Visual Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) induces T cell, B cell, and Ab responses that are detected for several months in recovered individuals. Whether this response resembles a typical respiratory viral infection is a matter of debate. In this study, we followed T cell and Ab responses in 24 mainly nonhospitalized human subjects who had recovered from PCR-confirmed SARS-CoV-2 infection at two time points (median of 45 and 145 d after symptom onset). Ab responses were detected in 95% of subjects, with a strong correlation between plasma and salivary anti-spike (anti-S) and anti—receptor binding domain IgG, as well as a correlation between circulating T follicular helper cells and the SARS-CoV-2–specific IgG response. T cell responses to SARS-CoV-2 peptides were determined using intracellular cytokine staining, activation markers, proliferation, and cytokine secretion. All study subjects had a T cell response to at least one SARS-CoV-2 Ag based on at least one T cell assay. CD4+ responses were largely of the Th1 phenotype, but with a lower ratio of IFN-γ– to IL-2–producing cells and a lower frequency of CD8+:CD4+ T cells than in influenza A virus (IAV)-specific memory responses within the same subjects. Analysis of secreted molecules also revealed a lower ratio of IFN-γ to IL-2 and an altered cytotoxic profile for SARS-CoV-2 S- and nucleocapsid-specific responses compared with IAV-specific responses. These data suggest that the memory T cell phenotype after a single infection with SARS-CoV-2 persists over time, with an altered cytokine and cytotoxicity profile compared with long-term memory to whole IAV within the same subjects.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), a pandemic respiratory virus, continues to circulate in many regions around the world (https://coronavirus.jhu.edu/map.html). Immunity to SARS-CoV-2 in a high proportion of the population will be required to control this pandemic (1). Although vaccine-induced immunity is an important component of protection against SARS-CoV-2, a substantial number of people have recovered from coronavirus disease 2019 (COVID-19), and there is a need to understand the persistence and quality of immunity to SARS-CoV-2 in these individuals. All successful viruses must suppress the host innate immune response to some extent; however, SARS-CoV-2 is particularly adept at evading type I and III IFN responses (25), and people with defects in IFN signaling are overrepresented among severe cases (6, 7). Whether these unique features of the early response to SARS-CoV-2 impact long-term immunity to SARS-CoV-2 compared with other respiratory viruses remains unknown.

There is now extensive evidence that CD4+ and CD8+ T cell as well as Ab responses are induced following asymptomatic, mild, or severe COVID-19 infection, with memory T cell responses established within a few weeks of infection (825). CD4+ T cell responses to SARS-CoV-2 are largely of the Th1 type, with cytotoxic CD4+ T cells and circulating T follicular helper (cTfh) cells also detected in some studies (8, 11, 16, 19). CD8+ T cells are less consistently detected (8, 11, 16, 19). Spike (S), nucleocapsid (N), membrane (M), and open reading frame 3 (Orf3) are all major sources of CD4+ and CD8+ T cell epitopes (8, 11, 16, 18, 20, 21, 26). SARS-CoV-2 infection can lead to mild or severe outcomes with early T cell responses, increased frequency of airway T cells, and early S-specific neutralizing and nonneutralizing Ab responses correlating with earlier viral control and better outcomes (16, 17, 20, 2733). Several studies have demonstrated the persistence of T cell and/or Ab responses to SARS-CoV-2 out to 6–10 mo postinfection (3440). For example, Dan et al. reported that T cell responses to SARS-CoV-2 decline with a t1/2 of 3–5 mo, with B cell memory and Ab responses relatively stable over 6 mo (34).

In a previous study, we examined the memory response to SARS-CoV-2 in the early convalescent phase in a cohort of 13 subjects, 46% of whom were categorized as having moderate to severe disease based on admission to a hospital or an intensive care unit (ICU), respectively (19). We compared T cell responses to SARS CoV-2 Ags with memory responses to whole H1N1 influenza A virus (IAV) within the same subjects. At the early convalescent stage, CD4+ T cell responses predominated over CD8+ T cell responses, in contrast to the response to IAV, which was dominated by a stronger CD8+ T cell response (19). In addition, we noted that the frequency of S-specific IFN-γ CD4+ T cell responses was lower than the frequency of S-specific IL-2 specific CD4+ T cell responses, and distinct from the response to IAV, which was dominated by IFN-γ responses (19). However, these findings were limited by the small sample size and were potentially impacted by the predominance of severe cases in the cohort. Here we examined the persistence and phenotype of T cell responses to SARS-CoV-2 at two time points over a period of 9 mo postsymptom onset (PSO) in a cohort of 24 confirmed COVID-19 convalescent subjects, 75% of whom were not hospitalized. The results show that even after mild COVID-19, T cell recall responses continue to be dominated by a higher CD4 response over the CD8 T cell response and a higher proportion of IL-2–producing cells than IFN-γ–producing CD4 T cells out to as late as 9 mo after infection. Plasma IgG responses to SARS-CoV-2 declined only slightly over the course of the study, whereas salivary Ab responses fell off more rapidly, albeit plasma and salivary IgG responses were strongly correlated overall.

This study was designed to investigate the persistence and immunophenotype of T cell and Ab responses to SARS-CoV-2 in subjects who had recovered from PCR-confirmed COVID-19 and to compare this phenotype with IAV-specific memory in the same subjects. Each subject donated blood and saliva at two time points, the first between 30 and 154 d PSO, median 45 d, and the second between 55 and 249 d PSO, median 145 d. Saliva and plasma were used for analysis of Ab responses, whereas PBMCs were stimulated with SARS-CoV-2 peptide pools or whole IAV and assayed using three complementary assays: a 16-parameter flow cytometry panel, in which cytokine production was used to identify Ag-specific T cells after 18 h of stimulation; a proliferation assay, in which dilution of CFSE by CD4+CD3+ or CD8+ CD3+ was measured after 6 d of stimulation; and a commercial multiplex bead array to analyze 13 secreted cytokines and cytotoxic molecules in the supernatant after 48 h of stimulation.

Individuals who had recovered from COVID-19 as confirmed by positive nasopharyngeal COVID-19 PCR upon presentation were recruited to the Risk of Environmental Surface and Air Contamination in COVID-19 (RISC-CoV) study through participating hospitals in the Toronto Invasive Bacterial Diseases Network, with informed consent, to donate blood and saliva, as approved by the University of Toronto Research Ethics Board (REB protocol number 00027673 to T.H.W.). All human subjects research was done in compliance with the Declaration of Helsinki. Whole blood was collected from healthy human donors by venipuncture, and plasma was separated. Saliva was collected in Salivette tubes, as described previously (12). Saliva and plasma aliquots were stored frozen at −80°C until use. PBMCs were isolated by density gradient centrifugation using Ficoll-Paque PLUS (GE Healthcare). PBMCs were stored at −150°C in AIM V (Life Technologies) with 10% DMSO until use.

Cryopreserved PBMCs were thawed at 37°C, washed twice with PBS, and cultured in complete media (RPMI 1640 supplemented with 10% FBS, 2-ME, sodium pyruvate, penicillin, streptomycin, and nonessential amino acids; Life Technologies) at 37°C with 5% CO2. A total of 2 × 106 PBMCs were plated per well in 96-well round-bottomed plates for 18 h with 1 μg/ml of SARS-CoV-2 S1, S2, N, M, or E peptide pools (PepMix, S1/S2: 90% purity, N/M/E: 70% purity; JPT) or 100 hemagglutination units (HAU)/ml of live IAV PR8/34. PBMCs were cultured with equimolar DMSO as a negative control. GolgiStop (BD Biosciences, San Jose, CA) containing monensin was added in the last 6 h of the culture. As a positive control, 50 ng/ml PMA (Sigma-Aldrich), 1 μg/ml ionomycin (Sigma-Aldrich), and GolgiStop were added to PBMCs cultured with complete media in the last 6 h of culture.

After culture, PBMCs were washed with PBS containing 2% FBS (FACS buffer). Cells were first stained with antihuman CCR7 (clone G043H7; BioLegend, San Diego, CA) at 37°C for 10 min, followed by staining with Fixable Viability Dye eFluor 506 (eBioscience, Thermo Fisher Scientific, Mississauga, ON, Canada) to discern viable cells, antihuman CD3 (clone UCHT1; BioLegend), CXCR5 (clone J252D4; BioLegend), 4-1BB (clone 4B4-1; BioLegend), HLA-DR (clone L243; BioLegend), CD4 (clone SK3; BD Biosciences), CD27 (clone L128; BD Biosciences), CD8 (clone Sk1; eBioscience), PD-1 (clone EH12.2H7; BioLegend), OX40 (clone Ber-ACT35; BioLegend), CD69 (clone FN50; eBioscience), and CD45RA (clone HI100; BD Biosciences) for 20 min at 4°C. Cells were washed twice with FACS buffer, then fixed with BD Cytofix/Cytoperm buffer (BD Biosciences) for 20 min at 4°C. Following fixation and permeabilization, cells were washed twice with 1× BD Perm/Wash buffer (BD Biosciences) and stained with antihuman IFN-γ (clone 4S.B3; BioLegend), TNF (clone Mab11; BioLegend), and IL-2 (clone MQ1-17H12; eBioscience) for 15 min at 4°C. Samples were washed twice, then resuspended in FACS buffer. All events (∼1 × 106–1.3 × 106 events per sample) were acquired on the BD LSRFortessa X-20 flow cytometer using FACSDiva software. In the representative sample, which shows the median response, 49% of live CD3+ lymphocytes were CD4+ T cells, and 74 events were collected in the IL-2+ gate for the S1-stimulated sample (Fig. 1A).

FIGURE 1.

ICC production by SARS-CoV-2– or IAV-specific CD4+ T cells over time. Cytokine production by virus-specific CD4+ T cells was assessed by flow cytometry following 18 h of stimulation with SARS-CoV-2 peptide pools or IAV. Representative flow cytometry plots (top) and longitudinal analyses after subtraction of background for each sample (Δ) show the frequency of CD4+ T cells expressing (A) IL-2, (B) IFN-γ, and (C) TNF. Boxed numbers in (A) show the average rate of decay between time points for each response, calculated for donors whose response was 2 SDs above the mean background. Violin plots show the median (red line) frequency of CD4+ T cells expressing (D) IL-2, (E) IFN-γ, and (F) TNF to each Ag binned by time PSO (4–8 wk or 15–35 wk). (G) Sum of the frequency of CD4+ T cells expressing IL-2 or IFN-γ in response to all SARS-CoV-2 peptide pools or IAV. All plots show data from 24 donors. Comparisons between each Ag were made using Dunn’s multiple comparisons test. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

FIGURE 1.

