Isolation of a Monoclonal Antibody That Targets the Alpha-2 Helix of gp120 and Represents the Initial Autologous Neutralizing-Antibody Response in an HIV-1 Subtype C-Infected Individual ^▿

Human samples and envelope clones.

The stored plasma and peripheral blood mononuclear cell (PBMC) samples analyzed in this study were obtained from participant CAP88 from the CAPRISA 002 Acute Infection Cohort. This study was reviewed and approved by the research ethics committees of the University of KwaZulu Natal and the University of the Witwatersrand in South Africa. The envelope genes were either previously cloned in our laboratory (9, 21) or obtained from the NIH AIDS Research and Reference Reagent Program or the Programme EVA Centre for AIDS Reagents, NIBSC, United Kingdom.

Cell lines.

Cells of the JC53bl-13 cell line, also known as TZM-bl cells, engineered by J. Kappes and X. Wu, were obtained from the NIH AIDS Research and Reference Reagent Program. Human embryonic kidney 293T cells were obtained from George Shaw (University of Alabama, Birmingham, AL), and N-acetylglucosamine transferase I-deficient 293S cells were obtained from James Binley (Torrey-Pines Institute). Both cell lines were cultured in Dulbecco modified Eagle medium (DMEM; Gibco, Invitrogen, Carlsbad, CA) containing 10% heat-inactivated fetal bovine serum (FBS) and 50 μg/ml gentamicin (Sigma-Aldrich, St. Louis, MO). Cell monolayers were disrupted at confluence by treatment with 0.25% trypsin in 1 mM EDTA.

Generation of chimeras and mutant envelopes.

Chimeric envelopes were created using an overlapping PCR strategy with the inserts and flanking regions amplified in separate reactions. After linkage, the 3-kb chimeric PCR fragments, generated using EnvAdir and EnvM primers (5), were cloned into plasmid pcDNA3.1 (directional; Invitrogen) and screened for function as previously described (8). Chimerism was confirmed by sequence analysis. Site-directed mutagenesis was performed using a Stratagene QuikChange II kit (La Jolla, CA).

Production of pseudoviruses.

Envelope-pseudotyped viruses were obtained by cotransfecting the envelope plasmids with pSG3ΔEnv (33) using FuGene 6 transfection reagent (Roche, Basel, Switzerland) as previously described (9). In some cases pseudoviruses were generated either in 293T cells treated with the glycosidase inhibitor kifunensine at 23 μg/ml (100 μM) before transfection or in GnTI-deficient 293S cells.

Production of infectious envelope clones.

Replicating viruses were prepared from specific envelope clones as described elsewhere (16). The vector pHIVΔenvBstEIInef-hisD was a gift from D. R. Kuritzkes. Briefly, linearized vector pHIVΔenvBstEIInef-hisD and full gp160 envelope PCR products were cotransfected into 293T cells with FuGene 6. Supernatants were harvested on day 4 and were used to infect activated CD8-depleted PBMCs via spinoculation. Viruses were harvested from the culture supernatants every second day until day 10 and tested for infectivity in JC53bl-13 cells. The 50% tissue culture infective dose in PBMCs was determined for each virus preparation as described elsewhere (18).

gp120 production and isolation.

Recombinant gp120 proteins were produced in 293T cells by transfection using codon-optimized gp120 genes cloned in the expression vector in plasmid pcDNA3.1 (Invitrogen). gp120 was isolated using Galanthus nivalis lectin agarose matrix (Sigma) and eluted with 1 M methyl-α-d-mannopyranoside (Sigma). Remaining protein contaminants were eliminated through ion-exchange chromatography using FastFlow Q-Sepharose (GE Healthcare Life Science, Piscataway, NJ), equilibrated in phosphate-buffered saline (PBS), and reconstituted in 2 M NaCl-PBS. The purified gp120s were collected in the flowthrough, washed in PBS, and concentrated to 5 mg/ml. The purities of the final gp120 preparations were tested by running 10 μg of protein in an SDS-polyacrylamide gel. Protein preparations with purity higher than 99% were used in subsequent experiments.

Adsorption of gp120-binding antibodies in plasma.

