Peripheral and lung resident T cell responses against SARS-CoV-2

liquid nitrogen until use assays. assay flask RPMI 1640 (Gibco) 10% Fetal Bovine Serum (FBS) (Gibco), 100µg/ml streptomycin Scientific) and 100 U/ml penicillin (Fisher Scientific) (R10). Next previous to SARS-CoV-2 pool stimulation, cells were for CCR7 (PE-CF594, BD Biosciences) and CXCR3 (BV650, BD Biosciences) for 30 min at 37ºC. After a washing with PBS, PBMCs were stimulated a round bottom 96-well plate for 5h at 37°C with 1µg/ml of SARS-CoV-2 peptides (PepTivator M, N and S, Miltenyi Biotec) in the p resence of 1 μl/ml of Brefeldin A (BD 30 Biosciences), 0.7 μl/ml of Monensin (BD Biosciences) and 3 μl/ml of α -CD28/CD49d (clones L293 and L25, BD Biosciences). Anti-CD107a (PE-Cy7, BD) was also added at this time. For each a negative control, cells treated with medium, and positive control, cells incubated in the presence of 81nM PMA and 1μM Ionomycin, were After stimulation, cells were


1
We are currently facing a global health emergency, the COVID-19 pandemic. While a great effort 2 is focused on vaccine development, many questions remain unanswered that are necessary to 3 properly manage patients and inform vaccine assessment. To this end, identifying the 4 development of a protective immune response after natural infection and characterizing the 5 correlates of protection would greatly inform on the best strategy to stimulate a protective 6 response by immunization. Moreover, identifying specific immunological parameters capable of 19 one patient from the severe group corresponded to a fatal case. Samples were obtained between 20 7-16 days after symptom onset and no differences between groups were detected ( Figure S1A). Table S1 shows a summary of the participant characteristics and baseline determinations, in 22 which significant differences are evidenced between the three groups. As previously reported 26, 23 28, 29, 35 36 , age, lymphopenia, and biochemical parameters such as D-dimer, IL-6, and ferritin were 24 associated with disease severity. Quantification of the viral load between days 5 and 15 after 25 symptom onset is also reported; however, values between the groups were not statistically 26 significant. Some of the clinical parameters used to stratify mild and severe hospitalized cases 27 are shown in Figure S1B: days to discharge since symptoms onset (p<0.0001), baseline IL-6 28 (p=0.008) and the percentage of oxyhemoglobin saturation in arterial blood /fraction of inspired 29 oxygen (SAFI) ratio at baseline (p=0.0052) and after 48h (p=0.0002). These parameters were 30 used to address associations between immunological parameters and disease severity. In some 31 analyses, 12 control individuals sampled before the COVID-19 pandemic were studied in parallel.

32
To determine if whole plasma cytokine levels in our groups of COVID-19 patients were 33 similar to previously defined patterns reported before 29, 37, 38, 39, 40 , we analyzed cytokine plasma 34 5 levels by the ELLA® microfluidics platform in the same samples. Levels of IL-1ra, IL-2, IL-6, IL-1 10, IL-15, CXCL10 (IP-10), IFNg, granzyme B and TNFa were elevated in the plasma of 2 hospitalized groups compared to non-hospitalized patients, with higher levels associated with 3 disease severity ( Figure S1C). Of note, the deceased patient from the severe cohort (circled in 4 green) had very high levels of some of the molecules associated with severity and fatality 5 prediction, namely IL-6 and IL-1ra 38, 40 . Further, CXCL10 was the most significant predictor of 6 hospitalization during acute infection, potentially related to impaired T cell responses as 7 suggested 13 . CCL2, also referred to as monocyte chemoattractant protein 1, was significantly 8 higher in the severe-hospitalized group compared to the non-hospitalized patients, while IL-4, IL-9 7, IL-13, IL-17A, GM-CSF were similar among all three groups of patients ( Figure S1C). Strikingly, 10 the levels of IL-12p70 were higher in the plasma of the non-hospitalized compared to the mild 11 COVID-19 group, and the deceased patient had the second lowest level of IL-12p70 of the severe 12 group (0.093pg/mL; Figure S1B). for each peptide set (background subtracted) and compared these antigen-specific T cell 23 responses among all three groups ( Figure 1A). This way, differences on the frequency of IFNg-24 secreting antigen-specific T cells were significantly higher among the hospitalized groups 25 compared to the outpatients (in CD4 + T cells: p=0.020 for M and S peptides in the mild group; 26 p=0.004 for M, p=0.011 for N and p=0.007 for S peptides in the severe group; Figure 1A). Of note, 27 while non-hospitalized patients did not show a significant increase in the production of IFNg as a 28 group, some individuals did show an increase in their response (>0.02% after background 29 subtraction) representing 43% of responders, which was lower than the 80% and 92% of 30 responders observed in mild and severe hospitalized groups. In contrast degranulation, measured 31 by CD107a expression, was less detected in general and significance among the groups was only 32 reached in response to M peptides in severe patients compared to non-hospitalized patients 33 (p=0.036) ( Figure 1A). We also calculated double positive IFNg/CD107a CD8 + T cells, as a 34 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021. ; https://doi.org/10.1101/2020.12.02.20238907 doi: medRxiv preprint CD107a + IFNg + CD8 + T cells responding to the M peptides positively correlated with viral load, the 3 same subset specific for N peptides inversely correlated with baseline levels of IL-6 within the 4 hospitalized cohort ( Figure S3A).

