A heterologous AZD1222 priming and BNT162b2 boosting regimen more efficiently elicits neutralizing antibodies, but not memory T cells, than the homologous BNT162b2 regimen

Abstract

Background

Comparative analyses of SARS-CoV-2-specific immune responses elicited by diverse prime-boost regimens are required to establish efficient regimens for the control of COVID-19.

Method

In this prospective observational cohort study, spike-specific immunoglobulin G (IgG) and neutralizing antibodies (nAbs) alongside spike-specific T-cell responses in age-matched groups of homologous BNT162b2/BNT162b2 or AZD1222/AZD1222 vaccination, heterologous AZD1222/BNT162b2 vaccination, and prior wild-type SARS-CoV-2 infection/vaccination were evaluated.

Results

Peak immune responses were achieved after the second vaccine dose in the naïve vaccinated groups and after the first dose in the prior infection/vaccination group. Peak titers of anti-spike IgG and nAb were significantly higher in the AZD1222/BNT162b2 vaccination and prior infection/vaccination groups than in the BNT162b2/BNT162b2 or AZD1222/AZD1222 groups. However, the frequency of interferon-γ-producing CD4⁺ T cells was highest in the BNT162b2/BNT162b2 vaccination group. Similar results were observed in the analysis of polyfunctional T cells. When nAb and CD4⁺T-cell responses against the Delta variant were analyzed, the prior infection/vaccination group exhibited higher responses than the groups of other homologous or heterologous vaccination regimens.

Conclusion

nAbs are efficiently elicited by heterologous AZD1222/BNT162b2 vaccination, as well as prior infection/vaccination, whereas spike-specific CD4⁺T-cell responses are efficiently elicited by homologous BNT162b2 vaccination. Variant-recognizing immunity is more efficiently generated by prior infection/vaccination than the other homologous or heterologous vaccination regimens.

1. Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has been spreading globally since its emergence in December 2019, and the development of effective vaccines against coronavirus disease 2019 (COVID-19) has been accelerated [1]. The success of the endeavor was seen relatively quickly, and several vaccines employing spike (S) protein as a vaccine antigen successfully demonstrated safety and efficacy in preventing symptomatic COVID-19. Specifically, mRNA-based vaccine BNT162b2 (Comirnaty®, Pfizer-BioNTech) and adenovirus vector-based vaccine AZD1222 (Vaxzevria®, AstraZeneca) were initially approved and widely administered as a two-dose prime-boost regimen. Two doses of BNT162b2 had 95 % (95 % confidence interval [CI], 90–98 %) efficacy against symptomatic COVID-19 [2], and two doses of AZD1222 had 70 % (95 % CI, 55–81 %) efficacy [3].

Intriguingly, heterologous prime-boost regimens of licensed COVID-19 vaccines gained substantial interest and were subsequently commenced in several countries despite insufficient data regarding immunogenicity and safety. BNT162b2 given as a boost dose in individuals prime vaccinated with AZD1222 induced robust S-specific immunoglobulin G (IgG) antibody (Ab) titers, neutralizing antibody (nAb) responses, and interferon-γ⁺ (IFN-γ⁺) T cells [[4], [5]]. However, few studies have evaluated the efficacy of heterologous AZD1222/BNT162b2 vaccination as compared to homologous BNT162b2/BNT162b2 vaccination [[6], [7], [8]], and these studies lacked comprehensive evaluation of S-specific T-cell responses. Accordingly, whether heterologous AZD1222/BNT162b2 vaccination results in potent cross-variant nAb or T-cell responses remains unclear.

Another example of heterologous priming-boosting can be seen in individuals who had been infected with SARS-CoV-2 and later vaccinated against the virus. Several studies suggested that S-specific binding Ab and nAb levels are detected in high levels in individuals with prior infection and vaccination, although there were no comparative analyses [[9], [10]]. Healthcare workers with prior SARS-CoV-2 infection have higher Ab titers in response to a single dose of mRNA vaccine than those who were not previously infected [11]. Moreover, individuals with prior infection exhibit enhanced S-specific T-cell responses and Ab-secreting memory B-cell responses following the prime dose without the boost dose [12].

