2.1. Integrated analysis of scRNA-seq data of peripheral immune cells from convalescent patients with COVID-19.
We collected samples between 3.5 and 4.2 months following first symptom onset or between 0 and 3.3 months following closest discharge. The demographics and clinical features of the patients are listed in Table 1. The laboratory findings of enrolled patients are shown in Supplementary Table 1. Thirteen convalescent patients with COVID-19 were classified into two clinical conditions: convalescent patients with moderate COVID-19 (n=5) and convalescent patients with severe COVID-19 (n=8) (Table 1; Supplementary Table 1). The five patients with moderate COVID-19 profiled were four males and one female, aged 28 to 39 years, at an average age of 35.4 years. The eight patients with severe COVID-19 profiled were six males and two females, aged 29 to 53 years, at an average age of 38.9 years. The healthy donors were four males and one female, aged 27 to 30 years, at an average age of 28.8 years. All healthy donors were also confirmed without SARS-CoV-2 infection by rRT-PCR. All the patients were confirmed between Jan 18th, 2020 and Feb 1st, 2020. Four of thirteen samples were collected from ventilated COVID-19 patients who were diagnosed with acute respiratory distress syndrome (ARDS; Supplementary Table 1). No patients received azithromycin, which has potential immunomodulatory effects, prior to sampling (Supplementary Table 1). No patients received Remdesivir during hospitalization prior to sampling.
There were no significant differences in the absolute number of leukocytes, neutrophils, monocytes and platelets counts between moderate and severe group. However, the severe group exhibited a significantly lower absolute lymphocyte count and albumin than the moderate group (P =0.003 and P =0, respectively) (Supplementary Table 1). These findings were in accordance with the results of previous study (PMID: 32641700). All the patients were shown positive in SARS-CoV-2 RNA testing, but 60% (3 of 5) and 50% (4 of 8) of patients were seropositive in moderate and severe respectively. Interestingly, however, we found no patients were seropositive on the days performed SARS-CoV-2 RNA testing (Supplementary Table 2).
All samples were subjected to scRNA-seq based on the 10X Genomics 5’ sequencing platform to generate both the gene expression and T cell receptor (TCR) or B cell receptor (BCR) data of peripheral blood mononuclear cells (PBMCs) from the subjects e.g. convalescent patients with moderate COVID-19 (n=5) and convalescent patients with severe COVID-19 (n=8) and 5 healthy donors (HDs; n=5) (Fig. 1a and b). The scRNA-seq raw data were analyzed by a unified analysis pipeline, including the kallisto and bustools programs, to obtain the gene expression data of individual cells and by the CellRanger program to obtain TCR and BCR sequences. We applied a common set of stringent quality control criteria to ensure that the selected data were from single and live cells and that their transcriptomic phenotypes were comprehensively characterized. We abandoned one sample (S02 in HD group) due to the unsatisfactory health records. A total of 122,726 high-quality single cells were ultimately obtained from 17 samples, with an average of 7,219 cells per sample (Supplementary Table 3). After the unified single-cell analysis pipeline (see Methods),11,7102 cells were used for identification of cell types (Supplementary Table 3). Among these (11,7102) cells, 27,781 cells (23.72%) were from the healthy donors, 