Main

Efforts to develop a vaccine against SARS-CoV-2 to control the global COVID-19 disease pandemic have been underway since January 2020, with more than 40 vaccine candidates in clinical trials by October 20201. The past decade has seen an expansion and acceleration in the development of tools to support pandemic preparedness, including the development of vaccines against novel and emerging pathogens8,9. This acceleration, spurred on by numerous outbreaks of diseases, including SARS-CoV, MERS-CoV, Ebola and Zika, has leveraged the use of platform technologies and blueprints for target product profiles for priority diseases10. Replication-deficient adenoviruses11 are attractive for use as COVID-19 vaccine candidates, as they can be manufactured at scale, have favorable safety profiles and are highly immunogenic. Importantly, viral vectored vaccines can induce strong immune responses in older adults and immunocompromised individuals12,13. Replication-deficient adenovirus vectors are also potent inducers of both antibodies as well as cytotoxic T cells; the latter can clear virus-infected host cells and contribute to the control of infection, alleviating disease symptoms4,14. Importantly, high-frequency T cell responses targeting the SARS-CoV-2 spike protein have been detected in patients who recover from COVID-19, with recent data suggesting a role for T cells during COVID-19 (refs. 15,16,17).

Previous efforts to develop vaccines against human coronaviruses have faced challenges, with several preclinical studies demonstrating disease enhancement in vaccinated animals after viral challenge. This was characterized by eosinophilic infiltrates resulting in immunopathology, after the induction of a T helper cell type 2 (Th2)-biased response, or a weak neutralizing antibody response that might contribute to antibody-dependent enhancement of infection3. In-depth analysis of SARS-CoV-2 vaccines are being conducted to determine whether responses are Th1 or Th2 dominated; these types of studies are being implemented in several COVID-19 vaccine trials18,19,20,21.

ChAdOx1 nCoV-19 (AZD1222) is a replication-deficient simian adenoviral vector that expresses the full-length SARS-CoV-2 spike protein. In preclinical studies, either a single dose or two doses of ChAdOx1 nCoV-19 vaccination prevented SARS-CoV-2-mediated pneumonia in rhesus macaques22. We previously reported safety data from phase 1/2 studies and demonstrated induction of SARS-CoV-2 spike-specific antibodies after vaccination, with boosting of binding and neutralizing titers after a second dose7. These data supported progression to phase 3 trials with a two-dose regimen, and we have now expanded our immunogenicity analysis to explore a wider range of the immunological phenotypes induced. In an accompanying paper23, we present detailed functional antibody profiling of responses to prime-boost regimens with differing doses and intervals.

Currently, there are no defined correlates of protection against COVID-19 infection, and the immunological thresholds required for vaccine efficacy remain undefined24. Clinical studies have suggested a protective role for both humoral and cell-mediated immunity in recovery from SARS-CoV-2 infection5,6,25. Here we provide a detailed description of the immune response after administration of one dose of ChAdOx1 nCoV-19. We define, in detail, the isotypes, subclasses and antibody avidity induced after vaccination and also perform multiplex cytokine profiling and intracellular cytokine staining (ICS) analysis, demonstrating that ChAdOx1 nCoV-19 vaccination induces a predominantly Th1-type response.

Results

Study participants

Recruitment, vaccination and demographics of the study participants were previously reported, with interim safety and immunogenicity data7. Healthy adults aged 18–55 years (n = 88) were randomized to receive either 5 × 1010 viral particles of ChAdOx1 nCov-19 or control vaccine (MenACWY) (Group 1; Supplementary Fig. 1). Blood samples were collected on the day of vaccination and 7, 14, 28 and 56 d after vaccination. Supplementary Table 1 summarizes the number of individuals assessed in each assay.

Immune cell activation induced by ChAdOx1 nCoV-19 vaccination with Th1-biased cytokine secretion

An unbiased approach was applied to measure gross phenotypic and cellular changes on days 7, 14 and 28 after vaccination (Fig. 1a–e). Flow cytometric and combined t-distributed stochastic neighbor embedding (tSNE) analysis of 26 randomly selected ChAdOx1 nCoV-19 vaccinated volunteers showed discrete populations of T cells, natural killer (NK) cells and B cells. Within these clusters, distinct populations of proliferating (Ki-67+) or activated (CD69+) cells were identified (Fig. 1b–e). B cells, especially the IgG+ B cell population, upregulated Ki-67 at all post-vaccination time points (Fig. 1f,g). Within the total B cell population, the shift toward an activated phenotype peaked on days 7–28 and for the IgG+ B cell population on days 7–14 (Fig. 1f,g).

Fig. 1: Activation of lymphocyte populations after ChAdOx1 nCoV-19 vaccination.
figure 1

