Nuclear Import of Epstein-Barr virus BLLF2 is Mediated by an Importin β1-Dependent Mechanism

Background: Epstein-Barr virus (EBV), the pathogen of several human malignancies, encodes many proteins that require to be transported into the nucleus for viral DNA reproduction and nucleocapsids assembly in the lytic replication cycle. A nuclear membrane phosphoprotein encoded by EBV BLLF2, is believed to associate with viral DNA packaging and primary egress across the nuclear membrane. Results: Here, uorescence microscope, mutation analysis, interspecies heterokaryon assays, co-immunoprecipitation assays and western blot were performed to explore the nuclear import mechanism of BLLF2. As results, BLLF2 was shown to be a nucleocytoplasmic shuttling protein, which was mediated neither by chromosomal region maintenance 1 (CRM1)- nor transporter associated with antigen processing (TAP)-dependent pathway. Yet, two functional nuclear localization signals (NLSs) of BLLF2, NLS1 ( 16 KRQALETVPHPQNRGR 31 ) and NLS2 ( 48 PPVAKRRR 58 ), were identied, whereas the predicted NES was nonfunctional. Finally, BLLF2 was proved to transport into the nucleus via Ran-dependent and importin β1-dependent pathway. Conclusions: This mechanism may contribute to a more extensive insight of the assembly and synthesis of EB virions in the nucleus, thus affording a new direction for the treatment of viruses.

molecule is accomplished by various cellular transporter exportins and importins, by distinguishing favourable motifs on the target proteins named nuclear localization signal (NLS) and nuclear export signal (NES) [4].
Nuclear tra cking of a de nitive protein is often achieved by the canonical importin α/β-reliant nuclear translocation pathway. As an adaptor protein, importin α binds to target protein encompassing with NLS and then heterodimerizes with importin β, to assemble the heterotrimeric importin α/β/NLS-cargo complex that penetrates through NPC and delivers NLS-cargo into the nucleus [5,6]. Additionally, another crucial component of the nuclear import pathway, Ras-related nuclear protein (Ran), is an eukaryotic evolutionarily preserved small GTPase. Both effective export and import are regulated by a gradient of Ran in GTP-and GDP-combined states between the cytoplasm and the nucleus. After attaching with RanGTP, the nuclear transport receptors binding cargo can be tra cked from the cytoplasm to the nucleus, while export receptors can release goods via binding to RanGTP and give off them from the nucleus into the cytoplasm after GTP hydrolysis [5].
Nuclear export of proteins is largely ful lled by a leucine-rich NES, which is bound by the major nuclear export receptor chromosomal region maintenance 1 (CRM1) (exportin 1, XPO1) that belongs to the the karyopherin-β family. Ttransporter associated with antigen processing (TAP/NXF1) is also a major export receptor, both of them export mRNA from the nucleus to cytoplasm in metazoan cells have been profoundly investigated. Despite TAP is not a strong RNA-binding protein, it mainly binds to the Aly/REF mRNA adaptor protein, an element of the messenger ribonucleoprotein particles (mRNPs), which is carried out of the nucleus via direct combination with nucleoporins imbedding into the nuclear pore [7]. TAP/NXF1 also can improve the nuclear export of some proteins [8][9][10], whereas CRM1 can export hundreds of cargo proteins out of the nucleus, by combining to their classical leucine-rich NESs [11]. The CRM1-relied export pathway is widely utilized to export proteins and non-coding RNAs, including ribosomal RNAs (rRNAs) and small nuclear RNAs (snRNAs), while only a minority of cellular mRNAs employ this pathway [11,12]. In addition to CRM1 and TAP, other exportins are also involved in the nuclear export process. Exportin 4 is in charge of the nuclear exports of eukaryotic mothers against decapentaplegic homolog 3 (Smad3) and translation initiation factor 5A (eIF5A) [13]. Exportin 5 can mediate the export of dsRNA and precursor microRNA, while exportin-t can help the export of tRNA [14][15][16][17]. Importantly, exportin cellular apoptosis susceptibility protein (CAS) can assist the export of importin α for a new cycle of protein nuclear translocation [18]. Besides, exportin 7 is a nuclear export mediator with broad substrate speci city [19].
