Isolating the Epstein-Barr Virus gp350/220 Binding Site on Complement Receptor Type 2 (CR2/CD21)*

Complement receptor type 2 (CR2/CD21) is essential for the attachment of Epstein-Barr virus (EBV) to the surface of B-lymphocytes in an interaction mediated by the viral envelope glycoprotein gp350. The heavily glycosylated structure of EBV gp350 has recently been elucidated by x-ray crystallography, and the CR2 binding site on this protein has been characterized. To identify the corresponding gp350 binding site on CR2, we have undertaken a site-directed mutagenesis study targeting regions of CR2 that have previously been implicated in the binding of CR2 to the C3d/C3dg fragments of complement component C3. Wild-type or mutant forms of CR2 were expressed on K562 cells, and the ability of these CR2-expressing cells to bind gp350 was measured using flow cytometry. Mutations directed toward the two N-terminal extracellular domains of CR2 (SCR1-2) reveal that a large contiguous surface of CR2 SCR1-2 is involved in gp350 binding, including a number of positively charged residues (Arg-13, (Arg-28, (Arg-36, Lys-41, Lys-57, Lys-67, and Arg-83). These data appear to complement the CR2 binding site on gp350, which is characterized by a preponderance of negative charge. In addition to identifying the importance of charge in the formation of a CR2-gp350 complex, we also provide evidence that both SCR1 and SCR2 make contact with gp350. Specifically, two anti-CR2 monoclonal antibodies, designated as monoclonal antibodies 171 and 1048 whose primary epitopes are located within SCR2, inhibit binding of wild-type CR2 to EBV gp350; with regard to SCR1, both K562 cells expressing an S15P mutation and recombinant S15P CR2 proteins exhibit diminished gp350 binding.

Primary infection normally occurs within the first few years of life and is usually asymptomatic, but if infection is delayed until adolescence or later ages, then it may present as infectious mononucleosis. Long term carriage of EBV has been implicated in the development of a number of other more serious disease states including B cell lymphomas, nasopharyngeal carcinoma, and gastric carcinoma (for review, see Refs. [1][2][3][4]. EBV has also been linked as a potential etiologic agent in the development of a number of autoimmune conditions including systemic lupus erythematosus, rheumatoid arthritis, and multiple sclerosis (5)(6)(7). EBV predominantly infects two major cell types: B lymphocytes and epithelial cells. Generally, more is known about the mechanism of entry of EBV into B lymphocytes than into epithelial cells. Preferential binding to B-lymphocytes is primarily mediated by the binding of the EBV viral envelope protein gp350 to complement receptor type 2 (CR2/CD21) on the surface of B cells (8 -14) and through the binding of a second glycoprotein gp42 to human leukocyte antigen class II molecules (15,16) which triggers fusion with the host cell in a process mediated by three additional viral glycoproteins, gB, gH, and gL (17)(18)(19). Additional viral ligands have also been implicated in the attachment and invasion of cells by EBV independent of the gp350/CR2 and gp42/human leukocyte antigen II pathways. EBV has been demonstrated in vitro to infect CR2(Ϫ) cells, and gp350 knock-out mutant forms of EBV have also been shown to infect primary B-lymphocytes, lymphoid cell lines, and epithelial cell lines that are susceptible to wild-type EBV infection, albeit with a significantly reduced efficiency of infection (20). EBV gp350/220 is an extensively glycosylated polypeptide (907 residues) that is expressed in two alternatively spliced forms of ϳ350 and 220 kDa (21) and has been identified as the dominant protein in the extracellular virus envelope (22). The N-terminal 470 residues of gp350 have been shown to be essential for the binding interaction with CR2 (14). In addition, monoclonal antibodies generated against gp350 have been shown to effectively inhibit infectivity of peripheral B lymphocytes with EBV (23), and some forms of this protein have been shown to act successfully as a vaccine against EBV-associated diseases in animal models (24,25). The three-dimensional structure of the CR2 receptor binding domain of gp350 has recently been determined by x-ray crystallography (26). Three distinct domains were identified, each comprising an anti-parallel ␤-barrel structure and adjoined by two long linker regions, each of 11 amino acids. The domains are tightly packed against each other, forming a distinctive L-shaped arrangement that is almost uniformly glycosylated. Site-directed mutagenesis targeting one of the few glycan-free areas of gp350 has identified the CR2 binding site to be located within a negatively charged region of this molecule, with mutant forms of gp350 exhibiting diminished or negligible ability to bind CR2 (26). Convincing evidence that this is indeed the correct receptor binding site is provided by the fact that the mutations targeting the CR2 binding site also disrupt the ability of gp350 to recognize its major neutralizing monoclonal antibody, 72A1 mAb (26,27). The epitope for 72A1 mAb within gp350 has been identified as one of the key areas involved in EBV gp350 binding to host cells because 72A1 mAb can successfully inhibit EBV binding and invasion of B-lymphocytes, Raji cells, monocytes, neutrophils, and T-cells (14, 28 -30). Also, a recent peptide mapping study by Urquiza et al. (31) has identified three gp350 regions that are involved in EBV binding to B-lymphocytes, two of which directly overlap with the CR2 binding site identified in the crystal structure of EBV gp350.
