Participants were selected based on age (<75 years) and preceding prime vaccination with either a mRNA-based or vector-based SARS-CoV-2 vaccine and planned boost vaccination. The control group was selected according to a homologous vaccination scheme whereas the study group was selected according to a heterologous vaccination scheme.

The study was designed to elucidate the humoral and cellular immunity against SARS-CoV-2 of a heterologous prime-boost and prime-boost-boost vaccination scheme. The study was authorized by the local ethics committee of the Ruhr-University Bochum (21-7260 and 20-6886). After written informed consent, the collected samples were included in this study. For this, whole blood was collected and PBMC and plasma isolated according to previously published studies (34). Initially, we recruited persons with one vaccination (ChAdOx1, vector (V) n=32; BNT162b2, mRNA (M) n=25), which was followed by a homologous prime-boost vaccination (VV n=7; MM n=25) or a heterologous prime-boost vaccination (T2; VM n=24). Samples were collected before another mRNA vaccination (VV n=4; VM n=10; MM n=11) and after another mRNA vaccination (VVM n=5; VMM n=18; MMM n=18). The study consists of four sample collection timepoints: prior and post prime-boost vaccination (T1, T2) and prior and post prime-boost-boost vaccination (T3, T4) ( Figure 1A ).

Serum was collected from whole blood samples by centrifugation of monovettes at 1500 g for 10 min and stored at -20 °C. SARS-CoV-2 neutralizing antibodies were determined using propagation-defective vesicular stomatitis virus (VSV) pseudotypes and the virus neutalization assay was performed as previously described (35). Briefly, BHK-G43 cells were treated with mifepristone to induce the expression of the VSV Glycoprotein (G) on the surface. Afterwards, cells were incubated with VSV*ΔG(FLuc) and trans-complemented with the G protein (VSV*ΔG(FLuc)+G). The virus particles were used to produce pseudotype virus complemented with the SARS-CoV-2 Spike (S) protein. Therefore, HEK 293T cells were transfected with the appropriate SARS-CoV-2 S expression plasmid for the wildtype (YP_009724390.1) and the VOCs Delta (EPI_ISL_1921353) and Omicron (BA1: EPI_ISL_6640919; BA4/5: EPI_ISL_11550739/EPI_ISL_12029894). After inoculation of the transfected cells with VSV*ΔG(FLuc)+G, the virus particles harbor the respective Spike protein on their surface. To determine the pseudotype virus neutralization (PVN), patient sera were inactivated at 56 °C for 30 min. Sera were pre-diluted 1:10 and 120 µl per donor were pipetted into the first row of a 96 well cell culture plate in triplicates. A twofold dilution was performed and pseudotype virus was added to each well, followed by incubation for 1 h at 37 °C. The suspension was transferred to Vero E6 cells, previously seeded in a density of 1x10⁵ cells per mL in a 96 well plate and incubated overnight. On the next day, the supernatant was aspirated and the cells were lysed with 35 µL luciferase lysis buffer per well. After freezing to the core and thawing, 20 µL of the lysate were pipetted into microtiter plates and luminescence was measured with a plate luminometer. Finally, the antibody dilution resulting in a luminescence reduction of 50%, representing 50% PVN, was calculated (PVND₅₀; lower limit of detection: 20 PVND₅₀; upper limit of detection: 2560 PVND₅₀).

Binding antibody units (BAU/mL) against the receptor binding domain (RBD) of the SARS-CoV-2 Spike from the Wuhan strain amongst the groups were measured as previously described (36). Briefly, 96 well plates were coated with recombinant RBD of SARS-CoV-2 WT spike overnight at 4°C and subsequently blocked with ELISA Diluent (Biolegend) for 1 h at 37 °C. Serum samples were serially diluted, including a negative serum control, anti-S antibody as positive control (Dianova, CSB-RA332450A0GMY), and a calibrator. Pre-diluted samples and controls were incubated on the coated plate for 1 h at 37 °C. After washing, HRP conjugated secondary antibody (goat anti human IgG Fc gamma fragment specific, Dianova) was added and incubated for 1 h at 37 °C. After washing, the plate was tapped dry and incubated with substrate (1 Step Ultra TMB, Pierce) for 5-10 min at RT in the dark until the positive control showed distinct blue staining. The reaction was stopped with 2 M H₂SO₄ and absorbance was measured at 450 nm. Normalization was performed by: (sample-negative control)/(calibrator-negative control). The sample dilution was used to calculate sample BAU/mL by fitting hyperbolic curves in GraphPad Prism using the correction factor of the WHO Standard 20/136 measurements, which is defined as 1000 BAU/mL.

