The study included 6 COVID-19 convalescent donors and 2 healthy controls (Table 1). The convalescent donors were all tested for SARS-CoV-2 using reverse transcriptase polymerase chain reaction (RT-PCR) in nasopharyngeal samples between March and April 2020. The eligibility criteria included an age of 21–65 years, a history of COVID-19 with a documented positive RT-PCR test for SARS-CoV-2. At the time of this study, all donors tested negative for SARS-CoV-2. The median age of the convalescent donors was 37 years (range 23–41), 3 were women and 3 were men. The median duration until a negative PCR for SARS-CoV-2 was 13 days (range 5–17). Two of the donors presented with bilateral pneumonia but did not require hospitalization. Only 1 of the donors underwent treatment with oral hydroxychloroquine plus azithromycin plus lopinavir/ritonavir, while the other was treated with oral hydroxychloroquine plus azithromycin. The study enrolled two healthy donors who had not been exposed to COVID-19 patients and tested negative for anti-SARS-CoV-2 antibodies in June 2020. All participants granted their written consent, and the study was approved by the Hospital Institution Review Board (IRB number: 254/20).

Peripheral blood mononuclear cells (PBMCs) from healthy donors and convalescent donors were isolated from their peripheral blood by density gradient centrifugation using Ficoll-Paque (GE Healthcare, Chicago, IL, United States). Briefly, the cells were rested overnight (o/n) at 37°C in TexMACS Medium (Miltenyi Biotec, Bergisch Gladbach, Germany) supplemented with 10% AB serum (Sigma-Aldrich, Saint Louis, MO, United States) and 1% penicillin/streptomycin (Sigma-Aldrich). The following day 1 × 10⁶ cells were stimulated with pooled or individual overlapping SARS-CoV-2 peptides at a final concentration of 0.6 nmol/mL. For the positive control, 1 × 10⁶ cells were stimulated in the presence of the plate-bound stimulator OKT3 at a final concentration of 2.8 μg/mL (mouse anti-human CD3 Clone OKT3, BD Biosciences). Cells with SARS-CoV-2 peptides and positive control were co-stimulated with CD28/CD49d at a final concentration of 5 μg/mL (anti-human CD28/CD49d Purified Clone L293 L25, BD Biosciences). Basal interferon gamma (IFN-γ) production by PBMCs was included as a background control in the absence of stimulation and co-stimulation. The peptide pools were short 15-mer peptides with 11 amino acid overlaps that can bind MHC class I and class II complexes and were therefore able to stimulate CD4⁺ and CD8⁺ T cells. The peptides cover the immunodominant sequence domains of the surface glycoprotein S, the complete sequence of the nucleocapsid phosphoprotein N and the membrane glycoprotein M (GenBank MN908947.3, Protein QHD43416.1, Protein QHD43423.2, Protein QHD43419.1; Miltenyi Biotec, Germany). After 5 h of stimulation, the cells were labeled with IFN-γ Catch Reagent (IFN-γ Secretion Assay-Detection Kit, human Miltenyi Biotec) containing bispecific antibodies for CD45 and IFN-γ, which were secreted by the stimulated target cells. After the secretion phase, the cell surface-bound IFN-γ was targeted using the IFN-γ PE antibody included in the kit.

We analyzed the cell composition of the T cells specifically activated by SARS-CoV-2 in the IFN-γ assay by subtracting the basal cytokine response from the background control. We stained the cell surface for 20 min at 4°C using the following fluorochrome-conjugated antibodies titrated to their optimal concentrations: CD45RA FITC (BD Pharmingen), CD27 APC (BD Pharmingen), CD3 VioGreen (Miltenyi Biotec), CD4 PECy7 (BD Pharmingen), CD8 APC Cy7 (BD Pharmingen), and 7AAD (BD Horizon). For the Treg panel CD25 BV421 (BD Horizon) and CD127 PE-CF594 (BD Horizon), were used. For the activation panel, HLA-DR BV 421 (BD Pharmingen), CD69 BV421 (Biolegend), and CD25 BV421 (BD Horizon), were used. For the exhaustion panel PD1 AF700 (Biolegend) and NKG2A BV421 (Biolegend) were used. For the chemokine panel, CD103 BV421 (BD Horizon) and CCR7 PE-CF594 (BD Horizon) were used. Cell acquisition was performed using a Navios cytometer (Beckman Coulter), acquiring an average of 200,000 cells. The analysis was performed using FlowJo 10.7.1 (FlowJo LLC).

