HEK293 (Delecluse et al., 1998) and HeLa (Kawashima et al., 2013) cells were maintained in Dulbecco’s modified Eagle medium (DMEM, Gibco BRL) supplemented with 10% fetal bovine serum (FBS). HEK293 cells infected with EBV BRRF2 knockout virus (Watanabe et al., 2015b) were cultured with DMEM containing 150 μg/ml hygromycin B. Akata EBV negative cells (Akata(-)), B95-8, and AGS-EBV cells (Katsumura et al., 2009) were cultured in RPMI 1640 medium containing 10% FBS.

Rabbit antibodies against BZLF1 and BALF2 were used as reported previously (Murata et al., 2009; Sugimoto et al., 2013). Mouse anti-FLAG antibody was purchased from Sigma-Aldrich, and rabbit anti-Giantin antibody was from Abcam. Rabbit Anti-tubulin, -Calnexin, and -Rab5 antibodies were purchased from Cell Signaling Technology. Mouse anti-BMRF1 antibody was purchased from Novocastra. Horseradish peroxidase-linked goat antibodies to mouse or rabbit IgG were from Amersham Biosciences.

Two expression vectors, pcDNABZLF1 and pcDNAFLAGBRRF2, were reported previously (Watanabe et al., 2015b). All mutant vectors with deletions or substitutions in BRRF2 were generated by inverse PCR using Flag-tagged wild type BRRF2 expression plasmid as a template. To generate the deletion mutants shown in Figures 1A and 2A, we used the following primers: for Δ1-96: 5′-TGTCTGGGTGTTATTCCAGG-3′ (forward), and 5′-CATGAATTCCTTGTCATCGT-3′ (reverse), for Δ97–180: 5′-CATGTGCAGCTTGCATACGA-3′ (forward), and 5′-TAAGTTGTGGGCCGTGGTTA-3′ (reverse), for Δ181-272: 5′-TTTTATCTCCCGACGACAGG-3′ (forward), and 5′-ACCACGTAAAGCCACAAGCT-3′ (reverse), for Δ273-341: 5′-TTAACGAGCAGAGGGAATGA-3′ (forward), and 5′-GTGATTATCCGAATTTCGGA-3′ (reverse), for Δ342-441: 5′-GACCTTGAAGAGGAAGAGGA-3′ (forward), and 5′-CTGAGAGCATGTCTGCTGGG-3′ (reverse), for Δ442-537: 5′-TAACTCGAGCATGCATCTAG-3′ (forward), and 5′-ACCAGACAGAGATGCTACTG-3′ (reverse). The C-terminal deletion mutants were generated using the following primers: for Δ442-470: 5′-GAGGCTCAGGACGCAAATCT-3′ (forward), and 5′- ACCAGACAGAGATGCTACTG-3′ (reverse), for Δ471-498: 5′-GAGGATGGGGAATTTTCAGA-3′ (forward), and 5′-AAGCCAGGCCTCAAATTCCT-3′ (reverse), for Δ499-537: 5′-GACCATGAAGGGGATGAGG-3′ (forward), and 5′-CGAACCATCCTCGTCCTCAT-3’ (reverse), for Δ512-537: 5′-TAACTCGAGCATGCATCTAG-3′ (forward), and 5′-GCTGTCAGACAGATCCA-3′ (reverse). The mutants harboring the serine-to-alanine substitution were prepared in the same fashion using the following primers: for S498A: 5′-GCGGAGGATGGGGAATTTTC-3′ (forward), and 5′- ACCATCCTCGTCCTCATCAC-3′ (reverse), for S504A: 5′- GCAGACCTGGATCTGTCTGA-3′ (forward), and 5′- AAATTCCCCATCCTCCGAAC-3′ (reverse), for S509A: 5′-GACAGCGACCATGAAGGGGA-3′ (forward), and 5′- AGCCAGATCCAGGTCTGAAA-3′ (reverse), for S511A: 5′- GCCGACCATGAAGGGGATGA-3′ (forward), and 5′- GTCAGACAGATCCAGGTCTG-3′ (reverse).

To induce lytic replication in HEK293 cells carrying EBV BAC clones, cells were transfected with pcDNABZLF1 using the Neon transfection system (Invitrogen). At 72 h after transfection, cells and culture medium were collected, followed by freezing and thawing. After centrifugation at 13,000 rpm, the supernatants were mixed to Akata(-) cells for 3 h at room temperature with rotation, and the medium was replaced with fresh medium. At 48 h post-infection, cells were fixed with 1% formaldehyde, washed with PBS, and suspended in PBS. GFP-positive cells were counted using the FACSCalibur G5 system (Becton-Dickinson), according to the manufacturer’s instructions.

