Monoclonal Antibody 2C6 Targets a Cross-Clade Conformational Epitope in gp41 with Highly Active Antibody-Dependent Cell Cytotoxicity

This paper further defines the function and area of the HIV trimeric envelope protein targeted by the monoclonal antibody 2C6. 2C6 binding is influenced by amino acid mutations across two separate gp41 sections of the envelope trimer. This epitope is recognized on multiple clades (variant groups of circulating viruses) of gp41, gp140 trimers, and SOSIP trimers. For the clades tested, 2C6 has robust ADCC. As the target of 2C6 is available in the major clades of HIV and has robust ADCC activity, further definition and appreciation of targeting of antibodies similar to 2C6 during vaccine development should be considered.

ABSTRACT

Previous studies in our laboratory characterized a panel of highly mutated HIV-specific conformational epitope-targeting antibodies (Abs) from a panel of HIV-infected long-term nonprogressors (LTNPs). Despite binding HIV envelope protein and having a high number of somatic amino acid mutations, these Abs had poor neutralizing activity. Because of the evidence of antigen-driven selection and the long CDR3 region (21 amino acids [aa]), we further characterized the epitope targeting of monoclonal Ab (MAb) 76-Q3-2C6 (2C6). We confirmed that 2C6 binds preferentially to trimeric envelope and recognizes the clades A, B, and C SOSIP trimers. 2C6 binds gp140 constructs of clades A, B, C, and D, suggesting a conserved binding site that we localized to the ectodomain of gp41. Ab competition with MAb 50-69 suggested this epitope localizes near aa 579 to 613 (referenced to HXB2 gp160). Peptide library scanning showed consistent binding in this region but to only a single peptide. Lack of overlapping peptide binding supported a nonlinear epitope structure. The significance of this site is supported by 2C6 having Ab-dependent cell cytotoxicity (ADCC) against envelope proteins from two clades. Using 2C6 and variants, alanine scanning mutagenesis identified three amino acids (aa 592, 595, and 596) in the overlapping region of the previously identified peptide. Additional amino acids at sites 524 and 579 were also identified, helping explain its conformational requirement. The fact that different amino acids were included in the epitope depending on the targeted protein supports the conclusion that 2C6 targets a native conformational epitope. When we mapped these amino acids on the trimerized structure, they spanned across oligomers, supporting the notion that the epitope targeted by 2C6 lies in a recessed pocket between two gp41 oligomers. A complete understanding of the epitope specificity of ADCC-mediating Abs is essential for developing effective immunization strategies that optimize protection by these Abs.

IMPORTANCE This paper further defines the function and area of the HIV trimeric envelope protein targeted by the monoclonal antibody 2C6. 2C6 binding is influenced by amino acid mutations across two separate gp41 sections of the envelope trimer. This epitope is recognized on multiple clades (variant groups of circulating viruses) of gp41, gp140 trimers, and SOSIP trimers. For the clades tested, 2C6 has robust ADCC. As the target of 2C6 is available in the major clades of HIV and has robust ADCC activity, further definition and appreciation of targeting of antibodies similar to 2C6 during vaccine development should be considered.

INTRODUCTION

Most neutralizing antibodies (Abs) target epitopes on cleaved trimerized human immunodeficiency virus (HIV) envelope (Env) proteins (1). Env epitopes that induce neutralizing Abs are shielded, and some are dependent on the presentation of quaternary epitopes during trimerization (2). Developing optimal neutralizing Abs remains the major focus of many vaccine strategies (3).

Neutralization is not the only useful function of anti-HIV Abs. Other functions, including Ab-dependent cell-mediated cytotoxicity (ADCC), were correlated with the protection seen in the recent RV144 vaccine trial (4–6). Mutation of the Fc receptor, on which ADCC depends, reduced the protective effect of the neutralizing Ab B12 in a simian-human immunodeficiency virus (SHIV) challenge macaque model (7). Recent data suggest that more potent neutralizing Abs, such as PGT121, can overcome such Fc receptor mutations (8). Nonneutralizing Abs (nnAbs) targeting conformational epitopes in gp41 with significant ADCC function are able to limit cell-to-cell spread of infection (9). Other nnAbs targeting this same gp41 epitope, such as F240 (10) and 7B2 (11), showed partial protection in macaques during SHIV vaginal challenge and an effect on viral set point (12). These limited studies support that ADCC may be able to assist in protection and control. Improved understanding of the epitope specificity of ADCC-mediating Abs is essential for developing effective immunization strategies that optimize protection by these Abs.

Newer vaccine reagents, such as the BG505 SOSIP trimer, have been shown to natively replicate the neutralizing epitopes seen on native HIV Env (13). However, only a limited number of gp41 epitopes have been characterized on the SOSIP trimer. Mapping of ADCC epitopes on the SOSIP is also lacking. It is unclear if nnAbs with these alternative functions would target this construct and, if so, would be beneficial targets in vaccine regimens.

To enhance isolation of quaternary-targeting Abs (QtAbs), we previously used HIV-like particles (VLPs) to bind and sort individual B cells from long-term nonprogressor (LTNP) subjects to identify a panel of monoclonal Abs (MAbs) (14, 15). When recombinant full-length MAbs were expressed using natively paired heavy- and light-chain sequences, a subset of these MAbs exhibited the binding profiles of QtAbs, preferably binding trimerized free and VLP-associated Env rather than monoclonal Env. Despite the highly mutated nature of these Abs, which is often associated with neutralization for HIV-specific Abs, they are generally nnAbs. Previous competition studies using the most mutated (80 to 88% nucleotide homology to inferred germ line) of the Abs suggested targeting of three distinct epitopes and that these epitopes are conformationally dependent (14). Two of the structurally influenced epitopes were resolved, but the epitope bound by MAbs 76-Q3-2C6 and 76-Q3-6B8 (here termed 2C6 and 6B8, respectively) could not be resolved initially.

