Defining the Antigenic Topology and Prospective Binding Breadth of Vaccination-induced SARS-CoV-2 Neutralizing Antibodies

Reviewer #1 (Public review):

Summary:

The authors provide high-resolution cryoEM structures to map and functionally characterize human antibodies against SARS-CoV-2 elicited by a standard mRNA vaccine. Here, they report high-resolution structural information on seven previously documented neutralizing antibodies from this response, which were produced from early plasmablasts and which engage diverse targets on the viral spike glycoprotein. This structural information is then integrated with functional assays to define how antibody epitope specificity, geometry, and conformational dynamics may shape neutralization outcomes.

Strengths:

A core strength of the study is a technically-well executed analysis of multiple 'ectopically balanced' mAbs elicited by early B cell plasmablast responses. These antibodies engage different neutralizing targets on the S-trimer of SARS-CoV-2, including the RBD and NTD domains. This has resolved a core distinction in terms of how nAbs engaging these features (and subfeatures, e.g., more conserved hydrophobic pocket within NTD) neutralize the virus.

Weaknesses:

A general weakness is that these antibody classes have been structurally characterized already (albeit individually), and much of this work has been done in the context of understanding susceptibility to escape mutations (delta, omicron, and subvariants therein; class I-IV antibody crossreactivity on Wuhan SARS-CoV-2 to present). It is exceptionally fine technical work presenting the antibodies in a collection like this, but perhaps the new predictive power of this analysis is somewhat overstated.

The early plasmablast angle seems like it could be better fleshed out. Many of the known SARS-CoV-2 nAbs are from the plasmablast pool, but how does this predict the antibody profile at latter stages, as per the stated goal and claim of the current study? Does the paratope pattern of plasmablast antibodies then change within the immune sera at later time points? New or existing cryoEMPEM data could shed light on this.

Reviewer #2 (Public review):

Summary:

This manuscript provides important insights into the interaction between early vaccine-elicited antibodies and SARS‑CoV‑2 evolution. The work will be of broad interest to researchers in structural virology, immunology, and vaccine development. However, several conclusions-particularly those involving neutralization breadth and spike destabilization-require additional functional and biophysical validation.

Strengths:

The manuscript provides an unusually comprehensive structural dataset, resolving all neutralizing antibodies in complex with the SARS‑CoV‑2 spike and enabling direct mechanistic comparison across epitope classes. Its integration of cryo‑EM structures with variant binding, sequence analysis, and fusion‑inhibition assays offers a coherent, multidimensional explanation for antibody breadth and escape. Notably, the identification of a conserved NTD hydrophobic pocket targeted by broad-reactive antibodies represents a conceptually important advance with clear implications for future vaccine design.

Weaknesses:

The study lacks variant-specific neutralization assays, limiting the ability to directly correlate binding breadth with functional viral inhibition. It also omits kinetic affinity measurements, leaving important mechanistic questions, such as why certain antibodies retain breadth, only partially resolved. Additionally, reliance on HEK293T-based spike display raises concerns about glycosylation-related artifacts, especially for NTD loop-dependent antibodies.

Reviewer #3 (Public review):

Summary:

In this manuscript by Jaiswal et al., the authors used structural biology combined with cellular assays to determine the molecular basis underlying the neutralizing ability of the SARS-CoV-2 antibodies. The authors compared the binding mode of the neutralizing antibodies that have two distinct binding interfaces and identified key sites that determine their vulnerability to virus evolution. Interestingly, the author also demonstrated that the trimer-disrupting antibody has the broadest activity as the variations at the trimer interface are limited in evolution.

Strengths:

This manuscript reported a large collection of structures and covered a broad range of binding modes and mechanisms of action. Many of the cryo-EM structures are of good quality. The authors' hypothesis regarding the molecular determinants of evolution vulnerability is solid.

Weaknesses:

However, in my opinion, several points listed below need to be addressed.

(1) At the beginning of the results section, the authors started by determining the structures of Fab PVI.V3-9 and Fab PVI.V6-4 in complex with the ancestral SARS-CoV-2 spike. However, the readers could benefit from a brief introduction of the Fabs PVI.V3-9 and PVI.V6-4. The same applies to the anti-NTD Fabs.

(2) In Figure 1A and E, the spike protein is shown with two different views. It is best to show the same view for comparison.

(3) Throughout the manuscript, the map quality of some Fabs (e.g., V6-11, V6-7, V6-2) is suboptimal. Does the map support the claims on the residues that form the interface? The authors should provide a figure showing the cryo-EM density for all side-chain residues involved at the interface.

(4) Line 152, the terminology "NTD top binders" could be ambiguous, as it could mean those Fabs have the strongest binding affinity. Maybe the authors can change the "top" to "tip".

(5) The authors described the interface between the spike protein and the Fabs in great detail. However, it would be nice if the authors could summarize the common binding strategy for each group of antibodies that utilize the same binding surface.

(6) Line 275, the authors should define what strain of Omicron is in Figure 4. The authors should also explain that the strains in Figure 4A are ordered by evolutionary age.

(7) Lines 286-287, isn't this conclusion already made from the cell-based flow cytometry binding assay? This sentence could be deleted.

(8) In both Figures S10 and S11, the readers could benefit from an additional row highlighting the residues interacting with ACE2.

(9) Lines 298-301, based on Figure S11, no contact is made between the N2 loop and the Fab. The authors may elaborate on why the mutations observed in the N2 loop indirectly influenced Fab recognition.

(10) Lines 321-323, even though this is a well-established assay, it is probably better to clearly explain that one pool of cells expresses the spike and the other pool of cells expresses ACE2.