5
Assessment of two other functions, IL-4 and IL-10, demonstrated major dominance of 6 these responses based on the cohort and the viral target. A general induction of an IL-4-specific 7 CD8 + T cell response was observed in response to the viral spike in hospitalized patients 8 compared to the non-hospitalized individuals (p=0.004 and p=0.003 for mild and severe patients, 9 respectively; Figure 1A). Of note, higher levels of spontaneous secretion of IL-4 (in unstimulated 10 conditions) were observed in hospitalized patients, which essentially correlated with the number 11 of days since symptoms onset to discharge and baseline IL-6 levels ( Figure S3B). Moreover, 12 SARS-CoV-2 viral load positively correlated with the overall capacity of CD4 + T cells to secrete IL-13 4 in response to TCR independent unspecific activation with PMA/ionomycin (PMA/Io) ( Figure   14 S3C). In contrast, the expression of IL-10, a prototypical regulatory cytokine, was significantly 15 increased in CD4 + T cells from non-hospitalized patients after stimulation with M peptides when 16 compared to the mild COVID-19 group (p=0.035; Figure 1A).

17
Correlations between the net frequency of a given function and clinical parameters were 18 consistent with more CD4 + T cells secreting IFNg and more CD8 + T cells secreting IL-4 in response 19 to M and S peptides associated with disease severity (Table S2). Even the total CD4 + or CD8 + T 20 cell IFNg response and the total IL-4 secretion by CD8 + T cells against any of the three viral 21 proteins (all peptides) correlated with more days at the hospital for IFNg or with other clinical 22 parameters for IL-4 (Table S2). Further, antigen-specific CD4 + T cells degranulating in response 23 to all viral peptides and, in the case of CD8 + T cells, in response to M peptides also correlated with 24 higher levels of IL-6 (Table S2). In contrast, the percentage of M-specific CD4 + T cells secreting 25 IL-10 correlated with better prognosis in all clinical parameters (Table S2) and for N-specific 26 positively with better oxygenation at 48h (Table S2).

27
Actually, when the overall response, including all functions, was represented as donut 28 charts displaying the mean frequency of responses including all individuals (responders and non-29 responders), differences among groups in response to each peptide set were emphasized ( Figure   30 1B). This was, M peptides were shown to mostly stimulate IL-10 secretion in non-hospitalized 31 patients, while in hospitalized cases, increasing amounts of IFNg for CD4 + T cells and of IL-4 and 32 degranulation for CD8 + T cell were observed ( Figure 1B). In addition, N peptides induced higher 33 frequencies of antigen-specific CD8 + T cells degranulating in mild and non-hospitalized cases, 34 7 while S peptides stimulated IL-4 secretion mainly in the hospitalized groups ( Figure 1B). Overall, 1 our analyses indicated, on one hand, group-based differences, where a dominance of IL-4 and 2 IFNg SARS-CoV-2-specific responses were associated with disease severity and of IL-10 to 3 minor disease; on the other hand, we observed targeted protein based-differences, where M and 4 N peptides induced a Th1 profile exemplified by IFNg in CD4 + T cells and degranulation (CD107a) 5 in CD8 + T cells, respectively, and S peptides induced a biased Th2 profile exemplified by IL-4.

6
This pattern was shown in an exaggerated manner in the deceased patient, in which IL-10 7 responses were absent and IL-4 together with some IFNg dominated antigen-specific responses 8 ( Figure 1C).

