Persistence of SARS-CoV-2-specific B and T cell responses in convalescent COVID-19 patients 6–8 months after the infection

Background Monitoring the adaptive immune responses during the natural course of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection provides useful information for the development of vaccination strategies against this virus and its emerging variants. We thus profiled the serum anti-SARS-CoV-2 antibody (Ab) levels and specific memory B and T cell responses in convalescent coronavirus disease 2019 (COVID-19) patients. Methods A total of 119 samples from 88 convalescent donors who experienced mild to critical disease were tested for the presence of elevated anti-spike and anti-receptor binding domain Ab levels over a period of 8 months. In addition, the levels of SARS-CoV-2 neutralizing Abs and specific memory B and T cell responses were tested in a subset of samples. Findings Anti-SARS-CoV-2 Abs were present in 85% of the samples collected within 4 weeks after the onset of symptoms in COVID-19 patients. Levels of specific immunoglobulin M (IgM)/IgA Abs declined after 1 month, while levels of specific IgG Abs and plasma neutralizing activities remained relatively stable up to 6 months after diagnosis. Anti-SARS-CoV-2 IgG Abs were still present, although at a significantly lower level, in 80% of the samples collected at 6–8 months after symptom onset. SARS-CoV-2-specific memory B and T cell responses developed with time and were persistent in all of the patients followed up for 6–8 months. Conclusions Our data suggest that protective adaptive immunity following natural infection of SARS-CoV-2 may persist for at least 6–8 months, regardless of disease severity. Development of medium- or long-term protective immunity through vaccination may thus be possible. Funding This project was supported by the European Union’s Horizon 2020 research and innovation programme (ATAC, no. 101003650), the Italian Ministry of Health (Ricerca Finalizzata grant no. GR-2013-02358399), the Center for Innovative Medicine, and the Swedish Research Council. J.A. was supported by the SciLifeLab/KAW national COVID-19 research program project grant 2020.


SUMMARY
Background: Monitoring the adaptive immune responses during the natural course of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection provides useful information for the development of vaccination strategies against this virus and its emerging variants. We thus profiled the serum anti-SARS-CoV-2 antibody (Ab) levels and specific memory B and T cell responses in convalescent coronavirus disease 2019 (COVID-19) patients. Methods: A total of 119 samples from 88 convalescent donors who experienced mild to critical disease were tested for the presence of elevated anti-spike and anti-receptor binding domain Ab levels over a period of 8 months. In addition, the levels of SARS-CoV-2 neutralizing Abs and specific memory B and T cell responses were tested in a subset of samples. Findings: Anti-SARS-CoV-2 Abs were present in 85% of the samples collected within 4 weeks after the onset of symptoms in COVID-19 patients. Levels of specific immunoglobulin M (IgM)/IgA Abs declined after 1 month, while levels of specific IgG Abs and plasma neutralizing activities remained relatively stable up to 6 months after diagnosis. Anti-SARS-CoV-2 IgG Abs were still present, although at a significantly lower level, in 80% of the samples collected at 6-8 months after symptom onset. SARS-CoV-2-specific memory B and T cell responses developed with time and were persistent in all of the patients followed up for 6-8 months.
Conclusions: Our data suggest that protective adaptive immunity following natural infection of SARS-CoV-2 may persist for at least 6-8 months, regardless of disease severity. Development of medium-or long-term protective immunity through vaccination may thus be possible.

Context and significance
Studies on the longevity of the adaptive immune response in convalescent coronavirus disease 2019 (COVID-19) patients may facilitate the understanding of how immune protection develops and persists during the natural course of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection and thus provide useful information for the evaluation of vaccines against this emerging virus. Although the serum levels of anti-SARS-CoV-2 immunoglobulin G (IgG) antibodies declined significantly 6 months after infection, virusspecific T and/or memory B cell responses developed and were maintained at a relatively high level 6-8 months after the onset of symptoms in the majority of convalescent patients. These data suggest that protective adaptive immunity following natural infection of SARS-CoV-2 may persist for at least 6-8 months.