CR2 is a ϳ145-kDa transmembrane glycoprotein expressed mainly on B-lymphocytes, follicular dendritic cells, and some T-cell subtypes. On B-cells CR2 either forms part of a CR2/ CD81/CD19 complex or is otherwise associated with CR1 and plays an essential role in cell activation and the initiation of normal immune responses to pathogens (32)(33)(34)(35). Interaction of foreign antigen that is coated by covalently attached C3d with CR2 results in a cell-signaling event occurring through CD19. This significantly lowers the signaling threshold for B-cell activation via a signal transduction cascade. CR2 also plays a critical role in the development of humoral immune responses to Tdependent antigens. Thus, CR2 acts as a bridge between the adaptive and innate immune systems (36). CR2 is a member of the structural family of C3/C4 receptor and regulatory proteins known as the regulators of complement activation (RCA family). Members of this family are characterized by the presence of short repeating domains of ϳ70 amino acids, known as SCR or CCP modules. Each of these compact units contains a number of conserved residues including four cysteines and an invariant tryptophan residue. The conserved cysteines form a pattern of disulfide bridges that connect Cys-1-Cys-3 and Cys-2-Cys-4. The modular composition of CR2 is well known and consists of a 15-or 16-SCR extracellular domain, a 24-amino acid transmembrane domain, and a short 34-amino acid intracellular C-terminal tail. In addition to EBV gp350, high affinity ligands for CR2 include the proteolytic fragments of complement component C3, C3d, C3dg, and iC3b (37,38), the low-affinity IgE receptor, CD23 (39,40), and interferon ␣ (41,42). Peptide and epitope mapping experiments have indicated that all known CR2 ligands bind within the first two N-terminal SCR domains (SCR1-2) at overlapping binding sites (43)(44)(45), although an additional glycosylation-dependent interaction with CD23 also involves SCR5-8 (40).
Recently, we have utilized a site-directed mutagenesis strategy to examine the binding interaction between C3dg and fulllength CR2 (46). This work involved expressing wild-type or single site mutant forms of CR2 on the surface of K562 erythroleukemia cells and then measuring the relative binding affinities of these cells for recombinant C3dg-biotin tetramers using multicolor flow cytometry. Mutagenesis targets for this study were directed by previously reported solution phase studies of the CR2-C3d interaction and by the available co-crystal structure of the CR2 SCR1-2-C3d complex (43)(44)(45)(47)(48)(49)(50). Our data revealed that in addition to the CR2-C3d interface observed in the co-crystal structure, which involved only SCR2 of CR2, a number of positively charged residues on the SCR1 domain were also essential for a significant CR2-C3dg interaction to occur. Several of these residues had not been implicated in CR2-ligand binding by previous experimental studies, although a recent study utilizing theoretical electrostatic potential and apparent pK a calculations observed that the association between CR2 and C3d was dependent not only on the apparent CR2 SCR2-C3d interface observed in the structural elucidation but also on the overall respective net charges of CR2 SCR1-2 and C3d (51). The binding interactions between CR2 and C3d and between CR2 and gp350 have previously been indicated to occur at overlapping binding sites (43)(44)(45). However, a number of intriguing differences between these binding sites or the nature of the binding interaction between CR2 and these ligands must occur. Murine CR2 SCR1-2 shares a ϳ61% sequence identity with human CR2 SCR1-2, and although murine CR2 can bind human C3d and iC3b with an equal affinity to the human form, it cannot bind EBV (43,47). Data reported by Martin et al. 52 and Prota et al. 53 have proposed that two areas on CR2 are critical in the binding interaction with EBV gp350. Substitutions among residues 8 -15 within the first inter-cysteine region of SCR1 have suggested this area is a structural determinant in gp350 binding; murine CR2 contains a N-linked glycosylation site at position 66 within the linker region connecting SCR1 and SCR2, and this has been proposed to sterically inhibit the interaction between CR2 and EBV gp350. To gain a better understanding of the binding interaction between CR2 and its ligand EBV gp350 we have utilized a combination of site-directed mutagenesis and epitope mapping to determine those residues on CR2 that are likely to play a role in the adsorption of EBV to B-lymphocytes.

