Uracil DNA Glycosylase BKRF3 Contributes to Epstein-Barr Virus DNA Replication through Physical Interactions with Proteins in Viral DNA Replication Complex

ABSTRACT

Epstein-Barr virus (EBV) BKRF3 shares sequence homology with members of the uracil-N-glycosylase (UNG) protein family and has DNA glycosylase activity. Here, we explored how BKRF3 participates in the DNA replication complex and contributes to viral DNA replication. Exogenously expressed Flag-BKRF3 was distributed mostly in the cytoplasm, whereas BKRF3 was translocated into the nucleus and colocalized with the EBV DNA polymerase BALF5 in the replication compartment during EBV lytic replication. The expression level of BKRF3 increased gradually during viral replication, coupled with a decrease of cellular UNG2, suggesting BKRF3 enzyme activity compensates for UNG2 and ensures the fidelity of viral DNA replication. In immunoprecipitation-Western blotting, BKRF3 was coimmunoprecipitated with BALF5, the polymerase processivity factor BMRF1, and the immediate-early transactivator Rta. Coexpression of BMRF1 appeared to facilitate the nuclear targeting of BKRF3 in immunofluorescence staining. Residues 164 to 255 of BKRF3 were required for interaction with Rta and BALF5, whereas residues 81 to 166 of BKRF3 were critical for BMRF1 interaction in glutathione S-transferase (GST) pulldown experiments. Viral DNA replication was defective in cells harboring BKRF3 knockout EBV bacmids. In complementation assays, the catalytic mutant BKRF3(Q90L,D91N) restored viral DNA replication, whereas the leucine loop mutant BKRF3(H213L) only partially rescued viral DNA replication, coupled with a reduced ability to interact with the viral DNA polymerase and Rta. Our data suggest that BKRF3 plays a critical role in viral DNA synthesis predominantly through its interactions with viral proteins in the DNA replication compartment, while its enzymatic activity may be supplementary for uracil DNA glycosylase (UDG) function during virus replication.

IMPORTANCE Catalytic activities of both cellular UDG UNG2 and viral UDGs contribute to herpesviral DNA replication. To ensure that the enzyme activity executes at the right time and the right place in DNA replication forks, complex formation with other components in the DNA replication machinery provides an important regulation for UDG function. In this study, we provide the mechanism for EBV UDG BKRF3 nuclear targeting and the interacting domains of BKRF3 with viral DNA replication proteins. Through knockout and complementation approaches, we further demonstrate that in addition to UDG activity, the interaction of BKRF3 with viral proteins in the replication compartment is crucial for efficient viral DNA replication.

INTRODUCTION

Epstein-Barr virus (EBV) is a human gammaherpesvirus that causes infectious mononucleosis and is associated with many lymphoproliferative diseases, such as Burkitt's lymphoma, Hodgkin's disease, and posttransplantation lymphoproliferative disease (PTLD), and epithelial cancers, including nasopharyngeal and gastric carcinomas (1). After primary infection, the viral genome persists in a latent form in memory B cells and is reactivated periodically. Upon the induction of the lytic cycle, immediate-early transactivators Zta and Rta turn on the expression of viral proteins for viral DNA replication and, later, the structural components for virion production. The core viral DNA replication complex contains eight virus-encoded proteins, including BZLF1 (oriLyt binding protein), BRLF1 (immediate-early transactivator), BALF5 (DNA polymerase), BMRF1 (also called EA-D; polymerase processivity factor), BALF2 (single-stranded-DNA binding protein), BBLF4 (helicase), BSLF1 (primase), and BBLF2/3 (helicase-primase-associated protein) (2, 3). In addition, EBV encodes a viral uracil DNA glycosylase (UDG), BKRF3, which is a homolog of the human uracil-N-glycosylase (UNG) family (4).

The UDG superfamily comprises five protein families which share a similar structural organization but lack sequence homology at their active sites (5, 6). Among them, the UNG family (family 1) is the most ubiquitous and conserved across species (7). Mammalian UNG2, one of the two isoforms encoded by the UNG gene through alternative splicing (8), is the major UNG localized in the nucleus (9). The other isoform is the mitochondrion-specific UNG1. These two isoforms are encoded by the same gene and differ in the first 35 and 44 residues on their N termini but are identical in the remaining 269 residues (8). Under normal physiological conditions, uracils possibly are introduced into DNA by two major processes, including misincorporation of dUMP and spontaneous deamination of cytosine within DNA. The replicative incorporation of dUMP generates a U·A base pair (10). Alternatively, deamination of cytosine yields a G:U mismatch and leads to a G·C-to-A·T transition if the lesion is not repaired before the next round of replication. Normally, T-to-U or C-to-U nucleotide changes are corrected by one of the two base excision repair (BER) pathways, namely, the short- and long-patch pathways, which are initiated following recognition of uracil by UNG (11). The N-glycosylic bond between uracil and deoxyribose then is hydrolyzed by UNG, creating an apurinic/apyrimidinic (AP) site (12, 13). The 5′ end of the AP site is cleaved by AP endonuclease, and the resulting single-strand break subsequently can be processed via either a short-patch or a long-patch repair pathway (14, 15).

To operate together with the DNA replication machinery, different UNG molecules use various strategies to translocate into the nucleus. For example, human UNG2 is transported to the nucleus by an unusual nuclear localization signal (NLS) in the N terminus (16) and recruited to replication foci through the physical interactions of its N-terminal noncatalytic domains with PCNA and replication protein A (RPA) to benefit DNA replication (17). UNG2 expression is highly regulated by the cell cycle, with maximum levels and enzyme activities being detectable during late G₁ to early S phase (18). The cellular turnover, association with RPA, and modulation of catalytic activity of UNG2 are regulated through distinct CDK-mediated phosphorylation (19). The interactions of UNG2 with PCNA and RPA contribute to efficient postreplicative repair of misincorporated uracils in newly synthesized DNA (20). In addition, UNG2 also functions in prereplicative repair of U:G mismatch through direct interaction with DNA repair protein XRCC1 (21). Previously, it was found that overexpression of human UNG2 causes cell cycle delay and increases DNA damage in fission yeast, suggesting uncoordinated UNG2 activity induces DNA damage (22). Thus, specific interactions with various DNA replication or repair proteins may provide a sophisticated regulation of UDG function.

During herpesvirus infections, various cellular components of the DNA repair machineries also participate in viral replication compartments to either stimulate or inhibit viral DNA replication. Both nonhomologous end joining (NHEJ) and homologous recombination repair (HRR) and chromatin remodeling factors accumulate in herpes simplex virus type 1 (HSV-1) replication compartments (23). Mismatch repair (MMR) and HRR factors were found colocalized within EBV replication compartments (24, 25). Additionally, it was suggested that the modulation of the cellular BER pathway plays an important role in human cytomegalovirus (HCMV) replication (26). Depletion of UNG2 with a short hairpin RNA (shRNA) approach attenuated the viral DNA replication and virion production in Kaposi's sarcoma-associated herpesvirus (KSHV)-positive cells that were induced by 12-O-tetradecanoylphorbol-13-acetate (TPA) and sodium butyrate for lytic replication (27). In our previous study, EBV replication was dramatically reduced in the presence of Ugi, which can block both cellular and viral UDG activities (4). These observations indicate that cellular UNG2 contributes to viral DNA replication.

All human herpesviruses encode conserved UDGs that were found in viral DNA replication compartments. Different UDG mutant viruses also were generated to explore the exact function of herpesviral UDG during replication. For example, the UDG of HSV-1, UL2, was shown to be associated with the viral DNA polymerase UL30 and was required for efficient virus replication and reactivation in neural cells (28, 29). In a mouse infection model, recombinant HSV-1 with a truncative mutant of UL2 replicates less efficiently in the nervous system than the wild-type virus does (30). HCMV UDG UL114 associates with viral polymerase processivity factor UL44 for targeting the nucleus and forms complexes with viral DNA polymerase UL54 (31, 32). The association with UL44 enhances the UDG activity of UL114 and increases the efficiency of viral DNA synthesis (33). Deletion of HCMV UL114 in recombinant virus delayed viral replication in quiescent fibroblasts (34, 35). However, the varicella-zoster virus (VZV) UDG mutant virus, with a deletion of amino acids (aa) 160 to 282 of ORF59, grew to titers similar to those of the parental virus in vitro (36). These observations suggest that viral UDGs are equipped for viruses to replicate well in quiescent cells, where cellular UNG expression is limited. Nevertheless, individual herpesviral UDG may have unique features to promote viral DNA replication. Previously, we found EBV UDG BKRF3 colocalized with viral DNA polymerase BALF5 by immunofluorescence microscopy (4). However, the nuclear targeting mechanism of BKRF3 and the function of BKRF3 in the viral replication compartment remain to be elucidated.

