Purification of Host Cell Enzymes Involved in Adeno-Associated Virus DNA Replication

Abstract

Adeno-associated virus (AAV) replicates its DNA by a modified rolling-circle mechanism that exclusively uses leading strand displacement synthesis. To identify the enzymes directly involved in AAV DNA replication, we fractionated adenovirus-infected crude extracts and tested them in an in vitro replication system that required the presence of the AAV-encoded Rep protein and the AAV origins of DNA replication, thus faithfully reproducing in vivo viral DNA replication. Fractions that contained replication factor C (RFC) and proliferating cell nuclear antigen (PCNA) were found to be essential for reconstituting AAV DNA replication. These could be replaced by purified PCNA and RFC to retain full activity. We also found that fractions containing polymerase δ, but not polymerase ɛ or α, were capable of replicating AAV DNA in vitro. This was confirmed when highly purified polymerase δ complex purified from baculovirus expression clones was used. Curiously, as the components of the DNA replication system were purified, neither the cellular single-stranded DNA binding protein (RPA) nor the adenovirus-encoded DNA binding protein was found to be essential for DNA replication; both only modestly stimulated DNA synthesis on an AAV template. Also, in addition to polymerase δ, RFC, and PCNA, an as yet unidentified factor(s) is required for AAV DNA replication, which appeared to be enriched in adenovirus-infected cells. Finally, the absence of any apparent cellular DNA helicase requirement led us to develop an artificial AAV replication system in which polymerase δ, RFC, and PCNA were replaced with T4 DNA polymerase and gp32 protein. This system was capable of supporting AAV DNA replication, demonstrating that under some conditions the Rep helicase activity can function to unwind duplex DNA during strand displacement synthesis.

Adeno-associated virus (AAV), like all parvoviruses, replicates by a strand displacement method using a hairpined terminal repeat (TR) as a primer. The hairpin primer is used to synthesize a duplex DNA molecule that is covalently closed at one or both ends (Fig. 1). In a process called terminal resolution the TR is cleaved at a unique site on one strand (the terminal resolution site [trs]) and the hairpined TR is repaired to make an open-ended duplex intermediate. The repaired TR is denatured and reannealed to form a double-hairpin intermediate that initiates strand displacement synthesis (Fig. 1, reinitiation), and strand displacement from this intermediate then generates a single-stranded genome that is packaged. The virally encoded Rep78 and Rep68 proteins have been shown to have the site-specific DNA helicase and endonuclease activities required to carry out both terminal resolution and reinitiation in vitro (17, 19, 35). However, it is not yet clear what other enzymes are necessary for AAV DNA synthesis.

FIG. 1.

Mechanism of AAV DNA replication. The scheme illustrates the key steps during AAV DNA replication with NE DNA substrate (NE). The alternative linear dsDNA substrate generated from the PvuII digestion of psub201 is also shown (PvuII). Horizontal arrows indicate 3′ ends, black dots represent the 5′ ends, and trs indicates the terminal resolution sites. Note that the monomer duplex (md) intermediate exists in two forms, with and without a covalently closed, hairpinned end.

AAV requires a helper virus for efficient DNA replication and viral propagation (26). AAV infection in the absence of a helper virus produces little AAV gene expression and virtually no DNA amplification (26). It is believed that only complementary (second-strand) synthesis occurs in the absence of helper virus, although this is often an inefficient process, which takes weeks or months to complete (36). In contrast, in the presence of helper, AAV replication and gene expression initiate within 8 h of cell infection and rapidly generate 500,000 to 1,000,000 viral particles and severalfold more genomes that are not packaged (5, 27). This represents at least a 100-fold increase in AAV DNA replication over a variety of methods that have been shown to induce AAV DNA replication in the absence of helper virus, including UV, heat shock, and topoisomerase inhibitors (51, 52).

It is believed that adenovirus (Ad) is the normal helper in the natural environment because AAV has been isolated most often as a contaminant of clinical Ad isolates (1, 2, 32). Studies of Ad mutants suggest that Ad genes are primarily responsible for insuring AAV gene expression and perhaps for inducing and maintaining cells in S phase (26). The requirement for S phase would be consistent with the fact that the related autonomous parvoviruses typically replicate only after cells enter S phase. With one exception, the Ad DNA binding protein (DBP) gene, none of the Ad genes required by AAV are directly involved in DNA replication. The Ad DNA polymerase and Ad terminal protein are dispensable for AAV DNA replication (20, 21, 27, 38). This suggests that AAV primarily uses cellular replication enzymes when Ad is the helper.

The requirement for Ad DBP in AAV DNA replication has been controversial (6, 20, 21, 27, 38, 46). Mutations in the Ad DBP gene, including deletion of the gene, have a modest effect on DNA replication, lowering replication three- to fourfold in cells infected with AAV and the Ad DBP deletion mutant (6). Since Ad DBP seems to be required for efficient Rep gene transcription (7, 20) and since Rep protein levels are reduced when an Ad DBP deletion mutant is used as a helper (6), it is possible that the only effect of Ad DBP is an indirect effect on gene expression of both Rep and the AAV capsid genes. Several groups, however, have argued that the single-stranded DNA (ssDNA) binding protein may be the critical protein that allows AAV to replicate when helper virus is present (37, 46). First, Ad DBP, cellular replication protein A (RPA), and the herpesvirus UL29 protein, all of which are ssDNA binding proteins, are found to colocalize in infected nuclei with the AAV Rep protein and AAV DNA (14-16, 37, 49). Second, Rep protein has been shown to interact directly with RPA, Ad DBP, and UL29 (15, 37), and this interaction increases AAV DNA replication in vitro about fivefold by increasing processivity (45, 46) and by improving the site-specific endonuclease activity at the trs (37).