INTRODUCTION
The emergence and spread of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has led to a pandemic with a major impact on global health. The genome of this novel coronavirus encodes 4 major structural proteins, including the spike (S) protein, nucleoprotein (N), membrane (M) protein, envelope (E) protein, and other proteins (ORF3a, ORF7a). 1 A robust adaptive immune response with presence of S-specific neutralizing antibodies (Abs), memory B cells, and circulating follicular helper T cells have been found in patients who have recovered from the infection. [2][3][4] Furthermore, a large number of S protein receptor-binding domain (RBD)-specific monoclonal Abs have been isolated from convalescent patients and been tested in animal models 3,5,6 and clinical trials for the development of potential passive immunotherapy. [7][8][9] It is, however, still unclear how long the adaptive immunity to SARS-CoV-2 lasts after the natural infection, and furthermore, whether the Ab titer is a marker for protective immunity. A relationship between a humoral immune response to SARS-CoV-2 infection and protection against reinfection has been shown in rhesus macaques, 10 but remains to be determined in humans. While a recent study in Iceland showed that the Ab response was maintained in 90% of convalescent patients for >4 months after onset of the disease, 11 other studies have suggested a rapid decay of anti-SARS-CoV-2 immunoglobulin G (IgG) in individuals with mild illness. 12,13 Nevertheless, long-lived memory T and B cells could be present and reactivated following a second exposure, thus providing immune protection.
Studies on the longevity of the adaptive immune response in convalescent coronavirus disease 2019 (COVID-19) patients may facilitate the understanding of how immune protection develops and persists during the natural course of SARS-CoV-2 infection and provide useful information for the development and evaluation of vaccines against this emerging virus. In this study, we aimed to assess the dynamics and longevity of the SARS-CoV-2-specific immune responses in COVID-19 patients with a broad spectrum of disease scores. The levels and Ig class of SARS-CoV-2-specific Abs and development of memory B and T cells were evaluated in samples collected from 88 patients at different time points (TPs) during a period of 8 months following initial symptoms.

RESULTS
The dynamics of the anti-SARS-CoV-2 Ab response in COVID-19 patients To evaluate the Ab response to SARS-CoV-2, 119 serum or plasma samples from 88 COVID-19 patients (78 from Italy and 10 from Sweden; Figure 1; Tables 1 and S1) were tested by an in-house enzyme-linked immunosorbent assay (ELISA) for the presence of anti-S and anti-RBD Abs (for details, see Method details). We examined the SARS-CoV-2-specific IgM, IgA, and IgG Abs in 55 samples from COVID-19 patients collected during the early phases of recovery (between 7 and 28 days after the onset of disease symptoms) and 108 historical controls (samples collected before the SARS-CoV-2 pandemic). Significantly higher levels of anti-S and anti-RBD IgM, IgA, and IgG Abs (p < 0.0001 for all groups) were detected in patients as compared to historical controls (Figures 2A-2C). Anti-S IgM, IgA, and IgG levels were increased in 76%, 78%, and 85% of patients, respectively. A similar proportion of patients had elevated anti-RBD IgM, IgA, and IgG levels, with 67%, 62%, and 85% being positive, respectively. Titers of anti-S and anti-RBD Abs were highly correlated for all isotypes (r = 0.88 for IgM, r = 0.77 for IgA, r = 0.95 for IgG) (Figures 2D-2F).
