Linear Epitopes in Vaccinia Virus A27 Are Targets of Protective Antibodies Induced by Vaccination against Smallpox

ABSTRACT

Vaccinia virus (VACV) A27 is a target for viral neutralization and part of the Dryvax smallpox vaccine. A27 is one of the three glycosaminoglycan (GAG) adhesion molecules and binds to heparan sulfate. To understand the function of anti-A27 antibodies, especially their protective capacity and their interaction with A27, we generated and subsequently characterized 7 murine monoclonal antibodies (MAbs), which fell into 4 distinct epitope groups (groups I to IV). The MAbs in three groups (groups I, III, and IV) bound to linear peptides, while the MAbs in group II bound only to VACV lysate and recombinant A27, suggesting that they recognized a conformational and discontinuous epitope. Only group I antibodies neutralized the mature virion in a complement-dependent manner and protected against VACV challenge, while a group II MAb partially protected against VACV challenge but did not neutralize the mature virion. The epitope for group I MAbs was mapped to a region adjacent to the GAG binding site, a finding which suggests that group I MAbs could potentially interfere with the cellular adhesion of A27. We further determined the crystal structure of the neutralizing group I MAb 1G6, as well as the nonneutralizing group IV MAb 8E3, bound to the corresponding linear epitope-containing peptides. Both the light and the heavy chains of the antibodies are important in binding to their antigens. For both antibodies, the L1 loop seems to dominate the overall polar interactions with the antigen, while for MAb 8E3, the light chain generally appears to make more contacts with the antigen.

IMPORTANCE Vaccinia virus is a powerful model to study antibody responses upon vaccination, since its use as the smallpox vaccine led to the eradication of one of the world's greatest killers. The immunodominant antigens that elicit the protective antibodies are known, yet for many of these antigens, little information about their precise interaction with antibodies is available. In an attempt to better understand the interplay between the antibodies and their antigens, we generated and functionally characterized a panel of anti-A27 antibodies and studied their interaction with the epitope using X-ray crystallography. We identified one protective antibody that binds adjacent to the heparan sulfate binding site of A27, likely affecting ligand binding. Analysis of the antibody-antigen interaction supports a model in which antibodies that can interfere with the functional activity of the antigen are more likely to confer protection than those that bind at the extremities of the antigen.

INTRODUCTION

Inoculation with vaccinia virus (VACV) elicits neutralizing antibodies against major antigens, including A27, A33, B5, D8, H3, and L1, on both the extracellular enveloped virus (EV) and the intracellular mature virion (MV or IMV), conferring protection against smallpox (1–5). As a result, widespread vaccination against smallpox (which is caused by variola virus [VARV]) led to the first eradication of a viral pathogen from nature (6). Among the major immunodominant antigens of the IMV, A27, H3, and D8, are adhesion molecules that bind to the glycosaminoglycans (GAGs) heparan sulfate (A27 and H3) and chondroitin sulfate (D8) (7–10). We have previously shown that anti-D8 antibodies can prevent the binding of D8 to chondroitin sulfate. Besides its binding to heparan sulfate, little is known about the function of H3 (9). However, human antibodies that target H3 in combination with those that target B5 provide significantly better protection than either antibody by itself and are promising for the treatment of smallpox in human (11). Since the general human population lacks protection against smallpox due to the cessation of smallpox vaccination, protective antibodies can be used to treat VARV-infected patients. While neutralizing anti-A27 antibodies protect against infection, they represent only a minor component of the Dryvax vaccine-induced immune response (12).

A27 is a homotrimeric extracellular protein that is attached to the viral membrane by binding to the transmembrane protein A17 through its C-terminal leucine zipper domain (residues 80 to 101). The GAG binding site is located at the N terminus, downstream of the signal sequence (residues 21 to 30) (13, 14). The central region of A27 consists of a coiled coil domain (residues 43 to 84), which is used to interact with the membrane fusion suppressor protein A26 through intermolecular disulfide bond formation (Cys71, Cys72). The crystal structure of an N-terminal fragment of A27 containing the heparan sulfate binding site and coiled coil domain (residues 21 to 84) was recently determined; however, only the central fragment (residues 47 to 84) is ordered, suggesting flexibility of the N-terminal GAG binding domain (15). The A27 structure illustrates the complexity and antiparallel nature of the A27 homotrimer, yet structural information about the N-terminal and C-terminal extremities is missing.

In this study, we produced a panel of anti-A27 antibodies by immunizing mice with VACV. We have identified 4 antibody groups (groups I to IV) based on cross-blocking experiments and identified the epitope using a peptide/protein enzyme-linked immunosorbent assay (ELISA). Group I, II, and IV antibodies recognized both the VACV lysate and synthetic peptides, suggesting that the epitope of these antibodies can be recapitulated using linear peptides. We have further determined the crystal structure of a protective antibody, 1G6 (group I), and a nonprotective antibody, 8E3 (group IV), in complex with their respective linear peptide epitopes, which are located at the N- and C-terminal extremities of A27, shedding light on the structural basis of A27 recognition.

MATERIALS AND METHODS

Viruses and antibodies.

VACV_WR stocks were grown on HeLa cells in T175 flasks, in which the HeLa cells were infected at a multiplicity of infection of 0.5. The cells were harvested at 60 h, and virus was isolated by rapidly freeze-thawing the cell pellet three times in 2.3 ml RPMI plus 1% fetal calf serum (FCS). Subsequently, the cell debris was removed by centrifugation. The clarified supernatant was frozen at −80°C and used as a stock. The titers of the VACV_WR stocks were determined on Vero cells (∼2 × 10⁸ PFU/ml). VACV_ACAM2000 was obtained from the CDC. The monoclonal antibodies (MAbs) used in the study were anti-H3#41; anti-D8 MAb Ab12.1; anti-L1 MAb M12B9; and anti-A27 MAbs 1G6, 12G2, 8H10, 6F11, 4G5, 12C3, and 8E3.

Hybridoma generation and characterization.