a–e, tSNE analysis of 9,600,000 live lymphocytes from 26 ChAdOx1 nCoV-19 vaccine trial participants across four time points (D0 n = 24, D7 n = 23, D14 n = 25 and D28 n = 24). Two samples were not available, and six samples with fewer than 100,000 live lymphocytes were excluded. The tSNE plot was generated by concatenation of samples containing 100,000 randomly selected live lymphocytes from each sample. a, Global clustering of immune cells across all samples. b–e, tSNE population analysis at day 0 and days 7, 14 and 28 after vaccination. Areas of Ki-67+ activity (yellow) cluster in IgG+ B cells (1), NK cells (2) and CD4+ T cells (3) after vaccination. Analysis conducted on unstimulated cells. f–h, Heat map analysis of activation markers expressed by immune cells at days 0, 7, 14 and 28 after ChAdOx1 nCoV-19 vaccination (D0 n = 24, D7 n = 23, D14 n = 25 and D28 n = 24). f, Expression of Ki-67 by IgG+ B cells and NK cells (top two rows) and NK cells (top three rows). Expression of CD69 by CD4+ T cells and CD8+ T cells (bottom two rows). g, Expression of Ki-67 by B cells, CD4+ T cells and CD8+ T cells. h, Expression of TNF-α and IFN-γ by CD4+ T cells. Analysis of TNF-α and IFN-γ expression conducted on cells stimulated with spike glycoprotein peptide pools with unstimulated values subtracted. All other analysis was conducted on unstimulated cells. i, A multiplex cytokine analysis was performed on day 7 after vaccination using supernatants after antigen-specific stimulation of PBMCs from ChAdOx1 nCov-19 (red) and MenACWY (blue). Number of samples presented: MenACWY–ChAdOx1 nCov-19: IFN-γ (n = 40,40); IL-2 (n = 42,42); TNF-α (n = 40,41); IL-1β (n = 41,42); IL-12p70 (n = 38,28); IL-4 (n = 38,38); IL-10 (n = 41,39); IL-13 (n = 31,36); and IL-8 (n = 42,41). Individual data points are shown here as an aligned dot plot with lines showing the median with IQR. Significant differences were determined by two-tailed Mann–Whitney test with Bonferroni correction for multiple comparisons (***P < 0.001; **P < 0.01; *P < 0.05).

Full size image

CD4+ T cells had increased expression of CD69 on days 7–28 after vaccination and a trend toward increased Ki-67 expression at days 7 and 14 after vaccination (Fig. 1f,g)26. CD8+ T cells expressed a similar pattern of Ki-67 and CD69 expression between days 7 and 28 after vaccination (Fig. 1f,g). We did not detect increases in expression of terminal differentiation markers CD57 and KLRG1 in post-vaccination CD8+ T cells (Supplementary Fig. 2), which would indicate a reduction in post-vaccination cytotoxic capacity27. After peptide stimulation, an increase in tumor necrosis factor (TNF)-α and interferon (IFN)-γ production by CD4+ T cells was also observed at day 14 (Fig. 1h).

NK cells can elicit a cytotoxic response to viral infection or vaccination28,29. Total expression of Ki-67 by NK cells increased steadily to a peak at day 28 (Fig. 1f). There was no significant change in the expression of CD57 or the activating receptor NKG2C (Supplementary Fig. 2).

Multiplex cytokine analysis was performed on day 7 after vaccination after antigen-specific stimulation of peripheral blood mononuclear cells (PBMCs) with pooled SARS-CoV-2 spike peptides. Of the nine cytokines analyzed, five (IL-1β, IL-12p70, IL-4, IL-13 and IL-8) showed no difference in expression levels after stimulation. IFN-γ and IL-2 levels after PBMC stimulation were significantly increased in individuals who received the ChAdOx1 nCoV-19 vaccine compared to MenACWY controls (***P = 0.0009 and **P = 0.0027, respectively, two-tailed Mann–Whitney test). IL-4 and IL-13 levels after PBMC stimulation were not elevated in individuals who received the ChAdOx1 nCoV-19 vaccine after stimulation of PBMCs (P > 0.05 for both, two-tailed Mann–Whitney test), but a modest increase in IL-10 was measured (*P = 0.045, two-tailed Mann–Whitney test). The magnitude of cytokine secretion measured in PBMC supernatant in individuals who received the ChAdOx1 nCoV-19 vaccine was greater for IFN-γ (median 36.4 pg ml−1, interquartile range (IQR) 15–67) and IL-2 (median 10.7 pg ml−1, IQR 1.7–22) than for IL-10 (median 1.4 pg ml−1, IQR 0.9–2.6), indicating a strong potential bias toward secretion of Th1 cytokines in blood in response to stimulation with SARS-CoV-2 spike peptides (Fig. 1i).

Humoral and cellular immune responses to ChAdOx1 nCoV-19 do not differ by sex

Robust immunity induced by ChAdOx1 nCOV-19 against the SARS-CoV-2 spike antigen, measured by ex vivo IFN-γ ELISpot and total IgG ELISA, was previously reported7. We analyzed these two main immunological outcome measures by sex and age (Supplementary Fig. 3). We found no sex difference in vaccine response at any of the time points measured (Supplementary Fig. 3a,b; P > 0.05, two-tailed Mann–Whitney test). We detected no association between age and magnitude of immune response for either outcome measure (Supplementary Fig. 3c) in this population aged between 18 and 55 years.

ChAdOx1 nCoV-19 vaccination induces SARS-CoV-2-specific IgM and IgA levels

Anti-SARS-CoV-2 IgG responses were detectable at day 14, peaked at day 28 and were maintained at day 56, as reported previously7. Here we show that vaccination with ChAdOx1 nCoV-19 also generated increased levels of SARS-CoV-2 spike-specific IgM and IgA, with peak responses at day 14 or day 28, respectively (Fig. 2a,b and Supplementary Table 2). Low SARS-CoV-2 spike-specific IgE was detected after vaccination with ChAdOx1 nCoV-19, similar to that in convalescent patients with SARS-CoV-2 (Supplementary Fig. 4).

