A third dose of inactivated vaccine augments the potency, breadth, and duration of anamnestic responses against SARS-CoV-2

Abstract

Dear Editor, The ongoing coronavirus disease 2019 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has lasted for more than four years, resulting in an unprecedented global public health crisis. Progress in halting this pandemic seems slow due to the emergence of variants of concern, such as the B.1.1.7 (Alpha), B.1.351 (Beta), P.1 (Gamma, also known as B.1.1.28.1), B.1.617.2 (Delta), and B.1.1.529 (Omicron), that appear to be high transmissible and more resistant to neutralizing antibodies (Wang et al., 2021g). New variants are thought to be responsible for re-infections (Hacisuleyman et al., 2021). A general decrease in immune protection against SARS-CoV-2 variants within 6–12 months after the primary infection or vaccination is also observed (Widge et al., 2021). However, not much is known about the immunogenic features of such a booster dose of a COVID-19 vaccine. In addition, there are large gaps in our understanding ofcorrelating immunogenic findings from surrogate endpoints to gauge vaccine efficacy. The CoronaVac, a 3-dose β-propiolactone-inactivated vaccine against COVID-19, has been approved for emergency use by the World Health Organization (Gao et al., 2020). To evaluate immune features, we recruited 22 COVID-19 convalescents, 6 healthy participants (SARS-CoV-2 negative, confirmed by RT-PCR), and 38 volunteers who received either 2 or 3 doses of the Coronavac vaccine for blood donation. None of the volunteers recruited for vaccination was infected by SARS-CoV-2 before the study. Blood samples from convalescents and vaccinees collected 1.3 months after infection and the indicated times after vaccination were used in this study, respectively, to compare humoral immune responses elicited against circulating SARS-CoV-2 variants. Neutralizing antibodies (NAbs) are a major correlate of protection for many viruses, including SARS-CoV-2. Neutralizing activity of plasma samples from 66 participants was measured against WT, B.1.351, P.1 and B.1.617.2 using live SARS-CoV-2 and VSV-pseudoviruses with the S from WT, B.1.1.7, P.1 variants and SARS-CoV (Fig. 1). The geometric mean half-maximal neutralizing titers (GMT NT50) against live SARS-CoV-2 in plasma obtained from convalescents and from vaccinees suggest an approximately 60% higher neutralizing activity against WT after 3-dose inoculation when compared with 2-dose administration, and 20% higher than those from convalescents (Fig. 1A). Interestingly, for the samples from the convalescents, 2-dose and 3-dose vaccinees, neutralizing titers against B.1.351 were, on average, 7.7-fold, 5.7-fold and 3.0-fold reduced, respectively, compared with WT (Fig. 1A). Similarly, fold decreases in neutralization ID50 titers against P.1 and B.1.617.2 for the three cohorts were 5.3, 4.3 and 3.1, and 5.3, 3.7 and 2.3, respectively (Fig. 1A). Overall, plasma of the 3-dose vaccinees displayed minimal reduction in neutralization titers against several authentic VOCs compared to the convalescents and 2-dose vaccinees (Fig. 1A). In line with the results of live SARS-CoV-2 neutralization assay, the mean fold decrease in the neutralization of B.1.1.7 relative to the WT was 2.8-fold for convalescents, 2.2-fold for 2-dose vaccinees and 1.7-fold for 3-dose vaccinees by using pseudo-typed viruses (Fig. 1B). Similarly, plasma from convalescents, 2-dose and 3-dose vaccinees exhibited a 4.5-fold, 2.9-fold and 2.4-fold reduction, in NAb titers against P.1, respectively, when compared to the WT (Fig. 1B). These results reveal that a third-dose boost of inactivated vaccine leads to enhanced neutralizing breadth to SARS-CoV-2 variants when compared to convalescent plasma. Landscape of antibodies elicited by a 3rd-dose booster of an inactivated vaccine. (A) Plasma neutralizing activity evaluated by authentic SARS-CoV-2 and (B) pseudo-typed SARS-CoV-2 neutralization assays. Left: half-maximal neutralizing titer (NT50) values for plasma from COVID-19 convalescents, 2-dose, 3-dose CoronaVac vaccine recipients (at week 4 after the last dose of vaccination) and negative controls (pre-COVID-19 historical control) against live SARS-CoV-2 WT, B.1.351, P.1 and B.1.617.2, and VSV-based SARS-CoV-2 pseudoviruses bearing WT or B.1.1.7 or P.1 S protein. Black bars and indicated values represent geometric mean NT50 values. Statistical significance was determined using the two-tailed Wilcoxon matched-pairs test. Experiments were