Characterization of the subcellular localization of Epstein-Barr virus encoded proteins in live cells

Epstein-Barr virus (EBV) is the pathogenic factor of numerous human tumors, yet certain of its encoded proteins have not been studied. As a first step for functional identification, we presented the construction of a library of expression constructs for most of the EBV encoded proteins and an explicit subcellular localization map of 81 proteins encoded by EBV in mammalian cells. Viral open reading frames were fused with enhanced yellow fluorescent protein (EYFP) tag in eukaryotic expression plasmid then expressed in COS-7 live cells, and protein localizations were observed by fluorescence microscopy. As results, 34.57% (28 proteins) of all proteins showed pan-nuclear or subnuclear localization, 39.51% (32 proteins) exhibitted pan-cytoplasmic or subcytoplasmic localization, and 25.93% (21 proteins) were found in both the nucleus and cytoplasm. Interestingly, most envelope proteins presented pan-cytoplasmic or membranous localization, and most capsid proteins displayed enriched or complete localization in the nucleus, indicating that the subcellular localization of specific proteins are associated with their roles during viral replication. Taken together, the subcellular localization map of EBV proteins in live cells may lay the foundation for further illustrating the functions of EBV-encoded genes in human diseases especially in its relevant tumors.


INTRODUCTION
Epstein-Barr virus (EBV), a representative member of family gammaherpesvirus (large DNA virus) that can afford interesting model system for viral-host interactions, is related with some severe human health problems such as infectious mononucleosis, various malignancies including nasopharyngeal carcinoma (NPC), gastric carcinoma, diffuse large B-cell lymphoma, NK/T-cell lymphomas, as well as Hodgkin Lymphoma (HL), Burkitt Lymphoma (BL) and post-transplant lymphoproliferative disease (PTLD) [1][2][3]. The EBV genome is composed of linear double-stranded DNA, about 172 kilobase pairs (kb) in length. Until now, it is reported that EBV encodes approximately 80 to 86 proteins (depending on the strain) [4][5][6][7], among which at least 33 are structural components of the virion [8]. However, the absolute number of EBVencoded proteins is still under investigation because of the examination of different gene transcript products in varied types of infected cells [4,7,[9][10][11][12][13][14]. In terms of Research Paper their localization in the virions, the proteins could be divided into five groups: envelope, tegument, capsid, and unclassified and nonstructural proteins. This complicated network of proteins leads to various consequences on the host cell and guarantees efficient proliferation of the virus.
Although the features of a set of these proteins are already well studied, there is little or nothing report concerning the role of many of the proteins. Therefore, more exhaustive analyses might contribute to figure out the complexity of the EBV pathogenic repertoire. A better perceiving of the individual protein functions is not only conducive to establish the complex replication and potentially approaches for inhibition of propagation, but also helpful for revealing the mechanisms that manipulate vital cellular processes by EBV. Subcellular localization of a protein reflects and is closely correlated with its function, and establishment of the subcellular localization has been demonstrated to be a helpful fashion to estimate the potential functions of a great number of proteins [15], where individual proteins were expressed fused to epitope or to GFP tags [16,17]. For most proteins, localization results were disclosed to be consistent in different survey regardless of the tag used or expression level. Subcellular localization screening is also a proper starting point to identify the roles of a certain protein of herpesvirus, giving insight into the potential effect of the protein in the course of infection, as well as the cellular processes that may be regulated.
The subcellular localization of part of the EBVencoded proteins is not known yet. To acquire a more comprehensive knowledge of the functions of EBV proteins, we constructed a mammalian expression library that is composed of most of the open reading frames (ORFs) of EBV. These proteins fused to the C terminus of enhanced yellow fluorescent protein (EYFP) tag that reasonable for protein localization exploration were expressed in COS-7 live cells, and we presented the subcellular localization for all of these EBV-encoded proteins.