Fig. 2: Immunoglobulin isotype responses induced by ChAdOx1 nCoV-19 vaccination.
figure 2

Volunteers received ChAdOx1 nCoV-19 (red) or MenACWY (blue) vaccination at day 0. SARS-CoV-2 spike trimer-specific IgM (a) and IgA (b) responses were quantified by standardized ELISA. Individual data points were expressed as ELISA units (EU) and shown here as an aligned dot plot with lines showing the median with IQR. Assays were performed at D0, D7, D14, D28 and D56 after vaccination (n = 44 for all time points in both groups of volunteers, except for D56 in ChADOx1 nCoV19 where samples from only 43 volunteers were available). Dotted lines are shown at the limit of quantification of each assay. c, Avidity of SARS-CoV-2 spike trimer-specific IgG responses was measured using a NaSCN chemical displacement ELISA. Individual data points were expressed as an IC50 and shown here as scatter dot plots with lines showing the median with IQR. Assay was performed at D28 (n = 44) and D56 after vaccination with ChAdOx1 nCoV-19 (n = 43). Significant difference was determined by two-tailed Wilcoxon test (***P < 0.001). CONV, convalescent plasma samples from patients with SARS-CoV-2 who recovered are shown here as a scatter dot plot with the line showing median with IQR (SARS-CoV-2 spike trimer-specific IgM, IgA n = 50; IgG avidity n = 49). CONV symbols are colored by disease severity: green, asymptomatic; gray, mildly symptomatic; yellow, moderately symptomatic; orange, severely symptomatic; and red, critically symptomatic.

Full size image

Anti-SARS-CoV-2 spike-specific IgG avidity increased significantly between day 28 (median 0.66, IQR 0.60–0.76; n = 44) and day 56 (median 0.88, IQR 0.74–0.94; n = 44) after vaccination (***P < 0.001, two-tailed Wilcoxon test) (Fig. 2c). At day 56, IgG avidity induced by ChAdOx1 nCoV-19 vaccination was similar to that measured in plasma from convalescent patients with SARS-CoV-2 (median 0.77, IQR 0.62–0.92; n = 49).

Subclass analysis after vaccination with ChAdOx1 nCoV-19

Specific IgG1 and IgG3 responses were readily detectable at day 14, increased by day 28 and returned to a similar level to that measured on day 14 by day 56 (Fig. 3a,b and Supplementary Table 2). Although IgG3 responses were quantifiable in nearly all individuals who received the vaccine (day 14, 39/44; day 28, 42/44; and day 56, 39/44), IgG1 responses were quantifiable in approximately half (day 14, 24/44; day 28, 23/44; and day 56, 22/44). Median levels of IgG2 and IgG4 were low across all time points (Fig. 3c,d). A similar IgG3/IgG1 profile with low levels of IgG2 and IgG4 was measured in convalescent plasma samples. In agreement with previously reported data30, SARS-CoV-2 spike-specific IgG1 was below the limit of quantitation in some convalescent plasma samples (Fig. 3a).

Fig. 3: IgG subclass responses induced by ChAdOx1 nCoV-19 vaccination.
figure 3

Volunteers with measurable SARS-CoV-2 spike-specific IgG at day 14 were assayed for IgG subclasses. SARS-CoV-2 spike trimer-specific IgG1 (a) and IgG3 (b) responses were quantified by standardized ELISA and expressed as ELISA units (EU). IgG2 (c) and IgG4 (d) were measured by indirect ELISA and expressed as OD405. Individual data points are shown here as an aligned dot plot with lines showing the median with IQR. Assays were performed at D14, D28 and D56 after vaccination (n = 44 for all time points, except for D56 where there were only 43 volunteers available). Dotted lines are shown at the limit of quantification of each assay. CONV, convalescent plasma samples from patients with SARS-CoV-2 who recovered are shown here as a scatter dot plot with lines showing the median with IQR (n = 50). CONV symbols are shown as diamonds and colored by disease severity: green, asymptomatic; gray, mildly symptomatic; yellow, moderately symptomatic; orange, severely symptomatic; and red, critically symptomatic.

Full size image

ChAdOx1 nCoV-19 induces a broad T cell response to the S1 and S2 subunits of the SARS-CoV-2 spike antigen

Vaccine-specific T cell responses were measured by IFN-γ ELISpot before and after vaccination with ChAdOx1 nCoV-19, peaking at day 14 (ref. 7), and summed T cell responses to the peptide pools for this cohort have been previously reported7. Responses were assayed against 13 pools of overlapping peptides (Supplementary Table 3) spanning the length of the vaccine antigen insert, which includes the S1 and S2 subunits, and an exogenous human tissue plasminogen activator (tPA) leader signal sequence peptide previously shown to enhance immunogenicity of a MERS-CoV vaccine candidate in mice31. There was a significant increase in response against both subunits between D0 and D14 (Fig. 4a; n = 42 participants, P < 0.0001 for both S1 and S2 comparing D0 to D14, two-tailed Wilcoxon matched pairs test). All pools except tPA elicited a positive response in at least 24% of participants (defined as the median of the negative control plus four standard deviations), indicating recognition of multiple epitopes across the spike antigen (Fig. 4b; n = 42). The most frequently recognized pools were 4 and 2, which span amino acids 311–430 and 101–200 of the S1 domain and generated a positive response by IFN-γ ELISpot in 35/42 (83%) and 33/42 (78%) participants, respectively. Responses at D14 were also plotted as fold change from D0 (Fig. 4b and Supplementary Fig. 5), and the greatest increases were to pools 4 and 5. These pools elicited a median response of 146 spot-forming cells (SFCs)/106 PBMCs and 80 SFCs/106 PBMCs, respectively, at day 14, equating to a median of a 27- and 18-fold change from baseline.