Based on the symptoms presented at the time of COVID-19 diagnosis, patients were given a disease score ranging from mild to moderate to severe to critical (see Method details). Analysis of anti-SARS-CoV-2 Ab levels did not show statistically significant differences in patients presenting with severe and critical disease scores compared to mild or moderate disease groups (p = 0.1444 and p = 0.2943 for anti-S IgG, p = 0.1203 and p = 0.4672 for anti-RBD IgG) ( Figures 3A  and 3B). Notably, in 6 patients (11%), rather low levels or even an absence of anti-S and anti-RBD IgM, IgA, and IgG Abs were observed. Sera from these individuals were obtained at median day 10.5 after symptom onset (range 7-22). A second sample taken at a later TP was available for 2 of these patients (86 and 226 days after symptom onset), in which both had become anti-S and anti-RBD IgG + . For the other 4 samples (7%), no second sample was available for analysis. These 4 individuals had a higher median age (79 years) compared to the entire patient cohort (62 years), had severe disease scores, and 2 of them later died from COVID-19 complications. When patients were divided based on sex, no statistically significant differences were observed in anti-S and anti-RBD Ab levels for all isotypes, except for anti-RBD IgG, in which significantly higher levels were present in males compared to females with severe/critical disease scores (p = 0.0306) ( Figures 3C and 3D).
To examine the longevity of the anti-SARS-CoV-2 Ab response, we subsequently analyzed all 119 samples from the 88 patients collected at different TPs (7-240 days after symptom onset). Anti-S and anti-RBD Ab levels were significantly increased already 7-14 days after symptom onset (p < 0.0001 for all isotypes) and reached maximum at days 15-28 ( Figure 4) (days 7-14 to days 15-28, p = 0.0087 and p = 0.0309 for IgM, p = 0.0080 and p = 0.0029 for IgA, p = 0.0159 and p = 0.0049 for IgG). After day 28, a significant decrease in anti-S and anti-RBD IgM and IgA Ab levels was observed (days 15-28 to days 29-90, p = 0.0184 for anti-RBD IgA; days 15-28 to days 91-180, p < 0.0001 for anti-S and anti-RBD IgM and IgA; days 15-28 to days 181-240, p < 0.0001 for anti-S and anti-RBD IgM and IgA). No significant decrease in anti-S and anti-RBD IgG levels  was present by days 91-180 (days 15-28 to days 91-180, p = 0.1847 and p = 0.0544, respectively); however, a significant decline was observed by days 181-240 (days 15-28 to days 181-240, p = 0.0003 and p = 0.0002, respectively). Importantly, a prominent anti-SARS-CoV-2 IgG response was still present in 80% (12/15) of the patients who were followed 181-240 days (6-8 months) after the onset of symptoms. These patients had mild (n = 5), moderate (n = 2), and severe (n = 5) disease scores at the time of diagnosis.
To further evaluate the dynamics of the anti-SARS-CoV-2 Ab response, we compared the Ab levels in paired samples from 27 patients. The first sample (TP1) was taken at median day 21 (range 7-64) after the onset of symptoms, and the paired second sample (TP2) was taken at median day 126 (range 57-234). This analysis showed a significant decrease in anti-S IgM and IgA levels (p < 0.0001 and p = 0.0008, respectively), as well as in anti-RBD IgM and IgA levels (p = 0.0002 and p = 0.0052, respectively) in the paired second samples ( Figure 5). No significant decline in anti-S and anti-RBD IgG levels, however, was observed (p = 0.1551).   response was induced in a majority of COVID-19 patients in our study cohort, and although IgM and IgA Abs rapidly declined, IgG Abs and plasma neutralizing activity remained present, albeit at lower levels, for at least 6-8 months after diagnosis.