The Abs 2C6 and 6B8 both have excessively long CDR3 regions (21 amino acids [aa]), which is theorized as being crucial for broadly neutralizing antibodies (16). Because 2C6 and 6B8 are also highly mutated, we pursued functional studies and epitope characterization. We initially confirmed that our QtAb 2C6 bound gp41 ectodomain of trimeric forms from a number of clades, including clade A, B, and C SOSIP constructs. 2C6 exhibited ADCC activity using both clade B and clade C gp41 targets. Peptide mapping failed to reveal an overlapping peptide binding, as is typically found with linear epitope targeting. Numerous single peptides between the immunodominant loop and N-terminal heptad repeat (NHR) defined a minimal 13-amino-acid region recognized by 2C6. Alanine scanning mutagenesis with mutants of 2C6 resolved the amino acids in the previously identified peptide region. This technique also identified additional amino acids bound by 2C6 that were outside the area of the peptide region. The mapping data support the conclusion that 2C6 targets a structural epitope spanning multiple gp41 oligomers. This functionally significant epitope is present on multiclade trimeric forms of the HIV Env protein.

RESULTS

The MAbs 2C6 and 6B8 were originally cloned from an LTNP subject by flow-cytometric capture assays with the clade B BaL HIV Env expressed on fluorescent VLPs (15). 2C6 initially was shown to neutralize the BaL strain when expressed as a Fab construct (17); however, full-length expression failed to show significant neutralization (15). The clonal relationship of 2C6 and 6B8 was defined by the overall nucleotide homology (96% identity) and homology of the expressed CDR3 region (100%). Due to this homology, experiments described here were performed solely with 2C6.

Previously, we showed that 2C6 recognized clade B BaL VLP-associated trimeric Env and free gp140 foldon trimeric Env but had significantly diminished binding to a recombinant gp41 protein (17, 18). To further explore gp140 binding, we obtained a cross-clade panel of gp140 proteins (from the NIH AIDS Reagent Program: clade A, UG 37; clade B, JRFL; clade C, CN54; clade D, UG21; and clade F, Bro29). Utilizing enzyme-linked immunosorbent assay (ELISA), all of the gp140 trimers were bound by 2C6 except for clade F protein (Fig. 1A). 2C6 consistently had the highest binding against clade A, which is intriguing, since this LTNP subject was infected with a clade B strain and showed clade B neutralization (19). Of the two clade B Env, BaL was more readily bound than JRFL.

FIG 1.

2C6 recognizes the gp41 ectodomain of multiple clades, including the SOSIP trimer. (A) Epitope of 2C6 is present on gp140 constructs from multiple clades. (B) Native gel electrophoresis and Western blotting after transfer to nitrocellulose membrane probed with 2C6, shown in duplicate. (C) SDS-PAGE and subsequent Western blotting of nonreduced and reduced samples probed with 2C6. (D) Binding of 2C6 and controls to SOSIP trimers from three different clades. (E) 2C6 recognizes clade B MN gp41 full protein and clade C ZA1197 gp41 ectodomain. Anti-influenza Ab FI6_v3 is the control. (F) ELISA showing direct binding to ectodomain of clade B gp41 and that the gp41 ectodomain can compete with 2C6 clade B BaL gp140 trimer binding.

2C6 binding to various constructs.

Since the initial cloning of 2C6, there have been a number of advancements in vaccine constructs. Recent development of the BG505 SOSIP.664 gp140 trimer (here termed SOSIP) (13) construct has advanced the ability to recapitulate the antigenic presentation of functional viral Env protein (20). Because of the extensive mutations of 2C6 (21-aa replacement of the heavy chain) and a long heavy-chain CDR3 (21 aa), we wanted to asses if this Ab could bind SOSIP constructs. Native gel electrophoresis shows 2C6 binds the clade A BG505 SOSIP construct (Fig. 1B). On SDS-PAGE and subsequent Western blotting (Fig. 1C), under nonreduced conditions, 2C6 binds the full clade B BaL gp140 foldon trimeric construct. The SOSIP trimer remains intact and shows monomeric forms, both of which are recognized by 2C6. This binding is specific to the gp41 region, as shown by inclusion of clade B MN gp41 protein. SDS-PAGE analysis and subsequent Western blotting under reduced conditions shows recognition of the gp140 monomeric form of the foldon trimers, which lack the cleavage site. SOSIP under reduced conditions, which retains the cleavage site but lacks the membrane-proximal external region (MPER), resolves an appropriately sized band representing the portion of the gp41 ectodomain lacking the MPER.

To explore if multiclade recognition of SOSIP trimers occurs similarly to that of gp140 foldon trimers, ELISA binding assays of 2C6 were performed on four SOSIP trimer constructs (AMC008 v4.2 and B41 v4.2 [clade B], BG505 v5.2 [clade A], and Zm197m v4.1 [clade C]) (Fig. 1D). 2C6 readily recognized all four constructs. Since the Western blot resolved gp41 targeting, multiple-clade targeting of gp41 was further shown by ELISA for gp41 constructs (Fig. 1E). Clade C ZA1197 lacks the MPER domain, which is also lacking in the gp41 portion of the BG505 SOSIP trimer (Fig. 1B). The immunodominant 1 loop-targeting QtAbs 76-Q4-5F4, 76-Q5-5C2, 8F6, and 76-Q6-7B6, previously cloned in our laboratory (15) (here referred to as 5F4, 5C2, 8F6, and 7B6), are used as controls and also recognized gp41 in both clade B and C constructs. FI6_v3 is an anti-influenza Ab used as a negative control. Localization to the ectodomain of gp41 was confirmed by binding a commercially available construct (gp41 aa 541 to 682; number 30-AH83; Fitzgerald Industries International, Acton, MD) (Fig. 1E). Preincubating 2C6 with this ectodomain can interfere with 2C6 recognition of gp140 foldon trimers.