The hybridomas were generated as described previously (16). In brief, a 6-week-old BALB/c mouse was infected intranasally with 5 × 10³ PFU of VACV_WR. At 7 weeks after the infection, the mouse was injected intravenously with 7 × 10⁷ PFU of UV-inactivated WR virus. Three days later, the spleen of the mouse was harvested for hybridoma generation. Hybridomas that secreted anti-A27 antibodies were identified by immunofluorescence assays of HeLa cells infected with the wild-type VACV strain or a VACV strain with an A27 deletion (17). IgGs were purified from the hybridoma culture medium using protein G affinity chromatography, and purity was assessed by reducing and nonreducing SDS-PAGE. All experiments were performed using purified antibodies.

A27 expression and purification.

A27 with a C-terminal hexahistidine tag was cloned into pET15b (Invitrogen) and transformed into CodonPlus BL21 cells (Agilent). After the culture growth reached an optical density (OD) at 600 nm (OD₆₀₀) of 0.6, A27 expression was induced by induction with 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) at 37°C for 4 h. Cells typically from 1 liter were spun down (10 min, 4,000 × g) and resuspended in 50 ml lysis buffer (50 mM Tris, pH 8.0, 5 mM EDTA). Cells were sheared with a Microfluidizer apparatus (3 rounds at 1.4 × 10⁸ Pa; Microfluidics), and the crude lysate was clarified by centrifugation (1 h, 50,000 × g). Soluble A27 was purified by ion metal affinity chromatography (IMAC) using a HisTrap 5-ml column (GE Healthcare). The supernatant was passed through the HisTrap column and washed with three column volumes (15 ml) of wash buffer (50 mM Tris, pH 8.0, 300 mM NaCl, 20 mM imidazole). After a final wash with 50 mM imidazole, A27 was eluted with the same buffer containing ∼250 mM imidazole and subsequently dialyzed twice against 10 mM Tris, pH 8.0, 200 mM NaCl prior to size-exclusion chromatography (SEC) for characterization.

Protein ELISA.

Flat-bottom 96-well microtiter plates were coated with 100 μl of recombinant VACV A27 protein (either C-terminally truncated A27 from residues 16 to 100 [A27_16–100] or full-length A27 from residues 16 to 110 [A27_16–110]) at 1 mg/ml diluted in phosphate-buffered saline (PBS) overnight at 4°C (Thermo Scientific Pierce) and washed with washing buffer (PBS, pH 7.2, plus 0.05% Tween 20). Subsequently, the plates were blocked with blocking buffer (PBS, pH 7.2, plus 1% bovine serum albumin [BSA] and 0.1% Tween 20) for 2 h at room temperature (RT). The plates were washed and incubated with purified MAb at 10 μg/ml for 90 min at RT. The plates were washed, and the bound MAb was detected by addition of a streptavidin-horseradish peroxidase (HRP)-conjugated secondary antibody to mouse immunoglobulin G (Invitrogen) and incubation for 60 min at RT, followed by addition of o-phenylenediamine (OPD) substrate (Sigma-Aldrich).

Cross-blocking ELISA.

Recombinant A27_16–110 was prepared at 0.5 μg/ml and was used at 100 μl per well to coat Nunc Polysorbent flat-bottom 96-well plates. The plates were incubated overnight at 4°C and subsequently washed four times with PBS plus 0.05% Tween 20. One hundred microliters of blocking buffer (PBS plus 10% fetal bovine serum) was added to each well of the plate, and the plate was blocked for 90 min at room temperature. The blocking buffer was discarded, 100 μl of purified and unmodified antibodies of interest (at 10 μg/ml) was added to the plate, and the plate was incubated for 90 min to allow binding to recombinant A27. HRP-conjugated antibodies of interest (which were conjugated by use of an Innova Biosciences Lightning-Link HRP conjugation kit) were prepared at 0.5 μg/ml and added to the plates for 20 min without prior washing of the plate. The ability of the conjugated antibody to bind to A27 in the presence of a prebound antibody was assessed using an optical assay. The plates were developed using OPD, and the OD₄₉₀ was read on a SpectraMax 250 spectrophotometer (Molecular Devices).

Flow cytometry-based in vitro neutralization.

Vero E6 cells (1 × 10⁵ cells/well) were seeded in 96-well Costar plates (Corning Inc., Corning, NY) and incubated for 5 h to adhere. Subsequently, the cells were infected with 12.5 μl purified VACV expressing green fluorescent protein (GFP) at 1 × 10⁶ PFU/ml (final titer, 1.25 × 10⁴ PFU) and treated with 12.5 μl MAbs at 80 μg/ml (final concentration, 20 μg/ml) for 12 h at 37°C in a 5% CO₂ atmosphere in a total volume of 50 μl in the presence (final concentration, 2%) or absence of sterile baby rabbit complement (Cedarlane). Samples were prepared in duplicate. The cells were subsequently tested using flow cytometry as described previously (18, 19).

Vaccinia virus intravenous infection protection studies.

To infect mice, 1 × 10⁵ PFU of VACV_ACAM2000 was injected retro-orbitally. After the animals were infected on day 0, their weights were taken on day 0 for the initial body weight measurement. Beginning at 3 days postinfection, their body weights were taken every other day. Body weights were recorded until the animal reached 75% of its initial body weight, at which time it was euthanized, or until other external health variables for which euthanasia was the only humane course of action became present. The clinical score, a composite score of the pox lesion abundance on the four paws plus the tail, was evaluated as described previously (11). For A27 MAb protection studies, mice were inoculated intraperitoneally with 100 μg of antibodies at 1 day before infection. Control mice received PBS. An additional group received anti-H3#41 as a positive control. Animal husbandry and experimental procedures were approved by the Department of Laboratory Animal Care and the Animal Care Committee of the La Jolla Institute for Allergy and Immunology.

Epitope mapping by peptide ELISA.