In the present study, we comprehensively examined the humoral and cellular immunogenicity of homologous and heterologous prime-boost regimens in a prospective observational cohort with side-by-side comparisons of S-specific binding Ab, nAb, and S-specific CD4⁺ T-cell responses. We also evaluated the immunological outcomes in participants who had been infected with wild-type SARS-CoV-2 and received COVID-19 vaccines to determine whether prior infection and vaccination efficiently elicits immune responses against the Delta variant.

2. Methods

2.1. Study design

A multicenter prospective cohort was established in March 2021 to study immune responses to COVID-19 vaccination. Participants in this cohort were voluntarily recruited among healthcare workers and initially scheduled to receive two doses of BNT162b2 at a 3-week interval or AZD1222 at a 12-week interval. However, a portion of individuals prime vaccinated with AZD1222 received the booster dose of BNT162b2 between May and June 2021, and were included as the AZD1222/BNT162b2 vaccination group. Among the participants in the cohort, those who had prior wild-type SARS-CoV-2 infection and received BNT162b2 or AZD1222 were included as the prior infection/vaccination group. All vaccine recipients analyzed in this study were age-matched among the groups.

Peripheral blood samples were collected from all participants in the cohort at each time point (T): before and 3 weeks after the 1st vaccine dose (T1 and T2, respectively), 5 weeks after the 1st BNT162b2 dose or 11 weeks after the 1st AZD1222 dose (T3), and 14 weeks after the 1st vaccine dose (T4). A schematic representation of the study protocol is provided in Fig. 1 .

Fig. 1.

Schematic representation of the vaccine recipient cohorts. Study scheme for the detection of SARS-CoV-2 spike-specific binding antibodies, neutralizing antibodies, and T-cell responses in peripheral blood samples from individuals vaccinated for COVID-19. The study includes BNT162b2/BNT162b2 (n = 22), AZD1222/AZD1222 (n = 20), AZD1222/BNT162b2 (n = 9), and prior wild-type SARS-CoV-2 infection/vaccination (n = 11) groups. Among the prior wild-type SARS-CoV-2 infection/vaccination, 2 and 9 were vaccinated with BNT162b2/BNT162b2 and AZD1222/AZD1222, respectively. History of SARS-CoV-2 infection was confirmed by a positive anti-SARS-CoV-2 nucleocapsid antibody test. T, time point; wks, weeks.

For the purpose of comparison, convalescent samples were obtained from unvaccinated patients infected with the SARS-CoV-2 Delta variant confirmed by real-time reverse transcription polymerase chain reaction using the Seegene Allplex 2019-nCoV Assay kit (Seegene, Korea) and Illumina BTSeq SARS-CoV-2 whole-genome sequencing kit (Celemics Inc, Korea) on a MiSeq sequencer (150-bp paired-end mode; Illumina, San Diego, CA, USA). The base sequence obtained through BTseq was compared to NCBI reference sequence NC_045512.2 using the IGV 2.10 software to confirm the Delta variant.

The study was approved by the institutional review boards of each participating institute and conducted according to the principles of the Declaration of Helsinki. All participants provided written informed consent before enrollment.

2.2. Study participants

This cohort comprised 22 participants who received homologous BNT162b2/BNT162b2 vaccination (mean age, 38.0 years; 31.8 % men), 20 participants who received homologous AZD1222/AZD1222 vaccination (mean age, 39.7 years; 23.8 % men), and 9 participants who received heterologous AZD1222/BNT162b2 vaccination (mean age, 38.2 years; 33.3 % men). Of the 11 participants with prior wild-type SARS-CoV-2 infection (mean age, 38.5 years; 27.3 % men), 2 and 9 were vaccinated with BNT162b2/BNT162b2 and AZD1222/AZD1222, respectively. The demographics of the different vaccinated groups are provided in Supplementary Table 2. All participants with prior wild-type SARS-CoV-2 infection experienced asymptomatic or mild COVID-19, and 3 of them had pneumonia. Median time from confirmation of COVID-19 to the first vaccination was 121.5 (interquartile range [IQR], 102.3–252.5) days. The demographics of the participants with prior wild-type SARS-CoV-2 infection and vaccination are provided in Supplementary Table 1.

