This study was approved by the National Commission of Ethics in Research (certificate CAAE: 31354720.0.0000.5188). All experiments were performed in compliance with the relevant regulations, institutional guidelines, and ethical standards of the Declaration of Helsinki. Informed consent was obtained from all the enrolled volunteers.

This study was conducted from May 2020 to May 2021 in Brazil. The study began around the time when the SARS-CoV-2 virus of the original lineage was still circulating and continued with the alpha and gamma variants being first detected in the end of 2020 (between October and December). These studies ended in May 2021, a few days after the delta variant was identified for the first time in Brazil (April 26, 2021) (17, 18). Sixteen nonvaccinated volunteers who presented COVID-19 symptoms between May 2020 and May 2021 and had subsequently recovered were recruited. The SARS-CoV-2 diagnosis was based on RT qPCR for COVID-19 during the acute phase and IgG serological tests (Euroimmun Anti‐SARS‐CoV‐2 assay, Perkin Elmer Company) after recovery. The recovered COVID-19 volunteers were assigned to either the Mild Recovered (did not need hospitalization, n = 9) or Severe Recovered (presented pulmonary symptoms requiring hospitalization, n = 7) group. For epidemiologic purposes, severe dyspnea, a respiratory rate of 30 or more breaths per minute, a blood oxygen saturation of 93% or less, and infiltrates in more than 50% of the lung were considered severe symptoms of COVID-19 (19). In addition, a healthy control group (HC, n = 8) was included. Volunteers in the HC group were not previously vaccinated against COVID-19, were reportedly asymptomatic for the last 10 weeks, were negative by certified SARS‐CoV‐2 antibody test (Euroimmun Anti‐SARS‐CoV‐2 assay Perkin Elmer Company), and had a negative RT qPCR test for SARS‐CoV‐2. Whole blood for flow cytometry analysis was collected in sodium heparin-coated vacutainers and gently agitated until processing. The blood samples were processed on the day of collection.

Quantitative real-time PCR (RT qPCR) was performed as described previously (20) at the time of sample collection for flow cytometry experiments. This procedure was essential to ensure that no SARS-CoV-2 infection or reinfection occurred in the HC and recovered volunteers. The RNA isolated from nasopharyngeal or oropharyngeal swabs was extracted (QIAprep&amp Viral RNA UM Kit, QIAGEN, USA) and amplified by one-step RT qPCR (SARS-CoV-2 N1+N2 Assay Kits, QIAGEN, USA).

For peptide design, we identify the most frequent HLA-class I/II alleles among the Brazilian population and most frequently in the world in the IEDB tools. We then performed MHC-I and II binding affinity and antigenicity predictions and analyzed the peptide molecular dynamics of the best-fitted MHC-I and II/protein of SARS-CoV-2 complexes. SARS-CoV-2 virus-specific CD4 and CD8 peptides ( Table 1 ) (patent number: BR 10 2022 005518 1) were synthesized (GenOne Biotechnologies, Rio de Janeiro, Brazil), lyophilized, and stored at -20°C for subsequent analysis. Subsequently, the peptides were resuspended in H₂O (MiliQ) and pooled into groups (Pool CoV-2 contained peptides from the spike protein and non-spike proteins, whereas Pool Spike CoV-2 contained peptides from the spike protein only, Table 1 ).