Overlapping 20-mer peptides of the A27 antigen were synthesized (AnaSpec) and tested for MAb binding using an ELISA. Flat-bottom 96-well microtiter plates were coated overnight at 4°C with 100 μl of NeutrAvidin biotin-binding protein (1 mg/ml in PBS) (Thermo Scientific Pierce), and washed with washing buffer (PBS, pH 7.2, plus 0.05% Tween 20). Subsequently, the plates were blocked with blocking buffer (PBS, pH 7.2, plus 1% BSA and 0.1% Tween 20) for 2 h at RT. The plates were incubated with 100 μl of overlapping linear biotinylated peptides (200 ng/ml) in blocking buffer for 90 min at RT. The plates were washed and incubated with purified MAb at 10 μg/ml for 90 min at RT. The plates were washed, the bound MAb was detected by adding a streptavidin-HRP-conjugated secondary antibody to mouse immunoglobulin G (Invitrogen), and the plates were incubated for 60 min at RT, followed by the addition of OPD substrate (Sigma-Aldrich).

Peptide truncation and alanine scan.

Variant peptides with N- or C-terminal truncations and/or alanine substitutions were tested for their ability to block binding to the parent 20-mer peptides in an ELISA. In case the peptide in question contained an alanine, the alanine was replaced by a serine instead. Ninety-six-well plates were coated with 100 μl NeutrAvidin per well at a concentration of 0.5 μg/ml. The plates were incubated overnight at 4°C and washed 4 times with PBS plus 0.05% Tween 20. One hundred microliters of blocking buffer (PBS plus 10% FBS) was added to the plates, and the plates were incubated for 90 min at 4°C. Subsequently, the blocking buffer was discarded, 100 μl of biotinylated 20-mer peptide at 200 ng/ml was added to the plate, and the plates were incubated for 90 min at 4°C. Selected antibodies were simultaneously incubated with variant peptides. We used 30 μl/well of MAb at 600 ng/ml and incubated it with 30 μl/well of alanine-modified peptides at 100 μg/ml for 90 min at 4°C. After the plates were washed, 50 μl of the antibody-alanine peptide mix was added to the plate-bound peptides and the plates were incubated for 20 min at 4°C. The plates were washed, 100 μl/well of secondary goat anti-mouse IgGγ conjugated to HRP (diluted 1:1,000 in blocking buffer) was added, and the plates were incubated for 90 min at 4°C. A final wash step was performed, the plates were developed using o-phenylenediamine, and the OD₄₉₀ was read on a SpectraMax 250 spectrophotometer (Molecular Devices).

Antibody sequencing.

Total RNA from 300 μl hybridoma cells in solution was isolated using a NucleoSpin RNA II kit according to the manufacturer's instructions (Macherey-Nagel). cDNA was amplified using a OneStep reverse transcription-PCR kit (Qiagen). The reverse transcription-PCR was performed using primers 5′MsVHE and 3′Cy1 (for isotype IgG1 MAb 6F11) and primers 3′Cy2c outer (for isotype IgG2a MAbs 1G6, 12G2, 12C3, and 4G5) or 3′Cy2b outer (for isotype IgG2b MAb 8H10) for the heavy (H) chains, primers 5′mVkappa and 3′mCκ for the kappa light (L) chains (for MAbs 1G6, 12G2, 8H10, 6F11, 12C3, and 4G5), and primers 5′mVλ1/2 and 3′mCλ outer for the lambda light chain (for MAb 8E3) (20). The cycling profile was slightly modified from the manufacturer's recommendations and set up as follows: 1 cycle of 30 min at 50°C and 15 min at 95°C and 40 cycles of 30 s at 94°C, 45 s at 60°C (for heavy chains) or 58°C (for light chains), and 55 s at 72°C, followed by 1 cycle of 10 min at 72°C and a 12°C cooldown. The PCR products were verified by gel electrophoresis, with an ∼500-bp product being obtained for the heavy chains and an ∼450-bp product being obtained for the light chains. Afterwards, the PCR products were purified using a QIAquick PCR purification kit (Qiagen) and then sequenced by Invitrogen (which was provided with the respective 5′ primer for the heavy and light chains). The sequences included variable (V)-diversity (D)-joining (J) regions for heavy chains and V-J regions for light chains. Finally, antibody germ lines were determined using the International Immunogenetics Information System's V-Quest service (21).

Peptide-Fab complex preparation.

Purified MAbs 1G6 and 8E3 (1 mg/ml in 50 mM sodium acetate [NaOAc], pH 5.5) were incubated with 4% (MAb 1G6) and 2% (MAb 8E3) (wt/wt) activated papain (catalog number P3125; Sigma) for 4 h at 37°C in 1× digestion buffer. Papain was activated by incubating 20.80 μl papain with 100 μl 10× digestion buffer (1 M NaOAc, pH 5.5, 12 mM EDTA) and 100 μl cysteine (12.2 mg/ml) in a total volume of 1 ml for 15 min at 37°C. The papain digestion was stopped by adding 20 mM iodoacetamide (IAA). Digestion mixtures were dialyzed against PBS for subsequent protein A purification to remove undigested IgG and Fc. The protein A flowthrough containing Fabs was concentrated and purified by size-exclusion chromatography on a Superdex S200 GL10/300 column (GE Healthcare), using 50 mM HEPES, pH 7.5, 150 mM NaCl as the running buffer. Fabs were incubated with a 2× molar excess of the corresponding A27 peptides for 1 h at 4°C and concentrated using 30-kDa centrifugal filtration units to remove unbound peptides.

Crystallization and structure determination.

