Prior SARS-Cov-2 infection affects adaptive immune responses to Omicron BA.4/BA.5 mRNA booster

Abstract

Background

Bivalent COVID-19 mRNA vaccines encoding Wuhan-1 and Omicron BA.4/5 spike proteins can prevent SARS-CoV-2 infection, but the quality of adaptive immune responses and the importance of hybrid immunity are not well-documented.

Objectives

Adaptive immune responses to the bivalent vaccine were studied in 40 healthy participants with (COVID+) or without (COVID-) prior history of SARS-CoV-2 infection.

Methods

We analyzed anti-N and anti-S IgG titers and surrogate virus neutralization capacity against variants of concern (VOCs) and assessed SARS-CoV-2 specific B and T cell responses by high-dimensional spectral flow cytometry, intracellular cytokine staining assay upon stimulation with SARS-CoV-2 peptides and TRB and IGH repertoire analysis.

Results

The COVID+ group had higher anti-S IgG levels pre- and post-booster and higher neutralization activity against BA.4/5 than the COVID− group. Spike antibody levels positively correlated with neutralizing activity against Omicron VOCs in all participants. For VOCs, lowest neutralization capacity was against XBB1.5. At baseline, the proportion of S1+ RBD+ B cells was higher in COVID+ than in COVID− subjects, but an increase of these cells post-boost was detected only in the COVID− group. Consistent with natural infection, COVID+ subjects had a higher frequency of IgA+ CXCR3+S1+RBD+ B cells at baseline than COVID− subjects. CD4+ memory T cells responses and breath of class II epitope SARS-CoV-2 specific clonotypes were increased post-boost only in COVID− participants.

Conclusions

The bivalent vaccine induces robust adaptive immune responses against the Omicron variant. Prior SARS-CoV-2 infection provides increased protection, but optimal timing of booster administration after natural infection should be defined to maximize benefits.

Graphical Abstract

Graphical abstract showing study design, timeline, methods and results.

Capsule summary

In those vaccinated with the original COVID-19 mRNA vaccine, hybrid immunity offers superior protection against SARS-CoV-2 variants; boosters should not be given too soon after natural infection to maximize bivalent vaccine benefits.

Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) mRNA vaccines generate efficient humoral and cellular immune responses that protect against symptomatic COVID-19 infection (1, 2). Two-doses of the original mRNA-1273 (Moderna) or BNT162b2 (Pfizer-BioNTech) elicit anti-spike antibody titers that neutralize the vaccine -homologous virus (3, 4). Booster doses of the same vaccine enhance the magnitude and the durability of the neutralizing antibody response and induce strong memory T and B cell responses against some variants of concern (VOCs) (5). Hybrid immunity, resulting from prior infection followed by vaccination, offers some enhanced protection against VOCs (1, 6). Altarawneh et al. (7) showed that hybrid immunity from infection and three doses of the original vaccine prevented Omicron BA.1 and BA.2 infections in 56% of cases. Individuals recovering from infection prior to immunization benefit from imprinted effector CD8⁺ T cells with improved antiviral properties and imprinted memory CD4⁺ T cell that enhance SARS-CoV-2 specific antibody responses (8, 9).

Furthermore, vaccine doses number and and the variant causing infection influence immune response to boosters (10). However, multiple antigen stimulations may impair immune responses due to antibody-mediated feedback (11, 12). Occurrence of natural infection close to booster vaccination can attenuate post-booster B-cell immune responses (13). Immune escape depends on neutralizing antibody specificity, necessitating updated boosters for population-level protection (14, 15).

Limited data exist on T-cell responses against SARS-CoV-2 variants after multiple antigen exposures, including hybrid immunity. T-cell responses remain largely preserved across variants (16–18). Given the vast number of SARS-CoV-2 epitopes, significant mutation without losing viral fitness is difficult. Hybrid immunity broadens T-cell epitope recognition, particularly for non-spike proteins (2). However, repeated exposure to SARS-CoV-2 antigens from multiple doses of the same vaccine may lead to more limited subsequent boosting of T-cell responses (19–22).

The emergence of antigenically divergent Omicron variants in 2022 required updated booster strategies (23–25), leading to the authorization of bivalent Omicron BA.4/BA.5 boosters for fall 2022 (US FDA). These boosters reduced Omicron-related morbidity and mortality, though their effects on cellular and humoral immunity remain under investigation.

To investigate this aspect and to analyze the impact of prior SARS-CoV-2 infection history, we studied antibody production, and B and T cell responses to the Omicron BA.4/BA.5-containing bivalent BNT162b2 and mRNA-1273.222 vaccines in a cohort of SARS-CoV-2 naïve and recovered individuals who had previously received at least 3 doses of the original monovalent SARS-CoV-2 mRNA vaccine.

Methods

Participants recruitment

Peripheral blood samples were obtained from 40 healthy volunteers (samples, n=80), before and 3 weeks after administration of Omicron BA.4/5 bivalent mRNA vaccine. All subjects and their guardians provide written informed consent in accordance with the Declaration of Helsinki, and with protocols approved by local ethical committees (Protocol CORONAVax NP4590, approved by Comitato Etico Provinciale, Italy or NIH institutional review board (IRB) approved protocols NCT03394053, and NCT03610802). Plasma was separated and DNA was extracted from whole blood samples. Information regarding age and sex was collected for all subjects (Table 1). Participants with anti-Nucleocapsid or anti-ORF8 serum antibodies and/or a clinical history of PCR/antigen confirmed infection were classified as COVID+ (vaccinated and recovered from SARS-CoV-2 infection), while those with no evidence of infection were in the COVID− group (vaccinated only).

