Vaccination against the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the key strategy to control the coronavirus disease 2019 (COVID-19) pandemic (1) and has already reduced incidences, hospitalizations, and deaths in several countries (2). Unfortunately, waning humoral immunity over time (3) and the emergence of immune evasive SARS-CoV-2 variants of concern (VOC) (4) impair vaccine effectiveness (5) and allow rebounds in infection rates (6, 7). The winter of 2021/2022 and the following summer came with the challenge of decreasing population immunity as initial vaccinations date back to early 2021 and the sudden appearance and rapid spread of the highly mutated immune evasive Omicron VOC (PANGO lineages B.1.1.529; BA.1, BA.2 and BA.3, BA.4, BA.5) (8–16). Therefore, ‘booster’ vaccinations are of enormous relevance to reestablish efficient protection (17, 18) and have been shown to induce humoral and cellular immune responses also against the Omicron VOC (9–14, 16, 19–21). ‘Boosters’ are performed as additional single vaccinations with a vaccine not necessarily matching the previous regimen. Generally, boosting triggers humoral and cellular responses. However, the degree might vary dependent on the specific combination of the initial vaccination regimen and the ‘booster’ vaccine (22). In at least 11 states of the European Union, individuals have received an initially unscheduled heterologous primary vaccination regimen consisting of a ChAdOx1 nCoV-19 (Vaxzevria, AstraZeneca) prime followed by a BNT162b2 (Comirnaty, BioNTech/Pfizer) boost after 8–12 weeks (23). This schedule had not been evaluated in clinical trials before application, but proven effective (24). The immunological responses were even superior to homologous vaccinations (25–28). However, the effect of a ‘booster’ following this regimen has not yet been described.

Here, we closely monitored the antibody titers and memory T cell immunity in a heterologously vaccinated cohort of young adults (25) over 7 months of follow-up and assessed the effect of a BNT162b2 ‘booster’. Our data show that immunity gradually declines over the course of 5.5 months but antibody and memory T cell responses are restored and increased after the ‘booster’. Responsive T cells recognized all SARS-CoV-2 variants, while the Omicron BA.1 VOC efficiently evaded neutralization by antibodies induced by initial vaccination. Strikingly, in this young cohort of heterologously vaccinated individuals where the primary vaccination had a longer interval than typically in homologous vaccinations, the ‘booster’ induced humoral immune responses that neutralized the Omicron BA.1 VOC almost as effectively as the early B.1 strain.

A previously described cohort of 26 individuals, of whom one had a history of SARS-CoV-2 infection, received a primary vaccination of a heterologous ChAdOx1 nCoV-19 prime and BNT162b2 boost within an 8-week interval in early 2021 (25). We performed a follow-up study of this young cohort (23 participants, median age 29.5 years, Table 1) for a duration of 7 months after primary vaccination and analysed humoral and cellular immunity over time, as well as reactogenicity and immune responses after a BNT162b2 ‘booster’ vaccination 6.5 months later (18 participants).

The ‘booster’ vaccination was well tolerated and associated with a lower overall reactogenicity compared to the initial two heterologous doses (25). The major solicited adverse reactions were pain at the injection site (94%, 17/18 participants), fatigue (44.4%, 8/18), and headache (33.3%, 6/18). No serious adverse events were observed (Figure 1).

As described previously (25), 2 weeks after primary vaccination, the cohort showed median cumulative anti-SARS-CoV-2-spike IgM and IgG (IgM/G) titers of 8,815 (1,206–19,046) BAU/ml, which decreased to 2,039 (235–5,926) BAU/ml over the course of 3 months. In the (slightly smaller) follow-up cohort, they further declined to 1,120 (125–3,287) BAU/ml after 5.5 months, corresponding to an eightfold decrease (Figure 2A). After 6.5 months, 18 of the participants (median 1,243 BAU/ml) received a BNT162b2 ‘booster’. Two weeks later, the median IgM/G titers had increased by 21-fold to 25,775 BAU/ml (2,092–49,627; p < 0.001, mixed model) in boosted individuals, while further decreasing to 753 (474–3,076, p = 0.0186, mixed model) BAU/ml in non-boosted participants (Figure 2A). This corresponds to a 34-fold higher median titre in the boosted versus non-boosted group at the 7-month time point (p = 0.0033, mixed model) and exceeds the initial titers determined after the primary vaccination by ~2-fold [median 11,339 BAU/ml (25)].