It's well documented that the nuclear accumulation of herpesviral proteins is signi cant for virus propagation, assembly and dissemination, whereas the nuclear transport mechanisms of the majority of virus-encoded proteins are less well explored. As a transcriptional co-activator of EBV nuclear antigen 2 (EBNA2), EBNA-LP interplays with importin α1 through functional NLS, to facilitate its e cient nuclear localization for its interaction with EBNA2 [20]. The N-terminus functional NES of EBV early protein EB2 (also designated BMLF1, SM or Mta) can advance the nucleocytoplasmic export of a number of early and late viral mRNAs (come from intronless genes) by precisely binding to TAP/NXF1, which is indispensable for the proliferation of infectious virions [10,21]. EBNA1 can interact with importin α1 and importin α5 via its NLS, to accelerate its nuclear import [6], which may be fundamental for the sustainment, proliferation and transcription of the EBV-positive tumor cells. BFLF2 is exhibited to be correlated with TAP for nuclear export, and interplayed with importin α7, importin β1 and transportin-1 for its nuclear accumulation, which may be meaningful for the e cient viral DNA packaging and primary release across the nuclear membrane. Furthermore, BGLF4 protein kinase can promote the nuclear accumulation of a few non-NLScontaining EBV proteins, including major capsid protein (VCA) and the viral DNA replication enzymes BBLF2/3, BBLF4 and BSLF1 [22]. However, the functional correlation of nucleocytoplasmic shuttling of most of the EBV proteins require to be probed.
EBV tegument protein BLLF2 is a nuclear membrane phosphoprotein, which may exert some roles in viral reduplication or virions assembling. Our previous study manifested that BLLF2 absolutely localizes in the nucleus [23], nonetheless the de nite mechanism for its subcellular localization was not well known. In our preliminary experiment, we established that BLLF2 could shuttle between the cytoplasm and nucleus, which propels us to investigate the nucleocytoplasmic transport mechanism of BLLF2.

Nucleocytoplasmic shuttling of BLLF2
Our previous study revealed that BLLF2 is de nitely located in the nucleus [23]. To dissect the subcellular transport mechanism of BLLF2, bioinformatics analysis was initially implemented and showed that BLLF2 possesses four predicted potential NLS motifs (pat4 44 RRPR 47 and 52 KRRR 55 , and pat7 48 PPVAKRR 54 and 49 PVAKRRR 55 ) and two supposed NES, that is NES1 ( 83 VQSPPQITAVIQL 95 ) and NES2 ( 102 MRPPIYL 108 ) (Fig. 1A). Subsequently, BLLF2 was again con rmed completely located in the nucleus in COS-7 cells transfected with EYFP-BLLF2 plasmid (Fig. 1B). In order to exclude the in uence of the big tag EYFP on the subcellular localization of BLLF2, BLLF2-EYFP and BLLF2-Myc expression plasmids were also constructed, and uorescence microscope demonstrated that the subcellular localization patterns of BLLF2-EYFP and BLLF2-Myc were similar with that of EYFP-BLLF2 (Fig. 1C). Besides, coexpression of nucleolar marker pNucleolin-EGFP with pEYFP-BLLF2 showed that BLLF2 also located in the nucleolus (Fig. 1D).
A nuclear localization protein equipped with functional NES and NLS can be theoretically transported from a donor to a recipient nucleus [24]. With regard to the heterokaryon assays, the shuttling characteristic of a certain protein can be established through the species-speci c DAPI staining fashions of the nucleus under uorescence microscope, with speckles in the nucleus of mouse NIH3T3 cells (Fig.  1E) [8,25]. Since BLLF2 holds predicted NES and NLS, we wondered if BLLF2 can shuttle between the cytoplasm and nucleus. As results (Fig. 1F), COS-7 cells expressing EYFP-BLLF2 were fused with a considerable number of NIH3T3 cells in the presence of protein synthesis inhibitor cycloheximide. After fusing, BLLF2 could also be detected with a distinguishing speckled fashion in the nucleus of NIH3T3 cells. As the nucleocytoplasmic shuttling positive control [25], EYFP-BFLF2 could shuttle between COS-7 cells and NIH3T3 cells, suggesting BLLF2 is a genuine nucleocytoplasmic shuttling protein.

