Donors were required to be 20 years of age or older. Blood samples were collected at Kyoto University Hospital, and HIV, HTLV-1, HBV, and HCV negative samples were used. All donors were recruited on the condition that they were healthy and did not have any serious medical conditions nor history including cancer, gastrointestinal, liver, kidney, cardiovascular, hematologic or endocrine disease. Samples were de-identified with an anonymous code assigned to each sample. Their characteristics, including age, gender, complete blood count and serology, are summarized in Table 1. Samples were obtained between July and September 2020 during the COVID-19 pandemic. Blood samples from convalescents were obtained 6 months after confirmation of SARS-CoV-2 negativity by PCR tests.

Whole blood was drawn into a BD Vacutainer CPT^TM Cell Preparation Tube with sodium heparin and processed within 2 h to isolate peripheral blood mononuclear cells (PBMCs) according to the manufacturer’s instructions. Isolated PBMCs were cryopreserved in CELLBANKER (ZENOGEN PHARMA) 1 at 8 × 10⁶ cells/ml and stored in a −150°C ultra-low temperature freezer until used in the assays. Cryopreserved PBMCs were thawed in pre-warmed X-VIVO15 (LONZA) without serum. After centrifugation, the cells were washed once and directly used for the assay as described below.

Whole blood was collected in an EDTA-2Na tube. The analysis was performed using the Automated Hematology Analyzer XN-9000 (Sysmex Corporation) at the Department of Clinical Laboratory, Kyoto University Hospital.

Whole blood was collected in a blood collection vessel, Venoject VP-P075K (Terumo), for serum isolation. Serum separator tubes were centrifuged for 4 min at 1,100 rcf at 4°C. The serum was then removed from the upper portion of the tube, aliquoted, and stored at −80°C. Anti-SARS-CoV-2 IgM/IgG levels in the serum were measured using Elecsys Anti-SARS-CoV-2 with cobas8000 (Roche Diagnostics K.K.) at the Department of Clinical Laboratory, Kyoto University Hospital. Anti-CMV IgG levels were measured using chemiluminescence immunoassays (CLIA) at LSI medience (Tokyo, Japan). The cut-off values for Anti-SARS-CoV-2 IgM/IgG and Anti-CMV IgG were 1.0 cutoff index (COI) and 6.0 (AU/ml), respectively.

PepTivator SARS-CoV-2 peptide pools (Miltenyi Biotech) were diluted in distilled water (DW) and used for the SARS-CoV-2-reactive T cell stimulation. The S peptide pool contained 15-mer peptides that overlapped by 11 amino acids and spanned the immunodominant sequence domains of the spike (S) glycoprotein of SARS-CoV-2. The S1, N and M peptide pools covered the entire sequence of the corresponding proteins. All four peptide pools were equally combined at the final concentration of 0.6 nmol/ml in the culture medium. CMV pp65 peptide pools (Miltenyi Biotech) were used for CMV-reactive T cell stimulation.

PBMCs were cultured in 150 μl X-VIVO15 medium supplemented with 5% human AB serum (Sigma) for 20 h at 37°C in the presence of SARS-CoV-2 or CMV peptide pools (0.6 nmol/ml) and CD40 blocking antibody (Miltenyi Biotech) (0.5 ug/ml) in 96-well U-bottom plates at 2 × 10⁶ PBMCs per well. Stimulation with DW at an equal volume was performed as a negative control. After the stimulation, the cells were stained with Ghost Dye^TM Red 710 (TONBO) to discriminate viable from non-viable cells. The cells were then washed and stained with fluorochrome-conjugated surface antibodies at pre-titrated concentrations in the presence of FcR blocking (Miltenyi Biotech) for 20 min at 4°C. Antibodies used in the AIM assay are listed in Supplementary Table 1. Stained cells were fixed and permeabilized for 20 min at room temperature using the eBioscience Foxp3/Transcription Factor Staining Buffer Set. The cells were then stained with anti-IRF4 antibody (1:500 dilution) for 30 min at room temperature. After the final wash, the cells were resuspended in 200 μl PBS with 2% FBS (FACS buffer) for flow cytometry analysis. Supernatants were harvested at 20 h post-stimulation for the multiplex detection of cytokines.

