Polysaccharide mimicry of the epitope of the broadly neutralizing anti-HIV antibody, 2G12, induces enhanced antibody responses to self oligomannose glycans

Abstract

Immunologically, "self" carbohydrates protect the HIV-1 surface glycoprotein, gp120, from antibody recognition. However, one broadly neutralizing antibody, 2G12, neutralizes primary viral isolates by direct recognition of Manalpha1-->2Man motifs formed by the host-derived oligomannose glycans of the viral envelope. Immunogens, capable of eliciting antibodies of similar specificity to 2G12, are therefore candidates for HIV/AIDS vaccine development. In this context, it is known that the yeast mannan polysaccharides exhibit significant antigenic mimicry with the glycans of HIV-1. Here, we report that modulation of yeast polysaccharide biosynthesis directly controls the molecular specificity of cross-reactive antibodies to self oligomannose glycans. Saccharomyces cerevisiae mannans are typically terminated by alpha1-->3-linked mannoses that cap a Manalpha1-->2Man motif that otherwise closely resembles the part of the oligomannose epitope recognized by 2G12. Immunization with S. cerevisiae deficient for the alpha1-->3 mannosyltransferase gene (DeltaMnn1), but not with wild-type S. cerevisiae, reproducibly elicited antibodies to the self oligomannose glycans. Carbohydrate microarray analysis of DeltaMnn1 immune sera revealed fine carbohydrate specificity to Manalpha1-->2Man units, closely matching that of 2G12. These specificities were further corroborated by enzyme-linked immunosorbent assay with chemically defined glycoforms of gp120. These antibodies exhibited remarkable similarity in the carbohydrate specificity to 2G12 and displayed statistically significant, albeit extremely weak, neutralization of HIV-1 compared to control immune sera. These data confirm the Manalpha1-->2Man motif as the primary carbohydrate neutralization determinant of HIV-1 and show that the genetic modulation of microbial polysaccharides is a route towards immunogens capable of eliciting antibody responses to the glycans of HIV-1.

Fig. 1

Natural variation in pre-immune reactivities to carbohydrate. Fluorescence intensities from the CFG glycan array are plotted from six pre-immune rabbits between the minimum and maximum signal for all glycans (A) and for oligomannose glycans (B). The average reactivity against each glycan is depicted by a black diamond. Symbols used for the structural formulae: ○, Man. The linkage position is shown by the angle of the lines linking the sugar residues (vertical line, 2-link; forward slash, 3-link; horizontal line, 4-link; back slash, 6-link). Anomericity is indicated by full lines for β-bonds and broken lines for α-bonds (Harvey et al. 2009). The nonreducing terminal D1, D2 and D3 mannoses are indicated in panel B. Unprocessed array data and structural descriptions of glycans corresponding to each CFG array number are available in Supplementary Information online.

Fig. 2

Comparative oligomannose glycoform specificities of rabbit sera following immunization with S. cerevisiae WT or ΔMnn1 determined by the CFG glycan array. Detection of serum reactivities to defined oligomannose glycans on glass slide format carbohydrate microarray, detected by fluorescent secondary antibody (anti-rabbit IgG, Alexa488). Sera from animals immunized with ΔMnn1 show significantly greater affinity for Manα1→2Manα1→2Manα1→6(Manα1→3)Manα1→R (CFG glycan number 195) and Manα1→2Manα1→2Manα1→3[Manα1→2Manα1→6(Manα1→2Manα1→3)Manα1→6] Manα1→R (CFG glycan number 311), shown in panels A and B, respectively. The nonreducing terminal D1, D2 and D3 mannoses are indicated in panel B.

Fig. 3

Cross-clade neutralization screen of immune sera against HIV isolates. The ability of WT or ΔMnn1 sera to neutralize a panel of HIV-1 isolates was tested using a high-throughput neutralization assay. The serum dilution above background required for 50% neutralization is shown for WT or ΔMnn1 sera. Although most serum neutralizations fell below the threshold for detection, there was a weak, but significant, pattern of modest neutralization for those sera immunized with ΔMnn1 but not those with WT. aMLV, murine leukemia virus (the cutoff of IC₅₀ detection in this assay is at a 5-fold serum dilution; axis adjusted to this threshold; see Supplementary Information for full neutralization data).