Identi cation of the functional NES in BLLF2
On the basis of the predicted motifs of NESs and NLSs, full-length of BLLF2 was rstly cut into two fragments (amino acids (aa)1-58 and aa59-148) and then fused to the C-terminus of EYFP ( Fig. 2A), to detect their subcellular localizations. As shown in Fig. 2B, aa1-58 was perfectly located in the nucleus and nucleolus, while aa59-148 displayed disperse dissemination thoughout the cytoplasm and nucleus. These results disclosed that aa1-58 contains functional NLS, and aa59-148 may not contain functional NES.
These data con rmed that 44 RRPR 47 has nucleolus localization function.
Nuclear import mechanism of BLLF2 Ran (Ras-associated nuclear protein), a small GTPase belonging to the RAS superfamily, is specialized and crucial for the nuclear accumulation of proteins with a canonical NLS [26]. Here, dominant negative (DN) RanGTP containing Q69 mutation (Ran-Q69L), without the competence of GTP hydrolysis [8], was applied to dissect whether Ran is vital for the nuclear tra cking of BLLF2. Compared to the cells cotransfected with mCherry vector and pEYFP-BLLF2, the nuclear import of BLLF2 was undoubtedly con ned in cells co-expressing Ran-Q69L-mCherry and EYFP-BLLF2 (Fig. 4B), suggesting the nuclear translocation of BLLF2 is Ran-relied and requires Ran GTP hydrolysis.
It's universally established that the importin α/β heterodimer can discern canonical NLS and facilitate the nucleocytoplasmic transport of particular target proteins [27]. In order to resolve which receptor takes part in the nuclear translocation of BLLF2, DN importin α5 (kα1, awed in the binding to importin β [28]), as well as DN importin β1 (kβ1, faulty in associating with Ran [28]), were exploited. Moreover, the nuclear import inhibitors of M9M (speci cally impedes importin β2 to attach to NLS) and Bimax2 (suppresses the functions of importin α1, α3 and α7 [8,25]) were also applied. Compared to the negative controls of DNs or competitive inhibitors (Fig. 4A), BLLF2 was relocalized to the cytoplasm by DN kβ1, but not DN kα1, M9M, Bimax2 or mCherry (Fig. 4B), suggesting BLLF2 may be transported into the nucleus via Ran-and importin β1-dependent pathway.
Nuclear export mechanism of BLLF2 CRM1 (Exportin1/XPO1), a member of the importin β family, mediates the nuclear export of proteins by combining to their classical NESs [11]. Therefore, we go on to assess whether the nuclear export of BLLF2 can be blocked by CRM1 speci c inhibitor leptomycin B (LMB) [8,25]. As CRM1-dependent positive control [29], the nuclear accumulation of UL4 was inhibited by LMB treatment (Fig. 5A), whereas the CRM1-independent negative control of EYFP vector was incapable to achieve the nucleocytoplasmic shuttling with or without the presence of LMB (Fig. 5B). Upon LMB treatment, EYFP-BLLF2 was also transported from monkey nuclei to mouse nuclei (Fig. 5C), indicating BLLF2 has the capability to shuttle between varied cells, and the nuclear export of BLLF2 may take place independently on CRM1.
It is reported that CRM1 overexpression can advance the nuclear export of CRM1-dependent proteins [25]. As results, the nuclear export of the CRM1-dependent positive control HSV-1 UL4 [29] was promoted by co-expression of CRM1-mCherry, yet BLLF2 remain thoroughly located in the nucleus and nucleolus in the existence of CRM1 when COS-7 cells were co-transfected with pCRM1-mCherry and pEYFP-BLLF2 ( Fig. 5D), certifying CRM1 is not fundamental for the nuclear export of BLLF2.
Besides CRM1, TAP (NXF1), the critical mRNA export receptor, is also related to the nuclear export of distinct proteins [30], which also can expedite the nuclear export of TAP-relied protein when it is overexpressed with the target protein [25]. As shown in Fig. 5E, the nuclear export of the TAP-dependent positive control BFLF2 [25] was obviously boosted by co-expression of TAP-mCherry, whereas TAP could not transport BLLF2 from the nucleus to cytoplasm when COS-7 cells were co-transfected with EYFP-BLLF2 and TAP-mCherry expression plasmids, testifying the nuclear export of BLLF2 was also independent on TAP.