EXPERIMENTAL PROCEDURES
Cloning and Expression of EBV gp350-EBV genomic DNA was extracted from previously obtained cell supernatants of the marmoset HB95-8 leukocyte cell-line (American type Culture Collection) using a QIAamp UltraSens virus kit (Qiagen). DNA corresponding to residues 1-470 of EBV gp350/220 was PCRamplified using the primers 5Ј-GCG GCC CAG CCG GCC GAG GCA GCC TTG CTT G-3Ј (incorporating an Sfi site) and 5Ј-G ATA GTT TAG CGG CCG CAT TCT TAT GGT GGA TAC AGT GGG-3Ј (incorporating a NotI site). In addition, a fragment of the Escherichia coli biotin carboxyl carrier protein corresponding to residues 70 -156 was also PCR-amplified using the primers 5Ј-CGG GGG GCC CCC CAT GGA AGC GCC AGC AGC-3Ј and 5Ј-CTG GGG CCC CTA CTC GAT CGA GAC CAG C-3Ј both of which incorporate a DraII site for cloning. PCR fragments were ligated into the pSecTag2/Hygro B eukaryotic expression vector (Invitrogen), which encodes an Myc epitope and a hexahistidine tag at the 3Ј-end of the inserted DNA. cDNA for the gp350-biotin construct was sequenced by the University of Colorado Cancer Center Sequencing Core (Denver, CO) to confirm the insert was in-frame with the plasmid DNA and also that it contained the designated sequences for the gp350 and biotin carboxyl carrier protein proteins outlined above. Plasmid DNA was subsequently transfected into human embryonic kidney 293f freestyle cells (Invitrogen) for soluble expression of recombinant EBV gp350 into the media. The recombinant EBV gp350 was concentrated and concurrently exchanged into a 20 mM sodium phosphate buffer, pH 7.4, containing 10 mM imidazole and 0.5 M NaCl and applied to a Ni 2ϩ charged HisTrap column (GE Healthcare). Bound protein was eluted with a 0.5 M imidazole gradient, concentrated, and subsequently purified by gel filtration on a Hiload 26/60 Superdex 200 prep grade column using an AKTA fast protein liquid chromatography system (GE Healthcare). After purification the EBV gp350 was biotinylated (EBV gp350-biotin) using biotin ligase (Avidity) according the manufacturer's instructions and then conjugated to Phycoerythrin-NeutrAvidin (Molecular Probes), generating fluorochrome tagged-gp350 monomers for flow cytometric binding analysis.
Expression of Wild-type, N11A, R13A, S15P, R28A, R36A, K41A, K57A, K67A, and R83A CR2 SCR1-2 Recombinant Proteins in E. coli-DNA corresponding to residues 1-133 of wildtype CR2 (SCR1-2) was PCR-amplified using the primers 5Ј-CCG GAA TTC CGG ATT TCT TGT GGC TCT CCT-3Ј incorporating an EcoRI site and 5Ј-CCC AAG CCT GGG TCA TCA CTC GAG AGG GAA AAC ACT-3Ј incorporating two stop sequences and a HindIII site after the codon correspond-ing to residue 133. Resulting PCR fragments were ligated into the prokaryotic expression vector pMAL-p2x (New England Biolabs) as previously described (26), which encodes a maltosebinding protein (MBP) tag at the 5Ј-end of the inserted DNA. This plasmid includes the malE gene with its signal sequence allowing the export of fusion proteins to the bacterial periplasm, facilitating folding and disulfide bond formation to take place. Plasmid DNA was subsequently transformed into E. coli BL21 cells, and wild-type recombinant MBP-CR2 SCR1-2 was produced as follows. Ampicillin-resistant colonies were used to start overnight cultures that were expanded to 1-3 liters and grown at 37°C until an A 600 of 0.3 was obtained. Cultures were induced with 0.3 mM isopropyl-␤-D-thiogalactoside at 30°C for 4 h before harvesting by centrifugation. Harvested pellets were resuspended in a column buffer comprising 20 mM Tris-HCl, pH 7.4, 0.2 M NaCl, 1 mM EDTA and lysed by sonication. The lysate was clarified by centrifugation and then applied to an amylose resin column. Bound MBP-CR2 SCR1-2 was eluted from the column using column buffer containing 10 mM maltose. Finally, the MBP-CR2 SCR1-2 was purified by size exclusion chromatography using the protocol described above for EBV gp350.
Flow Cytometry-Flow cytometric experiments were carried out using K562 erythroleukemia cells transfected with fulllength wild-type or mutant human CR2. Binding analyses were carried out using gp350-biotin. For each condition, 5 ϫ 10 5 human CR2-transfected K562 cells were first incubated with fluorescein isothiocyanate (FITC)-conjugated anti-CR2 HB5 mAb at 1 g/ml on ice for 1 h. The primary epitope for HB5 has been identified within the N-terminal SCR3-4 extracellular domains of CR2 and, accordingly, does not interfere with ligand binding. During this incubation 100 l of gp350-biotin monomers in PBS, 0.1% bovine serum albumin , 0.01% sodium azide were prepared for each condition by adding the appropriate amount of recombinant gp350-biotin and 0.4 g of PE-conjugated NeutrAvidin (Molecular Probes) and incubating at room temperature for 30 min. gp350-biotin concentrations used were 0.5g, 0.25, 0.125, 0.0625, and 0.03125 g. After washing of the FITC-stained K562 cells, 100 l of monomeric PEconjugated gp350-biotin were added to each sample of cells and incubated for 30 min on ice. After washing, the cells were fixed and analyzed by multicolor flow cytometry in the University of Colorado Cancer Center Flow Cytometry Core Facility (Denver, CO).
Cells were divided into high, medium, and low CR2 expression using 10 1 as the lower limit and gating on the intermediate 18% of CR2 expressing cells (FITC-positive). Monomeric gp350-biotin binding was determined by PE mean channel fluorescence. A minimum of four separate experiments was car-ried out for each mutation. The whole cell populations and/or the normalized gp350-biotin binding data for each of the intermediate wild-type and mutant populations are given.
