The Large Groove Found in the gH/gL Structure Is an Important Functional Domain for Epstein-Barr Virus Fusion

Abstract

Epstein-Barr virus (EBV) mediates viral entry into cells using four glycoproteins—gB, the gH/gL complex, and gp42—and fusion is cell type specific. gB and gH/gL are required for epithelial cell fusion; B cell fusion also requires gp42. To investigate functional domains within the gH/gL structure, we constructed site-directed EBV gH/gL mutants with alterations of residues located in a large groove that separates domain I (D-I) from domain II (D-II) within the gH/gL structure. We found that substitution of alanine for leucine 207 reduces both epithelial and B cell fusion and is accompanied by reduced gp42 binding. We also observed that substitution of alanine for arginine 152, histidine 154, or threonine 174 reduces fusion with epithelial cells but not with B cells. To test whether flexibility of the region between D-I and D-II of gH/gL could be important for membrane fusion activity and to allow potential interactions across the D-I/D-II groove, we mutated D-I amino acids V47, P48, and G49 to cysteine, allowing novel intersubunit disulfide bonds to form with the free C153 located in D-II. We found that the G49C mutant, predicted to bridge D-I and D-II with C153 of gH/gL, had normal B cell fusion activity but reduced epithelial cell fusion activity, which was partially restored by treatment with dithiothreitol. We conclude that structural rearrangements and/or interactions across the D-I/D-II groove of gH/gL are required for fusion with epithelial cells but not for fusion with B cells.

INTRODUCTION

Epstein-Barr virus (EBV), a member of the gammaherpesvirus family, was first identified in tumor biopsy specimens obtained from young children with Burkitt lymphoma (1–3). Along with this cancer, EBV has been associated with a variety of other cancers, including Hodgkin lymphoma and gastric carcinoma. EBV is also involved in a number of important disorders associated with diminished immunity, including AIDS-related malignancies, posttransplant lymphoproliferative disease (PTLD), and oral hairy leukoplakia (2, 3). EBV replicates in epithelial cells and establishes long-term latency in lymphocytes (2, 3). The binding of EBV and the subsequent fusion of the virion envelope with a host cell is mediated by multiple EBV-encoded glycoproteins and requires multiple steps, culminating with the release of the virus capsid into the cytoplasm.

The EBV fusion machinery consists of gB, the heterodimeric gH/gL complex, and glycoprotein 42 (gp42), all of which are required for the fusion of EBV with B cells. The fusion of EBV with epithelial cells requires only gB and the gH/gL complex, which forms the core fusion machinery found in all herpesviruses (4, 5). The presence of gp42 inhibits epithelial cell fusion, thus acting as a switch directing the entry of EBV into B cells or epithelial cells (6). The crystal structure of the EBV gH/gL complex was recently resolved, and a KGD motif was found to be prominently located on the surface of domain II (D-II) of gH. Also found in the gH/gL crystal structure was a large groove between domain I (D-I) and D-II, adjacent to the gH/gL KGD motif (7). We recently found that the gH/gL KGD motif is bifunctional, orchestrating the infection of B cells and epithelial cells by interaction with the epithelial cell receptor or gp42 (8). The D-I/D-II groove adjacent to the KGD motif also appears well suited to participate in B cell and epithelial cell fusion and was investigated in the current study by using a structure-based mutagenesis approach to further define the functional role of this region in EBV fusion.

MATERIALS AND METHODS

Cell culture.

Chinese hamster ovary cells (CHO-K1 cells) were grown in Ham's F-12 medium (BioWhittaker) containing 10% FetalPlex animal serum complex (Gemini Bio Products) and 1% penicillin-streptomycin (100 U penicillin/ml, 100 μg streptomycin/ml; BioWhittaker). The Daudi 29 cell line (for B cell fusion) and human embryonic kidney (HEK) 293 cells (for epithelial cell fusion) stably expressing T7 RNA polymerase (9, 10) were grown in RPMI 1640 medium with 900 μg/ml G418 (Sigma) and in Dulbecco's modified Eagle medium (DMEM) with zeocin, respectively, containing 10% FetalPlex animal serum complex and 1% penicillin-streptomycin.

Construction.

Mutations of gH (E58A, S154A, Q150A, R152A, C153A, H154A, T174A, D203A, L207A, S212A, T217A, Q220A, V47A, V47C, P48A, P48C, G49A, G49C, and G49S; numbers indicate the wild-type [wt] gH residue mutated to alanine, cysteine, or serine) were generated using the QuikChange site-directed mutagenesis kit (Stratagene). The primers used are shown in Table 1. Flag-tagged gH with a substitution of alanine for arginine at residue 152 (Flag-gH-R152A) was constructed by using the same R152A primer set and Flag-gH expression construct as those published previously (11). Sequencing was carried out to confirm the presence of the mutation and the absence of other mutations.

Table 1.

