How different doses of the original vaccine affect immunity to BA.5, illness severity, and risk of XBB reinfection

The baseline characteristics of the participants are presented in Fig. 1A. We conducted analyses to investigate the impact of inactivated virus and S protein-based vaccines on the severity of COVID-19 caused by the SARS-CoV-2 BA.5 variant during the acute phase. Our results indicated that the two vaccine types, when administered at equivalent doses, did not show significant differences in terms of disease severity (Fig. 1B). Furthermore, no significant differences were observed in terms of disease severity between females and males who received identical doses (Fig. 1C).

To investigate whether different vaccination doses influenced the severity of COVID-19 induced by the BA.5 variant, we conducted a study assessing disease severity among 359 patients who had received zero, one, two, or three vaccine doses prior to becoming infected with the BA.5 variant (Fig. 1D). The incidence of severe or critical illness was lower (P < 0.0001) among vaccinated individuals (Vac-1, Vac-2, and Vac-3) than among unvaccinated ones (Un-vac), and the proportion of mild illness was higher in those who received ≥2 doses than in those who received only one (P = 0.001). However, no significant difference (P = 0.673) was observed in terms of disease severity between individuals who received two vs three doses. These data suggested that vaccination against WT SARS-CoV-2 mitigated the severity of COVID-19 attributable to BA.5 and that ≥2 vaccine doses significantly reduced the risk of developing pneumonia.

Although WT vaccination could not induce anti-BA.5 NAbs even at 14 and 30 days after vaccination (11), it remained unclear whether it could affect infection-induced NAbs against the BA.5 and subsequent XBB strains. We, therefore, measured the patients' NAb titers against the WT, BA.5, and XBB1.9.1 strains before and after BA.5 infection (Fig. 2). Two months before the BA.5 outbreak, those who received three doses of WT vaccine had significantly higher levels of anti-WT NAbs than those who received zero to two doses (Fig. 2A). However, anti-BA.5 and anti-XBB1.9.1 NAb levels were both below the detectable limit in all of the groups (Fig. 2A).

At 20–50 days post-BA.5 infection, the group that received three vaccine doses showed significantly higher NAb levels than the groups that received zero, one, or two doses (Fig. 2B). Furthermore, those who received two doses exhibited significantly higher levels of anti-BA.5 NAbs compared to those who were unvaccinated. However, no significant differences were observed between the two-dose and one-dose groups or between the one-dose and zero-dose groups. The effect of vaccine dose on NAb titer also varied significantly among the different variants. For the anti-WT NAb titer, a clear increase was observed at higher vaccination doses (3 > 2 > 1 > 0; Fig. 2B). The anti-XBB1.9.1 titer for the three-dose group was notably higher than that for the other doses, whereas no significant difference was observed between the two-, one-, and zero-dose groups (Fig. 2B). Notably, different anti-BA.5 NAb doses were only observed between the ≥2 doses and <2 doses at 3 months post-infection (Fig. 2C), indicating that ≥2 vaccine doses were necessary for long-lasting immunity.

To investigate whether vaccine doses affect the magnitude of NAbs against different strains and variants, as well as how rapidly they are generated, we used the ggplot2 package for R (version 4.2.1) to generate fitted curves based on NAb titers at certain time points. Our data revealed that anti-BA.5 NAbs in the two- or three-dose groups were induced earlier and at higher levels (within 1 week) vs those in the one- or zero-dose groups (Fig. 3A). A similar trend was observed for anti-WT NAbs (Fig. 3B) but not for anti-XBB1.9.1 NAbs (Fig. 3C).

To investigate the potential impact of the number of vaccination doses on immune imprinting following BA.5 infection, we calculated the ratios of anti-WT, anti-BA.5, and anti-XBB1.9.1 NAbs at 3 months following BA.5 infection, as an indicator of immune imprinting. The ratios in the patients who received three, two, one, and zero doses were 31.4:8.1:1, 20.2:7.3:1, 2.4:5.1:1, and 1.5:4.4:1, respectively (Fig. S1). Our findings clearly show that different numbers of vaccine doses have different impacts on immune imprinting following BA.5 infection, with higher doses of WT vaccines resulting in more dominant anti-WT immune imprinting.

