Glycoprotein gp110 of Epstein–Barr virus determines viral tropism and efficiency of infection

B95.8 is an EBV-immortalized marmoset monkey lymphoblastoid cell line (12), and 293 was generated by transfection of the adenovirus type 5 E1a and E1b genes into human embryonic epithelial kidney cells (13). Raji, Akata, P3HR1, and BJAB are human Burkitt's lymphoma cell lines (14, 15). RJ2.2.5 is an HLA class II negative mutant of the Raji cell line (16). M-ABA is a lymphoblastoid cell line established with a virus isolated from a nasopharyngeal carcinoma (17). HeLa is a human cervix adenocarcinoma cell line. Molt-4 is derived from a peripheral T cell lymphoma (18). All cell lines with the exception of HeLa cells were grown in RPMI 1640 medium supplemented with 10% FCS. HeLa cells were grown in DMEM/25 mM Hepes medium supplemented with 10% FCS.

The EBV lytic cycle was induced by transfection of the BZLF1 viral transactivator (19, 20). The B95.8 BALF4 gene, which encodes the gp110 glycoprotein, was inserted into the pRK5 expression plasmid p2670. The B95.8 BALF4 ORF was amplified by PCR with primers (5′-CATATGACTCGGCGTAGGGT-3′ and 5′-CAATTGAACTCAGTCTCTGCCT-3′), cleaved with SmaI, and inserted into pUC19 to yield p2375. After complete sequencing, the NdeI/MunI fragment of p2375 was treated with the Klenow fragment of DNA polymerase and ligated into cytomegalovirus-promoter-containing pRK5 that had previously been cleaved with EcoRI/HindIII and treated with Klenow to yield p2650. p2303 encompasses the EBV gp350 gene (9).

Fixed (5 min, 1% paraformaldehyde) or living cells were incubated for 30 min with a mouse mAb (Chemicon) against gp110 (dilution 1:500 in PBS/5% FCS). After three washings, a Cy5-coupled sheep anti-mouse Ab was applied to living cells for 30 min at 37°C. FITC-labeled donkey anti-mouse Ab was used as a secondary Ab in fixed cells. After repeated washings in PBS, staining was recorded by using a laser scanning microscope (Zeiss).

The 293 cells containing the EBV recombinant virus (2089) were induced by transient transfection of the BZLF1 expression plasmid p509 with or without cotransfection of the BALF4 expression plasmid (21). B95.8, P3HR1, and M-ABA were induced with phorbol 12-tetradecanoate 13-acetate (TPA; 20 ng/ml final concentration) and butyrate (3 mM final concentration), whereas Akata was induced with anti-IgG immunoglobulins (22). Supernatants were harvested 3 days after induction and kept frozen at −80°C. Target cells were infected with 1 ml of filtered (0.8 μm) supernatants containing infectious viruses and kept in culture in cluster plates for 3 days before evaluation of GFP expression by UV microscopy. To detect viruses bound to their target cells (see below), BJAB and 293 cells were incubated with viral supernatants for 3 h or overnight at 4°C, washed twice in PBS, and further used for Western blotting or for Gardella gel analysis. In other experiments, viruses were precipitated in a polyethyleneglycol (PEG)-containing solution (0.5% wt/vol PEG 6000 in 0.5 M NaCl) and collected by centrifugation at 9,000 × g for 20 min.

Plasmid DNA was introduced into cell lines by using lipid micelles (Lipofectamine, GIBCO/BRL). Cells were seeded into six-well cluster plates 1 day before transfection. For transfection, 70% confluent cells were placed in OptiMEM (Invitrogen) minimal medium for 2 h and incubated with DNA embedded in lipid micelles for 4 h.

Viral linear DNA from viruses present in cell-free supernatants or bound to EBV-negative target cells or from lytically induced EBV-positive cells was detected by using the agarose gel electrophoresis described by Gardella et al. (23). In this method, PEG-precipitated viruses (from 1 ml of supernatant) or 10⁶ lytically induced cells are directly lysed in gel slots to avoid shearing of the viral DNA. Southern blot hybridization has been described (21).

