when assayed under the same conditions used to initiate SV40 chromatin replication in vitro except that cellular and viral fractions were absent to avoid measuring endogenous DNA polymerase activities. These results reveal either the presence of a novel form of DNA polymerase-a that is associated specifically with replicating chromosomes, or the presence of a unique replicative enzyme, such as DNA polymerase4 (16,40), that is resistant to BuPdGTP but sensitive to aphidicolin. pSVori replication in high-salt nucleosol plus PEG was also tested for its sensitivity to BuPdGTP. Remarkably, pSVori replication was reduced 50% by 15 p~ BuPdGTP, intermediate between the sensitivity of SV40 chromatin and DNA polymerase-a. The sensitivity of DNA polymerase-a, pSVori replication, and SV40 chromatin replication to aphidicolin was essentially the same under all conditions; 5-10 p M aphidicolin inhibited dNMP incorporation 50% (Fig. SB), as previously reported for SV40 DNA replication in isolated nuclei and nuclear extracts (39).

DISCUSSION
De novo initiation of DNA replication in viral chromosomes isolated from infected cells was made evident by comparing the properties of subcellular systems that do not initiate replication with the properties of one that does and by comparing the replication of chromosomes with that of purified plasmid DNA. Incubation of SV40 chromatin in a high-salt nucleosol from virus-infected cells, a cytosol fraction, polyethylene glycol, and a buffered salts solution containing all four dNTPs resulted in a rate of DNA synthesis that was 7-to 12fold higher in a 311-bp DNA restriction fragment containing ori than it was at an equivalent length of DNA that was located about 1800 bp from ori (Figs. 2-4). Since ori constitutes a minimum of 64 bp (ori-core) and a maximum of about 155 bp (ori-core plus flanking auxiliary sequences (1,2)), from 21 to 50% of the DNA synthesis in this fragment resided within ori itself. This observation was confirmed by the appearance of early RI with the same electrophoretic mobility that is observed in vivo (slower than Form I DNA but faster than Form I1 DNA, Fig. 1). Therefore, these newly initiated molecules contained covalently closed parental DNA strands that, following DNA purification, expressed the superhelical turns imposed by their organization into nucleosomes (1,3). DNA replication proceeded bidirectionally from ori in either the presence (11) or absence (Fig. 4, top panel) of aphidicolin, but accumulation of early RI was most easily observed in the presence of aphidicolin which rapidly suppressed DNA synthesis proceeding beyond the ori-region (Fig. 5 and Ref. 11). However, when either the PEG or high-salt nuclear extract were omitted from the in vitro reaction, little or no preference for DNA synthesis at ori was observed, no discrete population of early RI was observed either in the presence or absence of aphidicolin, and newly synthesized DNA was predominantly located in middle-to-late RI that had undergone bidirectional replication from ori ( Fig. 1-4 and Ref. 11). These data were consistent with continued elongation of RI that were initiated in vivo in the absence of new initiation events in vivo (3,13,14,18). Therefore, PEG and high-salt nuclear extract appeared to specifically stimulate a low-salt cytoplasmic extract to initiate new rounds of DNA replication.
I n vitro DNA replication in SV40 chromatin also required T-ag which was generally provided by the high-salt nucleosol from infected cells. Replication was inhibited either by omitting this fraction or by addition of a monoclonal antibody directed against SV40 T-ag (Fig. 5). Addition of purified Tag restored most of the activity (11). Several monoclonal antibodies directed against different T-ag domains, including PAb419, inhibited DNA synthesis at ori without inhibiting DNA synthesis at regions distal to ori.' Although it was not possible to alter the ori sequence in endogenous viral chromosomes, it was possible to show that ori was required to replicate purified plasmid DNA under the same conditions used to replicate viral chromatin (Fig. 6). Since plasmid DNA replication, like chromatin replication, required sV40 T-ag (6),* initiation of chromatin replication must also require a functional ori. Furthermore, both oridependent plasmid replication and SV40 chromatin replication continued to replicate in vitro for at least 3 h at similar rates of DNA synthesis (Table I), whereas subcellular systems that simply continue replication of pre-initiated RI expire after 30-60 min (3,13,14,18). Plasmid DNA became MboI sensitive during this period and therefore had reinitiated replication (Fig. 7, lanes j-y) which resulted in accumulation of newly replicated DNA larger than RI* (Fig. 7, lanes a-i). Similarly, chromatin continued to undergo DNA synthesis at ori (Fig. 3B) which