Heterologous prime-boost vaccination drives early maturation of HIV broadly neutralizing antibody precursors in humanized mice

Abstract

A protective HIV vaccine will likely need to induce broadly neutralizing antibodies (bnAbs). Vaccination with the germline-targeting immunogen eOD-GT8 60mer adjuvanted with AS01_B was found to induce VRC01-class bnAb precursors in 97% of vaccine recipients in the IAVI G001 phase 1 clinical trial; however, heterologous boost immunizations with antigens more similar to the native glycoprotein will be required to induce bnAbs. Therefore, we designed core-g28v2 60mer, a nanoparticle immunogen to be used as a first boost after eOD-GT8 60mer priming. We found, using a humanized mouse model approximating human conditions of VRC01-class precursor B cell diversity, affinity, and frequency, that both protein- and mRNA-based heterologous prime-boost regimens induced VRC01-class antibodies that gained key mutations and bound to near-native HIV envelope trimers lacking the N276 glycan. We further showed that VRC01-class antibodies induced by mRNA-based regimens could neutralize pseudoviruses lacking the N276 glycan. These results demonstrated that heterologous boosting can drive maturation toward VRC01-class bnAb development and supported the initiation of the IAVI G002 phase 1 trial testing mRNA-encoded nanoparticle prime-boost regimens.

Fig. 1.. Design and characterization of the core-g28v2 60mer boost immunogen.

(A) Shown is the iterative immunogen design workflow diagram to improve upon the starting HxB2 core-e-2cc N276D immunogen. (B) Shown is a surface representation of a computational model of core-g28v2 monomer colored with CD4bs (yellow), N276D and T278M mutations (orange), resurfaced residues (pink), natural glycosylation sites (green), engineered glycosylation sites (blue), and mutations to add potential CD4⁺ T helper cell epitopes conserved with the HIV env trimer (TH6 residues, purple). (C) K_D values were measured by SPR for mAbs elicited by eOD-GT8 60mer protein in humans and SE09 mice for first-boost immunogen candidates eOD-GT6v2-cRSF, core-g28v2, and 191084-N276D. Thick lines indicate median values, boxes show 25 and 75% percentile values. *The low-capture IgG SPR method may include some avidity for trimeric analytes.

Fig. 2.. Comparison of protein boost immunogens.

(A) Shown is the immunization scheme for evaluating boost immunogen candidates delivered as proteins plus adjuvant in SE09 mice. (B) The frequency of antigen⁺⁺ MBCs among total MBCs was analyzed by flow cytometry. Each group was sorted with matched antigens, except for the PBS placebo group which was sorted with the core-g28v2 probe. (C to E) Shown is the frequency of VRC01-class MBCs among antigen⁺⁺ MBCs (C), the frequency of VRC01-class MBCs among total MBCs (D), and the frequency of VRC01-class MBCs with human V_K1–33 light chains among total MBCs (E). The blue dashed bar in (E) indicates the overall frequency of VRC01-class MBCs with human V_K1–33 light chains among total MBCs within each group. Red bars indicate medians and each point represents an individual mouse for (B to E). (F and G) The median percent aa SHM in the V_H gene (F) and in the V_K/V_L genes (G) is shown for all VRC01-class MBCs. (H and I) The median percent aa SHM in the V_H gene (H) and in the V_K/V_L genes (I) is shown for non-VRC01-class MBCs. Each point represents the median per mouse and the red bars indicate the median of medians for panels (F to I). Statistical comparisons were made by Kruskal-Wallis test followed by Dunn’s test for multiple comparisons. *p<0.05, **p<0.01, ***p<0.001; ns, not significant.

Fig. 3.. mAb SPR and serum antibody binding responses after eOD-GT8 60mer priming and core-g28v2 boosting.

(A) Monovalent K_D values were measured by SPR of VRC01-class mAbs from naïve SE09 mice, eOD-GT8 60mer primed sE09 mice, and eOD-GT8 60mer primed, core-g28v2 60mer boosted SE09 mice for eOD-GT8, core-g28v2, and core-g28v2-N276⁺. Red bars indicate median affinities and include non-binders. Geomean affinities were calculated among binders only. (B) Shown is serum IgG ELISA binding to eOD-GT8, eOD-GT8-KO, core-g28v2, and core-g28v2-KO for SE09 mice primed with eOD-GT8 60mer protein (purple) alone or SE09 mice primed with eOD-GT8 60mer protein and boosted with core-g28v2 60mer protein (pink). Each point represents the half-maximal effective dilution (ED₅₀) of serum per mouse. Red bars indicate median ED₅₀ values. Statistical comparisons were made by Kruskal-Wallis test followed by Dunn’s test for multiple comparisons. **p<0.01, ***p<0.001; ns, not significant.

Fig. 4.. Key VRC01-class heavy chain residues after protein immunization.