Crystals of the complex of MAb 1G6 and A27 from residues 31 to 40 (A27_31–40) were grown over several days at 22°C by sitting drop vapor diffusion while mixing 0.5 μl protein (5.2 mg/ml) with 0.5 μl precipitant (20% polyethylene glycol [PEG] 4000, 200 sodium phosphate dibasic). The crystals were flash cooled at 100 K in the mother liquor containing 20% glycerol. Crystals of the 8E3–A27_101–110 complex were grown over several days at 4°C by sitting drop vapor diffusion while mixing 0.5 μl protein (14 mg/ml) with 0.5 μl precipitant (10% PEG 3000, 200 mM magnesium chloride, 100 mM cacodylate, pH 6.5). The crystals were flash cooled at 100 K in the mother liquor containing 20% glycerol. Diffraction data were collected at beamline 11-1 of the Stanford Synchrotron Radiation Laboratory (SSRL; 1G6 Fab) and beamline 5.0.1 of the Advanced Light Source (ALS; 8E3 Fab) and processed with the iMosflm and Scala software as part of the ccp4 suite of programs (22, 23). The crystal structure of 8E3 was determined by molecular replacement (MR) using the PHASER program (24) and the Fab of a quorum-quenching antibody (PDB accession number 2NTF), separated in constant and variable domains. The structure of MAb 1G6 was similarly obtained by MR, using the Protein Data Bank (PDB) coordinates of MAb 8E3 with the complementarity-determining region (CDR) loops removed. The model was rebuilt into σ_A-weighted 2F_o – F_c and F_o – F_c difference electron density maps using the program COOT (25). The peptides were manually built in COOT during later stages of refinement. The 1G6–A27_31–40 structure was refined to 1.95 Å to R and R_free values of 19.8% and 23.6%, respectively. The 8E3–A27_101–110 structure was refined to 2.27 Å to R and R_free values of 20.3% and 22.4%, respectively. The quality of the models was examined with the program MolProbity (26).

Protein structure accession numbers.

The Protein Data Bank accession numbers for Fab 1G6–A27_31–40 and Fab 8E3–A27_101–110 are 5EOQ and 5EOR, respectively.

RESULTS

Generation of anti-A27 MAbs.

Hybridomas were generated from a mouse that had been infected with a sublethal dose of VACV. Among them, 12 secreted antibodies reacted with the wild-type VACV but not with a VACV mutant with an A27 deletion in an immunofluorescence assay of infected HeLa cells. We decided to further characterize seven of those antibodies (MAbs 1G6 [IgG2a], 12G2 [IgG2a], 8H10 [IgG2b], 6F11 [IgG1], 12C3 [IgG2a], 4G5 [IgG2a], and 8E3 [IgG2a]). Out of these, six antibodies bound the C-terminally truncated A27_16–100 protein in an ELISA, whereas the other antibody (8E3) did not bind A27 (Fig. 1A). To assess whether 8E3 binds to the C-terminal extremity of A27, we prepared full-length A27_16–110 and repeated the ELISA with four selected MAbs, including MAb 8E3. All tested antibodies were found to bind the new construct, suggesting that 8E3 binds to the C-terminal extremity of A27 (Fig. 1B).

FIG 1.

Protein ELISA with A27 MAbs. (A) Seven anti-A27 MAbs were tested for their ability to bind purified A27_16–100 protein. All MAbs except MAb 8E3 successfully bound the construct. Anti-D8 MAb Ab12.1 (aD8 Ab12.1) was used as a negative control. (B) Direct comparison of four selected anti-A27 MAbs binding the A27_16–100 protein versus the A27_16–110 protein. All four tested anti-A27 MAbs, including MAb 8E3, bound the longer construct. Anti-L1 MAb M12B9 (aL1 MAb M12B9) was used as a negative control. Dashed lines, the cutoffs for positive results (OD, 1). The experiments were performed three times.

Cross-blocking ELISA was performed, and antibodies 1G6, 12G2, and 8H10 clustered into one distinct cluster (group I). Additionally, MAbs 4G5 and 12C3 formed another, distinct cluster (group III). MAbs 6F11 and 8E3 could not be cross blocked by any of the other MAbs and were assigned to groups II and IV, respectively (Fig. 2).

FIG 2.

Cross-blocking results for seven A27 MAbs. Anti-A27 antibodies were tested for cross-blocking ability. MAbs 1G6, 12G2, and 8H10 were found to exclusively bind site I and made up the group I MAbs. MAbs 4G5 and 12C3 bound to site II and clustered into another distinct group (group II MAbs). MAbs 8E3 and 6F11 did not cross-react with any other MAbs and bound sites III and IV, respectively, making up group III and IV MAbs, respectively). The experiments were performed twice.

Group I anti-A27 MAbs neutralize MV in a complement-dependent manner.

We then tested the ability of the anti-A27 MAbs to interfere with VACV MV infection in an in vitro neutralization assay (Fig. 3). Vero E6 cells were incubated overnight with purified VACV_WR MV expressing a green fluorescent protein (GFP) in the presence or absence of antibody and complement. The samples were evaluated on the following day using a flow cytometer. Group I MAbs (MAbs 1G6, 12G2, and 8H10) neutralized more than 90% of the viruses at 20 μg/ml in the presence of complement. In contrast, group II, III, and IV MAbs neutralized the virus by only 20% or less in the presence of complement (Fig. 3, bottom). Except for MAb 6F11 (IgG1), all antibodies were able to bind complement (IgG2a and IgG2b). None of the tested MAbs was able to neutralize in the absence of complement (Fig. 3, top). Anti-L1 MAb M12B9 (19), an antibody know to strongly neutralize (>90%) in a complement-independent manner, was used as a positive control.

FIG 3.

In vitro neutralization assay with anti-A27 MAbs. Anti-A27 antibodies were tested for their ability to neutralize in vitro in a fluorescence-activated cell sorting-based neutralization assay featuring MAbs 1G6, 12G2, 8H10, 6F11, 4G5, 12C3, and 8E3. Anti-L1 MAb M12B9 was used as a positive control. All antibodies were used at a final concentration of 20 μg/ml. Anti-A27 MAbs 1G6, 12G2, and 8H10 (group I) were capable of strong neutralization (>90%) in the presence of complement (bottom) but not in its absence (top). Antibodies 6F11 (group II), 4B5 and 12C3 (group III), and 8E3 (group IV) did not neutralize in the absence of complement and showed weak or no neutralization ability (<20%) in the presence of complement. The positive control, anti-L1 MAb M12B9, strongly neutralized (>95%) the virus in a complement-independent manner. Dashed lines, 50% neutralization. The experiments were performed at least twice.