Peripheral blood mononuclear cells (PBMC) were isolated from 30 mL blood per donor using standard Ficoll Hypaque density gradient technique as described previously (37) and frozen at a density of 10x10⁶ cells/mL. For flow cytometric analysis of different T cell subpopulations, 2.5 x 10⁶ PBMCs were stained with the viability dye LIVE/DEAD™ Fixable Blue Dead Cell Stain Kit (L23105, Thermo Fisher) for 30 min at 4°C. Subsequently, PBMCs were incubated with the Fc receptor blocking solution Human TruStain FcX™ (422302, Biolegend) for 15 min at 4°C. Afterwards, surface markers were stained for 15 min at 4°C. Fixation, permeabilization and Foxp3 staining was performed using the Foxp3 staining buffer set (130-093-142, Miltenyi Biotec) according to manufacturer’s recommendations. Further information about antibodies is provided in Supplementary Table S1 . Samples were measured in a Cytoflex LX (Beckman Coulter) and 1.5-2 x 10⁶ events per sample were acquired. Furthermore, SARS-CoV-2 reactive T cells were determined as previously described (37, 38). Briefly, PBMCs were stimulated in the presence of overlapping peptide pools of the WT and Omicron BA.1 Spike SARS-CoV-2 (JPT Peptide Technologies) for 16 h. Brefeldin A (1 μg/mL, Sigma-Aldrich) was added after 2 h. An unstimulated sample served as negative control and stimulation with staphylococcal enterotoxin B (1 μg/mL, Sigma-Aldrich) as positive control. After stimulation, the cells were harvested and stained with optimal concentrations of antibodies ( Supplementary Table S2 ) for 10 min at room temperature in the dark. Stained cells were washed twice with PBS/BSA before preparation for intracellular staining using the Intracellular Fixation & Permeabilization Buffer Set (Thermo Fisher Scientific) according to the manufacturer’s instructions. Fixed and permeabilized cells were stained for 30 min at room temperature in the dark with an optimal dilution of antibodies against the intracellular antigen. Stained samples were acquired immediately on a CytoFLEX flow cytometer (Beckman Coulter), collecting at least 25,000 CD3⁺ events. Quality control was performed daily using the recommended CytoFLEX daily QC fluorospheres (Beckman Coulter). No modification to the compensation matrices was required throughout the study. Antigen-reactive responses were considered positive after the non-reactive background was subtracted, and more than 0.01% were detectable. Negative values were set to zero.

Statistical analysis was performed using Tukey’s multiple comparison test to compare the vaccine regimens with each other and Šidák’s multiple comparison test to compare WT vs. Omicron for each vaccine regimen (* p=0.05; ǂ p=0.01; ß p=0.001, # p=0.0001) in GraphPad Prism 9.4.1. Flow cytometry data were analyzed with FlowJo™ (Becton Dickinson & Company, version 10.8.0 for unstimulated T cells and version 10.7.1 for stimulated T cells). The gating strategies and representative dot plots are shown in Supplementary Figure S1 . Binding antibody units were calculated in GraphPad Prism 9.

Since COVID-19 vaccination is available and different vaccination strategies have been applied, the resulting immune response in different healthy and diseased cohorts has since been investigated (34, 39–43). However, further in-depth analysis of the humoral and cellular immune response can add to the existing knowledge, especially in light of the currently dominant Omicron variant. Thus, our study was designed to assess the immune responses of a cohort upon different vaccine regimens after a second dose (prime-boost) and third dose (prime-boost-boost). In total, 57 participants were recruited (65% male, 35% female), with a mean age of 35 ± 12 years. Subjects were general healthy without any evidence of immune deficiency or chronic diseases. Subjects were assigned into three groups, based on the vaccine strategy ( Figure 1A ). Initially, participants have received prime vaccination with the mRNA vaccine COMIRNATY (BNT162b2, Pfizer-BioNTech, here referred to as “M”) or the vector-based vaccine Vaxzevria (ChAdOx1-S, AstraZeneca, here referred to as “V”) prior to participation in the study and did not report previous SARS-CoV-2 infection. For prime-boost vaccination (second dose), the participants either received a homologous boost of the initial vaccine, or a heterologous boost. All of the participants were further vaccinated with COMIRNATY as prime-boost-boost (third dose), except one participant, who received Spikevax (mRNA-1273, Moderna). Accordingly, our cohort is divided into the three groups VVM, VMM and MMM.

