ARTICLE https://doi.org/10.1038/s41467-021-23333-3 OPEN Peripheral and lung resident memory T cell responses against SARS-CoV-2 Judith Grau-Expósito 1,9, Nerea Sánchez-Gaona1,9, Núria Massana1, Marina Suppi 1, Antonio Astorga-Gamaza1, David Perea1, Joel Rosado2, Anna Falcó1, Cristina Kirkegaard1, Ariadna Torrella 1, Bibiana Planas1, Jordi Navarro 1, Paula Suanzes 1, Daniel Álvarez-Sierra 3, Alfonso Ayora4, Irene Sansano5,6, Juliana Esperalba7, Cristina Andrés 7, Andrés Antón 7, Santiago Ramón y Cajal 5,6, Benito Almirante1, Ricardo Pujol-Borrell 3,8, Vicenç Falcó 1, Joaquín Burgos 1, María J. Buzón 1,10 ✉ & 1234567890():,; Meritxell Genescà 1,10,10 ✉ Resident memory T cells (TRM) positioned within the respiratory tract are probably required to limit SARS-CoV-2 spread and COVID-19. Importantly, TRM are mostly non-recirculating, which reduces the window of opportunity to examine these cells in the blood as they move to the lung parenchyma. Here, we identify circulating virus-specific T cell responses during acute infection with functional, migratory and apoptotic patterns modulated by viral proteins and associated with clinical outcome. Disease severity is associated predominantly with IFNγ and IL-4 responses, increased responses against S peptides and apoptosis, whereas non- hospitalized patients have increased IL-12p70 levels, degranulation in response to N peptides and SARS-CoV-2-specific CCR7+ T cells secreting IL-10. In convalescent patients, lung-TRM are frequently detected even 10 months after initial infection, in which contemporaneous blood does not reflect tissue-resident profiles. Our study highlights a balanced anti- inflammatory antiviral response associated with a better outcome and persisting TRM cells as important for future protection against SARS-CoV-2 infection. 1 Infectious Diseases Department, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Vall d’Hebron Barcelona Hospital Campus, Barcelona, Spain. 2 Thoracic Surgery and Lung Transplantation Department, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Vall d’Hebron Barcelona Hospital Campus, Barcelona, Spain. 3 Diagnostic Immunology Group, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Vall d’Hebron Barcelona Hospital Campus, Barcelona, Spain. 4 Occupational Risk Prevention Unit, Vall d’Hebron Hospital Universitari, Vall d’Hebron Barcelona Hospital Campus, Barcelona, Spain. 5 Pathology Department, Vall d’Hebron Hospital Universitari, Vall d’Hebron Barcelona Hospital Campus, Barcelona, Spain. 6 Departament de Ciències morfològiques, Universitat Autònoma de Barcelona, Universitat Autònoma de Barcelona, Bellaterra, Spain. 7 Respiratory Viruses Unit, Microbiology Department, Vall d’Hebron Institut de Recerca (VHIR), Vall d’Hebron Hospital Universitari, Vall d’Hebron Barcelona Hospital Campus, Barcelona, Spain. 8 FOCIS Center of Excellence, Vall d’Hebron Hospital Universitari, Vall d’Hebron Barcelona Hospital Campus, Barcelona, Spain. 9These authors contributed equally: Judith Grau-Expósito, Nerea Sánchez-Gaona 10The authors jointly supervised this work: María J. Buzón, Meritxell Genescà ✉email: mariajose.buzon@vhir.org; meritxell.genesca@vhir.org NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 T he outbreak of a novel coronavirus SARS-CoV-2 led to a development and monitoring of vaccinated individuals towards global health emergency, the COVID-19 pandemic. While prediction of immune control. a great effort has been focused on vaccine development, many questions remain unanswered that are necessary to prop- Results erly manage patients and inform vaccine assessment. To this end, Cohort characteristics. Patients were recruited during the first identifying the development of a protective immune response pandemic wave of SARS-CoV-2 (spring 2020) at the Hospital after natural infection and characterizing the correlates of pro- Universitari Vall d’Hebron in Barcelona. A total of 46 patients tection would greatly inform on the best strategy to stimulate a were included, in which 14 individuals were symptomatic non- protective response by immunization. Moreover, identifying hospitalized cases, 20 individuals corresponded to mild- specific immunological parameters capable of predicting disease hospitalized cases, and 12 to severe-hospitalized cases. Only one control (i.e., no hospitalization) could provide new biomarkers to patient from the severe group corresponded to a fatal case. support medical decisions for patients. Most efforts to measure or Samples were obtained between 7 and 16 days after symptom induce immunity rely on neutralizing antibodies, which can onset and no differences between groups were detected (Supple- certainly limit infection; however, antibody detection is not only mentary Fig. 1a). Supplementary table 1 shows a summary of the inconsistent in infected or convalescent patients1–4 but may also participant characteristics and baseline determinations, in which wane with time as shown for other coronaviruses5,6, although significant differences are evidenced between the three groups. As their stability may also depend on the antigen targeted7. previously reported26,28,29,35,36, age, lymphopenia, and bio- Virus-specific T cells against SARS-CoV-2 have been shown to chemical parameters such as D-dimer, IL-6, and ferritin were develop against coronaviruses4,8–14 and specific memory T cells associated with disease severity. Quantification of the viral load persisted in SARS-recovered patients for up to 6 years post- between days 5 and 15 after symptom onset is also reported; infection5. Thus, T cells may potentially provide long-term however, values between the groups were not statistically sig- immunity as demonstrated for other viral infections such as SARS nificant. Some of the clinical parameters used to stratify mild and or influenza15–18. In this sense, mouse models of SARS-CoV-1 severe-hospitalized cases are shown in Supplementary Fig. 1b: infection demonstrated that both CD8+T and CD4+T cells are days to discharge since symptoms onset (p < 0.0001, by two-tailed critical for viral clearance15,17. Considering that SARS-CoV-2 Mann–Whitney U-test), baseline IL-6 (p = 0.008, by two-tailed interacts with the host at the respiratory tract mucosal interface, Mann–Whitney U-test) and the percentage of oxyhemoglobin T cells strategically positioned within these surfaces, may be saturation in arterial blood /fraction of inspired oxygen (SAFI) essential to limiting infection. A key role for resident memory ratio at baseline and after 48 h (p = 0.0002, by two-tailed T cells (TRM) in protection against pathogen challenge has been Mann–Whitney U-test). These parameters were used to address established for many tissues, including the lung5,17,19–22. These associations between immunological parameters and disease cells, which are strategically located both in the lung airways and severity. In some analyses, 12 control individuals sampled before in lung interstitial tissue, include CD4+ and CD8+T cells the COVID-19 pandemic were studied in parallel. designed to limit re-infections locally. In the context of respira- To determine if whole plasma cytokine levels in our groups of tory infection models such as influenza, CD8+TRM have shown to COVID-19 patients were similar to previously defined patterns confer cross-protection against different strains22, and both reported before29,37–40, we analyzed cytokine plasma levels by the influenza-specific CD4+ and CD8+TRM have been identified16,23. ELLA microfluidics platform in the same samples. Levels of IL- Importantly, optimal protection against SARS-CoV-infected mice 1ra, IL-2, IL-6, IL-10, IL-15, chemokine (C-X-C motif) ligand was conferred by airway memory CD4+T cells secreting both, (CXCL)10 (IP-10), IFNγ, granzyme B, and tumor necrosis factor proinflammatory interferon (IFN) γ and anti-inflammatory (TNF) were elevated in the plasma of hospitalized groups interleukin (IL)-10 molecules17. Thus, a broader spectrum of T compared to nonhospitalized patients, with higher levels helper (Th) profiles should be included when addressing associated with disease severity (Supplementary Fig. 1c). Of note, virus-specific T cells. Further, recruitment of these cells from the deceased patient from the severe cohort (circled in green) had circulation may depend on the expression of molecules such as very high levels of some of the molecules associated with severity chemokine C-X-C motif receptor (CXCR) 3, which, besides and fatality prediction, namely IL-6 and IL-1ra38,40. Further, mediating chemotaxis toward inflamed tissue of a biased Th1 CXCL10 was the most significant predictor of hospitalization profile, appears critical for the recruitment of pulmonary T cells during acute infection, as previously observed13, which may be a that control infection18,24,25. While such recruitment could reflection of increased IFNγ levels in these patients. C–C partially contribute to the decrease of circulating lymphocytes chemokine ligand 2 (CCL2), also referred to as monocyte and thus favor tissue infiltration, many other factors may explain chemoattractant protein 1, was significantly higher in the the observed lymphopenia associated with COVID-19 disease severe-hospitalized group compared to the nonhospitalized severity26–29. In this regard, increased susceptibility of both patients, while IL-4, IL-7, IL-13, IL-17A, granulocyte- specific and bystander CD4+ and CD8+T lymphocytes to macrophage colony stimulating factor (GM-CSF) were similar apoptotic cell death has been observed in other viral among all three groups of patients (Supplementary Fig. 1c). infections30–32, which could be linked to increased glycolysis33,34. Strikingly, the levels of IL-12p70 were higher in the plasma of the Here, we address several important questions related to early nonhospitalized compared to the mild COVID-19 group, and the control of SARS-CoV-2 infection mediated by cellular immunity deceased patient had the second lowest level of IL-12p70 of and long-term protection: (1) the functional profile of antigen- the severe group (0.093 pg/mL; Supplementary Fig. 1c). specific T cells associated with disease control; (2) if apoptosis is involved in disease severity; (3) if antigen-specific T cell responses have the potential to migrate to the lung and eventually become Disease severity and targeted antigen shape T cell function in TRM cells. To this end, we perform detailed phenotypic and acute infection. The functional capabilities of specific CD4+ and functional analyses in clinically defined groups of patients CD8+T cells against SARS-CoV-2 were measured by intracellular recruited during the first wave of SARS-CoV-2 infection, cytokine staining in samples ranging from 7 to 16 days-post- including the assessment of TRM cells in the lungs of convalescent symptom onset (mean of 12 days) from all three groups. For that, patients. Informing on the immunological parameters associated we stimulated peripheral blood mononuclear cells (PBMC) with with disease control and patient prognosis will aid vaccine overlapping membrane (M), nucleocapsid (N), and spike (S) 2 NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 ARTICLE a ** NON-HOSPITALIZED 1.2 1.0 MILD * SEVERE 0.8 * 0.8 % of CD107a+ % of IFNγ + 0.5 0.6 0.4 ** 0.4 0.3 * 0.2 * 0.2 0.1 0.0 0.0 M M N N S S M N M N S S CD4+ CD8+ CD4+ CD8+ 10 3 * 5 ** 2 % of IL-10+ % of IL-4+ 1.6 1.0 1.2 0.8 0.6 0.8 0.4 0.4 0.2 0.0 0.0 M N S M N S M N M N S S + + + + CD4 CD8 CD4 CD8 IFNγ CD107a CD4 + T cells + CD8 T cells IL-4 b IL-10 M N S M N S Non-hospitalized * Total (%) =0.59 Total (%) =0.66 Total (%) =0.26 =0.2 Total (%) =0.23 Total (%) =0.57 Total (%) =0.06 * * Mild ** Total (%)=0.50 Total (%) =0.51 Total (%) =0.44 Total (%) =0.43 Total (%) =0.52 Total (%) =0.43 * ** * ** Severe ** Total (%) =0.42 Total (%) =0.40 Total (%) =0.43 Total (%) =1.4 Total (%) =0.63 Total (%) =0.62 c M N S M N S Total (%) =0.22 Total (%) =0.22 Total (%) =0.12 Total (%) =1.25 Total (%) =1.96 Total (%) =1.62 peptide sets and determined the expression of IFNγ, IL-4, and IL- response, which included responses to all peptides, for each 