Studies of AAV DNA replication in herpesvirus-infected cells have suggested that two sets of proteins are necessary for AAV helper function, the herpesvirus helicase primase complex, which consists of the UL5, UL8, and UL52 proteins, and the herpesvirus ssDNA binding protein, UL29. Mutations in either complex produce profound effects on AAV replication in cell culture (48). Presumably, a cellular DNA polymerase is used in the case of herpesvirus coinfection as it is in the case of Ad. However, mutations in the herpesvirus DNA polymerase have also been shown to reduce AAV DNA synthesis by approximately 3 logs in cell culture (48). More recently, studies of helicase primase mutants have suggested that this complex may function to position AAV DNA at replication centers within the nucleus or that its helicase activity may stimulate AAV DNA synthesis in cell culture or both (34, 37). In addition, in vitro studies have shown that robust AAV DNA replication can be achieved with just the AAV Rep protein, the herpesvirus DNA polymerase complex (UL30/42), and UL29 (47).

Finally, AAV can also replicate its DNA and propagate efficiently in insect cells that have been infected with baculovirus vectors expressing the essential AAV rep and cap genes (23, 43, 44). Such cells can generate viral titers that are equal to those seen in human cells infected with wild-type AAV and either Ad or herpesvirus. This suggests that baculovirus-infected insect cells are a fully permissive environment for AAV DNA replication, but as yet nothing is known about the cellular or baculovirus factors that are essential.

Two groups have developed in vitro AAV DNA replication assays that faithfully recapitulate several aspects of AAV DNA synthesis in vivo (29, 45). These assays use AAV linear DNA that contains either open or covalently closed TRs at both ends (Fig. 1) and depend on the presence of both the AAV TR and Rep78 or Rep68 for efficient DNA synthesis. The assays appear, therefore, to be AAV specific. Both groups have also observed that Ad-infected-cell crude extracts are much more efficient (20- to 50-fold) in synthesizing full-length AAV DNA than uninfected extracts, and Ni et al. (29) have shown that all of the intermediates seen in vivo during AAV DNA replication are recapitulated in vitro. Ward et al. (46) supplemented crude uninfected extracts with purified Ad DBP and demonstrated a four- to fivefold stimulation of AAV replication activity. Ni et al. (28) used antibody inhibition of crude uninfected extracts and reconstitution assays using partially purified uninfected extracts to determine what activities might be necessary for AAV DNA replication. They concluded that RPA, PCNA, replication factor C (RFC), and an aphidocolin-sensitive polymerase could partially reconstitute AAV DNA synthesis in vitro.

In this report, we fractionate Ad-infected-cell extracts and use the in vitro replication assay to determine what cellular or Ad-encoded activities are necessary for reconstituting AAV DNA replication. We find that, in addition to RFC and PCNA, AAV DNA replication requires polymerase δ (Pol δ), but not Pol ɛ or Pol α. Furthermore, as the system is purified, RPA and Ad DBP are no longer essential. We also present evidence that AAV Rep protein is capable of carrying out the strand displacement reaction ahead of the replication fork. Finally, we demonstrate that there are as yet unidentified cellular factors necessary for AAV DNA replication, which appear to be enriched in Ad-infected cells.

MATERIALS AND METHODS

Reagents.

ECL Western blotting detection reagents were purchased from Amersham. [α-³²P]dATP (3,000 Ci/mmol) and [α-³H]TTP were purchased from PerkinElmer. Phosphocellulose (P-Cell; P11) was acquired from Whatman. Bio-Gel HTP was purchased from Bio-Rad. Mono S, phenyl-Sepharose, DEAE-Sepharose, protein A- and protein G-Sepharose, CNBr-activated Sepharose, Q-Sepharose, poly(dA), and oligo(dT) were purchased from Amersham. Ni-nitrilotriacetic acid (NTA) agarose was purchased from QIAGEN. Nucleotides and salmon type III DNA were obtained from Sigma. HindIII, DpnI, and λ DNA were purchased from New England Biolabs.

Antibodies.

Mouse monoclonal PCNA antibody (PC10) was purchased from Santa Cruz Biotechnology. RPA antibody was purchased from either Neomarkers or Oncogene. Pol δ antibodies were purchased from Transduction Laboratories. Pol α antibody was purified from ATCC hybridoma SJK-237-71 with a protein G-Sepharose column. Pol ɛ antibody was obtained from an ascitic preparation of ATCC hybridoma 3C5.1 and purified using a protein A-Sepharose column. Bruce Stillman (Cold Spring Harbor Laboratories) kindly provided RFC antibodies (to the 140-kDa subunit). Ad DBP antibody was provided by Arnold Levine (Princeton University).

Preparation of 293 cell extracts.

Suspension cultures of human 293 cells were grown at 37°C in Jokliks modified essential medium supplemented with 10% bovine calf serum. Ad type 5 stocks were prepared as previously described and assayed on monolayers of 293 cells to determine titers (18). S100 extracts (also called crude extracts) were essentially prepared as described by Ni et al. (29). Briefly, 8 liters of 293 cells (5 × 10⁵/ml) was infected with Ad (multiplicity of infection [MOI] of 5) and cultured for 40 h at 37°C. The infected cells were harvested at 1,000 × g for 10 min and then subjected to hypotonic lysis (20 mM HEPES, pH 7.5, 5 mM KCl, 1.5 mM MgCl₂, 1.0 mM dithiothreitol [DTT]) and Dounce homogenization. NaCl was added to 0.2 M final concentration, and the lysed cells were incubated on ice for 15 min. Extracts were centrifuged 100,000 × g for 1 h at 4°C. The supernatants were frozen in liquid nitrogen and stored at −80°C. Uninfected-cell extracts were prepared in the same manner except the cells were grown to a density of 1 × 10⁶ cells/ml. All subsequent fractionation was performed at 4°C.

P-Cell chromatography of crude extract.

P-Cell fractionation of the S100 extract was performed as described by Tsurimoto and Stillman (40, 41) (Fig. 2). A column was equilibrated with buffer A (25 mM Tris-HCl, pH 7.5, 1 mM EDTA, 10% glycerol, 0.01% NP-40, 1 mM DTT, 0.1 mM phenylmethylsulfonyl fluoride, 0.5 μg/ml leupeptin, 0.7 μg/ml pepstatin A) containing 0.2 M NaCl before loading the extract. The flowthrough fraction was designated fraction I. The column was washed with at least 5 volumes of the equilibration buffer before the elution of the bound proteins with a four-step gradient. Elution was performed using a step gradient with NaCl concentrations of 0.33 M (fraction IIA), 0.4 M (fraction IIB), 0.66 M (fraction IIC), and 1.0 M (fraction IID). The peak fractions were pooled and dialyzed against buffer A containing 25 mM NaCl and 20% sucrose.