BKRF3 first was shown to enhance EBV oriLyt-initiated plasmid replication in a transient cotransfection replication assay (2). Crystal structure analysis of BKRF3 in complex with the UDG inhibitor protein (Ugi) revealed that BKRF3 shares considerable similarity in overall structure with proteins in the family 1 UNGs. Four out of the five catalytic motifs are conserved completely, whereas the fifth domain carries a seven-residue insertion in the leucine loop, indicating that the leucine loop of BKRF3 plays a role in viral replication (37). Previously, we characterized the biochemical properties of BKRF3 with DNA glycosylase and mutator assays. BKRF3 was able to complement the phenotype of an Escherichia coli ung mutant and showed a higher efficiency in removing uracil from an artificial single-stranded DNA (ssDNA) probe than double-stranded DNA (dsDNA) in vitro. Using short interfering RNA (siRNA) to knock down BKRF3 expression caused an approximately 20% decrease in viral DNA replication in EBV-positive NA cells (4). In this study, we aimed to examine further the role and functional domains of BKRF3 in EBV lytic replication. We show that BKRF3 is translocated into the nucleus and colocalizes with the replication loci when EBV-positive cells are induced into the lytic cycle. This observation is supported by the fact that BKRF3 was immunoprecipitated with other EBV proteins within the replication compartment. Using a bacmid system, we show that a BKRF3 knockout recombinant EBV is defective for lytic DNA replication, and this defect can be restored by the expression of wild-type BKRF3 or the catalytic domain mutant of BKRF3 but not the leucine loop mutant of BKRF3. In addition to UDG activity, the interaction of BKRF3 with the viral DNA replication compartment is critical for EBV lytic DNA replication.

MATERIALS AND METHODS

Plasmids.

The full-length BKRF3 PCR products were amplified by primers 5′-CGGGATCCATGGCATCGCGGGGGC-3′ and 5′-CGGGATCCCTACAGCCTCCAATCTATC-3′ using Akata cell lysates as the template. BKRF3 products were digested with BamHI and were cloned into the BamHI site of pCMV-Tag2B (Stratagene) to generate pFlag-BKRF3 (pNC1). To generate BKRF3 enzymatically dead plasmid pFlag-BKRF3(Q90L,D91N) (pCYC11) or BKRF3 leucine loop mutant plasmid pFlag-BKRF3(H213L) (pCYC12), single-primer mutagenesis (38) was performed using the primer 5′-TGGTTATTTTGGGCCTCAACCCCTATCACGGGG-3′ or 5′-TTCTGACCTCTCAGCTTCCCTCTCCCCTGGC-3′. In addition, these three BKRF3 constructs were individually subcloned into the pHY25 plasmid, which is a pSG5 (Stratagene) derivative inserted with the hemagglutinin (HA) sequence and multiple cloning sites, to generate HA-BKRF3 (pIH4), HA-BKRF3(Q90L,D91N) (pIH5), and HA-BKRF3(H213L) (pIH6). For glutathione S-transferase (GST) pulldown assay, BKRF3 was subcloned into the EcoRI site of pGEX-4T1 (GE Healthcare) to generate pGEX-4T1-BKRF3 (pSMC8). The pGEX-4T1-BKRF3 mutants, including GST-d(1-30) (pSMC9), GST-d(28-83) (pSMC10), GST-d(81-166) (pSMC11), and GST-d(164-255) (pSMC12), were constructed by single-primer mutagenesis using primers LMRC654 (5′-TCCCCGGAATTCATGCTCCCCGACTTATGG-3′), LMRC655 (5′-GGTGTGAAAGGAGAAAATGACCCCTCTGATATTAAG-3′), LMRC656 (5′-CTGGGCCCGCTTTTGCTGGGCGTGGTTTACTG-3′), and LMRC657 (5′-GCCCGGCTCGCACGCATAGGAATTCCCGGGTC-3′). pMTS8, a pCR3.1 (Invitrogen)-based plasmid expressing HA-tagged BMRF1, was generated by inserting the HA sequence into pYPW88 (pCR3.1-BMRF1) (39). The Rta expression plasmid RTS15 and BALF5 expression plasmid pDH312 were gifts from Diane Hayward (40) and Gao et al. (41).

Cell lines and transfection.

The HEK293T (293T) cell line is a derivative of a human kidney epithelial cell line (CRL-1573; ATCC). The HeLa cell line was derived from human cervical epithelial cells (CCL-2; ATCC). The NA cell line, an EBV-positive cell line latently infected with recombinant Akata strain EBV (42), was selected from its parental cell line, NPC-TW01, an EBV-negative nasopharyngeal carcinoma (NPC) epithelial cell line (43). The EBV lytic cycle can be induced in NA cells with the treatment of 40 ng/ml 12-O-tetradecanoylphorbol-13-acetate (TPA) and 3 mM sodium butyrate (SB) (42). The 293TetER cell line, an Rta-inducible cell line constructed from the T-Rex 293 cell line (Invitrogen), carries an inducible Flag-Rta plasmid (44). The EREV8 cell line, constructed from the T-REx 293 cell line, carries inducible Flag-Rta plasmid DNA and also contains the Akata EBV genome. Doxycycline treatment of EREV8 cells induces Rta expression, which in turn triggers the EBV lytic cycle (45). DNA transfection was performed using Lipofectamine 2000 transfection reagent (Invitrogen) according to the protocol suggested by the manufacturer. The cells were incubated for the indicated periods of time at 37°C with 5% CO₂. Complementation of BKRF3 plasmids into 293TetER/p2089BKRF3STOP cells was performed with the calcium phosphate–N,N-bis(2-hydroxyethyl)-2-aminoethanesulfonic acid (BES)-buffered solution (BBS) transfection protocol (46).

Immunofluorescence assay.

Cells were cultured on fluorescence-negative glass slides and transfected as required. At the harvest time points indicated, the slides were air dried and fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) at room temperature for 20 min. The slides were washed with PBS and incubated with anti-Flag (M2; Sigma-Aldrich), anti-HA (HA.11; Covance), or rabbit anti-BALF5 serum (DP-1) at 37°C for 1.5 h, washed, and subsequently incubated with a fluorophore-conjugated secondary antibody at 37°C for 1 h. Finally, the slides were stained with 100 ng/ml Hoechst 33258 at room temperature for 1 min, covered with H1000 mounting medium (Vector Laboratories), and observed by fluorescence (Zeiss, Axioskop 40 FL) or LSM 510 META confocal microscopy (Zeiss).

Western blotting.

Western blotting was performed as described previously (47). To detect the EBV lytic proteins, the primary antibodies used were laboratory-made mouse anti-BKRF3 serum 3, mouse anti-Zta 1B4, anti-Rta 467, anti-BGLF4 2616, and anti-BMRF 88A9, as described previously (4, 48). The BALF2 antibody (OT13B) was kindly provided by J. M. Middeldorp (49). Other primary antibodies used were anti-gp350/220 (72A1; ATCC), anti-HA antibody (HA.11; Covance), anti-PARP (F-2; Santa Cruz Biotechnology), anti-α-tubulin (DM1A; Calbiochem), anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Biodesign), anti-actin (Sigma-Aldrich), anti-myc (9E10) (50), and UNG polyclonal antibody (ab23926; Abcam).

Subcellular fractionation.