Ab competition.

In our previous studies, 2C6 interfered with biotinylated 8F6 binding, which targets the immunodominant loop of gp41, but the reciprocal was not shown (14). To identify the binding domain of 2C6 on gp41, a competition ELISA was performed using well-characterized anti-gp41 monoclonal Abs. Consistent with our previous study, our immundominant 1 loop-targeting Abs (5C2, 5F4, 7B6, and 8F6) did not interfere with 2C6. Other Abs that failed to interfere are T32, which maps to immunodominant 1 loop aa 597 to 613 (21), 98-6, which targets another conformational epitope mapped to aa 579 to 613 (22), and F240, an immunodominant 1 loop binder (23) with ADCC activity (24). The 50-69 Ab, known to have significant ADCC activity, (25), competed with 2C6 (Fig. 2A). 50-69 targets a structural epitope spanning aa 579 to 613 at the end of the NHR and continuing into the immunodominant 1 loop of gp41. Our immunodominant 1 loop binding Abs did not interfere with 50-69 binding (data not shown). Similar to the gp140 BaL foldon construct, 50-69 also showed interference against 2C6 binding when using the BG505 SOSIP trimer as a target in ELISA. Both trimers are specifically recognized by 2C6, although there is a roughly 10-fold higher estimated 50% effective concentration (EC₅₀) for the SOSIP trimer (Fig. 2B).

FIG 2.

Ab competition of 2C6 on gp140 and SOSIP BG505 trimers. MAbs (as labeled; obtained from NIH AIDS Reagent Program) were used to compete against 2C6 binding to BaL gp140 foldon trimers (A) and His-tagged SOSIP trimers bound to a nickel-coated plate (B).

Peptide binding.

We had previously utilized a group M consensus peptide set to attempt to map 2C6 binding to a linear epitope (14). We did not find a set of overlapping peptides that would be typical for a linear epitope targeting Ab. This suggested that the epitope of 2C6 is not a simple linear epitope. There were two single peptides that had relatively low but specific binding on multiple experiments. One peptide, representing aa 216 to 230, was within the gp120 region and is not known to be involved in 50-69 binding. The other, representing aa 585 to 599, is within the known epitope of 50-69.

To attempt to confirm this binding, peptides from the gp41 domain regions of two other overlapping peptide sets were screened (Fig. 3, gp41 portion of ectodomain region). Similar to the group M set, the clade B-specific MN envelop peptide and the clade A BG505.W6M.ENV.C2 peptide set are peptides of 15 aa in length, with 11-aa overlaps between sequential peptides. Within each set, the MAb 2F5, which recognizes the linear epitope ELDKWA, was used as an internal positive control and shows binding to 2 to 3 overlapping peptides, representing linear epitope targeting (peptide sequences are shown in Fig. 3). The negative controls used in the figure were a gp41 C-terminal heptad repeat structure-targeting Ab 76-Q7-7C6 (14) and the CD4 binding site Ab VRC01. Notably, certain peptides in group M and MN sets showed nonspecific binding, so data are shown normalized to the internal negative-control Ab. 2C6 bound specifically and repeatedly to a single peptide within each of the three peptide sets (Fig. 3, left, peptide sequences are beneath the binding data). Affinities relative to those of 2F5 control binding are noted in Table 1. In all three peptide sets, 2C6 failed to bind overlapping peptides within the ectodomain of gp41.

FIG 3.

Binding of 2C6 to HIV-1 gp41 ectodomain peptides. Peptides spanning from amino acid 572 (referenced to gp160 HXB2 strain) to the transmembrane domain are shown. Peptides were bound with 2C6 and a variety of control MAbs to distinguish specific from nonspecific binding. 2F5 recognizes the linear epitope ELDKWA and is used as a positive control. Sequences of peptides showing specific binding are shown below the binding data for each peptide set.

TABLE 1.

Protein construct recognition by 2C6

From an additional peptide set (PTE peptides; NIH no. 11551, lot no. 15003), we identified peptide PTE 105, representing sites 583 to 597 (VERYLRDQQLLGIWG), that was robustly bound by 2C6 (Fig. 4). This binding was ablated by a change from R588K (Table 1, changes from R588 are in boldface), represented by PTE peptide 25. Similarly, a JRFL sequence peptide also was not recognized by 2C6 and contained an R588G replacement. Notably, gp140 binding of clade B JRFL was less efficient than that of BaL (Fig. 2). Although there are other sequence differences between JRFL and BaL outside this region, this finding suggests that the amino acid location of 588 is involved in this epitope. However, the structural context of this region apparently plays a role, as a 2-aa shift in the peptide (as seen in group M peptides) can bind robustly despite this R588K mutation, and even a single-amino-acid shift seen in clade M peptide shows recognition. Overall, these overlapping peptides define a minimal 13-aa epitope (Table 1); however, these data support that the epitope of 2C6 is not a simple linear epitope.

FIG 4.

Effect of amino acid gp160 588 mutations on peptide recognition by 2C6. (A) PTE 105 peptide (583-597), which has R588, shows specific binding compared to two overlapping peptides with K588 (PTE 25) and G588 (JRFL peptide). (B) Western blot analysis of 2C6 Ab shows specific recognition of PTE 105.

Activity.