Induction of SARS-CoV-2-specific memory B and T cells
To address the question of whether SARS-CoV-2-specific memory B and T cells were generated and how long the B and T cell-mediated responses persist in  Table S1). A total of 17 patients were sampled at a single time, whereas 6 patients had samples collected at 2 TPs and 1 at 3 TPs. We further divided samples into 3 groups based on the TPs (collected at 14-35 days, 85-180 days, and 181-236 days post-onset of symptoms, $2-4 weeks, 3-6 months, and 6-8 months, respectively). No, or a negligible number of B cells secreting RBD-specific IgG Abs were detected by ELISpot (for details, see Method details) in samples from 4 healthy individuals and 7 pre-pandemic buffy coats. Using the highest value observed from all of the negative controls as a cutoff, RBD-specific IgG-producing B cells were detected in 33% (2/6), 93% (14/15), and 100% (11/ 11) of the patient samples collected at 2-4 weeks, 3-6 months, and 6-8 months after the onset of symptoms, respectively ( Figure 6A). Compared to the early TP samples, the number of RBD-specific IgG-producing cells was significantly higher in the latter TP groups (p = 0.0001 and p = 0.0006, for 3-6 or 6-8 months versus 2-4 weeks, respectively) ( Figure S2A; Table S2). The B cell response seemed to reach a maximum at 3-6 months post-infection, although there was no significant decrease observed at 6-8 months ( Figures 6B and S2A, p = 0.1982; Table S2). Furthermore, no significant difference was observed for the number of specific B cells among patients with different disease scores ( Figure S3A; Table S3). Thus, although the decline in anti-RBD IgG levels was observed over time, SARS-CoV-2-specific memory B cells developed and remained present in almost all of the patients followed up until the latest date of the study, regardless of the disease severity ( Figures 6A, 6B, S2A, S2B, S3A, and S3B).  Figure 2). Significant differences between the groups were marked by bars on the top. Mann-Whitney U test. *p % 0.05, **p % 0.001, ***p % 0.001, and ****p < 0.0001. See also Figure S1.
Furthermore, while no or a negligible number of interleukin-2 (IL-2), interferon-g (IFN-g), or IL-2/IFN-g-producing T cells against 3 SARS-CoV-2-derived peptide pools were detected by a FluoroSpot assay (for details, see Method details) in the negative controls, such T cells were observed at a level above the cutoff in 17%-100% of the patient samples tested, depending on the peptide pool tested (S1, S2 N, or S N M O protein derived) and in which cytokines were analyzed ( Figures  6C-6E). In 2 patients, a high number of IFN-g-and/or IL-2-producing T cells was also detected in the control culture without adding the SARS-CoV-2 peptides, which may suggest ongoing inflammation. Overall, a T cell response against at least one of the SARS-CoV-2 peptide pools was detectable in all of the patients (n = 6) tested at the early TP (2-4 weeks), and the response was maintained in the vast majority of patients analyzed at the later TPs (93% [14/15] and 100% [11/11] at 3-6 or 6-8 months after the onset of symptoms, respectively). Notably, the only patient who had no T cell response at 4 months had a detectable memory B cell response. Compared to the early TP samples, the number of S1-specific IL-2, IFN-g, and IL-2/IFN-g-producing T cells was significantly higher in the later TP samples, especially those collected at 6 Table S3). While anti-RBD IgG levels decreased over time, SARS-CoV-2-specific

OPEN ACCESS
T cells, in addition to specific memory B cells, were also generated and remained present in the vast majority of patients followed to date, regardless of disease severity ( Figure 6F).
To further investigate the dynamics of virus specific B and T cell responses, patients with samples from at least 2 TPs were analyzed separately ( Figure 6G). In 6 of the 7 patients, there was an increase in the number of virus-specific B and T cells at the second TP (TP1 = 14-116 days, TP2 = 124-236 days; Figure 6G). In the remaining patient (Convalescent Patient [CP]19), there was a slight decrease in virus-specific B and T cells at TP2 (TP1 = 122 days, TP2 = 228 days).
Finally, for patients who have been evaluated for SARS-CoV-2-specific adaptive immunity in all assays (i.e., virus-specific Abs, memory B and T cells), we have plotted normalized log2 values of levels of anti-RBD-specific IgM, IgA, and IgG, number of RBD-specific IgG-producing B cells, and S1 peptide pool-specific IL-2 and/or IFN-g-producing T cells for each individual ( Figure S4). Samples with different TPs from the same individual were plotted together with different color codes ( Figure S4). Although most samples presented a unique pattern, 6 and 17 of the 24 patients have developed 2 or all ''arms'' of adaptive immunity at variable levels, in at least 1 TP, and the remaining single patient had only memory B cells detected (CP10, sampled at 126 days). In patients with samples from >1 TP, a clear shift from the production of specific Abs, including IgM/IgA at the early TP, to the generation of memory B and T cells at the later TP(s) was observed (Figure S4). Although having individual differences, SARS-CoV-2-specific memory B and T cells were generated and maintained in the vast majority of tested convalescent COVID-19 patients, regardless of the initial disease severity, suggesting that the adaptive immunity against SARS-CoV-2 during the natural course of infection is maintained at least for 6-8 months.