RESULTS
As a substantial step to figure out the detailed function of EBV-encoded proteins, a research has been performed to establish the genome-wide subcellular localization map of EBV-encoded proteins. To investigate the subcellular localization, almost each of the predicted ORFs from EBV, were isolated, and inserted into the expression plasmid, in frame fused with the C terminus of a 27 kDa EYFP tag-encoding sequence, which is demonstrated may but not exert significant influences on protein subcellular localization and favourable for direct fluorescence detection of the respective proteins in live cells. The constructed plasmids were confirmed and transfected into COS-7 cells for expression. COS-7 cells were used as a cell system for these studies because they exhibit a larger cytoplasm than HEK293 cells, which are the more commonly utilized cell types for subcellular localization study. Furthermore, the subcellular localizations of some representative EBV proteins fused with Flag tag were also detected using indirect immunofluorescence (IFA), to make the data more convincing. To get a general overview of the subcellular localization of each protein, Zeiss Axiovert 200M microscope was employed. With this procedure the EBV proteins could be clearly observed in considerable transfected cells. Although not every potential ORF was recovered in the high through-put cloning effort, we produced expression constructs for the majority of EBV, namely 81 EBV ORFs. Among these EBV ORFs, 78 were cloned from B95-8 strain of EBV (174-kb BAC), and the rest ORFs (LF1, LF2 and LF3), which could not isolate from 174-kb BAC, were cloned from the BAC DNA of EBV Akata strain of EBV (AK-BAC).
Protein localizations were broadly classified as pan-cellular (means protein diffusely localized throughout the cytoplasm and nucleoplasm), pan-cytoplasmic (means protein diffusely localized through-out the cytoplasm) or subcytoplasmic (means protein can form spots or concentrated at some subcellular organelles in the cytoplasm), pan-nuclear (means protein diffusely localized through-out the nucleoplasm) or subnuclear (means protein can form spots or concentrated at some subcellular organelles in the nucleus). Some proteins possess more complicated localization patterns and can be divided into multiple categories and, in these cases, the protein was presented in the most prevailing category, since they certainly have the capability to form subcellular structures in some instances, perhaps at higher expression levels. A comprehensive list of all proteins expressed in this study are summarized in Figures 1 to 4 and detailed in Tables 1 to 3. In terms of their localizations, the EBV-encoded proteins could be generally fell into three groups: 28 proteins (34.57%) with nuclear or subnuclear localization ( Figure 1 and Table 1), 32 proteins (39.51%) with cytoplasmic or subcytoplasmic localization (pancytoplasmic) ( Figure 2 and Table 2), and 21 proteins (25.93%) were localized in both the cytoplasm and the nucleus (pan-cellular) ( Figure 3 and Table 3). As a control, the fluorescence of EYFP from cells transfected with pEYFP-C1 was presented evenly distributed throughout the cytoplasm and the nucleoplasm but not the nucleolar structures ( Figure 1).
Of the 81 viral proteins we tested, 20 EBV proteins have not been previously characterized with respect to their localization and most of these proteins also lack any functional identification, 52 EBV proteins have previously published subcellular localization data, and these results are in consonance with previous reports (see Tables 1 to 3 for individual protein results). Meanwhile, www.impactjournals.com/oncotarget Figure 1: Nuclear localization summary of EBV-encoded proteins. 28 EYFP-fused EBV proteins were expressed in COS-7 cells, and cells were subjected to fluorescence microscope analysis in live cells 24 h after transfection. As a negative control, cells were transfected with the vector control (pEYFP-C1). Pictures were obtained using a Zeiss Axiovert 200M microscope. The same magnification was used in all panels. Representative fluorescence images of the vast majority live cells expressing indicated fusion protein were shown. Cells were counterstained with Hoechst to visualize the nuclear DNA. All scale bars indicate 10 μm. www.impactjournals.com/oncotarget     [5] and Salsman et al. [6].
minor discrepancies in localization were detected for 4 proteins (BLLF2, BKRF4, BLRF2 and BBRF1) diverse from previous studies, however, significant discrepancies in localization were observed for 5 proteins (BDLF4, BSRF1, BBLF1, BALF5 and BTRF1) distinct from previous results. Moreover, the subcellular localizations of some Flag-fused representative proteins from each category (BFRF3 and BMRF1 from nuclear localization, BSLF1 and BGLF1 from cytoplasmic localization, BGLF2 and BSRF1 from pan-cellular localization) were consistent with the subcellular localizations of those fused with EYFP tag (Figure 4), further making the results more convincing. Some discrepancies could not be unexpected with localizations assessed in the course of viral infection since the presence of interplays between viral proteins can change the localization of the individual proteins. The localization discrepancies of specific proteins might also be associated with protein expression levels or the presence of a tag. Not surprised, this initial analysis   [6].
demonstrated that such high through-put localization screening will not precisely clarify the localization of every viral protein in the context of infection, but it did present that such an approach is suitable for analyzing the localization of the vast majority of viral proteins even during viral infection.