Protein extracts from three different sources were analyzed in this work. Extracts from 5 × 10⁶ lytically induced cells were generated by resuspending cells in an extraction buffer (final concentration, 50 mM Tris at pH 7.5/150 mM NaCl/0.1% SDS/1% sodium deoxycholate/1% Triton X-100) containing glass powder to shear DNA. After vigorous vortexing for 30 sec, cell debris was spun down at 16,000 × g for 5 min and supernatants were directly loaded without denaturation onto an SDS/7.5% acrylamide gel. A similar method was applied to viruses bound on target cells: 5 × 10⁶ EBV-negative cells, e.g., BJAB or 293, were incubated with filtered supernatants and protein extracts were generated. A third assay assessed the presence of viral glycoproteins within the mature virion. Filtered supernatants were ultracentrifuged at 20,000 × g for 2 h. Virus pellets were then resuspended in 50 μl of extraction buffer, and debris was discarded after centrifugation at 16,000 × g for 5 min. After electrophoresis, proteins were electroblotted onto an ECL-Membrane (Amersham Pharmacia) for 1 h at 500 mA. After preincubation of the blot in PBS/5% dry milk powder containing 0.1% Tween 20, a mAb against gp110 (Chemicon MAB8184 dilution 1:1,000), or gp350 (American Type Culture Collection murine hybridoma 72A1) was added overnight at 4°C. After extensive washings in PBS, the blot was incubated for 1 h with a goat anti-mouse horseradish peroxidase-coupled secondary Ab (Promega, final dilution 1:10,000). Bound Abs were revealed by using the ECL detection reagent (Amersham Pharmacia) with exposure times varying from 10 to 120 sec. Relative amounts of antigen in Western blots or of DNA in Southern blots were estimated by scanning blots on a Heidelberg 1200 scanner at 1,200 dpi followed by analysis on a Macintosh G3 computer by using the public domain nih image program (developed at the U.S. National Institutes of Health and available on the Internet at http://rsb.info.nih.gov/nih-image/).

EBV usually persists latently in infected cells but can be induced to enter the lytic phase in permissive cells. Activators of the lytic phase of EBV include chemicals such as phorbol 12-tetradecanoate 13-acetate or butyrate, immunoglobulins directed against human IgG, and the EBV transactivator BZLF1. The 2089 cell line is a 293 cell clone stably transfected with a recombinant EBV from which the GFP gene is expressed (21). Viruses carrying the recombinant EBV DNA can be easily recovered from the 2089 cell line on induction of lytic replication. Cells infected with these recombinant EBV express GFP and are easily identified.

Deletion of BALF4 has been shown to completely inhibit virus production by impeding viral maturation (11), suggesting that BALF4 knockouts are unlikely to provide appropriate tools to assess BALF4 functions. We therefore constructed the BALF4 expression plasmid p2670 and assessed viral titers by infecting different target cells with supernatants obtained from BZLF1-induced 2089 cells cotransfected with p2670 or a control plasmid. Surprisingly, we found that infection of Raji cells was enhanced >10-fold and up to 100-fold after transfection of BALF4, as evaluated by counting the number of GFP-positive cells 3 days after infection (Fig. 1A).

A similar effect was observed with primary B cells. Primary B cells were plated at 10⁴ cells per well in 96-well cluster plates in the presence of virus stocks obtained with BALF4- or control-transfected 2089 cells. After 6 wk, the number of wells containing immortalized B cells was counted. As shown in Fig. 1B, supernatants obtained by cotransfection of BALF4 were much more effective in inducing B cell proliferation. Such supernatants contained ≈100 times more immortalizing virions per volume than supernatants from 2089 cells generated by transfection of BZFL1 only.

