Mapping of human monoclonal antibody responses to XBB.1.5 COVID-19 monovalent vaccines: a B cell analysis

Abstract

Background: The rapid emergence of highly transmissible and immune-evasive SARS-CoV-2 variants has required the reformulation of COVID-19 vaccines to target these evolving threats. Although previous infections and booster vaccinations can boost variant neutralisation, it remains uncertain whether monovalent vaccines-delivered via mRNA or protein-based platforms-can trigger novel B-cell responses specific to omicron XBB.1.5 variants. We sought to address this uncertainty by characterising the antibody repertoire of individuals receiving a monovalent booster vaccine.

Methods: In this observational study, we analysed the genetic antibody repertoire of 603 individual plasmablasts from five individuals (recruited at the Icahn School of Medicine at Mount Sinai, New York, NY, USA, from STUDY-16-01215/IRB-16-00971 and STUDY-20-00442/IRB-20-03374) vaccinated with a monovalent XBB.1.5 vaccine, either through mRNA (Moderna or Pfizer-BioNTech; participants 1, 2, and 3) or adjuvanted protein (Novavax; participants 4 and 5) platforms. Before XBB.1.5 booster vaccination, all participants received mRNA-based priming and booster vaccine with ancestral SARS-CoV-2 and four of the five participants had a breakthrough omicron variant infection. We expressed 100 human monoclonal antibodies (mAbs; 50 from participants 1, 2, and 3, and 50 from participants 4 and 5) and evaluated their binding and neutralisation against various SARS-CoV-2 variants, including JN.1. We then selected four mAbs for in-vivo protection experiments by passive immunisation and viral challenge, and cryo-electron microscopy with two selected mAbs complexed with the XBB.1.5 spike (S) protein to determine their structures and binding interactions.

Findings: Between October and November, 2023, we enrolled three male and two female participants (mean age 46 years) all of whom were White. We identified 21 binding mAbs and tested their neutralisation capacity against ancestral SARS-CoV-2, XBB.1.5, and JN.1. From the six neutralising mAbs we characterised, we selected three (M2, M27, and M39) for in-vivo protection studies, along with one broadly binding antibody (M15), finding that three neutralising mAbs offered full protection against morbidity from XBB.1.5. M27 also displayed robust protection against the ancestral and JN.1 strains, and M39 offered partial protection from JN.1. Among these, we identified two standout antibodies: M2 and M39. M2 was uniquely specific to XBB.1.5, and M39 demonstrated the ability to bind and neutralise both XBB.1.5 and JN.1 strains. Using high-resolution cryo-electron microscopy, we mapped the binding sites of M2 and M39 on the XBB.1.5 S glycoprotein, uncovering the precise molecular interactions that dictate their specificity.

Interpretation: Our findings offer key molecular insights into whether strain-specific boosters elicit sufficient protection against SARS-CoV-2 emerging variants. This knowledge can inform decisions on booster design and strategies to enhance preparedness to evolving viral threats.

Funding: Icahn School of Medicine at Mount Sinai; National Institutes of Health (NIH) FIRST; Laura and Isaac Perlmutter Cancer Center Support Grant; National Institute of Allergy and Infectious Diseases; Human Immunology Project Consortium by NIH; the São Paulo Research Foundation; the National Heart, Lung, and Blood Institute of the NIH; Irma T Hirschl and Monique Weill-Caulier Trust; and the Collaborative Influenza Vaccine Innovation Centers.

Figure 1:. Antibody gene repertoire of XBB.1.5 monovalent vaccines

(A) A boxplot showing the frequency of nucleotide mutations in the V gene of heavy loci of plasmablasts from participants immunised with mRNA vaccines or recombinant protein monovalent vaccines. (B) A boxplot showing the heavy chain CDR3 length, measured by total amino acids. (C) Antibody isotype frequency per participant. IGHD and IGHE information was removed from visualisation. Boxplots show the IQR, with the median indicated by the central line. The whiskers extend 1·5 times the IQR from the upper and lower quartiles. Points beyond this range are considered outliers.