30,447 cells (26%) were from the convalescent moderate condition and 58,874 cells (50.3%) were from the convalescent severe conditions. All high-quality cells were integrated into an unbatched and comparable dataset and subjected to principal component analysis after correction for read depth and mitochondrial read counts (Supplementary Table 4). Using graph-based clustering of uniform manifold approximation and projection (UMAP), we identified 47 major cell types (Supplementary Table 5) and captured 27 clusters according to the expression of canonical gene markers (Fig. 1c-f, Supplementary Table 5). These cell types included CD14 (+) monocyte (CD14+ Monocyte; CD14+LYZ+S100A8+), FCGR3A (+) monocyte (FCGR3A+ Monocyte; FCGR3A+IFITM3+CD68+), myeloid dendritic cells (Myeloid DC; CD1C+CST3+FCER1A+CLEC10A+), plasmacytoid dendritic cells (Plasmacytoid DC; GZMB+IL3RA+ITM2C+PLD4+), basophiles (Basophile; CLC+SOCS2+HDC+), CD34(+) hematopoietic stem and progenitor cells (HSPC; DEFA3+LYZ+SOX4+SPINK2+CD34+), neutrophils (Neutrophil; CSF3R+), naïve CD4(+) T cells (CD4+ naïve; CD4+CCR7+LEF1+SELL+), CD4(+) T cell memory (CD4+ memory; CD4+IL7R+LTB+S100A4+), CD4(+) effector memory T cells (CD4+ effector memory; CD4+AHNAK+RORA+S100A4+), T helper 2 (Th2; CD4+CCR6+GATA3+), regulatory T (Treg) cells (Treg; CD4+FOXP3+CTLA4+TIGIT+), naïve CD8(+) T cells (CD8+ naïve; CD8A+CCR7+LEF1+NELL2+), CD8(+) effector-GNLY (CD8+ effector-GNLY; CD8A+GNLY+FGFBP2+FCGR3A+), CD8(+) effector- GZMK (CD8+ effector-GZMK; CD8A+GZMK+), CD8(+) effector memory T cells (CD8+ effector memory; CD8A+FGFBP2+GZMH+NKG7+), gammadeltaT cells (gdT; CD3E+TRDC+CD8A-CD4-), natural killer T (NKT) cells (NKT; NCAM1+KLRD1+NCR1+KLRB1+SH2D1B+KIR3DL2+KIR2DL1+CD8A+), CD160 (+) natural killer T (NKT) cells (NKT-CD160; NCAM1+KLRD1+NCR1+KLRB1+CD160+SH2D1B+KIR3DL2+KIR2DL1+CD8A+), mucosal-associated invariant T (MAIT) cells (MAIT; CD3E+SLC4A10+), pre-B cells (Pre B; CD79A+MS4A1+CD19+IL7R+CD34-), maginal zone B cells (Maginal zone B; CD79A+MS4A1+CD19+CD1C+), follicular B cells (Follicular B; CD79A+MS4A1+CD19+CD22+FCER2+CD27-), plasma B cells (Plasma B; CD79A+MZB1+CD38+IGKC+), memory B cells (Memory B; CD79A+MS4A1+CD27+IGHD-), Proliferating (MKI67+), Platelet (PPBP+PF4+TUBB1+).
2.2. Phenotype characteristics of cell types in convalescent patients with moderate and severe COVID-19.
To show the differences in cell composition across two conditions, e.g. convalescent patients with moderate or severe COVID-19, and to compare with that in healthy donors, the relative percentage of the 27 major cell types in the PBMCs of each individual were calculated on the basis of scRNA-seq data. Notable differences could be observed in the immune compositions of healthy donors and the convalescent COVID-19 patients based on UMAP (Fig. 2a-c). The relative percentage of of CD4+ memory cluster was increased significantly in convalescent moderate COVID-19 patients compared with HD (P=0.0062) and decreased significantly in convalescent severe COVID-19 patients compared with moderate (P=0.0027). The relative percentage of the CD8+ effector memory cell cluster increased significantly with disease severity (HD vs. convalescent severe patients, P=0.04). In contrast, the relative percentage of the NKT-CD160 cell cluster was decreased significantly (HD vs. convalescent moderate patients, P=0.024; HD vs. convalescent severe patients, P=0.024). Compared with convalescent moderate patients, Maginal zone B cluster was decreased significantly in convalescent severe patients (P=0.000089).
