BALB/c mice were purchased from SPF (Beijing) Biotechnology Co., Ltd. All studies were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the Beijing Institute of Biological Products Co., Ltd., Beijing, China (No. BSYYF20230130002). All mice in this paper were six- to eight-week-old female mice on a BALB/c background.

For the vaccination regimens, 0.5 μg prototyped SARS-CoV-2 inactivated vaccines (BBIBP-corv) combined with 22.5 μg aluminum hydroxide adjuvant (Croda) or 0.5×10¹⁰ viral particles (VP) of adenovirus 5 vector vaccines (Ad) encoding full-length prototyped SARS-CoV-2 spike (Convidecia) were injected intraperitoneally in 500 μL 0.01 M PBS. The control mice were vaccinated with 0.01 M PBS containing an equal concentration of aluminum hydroxide adjuvant. The first two doses were administered at D0/D21, and the subsequent booster doses were administered every 14 days ( Figure 1A ). Blood from each vaccination and spleen samples from 4In and 3InAd were collected on day 10 after vaccination ( Figure 1A ).

IgG titers specific to spike, S1 (a variable region in spike), S2 (a conserved region in spike) or NTD (a conserved region in S1) in serum and spike- or S2-specific IgG titers in supernatants were determined. 96-well plates (Costar, cat 42594) were coated with 100 μL of SARS-CoV-2 spike (Sino, cat 40589-V08H4) or S1 (Sino, cat 40591-V08B1), S2 (Sino, cat 40590-V08H1), or NTD (Sino, cat 40591-V49H) at 0.5 μg/mL overnight at 4°C. After 2 h of blocking with 3% BSA in PBS, 100 μL of each diluted serum sample or culture supernatant was incubated for 2 h at room temperature. The serum dilutions used were 1:10000 and 1:40000 (for spike, S1 and S2) and 1:1000 and 1:4000 (for NTD). The supernatants were not diluted. Standard curves were obtained by serial dilution of standard serums via titration, as previously described (19). The secondary antibody, anti-mouse IgG conjugated to HRP (Cytiva, cat. NA931V), was diluted 1:4000 in PBS-3% BSA and incubated for 1 h at room temperature. The reactions were developed with HRP substrate, and the absorbance was measured at 405 nm and 630 nm.

NAb were determined using authentic and pseudo-SARS-CoV-2 virus, and the 50% inhibitory dilution (ID50) was defined as the serum dilution. NAb against authentic viruses (prototype and XBB.1.16 virus) were determined via a microcytopathogenic effect assay with a minimum eight-fold dilution and 2-fold serial dilutions, as previously described (20). NAb against pseudoviruses (prototype and XBB.1.16 virus) were determined with a minimum twenty-fold dilution and 3-fold serial dilutions, as previously described (12).

Spleens from the 3InAd, 4In and control groups were separated and stored in liquid nitrogen. All RNA extraction, library preparation, and sequencing were performed by BGI Genomics (Wuhan) Co., Ltd. DEGs were analyzed using DESeq2 (v1.4.5). KEGG and GO enrichment analyses were carried out as previously described (21, 22).

The fixable viability dye eFluor™ 506 (eBioscience, cat 65–0866-14) was used to exclude dead cells according to the manufacturer’s instructions. The following antibodies were used for surface staining at 4°C for 30 min: α-TCR (BioLegend, clone H57–597), α-CD4 (BioLegend, clone RM4–5), α-PD-1 (eBioscience, clone J43), α-CXCR5 (BioLegend, clone L138D7), α-B220 (BioLegend, clone RA3–6B2), α-Gl7 (BioLegend, clone GL7), and α-CD95 (BioLegend, clone SA367H8). The data were analyzed using FlowJo v10.8 or CytExpert.

B cells and CD4+ T cells were separated from splenetic cell suspensions using magnetic isolation kits (Stemcell, cat. 19854 for B cells, cat. 19852 for CD4+ T cells) according to the manufacturer’s instructions. B and CD4+T cells were sorted to 90% purity and utilized in the subsequent experiment.