Since the beginning of 2022, Omicron (B.1.1.529) is the dominating variant in Europe (44). To elucidate the humoral immune response after vaccination against Omicron in comparison to the ancestral SARS-CoV-2 wildtype (WT) strain, we analyzed binding and neutralizing antibody levels. Binding antibodies were determined as BAU/mL by ELISA (45). All three groups showed increased levels of anti-SARS-CoV-2-Spike binding antibodies 21d-28d after prime-boost (T2) vaccination that decreased within five to seven months post prime-boost (T3) ( Figure 1B ), which is in line with previous studies (9, 46, 47). Interestingly, homologous MM and heterologous VM vaccinated study participants displayed on median significantly higher binding antibody levels compared to VV vaccinated participants after the first boost (median 3451 BAU/mL [MM] and 1704 BAU/mL [VM] vs. 211 BAU/mL [VV]). In addition, MMM vacinees showed significantly higher binding antibody levels compared to VMM after the second boost (median 12856.5 BAU/mL vs. 6212.5 BAU/mL) ( Figure 1B ). Next, we determined neutralizing antibodies against a replication vesicular stomatitis virus pseudotype, which is known to correlate with the classical neutralization against live virus (35). Similar to binding antibody levels, measurement of pseudo virus neutralizing antibody titers (PVND₅₀) against the SARS-CoV-2 WT revealed significantly higher neutralizing antibodies in MM vacinees compared to VV vaccinated study participants on T2, as well as for the MMM group in comparison to the VMM group on T4 ( Figure 1C ; Supplementary Table S3 ). Moreover, the three analyzed vaccine regimens displayed reduced neutralizing capacity against Omicron BA.4/5 compared to the Delta variant at T4. Most importantly, neutralizing capacities against BA.1 were significantly reduced in VM and MM on T2. While neutralization was also significantly reduced in VVM, neutralization capacity against BA.1 and BA.4/5 in the VMM and MMM groups was comparable to SARS-CoV-2 WT at T4 ( Figure 1C ; Supplementary Table S3 ). Furthermore, B cell and T cell populations were analyzed after prime-boost (T2) and prime-boost-boost (T4) vaccination by flow cytometry. The total PBMC population showed constant CD19⁺ B cell levels crucial for antibody secretion on T4 compared to T2 for the vector-primed and -boosted group (VVM), whereas the B cell population was reduced for the vector-primed and mRNA-boosted group (VMM) and the homologous mRNA group (MMM). Interestingly, the proportion of CD3⁺ T cells was reduced on T4 for all vaccine regimens ( Figure 1D ). To conclude, humoral immune responses were induced after first and second booster vaccination independent of the vaccine regimen.

T cells are crucial components for effective cellular immunity. They are distinguished into two major subtypes: CD4⁺ T helper cells (T_H) and CD8⁺ cytotoxic T cells (T_C). Both subtypes can develop from a naïve, i.e. antigen-inexperienced state, into memory cells that are antigen-experienced and, upon antigen re-encounter, respond with a faster and stronger immune response. Furthermore, follicular T helper cells (T_FH) and regulatory T cells (T_REG), both CD4⁺ T cell subsets, are essential for B cell-derived high-affinity antibodies and suppression of over-shooting immune responses, respectively (48–50). To study T cell responses, several T cell subsets were analyzed in whole blood PBMC using flow cytometry and gated accordingly ( Supplementary Figure S1 ). First, levels of CD4⁺ cells appeared similar amongst the different groups on T2 and T4, however, the vector primed and boosted group showed a slightly elevated median ( Figure 2A ). Remarkably, the follicular helper T cells (T_FH) population, characterized by CXCR5 and PD-1 expression, decreased after the second boost compared to the first boost ( Figure 2B ). Moreover, levels of CD4⁺ T cells, which highly express PD-1, a marker of T cell activation and exhaustion (T_EX) were similar amongst the groups ( Figure 2C ). Regulatory T cells can be subdivided into naïve (naïve T_REG), non-suppressive (ns T_REG) and effector T_REG (e T_REG) ( Figures 2D–F ). Interestingly, while levels of naïve and non-suppressive T_REG were comparable among the vaccine regimens on T2 and reduced in the MMM group compared to the VMM group on T4 ( Figures 2D, E ), significantly higher amounts of effector T_REG were detected on T4 in the homologous mRNA vaccinated (MMM) group compared to the VMM group ( Figure 2F ). In terms of CD8⁺ cells, levels were reduced in the vector primed and boosted group (VVM) on T4, reaching statistical significance in comparison to the heterologous VMM group ( Figure 2G ), only a very small percentage could be identified as degranulated CD107a⁺ cytotoxic T cells (T_DEG) ( Figure 2H ). No detectable difference was observed in levels of PD-1⁺ exhausted cytotoxic T cells (T_EX) ( Figure 2I ).