SARS-CoV-2 epitopes were predicted using protein sequences derived from the Brazilian SARS-CoV-2 sequences deposited in GenBank and IEDB analysis resources. CD4 SARS-CoV-2 epitope prediction was performed as previously reported, using the NetMHCIIpan 4.0 algorithm and CD4 immunogenicity prediction (IEDB), HLA-DRB*1:03:01, HLA-DRB*1:07:01, HLA-DRB*1:15:01, HLA-DRB*3:01:01, HLA-DRB*3:02:02, HLA-DR*4:01:01, and HLA-DRB*5:01:01. Due to extremely high heterogenicity of the Brazilian population, we carefully included the frequently identified HLA I and II in the Brazilian population along with the most frequently identified HLA populations in the world. CD8 SARS-CoV-2 epitope prediction was performed as previously reported using the NetMHCpan4.0 algorithm (HLA-A* 01: 01; HLA-A* 02: 01; HLA-A* 11: 01; HLA-A* 24: 02; HLA-A* 68: 01; HLA-A* 23: 01; HLA-A* 26: 01; HLA-A* 30: 02; HLA-A* 31: 01; HLA-B* 07: 02; HLA-B* 51: 01; HLA-B* 35: 01; HLA-B* 44: 02; HLA-B* 35: 03; HLA-C* 05: 01; HLA- C* 07: 01 e HLA-C* 15: 02). Despite inferring cellular populations of T lymphocytes, this strategy extrapolates to lymphocyte-like cells, such as natural killer (NK) cells, because a specific prediction program for NK cell populations is not yet available.

Plasma samples were collected after centrifugation of whole blood at 400 × g for 10 min at room temperature (25°C). The undiluted plasma was then transferred to cryotubes, aliquoted, and stored at -80°C for subsequent analysis.

PBMC from volunteers inserted in the HC and volunteers who had recovered from mild and severe COVID-19 (Mild Recovered and Severe Recovered groups, respectively) were obtained from heparinized venous blood using density gradient centrifugation (Ficoll-Paque ™ Plus, GE Healthcare, Life Sciences, Pittsburgh, PA, USA). PBMC were centrifuged for 40 min at 400 × g and washed three times with phosphate-buffered saline (PBS) before counting. PBMC were maintained in RPMI-1640 (Sigma-Aldrich) supplemented with antibiotics (penicillin, 200 U/mL; streptomycin, 0.1 mg/mL), 1 mM L-glutamine (1 mM), and 10% heat-inactivated AB Rh+ human serum (Sigma-Aldrich), hereafter referred to as medium. Cultures were set up at a concentration of 2.5 X 10⁵ cells in 96-well plates in the presence or absence of SARS-CoV-2 antigens and Staphylococcal enterotoxin B from Staphylococcus aureus (SEB, Sigma-Aldrich). PBMC were subjected to four different conditions: unstimulated (medium), stimulated with SARS-CoV-2 antigens (Pool Spike Cov-2 and Pool Cov-2, each at 1 µg/well), and stimulated with SEB (1 ug/well). The cells were incubated under 5% CO₂ in an incubator at 37°C for 16 h. Next, brefeldin-A (1 mg/mL, Sigma-Aldrich) was added and the samples were incubated under 5% CO₂ in an incubator at 37°C for 4 h.