ICC production by SARS-CoV-2– or IAV-specific CD4+ T cells over time. Cytokine production by virus-specific CD4+ T cells was assessed by flow cytometry following 18 h of stimulation with SARS-CoV-2 peptide pools or IAV. Representative flow cytometry plots (top) and longitudinal analyses after subtraction of background for each sample (Δ) show the frequency of CD4+ T cells expressing (A) IL-2, (B) IFN-γ, and (C) TNF. Boxed numbers in (A) show the average rate of decay between time points for each response, calculated for donors whose response was 2 SDs above the mean background. Violin plots show the median (red line) frequency of CD4+ T cells expressing (D) IL-2, (E) IFN-γ, and (F) TNF to each Ag binned by time PSO (4–8 wk or 15–35 wk). (G) Sum of the frequency of CD4+ T cells expressing IL-2 or IFN-γ in response to all SARS-CoV-2 peptide pools or IAV. All plots show data from 24 donors. Comparisons between each Ag were made using Dunn’s multiple comparisons test. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Close modal

A total of 1 × 106 PBMCs were seeded per well in 96-well round-bottomed plates with 1 μg/ml each of SARS-CoV-2 S1, S2, N, M, or E peptide pools (PepMix; JPT) or 100 HAU/ml live IAV. PBMCs were cultured with equimolar DMSO as a negative control. Cell culture supernatants were collected after 48 h of incubation at 37°C and stored at −80°C. Cytokines in the supernatants were measured using the Human CD8/NK Cytokine Panel (13-plex) LEGENDplex kit (BioLegend) with capture reagents specific for IL-2, IL-4, IL-10, IL-6, IL-17A, TNF, sFas, sFasL, IFN-γ, granzyme A, granzyme B, perforin, and granulysin. The assay was performed as per the manufacturer’s instructions using a V-bottomed plate. Samples were acquired on the BD LSRFortessa X-20 flow cytometer. Data are reported after subtraction of the background controls (DMSO-containing media only) for each sample.

PBMCs were labeled with 10 μM CFSE (Thermo Fisher Scientific) in PBS for 3 min at room temperature. Excess CFSE dye was quenched by adding a 10× volume of cold complete media and incubated on ice for 5 min. Cells were then washed twice, resuspended in complete media, and plated at 4 × 105 cells/well in 96-well round-bottomed plates. CFSE-labeled PBMCs were incubated with 1 μg/ml of SARS-CoV-2 S1, S2, N, M, or E peptide pools (PepMix; JPT), 100 HAU/ml live PR8, or equimolar DMSO (negative control) for 6 d at 37°C. On day 6, cells were washed with FACS buffer and stained with Fixable Viability Dye eFluor 506 (eBioscience) to discern viable cells, antihuman CD3 (clone UCHT1; BioLegend), CD4 (clone RPA-T4; BD Biosciences), CD8 (clone SK1; eBioscience), CD45RA (clone HI100; BD Biosciences), and CD27 (clone L128; BD Biosciences) for 20 min at 4°C. Cells were washed twice with FACS buffer, then fixed with BD Cytofix for 20 min at 4°C. Samples were washed twice, then resuspended in FACS buffer and acquired on the BD LSRFortessa X-20 flow cytometer using FACSDiva software.

An automated ELISA chemiluminescence assay was used to analyze the levels of IgG, IgA, and IgM Abs to the S trimer, its receptor binding domain (RBD), and the N, essentially as in (12) with the following modifications. The N (PRO47; 7 ng/well) and RBD (PRO1151; 20 ng/well) Ags were produced in Chinese hamster ovary cells and were a kind gift from Dr. Yves Durocher, National Research Council of Canada (NRC). For IgG, the secondary Ab was an IgG–HRP fusion (PRO1146; 1:6700 or 0.9 ng/well) provided by the NRC. A standard curve of the VHH72 monobody (41) fused to human IgG1 Fc domain (also from the NRC) was generated for calibrating the anti-S and anti-RBD IgG response. All other Ags, detection reagents, and calibration reagents were as previously described (12). The analysis also proceeded largely as in (12), with the following exceptions. Blanks were not subtracted from the chemiluminescence raw values of the samples, and the raw values were normalized to a blank-subtracted point in the linear range of the calibration standard curve. (For S and RBD, the reference point was 0.0156 μg/ml, and for N, it was 0.0625 μg/ml). The final results are represented as a “relative ratio” to this calibration standard point. To define the cutoff for positive Ab calls for each Ag for IgG, 3 SDs from the mean of the log negative control distribution from >20 different runs collected over 4 mo was used as a cutoff to define positivity in each individual assay. For IgA, negatives from two different runs over 1 mo and for IgM negatives from three runs over 2 mo were used for this. In all cases, this corresponds to a <2% false-positive rate assessment, based on receiver operating characteristic curves. The selected cutoff for positivity was drawn as a dashed line on the figure.

The expression, purification, and biotinylation of the SARS-CoV-2 RBD and S ectodomain were performed as previously described (12, 42). ELISAs to detect S- and RBD-specific IgG and IgA Abs in saliva were completed as described previously (12). To analyze the data, raw OD450 measurements obtained from PBS-coated wells corresponding to each sample diluted at 1:5 (“background signal”) were subtracted from readings obtained from Ag-coated wells at each of the three dilutions described (1:5, 1:10, 1:20). For each plate, a sample of pooled saliva from COVID-19 acute and convalescent subjects was plated at 1:5 with no Ag (PBS control), as well as with Ags at 1:5, 1:10, and 1:20. The area under the curve was calculated based on the background subtracted values from all three dilutions for each sample and for the pooled sample of positive control saliva run on each plate. Each sample within a given plate was then normalized to the pooled positive control saliva for that particular plate and expressed as a percentage.

Flow cytometric data were analyzed using FlowJo version 10.7.1 (BD Biosciences). Multiplex cytokine bead data were analyzed using the LEGENDplex Data Analysis Software Suite (BioLegend). Statistical analyses were performed using GraphPad Prism version 9.1.1. Δ in figures represents values for which the corresponding background signal for each donor had been subtracted. “Background signal” is defined as the frequency of cells expressing the parameter being analyzed, or concentration of an analyte, in negative control wells. For each donor, the response is reported as the Δ after background subtraction for that donor. A T cell response was defined as positive if the value was two times background for that sample, and above a minimal level of 0.003%, which is calculated as the median of all DMSO controls. To calculate the rate of decay of the CD4 T cell responses based on ICC responses, we focused only on the stronger responses, and we only included donors whose response was 2 SDs above the median background. For proliferation data, in calculating the distribution of CD4 or CD8 subsets, only positive results were included, defined as above. For the multiplex data, the limit of detection was established using a standard curve for each analyte. Pairwise comparisons were made by a two-tailed Wilcoxon test or a nonparametric Dunn’s multiple comparisons test. Correlation analyses were performed by computing the Spearman correlation coefficient. For longitudinal data, the percentage change over time and t1/2 were calculated after linear regression was performed for each donor. We report the average percentage change per week and the average t1/2. Cluster analysis for multiplex data were performed by first calculating fold change in response to an Ag and then performing hierarchical agglomerative clustering on log-transformed data. Values across donors were standardized by minimum/maximum normalization such that, for each analyte, the minimum is subtracted, then divided by its maximum. These data were visualized using the Seaborn data visualization library for Python (https://joss.theoj.org/papers/10.21105/joss.03021).

A total of 24 COVID-19 convalescent subjects, who had recovered from COVID-19 as confirmed by a positive nasopharyngeal COVID-19 PCR test result upon presentation, consented to undergo a blood draw and saliva sampling. Samples were collected at two time points, the first between 30 and 154 d PSO, median 45 d, and the second between 55 and 249 d PSO, median 145 d. Of the study sample, 18 subjects had mild disease (not hospitalized), 1 subject had moderate disease (hospitalized, non-ICU), and 5 subjects had severe disease (ICU). The median age was 46.5 y, and 50% were female (Table I).

Table I.

Participant characteristics

CharacteristicValue
Age, y 22–67 (median, 46.5) 
Sex, n (%)  
 Male 12 (50%) 
 Female 12 (50%) 
Days PSO  
 First visit 30–154 (median, 45) 
 Second visit 55–249 (median, 145) 
Clinical features  
 ICU 5 of 24 (20.8%) 
 Hospitalized (non-ICU) 1 of 24 (4.2%) 
 Mild 18 of 24 (75%) 
CharacteristicValue
Age, y 22–67 (median, 46.5) 
Sex, n (%)  
 Male 12 (50%) 
 Female 12 (50%) 
Days PSO  
 First visit 30–154 (median, 45) 
 Second visit 55–249 (median, 145) 
Clinical features  
 ICU 5 of 24 (20.8%) 
 Hospitalized (non-ICU) 1 of 24 (4.2%) 
 Mild 18 of 24 (75%) 

To assess the recall response to SARS-CoV-2 in convalescent subjects, we used 15-mer peptide pools with 11-aa overlap, comprising the S, M, N, and envelope (E) of SARS-CoV-2, to stimulate freshly thawed PBMCs. Due to the large size of S, two S peptide pools (S1 and S2 pools) were analyzed separately to reduce competition for Ag presentation. After 18 h of stimulation, SARS-CoV-2–specific CD4+ T cells were detected based on the production of IL-2, IFN-γ, or TNF by ICC staining as determined by the gating strategy shown in Supplemental Fig. 1A. Because most adults are expected to have a memory response to IAV, we used IAV strain PR8/34/H1N1 as an internal positive control to ensure the ability of each patient sample to mount a T cell recall response dependent on APCs; PMA/ionomycin was used as a secondary nonspecific stimulus to validate overall T cell quality. Only samples with a response to PMA/ionomycin were included in the study, and graphs show the Δ value after subtraction of background for each donor. For the study population overall, there was a significant increase in IL-2, IFN-γ, and TNF in response to S1 peptide pools, as well as whole IAV, compared with background control, whereas significant CD4+ T cell responses to N were detected based on IL-2 only (Supplemental Fig. 1B).

IL-2–producing CD4+ T cells were the most abundantly detected of the cytokine-producing CD4+ T cells in response to SARS-CoV-2 peptide pools. IL-2 responses were still detectable to 35 wk for two of the four donors at that time point (Fig. 1A). IFN-γ–producing CD4+ cells were detected at a lower frequency than IL-2–producing CD4+ T cells in response to SARS-CoV-2 peptide pools (Fig. 1B). TNF-producing cells were the lowest in frequency compared with IL-2– or IFN-γ–producing cells and detected in only a few study subjects (Fig. 1C; Supplemental Fig. 1B). IL-2–, IFN-γ–, and TNF-producing cells were detected in response to S2 and M peptide pools in only a few donors and were generally weaker (Supplemental Fig. 1C–1E). Of the five SARS-CoV-2 peptide pools tested, the highest frequency of CD4+ T cell responses was detected against S, and the lowest frequency of responses was detected against E, based on the three cytokines measured (Fig. 1D1F). In general, donors who had the strongest responses to S1 also had the strongest responses to all the peptide pools tested.