Fig. 4: T cell responses to SARS-CoV-2 spike peptides measured by IFN-γ ELISpot and flow cytometry with ICS.
figure 4

Only data from individuals who received the ChAdOx1 nCoV-19 vaccine are shown. a, Total response to S1 and S2 (sum of six peptide pools each) at D0 and D14 after vaccination. Individual data points are shown here as a scatter dot plot with lines showing the median with IQR (n = 42 for all time points). The dashed line represents the lower limit of detection of the assay (48 SFCs). Significant difference was determined by two-tailed Wilcoxon test (***P < 0.001). b, Heat map of fold change in SFCs to each peptide pool for every participant from D0 to D14 after vaccination (n = 42 for all time points). c, Frequency of CD4+ or CD8+ T cells expressing IFN-γ, IL-2 or TNF-α (n = 34) and (d) frequency of CD8+ T cells expressing CD107a+ at D14 after vaccination (n = 34). Individual data points are shown here as a scatter dot plot with lines showing the median with IQR. e, Frequency of Th1 and Th2 cytokine secretion by CD4+ T cells at D14 after vaccination (n = 34). Individual data points are shown as a box plot with whiskers from the minimum to the maximum value. f, Frequency of CD4+ or CD8+ T cells expressing relevant individual cytokines at D14 after vaccination. Individual data points are shown here as a scatter dot plot with lines showing the median with IQR (n = 32). Dotted lines show the lower limit of detection. Dark green circles indicate participants with an ELISpot response to the summed spike peptides pools >200 SFCs/106 PBMCs before vaccination (n = 2); light green circles indicate participants with an ELISpot response to the summed spike peptides pools >200 SFCs/106 PBMCs who were also seropositive before vaccination (n = 3). Clear circles represent participants with an ELISpot response <200 SFCs/106 PBMCs who were also seronegative before vaccination (n = 27). g, Pie charts indicating the expression of cytokine combinations from CD4+ and CD8+ T cells (n = 32, all panels). LLD, lower limit of detection.

Full size image

Vaccination induces a Th1-biased CD4+ and CD8+ T cell response against SARS-CoV-2 spike peptides

Flow cytometry with ICS of PBMCs stimulated with peptides spanning the S1 and S2 subunits of SARS-CoV-2 spike protein demonstrated antigen-specific cytokine secretion from both CD4+ (median 0.12, IQR 0.061–0.16) and CD8+ (median 0.074, IQR 0.036–0.12) T cells 14 d after a single dose of ChAdOx1 nCoV-19 (Fig. 4c). CD8+ T cells expressing the degranulation marker CD107a, indicating cytotoxic function, were detected after vaccination (median 0.038, IQR 0.012–0.066; Fig. 4d). CD4+ responses were heavily biased toward secretion of Th1 cytokines (IFN-γ and IL-2) rather than Th2 (IL-5 and IL-13; Fig. 4e) The frequency of cytokine-positive cells was generally higher in the CD4+ T cell population than the CD8+ T cell population, and cytokine responses were detected at day 14 from participants with positive pre-vaccination T cell and antibody responses to SARS-CoV-2 (Fig. 4f). When combinations of cytokines were assessed, few multifunctional T cells were detected in either the CD4+ or CD8+ T cell populations (Fig. 4g). Responses were dominated by T cells expressing single cytokines, particularly monofunctional IFN-γ+ CD8+ T cells.

Discussion

An effective vaccine against COVID-19 will likely require both neutralizing antibodies and a Th1-driven cellular component. Analyzing the induction of immune responses after vaccination is driven, in part, by concerns about enhanced disease from potentially immunopathologic Th2 responses, as seen in animal studies of vaccines against other coronaviruses3,18,19,20,21. Vaccine-enhanced disease was also observed in early development of inactivated vaccines against respiratory syncytial virus, wherein pathology was associated with a high ratio of non-neutralizing antibodies to neutralizing antibodies, infiltration of neutrophils and eosinophils and predominantly a Th2-biased response32. We showed that antibodies induced after the first dose of ChAdOx1 nCoV-19 are neutralizing and are further increased after a second dose7 and were associated with reduced disease in vaccinated and challenged non-human primates22.

We have described here the profile of cytokine expression from both CD4+ and CD8+ T cells and the IgG subclass composition of the antibody response after administration of a single dose of the ChAdOx1 nCoV-19 vaccine. Robust B cell activation and proliferation was observed after vaccination, and anti-IgA and IgG antibodies to the SARS-CoV-2 spike protein were readily detected in sera from vaccinated volunteers7. Anti-spike IgG responses at the peak of the response after vaccination show a polarized IgG1 response, consistent with naturally acquired antibodies against SARS-CoV-2, as well as an IgG3 response in most vaccinees. Produced early after viral infections, IgG3 coordinates multiple antibody effector functions and might contribute to recovery after SARS-CoV-2 infection33,34. A mixed IgG1 and IgG3 response, with low levels of IgG2 and little detectable IgG4, is in agreement with previously published reports describing the induction of Th1-type human IgG subclasses (IgG1 and IgG3) after adenoviral priming35,36.

ChAdOx1 nCoV-19 induces a broad and robust T cell response to both S antigen subunits. The functionality of the T cell response observed here is similar in phenotype to that observed with other replication-deficient adenoviral vectors, with responses dominated by individual T cells secreting single, rather than multiple, cytokines20. Whether vaccine-induced monofunctional or polyfunctional T cells are of greater protective value appears to vary by disease53,54 and is unclear for SARS-CoV-2 infection. Analysis of cytokine secretion after peptide stimulation of PBMCs demonstrated that IFN-γ and IL-2 secretion were increased in individuals who received the ChAdOx1 vaccine compared to controls, and, notably, IL-4 and IL-13 levels were not increased. Similarly, phenotyping by flow cytometry demonstrated that CD4+ T cells secreted predominantly Th1 cytokines (IFN-γ, IL-2 and TNF-α) rather than Th2 (IL-5 and IL-13). Importantly, we demonstrate, with several methodologies (multiplex cytokine profiling, ICS analysis and antibody isotype profiling), that vaccination with ChAdOx1 nCoV-19 induces a predominantly Th1 response.