Adsorption of anti-gp120 antibodies was done as previously described (14), using gp120 covalently coupled to tosyl-activated magnetic beads. Anti-gp120 activity was measured by enzyme-linked immunosorbent assay (ELISA) as described elsewhere (14).

Neutralization assays.

Neutralization was measured as a reduction in luciferase gene expression after a single-round infection of TZM-bl cells with envelope-pseudotyped viruses (18). Titers were calculated as the inhibitor concentration (IC₅₀) or reciprocal plasma/serum dilution (ID₅₀) causing a 50% reduction of relative light units (RLU) with respect to the virus control wells (untreated virus). Neutralizing activity was also measured as a reduction in PBMC infection by measuring p24 Gag antigen synthesis as described previously (3).

Staining and sorting B-cell populations.

For antigen-specific memory-B-cell sorting, CAP63 gp120 and a chimeric protein containing the C3 region of CAP88 in the CAP63 backbone (63-88C3) were labeled with Pacific Blue and Alexa Fluor 647, respectively, using fluorochrome labeling kits (Invitrogen, Carlsbad, CA). Thawed PBMCs were stained using a viability marker (Aqua vital dye; Invitrogen, Carlsbad, CA) and a combination of the following antibodies: CD3 phycoerythrin (PE)-Cy5, CD14 PE-Cy5, CD16 PE-Cy5, CD235a PE-Cy5, CD19 allophycocyanin (APC)-Cy7, CD27 PE-Cy7, CD38 APC-Cy5.5, anti- immunoglobulin M (anti-IgM) fluorescein isothiocyanate (FITC), and anti-IgD PE (BD Biosciences, Mountain View, CA, and Invitrogen, Carlsbad, CA). The titers of all antibodies were determined and used at optimal concentrations for flow cytometry. During the sort, forward- versus side-scatter gating was used to select for lymphocytes, geometric gates were used to eliminate doublet events, and viability dye gating was used to select for viable cells. Memory B cells were gated as CD3⁻, CD14⁻, CD16⁻, CD235a⁻, CD19⁺, and IgD negative (IgD⁻). gp120-stained B cells were sorted as single cells into wells of a 96-well plate, selecting those cells that were labeled by the 63-88C3 gp120 but not by the CAP63 gp120. Cells were stored in room temperature reaction buffer at −80°C until use. Flow cytometric data were acquired on a BD Aria2 fluorescence-activate cell sorter (FACS), and the data were analyzed using FlowJo software (Tree Star, Ashland, OR).

Isolation of Ig variable gene transcripts.

The genes encoding Ig variable region gene segments of heavy chain (V_H) and light chain (V_L) were amplified by reverse transcription-PCR using a modification of the method described by Tiller and coworkers (29). Briefly, RNA from single B cells sorted in 96-well plates were reverse transcribed using Superscript III in the presence of primers specific for human IgG, IgM, IgD, IgA1, IgA2, kappa (K), and lambda (L) constant gene regions (15). The V_H, V_K, and V_L genes were then amplified from the transcribed cDNA separately by nested PCR, as described elsewhere, and analyzed on 1.2% agarose gels (15). The second-round PCR includes tag sequences at the 5′ end of each primer pair which permit assembling of the V_H and V_L genes into functional linear Ig gene expression cassettes (see below). PCR products were purified and sequenced. The variable gene segments, isotype, and potential functionality of the immunoglobulin were determined using the SoDA program (31).

Expression of recombinant antibodies from linear expression cassettes.

Three linear Ig expression cassettes, each containing the cytomegalovirus promoter and human Ig leader as one fragment, were used for small-scale expression and specificity analysis (15). Fragments for the heavy and light chains comprised either the IgG1 constant region, Ig kappa constant region, or Ig lambda constant region attached to poly(A) signal sequences. These two fragments plus either the V_H, V_K, or V_L gene amplified from single B cells as described above were assembled by overlapping PCR. PCR products containing linear full-length immunoglobulin heavy- and light-chain genes were purified, and the paired Ig heavy- and light-chain products were cotransfected into 293T cells grown in 6-well plates using Effectene transfection reagent (Qiagen, Valencia, CA). Cultures were fed 6 to 12 h later with 2 ml fresh medium containing 2% fetal calf serum (FCS) and incubated for 72 h at 37°C in a 5% CO₂ incubator. Thereafter, culture supernatants were harvested for antibody characterization.