DISCUSSION
Herpesviruses are large DNA viruses that encode a variety of proteins for complicated interactions with host. For the sake of taking an investigation on the possible roles of the many uncharacterized EBV-encoded proteins, the genomic expression libraries of EBV were constructed and screened to explore the complete intracellular localization map of almost all EBV proteins in mammalian cells. However, when we were cloning EBV genes, the subcellular localizations of 61 from 81 proteins have been probed by others to our knowledge (Tables 1 to 3), and with only slight differences, our results were good correlated with previously reported protein localizations, while new localization data was yielded for approximately 20 previously unlocalized proteins.
Almost one-third of EBV-encoded proteins localized principally in the nucleus (28 proteins) ( Figure 1 and Table  1). Furthermore, 32 proteins primarily showed cytoplasmic or subcytoplasmic localization ( Figure 2 and Table 2), and others localized throughout both compartments (often at unequal levels, 21 proteins) ( Figure 3 and Table 3). It is of interest that 34.57% of the EBV-encoded proteins were detected in the nucleus, whereas only 12% of randomly selected cellular proteins showed nuclear localization [17]. A recent genome-wide subcellular localization study reported that human herpesvirus 8 (HHV-8, gammaherpesvirus) was found to have 51% cytoplasmic and 22% nuclear proteins (with 27% in both compartments) [18], indicates that EBV has a higher proportion of nuclear proteins than HHV-8. Nuclear predominance of EBVencoded proteins is in good consonance with the viral life cycle, which is preferentially associated with the nucleus.
In the present study, some of the results from transfection may be different from infection, because of the interactions of viral proteins during infection (data as shown in Tables 1 to 3). While these transfection results might not uncover the accurate subcellular localization of every viral proteins during infection, the subcellular localization map of individually expressed EBV-encoded proteins could offer helpful data to further examine the mechanism by which individual EBV-encoded protein effects on EBV pathogenesis.
Minor discrepancies were observed for only 4 proteins (BLLF2, BKRF4, BLRF2 and BBRF1) in this study [6,19]. Specifically, BLLF2 and BKRF4 were detected exclusively in the nucleus by other investigators, whereas in our study BLLF2 showed subnuclear localization (nucleolus like), and BKRF4 showed obviously nuclear localization with multiple small foci. The minor capsid protein BBRF1 showed subcytoplasmic (with speckles in the cytoplasmic), which is different from the pan-cytoplasmic localization in previous reports. BLRF2 showed the fluorescence of nucleus is more than cytoplasm (with speckles), but this contrary to previous study in plasmid transfection, which may relocalized from the nucleus to the cytoplasm relied on the interaction with other viral proteins in the course of viral infection.
Significant differences were also observed for 5 proteins (BDLF4, BSRF1, BBLF1, BALF5 and BTRF1). In our study, BDLF4 showed the fluorescence in nucleus is more than cytoplasm, which is different from the pan-cellular localization reported by others [6]. The myristoylated phosphoprotein BBLF1 was found obviously nuclear localization without nucleolus that different with pan-cellular or cytoplasmic localization in previous study, which may depend on its myristoylation modification [6,20]. BSRF1 showed pan-cellular localization, whereas it gave a subcytoplasmic or perinuclear concentration localization in a previous report [6], which may be explained by subtle difference in employ of a different cell lines. The DNA polymerase BALF5 localized exclusively in the cytoplasm, which is different from the predominantly nuclear localization during EBV infection [21]. It's reported that the nuclear transport of HHV-8 DNA polymerase holoenzyme is dependent on the nuclear localization signal (NLS) present on the processivity factor PF-8, since the catalytic subunit pol-8 lacks a functional NLS, and hence the two subunits are targeted into the nucleus as a complex [22]. Therefore we speculated that the transport of BALF5 from cytoplasm to nucleus requires the expression of additional viral factors, which could help BALF5 target into the nucleus for executing its function in viral DNA synthesis. In addition, BTRF1 was also detected only in the cytoplasm, which is demonstrated to be a nuclear-targeted protein (nuclear > cytoplasmic) previously.