10
Expression patterns of chemokine receptors associated to SARS-CoV-2-infected patients 11 Next, we aimed to determine if part of the specific T cell response was potentially migrating 12 towards the infected tissues by assessing the proportion of CCR7 and CXCR3 expression within 13 the same analyses. In peripheral blood, CCR7 distinguishes T cells homing to lymph node (LN) 14 when expressed, or effector memory (EM) T cell subsets migrating to tissues when absent 41 , 15 while CXCR3 may help define antiviral T cells infiltrating inflamed tissues, including the lung 16 parenchyma 25 . CD4 + T cells showed only two evident subsets in most patients based on CCR7 17 expression, since CXCR3 was homogeneously dimly expressed in these two subsets ( Figure   18 S2A) and no differences between the different study groups were observed ( Figure 2A). In 19 contrast, CD8 + T cells presented five subsets based on these chemokine receptors ( Figure S2A 20 and 2C), and significant differences among the groups were observed ( Figure 2B). Non-21 hospitalized patients showed increased frequencies of CCR7 h CXCR3 d CD8 + T cells, while severe 22 patients presented increased frequencies of EM CCR7 -CXCR3 + T cells (p=0.0012 and p=0.0034 23 respectively, Figure 2B and 2C). In fact, the accumulation of CCR7 h CXCR3 d CD8 + T cells indicated 24 good prognosis and negatively correlated with the number of days to discharge since symptoms 25 onset and with IL-6 levels at hospital entry ( Figure 2D), while the frequency of EM CXCR3 + CD8 + T 26 cells significantly correlated with disease severity parameters ( Figure 2E).

27
We then focused on the distribution of the net antigen-response for each peptide and 28 cohort among these CCR7/CXCR3 subsets. Overall, antigen-specific CD4 + T cells showed a 29 distinct pattern based on the function assessed: IFNg, degranulation (CD107a) and IL-4 were 30 significantly associated to EM CXCR3 + CD4 + T cells across the different groups and proteins, while 31 IL-10 was associated to the LN-homing fraction (CCR7 + CXCR3 + ) in response to M and N protein 32 peptides in the non-hospitalized group ( Figure 3A). Of note, in general, most individuals in the 33 hospitalized groups also showed this trend for the IL-10 response, although statistical significance 34 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 5, 2021. ; https://doi.org/10.1101/2020.12.02.20238907 doi: medRxiv preprint 8 was not reached as a group. Moreover, several subsets out of these antigen-specific CD4 + T cells, 1 mostly the ones secreting IFNg or IL-4, correlated with worse prognosis in the clinical parameters 2 assessed before, and some examples are shown in Figure 3B-3E. In general, stronger 3 associations were observed within the CCR7 + CXCR3 + subset, which correlated with severity, 4 except if this subset was secreting IL-10 against M peptides ( Figure 3F). Moreover, SARS-CoV-5 2 viral load was negatively associated with the overall capacity of EM CXCR3 + CD4 + T cells to 6 secrete IL-10 in response to TCR independent unspecific activation with PMA/Io ( Figure 3G).

7
In the analyses of the proportion of antigen-specific CD8 + T cells in each CCR7/CXCR3 8 subset, we did not consider the EM CXCR3subset, which represented <1% in most patients 9 ( Figure 2B). As expected, IFNg antigen-specific CD8 + T cells were more frequent among the 10 CXCR3 + subsets, with some individual exception, such as N-specific CCR7 + CXCR3 -T cells within 11 the non-hospitalized group ( Figure S4A). Further, IFNg secreting CCR7 h CXCR3 d CD8 + T cells in 12 response to N peptides correlated negatively with days to hospital discharge ( Figure S4B).

13
Degranulation, which was enhanced also after N stimulation, was unexpectedly detected in all 14 CD8 + CCR7/CXCR3 subpopulations ( Figure S5A). Though only in the LN-homing 15 CCR7 + CXCR3 + CD8 + T cell subset in response to S peptides, degranulation was associated with 16 higher viral load ( Figure S5B). With respect to IL-4 secreting antigen-specific CD8 + T cells, in 17 general these responses were more frequent in CCR7 + LN-homing subsets, and as mentioned 18 before, they increased with disease severity ( Figure S6A). Consequently, the frequency of IL-4 19 detected in response to M or S peptides in several of these fractions correlated with disease 20 severity ( Figure S6B). Remarkably, SARS-CoV-2-specific CD8 + T cells secreting IL-10 were 21 strongly represented among the CCR7 h CXCR3 d subset, reaching statistical significance in 22 response to any of the viral proteins within the mild disease cohort and in response to N peptides 23 in non-hospitalized patients, but not in the hospitalized group with severe disease ( Figure 4A).