PBMCs were first stained with 2 µM Cell Trace Violet (CTV) (Thermo Fisher Scientific) for 15 min and washed with X-VIVO15 medium. The labeled cells were cultured in 200 μl X-VIVO15 medium supplemented with 2% human AB serum for 6 days at 37°C in the presence of SARS-CoV-2 peptide pools (0.6 nmol/ml) at 1 × 10⁶ PBMCs per well or Dynabeads Human T-Activator CD3/CD28 (Thermo Fisher Scientific) at 1 × 10⁵ PBMCs per well in 96-well U-bottom plates. A stimulation with an equal volume of DW was performed as the negative control. The medium change was conducted every three days. Five hours before ICS, the medium was replaced to fresh medium with the respective peptide pool and brefeldin A (BioLegend) (1:1,000 dilution). After stimulation, the cells were stained with Ghost Dye^TM Red 710 to discriminate viable from non-viable cells. Cells were then washed and stained with fluorochrome-conjugated surface antibodies at pre-titrated concentrations in the presence of FcR blocking for 20 min at 4°C. The antibodies used in the ICS assay are listed in Supplementary Table 2. Stained cells were then fixed and permeabilized using IC fixation buffer and Permeabilization buffer for 20 min at room temperature. The cells were then stained for intracellular IL-2, IL-17A, IFNγ, and TNFα (1:25 dilution for each) for 30 min at room temperature. After the final wash, the cells were resuspended in 200 μl FACS buffer for flow cytometry analysis.

All the samples in the AIM and ICS assays were acquired using a BD FACSAria™ II Cell Sorter (BD Biosciences). FCS 3.0 data files were exported and analyzed with FlowJo software version 10.6.1. Detailed gating strategies for individual markers are described in Supplementary Figures 2, 4, and 5.

Supernatants were collected from the culture plate after 20 h of AIM assays and after 3 or 6 days of peptide pool stimulation. They were then stored in 96-well plates at −80 °C until use. Cytokines in the cell culture supernatants were quantified using a human Th cytokine panel (12-plex) kit (BioLegend) according to the manufacturer’s instruction. Briefly, the supernatants were mixed with beads coated with capture antibodies specific for IL-5, IL-13, IL-2, IL-6, IL-9, IL-10, IFNγ, TNFα, IL-17A, IL-17F, IL-4 and IL-22 and incubated on a 96-well filter plate for 2 h. The beads were washed and incubated with biotin-labeled detection antibodies for 1 h, followed by a final incubation with streptavidin-PE. The beads were analyzed by flow cytometry using a BD LSR flow cytometer (BD Biosciences). Analysis was performed using the LEGENDplex analysis software v8.0, which distinguishes the 12 different analytes based on the bead size and internal dye.

Dimensionality reduction of multi-color flow cytometry data obtained from the AIM assays was performed using OMIQ software. FCS 3.0 data from all donors were imported. Up to 117 cells of AIM⁺CD4⁺ and 37 cells of AIM⁺CD8⁺ T cells from each donor were subsampled and merged for the analysis. These subsampling counts were derived from the median in the corresponding subsets. The markers we applied to the opt-SNE analysis are described in the figure legends. The parameters used were as follows: Max Iterations = 1,000, opt-SNE End = 5,000, Perplexity = 30, Theta = 0.5, Components = 2, Random Seed = 6,925, Verbosity = 25.

Subset definitions and gating strategies are outlined in the text and/or figure legends. The absolute numbers of a defined subset (e.g., NP, CM, EM, TEMRA, and AIM⁺ cells) for CD4⁺ and CD8⁺ T cells were obtained by individually multiplying the number of CD4⁺ or CD8⁺ T cells by the percentage of the corresponding subsets. The percentage of AIM⁺ T cells was calculated by subtracting the percentage of AIM⁺ cells after DW treatment from that after peptide stimulation. To calculate the percentages of each subset, samples that a percentage of AIM⁺ cells in CD4 or CD8 T cells of at least 0.001% are indicated in the graphs, since a very low frequency of AIM⁺ cells influences the results. The stimulation index (SI) was calculated by dividing the percentage of AIM⁺ cells after SARS-CoV-2 peptide pool stimulation with that after DW treatment. If the percentage of AIM⁺ cells after DW stimulation equaled 0, the minimum value across each cohort was used instead.

