|Home | About | Journals | Submit | Contact Us | Français|
Broad and potent neutralizing antibody (BNAb) responses are rare in people infected by human immunodeficiency virus type 1 (HIV-1). Clearly defining the nature of BNAb epitopes on HIV-1 envelope glycoproteins (Envs) targeted in vivo is critical for future directions of anti-HIV-1 vaccine development. Conventional techniques are successful in defining neutralizing epitopes in a small number of individual subjects but fail in studying large groups of subjects. Two independent methods were employed to investigate the nature of NAb epitopes targeted in 9 subjects, identified by the NIAID Center for HIV/AIDS Vaccine Immunology (CHAVI) 001 and 008 clinical teams, known to make a strong BNAb response. Neutralizing activity from 8/9 subjects was enhanced by enriching high-mannose N-linked glycan (HM-glycan) of HIV-1 glycoproteins on neutralization target viruses and was sensitive to specific glycan deletion mutations of HIV-1 glycoproteins, indicating that HM-glycan-dependent epitopes are targeted by BNAb responses in these subjects. This discovery adds to accumulating evidence supporting the hypothesis that glycans are important targets on HIV-1 glycoproteins for BNAb responses in vivo, providing an important lead for future directions in developing NAb-based anti-HIV-1 vaccines.
The envelope glycoproteins (Envs) of human immunodeficiency virus type 1 (HIV-1) are the sole viral target for neutralizing antibodies (NAbs) and are thus the central target for NAb-based vaccine development. A successful vaccine must be able to elicit broad and potent NAbs (BNAbs) capable of neutralizing diverse strains of primary HIV-1. The elusiveness of this goal is rooted in the rareness of BNAb responses during natural HIV-1 infections to serve as a road map for designing efficacious immunogens, adjuvants, and vaccination strategies. Fortunately, some HIV-1-infected people do successfully make BNAb responses after prolonged infection, providing limited but critical resources for characterization of natural BNAb responses to HIV-1 (5, 13, 18, 24, 33, 38, 40, 46, 55). Clearly defining the epitopes on HIV-1 Envs targeted by natural BNAb responses is a critical first step in this process and has been under intense pursuit in the field. The generation of monoclonal antibodies (MAbs) is a successful technique for defining NAb epitopes in individual patients. Characterization of MAbs such as b12, VRC01, 2G12, PG9/PG16, 2F5, 4E10, etc., has helped in the identification of major neutralizing epitopes, including CD4BS, CD4i, MPER, and high-mannose glycan (HM-glycan) (9–11, 42, 51, 55, 60). Recently, a panel of glycan-dependent NAbs was characterized for 4 subjects in a large-scale collective effort (54). Nevertheless, it is still not clear whether these known epitopes account for the major proportion of BNAb responses in the HIV-1-infected population (5, 32, 38, 44, 54, 56). Various approaches have been developed by modifying these known epitopes in recombinant HIV-1 Envs or in neutralization target viruses. Vigorous application of these methods has permitted characterization of the epitope specificity of only a few BNAb responses, failing in most cases (5, 38, 59). Thus, defining NAb epitopes on a population-wide scale has yet to be accomplished. As a consequence, the dominant form of BNAb responses during natural HIV-1 infection is not clear.
HIV-1 Envs have about 24 to 28 potential N-linked glycosylation signals. Virus-associated Envs contain almost exclusively N-linked glycans (N-glycans), with a minimal level of O-linked glycans (4, 15, 21, 36, 37, 47). N-glycans have distinct forms which can roughly be categorized as high-mannose N-glycan (HM-glycan) and complex-type N-glycan (C-glycan). After the initial glycosylation reaction (coupled with peptide translation) in the endoplasmic reticulum, trimming of the termini of the protoglycan results in a glycan that has mannose as the terminal residues, known as HM-glycan (28, 53). Most glycans are further modified by the addition of various monosaccharide residues, gaining complexity in terms of residue types/modifications and branching structure, and thus are known as C-glycans. HM-glycan and C-glycan are different in size and electric charge and thus perform different biological functions. Natural mammalian glycoproteins contain predominantly C-glycan (53), so a high content of HM-glycan on the virion-associated Env of HIV-1 logically makes it somewhat foreign to the human immune system, as well as potentially immunogenic (15, 17, 37). In contrast, recombinant HIV-1 Envs carry mainly C-glycan and a relatively low level of HM-glycan even when expressed in mammalian cells such as 293T cells and CHO cells (21–23, 30, 37, 43). The difference in glycan profiles between recombinant Envs and virion-associated Envs could provide a testable hypothesis for the failure of recombinant Envs as immunogens to induce NAb responses. Furthermore, the glycan profile difference also provides a simple explanation for the ineffectiveness of experimental techniques based on recombinant HIV-1 Envs in characterizing natural NAb responses if the majority of BNAb responses target HM-glycan. In this study, we tested the hypothesis that naturally occurring BNAb responses during HIV-1 infection frequently target epitopes composed of HM-glycan.