Table 1:

Cohort Demographics and Breakthrough

Characteristics	COVID +	COVID −
Donors, n	25	15
Sex, n
Male	8	4
Female	17	11
Age, years (mean ± SD)	45 ± 16	44 ± 15
Bivalent Vaccine Type, n
Pfizer	18	11
Moderna	7	4
Time since last vaccination, months (mean ± SD)	11 ± 3	10 ± 2
Time since natural infection (mean and range)	8.4 [3–12]*
Breakthrough within 6 mo post bivalent booster	2	4

^*In 23 participants, unknown in 2 participants with anti-ORF8 only positive serology indicating distal natural infection.

Measurement of SARS-CoV-2 specific anti-S and anti-N antibodies.

Plasma was separated from peripheral blood samples anticoagulated with EDTA that were collected pre-immunization, and 3 weeks after Omicron BA.4/5 bivalent mRNA vaccine (Pfizer-BioNTech or Moderna).

Testing for anti-spike and anti-nucleocapsid IgG was performed via luciferase immunoprecipitation systems assay (LIPS) that employs light-emitting luciferase-antigen fusion proteins of SARS-CoV-2 as described previously (26, 27) and in the Online Repository.

Seroreactivity of samples to SARS-CoV-2 ORF8 by ELISA

96 well Immulon plates (Thomas Scientific, Cat #3455) were coated with 20 ng/100 μL of recombinant ORF8 from WA1/2020 (recombinant protein was produced in house in the lab) in PBS overnight at 4°C. Starting at a 1:20 dilution, serum samples were serially diluted 5-fold and applied to the coated well for 1 hr at ambient temperature. Serum samples were assayed in duplicate, as described before (28, 29) and in the Online Repository.

SARS-CoV-2 Neutralization assays

Samples were evaluated in a qualified SARS-CoV-2 pseudovirion neutralization assay (PsVNA) using SARS-CoV-2 WA1/2020 strain and Omicron BA.4/5, BQ.1, XBB.1.5 Omicron subvariants. SARS-CoV-2 PsV’s were produced in-house as described in previous publications (29–31). Qualified means that the PsVNA assay meet specific criteria for assay qualification including, reproducibility, robustness, precision, accuracy, linearity etc. performed in in house in the lab. SARS- CoV-2 neutralizing activity measured by PsVNA correlates with PRNT (plaque reduction neutralization test with authentic SARS-CoV-2 virus) in previous studies (29–31). Neutralization assays were performed as previously described (29, 32) and in the Online Repository.

SARS-CoV-2 specific B cell characterization with flow cytometry

5 million cells per sample of PBMC were resuspended in PBS with 2% FBS and 2 mM EDTA (FACS buffer). Biotinylated probes to SARS-CoV-2 were crosslinked with fluorochrome-conjugated streptavidin in a molar ratio of 4:1. Fluorochrome-conjugated streptavidin was split into 5 aliquots and conjugated to biotinylated probes by mixing for 20 min/aliquot at 4°C. Four fluorochrome-conjugated probes were used in this assay, including S1 from the Wuhan strain (Biolegend) conjugated to APC, RBD from the Wuhan strain (Biolegend) conjugated to BV421, RBD from the Omicron variant (Acro) conjugated to BUV615, and RBD from the omicron variant (Acro) conjugated to PE. Cells were stained as described in the Online Repository.

Unsupervised analysis of high dimensional flow cytometry data

S1⁺RBD⁺ B cells from the SARS-CoV-2 antigen specific B cell panel with 29 parameters were analyzed with unsupervised clustering of surface antibody staining. S1⁺RBD⁺ live CD45⁺CD3⁻ CD14⁻CD19⁺ B cells from COVID- and COVID+ groups were analyzed. B cell analysis was based on surface expression of multiple surface markers as described in the Online Repository.

Intracellular cytokine staining of activated T cells

T cell stimulation and intracellular cytokine staining were performed similarly to previously described methods (33) (see Online Repository).

TCR and BCR repertoire analysis

Sequences of TCR-β and IGH CDR3 chains present in magnetic beads sorted CD4⁺, CD8⁺, and CD19⁺ lymphocytes (Miltenyi Biotec, MicroBeads human Lyophilized) derived from PBMC samples were studied using HTS through the ImmunoSEQ assay (see Online Repository). All data are deposited at https://zenodo.org/records/13835339, whereas the R scripts used in generating the uploaded files are under https://github.com/cihangenome/COVID-booster. The DOI is 10.5281/zenodo.13835338 for the Zenodo repository.