also resulted in accumulation of high molecular weight material (Fig. 1, lanes f-0). This high molecular weight DNA was not observed under in vitro conditions that did not permit initiation of replication (Fig. 1, lanes ae). Taken together, the above results suggest that both plasmid DNA and endogenous SV40 chromatin underwent multiple rounds of SV40 ori-dependent, T-ag-dependent DNA replication in vitro. Four significant differences were observed between in vitro replication of chromatin and plasmid DNA. First, initiation of DNA replication in plasmids occurred on non-nucleosomal DNA, because newly initiated plasmid DNA in this (Fig. 7) and other in vitro systems (29,41)  I) was produced, consistent with assembly of nascent DNA into nucleosomes (42), but most of the DNA product was high molecular weight material that may or may not have been assembled into nucleosomes (Fig. 7). In fact, other in vitro conditions that are optimal for assembling newly replicated DNA into nucleosomes are inhibitory for initiation of SV40  DNA replication (41, 42). Second, conditions that were optimal for initiation of DNA replication were suboptimal for separating sibling molecules, although plasmid DNA replication was more efficient at completing replication and producing Form I and I1 DNA than was chromatin replication. PEG specifically stimulated initiation of replication but inhibited completion of replication. PEG inhibited replication of endogenous SV40 chromatin in isolated nuclei where initiation of replication did not occur. Under conditions where initiation did occur, the high molecular weight DNA produced was converted into genomic length monomers upon cleavage at a single restriction endonuclease site (Figs. 1, 6, 7 and Refs. 6, 11). Therefore, this material had completed replication but remained as either interlocked circles or linear concatemers. Third, the data of Yamaguchi and DePamphilis (6) show that increasing concentrations of DNA containing the SV40 oriregion inhibit SV40 ori-dependent plasmid DNA replication but not replication of endogenous SV40 chromatin, indicating that cellular initiation factors are bound more tightly to the endogenous chromosomes. Finally, plasmid DNA replication was 10 times more sensitive to inhibition by BuPdGTP than was SV40 chromatin replication ( Fig. 8.4) even though both were equally as sensitive to inhibition by aphidicolin (Fig.  8B), consistent with a recent report that DNA replication in permeabilized human fibroblasts is about 500 times more resistant to BuPdGTP than is purified DNA polymerase-a (43). Therefore, either plasmid DNA replication is deficient in one or more cofactors that modify the behavior of DNA polymerase-a, or plasmid replication utilizes a mixture of DNA polymerases-a and -6, while chromatin replication utilizes DNA polymerase4 exclusively (16, 40).
Previously published data supporting DNA polymerase-a as the enzyme solely responsible for DNA synthesis during SV40, polyoma virus, and mammalian chromosome replication (1-3) are also consistent with DNA polymerase-& (16,37, 40, 44-46). First, with the exception of BuPdGTP (15, 16, 40), DNA polymerases-a and -6 are remarkably similar in their sensitivities to inhibitors. Second, in virtually all of the studies attempting to correlate polymerase-a activity with DNA replication activity, polymerase-6 activity was not considered. Third, the most direct evidence that DNA polymerase-a is involved in SV40 DNA replication comes from elimination of polymerase-a activity in cellular extracts either by treatment with N-ethylmaleimide (39) or by passing them over immobilized monoclonal antibody SJK-287 directed against polymerase-a (7) and then demonstrating that only purified polymerase-a will restore activity. However, SJK-287 partially cross-reacts with polymerase-& (44), and the absence of polymerase4 in the preparations of polymerase-a was not established. Furthermore, the difference between DNA replication in SV40 ori-containing plasmids and SV40 chromatin in sensitivity to BuPdGTP (Fig. 8) indicates that polymerase-6 may specifically initiate and carry out replication only when the DNA substrate is organized into a unique chromatin structure. Thus, it is possible that DNA polymerase-6 is the replicative enzyme in uiuo. Alternatively, the sensitivity of polymerase-a to BuPdGTP may depend on the structure of the DNA template. This notion is reflected in the decreasing sensitivity of DNA synthesis to BuPdGTP on activated DNA, plasmid DNA, and chromatin (Fig. 8), as well as the fact that utilization of dNTP substrates varies 200-fold depending on whether polymerase-a is assayed on singlestranded DNA, double-stranded DNA, or as "replicase" activity in isolated nuclei (47). A more detailed comparison of the properties of DNA polymerases-a and -6 will be needed to resolve this question.