All patients were recruited and samples collected under the direction of the Center for HIV/AIDS Vaccine Immunology (CHAVI), and study of these samples was protected by IRB protocol 2002P-000182, approved by the Committee for Clinical Research at Beth Israel Deaconess Medical Center.
Recombinant HIV-1 encoding firefly luciferase and pseudotyped with wild-type (WT) or mutant Envs was produced as previously described (65). Briefly, 293T cells in 100-mm tissue culture dishes were cotransfected overnight with the pHIV-1-Luc luciferase genomic vector, the pCMVΔP1ΔenvpA packaging vector, and a pSVIIIenv vector expressing the selected Env at a ratio of 3:1:1, using LipoD293 DNA in vitro transfection reagent (SignaGen Laboratories). The transfection mixture was removed after overnight incubation, and 10 ml of fresh medium was added for virus production for 24 to 36 h. To produce recombinant HIV-1 reporter viruses under the selection of α-mannosidase inhibitors, fresh medium containing the desired concentration(s) of kifunensine (Tocris Bioscience) or swainsonine (Cayman Chemical) was added for virus production. All viral stocks to be compared directly were prepared as a set.
The infectivity of HIV-1 reporter viruses was measured in a single-round entry assay by incubation of the viruses with Cf2Th-CD4/CCR5 target cells in a 96-well format, using standard protocols as described previously (38). To quantify virus infectivity, the mean value and range of variation of luciferase activity from the duplicate wells were measured and reported in arbitrary luciferase units.
For neutralization assays, serial dilutions of the neutralizing agent, i.e., antiserum/plasma or MAb, were made in cell culture medium in such a volume as to produce the designated final concentration after the target virus was added. The virus-antibody mixture was incubated at 37°C for 2 h, and its residual infectivity was determined using the single-round entry assay described above. The residual infectivity (%) was defined as the infectivity measured at a given concentration of the neutralizing agent divided by the infectivity of the same virus mock treated with cell culture medium. All experiments were performed at least three times. Comparable results were achieved, and a typical set of results are reported. Sources for MAbs were as follows: 2G12, b12, and 2F5 were obtained from Polymun Scientific; E51, 17b, and 48d were a gift from James E. Robinson (Tulane University); F105 was a gift from Joseph Sodroski (Dana-Farber Cancer Institute) (41); VRC01 (HIV-1 gp120 MAb) was obtained through the HIV AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH, and was a gift from John Mascola (Vaccine Research Center, NIH) (59); and PG9 and PG16 were obtained through the International AIDS Vaccine Initiative (IAVI), New York, NY, and were a gift from Dennis Burton (The Scripps Research Institute) (55).
Wild-type and glycan deletion mutant HIV-1YU2 and HIV-1JR-FL gp160s were expressed from the pSVIIIenv vector (48). Mutants were created by the PCR-based QuikChange protocol (Stratagene). Glycan deletion mutations were designed to replace the asparagine residue in the canonical NXS/T glycosylation signal with different residues used in some strains of HIV-1. The integrity of construction was confirmed by DNA sequencing of the entire env reading frame. The names of the mutants designate the wild-type amino acid residue in single-letter code, the residue number, and the substituted amino acid. Residue numbering is based on that of the prototypic HIV-1HXBc2 gp160, according to current conventions (27).
To map all potential N-linked glycans targeted by BNAb responses in subject antisera, a screening assay was designed by modifying the neutralization assay described above. All glycan deletion mutants and the parental wild-type Envs were tested as a set in a single experimental session of neutralization with a single dilution of a given antiserum. The concentration of antiserum used was close to the 50% inhibitory concentration (IC50) for wild-type Env of a given antiserum, as determined in preliminary experiments. The residual infectivity (RI%) of the virus was determined using the single-round entry assay. In each experimental session or test, the RI% of the wild-type virus was used as the baseline of neutralization of an antiserum, on which the RI% of all derivative mutants was judged. A mutation was deemed to have no effect on neutralization of a given antiserum if the following was true: 1/2 × RI% of WT < mutant RI% < [100% - 1/2(100% − RI% of WT)]. A mutant was judged to be significantly resistant (R) to an antiserum relative to its parental Env if the mutant RI% was ≥[(100% − 1/4(100% − RI% of WT)]. A mutant that was relatively less resistant than the R mutants but more resistant than the “no effect” ones was marked as marginally resistant (MR). Conversely, a mutant was read as significantly sensitive (S) to an antiserum relative to its parental Env if the mutant RI% was ≤1/4 × RI% of WT. Similarly, marginally sensitive (MS) mutants had a neutralization sensitivity between those of the S and “no effect” mutants. Each subject antiserum was tested a total of 4 times, and a mutant was designated resistant or sensitive (or variations thereof) based on concordant findings in 3 of 4 repeat tests. The results from this screening assay were further validated in a neutralization assay in which a full range of dilutions of antiserum was used.