Anti-CR2 mAb Inhibition of EBV gp350 Binding to Fulllength CR2-Wild-type CR2-expressing cells were incubated with FITC-conjugated anti-CR2 HB5 mAb at a concentration of 1 g/ml on ice for 1 h. After this initial incubation, cells were washed, and 100 l of PBS solution containing one of the anti-CR2 monoclonal antibodies, 171, 1048, or 629 mAb, was added to the cell mix at a concentration of 0.5 g/ml and allowed to stand on ice for 1 h. After this period cells were again washed, and 100 l of EBV gp350-biotin monomers conjugated to Phycoerythrin-NeutrAvidin was added to the CR2-expressing K562 cells as described above. Cells were then fixed, and EBV gp350-biotin binding was measured by multicolor flow cytometry.
Production of the Anti-gp350 Monoclonal Antibody 72A1-The HB-168 hybridoma cell line was obtained from the ATCC. The anti-gp350 monoclonal antibody 72A1 was subsequently obtained from the spent culture medium of hybridoma cells grown in RPMI 1640 supplemented with 2 mM L-glutamine and adjusted to contain 1.5 g/liter sodium bicarbonate, 4.5 g/liter glucose, 10 mM HEPES, 1.0 mM sodium pyruvate, and 10% fetal bovine serum. The antibody was then purified by affinity chromatography using protein G-Sepharose 4 Fast Flow resin (GE Healthcare) according to the manufacturer's instructions. The resulting purified 72A1 mAb was exchanged into PBS, pH 7.4, and finally concentrated to give a stock solution containing 1 mg/ml antibody as determined by UV-visible spectrophotometry and stored at Ϫ20°C until required.
EBV Pull-down Binding Assay-The B95-8 strain of EBV was obtained from the clarified culture medium of marmoset B-lymphocytes (GM07404) obtained from Coriell Cell Repositories. EBV-containing supernatant was subsequently passed through a 0.88-m (pore size) filter and then ultracentrifuged at 25,000 ϫ g at 4°C for 4 h. Virus-rich pellets were then resuspended in 1/20 original volume PBS, pH 7.4. 1 ml of the resulting concentrated EBV solution was incubated with 10 g of recombinant wild-type or mutant (N11A, R13A, S15P, R28A, R36A, K41A, K57A, K67A, and R83A) MBP-CR2 SCR1-2 at room temperature for 30 min. Subsequent to this stage, 40 l of a pre-equilibrated 50% slurry of amylose beads were added to the MBP-CR2 SCR1-2/EBV solution, and the mixture allowed to incubate for another 20 min. The amylose beads were then pelleted by centrifugation at 10,000 ϫ g for 30 s, and the supernatant containing unbound virus was aspirated off. The amylose beads were subsequently washed a total of three times in PBS. After washing, 30 l of an elution buffer (20 mM Tris-HCl, pH 7.4, 0.2 M NaCl, 1 mM EDTA, 10 mM maltose) was added to the beads, and after 10 min, the beads were pelleted a final time and the elution buffer containing recombinant wild-type or mutant MBP-CR2 SCR1-2 and EBV virus that had been bound was removed for SDS-PAGE and Western blotting.
Samples obtained from the pulldown experiment were diluted with loading buffer (Invitrogen), and any virus present was inactivated by heating at 80°C for 10 min. For detection of wild-type and mutant forms of MBP-CR2, samples were applied to a 10% Bis-Tris gel (Invitrogen), and levels of CR2 in the eluent were detected by Coomassie staining. To detect EBV, samples were first separated on 3-8% Tris acetate gels (Invitrogen) by electrophoresis before being transferred to nitrocellulose membranes. After blocking overnight using a 10% milk solution, the membranes were washed in PBS-Tween (0.5%) and then incubated with the anti-gp350 mAb, 72A1, at a concentration of 1 g/ml. Upon further washing, the membrane was incubated with goat anti-mouse heavy and light chain peroxidase-conjugated IgG at a concentration of 0.5 g/ml (Jackson ImmunoResearch Laboratories), washed again, and finally developed using ECL reagents (GE Healthcare).

