Distinctive Dynamics and Functions of the CD4⁺CD25⁺FOXP3⁺ Regulatory T Cell Population in Patients with Severe and Mild COVID-19

Since the first report of COVID-19 caused by SARS-CoV-2 infection in December 2019 in Wuhan, China (1, 2), 6 million cases of mortality had been reported globally as of April 10, 2022 (3). SARS-CoV-2 infection manifests as various forms of clinical illness, from asymptomatic infection to severe/critical illness (4, 5). Severe/critical COVID-19 is mediated by exaggerated inflammatory responses and accompanied by acute respiratory distress syndrome and multiple organ dysfunction.

Since the emergence of COVID-19, many studies have examined the host immune responses against SARS-CoV-2, including neutralizing Abs, CD4⁺ and CD8⁺ T cell responses (6–12), and inflammatory responses, including hyperactivation of monocytes/macrophages and neutrophils and the production of proinflammatory cytokines (13–16). However, the changes in suppressive immune cells during COVID-19 remain to be fully elucidated.

CD4⁺CD25⁺FOXP3⁺ regulatory T (T_REG) cells are representative suppressive immune cells that maintain homeostasis in the immune system (17–19). T_REG cells exert immunosuppressive functions via CTLA-4, which competes with the CD28 expressed by effector T cells for binding to B7 (20–22), and immunosuppressive cytokines, such as IL-10, IL-35, and TGF-β (23–26). In humans, the CD4⁺CD25⁺FOXP3⁺ T_REG cell population includes distinct subpopulations, such as CD45RA⁻FOXP3^hi activated T_REG cells and CD45RA⁺FOXP3^lo resting T_REG cells (27). CD45RA⁻FOXP3^hi activated T_REG cells exert stronger suppressive functions than CD45RA⁺FOXP3^lo resting T_REG cells but are vulnerable to apoptosis (27, 28).

Changes in the T_REG cell population have been reported in various viral diseases, particularly viral hepatitis. In patients with hepatitis C virus (HCV) infection, the frequency of T_REG cells increases in the peripheral blood (PB) and liver (29–31), and the increase in T_REG cell frequency has been related to CD8⁺ T cell dysfunction and high viral load, indicating that T_REG cells contribute to viral persistence by suppressing T cell responses during HCV infection. In contrast, the frequency of T_REG cells decreases in the PB via Fas-mediated apoptosis in patients with hepatitis A virus (HAV) infection. A decreased T_REG cell frequency is associated with severe liver injury (32), which is mediated by an immune-mediated mechanism during HAV infection (33), indicating a protective role of T_REG cells against immune-mediated host injury during viral infection.

Dual roles of T_REG cells in viral infection have also been demonstrated in mouse models. Depletion of T_REG cells results in enhanced CD8⁺ T cell responses against infecting viruses and accelerated viral clearance in mouse models of infection with HSV (34) or lymphocytic choriomeningitis virus (35), indicating that T_REG cells control virus-specific cellular immune responses during viral infection. In contrast, depletion of T_REG cells results in greater disease severity because of an increased immunopathological reaction in mouse models of corneal infection with HSV (36) or infection with respiratory syncytial virus or West Nile virus (37–39), supporting host protective roles of T_REG cells during viral infection. These previous findings show both immunosuppressive and host protective roles of T_REG cells during viral infection (40–42).

Perturbations in the T_REG cell population were recently reported among patients with COVID-19 (43–45). However, this cross-sectional study has not directly investigated suppressive functions of T_REG cells and relationships between T_REG cells and effector T cell functions. In this study, we examined the quantitative and qualitative changes in the T_REG cell population in patients with COVID-19 and healthy donors. We analyzed the frequency of the T_REG cell population and subpopulation and their phenotypes during the course of COVID-19 using longitudinal PB samples. We also compared the frequency, phenotype, and suppressive function of T_REG cells among patients with severe and mild COVID-19 and healthy donors and found that the T_REG cell population is distinctively modulated in patients with severe and mild disease.

For this study, we enrolled PCR-confirmed COVID-19 patients in the year 2020 with severe (n = 21; male/female ratio = 13:8; mean age = 62.9 y, range = 41–81 y) or mild (n = 35; male/female ratio = 16:19; mean age = 47.8 y, range = 18–96 y) disease. SARS-CoV-2 infection was confirmed using the PowerChek2019-nCoV Real-time PCR Kit (Kogene Biotech) at Chungnam National University Hospital (28 patients) or Allplex SARS-CoV-2 Assay (Seegene) at Chungbuk National University Hospital (28 patients; Supplemental Table I). The day of symptom onset could be defined in each patient, and thus days postsymptom onset (DPSO) could be determined for each sampling date. Three to four follow-up PB samples were obtained from each patient. The mild patients included patients with mild and moderate symptoms at peak severity, and severe patients included severe and critical patients according to the National Institutes of Health severity scale (46). PB samples from 23 SARS-CoV-2–unexposed healthy donors were used as controls. Clinical information of each patient, including the date of sampling, comorbidities, and treatment during COVID-19, is presented in Supplemental Table I. We also enrolled healthy donors (n = 23; mean age = 39.9 y, range = 24–62 y). This study was reviewed and approved by the institutional review board of all participating institutions and conducted according to the principles of the Declaration of Helsinki. Informed consent was obtained from all donors and patients.