We studied a population of 308 chronically HIV-1-infected subjects from several African countries, the United Kingdom, and the United States (recruited by the clinical team of CHAVI). The NAb responses were measured using a conventional neutralization assay based on single-round entry into Tzm.bl reporter cells by recombinant viruses pseudotyped with HIV-1 Envs of different strains (31). In this study, we selected 9 subjects whose antisera neutralized at least 10/12 primary clade B and C HIV-1 strains with an IC50 titer of 1:100 and at least 8/12 strains with a titer of 1:200; the exact neutralization breadth and potency of these 9 subjects have been presented previously (49). This level of neutralizing activity represents one of the best panels of patients with strong BNAb responses to HIV-1 identified so far. These 9 subjects included 8 individuals from Africa and 1 from the United Kingdom and were mostly women infected by clade A and C HIV-1 strains. We employed two independent methods to investigate whether HM-glycan is targeted in these 9 subjects.
To determine whether specific glycans are a part of the BNAb epitopes targeted in our subjects, we created a panel of 28 HIV-1YU2 Env mutants in which all potential N-linked glycosylation sites were deleted individually. Most of these mutants were fully functional, with entry levels within a 2-fold range of that of wild-type HIV-1YU2 Env (Fig. 1A). The N611S mutation caused a reduction of entry of close to 13-fold compared with wild-type Env, but the resultant mutant Env was competent enough to produce viral stocks for neutralization assays. N262S, N332S, and N386T mutations significantly disrupted the entry function of HIV-1YU2 Env, and these mutants could not produce viral stocks for use in neutralization assays. To cover the glycans at these locations, 3 analogous mutants were created in HIV-1JR-FL Env. The N332S and N386T mutants had reductions of entry function of 3.5- and 2-fold, respectively, compared with wild-type HIV-1JR-FL Env (Fig. 1B). The N262S mutation resulted in a nonfunctional HIV-1JR-FL Env, preventing further investigation of the glycan at this position (Fig. 1A and B). Combined, this panel of glycan deletion mutants covered all conserved N-glycans of clade B HIV-1 Env except for the one at position 262. If a specific glycan was a major component of a BNAb epitope targeted in a study subject, deletion of the glycan at that position was expected to cause an increase in neutralization resistance to the given subject antiserum compared to the wild-type parental Env of either HIV-1YU2 or HIV-1JR-FL.
We developed a screening assay to conserve antisera due to the limited supply of such valuable reagents. In this assay, the wild-type and mutant Envs of HIV-1YU2 and HIV-1JR-FL were compared in a single experimental set with an individual antiserum at a concentration close to its IC50 for the wild-type Envs. Compared with its corresponding wild-type Env, an Env mutant was judged to have no effect on neutralization by a given antiserum if its neutralization was within a 2-fold margin of that of the wild-type Env control. A mutant was deemed to have a significant increase in neutralization resistance or an increase in neutralization sensitivity if the difference of neutralization between the mutant and parental wild-type Envs was ≥4-fold. Otherwise, a marginal increase in neutralization resistance or neutralization sensitivity was recorded if the difference of neutralization was 2- to 4-fold. The exact same experiment was performed four times for each study subject. Table 1 presents sample results of this screening assay for subject C1-536. If a mutant had a significant increase in neutralization resistance by an antiserum in at least 3 of the 4 repeat experiments, the glycan at that specific location was judged to be a part of the BNAb epitope targeted in the subject. Conversely, if a mutant had a significant increase in neutralization sensitivity by an antiserum in at least 3 of the 4 repeat experiments, the glycan at that specific location was considered to have a significant influence on the BNAb epitope but not to be part of the epitope itself. Table 2 summarizes the screening results for all 9 subjects. For 8 of 9 subjects, at least one glycan deletion mutant resulted in a significant increase in neutralization resistance, suggesting that glycans are targeted for BNAb responses in these subjects. For subjects C1-219, C1-457, C1-269, and C1-763, some glycan deletion mutations led to an increase in neutralization in a subject-specific fashion, suggesting that these glycans might be proximal to or otherwise capable of influencing the BNAb epitope(s) targeted in these subjects. Interestingly, antisera from these subjects had relatively higher neutralization potencies among the panel of 9 subjects (data not shown). Mutations at glycans 160, 197, 301, and 616 resulted in an increase in neutralization sensitivity to almost all BNAb antisera; thus, glycosylation at these sites likely has a global impact on the neutralization sensitivity of HIV-1 Env.
