Elderly patients have a high risk of Lymphopenia.
A high rate of severe COVID-19 was reported in immunocompromised patients, suggesting an insufficient rather than an overactive antiviral immunity could be the basis for the development of this disease3,15. Meanwhile, Lymphopenia, a reduction in the number of lymphocytes in the blood, was associated with the disease severity of COVID-193,16. We analyzed the incidence of lymphopenia in 284 patients infected with SARS-CoV-2 collected from designated hospitals in Fujian province of China (basic information of the patients was listed in Table S1), and found that a reduction of lymphocytes was more frequently observed in aged patients (Fig. 1a). These findings denote the pivotal role of the adaptive immunity for the viral clearance and disease control.
scRNA-seq analysis identified main clusters in lymphocyte.
We hypothesized that single-cell transcriptomic analysis of peripheral lymphocytes during the course of the disease might clarify the role of the adaptive immune system in this disease. Therefore 13 samples of peripheral blood mononuclear cells (PBMCs) were collected from 10 patients at different stages of the disease, namely pre-severe disease (PR, 1 sample), severe disease (SD, 3 samples), post severe disease (PS, 3 samples), post mild disease (PM, 3 samples) and convalescence of mild disease (CM, 3 samples), which also included seven samples from our previous report (samples without enough number of lymphocytes were not included)7. Among all enrolled patients, four of them experienced severe disease under intensive care and the other six showed fever, cough, chill, and fatigue etc mild symptoms. Results of COVID-19 examination and times of sample collection were shown in Fig. 1b. Patient information was listed in Table S1. Samples of PM were collected from patients during hospitalization whose viral testing turned negative after treatment. Convalescent samples were collected within the first 5 days after hospital discharge. Three normal PBMCs from healthy donors (HD) were used as controls (named HD-1, HD-2, and HD-3).
We performed single-cell mRNA sequencing (scRNA-seq) of these samples on the 10x genomics platform. 243 million RNA transcripts in 91,649 cells were obtained after filtering cells with low quality. We then used t-Distributed Stochastic Neighbor Embedding (t-SNE) to visualize the clusters of all the cells identified by a shared nearest neighbor (SNN) modularity optimization based clustering algorithm implemented in Seurat17 (Fig. 1c). The three major groups of immune cells in PBMCs, namely myeloid, B and T cells were clearly identified by specific gene expression signatures (Fig. 1d). Classic lineage markers further confirmed the identities of these clusters (Fig. 1e). We then focused on adaptive immune cells, and grouped the lymphocytes into B cells, CD4 and CD8 T cells (Fig. 1f and S1a). Analysis of the lymphocyte composition at different disease stages revealed a significant reduction in T cells along with increase in B cells in SD and PS samples (Fig. 1h), which indicates the abnormal dynamics of adaptive immune cells in patients with severe disease.
Humoral and cellular immune responses in patients with severe disease.
Effector cells carry out the major functions of the adaptive immune system, and decide the efficacy of adaptive immunity against viral infection. We checked the distribution of effectors in the clusters of lymphocytes. Despite the small numbers of cells identified, the plasma cells were prominently grouped (Fig. 1f). The T effectors were identified by the expression of GZMB, an effective molecule specifically expressed in activated T cells. As shown in Fig. 1g, most of the effector T cells were located in the CD8 cluster, while CD4 effectors were barely detected.
In order to examine changes of effectors in different types of adaptive immune cells, we performed separate cluster analysis on the B, CD4 and CD8 T cells. B cells were grouped into 4 major clusters, namely mature, memory, activated B cells and plasma cells (Fig. 2a, S1b and S1e). CD4 T cells consisted of naïve, memory, effector and regulatory T cell clusters (Fig. 2c and S1c), while clusters of naïve, memory, effector and mucosal-associated invariant T cells (MAIT) were found In CD8 T cells (Fig. 2e and S1d). We found that the levels of effectors in these cells displayed different dynamics during the disease. The frequency of effectors in B cells was significantly increased at the SD stage but substantially reduced at the PS stage (Fig. 2b). Among CD4 T cells, Low levels of effectors were observed in all of the samples except for the one from PR (Fig. 2d). Consistent with the above observation (Fig. 1g), a large portion of effectors was found in CD8 T cells (Fig. 2e), but a significant low level of CD8 effectors was found at the PS stage (Fig. 2f).