We then conducted an analysis to assess the impacts of inactivated virus and S protein-based vaccines on NAb levels 3 months following infection. Our findings revealed no statistically significant difference in the anti-BA.5 or anti-WT immune responses elicited by these two vaccine types when administered in equivalent doses (Fig. S2A and B).

To investigate the impact of WT vaccine doses on XBB variant reinfection risk, we conducted an observational study in a hospital during the XBB reinfection wave among 843 hospitalized patients over 1 week. Notably, none of the patients were infected with SARS-CoV-2 when they were admitted to the hospital. They had varying vaccination histories—including zero, one, two, three, and four vaccine doses—and had previously contracted BA.5 infections 5 months prior. Throughout the observation period, XBB reinfection was confirmed through polymerase chain reaction (PCR) testing. Upon hospital admission, we performed a serum test to measure NAbs against the BA.5 and XBB.1.9.1 variants. Our findings revealed that those who had received either zero or one dose had significantly lower NAb levels compared to those who had taken ≥2 doses (Fig. 4A). Notably, the rate of XBB reinfection over the 1-week observation period was significantly lower among those who had received ≥2 doses (5/645, 0.78%) compared to those who were either unvaccinated or had received only one dose (6/198, 3.03%, P = 0.014; Fig. 4B). These results underscored the effectiveness of administering ≥2 doses of the WT vaccine in terms of reducing the risk of XBB reinfection.

To investigate the impact of vaccine doses on viral clearance and the inflammatory response, SARS-CoV-2 nucleic acids and inflammation-associated cytokines were analyzed in different vaccination groups on the 14th day post-infection (Tables S1 and S2). The data revealed that those who had been vaccinated with two or three doses exhibited a significantly higher rate of negative nucleic acid conversion (>70%) than the unvaccinated population (45.45%) at 14 days after symptom onset (Fig. 5A). The levels of the antiviral cytokine interferon (IFN)-α were higher in the three-dose group compared to both the unvaccinated and two-dose groups (Fig. 5B). The concentration of Regulated upon Activation, Normal T Cell Expressed and Presumably Secreted (RANTES), a major human immunodeficiency virus-suppressive factor (19), exhibited a statistically significant increase in the two- and three-dose groups compared to the unvaccinated one (Fig. 5C). Furthermore, the levels of the proinflammatory factors IL-6 and MIP-1α were comparatively lower in the vaccinated populations compared to the unvaccinated ones (Fig. 5D and E). However, no significant differences were observed in terms of IL-2 and IL-1β levels among the three groups (Fig. S3A and B). These findings suggested that the administration of multiple WT vaccine doses effectively promoted viral clearance and attenuated the inflammatory response.

To investigate the effect of vaccine doses on vaccination on the immunity induced by BA.5 infection at the molecular level, we performed RNA-seq analysis on peripheral blood mononuclear cell (PBMC) samples collected 14 days after symptom onset (Table S3). Principal component analysis revealed significant overall differences in the transcriptomes between the vaccinated (one, two, or three doses) and unvaccinated groups (Fig. 6A). The Vac-2 group could not be distinguished from the Vac-3 group. Compared to the Un-vac group, we observed the downregulation of 915 genes (fold change > 2, P < 0.05) and the upregulation of 1,030 genes in the Vac-2/3 group (Fig. 6B). Six genes were chosen for further validation by quantitative real-time PCR (qPCR). The results showed a concordant expression pattern with the RNA-seq data (Fig. 6C; Table S4), thereby confirming the reliability of the results. The differentially expressed genes (DEGs) we identified were predominantly enriched in various immune-related pathways, with the strongest enrichment being in the hematopoietic cell lineage (P = 1 × 10⁻⁶; Fig. 6D). The upregulation of genes encoding B cell surface molecules—including CD19, MS4A1, CD22, FCER2, and CR1—may be involved in the proliferation and differentiation of memory B cells induced by the vaccine doses, following BA.5 infection (Fig. 6E). The monocyte surface molecular genes HLA-DOA, HLA-DMB, HLA-DRB5, CD33, FCGR1A, and CSF1R were found to be upregulated in the Vac-2/3 group compared to the Un-vac one (Fig. 6F). By contrast, CSF2 expression was downregulated (Fig. 6F). Changes in these genes may be involved in the regulation of monocyte activation induced by vaccination. Furthermore, the Vac-2/3 groups exhibited upregulation of the complement components C4, C5, C3AR1, CR3, and CR4, along with the upregulation of the negative regulatory factors C1INH, CR1, and Clusterin in the complement pathway. This suggested that vaccination with two or three doses may promote complement pathway activation under precise regulation (Fig. 6G). Meanwhile, a pattern of downregulation was identified for most genes enriched in cytokine-cytokine receptor interactions (Fig. 6H) and chemokine signaling pathways (Fig. 6I). Alterations in these genes may play a role in the attenuation of inflammation induced by two or three vaccine doses.