Fig. 4

MALDI-MS analysis of gp120 glycoforms. (A) Spectra of released N-glycans from gp120 expressed in HEK 293T cells; a table of glycan masses is presented in Supplementary Information. (B) Spectra of released N-glycans from gp120 expressed in HEK 293T cells in the presence of kifunensine. (C) Spectra of released N-glycans from gp120 in HEK 293T cells in the presence of kifunensine and cleaved with ER α-mannosidase I. (D) Spectra of released N-glycans from gp120 expressed in GnT I-deficient HEK 293S cells. ○, Man; ▪, GlcNAc.

Fig. 5

Negative ion MS/MS spectra. Fragmentation analysis of Man₉GlcNAc₂ from Kif-treated 293T cells (A) and the corresponding spectrum of Man₈GlcNAc₂ obtained by treatment with ER α1→2 mannosidase I (B). The negative ion ESI-MS/MS of a reference sample of D1,D3-Man₈GlcNAc₂ (C). Although the Man₈GlcNAc₂ spectra are of chloride adducts rather than the phosphate adduct shown for Man₉GlcNAc₂, the nature of the adduct does not affect the fragment ions because they are all formed by abstraction of a proton by the adduct in the first stage of the fragmentation. The nomenclature of both the glycan symbols and the fragmentation ions follows that previously described (Domon and Costello 1988; Harvey et al. 2009).

Fig. 6

Oligomannose structures of gp120 expressed in GnT I-deficient HEK 293S cells. MALDI-MS of N-linked glycans with the region of the spectra between m/z 1370 and 1945 displayed with a 5-fold magnification.

Fig. 7

Serum reactivity of animals immunized with ΔMnn1 (▪) and WT (▲) to the envelope glycoform array. ELISA plates were coated with equal amounts of glycoform-specific variants of gp120_BaL expressed in: HEK 293T cells (A), HEK 293T cells grown in the presence of kifunensine (B), HEK 293T cells grown in the presence of kifunensine, with gp120 then cleaved with ER α-mannosidase I (C) or GnT I-deficient HEK 293S (Lec1) cells (D). Sera from WT and ΔMnn1 immunizations were assayed for binding in serial dilution. Corresponding serum pre-bleed reactivity is indicated for both ΔMnn1 (□) and WT (△). Each data point is the mean for n = 4 animals. Assay repeated three times.

Fig. 8

Binding to neoglycolipid array. Serum and 2G12 was probed against a neoglycolipid array. The reactivity of 2G12 to Manα1→2Man terminating glycans is evident (A) but only in those glycans with an exposed D1 arm (for example, Man₉GlcNAc₂ and Man_7(D1)GlcNAc₂ but not Glc₁Man₉GlcNAc₂). The binding of a representative serum from each of the ΔMnn1 (B) and WT (C) groups is included for comparison and to validate the assay in comparison with serum analyzed on the CFG array. The marked reactivity of the WT but not ΔMnn1 to the Manα1→3Man terminating structure is consistent with the deletion of this motif in the ΔMnn1 immunogen. The overall pattern of relative reactivities of ΔMnn1 more closely resembles that of 2G12 than does WT serum. GlcNAc is abbreviated to GN and AO, aminoxy linker.

Fig. 9

Molecular basis of 2G12-carbohydrate recognition. Glycans are represented in sticks and the protein surface in gray. (A) The crystal structure of 2G12 in complex with Man₉GlcNAc₂ (Calarese et al. 2003). (B) The crystal structure of 2G12 in complex with the D3-terminating glycan, Man(D3)α1→2Manα1→6[Manα1→2Manα1→3]Man (green) (Calarese et al. 2005) with the remaining saccharide residues generated by docking the solution-state NMR structure of Man₉GlcNAc₂ onto the reducing terminal mannose (cyan). (C) The crystal structure of 2G12 in complex with Manα1→2Man (yellow) (Calarese et al. 2003). (D) The Manα1→3Man terminal residue of a hybrid-type glycan (cyan) docked onto the crystal structure of 2G12 in complex with Manα1→2Man using the most abundant torsion angles (Petrescu et al. 1999; Wormald et al. 2002).