10, along with the degranulation marker CD107a in CD4+ and function (Supplementary Fig. 2c). This way, differences in the CD8+T cells (Supplementary Fig. 2a, b). For each function we frequency of IFNγ-secreting antigen-specific T cells were sig- calculated the net response for each peptide set (background nificantly higher among the hospitalized groups compared to the subtracted) and compared these antigen-specific T cell responses outpatients (in CD4+T cells: p = 0.020 for M and S peptides in among all three groups (Fig. 1a). We also analyzed the overall net the mild group; p = 0.004 for M, p = 0.011 for N, and p = 0.007 NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 Fig. 1 Functional characteristics of acute SARS-CoV-2-specific T cells. a Comparison of the net frequency (background subtracted) of interferon (IFN)γ, CD107a, interleukin (IL)-4, and IL-10 expression in SARS-CoV-2-specific CD4+ and CD8+T cells in response to viral proteins (membrane (M), nucleocapsid (N), and spike (S)) between study groups (nonhospitalized n = 14 in orange; mild n = 20 in blue and severe n = 12 in pink). Statistical comparisons were performed using Kruskal–Wallis rank-sum test with Dunn’s multiple comparison test (two-sided): CD4+ IFNγ (M, p = 0.020 and p = 0.004; N, p = 0.011 and S, p = 0.020 and p = 0.007) and IL-10 (M, p = 0.035); CD8+ CD107a (M, p = 0.037), IL-4 (S, p = 0.004 and p = 0.003). b Donut charts summarizing the contribution of each function to the overall CD4+- and CD8+-specific T cell response by targeted viral protein and individual group of patients. Data represent the mean value of the net frequency of each function indicated by color code considering all patients, responders and nonresponders. Total response value (%) is shown under each pie chart and represents the cohort average of the overall net frequency considering all individuals and adding up all functions (nonhospitalized n = 14; mild n = 20 and severe n = 12). Statistical comparisons as performed in a are indicated with * and ** symbols. c Donut charts summarizing the distribution of individual functions among specific CD4+ and CD8+T cells to either the M, N, or S protein from the only fatal case within the severe COVID-19 group. for S peptides in the severe group; Fig. 1a, by Kruskal–Wallis one- with better prognosis in all clinical parameters (Supplementary way ANOVA with Dunn’s multiple comparisons). Of note, while Table 2) and for N-specific positively with better oxygenation at nonhospitalized patients did not show a significant increase in the 48 h (Supplementary Table 2). production of IFNγ as a group, some individuals did show an Actually, when the overall response, including all functions, increase in their response (>0.02% after background subtraction) was represented as donut charts displaying the mean frequency of representing 43% of responders, which was lower than the 80 and responses including all individuals (responders and nonrespon- 92% of responders observed in mild and severe-hospitalized ders), differences among groups in response to each peptide set groups. In contrast degranulation, measured by CD107a expres- were emphasized (Fig. 1b). This was, M peptides were shown to sion, was less detected in general and significance among the mostly stimulate IL-10 secretion in nonhospitalized patients, groups was only reached in response to M peptides in severe while in hospitalized cases, increasing amounts of IFNγ for CD4 patients compared to nonhospitalized patients (Fig. 1a). We also +T cells and of IL-4 and degranulation for CD8+T cell were calculated double positive IFNγ/CD107a CD8+T cells, as a sur- observed (Fig. 1b). In addition, N peptides induced higher rogate of cytotoxic polyfunctional cells, since double positive cells frequencies of antigen-specific CD8+T cells degranulating in mild could be detected in some patients, as exemplified in Supple- and nonhospitalized cases, while S peptides stimulated IL-4 mentary Fig. 2. Interestingly, while the frequency of double secretion mainly in the hospitalized groups (Fig. 1b). Overall, our positive CD107a+IFNγ+CD8+T cells responding to the M pep- analyses indicated, on one hand, group-based differences, where a tides positively correlated with viral load, the same subset specific dominance of IL-4 and IFNγ SARS-CoV-2-specific responses for N peptides inversely correlated with baseline levels of IL-6 were associated with disease severity and of IL-10 to minor within the hospitalized cohort (Supplementary Fig. 3a). disease; on the other hand, we observed targeted protein based Assessment of two other functions, IL-4 and IL-10, demon- differences, where M and N peptides induced a Th1 profile strated major dominance of these responses based on the cohort exemplified by IFNγ in CD4+T cells and degranulation (CD107a) and the viral target. A general induction of an IL-4-specific CD8 in CD8+T cells, respectively, and S peptides induced a biased Th2 +T cell response was observed in response to the viral spike in profile exemplified by IL-4. This pattern was shown in an hospitalized patients compared to the nonhospitalized individuals exaggerated manner in the deceased patient, in which IL-10 (p = 0.004 and p = 0.003 for mild and severe patients, respec- responses were absent and IL-4 together with some IFNγ tively, by Kruskal–Wallis one-way ANOVA with Dunn’s multiple dominated antigen-specific responses (Fig. 1c). comparisons; Fig. 1a). Of note, higher levels of spontaneous secretion of IL-4 (in unstimulated conditions) were observed in hospitalized patients, which essentially correlated with the Patterns of chemokine receptors associated with SARS-CoV-2- number of days since symptoms onset to discharge and baseline infected patients. Next, we aimed to determine if part of the IL-6 levels (Supplementary Fig. 3b). Moreover, SARS-CoV-2 viral specific T cell response was potentially migrating towards the load positively correlated with the overall capacity of CD4+T cells infected tissues by assessing the proportion of CCR7 and CXCR3 to secrete IL-4 in response to TCR independent unspecific expression within the same analyses. In peripheral blood, CCR7 activation with PMA/ionomycin (PMA/Io) (Supplementary distinguishes T cells homing to lymph node (LN) when expres- Fig. 3c). In contrast, the expression of IL-10, a prototypical sed, or effector memory (EM) T cell subsets migrating to tissues regulatory cytokine, was significantly increased in CD4+T cells when absent41, while CXCR3 may help define antiviral T cells from nonhospitalized patients after stimulation with M peptides infiltrating inflamed tissues, including the lung parenchyma25. when compared to the mild COVID-19 group (p = 0.035; CD4+T cells showed only two evident subsets in most patients Fig. 1a). based on CCR7 expression, since CXCR3 was homogeneously Correlations between the net frequency of a given function and dimly expressed in these two subsets (Supplementary Fig. 2a clinical parameters were consistent with more CD4+T cells and 2c) and no differences between the different study groups secreting IFNγ and more CD8+T cells secreting IL-4 in response were observed (Fig. 2a). In contrast, CD8+T cells presented five to M and S peptides associated with disease severity (Supple- subsets based on these chemokine receptors (Supplementary mentary Table 2). Even the total CD4+ or CD8+T cell IFNγ Fig. 2a and 2c), and significant differences among the groups were response and the total IL-4 secretion by CD8+T cells against any observed (Fig. 2b). Non-hospitalized patients showed increased of the three viral proteins (all peptides) correlated with more days frequencies of CCR7hCXCR3dCD8+T cells, while severe patients at the hospital for IFNγ or with other clinical parameters for IL-4 presented increased frequencies of EM CCR7-CXCR3+T cells (Supplementary Table 2). Further, antigen-specific CD4+T cells (p = 0.0012 and p = 0.0034, respectively, by Kruskal–Wallis one- degranulating in response to all viral peptides and, in the case of way ANOVA, Fig. 2b, c). In fact, the accumulation of CD8+T cells, in response to M peptides also correlated with CCR7hCXCR3dCD8+T cells indicated good prognosis and higher levels of IL-6 (Supplementary Table 2). In contrast, the negatively correlated with the number of days to discharge since percentage of M-specific CD4+T cells secreting IL-10 correlated symptoms onset and with IL-6 levels at hospital entry (Fig. 2d), 4 NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 ARTICLE Fig. 2 T cell migratory patterns during acute SARS-CoV-2 infection. a, b The frequency of various T cell subsets defined by CCR7 and CXCR3 within CD4+ (a) and CD8+T cells (b). Each dot represents one patient of a specific cohort, indicated by color code (normal donors n = 12; nonhospitalized n = 14; mild n = 20 and severe n = 12). Data are shown as individual patients and boxes and error bars represent median and interquartile range (IQR). Statistical comparisons were performed using Kruskal–Wallis rank-sum test with Dunn’s multiple comparison test (two-sided): CCR7hCXCR3d (p = 0.029 and p = 0.0007) and EM CXCR3+ (p = 0.002). c Representative flow cytometry plots gating the different CD8+T cell subsets in a nonhospitalized (top) and a severe patient (bottom). d, e Correlations between days to discharge since symptoms onset or interleukin (IL)-6 baseline levels and the frequency of CD8+ CCR7hCXCR3d (d) and CD8+ EM CXCR3+ (e) subpopulations. Two-sided spearman rank correlation (n = 46). while the frequency of EM CXCR3+ CD8+T cells significantly nonhospitalized group (Fig. 3a and Supplementary Fig. 3d). Of correlated with disease severity parameters (Fig. 2e), results that note, in general, most individuals in the hospitalized groups also have also been observed in SARS-CoV-2-specific T cells42. showed this trend for the IL-10 response, although statistical We then focused on the distribution of the net antigen- significance was not reached as a group. Moreover, several subsets response for each peptide and cohort among these CCR7/ out of these antigen-specific CD4+T cells, mostly the ones CXCR3 subsets. Overall, antigen-specific CD4+T cells showed a secreting IFNγ or IL-4, correlated with worse prognosis in the distinct pattern based on the function assessed: IFNγ, degranula- clinical parameters assessed before, and some examples are shown tion (CD107a) and IL-4 were significantly associated to EM in Fig. 3b–3e. In general, stronger associations were observed CXCR3+ CD4+T cells across the different groups and proteins, within the CCR7+CXCR3+ subset, which correlated with while IL-10 was associated to the LN-homing fraction (CCR7 severity, except if this subset was secreting IL-10 against M +CXCR3+) in response to M and N protein peptides in the peptides (Fig. 3f). Moreover, SARS-CoV-2 viral load was NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 a IFNγ CD107a IL-4 IL-10 M N 2.5 * * 1.0 * 2.0 ** 2.5 S ** *** 2.0 0.8 1.5 2.0 %of CD107a+ %of IFNγ + %of IL-4+ %of IL-10+ 1.5 0.6 1.5 1.0 1.0 0.4 1.0 NON- 0.5 0.2 0.5 HOSPITALIZED 0.5 0.0 0.0 0.0 0.0 7+ 7- 7+ 7- 7+ 7- + + + + + + + + + + + + 3 3 3 3 3 3 + + + + + + R R R R R R R R R R R R 3 3 3 3 3 3 C C C 3 3 3 3 3 3 C C C XC XC XC XC XC XC R R R R R R C C C R R R R R R C C C 3+ 3+ 3+ 3+ 3+ 3+ XC XC XC XC XC XC XC XC XC XC XC XC C C C + C- + C- + C- R R R C C C R R R 7 7 7 7 7 7 + C- + C- + C- C C C XC XC XC + R R R C- + + C- C- XC XC XC R R R 7 7 7 7 7 7 7 7 7 C C C 7 7 7 C C C R R R R R R R R R C C C C C C R R R C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C 1.0 15 2.5 2.5 ** ** ** %of CD107a+ * ** ** ** 2.0 0.8 2.0 %of IL-10+ %of IFNγ + %of IL-4+ 10 1.5 0.6 1.5 1.0 0.4 1.0 5 0.5 0.2 0.5 MILD 0.0 0.0 0 0.0 + + + + + + 7+ 7- 7+ 7- 7+ 7- + + + + + + + + + + + + 3 3 3 3 3 3 3 3 3 3 3 3 R R R 3 3 3 3 3 3 R R R R R R R R R R R R R R R R R R R R R C C C C C C XC XC XC XC XC XC XC XC XC XC XC XC XC XC XC XC XC XC C C C C C C 3+ 3+ 3+ 3+ 3+ 3+ C C C C C C + - + - + - C C C C C C C C + C- + C- + - + C- + - + C- R R R 7 7 7 R R R 7 7 7 7 7 7 7 7 7 R R R XC XC XC 7 7 7 7 7 7 R R R XC XC