An important aspect in the epidemiology of COVID-19 disease is the marked difference in the mortality rates from disease between males and females, despite similar case rates37. We, therefore, disaggregated the data by sex and demonstrated no difference in the magnitude of either cellular or total IgG antibody responses between male and female participants. Other significant demographic risk factors for COVID-19 disease have been shown to include age and ethnicity38. The sample size in this cohort was relatively small, age was limited to 18–55 years and the vast majority of participants were white, limiting the ability to investigate these variables. It will be necessary to continue disaggregated analysis of the larger phase 2 and 3 cohorts, powered for subgroup analysis. It will also be important to continue to assess immune response durability over time, with consideration given to comorbidities that might further influence vaccine-induced immunity39.

Although there are no defined immune correlates of protection against COVID-19, it is generally accepted that high-titer neutralizing antibodies with a robust cytotoxic CD8+ T cell response and Th1-biased CD4+ effector response will be optimal for protective immunity after SARS-CoV-2 exposure40. Determining the precise threshold and phenotype of immune responses associated with protection will be crucial for bridging between populations and vaccines for any vaccine that demonstrates useful efficacy against infection or disease. If the immunogenicity of current vaccine candidates is insufficient, alternative prime-boost regimens using technologies that are rapidly and sustainably scalable, such as heterologous adenoviral prime-boost regimens, or combinations of viral vectors with approaches, such as messenger RNA (mRNA) vaccines, might be implemented. Although adenovirus-based viral vectors and mRNA vaccines have been in preclinical development for some time, few have progressed to phase 3 and subsequent market authorization; therefore, relatively little is known about effectiveness when compared to traditional vaccine platforms.

Although the number of participants studied here was relatively small, the detailed immunophenotyping of vaccine-induced immunity described here demonstrates strong humoral and cellular immune responses after a single dose, characterized by a Th1-dominated response. Importantly, several other COVID-19 vaccine candidates in clinical development have also reported neutralizing antibody responses41 and induction of Th1-biased cell-mediated immunity.

These data further support the ongoing evaluation of the ChAdOx1 nCoV-19 vaccine candidate in phase 2 and 3 clinical trials.

Methods

Study procedures and sample processing

Full details on the conduct of the phase 1/2 randomized controlled trial of ChAdOx1 nCoV-19 (AZD1222), including the trial protocol, were previously published7. This study was registered at ISRCTN (15281137) and ClinicalTrials.gov (NCT04324606). Only data from single-dose vaccinated volunteers are included in this paper. Before enrolment, all participants gave written informed consent. The trial was conducted according to the principles of Good Clinical Practice, and approval was obtained from a national ethics committee (South Central Berkshire Research Ethics Committee, reference 20/SC/0145) and a regulatory agency in the United Kingdom (the Medicines and Healthcare Products Regulatory Agency). An independent data safety monitoring board was appointed before recruitment began.

Blood samples were collected on the day of vaccination and 7, 14, 28 and 56 d after vaccination. At time points for immunological analyses, blood samples were taken in both plain and heparinized collection tubes. Samples were processed within 4 h of the blood draw. Plain tubes were processed for the collection of blood serum. Tubes were centrifuged at 1,800 r.p.m. for 5 min, and the serum was harvested for storage at −80 °C until required. Heparinized tubes were processed for the collection of PBMCs and blood plasma by density gradient centrifugation. Blood was decanted into Leucosep tubes (Greiner Bio-One) containing Lymphoprep (STEMCELL Technologies) and centrifuged at 1,000g for 13 min with the brake off. A fraction of blood plasma was collected and stored at −80 °C, while the remaining sample was decanted into a fresh Falcon tube and topped up with R0 media (RPMI-1640 cell culture media containing 1% penicillin–streptomycin and 2 mM L-glutamine (all Sigma-Aldrich). Samples were centrifuged again at 1,800 r.p.m. for 5 min; the supernatant was poured off; and the cell pellet was resuspended once more in R0 media for washing. After centrifugation, the cell pellet was resuspended in 10 ml of R10 media (RPMI-1640 containing 1% penicillin–streptomycin, 2 mM L-glutamine and 10% fetal calf serum (FCS, Labtech) for counting.

Cells were counted using a CasyCounter (OMNI Life Science) for use in fresh assays or for cryopreservation. The assays performed on fresh cells were ELISpot and ICS only (described below). All remaining cells were frozen at a concentration of 8–12 × 106 PBMCs per ml. After centrifugation (1800 r.p.m., 5 min) cells were resuspended in cold FCS at half the total freeze-down volume. Cells were placed in a refrigerator (2–8 °C) for 20 min before an equal volume of cold FCS containing 20% dimethylsulphoxide was added. One-milliliter aliquots were prepared and quickly transferred to CoolCells (Corning) for freezing at −80 °C overnight. Tubes were then transferred to a −150 °C ultra-low temperature freezer until required.

Convalescent plasma samples were obtained from hospitalized adult (≥18 years) patients admitted with polymerase chain reaction-positive SARS-CoV-2 infection or from healthcare workers enrolled in COVID-19 surveillance studies. Studies were approved by the following committees: Gastrointestinal Illness in Oxford: COVID substudy (Sheffield Research Ethics Committee, reference 16/YH/0247); ISARIC/WHO Clinical Characterisation Protocol for Severe Emerging Infections (Oxford Research Ethics Committee C, reference 13/SC/0149); and Sepsis Immunomics Project (Oxford Research Ethics Committee C, reference 19/SC/0296). Both asymptomatic and symptomatic participants were tested for each assay. Additional details on experimental procedures performed on convalescent plasma samples were described previously7.