Latency is a regulatory status, which may mainly rely on nuclear proteins to manipulate host cell and viral transcription. Compared with the intracellular localization map of HHV-8 [18], which found that all latencyassociated proteins showed a nuclear staining pattern, we detected only two latent related proteins EBNA3A (BLRF3/BERF1) and EBNA1 (BKRF1) localized in the nucleus, whereas the nuclear phosphoprotein BWRF1, latent membrane protein BNLF1 (LMP1), LMP2A and LMP2B showed pan-cellular (with speckles in cytoplasm and perinuclear), subcytoplasmic (with speckle cytoplasmic structures), perinuclear speckle concentration and nuclear membrane (like trans-Golgi network) localization, respectively. This disparity may be due to the different viral life cycle between EBV and HHV-8.
The assembly compartment for viral proteins in the host cell is probably relevant to its subcellular localization. EBV contains 13 membrane related protein, and 7 glycoproteins (BLLF1/gp350, BXLF2/ gH, BALF4/gB, BBRF3/gM, BLRF1/gN, BDLF3/gp150 and BKRF2/gL) are incorporated into its envelope [5,8], which play crucial roles during virus infection. Besides, BILF2/gp55/80, a predicted membrane protein, also perhaps targets to the envelope. In this study, most of these proteins exhibited pan-cytoplasmic, subcytoplasmic, membraneous or pan-cellular localization, without nuclear or subnuclear localization (Figures 2 and 3, Tables 2 and 3). This is high concordance with the compartment where the envelope assembling occurs. Other 4 membrane proteins (BZLF2/ gp42, BILF1/G-PCR, BMRF2 and BFRF1) also showed plasma membrane, pan-cytoplasmic or subcytoplasmic and perinuclear localization (with speckle cytoplasmic structures), but only BFLF2 localized absolutely to the nucleus, which is in accordance with the subcellular localization of its homologue HSV-1 UL31.
In addition, it has been established that EBV virion has 6 proteins in its capsid (BDLF1, BORF1, BBRF1, BVRF2, BdRF1 and BFRF3) [5,8], and BTRF1 is also predicted to be a potential protein that implicated in capsid maturation. In this work, BVRF2 and BFRF3 displayed enriched localization in the nucleus, and BORF1 (subnuclear) and BdRF1 showed complete localization in the nucleus, where the capsid assembling takes place. However, BDLF1, BBRF1 and BTRF1 demonstrated pan-cellular, pan-cytoplasmic or subcytoplasmic localization.
EBV is the first known human tumor virus to play an essential role in the induction of a broad spectrum of human lymphoid and epithelial malignancies [23][24][25][26], yet the fundamental mechanism of how EBV contributes to cancer remain unknown. Unlike other herpesviruses, the development of EBV-related tumorous diseases is relevant with the latent cycle, by virtue of the immune system is incompetent to monitor latently infected cells. It's shown that EBV can encode some viral oncoproteins that involved in EBV tumorigenicity (including EBNA1, EBNA2, LMP1, LMP2, EBNA3A, EBNA3C, BARF0, BALF1, RPMS1, BARF1 and BNLF1) [27][28][29][30][31][32][33][34][35][36][37][38], which are associated with Burkitt's lymphoma (EBNA1, LMP2A and BARF0), Hodgkin's disease, nasopharyngeal carcinoma and NK/T-cell lymphoma (EBNA1, LMP1, LMP2A and LMP2B), gastric carcinomas (BARF1), post-transplant lymphoproliferative disorders and AIDS-related lymphomas (EBNA1, EBNA3A, LMP1, LMP2A and LMP2B). It's well known the subcellular localization plays a critical role in the function execution of a specific protein, and therefore we speculated the subcellular localization of these mentioned EBV proteins might take some potential roles in the EBV-related lymphoid and epithelial malignancies, e.g. the subcytoplasmic localization of LMP1 may be crucial for it to lead to changes that is connected with B-cell activation, including B-cell fusion, increase of CD23, CD39, CD40, CD44 expression and apoptosis-restraining effects [39,40]. Furthermore, this localization may also important for LMP1 to promote oncogenesis and transformation of primary rodent fibroblasts and to impede differentiation of a squamous carcinoma cell line [41]. However, the exact pathological roles in EBV-related malignancies are not fully elucidated, this need further in-depth study.