XC R R R R R R R R R C C C R R R C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C 1.0 5 * 1.5 2.5 ** * * 4 2.0 0.8 %of CD107a+ %of IL-10+ 1.0 %of IL-4+ %of IFNγ + 0.6 3 1.5 2 SEVERE 1.0 0.4 0.5 0.5 0.2 1 0.0 0.0 0 0.0 + + + + + + + + + + + + + + + + + + 7+ 7- 7+ 7+ 7- 7- 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 R R R R R R R R R R R R R R R R R R R R R R R R XC XC XC XC XC XC C C C XC XC XC XC XC XC XC XC XC XC XC XC C C C C C C C C C 3+ 3+ 3+ 3+ 3+ 3+ C C C C C C C C C C C C C C C + - + - + - + - + - + - + C- + C- + C- 7 7 7 R R R 7 7 7 7 7 7 7 7 7 R R R 7 7 7 7 7 7 R R R R R R R R R XC XC XC R R R R R R R R R XC XC XC C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C C b c d 0.10 r = 0.56 0.20 r = 0.36 0.8 r = 0.34 % of S-sp CD4+ CCR7+ p = 0.0001 p = 0.014 p = 0.022 + + % of N-sp CD4 CCR7 % of S-sp CD4 CCR7 0.08 CXCR3+ CD107a+ 0.15 0.6 CXCR3+ IFNγ + CXCR3+ IL-4+ + + 0.06 0.10 0.4 0.04 0.05 0.2 0.02 0.00 0.00 0.0 0 20 40 60 80 0 50 100 150 200 1500 0 50 100 150 200 1500 Days to discharge since symptoms onset IL-6 (pg/mL) IL-6 (pg/mL) e f g 2.5 2.5 2.0 r = 0.44 r = -0.34 r = -0.40 % of M-sp CD4+ CCR7+ % of CD4+ EM CXCR3+ - p = 0.002 p = 0.022 p = 0.020 % of M-sp CD4 CCR7 2.0 2.0 1.5 IL-10+ (PMA/Io) CXCR3+ IL-10+ CXCR3+ IFNγ + + 1.5 1.5 1.0 1.0 1.0 0.5 0.5 0.5 0.0 0.0 0.0 0 50 100 150 200 1500 0 50 100 150 200 1500 0 200 400 600 IL-6 (pg/mL) IL-6 (pg/mL) Viral load (relative units) negatively associated with the overall capacity of EM CXCR3+ secreting CCR7hCXCR3dCD8+T cells in response to N peptides CD4+T cells to secrete IL-10 in response to TCR independent correlated negatively with days to hospital discharge (Supple- unspecific activation with PMA/Io (Fig. 3g). mentary Fig. 4b). Degranulation, which was enhanced also after In the analyses of the proportion of antigen-specific N stimulation, was unexpectedly detected in all CD8+ CCR7/ CD8+T cells in each CCR7/CXCR3 subset, we did not consider CXCR3 subpopulations (Supplementary Fig. 5a). Though only in the EM CXCR3- subset, which represented <1% in most patients the LN-homing CCR7+CXCR3+CD8+T cell subset in response (Fig. 2b). As expected, IFNγ antigen-specific CD8+T cells were to S peptides, degranulation was associated with higher viral load more frequent among the CXCR3+ subsets, with some individual (Supplementary Fig. 5b). With respect to IL-4 secreting antigen- exception, such as N-specific CCR7+CXCR3- T cells within the specific CD8+T cells, in general these responses were more nonhospitalized group (Supplementary Fig. 4a). Further, IFNγ frequent in CCR7+ LN-homing subsets, and as mentioned before, 6 NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 ARTICLE Fig. 3 Migratory patterns of acute SARS-CoV-2-specific CD4+T cells. a Net frequency of interferon (IFN)γ, CD107a, interleukin (IL)-4 and IL-10 expression in SARS-CoV-2-specific CD4+ T cells based on CXCR3+CCR7+ and CXCR3+CCR7- subsets for each individual patient (nonhospitalized n = 14; mild n = 20 and severe n = 12). Viral proteins are shown in color green (membrane protein, M), orange (nucleocapsid protein, N) and purple (spike protein, S). Dots connected by the same line represent the same individual. Statistical comparisons were performed using two-sided nonparametric Wilcoxon matched-pairs signed rank test to compare the two groups (CXCR3 + CCR7+ vs. CXCR3+CCR7−): nonhospitalized (IFNγ, p = 0.016 and p = 0.016; CD107a, p = 0.031; IL-4, p = 0.008; IL-10, p = 0.008, and p = 0.0005), mild (IFNγ, p = 0.006, p = 0.005, and p = 0.006; CD107a, p = 0.035; IL-4, p = 0.004, p = 0.008, and p = 0.002), and severe (IFNγ, p = 0.005, p = 0.024, and p = 0.039; IL-4, p = 0.042). b–d Correlation between the days to discharge since symptoms onset or IL-6 and the frequency of nucleocapsid or spike-specific CD4+ CXCR3+CCR7+ expressing IFNγ (b), CD107a (c), or IL- 4 (d). e, f Correlation between IL-6 and the frequency of membrane-specific CD4+ CXCR3+CCR7−/+ expressing IFNγ (e) or IL-10 (f). g Correlation between the viral load and the frequency of CD4+ EM CXCR3+ expressing IL-10 after PMA/Ionomycin stimulation. Two-sided spearman rank correlation (n = 46 for all correlations except for viral load (g), which corresponds to n = 33). they increased with disease severity (Supplementary Fig. 6a). Regarding antigen-specific T cells we detected remarkable Consequently, the frequency of IL-4 detected in response to M or differences within the IL-10 secreting T cells. In this sense, S peptides in several of these fractions correlated with disease increased expression of caspase-3 was distinguished in the severity (Supplementary Fig. 6b). Remarkably, SARS-CoV-2- hospitalized severe group compared to the nonhospitalized in specific CD8+T cells secreting IL-10 were strongly represented S-specific IL-10+ CD4+T cells (p = 0.003), N-specific IL-10+ among the CCR7hCXCR3d subset, reaching statistical signifi- CD8+T cells (p = 0.013) and even in the overall IL-10 secretion cance in response to any of the viral proteins within the mild capacity in response to PMA/Io response (p = 0.036 for CD4+ disease cohort and in response to N peptides in nonhospitalized and p = 0.004 for CD8+, by Kruskal–Wallis one-way ANOVA patients, but not in the hospitalized group with severe disease with Dunn’s multiple comparisons; Fig. 5e). Further, significant (Fig. 4a). Finally, we detected two additional correlations within correlations between apoptosis in IL-10 antigen-specific T cells the CD8+T cell compartment that were of interest: overall and several clinical parameters supported these results (Fig. 5f). antigen-specific EM CXCR3+ CD8+T cells correlated with As for the other functions, we only detected positive correlations higher viral loads if responding to M peptides (Fig. 4b), while for the expression of caspase-3 in baseline and N-specific CD107a the same subset responding to N peptides negatively correlated + CD4+T cells with viral load and IL-6 levels, respectively with IL-6 (Fig. 4c). All together, these results demonstrate (Fig. 5g). These results indicate increased activation-induced cell individual migratory patterns associated with a given function: death associated with viral replication and disease severity whilst most of the functions assessed here were associated with affecting total CD4+ and CD8+T cells, a phenomenon that lung homing subsets (CCR7-), IL-10-specific T cells expressed seems to be modulated by the viral protein targeted. Moreover, high levels of CCR7. In fact, a strong association towards better CD8+CCR7hCXCR3d T cells, a major producers of IL-10, disease prognosis was established for an increased proportion of appeared to be one of the most affected population. CCR7hCXCR3d CD8+T cells, which represented a major constant source of IL-10 in CD8+T cells. Further, antigenic stimulation could be driving CCR7- effector immune responses Ag-specific TRM responses are present in the lungs of con- towards the lung, yet under uncontrolled disease progression, valescent patients. In order to demonstrate that antigen-specific such effector functions seemed to increase in LN-homing CCR7+ T cells detected during acute SARS-CoV-2 infection, not only subsets. migrate into the lung parenchyma but also persist as TRM, we measured their frequency in lung tissue of seven patients (Sup- plementary Table 3). These patients, who strongly differed in Apoptosis is enhanced in T cells during severe infection. We their SARS-CoV-2 infection profile, successfully recovered and included caspase-3 in the flow cytometry panel as a surrogate SARS-CoV-2 was not detected in the respiratory tract by RT-PCR marker of apoptotic cell death activation43, which was quanti- before they underwent thoracic surgery for different reasons. fied in both antigen-specific and bystander T cells from the Briefly, HL24 patient was a young asymptomatic patient whose different subsets of the study groups (Fig. 1a). Overall expres- blood and lung samples were analyzed 21 days after SARS-CoV-2 sion of caspase-3 in response to stimulation was increased in laboratory confirmation by RT-PCR. In contrast, samples were total CD4+T cells of the severe group after M and PMA/Io analyzed between 6 and 10 months after initial SARS-CoV-2 RT- stimulation and in CD8+T cells after S stimulation in com- PCR confirmation for three of the mild cases (HL52, HL65, parison to the nonhospitalized group (Fig. 5a). Moreover, HL75) and the two severe cases (HL27 and HL69). HL81, another caspase-3 expression in CD4+ and CD8+T cells after stimula- mild case, was analyzed close to 4 months after initial infection. tion with S-peptides positively correlated with baseline IL-6 and Of note, a more detailed COVID-19 clinical history can be found with the number of hospitalization days for CD8+T cells in the methods section. (Fig. 5b). In addition, the overall frequency of CD4+T cells Antigen-specific T cell responses were analyzed in total lung expressing caspase-3 in response to PMA/Io positively corre- CD4+ and CD8+T cells and by three different fractions: CD69- lated with viral load (Fig. 5b). Further, an increased expression (non-TRM), CD69+ (TRM) and a subset within CD69+ cells of caspase-3 within the CCR7hCXCR3d subset was, in general, expressing CD103+ (TRM CD103+) (Supplementary Fig. 7A). Of linked to the disease severity (Fig. 5c), reaching statistical sig- note, CD69+T cells down-regulated T-bet expression, which has nificance after PMA/Io stimulation when comparing the severe been associated to tissue residency44. As shown for one of the and the nonhospitalized groups (Fig. 5c). Consequently, those convalescent patients with previous severe disease (HL27; Fig. 6a), frequencies in response to stimulation (N, S, or PMA/Io) cor- antigen-specific T cells secreting IFNγ were restricted to the TRM related positively with the number of days at the hospital and fractions, which in the case of S-specific CD4+T cells represented with baseline IL-6 (Fig. 5d). Expression of caspase-3 in other up to 3.47% of the TRM CD103+ subset. Similarly, in the lung of a CCR7+CD8+T cell subsets in response to N and S peptides also mild convalescent patient (HL52; Fig. 6b) N-specific CD8+T cells correlated with IL-6 baseline levels in patients (Fig. 5d). secreting IFNγ were restricted to the TRM fractions and >40% of NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 a M N S 25 25 25 * 15 15 15 %of IL-10+ %of IL-10+ %of IL-10+ 5 5 5 4 4 4 NON-HOSPITALIZED 3 3 3 2 2 2 1 1 1 0 0 0 - + + d - + + d - + + d 3 3 3 3 3 3 3 3 3 3 3 3 R R R R R R R R R R R R XC XC XC XC XC XC XC XC XC XC XC XC C C C C C C + C C C C C + C + + + h + h h EM EM 7 EM 7 7 7 7 7 7 R 7 7 R R R R R R R R C C C C C C C C C C C C C C C C C C 25 25 25 ** ** * 15 15 15 %of IL-10+ %of IL-10+ %of IL-10+ 5 5 5 MILD 4 4 4 3 3 3 2 2 2 1 1 1 0 0 0 - + + d - + + - + + d d 3 3 3 3 3 3 3 3 3 3 3 3 R R R R R R R R R R R R XC XC XC XC XC XC XC XC XC XC XC XC C C C C C C C C C C C C + + + + + + h h h EM EM EM 7 7 7 7 7 7 7 7 7 R R R R R R R R R C C C C C C C C C C C C C C C C C C 35 25 25 25 15 15 15 %of IL-10+ %of IL-10+ %of IL-10+ 5 5 5 SEVERE 4 4 4 3 3 3 2 2 2 1 1 1 0 0 0 - + + d - + + - + + d d 3 3 3 3 3 3 3 3 3 3 3 3 R R R R R R R R R R R R XC XC XC XC XC XC XC XC XC XC XC XC C C C C C C C C C C C C + + + + + + h h h EM EM EM 7 7 7 7 7 7 7 7 7 R R R R R R R R R C C C C C C C C C C C C C C C C C C b c 15 r = 0.41 4 r = - 0.39 %of N-sp CD8+ EM CXCR3+ + p = 0.016 p = 0.030 % of M-sp CD8 EM CXCR3 3 (all functions) (all functions) 10 + 2 5 1 0 0 0 200 400 600 0 50 100 150 200 1000 1500 Viral load (relative units) IL6 (pg/mL) Fig. 4 IL-10 expression in acute SARS-CoV-2-specific CD8+T cell subsets. a Net frequency of interleukin (IL)-10 expression in CCR7+CXCR3−, CCR7+ CXCR3+, CCR7hCXCR3d, and EM CXCR3+ subsets within CD8+ T cells after stimulation with any of the three viral proteins (membrane (M), nucleocapsid (N), and spike (S) proteins). Data are shown as median and upper range, where each dot represents an individual patient for each group (nonhospitalized n = 14; mild n = 20, and severe n = 12). Statistical comparisons were performed using Kruskal–Wallis rank-sum test with Dunn’s multiple comparison test (two-sided): nonhospitalized (N, p = 0.041) and mild (M, p = 0.003; N, p = 0.007 and S, p = 0.042). b, c Correlation between CD8+ EM CXCR3+T cells responding with any function (added net response for interferon (IFN)γ, CD107a, IL-4 and IL-10) against M peptides and viral load (b) and against N peptides and baseline IL-6 levels (c). Two-tailed spearman rank correlation (n = 33 for viral load and n = 46 for IL-6). 