Using vesicular stomatitis virus (VSV)-based pseudoviruses (PVs) carrying the SARS-CoV-2 spike protein, we analysed the neutralizing activity of the sera. Two weeks after the primary vaccination, median 50% pseudovirus neutralization (PVNT50) titers against PV carrying the SARS-CoV-2 Wuhan-Hu-1 D614G (B.1) spike protein were 2,418 (350–6,383). PVNT50 titers remained stable for 1.5 months but decreased 12-fold to 204 (24–601) over the course of 5.5 months (Figure 2B, p < 0.0001, mixed model). In comparison, titers against the Delta VOC after 5.5 months were eightfold lower with median titers of 24 (<20–481) (Figures 2B, S1, p < 0.0001, mixed model). In contrast to studies on homologous vaccinations, but in line with other studies on sera from heterologously vaccinated individuals (9, 15, 37), median neutralization titers of 345 (<20–4541) were already detected against the Omicron BA.1 VOC 2 weeks after primary vaccination in 15/16 (94%) participants. These titers decayed to <20 (<20–299) after 5.5 months, with 12/22 (55%) participants showing no detectable neutralizing activity at all. This corresponds to a 7–20-fold immune evasion compared to B.1 (Figures 2B, S1, p < 0.0001, mixed model). After the ‘booster’, titers against the B.1 variant increased ninefold to 1,929 (474–4,942), 45-fold to 1,094 (51–2,895) for Delta, and >88-fold to 1,768 (<20–3,760) against BA.1 (Figures 2B, C, S1, p < 0.0001, mixed model). Strikingly, the neutralizing titers 2 weeks after the ‘booster’ against Delta and BA.1 were similar and only slightly lower than for B.1 (Figures 2B, C, S1, p = 0.9608, p = 0.0198, p = 0.0211, mixed model). At 5.5 months after primary vaccination, the neutralizing activity correlated weakly with IgM/G titers; however, after the ‘booster’ (7 months after primary vaccination), the correlation was highly significant for all variants, indicating that induction of high titers is associated with potent neutralization (Figure 2D, r ≥ 0.73, Spearman). Notably, the neutralizing titers obtained for the three variants correlated only weakly before but became strongly significant after the ‘booster’ immunization ( Figure 2E , r ≥ 0.73, Spearman). These results indicate that the ‘booster’ induces broadly neutralizing antibodies that are even effective against the highly divergent Omicron BA.1 variant. Results were not confounded by participant age or sex.

To evaluate cellular immunity, we isolated peripheral blood mononuclear cells (PBMCs) from blood samples provided by 12 participants 5.5 months after the primary vaccination, as well as samples 2 weeks after the ‘booster’. Cells were exposed to pools of 315 peptides spanning the spike sequences of SARS-CoV-2 Wuhan-Hu-1 (Wu), Delta (B.1.617.2), or Omicron (B.1.1.529.1; BA.1) and analysed for intracellular cytokines IFNγ, IL-2, and TNFα. Increased cytokine production upon peptide stimulation was evaluated to determine responsive and spike-specific CD4⁺ and CD8⁺ memory T cells (Figures S2, S3). At 5.5 months after primary vaccination, only five of the 12 donors showed remaining CD4⁺ memory T cells responding to either spike peptide stimulation by IL-2 or TNFα production, respectively (Figure 3). In contrast, most participants (10 of 12) showed remaining memory CD4⁺ T cells responding by IFNγ production (median 0.005%–0.011% reactive cells). Notably, the magnitude of responses to SARS-CoV-2 BA.1 spike did not differ from Wu or Delta (Figure 3, p > 0.05, Mann–Whitney-U), indicating efficient cross-reactivity. After the ‘booster’, spike-specific IFNγ CD4⁺ memory T cell responses and the fraction of reactive cells further increased to 11 of 12 participants and the median ranged from 0.02% to 0.04% responsive cells for the spike peptide variants (p = 0.0273, p = 0.0137, p = 0.0098; Wilcoxon signed-rank). However, CD4⁺ T cells responding by IL-2 or TNFα secretion were not affected by the ‘booster’. CD8⁺ memory T cells showed a longer durability and typically remained reactive over the course of 5.5 months, with 8 of 12 participants responding to Wu, Delta, or BA.1 spike peptide challenge by IL-2 (0.004%–0.009%), all by TNFα (0.027%–0.052%) and IFNγ production (0.055%–0.081%) (Figure 3). Again, the ‘booster’ significantly enhanced IFNγ responses for all variants (0.105%–0.208%) (Figure 3, p = 0.0005, p = 0.0049, p = 0.0137; Wilcoxon signed-rank). IL-2 and TNFα responses also showed an increase, but not significant. Stimulation with a pool of 176 peptide epitopes from 18 infectious agents (CEFX) confirmed that the ‘booster’ did not unspecifically affect T cell responses. Of note, 11 of 12 participants developed CD4⁺ and all participants CD8⁺ T cell memory against BA.1. Altogether, 5.5 months after heterologous vaccination participants showed stable CD8⁺ memory T cell levels but a decreased humoral and CD4⁺ memory T cell immunity. A BNT162b2 ‘booster’, however, reactivated and enhanced T cell immunity and induced potent antibody responses also against the Omicron BA.1 VOC.