HeLa cells were transfected with the indicated expression vectors using lipofectamine 2000 reagent (Invitrogen) and fixed with 70% ethanol. Cells were permeabilized using 0.1% Triton-X100, blocked using 10% normal goat serum (Funakoshi), and stained with anti-FLAG mouse antibody (Sigma), followed by washing and treatment with secondary antibody (Alexa 555 anti-mouse IgG, Invitrogen). AGS-EBV cells were transfected with BZLF1 and FLAG-BRRF2 using lipofectamine 2000 reagent. The cells were fixed with 70% ethanol on glass slides, permeabilized, and blocked. After staining with anti-FLAG mouse antibody and anti-BALF2 rabbit antibody, cells were reacted with secondary antibodies (Alexa Fluor 555 anti-mouse and anti-rabbit IgG, Invitrogen). B95-8 cells were transfected using the Neon transfection system (Invitrogen) and stained with mouse anti-BMRF1 antibody directly labeled with Alexa Fluor 488 using Zenon Alexa IgG labeling reagent (Invitrogen). After washing with PBS, cells were mounted using ProLong Diamond Antifade Mountant with DAPI (Invitrogen). Confocal microscopy (LSM880, Zeiss) was used to analyze the samples.

For conventional SDS-PAGE, HEK293 cells carrying each BAC clone were transfected with the BZLF1 and BRRF2 mutant vectors using the Neon transfection system to induce lytic infection. After 2 days, cells were washed with PBS and lysed using sample buffer for SDS-PAGE. Electrophoresis, transfer, and staining were carried out as described previously (Narita et al., 2015).

HEK293 cells latently infected with BRRF2-knockout EBV were transfected with pcDNABZLF1 together with expression vector for wild-type or mutant FLAG-tagged BRRF2. After one day, cells were lysed in phosphatase buffer (Santa Cruz). Lysed samples were reacted with or without Lambda phosphatase (Santa Cruz). Protein samples were separated on a 7.5% Phos-tag gel (Wako chemicals). After electrophoresis, the Phos-tag gel was washed with transfer buffer containing EDTA to remove zinc ions from the gel, according to the manufacturer’s instructions, followed by immunoblotting as specified above.

To identify the functional domain(s) of BRRF2 involved in virus production, we determined whether BRRF2 mutants lacking various amino acid regions could complement the decreased viral titer caused by disruption of the BRRF2 gene. We prepared BRRF2 mutant vectors, each lacking ∼90 amino acids in various regions, as depicted in Figure 1A. These FLAG-tagged deletion mutant vectors were transfected into HEK293 cells latently infected with the BRRF2-knockout EBV BAC clone established previously (Watanabe et al., 2015b). Wild-type BRRF2 increased the virus yield of the BRRF2-knockout virus by two- or three-fold (Figure 1B) (Watanabe et al., 2015b). Likewise, mutants lacking the central regions of BRRF2 (Δ181–272, Δ273–341, and Δ342–441) successfully restored the impaired viral titer (Figure 1B). However, BRRF2 mutants lacking residues 1–180 at the N-terminus (Δ1–96 and Δ97–180) and 442–537 at the C-terminus (Δ442–537) failed to restore the reduced virus yield (Figure 1B). This result suggested that EBV BRRF2 has two important domains in both the N- and C-terminal regions (residues 1–180 and 442–537) required for efficient production of virus progeny.

We demonstrated previously that BRRF2 is localized predominantly in the cytoplasm of infected and transfected cells (Watanabe et al., 2015b). To investigate whether BRRF2 deletions alter the intracellular localization of the protein, we used immunofluorescence assays to observe the distribution of the mutant proteins. HeLa cells were transfected with vectors harboring the deletion mutants (shown in Figure 1A), stained with anti-FLAG antibody, and observed by confocal microscopy. Although BRRF2 lacking the N-terminal or central regions was localized in the cytoplasm in a manner similar to that of wild-type BRRF2, C-terminally truncated BRRF2 (Δ442–537) was found not only in the cytoplasm but also in the nucleus (Figure 1C). This result indicates that the C-terminal region of BRRF2 is required for its predominant localization in the cytoplasm, at least when produced in the absence of other viral proteins.

