Bispecific Anti-HIV-1 Antibodies with Enhanced Breadth and Potency

Abstract

Broadly neutralizing antibodies (bNAbs) against the HIV-1 envelope glycoprotein (Env) suppress viremia in animal models of HIV-1 and humans. To achieve potent activity without the emergence of viral escape mutants, co-administration of different bNAbs is necessary to target distinct epitopes essential for viral fitness. Here, we report the development of bispecific anti-Env neutralizing antibodies (biNAbs) with potent activity. Synergistic activity of biNAbs was achieved by combining an engineered hinge domain of IgG3 to increase Fab domain flexibility necessary for hetero-bivalent binding to the Env trimer while retaining the functional properties of the IgG1-Fc. Compared to unmodified biNAbs, hinge domain variants exhibited substantially improved neutralization activity, with particular combinations showing evidence of synergistic neutralization potency in vitro and enhanced in vivo therapeutic activity in HIV-1-infected humanized mice. These findings suggest innovative strategies for generating biNAbs with enhanced neutralization breadth and potency, representing ideal candidate molecules for the control of HIV-1 infection.

Figure 1. Overview of biNAb generation and neutralization activity of 3BNC117/10-1074 IgG1 biNAbs

(A) Overview of the strategy for generating biNAbs to ensure proper heavy-light chain pairing and heterodimerization. For correct heavy-light chain pairing, one of the parental mAbs was expressed in the CrossMab format (CH1-CL swapping), while for the other mAb, the wild-type domain architecture was maintained. Heavy chain heterodimerization was achieved by introduction of point mutations in the CH3 domain of the two mAbs. (B) Binding specificity for gp140 and (C) in vitro neutralization activity of wild-type and CrossMab variant of 10-1074 was assessed by ELISA using recombinant gp140 and standardized TZMbl neutralization assay, respectively. No difference in the antigen specificity and in vitro neutralization activity was noted between the different 10-1074 mAb variants. (D–F) ELISA assays to determine dual specificity of the 3BNC117/10-1074 biNAb. (D) Competition ELISA with increasing concentrations of each of the parental bNAbs or a mixture (1:1) of the two bNAbs to determine dual specificity of the biNAb. (E) 3BNC117/10-1074 biNAb, 3BNC117 and 10-1074 bNAbs were immobilized to gp140-coated microtiter plates and detected using Fc domain- or light chain (κ or λ) subclass-specific secondary IgG. BiNAb was detected with both the anti-kappa and anti-lambda secondary antibodies, whereas 3BNC117 and 10-1074 only with anti-kappa or anti-lambda, respectively. (F) Epitope-specific ELISA using a CD4bs antigen (2-CC Core) for capture and an anti-lambda detection antibody to confirm bispecific activity. Data are represented as mean ± SEM. (G) In vitro neutralization breadth and potency plot of 3BNC117/10-1074 IgG1 biNAb against an extended (120 strain) multiclade virus panel. Neutralization activity of their respective parental IgG1 bNAbs was included for comparison. 3BNC117/10-1074 IgG1 biNAb exhibited marked reduction in neutralization potency compared to the activity of a mix of their parental mAbs (determined based on the activity of the most potent parental mAb for a given virus strain) See also related Table S1.

Figure 2. In vitro neutralization activity of IgG1 biNAbs targeting different HIV-1 Env epitopes

In vitro neutralization activity against an extended multiclade virus panel was assessed for IgG biNAbs targeting different, non-overlapping epitopes on HIV-1 Env, using standardized TZMbl neutralization assays. Breadth (% viruses neutralized) vs. potency (IC₅₀ titer (μg/ml)) plots of (A) PG16/PGT121, (B) PG16/PGT128, (C) PG16/10-1074, and (D) PGT151/35O22 IgG1 biNAbs are presented. Neutralization activity of their respective parental IgG1 bNAbs and the predicted activity of the mix of the two parental bNAbs (determined based on the activity of the most potent parental mAb for a given virus strain) was included for comparison. Neutralization data are presented in Tables S2–S5.

Figure 3. Generation and characterization of biNAb hinge variants

(A) Schematic representation and primary amino acid sequence of the hinge domain of IgG1, IgG3 and the generated variant, IgG3C-. Cysteine residues participating in inter-chain disulfide bonds are depicted in red. IgG1 hinge variants of 10-1074 bNAb were generated by switching the IgG1 hinge region with either the IgG3 or IgG3C- hinge. (B) In vivo stability and half-life was compared between the various 10-1074 hinge-domain variants (mean±SEM, n=4 mice/group). No differences among the hinge domain variants were noted in terms of in vivo pharmacokinetics (B) and protein stability (see also related Figure S1). (C) Comparison of the in vitro neutralization activity against an extended (120 strains) multiclade virus panel of 3BNC117/10-1074 biNAbs expressed as IgG1 or IgG3C- hinge variants (IC_50/80 titers are presented in Table S6). As control, the predicted neutralization of the mix of the two parental bNAbs (3BNC117 and 10-1074) was included. (D) Representative IC₅₀ titers (μg/ml) of IgG1 and IgG3C- hinge domain variants of 3BNC117/10-1074 biNAbs and the parental bNAbs (3BNC117 and 10-1074) showing the improved neutralization activity of the IgG3C- hinge variant compared to the wild-type, unmodified IgG1 biNAb.