Sequences of primers for the mutants

Primer no.	Primer designation	Primer sequence
1	EBV gH-E58A F	5′-CTGTGGAGAGCAGCAAATGTC-3′
2	EBV gH-E58A R	5′-GACATTTGCTGCTCTCCACAG-3′
3	EBV gH-S145A F	5′-CGGCCACACGCATATGTCTTT-3′
4	EBV gH-S145A R	5′-AAAGACATATGCGTGTGGCCG-3′
5	EBV gH-Q150A F	5′-GTCTTTTATGCACTGCGCTGT-3′
6	EBV gH-Q150A R	5′-ACAGCGCAGTGCATAAAAGAC-3′
7	EBV gH-R152A F	5′-TATCAGCTGGCATGTCACTTG-3′
8	EBV gH-R152A R	5′-CAAGTGACATGCCAGCTGATA-3′
9	EBV gH-C153A F	5′-CAGCTGCGCGCACACTTGTCT-3′
10	EBV gH-C153A R	5′-AGACAAGTGTGCGCGCAGCTG-3′
11	EBV gH-H154A F	5′-CTGCGCTGTGCATTGTCTTAT-3′
12	EBV gH-H154A R	5′-ATAAGACAATGCACAGCGCAG-3′
13	EBV gH-T174A F	5′-GGGGCCATGGCATCTAAATTT-3′
14	EBV gH-T174A R	5′-AAATTTAGATGCCATGGCCCC-3′
15	EBV gH-D203A F	5′-AAGACGAAGGCACTGCCGGAT-3′
16	EBV gH-D203A R	5′-ATCCGGCAGTGCCTTCGTCTT-3′
17	EBV gH-L207A F	5′-CTGCCGGATGCAAGGGGGCCT-3′
18	EBV gH-L207A R	5′-AGGCCCCCTTGCATCCGGCAG-3′
19	EBV gH-S212A F	5′-GGGCCTTTTGCATACCCATCC-3′
20	EBV gH-S212A R	5′-GGATGGGTATGCAAAAGGCCC-3′
21	EBV gH-T217A F	5′-CCATCCTTAGCAAGTGCCCAA-3′
22	EBV gH-T217A R	5′-TTGGGCACTTGCTAAGGATGG-3′
23	EBV gH-Q220A F	5′-ACCAGTGCCGCAAGCGGGGAC-3′
24	EBV gH-Q220A R	5′-GTCCCCGCTTGCGGCACTGGT-3′
25	EBV gH-V47A F	5′-CTGATGGCAAAGGCACCAGGCCTTAGC-3′
26	EBV gH-V47A R	5′-GCTAAGGCCTGGTGCCTTTGCCATCAG-3′
27	EBV gH-P48A F	5′-ATGGCAAAGGTCGCAGGCCTTAGCCCA-3′
28	EBV gH-P48A R	5′-TGGGCTAAGGCCTGCGACCTTTGCCAT-3′
29	EBV gH-G49A F	5′-GCAAAGGTCCCAGCACTTAGCCCAGAG-3′
30	EBV gH-G49A R	5′-CTCTGGGCTAAGTGCTGGGACCTTTGC-3′
31	EBV gH-V47C F	5′-CTGATGGCAAAGTGTCCAGGCCTTAGC-3′
32	EBV gH-V47C R	5′-GCTAAGGCCTGGACACTTTGCCATCAG-3′
33	EBV gH-P48C F	5′-ATGGCAAAGGTCTGTGGCCTTAGCCCA-3′
34	EBV gH-P48C R	5′-TGGGCTAAGGCCACAGACCTTTGCCAT-3′
35	EBV gH-G49C F	5′-GCAAAGGTCCCATGTCTTAGCCCAGAG-3′
36	EBV gH-G49C R	5′-CTCTGGGCTAAGACATGGGACCTTTGC-3′
37	EBV gH-G49S F	5′-GCAAAGGTCCCATCTCTTAGCCCAGAG-3′
38	EBV gH-G49S R	5′-CTCTGGGCTAAGAGATGGGACCTTTGC-3′

Transfection.

CHO-K1 cells, grown to approximately 80% confluence, were transiently transfected with plasmids expressing the mutants and other glycoproteins essential for fusion, including gB (0.8 μg), gH (0.5 μg), gL (0.5 μg), and gp42 (0.8 μg), and with a luciferase reporter plasmid with a T7 promoter (0.8 μg), by using the Lipofectamine 2000 transfection reagent (Invitrogen) in Opti-MEM (Gibco) as described previously (8). Equal amounts of wt gH and gH mutant DNAs were used in each experiment.

Soluble gH binding assays.