In previous studies, 2C6 showed restricted neutralization only after CD4 induction using solely the Fab form (17). Since 2C6 targets gp41 (Fig. 1 and 3) in the epitope binding region of 50-69 and F240, which have ADCC activity, we tested 2C6 and our collection of previously mapped immunodominant 1 loop-targeting Abs for ADCC activity. Targets used were recombinant gp41 proteins recognized by 2C6 (Fig. 1E). We screened Abs, including 2C6 for ADCC activity, using a rapid fluorometric ADCC assay. The lysis of gp41-coated target cells measured by this assay reflects both the cytotoxic activity of natural killer cells as well as phagocytic activity. Results are compared to values for HIV immunoglobulin and control Abs FI6_v3 (influenza) and NC-1 (26) at 500 μg/ml. Using the MN gp41 monomer as a target, 2C6 showed robust ADCC activity (Fig. 5). Similar to the immunodominant 1 region-targeting MAb F240, MAbs 5F4, 5C2, 8F6, and 7B6, which target the immunodominant 1 loop, also exhibited ADCC activity but to a lesser degree than 2C6. An anti-Gag Ab cloned from the same LTNP subject and expressed in the same manner did not show any activity.

FIG 5.

ADCC activity of 2C6 Ab using MN gp41, which has MPER. Test Ab (as labeled) concentrations are 10, 50, and 500 ng/ml, where the peak of the triangle corresponds to the greatest concentration. Data are normalized to HIV immunoglobulin (HIVIG), with influenza-specific MAb FI6_v3 Ab and HIV NC-1 Ab as controls at 500 ng/ml.

Ab structure.

To explore which regions of 2C6 contribute to its binding and function, a number of 2C6 mutants with regions converted to the inferred germ line were created (germ line reference to the IMGT database [27]). These were created by replacing the 2C6 sequence in framework 1 and CDR1, framework 2 and CDR2, framework 3, and CDR3 (mutA and mutB) with nearly identical to germ line sequence from unrelated anti-HIV Ab 47e. 47e uses the same heavy chain (VH1-69) and a similar light chain and has shared homology in heavy-chain CDR3 (Fig. 6A to C). For example, CDR1mut contains the majority of the 2C6 sequence, but the portion (Fig. 6A) encompassing CDR1 is sequence derived from 47e.

FIG 6.

Binding and function of 2C6 and lower-affinity mutants. (A to C) Heavy-chain alignment (A and B) and light-chain alignment (C) of anti-HIV Ab 47e and 76-Q3 clones 2C6 and 6B8. (D) Binding of 2C6 and mutants to clade B BaL foldon trimer. (E) ADCC activity of 2C6 and mutants of 2C6 (10, 50, and 500 ng/ml concentrations, represented by the wedge) using clade B MN gp41 and clade C ZA1197 gp41 as targets. Data are normalized to HIV immunoglobulin with influenza-specific MAb FI6_v3 Ab and HIV NC-1 Ab as controls at 500 ng/ml.

Variable light chains were generally less mutated for both 47e and 2C6 and were predicted to be derived from two different germ line sequences, VK1-27 and VK1-39. However, these variable gene sequences are very similar. Of the six germ line amino acids that distinguish these two inferred germ line sequences, 2C6 had two of six amino acids consistent with VK1-39 despite a predicted germ line sequence of VK1-27. Expression with either light chain did not affect 2C6 targeting (data not shown), so all these chimeric mutant Abs were expressed with the light chain of 2C6.

In binding to trimeric Env, overall the mutation analysis shows that the importance of the major structural areas are in the order CDR3 ≫ CDR1 > CDR2 > framework 3, despite the high number of framework 3 region mutations (Fig. 6D). To explore how these structures would affect function, these mutants were assessed for ADCC activity. Using clade B gp41 as a target, most mutations severely affected ADCC activity, except for the framework 3 mutation (Fig. 6E). We also assessed if ADCC activity was equivalent for different clades, similar to binding (Fig. 1).

Using the clade C ZA1197 gp41, ADCC activity of 2C6 was greatly increased compared to the control levels (Fig. 6E). For clade C, only exchange of the first portion of the CDR3 of 2C6 with the 47e sequence completely ablated function. Notably, the second CDR3 section replacement was comparable to other sectional replacements. Other regions did not follow clade B trimer affinity, as they all appeared to be relatively equal in disrupting ADCC activity. Of the slight differences seen, the relative order of importance was CDR2, framework 3, and then CDR1 (Fig. 6E).

Alanine scanning shotgun mutagenesis.

Although we determined a 13-amino-acid peptide sequence that can be identified by 2C6, the lack of overlapping peptides and the differential binding to the JRFL-derived peptide (Table 1) versus the trimer (Fig. 1C) supports the conclusion that 2C6 binds a conformational epitope. Therefore, to further resolve the binding site, we performed epitope mapping by alanine scanning shotgun mutagenesis. We constructed a comprehensive mutation library of gp160 variants (28), measured expression by immunofluorescent Ab staining, and detected Ab binding by flow cytometry with comparison to control Ab binding to distinguish specific amino acids from mutations that globally affected folding and trimer formation (Fig. 7). For linear epitope-targeting Abs in this system, critical residues are generally defined as clones with reactivity of <30% relative to that of wild-type (WT) HIV gp160. Initial mapping with 2C6 showed no specific residues that reduced binding to <30%. However, residues G524A, L592A, and W596A showed consistent reductions (Fig. 7A). The conformational epitope binding Ab 76-Q13-6F5 (14) is shown as a control.

FIG 7.

Critical residues and epitope map for MAb 2C6 chimers using alanine scanning mutagenesis. (A) Binding data for WT and mutant forms of 2C6 for the identified critical residues compared to control Ab 76-Q13-6F5. (B) Visualization of MAb 2C6 epitope on the crystal structure of the BG505 SOSIP.664 HIV-1 Env trimer (PDB entry 4ZMJ). Residues G524, R579, R588, L592, I595, and W596 are highlighted.