Next, to explore the antiviral and pathogenic immune responses during infection, we performed expression levels of three important Gene Ontology (GO) biological process terms e.g. acute inflammatory response (GO0002526), response to interferon (IFN)-α (GO0035455) and immune response-activating cell surface receptor signaling pathway (GO0002429) in major cell types across three conditions. We found that the acute inflammatory response was significantly upregulated in CD8+ effector memory with disease severity. This response in NKT and NKT-CD160 significantly upregulated in moderate and significantly downregulated in severe. In contrast, it was significantly downregulated in Myeloid DC in moderate and significantly upregulated in severe. This response in CD14 Monocyte and FCGR3A+ Monocyte significantly upregulated in severe (Fig 2d; Supplementary Table 6). The response to IFN-α was significantly upregulated in CD4+ naïve, CD8+ naïve and NKT-CD160 with disease severity. In contrast, it was significantly downregulated in CD8+ effector-GNLY and Neutrophil. In gdT and NKT, it was significantly upregulated in moderate and significantly downregulated in severe. In contrast, in CD14+ Monocyte and FCGR3A+ Monocyte it was significantly downregulated in moderate and significantly upregulated in severe (Fig 2d; Supplementary Table 6). The immune response-activating cell surface receptor signaling pathway was found significantly upregulated in Follicular B, CD4+ naïve, CD4+ effector memory, CD8+ naïve, CD8+ effector-GNLY and CD8+ effector-GZMK with disease severity. But it was found significantly upregulated in CD4+ memory, Th2 and CD14 Monocyte in moderate and significantly downregulated in severe. In contrast, it was found significantly downregulated in NKT and NKT-CD160 in moderate and significantly upregulated in severe (Fig 2d; Supplementary Table 6).
2.3. Attenuated innate immune responses were observed in convalescent COVID-19 patients.
To further probe the transcriptomic changes of innate immune cells post infection with SARS-CoV-2, we compared the expression patterns of the convalescent moderate or convalescent severe condition with that of the HD in cell subtypes relevant to innate immunity. Using UMAP, we identified 5 major cell types according to the expression of canonical gene markers. These cell types included CD14+ Monocyte, FCGR3A+ Monocyte, Myeloid DC, Plasmacytoid DC and Neutrophil (Fig. 3a, b).
We analyzed differentially expressed genes (DEGs) from innate immune cell types like monocytes, dendritic cells and neutrophils from convalescent COVID-19 patients relative to healthy donors and used these genes to identify enriched Gene Ontology. We found that 70 genes related to innate immune responses were significantly differentially expressed across disease conditions (Fig. 3c; Supplementary Table 7 and 8). We found that significantly DEGs were involved in neutrophil response including neutrophil chemotaxis (GO: 0030593), neutrophil migration (GO: 1990266), neutrophil activation (GO: 0042119), neutrophil degranulation (GO:0043312), neutrophil activation involved in immune response (GO: 0002283) and neutrophil mediated immunity (GO: 0002446), such interferon-gamma response as response to interferon-gamma (GO:0034341), cellular response to interferon-gamma (GO:0071346) and interferon-gamma-mediated signaling pathway (GO:0060333) and such humoral response as antimicrobial humoral response (GO:0019730) and humoral immune response (GO:0006959) (Fig. 3d).
To unbias elucidate the status of innate cell subsets, which always reflected by specific mode of gene expression, from two kinds of convalescent COVID-19 patients, we deeply explored the expression pattern of DEGs involved in top 20 GO terms based on classification of their functions. We classified these genes into three categories including (1) response to interferon-gamma, (2) leukocyte migration and (3) neutrophil activation, migration and degranulation and it mediated immune response. We found substantial amounts of genes regarding response to interferon-gamma were downregulated in several innate cell subsets with the exception of some HLA class II molecules like HLA-DQA1, HLA-DQB1 and HLA-DPB1, which were upregulated in CD14+ Monocytes and FCGR3A+ Monocytes from convalescent patients with COVID-19. Downregulation of these kinds of genes were interferon-stimulated genes (ISGs) e.g. IFITM1, IFITM2, IFITM3, IRF7 and HLA class II genes HLA-DRA and HLA-DRB1 in CD14+ Monocytes, FCGR3A+ Monocytes, Myeloid DCs and Plasmacytoid DCs from convalescent patients with COVID-19. The gene, HMOX1, involved in leukocyte migration was also shown decreased its’ expression in CD14+ Monocytes, FCGR3A+ Monocytes and Myeloid DCs from convalescent patients with COVID-19. With the same trend, multiple genes, SERPINA1 and S100P, which regulate neutrophil activation, migration and degranulation and it mediated immune response, were downregulated in CD14+ Monocytes, FCGR3A+ Monocytes, Neutrophils and Myeloid DCs and GCA and CD36 were only downregulated Myeloid DCs and Plasmacytoid DCs from convalescent patients with COVID-19 (Fig. 3c, d, f; Supplementary Table 7 and 8).