Convalescent samples were obtained from 17 patients (mean age, 53 years; 52.9 % men) infected with the SARS-CoV-2 Delta variant and hospitalized for moderate to critical COVID-19 (moderate, 17.6 %; severe, 35.3 %; critically ill, 47.1 %). The median time from the onset of COVID-19-related symptoms to sampling was 24 (IQR, 22–32) days. The demographics of the SARS-CoV-2 Delta convalescent participants are provided in Supplementary Table 2 and Table 3.

2.3. Binding antibodies

SARS-CoV-2 S-specific binding Abs were analyzed using a commercial immunoassay. The Elecsys anti-S protein assay (Roche Diagnostics, Germany) is an electrochemiluminescence immunoassay used to detect SARS-CoV-2 S-specific antibodies on the Cobas e411 analyzer (Roche Diagnostics, Germany), with a measurement range from 0.4 U/mL to 250 U/mL (up to 2500 U/mL with onboard 1:10 dilution, and up to 12,500 U/mL with onboard 1:50 dilution). The recombinant receptor binding domain of S protein was used with a double antigen sandwich principle. The antigens in the reagent capture predominantly anti-SARS-CoV-2 IgG. The Elecsys anti-nucleocapsid protein assay (Roche Diagnostics, Germany) was used to detect previous SARS-CoV-2 infection, with a cut-off index ≥ 1.0.

2.4. Plaque-reduction neutralization assay

Plaque-reduction neutralization assays for wild-type SARS-CoV-2 virus (BetaCoV/Korea/KCDC03/2020) and the Delta variant strain YS117 (GenBank accession number MZ798798 and GISAID accession number EPI_ISL_3411836) were performed using plasma samples. The virus was grown on Vero E6 cells in a biosafety level 3 laboratory at Avison Biomedical Research Center in Seoul, Korea. Inactivated plasma samples (135 µL) of inactivated plasma samples were 2-fold serially diluted (1:40 to 1:1280) and added to an equal volume of 100-times the 50 % tissue culture infectious dose (TCID50) of the virus. The isolates containing 100-times the TCID50 were incubated in 96-well plates for 60 min at 37 °C. Virus-plasma mixtures were incubated for 60 min at 37 °C in a CO₂ incubator. The virus-plasma mixtures were then added to Vero E6 cells seeded in a 24-well plate with 1x10⁵ cells/well and incubated for 1 h at 37 °C. Overlay media composed of 2 % fetal bovine serum (FBS) in DMEM and 1 % agar was added to the E6 cells and incubated for 3 days in 37 °C in a CO₂ incubator. After a plaque was formed, 10 % formaldehyde solution and 0.33 % neutral red in phosphate buffered saline (PBS) were used to fix and stain the cell culture layer. Neutralizing dilution of each sample was determined by identifying the well with the highest plasma dilution without an observable cytopathic effect. The 50 % neutralization doses (ND50) was expressed as the reciprocal dilution of plasma, which resulted in a 50 % reduction in plaque numbers compared to the positive virus control. The Spearman–Kärber method was used to calculate the ND50 titers. By a comparison to WHO international standard sera (NIBSC 20–136; nAb titer: 1000 IU/mL), nAb titers were converted into international units per milliliter (IU/mL).

2.5. T-cell assays

Peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation using Leucosep tubes (Greiner Bio-One) containing lymphocyte separation medium. After centrifugation at 800g for 15 min with the brake off, the enriched cell fraction consisting of PBMCs was harvested and washed with 20 mL of PBS. After centrifugation at 300g for 10 min, the cell pellet was resuspended in 10 mL of PBS for counting. After centrifugation at 300g for 10 min, cells were cryopreserved in FBS with 10 % dimethyl sulfoxide at − 180 °C until use.

