Engaging an HIV vaccine target through the acquisition of low B cell affinity

Abstract

Low affinity is common for germline B cell receptors (BCR) seeding development of broadly neutralizing antibodies (bnAbs) that engage hypervariable viruses, including HIV. Antibody affinity selection is also non-homogenizing, insuring the survival of low affinity B cell clones. To explore whether this provides a natural window for expanding human B cell lineages against conserved vaccine targets, we deploy transgenic mice mimicking human antibody diversity and somatic hypermutation (SHM) and immunize with simple monomeric HIV glycoprotein envelope immunogens. We report an immunization regimen that focuses B cell memory upon the conserved CD4 binding site (CD4bs) through both conventional affinity maturation and reproducible expansion of low affinity BCR clones with public patterns in SHM. In the latter instance, SHM facilitates target acquisition by decreasing binding strength. This suggests that permissive B cell selection enables the discovery of antibody epitopes, in this case an HIV bnAb site.

Fig. 1. Vaccine-elicitation of D386R-sensitive antibodies targeting the CD4bs in humanized mice.

A Phylogenetically distinct HIV Env monomers deployed in this study (strains TH976.17, HT593.1, YU2 and ZM215.8, referred to as: 122E, 45B, YU2, and 92 C, respectively; see also Supplementary Fig. 1A). B Serum antibodies elicited in IGHV1-2 HC2 mice following sequential immunization with heterologous Env (mean ± SD, serum antibody reactivity against WT vs D368R forms of 122E Env, n = 5 biologically independent animals). C Corresponding serum antibody reactivity against WT vs D368R forms of 122E Env as compared by area under the curve (AUC, n = 5 biologically independent animals, P = 0.0208, two-sided paired T test of ratio, see also Supplementary Fig. 2 for serum reactivity to the other Env strains). D Serum antibodies elicited in IGHV1-2 mice following sequential immunization with homologous 122E Env (mean ± SD, serum antibody reactivity against WT vs D368R forms of 122E Env, n = 5 biologically independent animals). E Corresponding serum antibody reactivity against WT vs D368R forms of 122E Env as compared by AUC (n = 5 biologically independent animals, P = 0.908, two-sided paired T test of ratio). F D386R-sensitive versus D368R-insensitive 122E specific memory IgG B cells expanded following immunization with heterologous Env (gated on: CD3^-/CD19⁺/IgM^-/IgD^-/IgG⁺/GL7^-/CD38⁺/122E Env-PE+/122E Env-APC-Cy7⁺, see Supplementary Fig. 3 for full gating; see also Supplementary Fig. 5 for memory BCR reactivity to the other Env sequences used in the vaccine regimen). G D386R-sensitive versus D368R-insensitive Env specific memory IgG B cells expanded following immunization with homologous Env (gated on: CD3^-/CD19⁺/IgM^-/IgD^-/IgG⁺/GL7^-/CD38⁺/122E Env-PE+/122E Env-APC-Cy7⁺). H Ratio of D368R-sensitive to D368R-insensitive IgG B cells expanded by the homologous (homo) versus heterologous (hetero) vaccination regimens (mean ± SD, n = 5 biologically independent animals, P = 0.0007, two-sided Student’s T test; see also Supplementary Fig. 4 for comparison to sequential immunization with a cocktail of the four Env antigens). Data are derived from two independent immunization experiments (Experiment 1 = (B–E); Experiment 2 = (F–H) where mice were pre-bled and sequentially immunized through the intraperitoneal route with individual Env antigens at weeks 0, 3, 6 and 9. Blood was collected two weeks after each immunization and then three weeks after the final inoculation. Spleen was also collected at this final time point.

Fig. 2. Expansion of public IGHV1-2 BCR lineages with both affinity matured and non-detectable binding underscore the elicitation of D368R sensitive antibodies.