PBMCs were isolated from whole blood by density gradient centrifugation using Lymphocyte Separation Medium (Corning).

Total PBMCs were incubated at room temperature for 15 min with fluorochrome-conjugated Abs against cell surface markers. To exclude dead cells, we used a Live/Dead Cell Stain Kit (Invitrogen). For intracellular staining, fixation and permeabilization were achieved with the Foxp3/Transcription Factor Staining Buffer Kit (eBioscience) for 15 min, and Abs specific for Foxp3, Ki-67, and CTLA-4 were added for another 15 min (Table I). An LSR II instrument (BD Bioscience) was used for flow cytometry. The gating strategy for phenotyping the T_REG cell population is presented in Supplemental Fig. 1A.

Total blood lymphocyte counts were obtained by clinical laboratories at the time of blood sampling. We calculated the absolute counts of total CD3⁺ and CD4⁺ cells, T_REG cells, activated T_REG cells, and CTLA-4⁺ T_REG cells based on the blood lymphocyte counts and the percentages of each cell type obtained from flow cytometry.

Recently published publicly available datasets were used in the reanalysis (47) (GSE152522; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE152522). The original datasets were obtained by Ag-reactive T cell enrichment. In this assay, SARS-CoV-2–reactive CD4⁺ T cells were enriched based on the expression of surface activation-induced markers (CD137 and CD69) after ex vivo stimulation of PBMCs for 24 h with overlapping peptide (OLP) pools targeting the immunogenic domains of the spike and membrane proteins of SARS-CoV-2 from 30 acute COVID-19 patients (mild, n = 21; severe, n = 9). After downloading, the datasets were preprocessed and analyzed using the Seurat package (48). To correct the batch effect according to donor origin, dimensional reduction was performed using the mutual nearest neighbor (MNN) algorithm with 2000 variable genes (49) (RunFastMNN function). Uniform manifold approximation and projection (UMAP) was performed with the top MNN components (total CD4⁺ T cells, dimensions = 20; T_REG cells, dimensions = 10) for cell clustering and visualization (RunUMAP and FindNeighbors function). Next, the cells were clustered by an unsupervised clustering method (total CD4⁺ T cells, resolution = 0.1; T_REG cell resolution = 0.2; FindClusters function).

Differentially expressed genes (DEGs) were calculated based on the Wilcoxon rank sum test with default parameters (FindAllMarkers function). To characterize the immunophenotype of each cluster, we carried out gene set enrichment analysis (GSEA) with publicly available gene sets, including Gene Ontology: Biological Process database, using the piano package (50). Ontology terms related to immune responses were filtered using the following inclusion criteria: “T_CELL,” “IMMUNE,” “INNATE,” “ADAPTIVE,” “INFLAM,” “INTERLEUKIN,” “INTERFERON,” “NECROSIS,” “APOPTOSIS,” “SENESCENCE,” “NATURAL,” “LYMPHOCYTE,” “LEUKOCYTE,” “TRANSFORMING,” “CHEMO,” “CYTOTO,” “CYTOKINE,” “ANTIGEN,” and “SIGNALING.” To evaluate apoptosis-associated gene set enrichment, cells with active gene sets were identified using the AUCell package (51). The subcluster proportion was calculated in patients (mild, n = 15; severe, n = 7) whose T_REG cell numbers were >50.

PBMCs were stimulated with soluble anti-CD3 (0.1 μg/ml) and anti-CD28 (1 μg/ml) mAbs and cultured for 6 h at 37°C. After stimulation, apoptotic cells were stained with fluorochrome-conjugated annexin V Ab and analyzed by flow cytometry.

To evaluate apoptotic cells among T_REG cells, we stained PBMCs using a Live/Dead Cell Stain Kit and fluorochrome-conjugated Abs against surface markers, CD3, CD4, CD25, and CD127. After washing with 1× binding buffer, cells were stained with PE-conjugated annexin V (BD Biosciences). Fixation and permeabilization were performed before Foxp3 and active caspase-3 staining (Table I). The percentages of late apoptotic (annexin V⁺Live/Dead dye⁺) and active caspase-3⁺ cells were analyzed by flow cytometry in the gate of T_REG cells (CD4⁺CD25⁺CD127^loFOXP3⁺) or non-T_REG cells (CD4⁺CD25⁻) T cells.