Results

Cohort description

We enrolled 40 study participants receiving either the Pfizer (N=29) or Moderna (N=11) SARS-CoV-2 bivalent booster vaccine encoding for both ancestral Wuhan-1 original spike protein (WT) and the BA.4/BA.5 Omicron variant spike proteins (Table 1). To investigate the induction of antigen-specific immune responses by the bivalent vaccine, paired plasma and peripheral blood mononuclear cell (PBMC) were collected from 20 individuals and only plasma samples from 20 additional individuals at 2 different timepoints: pre-vaccination (baseline) and 3 weeks post-booster vaccination. Subjects were divided into 2 groups based on prior SARS-CoV-2 infection. All participants with a positive history of either, 1) SARS-CoV-2 PCR; or, 2) anti-nucleocapsid (anti-N) IgG titers >135.000 Light Unit (LU); or, 3) anti-ORF8 IgG titers with AUC>200 were defined as SARS-CoV-2 recovered (COVID+, n=25). Among these participants, three had asymptomatic infection, as they were positive for anti-ORF8 IgG only, with negative anti-N specific IgG. All other COVID+ subjects had developed symptomatic natural infection with positive PCR or COVID antigen testing between 3 to 12 months prior to bivalent booster vaccination (mean 8.4 months). Participants with no prior history of SARS-CoV-2 infection and negative serologies for anti-N and anti-ORF-8 IgG were considered SARS-CoV-2 naïve (COVID-, n=15). At the time of the study, all participants had received 3 to 4 doses of the Pfizer BNT162b2 or Moderna mRNA-1273 SARS-CoV-2 vaccines with the last dose administered between 8 and 14 months prior to the bivalent booster dose. Age and sex were balanced in both groups (Table 1).

Antibody responses to Omicron BA.4/5 bivalent mRNA vaccine

We first sought to measure anti-spike antibody responses to Omicron BA.4/5 bivalent mRNA vaccine in plasma samples. Administration of the bivalent vaccine induced robust circulating antibody responses to the SARS-CoV-2 spike protein in both SARS-CoV-2–naïve (COVID-) individuals (p≤ 0.0001) and recovered (COVID+) individuals (p≤ 0.0001) (Fig. 1A left panel, Fig. S1A). The COVID+ vaccinees had higher anti-S IgG levels at baseline than the COVID-vaccinees; however, post-booster levels were not significantly different between the 2 groups (Fig. 1A left panel and Fig. S1A, top panel). The fold-change of anti-spike IgG was significantly higher in the COVID− than in the COVID+ (p ≤ 0.01) (Fig. 1A right panel). As expected, no significant changes in anti-Nucleocapsid antibody levels were detected in plasma samples after bivalent booster (Fig. S1A, bottom panel).

Figure 1: Anti-Spike IgG titers and neutralization after.

Omicron BA.4/5 bivalent mRNA vaccine

(A) Anti-Spike Ig-G levels expressed in Light units in plasma samples from vaccinated subjects (Left panel). See also Figure S1A. Fold change over baseline of anti-Spike IgG levels in COVID+ (n=25) and COVID− subjects (n=15) (Right panel).

(B) Pseudovirus neutralization titers pre and post boost against BQ.1, BA.4/BA.5, or XBB.1.5 SARS-CoV-2 variants Spike proteins in plasma samples from COVID+ (n=25) and COVID− (n=15) subjects.

See also Figure S1 (B-C) neutralization titers against BA.1, Alpha, or Delta variants.

(C) Comparison of neutralization activity post booster in all subjects against BQ.1, BA.4/BA.5, or XBB.1.5 variants. See also Figure S1D showing neutralization activity post booster in COVID− (n=15) and COVID+ (n=25) groups separated.

(D) Correlation between post booster spike antibody levels and neutralization activity against the BQ.1, BA.4/BA.5, or XBB.1.5 variants.

Horizontal lines represent the mean (A), (B), (C). Whiskers represent standard deviation (A)(B), (C). Statistical analysis by Wilcoxon matched-pairs signed rank test or Mann-Whitney test (A), (B), (C) or Spearman rank order correlation (D). *p ≤ 0.05, **p ≤ 0.01, ***p≤ 0.001, ****p < 0.0001. Not significant, ns > 0.05.

To examine the functional quality of circulating antibodies, we used neutralization assays. Initial screening of samples was performed by competition assay at 10-fold dilution by ELISA for alpha (B.1.1.7), delta (B.1.617) or the Omicron BA.1 variant (Fig. S1B and S1C) while for the more recent circulating strains we used pseudotyped virus expressing BA.4/BA.5, BQ.1, XBB.1.5 Spike proteins of Omicron VOC to determine 50% neutralization (PsVNA50) titers using the Pseudovirus neutralization assay, (Fig. 1B). For all variants, pre-boost neutralizing activity was higher in the COVID+ than in the COVID− group (Fig. 1B, Fig. S1B). For the alpha and delta variants, the post-boost increase in percentage of binding capacity was significant only in the COVID− vaccinees but not in the COVID+ group, since the latter had already maximized neutralizing capacity resulting from infection plus 3 or 4 doses of prototype mRNA vaccine prior to the bivalent boost (Fig. S1B and S1C). In both COVID- and COVID + groups, the bivalent vaccine strongly boosted the neutralizing capacity to the BA.1 (Fig. S1B and S1C), BQ.1, BA.4/BA.5 and XBB.1.5 (Fig.1B) Omicron variants. No significant post-boost difference between COVID- and + vaccinees was detected for neutralizing capacity to BA.1, BQ.1, XBB.1.5 (Fig. 1B and Fig. S1B). Only for the BA.4/BA.5 variants the COVID+ group had higher post boost neutralizing titers compared with the COVID− individuals (Fig. 1B), as expected given that the vaccine strain was matched to the antigen likely encountered given the timing of infection. Next, to investigate the relative neutralizing activity of the bivalent vaccine against Omicron sublineage VOC, we compared post-boost neutralizing titers against the BA.4/BA.5, BQ.1, and XBB.1.5 variants. Higher neutralizing titers were detected against the BA.4/BA.5 compared with BQ.1 and XBB.1.5 variants, with the latter being associated with the lowest neutralizing activity (Fig.1C). A similar pattern was observed when splitting COVID- and COVID+ participants (Fig. S1D). These findings are consistent with the high rates of breakthrough infection due to the XBB1.5 variant, even after bivalent booster dose (34). Finally, the fold-change in neutralization capacity against VOCs at 3 weeks post-boost over baseline was significantly higher (p ≤ 0.01) in COVID− than COVID+ only for BA.4/BA.5 but not BQ.1 or XBB1.5 (Fig. S1E).