FIG. 2.

Fractionation scheme of cellular replication proteins. The P-Cell and Q-Sepharose chromatography steps used to isolate DNA replication factors from Ad-infected-cell extracts (S100) are outlined. The protein fractions generated following each chromatography step (P-Cell I, IIA, IIC, and IID; Q-Sepharose, IA, IB, IC, and ID) are indicated. The NaCl concentrations used to elute each fraction are also noted. In some instances the 0.4 M NaCl elution was not performed and the elution with 0.66 M NaCl was termed fraction IIC′ (not shown). Additional chromatography steps used for the purification of Pol δ, Pol ɛ, and RFC are also shown.

Q-Sepharose chromatography of fraction I.

Fraction I was further purified on a Q-Sepharose column (Fig. 2) using a modified method of Prelich et al. (30). The column was preequilibrated with buffer A containing 0.1 M NaCl. The sample was also adjusted to 0.1 M NaCl. The flowthrough (fraction IA) was collected and the column washed with the equilibration buffer. The bound proteins were eluted with a step gradient of 0.25 M (IB), 0.45 M (IC), and 1.0 M (ID) NaCl. The peak fractions were pooled and dialyzed as described above.

Replication proteins.

Baculovirus-expressed AAV Rep78 was purified as previously described (28). Baculovirus clones of RPA subunits were kindly provided by Jesper Christensen (8). The baculoviruses were used to infect Sf9 cells for expression of RPA. Typically, the three baculoviruses containing a His-tagged RPA70, RPA34, or RPA12 were used to infect 1 liter of Sf9 cells (2 × 10⁶ cells/ml) with a MOI of 5 (ratio of viruses, 2:2:3). The cells were harvested by centrifugation at 500 × g 40 h postinfection. The RPA was then purified with a Ni-NTA agarose column equilibrated with buffer 1 (25 mM HEPES-KOH, pH 7.7, 1.5 mM MgCl₂, 5 mM imidazole, 10% glycerol, and 0.4 M NaCl) and eluted with 100 mM imidazole in buffer 1. RPA-containing fractions were then pooled and loaded onto a Mono Q column equilibrated with buffer 1 and eluted with a 50 to 500 mM NaCl gradient. PCNA was purified from a bacterial expression clone, as described by Fien and Stillman (13). RFC was purified as described by Tsurimoto and Stillman (41) (Fig. 2). Highly purified RFC (the ssDNA chromatography fraction) was used in some assays to confirm the requirement for RFC but, typically, RFC purified from the second P-Cell chromatography was used in replication assays, as this fraction is stably stored. Ad DBP was purified from Ad-infected-293 cell extracts as described by Brush et al. (4).

DNA polymerase purification.

DNA Pol α was purified from the P-Cell IIB fraction of an Ad-infected-293 cell extract using immunoaffinity chromatography (SJK237-71 antibody) as described previously (42) except that the antibody was coupled with CNBr-activated Sepharose. The final concentration was 15 μg/ml, with a primase activity of 7.9 nmol/h/μl. Immunodepletion of Pol α was performed by chromatography of an Ad-infected cell extract on a Pol α immunoaffinity column (SJK237-71 antibody) as previously described (25). Western blot and primase assays were used to confirm Pol α depletion.

DNA Pol ɛ was purified from the P-Cell IIA fraction as described by Syvaoja and Linn (39) (specific activity, 348 nmol/h/mg [³H]dTTP incorporated). Pol ɛ did not contain any Pol α or δ, as judged by Western analysis. Pol δ was partially purified from P-Cell IIA fraction on a DEAE-Sepharose column as for the Pol ɛ purification. Fractions that showed significant absorbance at 280 nm were examined by Western analysis and tested for polymerase activity using activated DNA. The Pol δ fractions did not contain either Pol α or Pol ɛ, as seen by Western analysis. Baculovirus expression clones of the Pol δ subunits were kindly provided by Jesper Christensen (8). Typically, the four baculoviruses individually containing Pol δ 125, His-tagged Pol δ 66, Pol δ 50, and Pol δ 12 were used to infect 1 liter of Sf9 cells (2 × 10⁶ cells/ml) with a MOI of 5 (ratio of viruses, 3:2:1:2). The cells were harvested by centrifugation at 500 × g 40 h postinfection. The Pol δ complex was purified with P-Cell chromatography as described above for crude 293 cell extracts except that the polymerase was eluted using 0.45 M NaCl (essentially pooling fractions IIA and IIB). This fraction was then purified on a Ni-NTA column as described for RPA.

AAV DNA replication assay.

No-end (NE) substrate was prepared as previously described by Snyder et al. (35) (Fig. 1). In some cases, a linear double-stranded DNA (dsDNA) substrate was used. This substrate was generated from a PvuII digest of the plasmid psub201 as described by Samulski et al. (31).

The standard replication reaction was done as previously described by Ni et al. (28, 29), and the reaction mixture contained, in 15 μl, 30 mM HEPES, 7 mM MgCl₂, 0.5 mM DTT, 4 mM ATP, 6 μCi [α-³²P]dATP, 100 μM deoxynucleoside triphosphate, 40 mM creatine phosphate, 33 ng creatine phosphokinase, 0.05 μg NE, and 0.1 to 1.0 μg Rep78 and crude extract or purified replication proteins. Reaction mixtures were incubated for 3 h at 37°C, and then reactions were stopped with 35 μl of stop solution (0.3% sodium dodecyl sulfate [SDS], 17 mM EDTA, 0.7 μg/ml proteinase K). The samples were extracted with phenol-chloroform and precipitated with ethanol. The DNA was digested with DpnI for 1 h at 37°C and fractionated on a 0.8% agarose gel. The gels were dried and exposed to X-ray film and/or a Fuji phosphorimager screen. Where indicated in the figure legends, an aliquot of the reaction products was also was spotted onto DE-81 filter paper. The paper was then washed with 0.5 M NaH₂PO₄ before counting the dAMP incorporation with a scintillation counter.

Polymerase assays.