CD45RA^– memory T lymphocytes from the convalescent donor were thawed and stimulated with interleukin (IL)-15 to obtain an activated phenotype. Cells were incubated in TexMACS Medium (Miltenyi Biotec, Germany) supplemented with 5% AB serum (Sigma-Aldrich, Saint Louis, MO, United States), 1% penicillin/streptomycin (Sigma-Aldrich, Saint Louis, MO, United States) and 50 ng/mL of IL-15 o/n and for 72 h. After that time the cells were harvested and the phenotypic assay was performed. The same culture without IL-15 was run in parallel as a control.

The criteria for selecting convalescent donors were as follows: (1) IFN-γ secretion upon activation with the three SARS-CoV-2-specific peptides (M, N, S) and, (2) the most frequent human leukocyte antigen (HLA) typing to cover most of the population. The HLA phenotype of the convalescent donor was performed at the Centro de Transfusión of the Comunidad of Madrid on two independent samples by SSO and NGS: A^∗02:01,A^∗24:02/B^∗44:02,B^∗51:01/C^∗16:02,C^∗16:04/DRB1^∗07:01,DRB1^∗11:03/DQB1^∗02:02,DQB1^∗03:01.

Non-mobilized apheresis was performed at the Bone Marrow Transplantation and Cell Therapy Unit of University Hospital La Paz (Madrid, Spain) using a CliniMACS Plus cell separation system (Miltenyi Biotec). The donor provided written informed consent, and the study was conducted according to the Declaration of Helsinki protocol and the guidelines of the local ethics committee (IRB number 5579). The Unit was responsible for complying with the requirements regarding the quality and safety of the donation, obtention, storage, distribution, and preservation of human cells and tissues under the Spanish specific regulation. Following apheresis, CD45RA⁺ cells were depleted by immunomagnetic separation using a CliniMACS CD45RA Reagent and the CliniMACS Plus system (both from Miltenyi Biotec), following the manufacturer’s instructions. CD45RA^– cells were frozen using autologous plasma plus 5% dimethyl sulfoxide (DMSO) and stored. We were able to cryopreserve 30 aliquots at various doses according to the trial design. The viability, purity, phenotype and spectratyping of the CD45RA^– fraction were analyzed by flow cytometry (FCM).

Most of the CDR3-encoding regions of the TCRV- and TCRV-γ genes were amplified using 2 V-J multi-primer sets for each locus and one additional multi-primer set covering D-J TCRV-β (Vitro, Master Diagnostica, Spain). Primers marked at their 5′ end with 6-FAM fluorochrome enabled the denatured fragment size analysis by capillary electrophoresis (ABI3130 DNA-analyzer) and Genemapper software (Thermo Fisher Scientific, United States).

The quantitative variables are expressed as mean ± standard deviation (SD), while the qualitative variables are expressed as percentages (%). A two-tailed Mann–Whitney non-parametric test was used for comparison means for the non-paired samples using GraphPad Prism (version 8.0.0 for Windows, GraphPad Software, San Diego, CA, United States). A P-value < 0.05 was considered statistically significant.

We detected the presence of a SARS-CoV-2-specific population in both subsets of naïve CD45RA⁺ and memory CD45RA^– T cells in the PBMCs of the convalescent donors but not in the healthy controls (Table 2 and Figure 1). The subsets also showed reactivity for the single peptides M, N, and S (data not shown). The mean CD45RA^–CD3⁺ population in the convalescent donors was 90.01%. IFN-γ expression within the CD45RA^–CD3⁺ population was 1.12%, whereas IFN-γ expression within the CD45RA⁺CD3⁺ population was 0.40% (P = 0.065) (Table 2). We detected no IFN-γ expression in the healthy individuals. Despite the cohort’s small size, we found no synergistic effect on the percentage of IFN-γ when the three peptides were mixed when compared with the single peptides (data not shown).

We then sought to determine whether both CD8⁺ and CD4⁺ subsets contained specific SARS-CoV-2 T cells within the PBMCs of the convalescent donors. Among all subsets studied, we observed that CD45RA^–CD4⁺ and CD45RA^–CD8⁺ cells expressed 1.25 and 0.85% of IFN-γ, respectively (P = 0.132) (Table 2, Figure 2, and data not shown). Thus, all of the convalescent donors who recovered from COVID-19 generated CD4⁺ T and CD8⁺ T-cell responses against SARS-CoV-2 within the memory CD45RA^– T-cell population.

We then analyzed the T central memory (CM) (CD45RA^–CD3⁺CD27⁺) and T effector memory (EM) (CD45RA^–CD3⁺CD27^–) compartments. Although there were no significant differences, we observed responses to the SARS-CoV-2-specific peptides within all subpopulations. Within the CD4⁺ and CD8⁺ CM T-cell subsets, we detected a mean of 1.26 and 0.79% of IFN-γ⁺ cells, respectively (P = 0.132). When examining the CD4⁺ and CD8⁺ EM T-cell subsets, we detected a mean of 1.25 and 1.06% of IFN-γ⁺ cells, respectively (P = 0.108) (Table 2 and Figure 3). These data demonstrate the presence of a population of memory T cells specific for SARS-CoV-2 within the CD45RA^–CD3⁺ memory T cells.

