Abstract
Rabies vaccines require repeated immunization to robustly elicit neutralizing antibodies that prevent fatal diseases. Here, we analyzed rabies glycoprotein antibody repertoires at both polyclonal and monoclonal levels following repeated vaccination. Booster vaccination dramatically elevated the neutralizing activity of recalled antibodies, primarily targeting an immunodominant site III epitope with hydrophilic and rugged structures. Strikingly, the majority of site III-directed antibodies in the recall response used a convergent VH gene (IGHV3-30), and they exhibited more hydrophilic and shorter paratopes than non-site III antibodies, providing physicochemical advantages for binding to site III. Additionally, several amino acids on heavy chain CDR3 were identified as key sites for acquiring an ultrapotent neutralizing activity through site III binding. Our in-depth analysis of antibody repertoires revealed the molecular signatures of neutralizing antibodies generated by repeated rabies vaccination, possibly as a result of adaptive convergence.
Similar content being viewed by others
Introduction
Rabies is a highly fatal zoonotic infectious disease caused by lyssaviruses, among which the rabies virus (RABV) is the most common causative virus of human rabies1. According to the World Health Organization (WHO)2, rabies is estimated to cause 59,000 human deaths every year, with 95% of cases occurring in Africa and Asia3,4. Despite an extremely high fatality rate (nearly 100%) after symptom onset, rabies infection is preventable by vaccination in both pre- and post-exposure settings5,6,7. At present, the inactivated vaccine platform is the most widely used for humans, and repeated vaccination is currently required to elevate the neutralizing antibody titer above 0.5 international unit/mL2, which is established as the global cut-off required for antibody-mediated protection. Although the neutralizing antibody titers persist for several years after primary PrEP or PEP vaccination8,9, booster vaccination is more efficient in maintaining an adequate neutralizing antibody titer for a longer period of time10. Therefore, booster vaccination is recommended for sustained protection when the serum antibody titer declines below the internationally established protective threshold.
RABV is a single-stranded RNA virus belonging to the family Rhabdoviridae and the genus Lyssavirus11. The RABV genome encodes five proteins, nucleoprotein (N), phosphoprotein (P), matrix protein (M), glycoprotein (G), and RNA-dependent RNA polymerase (L)12. RABV glycoprotein (RABV-G) is a trimeric class III protein that exists on the viral surface, undergoing a reversible transition between pre- and post-fusion conformations in a pH-dependent manner. Prefusion RABV-G mediates viral attachment to cellular receptors and the post-fusion conformation triggered by acidic pH facilitates fusion for host-cell entry13,14. With these antigenic properties, RABV-G has been identified as the sole antigen targeted by neutralizing antibodies so far. The antigenic structure of RABV-G has been extensively studied, and antigenic sites I, II, and III have been shown to play important roles in virus neutralization among four major antigenic sites (I, II, III, and IV)15,16,17,18. Antigenic sites I and III are often targeted by neutralizing antibodies against phylogroup I lyssaviruses, including rabies virus, while neutralizing antibodies against antigenic site II, composed of discontinuous two regions, appear to play more dominant roles in phylogroup II neutralization16,19.
Although structural data of trimeric pre-fusion RABV-G in complex with a handful of neutralizing antibodies provide mechanistic insights into antibody neutralization, comprehensive analysis of the neutralizing antibody repertoires elicited by vaccination remains largely limited to serological analysis. In this study, we performed in-depth profiling of RABV-G antibodies, not only at the polyclonal level but also at the monoclonal level, through the isolation of a monoclonal antibody panel from memory B (Bmem) cells before and after the booster vaccination. The antibody repertoire analysis is based on the following parameters: neutralizing activities, epitope specificity, binding properties, variable region gene usages, and hydropathy. We found profound improvement in the neutralizing activities of secreted antibodies, as well as B-cell antigen receptors on Bmem cells after the booster vaccination. The neutralizing antibodies extensively focused on an immunogenic site III epitope through a particular VH gene (IGHV3-30) which is frequently used in human antibody repertoires carrying hydrophilic CDRs. Our results suggest that epitope recognition through convergent IGHV3-30-encoded antibodies provides an inherent advantage in human immunity for promptly eliciting rabies-neutralizing antibodies in a wide population.
Results
Study cohort
We voluntarily recruited 20 healthcare workers who received the inactivated rabies vaccine (PVRV, SPEEDA®) and longitudinally provided blood samples. The vaccine group was divided into a prime group (n = 10) and a boost group (n = 10) (Fig. 1a). Circulating antibodies and Bmem cells in the peripheral blood were analyzed by multiple parameters. The median ages in the prime and boost groups were 56.5 (33–70) and 61.0 (37–66) years, respectively. Most participants (19 out of 20) were male. The baseline vaccination history and neutralizing antibody status of all participants were investigated in our previous study20. None of the participants had a history of rabies vaccination or anti-rabies antibody production prior to the prime regimen.
a Schematic diagram of the sample collection. Subjects who had not received the inactivated rabies vaccine (SPEEDA) (prime group; n = 10, median age 56.5 years, ranging from 33 years to 70 years) and those who had received it one year prior (boost group; n = 10, median age 61.0 years, ranging from 37 years to 66 years) were enrolled. Plasma and PBMCs were collected before (day 0) and after (day 28) the vaccination. b RABV-G binding IgG titers and avidity in plasma were longitudinally quantified by ELISA. Avidity was assessed by using 3.5 M urea which facilitates the detachment of the low avidity antibodies. Data were pooled from two independent experiments. Fold changes are indicated above. Each dot represents the data from each individual, and dots connected with lines indicate data from the same individuals. c Schematic diagram of RABV-G showing non-overlapping antigenic site locations of RVC20 (antigenic site I, yellow), M777-16-3 (antigenic site II, purple), and CR4098 (antigenic site III, red). Numbering corresponds to the mature protein after cleavage from its secretion signal peptide (left). The locations on the 3D structure of the RABV-G trimer (PDB: 7U9G are highlighted with the same colors (right). d Site-specific IgG titers in plasma were longitudinally quantified by CEE-cELISA. Monoclonal antibodies with a known antigenic site (RVC20, M777-16-3, and CR4098) were used to quantify the site-specific IgG titers. The cut-off value set the 30% inhibition. Data were pooled from two independent experiments. Fold changes are indicated above. Each dot represents the data from each individual, and dots connected with lines indicate data from the same individuals. Statistical analyses were conducted using the Wilcoxon test (ns: p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001) (b, d).
RABV-G site III is an immunodominant epitope of vaccine-elicited neutralizing antibodies
Rabies vaccines are given in multiple doses over a short period to raise neutralizing antibody titers above the protection cut-off. In this vaccination cohort, inactivated rabies vaccines were given intradermally on both arms (0.1 mL per arm) on days 0 and 7 for priming, based on the WHO guideline 201721. Four weeks later, the priming group donated peripheral blood. The boost group received further intradermal vaccination on either arm at 0.1 mL per arm approximately one year after the priming. The peripheral blood was collected on day 0 and day 28 post-vaccination for both groups (Fig. 1a). Plasma antibodies were subjected to ELISA to quantify the amounts of RABV-G-binding IgG titers and to assess the avidity index based on urea treatment for detaching the plate-bound IgGs. Whereas RABV-G IgG titers modestly increased on day 28 after the priming, the additional booster at a one-year interval robustly recalled IgG titers, leading to a 33-fold increase (Fig. 1b). Despite the remarkable recall response after the booster, the overall increase in avidity indexes remained quite modest (1.4-fold).