To estimate the rate of decay of the T cell response to SARS-CoV-2, we analyzed the frequency of IL-2–producing cells at the two time points for those donors whose response was 2 SDs above background. For S1- and N-specific responses, the rate of decay between the two time points was 3.5–3.8% per week, equivalent to a t1/2 of ∼14 wk (Fig. 1A). This decay was not impacted by any change in the frequency of APCs between the two time points, because there was no difference in the frequency of HLA-DR+CD3 cells over time (Supplemental Fig. 1F). The rate of decay of the T cell response to IAV between time points was lower than that in response to SARS-CoV-2 S (Fig. 1A), which is to be expected when comparing a recent primary response to long-term memory, likely representing multiple exposures.

As expected, responses against whole IAV were generally higher than those against individual SARS-CoV-2 peptide pools. Our previous studies showed that both trivalent inactivated influenza A vaccine, which lacks nucleic acids, and live IAV gave a similar response to S (19). Thus, the larger overall response to IAV is not likely to be due to innate immune responses but is likely due to whole IAV containing the full complement of epitopes. Therefore, we also analyzed SARS-CoV-2–specific CD4+ T cell frequencies as a combined sum of the IL-2 or IFN-γ responses against all five SARS-CoV-2 peptide pools analyzed (Fig. 1G). The median frequency of IL-2–producing T cells did not differ significantly between SARS-CoV-2 and IAV. However, the median frequency of IFN-γ–producing CD4+ T cells was significantly reduced in response to SARS-CoV-2 compared with IAV (Fig. 1G). Overall, these results showed that 83% of study subjects have a T cell response to any SARS-CoV-2 peptide pool based on IL-2 or IFN-γ cytokine production, with IL-2–producing CD4+ T cells representing the most abundant S- and N-specific T cells.

We next characterized the phenotype of the SARS-CoV-2–specific CD4+ T cells for production of one, two, or three cytokines per cell, as well as surface marker upregulation. Eighty-four percent of S-specific CD4+ T cells produced only one cytokine at 4–8 wk PSO, and this was similar at 15–35 wk (Fig. 2A). IAV-specific CD4+ T cell responses were only slightly more multifunctional, with 4.6% of IAV-specific CD4+ T cells producing three cytokines at 4–8 wk PSO; this percentage was 5.7% at 15–35 wk PSO. This was significantly greater than S1-specific CD4+ T cell responses, where only 2% of cells produced three cytokines at 4–8 wk PSO, and this percentage was 3.2% at 15–35 wk PSO (Fig. 2A). The ratio of IFN-γ– to IL-2–producing CD4+ T cells was also significantly lower in response to both S1 and N than to IAV at 4–8 wk PSO within the same donor (Fig. 2B). Moreover, this diminished IFN-γ/IL-2 ratio in the response to the S1 peptide pool compared with IAV persisted to 35 wk (Fig. 2B).

FIGURE 2.

Immunophenotype of SARS-CoV-2– and IAV-specific CD4+ T cell responses over time. (A) Frequency of CD4+ T cells expressing one, two, or three of IL-2, IFN-γ, and/or TNF as a proportion of cytokine-producing cells. (B) Ratio of the percentage of IFN-γ+ to the percentage of IL-2+ CD4+ T cells in donors producing both cytokines in response to S, N, or IAV, showing comparison between S, N, and IAV at 4–8 wk (left) and the ratio over time up to 35 wk (right); n = 24. (C) Representative flow plot of CD45RA and CD27 expression by IL-2–producing CD4+ T cells in response to stimulation with S1 or IAV (top). Bar graph depicts each memory subset (Tem, Temra, Tsc, Tcm) as a proportion of IL-2+CD4+ T cells (bottom); n = 20. (D) Representative flow plots of 4-1BB and OX40 expression by CD4+ T cells in response to stimulation with S1, N, or IAV. Graphs depict the percentage of 4-1BB+OX40+ CD4+ T cells in response to S1 or N peptide pools or whole IAV in comparison with DMSO negative control; n = 24. Comparisons in (A) and (D) were made by pairwise two-tailed Wilcoxon test. Comparisons in (B) were made by Dunn’s multiple comparisons test. *p ≤ 0.05, ***p ≤ 0.001, ****p ≤ 0.0001. All data are plotted after subtraction of background values from unstimulated controls for each sample.

FIGURE 2.

Immunophenotype of SARS-CoV-2– and IAV-specific CD4+ T cell responses over time. (A) Frequency of CD4+ T cells expressing one, two, or three of IL-2, IFN-γ, and/or TNF as a proportion of cytokine-producing cells. (B) Ratio of the percentage of IFN-γ+ to the percentage of IL-2+ CD4+ T cells in donors producing both cytokines in response to S, N, or IAV, showing comparison between S, N, and IAV at 4–8 wk (left) and the ratio over time up to 35 wk (right); n = 24. (C) Representative flow plot of CD45RA and CD27 expression by IL-2–producing CD4+ T cells in response to stimulation with S1 or IAV (top). Bar graph depicts each memory subset (Tem, Temra, Tsc, Tcm) as a proportion of IL-2+CD4+ T cells (bottom); n = 20. (D) Representative flow plots of 4-1BB and OX40 expression by CD4+ T cells in response to stimulation with S1, N, or IAV. Graphs depict the percentage of 4-1BB+OX40+ CD4+ T cells in response to S1 or N peptide pools or whole IAV in comparison with DMSO negative control; n = 24. Comparisons in (A) and (D) were made by pairwise two-tailed Wilcoxon test. Comparisons in (B) were made by Dunn’s multiple comparisons test. *p ≤ 0.05, ***p ≤ 0.001, ****p ≤ 0.0001. All data are plotted after subtraction of background values from unstimulated controls for each sample.

Close modal

Based on the expression of CD27 and CD45RA by the IL-2–producing CD4+ T cells, all four subtypes of memory T cells were detected, with the largest fraction of IAV-specific and S-specific CD4+ T cells comprising the central memory (Tcm) phenotype (CD45RACD27+) (Fig. 2C). Compared with IAV-specific CD4+ T cells, S-specific CD4+ T cells included a higher proportion of Tcm and lower proportions of effector memory (Tem; CD45RACD27) and T effector memory RA+ (CD45RA+CD27; Temra) cells, thought to represent more differentiated Tem cells (Fig. 2C; Supplemental Fig. 2A). We also detected cytokine-producing CD45RA+CD27+ cells among the SARS-CoV-2– and IAV-specific T cells. These are likely the CD45RA+CD45ROCD27+ stem cell memory T (Tsc) cells described by Gattinoni et al. (43). The activation markers 4-1BB and OX40 were significantly upregulated in response to S1, N, and IAV at both time points (Fig. 2D). Based on 4-1BB/OX40 coexpression, 88% of subjects showed responses to S or N. In sum, although the IAV-specific and SARS-CoV-2–specific cells had relatively similar proportions of memory subsets, SARS-CoV-2 S- and N-specific T cell frequency showed a lower IFN-γ/IL-2 ratio, and the T cells were slightly less multifunctional than IAV-specific T cells within the same subjects.

Next, we analyzed CD8+ T cell responses in the PBMC cultures following stimulation with SARS-CoV-2 S, N, M, and E peptide pools compared with IAV at each time point. In general, the frequency of SARS-CoV-2–specific CD8+ T cell responses detected based on IFN-γ production was lower than the response to intact IAV (Fig. 3A). As with the CD4+ T cell responses, SARS-CoV-2–specific IFN-γ–producing CD8+ T cells were most readily detected in response to S1 peptide pools and were detected in very few study subjects in response to N, M, or E peptide pools (Fig. 3B). Based on the expression of CD27 and CD45RA by IFN-γ+CD8+ T cells, both IAV and SARS-CoV-2 recall responses were predominantly of the more terminally differentiated Temra phenotype, and there was no difference in the distribution of memory subsets between S-specific and IAV-specific CD8+ T cells (Fig. 3C; Supplemental Fig. 2B). In summary, CD8+ T cell responses are present at low frequencies in response to SARS-CoV-2 peptide pools based on ICC staining, with the response to S predominating over the other Ags tested at both time points.

FIGURE 3.

CD8+ T cell responses to SARS-CoV-2 are low in frequency. (A) Representative flow cytometry plots and longitudinal analyses after subtraction of background (unstimulated control) for each sample (Δ) show the frequency of CD8+ T cells expressing IFN-γ to the indicated Ag. (B) Violin plots show the median frequency (red line) of CD8+ T cells expressing IFN-γ to each Ag binned by time PSO (4–8 wk or 15–35 wk). (A) and (B) show data from 24 donors. Comparisons were made by Dunn’s multiple comparisons test. (C) Representative flow plot of CD45RA and CD27 expression by total and IFN-γ–producing CD8+ T cells (left). Bar graph depicts each memory subset as a proportion of IFN-γ+CD8+ T cells (right); n = 13. *p ≤ 0.05, ***p ≤ 0.001, ****p ≤ 0.0001.

FIGURE 3.

CD8+ T cell responses to SARS-CoV-2 are low in frequency. (A) Representative flow cytometry plots and longitudinal analyses after subtraction of background (unstimulated control) for each sample (Δ) show the frequency of CD8+ T cells expressing IFN-γ to the indicated Ag. (B) Violin plots show the median frequency (red line) of CD8+ T cells expressing IFN-γ to each Ag binned by time PSO (4–8 wk or 15–35 wk). (A) and (B) show data from 24 donors. Comparisons were made by Dunn’s multiple comparisons test. (C) Representative flow plot of CD45RA and CD27 expression by total and IFN-γ–producing CD8+ T cells (left). Bar graph depicts each memory subset as a proportion of IFN-γ+CD8+ T cells (right); n = 13. *p ≤ 0.05, ***p ≤ 0.001, ****p ≤ 0.0001.

Close modal

Because CD8+ T cell responses to SARS-CoV-2 were low in frequency as detected by overnight stimulation, we next asked if these cells could be better detected based on Ag-specific proliferation, an important indicator of potential recall responses. To this end, total PBMCs were labeled with CFSE, then stimulated for 6 d with SARS-CoV-2 peptide pools or with IAV, and T cell proliferation was assessed based on CFSE dilution, using the gating strategy shown in Supplemental Fig. 3A. CD4 and CD8 proliferative responses in response to S and N persisted for up to 35 wk in three of the four donors at this time point (Fig. 4A and 4B). T cell proliferative responses were also detected in response to S2 and M in both CD4+ T cells and CD8+ T cells (Fig. 4C and 4D; Supplemental Fig. 3B and 3C). Proliferative responses to S were most often dominated by CD4+ responses, whereas responses to N peptides were approximately equally distributed between CD4+ and CD8+ T cells. In contrast, proliferative responses to IAV were dominated by CD8+ T cells (Fig. 4E). Altogether, the data revealed a proliferative response upon reexposure to SARS-CoV-2 peptides in ∼79% of subjects, with the CD8+ proliferative responses generally less robust than CD4+ proliferative responses, and responses to S dominated by CD4+ T cells.

FIGURE 4.

