In this study, we characterized the cell-specific glycoforms of recombinant trimeric consensus B gp120 with respect to their recognition by monoclonal gp120-specific neutralizing antibodies and polyclonal serum antibodies from persons infected with HIV-1 subtype A/C. For this purpose, we used recombinant gp120 expressed in the N-terminal fusion with a 62-amino acid non-glycosylated fragment of mannan-binding lectin to drive gp120 trimerization and secretion and C-terminal His tag and V5 tag [21, 86]. These modifications of gp120 sequence contain only naturally occurring PNGS present in consensus B gp120, and, because they are the same for all preparations, it is unlikely that protein backbone impacted the observed differences in antibody reactivities. The gp120 glycoforms were differentially recognized by the monoclonal as well as polyclonal antibodies. As the gp120 produced in different cell types had the same amino-acid sequence, it implicates that differential cell type-specific glycosylation of gp120 is the solely cause of the variation in antibody binding. Cell-specific differential glycosylation of Env was reported to affect both epitope availability and epitope formation, as N-glycans contribute to the proper folding of gp120 .
The oligomeric consensus B gp120 that we used in this study is glycosylated, depending on the producer cell type, as we reported earlier . We previously showed that human T cells (Jurkat) produce gp120 with the largest proportion of high-mannose and hybrid N-glycans, whereas HepG2 cells secrete gp120 with more complex N-glycans and less high-mannose and hybrid oligosaccharides. The greatest variability in glycosylation of gp120 produced by different cell types was related to the relative contribution of complex and hybrid N-glycans . Based on the relative abundance of high-mannose oligosaccharides, the oligomeric gp120 expressed in Jurkat and RD resembled, at least partially, gp120 isolated from HIV-1 virions produced by peripheral blood mononuclear cells [56, 88].
Here, we extended our previous analyses characterizing differential reactivities of antibodies in sera from HIV-1 clade B-infected individuals  to the antibodies in sera from HIV-1 A/C-infected persons and a panel of twelve monoclonal antibodies with five gp120 glycoforms, produced in HEK 293T, Jurkat, RD, HepG2, and CHO cells. Unlike the antibodies in sera from HIV-1 subtype B-infected patients , the antibodies in sera from HIV-1 A/C-infected persons reacted similarly with all gp120 glycoforms. We can speculate that the observed similar reactivities of sera from HIV-1 A/C-infected persons with our set of consensus B gp120 glycoforms are due to preferential recognition of highly conserved epitopes on gp120 which are less affected by differential glycosylation than the clade-specific epitopes recognized by sera from HIV-1 clade B-infected persons. Such epitopes likely include also glycopeptides, as several PNGS and the attached N-glycans are highly conserved in gp120 of HIV-1 [37, 89]. Nevertheless, such glycopeptides are probably not generally recognized by broadly neutralizing antibodies, as the frequency of broadly neutralizing glycopeptide-specific antibodies in HIV-1-infected individuals is quite low [40, 44, 45, 90]. However, when testing the binding of a panel of monoclonal gp120-specific antibodies to gp120 glycoforms, we noted marked differences for some of them. This difference between polyclonal and monoclonal antibodies is probably due to recognition of a broad epitope spectrum by serum antibodies vs. single epitope recognized by individual monoclonal antibodies. After partial PNGase F-driven deglycosylation of gp120, the differences in the recognition of individual gp120 glycoforms by serum antibodies increased, probably due to uneven degree of PNGase F-driven deglycosylation and uncovering of several glycan-shielded epitopes. In contrast, the differences in reactivity of monoclonal antibodies with partially deglycosylated gp120 were dependent on localization of the particular epitopes. Specifically, the recognition of gp120 glycoforms by monoclonal antibodies against gp120 V3 loop apex (268 D4, F425 B4e8, 257 D4, 447-52D, 19b) and Manα1, 2Man-linked glycan-specific (2G12) increased markedly. In contrast, partial deglycosylation of gp120 glycoforms differentially impacted the recognition by CD4-binding site-specific antibodies (b12, VRC03, and VRC01): the binding of b12 did not change, VRC03 reactivity decreased, whereas reactivity of VRC01 increased. Recent study using surface plasmon resonance analyses demonstrated that VRC01 binds gp120 with high affinity whereas VRC03 reacts with a 10-fold lower affinity . Thus, VRC01 could act as a partial CD4 agonist in the interaction with gp120, whereas VRC03 does not display this effect . The different VRC03 and VRC01 sensitivity to gp120 deglycosylation observed in this study suggests that VRC03 requires native gp120 glycosylation for effective binding. Due to unchanged proportions in VRC03 reactivities with individual gp120 glycoforms before and after deglycosylation and the observed overall decrease in VRC03 reactivities after deglycosylation, it could be proposed that glycans involved in VRC03 epitope formation are similarly sensitive to PNGase F treatment in all tested gp120 glycoforms. Furthermore, we showed that PG9 and PG16 antibodies, specific to quaternary epitope composed of the gp120 V1/V2 and V3 loops and the associated glycans [40, 91], exhibited reduced reactivities with gp120 after partial deglycosylation. Recently, it was reported that changing the glycan profile on the HIV-1 trimer using glycosidase inhibitors or a mutant producer cell line resulted in HIV-1 strain-specific changes in sensitivity to the neutralization activity of PG-family of antibodies . This observation is consistent with disproportional changes in PG9 reactivities with our set of gp120 glycoforms before and after partial glycan removal. The reactivity of another monoclonal antibody tested in our study, PGT121, specific for discontinuous epitope composed of V3 base, the N332 oligomannose glycan and surrounding glycans [47, 92], decreased after partial deglycosylation of gp120. This suggests that in contrast to V3 loop which is known to protrude from Env , epitopes for monoclonal antibodies which are located in various Env cavities are surrounded by glycans which contribute to epitope formation as well as shielding. This conclusion is supported by the observed changes in reactivities of individual monoclonal antibodies before and after partial deglycosylation of gp120 antigens. Comparison of differences in the reactivities of 2G12 and PGT121 with native and PNGase F-treated gp120 glycoforms suggested that conformational changes in V3 base and/or removal of V1/V2-associated complex glycans could explain the diminished reactivity of PGT121 with PNGase F-treated gp120. Conversely, V3-base glycans or at least some of those glycans involved in binding of 2G12 (N295, N332) are not affected by PNGase F treatment under native condition [47, 75, 76, 83, 92].