Briefly, freshly isolated PBMC were plated at a concentration of 2.5 x 10⁵ cells per well in a 96-well U-bottom plate. Cocktails of Antibodies for extracellular staining were added and incubated for 30 min at 4°C. For extracellular staining, the following antibodies from BD Biosciences were used: extracellular PE-Cy5 conjugated anti-CD3 (clone UCHT1, isotype. Mouse BALB/c IgG1, κ, cat. 555334, specie. mouse), APC-Cy7 conjugated anti-CD4 (clone RPA-T4, isotype. Mouse IgG1, κ, cat. 557871, specie. mouse), PE-Cy7 conjugated anti-CD25 (clone M-A251, isotype. Mouse BALB/c IgG1, κ, cat. 557741, specie. mouse), PE-conjugated anti-CD39 (clone TU66, Isotype: Mouse IgG2b, κ, cat. 555464, specie. mouse), APC-conjugated anti-CD73 (clone AD2, isotype: Mouse IgG1, κ, cat. 560847, specie. mouse), PE-conjugated anti-PD1 (clone MIH4, isotype. Mouse IgG1, κ, cat. 557946, specie. mouse), PE-conjugated anti-LAP (clone TW4-2F8, isotype. Mouse BALB/c IgG1, κ, cat. 562260, specie. mouse), IgG isotypes control antibodies such as IgG3-FITC (clone J606, isotype. Mouse BALB/c IgG3, κ, cat. 555578, specie. mouse), IgG3-Alexa Fluor 647 (clone J606, isotype. Mouse BALB/c IgG3, κ, cat. 560803, specie. mouse), IgG1-APC-H7 (clone MOPC-21, isotype. Mouse IgG1, κ, cat. 560167, specie. mouse), IgG1-PE-Cy-5 (clone MOPC-21, isotype. Mouse IgG1, κ, cat. 555750, specie. mouse), and IgG1-PE-Cy-7 (clone MOPC-21, isotype. Mouse IgG1, κ, cat. 557872, specie. mouse). After incubation with antibodies for 30 min at 4°C, the cells were washed with 150 μL of PBS. The plate was centrifuged (8 min, 244 × g, 4°C), the supernatant was removed, and 100 μL of 4% formaldehyde and 100 μL of PBS were added to the wells. The plate was incubated at room temperature (25°C) for 20 min to fix the extracellular staining. After centrifugation (8 min, 244 × g, 4°C), the supernatant was discarded and the samples were washed with 150 μL of PBS. The plate was centrifuged yet again (8 min, 244 × g, 4°C) and the supernatant was discarded. For intracellular staining, the cells were permeabilized with 150 μL of permeabilization buffer (0.5% bovine serum albumin (BSA), w/v and 0.5% saponin, w/v in PBS) for 10 min at room temperature (25°C). After centrifugation (8 min, 244 × g, 4°C), the supernatant was removed, and intracellular staining was performed. The following conjugated antibodies from BD Biosciences were used to detect intracellular molecules: Alexa Fluor 647 conjugated anti-perforin (clone δG9, isotype. Mouse BALB/c IgG2b, κ, cat. 563576, specie. mouse), PE-conjugated anti-granzyme B (clone GB11, isotype. Mouse BALB/c IgG1, κ, cat. 561142, specie. mouse), PE-conjugated anti-CTLA-4 (clone BNI3, isotype. Mouse BALB/c IgG2a, κ, cat. 555853, specie. mouse), APC-conjugated anti-IL10 (clone JES3-19F1, isotype. Rat IgG2a, κ, cat. 554707, specie. rat), PE-conjugated anti-IL-17 (clone SCPL1362, isotype. Mouse IgG1, κ, cat. 560436, specie. mouse), Alexa Fluor 488 conjugated anti-FOXP3 (clone 236A/E7, isotype. Mouse BALB/c IgG1, κ, cat. 561181, specie. mouse), and IgG isotype control antibodies IgG1-PE (clone MOPC-21, isotype. Mouse IgG1, κ, cat. 559320, specie. mouse). All antibodies were used according to the manufacturer’s instructions. The plate was then incubated for 45 min at room temperature (25°C), and 150 μL of permeabilization buffer was added. After centrifugation, the supernatant was removed (8 min, 244 × g, 4°C). Finally, 200 μL of wash B (PBS/BSA) was added, and the samples were transferred to fluorescence-activated cell sorting (FACS) tubes and maintained at 4°C. At least 70,000 gated events were acquired using FACS CANTO II (BD Biosciences) and analyzed using the FlowJo v.10.8 software (BD, Ashland - USA).