CHO-K1 cells were transfected with gH/gL or gH-R152A/gL as described above. Twenty-four hours later, 2 × 10⁶ cells were collected, resuspended in 1 ml medium, subjected to 3 freeze-thaw cycles in liquid nitrogen, and sonicated for 10 s. Insoluble cellular debris was removed by centrifugation at 1,500 × g for 5 min. HEK 293 cells were overlaid with clarified supernatants (100 μl) in a 96-well dish in triplicate. After incubation for 4 h at 4°C, gH/gL binding was examined using a cell-based enzyme-linked immunosorbent assay (CELISA) as described below. For the detection of binding by Western blotting, we applied 1-ml portions of clarified supernatants containing Flag-tagged gH/gL or Flag-tagged gH-R152A/gL to the HEK cells seeded in 6-well plates and incubated the mixture for 1 h at 4°C. The cells were then washed four times with ice-cold phosphate-buffered saline (PBS) and were lysed with 200 μl of sodium dodecyl sulfate (SDS) sample buffer. Proteins were separated on 10% Mini-Protean TGX gels (Bio-Rad) after boiling for 10 min under reducing conditions. Western blot analyses were performed using a polyclonal anti-Flag antibody (F7425; Sigma) and a polyclonal anti-glyceraldehyde-3-phosphate dehydrogenase (anti-GAPDH) antibody (Abcam) at 1:1,000.

Fusion assay.

The virus-free cell-based fusion assay was performed as described previously (8). Briefly, effector CHO-K1 cells were transfected as described above. Twenty-four hours posttransfection, the cells were detached, counted, and mixed 1:1 with target cells (Daudi 29 B cells or HEK 293 cells; 0.25 × 10⁶ per sample) into a 24-well plate in 1 ml Ham's F-12 medium. Twenty-four hours later, the cells were washed once with PBS and were lysed with 100 μl of passive lysis buffer (Promega). Luciferase was quantified in duplicate by transferring 20 μl of lysed cells to a 96-well plate, adding 100 μl of luciferase assay reagent (Promega), and measuring luminescence on a Perkin-Elmer Victor plate reader.

CELISA.

The expression of the various mutants in the plasma membrane was determined by CELISA as described in previous reports (8). CHO-K1 cells were transfected with various glycoproteins and mutants. Transfected cells were split for use in the fusion assay (above) and CELISA. Twenty-four hours posttransfection, 4 × 10⁴ cells/well were transferred to a 96-well plate and were incubated for another 24 h. The expression of each glycoprotein (including mutants) was evaluated using conformation-specific antibodies (E1D1 for gH/gL). After incubation with the primary antibody for 30 min and fixation with 2% formaldehyde–0.2% glutaraldehyde in PBS for 15 min, the biotin-labeled secondary antibody was added at a 1:500 dilution for 30 min. After washing, streptavidin-labeled horseradish peroxidase (1:20,000) was further incubated with the fixed cells for 30 min. The peroxidase substrate was added, and the amount of cell surface staining was determined by measurement at 380 nm with a Perkin-Elmer Victor plate reader.

Monolayer binding assay.

CHO-K1 cells, grown to 80% confluence, were transfected with soluble Flag-gp42 (12) in three wells. The other three wells were transfected with a control, wt gH/gL, or mutant gH/gL by using Lipofectamine 2000, as described above. Twenty-four hours posttransfection, the supernatants of the cells transfected with soluble Flag-gp42 were collected and were centrifuged so as to collect the protein in the supernatants. The cells transfected with a control, wt gH/gL, or mutant gH/gL were washed twice with ice-cold PBS and were incubated with Flag-gp42 for 1 h at 4°C. The cells were then washed four times with ice-cold PBS and were lysed with 200 μl of SDS sample buffer. Proteins were separated on 10% Mini-Protean TGX gels (Bio-Rad) after boiling for 10 min under reducing conditions. Western blot analyses were performed using a polyclonal anti-Flag antibody (F7425; Sigma) at 1:1,000 or a polyclonal anti-GAPDH antibody (Abcam) at 1:1,000.

SDS-polyacrylamide gel electrophoresis (PAGE) migration assay for gH.

CHO-K1 cells were transfected with a control, Flag-tagged wt gH, or Flag-tagged gH-G49C together with gL and gB. After 24 h, transfected cells in 6-well plates were collected and were lysed in 200 μl lysis buffer (20 mM Tris-HCl [pH 7.4], 100 mM NaCl, 1 mM EDTA, 5 mM MgCl₂, 1% Triton X-100, and Calbiochem's 1× protease inhibitor cocktail set I). The cell lysates were cleared of debris by centrifugation. One hundred microliters of lysates was mixed with 100 μl 2× SDS loading buffer (60 mM Tris-Cl [pH 6.8], 0.2% SDS, 25% glycerol, 0.01% bromophenol blue). The samples were either left untreated or treated with 100 mM dithiothreitol (DTT) at room temperature (RT) for 10 min and were loaded onto a 7.5% Mini-Protean TGX gel (Bio-Rad) for Western blotting. An anti-Flag antibody was used for detection of the Flag-labeled gH band.