CD4+ and CD8+ T cell proliferative responses to SARS-CoV-2 up to 35 wk PSO. (A) Representative flow cytometry plots and longitudinal analyses after subtraction of background (unstimulated control) subtracted values (Δ) show the frequency of CFSEloCD4+ T cells after stimulation with the indicated Ag. (B) Representative flow plots and longitudinal analyses of background subtracted values (Δ) showing the frequency of CFSEloCD8+ T cells. (C) Violin plot showing the median (red line) percentage of CFSEloCD4+ T cells binned by time PSO. (D) Violin plot showing the median percentage of CFSEloCD8+ T cells binned by time PSO. (A–D) n = 24. (C and D) Comparisons were made by Dunn’s multiple comparisons test. (E) Comparison of the percentage of CFSElo CD4+ or CD8+ T cells in response to each Ag using pairwise two-tailed Wilcoxon test. Right two graphs show the percentage of CFSElo CD4+ or CD8+ T cells stimulated with S1 or N peptide pools by individual donors. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

FIGURE 4.

CD4+ and CD8+ T cell proliferative responses to SARS-CoV-2 up to 35 wk PSO. (A) Representative flow cytometry plots and longitudinal analyses after subtraction of background (unstimulated control) subtracted values (Δ) show the frequency of CFSEloCD4+ T cells after stimulation with the indicated Ag. (B) Representative flow plots and longitudinal analyses of background subtracted values (Δ) showing the frequency of CFSEloCD8+ T cells. (C) Violin plot showing the median (red line) percentage of CFSEloCD4+ T cells binned by time PSO. (D) Violin plot showing the median percentage of CFSEloCD8+ T cells binned by time PSO. (A–D) n = 24. (C and D) Comparisons were made by Dunn’s multiple comparisons test. (E) Comparison of the percentage of CFSElo CD4+ or CD8+ T cells in response to each Ag using pairwise two-tailed Wilcoxon test. Right two graphs show the percentage of CFSElo CD4+ or CD8+ T cells stimulated with S1 or N peptide pools by individual donors. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Close modal

Abs are often an important correlate of protective immunity, and evidence suggests that this is the case in protection from COVID-19 (27, 28, 30, 4446). Therefore, we assayed plasma levels of S-, RBD-, and N-specific IgG, IgA, and IgM for 19 study subjects for whom these samples were available. We found that most COVID-19 convalescent subjects (95%) mounted an IgG response against S trimer at the first visit, 79% against RBD, and 53% against N, whereas IgA responses were lower, and few subjects demonstrated an IgM response to SARS-COV-2 S, RBD, or N, which is not unexpected, given that our first time point is several weeks past initial infection. IgG responses against S trimer were stable and persisted for up to 35 wk, decaying at ∼0.50%/wk (t1/2 = 100 wk), whereas IgG responses against RBD and N were estimated to decay at rates of 1.8%/wk (t1/2 = 28 wk) and 1.1%/wk (t1/2 = 45 wk), respectively (Fig. 5A).

FIGURE 5.

Persistence of anti–SARS-CoV-2 Abs in the plasma up to 35 wk PSO. (A) Longitudinal analysis of the relative ratio (to a synthetic standard; see Materials and Methods) of IgG (top row), IgA (middle row), and IgM (bottom row) to S trimer, RBD, or N in the plasma of COVID-19 convalescent individuals over time (n = 19). Boxed numbers in the IgG panels show the average rate of decay between time points for each response. Dotted line indicates 3 SDs over the mean of the log negative control distribution (see Materials and Methods). (B) Longitudinal analysis of the percentage area under the curve (%AUC) IgG (top row) and IgA (bottom row) to S trimer or RBD in the saliva of COVID-19 convalescent individuals over time (n = 19). (C) Correlation between plasma and salivary IgG to S trimer (top) or RBD (bottom). %AUC refers to area under the curve for experimental sample over AUC of positive control. Both time points for each donor are shown (n = 19 per time point). Analysis was performed by Spearman correlation. ***p ≤ 0.001.

FIGURE 5.

Persistence of anti–SARS-CoV-2 Abs in the plasma up to 35 wk PSO. (A) Longitudinal analysis of the relative ratio (to a synthetic standard; see Materials and Methods) of IgG (top row), IgA (middle row), and IgM (bottom row) to S trimer, RBD, or N in the plasma of COVID-19 convalescent individuals over time (n = 19). Boxed numbers in the IgG panels show the average rate of decay between time points for each response. Dotted line indicates 3 SDs over the mean of the log negative control distribution (see Materials and Methods). (B) Longitudinal analysis of the percentage area under the curve (%AUC) IgG (top row) and IgA (bottom row) to S trimer or RBD in the saliva of COVID-19 convalescent individuals over time (n = 19). (C) Correlation between plasma and salivary IgG to S trimer (top) or RBD (bottom). %AUC refers to area under the curve for experimental sample over AUC of positive control. Both time points for each donor are shown (n = 19 per time point). Analysis was performed by Spearman correlation. ***p ≤ 0.001.

Close modal

We also measured Abs in saliva and found that IgG responses in saliva decayed more rapidly than in the plasma, decreasing at 24–30%/wk overall (Fig. 5B). IgA responses also decayed rapidly in saliva, decaying at a rate of 17%/wk against S trimer (t1/2 = 3.0 wk) and 27%/wk against RBD (t1/2 = 1.9 wk) (Fig. 5B). Plasma IgG strongly correlated with salivary IgG levels against both S trimer (r = 0.76) and RBD (r = 0.73) (Fig. 5C). One of 19 donors did not mount any Ab response to any of the Ags assayed, albeit T cell responses were detected in this individual. Taken together, our data show that the vast majority of COVID-19 convalescent subjects mounted an Ab response in the plasma and saliva to at least one SARS-CoV-2 protein, with little decay of anti-S plasma IgG over the course of the study and a strong correlation between salivary and plasma IgG responses, as previously reported (12).

Tfh cells are important in the formation of memory B cells and long-lived Ab-producing plasma cells (47). Although these cells typically reside in the germinal centers of lymphoid organs, cTfh cells are the peripheral counterparts of germinal center Tfh; cTfh cells are able to efficiently induce Ab secretion by B cells, contribute to Ab generation following vaccination, and correlate with broadly neutralizing Ab responses against some viruses (4850). Because IL-21 is difficult to detect by flow cytometry, we identified SARS-CoV-2–specific cTfh cells by ICC staining as IL-2–producing CD45RACXCR5+CCR7+CD4+ T cells as defined previously (4850), with gating shown in Supplemental Fig. 1A. IL-2–producing cells with a cTfh surface phenotype were detected specifically in response to S1 and IAV at both time points, whereas for N-stimulated cultures, this response was only specific at time point 2 (Supplemental Fig. 3D). Although these cTfh responses to S1 were detected out to 35 wk PSO, there were large variations and fluctuations in N- and IAV-specific cTfh responses over time (Fig. 6A), albeit this may reflect the low frequency of responses. SARS-CoV-2 S2 and M peptide pools induced only minimal cTfh responses, and no responses to E peptide pools were detected (Supplemental Fig. 3D). As expected, based on whole IAV having the entire complement of Ags, the frequency of IL-2+ cTfh cells was greatest after IAV stimulation (Fig. 6B).

FIGURE 6.

cTfh and CD4+ Teff cell responses correlate with plasma Ab levels. (A) Representative flow plots and longitudinal analyses displaying results after subtraction of background (unstimulated control) values (Δ) for each sample, showing the frequency of cTfh-expressing IL-2 to the indicated Ag (n = 24). (B) Violin plots show the median frequency (red line) of cTfh expressing IL-2 to each Ag binned by time PSO (4–8 wk or 15–35 wk). Comparisons were made by Dunn’s multiplex comparisons test (n = 24). (C) Correlation between S-specific IL-2+cTfh and anti-S trimer or anti-RBD IgG in the plasma (top and middle rows), and correlation between N-specific IL-2+cTfh and anti-N IgG in the plasma (bottom row). (D) Correlation between S-specific IL-2+CD4+ Teff and S-/RBD-specific IgG (top) or IgA (bottom) in the plasma. (E) Correlation between N-specific IL-2+CD4+ Teff and N-specific IgG (top) or IgA (bottom). (C and D) Correlation analysis was performed by Spearman correlation, with each plot depicting both time points for each donor (n = 19 per time point). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

FIGURE 6.

cTfh and CD4+ Teff cell responses correlate with plasma Ab levels. (A) Representative flow plots and longitudinal analyses displaying results after subtraction of background (unstimulated control) values (Δ) for each sample, showing the frequency of cTfh-expressing IL-2 to the indicated Ag (n = 24). (B) Violin plots show the median frequency (red line) of cTfh expressing IL-2 to each Ag binned by time PSO (4–8 wk or 15–35 wk). Comparisons were made by Dunn’s multiplex comparisons test (n = 24). (C) Correlation between S-specific IL-2+cTfh and anti-S trimer or anti-RBD IgG in the plasma (top and middle rows), and correlation between N-specific IL-2+cTfh and anti-N IgG in the plasma (bottom row). (D) Correlation between S-specific IL-2+CD4+ Teff and S-/RBD-specific IgG (top) or IgA (bottom) in the plasma. (E) Correlation between N-specific IL-2+CD4+ Teff and N-specific IgG (top) or IgA (bottom). (C and D) Correlation analysis was performed by Spearman correlation, with each plot depicting both time points for each donor (n = 19 per time point). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Close modal

There was a positive correlation between the frequency of S-specific IL-2+ cTfh and plasma IgG responses mounted against S trimer and S RBD, as well as plasma anti-S trimer IgA (Fig. 6C). However, there was no correlation between Tfh and anti-RBD IgA in the plasma. Likewise, the frequency of N-specific IL-2+ cTfh cells significantly correlated with anti-N IgG in the plasma, but not with anti-N IgA (r = 0.47), although there was a positive trend (Fig. 6D). Similarly, IL-2+ S-specific CD4+ Teff cell responses (defined as the non-Tfh IL-2–producing cells) significantly correlated with both plasma anti-S trimer IgG, IgA, and anti-RBD IgG. Again, there was no correlation with plasma anti-RBD IgA (Fig. 6D). Analysis of IL-2+ N-specific CD4+ Teff responses also showed significant correlation with anti-N IgG in the plasma (r = 0.43; p < 0.01) and likewise did not correlate with levels of anti-N IgA (Fig. 6E). In sum, cTfh responses can be detected in response to SARS-CoV-2 S, N, and M peptide pools for most of the subjects, with some analyzed up to 9 mo PSO. Although these responses were of low frequency, both the SARS-CoV-2–specific cTfh and CD4+ Teff responses correlated with plasma Ab titers.