RABV-G has three major sites as neutralizing epitopes: sites I, II, and III (Fig. 1c)22. Monoclonal antibodies against each antigenic site have been identified in mice and humans, but it remains unknown which sites are immunodominant in eliciting neutralizing antibodies after vaccination in this regimen. Here, we created an epitope map of the RABV-G antibodies based on the competition between circulating antibodies in plasma and reference monoclonal antibodies against individual epitopes. Because site I and site III are considered as the major neutralizing epitopes for RABV-G, competition assays using the site I (RVC20) and site III (CR4098) monoclonal antibodies have been performed23,24. Additionally, plasma samples were subjected to a competition assay using a site II antibody (M777-16-3)25. Whereas the site I and site II antibodies remained undetectable in the priming group, the site III antibodies were elicited above detectable levels in many, but not all, individuals (Fig. 1d). Higher amounts of site III antibodies were also observed in the booster group, revealing the immunodominant nature of the site III epitope in the vaccine-elicited RABV-G antibodies.
Induction and restimulation of RABV-G-binding Bmem cells after the vaccination
We then tracked the frequencies and phenotypes of RABV-G-binding Bmem cells in peripheral blood by flow cytometry. RABV-G-binding B cells were detected using fluorescent-coupled RABV-G baits (Fig. 2a). Two types of RABV-G baits coupled with different dyes were utilized to exclude the cells bound to the dyes rather than RABV-G itself. RABV-G Bmem cells (CD19+CD20+RABV-G+) were selected after gating out dead cells and dump+ cells mainly composed of non-B-cells. RABV-G-binding Bmem cells were detected at elevated levels in both the prime (0.068%) and boost (0.088%) groups compared to control specimens from pre-vaccination (5.5 ⨯10−3%) (Fig. 2b). For the prime group, both IgM+ and IgG+ fractions were equally present in the RABV-G-binding Bmem population, whereas IgG+ fraction became dominant in the boost group (Fig. 2c). Then, IgG+ Bmem cells were further divided into four subsets based on CD21/CD27/CD11c/FcRL5 expression (Fig. 2d); resting or classical subset (CD21+CD27+), activated subset (CD21−CD27+), CD27low subset (CD21+CD27low/−), and atypical subset (CD21−CD27low/−CD11c+FcRL5+)26,27. The percentages of each Bmem subset were compared between the prime and boost groups at day 28 after vaccination (Fig. 2e). In both the prime and boost groups, activated subset, which is normally a minor population at steady state28,29, occupied about half of the RABV-G-binding IgG+ Bmem population in peripheral blood, suggesting the efficient B-cell stimulation via the priming and boosting by intradermal vaccination.
a Gating strategy for FACS sorting of RABV-G-binding memory B cell. b Representative FACS plots of RABV-G-binding B cells in PBMCs of naïve and vaccinated groups (day 28 after the prime and boost vaccination). c The frequencies of RABV-G binding B cells expressing IgG, IgM, and IgA after the prime and boost vaccinations were plotted (n = 5). Each dot represents the data from each individual. Statistical analyses were conducted using the Kruskal–Wallis test (*p < 0.05; **p < 0.01). d RABV-G binding IgG+ B cells (CD19+CD20+ RABV-G+IgG+ cells) were subdivided into four Bmem subsets, including activated (CD21−CD27high), atypical (CD21−CD27lowCD11c+FcRL5+), CD27low (CD21+CD27llow), and resting (CD21+CD27high) Bmem cells. e The frequencies of the four Bmem subsets were determined as gated in (d). Each dot represents the data from each individual, and dots connected with lines indicate data from the same individuals (n = 5). Statistical analyses were conducted using the Wilcoxon test (*p < 0.05) (upper). Pie charts showing the percentages of four Bmem subsets among RABV-G binding IgG+ B cells are depicted. Statistical analyses were conducted using the Friedman test and chi-square test (ns: p > 0.05) (lower).
Robust increase in the neutralizing potency of Bmem cell repertoire following the booster vaccination
To characterize the neutralizing potencies, epitopes, and immunoglobulin genes of the antibodies preserved in IgG+ Bmem cells, we sorted RABV-G-binding IgG+ Bmem cells from the prime and boost groups for single-cell Nojima culture analysis (Fig. 3a)30. We then screened for B cells producing RABV-G-binding IgG in the culture supernatant and obtained 107 antibody-positive clones (prime:37, boost:70), reflecting higher numbers of Bmem cells per donor from the boost group (prime:52 cells per donor, boost:114 cells per donor). Finally, 79 VH/VL sequence datasets were successfully collected for producing recombinant monoclonal antibodies (prime:22, boost:59). These IgG antibodies were reconstructed, and their neutralizing potencies were examined using a replication-competent RABV that carries a reporter gene31. Only 23% of IgG clones from the prime group showed detectable levels of neutralizing activities (<9 µg/mL) (Figs. 3b, c), possibly reflecting the lower RABV-G binding of the prime group IgG clones due to insufficient affinity maturation (Fig. 3d). In contrast, neutralizing activities were detected in 95% of the booster group clones with a high neutralizing potency (IC50: 31.0 ng/mL), which was over 100-fold higher than those from the prime group. Additionally, there was a significant difference in the number of somatic hypermutations, with a remarkable accumulation of somatic hypermutations in the boost group (Fig. 3e).
a Schematic illustration of the experimental workflow for the analysis of RABV-G binding antibodies preserved in Bmem cells at the monoclonal level. RABV-G binding IgG+ Bmem cells were FACS sorted and cultured at single cell level and RABV-G antibody-producing clones were determined using the culture supernatants. b The neutralizing antibody titers against rabies virus (CVS-AcGFP) were compared between monoclonal antibodies established from Bmem cells of the prime group (n = 22) and those from the boost group (n = 57). Each dot represents the data from each monoclonal antibody. IC50 < 9 µg/mL was considered NT+. c Pie chart showing the percentages of NT+ clones of both prime and boost groups are depicted. d The area under the curve (AUC) of each monoclonal antibody against RABV-G was compared between the prime (n = 22) and boost (n = 57) groups. Each dot represents the data from each monoclonal antibody. e Distribution of somatic hypermutation (SHM) for each monoclonal antibody. Statistical analyses were conducted using the two-tailed Mann–Whitney test (b, d, and e) and chi-square test (c) (****p < 0.0001).
Dominant site III recognition by Bmem cell repertoire following booster vaccination
Repeated vaccinations with COVID-19 and other vaccines can sometimes alter the epitope preference of antibodies present in plasma and Bmem cells, shifting from immunodominant to subdominant epitopes, as the number of the vaccine doses increases32,33. To examine whether an epitope shift occurs after repeated rabies vaccination, we determined the binding epitopes of Bmem-derived IgG clones by competitive ELISA as described. First, competitive ELISA was performed using site I and site III reference antibodies, as these sites are dominantly targeted by RABV-neutralizing antibodies. Thereafter, IgG clones not competing with the site I and site III antibodies were subjected to a site II competition assay using the M777-16-3 clone to categorize them into either the site II or undetermined group. The threshold for the competition was set at 75% (Supplementary Fig. 1). The competition by high-affinity IgG clones sometimes hampers the mapping of low-affinity IgG clones, due to the detachment of pre-bound IgG clones by high-affinity reference antibodies. Indeed, more than half of the prime group IgG clones did not compete with any site of reference IgG clones, resulting in half of IgG clones being categorized into the undetermined group (Fig. 4a). The remaining clones from the prime group were categorized into either site II or III epitopes, with none into the site I epitope. In contrast, the majority of boost group clones were categorized into either site I, II, or III and only 7.0% of IgG clones remained undetermined for the epitopes, presumably reflecting their higher-affinity binding compared to the prime group clones. Consistent with the immunodominant nature of site III epitopes in circulating antibodies (Fig. 1d), the site III IgG clones were most frequently present in the booster group, occupying about half of the IgG clones. Despite the quantitative dominance of site III antibodies, the neutralizing potencies of the site III-directed antibodies were comparable to those targeting site I and site II (Fig. 4b). Collectively, the RABV booster vaccination recalled potently neutralizing antibodies focused on the site III epitope on RABV-G, the primary target of neutralizing antibodies against RABV.
a Epitope distributions of the monoclonal antibodies from both prime (n = 22) and boost (n = 57) groups are shown. Monoclonal antibodies competed with RVC20, M777-16-3, and CR4098 and were categorized as site I, site II, and site III, respectively. b Neutralizing antibody titers of the monoclonal antibodies that recognize each antigenic site are shown. Statistical analyses were conducted using the Kruskal–Wallis test (ns: p > 0.05).