8 NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 ARTICLE a %of Caspase-3 after PMA/Ionomycin %of Caspase-3 after N stimulation %of Caspase-3 after S stimulation %of Caspase-3 after M stimulation 60 **** 15 15 30 * NON-HOSPITALIZED MILD ** SEVERE 40 10 10 20 20 5 5 10 0 0 0 0 4+ 8+ 4+ 8+ 4+ 8+ 4+ 8+ D D D D D D D D C C C C C C C C b c 100 100 100 r = 0.43 3 8 r = 0.36 r = 0.42 r = 0.48 60 60 60 % of caspase-3 in S-sp % of caspase-3 in S-sp % of caspase-3 in S-sp CD8+ CCR7h CXCR3d p = 0.014 p = 0.004 p = 0.003 p = 0.005 * 20 %of Caspase-3 in 20 20 % of caspase-3 in 10 10 10 6 CD4+ (PMA/Io) 2 8 8 8 CD4+ CD8+ CD8+ 4 6 6 6 1 4 4 4 2 2 2 2 0 0 0 0 0 0 50 100 150 200 1500 0 50 100 150 200 1500 0 20 40 60 80 0 200 400 600 pe al s /I o s es de de as id A IL-6 (pg/mL) Viral load (relative units) i i PM B pt pt pt IL-6 (pg/mL) Days to discharge pe pe since symptoms onset N M S d CD8+ CCR7h CXCR3d (PMA/Io) 5 r = 0.33 5 r = 0.39 5 r = 0.30 8 r = 0.47 100 r = 0.35 p = 0.024 p = 0.008 p = 0.045 p = 0.019 % of caspase-3 in N-sp p = 0.0009 % of caspase-3 in S-sp % of caspase-3 in S-sp % of caspase-3 in N-sp CD8+ CCR7+ CXCR3 - CD8+ CCR7h CXCR3d CD8+ CCR7h CXCR3d CD8+ CCR7h CXCR3d 4 4 80 % of caspase-3 in 4 6 3 3 3 60 4 2 2 2 40 2 1 1 1 20 0 0 0 0 0 0 20 40 60 80 0 50 100 150 200 1500 0 20 40 60 80 0 50 100 150 200 1500 0 50 100 150 200 1500 Days to discharge IL-6 (pg/mL) Days to discharge IL-6 (pg/mL) IL-6 (pg/mL) since symptoms onset since symptoms onset e %of Caspase-3 in PMA/Ionomycin 100 100 * r = 0.35 r = 0.30 %of Caspase-3 in N-specific %of Caspase-3 in S-specific 60 100 100 100 % of caspase-3 in S-sp p = 0.018 % of caspase-3 in S-sp p = 0.048 CD8+ CCR7+ CXCR3+ CD8+ CCR7+ CXCR3- 20 80 10 80 80 80 8 60 60 ** 60 60 * ** 6 40 4 40 40 40 2 20 20 20 20 0 0 50 100 150 200 1500 0 0 50 100 150 200 1500 0 0 0 IL-6 (pg/mL) 0+ 0+ 0+ 0+ 0+ 0+ IL-6 (pg/mL) -1 -1 -1 -1 -1 -1 IL IL IL IL IL IL 4+ 8+ 4+ 8+ 4+ 8+ f D D D D D D C C C C C C 100 100 40 r = 0.48 150 r = 0.33 150 r = 0.37 r = 0.43 r = 0.31 60 p = 0.0007 % of caspase-3 in N-sp p = 0.024 % of caspase-3 in S-sp % of caspase-3 in S-sp % of caspase-3 in N-sp % of caspase-3 in S-sp p = 0.011 p = 0.003 p = 0.036 80 20 10 30 CD8+ IL-10+ CD4+ IL-10+ CD4+ IL-10+ CD8+ IL-10+ CD8+ IL-10+ 100 100 60 8 20 6 40 4 50 50 10 20 2 0 0 0 0 50 100 150 200 1500 0 0 0 50 100 150 200 1500 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 IL-6 (pg/mL) IL-6 (pg/mL) Days to discharge Days to discharge Days to discharge since symptoms onset since symptoms onset g since symptoms onset 60 r = 0.42 150 100 r = 0.35 r = 0.40 p = 0.003 in basal CD4+ CD107a+ p = 0.044 % of caspase-3 in N-sp p = 0.006 CD8+ IL-10+ PMA/Io) % of caspase-3 in 80 CD4+ CD107a+ % caspase-3 40 100 60 20 50 40 20 0 0 0 20 40 60 80 0 200 400 600 0 Days to discharge 0 50 100 150 200 1500 Viral load (relative units) since symptoms onset IL-6 (pg/mL) them were simultaneously degranulating (CD107a+). In fact, the Friedman test; Fig. 6d). Regardless of the variability observed assessment of the frequency of cells responding to each peptide among study patients, the frequency of specific T cell responses pool demonstrated a general trend for more antigen-specific increased with disease severity (from left to right; Fig. 6c), while T cells belonging to the TRM subsets for all functions and all their magnitude was in general low and less consistent for IL-4 or patients (Fig. 6c and Supplementary Fig. 7b), which was IL-10 responses (Fig. 6c and Supplementary Fig. 7b). Importantly, significant for total CD3+ T cells secreting IFNγ (p = 0.016 by a consistent polyfunctional IFNγ+CD107a+ T cell response, NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 Fig. 5 Caspase-3 expression in T cells during acute SARS-CoV-2 infection. a Frequency of caspase-3 expression in CD4+ and CD8+T cells after stimulation with membrane (M), nucleocapsid (N) or spike protein (S) and PMA/Ionomycin, in nonhospitalized (orange, n = 14), mild (blue, n = 20), and severe (pink, n = 12) COVID-19 patients. Data are shown as individual patients and boxes and error bars represent median and interquartile range (IQR). Statistical comparisons were performed using Kruskal–Wallis rank-sum test with Dunn’s multiple comparison test (two-sided): CD4+ (M, p < 0.0001 and p < 0.0001; PMA/Io p = 0.036) and CD8+ (S, p = 0.007 and p = 0.002). b Correlation between days to hospital discharge since symptoms onset, viral load or baseline interleukin (IL)-6 levels (pg/mL) and the net frequency (background subtracted) of caspase-3 in CD4+ and CD8+T cells after stimulation with the spike protein or PMA/Ionomycin. c Frequency of caspase-3 expression in CD8+ CCR7hCXCR3dT cells after stimulation in the same groups as in a. Statistical comparisons were performed using Kruskal–Wallis rank-sum test with Dunn’s multiple comparison test (two-sided): p = 0.026. d Correlations between clinical parameters and the net frequency of caspase-3 expression in CD8+ CCR7+ T cell subsets after stimulation. e Frequency of caspase-3 expression in IL-10-secreting SARS-CoV-2-specific CD4+ and CD8+T cells responding to the spike protein, the nucleocapsid protein or to PMA/Ionomycin in the same groups as in a. Statistical comparisons were performed using Kruskal–Wallis rank-sum test with Dunn’s multiple comparison test (two-sided): CD4+ (S, p = 0.003; PMA/Io, p = 0.036) and CD8+ (N, p = 0.013; PMA/Io, p = 0.004). f Correlation between clinical parameters and IL-10-expressing SARS-CoV-2-specific CD4+ or CD8+T cells, or after PMA/Ionomycin stimulation. g Correlation between viral load and the frequency of caspase-3 expression in basal CD107a+ degranulating CD4+T cells and between IL-6 and the net frequency of caspase-3 expression in CD107a-expressing CD4+ in response to N peptides. Two-tailed spearman rank correlation (n = 46 for all correlations except for viral load, which corresponds to n = 33). which represented between a median of 0.022 and 0.051% of all all, we show that the acute response of nonhospitalized infected CD3+T cells and was mostly associated with the TRM fraction patients is characterized by CD4+ and, to less extent, CD8+ (>75%), was detected against N peptides in all patients except in SARS-CoV-2-specific T cells secreting IL-10, to which subsets the asymptomatic (HL24) and the oldest patient (HL75) (Fig. 6e). expressing high levels of CCR7 contribute abundantly. In con- Of note, the individual patient with a frequent polyfunctional trast, hospitalized patients show a bias towards an effector response in blood was HL65, who tested SARS-CoV-2 positive for response characterized by IFNγ and IL-4 secretion, being the 4 months, main functions as severity increases. Second, depending on the The comparison between the overall SARS-CoV-2-specific T SARS-CoV-2 viral protein targeted, different CD4+ and CD8+T cell responses detected in lungs with the ones found in cell functional profiles are generated, which have clear implica- contemporaneous peripheral blood samples highlighted strong tions for vaccine design. Third, lymphopenia is partially a con- differences between these two compartments (Fig. 7). For sequence of increased apoptosis in antigen-specific and example, in the asymptomatic HL24-patient IFNγ and IL-10 nonspecific T cells, which is associated with disease severity and responses were more frequent in blood than in lung, IL-4 was where SARS-CoV-2-specific subsets, such as IL-10 secreting absent, and T cell degranulation was only detectable in lung T cells, appear to be more susceptible. Finally, and most impor- (Fig. 7). Importantly, this patient was closer to the initial RT- tant, SARS-CoV-2 TRM can be established and persist for PCR-based laboratory confirmation (3 weeks after), and his 10 months after infection; nonetheless, the magnitude and profile profile was more consistent with a nonhospitalized patient. A of the lung SARS-CoV-2-specific T cells strongly differ from the different pattern was observed for one of the severe convalescent response detected in blood. patients (HL27), who had persistent TRM in the lung over- Major efforts have recently centered on the identification and represented by S-specific CD4+T cells secreting IFNγ, which characterization of SARS-CoV-2-specific T cells4,8–14. Several of represented up to 1.58% of the total CD4+T cells, while only these studies have focused on defining the viral proteins more 0.082% of the circulating CD4+T cells secreted IFNγ and, in this often targeted by specific T cells, concluding that after infection a case, in response to M peptides (Fig. 7). In fact, whereas all four broad cell response against multiple structural and non-structural functions were detected in T cells from lung after 6 months since regions of SARS-CoV-2 is detected in most convalescent initial infection for this patient, they were barely detectable in patients4,9,12. More recently, it has been reported that SARS- blood. Last, may be worth mentioning that the highest levels of CoV-2-specific T cells appear to be weaker and less frequent IL-4 secretion were detected in response to S peptides in the lung during acute infection13. In this sense, in our study, while the of HL65, who tested SARS-CoV-2 positive for 4 months (Fig. 7). frequency of responders based on CD4+T cells specifically Overall, viral-specific TRM responses were detected in all patients. secreting IFNγ was similar to previous reports, they were indeed However, no consistent patterns were observed among patients in weak in terms of the amount of IFNγ or cytotoxicity. However, terms of viral proteins targeted and functions between blood and SARS-CoV-2-specific T cell response during acute infection lung compartments, except for the polyfunctional response appeared to be dominated by IL-4 secretion in hospitalized detected in tissue against N peptides. Of note, the asymptomatic patients and by IL-10 in nonhospitalized patients. A recent report patient had no detectable antibodies, while all the other patients highlights that, in contrast to other cytokines, increased levels of had detectable antibodies against SARS-CoV-2 in a concomitant SARS-CoV-2-specific CD4+T cells secreting IL-10 were mainly plasma sample (measured as total and as IgG fraction). detected during active disease45. However, these functions have Furthermore, no virus was detected in any of the lung samples rarely been assessed as part of the specific intracellular T cell by immunofluorescence or viral RNA in situ hybridization. response, and others have measured IL-4 in the supernatant of Overall, our results highlight the establishment and persistence of stimulated PBMC without detecting any increase9,13,14. In con- lung-resident T cell immunity against SARS-CoV-2 viral infec- trast, higher serum levels of IL-4, IFNγ and IL-10 cytokines, tion. However, in most cases, antigen-specific TRM patterns could among others, have been associated with COVID-1929,36,37,39. not be identified in T cells from blood. Differences in methodology and sample timing may account for these discrepancies. In fact, we did not detect changes in plasma Discussion levels of IL-4, highlighting differences between measuring the This study identifies unique features of the cellular immunolo- overall level of a given cytokine released systemically rather than gical response against SARS-CoV-2 relevant to infection control the capacity of a small frequency of antigen-specific T cells to and disease progression, which may be critical to informing quickly secrete such cytokine. Importantly, several studies have vaccine assessment and development of new prototypes. First of detected a negative impact