Broadening sarbecovirus neutralization with bispecific antibodies combining distinct conserved targets on the receptor binding domain

Abstract

Monoclonal neutralizing antibodies (mAbs) are considered an important prophylactic against SARS-CoV-2 infection in at-risk populations and a strategy to counteract future sarbecovirus-induced disease. However, most mAbs isolated so far neutralize only a few sarbecovirus strains. Therefore, there is a growing interest in bispecific antibodies (bsAbs) which can simultaneously target different spike epitopes and thereby increase neutralizing breadth and prevent viral escape. Here, we generate and characterize a panel of 30 novel broadly reactive bsAbs using an efficient controlled Fab-arm exchange protocol. We specifically combine some of the broadest mAbs described so far, which target conserved epitopes on the receptor binding domain (RBD). Several bsAbs show superior cross-binding and neutralization compared to the parental mAbs and cocktails against sarbecoviruses from diverse clades, including recent SARS-CoV-2 variants. BsAbs which include mAb COVA2-02 are among the most potent and broad combinations. As a result, we study the unknown epitope of COVA2-02 and show that this mAb targets a distinct conserved region at the base of the RBD, which could be of interest when designing next-generation bsAb constructs to contribute to a better pandemic preparedness.

Keywords: SARS-CoV-2; bispecific antibodies; breadth; cross-reactivity; neutralization; sarbecoviruses; variants.

Figure 1.

Structural representation of the SARS-CoV-2 RBD with broadly neutralizing mAbs. Depicted is the SARS-CoV-2 S protein (PDB: 7BNN), with focus on one of the three RBD subunits in the “up” conformation, in complex with the Fab regions of mAbs S309 (PDB: 7R6X), Ly-CoV1404 (PDB: 7MMO), CR3022 (PDB: 6YLA), 10-40 (PDB: 7SD5), COVA1-16 (PDB: 7LMW) and the human ACE2 receptor (PDB: 7DF4). A darker color indicates the heavy chain, and a lighter color represents the light chain variable domains.

Figure 2.

BLI affinity measurements of all the bsAbs and corresponding mAbs against S from different sarbecoviruses. Samples were grouped as follows: mAbs, bsAbs with one COVA2-02 arm, bsAbs including one class 3-specific arm, bsAbs including one class 4-specific arm. The heatmap shows area under the curve (AUC) values. S of different SARS-CoV-2 variants (WT, Omicron BA.2 and Omicron BA.4/5), as well as multiple sarbecovirus clades (SARS-CoV, SHC014, WIV1, Pangolin GX 2017, Rf1 and Khosta-2) were included in the assay.

Figure 3.

BsAbs neutralization of SARS-CoV-2, its variants and other sarbecoviruses. (a) Dot graph showing the neutralization for mAbs and bsAbs for the three different concentrations measured in duplo (related to Supplementary Figure 2a-b), and indicated by the % AUC. The mean of all values for a certain group is indicated with a horizontal stripe. Groups were compared using the Mann-Whitney U test, *= p < 0.05. Where not indicated, the difference was not statistically significant. (b) Table showing IC₅₀ values of selected mAbs, bsAbs and corresponding cocktails, as measured in triplicates by pseudovirus neutralization assay. Each IC₅₀ value represents the mean of at least three replicates per mAb/bsAb/cocktail. (c) Ratio of IC₅₀ values between the bsAbs and the most potent of the two parental mAbs (left, circles), and between the bsAbs and corresponding cocktails (right, squares). Values below 1 indicate a neutralization improvement of the bsAb compared to the more potent mAb or cocktail. (d) Geometric means of the IC₅₀ values of all tested viruses. Each symbol represents the GeoMean for one tested mAb (blue triangle), bsAb (red circle) or cocktail (green square), ranked from smallest to highest IC₅₀ value.

Figure 4.

Study of COVA2-02 target epitope. (a) Schematic setup of the BLI competition experiment with the SARS-CoV-2 WT RBD and known class 3 and 4 mAbs. (b) BLI competition curves. S309 and 10-40 were used as representative capture mAbs of class 3 and 4, respectively. The curves represent the second association-residual binding of the competitor mAb to the RBD. (c) Setup of the BLI competition experiment of known class 3 and 4 mAbs and COVA2-02 with the human ACE2 receptor for SARS-CoV-2 WT RBD binding. (d) BLI competition curves. SARS-CoV-2 WT RBD was loaded onto the biosensor, followed by class 3 and 4 mAbs, or COVA2-02. COVA1-18 and COVA2-15, RBS-targeting mAbs were used as control. The curves represent the second association-residual binding of the human ACE2 receptor to the RBD. (e) Representative 2D class averages from NS-EM analysis of COVA2-02 IgG bound to SARS-CoV-2 S. Trimer degradation caused by COVA2-02 binding is highlighted with red circles. Due to inherent flexibility and heterogeneity, particles did not converge to a stable 3D class.