A group I anti-A27 MAb protects against vaccinia virus infection.

Next, anti-A27 MAbs were tested in an in vivo vaccinia virus protection system. SCID mice were infected with 10⁵ PFU ACAM2000 retro-orbitally (Fig. 4). In this model, MAbs 1G6 (group I), 6F11 (group II), and 8E3 (group IV) were tested for their ability to protect against moribundity, weight loss, and pox lesions (referred to as the clinical score) (27). The animals were euthanized when their body weight became 75% of their initial body weight (the endpoint criterion). Anti-H3#41 was used as a positive control; at day 49, it provided good protection against weight loss (P = 0.0024) (Fig. 4A) and mortality (P = 0.0008) (Fig. 4B) when the results obtained with anti-H3#41 were compared to those obtained with control mice receiving no MAb. Anti-H3#41 also provided robust protection against the development of pox lesions (P < 0.0001) (Fig. 4C). MAb 1G6 showed results comparable to those achieved with anti-H3#41 (P = 0.02 for weight loss and P = 0.001 and P < 0.0001 for the clinical score compared to the results for the control group). MAb 6F11 had some measurable activity protecting against weight loss (P = 0.047) and moribundity (P = 0.04) and offered substantial protection against pox development (P < 0.0001). Group IV MAb 8E3 did not significantly protect against weight loss (P = 0.2) or moribundity (P = 0.8) and provided only poor protection against the development of pox lesions (P = 0.02). From these results, we conclude that anti-A27 MAb 1G6 is highly effective in vivo at neutralizing MV and provides substantial protection from disease pathogenesis. Group II MAb 6F11 and group IV MAb 8E3 had modest beneficial impacts in vivo. These results agree well with the in vitro neutralization data described above.

FIG 4.

In vivo protection assays with anti-A27 MAbs. Protection of SCID mice against VACV ACAM2000. In this study, we tested anti-A27 MAbs 1G6 (group I), 6F11 (group II, conformational epitope), and 8E3 (group IV). Anti-H3#41 was used as a positive control. Body weight (A), survival (B), and clinical scores (C) over time are shown for 8 mice per group. Statistical significance was assessed at the final time point. Significance ranges are indicated by asterisks. *, P = 0.05 to 0.01; **, P = 0.01 to 0.005; ***, P = 0.005 to 0.0001; ****, P < 0.0001; ns, not significant. Dashed lines in panel A, initial body weight (100%) and minimum cutoff weight (75% initial body weight). The experiments were performed at least twice.

Epitope mapping of anti-A27 MAbs reveals three linear binding sites.

In order to understand why the anti-A27 MAbs showed such a big difference in their ability to neutralize in vitro and provide protection in vivo, we proceeded to characterize those MAbs by mapping their respective epitopes using ELISAs. All MAbs recognized the control VACV lysate and recombinant A27 protein. In parallel, we tested a series of 20-mer peptides spanning the A27_16–110 protein, overlapping by 10 residues at a time (Fig. 5). All three group I MAbs (MAbs 1G6, 12G2, and 8H10) bound a linear peptide spanning from residues 21 to 40 (Fig. 5A). Both group III MAbs (MAbs 12C3 and 4G5) recognized a different peptide spanning from residues 81 to 100, with MAb 12C3 also showing reduced but reproducible binding to the peptide spanning from residues 21 to 40, the same one recognized by group I MAbs (Fig. 5C). Group IV MAb 8E3 strongly recognized the peptide spanning from residues 91 to 110 (Fig. 5D), which explained our initial ELISA results where MAb 8E3 failed to bind A27_16–100 but bound to the full-length version, A27_16–110. Group II antibody 6F11 did not recognize any of the linear peptides (Fig. 5B), indicating that it bound a discontinuous epitope instead.

FIG 5.

Determination of linear epitope binding for A27 MAbs. (A) Group I MAbs (MAbs 1G6, 12G2, and 8H10) bound the linear peptide spanning from residues 21 to 40. (B) Group II MAb 6F11 did not a bind a linear epitope; thus, we hypothesized that it bound a conformational, discontinuous epitope instead. (C) Group III MAbs (MAbs 12C3 and 4G5) bound the peptide spanning from residues 81 to 100. Additionally, MAb 12C3 also bound the peptide spanning from residues 21 to 40. (D) Group IV MAb 8E3 bound the peptide spanning from residues 91 to 110. The experiments were performed three times.

To map the epitopes at a finer level, we performed truncation analyses of the identified 20-mer peptides. These showed that the binding site for group I MAbs was contained in residues 31 to 40 (Fig. 6A), while the binding site for group IV MAbs lay in residues 101 to 110 (Fig. 6C). Alanine substitutions of peptides 31 to 40 identified five residues (E33, I35, V36, K37, and D39) to be critical for binding by group I MAbs (Fig. 6B), whereas group IV MAb 8E3 required four residues (G105, R107, P108, and Y109) for proper binding. Arguably, residue E110 appeared to be of importance as well (Fig. 6D). For the group III antibodies (MAbs 12C3 and 4G5), peptide truncation assays showed the loss of antibody binding, which could indicate that truncation of the peptide led to the loss of the peptide secondary structure, which is necessary for antibody binding. It should be pointed out that while group I, III, and IV MAbs bound to linear peptides, they also bound to recombinant protein, suggesting that the linear peptide forms the conformational and continuous epitope found within the recombinant A27 protein.

FIG 6.

Truncation and alanine scan of A27 linear epitopes. (A and C) A truncation assay discovered the epitope of group I MAbs (MAbs 1G6, 12G2, and 8H10) to be residues 31 to 40 of the peptide spanning from residues 21 to 40. The epitope of group IV MAb 8E3 was residues 101 to 110 of the peptide spanning from residues 91 to 110. A low OD indicated that the peptide fragment preincubated with the antibody fully occupied the antibody's binding interface and prevented it from binding to plate-bound full-length peptide. (B and D) An alanine scan of the linear epitopes revealed residues E33, I35, V36, K37, and D39 (the linear epitope from residues 31 to 40) and residues G105, R107, P108, Y109, and (arguably) E110 (the linear epitope from residues 101 to 110) to be key residues for antibody binding. A low OD indicates a decreased ability of the antibody to bind to that particular peptide; thus, the residue in the original peptide replaced by an alanine had a strong impact on the binding ability. When the peptide sequence contained an alanine to begin with, the alanine was replaced by a serine instead. Dashed lines, cutoff for positive results (ODs, 0.5 for the truncation assay and 1.0 for the alanine scan). The experiments were performed three times.