The subcellular fractionation protocol was modified from a previous study (4). Briefly, trypsinized cells were harvested, washed, and incubated with 1 ml hypotonic buffer (5 mM Tris-HCl, pH 7.4, 5 mM KCl, 1.5 mM MgCl₂, 0.1 mM EGTA, 1 mM dithiothreitol [DTT], and 1 mM phenylmethylsulfonyl fluoride [PMSF]) with gentle shaking at 4°C for 1 h. The cell suspension then was passed 15 times through a 26-gauge needle and centrifuged at 500 × g at 4°C for 5 min. The pellet was collected as the nuclear fraction. The supernatant was mixed with a 0.1× volume of 72% trichloroacetic acid and a 0.1× volume of 0.15% deoxycholic acid with gentle shaking at 4°C for 90 min and was centrifuged at 15,000 × g at 4°C for 20 min. The resulting pellet was washed with 95% ethanol twice and air dried as the cytoplasmic fraction.

Coimmunoprecipitation assay.

About 1 × 10⁷ cells were lysed with 1 ml of CSK buffer (48) with gentle shaking at 4°C for 2 h. Cell debris was precipitated by centrifugation at 16,000 × g at 4°C for 10 min, and supernatant was collected. The supernatant was precleaned with 100 μl of 20% protein A-Sepharose beads (GE Healthcare) for 1 h at 4°C. The precleaned lysate then was incubated with 1 μg of anti-Flag, anti-Rta, anti-GST, or anti-HA antibody for 2 h at 4°C, followed by incubation with 100 μl of 20% protein A-Sepharose beads for 1 h at 4°C. Immunocomplexes were collected and washed with CSK buffer once and with cold PBS three times. Immunocomplexes were disrupted by SDS-sample buffer, and the interacting proteins were detected by Western blot analysis.

GST pulldown assay.

Recombinant protein expression in E. coli transformants was induced by adding isopropyl-β-d-thiogalactopyranoside (IPTG) at a concentration of 0.1 mM to 100 ml of bacterial cultures of an optical density at 600 nm (OD₆₀₀) of 0.6 to 0.8 and then incubating it at 25°C for 2 h. The bacteria then were harvested and resuspended in 2 ml of PBST (1× PBS, 1% Triton X-100, and 1× protease inhibitor) with 1 mg/ml lysozyme, incubated on ice for 1 h, and subjected to a −80 to 37°C freeze-thaw cycle. The bacteria were disrupted with sonication on ice, and the insoluble portion was removed by centrifugation at 12,000 × g for 10 min. The buffer-soluble supernatant was mixed with 150 μl of 50% glutathione-Sepharose beads (GE Healthcare) and incubated on a rotating platform with a speed of 5 rpm at 4°C for 2 h. The beads were washed three times with 1 ml of cold PBST with 10 rpm rotation at 4°C for 5 min. Cell lysates from Rta- or BALF5-transfected 293T cells were added and incubated with the beads in binding buffer (1% Triton X-100, 20 mM Tris-HCl, pH 8.0, 140 mM NaCl, 1 mM EGTA, 10% glycerol, 1.5 mM MgCl₂, 1 mM DTT, 10 μg/ml PMSF, and 1× protease inhibitor) at 4°C for 18 h. The beads then were washed three times with 1 ml of washing buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 1% NP-40) and subjected to SDS-PAGE analysis.

UDG assay.

The single-stranded oligonucleotide LMRC-U (5′-AGCTACCATGCCTGCACGAAUTAAGCAATTCGTAATCATGGTCAT-3′) was labeled with [γ-³²P]ATP at the 5′ end, purified, and quantified as previously described (4, 51). Typically, labeled oligonucleotides equivalent to 4 × 10⁵ cpm were incubated with E. coli UNG enzyme (NEB) or cell lysates at 37°C for 10 min. A standard assay was carried out in 20 μl of the buffer containing 1 mM EDTA, 1 mM DTT, and 20 mM Tris-HCl, pH 8.0. The UDG activity was stopped by heating the reaction mixtures at 95°C for 5 min. After glycosylase cleavage, abasic sites were incised by 0.1 mM NaOH treatment at 95°C for 5 min. Reaction products were analyzed with electrophoresis on 15% (wt/vol) polyacrylamide denaturing gels (7 M urea, 1× Tris-borate-EDTA), and the gels were dried and subjected to autoradiography. The cleavage percentage of U probe was quantified with ImageQuant (GE Healthcare).

Construction of the BKRF3 knockout and revertant EBV bacmids and selection of doxycycline-inducible cells containing EBV bacmid DNA.

The shuttle vector pGS284, donor strain S17λpir E. coli (GS111), and recipient strain GS500 (recA⁺), used for EBV allelic exchange, were generous gifts from R. Sun (University of California, Los Angeles) (52). The 5′- and 3′-flanking regions of BKRF3 (B95.8 strain 110072-111072) were amplified by PCR using primers LMRC785 (5′-GAAGATCTCTTCTCGCGTTGGAAAACATTAGCGAC-3′) and LMRC786 (5′-AAGATCTTTAGCGAGGACAAAGTGGTTGTTGCCC-3′). The PCR product was digested and cloned into the BglII site of pGS284 to generate pIH1 (pGS284/BKRF2-3). Sequentially, the stop cassette containing triple-open reading frame (ORF) nonsense codons and an NheI site was inserted into pIH1 between nucleotides 110572 and 110573 of the B95.8 genome by double-primer PCR mutagenesis (53) using primers LMRC808 (5′-CATGTGCATGGCCGCTAGCTTGATTAATTGATGGGCCCGCTTTT-3′) and LMRC809 (5′-AAAAGCGGGCCCATCAATTAATCAAGCTAGCGGCCATGCACATG-3′). The resulting plasmid, pIH3 (pGS284/BKRF3STOP), was electroporated into E. coli strain GS111 for allelic exchange. The EBV bacmid p2089 (54), a kind gift from H. J. Delecluse (DKFZ unit F100, Heidelberg, Germany), was electroporated into E. coli strain GS500 (recA⁺). For allelic exchange, conjugation was performed by cross-streaking GS500/p2089 and GS111/pIH3 on LB agar at 37°C for 16 h, and recombinant bacmids were selected according to a previously described procedure (55). The incorporation of the stop codon in the BKRF3 open reading frame was determined by colony PCR using primers LMRC785 and LMRC786 and restriction enzyme digestion for the insertion of the NheI site. Furthermore, the GS500/p2089BKRF3STOP bacmid was cross-streaked with GS111/pIH1 to generate the revertant bacmid of p2089BKRF3STOP by allelic exchange.

To select doxycycline-inducible EBV bacmid-positive cell lines, 293TetER cells (5 × 10⁵ cells/well) were seeded in a 6-well culture dish and transfected with 7 μg of p2089, p2089BKRF3STOP, or BKRF3 revertant (K3R) using T-Pro NTRII transfection reagent (T-Pro Biotechnology). At 72 h posttransfection, transfected cells were split into two 10-cm culture dishes and selected with hygromycin B (100 μg/ml) for 1 month. Four to 6 green fluorescent protein (GFP)-positive cell colonies were picked up to obtain pool clones. More than 10 pooled clones of individual transfected cells were selected. The selected 293TetER/p2089, 293TetER/p2089BKRF3STOP, or 293TetER/K3R pool clones were treated with doxycycline (50 ng/ml) to confirm successful lytic induction by Western blotting.

Immunoprecipitation-UDG assays.

Cells were harvested and resuspended in extraction buffer (10 mM Tris-HCl, pH 8.0, 200 mM KCl, 1 mM EDTA, 20% glycerol, 0.25% NP-40, 1 mM DTT, 1× protease inhibitor). The cell mixture was rotated at 4°C for 3 h, and cell debris was removed by centrifugation at 16,000 × g for 5 min. Two hundred micrograms of cell extract was incubated with 1 μg anti-HA antibody on a rotating shaker at 3 rpm at 4°C for 3 h. Subsequently, 100 μl of 20% protein A-Sepharose beads was added to the mixture, which then was incubated at 4°C for 1 h. Immunocomplexes were collected and washed three times with 10 mM Tris-HCl (pH 7.4) at 4°C for 15 min. The protein A bead-bound immunocomplexes that comprised 10% of the total immunocomplexes were resuspended in 2 μl UDG extraction buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 250 mM NaCl, and 1 mM DTT) for the UDG assay; the other 90% of immunocomplexes were analyzed by Western blotting.