Peptides and stimulations

Peptides spanning the full length of the SARS-CoV-2 spike protein sequence were synthesised for use in antigen-specific T cell assays (ProImmune). A total of 253 peptides were synthesized as 15-mers overlapping by ten amino acids. Peptides were also synthesized for the N-terminal tPA leader sequence, which was included to increase expression of the vaccine antigen from the adenoviral vector. Details of peptide sequences and pooling for assays are shown in Supplementary Table 3. Briefly, for the Cytek Aurora flow cytometry assay, Meso Scale Discovery (MSD) Th1/Th2 cytokine profiling assay and ICS, two separate peptide pools were made spanning the S1 (134 peptides) and S2 (119 peptides) subunits of the SARS-CoV-2 spike protein. For the ELISpot assay, 12 pools of 18–24 peptides were made consisting of six pools each for the S1 and S2 subunits. A separate tPA leader sequence pool (five peptides) was included in this assay.

Flow cytometry conducted on a Cytek Aurora spectral analyzer

Flow cytometry was performed from frozen aliquots of PBMCs of donors from days 0, 7, 14 and 28 after vaccination with ChAdOx1 nCoV19 (D0 n = 24, D7 n = 23, D14 n = 25 and D28 n = 24). Cells were defrosted in media containing >5 U ml−1 of benzonase and resuspended in complete RPMI media supplemented with 10% FCS, L-glutamine and penicillin–streptomycin at a concentration of 2 × 107 cells per ml. Then, 2 × 106 PBMCs per well were plated in a 96-well plate and stimulated with synthetic peptides spanning the SARS-CoV-2 spike protein split into two separate pools for the S1 and S2 subunits (Supplementary Table 3) at a final concentration of 2 µg ml−1 or media as a control. One well per donor was stimulated with phorbol 12-myristate 13-acetate and ionomycin (Cell Activation Cocktail, BioLegend) as a positive control. PBMCs were co-stimulated in the presence of anti-human CD28, CD49d (1 µg ml−1; Life Technologies) and CD107a-BV785 (BioLegend) for 2 h at 37 °C with 5% CO2 and then incubated for an additional 16 h after the addition of 1 µg ml−1 of brefeldin A and monensin to each well (BioLegend).

PBMCs were washed in FACS buffer (phosphate-buffered saline with 0.5% bovine serum albumin and 1% EDTA) and stained with a cocktail of surface antibodies, including anti-human Live/Dead-Zombie UV, CD4-AF700, CD19-Spark NIR 685, CD56-APC, CCR7-PerCP/Cy5.5, PD1-PE/Dazzle 594, CD57-PE/Cy7(BioLegend) CD8-AF405, CD45RA-SuperBright 702, CD27-PerCP eF710 and CD20-AF532 (Thermo Fisher Scientific); CD16-BUV495, CD3-BUV661, CD138-BUV805, NKG2A-BV480 and IgM-BB515 (BD Biosciences); and NKG2C-PE and KLRG1-VioBlue (Miltenyi) in FACS buffer with 10% Brilliant Stain Buffer Plus (BD Biosciences). PBMCs were incubated at 4 °C in the dark for 30 min and then washed twice in FACS buffer. PBMCs were then incubated in CytoFix/CytoPerm solution (BD Biosciences) at 4 °C in the dark for 30 min and then washed twice in Perm/Wash buffer and then stained with a cocktail of intracellular antibodies, including anti-human IFN-γ-BV650 and IL-2-BV605 (BioLegend); IgG-BV421, TNF-α-BUV395, CD69-BV750, CD71-BUV563 and CD25-BV737 (BD Biosciences); and Ki-67-APC eF780 (Thermo Fisher Scientific) in Perm/Wash. PBMCs were incubated at 4 °C in the dark for 30 min, washed twice in Perm/Wash buffer, once in FACS buffer and then re-suspended in 200 µl of FACS buffer for acquisition on a custom four-laser Cytek Aurora spectral analyzer using SpectroFlo v2.2 (Cytek Biosciences).

Single-fluorochrome compensation was calculated on beads (BD Biosciences and Miltenyi) or human PBMCs. Analysis of data was conducted in FlowJo (v10.6.2) by a hierarchical gating strategy (Supplementary Fig. 6) and Prism 8 (GraphPad). Peptide-specific responses were calculated by subtraction of the unstimulated controls from the peptide-stimulated samples.

Downsampling and tSNE analysis were conducted on gated live lymphocytes in FlowJo v.10.7.1. A random sample of 100,000 cells per donor and time point was collected and concatenated into a single file. All fluorochrome colors and the sample time point were included as parameters. The tSNE analysis was implemented in FlowJo v.10.7.1 with 100,000 iterations and a perplexity of 30 and using Barnes–Hut gradient algorithm.

MSD Th1/Th2 cytokine profiling

Th1/Th2 cytokine responses were measured in tissue culture supernatants from the stimulation of PBMCs with synthetic peptides covering the spike protein. Then, 5 × 105 freshly isolated PBMCs were resuspended in 250 µl of R10 media in 96-well U-bottom plates and supplemented with 1 µg ml−1 of anti-human CD28 and CD49d. Peptides spanning the S1 and S2 subunits of the SARS-CoV-2 spike protein (Supplementary Table 3) were added to separate wells at a concentration of 2 µg ml−1. Each sample also included an unstimulated (media-only) control. After a 16–18-h incubation at 37 °C with 5% CO2, cells were pelleted by centrifugation (1,800 r.p.m., 5 min), and 200 µl of supernatant was harvested. Supernatants from the S1 and S2 stimulations were combined and stored at −80 °C until required.