This study aimed to investigate the impact of WT SARS-CoV-2 vaccination schedules on the immune response and COVID-19 disease severity following BA.5 infection, as well as the risk of reinfection. All of the participants (n = 2,216) were recruited from the Guangzhou First People's Hospital, and their basic information and vaccination records were obtained from the hospital's medical records. Blood samples were systematically gathered from the participants during four distinct phases—the pre-infection period, the acute phase, and at 3 and 5 months post-infection—to facilitate comprehensive analyses across the disease course.

As of October 2022, 214 samples had been obtained from a cohort of vaccinated or unvaccinated individuals who had not previously contracted SARS-CoV-2. Among them, 29 had not received any COVID-19 vaccination and 17, 57, and 111 had received one, two, and three doses, respectively (Fig. 1A, pre-infection). During the BA.5 epidemic, a total of 592 samples were collected from 359 patients with COVID-19 confirmed using a SARS-CoV-2 PCR test administered at 0–50 days following symptom onset (Fig. 1A, acute phase). Among them, 106 had not received any COVID-19 vaccination and 68, 70, and 115 had received one, two, and three doses, respectively. Their samples were used to evaluate their titers of SARS-CoV-2-specific NAbs and cytokine profiles during the acute phase of the disease. Three months following infection, a total of 800 participants were analyzed—of whom 139 had received no COVID-19 vaccination and 34, 126, and 501 had received one, two, and three doses, respectively (Fig. 1A, 3 months p.i.). Convalescent specimens were used to track their waning NAb levels. During the subsequent XBB wave, 843 samples were collected from hospitalized patients without XBB variant infections—of whom 162 had received no COVID-19 vaccination and 36, 135, 492, and 18 had received one, two, three, and four doses, respectively (Fig. 1A, 5 months p.i.). We tracked their SARS-CoV-2 infection statuses during the XBB outbreak via PCR confirmation at the same hospital over a 1-week period.

COVID-19 severity was defined using the Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia (Trial Version 10) published by the National Health Commission of China. These definitions encompass (i) mild: symptomatic patients who meet the COVID-19 case definition but do not show signs of viral pneumonia or hypoxia; (ii) moderate: patients with pneumonia detected via imaging, respiratory rate of <30 breaths/min and pulse oxygen saturation (SpO₂)of <93% on room air; (iii) severe: patient with pneumonia plus one of the following: progressive worsening of clinical symptoms and imaging tests showing a > 50% increase in pulmonary lesions within 24–48 h, arterial partial pressure of oxygen (PaO₂)/fraction of inspired oxygen (FiO₂) of ≤300 mmHg, respiratory rate of ≥30 breaths/min, or oxygen saturation ≤93% on room air; (iv) critical: patients with criteria for acute respiratory distress syndrome necessitating mechanical ventilation, septic shock, or multiple organ failure requiring intensive care unit monitoring and treatment.

The SARS-CoV-2 conventional neutralization assay was performed according to a previously described protocol (39). Heat-inactivated plasma samples were initially tested at a dilution of 1:8 for the WT and BA.5 strains, whereas a dilution of 1:4 was used for XBB.1.9.1. Because of antibody degradation, the samples collected 5 months after infection were diluted 1:4 for the BA.5 and XBB.1.9.1 strains. Stepwise 1:2 dilution was then performed to obtain eight data points. The diluted samples were subsequently mixed with 50 µL of viral solution containing SARS-CoV-2 WT, BA.5, or XBB.1.9.1 variants. Following incubation at 37°C for 1 h, Vero E6 cells (1.2 × 10⁴; ATCC, USA) were added to the mixtures. The cells were then incubated at 37°C in a humidified environment with 5% CO₂ for 4 days. The cytopathic effects were examined using a Celigo Imaging Cytometer (Nexcelom Bioscience, Lawrence, MA, USA). The presence or absence of cytopathic effect was determined through comparisons with positive and negative controls in each well. When the NAb titer fell below the limit of detection, we assigned an NAb titer of 50% inhibitory dilution (EC50) of four for specimens diluted at a ratio of 1:8, and one of two for those diluted 1:4.