Comparison of A27 MAb sequences.

Next we compared the sequences of the antibodies within and between each cross-blocking group. The group I MAbs are highly potent MV neutralization antibodies and target the same linear epitope (residues 31 to 40), with residues E33, I35, V36, K37, D39, and E40 having the most impact on binding. While CDR1, CDR2, and CDR3 of the heavy chains were quite divergent throughout the seven MAbs, antibodies within a given cross-blocking group showed highly conserved CDRs in their respective light chains (Table 1). Group III MAbs 12C3 and 4G5 displayed identical CDRs in their light chains, while group III MAbs 1G6, 12G2, and 8H10 all shared identical light-chain CDRs. This indicated that the light chain may contribute significantly to epitope binding, since it would be surprising if divergent heavy-chain sequences within a cross-blocking group were to give rise to highly similar epitope specificities.

TABLE 1.

Sequence characteristics of anti-A27 MAbs

Chain and group	Sequence identifiera	Mus musculus V gene and allele	Mus musculus J gene and allele	CDR sequence
				CDR1	CDR2	CDR3
Heavy chain
I	1G6_HC	IGHV8-8*01 F	IGHJ4*01 F	GFSLSTSGMG	IWWDDDK	AHDRGYYAMDY
I	12G2_HC	NAb	NA	NA	NA	NA
II	8H10_HC	IGHV9-2-1*01 F	IGHJ2*01 F	TFRDYS	INTETGDP	ARLRGDC
II	6F11_HC	IGHV7-3*02 F	IGHJ3*01 F	GFTFTDYY	IRNKPNGYTT	ARDYRFDGAWFAY
III	12C3_HC	IGHV9-2-1*01 F	IGHJ2*01 F	GYTFRDYS	INTETGDP	ARLRGDC
III	4G5_HC	IGHV4-1*02 F	IGHJ4*01 F	GFDFSRYW	INPDSSTI	ARPGWLPFYSGLDY
IV	8E3_HC	IGHV2-4-1*01 F	IGHJ3*01 F	GLSLTSYG	IWSGGNT	AIYYRYGAY
Light chain
I	1G6_LC	IGKV1-110*01 F	IGKJ1*01 F	QSLVHSNGNTY	KVS	SQSTHVPPT
I	12G2_LC	IGKV1-110*01 F	IGKJ1*01 F	QSLVHSNGNTY	KVS	SQSTHVPPT
II	8H10_LC	IGKV1-110*01 F	IGKJ1*01 F	QSLVHSNGNTY	KVS	SQSTHVPPT
II	6F11_LC	IGKV6-23*01 F	IGKJ2*01 F	QDVGTS	WAS	QQYSSWYT
III	12C3_LC	IGKV4-59*01 F	IGKJ2*01 F	SSVSY	DTS	QQWSNDPYT
III	4G5_LC	IGKV4-59*01 F	IGKJ2*01 F	SSVSY	DTS	QQWSNDPYT
IV	8E3_LC	IGLV3*01 F	IGLJ2*01 F	SQHSTYT	LRRDGSH	GVGDTIKEQFVYV

^aHC, heavy chain; LC, light chain.

^bNA, not available.

Crystal structure of MAb 1G6-peptide complex and MAb 8E3-peptide complex.

To identify the exact epitope and characterize its interaction with the antibody, we crystallized the Fab of neutralizing antibody 1G6 (group I) bound to peptide A27_31–40 (KREAIVKADE) and determined the structure to a resolution of 1.95 Å. We further determined the crystal structure of nonneutralizing antibody 8E3 (group IV) in complex with the linear peptide A27_101–110 (DVQTGRRPYE) at a resolution of 2.27 Å (Table 2; Fig. 7A). Both peptides are bound in a shallow binding crevice between the L and H chains of the Fab (Fig. 7B). Both the L and H chains of MAb 1G6 contributed equally to the binding of the peptide, as indicated by the buried surface areas between the peptide and the antibody. The H chain buried 385 Ų on the peptide, while the L chain buried 366 Ų. In the case of MAb 8E3, the H chain interacted less broadly with the antigen and buried 355 Ų, while the L chain buried 480 Ų on the peptide.

TABLE 2.

Data collection and refinement statistics

Parametera	Value(s) forb:
	Fab 1G6–A2731–40	Fab 8E3–A27101–110
Data collection statistics
Space group	P6	P6
a, b, c unit cell dimensions (Å)	118.3, 118.3, 198.7	99.3, 99.3, 117.6
Resolution range (Å)	49.7–1.95 (2.06–1.95)	55.6–2.27 (1.53–2.27)
Completeness (%)	95.7 (96.0)	97.8 (98.1)
No. of unique reflections	45,752	70,232
Redundancy	8.6	2.5 (2.5)
Rpim (%)	4.2 (40.6)	7.3 (38.2)
I/σ	12.3 (2.0)	6.3 (2.1)
Refinement statistics
Space group	P65	P6122
No. of reflections (> 0)	43,427	35,641
Maximum resolution (Å)	1.95	2.27
Rcryst (%)	19.8 (32.1)	20.3 (31.1)
Rfree (%)	23.6 (34.7)	22.4 (35.4)
No. of atoms	3,779	3,244
Protein	3,318	3,004
Peptide	65	77
Water	396	205
Ramachandran statistics (%)
Favored	98.2	98.8
Outliers	0.0	0.0
RMSD from ideal geometry
Bond length (Å)	0.009	0.011
Bond angles (°)	1.18	1.36
Avg B value (Å2)
Protein	31.6	32.7
Peptide	28.8	29.2
Water molecules	40.9	39.7

^aI, intensity of a reflection; R_pim, precision-indicating merging R factor; RMSD, root mean square deviation.