To specifically analyze SARS-CoV-2-reactive CD4⁺ and CD8⁺ T cells, PBMCs were stimulated with an overlapping peptide pool (OPP) of the SARS-CoV-2 Spike (S) protein. As the original vaccination was developed against the Wuhan SARS-CoV-2 strain (Wuhan-Hu-1, USA_WA1/2020 (51)), the cellular immune reaction against VOCs is of particular interest. To identify possible differences in cellular responses, WT and Omicron BA.1 Spike peptides were selected for stimulation, with Staphylococcal Enterotoxin B serving as positive control. Amongst the different vaccination groups, the proportion of activated CD4⁺ cells was similar for WT and Omicron stimulated PBMCs, however, the activation markers CD154 and CD137 were more abundant upon WT Spike stimulation ( Figures 2J, K ). Moreover, the VMM group displayed significantly decreased levels of activated T cells upon WT Spike stimulation ( Figure 2K ). Comparably, the proportion of CD8⁺ cells was similar amongst the vaccination groups, with lower levels of the activation markers CD137 and CD69 in Omicron Spike stimulated cells ( Figures 2L, M ).

The activation markers CD154 and CD137, or CD137 and CD69 are co-expressed with cytokines in SARS-CoV-2-reactive CD4⁺ or CD8⁺ T cells, respectively. Thus, the production of interferon γ (IFNγ), interleukin 2 (IL-2), tumor necrosis factor α (TNFα) or granzyme B (GrzB) and combinations thereof were measured in activated T cells upon Spike stimulation using flow cytometry and gated accordingly ( Figure 3 ; Supplementary Figure S1 ). Interestingly, WT stimulation resulted in a higher percentage of CD4⁺ TNFα⁺/TNFα⁺IL-2⁺ T cells in the homologous MMM vaccine regimen compared to the two heterologous vaccine regimens VVM and VMM, or VMM, respectively ( Figure 3A ). Similarly, Omicron stimulation resulted in a higher percentage of CD4⁺ TNFα⁺/TNFα⁺IL-2⁺/IL-2⁺ T cells in the homologous MMM vaccine regimen compared to VMM and VMM (for TNFα⁺) or VMM (for TNFα⁺IL-2⁺/IL-2⁺) ( Figure 3B ). Notably, this elevated cytokine response upon homologous vaccination complements the observation of increased effector T_REG cells in this group to prevent excessive immune reactions compared to the heterologous vaccinations ( Figure 2F ). For the protease GrzB and the cytokine IFNγ, no difference was observed in the different study groups. Similarly, the cytokine production in activated CD8⁺ cells was comparable amongst the vaccine regimens after WT and Omicron Spike stimulation, solely the VMM group displayed significantly higher levels of GrzB⁺ T cells, suggesting increased cytotoxicity of CD8⁺ T cells ( Figures 3C, D ).

Responding T cell subsets, including naïve CD4⁺ T cells (T_naïve), CD4⁺ effector memory T cells (T_EM), CD4⁺ central memory T cells (T_CM), and the subset of T effector memory, re‐expressing CD45RA (T_EMRA), which reside in secondary lymphoid organs (T_CM), circulate through the blood stream (T_EM) or exhibit a terminally differentiated phenotype (T_EMRA), were additionally analyzed in this study ( Figure 4 ). Interestingly, CD4⁺ central memory T cells were most abundant amongst the memory cells, with higher frequencies in the VVM vaccine regimen compared to the other groups on T2 and T4, whereas the frequencies of the other responding T cell subsets were comparable among the groups ( Figure 4A ). In contrast, in the CD8⁺ memory cell population, T_EM cells were most abundant and the frequency in the VVM vaccine regimen was lower compared to the other groups on T2 and T4. Overall however, the T_naïve was the most abundant CD8⁺ T cell subset in VV and MM vaccinees on T2 and VVM vaccinees on T4 ( Figure 4B ). Moreover, after stimulation with S-protein OPP, T cell subsets were analyzed after the second boost, where activated CD4⁺ T_CM cells were higher upon WT stimulation compared to Omicron, however, no significant difference was detected between the vaccine regimens except for the VVM group ( Figure 4C ). For CD8⁺ naïve T cells, the VVM group reached significantly higher levels compared to the other groups upon WT stimulation, also compared to the respective Omicron stimulated VVM group. Strikingly, CD8⁺ T_EM cells were higher upon Omicron stimulation with the highest in the homologous mRNA vaccinated (MMM) group, whereas the population of T_EMRA cells was lower in this group ( Figure 4D ).