IgG-like bispecific antibodies with potent and synergistic neutralization against circulating SARS-CoV-2 variants of concern

Abstract

Monoclonal antibodies are a promising approach to treat COVID-19, however the emergence of SARS-CoV-2 variants has challenged the efficacy and future of these therapies. Antibody cocktails are being employed to mitigate these challenges, but neutralization escape remains a major challenge and alternative strategies are needed. Here we present two anti-SARS-CoV-2 spike binding antibodies, one Class 1 and one Class 4, selected from our non-immune human single-chain variable fragment (scFv) phage library, that are engineered into four, fully-human IgG-like bispecific antibodies (BsAb). Prophylaxis of hACE2 mice and post-infection treatment of golden hamsters demonstrates the efficacy of the monospecific antibodies against the original Wuhan strain, while promising in vitro results with the BsAbs demonstrate enhanced binding and distinct synergistic effects on neutralizing activity against circulating variants of concern. In particular, one BsAb engineered in a tandem scFv-Fc configuration shows synergistic neutralization activity against several variants of concern including B.1.617.2. This work provides evidence that synergistic neutralization can be achieved using a BsAb scaffold, and serves as a foundation for the future development of broadly reactive BsAbs against emerging variants of concern.

Fig. 1. Biochemical characterization of Ab 12 and Ab 2-7 scFv-Fcs.

A Competition matrix showing percent blockade of CR3022 and ACE2 binding to SARS-CoV-2 RBD. B Antibody binding at 2 ug/ml was measured by ELISA against Tor2 SARS-CoV spike (Toronto, 2003). n = 2 biologically independent samples and the mean is presented. C PRNT neutralization assays performed with SARS-CoV-2 (isolate USA‐WA1/2020) demonstrate that Ab 12 is the more potent of the two scFv-Fcs. Samples were tested as 3 biologically independent samples with 3 technical replicates per sample and the data is presented as the mean with standard deviation. D FACS based measurement of spike shedding induced by binding of Ab 12, Ab 2-7, CR3022 IgG, and solACE2 at 37 °C. A decrease in median fluorescence indicates an increase in spike shedding whereas an increase in fluorescence indicates minimal shedding is observed (n = 2 biologically independent samples with the mean represented). E Western blot detecting shed S1 in supernatant from Ab 12 IgG spike shedding experiment confirming decreasing fluorescence is at least partially a result of shedding. Experiment was repeated once and representative data is presented. Source data are provided as a Source Data file.

Fig. 2. High resolution cryo-EM structure of Ab 2-7 scFv and Ab 12 Fab complexed SARS-CoV-2 S D614G.

A Side and (B) top views of a composite 3D cryo-EM reconstruction of Ab 2-7 scFvs bound to the receptor binding domains of SARS-CoV-2 spike ectodomains at 3.0 Å resolution. Individual spike protein protomers are colored red, blue and green. The heavy chain of Ab 2-7 scFv is colored yellow, the light chain is colored tan. C Locally refined 3D cryo-EM reconstruction of antibody binding interface. D Atomic model of the binding interface between Ab 2-7 scFv and the spike RBD. E Side view of 3D EM reconstruction of Ab 12 Fab (purple) bound to the RBD of SARS-CoV-2 spike ectodomain. Overlay of the Ab 2-7 and Ab 12 structures demonstrates that the two antibodies target spatially discrete, non-overlapping epitopes (insert, with the Ab 12 cryo-EM reconstruction shown in gray).

Fig. 3. Ab 2-7 can only bind as an scFv and Ab 12 competes with ACE2.

A Cryo-EM reconstruction of unbound S-D614G with one RBD in the up conformation (3.0 Å). Coloring is the same as in Fig. 2. The inset at the top right shows that Ab 2-7 would clash with the “down” RBD of the neighboring spike protomer (green cryo-EM reconstruction, filtered to 6 Å). The inset below shows Ab 2-7 modeled as a Fab, wherein the CH region would clash with the spike NTD (blue cryo-EM reconstruction, filtered to 6 Å). Clashes are indicated as white arrows. B Overlay of ACE2 bound to the RBD confirms that Ab 12 directly competes for the receptor binding motif (RBM). Ab 2-7 binds the conserved RBD core and does not directly compete with ACE2. Clashes are indicated as white arrows.

Fig. 4. Therapeutic efficacy of Ab 12 IgG and Ab 2-7 scFv-Fc in Syrian golden hamster model.

A Therapeutic treatment of Syrian golden hamsters post-infection with Ab 12 IgG or Ab 2-7 scFv-Fc leads to a 513.9- and 5.2-fold reduction of viral loads, respectively, compared to control (PBS) treated animals B Pathology scores for animals treated with PBS, Ab 12 IgG, or Ab 2-7 scFv-Fc. Significant reduction in overall, consolidation, and septal thickening scores were seen for Ab 12 treatment. Scores were determined based on the criteria in Supplementary Table 2. Data points represent individual mice with geometric mean (A) or mean ± standard deviation in (B). Statistical analysis was performed using Mann–Whitney test in (A) and Kruskall–Wallis test with Dunn’s post-hoc correction in (B). Source data are provided as a Source Data file.

Fig. 5. Design and in vitro characterization of anti-SARS-CoV-2 BsAbs.

A Design of the four anti-SARS-CoV-2 BsAbs. Constant regions are colored in gray, Ab 12 binding domains are red/pink, and Ab 2-7 binding domains are blue/cyan. In vitro neutralization was performed with the IgG fusion BsAbs (B) and tandem scFv-Fcs (C). Samples were tested with n = 3 biologically independent samples and data are presented as mean ± standard deviation. Source data are provided as a Source Data file.

Fig. 6. BLI binding and competition analysis for mono- and bispecific constructs with various variant spike/RBD proteins.

A AHC sensors were loaded with anti-SARS-CoV-2 mono- and bispecific Abs and then dipped into wells containing different Spike variants. Amplitude of the binding signal for each variant was normalized to that Ab format’s binding of the Wuhan spike. Though the parental Abs, IgG fusions, and Ab 12/Ab 2-7 tandem BsAbs exhibit marginal affinity to the variants tested, Ab 2-7/Ab 12 constantly binds at a greater level than either of the parental Abs, indicating a synergistic binding pattern. B Competition with ACE2 for binding to Wuhan, B.1.351, B.1.617.2, and B.1.1.529 RBDs was performed with biotinylated RBD proteins and soluble ACE2. Though our BsAbs weakly bind mutant spikes at low concentration, at high concentrations they are able to effectively block ACE2 binding to the RBD. Two chimeric antibodies from Acro Biosystems were used as experimental controls, as both are cross-reactive but only AM122 has been demonstrated to compete with ACE2 binding. The percent binding experiment was performed in triplicate and mean values are shown. Competition was performed as a single experiment. Source data are provided as a Source Data file.

Fig. 7. In vitro neutralization curves using GFP engineered SARS-CoV-2 variant strains.

Neutralization curves were performed in triplicate using the mono- and bispecific Ab constructs against the D614G variant (A) and the subsequent B.1.1.7 (B), P.1 (C), B.1.351 (D), and B.1.617.2 (E) variants of concern which contain a number of additional spike mutations. The greatest effect of these mutations can be seen in the B.1.351 and B.1.617.2 neutralization plots where all curves display a substantial shift to the right. The exception to this is the Ab 2-7/Ab 12 tandem scFv-Fc, which demonstrates remarkable breadth of neutralization across the four variants of concern. Samples were tested with n = 3 biologically independent samples and data are presented as mean ± standard deviation. Source data are provided as a Source Data file.