Due to the high affinity of 2C6, we also performed this assay with lower-affinity mutants, which were created by exchanging portions with 47e (as shown in Fig. 6). Binding of the Fr3 mutant, which retained some ADCC activity in clade B data, identified residues 524 and 596 as being involved in the epitope. The mutCDR2+Fr3 mutant additionally identified residue 596, which suggested involvement of site 579. Site 592 was identified as a residue by the CDR1 mutant. Site 588 was not resolved with any construct (Fig. 7A).

Projecting these amino acids onto the prefusion crystal structure of the HIV-1 Env trimer, we found that the side chains of residues 588, 592, and 596 all were oriented on the same face of the alpha-helical structure (Fig. 7B), whereas the side chain of residue 579 was oriented in the opposite direction. When residues 524, 579, 588, 592, and 596 were mapped on a three-dimensional structure (Fig. 7B), the result suggested that they contribute to the formation of a pocket that depended on the juxtaposition of two separate gp41 oligomers.

Exploring identified amino acids.

To further explore the binding of the epitope in the context of the alanine scanning mutagenesis data, we constructed a 25-mer peptide that encompassed the linear peptide identified and additionally includes R579 (Fig. 8). This peptide spans all amino acids identified by alanine scanning except G524. The peptide is recognized by 2C6 but is not bound by ID1 dominant Abs, the anti-Gag Ab (8B10), or 50-69 (Fig. 8B).

FIG 8.

Binding of various Abs to 25-mer synthetic peptide. (A) A 25-mer construct that encompassed the M peptide 145 and the arginine at position 579 was constructed. Mutants are labeled with numbering relative to reference sequence HXB2. Homology to the germ line sequence is represented by a dash. (B) 2C6 and other gp41-targeting Abs binding to the 25-mer germ line peptide. (C) 2C6 (black) and mutant Abs (shades of gray) binding to 25-mer and alanine replacement mutants.

Amino acid mutations at aa 592, 595, and 596 all ablated 2C6 binding. The R579A-mutated 25-mer demonstrated a significant decrease in 2C6 binding. For the mutants of 2C6 (Fig. 8C), exchanging the CDR2 region and either portion of the CDR3 ablates 2C6 recognition most severely. Similar to previous findings, framework 3 mutations were largely tolerated but still had diminished binding compared to that of native 2C6.

Binding to the 25-mer WT peptide and loss of binding in various mutated 25-mer peptides confirm the importance of amino acids at 579, 592, 595, and 596. This is consistent with our previous findings of 2C6 binding to number 9118 (RYLKDQQLLGIWGCS) but not to number 9119 (DQQLLGIWGCSGKLI) (Fig. 8A), despite both containing the identified sites 592, 595, and 596 (boldface text within peptides).

DISCUSSION

Here, we show that the epitope targeted by 2C6 is readily available and functional across multiple clades. We have mapped the epitope of our anti-HIV Ab 2C6 to a region at the end of the NHR and continuing into the beginning of the gp41 immune dominant loop, which lacks the necessity of the full C-C loop. Resolution of G524 by alanine scanning mutagenesis and mapping on three-dimensional HIV Env structures support that a pocket epitope is targeted by 2C6. To our knowledge, this structural epitope has not been described previously. Since this epitope can be targeted by high-affinity functional Abs, is present in multiple clades, is present in advanced vaccine constructs, such as SOSIP, and supports ADCC activity, this may be a significant site to consider in vaccine construct development. A more complete understanding of the epitope specificity of ADCC-mediating Abs, such as 2C6, is essential for developing effective immunization strategies that optimize protection by these Abs.

The main structure targeted by 2C6 appears to be an alpha helix on gp41, which is consistent with the resolution of the linearly separated amino acids (592, 595, and 596) that would align their side chains in the same orientation. Site 588 would also align along the same face of the alpha helix. However, aa 588 seems to affect binding only when the core peptide is shortened, as MN peptide (9118) shows robust binding and has the same amino acid at site 588 but is shifted by only two amino acids. Site 579 also maps to that same alpha helix but has mixed effects depending on the affinity of the 2C6 mutant. Notably, the side chain of aa 579 is not oriented in the same manner. It is possible aa 579 would affect the overall positioning of the helix.

The initial identification of site 524, on the C-terminal end of the fusion domain, was difficult to explain until projection onto trimerized Env proximally localized these regions across two oligomers of gp41. This epitope is most similar to interface binding Abs that include the broadly neutralizing fusion binding Abs PGT151 and VRC34 (3, 29). These Abs target the N-terminal portion of the fusion domain from sites 512 to 522 but do not have contact with 524 on X-ray diffraction analysis (30). Since PGT151 does resolve L592 but not any of the other amino acids associated with 2C6 binding, it is possible that the epitopes of PGT151 and 2C6 partially overlap. By peptide scanning, we did see repeated binding to a peptide within gp120; however, alanine scanning in this area did not show any significant results. Projection of this peptide onto the three-dimensional structure does not reveal an obvious area of accessibility (data not shown). We did not identify N88 by alanine scanning mutagenesis, which has been shown to be crucial for VRC34.01 binding (29). As longer peptides are recognized by 2C6, how much this epitope is truly dependent on interaction with amino acid 524 remains unclear. Further definition of epitope targeting in peptides and oligomers by cocrystallization studies are currently being pursued.