Production of purified recombinant MAbs.

The selected immunoglobulin V_H and V_L genes from CAP88-CH06 were cloned into human Ig(γ) and Ig(λ) expression vectors in pcDNA3.3 (15). Clones with inserts of the correct size were sequenced to confirm identity with the original PCR product. For production of purified antibodies of CAP88-CH06 by batch transient transfections, 293T cells in 10 to 20 T-175 flasks or a Hyperflask (Corning) grown to 80 to 90% confluence in DMEM supplemented with 10% FCS were cotransfected with plasmids expressing HIV-1-specific Ig heavy- and light-chain genes using polyethylenimine. Recombinant antibodies were purified using protein A-agarose columns (Thermo, Rockford, IL).

Antibody binding assay.

Supernatants from the small-scale transfections were tested for the amount of IgG present by ELISA and for binding to HIV antigens by ELISA using standard techniques (1) and Luminex and using a standardized custom assay (30). Antibodies were tested for reactivity with a panel of HIV-1 Env and other HIV-1 proteins. Samples were serially diluted for assay, and data were analyzed using 5-parameter curve fitting; endpoint titers were calculated to be 3-fold above assay background. ELISA testing for antigens was considered positive if the optical density reading was above 0.13 unit. Affinity of recombinant MAbs was determined using a 4-parameter curve fit of ELISA titer data. Luminex assays were considered positive if the value obtained after the value for the blank bead was subtracted was greater than 20 units and greater than 10 divided by the IgG concentration in % g/ml. Transiently expressed MAbs with low IgG concentrations were retransfected and reassayed. In cases where multiple tests of the same antibody were available, the final data set contained data from the assay with the highest concentration of MAb.

MAb binding to pseudovirus gp120.

Pseudovirus supernatants were lysed in 1% Triton X-100 for 1 h. The gp120 molecules in the lysate were captured onto a solid phase via adsorbed antibody D7324 (Aalto Bio Reagents, Dublin, Ireland). MAbs were bound to gp120 in 5% milk–0.05% Tween 20 in PBS. Bound antibodies were detected using a total anti-human immunoglobulin biotinylated at 1:3,000 (KPL, Gaithersburg, MD) and antibiotin antibody-horseradish peroxidase conjugated at 1:1,000 (Calbiochem, Merck, Darmstadt, Germany). The ELISA was developed with tetramethylbenzidine substrate (Thermo Fisher Scientific, Rockford, IL) and stopped with 1 M H₂SO₄. Plates were read at 450 nm on a microplate reader (Molecular Devices, Sunnyvale, CA).

The anti-C3 neutralizing activity can be depleted by monomeric gp120.

In a previous study we demonstrated that CAP88, an HIV-1 subtype C-infected woman, developed potent autologous neutralizing antibodies first detected at 11 weeks of infection (21). The targets of these early antibodies were mapped by making chimeras between the CAP88 virus and an unrelated heterologous CAP63 virus that was insensitive to CAP88 serum antibodies. Transfer of the C3 region from the early virus of CAP88 (envelope clone 88.2mo.B5 isolated 2 months postinfection) into a CAP63 backbone resulted in a virus (clone 63-88C3) that was as sensitive to the earliest autologous neutralizing response as the CAP88 parental virus, with titers peaking at 26 weeks (Fig. 1 A, gray shading) (21). This indicated that the targets of the earliest autologous response were entirely within C3. As the virus escaped these anti-C3 antibodies, a new response targeting the V1V2 loop appeared at 26 weeks, becoming dominant at 54 weeks postinfection, as demonstrated by the neutralization of the V1V2 chimera (63-88V1V2) (Fig. 1A, striped shading). Analysis of the escape mutations to the first response mapped the epitope in C3 to the alpha-2 helix in the outer domain of gp120 (21).