Genomic DNA extraction and quantitative real-time PCR for EBV copy number.

Cells were lysed in 400 μl digestion buffer (100 mM NaCl, 10 mM Tris-Cl, pH 8.0, 25 mM EDTA, pH 8.0, 0.5% SDS, 0.1 mg/ml proteinase K) and incubated at 55°C for 3 h. RNase A (0.5 mg/ml) then was added and incubated at 55°C for another 20 h. The genomic DNA was extracted with an equal volume of phenol-chloroform-isoamyl alcohol (25:24:1) and precipitated by adding 2 volumes of 100% ethanol and a 1/10 volume of 3 M sodium acetate (pH 5.2). Twenty-five ng of total DNA was evaluated for EBV genomic DNA by PCR or quantitative PCR (qPCR) using the following BamHI W primers: 5′-CCCTGGTATAAAGTGGTCCT-3′ and 5′-AAGTCCACTTACCTCTGG-3′. The copy number of cellular β-globin was detected for normalization using following primers: 5′-GGTTGGCCAATCTACTCCCAGG-3′ and 5′-GCTCACTCAGTGGCAAAG-3′. For qPCR, EBV DNA was quantified using a SensiFAST SYBR No-ROX kit (Bioline). The standard curve for qPCR was generated by a 10-fold serial dilution of a mixture of 10⁴ copies of genomic DNA of 293TetER cells and 2.5 × 10⁶ copies of purified EBV bacmid DNA.

DNA binding assay with single-stranded DNA cellulose.

The [³⁵S]methionine-labeled wild-type and mutant HA-BKRF3 proteins, prepared by in vitro TNT quick coupled transcription/translation systems (Promega), were used for ssDNA cellulose chromatography. The single-stranded calf thymus genomic DNA cellulose resin (Sigma-Aldrich) was equilibrated with the binding buffer (50 mM Tris-HCl, pH 8.0, 1 mM PMSF, and 10% glycerol). Thirty μl of in vitro transcription/translation products was diluted 1:10 in binding buffer containing 100 μg RNase A and applied to a 200-μl bed volume of ssDNA cellulose resin in a Poly-Prep chromatography column (Bio-Rad). After washing with 300 μl binding buffer, the bound proteins were eluted with 300 μl serial gradients of binding buffer containing 0.1, 0.2, 0.3, 0.4, and 0.5 M NaCl. Three μl of transcription/translation products and 30 μl of eluents were subjected to electrophoresis in 15% SDS-PAGE gels and analyzed with a Typhoon Trio variable-mode imager (GE Healthcare).

RESULTS

BKRF3 was translocated into the nucleus during EBV lytic replication.

Previously, we showed that BKRF3 expression could be detected in EBV-positive NA cells upon EBV reactivation in both the nuclear and cytosolic fractions (4). However, when BKRF3 was transiently expressed alone in EBV-negative 293 cells, it was located only in the cytoplasm, suggesting that interaction of BKRF3 with other viral factors is required for BKRF3 nuclear targeting. Here, we examined further the localization of BKRF3 with or without EBV lytic cycle induction to gain a better understanding of the expression dynamics of proteins interacting with BKRF3 in the cells. A Flag-BKRF3 expression plasmid was transfected into an Rta-inducible EBV-positive 293-derived cell line, EREV8, and the lytic cycle was induced with doxycycline at 24 h after transfection. At 48 h postinduction, protein expression was detected using an immunofluorescence assay. As shown in Fig. 1A, without Rta induction, BKRF3 was distributed mainly in the cytoplasm, whereas with Rta induction, BKRF3 distribution moved into the nuclei, as evidenced by the superimposed BKRF3 and Hoechst-stained images. Moreover, BKRF3 showed a colocalized distribution with the EBV DNA polymerase, BALF5. In subcellular fractionation and immunoblot analysis, most of the BKRF3 protein was detected in the cytoplasmic fraction before Rta induction by doxycycline and shifted to the nuclear fraction after induction (Fig. 1B). Similarly, transfection of Flag-BKRF3 and Rta plasmids into an EBV-positive NPC cell line, NA, resulted in a cytoplasmic distribution of BKRF3 in the absence of Rta expression. When Rta expression was induced by doxycycline, BKRF3 was translocated into the nuclei and colocalized with BALF5 at discrete sites (Fig. 1C), possibly the replication compartments (56). More than 90% of Flag-BKRF3-expressing cells showed a nuclear staining pattern and colocalized with BALF5, suggesting that BKRF3 is recruited to the nuclei through association with the viral DNA replication machinery.

FIG 1.

Translocalization of BKRF3 during the lytic cycle. (A) In EREV8 cells, an Rta-inducible EBV-positive cell line, BKRF3 protein was mostly distributed in the cytoplasm before Rta expression is induced (left) and is transported into the nucleus and colocalized with BALF5 DNA polymerase after induction with 100 ng/ml doxycycline for 48 h (right). (B) Western blotting of fractionations of cell lysates indicates that in noninduced EREV8 cells, BKRF3 was mainly in the cytoplasmic (C) fraction, whereas after induction, a greater amount of BKRF3 was also detected in the nuclear (N) fraction. T, total; PARP, poly-ADP-ribose polymerase, a nuclear protein marker. α-Tubulin serves as a cytoplasmic marker. (C) In EBV-positive NA cells cotransfected with vector control, Flag-BKRF3 was distributed in the cytoplasm at 48 h posttransfection (left) and was translocated to the nucleus when viral replication was induced by cotransfection of the Rta expression plasmid (right). The BKRF3 protein may interact with the viral DNA replication-associated proteins, as evidenced by its colocalization with BALF5 DNA polymerase in cells with virus replication. (D) To monitor temporal expression of EBV BKRF3 and cellular UNGs, lytic replication was induced in EBV-positive NA cells with TPA-SB. BKRF3 expression was detected at 12 h and significantly upregulated at 36 h postinduction. The protein expression levels of mitochondrial UNG1 remained stable, and nuclear UNG2 decreased through the time course. (E) The upregulated BKRF3 may compensate for the reduced amount of cellular UNG2 according to the results of the UDG activity analysis.

To monitor the temporal expression profile of BKRF3 during the stages of EBV replication, EBV-positive NA cells were induced into lytic replication using TPA-SB. Protein expression of several lytic-cycle proteins, Rta, Zta, BMRF1, BGLF4, and BKRF3, as well as cellular UNG1 and UNG2, was determined at various time points up to 60 h postinduction. As shown in Fig. 1D, expression of BKRF3 in NA cells was detectable from 12 h postinduction and increased significantly between 24 and 36 h postinduction. After that, the expression of BKRF3 remained relatively stable until the end of the time course. Expression of UNG1, which is one of the UDG products located in the mitochondria, was not affected by the expression of BKRF3 or other viral proteins. On the contrary, expression of UDG2, which is the nucleus-localized UDG, gradually decreased over the time course of induction. To determine whether this decrease of UDG2 expression caused an overall reduction of glycosylase activities, total cell lysates were collected at each time point and subjected to a DNA glycosylase assay (4). As illustrated in Fig. 1E, incubating ³²P-labeled 45-mer probes with cell lysates containing DNA uracil glycosylases results in AP site cleavage following alkaline treatment, generating ³²P-labeled 20-mer fragments. The probe cleavage rate was about 76% before induction and increased to 90% at 60 h after TPA-SB induction (Fig. 1E). The time course assay indicated that, after the EBV lytic cycle was induced, total DNA glycosylase activities increased slightly, even though the expression of cellular UNG2 was reduced, and the gain of total UDG activities likely is from the increased expression of BKRF3 during EBV lytic replication.

Physical interactions of BKRF3 with other EBV DNA replication proteins and the nuclear targeting of BKRF3.