A full titration of the neutralization reaction was conducted to compare the titers of a given antiserum in neutralizing the wild-type parental Envs and the Env mutant(s) selected as having a significant increase in neutralization resistance (Table 2). The purpose was to validate the results from the above screening assay. The data would also help to assess if the neutralizing activity attributed to the glycan-dependent epitopes represents the dominant BNAb responses in the study subjects. In all cases, the glycan deletion mutation resulted in a neutralization titer decrease of at least 2 1:2 dilutions at some point on the neutralization curve; in fact, most cases were close to full escape from neutralization by the subject antiserum (Fig. 2). These results confirm that glycans at these locations contribute to the neutralizing epitopes targeted by BNAbs in these eight subjects. A total of five glycans are involved in the BNAb epitopes targeted in these subjects.
Glycans 332 and 386 are located close to each other on the gp120 outer domain and thus are likely to form an epitope similar to the 2G12 epitope (29, 60). Epitopes involving these two glycans may include a cluster of glycans around these two glycan residues, which we have designated the outer domain glycan cluster (ODGC). To test whether epitope mapping using the N332S and N386T mutants is specific for the ODGC, we investigated whether these two mutations led to neutralization escape from well-known NAbs. Both N332S and N386T mutant Envs fully escaped neutralization by 2G12 compared with the parental HIV-1JR-FL Env, as expected (Fig. 3, top graph). Importantly, none of the other NAbs tested neutralized either of these two mutant Envs differently from the wild-type control. Thus, data from the N332S and N386T mutants permitted faithful mapping of the ODGC-related epitopes.
Glycans 234 and 241 are spatially close to each other and are located on the β-sandwich structure of the HIV-1 Env protein (20, 39, 64). The β-sandwich is located in the gp120 inner domain and is important for trimeric association between the gp41 and gp120 subunits (63). Glycans 234 and 241 abut the CD4BS and likely reside between the CD4BS and the gp41 ectodomain/viral membrane, according to the current understanding of HIV-1 Env trimer structure. Glycans 234 and 241, possibly in conjunction with some adjacent glycans, form a cluster of epitopes which we have designated the inner domain glycan cluster (IDGC), because they are in the gp120 inner domain. The IDGC is spatially distant from the ODGC on X-ray crystal structures of the HIV-1 gp120 core and thus is likely distinct from the ODGC (29, 39). On X-ray crystal structures of the HIV-1 gp120 core, glycan 362 is rather apart from and located on a different facet of gp120 from glycans 234 and 241 (39), and thus it is uncertain whether glycan 362 could form an integrated epitope with glycans 234 and 241. To test if the N234S, N241S, and N362K mutants have specificity for mapping IDGC epitopes, we investigated whether these mutations led to a change in neutralization by well-known NAbs. The N234S and N241S mutations had no influence on neutralization by most NAbs tested, except that the N241S mutation had an increase in neutralization sensitivity to NAb 17b (Fig. 4). It is possible that loss of glycan 241 might have affected the CD4i site, specifically the 17b epitope, through subtle conformational changes in the Env trimer. Because the N241S mutant gained sensitivity to 17b, this factor does not substantially impact the specificity of mapping the IDGC based on an increase of neutralization resistance by the N234S and N241S mutants in our study. The N362K mutation did not affect neutralization by selected MAbs in most cases but gave an increase in neutralization sensitivity to 17b and an increase in neutralization resistance to 48d, to a minor degree (Fig. 4, bottom right panel). Both 17b and 48d target epitopes of the CD4i site, suggesting that the CD4i epitopes have been perturbed considerably in this mutant. The effect on 48d neutralization suggested that the N362K mutant could not be employed for mapping the IDGC-related epitopes with perfect specificity. Fortunately, the mutant was involved in results for subject C8-258 only. Importantly, the N241S mutation led to neutralization escape from the C8-258 antiserum by about 10-fold, indicating that the IDGC is targeted in this subject. In summary, glycan deletion mutations at positions 234 and 241 can faithfully map the IDGC-related BNAb epitopes.