The reaction mixture contained, in 15 μl, 30 mM HEPES, 7 mM MgCl₂, 10% glycerol, 0.5 mM DTT, 4 mM ATP, 30 ng poly(dA)/oligo(dT) (Pharmacia; A-T ratio, 1:1), 0.75 μCi [³H]dTTP, and 1 to 20 μg of polymerase. Reaction mixtures were incubated at 37°C for 30 min before the reaction was stopped by chilling on ice. An aliquot (5 to 8 μl) was spotted onto DE-81 paper, washed, and quantified in a scintillation counter. In some cases activated salmon sperm DNA (Pharmacia, 50 ng DNA per reaction) was used as a substrate in place of the poly(dA)/oligo(dT). Pol α primase assays were performed as previously described by Chui and Linn (9). One unit of polymerase activity equals 1 nmol [³H]dTTP/h.

RESULTS

P-Cell chromatography of Ad-infected-cell extracts.

To identify viral and cellular replication factors that play a role in AAV DNA replication, we fractionated Ad-infected-cell extracts using a purification scheme similar to that used to study simian virus 40 (SV40) DNA replication (Fig. 2). Ad-infected-cell extracts were used because they consistently achieved 10- to 20-fold-higher synthesis of full-length DpnI-resistant AAV DNA in the Rep dependent AAV DNA replication assay. This mirrored a better-than-100-fold difference in DNA replication seen in vivo between uninfected and Ad-infected cells. Extracts derived from Ad-infected 293 cells were subjected to P-Cell chromatography to generate fractions I, IIA, IIB, IIC, and IID. In some instances the 0.4 M NaCl elution was not performed and the fraction eluted with 0.66 M NaCl was termed fraction IIC′. IIC′, therefore, contained what was usually in IIB and IIC.

The location of known cellular DNA replication factors within each fraction was then determined by Western blot analysis (Fig. 3A). As expected, RPA and PCNA were observed in fraction I and RFC was present within fraction IIC (Fig. 3A). Ad DBP was mostly located in fraction IIB but was also present in IIA and IIC. In addition, we analyzed each P-Cell fraction for DNA polymerase activity using activated salmon sperm DNA as the template. The majority of the DNA polymerase activity, as judged by incorporation of radioactive deoxynucleotides into DNA, was recovered in fraction IIA; little or no activity could be detected in fraction I, IIC, or IID. These results were consistent with the known fractionation patterns of cellular and viral DNA polymerases on P-Cell. A silver-stained SDS-polyacrylamide gel of the P-Cell fractions is shown (see Fig. 5A).

FIG. 3.

P-Cell separation of replication proteins from Ad-infected cells. (A) Western blots of fractions I, IIA, IIB, and IIC with Pol δ, Pol ɛ, Pol α, Ad DBP, RFC, RPA, and PCNA antibodies. (B) Reconstitution of AAV DNA replication in vitro using fractionated Ad-infected-cell extracts. Standard replication reaction mixtures (30 μl) contained either crude Ad-infected-cell extract (C; 200 μg) or P-Cell fractions (I, 160 μg; IIA, 36 μg; IIC′, 30 μg; IID, 9 μg) and Rep78 (3.4 μg), as indicated. Reaction mixtures were incubated for 2 h at 37°C, processed, and analyzed as described in Materials and Methods. The DNA products from each reaction were digested with DpnI and analyzed on 0.8% Tris-borate-EDTA (TBE)-agarose. md and dd indicate monomer duplex and dimer duplex DNA species, respectively, which are resistant to DpnI digestion. DNA products sensitive to DpnI digestion are denoted by the black line. (C) Optimization of AAV DNA replication with RFC and P-Cell fractions I and IIA. Standard replication reaction mixtures (15 μl) contained either crude Ad-infected-cell extract (C; 100 μg) or P-Cell fraction I (80 μg), RFC (0.5 μg), Rep78 (1.7 μg), and various amounts of fraction IIA (IIA total protein concentration, 6 mg/ml), as indicated. Reaction mixtures were incubated for 2 h at 37°C, processed, and analyzed as described in Materials and Methods. Half of each reaction mixture was subjected to DpnI digestion and analyzed on 0.8% TBE-agarose; the remaining half was used to quantify [³²P]dAMP incorporation using a DE-81 filter binding assay (expressed as pmol of dAMP incorporated per 15 μl reaction mixture). The migration pattern and fragment size of radiolabeled HindIII-digested lambda DNA are indicated on the left.

FIG. 5.

(A) Silver-stained polyacrylamide gel of fractions used in this study and purified from Ad-infected-cell extracts. Lane C, Ad-infected-cell extract (2 μl, 92 μg); IIA, 5 μl, 55 μg; IIB, 5 μl, 30 μg; IIC, 5 μl, 25 μg; I, 5 μl, 115 μg; IA, 5 μl, 34 μg; IB, 5 μl, 40 μg; IC, 5 μl, 4.7 μg. (B) Silver stain of an SDS-polyacrylamide gel of purified protein fractions: DEAE Pol δ from uninfected extract (15 μl, 16.5 μg), Ad BDP (5 μl, 0.65 μg), PCNA (2.5 μl, 0.7 μg), baculovirus-expressed RPA (20 μl, 5 μg), RFC from Ad-infected-cell extract (40 μl, 0.6 μg), Pol ɛ from uninfected extract (40 μl, 3.8 μg), and baculovirus-expressed Pol δ (30 μl, 3.2 μg). Arrows indicate the molecular masses of the expected subunits in each complex.

Reconstitution of AAV DNA replication with P-Cell fractions I, IIA, and IIC.

Replication reactions were performed with the substrate NE DNA, AAV Rep78 purified from baculovirus-infected SF9 cells, and various combinations of the P-Cell fractions described above. NE DNA contained no free 3′OH ends (both ends are covalently closed in the hairpin configuration) and was shown previously by us to require Rep for initiation of DNA replication from the AAV TR (29, 35). The use of NE DNA helped to insure that the replication was TR (that is, ori) specific. However, similar results were obtained when a linear duplex AAV DNA genome that contained open ends was used as shown by others (45).