CD4+CXCR5+PD-1+ Tfh cells or B220+CD95+Gl7+ GC B cells were sorted via the following steps. In brief, magnetically isolated CD4+ T or total B cells were stained with viability dye (eBioscience, cat 65–0866-14) and antibodies as above for 30 min at 4°C. Then, the cell suspensions were passed through 70-micron filters and resuspended in RPMI 1640 (Gibco, cat 11875093) supplemented with 10% FBS, 1 mM EDTA, 100 U penicillin, and 100 mg/ml streptomycin. Cell sorting was performed on an Aria III instrument.

Isolated B cells and Tfh cells were cocultured as previously described with minor modifications (23, 24). 3×10^5/well B cells were cultured with or without 1.5×10^5/well Tfh cells in the presence of 5μg/ml spike peptide pools (GenScript, cat RP30020) in RPMI 1640 (Gibco, cat 11875093) supplemented with 10% heat-inactivated fetal bovine serum (Gibco, 16140071), 2 mM L-glutamine, 1 mM sodium pyruvate, 50 mM 2-mercaptoethanol, 100 U penicillin, and 100 mg/ml streptomycin. Round-bottom tissue culture plates were precultured with 2μg/ml anti-CD3 antibody (BD Biosciences, cat 553057) and 5μg/ml anti-IgM (Invitrogen, cat 16–5092-85) in PBS at 4°C overnight. Six days after coculture, the cells were stained with α-IgG1 (BioLegend, clone RMG1–1), α-IgG2b (BioLegend, clone RMG2b-1), and α-Gl7 (BioLegend, clone GL7). The supernatants were harvested, and the anti-spike IgG titers were assessed via ELISA.

CD4+T or B cells were magnetically sorted to 90% purity and used for RNA extraction. Total RNA was extracted from isolated cells using TRIzol reagent (Invitrogen, 15596026). First-strand cDNA was transcribed (Vazyme, cat R312–3), and real-time PCR was performed using universal qPCR SYBR Green Master Mix (Vazyme, cat Q711–03) according to the manufacturer’s protocols. The experiment was performed on a CFX96 instrument (Roche). The data were normalized to the housekeeping gene Gapdh and analyzed by the 2^(–△△CT) method.

The primers used for cDNA amplification were obtained from previous methods (25–28) and are listed as follows:

Gapdh, (forward) 5’-GTGAAGGTCGGTGTGAACGGATT-3’ and (reverse) 5’-GGAGATGATGACCCTTTTGGCTC-3’; Il21, (forward) 5’-GCCAGATCGCCTCCTGATTA-3’ and (reverse) 5’-CATGCTCACAGTGCCCCTTT-3’; Bcl6, (forward) 5’-CCGGCTCAATAATCTCGTGAA-3’ and (reverse) 5’-GGTGCATGTAGAGTGGTGAGTGA-3’; Cxcr5, (forward) 5’-ACTCCTTACCACAGTGCACCTT-3’ and (reverse) 5’-GGAAACGGGAGGTGAACCA-3’; Cxcr4, (forward) 5’-TCCAACAAGGAACCCTGCTTC-3’ and (reverse) 5’-TTGCCGACTATGCCAGTCAAG-3’; Aicda, (forward) 5’-GGCATGAGACCTACCTCTGC-3’ and (reverse) 5’-CAGGAGGTGAACCAGGTGAC-3’.

Antigen-specific B cells were labeled and sorted as previously described with minor modifications (29, 30). In brief, biotinylated S1 (Acro, cat S1N-C82E8), S2 (Acro, cat S2N-C52E8) and NTD (Acro, cat S1D-C52E2) proteins were conjugated to two different streptavidin-fluorochrome conjugates, streptavidin-APC (BioLegend, cat 405207) and streptavidin-PE (BioLegend, cat 405203), in equimolar ratios. The cells were incubated with S1, S2, NTD, viability dye (eBioscience, cat. 65–0866-14) or α-CD19 (BioLegend, clone 1D3/CD19) at 4°C for 30 minutes. Then, the cells were sorted as above. Sorted cells were captured in droplets to generate nanoliter-scale gel beads in EMulsions (GEMs). The 5’ gene expression library and BCR library were prepared, and sequencing was performed by Abiosciences (Beijing) Co., Ltd. The mRNA sequencing data were processed using CellRanger version 7.0. R version 4.3.1 and the Seurat package (v5.0.0) were used for downstream analysis. The following marker genes were examined: Bcl2 and Itgax (CD11c) for memory B cells (31, 32); Fas and Mki67 for GC B cells; Xbp1 and Mzb1 for plasma cells, while other non-B cells were removed. The SHM frequency was calculated for each heavy chain sequence in the variable segment leading up to the CDR3.