Discussion
Subcellular distribution of nuclear protein was rstly disclosed to shuttle back and forth between the cytoplasm and nucleus by transplantation experiment in Amoeba proteus [32], and a increasing number of proteins are proved to have the ability of nucleocytoplasmic shuttling [33]. Nonetheless, the nucleocytoplasmic shuttling feature of protein from higher eukaryotic cells is shown based on cell fusion experiment. In interspecies heterokaryon assays, we found that the phosphorylated nuclear membrane protein BLLF2, which may act a vital role for the viral proliferation or virions assembling, could locate in the nucleus and nucleolus, and shuttle between the cytoplasm and nucleus.
When herpesvirus invades cells, the tegument proteins are delivered into the cytoplasm, which may stay in the cytoplasm or transfer into the nucleus to achieve their corresponding functions, such as inhibiting the transcription and translation of the host cells, restraining the host innate immunity, etc. to facilitate the propagation of the virus. After replication, the virus will synthesize a series of structural proteins to assemble progeny virions, which also play different roles in the life cycle of the virus. In the case of transient transfection, the newly synthesized BLLF2 fusion protein will also locate in the speci c cell compartments to perform its function(s), which can simulate the function(s) and/or characteristics of BLLF2 to a certain extent during EBV infection. As we known, speci c antibody is the key tool to investigate the function of target protein. In our previous experiment design, we had considered we could use speci c BLLF2 antibody to detect the time course expression and subcellular localization of BLLF2 during EBV lytic infection induced from EBV latent cells, which can be used to analyze the correlation between the function(s) and characteristics of BLLF2 during EBV lytic infection and BLLF2 transient expression. We had tried to induce the expression of BLLF2 in prokaryocyte (E.coli) to prepare the BLLF2 antibody by using the mature antibody preparation technology in our laboratory [34][35][36][37], but the expression of BLLF2 could not be effectively induced after many attempts. Thus, the BLLF2 antibody was not successfully prepared. Besides, there is no commercial BLLF2 antibody available, we therefore unable to analyze the correlation between the function(s) and characteristics of BLLF2 during EBV lytic infection and BLLF2 transient expression.
For accomplishing nucleocytoplasmic shuttling, nuclear localization of a speci c protein demands the engagement of functional NES and NLS. Bioinformatics analysis revealed that BLLF2 carries four potential NLS motifs (pat4 44 RRPR 47 and 52 KRRR 55 , and pat7 48 PPVAKRR 54 and 49 PVAKRRR 55 ) and two potential NES 83 VQSPPQITAVIQL 95 (NES1) and 102 MRPPIYL 108 (NES2). In the present study, aa1-82 was divided into two regions aa1-47 and aa48-82. In aa1-47, aa16-31 was identi ed as a functional NLS (NLS1). Moreover, aa48-58 was also established as a functional NLS (NLS2) in aa48-82. Since these two functional NLSs are not adjacent to each other, and the functional NLS is generally not long. Thus, the functional NLSs of BLLF2 are aa16-31 and aa48-58. Besides, aa44-47 exerts a substantial role for the nucleolus localization of BLLF2. However, no functional NES was found.
Nuclear export is a strikingly sophisticated and extremely regulated procedure in cells. The members of the importin/exportin family of nucleocytoplasmic transport receptors, such as TAP and CRM1, are essential for the nuclear export of proteins. For example, the nuclear exports of EBV BFLF2 and EB2 (Mta) are accomplished by directly associating with TAP/NXF1 [10,21]. HSV-1 ICP27, varicella-zoster virus (VZV) IE4 and kaposi's sarcoma-associated herpesvirus (KSHV) ORF57 can bind to several cellular export factors, including SRp20, ASF/SF2, Aly, 9G8 and TAP, to accelerate the export of viral mRNAs via the TAP/NXF1 export pathway [9,38]. Pseudorabies virus (PRV) UL54, in uenza A virus (IAV) NS1 and herpesvirus saimiri ORF57 are also demonstrated to achieve mRNA export from the nucleus to cytoplasm by TAP pathway [8,[39][40][41]. Additionally, HSV-1 UL4, UL47 and VP19C are manifested to transport to the cytoplasm through functional NES mediated by CRM1-dependent pathway [29,42,43], which is also exploited by KSHV ORF9, LANA2 and human cytomegalovirus UL94 to ful ll their nuclear exports via a classical NES [44][45][46] (Fig. 7). Besides, the nuclear export activities of chicken anemia virus VP1, IAV NS2 and human immunode ciency virus type 1 Rev are also modulated via CRM1-mediated pathway [47][48][49]. In this study, we found that EBV BLLF2 could shuttle between the cytoplasm and nucleus. Although we did not identify the classical NES in BLLF2, it could not rule out the nuclear export of BLLF2 is mediated through a non-classical pathway or through the interaction with other proteins, or the nucleocytoplasmic shuttling of BLLF2 is mediated by the spatial constituted functional NES. Accordingly, we continued to identify the nuclear export mechanism of BLLF2, and found that its nuclear export neither depends on CRM1 nor TAP. Therefore, the exact nuclear expert mechanism of BLLF2 needs to be further explored in the future study.