We next used a quantitative multiplex bead array to determine the levels of 13 cytokines/cytolytic molecules released into the supernatant after 48-h stimulation of PBMCs with SARS-CoV-2 peptide pools or with IAV. Cytokines produced in response to S and N included IFN-γ, IL-2, and TNF and were detectable up to 35 wk PSO, albeit with some decline over time. IL-2, IL-10, and TNF induction by IAV were stable up to 35 wk, as was IL-10 induction by N peptides. (Fig. 7A). IL-4 and IL-17a were detected at low levels in some donors, with no consistent trend observed over time (Supplemental Fig. 4A). Overall, 83% of donors showed a cytokine response of at least two times background and above the limit of detection.

FIGURE 7.

Analysis of cytokine secretion by SARS-CoV-2 convalescent PBMCs compared with IAV. Cytokines in the supernatant of COVID-19 convalescent PBMC cultures were measured using a multiplex bead array after 48-h stimulation with S1 or N peptide pools or IAV (n = 24). Graphs show longitudinal analysis after subtraction of background based on unstimulated controls for each PBMC sample (Δ) of secreted (A) IFN-γ, IL-2, IL-10, TNF, IL-6; (B) granzyme A, granzyme B, granulysin, perforin; and (C) sFas, sFasL.

FIGURE 7.

Analysis of cytokine secretion by SARS-CoV-2 convalescent PBMCs compared with IAV. Cytokines in the supernatant of COVID-19 convalescent PBMC cultures were measured using a multiplex bead array after 48-h stimulation with S1 or N peptide pools or IAV (n = 24). Graphs show longitudinal analysis after subtraction of background based on unstimulated controls for each PBMC sample (Δ) of secreted (A) IFN-γ, IL-2, IL-10, TNF, IL-6; (B) granzyme A, granzyme B, granulysin, perforin; and (C) sFas, sFasL.

Close modal

Molecules associated with cytotoxicity, including granzyme A, granzyme B, granulysin, perforin, FAS, and FASL, were detected in response to both S and N peptide pools. However, the response to S and N showed a wider range of responses in general compared with IAV (Fig. 7B and 7C). sFas and sFasL were both upregulated in all subjects in response to IAV, with minimal decay over time, but less consistently in response to SARS-CoV-2 (Fig. 7C).

We next sought to characterize differences in the cytokine profile induced by each viral Ag. To this end, we calculated fold change in response to S peptide or IAV and performed hierarchical agglomerative clustering on log-transformed data. Values across donors were normalized per cytokine as described in Materials and Methods. At 4–8 wk PSO, we found that S1- and IAV-stimulated samples did not completely cluster based on Ag stimulation, whereas at 15–35 wk PSO, samples clustered almost completely by type of Ag (Fig. 8A and 8B). Analysis of the clusters revealed that although most donors upregulated IFN-γ, granzyme A, granzyme B, and IL-10 (associated with typical antiviral responses), there was a cluster of study subjects who had noticeably diminished expression of granzyme A, granzyme B, and IL-10 (module 1). S1-specific responses also exhibited lower expression of sFas, sFasL, granulysin, and perforin (module 2) than IAV within the same set of donors. TNF and IL-2 were similarly upregulated between S1- and IAV-stimulated samples, whereas IL-17A and IL-4 were not broadly upregulated in response to either Ag (Fig. 8A). We observed that these two cytokine modules were conserved and more apparent at 15–35 wk PSO (Fig. 8B).

FIGURE 8.

SARS-CoV-2 convalescent PBMCs exhibit an altered secreted cytokine profile compared with IAV. Agglomerative hierarchical clustering analysis of the fold change in secreted cytokines was performed on log-transformed data, comparing responses to S1 peptide pool or IAV in each donor at (A) 4–8 wk and (B) 15–35 wk. Each row represents one donor, and each column represents one cytokine. Values across donors for each cytokine were standardized by minimum/maximum normalization. (C) Ratio of IFN-γ to IL-2 at the early convalescent phase 4–8 wk PSO (right) and over time to S1 peptide pools (left); n = 20. (D) Ratio of granzyme A to perforin, sFasL, sFas, and granulysin at the first visit (n = 24). (E) Ratio of IFN-γ to TNF, sFasL, and granzyme A at the first visit (n = 24). Comparisons in (C)–(E) were made by Dunn’s multiple comparisons test. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. Dotted blue line indicates 1:1 ratio. Δ indicates values with background subtracted.

FIGURE 8.

SARS-CoV-2 convalescent PBMCs exhibit an altered secreted cytokine profile compared with IAV. Agglomerative hierarchical clustering analysis of the fold change in secreted cytokines was performed on log-transformed data, comparing responses to S1 peptide pool or IAV in each donor at (A) 4–8 wk and (B) 15–35 wk. Each row represents one donor, and each column represents one cytokine. Values across donors for each cytokine were standardized by minimum/maximum normalization. (C) Ratio of IFN-γ to IL-2 at the early convalescent phase 4–8 wk PSO (right) and over time to S1 peptide pools (left); n = 20. (D) Ratio of granzyme A to perforin, sFasL, sFas, and granulysin at the first visit (n = 24). (E) Ratio of IFN-γ to TNF, sFasL, and granzyme A at the first visit (n = 24). Comparisons in (C)–(E) were made by Dunn’s multiple comparisons test. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. Dotted blue line indicates 1:1 ratio. Δ indicates values with background subtracted.

Close modal

Because the overall response to whole IAV is higher than the response to each of the SARS-CoV-2 peptide pools analyzed, to determine if there was indeed a different profile of cytokine response to SARS-CoV-2 Ags compared with IAV, we determined the ratio of IFN-γ to IL-2 within each donor for IAV- or SARS-COV-2–specific responses. Accordingly, we observed a reduced IFN-γ/IL-2 ratio in response to S and N peptide pools compared with IAV stimulation. Furthermore, this reduced ratio of IFN-γ to IL-2 was retained over time, recapitulating our observations based on ICC analysis (Figs. 2B and 8C; Supplemental Fig. 4B).

To investigate whether the profile of cytotoxic molecules was also distinct between SARS-CoV-2–specific responses and IAV-specific responses independently of the overall magnitude of the response, we analyzed the proportion of cytotoxic molecules secreted in response to S, N, or IAV relative to levels of granzyme A and IFN-γ within each donor for each Ag. S-specific responses resulted in lower levels of perforin, sFas, sFasL, and granulysin, relative to granzyme A, than for IAV, whereas N-specific responses differed from IAV-specific responses in perforin and sFasL only (Fig. 8D). Relative to the level of IFN-γ in the cultures, IAV-specific responses showed differences in TNF, sFasL, and granzyme A relative to SARS-CoV-2–specific responses (Fig. 8E). Therefore, after normalizing to account for different overall level of response, data from the cytokine/cytotoxic molecule secretion assay demonstrate an altered antiviral profile in response to SARS-CoV-2 compared with IAV, which is less IFN-γ dominated, with an altered cytotoxic profile that might indicate different antiviral mechanisms.

In this study, we investigated the longevity of Ab and T cell responses to SARS-CoV-2 by comparing two memory time points in a cohort of 24 PCR-confirmed SARS-CoV-2 convalescent subjects, the majority of whom were nonhospitalized. Based on ICC staining, we detected CD4+ Teff cell responses to S and N in the majority of subjects, with a lower frequency of CD8 and cTfh responses detected.

A similar proportion of subjects showed a positive response to SARS-CoV-2 based on intracellular IL-2 or IFN-γ (83%), activation markers (88%), or cytokine secretion (83%). Although there was a wide range in the magnitude of responses, all subjects showed a positive response to at least one SARS-CoV-2 Ag in at least one T cell assay. The SARS-CoV-2–specific T cells showed a CD4-dominant response and a lower frequency of IFN-γ–producing compared with IL-2–producing CD4+ T cells based on proliferation and ICC data. Analysis of secreted molecules from total PBMC cultures showed a strong Th1 profile dominated by IFN-γ, albeit with a lower IFN-γ/IL-2 ratio than the response to whole IAV within the same subjects. Ab responses to SARS-CoV-2 were detected in 95% of subjects, and there was a correlation between saliva and plasma IgG. Our findings demonstrate that most individuals in a cohort of recovered, mainly mild cases have detectable T and Ab responses to SARS-CoV-2, with evidence for a subset of subjects in whom these responses last as long as 9 mo postinfection. These findings are consistent with studies showing 80–85% protection from reinfection with SARS-CoV-2 in the under-65 age group with up to 7 mo of follow-up (5154).

Plasma IgG responses to S, although variable between individuals, were remarkably stable over the course of the study, decaying at 0.5%/wk (t1/2 = 23 mo), whereas anti-RBD IgG response had a t1/2 of ∼6 mo. S- and RBD-specific Ab responses in saliva fell off more rapidly. Nonetheless, there was a correlation between the plasma and saliva anti-S and anti-RBD IgG levels, as previously noted (12). Given that anti–SARS-CoV-2 Ab titers in the saliva are lower than in the plasma at all time points, it is possible that the more rapid decline in saliva may reflect levels of Ab in the saliva that drop below the limit of detection of the assay. There was also a correlation between levels of plasma IgG and IgA and the cTfh and Teff cell responses. The longevity of IgG responses to S and the slightly more rapid decay of anti-RBD responses, as well as the more transient IgA and IgM responses, are similar to results reported in earlier studies (12, 22, 34).

On average, we found that CD4+ T cells specific for S or N decayed at a rate of ∼3–4%/wk (t1/2 = 3–4 mo) between the two time points, based on IL-2 secretion. A caveat to these estimates is that the T cell frequencies based on ICC were low, potentially rendering them less quantitative. For this reason, we restricted our analysis of decay rate to those donors with a response of at least 2 SDs above the median background. Of note, our results are quite similar to the results of Dan et al. (34), who reported t1/2 values of 6 mo for CD8+ T cell responses and 3 mo for CD4+ T cell responses, by analyzing paired samples based on activation markers. SARS-CoV-2–specific proliferative responses of CD4+ T cells were more stable between time points 1 and 2 than the proliferative responses of CD8+ T cells, perhaps reflecting the higher proportion of Tcm cells in the CD4+ population and a higher proportion of Temra cells in the CD8+ T cell populations. Bilich et al. reported stable CD8+ T cell responses and increasing CD4+ T cell responses over two time points (median, 40 d and 159 d, respectively) based on ICC or ELISPOT assays in a cohort of SARS-CoV-2 convalescent subjects, most with mild disease (36). However, the study of Bilich et al. (36) was focused on a series of defined epitopes, whereas the study of Dan et al. (34) and our present study used overlapping peptide pools covering the full Ag sequences. Thus, although the overall T cell response to SARS-CoV-2 declines over time, it is possible that subsets of epitope-specific T cells form longer-lived subsets, which displace other T cells over time.