Cryopreserved PBMCs were incubated for 12 h at 37°C after thawing. To deplete CD25⁺ T_REG cells from PBMCs, we attached CD25 Microbeads II (Miltenyi Biotec) to total PBMCs, and labeled cells were depleted by an LD column (Miltenyi Biotec) and MACS separator (Miltenyi Biotec). Unlabeled cells that flew through the LD column were collected and used as CD25⁺ cell–depleted PBMCs. The magnetically labeled CD25⁺ cells remaining in the LD column were flushed out using a plunger. CD25⁺ cell–depleted PBMCs were labeled with 5 μM CellTrace Violet (CTV; CellTrace Violet Cell Proliferation Kit; Invitrogen). CD25⁺ cell–depleted, CTV-labeled PBMCs, serving as responder T (T_RESP), were stimulated with soluble anti-CD3 (0.1 μg/ml; BD Pharmingen) and anti-CD28 (0.1 μg/ml; BD Pharmingen) with or without readdition of the depleted CD25⁺ cells. After 96 h of culture, proliferation of CD4⁺ and CD8⁺ T_RESP cells was analyzed by assessing the percentage of CTV^lowKi-67⁺ cells in the gate of CD4⁺ or CD8⁺ T cells by flow cytometry. Data are presented as percent suppression and determined as follows: % suppression = [1 − (% T_RESP cell proliferation in the CD25⁺ cell–reconstituted PBMCs/% T_RESP cell proliferation in the CD25⁺ cell–depleted PBMCs)] × 100. In this assay, we could evaluate suppressive activity of the total CD25⁺ T_REG cell population, not suppressive activity of T_REG cells on a per-cell basis. The depletion of T_REG cells from PBMCs and suppression assay were described previously (32). The suppression data were also analyzed by the mitotic index on the basis of the number of mitotic events of T_RESP cells. The mitotic index was calculated as described previously (52, 53).

CD4⁺ cells were isolated from PBMCs using CD4 microbeads (Miltenyi Biotec), LS columns (Miltenyi Biotec), and MACS separator (Miltenyi Biotec). The magnetically labeled CD4⁺ cells were incubated with fluorochrome-conjugated Abs against cell surface markers (CD25, Live/Dead, CD4, and CD127) at room temperature for 15 min. Using BD FACSAria, we sorted CD4⁺ T_REG cells and non-T_REG CD4⁺ cells. Non-T_REG CD4⁺ cells were labeled with 5 μM CTV (CellTrace Violet Cell Proliferation Kit; Invitrogen) and stimulated with T_REG Suppression Inspector (130-092-909; Miltenyi Biotec) with or without the addition of CD4⁺ T_REG cells at a 5:1 ratio. After 96 h of culture, proliferation of CD4⁺ T_RESP cells was analyzed by assessing the percentage of CTV_lowKi-67⁺ cells in the gate of CD4⁺ T cells by flow cytometry. Data are presented as percent suppression and determined as follows: % suppression = [1 − (% T_RESP cell proliferation with T_REG cells/% T_RESP cell proliferation without T_REG cells)] × 100.

PBMCs were cultured for 6 h at 37°C with OLP mix for SARS-CoV-2 spike protein (1 µg/ml for each peptide; Miltenyi Biotec). In the culture, we also added anti-human CD28 and CD49d mAbs (1 µg/ml; BD Biosciences) to stimulate costimulatory and adhesion proteins, enhancing the detection of cytokine-producing T cells in intracellular cytokine staining (ICS) assays (54, 55). One hour after the initial stimulation, brefeldin A (GolgiPlug; BD Biosciences) and monensin (GolgiStop; BD Biosciences) were added.

Corrplot package (v0.84) (56) in R (4.0.1) was used to visualize the Spearman rank correlation coefficients (r) and p values between all parameters. Spearman rank two-tailed p values were calculated using rcorr (Hmisc v4.4-2).