Further studies to more clearly define the structure of this epitope are warranted, as this epitope is near other significant sites on the Env. The Ab 50-69 competed with 2C6, but data support that these Abs do not target the same epitope. In peptide binding, 50-69 is dependent on a disulfide bridge (31). Our Ab 2C6 binds to foldon trimer, SOSIP trimer, and gp41 under reduced conditions, further confirming that 2C6 binding is independent of a disulfide bridge. The epitope of Ab QA255.067 (from 952-606; LLGIWGCSGKLICTT), identified by binding to similarly constructed VLPs (32), is near 2C6 but is also dependent on the disulfide bridge of the C-C loop sequence. Besides the fusion peptide-targeting Abs, 2C6 also maps near the hydrophobic pocket targeted by enfuvirtide (T20) and other fusion inhibitors in development (33). A number of our QtAbs can interfere with enfuvirtide in a fusion assay, but 2C6 failed to interfere with the function of T20 (34). Abs D5 (35) and HK20 have been shown to bind the region that is targeted by these fusion inhibitors. HK20 has similarity to 2C6, as it binds the hydrophobic pocket, resolves site 579 in cocrystallization studies (36), and has improved neutralization with Fab forms (17). The site at 579 has also been implicated in the protective IgA response in an HIV-exposed seronegative cohort, specifically mapping to the sequence LQAR (aa 576 to 579) (37). However, this immune protection is controversial (38).

As our Abs were cloned from an LTNP, it is intriguing to consider a role for specific epitope targeting contributing to this state. Abs can show protection in challenge models for both neutralizing (7, 39, 40) and nnAbs (10, 11). Cervical IgA responses against gp41 (41) have been confirmed in cohorts of chronically exposed noninfected individuals (42). Similar responses can be raised by gp41 vaccination and have shown protection in primate SHIV mucosal challenge (43). Most Abs have no effect on viremia (44), but recent studies suggest Abs targeting specific epitopes affect viremia. The most compelling evidence has come from recent trials on recombinant neutralizing Ab infusions. Initially shown with PGT121 in SHIV-infected rhesus monkeys (45), suppression of viremia was also demonstrated in humans after infusion of the broadly neutralizing Abs 3BNC117 (46) and 10-1074 (47). Infusions have also been shown to enhance neutralization (48) and CD8 responses and provide long-lasting stabilization of CD4 numbers (49). These data support a role for immunotherapies as prophylaxis and provide proof of concept that Abs that target specific epitopes can control viremia.

One of the most important restrictive factors in developing a vaccine is the challenge of replicating the native Env protein. The recent development of the BG505 SOSIP.664 gp140 trimer construct has advanced the ability to recapitulate the antigenic presentation of functional viral Env protein (20, 50). We show 2C6 binds SOSIP trimers with reasonable activity, which contradicts the initial theory that the SOSIP trimer is exclusively recognized by neutralizing Abs (13). Our finding is consistent with vaccination studies showing neutralizing Abs and nnAbs are produced by SOSIP trimers (51). Our QtAbs were identified and characterized initially with theoretically less native trimers, including uncleaved gp140 on VLPs and gp140 with foldon trimerization (17). Cleavage mutations create variable trimers with more open conformation, allowing for increased gp41 accessibility (52). Data supporting uncleaved gp140 foldon-based trimer recapitulation of native structure have not been corroborated (53). The fact that trimers embedded into virions or VLPs show differences in presentation of Env epitopes adds another layer of complexity to antigen design (54). As numerous SOSIP-based constructs are advancing as vaccine candidates (13, 55, 56) and 2C6 shows significant cross-clade ADCC activity, identification of the 2C6 epitope region may aid in understanding immunogenicity in ongoing trials. Since functionality and the pocket epitope of this region have not been previously known, the significance of targeting this epitope during vaccination is a question remaining to be determined.

Overall, we have defined a minimal thirteen-amino-acid domain for 2C6 peptide binding that contributes to an alpha helix at the base of a pocket between two gp41 oligomers on trimerized Env. Alanine scanning mutagenesis data, including resolving the remote amino acid at 524, further support this as an epitope enhanced in quaternary form on the native Env. As ADCC may protect against acquisition of HIV-1, a thorough exploration of Abs that have ADDC function is warranted. Due to the associated functional activity and its accessibility to a number of clades, further characterization of the structure of this epitope is warranted.

MATERIALS AND METHODS

Full-length Ab expression.

Natively paired heavy- and light-chain Ab plasmid DNA were amplified using the PureYield plasmid maxiprep system (A2392; Promega) by following the manufacturer’s instructions. Plasmids were then cotransfected into Free Style 293-F cells. Briefly, 50 μg of DNA per 100 ml of culture was mixed with 3 ml of Opti-MEM I reduced-serum medium (31985062; Thermo Scientific) and sterile filtered. A volume of 1 ml of 1 mg/ml polyethylenimine (PEI) per 100 ml of culture next was added to 3 ml of Opti-MEM I reduced-serum medium and sterile filtered. The DNA and PEI solutions were then mixed dropwise, vortexed, and incubated for 12 min at room temperature (RT). The mixture was added to 100 ml of Free Style 293-F cells seeded in Free Style 293 expression medium (12338018; Thermo Scientific) at a density of 1 × 10⁶ cells/ml and left to incubate for 6 days at 37°C and 8% CO₂ on a shaker at 130 rpm.

Ab purification.

On the sixth day of transfection, culture supernatants were collected and cleared by centrifugation at 3,000 × g for 15 min at 4°C and then filtered through a 0.2-μm filter. Following filtration, immunoglobulins were purified using a 5 ml HiTrap Protein G column (17-0404-01; GE Healthcare) on an AKTA fast protein liquid chromatograph (GE Healthcare, Piscataway, NJ). The IgG fraction was recovered from the column by elution with glycine buffer (100 mM, pH 2.5). The fractions containing IgG were pooled and exchanged with 1× phosphate-buffered saline (PBS) (pH 7.4) via 3 washes using Amicon Ultra-2 ml Ultracel with a 10-kDa cutoff (Millipore Corp., Bedford, MA). A Tris-HCl neutralization step was avoided as it may interfere with later biotinylation steps. Ab concentrations were measured on a NanoDrop Lite spectrophotometer (Thermo Scientific).