All parameters evaluated in Tregs were analyzed using FlowJo software v.10.8 (BD, Ashland, USA). Limits for quadrant markers were set based on negative populations (cells), isotype controls, and fluorescence minus one (FMO), when appropriate. Five or six different fluorochromes were identified in each analysis. Four fluorochromes, anti-CD3 PE-Cy5, anti-CD4 APC-Cy7, anti-CD25 PE-Cy7, and anti-FOXP3 Alexa Fluor 488, were used to identify Tregs (CD3⁺CD4⁺CD25⁺FOXP3⁺). We first selected the total lymphocyte gate through size-FSC-A and granularity-SSC-A profiles, followed by singlet separation using the FSC-A × FSC-H parameters. Next, we set the CD3⁺CD4⁺ cells and identified CD25 + T cells within this subpopulation. Subsequently, CD3⁺CD4⁺CD25⁺FOXP3⁺ T-cells were detected. Two fluorochromes, anti-X PE and anti-Y APC (or Alexa Fluor 647), were used to assess the intracellular or surface markers. The following markers were used: IL-10, IL-17, latency-associated peptide (LAP), perforin, granzyme B, CTLA-4, PD-1, CD39, and CD73. The analytical strategy is shown in Supplementary Figures 1–5 .

The Shapiro–Wilk normality test was performed for all parameters evaluated using flow cytometry. If normally distributed, multiple comparisons were performed using two-way ANOVA and Tukey’s post-hoc test or one-way ANOVA and Tukey’s post-hoc test. Kruskal-Wallis with Dunn’s multiple comparisons test was performed if the data were not normally distributed. The immunological data between volunteers who had and had not developed certain symptoms in the Mild Recovered group was compared and analyzed based on the following clinical data: dyspnea, sore throat, nasal obstruction, myalgia, arthralgia, fatigue, diarrhea, anosmia, and ageusia. These comparisons were performed using one-way ANOVA and Tukey’s post hoc test or Kruskal-Wallis with Dunn’s multiple comparisons tests. Differences were considered statistically significant at p < 0.05. Statistical analyses were performed using GraphPad Prism 8.4.3. To create a heat map matrix regarding immunological marker expression in the HC, mild recovered, and severe recovered groups, we used Morpheus software and adjusted it to Z-score (22).

For this study, we recruited 24 volunteers and distributed them into three groups. In the HC group, 8 volunteers (4 male and 4 female) with a mean age of 35.62 ( ± 8.63) were enrolled. The Mild Recovered group included 9 volunteers (5 male and 4 female) with a mean age of 33.22 ( ± 6.70). Finally, 7 volunteers were enrolled in the Severe Recovered group (4 male and 3 female) with a mean age of 36.85 ( ± 7.64) ( Tables 2 – 4 ). No significant differences in age were found between the groups.

Most volunteers who experienced mild COVID-19 reported headache, runny nose, and fatigue ( Table 3 ). All volunteers in the severe recovery group experienced pulmonary infiltration, low oxygen saturation, and headache, whereas almost all had high fever, tachypnea, and fatigue ( Table 4 ). Regarding comorbidities, obesity was a significant chronic disease observed among 56% of the recovered COVID-19 cases.

Finally, volunteers were asked to self-identify themselves among the following ethnic–racial classifications: white, brown, black, yellow (Asian), and indigenous (Native American). Volunteers who self-classified themselves as brown and white comprised 58.33% and 33.33%, respectively, whereas those of black and indigenous self-classification comprised 4.1% each. Thus, the demographic of the HC group was white (62.5%), brown (25%), and indigenous (12.5%); that of Mild Recovered group was brown (88.88%) and white (11.11%), whereas that of Severe Recovered group was brown (57.14%), white (28.57%), and black (14.28%). The demographic and clinical details of the volunteers are shown in Tables 2 – 4 .

Analysis of PBMC from volunteers recovered from COVID-19 revealed a higher frequency of occurrence of Tregs in Mild Recovered group than in HC for non-stimulated samples ( Figure 1A ). In addition, pool CoV-2-stimulated samples from Mild Recovered and Severe Recovered groups presented more Tregs than did HC ( Figure 1A ). We also compared the frequency of occurrence of the studied lymphocyte subset among the Mild Recovered volunteers between those who had and had not experienced the following symptoms during acute COVID-19: dyspnea, myalgia, sore throat, arthralgia, fatigue, nasal obstruction, diarrhea, anosmia, and ageusia. Our data showed that those who did not experience nasal obstruction had more Tregs among the non-stimulated and Pool Spike CoV-2 stimulated PBMC than in those with this symptom ( Figure 1B ). A comparison of the frequency of occurrence of Treg among the Mild Recovered volunteers between those who had and had not experienced the other symptoms was also analyzed, but no significant results were found ( Supplementary Figure 1 ).