All statistical analyses in this study were performed using GraphPad Prism 9.0. The statistical details, such as number of subjects, cohorts, statistical test and significance of the experiments, are provided in the respective figure legends. Non-parametric tests were primarily used for the statistical analysis in this study, because most of the sample values had some outliers and did not follow a Gaussian distribution, which was assessed by the Shapiro-Wilk normality test. Two-tailed Mann-Whitney and Wilcoxon tests were applied for unpaired and paired group comparisons, respectively. Correlation analyses were performed using the Spearman rank correlation test.

Thirty young (median age, 22 years) and twenty-six elderly (median age, 72 years), healthy, uninfected participants were recruited (Table 1). All donors were anti-SARS-CoV-2 Ig negative (Table 1). PBMCs were isolated and used for further analyses. The numbers of white blood cells and monocytes did not differ between the two groups, but lymphocyte and CD3⁺ T cell numbers were approximately 20% lower in the elderly (Table 1; Supplementary Figure 1). The number of CD4⁺ T cells was comparable between the two groups, while the number of CD8⁺ T cells in the elderly was approximately 50% that of the young (Table 1; Supplementary Figure 1) 1), consistent with previous reports showing a more profound reduction of CD8⁺ T cells with age (Czesnikiewicz-Guzik et al., 2008; Wertheimer et al., 2014).

To identify and quantify SARS-CoV-2-reactive T cells in those samples, we utilized TCR-dependent AIM assays, a commonly used and cytokine-independent system to detect antigen-specific T cells (Dan et al., 2016; Havenar-Daughton et al., 2016; Herati et al., 2017; Reiss et al., 2017; Morou et al., 2019), including SARS-CoV-2-reactive T cells in healthy individuals (Bacher et al., 2020; Braun et al., 2020; Grifoni et al., 2020; Mateus et al., 2020; Weiskopf et al., 2020). T cell responses have been detected mainly against the spike (S), membrane (M), and nucleocapsid (N) proteins of SARS-CoV-2 (Braun et al., 2020; Grifoni et al., 2020; Le Bert et al., 2020; Mateus et al., 2020; Sekine et al., 2020; Thieme et al., 2020; Weiskopf et al., 2020). Therefore, to reduce the required quantity of blood samples from an ethical perspective, we used overlapping peptide mixtures that could stimulate both CD4⁺ and CD8⁺ T cells and cover immunodominant sequences of SARS-CoV-2, including the N-terminal and C-terminal regions of S (amino acid residues 1–692 and 885–1,273, respectively), as well as N and M proteins, for PBMC stimulation. Flow cytometry identified antigen-specific AIM⁺ cells and their phenotypes in stimulated PBMCs. The gating strategy is shown in Supplementary Figure 2A. AIM⁺ cells were defined by the co-expression of CD137 and IRF4, which sensitively marks peptide-specific T cells with low background and reflects TCR signal strength (Wu et al., 2019). Utilizing CMV-pp65 peptide stimulation, AIM⁺(CD137⁺IRF4⁺) CD4⁺ and CD8⁺ T cells were substantially detected in CMV seropositive (CMV⁺) individuals but not in CMV seronegative (CMV⁻) individuals (Supplementary Figure 2B). Similar results were obtained using more common AIMs (CD137 and CD154) (Reiss et al., 2017; Grifoni et al., 2020) in CD4⁺ T cells (Supplementary Figure 2B). Furthermore, based on SARS-COV-2 peptide stimulations, more than approximately 30-fold higher frequencies of AIM⁺ CD4⁺ and CD8⁺ T cells were detected in convalescent patients with COVID-19 compared to unexposed individuals (Supplementary Figure 2B), confirming the accuracy of the detection system and the AIMs.

SARS-CoV-2-specific responses (the percentage of peptide-stimulated AIM⁺ cells subtracted by DW-stimulated cells) were detected in CD4⁺ and CD8⁺ T cells from both young and elderly unexposed individuals (Figures 1A,B). The frequencies and numbers of SARS-CoV-2-reactive CD4⁺ T cells were comparable between the two groups (Figure 1C). Similar results were obtained using another AIM combination (CD137 and CD154) (Supplementary Figure 2C, D), and AIM⁺ frequencies in total CD4⁺ T cells using these two AIM combinations exhibited a positive correlation (Supplementary Figure 2E). On the other hand, the frequency of SARS-CoV-2-reactive cells among total CD8⁺ T cells was significantly higher in the elderly than in the young participants, although the number of SARS-CoV-2-reactive CD8⁺ T cells/L in the blood was not different between the two groups owing to the low CD8⁺ T cell count in the peripheral blood of the elderly (Figure 1C; Table 1; Supplementary Figure 1). Based on the definition of responders using the stimulation index, which was calculated as peptide-specific responses/DW responses (Grifoni et al., 2020; Weiskopf et al., 2020) > 3, we found that CD4⁺ T cells in 50.0% (15/30) and 34.6% (9/26) and CD8⁺ T cells in 30.0% (9/30) and 30.8% (8/23) of young and elderly individuals, respectively, showed a significant level of pre-existing SARS-CoV-2-reactive T cells (Figure 1D). These results indicate that young and elderly unexposed individuals have comparable numbers of SARS-CoV-2-reactive T cells.