Figure 2:. Binding and neutralising capacity of mAbs isolated from participants vaccinated with either mRNA or protein-based SARS-CoV-2 XBB.1.5 vaccines

Binding activity of the mAbs against ancestral, XBB.1.5, and JN.1 S protein (A–C) and of ancestral, XBB.1.5, and JN.1 RBD (D–F). Antibodies that did not bind any antigen are not shown. Neutralising capacity of binding mAbs against ancestral, XBB.1.5, and JN.1 SARS-CoV-2 (G–I). Only mAbs binding to at least one antigen were selected for the in-vitro neutralisation assays and only the antibodies that neutralised at least one of the antigens tested are shown. The dashed line represents the limit of detection, which is set at the starting dilution of 30 μg/mL. Binding is defined by minimum binding concentration (μg/mL) and neutralisation is determined by IC₅₀. mAbs with minimum binding concentration /r IC₅₀≥30 μg/mL, or both, were considered negative and were assigned half the limit of detection for graphing purposes. Each experimental replicate is plotted, with error bars showing the mean and SD. For ELISAs and microneutralisations, an isotype-matched influenza virus anti-haemagglutinin mAb, CR9114, was utilised as a negative control. In the ELISA assays, an anti-6×His-tag was used as a positive control. For the microneutralisation assays remdesivir was used as a positive control. Results for M2, M15, M27, and M39 are summarised in table 2. IC₅₀=half-maximal inhibitory concentration. mAbs=monoclonal antibodies. P=protein-based antibodies. S=spike. RBD=receptor-binding domain.

Figure 3:. Protection in vivo by prophylactic treatment with IgG1 mAbs M2, M15, M27, and M39

Weight loss in mice treated with 10mg/kg (intraperitoneally) of M2, M15, M27, or M39mAbs 2 h before challenge with a 3 × LD₅₀ dose ofWA1/2020 (A), XBB.1.5 (B), and JN.1 (C). Survival curves of mice treated with M2, M15, M27, or M39 mAbs and infected with WA1/2020 (D), XBB.1.5 (E), and JN.1 (F). We used the isotype-matched influenza virus anti-haemagglutinin mAb, CR9114, as a negative control. Challenge experiments were completed once and there were five mice per experimental condition, except for the JN.1 challenge where three mice were used for M2 and four mice were used for the negative control (CR9114). Mean values with SDs are plotted. Survival of the mice treated with M2, M15, M27, and M39 was statistically compared to survival of the mice treated with CR9114 using the Mantel–Cox test. Numbers at risk at each timepoint and full protection results are summarised in table 2 and the appendix (p 25). LD₅₀=the dose at which a substance is lethal for 50% of animals tested. mAbs=monoclonal antibodies.

Figure 4:. Cryo-electron microscopy structures of M2 and M39 Fabs in complex with XBB.1.5 S

(A) Surface regions of the SARS-CoV-2 S protein contacted by the two antibodies M2 and M39. The global composite cryo-electron microscope map is shown as a transparent surface, with the NTD:M2 Fab and RBD:M39 Fab complexes docked and shown in cartoon representation. (B) The M2 Fab defines an epitope on the top side of the NTD. M2 creates an extensive surface contact area and engages NTD with both its heavy and light chains. (C) The M39 Fab defines an epitope on RBD engaged by numerous published class 3 antibodies. (D–E) 180° views along the y-axis that show details of the intermolecular interactions between the M2 Fab and NTD with numerous polar and non-polar interactions. (F–G) 180° views along the y-axis that show details of the M39 Fab:RBD molecular interface with very few interacting residues forming salt bridges. Spike residues are underlined. Fab=fragment antigen-binding region. NTD=N-terminal domain. pGLU=pyroglutamate. RBD=receptor-binding domain.