Convergent V gene usage by site III-directed Bmem cells
The site III immunodominance following the booster vaccination is an intriguing phenomenon that is potentially relevant to vaccination regimens aimed at promptly elevating neutralizing antibody titers above the protective cut-off. To gain insights into the underlying mechanism for this epitope focus, we determined the VH/VL gene usage in the Bmem cell repertoires following the booster vaccination. While the site I and II antibodies utilized a wide range of VH genes without apparent bias towards specific VH genes, about 72.4% of site III antibodies were encoded by a single VH gene, IGHV3-30 (Fig. 5a). Light chain analysis of IGHV3-30+ clones also revealed a bias toward IGKV1-9 and IGKV1-16 (Fig. 5b), indicating highly convergent V gene usage by the site III antibodies.
a The IGHV gene usages for each site-specific antibody are shown. Percentages of the IGHV gene usage among site I + site II (n = 24) and site III antibodies (n = 29) in the boosting group. IgBLAST (https://www.ncbi.nlm.nih.gov/igblast/) was used to assign the IGHV for each antibody. The most frequently used IGHV3-30 in site III antibodies is highlighted in red. b IGLV gene usages in IGHV3-30 possessing clones among site III antibodies. c, d Spearman’s correlation analysis between IC50 vs RABV-G binding activity (AUC) of all antibodies (left) and potent neutralizing antibodies (right, IC50 < 100 ng/mL) for site I + site II (c) or site III (d) antibodies are shown. Each dot represents each monoclonal antibody. The most potent neutralizing clone, 4G9, was plotted in red (d). The Spearman r- and p-values are shown.
To assess the biological impact of this convergent site III antibody responses, we examined the correlation between binding capability and neutralizing potencies of antibodies against site III and non-site III epitopes (Fig. 5c, d). There was no correlation between the two antibody parameters among the non-site III antibodies; however, the neutralizing potencies of site III antibodies correlated with their binding capability (r = 0.389, p = 0.037), and highly potent antibodies (IC50 < 100 ng/mL) such as the 4G9 clone (red dot) showed an even stronger correlation (r = 0.676, p = 0.0004). The homologous antibody response due to convergent V gene usage may enhance the linkage between binding and neutralization, leading to a robust increase in neutralizing potencies in a manner that correlates with affinity maturation, which is evident after the booster vaccination.
The site III recognition by hydrophilic and short paratopes in convergent antibodies
Next, we examined the potential advantages of the preferential use of IGHV3-30 against the site III epitope. The prevalence of IGHV3-30 in the naive IgM repertoire contributes to its frequent involvement in antibody responses against various antigens34, thereby raising the likelihood of IGHV3-30 usage by vaccine-elicited antibodies. In addition, hydropathy is one of the key biochemical determinants of antibody-antigen binding and the selection of antigen-reactive B-cells35,36,37, as observed in SARS-CoV-2 neutralizing antibodies binding to hydrophobic epitopes in the receptor-binding domain37. To gain insights into the underlying physicochemical mechanisms for the dominant IGHV3-30 usage against site III, we calculated hydropathy using spatial aggregation propensity38 for the trimeric RABV-G (PDB: 7U9G, chains A, B, and C). This calculation revealed that sites II and III are hydrophilic epitopes, with comparable scores, whereas site I is relatively more hydrophobic (Fig. 6a, b).
a Distribution of hydropathy on the surface of the RABV-G 3D structure (PDB: 7U9G, chains A–C). The tendency for hydrophilicity (blue) and hydrophobicity (red) is indicated in the scale bar. Each antigenic site is delineated by lines, with site I in yellow, site II in purple, and site III in red. b Hydropathy assessment of RABV-G epitope sites. Each dot represents an amino acid comprising each epitope. Data are presented as median value ± SD. c, d Hydropathy assessment of CDR-H2 for all antibodies in the prime group (n = 22) and boost group (n = 57). The scores are compared between IGHV3-30 and other VH genes in the prime and boost groups (c), and among each site (d). Site III antibodies were subdivided into two groups depending on the use of IGVH3-30 or other VH genes (d). The most potent neutralizing clone, 4G9, was plotted in red (d). Data are presented as median value ± SD. Statistical analyses were conducted using the Kruskal–Wallis test (b) and two-tailed Mann–Whitney test (c, d) (ns: p > 0.05; ****p < 0.0001).
We also computed the hydropathy of CDRs in antibodies from the prime and boost groups. The structures of the antibodies were predicted using Ig-Fold39. To explore the potential advantages of the IGHV3-30 gene, we further subdivided each group into IGHV3-30-encoded and non-encoded antibodies (Fig. 6c and Supplementary Figs. 2a and 3a). CDR-H2 from the IGHV3-30-encoded antibodies tended to be more hydrophilic than those from non-IGHV3-30 antibodies in the prime group, suggesting the inherent hydrophilic nature of IGHV3-30 CDR-H2. A similar trend was observed in the boost group. On the other hand, CDR-L2 from IGHV3-30-encoded antibodies was more hydrophobic than those from non-IGHV3-30 antibodies; this is largely attributed to the preference usage of hydrophobic kappa light chains (IGKV1-9 and IGKV1-16) by the IGHV3-30-encoded antibodies (Fig. 5b).
We then recategorized these hydropathy scores by each epitope site (Fig. 6d and Supplementary Figs. 2b and 3b). Again, CDR-H2 hydropathy tended to be more hydrophilic in the site III antibodies than in non-site III antibodies, with this difference more pronounced in IGHV3-30-encoded antibodies (Fig. 6d). Thus, these data suggest a potential role of CDR-H2 hydrophilicity in the predominant recall of IGVH3-30-encoded antibodies following the booster vaccination. Interestingly, the most potently neutralizing site III clone, 4G9 (red dot in Fig. 6d), exhibited the lowest CDR-H2 hydrophobicity among the IGHV3-30-encoded antibodies against site III (Fig. 6d), further highlighting the importance of CDR-H2 hydrophilicity for binding to site III and enhancing neutralization potency. In general, the contribution of CDR-L2 to antigen recognition is very limited, with CDR-H2 playing a much more significant role40. Therefore, while a selective advantage of the IGHV3-30 gene likely stems from the hydrophilicity of CDR-H2 towards the hydrophilic site III epitope, the selection of a hydrophobic CDR-L2 from IGKV1-9 and IGKV1-16 genes may be due to other factors, which are further discussed below.