We then analyzed the correlation between post-booster anti-spike IgG levels and neutralization activity against the BQ.1, BA.4/BA.5, and XBB.1.5 Omicron VOC. Irrespective of COVID status, a positive correlation was observed between anti-S IgG levels and neutralizing capacity for all variants (Fig. 1D). However, the correlation index and the significance values were higher in the COVID+ vaccinees. These data indicate that both booster immunization with bivalent mRNA vaccines and hybrid immunity generate a functional humoral response against multiple Omicron sublineage VOC, despite variable levels of neutralization activity to individual variants.

Memory B Cell Responses to Omicron BA.4/5 bivalent mRNA vaccine

Next, we used high dimensional spectral flow cytometry to determine the frequencies of SARS-CoV-2 S1 domain and receptor-binding domain (RBD)-specific (S1+ RBD+) total circulating B cells and switched memory B cells (smB, defined as CD27⁺IgD⁻ excluding CD38^hiCD27^hi plasmablast and plasma cells) specific for the Wuhan wild-type (WT) and the B.A.1 Omicron variant. To this purpose, we analyzed the peripheral B cell compartment of 20 vaccinees, after staining PBMC with fluorescently labelled probes for RBD and S1 portion of the spike protein from the WT strain and the BA.1 variant (Fig. 2A).

Figure 2: SARS-CoV-2 specific B cell characterization with spectral flow cytometry.

(A) Representative gating strategy to identify S1+RBD+ CD19+ or IgD- CD27+ switched memory B cells specific for wild type (WT) or Omicron variant in PBMC.

(B) Percentage of S1⁺RBD⁺ cells among CD19⁺ B cells and IgD- CD27+ switched memory B cells against WT or Omicron variant pre and post boost in COVID+ (n=11) and COVID− subjects (n=9).

(C) Correlation between Omicron specific post boost switched memory B cells and post boost neutralizing activity for the BQ.1, BA.4/BA.5, or XBB.1.5 variants. See also Figure S2 A–C the same correlation comparing COVID− (n=9) and COVID+ (n=11) for the BQ.1, BA.4/BA.5, or XBB.1.5 variants.

(D) Frequency of CD27+CD21-CD11c+ B cells specific for Omicron or WT variant in PBMC of 20 vaccinated subjects (COVID+ n=9; COVID− n=11) whose samples were obtained both pre and post boost.

(E) Frequency of IgA+ switched memory B cells specific for either Omicron or WT variants in PBMC of 20 vaccinated subjects (COVID+ n=11; COVID− n=9) whose samples were obtained both pre and post boost.

Horizontal lines represent the mean in (B), (E) and whiskers the standard deviation (B). Statistical analysis by Wilcoxon matched-pairs signed rank test or Mann-Whitney test (B), (D), (E) or Spearman rank order correlation (C). *p ≤ 0.05, **p ≤ 0.01, ***p≤ 0.001, ****p < 0.0001. Not significant, ns > 0.05

S1⁺RBD⁺ CD19⁺ B cells and S1⁺RBD⁺ smB recognizing RBD from WT and BA.1 were detected in all individuals at baseline. Prior to boost administration, COVID-naive subjects had a lower frequency of S1⁺ RBD⁺ smB cells reactive against the WT virus compared to COVID+ individuals, and a similar trend was observed also for smB cells reactive to the Omicron variant (Fig. 2B). The population of total B cells and smB cells specific for the WT and for the BA.1 variant was significantly boosted by the bivalent vaccine only in the COVID− group but not in the COVID+ group (Fig. 2B). A positive correlation between the frequency of S1+ RBD+ smB cells and neutralization titers post-boost was observed in the entire cohort of 20 subjects in which this analysis was performed (Fig. 2C). However, a positive correlation between the frequency of S1+ RBD+ smB cells and neutralization titers post-boost was observed only in the COVID-group of subjects (Fig. S2A–C).