Ab biotinylation.

Ab biotinylation was performed as previously described (14). First, 100 to 200 μl of 3-mg/ml Abs was equilibrated in carbonate buffer (pH 9.3) using a 10- or 50-kDa-cutoff Amicon Ultra-4 centrifugal filter unit (Millipore Corp., Billerica, MA) per the manufacturer’s instructions. Ab concentrations were measured using a NanoDrop spectrophotometer (Thermo Scientific). Sulfosuccinimide biotin (Sigma-Aldrich) was added at a 30-fold molar excess over the concentration of Ab (approximately 90 μg/mg of IgG). The mixture was incubated for 30 min to 4 h with gentle shaking, protected from light, at RT. Tris-HCl (1 M, pH 7.5) was added up to 10% by volume to bind residual biotin molecules. The biotinylated Abs were separated from residual biotin complexes and equilibrated to PBS (pH 7.4) using a Pierce 7,000-molecular-weight-cutoff polyacrylamide desalting spin column (89849; Thermo Scientific) per the manufacturer’s protocol.

General Ab binding ELISAs.

In brief, Immulon 2-HB plates were coated with respective antigen in 0.1 M sodium bicarbonate buffer, pH 8.4, overnight at 4°C. Plates were blocked with 10% fetal bovine serum (FBS) in PBS for 2 h at 4°C. After removing blocking buffer, 100 μl of the respective Ab diluted in 10% FBS was added and incubated at 37°C for 1 h. After 1 h, plates were washed with washing buffer (PBS containing 0.1% Tween 20) three times. After washing, 100 μl of horseradish peroxidase (HRP)-conjugated secondary Ab in blocking buffer was added and incubated at RT for 1 h. Plates were washed three times with washing buffer, 50 μl Ultra-TMB (Thermo Scientific) substrate was added to each well, and reactions were stopped by adding 50 μl of 2 N sulfuric acid. Absorbance was measured at an optical density at 450 nm (OD₄₅₀) within 30 min.

SOSIP trimer binding ELISA.

SOSIP trimer creation was previously characterized in respective references for version four (Zm197m v4.1, AMC008 v4.2, and B41 v4.2 [57]) and version five (BG505 v5.2; [58]). In brief, half-area plates were coated overnight with Galanthus nivalis lectin (Vector Laboratories) in 0.1 M sodium hydrogen carbonate (NaHCO₃) and blocked in 1% casein in Tris-buffered saline (TBS) (ThermoFisher) at RT for 1 h. All protein dilutions were made in casein buffer, and all wash steps were performed with 1× TBS. After washing, Env protein constructs were added at 2 μg/ml and incubated at RT for 2 h. Plates were washed and Abs were added for 2 h at RT with concentrations ranging from 1 μg/ml up to 5 μg/ml. Secondary goat-anti-human IgG antibody (1:3,000) (Sera Care) was added after washing and incubated at RT for 1 h. After washing with 1× TBS plus 0.05% Tween, development reactions were started with 0.1 M sodium acetate–0.1 M citric acid with TMB (1:100) (Sigma-Aldrich) and H₂O₂ (1:3,000) and stopped with 0.8 M H₂SO₄. OD₄₅₀ was measured using a SPECTROstar Nano spectrophotometer from BMG Labtech.

Competition ELISA.

Ab was equilibrated first to carbonate buffer (pH 9.3) using 50-kDa-cutoff AmiconUltra-4 centrifugal filter units (Millipore Corp., Billerica, MA) per the manufacturer’s instructions. Concentration of the residual Ab was determined using a NanoDrop spectrophotometer (Thermo Scientific). Sulfosuccinimide biotin (Sigma-Aldrich, St. Louis, MO) was added at a 30-fold molar excess over the concentration of Ab (approximately 90 μg per mg of Ab). This solution was mixed and incubated for 30 min with gentle shaking at RT. Trizma-HCl (pH 7.5) was added up to 10% by volume to bind residual biotin molecules. The biotinylated Ab was separated from residual biotin complexes and equilibrated with PBS (pH 7.4) using a G-25 Sephadex Nap-5 (GE Healthcare, Piscataway NJ) column per the manufacturer’s protocol. Biotinylated Abs were competed with a panel of known unlabeled HIV-specific Abs (T32, 98-6, F240, 50-69, and 2F5), as listed in the Results. EC₅₀ values were used to normalize Ab competition levels. ELISAs were performed similarly to the method of our previously described work (17) but using streptavidin-HRP (Southern Biotech, Birmingham, AL)-conjugated secondary Ab reagent for detection of binding.

For the BG505 SOSIP competitions, His-tagged SOSIP ELISA nickel-nitrilotriacetic acid (Ni-NTA) plates from Qiagen, Inc., were used. In brief, SOSIP in TBS at 1,500 ng/ml was incubated on Ni-NTA plates at 4°C overnight. The next day, plates were washed three times with TBS and blocked with 2% milk in TBS for 30 min at RT. This was washed thrice with TBS, and respective Abs were added and incubated at RT for 2 h. After incubation, plates were washed thrice with TBS, and then peroxidase-labeled secondary Ab in TBS with 2% milk was added and incubated for 1 h at RT. After washing thrice with PBS with Tween 20 (PBST), 100 μl of TMB substrate was added, the reaction was stopped with 2 M sulfuric acid, and absorbance was measured at 450 nm.

Immunoblot analysis.