For intracellular cytokine staining (ICS) of IFN-γ, interleukin-2 (IL-2), and tumor necrosis factor (TNF), we used peptide pools of 15-mer peptides overlapping by 11 amino acids covering the immunodominant parts of SARS-CoV-2 Wuhan S protein (aa 304–338, 421–475, 492–519, 683–707, 741–770, 785–802, and 885–1273; Miltenyi Biotec). To compare T-cell responses against the Wuhan and Delta strains, we used two different sets of S peptide pools comprising non-conserved regions of S sequences between the Wuhan and Delta (B.1.617.2 lineage) strains (Miltenyi Biotec). After thawing PBMCs and resting them overnight, 1 × 10⁶ PBMCs were stimulated with peptide pools at a final concentration of 1 µg/mL and anti-human CD28/CD49d (1 µg/mL; BD Biosciences) for 1 h in a 96-well plate in RPMI-1640 media containing 1 % penicillin–streptomycin, 2 mM l-glutamine, and 10 % FBS. PBMC culture was incubated for another 5  h after the addition of brefeldin A and monensin (BD Biosciences) at 37 °C in 5 % CO₂.

After the stimulation, cells were washed with PBS and stained with Live/Dead-Aqua (Invitrogen) for 20 min at room temperature, followed by a cocktail of surface Abs for 20 min at room temperature, including anti-CD14-BV510, anti-CD19-BV510, anti-CD4-BV650, anti-CD3-BV786, and anti-CD8-APC-Cy7 (BD Bioscience). Cells were fixed and permeabilized using a Foxp3/Transcription Factor Staining Buffer Set (eBioscience) and then stained with a cocktail of intracellular Abs, including anti-TNF-FITC, anti-IL-2-PE (eBioscience), and anti-IFN-γ-APC (BD Biosciences) for 20 min at room temperature. Upon completion of staining, cells were analyzed on an LSR II instrument with FACSDiva (BD Biosciences) and the data analyzed in FlowJo software (FlowJo LLC).

2.6. Statistical analysis

Data plotted in linear scale were expressed as mean ± standard deviation. Data plotted in logarithmic scales were expressed as geometric mean ± geometric standard deviation. Mann–Whitney U or Wilcoxon tests were applied for unpaired or paired comparisons, respectively. Statistical analyses were 2-tailed, with P < 0.05 considered significant. All analyses were performed in R studio or GraphPad Prism v9.0.

3. Results

3.1. Kinetics of immune responses elicited by different vaccination regimens

First, we examined anti-S IgG titers following vaccination with different regimens. As expected, the first dose significantly increased anti-S IgG titers (T2), and the second dose further significantly increased anti-S IgG titers in the homologous BNT162b2/BNT162b2 (T3) and AZD1222/AZD1222 vaccination (T4) and heterologous AZD1222/BNT162b2 vaccination (T4) groups (Fig. 2 A). In the prior infection/vaccination group, the first dose maximally increased the anti-S IgG titers (T3), but the second dose did not further increase the anti-S IgG titers (T4).

Fig. 2.

Antibody and T-cell responses elicited by diverse prime-boost regimens. Longitudinal immune responses according to the sampling schedule from pre-vaccination to 14 weeks post-first vaccination. (A) SARS-CoV-2 spike-specific binding antibodies were measured in BNT162b2/BNT162b2 (n = 22), AZD1222/AZD1222 (n = 20), AZD1222/BNT162b2 (n = 9), and prior infection/vaccination (n = 6) groups. (B) SARS-CoV-2 spike-specific CD4⁺T-cell responses were measured in BNT162b2/BNT162b2 (n = 19), AZD1222/AZD1222 (n = 18), AZD1222/BNT162b2 (n = 9), and prior infection/vaccination (n = 7) groups. Data are presented as geometric mean ± geometric SD (A) or mean ± SD (B). Statistical analyses between time points within each group were carried out using the Wilcoxon signed-rank test. ns, not significant; *P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