Single D368R-sensitive memory IgG B cells (CD19⁺/IgD^-/IgM^-/IgG⁺/ GL7^-/CD38⁺/122E-PE⁺/122E-APC-Cy7⁺/122E D368R-APC^-) from animals receiving the heterologous immunization regimen (see also Fig. 1F, Supplementary Figs. 3–5) were sorted from n = 3 biologically independent IGHV1-2 HC2 animals. A Sequencing the memory B cells yielded 68 BCRs in Mouse A, 70 BCRs in Mouse B, and 68 BCRs in Mouse C (see also Supplementary Data 1). Public B cell pathways were first assigned based on clonotype (usage of a shared CDRH3 sequence) using Cloanalyst software^,–. These public B cell pathways were then further subdivided into public clonal lineages (public CDRH3 + shared CDRL3) revealing four clonal lineages that were reproducibly expanded in all the mice (see also Supplementary Data 1). B Fabs from each public clonal lineage recombinantly and then tested for affinity to original sorting probes (122E Env or 122E Env-D368R). Binding was evaluated using bilayer interferometry (BLI). The equilibrium dissociation constant (KD) values were calculated by applying a 1:1 binding isotherm using vendor-supplied software. KD values above 100 µM are beyond the limit of detection for this instrument. The low affinity Lin1 and Lin2 BCR sequences chosen for this analysis contain a conserved pattern of SHM that was independently reproduced within each public clonal lineage within each mouse (Supplementary Figs. 7, 8). Data are from one vaccination experiment that yielded antigen specific IgG memory BCRs in the three mice; details of each individual B cell clone are provided in Supplementary Data 1.

Fig. 3. BCR antigenicity and signaling within Public Clonal Lineage 1 (Lin1).

A sHsL and gHgL BCRs in Lineage 1, along with VRC01 BCRs were expressed in our B cell reporter system and evaluated by flow cytometry for binding to fluorescent versions of WT Env (red lines) vs D368R Env (blue lines; D368R + A281R + G366R + P369R in the case of YU2¹²) for each antigen used in the immunization regimen. Gray histograms represent binding to BCR isotype control or binding to surface BCR negative for LC expression. B Acquisition of D368R sensitivity, as further resolved by gHgL BCR vs sHsL BCR triggering in response to bivalent Env-Fc, Env-D368R-Fc or anti-IgM as a positive control. All Env variants used in the heterologous immunization regimen were tested. Presented is the Ca²⁺ flux activity, measured kinetically by the ratiometric Ca²⁺ sensing dye fura red and normalized to total flux capacity, as defined by the ionophore ionomycin. (A) vs (B) represents two independent experiments to evaluate BCR antigen recognition via the two orthogonal methods (binding to membrane presented BCR and BCR triggering following antigen exposure). Antigenicity of sHsL vs gHgL BCR was independently confirmed by reversing the fluorescent label on Env vs Env-D368R (Supplementary Fig. 9), and LC vs HC contribution to the acquisition of D368R sensitivity was evaluated by independently comparing antigenicity of sHgL vs sHsL BCR (Supplementary Fig. 10). BCR triggering was first established in response to 122E Env-Fc vs 122E Env-D368R-Fc (Supplementary Fig. 11), and then independently evaluated for Fc-presentation of all the Env strains used in the heterologous immunization regimen, as shown in this Figure (B).

Fig. 4. Scanning the antigen surface through SHM and low affinity in Public B cell Lineage 1 (Lin1).

A sHsL, gHgL and inferred intermediate sequences within B cell Lineage 1 (Lin1_mut1-4). Lin 1 is public clonal lineage that also uses a public pattern of SHM that is conserved across vaccine recipients (Supplementary Figs. 7, 12). B sHsL, gHgL and intermediate BCR sequences were expressed as IgM BCRs in a B cell reporter system and evaluated by flow cytometry for binding to fluorescent versions of WT Env (red lines) vs D368R Env (blue lines). Gray histograms represent binding to BCR isotype control or binding to surface BCR negative for LC expression. Data presented represents one experiment. C sHsL, gHgL and Lin1_mut1-4 expressed as Fabs and then evaluated for binding to Env vs Env-D368R using bilayer interferometry (BLI). The equilibrium dissociation constant (KD) values were calculated by applying a 1:1 binding isotherm using vendor-supplied software (see also Supplementary Table 1). KD values above 100 µM are beyond the limit of detection for this instrument. Data presented represents one experiment. D Permissiveness in B cell affinity selection as a window for scanning the antigen surface through SHM. Acquisition of D368R sensitivity in the membrane BCR format (see B) is underscored by lower affinity for cognate antigen, as measured in the Fab format (see C). B, C represent independent experiments to compare BCR recognition by two orthogonal methods (binding by membrane presented BCR versus monomeric antibody affinity). A phylogenetic tree denoting the positions of sHsL, gHgL and intermediates is presented in Supplementary Fig. 12.