Native gels were run with using a Bio-Rad miniformat one-dimensional electrophoresis system with 4 to 20% gradient gels (Mini Protean gel; 456-1094; Bio-Rad). Samples (3 μg) were mixed with 2× native sample buffer (Bio-Rad) to a final volume of 10 μl. Native gels and nonnative gels (in the presence of SDS with and without beta-mercaptoethanol) were run at 100 V until running dye crossed to the bottom of the gel. Proteins were transferred to polyvinylidene difluoride membrane using an iBlot dry blotting system from Invitrogen by Life Technologies (IB10001) according to the manufacturer’s instructions. The blot then was blocked with 1% bovine serum albumin for 1 h at RT. The blot was rinsed with wash buffer (PBS with 0.1% Tween), and then primary Ab diluted in the PBS containing 1% bovine serum albumin was added. This was incubated overnight at 4°C on a rocker. The blot was then washed extensively in wash buffer (3 times for 10 min each time) with gentle agitation. Blots were probed with secondary Ab (alkaline phosphatase-conjugated goat-anti-human Ig; 2010-04; Southern Biotech, Birmingham, AL) in 1% bovine serum albumin and incubated for 1 h at RT with gentle agitation. Membrane was washed with gentle agitation 4 times for 5 min each time in wash buffer, three times for 5 min each time in PBST, and twice for 5 min each time in PBS. One-step nitroblue tetrazolium–5-bromo-4-chloro-3-indolylphosphate color developer (34042; Thermo Scientific) solution was added and incubated as recommended by the manufacturer to visualize protein bands.

Ab-dependent cellular cytotoxicity assay.

Rapid and fluorometric ADCC (RFADCC) assay was performed as described previously (59–62). Briefly, CEM-NKr target cells were double stained with PKH (a membrane stain) and carboxyfluorescein succinimidyl ester (CFSE; a cytosolic stain) and coated with 50 μg/ml of either MN gp41 protein (NIH AIDS Reagent Program) or ZA1197 gp41 protein (IT-001-0052p; Immune Technology, New York, NY). The coated target cells (5,000 cells/well) were incubated with the Ab for 10 min to allow binding of the Ab to gp41. Effector cells (HIV-negative donor peripheral blood mononuclear cells) next were added at an effector-target ratio of 50:1. The mixture was incubated for 4 h at 37°C. After the incubation, the cells were fixed in 1% paraformaldehyde in PBS. When ADCC occurs, the target cell membrane is permeabilized, leading to the loss of CFSE and retention of PKH. ADCC was measured by flow cytometry as the percentage of phycoerythrin (PE)-positive, CFSE-negative cells among all PE-positive cells with background (average ADCC mediated against uncoated target cells) subtracted. Background was set to 3 to 5%. Influenza Ab was included as a negative control and HIVIG as a positive control (both used at 500 ng/μl). For our analysis, we set the percent killing as observed with HIVIG to 100% and normalized the rest of the samples to HIVIG. Relative ADCC activity was calculated by dividing the percent activity at that dilution minus background.

Shotgun mutagenesis epitope mapping.

To perform shotgun mutagenesis epitope mapping, an alanine scanning mutagenesis library was created for HIV-1 (strain KNH1144; GenBank accession no. JQ715384), as previously described (28). Briefly, parental plasmid-expressing codon-optimized gp160 Env protein was used as a template to make the mutation library. Each residue from amino acids 30 to 708 was individually mutated to alanine, with alanine residues mutated to serine. In total, 678 HIV Env mutants were generated (99.9% coverage). Sequences were confirmed, and mutants were transiently expressed in human HEK-293T cells for 22 h in 384-well plates. Each plate included eight positive- and four negative-control wells. Cells were fixed in 4% (vol/vol) paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) in PBS plus calcium and magnesium (PBS++).

Cells were stained with purified MAbs at 0.1 to 2.0 μg/ml diluted in 10% normal goat serum (NGS) (Sigma-Aldrich, St. Louis, MO). The optimal primary Ab concentration was separately determined for each Ab by titrating the Ab immunofluorescence against WT gp160 to ensure that signals were within the linear range of detection and to ensure a ≥5-fold difference between signal and background. Abs were detected by using 3.75 μg/ml Alexa Fluor 488-conjugated secondary Ab (Jackson Immuno Research Laboratories, Pike West Grove, PA) in 10% NGS. Cells were washed three times with PBS. Mean cellular fluorescence was detected by using an Intellicyt high-throughput flow cytometer (Albuquerque, NM). Ab reactivities against each mutant gp160 clone were calculated relative to WT HIV Env reactivity by subtracting the signal from mock-transfected controls and normalizing to the signal from WT HIV Env-transfected controls.

Residues required for MAb binding were identified as being critical to the MAb epitope if reactivity of the test MAb was lost (<20 to 35% of the value for the WT), but reactivity of control Abs (>70%) was retained. This counterscreening strategy enabled exclusion of gp160 mutants that were locally misfolded or had defects in trafficking or expression (28). Critical amino acids required for Ab binding were visualized on the crystal structure of the BG505 SOSIP.664 HIV-1 Env trimer (PDB entry 4ZMJ) (63).

Peptide binding assay.

Group M consensus peptides (HIV-1 consensus group M Env peptide set 9487; obtained through the NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH) were dissolved in 10% dimethyl sulfoxide in PBS. ELISA plates were coated with 50 μg per well of peptide and incubated overnight at 4°C on a rocking platform. Excess peptide was washed off, and plates were blocked with 10% bovine serum albumin in PBS. Test Abs at a concentration of 100 ng/ml then were added and incubated at 37°C for 1 h. After washing, secondary goat anti-human IgG (H+L) (Southern Biotech, Birmingham, AL) then was incubated for 1 h at RT. After washing, TMB substrate (Pierce, Loves Park, IL) was added and color development was halted with 2 N sulfuric acid. Optical density was read at 450 nm, and data were analyzed with Prism Software (GraphPad, La Jolla, CA).