Statistical analyses were performed in Prism software version 9.0 for MAC (GraphPad, La Jolla, CA). Data did not follow a normal distribution. To compare values between healthy donors and severe and mild patients, we used one-way ANOVA. To compare paired values for each patient between different time periods (1–14 versus 15–28 DPSO), we performed the mixed-effects analysis test. For patients whose blood samples were drawn on different days within a single time period (1–14 or 15–28 DPSO), the two values were averaged. For cross-sectional analyses, the local regression (LOESS) function was used with a confidence level of 95%. We also performed a linear regression analysis with longitudinally obtained data. Correlations between two parameters were evaluated by a nonparametric two-tailed Spearman correlation test with a 95% confidence interval. A p value ≤0.05 was considered significant.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

First, we examined the frequency of T_REG cells using PBMCs obtained from patients with severe and mild COVID-19 using flow cytometry (Table I). CD3⁺CD4⁺FOXP3⁺CD25^hiCD127^lo cells were defined as T_REG cells (Supplemental Fig. 1A). The relative frequency of T_REG cells among CD4⁺ T cells was plotted during the course of COVID-19 up to 46 DPSO (Fig. 1A) and presented longitudinally for each patient (Fig. 1B). In a linear regression analysis, severe patients exhibited a significant increase in the frequency of T_REG cells among CD4⁺ T cells during the period of 1–21 DPSO, but mild patients did not (Fig. 1C). We determined the mean values obtained between 1 and 14 DPSO for the severe (n = 17) and mild (n = 18) patient groups and found that the T_REG frequency was significantly reduced in patients with mild COVID-19 compared with healthy donors (Fig. 1D). We also examined changes in T_REG frequency during the course of COVID-19. T_REG frequency significantly increased in patients with severe COVID-19 15–28 DPSO compared with 1–14 DPSO, but we found no significant change in patients with mild COVID-19 (Fig. 1E). At 15–28 DPSO, patients with severe disease had a significantly higher T_REG frequency than patients with mild disease. We also compared the frequencies of total T_REG cells in COVID-19 patients with those in healthy donors 15–28 DPSO (Supplemental Fig. 1B). However, we found no significant difference between patients with severe or mild COVID-19 and healthy donors.

We examined the frequency of CD45RA⁻FOXP3^hi activated T_REG cells among CD3⁺CD4⁺FOXP3⁺CD25^hiCD127^lo T_REG cells (Supplemental Fig. 1A). Representative flow cytometry plots from healthy donors and patients with mild and severe COVID-19 are presented in Fig. 1F. The relative frequency of CD45RA⁻FOXP3^hi activated T_REG cells among CD4⁺ T cells was plotted during the course of COVID-19 (Fig. 1G) and presented longitudinally for each patient (Fig. 1H). In a linear regression analysis, severe patients exhibited a significant increase in the frequency of activated T_REG cells among CD4⁺ T cells during the period of 1–21 DPSO, but mild patients did not (Fig. 1I). At 1–14 DPSO, the activated T_REG cell frequency was significantly reduced in patients with mild COVID-19 compared with patients with severe COVID-19 (Fig. 1J). The frequency of activated T_REG cells significantly increased in patients with severe COVID-19 15–28 DPSO compared with 1–14 DPSO (Fig. 1K). At 15–28 DPSO, patients with severe disease had a significantly higher frequency of activated T_REG cells than patients with mild disease. We also compared the frequencies of activated T_REG cells in COVID-19 patients with those in healthy donors 15–28 DPSO (Supplemental Fig. 1C). Patients with severe COVID-19 exhibited significantly higher values than healthy donors or patients with mild COVID-19.

We also analyzed the percentage of activated T_REG cells among total T_REG cells and found that the activated T_REG cell percentage was significantly increased in patients with severe COVID-19 compared with patients with mild COVID-19 (Supplemental Fig. 1D). In addition, at 1–14 and 15–28 DPSO, patients with severe disease had a significantly higher percentage of activated T_REG cells among total T_REG cells than patients with mild disease (Supplemental Fig. 1E). Thus, expansion of the CD45RA⁻FOXP3^hi activated T_REG cell population was a characteristic feature of severe COVID-19.

We obtained the absolute numbers of lymphocytes (Supplemental Fig. 1F), total CD3⁺ (Supplemental Fig. 1G) and CD4⁺ cells (Supplemental Fig. 1H), T_REG cells (Supplemental Fig. 1I), and activated T_REG cells (Supplemental Fig. 1J) and compared the numbers between the severe and mild patient groups. As reported previously, the absolute numbers of total lymphocytes and CD3⁺ and CD4⁺ cells were significantly lower in the severe group than in the mild group (4, 57). The absolute numbers of total lymphocytes and CD3⁺ and CD4⁺ cells were significantly increased 15–28 DPSO only in the mild group. The absolute number of T_REG cells did not differ between the two patient groups, but it significantly increased in each group 15–28 DPSO. Importantly, the absolute number of activated T_REG cells was significantly higher in severe COVID-19 patients compared with mild COVID-19 patients and significantly increased 15–28 DPSO only in the severe group.

We also analyzed the normalized geometric mean fluorescence intensity of FOXP3 in each sample. There was no significant difference in the normalized geometric mean fluorescence intensity of FOXP3 in T_REG cells among healthy controls and patients with severe and mild COVID-19, although it tended to be higher in patients with severe disease (Supplemental Fig. 1K, 1L).