We also found that CD14+ Monocytes, FCGR3A+ Monocytes, Myeloid DCs and Plasmacytoid DCs significantly downregulated regulation of innate immune response and positive regulation of cytokine production in severe patients compared to HD group (Fig. 3e). These results suggest that innate immune responses including neutrophil response and interferon-gamma response were attenuated in convalescent COVID-19-19 patients. Taken together, innate cell subsets decreased expression of genes involved in interferon-gamma, leukocyte migration and neutrophil mediated immune response in convalescent COVID-19 patients.
To further characterize the functions of DEGs in innate immune cell subsets, we performed analyses CD14+ Monocyte, FCGR3A+ Monocyte, Myeloid DC, Plasmacytoid DC and Neutrophil of convalescent moderate or severe patients in comparison with those of HDs. Total DEGs in 5 innate immune cell subsets from convalescent moderate or severe patients in comparison with those of HDs were shown in Supplementary Table 9 and the DEGs shared by both comparisons were presented in Fig. 3g. We found DEGs were involved in phagocytosis, neutrophil mediated immune response and immune response-activating cell surface receptor signaling pathway in CD14+ Monocyte, T cell activation, differentiation and T cell receptor signaling pathway and leukocyte cell-cell adhesion in FCGR3A+ Monocyte, regulation of T and leukocyte cell-cell adhesion in Myeloid DC, viral transcription and gene expression in Plasmacytoid DC, and neutrophil mediated immune response and regulation of cytokine production in Neutrophil in convalescent COVID-19 patients (Supplementary Fig. 1).
2.4. Strengthened T cell and lymphocyte activation, differentiation and cell-cell adhesion in convalescent COVID-19 patients
To characterize changes in individual T cell subsets after SARS-CoV-2 infection among the subjects across three conditions, we subclustered T cells from PBMCs and obtained 13 subsets according to the expression and distribution of canonical T cell markers using UMAP. These cell types included CD4+ naïve, CD4+ memory, CD4+ effector memory, Th2, Treg, CD8+ naïve, CD8+ effector-GNLY, CD8+ effector-GZMK, CD8+ effector memory, gdT, NKT, NKT-CD160 and MAIT (Fig. 4a-c; Supplementary Figure 2).
It demonstrated that 142 genes related to T cell responses were significantly differentially expressed across disease conditions (Fig. 4d; Supplementary Table 10 and 11). We found that significantly DEGs were involved in T cell and lymphocyte activation, differentiation and proliferation including T cell activation (GO:0042110), positive regulation of T cell activation (GO:0050870), regulation of T cell activation (GO:0050863), T cell differentiation (GO:0030217), regulation of lymphocyte activation (GO:0051249), positive regulation of lymphocyte activation (GO:0051251), lymphocyte differentiation (GO:0030098) and lymphocyte proliferation (GO:0046651), cell-cell adhesion such as leukocyte cell-cell adhesion (GO:0007159), positive regulation of leukocyte cell-cell adhesion (GO:1903039), regulation of leukocyte cell-cell adhesion (GO:1903037), positive regulation of cell-cell adhesion (GO:0022409), regulation of cell-cell adhesion (GO:0022407) and positive regulation of cell adhesion (GO:0045785) and myeloid cell responses including myeloid cell differentiation (GO:0030099),regulation of hemopoiesis (GO:1903706) and regulation of myeloid cell differentiation (GO:0045637) (Fig. 4e).