The responses to IAV in our study were generally more stable than the response to SARS-CoV-2 within the same subjects. On the one hand, because boosting is known to increase the duration of T cell immunity (55), this finding likely reflects that the IAV-specific memory has been boosted over a lifetime of exposure or vaccination, whereas the response to SARS-CoV-2 represents the memory response to a single infection. On the other hand, the primary response to the yellow fever vaccine strain YF17D in humans gives rise to long-lived HLA-A2/NS4b–specific CD8+ T cells with a t1/2 of 493 d (56). Thus, the longevity of T cell responses likely reflects the stimulation history as well as the particular virus and epitope studied. A limitation of our analysis is that we only examined two time points during the first 9 mo of infection. We do not expect the decay to be linear, and therefore these data cannot be extrapolated to predict the longevity of the response.

We also compared the immunophenotype of SARS-CoV-2–specific and IAV-specific recall responses in this study. A caveat to this comparison is that IAV recall responses represent a lifetime of exposures to IAV compared with a new infection with SARS-CoV-2, and we compared live IAV stimulation with the response to SARS-CoV-2 peptide pools. Both IAV-specific and SARS-CoV-2–specific T cells showed a mixture of memory phenotypes, with SARS-CoV-2 S-specific CD4+ T cell responses showing a higher proportion of Tcm cells (CD45RACD27+) than IAV-specific responses, whereas IAV-specific responses showed a higher proportion of Tem cells (CD45RACD27) at both time points. Both IAV- and SARS-CoV-2–specific T cells included a proportion of CD45RA+CD27+ T cells. These cells were detected based on Ag-specific cytokine production, so they likely represent the Tsc phenotype defined by Gattinoni et al. because these cells have a surface phenotype that is very similar to that of naive T cells (43).

Based on ICC staining, S1-, S2-, N-, and M-specific CD4+ T cells were detected in most donors, whereas responses to E were quite rare. More than 80% of these Ag-specific CD4+ T cells produced only one cytokine, with IL-2–producing T cells most frequent, followed by IFN-γ and then TNF. Within the same samples, IFN-γ–producing cells predominated over IL-2–producing CD4+ T cells following stimulation of the PBMCs with IAV. Moreover, within each donor, there was a higher frequency of CD8+ over CD4+ IAV-specific T cells, whereas SARS-CoV-2 S-specific responses showed a predominance of CD4+ over CD8+ responses, as previously noted (8, 11, 16, 19). IAV stimulation involves whole virus and therefore contains epitopes from multiple viral proteins. To address this limitation, we compared the sum of all responses with the five SARS-CoV-2 peptide pools with the response to whole IAV. Although there was a similar frequency of IL-2–producing cells in response to SARS-CoV-2 or IAV, the IAV response had a higher proportion of IFN-γ–producing cells. We do not think this altered cytokine profile reflects innate immune responses to the live IAV in the PBMC cultures, because our previous work showed a similar frequency of IFN-γ–producing cells in response to IAV and trivalent inactivated influenza A vaccine, which contains no nucleic acid (19). Moreover, IAV does not productively infect PBMCs. Thus, the different proportion of cytokines within each response suggests a different phenotype in the SARS-CoV-2 S- or N-specific versus IAV-specific CD4+ response that is not explained by the different magnitude of the response.

In a previous study, in which about half the study subjects were hospitalized or ICU subjects, we also observed that the most frequent SARS-CoV-2–specific T cells produced IL-2 (19). However, in that study, TNF-producing CD4+ T cells were also more frequent than IFN-γ–producing cells (19), whereas in the present study, TNF-producing CD4+ T cells were less frequent. It is possible that the higher proportion of hospitalized cases in the previous study impacted the levels of TNF-producing cells observed (19). Because the median time of sampling in the previous study, 37 d (range, 27–90 d), was not that different from time point 1 of the present study, median 45 d (range, 30–154 d), we think it is unlikely that time since infection impacts the proportion of TNF-producing cells, albeit this cannot be ruled out.

Based on the multiplex bead assay, the most predominant cytokines secreted in response to S and N were IFN-γ and IL-6. Whether IL-6 was produced directly by activated CD4+ or CD8+ T cells or by other cells responding to activated T cells or the molecules they secrete cannot be deduced from these data. Unsupervised clustering of the 13 secreted molecules largely segregated SARS-CoV-2–specific from IAV-specific responses, particularly at the later time point. IAV-specific responses were higher overall and showed a higher cytotoxicity profile, likely due to the stronger stimulation from whole virus compared with individual peptide pools. However, by looking at the ratio of cytokines produced within each donor, we observed an altered cytokine profile for SARS-CoV-2–specific as compared with IAV-specific responses, findings that recapitulated the results from ICC analysis. Notably, within each donor, the ratio of IFN-γ to IL-2 was lower for S- and N-specific T cell responses than for IAV-specific responses.

In this study we used three different assays to characterize T cell responses to SARS-CoV-2. The flow cytometry–based ICC assay, which detects Ag-specific responses based on 18 h of restimulation, has the advantage of providing single-cell data on cytokine production and other markers at a time point before cells start to divide, thus providing an estimate of Ag-specific T cells in the blood immediately ex vivo. However, the ICC assay has the disadvantage that the frequency of responses we measured was quite low in many of the subjects, particularly for the CD8+ T cell responses, rendering the analysis less quantitative. To compensate for this, we also used a proliferation assay, which confirmed the lower CD8+ to CD4+ ratio for SARS-CoV-2 S-specific responses than for IAV-specific responses, observed by ICC. A caveat to the proliferation assay, however, is that the long period of T cell expansion could amplify weak cross-reactive T cell responses to seasonal CoV, as pointed out in a recent study (57). We also measured cytokines produced in the culture supernatants after 48 h of Ag stimulation using a quantitative commercial bead-based assay. A limitation of this assay is that it does not indicate which cell produced the cytokine in question, albeit the cytokines were specifically detected in response to Ag stimulation. Despite these limitations, the altered cytokine profile detected in SARS-CoV-2–specific as compared with IAV-specific T cells by ICC staining was recapitulated by the multiplex cytokine assays, with a similar proportion of subjects showing a positive response and a similar alteration in IL-2/IFN-γ ratio compared with IAV. The multiplex assay may be useful for larger studies due to its ease of high-throughput analysis.

In summary, we report that SARS-CoV-2–specific T cell responses in a cohort of Toronto donors predominated by nonhospitalized cases have a high proportion of IL-2–producing CD4+ T cells, with a less frequent CD8+ T cell response and a potentially altered cytotoxic mechanism. A limitation of our study is that the small number of study subjects (n = 24) could result in sampling errors such that the data are not representative of the population as a whole, albeit other studies have shown a similar preponderance of CD4+ over CD8+ T cells in response to SARS-CoV-2 Ags (8, 11, 16, 19, 35) and a high proportion of CD4 IL-2–producing cells in response to S (35). Another limitation of our study is that the IAV-specific responses measured represent a lifetime of exposure to IAV and stimulation with whole virus, whereas SARS-CoV-2–specific responses represent the response to a new infection and peptide pools representing a subset of Ags. Nonetheless, the data suggest that SARS-CoV-2–specific T cell responses are distinct from the typical response to the respiratory pathogen IAV.

We thank Birinder Ghumman for technical assistance; Andrew Law for assistance with data analysis and Seaborn implementation; Nathalie Simard and Janine Charron for assistance with flow cytometry; Yves Durocher at the NRC for antigens, calibration standards, and secondary antibodies; members of the Gingras laboratory, Adrian Pasculescu, and the Network Biology Collaborative Centre for assistance with sample intake and Ab data generation and analysis; Thierry Mallevaey for critical reading of the manuscript; and all the study subjects for participating and donating samples to make this study possible.

This work was funded by Canadian Institutes of Health Research Grant VR1-172711 to T.H.W., J.L.G., J.M.R., M.A.O., A.-C.G., S.M., and A.J.M. with an additional supplement from the COVID-19 Immunity Task Force. J.L.G. received funding from an Ontario Together grant. Patient recruitment was funded by Canadian Institutes of Health Research Grant RN419944-439999 and by the Juan and Stefania Speck fund for COVID-19 and other viral infections (to M.A.O.). J.C.L. was funded by a Queen Elizabeth II/Aventis Pasteur Graduate Scholarship in Science and Technology at the University of Toronto.

The online version of this article contains supplemental material.

