Nanobody engineering for SARS-CoV-2 neutralization and detection

ABSTRACT In response to the ongoing COVID-19 pandemic, the quest for coronavirus inhibitors has inspired research on a variety of small proteins beyond conventional antibodies, including robust single-domain antibody fragments, i.e., “nanobodies.” Here, we explore the potential of nanobody engineering in the development of antivirals and diagnostic tools. Through fusion of nanobody domains that target distinct binding sites, we engineered multimodular nanobody constructs that neutralize wild-type SARS-CoV-2 and the Alpha and Delta variants at high potency, with IC50 values as low as 50 pM. Despite simultaneous binding to distinct epitopes, Beta and Omicron variants were more resistant to neutralization by the multimodular nanobodies, which highlights the importance of accounting for antigenic drift in the design of biologics. To further explore the applications of nanobody engineering in outbreak management, we present an assay based on fusions of nanobodies with fragments of NanoLuc luciferase that can detect sub-nanomolar quantities of the SARS-CoV-2 spike protein in a single step. Our work showcases the potential of nanobody engineering to combat emerging infectious diseases. IMPORTANCE Nanobodies, small protein binders derived from the camelid antibody, are highly potent inhibitors of respiratory viruses that offer several advantages over conventional antibodies as candidates for specific therapies, including high stability and low production costs. In this work, we leverage the unique properties of nanobodies and apply them as building blocks for new therapeutic and diagnostic tools. We report ultra-potent SARS-CoV-2 inhibition by engineered nanobodies comprising multiple modules in structure-guided combinations and develop nanobodies that carry signal molecules, allowing rapid detection of the SARS-CoV-2 spike protein. Our results highlight the potential of engineered nanobodies in the development of effective countermeasures, both therapeutic and diagnostic, to manage outbreaks of emerging viruses.

. Reported binding and neutralization properties of nanobody modules Table S2.Parameters for the molecular dynamics simulations Table S3.Cryo-EM data collection and processing statistics   F-G) shows a nanobody-spike conformation that are unlikely to be allowed, due to a steric clash (panel F), or due to a distance between nanobodies that the (GGGGS)4 linker may not bridge (panel H).These models demonstrate that the linker length is permits multivalent binding to most RBD conformations.PDB models used for the analysis were 6ZXN (spike in 1-up conformation), 7DX8 (spike in 2-up conformation), 6ZXN (Ty1 in complex with Spike), 7CAN (MR17-K99Y in complex with RBD), 6XHD (H11-H4 in complex with Spike) and 7KN6 (VHH V in complex with RBD).

Figure S1 .Figure S2 .Figure S3 .
Figure S1.Distances between key residues in SARS-CoV-2 spike Figure S2.Modelling of spike-bound nanobody modules connected by (GGGGS)4 linkers Figure S3.Neutralization of SARS-CoV-2 WT at 1 MOI by multimodular nanobodies Figure S4.RT-qPCR of hamster lung samples and individual hamster weights Figure S5.Resolution estimates of cryo-EM reconstructions Figure S6.Nanobody-bound spike cryo-EM maps rendered at different contour levels Figure S7.Fit of spike and nanobody PDB models in the cryo-EM maps Figure S8.RMSD of protein Cα atoms over MD simulation and distance between charged residues (E/K and R) in MD simulations Figure S9.Binding enthalpy (ΔH) of each tri-TMH module and SARS-CoV-2 RBD.

Figure S1 .
Figure S1.Distances (Å) between key residues in the receptor-binding domains in the different conformations of the SARS-CoV-2 spike illustrate the distance bridged by nanobody modules connected by 20-AA linkers.A) Distances between the central residues (493) of the ACE2 binding sites in the 3 RBDs, with spike in the 2-up conformation (left) or 1-up conformation (right).B) Distances between the central residues (375) of the VHH V nanobody epitopes in the 3 RBDs, with spike in the 2up conformation (left) or the 1-up conformation (right).