Figure 4. Neutralization activity of a panel of hinge domain engineered (IgG3C-) biNAbs

A panel of bNAbs were selected targeting distinct epitopes on Env and biNAb combinations encompassing the IgG3C- hinge variant were generated with non-overlapping epitope specificities. (A) Epitope mapping on the surface of the HIV-1 Env trimer showing the binding sites of the selected bNAbs (color-coded to match their respective HIV-1 Env epitope). In vitro neutralization activity against a cross-clade, tier 2/3 virus panel (7 strains) was assessed and combinations with variable degree of synergy were identified (B: grid showing the fold change (log) in activity of biNAb over the respective parental bNAb with the most potent (lowest IC₅₀) activity for a given virus strain; C: IC₅₀ titers (μg/ml) of example biNAb combinations exhibiting synergistic, neutral or inhibitory activity). Complete neutralization data are presented in Table S7.

Figure 5. In vitro neutralization breadth and potency of selected IgG3C- hinge biNAbs

In vitro neutralization activity against an extended multiclade virus panel (119 strains) was assessed for IgG3C- hinge variants of PGT151/10-1074 (A), 8ANC195/PGT128 (B), 3BNC117/PGT135 (C) biNAbs and their respective parental bNAbs (also as IgG3C- hinge variants). Complete neutralization data are presented in Tables S8–S10. Predicted neutralization activity of the mix of the two parental bNAbs (determined based on the activity of the most potent parental mAb for a given virus strain) was included for comparison. (D) Representative IC_50/80 titers (μg/ml) of 3BNC117, PGT135, and 3BNC117/PGT135 biNAb showing substantially improved neutralization activity of the biNAb over the parental bNAbs.

Figure 6. Hinge length and flexibility requirements for the enhanced neutralization activity of the 3BNC117/PGT135 biNAb

(A) Side and (B) top view of 3BNC117 (cyan; PDB ID: 4JPV) and PGT135 (red; PDB ID: 4JM2) Fabs bound to the Env trimer (PDB ID: 4NCO). The crystal structures of PGT135 and 3BNC117 Fab complexed with gp120 (JRFL and 93TH057, respectively) were aligned onto the crystal structure of the BG505.SOSIP.664 trimer (PDB ID: 4NCO). The distance between the ends of the two Fabs was calculated to be 67Å, suggesting that the IgG3C- hinge variant might facilitate bivalent, intra-trimeric interactions of the 3BNC117/PGT135 biNAb. (C–D) In vitro neutralization activity (IC₅₀ titers (μg/ml)) of 3BNC117 + PGT135 bNAb mix or 3BNC117/PGT135 biNAb expressed with the following hinge domain structures: IgG1, IgG3 and IgG3C- (see also related Fig. 3A for sequence). 3BNC117/PGT135 IgG3C- biNAb exhibited significantly improved neutralization activity (**p<0.001) compared to IgG1 or IgG3 biNAb hinge variants. IC₅₀ titer (μg/ml) results are presented as geometric mean±95%CI. (E–F) IC₅₀ and IC₈₀ titers (μg/ml) of 3BNC117/PGT135 IgG3C- biNAbs with variable hinge domain length. Shorter variants of IgG3C- (“Full length”) were generated by deleting either one (“-15mer”) or two (“-2x15mer”) of the three 15-mer repeats (EPKSSDTPPPSPRSP). (F) Comparison of the neutralization activity between 3BNC117/PGT135 IgG3C- biNAb (full length) and shortened hinge domain variants revealed that enhanced neutralization activity is correlated with hinge domain length. IC₅₀ titer (μg/ml) results are presented as geometric mean±95%CI. *p<0.05; **p<0.001.

Figure 7. In vivo evaluation of the therapeutic activity of the 3BNC117/PGT135 IgG3C-biNAb in HIV-1-infected humanized mice

(A–B) The in vivo therapeutic activity of 3BNC117/PGT135 IgG3C- biNAb was compared to that of a mix of 3BNC117 + PGT135 IgG3C- bNAbs in HIV-1-infected humanized mice. Mice with established viremia were treated for 4 weeks (red shaded area) with bNAb mix (A) or biNAb (B) and plasma viremia was monitored to assess the in vivo protective activity of antibodies. (C) Comparison of viremia levels or (D) change from baseline at the end of the treatment period revealed significant differences between the mix- and biNAb-treated mice (**p<0.005; mean±SEM, n=7 or 9). See also related Figure S3.