Statistical analysis was performed utilizing GraphPad 8.0. The Mann−Whitney test was used and is explicitly described in the figure legends.

To compare the antibody responses induced by heterologous and homologous vaccination, we designed different immunization strategies ( Figure 1A ). Mice were vaccinated intraperitoneally with SARS-CoV-2 inactivated vaccines (In) or adenovirus 5 vector vaccines (Ad) encoding the full-length SARS-CoV-2 spike protein. The homologous vaccination regimen was 1 to 5 In doses (marked as In, 2In, 3In, 4In and 5In), and the heterologous immunization regimen was 3 doses of In followed by one dose of Ad (3InAd). Blood was collected on day 10 after the last homologous or heterologous booster dose. IgG titers specific to spike, S1 (a variable region in spike), S2 (a conserved region in spike), the N-terminal domain (NTD, a conserved region in S1), and neutralizing antibodies (NAb) against spike pseudovirus, as well as the authentic virus, were determined to explore the antibody dynamics after heterologous and homologous vaccination.

To assess the antibody responses in the homologous group, IgG and NAb titers were measured after each vaccination ( Supplementary Figure 1 ). For SARS-CoV-2 vaccination groups, the spike-specific IgG titer peaked after the 3rd dose of the vaccine and remained at this level even after the following doses ( Supplementary Figure 1A ). The S1-specific antibody showed similar changes ( Supplementary Figure 1A ). Interestingly, S2-specific IgG levels peaked after the 4th dose and decreased after the 5th dose ( Supplementary Figure 1A ), and NTD-specific IgG levels consistently increased ( Supplementary Figure 1A ). These results indicate that more than 3 doses of homologous prime-boosts with In may induce higher titers of IgG specific to the conserved spike protein regions ( Supplementary Figure 1A ). Moreover, the titers of NAb against the prototype spike pseudovirus slightly increased after the 5th dose ( Figure 1F and Supplementary Figure 1B ), while the titers of NAb against the prototype authentic virus peaked after the 4th dose ( Figure 1H and Supplementary Figure 1C ). Additionally, we found that every dose can induce antibody responses to the XBB.1.16 virus ( Supplementary Figures 1B, C ).

We next compared antibody responses between heterologous and homologous vaccination ( Figure 1 ). Compared with those in the 4In group, the spike-, S1-, S2-, and NTD-specific IgG levels were significantly higher in the 3InAd group ( Figures 1B-E ). Similar results were obtained for the titers of NAb against the prototype spike pseudovirus and the prototype authentic virus ( Figures 1F, H ). Compared with those in the 5In group, the spike-, S1-, and S2-specific IgG titers were also higher in the 3InAd group ( Figures 1B-D ), but there was no difference in the NTD-specific antibody titers ( Figure 1E ). Moreover, we found that the titers of NAb against the prototype pseudovirus or authentic virus were significantly higher in 3InAd ( Figures 1F, H ). Heterologous immunization showed stronger neutralizing effects on authentic XBB.1.16 virus ( Figure 1I ), although no differences were found on XBB.1.16 pseudovirus ( Figure 1G ). These data suggest that heterologous vaccination induced more robust IgG and NAb responses than homologous vaccination, and even the 5In homologous vaccination regimen could not induce the same antibody response as that induced by heterologous vaccination. In the following study, we further compared the 4In and 3InAd groups.