repeated in triplicate. Dotted lines indicate the limit of detection. Right: fold decrease in neutralization for each variant relative to WT for each cohort of plasma samples (calculated from the left datasets) is shown. (C) Longitudinal neutralizing titers of plasma from 3-dose vaccinees at days 0, 7, 14, 28, 90, and 180 post the 3rd-dose vaccination. The geometric mean NT50 values are labeled. (D) Kinetics of the 3rd-dose booster elicited recall response as indicated during monitoring of NAb titers at different time points. The green and blue curves show the changes in kinetics of NAb titers for pre-3rd-dose and post-3rd-dose vaccination, respectively. (E) Pie charts represent the distribution of antibody sequences from the four 3-dose vaccinees. The number in the inner circle is the number of sequences analyzed here. Pie-slice size is proportional to the number of clonally related sequences. The black outline indicates the frequency of clonally expanded sequences detected individually. Colored slices reveal clones that share the same IGHV and IGLV genes. (F) Graph shows relative clonality among seven individuals who received 2-dose or 3-dose of inactivated vaccines. Relative clonality for COVID-19 convalescents assayed at 1.3, 6.2, and 12 months after infection, as well as 2-dose mRNA vaccine recipients (Wang et al., 2021f, 2021g), previously described by Michel’s group, was compared. Black horizontal bars indicate mean values. Statistical significance was determined using two-tailed t-test. (G) Number of somatic nucleotide mutations in the IGHV (left) and IGLV (right) in antibodies from vaccinees, including 2-dose or 3-dose of inactivated vaccines and 2-dose of mRNA vaccines and COVID-19 convalescents assayed at 1.3, 6.2 and 12 months after infection (Wang et al., 2021f, 2021g). (H) Normalized ELISA binding (EC50) by antibodies isolated from the 3-dose inactivated and 2-dose mRNA vaccinees (ref) as well as COVID-19 convalescents to SARS-CoV-2 S trimer (left) and normalized pseudovirus neutralization activity (IC50) (right) against SARS-CoV-2 assayed at 1.3, 6.2, and 12 months after infection (ref). Among these, eight antibodies reported by Michel’s group were expressed and assessed for both binding by ELISA and pseudovirus neutralization activity for normalized comparison here. Black horizontal bars indicate mean values. (I) Pseudo-typed virus neutralization by antibodies isolated from the 3-dose vaccinees to circulating SARS-CoV-2 variants. Color gradient for bottom panel indicates IC50 values ranging from 0 (green), through 20 (yellow) and 200 (red) to 2,000 ng/mL (purple). Gray suggests no/very limited neutralizing activity (>2,000 ng/mL). To further characterize the expeditiousness, longevity, and immunological kinetics of recall response stimulated by the third-dose immunization, neutralizing potencies at days 0, 7, 14, 28, 90, and 180 post the third-dose vaccination were determined (Fig. 1C and 1D). Remarkably, NAb titer surged by ~8-fold (from 7 to 53) at week 1, and peaked by ~25-fold increase (up to 177) at week 2 after the 3rd-booster and slowly decreased over time (Fig. 1C). Notably, NAb titer was maintained at around 60 on 180 days post the 3rd-booster, comparable to the high level of NAb titer elicited by the 2-dose administration (Fig. 1D). Taken together, these serological results reveal a third-dose booster can elicit an expeditious, robust and long-lasting recall humoral response. We used flow cytometry to sort the SARS-CoV-2 S-trimer-specific memory B cells from the blood of seven selected CoronaVac vaccinees (Figs. S1–2). The gated double-positive cells were single-cell sorted and immunoglobulin heavy (IGH; IgG isotype) and light (IGL or IGK) chain genes were amplified by nested PCR. Overall, we obtained 422 and 132 paired heavy and light chain variable regions from S-binding IgG+ memory B cells from four 3-dose and three 2-dose vaccinees, respectively (Figs. 1E and S2). Surprisingly, expanded clones of cells comprised 45%–61% of the overall S-binding memory B compartment in 3-dose vaccinees, which is approximately 2-fold higher than those in COVID-19 convalescents and in mRNA or 2-dose vaccinated individuals, revealing an ongoing clonal evolution (Fig. 1E and 1F). Shared antibodies with the same combination of IGHV and IGLV genes in 3-dose vaccinees comprised ~20% of all the clonal sequences. Similar to natural infection and mRNA vaccination (Wang et al., 2021g), IGHV3-30, IGHV3-53, and IGHV1-69 remained significantly over-represented in 3-dose