(A) Shown is a ribbon diagram of the VRC01 variable fragment (Fv) with key VRC01-class heavy chain residues colored green for non-paratope residues and pink for paratope residues. (B) A molecular surface representation of HIV Env shows contact residues of key VRC01-class heavy chain paratope residues (red). (C) Shown is the 90^th percentile number of key VRC01-class heavy chain residues elicited during immunization experiment described in Fig. 2A. Each point is the 90^th percentile number of key VRC01-class heavy chain residues for each mouse with the red bars indicating the median of the 90^th percentile values for each group. Statistical comparisons were made by Kruskal-Wallis test followed by Dunn’s test for multiple comparisons. **p<0.01; ns, not significant. (D) Shown are key VRC01-class heavy chain residues for all VRC01-class sequences recovered for placebo and core-g28v2 boost groups. Numbers at top indicate positions within the antibody, colored as in (A); column at left indicates data from each mouse in a different color; each row indicates a single VRC01-class sequence, with red boxes indicating the presence of non-germline key VRC01-class heavy chain residues. Residue numbers in panels (A), (B), and (D) use the Kabat antibody numbering scheme (56). (E to G) K_D values between core-g28v2 and mAbs isolated after core-g28v2 60mer boosting were correlated with number of key VRC01-class heavy chain residues (E), percent V_H SHM (F), and percent V_K SHM (G). In (E to G), solid lines show correlations, dashed lines show 95% confidence internal, S represents slope, and p represent the P-value from a simple linear regression test. ns, not significant.

Fig. 5.. Immune response elicited using mRNA/LNP immunization.

(A) Shown is the immunization scheme for evaluating immunogens delivered using mRNA/LNPs in SE09 mice. (B) The frequency of antigen⁺⁺ MBCs among total MBCs was analyzed by flow cytometry. (C to E) Shown is the frequency of VRC01-class MBCs among antigen⁺⁺ MBCs (C), the frequency of VRC01-class MBCs among total MBCs (D), and the frequency of VRC01-class MBCs with human V_K1–33 light chains among total MBCs (E). Red bars indicate medians and each point represents an individual mouse for panels (B to E). Statistical comparisons were made by Kruskal-Wallis test followed by Dunn’s test for multiple comparisons. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001; ns, not significant.

Fig. 6.. Key VRC01-class heavy chain residues, mAb SPR, and serum antibody binding after mRNA/LNP immunization.

(A) Shown are monovalent K_D values measured by SPR of VRC01-class mAbs from naïve SE09 mice, eOD-GT8 60mer mRNA/LNP primed SE09 mice, and eOD-GT8 60mer mRNA/LNP primed, core-g28v2 60mer mRNA/LNP boosted SE09 mice for eOD-GT8, core-g28v2, and core-g28v2-N276⁺. Red bars indicate median affinities and include non-binders. Geomean affinities were calculated among binders only. (B) Shown is serum IgG ELISA binding to eOD-GT8, eOD-GT8-KO, core-g28v2, and core-g28v2-KO for SE09 mice primed with eOD-GT8 60mer mRNA/LNPs (magenta) only or primed and boosted with core-g28v2 60mer mRNA/LNPs (green). Each point represents the ED₅₀ value of serum per mouse. Red bars indicate median ED₅₀ values. (C) Shown are the 90^th percentile values for key VRC01-class heavy chain residues elicited during mRNA/LNP immunization experiment described in Fig. 5A. Each point is the 90^th percentile key VRC01-class heavy chain residues for each mouse with the red bars indicating the median of the 90^th percentile values for each group. Statistical comparisons in (B and C) were made by Kruskal-Wallis test followed by Dunn’s test for multiple comparisons. *p<0.05, **p<0.01, ****p<0.0001; ns, not significant. (D) Key VRC01-class heavy chain residues for all VRC01-class sequences recovered for placebo and core-g28v2 boost groups. Numbers at top indicate positions within the antibody, colored as in (Fig. 4A); column at left indicates data from each mouse in a different color; each row indicates a single VRC01-class sequence, with red boxes indicating the presence of non-germline key VRC01-class heavy chain residues. Residue numbers use the Kabat antibody numbering scheme (56). (E to G) K_D values between core-g28v2 and mAbs isolated after core-g28v2 60mer mRNA/LNP boosting were correlated with number of key VRC01-class heavy chain residues (E), percent V_H SHM (F), and percent V_K SHM (G). In (E to G), solid lines show correlations, dashed lines show 95% confidence internal, S represents slope, and p represent the P-value from a simple linear regression test.

Fig. 7.. Post core-g28v2 mAbs bind N276(–) native-like HIV Env trimers and neutralize corresponding pseudoviruses.

(A) Shown are apparent K_D values measured by SPR of VRC01-class mAbs elicited in SE09 mice after priming with eOD-GT8 60mer mRNA/LNPs and boosting with core-g28v2 mRNA/LNPs, for wildtype and N276(–) native-like HIV Env trimers. (B) Shown are apparent K_D values measured by SPR of VRC01-class mAbs elicited in SE09 mice after priming with eOD-GT8 60mer protein + adjuvant and boosting with core-g28v2 60mer protein + adjuvant, for wildtype and N276(–) native-like HIV Env trimers. Thick lines indicate median values, boxes show 25 and 75% percentile values (A and B). (C) Pseudoviruses neutralization was tested using a panel of 15 VRC01-class mAbs elicited in SE09 mice after eOD-GT8 60mer mRNA/LNP priming and core-g28v2 60mer mRNA/LNP boosting. Murine leukemia virus (MLV) was used as a negative control. No neutralization: NN; IC₅₀ > 50 μg/mL.