In summary, our results demonstrate that specific glycans are involved in the BNAb epitopes targeted in 8 of 9 study subjects. The IDGC is involved in a novel glycan-dependent epitope centered on glycans 234 and 241, selectively targeted in subjects C1-536 and C8-258. The ODGC is centered on glycans 332 and 386, similar to the well-known HM-glycan-dependent epitope of 2G12, and is specifically targeted in subjects C1-219, C1-175, and C1-763. Both IDGC-related and ODGC-related epitopes are targeted by BNAb responses in subjects C1-457, C1-534, and C1-269.
Kifunensine and swainsonine are plant alkaloids that are effective inhibitors of α-mannosidases I and II, respectively, in mammalian cells (53). Treating cells with a sufficient concentration of kifunensine inhibits cellular α-mannosidase I activity and blocks trimming of the termini of protoglycans and subsequent steps of glycan modifications for C-glycan formation (45). The major consequence is enrichment of HM-glycan, primarily the Man9GlcNAc2 species, composed of 9 mannose residues and 2 N-acetylglucosamine residues, on glycoproteins produced in the cells under selection. Swainsonine inhibits α-mannosidase II activity selectively, as well as α-mannosidase I activity at very high concentrations (7, 52). Swainsonine treatment of cells also results in the enrichment of HM-glycan, but mainly of the Man(4-6)GlcNAc2 species. Both kifunensine and swainsonine treatments enrich HM-glycan on HIV-1 Envs, leading to higher neutralization sensitivity to MAb 2G12, which binds specifically to the terminal mannose residues of HM-glycan (45). We took advantage of these experimental tools to investigate whether enriching HM-glycan on Envs of recombinant HIV-1 reporter viruses could enhance their neutralization sensitivity to our subject antisera. An affirmative answer would provide further evidence in support of the above finding that N-glycans are involved in BNAb responses in our study subjects. Although kifunensine and swainsonine enrich different species of HM-glycan, their effects on HIV-1 Env were indistinguishable in all experiments performed in our study (data not shown). We report only data obtained with kifunensine here for the sake of space and clarity of presentation.
First, we investigated the impact of enriching HM-glycan by kifunensine treatment on the structural and functional integrity of HIV-1 Envs as well as on their antigenicity in a neutralization assay. We performed preliminary experiments by titrating the effect on virus entry and neutralization sensitivity by treating virus-producing cells with 3.125 to 400 μM kifunensine. Under our experimental conditions, treatment with concentrations of kifunensine from 50 to 400 μM produced stable results in studying both HIV-1YU2 and HIV-1JR-FL, whereas lower concentrations of kifunensine did not (data not shown). We therefore chose to further investigate viruses produced in the presence of kifunensine at a concentration of 50 or 200 μM. We confirmed that treatment with kifunensine at concentrations from 25 to 400 μM led to significant enhancement of the neutralization sensitivity of viruses carrying HIV-1YU2 gp160 to 2G12, while mock-treated HIV-1YU2, as previously reported, was not sensitive to neutralization by 2G12 (Fig. 5A) (61, 65). Enhancement of 2G12 neutralization by kifunensine treatment was observed with HIV-1JR-FL as well (see Fig. S1A in the supplemental material). Enhancement of 2G12 neutralization was appreciably heightened by 200 μM kifunensine compared with 50 μM kifunensine. This may be due to better enrichment of HM-glycan on Env at higher concentrations of kifunensine. However, it may also be that absolute blocking of the glycan terminus trimming activity by higher concentrations of kifunensine has some influence on conformational maturation of the HIV-1 Env trimer within the endoplasmic reticulum, consequently changing the neutralization sensitivity. These results demonstrated that kifunensine treatment enriches neutralizing epitopes composed of HM-glycan, and they also served as an experimental control for the successful enrichment of HM-glycan in the following experiments.