DISCUSSION
In the present study, we measured anti-S and anti-RDB IgM, IgA, and IgG Ab levels using normalization against a serially diluted highly positive reference serum pool and by setting a cutoff value based on historical control samples. Our data showed that an anti-SARS-CoV-2 Ab response was present in the majority of COVID-19 patients as early as 2 weeks after onset of symptoms, and the level of anti-S and anti-RDB IgG remained relatively stable up to 6 months after diagnosis, followed by a significant decline at months 6-8, while a decrease in anti-S and anti-RDB IgM and IgA levels was observed already between 1 and 3 months after the onset of disease. Our results are in line with previous studies showing a similar longevity and pattern of anti-SARS-CoV-2 Ab responses, with Ab levels reaching a peak at 23 days following symptom onset and being maintained for at least 4 months, 11,14-20 yet contradictory to others, in which a low prevalence and rapid decay (within 3 months) of anti-SARS-CoV-2 Abs in COVID-19 patients with either mild or severe disease were observed. 12,21 In agreement with other reports, we also observed higher anti-RBD IgG Ab titers in males who were more severely affected by SARS-CoV-2 infection, while 7% of patients with a severe disease score in our study did not develop or had extremely low levels of Abs, suggesting that they mounted a weaker antiviral immune response. 11,13,21 Higher levels of specific Abs, or the other end of spectrum (i.e., absence of specific Abs), in individuals with severe diseases and in male patients may be due to a higher viral load, longer duration of viral shedding, 22,23 or other host/genetic factors. 24,25 Previously reported conflicting findings in the prevalence and longevity of anti-SARS-CoV-2 Ab response may result from an absence of a standard assay to measure anti-SARS-CoV-2 Abs, as a majority of reported studies are based on using different types of SARS-CoV-2 antigens (RBD, S or N proteins). The discrepant results between studies could also be explained by the differences between COVID-19 cohorts, as the number of patients included the sample collection TPs, the treatment used, and the age and sex of study subjects all could affect the analysis. 26 Furthermore, the use of negative controls and setting of cutoffs may also contribute to the differences among studies. In our study, we have tested patients with both mild and severe disease, using the RBD and S antigens and compared the patients to historical controls (pre-pandemic samples) and set the cutoffs accordingly. Under these conditions, we demonstrated that the level of anti-S and anti-RDB IgG remained relatively stable-in other words, no significant decline up to 6 months after diagnosis, regardless of disease severity. Although the protective role of Abs against SARS-CoV-2 remains uncertain, in agreement with previous reports, we observed that the levels of anti-RBD Abs correlated with the plasma neutralization activity. 17,19,27 A recent study suggested that despite the anti-RBD, S2, and N protein Ab levels declining over the 6-to 8-month period, the neutralizing titers showed an insignificant decay. 28 In our study, the decline in anti-RBD and anti-S IgG Abs observed after 6 months was paralleled by a ll OPEN ACCESS corresponding decrease in neutralization titers, which is in agreement with another recent report. 32 It is important to point out that in both cases, the anti-RBD Abs and plasma neutralizing activity were reduced but remained detectable 6 months after the infection. While our study cohort is relatively small, the inclusion of Italian and Swedish patients with a different spectrum of disease and the long follow-up time (up to 6-8 months) can help to solve the current debate about the persistence of SARS-CoV-2-specific Abs. It has been reported that Abs against the 2 other coronaviruses, SARS-CoV and Middle Eastern respiratory syndrome (MERS)-CoV, could still be detected 1-3 years after infection onset, 22 suggesting that SARS-CoV-2-specific Abs may be present for an even longer period than we have observed thus far.