Abbreviations used in this article

COVID-19

coronavirus disease 2019

cTfh

circulating T follicular helper

E

envelope

HAU

hemagglutination unit

IAV

influenza A virus

ICC

intracellular cytokine

ICU

intensive care unit

M

membrane

N

nucleocapsid

PSO

postsympton onset

RBD

receptor binding domain

S

spike

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2

Tcm

central memory T

Teff

T effector

Tem

effector memory T

Temra

effector memory T expressing CD45RA

Tsc

stem cell memory T

1.
Anderson
R. M.
,
C.
Vegvari
,
J.
Truscott
,
B. S.
Collyer
.
2020
.
Challenges in creating herd immunity to SARS-CoV-2 infection by mass vaccination.
Lancet
396
:
1614
1616
.
2.
Blanco-Melo
D.
,
B. E.
Nilsson-Payant
,
W. C.
Liu
,
S.
Uhl
,
D.
Hoagland
,
R.
Møller
,
T. X.
Jordan
,
K.
Oishi
,
M.
Panis
,
D.
Sachs
, et al
2020
.
Imbalanced host response to SARS-CoV-2 drives development of COVID-19.
Cell
181
:
1036
1045.e9
.
3.
Lei
X.
,
X.
Dong
,
R.
Ma
,
W.
Wang
,
X.
Xiao
,
Z.
Tian
,
C.
Wang
,
Y.
Wang
,
L.
Li
,
L.
Ren
, et al
2020
.
Activation and evasion of type I interferon responses by SARS-CoV-2.
Nat. Commun.
11
:
3810
.
4.
Xia
H.
,
Z.
Cao
,
X.
Xie
,
X.
Zhang
,
J. Y.
Chen
,
H.
Wang
,
V. D.
Menachery
,
R.
Rajsbaum
,
P. Y.
Shi
.
2020
.
Evasion of type I interferon by SARS-CoV-2.
Cell Rep.
33
:
108234
.
5.
Miorin
L.
,
T.
Kehrer
,
M. T.
Sanchez-Aparicio
,
K.
Zhang
,
P.
Cohen
,
R. S.
Patel
,
A.
Cupic
,
T.
Makio
,
M.
Mei
,
E.
Moreno
, et al
2020
.
SARS-CoV-2 Orf6 hijacks Nup98 to block STAT nuclear import and antagonize interferon signaling.
Proc. Natl. Acad. Sci. USA
117
:
28344
28354
.
6.
Bastard
P.
,
L. B.
Rosen
,
Q.
Zhang
,
E.
Michailidis
,
H. H.
Hoffmann
,
Y.
Zhang
,
K.
Dorgham
,
Q.
Philippot
,
J.
Rosain
,
V.
Béziat
, et al
COVID Human Genetic Effort
.
2020
.
Autoantibodies against type I IFNs in patients with life-threatening COVID-19.
Science
370
:
eabd4585
.
7.
Zhang
Q.
,
P.
Bastard
,
Z.
Liu
,
J.
Le Pen
,
M.
Moncada-Velez
,
J.
Chen
,
M.
Ogishi
,
I. K. D.
Sabli
,
S.
Hodeib
,
C.
Korol
, et al
NIAID-USUHS/TAGC COVID Immunity Group
.
2020
.
Inborn errors of type I IFN immunity in patients with life-threatening COVID-19.
Science
370
:
eabd4570
.
8.
Sekine
T.
,
A.
Perez-Potti
,
O.
Rivera-Ballesteros
,
K.
Strålin
,
J. B.
Gorin
,
A.
Olsson
,
S.
Llewellyn-Lacey
,
H.
Kamal
,
G.
Bogdanovic
,
S.
Muschiol
, et al
Karolinska COVID-19 Study Group
.
2020
.
Robust T cell immunity in convalescent individuals with asymptomatic or mild COVID-19.
Cell
183
:
158
168.e14
.
9.
Weiskopf
D.
,
K. S.
Schmitz
,
M. P.
Raadsen
,
A.
Grifoni
,
N. M. A.
Okba
,
H.
Endeman
,
J. P. C.
van den Akker
,
R.
Molenkamp
,
M. P. G.
Koopmans
,
E. C. M.
van Gorp
, et al
2020
.
Phenotype and kinetics of SARS-CoV-2-specific T cells in COVID-19 patients with acute respiratory distress syndrome.
Sci. Immunol.
5
:
eabd2071
.
10.
Crawford
K. H. D.
,
A. S.
Dingens
,
R.
Eguia
,
C. R.
Wolf
,
N.
Wilcox
,
J. K.
Logue
,
K.
Shuey
,
A. M.
Casto
,
B.
Fiala
,
S.
Wrenn
, et al
2021
.
Dynamics of neutralizing antibody titers in the months after severe acute respiratory syndrome coronavirus 2 infection.
J. Infect. Dis.
223
:
197
205
.
11.
Grifoni
A.
,
D.
Weiskopf
,
S. I.
Ramirez
,
J.
Mateus
,
J. M.
Dan
,
C. R.
Moderbacher
,
S. A.
Rawlings
,
A.
Sutherland
,
L.
Premkumar
,
R. S.
Jadi
, et al
2020
.
Targets of T cell responses to SARS-CoV-2 coronavirus in humans with COVID-19 disease and unexposed individuals.
Cell
181
:
1489
1501.e15
.
12.
Isho
B.
,
K. T.
Abe
,
M.
Zuo
,
A. J.
Jamal
,
B.
Rathod
,
J. H.
Wang
,
Z.
Li
,
G.
Chao
,
O. L.
Rojas
,
Y. M.
Bang
, et al
2020
.
Persistence of serum and saliva antibody responses to SARS-CoV-2 spike antigens in COVID-19 patients.
Sci. Immunol.
5
:
eabe5511
.
13.
Le Bert
N.
,
A. T.
Tan
,
K.
Kunasegaran
,
C. Y. L.
Tham
,
M.
Hafezi
,
A.
Chia
,
M. H. Y.
Chng
,
M.
Lin
,
N.
Tan
,
M.
Linster
, et al
2020
.
SARS-CoV-2-specific T cell immunity in cases of COVID-19 and SARS, and uninfected controls.
Nature
584
:
457
462
.
14.
Ni
L.
,
F.
Ye
,
M. L.
Cheng
,
Y.
Feng
,
Y. Q.
Deng
,
H.
Zhao
,
P.
Wei
,
J.
Ge
,
M.
Gou
,
X.
Li
, et al
2020
.
Detection of SARS-CoV-2-specific humoral and cellular immunity in COVID-19 convalescent individuals.
Immunity
52
:
971
977.e3
.
15.
Rodda
L. B.
,
J.
Netland
,
L.
Shehata
,
K. B.
Pruner
,
P. A.
Morawski
,
C. D.
Thouvenel
,
K. K.
Takehara
,
J.
Eggenberger
,
E. A.
Hemann
,
H. R.
Waterman
, et al
2021
.
Functional SARS-CoV-2-specific immune memory persists after mild COVID-19.
Cell
184
:
169
183.e17
.
16.
Rydyznski Moderbacher
C.
,
S. I.
Ramirez
,
J. M.
Dan
,
A.
Grifoni
,
K. M.
Hastie
,
D.
Weiskopf
,
S.
Belanger
,
R. K.
Abbott
,
C.
Kim
,
J.
Choi
, et al
2020
.
Antigen-specific adaptive immunity to SARS-CoV-2 in acute COVID-19 and associations with age and disease severity.
Cell
183
:
996
1012.e19
.
17.
Tan
A. T.
,
M.
Linster
,
C. W.
Tan
,
N.
Le Bert
,
W. N.
Chia
,
K.
Kunasegaran
,
Y.
Zhuang
,
C. Y. L.
Tham
,
A.
Chia
,
G. J. D.
Smith
, et al
2021
.
Early induction of functional SARS-CoV-2-specific T cells associates with rapid viral clearance and mild disease in COVID-19 patients.
Cell Rep.
34
:
108728
.
18.
Peng
Y.
,
A. J.
Mentzer
,
G.
Liu
,
X.
Yao
,
Z.
Yin
,
D.
Dong
,
W.
Dejnirattisai
,
T.
Rostron
,
P.
Supasa
,
C.
Liu
, et al
ISARIC4C Investigators
.
2020
.
Broad and strong memory CD4+ and CD8+ T cells induced by SARS-CoV-2 in UK convalescent individuals following COVID-19.
Nat. Immunol.
21
:
1336
1345
.
19.
Law
J. C.
,
W. H.
Koh
,
P.
Budylowski
,
J.
Lin
,
F.
Yue
,
K. T.
Abe
,
B.
Rathod
,
M.
Girard
,
Z.
Li
,
J. M.
Rini
, et al
2021
.
Systematic examination of antigen-specific recall T cell responses to SARS-CoV-2 versus influenza virus reveals a distinct inflammatory profile.
J. Immunol.
206
:
37
50
.
20.
Le Bert
N.
,
H. E.
Clapham
,
A. T.
Tan
,
W. N.
Chia
,
C. Y. L.
Tham
,
J. M.
Lim
,
K.
Kunasegaran
,
L. W. L.
Tan
,
C. A.
Dutertre
,
N.
Shankar
, et al
2021
.
Highly functional virus-specific cellular immune response in asymptomatic SARS-CoV-2 infection.
J. Exp. Med.
218
:
e20202617
.
21.
Sette
A.
,
S.
Crotty
.
2021
.
Adaptive immunity to SARS-CoV-2 and COVID-19.
Cell
184
:
861
880
.
22.
Iyer
A. S.
,
F. K.
Jones
,
A.
Nodoushani
,
M.
Kelly
,
M.
Becker
,
D.
Slater
,
R.
Mills
,
E.
Teng
,
M.
Kamruzzaman
,
W. F.
Garcia-Beltran
, et al
2020
.
Persistence and decay of human antibody responses to the receptor binding domain of SARS-CoV-2 spike protein in COVID-19 patients.
Sci. Immunol.
5
:
eabe0367
.
23.
Long
Q. X.
,
B. Z.
Liu
,
H. J.
Deng
,
G. C.
Wu
,
K.
Deng
,
Y. K.
Chen
,
P.
Liao
,
J. F.
Qiu
,
Y.
Lin
,
X. F.
Cai
, et al
2020
.
Antibody responses to SARS-CoV-2 in patients with COVID-19.
Nat. Med.
26
:
845
848
.
24.
Muecksch
F.
,
H.
Wise
,
B.
Batchelor
,
M.
Squires
,
E.
Semple
,
C.
Richardson
,
J.
McGuire
,
S.
Clearly
,
E.
Furrie
,
N.
Greig
, et al
2021
.
Longitudinal serological analysis and neutralizing antibody levels in coronavirus disease 2019 convalescent patients.
J. Infect. Dis.
223
:
389
398
.
25.
Wheatley
A. K.
,
J. A.
Juno
,
J. J.
Wang
,
K. J.
Selva
,
A.
Reynaldi
,
H. X.
Tan
,
W. S.
Lee
,
K. M.
Wragg
,
H. G.
Kelly
,
R.
Esterbauer
, et al
2021
.
Evolution of immune responses to SARS-CoV-2 in mild-moderate COVID-19.
Nat. Commun.
12
:
1162
.
26.
Nelde
A.
,
T.
Bilich
,
J. S.
Heitmann
,
Y.
Maringer
,
H. R.
Salih
,
M.
Roerden
,
M.
Lübke
,
J.
Bauer
,
J.
Rieth
,
M.
Wacker
, et al
2021
.
SARS-CoV-2-derived peptides define heterologous and COVID-19-induced T cell recognition.
Nat. Immunol.
22
:
74
85
.
27.
Atyeo
C.
,
S.
Fischinger
,
T.
Zohar
,
M. D.
Slein
,
J.
Burke
,
C.
Loos
,
D. J.
McCulloch
,
K. L.
Newman
,
C.
Wolf
,
J.
Yu
, et al
2020
.
Distinct early serological signatures track with SARS-CoV-2 survival.
Immunity
53
:
524
532.e4
.
28.
Garcia-Beltran
W. F.
,
E. C.
Lam
,
M. G.
Astudillo
,
D.
Yang
,
T. E.
Miller
,
J.
Feldman
,
B. M.
Hauser
,
T. M.
Caradonna
,
K. L.
Clayton
,
A. D.
Nitido
, et al
2021
.
COVID-19-neutralizing antibodies predict disease severity and survival.
Cell
184
:
476
488.e11
.
29.
Bartsch
Y. C.
,
C.
Wang
,
T.
Zohar
,
S.
Fischinger
,
C.
Atyeo
,
J. S.
Burke
,
J.
Kang
,
A. G.