To investigate the impact of kifunensine treatment of virus-producing cells on the structural and functional integrity of HIV-1 Envs, we investigated whether kifunensine treatment influenced the entry function of HIV-1 Envs and their sensitivity to known MAbs. Treatment with kifunensine had no impact on the entry function of HIV-1YU2 and HIV-1JR-FL Envs in a single-round entry assay conducted within the linear range of titration (Fig. 5B; see Fig. S1B in the supplemental material). Next, we investigated whether kifunensine treatment affected binding and neutralization by recombinant CD4, the natural ligand of HIV-1 Env. CD4 immunoglobulin (Ig) contains the 4 ectodomains of human CD4 fused to the Fc fragment of human IgG1, which is capable of blocking HIV-1 entry, in semblance to NAbs (12, 50, 58). Treatment with 50 μM kifunensine did not significantly affect neutralization of HIV-1YU2 by CD4 Ig (Fig. 5C). Interestingly, however, 200 μM kifunensine slightly increased HIV-1YU2 Env's sensitivity to inhibition by CD4 Ig. Finally, we investigated whether enrichment of HM-glycan significantly changed the neutralization sensitivity of HIV-1 Env to known NAbs. Kifunensine treatment at either 50 μM or 200 μM did not change the neutralization sensitivity of HIV-1YU2 to NAbs targeting CD4BS epitopes (b12, VRC01, and F105), CD4i epitopes (E51 and 17b), or MPER (2F5) (Fig. 5D, top and middle graphs). Similar findings were also observed with HIV-1JR-FL (see Fig. S1C in the supplemental material). However, kifunensine treatment resulted in a reduction of neutralization sensitivity to MAb 48d, targeting another CD4i epitope (Fig. 5D, lower left graph). Interestingly, the impact on the CD4i site was strain specific, because kifunensine treatment of HIV-1JR-FL did not affect the 48d epitope but suppressed neutralization by 17b (see Fig. S1C in the supplemental material). While 50 μM kifunensine suppressed the neutralization potency of PG9 and PG16 with HIV-1YU2, as reported previously (16), 200 μM kifunensine had significantly less suppression of PG9 neutralization and even slightly enhanced neutralization of PG16 (Fig. 5D).
Taken together, the data show that kifunensine treatment had no appreciable deleterious impact on the functional integrity of HIV-1 Env and did not change its neutralization sensitivity to most NAbs whose epitope does not involve glycan. At lower concentrations of kifunensine, e.g., 50 μM, the main effect was enrichment of HM-glycan, leading to enhanced neutralization by antibodies such as 2G12. At higher concentrations of kifunensine, e.g., 200 μM, the effect may have two components, i.e., enriching HM-glycan and causing some conformational changes in the HIV-1 Env trimer. Both of these effects may contribute to the influence on neutralization at glycan-dependent epitopes such as PG9 and PG16, whose epitopes contain both glycan and peptic elements (16, 55). The effect on Env conformation induced by a high concentration of kifunensine is indicated by reduced neutralization sensitivity to 48d at the CD4i site (62). Depending on the nature or composition of NAbs in a BNAb antiserum, the final outcome of the effect on neutralization of viruses selected by high concentrations of kifunensine is likely a congregated impact from both effector components. Within our experimental system using HIV-1YU2 Env as a model, kifunensine treatment caused no change of neutralization for most NAbs, and whenever it had a detectable influence, kifunensine treatment led to a reduced level of neutralization by known NAbs. This formed the basic framework of appraisal in the following experiments.
The kifunensine-selected viruses had significant increases in neutralization sensitivity to antisera from all but 1 (C1-534) of the 9 study subjects (Fig. 6). A significant increase in neutralization sensitivity was defined by the presence of a difference of 2 1:2 dilutions at one or more points in the neutralization titration curves between the mock-treated virus and those treated with kifunensine. For subjects C1-175, C1-269, and C8-258, neutralization was enhanced by both 50 μM and 200 μM kifunensine (Fig. 6, top three graphs). For subjects C1-219, C1-763, and C1-536, 50 μM kifunensine enhanced neutralization, indicating a glycan-related component of the target epitopes (Fig. 6, middle three graphs). Interestingly, their neutralization potency was not changed significantly by 200 μM kifunensine. This might be the result of a congregated impact on both glycan and conformational elements by high concentrations of kifunensine, as discussed above. Neutralizing activity in subjects C1-457 and C1-440 was enhanced by 200 μM kifunensine only, not by 50 μM kifunensine (Fig. 6, bottom left two graphs). Thus, enrichment of HM-glycan by itself was not sufficient to enhance neutralization by these BNAbs, and the conformational impact induced by a high concentration of kifunensine was probably responsible for the outcomes, maybe working on the basis of enriched HM-glycan. Finally, neutralization of antiserum from subject C1-534 at low dilutions was suppressed by both 50 μM kifunensine and 200 μM kifunensine (Fig. 6, bottom right graph). However, neutralization of HIV-1JR-FL by antiserum from subject C1-534 at low dilutions was suppressed by 50 μM kifunensine but not by 200 μM kifunensine (see Fig. S2, bottom right graph, in the supplemental material). The magnitude of the effect was close to but short of 2 1:2 dilutions, but this effect was observed consistently in 4 repeat experiments. The effect of kifunensine treatment on this antiserum was similar to that on MAbs PG9 and PG16, which bind to a glycan-dependent epitope (16). Similar results were obtained for the 9 study subjects when HIV-1JR-FL was used as the neutralization target (see Fig. S2 in the supplemental material). In summary, our data indicate that the neutralization activity of BNAbs in all 9 study subjects is affected by kifunensine treatment, and therefore HM-glycan is specifically involved in the BNAb epitopes targeted in these subjects.