To further investigate differences in the S1+ RBD+ B cell response in COVID-naïve versus COVID+ recovered individuals, we analyzed antigen-specific smB cells pre- and post-boost by assessing markers that are tipically increased during SARS-CoV-2 acute infection including CD69, CD62L, FCRL3/FCRL5, CD86, CD71, and CD11c surface expression. A significant increase of activated memory B cell populations was detected post- compared to pre-boost in both groups (Fig. S3A). In particular, a sharp increase in the frequency of CD27⁺ CD21^low CD11c⁺ B cells was evident 3 weeks post-immunization irrespective of COVID status (Fig. 2D). This increase was observed in all subjects except a single COVID-negative individual, who had a high frequency of atypical B cells already at baseline, suggestive of a possible underlying immune dysregulation and/or heighten response to previous SARS-CoV-2 vaccinations. The inhibitory surface co-receptor molecule FCRL3/FCRL5 was also increased post-boost in both groups (Fig S3A).

Then, we interrogated S1⁺RBD⁺ smB cells for the expression of surface immunoglobulin isotypes. The vast majority of the cells were IgG+, with fewer cells being IgA+, and virtual absence of IgM+ smB cells (Fig. S3B). While IgM+ and IgA+ smB cells are known to represent a minor fraction of the overall response in the blood, we identified a significant relative increase in IgA+ smB cells in the COVID+ versus the COVID− group both pre and post boost for WT and BA.1 variants (Fig. 2E, Fig. S3B).

To further compare the phenotypes of S1⁺RBD⁺ B cells between COVID naïve and recovered vaccinees pre- and post-boost, we concatenated the S1⁺RBD⁺ B cells for all variants and performed unsupervised clustering analysis based on the expression of 19 different B cell surface markers. Unsupervised clustering of circulating S1⁺RBD⁺ B cells from PBMCs generated 15 phenotypically distinct clusters, 12 of which represented IgD⁻ smB cells, one cluster (cluster 11) containing both IgM⁺IgD⁺CD27⁻ naïve and IgM⁺ IgD⁺CD27⁺ unswitched memory B cells, one cluster (cluster 7) of IgD⁻ CD27⁻ B cells, and one cluster (cluster 13) representing IgA+ plasmablasts (Fig. 3A–D). We then looked at the effect of COVID exposure history and of pre-versus post-booster status on the proportion of each cluster estimated under a linear model controlling for age and sex (Fig. 3C, 3E, 3F). COVID+ subjects had higher frequencies of S1⁺RBD⁺ B cells pre-boost in cluster 10 (representing IgA⁺ CXCR3⁺ memory B cells) than COVID naïve subjects (Fig. 3C, 3F). By contrast, post-boost COVID− vaccinees had a significantly higher proportion of S1⁺RBD⁺ B cells in cluster 3 than COVID+ vaccinees. This cluster includes both CD21⁻CD27⁺ activated memory B cells and CD21⁺CD27⁺ classical memory B cells. However, cells belonging to this cluster were CXCR3⁻ CD62L^low (Fig. 3B, 3D), consistent with absent signaling to home to secondary lymphoid organs and tissues. We used a linear mixed model to estimate the impact of COVID exposure or boost on the relative frequency of cluster representation. Booster vaccine administration had a clear impact on the frequency of cells in cluster 1, 2, 5, 6, and 8 (Fig. 3E, 3F, 3G, Fig. S3C). The relative frequency of cells in clusters 1, 2 and 5 (both representing subsets of CD21⁺ memory B cells) significantly decreased post-boost in the COVID− group (cluster 2 and 5) or in both COVID groups (cluster 1) (Fig. 3F, Fig. S3C), consistent with contraction of the resting memory B cell compartment after immunization. By contrast, the relative proportion of cells in clusters 6 and 8 (representing memory B cells with lower expression of CD21 and higher surface expression of the markers CD62L (cluster 6) and CD11c and FCRL5 (cluster 8), increased post-boost irrespective of COVID exposure (Fig. 3F, Fig. S3C). This mixed model analysis also confirmed that cluster 10 was more represented in COVID+ vaccinees compared to naïve subjects, especially pre-boost (Fig. 3C, 3F–G).

Figure 3: Unsupervised analysis of S1+RBD+ B cells in peripheral blood.

(A) UMAP projection of unsupervised clustering based on surface markers from flow cytometric analysis of S1+RBD+ CD19+ B cells to define subpopulations B cell clusters. Pre and post boost PBMC samples from 20 vaccinated subjects (COVID+ n=11; COVID− n=9) were merged and processed together.

(B) Heatmap of marker expression in each cluster.

(C) Linear model comparing the result of infection status on the proportion of clusters in PBMC of 20 vaccinated subjects (COVID+ n=11; COVID− n=9) whose samples were obtained both pre and post boost. Model controls for age and sex of vaccinees.

(D) UMAP overlay of cluster localization by marker.

(E) UMAP overlay of proportion of clusters in both groups of vaccinees (COVID+ n=11; COVID− n=9) pre and post boost.

(F) Proportion of each cluster in each group (COVID+ n=11; COVID− n=9) evaluated in the same patients pre and post boost.

(G) Significant variations between cluster frequencies as a result of booster or infection status. Comparisons of both groups (COVID+ n=11; COVID− n=9) along with regression coefficients ± 95% confidence interval (CI) and p values are shown. Statistically significant clusters (p<0.05) are in red. Samples with fewer than 25 S1+RBD+ B cells were excluded from the analysis.