We also examined S-specific CD4⁺ T-cell responses following direct ex vivo stimulation with peptide pools covering the SARS-CoV-2 Wuhan wild-type S protein sequence. Based on ICS, the first dose significantly increased the frequency of IFN-γ-producing cells among CD4⁺ T cells, and the second dose further significantly increased the frequency in the BNT162b2/BNT162b2 (T3) and AZD1222/BNT162b2 vaccination (T4) groups (Fig. 2B). In the AZD1222/AZD1222 and prior infection/vaccination groups, the first dose significantly increased the frequency of IFN-γ⁺ cells among CD4⁺ T cells, but the second dose did not further increase the frequency.

3.2. Comparison of Ab titers among different regimens

Next, we compared the peak titers of binding Ab among the different vaccination regimens. As peak titers, we selected anti-S IgG titers after the second dose in the BNT162b2/BNT162b2, AZD1222/AZD1222, and AZD1222/BNT162b2 vaccination groups and after the first dose in the prior infection/vaccination group. Peak titers of anti-S IgG were significantly higher in the AZD1222/BNT162b2 vaccination group than in the BNT162b2/BNT162b2 or AZD1222/AZD1222 vaccination groups (Fig. 3 A). Similarly, the peak titers of anti-S IgG were significantly higher in the prior infection/vaccination group than in the BNT162b2/BNT162b2 or AZD1222/AZD1222 vaccination groups. There was no difference in anti-S IgG titers between the AZD1222/BNT162b2 and prior infection/vaccination groups.

Fig. 3.

Humoral immune responses at peak time points. (A) SARS-CoV-2 spike-specific binding antibodies were compared among BNT162b2/BNT162b2 (n = 22), AZD1222/AZD1222 (n = 20), AZD1222/BNT162b2 (n = 9), and prior infection/vaccination (n = 6) groups at peak time points. (B) Neutralizing antibodies were compared among BNT162b2/BNT162b2 (n = 22), AZD1222/AZD1222 (n = 20), AZD1222/BNT162b2 (n = 9), and prior infection/vaccination (n = 11) groups at peak time points. Peak time points were selected as follows: after the second dose in the BNT162b2/BNT162b2, AZD1222/AZD1222, and AZD1222/BNT162b2 vaccination groups and after the first dose in the prior infection/vaccination group. Data are presented as geometric mean ± geometric SD and analyzed for significance using the Mann-Whitney test. (C and D) The percentages of participants with varying levels of humoral immune responses. Quantification of the type of responder (strong, moderate, and none/weak) is expressed as a fraction of the number of vaccine recipients in each group. Types of responders were determined by the percentile score calculated from all vaccinated participants. ns, not significant; *P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

We also evaluated nAb titers at peak time points: after the second dose in the BNT162b2/BNT162b2, AZD1222/AZD1222, and AZD1222/BNT162b2 vaccination groups and after the first dose in the prior infection/vaccination group. Remarkably, nAb titers were significantly higher in the AZD1222/BNT162b2 vaccination group than in the BNT162b2/BNT162b2 or AZD1222/AZD1222 vaccination groups (Fig. 3B). In addition, nAb titers were significantly higher in the prior infection/vaccination group than in the BNT162b2/BNT162b2 or AZD1222/AZD1222 vaccination groups. There was no difference in nAb titers between the AZD1222/BNT162b2 and prior infection/vaccination groups.

The level of humoral response elicited by diverse vaccination regimens was categorized into strong, moderate, and none/weak responders based on the percentile ranking. The proportion of strong responders was highest in the prior infection/vaccination group, followed by the AZD1222/BNT162b2 vaccination group with respect to the analysis of anti-S IgG (Fig. 3C) and nAb (Fig. 3D). Taken together, these results indicate that heterologous priming and boosting, including both AZD1222/BNT162b2 vaccination and prior infection/vaccination, is more efficient in the generation of humoral responses than homologous priming and boosting.