Western blotting.

Protein samples were loaded onto a 10% Mini-Protean TGX gel (Bio-Rad). After electrophoresis, proteins were transferred to nitrocellulose membranes (Schleicher & Schuell, Keene, NH). The blots were blocked with 5% nonfat dry milk in TBS (20 mM Tris-HCl [pH 7.6], 137 mM NaCl) for 2 h at RT. The blots were washed with TBS and were incubated with primary antibodies overnight at 4°C. An IRDye 800-conjugated anti-rabbit or IRDye 680-conjugated anti-mouse secondary antibody (Li-Cor Biosciences, Lincoln, NE) was added to the membranes at a dilution ratio of 1:10,000, and incubation was continued for 1 h at RT. The protein bands on the membrane were visualized with the Odyssey Fc Western blotting imager using Image Studio, version 2.0 (Li-Cor Biosciences, Lincoln, NE).

RESULTS

Amino acids on the surface of the large groove of gH/gL selectively affect epithelial and B cell fusion.

To determine the potential function of the large groove between D-I and D-II of gH/gL, we generated a panel of structure-based gH/gL mutants. The residues located in the gH/gL groove are largely surface exposed, and many surround smaller subsite pockets that could engage gp42 side chains or, potentially, the EBV epithelial receptor integrins (Fig. 1A). Mutations were designed to disrupt local interactions but to minimize any large-scale effects on domain folding. A total of 12 point mutants were made in which alanine was substituted for residues present in gH, as indicated in Fig. 1A and B. The mutants were constructed in a gH expression vector by using site-directed mutagenesis and were sequenced in order to confirm the presence of the mutated amino acid and the absence of other mutations. Effector Chinese hamster ovary cells (CHO-K1 cells) were transfected with a luciferase reporter construct under the control of the T7 promoter and with glycoproteins gB, gL, and gH (wt or mutant) for epithelial cell fusion. The transfected effector cells were then overlaid with HEK 293 target cells stably expressing T7 polymerase in order to monitor epithelial cell fusion. Luciferase expression was measured to quantify the extent of membrane fusion. We found that of the 12 gH mutants, 5 (the R152A, H154A, T174A, L207A, and Q220A mutants) had decreased fusion activity with epithelial cells (Fig. 1D). The fusion activities of the R152A and Q220A mutants were reduced to levels nearly comparable to that of the vector control (Fig. 1D). To exclude the possibility that the gH mutants with lower fusion activity had lower surface expression than wt gH, we verified cell surface expression using a cell-based enzyme-linked immunosorbent assay (CELISA). We found that except for the Q220A mutant, all the gH mutants were expressed on the cell surface at levels similar to that for wt gH (Fig. 1C). The decreased fusion activity for the Q220 mutant, but not for the other mutants, results from the lack of cell surface expression, probably caused by incorrect folding and transport of gH/gL to the cell surface. Although interpretation of the gH/gL structure is limited by the low resolution of the crystal structure, it appears that Q220 could be involved in a hydrogen bond network involving residues T201, T217, and S225. These interactions occur at the N- and C-terminal ends of an extended loop (residues 201 to 217) that projects from D-II into the D-I/D-II groove and toward D-I. Perturbation of this network and this loop region could affect the overall folding of D-II and/or the D-I/D-II interface in a manner that interferes with the proper folding and surface expression of the mutant.

Fig 1.

Mutation of EBV gH/gL identifies residues in gH that affect EBV fusion tropism. (A) Structure of EBV gH/gL, indicating domains I through IV (D-I, D-II, D-III, and D-IV) and the D-I/D-II groove. gH is colored blue in D-I, magenta in D-II, green in D-III, and yellow in D-IV. gL is colored red. (B) Close-up view of the D-I/D-II groove and sites of point mutations that were tested. Residues shown in green and with underlined labels exhibit functional effects. The view is 90° from that for panel A, looking “down” into the interdomain groove. (C and D) CHO-K1 cells were transiently transfected with a T7 luciferase plasmid, either with a vector plasmid alone (control) or with EBV gB and gL, together with wt gH or a gH mutant. (C) Expression of gH/gL or gH/gL mutants. Twenty-four hours posttransfection, 4 × 10⁴ cells were seeded into 96-well plates in triplicate. CELISA was performed with anti-gH/gL antibodies (E1D1). (D) Transfected CHO-K1 cells were overlaid with epithelial cells expressing T7 polymerase. Luciferase activity was monitored after 24 h of overlay and was normalized to that of cells with wt gH/gL levels (set at 100%). (E) CHO-K1 cells were transiently transfected with a T7 luciferase plasmid, either with a vector plasmid alone (control) or with EBV gB, gL, and gp42, together with wt gH or gH mutants. Twenty-four hours posttransfection, CHO-K1 cells were overlaid with Daudi B cells expressing T7 polymerase. Luciferase activity was monitored after 24 h of overlay and was normalized to that of cells with wt gH/gL levels (set at 100%). Data are means plus standard errors of the means for three independent experiments.

