Identification of Cellular Proteins That Interact with the Adeno-Associated Virus Rep Protein

Abstract

Adeno-associated virus (AAV) codes for four related nonstructural Rep proteins. AAV both replicates and assembles in the nucleus and requires coinfection with a helper virus, either adenovirus (Ad) or herpesvirus, for a productive infection. Like other more complex DNA viruses, it is believed that AAV interacts or modifies host cell proteins to carry out its infection cycle. To date, relatively little is known about the host proteins that interact with the viral Rep proteins, which are known to be directly involved in DNA replication, control of viral and cellular transcription, splicing, and protein translation. In this study, we used affinity-tagged Rep protein to purify cellular protein complexes that were associated with Rep in cells that had been infected with Ad and AAV. In all, we identified 188 cellular proteins from 16 functional categories, including 14 transcription factors, 6 translation factors, 15 potential splicing proteins, 5 proteins involved in protein degradation, and 13 proteins involved in DNA replication or repair. This dramatically increases the number of potential interactions over the current number of approximately 26. Twelve of the novel proteins found were further tested by coimmunoprecipitation or colocalization using confocal immunomicroscopy. Of these, 10 were confirmed as proteins that formed complexes with Rep, including proteins of the MCM complex (DNA replication), RCN1 (membrane transport), SMC2 (chromatin dynamics), EDD1 (ubiquitin ligase), IRS4 (signal transduction), and FUS (splicing). Computer analysis suggested that 45 and 28 of the 188 proteins could be placed in a pathway of interacting proteins involved in DNA replication and protein synthesis, respectively. Of the proteins involved in DNA replication, all of the previously identified proteins involved in AAV DNA replication were found, except Ad DBP. The only Ad protein found to interact with Rep was the E1b55K protein. In addition, we confirmed that Rep interacts with Ku70/80 helicase. In vitro DNA synthesis assays demonstrated that although Ku helicase activity could substitute for MCM to promote strand displacement synthesis, its presence was not essential. Our study suggests that the interaction of AAV with cellular proteins is much more complex than previously suspected and provides a resource for further studies of the AAV life cycle.

Adeno-associated virus (AAV) is a single-stranded DNA virus with a 4.7-kb genome consisting of two open reading frames, rep and cap, flanked by two inverted terminal repeat (ITR) sequences. Coinfection with either adenovirus (Ad) or herpesvirus is required for efficient AAV replication and virus propagation (for a review, see reference 11). The rep gene codes for a family of four proteins with overlapping coding regions. In the presence of helper virus, the two larger Rep proteins, Rep78 and Rep68, are required for AAV DNA replication (40, 48, 101), control of AAV transcription (9, 58, 60, 61, 66, 77, 111, 129), alternative splicing of viral RNA (86), viral DNA packaging (24, 31, 54, 83, 119), and site-specific integration of viral DNA into human chromosome 19 (3, 56). In addition, the expression of Rep proteins has been shown to inhibit Ad (21, 22, 51), simian virus 40 (SV40) (4, 127), bovine papillomavirus (38), human immunodeficiency virus (1), and herpesvirus propagation (51, 55); inhibit transcription from a variety of cellular and viral promoters (1, 37, 39, 44, 53, 59, 121, 122); and efficiently arrest cells in the S phase (12, 36, 92, 128). To accomplish these tasks, Rep is believed to interact with a variety of cellular and helper virus proteins, which have thus far been poorly defined.

AAV replicates in the nucleus by a strand displacement method using a hairpin ITR as a primer (11). The hairpin primer is used to synthesize a duplex DNA molecule that is covalently closed at one or both ends. The ITR is cleaved at a unique site on one strand, the terminal resolution site (trs), in a process called terminal resolution. This nicked hairpin ITR is repaired to make an open-ended duplex intermediate which can be denatured and reannealed to form a double hairpin intermediate that initiates a further round of DNA replication and generates single strands for packaging. In vitro reconstitution experiments with purified enzymes and in vivo genetic experiments suggest that in the presence of Ad helper virus, this process requires a minimum set of four cellular DNA replication complexes. These are polymerase δ (pol δ), proliferating cell nuclear antigen (PCNA), replication factor C (RFC), and minichromosome maintenance complex (MCM) (69, 70). In addition, the cellular single-stranded DNA binding protein RPA and the Ad DNA binding protein DBP may have a role in AAV DNA replication (72, 106, 114).

The viral-encoded Rep78 and Rep68 proteins have been shown to have site-specific DNA helicase and endonuclease activities required to carry out both terminal resolution and reinitiation in vitro (48, 99, 101, 130), and mutations in the rep gene result in defects in AAV replication in vivo (40, 110). The N-terminal region that is unique to the larger Rep protein has a site-specific DNA binding domain for a core 22-bp sequence within the AAV ITR, termed the Rep binding element (RBE) (25, 67, 68, 75, 89, 100, 117). Another ITR recognition sequence, RBE′, increases the binding affinity (68, 89) and stimulates the Rep helicase activity that is necessary for initiating site-specific nicking (16, 48). Helicase activity is believed to be necessary to unwind the duplex trs and extrude a stem-loop structure, which is the subsequent substrate for the nicking reaction (17, 50). Strand-specific nicking of the ITR occurs at the trs, 3′-CCGGT/TG-5′, which results in a phosphotyrosyl linkage between Rep and the nicking site (17, 48, 99). Although Rep78 and Rep68 both have trs endonuclease activity, in vitro assays have suggested that several cellular proteins, including RPA (106), nucleophosmin (13), and high mobility group 1 (HMG1) (26), can stimulate the trs nicking reaction and Rep binding to the ITR. The interaction with 14-3-3 γ and ɛ proteins, on the other hand, seems to reduce the binding of Rep68 to the ITR and reduces DNA replication (35). Rep also has been shown to initiate DNA replication (albeit less efficiently) from an alternative RBE in the p5 promoter (73, 74, 108, 112), which initiates mRNA transcripts that code for the large Rep proteins. The initiation of DNA replication from this site requires interaction with the TATA binding protein (33) and may be involved in AAV integration at chromosome 19 (126). Rep is also involved in site-specific recombination. In this respect, Rep has been shown to interact with TRP-185, which binds to the chromosome 19 integration site and modifies the DNA binding and helicase activities of Rep (125), and it has been shown to copurifiy with Ku70/80 in a DNA-dependent manner.

In the absence of helper virus, the expression of Rep78/68 has been shown to arrest cells in the S phase (12, 36, 92, 128). This is believed to be partly responsible for the antioncogenic property of AAV infection. The effect may be due to an increase in pRb protein in cells that constitutively express Rep78 and requires a Zn binding motif in the C terminus of Rep78 (44, 92). In the presence of Ad helper virus infection, pRB induction is largely overcome by the expression of the Ad E1a protein (92). The binding of E1a to the pRB complex releases E2F, a transcription complex that initiates S phase by inducing the synthesis of proteins required for cellular as well as Ad and AAV DNA replication (reviewed in reference 10). Rep protein has been shown to interact with pRB, pRB-E2F1 complexes, and the Ad E1A protein and to inhibit the E1a-induced release of E2F from pRB complexes (5, 6, 92). Subsequent to S phase induction by E1a, the Ad E1b55K/E4orf6 complex inhibits apoptosis by promoting the degradation of p53 (10), and E4orf6 protein freezes cells in S phase by promoting the degradation of cyclin A (34), thereby providing a window for AAV DNA replication. Rep independently also achieves a block in S phase by binding to Cdc25A (12). Curiously, the E1b/E4-induced degradation of p53 is apparently inhibited by p53 binding to Rep (7). Additionally, Rep78/52 proteins contain an inhibitory domain similar to that of the protein kinase A (PKA) inhibitor PKI, which is able to inhibit PKA and PrKX. Mutated AAV2 genomes lacking this domain fail to inhibit Ad replication, suggesting that the modulation of PKA activity is a mechanism by which AAV2 perturbs normal cyclic AMP signal transduction and interferes with helper virus replication (28, 29, 94).