3.3. Comparison of S-specific CD4⁺T-cell responses among different regimens

We also compared S-specific CD4⁺ T-cell responses among the different vaccination regimens at peak time points: after the second dose in the BNT162b2/BNT162b2, AZD1222/AZD1222, and AZD1222/BNT162b2 vaccination groups and after the first dose in the prior infection/vaccination group. Unlike anti-S IgG and nAb titers, the frequency of IFN-γ⁺ cells among CD4⁺ T cells was highest in the BNT162b2/BNT162b2 vaccination group (Fig. 4 A). The proportion of strong responders was also highest in the BNT162b2/BNT162b2 vaccination group, followed by the prior infection/vaccination group (Fig. 4B). These results show that immunization with the second homologous boost of BNT162b2 more efficiently elicits memory T-cell responses than the second heterologous boost vaccination or the first dose vaccination following natural SARS-CoV-2 infection. On the contrary, heterologous prime-boost regimens more likely to develop higher levels of humoral immune responses.

Fig. 4.

Cellular immune responses at peak time points. (A) The frequencies of spike-specific IFN-γ⁺ cells among CD4⁺ T cells were compared among BNT162b2/BNT162b2 (n = 19), AZD1222/AZD1222 (n = 18), AZD1222/BNT162b2 (n = 9), and prior infection/vaccination (n = 7) groups at peak time points. Peak time points were selected as follows: after the second dose in the BNT162b2/BNT162b2, AZD1222/AZD1222, and AZD1222/BNT162b2 vaccination groups and after the first dose in the prior infection/vaccination group. (B) The percentages of participants with varying levels of cellular immune responses. Quantification of the type of responder (strong, moderate, and none/weak) is expressed as a fraction of the number of vaccine recipients in each group. Types of responders were determined by the percentile score calculated from all vaccinated participants. (C) The frequencies of polyfunctional spike-specific CD4⁺ T cells producing any combination of IFN-γ, IL-2, or TNF in each group of vaccinated participants. (D) The fraction of spike-specific CD4⁺ T cells positive for a given number of cytokines. Data are presented as mean ± SD. Significance was analyzed using the Mann-Whitney test (A and C). ns, not significant; *P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

We also analyzed the relative frequency of polyfunctional CD4⁺ T cells that simultaneously produce multiple cytokines (i.e., IFN-γ, IL-2, and/or TNF). The BNT162b2/BNT162b2 vaccination group exhibited significantly higher frequencies of polyfunctional cells producing ≥ 2 cytokines among S-specific CD4⁺ T cells than the AZD1222/AZD1222 or AZD1222/BNT162b2 vaccination groups (Fig. 4C). There was no difference in the frequency between the BNT162b2/BNT162b2 vaccination and prior infection/vaccination groups. The frequency of polyfunctional cells was significantly higher in the AZD1222/BNT162b2 vaccination group than in the AZD1222/AZD1222 vaccination group. Pie graphs representing single-, double-, and triple-positive CD4⁺ T cells in terms of cytokine production responding to S overlapping peptide pools also indicated a relatively high frequency of double- or triple-positive cells in the BNT162b2/BNT162b2 vaccination and prior infection/vaccination groups (Fig. 4D).

3.4. Immune responses against the Delta variant among different regimens

Finally, we analyzed immune responses against the Delta variant. In this analysis, we included convalescent individuals who recovered from SARS-CoV-2 Delta infection for the purpose of comparison. Delta variant-specific nAb titers were significantly higher in the prior infection/vaccination group than in the BNT162b2/BNT162b2, AZD1222/AZD1222, or AZD1222/BNT162b2 vaccination groups, and were even higher than those in the Delta convalescent individuals (Fig. 5 A). We also examined CD4⁺ T-cell responses against the S protein of the Delta variant. In this analysis, we performed ICS of IFN-γ following direct ex vivo stimulation using two different sets of S peptide pools comprising non-conserved regions of S sequences between the Wuhan and Delta strains and calculated the ratio of the frequency of Delta S-specific IFN-γ⁺ cells to the frequency of Wuhan S-specific IFN-γ⁺ cells. As expected, the ratio was highest in the Delta convalescent individuals. Intriguingly, the ratio was higher in the prior infection/vaccination group than in the BNT162b2/BNT162b2 vaccination group (Fig. 5B), though Wuhan S-specific CD4⁺ T-cell responses were highest in the BNT162b2/BNT162b2 vaccination group (Fig. 4A). Taken together, these results indicate that nAb and CD4⁺ T-cell responses that cross-recognize the Delta variant are more efficiently generated by prior infection/vaccination than the homologous or heterologous vaccination regimens.