Because participation of UDG in the DNA replication core machinery was seen for both mammalian UNG2 and some viral UDGs, we performed coimmunoprecipitation assays to examine the physical association between BKRF3 and EBV replication-associated proteins. In a preliminary test with a doxycycline-induced EREV8 lysate and anti-Flag antibody, we observed that the viral DNA polymerase and Rta, but not Zta, were coimmunoprecipitated with Flag-BKRF3 (data not shown). To analyze the interactions further, Myc-BALF5, Rta, and Flag-BKRF3 expression plasmids were transfected into 293T cells, and protein expression was detected by Western blotting (Fig. 2A). In the coimmunoprecipitation reaction with anti-Flag antibody, Rta and Myc-BALF5 were detected on the immunoblot with the respective antibodies, indicating a complex comprised of at least these three proteins (Fig. 2B, lane 7). Moreover, Rta or Myc-BALF5 also formed a protein complex with Flag-BKRF3 (Fig. 2B, lanes 4 and 5). Alternatively, the immunocomplexes were captured with anti-Rta antibody and immunoblotted, which indicates interaction between Rta and Myc-BALF5 (Fig. 2C, lane 6). Together, these results suggest that Flag-BKRF3, Rta, and Myc-BALF5 form a complex through direct or indirect interactions. However, Flag-BKRF3 was not detected in the Rta antibody-captured immunocomplex, possibly because Rta antibody interferes with complex formation or the interaction among Rta, BALF5, and BKRF3 prevents epitope recognition by Rta antibody.

FIG 2.

BKRF3 interacts with viral DNA replication-associated proteins. Immunoprecipitation (IP) and Western blotting assays were used to examine protein interactions among BKRF3, BALF5, and Rta in a transient cotransfection system. (A) Transfection of plasmid DNA expressing BALF5, Rta, or BKRF3 into 293T cells shows the expression of each protein. Vector plasmids were supplemented to ensure equal amounts of total DNA in each reaction. Cell lysates of transfected cells were displayed by Western blotting and immunoprecipitated with anti-Flag (B) or anti-Rta (C) and detected with anti-BALF5, anti-Rta, or anti-Flag in the subsequent Western blotting.

Since BKRF3-interacting proteins Rta and BALF5 localize to replication compartments and are required for lytic EBV replication (3, 57, 58), we examined whether Rta or BALF5 can translocate BKRF3 into the nucleus. To this end, HA-BKRF3 plasmid was cotransfected with vector control or plasmid expressing Rta or Myc-BALF5 into HeLa cells and examined with immunofluorescence staining. In Rta-expressing cells, BKRF3 distributed in the cytoplasm predominantly the same as in BKRF3-only cells, indicating expression of Rta did not change the cytoplasmic distribution of BKRF3 (Fig. 3A, upper). Interestingly, we found that Myc-BALF5 displayed a partial colocalization pattern with BKRF3 in the cytoplasm (Fig. 3A, lower). We searched for candidates that can promote the nuclear targeting of BKRF3. Previously, UL114, the UDG of HCMV, was found to associate with DNA polymerase processivity factor UL44 in HCMV-infected cells (33). A recent study further revealed that EBV DNA polymerase processivity factor BMRF1 interacts with BALF5 through the assistance of HSP90 to promote BALF5 nuclear targeting (59). Thus, we explored whether BMRF1 can directly interact with and promote the nuclear targeting of BKRF3 or if the complex formation of BMRF1-HSP90-BALF5 is required for nuclear targeting of BKRF3. In immunofluorescence analysis, we found that coexpression of BMRF1 can effectively promote the translocation of BKRF3 from the cytoplasm into the nucleus in about 90% of coexpression cells (Fig. 3B), suggesting the interaction with BALF5 is not required for this process. In coimmunoprecipitation assays, BKRF3 was detected in the immunocomplexes captured with anti-HA antibody in HA-BMRF1- and Flag-BKRF3-coexpressed cells (Fig. 3C lane 5). Furthermore, BKRF3 and BALF5 also were detected in BMRF1-associated immunocomplexes in the presence of all three proteins (Fig. 3C, lane 7). Reciprocally, BMRF1 and BALF5 also were detected in the anti-Flag antibody pulldown immunocomplexes (Fig. 3D, lanes 5 and 6). Interestingly, the amount of HA-BMRF1 pulled down was less and Myc-BALF5 was not captured by anti-Flag antibody in the presence of all three proteins, suggesting the triple complex is detected by anti-Flag antibody less efficiently (Fig. 3D, lane 7). Taken together, data here indicate that BKRF3 interacts with viral replication-associated proteins, including BALF5, Rta, and BMRF1, and the nuclear targeting of BKRF3 is regulated by BMRF1.

FIG 3.

BKRF3 is translocated into the nucleus in the presence of BMRF1. (A) HA-BKRF3 plasmid was transfected with vector control, Rta, or Myc-BALF5 plasmid into HeLa cells and detected with anti-HA, anti-Rta (467), or anti-Myc (9E10) antibody at 24 h posttransfection in an immunofluorescence assay. DNA was stained with Hoechst 33258. (B) Flag-BKRF3 was transfected into HeLa cells with HA-BMRF1 or control plasmid. At 24 h posttransfection, cells were fixed and stained for BKRF3, BMRF1, and DNA with anti-Flag antibody, anti-HA antibody, and Hoechst 33258, respectively. (C and D) 293T cells were transfected with the plasmids indicated for 24 h and harvested using radioimmunoprecipitation assay (RIPA) buffer. Cell lysates were incubated with anti-HA (C) or anti-Flag (D) antibody to precipitate BMRF1-associated or BKRF3-associated immunocomplexes and examined for BALF5, BMRF1, or BKRF3 by Western blotting. *, Ig light chain; **, Ig heavy chain.

We then used a GST pulldown assay to map different domains on BKRF3 responsible for interacting with different DNA replication proteins. The GST-BKRF3 wild type and a series of deletion mutants were generated (Fig. 4A). Indeed, BALF5, Rta, or BMRF1 expressed in 293T cells was pulled down by bacterially expressed GST-tagged wild-type BKRF3 (Fig. 4B to D). GST-fused wild-type and mutant BKRF3 proteins were used in GST pulldown assays to identify the domain important for interacting with BALF5 (Fig. 4E), Rta (Fig. 4F), or BMRF1 (Fig. 4G). BALF5 and Rta were pulled down by GST-tagged wild-type BKRF3 and BKRF3 mutants d(1-30), d(28-83), and d(81-166) but not by d(164-255). This suggests that residues 164 to 255 of BKRF3 are crucial for its interaction with Rta and BALF5. On the other hand, BMRF1 was pulled down by the GST-tagged wild-type BKRF3 and all BKRF3 mutants except d(81-166), suggesting BKRF3 forms complexes with BMRF1 with a different interacting domain. The interacting abilities of different GST deletion clones with BALF5, Rta, or BMRF1 are shown in Fig. 4A. In summary, the C-terminal region of BKRF3 is important for interaction with BALF5 and Rta, whereas the middle region of BKRF3 is critical for interacting with BMRF1.

FIG 4.

Mapping the interaction domains of BKRF3 using a GST pulldown assay. (A) Four GST-fused deletion mutants of BKRF3 were constructed, d(1-30), d(28-83), d(81-166), and d(164-255). The relative BKRF3-interacting abilities of individual GST fusion proteins are summarized on the right. (B to G) Bacterially expressed GST or GST-fused BKRF3 proteins were purified with glutathione beads and incubated with lysates of 293T cells transfected with Myc-BALF5, Rta, or HA-BMRF1 for 18 h. The beads were washed and examined with anti-Myc (B and E), anti-Rta (C and F), and anti-HA (D and G) antibodies by Western blotting.

Complex formation of BKRF3 with viral proteins correlated with higher UDG activity of BKRF3 during EBV reactivation.