vaccinees (Fig. S3). Additionally, the number of nucleotide mutations in the V gene in 3-dose vaccinees is higher than those in both 2-dose vaccinees and naturally infected individuals assayed after 1.3 and 6.2 months, but slightly lower than those in convalescent individuals 1 year after infection (Fig. 1G), revealing ongoing somatic hypermutation of antibody genes. To further explore the immunogenic characteristics of the antibodies obtained from memory B cells in 3-dose vaccinees, 48 clonal antibodies, designated as XGv01 to XGv50 (no expression for XGv37 and XGv48) were expressed and their antigen binding abilities verified by ELISA (Fig. S4). Biolayer interferometry affinities (BLI) measurements demonstrated that all antibodies bound to WT SARS-CoV-2 at sub-nM levels (Table S1). The normalized geometric mean ELISA half-maximal concentration (EC50) revealed that all antibodies (EC50 = 4.5 ng/mL) obtained from 3-dose vaccinees, in particular RBD-specific mAbs (EC50 = 3.5 ng/mL), possessed higher binding activities than RBD-mAbs from early convalescents (at 1.3 and 6.2 months after infection, EC50 = 5.0 and 6.8 ng/mL, respectively) and mRNA (EC50 = 4.4 ng/mL) vaccinated individuals, but were comparable to those from late convalescent individuals (EC50 = 2.6 ng/mL) assessed at 12 months after infection (Fig. 1H). These results indicate the possibility of the loss of antibodies with low binding affinities over time or an ongoing increase in affinity under repeated exposures to antigen. Among these antibodies tested, 26 were bound to RBD, 16 targeted NTD, and 6 interacted with neither RBD nor NTD, but bound S1 (S1/non-RBD-NTD) (Table S1). Pseudovirus neutralization assay revealed that all RBD-specific antibodies, 10 (~60%) of the 16 NTD-directed antibodies and 3 (~50%) of the 6 S1/non-RBD-NTD antibodies were neutralizing, presenting a relatively high ratio for NAbs (Figs. 1I, S5 and Table S2). Authentic SARS-CoV-2 neutralization assay results largely verified their neutralizing activities, albeit higher concentrations were required for some NAbs (Fig. S6). In line with binding affinity, the normalized geometric mean IC50 of the RBD antibodies of 3-dose vaccinees was 80 ng/mL, substantially lower than those from naturally infected individuals (ranging from 1.3 to 6.2 months, IC50 = 130–160 ng/mL) and mRNA vaccinated individuals (IC50 = 150 ng/mL), but similar to those from late convalescents (IC50 = 78 ng/mL) (Fig. 1H). The overall increased neutralizing potency might have resulted from the ongoing accumulation of clones expressing antibodies with tight binding. RBD is one of the main targets of neutralization in SARS-CoV-2. RBD exists in either an “open” or “closed” configuration (Walls et al., 2020), bearing antigenic sites with distinct “neutralizing sensitivity.” To dissect the nature of the epitopes of RBD targeted by NAbs, 171 SARS-CoV-2 RBD-targeting NAbs with available structures (Lv et al., 2020; Zhou et al., 2021), including cryo-EM structures determined in this manuscript (Figs. S7–8 and Table S3), were examined. By using cluster analysis on epitope structures, the antibodies were primarily classified into six sites (Ⅰ, Ⅱ, Ⅲ, Ⅳ, Ⅴ, and Ⅵ) (Figs. 2A and S9). Additionally, we superimposed structures of RBDs from these complex structures and calculated the clash areas between any 2 NAbs (Fig. 2B). Both strategies yielded identical results. Combining the results of the characterization of binding and neutralization studies reported previously with those determined here, the key structure-activity correlates for the six classes of antibodies were analyzed (Fig. 2). Antibodies with sites Ⅰ, Ⅱ, and Ⅲ, most frequently elicited by SARS-CoV-2 early infection, target the receptor-binding motif (RBM), and potently neutralize the virus by blocking the interactions between SARS-CoV-2 and ACE2 (Fig. 2C and 2D). Class I antibodies, mostly derived from IGHV3-53/IGHV3-66 with short HCDR3s (generally < 15 residues), recognize only the “open” RBD, and make major contacts with K417 and N501 (Figs. 1F, 1G, and S9–10). Approximately ~75% and 60% of class I NAbs were significantly impaired in binding and neutralizing activities against B.1.351 as well as P.1, respectively, due to the combined mutations of K417N/T and N501Y (Figs. 2D, S11 and S12). Contrarily, Class III antibodies bound to RBD either in “open” or “closed” conformation, extensively associated with E484, and partially with L452 (Figs. 2D and S10C). Disastrously, over 90% of class III antibodies showed a complete loss of