Heterologous primary ChAdOx1 nCoV-19 prime, BNT162b2 boost vaccination induces potent immune responses against SARS-CoV-2 (25–28) resulting in effective protection from COVID-19 (24). Data about long-term immunity and protection conferred by this vaccination regimen, as well as reaction toward a ‘booster’ vaccination and its efficacy toward the Omicron VOC, are, however, scarce, in particular for younger individuals (24). T cells generally show broad cross-reactivity against SARS-CoV-2 variants (25, 38) including Omicron (39–41), which is expected, because the majority of mutations in the Omicron spike are not located in known T cell epitopes (42, 43) and because the large HLA allele diversity on population level makes T cell evasion unlikely (44). In contrast, BA.1 showed neutralization-evading properties (9–14, 20, 21) which consequently results in loss of protection from symptomatic infection (45–47). Thus, ‘booster’ vaccinations are performed aiming for enhanced immune protection especially from Omicron (9–14, 16, 19–21). ‘Booster’ vaccinations after homologous BNT162b2 or ChAdOx1 nCoV-19 vaccination regimen have been described as safe (22) and shown to restore protection from the Delta VOC (17, 18) and to reduce Omicron breakthrough infections and the secondary attack rate (48).

We here show for a young cohort with a median age of 29.5 years that heterologous primary vaccination already resulted in antibody titers with moderate Omicron BA.1-neutralizing activity, which declined over the course of 5.5 months to levels hardly neutralizing this VOC. In line, spike-specific CD4⁺ memory T cells showed remaining but limited reactivity 5.5 months after heterologous vaccination. However, CD8⁺ memory T cells remained responsive and also reacted to BA.1 spike epitopes. This is in line with the observation that individuals that received homologous primary vaccination remain partly protected from hospitalization upon Omicron infection (49–51) but also with the fact that Omicron shows increased breakthrough infections (52). The ‘booster’ resulted in lower reactogenicity than determined in the first two vaccinations (25) and elicited both high antibody titers and enhanced memory T cell responses against the tested SARS-CoV-2 variants including BA.1. Strikingly, the induced neutralizing antibodies were as potent against BA.1 as against Delta and almost as potent as against B.1. This is in contrast to earlier studies focusing on homologous short-interval primary vaccinations that also found ‘boosters’ to induce BA.1-neutralizing titers but where this variant still shows some degree of evasion (9–14, 20, 21). This highlights the major benefits of a third dose regimen (50, 51, 53) especially after heterologous primary vaccination in young adults to protective humoral and cellular immunity against SARS-CoV-2 variants.