To obtain a global view of the SARS-CoV-2-reactive T cell phenotype, the data were visualized by optimized t-Distributed Stochastic Neighbor Embedding (opt-SNE) (Belkina et al., 2019). AIM⁺ cells from the young and elderly were overlayed on total AIM⁺ cells from both groups to visualize marker expression patterns. Phenotypically related cells, namely NP, central memory (CM), effector memory (EM), and terminally differentiated effector memory T cells re-expressing CD45RA (TEMRA), were identified using opt-SNE based on CCR7 and CD45RA expression (Figure 2). Most AIM⁺CD4⁺ T cells expressed CD28 but not CD57 (Figure 2A), a marker of T cell senescence (Brenchley et al., 2003). PD-1^high cells were observed in memory fractions (Figure 2A), with a portion of them probably representing follicular helper T cells, as previously reported in COVID-19 patients (Mathew et al., 2020). Nevertheless, the distribution of AIM⁺CD4⁺ T cells was similar between the elderly and young (Figure 2A; Supplementary Figure 3). However, PD-1^high MP T cells were rare, and CD57^high senescent-like T cells were abundant among AIM⁺CD8⁺ T cells, suggesting that SARS-CoV-2 cross-reactive CD8⁺ T cells tend to be senescent rather than exhausted (Figure 2B). Notably, AIM⁺CD8⁺ T cells from the elderly exhibited less distribution in the CCR7⁺CD45RA⁺ NP region and more accumulation in the CD57^high region (Figure 2B). Furthermore, IFR4^high CD4⁺ T cells were distributed mainly in the memory fraction, whereas IFR4^high CD8⁺ T cells accumulated in the NP region (Figures 2A,B, and Supplementary Figure 3). These results suggest that SARS-CoV-2-reactive T cell phenotypes from the young and elderly are similar in CD4⁺ T cells but different for CD8⁺ T cells. Thus, NP CD8⁺ T cells with a high affinity towards SARS-CoV-2 and CD57⁺ senescent SARS-CoV-2-reactive T cell populations were lower and higher, respectively, in the elderly compared with those in the young.

Next, we compared the exact frequency and number of each SARS-CoV-2-reactive T cell subpopulation between young and elderly individuals using a manual gating strategy on the flow cytometry data. T cell populations were divided into four fractions based on their expression markers: NP (CD45RA⁺CCD7⁺CD28⁺), CM (CD45RA⁺CCR7⁻), EM (CCR7 CD45RA⁻), and TEMRA (CD45RA⁺CCD7⁻) (Jameson and Masopust, 2018; Thome et al., 2014) (Supplementary Figure 4A). Following the stimulation of PBMCs with CMV peptide, less NP and more MP (EM and TEMRA) AIM⁺ CD4⁺ and CD8⁺ T cells were found in CMV⁺ individuals compared to CMV⁻ individuals (Supplementary Figure 4B). A similar trend was observed in T cells from convalescent patients and uninfected individuals upon SARS-CoV-2 peptide stimulation (Supplementary Figure 4B), consistent with a previous report (Braun et al., 2020).