To measure the extent of DNA replication, the DNA products from each reaction were resolved on neutral agarose following restriction analysis with DpnI. As seen in Fig. 3B, full-length AAV DpnI-resistant monomer and dimer products were readily detectable in reactions carried out with crude Ad-infected-cell extracts. We also observed radiolabeled DpnI-sensitive DNA from these reactions. The amount of these unmethylated products was variable, and we are not certain what caused them. They may reflect initiation events that do not lead to elongation (Fig. 3B) or interruption of elongation by DNA endonucleases or proteins (e.g., transcription factors) that block elongation.

When crude extracts were replaced with various P-Cell fractions, we discovered that fractions I, IIA, and IIC′ were all that was required for recovery of significant levels of AAV DNA replication (Fig. 3B). These findings parallel the known requirements for reconstitution of SV40 DNA replication using similar P-Cell fractions derived from 293 cellular extracts. The absence of Rep78 protein gave no replication with the NE substrate as expected (Fig. 3B, lane 8).

Studies of SV40 DNA replication identified RFC as the essential component in fraction IIC (41). We therefore, replaced fraction IIC with RFC that was purified from uninfected 293 cells. The purified RFC restored the level of DNA replication to that of the crude extract once an appropriate level of polymerase fraction IIA was used (Fig. 3C, compare lanes 1, 8, and 9). This demonstrated that RFC was probably the only component from fraction IIC that was required for DNA replication. In the absence of RFC there was a small amount of monomer duplex product (Fig. 3C, lane 11) which was DpnI resistant. We believe that this was probably due to a small amount of RFC contamination in fraction IIA or possibly incomplete DpnI digestion of the DNA.

P-Cell fraction I.

To analyze fraction I further, we used Q-Sepharose chromatography to generate four fractions (IA, IB, IC, and ID; Fig. 2 and 3). Western blot analysis revealed that fraction IB contained RPA and fraction IC contained PCNA (Fig. 4A). There was no Ad DBP, Pol δ, or Pol ɛ in fraction I (Fig. 3A) or in the fractions derived from fraction I. Silver-stained SDS-polyacrylamide gel for these fractions (IA, IB, and IC) as well as the P-Cell II fractions is shown in Fig. 5A. In our replication reconstitution analysis, we found that only IA and IC were essential for DNA replication (Fig. 4B, compare lanes 5 and 7 with lanes 3 and 8). When IA and IC were combined with Rep78, RFC, and polymerase fraction IIA, nearly all of the replication activity was restored. Fraction IB, which contained all of the RPA (Fig. 4A), was not required.

FIG. 4.

Q-Sepharose chromatography of P-Cell fraction I from Ad-infected cells. (A) Western blot analysis of Q-Sepharose fractions IA, IB, IC, and ID. RPA (lane 1, 0.7 μg), PCNA (lane 2, 1.7 μg), P-Cell fraction I (400 μg), and Q-Sepharose fractions IA (250 μg), IB (188 μg), IC (60 μg), and ID (6.3 μg) were resolved on 10% SDS-polyacrylamide and then transferred to nitrocellulose. The nitrocellulose filter was probed with RPA (70-kDa subunit) and PCNA monoclonal antibodies. (B) Reconstitution of AAV DNA replication with Q-Sepharose fractions. Standard replication reaction mixtures (15 μl) contained P-Cell fractions I (80 μg) and IIA (24 μg); RFC (0.5 μg); Rep78 (1.7 μg); and Q-Sepharose fractions IA (50 μg), IB (38 μg), IC (12 μg), and ID (1.3 μg), as indicated. Reaction mixtures were incubated for 2 h at 37°C and processed as described in Materials and Methods. md and dd indicate monomer duplex and dimer duplex DNA species, respectively. (C) Replacement of Q-Sepharose fraction IC with PCNA. Standard replication reaction mixtures (15 μl) contained P-Cell fraction IIA (24 μg), RFC (0.5 μg), Rep78 (1.7 μg), Q-Sepharose fractions IA (50 μg) and IC (12 μg), RPA (0.7 μg), and PCNA (lanes 6 to 10, 17 ng, 34 ng, 85 ng, 170 ng, and 340 ng, respectively; lanes 11 and 12; 340 ng), as indicated. Reaction mixtures were incubated for 2 h at 37°C and processed as described in Materials and Methods.

Since fraction IC contains PCNA, we performed a substitution experiment where we replaced fraction IC with pure PCNA (Fig. 4C). PCNA successfully restored the activity of the IC fraction. However, fraction IA was still required for full activity (Fig. 4C, compare lanes 3 and 4, and lanes 10 and 11). Examination of fraction IA by Western analysis showed that this fraction was devoid of proteins likely to be involved in leading strand DNA replication (Fig. 3A and 4A). Replacement of IA with purified RPA was unsuccessful at recovering the activity of the IA fraction, so the IA component was not likely to be an ssDNA binding protein (Fig. 4C, lane 12). Thus, PCNA and an unknown factor(s) in fraction IA are essential for reconstituting AAV DNA replication from fraction I.

Fraction IA.

Fraction IA contained a small amount of material that reacted with Pol α antibody (Fig. 3A). In our previous work, high concentrations of Pol α neutralizing antibody appeared to inhibit AAV replication in vitro. Although there is no role for a primase in the known mechanism for AAV DNA replication, it was conceivable that Pol α or a protein complexed with a subset of Pol α might be necessary for AAV DNA replication. We therefore, used a monoclonal antibody to Pol α to quantitatively immunodeplete Pol α from an Ad-infected-cell crude extract. The extract that was depleted of Pol α did not show any decrease in AAV replication (data not shown). Moreover, addition of immunoaffinity-purified and concentrated Pol α back to the depleted extract did not enhance replication activity. Thus, neither the absence of Pol α nor the addition of factors that might purify when complexed to Pol α appeared to affect replication activity. Pol α was also unable to reconstitute replication when added with other purified replication factors to a replication assay (data not shown). Therefore, we concluded that Pol α was not the factor in fraction IA that seemed to be required for AAV DNA replication. Using similar techniques, we examined fraction IA for the presence of a variety of other DNA related enzymes. These included the BLM helicase, the WRN helicase, high mobility group proteins 1 and 2 (HMG1 and -2), and Cdk1. None of these proteins, when added back to the reaction by themselves or inhibited with a commercial inhibitor, were able to substitute for fraction IA (data not shown). Rep52 also did not reconstitute the activity of fraction IA (data not shown).