We hypothesized that this epitope in C3 might be apparent on monomeric gp120. To prove this, we cloned, expressed, and purified gp120 from the 88.2mo.B5 virus and used this in adsorption/depletion studies. Plasma samples collected at 26 and 54 weeks postinfection were adsorbed with CAP88 and CAP63 gp120-coated beads. Blank beads were used as negative controls. Adsorbed plasma samples were tested for neutralization against the sensitive autologous virus 88.2mo.B5 (Fig. 1B and D) and for binding to CAP88 and CAP63 gp120 (Fig. 1C and E). CAP88 gp120-coated beads completely adsorbed the autologous neutralizing activity from plasma collected at 26 weeks postinfection (Fig. 1B), and this corresponded with almost total depletion of CAP88 gp120-binding antibodies (Fig. 1C). This suggested that the epitope targeted by the first autologous antibody response was wholly presented on monomeric gp120. However, adsorption of the neutralizing antibodies in plasma collected at 54 weeks postinfection was less efficient (Fig. 1D), despite the fact that most antibodies binding to monomeric CAP88 gp120 were removed (Fig. 1E). This might be due to the fact that the C3 antibody had started to wane and was being replaced by antibodies targeting an epitope in V1V2 (Fig. 1A) (21) that was presumably not adsorbed by monomeric gp120. A chimeric 63-88C3 gp120 was produced to specifically target the anti-C3 antibodies in CAP88 plasma. This protein efficiently depleted antibodies binding to the 63-88C3 chimeric and CAP63 gp120 proteins in the plasma collected at 26 weeks (Fig. 1C), which resulted in a 96% reduction of neutralizing activity against the 88.2mo.B5 clone, from a titer of 1:1,637 to one of 1:55 (Fig. 1B). However, this depletion was not as efficient as that for the full CAP88 gp120, perhaps because another specificity targeting an epitope outside C3 but within monomeric gp120 made a minor contribution to the autologous neutralization. This was also suggested by the significant levels of binding of the 63-88C3-depleted plasma to CAP88 gp120 (Fig. 1C). Nevertheless, the use of a chimeric protein where only the target of interest was present likely increased the specificity by reducing the binding of non-C3 antibodies. Indeed, the heterologous CAP63 gp120 did not adsorb any neutralizing activity in the plasma from either time point and therefore served as a suitable negative control (Fig. 1B and D). This protein, however, very efficiently depleted the CAP63 binding antibodies in this plasma, as expected (Fig. 1C and E). Little or no difference in binding to CAP88 gp120 was observed after adsorption with CAP63 gp120-coated beads (Fig. 1C and E). This might be explained by the presence of a large quantity of CAP88 type-specific binding antibodies in these plasma samples that were not adsorbed by CAP63 gp120, which generated a saturating signal in the ELISA.

Epitope-specific memory-B-cell sorting.

In order to isolate the B cells that produced these anti-C3 antibodies, a differential sorting strategy using the chimeric and heterologous gp120s labeled with different fluorochromes was designed. The chimeric 63-88C3 gp120 was used to identify B cells that specifically recognized the C3 epitope, while the CAP63 gp120 was used to exclude B cells expressing antibodies that bound to regions outside the C3. Given that the anti-C3 antibody titers in plasma peaked at 26 weeks postinfection (Fig. 1A) and were effectively removed by the chimeric gp120, we proposed that PBMCs collected after this time should contain memory B cells specific for this epitope. Thus, PBMCs collected at 34 weeks postinfection were stained with a panel of antibodies as well as CAP63 gp120 conjugated to Pacific Blue and 63-88C3 gp120 conjugated to Alexa Fluor 647. Memory B cells (CD19⁺, IgD⁻) positively stained with 63-88C3 gp120 but negative for the heterologous CAP63 gp120 were sorted into single cells and placed into a 96-well plate for amplification of heavy-chain (V_H) and light-chain (V_L) antibody genes (Fig. 2). The frequency of epitope-specific cells acquired from this sorting was approximately 16/10,000 memory B cells. However, upon further analysis, it was noted that a group of double-positive cells was inadvertently included in the sorting gate. Of the 64 cells sorted, 16 functional heavy- and light-chain pairs were identified and expressed as full-length IgG1 recombinant antibodies in 293T cells by transient transfection using linear Ig gene expression cassettes (15). The supernatants containing the expressed antibodies were tested for binding to the 63-88C3 and CAP63 gp120s in a multiplex Luminex binding antibody assay. Seven of the 16 expressed recombinant antibodies bound 63-88C3 gp120, and 6 bound the CAP63 gp120 to various levels, possibly due to the inclusion of the double-positive population (Fig. 3 A). The one MAb which failed to bind CAP63, 3469L, had the highest binding to 63-88C3 gp120, suggesting that it specifically recognized the C3 domain of CAP88. Confirmation of this came from the observation that the 3469L supernatant was the only one that neutralized the autologous virus 88.2mo.B5 significantly above background levels (Fig. 3B). One of the other MAbs isolated, CAP88-3468L, which bound strongly to both wild-type and chimeric proteins (Fig. 3A), was shown on further screening to recognize the V3 loop of gp120 (unpublished data). This anti-V3 MAb, however, showed no neutralization activity against the CAP88 virus.