Next, we examined changes of effectors in the humoral and cellular immune responses by measuring the frequencies of B (plasma cells plus activated B cells) and T effectors (CD4 plus CD8 effectors) in total lymphocytes respectively. In consistent with the observation in B cells, high levels of B effectors were observed in lymphocytes at the SD stage (Fig. 2g). For T effectors, however, low frequencies were found in lymphocytes at both the SD and PS stages (Fig. 2h), which was different from the observation in CD8 T cells among which low levels of effectors were only observed at the PS stage (Fig. 2f), This could be explained by the reduced portion of total T cells in lymphocytes at these two stages (Fig. 1h). Taken together, the dynamics of effector levels in lymphocytes revealed a coincidence of high humoral and low cellular immune responses at the stage of SD, which likely contributes to the progression of severe COVID-19.
Excessive activation led to exhaustion of CD8 T effector cells in sever COVID-19.
In order to understand the reduction of effector cells, we increased the modularity to group more distinct subsets, which allow us to explore the changes inside the major clusters. In CD8 T cells, two naïve, three memory and two effector clusters were identified after stringent grouping (Fig. 3a). No sub-clusters were detected in the MAIT cells. Gene expression signatures and pseudotime analysis demonstrated unbiased cellular dynamic processes of these sub-clusters (Fig. S2a and S2b). We then compared the two groups of effector T cells. Differentially expressed gene (DEG) and pathway enrichment analysis showed that one group was excessively activated, which was then named as Tea (excessively activate T cells). Compared to the normal effectors (Te), Tea cells had higher expression of activation genes (FCER1G, KLRB1, KLRF1, NKG7 and IGFBP7), co-inhibitory receptors (LAG3, CD300A, CD244, CD160 and HVCR2), IFN downstream signaling molecules (JAK1, IFITM2, IFITM3, TYK and IRF8), and chemokines (CCL3 and CCL5) (Fig. 3b and 3c). Significant reductions in TCR signaling molecules (CD3D, CD3G, CD8A, CD8B and ZNF683), and cytokine/receptor genes (IL7R, IL2RB, IL32 and IFNG) were found in Tea cells. For effector molecules, Tea cells expressed less GZMB but more PRF1 than Te cells. Although it is known that activation induces cell death in T cells18-21, we did not observe any significant difference in expression of apoptosis genes or pathways between these two clusters. This might have been due to the process of single cell analysis which excluded apoptotic cells to ensure the accuracy of clustering. Indeed, although the Tea cells expressed slightly more FAS and FASLG, both Tea and Te clusters expressed extremely low levels of these two genes (Fig. S2d). IFN signaling might be involved in the differentiation of Tea cells. Although CCL3 and CCL5 secreted by Tea cells may promote innate immunity, the slight increase of these two chemokines argued they were an effectual reservoir. As shown in Fig. 3d, CD8 effectors were dominated by Tea cells in SD samples, indicating that the adaptive immune system had been overwhelmed by these dysfunctional cells. Together, these data support that the Tea cells are a group of excessively activated T cells with exhausted phenotype and diminished function of antigen recognition. Excessive activation-induced continuous expansion of Tea cells may waste the majority of T cells, and lead to lymphopenia which paralyzes the adaptive immune system.
We then checked the CD8 memory T cells. The three sub-clusters were named as Tm-1, Tm-2 and Tm-3. Significant accumulation of Tm-3 was observed in the SD and PS stage (Fig. 3e), suggesting that this group of cells was specifically generated during severe disease. The percentage of Tm-3 in CD8 memory T cells was strongly correlated with the proportion of Tea in the CD8 effector cells (R2=0.7813) (Fig. 3f), suggesting that Tm-3 might be a group of memory cells derived from the Tea cells during excessive activation. Differential gene expression analysis and pathway enrichment analysis showed that Tm-3 was a cluster of highly proliferating cells with features of memory stem cells (Fig. S2e and S2f). Compared to Tm-1 and Tm-2, Tm-3 had high expression of T memory stem cell markers (SELL, CXCR3, CCR7, FAS, CD27 and CD28), and the proliferation gene Ki67 (Fig. 3g)22. Furthermore, a slightly higher expression of GZMA and lower level of IL7R in the Tm-3, as compared to Tm-1 and Tm-2, also showed that this was a group of recently developed memory T cells. The low level of Tm-3 at the PR stage further confirmed its differentiation during severe disease stages (Fig. 3e).