To test whether the enhancement of neutralization by kifunensine treatment is specific to Env-binding antibodies, antibodies from the 9 subject antisera were purified using protein A beads and designated purified Ig. The residual fraction of antiserum was also collected and referred to as antibody-depleted antiserum. In all 9 subjects, purified Ig was indistinguishable from whole antiserum in neutralizing wild-type or kifunensine-selected HIV-1YU2 reporter viruses (see Fig. S3 in the supplemental material). Complementing these data, the antibody-depleted antiserum was not neutralizing to any of the target viruses, including the kifunensine-selected ones. Therefore, the neutralizing activity sensitive to kifunensine treatment of virus-producing cells is not due to nonantibody serum factors that can bind nonspecifically to HM-glycans, such as mannose-binding lectins, etc.
It has also been reported that antibodies from some “healthy” human volunteers recognize synthetic mannose ligands (6). This would theoretically make it possible that enrichment of mannose on HIV Env gp160 by kifunensine treatment of virus-producing cells could create binding targets on HIV-1 Env for nonspecific antibodies. To test this possibility, sera from 4 healthy volunteers, as well as purified Ig and antibody-depleted sera, were tested and were not able to appreciably neutralize either wild-type or kifunensine-selected HIV-1YU2 reporter viruses (data not shown; see Fig. S3 in the supplemental material). Additionally, subject antisera did not neutralize HIV-1 reporter viruses carrying vesicular stomatitis virus glycoprotein G (VSV-G), with and without kifunensine treatment of virus-producing cells (data not shown). Taken together, the data indicate that the enhancement of neutralization by kifunensine treatment of virus-producing cells cannot be explained by the potential presence of non-Env-specific antibodies in the subject antisera.
We studied 9 subjects who make strong BNAb responses during chronic HIV-1 infection. Evidence from two experimental approaches supports the hypothesis that HM-glycan is involved in the epitope(s) targeted by BNAb responses in 8 of the 9 study subjects. For subject C1-440, no specific glycan deletion mutation was identified that caused neutralization resistance to this antiserum (Table 2), but the NAb activity was enhanced by kifunensine treatment (Fig. 6, bottom middle graph). The simplest interpretation is that HM-glycan is not directly involved in this BNAb response, and the outcome could be due to a methodological weakness of kifunensine treatment (see below). It is also possible that this NAb response is very polyclonal, involving a diverse array of N-glycans, and that this diversity is able to overcome the sensitivity of our screening assay using Env point mutants. A third possibility is that the BNAb activity in this subject may target a glycan-dependent epitope centered at glycan 262; however, since mutation of the potential glycosylation site at position 262 significantly disrupted the function of model Envs of both HIV-1YU2 and HIV-1JR-FL (Fig. 1), the BNAb activity in this subject may elude our mapping with glycan deletion mutants. In summary, our results demonstrate that a vast majority of BNAb responses during chronic HIV-1 infection target HM-glycan-dependent epitopes in our study subjects. For one subject (C1-219), monoclonal antibodies have been isolated, and the neutralization activity of the neutralizing ones is sensitive to changes in glycosylation (8). That finding provides validation of our conclusion that glycan-dependent BNAbs are present in the 9 study subjects. Significantly, this result is also in full concordance with the recent isolation of a large panel of glycan-dependent NAbs from 4 independent subjects (54). Our results are able to attribute a unified phenotype to a large panel of BNAb responses in HIV-1 infection. Collectively, these studies establish that glycan-specific NAb responses are produced naturally and successfully in some HIV-1-infected subjects, providing an important basis for vaccine development.
Importantly, our study used a discriminating factor of 2 1:2 dilutions in neutralization curves of wild-type and mutant Envs. Although not conventional in the field, this requirement ensured that only the dominant NAb activity in the study subjects was identified. Thus, our finding does not preclude the existence of subdominant NAb responses in these subjects that are specific for other epitopes, such as the CD4BS and MPER, and this is under continued investigation.