In addition to hydropathy, shape complementarity at the interface is another important factor driving antibody-antigen interactions41. Although precise analysis of such complementarity requires 3D structures of antibody-antigen complexes, a repertoire-scale analysis based on 3D structure or accurate prediction is not feasible. Instead, we took advantage of the notion that the length of CDRs significantly influences the shape of the paratope42. Therefore, we analyzed the length of each CDR in the neutralizing antibodies. Interestingly, CDR-L1, L3, and H3 in the site III antibodies were shorter than those in site I/II antibodies (Fig. 7 and Supplementary Fig. 4). Specifically, the shorter CDR-L1 and L3 lengths were attributed to the preferential use of the kappa light chain in the site III antibodies; 79% (27/34) of these antibodies were paired with kappa light chain, while only 23% (3/13) and 44% (7/16) of the site I and II antibodies used kappa light chain, respectively. Comparison of the predicted Fv model structures, along with an experimentally determined structure from an IGHV3-30-encoded antibody in the PDB (PDB: 8A1E, chains D, E), clearly demonstrated that the shorter paratopes, characterized by a 6-residue CDR-L1 and a 9-residue CDR-L3, contribute to flatter paratopes in the site III antibodies. Notably, unlike antibodies targeting other epitopes, which exhibit more diverse CDR lengths, 19 out of 23 the site III antibody clones encoded by IGHV3-30 possess an 11- or 12-residue CDR-H3, resulting in even flatter paratopes (Fig. 7d). Despite their shorter CDR-H3 lengths, these paratopes exhibited varied conformations of CDR-H3, including a variety of side chains, potentially contributing to more planar and yet rugged paratopes compared to those found in the site I/II antibodies. The site III epitope is located on the upper side of RABV-G at the foot of the protruding region that includes site II, making the molecular surface rugged (Fig. 1c), attributed to an Arg residue protruding into the solvent at site III. Therefore, these shorter CDR lengths of IGHV3-30-encoded antibodies against site III are largely due to the preference for kappa light chains encoded by IGKV1-9 and IGKV1-16, and are likely selected to adapt to the rugged, planar nature of the site III surface.
a–c Distribution of CDR lengths for each site-specific antibody. CDR-H3 length distribution (a), CDR-L1 length distribution (b), and CDR-L3 length distribution (c) are shown (for other CDRs, see Supplementary Fig. 4). Statistical analyses were conducted using the two-tailed Mann–Whitney test (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). d Structural comparison of antibodies targeting each site, arranged from top to bottom: 19 site III antibodies from IGHV3-30 featuring a 6-residue CDR-L1, a 9-residue CDR-L3, and either an 11- or 12-residue CDR-H3. The cryo-EM structure (8A1E, chains D, E) is shown in gray; 9 site III antibodies from other genes with a 6-residue CDR-L1 and a 9-residue CDR-L3; 12 site I antibodies; and 14 site II antibodies. All antibodies were superimposed onto a site I antibody, 1D7, using the Matchmaker tool of ChimeraX68. Heavy and light chains are shown in pink and green, respectively.
The site III recognition by electrostatic complementarity in convergent antibodies
Another important factor that determines antibody-antigen interactions is the formation of hydrogen bonds at the interface or, more broadly, electrostatic interactions. These electrostatic properties are often quantified by calculating electrostatic potentials, which involve solving continuum electrostatics equations using protein structures43 Since structures of antibody-antigen complexes are not available, we focused on the electrostatic properties of molecular surfaces by calculating electrostatic potentials, based solely on antibody or antigen structures alone.
We employed the APBS software44 to compute the electrostatic potential of molecular surfaces. On average, the electrostatic potential of each epitope site on the RABV-G surface was nearly neutral but exhibited some variations, with site III being more negative than the other epitope sites (Supplementary Fig. 5). When mapping the electrostatic potential onto the antibody surfaces, we observed that some CDR-H1 and CDR-H2 of IGHV3-30 antibodies tended to exhibit more positive characters than those of other antibodies (Supplementary Fig. 5). This may suggest another advantage of the IGHV3-30 gene in terms of electrostatic complementarity at the antibody-antigen interfaces, though these electrostatic effects are not as well defined as hydropathic interactions (Supplementary Figs. 5 and 6).
Key amino acids in CDR-H3 for site III binding by an ultrapotent neutralizing antibody
A previous study demonstrated that potent neutralizing antibodies to SARS-CoV-2 tend to exhibit more hydrophobic characteristics in CDR-H337. Indeed, neutralizing antibodies against RABV-G in this study also showed greater hydrophobicity in CDR-H3, compared to the other CDRs (Supplementary Fig. 2a). Although CDR- L1, L3, and H3 tend to be shorter in the convergent IGHV3-30-encoded, site III antibodies (Fig. 7), and the highly potent 4G9 clone possesses a hydrophilic CDR-H2 (Fig. 6d and Supplementary Fig. 2b), it is still possible that the site III antibodies utilize hydrophobicity in CDR-H3 for binding and neutralizing potency. First, we compared the CDR-H3 sequence of all anti-rabies neutralizing antibodies in this study, including 4G9, with those deposited in the Protein Data Bank (PDB). We identified RVA122, which has a 17-residue CDR-H3 containing five Tyr residues, in complex with RABV-G (PDB: 7U9G) (Fig. 8a). Interestingly, RVA122, also derived from a vaccinated individual but from a different germline origin, strategically positions its CDR-H3 between site II and III epitopes, with all Tyr residues interacting with the antigen (Fig. 8b). Notably, RVA122 features an Asp residue at position 107 in CDR-H3 (IMGT numbering), which forms a salt bridge with an Arg residue at the site III epitope of RABV-G. A similarly charged residue is found in the corresponding position of CDR-H3 in 4G9, suggesting its significance in antigen recognition. Additionally, the CDR-L3 of 4G9 and RVA122, both derived from lambda light chains, showed high sequence identity (64% over 11 residues), and their differing amino acids shared similar physicochemical properties (Fig. 8a). These observations suggest a similar, perhaps functionally convergent, binding mechanism between 4G9 and RVA122 to RABV-G. To test the role of CDR-H3 Tyr residues shared between 4G9 and RVA122, we conducted a mutational analysis on CDR-H3 of 4G9, replacing Tyr and other charged amino acids with alanine (Fig. 8c). This analysis revealed that seven of the eight mutants exhibited reduced binding, highlighting the critical role of these Tyr and charged Asp residues in site III recognition (Fig. 8d). Put together, the epitope specificity, neutralizing potency, V gene usage, and structural comparisons suggest convergent evolutionary trends in RABV-neutralizing antibodies, possibly as a result of adaptation to the immunodominant site III epitope with hydrophilic and rugged features.
a Phylogenetic tree of clustered CDR-H3 sequences from both the prime and boost groups. Homology verification using CDR-H3 amino acid sequences showed that 4G9 and RVA122 are closest in CDR-H3 sequence among all clones. b The structural image of the binding mode between RVA122 and RABV-G (PDB: 7U9G). c Amino acid sequence alignment of CDR-H3 in mutation 4G9 for alanine scanning. d Binding of each clone to RABV-G. The AUC was calculated from this graph to evaluate binding activity. Data were pooled from three independent experiments. Bars represent mean ± SD. Statistical analyses were conducted with the unpaired t-test (ns: p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001).
Discussion
Prompt and robust induction of neutralizing antibodies above the protection threshold is the primary aim of repeated vaccination with inactivated rabies vaccine. Although the immunogenicity of inactivated vaccines may be inferior to that of mRNA vaccines45,46, the booster vaccination with a one-year interval in this cohort robustly recalled the neutralizing antibodies, with prominent increases in both binding affinity and neutralizing potency, possibly due to the temporal maturation of humoral immune memory28,47. This approach, which involves a specific immunization schedule and dose, has been shown to be effective in inducing robust immune responses, as evidenced by our previous studies20,48. Here, we identified a previously unappreciated feature of RABV-neutralizing antibody responses that dominantly targeted the site III epitope only after the booster vaccination. Repeated vaccination can sometimes induce antibody feedback, whereby antibodies produced after the priming lead to at least two types of biases in the selection of B cells after the booster. One is the preferential selection of higher affinity antibodies recognizing the same epitope49, and the other is a shift of antibody epitopes from dominant to sub-dominant ones, as observed with SARS-CoV-2 mRNA vaccines33. The dominant appearance of site III antibodies in plasma after the priming and booster vaccination decreases the likelihood of an epitope shift after repeated rabies vaccinations. Instead, the antibody feedback in this rabies vaccine regimen is more likely to contribute to increased potencies of neutralizing antibodies against the immunodominant site III epitope, as represented by a significant increase in the neutralizing potencies of site III-directed antibodies after the boost vaccination (Fig. 4b).
