Defining criteria for oligomannose immunogens for HIV using icosahedral virus capsid scaffolds

Abstract

The broadly neutralizing antibody 2G12 recognizes a conserved cluster of high-mannose glycans on the surface envelope spike of HIV, suggesting that the "glycan shield" defense of the virus can be breached and may, under the right circumstances, serve as a vaccine target. In an attempt to recreate features of the glycan shield semisynthetically, oligomannosides were coupled to surface lysines on the icosahedral capsids of bacteriophage Q beta and cowpea mosaic virus (CPMV). The Q beta glycoconjugates, but not CPMV, presented oligomannose clusters that bind the antibody 2G12 with high affinity. However, antibodies against these 2G12 epitopes were not detected in immunized rabbits. Rather, alternative oligomannose epitopes on the conjugates were immunodominant and elicited high titers of anti-mannose antibodies that do not crossreact with the HIV envelope. The results presented reveal important design considerations for a carbohydrate-based vaccine component for HIV.

Figure 1. Presentation of reactive amino groups on the capsid surfaces of CPMV and Qβ

Capsid structures have been determined by X-ray crystallography. (A) The CPMV virus surface bears two reactive Lys residues on each of the 60 small (S) subunits and three reactive Lys residues on each of the 60 large (L) subunits; K38(S) is the most reactive, followed by K99(L) (Chatterji et al., 2004; Wang et al., 2002). (B) Qβ displays three accessible Lys residues plus the N-terminal amino group of each of its 180 subunits. The clustered arrangement of K2 at the 3-fold axis prevents complete loading of this position. Approximate distances between symmetry-related pairs of reactive amino groups are indicated. These images were made using the coordinates provided by the VIPER database (http://viperdb.scripps.edu/) using the program VMD.

Figure 2. Synthesis of virus glycoconjugates

(Top panel) Synthesis of Qβwt, QβK16M and CPMV conjugates. In structures 8–10, the D1, D2 and D3 arms are labeled. (Bottom panel) Stepwise synthesis of QβHPG glycoconjugates QβHPG-Man₈ (11) and QβHPG-Man₈/Man₉ (12).

Figure 3. Representations of the display patterns around the 3-fold symmetry axis on the surface of Qβ wt (A) and QβK16M (B)

The colored spheres represent triazole linkages to the following: K12 (blue), K16 (green), K2 (red) and the N-terminus (black). Because K12 and K16 are very close to each other (~ 8 A° apart), it is unlikely that both bear attached glycans in each subunit of the wild-type structure. The oval shapes in A, therefore, represent a triazole linkage made at one of the two residues (K12 or K16). Based on this model, the overall glycan display patterns for Qβwt and QβK16M glycoconjugates are very similar.

Figure 4. Polyvalent displays on icosahedral virus capsids mimic the glycan shield

The relative affinities of 2G12 for Qβwt, QβK16M, CPMV, and QβHPG glycoconjugates compared to that of 2G12 for gp120_JR-FL were estimated by a conventional ELISA format (A). Serial dilutions of 2G12 IgG were allowed to bind antigens coated onto ELISA wells. 2G12 was detected with alkaline phosphatase conjugated anti-human IgG. The relative affinity and number of 2G12 epitope mimics displayed on the surface of QβK16M, Qβwt, and QβHPG was examined in a 2G12 sandwich assay (B). Titrations of 2G12 were coated onto ELISA wells upon which 5 μg/mL solutions of glycoconjugates were captured and then detected by biotinylated 2G12. In these assays, the various glycoconjugates may be compared relative to one another based on both the number of displayed epitope mimics (saturation binding level) and their overall apparent affinity (half-max binding level).

Figure 5. Representative immunogenicity profiles for QβK16M-Man₄ and QβK16M-Man₉

Serum Ab titers from rabbits immunized with QβK16M virus glycoconjugates or naked QβK16M particles. Symbol keys below each panel of graphs indicate the rabbit ID numbers, open symbols correspond to pre-immune sera and filled symbols correspond to 4^th bleed immune sera. Antigens are denoted on the left of each graph. Man₇ corresponds to the D1/D2 motif, i.e. Manα1–2Manα1–2Manα1–3Man(Manα1–2Manα1–3Manα1–6)Man.

Figure 6. Glycan array analysis of the anti-mannose specificities elicited by QβK16M glycoconjugates

Serum IgG recognition of α-D-mannose and various synthetic and naturally derived oligomannosides present on the printed glycan array V3.0 (A). Binding of pre-bleed and final bleed sera, diluted 1:200, to oligomannose glycans as measured by fluorescence (y axis). Individual glycans assayed are referenced by the glycan ID numbers on the printed array (x axis). Pre-immune serum data corresponding to a given sugar are plotted above the asterisks preceding each glycan ID number, while immune serum data for a given glycan are plotted above the denoted glycan ID number. Each symbol corresponds to a single rabbit in the QβK16M, QβK16M-Man₄ or QβK16M-Man₉ group. Symbolic representations of the oligosaccharides corresponding to the glycan ID numbers in (A) shown in the context of Man₉ (GlcNAc)₂ (B). Note glycan 9 is not depicted as it represents the monosaccharide α-D-mannose.

Figure 7. Recognition of distinct carbohydrate epitopes by 2G12 and serum anti-mannose IgG

ELISA binding assays to Qβwt-Man₄ (left panels) and Qβwt-Man₉ (right panels) glycoconjugates were performed with 1:700 diluted sera from the 4^th bleed which were pre-incubated with naked Qβwt particles. Sera from QβK16M-Man₄ and QβK16M-Man₉ immunizations were assayed on Qβwt-Man₄ and Qβwt-Man₉, respectively. (A) Binding of immune serum IgG to Qβwt glycoconjugates compared to underivatized Qβwt particles. (B) Residual binding (%) of immune serum IgG to Qβwt glycoconjugates in the presence of serially diluted 2G12. (C) Binding of serially diluted 2G12 to Qβwt glycoconjugates in the presence and absence of QβK16M-Man₄ and QβK16M-Man₉ sera. Error bars indicate the standard deviation between duplicate points. Representative data are shown from repeat experiments.