In addition to its role in DNA replication and the inhibition of cellular functions, Rep is both a repressor and activator of AAV transcription (9, 44, 58, 60, 67, 77-79). In the absence of helper virus, Rep78 and -68 bind to their own promoter and suppress transcription (44, 58, 77). In the presence of Ad E1a, Rep activates all three AAV promoters as much as 200-fold (9, 60, 61, 66, 77-79). This is presumably accomplished by interaction with cellular transcription factors and chromosome remodeling factors as well as the p5 and ITR RBEs, but to date, the mechanism is not clear. Rep has been shown to interact with transcription factors Sp1 (41, 79), E2F1, TATA binding protein (6, 42, 71), c-Jun (82), and p53 (7, 115). Rep also interacts with the RNA polymerase II coactivator PC4 and the chromatin remodeling complex TAF1/Set/ANP32A/B (76, 116). These interactions have typically shown a downregulation of the promoter activity when reporter assays are used (for example, c-Jun [82]). The interaction with Sp1, on the other hand, has been proposed to be essential for the Rep-mediated activation of transcription from the AAV p19 and p40 promoters. This is believed to occur via a DNA loop between Rep bound at the p5 promoter and Sp1 bound at the p19 and p40 promoters (61). Similarly, the interaction of Rep with Topors, a p53 and topoisomerase 1 binding protein, also appears to stimulate AAV transcription (115).

In all, some 26 proteins, some existing as multiprotein complexes, that interact with one or more Rep proteins, have been identified (6, 7, 12, 13, 20, 26, 28, 29, 35, 41, 42, 71, 76, 78, 82, 85, 94, 106, 115, 116, 125). However, for the most part, previous searches for Rep-interacting proteins have relied on methods that would require a direct interaction, and these were not likely to capture all of the interactions that involved Rep during productive infections. In an attempt to obtain more information about multiprotein complexes that interacted with Rep and might be involved in AAV DNA replication, we used the tandem affinity purification (TAP) approach (84, 88). In this paper, we characterize the proteins that Rep interacts with during a productive AAV infection, specifically in the presence of Ad helper virus. Using the TAP approach, we generated a Rep-TAP fusion protein which enabled us to isolate cellular complexes containing Rep78. Analysis of these complexes by mass spectroscopy identified 188 proteins that were unique to the Rep-TAP extracts. These were involved in a number of different biological processes, including DNA replication, transcription, translation, protein degradation, and RNA splicing. We confirmed the interaction of a selection of these proteins using both coimmunoprecipitation and immunocytochemistry. Our results suggest that Rep modifies multiple pathways within the cell during a productive infection.

MATERIALS AND METHODS

Antibodies and chromatography reagents.

Antibodies were purchased from Abcam (NTF97/importin), Bethyl Laboratories (FUS, RCN, SMC2, and MCM2 to -6), Upstate (IRS4 and DNA-dependent protein kinase [DNA-PK]), Sigma (PCNA), Novus Biologicals (EDD-1, Ku70, and Ku80), BD Pharmingen (MCM7), and Invitrogen (Alexa Fluor 488 goat anti-rabbit and Alexa Fluor 633 goat anti-mouse antibodies). Mouse monoclonal anti-Rep 78/68 (7B73.2) antibody, which detects only the two large Rep proteins, and anti-Rep 52/40 (1F11) antibody, which detects all four Rep proteins, were a kind gift from R. J. Samulski (46). Mouse monoclonal anti-AAV2 capsid C37 and D3 antibodies, which recognize both intact capsids and capsid intermediates, were a kind gift from J. Kleinschmidt (120). Immunoblotting detection reagents were purchased from Millipore. [α-³²P]dATP (3,000 Ci/mmol) was purchased from Perkin Elmer. DE52 cellulose was acquired from Whatman. Phenyl-Sepharose and protein A and G Sepharose were purchased from Amersham. Ni-nitrilotriacetic acid agarose was purchased from Qiagen. HindIII, DpnI, and λ DNA were purchased from New England Biolabs. DNA-PK was purchased from Trevigen.

Replication proteins.

Replication proteins PCNA, pol δ, RPA, Rep78, and RFC were purified as described before (70). Ad-infected crude cell extracts from the Chinese hamster ovary (CHO) cell line xrs-5 and the parental CHO-K cell line were fractionated on phosphocellulose as described previously for crude cell extracts (69). Baculovirus expression clones of Ku70 and Ku80 were kindly provided by L. Comai. For the purification of Ku, typically 1 liter of Sf9 cells at 1 × 10⁶ cells/ml was infected at a multiplicity of infection (MOI) of 2 to 5. After 48 h, the cells were harvested by centrifugation and lysed by Dounce homogenization with buffer C (20 mM HEPES [pH 7.4], 10% glycerol, 25 mM NaCl, 1 mM dithiothreitol (DTT), Complete protease inhibitors [Roche Diagnostics, GmbH]). The lysate was clarified by centrifugation at 10,000 × g and then loaded onto a DE52 column equilibrated with buffer C and eluted with a step gradient of 0.15 M NaCl, 0.3 M NaCl, and 1.0 M NaCl. Ku was eluted in the 0.15 M NaCl fraction and was dialyzed in buffer C. The sample was then loaded onto a single-stranded DNA column (buffer C) and eluted with a gradient of 0.025 to 1.0 M NaCl. To the pooled fractions containing Ku was added (NH₄)₂SO₄ to a 0.7 M final concentration. This was loaded onto a phenyl Sepharose column equilibrated with buffer [50 mM Tris (pH 7.9), 1 mM EDTA, 5% glycerol, 0.02% Tween, protease inhibitors, 0.7 M (NH₄)₂SO₄]. Ku was eluted with a gradient of 0.7 M to 0 M (NH₄)₂SO₄.

Generating the Rep-TAP plasmid.

An in-frame C-terminal TAP tag (84, 88) was cloned into the 3′ end of the AAV Rep78/52 gene in the infectious wild-type (wt) AAV2 clone pSM620 (90). The C terminus of the Rep gene was amplified with the oligonucleotide primers SM620.1522-41 (GACCAGAAATGCAGTCCTC) and RepTap.rev (CTCCTCCCATGGCTTGTTCAAAGATGCAGTCATCC) to allow the ligation of the TAP tag in frame with the Rep gene. The TAP tag was removed from plasmid pBS1479 (88) with the NcoI and SmaI restriction enzymes. To make the vector capable of packaging the Rep-TAP genome, a 1,071-bp deletion was made which extended from the AAV intron (nucleotide 2183) to a position 3′ of the start codons of the three AAV capsid genes (nucleotide 3253). This essentially resulted in an AAV2 genome that was replication competent (ITR positive and Rep-TAP positive) but did not express any of the AAV capsid genes. The clones were sequenced to confirm identity.

Virus production.

AAV2 Rep-TAP virus was generated by the cotransfection of the Rep-TAP and pDG plasmids into 293 cells as previously described (129). The cells were harvested at 60 h posttransfection. The cell pellets were freeze-thawed three times in lysis buffer (0.15 M NaCl, 50 mM Tris-HCl [pH 8.0]). After benzonase digestion, the virus was purified by iodixanol gradient and heparin column chromatography (131). The virus titer was measured by both A20 enzyme-linked immunosorbent assay and dot blotting. Ad5 virus was made by the infection of 293 cells at an MOI of 10 and purified by CsCl gradient as described previously (49). Ad5 titer was measured by plaque assay.

Expression of Rep-TAP.