of IL-4 mediated responses on 10 NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 ARTICLE a b non-TRM TRM TRM CD103 + non-TRM TRM TRM CD103 + 10 5 1.14 10 5 1.74 10 5 1.59 10 5 0.47 10 5 1.09 10 5 0.97 4 4 4 4 4 4 10 10 10 10 10 10 3 3 3 Ctrl 3 3 3 Ctrl 10 10 10 10 10 10 0 0 0 0 0 0 3 0.026 3 0.027 3 0 3 0 3 0 3 0 CD107a PE-Cy5 -10 -10 -10 -10 -10 -10 CD107a PE-Cy5 CD4+ 3 3 4 5 3 3 4 5 3 3 4 5 3 3 4 5 3 3 4 5 3 3 4 5 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 10 5 1.25 10 5 2.30 10 5 2.97 10 5 0.36 10 5 1.17 10 5 1.07 CD8+ 4 4 4 4 4 4 10 10 10 10 10 10 S N 3 3 3 3 3 3 10 10 10 10 10 10 0 0 0 0 0 0 -10 3 0.057 -10 3 1.87 3 3.47 3 0 3 0.13 3 0.044 -10 -10 -10 -10 3 3 4 5 3 3 4 5 3 3 4 5 3 3 4 5 3 3 4 5 3 3 4 5 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 IFN Alexa-700 IFN Alexa-700 c CD4⁺ T cells CD8⁺ T cells M N S M N S IFNγ ⁺ CD107a⁺ Non-TRM IL-4⁺ IL-10⁺ 0 IFNγ ⁺ >0 - ≤0,02 CD107a⁺ >0,02 - ≤0,05 TRM IL-4⁺ >0,05 - ≤0,1 IL-10⁺ >0,1 - ≤0,5 IFNγ ⁺ >0,5 - ≤1,5 CD107a⁺ >1,5 CD103+ TRM IL-4⁺ IL-10⁺ HLTE 24 81 52 75 65 69 27 24 81 52 75 65 69 27 24 81 52 75 65 69 27 24 81 52 75 65 69 27 24 81 52 75 65 69 27 24 81 52 75 65 69 27 d 12 6 0.25 0.20 10 8 0.20 % of CD107a+ 0.15 % of IL-10+ % of IFNγ + % of IL-4+ 6 * 4 0.15 0.10 4 0.10 2 2 0.05 0.05 0 0 0.00 0.00 3+ + + + M M M M M M M M 3 3 3 R TR R R R R R R n- T T 10 n- T T 10 n- T T 10 10 T n- CD CD CD D no no no no C M M M M TR R R R T T T e Lung Blood 0.3 M % of IFNγ +CD107a+ 0.2 N S 0.1 0.04 0.02 0.00 Fig. 6 Functional analysis of lung-resident SARS-CoV-2-specific T cells. a, b Flow cytometry plots showing the frequency of interferon (IFN)γ and degranulation (CD107a) by non-resident memory T cells (TRM), TRM or CD103+ TRM in CD4+ from HL27 after spike stimulation and control (a) and in CD8+ from HL52 after nucleocapsid stimulation and control (b). c Heatmaps summarizing the net frequencies of IFNγ, CD107a, interleukin (IL)-4, and IL- 10 SARS-CoV-2-specific CD4+ or CD8+ non-TRM, TRM, and TRM CD103+ from seven different SARS-CoV-2 recovered patients. Cytokine production or degranulation are displayed as colors ranging from yellow to blue, based on the frequency, as shown in the key. d Net frequency of SARS-CoV-2-specific CD3+T cells in response to all viral proteins (membrane (M), nucleocapsid (N) and spike (S)) from from lung by non-TRM, TRM or CD103+ TRM are shown as mean ± SEM (n = 7). Statistical comparisons were performed using Kruskal–Wallis rank-sum test with Dunn’s multiple comparison test (two-sided): p = 0.023. e Net frequency of double positive IFNγ/CD107a CD3+T cells from lung or blood after stimulation with membrane (M; green), nucleocapsid (N; orange), or spike protein (S; purple) (n = 7). NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 M Lung Blood Lung Blood Lung Blood Lung Blood N 0.10 0.10 0.10 0.10 S 0.08 0.08 0.08 0.08 % of CD107a+ % of IL-10+ % of IFNγ + % of IL-4+ 0.06 0.06 0.06 0.06 HL24 0.04 0.04 0.04 0.04 (asymptomatic, 0.7 mo) 0.02 0.02 0.02 0.02 0.00 0.00 0.00 0.00 0.30 0.4 0.10 0.10 0.25 0.08 0.08 % of CD107a+ 0.3 % of IL-10+ % of IFNγ + % of IL-4+ 0.20 0.06 0.06 0.15 0.2 HL81 0.04 0.04 (mild, 3.7 mo) 0.10 0.1 0.05 0.02 0.02 0.00 0.0 0.00 0.00 0.30 0.10 0.10 0.10 0.25 0.08 0.08 0.08 % of CD107a+ % of IL-10+ % of IFNγ + % of IL-4+ 0.20 0.06 0.06 0.06 0.15 HL52 0.04 0.04 0.04 (mild, 7.5 mo) 0.10 0.05 0.02 0.02 0.02 0.00 0.00 0.00 0.00 0.30 0.4 0.8 0.4 0.25 % of CD107a+ 0.3 0.6 0.3 % of IL-10+ % of IFNγ + % of IL-4+ 0.20 HL75 0.15 0.2 0.4 0.2 (mild, 10.5 mo) 0.10 0.1 0.2 0.1 0.05 0.00 0.0 0.0 0.0 0.30 0.4 0.16 0.10 0.25 0.08 % of CD107a+ 0.3 0.12 % of IL-10+ % of IFNγ + % of IL-4+ 0.20 0.06 0.15 0.2 0.08 0.04 HL65 0.10 (mild, 6 mo) 0.1 0.04 0.05 0.02 0.00 0.0 0.00 0.00 0.30 0.15 0.10 0.4 0.25 0.08 % of CD107a+ 0.3 % of IL-10+ % of IFNγ + % of IL-4+ 0.20 0.10 0.06 0.15 0.2 HL69 0.04 (severe, 10 mo) 0.10 0.05 0.02 0.1 0.05 0.00 0.00 0.00 0.0 2 1.0 0.10 0.10 0.8 0.08 0.08 % of CD107a+ % of IL-10+ % of IFNγ + % of IL-4+ 1 0.6 0.06 0.06 0.6 HL27 0.4 0.04 0.04 0.4 (severe, 10 mo) 0.2 0.02 0.02 0.2 0.0 0.0 0.00 0.00 CD4+CD8+ CD4+ CD8+ CD4+ CD8+ CD4+ CD8+ CD4+ CD8+ CD4+ CD8+ CD4+ CD8+ CD4+ CD8+ Fig. 7 SARS-CoV-2-specific T cells in lungs and blood of convalescent patients. Total CD4+ and CD8+T cell net frequencies of interferon (IFN)γ, CD107a, interleukin (IL)-4, and IL-10 expression in SARS-CoV-2-specific T cells derived from lung or blood from the same patient. Each patient (n = 7) is labeled on the right and includes disease profile and sampling time: months (mo) after initial SARS-CoV-2 detection by PCR. Viral proteins are shown in color green (membrane protein, M), orange (nucleocapsid protein, N), and purple (spike protein, S). 12 NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 ARTICLE immune protection46. Furthermore, an increased expression of represented a hallmark of disease severity, being the sole response IL-4 was detected in the lungs of patients who died from SARS- in the fatal case. Further, the observation that N-specific EM CoV-2 infection47. In our cohort, spontaneous IL-4 secretion CXCR3+CD8+T cell responses with an IFNγ+CD107a+ Th1 from both T cell subsets correlated with disease severity, and profile were associated with nonhospitalization during sympto- responses against the spike protein strongly stimulated this matic COVID-19 suggests a favorable immune response when response, potentially suggesting the induction of a stronger the N protein is targeted by CD8+T cells, and that these antibody-directed response. Last, the patient with mild infection responses can migrate towards infected tissues. The CXCR3- who tested positive for 4 months (HL65) showed the most fre- CXCL10 axis appears critical for the recruitment of CD4+ and quent IL-4 response in the lung, which was against S peptides. CD8+T cells that control influenza and tuberculosis infection in Caspase-mediated apoptosis in the immune system is a major the lung, respectively25,53. Consequently, lung T cell recruitment contributor to immune homeostasis in a process termed may partially contribute to the lymphopenia detected in activation-induced cell death48, which could potentially con- patients54, where this antiviral response will likely establish as tribute to minimize an overwhelming cytokine response. Our resident memory cells. In fact, N-specific IFNγ+CD107+ TRM results are consistent with apoptosis significantly contributing to were detected in five convalescent patients several months after the lymphopenia detected in COVID-19 patients, and the pre- infection. ferential loss of CD8+T cells is accompanied by an increase in Recent deep immune profiles of COVID-19 patients have caspase-3 within this compartment, consistent with a pre- identified T-bet expression as a transcriptional factor associated publication49. Since inflammatory molecules can often be to patients with better prognosis55. Importantly, T-bet is not only potent activators of cell death, increasing levels of IL-10 may a key regulator of Th1 immune effector responses and CXCR3 moderate the extent of apoptosis induced, as occurs in mouse expression, essential for effective clearance of pathogens and models of bacterial infection50. Moreover, not only may the maintenance of immunity56, but also crucial for migration, pro- inflammatory environment contribute to a higher proportion of liferation and survival of T regulatory (Treg) cells during Th1- bystander T cells succumbing to apoptotic cell death, but viral mediated immune responses in vivo57. Indeed, CXCR3 is also proteins may also induce apoptosis in both antigen and non- found on a subset of CD4+Foxp3+T cells, and the control of antigen-specific T cells. While effector T cells with degranulation inflammatory responses at mucosal surfaces requires IL-10 pro- capacity are expected to be more terminally differentiated and, ducing Treg cells57. One of the main correlates of disease control consequently, may be more prone to activation-induced cell during acute infection was IL-10 secretion, which dominated the death, the fact that other subsets (i.e., CCR7hCXCR3d/IL- specific immune response of nonhospitalized patients. Two pre- 10 secreting) were more affected is intriguing and requires vious studies have detected reduced frequencies of Treg cells in further study. severe COVID-19 cases26,58; however, IL-10 was detected in Critically, the only cytokine that was higher in nonhospitalized supernatants from stimulated PBMC of severe patients during compared to hospitalized groups was IL-12p70 (Supplementary acute infection14, and a similar trend was observed in a cohort of Fig. 1c), suggesting that, as occurs with other respiratory acute patients with a wider range of sampling (4–37 days)13. viruses51, cell-mediated Th1 immunity is necessary for recovery While, as detected in the plasma of our own cohort of patients, from respiratory infection. A general predominance of a Th1 increased serum levels of IL-10 have been widely associated with profile does not compromise the generation of neutralizing COVID-19 and disease severity (reviewed in37), several factors antibodies46. Further, it is increasingly accepted that Th-cell may explain these results. The plasma source of IL-10 can have subsets are plastic, especially during responses to pathogens multiple cellular origins, ranging from myeloid subsets to epi- in vivo, and even cytokines such as IL-10 can be produced by thelial cells, since not only may almost all leukocytes produce IL- subpopulations of cells within multiple effector subsets46. How- 10, but also the range amplifies during inflammation50. The ever, SARS-CoV-2-specific T cells from acute responders systemic increase in IL-10 may still act as a compensatory demonstrated a biased Th1 profile, where IFNγ-secreting T cells response to limit massive ongoing inflammation in severe could also secrete IL-2 in the case of CD4+, and TNF and patients59. We propose that an early effector-specific T cell granzyme B (but not IL-10) in the case of CD8+[13]. Our data response coordinated with engagement of other immune profiles also show marginal co-expression between IFNγ, IL-4, and IL-10, limiting inflammation may aid at promoting infection resolution. suggesting multiple polarizations of antigen-responding T cells46, This notion is supported by comprehensive analyses of common which is also supported by different CCR7/CXCR3 subsets being immune correlates of protection from mortality in mouse models the main contributors to a given response. Moreover, this of influenza and SARS-CoV infection, which revealed a unique T polarization was partially induced by the targeted protein, where regulatory suppressive profile that contributed to this balance60. M peptides induced the strongest IFNγ secretion in CD4+T cells, Furthermore, it has previously been shown that Treg activity is N peptides enhanced cytotoxicity in CD8+T cells and S peptides required during viral infections to allow for appropriate genera- had an overall predominant Th2 profile (IL-4). Similar to what tion and migration of immune effector cells to the site of was observed for convalescent patients9, M and N-specific infection61,62, while blocking the action of the IL-10 secreted by responses dominated in nonhospitalized and mild-hospitalized antiviral T cells results in enhanced pulmonary inflammation and cases, indicating that a candidate vaccine including only SARS- lethal injury17,50,63. In fact, during influenza infections, type I IFN CoV-2 spike would limit the array of responses during natural signaling may contribute to IL-10-producing lymphocyte infection9. Furthermore, a comparison between the functional recruitment to the site of infection to moderate excessive profile of antigen-specific CD4+ and CD8+T cells based on the inflammation, which will be coincident with the onset of the viral protein targeted evidenced a wider functional profile for adaptive immune response50, being CD8+T cells a primary CD8+T cells targeting M or N peptides compared to S, which was source of IL-10 production in the respiratory tract64. Accordingly, more often observed in milder