Although NP cells represented a dominant fraction of CD4⁺ T cells in both the young and elderly, the percentages [median (interquartile range: IQR)] of the NP [54.7 (15.9)% in the young, 41.3 (21.5)% in the elderly] and CM [28.0 (8.4)% in the young, 34.2 (15.9)% in the elderly] fractions of total CD4⁺ T cells were significantly lower and higher, respectively, in the elderly compared to the young participants, while the EM and TEMRA fractions were comparable between the two groups (Figures 3A,B). A similar trend was observed in the number of cells in each fraction of whole blood between the two groups (Figure 3B). On the other hand, SARS-CoV-2-reactive CD4⁺ T cells were substantially detected in NP fraction in both groups, but were less abundant [16.6 (20.6)% in the young, 10.8 (15.9)% in the elderly] but were highly enriched in the memory fractions (Figures 3C,D; Supplementary Figure 4C), consistent with previous reports (Bacher et al., 2020; Braun et al., 2020; Mateus et al., 2020; Weiskopf et al., 2020). These pre-existing MP SARS-CoV-2-reactive T cells, so-called cross-reactive T cells, were enriched in the CM [44.4 (16.2)% in the young, 52.8 (25.2)% in the elderly] and EM [28.3 (21.9)% in the young, 26.2 (20.4)% in the elderly] fractions (Figures 3C,D). The percentages of NP, CM, and EM cells of the total CD4⁺ T cell population in each individual exhibited positive correlations with their percentages of AIM⁺ cells in each fraction (Figure 3E). However, there were no significant differences in the frequency or number of SARS-CoV-2-reactive CD4⁺ T cells between the young and elderly in any fraction (Figures 3C,D). These results indicate that SARS-CoV-2-reactive T cells are mainly detected in antigen-primed CM fractions in CD4⁺ T cells and that these MP CD4⁺ T cells that were cross-reactive to SARS-CoV-2 were well retained irrespective of age.

Next, we examined the cytokine production and proliferation in response to SARS-CoV-2 peptide stimulation to assess the functionality and polarization of the SARS-CoV-2-specific CD4⁺ T cell response. Previous reports have shown that SARS-CoV-2 peptide stimulation significantly induces IFNγ production by CD4⁺ T cells from unexposed individuals (Braun et al., 2020; Grifoni et al., 2020; Le Bert et al., 2020; Mateus et al., 2020; Meckiff et al., 2020; Sekine et al., 2020; Weiskopf et al., 2020) and that the Th1 response is associated with an effective resolution of infection and symptoms in COVID-19 (Chen et al., 2020; Neidleman et al., 2020). CTV-labeled PBMCs were cultured with the peptide pools, and cytokine expression and proliferation estimated by CTV dilution were analyzed at days 3 and 6. The gating strategy is shown in Supplementary Figure 5A. CD4⁺ T cells expressing intracellular Th1 cytokines, including IFNγ, TNFα, and IL-2, were detected in both the young and elderly (Figure 4A and Supplementary Figures 5A,C), and these cytokines were significantly detected in culture supernatants (Supplementary Figure 5D). Th2 cytokines, such as IL-4 and IL-5, as well as IL-17, were produced at markedly low levels (Supplementary Figure 5D), consistent with a previous report (Grifoni et al., 2020). Although the frequencies of intracellular IFNγ⁺ cells tended to be lower and those of IL-17⁺ cells tended to be higher in the elderly, there was no statistically significant difference between elderly and young participants (Figure 4A). Indeed, most participants in the two groups showed similar frequencies (Figure 4A). Furthermore, no significant difference in the percentages of CTV-diluted FSC^high proliferated T cells after peptide stimulation between the young and elderly was observed (Figure 4B), suggesting that SARS-CoV-2-reactive CD4⁺ T cells have comparable proliferation capacity in the two populations. Cytokine-producing cells were mainly detected in the CTV^low cell fraction of both groups, but the percentages of IFNγ⁺ cells tended to be lower in the elderly compared to the young (Supplementary Figure 5E). These data strongly suggest that SARS-CoV-2-reactive CD4⁺ T cells in the elderly can proliferate and produce helper cytokines upon SARS-CoV-2 recognition despite the tendency for less IFNγ production.