As shown in Fig. 1, the reduced cellular UNG2 coupled with upregulated BKRF3 during the EBV lytic cycle, implying BKRF3 restores the UDG activity in virus-replicating cells. Because we have shown that the UDG activity of purified recombinant BKRF3 is about 10-fold less than that of E. coli UDG in an in vitro UDG assay (4), we wondered whether BKRF3 functions through interaction with proteins in the viral DNA replication complex to achieve a stronger UDG activity in vivo. To test this hypothesis, Flag-BKRF3 proteins were enriched from Rta or vector-transfected NA cell lysates by immunoprecipitation with anti-Flag antibodies, and immunocomplexes were subjected to UDG assays. The immunoprecipitation-Western blotting indicates that BKRF3 forms a complex with BALF5, Rta, and BMRF1 as described above (Fig. 5A, lane 8). Simultaneously, the Rta-transfected immunocomplex associated with BKRF3 conferred a significantly greater level of UDG activity (92% of probe cleavage) than the lysate without Rta transfection (49% of probe cleavage) (Fig. 5B, lanes 7 and 8). The BKRF3 complexes associated with viral replicating proteins, including BALF5, Rta, and BMRF1, displayed enhanced UDG activity. Although it cannot be excluded that the enhancement is attributed to interaction with other viral proteins resulting in posttranslational modification or enhanced stability of BKRF3, it is evident that the formation of BKRF3-associated complexes enhanced BKRF3 UDG activity during EBV reactivation.

FIG 5.

Complex formation with viral DNA replication machinery in Rta-reactivated NA cells stimulates BKRF3 UDG activity. NA cells were seeded and transfected with vector or BKRF3-expressing plasmid, coupled with Rta or uncoupled, to induce lytic replication. At 60 h posttransfection, whole-cell lysates were harvested and Flag-BKRF3 was immunoprecipitated by anti-Flag (M2) antibody for coimmunoprecipitation assay and IP-UDG assay. (A) Ninety percent of Flag-BKRF3-associated immunocomplexes were examined for viral lytic proteins, including Rta, BALF5, and BMRF1 by immunoblotting assay. (B) Ten percent bed volumes of immunoprecipitated products were subjected to IP-UDG assay. The percentage of U-probe cleavage was quantified by ImageQuant (GE Healthcare) and normalized to the relative fold of immunoprecipitated Flag-BKRF3, as shown in panel A. Data are representative of two independent experiments.

Generation of a BKRF3 knockout and the revertant EBV bacmid clones.

To investigate further the involvement of BKRF3 in DNA replication during EBV lytic replication, we constructed the p2089BKRF3STOP EBV bacmid (BKRF3STOP) by an allelic exchange procedure (52). A three-frame stop codon cassette was inserted at amino acid 73 of BKRF3, and the incorporation of the cassette in BKRF3 was confirmed by restriction enzyme digestion of the inserted NheI site next to the stop codons (Fig. 6A). Using BKRF3STOP bacmid as the backbone, BKRF3 revertant bacmid (K3R) was constructed by allelic exchange with wild-type BKRF3 sequence (Fig. 6B). Compared to the wild-type bacmid p2089, the 24.6-kb DNA fragment containing BKRF3 was cut into two fragments of 16.8 kb and 7.8 kb in the BKRF3STOP bacmid by NheI digestion. The stop codon cassette was replaced with wild-type sequence of BKRF3 in revertant K3R and resulted in the 24.6-kb BKRF3-containing fragment (Fig. 6B and C, right). Furthermore, the BamHI fragmentation analysis was performed to confirm that no other recombination sites exist except the BKRF3 target site (Fig. 6C, left). Wild-type p2089, BKRF3STOP, and K3R bacmids were transfected into 293TetER cells, in which Rta expression can be induced by tetracycline for the subsequent progression of the lytic cycle genes (44), to establish 293TetER/p2089 (p2089), 293TetER/p2089BKRF3STOP (BKRF3STOP), and 293TetER/K3R (K3R) stable clones.

FIG 6.

Construction and characterization of BKRF3 knockout and the revertant EBV bacmids. (A) Schematic summary of the BKRF3 mutant EBV bacmid cloning strategy. The termination cassette containing an NheI site and translation stop codons was inserted into nucleotide 110572 of the EBV B95.8 genome (GenBank accession no. V01555.2) by allelic exchange, as described in Materials and Methods. (B) Recombination of allelic exchange led to a size change of NheI fragments from 24.6 kb to 16.8 and 7.8 kb, resulting in the disruption of BKRF3 without interfering with the coding sequence of BKRF2. The BKRF3 revertant bacmid (K3R), which contains the same genomic pattern with which the wild-type bacmid was generated, used BKRF3STOP bacmid as the backbone. (C) The wild-type, BKRF3STOP, and K3R bacmids were digested with BamHI or NheI and displayed by agarose gel electrophoresis. The fragmented viral DNA of wild-type, BKRF3STOP, and K3R bacmids were identical in BamHI digestion (left). In addition, cleavage of the BKRF3STOP bacmid produced 16.8- and 7.8-kb fragments (right, lane 2) that replaced the wild-type 24.6-kb fragment (right, lane 1), and the NheI-STOP cassette was replaced with the wild-type genome by allelic exchange-generated K3R (right, lane 3).

BKRF3 was required for EBV genome replication.

To monitor the BKRF3 knockout effect on viral DNA replication, EBV genome copy numbers were measured in doxycycline-treated p2089, BKRF3STOP, and K3R cells at 72 h postinduction using quantitative real-time PCR. After doxycycline induction, immunoblotting was performed to confirm that BKRF3 was detectable in p2089 and K3R stable cells but not in BKRF3STOP cells (Fig. 7A). Simultaneously, induction of Rta successfully turned on viral lytic protein expression in p2089, BKRF3STOP, and K3R cells, including Zta, BMRF1, BGLF4, BALF5, and gp350/220 (Fig. 7A). The EBV genome copy numbers increased 17- and 28-fold in p2089 PL1 and PL2 cells, respectively, at 72 h postinduction, whereas EBV DNA copy numbers did not significantly increase at 72 h postinduction in any of the BKRF3STOP pool clones (Fig. 7B, lanes 1 to 10). In the BKRF3 revertant (K3R) clones, the EBV genome copy number increased 18-, 44-, and 23-fold in K3R PL5, PL6, and PL14 cells, respectively, at 72 h postinduction, indicating that BKRF3 contributes to viral DNA replication and that there are no additional mutations in BKRF3STOP (Fig. 7B, lanes 11 to 16). Notably, the expression of cellular UNG2 was significantly reduced at 72 h postinduction in p2089, BKRF3STOP, and K3R cells (Fig. 7A), very likely due to virus-mediated host shutoff effects. Thus, BKRF3 may compensate for cellular UNG2 to benefit EBV replication at this time point.

FIG 7.

EBV genome replication was deficient in cells containing BKRF3 knockout bacmids. (A) Wild-type p2089, BKRF3STOP, and K3R EBV bacmids were transfected into 293TetER cells, and stable clones were selected with hygromycin (50 μg/ml). EBV bacmid stable cells (p2089, BKRF3STOP, and K3R) were treated with doxycycline (50 ng/ml) to induce the EBV lytic cycle for 72 h, and viral lytic proteins were detected by immunoblotting. (B) The EBV genome copy number of wild-type p2089, BKRF3STOP, and K3R cells with or without doxycycline induction was determined by qPCR, using β-globin as an internal control, and results were compared to those for a 10-fold serial dilution of standard DNA as described in Materials and Methods.

The interaction of BKRF3 with viral proteins in replication complexes promoted EBV DNA replication through enzymatic activity-independent pathways.