cases than in severe ones12 and while IL-10 plasma levels increased with disease severity, the fatal agrees with recent identification of responses to frequently case had one of the lowest levels in plasma, which was accom- recognized CD8+T cell epitopes52. Overall, our results concur panied by an absolute lack of IL-10 secretion by antigen-specific with a broader Th profile in CD8+T cells induced by the non- T cells. spike viral proteins and associated with a less severe infection, TRM strategically residing in peripheral tissues are key to while spike responses dominated by IL-4+ CD8+T cells controlling mucosal infections and providing rapid and durable NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 immunity against reinfection20,65. Previous studies in SARS- beyond the spike protein, in particular nucleocapsid peptides, recovered patients already pointed towards persistence of a which should broaden and balance the functional profile of memory T cell response for up to 6 years after infection, and memory T cells, resembling control of natural infection. suggested vaccine-mediated induction of TRM as a long-term protection strategy5. In concordance, a larger proportion of CD8 Methods +T cell effectors with T RM characteristics were present in Ethics statement. This study was performed in accordance with the Declaration of bronchoalveolar lavages from patients with moderate infection Helsinki and approved by the corresponding Institutional Review Board (PR(AG) compared to severe-infected patients66. We indeed report the 192/2020, PR(AG)212/2020, PR(AG)116/2018, and PR(AG)117-2018) of the Vall d’Hebron University Hospital (HUVH), Barcelona, Spain. Written informed existence of a high frequency of SARS-CoV-2-specific TRM in the consent was provided by all patients recruited to this study and samples were lung of some patients infected 6–10 months before. In particular, prospectively collected and cryopreserved in the Vall d’Hebron Research Institute. for the severe and durable COVID-19 patient (HL27), while all functions were represented in TRM in a remarkably higher pro- Healthy donors. Blood samples from healthy adult donors were obtained via portion than in blood, IFNγ in response to S peptides dominated. phlebotomy. These blood samples were collected for studies unrelated to COVID- In this sense, high frequency of spike protein-specific CD4+T cell 19 between September 2018 and June 2019. At the time of enrollment in the initial responses was observed in blood in patients who had recovered studies, all individual donors provided informed consent that their samples could from COVID-194,9,12. Importantly, CD4+T cells are necessary for be used for future studies. These samples were considered to be from unexposed controls given that SARS-CoV-2 emerged as a novel pathogen in December 2019 the formation of protective CD8+TRM during influenza infection, and these samples were largely collected before this date. These donors were and cytokines, such as IFNγ, are necessary signals for this considered healthy in that they had no known history of any significant systemic process20. However, CD4+T cells themselves can be cytotoxic illnesses. The cohort of healthy donors includes 12 individuals. and, actually, have been shown to confer protection against influenza20. We also detected degranulation in response to viral Patients with SARS-CoV-2 infection. Adult patients, 18 years old and older, peptides in CD4+ and even more so in CD8+T cells from the diagnosed with acute COVID-19 were recruited at the Vall d’Hebron Hospital during the first COVID-19 outbreak between March and May 2020. Diagnosis of lung, which in the case of the asymptomatic young patient were acute COVID-19 was defined by symptomatology and/or clinical findings and completely absent in blood. Lack of degranulation in blood from confirmed by positive reverse-transcriptase polymerase chain reaction (RT-PCR) convalescent patients has also been reported12. Further, the fact for SARS-CoV-2 in a respiratory tract specimen. Immunocompromised patients in that the lung sample of this young asymptomatic patient was which the immune response may be affected were excluded of the study. Twenty milliliters of blood were collected at baseline by phlebotomy in two EDTA tubes 3 weeks after RT-PCR laboratory confirmation of infection, and stored at room temperature briefly prior to processing for PBMC and plasma suggests early recruitment of cytotoxic T cells to the lung even in isolation. Samples were obtained between 7 and 16 days after symptoms onset. asymptomatic cases. Moreover, the most frequent responses by Biochemistry analyses were measured at baseline in all patients and, circulating T cells from this asymptomatic patient were CD4+ subsequently, according to the clinical care needs of each patient during infection. and CD8+T cells secreting IL-10, which concurs with the dom- Routine clinical laboratory analyses included complete blood count, coagulation testing (including D-dimer measurement), liver and renal function, electrolytes and inating pattern in nonhospitalized patients during acute infection. inflammatory profile (including C-reactive protein, fibrinogen, ferritin, and IL-6). We acknowledge that our study has several limitations, one The study cohort consisted of 12 patients with severe disease, 20 with mild disease, being that sample size for the different groups was small to be and 14 nonhospitalized individuals. Patient information is summarized in conclusive. However, this was compensated by a narrow window Supplementary table 1. According to disease severity patients, at the discretion of the treating physician, patients were classified in three groups: of sampling during acute infection (7–16 days, post-symptoms onset) and by a comprehensive clinical characterization to stratify (a) Patients with severe disease: individuals with radiologically confirmed patients to study groups. In this sense, multiple correlations pneumonia that required hospitalization and had acute respiratory failure and/or analytical parameters of severity and/or extensive radiological support our main findings and provide strength to our data, involvement. which is also largely supported by current literature. Further, (b) Patients with mild disease: individuals with radiologically confirmed identification of the precise phenotypes generating antigen- pneumonia that required hospital admission but without criteria of severity. specific T responses, such as Treg cells for IL-10, or the con- (c) Nonhospitalized patients: individuals without pneumonia and with pauci- asymptomatic disease that did not require hospitalization and managed on sideration of other T lymphocytes such as γδ T cells should also an outpatient clinic. be considered in future studies. Last, only seven lung samples obtained from different COVID-19 convalescent patients could Data were collected prospectively from the medical charts of the patients. We be studied. While these patients are so far scarce, the immune collected sociodemographic characteristics, past medical records, Charlson responses identified in those samples not only contributed to comorbidity score, concomitant medication, treatments against SARS-CoV-2 round out the present report, but also represent the first evidence infection, adverse drug events, blood test results, imaging studies, microbiological tests, and supportive measures needed. Vital signs, symptoms and physical to our knowledge of persisting SARS-CoV-2-specific TRM in the examination were recorded. Laboratory, microbiology, and imaging studies were lung. Disease severity during acute SARS-CoV-2 infection is performed according to the clinical care needs of each patient. associated with strong peripheral T and B cell responses4,27, which not only may relate to antigenic burden but, also could Lung samples. Lung tissue was obtained from seven patients recovered from potentially translate into a significant proportion of antigen- SARS-CoV-2 infection who needed a lung resection (Supplementary table 3). specific TRM in the lung once the patient recovers. Remaining Analyses were performed using healthy areas from the lung resection. HL24 sample important questions concern the level of viral replication and corresponded to a 21-year-old smoking man who had detectable viral load by RT- PCR without symptoms, followed by two negative RT-PCR measurements (8 and associated symptomatology that will stimulate an effective 18 days after the positive result). He underwent surgery for pneumothorax 21 days immune response at the respiratory tract, and also, how quick this after the positive SARS-CoV-2 detection. HL52, HL75, and HL81 corresponded to response will be established. However, the fact that an asymp- a 71, 77, and a 64-year-old men hospitalized for 5, 13, and 9 days, respectively, for tomatic patient had degranulating antigen-specific T cells in the mild disease (baseline IL-6 of 26.15, 9.13, and 309 pg/mL, respectively). Both patients underwent thoracic surgery due to lung carcinoma 7.5, 10.5, and lung 3 weeks after infection is at least encouraging. Overall, a 3.7 months after confirmatory RT-PCR. HL65 was a 52-year-old woman hospi- balanced effector/anti-inflammatory response may be key for talized for three days (without oxygen requirements), with confirmatory SARS- early viral containment, where antigen-specific IL-10+ T cells CoV-2 RT-PCR and a baseline IL-6 of 5.25 pg/mL. During the next 4 months, she could be determinant in limiting inflammation. Thus, the possi- tested positive for RT-PCR. She underwent thoracic surgery for a pulmonary nodule 2 months after testing negative for RT-PCR (and 6 months after initial bility that overstimulated proinflammatory T cells contribute to discharge). HL69 was a 69-year-old man hospitalized for severe COVID-19 for disease severity cannot be ruled out. Our findings encourage next- 35 days (baseline IL-6 of 34.66 pg/mL) and treated for the symptomatology derived generation vaccine designs to consider including viral proteins of the SARS-CoV-2 infection (with confirmatory RT-PCR). During a post-COVID 14 NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 ARTICLE examination, a pulmonary nodule was diagnosed, which instigated thoracic surgery Phenotyping and intracellular cytokine staining in lung. Lung tissues were 10 months after initial infection. HL27 was a 68-year-old ex-smoker man hospi- collected in antibiotic-containing RPMI 1640 medium from the Thoracic Surgery talized for one month due to respiratory insufficiency caused by SARS-CoV-2 Service at the Vall d’Hebron University Hospital. Immediately following surgery, infection (with confirmatory RT-PCR and a baseline IL-6 of 133.7 pg/mL). One the tissue was dissected into approximately 8-mm3 blocks. These blocks were first month after discharge, he presented a rash in photoexposed skin, potentially enzymatically digested with 5 mg/ml collagenase IV (Gibco) and 100 µg/ml of related to persistent SARS-CoV-2 infection67 or treatment with DNase I (Roche) for 30 min at 37 °C and 400 rpm and, then, mechanically digested hydroxychloroquine68. RT-PCR for SARS-CoV-2 tested positive again, turning with a pestle. The resulting cellular suspension was filtered through a 70 µm pore into a RT-PCR negative 2 weeks after. During this time, a lung carcinoma was size cell strainer (Labclinics), washed twice with PBS and cultivated with R10 in a diagnosed, which instigated thoracic surgery 3 months after the negative RT-PCR round-bottom 96-well plate overnight at 37 °C with 1 µg/ml of SARS-CoV-2 (and 5 months after initial discharge). peptides (M, N, and S) in the presence of 3 μL/mL α-CD28/CD49d (clones L293 Concomitant to the lung samples, blood samples were also collected. Thus and L25, BD Biosciences), 0.5 μL/mL Brefeldin A (BD Biosciences), 0.35 μL/mL analyzed paired blood and tissue samples corresponded to 21 days after the first Monensin (BD Biosciences) and 5 μL/mL anti-CD107a-PE-Cy5. For each patient, a positive SARS-CoV-2 detection for the HL24 patient (asymptomatic), 3.7 months negative