Immune responses against SARS-CoV-2 variants after heterologous and homologous ChAdOx1 nCoV-19/BNT162b2 vaccination

Currently approved viral vector-based and mRNA-based vaccine approaches against coronavirus disease 2019 (COVID-19) consider only homologous prime-boost vaccination. After reports of thromboembolic events, several European governments recommended using AstraZeneca’s ChAdOx1-nCov-19 (ChAd) only in individuals older than 60 years, leaving millions of already ChAd-primed individuals with the decision to receive either a second shot of ChAd or a heterologous boost with mRNA-based vaccines. However, such combinations have not been tested so far. We used Hannover Medical School’s COVID-19 Contact Study cohort of healthcare professionals to monitor ChAd-primed immune responses before and 3 weeks after booster with ChAd (n = 32) or BioNTech/Pfizer’s BNT162b2 (n = 55). Although both vaccines boosted prime-induced immunity, BNT162b2 induced significantly higher frequencies of spike-specific CD4+ and CD8+ T cells and, in particular, high titers of neutralizing antibodies against the B.1.1.7, B.1.351 and P.1 variants of concern of severe acute respiratory syndrome coronavirus 2.

However, in the first half of March 2021, vaccinations with ChAd were halted after reports of moderate to severe thrombocytopenia and unusual thrombosis cases in vaccinees 5,7 . This new syndrome, termed vaccine-induced thrombotic thrombocytopenia 6 or thrombosis with thrombocytopenia syndrome, appears to be induced by antibodies directed against platelet factor 4 that lead to platelet activation 8 . Despite concerns, the European Medicines Agency concluded that the benefits of ChAd vaccination outweigh the potential risks for an individual (https://www.ema.europa.eu/en/news/ astrazenecas-covid-19-vaccine-benefits-risks-context; accessed 17 June 2021), and ChAd remains a valuable tool against COVID-19. However, many countries recommended to vaccinees, who received the first ChAd dose, to have an mRNA vaccine or to choose between ChAd-based or mRNA-based vaccines as a second (boost) dose. An initial report from the United Kingdom randomized Com-COV Study suggested more short-term reactogenicity of heterologous prime-boost schedules 9 .
It remains to be determined whether heterologous prime-boost regimens can induce equal or even stronger immune responses against the novel viral variants compared to the homologous primeboost regimens.
To analyze the efficacy of the heterologous prime-boost vaccination schedule, we used our COVID-19 Contact (CoCo) Study cohort of healthcare professionals (HCPs) 17,18 and monitored responses to homologous and heterologous prime-boost COVID-19 vaccine treatment schedules (Methods). Hannover Medical School HCP vaccinees who received one dose of ChAd were offered a choice between the ChAd and BNT vaccines for a second dose.
To determine immunogenicity of the homologous and heterologous immune regimens, we studied 129 ChAd-primed vaccinees without previous SARS-CoV-2 infection, of whom 32 chose homologous boosting and 55 chose heterologous boosting. For comparison, we included a group of 46 BNT/BNT vaccinated HCPs. The vaccination and blood collection schedule is depicted in Fig. 1a, with additional demographic information (age and sex) in Extended Data Fig. 1a-c. A retrospective analysis revealed that the mean anti-SARS-CoV-2 spike IgG (anti-S IgG) and IgA had declined by 42% and 66%, respectively, from mean titers 30 d after ChAd prime to shortly before boosting, which is similar to declines in BNT/BNT vaccinated individuals (Extended Data Fig. 2a,b). Notably, we found similar levels of anti-S IgG and IgA antibodies in the ChAd/ChAd and the ChAd/BNT groups before the booster, indicating that both groups responded equally well after priming with ChAd (Fig. 1b).