In this study, we found that the sequence of aa 81 to 255 of BKRF3, which contains a catalytic domain and DNA binding domain, was crucial for association with other viral replicating components. To determine whether BKRF3 plays other roles in addition to providing the UDG activity in EBV DNA replication, trans-complementation assays were performed in BKRF3STOP cells with wild-type or mutant BKRF3 expression plasmids. To this end, two Flag-BKRF3 mutants were generated, one mutated at the catalytic site (Q90L,D91N) and the other at the leucine loop (H213L) (Fig. 8A) (37). As indicated by the immunoblot, viral lytic protein expression was comparable in cells complemented with plasmids expressing wild-type or mutant BKRF3 (Fig. 8B). In the immunoprecipitation-UDG assay, the UDG activity of BKRF3 was normalized to the relative level of immunoprecipitated HA-tagged BKRF3. The UDG activity of wild-type BKRF3 increased after lytic induction (Fig. 8C, lanes 5 and 6), whereas the enzymatically dead BKRF3(Q90L,D91N) failed to remove uracils (Fig. 8C, lanes 7 and 8). Interestingly, the leucine loop mutant BKRF3(H213L), which contains a wild-type catalytic domain, also was defective in UDG activity (Fig. 8C, lanes 9 and 10). According to the 3D structural analyses of BKRF3 and human UNG (37, 60), the leucine loop of UDG is within its DNA-contacting region (resides 213 to 229 of BKRF3). Therefore, single-stranded DNA cellulose chromatography was performed to determine whether the replacement of histidine 213 with leucine affects the general DNA binding ability of BKRF3. After binding of in vitro-transcribed/translated wild-type or mutant HA-BKRF3 to the DNA cellulose, proteins were eluted with step gradients of NaCl to indicate their relative DNA binding abilities (Fig. 8D). The peaks of both wild-type BKRF3 and BKRF3(H213L) were mostly detected in the 0.2 M NaCl fractions, indicating that the general DNA binding affinity of BKRF3(H213L) is very close to that of the wild type and that another mechanism is involved in the inability of BKRF3(H213L) to remove uracil base. On the other hand, BKRF3(Q90L,D91N) was eluted at a high level in the 0.3 M NaCl fraction (Fig. 8D). According to PyMOL 3D structure prediction, mutation sites of BKRF3(Q90L,D91N) are close to the β1 sheet of BKRF3 and may create a conformational change of BKRF3 (data not shown), leading to a slight increase in DNA binding affinity.

FIG 8.

Trans-complementation by wild-type or enzymatically dead BKRF3(Q90L,D91N), but not leucine loop mutant BKRF3(H213L), bacmids rescued lytic viral DNA replication in BKRF3STOP cells. (A) Sequence alignment of the catalytic domain and DNA binding domain of UDG within the herpesviruses, E. coli, and humans. Residues conserved within UNG (family 1) are marked in gray boxes. The domains of catalytic activity and DNA binding are indicated at the bottom, and the leucine loop involved in the DNA binding domain is indicated at the top. Secondary structures of EBV UDG are illustrated above the alignment. The mutation sites of enzymatically dead BKRF3(Q90L,D91N) and leucine loop mutant BKRF3(H213L) are indicated. (B) p2089 and BKRF3STOP cells were transfected with 7, 14, or 4 μg of HA-tagged wild-type BKRF3 (WT), BKRF3(Q90L,D91N), or BKRF3(H213L) plasmid, respectively, and treated with doxycycline (50 ng/ml) at 12 h posttransfection. After Doxycycline (Dox) treatment for 60 h, cells were harvested for Western blotting of viral lytic proteins. (C) The UDG activities of HA-tagged BKRF3 WT, BKRF3(Q90L,D91N), and BKRF3(H213L) in BKRF3STOP cells treated with doxycycline or left untreated. The transfected cell lysates were prepared for IP-UDG assays as described in Materials and Methods. E. coli UDG (10 U) was used as a positive control. The cleavage percentage of input ³²P-labeled U probe was normalized to relative amounts of immunoprecipitated BKRF3. (D) The DNA binding activities of in vitro-transcribed/translated [³⁵S]methionine-labeled BKRF3 WT, BKRF3(Q90L,D91N), and BKRF3(H213L) were measured in the binding buffer with ssDNA cellulose chromatography, eluted with binding buffer with step gradients of NaCl, and subjected to electrophoresis with 15% SDS-PAGE. The DNA-bound [³⁵S]methionine-labeled proteins in each eluent were measured by phosphorimaging and normalized to the total input. Data are representative of two independent experiments. (E) BKRF3STOP pool clones (PL1, PL5, and PL11) were transfected with control vector (vec), HA-BKRF3 WT, BKRF3(Q90L,D91N), or BKRF3(H213L). At 24 h posttransfection, doxycycline (Dox; 50 ng/ml) was added to induce lytic replication for 60 h. Subsequently, intracellular EBV DNA copy numbers of trans-complementation cells were quantified with qPCR for the EBV BamHI W fragment and human beta-globin (HBG) as described in Materials and Methods. Data are representative of three independent experiments.

Consequently, the EBV genome copy number was monitored to verify the biological function of BKRF3 in BKRF3STOP cells after complementation. Compared to those of vector control-transfected cells, EBV genome copy numbers increased 29-, 34-, and 33-fold at 60 h postinduction in BKRF3STOP PL1, PL5, and PL11 cells, respectively, with the complementation of wild-type BKRF3, indicating that BKRF3 is crucial for EBV replication (Fig. 8E, lanes 3, 8, and 13). Notably, complementation of enzymatically dead BKRF3(Q90L,D91N) also increased viral DNA copies by 22-, 31-, and 18-fold in doxycycline-treated BKRF3STOP PL1, PL5, and PL11 cells (Fig. 8E, lanes 4, 9, and 14), suggesting that BKRF3 plays an essential role in the DNA replication process in addition to its UDG activity. In cells complemented with the leucine loop mutant BKRF3(H213L), the level of viral DNA increased 5-, 10-, and 6-fold in doxycycline-treated BKRF3STOP PL1, PL5, and PL11 cells, respectively (Fig. 8E, lanes 5, 10, and 15). Leucine loops of gammaherpesviral UDGs are seven amino acid longer than those of alpha- and betaherpesviruses (Fig. 8A). We found that the mutation of histidine 213 in BKRF3 attenuated its ability to rescue viral DNA replication in BKRF3STOP cells, indicating that the leucine loop of BKRF3 is important for viral DNA replication. Thus, data here suggest that the proper overall structure of BKRF3 contributes more to EBV genome replication in our current system than does enzyme activity.

To determine whether the recruitment of BKRF3 to the viral replication compartment was affected in BKRF3 mutants, immunofluorescence staining was performed to visualize the colocalization of BKRF3 with the DNA polymerase BALF5 in NA cells (Fig. 9A). With the transfection of Rta expression plasmid, the colocalization intensity of BKRF3(H213L) with BALF5 was weaker than that of WT BKRF3 in reactivated NA cells in confocal analysis (Fig. 9A). Although BKRF3(H213L) displayed mainly intranuclear distribution, reduced colocalization with BALF5 was observed in about 80% of NA cells with lytic cycle progression in the immunofluorescence assay (data not shown). Consistent with this finding, the amounts of HA-BKRF3(H213L)-associated Rta and BALF5 were reduced (Fig. 9B, lanes 7 and 8). BKRF3(H213L) still interacted with BMRF1 for its nuclear localization (Fig. 9C); however, its targeting to DNA replication complexes was attenuated. Interestingly, the amount of BALF5 DNA polymerase being coimmunoprecipitated with HA-BKRF3(Q90L,D91N) was about 2-fold higher than that of WT BKRF3 (Fig. 9B, lanes 3 to 6). In the middle rows of Fig. 9A, more distinct colocalization signals of HA-BKRF3(Q90L,D91N) and BALF5 also were observed. Because both Rta and BALF5 were mapped to interact with the C terminus of BKRF3 (Fig. 4), it is possible that mutation of H213L affects the ability of BKRF3 to interact with these proteins. Overall, data here suggest that the interaction of BKRF3 with viral proteins in replication compartments correlated with its ability to support viral DNA replication.

FIG 9.

Point mutation at His213 of BKRF3 attenuated its recruitment to the viral DNA replication compartment and its ability to interact with the viral DNA polymerase BALF5 and immediate-early gene, Rta. (A) NA cells were seeded at 3 × 10⁶ cells per 10-cm dish for slide culture and transfected with Flag-BKRF3 or control plasmid combined with vector or Rta expression plasmid. At 48 h posttransfection, cells were fixed with 4% paraformaldehyde and stained for BKRF3, BALF5, and DNA with mouse anti-Flag antibody, rabbit BALF5 antiserum, and Hoechst 33258, followed by confocal microscopy analysis. (B) To measure the ability of wild-type or mutant BKRF3 to interact with viral DNA replication-associated proteins, NA cells transfected with 7, 14, or 4 μg of HA-BKRF3 WT, BKRF3(Q90L,D91N), or BKRF3(H213L) were induced into the lytic cycle by Rta transfection. At 60 h posttransfection, cell lysates were harvested for coimmunoprecipitation assay. The HA-BKRF3-associated complexes were precipitated with anti-HA and detected with anti-Rta, anti-BALF5, and anti-HA antibodies by Western blotting. This is a representative result of triplicate experiments. (C) To verify the interacting abilities of BMRF1 with BKRF3 WT, BKRF3(Q90L,D91N), and BKRF3(H213L), 293T cells were cotransfected with expression plasmids as indicated for 48 h. Whole-cell lysates were subjected to a coimmunoprecipitation assay. The HA-BMRF1-associated protein complexes were precipitated with anti-HA and detected with anti-Flag by Western blotting. This is a representative result of duplicate experiments. *, Ig heavy chain; **, Ig light chain.