Same as analysis procedure of DEGs in innate cell subsets, we firstly classified DEGs in T cell subsets into four categories including (1) regulation of T cell activation, (2) regulation of lymphocyte and leukocyte activation (3) T cell, lymphocyte and myeloid cell differentiation and (4) regulation of cell-cell adhesion and then deeply explored the expression pattern of these genes involved in top 20 GO terms based on the classification. It depicted enhanced functions of activation, differentiation and adhesion of multiple T cell compartments from convalescent patients with COVID-19. Genes involved in biological processes of regulation of T cell activation and regulation of lymphocyte and leukocyte activation remarkably increased their expression from both convalescent patients with moderate and severe COVID-19. We found that CD8B were upregulated in CD8+ effector-GNLY, CD8+ effector memory, gdT and NKT and TRDC were upregulated in CD4+ effector memory, CD8+ naïve, CD8+ effector memory, NKT and NKT-CD160. SELL and S100A10 were upregulated in nearly more than 10 T cell subsets in convalescent patients with severe COVID-19. With the similar trend, genes, FOS, HSPA1A and HSPA1B, relating to biological process of T cell, lymphocyte and myeloid cell differentiation, were highly expressed in more than 10 T cell subsets in convalescent patients with severe COVID-19 while expression of Ly-1 antibody reactive clone (LYAR) was raised in CD4+ naïve, CD4+ memory, Th2, Treg, CD8+ naïve and NKT-CD160 from both kinds of convalescent patients. However, we observed a prominent feature that JUNB and ZFP36 declined in a wide range of T cell subsets from both kinds of convalescent subjects. The previous studies revealed that JUNB inhibits CD4+ T cell by promoting Th17 cell[38, 39] and ZFP36 RBPs in restraining T cell expansion and effector functions[40] (Fig. 5a; Supplementary Table 10 and 11). Combined with the above results, we summarize that functions of T cell were retained even strengthened in convalescent COVID-19 patients by clear endorsement of increased expression of genes involved in biological processes of regulation of T cell activation, regulation of lymphocyte and leukocyte activation, T cell, lymphocyte and myeloid cell differentiation and regulation of cell-cell adhesion.
To further characterize the functions of DEGs in 13 T cell subsets, we performed analyses of T cells from moderate or severe convalescent patients in comparison with those of HDs. Total DEGs from moderate or severe convalescent patients in comparison with those of HDs were shown in Supplementary Table 12 and the DEGs shared by both comparisons were presented in Fig. 5b and Supplementary Fig. 3. We found DEGs were involved in myeloid cell and myeloid leukocyte differentiation in CD4+ naïve, regulation of microtubule polymerization and Fc receptor signaling pathway in CD4+ memory (Fig. 5c), mRNA and RNA catabolic process in CD4+ effector memory, regulation of protein transport, localization and complex assembly in Treg, RNA catabolic process in CD8+ naïve, T cell activation and T cell receptor signaling pathway in CD8+ effector-GNLY, posttranslational protein folding in Th2, CD8+ effector-GZMK and CD8+ effector memory (Fig. 5c), cellular defense response and lymphocyte differentiation in gdT, T cell activation and regulation of hemopoiesis in NKT, T cell receptor signaling pathway and antigen receptor-mediated signaling pathway in NKT-CD160 (Fig. 5c), and neutrophil activation involved in immune response in MAIT in convalescent COVID-19 patients (Fig. 5c; Supplementary Fig 4).
Classical HLA class I genes e.g. HLA-A and HLA-B were downregulated in CD4+ naïve, CD4+ memory, CD4+ effector memory, Th2, Treg, CD8+ naïve, CD8+ effector-GNLY, CD8+ effector-GZMK, CD8+ effector memory, gdT, NKT and NKT-CD160 subsets and HLA-C was were downregulated in CD4+ memory, Th2, CD8+ naïve, CD8+ effector-GZMK, CD8+ effector memory, gdT, while Non-classical HLA class I genes e.g. HLA-E was also downregulated in CD4+ naïve, CD4+ memory, CD4+ effector memory, Th2, CD8+ naïve, CD8+ effector-GNLY, CD8+ effector memory, NKT and NKT-CD160 subsets of convalescent severe COVID-19 patients (Fig. 5d and Supplementary Table 13). However, three genes e.g. HLA-DPA1, HLA-DPB1 and HLA-DQB1 encoding HLA class II molecules were upregulated in CD4+ naïve, CD4+ memory, Th2, CD8+ naïve, CD8+ effector-GNLY, CD8+ effector-GZMK, CD8+ effector memory, NKT and NKT-CD160 subsets and other HLA class II molecules, e.g. HLA-DQA1 and HLA-DRB5 were upregulated in CD4+ effector memory, CD8+ effector-GNLY, CD8+ effector-GZMK, CD8+ effector memory, NKT and NKT-CD160 subset of convalescent severe COVID-19 patients (Fig. 5d and Supplementary Table 13).