After the booster immunization, increased anti-spike (S) IgG and IgA responses were found in both groups. Heterologous ChAd/ BNT vaccination led to a significant 11.5-fold increase for anti-S IgG (P < 0.0001) compared to a 2.9-fold increase after homologous ChAd vaccination (P < 0.0001) ( Fig. 1b and Extended Data Fig. 2c). Differences in anti-S IgG were not significantly influenced by age or sex of participants (Extended Data Fig. 2d). We observed similar changes for anti-S IgA (Fig. 1b), indicating better humoral immune responses after heterologous prime-boost immunization. Anti-S IgG and IgA concentrations after ChAd/BNT vaccination were within the range of fully BNT/BNT vaccinated individuals (Extended Data Fig. 2b and Extended Data Fig. 3a,b).
We next determined the frequency and phenotype of B cells carrying membrane-bound immunoglobulins specific for the spike protein (Methods and Extended Data Fig. 4). Interestingly, in samples taken before booster vaccination, spike-specific memory B cells could be detected in only 53.1% (17/32) of vaccinees from the ChAd/ChAd group and in only 43.6% (24/55) of vaccinees from the ChAd/BNT group. Moreover, if present, spike-specific memory B cells represented only a rare (~0.003%) population of whole-blood B cells, with no significant difference between the ChAd/ChAd and the ChAd/BNT groups (Fig. 1c, open circles). In contrast, spike-specific memory B cells were significantly increased in all vaccinees from both the ChAd/ChAd and ChAd/BNT groups after booster vaccination (Fig. 1c, filled dots). In contrast to the anti-S antibody responses, heterologous ChAd/BNT and homologous ChAd/ChAd vaccinations led to expansion of spike-specific memory B cells to a similar extent. The increased frequencies of spike-specific memory B cells after booster immunization, combined with increased amounts of spike-specific antibodies, highlight the importance of the booster vaccination for full protection from SARS-CoV-2 infection.
To test for neutralizing activity of antibodies induced by infection or vaccination, we recently developed an enzyme-linked immunosorbent assay (ELISA)-based surrogate virus neutralization test (sVNT) 19 . We adapted the sVNT to include spike proteins of the B.1.1.7, P.1 and B.1.351 VoC (Methods). To validate these new assays, we applied sera from vaccinees who had been recently tested for their neutralizing capacity, applying the vesicular stomatitis virus (VSV)-based pseudotyped virus neutralization test (pVNT) 12 . Comparing results obtained using pVNT with those of the newly developed sVNTs, we observed a high degree of correlation between both assays, with R 2 values ranging between 0.50 and 0.69 (Extended Data Fig. 5). These findings demonstrate that the sVNT is suited to quantitatively assess the neutralization capacity of vaccination-induced antibodies, not only against the Wuhan strain but also against the B.1.1.7, P.1 and B.1.351 variants of SARS-CoV-2.
Applying sVNT assays, we found that 81 of 88 participants had neutralizing antibodies against the Wuhan strain in pre-boost plasma. In contrast, neutralizing antibodies against the B.1.1.7 (17/88), P.1 (12/88) and B.1.351 (5/88) variants were less frequent ( Fig. 1d and Extended Data Fig. 6). At 2-3 weeks after the booster immunization, frequencies and titers of neutralizing antibodies against the Wuhan strain increased in the ChAd/ChAd and ChAd/ BNT groups, with titers reaching higher values in the latter group ( Fig. 1d and Extended Data Fig. 6). Differences between the ChAd and the BNT booster vaccination became even more evident when analyzing the neutralization capacity of antibodies induced against the VoC. In the ChAd/ChAd group, booster immunization increased neutralization of the B.1.1.7 variant in some individuals but showed no effect against the P.1 and B.1.351 variants ( Fig. 1d and Extended Data Fig. 6). In contrast, booster immunization with BNT induced neutralizing antibodies at high frequencies against all analyzed VoC. In the ChAd/BNT group, all participants had neutralizing antibodies against the B.1.1.7 and P.1 variants, and all but two participants also had neutralizing antibodies against the B.1.351 variant ( Fig. 1d and Extended Data Fig. 6). In the ChAd/BNT group, the post-boost neutralization capacity was highest against the Wuhan strain, followed by the B.1.1.7 variant and less efficient against the P.1 and B.1.351 variants ( Fig. 1d and Extended Data Fig. 6). Altogether, these data indicate that the booster immunization led to an increase of neutralizing antibodies in both vaccination groups and that the heterologous BNT booster vaccination efficiently induced neutralizing antibodies against all tested VoC.