We used two independent methods to demonstrate that the BNAb epitopes targeted in study subjects contain N-glycan, specifically HM-glycan. Epitope mapping using glycan deletion mutations was straightforward, demonstrating that BNAbs in 8 of 9 study subjects target epitopes composed of or at least having a component of N-glycan. Our interpretation is the simplest explanation for data in the form of a phenotypic change, i.e., neutralization resistance, in response to a genetic change, i.e., removal of a potential glycosylation signal. However, it is conceivable that an off-target effect of removal of a specific glycan on local peptic structure or on glycosylation at other potential N-glycan sites may contribute to the phenotypic change. Also, due to the usage of the screening assay and the high discriminating factor of a 4-fold difference between a mutant and its parental Env, our method of epitope mapping theoretically may miss some subjects who make glycan-specific BNAb responses that involve a diverse array of glycans (including HM-glycan and C-glycan). Enrichment of HM-glycan by use of α-mannosidase inhibitors is an established method (53). Kifunensine treatment has no impact on neutralization of HIV-1YU2 by most known MAbs whose epitopes do not involve N-glycan. (One notable MAb not reported in this publication is MAb 4E10. Kifunensine treatment of HIV-1YU2 and HIV-1JR-FL had the opposite effect on neutralization by MAb 4E10 from two independent sources. The interpretation for this is not clear, and thus the results were intentionally excluded.) Kifunensine treatment led to enrichment of HM-glycan-dependent epitopes and enhanced neutralization by MAbs such as 2G12, but its effect on complex epitopes with an N-glycan-dependent component, such as PG9 and PG16, was more complicated and cannot be appreciated straightforwardly. Enrichment of HM-glycan by blocking with α-mannosidase inhibitors such as kifunensine may lead to an enhancement of neutralization in different ways. The most obvious and consistent effect of kifunensine treatment is to increase the epitope frequency for HM-glycan-binding NAbs such as 2G12 in a large population of HIV-1 strains with diverse patterns of glycosylation. Enriching HM-glycan can also make low-affinity HM-glycan-specific antibodies gain neutralization potency (14). In addition, modification of glycosylation may lead to conformational changes in HIV-1 Env, a possibility which can be inferred from our studies of PG9 and PG16 (see Results) (15). Our data do not allow for discrimination between these various possibilities. Finally, kifunensine treatment can have different influences on different HIV-1 strains under testing conditions (E. Gray and L. Morris, unpublished results). Notwithstanding these caveats, kifunensine treatment of HIV-1YU2 Env studied under the conditions reported here produced interpretable results within our experimental system: whereas kifunensine treatment caused no increase in the neutralization by known NAbs, except for 2G12, it enhanced neutralization of HIV-1YU2 by 8 of 9 subject antisera. These data support the hypothesis that HM-glycan is involved in BNAb epitopes in a majority of our study subjects. Most importantly, consistent results from two complementary approaches synergize to support the observation that BNAb responses in the majority of our study subjects target HM-glycan-dependent epitopes.
It is a striking finding that the majority of our study subjects make BNAb responses to HM-glycan-dependent epitopes. Given its novelty and its enormous implications for the future direction of HIV-1 vaccine development, validating this finding for independent cohorts of HIV-1-infected patients will be critical. Glycosylation patterns are defined primarily by the presence of the NXS/T consensus signal. However, the profile of different “glycoforms” is heavily influenced by the host cell type and even by cell culture conditions (53). The nature of glycans on HIV-1 can be influenced by the host cells and culture conditions, including methods of activation and differentiation of lymphocytes or macrophages. In most studies investigating neutralization and neutralization escape mechanisms of HIV-1 Env or measuring NAb responses to vaccination, the neutralization target viruses are produced in immortalized lymphocytes or 293T cells and thus are likely to have different N-glycan patterns from HIV-1 strains replicating in human hosts. In light of the findings reported here, necessary caution should be used in interpreting data on NAb responses to HIV-1 infection or to vaccinations, especially those results produced in experiments using pseudotyped reporter viruses. Glycan-specific antibody responses can be an attractive line of vaccine development, since some glycan-specific MAbs demonstrate great neutralizing breadth and potency (51, 54). It has also been shown that glycan-binding antibodies can be induced by synthetic glycans (2, 3, 17, 19, 26, 57) or HM-glycan on non-HIV-1 proteins (1, 25, 34, 35). Unfortunately, a glycan-binding antibody capable of neutralizing diverse strains of primary HIV-1 has yet to be achieved. Thus, it will not be a straightforward task to develop anti-HIV-1 vaccines by eliciting glycan-specific NAbs. However, our finding should add to the thinking in the overall direction of vaccine development against HIV-1 infection.
This work was supported by the NIAID Center for HIV/AIDS Vaccine Immunology (grant AI067854) and by a grant (RO1 AI073133) from the NIH to X.Y.
We acknowledge the critical contributions of identifying and collecting samples from study subjects by the CHAVI 001 and 008 clinical working teams. We thank James Robinson, John Mascola, Dennis Burton, and the International AIDS Vaccine Initiative for their generosity in providing monoclonal antibodies.
Published ahead of print 7 December 2011
Supplemental material for this article may be found at http://jvi.asm.org/.