activity against B.1.351 as well as P.1 largely owing to an E484K mutation (Fig. S12). Against B.1.617.2, the substantially decreased activity of ~half of the class III antibodies is presumably mediated by L452R (Fig. S12). Class II antibodies use more diverse VH-genes and target the patch lying between sites I and III (Figs. 2D and S13). As expected, the effects of mutations on the activity of class II antibodies were severe, two-thirds of these antibodies had >10-fold fall in neutralization activities against VOCs (Fig. S12). Overall, the above analysis reveals that the RBD mutations identified in several VOCs can significantly reduce and, in some cases, even abolish the binding and neutralization of classes I to III antibodies, albeit being the most potent neutralizing antibodies against WT SARS-CoV-2. Structural, immunogenic, and evolutionary features of RBD Nabs. (A) Structure-based antigenic clustering of SARS-CoV-2 RBD NAbs. A total of 171 RBD NAbs with available structures were classified into six clusters (Ⅰ, Ⅱ, Ⅲ, Ⅳ, Ⅴ, and Ⅵ). NAbs that can block ACE2 binding or not are outlined in light pink and light yellow, respectively. NAbs that can attach to the closed RBD or not are outlined by gray blue and gray green, respectively. (B) Superimposition matrix of 171 RBD NAb structures’ output from clashed areas (Å2) between variable regions of any two Fab fragments showing the clustering into six antibody classes. (C) Surface representative model of six types of NAbs bound to the RBD. Fab fragments of six representative antibodies are shown in different colors and the RBD is colored in gray. Insets illustrate the antigenic patches targeted by six representative antibodies. Dashed dots indicate the overlaps between two adjacent antigenic patches. (D) Structural landscapes of the six classes of RBD NAbs (upper panel). Antigenic patches (with targeting frequency > 30%) recognized by six classes of NAbs are outlined in the assigned color scheme (same as Fig. 2C), among which residues with “hot targeting frequency” (generally over 65%, but over 85% in class I) are shown in bright colors corresponding to the patches they belong to. Residues involved in two (such as Y489, L452) or three (such as F486) neighboring antigenic patches are presented in a mixed color. Representative “hot” antigenic residues are labeled. Middle: hot map for antigenic residues on the RBD. Per residue frequency recognized by the 171 NAbs were calculated and shown. The top 9 of the hottest antigenic residues and key residues with substitutions in several VOCs are and labeled. hot map for circulating variants with mutations on the RBD. frequency for each residue was calculated on the from (E) and the antibodies as the early time group (left) and late time group The antibodies are colored on their cluster by the clustering Antibodies from I to III and to are in and gray blue respectively. Pie charts represent the frequency distribution of antibodies to I to III and to Antibodies isolated from 3-dose vaccinees are outlined by black (F) of the antibodies from I to class antibodies are in colors corresponding to the classes they belong to. The color scheme is same as Fig. are shown in Fig. (G) The measured (left) and (right) of antibodies from I and II are shown. and antibodies to classes I and II are colored in and green, respectively. The curves and lines are the of the to the with the values of the for and antibodies. The black curves and lines indicate the of antibodies from I or the and green suggest the of and antibodies, respectively. The lines are the 90% By antibodies of the three classes recognize regions distinct from the and some of these are with et al., 2021). The binding of class antibodies, albeit to the of the is on a patch that residues which are not related to mutations observed in early VOCs (Figs. 2C and S9). Interestingly, class antibodies can their et al., Class antibodies, and the for high to most V is the to the and adjacent to the None of the class V antibodies with ACE2 binding (Figs. 2D and to targeting frequency to B.1.617.2, but not partially decreased the activities of some class V antibodies (Fig. S12). Class antibodies recognize a patch on one of the RBD, from the Among these, some with ACE2 some and this largely on the of the antibodies In of antibodies targeting sites V to are mostly to the To further the of we immunogenic and for RBD using the 171 NAb complex structures to in on the RBD and mutation (calculated from the in the respectively (Figs. 2D and S13). revealed