In addition to humoral immune responses, we also analyzed frequencies and phenotypes of spike-specific T cells (Methods and Extended Data Fig. 7). The frequencies of spike-specific CD4 + T cells in blood samples collected before booster vaccination were significantly higher for both vaccination groups compared to the MNE (control) peptides or DMSO alone ( Fig. 2a and Extended Data Fig. 8). No significant differences were found between the ChAd/ ChAd and ChAd/BNT groups (Fig. 2a, open circles). After boosting, the frequencies for spike-specific CD4 + T cells increased in both groups and were significantly higher in the ChAd/BNT group (Fig. 2a, filled dots). The same effect was observed for spike-specific CD8 + T cells. These cells were present at similar frequencies in both groups before boosting and increased in frequencies after boosting. Again, boosting with BNT induced higher frequencies than boosting with ChAd ( Fig. 2b, filled dots). Regarding the distribution of spike-specific CD8 + T cells producing interferon (IFN)-γ or tumor necrosis factor (TNF)-α, application of both booster vaccines led to an increase in the proportion of cells producing both cytokines simultaneously (Fig. 2c). Significant increase in spike-specific IFN-γ-producing T cells in the ChAd/BNT group but not in the ChAd/ChAd group was confirmed by cytokine measurement in supernatants after SARS-CoV-2 spike peptide stimulation (Fig. 2d).
Due to the abrupt recommendation of several European governments to discontinue the use of ChAd in the young and middle-aged population, a unique situation was created in which heterologous prime-boost vaccination regimens were applied despite the lack of any information available regarding immunogenicity and safety aspects. This study provides insights into the immunogenic outcome of homologous and heterologous vaccination protocols with two vaccines: BNT and ChAd. Head-to-head comparison of ChAd-primed vaccinees who received either a ChAd or BNT booster immunization revealed that both regimens after boost enhanced both humoral and cellular immune responses. Although this setup did not allow for randomization of the participants, and we are, thus, unable to completely exclude confounding factors, our study revealed that the group boosted with BNT showed significantly stronger immune responses than the group boosted with ChAd. CD4 + and CD8 + T cell responses directed against spike protein epitopes were higher in frequencies, and cells produced more IFN-γ upon re-stimulation. Likewise, the group boosted with BNT developed higher titers of anti-spike protein antibodies of both the IgG and IgA subclasses, and these differences were not significantly influenced by age or sex. It should be noted that these antibodies were highly efficient in neutralizing all three VoC tested in the present study. It was previously reported that vaccinees immunized with BNT/BNT also develop neutralizing antibodies against the VoC 20 .
We confirmed these findings in the present study using data from the participants of the CoCo Study cohort who were also immunized with BNT/BNT. Our data indicate that BNT/BNT-vaccinated and ChAd/BNT-vaccinated individuals develop neutralizing antibodies to similar degrees 2-3 weeks after booster vaccination. Likewise, immune responses of the ChAd/ChAd group were in the range of earlier reported results [11][12][13]21 . Although it would have been interesting to also characterize immune responses in a cohort of people immunized with BNT/ChAd, such individuals were not available to us. We want to emphasize that our data obtained in mostly healthy and relatively young HCPs cannot be generalized to elderly people or to specific patient groups. Another limitation of our study is that we were unable to test neutralizing activity against the Delta variant and to collect data on safety and reactogenicity after vaccination. Extended studies, ideally including clinical endpoints, are needed to further characterize immune responses not only in   heterologously immunized cohorts. It will be of particular importance to examine neutralizing activity against novel VoC, such as the Delta variant, and how long protective immune responses are maintained, in both individuals who are at elevated risk for developing severe COVID-19 and individuals who are known for mounting impaired immune responses.