^bNumbers in parentheses refer to the highest-resolution shell.

FIG 7.

Crystal structures of Fabs 1G6 and 8E3 in complex with A27 peptides. (A) Overview of the Fab-peptide complex of Fabs 1G6 (top row) and 8E3 (bottom row). Green, heavy chain; orange, light chain. (B) Top view onto the antigen binding site of both Fabs, shown as a molecular surface and colored by electrostatic surface potential (red, negative; blue, positive; contoured from −30 to +30 kT/electron). Bound A27 peptides are shown as white sticks, and the 2F_o − F_c electron density is drawn as a blue mesh contoured at 1σ. (C) Polar interactions between Fab residues and the peptide. Blue dotted lines, H bonds and salt bridges.

In both structures, the first amino acid was disordered, suggesting that it did not interact tightly with the Fab. Also, the side chain of Arg32 in peptide A27_31–40 could not be modeled due to a lack of defined electron density (it was truncated to alanine in the MAb 1G6 structure). All other amino acids gave rise to well-defined electron densities, suggesting a tight interaction with the Fab (Fig. 7B).

Antibody 1G6 formed polar contacts with 7 amino acids of the A27 peptide, with 5 residues of the L chain interacting with 4 amino acids of the antigen (R32, E33, A34, and D39), while 5 residues of the H chain interacted with 4 amino acids (K37, A38, D39, E40) of the antigen (Table 3; Fig. 7C). L1 appeared to form the majority of the polar interactions with A27, while the H chain used H2 and H3 to bind the antigen. The side chain interactions of the antigen with the antibody explain the results of the peptide alanine scan. A27 residues E33, I35, V36, K37, and D39 were sensitive to alanine substitution. E33 formed an H bond with S32 of L1 (L1:S32), K37 formed a salt bridge with both H2:D59 and H2:D61, D39 formed a salt bridge with L2:K55 and an H bond with L1:Y37, and E40 formed a salt bridge with H3:R104 and an H bond with H3:Y106. Disruption of any of these contacts led to the loss of binding, suggesting that these interactions were equally important for binding. I35 and V36 are hydrophobic amino acids, and their side chains do not engage in polar interactions; however, I33 packed against H2:Tyr63, while Val36 packed against L1:H31 and L1:Y37. These hydrophobic interactions appear to be critical to the overall stability of the complex (Fig. 7B and C).

TABLE 3.

Antibody-antigen contact list

Antigen	Antibody and H or L chain	CDR	Interactiona
A27	1G6
Arg32O	Val99LN	L3	H
Glu33OE1	Ser32LOG, His31LNE2	L1	H, S
Glu33OE2	Ser32LN, His31LND1	L1	H, S
Ala34N	Thr97LO	L3	H
Ala34O	His31LNE2	L1	H
Lys37N	Gly105HO	H3	H
Lys37NZ	Asp59HOD1	H2	S
Lys37NZ	Asp59HOD2	H2	S
Lys37NZ	Asp61HOD2	H2	S
Ala38N	Gly105HO	H3	H
Asp39OD1	Lys55LNZ	L2	S
Asp39OD2	Lys55LNZ	L2	S
Asp39OD2	Tyr37LOH	L1	H
Glu40O	Lys55LNZ	L2	H
Glu40OE1	Arg104HNH2	H3	S
Glu40OE2	Arg104HNE	H3	S
Glu40OE2	Tyr106HOH	H3	H
A27	8E3
Gln103O	Thr31LOG1	L1	H
Thr104O	Thr33LN	L1	H
Arg106N	Gly96LO	L3	H
Arg106NH1	Thr98LOG1	L3	H
Arg107NH1	Thr33LO	L1	H
Arg107NH1	Asp35LOD2	L1	S
Arg107NH2	Glu50LOE1	L2	S
Arg107NH2	Asp35LOD1	L1	S
Arg107NE	Glu50LOE1	L2	S
Pro108O	His36HNE2	H1	H
Tyr109OH	Glu50LOE1	L2	H
Glu110O	Arg101HNH1	H3	H
Gln110O	Arg101HNH2	H3	H
Glu110OE1	Ser54HOG	H2	H
Glu110OE2	Gly34HN	H1	H
Glu110OE2	Ser54HN	H2	H

^aH, H bond; S, salt bridge; superscript, atom name.

Antibody 8E3 also formed polar contacts with 7 amino acids of the antigen. Six residues of the L chain interacted with 5 amino acids of the peptide (Q103, T104, R106, R107, and Y109), while 4 residues of the H chain interacted with 2 amino acids of the peptide (P108 and E110) (Table 3; Fig. 7C). While the L chain appeared to make more contacts with the peptide overall, alanine scanning of the peptide suggested that the N-terminal half of the peptide did not form specific interactions with the antibody. H-bond interactions with A27 residues Q103 and G105 were formed with backbone oxygens, indicating that they are independent of the amino acid sequence. The G105-to-alanine substitution likely induced a steric clash with the antibody surface. As a result, the amino acids that formed specific interactions with the antibody were located within the C-terminal half and included R107, Y109, and, possibly, E110. The P108-to-alanine exchange also likely induced a steric clash with the antibody. R107 and Y109 were involved in a salt bridge and an H bond with L2:E50, which therefore seems to be a major hot spot for binding the antigen, while E110 formed an H bond with H2:S54 and H3:R101, which, on the basis of the alanine scanning data (Fig. 6D and 7C), appeared to be less crucial for binding. Both MAb 1G6 and MAb 8E3 formed polar contacts with 3 A27 residues exclusively through the backbone atoms of the antigen, while 4 amino acids were in polar contact with the antibody in a sequence-specific manner. However, antigen side chains formed important hydrophobic interactions with antibody 1G6, while these additional contacts seemed to be dispensable for MAb 8E3. In summary, the combined data suggest that the binding of MAb 8E3 to the A27 peptide is less sequence specific than antigen recognition by MAb 1G6 and a few polar contacts dominate the overall binding, while MAb 1G6 recognizes the amino acid sequence of the antigen using both directional polar contacts and hydrophobic interactions with amino acid side chains.