Our approach could not distinguish between the glycan- vs. protein-associated epitopes, as evidenced by the partially retained reactivities of antibodies recognizing glycans on the partially deglycosylated gp120 (PGT121, 2G12, PG16, PG9) [40, 47, 48, 83, 91]. However, our tests characterized overall interactions of antibodies with differentially glycosylated gp120 oligomers and discriminated between various glycoforms of gp120 as targets for a panel of well characterized broadly neutralizing monoclonal antibodies. For PGT121, the data suggested that glycans removed by PNGase F under native conditions contributed substantially to PGT121 epitope formation. Moreover, the variability in PGT121 binding to individual gp120 glycoforms after partial deglycosylation suggested that some glycans were still present and involved in epitope formation (Table 1). This interpretation is consistent with our previous finding that complete removal of all glycans requires denaturing conditions  as well as with reports of others  indicating that some glycans in oligomeric gp140 and gp120/41 Env can be resistant to endoglycosidase cleavage under native conditions.
A partial removal of N-glycans reduced differences in monoclonal antibody binding to native gp120, indicating that differential N-glycosylation affected formation and/or accessibility of Env-specific epitopes, extending previous observations for monoclonal antibody 19b [49, 50]. In contrast, the binding of polyclonal antibodies from sera of subtype A/C HIV-1-infected subjects to native recombinant gp120 glycoproteins was similar for seven of nine sera analyzed by ELISA. However, after partial removal of N-glycans binding differences were detected for eight of the nine sera. The most notable increase in reactivity of polyclonal antibodies occurred after deglycosylation of gp120 produced by CHO and HEK 293T (Figure 5), the cells commonly used for production of vaccine antigens for both experimental and clinical trials, including the first partially successful RV144 trial [38, 94]; this observation supports the concept that glycans may shield various Env epitopes . The antibody reactivity after gp120 deglycosylation was least pronounced for gp120 produced by T cells, the primary host cell type infected by HIV-1. Together, these findings underscore the importance of the cell type for the production of gp120 as a vaccine antigen for eliciting neutralizing antibodies or binding antibodies for antibody-dependent cell-mediated viral inhibition (ADCVI) .
Antibodies specific for glycans are associated mostly with IgG2 and IgA2 subclasses, whereas antibodies specific for proteins are predominantly IgG1 and IgA1 isotypes . Thus, for vaccination purposes, the glycosylation of gp120 antigen should be considered in the strategies to induce isotype-specific immune responses in the systemic and mucosal compartments. In this regard, the results of the recent RV144 trial suggest that IgG antibodies targeting V1/V2 gp120 region may have contributed to protection against HIV-1 infection, whereas high levels of Env-specific IgA antibodies in serum may have mitigated the effects of protective antibodies [39, 97]. Studies addressing this point are in progress in our laboratories.
To test the hypothesis that differential gp120 glycosylation affects HIV-1 entry into host cells, and consequently infectivity, we analyzed the capacity of recombinant gp120 glycoforms to inhibit the in vitro infection of reporter cells by two HIV-1 pseudotyped viruses, SF162 and YU.2. The gp120 produced in Jurkat and CHO cells was the most effective at inhibiting infection by both HIV-1 strains. As we reported earlier, the greatest variability in glycosylation of gp120 produced by different cell lines was the relative abundance of complex and high-mannose and hybrid N-glycans. Notably, the two gp120 glycoforms from Jurkat and CHO cell lines contained less complex and hybrid glycans compared to high-mannose glycans , resembling the glycan patterns of gp120 on virions [56, 88]. Potentially a more direct and biologically relevant way to test the effects of cell-type specific glycosylation of HIV-1 gp120 would be to produce HIV-1 pseudotyped viruses in different cell types and then characterize the differences in their infectivity or sensitivity to broadly neutralizing antibodies. Although we attempted this very challenging approach, we have not succeeded to produce sufficient amount of pseudotyped viruses for our experiments, in contrast to the production of recombinant gp120 in stably transfected cells. Furthermore, it would not be possible to unambiguously attribute all potentially observed changes to cell-type specific glycosylation differences because each cell type may be distinctly sensitive to HIV-1 viral regulatory proteins. The main goal of this study was to assess how the differences in glycosylation, according to the producer cell type, affect recognition by HIV-1 gp120-specific neutralizing and serum antibodies and how they affect interaction with HIV receptors on the reporter cells.
In conclusion, although the production of recombinant gp120 is higher in standard cell types used for biotechnology applications (e.g., HEK293), other cell types, such as T cells (Jurkat) produce gp120 antigen with a glycosylation profile resembling that on HIV-1 virions . Jurkat-produced gp120 is well recognized by broadly neutralizing monoclonal antibodies, including the glycan-dependent antibodies PG9 and PGT121, and effectively competes with HIV-1 virions in binding the receptors on reporter cells.