Studies reported up to now have mainly been focused on the longevity of the specific antiviral Ab response. However, the development of memory B and T cell is critical for long-term protection, and the longitudinal dynamics of these memory cells remain poorly resolved. Our results show that the majority of patients, irrespective of disease severity, can mount specific memory B cell responses, which remain present at least 6-8 months post-symptom onset. These findings are consistent with recent studies (including preprints) predominantly based on mild to moderate disease cohorts, and where flow cytometry assays were used to measure the RBD-specific memory B cells. 18,20,[29][30][31][32] In addition, we showed that these memory B cells could secrete RBD-specific IgG Abs following stimulation.
Previously, it has been shown that S1 and other SARS-CoV-2 protein-derived peptides induce specific T cell responses in patients with mild to severe disease 1-3 months post-symptoms. 2,29,[33][34][35] More recently, it was also reported (including in preprints) that SARS-CoV-2-specific T cells are maintained at least 6 months following primary infection in all tested COVID-19 patients in predominantly mild to moderate disease cohorts. 20,[36][37][38] Our results confirm and extend earlier findings, showing that SARS-CoV-2-specific IL-2 and/or IFN-gÀproducing cells are present in a vast majority of samples collected from patients with disease scores ranging from mild to severe. Furthermore, we showed that specific T cell responses were not only detectable but also appeared to be maintained 6-8 months after the onset of disease (the longest follow-up time in the literature). This is somewhat in contrast to recent studies showing that the T cell response declined over the 6-to 8-month period in asymptomatic as well as symptomatic COVID-19 patients. 20,38 The difference could be explained by the small number of samples included in our study, especially in the early disease phase, and/or the higher proportion of aged and severe patients in our cohort, as SARS-CoV-2-specific T cell response may be stronger and persist longer in patients with severe disease due to a higher viral load and/or longer persistence of viral antigens. 20,38 Some studies also noted that T cell responses directed against the S and/or M protein of SARS-CoV-2 are present in 25%-50% of unexposed healthy blood donors, which is consistent with a high degree of potentially cross-reactive T cell immunity in the general population. 2,34,35 The reason we did not detect a significant number of cross-reactive T cells against the virus in unaffected controls may be due to different experimental conditions, including the negative controls used and a longer in vitro stimulation with the peptides in some studies 39 ; however, cross-reactive T cells have also been observed in studies using in vitro stimulation for a shorter time (9-24 h). 2,34,35 The detection of memory S1-specific T cells marked by the production of IL-2 and IFN-g, and the dual production of those cytokines is indicative of induction of T cells with both effector and proliferative potential in vivo. IFN-g-producing T cells is a hallmark of immunity against intracellular pathogens, and although it was not tested in our study, SARS-CoV-2-specific IFN-g-producing T cells were previously shown to be of the CD4 + (T helper 1 [Th1]-like) or CD8 + cytotoxic phenotype. 2,34 It was shown that convalescent patient donors with undetectable Abs against the S1 protein of SARS-CoV-2 had T cell responses more strongly directed against the M than the S1 protein. 40 Furthermore, a Th1-biased cellular immune response of S-specific IFN-g + CD4 + T cells to pooled S peptides was detected in a majority of monkeys vaccinated with S protein and was associated with the induction of specific and neutralizing anti-S Abs. 41 Our results thus suggest that the use of S protein as an immunogen for vaccination has the potential to induce memory T and B cells specific for the S protein and RBD in humans.