Neutralization profile of CAP88-CH06 MAb.

The variable heavy- and light-chain regions of 3469L, from now referred to as CAP88-CH06, were cloned into an IgG1-expressing vector and produced on a large scale for further assessment of neutralizing activity. In the TZM-bl pseudovirus assay, the CAP88-CH06 MAb neutralized the early-infection clone 88.2mo.B5 very potently with an IC₅₀ of 13 ng/ml (Fig. 4 A). A similar level of activity was noted against the autologous transmitted/founder virus from CAP88 represented by clone 88.1mo.17, which had a gp120 amino acid sequence identical to that of 88.2mo.B5 gp120. This MAb, however, did not neutralize eight CAP88 viruses amplified from 6 and 12 months postinfection, when escape from this antibody specificity had already occurred (21) (Fig. 4A). Furthermore, no neutralizing activity was recorded at 50 μg/ml against a panel of 46 heterologous viruses, which included the subtype A (n = 6), B (n = 12), and C (n = 12) tier 2 panels (27) and the neutralization-sensitive strains MW965.26, SF162, and HxB2 (data not shown). It also failed to neutralize any of the 12 viruses from within the CAPRISA cohort (9) or the consensus C strain (10).

A similar pattern was observed in the PBMC neutralization assay. The MAb CAP88-CH06, was capable of neutralizing an infectious clone containing the envelope from the transmitted/founder autologous virus, 88.1mo.17, but not the escaped virus, 88.6mo.10, isolated at 6 months of infection (Fig. 4B). Interestingly, neutralization of the 88.1mo.17 virus was ∼8 times more potent in the PBMC assay than in the TZM-bl pseudovirus assay, with an IC₅₀ of 1.7 ng/ml, although the curves were flatter in the PBMC assay. The escaped 88.6mo.10 virus contained 2 amino acid mutations (I339N and E350K) in the alpha-2 helix of C3 which were previously shown to mediate escape from autologous plasma (21). Here we found that the CAP88-CH06 MAb was able to neutralize the escape virus obtained at 6 months in which these 2 sites were backmutated (88.6mo.10.N339I K350E), providing direct proof that this MAb targeted the predicted C3 epitope (Fig. 4C).

The CAP88-CH06 epitope is occluded by glycans.

We previously showed that escape from plasma neutralization in CAP88 was mediated by the I339N mutation, which introduced an N-linked glycosylation site, in combination with either an E343K or E350K charge change in the alpha-2 helix (21). These three residues, which are separated on a linear sequence, lie adjacent to one another on the solvent-exposed side of the helix, an arrangement that likely contributed to the formation of this epitope (highlighted in red in Fig. 5).

To further explore the role of these changes in the escape from CAP88-CH06 neutralization, we introduced combinations of these 3 escape mutations into the sensitive 88.2mo.B5 virus and measured the acquisition of neutralization resistance (as opposed to measuring acquisition of sensitivity, as shown in Fig. 4C). The isoleucine (I) at position 339 in the autologous clone 88.2mo.B5 was mutated to an asparagine (N), alone or in combination with a charge change from glutamic acid (E) to lysine (K) at position 343 or 350. In addition, the isoleucine 339 was also changed to a glutamine (Q) to assess whether glycosylation per se or just the change from a hydrophobic to a hydrophilic residue at this position could mediate escape. The I339N substitution mediated partial resistance, with a 7-fold increase in IC₅₀ from 13.2 to 102.9 ng/ml (Fig. 6 A). However, the I339Q substitution had a minimal effect on neutralization resistance, emphasizing the contribution of the N-linked glycan at this position in conferring escape from this antibody. The addition of either E343K or E350K in combination with the N339 glycan resulted in complete resistance to neutralization by the CAP88-CH06 MAb, even at 10 μg/ml. Interestingly, while the E343K mutant showed a 43-fold increase in IC₅₀ but remained neutralization sensitive, the introduction of the E350K mutation on its own resulted in a completely resistant variant. However, no naturally occurring sequences with only the E350K mutation were found among the 7 clones that were analyzed (21). This suggested that resistance to this MAb was mediated by the combined effects of both the charge change at position 343 or 350 and the N-linked glycan at position 339.