The site III-directed antibodies were remarkably convergent and largely composed of a single VH gene, IGHV3-30, which is commonly expressed in a wide human population. These data suggest the inherent adaptation of the human antibody repertoire to the RABV-G site III epitope, as well as to SARS-CoV-2 spike or influenza hemagglutinin epitopes50,51,52, which leads to the frequent usage of IGHV3-30 gene in the neutralizing antibodies against these viruses. It should be stressed that the IGHV3-30 usage in the RABV-neutralizing antibodies was more frequent (72.4%) than in the I B cells (approximately 6%) and the cells specific to other viral antigens53, supporting the additional selection of this VH gene by RABV-G. Concordant with our findings, a previous study using a phage display approach identified IGHV3-30-encoded antibodies at the highest frequency among the RABV-G-binding fraction23, further strengthening the advantage of this VH gene for RABV-G binding.
The possible molecular basis for the preference for IGHV3-30 is its hydrophilic, polar CDR-H2, which likely enhances the recognition of the hydrophilic site III epitope. Another notable feature of IGHV3-30-encoded, site III antibodies is their shorter paratopes. While longer CDR-L1 contributes to pocket-like paratopes, shorter ones suggest flatter paratopes42, potentially making the antibodies more suitable for recognizing flatter epitopes, including RABV-G site III. Furthermore, CDR-L3 and H3, located geometrically at the center of the antigen-binding sites, suggest that shorter lengths of these CDR3s are beneficial for adapting to the planar site III epitope. Based on the geometric features of each epitope, site I is located on the side of RABV-G, site II at the rim of the upper part of RABV-G, and site III on the upper side at the foot of the protruding region that includes site II. Therefore, it is tempting to speculate that although antibodies may favor site II due to its increased accessibility via exposure to the solvent, they bind more avidly to site III because it is wider, flatter, and rugged. To adapt to these molecular features of site III, antibodies with shorter paratopes, such as those encoded by IGHV3-30, may be preferentially selected.
The question remains how such molecular differences in the IGHV3-30-encoded antibodies result in their prominent selection as site III binders only after the booster vaccination, not after the primary vaccination. Hydrophilic CDR-H2 and short paratopes already exist in the IGHV3-30-encoded antibodies in the native repertoires, possibly increasing the selective advantage over non-IGHV3-30-encoded antibodies after the initial priming. One possible scenario accounting for the observed IGHV3-30 preference solely after the booster vaccination would be the less stringent selection of Bmem cell recruitment from naïve B cells based on BCR affinity after the priming. Less stringent selection is known to permit the development of Bmem cells with low-affinity BCRs toward selecting antigens54,55, thereby decreasing the competitive advantage among B cells with modest differences in affinity. In contrast, the boosted antigen restimulates only a limited number of higher affinity Bmem cells that have a competitive advantage over other Bmem cells56,57, which likely increases the affinity threshold for Bmem cell restimulation. Therefore, we speculate that the slight advantage provided by hydrophilicity, shorter paratope, and other physical properties, such as electrostatic properties, of IGHV3-30-encoded antibodies eventually contribute to the prominent selection of Bmem cells carrying this VH gene after the booster.
Alanine scanning of the ultrapotent neutralizing antibody 4G9 revealed that contiguous Tyr residues in the CDR-H3 are involved in binding to site III. This role of CDR-H3 Tyr in site III binding is further supported by the structural analysis of another ultrapotent neutralizing antibody clone, RVA122. Although structural analysis of the 4G9 and RABV-G complex is essential for a deeper understanding of the possible roles of CDR-H3 Tyr, it is important to note that Tyr is naturally abundant in antibodies and generally crucial for antigen recognition58. Despite being categorized as a hydrophobic amino acid in our calculations, Tyr also exhibits hydrophilic properties due to its ability to form hydrogen bonds with surrounding residues or water molecules through its hydroxyl group. This amphiphilic nature makes Tyr ideally suited for antigen recognition.
In this study, this vaccination schedule has been shown to maintain effective neutralizing antibody titers in the body, even though the number of doses has been reduced compared to the conventional multi-dose schedule. This reduction in the number of doses could help lower vaccination costs and facilitate broader vaccine access, particularly in developing countries. In addition, booster vaccination produced antibodies that recognize additional epitope sites, which we hypothesize could provide enhanced protection against infection. Given these findings, this study is significant for future rabies prevention efforts, as it clarifies the mechanism of neutralizing antibody production by inactivated rabies vaccine and demonstrates that effective neutralizing antibodies can be generated even with a reduced vaccination schedule.
The mechanisms of action of neutralizing antibodies against rabies remain under debate59. Site III-directed antibodies may block both the conformational transitions between pre- and post-fusion forms and receptor interactions required for infection. This study highlights the importance of neutralizing antibodies targeting the site III epitope for future vaccine design. Our findings suggest that site III is a critical target for neutralizing antibodies, and focusing on this epitope could improve vaccine efficacy. Rationally designing immunogens to stabilize site III or present it in an optimized configuration could drive a stronger and more focused antibody response. Understanding the structural features of site III-binding antibodies could further guide the engineering of epitope mimics to preferentially elicit such antibodies.
The study was limited in cohort size due to the updated vaccination schedule based on the WHO guideline21. Besides, in our study cohort, 19 of 20 participants were male, which reflects the demographic characteristics of Village or Livestock Volunteers (VLVs) in a village in Thailand. This gender imbalance represents a limitation of our study, as immune responses could differ between males and females due to hormonal, genetic, and environmental factors. Although no significant deviations were observed in circulating antibody titers in the single female participant compared to male participants in the same prime group, a single female subject is insufficient to robustly assess potential sex-based differences in the vaccine responses. Future studies should aim to recruit a more balanced cohort to investigate potential sex-specific differences in immune responses to rabies vaccination and improve the generalizability of findings.
Furthermore, the exact contribution of the hydrophilic, polar CDR-H2 in IGHV3-30-encoded antibodies needs to be assessed through biochemical and structural approaches. It is also important to keep in mind that IGHV3-30 usage observed after vaccination with COVID-19 mRNA and other vaccines may have a different mechanistic basis.
Methods
Human samples
Healthy adult volunteers (n = 20) were vaccinated with the inactivated rabies vaccine (PVRV, SPEEDA®), which is chromatographically purified L. Pasteur PV206 strain grown in Vero cells. According to WHO guideline, PrEP recommendations for individuals at higher risk due to occupation or for sub-populations in remote rabies-endemic settings were updated to account for: (i) timely access to rabies biologicals; (ii) access to rabies serological testing; (iii) requirements for booster vaccination; and (iv) the presence of rabies in wildlife reservoirs. Based on these criteria, 20 villages or livestock volunteers (VLVs), who assist in providing rabies vaccinations for cats and dogs in their villages, were enrolled in the study. The study participants were at risk of exposure, especially VLVs from the Mae Kha subdistrict, San Pa Tong district, Doi Lor subdistrict, Doi Lor distinct, and Chiang Mai province. All participants were age-matched, and none had signs of acute infectious disease within three months prior to enrollment. Participants were separated into a prime group (n = 10, median age 56.5 (33–70)) and a boost group (n = 10, median age 61.0 (37–66)) and intradermally received the vaccine in both arms (0.1 mL per arm) at days 0 and 7. For the boost group, participants received a booster vaccination approximately after 1 year of priming on either arm at 0.1 mL per arm. Blood samples were collected longitudinally before receiving the vaccination at days 0 and 28 after the priming (prime group) and the boosting (boost group). The study was approved by the ethics committee of the Faculty of Associated Medical Sciences, Chiang Mai University (AMSEC-63FB-005) and the Institutional Review Board of the National Institute of Infectious Diseases (#1263). They were performed according to the Declaration of Helsinki. All volunteers provided written informed consent before the enrollment.