2.5. Clonal expansion in T cells and preferred usage of V(D)J genes in convalescent COVID-19 patients.
Next, to gain insight into the clonal relationship among individual T cells and usage of V(D)J genes across three conditions, we reconstructed TCR sequences from the TCR sequencing. Briefly, there were more than 75% of cells in CD4+ memory, Th2 and Treg subsets, more than 70% of cells in CD4+ naïve, CD4+ effectory memory, CD8+ naïve and CD8+ effector-GNLY subsets, more than 50% of cells in CD8+ effectory memory and MAIT subsets with matched TCR sequences, except for the three γδT, NKT and NKT-CD160 subsets (Fig. 6a, b). Compared to the HDs, increased clonal expansion was found in patients with COVID-19 (Fig. 6c-e). The clonal expansion in the severe condition was higher than that of the HD and moderate conditions. Meanwhile, the clonal expansions (clonal size 50-100) were observed both in moderate and severe condition whereas clonal expansions (clonal size >100) were only observed in severe condition (Fig. 6e), revealing that clonal expansion of effector T cells might be active in severe patients. We surveyed different degrees of clonal expansion among T cell subsets (Fig. 6c and d). CD4+ effectory memory, CD8+ effectory-GZMK and CD4+ effectory memory showed high proportions of clonal cells both in patients increased with disease severity (Fig. 6d), suggesting that effector T cells underwent dynamic state transitions. We used STARTRAC to analyze the expansion and transition of T cell subsets. The results shown that CD4+ effectory memory and CD8+ effector-GNLY increased transition in severe than those in moderate and HD respectively (Fig. 6f; Supplementary Table 14).
To study the dynamics and gene preference of TCRs in COVID-19 patients and HDs, we compared the usage of V(D)J genes across three conditions. The top 10 complementarity determining region 3 (CDR3) sequences were different across three conditions. The moderate and HD shared two CDR3 sequences. The usage percentage of the top 10 CDR3 sequences in the HD condition was lower and more balanced compared to moderate and severe conditions (Fig. 6h). Of note, we discovered a different usage of V(D)J genes with decreased diversity in patients with COVID-19, which was more pronounced in TRA genes (Fig. 6i).
We found that convalescent COVID-19 severe patients exhibit a preferred TCRVα-region bias toward TRAJ13, TRAJ23, TRAJ4, TRAJ45, TRAJ53 and TCRJ-region bias toward TRAJ57 and TRAV12-1, TRAV13-1, TRAV14-DV4, TRAV26-1, TRAV35, TRAV4 and TRAV83 gene segments, respectively (Fig. 6g and Supplementary Table 15). Meanwhile, convalescent COVID-19 severe patients shown a preferred TCRVβ-region bias toward TRBJ1-6 and TRBJ2-7 TRVB27, TRVB30, TRVB5-6 and TRVB6-1 (Fig. 6g and Supplementary Table 15).
To study biased V(D)J rearrangements of the TCR, we compared the usage of V(D)J genes across three conditions (Fig. 6g and Supplementary Table 15). We observed more specific V(D)J usage in the severe patients compared with the other two groups, indicating that T cells might have undergone unique and specific V(D)J rearrangements in severe COVID-19 patients (Fig. 6g, i). We also discovered paired genes of TRAV1-2/TRAJ33 in all HDs and COVID-19 patients (Fig. 6i). However, there were some unique V-J pairs in severe group. The shared V-J pairs e.g. TRAV27/TRAJ2-5, TRAV19/TRAJ8, TRAV19/TRAJ6, TRAV19/TRAJ57, TRAV21/TRAJ48, TRAV8-3/TRAJ43, TRAV19/TRAJ8, TRAV38-2/DV8/TRAJ40, TRAV14/DV4/TRAJ36 and TRAV12-1/TRAJ8 were only found in in severe group (Fig. 6i). In brief, increased T cell clonality and skewed usage of the TRAV and TRAJ genes in severe subjects indicate that infection with SARS-CoV-2 is closely associated with V(D)J rearrangements in T cells of the hosts. Notably, selective usage of dominant genes, especially TRAV27/TRAJ2-5, TRAV19/TRAJ8, TRAV19/TRAJ6, TRAV19/TRAJ57, TRAV21/TRAJ48, TRAV8-3/TRAJ43, TRAV19/TRAJ8, TRAV38-2/DV8/TRAJ40, TRAV14/DV4/TRAJ36 and TRAV12-1/TRAJ8 in severe patients, may facilitate the design of vaccines or therapeutics.