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41591-021-01449-9.

Methods
Participants. Participants for this analysis were from the CoCo Study (German Clinical Trial Registry, DRKS00021152), which started in March 2020 and is an ongoing, prospective, observational study monitoring anti-SARS-CoV-2 IgG immunoglobulin and immune responses in 1,493 HCPs at Hannover Medical School and in individuals with potential contact to SARS-CoV-2 (refs. 18,22
Pre-incubation of the spike protein with serum or plasma of convalescent patients or vaccinees prevents subsequent binding to ACE2 to various degrees, depending on the amount of neutralizing antibodies present. In detail, Maxisorp 96F plates (Nunc) were coated with recombinant soluble hACE2-Fc(IgG1) protein at 300 ng per well in 50 μl of coating buffer (30 mM Na 2 CO 3 , 70 mM NaHCO 3 , pH 9.6) at 4 °C overnight. After blocking with hACE2-Fc(IgG1), plates were washed with PBS, 0.05% Tween-20 (PBST) and blocked with BD OptEIA Assay Diluent for 1.5 h at 37 °C. In the meantime, plasma samples were serially diluted three-fold starting at 1:20 and then pre-incubated for 1 h at 37 °C with 1.  Flow cytometric analysis of spike-specific B cells. Total leukocytes were isolated from whole blood using erythrolysis in 0.83% ammonium chloride solution. Isolated cells were then washed, counted and resuspended in PBS and stained for 20 min at room temperature with an antibody mix containing antibodies listed in Extended Data Fig. 4 together with spike-mNEONGreen protein (5 μg per reaction; production will be described elsewhere). After one wash, samples were acquired on a spectral flow cytometer, and the data were analyzed as described above.
Quantification of IFN-γ and TNF-α release. Next, 0.5 ml of full blood was stimulated with manufacturer's selected parts of the SARS-CoV-2 S1 domain of the spike protein for a period of 20-24 h. We carried out negative and positive controls according to the manufacturer's instructions (SARS-CoV-2 Interferon Gamma Release Assay (Euroimmun)). After stimulation, supernatants were collected after centrifugation and examined by the LEGENDplex kit (BioLegend) according to the manufacturer's instructions. Data from duplicate measurements were acquired with a LSR II flow cytometer (BD Biosciences) using BD's FACSDiva version 8.0.1 software and analyzed with the LEGENDplex Data Analysis Software Suite, Gen5 version 2.01 software.
Statistics. Statistical analysis was performed using GraphPad Prism version 8.4 (GraphPad Software) and SPSS version 20.0.0 (IBM, SPSS Statistics). For comparison of spike-specific IgG and IgA levels, as well as for comparison of percentages of cytokine-secreting T cells or cytokine concentrations in serum, we used the paired t-test (within groups) or two-way analysis of variance (ANOVA) followed by Sidak's multiple comparison test (between groups). For comparison of sVNT titers, we used the chi-square test for trend. Differences were considered significant if P < 0.05. Correlation between sVNT and pVNT values was calculated using single linear regression analysis.
Ethics committee approval. The CoCo Study and the analysis conducted for this article were approved by the Internal Review Board of Hannover Medical School (institutional review board no. 8973_BO-K_2020, amendment December 2020).
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
All requests for raw and analyzed data that underlie the results reported in this article will be reviewed by the CoCo Study team at Hannover Medical School (cocostudie@mh-hannover.de) to determine whether the request is subject to confidentiality and data protection obligations. Data that can be shared will be released via a material transfer agreement. Source data are provided with this paper.