Cytokine responses were analyzed using the MSD V-PLEX Proinflammatory Cytokine (human) Panel 1 Kit, validated by MSD. Each plate is coated with nine different capture monoclonal antibodies against nine different cytokines arranged in independent spots on the base of each well. Cytokines IFN-y, IL-1β, IL-2, IL-4, IL-8, IL-10, IL-12p70, IL-13 and TNF-α were associated with either a Th1- or Th2-type T cell response.

Supernatants were diluted 1:2 for unstimulated samples and 1:10 for S1/S2 stimulated samples in MSD Diluent 2. The kit provides a multi-analyte lyophilized calibrator that, when reconstituted, will be used as the standard curve using a four-fold serial dilution to form an eight-point standard curve plated out in duplicate. Cytokine measurements were carried out according to the manufacturer’s instructions. Plates were read on an MSD reader within 15 min of adding the read buffer.

Data were analyzed using MSD Discovery Workbench 4.0. Samples were repeated if any sample had a replicate with a coefficient of variation greater than 20%. Replicates were read off the standard curve and multiplied by the dilution factor, and concentration was reported as the average of the replicates in pg ml−1. Concentration from unstimulated sample was subtracted from concentration from stimulated (background subtract). Negative values of background subtracts were replaced by zeros. An arbitrary value of 0.0001 was added to the background subtracts across all the samples to overcome the presence of null values raised from samples too low to be read off the standard curve.

Isotype and subclass standardized ELISA

Samples from participants vaccinated with ChAdOx1 nCoV-19 and convalescent plasma samples were assayed for anti-spike IgG1, IgG3, IgA and IgM. Samples from participants vaccinated with MenACWY were assayed for anti-spike IgA and IgM antibodies only. Standardized ELISA was used to quantify circulating SARS-CoV-2 spike-specific IgG1, IgG3, IgA and IgM responses. Full methodological details for this assay were previously published23. Briefly, ELISA plates were coated overnight with 5 µg ml−1 of SARS-CoV-2 full-length spike protein. After blocking with Blocker Casein in PBS (Thermo Fisher Scientific), samples (minimum 1:50 dilution) were incubated for 2 h at 37 °C with 300 r.p.m. shaking. Standard curve and internal controls were created from reference serum using a pool of high-titer donor serum. An alkaline phosphatase-conjugated secondary antibody (dependent on the immunoglobulin subclass or isotype being detected) was then added and incubated for 1 h at 37 °C with 300 r.p.m. shaking. Plates were developed using PNPP alkaline phosphatase substrate (Thermo Fisher Scientifc) for 1–4 h at 37 °C with 300 r.p.m. shaking and read at 405 nm when the internal control reached an OD405 of 1. Plate pass/fail criteria are described in ref. 23.

Isotype and subclass optical density ELISA

Antigen-specific IgG2, IgG4 and IgE responses were detected in the absence of an antigen-specific serum control by optical density (OD) ELISA. Detailed procedures for this assay were previously described23. Briefly, ELISA plates were coated overnight with 5 µg ml−1 of SARS-CoV-2 full-length spike protein, plus a commercial human immunoglobulin control for the antibody isotype or subclass being assayed. After blocking with Blocker Casein in PBS, test samples and pre-pandemic negative controls (minimum 1:50 dilution) were plated out for 2 h at 37 °C with 300 r.p.m. shaking. Different alkaline phosphatase-conjugated secondary antibodies were added depending on the immunoglobulin isotype or subclass being assayed for 1 h at 37 °C with 300 r.p.m. shaking. Plates were developed using PNPP alkaline phosphatase substrate for 1–4 h at 37 °C with 300 r.p.m. shaking and read at 405 nm when the immunoglobulin control reached a specified OD405. Negative cutoff calculations are described in ref. 23.

Avidity ELISA

The avidity of SARS-CoV-2 spike-specific IgG from volunteers who had a quantifiable response at day 28 was assessed. Anti-SARS-CoV-2 spike-specific total IgG antibody avidity of donor serum was assessed by sodium thiocyanate (NaSCN)-displacement ELISA. Nunc MaxiSorp ELISA plates (Thermo Fisher Scientific) were coated overnight (≥16 h) at 4 °C with 50 µl per well of 2 µg ml−1 of SARS-CoV-2 trimeric spike protein diluted in PBS. Plates were washed three times with PBS/Tween (0.05%) (PBS/T) and tapped dry. Plates were blocked for 1 h with 100 µl per well of Blocker Casein in PBS (Thermo Fisher Scientific) at 20 °C. Test samples and a positive control serum pool were diluted in blocking buffer to normalize them to an OD405 of 1, and 50 μl per well was added in duplicate to each row of the plate (except the last row, where only blocking buffer was added). Plates were incubated for 2 h at 20 °C and then washed three times with PBS/T and tapped dry. Increasing concentrations of NaSCN (Sigma-Aldrich) diluted in PBS were added at 50 μl per well to each row down the plate (1 M, 2 M, 3 M, 4 M, 5 M and 6 M) except for the first and last row, where only PBS was added. Plates were incubated for 15 min at 20 °C and then washed six times with PBS/T and tapped dry. Anti-human IgG (γ-chain specific) alkaline phosphatase antibody produced in goat (Sigma-Aldrich) was diluted 1:1,000 in blocking buffer, and 50 μl per well was added to the plate. Plates were incubated for 1 h at 20 °C and then washed three times with PBS/T and tapped dry. Then, 100 μl per well of PNPP alkaline phosphatase substrate (Thermo Fisher Scientific) was added, and plates were incubated at 20 °C. OD at 405 nm (OD405) was measured using an ELx808 absorbance reader (BioTek) until the untreated sample wells reached an OD405 of 1 (0.8–2.0). Gen5 ELISA software v3.09 (BioTek) was used to plot the test sample OD405 against concentration of NaSCN, and a spline function with smoothing factor 0.001 was fitted to the data. For each sample, concentration of NaSCN required to reduce the OD405 to 50% of that without NaSCN (IC50) was interpolated from this function and reported as a measure of avidity.