To comprehensively understand the molecular characteristics of T_REG cells in patients with COVID-19, we reanalyzed publicly available single-cell RNA sequencing (scRNA-seq) data for SARS-CoV-2–reactive CD4⁺ T cells (GSE152522; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE152522) (47). In the previous study, SARS-CoV-2–reactive CD4⁺ T cells had been enriched based on the upregulation of activation-induced markers, such as CD137 and CD69, after ex vivo stimulation of PBMCs with SARS-CoV-2 Ags, such as membrane and spike. The enriched SARS-CoV-2–reactive CD4⁺ T cells from patients with COVID-19 (mild, n = 21; severe, n = 9) had been subjected to scRNA-seq analysis. In this study, we reanalyzed the raw count matrices and metadata. The MNN algorithm was used to correct the batch effect according to donor origin (49). We clustered the total SARS-CoV-2–reactive CD4⁺ T cells using the UMAP algorithm with the Seurat package (48). T_REG cells with FOXP3 expression clustered among CD4⁺ T cells (Supplemental Fig. 2A, 2B). T_REG cells were further subclustered into four different populations (Fig. 2A). These subclusters were characterized according to the expression of differentially expressed marker genes (Fig. 2B, Supplemental Fig. 2C, 2D).

To further investigate the molecular characteristics of each cluster, we performed GSEA with publicly available gene sets related to immune responses (Fig. 2C). Cluster-1 was characterized by gene sets related to the response to type I IFN and the IFN-γ and cytokine response. Cluster-2 was mainly enriched with gene sets related to apoptosis, especially the intrinsic apoptotic pathway. In further analyses, cluster-2 was also enriched with gene sets related to S phase of the cell cycle (Fig. 2D). Collectively, cluster-2 exhibits proliferating but apoptosis-prone features, indicating that T_REG cells in cluster-2 are recently activated but vulnerable to apoptotic stimulation. Cluster-3 was characterized by genes related to leukocyte chemotaxis, and cluster-4 was characterized by marked upregulation of CCR8 in addition to gene sets related to the regulation of diverse immune effector processes.

Next, we compared the proportion of each cluster according to disease severity. Cluster-2 was significantly increased in patients with severe COVID-19 (mean 54.2%) compared with patients with mild COVID-19 (mean 38.4%), whereas cluster-4 was significantly decreased in patients with severe disease (mean 1.8%) compared with patients with mild disease (mean 5.4%; Fig. 2E). The significant increase in the proportion of cluster-2 in patients with severe COVID-19 indicates that T_REG cells are activated during severe COVID-19 but become vulnerable to apoptosis. To further analyze the enrichment of gene sets related to apoptosis, we identified T_REG cells with active gene sets using the AUCell method (51). When we plotted cells with a gene set enrichment score above the threshold, cells with various apoptotic features were mainly located in cluster-2 (Supplemental Fig. 2E). In addition, the expression of various apoptosis-related gene sets was enriched in patients with severe disease compared with patients with mild disease (Fig. 2F). Collectively, these findings demonstrate that T_REG cells from patients with severe COVID-19 exhibit proliferating but apoptosis-prone features.

We validated the results of the scRNA-seq analysis using PBMCs obtained from our cohort. First, we examined the expression of Ki-67, a marker of proliferating cells, among T_REG cells (Fig. 3A). The relative frequency of Ki-67⁺ cells among T_REG cells was plotted during the course of COVID-19 (Fig. 3B) and presented longitudinally for each patient (Fig. 3C). In a linear regression analysis, severe patients exhibited a significant increase in the frequency of Ki-67⁺ cells among T_REG cells during the period of 1–21 DPSO, but mild patients did not (Fig. 3D). At 1–14 DPSO, the frequency of Ki-67⁺ cells among T_REG cells showed no significant differences between healthy donors and patients with COVID-19 (Fig. 3E). Although the frequency of Ki-67⁺ cells among T_REG cells was not increased in patients with severe COVID-19 1–14 DPSO, it significantly increased in patients with severe COVID-19 at 15–28 DPSO compared with 1–14 DPSO, but there was no significant change in patients with mild COVID-19 (Fig. 3F). At 15–28 DPSO, patients with severe disease had a significantly higher frequency of Ki-67⁺ cells than patients with mild disease. We also compared the frequencies of Ki-67⁺ T_REG cells in COVID-19 patients with those in healthy donors 15–28 DPSO (Fig. 3G). Patients with mild COVID-19 exhibited significantly lower values than patients with severe COVID-19. However, we found no difference between patients with severe COVID-19 and healthy donors. Thus, T_REG cells undergo proliferation in patients with severe COVID-19, corroborating the results from the scRNA-seq analysis.