24
Finally, we detected two additional correlations within the CD8 + T cell compartment that were of 25 interest: overall antigen-specific EM CXCR3 + CD8 + T cells correlated with higher viral loads if 26 responding to M peptides ( Figure 4B), while the same subset responding to N peptides negatively 27 correlated with IL-6 ( Figure 4C). All together, these results demonstrate individual migratory 28 patterns associated with a given function: whilst most of the functions assessed here were 29 associated with lung homing subsets (CCR7 -), IL-10-specific T cells expressed high levels of 30 CCR7. In fact, a strong association towards better disease prognosis was established for an 31 increased proportion of CCR7 h CXCR3 d CD8 + T cells, which represented a major constant source 32 of IL-10 in CD8 + T cells. Further, antigenic stimulation could be driving CCR7effector immune 33 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 5, 2021. 9 responses towards the lung, yet under uncontrolled disease progression, such effector functions 1 seemed to increase in LN-homing CCR7 + subsets.

4
We included caspase-3 in the flow cytometry panel as a surrogate marker of apoptotic cell death 5 activation 42 , which was quantified in both antigen-specific and bystander T cells from the different 6 subsets of the study groups ( Figure 1A). Overall expression of caspase-3 in response to 7 stimulation was increased in total CD4 + T cells of the severe group after M and PMA/Io stimulation 8 (p<0.0001 and p=0.032, respectively) and in CD8 + T cells after S stimulation in comparison to the 9 non-hospitalized group (p=0.0009) ( Figure 5A). Moreover, caspase-3 expression in CD4 + and 10 CD8 + T cells after stimulation with S-peptides positively correlated with baseline IL-6 and with the 11 number of hospitalization days for CD8 + T cells ( Figure 5B). In addition, the overall frequency of 12 CD4 + T cells expressing caspase-3 in response to PMA/Io positively correlated with viral load 13 ( Figure 5B). Further, an increased expression of caspase-3 within the CCR7 h CXCR3 d subset was, 14 in general, linked to the disease severity ( Figure 5C), reaching statistical significance after PMA/Io 15 stimulation when comparing the severe and the non-hospitalized groups (p=0.031; Figure 5C).

16
Consequently, those frequencies in response to stimulation (N, S or PMA/Io) correlated positively 17 with the number of days at the hospital and with baseline IL-6 ( Figure 5D). Expression of caspase-18 3 in other CCR7 + CD8 + T cell subsets in response to N and S peptides also correlated with IL-6 19 baseline levels in patients ( Figure 5D).

20
Regarding antigen-specific T cells we detected remarkable differences within the IL-10 21 secreting T cells. In this sense, increased expression of caspase-3 was distinguished in the 22 hospitalized severe group compared to the non-hospitalized in S-specific IL-10 + CD4 + T cells 23 (p=0.004), N-specific IL-10 + CD8 + T cells (p=0.016) and even in the overall IL-10 secretion 24 capacity in response to PMA/Io response (p=0.031 for CD4 + and p=0.006 for CD8 + ; Figure 5E).

25
Further, significant correlations between apoptosis in IL-10 antigen-specific T cells and several 26 clinical parameters supported these results ( Figure 5F). As for the other functions, we only 27 detected positive correlations for the expression of caspase-3 in baseline and N-specific CD107a +

28
CD4 + T cells with viral load and IL-6 levels, respectively ( Figure 5G). These results indicate 29 increased activation-induced cell death associated with viral replication and disease severity 30 affecting total CD4 + and CD8 + T cells, a phenomenon that seems to be modulated by the viral 31 protein targeted. Moreover, CD8 + CCR7 h CXCR3 d T cells, a major producers of IL-10, appeared to 32 be one of the most affected population.

34
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) not only migrate into the lung parenchyma but also persist as TRM, we measured their frequency 3 in lung biopsies of five patients. These patients, who strongly differed in their SARS-CoV-2 4 infection profile, successfully recovered and SARS-CoV-2 was not detected in the respiratory tract 5 by RT-PCR before they underwent thoracic surgery for different reasons. Briefly, HL24 patient 6 was a young-asymptomatic patient whose blood and lung samples were analyzed 21 days after 7 SARS-CoV-2 laboratory-confirmation by RT-PCR. In contrast, samples were analyzed between 8 6 and 10 months after initial SARS-CoV-2 RT-PCR confirmation for the two mild cases (HL52 and 9 HL65) and the two severe cases (HL27 and HL69). Of note, a more detailed COVID-19 clinical 10 history can be found in the methods section.