Ex vivo IFN-γ ELISpot assays

ELISpot assays were performed on freshly isolated PBMCs before and 14 d after vaccination with ChAdOx1 nCoV19, as previously described7. Assays were performed using Multiscreen IP ELISpot plates (Millipore) and were coated overnight at 4 °C with 10 μg ml−1 of human anti-IFN-γ coating antibody (clone 1-D1K, Mabtech) in carbonate buffer, before washing three times with PBS and blocking with R10 media for 2–8 h. Then, 2.5 × 105 PBMCs were added to each well of the plate, along with 13 pools of peptides covering the SARS-CoV-2 spike protein and the N-terminal tPA leader sequence at a final concentration of 10 µg ml−1 (Supplementary Table 3). Each assay was performed in triplicate and incubated for 16–18 h at 37 °C with 5% CO2.

Plates were then developed by washing six times with PBS/T, followed by the addition of 1 μg ml−1 of anti-IFN-γ detector antibody (7-B6-1-Biotin) to each well. After a 2–4-h incubation, plates were washed again, and 1:1,000 SA-ALP was added for 1–2 h. After a final wash step, plates were developed using BCIP NBT-plus chromogenic substrate (Moss).

ELISpot plates were counted using an AID automated ELISpot counter (AID Diagnostika, algorithm C), using identical settings for all plates, and spot counts were adjusted only to remove artifacts. Responses were averaged across triplicate wells, and the mean response of the unstimulated (negative control) wells was subtracted. Results are expressed as SFCs/106 PBMCs. Responses to a peptide were considered positive if background subtracted responses were >40 SFUs/106 PBMCs. If responses were >80 SFCs/106 PBMCs in the negative control wells (PBMCs without antigen) or <800 SFCs/106 PBMCs in the positive control wells (pooled Staphylococcal enterotoxin B at 0.02 μg ml−1 and phytohaemagglutinin-L at 10 μg ml−1), results were excluded from further analysis.

ICS

ICS was performed on freshly isolated PBMCs stimulated with pooled S1 and S2 peptides. Then, 3 × 106 PBMCs were resuspended in 5 ml of polypropylene FACS tubes to a volume of 1 ml in R10 media supplemented with 1 µg ml−1 of anti-human CD28 and CD49d and 1 µl of CD107a PE-Cy5 (eBioscience). S1 and S2 peptide pools (Supplementary Table 3) were added at a concentration of 2 µg ml−1. Each sample also included a positive control (S. enterotoxin B at 1 µg ml−1; Sigma Aldrich) and an unstimulated (media-only) control. Cells were incubated at 37 °C with 5% CO2 for 16–20 h, with brefeldin A (3 µg ml−1) and monensin (2 mM) (eBioscience) added after 2 h

At the end of the incubation, cells were washed in FACS buffer (PBS containing 0.1% bovine serum albumin and 0.01% NaN3) and transferred to a 96-well U-bottom tissue culture plate for staining. A surface staining cocktail was first added containing 2.5 µl of a 1:40 dilution of Aqua Live/Dead stain (Thermo Fisher Scientific) and 1 µl of BV711 CCR7 (BioLegend) in 46.5 µl of FACS buffer. Cells were incubated in the dark for 20 min and washed with FACS buffer. Then, 100 µl of CytoFix/CytoPerm solution (BD Biosciences) was added to each well and left to incubate for an additional 20 min. Cells were then washed with Perm/Wash buffer before ICS. The ICS cocktail contained 0.025 µl of CD45RA BV605, 0.025 µl of TNF-α PE-Cy7, 0.1 µl of IFN-γ FITC, 0.025 µl of CD14 e450, 0.025 µl of CD19 e450, 0.5 µl of CD3 AF700, 1 µl of IL-2 BV650, 1.25 µl of IL-5 PE, 2.5 µl of IL-13 APC, 3.5 µl of CD4 PerCP Cy5.5 and 5 µl of CD8 APC-eF780, to a total volume of 50 µl diluted in FACS buffer. Samples were stained in the dark for 30 min. Cells were washed twice with Perm/Wash buffer and twice with FACS buffer before being resuspended in 100 µl of 1% paraformaldehyde.

Compensation controls were prepared fresh for each batch using OneComp eBeads (eBioscience). Cells were kept on ice and strained through a 35-µm filter before acquisition. Cells were acquired on a five-laser BD LSRFortessa flow cytometer (BD Biosciences) using FACSDiva v8.02 (BD Biosciences), and data were analyzed in FlowJo v10.7. A hierarchical gating strategy was applied for sample analysis (Supplementary Fig. 7). A quality control process was applied to remove samples with fewer than 100,000 events in the live CD3+ gate and samples with <1% cytokine response to S. enterotoxin B (CD4+ and CD8+ IFN-γ+, CD8+ TNF-α+). A lower limit of detection was applied, and only samples with an ELISpot response greater than 200 SFCs/106 PBMCs were included in the analysis.

Statistical analysis

All statistical tests, as well as all graphical representation of the data, were performed in GraphPad Prism 8.4.3. Data are presented as medians with IQRs. To check for the normality of the data, d’Agostino–Pearson tests were used. Unpaired samples were compared using Mann–Whitney U tests, and paired samples were compared with the Wilcoxon test. All tests were two tailed, with a 5% per-comparison error rate. Bonferroni correction was used to correct for multiple comparisons. Correlations were analyzed using Spearman’s rank test. P values less than 0.05 were considered significant.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.