Blood sample preparation
The plasma samples were isolated via centrifugation and kept frozen at -80 °C until testing. The peripheral blood mononuclear cells (PBMCs) were isolated by ficoll density gradient centrifugation from heparinized blood. The separated PBMCs were washed twice with phosphate-buffered saline adjusted to a concentration of 1 × 107 cells/mL and cryopreserved.
Recombinant RABV-G
Human codon-optimized DNA fragment coding for the RABV-G of the rabies virus (GeneBank: BAA03837) was commercially synthesized (GenScript). This sequence along with signal peptide (amino acids 1-19; MVPQALLLVPILGFSLCFG), including a T4 foldon trimerization motif, histidine-tag, and Strep-tag were cloned into mammalian expression vector pCMV. Recombinant proteins were produced using Expi293F cells according to the manufacturer’s instruction (ExpiFectamine™ 293 Transfection Kit, Thermo Fisher Scientific). Supernatants from transfected cells were harvested at day 5 post-transfection and recombinant proteins were purified using Ni-NTA Agarose (QIAGEN). For the preparation of RABV-G fluorescent probes, Strep-tagged–RABV-G was incubated with PE or APC-labeled Strep-Tactin (IBA Lifesciences) at a 1:4 ratio overnight at 4 °C before using for FACS analysis.
Recombinant antibodies
Recombinant monoclonal antibodies and Fabs were produced as described previously37,60. Briefly, variable regions of immunoglobulin heavy and light chain from single-cell cultures or published monoclonal antibodies (RVC2061, M777-16-325, and CR409823) were cloned into the expression vectors of human IgG1 heavy chain, heavy chain CH1, kappa, or lambda light chain. Pairs of heavy and light chain vectors were co-expressed on the HEK293F system (Thermo Fisher Scientific), IgG1 was purified with a protein G column (Thermo Fisher Scientific), and Fab was purified using a Talon column (Clontech).
Flow cytometry
PBMCs were thawed at 37 °C and immediately washed with DMEM supplemented with 2% FBS, followed by staining with the RABV G-protein probes in DMEM supplemented with 2% FBS for 30 min at room temperature. The cells were washed with the medium and stained with subsequent antibodies/reagents using the Brilliant Stain Buffer Plus (BD Biosciences) for 30 min at room temperature: anti-IgA FITC [polyclonal rabbit F(ab′)2, Dako], anti-IgG-BV421 (G18-145, BD Biosciences), anti-human CD2-BV510 (RPA-2.10, BioLegend), anti-human CD4-BV510 (RPA-T4, BioLegend), anti-human CD10-BV510 (HI10a, BioLegend), anti-human CD14-BV510 (M5E2, BioLegend), LIVE/DEAD Fixable Yellow Dead Cell Stain kit (Thermo Fisher Scientific), anti-CD27-BV605 (O323, BD Biosciences), anti-FcRL5-BV650 (509F6, BD Biosciences), anti-CD19-BUV395 (HIB19, BD Biosciences), anti-CD20-BUV496 (2H7, BD Biosciences), anti-IgM-BUV563 (UCH-B1, BD Biosciences), anti-CD11c-BUV615 (3.9, BD Biosciences), and anti-CD21-BUV737 (B-ly4, BD Biosciences). The cells were washed with DMEM supplemented with 2% FBS, followed by resuspension in the medium, and analyzed using FACS Symphony S6 (BD Biosciences). Data were analyzed using FlowJo software (BD Biosciences).
Single B-cell culture
A single B cell culture was performed as described previously62. Briefly, CD19+ CD2- CD4- CD10− CD14− IgD− CD27+ IgG+ RABV-G+ B cells were single-cell sorted onto precultured MS40L-low feeder cells in 96-well plates containing RPMI 1640 medium supplemented with 10% FBS, 55 μM 2-mercaptoethanol (2-ME), penicillin (100 U/mL), streptomycin (100 μg/mL), 10 mM HEPES, 1 mM sodium pyruvate, 1% minimal essential medium non-essential amino acids, recombinant human interleukin-2 (IL-2; 50 ng/mL; PeproTech), recombinant human IL-4 (10 ng/ml; PeproTech), recombinant human IL-21 (10 ng/mL; PeproTech), recombinant human B cell activating factor belonging to the TNF family (BAFF, 10 ng/mL; PeproTech) for FACS Symphony S6 (BD Biosciences). The plates were incubated at 37 °C in a humid atmosphere with 5% CO2. The medium was half-replenished on days 4, 8, 12, 15, and 21. The supernatants were harvested on day 24 and the RNA was extracted from the cells using Direct-zol RNA miniprep (Zymo research) for immunoglobulin gene cloning and sequencing. Immunoglobulin genes (heavy chain and light chain genes) were amplified by two rounds of nested PCR with established primers and methods37,60,63. The V(D)J rearrangement and the number of somatic hypermutations were identified using IgBlast IMGT.
ELISAs
To quantify RABV-G binding IgG titers in plasma or single-cell culture supernatants, MaxiSorp 96- or 384-well ELISA plates (Thermo Fisher Scientific) were coated with 1 µg/mL RABV-G at 4 °C overnight. After blocking with ChonBlock™ Blocking/Sample Dilution Buffer (Chondrex), serially diluted plasma samples or single-cell culture supernatants were applied to the plate and incubated for 2 h at room temperature. After washing with PBS containing 0.1% Tween-20, 1/5000 anti-human IgG-HRP (SouthenBiotech) diluted with Can Get Signal #2 (TOYOBO) was added, and HRP activity was visualized with o-phenylenediamine (OPD, Sigma-Aldrich). After stopping the reaction with 2 N H2SO4, OD490 was measured in a microplate reader (Epoch 2, BioTek). A calibration curve was prepared using a control antibody (mAb15-1364) and the concentration of anti-RABV-G antibody in each plasma sample was calculated.
The site-specific IgG titers in plasma samples were quantified by CEE-cELISA using reference monoclonal antibodies (Site I: RVC20, Site II: M777-16-3, and Site III: CR4098) as previously described in ref. 65. Percentage inhibitions of plasma sample against reference antibodies were calculated as (1 − ODplasma/ODreference mAb) × 100 and the dilution factors showing65 30% inhibition were determined. Similarly, the binding site of the recombinant antibodies was determined by competition ELISA using Fabs of the same reference monoclonal antibodies, and antibodies showing < 75% inhibition against each reference antibody were classified into each site.
Virus infectivity neutralizing (NT) assay
Neutralizing activities of recombinant antibodies were tested by NT assay with replication-competent RABV challenge virus standard strain carrying fluorescent AcGFP gene (CVS-AcGFP). CVS-AcGFP was generated by the insertion of the AcGFP gene in the N/P intergenic region as described previously31. Briefly, recombinant antibodies were serially diluted 3-fold in MEM containing 2% FBS and mixed with 100 focus forming units (ffu) of CVS-AcGFP in 50 µL. After incubation at 37 °C for 30 min, the antibody-virus mixtures were added to BHK-21 C13 cells (JCRB, JCRB9020) in 96-well plates. After 4 days in culture, cells were fixed with a 4% paraformaldehyde phosphate buffer solution and stained with Hoechst 33342 to visualize cell nuclei. Fluorescent images were captured using a fluorescence microscope IX73 (Olympus) or an automated microscope IN Cell Analyzer 2500 HS (Cytiva). The IC50 values were calculated using GraphPad Prism 9 (GraphPad).