DISCUSSION

In this study, we have characterized seven antibodies specific for the VACV A27 protein at different levels of detail (Table 4). Group I MAbs (MAbs 1G6, 12G2, and 8H10) potently neutralized VACV in a complement-dependent manner, while group II, III, and IV MAbs failed to neutralize VACV in the presence or absence of complement. By using a panel of different techniques, we were able to precisely map the epitopes of the group I and IV MAbs. While the group II epitope was conformational and discontinuous, we did not pursue characterization further, due to the complexity of the trimeric A27 protein. Epitopes for group III MAbs were characterized but showed some ambiguity in regard to what peptides were bound. Both MAb 12C3 and MAb 4G5 bound linear peptide A27_81–100 in the peptide ELISA. In addition, MAb 12C3 also bound peptide A27_21–40, which overlaps the epitope identified for group I MAbs. Since the A27 crystal structure suggested that A27 is an antiparallel and extended coiled coil trimer, both peptides could be located next to each other in the native protein. This region is not contained in the A27 protein that was crystallized (15). Since MAb 4G5 did not recognize peptide A27_21–40, we speculate that the overlap in the binding specificity that leads to cross-blocking between the two MAbs is restricted to amino acids in the peptide spanning from residues 81 to 100. As both antibodies share the identical L chain, the L chain could be responsible for binding to peptide A27_81–100, while sequence differences in the H chain dictate whether peptide A27_21–40 is bound or not. However, although only one peptide was required for the binding of MAb 12C3 in a peptide ELISA, combining both peptides did not seem to improve overall binding, suggesting that peptide A27_21–40 could be a secondary binding site (data not shown).

TABLE 4.

Summary of anti-A27 MAbs

Group	MAb clone	Isotype	Epitope	In vitro neutralization	In vivo protection
I	1G6	IgG2a	A2731–40 (KREAIVKADE)	Strong (>90%) with complement	Strong
I	12G2	IgG2a	A2731–40 (KREAIVKADE)	Strong (>90%) with complement	Not tested
I	8H10	IgG2b	A2731–40 (KREAIVKADE)	Strong (>90%) with complement	Not tested
II	6F11	IgG1	Conformational	Weak (<20%) with complement	Modest
III	12C3	IgG2a	A2731–40 (KREAIVKADE)	Weak (<20%) with complement	Not tested
III			A2781–100 (RLENHAETLRAAMISLAKKI)a
III	4G5	IgG2a	A2781–100 (RLENHAETLRAAMISLAKKI)a	Weak (<20%) with complement	Not tested
IV	8E3	IgG2a	A27101–110 (DVQTGRRPYE)	Weak (<20%) with complement	Modest

^aCommon epitope of group III MAbs.

A27 is one of the immunodominant antigens of VACV, and antibodies targeting A27 have been extensively studied (28–30). Group I antibodies had especially been previously identified, and the epitope mapped to the heparan binding domain (28). Therefore, the group I epitope appears to constitute a major immunogenic site for antibody recognition. In our hands, group I MAbs were also the only antibodies that were neutralizing in vitro and provided protection against VACV in SCID mice. While anti-A27 antibodies have been tested using different in vivo models, the SCID model used in this study, and the well-characterized intranasal challenge model (31), it is difficult to compare their efficacy side by side. However, based on the protective activity of the best anti-A27 MAbs in the model of intravenous infection of SCID mice with strain ACAM2000 shown in Fig. 4, which was comparable to that of anti-H3#41, we predict that they would perform similarly to anti-H3#41 (16) and provide minimal protection against morbidity after a respiratory VACV_WR infection but would provide a moderate amount of protection against death.

In this study, we continued to identify the exact epitope of group I and group IV antibodies, for which no structural information was available. The epitopes for both a group I MAb (MAb 1G6) and a group IV MAb (MAb 8E3) were determined by X-ray crystallography. MAb 1G6 binds peptide A27_31–40, while MAb 8E3 binds peptide A27_100–110. The crystal structure of the MAb 1G6–A27_31–40 complex revealed that both the heavy chain and the light chain contributed equally to the binding of the epitope. Ten residues (5 on heavy chains, 5 on light chains) have been identified to interact with a total of 7 amino acids of the antigen, with E33, I35, V36, K37, and D39 being important key residues. Since group I MAbs bind closely to the GAG binding site, we hypothesized that antibody binding potentially interfered directly with the adhesion of A27 to GAGs or generally impaired the function of A27. In contrast, nonprotective group IV antibody 8E3 bound at the C-terminal extremity of A27 and did not confer protection. Structural comparison of MAb 1G6 with MAb 8E3 did not illuminate drastic differences in the binding of the respective A27 peptide antigens. We therefore hypothesize that the location of the epitope in relation to the protein's function, rather than the binding chemistry, is a key factor in determining whether an anti-A27 antibody is protective or not. This model is in contrast to that for the VACV adhesion molecule D8, for which we have previously characterized epitopes for different antibody groups. Here, all the antibodies were protective in the presence of complement, irrespective of whether they competed with adhesion to its host ligand or not (10, 32). However, the structure of the GAG binding domain of D8 is more compact, and antibody binding in general could interfere with the D8 function, since recruitment of complement could lead to the steric hindrance of GAG binding regardless of where the antibody binds. As VACV can also use the H3 protein for adhesion to heparan sulfate, it could be that a functional A27 is dispensable for VACV attachment.

In summary, it appears that while antibodies can bind to nearly any accessible surface of a given protein, those that interfere with the function of the protein (MAb 1G6) are more likely to be protective than those that bind at remote areas of the protein (MAb 8E3).

Funding Statement

This work, including the efforts of all authors, was funded by HHS | NIH | National Institute of Allergy and Infectious Diseases (NIAID) (HHSN272200900048C). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.