To analyze BRRF2 localization in the context of infection, we stained B95-8 and AGS-EBV cells after lytic induction (Figure 1D). The BMRF1 protein was also stained to monitor the lytic induction. Regardless of the cell type or presence of truncated BRRF2, BMRF1 protein was clearly localized in the nucleus as expected (Figure 1D). Whereas wild-type BRRF2 was detected mainly in the cytoplasm, the C-terminally truncated mutant (Δ442–537) was distributed in the nucleus and cytoplasm (Figure 1D). This result indicated that the BRRF2 C-terminal region is important for its predominant localization in the cytoplasm, even in the context of infection.

Interestingly, residues 442–537 contain three stretches of acidic amino acid clusters, as shown in Figure 1A and 2A. As several herpesvirus gene products, such as EBV BBLF1 (Chiu et al., 2012), retain acidic amino acid clusters crucial for localization in the cytoplasm (including the Golgi apparatus), we speculated that these acidic residues in BRRF2 might also play a role in cytoplasmic localization.

To narrow down the C-terminal residues of BRRF2 important for efficient progeny production and cytoplasmic distribution, we further constructed four additional deletion mutant vectors between residues 442 and 537 (Δ442–470, Δ471–498, Δ499–537, and Δ512–537), as shown in Figure 2A. Wild-type BRRF2 successfully augmented the viral titer, but the Δ442–537 mutant failed to do so (Figure 2B). However, all four deletion mutants of BRRF2 increased the yield almost as efficiently as did wild-type BRRF2 (Figure 2B). In addition, all four mutant proteins were localized in the cytoplasm, similar to the wild-type protein (Figure 2C). These results suggest the presence of redundant acidic amino acids in the BRRF2 protein.

Previous reports have shown that the tegument protein BRRF2 is phosphorylated in infected cells (Johannsen et al., 2004; Watanabe et al., 2015b). To identify the phosphorylation site of BRRF2, we used the phos-tag electrophoresis system, in which phosphate residues are bound by phos-tag molecules, thus retarding the mobility of phosphoproteins during electrophoresis (Kinoshita-Kikuta et al., 2007). Phosphoproteins treated with phosphatase migrate through the gel faster than those without phosphatase treatment.

We first analyzed the phosphorylation state of BRRF2 mutants lacking various amino acid regions (see Figure 1A) with or without phosphatase treatment (Figure 3A). HEK293 cells expressing EBV BRRF2 lacking residues 442–537 migrated at the same speed in the phos-tag gel regardless of phosphatase treatment (Figure 3A, Δ442–537). In contrast, wild-type or other deletion mutant BRRF2 proteins (Δ1–96, Δ97–180, Δ181–272, Δ273–341, and Δ342–441) migrated faster when de-phosphorylated (Figure 3A). Thus, residues 442–537 contain phosphorylated amino acid.

Next, we examined the deletion mutants (Δ442–470, Δ471–498, Δ499–537, and Δ512–537) to determine which specific region is phosphorylated (see Figure 2A). BRRF2 Δ512–537 migrated faster when treated with phosphatase, whereas the mobility of BRRF2 Δ499–537 did not change in the presence or absence of phosphatase (Figure 3B), indicating that phosphorylated amino acid is between residues 499 and 511.

We thus created four mutants with alanine substitutions at each of the four serine residues in this region as shown in Figure 3C and evaluated whether these mutants were phosphorylated (Figure 3D). The results showed that BRRF2 S511A protein migrated similarly regardless of phosphatase treatment (Figure 3D), indicating that serine 511 undergoes phosphorylation.

We then subjected the BRRF2 S511A mutant to a complementation assay and found that this mutant restored the production of progeny as efficiently as did wild-type BRRF2 (Figure 3E left panels). In addition, BRRF2 S511A protein showed a similar localization pattern as that observed with wild-type BRRF2 (Figure 3E right panels). These results suggested that BRRF2 serine 511 is phosphorylated during lytic infection; however, this phosphorylation is not essential for efficient production of viral progeny.

To clarify the localization of BRRF2 in the cytoplasm, we transfected HeLa cells with wild-type BRRF2 and co-stained with markers of cytoplasmic organelles (Figure 4). The pattern of Giantin, which is known to localize in the Golgi apparatus, was different from that of BRRF2 although a part of BRRF2 protein might colocalize (Figure 4 upper panels). The Calnexin resides in endoplasmic reticulum (ER). It might colocalize with BRRF2 protein, albeit only partly (Figure 4 middle panels). Interestingly, localization of BRRF2 matched at least partially with Rab5, which is a marker of endosomes, especially of early endosome (Figure 4 lower panels). Therefore, unlike BBLF1 (Chiu et al., 2012), BRRF2 likely localizes more efficiently to endosomes, than Golgi apparatus and ER.