In this cohort, the BNT162b2 ‘booster’ induced titers that were neutralizing BA.1 almost as potently as B.1. In light of the immune evasive properties of Omicron and the results from studies on homologous vaccinations, this finding is somewhat surprising. As potential confounding factors, unnoticed SARS-CoV-2 infection of the participants was excluded by nucleocapsid antibody detection and adequate sensitivity of the used neutralization assay has been validated previously (54). Therefore, this exceptionally efficient neutralization of BA.1 is most likely due to the extremely high antibody levels after the ‘booster’ as well as the long-term germinal center reaction, ongoing affinity maturation after vaccination (55), and reactivation of memory B cells (56). The Delta VOC was also potently neutralized after the ‘booster’, resulting in titers that were lower than against B.1 but similar to BA.1, indicating immune evasion by these VOCs but at the same time a broad cross-reactivity of ‘booster’-induced antibodies. These remarkable potent and broadly active antibodies might be a result of the heterologous vaccination where DNA and mRNA vaccines encoding non- and pre-fusion-stabilized spike protein variants are mixed (57). Another explanation for the strikingly effective neutralization of the Omicron BA.1 VOC might be the relatively young age and associated potent immunity (58–61) of the here analysed cohort with median age of 29.5 years. Also timing might play a role, as the intervals in the primary vaccination and between the ‘booster’ influence humoral as well as cellular responses (62, 63). Studies on homologous primary BNT162b2 vaccinations with a 3-week interval detected neutralizing titers of 306–604 against B.1 but only undetectable-13 against BA.1 (9–11, 20, 21). In our study, we already detected a titre of 345 after primary vaccination, and similar results have been obtained in other studies looking at BA.1 neutralization of sera from heterologously ChAdOx1-BNT162b2-vaccinated individuals with an interval of 8–12 weeks (9, 15, 37). This is in line with the finding that longer intervals in heterologous or homologous primary vaccinations result in higher neutralization capacities of SARS-CoV-2 VOCs (21, 28, 62, 64–66). This might be attributed to ongoing antibody maturation before primary boost (55, 56, 67) as has also been observed for vaccination against influenza virus (68). Thus, the typically longer interval within heterologous primary vaccination might result in affinity maturation already before the second vaccine dose and explain the strikingly potent cross-neutralization of BA.1 after the ‘booster’.

After the ‘booster’, memory CD4⁺ T cells became strongly reactive toward SARS-CoV-2 spike peptides of all variants. Also the reactivity of the residual CD8⁺ T cells was further enhanced by the ‘booster’ dose, which agrees with data after homologous primary vaccinations showing that a third dose enhances preexisting cellular responses (69, 70). The finding that SARS-CoV-2-specific T cells are generally reactive to spike peptides derived from the Omicron VOC is supported by recent data (19, 40, 70) and suggests that the ‘booster’ might also enhance protection of Omicron-infected individuals from severe disease (71).

The Omicron VOC seems to form a new antigenic cluster (72–74), of which the BA.2 variant has rapidly expanded, now followed by BA.5. First results indicate that neutralization capacity is similar between BA.1 and BA.2 (75, 76) and lower against BA.5 (77, 78); however, the general cross-reactivity of T cells suggests that the ‘booster’ is likely also effective against these variants. Therefore, it will now be of importance to elucidate the longevity of humoral and cellular immunity after the ‘booster’ against circulating variants and most importantly its durable effectiveness in preventing infection and disease (79). In addition, fourth vaccine doses (80) or an adaptation of the vaccines to BA.1 (81) need yet to be proven useful to provide effective protection from these old and new variants. Yet unvaccinated individuals might benefit from an updated vaccine that establishes a high degree of protection against Omicron already after two doses. Altogether, our results suggest that a ‘booster’ 6 months after initial heterologous vaccination of young adults induces good humoral and cellular protection against SARS-CoV-2 Omicron, even though the antigen of immunization is the ancestral Wuhan-Hu-1 spike. Thus, a ‘booster’ following the heterologous vaccination is highly warranted especially in the light of the immune evading Omicron variants.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the ethics committee of Ulm University. The patients/participants provided their written informed consent to participate in this study.

Conceptualization, JMü, JM, AS, MZ, RG. Funding acquisition, JMü, JM, FK, SP. Investigation, AS, MZ, RG, DK, SE-E, DA, CC, SL, PM, TW, AB, JK, JMü. Essential resources, BM, MH, SP, FK, JM, JMü. Writing, JMü, AS, MZ, RG. Review and editing, all authors. AS, MZ, RG contributed equally. AS, MZ, RG and JMü verified the underlying data. All authors read and approved the final version of the manuscript.

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement no. 101003555 (Fight-nCoV) to JM, the German Research Foundation (CRC 1279) to FK and JM, MU 4485/1-1 to JMü, the BMBF to FK (Restrict SARS-CoV-2), and the Robert Koch Institute (RKI). Additional funding was provided to SP from BMBF (01KI1723D, 01KI20328A, 01KI20396, 01KX2021) and the county of Lower Saxony (14-76103-184, MWK HZI COVID-19) and to JM by the DFG (KL 2544/8-1, FR 2974/3-1).

We thank all participants for regular blood donations. We thank Nicola Schrott, Regina Burger, Jana Romana Fischer, and Birgit Ott for skilful laboratory assistance. We thank Sarah Warth and Simona Ursu of the flow cytometry core facility for support with flow cytometric analysis. R.G., A.S., C.C, and T.W. are part of the International Graduate School in Molecular Medicine Ulm.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.