We also evaluated cytokine expression by Tregs in all the groups. Mild Recovered non-stimulated samples had higher levels of IL-10-producing Tregs than did the HC group ( Figure 2A ). Moreover, in Pool CoV-2-stimulated cells, the frequency of CD3⁺CD4⁺CD25⁺FOXP3⁺IL-10⁺ was lower in the Severe Recovered group than in the Mild Recovered group ( Figure 2A ). In the Mild Recovered group, the IL-10 expression by Tregs due to Pool CoV-2 peptide stimulus was higher compared with that due to Pool Spike CoV-2 stimulus ( Figure 2A ).

We also detected an elevated IL-17⁺ Treg frequency in PBMC from volunteers in the Mild Recovered group stimulated with SARS-CoV-2 related peptides compared with those from HC and Severe Recovered volunteers ( Figure 2B ). Severe Recovered group also presented higher IL-17 expression by Tregs than did the Pool CoV-2-stimulated PBMC in HC ( Figure 2B ). In addition, in HC samples incubated with SEB, IL-17⁺ Treg expression was more pronounced than in PBMC incubated with SARS-CoV-2-related peptides ( Figure 2B ). In contrast, Pool Spike CoV-2 reduced IL-17 expression by Tregs in the Severe Recovered group ( Figure 2B ). Finally, the levels of LAP⁺ Tregs did not differ between the HC, Mild Recovered, and Severe Recovered groups ( Figure 2C ). However, in the Pool CoV-2-stimulated PBMC, the frequency of Tregs expressing LAP was more accentuated than in Pool Spike CoV-2-stimulated samples ( Figure 2C ).

We also compared cytokine expression by Tregs among Mild Recovered volunteers between those who had and had not experienced certain symptoms of acute COVID-19. The Pool Spike CoV-2-stimulated samples of Mild Recovered volunteers who had not developed myalgia, fatigue, or nasal obstruction presented lower IL-10⁺ Treg frequency than did the unstimulated cells ( Figure 2D ). Diminished levels of IL-10 expression were observed in Pool Spike CoV-2-stimulated-Tregs compared with that of Pool CoV-2-stimulated Tregs in volunteers who had not developed myalgia, diarrhea, or anosmia during acute disease ( Figure 2D ). Finally, in unstimulated samples, the expression of IL-10 by Tregs was higher and lower in volunteers with arthralgia and fatigue, respectively, than in those who had not developed these symptoms ( Figure 2D ).

Interestingly, compared to unstimulated samples, Pool CoV-2 elicited the expression of IL-17 by Tregs in the Mild Recovered volunteers who did not present with myalgia and diminished the frequency of IL-17⁺ Treg in those who developed this symptom ( Figure 2E ). In the unstimulated samples, expression of IL-17 by Tregs from volunteers who did not have myalgia was lower than that by Tregs from those with this symptom ( Figure 2E ). Our data showed that IL-17 expression was less prevalent in volunteers with fatigue than in those without fatigue in samples stimulated with Pool CoV-2 ( Figure 2E ). For volunteers who experienced anosmia, the Pool Spike CoV-2 stimulus reduced the expression of IL-17 in Tregs compared with that of unstimulated samples ( Figure 2E ).

We also found elevated LAP⁺ Treg levels in Pool CoV-2-stimulated samples from Mild Recovered volunteers with dyspnea, whereas the opposite was observed in nasal obstruction ( Figure 2F ). In addition, unstimulated PBMC from Mild Recovered volunteers who had not experienced sore throat expressed a higher frequency of LAP-expressing Tregs than did the PBMC from those who had this symptom ( Figure 2F ). A comparison of cytokine expression among Mild Recovered volunteers between those who had and had not experienced the other symptoms was also analyzed, but no significant results were found ( Supplementary Figure 2 ).