11
Antigen-specific T cell responses were analyzed in total lung CD4 + and CD8 + T cells and

22
of the variability observed among study patients, the frequency of specific T cell responses 23 increased with disease severity (from left to right; Figure 6C), while their magnitude was in general 24 low and less consistent for IL-4 or IL-10 responses ( Figure 6C and S7B-F). Importantly, a 25 consistent polyfunctional IFNg + CD107a + T cell response, which represented between 0.025 and 26 0.051% of all CD3 + T cells and was mostly associated with the TRM fraction (>75%), was detected 27 against N peptides in all patients except in the asymptomatic one (HL24; Figure 6D).

28
The comparison between the overall SARS-CoV-2-specific T cell responses detected in 29 lungs with the ones found in contemporary peripheral blood samples highlighted strong 30 differences between these two compartments ( Figure 7). For example, in the asymptomatic HL24-31 patient IFNg and IL-10 responses were more frequent in blood than in lung, IL-4 was absent, and 32 T cell degranulation was only detectable in lung ( Figure 7A). Importantly, this patient was closer 33 to the initial RT-PCR-based laboratory confirmation (3 weeks after), and his profile was more 34 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 5, 2021. 11 consistent with a non-hospitalized patient. A different pattern was observed for one of the severe 1 convalescent patients (HL27), who had persistent TRM in the lung overrepresented by S-specific 2 CD4 + T cells secreting IFNg, which represented up to 1.58% of the total CD4 + T cells, while only 3 0.082% of the circulating CD4 + T cells secreted IFNg and, in this case, in response to M peptides 4 ( Figure 7E). In fact, whereas all four functions were detected in T cells from lung after 6 months 5 since initial infection for this patient, they were barely detectable in blood. Overall, viral-specific 6 TRM responses were detected in all patients. However, no consistent patterns were observed 7 among patients in terms of viral proteins targeted and functions between blood and lung 8 compartments, except for the polyfunctional response detected in tissue against N peptides. Of 9 note, the asymptomatic patient had no detectable antibodies, whilst the two severe and two mild

33
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) 1 CoV-2 specific T cells 4,8,9,10,11,12,13,14 . Several of these studies have focused on defining the viral 2 proteins more often targeted by specific T cells, concluding that after infection a broad cell 3 response against multiple structural and non-structural regions of SARS-CoV-2 is detected in 4 most convalescent patients 4, 9, 12 . More recently, it has been reported that SARS-CoV-2 specific 5 T cells appear to be weaker and less frequent during acute infection 13 . In this sense, in our study, 6 while the frequency of responders based on CD4 + T cells specifically secreting IFNg was similar 7 to previous reports, they were indeed weak in terms of the amount of IFNg or cytotoxicity.   CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021. 1 apoptotic cell death, but viral proteins may also induce apoptosis in both antigen and non-antigen 2 specific T cells. While effector T cells with degranulation capacity are expected to be more 3 terminally differentiated and, consequently, may be more prone to activation-induced cell death, 4 the fact that other subsets (i.e. CCR7 h CXCR3 d /IL-10 secreting) were more affected is intriguing 5 and requires further study.

6
Critically, the only cytokine that was higher in non-hospitalized compared to hospitalized 7 groups was IL-12p70 ( Figure S1B), suggesting that, as occurs with other respiratory viruses 49 ,  CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021. 14 axis appears critical for the recruitment of CD4 + and CD8 + T cells that control influenza and 1 tuberculosis infection in the lung respectively 25, 51 . Consequently, lung-T cell recruitment may 2 partially contribute to the lymphopenia detected in patients 52 , where this antiviral response will 3 likely establish as resident memory cells. In fact, N-specific IFNg + CD107 + TRM were detected in all 4 four convalescent patients 6 to 10 months after infection.

5
Recent deep immune profiles of COVID19 patients have identified T-bet expression as a 6 transcriptional factor associated to patients with better prognosis 53 . Importantly, T-bet is not only 7 a key regulator of Th1 immune effector responses and CXCR3 expression, essential for effective 8 clearance of pathogens and maintenance of immunity 54 , but also crucial for migration, proliferation CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021. 1 severity, the fatal case had one of the lowest levels in plasma, which was accompanied by an absolute lack of IL-10 secretion by antigen-specific T cells.