After depletion of the CD45RA⁺ cells (as described in the Materials and Methods section), 99.8% of the cells were CD45RA^–CD3⁺, and most of the CD45RO⁺ cells were CD4⁺ with a high CD4/CD8 ratio. The viability of the cells after thawing was 98–99% (data not shown).

After CD45RA depletion, the percentage of CD45RA^–CD3⁺ cells were 80.05%. Within that population, 83.56% of the cells were CD4⁺ and 14.37% were CD8⁺ T cells. After exposure to the three SARS-CoV-2-specific peptides, the CD3⁺, CD3⁺CD4⁺, and CD3⁺CD8⁺ subsets expressed IFN-γ at a rate of 0.36, 0.38, and 0.31%, respectively (Table 3).

Most of the CD4⁺ and CD8⁺ cells had a CM phenotype (89.1 and 61.6%, respectively). Both the CM and EM subpopulations expressed IFN-γ after exposure to the three peptides. Thus, we found that 0.38, 0.70, and 0.39% of the cells within the CD4⁺ CM, CD8⁺ CM, and CD4⁺ EM subsets expressed IFN-γ, respectively. We found no specific IFN-γ expression within the CD8⁺ EM subset (Table 3).

We then characterized in-depth the phenotypic expression of the activation, exhaustion, and Treg markers within the CD45RA^– population of the convalescent donor. The percentage of Treg defined by CD127^lowCD25⁺ was 11.26%. We found that the CD45RA^–CD3⁺ cells expressed the activation marker HLA-DR (19.40%) in both subsets of CD4⁺ (16.70%) and CD8⁺ cells (29.19%). After exposure to the three SARS-CoV-2-specific peptides, we observed that 0.93, 0.70, and 1.21% of the cells within the CD3⁺HLA^–DR⁺, CD4⁺HLA^–DR⁺, and CD8⁺HLA-DR⁺ populations expressed IFN-γ, respectively. The CD45RA^–CD3⁺ cells also expressed the exhaustion marker NKG2A (1.88%), and this expression was 10.24% within the CD8⁺ subset, whereas it was nearly undetectable in the CD4⁺ cells (0.14%). We found no IFN-γ expression within the NKG2A⁺ population. In addition, 0.70% of the CD45RA^–CD3⁺ cells expressed the exhaustion marker PD-1, and this expression was 1.43% within the CD4⁺ subset and 0.13% within the CD8⁺ cells. We found 0.62% of IFN-γ⁺ cells within the CD45RA^–CD3⁺CD4⁺PD-1⁺ population after exposure to the three peptides (Table 3).

The polyclonal distribution for both TCR-β and TCR-γ CDR3-encoding regions was almost identical among the controls and the CD45RA^– population, whereas different oligoclonal fragments were observed in the CD45RA⁺ population. Three of the oligoclonal fragments seen in the CD45RA⁺ cells were also identified in the CD45RA^– cells (Figure 4).

IL-15 is an essential cytokine for memory T cells that induces the activation, proliferation, and survival of T cells. After incubating the CD45RA^– T cells with IL-15, we observed an increase in the activation markers HLA-DR, CD69, and CD25 after 72 h of incubation (2.63-fold, 29.59-fold, and 8.52-fold, respectively) when compared with the o/n incubation (1.15-fold, 3.18-fold, and 1.50-fold) (Table 3). The expression of the exhaustion markers NKG2A and PD-1 was also higher after 72 h of incubation (Supplementary Figures 1–5). We then examined the expression of chemokine CCR7 and integrin CD103, which are important for the homing of T-cells to the respiratory tract (Campbell et al., 2001; Uss et al., 2006). We observed that most of the CD3⁺CD4⁺ cells expressed CCR7⁺ cells (88.13%), whereas the CD3⁺CD8⁺ subpopulation expressed 45.42% of CCR7⁺ cells. As expected, CD103 expression was low in the peripheral blood (Uss et al., 2006). We detected an expression of 2.18, 5.96, and 1.38% in the CD3⁺CD103⁺, CD3⁺CD8⁺CD103⁺, and CD3⁺CD4⁺CD103⁺ compartments, respectively. Although the fold increase was not particularly remarkable in the CD45RA^–CD3⁺ cells after 72 h of incubation (1.00-fold CCR7 and 1.24-fold CD103), the increase was higher (1.40) within the CD8⁺ subpopulation (Table 3 and Supplementary Figures 6,7).