The present study also analyzed the production of cytotoxic granules by Tregs. Pool CoV-2-stimulated samples presented a higher Perforin⁺ Treg frequency in the Mild Recovered group than in the other groups ( Figure 3A ). In contrast, in Pool Spike CoV-2 stimulus Tregs in the Mild Recovered group produced more perforin than that in the Severe Recovered group ( Figure 3A ). In addition, SEB-incubated PBMC from HC presented higher levels of Perforin⁺ Tregs than in cells from Severe Recovered group ( Figure 3A ). Finally, in the Mild Recovered group, the frequency of Tregs that produced perforin was lower in the SEB-stimulated cells than in the Pool CoV-2-stimulated samples ( Figure 3A ).

We also found elevated levels of granzyme B producing Treg in Mild Recovered group than in the HC in unstimulated or SARS-CoV-2-related peptide-stimulated samples ( Figure 3B ). When perforin and granzyme B co-expression was analyzed, we observed that SEB-stimulated PBMC from HC presented higher levels of Tregs co-expressing both granules than did the cells from the Severe Recovered group ( Figure 3C ). In the HC, Pool CoV-2 stimulus induced a higher frequency of perforin⁺granzyme B⁺ Tregs compared to that in the unstimulated samples ( Figure 3C ). In contrast, in the Severe Recovered group, the SEB-incubated samples presented lower perforin and granzyme B co-expression by Tregs than in the other conditions ( Figure 3C ).

We compared and analyzed the expression of perforin by Tregs among Mild Recovered volunteers between those who had and had not experienced certain symptoms and found higher Perforin⁺ Treg levels in unstimulated and Pool CoV-2-stimulated PBMC from volunteers who presented with dyspnea during acute COVID-19 than in those who did not ( Figure 3D ). We also detected higher levels of perforin produced by unstimulated Tregs from volunteers who had ageusia than in those who did not ( Figure 3D ). Elevated granzyme B expression was observed in Tregs in samples stimulated with Pool Spike CoV-2 in Mild Recovered volunteers who did not have arthralgia during acute COVID-19 than in those who did ( Figure 3E ).

In samples unstimulated and stimulated with peptides related to SARS CoV-2, the co-expression of cytotoxic granules by Tregs was higher in Mild Recovered volunteers who experienced dyspnea than in those who did not ( Figure 3F ). Higher levels of Tregs Perforin⁺Granzyme B⁺ cells were detected in Pool CoV-2- stimulated PBMC from volunteers who had arthralgia than in those who did not ( Figure 3F ). However, for those who developed nasal obstruction during acute COVID-19, we observed a mild Pool Spike CoV-2-induced co-expression of cytotoxic granules by Tregs ( Figure 3F ). Comparison of the cytotoxic granule expression among Mild Recovered volunteers between those who had and had not experienced other symptoms were also analyzed, but no significant results were found ( Supplementary Figure 3 ).

This study evaluated the expression of inhibitory receptors such as PD-1 and CTLA-4. Although CTLA-4 did not show significant changes in any of the studied groups ( Supplementary Table 1 ), the frequency of Tregs expressing PD-1 was lower in the Severe Recovered group than in the Mild Recovered group in PBMC stimulated with Pool Spike CoV-2 ( Supplementary Table 1 ). The Pool Spike CoV-2-stimulated samples in the Mild Recovered group presented a higher frequency of PD-1-producing Tregs than did those incubated with Pool CoV-2 peptide ( Supplementary Table 1 ).