RESULTS
Anti-CR2 SCR1-2 Monoclonal Antibody Inhibition-Wildtype CR2 expressing K562 erythroleukemia cells that were initially incubated with the FITC-conjugated anti-CR2 monoclonal antibody HB5 and subsequently incubated with gp350-biotin PE-NeutAvidin at a range of gp350-biotin concentrations ranging from 0.5 to 0.0313 g exhibited a typical dose-dependent binding curve as measured by dual-color flow cytometry (Fig. 1, A and E). However, when cells were preincubated with the anti-CR2 SCR1-2 monoclonal antibodies 629, 171, and 1048 before adding gp350-biotin, the latter two antibodies were found to exhibit significantly compromise gp350biotin binding (Fig. 1, B-E). Preincubation with 171 mAb in particular inhibited binding of gp350-biotin to wild-type CR2 with an mean fluorescence intensity (MFI) of only ϳ3%, whereas 1048 mAb demonstrated an MFI of 19% relative to wild-type CR2 (at a concentration of 0.5 g of gp350-biotin/0.4 g of PE-NeutrAvidin) ( Table 1). No decrease in gp350-biotin binding was observed as a result of preincubating K562 cells with 629 mAb. The primary epitopes for 171 and 1048 mAbs have previously been identified on the SCR2 extracellular domain of CR2, specifically 86 TPYRH 90 for 171 mAb and 111 WCQANNMW 118 for mAb 1048, as determined by peptide mapping (54). It should be noted that the epitope for 171 mAb on SCR2 partly overlaps with that of the C3d binding site on SCR2, which involves residues 80 YKIRGSTP 88 , as revealed by the co-crystal structure of the CR2-C3d complex (50). These data coupled with previous studies are suggestive that the gp350 binding site is located immediately adjacent to, or directly overlapping with both the 171 mAb and C3d binding sites. In contrast, the 1048 mAb primary epitope is geographically located on the opposite side of the CR2 molecule relative to the 171 mAb and the C3d binding sites. Only a single residue, Trp-118, which forms part of the interface between SCR1 and SCR2, is located on the same visage as both the C3d and 171 mAb sites. It is likely, therefore, that 1048 mAb inhibits the interaction between CR2 and gp350 via steric or allosteric means. A monoclonal antibody, which was not part of the current study, designated as 944 mAb has previously been found to inhibit the interaction between CR2 and its ligands, and this monoclonal antibody shares most of its primary epitope ( 110 VWCQANNM 117 ) with that of 1048 mAb (54).
SCR2-C3d Interaction Site and the Linker Region Connecting SCR1 and SCR2-The SCR2 interface with C3d is dominated by main-chain interactions, often mediated by water molecules. From the CR2 perspective, only Arg-83 appears to contribute a significant side-chain interaction to C3d binding. In the CR2-C3d co-crystal structure the side chain of Arg-83 is inserted into an anion hole formed by the carbonyl atoms of residues Ile-115, Leu-116, Glu-117, and Gln-119. To investigate the degree of overlap between the C3d and gp350 binding sites on CR2, we generated R83A and R83E CR2 mutants expressed on the surface of K562 cells and assessed their capacity to interact with gp350-biotin. Both of these mutants exhibited decreased ability to bind gp350-biotin relative to wild-type CR2 with an MFI of ϳ33 and 41%, respectively (Table 1, Fig. 2). These data taken in context with the monoclonal antibody inhibition data  In the CR2-C3d co-crystal structure Arg-83 was identified as providing the only substantial side-chain contribution from the CR2 side of the CR2-C3d interaction. The numbering system used within this article for CR2 is based on the previously reported amino acid sequence for the mature protein (55).  53 has implicated this linker region in playing a significant role in EBV binding. Specifically, the sequence 66 NKY 68 , as found in human CR2, when mutated to a sequence identical to the corresponding murine sequence of 66 NKT 68 , causes glycosylation of Asn-66. Significantly, murine CR2 is unable to bind EBV or gp350. Subsequent introduction of this glycan moiety to human CR2 either in conjunction with mutations targeting residues 8 PILN-GRIS 15 of the first inter-cysteine region of SCR1 or in their absence has been demonstrated to cause a substantial decrease in gp350 binding to CR2 (52,53). To test the importance of the SCR1-SCR2 linker region in binding to gp350biotin, we generated four point mutations targeting a total of three residues within the linker region. These mutations were Y64A, K67A, K67E, and Y68A. Cells transformed with plasmid DNA containing an additional Y68T mutation generated with the goal of causing Asn-66 to be glycosylated failed to sort with streptavidin-coated magnetic beads after being incubated with biotinylated anti-CR2 HB5 mAb and were, accordingly, rejected from the study. Whole cell populations of K562 cells expressing Y64A, K67A, K67E, and Y68A mutants are shown (Table 1, Fig. 3). Neither Y64A nor Y68A mutants demonstrated any pronounced decrease in gp350-biotin binding affinity relative to wildtype CR2. Binding curves for gp350biotin binding to Y64A and Y68A mutants over the full range of gp350-biotin concentrations are shown in Fig. 3E. However, alanine and opposite charge substitutions targeting Lys-67, also located within the linker region (K67A and K67E), showed a decreased capacity for binding gp350-biotin relative to wild-type CR2 with MFI values of 42 and ϳ23% for K67A and K67E mutants, respectively (Fig. 3F). Interestingly, in the available structures of CR2, Lys-67 is oriented toward the same face on CR2 as Arg-83 and the 171 mAb epitopes, whereas the side chains of Tyr-64 and Tyr-68 are oriented behind and away from this region (50,53). 