Rep-TAP cell extracts were generated from 293 cells infected with AAV2-Rep-TAP virus at an MOI of 100 vector genomes/cell and Ad5 at an MOI of 5 PFU/cell. Parallel infections with wt AAV2 (MOI = 20) and Ad (MOI = 5), which are expected to make the normal complement of Ad and AAV proteins but no TAP tag, were used to make the control extracts. Three other types of control lysates were isolated and analyzed: HeLa C12 cells infected with AAV2-Rep-TAP virus (MOI = 100), which are expected to make small amounts of Rep-TAP; 293 cells infected with Ad (MOI = 5); and 293 cells constitutively expressing an unrelated TAP fusion protein, α synuclein-TAP. In addition, Rep-TAP complexes were isolated from 293 cells transfected with the Rep-TAP plasmid using calcium phosphate. In this case, parallel transfections with the ITR-negative AAV2 plasmid pIM45 were used to make the control extracts that contained wt Rep protein with no TAP tag. In all cases, the cells were harvested at 48 h posttransfection or postinfection. The cell pellets were lysed in TAP buffer (10 mM Tris-HCl [pH 8.0], 150 mM NaCl, 0.1% NP-40, 1 mM DTT containing protease inhibitors [0.1 mM phenylmethylsulfonyl fluoride, 10 μg/ml leupeptin, 7 μg/ml pepstatin]) and incubated on ice for 30 min. The lysates were centrifuged at 8,000 × g for 1 h at 4°C, and the protein concentrations were measured by a Bradford assay using bovine serum albumin as the standard.

Purification of Rep-TAP.

The purification of TAP complexes was done essentially as described previously (84, 88). However, it was discovered early in the study that Rep protein binds immunoglobulin G (IgG)-Sepharose independently of the protein A sequence in the TAP tag, and this often resulted in a low yield of TAP-containing complexes for chromatography on the calmodulin column. A total of 200 mg of protein extract was incubated with 1 ml of IgG-Sepharose beads (Amersham) overnight at 4°C. The IgG-Sepharose was washed five times with TAP buffer and once with TEV buffer (10 mM Tris-HCl [pH 8.0], 150 mM NaCl, 0.1% NP-40, 1 mM DTT, and 0.5 mM EDTA). TEV cleavage was performed by addition of 1.2 ml of TEV buffer and 150 U of TEV protease (Invitrogen) and incubation at 4°C for 6 h. The eluate was recovered by gravity flow and mixed with 3 volumes of calmodulin binding (CB) buffer (10 mM Tris-HCl [pH 8.0], 15 mM NaCl, 10 mM 2-mercaptoethanol, 1 mM magnesium acetate, 1 mM imidazole, 0.1% NP-40, and 2 mM calcium chloride). The mixture was adjusted to a 2 mM CaCl₂ final concentration by adding 1 M CaCl₂ and then transferred to the column containing equilibrated calmodulin beads (Stratagene). After incubation for 2 h at 4°C, the beads were washed three times with CB buffer and eluted with CB elution buffer (10 mM Tris-HCl [pH 8.0], 15 mM NaCl, 10 mM 2-mercaptoethanol, 1 mM magnesium acetate, 1 mM imidazole, 0.1% NP-40, and 2 mM EGTA). Immunoblotting was used to detect the efficiency of each step by Rep antibody (1F; 1:3,000).

Protein complex identification.

TAP-purified proteins (CB eluates) were loaded on sodium dodecyl sulfate (SDS)-polyacrylamide gel and visualized by silver stain. Bands were cut for liquid chromatography-microspray mass spectrometry (LC-MS/MS) protein identification (performed by the University of Florida Institute for Biotechnology Research protein core). The Scaffold program (www.proteomesoftware.com) was used for protein identification and data analysis. Alternatively, TAP-purified proteins (CB eluates) were also subjected to buffer exchange into 10 mM NH₄HCO₃ solution, then trypsinized, and subsequently analyzed with LC-MS/MS.

Coimmunoprecipitation.

The 293 cells were infected with wt AAV2 at an MOI of 20 and with Ad5 at an MOI of 5. The cell lysates for immunoprecipitation were made as described above. A total of 500 μg of proteins was added to a total of 500 μl TAP buffer containing 50 μl of protein A or G Sepharose slurry and 5 μg of primary antibody. The immune complexes were rotated overnight at 4°C and then washed five times with 1 ml of TAP buffer. The immunoprecipitates were resuspended with 30 μl of 2× SDS loading buffer and immune complexes were boiled at 95°C for immunoblot analysis.

Immunoblot analysis.

Samples were prepared for SDS-polyacrylamide gel electrophoresis (PAGE) by boiling in Laemmli sample buffer. After SDS-PAGE, the proteins were transferred onto a polyvinylidene difluoride membrane (Millipore), and the blots were blocked in 5% nonfat milk in phosphate-buffered saline (PBS) with 0.05% Tween-20 for 1 h at room temperature (RT). The primary antibody was incubated with blots for 1 h at RT or overnight at 4°C. Following three washes, the secondary antibody was incubated with blots for an hour. The protein bands were visualized by enhanced chemiluminescence (Millipore). In some instances, the membranes were stripped with stripping buffer (Thermo Scientific) and were reblotted with other antibodies as necessary.

Confocal microscopy.

Tissue culture glass slides were seeded with HeLa cells. Once the cells were ∼50% confluent, they were infected with wt AAV2 (MOI = 100) and Ad5 (MOI = 5) in serum-free medium for 10 min prior to the addition of serum-containing medium. At 16 h postinfection, the cells were washed with PBS and fixed in 4% paraformaldehyde for 10 min at RT. The cells were washed with PBS and PBS-Triton X-100 (0.1%) for 5 min and then blocked with PBS-Tween 20 plus low IgG bovine serum albumin (5 mg/ml). The primary antibodies were incubated for 2 h at RT (1/300 dilution), followed by the secondary Alexa-Fluor antibodies for 1.5 h at RT (1/200 dilution). Anti-mouse Alexa-Fluor 633 antibody was used to detect the anti-Rep78/68 (7B73.2) antibody (46), which detects only the large Rep proteins, or anti-capsid C37 or D3 antibody (120), which detects both intact capsids and capsid intermediates. Anti-rabbit Alexa-Fluor 488 was used to detect the rabbit polyclonal antibodies of the other proteins examined. Confocal microscopy images were acquired using a Leica TCS SP5 laser-scanning microscope with a 63× objective and 0.2-μm sections.

AAV DNA replication assay.

No-end (NE) substrate was prepared as previously described by Snyder et al. (101). The standard replication reaction was done as previously described by Nash et al. (70) and contained the following 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, 0.1 to 1.0 μg Rep78, and crude extract or purified replication proteins. The reaction mixtures were incubated for 3 h at 37°C, and then the reactions were stopped with 35 μl of stop solution (0.3% 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.

Protein network analysis.

For the 210 Rep-TAP-associated proteins identified by LC-MS/MS, we used a combination of bioinformatics databases and software that included E! Ensemble (www.ensembl.org/Homo_sapiens), SwissProt (http://ca.expasy.org/sprot/sprot-top.html), Protonet (www.protonet.cs.huji.ac.il), Pandora (www.pandora.cs.huji.ac.il), and Pubmed (www.ncbi.nlm.nih.gov/sites/entrez). Using this approach, we were able to identify probable functions in some entries that were initially unknown. In addition, the functional analysis algorithm from Ingenuity Pathway Analysis (Ingenuity_Systems, www.ingenuity.com) was used to identify the biological functions that were most significant to the data set. The 210 genes from the data set associated with biological functions in the Ingenuity Pathway knowledge base were considered for the analysis. Fisher's exact test was used to calculate a P value determining the probability that each biological function and/or disease assigned to that data set is due to chance alone. The two functional pathways identified, the replication pathway and the protein synthesis pathway, both had P values of <0.001.

RESULTS AND DISCUSSION

To examine the proteins that are associated with the AAV Rep protein in vivo, we used the TAP tag method for pulling down protein complexes. The 579-bp TAP tag was placed at the C terminus of the rep open reading frame using the infectious AAV2 plasmid construct pSM620 (90) (Fig. 1). The insertion of the TAP tag increased the size of the AAV genome to approximately 112% of the wt size. As this was not likely to be packaged efficiently, a deletion was constructed in the clone; it extended from AAV2 nucleotide 2183, within the intron, to nucleotide 3253 it within the cap open reading frame. This deletion removed the 3′ splice site of the Rep68/40 intron; the 3′ splice acceptor sites of VP1, VP2, and VP3; the first 351 amino acids of VP1; and the 213 and 148 N-terminal amino acids of VP2 and VP3, respectively. The Rep-TAP fusion construct, therefore, only expressed Rep78 and Rep52 (Fig. 1), and only cellular proteins that interact with these Rep species could be identified. Rep68, Rep40, and all three of the capsid proteins were not expressed (Fig. 1 and data not shown).