Analyzing the PD-1⁺ Treg frequency between Mild Recovered volunteers with or without certain symptoms revealed that those who experienced dyspnea presented lower levels of PD-1⁺ Tregs in PBMC stimulated with Pool CoV-2 than with Pool Spike CoV-2 ( Figure 4A ). In addition, the volunteers who did not develop sore throat or nasal obstruction presented a higher frequency of PD-1⁺ Treg in unstimulated PBMC than in those stimulated with Pool CoV-2 ( Figure 4A ). Finally, unstimulated samples from volunteers who did not experience sore throat or fatigue presented a higher frequency of PD-1 producing Tregs than the cells from volunteers who developed these symptoms ( Figure 4A ).

The frequency of CTLA-4⁺ Treg among Mild Recovered volunteers between those who had and had not experienced certain symptoms during acute COVID-19 was compared, and we observed that those who had not experienced sore throat, nasal obstruction, diarrhea, and ageusia had a lower Pool Spike CoV-2 peptide-induced CTLA-4 expression in Tregs compared with that in unstimulated cells ( Figure 4B ). We observed higher levels of CTLA-4⁺ Tregs in PBMC stimulated with Pool CoV-2 than with Pool Spike CoV-2 among Mild Recovered volunteers who had not experienced fatigue ( Figure 4B ). Pool CoV-2-stimulated PBMC presented fewer Tregs expressing CTLA-4 in those who had developed anosmia when compared with that of the unstimulated samples ( Figure 4B ). We also observed a lower Treg CTLA-4⁺ frequency in Pool CoV-2-stimulated PBMC from Mild Recovered volunteers who had experienced fatigue, anosmia, and ageusia than in those who had not developed these symptoms ( Figure 4B ). The Treg CTLA-4⁺ levels in the Pool Spike CoV-2-stimulated PBMC were higher in volunteers who had developed diarrhea than in those who had not ( Figure 4B ). A comparison of the expression of the inhibitory receptors among Mild Recovered volunteers between those who had and had not experienced other symptoms was also analyzed, but no significant results were found ( Supplementary Figure 4 ).

Finally, the expression of CD39 and CD73 by Tregs was analyzed. As shown in Figures 5A, B ), no significant changes in the levels of Tregs expressing CD39 or CD73 were observed between the studied groups. However, when the co-expression of CD39⁺CD73⁺ was analyzed, Mild Recovered samples incubated with Pool CoV-2 had a higher frequency of CD39⁺CD73⁺ Treg expression than did the HC ( Figure 5C ). In the HC group, we also observed the high co-expression by Tregs in samples stimulated with SEB that with Pool CoV-2 ( Figure 5C ).

The comparison of ectonucleotidase expression among Mild Recovered volunteers between those who had and had not experienced certain symptoms showed a higher CD39⁺ Treg frequency in unstimulated PBMC from volunteers who had not developed myalgia and arthralgia than in those who had ( Figure 5D ). In addition, the levels of CD39 expression by Tregs in volunteers who had presented with fatigue and anosmia were more prevalent in the unstimulated samples than in Pool Spike CoV-2-stimulated samples ( Figure 5D ). The anosmia-positive volunteers were also marked by a lower expression of CD39 by Tregs in Pool CoV-2 stimulus compared to those that had not experienced this symptom ( Figure 5D ).

In Pool Spike CoV-2-incubated samples from Mild Recovered volunteers who had myalgia, arthralgia, and diarrhea, we observed a higher level of CD73⁺ Treg frequency than in those who had not presented these symptoms ( Figure 5E ). Volunteers who experienced arthralgia also presented higher CD73 expression by Tregs in unstimulated samples than in those without arthralgia ( Figure 5E ). Interestingly, CD39⁺CD73⁺ Tregs were more prevalent in all conditions for volunteers who had developed myalgia than for those who had not ( Figure 5F ). A similar result was observed for diarrhea with the exception of unstimulated samples; no significant differences were found among volunteers between those who had and had not developed this symptom ( Figure 5F ). No significant results were found in the comparison of ectonucleotidase expression among Mild Recovered volunteers between those who had and had not experienced other symptoms ( Supplementary Figure 5 ).