The classical importin-dependent mechanism for nuclear transport is well known for importin α/β/NLScargo complex, of which importin α discriminates the NLS, and importin β executes the association with small regulatory Ran-GTP to ship the complex into the nucleus [5,22,[50][51][52]. Moreover, proteins imported into the nucleus also can directly attach to importin β beyond the engagement of importin α-like adaptor [53]. Up to now, some herpesvirus-encoded proteins are reported to be transported into the nucleus by diverse mechanisms. The nuclear translocation of EBV BFLF2 is mediated through Ran-, importin α7-, importin β1-and transportin-1-dependent pathway. EBV EBNA-LP interacts with importin α1 [20], and EBNA1 combines to importin α1 and importin α5 [6], to complete their nuclear tra ckings. HSV-1 UL31 is imported into the nucleus through Ran-, importin α1-and transportin-1-mediated pathway [54].
HSV-1 UL2 is also described to be assisted into the nucleus through Ran-, importin α1-, α5-, α7-, β1-and transportin-1 cellular transport receptors [55]. PRV ICP22 is demonstrated to be targeted to the nucleus via Ran-, importin α1-, and α7-mediated pathway [56]. PRV UL54 is proved to accumulate in the nucleus through a classic Ran-, importin β1-and α5-dependent mechanism [8]. In addition, VZV ORF9 and HSV-1 US11 are shown to be transported into the nucleus via Ran-and importin β-dependent pathway [44,57] ( Fig. 7). Here, our data disclosed that the nuclear tra cking of BLLF2 was restrained by the Ran-GTP Q69L mutant, indicating BLLF2 is a Ran-dependent protein. Furthermore, co-transfection of DNs or inhibitors of importins demonstrated that the nuclear import of BLLF2 is mediated by importin β1.

Conclusion
In conclusion, we identi ed EBV BLLF2 was de nitely located in the nucleus and nucleolus, which were achieved by two functional NLSs 16 KRQALETVPHPQNRGR 31 (NLS1) and 44 RRPRPPVAKRRR 58 (NLS2) and the nucleolus localization signal 44 RRPR 47 . Additionally, BLLF2 was demonstrated to tra c from the cytoplasm to nucleus via Ran-and importin β1-dependent mechanism, without the participation of importin α. However, the nuclear expert of BLLF2 was mediated neither by CRM1-nor TAP-dependent pathway.

Enzymes and antibodies
All cloning-relevant enzymes were supplied by Thermo Scienti c, with the exclusion of KOD-Plus-Neo DNA polymerase and T4 DNA Ligase were afforded by TOYOBO (Japan) and Takara (Beijing, China), respectively. Mouse anti-Flag monoclonal antibody (mAb) and rabbit anti-YFP polyclonal antibody (pAb) were offered by Abmart (Shanghai, China) and RayBiotech, respectively. Mouse nonspeci c IgG was purchased from eBioscience (Thermo Fisher Scienti c, USA). Alkaline phosphatase (AP)-linked anti-rabbit IgG and anti-mouse IgG were provided by Cell Signaling Technology (MA, USA), and protein A/G PLUS-Agarose was bought from Santa Cruz (Texas, USA).
Confocal microscopy analysis COS-7 cells, cultured on cover-slips in the 24 well plate (Corning, USA), were co-transfected with the indicated expression plasmids for 24 h. Transfected cells were then xed with 4% paraformaldehyde (Tianjun biotechnology, China) for 30 min. Subsequently, the xed cells were washed three times with PBS (Boster, Wuhan, China), followed by staining with DAPI for 5 min at 37 °C. Then, the stained cells were placed on the microscope slides (Biosharp, Shanghai, China) for confocal microscope analysis.