FIG. 1.

Rep-TAP construct. (A) Diagrammatic representation of AAV2 and the Rep-TAP viral construct used to infect HEK293 cells for the expression of Rep78/52-TAP protein. (B) Western blot using a Rep antibody against crude extract from wt AAV2/Ad infection (lane 1) and crude extract from Rep-TAP/Ad infection (lane 2).

Isolation of Rep-TAP interacting complexes.

To examine the proteins that Rep interacts with during a typical Ad-AAV viral coinfection, we first packaged the Rep-TAP fusion recombinant AAV (rAAV) into AAV2 capsids by cotransfecting the Rep-TAP rAAV plasmid into 293 cells with pDG, a plasmid that supplied the missing Ad and AAV gene products, as described in Materials and Methods. The resulting Rep-TAP fusion virus was then used to infect HEK293 cells with Ad5 helper virus, and the cell lysate was isolated at 48 h postinfection. As expected from previous work, the Rep-TAP protein was functional for DNA replication, and the rAAV genome underwent DNA replication in the presence of the Ad helper functions and in the absence of capsid proteins (40, 110; data not shown). The 48-h cell lysate was isolated as described in Materials and Methods and subjected to successive column purification using the protein A and calmodulin columns.

Four types of negative-control extracts were used: (i) cells that had been infected with wt AAV2, which contains no TAP motif; (ii) cells that had been infected with Ad5; (iii) cells that had been infected with wt AAV2 and Ad5; and (iv) cells that had been infected with Ad5 and an rAAV virus that expressed an α synuclein-TAP fusion protein. The α synuclein-expressing virus provided an Ad-positive extract that expressed an irrelevant protein tagged with TAP as a negative control. Cell extracts were also prepared and purified from 293 cells that had been transfected with Rep-TAP plasmid DNA in the absence of Ad virus coinfection. The transfection of the double-stranded Rep-TAP plasmid resulted in the limited expression of AAV Rep-TAP in the absence of the complete set of Ad helper functions in 293 cells. In this case, the pIM45 plasmid, an AAV2 plasmid that expresses wt Rep without a TAP tag, was used as a negative control.

Initial attempts to isolate Rep-containing complexes from the protein A column resulted in poor yields of protein due to the fact that Rep apparently could bind protein A irrespective of the IgG motif in the TAP tag and that the bound complex was not eluted following cleavage with TEV. Subsequent attempts, therefore, only used the calmodulin column affinity purification step. The calmodulin column eluate was then subjected to LC-MS/MS to identify the proteins that coelute with Rep. All proteins that were common to both the Rep-TAP extract and the negative controls were eliminated from further consideration (see below). We note also that no attempt was made to remove nucleic acid from the extracts prior to chromatography. Thus, it is possible that some protein associations occurred through a DNA- or RNA-containing intermediate.

We identified 210 proteins by LC-MS/MS in extracts from cells infected with Ad and the AAV Rep-TAP virus that appeared to interact with Rep. Six of these were common with α synuclein-TAP extracts, primarily proteins from the keratin family or CB proteins. When the remaining 204 proteins were compared with extracts from cells infected with wt AAV and Ad, an additional 16 proteins were common. This left a total of 188 proteins that appeared to be unique to the extract expressing Rep-TAP. The complete list of proteins identified is presented in VirHostNet (http://pbildb1.univ-lyon1.fr/virhostnet) a public database for virus-host protein interaction data management and analysis. A selection of these proteins is shown in Table 1. Of the 188 proteins, 70 fulfilled the standard criteria of being identified by at least two unique peptides at a confidence level better than 95% (see http://pbildb1.univ-lyon1.fr/virhostnet, # peptides). The remaining proteins, which were identified by a single peptide, are included because most of these have been shown to participate in complexes with the 70 proteins mentioned above and, thus, are likely to be authentic interacting partners. For example, MCM7 and MCM5 were unequivocally identified through five and two unique peptides, respectively, while MCM6 was identified with only one peptide and the remaining proteins in the MCM complex, MCM2, -3, -4, and -7, were not detected. Nevertheless, MCM2 to -7 can be purified as a complex (65, 70) and are likely to participate in AAV DNA replication. The lack of the remaining MCM proteins might be a function of their molecular masses. For example, in the case of RPA, the largest subunit (70 kDa) was identified with four unique peptides, while the 34-kDa subunit yielded only one peptide and the smallest of the three RPA subunits (11 kDa) was not detected. Alternatively, our ability to detect known subunits of a complex may reflect the relative abundance of subcomplexes. This will have to be evaluated with future experiments.

TABLE 1.

Proteins interacting with Rep78^a

Group and protein (n)	ID	RP
DNA replication/repair (13)
DNA replication licensing factor MCM5	MCM5	X
DNA replication licensing factor MCM6	MCM6	X
DNA replication licensing factor MCM7	MCM7	X
70-kDa replication protein A	RPA1
32-kDa replication protein A	RPA2	X
Proliferating cell nuclear antigen	PCNA	X
DNA-dependent protein kinase	DNA-PK	X
Ku70	KU70	X
Ku80	KU80	X
RAD50	RAD50
Structural maintenance of chromosome 2	SMC2
Telomere-associated protein RIF1	RIF1
Poly-[ADP-ribose] polymerase 1	PARP1	X
Membrane transport/vesicle (15)
Sodium/potassium-transporting ATPase α	ATP1A1
Reticulocalbin-1 precursor	RCN1
ADP-ribosylation factor 1	ARF1
ATP-binding cassette subfamily D member 3	ABCD3
ATP-binding cassette subfamily E member 1	ABCE1
Coatomer subunit α	COPA
Coatomer ADP-ribosylation factor 4	ARF4
Spectrin β chain, brain 2	SPTN2
SERCA2A	AT2A2
Voltage-dependent anion channel protein 2	VDAC2
Kinesin heavy chain	KINH
Kinesin light chain 2	KLC2
Dynein heavy chain, cytosolic	DYHC1
BC002942 protein	LMF2
Dolichyl-diphosphooligosaccharide transferase	OST48
Nuclear pore (4)
Ran-binding protein 2	RBP2
Importin-7	IPO7	X
GTP-binding nuclear protein RAN	RAN
Nucleoporin 85	NUP85
Nucleolus (3)
Nucleolin	NUCL	X
Nucleophosmin	NPM	X
KIAA0690	KIAA069
Protein degradation (5)
Ubiquitin-protein ligase EDD1	EDD1	X
Proteasome non-ATPase regulatory subunit 2	PSMD2
26S protease regulatory subunit 7	PSMC2
Ubiquitin-activating enzyme E1	UBA1
Lysozyme C precursor	LYSC
Signal transduction (10)
A-kinase anchor protein 12	AKAP12
Insulin receptor substrate 4	IRS4	X
ELG protein variant (fragment)	ELG
Calcium-dependent protein kinase type II γ	CAMK2G	X
Calcium-dependent protein kinase type II δ	CAMK2D
Wolframin	WFS1
14-3-3 Dedicator of cytokinesis protein 7	DOCK7
14-3-3 Protein ζ/δ	1433Z	X
14-3-3 Protein ε	1433E	X
14-3-3 Protein θ	1433T	X
Splicing or RNA/single-stranded DNA binding
proteins (13)
RNA-binding protein FUS	FUS	X
KH-type splicing regulatory protein	KHSRP
U3 snRNA-associated protein 14 A	UT14A
U4/U6 small nuclear ribonucleoprotein Prp4	PRP4
Splicing factor, arginine/serine-rich 4	SFRS4
Splicing factor, arginine/serine-rich 1	SFRS1
Splicing factor, arginine/serine-rich 2	SFRS2
Putative splicing factor, arginine/serine-rich 14	SFR14
Splicing factor 3B subunit 4	SF3B4
Splicing factor U2AF 35-kDa subunit	U2AF1
KH-type splicing regulatory protein	FUBP2
Splicing coactivator subunit SRm300	SRRM2
Heterogenous nuclear ribonucleoprotein U	HNRPU	X
Transcription (14)
Transcription intermediary factor 1-α	TIF1A
SNW domain-containing protein 1	SNW1	X
HINT4 (histidine triad protein 3)	HINT3
Zinc finger CCHC domain-containing protein 3	ZCCHC3
RuvB-like 1	RUVBL1	X
DNA methyltransferase 1-associated protein 1	DMAP1	X
Prohibitin-2	PHB2
Transcription elongation factor SPT6	SUPT6H
DEAH (Asp-Glu-Ala-His) box polypeptide 9	DHX9	X
RNA polymerase II coactivator p15 (PC4)	PC4
Histone deacetylase 2 variant (fragment)	HDAC2
Nuclease sensitive element-binding protein 1	YBOX1
Lupus La protein	LA	X
Histone H1.2	H12
Translation (6)
Elongation factor 2	EF2	X
Elongation factor 1-γ	EF1G	X
Nuclease sensitive element-binding protein 1	YBOX1
Aspartyl-tRNA synthetase	SYDC
Interleukin enhancer-binding factor 3	ILF3
Eukaryotic initiation factor 4A-I	IF4A1