A similar analysis was performed on CD8⁺ T cells. Compared to CD4⁺ T cells, the percentage of NP cells among total CD8⁺ T cells was significantly lower in the elderly [8.2 (15.2)%] compared to the young [49.5 (26.7)%]; however, the percentages of CM and TEMRA cells were higher in the elderly [17.7 (7.8)% and 34.6 (21.9)%, respectively] compared to the young [5.9 (3.0)% and 14.9 (17.0)%, respectively] (Figures 5A,B). Moreover, the number of NP cells was remarkably fewer in the elderly (Figure 5B). On the other hand, SARS-CoV-2-reactive AIM⁺ CD8⁺ T cells were more diffusely distributed in each phenotype with significant individual variability (Figures 5C,D). Notably, elderly individuals had a significantly lower frequency and number of NP SARS-CoV-2-reactive CD8⁺ T cells compared with the young [28.6 (36.2)% and 6.19 (1.7) × 10³ cells/L in young; 0.0 (7.9)% and 0.0 (3.3) × 10³ cells/L in elderly] (Figures 5C,D). On the other hand, the percentages as well as numbers of MP (cross-reactive) SARS-CoV-2-reactive CD8⁺ T cell were significantly higher in the elderly (Supplementary Figure 4C). Furthermore, SARS-CoV-2-reactive CD8⁺ T cells were most frequently observed in the TEMRA fraction from the elderly [49.0 (36.0)% and 10.9 (14.3) × 10³ cells/L), and their prevalence was significantly higher in the elderly than that in the young [21.1 (36.0)% and 4.3 (11.3) × 10³ cells/L] (Figures 5C,D). The percentages of each phenotype of CD8⁺ T cells exhibited weak correlations with the percentages of AIM⁺ cells in each fraction (Figure 5E). These data indicate that young individuals have more SARS-CoV-2-reactive NP CD8⁺ T cells than do elderly individuals. Additionally, MP CD8⁺ T cells cross-reactive to SARS-CoV-2 in the elderly were biased towards TEMRA cells, which exhibit highly differentiated phenotypes (Thome et al., 2014; Jameson and Masopust, 2018).

Regarding functionality, CD8⁺ T cells from the elderly showed higher proportions of IFNγ⁺ and TNFα⁺ cells compared to those from young participants after 6 days of culture with SARS-CoV-2 peptides (Figure 6A and Supplementary Figures 6A,B). The percentage of CTV-diluted FSC^high cells did not differ between the two populations (Figure 6B). Considering that the percentage in AIM⁺ CD8⁺ T cells was higher in the elderly than in young individuals (Figure 1C), SARS-CoV-2-reactive CD8⁺ T cells from the elderly may have a lower proliferative capacity. In contrast to CD4⁺ T cells, the percentage of cytokine⁺ cells was higher in less proliferated CTV^high cells, with this tendency more obvious in the elderly (Supplementary Figures 6C). These features may be consistent with the elderly SARS-CoV-2 reactive CD8⁺ T cells being highly enriched in the TEMRA fraction (Figure 5D).

TEMRA cells include CD28⁻ terminally differentiated T cells with compromised proliferation capacities, as well as CD57⁺ senescent T cells with a high potential for inflammatory cytokine production (Brenchley et al., 2003; Henson et al., 2015). Therefore, we further assessed CD28 and CD57 expression in AIM⁺ T cells. We found that the percentage and number of CD57⁺ cells, but not CD28⁻ cells, among AIM⁺ CD8⁺ T cells from the elderly were significantly higher than those from young participants (Figure 6C) and that CD28⁻ or CD57⁺ AIM⁺ CD4⁺ T cells were rare (Supplementary Figure 7). These findings suggest that senescent SARS-CoV-2-reactive CD8⁺ T cells with compromised proliferation capacity and high effector functions tended to accumulate in the elderly.

We noticed that young individuals are divided into two groups based on the percentage of CD57⁺ cells (high, > 60%, n = 11 and low, < 50%, n = 19) in SARS-CoV-2-reactive CD8⁺ T cells (Figure 6C). All elderly individuals were CMV⁺, but approximately half (46.7%) of young individuals were CMV⁺ (Table 1). Because CMV infection markedly affects CD8⁺ T cell homeostasis and the generation of senescent T cells (Nikolich-Zugich, 2008; Wertheimer et al., 2014; Klenerman and Oxenius, 2016), we re-analyzed the AIM⁺ cell phenotypes from young individuals as CMV⁺ (n = 14) and CMV⁻ (n = 16). Consistent with previous studies (Nikolich-Zugich, 2008; Wertheimer et al., 2014), young CMV⁺ individuals tended to have proportionally less NP and more TEMRA CD8⁺ T cells (Figure 7A). Importantly, young CMV⁺ individuals showed significantly more TEMRA as well as CD57⁺ senescent SARS-CoV-2-reactive CD8⁺ T cells (Figures 7B,C). SARS-CoV-2-reactive TEMRA CD4⁺ T cells were rare, and no marked difference was observed in CD57⁺CD4⁺ T cells (Supplementary Figure 8). These results indicate that CMV infection affects the SARS-CoV-2-reactive CD8⁺ T cell frequency in young individuals.