To explore the heterogeneity of homologous and heterologous immunization strategies, RNA-seq was performed on the spleens from the heterologous and homologous group mice to compare discrepancies in the GC response according to gene expression profiling.

One hundred fifty-two differentially expressed genes (DEGs) were screened between the 3InAd and 4In groups ( Figure 2A ). Among these DEGs, 38 immune-related genes were found using the GP_CFP and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway classification methods ( Figure 2B ). Then, through GO biological process (BP) and KEGG pathway enrichment analyses, we found that these immune-related DEGs were enriched in somatic hypermutation and production of immunoglobulin (Nuggc, Aicda, Il21, Il6), Tfh differentiation (Il21, Il6), germinal center B cell differentiation (Il6, Il21), cellular response to virus (Gli2, Il6, Il21) ( Figures 2A, C, D ). To further explore signaling changes in the 3InAd group, we next performed gene set enrichment analysis (GSEA) using transcripts per million (TPM) values. We found that genes associated with the positive regulation of cell migration, the immunoglobulin-mediated immune response, and the positive regulation of the B-cell receptor signaling pathway were upregulated in the 3InAd group compared to the 4In group ( Figures 2E, F ). These gene expression profiling results suggest that heterologous vaccination enhances B-cell and Tfh cell responses and GC activation.

Tfh cells are specialized B helper cells that enable the proliferation, survival, and differentiation of GC B cells through the delivery of cytokines and costimulatory signals (12). To further investigate the changes in Tfh cells and GC responses after SARS-CoV-2 heterologous and homologous vaccination, the number and function of Tfh cells were analyzed.

We found that the 3InAd group showed higher Tfh cell populations than the 4In group ( Figures 3A, B ), indicating that heterologous vaccination induced more robust Tfh cell expansion. Moreover, to evaluate Tfh cell function in the heterologous group, PD1+CXCR5+ Tfh cells sorted from 3InAd or 4In group mice were cocultured with mature splenic B cells from 2In group mice (mice vaccinated with 2 doses of In), and B-cell activation defined by expression of Gl7 marker and isotype switching were detected. The coculture strategy is shown in Figure 3C . When B cells cocultured with 3InAd Tfh cells, we only found a slight increase of activation in the Gl7+IgG1+B group ( Figures 3D, E ), but both groups exhibit no statistical differences to the 4In Tfh cells, showing no discernible differences in Tfh helper function between heterologous and homologous groups. However, when antibodies in the coculture supernatants were further detected, a significant increase in spike-specific IgG titers was identified in B cells cocultured with 3InAd Tfh cells ( Figure 3G ), indicating an increase in IgG titers in the absence of significant B-cell activation. This discrepancy may be attributed to the different sensitivity between antibody detection and other immunologic parameters examination during the in vitro experiment (33). Additionally, in our study, Tfh cells were repeatedly activated by 3 doses of inactivated vaccination before the 4th-dose vaccination, therefore, the distinction of Tfh function between two groups may not be optimized.

It has been reported that IL-21, BCL-6 and CXCR5 play key roles in regulating Tfh cell differentiation and function. To compare Tfh cell proliferation and function at the gene level, we measured the mRNA expression of Il21, Bcl6 and Cxcr5 in the 3InAd and 4In groups on day 10 after vaccination. Il21, Bcl6 and Cxcr5 expression was significantly higher in the 3InAd group than that in the 4In group ( Figures 3H-J ), suggesting that enhanced Tfh cells populations and function in the heterologous group may be related to upregulating the expression of Il21, Bcl6 and Cxcr5.

We have proven that Tfh cell function is enhanced by heterologous vaccination. Tfh cells provide stimulatory signals to B cells, which promote their survival and ongoing proliferation (34). To compare B cells between the heterologous and homologous groups, we analyzed their proliferation and function.