To examine fusion activity with B cells, gp42 was transfected into CHO-K1 cells together with gH/gL, gB, and T7-luciferase, and B cell fusion function was monitored by the addition of Daudi B cells expressing T7 polymerase. Interestingly, only the L207A and Q220A mutants exhibited fusion activity lower than that of the wt, in contrast to the results of fusion assays with 293 cells (Fig. 1E). The absence of fusion with the Q220A mutant parallels that observed with 293 cells and is due to lack of cell surface expression of the Q220A mutant. Since gp42 binding plays an important role in B cell fusion, the approximately 40% decrease in the fusion of the gH/gL L207A mutant with B cells may be related to a decreased level of association with gp42.

The gH-L207A/gL mutation results in reduced association with gp42.

Previously, we proposed that the large groove of gH/gL between D-I and D-II is important for gp42 binding (7, 8). However, of the five mutants with normal cell surface expression and with epithelial cell fusion activity lower than that of the wt, only one, the L207A gH/gL mutant, had lower B cell fusion activity (Fig. 1E). In order to determine whether the groove mutations altered gp42 binding, we monitored gp42 binding by using a monolayer cell binding assay. We investigated only the gH mutants with altered fusion activity. Supernatants containing soluble Flag-tagged gp42 were applied to cells transfected with gH/gL; the cells were washed extensively; and gp42 binding was determined by Western blotting. Of all the gH mutants tested, only the L207A mutant, with decreased B cell fusion activity, showed reduced gp42 association (Fig. 2A). All the other mutants with decreased fusion activity in epithelial cells showed no change in gp42 binding (Fig. 2A). We quantified the Western blot data by using Image Studio software (version 2.0; Li-Cor Biosciences, Lincoln, NE). The binding of the L207A mutant to gp42 was reduced about 50% from that of the wt (Fig. 2B), a reduction similar to that in fusion activity for this mutant (about 40%). This finding indicates that L207A, within the groove between D-I and D-II, affects gp42 binding, supporting our hypothesis that the D-I/D-II groove is in part a binding domain for gp42.

Fig 2.

The gH/gL L207A mutant has decreased association with gp42. (A) CHO-K1 cells seeded in six-well plates were transfected with a control, wt gH/gL, or a gH/gL mutant as indicated. Twenty-four hours posttransfection, the cells were washed twice with ice-cold PBS and were incubated for 1 h at 4°C with a medium containing Flag-tagged gp42 (isolated 24 h posttransfection). The cells were then washed four times with ice-cold PBS and were lysed with 200 μl of 1× SDS lysis buffer. Proteins were separated on 10% Mini-Protean TGX gels (Bio-Rad). Western blot analyses with an anti-Flag polyclonal antibody (F7425; dilution, 1:1,000) were performed to detect the Flag-tagged gp42 that bound to the transfected cells. Anti-GAPDH was used as a loading control. (B) The relative level of gp42 binding was quantified and was normalized to the level of GAPDH by using Image Studio software. The binding of wt gH/gL to gp42 was set at 100%. Data are means plus standard errors of the means for three independent experiments.

Conformational flexibility between D-I and D-II of gH/gL is important for epithelial cell fusion but not for B cell fusion.

We were intrigued by the R152A mutant, which was nearly negative for epithelial cell fusion but was expressed well and had levels of B cell fusion similar to those of wt gH/gL. To identify the defect of this mutant for epithelial cell fusion, we first determined whether binding to the epithelial receptor integrins (13, 14) was deficient for this mutant. CHO-K1 cells were transfected with either a control plasmid, wt gH/gL, or the gH R152A mutant. The cells were then lysed by freeze-thawing, and the supernatants containing solubilized gH/gL were collected. The supernatants were then overlaid on HEK 293 cells, and gH/gL binding to the cells was detected by CELISA (Fig. 3A) or by Western blotting using Flag-tagged gH/gL or gH-R152A/gL (Fig. 3B). gH-R152A/gL showed binding equivalent to that of wt gH/gL (Fig. 3A and B). These data indicate that the reduced fusion activity of gH-R152A/gL is not likely a result of reduced receptor binding.

Fig 3.