The increased virus titer observed upon heterologous expression of BALF4 during the lytic phase of 2089 cells might result from an improved production of progeny virus. To test this hypothesis, 2089 cells were cotransfected with BZLF1 and BALF4 or BZLF1 together with a control plasmid and were analyzed in terms of late viral protein expression and viral DNA replication. Similarly, virus stocks were investigated for their content of viral structural proteins and unit-length EBV genomic DNA.

Western blot analysis was used to analyze late viral protein expression in 2089 cells after induction of the lytic cycle. The steady-state level of the EBV late gene product gp350 in the producer cells was found to be the same whether BALF4 was overexpressed or not, showing that BALF4 does not influence late gene protein expression (Fig. 2A). Similarly, no major difference was found in the amount of gp350 in viruses sedimented by ultracentrifugation from supernatants generated in the presence or absence of exogenous BALF4 (Fig. 2A). gp350 is a major constituent of the virion and abundant in the viral envelope, suggesting that the physical concentration of virus particles in the supernatants was not influenced by BALF4 overexpression during the lytic cycle of EBV. To confirm these observations, viral DNA content in supernatants and the amount of linear unit-length virion DNA in 2089 cells after induction of the lytic cycle were analyzed by Gardella gel analysis, followed by Southern blotting with the EBV-specific BamHI W probe (Fig. 2B). Again, no difference could be found between cells and supernatants with or without transfection of a BALF4 expression plasmid. These results demonstrated that gp110 overexpression during the lytic cycle of EBV in 2089 cells does not increase the number of viral particles in the virus stocks. Instead, transfection of the BALF4 expression plasmid led to a clear increase of gp110 incorporated into virions (Fig. 2A). Therefore, viruses obtained from BALF4 or control transfected 2089 cells were termed gp110^high and gp110^low, respectively.

The level of gp110 present in viral particles might influence binding to the target cells of EBV. To test this hypothesis, various cell lines were incubated with gp110^high and gp110^low viruses for 3 h at 4°C followed by extensive washings. The amount of bound virus was monitored by Gardella gel analysis followed by Southern blot hybridization to determine the amount of genomic virion DNA (Fig. 2C) or Western blot detection of gp350 by using protein extracts of the infected cells (not shown). With both methods, we could not detect any differences between gp110^high and gp110^low viruses in terms of the amount of bound virus, suggesting that gp110 does not influence binding of EBV to its target cells.

Studies performed mainly on herpes simplex virus type 1 (HSV-1) have shown that subcellular location and the function of viral glycoproteins during lytic cycle are related (24). Glycoproteins involved in the first round of envelopment of the viral capsid are found at the nuclear membrane, whereas those expressed at the cellular membrane are eventually incorporated into the mature viral particle just before egress. Immunoelectron microscopy has been used to assess the distribution of gp110 in cells carrying the EBV laboratory strain B95.8 (10). In this work, gp110 was found to be restricted to the nuclear membrane and the near-nuclear endoplasmic reticulum. Prompted by our finding that gp110 can be detected in EBV virion preparations, we reevaluated the subcellular localization of this glycoprotein. To this aim, we stained living or unpermeabilized paraformaldehyde-fixed lytically induced cells with an Ab specific for gp110. In B95.8 cells, which support the lytic cycle of EBV spontaneously, and 2089 cells, which harbor a recombinant EBV genome derived from the B95.8 genome, we were not able to detect gp110 at the plasma membrane. In contrast, lytically induced M-ABA, Akata, P3HR1, or 2089 cells transfected with the expression plasmid encoding BALF4 were found to unequivocally express gp110 at the plasma membrane (Fig. 3 and data not shown). It therefore appears that gp110 cell surface expression during the lytic phase of EBV is not common to all EBV strains.