We also evaluated apoptosis in T_REG cells from COVID-19 patients with or without anti-CD3/anti-CD28 stimulation, which mimics TCR-mediated stimulation using PBMCs obtained 15–28 DPSO. When apoptotic cells were detected by staining with annexin V and Live/Dead dye, the frequency of apoptotic cells among T_REG cells was significantly increased by anti-CD3/anti-CD28 stimulation in patients with severe disease, but not in patients with mild disease (Fig. 3H, 3I). The frequency of apoptotic cells after stimulation was significantly higher in patients with severe disease than in patients with mild disease. Similar results were observed when apoptotic cells were detected by immunostaining active, cleaved caspase-3 (Fig. 3J, 3K). These findings confirmed the results from scRNA-seq analysis that T_REG cells from patients with severe COVID-19 are prone to apoptosis. Taken together, T_REG cells from patients with severe COVID-19 easily undergo apoptosis upon stimulation, although they have features of activation and proliferation.

Because the T_REG cell population underwent dynamic changes in patients with severe COVID-19, including an increased frequency of activated T_REG cells and enhanced apoptotic features, we examined the suppressive functions of the T_REG population. First, we investigated the expression of CTLA-4 (Fig. 4A), which is a main mediator of the suppressive functions of T_REG cells (20–22). The relative frequency of CTLA-4⁺ cells among T_REG cells was plotted during the course of COVID-19 (Fig. 4B) and presented longitudinally for each patient (Fig. 4C). In a linear regression analysis, severe patients tended to have an increased frequency of CTLA-4⁺ cells among T_REG cells during the period of 1–21 DPSO, but mild patients did not (Fig. 4D). The frequency of CTLA-4⁺ cells significantly increased 1–14 DPSO in patients with severe COVID-19 compared with healthy donors and patients with mild COVID-19 (Fig. 4E), supporting a previous report that genes encoding suppressive molecules, such as CTLA-4, are up-regulated in T_REG cells from patients with severe COVID-19 (44). At 1–14 and 15–28 DPSO, patients with severe disease had a significantly higher frequency of CTLA-4⁺ cells than patients with mild disease (Fig. 4F). The absolute number of CTLA-4⁺ T_REG cells was significantly higher in the severe group compared with the mild group at 15–28 DPSO and significantly increased at 15–28 DPSO only in the severe group (Fig. 4G).

Next, we evaluated the suppressive activity of the T_REG population by assessing the anti-CD3/CD28–stimulated proliferation of T_REG-depleted PBMCs and T_REG-reconstituted PBMCs. The proliferation of non-T_REG CD4⁺ and CD8⁺ T cells was measured by CTV dilution (Fig. 4H). In this assay, we evaluated the suppressive activity of the total T_REG population, not the suppressive activity of T_REG cells on a per-cell basis. We found that the suppressive activity of the T_REG population was not significantly different among healthy donors and patients with severe or mild COVID-19 when the proliferation of non-T_REG CD4⁺ T cells (Fig. 4I) and CD8⁺ T cells (Fig. 4J) was assessed. Similar results were obtained when we analyzed the suppression data by the mitotic index on the basis of the number of mitotic events of T_RESP cells (Supplemental Fig. 3A, 3B).

We also performed conventional T_REG cell suppression assays on a per-cell basis using sorted T_REG cells and non-T_REG CD4⁺ T cells and found that the suppressive activity of T_REG cells was not significantly different among healthy donors and patients with severe or mild COVID-19 (Supplemental Fig. 3C).

Collectively, these results demonstrate that the suppressive functions of the T_REG cell population are not enhanced in patients with severe COVID-19, although the frequencies of activated T_REG cells and CTLA-4⁺ T_REG cells are increased in patients with severe COVID-19. This finding can be explained by increased apoptosis of T_REG cells in patients with severe COVID-19. The T_REG cell population is remarkably deranged during severe COVID-19, but the overall suppressive activity of the T_REG cell population is not altered.

ICS was performed to assess the SARS-CoV-2–specific effector functions of CD4⁺ and CD8⁺ T cells. In ICS, PBMCs were ex vivo stimulated with OLP mix for SARS-CoV-2 spike protein and stained for IFN-γ, IL-2, and TNF (Supplemental Fig. 3D). The frequency of IFN-γ– or TNF-producing T cells was not different between patients with severe and mild COVID-19, whereas the frequency of IL-2–producing T cells was significantly higher in patients with severe COVID-19 (Supplemental Fig. 3E, 3F). The absolute count of TNF-producing T cells was significantly higher in patients with mild COVID-19 (Supplemental Fig. 3G, 3H). The ratio of IFN-γ–producing CD4⁺ T cells or TNF-producing CD8⁺ T cells to T_REG cells was significantly higher in patients with mild COVID-19 (Supplemental Fig. 3I, 3J).