The binding of gH to epithelial cells is not decreased for the gH-R152A/gL mutant. (A) CHO-K1 cells were transiently transfected with either a control, gH/gL, or gH-R152A/gL. Soluble gH/gL and gH-R152A/gL were prepared as described in Materials and Methods and were used to overlay epithelial cells in 96 wells in triplicate. After incubation for 4 h at 4°C, the binding in 96 wells was examined by CELISA using the E1D1 monoclonal antibody, which recognizes the gH/gL complex. (B) CHO-K1 cells were transiently transfected with either a control, Flag-tagged gH/gL, or Flag-tagged gH-R152A/gL. Soluble Flag-gH/gL and Flag-gH-R152A/gL were prepared as described in Materials and Methods and were used to overlay epithelial cells in 6-well plates. After incubation for 4 h at 4°C, the cells were washed twice with ice-cold PBS and were lysed with 200 μl of 1× SDS lysis buffer. Proteins were separated on 10% Mini-Protean TGX gels (Bio-Rad). Western blot analyses were performed with a polyclonal anti-Flag antibody (F7425; dilution, 1:1,000) to detect the Flag gH/gL or Flag gH-R152/gL that bound to the HEK 293 cells.

In the crystal structure of gH/gL, R152 is located on the surface of D-II and is in close proximity to the amino acids V47, P48, and G49, located on the surface of D-I (Fig. 4A). We hypothesized that R152, located in D-II of gH/gL, may interact with D-I of gH/gL and that noncovalent interactions and flexibility between D-I and D-II of gH/gL may be important for fusion with epithelial cells. In order to test this hypothesis, we introduced cysteine mutations at V47, P48, and G49 to generate potential interdomain disulfide bonds linking D-I and D-II with C153, located across the D-I/D-II groove, potentially reducing the flexibility between D-I and D-II. Previous studies have indicated that C153 is a free cysteine and can be labeled with acrylodan and IANBD (7, 14). Our data also showed that the C153A mutant had no defect in cell surface expression or fusion activity for epithelial cells or B cells (Fig. 1C, D, and E), further confirming that C153 is a free cysteine. Following site-directed mutagenesis and confirmation by DNA sequencing, wt gH and the V47C, P48C, G49C, and R152A gH mutants were found by CELISA to be expressed similarly to wt gH/gL (Fig. 4D) and were tested for fusion function. In agreement with our hypothesis that flexibility across the D-I/D-II groove is important for epithelial cell fusion, all of the cysteine mutants had decreased epithelial cell fusion activity (Fig. 4B); the G49C mutant had the most dramatic loss of fusion function, similar to that of the R152A mutant. In contrast, all the gH mutants had fusion activity similar to that of wild type gH when fusion with Daudi B cells was analyzed (Fig. 4C).

Fig 4.

EBV gH/gL cysteine mutants have decreased cell fusion activity, similar to that of the gH-R152A/gL mutant. (A) The gH/gL residues V47, P48, and G49 in D-I and R152 and C153 in D-II are shown in stick format, with carbon atoms colored yellow, beneath a transparent surface of the protein (7). (B through D) CHO-K1 cells were transiently transfected with a T7 luciferase plasmid either with a vector plasmid alone (control) or with EBV gB and gL, together with gH or gH mutants, as indicated, with gp42 (C) or without gp42 (B and D). (B and C) Transfected CHO-K1 cells were overlaid with epithelial cells (B) or Daudi B cells (C) expressing T7 polymerase and were monitored for luciferase activity after 24 h. Luciferase activity was normalized to that of cells with wt gH/gL levels (set at 100%). Data are means plus standard errors of the means for three independent experiments. (D) Expression of wt gH/gL or gH/gL mutants. After 24 h of transfection, 4 × 10⁴ cells were seeded into 96-well plates in triplicate. CELISA was performed with the gH/gL antibodies E1D1.

Since the G49C mutant had the most dramatic phenotype, we next tested the effect of adding the reducing agent dithiothreitol (DTT) to reduce the potential disulfide bond formed between C49 and C153 in order to determine whether reduction of this bond could restore fusion function. Cells transfected with a control, wt gH/gL, or gH-G49C/gL, along with gB and the T7 luciferase plasmid, were treated for 30 min with concentrations of DTT ranging from 3.125 to 25 mM (Fig. 5). Following treatment, cells were washed twice with iodoacetamide (1 mM) to alkylate free cysteine residues. The cells were then overlaid with epithelial target cells, and fusion was monitored by luciferase activity. In agreement with the reduction of a disulfide bond between C49 and C153, we found that with increasing amounts of DTT, the epithelial cell fusion activity of gH/gL G49C was restored (Fig. 5A). In contrast, the addition of DTT did not increase wt gH/gL activity but resulted in a very modest decrease in fusion activity at higher concentrations, likely due to cellular toxicity or the reduction of other functionally important disulfide bonds (Fig. 5A). To investigate whether a disulfide bond was formed between C49 and C153 and whether it might result in altered gH/gL mobility, we compared the migration of Flag-tagged wt gH/gL with that of Flag-tagged gH-G49C/gL under nonreducing and reducing conditions. We found that under nonreducing conditions, Flag-tagged gH-G49C/gL migrated more rapidly than Flag-tagged wt gH/gL, suggesting an altered conformation for gH-G49C/gL, compatible with the presence of an additional disulfide bond. As expected, when all disulfide bonds were reduced by the addition of 100 mM DTT (Fig. 5B), wt gH/gL and the gH-G49C/gL mutant migrated similarly. To further confirm the role of disulfide bond formation, rather than disruption of function, due to the introduction of the cysteine, we generated alanine mutations at residues V47, P48, and G49 and tested the fusion activities of the mutants. We found that alanine mutation at these amino acids did not alter either epithelial cell or B cell fusion (Fig. 6). In addition, we also generated a G49S mutant, since the structures of serine and cysteine are very similar. Since serines are unable to form disulfide bonds, the G49S mutant further tests whether the absence of function is due to disulfide bond formation rather than to replacement of the glycine. The G49S mutant behaved similarly to wt gH, further confirming that flexibility between D-I and D-II is important for epithelial cell fusion, in contrast to B cell fusion.