As already mentioned, expression of glycoproteins at the cell surface is usually followed by their incorporation in mature virions. Because the amount of gp110 expressed at the cell surface during the lytic phase differs in various cell lines, we expected variations in the amounts of gp110 incorporated into the virions of various EBV strains. We analyzed virions from several commonly used laboratory virus strains (P3HR1, B95.8, Akata, and M-ABA) by Western blotting with a gp110-specific Ab (Fig. 4 Top Left). Included in the panel were 2089 virus stocks obtained with or without concomitant heterologous expression of BALF4. The same samples were also examined for gp350 levels and for DNA content to estimate virus titers (Fig. 4 Middle Left and Bottom Left). Scanning of the blots indicated that there was a good if not perfect correlation between the amount of gp350 and DNA content of the virions. In contrast to gp350, which is present in similar amounts in different strains, the amount of gp110 varied considerably among different strains. M-ABA was found to carry the highest amount of gp110, even if the DNA content was slightly higher than that of other viral strains. In contrast, gp110 was barely detectable in B95.8 and 2089. Although Akata and P3HR1 viral titers were much lower than those obtained with the other cell lines, Akata viruses were clearly positive for gp110, which suggests that this viral strain is a high gp110 producer. P3HR1 did not show strong gp110 expression in the virions, but this is probably related to the low viral titers. Supernatants from lytically induced 2089 cells transfected with the BALF4 expression plasmid p2650 showed a relatively strong gp110 signal, confirming that overexpression of BALF4 in the producer cell line during the lytic phase of EBV increased incorporation of the glycoprotein into mature virions. Notably, gp110^high 2089 virus still contains less gp110 than some of the laboratory strains such as M-ABA. It therefore appears that the amount of gp110 incorporated in EBV virions after transfection lies within the range observed for other gp110-positive EBV strains. Results gained from experiments performed with gp110^high viruses can therefore be extended to wild-type viruses. gp110 and gp350 were positive in all induced cell lines by immunostaining (data not shown), and protein synthesis was further tested by using Western blotting (Fig. 4 Right). gp110 was massively produced after transfection into 2089 and appears to be very low in Akata, which is probably related to the low number of induced cells. The remaining cell lines showed comparable expression levels. There was therefore no direct relationship between the amount of gp110 produced in the replicating cell lines and the amount of gp110 present in the mature particle. However, massive overexpression of gp110, even if inefficient, enhanced incorporation.

The tropism of EBV in vitro is restricted almost entirely to B cells, with some notable exceptions such as primary gastric epithelial cells. Akata, a viral strain capable of infecting gastric epithelial cells (3), also expressed high amounts of gp110 (Fig. 4), raising the possibility that gp110 might be critically involved in infection of cells other than human B cells. We therefore extended our infection experiments to non-B cell lines. We first infected 293 cells, a cell line reported to express subliminal amounts of CD21 (9, 25), and found that infection with gp110^high virus stocks increased the number of infected cells 100-fold as compared with gp110^low stocks (Fig. 5 a and b). HeLa is completely CD21 negative and has been reported to be completely refractory to infection by EBV (8, 9, 26). Incubation of HeLa cells with gp110^high supernatants resulted in successful infection of 5–10% of the cells, whereas the control experiment with gp110^low supernatants was negative (Fig. 5c and data not shown). A low degree of infection (0.1%) was obtained with the T cell lymphoma cell line MOLT-4. This cell line is considered to be completely resistant to EBV infection (27), but a few GFP-positive cells were observed after infection with supernatants obtained from induced 2089 overexpressing gp110 only (Fig. 5d).

Interaction between the gH/gL EBV glycoproteins and HLA class II molecules has been shown to be an important modulator of viral infection (28). To assess possible cross-talk between this infection pathway and the enhancing effect of gp110 on infection, we infected RJ2.2.5 cells, an HLA class II-negative mutant of the B cell lymphoma Raji cell line, with both gp110^high and gp110^low supernatants. Enhanced incorporation of gp110 showed increased infection of the target to the same extent as with the HLA class II-positive Raji parental cell line (data not shown). The effect of gp110 on infection is therefore independent of the gH/gL-HLA class II infection.