2.6. Dual functions of B cell subsets in convalescent COVID-19 patients
To trace the dynamic changes of different B subtypes, we subclustered B cells into five subsets according to the expression and distribution of canonical B cell markers. We identified Pre B, Maginal zone B, Follicular B, Plasma B and Memory B (Fig. 7a and b and Supplementary Fig 5 and 6). Notably, the proportions of plasma B and Follicular B subsets increased in convalescent severe COVID-19 patients in comparison with those of HDs. In contrast, the proportion of memory B cells decreased in convalescent severe COVID-19 patients (Fig. 7c).
To further investigate diverse transcriptomic changes in B cells post SARS-CoV-2 infection, we compared the expression profiles of B cells of the moderate or severe condition to those of the HD condition. It demonstrated that 82 genes related to B cell responses were significantly differentially expressed across disease conditions (Fig. 7d; Supplementary Table 16 and 17). We found that significantly DEGs were involved in humoral immune response (GO:0006959), B cell activation (GO:0042113; GO:0050864) and cell surface signaling pathway including immune response-activating cell surface receptor signaling pathway (GO:0002429), antigen receptor-mediated signaling pathway (GO:0050851) and B cell receptor signaling pathway (GO:0050853) (Fig. 7e).
Humoral immunity plays an important role against viral infection. Then we further unbiasly mapped the status of B cell subtypes, using the same analysis procedures as innate and T cells, by evaluating the expression pattern of DEGs involved in top 20 GO terms based on classification of their functions. We classified these genes into two categories e.g. B cell activation and humoral immune response. Two genes, Kruppel-like factor 6 (KLF6) and Fc receptor-like 1 (FcRL1), relevant to B cell activation, were observed to be upregulated in B cells from moderate convalescent COVID-19 patients, with KLF6 increased in Follicular B and Plasma B whereas FcRL1 increased in Follicular B, Plasma B and Memory B. It is noteworthy that FcRL1 was upregulated in Memory B only from severe convalescent COVID-19 patients. There is a study shows that FcRL1 enhances B cell activation and function[41]. In another panel of genes relevant to humoral immune response, we found two molecules S100A8 and S100A9 upregulated in Plasma B only from severe subjects. However, two genes showed an opposite status, revealed S100A9 were upregulated and POU2AF1 were downregulated in Memory B both from moderate and severe convalescent COVID-19 patients. POU2AF1 and ACTG1 demonstrated a decreased expression in Memory B and Plasma B from moderate subjects. In contrary, JCHAIN was upregulated in Maginal zone B, Follicular B and Plasma B from severe convalescent subjects (Fig. 7d, e, f; Supplementary Table 16 and 17). Token together, the findings revealed the sophisticated and dual functions of B cells during convalescence of COVID-19, providing a novel mechanism that B cell activation was retained especially in moderate while humoral immune response was weakened.
We also found that Plasma B, Follicular B, Memory B and Maginal zone B significantly downregulated B cell activation (GO: 0042113), B cell differentiation (GO: 0030183) and lymphocyte differentiation (GO: 0030098) in severe patients compared to HD group (Supplementary Figure 6). These results suggest that B cell activation and differentiation responses including neutrophil response and interferon-gamma response were attenuated in convalescent COVID-19 patients. To further characterize the functions of DEGs in B cells, we performed analyses B cells of moderate or severe convalescent patients in comparison with those of HDs. Total 82 DEGs from moderate or severe convalescent patients in comparison with those of HDs were shown in Supplementary Table 18 and the DEGs shared by both comparisons were presented in Fig. 7g. We found DEGs were involved in neutrophil activation involved in immune response in Plasma B, regulation of mRNA processing in Pre B, B cell receptor signaling pathway in Follicular B in convalescent COVID-19 patients (Fig. 7h).