DISCUSSION

Herpesviral DNA replication-associated enzymes have been considered good antiviral targets. The intranuclear viral DNA replication compartments contain multiple viral and cellular DNA replication and repair enzymes and chromatin modifiers that can coordinately function together for efficient virus replication. In this study, we revealed how BKRF3 is recruited into the nucleus by viral DNA polymerase processivity factor BMRF1. We also showed the regulation of the enzyme activity of BKRF3 and its interaction with viral DNA replication protein complexes and viral DNA replication. The inability of the BKRF3STOP bacmid to replicate following lytic induction indicates that BKRF3 is crucial for viral DNA replication. The distinct outcomes of complementation with BKRF3(Q90L,D91N) and BKRF3(H213L) suggest the recruitment function of BKRF3 via its C-terminal leucine loop is important for EBV replication.

Here, we observed that BKRF3 was translocated from the cytoplasm into the nucleus upon lytic induction and colocalized with the DNA polymerase BALF5 in both NPC NA cells and 293-derived EREV8 cells (Fig. 1A to C). We also found, upon the induction of lytic replication, that the expression levels of UNG2 decreased along with increasing expression of viral lytic replication proteins (Fig. 1D and E). The exact mechanism leading to UNG decrease during virus replication is not clear. However, the expression of UNG2 was reported to peak in and throughout S phase and then decline to undetectable levels until the next S phase through the ubiquitin-proteasome pathway (61). Taking into account that EBV replication causes cell cycle arrest at the G₁/S transition without cellular DNA replication (62, 63), we suspect the decrease of UNG2 is caused by downregulated transcription or increased protein degradation.

Unlike cellular UNG2, which contains a unique NLS as well as subnuclear targeting signals (16), we found that the interaction of BKRF3 with viral DNA polymerase processivity factor BMRF1 promoted its nuclear transport (Fig. 3). The nuclear targeting of BKRF3 is similar to that of HCMV UL114, which associates with UL44, the homolog of BMRF1 (31), suggesting interactions with nuclear replicating proteins are required for herpesviral UDGs to participate in the viral replication compartment. The nuclear targeting of BALF5 was recently reported to be dependent on BMRF1 and cellular chaperone protein Hsp90 (59). It is not clear whether other cellular proteins are required for BMRF1-dependent nuclear targeting of BKRF3.

Interaction of cellular UNG2 with PCNA and RPA replicating complex ensures postreplicative repair of misincorporated uracil (U·A) in DNA (64). Similarly, the association between UDGs and viral DNA polymerases to function in replication-coupled base excision repair has been described for HSV-1 and HCMV (29, 32). HCMV UL44 facilitates loading of UL114 onto DNA and promotes UL114 to remove uracil from DNA (33). In our study, BKRF3 interacted with the immediate-early transactivator Rta, DNA polymerase BALF5, and DNA polymerase processivity factor BMRF1 in coimmunoprecipitation and GST pulldown assays (Fig. 2 to 4). We suggest the interaction with BALF5 and BMRF1 helps BKRF3 to form complexes with other replication proteins to enhance UDG activity of BKRF3 and carry out repairs at DNA replication loci. Because Rta was found to promote the nuclear targeting of the DNA primase-associated factor BBLF2/3, which is the linker for proteins targeting viral DNA replication origin (65), it is possible that Rta also helps the stabilization of BKRF3 to the DNA replication complex through network interactions with other viral replication proteins. Furthermore, the UDG activity of HA-BKRF3 was stimulated when it formed complexes with other viral DNA replication proteins (Fig. 5). In this way, viral UDG activity may be regulated to function along with viral DNA replication, avoiding nonspecific DNA damages.

A search through the literature found that the UDG homologs are dispensable in alphaherpesviruses for infection of actively growing fibroblasts but are essential for infection and reactivation in the murine nervous system (29, 30). The UDG of HCMV, UL114, is required for full virus replication in a culture system. Viral replication of the HCMV UL114 mutant is retarded in quiescent fibroblasts but proceeded smoothly in actively growing fibroblasts (35). Additionally, using a random signature-tagged mutagenesis approach to select the viral genes essential for the murine gammaherpesvirus 68 life cycle, Song et al. found viral UDG ORF46 is not essential for viral replication in a mouse fibroblast cell line (66). These observations suggest that cellular UDG activity is able to compensate for viral UDG deficiency to a certain extent. Previously, we found viral DNA replication was decreased about 10 to 20% at 48 h postinduction when BKRF3 was knocked down by siRNA (4). It is possible that residual BKRF3 was sufficient to support EBV DNA replication complex formation in that setting. In this study, we demonstrated cellular UNG2 was undetectable at 60 h post-EBV reactivation, while expression of BKRF3 gradually increased and the total UDG activity was sustained in NA cells (Fig. 1E). Thus, the BKRF3 UDG activity may function in cells with low levels of UNG2 expression.

Notably, the presence of BKRF3 in the viral DNA replication compartment is correlated with its function in DNA replication. In bacmid experiments, we found the EBV DNA copy number significantly increased about 20- to 40-fold following the induction of Rta by doxycycline in p2089 and in K3R revertant cells, but not in BKRF3STOP cells, at 72 h postinduction (Fig. 7). In the sequence alignment, the extended leucine loop at the C-terminal region is conserved among gammaherpesviruses (Fig. 8A), suggesting functions in facilitating viral DNA replication. BKRF3STOP virus, which was inserted with a stop codon cassette at aa 73 of BKRF3, did not replicate after lytic induction. However, MHV-68 ORF46 mutant virus, which was inserted with a transposon at aa 103 of the viral UDG, still produced infectious viruses (66). We suspect that BKRF3 plays a unique role in EBV DNA replication or the residual fragment of truncated UDG in MHV-68 still facilitates viral DNA replication. Nevertheless, expression levels of UNG2 in virus-replicating cells may affect the outcome of viral UDG defective viruses. Thus, it would be interesting to address whether UNG2 also participates in the viral DNA replication compartment and tethers other cellular DNA replication factors to enhance viral DNA replication.

Complementary expression of both the wild type and two BKRF3 mutants in BKRF3STOP cells further demonstrated the contribution of different functional domains of BKRF3 in viral DNA replication. Previously, we showed that blocking cellular and viral UDG activity by the inhibitor Ugi decreases 50% of EBV DNA replication. Here, the fact that enzymatically dead mutant BKRF3(Q90L,D91N) restored EBV genome replication in BKRF3STOP cells at a slightly lower level than WT BKRF3 strongly suggests that BKRF3 has a function other than its enzymatic activity for EBV genome replication (Fig. 8E). With nuclear distribution upon lytic induction, BKRF3(H213L) interacts with proteins in the viral DNA replication complex less efficiently, as revealed by immunofluorescence and coimmunoprecipitation experiments (Fig. 9). Data here hint that precise associations with other components in the complex affect the enzymatic activities and function(s) of BKRF3. It is possible that BKRF3 promotes a conformational change of BALF5/BMRF1 complexes to enhance DNA synthesis. Such interactions also ensure viral UDG is recruited to viral DNA replication forks before DNA synthesis.

Regulation through specific interaction with DNA binding proteins were founded in cellular and viral UDGs. For example, UNG2 directly interacts with XRCC1 to enhance the repair efficiency of the XRCC1-mediated repair pathway (64, 67). Likewise, the HCMV UDG, UL114, was found to interact with the chromatin remodeling factor SMARCB1 and to participate in the recruitment of the chromatin remodeling complex onto replication foci (33). Although no BKRF3-interacting cellular proteins were identified in two previous yeast-two hybrid screening studies (68, 69), it is still possible that BKRF3 also recruits the cellular machinery to support viral DNA replication.