Fig 5.

A reducing agent restores the fusion function of gH-G49C/gL. (A) CHO-K1 cells were transiently transfected with the T7 luciferase plasmid either with the vector plasmid alone (control) or with EBV gB and either wt gH/gL or gH-G49C/gL, as indicated. Transfected CHO-K1 cells were incubated with increasing amounts of DTT, as indicated, for 30 min on ice, washed twice with 1 mM iodoacetamide, and overlaid with epithelial cells, and luciferase activity was measured. Data are means plus standard errors of the means for three independent experiments. (B) CHO-K1 cells were transiently transfected with a control, Flag-tagged gH/gL, or Flag-tagged gH-G49C/gL, as indicated, along with EBV gB. After 24 h of transfection, 1 × 10⁶ cells were first lysed in 100 μl lysis buffer (20 mM Tris-HCl [pH 7.4], 100 mM NaCl, 1 mM EDTA, 5 mM MgCl₂, 1% Triton X-100, 1× protease inhibitor) and then centrifuged at 13,000 rpm for 5 min. Equal amounts of 2× 0.2% SDS sample buffer were added to the supernatants, which were then either left untreated or treated with 100 mM DTT at RT for 10 min. The samples were then analyzed on 7.5% Mini-Protean TGX gels (Bio-Rad). Data are representative of four independent experiments.

Fig 6.

Alanine or serine substitutions at the sites of the cysteine mutations do not alter the fusion function of gH/gL. CHO-K1 cells were transiently transfected with the T7 luciferase plasmid either with the vector plasmid alone (control) or with EBV gB and gL, together with wt gH or gH mutants, as indicated, with gp42 (A) or without gp42 (B and C). (A and B) Transfected CHO-K1 cells were overlaid with Daudi B cells (A) or epithelial cells (B) expressing T7 polymerase and were monitored for luciferase activity after 24 h. Luciferase activity was normalized to that of cells with wt gH/gL (set at 100%). (C) The expression of wt gH/gL or gH/gL mutants was detected by CELISA with anti-gH/gL antibodies (E1D1). Data are means plus standard errors of the means for three independent experiments.

DISCUSSION

In the current study, we investigated the functional role of the large groove located between D-I and D-II of gH/gL by using structure-based mutagenesis. We identified several mutations in this region that altered fusion function. Our previous studies found that a KGD motif located on the surface of gH/gL D-II close to the large groove between D-I and D-II plays an important role in binding gp42 and the EBV epithelial receptor (8). Thus, we hypothesized that the groove between gH/gL D-I and D-II found in the gH/gL crystal structure may also participate in gp42 binding and/or binding to an epithelial receptor. Of 12 single amino acid mutants located on the surface of the large groove between gH/gL D-I and D-II with normal cell surface expression, we found only 1, the L207A mutant, with fusion activity lower than that of wt gH/gL for B cells; this mutant was also associated with decreased gp42 binding, suggesting that the defect in B cell fusion was related to an impaired ability to bind gp42. The L207A mutant also exhibited reduced epithelial cell fusion, suggesting that L207 may also be important for binding to integrins or for a post-receptor binding event important for epithelial cell fusion.

In addition to the L207A mutant, which exhibited altered epithelial and B cell fusion, we found three mutants with normal cell surface expression and with epithelial cell fusion activity lower than that of the wt (the R152A, H154A, and T174A mutants). Each of these mutants had wt levels of B cell fusion. Supporting earlier results, our current findings suggest that the mechanism employed by gH/gL to trigger fusion involves different regions and residues of gH/gL for epithelial fusion versus B cell fusion. Previous studies with gH/gL mutants found that an E595A mutant had lower epithelial cell fusion activity than wt gH/gL but 5-fold-higher B cell fusion activity (15). In these studies, it was also found that the R597A and R607A mutants had lower fusion activity for B cells than for epithelial cells. If a serine and an arginine were inserted between amino acids 607 and 608 of wt gH, there was a higher level of B cell fusion and no epithelial cell fusion (16). Interestingly, these mutations are in gH/gL D-IV, distant from the mutations that we describe in our current study. Taken together with our data, this suggests that the mechanisms by which gH/gL integrates the binding of an epithelial receptor and the binding of gp42 to HLA class II to trigger epithelial cell fusion and B cell fusion, respectively, require different gH/gL conformations and/or flexibility.