Fig. 5.

Cross-variant immune responses against the Delta elicited by diverse prime-boost regimens. (A) Plasma samples were analyzed by plaque-reduction neutralization assays for the SARS-CoV-2 Delta variant in BNT162b2/BNT162b2 (n = 22), AZD1222/AZD1222 (n = 20), AZD1222/BNT162b2 (n = 9), prior infection/vaccination (n = 11), and Delta-convalescent (n = 17) groups. Delta-neutralizing antibodies were compared among groups at following time points: after the second dose in the BNT162b2/BNT162b2, AZD1222/AZD1222, and AZD1222/BNT162b2 vaccination groups and after the first dose in the prior infection/vaccination group. (B) PBMCs were analyzed by intracellular staining for IFN-γ following direct ex vivo stimulation using two different sets of S peptide pools comprising non-conserved regions of spike sequences between the Wuhan and Delta strains in BNT162b2/BNT162b2 (n = 19), AZD1222/AZD1222 (n = 18), AZD1222/BNT162b2 (n = 9), prior infection/vaccination (n = 7), and Delta-convalescent (n = 17) groups. The ratio of the frequency of IFN-γ-producing CD4⁺ T cells against the Delta spike to the frequency of IFN-γ-producing CD4⁺ T cells against the Wuhan spike was calculated in each group. Data are presented as the geometric mean ± geometric SD (A) or mean ± SD (B) and analyzed for significance using the Mann-Whitney test. ns, not significant; *P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

4. Discussion

The currently authorized mRNA vaccines and adenovirus vector vaccines have shown remarkable efficacy in preventing COVID-19. However, heterologous prime-boost vaccination has been considered to improve immunogenicity and mitigate intermittent supply shortages [13]. In addition, safety issues have been raised with the adenovirus vector vaccines after recognition of severe adverse events, including life-threatening venous thrombosis associated with AZD1222 vaccination, especially in younger women [[14], [15]]. These have led to heterologous mRNA booster immunization for those who had already received the first dose of AZD1222. Although no efficacy data were available, many national health authorities adopted heterologous regimens of authorized COVID-19 vaccines. Thus, a comparative study of humoral and cellular immune responses elicited by diverse prime-boost vaccination regimens is informative and aids in better understanding the real-word challenges.

In the present study, we demonstrated differential induction of humoral and cellular immunity by heterologous prime-boost vaccinations (Supplementary Table 4). In stark contrast to both S-specific binding Ab levels and neutralization titers, which were highest in the AZD1222/BNT162b2 vaccination group, S-specific CD4⁺ T-cell responses were highest in the BNT162b2/BNT162b2 vaccination group, not in the AZD1222/BNT162b2 vaccination group. The relatively low levels of S-specific CD4⁺ T cell responses in AZD1222/BNT162b2-vaccinated participants were unexpected and may still put these individuals at risk of severe SARS-CoV-2 infection. Our findings differ from those reported by Schmidt and colleagues, who suggested that, upon heterologous BNT162b2 boosting, nAb titers and T-cell responses were significantly increased in AZD1222-primed individuals [7]. However, the absolute frequency of CD69⁺IFN-γ⁺ cells among CD4⁺ T cells was only marginally increased, and this study precluded in-depth analyses of the polyfunctionality of S-specific T-cell responses.