Importantly, although serum IgG Ab levels decreased over time in our study, specific memory B and T cells were maintained over the 6-to 8-month period and could contribute to protection from SARS-CoV-2 reinfection. 29,31 However, the detection of Abs to SARS-CoV-2, including neutralizing Abs, as well as memory B cells and T cells over a long period does not necessarily indicate protective and long-term immunity, and a correlate of protection still needs to be established. Studies on common human coronaviruses show that neutralizing Abs are induced, and a recent report showed that the duration of protective immunity against the common cold coronavirus may last 6-12 months. 42 Reinfections with all seasonal coronaviruses usually occur within 3 years, 22 and repeat infections are generally associated with milder symptoms and a lower viral load. 22,43 Single intravenous administration of neutralizing Abs against the spike protein in patients with mild or moderate COVID-19 has been shown to reduce the viral load and a shortened length of hospitalization. 8 Furthermore, infection with SARS-CoV-2 and vaccine against the S protein can protect rhesus macaques from a challenge infection. 10 It is thus likely that Abs and cell-mediated immunity will decrease the risk of reinfection and attenuate the severity in case of reinfection. We thus plan to expand our cohort and follow it over longer intervals of time to evaluate the maintenance of immunological memory.
In conclusion, the presence of high levels of SARS-CoV-2-specific memory B and T cells in the majority of patients 6-8 months after infection, suggests that immunity after natural infection may be at least transiently protective and that the development of medium-or long-term protective immunity through vaccination may thus be possible.
The discovery of T cell reactivity against S protein epitopes and Abs against the RBD domain suggests that vaccine development using the S protein to induce Abs that target the RBD is a plausible approach. 33 It is, however, unclear at the moment whether the neutralizing Abs and B and T cell responses developed during earlier waves of infection and current vaccination strategies are protective for the emerging SARS-CoV-2 variants such as 501Y.V2. 44 To meet the urgent need for SARS-CoV-2 vaccine development and evaluation, we propose that in addition to the analysis of specific Ab responses and their longevity, the measurement of specific memory B and T cells, the main components of long-term immunity, as well as correlation with protection from reinfection (with the same or different variants), should be considered.

Limitations of study
Our study has certain limitations, including the relatively small patient cohort size (119 samples from 88 patients) and the low number of samples analyzed for T and B cells (32 samples, including paired samples from 7 individuals). Furthermore, the first TP samples mainly included those during the early phase of convalescence ll OPEN ACCESS (days 14-28 after onset of symptoms), and fewer samples were analyzed for B and T cells between 1 and 3 months after diagnosis. The level of specific T cells measured at an early TP was lower than that previously measured in convalescent patients 1-2 months after onset of symptoms, suggesting that the number of memory cells may increase during this period. 34,45 In addition, due to the experimental settings, it was not possible to distinguish whether the T cells measured at early TPs were effector or memory cells. To assess the development and persistence of memory B and T cells, a higher number of samples should be evaluated and followed at multiple TPs during the study period, and cell phenotyping should be performed.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

ACKNOWLEDGMENTS
We thank all of the patients, blood donors, and clinicians for their contributions.

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Qiang Pan-Hammarströ m (qiang.pan-hammarstrom@ki.se).
Materials availability RBD and S1-S2 proteins can be generated and shared on a collaborative basis. Blood samples (convalescent) Italy, this paper Table S1 Blood samples (convalescent, healthy) Sweden, this paper Table S1 Serum (convalescent, healthy) Sweden, this paper Table S1 Buffy coat (healthy donor) Sweden, this paper  45 The SARS-CoV-2 S2 N defined peptide pool contains 41 synthetic peptides binding to human HLA derived from the S and N proteins of the SARS-CoV-2 virus (#3620-1, Mabtech AB). 53 Results of ELISpot and Fluorospot assays were evaluated using an IRIS-reader and analyzed by the IRIS software version 1.1.9 (Mabtech AB). The results were expressed as the number of spots per 300 000 seeded cells after subtracting the background spots of the negative control. The cut-off value was set at the highest number of specific B-and T cell spots from the negative controls.

QUANTIFICATION AND STATISTICAL ANALYSIS
Mann-Whitney U test was used for comparisons between groups in anti-SARS-CoV-2 antibody levels and numbers of specific memory B-and T cells. Correlation analysis was performed using Spearman's rank correlation. A Wilcoxon signed-rank test was used for comparison paired samples. All analyses and data plotting were performed using GraphPad or R version 3.6.1.