To further explore the role of N-linked glycosylation in CAP88-CH06 MAb neutralization, the sensitive virus 88.2mo.B5 was produced in 293T cells in the presence of kifunensine or in cells that lacked N-acetylglucosamine transferase I (GnTI-deficient 293S). Kifunensine inhibits the enzyme mannosidase I, which is responsible for trimming the Man9 precursor into Man5. This is the preliminary step in the synthesis of hybrid and complex glycans, and treatment with kifunensine results in a virus rich in large high-mannose Man9 glycans (2). Conversely, viruses grown in GnTI-deficient 293S cells mainly carry Man5 glycans, due to the failure of these cells to produce hybrid or complex glycans. It has been proposed that viruses produced in 293T cells express Man5-rich glycans on their envelope glycoproteins and therefore do not substantiality differ from viruses grown in GnTI-deficient 293S cells (4). The virus produced in the presence of kifunensine was resistant to neutralization by CAP88-CH06, while the virus produced in GnTI-deficient 293S cells was more sensitive than the control virus grown in 293T without kifunensine (Fig. 6B). Altogether, these data suggested that the exposure of the CAP88-CH06 epitope is affected by the glycan moieties, which may have been exploited by the virus to escape from this antibody response.

Escape did not result in loss of binding to recombinant gp120.

Having found mutations in the C3 region that totally abrogated neutralization by CAP88-CH06, we investigated whether these escape mutations resulted in the loss of binding to gp120. Monomeric gp120 proteins from the clone 88.2mo.B5 and the escape mutant 88.2mo.B5.I339N E343K were produced in 293T cells and tested for binding to the MAb CAP88-CH06. CAP63 gp120 was used as a negative control. The anti-V3 MAb CAP88-3468L, which was used as a positive control, bound all three proteins equally well, as expected (Fig. 7 B). The autologous neutralizing antibody CAP88-CH06 did not bind the CAP63 gp120, confirming the isolate-specific nature of this MAb (Fig. 7A). However, CAP88-CH06 bound both CAP88 gp120s, including the one containing the two escape mutations, albeit at slightly lower levels. This indicated that neutralization escape does not necessarily preclude binding to recombinant gp120. Given this dichotomy between neutralization and binding, we captured gp120 from lysed pseudovirus supernatants grown in 293T cells and tested them for binding to the MAbs. Interestingly, the MAb CAP88-CH06 was not able to bind the gp120 from the escape mutant or the heterologous gp120, while it efficiently bound the gp120 derived from the enrollment virus (Fig. 7C), consistent with the neutralization profile and in contrast to the binding to recombinant gp120. The anti-V3 MAb was able to bind all three proteins similarly, confirming that the differences in CAP88-CH06 binding were not the result of variations in the amount of captured gp120.

Sequence analysis of variable region of the MAb CAP88-CH06.

The V_H and V_L sequences of CAP88-CH06 were analyzed using the SoDA (31) and JoinSolver (28) programs. The heavy chain was shown to belong to the IgA1 subclass, and the maximum-probability inferred gene recombinations were V_H4∼39*01, D_H3∼3*01, and J_H4*02 genes for the heavy chain and V_λ3∼21*03 and J_λ2*01 genes for the lambda chain. The antibody sequences were compared to the closest reported germ line sequence and indicated that CAP88-CH06 had 21 mutations (5.9%) in the heavy-chain gene with insertions of 6 nucleotides in the CDR2 and 8 mutations (3.3%) in the light-chain gene. A high ratio of synonymous to nonsynonymous mutations was observed, suggesting that a certain level of affinity maturation from the germ line sequence took place (Fig. 8). However, it is not known if all these mutations were required to achieve the binding threshold required for neutralization.