To further characterize SARS-CoV-2 specific B cell responses, we performed high-throughput sequencing (HTS) of the IGH repertoire of CD19⁺ B cells. Upon searching for IGH-CDR3 sequences of the vaccinees that share at least 80% similarity (based on CD-HIT) (35) with published SARS-CoV-2 specific BCR sequences in the COVID AbDab database (35, 36) to detect SARS-CoV-2-specific hits, we identified 110 SARS-CoV-2 specific IGVH clonotype clusters in the COVID-recovered vaccinees, and 105 clusters in the COVID-naïve group (Suppl. Table 1). No statistically significant differences were observed with respect to the breadth and summed frequency of the SARS-CoV-2 specific B cell clonotypes pre- and post-boost between COVID naïve and recovered vaccinees (Fig. S4A, S4B). Similarly, the rate of somatic hypermutation (SHM) among SARS-COV-2 spike-specific IGVH clonotypes did not show a statistically significant difference in the COVID+ group compared to the COVID− (Fig. S4C).

Memory T Cell Responses to Omicron BA.4/5 bivalent mRNA vaccine

Memory T cells generated after immunization or natural infection can also contribute to protection against SARS-CoV-2 infection (2), and are less influenced by VOC than humoral immune responses (16, 37). To determine whether mRNA vaccination induced antigen-specific memory T cell responses, we stimulated PBMC in vitro with peptide pools containing spike epitopes for both WT and the Omicron BA.1.1 variant, followed by intracellular cytokine staining (ICS) (33). Antigen-specific responses were quantified as the frequency of cytokine producing memory CD4⁺ or CD8⁺ T cells in stimulated samples, with background subtraction from paired unstimulated controls. In particular, antigen-specific memory CD4⁺ T cells were defined based on the co-expression of CD154 and at least one intracellular cytokine among IFN-γ, TNF-α, and IL-2. Antigen-specific memory CD8⁺ T cells were defined based on the expression of CD69 and at least one intracellular cytokine among IFN-γ, TNF-α, and IL-2, as previously described (33) (Fig. S5A–C). Consistent with previous studies, we detected Th-1 biased responses in CD4⁺ T cells and low level CD8⁺ T cell responses (Fig. 4A–B). The frequency of antigen-specific memory CD4⁺ and CD8⁺ cells was similar in COVID- and COVID+ individuals pre-boost (Fig. 4A and 4B)(38). After boost, a significant increase in the frequency of antigen-specific memory CD4⁺ T cells was observed in the COVID-naïve group, but not among COVID-recovered vaccinees (Fig. 4A)(38). Taken together, these data indicate that mRNA bivalent boost vaccination generates SARS-CoV-2-specific CD4⁺ T cell memory responses in individuals who were not previously infected with SARS-CoV-2 but does not significantly amplify these responses in previously infected individuals.

Figure 4: Memory T cell responses following stimulation with SARS-CoV2 wild type or Omicron B.1.1.529/BA.4 & BA.5 spike peptides.

(A) CD4+ T cells: Memory CD4+ T cell responses (CD154+ IFNγ+ only or CD154+ IFNγ+/TNFα+/IL-2+) in 20 vaccinated subjects (COVID+ n=11; COVID− n=9) whose samples were obtained both pre and post bivalent boost following Omicron or WT SARS-CoV-2 peptide stimulation. See Figure S5 (A) for gating strategy. (B) CD8+ T cell: Memory CD8+ T cell responses (CD69+ IFNγ+ only or IFNγ+/TNFα+/IL-2+) in 20 vaccinated subjects (COVID+ n=11; COVID− n=9) whose samples were obtained both pre and post bivalent boost following Omicron or WT SARS-CoV-2 peptide stimulation. See Figure S5 (A) for gating strategy. Color scheme described in (A) applies to (B). Statistical analysis by Wilcoxon matched-pairs signed rank test or Mann-Whitney test (A), (B). *p ≤ 0.05, **p ≤ 0.01, ***p≤ 0.001, ****p > 0.0001. Not significant, ns > 0.05.

We next performed HTS of the T-cell repertoire on sorted total CD4⁺ and CD8⁺ T cells, with the aim of evaluating SARS-CoV-2 specific responses targeting Human Leukocyte Antigen (HLA) class I and class II epitopes pre- and post- boost in both COVID- and COVID+ vaccinees. To identify the SARS-CoV-2-specific signatures in the repertoire data, we utilized the criterion of exact matching between the TRB CDR3 sequences identified in vaccinees and the SARS-CoV-2-specific TRB CDR3 sequences reported in the ImmuneCODE database (39), that had been developed using the multiplex identification of antigen-specific T-cell receptors (MIRA) assay (40). We identified over 2000 unique clonotypes (with ≥2 copies per sample) targeting distinct HLA class-I and class II-restricted epitopes within the spike protein, and computed breadth and summed frequency (Suppl. Table 2). Breadth indicates the proportion of distinct SARS-CoV-2 spike-specific T-cell clonotypes that are present in a TCR repertoire sample, while summed frequency is the cumulative frequency of the SARS-CoV-2 spike-specific clonotypes in the sample. As expected, given the nature of the immunization stimulus, breadth and summed frequency of Class II, but not Class I, spike-specific responses were significantly boosted (p ≤ 0.01 and p ≤ 0.05 respectively) by the bivalent vaccine in the COVID-naïve group only (Fig.5A). In the COVID+ group, the breadth of class II-restricted spike-specific responses was significantly higher compared to the COVID-naïve group at baseline but did not further increase after boosting (Fig.5A). We also looked at the distribution and number of unique clonotypes HLA class-I and class II-restricted epitopes across the entire SARS-CoV-2 proteome, with each bar in Fig. 5B corresponding to a MIRA interval of amino acid residues containing one or more contiguous epitopes. HLA class I and II-restricted epitopes were scattered throughout the entire viral proteome for both COVID+ and – participants (Fig. 5B), likely due to cross reactivity with seasonal coronaviruses prior to SARS-CoV2 (41).