We analyzed whether the frequency of T_REG cells correlates with immunological parameters. We analyzed 21 parameters, including demographic, leukocyte, T_REG, CD4 effector function, and CD8 effector function parameters, in correlograms. The absolute number of T_REG cells positively correlated with the absolute number of CD3⁺, CD4⁺, and CD8⁺ T cells in both mild and severe patients. However, the absolute number of neutrophils and the SARS-CoV-2 neutralizing Ab titers did not significantly correlate with the T_REG cell population regardless of the severity of the disease. In patients with mild COVID-19, the T_REG-related parameters exhibited significant inverse correlations with CD4 T cell effector functions (blue box in Fig. 5A). However, such correlations were not observed in patients with severe COVID-19 (blue box in Fig. 5B).

When regression analyses were performed between T_REG cell frequency and CD4 T cell effector functions among all patients with COVID-19, no significant correlation was observed (Fig. 5C, 5D). In patients with mild COVID-19, the frequency of T_REG cells among CD4⁺ T cells inversely correlated with the frequency of spike-specific CD4⁺ T cells that produced effector cytokines, including IFN-γ, IL-2, and TNF, among CD4⁺ T cells (Fig. 5E). We also analyzed the frequency of polyfunctional CD4⁺ T cells that produce ≥2 cytokines simultaneously. We found that the frequency of T_REG cells among CD4⁺ T cells inversely correlated with the frequency of spike-specific polyfunctional CD4⁺ T cells in patients with mild COVID-19 (Fig. 5F). However, such correlations were not observed in patients with severe COVID-19 (Fig. 5G, 5H). These findings suggest that T_REG cells negatively control the SARS-CoV-2–specific effector functions of CD4⁺ T cells during mild COVID-19. However, such negative correlations were not observed in patients with severe COVID-19. The lack of correlation might be because of functional derangement of the T_REG cell population during severe COVID-19 (Table II).

This longitudinal study demonstrated the phenotypes and functions of the CD4⁺CD25⁺FOXP3⁺ T_REG cell population in patients with severe and mild COVID-19 as summarized in Table II. In patients with severe COVID-19, the frequencies of total T_REG cells and CD45RA⁻FOXP3^hi activated T_REG cells among CD4⁺ T cells significantly increased 15–28 DPSO. At this time point, T_REG cells from severe patients showed not only increased proliferation but also enhanced apoptotic features. Although the expression of CTLA-4, an effector molecule for immunosuppressive functions, significantly increased 15–28 DPSO in T_REG cells from severe patients, the actual suppressive function of the T_REG cell population in severe patients was comparable with that of healthy donors or mild patients. Interestingly, the frequency of T_REG cells among CD4⁺ T cells inversely correlated with SARS-CoV-2–specific cytokine production and the polyfunctionality of CD4⁺ T cells in mild patients, but not severe patients. These data suggest that T_REG cells are major regulators of virus-specific CD4⁺ T cell responses during mild COVID-19, whereas the T_REG cell population is functionally deranged during severe COVID-19. Our findings suggest that the T_REG cell population alters distinctively in patients with severe and mild COVID-19 during the course of disease.

Several studies have examined the landscape of immune cell populations in patients with COVID-19 by performing transcriptomic analyses and reported alteration in the T_REG cell population according to the severity of the disease (43, 44, 58). Recent studies showed consistent results that the proportion of T_REG cells and their FOXP3 expression level increased in patients with severe COVID-19, corroborating our current finding (43, 44). In addition, transcriptomic analysis showed proinflammatory features and suppressive features in T_REG cells from patients with severe COVID-19 (43). They suggest that IL-6 and IL-18 might cause the perturbation in the T_REG cell population. Another study reported that T_REG cells from patients with COVID-19 exhibit activated signatures, which increases with the disease severity (44). Relatively decreased frequencies of T_REG cells were reported in convalescent patients after mild COVID-19 (45). However, previous studies have not directly investigated the suppressive capacity of T_REG cells and the relationship between T_REG cells and SARS-CoV-2–specific effector T cell functions. In addition, previous studies examined T_REG cells only cross-sectionally and lacked longitudinal analysis of each COVID-19 patient.

In this study, we reanalyzed publicly available scRNA-seq data for SARS-CoV-2–reactive CD4⁺ T cells that had been enriched based on the upregulation of activation-induced markers CD137 and CD69 after ex vivo stimulation of PBMCs with SARS-CoV-2 Ags (47). T_REG cells from patients with severe COVID-19 exhibited apoptosis-prone features. However, ex vivo stimulation with SARS-CoV-2 Ags may alter the transcriptomic features of T_REG cells. To overcome this limitation, we performed additional scRNA-seq reanalysis using another set of scRNA-seq data from patients with COVID-19 (59) and confirmed the apoptosis-prone features of T_REG cells from patients with severe COVID-19.