^aThe table shows only the 84 proteins identified in the categories listed. For a complete list of all categories and all 188 proteins, see Table S1 in the supplemental material. Protein identification names (ID) that are shown in bold were confirmed to interact with Rep78 by coimmunoprecipitation experiments (see the text and figures). An X in the replication pathway (RP) column indicates proteins that were assigned to an interacting group of proteins involved in DNA replication, as shown in Fig. 8, using Ingenuity software. The complete set of RP proteins is shown in Table S1 in the supplemental material and in Fig. 8.

As mentioned earlier, previous work had identified 26 proteins that interacted with the Rep protein. Of these, only 9 were present among the 188 proteins found in this study (see http://pbildb1.univ-lyon1.fr/virhostnet). They were RPA70, RPA32, PC4, 14-3-3, NPM1, nucleolin, importin, Ku70, and Ku80. This suggests that the efficiency of this approach for identifying proteins that interact with Rep was approximately 40% (9/26). Alternatively, it is possible that some of the proteins previously identified by in vitro methods do not interact in vivo. In addition, some of the proteins previously identified as interacting with Rep were identified under conditions that did not involve a productive AAV infection cycle in the presence of Ad helper virus and, thus, might not be predictive for the conditions used here (104).

Proteins involved in DNA replication.

The proteins could be classified into 16 categories, which are illustrated in Fig. 2 and at http://pbildb1.univ-lyon1.fr/virhostnet. Our primary interest was in protein complexes that might be involved in DNA replication or in control of the cell cycle. As mentioned earlier, we have identified a minimum set of cellular protein complexes required for AAV DNA replication (pol δ, RFC, PCNA, RPA, and MCM) (69, 70); however, we were interested in whether additional complexes might be involved and also whether Rep interacts with complexes that might be involved in AAV integration, rAAV episome establishment, or the antioncogenic effect of Rep. A group of 13 proteins were identified that could be classified as DNA replication, repair, or recombination proteins (Table 1; see also http://pbildb1.univ-lyon1.fr/virhostnet). As expected, proteins of the MCM complex, the RPA complex, and PCNA copurified with Rep-TAP; these proteins had all been shown to have a role in AAV DNA replication (69, 70, 72, 106). In contrast, none of the RFC or pol δ subunits were identified. Given that the PCNA homotrimer complex is an essential accessory factor for pol δ, it is unlikely that PCNA (and MCM and RPA) but not pol δ copurified with Rep via a DNA-containing intermediate. To confirm this, we used antibodies to RFC, RPA, PCNA, pol δ, Ku, and DNA-PK to see if we could coimmunoprecipitate Rep protein (Fig. 3) from a cell extract that had been infected with wt AAV2 and Ad5, either before or after the extract was treated with DNase. As expected from the work of Stracker et al. (105), a mixture of antibodies to RPA70 and RPA34 precipitated Rep78 and -68. Antibody to PCNA also precipitated a small amount of Rep protein, but the bulk of this Rep was in a high-molecular-weight form of the Rep complex that was insensitive to SDS, similar to that seen with Rep antibody precipitation (Fig. 3A, compare the PCNA lanes with the Rep lanes). This was consistent with previous immunofluorescence experiments, which had suggested that PCNA and Rep colocalized (45). Surprisingly, similar high-molecular-weight complexes of Rep were seen when pol δ antibody was used (Fig. 3A) and when DNA-PK antibody was used (Fig. 3B, bands labeled C) but not when RPA (Fig. 3A), Ku (Fig. 3B), or MCM (not shown) antibodies were used. In contrast, antibodies to RPA and Ku precipitated only monomer Rep78 and -68 (Fig. 3A and B). Antibodies to RFC, a complex that assembles PCNA at a 3′ OH primer site, did not precipitate Rep. This was consistent with the fact that the interaction of RFC with the 3′ OH primer is transient.

FIG. 2.

The distribution of proteins that copurified with Rep-TAP within 16 functional categories.

FIG. 3.

Interaction of Rep78 with proteins involved in AAV DNA replication. (A) AAV- and Ad-infected 293 extract was immunoprecipitated with anti-Rep 78/68, anti-RPA (70 and 34 subunits), anti-pol δ, anti-RFC, and anti-PCNA antibodies in the presence (+) or absence (−) of DNase. A total of 0.5 mg of AAV- and Ad-infected 293 cell extract was incubated with 2 to 5 μg of antibody and 15 μl of protein A agarose beads (see Materials and Methods). The immunoprecipitates were then immunoblotted with anti-Rep 52/40 antibody. (B) The same experiment as described in panel A was used except that the AAV and Ad extract was immunoprecipitated with anti-Ku (70 and 80 subunits) or anti-DNA-PK antibodies. The No Ab lanes show the incubation of AAV/Ad-infected 293 cell extract with no antibody present. The positions of Rep78 and -68 monomers are shown on the right. C indicates the positions of the Rep concatemers that are stable in SDS.

The high-molecular-weight Rep complexes have been seen previously by us and others (62) when purified Rep was stored at high concentrations and appear to be Rep-Rep protein concatemers containing two to six Rep proteins. These concatemers can form in the absence of DNA and can persist in SDS. Their appearance in some of the antibody precipitations suggests that Rep concatemers exist in a complex with PCNA and pol δ (as well as DNA-PK) and form a Rep aggregate upon the addition of SDS. The significance of this is not clear. It was also not clear why immunoprecipitation with pol δ and PCNA antibody readily demonstrated Rep interaction but only PCNA had copurified with Rep-TAP in the initial screen. In all cases, the numbers and quantities of Rep complexes precipitated by antibodies to various replication/repair proteins were approximately the same regardless of whether the extract had been treated with DNase prior to immunoprecipitation (Fig. 3, compare the DNase + and − lanes). This does not rule out the possibility that a small fragment of DNA was part of some of the complexes immunoprecipitated with Rep but does suggest that there was likely to be direct contact between Rep and pol δ/PCNA and between Rep and Ku70 and -80/DNA-PK.