Samples were assayed through a Leica confocal laser scanning microscope (Leica SP8), by using 63×1.4NA immersion oil lenses, with excitation wavelength at 512 nm for YFP/EGFP, 405 nm for DAPI and 555-580 nm for mCherry/RFP. All confocal microscopy experiments were performed at least 3 times, and similar results were obtained.

Interspecies heterokaryon assays
The interspecies heterokaryon assays were performed in accordance with previous studies [8,10,25,[62][63][64][65][66]. In short, monkey COS-7 cells were seeded in 6 well plate and transfected with the indicated plasmids. 18 h post-transfection, mouse NIH3T3 cells were added into the COS-7 cells containing cycloheximide (50 μg/mL, inhibiting the synthesis of new proteins) (Sigma, Shanghai, China), which was used to prevent the residual target plasmid entering NIH3T3 cells for expression, to help us analyze whether the target protein can shuttle from COS-7 cells to NIH3T3 cells. In the experiments, cells were treated with or without 20 ng/mL LMB (Sigma, Shanghai, China). 4 h later, polyethylene glycol (Sigma, Shanghai, China) was employed to fuse COS-7 cells with NIH3T3 cells. After hatching for 1 h, cells were stained with DAPI and imaged by uorescence microscopy. All interspecies heterokaryon assays were performed at least 3 times, and similar results were obtained.

Co-IP and western blot assays
Co-immunoprecipitation (Co-IP) and western blot (WB) assays were performed as described previously [8, 44, 54-56, 58, 59, 67, 68]. Brie y, HEK293T cells co-transfected with different expression plasmids combination were collected and lysed with 600 to 800 μL lysis buffer on ice for 30 min, then centrifuged at 12000 g at 4 ℃ for 15 min. For each Co-IP, supernatants were immunoprecipitated with anti-YFP pAb, anti-Flag mAb or mouse nonspeci c IgG at 4 ℃ for 2 to 4 h, and then incubated overnight with protein A/G beads. The beads were then washed four times with PBS buffer at 2500 g at 4 ℃ for 5 min. Subsequently, the complex with SDS-PAGE loading buffer was boiled for 10 min and subjected to SDS-PAGE/WB analysis after centrifuging at 12 000 g for 5 min. For WB analysis, the immunoprecipitated proteins were stained with the indicated Abs. Cell lysates were also directly subjected to WB analysis, to verify the expression of speci c proteins. All Co-IP and WB assays were performed at least 3 times, and similar results were obtained. Figure 1 Subcellular localization and nucleocytoplasmic shuttling of BLLF2. (A) Potential NESs and NLSs of BLLF2 were predicted by the bioinformatics softwares NetNES 1.1 and PSORT II, respectively. Proteins assigned for tra cking into the nucleus encompass aa targeting sequences termed NLSs [69], and proteins bound for delivery out of the nucleus consist of aa targeting sequences named NESs [70]. The basic aas are arginine (R), histidine (H) and lysine (K). PSORT II adopts the following two standards to dissect target protein: 4 residue pattern (termed 'pat4') formed by 4 basic aa (K or R), or formed by three basic aa (K or R) and either H or P; the other (termed 'pat7') is a pattern beginning with P and followed within 3 residues by a basic segment including 3 K/R residues out of 4. The identi ed NLSs 16KRQALETVPHPQNRGR31 (NLS1) and 44RRPRPPVAKRRR58 (NLS2) were also indicated.     Nuclear export mechanism of BLLF2. Interspecies heterokaryon assays were performed to analyze nuclear export of BLLF2. Mouse NIH3T3 cells were plated onto the CRM1-dependent positive control UL4 (A), CRM1-independent negative control EYFP vector (B) or pEYFP-BLLF2 (C) transfected COS-7 cells, with or without LMB treatment, as described in Material and methods. Cells were then stained with DAPI and imaged by uorescence microscopy. NIH3T3 cells were identi ed by their speckled nuclei (red arrowhead). (D) COS-7 cells were individually transfected with UL4-EYFP or co-transfected with expression plasmids CRM1-mCherry/UL4-EYFP or CRM1-mCherry/EYFP-BLLF2, then examined by confocal microscopy. (E) COS-7 cells were individually transfected with the TAP-dependent positive control EYFP-BFLF2 or co-transfected with expression plasmids TAP-mCherry/EYFP-BFLF2 or TAP-mCherry/EYFP-BLLF2, then examined by confocal microscopy. All scale bars indicate 20 um.