3
TRM strategically residing in peripheral tissues are key to controlling mucosal infections 4 and providing rapid and durable immunity against reinfection 20, 63 . Previous studies in SARS-5 recovered patients already pointed towards persistence of a memory T cell response for up to 6 6 years after infection, and suggested vaccine-mediated induction of TRM as a long-term protection 7 strategy 5 . In concordance, a larger proportion of CD8 + T cell effectors with TRM characteristics were 8 present in bronchoalveolar lavages from patients with moderate infection compared to severe-9 infected patients 64 . We indeed report the existence of a high frequency of TRM in the lung of a 10 patient who was infected almost 6 months before, yet had a severe and durable infection (HL27).

11
In this patient, while all functions were represented in TRM in a remarkably higher proportion than 12 in blood, IFNg in response to S peptides dominated. In this sense, particularly high frequency of 13 spike protein-specific CD4 + T cell responses was observed in blood in patients who had recovered 14 from COVID-19 4, 9, 12 . Importantly, CD4 + T cells are necessary for the formation of protective 15 CD8 + TRM during influenza infection, and cytokines, such as IFNg, are necessary signals for this 16 process 20 . However, CD4 + T cells themselves can be cytotoxic and, actually, have been shown to 17 confer protection against influenza 20 . We also detected degranulation in response to viral peptides 18 in CD4 + and even more so in CD8 + T cells from the lung, which in the case of the asymptomatic 19 young patient were completely absent in blood. Lack of degranulation in blood from convalescent 20 patients has also been reported 12 . Further, the fact that the lung biopsy of this young 21 asymptomatic patient was at 3 weeks RT-PCR laboratory-confirmation of infection, suggests early 22 recruitment of cytotoxic T cells to the lung even in asymptomatic cases. Moreover, the most 23 frequent responses by circulating T cells from this asymptomatic patient were CD4 + and CD8 + T 24 cells secreting IL-10, which concurs with the dominating pattern in non-hospitalized patients 25 during acute infection.

26
We acknowledge that our study has several limitations, one being that sample size for the 27 different groups was small to be conclusive. However, this was compensated by a narrow window 28 of sampling during acute infection (7-16 days, post-symptoms onset) and by a comprehensive 29 clinical characterization to stratify patients to study groups. In this sense, multiple correlations 30 support our main findings and provide strength to our data, which is also largely supported by CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021. ; https://doi.org/10.1101/2020.12.02.20238907 doi: medRxiv preprint 16 different COVID-19 convalescent patients could be studied. While these patients are so far 1 scarce, the immune responses identified in those samples not only contributed to round out the 2 present report, but also represent the first evidence to our knowledge of persisting SARS-CoV-2-3 specific TRM in the lung. Disease severity during acute SARS-CoV-2 infection is associated with 4 strong peripheral T and B cell responses 4, 27 , which not only may relate to antigenic burden but, 5 also could potentially translate into a significant proportion of antigen-specific TRM in the lung once 6 the patient recovers. Remaining important questions concern the level of viral replication and 7 associated symptomatology that will stimulate an effective immune response at the respiratory 8 tract, and also, how quick this response will be established. However, the fact that an   These samples were considered to be from unexposed controls given that SARS-CoV-2 emerged 33 as a novel pathogen in December 2019 and these samples were largely collected before this 34 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021.
date. These donors were considered healthy in that they had no known history of any significant 1 systemic illnesses. The cohort of healthy donors includes 12 individuals.  Table   19 S1. According to disease severity patients, at the discretion of the treating physician, patients 20 were classified in three groups:

34
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
Lung biopsies were obtained from five patients recovered from SARS-CoV-2 infection who 3 needed a lung resection. Analyses were performed using healthy areas from the lung resection.         CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021. We would like to thank all the patients who participated in the study. We also thank Prof. Shawn

34
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021.

47
Associations with Age and Disease Severity. Cell 2020.

49
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021.

51
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021.

49
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021.

46
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021.        is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021.     is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted February 5, 2021. ; https://doi.org/10.1101/2020.12.02.20238907 doi: medRxiv preprint C Figure 4 A B C C    is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 5, 2021. ; https://doi.org/10.1101/2020.12.02.20238907 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 5, 2021. ; https://doi.org/10.1101/2020.12.02.20238907 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 5, 2021. ; https://doi.org/10.1101/2020.12.02.20238907 doi: medRxiv preprint