Classical HLA class I genes HLA-A and HLA-B were significantly downregulated in Maginal zone B, Follicular B, Plasma B and Memory B subsets from both moderate or severe convalescent patients and HLA-C was were significantly upregulated in Maginal zone B, Follicular B, Plasma B and Memory B subsets from moderate convalescent patients but downregulated in Maginal zone B, Plasma B and Memory B subsets from severe convalescent patients. Non-classical HLA class I genes HLA-DPA1, HLA-DPB1, HLA-DQB1, HLA-DRA and HLA-DRB1 were significantly upregulated in Maginal zone B and Memory B subsets from severe convalescent patients while HLA-DPA1, HLA-DQB1 and HLA-DRB1 were significantly downregulated in Follicular B subsets from severe convalescent patients. Furthermore, we found that HLA-DRB5 was significantly upregulated in Maginal zone B, Follicular B and Memory B subsets from both moderate or severe convalescent patients (Fig. 7i; Supplementary Table 19).
2.7. Expanded B cells and specific rearrangements of V(D)J genes in severe patients.
Next, to gain insight into the clonal relationship among individual B cells and usage of V(D)J genes across three conditions, we reconstructed TCR sequences from the BCR sequencing. Briefly, there were more than 75% of BCRs were detected of cells in Maginal zone B, Follicular B, Plasma B and Memory B subsets (Fig 8a, b). Compared to the HDs, it shown an inconsistently tendency in clonal expansion between T and B cell subsets. Although the clonal expansion was obviously detected in Plasma B subsets from severe convalescent patients, in other B cell subsets e.g. Maginal zone B, Follicular B and Memory B, the clonal expansion level was not significantly increased from both moderate or severe convalescent patients, even weakened in later patients (Fig. 8c-d).
Meanwhile, clonal expansions (clonal size >100) were decreased both in moderate and severe condition, especially in moderate group (Fig. 8e), indicating that convalescent COVID-19 patients might have a faded clonal expansion of effector B cells. We also used STARTRAC to analyze the expansion and transition of B cell subsets. The results shown there was no significant difference in expansion and transition among subjects (Supplementary Fig. 7a and b; Supplementary Table 20).
Next, we explored the distribution of IgA, IgD, IgG and IgM (IgE not detected) in each patient at moderate or severe convalescent conditions, respectively. In most patients, IgM was the predominant immunoglobulin. Compared to moderate convalescent patients, the abundance of IgA was significantly decreased, whereas IgM increased in severe ones (Fig. 8f). The top 10 CDR3 sequences were differently across three conditions (Fig. 8h). To study biased V(D)J rearrangements of the BCR, we compared the usage of V(D)J genes across three conditions. We observed more specific V(D)J usage in the severe condition compared with the other two groups, indicating that B cells might have undergone unique and specific V(D)J rearrangements in severe COVID-19 patients (Fig. 8g and i). We also found comprehensive usage of IGHJ4, IGKJ1 and IGLJ2 in all HDs and patients, but the paired genes among IGHV/IGHJ4, IGKV/IGKJ1 and IGLV/IGLJ2 were different in patients compared with HDs. The shared V-J pairs IGHV3-33/IGHJ4, IGHV4-39/IGHJ4, IGHV3-23/IGHJ4, IGKV1-5/IGKJ1, IGKV1D-39/IGKJ1, IGKV3-20/IGKJ1 and IGLV2-14/IGLJ2 were detected in both moderate and severe patients whereas IGHV3-11/IGHJ4 and IGLV3-21/IGLJ2 were only observed in HDs, respectively. However, it attracted our attention that there were unique V-J pairs in severe group, such as IGHV4-59/IGHJ1, IGHV4-61/IGHJ3 and IGLV2-18/IGLJ2. similarly with T cells, increased B cell clonality and skewed usage of the IGHV and IGKJ genes in severe COVID-19 patients suggest that SARS-CoV-2 infection is closely associated with V(D)J rearrangements in B cells of the host. Notably, selective usage of dominant IGV genes, especially IGHV3-7/IGHJ4,IGHV3-33/IGHJ4, IGHV4-61/IGHJ4, IGHV4-39/IGHJ4, IGHV3-23/IGHJ4, IGLV2-14/IGLJ2, IGLV2-18/IGLJ2, IGKV1-5/IGKJ1, IGKV1D-39/IGKJ1, IGKV3-20/IGKJ1, IGHV4-59/IGHJ1 and IGHV4-61/IGHJ3 in severe patients, may facilitate the design of vaccines.