Several kinds of MCM complexes are known to exist in cells (reviewed in references 14 and 65). To confirm that Rep existed in a complex with MCM proteins in vivo, Ad- and AAV-infected extracts were immunoprecipitated with antibody to MCM2 or MCM5. In both cases (Fig. 4A), Rep78, Rep68, and Rep52 coprecipitated with the MCM complex. The two antibodies precipitated somewhat different MCM complexes; both precipitated MCM2, -5, and -6, but antibody to MCM5 also precipitated MCM7 (Fig. 4B). To see if Rep-MCM complexes could form in vitro, fraction 1A purified from an Ad-infected 293 cell extract was incubated with purified Rep78 and then treated with antibody to MCM2 or MCM5 (Fig. 4C). Fraction 1A was previously shown to contain MCM complex proteins and to be capable of supporting in vitro AAV DNA replication (70). Both MCM antibodies appeared to precipitate Rep compared to the controls (Fig. 4C), but MCM5 antibody appeared to precipitate much more Rep than MCM2 antibody. Finally, to see if complexes could be formed between purified proteins, we incubated purified Rep78 with purified His-tagged MCM2 affinity-purified complexes or His-tagged MCM5 affinity-purified complexes. When this was done, we could not demonstrate an interaction between MCM and Rep78 (not shown). However, when purified MCM and Rep78 were mixed in the presence of a supercoiled plasmid (2-D plasmid [123]) that contained the AAV ITR flanked by two D sequences, antibodies to both of these His-tagged MCM complexes immunoprecipitated Rep78 (Fig. 4D), and treatment of the complex with DNase did not affect binding to Rep78 (Fig. 4D, lanes 3 and 5). This result was consistent with a direct interaction between some component of the MCM complex and Rep78, but the interaction apparently required the presence of DNA, perhaps because DNA binding modified the conformation of one or the other protein complex.

FIG. 4.

Interaction of Rep and the MCM complex. (A) AAV- and Ad-infected 293 cell extract (0.5 mg) was incubated with 2 μg of MCM2 or MCM5 antibody and 20 μl of protein A agarose beads (see Materials and Methods). The immunoprecipitate fractions were then immunoblotted with anti-Rep 52/40 antibody. The No Ab lane shows the incubation of AAV/Ad-infected 293 cell extract with no antibody present. (B) The immunoblot in panel A was reprobed with a mixture of anti-MCM2, -5, -6, and -7 antibodies. (C) Purified baculovirus-expressed Rep78 was mixed with fraction 1A (which contains MCM) prepared from an Ad-infected 293 cell extract as described in Materials and Methods and then immunoprecipitated with MCM2 or MCM5 antibody. The immunoprecipitate fractions were then immunoblotted with anti-Rep52/40 antibody. Lane 1 contained no Rep or antibody (Ab), and lane 2 contained no antibody. (D) Purified baculovirus-expressed Rep78 was mixed with affinity-purified MCM complex from HEK293 cell extracts and 2-D plasmid DNA. The mixture was immunoprecipitated with either anti-MCM2 or anti-MCM5 antibodies, and the immunoprecipitates were immunoblotted with anti-Rep52/40 antibody. Lanes 1 and 2 contained only Rep78 and MCM, respectively. Lane 5 was incubated with DNase I after incubation with MCM2 antibody. α, anti.

Additional support for the interaction between MCM and Rep came from colocalization studies using confocal immunomicroscopy (Fig. 5). Cells infected with both AAV and Ad were stained with antibodies to MCM5 or MCM7 and Rep. In both cases, MCM antibodies produced a punctate pattern consistent with the distribution of MCM on chromosome origins and replication forks. In addition, there was significant overlap with positions within the nucleus occupied by Rep protein. Because Rep has previously been shown to bind and colocalize with AAV capsid proteins (46, 57, 83, 119), we also examined the distribution of MCM and capsid protein using capsid monoclonal antibodies C37 and D3, which recognize both intact capsids and capsid intermediates (120). Again, there was partial overlap between the positions of capsid proteins (VP) and MCM6 and -7.

FIG. 5.

Confocal microscopy images of HeLa cells infected with AAV and Ad 16 h postinfection, which were stained with Rep or capsid (VP) antibody (C37 with MCM7; D3 with MCM6) and rabbit polyclonal antibodies to the indicated potential Rep-interacting proteins. Alexa Fluor 488 goat anti-rabbit IgG (green) and Alexa Fluor 633 goat anti-mouse IgG (red) secondary antibodies were used for detection. Images were acquired using a Leica TCS SP5 laser-scanning microscope with 0.2-μm sections. Scale shown is 5 μm.

Nonhomologous end joining.

Five of the proteins identified in the DNA replication/repair group are involved in recognizing double-strand breaks in DNA and promoting nonhomologous end joining (NHEJ). These are Ku70 and -80, DNA-PK, Rif1, and Rad50. There is substantial evidence that Ku70 and -80 and DNA-PK are involved in one of several pathways for converting linear input rAAV genomes into circular episomes and concatemers (30, 102, 103). DNA-PK and Ku were identified in a previous study as possible Rep-interacting proteins (76), but that study had suggested that the interaction required DNA binding. In our study, this did not appear to be the case (Fig. 3B). In addition, interleukin enhancer-binding factor 3 (Ilf3), a protein that inhibits translation and forms a complex with Ku70/80 and DNA-PK (109, 124), was identified in the screen (Table 1).

Rad50 is part of the Mre11-Rad50-NBS1 (MRN) complex. This complex binds to double-strand breaks and contains single-stranded endonuclease and double-stranded 3′-5′ exonuclease activity (reviewed in references 97 and 118). The complex also acts as a double-strand-break sensor that initiates the activation of ataxia-telangiectasia mutated kinase (ATM), which phosphorylates more than 30 proteins involved in cell-cycle control, including p53, SMC1, and BRCA1. It is worth noting that the MRN complex is a target of the Ad E1b 55K T antigen complex with the E4orf6 protein (105). The E1b55K/E4orf6 complex binds to the MRN complex and to p53 as well as an E3 ligase complex (elongins B and C, cullin 5, and Rbx1), thereby targeting MRN and p53 for protein degradation. The Ad E4orf3 protein also independently targets the MRN complex (2, 107). Although Rad50 emerged as a Rep78-interacting partner in our screen, neither Mre11, which should be present at the same molar ratio in the MRN complex, nor NBS1 was detected.

The degradation of the MRN complex is believed to serve several purposes. The most important of these is likely to prevent the formation of Ad DNA concatemers that would inhibit Ad infection and reduce the yield of Ad virus (95, 105). Because AAV also creates linear double-stranded DNA intermediates during replication, it too has been shown to be a target of the MRN complex (96). The interaction of Rad50 with Rep, therefore, may be a redundant mechanism for inhibiting MRN function that functions in addition to the Ad E1b and E4 proteins and may be more relevant when other helper viruses, such as herpes, are used. In this respect, poly-ADP ribose polymerase 1 (Parp1) was also shown to interact with Rep (Table 1). Parp1 is an essential element of the base excision repair pathway. However, recently it has also been shown to participate in an alternative NHEJ mechanism (113) that is active in the absence of Ku. Thus, Rep binding to Parp1 may serve to inhibit an alternative NHEJ pathway. Alternatively, Rep has also been shown to be necessary for the site-specific integration of AAV genomes into chromosome 19 (63, 64, 81, 117). This presumably uses an NHEJ mechanism, and the interaction with Rad50 (or Parp1) may be part of this mechanism. Rad50 has been shown to be a protein that holds the ends of broken DNA together (reviewed in reference 118). Since Rep has been shown to cleave chromosome DNA when expressed in cells either with or without Ad coinfection (12, 47), it is possible that the role of Rad50 is either to promote AAV integration in conjunction with Rep protein or to prevent chromosome inactivation by Rep-mediated nicking prior to the completion of AAV infection.

Finally, two other proteins involved in DNA replication or repair were identified. Telomere-associated protein Rif1 (Rap interacting factor 1) has been shown to be an ATM-dependent intra-S phase checkpoint that is independent of both MRN and Chk2 (98). The structural maintenance of chromosome 2 (Smc2) is part of the condensin complex that is involved in the conversion of interphase chromosomes into mitotic chromosomes prior to mitosis (43). It is not clear what Rep association with these proteins might accomplish for the AAV life cycle; however, in the case of Smc2, the interaction with Rep was confirmed by coimmunoprecipitation and confocal immunomicroscopy (see below).

Ku stimulates in vitro AAV DNA replication.