Figure 5: SARS-CoV-2 Spike specific TRB repertoire pre and post.

Omicron BA.4/5 bivalent mRNA vaccine

(A) SARS-CoV-2 spike protein specific class 1 or class 2 T cell receptor beta repertoire breadth or summed frequency in 21 vaccinated subjects (COVID+ n=12; COVID− n=9) whose samples were obtained both pre and post bivalent boost.

(B) ImmunoCODE and VDJdb based SARS-CoV-2 specific clonotype hits associated with Class 1 and 2 T cell clonotype across the SARS-CoV-2 peptidome for the COVID + (n=12) and COVID – (n=9) vaccinees.

Bar length denotes number of clonotypes. Each bar represents a MIRA interval of amino acid residues containing a set of nested epitopes. Bar colors represent location on the viral proteome. Bar color representing different regions of the viral proteome recognized by SARS-CoV-2 specific clonotypes. A scheme of the viral open reading frame (ORF) is shown at the bottom. In (A) p value derived from two-tailed Wilcoxon rank-sum test. Box plots show the median, first and third quantiles (lower and upper hinges) and smallest (lower hinge − 1.5× IQR) and largest (upper hinge + 1.5× IQR) values (lower and upper whiskers). Significance value is expressed as follows: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Last, breakthrough infections, within a period of sixth months post bivalent boost, occurred in 4 out of 15 COVID− individuals versus 2 out of 25 COVID+. Of note both COVID+ individuals that got infected had negative anti-nucleocapsid and positive anti-ORF-8 at baseline, indicating distal infection most likely before the omicron era (Table1).

Discussion

It has been shown that mRNA booster vaccines induce SARS-CoV-2-specific humoral and cellular B and T cell responses (5, 42). However, the strength of these responses is influenced by prior natural infection status (5, 10, 13). To better dissect the impact of previous natural infection, we studied two cohorts: those with (COVID+ group), and those without (COVID-) a history of natural SARS-CoV-2 infection. This design helped identify qualitative and quantitative differences between vaccine-only versus hybrid immunity in a real-world setting.

We observed that prior to bivalent vaccine boost, anti-Spike IgG levels, antibody neutralizing activity and specific breadth of class II-restricted SARS-CoV-2-specific TCR repertoire, were higher in the COVID+. Since both groups had received at least three doses of BNT162b2 or mRNA-1273 between 8 and 14 months prior to the bivalent vaccine, the lower levels of anti-spike IgG, antibody neutralizing activity, and reduced TCR diversity in COVID− individuals suggest shorter durability of vaccine-induced humoral and cellular protection compared to natural infection (43–45). Progressive waning of antiviral antibody response to SASRS-CoV-2 has been reported beyond 5 to 9 months after administration of COVID-19 mRNA vaccines (46). The interval (8–14 months) between the last vaccine dose and the bivalent vaccine may have contributed to inferior humoral and cellular immune responses in the COVID− group at baseline in this study.

The bivalent booster significantly increased anti-spike IgG levels and diversity of virus-specific TCR repertoire in both groups; however, a higher fold-increase of anti-spike IgG and of neutralizing ability against BA.4/BA.5 was seen in COVID− individuals. Furthermore, neutralizing activity against the α and δ variants increased in the COVID− group but not in the COVID+ group. Conversely, the bivalent vaccine significantly boosted neutralizing activity against BA.4/BA.5, BQ.1, and XBB.1.5 in both groups, reinforcing the need for updated vaccines to combat emerging variants (6, 14, 25).

The correlation index between post-boost anti-Spike IgG levels and neutralization capacity against all Omicron VOCs tested was higher in COVID+ than in COVID− vaccinees. The same anti-spike antibody levels resulted in better neutralization in recovered individuals, suggesting hybrid immunity provides a qualitative advantage. This aligns with previous studies conducted upon vaccination with a booster dose of the original vaccine (7, 47, 48) and findings that hybrid immunity protects against breakthrough Omicron infections post-booster (49). In our study, breakthrough infections, within a period of sixth months post bivalent boost, occurred in four of 15 COVID− individuals versus two of 25 COVID+ (both negative for anti-nucleocapsid and positive anti-ORF-8 indicating distal infection) (Table1). The higher frequency of breakthrough infections in the COVID− individuals further supports hybrid immunity’s superiority.

Higher neutralizing efficiency in COVID+ individuals appears independent of bivalent vaccination, given the lack of significant post-boost increases in neutralizing antibody levels for most VOCs. The exception was BA.4/BA.5, where post-boost neutralization was higher in COVID+ individuals, suggesting some benefit when the vaccine matches the antigen (24, 50). Comparing relative post-boost neutralization across VOCs, the lowest levels were observed against XBB.1.5, consistent with prior studies (51, 52) and its high breakthrough infection rates post-bivalent booster (34, 53).