DNA polymerase required for AAV DNA replication.

To identify the DNA polymerase(s) necessary for AAV DNA replication, fraction IIA was fractionated further. The DNA polymerase activity of IIA was resolved on Q-Sepharose using a linear gradient, and three DNA polymerase peaks were pooled, IIA1, IIA2, and IIA3 (Fig. 6A). The bulk of the polymerase activity was in fractions IIA2 and IIA3, and these were tested for their dependence on PCNA (Fig. 6B and C). The IIA2 fraction displayed PCNA-dependent polymerase activity, which is consistent with the presence of Pol δ (Fig. 6B), and the IIA3 fraction displayed PCNA-independent polymerase activity, which suggests Pol ɛ activity (Fig. 6C). When the Q-Sepharose fractions were used in reconstitution assays, it was clear that only fraction IIA2 had significant DNA replication activity on the AAV template (Fig. 6D). This was confirmed further with a titration of fraction IIA2 which showed increased replication levels with increasing amounts of fraction IIA2 (Fig. 6E). These data suggested that Pol δ is the active polymerase for AAV DNA replication.

FIG. 6.

Q-Sepharose chromatography of P-Cell fraction IIA from Ad-infected cells. (A) DNA polymerase activity profile of Q-Sepharose fractions. DNA polymerase activity of each fraction was assayed by measuring the incorporation of dAMP into activated DNA. Standard reaction mixtures (25 μl) contained 3 μl of each fraction and were incubated for 30 min at 37°C. Incorporation of dAMP into DNA was quantified using a DE-81 filter binding assay and is expressed as dAMP (pmol) incorporated per 25 μl reaction mixture. IIA1, IIA2, and IIA3 represent pooled fractions used to reconstitute DNA replication. (B) PCNA-dependent and independent DNA polymerase activities of Q-Sepharose fraction IIA-2 as described in Materials and Methods (PCNA, 255 ng). (C) PCNA-dependent and independent DNA polymerase activities of Q-Sepharose fraction IIA-3 as described in Materials and Methods (PCNA, 255 ng). (D) Reconstitution of AAV DNA replication with Q-Sepharose fractions. Standard replication reaction mixtures (15 μl) contained P-Cell fraction I (80 μg); RFC (0.5 μg); Rep78 (1.7 μg); IIA (24 μg); or Q-Sepharose fractions IIA1 (5 μg), IIA2 (10 μg), and IIA3 (7.5 μg), as indicated. md and dd indicate monomer duplex and dimer duplex DNA species, respectively. (E) Titration of Q-Sepharose fraction IIA2. Standard replication reaction mixtures (15 μl) were as described above except that various amounts of IIA2 (lanes 5 to 8, 3 μg, 6 μg, 12 μg, and 18 μg, respectively) were used to substitute for fraction IIA. All reaction mixtures were incubated for 4 h at 37°C and processed as described in Materials and Methods. [³²P]dAMP incorporation was measured as described for Fig. 3.

Fraction IIA, as shown in Fig. 3A, contains Ad DBP. In our attempts to further purify Pol δ from fraction IIA we were never successful in separating Pol δ from Ad DBP. Therefore, to eliminate the possible contribution of Ad DBP, we fractionated uninfected-293 cell extracts. A DEAE-Sepharose fractionation of uninfected-cell IIA proteins yielded two peaks of polymerase activity (Fig. 7A). Western analysis showed that peak 1 is due to Pol δ activity and that peak 2 activity is mostly due to Pol ɛ but also contained some Pol δ (Fig. 7B). Reconstitution of in vitro DNA replication showed that the Pol δ peak fractions (pool of fractions 10 and 11) restored DNA replication (Fig. 7C). The Pol ɛ-containing fraction showed only a small amount of DNA replication activity. To determine if this activity was due to Pol ɛ or the contaminating Pol δ, we further purified Pol ɛ (Fig. 2) to remove any traces of Pol δ (Fig. 8A) and then examined DNA replication (Fig. 8B, lanes 12 to 18). Unlike purified Pol δ, the purified Pol ɛ did not show any AAV DNA replication. Figure 8B also shows that Pol δ requires the factors that we had already identified from previous fractions: PCNA, RFC, and IA. A silver-stained polyacrylamide fractionation gel of the purified proteins is shown in Fig. 5B.

FIG. 7.

DEAE-Sepharose fractionation of DNA polymerases from P-Cell fraction IIA of uninfected-293 cell extract. (A) DNA polymerase activity of fractions from DEAE Sepharose (see Materials and Methods). (B) Western blot of DEAE fractions using anti-Pol δ and anti-Pol ɛ antibodies which were probed for sequentially without stripping the blot between antibodies. (C) In vitro AAV DNA replication. Standard replication reaction mixtures (15 μl) contained, as shown, the following components: crude extract (60 μg), RFC (0.01 μg), Rep78 (0.2 μg), PCNA (0.4 μg), IA (6 μg), DEAE-Sepharose fractions 10 and 11 of the Pol δ pool (0.44 mg/ml), and fractions 13 to 15 of the Pol ɛ pool (2.6 mg/ml), as indicated. md and dd indicate monomer duplex and dimer duplex DNA species, respectively, which are resistant to DpnI digestion.

FIG. 8.

Comparison of DNA Pol δ and Pol ɛ. (A) Western blot using anti-Pol δ and anti-Pol ɛ (blot probed sequentially with antibodies) shows that fractions of the DEAE fraction of Pol δ (shown in Fig. 7) did not contain Pol ɛ and that the Mono S fraction of Pol ɛ did not contain Pol δ. (B) In vitro AAV DNA replication with various purified DNA replication components. Standard replication reaction mixtures (15 μl) contained crude extract (60 μg), RFC (0.01 μg), Rep78 (0.2 μg), PCNA (0.4 μg), IA (6 μg), DEAE Pol δ (2 μg, 16 nmol [³H]dTTP/h/μl), and Mono S Pol ɛ (0.4 μg, 348 nmol [³H]dTTP/h/μl), as indicated. md and dd indicate monomer duplex and dimer duplex DNA species, respectively.