The ggplot2 package for R was used to generate a fitting curve illustrating the NAb data across various vaccination schedules during the initial wave of infection.

Oropharyngeal swab samples were collected to measure viral RNA. The presence of viral RNA was detected using commercially available kits (Sansure Biotech, Changsha, China) targeting the ORF1ab and N genes, according to the manufacturer's instructions. This detection method uses qPCR. A positive result was determined if the Ct value was <40, whereas a negative or undetermined outcome was assigned to Ct values of ≥40.

This method was also used to detect the mRNA expression levels of the CSF2, CR1, ITGAM, C5, ITGB2, and C3AR1 PBMC genes. Total RNA was extracted using a Rapid RNA Extraction Kit (Goonie, Guangzhou, China). Subsequently, cDNA was obtained by reverse transcription using HiScript III RT SuperMix (Vazyme, Nanjing, China). The ChamQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China) was then used to perform qPCR according to the manufacturer's instructions, with the appropriate primers (Table S4). The reaction conditions were 95°C for 30 s, 95°C for 10 s, and 60°C for 30 s, over a total of 40 cycles. Relative quantitation was performed via the ΔΔCt method, and the expression of detectable genes was normalized to the 28S RNA reference gene.

Cytokine levels were assessed using a cytometric bead array (CBA; CBA Human Soluble Protein Flex Set System; BD Biosciences) on an LSR Fortessa instrument (BD Biosciences). Interleukin (IL)-6, IL-2, IL-1β, macrophage inflammatory protein-1α, and IFN-α levels were determined using a CBA assay. RANTES levels were measured via another CBA assay, using a 10-fold dilution. The IL-6 and RANTES levels were then re-evaluated using 100-fold dilutions when their concentrations exceeded the detection range. All of the experiments were performed according to the manufacturer's instructions. Analyte signal intensities were determined relative to their corresponding standards, and absolute concentrations of individual analytes were calculated using BD FCAP Array Software (BD Biosciences).

Approximately 1 million live PBMCs were thawed and dissolved in 1 mL of TRIzol (Invitrogen, USA). The manufacturer's instructions were followed to extract total RNA from the PBMCs using the reagent. An RNA-seq transcriptome library was prepared using a Hieff NGS Ultima Dual-mode mRNA Library Prep Kit (Yeasen, China) using 1 µg of total RNA. The library was sequenced using a NovaSeq 6000 instrument (Illumina). Raw sequence reads were subjected to fastp version 0.18.0 for filtering, which entailed the removal of reads containing adapters or >10% unknown nucleotides, and the exclusion of low-quality reads comprising >50% bases with Q-values of ≤20. The short-read alignment tool Bowtie2 version 2.2.8 was used to map reads against the ribosomal RNA database, followed by the removal of rRNA-mapped reads. Quality-filtered reads were subsequently mapped to the human genome using HISAT 2.2.4, and the other parameters were set as the default. The mapped reads of each sample were assembled using StringTie version 1.3.1 via a reference-based approach. For each transcription region, a transcript per kilobase of the exon model per million mapped reads value was calculated to quantify its expression abundance and variations using RSEM software (40). Differential RNA expression analysis was performed using DESeq2 software between pairs of different groups (41). The genes and transcripts with P values of <0.05 and absolute fold changes of ≥2 were considered DEGs or differentially expressed transcripts.

Statistical analyses were performed using SPSS version 20.0 and GraphPad Prism version 6.01. NAbs are reported as geometric mean titers. The Mann-Whitney U test was used to compare the central tendencies between pairs of groups. The Wilcoxon matched-pair signed-rank test was used to compare NAb levels between pairs of groups within the same cohort. The counts were analyzed using the Chi-squared test with a significance level set at P < 0.05 (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001).