8 PILNGRIS 15 Region of SCR1-The region within the first inter-cysteine area of SCR1, previously defined as one of the major epitopes of the inhibitory monoclonal antibody OKB7 and located within residues Pro-8 -Ser-15, has been implicated in a number of studies as being essential for the ligation of CR2 to C3d/C3dg and also to EBV/gp350 (44 -47, 52). In particular, work by Martin et al. 52 demonstrated it was possible to get murine forms of CR2 to bind EBV when residues 8 -15 ( 8 EVKNARKP 15 ) contained a single point mutation of P15S. Constructs of this P15S mutant form of murine CR2 were able to bind EBV irrespective of whether or not the N-glycan moiety at Asn-66 was removed by site-directed mutagenesis (52). To investigate the role that residues 8 -15 play in binding gp350, we generated a total of 6 mutations targeting 5 residues within or in close spatial proximity to the 8 PILNGRIS 15 region of SCR1. Mutations generated were N11A, R13A, R13E, S15P, Y16A, and S32A. Whole cell populations for the four mutations within the Pro-8 -Ser-15 region are shown in Fig. 4, A-D. In addition, dosedependent binding curves for all of the mutants over the range of 0.5-0.0313 g of gp350-biotin are shown in Fig. 4, E-H. These data reveal that a number of residues within the OKB7 primary epitope are essential for binding gp350. R13A and R13E mutants both demonstrate greatly decreased binding affinities relative to wild-type CR2, with binding levels of ϳ7%. In addition, when Ser-15 was mutated to its corresponding murine amino acid, Pro-15, a minimal value of ϳ5% was observed for gp350-biotin binding. Of the other mutants studied, N11A and S32A both demonstrated similar binding curves of gp350-biotin to that of wild-type CR2, whereas Y16A showed slightly reduced binding with an MFI of ϳ67% relative to wild-type CR2.
Because Arg-13 has not previously been implicated in the association of gp350 with CR2 and other studies have postulated that the linker region between SCR1 and SCR2 is the major site of attachment for gp350, we decided to generate a limited number of additional mutations targeting residues Arg-13 and Ser-15 and expressed using an E. coli system. Equal quantities of purified wild-type, R13A, and S15P forms of MBP-CR2 SCR1-2 were assessed by SDS-PAGE, and their capacity to bind plate-bound gp350biotin or plate-bound C3d was subsequently measured by ELISA analysis (Fig. 5). Both R13A and S15P mutants demonstrated levels of gp350 ligand binding consistent with that seen using the K562 mutant analysis. R13A exhibited insignificant gp350-biotin binding at all of the MBP-CR2 SCR1-2 concentrations used in the ELISA (2-0.016 g/ml), whereas S15P also exhibited a considerably reduced capacity to bind gp350-biotin, especially at concentrations below 0.5 g/ml. It is not known whether the decrease in binding affinity associated with the S15P mutation is the result of a localized structural rearrange- ment disrupting a binding interaction with gp350 or whether Ser-15 itself is directly involved in gp350 ligation. However, these data are highly suggestive that SCR1 undergoes a direct interaction with the surface of gp350 especially around Arg-13. In contrast to these data the R13A binding curve for platebound C3d demonstrated appreciable, although reduced binding at levels similar to those seen in our previous CR2-C3dg tetramer binding studies (46), whereas the S15P mutant exhibited C3d binding similar to that seen for wild-type CR2. Comparison of the CR2-gp350 and the CR2-C3d ELISA binding data suggest that there are likely to be intrinsic differences between the binding sites on CR2 for gp350 and for C3d.
Positively Charged Surface of SCR1-Human CR2 SCR1-2 contains a number of conserved positively charged residues, primarily concentrated on a single external face of SCR1. We have previously used site-directed mutagenesis to highlight the importance that positive charge plays in the ligation of CR2 to C3d (46). To evaluate the role that this same region may also play in the binding of EBV to CR2, we decided to utilize our previously generated alanine and opposite charge substitutions targeting residues Arg-28, Arg-36, Lys-41, Lys-50, and Lys-57 on SCR1 for our gp350-biotin binding analysis. Data for R28A, R28E, R36A, R36E, K41A, K41E, K50A, K50E, K57A, and K57E are given (Table 1, Fig. 6). Three of these pairs of mutations in particular appeared to demonstrate decreased gp350-biotin binding; that is, R28A and R28E, R36A and R36E, and K41A and K41E all demonstrated less than 30% gp350-biotin binding relative to wild-type CR2 for the alanine substitutions and less than or equal to 12% for the glutamic acid point mutations. Each of these three sets of mutations demonstrated a substantial decrease in gp350-biotin binding with the transition from the wild-type arginine or lysine positively charged side-chain group to a neutral alanine side-chain followed by a further decrease in binding as a negatively charged glutamic acid side chain was introduced. This stepwise decrease in gp350-biotin binding suggests that electrostatic interactions involving residues Arg-28, Arg-36, and Lys-41 in conjunction with residue Arg-13 from the previously defined OKB7 mAb epitope described above, as shown in Fig.  7A, are likely to play an important role in the association of gp350 to CR2. The remaining two pairs of mutants used in this analysis, pair K50A and K50E and pair K57A and K57E, also show decreased gp350-biotin binding relative to wild-type CR2 SCR1-2, but to a lesser degree. K50A and K50E mutants demonstrated an MFI of ϳ59 and ϳ51%, respectively, whereas K57A and K57E exhibited an MFI of ϳ37 and ϳ64%, respectively. The data for the Lys-57 mutations are unusual in that the opposite charge point mutant has a demonstrably higher binding affinity than the alanine substitution mutant and as such should be treated with caution. However, in general it would appear that Lys-50 and Lys-57 are less critical to the CR2-gp350 interaction either from the perspective of being proximal rather than central to a gp350 binding site or as weaker contributors to a long-range electrostatic interaction leading up to the formation of an encounter complex.