To confirm that Pol δ is indeed the polymerase required for replication, Pol δ purified from uninfected 293 cell extracts was replaced with Pol δ purified from baculovirus-infected SF9 cells. The baculovirus-expressed Pol δ was capable of substituting for the activity of the 293 cell extract Pol δ (Fig. 9), and the amounts of AAV replication with Pol δ from the 293 cell extract (0.38 pmol [³H]dTTP/U of enzyme) and the baculovirus-expressed Pol δ (0.39 pmol ³H-dTTP/U of enzyme) were equivalent. These findings are consistent with Pol δ, and not Pol α or ɛ, as the AAV replicative polymerase.

FIG. 9.

Replacement of Pol δ purified from 293 cell extracts with baculovirus-expressed Pol δ in the in vitro DNA replication assay. Standard replication reaction mixtures (15 μl) contained crude RFC (0.01 μg), Rep78 (0.2 μg), PCNA (0.4 μg), IA (6 μg), DEAE Pol δ (0.43 mg/ml, 7 nmol [³H]dTTP/h/μl), baculovirus-expressed Pol δ (0.024 mg/ml, 1.7 nmol [³H]dTTP/h/μl), RPA (30 to 260 ng), and Ad DBP (15 to 130 ng), as indicated. The amount of ³²P incorporation for the monomer AAV replication species was determined by phosphorimager analysis and is indicated for each lane. md and dd indicate monomer duplex and dimer duplex DNA species, respectively.

ssDNA binding proteins.

Since we avoided using Ad DBP by using a Pol δ preparation from uninfected-cell extracts and since the only partially purified fraction that remained (fraction IA) was negative for both RPA and Ad DBP, our reconstituted assay did not contain ssDNA binding protein. Because several groups have suggested that ssDNA binding proteins were necessary for AAV DNA replication (28, 37, 46), we examined the effect of adding back either purified RPA or purified Ad DBP. Addition of purified RPA was found to enhance replication ∼1.5-fold (Fig. 9). Similarly, Ad DBP stimulated replication to the same extent as RPA (Fig. 9). However, in all experiments with Ad DBP, we saw inhibition of DNA replication at the highest DBP concentration (Fig. 9, lane 12). As we could not detect other proteins in our purified Ad DBP preparations, we do not believe this was due to a contaminating inhibitor in the purified Ad DBP. We also tested if the addition of either Ad DBP or RPA to Pol ɛ could reconstitute replication, and it did not (data not shown). Therefore, we concluded that DNA binding proteins appeared to increase DNA replication in vitro approximately 50%, but were not essential for activity.

The DNA helicase requirement.

During AAV DNA replication, there are three steps that could require a DNA helicase activity. The first is the terminal resolution step, in which Rep creates a single-stranded bubble at the terminal resolution site and cleaves it (3). Following this event, Rep is also capable of unwinding the TR sequence so that DNA polymerase can fill in the TR to synthesize a complete duplex molecule (35). The second step involves unwinding the duplex TR to form a hairpin that initiates strand displacement synthesis in a process we have called reinitiation (Fig. 1). Finally, a helicase (or DNA binding protein) is necessary to unwind the duplex molecule during strand displacement synthesis. These last two steps are coupled; if the TR is not denatured and reannealed to form a hairpin primer, strand displacement cannot proceed. We have shown previously that Rep protein can bind to a linear TR and unwind it (53).

To determine if Rep68 is able to unwind double-stranded linear AAV DNA and allow strand displacement synthesis to take place, we incubated Rep68 with PuvII-ClaI-digested psub201 DNA. PuvII-ClaI-psub201 DNA digests contain a linear AAV genome, which is missing only 13 bases from either end, and two cloning vector fragments (Fig. 1). The two vector fragments were purposely not separated from the AAV DNA to serve as internal controls because they, unlike the AAV DNA, do not have TR sequences and thus should be incapable of undergoing self-primed replication. As a result, the two vector fragments would be sensitive to DpnI digestion, whereas any replicated AAV DNA would become resistant to the treatment. The missing 13 bases of AAV sequence have been shown previously not to affect Rep binding or DNA replication activity, and they are replaced immediately after the hairpin primer is extended (31). We have shown previously that Rep68 and Rep78 have similar biochemical activities and are essentially indistinguishable in the in vitro DNA replication assay (28).

As shown in Fig. 10A, neither Ad-infected- nor uninfected-cell extracts were capable of stimulating strand displacement synthesis on linear AAV DNA in the absence of Rep protein. However, when Rep68 was added to the reactions, AAV DNA replication took place, suggesting that Rep68, but not cellular helicases, was required to unwind the TR to form the hairpin primer and possibly unwind the remainder of the DNA ahead of the replication fork.

FIG. 10.

Examining the helicase activity of Rep in AAV DNA replication using linear dsDNA substrate (0.1 μg of PvuII-digested psub201). (A) Standard replication assay with either Ad-infected-cell or uninfected-cell crude extract (487 μg) and Rep68 (67 μg), as indicated. (B) Standard replication assay with T4 DNA polymerase (1.5 U; New England Biolabs), gene 32 protein (gp32) (1 μg; Pharmacia), and Rep68 (13 μg), as indicated. md indicates the monomer duplex DNA species.

To determine if Rep alone was capable of unwinding the genome during strand displacement synthesis, we designed an artificial replication system that contained only purified Rep68, T4 DNA polymerase, and the T4 DNA binding protein, gene 32 protein (gp32). As shown in Fig. 10B, the addition of Rep68 to the reaction enabled strand displacement synthesis of AAV DNA by T4 DNA polymerase to proceed efficiently. Omission of gp32 showed that Rep68 and T4 DNA polymerase alone were sufficient to generate significant amounts of fully replicated AAV DNA (Fig. 10C). The level of monomer duplex DNA synthesized with T4 polymerase and gene 32 protein was approximately 30% of that seen with Ad-infected-cell crude extracts. Finally, no incorporation of labeled nucleotides was seen in the two vector fragments that were also present as templates in the reaction mixture. This demonstrated that the replication reaction was specific for DNA containing an AAV TR.

DISCUSSION

The replicative polymerase.