For individuals who were previously infected, the first vaccine dose may act as a booster immunization and potently induced humoral and cellular responses [[16], [17]]. As the nAb titer is presumed to be a key correlate of protection against SARS-CoV-2 infection [18], our results fit well with the recent finding of a beneficial boost to infection-acquired immunity, with an adjusted vaccine effectiveness of > 90 % after the first dose in vaccine recipients who had previous SARS-CoV-2 infection [19]. Furthermore, despite the considerable escape of nAb responses against the Delta variant among study participants, we observed the highest levels of nAb cross-recognizing the Delta variant in the group of participants with prior wild-type SARS-CoV-2 infection and vaccination. However, in line with the heterologous AZD1222/BNT162b2 vaccination group, those with prior infection who were vaccinated had a weaker S-specific CD4⁺ T-cell response than individuals with homologous BNT162b2/BNT162b2 vaccination without prior infection.

Notably, our results show that two doses of BNT162b2 leads to reliable and robust induction of S-specific CD4⁺  T-cell responses, significantly higher than other prime-boost vaccination regimens. These findings are consistent with a recent report on four different COVID-19 vaccines [20], while our analysis included heterologous prime-boost regimens and hybrid immunity from prior infection and vaccination. Accordingly, mRNA vaccine platforms might be selected for their ability to induce robust memory CD4⁺ T cell responses.As the incidence of breakthrough infection after vaccination increases worldwide, there is increasing interest in immunity protecting from severe COVID-19. As new SARS-CoV-2 variants continue to emerge and evade humoral immune defenses, cellular immunity, which targets a broad spectrum of SARS-CoV-2 epitopes recognized by T cells, may provide stable protective immunity [21], though determinants of cellular responses in vaccine-mediated protection are not yet accurately defined. Therefore, whether the increased humoral immunogenicity of heterologous AZD1222/BNT162b2 vaccination consistently translates into improved protection remains unclear, and this requires further study. In addition, whether people who have developed poor T-cell responses after vaccination, such as those primed with AZD1222, may benefit from T-cell vaccine formulations needs to be investigated [22].

Both infection-induced and vaccine-induced immunity naturally wane over time; however, greater protection appears to be conferred to persons who had been previously infected with SARS-CoV-2 [23]. In vaccinated individuals, breakthrough infection with the Omicron variant induces smaller nAb responses than infection with the Delta variant [24], and relatively little is known about the potential consequences of hybrid immunity conferred by infection and vaccination with different strains. Nonetheless, the presence of polyfunctional S-specific T cells in previously infected and vaccinated participants in our study is consistent with a slower decline in protection against severe COVID-19 in this population and may contribute to modifying disease severity following reinfection with SARS-CoV-2 variants [[25], [26]].

The current study has several limitations. First, the small sample sizes in the cohort makes robust statistical analysis challenging; therefore, our findings will need to be further validated in larger cohorts. Second, we were not able to test samples at later time points to determine the durability and waning of humoral and cellular immune responses. Third, our results may underestimate the impact of subdominant T-cell epitopes outside the dominant epitopes. Fourth, we only studied circulating lymphocytes and future studies of draining lymph nodes after vaccination may be required to understand the mechanisms underlying the differential responses observed here.

In summary, the current study with direct, age-matched comparisons sheds light on the discordant humoral and cellular immunogenicity of heterologous priming and boosting in the context of SARS-CoV-2 antigen exposures. Our findings highlight how individual immunological experience shapes the subsequent immune response elicited by COVID-19 vaccination. Additional real-world effectiveness data and laboratory investigations will further inform the most appropriate vaccination strategy for protection from emerging SARS-CoV-2 variants, with implications for the development of next-generation vaccine platforms.

Funding statement

This work was supported by the National Institute of Infectious Diseases, National Institute of Health, Korea Disease Control and Prevention Agency (#2021-ER1902–0 and #2021-ER2601-00), a grant from the Ministry of Health & Welfare, Republic of Korea (Grant No. HI14C1324), the 2020 Joint Research Project of Institutes of Science and Technology (J.Y.C) and the Institute for Basic Science (IBS), Republic of Korea, under project code IBS-R801-D2 (E.-C.S.).

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

The following are the Supplementary data to this article:

Data availability

Data will be made available on request.