control, cells treated with medium, and positive control, cells incubated in after initial infection for HL81 (mild), and between 6 and 10 months after initial the presence of 40.5 nM PMA and 0.5 μM Ionomycin, were included. Next day, infection for HL65 and HL75 (mild), HL52 (mild-PCR+ for 4 months), HL69 cellular suspensions were stained with Live/Dead Aqua (Invitrogen) and anti- (severe), and HL27 (severe-PCR+ for 2 months) patients. Blood samples were CD103 (FITC, Biolegend), anti-CD69 (PE-CF594, BD Biosciences), anti-CD40 immediately processed for PBMC and serum isolation, which was used for SARS- (APC-Cy7, Biolegend), anti-CD8 (APC, BD Biosciences), anti-CD3 (BV650, BD CoV-2 antibody detection. Biosciences), and anti-CD45 (BV605, BD Biosciences) antibodies. Cells were subsequently fixed and permeabilized using the Foxp3 Staining Buffer Set (Thermo Fisher Scientific, Cat. #00-5523-00) and stained with anti-IL-4 (PE-Cy7, SARS-CoV-2 RT-qPCR. Upper (naso/oropharyngeal swabs) and lower (bronch- eBioscience), anti-IL-10 (PE, BD Biosciences), anti-T-bet (BV421, Biolegend), and oalveolar lavage, tracheal aspirate, sputum, or bronchoaspirate) respiratory tract anti-IFNγ (AF700, Invitrogen) antibodies. After fixation with PBS 2% PFA, cells specimens from subjects with suspicion of COVID-19 were received and tested at were acquired in a BD LSR Fortessa flow cytometer. FACS Diva software 6.2. the Respiratory Viruses Unit of the Microbiology Department of the HUVH. (firmware version 1.2) was used for flow cytometry sample acquisition and ana- COVID-19 diagnosis was performed by two commercial RT-PCR-based assays, lyzed with FlowJo v10.7.1 software (TreeStar). Allplex™ 2019-nCoV (Seegene, Korea, Cat. #RP10244Y) or Cobas® SARS-CoV-2 (Roche Diagnostics, USA, Cat. #09175431190) tests. In addition, an in-house PCR assay using the primer/probe set targeting the nucleocapsid protein (N1) and the human RNase P (housekeeping gene), from the CDC 2019-nCoV Real-Time RT- SARS-CoV-2 serology. Serological status of HL24, HL27, HL52, HL65, HL69, PCR Diagnostic Panel (Qiagen, Hilden, Germany), was performed. Details on the HL75 and HL81 patients was determined in serum using two commercial che- primers and probes used are shown in Supplementary Table 4. In order to mini- miluminescence immunoassays (CLIA) targeting specific SARS-CoV-2 antibodies: mize variations due to a nonstandardized collection of a heterogeneous specimen, (1) Elecsys Anti-SARS-CoV-2 (Roche Diagnostics, USA, Cat. #09203079190) was the Ct values of the viral target were normalized to the housekeeping gene based on performed on the Cobas 8800 system (Roche Diagnostics, USA) for qualitative the 2–ΔCt method, where ΔCt corresponds to the formula Ct sample − Ct determination of total antibodies (including IgG, IgM, and IgA) against nucleo- housekeeping. capsid SARS-CoV-2 protein; and (2) Liaison SARS-CoV-2 S1/S2 IgG (DiaSorin, Italy, Cat. #311510) was performed on the LIAISON XL Analyzer (DiaSorin, Italy) for quantitative determination of IgG against the spike (S) glycoprotein subunits 1 and 2 (S1/S2). Plasma cytokine quantification. Plasma obtained from our three cohorts of COVID-19 patients (n = 46) was analyzed using Ella® platform (Bio-Techne, Minneapolis, Minnesota, USA) for the quantification of the following cytokines and chemokines: CCL2, GM-CSF, IL-10, IL-12p70, IL-1ra, IL-6, IL-7, TNF, SARS-CoV-2 detection by immunofluorescence and RNA hybridization. CXCL10, Granzyme B, IFNγ, IL-13, IL-15, IL-17A, and IL-4. Samples were 1:2 Paraffin-embedded lung tissue samples were processed and analyzed at the diluted with sample diluent provided by the manufacturer and loaded onto mul- Pathology Department of the HUVH. For SARS-CoV-2, lung tissue sections of tiplex cartridges according to manufacturer’s instructions prior to their analysis. 3 μm were deparaffinized with xylene and dehydrated in ethanol. Samples were Results are expressed as pg/mL. pretreated with CC2 (pH = 6), and rinsed with working PBS. SARS-CoV-2 (SARS- CoV Nucleoprotein/NP Antibody, Rabbit PAb, 6F10, Sino Biological, dilution at 1:1000) antibody was applied and incubated during 1 h. After washing in PBS, Phenotyping and intracellular cytokine staining in blood. PBMCs were isolated slides were mounted in 80% glycerol and sealed. Images were taken using from blood by density-gradient centrifugation using Ficoll-Paque and immediately BenchMArk Ultra Ventana System. cryopreserved and stored in liquid nitrogen until use in the assays. Cells were RNA hybridization was performed using RNAscope VS Universal Assays and thawed the day before the assay and cultured in a T-25 flask at 37 °C with RPMI the Ventana Discovery Ultra System. A high sensitivity target-specific probes to 1640 (Gibco) supplemented with 10% Fetal Bovine Serum (FBS) (Gibco), 100 SARS-CoV-2 mRNA sequence (probe V-nCoV2019-S, ADC Biotecnec biology) µg/ml streptomycin (Fisher Scientific), and 100 U/ml penicillin (Fisher Scientific) were used. Lung tissue sections of 3 μm tissue sections were mounted on Superfrost (R10). Next day, previous to SARS-CoV-2-peptide pool stimulation, cells were Plus microscope slides (Fisher Scientific). The assay was performed according to stained for CCR7 (PE-CF594, BD Biosciences) and CXCR3 (BV650, BD Bios- manufacturer’s instructions. Briefly, samples were deparaffinized and pretreated as ciences) for 30 min at 37 °C. After a washing with PBS, PBMCs were stimulated in mentioned. Next, probes were incubated for 2 h at 40 °C and samples were stored a round-bottom 96-well plate for 5 h at 37 °C with 1 µg/ml of SARS-CoV-2 pep- overnight in 5× saline sodium citrate buffer. The following day, amplification and tides (PepTivator SARS-CoV-2 M, N and S, Miltenyi Biotec) in the presence of 1 signal development was performed by sequential incubation of Pre-Amplifiers, μl/ml of Brefeldin A (BD Biosciences), 0.7 μl/ml of Monensin (BD Biosciences) and Amplifiers and label probe according to the manufacturer’s instructions (Kit 3 μl/ml of α-CD28/CD49d (clones L293 and L25, BD Biosciences). Anti-CD107a DISCOVERY mRNA, Roche, Cat. #06614353001). Lastly, samples were revealed (PE-Cy5, BD) was also added at this time. For each patient, a negative control, cells with DAB staining (3,3′-Diaminobenzidine). The experiment controls used were treated with medium, and positive control, cells incubated in the presence of 81 nM infected and non-infected HeLA cells. PMA and 1 μM Ionomycin, were included. After stimulation, cells were washed twice with PBS and stained with Aqua LIVE/DEAD fixable dead cell stain kit (Invitrogen, Cat. #L34966). Cell surface antibody staining included anti-CD3 (Per- CP), anti-CD4 (BV605), and anti-CD56 (FITC) (all from BD Biosciences). Cells Statistical analyses and reproducibility. All analysis and figures are repre- were subsequently fixed and permeabilized using the Cytofix/Cytoperm kit (BD sentative of a single probing experiment. Flow cytometry data were analyzed using Biosciences, Cat. #554714) and stained with anti-Caspase-3 (AF647, BD Bios- FlowJo v10.7.1 software (TreeStar). Data and statistical analyses were performed ciences), anti-Bcl-2 (BV421, Biolegend), anti-IL-4 (PE-Cy7, eBioscience), anti-IL- using Prism 7.0 (GraphPad Software, La Jolla, CA, USA), unless otherwise stated. 10 (PE, BD Biosciences), and anti-IFNγ (AF700, Invitrogen) for 30 min. Cells were The statistical specifics of the experiments are provided in the respective figure then fixed with PBS 2% PFA and acquired in a BD LSR Fortessa flow cytometer legends. Data plotted in linear scale were expressed as median + Interquartile (Cytomics Platform, High Technology Unit, Vall d’Hebron Institut de Recerca). (IQR) or Min to Max range, unless otherwise stated. Correlation analyses were FACS Diva software 6.2. (firmware version 1.2) was used for flow cytometry sample performed using Spearman rank correlation. For correlations in Supplementary acquisition and analyzed with FlowJo v10.7.1 software (TreeStar). FMO controls table 2, a Benjamini–Hochberg False Discovery Rate was also calculated by R Core were used to draw the gates for each function. Team (2020). Mann–Whitney and Wilcoxon tests were applied for unpaired or For the patients with lung samples, their contemporary blood sample was paired comparisons, respectively, while Kruskal–Wallis rank-sum test with Dunn’s processed immediately and the T cell response assay was performed without a post-hoc test was used for multiple comparisons. A P-value <0.05 was considered previous cryopreservation step. Isolated PBMC were rested for 4 h in the incubator significant. For most analyses, antigen-specific T cell data have been calculated as and then stimulated with the same SARS-CoV-2 peptides (M, N, and S) overnight, the net frequency, where the individual percentage of expression for a given following the same protocol described above with half the amount of Brefeldin A molecule in the control condition (vehicle) has been subtracted from the corre- and Monensin to avoid toxicity. sponding SARS-CoV-2-peptide stimulated conditions. NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications 15 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 Reporting summary. Further information on research design is available in the Nature 26. Chen, G. et al. Clinical and immunological features of severe and moderate Research Reporting Summary linked to this article. coronavirus disease 2019. J. Clin. Invest. 130, 2620–2629 (2020). 27. Chen, Z. & John Wherry, E. T cell responses in patients with COVID-19. Nat. Rev. Immunol. 20, 529–536 (2020). Data availability 28. Giamarellos-Bourboulis, E. J. et al. Complex immune dysregulation in The authors declare that the data supporting the findings of this study are available COVID-19 patients with severe respiratory failure. Cell Host Microbe 27, within the paper and its supplementary information files. Source data are provided with 992–1000 e1003 (2020). this paper. 29. Huang, C. et al. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet 395, 497–506 (2020). Received: 1 February 2021; Accepted: 24 April 2021; 30. Barathan, M. et al. Chronic hepatitis C virus infection triggers spontaneous differential expression of biosignatures associated with T cell exhaustion and apoptosis signaling in peripheral blood mononucleocytes. Apoptosis 20, 466–480 (2015). 31. Genesca, M. et al. Live attenuated lentivirus infection elicits polyfunctional simian immunodeficiency virus Gag-specific CD8+ T cells with reduced apoptotic susceptibility in rhesus macaques that control virus replication after References challenge with pathogenic SIVmac239. J. Immunol. 179, 4732–4740 (2007). 1. Long, Q. X. et al. Clinical and immunological assessment of asymptomatic 32. Tan, Y. J., Lim, S. G. & Hong, W. Regulation of cell death during infection by SARS-CoV-2 infections. Nat. Med. 26, 1200–1204 (2020). the severe acute respiratory syndrome coronavirus and other coronaviruses. 2. Mallapaty, S. Will antibody tests for the coronavirus really change everything? Cell. Microbiol. 9, 2552–2561 (2007). Nature 580, 571–572 (2020). 33. Codo, A. C. et al. Elevated glucose levels favor SARS-CoV-2 infection and 3. Robbiani, D. F. et al. Convergent antibody responses to SARS-CoV-2 in monocyte response through a HIF-1alpha/glycolysis-dependent axis. Cell convalescent individuals. Nature 584, 437–442 (2020). Metab. 32, 498–499 (2020). 4. Sekine, T. et al. Robust T cell immunity in convalescent individuals with 34. Secinaro, M. A. et al. Glycolysis promotes caspase-3 activation in lipid rafts in asymptomatic or mild COVID-19. Cell 183, 158–168 e114 (2020). T cells. Cell Death Dis. 9, 62 (2018). 5. Channappanavar, R., Zhao, J. & Perlman, S. T cell-mediated immune response 35. Tan, L. et al. Lymphopenia predicts disease severity of COVID-19: a to respiratory coronaviruses. Immunol. Res. 59, 118–128 (2014). descriptive and predictive study. Signal Transduct. Target Ther. 5, 33 (2020). 6. Huang, A. T. et al. A systematic review of antibody mediated immunity to 36. Vabret, N. et al. Immunology of COVID-19: current state of the science. coronaviruses: kinetics, correlates of protection, and association with severity. Immunity 52, 910–941 (2020). Nat. Commun. 