Previous studies regarding the dynamics of the T_REG cell population during viral infection have shown that increased T_REG cell frequencies or suppressive activity may be disadvantageous to the host in regard to eliminating infecting viruses (34, 37, 38). The frequencies of T_REG cells in the PB and liver have been reported to be higher in patients with HCV infection (29–31). Attenuated CD8⁺ T cell functions are related to increased T_REG cell frequency during HCV infection. In addition, therapeutic vaccine-induced enhancement of HCV-specific T cell responses is associated with an IFNL3 adjuvant-related reduction in T_REG cell frequency in patients with chronic HCV infection (60), indicating that T_REG cells are continuously suppressing virus-specific T cell responses during chronic HCV infection. In this study, the frequency of T_REG cells was lower in patients with mild COVID-19 than in healthy donors or severe patients, and these cells exhibited less activated phenotypes. Moreover, the frequency of T_REG cells inversely correlated with SARS-CoV-2–specific cytokine production and the polyfunctionality of CD4⁺ T cells in mild COVID-19 patients. Our study supports that the T_REG cell population from mild COVID-19 patients is not abnormally disturbed and able to regulate the effector functions of virus-specific effector T cells.

Our study has limitations. First, patients with severe and mild COVID-19 differed in their age distribution (average age 62.9 and 47.8 y in severe and mild patients, respectively). In addition, the two patient groups differed in their gender distribution. Further studies of T_REG cells with age- and sex-matched patient groups are required. Second, we could not examine whether our current findings are generalizable to patients with other severe respiratory viral diseases, such as influenza and respiratory syncytial virus infection.

In this study, the size of the T_REG cell population did not correlate with the effector functions of virus-specific CD4⁺ T cells during severe COVID-19, showing that the T_REG cell population is aberrantly deranged. The T_REG cell population exhibited enhanced apoptosis in patients with severe COVID-19. Moreover, the scRNA-seq analysis of public data revealed that T_REG cells are prone to intrinsic pathway-mediated apoptosis. A previous study demonstrated that the frequency of T_REG cells is decreased because of upregulation of Fas receptor during acute HAV infection (32). Importantly, a decreased frequency of T_REG cells was significantly associated with enhanced liver injury in patients with acute HAV infection. Given that the liver injury is caused by a T cell–mediated immunopathological mechanism during acute HAV infection (33), enhanced apoptosis of T_REG cells results in a loss of immunosuppression and exaggerated immunopathological host injury in patients with acute HAV infection. Other studies support the T_REG cell population being regulated mainly by susceptibility to Fas-mediated apoptosis (61–63). Although our findings demonstrate increased cleavage of capsase-3, which is a downstream caspase, in the expanded T_REG cell population from severe COVID-19 patients, the precise mechanism of how apoptosis is initiated in T_REG cells needs to be investigated further.

Immunosuppressive cells other than T_REG cells also contribute to the maintenance of immune homeostasis, including myeloid-derived suppressor cells (MDSCs). MDSCs are known to exert their immunosuppressive activity on T cells, macrophages, dendritic cells, and NK cells in pathological environments through reactive oxygen species or l-arginine. In viral infection and cancer, MDSCs have been shown to suppress the functions of other immune effector cells and increase the number of T_REG cells (64–67). In patients infected with SARS-CoV-2, increased proinflammatory cytokines in severe COVID-19 patients (16) may cause the expansion of MDSCs, which are capable of inhibiting the proliferation and function of effector T cells (68–70). Severe progression of COVID-19 may be associated with perturbation and derangement of the major two immunosuppressive cell types, including T_REG cells and MDSCs.

In this study, we found that the T_REG cell population from patients with severe COVID-19 expanded significantly and exhibited not only more proliferative features but also elevated apoptotic features compared with the population from mild patients at 15–28 DPSO. Despite the increased activation, we observed comparable suppressive activity of the T_REG cell populations of severe and mild COVID-19 patients. The frequency of T_REG cells in mild patients inversely correlated with the functionality of SARS-CoV-2–specific CD4⁺ T cells, supporting the importance of T_REG cells in the regulation of virus-specific CD4⁺ T cells in mild COVID-19. However, no correlations were observed in patients with severe disease. Our findings suggest that the dynamics and functions of the T_REG cell population are distinctive in patients with severe and mild COVID-19.

The authors have no financial conflicts of interest.

We appreciate laboratory members for support and critical opinions.