In support of this idea, the reductions in epithelial cell fusion that we observed with our R152A, H154A, and T174A mutants indicate that the flexibility between D-I and D-II might be important for epithelial cell fusion and not for B cell fusion. To investigate this possibility, we introduced cysteine mutations at amino acids V47, P48, and G49 in gH/gL, across the D-I/D-II groove from a free cysteine at amino acid 153 adjacent to the R152A mutation. If disulfide bonds with the introduced cysteines form across the groove, flexibility should be reduced. Each of the cysteine mutants demonstrated normal surface expression of the gH/gL complex as judged by CELISA, indicating no defects in the formation and transport of the mutant gH/gL complex to the plasma membrane relative to those of the wt. All the cysteine mutants, especially the G49C mutant, showed fusion activity lower than that of the wt for epithelial cells and normal fusion activity for B cells. To verify that formation of a disulfide bond within gH/gL was likely responsible for this decreased epithelial cell fusion, increasing amounts of the reducing agent DTT were added in the fusion assays. In these experiments, fusion with gH/gL G49C was restored to levels similar to those of wt gH/gL. We also found that when V47, P48, and G49 were mutated to alanine instead of to cysteine, or when G49 was mutated to serine instead of to cysteine, there was no change in either epithelial cell or B cell fusion, indicating that substitution of the amino acid was likely not responsible for the loss of function; rather, disulfide bond formation between D-I and D-II was responsible. In support of the formation of a new disulfide bond, gH-G49C/gL migrated faster than wt gH/gL under nonreducing conditions but migrated similarly to wt gH/gL under reducing conditions. These data indicate that when structural rearrangements or access to the groove between gH/gL D-I and D-II is prevented, gH/gL-mediated epithelial cell fusion is blocked, supporting the conclusion that groove access or rearrangement is required for epithelial cell fusion but not for B cell fusion. Our G49C mutant data indicate that flexibility between D-I and D-II is important for epithelial cell fusion but not for B cell fusion.

We speculate that the triggers of gB-mediated B cell fusion and epithelial cell fusion are different. For B cell fusion, binding of gp42 to HLA class II triggers a conformational change in gp42, which leads to gB activation and fusion. The current studies indicate that a subsequent conformational change induced in gH/gL by gp42 may be different from that which leads to epithelial cell fusion, since the flexibility between D-I and D-II can be restricted. For epithelial cell fusion, a gH/gL conformational change is triggered by epithelial receptor binding, and flexibility between D-I and D-II is particularly important for this fusion activation. Further support for a conformational change in this region comes from recent site-specific fluorescence studies of gH/gL labeled at C153 binding directly to integrins, which indicated receptor-induced conformational changes in gH/gL in the process of epithelial cell entry (14). The blue shift in the fluorescence of residue 153 labeled with the environmentally sensitive probes acrylodan and IANBD was seen upon binding to the integrin, which was consistent with the movement of residue 153 into a more hydrophobic environment and with a potential conformational shift in gH/gL at the interface of D-I and D-II (14). Structural comparisons of the EBV and herpes simplex virus 2 (HSV-2) gH/gL complexes also show large conformational differences in D-I/D-II orientations (7, 17), consistent with the possibility that conformational changes in this region may occur upon receptor binding, providing a signal for gB activation. Such conformational changes may be similar to the changes observed upon receptor engagement of gD or gp42 (18), but it is not likely that the changes are comparable to the conformational changes found in fusion proteins in which domains are dramatically refolded or rearranged. However, our data may suggest the existence of gH/gL pre- and postfusion forms associated with receptor binding. Based on the finding that cross-linking of D-I and D-II of gH/gL blocks epithelial cell fusion and not B cell fusion, it may be possible to find small-molecule inhibitors or monoclonal antibodies that stabilize the interface in this region to prevent infection of epithelial cells by EBV without altering B cell infection. Previous studies have shown that the binding site of gp42 with gH/gL maps to a minimal peptide of 33 amino acids corresponding to residues 44 to 61 and 67 to 81 of gp42 (19). However, the corresponding binding site on gH/gL is not well described. Our previous studies indicate that the KGD motif of gH/gL may directly bind to amino acids 62 to 66 of gp42 (8). Our current studies suggest that L207 of gH/gL may also play a role in gp42 binding, either by directly binding to gp42 or by altering the structure of gH and decreasing the affinity of the KGD motif for gp42. Thus, our current studies, as well as our previous studies analyzing the gH/gL KGD motif, support the conclusion that this region of gH/gL is key to gH/gL fusion function.