Pol δ was an obvious candidate for the replicative polymerase required for AAV DNA replication. Studies of SV40 DNA replication in vitro had shown that Pol δ was the leading strand DNA polymerase, and AAV DNA replicates exclusively by a leading strand mechanism. The requirement for PCNA and RFC is also consistent with Pol δ as the replicative polymerase. PCNA was originally identified as a Pol δ processivity factor, and RFC assembles PCNA onto 3′OH primers. Our evidence that these two factors were necessary for in vitro AAV DNA replication in Ad-infected-cell extracts confirms our earlier work with uninfected extracts.

Nevertheless, SV40 DNA can be replicated in vitro with Pol α alone (50), and PCNA has also been shown to be an accessory protein for Pol ɛ (33). Furthermore, in our previous work with uninfected extracts we could not substitute purified Pol δ, Pol α, Pol ɛ, or any combination of these for the polymerase fraction (28). Therefore, it was important to identify which polymerase fractionated with AAV DNA replication activity, and that proved to be Pol δ. Not only did Pol δ emerge as the only polymerase that fractionated with AAV replication activity, but it also functioned when it was purified from heterologous sources and added back as a purified protein.

Finally, it was important as well to eliminate the possible contribution of Pol α to AAV DNA replication. Pol α was the first DNA polymerase that was shown to be capable of replicating a parvovirus genome in vitro (24), and it has been suggested that the autonomous parvovirus minute virus of mice (MVM) may have an internal bidirectional origin for DNA replication that may function in addition to its terminal sequences (11, 12).

The identification of Pol δ as the replicative AAV polymerase is consistent with two other observations. Christensen and Tattersall (8) found that purified Pol δ, RPA, PCNA, and RFC were the minimum cellular factors required for rolling-circle replication from a 3′-dimer junction of MVM when added back to an in vitro reaction mixture that also contained the MVM NS1 protein and the cellular PIF protein. MVM dimer junctions are believed to be one of the key replicative intermediates that lead to the synthesis of progeny DNA. In addition, Jurvansuu et al. (22) also implicated Pol δ in an AAV-induced repair reaction. In these experiments UV-treated AAV infection arrested cells at G₂/M. When infected cells were processed for chromatin immunoprecipitation, AAV DNA was found to be associated with RPA and Pol δ but not Pol α or ɛ.

The identification of Pol δ as the replicative polymerase would also explain why AAV is capable of replication in cells from a variety of species ranging from insects to humans. Pol δ, PCNA, and RFC are highly conserved enzymes, and interaction of Rep with these cellular replicative enzymes would explain AAV's species promiscuity both during productive infections and during transduction with AAV vectors when a limited second strand synthesis is believed to occur.

No role for ssDNA binding protein.

As we purified the components of the AAV DNA replication reaction, it became clear that ssDNA binding protein (neither RPA nor Ad DBP) was not required for replication. In contrast, experiments with crude extracts showed some role for ssDNA binding proteins (28, 46). We can imagine several possible explanations. First, ssDNA binding proteins may be required in vivo and in crude extracts to protect single-stranded template DNA or the single-stranded progeny DNA from random nicking during strand displacement synthesis or from binding by other proteins that might interfere with DNA replication. This is believed to be the primary role for ssDNA binding proteins in vivo (for a review see reference 10). Once the replication enzymes are purified, RPA, Ad DBP, and the herpesvirus UL29 protein may no longer be essential. We note, for example, that, in spite of the apparent tight requirement for UL29 in herpesvirus-infected cells and in crude extract experiments in vitro (37, 48), only a twofold reduction in DNA synthesis was seen when UL29 was omitted from in vitro reaction mixtures that contained only purified herpesvirus replication proteins (47). It is also worth noting that RPA interacts with a wide variety of cellular DNA replication and repair enzymes in addition to the Rep protein (10). It is believed that these interactions are used in part to displace RPA from DNA so that the polymerase complex can use exposed ssDNA as a template (10). Since AAV clearly uses cellular enzymes for DNA replication, this may explain why Ad DBP and not RPA inhibits AAV DNA replication at high concentrations. Ad DBP may not have the correct set of interactions with the replicative complex during strand displacement that are required to remove Ad DBP ahead of the polymerase complex.

Second, it is possible that an excess of Rep78 or Rep68, which are both ssDNA binding proteins and DNA helicases, may be sufficient to both protect and unwind the template, thereby substituting for ssDNA binding protein. This might also explain the puzzling fact that herpesvirus helicase primase (UL5, -8, and -52) did not affect in vitro replication when it was omitted from a reaction mixture that contained Rep, the herpesvirus polymerase complex, and the herpesvirus ssDNA binding protein (47).

Third, it may be that the intracellular role of ssDNA binding proteins is to help colocalize Rep and cellular replication enzymes onto ssDNA in replication foci within the nucleus. This has been suggested by several groups and would account for the ssDNA-stimulated interaction of Rep and RPA, Rep and DBP (37), and Rep and UL29 (15). However, it leaves the puzzling question of why there is clear evidence of a genetic requirement for AAV DNA replication only in the case of UL29 (15, 37, 48).

The missing factor(s) in P-Cell IA.

We could not identify the factor(s) in fraction IA that was responsible for stimulation of AAV DNA replication to the levels seen in Ad-infected-cell crude extracts. Attempts to purify the factor failed, suggesting either that the factor was unstable or consisted of a complex that fractionated into multiple parts upon further fractionation. This factor is of considerable interest since it remains a puzzle as to what factor is induced by the Ad helper that stimulates AAV DNA synthesis in vitro by more than 10-fold and in vivo by over a 100-fold.

One potential activity contained within fraction IA could be a DNA helicase. The fact that we were able to replicate AAV DNA in vitro with just three purified enzymes (T4 DNA polymerase, gp32, and Rep) suggests that Rep protein alone could supply the necessary DNA helicase activity for strand displacement synthesis. Ward et al. (47) showed a similar phenomenon when they were able to replicate AAV DNA in vitro with only herpesvirus DNA polymerase, UL29, and Rep protein. However, Rep helicase activity was not sufficient to produce detectable replication in the context of a reaction mixture containing Pol δ, PCNA, and RFC (and RPA or Ad DBP). Thus, one or more cellular helicase activities may be the missing factor in fraction IA.