that the epitope residues of sites I to III showed higher NAb and antibodies residue on for and respectively) compared with those of sites to and antibodies residues on for and that class I to III antibody epitopes are “hot” immunogenic sites (Figs. 2D and S13). In line with residues within sites I to III exhibited higher mutation as revealed in circulating variants that mutations of E484, and N501 residues (Figs. 2D and S13). Surprisingly, of the top 9 hottest immunogenic residues had a high mutation In such as Y489, with large exhibited low mutation in circulating SARS-CoV-2 (Figs. 2D and is that all these residues are extensively involved in the of and immunogenic analysis mutations at these not be To changes in the frequency of distribution of the six types of RBD antibodies are associated with evolution we and SARS-CoV-2 NAbs from available antibody we combined and matrix analysis for RBD NAbs in total (Figs. and In the studies 2020), NAbs to classes I to III were identified in early COVID-19 convalescent and 2-dose vaccinated individuals as early time for to of total antibodies. By a low ratio of NAbs from to was reported due to their potent activities at the early time (Fig. In a of 2020), NAbs with enhanced neutralizing potency and breadth from to have substantially been in the late convalescents or 3-dose vaccinees, in frequency to antibodies from I to III and further in individuals with 3 doses of inactivated vaccine (Fig. These results suggest that memory B cells clonal after about 6 months, resulting in changes in the of antibodies in B and partially to enhanced activities of antibodies in the plasma over To explore the we measured the binding affinities of antibodies that are also further into early and late time (Table S1). the late time group, there was a fold increase in binding affinity for compared to those in the early time group (Fig. In the early time group, antibodies from to exhibited higher binding affinities to the RBD than those from I to in antibodies from V and limited (Fig. most antibodies from V and with low affinities and activities might be in the early time In the late group, sub-nM binding affinities for class antibodies with distinct were ongoing affinity over time (Fig. S12). antibody clustering and V gene analysis suggests that class antibodies can be derived from V genes and the V gene antibodies belong to different classes. To the in the between binding affinity and somatic hypermutation we determined the relative affinity and calculated the of antibodies that are by the same V gene and belong to the same The measured and of class I antibodies = including NAbs derived from IGHV3-53, show of to a with a of for antibodies for all class I (Fig. The of corresponding to a of is = and = Antibodies with from II and III exhibited similar by a among which antibodies in class II yielded a of six involved in the (Fig. These indicate that as the the binding and As of the of this the B.1.617.2 variant had to in COVID-19 of among vaccinated were observed when variants and including and such as and et al., et al., results that a third-dose booster of inactivated vaccine can elicit an expeditious, and long-lasting recall humoral response which to with ongoing accumulation of somatic emergence of and affinities of antibodies to enhanced neutralizing and the emergence of which have been to largely over the immune responses by repeated of all types of vaccines have been However, the of variant booster dose has shown a immune response to the infection and has been by the of booster dose against SARS-CoV-2 et al., similar to our observed antibody after the booster studies have that the against the SARS-CoV-2 variants by the booster dose is largely due to the somatic hypermutation and affinity of memory B antibodies et al., et al., our findings the use of 3-dose vaccination into the of booster the of SARS-CoV-2 the We and for cryo-EM the for in the of for and for with We also and for the of the sequences of two have was by the of and of and and the and was by for and the and and the and and antibodies. and expressed and all antigen used in this and pseudovirus and authentic virus neutralization and study. and and flow cytometry and and and recruited volunteers and the of blood analyzed the manuscript with from all of the SARS-CoV-2 S trimer in complex with or the SARS-CoV-2 S trimer in complex with and the SARS-CoV-2 S trimer in complex with and and the SARS-CoV-2 S trimer in complex with and have been at the with and respectively.

Topics & Concepts