Hydropathy calculations and structure predictions
Hydropathy of the CDRs of each antibody and RABV-G was assessed using the spatial aggregation propensity (SAP) score38, calculated with the CHARMM program66. Electrostatic potentials were computed using the APBS software44, in which all calculations were performed at an ionic strength of 0.15 M in a 1:1 salt, with a protein dielectric constant of 2, a solvent dielectric constant of 78.54, and a temperature of 298.15 K. These analyses were based on a cryo-EM structure of RABV-G (7U9G, chain A, B, and C) and predicted models of antibodies generated with Ig-Fold39. To accommodate the uncertainties of the cryo-EM structure and predicted models, the SAP radius was set to 10 Å, enabling the identification of broader, low-resolution patches. The CDR definition in this study is based on the IMGT scheme67.
Data analysis
The numerical data were statistically analyzed and visualized with GraphPad Prism 9 software (GraphPad). Wilcoxon test, Mann–Whitney test, Kruskal–Wallis test, Friedman test, Chi-square test, and Dann’s multiple comparison test were performed. Differences with p-values less than 0.05 were considered significant: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.
Data availability
Sequence data for several anti-RABV-G monoclonal antibodies have been deposited in GenBank under accession numbers LC812179–LC812257 (VH) and LC812292–LC812370 (VL). Data are publicly available as of the date of publication. The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
References
Hemachudha, T. et al. Human rabies: neuropathogenesis, diagnosis, and management. Lancet Neurol. 12, 498–513 (2013).
World Health Organization Rabies vaccines: WHO position paper, April 2018—recommendations. Vaccine 36, 5500–5503 (2018).
WHO. Rabies. http://www.who.int/health-topics/rabies (2023).
Pattnaik, P. et al. Alarming rise in global rabies cases calls for urgent attention: current vaccination status and suggested key countermeasures. Cureus 15, e50424 (2023).
Giesen, A., Gniel, D. & Malerczyk, C. 30 years of rabies vaccination with Rabipur: a summary of clinical data and global experience. Expert Rev. Vaccines 14, 351–367 (2015).
Fooks, A. R. et al. Current status of rabies and prospects for elimination. Lancet 384, 1389–1399 (2014).
Xu, C. et al. Immunogenicity after pre- and post-exposure rabies vaccination: a systematic review and dose-response meta-analysis. Vaccine 39, 1044–1050 (2021).
Briggs, D. J. & Schwenke, J. R. Longevity of rabies antibody titre in recipients of human diploid cell rabies vaccine. Vaccine 10, 125–129 (1992).
World Health Organization. The Immunological Basis for Immunization Series: Module 17: Rabies. http://iris.who.int/handle/10665/259511 (2017).
Strady, A. et al. Antibody persistence following preexposure regimens of cell-culture rabies vaccines: 10-year follow-up and proposal for a new booster policy and Pasteur Mérieux Connaught, purpose and design, which was circulated by the French Caiss. J. Infect. Dis. 177, 1290–1295 (1998).
Dietzgen, R. G., Kondo, H., Goodin, M. M., Kurath, G. & Vasilakis, N. The family Rhabdoviridae: mono- and bipartite negative-sense RNA viruses with diverse genome organization and common evolutionary origins. Virus Res. 227, 158–170 (2017).
Schnell, M. J., McGettigan, J. P., Wirblich, C. & Papaneri, A. The cell biology of rabies virus: using stealth to reach the brain. Nat. Rev. Microbiol. 8, 51–61 (2010).
Albertini, A. A. V., Baquero, E., Ferlin, A. & Gaudin, Y. Molecular and cellular aspects of rhabdovirus entry. Viruses 4, 117–139 (2012).
Yang, F. et al. Structural analysis of rabies virus glycoprotein reveals pH-dependent conformational changes and interactions with a neutralizing antibody. Cell Host Microbe 27, 441–453.e7 (2020).
de Melo, G. D., Hellert, J., Gupta, R., Corti, D. & Bourhy, H. Monoclonal antibodies against rabies: current uses in prophylaxis and in therapy. Curr. Opin. Virol. 53, 101204 (2022).
Prehaud, C., Coulon, P., LaFay, F., Thiers, C. & Flamand, A. Antigenic site II of the rabies virus glycoprotein: structure and role in viral virulence. J. Virol. 62, 1–7 (1988).
Seif, I., Coulon, P., Rollin, P. E. & Flamand, A. Rabies virulence: effect on pathogenicity and sequence characterization of rabies virus mutations affecting antigenic site III of the glycoprotein. J. Virol. 53, 926–934 (1985).
Kuzmina, N. A., Kuzmin, I. V., Ellison, J. A. & Rupprecht, C. E. Conservation of binding epitopes for Monoclonal Antibodies on the rabies virus glycoprotein. J. Antivir. Antiretrovir 5, 037–043 (2013).
Evan, J. S. et al. Antigenic site changes in the rabies virus glycoprotein dictates functionality and neutralizing capability against divergent lyssaviruses. J. Gen. Virol. 99, 169–180 (2018).
Kewcharoenwong, C. et al. One-dose intradermal rabies booster enhances rabies antibody production and avidity maturation. Med. Microbiol. Immunol. 213, 7 (2024).
WHO. Rabies Vaccines and Immunoglobulins: WHO Position SUMMARY OF 2017 UPDATES. http://www.who.int/publications/i/item/WHO-CDS-NTD-NZD-2018.04 (2017).
Shi, C. et al. Research progress on neutralizing epitopes and antibodies for the Rabies virus. Infect. Med. 1, 262–271 (2022).
Kramer, R. A. et al. The human antibody repertoire specific for rabies virus glycoprotein as selected from immune libraries. Eur. J. Immunol. 35, 2131–2145 (2005).
De Benedictis, P. et al. Development of broad-spectrum human monoclonal antibodies for rabies post-exposure prophylaxis. EMBO Mol. Med. 8, 407–421 (2016).
Müller, T. et al. Development of a mouse monoclonal antibody cocktail for post-exposure rabies prophylaxis in humans. PLoS Negl. Trop. Dis. 3, e542 (2009).
Pape, K. A. et al. High-affinity memory B cells induced by SARS-CoV-2 infection produce more plasmablasts and atypical memory B cells than those primed by mRNA vaccines. Cell Rep. 37, 109823 (2021).
Ellebedy, A. H. et al. Defining antigen-specific plasmablast and memory B cell subsets in human blood after viral infection or vaccination. Nat. Immunol. 17, 1226–1234 (2016).
Kotaki, R. et al. SARS-CoV-2 Omicron-neutralizing memory B cells are elicited by two doses of BNT162b2 mRNA vaccine. Sci. Immunol. 7, 1–14 (2022).
Zurbuchen, Y. et al. Human memory B cells show plasticity and adopt multiple fates upon recall response to SARS-CoV-2. Nat. Immunol. 24, 955–965 (2023).
McCarthy, K. R. et al. Memory B cells that cross-react with group 1 and group 2 Influenza A viruses are abundant in adult human repertoires. Immunity 48, 174–184.e9 (2018).
Anindita, P. D. et al. Generation of recombinant rabies viruses encoding NanoLuc luciferase for antiviral activity assays. Virus Res. 215, 121–128 (2016).
Muecksch, F. et al. Increased memory B cell potency and breadth after a SARS-CoV-2 mRNA boost. Nature 607, 128–134 (2022).
Schaefer-Babajew, D. et al. Antibody feedback regulates immune memory after SARS-CoV-2 mRNA vaccination. Nature 613, 735–742 (2023).