In conclusion, this study on the construction of a library of expression clones for the EBV proteome, we believe, will be a remarkably essential work in producing highly valuable platform for further studies of the viral life cycle and mechanistic pathogenesis in the future. Additionally, it will also be applicable for screening the possible viral proteins or host cellular factors that may interact with viral proteins.

Cloning of EBV genes
The enzymes used for cloning programs were purchased from Thermo Scientific except DNA polymerase KOD-Plus-Neo from TOYOBO and T4 DNA Ligase from Takara. The 81 ORFs of EBV from the NCBI entries, including the start methionine, were amplified by PCR from the BAC DNA of B95-8 strain of EBV (174-kb BAC) except LF1, LF2 and LF3 from the BAC DNA of Akata strain of EBV (AK-BAC) [4], using specific primers with suitable overhanging restriction enzyme motifs (contain HindIII and BamHI sites unless otherwise specified). Due to EBV genome contains high GC, the annealing temperature for PCR reaction is generally high. EYFP has been widely employed as a reporter to visualize EYFP-tagged proteins in live cells. Therefore, the amplified DNA products were digested with HindIII and BamHI and inserted into the multicloning site of pEYFP-C1 (Clontech, BD Biosciences) in frame with an EYFP tag at the C terminus, which is digested with appropriate restriction enzymes, with the aim to yield corresponding EYFP fusion protein to allow direct observation of the subcellular localization of each protein. Furthermore, some representative proteins from each category (BFRF3 and BMRF1 from nuclear localization, BSLF1 and BGLF1 from cytoplasmic localization and BGLF2 and BSRF1 from pan-cellular localization) were also subcloned into pCMV-Flag-N1 (Beyotime Biotechnology) in frame with a Flag tag at the N terminus. All constructs described above were verified by plasmid PCR, restriction analysis and full-length DNA sequencing, and all primers used in this research are available upon request.

Plasmid transfection and fluorescence microscopy
To test the subcellular localization of EBV proteins in live cells, plasmid transfection and fluorescence microscopy assays were performed as described in our previous studies [42][43][44][45][46]. Briefly, COS-7 cells were plated onto 12 well plates (Corning, USA) and cultured in DMEM with 10% FBS overnight to reach the confluency 60-80% before transfection. The next day, monolayer cells were transfected with 1.5 μg of assigned plasmid DNA mixed with TurboFect Transfection Reagent (Thermo Scientific) as per the manufacturer's instructions. After transfection for 24 h, the live cells were beard for fluorescence microscopy. To make the data more convincing, the subcellular localizations of some representative proteins fused with Flag tag from each category were detected by IFA, using anti-Flag monoclonal antibody (mAb) (ABmart) and fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG (Sigma-Aldrich), as described in previous studies [47]. In the same observation, each transfection was carried out for at least two times. Data shown were from one illustrative experiment. Fluorescences were analyzed using a Zeiss Axiovert 200 M inverted fluorescence microscope (Carl Zeiss, Germany), equipped with a halogen lamp (100 HAL, 12 V, 100 Watt) for transmitted light microscopy and an objective LD "Plan-Neofluar" with 40×/0.6 Corr M27 (D=0-1.5) lens (WD=3.3mm when D=0 and WD=2.5mm when D=1.5). The YFP (EX BP 500/20, BS FT 515, EM BP 535/30), FITC (EX BP 475/40, BS FT 500, EM BP 530/50) and DAPI (EX G 365, BS FT 395, EM BP 445/50) filtersets were used to detect EYFP labelled proteins, FITC labelled proteins and nuclear DNA labelled with Hoechst, respectively. Transmitted and fluorescence light images were captured under a digital camera (Axiocam; Carl Zeiss), with Zeiss AxioVision Rel. 4.8 software for controlling the image recording, microscope stage and image merge. Microscopic settings were kept constant for comparisons among different samples. All the pictures were taken under a magnification of 400×. Classification of subcellular localization of the proteins was determined by three researchers independently, and categorization was discussed until consensus was reached. Each picture represents most of the cells with similar subcellular localization. Light-translucent photomicrographs are introduced to show cellular morphology. Cells were counterstained with Hoechst to visualize the nuclear DNA. Fluorescent images of EYFP fusion proteins and FITC labelled protein were presented in pseudocolor green and genuine color green, respectively, and merged with Hoechst using Zeiss AxioVision Rel. 4.8 software. All scale bars indicate 10 μm, and images were processed using Adobe Photoshop.