11, 4704 (2020). 37. Costela-Ruiz, V. J., Illescas-Montes, R., Puerta-Puerta, J. M., Ruiz, C. & 7. Wajnberg, A. et al. Robust neutralizing antibodies to SARS-CoV-2 infection Melguizo-Rodriguez, L. SARS-CoV-2 infection: the role of cytokines in persist for months. Science 370, 1227–1230 (2020). COVID-19 disease. Cytokine Growth Factor Rev. 54, 62–75 (2020). 8. Braun, J. et al. SARS-CoV-2-reactive T cells in healthy donors and patients 38. Del Valle, D. M. et al. An inflammatory cytokine signature predicts COVID- with COVID-19. Nature 587, 270–274 (2020). 19 severity and survival. Nat. Med. 26, 1636–1643 (2020). 9. Grifoni, A. et al. Targets of T cell responses to SARS-CoV-2 coronavirus in 39. Han, H. et al. Profiling serum cytokines in COVID-19 patients reveals IL-6 humans with COVID-19 disease and unexposed individuals. Cell 181, and IL-10 are disease severity predictors. Emerg. Microbes Infect. 9, 1123–1130 1489–1501 e1415 (2020). (2020). 10. Le Bert, N. et al. SARS-CoV-2-specific T cell immunity in cases of COVID-19 40. Yang, Y. et al. Plasma IP-10 and MCP-3 levels are highly associated with and SARS, and uninfected controls. Nature 584, 457–462 (2020). disease severity and predict the progression of COVID-19. J. Allergy Clin. 11. Ni, L. et al. Detection of SARS-CoV-2-specific humoral and cellular immunity Immunol. 146, 119–127 e114 (2020). in COVID-19 convalescent individuals. Immunity 52, 971–977 e973 (2020). 41. Sallusto, F., Lenig, D., Forster, R., Lipp, M. & Lanzavecchia, A. Two subsets of 12. Peng, Y. et al. Broad and strong memory CD4(+) and CD8(+) T cells induced memory T lymphocytes with distinct homing potentials and effector by SARS-CoV-2 in UK convalescent individuals following COVID-19. Nat. functions. Nature 401, 708–712 (1999). Immunol. 21, 1336–1345 (2020). 42. Rha, M. S. et al. PD-1-expressing SARS-CoV-2-specific CD8(+) T cells are not 13. Rydyznski Moderbacher, C. et al. Antigen-specific adaptive immunity to exhausted, but functional in patients with COVID-19. Immunity 54, 44–52 SARS-CoV-2 in acute COVID-19 and associations with age and disease e43 (2021). severity. Cell 183, 996–1012 e1019 (2020). 43. Kumar, S. Caspase function in programmed cell death. Cell Death Differ. 14, 14. Weiskopf, D. et al. Phenotype and kinetics of SARS-CoV-2-specific T cells in 32–43 (2007). COVID-19 patients with acute respiratory distress syndrome. Sci. Immunol. 5, 44. Behr, F. M., Chuwonpad, A., Stark, R. & van Gisbergen, K. Armed and ready: eabd2071 (2020). transcriptional regulation of tissue-resident memory CD8 T cells. Front. 15. Channappanavar, R., Fett, C., Zhao, J., Meyerholz, D. K. & Perlman, S. Virus- Immunol. 9, 1770 (2018). specific memory CD8 T cells provide substantial protection from lethal severe 45. Bacher, P. et al. Low-avidity CD4(+) T cell responses to SARS-CoV-2 in acute respiratory syndrome coronavirus infection. J. Virol. 88, 11034–11044 unexposed individuals and humans with severe COVID-19. Immunity 53, (2014). 1258–1271 e1255 (2020). 16. Teijaro, J. R. et al. Cutting edge: Tissue-retentive lung memory CD4 T cells 46. Swain, S. L., McKinstry, K. K. & Strutt, T. M. Expanding roles for CD4(+) mediate optimal protection to respiratory virus infection. J. Immunol. 187, T cells in immunity to viruses. Nat. Rev. Immunol. 12, 136–148 (2012). 5510–5514 (2011). 47. Vaz de Paula, C. B. et al. IL-4/IL-13 remodeling pathway of COVID-19 lung 17. Zhao, J. et al. Airway memory CD4(+) T cells mediate protective immunity injury. Sci. Rep. 10, 18689 (2020). against emerging respiratory coronaviruses. Immunity 44, 1379–1391 (2016). 48. Lakhani, S. & Flavell, R. A. Caspases and T lymphocytes: a flip of the coin? 18. Slutter, B., Pewe, L. L., Kaech, S. M. & Harty, J. T. Lung airway-surveilling Immunol. Rev. 193, 22–30 (2003). CXCR3(hi) memory CD8(+) T cells are critical for protection against 49. Adamo, S. et al. Lymphopenia-induced T cell proliferation is a hallmark of influenza A virus. Immunity 39, 939–948 (2013). severe COVID-19. bioRxiv https://doi.org/10.1101/2020.08.04.236521 (2020). 19. Kohlmeier, J. E. & Woodland, D. L. Immunity to respiratory viruses. Annu. Preprint at. Rev. Immunol. 27, 61–82 (2009). 50. Iyer, S. S. & Cheng, G. Role of interleukin 10 transcriptional regulation in 20. Oja, A. E. et al. Trigger-happy resident memory CD4(+) T cells inhabit the inflammation and autoimmune disease. Crit. Rev. Immunol. 32, 23–63 (2012). human lungs. Mucosal Immunol. 11, 654–667 (2018). 51. Bont, L., Kavelaars, A., Heijnen, C. J., van Vught, A. J. & Kimpen, J. L. 21. Woodland, D. L. & Scott, I. T cell memory in the lung airways. Proc. Am. Monocyte interleukin-12 production is inversely related to duration of Thorac. Soc. 2, 126–131 (2005). respiratory failure in respiratory syncytial virus bronchiolitis. J. Infect. Dis. 22. Wu, T. et al. Lung-resident memory CD8 T cells (TRM) are indispensable for 181, 1772–1775 (2000). optimal cross-protection against pulmonary virus infection. J. Leukoc. Biol. 95, 52. Ferretti, A. P. et al. Unbiased screens show CD8(+) T cells of COVID-19 215–224 (2014). patients recognize shared epitopes in SARS-CoV-2 that largely reside outside 23. Pizzolla, A. et al. Influenza-specific lung-resident memory T cells are the spike protein. Immunity 53, 1095–1107 e1093 (2020). proliferative and polyfunctional and maintain diverse TCR profiles. J. Clin. 53. Moguche, A. O. et al. ICOS and Bcl6-dependent pathways maintain a CD4 T Invest. 128, 721–733 (2018). cell population with memory-like properties during tuberculosis. J. Exp. Med. 24. Groom, J. R. & Luster, A. D. CXCR3 in T cell function. Exp. Cell Res. 317, 212, 715–728 (2015). 620–631 (2011). 54. Tay, M. Z., Poh, C. M., Renia, L., MacAry, P. A. & Ng, L. F. P. The trinity of 25. Pejoski, D. et al. Site-specific DC surface signatures influence CD4(+) T cell COVID-19: immunity, inflammation and intervention. Nat. Rev. Immunol. co-stimulation and lung-homing. Front. Immunol. 10, 1650 (2019). 20, 363–374 (2020). 16 NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-23333-3 ARTICLE 55. Mathew, D. et al. Deep immune profiling of COVID-19 patients reveals funded by the Spanish Health Institute Carlos III (CP17/00179). N.M. is supported by a distinct immunotypes with therapeutic implications. Science 369, Ph.D. fellowship from the Vall d’Hebron Institut de Recerca (VHIR) and A.A.-G. and N. eabc8511 (2020). S.-G. are supported by a Ph.D. fellowship from the Spanish Secretariat of Science and 56. Kallies, A. & Good-Jacobson, K. L. Transcription factor T-bet orchestrates Innovation (BES-2016-076382, PRE2019-087393). The funders had no role in study lineage development and function in the immune system. Trends Immunol. design, data collection and analysis, the decision to publish, or preparation of the 38, 287–297 (2017). manuscript. 57. Koch, M. A. et al. The transcription factor T-bet controls regulatory T cell homeostasis and function during type 1 inflammation. Nat. Immunol. 10, 595–602 (2009). Author contributions Conceptualization, M.J.B. and M.G.; Patient Recruitment and Sample Collection, J.R., 58. Qin, C. et al. Dysregulation of immune response in patients with coronavirus A.T., B.P., J.N., P.S., A.L.A., B.A., V.F., J.B.; Methodology, A.F., C.K., J.G.-E., N.S.-G., 2019 (COVID-19) in Wuhan, China. Clin. Infect. Dis. 71, 762–768 (2020). N.M., M.S., J.E., A.N.A., and S.R.C.; Investigation, J.G.-E., N.S.-G., N.M., M.S., A.A.-G., 59. Gong, J. et al. Correlation analysis between disease severity and inflammation- D.P., D.A.-S., I.S., J.E., A.N.A., and M.G.; Formal Analysis, J.G.-E., N.S.-G., N.M., M.S., related parameters in patients with COVID-19: a retrospective study. BMC D.P., and M.G.; Writing-Original Draft J.G.-E., N.S.-G., N.M., M.S., and M.G; Writing- Infect. Dis. 20, 963 (2020). Review and Editing, R.P.-B., M.J.B., and M.G.; Funding Acquisition, M.J.B. and M.G.; 60. Graham, J. B. et al. Immune predictors of mortality after ribonucleic acid virus all authors revised the manuscript; Supervision, M.J.B. and M.G. N.M. and M.S. infection. J. Infect. Dis. 221, 882–889 (2020). contributed equally. 61. Lund, J. M., Hsing, L., Pham, T. T. & Rudensky, A. Y. Coordination of early protective immunity to viral infection by regulatory T cells. Science 320, 1220–1224 (2008). Competing interests 62. Soerens, A. G., Da Costa, A. & Lund, J. M. Regulatory T cells are essential to The authors declare no competing interests. promote proper CD4 T-cell priming upon mucosal infection. Mucosal Immunol. 9, 1395–1406 (2016). 63. Sun, J., Madan, R., Karp, C. L. & Braciale, T. J. Effector T cells control lung Additional information inflammation during acute influenza virus infection by producing IL-10. Nat. Supplementary information The online version contains supplementary material Med. 15, 277–284 (2009). available at https://doi.org/10.1038/s41467-021-23333-3. 64. Ejrnaes, M. et al. Resolution of a chronic viral infection after interleukin-10 receptor blockade. J. Exp. Med. 203, 2461–2472 (2006). Correspondence and requests for materials should be addressed to M.J.Bón. or M.Gà. 65. Turner, D. L. & Farber, D. L. Mucosal resident memory CD4 T cells in Peer review information Nature Communications thanks the anonymous reviewers for protection and immunopathology. Front. Immunol. 5, 331 (2014). their contribution to the peer review of this work. Peer review reports are available. 66. Liao, M. et al. Single-cell landscape of bronchoalveolar immune cells in patients with COVID-19. Nat. Med. 26, 842–844 (2020). Reprints and permission information is available at http://www.nature.com/reprints 67. Gisondi, P. et al. Cutaneous manifestations of SARS-CoV-2 infection: a clinical update. J. Eur. Acad. Dermatol. Venereol. 34, 2499–2504 (2020). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 68. Skroza, N. et al. A late-onset widespread skin rash in a previous COVID-19- infected patient: viral or multidrug effect? J. Eur. Acad. Dermatol. Venereol. published maps and institutional affiliations. 34, e438–e439 (2020). Open Access This article is licensed under a Creative Commons Acknowledgements Attribution 4.0 International License, which permits use, sharing, We would like to thank all the patients who participated in the study. We also thank adaptation, distribution and reproduction in any medium or format, as long as you give Prof. Shawn C. Kefauver for thoughtful review of the manuscript and Prof. Alex Sanchez- appropriate credit to the original author(s) and the source, provide a link to the Creative Pla from the Statistics and Bioinformatics Unit (UEB) at the Vall d’Hebron Institut de Commons license, and indicate if changes were made. The images or other third party Recerca (VHIR), for statistical analysis support. This work was primarily supported by a material in this article are included in the article’s Creative Commons license, unless grant from the Health department of the Government of Catalonia (DGRIS 1_5). This indicated otherwise in a credit line to the material. If material is not included in the work was additionally supported in part by the Spanish Health Institute Carlos III article’s Creative Commons license and your intended use is not permitted by statutory (ISCIII, PI17/01470, and ISCIII COV20/00416), the Spanish Secretariat of Science and regulation or exceeds the permitted use, you will need to obtain permission directly from Innovation and FEDER funds (grant RTI2018-101082-B-I00 [MINECO/FEDER]), the the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Spanish AIDS network Red Temática Cooperativa de Investigación en SIDA (RD16/ licenses/by/4.0/. 0025/0007), the European Regional Development Fund (ERDF), the Fundació La Marató TV3 (grants 201805-10FMTV3 and 201814-10FMTV3) and the Gilead fellowships GLD19/00084 and GLD18/00008. M.J.B. is supported by the Miguel Servet program © The Author(s) 2021 NATURE COMMUNICATIONS | (2021)12:3010 | https://doi.org/10.1038/s41467-021-23333-3 | www.nature.com/naturecommunications 17