Gidoni, M. et al. Mosaic deletion patterns of the human antibody heavy chain gene locus shown by Bayesian haplotyping. Nat. Commun. 10, 628 (2019).
Sinha, N., Mohan, S., Lipschultz, C. A. & Smith-Gill, S. J. Differences in electrostatic properties at antibody-antigen binding sites: implications for specificity and cross-reactivity. Biophys. J. 83, 2946–2968 (2002).
Lee, J. H., Yin, R., Ofek, G. & Pierce, B. G. Structural features of antibody-peptide recognition. Front. Immunol. 13, 910367 (2022).
Moriyama, S. et al. Structural delineation and computational design of SARS-CoV-2-neutralizing antibodies against Omicron subvariants. Nat. Commun. 14, 4198 (2023).
Chennamsetty, N., Voynov, V., Kayser, V., Helk, B. & Trout, B. L. Design of therapeutic proteins with enhanced stability. Proc. Natl. Acad. Sci. USA 106, 11937–11942 (2009).
Ruffolo, J. A., Chu, L. S., Mahajan, S. P. & Gray, J. J. Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies. Nat. Commun. 14, 2389 (2023).
Madsen, A. V. et al. Structural trends in antibody-antigen binding interfaces: a computational analysis of 1833 experimentally determined 3D structures. Comput. Struct. Biotechnol. J. 23, 199–211 (2024).
Kuroda, D. & Gray, J. J. Shape complementarity and hydrogen bond preferences in protein-protein interfaces: implications for antibody modeling and protein-protein docking. Bioinformatics 32, 2451–2456 (2016).
Raghunathan, G., Smart, J., Williams, J. & Almagro, J. C. Antigen-binding site anatomy and somatic mutations in antibodies that recognize different types of antigens. J. Mol. Recognit. 25, 103–113 (2012).
Nakamura, H. et al. Roles of electrostatic interaction in proteins. Q. Rev. Biophys.29, 1–90 (1996).
Jurrus, E. et al. Improvements to the APBS biomolecular solvation software suite. Protein Sci. 27, 112–128 (2018).
Premikha, M. et al. Comparative effectiveness of mRNA and inactivated whole-virus vaccines against coronavirus disease 2019 infection and severe disease in Singapore. Clin. Infect. Dis. 75, 1442–1445 (2022).
Lim, W. W., Mak, L., Leung, G. M., Cowling, B. J. & Peiris, M. Comparative immunogenicity of mRNA and inactivated vaccines against COVID-19. Lancet Microbe 2, e423 (2021).
Kim, W. et al. Germinal centre-driven maturation of B cell response to mRNA vaccination. Nature 604, 141–145 (2022).
Khamduang, W. et al. Antibody response to rabies pre-exposure vaccination among village health volunteers in a northern region of Thailand. Prev. Vet. Med. 199, 105555 (2022).
Victora, G. D. & Nussenzweig, M. C. Germinal centers. Annu Rev. Immunol. 40, 413–442 (2022).
Sakharkar, M. et al. Prolonged evolution of the human B cell response to SARS-CoV-2 infection. Sci. Immunol. 6, 1–14 (2021).
Kaku, C. I. et al. Recall of preexisting cross-reactive B cell memory after Omicron BA.1 breakthrough infection. Sci. Immunol. 7, eabq3511 (2022).
Harshbarger, W. D. et al. Unique structural solution from a VH 3-30 antibody targeting the hemagglutinin stem of influenza A viruses. Nat. Commun. 12, 1–13 (2021).
Robbiani, D. F. et al. Convergent antibody responses to SARS-CoV-2 in convalescent individuals. Nature 584, 437–442 (2020).
Shinnakasu, R. et al. Regulated selection of germinal-center cells into the memory B cell compartment. Nat. Immunol. 17, 861–869 (2016).
Viant, C. et al. Antibody affinity shapes the choice between memory and germinal center B cell fates. Cell 183, 1298–1311.e11 (2020).
Mesin, L. et al. Restricted clonality and limited germinal center reentry characterize memory B cell reactivation by boosting. Cell 180, 92–106.e11 (2020).
Blier, P. R. & Bothwell, A. A limited number of B cell lineages generates the heterogeneity of a secondary immune response. J. Immunol. 139, 3996–4006 (1987).
Fellouse, F. A., Wiesmann, C. & Sidhu, S. S. Synthetic antibodies from a four-amino-acid code: a dominant role for tyrosine in antigen recognition. Proc. Natl. Acad. Sci. USA 101, 12467–12472 (2004).
Ng, W. M. et al. Structure of trimeric pre-fusion rabies virus glycoprotein in complex with two protective antibodies. Cell Host Microbe 30, 1219–1230 (2022).
Onodera, T. et al. A SARS-CoV-2 antibody broadly neutralizes SARS-related coronaviruses and variants by coordinated recognition of a virus-vulnerable site. Immunity 54, 2385–2398.e10 (2021).
Hellert, J. et al. Structure of the prefusion-locking broadly neutralizing antibody RVC20 bound to the rabies virus glycoprotein. Nat. Commun. 11, 25–28 (2020).
Nojima, T. et al. In-vitro derived germinal centre B cells differentially generate memory B or plasma cells in vivo. Nat. Commun. 2, 465 (2011).
Tiller, T. et al. Efficient generation of monoclonal antibodies from single human B cells by single cell RT-PCR and expression vector cloning. J. Immunol. Methods 329, 112–124 (2008).
Luo, T. R. et al. A virus-neutralizing epitope on the glycoprotein of rabies virus that contains Trp251 is a linear epitope. Virus Res. 51, 35–41 (1997).
Aronthippaitoon, Y. et al. A cost effective easy competitive enzyme-linked immunosorbent assay suitable for monitoring protective immunity against the rabies virus in the serum of humans and dogs. Jpn J. Infect. Dis. 72, 99–105 (2019).
Brooks, B. R. et al. CHARMM: the biomolecular simulation program. J. Comput. Chem. 30, 1545–1614 (2009).
Lefranc, M.-P. & Lefranc, G. Antibody sequence and structure analyses using IMGT®: 30 years of immunoinformatics. Methods Mol. Biol. 2552, 3–59 (2023).
Meng, E. C. et al. UCSF ChimeraX: tools for structure building and analysis. Protein Sci. 32, e4792 (2023).
Acknowledgements
MS40L-low feeder cells were kindly provided by Garnett H. Kelsoe (Duke University, Durham, NC, USA). We thank the members of NIID for the discussion and support. The human sample collection of this project was supported by AMS CMU Challenge 2023/2024. This work was supported in part by the Japan Agency for Medical Research and Development (JP22fk0108141 and JP243fa627009) to Y.T., JP23gm1810004 to Y.T. and D.K., JP23wm0325047 and JP24wm0325075 to D.K., JP22fk0108571j0001 to T.O., and JP243fa627005 to M.S., Y.I., H.S., and Y.T.). The supercomputing resources in this study were provided in part by the Human Genome Center at the Institute of Medical Science, The University of Tokyo, and by the Research Center for Computational Science, Okazaki, Japan (Project: 24-IMS-C074).
Author information
Authors and Affiliations
Contributions
M.F., T.O., D.K., M.S., Y.I., K.Y. performed research, collected, and analyzed data. C.K., A.N., and G.L. collected human samples. N.I. and T.S. generated a key reagent. S.T., H.S., G.L., and Y.T. conceived, designed, and coordinated the study. M.F., T.O., D.K., and Y.T. wrote and edited the manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Fujisawa, M., Onodera, T., Kuroda, D. et al. Molecular convergence of neutralizing antibodies in human revealed by repeated rabies vaccination. npj Vaccines 10, 39 (2025). https://doi.org/10.1038/s41541-025-01073-5
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41541-025-01073-5