Figure S2 .
Figure S2.Distances between nanobody modules bound to different RBD conformations.The figure illustrates the length of the (GGGGS)4 linker connecting the nanobody modules, and the distances the (GGGGS)4 linkers must cover between the C-terminus of one nanobody module and the N-terminus of the next.A) A model of a multimodular nanobody bound to SARS-CoV-2 spike, with modules connected by (GGGGS)4 linkers.Linker residues were built in Coot (1) between nanobodies in a model of Ty1 binding the spike protein (PDB 6ZXN), resulting in a model of the spike-bound multimodular nanobody tri-Ty1.The selected 20-residue linker is sufficiently long to span a distance of approximately 70Å between nanobody modules.B)-E) Distances (Å) between the nanobody C termini (red spheres) and N termini (blue spheres) are shown.There are more than 20 possible module pair/RBD conformation combinations, and these panels show examples of nanobody modules binding adjacent RBDs in either up or down conformations.Linker length accommodates binding to RBD pairs in the prevalent downdown and down-up conformations, and also allows binding to some of the rarer up-up RBD conformations.PanelsF-G) shows a nanobody-spike conformation that are unlikely to be allowed, due to a steric clash (panel F), or due to a distance between nanobodies that the (GGGGS)4 linker may not bridge (panel H).These models demonstrate that the linker length is permits multivalent binding to most RBD conformations.PDB models used for the analysis were 6ZXN (spike in 1-up conformation), 7DX8 (spike in 2-up conformation), 6ZXN (Ty1 in complex with Spike), 7CAN (MR17-K99Y in complex with RBD), 6XHD (H11-H4 in complex with Spike) and 7KN6 (VHH V in complex with RBD).

Figure S3 .
Figure S3.Parallel neutralization experiments were performed with 30 000 pfu (1 MOI) of virus instead of 50 pfu (Main Figure3).Nanobodies tri-Ty1, tri-TMH, and tri-TMV show neutralization at these higher virus titers, but the calculated IC50 values are decreased.With higher amounts of virus used for infection, tri-TMH remains the most effective neutralizer, based on the calculated IC50 value.

FigureFigure S5 .AFigure S6 .
Figure S4.A) 35 µg Tri-TMH nanobody was given to hamsters intranasally six hours before infection, and SARS-CoV-2 RNA (RdRp, subE, and E) was quantified from lungs with RT-qPCR.Results are shown as the percentage of the average value for untreated individuals (n=6 in treated and n=4 in untreated hamsters).B) In a small pilot experiment (n = 3 per group), 20 µg Tri-TMH was administered intranasally to hamsters either at the same time as they were infected with SARS-CoV-2 (d0) or one day post infection (d1).SARS-CoV-2 RNA (N) was quantified with RT-qPCR from the d0 and d1 groups.Results are shown as the percentage of the average value for untreated individuals (n=4).Lines in all panels stand for the median.The body weight of the hamsters was recorded daily and the percentage weight change from baseline was plotted for each animal.

Figure S7 .
Figure S7.Fit of spike and nanobody PDB models in the cryo-EM maps.Model of the SARS-CoV-2 S trimer (PDB: 7A29) and models of nanobody-RBD complexes (PDB: 6ZHD, 6ZXN, 7CAN), displayed as ribbon representations, fitted into the cryo-EM maps.A) Map with all RBDs down.B) Localized refinement of each RBD-Nb region in the all-down (closed) conformation spike data did not improve nanobody density features, indicating a mixed population of different modules at each site in the reconstruction.Binding modes of the three nanobodies are shown and indicate that an average density of the three modules is likely to roughly coincide with the position of Ty1.Density features did not support the unambiguous identification of the nanobody modules in the reconstruction of the closed spike conformation.C) Map with one RBD up.D) Localized refinement of each RBD-Nb region in the partially open spike data improved nanobody density features and allowed the confident placement of each tri-TMH module using fitmap global search in ChimeraX.Cross-correlation scores of the top placement, fitting shown in panel D, were 0.86 for Ty1−RBD (from 6ZXN), 0.82 for MR17-K99Y−RBD (7CAN), and 0.87 for H11-H4−RBD.

Figure
Figure S8.A) RMSD (Root Mean Square Displacement) of protein Cα atoms over MD simulation using the zero frame as a reference.Data from all simulation replicas are combined, and the mean value is shown as a dot, with the standard deviation as error bars.B) Distance between charged residues (E/K and R) in MD simulations.The three simulation replicas are shown separately for WT (blue) and E484K (red).The bold lines represent a running average of the previous 20 ns simulation data.Distances were measured between the glutamic acid CD atoms or lysine NZ atoms, and arginine CZ atoms.

Table S1 .
Reported binding and neutralization properties of nanobody modules.

Table S2 .
Description of model systems and MD simulation time scales.Three replicas of 500 nanosecond simulations were run for each simulation setup.

Table S3 .
Cryo-EM data collection and processing statistics