Intact EBV Binding-Our cell binding data, summarized in Table 1 and Fig. 7, was used to direct the generation of further mutant forms of recombinant MBP-CR2 SCR1-2 (N11A, R13A, S15P, R28A, R36A, K41A, K57A, K67A, and R83A) expressed in E. coli. With the exception of N11A, which was made as a control mutation, all of the other mutants generated were predicted to exhibit compromised ability to bind intact EBV. The Western blot assay used to measure the capacity of mutant forms of MBP-CR2 SCR1-2 to bind EBV demonstrated that indeed the majority of the mutations generated displayed compromised binding capacity when incubated with concentrated virus (Fig. 8). Only N11A demonstrated binding levels similar to that of wild type. These data are indicative that our cell binding data mapping a potential gp350 binding interface on CR2 is also likely to be relevant to the interaction of CR2 with intact EBV.

DISCUSSION
We have used a large number of instructive full-length CR2 mutants expressed on the surface of K562 erythroleukemia cells to map out those residues within the two N-terminal extracellular SCR domains that are essential for the attachment of the EBV envelope gp350 protein to B-lymphocytes. The objective of this study was to complement the recent three-dimensional structural determination of gp350, which identified the major FIGURE 5. A, SDS-PAGE of wild-type MBP-CR2 SCR1-2, R13A MBP-CR2 SCR1-2, or S15P CR2 SCR1-2 recombinant proteins in which 5 g of sample has been loaded onto the gel matrix. B, ELISA demonstrating wild-type (WT), R13A MBP-CR2 SCR1-2, and S15P MBP-CR2 SCR1-2 binding to plate-bound gp350. C, ELISA demonstrating wild-type, R13A MBP-CR2 SCR1-2, and S15P MBP-CR2 SCR1-2 binding to plate-bound C3d. The average and S.E. of the normalized values relative to wild-type gp350 binding are given.
CR2 binding site within this molecule to be a glycan-free, negatively charged region within the N-terminal 470 residues (26). A summary of the site-directed mutagenesis binding data and the primary inhibitory monoclonal antibody epitopes used in this study is shown (Fig. 7, B and C). Both of the N-terminal extracellular SCR domains of CR2 appear to be essential in the binding interaction with gp350. Our epitope mapping data confirmed previous studies which demonstrated that the anti-CR2 monoclonal antibodies 1048 mAb, and particularly, 171 mAb, significantly inhibit binding of gp350 to CR2 (54). The primary epitope for 171 mAb overlaps directly with the previously determined C3d/ C3dg ligand binding site on SCR2, and these data along with the available R83A and R83E mutant-gp350-biotin binding curves allow us to propose that SCR2 plays an important role in the ligation of CR2 to gp350.
A number of studies have highlighted the ability of human CR2 to bind EBV or gp350, whereas the murine form of the protein is unable to do so. This decreased capacity to bind virus or viral protein on the part of murine CR2 has been linked to differences in two major areas of CR2; that is, the 8-amino acid linker region connecting SCR1 and SCR2 comprising residues 63 EYFNKYSS 70 for the human protein and 63 EESVNKTIS 70 for the murine protein and within residues 8 -15 of SCR1, which in human CR2 delineates one of the major epitopes for the no longer available anti-CR2 monoclonal antibody OKB7 mAb. The linker region adjoining the two SCR domains in murine CR2 differs from the human form of the protein with the presence of an additional N-glycan moiety at position 66. This glycan has been shown to interfere with EBV binding by studies in which receptor bearing cells expressing human CR2 containing a Y68T point mutation exhibited compromised EBV binding (53). Although we were unable to generate our own stable population of K562 cells expressing an identical Y68T mutant, we did successfully produce valuable gp350-biotin binding data of Y64A, K67A, K67E, and Y68A point mutants within the human CR2 linker region. Our data indicate that substitution of residues on the outside of the linker region (Y64A and Y68A) which are oriented away from the tight V-shaped structure resolved in the available x-ray structures of CR2 do not appear to have any deleterious effect on gp350-biotin binding. However, alanine and opposite charge substitutions targeting Lys-67, which is oriented toward the tightly packed structure, resulted in greater decreases in gp350-biotin binding (Fig. 7, B and C). Within the OKB7 mAb epitope of CR2, we were able to analyze gp350biotin binding for a large number of mutations directed at and immediately proximal to this area. R13A, R13E, and S15P mutant forms of CR2 expressed on K562 cells all exhibited insignificant levels of gp350-biotin binding. Other mutations targeting this Pro-8 -Ser-15 area all exhibited gp350-biotin curves binding identical to, or only slightly reduced from wildtype CR2. Using recombinant wild-type and mutant forms of MBP-CR2 SCR1-2 produced using a prokaryotic expression system, we were able to replicate our cell binding data for the R13A and S15P mutants by ELISA. Although we are unable to state if the negligible gp350 binding observed in this study for the S15P mutant is the result of a localized structural perturbation, resulting in an altered gp350 binding site or if Ser-15 is actually involved in a binding interaction with gp350 itself, the combined mutagenesis data available for Ser-15 and Arg-13 make it likely that SCR1 undergoes a contact interaction with gp350. Our MBP-CR2 SCR1-2 ELISA methodology when extended to probe C3d ligand binding also allowed us to isolate residues on CR2 that are essential for gp350 binding but not for that of C3d.