As mentioned earlier, in vitro studies of AAV DNA replication suggested that the helicase activity of the MCM complex might function as the replicative helicase. The helicase activity of Rep, when tested in the presence of pol δ, RFC, and PCNA, was not sufficient to efficiently promote strand displacement elongation during AAV DNA replication, unless an MCM complex was also present (70). However, the copurification of Ku70/80 and DNA-PK with Rep suggested that the helicase activity of Ku might also be capable of promoting strand displacement synthesis. Ku is a 3′ to 5′ DNA helicase which has been well demonstrated to recognize double-stranded DNA breaks in vivo. During AAV DNA replication, one intermediate is a blunt double-stranded DNA. Thus, Ku could potentially melt the ends of the duplex AAV DNA replication intermediate to form the double hairpin intermediate that is believed to initiate strand displacement synthesis. To test this possibility, we purified Ku70/80 to homogeneity (Fig. 6A) and tested its ability to substitute for fraction 1A in the in vitro AAV DNA replication assay as we had previously done for the MCM complex. Figure 5B shows that Ku could achieve only about 50% of the activity that we observed with PCNA, RFC, pol δ, and partially purified fraction 1A (which we previously have shown to contain the active MCM complex [70]) (Fig. 6B). This activity was not significantly increased by the addition of RPA and/or DNA-PK. To determine if Ku is essential for AAV DNA replication, we examined the activity of the Ku-containing fraction (phosphocellulose fraction 1) from xrs-5 cell extracts, which are defective for Ku, and compared it with the activity of the corresponding Ku-positive fraction from the parental CHO-K cell line (Fig. 6C). The activities were not significantly different between the lysates of these two cell lines, suggesting that Ku, although capable of acting as a replicative helicase when purified protein was added to an in vitro assay reconstituted with purified proteins, was not essential for AAV DNA replication.

FIG. 6.

Effect of Ku70/80 on AAV DNA replication. (A) Silver-stained SDS-PAGE of purified baculovirus-expressed Ku70/80 (see Materials and Methods). (B) In vitro AAV DNA replication with purified Ku70/80 protein and other purified DNA replication components. Standard replication reactions (15 μl) contained 30 mM HEPES, 7 mM MgCl₂, 0.5 mM DTT, 4 mM ATP, 6 μCi [α-³²P] dATP (3 μCi/pmol), 100 μM deoxynucleoside triphosphate, 33 ng creatine phosphokinase, 40 mM creatinine phosphate, and 100 ng NE template (pH 7.5). Reactions contained, where indicated, crude Ad-infected 293 cell extract (60 μg), RFC (0.01 μg), Rep78 (0.2 μg), PCNA (0.4 μg), fraction 1A (6 μg), pol δ (0.4 μg), RPA (0.1 μg), Ku70/80 (0.3 to 1.3 μg), and DNA-PK (1.2 μg). Replication products were subjected to DpnI digestion prior to electrophoresis. (C) In vitro DNA replication assay in which fraction 1 from CHO-K (parent cell line for xrs-5) or xrs-5 (Ku-negative mutant) cell lysates was substituted for fraction 1A from HEK293 crude extracts. Replication products were subjected to DpnI digestion prior to electrophoresis. Fraction 1 contains PCNA in addition to MCM complex.

Ad helper proteins.

Rep was expected to interact with Ad DBP. DBP has been shown to bind to Rep in vitro and to modestly stimulate in vitro DNA replication (106, 114). In addition, DBP has a role in transcriptional control and has been shown to stimulate the transcription of the AAV p5 promoter which initiates the mRNA for Rep78 and -68 (23) and to increase the translation of the AAV capsid mRNA. In vivo, the deletion of the Ad DBP gene has only a modest effect on AAV DNA replication (three- to fivefold less) and a major effect on viral yield (19). In spite of the fact that DBP is the most abundant Ad noncapsid protein in Ad-infected extracts, we could not detect an interaction between Ad DBP and Rep. This is consistent with the in vivo genetic analysis of DBP deletions and suggests that DBP is not directly involved in AAV DNA replication.

In contrast to DBP, we unexpectedly found a strong interaction with the Ad E1b55K protein, which had not been previously reported (see http://pbildb1.univ-lyon1.fr/virhostnet). In addition to its role in promoting the degradation of p53 and MRN, the 55K/E4orf6 34K complex has been shown to increase the transport of both Ad and AAV mRNA to the cytoplasm (15, 80, 91). The E4 34K protein also promotes the degradation of cyclin A and cdc2, which are necessary for S to G2M transition (34). In general, the roles of the Ad helper functions (E1a, E1b, E4, E2a, and VA) are (i) to induce the cellular S phase (E1a), which provides the necessary nucleotide precursors and replication proteins (e.g., pol δ and MCM); (ii) to freeze cells in the S phase by preventing mitosis (E4); (iii) to prevent apoptosis (E1b and VA), which provides sufficient time for AAV to replicate and package its DNA; and (iv) to increase the preferential transport and translation of AAV mRNA (E1b and E4). The apparent interaction of Rep with the E1b55K protein suggests that there may be additional roles for the 55K protein. Aside from the E1b55K protein, no other Ad protein copurified with Rep.

Ribosomes, nucleoli, and protein translation.

The largest category of Rep-interacting proteins consisted of ribosomal proteins (Fig. 2; see also http://pbildb1.univ-lyon1.fr/virhostnet). In addition, six proteins involved in translation and two proteins that are believed to be involved in ribosome biogenesis in the nucleolus, nucleophosmin (NPM1) and nucleolin, were also identified (Table 1; see also http://pbildb1.univ-lyon1.fr/virhostnet). The significance of this is not clear. Because of their abundance, ribosomal proteins are often found in proteomic screens. Nevertheless, no ribosomal proteins and only one of the translation factors were present in the control AAV plus Ad samples (see http://pbildb1.univ-lyon1.fr/virhostnet). Previous work had suggested that Rep expression changes the efficiency of translation of some AAV proteins. Trempe et al. (111) used p40 chloramphenicol acetyltransferase gene fusion constructs to show that the expression of Rep in 293 cells increased the steady-state abundance of p40 (capsid) mRNA but reduced the steady-state level of translated protein. Thus, Rep is able to inhibit the translation of the capsid p40 mRNA by an unknown mechanism, and the interaction with ribosomes and translation factors may be significant in this regard. In this respect, we note that Ilf3 has been shown to inhibit translation (124) as well as affect splicing (93) and interact with DNA-PK (109). An alternative explanation for the association with ribosomes may be the interaction of Rep with nucleophosmin (NPM1). This protein is believed to be involved in ribosome assembly and transport and binds single-stranded nucleic acid. Bevington and his colleagues (13) recently demonstrated that both the Rep and AAV capsid proteins bind and colocalize with NPM1. Given the possible involvement of NPM in ribosome biogenesis and transport, Rep interaction with NPM may be the source of the apparent interaction with ribosomal proteins seen in our study rather than a direct interaction of Rep with ribosomes during translation. NPM increases Rep binding to the ITR as well as trs endonuclease activity in vitro (13). The mechanism is not clear but suggests a more direct role for NPM in AAV replication. In addition to its role in DNA replication, Rep is also involved in DNA packaging. However, as no capsid proteins were synthesized in the extracts studied here, the interaction of NPM and Rep-TAP was not due to the presence of capsid protein but rather was due exclusively to an interaction with Rep.

Transcription factors and splicing proteins.

The expression of Rep has been shown both to affect the steady-state levels of the mRNAs initiated at the three AAV promoters (p5, p19, and p40) (9, 58, 60, 61, 66, 77, 111, 129) and to increase the level of spliced to unspliced p40 capsid mRNA (86). Therefore, it was not surprising that Rep copurified with a number of transcription factors and splicing factors (Table 1; see also http://pbildb1.univ-lyon1.fr/virhostnet). Many of the factors in the transcription and splicing groups appeared to have multiple functions. For example, SNW1 and YBOX1 have been shown to have possible roles in both transcription and splicing (8, 27, 32, 87). FUS appeared to be implicated in both cell cycle control and splicing through its interaction with ILF3, a protein from the translation group that forms a complex with DNA-PK (52, 109). A number of the factors were chromatin remodeling proteins involved in acetylation or methylation (RUVBL1, DMAP1, and prohibitin 2).