Fig. 5. BCR antigenicity and signaling within Public Clonal Lineage 2 (Lin2).

A sHsL and gHgL BCRs in Lineage 2, along with VRC01 BCRs were expressed in our B cell reporter system and evaluated by flow cytometry for binding to fluorescent versions of WT Env (red lines) vs D368R Env (blue lines; D368R + A281R + G366R + P369R in the case of YU2¹²) for each antigen used in the immunization regimen. Gray histograms represent binding to BCR isotype control. Data presented represents one experiment. B Acquisition of D368R sensitivity, was further resolved by gHgL BCR vs sHsL BCR triggering in response to bivalent Env-Fc, Env-D368R-Fc or anti-IgM as a positive control. All Env variants used in the heterologous immunization regimen were tested. Presented is the Ca²⁺ flux activity, measured kinetically by the ratiometric Ca²⁺ sensing dye fura red and normalized to total flux capacity, as defined by the ionophore ionomycin. Data presented represents one experiment. (A) vs (B) represents two independent experiments to evaluate BCR antigen recognition via the two orthogonal methods (binding to membrane presented BCR and BCR triggering following antigen exposure). Antigenicity of sHsL vs gHgL BCR was also independently confirmed by reversing the fluorescent label on Env vs Env-D368R (Supplementary Fig. 13), and LC vs HC contribution to the acquisition of D368R sensitivity was independently evaluated by comparing antigenicity of sHgL vs sHsL BCR (Supplementary Fig. 14). Additionally, BCR triggering was first established in response to 122E Env-Fc vs 122E Env-D368R-Fc (Supplementary Fig. 15), and then independently evaluated for Fc-presentation of all the Env strains used in the heterologous immunization regimen, as shown in this Figure (B).

Fig. 6. Scanning the antigen surface through SHM and low affinity in Public Clonal Lineage 2 (Lin2).

A sHsL, gHgL and inferred intermediate sequences within B cell Lineage 2 (Lin2_mut1-6). Lin 2 is public clonal lineage that also uses a public pattern of SHM that is conserved across vaccine recipients (Supplementary Figs. 8, 16). B sHsL, gHgL and intermediates were evaluated by flow cytometry for binding to WT Env (red lines) vs D368R Env (blue lines) in our B cell reporter system. Gray histograms depict binding to isotope control or binding to surface BCR negative for LC expression. Data presented represents one experiment. C sHsL, gHgL and Lin2_mut1-6 expressed as Fabs and then evaluated for binding to Env vs Env-D368R using BLI. The equilibrium dissociation constant (KD) values were calculated by applying a 1:1 binding isotherm using vendor-supplied software (see also Supplementary Table 1). KD values above 100 µM are beyond the limit of detection for this instrument. Data presented represents one experiment. D Permissiveness in B cell affinity selection as a window for scanning the antigen surface through SHM. Acquisition of D368R sensitivity in the membrane BCR format (see B) is underscored by lower affinity for cognate antigen, as measured in the Fab format (see C). B, C represent independent experiments to compare BCR recognition by two orthogonal methods (binding by membrane presented BCR versus monomeric antibody affinity). A phylogenetic tree denoting the positions of sHsL, gHgL and intermediates is presented in Supplementary Fig. 16.

Fig. 7. Asymptotic/one-step restricted accumulation of affinity following sequential immunization in mice.

WT C57Bl/6 and IGHV1-2 HC2 mice were sequentially immunized with NP-ovalbumin (4x) at weeks: 0, 3, 6 and 9 and ratio of NP2 to N23 binding by the serum IgG response was measured two weeks after each immunization [mean ± SD], n = 5 biologically independent animals per genotype, two-factor ANOVA. Factor 1= genotype; Factor 2 = immunization number. Following identification of an immunization number effect (P < 0.0001) but not genotype effect (P = 0.8238), factor 2 was further analyzed with Tukey’s test. The P values for the resultant pairwise comparisons are presented in the figure. Data are from one vaccination experiment.