One of the most striking features of our data is the importance that charge plays in the attachment of gp350 to CR2. In addition to residues Arg-13, Lys-67, and Arg-83 we also found extremely low gp350-biotin apparent binding affinities for a number of point mutations targeting positively charged amino acids on SCR1, namely Arg-28, Arg-36, and Lys-41 and, to a lesser extent, Lys-50 and Lys-57. In this study we cannot ascertain whether the observed charge dependence associated with the formation of a CR2-gp350 complex is a result of a longrange electrostatic attraction between the two molecules, which has previously been proposed as one of the mechanisms behind the CR2-C3d interaction (51) or is rather a consequence of localized ion-pair formation at a CR2-ligand binding inter- FIGURE 7. Surface representations of the CR2 SCR1-2 molecule in its C3d ligand-bound state (C3d not shown) as determined by x-ray crystallography (PDB accession code 1GHQ (49)). A and D, electrostatic representation of the structure of CR2 SCR1-2 demonstrating the preponderance of positive charge associated with this molecule. D, as for A except the molecule has been rotated about the y axis by 90°. B and E, alanine and proline substitutions mapped onto the surface of CR2 SCR1-2. The scheme used to color mutants shown in B, C, E, and F represents the percentage of gp350-biotin monomer binding of mutants relative to wild-type CR2 (at a concentration of 0.5 g of gp350-biotin, 0.4 g of PE-NeutrAvidin). E, as for B, except the molecule has been rotated about the y axis by 90°. C and F, negative charge substitutions and inhibitory monoclonal antibody epitopes mapped onto the surface of CR2 SCR1-2. The primary epitopes for 171 and 1048 mAbs on SCR2 of CR2 have been identified as 86 TPYRH 90 for 171 mAb and 111 WCQANNMW 118 for 1048 mAb and are shown in green and cyan, respectively (54). Also indicated is Trp-118, which is the only residue in the 1048 mAb epitope that is located on the same face as the C3d ligand binding site. F, as for C except the molecule has been rotated about the y axis by 90°. This figure was generated using the PYMOL Molecular Graphics System (Delano Scientific, San Carlos, CA). Upper panels in A and B represent levels of wild-type (WT) or mutant MBP-CR2 eluted from the amylose beads, whereas lower panels indicate levels of EBV identified using 72A1. A, shown are mutations targeting residues identified within SCR1 (R13A, S15P, R28A, R36A, K41A, K57A) and the linker region connecting SCR1 and SCR2 (K67A) as being important in the interaction between CR2 and gp350. Also shown is N11A, which demonstrates binding identical to wild-type as a positive control. B, shown is a mutant targeting Arg-83 (R83A) within SCR2 that has also been demonstrated to be important in the gp350 binding interaction and exhibits reduced binding relative to wild type. face. One likely scenario would involve a combination with long-range and short-range electrostatic interactions leading to the formation of a stable CR2-gp350 complex.
The overall binding site on CR2 delineated by our site-directed mutagenesis targeting the binding of the N-terminal three domains of gp350 also appears to reflect the binding site for the full-length form of gp350 and for intact EBV. Our EBV pulldown assay in which MBP-CR2 point mutants were incubated with concentrated virus demonstrated that all of the residues on CR2 implicated by our cell binding data to play a significant role in gp350 binding also exhibited a reduced capacity to bind intact virus. Our data also suggest that it is likely that the major CR2 binding site on gp350 is primarily located within the N-terminal 470 residues, in agreement with previous studies (14), although we cannot rule out the possibility that an additional binding site located on the C terminus of gp350 and comprising residues 822-841, as identified by Urquiza et al. (31), plays a peripheral role in binding.
In summary, our site-directed mutagenesis analysis of the CR2-gp350-biotin interaction has revealed a number of residues on both SCR1 and SCR2 that are critical for binding to occur. Many of these have not previously been associated with gp350 binding, particularly a number of positively charged residues on SCR1. It is interesting to note that the majority of these residues lie on a single face of CR2, although whether this face accurately delineates the gp350 binding site on this molecule depends on the mechanism of the electrostatic interaction between the two. It is hoped that our mutagenesis data defining the essential roles that residues Arg-13, Arg-28, Lys-41,and Arg-83 play in the binding of CR2 and EBV will assist in understanding of viral-receptor interactions.