Compared to those in the 4In group, the populations of GC B cells in the 3InAd group significantly increased ( Figures 4A, B ), indicating that a heterologous boost dose promote GC B-cell expansion. Then, we further compared the activation of B-cell from these two groups. CD19+ B cells separated from the 3InAd and 4In groups were cocultured with mature splenic Tfh cells sorted from the 2In group for 6 days in vitro ( Figure 4C ). We found that frequencies of Gl7+IgG1+B and Gl7+IgG2b+B cells was significantly greater in the 3InAd B cells cocultured with Tfh cells than in the 4In B cell coculture ( Figures 4D-F ). Next, antibody titers in the collected coculture supernatants were measured by ELISA. We found significantly increased anti-spike IgG titers in 3InAd B cells ( Figure 4G ). These results suggest that a heterologous booster dose induces more robust B-cell activation and greater antibody production.

To assess the relationship between GC B cells and antibody responses in both the heterologous and homologous groups, Spearman correlation analysis was used. We found a strong positive correlation between spike-specific IgG titers and the percentage of GC B cells in the 3InAd and 4In groups ( Supplementary Figure 2A ). This correlation was also observed between NAb titers against authentic virus and the frequencies of GC B cells in both the heterologous and homologous groups ( Supplementary Figure 2B ). These results suggest that heterologous vaccine induced more SARS-CoV-2-specific B cells, which led to higher antibody titers.

We have demonstrated that heterologous vaccination increases B-cell activation and antibody production in mice. B cells play a vital role in immune responses by generating NAb that defend SARS-CoV-2 infection and prevent reinfection (35). To compare the differences of antigen specific B-cell responses in heterologous and homologous vaccination, Libra-seq (linking B-cell receptor to antigen specificity through sequencing) was used to map spike-, S1-, S2- and NTD-specific BCR sequences. The number of SHMs and the expansion of antigen-specific B cells were evaluated.

Compared to those in the 4In group, the SHMs in the 3InAds group were increased in spike-, S1-, S2- and NTD-specific B cells ( Figure 5A ). S2-specific B cells exhibited 1.95-fold increases, respectively ( Figure 5A ). The 3InAd group exhibited increased SHMs in highly expanded clones (found more than 4 times), moderately expanded clones (found 2 to 3 times) and singleton clones ( Figures 5B, C , Supplementary Figure 3A ). These results suggest that a heterologous booster dose promoted higher SHM of B cells specific to conversed regions. We next assessed the SHM in phenotypes of the B cells above in heterologous and homologous groups. There was a slight change in the SHM of GC B cells in the 3InAd group compared to the 4In group ( Figure 5D ). The SHMs in the spike-, S1-, S2- and NTD-specific plasma cells from the 3InAds group exhibited 4.90-, 4.45-, 6.06- and 1.15-fold increases, respectively ( Figure 5E ). Additionally, we observed 1.98- and 1.34-fold SHM increases in spike- and S1-specific memory B cells, respectively ( Figure 5F ). These results suggest that the increased SHM of the heterologous group was mainly enriched in S2-specific plasma cells, which may contribute to high-affinity antibodies.

SHMs are regulated by activation-induced cytidine deaminase (AID), an enzyme that induces point mutations in CDR3 and leads to B-cell expansion (36–38). It is predominantly expressed in CXCR4^hiGC B cells⁴³. To further assess the source of SHMs in the B cells of heterologous and homologous groups, we measured the mRNA expression of Aicda (encoded AID) and Cxcr4 in GC B cells. We found that Aicda and Cxcr4 were significantly upregulated in the 3InAds group compared to the 4Ins group ( Figure 5G and Supplementary Figure 3B ), indicating that heterologous vaccination upregulated higher AID expression to increased SHMs in B cells than homologous group.

To measure the expansion and proliferation of B cells, we calculated the cell number ratio of each BCR clonotype to each antigen-specific B-cell number. There was a greater percentage of spike-specific top 2 BCR clones in the 3InAd group ( Figure 5H ). Similarly, the proportions of the top 10 BCR clones specific to S1 exhibited similar tendencies in the 3InAd and 4In groups ( Supplementary Figure 3C ), whereas the proportions of the S2- and NTD-specific clones in the 3InAd group were greater than those in homologous group ( Figures 5I, J ). These data demonstrated that heterologous vaccination induced greater expansion of BCRs enriched in the conserved regions in the spike protein.