Although both Tea and Tm-3 cells experienced excessive activation, exhaustion was observed in the Tea but not Tm-3 cells. We then performed DEG and pathway-enrichment analysis to understand the difference between these two groups of cells. The Tea cells had comparatively higher-level expression of T cell activation and effector genes (GNLY, NKG7 and GZMB) but almost had no expression of Ki67 (Fig. 3h and 3i), suggesting that they were end-differentiated cells and more prone to death. In contrast, the Tm-3 cells had high expression of genes involved in epigenetic modification (DNMT1 and EZH2), oxidative phosphorylation (NDUB6 and NDUFV2), regulation of telomerase, cell cycle and proliferation etc (Fig. 3h and 3i). These distinct signaling transduction and epigenetic changes might have shaped the stem-like memory phenotype of Tm-3. Interestingly, we found the expression of
Prothymosin alpha (PTMA) was also significantly increased in Tm-3 (Fig. 3h and 3j). Thymosin alpha-1 (Tα1), the first 28 amino acids of PTMA, was found to be involved in T cell development, and has been used for the treatment of certain infection diseases including COVID-1923. In addition, Tea expressed less PTMA than Te, and Tm-3 expressed the highest level of PTMA in all the sub-clusters of CD8 T cells (Fig. 3j). We thus suspected that Tα1 might protect T cells from excessive activation.
We further analyzed the subsets in naïve T cells but did not observed any significant change during the different disease stages (Data not shown). Since no sub-clusters were identified in the effector clusters of B cells and CD4 T cells, we did not analyze the subsets in these two types of adaptive immune cells (Data not shown).
Thymosin alpha-1 protected T cells from excessive activation.
Next, we tested the effect of Tα1 on T cell activation. PBMCs from healthy donors were activated by anti-CD3/CD28 antibodies in vitro and treated with 200ng/ml Tα1 for 3 days, followed by cultures with IL-2 (200U/ml) and Tα1 (200 ng/ml) for 6 more days. Compared to the control group, Tα1 had significantly increased T cell numbers at day 6 and 9 although a slightly decreased number in T cells was observed at day 3 (Fig. 4a), indicating that Tα1 promoted the proliferation of T cells after activation. After 3 days of activation, we found that cell size measured by FSC and SSC was slightly reduced in the group with Tα1 (Fig 4b), suggesting that these T cells were less activated. Tα1 treatment did not change the proportion of CD4 and CD8 T cells (Fig 4c), but reduced the production of IFNγ and TNFα, although no significant statistical difference was observed (Fig 4d). Significant reduction of granzyme B was observed in both CD4 and CD8 T cells, while no change in the expression of PD-1 was observed. These data showed that Tα1 could reduce T cell activation and promote the proliferation of T cells after activation, indicating it may protect the T cells from excessive activation.
The use of Tα1 in some COVID-19 patients may allow the evaluation of its impact on lymphocytes. The data from 25 severe and critical COVID-19 cases treated at the Huoshenshan hospital (Wuhan, China) were collected (basic information of the patients was listed in Table S1). Of them, 11 patients received daily Tα1 treatment for at least one week, while the other 14 patients were not treated with Tα1 during the hospitalization period. Compared to the non-treated patients, the lymphocyte counts of the treated patients were significantly increased after one week of Tα1 treatment (Fig. 4e). The fold change of lymphocytes counts in each patient was shown in Fig. 4f. These data showed that Tα1 enhanced the number of lymphocytes in patients with severe and critical disease. Due to the limited number of patients in this retrospective analysis, we were not able to clearly evaluate the clinical benefits of Tα1 treatment. Nevertheless, our data suggest that the administration of Tα1 could be a potential approach to protect T cells from excessive activation in COVID-19.