In our cohort, symptomatic SARS-CoV-2 infection had occurred between 3 and 12 months prior to bivalent booster (mean: 8.4 months). Previous studies had shown that B cells of individuals who were recently infected (approximately 6 months prior) with SARS-CoV-2 are less responsive to the prototype variant booster vaccine, and that the interval between prior infection and booster vaccination is a critical determinant of the immune response to the booster vaccine (13). Accordingly, B cell numbers did not significantly increase post-booster in COVID+ individuals. However, we found no correlation between time since infection and fold increases in anti-Spike IgG, neutralization capacity, or S1+ RBD1+ B cell numbers. Additionally, a positive correlation between SARS-CoV-2-specific B cell numbers and neutralizing antibody titers was only observed in the COVID− group, suggesting that increased neutralization in COVID+ individuals may be independent of B cell recall.

Use of high dimensional spectral flow cytometry revealed distinctive SARS-CoV-2-specific B cell features in COVID+ and COVID− subjects, both at baseline and after booster. At baseline, COVID+ individuals had a higher frequency of S1+RBD+ smB cells reactive to the parental strain. Moreover, these S1+RBD+ smB cells showed different surface immunoglobulin expression, with a higher fraction of IgA+ cells in the COVID+ group, a difference persisting post booster. This may reflect distinct memory B-cell developmental pathways depending on the exposure. In particular, induction of virus-specific mucosal IgA by natural infection promotes protection against subsequent SARS-CoV-2 re-infection, whereas mRNA vaccination generates weaker mucosal IgA responses in the respiratory tract compared to infection (32, 54–56). Other factors contributing to observed B cell response differences include antigen exposure breadth (spike-only vs. broader viral antigens) and antigen presentation (lipid nanoparticles vs. virions). COVID+ individuals had a higher fraction of IgA+ CXCR3+ S1+RBD+ smB cells, which facilitate migration to inflammation sites (57) and tissues (58). Enrichment in this cluster may account for the expansion of IgA⁺ S1⁺RBD⁺ smB cells in recovered vaccinees.

Certain virus-specific B cells features were similar in both groups. Post-booster, CD27+ CD21low CD11c+ B cells increased (Fig. 2D), consistent with their role as pre-antibody-secreting cells (59–61). A corresponding increase in FCRL3/5-expressing CD27+ CD21low CD11c+ B cells (cluster 8)—linked to rapid antibody secretion—was observed (62). In addition, there was an increased proportion of CD21^low CD62L⁺ memory B cells post-booster in both COVID+ and COVID− groups. An enrichment of circulating CD62L⁺ memory B cells has been previously reported following SARS-CoV-2 infection or vaccination and has been associated with potently neutralizing BCRs and positively correlated with neutralizing Ab titers (63).

Similar somatic hypermutation rates were detected among SARS-COV-2-specific IGVH clonotypes in COVID+ and COVID− vaccinees (Fig. S4C), consistent with previous observations (42).

For T-cell responses, PBMCs were stimulated in vitro with spike peptide pools, and CD4+/CD8+ activation markers and cytokine production were analyzed (64). Pre-boost, similar frequencies of spike-reactive memory CD4+ and CD8+ cells were found in both groups, indicating comparable epitope recognition breadth. However, the bivalent boost increased spike-specific memory CD4+ cells in COVID− individuals, while responses were more variable in COVID+ individuals. High-throughput TCR repertoire analysis showed pre-boost SARS-CoV-2-specific clonotype breadth and frequency were similar across groups, but post-boost class II-restricted responses increased only in COVID-naïve individuals. This aligns with prior studies where frequent antigen stimulations, particularly in hybrid immunity cases, may limit T cell responses (20–22, 65). Weaker memory CD8⁺ T-cell responses were detected across all vaccinees likely due to CD8⁺ T cells playing a lesser role in responses to non-live viral vaccines (22).

In summary, bivalent boosters induce robust adaptive responses against Omicron variants. Prior infection is linked to increased neutralizing activity and mucosal immunity signatures. Consideration should be given to defining an optimal time interval between natural infection and booster immunization to avoid inefficiency of immune responses to repeat doses of the vaccine. Our findings emphasize that booster-induced protection depends on the infection status of the individual.

Clinical Implications.

The Omicron BA.4/BA.5 bivalent mRNA booster induces strong immune responses. Protection is higher in COVID recovered vaccinees and optimal booster timing after natural infection should be defined to maximize benefits.

Abbreviations

BCR

B cell receptor

CDR3

complementarity determining region 3

COVID-19

coronavirus disease 2019

HTS

high throughput sequencing

IFN

interferon

Ig

immunoglobulin

IGH

immunoglobulin heavy chain

IQR

interquartile range

MHC

Major Histocompatibility Complex

MIRA

Multiplex Identification of T-cell Receptor Antigen Specificity

N

Nucleocapsid protein

NSP

non-structural protein

ORF

open reading frame

PBMCs

peripheral blood mononuclear cells

PCR

polymerase chain reaction

RBD

receptor binding domain

S

Spike protein

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2

TCR

T cell receptor

TRB

T cell receptor β

VOC

variant of concern

WT

Wild Type