Edlow
,
A.
Fasano
,
L. R.
Baden
, et al
2021
.
Humoral signatures of protective and pathological SARS-CoV-2 infection in children.
Nat. Med.
27
:
454
462
.
30.
Lucas
C.
,
J.
Klein
,
M. E.
Sundaram
,
F.
Liu
,
P.
Wong
,
J.
Silva
,
T.
Mao
,
J. E.
Oh
,
S.
Mohanty
,
J.
Huang
, et al
Yale IMPACT Research Team
.
2021
.
Delayed production of neutralizing antibodies correlates with fatal COVID-19. [Published erratum appears in 2021 Nat. Med. 27: 1309.]
Nat. Med.
27
:
1178
1186
.
31.
Weisberg
S. P.
,
T. J.
Connors
,
Y.
Zhu
,
M. R.
Baldwin
,
W. H.
Lin
,
S.
Wontakal
,
P. A.
Szabo
,
S. B.
Wells
,
P.
Dogra
,
J.
Gray
, et al
2021
.
Distinct antibody responses to SARS-CoV-2 in children and adults across the COVID-19 clinical spectrum.
Nat. Immunol.
22
:
25
31
.
32.
Szabo
P. A.
,
P.
Dogra
,
J. I.
Gray
,
S. B.
Wells
,
T. J.
Connors
,
S. P.
Weisberg
,
I.
Krupska
,
R.
Matsumoto
,
M. M. L.
Poon
,
E.
Idzikowski
, et al
2021
.
Longitudinal profiling of respiratory and systemic immune responses reveals myeloid cell-driven lung inflammation in severe COVID-19.
Immunity
54
:
797
814.e6
.
33.
Bergamaschi
L.
,
F.
Mescia
,
L.
Turner
,
A. L.
Hanson
,
P.
Kotagiri
,
B. J.
Dunmore
,
H.
Ruffieux
,
A.
De Sa
,
O.
Huhn
,
M. D.
Morgan
, et al
Cambridge Institute of Therapeutic Immunology and Infectious Disease-National Institute of Health Research (CITIID-NIHR) COVID BioResource Collaboration
.
2021
.
Longitudinal analysis reveals that delayed bystander CD8+ T cell activation and early immune pathology distinguish severe COVID-19 from mild disease.
Immunity
54
:
1257
1275.e8
.
34.
Dan
J. M.
,
J.
Mateus
,
Y.
Kato
,
K. M.
Hastie
,
E. D.
Yu
,
C. E.
Faliti
,
A.
Grifoni
,
S. I.
Ramirez
,
S.
Haupt
,
A.
Frazier
, et al
2021
.
Immunological memory to SARS-CoV-2 assessed for up to 8 months after infection.
Science
371
:
eabf4063
.
35.
Breton
G.
,
P.
Mendoza
,
T.
Hägglöf
,
T. Y.
Oliveira
,
D.
Schaefer-Babajew
,
C.
Gaebler
,
M.
Turroja
,
A.
Hurley
,
M.
Caskey
,
M. C.
Nussenzweig
.
2021
.
Persistent cellular immunity to SARS-CoV-2 infection.
J. Exp. Med.
218
:
e20202515
.
36.
Bilich
T.
,
A.
Nelde
,
J. S.
Heitmann
,
Y.
Maringer
,
M.
Roerden
,
J.
Bauer
,
J.
Rieth
,
M.
Wacker
,
A.
Peter
,
S.
Hörber
, et al
2021
.
T cell and antibody kinetics delineate SARS-CoV-2 peptides mediating long-term immune responses in COVID-19 convalescent individuals.
Sci. Transl. Med.
13
:
eabf7517
.
37.
Kang
C. K.
,
M.
Kim
,
S.
Lee
,
G.
Kim
,
P. G.
Choe
,
W. B.
Park
,
N. J.
Kim
,
C. H.
Lee
,
I. S.
Kim
,
K.
Jung
, et al
2021
.
Longitudinal analysis of human memory T-cell response according to the severity of illness up to 8 months after severe acute respiratory syndrome coronavirus 2 infection.
J. Infect. Dis.
224
:
39
48
.
38.
Anand
S. P.
,
J.
Prévost
,
M.
Nayrac
,
G.
Beaudoin-Bussières
,
M.
Benlarbi
,
R.
Gasser
,
N.
Brassard
,
A.
Laumaea
,
S. Y.
Gong
,
C.
Bourassa
, et al
2021
.
Longitudinal analysis of humoral immunity against SARS-CoV-2 spike in convalescent individuals up to 8 months post-symptom onset.
Cell Rep. Med.
2
:
100290
.
39.
Vanshylla
K.
,
V.
Di Cristanziano
,
F.
Kleipass
,
F.
Dewald
,
P.
Schommers
,
L.
Gieselmann
,
H.
Gruell
,
M.
Schlotz
,
M. S.
Ercanoglu
,
R.
Stumpf
, et al
2021
.
Kinetics and correlates of the neutralizing antibody response to SARS-CoV-2 infection in humans.
Cell Host Microbe
29
:
917
929.e4
.
40.
Chia
W. N.
,
F.
Zhu
,
S. W. X.
Ong
,
B. E.
Young
,
S. W.
Fong
,
N.
Le Bert
,
C. W.
Tan
,
C.
Tiu
,
J.
Zhang
,
S. Y.
Tan
, et al
2021
.
Dynamics of SARS-CoV-2 neutralising antibody responses and duration of immunity: a longitudinal study.
Lancet Microbe
2
:
e240
e249
.
41.
Wrapp
D.
,
D.
De Vlieger
,
K. S.
Corbett
,
G. M.
Torres
,
N.
Wang
,
W.
Van Breedam
,
K.
Roose
,
L.
van Schie
,
M.
Hoffmann
,
S.
Pöhlmann
, et al
VIB-CMB COVID-19 Response Team
.
2020
.
Structural basis for potent neutralization of betacoronaviruses by single-domain camelid antibodies. [Published erratum appears in 2020 Cell 181: 1436–1441.]
Cell
181
:
1004
1015.e15
.
42.
Abe
K. T.
,
Z.
Li
,
R.
Samson
,
P.
Samavarchi-Tehrani
,
E. J.
Valcourt
,
H.
Wood
,
P.
Budylowski
,
A. P.
Dupuis
II
,
R. C.
Girardin
,
B.
Rathod
, et al
2020
.
A simple protein-based surrogate neutralization assay for SARS-CoV-2.
JCI Insight
5
:
e142362
.
43.
Gattinoni
L.
,
E.
Lugli
,
Y.
Ji
,
Z.
Pos
,
C. M.
Paulos
,
M. F.
Quigley
,
J. R.
Almeida
,
E.
Gostick
,
Z.
Yu
,
C.
Carpenito
, et al
2011
.
A human memory T cell subset with stem cell-like properties.
Nat. Med.
17
:
1290
1297
.
44.
McMahan
K.
,
J.
Yu
,
N. B.
Mercado
,
C.
Loos
,
L. H.
Tostanoski
,
A.
Chandrashekar
,
J.
Liu
,
L.
Peter
,
C.
Atyeo
,
A.
Zhu
, et al
2021
.
Correlates of protection against SARS-CoV-2 in rhesus macaques.
Nature
590
:
630
634
.
45.
Khoury
D. S.
,
D.
Cromer
,
A.
Reynaldi
,
T. E.
Schlub
,
A. K.
Wheatley
,
J. A.
Juno
,
K.
Subbarao
,
S. J.
Kent
,
J. A.
Triccas
,
M. P.
Davenport
.
2021
.
Neutralizing antibody levels are highly predictive of immune protection from symptomatic SARS-CoV-2 infection.
Nat. Med.
27
:
1205
1211
.
46.
Voss
W. N.
,
Y. J.
Hou
,
N. V.
Johnson
,
G.
Delidakis
,
J. E.
Kim
,
K.
Javanmardi
,
A. P.
Horton
,
F.
Bartzoka
,
C. J.
Paresi
,
Y.
Tanno
, et al
2021
.
Prevalent, protective, and convergent IgG recognition of SARS-CoV-2 non-RBD spike epitopes.
Science
372
:
1108
1112
.
47.
Crotty
S.
2019
.
T follicular helper cell biology: a decade of discovery and diseases.
Immunity
50
:
1132
1148
.
48.
Morita
R.
,
N.
Schmitt
,
S. E.
Bentebibel
,
R.
Ranganathan
,
L.
Bourdery
,
G.
Zurawski
,
E.
Foucat
,
M.
Dullaers
,
S.
Oh
,
N.
Sabzghabaei
, et al
2011
.
Human blood CXCR5+CD4+ T cells are counterparts of T follicular cells and contain specific subsets that differentially support antibody secretion.
Immunity
34
:
108
121
.
49.
Locci
M.
,
C.
Havenar-Daughton
,
E.
Landais
,
J.
Wu
,
M. A.
Kroenke
,
C. L.
Arlehamn
,
L. F.
Su
,
R.
Cubas
,
M. M.
Davis
,
A.
Sette
, et al
International AIDS Vaccine Initiative Protocol C Principal Investigators
.
2013
.
Human circulating PD-1+CXCR3CXCR5+ memory Tfh cells are highly functional and correlate with broadly neutralizing HIV antibody responses.
Immunity
39
:
758
769
.
50.
Bentebibel
S. E.
,
S.
Lopez
,
G.
Obermoser
,
N.
Schmitt
,
C.
Mueller
,
C.
Harrod
,
E.
Flano
,
A.
Mejias
,
R. A.
Albrecht
,
D.
Blankenship
, et al
2013
.
Induction of ICOS+CXCR3+CXCR5+ TH cells correlates with antibody responses to influenza vaccination.
Sci. Transl. Med.
5
:
176ra32
.
51.
Hansen
C. H.
,
D.
Michlmayr
,
S. M.
Gubbels
,
K.
Mølbak
,
S.
Ethelberg
.
2021
.
Assessment of protection against reinfection with SARS-CoV-2 among 4 million PCR-tested individuals in Denmark in 2020: a population-level observational study.
Lancet
397
:
1204
1212
.
52.
Boyton
R. J.
,
D. M.
Altmann
.
2021
.
Risk of SARS-CoV-2 reinfection after natural infection.
Lancet
397
:
1161
1163
.
53.
Rennert
L.
,
C.
McMahan
.
2021
.
Risk of SARS-CoV-2 reinfection in a university student population.
Clin. Infect. Dis.
DOI: 10.1093/cid/ciab454.
54.
Lumley
S. F.
,
D.
O’Donnell
,
N. E.
Stoesser
,
P. C.
Matthews
,
A.
Howarth
,
S. B.
Hatch
,
B. D.
Marsden
,
S.
Cox
,
T.
James
,
F.
Warren
, et al
Oxford University Hospitals Staff Testing Group
.
2021
.
Antibody status and incidence of SARS-CoV-2 infection in health care workers.
N. Engl. J. Med.
384
:
533
540
.
55.
Harty
J. T.
,
V. P.
Badovinac
.
2008
.
Shaping and reshaping CD8+ T-cell memory.
Nat. Rev. Immunol.
8
:
107
119
.
56.
Akondy
R. S.
,
M.
Fitch
,
S.
Edupuganti
,
S.
Yang
,
H. T.
Kissick
,
K. W.
Li
,
B. A.
Youngblood
,
H. A.
Abdelsamed
,
D. J.
McGuire
,
K. W.
Cohen
, et al
2017
.
Origin and differentiation of human memory CD8 T cells after vaccination.
Nature
552
:
362
367
.
57.
Ogbe
A.
,
B.
Kronsteiner
,
D. T.
Skelly
,
M.
Pace
,
A.
Brown
,
E.
Adland
,
K.
Adair
,
H. D.
Akhter
,
M.
Ali
,
S. E.
Ali
, et al
Oxford Protective T Cell Immunology for COVID-19 (OPTIC) Clinical Team
.
2021
.
T cell assays differentiate clinical and subclinical SARS-CoV-2 infections from cross-reactive antiviral responses.
Nat. Commun.
12
:
2055
.

The authors have no financial conflicts of interest.

Supplementary data