Other interacting groups.

In addition to the seven groups already discussed, representatives of nine other groups were identified. These include proteins involved in cell structure, membrane transport and vesicle formation, metabolism, mitochondrial function, nuclear pore function, protein degradation, signal transduction, and stress response as well as a group of proteins with no known function. To confirm the association of Rep with at least some of the proteins identified by mass spectroscopy in these groups, we performed coimmunoprecipitation experiments on a number of different proteins with commercially available antibodies. Crude cell lysates from cells infected with Ad and AAV were incubated with these antibodies overnight at 4°C, and the immunocomplex was precipitated with protein A agarose beads. After the beads were washed, the proteins attached were analyzed by immunoblotting to determine if Rep was bound. With one exception, Rif1 (data not shown), all of the proteins tested coimmunoprecipitated Rep (Fig. 7). The proteins tested included reticulocalbin-1 (RCN1) from the membrane transport group, which is located in the endoplasmic reticulum; FUS, an RNA binding protein from the splicing group; insulin receptor substrate 4 (IRS4) from the signal transduction group; and EDD1, a ubiquitin-protein ligase from the protein degradation group (Fig. 7A). We also tested the structural maintenance of chromosomes protein 2 (SMC2) from the DNA replication group (Fig. 7B) and importin β/NTF from the nuclear pore group, the latter because it is known to bind Ran and is a member of the importin β family (Fig. 7C). Both of these also coimmunoprecipitated Rep. In all cases, there was a clear preference for interacting with Rep78 and -68 rather than Rep52 and -40. Importin β precipitated both monomer Rep proteins and Rep complexes that were insensitive to SDS.

FIG. 7.

Immunoprecipitation of potential Rep-interacting proteins. (A) RCN, FUS, IRS4, and EDD1 were immunoprecipitated from 1 mg of an AAV/Ad-infected 293 cell extract, and the immunoprecipitate was immunoblotted with anti-Rep52/40 antibody. (B) Immunoprecipitation with anti-Smc2 antibody and immunoblotting with Rep antibody as described for panel A. (C) Immunoprecipitation with anti-importin β antibody and immunoblotting with Rep as described for panel A. S, supernatant after immunodepletion; W, wash of agarose beads with bound antibody complex; B, protein A beads containing antibody complex; C, Rep concatemers. The 78, 68, 52, and 40 markers indicate the Rep monomeric species present. The No Ab (no antibody) lanes show the incubation of AAV/Ad-infected 293 cell extract with no antibody present.

To further confirm the association of Rep with these proteins during an active AAV and Ad infection, we examined the localization of some of these proteins using immunocytochemistry (Fig. 5). Using polyclonal rabbit antibodies to the cellular proteins and Rep monoclonal antibody (anti-78/68), we observed the partial colocalization of Rep with EDD1 and SMC2 (Fig. 5) and to a lesser extent with FUS (Fig. 5) and RCN (data not shown). We note, however, that the colocalization of EDD1, SMC2, and FUS with Rep was not as robust as it was in the case of MCM. The total list of proteins confirmed to interact with Rep by either coimmunoprecipitation or colocalization is listed at http://pbildb1.univ-lyon1.fr/virhostnet.

Computer analysis of pathways represented by Rep-interacting proteins.

To see if any particular cellular pathways were populated by the 188 proteins that copurified with Rep-TAP, we used Ingenuity software, which is designed to identify relationships between proteins based on known biochemical and functional interactions. When analyzed with Ingenuity software, two pathways that included a significant number of the 188 proteins that copurified with Rep-TAP were identified, the DNA replication pathway and the protein synthesis pathway. The proteins involved in the DNA replication pathway and their interactions are illustrated in Fig. 8 and listed in Table 1 and at http://pbildb1.univ-lyon1.fr/virhostnet in the column titled replication pathway (RP). Proteins involved in the protein synthesis pathway are illustrated at http://pbildb1.univ-lyon1.fr/virhostnet in the column titled protein synthesis pathway (PSP).

FIG. 8.

Ingenuity map showing the interactions of the replication pathway. Proteins represented by the shaded symbols are proteins that were identified in the Rep-TAP screen. Proteins represented by the open symbols were not identified in the Rep-TAP screen but are necessary to fill in the pathway. A line between two pathway members indicates that a physical or functional interaction that is either direct (solid line) or indirect (dotted line), has been demonstrated by previously published work. The symbol shapes indicate various functional categories (e.g., cytokine, enzyme, etc.) as shown in the legend.

The emergence of the protein synthesis pathway was not surprising given the large number of ribosome proteins that were identified. Indeed, 75% (21/28) of the proteins in this pathway were from the ribosome, translation, or splicing/RNA binding groups (see http://pbildb1.univ-lyon1.fr/virhostnet). In contrast, the DNA replication pathway, which was the larger of the two pathways (45 versus 28 proteins), was composed of a much more diverse group of proteins and included members of every group except the membrane transport/vesicle group. In fact, proteins directly involved in DNA replication, repair, or recombination were only 25% (9/45) of the total proteins in the replication pathway. The replication pathway included proteins involved in cell structure (3), metabolism (3), mitochondrial function (3), nuclear pore function (1), nucleolus function (2), protein degradation (1 [EDD1]), ribosome function (8), signal transduction (4), splicing (2, including FUS), stress response (1), transcription (5), and translation (2) and one protein of unknown function (Fig. 8 and Table 1; see also http://pbildb1.univ-lyon1.fr/virhostnet). Some of these interactions may be indirect and have no functional significance for the AAV life cycle. For example, we noted previously that tubulin, a cell structure protein, copurified with affinity-purified MCM complex (70). Its presence in Rep-TAP complexes (Table 1, cell structure group), therefore, may simply be due to its interaction with the MCM complex and may have no significance in the AAV life cycle. Nevertheless, the diversity of proteins identified suggests that the AAV life cycle is more complex than previously recognized. Finally, we note that, taken together, the protein synthesis and replication pathways included only 71 of the 188 total proteins identified. This suggests that additional cellular pathways are affected by Rep that were not identified by the Ingenuity software.

Summary.

In summary, we have identified 188 proteins that appear to interact with Rep protein during the coinfection of cells with AAV and Ad, its helper virus. In addition to proteins involved in DNA replication and nucleolus and nuclear pore function, which were anticipated, we have identified a number of candidate transcription factors, splicing factors, and translation factors. Rep protein has previously been shown to change the splicing pattern of its p40 mRNA (86) and translation of its capsid mRNA (111), and it is required for the transcriptional activation and repression of both its own promoters and cellular promoters (1, 9, 37, 39, 44, 53, 58, 60, 67, 77-79, 121, 122). The proteins identified in these categories are, therefore, candidate factors that may have a function in transcription, translation, and splicing during the AAV life cycle.

Surprisingly, a large set of additional proteins were identified that apparently interacted with Rep but were not anticipated from previous work. These proteins come from a diverse set of functional categories, including proteins involved in mitochondrial function, signal transduction, cell structure, metabolism, membrane transport, and protein degradation. Some of these could be associated with an extended network of proteins involved in DNA replication, but the majority represent potential new and, as yet, unexplored cellular functions that Rep modifies during its life cycle. Clearly, additional experiments will be needed to evaluate the role of these proteins, as only a few proteins have thus far been validated by immunoprecipitation or colocalization in this study. In this regard, it is worth mentioning that the proteins identified in this study interacted with Rep in cells that had been infected with Ad. However, herpes simplex virus has also been shown to be capable of supporting a productive AAV life cycle (18). Presumably, similar studies of herpes simplex virus-infected cells will find a common set of proteins that will focus future work on Rep-interacting factors.

ADDENDUM IN PROOF

An article about the VirHostNet database (V. Navratil, B. de Chassey, L. Meyniel, S. Delmotte, C. Gautier, P. André, V. Lotteau, and C. Rabourdin-Combe, Nucleic Acids Res., 4 November 2008 [Epub ahead of print]) appeared after the present study was submitted.